Iom Marc Nuwer PDF

HANDBOOK OF CLINICAL NEUROPHYSIOLOGY
VOLUME 8
INTRAOPERATIVE MONITORING OF NEURAL FUNCTION
This page intentionally left blank
Handbook of Clinical Neurophysiology

Series Editors
Jasper R. Daube
EMG Laboratory, Gonda 8, Mayo Clinic and Mayo Foundation, 200 First Street SW, Rochester,
MN 55905, USA
and
Francois Mauguie`re
Functional Neurology and Epilepsy Department, Hopital Neurologique Pierre Wertheimer,
59 Boulevard Pinel, F-69394 Lyon Cedex 03, France
Volume 8
Intraoperative Monitoring of Neural Function
Volume Editor
Marc R. Nuwer
Department of Neurology and Clinical Neurophysiology, UCLA School of Medicine, Reed
Neurological Research Center, Room 1194, 710 Westwood Plaza, Los Angeles, CA 90095, USA
2008
Amsterdam Boston Heidelberg London New York

San Francisco Singapore Sydney Tokyo
Oxford
Paris
San Diego
Radarweg 29, 1043 NX, Amsterdam, The Netherlands
# 2008, Elsevier B.V. All rights reserved.

The right of Jun Kimura to be identified as editor of this work has been asserted by him in accordance with the
Copyright, Designs and Patents Act 1988
No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any
means, electronic, mechanical, photocopying, recording or otherwise, without either the prior permission of the
publishers or a licence permitting restricted copying in the United Kingdom issued by the Copyright Licensing
Agency, 90 Tottenham Court Road, London W1T 4LP. Permissions may be sought directly from Elseviers
Health Sciences Rights Department in Philadelphia, USA: phone: (1) 215 238 3804, fax: (1) 215 238 3805,
e-mail: healthpermissions@elsevier.com. You may also complete your request on-line via the Elsevier
homepage (http://www.elsevier.com), by selecting Customer Support and then Obtaining Permissions.
This edition published 2008
ISBN 978-0-444-52905-3
ISSN 1567-4231
British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging in Publication Data
A catalog record for this book is available from the Library of Congress
Note
Knowledge and best practice in this field are constantly changing. As new research and experience broaden our
knowledge, changes in practice, treatment and drug therapy may become necessary or appropriate. Readers are
advised to check the most current information provided (i) on procedures featured or (ii) by the manufacturer of
each product to be administered, to verify the recommended dose or formula, the method and duration of
administration, and contraindications. It is the responsibility of the practitioner, relying on their own experience
and knowledge of the patient, to make diagnoses, to determine dosages and the best treatment for each
individual patient, and to take all appropriate safety precautions. To the fullest extent of the law, neither the
Publisher nor the authors assume any liability for any injury and/or damage to persons or property arising out of
or related to any use of the material contained in this book.
The Publisher
The
Publisher's
policy is to use
paper manufactured
from sustainable forests
Printed in the Netherlands
Dedication
This book is dedicated to Jamie, Charles, Stephen, and Catherine. They shared time with this project to
allow me the opportunity to organize and create this book.
Acknowledgments
This work would not have been possible without the dedicated work of my assistants, Amy McGann, Sylvia
Fong, and Benjamin Caplan. Their organizational and editorial assistance greatly facilitated production of this
volume.
The book was initiated at the request of series editor, Jasper Daube. His helpful guidance was instrumental
in establishing the format of the book. His many helpful suggestions greatly enhanced the content and directions
for this work.
My own role and expertise in this field also was facilitated and enhanced by my UCLA colleague, James
Packwood PhD. Together we ventured into this field even when many others thought that recording reliable
evoked potentials in the operating room was unachievable, and when no other neurologist provided such a
service. I wish him well in his recent retirement after three decades of work in intraoperative monitoring at
UCLA.
I appreciate too the authors of the individual chapters in this volume. This is a very diverse set of experts.
Compilation of such knowledge into one volume has been an unprecedented undertaking for this field of
intraoperative monitoring.
Marc R. Nuwer
Volume Editor
Foreword
Clinical neurophysiology encompasses the application of a wide variety of electrophysiologic methods to the
analysis and recording of normal function, as well as to the diagnosis and treatment of diseases involving
the central nervous system, peripheral nervous system, autonomic nervous system, and muscles. The steady
increase in the growth of subspecialty knowledge and skill in neurology has led to the need for a compilation
of the whole range of physiologic methods applied in each of the major categories of neurologic disease. While
some of the methods are applied to a single category of disease, most are useful in multiple clinical settings.
Each volume is designed to serve as the ultimate reference source for academic clinical neurophysiologists
and as a reference for specialists in the area. It will provide the information needed to fully understand the
physiology and pathophysiology of disorders in their patients. As such, these volumes will also serve as a major
teaching text for trainees in each of the subspecialties.
The Handbook volumes cover all of the clinical disorders served by clinical neurophysiology, including the
muscle and neuromuscular junction diseases, epilepsy, surgical epilepsy, motor system disorders, peripheral
nerve disease, autonomic dysfunction, somatosensory system disorders, behavioral disorders, visual and
auditory system disorders, and monitoring neural function. Each will focus on the advances in one of these
major areas of clinical neurophysiology. Each volume will include critical discussion of new knowledge in
basic neurophysiology, and its application to different central nervous systems.
Each volume will include an overview of the field, followed by a section that includes a detailed description of
each of the CNP techniques used, and a third section discussing electrophysiologic findings in specific situations.
The latter will include how to evaluate each along with a comparison of the relative contribution of each of the
methods. A final section will discuss ongoing research studies, and anticipated future advances.
The application of clinical neurophysiology methods to monitoring in the operating room and in intensive
care has expanded rapidly over the past 10 years. This is seen in many research publications in neurosurgery,
orthopedic, vascular, otolaryngology and neurology, as well as the publication of books dedicated to nervous
system monitoring. The Accreditation Council of Graduate Medical Education has recently recognized
monitoring neural function during surgery as a subspecialty of neurology.
We are privileged to have Marc Nuwer, a pioneer in the development of monitoring neural function, as the
volume editor. He has done a superb job of assembling world leaders in the description of the methods and in
their application to a wide range of diseases and settings.
The volume describes a multiplicity of electrophysiologic methods that are being applied to the many types
of surgery in which neural structures are at risk for loss of function. Special emphasis is on surgical procedures
where function critical structures, like the spinal cord, can be preserved. Wherever possibly applicable, the
information presented focuses on evidence-based medicine; the specificity and sensitivity of each mode of
monitoring is provided when known, along with comparison of their relative values.
Jasper R. Daube, MD
Rochester, MN, USA
Francois Mauguie`re, MD
Lyon, France
Series Editors
Preface
The best way to deal with paraplegia is to prevent it from happening in the first place. This is the goal of
intraoperative monitoring (IOM) and testing. The IOM mission now includes many techniques beyond spinal
cord monitoring. Analogous goals remain the mission across the many other IOM applications.
In the past three decades, IOM and testing has blossomed into a major area of clinical neurophysiology, now
widely used, and offering diverse techniques and applications. Most of the technology is familiar to clinical
neurophysiologists who practice routine outpatient testing. Applications have made inroads into many surgical
subspecialties, as reflected in this volumes Table of Contents. This volume presents the state of the art and
science, separated into several sections.
Section I reviews general issues of science and practice behind IOM. This includes an overview, some
history, and issues about staffing. Anatomy and physiology are complied as a convenient reference for users.
Effects of anesthetics are reviewed.
Section II presents the various techniques. Each chapter describes typical techniques, many with some
didactic examples. Descriptions are sufficiently detailed so that a new user would know how to stimulate
and record, and what changes to consider significant. The chapters also discuss the drawbacks or problems,
and how monitorists might cope with them.
Section III presents the applications to clinical disorders and situations. Each chapter reviews the relevant
literature. It discusses the clinical issues and any outcome studies for IOM and testing when used during a
particular surgical application.
Section IV digresses into the intensive care unit (ICU). Monitoring in surgery encouraged development of
monitoring for the immediate postoperative period in the ICU. Such monitoring also evolved to include uses
well beyond postoperative care. Chapters here include the search for nonconvulsive seizures in adults and
neonates.
Section V concludes the book with several related topics. Safety remains a concern for all monitoring, and
we adhere to the traditional medical dictum primum non nocere above all, do no harm. Infrared and ultrasound monitoring techniques are reviewed in their own chapters. Another chapter describes techniques for
monitoring depth of anesthesia.
Wilder Penfield took neurophysiology to the operating room to localize motor and sensory cortex with
direct cortical stimulation. When he did so in the 1920s, I wonder if he had any idea how much intraoperative
neurophysiology would grow over the next 80 years?
Marc R. Nuwer
Volume Editor
List of Contributors
Aglio, L.S.
Department of Anesthesiology and Perioperative Medicine, Brigham and

Womens Hospital, Harvard University, 75 Francis Street, Boston,
MA 02115, USA.
Ashram, Y.A.
D ABNM, University of Alexandria, 507 El Horreya Ave., School of

Medicine, Fleming, Alexandria, Egypt.
Ball, B.
Department of Orthopedic Surgery, Mayo Clinic, 200 First Street SW,

Rochester, MN 55905, USA.
Balzer, J.R.
Departments of Neurological Surgery and Neuroscience, University of

Pittsburgh Medical Center, Suite B-400, 200 Lothrop Street, Pittsburgh,
PA 15213, USA.
Becker, D.P.
Department of Surgery/Neurosurgery, University of California, Los Angeles,

CA 90095, USA.
Berenstein, A.
Center for Endovascular Surgery and Intraoperative Neurophysiology,

Institute for Neurology and Neurosurgery, Roosevelt Hospital, New York,
NY 10019, USA.
Bishop, A.T.
Mayo Clinic, Brachial Plexus Clinic, Rochester, MN 55905, USA.
Burke, D.
Office of Research and Development, College of Health Sciences, Level 2,

Medical Foundation Building K-25, University of Sydney, Sydney, NSW
2006, Australia.
Canalis, R.F.
Department of Surgery/Head and Neck Surgery, University of California,

Los Angeles, CA 90095, USA.
Cannestra, A.F.
Department of Surgery/Neurosurgery, University of California, Los Angeles,

CA 90095, USA.
Cascino, G.D.
Department of Neurology, Division of Epilepsy and EEG, Mayo Clinic, 200

First St. SW, Rochester, MN 55905, USA.
Claassen, J.
Division of Critical Care Neurology and Comprehensive Epilepsy Center,

Department of Neurology, Columbia University, Neurological Institute,
Box 91, 710 W 168th Street, New York, NY 10032, USA.
Cozzens, J.W.
Department of Neurosurgery, Feinberg School of Medicine, Northwestern

University, Evanston Hospital, Evanston, IL 60201, USA.
Crammond, D.
Department of Neurological Surgery, University of Pittsburgh Medical

Center, Suite B-400, 200 Lothrop Street, Pittsburgh, PA 15213, USA.
xii
LIST OF CONTRIBUTORS
Crum, B.A.
Department of Neurology, W8A, Mayo Clinic College of Medicine, 200

First Street SW, Rochester, MN 55905, USA.
Daube, J.R.
Department of Neurology, E8B, Mayo Clinic College of Medicine, 200 First

Street SW, Rochester, MN 55905, USA.
Deletis, V.
Intraoperative Neurophysiology, Institute for Neurology and Neurosurgery,

Suite 11G-78, St. Lukes Roosevelt Hospital, 1000 Tenth Avenue,
New York, NY 10019, USA.
De Vries, L.S.
Department of Neonatology, Wilhelmina Childrens Hospital, UMC, PO Box

85090, 3508 AB Utrecht, The Netherlands.
Dong, C.C.J.
Department of Surgery, Vancouver General Hospital, EEG Lab., CP Ground

Floor, 855 W 12th Avenue, Vancouver, BC V5Z 1M9, Canada.
Duffau, H.
Department of Neurosurgery, Hopital Gui de Chauliac, CHU de Montpellier,

80 Avenue Augustin Fliche, 34295 Montpellier Cedex 5, France.
Edmonds, H.L.
Cardiovascular Services, Surgical Monitoring Associates, Inc., 3712

Plymouth Road, Louisville, KY 40207, USA.
Eliashiv, D.
Department of Clinical Neurophysiology, Cedars Sinai Hospital, 8700

Beverly Boulevard, South Tower, Los Angeles, CA 90048, USA.
Emerson, R.G.
Department of Clinical Neurology, Columbia University College of

Physicians and Surgeons, New York, NY 10032, USA.
Engel, Jr., J.
Department of Neurology, Geffen School of Medicine, UCLA, 710

Westwood Plaza, Suite 1250, Los Angeles, CA 90095, USA.
Fee, D.
Department of Neurology, University of Kentucky Chandler Medical Center,

KY Clinic, Room L-445, Lexington, KY 40536, USA.
Fehlings, M.G.
Division of Neurosurgery, Toronto Western Hospital, 399 Bathurst Street,

Toronto, Ont. M5T 2S8, Canada.
Gonzalez, A.A.
Department of Neurology, Keck School of Medicine, University of Southern

California, Healthcare Consultation Center II, 1520 San Pablo Street, Suite
3000, Los Angeles, CA 90033-4606, USA.
Guerit, J.-M.
Stress and Craniofacial Pain Clinic, Clinique Edith Cavell, Unite dExplorations
Electrophysiologiques du Syste`me Nerveux, CHIREC, Avenue Louise 390,
B-1050 Brussels, Belgium.
Gugino, L.D.
Department of Anesthesiology and Perioperative Medicine, Brigham and

Womens Hospital, Harvard University, 75 Francis Street, Boston, MA
02115, USA.
Habeych, M.
Department of Neurological Surgery, University of Pittsburgh Medical

Center, Suite B-400, 200 Lothrop St., Pittsburgh, PA 15213, USA.
Haghighi, S.S.
Department of Clinical Neurodiagnostic, Sharp Memorial Hospital, 7901

Frost Street, San Diego, CA 92123, USA.
Hellstrom-Westas, L.
Neonatal Intensive Care Unit, Department of Pediatrics, University Hospital,

SE-22185 Lund, Sweden.
xiii
Hemmerling, T.M.
Department of Anesthesiology, Hotel Dieu, 3840 St. Urbain, Montreal, PQ

H2W 1T8, Canada.
Hirsch, L.J.
Comprehensive Epilepsy Center, Department of Neurology, Columbia

University, Neurological Institute, Box NI-135, 710 W 168th Street,
New York, NY 10032, USA.
Huang, T.C.
Department of Otolaryngology, New York University School of Medicine,

550 First Avenue, NBV 5E5-10, New York, NY 10016, USA.
Huddleston, P.M.
Department of Orthopedic Surgery, Mayo Clinic, 200 First Street SW,

Husain, A.M.
Duke University Medical Center, Box 3678, 202 Bell Building, Durham,
NC 27710, USA.
Jantti, V.
Department of Clinical Neurophysiology, Tampere University Hospital,

P.O. Box 2000, FIN-33521 Tampere, Finland.
Jones, S.
Department of Clinical Neurophysiology, The National Hospital for

Neurology and Neurosurgery, Queen Square, London WC1N 3BG, UK.
Journee, H.L.
Department of Neurosurgery, University Medical Center Groningen, P.O.

Box 30.001, 9700 RB Groningen, The Netherlands.
Kelleher, M.O.
Division of Neurosurgery and Spinal Program, Krembil Neuroscience

Center, University of Toronto, Toronto, Ont. M5T 2S8, Canada.
Kombos, T.
Charite-Universitatsmedizin
Berlin,
Campus
Hindenburgdamm 30, D-12200 Berlin, Germany.
Kothbauer, K.F.
Division of Neurosurgery, Kantonsspital Luzern, CH-6000 Lucerne 16,

Switzerland.
Lalwani, A.K.
Department of Otolaryngology, New York University School of Medicine,

550 First Avenue, NBV 5E5-10, New York, NY 10016, USA.
Legatt, A.D.
Department of Neurology, Montefiore Medical Center, 111 East 210th Street,

Bronx, NY 10467, USA.
Lopez, J.R.
Department of Neurology and Neurological Sciences, Intraoperative

Neurophysiologic Monitoring Program, Stanford University School of
Medicine, 300 Pasteur Drive, Room A343, Stanford, CA 94305, USA.
MacDonald, D.B.
Section of Clinical Neurophysiology, Department of Neurosciences, King

Faisal Specialist Hospital and Research Center, MBC 76, PO Box 3354,
Riyadh 11211, Saudi Arabia.
Mandir, A.S.
Department of Neurology, Georgetown University, 3800 Reservoir Road

NW, Washington, DC 20007, USA.
Mashour, G.A.
Department of Anesthesia and Critical Care, Massachusetts General Hospital

and Harvard Medical School, Boston, MA 02114, USA.
Matthies, C.
Section of Functional Neurosurgery, Department of Neurosurgery, JuliusMaximilians-University of Wurzburg, Josef-Schneider-Strasse 11, D-97080
Wurzburg, Germany.
Benjamin
Franklin,
xiv
Mendiratta, A.
Department of Clinical Neurology, Columbia University College of

Physicians and Surgeons, New York, NY 10032, USA.
Minahan, R.E.
Department of Neurology, Georgetown University, 3800 Reservoir Road

Moed, B.R.
Department of Orthopedic Surgery, Saint Louis University School of

Medicine, 3635 Vista Avenue, 7th Floor, Desloge Towers, St. Louis,
MO 63110, USA.
Mller, A.R.
School of Behavioral and Brain Sciences, University of Texas at Dallas, GR

41, P.O. Box 830688, Richardson, TX 75083-0688, USA.
Neuloh, G.
Neurochirurgische Universitatsklinik, D-53105 Bonn, Germany.
Niimi, Y.
Center for Endovascular Surgery, Institute for Neurology and Neurosurgery,

Roosevelt Hospital, Suite GG15, 1000 Tenth Avenue, New York, NY 10019,
USA.
Nuwer, M.R.
Department of Neurology and Clinical Neurophysiology, UCLA School of

Medicine, Reed Research Building, Room 1-190, 710 Westwood Plaza,
Packwood, J.W.
Department of Clinical Neurophysiology, UCLA School of Medicine, Reed

Research Building, Room 1-194, 710 Westwood Plaza, Los Angeles, CA
90095, USA. Present address: 2718 Yandall Drive, Austin, TX 78748, USA.
Quinonez, D.

Rodi, Z.
Division of Neurology, Institute of Clinical Neurophysiology, University

Medical Centre, Zaloska 7, SI-1525 Ljubljana, Slovenia.
Rosen, I.
Department of Clinical Neurophysiology, University Hospital, SE-22185

Lund, Sweden.
Rosow, C.
Department of Anesthesia and Critical Care, Harvard Medical School, 55

Fruit Street, Boston, MA 02114, USA.
Royter, V.

Rubinstein, E.H.
Department of Anesthesiology, University of California, Los Angeles,

CA 90095-1778, USA.
Sabet, A.
Department of Neurology, Gold Coast Hospital, Southport, Queensland,

Australia.
Sala, F.
Department of Neurological Sciences and Vision, Section of Neurosurgery,

University Hospital, 37100 Verona, Italy.
Schrader, L.M.
David Geffen School of Medicine, UCLA, 710 Westwood Plaza, Room

1-194, Los Angeles, CA 90095, USA. Present address: University of
Connecticut School of Medicine, 20 Old Stone Crossing, Hartford, CT
06117, USA.
Schramm, J.
Neurochirurgische Universitatsklinik, D-53105 Bonn, Germany.
Sclabassi, R.J.
xv
Department of Neurological Surgery, School of Medicine, University of

Pittsburgh Medical Center, Suite B-400, 200 Lothrop Street, Pittsburgh,
PA 15213, USA.
Shin, A.Y.
Mayo Clinic, Brachial Plexus Clinic, Rochester, MN 55905, USA.
Slimp, J.C.
Neuromonitoring Program, Department of Rehabilitation Medicine, Box

356490, University of Washington School of Medicine, Seattle, WA
98195, USA.
Sloan, T.B.
Department of Anesthesiology, University of Colorado HSC, 4200 E. 9th

Avenue, Campus Box B113, Denver, CO 80262, USA.
Spinner, R.J.
Mayo Clinic, Brachial Plexus Clinic, Gonda 8S, 200 First Street SW,
Stead, M.

Stern, J.M.
Department of Neurology, Geffen School of Medicine, UCLA, 710

Westwood Plaza, Suite 1250, Los Angeles, CA 90095, USA.
Strauss, C.
Neurochirurgische Universitatsklinik, University

Schwabachanlage 6, D-91054 Erlangen, Germany.
Strommen, J.A.
Department of Physical Medicine and Rehabilitation, Mayo Clinic College

of Medicine, 200 First Street SW, Rochester, MN 55905, USA.
Van Huffelen, A.C.
Department of Clinical Neurophysiology (F.02.230), University Medical

Center Utrecht, P.O. Box 85500, 3508 GA Utrecht, The Netherlands.
Vespa, P.M.
Department of Neurosurgery and Neurology, David Geffen School of

Medicine, CHS 18-218, UCLA Medical Center, Los Angeles, CA 90095,
USA.
Vincent, F.
Division of Neurosurgery and Spinal Program, Krembil Neuroscience

Center, University of Toronto, Toronto, Ont. M5T 2S8, Canada.
Vodusek, D.B.
Division of Neurology, Institute of Clinical Neurophysiology, University

Medical Centre, Zaloska 7, SI-1525 Ljubljana, Slovenia.
Wang, H.
Department of Hand Surgery, Huashan Hospital, Fudan University,

Shanghai, Peoples Rep. of China, and Neurologic Surgery, Mayo Clinic,
Guggenheim 15-21B, 200 First Street SW, Rochester, MN 55905, USA.
Worrell, G.A.

Yingling, C.D.
Stanford School of Medicine and Surgical Monitoring Services, 1001

Bridgeway #434, Sausalito, CA 94965, USA.
York, D.
Department of Neuroscience, St. Johns Mercy Medical Center, St. Louis,

MO 63141, USA.
Erlangen-Nuremberg,
Contents
Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
List of Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Section I. Overview and General Considerations
1. Overview and history
M.R. Nuwer (Los Angeles, CA, USA) . . . . . . . . . . . . . . .
2. Physiology
J.R. Daube (Rochester, MN, USA) . . . . . . . . . . . . . . . . .
3. Anatomy
J.R. Daube (Rochester, MN, USA) . . . . . . . . . . . . . . . . .
4. EEG and anesthetic effects
V. Jantti and T.B. Sloan (Tampere, Finland and Denver, CO, USA) .
5. Anesthetic effects on evoked potentials
T.B. Sloan and V. Jantti (Denver, CO, USA and Tampere, Finland) .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . . 44
. . . . . . . . . . . . . . 77
. . . . . . . . . . . . . . 94
Section II. CNP Techniques Used in Monitoring Neural Function in Surgery

6. Electroencephalography used in monitoring neural function during surgery
A.C. Van Huffelen (Utrecht, The Netherlands) . . . . . . . . . . . .
7. Electrocorticography
G.A. Worrell, M. Stead and G.D. Cascino (Rochester, MN, USA) . . .
8. Direct cortical stimulation to localize sensory, motor and language function
L.M. Schrader (Los Angeles, CA, USA) . . . . . . . . . . . . . . .
9. Intraoperative approaches for deep brain stimulation and targeting
A.S. Mandir and R. Minahan (Washington, DC, USA) . . . . . . . .
10. Visual evoked potentials during surgery
D. York (St. Louis, MO, USA) . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . 128
. . . . . . . . . . . . 141
. . . . . . . . . . . . 150
. . . . . . . . . . . . 163
. . . . . . . . . . . . 172
xviii
CONTENTS
Section II. 1. Somatosensory Evoked Potentials

11. Somatosensory evoked potential monitoring with scalp and cervical recording
M.R. Nuwer and J.W. Packwood (Los Angeles, CA, USA) . . . . . . . . . . . .
12. Somatosensory evoked potential monitoring with dermatomal stimulation
J.C. Slimp (Seattle, WA, USA) . . . . . . . . . . . . . . . . . . . . . . . . . .
13. Monitoring spinal epidural potentials to peripheral nerve stimulation
S. Jones (London, UK) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14. Somatosensory evoked potentials for intraoperative mapping of the sensorimotor cortex
T. Kombos (Berlin, Germany) . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . 180
. . . . . . 190
. . . . . . 202
. . . . . . 211
Section II. 2. Motor Evoked Potentials

15. Motor EP physiology, risks and specific anesthetic effects
H.L. Journee (Groningen, The Netherlands) . . . . . . . . . . . . . . . . . .
16. Corticospinal tract monitoring with D- and I-waves from the spinal cord and muscle
MEPs from limb muscles
V. Deletis and F. Sala (New York, NY, USA and Verona, Italy) . . . . . . .
17. Recording MEPs to transcranial electrical stimulation and SEPs to peripheral nerve
stimulation simultaneously from the spinal cord
D. Burke (Sydney, Australia) . . . . . . . . . . . . . . . . . . . . . . . . .
18. Transcranial electrical MEP with muscle recording
A. Mendiratta and R.G. Emerson (New York, NY, USA) . . . . . . . . . . .
19. Neurogenic mixed evoked potentials
R.E. Minahan and A.S. Mandir (Washington, DC, USA) . . . . . . . . . . .
20. Magnetic cortical stimulation techniques
L.D. Gugino, L.S. Aglio, H.L. Edmonds and A.A. Gonzalez
(Boston, MA, Louisville, KY and Los Angeles, CA, USA) . . . . . . . . . .
21. Mapping the corticospinal tract
C.D. Yingling (San Francisco, CA, USA) . . . . . . . . . . . . . . . . . . .
. . . . . . . . 218
. . . . . . . . 235
. . . . . . . . 252
. . . . . . . . 260
. . . . . . . . 273
. . . . . . . . 282
. . . . . . . . 319
Section II. 3. Brainstem and Auditory Evoked Potentials

22. BAEPs in surgery
A.D. Legatt (Bronx, NY, USA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
23. Mapping the brainstem: floor of the fourth ventricle
J.R. Lopez (Stanford, CA, USA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
Section II. 4. Electromyographic, Reflex and Nerve Conduction Monitoring
24. Intraoperative peripheral nerve stimulation and recording
B.A. Crum and J.A. Strommen (Rochester, MN, USA) . . . . . . . . . . . .
25. Intraoperative facial nerve monitoring
Y.A. Ashram and C.D. Yingling (Alexandria, Egypt and Stanford, CA, USA) .
26. Oculomotor and lower cranial nerve monitoring
J.R. Lopez (Stanford, CA, USA) . . . . . . . . . . . . . . . . . . . . . . .
27. Intraoperative monitoring with free-running EMG
J.A. Strommen and B.A. Crum (Rochester, MN, USA) . . . . . . . . . . . .
. . . . . . . . 364
. . . . . . . . 371
. . . . . . . . 384
. . . . . . . . 396
CONTENTS
xix
28. Intraoperative EMG during spinal pedicle screw instrumentation

J.R. Balzer, D. Crammond, M. Habeych and R.J. Sclabassi (Pittsburgh, PA, USA)
29. Sacral roots and nerves, and monitoring for neuro-urologic procedures
D.B. Vodusek and V. Deletis (Ljubljana, Slovenia and New York, NY, USA) . . .
30. Motor evoked potentials from external anal and urethral sphincter muscles
by transcranial cortical stimulation during surgery
S.S. Haghighi (San Diego, CA, USA) . . . . . . . . . . . . . . . . . . . . . . .
31. Selective dorsal rhizotomy
C.D. Yingling (San Francisco, CA, USA) . . . . . . . . . . . . . . . . . . . . .
32. Nerve root assessment with SEP and MEP
B.A. Crum and J.A. Strommen (Rochester, MN, USA) . . . . . . . . . . . . . .
33. Technological advances in intraoperative neurophysiological monitoring
R.J. Sclabassi, J. Balzer, D. Crammond and M. Habeych (Pittsburgh, PA, USA). .
. . . . . . 404
. . . . . . 423
. . . . . . 434
. . . . . . 439
. . . . . . 455
. . . . . . 464
Section III. Disorders Monitored during Surgery

Section III. 1. Cerebral Neurosurgery
34. Neurophysiology during epilepsy surgery
J.M. Stern and J. Engel, Jr. (Los Angeles, CA, USA) . . . . . . . . . . . . . . . . . .
35. Intraoperative neurophysiology during surgery for cerebral tumors
H. Duffau (Montpellier, France) . . . . . . . . . . . . . . . . . . . . . . . . . . . .
36. Movement disorders
J.W. Cozzens (Evanston, IL, USA) . . . . . . . . . . . . . . . . . . . . . . . . . . .
37. Mapping and monitoring for brainstem lesions
G. Neuloh, C. Strauss and J. Schramm (Bonn, Germany) . . . . . . . . . . . . . . . .
38. Cranial base surgery
A.F. Cannestra, R.F. Canalis, E.H. Rubinstein and D.P. Becker (Los Angeles, CA, USA)
39. Neurophysiology during microvascular facial nerve decompression
A.R. Mller (Dallas, TX, USA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
40. Neurophysiology during middle ear, mastoid and parotid surgery
T.C. Huang and A.K. Lalwani (New York, NY, USA) . . . . . . . . . . . . . . . . .
41. Monitoring during surgery around the acoustic and vestibular nerves
C. Matthies (Wurzburg, Germany) . . . . . . . . . . . . . . . . . . . . . . . . . . .
42. Monitoring neural function during surgery around the glossopharyngeal, vagus and
laryngeal nerves during neck (thyroid, larynx, carotid) and chest procedures
T.M. Hemmerling (Montreal, Canada) . . . . . . . . . . . . . . . . . . . . . . . . .
. . . 484
. . . 491
. . . 508
. . . 522
. . . 534
. . . 547
. . . 556
. . . 566
. . . 590
Section III. 2. Spine Surgery

43. Monitoring during the surgical treatment of scoliosis
M.R. Nuwer and J.W. Packwood (Los Angeles, CA, USA) . . . . . . .
44. Monitoring during spinal surgery for fractures and extramedullary tumors
F. Vincent, M.O. Kelleher and M.G. Fehlings (Toronto, Canada) . . . .
45. Intraoperative neurophysiological monitoring during surgery for intramedullary
F. Sala and K.F. Kothbauer (Verona, Italy and Lucerne, Switzerland) . .
46. Neurophysiological monitoring during spinal endovascular procedures
Y. Niimi, V. Deletis and A. Berenstein (New York, NY, USA) . . . . .
. . . . . . . . . . . 608
. . . . . . . . . . . 618
spinal cord tumors
. . . . . . . . . . . 632
. . . . . . . . . . . 651
xx
47. Monitoring the spinal cord during surgery for cervical myelopathy
S. Jones (London, UK) . . . . . . . . . . . . . . . . . . .
48. Monitoring during lumbar stenosis and fusion surgery
B. Ball and P.M. Huddleston (Rochester, MN, USA) . . . .
49. Surgery for tethered cord syndrome and other cauda equina lesions
A.M. Husain (Durham, NC, USA) . . . . . . . . . . . . .
50. Dorsal root entry zone procedures and other surgeries for pain
A.M. Husain (Durham, NC, USA) . . . . . . . . . . . . .
CONTENTS
. . . . . . . . . . . . . . . . . 671
. . . . . . . . . . . . . . . . . 678
. . . . . . . . . . . . . . . . . 689
. . . . . . . . . . . . . . . . . 700
Section III. 3. Peripheral Nerve Surgery

51. Intraoperative testing and monitoring during brachial plexus surgery
H. Wang, A.T. Bishop, A.Y. Shin and R.J. Spinner (Shanghai, China and
Rochester, MN, USA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
52. Intraoperative testing and monitoring during lumbosacral plexus surgery
R.E. Minahan and A.S. Mandir (Washington, DC, USA) . . . . . . . . . . . . . . . .
53. Monitoring of neural structures involved in micturition and sexual function
Z. Rodi and D.B. Vodusek (Ljubljana, Slovenia) . . . . . . . . . . . . . . . . . . . .
54. Monitoring during total hip arthroplasty
D. Fee and A. Sabet (Lexington, KY, USA and Southport, Australia) . . . . . . . . . .
55. Monitoring neural function during pelvic surgery
B.R. Moed (St. Louis, MO, USA). . . . . . . . . . . . . . . . . . . . . . . . . . . .
56. Intraoperative testing and monitoring during peripheral nerve surgery
H. Wang and R.J. Spinner (Shanghai, Peoples Rep. of China and Rochester, MN, USA)
. . . 720
. . . 731
. . . 739
. . . 748
. . . 752
. . . 764
Section III. 4. Vascular Surgery

57. Intraoperative monitoring during carotid endarterectomy
J.-M. Guerit (Brussels, Belgium) . . . . . . . . . . . . . . . . . . . . . . . . .
58. Monitoring during carotid balloon test occlusion
D. Eliashiv, V. Royter and D. Quinonez (Los Angeles, CA, USA). . . . . . . . .
59. Evoked potential monitoring during surgery for intracranial aneurysms
G. Neuloh and J. Schramm (Bonn, Germany) . . . . . . . . . . . . . . . . . . .
60. Spinal cord monitoring during descending aortic procedures
D.B. MacDonald and C.C.J. Dong (Riyadh, Saudi Arabia and Vancouver, Canada)
61. Intraoperative monitoring during cardiac surgery
J.-M. Guerit (Brussels, Belgium) . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . 776
. . . . . . 791
. . . . . . 801
. . . . . . 815
. . . . . . 829
Section IV. Intensive Care Monitoring

62. Neonatal ICU monitoring
L. Hellstrom-Westas, L.S. De Vries and I. Rosen (Lund, Sweden
and Utrecht, The Netherlands) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 840
63. ICU EEG monitoring for acute seizures and status epilepticus
P.M. Vespa (Los Angeles, CA, USA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 856
64. ICU EEG monitoring for vasospasm and other focal cortical disorders
J. Claassen and L.J. Hirsch (New York, NY, USA) . . . . . . . . . . . . . . . . . . . . . . 864
CONTENTS
xxi
Section V. Other Issues

65. Safety issues during surgical monitoring
D.B. MacDonald and V. Deletis (Riyadh, Saudi Arabia and New York, NY, USA)
66. Monitoring the depth of anesthesia
G.A. Mashour and C. Rosow (Boston, MA, USA). . . . . . . . . . . . . . . . .
67. Monitoring of cerebral perfusion with transcranial Doppler ultrasound
H.L. Edmonds (Louisville, KY, USA) . . . . . . . . . . . . . . . . . . . . . .
68. Cerebral oximetry as a tool in the operating room and intensive care unit
J.R. Balzer, D. Crammond, M. Habeych and R.J. Sclabassi (Pittsburgh, PA, USA)
. . . . . . 882
. . . . . . 899
. . . . . . 909
. . . . . . 924
Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 939
SECTION I
Overview and General Considerations

Handbook of Clinical Neurophysiology, Vol. 8
M.R. Nuwer (Ed.)
# 2008 Elsevier B.V. All rights reserved
CHAPTER 1
Overview and history

Marc R. Nuwer*
Department of Clinical Neurophysiology, UCLA School of Medicine, Los Angeles, CA 90095, USA
Over the past 30 years, neurophysiologic intraoperative monitoring (IOM) has grown from an interesting
investigational procedure into a widely used method
to protect patients from neurologic injury during
surgery.
IOM techniques include most neurophysiologic
modalities commonly used among outpatients. These
include electroencephalography (EEG), electromyography (EMG), evoked potentials (EPs), and nerve
conduction velocity (NCV) testing of various types.
IOM also includes some techniques not used in outpatients, such as transcranial electrical motor EPs
(tceMEPs). Most techniques are electrical. Some
other modalities, though, also have found some use
in surgery. The latter include oximetry and transcranial Doppler (TCD).
IOM helps in a number of ways. Most obviously,
it can warn the surgeon of a serious complication in
time to intervene and correct the problem before it
becomes permanent. Second, it sometimes identifies
a serious systemic problem that needs to be corrected. Third, the surgeon can feel comfortable about
the patients neurologic safety to that point in the
case, and therefore go forward to provide a more
thorough procedure. Fourth, with IOM the surgeons
can feel more confident about a procedures safety,
allowing surgery on a high-risk patient who might
otherwise be turned away. Fifth, the patient and his
or her family can take comfort that the very real neurologic risks of surgery are lessened by IOM.
IOM is not a perfect procedure. False positive cases
(false alarms) occur in a portion of cases. In scoliosis,
that rate is around 1% of procedures. In some other
types of procedures, the rate is higher. These may be
*
Correspondence to: Marc R. Nuwer, M.D., Ph.D., Department of Clinical Neurophysiology, UCLA School of
Medicine, Reed Research Building, Room 1-194,
710 Westwood Plaza, Los Angeles, CA 90095, USA.
Tel.: 1-310-206-3093; fax: 1-310-267-1157.
E-mail: mrn@ucla.edu (M.R. Nuwer).
due to problems with the technique itself, the difficulty

of obtaining good quality tracings from some patients,
or as a result of anesthetic changes. True positive cases
(true predictions of postoperative deficits) also occur.
Those are cases where IOM raises an alarm, any available interventions are accomplished, but the patient has
neurologic injury anyway. Just because an alarm is
raised does not necessarily allow for prompt, complete
correction of the problem. False negative cases are
those in which the patient suffers from a neurologic
injury that was not predicted by IOM changes. False
negative cases are rare, but do occur. Some are due to
immediate postoperative deterioration. Other cases are
a result of injury in pathways not monitored. Occasionally, they are due to errors by the IOM team, who failed
to recognize changes when they occurred.
1.1. History of monitoring
Early intraoperative use of neurophysiology dates
back to epilepsy surgery during the first half of the
twentieth century. Penfield (Penfield and Boldrey,
1937) used direct cortical stimulation to define the
homunculus of human motor and sensory cortex. Soon
thereafter, electrocorticography (ECoG) was used to
identify regions of epileptic discharges, slowing, or
lack of fast activity (Jasper, 1949; Marshall and
Walker, 1949). These initial recordings were from
the surface of the exposed cortex, and later acute depth
electrode recordings were added. This set of techniques has evolved in its details, but remains a valuable
clinical tool in the surgical treatment of patients with
medical refractory epilepsy even today.
The next clinical advances were not until several
decades later when routine EEG was carried out in
carotid endarterectomy (CEA) (Thompson, 1968;
Wylie and Ehrenfeld, 1970; Sharbrough et al., 1973)
In the early days of CEA, many patients were kept
awake during carotid clamping because of worries
about cortical ischemia during clamping. Better alternative ways were sought to evaluate for ischemia.
OVERVIEW AND GENERAL CONSIDERATIONS
EEG provided an excellent substitute. Studies at Mayo

Clinic (Sundt et al., 1974) showed how EEG changed
at degrees of ischemia that could be tolerated for a
while in surgery, leading to widespread acceptance of
these monitoring techniques. Competing techniques
to measure ischemia included stump pressure measurement, which could only be performed at a few
specific times not continuously. Some surgeons
preferred not to monitor at all. Overall, EEG developed in the 1970s into a commonly used, although
not universally accepted, technique to safeguard
patients during CEA.
Spinal cord monitoring grew out of research
investigations in the early 1970s. Japanese investigators reported monitoring the electrospinogram,
directly recorded spinal potentials from the epidural
space after direct spinal stimulation (Shimoji et al.,
1971; Imai, 1976). Given the neurologic risk of spine
surgery, these research techniques were evaluated
and validated as a method to monitor spinal cases
(Tamaki et al., 1972, 1981). This school of monitoring
used direct spinal epidural stimulation and recording
sites.
Somatosensory EPs (SEPs) from the scalp initially
were investigated in the mid-1970s. These were middle- and long-latency 50200-ms cortical potentials.
Nash and colleagues (Nash et al., 1974, 1977; Nash
and Brodkey, 1977) initially applied these SEPs in
the operating room, but were impeded by variability
of signals and sensitivity to anesthesia. Grundy
(1982) published a series of reports about anesthetic
effects, described techniques to reduce adverse effects,
and extended the techniques into neurosurgical procedures. This technique, as used by others in the earliest
stages of SEP cortical potential monitoring, used filters at 1100 Hz with middle- and long-latency potentials (e.g., Engler et al., 1978; Speilholz et al., 1979).
Problems remained for excess noise or irreproducible
background variability. By the late 1970s, Nuwer and
Dawson (1984) evaluated the causes of variability
and determined that use of short-latency SEP techniques, restricted filters, and other technical modifications substantially reduced background variability
and greatly improved reliability in the SEP tracings.
With these technical improvements, SEP became a
widely adopted method of spinal cord monitoring.
These techniques used ankle stimulation with scalp
and neck recordings.
A somewhat different approach was developed in
the UK, where Jones (Jones et al., 1982) used the spinal recordings but moved the stimulator to the
posterior tibial nerve. This avoided any concerns about

the safety of repeated spinal cord epidural electrical
stimulation. It still required placement of epidural
recording electrodes even in orthopedic cases.
For two decades, monitoring the sensory pathways
with SEPs and measuring the epidural spinal potentials were the techniques of choice for spinal cord
monitoring. Spurred by occasionally reported cases
of false negative monitoring, several investigators
sought ways to monitor directly from the corticospinal pathways. By the mid-1980s, magnetic cortical
stimulation techniques were tried as a method for
evoking corticospinal track potentials, which were
known to successfully stimulate cortex in awake outpatients. These were found difficult to conduct under
anesthesia. Burke subsequently popularized the use
of transcranial electrical stimulation as a practical corticospinal technique for use under anesthesia (Hicks
et al., 1991; Burke et al., 1992). Many monitoring teams
now use this technique to measure corticospinal pathways in the operating room. Controversy continues,
however, about where to measure the responses. Some
groups (e.g., Deletis) recommend measuring the
D-wave electrospinogram from the caudal spinal cord or
cauda equina, whereas others measure motor responses
from muscles.
Meanwhile, in the 1980s, methods were developed
and refined for auditory EPs and EMG during surgery around cranial nerves. These were applied to
monitor during posterior fossa surgery (Mller and
Mller, 1985), such as microvascular decompression
and acoustic neuroma resection.
In the early 1980s, the field consisted of a set of specific techniques applied by particular groups for specific surgeons. Commercial IOM equipment was not
available until 1981. Before that, clinical neurophysiologists had to adapt other equipment for their needs.
Growth of a more fully developed field took several
phases. The first general IOM service set up at UCLA
in 1979 offered a variety of techniques to any surgeon
in various surgical disciplines. In the early 1980s,
annual academic meetings of clinical neurophysiologists and certain surgical groups included reports about
IOM. By the mid-1980s, special IOM meetings were
offered widely to clinical neurophysiologists, surgeons, and technologists. Those meetings carried the
message of IOM to a much wider audience. The mid1980s also saw the first two IOM textbooks (Nuwer,
1986; Mller, 1988) appear, providing more detail to
a wider audience of general users. By the late 1980s,
IOM became an established technique in wide use.
These initial IOM services included EEG, ECoG,

EPs, EMG, and NCV techniques. This first wave
of IOM was based mostly on techniques familiar to
clinical neurophysiologists. Over the next decade, a
second wave of additional IOM techniques became
popular. This second wave included specific monitoring modalities such as motor EPs and pedicle screw
testing. It also included advances in the monitoring
equipment such as remote supervision and increase
channel number.
1.2. Technologist staffing and supervision
Two fundamentally different types of procedures are
used in surgery. They differ not by their techniques
but by their goals. IOM aims to watch over neurologic
pathways and to identify any signs of injury. Intraoperative testing aims to identify neurologic structures.
Monitoring occurs over many hours, whereas testing is
usually done at one or several discrete times. Despite
the differences between these two different concepts,
the two are often lumped together and referred to as
IOM.
Monitoring requires sufficient expertise by the
team and good communication with the surgeon.
Details depend upon the particular techniques, especially on whether the aim is testing or monitoring.
Testing to identify a particular neurological structure usually requires personal involvement of a clinical neurophysiologist. This is the case for ECoG, or
for localization of language or motor cortex. These
require professional judgment about what tissue to
resect, a level of skill greater than that of a technologist alone. The clinical neurophysiologist is in the
room to make decisions and discuss interpretations
and recommendations personally with the surgeon.
This is referred to as personal supervision.
Routine monitoring of the brain or spinal cord typically does not require the continuous personal inroom involvement of a clinical neurophysiologist
(Nuwer and Nuwer, 1997). This is the case for monitoring during scoliosis correction or clipping of an
aneurysm. A technologist often conducts ordinary
monitoring in the surgical suite, supervised by a clinical neurophysiologist who is nearby or online. The
degree of involvement is less than that for testing,
but the availability to intervene remains. This is
referred to as direct supervision. Modern internet
technology allows for remote online real-time monitoring by an expert clinical neurophysiologist who
is elsewhere in the building or even off-site.
M.R. NUWER
Some clinical neurophysiologists provide direct

supervision to more than one case simultaneously. In
such a circumstance, the supervising clinical neurophysiologist needs to be ready to become involved
intensely in any one case if needed. In order to pay
great attention to one case when problems occur, there
needs to be a plan for covering the other cases. Usually
only two or three cases reasonably could be supervised
simultaneously while still giving sufficient professional attention to supervising any one case.
A technologist generally lacks suitable skills,
knowledge, abilities, training, and experience to provide the IOM services without supervision by a clinical neurophysiologist. Generally, technologists also
lack the medical knowledge to advise the surgeon
about clinical options when changes do occur. Many
local authorities and institutions require physician
supervision of medical procedures including IOM.
In some institutions, advanced non-physician clinical
neurophysiologists provide supervision services. In
any case, the monitoring supervisor needs to have
staff privileges at the hospital where the monitoring
is conducted. The privileging should meet local public policies to review and approve each individuals
suitability to provide these clinical services.
Technologists are key members of the IOM team.
They need to have sufficient experience in neurophysiologic testing, techniques, basic sciences, and relevant clinical sciences before being left in an
operating room for IOM. Usually, this includes several years of experience conducting EEG and EPs
in regular inpatients and outpatients. That provides
a basis in experience for knowing what signals look
like, what various changes and abnormalities occur
among patients, and how to deal with technical problems. Even with such experience outside the
operating room, any technologist should be proctored
and subject to progressively lessened supervision
when introduced to IOM. The technologists supervision and privileging should be specific to individual
types of procedures. For example, a technologist
who knows how to monitor EEG during a CEA does
not necessarily know how to monitor somatosensory
EPs in the same case. It depends instead on the individuals own skills, knowledge, abilities, training,
and experience.
1.3. Training for physicians
Training of young physicians to conduct monitoring
needs to include formal basic and clinical learning
and experience. This should include training about

techniques as applied in the operating room as well
as training about how to apply those techniques in a
variety of surgical procedures. In general, this forms
a part of the training of young physicians in clinical
neurophysiology.
After training, the fellow should understand
basic sciences of anatomy, physiology, and phar-
macology of pathways monitored in surgery

various neurophysiological techniques used in sur-
gical monitoring and testing

how to use IOM to predict and prevent adverse
neurologic outcomes
how to use IOM to locate and identify neurologic
structures during surgery

effects of anesthesia, systemic disorders, and co-
morbidities
how to train and supervise technologists to conduct
The fellow also should have some additional experience in other occasionally used IOM techniques and
clinical situations, and preferably in other types of
neurophysiologic monitoring in the critical care unit,
epilepsy unit, and during radiological procedures.
Overall, the field of intraoperative neurophysiology has developed from disparate difficult techniques
in its earlier days three decades ago. Now, it has
become a discipline with a large number of techniques, applied to a variety of surgical and other procedures. The monitoring teams encompass many types
of specialists contributing their own expertise. The
goals remain the same as before: to enhance patient
care, avoid neurological deficits, allow for more
complete procedures, allow procedures even on some
high-risk patients, and provide feedback to the surgeon about actions that could injure the nervous
system.
monitoring
normal variations and criteria for abnormality and
References
alarms
The fellow should have extensive clinical experience in the use of common IOM techniques
including

EEG monitoring
ECoG
somatosensory EP spinal cord monitoring
somatosensory EP motor cortex localization
motor EP spinal cord monitoring
brainstem auditory EP monitoring
EMG cranial nerve monitoring
EMG monitoring of trunk and limbs
pedicle screw stimulation
NCV peripheral testing
deep brain stimulator placement
monitoring from remote sites
The fellow should have extensive clinical experience

with common clinical uses of IOM including

cerebral aneurysm clipping

cerebral tumor resection
epilepsy surgery
deep brain stimulator implantation
cranial base and posterior fossa tumor resection
scoliosis correction
cervical myelopathy decompression
peripheral nerve repair or decompression
CEA
aortic and cardiac surgery
Burke, D, Hicks, R, Stephen, J, Woodforth, I and Crawford, M (1992) Assessment of corticospinal and somatosensory conduction simultaneously during scoliosis
surgery. Electroencephalogr. Clin. Neurophysiol., 85:
388396.
Engler, GL, Speilholz, NI, Bernhard, WN, Danziger, F,
Merkin, H and Wolf, T (1978) Somatosensory evoked
potentials during Harrington instrumentation for scoliosis. J. Bone Joint Surg. (Am), 60: 528532.
Grundy, BL (1982) Monitoring of sensory evoked potentials during neurosurgical operations: methods and
applications. Neurosurgery, 11: 556575.
Hicks, RG, Burke, DJ and Stephen, JP (1991) Monitoring spinal cord function during scoliosis surgery with CotrelDubousset instrumentation. Med. J. Aust., 154: 8286.
Imai, T (1976) Human electrospinogram evoked by direct
stimulation on the spinal cord through epidural space. J.
Japanese Ortho. Assoc., 50: 10371056.
Jasper, HH (1949) Electrocorticograms in man. Electroencephalogr. Clin. Neurophysiol., Suppl. 2: 1629.
Jones, SJ, Edgar, MA and Ransford, AO (1982)
Sensory nerve conduction in the human spinal cord: epidural recordings made during scoliosis surgery. J.
Neurol. Neurosurg. Psychiatr., 45: 446451.
Marshall, C and Walker, AE (1949) Electrocorticography.
Bull. Johns Hopkins Hosp., 85: 344359.
Mller, AR (1988) Evoked Potentials in Intraoperative Monitoring, Williams & Wilkins, Baltimore, 224 pp.
Mller, MB and Mller, AR (1985) Loss of auditory function in microvascular decompression for hemifacial
spasm. Results in 143 consecutive cases. J. Neurosurg.,
63: 1720.
6
Nash, CL, Jr. and Brodkey, JS (1977) Clinical Application of
Spinal Cord Monitoring for Operative Treatment
of Spinal Disease. Case Western Reserve University,
Cleveland, 140 pp.
Nash, CL, Jr., Schatzinger, L and Lorig, R (1974) Intraopertive monitoring of spinal cord function during scoliosis
spine surgery. J. Bone Joint Surg. (Am), 56: 1765.
Nash, CL, Jr., Lorig, RA, Schatzinger, LA and Brown, RH
(1977) Spinal cord monitoring during operative treatment
of the spine. Clin. Orthop., 126: 100105.
Nuwer, MR (Ed.) (1986) Evoked Potential Monitoring in
the Operating Room. Raven Press, New York, 246 pp.
Nuwer, MR and Dawson, EC (1984) Intraoperative evoked
potential monitoring of the spinal cord: enhanced stability of cortical recordings. Electroencephalogr. Clin.
Neurophysiol., 59: 318327.
Nuwer, JM and Nuwer, MR (1997) Neurophysiologic surgical monitoring staffing patterns in the USA. Electroencephalogr. Clin. Neurophysiol., 103: 616620.
Penfield, W and Boldrey, E (1937) Somatic motor and sensory representation in the cerebral cortex of man as studied by electrical stimulation. Brain, 37: 389443.
Sharbrough, FW, Messick, JM, Jr. and Sundt, TM, Jr.
(1973) Correlation of continuous electroencephalograms with cerebral blood flow measurements during
carotid endarterectomy. Stroke, 4: 674683.
M.R. NUWER
Shimoji, K, Higashi, H and Kano, T (1971) Epidural
recording of spinal electrogram in man. Electroencephalogr. Clin. Neurophysiol., 30: 236239.
Speilholz, NI, Benjamin, MV, Engler, GL and Ransohoff, J
(1979) Somatosensory evoked potentials during decompression and stabilization of the spine: methods and
findings. Spine, 4: 500505.
Sundt, TM, Jr., Sharbrough, FW, Anderson, RE and Michenfelder, JD (1974) Cerebral blood flow measurements and
electroencephalograms during carotid endarterectomy. J.
Neurosurg., 41: 310320.
Tamaki, T, Yamashita, T, Kobayashi, H and Hirayama, H
(1972) Spinal cord monitoring. J. Jpn. Electroencephalogr. Electromyogr., 1: 196.
Tamaki, T, Tsuji, H, Inoue, S and Kobayashi, H (1981)
The prevention of iatrogenic spinal cord injury utilizing the evoked spinal cord potential. Int. Orthop., 4:
313317.
Thompson, JE (1968) Surgery for Cerebrovascular Insufficiency (Stroke) with Special Emphasis on Carotid
Endarterectomy. Charles C Thomas, Springfield, IL,
96 pp.
Wylie, EJ and Ehrenfeld, WK (1970) Extracranial
OcclusiveCerebrovascular Disease: Diagnosis and
Management. W.B. Saunders Company, Philadelphia,
231 p.

M.R. Nuwer (Ed.)
CHAPTER 2
Physiology
Jasper R. Daube*
Department of Neurology, Mayo Clinic College of Medicine, Rochester, MN 55905, USA
2.1. Overview
Knowing the location and function of the structural
components of the nervous system permits localization
of the site of a lesion. The temporal profile of the major
types of disease assists in identifying the cause of the
disorder. However, one temporal profile has not yet
been considered, that of the transient or rapidly reversible abnormality. Many diseases that produce signs or
symptoms of brief duration may not produce destructive changes in cells and may occur without demonstrable histologic abnormality of the involved
structures. To understand transient manifestations of
disease, it is necessary to understand the mechanism
by which the cells of the nervous system process information and to understand their physiology. Transient
alterations in the physiology of the cells cause transient symptoms and signs. This chapter provides an
introduction to the physiology of neurons, axons, and
muscle fibers, which is the basis for information transmission in the central and peripheral neural structures
and for the transient symptoms and signs that accompany disease states. The major function of the nervous
system is the transmission, storage, and processing of
information. This function is accomplished by the generation, conduction, and integration of electrical activity and by the synthesis and release of chemical agents.
Information is conducted from one region to another as
electrical activity, commonly known as nerve
impulses, which are generated by neuronal cell bodies
or axons and conducted by axons. Information is
transmitted between cells by neurochemical agents
that convey the signals from one cell to the net.
Information is integrated by the interaction of electrical activity in single cells and in groups of cells.
*
Correspondence to: Jasper R. Daube, M.D., E8B, Department of Neurology, Mayo Clinic, 200 First Street SW,
Tel.: 1-507-284-4409; fax: 1-507-284-4074.
E-mail: daube.jasper@mayo.edu (J.R. Daube).
Although this chapter discusses only the physiology

of single cells, it must be remembered that the activity of the central and peripheral nervous systems
never depends on the activity of a single neuron or
axon but is always mediated by a group of cells or
nerve fibers. Information is represented in the nervous system by a change in activity in a group of
cells or fibers as they respond to some change in
their input. The interactions of neurons in large
groups are considered in later sections.
Normal function in a single neuron as it participates
in the processing of information is manifested as electrical potentials. These potentials are called membrane
potentials. The membrane potential is the difference
in electrical potential between the inside and the outside of a cell. All neurons, axons, and muscle fibers
have a membrane potential. Membrane potentials
include resting potentials, action potentials, and local
potentials such as synaptic potentials, generator (or
receptor) potentials, and electrotonic potentials. All
membrane potentials result from ion flow through
channels in the membrane. Cell membranes separate
ions into different concentrations on the exterior and
interior of the cell. These concentration differences
produce an electrical potential across the membrane,
the membrane potential. The concentration gradients
are maintained by the cell membrane, a lipid bilayer
that is relatively impermeable to sodium, potassium,
chloride, and calcium ions, the ions involved in electrophysiologic activity and signal transmission. The
concentration of sodium, calcium, and chloride ions
is higher extracellularly and that of potassium ions
and impermeable anions (A) is higher intracellularly
(Table 1). The equilibrium potential of each ion is the
voltage difference across the membrane that exactly
offsets the tendency of the ion to move down its concentration gradient. Ions can move across the cell
membrane passively through ion channels or by adenosine triphosphate (ATP)-dependent binding to carrier
molecules. Some ion channels are open at rest, but
J.R. DAUBE
Table 1
Relative ionic concentrations in mammalian neurons
Internal concentration
External concentration
Resting permeability
Sodium
Potassium
Chloride
Calcium
Low
High
Low
High
Low
High
Low
High
Moderate
High
Low
Low
most open (or close) in response to stimuli, including

changes in membrane potential (voltage-gated ion
channels), binding receptor (ligand-gated channels),
or chemical changes in the cytoplasm (chemical-gated
ion channels). The opening of a channel for a particular
ion brings the membrane potential toward the equilibrium potential of that ion. Thus, at a given time, the
membrane potential is determined by the concentration gradient of the ions (which determines their
respective equilibrium potentials) and by any changes
in the permeability to individual ions across the membrane (Fig. 1). The survival and excitability of the cell
depend on the membrane potential. Maintaining this
potential requires energy metabolism for the ATPdependent sodium/potassium pump.
The resting potential is the baseline level of the
membrane potential when the cell is at rest and not
processing information. This potential depends primarily on the potassium channel. When a cell is
active in the processing of in formation, the membrane potential varies. These variations are either
local potentials or action potentials (Table 2).
Action potentials are the electrical signals, or
nerve impulses, by which information is conducted
from one area to another within a single cell. The
action potential is an all or none change in membrane
potential in the body or axon of a neuron or within a
muscle fiber. It either occurs fully or not at all and
depends on the sodium channel. The function of
action potentials is to conduct bits of information
from one place to another. The action potential is
initiated by one form of local potential, the electrotonic potential. Local potentials are localized
changes in a number of ion channels that change
the membrane potential in response to stimuli. They
are graded signals whose size varies in proportion
to the size of the stimulus. They remain localized in
the area of the cell in which they are generated; that
Fig. 1. Variables that determine the membrane potential. Transmembrane ion gradients determine the equilibrium potential
of a particular ion. The transmembrane gradients depend on the activity of adenosine triphosphate (ATP)-driven ion pumps
and the buffering effects of the astrocytes on extracellular fluid composition. Membrane permeability to a particular ion
depends on the opening of specific ion channels. This opening can be triggered by voltage (voltage-gated channels), neurotransmitters (ligand-gated channels), or intracellular chemicals such as calcium, ATP, or cyclic nucleotides (chemically
gated channels). Increased membrane permeability to a given ion (the opening of the ion channel) brings the membrane
potential toward the equilibrium potential of this ion.
Table 2
Characteristics of different membrane potentials
Characteristic
Graded and localized

All-or-none spread
Active membrane
channel
Initiated by
Local potentials
Action potential
Generator potential
Synaptic potential
Electrotonic potential

Na, K

Na, Ca2, K, Cl

None
Sensory stimulus
Neurotransmitter
Generator, synaptic,
or action potentials
is, they do not spread to involve the entire cell. Local

potentials can be summated and integrated by single
cells and are an integral part of the processing of
information by the nervous system. Changes in ion
channels underlying local potentials are generated by
neurochemical transmitters
synaptic and generator potentials or
the flow of electrical current, the electrotonic
potentials.
Generator, or receptor, potentials occur in receptors
those neural structures in the body, such as the touch
receptors in the skin or the light receptors in the eye
that respond to specific stimuli. Receptor potentials
are also local potentials caused by opening of ion
channels and are localized and graded. They can generate electrotonic potentials and thereby initiate
action potentials (Fig. 2). A localized change in
membrane potential results in current flow to surrounding areas of membrane. This current flow produces a small change in the membrane potential of
Na, K,
sometimes Ca2
Electronic potential
adjacent areas. This change is called an electrotonic

potential. Synaptic potentials are variations of the
membrane potential that occur at synapses, the
specialized areas where adjacent neurons are in intimate contact. Synaptic potentials are local potentials
arising from postsynaptic ion channels opening in
response to the action of a neurotransmitter released
by presynaptic cells. Neurotransmitters transmit
information from one cell to another by converting
the electrical signal (action potential) into a chemical
signal (neurotransmitter release) and then back into
an electrical signal (synaptic potential or membrane
potential change). In turn, synaptic potentials produce electrotonic potentials, which can then initiate
an other action potential. Chemical synapses are the
most common form of communication between neurons. They consist of a presynaptic axon terminal and
a postsynaptic element, which can be a dendrite, cell
body, or axon of the target neuron. The presynaptic
terminal contains the chemical transmitter, which is
stored in synaptic vesicles. The arrival of an action
Fig. 2. Local potentials and triggering of the action potential. Three types of local potentials are (1) receptor (or generator)
potential, triggered by the action of a sensory stimulus on a sensory receptor; (2) synaptic potential, triggered by the action
of a neurotransmitter; and (3) electrotonic potential, which consists of the passive movement of charges according to the
cable properties of a membrane. Both the generator and synaptic potentials give rise to electrotonic potentials, which depolarize the membrane to threshold for triggering an action potential. The action potential is a regenerating depolarizing stimulus that, via electrotonic potentials, propagates over a distance without decrement in its amplitude.
10
potential produces an influx of calcium ions into the

presynaptic terminal, and this triggers the vesicular
release of the chemical transmitter, an example of
the process known as exocytosis. Neurochemical
transmitters act on different types of postsynaptic
receptors to produce two different types of responses:
(1) the classic responses (referred to as classic
neurotransmission) are the postsynaptic potentials
described above and (2) neurochemicals may also
produce changes in the excitability and responses of
the postsynaptic neuron to other neurotransmitters, a
process called neuromodulation.
Transient alterations in function are the result of
reversible disturbances in neuronal excitability, the
ability to propagate action potentials, or communication via chemical synapses. Transient disorders reflect
abnormalities in resting, local, or action potentials
due to the failure of ion pumps to maintain electrochemical gradients, impaired function of ion channels, or alterations in the ionic composition of the
extracellular fluid. Transient disorders may be
generalized or focal and be manifested by excessive
activity, decreased activity, or both.
2.2. Cell membrane
2.2.1. Transmembrane ion gradients
The plasma membrane is a lipid bilayer with the
polar (hydrophilic) heads facing outward and the
nonpolar (hydrophobic) tails extending to the middle
of the bilayer. Embedded in this lipid bilayer are protein macromolecules, including ion channels, receptors, and ionic pumps, that are in contact with both
the extracellular fluid and the cytoplasm. The lipid
bilayer is relatively impermeable to water soluble
molecules, including ions such as sodium (Na),
potassium (K), chloride (C1), and calcium (Ca2).
These ions are involved in electrophysiologic activity
and signal transmission. The concentrations of sodium,
chloride, and calcium are higher extracellularly, and the
concentrations of potassium and impermeable anions
(A) are higher intracellularly (Fig. 3). Maintenance
of transmembrane ion concentration depends on the
balance between
passive diffusion of ions across ion channels, or
pores, of the membrane, driven by their concentration gradient and

active, energy (ATP)-dependent transport of ions
against their concentration gradient, via ATPdriven ion pumps.
J.R. DAUBE
Fig. 3. Transmembrane ion concentrations, equilibrium

potential, and resting membrane potential. The semipermeable cell membrane determines a differential distribution of ions in the intracellular and extracellular
compartments. Sodium and chloride predominate extracellularly, and potassium and nondiffusible ions (A)
predominate intracellularly. Transmembrane ion composition is maintained by the activity of ATP-dependent
pumps, particularly sodiumpotassium adenosine triphosphatase (ATPase). The different transmembrane concentrations of diffusible ions determine the equilibrium
potential of each ion (Eion). The contribution of each
ion to the membrane potential depends on the permeability of the membrane for that particular ion. Increased
permeability to an ion brings the membrane potential
toward the equilibrium potential of that ion. At rest, the
membrane is predominantly, but not exclusively, permeable to potassium. There is continuous leakage of potassium out of the cell and of sodium into the cell, driven by
both their concentration gradient and electrical gradient.
The ion gradient is restituted by the activity of sodium
potassium ATPase, which is electrogenic (as it
exchanges two potassium for three sodium ions) and
contributes to the maintenance of the resting potential.
In the central nervous system, astrocytes provide a

buffer system to prevent excessive accumulation of
extracellular potassium ions.
2.2.2. Active transport
Nerve and muscle cells obtain energy from glucose
and oxygen via the glycolytic pathways, the Krebs
cycle, and the electron transport system. These pathways provide the energy for normal cell function in
the form of ATP. ATP is partly consumed in generating the resting potential by a mechanism in the membrane, which moves potassium in and sodium out of
the cell, with slightly more sodium being moved than
potassium. This movement is referred to as active

transport, and the system through which it occurs is
the sodium pump. The sodium pump moves sodium
out of the cell and potassium into the cell against
their concentration gradients. Chloride moves out of
the cell passively with sodium.
2.2.3. Equilibrium potential
The diffusible ions (sodium, potassium, and chloride,
but not calcium) tend to move across the cell membrane according to their concentration gradient. The
molecular motion of ions is a source of energy known
as the diffusion pressure. For example, the intracellular concentration of potassium is 30 times higher than
the extracellular concentration, [K ]o; therefore,
potassium tends to diffuse from intracellular to extracellular fluid. The opposite occurs with sodium. As
ions diffuse across the cell membrane, a separation
of charges develops because the nondiffusible negatively charged intracellular ions (principally proteins) have a charge opposite that of the diffusible
ions. Two regions that accumulate different charges
have an electrical potential difference. The voltage
that develops as a diffusible ion moves across the
membrane produces an electrical pressure that
opposes the movement of the ion. The net ionic
movement continues until the electrical pressure
equals the diffusion pressure. At this time, the system
is in equilibrium. At equilibrium, random ionic
movement continues, but no net movement of ions
occurs. The electrical potential that develops across
the membrane at equilibrium is called the equilibrium
potential, and this potential is different for each ion.
The equilibrium potential of an ion (Eion) is the voltage difference across the membrane that exactly offsets the diffusion pressure of an ion to move down
its concentration gradient. Therefore, the equilibrium
potential is proportional to the difference between the
concentration of the ion in the extracellular fluid and
the concentration in the intracellular fluid. An algebraic representation of the equilibrium potential can
be derived because the physical determinants of the
diffusion pressure and electrical pressure expressed
are known. The final equation is the Nernst equation.
Electrical pressure is defined by
W e Em Z i F
in which We electrical pressure (work required to
move an ion against a voltage), Em absolute
11
membrane potential, Zi valence (number of

charges on the ion), and F faraday (number of coulombs per mol of ion).
Diffusion pressure is defined by
W d R T (ln Chi ln C1o
in which Wd diffusion pressure (work required to
move an ion against a concentration gradient), R
universal gas constant, T absolute temperature,
ln natural logarithm, [C]hi ion concentration on
the more concentrated side of the membrane, and
[C]lo ion concentration on the less concentrated side.
At equilibrium,
We Wd
and therefore

RT
ln Chi
Em
:
Zi F ln Clo
By rearrangement, the equilibrium potential is

RT
ln Chi

:
Em
F Zi
ln Clo
The Nernst equation is an important relationship that
defines the equilibrium potential, Em, inside the cell
for any ion in terms of its concentration on the two
sides of a membrane. From the Nernst equation, the
polarity of the potential depends on whether the ion
is an anion or a cation: Em for a cation will be ()
on the low concentration side, and Em for an anion
will be () on the low concentration side. By substituting for the constants at room temperature, converting to a base 10 logarithm, and converting to
millivolts, we get a useful form of the equation
Em 58log10 Chi for cations
For example,
Ena 58log0 140 43:3
Em 58log10 Chi for anions
We may use these equations to calculate the equilibrium
potential for any ion if we know the concentrations of
that ion on the two sides of the membrane. This potential only develops if the membrane is permeable to the
ion. The approximate neuronal equilibrium potentials
of the major ions are K 100 mV, Na
40 mV, Cl 75 mV, and Ca2 124 mV
(Fig. 3).
12
2.2.4. Membrane potential

The contribution of a given ion to the actual voltage
developed across the membrane with unequal concentrations of that ion depends not only on its concentration gradient but also on the permeability (P) of the
membrane to that ion. Permeability is the ease with
which an ion diffuses across the membrane and is a
reflection of the probability that the membrane channel that conducts the ion will open. For example, an
ion with a high concentration gradient that has very
low permeability (e.g., calcium) does not contribute
to the resting membrane potential. If a membrane is
permeable to multiple ions that are present in differing
concentrations on either side of the membrane, the
resultant membrane potential is a function of the concentrations of each of the ions and of their relative
permeabilities. The Goldman equation combines
these factors for the major ions that influence the
membrane potential in nerve and muscle cells. Such
calculations, on the basis of the actual ionic concentrations and ionic permeabilities, agree with measurements of these values in living cells. These equations
also show that a change in either ionic permeability or
ionic concentrations can alter membrane potential. If
the concentration gradient of an ion is reduced, there
will be a lower equilibrium potential for that ion. If
the resting membrane potential is determined by the
equilibrium potential of that ion, the resting potential
will decrease. In contrast, if the permeability for an
ion is increased by opening of channels for that ion,
the membrane potential will approach the equilibrium
potential of that ion, and it may increase or decrease,
depending on whether the membrane potential is
above or below the equilibrium potential. The movements of ions that occur with normal cellular activity
are not sufficient to produce significant concentration
changes; therefore, membrane potential fluctuations
are normally due to permeability changes caused by
channel opening and closing. Increased permeability
(i.e., opening of the channel) to a particular ion brings
the membrane potential toward the equilibrium potential of that ion. In an electrical model of the membrane, the concentration ratios of the different ions
are represented by their respective equilibrium potentials (ENa, EK, ECl); their ionic permeabilities are
represented by their respective conductances. The
conductance (the reciprocal of the resistance) for a
particular ion is the sum of the conductances of all
the open channels permeable to that ion. The movement of ions across the membrane is expressed as an
J.R. DAUBE
ion current. By Ohms law, this current depends on

two factors: the conductance of the ion and the driving
force for the ion. The driving force is the difference
between the membrane potential and the equilibrium
potential of that ion.
2.2.5. Ion channels
Ion channels are intrinsic membrane proteins that
form hydrophilic pores (aqueous pathways) through
the lipid bilayer membrane. They allow the passive
flow of selected ions across the membrane on the basis
of the electrochemical gradients of the ion and the
physical properties of the ion channel. Most channels
belong to one of several superfamilies of homologous
proteins with great heterogeneity in amino acid composition. They are defined on the basis of their ion
selectivity, conductance, gating, kinetics, and pharmacology. In general, the transmembrane portion of the
protein forms the pore, and the specific amino acids
in the region of the pore determine ion selectivity, conductance, and voltage sensitivity of the channel.
Amino acids in the extracellular or intracellular portion (or both) of the channel protein determine the gating mechanism and the kinetics of inactivation. Ion
channels vary in their selectivity; some are permeable
to cations (sodium, potassium, and calcium) and
others to anions (primarily chloride). The open state
predominates in the resting membrane for a few channels; these are mostly the potassium channels responsible for the resting membrane potential (see below).
Most ion channels are gated; that is, they open
in response to specific stimuli. According to their
gating stimuli, ion channels can be subdivided into
(1) voltage-gated channels, which respond to changes
in membrane potential; (2) ligand-gated channels,
which respond to the binding of a neurotransmitter to
the channel molecular complex; and (3) chemically
gated channels, which respond to intracellular molecules such as ATP, ions (particularly calcium), and
cyclic nucleotides. Important examples of chemically
gated channels include the cyclic nucleotide-gated
channels found in many sensory receptors (e.g., photoreceptors in the retina). Mechanoreceptors are activated
by mechanical distortion of the cell membrane and are
sometimes referred to as stretch-activated channels.
Gating stimuli may interact in some channels. For
example, the ion permeability of some ligand-gated
channels is affected by membrane voltage or intracellular factors (or both). Voltage-gated channels are
critical for neuronal function. They control excitability,
spontaneous neuronal activity, generation and conduction of action potentials, and neurotransmitter release.
Sensitivity to voltage is due to a voltage sensor at the
pore. A region of the pore acts as a selectivity filter,
which regulates ion permeability according to the size
and molecular structure of the ion. The range of voltage for activation and the rate of activation (opening)
and inactivation (closing) are important variables in
voltage-gated channels. Voltage-gated cation channels
are responsible for the maintenance of neuronal excitability, generation of action potentials, and neurotransmitter release (Table 3). They are members of a
superfamily of proteins with a common basic structure
consisting of a principal subunit and one or
more auxiliary subunits. The amino acid composition
of the subunit determines ion selectivity, voltage sensitivity, and inactivation kinetics of the channel. Voltage-gated sodium channels are critical for the
generation and transmission of information in the nervous system by action potentials. In neurons, sodium
channels are concentrated in the initial segment of
the axon (the site of generation of action potentials)
and in the nodes of Ranvier (involved in rapid conduction of action potentials). In muscle, these channels
participate in excitationcontraction coupling. There
are several varieties of voltage-gated calcium channels, and they have different distributions, physiology,
13
pharmacology, and functions (Table 4). Calcium

influx occurs not only through voltage-gated channels,
but also through ligand-gated and cyclic nucleotidegated channels. Calcium ions are important in the regulation of numerous processes in neurons, including
modulation of neuronal firing pattern, neurotransmitter release, signal transduction, enzyme activation,
intracellular transport, intermediate metabolism, and
gene expression. Intracellular calcium is also necessary for muscle contraction and glandular secretion.
These functions depend on levels of calcium in the
cytosol that are determined by the calcium influx
through various channels, release from intracellular
stores (particularly the endoplasmic reticulum), and
counter balancing active mechanisms of reuptake
and extrusion. Large numbers of voltage-gated potassium channels determine much of the pattern of activity generated by neurons. They are responsible
primarily for the resting membrane potential, repolarization of the action potential, and control of the probability of generation of repetitive action potentials.
Ligand-gated channels open in response to the binding
of neurotransmitters (Fig. 4). They include (1) nonselective cation channels permeable to sodium, potassium, and, in some cases, calcium and (2) anion
channels permeable to chloride. These channels are
discussed in relation to synaptic transmission.
Table 3
Examples of ion channels
Ion channel
Voltage-gated
Na
K
Equilibrium potential
Location
Function
35
Node of Ranvier
Initiation and conduction of

action potential
90
Axon hillock
Diffuse along internode
Diffuse in neurons
Ca2
Chemical-gated Cl
(GABA)
Cation channel (L-glutamate, acetylcholine)
GABA, g-aminobutyric acid.
200
75
Dendrite
Soma
Axon terminal
Dendrite
Soma
Dendrite
Repolarization of action
potential
Decrease neuronal
excitability and discharge
Slow depolarization
Burst firing
Oscillatory firing
Neurotransmitter release
Synaptic inhibition
Synaptic excitation
14
J.R. DAUBE
Table 4
Voltage-gated calcium channels
Channel type
Location
Function
T
L
Apical dendrites
Soma, dendritic shafts, and spines
Skeletal muscle
Synaptic terminals
Synaptic terminals
Dendritic shafts and spines
Rhythmic firing
Slow action potentials
Excitationcontraction coupling
Persistent depolarization
N
P/Q
2.2.6. Neuronal excitability

Neuronal excitability is defined as the ability of the
neuron to generate and transmit action potentials. It
depends on the membrane potential, which determines the gating of the sodium channels. The membrane potential depends on the transmembrane ion
concentration (which determines the equilibrium
potential) and ion permeability. Increased permeability to an ion moves the membrane potential toward
the equilibrium potential of that ion. In the absence
of a stimulus, the membrane potential of the neuron,
or resting membrane potential, is dominated by its
permeability to potassium, whose channels are open;
therefore, this potential varies between 60 and
80 mV. Because the threshold for opening voltage-
gated sodium channels that are needed to trigger

and propagate action potentials is 5055 mV, any
change of the membrane potential in this direction
will increase the probability of triggering an action
potential. An increase in membrane permeability to
sodium or calcium increases excitability, and an
increase in permeability to potassium or chloride will
decrease excitability (Table 5).
2.3. Resting potential
The resting potential is the absolute difference in
electrical potential between the inside and the outside
of an inactive neuron, axon, or muscle fiber. If an
electrical connection is made between the inside
and the outside of a neuron, the cell acts as a battery
Fig. 4. The plasma membrane consists of a phospholipid bilayer that provides a barrier to the passage of water-soluble
molecules, including ions. Passage of ions across the membrane depends on the presence of transmembrane proteins,
including ion channels and ion pumps. Ion channels provide an aqueous pore for the passage of ions across the membrane,
according to their concentration gradients. The opening of an ion channel, or pore, may be triggered, or gated, by several
stimuli, such as voltage (voltage-gated channel) or neurotransmitters (ligand-gated channel). In the example shown here,
a neurotransmitter (such as glutamate) binds to a specific ligand-gated cation channel, and this produces a change in the
spatial configuration of the channel protein, allowing the pore to open and the cation to pass through the membrane.
Changes in the amino acid composition of the ion channel protein affects its ion selectivity, gating mechanism, and kinetics
of channel opening (activation) and closing (inactivation).
15
Table 5
Ionic basis of local potentials
Ion
Equilibrium potential (mV)
Na
40
Ca2
K
Cl
200
90
75
Effect of increased permeability

on membrane potential
Examples of local potentials
Depolarization
Generator potentials
Excitatory postsynaptic potential
Excitatory postsynaptic potential
Inhibitory postsynaptic potential
Inhibitory postsynaptic potential
Depolarization
Hyperpolarization
Hyperpolarization, depolarization,
or no change
and an electrical current will flow. The potential is

generally between 60 and 80 mV, with the inside of
the cell negative with respect to the outside. The resting potential can be measured directly by using a
microelectrode. The tip of such an electrode must
be less than 1 mm in diameter to be inserted into a
nerve or muscle cell. By connecting the microelectrode to an appropriate amplifier, the membrane
potential can be recorded and displayed on an oscilloscope (Fig. 5). The oscilloscope registers the potential difference between the two electrical inputs and
is displayed as a vertical deflection of a spot of light
that moves continuously from left to right across the
cathode ray tube of the oscilloscope. A negative
membrane potential is registered as a downward
deflection; thus, when a microelectrode enters a neuron or muscle fiber, the oscilloscope beam moves
down to a new position. The resting membrane
potential is the transmembrane voltage at which there

is no net flow of current across the membrane. Its
value determines spontaneous neuronal activity and
neuronal activity in response to extrinsic input.
Because the resting potential is the absolute difference in potential between the inside and the outside
of the cell, it represents transmembrane polarity.
A decrease in the value of the resting membrane
potential means less negativity inside the cell and
the membrane potential moves toward zero; this constitutes depolarization. When the membrane potential
becomes more negative than the value of the resting
potential, the potential moves away from zero; this
is hyperpolarization. The resting membrane potential
depends on two factors: (1) the presence of leak ion
channels open at rest with markedly different permeabilities to sodium and potassium, making the cell
membrane a semipermeable membrane and (2) the
Fig. 5. Oscilloscopic recording of a membrane potential from a neuron.
16
presence of energy-dependent pumps, particularly the

sodium/potassium pump.
At rest, there is a continuous leak of potassium
outward and of sodium inward across the membrane.
Cells at rest have a permeability to sodium ions that
ranges from 1% to 10% of their permeability to
potassium. Thus, in the absence of synaptic activity,
the membrane potential is dominated by its high permeability to potassium, and the membrane potential
is drawn toward the equilibrium potential of this
ion (100 mV). However, the membrane at rest is also
permeable to sodium and chloride, so that the membrane potential is also pulled toward the equilibrium
potential of these ions. The resting potential varies
among different types of neurons, but it is typically
6080 mV. The continuous leaking of potassium outward and sodium inward is balanced by the activity
of the sodium/potassium pump.
2.3.1. Steady state
Potassium diffuses through the membrane most readily because potassium channels are more open and
potassium conductance is much higher than that of
other ions. Therefore, potassium is the largest source
of separation of positive and negative charges (voltage) as it diffuses out and leaves the large anions
behind. This is illustrated schematically in Fig. 6.
Small amounts of sodium entering the cells, driven
by both electrical and chemical forces, tend to depolarize the membrane. As a result, potassium is no longer in equilibrium and leaves the cell. Thus, the cell
is not in equilibrium but in a steady state dependent
on metabolic energy. In this steady state, the small
outward potassium leak must be exactly equal in
magnitude to the rate at which potassium is transported into the cell. The same is true also for sodium.
In this condition, the net movement of each ion
across the membrane is zero, an exact description
of the resting membrane potential.
2.3.2. Sodium pump
The sodium pump (Na/K adenosine triphosphatase
[ATPase]) maintains the intracellular concentrations of
sodium and potassium despite their constant leaking
through the membrane. The sodium/potassium pump
transports three sodium ions out of the cell for every
two potassium ions carried into the cell. Because the
pump is not electrically neutral, it contributes directly
to the resting potential; that is, it is electrogenic. The
J.R. DAUBE
contribution of the sodium/potassium pump steady state

to the resting potential is 11 mV. The cell membrane at
rest is permeable also to chloride ions. In most membranes, chloride reaches an equilibrium simply by
adjustment of its internal concentration to maintain electroneutrality, without affecting the steady-state membrane potential.
2.3.3. Role of extracellular calcium
The external surface of the cell membrane contains a
high density of negative charges because of the presence of glycoprotein residues. This produces a local
negative potential that contributes to the resting membrane potential. Divalent ions, such as extracellular calcium, alter the transmembrane potential by neutralizing
this local, negative surface potential. Neutralizing the
surface potential increases the contribution of the
transmembrane potential to the resting potential, and
this increases the threshold for opening voltage-gated
sodium channels. This explains the stabilizing effect
of extracellular calcium on membrane excitability
and the presence of increased spontaneous activity
(tetany) that occurs in patients with hypocalcemia or
alkalosis.
2.3.4. Role of glial cells
Astrocytes are important in controlling the extracellular
concentration of potassium. Astrocytes are highly permeable to potassium and are interconnected with each
other by gap junctions. When the extracellular concentration of potassium increases because of neuronal
activity, astrocytes incorporate potassium and transfer
it from one cell to another through gap junctions. This
prevents the extracellular accumulation of potassium
and maintains neuronal excitability. This is referred to
as spatial buffering of extracellular potassium.
2.4. Local potentials
In a normal nerve cell or muscle cell with adequate
sources of oxygen and glucose, the resting potential
is maintained at a stable, relatively unchanging level.
However, the resting potential readily changes in
response to stimuli. The membrane potential can
change from the resting state in only two ways. It
can either become more negative inside, hyperpolarization, or less negative inside, depolarization. Even if
the membrane potential reverses, so that the inside
becomes positive with respect to the outside, it is still
17
Fig. 6. A theoretical model of the generation of a membrane potential by diffusion across a semipermeable membrane.
Equal amounts of anions and cations are dissolved on each side of the membrane; no voltage gradient. The membrane is
permeable to all ions except large anions (A). B: K, Na, and Cl redistribute themselves solely by diffusion; this results
in a charge separation, with greater negativity inside. C: Electrical pressure due to charge separation and diffusion pressure
due to concentration differences are balanced at the resting membrane potential.
referred to as depolarization, because the potential is

less negative than the resting potential. The changes
in the membrane potential that occur with anoxia or
a change in the concentration of the ions on either
side of the membrane are relatively long lasting
(minutes to hours). In contrast, rapid changes (seconds or less) can occur in response to electrical,
mechanical, or chemical stimuli. These changes
occur as a result of current flow through the membrane. Transient currents in living tissues are due to
the movement of charged ions and can flow through
the membrane as a result of an applied voltage or
of a change in membrane conductance. A local
potential is a transient depolarizing or hyperpolarizing shift of the membrane potential in a localized

area of the cell. Local potentials result from the current flow due to localized change in ion channels that
alter permeability to one or more ions. Ion channel
opening or closing may result from
a chemical agent acting on the channel, a synaptic
potential
activation of a sensory receptor channel by a stimu-
lus, a receptor potential

a current from an eternally applied voltage, an
electrotonic potential
18
J.R. DAUBE
Synaptic potentials are the response to information

carried by a neurotransmitter released by an adjacent
neuron. Receptor potentials are the response to external stimuli. Electrotonic potentials participate in the
transfer of information throughout a cell by action
potentials.
membrane, the negative pole is commonly referred to

as the cathode; the positive pole is the anode. The current flow at the cathode depolarizes, whereas that at the
anode hyperpolarizes a membrane.
2.4.1. Ionic basis
All local potentials have certain characteristics in

common (Table 2). Importantly, the local potential
is a graded potential; that is, its amplitude is proportional to the size of the stimulus (Fig. 7). Measurement of a local potential uses the resting potential
as its base line. If the membranes resting potential is
depolarized from 80 to 70 mV during the local potential, the local potential has an amplitude of 10 mV.
This potential change is one of decreasing negativity
(or of depolarization), but it could also be one of
Local potentials result from the flow of current

through the membrane with a change in channels that
are open or closed in response to a chemical agent,
mechanical deformation, or an applied voltage. Neurotransmitters and neuromodulators produce synaptic
potentials by one of six mechanisms:
(1) Opening of potassium channels increases potassium conductance, resulting in hyperpolarization,
a relatively slow process.
(2) Opening of sodium channels increases sodium
conductance, resulting in depolarization, a relatively fast process.
(3) Opening of both potassium and sodium channels
increases the conductance of both ions, resulting
in a depolarization but to a lesser degree than in
item 2, above.
(4) Opening of chloride channels increases chloride
conductance, resulting in rapid stabilization or
hyperpolarization of the membrane voltage.
(5) Closing of potassium channels, resulting in a slow
depolarization.
(6) Opening of calcium channels, resulting in a slow
depolarization.
Generator potentials occur primarily by opening of
both sodium and potassium channels and increasing
conductance of both ions. This produces depolarization. Generator potentials also occur in response to
specific molecules that activate olfactory receptors
and to photic stimuli that activate photoreceptors in
the retina of the eye. Electrotonic potentials occur
in one of two ways. (1) Opening of sodium channels
by a current arising from a voltage in an adjacent
area of membrane. This produces depolarization.
(2) Opening or closing of several different ion channels by an eternally applied voltage.
Application of a negative voltage to the outside of
the membrane causes outward current flow and depolarization of the membrane. Application of a positive
potential to the outside of the membrane causes inward
current flow and hyperpolarization of the membrane.
When a voltage is applied to the outside of the axonal
2.4.2. Characteristics of local potentials
Fig. 7. Local potentials. These potentials are shown as an

upward deflection if they are a depolarization and as a
downward deflection if they are a hyperpolarization. Resting potential is 70 mV. At time zero, electrical currents of
varied polarities and voltage are applied to the membrane
(bottom). A is an anodal current; B, C, and D are cathodal
currents. A produces a transient hyperpolarization; B, C,
and D produce a transient depolarization that is graded
and proportional to the size of the stimulus. All of these
are local potentials. D produces an action potential E.
increasing negativity (or of hyperpolarization).

Because the local potential is a graded response proportional to the size of the stimulus, the occurrence
of a second stimulus before the first one subsides
results in a larger local potential. Therefore, local
potentials can be summated. They are summated
algebraically, so that similar potentials are additive
and hyperpolarizing and depolarizing potentials tend
to cancel out each other. Summated potentials may
reach threshold and produce an action potential when
single potentials individually are subthreshold. When
a stimulus is applied in a localized area of the membrane, the change in membrane potential has both a
temporal and a spatial distribution. A study of the
temporal course of the local potential (Fig. 7) shows
that the increase in the potential is not instantaneous
but develops over a few milliseconds. After the stimulus ends, the potential subsides over a few milliseconds as well. Therefore, local potentials have a
temporal course that outlasts the stimulus. The occurrence of a second stimulus at the same site shortly
19
after the first produces another local potential, which

summates with any residual of the earlier one that
has not yet subsided (Fig. 8). This summation of
local potentials occurring near each other in time is
called temporal summation (Fig. 8B). Different synaptic potentials have different time courses. Most
synaptic potentials range from 10 to 15 ms in duration; however, some are very brief, lasting less than
1 ms, but others may last several seconds or several
minutes. The longer the duration of the synaptic
potential, the greater the chance for temporal summation to occur. By means of temporal summation, the
cell can integrate signals that arrive at different
times.
Study of the spatial distribution of local potentials
reveals another of their characteristics. As their name
implies, they remain localized in the region where
the stimulus is applied; they do not spread throughout
the entire cell. However, the locally applied stimulus,
because of local current flow, has an effect on the
nearby membrane. The potential change is not
Fig. 8. Summation of local potentials in a neuron. A: Spatial summation occurs when increasing numbers of nerve terminals release more neurotransmitter to produce larger excitatory postsynaptic potentials (EPSPs). B: Temporal summation
occurs when a single terminal discharges repetitively more rapidly to produce larger EPSPs.
20
sharply confined to the area of the stimulus but falls

off over a finite distance along the membrane, usually a few millimeters. The application of a simultaneous second stimulus near the first (but not at the
same site) results in summation of the potentials in
the border zones spatial summation (Fig. 8). Thus,
the membrane of the cell can act as an integrator of
stimuli that arrive from different sources and impinge
on areas of membrane near one another. Spatial and
temporal summation are important mechanisms in
the processing of information by single neurons;
when summated local potentials reach threshold, they
initiate an action potential. If a current or voltage is
applied to a membrane for more than a few milliseconds, the ion channels revert to their resting state,
changing ionic conductances of the membrane in a
direction to restore the resting potential to baseline
value. This phenomenon is known as accommodation
(Fig. 9). Therefore, if an electrical stimulus is
increased slowly, accommodation can occur and no
change will be seen in the membrane potential. The
changes in conductance during accommodation
require several milliseconds, both to develop and to
subside. As a result, if an electrical stimulus is
applied gradually so that accommodation prevents a
change in resting potential, then when the stimulus
is suddenly turned off, the residual change in conductance will produce a transient change in resting
potential. Thus, accommodation can result in a cell
responding to the cessation of a stimulus.
2.5. Action potential
Action potentials have several advantages for the rapid
transfer of information in the nervous system. Because
action potentials are all-or-none, they either occur or
do not occur and they can transfer information without
loss over relatively long distances. Their all-or-none
feature also allows coding of information as frequency
rather than the less stable measure of amplitude. Also,
Fig. 9. Accommodation of the membrane potential to

applied stimulus of constant strength. Note the response
to sudden cessation of the stimulus.
J.R. DAUBE
their threshold eliminates the effects of small, random

changes in membrane potential.
2.5.1. Threshold
The membranes of neurons, axons, and muscle cells
have another characteristic that is basic to their ability to transmit information from one area to another
their excitability. If a membrane is depolarized by
a stimulus, there is a point at which suddenly many
sodium channels open. This point is known as the
threshold for excitation (Fig. 7). If the depolarization
does not reach threshold, the evoked activity is a local
potential. Threshold may be reached by a single local
potential or by summated local potentials. When
threshold is reached, there is a sudden increase in the
membranes permeability to sodium. This change in
conductance results in the action potential.
2.5.2. Ionic basis of action potential
In the resting state, many more potassium channels are
open, the conductance of sodium is much less than that
of potassium, and the resting potential is near the equilibrium potential of potassium. At threshold, many
sodium channels open so that the conductance of
sodium suddenly becomes greater than that of potassium, and the membrane potential shifts toward the
equilibrium potential of sodium, which is 60 mV.
This depolarization reverses the polarity of the membrane, the inside becoming positive with respect to
the outside. With the opening of the sodium channels
and increased sodium conductance, there is a flow of
current with the inward movement of sodium ions.
The change in sodium conductance is usually transient,
lasting only a few milliseconds, and is followed by
opening of potassium channels, an increase in the
potassium conductance, and an outward movement of
potassium ions. These three changes overlap, and the
potential of the membrane during these changes is a
function of the ratios of the conductances (Fig. 10).
Sodium conductance increases by several thousand
folds early in the process, whereas potassium conductance increases less, does so later, and persists longer.
The conductance changes for these two ions result in
ionic shifts and current flows that are associated with
a membrane potential change the action potential
(Fig. 10). The action potential is a sudden, short
duration, all-or-none change in the membrane potential that occurs if the membrane potential reaches
threshold (Table 6). Its components are shown in
21
Fig. 10. Conductance changes during action potential. A: Temporal sequence at a single site along an axon. Changes in
conductances (permeabilities) of sodium and potassium are plotted against time as they change with associated changes in
membrane potential. Note that sodium conductance changes by several thousand folds early in the process, whereas potassium conductance changes only about 30-fold during later stages and persists longer than sodium conductance changes.
B: Spatial distribution of an action potential over a length of axon at a single instant.
Table 6
Comparison of local potentials and action potentials
Characteristic
Local potentials
Action potential
Example
Generator
Synaptic
Electrotonic
5100
0.110
Local changes in permeability
to Na, K, Ca2, or Cl
No
Yes
No
Passive and decremental
Nerve impulse
Duration (ms)
Amplitude (mV)
Ionic mechanism
Threshold
Spatial and temporal summation
Refractory period
Propagation
Fig. 11. The initial portion of the membrane potential

change is the local potential. At threshold, the rising
phase of the action potential suddenly changes
because of the influx of positive ions. In most nerve
cells and skeletal muscle cells, the inward current
during the rising phase of the action potential is carried by sodium ions, because sodium conductance is
110
70110
Transient increase permeability to Na,
followed by increased permeability to K
Yes
No (all-or-none)
Yes
Active and nondecremental
markedly increased. The action potential also could

be carried by calcium ions if the calcium conductance increased sufficiently, as occurs in some dendrites. Repolarization begins as sodium conductance
decreases or potassium conductance increases (or
both). The decreased flow of sodium ions is followed
by an efflux of potassium ions. The rate of return of
22
J.R. DAUBE
Fig. 11. Component of an action potential with a resting potential of 70 mV. (A) Local electrotonic potential. (B) Threshold
level. (C) Spike. (D) Negative (depolarizing) afterpotential. (E) Positive (hyperpolarizing) afterpotential.
the membrane potential to the baseline slows after

sodium conductance has returned to baseline, producing a small residual on the negative component of the
action potential, which is called the negative afterpotential. In some myelinated axons, repolarization
occurs by a decrease in sodium conductance with
no change in potassium conductance. The afterpotential is positive when the membrane potential is
recorded with a microelectrode within the cell, but
it is called negative because it is negative when
recorded with an extracellular electrode. The increase
in potassium conductance persists and results in a
hyperpolarization after the spike component of the
action potential, the after-hyperpolarization. The
after-hyperpolarization is due to continued efflux of
potassium ions, with a greater than resting difference
in potential between the inside and outside of the
cell. The after-hyperpolarization is positive when
measured with extracellular electrodes and therefore
is called a positive afterpotential. During the positive
afterpotential, the membrane potential is near the
potassium equilibrium potential, and oxygen consumption is increased with increased activity of the
sodium pump. The amounts of sodium and potassium
that move across the membrane during the action
potential are small, buffered by surrounding astrocytes, and do not change the concentration enough
to result in a change in the resting potential. In addition, the sodium that moves in during the action
potential is continually removed by the sodium pump
during the relatively long intervals between action

potentials.
2.5.3. Excitability
The excitability of a membrane is the ease with
which an action potential can be generated and is
usually measured in terms of the voltage required to
initiate an action potential. During increased sodium
conductance, the membrane cannot be stimulated to
discharge again. A second stimulus at this time is
without effect; therefore, action potentials, unlike
local potentials, cannot summate. This period of
unresponsiveness is the absolute refractory period
(Fig. 12). As sodium conductance returns to normal,
the membrane again becomes excitable, but for a
short period, the relative refractory period, it requires
a larger stimulus to produce a smaller action potential. After the relative refractory period, while the
negative afterpotential is subsiding, the membrane
is partially depolarized, is closer to threshold, and
has an increased excitability. This period is the
supernormal period. Finally, during the positive
afterpotential, the membrane is hyperpolarized, and
stronger stimuli are required. This period is the subnormal period.
Up to now, the term threshold has been used to
refer to the membrane potential at which sodium
channels open and an action potential is generated.
The threshold of a membrane remains relatively
23
Fig. 12. Excitability changes during an action potential. The lower portion of the illustration shows the ease with which
another action potential can be elicited (change in threshold). During absolute and relative refractory periods, the amplitude
of the action potential evoked is low. Subsequently, it is normal.
constant. If the membrane potential becomes hyperpolarized, the membrane potential moves away from
threshold, and the membrane is less excitable. If the
membrane potential moves closer to threshold, the
membrane becomes more excitable and will generate
an action potential with a smaller stimulus. If the
Fig. 13. The effect of stimulation of a neuron at different

resting potentials as recorded with a microelectrode.
A: The membrane is hyperpolarized, and a stimulus produces a subthreshold local potential. B: The membrane is
normally polarized at 65 mV, and a stimulus produces a
local potential that reaches threshold and results in an
action potential. C: The membrane is depolarized beyond
threshold, and a stimulus produces only a small local
potential.
membrane potential is very near threshold, the cell

may fire spontaneously. If the membrane potential
remains more depolarized than threshold, however,
the membrane cannot be stimulated to fire another
action potential (Fig. 13). The term threshold is also
used to describe the voltage required to excite an
action potential with an eternally applied stimulus.
When threshold is used in this sense, an axon with
an increased excitability due to partial depolarization
may be said to have a lower threshold for stimulation, even though the actual threshold is unchanged.
The first meaning of threshold is used when intracellular recordings are considered, and the second is
used in reference to extracellular stimulation and
recording. The threshold of the nerve membrane differs in different parts of the neuron: it is high in the
dendrite and soma and lowest at the initial segment.
Thus, an action potential is usually generated in the
area of the axon hillock.
2.5.4. Propagation
Another important characteristic of action potentials
is their propagation. If an action potential is initiated
in an axon in the tip of the finger, for instance, the
potential spreads along the entire length of that axon
24
to its cell body in the dorsal root ganglion, and then

along the central axon, ascending in the spinal cord
to the brainstem. This characteristic permits the nervous system to transmit information from one area
to another. When an area of membrane is depolarized
during an action potential, ionic currents flow
(Fig. 14). In the area of depolarization, sodium ions
carry positive charges inward. There is also a longitudinal flow of current both inside and outside the
membrane. This flow of positive charges (current)
toward nondepolarized regions internally and toward
depolarized regions eternally tends to depolarize the
membrane in the areas that surround the region of
the action potential. This depolarization is an electrotonic potential. In normal tissue, this depolarization
is sufficient to shift the membrane potential to threshold and thereby generate an action potential in the
immediately adjacent membrane. Thus, the action
potential spreads away from its site of initiation
along an axon or muscle fiber. Because of the refractory period, the potential cannot reverse and spread
back into an area just depolarized. The rate of conduction of the action potential along the membrane
depends on the amount of longitudinal current flow
and on the amount of current needed to produce
depolarization in the adjacent membrane. The longitudinal current flow can be increased by increasing
the diameter of an axon or muscle fiber, because this
J.R. DAUBE
increase reduces the internal resistance, just as a

larger electrical wire has a lower electrical resistance.
However, many axons in the central and peripheral
nervous systems have an increased conduction
velocity because they are insulated with a myelin
sheath. A myelinated axon has its membrane bared
only at the nodes of Ranvier, so that transmembrane
current flow occurs almost exclusively at the nodal
area. When current flow opens enough sodium
channels to reach threshold in the nodal area, it
results in many more sodium channels opening and
an influx of sodium ions with a generation of an
action potential. The nodal area in the mammalian
nervous system is unique in that it consists almost
exclusively of sodium channels, with an almost
complete absence of potassium channels. The potassium channels are located at the paranodal regions
(adjacent to the node), which are covered
by myelin. The action potential generated at the
node consists predominantly of inward sodium currents with little outward potassium currents, and
repolarization is achieved by means of sodium inactivation and leakage currents. An action potential at
one node of Ranvier produces sufficient longitudinal current flow to depolarize adjacent nodes to
threshold, thereby propagating the action potential
along the nerve in a skipping manner called
saltatory conduction (Fig. 15).
Fig. 14. Current flow and voltage changes in an axon in the region of an action potential. The voltage changes along the
membrane are shown in the upper part of the figure and the spatial distribution of current flow is shown in the lower part
as arrows through the axon membrane.
25
Fig. 15. Saltatory conduction along an axon from left to right. A: The charge distribution along the axon is shown with
an action potential (depolarization) at the second node of Ranvier (N2). Current flow spreads to the net node (N3).
B: Membrane current flow along the axon. C: The portion of the action potential found at each node is indicated by
dotted lines.
2.5.5. Patterns of activity

Information in the nervous system is coded by the
number and type of axons that are active and by the
firing pattern of action potentials. This activity is
initiated in peripheral receptors or in neurons. Neuronal firing of action potentials may occur spontaneously or in response to eternal stimulation. Beating,
or pacing, neurons fire repetitively at a constant
frequency; their intrinsic firing rate may be increased
or decreased by eternal stimulation. Bursting neurons
generate regular bursts of action potentials separated
by hyperpolarization of the membrane. Such neurons
are important for rhythmic behavior such as breathing, walking, and chewing. Neurons that fire in
response to external stimulation may do so in one
of three ways. A sustained response neuron shows
repeated action potentials with a constant firing frequency that reflects the strength of the stimulus.
A delayed response neuron fires action potentials
only after stimulation of sufficient intensity. An
accommodation response neuron fires only a single

potential at the onset of stimulation and remains
silent thereafter. Some neurons (e.g., in the thalamus)
have the ability to discharge either in rhythmic bursts
or with typical action potentials. The firing pattern
depends on the level of the resting membrane potential. An important property of this type of neuron
is the presence of a particular class of calcium channel, the T channel. This channel can be activated
only if the membrane potential is relatively hyperpolarized (e.g., 80 mV). Under this condition, a stimulus opens the T channel and calcium enters the cell
and produces a small, brief calcium-based depolarizing potential change called the low-threshold calcium
spike. This calcium spike triggers the opening of
sodium channels, which produces a burst of repetitive action potentials. As calcium accumulates in
the cell, it opens a calcium-activated potassium channel that allows the efflux of potassium. The resulting
hyperpolarization (called after-hyperpolarization)
allows reactivation of the T channel, the entry of
26
sodium, and recurrence of the cycle. This sequence

generates rhythmic burst firing of the neuron. Thus,
T channels are an exception to the general rule
of neuronal excitability: hyperpolarization deinactivates T channels and increases the likelihood that
the neuron will discharge in rhythmic bursts of activity. Rhythmic burst firing in thalamic neurons that
project to the cerebral cortex impairs the encoding
of information by cortical neurons and interferes with
the transmission of sensory information. Inactive
states of the cerebral cortex occur during deep sleep
and in some types of seizures.
2.6. Synaptic transmission
A synapse is a specialized contact zone where one
neuron communicates with another neuron. The contact zone between a neuron and a nonneural effector
element (e.g., a muscle fiber) is referred to as a neuroeffector junction. The two types of synapses are
chemical and electrical. The most common form of
communication in the nervous system is through
chemical synapses. A chemical synapse consists of
a presynaptic component (containing synaptic vesicles), a postsynaptic component (dendrite, soma, or
axon), and an intervening space called the synaptic
cleft (Fig. 16). Many of the drugs used in clinical
medicine have their pharmacologic site of action at
the synapse. The mechanism underlying chemical
synaptic transmission should make it apparent that
synaptic transmission has three unique characteristics. First, conduction at a synapse is delayed because
of the brief interval of time required for the chemical
events to occur. Second, because the two sides of the
synapse are specialized to perform only one function,
transmission can occur in only one direction across
the synapse. Thus, neurons are polarized in the direction of impulse transmission. Third, because nerve
impulses from many sources impinge on single cells
in the central and peripheral nervous systems, synaptic potentials summate both temporally and spatially.
The membrane of a cell is continually bombarded
with neuromodulators and neurotransmitters, which
produce either excitatory postsynaptic potentials
(EPSPs) or inhibitory postsynaptic potentials (IPSPs)
of varying duration. When the membrane potential
reaches threshold, an action potential is generated.
Thus, a single neuron can integrate activity from many
sources. A summary of the electrical events in a single
cell underlying the transmission, integration, and
conduction of information is shown in Fig. 17.
J.R. DAUBE
2.6.1. Biosynthesis, storage, release, and reuptake

of neurochemical transmitters
Neurochemical transmitters include amino acids, acetylcholine, monoamines (catecholamines, serotonin,
histamine), neuropeptides, and purines (ATP and
adenosine). Amino acids include L-glutamate, the
most abundant excitatory neurotransmitter in the central nervous system, and g-aminobutyric acid
(GABA), the most abundant inhibitory neurotransmitter. Both of these neurotransmitters are synthesized from intermediate metabolites of the Krebs
cycle. Acetylcholine and monoamines are synthesized by specific enzymes from precursors that are
actively taken up by the presynaptic terminal. Neuropeptides are synthesized from messenger RNA in the
cell body and transported to the synaptic terminal.
Neurochemical transmitters are stored in special
intracellular organelles called synaptic vesicles.
Small clear synaptic vesicles store the classic neurotransmitters (amino acids, acetylcholine, monoamines), and large dense core secretory granules
store neuropeptides. Neurotransmitter release is triggered by the influx of calcium through voltage-gated
channels that open in response to the arrival of an
action potential in the presynaptic terminal. These
channels are clustered in specific regions of the presynaptic membrane called active zones (Fig. 16).
The synaptic vesicles are mobilized in the presynaptic terminal and dock close to the active zones.
In response to the influx of calcium, the vesicle
membrane fuses with the presynaptic membrane,
which allows the release of the neurotransmitter into
the synaptic cleft; this process is called exocytosis.
The mobilization, docking, and fusion of synaptic
vesicles depend on the interactions of various synaptic vesicle proteins with other components of the presynaptic terminal. A neuron can produce and release
different neurotransmitters. Neurons frequently contain a classic neurotransmitter (an amino acid or acetylcholine) and one or more neuropeptides. The
neuron can release a variable mixture of these neurotransmitters according to its firing pattern, a process
referred to as frequency-dependent chemical coding.
Classic neurotransmitters can be released after a single action potential; neuropeptides are released in
response to rapid, burst firing of a neuron. Two other
presynaptic mechanisms also regulate neurotransmitter release: (1) in many cases, the neurotransmitter
inhibits its own release, acting via presynaptic inhibitory autoreceptors and (2) in other cases, inhibitory
27
Fig. 16. Synaptic transmission. A: In a resting synapse, both the presynaptic axon terminal and the postsynaptic membrane
are normally polarized. B: In an active synapse, an action potential invades the axon terminal (from left in the diagram) and
depolarizes it. Depolarization of the axon terminal of a presynaptic neuron results in the release of neurotransmitter from
the terminal. The neurotransmitter diffuses across the synaptic cleft and produces local current flow and a synaptic potential
in the postsynaptic membrane, which initiates the effector activity (neuronal transmission, neurotransmitter release, hormonal secretion, or muscle contraction).
neurons (generally containing GABA) form axoaxonic synapses that inhibit the release of neurotransmitter from the postsynaptic axon, a process called
presynaptic inhibition (Fig. 18). The synaptic action
of neurotransmitters is terminated by several
mechanisms. Presynaptic reuptake, mediated by specific sodium-dependent and ATP-dependent neurotransmitter transporters, is the primary mechanism
of in activation of glutamate, GABA, and monoamines. Monoamines are metabolized after reuptake
by monoamine oxidases and methyl transferases.

Acetylcholine and neuropeptides do not undergo
reuptake but are rapidly inactivated by enzymatic
hydrolysis in the synaptic cleft.
2.6.2. Postsynaptic effects of neurochemical
transmitters
Postsynaptic effects are mediated by two main classes
of receptors. (1) Ligand-gated receptors or ion
28
J.R. DAUBE
Fig. 17. Neuronal electrical activity from its initiation by excitatory postsynaptic potentials (EPSPs) to its transmission as
an action potential to another area.
channels mediate rapid changes in ionic conductance

(ionotropic effect). (2) G-protein-coupled receptors
produce slower changes in neuronal excitability and
metabolism (metabotropic effect) (Table 7). These
changes not only modify the electrical behavior of
the neuron but they also may produce long-term
effects, such as use-dependent modification of synaptic efficacy, cytoskeletal changes during development
and repair, and control of genetic transcription.
2.6.2.1. Classic neurotransmission
Classic neurotransmission is used for fast, precise,
point-to-point transmission of excitatory or inhibitory
signals. It involves rapid, brief opening of ligand-gated
ion channel receptors. Neurotransmitters produce a
transient increase or decrease in ion channel conductance to the passive flow of a specific ion current.
These ionic currents produce local changes in the
membrane potential called postsynaptic potentials.
In most mammalian neurons, the resting membrane
potential is 6080 mV. The threshold for opening
voltage-gated sodium channels that trigger an action
potential is reached when the postsynaptic potentials
drive the membrane potential to a value that is about
10 mV less negative than the resting potential. Ion currents that increase the net positive charge of the membrane produce depolarization EPSPs because they
bring the membrane potential toward the threshold
for triggering an action potential. In classic neurotransmission, fast EPSPs result from the opening cation
channels (conducting sodium ions and, in some cases,
calcium ions). Ligand-gated cation channel receptors
that produce fast EPSPs include nicotinic acetylcholine receptors and several ionotropic glutamate receptors. Ion currents that increase the net negative
charge of the membrane produce IPSPs. Fast IPSPs
are produced by the opening of chloride channels.
GABA (via GABAA receptors) and glycine act via this
mechanism (Table 8 and Fig. 19).
2.6.2.2. Neuromodulation
Neurotransmitters acting through G-protein-coupled
receptors, second messengers, and protein phosphorylation cascades control the excitability and responsiveness of neurons to rapid synaptic signals, a
process called neuromodulation. G-protein-coupled
receptors include metabotropic glutamate receptors,
GABAB receptor, and receptors for catecholamines,
29
Fig. 18. A: Presynaptic inhibition of neuron 3 when axon 1 partially depolarizes axon 2. B: Response to axon 2 acting
alone. C: Response to axon 2 after depolarization of axon 1. In the latter case, there is less neurotransmitter and a smaller
excitatory postsynaptic potential (EPSP).
serotonin and histamine, neuropeptides, and adenosine.

Potassium channels are an important target of neuromodulatory signals. Potassium currents determine the
pattern of activity generated by neurons through control of the resting membrane potential, repolarization
of the action potential, and probability of generation
of repetitive action potentials. The opening of potassium channels brings the membrane potential toward
the equilibrium potential of potassium (100 mV) and
thus away from the threshold for triggering an action
potential. Closure of the potassium channels moves
the membrane away from the equilibrium potential of
potassium and thus closer to threshold. Neuromodulation involves the production of slow potentials. Activation of G-protein receptors that lead to closure of
potassium channels produces slow depolarization and

increased neuronal excitability. G-protein receptor
mechanisms that increase potassium permeability lead
to membrane hyperpolarization and reduce neuronal
excitability. The same neurotransmitter may act via
different receptor subtypes, each coupled to a distinct
transduction pathway. Also, different neurotransmitters, via their respective receptors, may activate a similar transduction pathway.
2.6.3. Electrical synapses
Although most synapses in the nervous system use
chemical transmitters, neurons with junctions that contain channels extending from the cytoplasm of the
30
J.R. DAUBE
Table 7
Comparison of classic neurotransmission and neuromodulation
Classic neurotransmission
Neuromodulation
Function
Receptor mechanism
Ionic mechanism
Rapid synaptic excitation or inhibition

Ion channel receptors
Opening of either cation channel (fast
EPSP) or Cl channel (fast IPSP)
Example
L-Glutamate (ionotropic)
GABA (GABAA)
Acetylcholine (nicotinic)
Systems
Relay systems, direct

Sensory
Motor
Modulation of neural excitability

G-protein-coupled receptors
Opening or closing of voltage-gated K or
Ca2 channels (slow IPSP and slow
EPSP)
L-Glutamate (metabotropic)
GABA (GABAB)
Acetylcholine (muscarinic)
Monamines
Neuropeptides
Adenosine
Diffuse systems, indirect
Internal regulation
Consciousness
EPSP, excitatory postsynaptic potential; GABA, g-aminobutyric acid; IPSP, inhibitory postsynaptic potential.
presynaptic neuron to that of the postsynaptic neuron

interact electrically. In these electrical synapses, the
bridging channels mediate ionic current flow from
one cell to the other. Transmission across the electrical
synapse is very rapid, without the synaptic delay of
chemical synapses. Electrical synapses are also bidirectional, in contrast to chemical synapses, which
transmit signals in only one direction.
2.7. Clinical correlations
2.7.1. Pathophysiologic mechanisms
The mechanisms responsible for neuronal excitability,
impulse conduction, and synaptic transmission in the
central and peripheral nervous system may be altered
transiently to produce either loss of activity or overactivity of neurons. A loss of activity results in a clinical
deficit of relatively short duration (seconds to hours);
overactivity results in extra movements or sensations.
Both types of transient alteration are usually reversible. These transient disorders may be focal or
generalized (Table 9) and may be due to different
mechanisms (Table 10). Transient disorders reflect
disturbances in neuronal excitability due to abnormalities in membrane potential.
2.7.1.1. Energy failure
Energy metabolism is necessary for maintenance of
the membrane potential by the ATP-coupled
sodium/potassium pump. Most of the ATP produced
in the nervous system by aerobic metabolism of glucose is used to maintain the activity of the sodium
pump. Conditions such as hypoxia, ischemia, hypoglycemia, or seizures affect the balance between
energy production and energy consumption of neurons and cause energy failure and thus impaired
activity of sodium/potassium ATPase. If the active
transport process stops, the cell accumulates sodium
and loses potassium and the membrane potential
Table 8
Postsynaptic potentials
Receptor (example)
Ionic mechanism
Effect
Nicotinic and ionotropic glutamate

GABAA and glycine
G protein-coupled receptors
Increase Na or Ca2 conductance

Increase Cl conductance
Decreased K conductance
Increased K conductance
Fast excitation
Fast inhibition
Slow excitation
Slow inhibition
31
Fig. 19. Postsynaptic inhibition in the neuron on the left occurs when the inhibitory and excitatory endings are active
simultaneously. On the right, a microelectrode recording shows two excitatory postsynaptic potentials (EPSPs) summating
to initiate an action potential. When there is a simultaneous occurrence of an inhibitory postsynaptic potential (IPSP),
depolarization is too low to reach threshold, and no action potential occurs.
progressively decreases. This depolarization has two

consequences. First, there may be a transient increase
in neuronal excitability as the membrane potential
moves closer to threshold for opening voltage-gated
sodium channels and triggering the action potential.
This may produce a paroxysmal discharge of the
neuron or in an axon. Second, if depolarization persists, the sodium channel remains inactivated and the
neuron becomes inexcitable. This is known as depolarization blockade and results in a focal deficit, such as
focal paralysis or anesthesia, or a generalized deficit,
such as paralysis or loss of consciousness (Fig. 20).
Table 9
Transient disorders of neuronal function
Table 10
Neuronal
excitability
Focal disorder
Generalized
disorder
Mechanisms of transient disorders
Increased
Focal seizure
Decreased
Tonic spasms
Muscle cramp
Paresthesia
Paroxysmal pain
Transient ischemic
attack
Migraine
Transient
mononeuropathy
Generalized
seizure
Tetany
Syncope
Concussion
Cataplexy
Periodic paralysis
Energy failure
Hypoxiaischemia
Hypoglycemia
Seizures
Spreading cortical depression
Trauma
Ion channel disorders
Mutation of channel protein (channelopathies)
Immune blockade
Drugs
Toxins
Electrolyte disorders
Demyelination
32
J.R. DAUBE
Fig. 20. Effects of increasing severity of energy failure (and ATP depletion) on activity of ATP-driven pumps, ionic concentrations in the intracellular and extracellular fluid, and neuronal electrical activity. With progressive failure of ATPdriven pumps, potassium accumulates in the extracellular fluid and sodium and calcium accumulate inside the neuron. This
produces progressive neuronal depolarization. With partial depolarization, the resting potential moves closer to the threshold for triggering an action potential; this results in a transient increase in neuronal excitability, which may be manifested
by paresthesias or seizures. With further depolarization, the membrane potential is at a level that maintains inactivation of
the sodium channel, preventing further generation of action potentials and, thus, reducing neuronal excitability. This constitutes a depolarization block, which manifests with transient and reversible deficits such as paralysis or loss of consciousness. If the energy failure is severe and prolonged, the excessive accumulation of intracellular calcium triggers various
enzymatic cascades that lead eventually to neuronal death and irreversible loss of function.
The neuron also uses ATP to maintain ion gradients

that allow active presynaptic reuptake of neurotransmitters, such as the excitatory amino acid L-glutamate.
Under conditions of energy failure, glutamate accumulates in the synapse and produces prolonged activation
of its postsynaptic receptors, leading to neuronal depolarization and the accumulation of calcium in the cytosol. Because the lack of ATP also impairs active
transport of calcium into the endoplasmic reticulum
or toward the extracellular fluid, calcium accumulates.
Essentially, all forms of neuronal injury involve to various extents
generation of free radicals and

mitochondrial failure.
accumulation of glutamate and activation of gluta-
2.7.1.2. Ion channel blockade

Voltage-gated sodium channels mediate the initiation
and conduction of action potentials. Voltage-gated
calcium channels mediate neurotransmitter release,
mate receptors
accumulation of cytosolic calcium and activation of
calcium triggered cascades
The consequences are functional and potentially

reversible (e.g., cell depolarization, pump failure,
and accumulation of intracellular sodium). If the
cause is not corrected, calcium accumulates and triggers irreversible changes, including destruction of
cellular and mitochondrial membranes, disorganization of the cytoskeleton, and degradation of DNA
by nucleases, and these effects lead to cell death by
necrosis or apoptosis.
33
Fig. 21. Abnormalities of synaptic transmission may occur (A). Types of transmission block include block of transmitter release
(block), block of transmitter binding to postsynaptic membrane (competitive inhibition), and binding of another depolarizing
agent to the membrane (depolarizing block). B: These types of abnormalities may occur at each neuronal synapse shown.
and ligand-gated cation (sodium and calcium) channels mediate EPSPs. All these channels may be
blocked by autoantibodies, drugs, or toxins. Examples
of the types of transmission block are illustrated in
Fig. 21. There may be presynaptic block of transmitter
release, or postsynaptic block by competitive or noncompetitive inhibition of postsynaptic receptors, or by
depolarizing substances. Blockade of sodium channels
at the node of Ranvier slows conduction velocity or
causes conduction block; this produces a reversible
focal deficit (weakness or anesthesia). For example,
the blockade of sodium channels in sensory axons by
local anesthetic agents produces anesthesia, and antibodies against ganglioside GM1 (associated with sodium
channels) in the nodes of Ranvier of motor axons produce focal paralysis. Autoantibodies may also block
ion channels involved in neuromuscular transmission
and produce reversible muscle fatigue or paralysis.
2.8. Summary
The transmission of information in the nervous system depends on the generation of a resting potential
that acts as a reserve of energy poised for release

when the valve is turned on. Ionic channels act as
the valve, controlling the energy in the ionic concentration gradient. The release of energy is seen either
as local graded potentials or as propagated action
potential that arise when local potentials reach
threshold. Information is moved from one area to
another as action potentials conducted by single cells.
The information is integrated in neurons by the interaction of local potentials generated in response to the
neurotransmitters released from depolarized nerve
terminals. In this system, information can be coded
either as the rate of discharge in individual cells
or axons or as the number and combination of
active cells. Both of these are important mechanisms, for although the activity of the nervous system can be conveniently described in terms of the
electrical activity of single cells, the combined
activity of large numbers of cells and axons determines the behavior of the organism. Each type of
alteration in neuronal or muscle cell physiology
can produce symptoms or signs of short duration,
transient disorders. The particular findings in a
34
patient depend on which cells are altered. If the

changes are in neurons that subserve sensation, there
may be a loss of sensation or an abnormal sensation
such as tingling, loss of vision, or seeing stars. In
other systems, there might be loss of strength,
twitching in muscles, loss of intellect, or abnormal
behavior. In all these cases, the physiologic alterations are not specific and may be the result of any
one of a number of diseases. Transient disorders do
not permit a pathologic or etiologic diagnosis. Any
type of disease (vascular, neoplastic, inflammatory)
may be associated with transient changes. Therefore, the pathology of a disorder cannot be deduced
when its temporal profile is solely that of transient
episodes. The transmission of information in the
nervous system depends on the generation of a resting potential that acts as a reserve of energy poised
for release when the valve is turned on. Ionic channels act as the valve, controlling the energy in the
ionic concentration gradient. The release of energy
is seen either as local graded potentials or as propagated action potentials that arise when the local
potentials reach threshold.
2.9. Peripheral and cranial nerve physiology
Nerves are the gross structures carrying motor,
sensory, and autonomic axons to the end organs.
J.R. DAUBE
(Some nerves contain only autonomic fibers, e.g.,

the vagus nerve.) Nerves, whether peripheral or
spinal, are made up of the axons traveling between
the central nervous system and the peripheral end
organ. They are similar in their microscopic features, their physiology, and their pathophysiologic
alterations with disease. The general features that
are common to all types of nerves are considered
first and the differences considered subsequently.
A nerve is composed of thousands of axons ranging
in size from less than 1 to 20 mm in diameter. In
each nerve trunk, individual fibers are surrounded
by a connective tissue sheath, the endoneurium.
Each of these is grouped with many other axons into
bundles of fascicles by the perineurium. Groups of
fascicles are bound together by an outer covering of
connective tissue, the epineurium (Fig. 22). Nerves
have their own blood supply. The nutrient arteries
enter at intervals along their length and form anastamotic channels within the connective tissue framework of the nerve. These anastomoses make nerves
relatively resistant to vascular disease. Nerves are
made up of afferent and efferent fibers. The axons
can be differentiated histologically on the basis of
their size and the presence or absence of myelin.
The unmyelinated fibers are small and include autonomic fibers and fibers carrying pain and temperature. Proprioceptive and somatic motor fibers are
Fig. 22. Histologic features of a peripheral nerve. A nerve is subdivided into fascicles by the perineurium, with multiple
motor and sensory nerve fibers intermingled in each fascicle.
large. However, these characteristics do not permit

the identification of the function of an individual
axon, because the afferent (carrying information centrally) axons and the efferent (carrying information
peripherally) axons have a similar microscopic
appearance. Each nerve fiber consists of an axon
embedded in a series of Schwann cells arranged longitudinally along the axon. Each Schwann cell covers
0.51.0 mm of axon. The junctions between
Schwann cells along the axon are seen as constrictions of the nerve fiber and are called the nodes of
Ranvier (Figs. 22 and 23). A single Schwann cell surrounds either a number of unmyelinated axons or one
myelinated axon. During development, either many
unmyelinated axons become embedded in the
Schwann cell or the Schwann cell wraps around one
axon in concentric circles to form the myelin of the
myelinated nerve fiber. Although the fibers in a
nerve are adjacent to one another, the electrical activity in each nerve fiber is independent of the activity
in all the other fibers in the nerve. The action potentials are isolated from each other by the endoneurium
and the myelin. As the Schwann cell encircles an
35
axon, layers of plasma membrane fuse to form myelin (Fig. 24). Myelin is thus a series of concentric
layers of lipids and proteins. The lipids include cerebrosides, sulfatides, proteolipids, sphingomyelin, inositol phosphatides, phosphatidylserine, glycolipids,
glycoproteins, and cholesterol. Myelin contains specific proteins expressed in myelin forming Schwann
cells (or in oligodendrocytes in the central nervous
system). These proteins are adhesion molecules
involved in the processes of wrapping and compaction of the myelin sheath. Myelin proteins in the
peripheral nervous system include protein zero (P0)
and peripheral myelin protein 22 (PMP22). The gap
junction protein connexin is expressed in Schwann
cells and is also critical for myelination. Mutations in
the genes encoding for myelin proteins produce several
types of hereditary sensory and motor neuropathies.
Although myelin is relatively inert metabolically, it
has a significant turnover and responds to various disease states. For instance, myelin may be lost (demyelination) in certain immunologic disorders. When myelin
is lost along a peripheral nerve, it is usually lost in the
region of a single Schwann cell, which extends from
Fig. 23. Histologic features of a myelinated motor nerve fiber. A: Single myelinated axon extends from a ventral horn cell
to nerve terminals on muscle fibers. B: Cross-section through an internode of a nerve fiber, with layers of myelin formed by
Schwann cell membrane wrapped around it. C: Longitudinal section of a node of Ranvier, with Schwann cell and myelin
terminations abutting around continuous central axon.
36
J.R. DAUBE
Fig. 24. Formation of myelin from layers of Schwann cell membrane. Major dense lines are formed from protein layers of
membrane and are separated by the lipid layer, which contains cholesterol, cerebroside, sphingomyelin, etc.
one node of Ranvier to another. This loss is called segmental demyelination and alters the function of a nerve
fiber. Myelin also may be formed abnormally or may
accumulate myelin metabolites in abnormal quantities.
This condition occurs in genetic disorders due to
enzyme defects, such as metachromatic leukodystrophy
in which a deficit of arylsulfatase A results in the accumulation of metachromatic sulfatides in nerve fibers
and loss of function in myelinated axons. The axons of
all nerve fibers consist of the axon membrane, or axolemma, and the axoplasm. The axoplasm contains
mitochondria, microtubules, microfilaments, and neurofilaments. The mitochondria mediate the generation
of energy needed to establish the concentration gradients across the axolemma. The microtubules participate
in the transport of proteins, enzymes, and other materials down the axon from the cell body to the periphery.
The function of the neurofilaments and microfilaments
is related to axonal transport and axon growth.
2.9.1. Axonal transport
The continuous and regulated flow of material from the
cell body to the axons and synaptic terminals (and in the
reverse direction) is critical for neuronal function and
survival. Transport of substances along axons is not random but directed by the microtubules, which give polarity to the transport. Axonal transport includes
anterograde transport, which allows the constant flow
of material synthesized in the cell bodies and dendrites
to reach the axon terminals, and retrograde transport,
which is the transport of material from axon terminals
to the cell body. Retrograde transport is a mechanism
for the cell body to sample the environment around
the synaptic terminals of its axons. Fast axonal anterograde transport occurs at a rate of 200400 mm/day
and is involved in the movement of proteins associated
with membrane vesicles. These include glycosylated
proteins that are delivered preferentially to synaptic
terminals (e.g., synaptic vesicles, ion channels, neuropeptides, and enzymes for neurotransmitter biosynthesis). Slow anterograde axonal transport occurs at a rate
of 0.14 mm/day and is involved in the movement of
cytoskeletal proteins (e.g., tubulin, actin, and neurofilament proteins). Retrograde transport occurs at a rate of
100200 mm/day and is an exaggerated manifestation
of the process of endocytosis. It is involved with the
incorporation and recycling of lysosomes, pinocytotic
vesicles, synaptic vesicle proteins, and neurotrophic
factors. Retrograde transport is the mechanism by
which some viruses (e.g., rabies and herpes simplex)
and toxins (e.g., tetanus and botulinum toxins) enter
the nervous system. Organic solvents (used in industry
for cleaning, extraction, laboratory work, paint, printing
ink), pesticides, and some antineoplastic drugs can produce axonal neuropathy by disrupting the normal
mechanisms of axonal transport and cytoskeletal
assembly. Overexposure to these toxic chemicals produces a distal symmetrical sensorimotor axonal neuropathy that affects large diameter sensory and motor axons
in peripheral nerves and, in severe cases, long tracts in
the spinal cord.
2.9.2. Physiology
The resting potential and action potentials in single
axons are described in detail earlier. Here, we focus
on the physiology of whole nerve trunks. The function
of the axons is to carry information in the form of

electrical activity from one area to another.
A measure of the ability of a nerve to perform this
function would be of major clinical value in the identification of disease involving a nerve. However, during normal function, the electrical activities of the
fibers in a nerve are asynchronous and cancel each
other out. The action potential of a single axon can
be recorded experimentally with an intracellular
microelectrode, which records the action potential
as a monophasic wave of depolarization. The electrical activity in a single nerve fiber also can be monitored by placing electrodes in the extracellular fluid
close to the nerve fiber. This method does not detect
transmembrane potential changes; rather, it senses
potential changes in the extracellular fluid that result
from longitudinal current flow between the depolarized and nondepolarized regions of the axon. The
extracellular recording is improved (a bigger voltage
change is measured) if the extracellular resistance is
artificially increased by recording from the nerve
experimentally in air or in oil. Extracellular recording from single axons is difficult because of their
small size; however, it is possible to record from
groups of axons or from whole nerve trunks, if all
axons discharge synchronously. This recording is
obtained experimentally and from patients by applying an electrical shock that activates all axons simultaneously. The potential recorded from a nerve
activated in this way is the compound action potential. The configuration of the signal obtained from
an extracellular recording of the nerve impulse
depends on the electrode arrangement. A monophasic
potential change is observed from nerve fibers conducting an impulse if only one of the electrodes is
placed over an active nerve. A biphasic potential is
recorded if both electrodes are placed over the active
nerve (Fig. 25). As in the stimulation of a single
axon, the whole nerve trunk is activated by passing
a current between the cathode (negative pole) and
the anode (positive pole). The cathode depolarizes the
underlying axons, and the anode hyperpolarizes
them. Depolarization requires current flow inside
the axons. Because large axons have lower internal
resistance, the threshold for activation is lowest for
the larger fibers. The threshold stimulus for a nerve
trunk is that which just excites the large fibers.
Supramaximal stimuli activate all fibers, including
the small fibers, and require greater current flow.
Excitability, therefore, depends on axon size. The
37
excitability of a nerve can be defined in terms of

the two variables of a stimulus voltage and duration.
If the strength of the current (or voltage) is plotted
against the duration of a stimulus needed to produce
excitation of a nerve, a curve is obtained, which is
called a strength duration curve. A shift in the
strength duration curve indicates a change in excitability and is seen in nerve diseases. The strength
duration curve is often characterized in terms of
two points. The rheobase is the minimal voltage
needed to produce excitation with a long stimulus
duration (usually 300 ms) and the chronaxie is the
time required to excite a nerve by a stimulus with a
voltage twice as large as the rheobase (Fig. 26).
The compound action potential recorded from a
nerve trunk after supramaximal stimulation is the
summation of action potentials from many axons.
Its amplitude can be graded by varying the strength
of the stimulus. A threshold stimulus evokes only a
small potential resulting from activity in a few large
fibers. As the stimulus strength is increased, more
fibers are excited, and their activity is added to the
compound action potential as each additionally activated fiber produces a small increment in the recorded
voltage. When all the fibers are excited, the amplitude
of the compound action potential is maximum; it will
not increase in amplitude with further increases in the
stimulus strength (supramaximal). The compound
action potential thus can be graded in amplitude, while
action potentials in single axons are not graded but
fire in an all-or-none fashion. Variation in axon diameter in a nerve trunk results in different conduction
velocities as well as different thresholds for activation.
The rate at which an axon conducts is a function of
the amount of longitudinal current flow and is greater
with larger axons. The conduction velocity is calculated by dividing the distance a potential travels by
the time it takes to travel that distance. It is approximately five times the axons diameter in microns, for
example, 5100 m/s for axons of 120 mm diameter.
If a nerve trunk is stimulated at a distance from the
recording electrodes, the compound action potential
exhibits several components (Fig. 27) because of
the dispersion of the potentials from fibers of different diameters. The impulses in the large fibers reach
the recording site first. The components of the compound action potential thus distinguish activity in
groups of fibers whose diameters are within certain
size ranges. The afferent fibers in cutaneous nerves
(to joints and skin) are subdivided into groups named
38
J.R. DAUBE
Fig. 25. Recording action potentials from a nerve trunk with electrodes (G1 and G2). A: With G2 over damaged nerve and
G1 over active nerve, a passes under G1. B: If both electrodes are over active nerve fibers, potentials of opposite polarity
are recorded as depolarization passes under G1 and G2, a biphasic action potential. When G1 and G2 are close together, the
two potentials fuse to form a smooth biphasic response.
by letters (Aa, Ad, and C). The afferent fibers in

muscle nerves (nerves to muscle) are subdivided into
groups designated by Roman numerals (I, II, III, and
IV). These are listed in Table 11. Nerves that innervate muscle contain both sensory and motor fibers.
The motor fibers arise from the alpha and gamma
motor neurons and innervate the extrafusal and intrafusal muscle fibers. The sensory fibers are the group
Ia and II fibers from muscle spindles and the group
Ib fibers from Golgi tendon organs. Cutaneous
nerves innervate joints and skin and are commonly
considered sensory nerves, although both they and
the muscle nerves contain efferent and afferent fibers
of the autonomic nervous system, the C or group
IV fibers. Table 12 lists by size the components of
a mixed nerve. In addition to transmitting action

potentials, axons move proteins along their length.
This process of axonal transport is important for
making available the enzymes needed for the production of neurotransmitter in the nerve terminal, for
maintaining the integrity of the distal parts of the axon,
and for the release of trophic factors from the nerve terminal. Trophic factors are released from nerve terminals and are necessary for the normal function of the
postsynaptic cell. Loss of these trophic factors occurs
in some nerve diseases and results in physiologic and
histologic abnormalities of the postsynaptic cell. A
nerve may be altered in several ways by disease processes. These can be classified as diseases of the axolemma, the axoplasm, or the myelin sheath.
39
Fig. 26. Strengthduration curve. Threshold voltage for each duration is plotted. Rheobase is 25 V and chronaxie
is 0.6 ms.
Fig. 27. Compound action potential recorded directly from a cutaneous nerve, showing peaks generated by different fiber
types.
40
J.R. DAUBE
Table 11
Nerve fiber types
Type
Diameter (mm)
Conduction
velocity (m/s)
Function
Muscle nerve afferents

Ia
1220
70120
Ib
II
1220
612
70120
3070
III
IV
26
<2
430
0.52.0
Afferents from muscle spindle (primary

endingsannulospiral)
Afferents from Golgi tendon organs
Afferents from muscle spindle (secondary
endingsflower spray)
Pressurepain afferents
Pain afferents
1220
612
70120
3070
26
<2
430
0.52.0
212
<2
470
0.22.0
Internal regulation receptors
1220
70120
Gamma
28
1050
B
C
<3
<1
330
0.52.0
Extrafusal skeletal muscle innervation from

alpha motor neurons
Intrafusal muscle spindle innervation from
gamma motor neurons
Preganglionic autonomic efferents
Postganglionic autonomic efferents
Cutaneous nerve afferents

Aa
Aa
Ad
C
Visceral nerve afferents
A
C
Efferents
Alpha
Table 12
Fiber types in a mixed nerve
Diameter (mm)
Conduction
velocity (m/s)
Type
1220
70120
612
3070
26
430
<2
0.52.0
Ia, Ib, Aa, alpha

efferent
II, Aa, gamma
efferent, visceral
afferent
III, Ad, gamma
efferent, visceral
afferent
IV, B, C
2.9.3. Axolemmal disorders

The axon membrane may undergo physicochemical
alterations that block conduction without destruction
or histologic alteration of the axon. Such alterations
Joint receptor afferents

Paccinian corpuscle and touch receptor
afferents
Touch, temperature, and pain afferents
Pain, temperature, and some
mechanoreceptors
may occur by electrical, pharmacologic, thermal, or

mechanical means. The alterations are usually transient and reversible and include the familiar phenomenon of a leg going to sleep. Electrical conduction
blocks do not occur clinically but can be obtained by
the hyperpolarizing (anodal block) currents to a
nerve fiber or nerve trunk. A depolarizing current
may initially evoke an action potential and then
block impulse transmission. An anodal block results
from a hyperpolarization of the axon membrane,
which moves the membrane potential away from
threshold. A clinically useful method of producing
conduction block is the application of pharmacologic
agents (local anesthetics) to a nerve. These include
compounds such as procaine hydrochloride (Novocain), benzocaine, cocaine, and other esters of
benzoic acid. Local anesthetics interfere with nerve
conduction by preventing the membrane permeability
changes that occur with depolarization (Table 13).
The membrane is said to be stabilized by local

Table 13
Sequence of events of local anesthetic block
Displacement of calcium ions from nerve receptor site
#
Binding of local anesthetic to receptor site
#
Blockade of sodium channel
#
Decrease in sodium conductance
#
Decreased depolarization of nerve membrane
#
Failure to achieve threshold potential level
#
Lack of development of propagated action potential
#
Conduction blockade
anesthetics. Small, unmyelinated nerve fibers, such as

those mediating pain, are more sensitive to local anesthetics than are the larger myelinated fibers and are
blocked at low concentrations of the drug that do not
appreciably affect large fibers. A transient, reversible
41
conduction block can be obtained by lowering the temperature of nerve fibers. This method of blocking nerve
impulse transmission is accomplished by the local
application of ice or an ethyl chloride spray and is used
clinically to produce superficial anesthesia. Mechanical
conduction blocks occur with distortion of a nerve and
may be due to alteration of the blood supply or to
changes in the configuration of the membrane, with secondary changes in its ionic permeability.
2.9.4. Axoplasmic disorders
Axons may be affected by acute or chronic disorders of
the axoplasm. An acute lesion is one in which the axon
is disrupted. This may occur with complete division of
the nerve in a laceration or with a severe local crush,
traction, or ischemia. In laceration, the connective tissue
framework is destroyed; in the other lesions, it remains
intact. In each instance, the continuity of the axons is
lost, the distal axon is deprived of axonal flow from
the neuron, and it undergoes dissolution in a process
called wallerian degeneration. Central chromatolysis
and peripheral muscle atrophy accompany wallerian
degeneration (Fig. 28). In most lesions other than
Fig. 28. Pathologic changes in peripheral nerve fibers. A: Normal axon. B: Wallerian degeneration occurs distal to local destruction of an axon and is associated with central chromatolysis and muscle fiber atrophy. Regeneration occurs along the connective
tissue path. C: Axonal dystrophy results in distal narrowing and dying back of nerve terminals due to either intrinsic axon or motor
neuron disease. D: Segmental demyelination destroys myelin at scattered internodes along the axon without axonal damage.
42
laceration, not all axons are destroyed and some function may remain. The smaller fibers are more resistant
to such injuries and are more likely to be spared. After
acute axonal disruption, recovery occurs only through
the growth of new axons. If a nerve is completely severed, reinnervation is poor because the axonal sprouts
have no pathway to follow. Axonal sprouts may grow
in the wrong direction and produce spirals or large bulbous tips. These sprouts, with their Schwann cells and
connective tissue, may form a neuroma. The neuroma
may not only prevent proper regrowth of the nerve but
may also be painful. The activity of Schwann cells in
the distal nerve stump provides an aid to reinnervation
across a gap, as they divide, elongate, and migrate
toward the proximal nerve stump. If axonal sprouts
manage to reach this Schwann cell outgrowth, they
may eventually reinnervate the denervated organs.
However, the amount of functional recovery is always
less than that seen in a crush injury. One reason for this
is that most axonal sprouts do not find their way along
the pathway followed originally by their parent fibers
and reinnervate an inappropriate organ. A motor axon
that establishes a connection with a sensory receptor
organ will not function, and a motor axon that reinnervates a muscle different from the one it originally
supplies cannot take part in the same reflex actions.
Synkinesis is the result of such aberrant reinnervation,
in which attempts to activate one group of muscles produce concomitant contraction in other muscles innervated by that nerve. The patient can no longer
selectively activate a muscle. In injuries to
long nerves, the end organs may atrophy before reinnervation can occur, thus preventing normal recovery. The
rate of nerve regeneration varies with the type of injury.
Recovery is quicker with crush injuries than with nerve
severance. The delay in recovery depends on axonal
growth, reversal of atrophy of the end organ, reinnervation of the end organ, and remyelinization
and maturation of the axon. In humans, the overall
rate of functional recovery under optimal conditions
is about 13 mm/day. The recovery rate in a limb
may be quicker proximally than distally. Axoplasmic disorders may be chronic or slow in evolution
and present with different findings than do acute
lesions. When the process is complete, the axons
degenerate just as they do in acute lesions; however, there are intermediate stages in which the
axons first lose their integrity distally, a so-called
dying back. This occurs first in the longest axons
and results in loss of function in the most distal
parts of the body. The axons also may atrophy or
J.R. DAUBE
become narrowed in chronic disorders, referred to

as axonal atrophies or axonal dystrophies
(Fig. 28). In either situation, there are abnormalities
of the axon before there are changes in the myelin.
Therefore, unless major narrowing of an axon is
present, the conduction of the axons is slowed very
little. Chronic axonal disorders occur with many
diseases, including genetic, toxic, and metabolic,
and deficiency states. The narrow axons seen in some
axoplasmic disorders also occur with local compression of a nerve and in regenerating fibers. Moderate
narrowing of an axon results in slowing of conduction
velocity, but by itself, it usually causes little functional impairment. A nerve with slowed conduction
velocity can still transmit impulses, though not at as
high rates as normal. High frequency impulses, such
as rapid vibrations, the output from muscle spindles,
and motor activity in strong muscle contraction, are
poorly transmitted so that vibratory sensation and
reflexes are lost.
2.9.5. Myelin disorders
Genetic, immunologic, and toxic disorders can produce primary damage to myelin. In these disorders,
myelin is usually lost at internodes, with normal
myelin remaining at other internodes. This scattered
loss of myelin is segmental demyelination (Fig. 28).
The loss of myelin results in slowing of conduction
velocity, with mild impairment of vibratory sensation, loss of reflexes, some loss of proprioceptive
sensation, and loss of strong muscle contractions.
However, with moderate demyelination, the action
potential is blocked, producing more severe deficits.
Genetic disorders can be associated with a lack of
myelin (hypomyelination) or abnormal myelin and
can result in functional disturbances similar to those
seen with segmental demyelination. In each of these
disorders, there may be varied severity of damage
and selective involvement of one or another fiber
type. In localized lesions, there is loss of function
in the areas supplied by the nerve. In generalized
nerve disease, the axons are affected randomly
throughout the cross-section of a nerve and randomly along the length of the nerve, so that the most
likely areas to lose function are the distal regions
supplied by the longest nerves. This produces a
characteristic distribution of abnormalities in the
distal portions of the extremities. This distal deficit
also occurs in primary neuronal disease in which
the neuron is unable to provide sufficient nutrients
43
Table 14
Examples of denervation hypersensitivity
Site
Clinical finding
Due to destruction of
Hypersensitivity to
Striated muscle
Ventral horn cell
Pupil
Pupil
Fibrillation
Spasticity, clonus, hyperreflexia
Miosis
Mydriasis
Alpha efferents
Descending pathway
Postganglionic sympathetic
Postganglionic parasympathetic
Acetylcholine
Local sensory input
Epinephrine analogues
Acetylcholine analogues
to the most distal portion of the nerve, with a resultant dying back of the distal portions of the long
nerves.
2.9.6. Nerve physiology summary
Because the function of peripheral nerves is to conduct action potentials from one area to another, three
general kinds of functional abnormality occur.
(1) The excitability of axons may be increased with
spontaneous or excessive firing of an axon. This
occurs in many disorders, but especially in ischemic or metabolic diseases.
(2) The axon may be unable to conduct an action
potential, because of either transient metabolic
changes or structural damage to the axon. If an
axon is severed, the distal portion undergoes wallerian de generation but is able to conduct an
impulse in the distal part of the nerve for 3
5 days. The proximal portion of the axon continues to function normally.
(3) The axon may conduct an impulse slowly or at low
rates of firing. This may occur from loss of myelin
or be due to narrowing and deformation of the axon.
The latter may be seen in the area of compression or
in regenerating fibers. Slow conduction results in
mild clinical symptoms or signs except for the
ability to carry high-frequency information such
as vibration, which is severely impaired.
The physiologic alterations seen with diseases of
the nerves are associated with two clinically important manifestations. The threshold for activation of
some portion of lower motor neurons may be low,

and they may discharge spontaneously. If this occurs,
all the muscle fibers in the motor unit will contract
simultaneously. Such a single, spontaneous contraction of a motor unit is visible as a small twitch under
the skin called a fasciculation. It is evidence of irritability of the motor unit and occurs in normal persons
and in many disorders. Similar irritability in sensory
fibers is perceived as paresthesia (tingling) in large
fibers or as pain in small fibers. A second important
manifestation is the result of the loss of trophic factors of the nerve acting on muscle. A denervated
muscle atrophies and undergoes change in its membrane. This change includes a hypersensitivity to acetylcholine. Normally, most of the acetylcholine
receptors are confined to the area immediately adjacent to the end plate. After denervation, the receptors
spread along the surface, until the entire fiber
responds to the drug. This is one form of denervation
hypersensitivity (Table 14). Muscle fibers undergo
denervation hypersensitivity and begin to discharge
and twitch 2 weeks after losing their innervation.
Such spontaneous, regular twitching of single muscle
fibers is called fibrillation.
Acknowledgments
All figures and tables adapted with permission from:
Benarroch EE, Westmoreland BF, Daube JR, Reagan
TJ, Sandok BA.4th Edition, Lippincott, Williams and
Wilkins, Phil. (1999) Medical Neurosciences: An
Approach to Anatomy, Pathology and Physiology
by Systems and Levels.

M.R. Nuwer (Ed.)
44
CHAPTER 3
Anatomy
Jasper R. Daube*
3.1. Orientation
The major structures of the central nervous system
the brain and the spinal cord are surrounded by
three fibrous connective tissue linings called
meninges and are encased in a protective bony skeleton. The brain is enclosed in the skull; the spinal cord
is in the spinal column. Cranial and peripheral nerves
must pass through these surrounding investments to
reach more peripheral structures. The nervous system
is generally divided into four major regions: supratentorial, posterior fossa, spinal, and peripheral.
The skull (Fig. 1) is formed by the union of a
number of bones and can be grossly subdivided into
(1) the facial bones and orbits, (2) the sinus cavities
within the bones that form the anterior aspect of the
skull, and (3) the cranial bones. The cranial bones
surround the brain in the cranial cavity and provide
a non-yielding protective covering for the brain. In
contrast to other protective structures in the body,
the cranial bones severely limit the expansion of the
brain, even when expansion becomes necessary in
response to diseases. The cranial cavity is formed
by the frontal, parietal, sphenoid, temporal, and
occipital bones. The floor of the human skull viewed
from above is divided into three distinct compartments (fossae) on each side: anterior, middle, and
posterior (Fig. 2). The cranial nerves emerge through
symmetrically placed holes (foramina) in the base of
the skull to innervate peripheral structures.
Reflections of a rigid membrane arising from the
skull, the dura mater, form the falx cerebri and tentorium cerebelli (Fig. 3). The falx and tentorium
divide the cranial cavity into spaces for and supports
the major components of the brain. The tentorium

lies in a nearly horizontal plane and attaches laterally
to the petrous ridges on the inside of the skull and
posteriorly to the occipital bone. The cerebral hemispheres in the anterior and middle fossae are above
the tentorium, and the cerebellum and brainstem in
the posterior fossa are below the tentorium cerebelli.
The tentorium and the foramen magnum (the opening of the skull to the spinal canal) in the base of the
skull define the major neurologic regions: supratentorial, posterior fossa, spinal, and peripheral (Fig. 4).
Structures located above the tentorium are supratentorial. The major supratentorial structures are the
cerebral hemispheres, basal ganglia, thalamus, hypothalamus, and cranial nerves I (olfactory) and II (optic).
Structures located within the skull below the tentorium, but above the foramen magnum are in the posterior fossa. The structures in the posterior fossa are
midbrain, pons, medulla, and cerebellum (Fig. 5). Cranial nerves III through XII are located in the posterior
fossa (Fig. 6).
The portions of the central nervous system located
below the foramen magnum of the skull that are
contained in the vertebral column are at the spinal
level from the skull to the sacrum (Fig. 7). However,
the spinal cord itself (the major structure at the spinal
level) does not extend that entire length. A series of
spinal nerves arise from the spinal cord and exit
through the intervertebral foramina (Figs. 8 and 9).
All neuromuscular structures located outside the
skull and vertebral column, including the peripheral
nerves, their branches, and the structures (including
muscle) that are innervated by these nerves are
peripheral. The skull, meninges, blood vessels, and
ventricular system are supporting structures present
a multiple levels.
Correspondence to: Jasper R. Daube, M.D., E8B, Department of Neurology, Mayo Clinic, 200 First Street SW,
Tel.: 1-507-284-4409; fax: 1-507-284-4074.
E-mail: daube.jasper@mayo.edu (J.R. Daube).
3.1.1. Meningeal coverings

The meninges are an important supporting elements
of the central nervous system and include the dura
45
Fig. 1. Anterior (left) and posterior (right) views illustrating major bones of the skull. Hollow sinus cavities are located
within frontal, ethmoid, sphenoid, and maxillary bones.
Fig. 2. Base of the cranial cavity viewed from above, illustrating the major cranial fossae, bones of the base of the skull,
and the foramina.
46
J.R. DAUBE
Fig. 3. Reflections of the dural mater forming the falx cerebri and the tentorium cerebelli (top). Structures located above
the tentorium are supratentorial; those below the tentorium but above the foramen magnum are part of the posterior fossa
(bottom).
mater, arachnoid, and pia mater. The outermost fibrous

membrane, the dura mater, consists of two layers of
connective tissue that are fused, except in certain
regions where they separate to form the intracranial
venous sinuses. The dura mater is folded into the cranial cavity in two areas to form distinct fibrous barriers: the falx cerebri, located between the two
cerebral hemispheres, and the tentorium cerebelli,
which demarcates the superior limit of the posterior
fossa. The delicate, filamentous arachnoid lies beneath
the dura mater and appears to be loosely applied to the
surface of the brain. Many of the major arteries can be
seen on the surface of the brain beneath the arachnoid.
The innermost layer, the pia mater, is composed of a
very thin layer of tissue that is so closely attached to

the brain surface that it cannot be seen.
Several spaces are found in association with the
meninges (Fig. 10). Between the bone and the dura
mater is the epidural space, and beneath the dura mater
is the subdural space. The bone, dura, and arachnoid
are normally closely applied to one another so that
the epidural and subdural spaces are potential spaces
in which blood or pus may accumulate. Beneath the
arachnoid is the subarachnoid space, which surrounds
the entire brain and spinal cord and is filled with cerebrospinal fluid (CSF). The subarachnoid space communicates with the interior of the brain via the
ventricular system (Figs. 11 and 12).
47
Fig. 4. Medial A: and lateral B: views of the brain and spinal cord illustrating the major levels: supratentorial (dark shading),
posterior fossa with brainstem (lines) and cerebellum (dots), and spinal (clear area). The peripheral nerves are not shown.
3.1.2. Ventricular system

Located within the depth of the brain is the ventricular
system (Fig. 12). The ventricular system is lined with
ciliated cuboidal epithelium called the ependyma.
Fig. 5. Cerebellum as viewed from its dorsal surface.
Choroid plexuses are found in the lateral, third, and

fourth ventricles. They are multitufted vascular
organs that arise by the ependyma, leptomeninges,
and blood vessels folding into the ventricles. These
structures, rich in the enzymes that are found in other
secretory organs, are the main source for the production of the CSF. A lateral ventricle (Fig. 11) is located
in each of the cerebral hemispheres and is divided into
an anterior horn located in the frontal lobe, body and
atrium located in the parietal lobe, posterior horn in
the occipital lobe, and inferior horn in the temporal
lobe. CSF is formed within the ventricles by the choroid plexus and circulates throughout the ventricles
and subarachnoid space. The lateral ventricles communicate with each other and the third ventricle of
the diencephalon via the interventricular foramina of
Monro. The aqueduct of Sylvius leads from the third
ventricle to the fourth ventricle, located dorsal to the
48
J.R. DAUBE
Fig. 6. Ventral aspect of the brainstem and cranial nerves. Cranial nerve I is not shown. It ends at the olfactory bulb, from
which the olfactory tract arises. Cranial nerves I and II arise above the tentorium; cranial nerves III through XII arise in the
posterior fossa.
pons and medulla. The communication between the

ventricular system and the subarachnoid space occurs
in the fourth ventricle via two foramina of Luschka
and the foramen of Magendie.
CSF is reabsorbed through the arachnoid granulations into the superior sagittal sinus. The rate of CSF
formation remains relatively constant at approximately
0.35 ml/min (500 ml/day). Most of the CSF is actively
secreted into the ventricular system by the choroid plexuses; some is also derived directly from the interstitial
fluid of the brain and crosses the ependyma to enter
the ventricles. Additional exchange may take place
between the neural tissue and the subarachnoid space
across the pia mater.
Circulation of the CSF is pulsatile and promoted
by the beating of the cilia of the ependymal cells
and the pulsatile changes in the volume of intracranial blood that occurs with cardiac systole and respiratory movements. The to-and-fro movement of CSF
results in a directional flow, that is, from the lateral,
third, and fourth ventricles to the subarachnoid space,

where it then circulates in two major directions
(Fig. 13). The more important pathway is rostrally
and dorsally toward the intracranial venous system,
where CSF exits through arachnoid villi that project
into the superior sagittal sinus. CSF also drains into
the spinal subarachnoid space with some resorption
through the dural sleeves of the nerve roots (Fig. 14).
3.1.3. Vascular system blood vessels
Blood enters the skull via two arterial systems
(Fig. 15). The brain is supplied posteriorly by the
vertebrobasilar system and anteriorly by the carotid
arteries. A series of anastamotic channels lying at
the base of the brain, known as the circle of Willis,
permits communication between these two systems.
The internal carotid artery and its major branches,
the anterior cerebral and middle cerebral arteries,
can be seen at the base of the brain (Fig. 16). The
Fig. 7. Structures at the spinal level include the spinal

cord, nerve roots contained in the vertebral column, and
the vertebral column itself.
anterior cerebral arteries are connected to each other

by the small anterior communicating artery and continue in the midline between the two hemispheres to
supply blood to their medial surfaces. The middle
cerebral artery courses laterally between the temporal
and frontal lobes and emerges from the insula
between the frontal and temporal lobes, where its
branches spread over and supply blood to the lateral
surface of the hemisphere (Fig. 17).
49
Additional blood is carried to the brain by the two

vertebral arteries, which enter the skull via the foramen
magnum and join at the caudal border of the pons to
form the basilar artery (Fig. 15). Branches from these
arteries normally provide the sole arterial supply to
the occipital lobe, undersurface of the temporal lobe,
thalamus, midbrain, pons, cerebellum, medulla, and
portions of the cervical spinal cord. The basilar artery
continues to cephalad until it divides into the posterior
cerebral arteries. The posterior communicating
arteries usually arise as branches of the posterior cerebral arteries and join those vessels with the internal
carotid arteries to complete the circle of Willis.
Blood leaves the head by way of veins (Fig. 18)
that course over the cerebral hemispheres to converge into large channels, the venous sinuses,
contained within the layers of the dura mater. The
most prominent of these sinuses are the superior sagittal sinus and inferior sagittal sinus, which run longitudinally from front to back in the falx cerebri
between the hemispheres. The major venous channels
merge in the occipital region and form the transverse
and sigmoid sinuses, which exit through the skull via
the jugular foramen as the internal jugular veins.
Each organ in the body must have blood vessels to
provide a relatively constant supply of oxygen and
other nutrients and to remove metabolic waste. Vascular structures are found at all levels of the nervous
system and include the arteries, arterioles, capillaries,
veins, and dural sinuses. These supply supratentorial,
posterior fossa, spinal, and peripheral structures.
All of the arteries that supply the supratentorial
and posterior fossa structures arise from the aortic
arch (Fig. 19). The innominate artery divides into
the right common carotid and the right subclavian
arteries. The left common carotid artery arises
directly from the apex of the aortic arch. The right
and left common carotid arteries ascend in the neck
lateral to the trachea. Slightly below the angle of
the jaw, the common carotid artery bifurcates into
the internal and external carotid arteries. The internal
carotid artery on each side enters the skull, without
branching, through the carotid canal located in the
petrous portion of the temporal bone.
The petrous segment of the internal carotid artery
enters the cranium via foramen lacerum. After entering the cranium, each internal carotid artery forms an
S-shaped curve, the carotid siphon, and lies within
the cavernous sinus. As the artery leaves the cavernous sinus, it pierces the cranial dura and arachnoid to
enter the subarachnoid space at the base of the brain.
50
J.R. DAUBE
Fig. 8. Cross-sections of the vertebral column at cervical and lumbar vertebral levels showing the exit of spinal nerves from
the spinal cord to the periphery.
Just distal to the cavernous segment within the subarachnoid space, the internal carotid artery gives rise
to the ophthalmic artery, which is an important anastamotic communication with branches of the external carotid artery. The internal carotid artery also
gives rise to the posterior communicating and anterior choroidal arteries (Fig. 15). The posterior communicating artery connects the internal carotid
artery to the posterior cerebral artery. The anterior
choroidal artery supplies the ipsilateral internal capsule and portion of the basal ganglia. At the carotid
terminus, each internal carotid artery then divides
into an anterior cerebral artery and a middle cerebral
artery.
The vertebral arteries arise as the first branches of
the right and left subclavian arteries. Each artery
ascends through foramina in the transverse processes
of the upper six cervical vertebrae and enters the subarachnoid space at the level of the upper cervical
spinal cord. The vertebral arteries enter the cranial
cavity through the foramen magnum. The two vertebral arteries are often of unequal caliber, one being
dominant.
The vertebral arteries enter the cranium and
ascend on the ventrolateral surface of the medulla
oblongata. At the lower border of the pons, they unite
to form the basilar artery (Fig. 16). Major branches
from the basilar artery includes: anterior inferior cerebellar artery, superior cerebellar artery, and multiple
median and paramedian perforators. At the level of
the midbrain, the basilar artery divides into the right
and left posterior cerebral arteries.
At the base of the brain, surrounding the optic

chiasm and pituitary stalk, anastamotic connections
occur between the internal carotid and vertebrobasilar arterial systems. This ring-like series of vessels
is called the circle of Willis (Fig. 15) and consists
of the anterior communicating artery, which unites
the two anterior cerebral arteries, and the posterior
communicating arteries, which join the internal
carotid arteries with the posterior cerebral arteries.
The brain is provided with blood from the anterior, middle, and posterior cerebral arteries (Fig. 17).
The anterior cerebral artery supplies the medial surface of the cerebrum and the superior border of the
frontal and parietal lobes. The middle cerebral artery
supplies most of the lateral surface of the cerebral
hemispheres, including the lateral portions of the
frontal lobe, the superior and lateral portions of the
temporal lobes, and the deep structures of the frontal
and parietal lobes. The posterior cerebral artery
supplies the entire occipital lobe and the inferior
and medial portions of the temporal lobe. The deeper
structures of the cerebral hemispheres are supplied by
penetrating branches of the larger arteries. Of notable
importance are the perforating lenticulostriate
arteries, which supply the basal ganglia and internal
capsule, and the perforating branches of the posterior
cerebral artery, which supply the thalamus.
The venous drainage of the brain is divided into
superficial and deep systems. The cerebral cortex
and outer half of the white matter drain into the
superficial system of veins located over the convexity of the brain in the subarachnoid space. The
51
Fig. 9. Dorsal view of the spinal level. The spinal cord terminates between vertebrae L-1 and L-2 and is enlarged at the
cervical and lumbosacral levels. These enlargements correspond to the segments that innervate the upper and lower limbs.
The roots form the spinal nerves that exit through the intervertebral foramina. The cervical roots exit above the
corresponding vertebra, and the eighth cervical root exits between vertebrae C-7 and T-1. The rest of the roots exit below
the corresponding vertebra. Because of the difference in length between the spinal cord and the spinal canal, the lumbar and
sacral roots of the conus medullaris travel a relatively long distance in the subarachnoid space before exiting through their
corresponding foramen. These roots form the cauda equina.
52
J.R. DAUBE
Fig. 10. Meninges and meningeal spaces. Coronal section through the paramedian region of the cerebral hemispheres.
superficial veins of the superior half of the brain drain

into the superior sagittal sinus, those from the inferior
half drain into the lateral sinuses. The deep white matter and deep nuclei of the brain drain into the deep
venous system, which includes the great cerebral vein

of Galen, inferior sagittal sinus, and straight sinus.
From these venous channels, blood empties into the
transverse sinuses, the sigmoid sinuses, and ultimately
Fig. 11. Ventricular system. A: Lateral view. B: Anterior view. The pons has been removed from the illustration to display
the anatomy of the fourth ventricle.
53
Fig. 12. Detailed relationships of the meninges and structures in the subarachnoid space. An arteriole carries pia mater into
the cerebral cortex.
the jugular veins (Fig. 18). Veins on the inferior surfaces of the cerebrum terminate directly or indirectly
in the cavernous sinus, an important dural structure
located on either side of the pituitary fossa containing
the carotid artery and cranial nerves III, IV, V, and VI.
The spinal cord is supplied with arterial blood by one
anterior and two posterolateral vessels that run along the
length of the cord and by an irregular plexus of segmentally arranged vessels that encircle the cord and interconnect the major vessels (Fig. 19). The anterior
spinal artery is a single vessel lying in the ventral
median fissure. It arises from a pair of small branches
of the vertebral arteries that fuse along the caudal
medulla and descend along the cervical spinal cord. A
series of six to eight ventral radicular arteries arising
from the intercostal, lumbar, and sacral arteries connect
with the anterior spinal artery at various levels along the
length of the spinal cord. The largest of these radicular
arteries enters at the low thoracic or upper lumbar
region. Because of this uneven blood supply, the spinal
cord is most vulnerable to ischemia at the mid-thoracic
and upper lumbar levels, as shown by the stippled areas
in Fig. 19. The posterior spinal arteries are paired
structures that run along the posterolateral aspect of

the cord near the dorsal roots. They receive contributions from the posterior radicular arteries and supply
the dorsal funiculus and dorsal gray horns. The spinal
cord is drained by an anastamotic venous plexus surrounding the dural sac. Veins drain outward along both
the dorsal and ventral roots into this plexus, which has
numerous connections with the veins of the thoracic,
abdominal, and pelvic cavities.
All neural structures must receive adequate arterial
blood supply in order to sustain life and to maintain their
integrity. Axons traveling to the periphery are gathered
into bundles or fascicles that have a connective tissue
covering. Within this covering, along the entire course
of the nerve, is a rich and highly anastamotic plexus of
small arterioles derived from the branches of the major
extremity vessels (Fig. 20). This dense anastomosis renders the peripheral nerve relatively immune to ischemic
vascular disease. Such abnormality, when noted in
peripheral neural structures, is usually associated with
either direct compression of a nerve or with multiple
segmental vascular lesions from small vessel arterial
disease.
54
J.R. DAUBE
Fig. 13. Major subarachnoid spaces, third and fourth ventricles, subarachnoid cisterns, and meningeal layers. CSF is produced primarily in the choroid plexus, and circulates from the lateral ventricles through the foramen of Monro into the third
ventricle, then out the aqueduct of Sylvius into the fourth ventricle. CSF flows out of the fourth ventricle through the foramina of Luschka and Magendie into the posterior fossa cisterns. Flow then proceeds both rostrally over the hemispheres and
caudally into the subdural spaces in the spine. Most of the CSF is reabsorbed through the arachnoid granulations into the
superior sagittal sinus.
3.1.4. Major neural structures

The major components of the central nervous system
are shown in Fig. 4. The cerebral hemispheres make
up the major portion of the brain. The longitudinal
(interhemispheric) fissure separates the cerebrum
into two cerebral hemispheres. The surface of each
hemisphere is convoluted into folds known as gyri

separated from one another by grooves, or sulci.
As shown in Fig. 21, deeper grooves known as fissures separate the four major lobes. The central sulcus separates the precentral gyrus of the frontal
lobe with motor function from the post-central gyrus
of the parietal lobe with sensory function. The fissure
55
Fig. 14. Meningeal relationships at the spinal level. Note investments of the spinal nerve root by dura mater as it leaves the
spinal canal. The denticulate ligaments attach the spinal cord to the dura mater laterally.
Fig. 15. Arterial supply to the brain. The major arterial supply is via the internal carotid and vertebrobasilar systems, which
communicate with each other via the posterior communicating arteries that complete the circle of Willis.
56
J.R. DAUBE
Fig. 16. Arterial supply as viewed from the base of the brain with a portion of the temporal lobe and left cerebellar hemisphere removed to show the middle cerebral artery and base of the occipital lobe.
of Sylvius demarcates the temporal lobe from the

frontal and parietal lobes. The corpus callosum
on the medial surface of the hemisphere is a prominent fiber tract transferring information from one
hemisphere to another. The remaining prominent
structures on the medial surface process memory

and emotion and are part of the limbic lobe.
The diencephalon represents a zone of transition
between the cerebral hemisphere at the supratentorial
level and the structures in the posterior fossa. The
Fig. 17. Lateral and medial surfaces of the cerebral hemisphere illustrating the distribution of the major arteries. The
anterior and middle cerebral arteries are branches of internal carotid arteries; posterior cerebral arteries are branches of
the basilar artery.
57
Fig. 18. Venous drainage of the cerebral hemispheres. Blood circulating over the cerebral cortex collects in the superior
sagittal sinus (left); blood from deeper structures enters other venous sinuses (right). The direction of flow is toward the
confluence of sinuses in the occipital region and then toward the internal jugular veins by way of the transverse and sigmoid
sinuses (S, sinus; V, vein).
diencephalon consists of the third ventricle and those

structures related to it, including the thalamus, hypothalamus, optic pathways, and pineal body. At the
base of the hypothalamus is an important neuroendocrine structure, the pituitary gland. It is located in the
middle of the skull in the bony sella turcica.
Cranial nerves I and II arise from the cerebral hemispheres. Cranial nerve I, the olfactory nerves, are
located at the base of each frontal lobe. Cranial nerve
II, the optic pathway, connects the eye to the optic
nerve, optic chiasm, and optic tract. The major structures contained in the posterior fossa are the brainstem,
the cerebellum, and cranial nerves III through XII. The
cerebellum consists of two hemispheres and a midline
vermis (Fig. 4). The cerebellum lies dorsal to the
fourth ventricle, the pons, and the medulla.
The brainstem consists of the portion of the brain
that remains after removal of the cerebral and cerebellar hemispheres (Fig. 4). The brainstem includes
the medulla, pons, and midbrain connected to the
spinal cord. The pons is separated from the midbrain
and medulla is made by a large bundle of prominent
crossing fibers on the ventral and lateral surfaces of
the brainstem that connect the brainstem to the cerebellum (Fig. 6). Emerging from the brainstem are the
remaining 10 pairs of cranial nerves.
The major structure in the spine is the spinal cord.

The spinal cord is surrounded by meninges similar to
those that surround the brain. The adult spinal cord
begins at the caudal margin of the medulla at the level
of the foramen magnum and terminates opposite the
caudal margin of the first lumbar vertebra. Throughout
much of the length of the cord, a spinal segment is not
adjacent to its corresponding vertebral segment.
Thirty-one pairs of spinal nerves are attached to the
spinal cord via dorsal (posterior) and ventral (anterior)
nerve roots: 8 cervical, 12 thoracic, 5 lumbar, 5 sacral,
and 1 coccygeal spinal nerve on each side. The collection of spinal nerve roots contained in the lumbosacral
spinal canal is known as the cauda equina.
The major peripheral structures are the somatic
nerves, the autonomic nerves and ganglia, the neuromuscular junction, the muscles, and the peripheral sensory receptors. Spinal nerves are formed by the joining
of dorsal and ventral roots and thus contain motor
and sensory nerve fibers. Fibers en route to the limbs
come together and are rearranged in plexuses. The
brachial plexus, located in the axillary region, redistributes the fibers to the major nerves of the upper arms:
median, ulnar, radial, axillary, and musculocutaneous
The lumbosacral plexus, located in the lower abdominal
cavity and pelvis, redistributes the fibers to the major
58
nerves in the lower legs, femoral, obturator, and sciatic,

which divides into the tibial and peroneal nerves.
The major functions of the nervous system are
organized into separate longitudinal systems carrying
information between the brain, spinal cord, nerves,
and muscles (Fig. 22). The motor system is efferent,
controlling movement of the muscles; the sensory
system in afferent, carrying sensation from the body
to the brain; the autonomic pathways are afferent
J.R. DAUBE
and efferent, controlling automatic functions like

blood flow, blood pressure, bladder and bowel function, sweating, and many other bodily activities.
3.2. The motor system
The motor system initiates and controls activity in the
somatic muscles; the functions of the motor system are
control of posture and movement. Components of this
(Fig. 19 continued)
59
Fig. 19. A: Anterior spinal artery. Radicular arteries are variable in location but are shown here at C-3, C-5, T-5, T-10,
L-2, and S-1. Stippled areas indicate zones of marginal blood supply at T-4 and L-1. B: Posterior spinal arteries supply
blood to the posterior aspect of the cord. They receive blood from radicular arteries at multiple levels.
system include motor cortex and other areas of the

frontal lobes; descending pathways that traverse the
internal capsule, cerebral peduncles, medullary pyramids, and other areas of the brainstem; portions of
the spinal cord, including the ventral horns; ventral
roots; efferent fibers traveling in both peripheral and
cranial nerves; and muscle, the major effector organ

of the motor system. Also included in the motor system
are the cerebellum and basal ganglia (Fig. 23). Thus,
the motor system is present throughout the neuraxis
and is directly involved in the performance of all motor
activity mediated by voluntary muscles. Weakness,
60
J.R. DAUBE
Fig. 20. Blood supply to peripheral nerve. Multiple anastamotic channels are derived from regional arteries.
paralysis, twitching, jerking, staggering, wasting,

shaking, stiffness, spasticity, and incoordination
involving the arms, legs, eyes, or muscles of speech
are all due to impairment of the motor system.
The motor pathway has three major components,
the corticospinal tract from the cortex to the spinal
cord, brainstem motor pathways from the brainstem
to the spinal cord, and the final common pathway
from motor neurons in the ventral horn of the spinal
cord (and motor nuclei of the brainstem) to the muscles located in the periphery. The motor neurons
receive segmental inputs from the limbs via primary
afferents and descending inputs from supraspinal
structures. Segmental and supraspinal inputs affect
motor neuron activity either directly or, more commonly, via interneurons. The descending pathways
originate from motor areas of the cerebral cortex
and the brainstem. Cortical motor neurons give rise
to the corticospinal tract, which controls fine movements of the distal portions of the limbs, particularly
the fingers. The descending pathways from the brainstem primarily control postural and reflex movements. The activity of these pathways is regulated
by two circuits centered in the basal ganglia and the
cerebellum. The basal ganglia are important in the
selection and initiation of specific motor programs.

The cerebellum controls the execution of the motor
acts and motor learning (Fig. 23). The cortical motor
areas project to both the basal ganglia and the cerebellum. Both of these structures project back to the
cortex via a relay in the motor nuclei of the thalamus.
3.2.1. Final common pathway
The somatic, or skeletal, muscles that perform the
work of moving parts of the body are all under direct
control of lower motor neurons and contract only in
response to activation by these neurons. Each muscle
performs a particular movement, but each muscle
may be involved in several different motor activities
organized by the lower motor neurons and their pools
of interneurons. Alpha motor neurons and their surrounding interneuronal pool integrate activity from
central and peripheral sources and transmit action
potentials to the muscles to produce the appropriate
level of contraction. Disease processes that impair
the function of a motor unit prevent the normal activation of muscle fibers in that motor unit. This is
manifested as an inability of the muscle to contract
fully (weakness or paralysis).
61
Fig. 21. The five lobes of the brain frontal, parietal, temporal, occipital, and limbic. A: Lateral view. B: Medial view.
The final common pathway is the effector mechanism by which all motor activity is mediated
(Fig. 24). It includes the motor neurons in the ventral
horn of the spinal cord and brainstem and their axons
that extend peripherally via nerves to innervate muscles. These motor neurons are called alpha motor
neurons. The axons from the motor neurons innervate
the muscle fibers that are responsible for skeletal
muscle contraction. This combination of alpha motor
neuron, peripheral axon, and all the muscle fibers
innervated by them is the motor unit, the basic functional component of the final common pathway.
The nerve terminals of a single motor axon innervate muscle fibers that may be distributed widely
throughout the muscle, intermingling with muscle

fibers innervated by other neurons. A muscle may
contain from 50 to 2,000 motor units. The size of
a motor unit is determined by the number of fibers
innervated by a single motor neuron, expressed as
the innervation ratio, the number of muscle fibers
per axon. The motor units of the powerful limb muscles each contain from 500 to 2,000 muscle fibers. In
contrast, motor units in intrinsic hand muscles have
innervation ratios of only 50400.
Diseases may affect the final common pathway at
the level of the ventral horn cell, the axon, or the
muscle fiber. Damage to any of these sites produces
weakness, atrophy, loss of reflexes, and loss of tone.
62
J.R. DAUBE
Fig. 22. Major nervous system connections. Long intersegmental pathways leading to and from higher centers and multiple,
short segmental pathways (cranial and peripheral nerves) to the periphery. Summary of the functions associated with the
major anatomical levels is listed on the right.
3.2.2. Corticospinal tract

The largest, best-defined motor pathway is a singleneuron pathway that extends from the cerebral cortex
to the spinal cord called the corticospinal tract. This
pathway provides a direct route by which information
can travel from the cerebral cortex to the brainstem
and spinal cord without an intervening synapse
(Fig. 25). Its major function is to effect voluntary
activity, in particular, skilled movements under conscious control. The corticospinal tract descends from
the cerebral cortex through the white matter of the
cerebral hemispheres, the pyramids in the brainstem,
and the spinal cord to end on ventral horn cells. The
axons cross the midline at the junction of the brainstem and spinal cord to end on the opposite side.
The effectiveness of this pathway depends on an
intact final common pathway to carry information
to the muscles.
The corticospinal tract is the route by which the
motor areas of the cerebral cortex in each hemisphere
control motor neurons in the ventral horn on the
opposite side of the spinal cord and in the motor

nuclei of the brainstem. The fibers in the corticospinal tract are corticospinal and corticobulbar. Those
traveling to the spinal cord are called corticospinal,
or pyramidal, tract. Those ending on brainstem nuclei
are corticobulbar fibers. The neurons from which
these tracts arise are known as upper motor neurons.
The major function of the corticospinal pathway is to
initiate and control skilled voluntary activity.
Each corticospinal tract arises primarily from cells
in the cortex of the frontal lobe of one hemisphere
and descends through the corona radiata into the internal capsule. The tract passes from the internal capsule
via the cerebral peduncles to the base of the brainstem,
where it forms the medullary pyramids. At the junction
between the medulla and spinal cord, most of the fibers
in each pyramid cross the midline (the corticospinal
decussation) to lie in the lateral funiculus of the opposite half of the spinal cord. These crossed fibers form
the lateral corticospinal tract of the cord.
The corticospinal tract is formed by axons
of neurons located in the primary motor cortex

Fig. 23. A: Outline of the motor system. B: Basic connections of the motor system. The motor neurons of the ventral horn of the spinal cord and of the motor nuclei in
the brain stem are the final common pathway for muscle control. They receive input from the contralateral motor cortex via the corticospinal tract and the corticobulbar
tract and from several brainstem nuclei. The cerebellum controls the ipsilateral limbs via connections with the spinal cord, brainstem, and contralateral motor cortex
through the thalamus. The basal ganglia participate in motor planning via reciprocal connections with ipsilateral motor cortex (contralateral to the limb).
63
64
Fig. 24. A single motor unit and its components: the lower
motor neuron and muscle fibers innervated by it. The final
common pathway contains hundreds of thousands of such
units innervating skeletal muscle.
(area 4) (Fig. 26), the lateral premotor cortex

(area 6a), the supplementary motor area (or medial
premotor cortex, area 6b), and the anterior cingulate
motor area (on the medial surface of the hemisphere).
All these areas are closely interconnected and project
to the ventral horn. The primary motor cortex occupies the anterior lip of the central sulcus of Rolando
and the adjacent precentral gyrus (area 4). The primary motor cortex integrates input from multiple
sources and has a somatotopic organization, with
the contralateral body represented upside down: the
head area is located above the fissure of Sylvius,
the upper extremity next (with the thumb and index
finger in proximity to the face), the trunk interposed
between the shoulder and hip areas high on the convexity, and the lower limb representation extending
onto the paracentral lobule in the longitudinal fissure.
The frontal eye field (area 8) contains neurons
involved in the generation of spontaneous and visually guided rapid eye movements. Brocas area is
immediately ventral to the motor area of the left cerebral hemisphere near where the face is represented.
Neurons in Brocas area participate in the motor programming necessary for speech.
Axons from the motor cortex converge in the
corona radiata toward the internal capsule, where
they are compactly gathered in a topographic localization. The corticobulbar fibers occupy a more anterior location in the posterior limb of the internal
capsule than the corticospinal fibers.
J.R. DAUBE
The pyramidal fibers remain grouped together as

they pass from the internal capsule to the cerebral
peduncle in the midbrain. In the midbrain, the corticospinal and corticobulbar fibers occupy the middle
two-thirds of the cerebral peduncle, with the corticobulbar fibers being more medial. During their course
in the brainstem, the corticobulbar fibers leave the
pyramidal pathway at several levels, some crossing
the midline and some remaining uncrossed. These
fibers synapse in the motor centers and nuclei of
the cranial nerves trigeminal, facial, vagus, spinal
accessory, and hypoglossal. The fibers in the medulla
form the medullary pyramids. At the lower border of
the medulla, the main pyramidal decussation occurs
with about 80% of the fibers crossing to the opposite
side of the spinal cord.
In the spinal cord, the crossed pyramidal fibers
occupy the lateral column (the lateral corticospinal
tract). Because of the decussation of most of the
fibers of the pyramidal tracts, the voluntary movements of one side of the body are under the control
of the opposite cerebral hemisphere.
More diffuse, extrapyramidal pathways act indirectly on the final common pathway, mediating the
enormous number of automatic activities involved
in normal motor function. For example, the maintenance of erect posture when sitting or standing
requires the coordinated contraction of many muscles. This coordination is under subconscious control
and is mediated by the reticulospinal, vestibulospinal,
and rubrospinal tracts.
Damage to the motor pathways result in characteristic clinical patterns. There is weakness or paralysis
of muscles, especially the distal muscles. The
impairment is greatest for fine movements, skilled
movements, and movements under voluntary control.
The distribution of the weakness is a function of the
site of the lesion. If the lesion is localized in a limited
area of cortex, then a single limb or one side of the
face only may be involved. If the lesion involves
only the pyramidal tract fibers in the pyramids of
the medulla, one side of the body below the level
of the lesion is affected. Spasticity and hyperreflexia
are the result of the loss of activity of inhibitory
interneurons (reticulospinal, the lateral vestibulospinal, and pontine reticulospinal). A lesion at any
corticospinal level produces the upper motor neuron
syndrome of distal weakness, loss of cutaneous
reflexes, and Babinskis sign with increased muscle
tone and reflexes.
65
Fig. 25. Corticospinal tract. The course of the fibers in the tract is shown descending through the cerebral hemispheres,
brainstem, and spinal cord. Some of the axons in the tract extend the entire length of the spinal cord as shown schematically
for a single neuron on the left.
66
J.R. DAUBE
Fig. 26. Motor areas of the cerebral cortex. A: Lateral view of the cerebral hemisphere showing the primary motor area
(area 4), lateral premotor area (lateral area 6), frontal eye fields (area 8), and Brocas area (area 44). B: Medial view of the
cerebral hemisphere showing the supplementary motor area (medial area 6) and the anterior cingulate motor area. The
numerals in parentheses refer to Brodmanns areas.
3.2.3. Control pathways

Two parallel pathways, the cerebellar and the basal
ganglia pathways, control and modify motor activity.
The cerebellum and basal ganglia both receive input
from several motor and sensory cortical areas and
send information back to the cortex through different
nuclei of the thalamus.
These systems are organized into several parallel
loops: cerebral cortexbasal gangliathalamuscerebral
cortex and cerebral cortexcerebellumthalamus
cerebral cortex. They integrate and modulate motor
activity primarily through the cerebral cortex and corticospinal tracts. However, the cerebellum and basal
ganglia also send information to the brainstem and
the extrapyramidal pathways.
The functions and connections of the basal ganglia
and cerebellar control circuits are different despite
the general features they have in common. The basal
ganglia are concerned with selective activation and
inhibition of specific motor programs necessary for
automatic performance of learned movements and

postural adaptations. The cerebellum is involved in
the control of the execution of motor acts, including
maintenance of balance and posture, planning and
execution of coordinated limb movements, adjustments of motor performance, and learning of new
motor tasks. Abnormalities of the control circuits
result in disorders of posture and coordination, at
times accompanied by tremor or other abnormal
involuntary movements. Control circuit damage does
not produce weakness.
The basal ganglia are concerned primarily with
learned, automatic behavior and with preparing and
maintaining the background support, or posture, needed
for voluntary motor activity. Components of the basal
ganglia include the striatum, globus pallidus, subthalamic nucleus, and substantia nigra. The striatum is the
receptive component of the basal ganglia and receives
input from the cerebral cortex (Fig. 27). The striatum
receives three main inputs: the cerebral cortex, the intralaminar thalamic nuclei, and the substantia nigra pars
Fig. 27. The main nuclei and connections of the basal ganglia circuit. The basal ganglia include the striatum (putamen,
caudate nucleus, and the accumbens nucleus [not shown]),
globus pallidus (including external and internal segments),
subthalamic nucleus, and substantia nigra. The striatum
(especially the putamen) is the receptive component of the
basal ganglia control circuit and it receives excitatory input
from the motor cortex. Neurons of the striatum contain
gamma amino butyric acid and project to both segments of
the globus pallidus and the substantia nigra. The internal segment of the globus pallidus contains neurons that constitute
the output of the basal ganglia. The main target is the ventral
anterior and other thalamic nuclei that project to the supplementary motor cortex and other areas of the frontal lobe. The
external segment also contains neurons that project to the
internal segment and the subthalamic nucleus. The subthalamic nucleus sends an excitatory projection to the internal
segment of the globus pallidus. The substantia nigra pars
compacta contains dopaminergic neurons that project to the
striatum. The basal ganglia circuits consist of extrinsic and
intrinsic connections. The extrinsic connections include
input from the cerebral cortex and substantia nigra pars compacta to the striatum (and subthalamic nucleus) and output
from the internal segment of the globus pallidus to the thalamus and brainstem motor nuclei. The intrinsic connections
include projections from the striatum to both segments of
the globus pallidus, reciprocal connections between the substantia nigra and the striatum, and reciprocal connections
between the subthalamic nucleus and the globus pallidus.
compacta. The most important of output projection

is to the thalamus nuclei that relay information to
the premotor and supplementary motor areas and
to other regions of the frontal lobe. The output of
67
the basal ganglia affects both the corticospinal and

the brainstem motor pathways.
The cerebellum accomplishes the coordination and
correction of movement errors of muscles during
active movements. The cerebellum and its connections
compose the second major control circuit. This control
is concerned with the planning and execution of movements, adaptation of motor performance, and motor
learning. Its functions include control of posture, balance, and eye movements necessary for maintaining
equilibrium; adjustment of ongoing execution of
movement; initiation, timing, and planning of coordinated limb movements; and learning new motor tasks.
Cerebellar inputs and outputs are side loops of pathways from the motor cortex, subcortical nuclei, and spinal
cord. Information about motor plans is provided to the
cerebellum by collateral projections from premotor cortex and motor cortex (relayed to the cerebellum via the
pontine nuclei) and from brainstem motor regions. Information about motor performance, or external feedback, is
provided by inputs from peripheral receptors via the
dorsal spinocerebellar tract. The cerebellum is subdivided
into flocculonodular lobe and the body of the cerebellum,
which includes the anterior lobe and the posterior lobe.
The midline portion of the anterior and posterior lobe is
called the vermis, and the lateral portions are the cerebellar hemispheres (Fig. 5). Nerve fibers enter or leave the
cerebellum in the three cerebellar peduncles. The inferior
cerebellar peduncle (or restiform body) connects the cerebellum with the medulla and spinal cord; the middle cerebellar peduncle (or brachium pontis) connects the
cerebellum with the pons; and the superior cerebellar
peduncle (or brachium conjunctivum) connects the cerebellum with the midbrain and cerebral hemispheres.
The cerebellum corrects motor performance
through its output to brainstem nuclei and to premotor cortex and motor cortex. The cerebellar output to
the motor cortex is relayed through the ventral lateral
nucleus of the thalamus. The cerebellum controls the
ipsilateral limbs. Therefore, it processes input from
the ipsilateral spinal cord and vestibular nuclei and
the contralateral cerebral hemisphere and red
nucleus. The cerebellar projections to the motor cortex and red nucleus travel in the superior cerebellar
peduncles, which decussate in the midbrain. Descending input from the cerebral hemispheres provides
information to the contralateral cerebellum via the
pontine nuclei; the crossed pontocerebellar axons
form the entire middle cerebellar peduncle. Projections from the dentate nucleus to the thalamus close
an important feedback loop between the cerebral
68
J.R. DAUBE
cortex and the cerebellum: the corticocerebellar

dentothalamocortical loop. This loop is thought to be
critical for the initiation, planning, and timing of motor
acts, including specification of the direction, pattern,
and intensity of movement of the upper extremity. The
posterior lobes, particularly the large lateral hemispheres, form a servomechanism for coordination of
skilled action. Lesions of the posterior lobes produce
irregular movements of the limbs, loss of muscle coordination, loss of ability to measure range of motion,
irregularity in alternate motion rate, and tremor with
voluntary activity (intention tremor). All these manifestations are ipsilateral to the side of the lesion.
3.3. Sensory system

The primary somatosensory cortex is located in the
post-central gyrus of the parietal lobe and is concerned
with discriminative aspects of reception and appreciation of somatic sensory impulses. It consists of at least
four functionally distinct areas, each containing a
complete somatotopic map. Fibers terminate in the
postcentral gyrus in an organized fashion, with the
lower extremity represented on the medial surface of
the hemisphere, and the arm and hand represented on
the lateral surface. The face, mouth, and tongue are
represented in the suprasylvian region (Fig. 28).
Fig. 28. Coronal section of cerebral hemisphere showing the distribution of sensory fibers in the postcentral gyrus.
Sensory signals reach the cortex from the peripheral

receptors through peripheral nerves, the spinal cord,
and brainstem via a number of parallel pathways
mediating different functions (Fig. 29).
Somatosensory receptors include cutaneous receptors, joint receptors, and muscle receptors. Cutaneous
receptors consist of low-threshold mechanoreceptors,
which are innervated by large myelinated fibers
and transmit touch sensation, and high-threshold
mechanoreceptors, chemoreceptors, and thermoreceptors, which are innervated by small myelinated
or unmyelinated fibers and mediate pain and
69
temperature sensation. Joint and muscle receptors

are mainly innervated by large, rapidly conducting
myelinated fibers. Muscle receptors include muscle
spindles, which signal muscle length and rate of
change in length, Golgi tendon organs, which
respond to changes in muscle tension, and free nerve
endings, which respond to muscle pressure and pain.
Information from somatic receptors is transmitted
to the spinal cord by the first-order neurons. The cell
bodies of these neurons are located in the dorsal root
ganglia. Each of these neurons has a single nerve process that divides into two branches. The distal, or
Fig. 29. Diagram of the pathway for discriminative touch, vibration, and proprioception (thick line) and for pain and
temperature (thin line) of the left arm. The first-order neurons are the large and small dorsal root ganglion neurons.
Large-diameter afferents for touch and proprioception ascend ipsilaterally in the left dorsal column at the cervical level
(fasciculus cuneatus), and synapse on second-order neurons in the lower medulla (nucleus cuneatus). Axons from the
second-order neurons decussate and ascend in the right (contralateral) medial lemniscus to synapse in the ventral posterolateral nucleus of the thalamus, which projects to the primary sensory area in the postcentral gyrus. Small-diameter afferents
for pain and temperature synapse on second-order neurons in the dorsal horn of the spinal cord. Axons of these secondorder neurons decussate in the anterior commissure and ascend, as the spinothalamic tract, in the contralateral ventrolateral
quadrant. This tract joins the medial lemniscus and terminates in the ventral posterolateral nucleus and other nuclei in
the thalamus.
70
peripheral, branch corresponds to the sensory afferent

that innervates the receptor. The proximal, or central,
branch enters the spinal cord via the dorsal root. The
areas of the skin innervated by individual dorsal roots
are called dermatomes. The nerve roots include axons
from peripheral nerves (Fig. 30A) and spinal nerves;
the latter are arranged in a highly ordered way on the
body surface (Fig. 30B) that differs from the peripheral
nerves.
J.R. DAUBE
The two main types of neurons in a dorsal root

ganglion are large neurons, with large myelinated
axons that innervate low-threshold mechanoreceptors
(touch) and proprioceptors, and small neurons, with
small myelinated or unmyelinated axons that
innervate nociceptors, thermoreceptors, and visceral
receptors. Touch fibers are intermediate in size;
these fibers in the dorsal root are separated medial
to lateral as they enter the dorsal horn of the
(Fig. 30 continued)
71
Fig. 30. A: Cutaneous distribution of the major peripheral nerves. B: Cutaneous and dermatomal distribution of spinal nerve
roots. Note considerable overlap between segments, and note that the distribution differs from that of peripheral nerves.
spinal cord (Fig. 31). This subdivision is relevant

clinically because diseases that selectively effect
large sensory fibers or large dorsal root ganglion
neurons produce severe loss of all tactile modalities
and proprioception but leave pain and temperature
sensation intact. Diseases of small sensory fibers
or small dorsal root ganglion neurons affect pain
and temperature but spare touch and proprioceptive
sensation.
The pathways for the different sensory modalities
diverge as they ascend in the spinal cord to higher centers (Fig. 32). The medially located large myelinated
fibers bifurcate into branches that may (1) ascend
directly in the ipsilateral dorsal columns, without synapsing in the spinal cord, as the direct dorsal column
pathway; (2) synapse on dorsal horn neurons that in
turn contribute axons to the dorsal column (the postsynaptic dorsal column pathway), dorsolateral funiculus, and spinothalamic tract; (3) synapse in the
intermediate gray matter on neurons that give rise to
the spinocerebellar tract; (4) synapse on interneurons
and motor neurons in the ventral horn for segmental,
or myotatic, reflexes; and (5) synapse in the dorsal
horn on interneurons that provide segmental modulation of pain transmission. The laterally located small
myelinated and unmyelinated fibers bifurcate into
ascending and descending branches that run longitudinally in Lissauers tract, part of the dorsolateral funiculus. Within several segments, these axons leave
Lissauers tract to enter the dorsal horn and the intermediate gray matter of the spinal cord. In the gray matter, they may (1) synapse on different groups of dorsal
72
J.R. DAUBE
Fig. 31. Dorsal root entry zone. Largest, most heavily myelinated fibers mediating proprioception occupy the medial division. Medium sized myelinated fibers mediating touch are located centrally, and finally myelinated fibers carrying pain and
temperature sensation are located in the lateral division.
horn and intermediate gray matter neurons that form

the spinothalamic and other tracts ascending in the
contralateral ventrolateral quadrant, (2) synapse on
dorsal horn interneurons involved in segmental
modulation of pain and in intrinsic (propriospinal)
intersegmental pathways, and (3) synapse on interneurons and activate somatic and preganglionic
autonomic motor neurons to initiate segmental visceral and somatic reflexes. The second-order spinal
somatosensory neurons occupy the dorsal horn and
Fig. 32. Cross-section of upper cervical spinal cord illustrating the location of the major ascending sensory pathways and
their relationship to the posterior, lateral, and anterior funiculi.
the intermediate gray matter of the spinal cord. These

neurons contribute to all somatosensory pathways
except the direct dorsal column pathway.
Somatosensory pathways can be subdivided into
three groups. First, the direct, contralateral, somatotopically organized pathways for tactile discrimination,
conscious proprioception and discriminative aspects
of pain and temperature that synapse in the ventral posterior complex of the thalamus. Second, the indirect
pathways with poor somatotopy that ascend bilaterally, have multiple interconnections with the reticular
formation and other subcortical regions, relay in midline thalamic nuclei, and affect limbic and paralimbic
cortical areas. These indirect pathways are not helpful
for localization, but they are important for transmission of affective arousal components of pain and visceral sensation and for initiation of reflex somatic,
autonomic, and hormonal responses to external stimuli. Third, the spinocerebellar tracts are two-neuron
pathways that transmit unconscious proprioceptive
information to the ipsilateral cerebellum.
The direct dorsal column pathway is the most
important component of the lemniscal system and consists of large myelinated, primary dorsal root axons
that ascend ipsilaterally to reach the dorsal column
nuclei in the medulla (Fig. 33). This pathway is critical
for spatiotemporal tactile discrimination and fine
motor control. The two major anatomical divisions of
the dorsal columns are the fasciculus gracilis, which
is medial and carries information from the lower extremities and the lower trunk (spinal segment T-7 and
lower), and the fasciculus cuneatus, which is lateral
and carries input from the upper extremities and the
upper trunk (spinal segment T-6 and higher). Cutaneous and proprioceptive inputs terminate in the nucleus
cuneatus of the medulla. Muscle receptor afferents
traveling in the fasciculus cuneatus leave this tract in
the medulla and terminate in the external, or accessory,
cuneate nucleus (analogous to Clarkes nucleus).
Therefore, the dorsal columns are functionally heterogeneous and carry mostly cutaneous and some proprioceptive inputs to the dorsal column nuclei and
proprioceptive input to cerebellar relay nuclei. Lesions
in the dorsal columns at any spinal cord level interfere
with input from rapidly adapting cutaneous mechanoreceptors, but lesions above the thoracic cord largely
spare input from muscle receptors in the lower trunk
and lower extremities.
The second-order neurons of the direct dorsal column pathway are located in the dorsal column nuclei
73
of the lower medulla. They are the nucleus gracilis,

which receives cutaneous inputs from the lower
extremity via the fasciculus gracilis, and the nucleus
cuneatus, which receives cutaneous and some proprioceptive inputs from the upper extremities via the
fasciculus cuneatus. The dorsal column nuclei are
not simple relay stations but are sites of modulation
of sensory transmission critical for sensory discrimination. Second-order axons from the dorsal column
nuclei cross to the opposite side in the lower medulla
as the internal arcuate fibers (the decussation of the
medial lemniscus) and form the medial lemniscus,
which ascends to the thalamus. The medial lemniscus
terminates in the ventral posterolateral nucleus and
other subdivisions of the ventral posterior complex
of the thalamus.
The sensations of pain and temperature are transmitted primarily via the spinothalamic tract, which
ascends in the ventrolateral quadrant of the spinal
cord contralateral to the side of entry of the primary
afferents (Fig. 34). The spinothalamic tract is complex and functionally heterogeneous. It mediates the
discriminative and arousal-emotional components of
pain sensation as well as thermal, visceral, and simple tactile information. The different components of
the spinothalamic tract include (1) a direct pathway,
the neospinothalamic pathway, which mediates the
discriminative aspect of pain and temperature and is
important for localization and (2) several indirect
pathways for the affective-arousal components of
pain; they form part of the core, or inner tube, of
the neuraxis.
The direct tract consists of second-order axons
from both nociceptive-specific and wide dynamic
range neurons. The axons cross through the ventral
white commissure and ascend strictly contralaterally
in the anterolateral quadrant of the spinal cord.
The tract has a somatotopic organization in the spinal
cord, with the sacral dermatomes represented
dorsolaterally and the cervical dermatomes ventromedially. The spinothalamic tract ascends in the
lateral portion of the brainstem. In the medulla, it
is dorsal to the lateral aspect of the inferior olivary
nucleus, and in the pons and midbrain, it is lateral
to the medial lemniscus. At the mesodiencephalic
junction, the spinothalamic tract and medial lemniscus join. Throughout its course, the spinothalamic
tract maintains a somatotopic organization. The
spinothalamic tract axons synapse on third-order
neurons in several thalamic nuclei, particularly
74
J.R. DAUBE
Fig. 33. Dorsal columnlemniscal pathway. Conscious proprioception and discriminative sensation.
the ventral posterolateral nucleus that, in turn, projects to the primary sensory cortex in the postcentral
gyrus.
The spinocerebellar tracts transmit information
about the activity of the effector muscles or motor
neuron pools to the cerebellum, where it is integrated
and processed. The cerebellum is capable of modifying the action of different muscle groups so that
movements are performed smoothly and accurately.

Because the information carried by these pathways
does not reach consciousness directly, it is referred
to as unconscious proprioception. The two spinal
cord pathways that convey unconscious proprioceptive information to the cerebellum are the dorsal
and ventral spinocerebellar tracts. They have some
features in common, but they also have important
Fig. 34. Spinothalamic tract. Pathway for pain and temperature sensation.
75
76
anatomical and functional differences. Both tracts

(1) originate from neurons in the intermediate gray
matter; (2) contain large-diameter, rapidly conducting secondary axons (among the fastest conducting
pathways in the nervous system); (3) transmit
information from the lower extremities; and (4)
provide input predominantly to the ipsilateral
cerebellum.
J.R. DAUBE
Acknowledgments
The figures are adapted with permission from: Benarroch EE, Westmoreland BF, Daube JR, Reagan TJ,
Sandok BA. 4th Edition, Lippincott, Williams and
Wilkins, Philadelphia (1999) Medical Neurosciences:
An Approach to Anatomy, Pathology and Physiology
by Systems and Levels.

M.R. Nuwer (Ed.)
77
CHAPTER 4
EEG and anesthetic effects

Ville Janttia,* and Tod B. Sloanb,1
a
Department of Clinical Neurophysiology, Tampere University Hospital, FIN-33521 Tampere, Finland

b
Department of Anesthesiology, University of Colorado HSC, Denver, CO 80262, USA
4.1. Introduction
Electroencephalography (EEG) is increasingly used
in monitoring during anesthesia and surgery. Modern
electronics and signal processing technology have
produced new methods for both intensive care unit
(ICU) and operating room (OR) monitoring. For
example, estimating the anesthetic effects with univariate indexes calculated from EEG and auditory
evoked potentials. Understanding the physiology
behind these monitoring techniques and their use in
identifying adverse effects, such as epileptic seizures,
has become increasingly important and should be
recognized by everyone who uses these techniques
for monitoring.
In this chapter, the effects of anesthetics on bioelectrical phenomena at the molecular level, at the
effector site of anesthetics (synaptic ion channels),
and on large neural systems, such as those involved
in controlling consciousness, will be reviewed.
4.2. Effects of anesthetic agents
The target organ of anesthetic agents, the human brain,
is the most complex biological system known. The
function of the brain must be studied at different levels
ranging from cell membranes, receptors, and ion channels, to increasingly complex systems which result in
the changes in consciousness and cognition. Even if
the effects of anesthetics on all its target molecules
were known, unconsciousness cannot be predicted at
Correspondence to: Dr. Ville Jantti, Department of Clinical Neurophysiology, Tampere University Hospital, P.O.
Box 2000, FIN-33521 Tampere, Finland.
Tel.: +350-40-75-85-401.
E-mail: Ville.Jantti@pshp.fi (V. Jantti).
Present Address: Department of Anesthesiology, University of Colorado, Denver, Aurora, CO 80045, USA.
the molecular level because of the complexity of the

end result. To understand the effects of anesthetics, it
is beneficial to review the effects from the molecular
level to the cognitive neuroscience of sleep and anesthesia (Hobson and Pace-Schott, 2002; Jantti, 2005).
At the cellular level (ion channels and receptors),
gamma amino butyric acid (GABAergic), acetylcholine
(cholinergic), and glutamate (glutaminergic) receptors
are considered to be the most important sites of action
of anesthetics. Two of the major neural systems of
importance and relevance for EEG are those mediating
physiological sleep and cortical suppression. Together,
they generate the typical evolution of EEG characteristic of most anesthetics and their combinations: the
anesthetic model from an awake EEG through slowing
to burst suppression and suppression.
4.2.1. Molecular sites of action of anesthetic
agents: membranes and ion channels
Although it is clear that EEG activity changes with
anesthesia, the actual mechanisms which produce general anesthesia remain unclear. Cell ion channels, critical regulators of synaptic transmission, are considered
a potential target for anesthetic-induced depression of
synaptic function. These channels can be opened or
closed by (1) specific direct action of drugs at receptors,
(2) action on receptors through changes in the protein
lipid milieu around it in the cell membrane, and (3) secondarily through intermediary messengers from other
receptors affected by the anesthetics (such as guanosine
nucleotide binding protein: G protein).
Synaptic function appears to be the major mechanism of anesthetic drug action as opposed to altered
axonal conduction (which occurs only at much higher
drug concentrations than those effecting synaptic function) (Ting et al., 2003). Interactions via changes in
the receptor through proteinlipid milieu have been of
particular interest with the inhalational agents since
their lipid solubility is a characteristic physical property.
78
This result has been demonstrated with the Na/K

ATPase membrane protein (Sinensky et al., 1979)
and the nicotinic acetylcholine receptor (Firestone
et al., 1994).
Regardless of the specific mechanism, it is clear
that alterations in presynaptic transmitter release, or
alteration in postsynaptic ion channels are a mechanism of drug action that could produce some of the
effects of anesthetic agents. General anesthetics can
be classified into two groups based on their putative
molecular targets of action: (1) ketamine, nitrous
oxide, and xenon which antagonize the N-methyl-Daspartate (NMDA) receptor and inhibit neuronal nicotinic acetylcholine receptors and (2) volatile inhalational anesthetics and most intravenous drugs
(barbiturates, benzodiazepines [BZDs], propofol, etomidate) which act primarily as GABAA agonists
(Goto et al., 2001).
The GABAA receptor is a prime target for anesthetic action because of its role as the major inhibitory neurotransmitter. When the GABAA receptor is
activated, it prolongs the inhibitory current through
the ion pore (Jones and Harrison, 1993) decreasing
synaptic response to excitatory activation and passing the message to other neurons. Volatile anesthetic agents appear to modulate GABAA cortical
receptor function, specifically enhancing GABA
(Cheng and Kendig, 2000) resulting in neuronal
depression. The GABAA receptor has specific binding sites for barbiturates, etomidate, althesin, propofol, and BZDs. Most intravenous anesthetic agents
modulate the GABAA receptor by enhancing GABA
gating, increasing the effectiveness of GABA in
depressing neuronal function (Hales and Lambert,
1991; Tomlin et al., 1998). Action at extrasynaptic
GABAA receptors may also contribute to neurodepression of volatile anesthetic agents (Hales and
Lambert, 1991).
The other major target for anesthetic agents is the
major excitatory neurotransmitter system, glutamate
receptors, which is divided into three classes based
on the binding of NMDA, alpha-amino-3-hydroxy-5methyl-isoxazole propionic acid (AMPA), and kainate. These receptors are also believed to be involved
in learning, memory, motor coordination, neurotoxicity, and neurodegenerative disorders. Inhibition of
the AMPA and kainate receptors would likely cause
profound changes in excitability of the brain. This
may be the primary method of action of some volatile
agents, non-halogenated inhaled agents (including
xenon and nitrous oxide), and ketamine that appear
to have the primary effect of depressing excitatory
NTTI AND T.B. SLOAN

V. JA
transmission presynaptically by reducing glutamate

release (Zeilhofer et al., 1992; Franks et al., 1998;
Jevtovic-Todorovic et al., 1998; Flood and Krasowski,
2000; Raines et al., 2001; Campagna et al., 2003;
Perouansky and Hemmings Jr., 2003; Hemmings
et al., 2005). A variety of evidence suggests that barbiturates upregulate the NMDA receptor making it less
sensitive to depolarization currents. The degree of
inhibition at the NMDA and AMPA/kainate receptors
is known to alter the potency of inhalational agents
(Raja et al., 1982; Puil and El-Beheiry, 1990; Carl
and Moroni, 1992; Lin et al., 1993).
Other synaptic transmitters are also altered by
anesthetic agents. Of particular interest is the nicotinic acetylcholine receptor which is involved in the
neuromuscular junction and also in the central nervous system (CNS). The cortical synapses may be a
location for the action of xenon and nitrous oxide
(Campagna et al., 2003). Neuronal nicotinic acetylcholine receptors are thought to be involved in
hippocampal-mediated amnesia (Flood et al., 1997;
Violet et al., 1997; Raines et al., 2001). Nitrous
oxide, ketamine, and xenon also act on opioid receptors (mu, kappa, and delta) to produce analgesia
without much effect on consciousness. A role for
potassium and calcium channels, as well as gap junctions, has been proposed, but their role in anesthesia
is unclear.
Differences in the action of individual anesthetic
agents may relate to the spectrum of receptors altered
and the effects on a given receptor type throughout the
nervous system. This results in different and sometimes even opposite effects of different anesthetics
on some nervous system functions, despite the common outcome, unconsciousness (Sloan, 1998).
4.2.2. Effects of anesthetics on neural systems
4.2.2.1. Sleep and arousal
Wakefulness is the result of facilitating neuromodulatory influences at both thalamic and cortical levels
and originates from monoaminergic and cholinergic
cells located in the brainstem and basal forebrain.
A decrease in the firing rate of these cells leads to
a progressive hyperpolarization of thalamic and
cortical neurons resulting in slow oscillations characteristic of slow wave sleep (SWS). These oscillations
are seen in EEG and they are paralleled by changes
in processing of sensory information, which is increasingly blocked at the thalamic level. Hence, evoked
responses to sensory stimuli decrease in amplitude
and slightly increase in latency cephalad to the
thalamus. Cellular recordings also reveal changes in

evoked electrical activity at thalamic level, but, probably due to the shape of electrical field, these have not
been recorded at scalp level (Steriade, 2000).
One of the major results of several anesthetic
agents involves actions on specific neural pathways.
This is particularly true in the hypothalamus by activation of GABAA receptors that underlie SWS (Nelson
et al., 2002). The sedative effect is, therefore, not
due to generalized inhibition throughout the CNS,
but rather due to inhibition of the arousal-producing
systems. This results in activation of mechanisms
characteristic of SWS, which produce the slow oscillation seen in EEG with many anesthetics (Steriade,
2000) (Fig. 1). This is probably the main mechanism
behind the slow waves that are seen in the EEG with
increasing anesthetic effect. Interestingly, the general
anesthetic dexmedetomidine, a selective alpha2 adrenoceptor agonist, also activates sleep-promoting pathways, causing SWS-like EEG patterns although the
drug receptors involved are different (Nelson et al.,
2003). Additional features of general anesthesia that
link it to physiological sleep are amnesia, dreaming
(Brandner et al., 1997; Hobson and Pace-Schott,
79
2002), and recovery from sleep deprivation during

anesthesia (Tung et al., 2004).
4.2.2.2. Toxic effects and burst suppression
With increasing concentrations, the anesthetics affect
cortical neurons directly producing the EEG pattern of
burst suppression. Relatively high-amplitude activity
is intermittently suppressed to varying degrees, usually
under 5 mV. The duration of individual spontaneous
bursts and suppressions is unpredictable and irregular. This pattern can be produced in isolated cortex
after surgery (Wennberg et al., 1997). But, in the
structurally intact brain, it occurs in close connection
with subcortical structures (the corticothalamic system in particular), so that the pattern is synchronous
in the whole cortex (Steriade et al., 1994a).
Bursts consist of mixed frequency activity in the
alpha or theta band, superimposed on slower activity.
Some of the waveforms are sharp, particularly in barbiturate anesthesia. Sometimes, the burst consists of
distinctly epileptiform spikes or triphasic waves
(Fig. 2). At the onset of the burst, the cortical surface
and scalp electrodes become positive compared to
subcortical electrodes. Although this slow wave
Cortex
Slow
Spindle
5s
A
Reticular
Thalamus
Delta
0.5 s
Relay
1s
Fig. 1. Pacemakers of cortical slow oscillation, thalamic delta activity, and spindles during physiological sleep. These modulate the activity of cortical pyramidal neurons which produce EEG waves. The slow wave oscillations are probably responsible for majority of slow activity which increases in deepening anesthesia. A: excitatory glutaminergic neocortical
connections produce slow cortical oscillation, 0.3 Hz; B: inhibitory GABAergic reticular thalamic neurons produce spindle
oscillations, 7 Hz; C: excitatory thallamocortical or relay nefromurons produce clock-like delta oscillations, 1.5 Hz. Direction of axons is indicated by arrows. Short- and long-scale intracortical connections are shown. Divergent axons of thalamic
reticular neurons are shown as broken lines. In the intact brain, these structures are interacting and their rhythms are
combined within complex wave sequences. Reproduced from Steriade et al. (1994b) with permission from Elsevier.
80
Fig. 2. Non-epileptiform and epileptiform electroencephalogram phenomena during induction of sevoflurane

anesthesia. Fast (A), slow (B), delta (C), and slow delta
(D) activity are normal patterns, typical of anesthesia
induction. Delta monophasic activity (E) resembles slow
oscillations of sleep and develops into slow delta monophasic activity with spikes (F), followed by burst suppression
(G). Burst suppression patterns can be epileptiform, like
burst suppression with spikes (H) or polyspikes (I) the patterns can also develop into rhythmic polyspikes (J) which,
again, can present as triphasic waves. The patterns (FJ)
are similar to those frequently seen in ICU during treatment
of status epilepticus together with generalized or focal seizures. Spikes are indicated with asterisks in trace (F). EEG
was filtered with bandpass 1.050 Hz, montage Fp1left
mastoid. Reproduced from Yli-Hankala et al. (1999) with
permission from Lippincott Williams and Wilkins.
NTTI AND T.B. SLOAN

V. JA
activity is seen synchronously across the whole cortex, its amplitude is not uniform. This slow wave is
often parietally negative with an ear or frontopolar
reference (Figs. 3 and 4).
During suppression, the EEG can be very flat, which
has prompted the use of misleading descriptions such as
isoelectric or inactive. However, the EEG is neither
inactive nor isoelectric during suppression. Several
kinds of harmless or pathologic, spontaneous or evoked
waveforms, often exceeding 5 mV, can be seen. This
may consist of arrhythmic or rhythmic, theta or delta.
Rhythmic spindles of beta or higher frequency can be
seen during propofol- and etomidate-induced suppression. They include, and are sometimes mainly, gamma
frequency with double that of the beta frequency of
the spindle (Fig. 5). Rhythmic focal epileptic seizure
activity and high-amplitude spikes, up to 1 mV can be
seen both during bursts and suppressions. Somatosensory evoked potentials (SEPs) may even be enhanced
during suppression (Jantti et al., 1998).
Bursts are often readily induced during suppression
by external stimuli, such as somatosensory, auditory,
or photic activity. When a long stimulus is given, both
the onset and the end of the stimulus can induce a
burst. The beginning of a burst may have a regularly
repeating and stimulus-specific waveform, being, thus,
essentially an evoked potential (Hartikainen et al.,
1995). With propofol, this is a separate negative
vertex-wave, which can also occur during continuous
suppression (Fig. 4) (Huotari et al., 2004).
Burst suppression indicates severe dysfunction of the
cerebral cortex, if not occurring during anesthesia. It is a
sign of deep unconsciousness when appearing in the
whole cortex of a structurally healthy brain (like in
anesthesia). It can also occur focally in a structurally
damaged brain and then the effects of damage and anesthesia are additive (Wennberg et al., 1997). It can be
seen unilaterally in a Wada test (when one hemisphere
is anesthetized). In these cases, burst suppression does
not indicate unconsciousness. During anesthesia, the
burst suppression pattern is closely connected to thalamic activity (Steriade et al., 1994a). It may even be
reflected in the heart rate, which is vagally inhibited
during suppression (Yli-Hankala et al., 1993).
4.3. Effects of anesthetics on EEG
The EEG is a recording of voltage fluctuations ranging in duration from hours to less than a millisecond,
and in amplitude from millivolts to a fraction of
81
Czdepth
CzFp1
100 V
1s
Fig. 3. Sharp wave, burst, and spindle in propofol anesthesia. Upper trace: Czdepth electrode in subthalamic nucleus.
Lower trace: CzFp1 (0.01670 Hz). Negative at Cz is up. Note that burst slow wave is positive in surface electrode Cz
compared with depth electrode but negative with frontopolar reference. The slow wave is synchronous and positive with
depth electrode reference in all surface electrodes, that is, in the whole cortex, but due to different amplitude can be positive
or negative between two scalp electrodes depending on the montage. Sharp wave is negative in both, as it is generated in a
small part of cortex around Cz. This shows that in order to analyze and quantify widespread potential fields like that of
burst suppression and related slow activity in anesthesia, depth electrodes are necessary as the scalp derivations only reflect
potential differences between scalp locations. Note the mixed frequency activity on the slow wave of burst. It contributes
significantly to gamma band and is the major source of noise in SEP recordings.
Fig. 4. Burst suppression in propofol anesthesia, three successive 20-s EEG segments, recorded from Pz with reference on
tip of nose (0.0534 Hz). Repeated painful electrical stimuli (arrows) produce a series of waves, a vertex sharp wave, a
burst of slow wave, and the spindle. Notice that these resemble the evoked K-complexes of slow wave sleep. Reproduced
from Huotari et al. (2004) with permission from Oxford University Press.
NTTI AND T.B. SLOAN

V. JA
82
C4 A2
C4 A1
100 V
1s
Fig. 5. Beta and gamma activity during burst suppression in etomidate anesthesia (0.01670 Hz). The mixed frequency
activity on the slow wave, burst, contributes to gamma frequency band. The following spindle is first of beta frequency
but changes into a rhythm with double frequency, being then in the gamma frequency range. These spindles can also be
nonsinusoidal with spiky look like the m-rhythm of motor cortex awake and some sleep spindles. They are all probably
generated by the same physiological oscillating system.
microvolt. Thus, in frequency analysis, the EEG

covers the range from 0 Hz to over 1,000 Hz. The
higher frequency components are generally lower in
amplitude with the fastest components being smaller
than the noise level of recordings. These can only
be analyzed by correlating them with events by using
signal averaging techniques to produce averaged event
related potentials, for example, evoked potentials.
Drugs used to produce general anesthesia change
the EEG in the whole frequency range in all scalp
and depth electrodes. Anesthetics and their combinations which affect GABAA receptors induce typical
EEG changes: slowing (resembling SWS) with burst
suppression at higher concentrations. Although there
are differences between agents, there are common
changes such that these features can be used to
roughly estimate the general anesthetic effect. Several different EEG features, generated in different
parts of the brain, can be used to estimate the effect
of these anesthetics. In particular is the loss of consciousness and changes associated with unnecessarily
high drug concentrations. This has led to the development of EEG-based monitors of anesthetic effect.
Many important features can only be reliably
detected in the time domain (i.e., original, raw, EEG),
for example, burst suppression and epileptiform patterns. In the time domain, amplitudes, intervals, and patterns are analyzed, as in the case of burst suppression.
EEG patterns in different scalp electrodes can vary considerably. Waveforms and topography of patterns are
important, although their monitoring is more demanding. Slowing, decrease of amplitude, and developing
asymmetries are, however, important in monitoring

for ischemia. Transformation to the frequency domain,
usually with Fourier transform makes quantification of
oscillations of different frequency easy and is the most
popular way of quantifying EEG patterns.
4.3.1. Patterns and topography of EEG changes: the
anesthetic model
Considerable differences exist in the effects of anesthetics on the EEG between individuals and in different species. Yet, some features are common to most
drugs and their combinations. After presenting a general pattern of anesthetic action, the specific action
within the different EEG frequency bands will be
reviewed. Some typical patterns of normal anesthesia
will be presented and changes indicative of pathophysiological phenomena will be described as they
should be recognized and accounted for. Finally,
the effects of specific anesthetic drugs and their combinations on EEG will be mentioned.
Most anesthetics produce a characteristic sequence
of events with increasing drug levels (Fig. 6) (Stockard
and Bickford, 1981). At induction, widespread rhythmic beta activity is seen. Resting alpha rhythms of alert
wakefulness in occipital (classical alpha rhythm),
parietal, and superior temporal cortex disappear. With
loss of consciousness from anesthetic and sedative
agents, a common set of observations include (1) a
marked drop in gamma-band activity (2550 Hz),
(2) increase in slower frequency patterns (theta, delta),
(3) a great increase in power in the frontal electrodes,
83
Integrated cortical output

Amplitude x persistence
Arbitrary units
Diagram of average pattern changes

during anethesia
Complex
200
150
Burst suppression (early)
Rhythmic
100
Fast
50
Alpha
High
Med
Low
Burst suppression (late)
Suppression
Resting
l ll
rhythmns
Start
anesthetic
lll
lV
V
EEG Levels
Vl
Fig. 6. Typical EEG changes with progressively increasing levels of inhalational anesthesia. As shown, the amplitude
increases as the variability of the EEG is reduced to synchronous activity in the 812 Hz range. As anesthesia deepens,
the amplitude and frequency decrease giving way to burst suppression and finally electrocerebral silence. Reprinted from
Stockard and Bickford (1981) with permission from Elsevier.
(4) a marked increase in synchrony of the EEG, and

(5) an uncoupling of the interaction between frontal
and parietal electrodes and across the midline. These
observations have recently been presented as a model
of anesthetic action (John and Prichep, 2005) (Fig. 7).
At deep surgical anesthetic concentrations, occasional short suppressions appear, gradually increasing
in duration. Finally, continuous suppression is seen
(i.e., relatively low amplitude activity). Modern volatile anesthetics produce burst suppression, usually at
REGULATION OF MEAN FREQUENCY
CORTICAL-SUBCORTICAL
CT-TC
LOOP
CINGULATE
N. RETICULARIS
HIPPOCAMPUS
DORSAL STRIATUM
THALAMUS
AMYGDALA
SEPTAL NUCLEI
ENTORHINAL CX
S. NIGRA
VENTRAL
TEGMENTAL AREA
SENSORY
INPUTS
RETICULAR
FORMATION
GENERATION OF THETA
Fig. 7. Proposed mechanism of six stages of the effects of anesthesia producing loss of consciousness. As described in the
text, the effects on the brainstem ascending reticular activating system (ARAS) (1) leads to depression of memory formation (2) and gating of sensory information at the thalamus (3). This blocks cortical information processing (4), uncoupling
of cortical interactions (5), and depression of the prefrontal cortex (PFC) (6). Reprinted from John and Prichep (2005) with
permission from Lippincott, Williams and Wilkins.
84
minimum alveolar concentration (MAC) values of 1.5

or higher (MAC is the concentration of volatile agents
which eliminate movement response to pain in 50% of
patients). However, this threshold is usually agent- and
species-specific and substantial individual variability
can occur. Note, the onset of burst suppression has historically been associated with Guedels surgical plane
of surgical anesthesia (stage III, planes 23) (March
and Muir, 2005). Most anesthetics in current clinical
use (isoflurane, desflurane, propofol) follow a depressive pattern; however, some volatile agents may produce EEG excitation or seizure-like activity under
some physiologic conditions (e.g., sevoflurane and
enflurane with hyperventilation). This general pattern
of EEG activation and depression is summarized by
Winters and shown in Fig. 8 (Winters, 1976).
Anesthetic agents who induce their effect through
other than GABAA receptors (mainly by inhibiting
glutaminergic NMDA receptors) do not produce this
sequence of events. They have not been reported to
produce burst suppression (likely because this requires
glutamate-mediated excitatory synaptic transmission)
NTTI AND T.B. SLOAN

V. JA
(Lukatch et al., 2005). For example, ketamine induces

theta activity and nitrous oxide induces beta activity.
Xenon induces slow activity, but of more posterior distribution than GABAergic anesthetics.
Differences in EEG patterns between drugs and
differences in reactivity and evoked responses show
that there are also significant differences between
anesthetic agents and their combinations, even when
they follow the typical anesthetic model. Therefore,
despite the similarities in EEG changes, the anesthetic
states produced are physiologically very different
between drugs. There is no general depth of anesthesia,
or even depth of hypnosis, because both refer to different kinds of complex and incompletely understood
physiological phenomena (Kissin, 1993).
Burst suppression is not specific to drugs; it is also
seen in many intoxications and metabolic disturbances
(Young and DeRubeis, 1998). While burst suppression
and some other EEG patterns during anesthesia are
pathological, they are not, however, a cause for concern because they reflect a completely reversible
situation.
Fig. 8. Anesthetic drugs tend to either excite or depress the EEG and follow a pattern summarized by Winters (1976). Cortical
EEG stages typical of anesthetic agents. Reprinted, with permission, from the Annual Review of Pharmacology and Toxicology, Volume 16 # 1976 by Annual Reviews www.annualreviews.org.
4.3.2. Effects of anesthetics on different frequency

bands of EEG
Anesthesia-induced changes in the traditional frequency bands of EEG have been described. Before
examining these, two issues of frequency characteristics of the EEG must be noted. First, care should be
taken to differentiate regular rhythms from irregular
activity in the frequency band caused partly by bandpass filtering of non-sinusoidal regular, irregular, or
transient patterns. Second, slow activity is typically
higher in amplitude than fast activity in EEG. Deepening
anesthesia causes relative and absolute increase in slow
activity, even though there may be increased activity
in other frequency bands, and even in deep anesthesia.
As pointed out (Steriade, 2006), we should think
in the terms of the physiology underlying EEG patterns and rhythms. Beta and gamma oscillations of
sleep spindles, also seen during anesthesia, represent
the same physiological process, although they have
spectral peaks in two different bands. Any band, for
instance delta oscillations, includes activity from
many different neurophysiological systems.
The frequency range of the EEG can be divided into
bands according to the ranges described with physiological sleep (Steriade, 2006). Some of the patterns seen
during anesthesia reflect the mechanisms responsible
for EEG patterns in SWS. Others, burst suppression in
particular, are not part of adult physiological sleep.
4.3.2.1. Very slow activity, 00.5 Hz
Amplifiers capable of recording very slow fluctuations in voltage are said to be direct current (DC)
coupled. In practice, due to limitations of electrode
technology, we are limited to recordings of hours,
although we could expect that slow changes can last
for days, particularly in ICU patients. During clinical
anesthesia, this technology is difficult to apply as a
result of external interference. Nevertheless, slow
voltage changes have been recorded during physiological sleep. Changes in PaCO2 level change the
DC potential. Glial cells, in addition to neurons, are
involved in its generation. It has been claimed that
changes in the bloodbrain barrier are also partly
responsible (Voipio et al., 2003). Spreading depression, a mechanism suggested to be involved in brain
damage, can induce DC shifts several millivolts in
magnitude (Somjen, 2001). Burst suppression pattern
as characteristically seen during deep anesthesia
involves DC-shifts between depth and scalp electrodes. At the onset of a burst, particularly with volatile
85
anesthetics, the EEG at scalp electrode jumps to a

more positive level (when referred to depth electrode). In recordings between scalp recordings, however, anesthesia-induced bursts are often negative
parietally but positive frontally with an ear reference
(Fig. 3). Pathological EEG activity, such as slow
activity of encephalitis, can occur in this frequency
range and epileptic seizures typically occur on a negative DC-shift.
4.3.2.2. Delta activity, 0.54 Hz
Physiological slow wave (non-REM) sleep is characterized by delta waves that are the basis of sleep classification to different levels. It has two main components: the
cortical component of 0.51 Hz and the 14 Hz delta
activity reflecting the intrinsic properties of thalamocortical neurons (Steriade, 2000) (Fig. 1). Part of the
slow activity during sedation and light anesthesia is
generated by sleep mechanisms. Several other mechanisms may also contribute to delta activity, including the
emerging burst suppression pattern where the first suppressions are typically of short duration range. Many
pathological conditions cause focal or generalized delta
activity, including seizures, and these can contribute to
or dominate the EEG during anesthesia.
4.3.2.3. Theta, 47 Hz
Arrhythmic widespread theta activity is characteristic of
drowsiness. Rhythmic theta activity in the frontal midline can be seen during wakefulness and has sometimes
been associated with hypnosis. With deepening anesthesia, rhythmic theta follows alpha activity, which follows
the initial fast activity at induction. In deeper anesthesia,
theta waves or rhythms are seen on delta activity. In ketamine anesthesia, theta is the dominant EEG pattern.
4.3.2.4. Alpha, 813 Hz
Rhythms of 10 Hz are characteristic of relaxed wakefulness, particularly in the visual, sensorimotor, and auditory cortices. Sensory stimulation, such as opening of
eyes, abolishes these rhythms in the corresponding cortical region. They also disappear during sleep and general anesthesia. The best known is the alpha rhythm of
the visual cortex, which is abolished by visual stimulus
or imagery. Often these rhythms are sinusoidal, but can
also be spiky, suggesting non-linearities or harmonics
in the generators (resulting in spectral peaks at harmonic frequencies). The 10-Hz rhythm in the motor
cortex is called m-rhythm, which refers to its waveform.
However, at the motor cortex, discrete rhythms at 20
and 40 Hz can be seen in addition to the 10-Hz rhythms.
NTTI AND T.B. SLOAN

V. JA
86
frontal 10 -11 Hz alpha

Fp2 A2
O2 A2
frontal14 -16 Hz beta

Fp2 A2
O2 A2
100 V
1s
Fig. 9. Alpha coma due to intoxication. Upper trace shows 11-Hz alpha activity at frontopolar (Fp2) electrode and much
lower amplitude activity at occipital electrode. The lower picture shows 14-Hz beta activity a moment later. Patient reacted
to stimulation by frowning and corresponding EMG activity and decrease of EEG amplitude. He later recovered fully.
Although this rhythm in waveform and frequency resembles the alpha rhythms of restful wakefulness in visual, somatomotor,
and auditory cortices, it is physiologically distinctly different, and probably closer to the frontally dominant alpha rhythms of
anesthesia. The alpha rhythm of anesthesia induction also has similar distribution. Their origin is poorly understood but they
might result from slowing of beta rhythm which is readily seen at induction of anesthesia (bandpass 570 Hz).
Several drugs produce alpha rhythm during anesthesia. Halothane, for instance, produces widespread
814 Hz rhythmic activity and propofol produces
mixed activity onto slow wave. Interestingly, intoxications can produce an alpha coma pattern with
frontally maximal rhythmic alpha activity, which,
particularly if it clearly reacts by amplitude decrease
after stimulation, is, prognostically, a good sign
(Fig. 9). Unreactive alpha coma in ischemic brain
damage, however, is an ominous sign. The rhythms
of wakefulness do not move frontally, they disappear
at the onset of anesthesia, and the frontal rhythms of
the same frequency are physiologically different.
4.3.2.5. Beta, 1330 Hz
Sedative drugs induce diffuse beta activity in patients
who are awake, which disappears in physiological
sleep. It is probably caused by slowing of faster
rhythms due to inhibitory effects (Jensen et al.,
2005). At induction of anesthesia, and also during
physiological sleep, frontal fast beta activity is often
seen. The fast beta activity then slows to alpha activity
in anesthesia. With nitrous oxide as the sole agent, fast
beta is the dominating rhythm of surgical anesthesia.
Rhythmic 1315 Hz beta activity appears in runs of
waves resembling the spindles of physiological sleep
in propofol and etomidate anesthesia even during continuous suppression (Figs. 35).
4.3.2.6. Gamma, 3060 Hz
Rhythmic activity in this frequency range is interesting
in anesthesiology, as it has been suggested to be the
mechanism of binding together the function of different
brain regions, necessary for, among other things,
waking consciousness. There is, however, a considerable amount of spontaneous, arrhythmic activity in this
frequency range. This occurs even during burst
suppression as part of the bursts (MacDonald et al.,
2005), and rhythm in this range, related to the spindle
oscillations, can be seen during suppression (Fig. 5).
This is not surprising, as gamma band oscillations are
also seen during physiological sleep. During wakefulness and physiological REM, gamma activity is continuous. During SWS and propofol as well as etomidate
anesthesia, it is phasic.
The mid-latency auditory evoked potential
(MLAEP) also falls into this frequency range, and has
proven to be a reliable indicator of unconsciousness.
Repeating the auditory stimulus 40 times per second
evokes a 40-Hz rhythm. This steady-state potential,
which is closely related to the MLAEP, is also a sensitive measure of anesthetic effect. Cortical components
SEPs are also partly in this frequency range.
4.3.2.7. Very fast rhythms, above gamma

Spontaneous EEG activity above 80200 Hz and up
to 400 Hz is superimposed on the slow oscillations
in the neocortex. They are, however, below the noise
level in scalp recordings. Some very sharp waveforms
such as epileptic spikes may contribute to this band.
Some evoked waveforms, ranging in frequency up
to hundreds of Hertz have raised interest recently,
and are affected by anesthesia. A series of oscillations of 600800 Hz is seen superimposed on the primary cortical (N2O) wave of median SEPs when
awake. It disappears in sleep and light anesthesia.
4.3.3. Pathological EEG during anesthesia:
epileptiform activity and periodic patterns
K-complex oscillations during sleep can turn into
polyspikes and even seizures (Steriade and Amzica,
2003). This seems to be the mechanism of epileptiform activity induced by sevoflurane, where rapid
mask induction with high concentrations can produce
slow oscillation which turns into polyspikes and then
triphasic waves (Yli-Hankala et al., 1999) (Fig. 2).
Hyperventilation enhances development of epileptiform patterns and even clinical seizures with enflurane and sevoflurane.
Periodic waves occur with regular intervals like
rhythmic patterns with definable and quantifiable
intervals (Hirsch et al., 2005). When epileptiform
patterns (i.e., waves of sharp contour) occur periodically, they are often called periodic epileptiform discharges (PED). When they are focal, the term
lateral is added (hence PLED). This is a sign of
disturbance which may develop into seizures. PED
are frequently seen in the ICU, particularly when
monitoring status epilepticus, but may also appear
in the EEG during anesthesia. If they are seen, they
are best taken to be signs of a potential seizure. Sharp
waves and spikes are sometimes seen during treatment of status epilepticus with barbiturates, and they
are usually harmless.
4.3.4. EEG patterns induced by different
anesthetics
In monoanesthesia (i.e., using only one anesthetic
drug), specific effects on the EEG and evoked potentials reveal specific mechanisms of action in the
CNS. Differences between drugs are often best seen
in deep anesthesia at burst-suppression level. In
87
practice, combinations of anesthetics are used and

only a limited number of clinically interesting combinations have been well studied.
In this section, we first discuss the anesthetics
which can produce hypnosis for surgery, and variably
also analgesia and relaxation. Analgesics and relaxants, which do not produce surgical anesthesia alone,
are then presented.
4.3.4.1. Hypnotics that specifically affect GABAA
receptors: propofol and etomidate
The hypnotic effect of propofol and etomidate depend
on GABAA receptors which inhibit the effects of brainstem arousal mechanisms (Nelson et al., 2002; Rudolph
and Antkowiak, 2004). The EEG patterns are similar,
resembling closely the EEG of SWS. This includes slow
oscillations, vertex sharp waves, and spindles, and
1217-Hz waxing and waning series of waves. Spindles
and vertex waves can be seen in deep levels, during
burst suppression. Spindles are still present during continuous suppression when the slow wave bursts have
ceased (Huotari et al., 2004). Spindles are often of spiky
waveform, contributing to gamma band, and can also
consist of oscillation at double the basic frequency
(Fig. 5). Bursts are slow waves, with superimposed
activity at 10 Hz. Both bursts and spindles, therefore,
contribute to the gamma band.
Propofol can also enhance epileptiform activity in
some patients, but has become popular in treatment
of status epilepticus due to faster arousal than after
barbiturate anesthesia. Despite its seizure suppressive
activity, propofol has become a popular agent for
conscious sedation during surgery to identify and
ablate seizure foci. Because of its rapid metabolism,
propofol can be used to induce sleep during portions
of the procedure; it is eliminated rapidly enough to
allow subsequent testing and electrocorticography
(ECoG) when the patient is awake.
Propofol minimally suppresses motor systems of
the spinal cord. Many patients move spontaneously
or in reaction to touch or stimulation even at burstsuppression level. These movements include cough,
hiccup, twitching, or tremor. The patient may react
to touching as if tickled. Electrical stimulation of
the motor cortex with scalp electrodes can produce
activation of face and neck muscles for minutes,
which can totally cover the EEG and make it unreadable even at burst-suppression level. In this aspect,
propofol anesthesia also resembles SWS. Despite
unconsciousness and amnesia, motor activity and
reactions to pain are well preserved.
88
Studies with transcranial electric motor evoked

potentials (tceMEP) or magnetic motor evoked potentials (tcmMEP) have demonstrated a depressant effect
on response amplitude consistent with a cortical effect
resembling SWS (Kalkman et al., 1992). However, in
many patients, tceMEP can still be recorded during
suppression, although with decreased amplitude. The
rapid metabolism of propofol makes it an excellent
drug for tightly controlled infusion anesthesia. Its
rapid metabolism allows the depth of anesthesia and
effects on evoked responses to be adjusted rapidly.
Therefore, propofol is probably the most popular intravenous anesthetic when SEPs and motor evoked
potentials (MEP) are monitored, for example, during
scoliosis surgery.
Etomidate produces a similar evolution of EEG
from awake to burst suppression as propofol (including spindles which are still present during suppression) (Fig. 5). Etomidate has also been studied with
sensory and motor evoked responses and demonstrated an amplitude increase in cortical components
after injection (Kochs et al., 1986), with no changes
in subcortical or peripheral sensory responses. This
amplitude increase appears coincident with the
cyclones observed with administration of the drug,
suggesting a heightened cortical excitability (no evidence of seizure activity was noted).
Etomidate can enhance epileptic activity at low
doses (0.1 mg/kg) and may produce seizures in patients
with epilepsy as do barbiturates such as methohexital
(Rampil, 1997). Etomidate has been used as an alternative to thiopental to produce a flat EEG for metabolic
suppression; however, its faster metabolism allows for
faster patient awakening. Whether metabolic suppression with etomidate is as protective a suppression agent
as with barbiturates is not known. Deep anesthesia at
burst-suppression level, however, requires high doses
of etomidate which may suppress the adrenal cortex.
4.3.4.2. GABAergic anesthetics with other significant
effects
This group of anesthetics includes both volatile and
intravenous anesthetics. They all show the EEG
changes of the anesthetic model, anterior slow activity, and most of them also show burst suppression at
high concentrations. The EEG patterns, particularly
at burst-suppression level, are distinctly different
from each other and more specifically of GABAergic
propofol and etomidate. Spindles are reported with
some of these anesthetics, but they are rare in
humans and are not seen during EEG suppression.
NTTI AND T.B. SLOAN

V. JA
4.3.4.2.1. Halogenated agents. Possibly the most

common anesthetics are halogenated volatile anesthetics. The commonly used volatile anesthetic agents
(such as isoflurane or sevoflurane) cause immobility
by modulating multiple molecular targets predominantly in the spinal cord, including GABAA receptors,
glycine receptors, glutamate receptors, and two-poredomain K channels (TREK-1) (Grasshoff et al.,
2005). They produce the typical anesthetic pattern of
slowing and then suppression. With increasing age,
the brain becomes increasingly sensitive to anesthetics,
and burst suppression is produced at lower concentrations. In children, the EEG may change abruptly from
continuous high-amplitude activity to continuous suppression, which then again abruptly returns back to continuous EEG without intervening burst suppression.
4.3.4.2.2. Halothane. Halothane is the oldest in the
group of halogenated drugs. It does not readily cause
burst suppression but rather a gradual decrease in
amplitude. EEG-based indexes, therefore, cannot use
the burst-suppression pattern for calculation of deep
levels. Characteristic of halothane is rhythmic widespread alpha activity (814 Hz), which is not seen
with, for instance, isoflurane. This rhythm is abolished
by the addition of nitrous oxide (Yli-Hankala, 1990).
4.3.4.2.3. Isoflurane. Isoflurane has been popular
during monitoring for cortical ischemia (Sharbrough
et al., 1973). Isoflurane produces burst suppression
at clinical levels where blood pressure is not excessively affected. It has, therefore, been used for metabolic suppression. It has also been used at burstsuppression level for treatment of status epilepticus
and even in treatment of clinical depression instead
of electroconvulsive therapy.
4.3.4.2.4. Enflurane and sevoflurane. Both enflurane and sevoflurane produce EEG changes similar to
those of isoflurane. High concentrations with hyperventilation can produce epileptiform spikes or even electrographic seizures (Fig. 2). Epileptiform activity does not
readily produce convulsions. It may be associated with
an increase in heart rate. A rapid, unexplained, rise in
heart rate may warn about an electrographic seizure
(Jantti and Yli-Hankala, 1990). Sevoflurane can also
induce epileptiform activity in patients with epilepsy
(Iijima et al., 2000). Rapid induction with sevoflurane,
particularly with hyperventilation, and prolonged
anesthesia with high concentrations can produce a
generalized or focal seizure (even in healthy subjects).
In the EEG, the evolution of sleep-like slow wave oscillations to polyspikes, triphasic waves, and periodic patterns are seen (Fig. 2). In general, halogenated agents
are considered to suppress both motor and SEPs.
4.3.4.2.5. Desflurane. Desflurane is another rapidly eliminated volatile anesthetic due to its low solubility. Its EEG patterns are similar to those of
isoflurane and sevoflurane, and it has not been
reported to produce epileptiform activity.
4.3.4.2.6. Barbiturates. Barbiturates are similar to
inhalational agents in that their effect on the EEG is the
production of mild activation (fast activity) at low doses
and a depressant effect leading to burst suppression and
suppression at higher doses. Bursts may have very sharp
waves, which should not be interpreted as signs of a
threatening seizure. Because barbiturates produce
EEG suppression, they have been used to decrease
metabolism (about 50% at burst suppression) from synaptic activity, thereby improving the balance of nutrient
supply and demand (barbiturate coma). This suppression technique has been used in intensive care situations
to treat status epilepticus and during surgery in which
ischemic injury may occur (e.g., intracranial vascular
surgery, cardiac surgery, and carotid surgery). In such
cases, the EEG can be used to monitor the drug dosage,
but monitoring for seizures or cerebral ischemia will be
limited. Interestingly, low-dose methohexital (0.5 mg/
kg) has been used to enhance epileptic spike activity
during ECoG in surgery to identify seizure foci (Rampil
1997).
4.3.4.2.7. Benzodiazepines. The BZDs produce
frontal beta activity with a decrease in alpha activity
at low doses; spike-like activity is manifest in some
patients. In subjects who are awake, BZDs induce
beta activity which is probably caused by slowing
of higher frequencies through inhibitory mechanisms
(Jensen et al., 2005). At higher doses, the BZD produce generalized slowing into the theta and delta
range, usually without burst suppression. As a potent
anticonvulsant, BZD should be avoided if ECoG is
planned for identification of seizure foci. However,
small doses appear to be acceptable for supplementation during cases in which monitoring for ischemia is
needed.
4.3.4.2.8. Midazolam. Midazolam produces burst
suppression and this is used in treatment of status
epilepticus for monitoring of effect. Because of
89
midazolams excellent amnestic qualities, an infusion

can be used to maintain a steady level of supplemental hypnosis during opioid analgesia. Similarly, midazolam is an excellent supplemental hypnotic when
used with ketamine; it may also be useful to reduce
associated hallucinations.
4.3.4.3. NMDA inhibitors
Non-halogenated inhalational anesthetics, nitrous
oxide and xenon, as well as the intravenous agent
ketamine produce anesthesia by inhibiting NMDA
receptors. They are all potent analgesics. Their EEG
effects do not follow the typical anesthetic model,
although the EEG changes in a dose-dependent manner. Although the drug effect can be evaluated from
EEG, its correlation with unconsciousness is less
clear than with GABAergic drugs. No EEG indexes
have been developed to predict unconsciousness
using these anesthetics as a sole agent. Burst suppression is not reported with monoanesthesia. When
combined with GABAergic drugs such as halogenated volatile agents, the typical pattern of slowing
and burst suppression is dominant.
4.3.4.3.1. Nitrous oxide. When used alone, nitrous
oxide (N2O) decreases the frequency and amplitude
of sensory alpha rhythms of restful wakefulness. It
then produces a frontally dominant high-frequency
(>30 Hz) activity which coincides with analgesia
and depressed consciousness. The concentrations
needed to suppress movement are high and it has
been suggested that 50% nitrous oxide alone is compatible with motor and sensory monitoring. When
used with halogenated inhalational agents, it can be
additive or antagonistic depending on the circumstances and the neurophysiological measure used.
For example, when high concentration of N2O are
added during continuous EEG, it increases the
amount of delta activity and appears additive. However, at concentrations of halogenated agents producing burst suppression, N2O reduces the fraction of
time in suppression and can even return the EEG to
continuous activity. At the same time, it further suppresses the SEPs. Thus, N2O may be context sensitive in its effects, similar to its effects on the EEG
(i.e., the actual effect may vary depending on the
other anesthetics already present). N2O has proconvulsant properties when combined with some agents,
but it may also suppress epileptic spike activity during ECoG, depending on the other agents used. Like
90
sevoflurane and desflurane, N2O is relatively insoluble. Therefore, anesthetic effects can change rapidly
when concentrations are varied.
4.3.4.3.2. Xenon. Xenon, at inspired concentrations
of 80%, produces EEG changes which resemble volatile anesthetics. The increase of slow activity, however,
is strongest in posterior regions. Xenon causes stronger
slowing of EEG than, for instance isoflurane, and, therefore, some EEG indexes of anesthetic depth may suggest
surgical anesthesia when the patient is actually awake
(Goto et al., 2001). This, together with posterior slowing,
suggests that this slowing is not similar to SWS. This is
comparable to slow activity of an encephalitis patient
whose EEG may show slow delta for days after return
of consciousness. At subanesthetic concentrations the
sensory alpha rhythms are suppressed and beta activity
increases. Xenon suppresses the MLAEP and this could,
therefore, be useful in monitoring xenon anesthesia.
4.3.4.3.3. Ketamine. The effects of ketamine on
the EEG and evoked responses also differ from those
of inhalational agents. It has been reported to provoke seizure activity in persons with epilepsy but
not in normal subjects. Ketamine provides excellent
analgesia and hypnosis, but the hallucinatory activity
in adults and the increases in intracranial pressure in
patients with intracranial abnormalities that it produces limit its clinical usefulness. Ketamine produces
initial suppression of awake alpha rhythms, and then
induces high-amplitude theta activity in EEG, with
an accompanying increase in beta activity (Kochs
et al., 1996). At large doses, polymorphic delta with
interspersed beta is reported (Hirota, 2006).
Although ketamine produces analgesia and sedation (dissociative anesthesia), it is usually accompanied in adults by a sedative agent (e.g., a BZD)
to reduce hallucinatory activity. Ketamine does not
change either the amplitude or latency of the
MLAEP. In a study by Plourde et al. (1997), the
40-Hz auditory steady-state response (ASSR) and
EEG revealed no consistent differences between conscious and unconscious patients. No relationship
could be demonstrated between the increase in
amplitude of the 40-Hz ASSR or of relative theta
power (the hallmark of ketamine effect) and loss of
responsiveness to commands. They concluded that
ketamine, unlike other anesthetics, increases the
amplitude of the 40-Hz ASSR. Therefore, EEG and
MLAEP may not be useful in predicting unconsciousness in ketamine anesthesia.
NTTI AND T.B. SLOAN

V. JA
4.3.4.4. Analgesic agents: opioids

Opioids are used to produce analgesia particularly
with drugs which have little anesthetic property of
their own. This is particularly important with propofol
which has little analgesic and relaxant property. They
potentiate the effect of propofol on EEG and additionally suppress motor reflexes and spontaneous activity.
The opioids produce a dose-related decrease in
frequency of the EEG until in the delta range, while
maintaining amplitude. Burst suppression or electrical silence does not appear at higher doses. Opioids
do not appear to produce an initial excitement phase
in the EEG. However, increased muscle tone and seizures have been observed in dogs, unrelated to EEG
seizure activity. Some clinicians have found alfentanil useful in enhancing epileptic spikes.
Opioids are frequently used as supplementation during anesthesia, when detection of ischemia is needed.
The slower frequencies associated with large-dose
opioids may reduce the ability to detect ischemia. However, because amplitude is maintained, detection is not
prevented. Because the opioids do not produce marked
suppression of the EEG, they are frequently used during
electrocorticography (ECoG) in surgery for seizure
focus ablation. Remifentanil, a rapidly metabolized
opioid, may be well suited for use by infusion, particularly during ECoG. Because opioid anesthesia is often
insufficient to produce sedation and lack of awareness,
it is usually combined with an inhalational agent
(halogenated or N2O) or sedative drug.
4.3.4.5. Other sedatives: droperidol and
dexmedetomidine
Sedatives which do not have direct influence on
GABAA or NMDA receptors include droperidol and
dexmedetomidine. Droperidol has little effect on the
EEG when used alone. However, it lowers seizure
threshold, probably by dopamine antagonism. It does
not appear to produce neuroexcitatory phenomena or
to induce seizures in epileptic patients. When combined with fentanyl (neurolept anesthesia), droperidol increases EEG alpha activity at low doses. At
higher doses, it produces high-amplitude beta and
delta activity. Droperidol has been a popular anesthetic drug for use during conscious sedation for seizure focus identification and ablation, and it has been
an acceptable anesthetic agent during monitoring of
both EEG and cortical SEPs. Concerns of malignant
ventricular arrhythmias have limited its use.
The general anesthetic dexmedetomidine, a selective
alpha2 adrenoceptor agonist, activates endogenous
91
non-rapid eye movement (NREM) sleep-promoting

pathways. It, therefore, effects through the same
mechanisms as propofol and etomidate, although the
receptors involved are different (Nelson et al., 2003).
Dexmedetomidine produces an EEG which is very similar to SWS, but is not reported to produce burst suppression alone. Dexmedetomidine has been shown to
reduce the thiopental dose requirement for electroencephalographic burst suppression by 30% (Buhrer
et al., 1994).
specific way in which each individual drug acts to

produce the end result is unclear. The mechanisms
are likely linked to the specific ionic, synaptic, and
molecular interactions that are known for each drug.
However, as described in this chapter, individual variations in action give rise to wide variability in
changes to the EEG.
4.3.4.6. Regional anesthesia

Regional anesthetics do not normally affect EEG.
Yet, they can sometimes affect CNS when local anesthetics are given inadvertently, intravenously, or
when large doses are absorbed by vascular structures
near neural structures being anesthetized. They then
produce slowing of EEG and can even produce temporal epileptic seizures.
Brandner, B, Blagrove, M, et al. (1997) Dreams, images

and emotions associated with propofol anaesthesia.
Anaesthesia, 52(8): 750755.
Buhrer, M, Mappes, A, et al. (1994) Dexmedetomidine
decreases thiopental dose requirement and alters distribution pharmacokinetics. Anesthesiology, 80(6): 12161227.
Campagna, JA, Miller, KW, et al. (2003) Mechanisms of
actions of inhaled anesthetics [see comment]. N. Engl.
J. Med., 348(21): 21102124.
Carl, V and Moroni, F (1992) General anesthetics inhibit
the responses induced by glutamate receptor agonists
in the mouse cortex. Neurosci. Lett., 146: 2124.
Cheng, G and Kendig, JJ (2000) Enflurane directly depresses
glutamate AMPA and NMDA currents in mouse spinal
cord motor neurons independent of actions on GABAA
or glycine receptors. Anesthesiology, 93(4): 10751084.
Firestone, LL, Alifimoff, JK, et al. (1994) Does general
anesthetic-induced desensitization of the Torpedo acetylcholine receptor correlate with lipid disordering?
Mol. Pharmacol., 46(3): 508515.
Flood, P and Krasowski, MD (2000) Intravenous anesthetics differentially modulate ligand-gated ion channels. Anesthesiology, 92(5): 14181425.
Flood, P, Ramirez-Latorre, J, et al. (1997) Alpha 4 beta
2 neuronal nicotinic acetylcholine receptors in the central nervous system are inhibited by isoflurane and propofol, but alpha 7-type nicotinic acetylcholine receptors
are unaffected. Anesthesiology, 86: 859865.
Fodale, V and Santamaria, LB (2002) Laudanosine, an
atracurium and cisatracurium metabolite. Eur. J. Anaesthesiol., 19(7): 466473.
Franks, NP, Dickinson, R, et al. (1998) How does xenon produce anaesthesia? Nature, 396(6709): 324.
Goto, T, Nakata, Y, et al. (2001) How does xenon produce
anesthesia? A perspective from electrophysiological
studies. Int. Anesthesiol. Clin., 39(2): 8594.
Grasshoff, C, Rudolph, U, et al. (2005) Molecular and systemic mechanisms of general anaesthesia: the multisite and multiple mechanisms concept. Curr. Opin.
Anaesthesiol., 18: 386391.
Hales, TG and Lambert, JJ (1991) The actions of propofol
on inhibitory amino acid receptors of bovine adrenomedullary chromaffin cells and rodent central neurones.
Br. J. Pharmacol., 104(3): 619628.
4.3.4.7. Muscle relaxants

Muscle relaxants are generally believed to have no
effect on the EEG. When excessive EMG is observed
in EEG electrodes, it is detected by many monitors
and suggested to be an indicator of inadequate anesthesia. It may, however, only be a sign of a reflex
action to pain when the patient is deeply unconscious
(such as burst-suppression level in propofol anesthesia). Relaxation can be for surgical reasons, not for
consciousness or memory. The decision about its
importance, therefore, depends on the anesthetics
used and the surgical context.
Theoretically, one muscle relaxant may affect the
EEG. Laudanosine is a metabolite of the neuromuscular-blocking drugs atracurium and cisatracurium with
potentially toxic systemic effects. It crosses the blood
brain barrier and may cause excitement and seizure
activity. Its interest in recent years has increased
because of the recognized interaction with GABA, opioid, and nicotinic acetylcholine receptors. However,
except for prolonged administration of atracurium in
ICUs, laudanosine accumulation and related toxicity
seem unlikely to be achieved in clinical practice (Fodale
and Santamaria, 2002).
4.4. Conclusion
Clearly, the end results of anesthesia (unconsciousness, amnesia, analgesia, and immobility) are common among the anesthetic agents. However, the
exact mechanism by which these occur and the
References
92
Hartikainen, KM, Rorarius, M, et al. (1995) Cortical reactivity during isoflurane burst-suppression anesthesia.
Anesth. Analg., 81: 12231228.
Hemmings, HC Jr., Akabas, MH, et al. (2005) Emerging
molecular mechanisms of general anesthetic action.
Trends Pharmacol. Sci., 26(10): 503510.
Hirota, K (2006) Special cases: ketamine, nitrous oxide and
xenon. Best Pract. Res. Clin. Anaesthesiol., 20(1):
6979.
Hirsch, LJ, Brenner, RP, et al. (2005) The ACNS subcommittee on research terminology for continuous EEG
monitoring: proposed standardized terminology for
rhythmic and periodic EEG patterns encountered in critically ill patients. J. Clin. Neurophysiol., 22(2):
128135.
Hobson, JA and Pace-Schott, EF (2002) The cognitive neuroscience of sleep: neuronal systems, consciousness and
learning. Nature Rev. Neurosci., 3(9): 679693.
Huotari, AM, Koskinen, M, et al. (2004) Evoked EEG patterns during burst suppression with propofol. Br. J.
Anaesth., 92(1): 1824.
Iijima, T, Nakamura, Z, et al. (2000) The epileptogenic properties of the volatile anesthetics sevoflurane and isoflurane in patients with epilepsy [see comment]. Anesth.
Analg., 91(4): 989995.
Jantti, V (2005) From crystal ball towards cognitive anaesthesiology. Acta Anaesthesiol. Scand., 49(3): 273276.
Jantti, V and Yli-Hankala, A (1990) Correlation of instantaneous heart rate and EEG suppression during enflurane
anaesthesia: synchronous inhibition of heart rate and cortical electrical activity? Electroencephalogr. Clin. Neurophysiol., 76(5): 476479.
Jantti, V, Sonkajarvi, E, et al. (1998) Single-sweep cortical
somatosensory evoked potentials: N20 and evoked bursts
in sevoflurane anaesthesia. Electroencephalogr. Clin.
Neurophysiol., 108(3): 320324.
Jensen, O, Goel, P, et al. (2005) On the human
sensorimotor-cortex beta rhythm: sources and modeling.
Neuroimage, 26(2): 347355.
Jevtovic-Todorovic, V, Todorovic, SM, et al. (1998)
Nitrous oxide (laughing gas) is an NMDA antagonist,
neuroprotectant and neurotoxin [see comment]. Nature
Med., 4(4): 460463.
John, ER and Prichep, LS (2005) The anesthetic cascade: a
theory of how anesthesia suppresses consciousness.
Anesthesiology, 102(2): 447471.
Jones, MV and Harrison, NL (1993) Effects of volatile
anesthetics on the kinetics of inhibitory postsynaptic
currents in cultured rat hippocampal neurons. J. Neurophysiol., 70(4): 13391349.
Kalkman, CJ, Drummond, JC, et al. (1992) Intraoperative
monitoring of tibialis anterior muscle motor evoked
responses to transcranial electrical stimulation during
partial neuromuscular blockade. Anesth. Analg., 75(4):
584589.
NTTI AND T.B. SLOAN

V. JA
Kissin, I (1993) General anesthetic action: an obsolete
notion? [see comment]. Anesth. Analg., 76(2): 215218.
Kochs, E, Treede, RD, et al. (1986) Increase in somatosensory evoked potentials during anesthesia induction with
etomidate. Anaesthesist, 35(6): 359364.
Kochs, E, Scharein, E, et al. (1996) Analgesic efficacy of
low-dose ketamine. Somatosensory-evoked responses in
relation to subjective pain ratings. Anesthesiology, 85
(2): 304314.
Lin, LH, Leonard, S, et al. (1993) Enflurane inhibits the
function
of
mouse
and
human
brain
phosphatidylinositol-linked acetylcholine and serotonin
receptors expressed in Xenopus oocytes. Mol. Pharmacol., 43(6): 941948.
Lukatch, HS, Kiddoo, CE, et al. (2005) Anesthetic-induced
burst suppression EEG activity requires glutamatemediated excitatory synaptic transmission. Cereb. Cortex, 15(9): 13221331.
MacDonald, DB, Al Zayed, Z, et al. (2005) Tibial somatosensory evoked potential intraoperative monitoring:
recommendations based on signal to noise ratio analysis
of popliteal fossa, optimized P37, standard P37, and P31
potentials. Clin. Neurophysiol., 116(8): 18581869.
March, PA and Muir, WW (2005) Bispectral analysis of the
electroencephalogram: a review of its development and
use in anesthesia. Vet. Anaesth. Analg., 32(5): 241255.
Nelson, LE, Guo, TZ, et al. (2002) The sedative component
of anesthesia is mediated by GABA(A) receptors in an
endogenous sleep pathway [see comment]. Nature
Neurosci., 5(10): 979984.
Nelson, LE, Lu, J., et al. (2003) The alpha2adrenoceptor agonist dexmedetomidine converges on an
endogenous sleep-promoting pathway to exert its sedative
effects. Anesthesiology, 98(2): 428436.
Perouansky, M and Hemmings, HC Jr. (2003) Presynaptic
Actions of General Anesthetics. Humana Press.
Plourde, G, Baribeau, J, et al. (1997) Ketamine increases
the amplitude of the 40-Hz auditory steady-state
response in humans. Br. J. Anaesth., 78(5): 524529.
Puil, E and El-Beheiry, H (1990) Anaesthetic suppression
of transmitter actions in neocortex. Br. J. Pharmacol.,
101(1): 6166.
Raines, DE, Claycomb, RJ, et al. (2001) Nonhalogenated
alkane anesthetics fail to potentiate agonist actions on two
ligand-gated ion channels. Anesthesiology, 95(2): 470477.
Raja, SN, Moscicki, JC, et al. (1982) The effect of acute
phencyclidine administration on cyclopropane requirement (MAC) in rats. Anesthesiology, 56(4): 275279.
Rampil, IJ (1997) Electroencephalogram. In: MA Albin
(Ed.), Textbook of Neuroanesthesia with Neurosurgical
and Neuroscience Perspectives. McGraw-Hill, New
York, pp. 193220.
Rudolph, U and Antkowiak, B (2004) Molecular and neuronal substrates for general anaesthetics. Nat. Rev. Neurosci., 5(9): 709720.

Sharbrough, FW, Messick, JM Jr., et al. (1973) Correlation
of continuous electroencephalograms with cerebral
blood flow measurements during carotid endarterectomy. Stroke, 4(4): 674683.
Sinensky, M, Pinkerton, F, et al. (1979) Rate limitation of
(Na K)-stimulated adenosinetriphosphatase by
membrane acyl chain ordering. Proc. Natl. Acad. Sci.
USA, 76(10): 48934897.
Sloan, TB (1998) Anesthetic effects on electrophysiologic
recordings. J. Clin. Neurophysiol., 15(3): 217226.
Somjen, GG (2001) Mechanisms of spreading depression
and hypoxic spreading depression-like depolarization.
Physiol. Rev., 81(3): 10651096.
Steriade, M (2000) Brain Electrical Activity and Sensory
Processing During Waking and Sleep Stated. W.B.
Saunders Company, Philadelphia.
Steriade, M (2006) Grouping of brain rhythms in
corticothalamic systems. Neuroscience, 137(4):
10871106.
Steriade, M and Amzica, F (2003) Sleep oscillations developing into seizures in corticothalamic systems. Epilepsia, 44(Suppl. 12): 920.
Steriade, M, Amzica, F, et al. (1994a) Cortical and thalamic cellular correlates of electroencephalographic
burst-suppression. Electroencephalogr. Clin. Neurophysiol., 90(1): 116.
Steriade, M, Contreras, D, et al. (1994b) Synchronized
sleep oscillations and their paroxysmal developments.
Trends Neurosci., 17(5): 199208.
Stockard, J and Bickford, R (1981) The neurophysiology of
anesthesia. In: E. Gordon (Ed.), A Basis and Practice of
Neuroanesthesia. Excerpta Medica, New York, pp. 350.
Ting, CH, Angel, A, et al. (2003) Neuronal network modelling of the effects of anaesthetic agents on somatosensory pathways. Biol. Cybern., 88(2): 99107.
Tomlin, SL, Jenkins, A, et al. (1998) Stereoselective effects
of etomidate optical isomers on gamma-aminobutyric
93
acid type A receptors and animals [see comment]. Anesthesiology, 88(3): 708717.
Tung, A, Bergmann, BM, et al. (2004) Recovery from
sleep deprivation occurs during propofol anesthesia
[see comment]. Anesthesiology, 100(6): 14191426.
Violet, J, Downie, D, et al. (1997) Differential sensitivities
of mammalian neuronal and muscle nicotinic acetylcholine receptors to general anesthetics. Anesthesiology, 86:
866874.
Voipio, J, Tallgren, P, et al. (2003) Millivolt-scale DC
shifts in the human scalp EEG: evidence for a nonneuronal generator. J. Neurophysiol., 89(4): 22082214.
Wennberg, RA, Quesney, LF, et al. (1997) Epileptiform
and non-epileptiform paroxysmal activity from isolated
cortex after functional hemispherectomy. Electroencephalogr. Clin. Neurophysiol., 102(5): 437442.
Winters, WD (1976) Effects of drugs on the electrical
activity of the brain: anesthetics. Annu. Rev. Pharmacol.
Toxicol., 16: 413426.
Yli-Hankala, A (1990) The effect of nitrous oxide on EEG
spectral power during halothane and isoflurane anaesthesia. Acta Anaesthesiol. Scand., 34(7): 579584.
Yli-Hankala, A, Loula, P, et al. (1993) Atropine abolishes
electroencephalogram-associated heart rate changes
without an effect on respiratory sinus arrhythmia during
anaesthesia in humans. Acta Physiol. Scand., 149(4):
435440.
Yli-Hankala, A, Vakkuri, A, et al. (1999) Epileptiform
electroencephalogram during mask induction of anesthesia with sevoflurane. Anesthesiology, 91(6):
15961603.
Young, GB and DeRubeis, DA (1998) Metabolic Encephalopathies. McGraw-Hill, New York.
Zeilhofer, HU, Swandulla, D, et al. (1992) Differential
effects of ketamine enantiomers on NMDA receptor
currents in cultured neurons. Eur. J. Pharmacol., 213(1):
155158.

M.R. Nuwer (Ed.)
94
CHAPTER 5
Anesthetic effects on evoked potentials

Tod B. Sloana,* and Ville Janttib
a
Department of Anesthesiology, University of Colorado HSC, Denver, CO 80262, USA
Department of Clinical Neurophysiology, Tampere University Hospital, FIN-33521 Tampere, Finland
As discussed in the previous chapter (Chapter 4, this

volume) on anesthetic effects and the electroencephalogram (EEG), the major target of anesthetic action
appears to be synaptic function. Since electrophysiological recordings that depend on these structures
will be most susceptible to depressant agents, the
changes from anesthetic agents can usually be predicted by examining the anatomy of the neural pathways involved. The net effect of anesthetic agents is
due to at least three synaptic-mediated effects as well
as changes in physiology caused by the agents.
First is the depression of synaptic function. This
effect can be generally predicted by knowing the
location of synapses within the neural pathway
involved and the specific synaptic receptors and peak
generators being affected by the drugs. Since synaptic effects will, like the effects on the EEG, result
in prominent anesthetic effects on the cortically generated responses, it is not surprising that anesthetic
effects on evoked responses parallel anesthetic
effects on the EEG (which is also cortically and synaptically derived). In 1967, Winters et al. (1967) proposed a schema for anesthesia effects on cortical
auditory evoked potential (AEP) that mimics a similar schema for anesthetic effects on the EEG (see
Chapter 4, this volume) (Fig. 1). This schema implies
that anesthetic agents have two major effects on cortical evoked potentials (EPs). First, some anesthetics
decrease the amplitude until the EPs is no longer
distinguishable from the background noise. Other
anesthetics increase the amplitude, perhaps by
*
Correspondence to: Dr. Tod B. Sloan, Department of
Anesthesiology, University of Colorado, Denver, Anschutz
Office West (AO1), P.O. Box 6511, 12631 E 17th Avenue,
Aurora, CO 80045, USA.
Tel.: +1-303-724-1751; fax: +1-303-724-1761.
E-mail: Tod.Sloan@uchsc.edu (T.B. Sloan).
increasing cortical excitability. These can then lead

to seizure activity or depression.
The changes of Winters imply that the major
effect is on EP amplitude; however, a consideration
of generators suggests the changes may be more
complex. An EP generator can refer to the anatomical structure from which the potential is believed to
originate, or some theoretical model of the source
of the potential (or both together) (Mauguie`re,
2004). From the point of view of the impact of
anesthetic agents, it is important to recognize that
any waveform or peak in cortical EPs represents the
sum of the activity of several different simultaneously active generators. The individual component
generators may be affected differently by the anesthetics leading to changes in peak morphology as
well as amplitude and latency. Typically, since
longer latency peaks generally have more synapses
involved, these peaks are effected more by anesthetics. This has been nicely shown with sevoflurane
at burst suppression where only the N20 wave of cortical median nerve somatosensory EPs (SEPs) is preserved (Jantti et al., 1998). This also means that the
anesthetic effects can be used to study the generator
sources.
The synaptic effect also changes the ability of the
synapse to recover after a depolarization since ionic currents may be prolonged (Chapter 4). As such, the anesthetic effect is stimulation-rate dependent. This leads to
a trade-off in response amplitude and signal averaging
time for response acquisition (i.e., a slower stimulation
rate produces larger amplitude responses but increases
the time to average for a set number of EP averages)
(Nuwer, 1986). This effect is seen in Fig. 2 where the
cortical N20 decreases in amplitude with a higher
frequency of stimulation (and adapts with the frequency of 1 stimulus/s). Hence, higher concentrations
of anesthetics may necessitate the use of lower stimulus
frequencies (Jantti et al., 1998). Ironically, a higher
95
Fig. 1. Cortical somatosensory evoked potential (SEP) stages typical of anesthesia. Reproduced with permission from
Winters et al. (1967) with permission from Lippincott, Williams and Wilkins.
Fig. 2. N20 waves after median nerve stimulation in sevoflurane-induced electroencephalogram (EEG) suppression. Short
spikes are stimulus artifacts, sharp wave upwards is the N20 wave. Note that the later cortical waves are abolished. The
amplitude of the N20 wave decreases, that is, adapts still at the frequency of 2 Hz, and to 1/3rd with the 5 Hz stimulation
frequency, which can be used in awake subject. On the other hand, due to the high signal to noise ratio, that is, even single
responses are visible, only a few responses need to be averaged for excellent quality somatosensory evoked potentials
(SEPs). This recording is from P3 to C3 and it is a grand average from six patients. Reproduced from Jantti et al. (1998)
with permission from the International Federation of Clinical Neurophysiology.
96
concentration usually cause EEG suppression which

improves the ratio of the EP signal to background
EEG noise reducing the number of averages required
(MacDonald et al., 2005). Hence, the optimal stimulation rate for each patient and anesthetic may need
to be explored in monitoring. As such, this effect may
also necessitate recording from different montages
simultaneously because of the individual variation of
potential fields of EPs.
Also of clinical relevance is that the impact on
synapses and generators can change the interwave
latencies because the latency of the later waves is
more substantially affected. As each peak in EPs
represents the sum of different generators, which
are not necessarily successive (i.e., may be activated
in a parallel fashion, instead of sequentially), intervals of peaks do not necessarily indicate conduction
times, and interpeak intervals may change with
anesthesia in addition to physiological parameters
such as neural ischemia.
The second type of anesthetic effect is the alteration in synaptic function of ancillary neural pathways that interact on the pathway that mediate the
response being measured. These effects could cause
additional depression, could release the current
state of depression, or could result in enhancement
of the responses. This effect may account for some
of the effects of anesthetics which increase responses
at low doses and result in depression at higher doses.
The third mechanism is the more global effect of
anesthetic agents that results in the state of unconsciousness and lack of movement to painful skin
incision referred to as general anesthesia. This latter effect is not well understood, but a recent model
suggests that general anesthetic agents result in
changes at two major locations that will impact on
evoked responses (John and Prichep, 2005). The first
of these locations is action at the reflex pathways
in the spinal cord. This effect is the well known effect
that mediates the lack of movement with inhalational
agents referred to as minimal alveolar concentration
(MAC) where 50% of patients do not move to skin
incision. This effect may alter spinal reflex activity
and motor EPs. The second location is blockage
of sensory information to the brain at the thalamus.
This thalamic gating may explain why cortical sensory responses are substantially blocked at anesthetic
doses associated with anesthesia and unconsciousness. This effect is intimately intertwined with
arousal mechanisms and occurs naturally with cortical EPs as depicted in Winterss schema as changes of
EPs with sleep.
NTTI
T.B. SLOAN AND V. JA
As discussed in Chapter 4 on EEG, current

evidence suggests that one of the major effects of
several anesthetic agents involves actions on the
hypothalamus that underlies slow wave sleep (SWS)
by activation of the alpha receptor of the g-aminobutyric acid (GABAA) receptors (Nelson et al.,
2002). The general anesthetic dexmedetomidine, a
selective alpha2 adrenoceptor agonist, activates the
same pathways although through different receptors
(Nelson et al., 2003). Slow wave sleep involves gating of sensory information at the level of thalamus,
causing changes in the waveform, amplitude, and
latency of cortical EPs. The highly specific GABAergic drugs etomidate and propofol as well as dexmedetomidine therefore produce a state which closely
resembles SWS. This is seen both in EEG, where slow
waves, vertex sharp waves, and spindles are seen.
Also, the changes in EPs are similar to those during
SWS (Bastuji and Garcia-Larrea, 1999; Steriade,
2000). Their latency increases, shape changes, and
amplitude decreases. It is therefore likely that the
changes in EPs caused by GABAergic drugs are partly
due to the sleep mechanisms, particularly gating of
sensory information at thalamic level.
Anesthetics with significant effects on other systems in addition to GABAergic pathways typically
affect the sensory EPs as well as motor potentials
more than those which act mainly on the GABAergic
sleep promoting systems, that is, propofol and etomidate. Anesthetics which affect mainly N-methyl Daspartic acid (NMDA) receptors do not produce the
typical EEG changes including sleep-like slow activity frontally and their effects on EPs are different.
Unfortunately, neither EEG or EPs give reliable estimate of consciousness with these latter drugs (N2O,
xenon, ketamine).
Although part of the arousal mechanism is
affected by anesthestics; part of them are still active
in deep anesthesia (i.e., at burst suppression level).
Hence, a minor novel stimulus, such as electrical
stimulation of peripheral nerve or even light touch
of palm or foot (Jantti et al., 1998) can produce a
burst during suppression. This means that some neural system, probably in the brainstem, detects the
novel stimulus and activate an arousal system, which
again activates the cerebral cortex. Interestingly, a
similar sensitivity to light touch is sometimes seen
after ischemic brain damage (post-ischemic myoclonus). Painful stimulation causes an arousal reaction
which may consist of increase or decrease of amplitude and frequency of the EEG. Both somatosensory
EP (SEP) and mid-latency AEPs (MLAEP) change
towards patterns of lighter anesthesia, although the

drug concentrations remain constant (Thornton,
1991; Rundshagen et al., 1995). At a burst suppression
level of the EEG, sensory stimuli readily produce
bursts, which is due to arousal (Derbyshire et al.,
1936; Hartikainen et al., 1995). In sevoflurane and
isoflurane anesthesia, the evoked responses to somatosensory stimulation are preceded by the short latency cortical component and the onset of burst is of
constant waveform (i.e., a long latency EP). With propofol, the evoked complex during suppression consists
of the short latency SEP, a vertex sharp wave, the slow
wave burst, and a spindle, each from different generators (Huotari et al., 2004). Painful stimulation can,
in fact, change burst suppression to continuous EEG
in isoflurane anesthesia.
Thus, to fully explain the effects of anesthesia,
EPs must be viewed in the context of effects on the
synapses, generators, ancillary pathways, and complex mechanisms of sleep and unconsciousness.
In addition, the nonneural effects, such as those due
to alterations in cardiovascular physiology (e.g.,
changes in blood flow or blood pressure) may also
result in neural changes. All of these, no doubt,
account for the differences between individual
agents, even if the specific actions are not completely
understood.
5.1. Inhalational anesthetic agents
The parallel between the effects of anesthetics on the
EEG and EPs has been observed in the SEP, where
agents alter the cortical SEP in parallel to their
effects on the EEG (Himwich, 1951). Since most
drugs in common use today produce a dose-related
depression of the EEG, they decrease the evoked
response (decrease in amplitude and increase in
latency), making the choice of anesthetic medications
challenging during intraoperative monitoring of the
cortical SEP.
The effects of anesthetic depression have been
shown in an extensive study by Angel (Angel and
Gratton, 1982), in which numerous anesthetic agents
were examined using the cortical SEP from forepaw
stimulation in the rat. All of the agents studied produced a dose-dependent decrease in amplitude and
increase in latency. An effective dose was calculated
for 50% depression of amplitude (ED50), which
correlated with the lipid solubility of the agents,
which is known to correlate with anesthetic potency
(Meyer-Overton theory). This suggested that the cortical EP changes paralleled anesthetic depth. This
97
creates the possibility that cortical evoked responses

can be used for the assessment of anesthetic effect
and has been used in a device based on the cortical
auditory response.
Although the effects of anesthetics on EPs appear
to parallel their anesthetic potency, specific anesthetic agents may differ depending on the specific
loci of neural structures that may be excited or
depressed. This effect was nicely demonstrated by
Rosner, who reviewed the dose-related effects of several anesthetics on different neural areas (notably the
mesencephalic reticular formation, thalamus, and
cerebral cortex) (Rosner and Clark, 1973). Rosner
demonstrated that differences in neural depression
and excitation correlated with differences in EEG
patterns with increasing doses of the agents studied.
Rosner ordered anesthetic agents based on the ability
to depress cortical evoked responses (nitrous oxide >
ether > chloroform > halothane, methoxyflurane,
and trichloroethylene).
Consistent with Winterss proposal and the effect
on the EEG, enflurane has the capability of causing
an increase in cortical excitability (including seizures
under some condition), which appears to enhance
cortical EPs. This effect has been observed in the
rat visual EP (VEP) and auditory brainstem response
(ABR) using depth electrodes at concentrations over
1.5% (Yeoman et al., 1980; Haghighi et al., 1990a).
Thus, at clinically (or surgically) equivalent depths
of anesthesia, some agents (nitrous oxide) may produce a greater degree of cortical EP depression than
other agents. The differences between drugs may also
be explained by differing profiles on receptor types
(e.g., GABA, NMDA, etc.), differing location of
action (i.e., pre or postsynaptic effects), and the
spectrum of effects on individual subtypes of these
receptors. The differing spectrum of actions can also
explain the differing action on specific neural pathways and modalities. For example, barbiturates and
nitrous oxide depress the anterolateral spinal cord
pathways more than the dorsal column pathways.
5.1.1. Halogenated inhalational agents
The most prominent anesthetic effects on evoked
responses during clinical anesthesia are those of the
potent inhalational agents; halogenated alkanes (halothane), or ethers (enflurane, isoflurane [ForaneW],
sevoflurane [UltaneW], desflurane [SupraneW]). These
drugs have a broad action on neural structures
including the GABA receptor in the synapses, on
GABA receptors extrasynaptically, they antagonize
NTTI
98
effect. Studies in children demonstrate that the predominant effect is above the level of the thalamus
as predicted (N19P22 and above) (Da Costa et al.,
2001) and specific studies of the spontaneous and
evoked output of the thalamic relay nuclei ventroposterior and lateral (VPM, VPL) suggest that these
nuclei may be an important location for the anesthetic
modulation of afferent stimuli (Detsch et al., 1999).
Since this level of anesthetic is 0.30.5 MAC, it may
explain why many cortical sensory evoked responses
(such as the SEP) can often be recorded with concentrations of about 0.51 MAC. Interestingly, the
nonlinear effect is also supported by neuronal network
modeling of the SEP effect based on the known effect
of anesthetic agents on neurons (Ting et al., 2003).
These predictions mirror what is seen in practice
(Shimoji et al., 1984). Shown in Fig. 3 are the effects
of isoflurane on the responses recorded in the epidural space, on the skin over the cervical spine (Fig. 4),
and over the sensory cortex (Fig. 5) after stimulation
of the posterior tibial nerve. As clearly seen, these
responses mirror the predictions. Further, also as
shown in Fig. 5, the loss of cortical amplitude is nonlinear at isoflurane concentrations just above those
where unconsciousness occurs.
This anesthetic effect is also seen with the auditory response (Dubois et al., 1982; James et al.,
1982; Thornton et al., 1983, 1984; Manninen et al.,
1985; Schmidt and Chraemmer-Jorgensen, 1986;
Sebel et al., 1986; Heneghan et al., 1987; Sainz
et al., 1987; Newton et al., 1989; Lloyd-Thomas
et al., 1990; Sharpe et al., 1997a,b). The ABR
Isoflurane % ET
0.3
0.6
0.9
1.2
Amplitude (mV)
the NMDA channel, and they act via interactions in

the hydrophobic region of the cell membrane bilayer
on the Na/K ATPase channel and neuronal nicotinic acetylcholine receptor. A variety of studies have
been done with these agents and an understanding
of their effects serves as a good reference for
comparison of the other agents.
As a sensory response, the somatosensory evoked
response from peripheral nerve stimulation follows
the synaptic model of anesthetic effect with depression of the EEG. In general, this synaptic model suggests that the lack of synapses between stimulation of
the peripheral nerve and the cervicomedullary junction should be associated with minimal changes in
the responses recorded in the peripheral nerve and
spinal responses. Studies of recordings at Erbs point
(brachial plexus from upper extremity stimulation)
and over the cervical spine (from lower extremity
stimulation) show minimal changes (09%), that are
not dose related (Peterson et al., 1986; Sebel et al.,
1987). Major changes are seen above the thalamus
(where the second synapse is located) and from the
cerebral cortex. Consistent with thalamic gating
of the anesthetic model, the responses above the
thalamus are disproportionately effected, as seen
in several studies (Hosick et al., 1971; Manninen
et al., 1985; Samra et al., 1987; Griffiths and
Norman, 1993).
As predicted, higher concentrations of these
agents also affect the spinal cord. Changes in the
H-reflex (Mavroudakis et al., 1994) confirm the
effect at the spinal level. Depth electrode studies in
the spinal cord suggest that halothane and nitrous
oxide may have effects in lamina IVI and thereby
account for the changes seen in epidural recordings
and cervical spinal recordings from posterior tibial
nerve stimulation.
Some studies of the subcortical responses show
anesthetic effects appear to plateau at low concentrations consistent with a minimal effect on pathways
without synapses. For example, the major latency
increase often occurs at 0.51% inspired isoflurane
with minimal effects at higher concentrations. These
suggest that the effect on cortical responses has a
marked effect above concentrations where the animal
falls asleep. This narrowed effect range has also been
observed in humans. Hence the effect in the evoked
responses correlate with the clinical effects on the
cortex and are consistent with a synaptic effect mediating both the effects of sedation as an anesthetic
effect and mediating the cortical evoked response
1.5
1.8
10
20
30
Time post stimulation (ms)
40
50
Fig. 3. Somatosensory evoked potential (SEP) responses

recorded in the epidural space of a baboon following posterior tibial nerve stimulation at various concentrations of
isoflurane.
99
Isoflurane % ET
0.25
Amplitude (mV)
0.50
0.75
1.00
1.25
1.50
10
15
20
25
30

recorded over the cervical spine of a baboon following posterior tibial nerve stimulation at various concentrations of
isoflurane.
(Fig. 6) shows a progressive increase in effect as the

number of synapses increases along the auditory
pathway, with a substantial increase in the effect
at the cortical level (Fig. 7). In fact, the effect of
anesthesia on the cortically generated MLAEPs,
is the basis of a device for monitoring the state of
anesthetic-induced unconsciousness (Plourde, 2006).
The effect on the visual evoked response is among
the most dramatic, perhaps also due to the multiple
synapses involved (Sebel et al., 1986) (Fig. 8).
Since the effect of the anesthetic agents on
synapses will also reduce the EEG activity, the SEP
responses are often recordable even when the EEG
Isoflurane % ET
0.3
Amplitude (mV)
0.6
0.9
1.2
Relative Amplitude
1
0.9
1.5
1.8
0.8
0.4
0.3
0
1
1.5
0.5
Isoflurane Conc (%)
10
20
30
40
50

recorded over the parietal cortex of a baboon following
posterior tibial nerve stimulation at various concentrations
of isoflurane. The inset on these graphs shows that the
decrease in amplitude is nonlinear (occurring over a rather
narrow range of concentrations).
Fig. 6. Influence of isoflurane on auditory brainstem

response (ABR). Latency of peaks III and IVV are
increased at 1.0% but plateau with increasing anesthetic
depth. Reproduced with permission from Manninen et al.
(1985) with permission from Lippincott Williams and
Wilkins.
is suppressed. This effect has been observed with

sevoflurane and the first cortical components are
enhanced at low stimulation rates (Jantti et al.,
1998; Rytky et al., 1999). Shown in Fig. 9, the stimulus to the median nerve (spike in lower marker
trace) induces an enhanced N20 cortical wave seen
regularly after every stimulus with later waves
almost totally abolished. With repeated stimuli during EEG suppression, the evoked response (N20) to
median nerve stimulation is seen following each
stimulus, but adapts strongly when stimulation rate
exceeds 1 Hz (Fig. 2). Due to the high signal-to-noise
ratio, that is, lack of high-amplitude EEG and electromyographic activity (EMG), single N20 responses are
visible (Fig. 9).
The synaptic model also helps explain the anesthetic effect in the motor pathways. Motor EPs are
susceptible to anesthetic agents at three sites. The
first is in the motor cortex. Stimulation of the motor
cortex pyramidal cells is either by direct stimulation
of these cells (leading to the production of
D waves) or indirect stimulation via internuncial
100
Fig. 7. Effect of increasing end-tidal isoflurane on the early

cortical components of the mid-latency auditory evoked
potential (AEP). The latencies of Pa and Nb increase, and
the amplitudes decrease, with increasing isoflurane concentration. Reproduced with permission from Heneghan et al.
(1987) # The Board of Management and Trustees of the
British Journal of Anaesthesia. Reproduced by permission
of Oxford University Press/British Journal of Anaesthesia.
Fig. 8. Visual evoked potentials during isoflurane anesthesia. Reproduced with permission from Sebel et al. (1986)
# The Board of Management and Trustees of the British
Journal of Anaesthesia. Reproduced by permission of
Oxford University Press/British Journal of Anaesthesia.
NTTI
neurons (leading to production of I waves). The D

waves are relatively unaffected by anesthetics since
no synapses are involved in their production (Deletis,
1993; Yamada et al., 1994; Stephen et al., 1996;
Gugino et al., 1997). I waves are generated through
synaptic activity, are substantially affected by anesthetics. This is seen in the spinal epidural responses
to motor cortex stimulation of Fig. 10.
Studies comparing motor EPs from transcranial
magnetic (tcMMEP) and electric (tcEMEP) stimulation suggest that the magnetic technique can be more
sensitive to the inhalational agents because magnetic
stimulation (especially weaker field strengths) rely
more on transsynaptic activation of the cortex (Sloan
and Angell, 1993). High magnetic strength tcMMEP
(which produces D waves) appears to overcome this
difference. The synaptic dependence of tcMMEP
likely also relates to the type of current pulse driving
the magnetic coil since biphasic or rapidly attenuated sine wave pulses may be more effective than
monophonic pulses on the production of D waves
(Taniguchi et al., 1993a; Loughnan and Fennelly,
1995).
The second site of anesthetic action in the motor
pathway is in the anterior horn cell. At this location,
the D and I waves summate temporally. If they
are able to reach threshold, the anterior horn cell
depolarizes producing a peripheral nerve action
potential. Partial synaptic blockade here, induced
by anesthetics, may make it more difficult to reach
threshold. The combined effect of anesthetics to
block I waves from the cortex, and synapses at
the spinal cord further reduce the probability of generating a compound muscle action potential (CMAP).
At higher anesthetic doses, an even more profound
synaptic block at the anterior horn cell may prohibit
synaptic transmission regardless of the composition
of the descending spinal cord volley of activity. This
has suggested that the most prominent anesthetic
effect on tcEMEP is at the a-motoneuron cell level
(Loughnan et al., 1989; Zentner et al., 1992).
Since this is a location for the anesthetic-induced
effect associated with lack of movement in response
to pain (MAC), it also explains the nonlinear
decrease in muscle responses associated with the
induction of anesthesia. Hence, motor EPs (MEP)
recorded in muscle (myogenic) are among the most
easily abolished evoked responses by halogenated
inhalational agents (Fig. 11). Single pulse stimulation
transcranial motor evoked myogenic potentials
(tcEMEP) appear to be so easily abolished by
101
Fig. 9. Evoked responses to median nerve stimulation in sevoflurane-induced suppression. The stimulus (spike in lower
marker trace) induces the enhanced N20 cortical wave, and the later waves are almost totally abolished. This is seen regularly after every stimulus. The burst, seen 200300 ms later does not follow every stimulus, but it is a nonlinear (onoff)
response probably induced by arousal mechanisms. Montage P3C3 for upper trace. Reproduced from Jantti et al. (1998)
with permission from the International Federation of Clinical Neurophysiology.
inhalational agents that they are often unrecordable

in the presence of these agents (Kalkman et al.,
1991b; Stone et al., 1992). When recordable, the
major effect may occur at low concentrations (e.g.,
less than 0.20.5% isoflurane) with loss of responses
above 0.30.5 MAC (Haghighi et al., 1990a,b;
Woodforth et al., 1996).
Because of the resistance of the D wave, the anesthetic effect at the a-motoneuron cell can be partially
overcome at low concentrations by multiple pulse
transcranial stimulation (Taylor et al., 1993; Machida
et al., 1995). In this circumstance, the multiple D
waves formed (and I waves if produced) summate
at the a-motoneuron resulting in a peripheral nerve
and motor response when cortical stimuli are placed
at an interstimulus interval (ISI) of 25 ms (Taniguchi
et al., 1993a; Taylor et al., 1993 As a consequence,
low concentrations of inhalational agents appear
acceptable when high-frequency transcranial stimulation is used in some patients with robust responses
(Kawaguchi et al., 1996; Pechstein et al., 1998;
Ubags et al., 1998). Alternatively, the anesthetic
effect can also be partially overcome by activation
of the H-reflex by peripheral nerve stimulation combined with transcranial stimulation (Taniguchi et al.,
1991) or by stimulation of the foot sole within the
receptive field of the withdrawal reflex of the tibialis
anterior muscle preceding the cortical stimulus by
50100 ms. This cutaneous input provides a spatial
facilitation of the cortically elicited response yielding
a larger and more reliable motor response (Andersson
and Ohlin, 1999).
The third site in the motor pathway is at the neuromuscular junction. Fortunately, with the exception
of neuromuscular blocking agents, anesthetic drugs
have little effect at the neuromuscular junction.
Isoflurane % ET
0.3
Isoflurane % ET
0.3
0.6
0.9
1.2
1.5
1.8
2.1
10
15
20
25
30
Fig. 10. Transcranial electrical motor evoked potential

(tcEMEP) responses recorded in the epidural space of a
baboon at various concentrations of isoflurane.
Amplitude (mV)
Amplitude (mV)
0.6
0.9
1.2
1.5
1.8
10
15
20
25
30
Fig. 11. Transcranial electrical motor evoked potential

(tcEMEP) responses recorded in the hypotenar muscles of
a baboon at various concentrations of isoflurane.
102
Muscle relaxants are preferred when there is recording from the epidural space or peripheral nerves but
should be controlled carefully when monitoring
recordings from muscles (and not used when recording spontaneous or mechanically elicited muscle
responses).
Studies with evoked responses produced by spinal
or epidural stimulation show minimal effects of
anesthesia on recording near the peripheral nerve
(neurogenic) or myogenic responses suggesting the
neurophysiology of the electrical activity arriving at
the a-motoneuron cell is different than from cortical
stimulation (Russell et al., 1994; Schwentker et al.,
1995; Bernard et al., 1996; Owen, 1997; Jou et al.,
2003a). Machida studied the responses in the peripheral nerve and muscle following epidural spinal
cord stimulation in the cat (Machida et al., 1995).
He noticed that single pulse stimulation produced a
response that was eliminated by pentobarbitol, by
low-dose isoflurane and by posterior column transection (but not lateral column transection). When a pair
of stimuli was used (ISI: 15 ms), a new complex in
the peripheral nerve response was seen. This complex and the CMAP were eliminated only by highdose isoflurane or by lateral spinal cord transection.
Machidas study suggests that the type of spinal cord
stimulation and the anesthetic used may alter the
balance of sensory and motor contributions to the
peripheral nerve and muscle response of spinal stimulation. Of interest is that the sensory tracts were
more easily stimulated than motor tracts. Recent
studies suggest that with isoflurane anesthesia, the
motor component is preferentially blocked, perhaps
by interaction at the synapses at the a-motoneuron
cell or by differential effects on conduction in the
spinal tracts in humans (Deletis, 1993). Based on
these studies, it is conceivable that spinal stimulation
techniques may monitor a mixture of sensory and
motor pathways that may change with the type and
dosage of the anesthetic agents used (Machida
et al., 1985; Kai et al., 1993).
The inhalational agents have differing profiles.
Since the anesthetic potency of inhalational agents
have been traditionally assessed by MAC (the minimal alveolar concentration when 50% of subjects
move in response to a painful skin incision), studies
have been conducted comparing the different agents
in their effect on SEP and TcEMEP at equi-MAC
values. This results in a relative potency based on
MAC equivalents in the order nitrous oxide (most
potent) > isoflurane sevoflurane desflurane >
NTTI
enflurane > halothane (McPherson et al., 1985; Salzman et al., 1986; Pathak et al., 1989; Thornton et al.,
1992; Lam et al., 1994). This difference in effect on
the cortex and spinal cord may explain why the cortical effects of the agents differ at equivalent MAC
levels (Rehberg et al., 1998). This difference in effect
has been seen in one study where desflurane
depressed the thalamocortical SEP amplitude more
than sevoflurane (Vaugha et al., 2001).
The other main difference between these agents is
their solubility in tissues (halothane > enflurane >
isoflurane > sevoflurane > desflurane). The more
the agent is insoluble, the more rapidly the concentration (and response effect) can be changed
(Ku et al., 2002). Hence, desflurane may have a faster onset of effect when introduced into an anesthetic.
Conversely, some studies have compared the
effects of the agents using comparable levels of the
cortical effect using processed EEG bispectral index
(BIS). When the BIS was adjusted to 60, the cortical
amplitude of the posterior tibial nerve SEP was
greater with isoflurane than with desflurane (Fletcher
et al., 2005). These studies suggest that the inhalational agents do not share equivalent profiles on all
of their various anesthetic effects similar to the dissimilarities in changes in cerebral physiology.
Consistent with the depression of movement by a
spinal action of anesthetics, studies of spinal reflexes
with inhalational agents also demonstrate depression.
One study compared transcranial motor evoked
responses with the H-reflex and F-wave (Kammer
et al., 2002). Sevoflurane was studied at subanesthetic concentrations (0.2% and 0.4% inhaled) where
the subjects were sedated but arousable at the higher
concentration (thought to be two-thirds of the concentration producing sleep). The study observed significant amplitude reductions of the spinal cord
responses (F response and H-reflex) to an extent
much less than recordings from the cortex (i.e., alteration in the processed EEG (BIS) and amplitude
reduction of the mid-latency auditory evoked
response). When the amplitude of the CMAP of the
transcranial MEP was reduced to 50%, the F-wave
amplitude was decreased by 40%, the H-reflex by
22%, the BIS by 7%, and the mid-latency auditory
evoked response was unchanged. In these studies,
the M wave was unaffected confirming the minimal
effect of low concentrations of inhalational agents
on the neuromuscular junction and peripheral nerve
conduction (Pereon et al., 1999). A similar study
using isoflurane also demonstrated the depression
with the MEP being more than the F-wave (Zhou and
Zhu, 2000).
This relative difference in sensitivity of the tcEMEP and F-wave and the knowledge that the
inhalational agents have minimal effects on axonal
conduction in nerve fibers (Bosnjak et al., 1982;
Berg-Johnsen and Langmoen, 1986) suggest the
inhalational agents decrease spinal motor neuron
excitability, perhaps through cortical effects. Further,
the prolongation of the tcEMEP but not F-wave
suggests suppression of synaptic transmission in the
polysynaptic motor pathways with the effect on
the tcEMEP being a possible combination of these
effects (Zhou and Zhu, 2000).
103
ON
5
10
15
OFF
19
5.1.2. Nitrous oxide

10
TIME
Nitrous oxide (N2O) is generally considered a weak

anesthetic compared to the potent inhalational agents
(based on MAC it is about 1/100th as potent). It is
believed to have actions of antagonizing the NMDA
receptor, inhibiting the neuronal nicotinic acetylcholine receptor, and exhibiting opioid-like effects on
the opioid receptors. Some of its actions may be
mediated through alpha2 adrenoreceptors, especially
in the locus coeruleus which has efferent neural connections to the thalamus and cerebral cortex (Ohara
et al., 1997). Nitrous oxide is a more potent depressant of the P15N20 SEP response than isoflurane
(Thornton et al., 1992). It has been postulated that
since this response is generated in the pontine thalamic region of the brain and the locus coeruleus
projects to this area, this may account for the difference between the inhalational agents and nitrous
oxide (Thornton et al., 1999).
The effects of nitrous oxide vary with the other
anesthetic agents being employed. When used alone,
nitrous oxide tends to produce graded amplitude
and latency changes in a dose-dependent manner
(Fenwick et al., 1979; Benedetti et al., 1982; Chapman et al., 1982; Harkins et al., 1982; Houston
et al., 1988; Zentner and Ebner, 1989), with minor
or no changes in subcortical responses (Peterson
et al., 1986; Schmidt and Chraemmer-Jorgensen,
1986). Because nitrous oxide is very insoluble, the
changes can occur rapidly as shown in Fig. 12. When
added to inhalational anesthetics, nitrous oxide may
cause additional changes in latency and amplitude
(Peterson et al., 1986) or have no apparent additive
effect (Manninen et al., 1985; Chi and Field, 1986).
Studies of equi-anesthetic mixtures of isoflurane
15
20
25
PTN
0
100
50% NITROUS OXIDE
Fig. 12. Effect of nitrous oxide on cortical recordings of

posterior tibial nerve somatosensory evoked potentials.
The amplitude of the response is markedly reduced over
the 1015 min following the introduction of nitrous oxide
and a return after agent is removed. Reproduced from
Sloan and Koht (1985) with permission by Lippincott,
Williams and Wilkins.
and nitrous oxide have demonstrated that the mixture

has a more potent effect on cortical SEP than would
be predicted by adding the individual effects of each
agent suggesting a synergism from different mechanisms of action (Sloan et al., 1995). In cases in which
nitrous oxide is added to intravenous agents, amplitude changes predominate, without latency change
(McPherson et al., 1985; Sloan and Koht, 1985;
Zentner et al., 1989; Schubert et al., 1990). Hence,
nitrous oxide may be more context sensitive in its
effects (i.e., the actual effect may vary with the other
anesthetics already present).
Despite its relatively weak anesthetic profile,
studies with tcMMEP (Firsching et al., 1991) and
tcEMEP (Jellinek et al., 1991b) show that nitrous
NTTI
104
5.2. Intravenous analgesic agents

Because the inhalational anesthetic agents have
marked depressant effects on cortical EPs and motor
EPs, anesthesiologists frequently choose intravenous
analgesics (opioids or ketamine) supplemented with
intravenous sedative agents (e.g., propofol) when
monitoring is required. The goal of a complete
anesthetic is to use a mixture of agents to provide
analgesia (pain relief), sedation, amnesia, and muscle
relaxation (in some circumstances).
Cervical
Cortical
0
0
1
2
3
4
5
6
7
8
9
Time after sufentanil injection (min)
Time after sufentanil injection (min)
oxide produces depression of myogenic responses

(Zentner et al., 1989; Jellinek et al., 1991b; Woodforth et al., 1996; Pechstein et al., 1998). When compared at equi-MAC anesthetic concentrations, nitrous
oxide produces more profound changes in myogenic
tcEMEP than any other inhalational anesthetic agent
(Sloan, 1997). However, one study suggests that
nitrous oxide is usually acceptable when used in
concentrations below 50% (Jellinek et al., 1991b).
Some studies have suggested nitrous oxide may be
acceptable for monitoring with multipulse stimulation techniques; however, the other anesthetics used
with it make a difference in the degree of depression
(Van Dongen et al., 1999a,c; Sakamoto et al., 2001).
Like the halogenated agents, the effects on the epidurally recorded MEP are minimal.
12
15
18
21
0
1
2
3
4
5
6
7
8
9
12
15
18
24
21
24
27
30
27
30
2 V
0
25
Time (ms)
60
Time (ms)
Fig. 13. Changes in median nerve cervical and cortical

somatosensory evoked potential (SEP) recording with time
in one patient after sufentanil 5 mg/kg. Two baseline
recordings at time zero are shown. Reproduced with permission from Kimovec et al. (1990) # The Board of Management and Trustees of the British Journal of Anaesthesia.
Reproduced by permission of Oxford University Press/
British Journal of Anaesthesia.
5.2.1. Opioid agents

Opioids (e.g., fentanyl, alfentanil [AlfentaW], sufentanil [SufentaW], morphine, meperidine [DemerolW],
remifentanil [UltivaW]) provide excellent analgesia
for anesthesia. The effect of the opioid analgesics
on evoked responses is generally mild. The difference between the opioid agents and the inhalational
agents likely is the result of opioid action on the opioid receptor pathways rather than the GABA and
NMDA pathways. This difference in action also
explains that opioids produce less sedation or unconsciousness compared to the inhalational and sedative
agents (below), consistent with fewer effects at the
thalamus and sensory gating.
As with minimal depression of the EEG, opioid
effects on sensory and motor evoked responses are
minimal on spinal or subcortical recordings. Depression of amplitude and increases of latency in cortical
responses and occasional loss of late cortical peaks
(over 100 ms) is seen at higher doses which produce
some sedation (Ghaly et al., 1991b; Gugino et al.,
1992; Kalkman et al., 1992b; Glassman et al.,

1993). Fig. 13 shows that a large dose of sufentanil
does produce some transient changes in the cortical
SEP which largely resolves as the drug is redistributed. The effects are reversed with naloxone, suggesting that the effect is a mu receptor effect (Velasco
et al., 1984; Chi et al., 1987; Lee, 1994).
The spinal application of morphine or fentanyl produces minimal changes in the SEP, H-reflex, or spinal
motor reflex (Chabal et al., 1988; Schubert et al.,
1990; Fernandez-Galinski et al., 1996; Van Dongen
et al., 1999a). Several studies have shown a minimal
depressant effect of clinical doses of opioids on the
tcEMEP (Levy et al., 1984; Zentner, 1989; Shields
et al., 1990; Firsching et al., 1991; Zentner, 1991a,b;
Kalkman et al., 1992b; Schmid et al., 1992;
Herdmann et al., 1993; Kalkman et al., 1993; Taniguchi et al., 1993b; Stinson et al., 1994; Yang et al.,
1994; Glassman et al., 1995; Jones et al., 1996; Lang
et al., 1996a; Nagle et al., 1996; Pechstein et al., 1996;

Stephen et al., 1996; Ubags et al., 1996; Watt et al.,
1996; De Haan et al., 1997; Gugino et al., 1997;
Morota et al., 1997; Owen, 1997; Calancie et al.,
1998; Pechstein et al., 1998). As a consequence of this
minimal effect, total intravenous anesthesia with
opioids and sedative drugs is often used when recording of responses is not possible in the presence of
inhalational agents.
5.2.2. Ketamine
An alternative analgesic to opioids and the inhalational agents is ketamine. A racemic mixture of a
phencyclidine derivative called ketamine acts by
decreasing NMDA receptor activity, inhibiting neuronal nicotinic acetylcholine receptors, decreasing
the presynaptic release of glutamate, and by opioidlike actions on the opioid receptors. It provides
excellent analgesia and hypnosis, but hallucinatory
activity and increases in intracranial pressure (ICP)
in patients with cortical abnormalities limits its usefulness. As seen in the EEG, ketamine is an excitatory agent (probably through its interaction at the
NMDA receptor) that may heighten synaptic function
rather than depress it. Ketamine has been reported to
increase cortical SEP amplitude (Schubert et al.,
1990; Schwender et al., 1993) (Fig. 14) and increase
the amplitude of muscle and spinal recorded
responses following spinal stimulation at doses that
105
do not produce spike and wave activity in the EEG

(Kano and Shimoji, 1974; Glassman et al., 1993;
Taniguchi et al., 1993b).
Ketamine has minimal effects been on ABR
(Cohen and Britt, 1982), cortical AEP (Schwender
et al., 1993; Schwender et al., 1996), VEP (Hetzler
and Berger, 1984), and in myogenic tcEMEP (Ghaly
et al., 1990a; Glassman et al., 1993; Kothbauer et al.,
1993; Kalkman et al., 1994; Ubags et al., 1997; Inoue
et al., 2002). Ketamine also increases the H-reflex
suggesting that a change in alpha motor neuron excitability may contribute to the tcEMEP enhancement
(Shimoji and Kano, 1973; Kano and Shimoji, 1974).
High dosages, however, produce depression of the
myogenic response consistent with its known property
of spinal axonal conduction block (Iida et al., 1997).
As such, ketamine has become a valuable adjunct
during some total intravenous anesthetic (TIVA)
techniques for recording muscle responses. In some
studies, ketamine has been used to reduce the dose of
other depressant sedatives in TIVA (e.g., propofol),
or used as the sole sedative agent with resulting
increase in SEP or tcEMEP (Agarwal et al., 1998;
Kawaguchi et al., 2000; Erb et al., 2005).
5.2.3. Dexmeditomidine
Another agent which produces analgesia and sedation
is dexmeditomidine (PrecedexW). As a central,
selective alpha2 adrenoreceptor agonist drug, it can
AMPLITUDE (2.5 V/div)
T30
T20
T15
T10
T5
T2
KETAMINE INDUCTION (2 mg/kg)
AWAKE
LATENCY (5 ms/div)
Fig. 14. Example of SEP waveforms before and after induction with ketamine at times 2, 5, 10, 15, 20, and 30 min.
Reproduced with permission from Schubert et al. (1990) with permission by Lippincott, Williams and Wilkins.
NTTI
106
provide analgesia, anxiolysis, hypnosis, and sedation.

It has been used for sedation of patients in intensive
care units where it provides sedation with preserved
neurologic examination on arousal. Side effects of
hypotension and bradycardia relate to its sympatholytic properties and limit the drug to a role as a supplement to other anesthetic agents. Dexmeditomidene
was studied as a supplement to isoflurane and it caused
no additional depression to the cortical mid-latency
auditory response and the cortical SEP (Thornton
et al., 1999). It has also been used as a supplement to
propofolfentanylnitrous oxide anesthetic where the
later cortical peaks (P25N35) of the SEP were affected
but the early cortical peak (N15P20) was unaffected
(Bloom et al., 2001) (Fig. 15).
Clonidine, also a central alpha1 and alpha2 agonist
(less alpha2 selective than dexmeditomidine) has also
been used as a supplement to opioid based anesthesia. Studies with it as an oral premedicant show no
significant changes in the interpeak latencies of the
ABR (Kumar et al., 1994) or on the median nerve
cortical SEP (Porkkala et al., 1998). Clonidine has
been given epidurally where minor changes in the
dermatomal evoked lumbar and sacral responses is
thought to be the result of the action of clonidine
on the dorsal root afferent neurons in the spinal cord
(Lund et al., 1989b).
5.3. Other analgesics
A variety of other intravenous drugs have been used
to produce analgesia. Tramadol (UltramW) is an analgesic compound that produces analgesia through opiate and adrenergic mechanisms. It is thought to exert
a local anesthetic type effect on peripheral nerves.
When studied in the rat by intrathecal administration,
it decreased the amplitude and increased the latency

of the SEP produced by sciatic nerve stimulation
and recording in the epidural space (Jou et al.,
2003b). It had similar effects on the CMAP recorded
after stimulation of the spinal cord via electrodes in
the interspinous space. Since these effects were
not reversed by naloxone (NarcanW), it indicates a
non-mu receptor effect, perhaps by a mechanism of
sodium channel blockade similar to local anesthetics
which is also seen with meperidine (FernandezGalinski et al., 1996; Jaffe and Rowe, 1996; Pang
et al., 1998).
5.3.1. Sedative-hypnotic drugs
In some patients, excellent anesthesia for cortical
evoked response recording can be provided with
analgesia from opioids or ketamine, supplemented
with nitrous oxide or low-dose inhalational agents.
However, in some patients, the depressant characteristics of these gaseous agents reduce the size of the
evoked response below that acceptable for monitoring (i.e., the desired response cannot be reliably
distinguished from background noise). In these circumstances, the anesthesiologist may choose to supplement with intravenous sedative agents rather
than the inhaled agents in a TIVA (e.g., opioids or
ketamine for analgesia and ketamine or sedativehypnotics for sedation).
5.3.2. Barbiturates
Barbiturates are thought to exert their synaptic
effects via the GABAA receptor and they have the
ability to upregulate the NMDA receptor desensitizing to stimulation. Studies demonstrate decreases
Fig. 15. Cortical somatosensory evoked responses recorded in a human during anesthesia (low-dose sevoflurane, 50% N20,
and a sufentanil infusion) before and after dexmeditomidine (1 mg/kg load followed by infusion 0.31 mg/kg/h). Courtesy of
Mary Sturaitus, MD.
in amplitude and increases in latency of cortical sensory responses with increasing effects on longer
latency waves and minimal effects on the brainstem
responses. Studies with thiopental (PentothalW), a
thiobarbiturate, demonstrate that the effect is minimal on subcortically recorded responses, with progressive effects on longer latency responses. For
example, ABR is virtually unaffected at doses of
pentobarbital that produce coma (Bobbin et al.,
1979; Cohen and Britt, 1982; Newlon et al., 1983).
Changes in the ABR are not seen until dosages are
sufficient to produce cardiovascular collapse (Marsh
et al., 1984). The SEP can similarly be recorded with
thiopental sufficient to produce a flat EEG (Newlon
et al., 1983; Drummond et al., 1985).
The tcMMEP, however, was more sensitive to
barbiturates, with effects of amplitude depression at
doses below that affecting the SEP and lasting for a
longer period of time after induction. Induction with
thiopental has significantly reduced (Sakamoto
et al., 2001) or eliminated the tcMMEP response for
as long as 4560 min (Glassman et al., 1993). An
infusion of thiopental sufficient to produce light
anesthesia abolished tcMMEP (Taniguchi et al.,
1993b). However, it has been successfully used in
some anesthetic regimes (Zentner, 1989, 1991a,b)
and given as intermittent boluses during the anesthetic (Zentner, 1991b). Methohexitol (BrevitolW)
may be unusual among the barbiturates in that it is
rapidly metabolized and activates seizure foci in
small doses. tcMMEP has been measured when it is
used in dogs, but human experience is not widely
published (Young et al., 1994).
5.3.4. Benzodiazepines
The benzodiazepines, notably midazolam (VersedW),
have been advocated as supplements to TIVA in routine
surgery because of excellent sedation and amnestic qualities (particularly to reduce the chance of hallucinogenic activity with ketamine). They are thought to
exert their effects via action at the synaptic and extrasynaptic GABAA receptors. Unlike the barbiturates, benzodiazepines have a less profound effect on the EEG
suggesting a different profile on the GABAA receptor.
Midazolam, in doses consistent with induction
of anesthesia and in the absence of other agents, produces a mild depression of cortical SEP (Koht et al.,
1988; Sloan et al., 1990). As with thiopental midazolam produces marked acute (Schonle et al., 1989;
Ghaly et al., 1990b; Kalkman et al., 1992b; Taniguchi
107
et al., 1993b) and prolonged depression of myogenic

tcMMEP (however, small doses for sedation appear
tolerated; Schonle et al., 1989; Ghaly et al., 1991a;
Zentner, 1991; Kalkman et al., 1992b; Scheufler and
Zentner, 2002).
In addition to possible cortical locations for the
benzodiazepine effect, an effect at the spinal cord
has been described as antinociceptive through actions
at the GABA receptors in laminae I and II in the
dorsal horn (Faull and Villiger, 1986; Crawford
et al., 1993). This action has been demonstrated by
a study of posterior tibial stimulation where diazepam produced a marked decrease in the amplitude
of the H-reflex with no effect on the M response
(Kaieda et al., 1981). Since the first peak of the electrospinogram was decreased, this is consistent with a
drug effect at the dorsal root. This effect has also
been seen with midazolam administered epidurally
in rabbits where higher doses caused significant
increases in the latency of the SEP responses
measured in the epidural space (Cicek et al., 2000).
5.3.5. Etomidate
As opposed to the barbiturates and benzodiazepines,
etomidate (AmidateW), an imidazole derivative, can
enhance synaptic activity at low doses, possibly by
changing the balance of inhibitory and excitatory
influences on neural pathways. It is also thought to
mediate its synaptic effects via the synaptic GABAA
receptors. At low doses, etomidate may produce seizures in patients with epilepsy (Rampil, 1997). This
effect has been used to enhance amplitude in both
sensory and motor evoked responses (Kochs et al.,
1986; McPherson et al., 1986; Russ et al., 1986; Koht
et al., 1988; Sloan et al., 1988; Langeron et al., 1997)
(Fig. 16). Fortunately, the enhancing activity occurs
at doses that are consistent with the desired degree
of sedation and amnesia needed for TIVA. This
amplitude increase appears coincident with the
myoclonus seen with the drug, suggesting a heightened cortical excitability (however, no evidence of
seizure activity was seen; Sloan et al., 1990). A sustained increase with constant drug infusion has been
used to enhance cortical recordings that were otherwise not monitorable (Sloan et al., 1988).
A cat study suggests that the location of enhancement is cortical (Samra and Sorkin, 1991) which is
consistent with clinical studies showing enhancement of cortical responses with no enhancement in
subcortical responses (Sloan et al., 1988). In one
NTTI
108
N22.5
N39
C3 F2
ZE:T
9 : 45
N22.5
10 : 05
N22.6
10 : 08
N22.5
N39
N23.7
10 : 12
N43
10 : 15 : 10 mg
ETOMIDAT i.v.
10 : 17
N23.1
N22.5
2 V
+
N42
10 : 21
N41
10 : 24
25
50
t (ms)
SOMATOSENSORISCH EVOZIERTE POTENTIALE (SEP)

NACH MEDIANUS STIMULATION
Fig. 16. Cortical somatosensory evoked potential (SEP) from median nerve stimulation before and following 10 mg etomidate. Note that the N20 wave and later cortical waves are enhanced. Reproduced from Russ et al. (1986) with kind permission of Springer Science and Business Media.
study, this amplitude enhancement had not resolved

when the patients awakened (Liang et al., 2004).
Higher doses of etomidate cause depression of the
evoked responses (similar to the EEG) suggesting a
biphasic effect (enhancement followed by depression). Since this would be inconsistent with a drug
effect at a single synaptic site, it suggests a modulation of the degree of depression or excitation of adjacent neural pathways on the SEP pathway. This
enhancement can also increase the amplitude of later
cortical waves; these would normally be depressed
by agents which decrease the cortical amplitude as
discussed above (Fig. 16).
Studies with transcranial elicited motor EPs have
suggested that etomidate is an excellent agent for
induction and monitoring of this modality (Lumenta,
1991; Kalkman et al., 1992b; Glassman et al., 1993;
Sloan and Levin, 1993; Taniguchi et al., 1993b; Yang
et al., 1994). Of several intravenous agents studied,
etomidate had the least degree of amplitude depression after induction doses or continual intravenous
infusion (Glassman et al., 1993). Latency (onset)
changes were not observed and amplitude
enhancement of muscle responses was not observed

except at very small dosages with depression at high
dosages (Sloan and Levin, 1993). This effect has also
been used to enhance amplitude in motor evoked
responses (Kochs et al., 1986; Sloan et al., 1988).
Etomidate is also unusual in that it depresses the
production of cortisol. This may not be an issue with
many surgeries in which steroid agents are given routinely. However, when not given as a part of the surgery, it is unclear if supplemental steroids should be
given when etomidate is used (Sloan et al., 1988).
Ketamine and etomidate are therefore unique
agents in the intravenous armamentarium, as they
have the ability to enhance cortical evoked responses
while contributing to anesthesia. It is interesting that
these two agents also increase H-reflex suggesting
a change in alpha motor neuron excitability (Kano
and Shimoji, 1974).
5.3.6. Propofol
Propofol (DiprovanW), an alkylphenol, is thought to
act on synapses via the GABAA receptor and
extrasynaptic GABAA receptors. The drug is very

rapidly metabolized such that the drug effect can usually be titrated down to levels compatible with adequate TIVA and MEP recording. It produces dosedependent depression of the EEG reminiscent of the
barbiturates and can produce burst suppression and
suppression at high doses, but low amplitude activity,
including spindles, can still be seen during suppression (Huotari et al., 2004).
Propofol induction produces amplitude depression
in cortical AEP (Savoia et al., 1988; Chassard et al.,
1989; Thornton et al., 1989; Raeder, 1996; Tooley
et al., 1996), VEP (Hamaguchi et al., 2005), and cortical SEP (Maurette et al., 1988; Freye et al., 1989;
Scheepstra et al., 1989) with rapid recovery after cessation of infusion (Fig. 17). Propofol does not generally appear to enhance cortical responses but one
report of propofol used without other major anesthetic agents demonstrated a 15% increase in cortical
SEP amplitude (Zentner et al., 1991). When the SEP
is recorded in the epidural space, propofol has no
109
significant effect. The latencies of the ABR were

increased without significant amplitude decreases
(Chassard et al., 1989; Purdie and Cullen, 1993).
This is consistent with the postulated site of anesthetic action of propofol on the cerebral cortex (Jellinek et al., 1991a; Angel and LeBeau, 1992; Kalkman
et al., 1992b; Keller et al., 1992; Taniguchi et al.,
1993b; Pechstein et al., 1998).
Another advantage of propofol compared to the
inhalational agents is that the dose response curve
on the SEP is substantially flattened. As indicated
above, the dose response curve of the inhaled agents
is nonlinear with a marked effect occurring over a
small range. Studies in rats indicate a more gradual
change in cortical SEP amplitude with propofol
blood concentrations suggesting more latitude in the
titration of propofol in a TIVA infusion technique
(Logginidou et al., 2003).
In addition to the practical experience with propofol which has shown its utility, studies have been done
to compare propofol with the inhaled agents. When
Fig. 17. Cortical (mid-latency) auditory evoked potential (AEP) before anesthesia and at different concentrations of propofol. Arrows indicate the position of waves V, Pa, and Nb. Reproduced with permission from Chassard et al. (1989) # The
Board of Management and Trustees of the British Journal of Anaesthesia. Reproduced by permission of Oxford University
Press/British Journal of Anaesthesia.
NTTI
110
the BIS is used as an endpoint for anesthesia adjustment, isoflurane produced more depression of the
cortical SEP amplitude from posterior tibial nerve
stimulation (Chen, 2004). In this study, the effect of
propofol plateaued below a BIS of 60 (where the
effect was similar on latency and amplitude), whereas
the effect of isoflurane continued to become more profound and the depression from isoflurane markedly
diverged from propofol. Similar findings were seen
with sevoflurane and propofol (delivered by a target
controlled infusion) (Boisseau et al., 2002). Another
study compared the effect of propofol versus isoflurane on the cortical SEP from posterior tibial nerve
stimulation when the BIS was held between 40 and
50 (Liu et al., 2005). This study demonstrated a significantly lower cortical amplitude, higher cortical
latency, and greater variability with isoflurane.
Another study compared propofol infusion with
0.40.6% isoflurane with and without 70% nitrous
oxide in patients undergoing scoliosis surgery with
SEP and BIS. Here, the level of cortical amplitude
was higher with propofol despite a lower BIS (44 vs.
6162) suggesting a superiority for the intravenous
agent (Clapcich et al., 2004). In the rat, the effect of
propofol on the amplitude of the SEP was nonlinear
with minimal depression of the cortical response at
20 mg/kg/h and near maximal effect at 60 mg/kg/h
with essentially no effect below 20 and no additional
effect above 60 (Logginidou et al., 2003).
Also consistent with a cortical effect of propofol
has been the observation that only very high concentration of propofol (9 mg/ml) cause depression of the
H-reflex in humans (Kerz et al., 2001). Similarly, the
M wave was unaffected. Other studies have observed
a dose-dependent decrease in the H-reflex amplitude
and F-wave with propofol concentrations in the clinical range (Kammer et al., 2002; Kakinohana et al.,
2005; Baars et al., 2006a,b). It is of interest that
the suppression of the F-wave occurs at much lower
concentrations of propofol (50% suppression at
1.5 mg/l) than the suppression of the BIS (50%
suppression at 3.3 mg/l) confirming the suggestion
that the action on the F-wave is likely by a different
mechanism than the cortical effect giving rise to the
BIS change (Baars et al., 2006b).
Consistent with the depression of movement by a
spinal action of anesthetics, studies of spinal reflexes
with low-dose propofol parallel those mentioned above
with sevoflurane (Kammer et al., 2002). Hence, propofol at subanesthetic concentrations can depress the spinal reflexes with minimal cortical effect. This action is
believed to be depression of spinal neuronal excitability

by suppressing L-type calcium channel plateau potentials through potentiation of GABAA receptors (Guertin
and Hounsgaard, 1999; Dong and Xu, 2002).
Studies with transcranial electric or magnetic elicited motor EPs have demonstrated a depressant effect
on the F-wave and CMAP response amplitude, also
consistent with a cortical effect (Kalkman et al.,
1992b; Keller et al., 1992; Taniguchi et al., 1993b). Propofol has been used in tcEMEP when the recordings are
epidural (Loughnan et al., 1989). As a component of
TIVA, induction of anesthesia can include propofol
(Pechstein et al., 1998) and infusions of propofol have
been combined with opioids (Jones et al., 1996; Pechstein et al., 1996, 1998; Calancie et al., 1998). However,
as a component of TIVA, infusions of propofol have
been combined with opioids and produced acceptable
conditions for myogenic tcEMEP monitoring, especially when a multipulse stimulation technique is used
(Jones et al., 1996; Pechstein et al., 1996, 1998; Calancie et al., 1998). Studies comparing a propofol TIVA
with isoflurane with nitrous oxide have demonstrated
the superiority of the TIVA technique (Pechstein
et al., 1998). In propofol monoanesthesia, tcEMEPs
can usually be recorded still at burst suppression level,
although their amplitude is lower than that at lighter
levels (MacDonald et al., 2005).
Although acceptable recording conditions can be
obtained with evoked responses, higher doses in
man and animals have depressed the responses so
that recording is not possible (Logginidou et al.,
2003). Because of this, some TIVA methods have
used ketamine to provide additional sedation so that
the dose of propofol can be reduced into an acceptable range (Kawaguchi and Furuya, 2004). Ketamine
also provides some analgesia in this regime and may
produce some response enhancement as noted above.
5.3.7. Droperidol
Droperidol (InapsineW) is a butyrophenone, and is a
potent D2 (dopamine receptor) antagonist with some
histamine and serotonin antagonist activity. It has little
effect on the EEG, but when combined with fentanyl
(neurolept anesthesia), it increases EEG alpha activity at low doses and it produces high-amplitude beta
and delta activity. Its anesthetic action is unknown
but it is believed to interact at the GABAA and neuronal nicotinic acetylcholine receptor (Flood and Coates,
2002). Droperidol has been used successfully during EP
monitoring. It appears to have minimal effects when
5.4. Local and regional anesthesia

Regional anesthesia with local anesthetics blocks
conduction in the neural pathways affected causing
loss of EPs. Epidural anesthesia with bupivicaine
(at L34) has been studied with posterior tibial SEP
and dermatomal stimulation at T10, L1, and S1 (Lund
et al., 1987; Lund et al., 1989a; Dahl et al., 1990;
Loughman et al., 1995). Changes were seen with
0.5% and 0.75% but not with 0.25%. The effect is
similar to studies with 2% lidocaine (Loughnan
et al., 1990). Loss of response has also been seen
with intravenous regional block (Lang et al., 1993),
specific nerve blocks (Benzon et al., 1986), thoracic
paravertebral blocks (Richardson et al., 1998), as
well as the topical application in areas where pain
is evoked cutaneously (Svensson et al., 1993). In addition to the specific effects of local anesthetics on
regional nerve function, systemically infused local
anesthesia agents can cause an effect probably due
to effects on sodium ion channels when given in a
high dose (Javel et al., 1982; Schubert et al., 1992).
5.5. Muscle relaxants
Since muscle relaxants have their major site of action
at the neuromuscular junction, they have little effect
on electrophysiologic recordings that do not derive
from muscle activity. Muscle relaxants are generally
thought to have no effect on the sensory evoked
responses (Domino and Corssen, 1964; Harker et al.,
1977; Sloan, 1994). In fact, SEPs may actually
improve with muscle relaxation because EMG interference is reduced in electrodes near muscle groups
such as over the cervical spine. This benefit to recording is also seen in recording transcranial motor EPs
with muscle relaxation
Amplitude (mV)
combined with an opioid on SEP (Bertens, 1988), VEP

(Russ et al., 1982), and tcMMEP (Ghaly et al., 1991b;
Kalkman et al., 1994). The use of a droperidol-opioid
(neurolept) technique has the additional advantage
of not depressing cortical seizure activity (thus
making it useful for seizure focus identification and
ablation). It appears to have minimal effects on myogenic tcMMEP when combined with opioids (Herdmann et al., 1993; Taniguchi et al., 1993b).
Recently, warnings have emerged regarding malignant
ventricular arrhythmias and torsade-de-pointe with the
use of droperidol (especially in patients with prolonged QT interval on the ECG) further reducing the
use of this drug in anesthesia.
111
without muscle relaxation
10
15
20
25
30
Fig. 18. Transcranial electrical motor evoked potentials

(tcEMEP) responses recorded in the epidural space of a
baboon at various concentrations of isoflurane with (upper)
and without (below) neuromuscular blockade.
epidurally where paraspinous muscle activity can

obscure recording and when recording neurogenic
responses from spinal stimulation (Schwentker et al.,
1995). This is true for epidural or peripheral nerve
recordings where the activity of overlying muscle
obscures the response from transcranial or spinal stimulation. For recording of epidural or neurogenic
responses, complete or near-complete neuromuscular
blockade is highly desirable (Levy et al., 1984; Rodi
et al., 1996; Stephen et al., 1996). Fig. 18 shows
recording from the epidural space from tcEMEP with
(top) and without (bottom) muscle relaxation. Note
the muscle artifact obscures the identification of I
waves. This effect may be responsible for the enhancement seen with low doses of propofol (Zentner et al.,
1991) and meperidine (Anonymous, 1980).
Certainly, complete neuromuscular blockade will
prevent recording of muscle responses (CMAP) during MEP. However, partial neuromuscular blockade
has the benefit of reducing a substantial portion of
the movement which accompanies the testing and
may facilitate some surgical procedures where muscle relaxation is needed for retraction of tissues. In
these cases, careful monitoring of the blockade of
the neuromuscular junction is critical in the muscles
being monitored (since not all muscles will respond
identically to the same dose of muscle relaxants).
When neuromuscular monitoring is quantitated
using the amplitude of the CMAP (T1) produced by
supramaximal stimulation of a peripheral motor
nerve (M response), successful monitoring of
tcEMEP myogenic responses have been accomplished
at 515% (Oro and Haghighi, 1992; Van Dongen et al.,
NTTI
112
It is important to note that the use of neuromuscular blockade is controversial during monitoring of
muscle responses from mechanical stimulation of
nerves and partial paralysis may reduce the ability
to record these responses (e.g., facial nerve monitoring or monitoring for pedicle screw placement). One
study of vocalis muscle monitoring (Streinzer et al.,
1986) suggested that the effect of vecuronium was
nonlinear with the response of the vocalis muscle
being reduced to 50% of the baseline when the twitch
height of the evoked adductor pollicis response was
20% of baseline under various degrees of neuromuscular blockade using accelerometry.
5.6. Physiological considerations
In addition to the specific action of anesthetic
agents, the intraoperative management of patients
may have physiological changes induced by anesthetic agents or surgery. Some of these changes
are associated with changes in evoked responses.
For example, numerous studies (Branston et al.,
1974, 1976; Astrup et al., 1977; Brierley and
Symon, 1979; Symon et al., 1984; Symon and
Murota, 1989) have demonstrated a threshold relationship between regional cerebral blood flow
(CBF) and cortical evoked responses. Although
clinical function becomes abnormal at about
25 cm3/min/100 g (the normal is 50 cm3/min/
100 g), electrical function generally remains
normal when the CBF exceeds the functional
threshold of about 22 cm3/min/100 g (Fig. 20).
1.2
NORMAL
ELECTRICAL ACTIVITY
0.6
M wave
0.4
0.2
(1820)
tcEMEP CMAP
0.8
ERE
1.0
ALT
Relative amplitude of CMAP or M wave
1999b), 10% (Nagle et al., 1996; Scheufler and

Zentner, 2002), 15% (Hargreaves and Watt, 2005),
1025% (Shields et al., 1990; Stinson et al., 1994),
20% (Herdmann et al., 1993; Glassman et al., 1995;
Lang et al., 1996a,b; De Haan et al., 1997), 25%
(Ubags et al., 1996), 3050% (Yang et al., 1994;
Lee et al., 1995; Gugino et al., 1997; Van Dongen
et al., 1999b), and 8090% (Kalkman et al., 1992b;
Herdmann et al., 1993; Tabaraud et al., 1993; Stinson
et al., 1994; Lee et al., 1995) of T1 compared to baseline. When neuromuscular blockade is assessed by
comparing the ratio of the fourth to the first twitch
when stimulated at a rate of 2 Hz (called a train of four
response), acceptable CMAP monitoring has been
conducted with only two of four responses remaining
(Pechstein et al., 1996; Calancie et al., 1998).
Although recording of myogenic responses is
possible with partial neuromuscular blockade, the
amplitude of the CMAP will be reduced by the
blockade. Studies suggest the actual reduction varies
from a linear reduction paralleling the %T1 to a
slightly decreased rate of reduction (Sloan and Erian,
1993a,b) (Fig. 19). As a consequence of the amplitude reduction, the ability to record with partial neuromuscular blockade will be dependant on the
neurological pathology in the pathway monitored that
may reduce the baseline CMAP response. This
reduction can impact on pedicle screw testing; one
study of pedicle screw stimulation suggests that
neuromuscular blockade exceeding 80% reduction
of the single twitch may falsely elevate stimulation
thresholds (Minahan et al., 2000).
SEP
ABSENT
0
0
0.6
0.8
0.2
0.4
Fractional amplitude of single twitch (T1)
Fig. 19. Comparison of the M wave from peripheral nerve

stimulation at various degrees of neuromuscular blockade
(dashed line) to the amplitude of the compound muscle
action potential (CMAP) from transcranial electrical
stimulation.
EEG
cc/min/100 gm
10
20
30 50
REGIONAL CEREBRAL BLOOD FLOW
Fig. 20. Relationship between the somatosensory evoked

potential (SEP) and electroencephalogram (EEG) electrical
response and regional cerebral blood flow. Reproduced
from Sloan (1985) with permission from Oxford University
Press.
Below this level, the EEG and SEP become abnormal (Sloan 1985; Florence et al., 2004). SEP
amplitude is reduced by desynchronization of
the responses or a loss of functional neurons. Below
this level, the impact on electrical function becomes
more profound with decreases in SEP amplitude
between 16 and 20 cm3/min/100 g. At lower levels,
electrical activity is lost (EEG at 715 cm3/min/
100 g and SEP lost between 12 and 15 cm3/min/
100 g). In some studies, the SEP is lost at CBF about
20% below the level which produces an isoelectric
EEG (Prior, 1985; Nuwer, 1988).
In some experimental studies, hypotension to less
than 40 mm Hg or ischemia to 2025% of normal
blood flow has not been associated with SEP changes
(Laschinger et al., 1988). However, in humans during
surgery, SEP changes have been observed at blood
pressures which would not ordinarily be associated
with neural ischemia (e.g., systolic blood pressures
above 90 mm Hg systolic) (May et al., 1996). This
has been thought to be the result of operative
mechanical stress combined with the blood pressure
reduction leading to a more profound effect than
predicted by blood pressure alone (Seyal and Mull,
2002). In several of these patients, an increase
in blood pressure restored the response (Brodkey
et al., 1972; Griffiths et al., 1979; Dolan et al.,
1980; Wiedemayer et al., 2002).
In addition to systemic hypotension, regional
hypoperfusion can be detected by the evoked
response if it involves the neural pathway generating
the response. Examples include peripheral nerve
ischemia from positioning, tourniquets or vascular
interruption (Yamada et al., 1981; Grundy et al.,
1982a,b; Mahla et al., 1984; McPherson et al.,
1984; Fava et al., 1988; Witzmann and Reisecker,
1989; North et al., 1991), spinal cord ischemia from
aortic interruption or mechanical distortion, carotid
artery interruption (Russ and Fraedrich, 1984), vertebrobasilar insufficiency aggravated by head extension, cerebral artery constriction by vasospasm, and
cerebral ischemia due to retractor pressure (Symon
and Murota, 1989).
Studies of ischemia and anoxia in peripheral
nerves show an action potential amplitude decrease
to 5060% of baseline during the first 20 min that
is thought to represent temporal dispersion with
slowing of the fast and slow conducting fibers rather
than conduction block (Laschinger et al., 1988).
These amplitude changes vary with the degree of
ischemia. Since the dorsal column pathways are a
113
continuation of the peripheral nerves, these effects

may explain the decline of SEP amplitude with
spinal cord ischemia. With anoxia, a latency increase
in peripheral nerves occurs within 1520 min. However, after 1 h of anoxia, the conduction velocity in
the spinal cord increases suggesting a preferential
loss of slow conducting pathways.
Both the white and gray matter of the spinal cord
is vulnerable to ischemia and anoxia. In studies of
animals which developed paraplegia following spinal
cord ischemia, the animals either had both gray and
white matter lesions or just white matter lesions with
prolonged SEP conduction time (Follis et al., 1993).
Once injured, there is usually more recovery in the
white matter than the gray matter. Compound action
potentials in myelinated axons (such as the dorsal
columns) are attenuated rapidly with anoxia and disappear within minutes (Waxman et al., 1991).
Demyelinated axons appear to have more resistance
to ischemia and appear to recover faster. Demyelinated peripheral nerves also appear more resistant
to ischemia (Imaizumi et al., 1998). This suggests
that the SEP may be a sensitive pathway to detect
spinal cord ischemia; however, if it includes pathology with demyelination it may be less sensitive than
other pathways.
Another physiological variable affecting the
evoked responses is raised ICP. Several studies have
shown that reductions in amplitude and increases in
latency of cortically generated visual, somatosensory,
and auditory evoked responses occur with increasing
ICP. ABR responses are altered as uncal herniation
occurs (Nagao et al., 1978). SEP has been used to
guide cerebrospinal fluid (CSF) pressure management
during thoracic aorta procedures to reduce the risk to
the spinal cord (Oka and Miyamoto, 1987; Grubbs
et al., 1988; Maeda et al., 1989). The relationship of
the VEP to ICP has suggested the VEP as a means of
noninvasive ICP testing (York et al., 1981).
Hypoxia (similar to cerebral ischemia) is associated with SEP latency increase and amplitude
decrease until the responses are lost (Colin et al.,
1978; Branston et al., 1984; Koscielniak-Nielsen
et al., 1998). Cortical SEP appear more sensitive to
hypoxemia than the EEG and are more sensitive than
the subcortical and spinal responses (Kayama, 1974;
Kobrine et al., 1980; Iwayama et al., 1986). The
ABR is unaffected with PaO2 levels as low as
60 mm Hg or O2 saturation levels of 45% (Mosko
et al., 1981; Samra et al., 1984); however, acute
hypoxemia to a PaO2 of 2030 mm Hg increased
114
ABR latency and decreased amplitude before loss

of the response (Sohmer et al., 1982, 1989; Pierelli
et al., 1986). These changes occurred despite
unchanged SEP suggesting cochlear dysfunction
(Sohmer et al., 1982, 1986). The VEP shows a
biphasic response to a PaO2 of 20 mm Hg with a
transient amplitude increase that precedes a decrease
(Kayama, 1974).
Since changes in hematocrit can alter both oxygen carrying capacity and blood viscosity, the maximum oxygen delivery is often thought to occur in a
mid-range hematocrit (3032%). Evoked response
changes with hematocrit are consistent with this optimum range (Nagao et al., 1978; Merton et al., 1982;
Dong et al., 1986). The combination of hemodilution
and hypotension has been associated with changes
that are not seen with each effect alone (Starr and
Achor, 1979).
Alterations in latency and amplitude of cortically
generated EPs have been observed as ventilation is
altered beyond the extremes of arterial or end-tidal carbon dioxide concentrations routinely employed during
anesthesia and surgery (Symon and Murota, 1989).
The most significant changes occur with the carbon
dioxide is extremely low (PaCO2 < 20 mm Hg)
and may indicate cerebral ischemia. Hypercapnea
(>100 mm Hg) is associated with an increased latency
(1530%) and decreased amplitude (6080%) of the
feline cortical SEP (Browning et al., 1992). Levels of
50 mm Hg have not been associated with changes in
human SEP (Kalkman et al., 1991a).
Hypothermia, either inadvertent (from a cold
operating room) or intentional (such as used to provide
neural protection such as with thoracic aortic aneurysm repair) is common in operating rooms where
monitoring is occurring. Hypothermia can alter
evoked responses by changing nerve depolarization
(increased action potential duration (Klee et al.,
1974), reduced conduction velocity (Kraft, 1972; Desmedt, 1989), and decreased synaptic function (Weight
and Erulkar, 1976), resulting in increases in latency
and decreases in amplitude of evoked responses
(Dolman et al., 1986). Neurotransmitter release is enhanced at synapses leading to higher end plate potentials (Lundberg, 1948). Nerve conduction velocity is
decreased increasing latency while the amplitude and
duration of the nerve action potential is increased leading to variable amplitude changes (Takaki et al., 1992;
MacKenzie et al., 1995). At temperatures less than
32 C (moderate hypothermia), synaptic transmission
is reduced due to impaired neurotransmitter release
NTTI
(Fay, 1959; Hubbard et al., 1971; Benita and Conde,

1972). Nerve conduction velocity is also impaired (to
an extent less than synaptic transmission; Hubbard
et al., 1971; Andersen et al., 1972; Sohmer et al.,
1989) due to decreases in resting membrane potential
and increases in sodiumpotassium channel activation
time (Klee et al., 1974). In general, peripheral nerve
conduction decreases by about 5% and central conduction by 812%/ C (Aren et al., 1985; Russ et al., 1987;
Zeitlhofer et al., 1990; Reynolds et al., 1991). The net
effect on response latency is the combined effect of
the conduction change and synaptic delays. Hence,
late cortical waves are markedly diminished due to
the cumulative effect on multiple synapses and later
waves are lost before early peaks (Florence et al.,
2004). The primary cortical sensory responses can be
consistently recorded to temperatures as low as 19 C
in cardiopulmonary bypass (Aren et al., 1985) with
the median nerve N20 being lost at about 1526 C
and the P14 being lost at 1220 C (for comparison,
the EEG becomes isoelectric at 2225 C) (Kochs,
1995; Guerit, 1999). Hypothermia appears to affect
synaptic function more than conduction (Budnick
et al., 1981), probably primarily by interference in
the postsynaptic membrane (Weight and Erulkar,
1976). Thus, changes are more prominent at the cephalic end of long neural tracts (such as the SEP) or in components of responses associated with multiple synaptic
elements. Hence, responses recorded from peripheral
nerves are minimally affected, while those produced
by cortical structures are markedly affected (Dolman
et al., 1986; Hume and Durkin, 1986; KottenbergAssenmacher et al., 2003). Later waves of the ABR
are similarly affected more than early waves (Stockard
et al., 1978; Hett et al., 1995).
Whole body hypothermia, either inadvertent or
intentional, is the most obvious temperature change
that occurs during surgery. In addition, changes in
regional temperature can occur and result in evoked
response alterations making the site of temperature
monitoring key to understanding the changes in
responses (i.e., the location may not identify the temperature change or may overrepresent the extent of
the temperature change). Hence, hypothermia of a
limb may delay peripheral nerve responses of the
SEP, but not be associated with central conduction
time changes if the core temperature is maintained
(Aren et al., 1985; Reynolds et al., 1991). For example,
cold irrigation solutions applied to the spinal cord
(Coles et al., 1983), brainstem, or cortex routinely
cause evoked response changes.
115
Changes in a variety of other physiological variables may produce alterations in the evoked
responses during surgical monitoring. Significant
reduction in blood volume can alter evoked responses
due to changes in blood flow distribution, despite
absence of significant blood pressure changes (e.g.,
extremity ischemia altering the SEP as blood flow
to central organs is spared). An increase in superior
vena caval pressure during cardiopulmonary bypass
has been associated with SEP changes (Hill et al.,
1987).
Other physiologic events may occur too slowly to
be noted as changes in the evoked response. For
example, changes in glucose (Deutsch et al., 1983),
sodium, potassium, and other electrolytes important
in the neurochemical environment and affecting neural depolarization and conduction are likely to result
in evoked response changes. For example, with
injury (such as blunt trauma to the spinal cord),
extracellular potassium increases (from 4 up to
80 mM/l) leading to axonal failure (above 10 mm/l)
(Young and Sakatani, 1990). Hence, the SEP could
be lost from potassium released from adjacent structures and does not require axonal disruption of the
pathway. Gradual clearing of the potassium will allow

restoration of the SEP with the time to recovery dependent on the initial rise in the potassium and the local
blood flow. Another effect is that of bilirubin neurotoxicity which can markedly alter the ABR but cannot
affect the SEP (Shapiro, 2002). This has been observed
by other authors (MacDonald et al., 2003; Skinner
et al., 2003) and similar changes have been seen with
the SEP (Maurette et al., 1988; Lubicky et al., 1989;
Kalkman et al., 1991c; Rappaport et al., 1994). Finally,
changes in cortical evoked responses have been
observed with pneumocephalus (Paisansathan et al.,
2003).
5.7. Conclusion
In general, the effect of anesthetic agents on the
evoked responses parallels the effects on the EEG
(Table 1). In most patients, an anesthetic suitable
for monitoring sensory and motor potentials can be
found if the anesthesiologist is familiar with the monitoring methods, the underlying physiology, and the
different effects of anesthetic agents. When appropriate responses are not recorded, technical changes in
Table 1
Summary of neurophysiological effects of hypnotics
Specific GABA agonist

GABA and others
Alpha 2 agonist
NMDA antagonist
Slow wave sleep
Propofol
Etomidate
Halothane
Isoflurane
Enflurane
Sevoflurane
Desflurane
Barbiturates
Clonidine
Dexmedetomidine
Nitrous oxide
Ketamine
Xenon
EEG
SEP
AEP
MEP
Spindles, vertex-wave, B-S

Spindles, vertex-wave, B-S
B-S variable
B-S
B-S, seizures
B-S, seizures
B-S
B-S, epileptiform patterns
Slow
Slow
Frontal beta
Theta
Central slow
Spindles, vertex-wave
#
"
##
##
##
##
##
##
#
#
##
"
#
#
##
##
##
##
##
##
##
##
?
?
#
#
#
#
##
##
##
##
##
##
#
#
##
#
#
#
Summary of neurophysiological effects of hypnotics in monoanesthesia at surgical level, that is, 1 MAC (minimal alveolar concentration)
or higher for volatile anesthetics. Slow wave sleep is included for comparison, as the EEG patterns and effect on somatosensory and motor
responses of specific g-amino butyric acid (GABAA) agonists and alpha-2 agonists are probably caused partly by the same mechanisms.
On the other hand, arousal changes all these neurophysiological measures towards awake patterns, although we only wake up from physiological sleep. Somatosensory evoked potential (SEP) refers mainly to the short latency cortically generated waves and mid-latency
auditory evoked potential (AEP) mainly to cortical mid-latency auditory evoked potentials. B-S, burst suppression.
116
recording and stimulation, such as different stimulus

parameters (rate or ISI with SEPs) and voltage numbers, and intervals between pulses in tcEMEP stimulation. When these do not give satisfactory results,
different anesthetic agents may be tried, such as
changing from inhalational agents to TIVA.
Many monitoring devices allow continuous monitoring of the EEG which gives an additional view
of cerebral function. The univariate measures such
as BIS, the detection of EMG activity, and identification of the fortunately rare occasions of epileptiform activity or ischemia can be helpful
companions to anesthesia and monitoring. By
understanding the physiological and pharmacological basis of anesthesia and monitoring, the operating
room team can considerably improve the safety of
the patient.
References
Agarwal, R, Roitman, KJ, et al. (1998) Improvement of
intraoperative somatosensory evoked potentials by ketamine. Paediatr. Anaesth., 8: 263266.
Andersen, P, Gjerstad, L, et al. (1972) Effect of cooling on
synaptic transmission through the cuneate nucleus. Acta
Physiol. Scand., 84: 433477.
Andersson, G and Ohlin, A (1999) Spatial facilitation of
motor evoked responses in monitoring during spinal
surgery. Clin. Neurophysiol., 110(4): 720724.
Angel, A and Gratton, DA (1982) The effect of anaesthetic
agents on cerebral cortical responses in the rat. Br. J.
Pharmacol., 76: 541549.
Angel, A and LeBeau, F (1992) A comparison of the
effects of propofol with other anaesthetic agents on
the centripetal transmission of sensory information.
Gen. Pharmacol. 23(6): 945963.
Anonymous (1980) Society proceedings: 32nd annual
meeting of the southern electroencephalographic society. Electroencephalogr. Clin. Neurophysiol., 50: 177.
Aren, C, Badr, G, et al. (1985) Somatosensory evoked
potentials and cerebral metabolism during cardiopulmonary bypass with special reference to hypotension
induced by prostacyclin infusion. J. Thorac. Cardiovasc. Surg., 90: 7379.
Astrup, J, Symon, L, et al. (1977) Cortical evoked potential
and extracellular K and H at critical levels of brain
ischemia. Stroke, 8(1): 5157.
Baars, JH, Dangel, C, et al. (2006a) Suppression of the
human spinal H-reflex by propofol: a quantitative analysis. Acta Anaesthesiol. Scand., 50: 193200.
Baars, JH, Tas, S, et al. (2006b) The suppression of spinal
F-waves by propofol does not predict immobility to
painful stimuli in humans. Br. J. Anaesth., 96: 118126.
NTTI
Bastuji, H and Garcia-Larrea, L (1999) Evoked potentials
as a tool for the investigation of human sleep. Sleep
Med. Rev., 3: 2345.
Benedetti, C, Chapman, CR, et al. (1982) Effect of nitrous
oxide concentration on event-related potentials during
painful tooth stimulation. Anesthesiology, 56(5):
360364.
Benita, M and Conde, H (1972) Effects of local cooling
upon conduction and synaptic transmission. Brain
Res., 36: 133151.
Benzon, HT, Toleikis, JR, et al. (1986) Somatosensory
evoked potential quantification of ulnar nerve blockade.
Anesth. Analg., 65(8): 843848.
Berg-Johnsen, J and Langmoen, IA (1986) The effect of
isoflurane on unmyelinated and myelinated fibers in
the rat brain. Acta Physiol. Scand, 127: 8793.
Bernard, JM, Pereon, Y, et al. (1996) Effects of isoflurane
and desflurane on neurogenic motor- and
somatosensory-evoked potential monitoring for scoliosis surgery. Anesthesiology, 85: 10131019.
Bertens, AP (1988) Effects of an analgesic, fentanyl and of
a sedative, droperidol, on the somatosensory evoked
potentials in dogs. Electromyogr. Clin. Neurophysiol.,
28(7/8): 433438.
Bloom, M, Beric, A, et al. (2001) Dexmedetomidine infusion
and
somatosensory
evoked
potentials.
J. Neurosurg. Anesthesiol., 13: 320322.
Bobbin, RP, May, JG, et al. (1979) Effects of pentobarbital
and ketamine on brainstem auditory potentials. Arch.
Otolaryngol., 105: 467470.
Boisseau, N, Madany, M, et al. (2002) Comparison of the
effects of sevoflurane and propofol on cortical somatosensory evoked potentials. Br. J. Anaesth., 88: 785789.
Bosnjak, ZJ, Seagard, J, et al. (1982) The effects of halothane on sympathetic ganglionic transmission. Anesthesiology, 57: 473479.
Branston, NM, Symon, L, et al. (1974) Relationship
between the cortical evoked potential and local cortical
blood flow following acute middle cerebral artery
occlusion in the baboon. Exp. Neurol., 45: 195208.
Branston, NM, Symon, L, et al. (1976) Recovery of the
cortical evoked response following temporary middle
cerebral artery occlusion in baboons: relation to local
blood flow and PO2. Stroke, 7: 151157.
Branston, NM, Ladds, A, et al. (1984) Comparison of the
effects of ischaemia on early components of the somatosensory evoked potential in brainstem, thalamus, and
cerebral cortex. J. Cereb. Blood Flow Metabol., 4: 6881.
Brierley, JN and Symon, L (1979) The extent of infarcts in
baboon brains 3 years after division of the middle cerebral artery. J. Neuropathol. Appl. Neurobiol., 3:
271277.
Brodkey, JS, Richards, DE, et al. (1972) Reversible spinal
cord trauma in cats: additive effects of direct pressure
and ischemia. J. Neurosurg., 37: 591593.

Browning, JL, Heizer, ML, et al. (1992) Variations in corticomotor and somatosensory evoked potentials: effects
of temperature, halothane anesthesia, and arterial partial
pressure of CO2. Anesth. Analg., 74: 643648.
Budnick, B, McKeown, KL, et al. (1981) Hypothermiainduced changes in rat short latency somatosensory evoked
potentials. Electroencephalogr. Clin. Neurophysiol., 51:
1931.
Calancie, B, Harris, W, et al. (1998) Threshold-level
multipulse transcranial electrical stimulation of motor
cortex for intraoperative monitoring of spinal motor
tracts: description of method and comparison to somatosensory evoked potential monitoring. J. Neurosurg., 88:
457470.
Chabal, C, Jacobson, L, et al. (1988) Effects of intrathecal
fentanyl and lidocaine on somatosensory evoked potentials, the H reflex, and clinical responses. Anesth.
Analg., 67: 509513.
Chapman, CR, Colpitts, YM, et al. (1982) Event-related
potential correlates of analgesia; comparison of fentanyl, acupuncture, and nitrous oxide. Pain, 14(4):
327337.
Chassard, D, Joubaub, A, et al. (1989) Auditory evoked
potentials during propofol anaesthesia in man. Br. J.
Anaesth., 62: 522526.
Chen, Z (2004) The effects of isoflurane and propofol on
intraoperative neurophysiological monitoring during
spinal surgery. J. Clin. Monit. Comput., 18: 303308.
Chi, OZ and Field, C (1986) Effects of isoflurane on visual
evoked potentials in humans. Anesthesiology, 65(3):
328330.
Chi, OZ, McCoy, CL, et al. (1987) Effects of fentanyl
anesthesia on visual evoked potentials in humans. Anesthesiology, 67: 827830.
Cicek, S, Attar, A, et al. (2000) Effects of different doses
of epidural midazolam on spinal somatosensory evoked
potentials. Acta Neurochir. (Wien), 142: 921927.
Clapcich, AJ, Emerson, RG, et al. (2004) The effects of propofol, small-dose isoflurane, and nitrous oxide on cortical
somatosensory evoked potential and bispectral index
monitoring in adolescents undergoing spinal fusion.
Anesth. Analg., 99(5): 13341340; table of contents.
Cohen, MS and Britt, RH (1982) Effects of sodium pentobarbital, ketamine, halothane, and chloralose on brainstem auditory evoked responses. Anesth. Analg., 61(4):
338343.
Coles, JG, Wilson, G, et al. (1983) Intraoperative management of thoracic aortic aneurysm. J. Thorac. Cardiovasc. Surg., 85: 292299.
Colin, F, Bourgain, R, et al. (1978) Progressive alteration
of somatosensory evoked potential waveforms with
lowering of cerebral tissue pO2 in the rabbit. Arch.
Int. Physiol. Biochem., 86: 677679.
Crawford, ME, Molkejensen, F, et al. (1993) Direct spinal
effect of intrathecal and extradural midazolam on
117
visceral noxius stimulation in rabbits. Br. J. Anaesth.,
70: 642646.
Da Costa, VV, Saraiva, RA, et al. (2001) The effect of
nitrous oxide on the inhibition of somatosensory evoked
potentials by sevoflurane in children. Anaesth. Intensive
Care, 56: 202207.
Dahl, JB, Rosenberg, J, et al. (1990) Effect of thoracic epidural bupivacaine 0.75% on somatosensory evoked
potentials after dermatomal stimulation. Reg. Anesth.,
15: 7375.
De Haan, P, Kalkman, CJ, et al. (1997) Efficacy of transcranial motor-evoked myogenic potentials to detect
spinal cord ischemia during operations for thoracoabdominal aneurysms. J. Thorac. Cardiovasc. Surg., 113:
87101.
Deletis, V (1993) Intraoperative monitoring of the functional integrity of the motor pathways. Adv. Neurol.,
63: 201214.
Derbyshire, AJ, Rempel, B, et al. (1936) The effect of
anesthetics on action potentials in the cerebral cortex
of the cat. Am. J. Physiol., 116: 577596.
Desmedt, JE (1989) Somatosensory evoked potentials in
neuromonitoring. In: JE Desmedt (Ed.), Neuromonitoring
for Surgery. Elsevier, Amsterdam, pp. 122.
Detsch, O, Vahle-Hinz, C, et al. (1999) Isoflurane induces
dose-dependent changes of thalamic somatosensory
information transfer. Brain Res., 829: 7789.
Deutsch, E, Sohmer, H, et al. (1983) Auditory nerve
brainstem evoked potentials and EEG during severe
hypoglycemia. Electroencephalogr. Clin. Neurophysiol., 55: 714716.
Dolan, EJ, Transfeld, EE, et al. (1980) The effect of spinal
distraction on regional blood flow in cats. J. Neurosurg., 53: 756764.
Dolman, J, Silvay, G, et al. (1986) The effect of temperature, mean arterial pressure, and cardiopulmonary bypass
flows on somatosensory evoked potential latency in man.
Thorac. Cardiovasc. Surg., 34: 217222.
Domino, EF and Corssen, G (1964) Visually evoked
response in anesthetized man with and without induced
muscle paralysis. Ann. N.Y. Acad. Sci., 112: 226229.
Dong, XP and Xu, T (2002) The actions of propofol on
gamma-aminobutyric acid-A and glycine receptors in
acutely dissociated spinal dorsal horn neurons of the
rat. Anesth. Analg., 95: 907914.
Dong, WK, Bledsoe, SW, et al. (1986) Electrical correlates
of brain injury resulting from severe hypotension and
hemodilution in monkeys. Anesthesiology, 65: 617625.
Drummond, JC, Todd, MM, et al. (1985) The effect of
high dose sodium thiopental on brainstem auditory and
median somatosensory evoked responses in humans.
Anesthesiology, 63: 249254.
Dubois, MY, Sato, S, et al. (1982) Effects of enflurane on
brainstem auditory evoked responses in humans.
Anesth. Analg., 61: 898902.
118
Erb, TO, Ryhult, SE, et al. (2005) Improvement of motorevoked potentials by ketamine and spatial facilitation
during spinal surgery in a young child. Anesth. Analg.,
100: 16341636.
Faull, RL and Villiger, JW (1986) Benzodiazepine receptors
in the human spinal cord: a detailed anatomical and pharmacological study. Neuroscience, 17(3): 791802.
Fava, E, Bortolani, EM, et al. (1988) Evaluation of spinal
cord function by means of lower limb somatosensory
evoked potentials in reparative aortic surgery. J. Cardiovasc. Surg., 29(4): 421427.
Fay, T (1959) Early experiences with local and generalized
refrigeration of the human brain. J. Neurosurg., 16:
239260.
Fenwick, P, Bushman, J, et al. (1979) Contingent negative
variation and evoked potential amplitude as a function
of inspired nitrous oxide concentration. Electroencephalogr. Clin. Neurophysiol., 47(4): 473482.
Fernandez-Galinski, SM, Monells, J, et al. (1996) Effects
of subarachnoid lidocaine, meperidine and fentanyl on
somatosensory and motor evoked responses in awake
humans. Acta Anaesthesiol. Scand., 40(1): 3946.
Firsching, R, Heinen-Lauten, M, et al. (1991) The effects of
halothane and nitrous oxide on transcranial magnetic
evoked potentials. Anasthesiol. Intensivmed. Notfallmed.
Schmerzther., 26(7): 381383.
Fletcher, JE, Hinn, AR, et al. (2005) The effects of isoflurane and desflurane titrated to a bispectral index of 60 on
the cortical somatosensory evoked potential during
pediatric scoliosis surgery. Anesth. Analg., 100(6):
17971803.
Flood, P and Coates, KM (2002) Droperidol inhibits
GABAA and neuronal nicotinic receptor activation.
Florence, G, Guerit, J-M, et al. (2004) Electroencephalography (EEG) and somatosensory evoked potentials
(SEP) to prevent cerebral ischaemia in the operating
room. Neurophysiol. Clin., 34(1): 1732.
Follis, F, Scremin, OU, et al. (1993) Selective vulnerability of white matter during spinal cord ischemia. J.
Cereb. Blood Flow Metab., 13(1): 170178.
Freye, E, Hartung, E, et al. (1989) Somatosensory-evoked
potentials during block of surgical stimulation with propofol. Br. J. Anaesth., 63(3): 357359.
Ghaly, RF, Stone, JL, et al. (1990a) Effects of incremental
ketamine hydrochloride dose on motor evoked potentials (MEPs) following transcranial magnetic stimulation: a primate study. J. Neurosurg. Anesthesiol., 2:
7985.
Ghaly, RF, Stone, JL, et al. (1990b) The effects of etomidate or midazolam hypnotic dose on motor evoked
potentials in the monkey. J. Neurosurg. Anesthesiol.,
2: 244247.
Ghaly, RF, Stone, JL, et al. (1991a) The effect of an anesthetic induction dose of midazolam on motor potentials
NTTI
evoked by transcranial magnetic stimulation in the monkey. J. Neurosurg. Anesthesiol., 3: 2025.
Ghaly, RF, Stone, JL, et al. (1991b) The effect of neuroleptanalgesia (droperidol-fentanyl) on motor potentials
evoked by transcranial magnetic stimulation in the monkey. J. Neurosurg. Anesthesiol., 3: 117119.
Glassman, SD, Shields, CB, et al. (1993) Anesthetic
effects on motor evoked potentials in dogs. Spine,
18(8): 10831089.
Glassman, SD, Zhang, YP, et al. (1995) Transcranial magnetic motor-evoked potentials in scoliosis surgery.
Orthopedics, 18(10): 10171023.
Griffiths, R and Norman, RI (1993) Effects of anaesthetics
on uptake, synthesis and release of transmitters. Br. J.
Anaesth., 71(1): 96107.
Griffiths, IR, Trench, JG, et al. (1979) Spinal cord blood
flow and conduction during experimental cord compression in normotensive and hypotensive dogs. J. Neurosurg., 50(3): 353360.
Grubbs, PE, Jr., Marini, C, et al. (1988) Somatosensory
evoked potentials and spinal cord perfusion pressure
are significant predictors of postoperative neurologic
dysfunction. Surgery, 104(2): 216223.
Grundy, BL, Jannetta, PJ, et al. (1982a) Intraoperative
monitoring of brainstem auditory evoked potentials. J.
Neurosurg., 57(5): 674681.
Grundy, BL, Procopio, PT, et al. (1982b) Evoked potential
changes produced by positioning for retromastoid craniectomy. Neurosurgery, 10(6 Pt 1): 766770.
Guerit, JM (1999) Medical technology assessment EEG
and evoked potentials in the intensive care unit. Neurophysiol. Clin., 29(4): 301317.
Guertin, PA and Hounsgaard, J (1999) Non-volatile general
anaesthetics reduce spinal activity by suppressing plateau potentials. Neuroscience, 88(2): 353358.
Gugino, LD, Kraus, KH, et al. (1992) Peripheral ischemia
as a complicating factor during somatosensory and
motor evoked potential monitoring of aortic surgery. J.
Cardiothorac. Vasc. Anesth., 6(6): 715719.
Gugino, LD, Aglio, LS, et al. (1997) Use of transcranial
magnetic stimulation for monitoring spinal cord motor
paths. Semin. Spine Surg., 9: 315336.
Haghighi, S, Madsen, R, et al. (1990a) Suppression of
motor evoked potentials by inhalation anesthetics. J.
Neurosurg. Anesthesiol., 2: 7376.
Haghighi, SS, Green, KD, et al. (1990b) Depressive effect
of isoflurane anesthesia on motor evoked potentials.
Neurosurgery, 26(6): 993997.
Hamaguchi, K, Nakagawa, I, et al. (2005) Effect of propofol on visual evoked potentials during neurosurgery.
Masui Japanese J. Anesthesiol., 54(9): 9981002.
Hargreaves, SJ and Watt, JWH (2005) Intravenous anaesthesia and repetitive transcranial magnetic stimulation
monitoring in spinal column surgery. Br. J. Anaesth.,
94(1): 7073.

Harker, LA, Hosick, E, et al. (1977) Influence of succinylcholine on middle component auditory evoked potentials. Arch. Otolaryngol., 103(3): 133137.
Harkins, SW, Benedetti, C, et al. (1982) Effects of nitrous
oxide inhalation on brain potentials evoked by auditory
and noxious dental stimulation. Prog. Neuropsychopharmacol. Biol. Psychiatry, 6(2): 167174.
Hartikainen, KM, Rorarius, M, et al. (1995) Cortical reactivity during isoflurane burst-suppression anesthesia.
Anesth Analg., 81: 12231228.
Heneghan, CP, Thornton, C, et al. (1987) Effect of isoflurane on the auditory evoked response in man. Br. J.
Anaesth., 59(3): 277282.
Herdmann, J, Lumenta, CB, et al. (1993) Magnetic stimulation for monitoring of motor pathways in spinal procedures. Spine, 18(5): 551559.
Hett, DA, Smith, DC, et al. (1995) Effect of temperature
and cardiopulmonary bypass on the auditory evoked
response. Br. J. Anaesth., 75(3): 293296.
Hetzler, BE and Berger, LK (1984) Ketamine-induced
modification of photic evoked potentials in the superior
colliculus of hooded rats. Neuropharmacology, 23(4):
473476.
Hill, R, Sebel, PS, et al. (1987) Alterations in somatosensory evoked potentials associated with inadequate
venous return during cardiopulmonary bypass. J. Cardiothorac. Anesth., 1(1): 4850.
Himwich, HE (1951) Brain Metabolism and Cerebral Disorders. Williams and Wilkins, Baltimore.
Hosick, EC, Clark, DL, et al. (1971) Neurophysiological
effects of different anesthetics in conscious man. J.
Appl. Physiol., 31(6): 892898.
Houston, HG, McClelland, RJ, et al. (1988) Effects of
nitrous oxide on auditory cortical evoked potentials
and subjective thresholds. Br. J. Anaesth., 61(5):
606610.
Hubbard, JI, Jones, SF, et al. (1971) The effect of temperature change upon transmitter release, facilitation and
post-tetanic potentiation. J. Physiol., 216(3): 591609.
Hume, AL and Durkin, MA (1986) Central and spinal
somatosensory conduction times during hypothermic
cardiopulmonary bypass and some observations on the
effects of fentanyl and isoflurane anesthesia. Electroencephalogr. Clin. Neurophysiol., 65(1): 4658.
Huotari, AM, Koskinen, M, et al. (2004) Evoked EEG patterns during burst suppression with propofol. Br. J.
Anaesth., 92(1): 1824.
Iida, H, Dohi, S, et al. (1997) Spinal conduction block by
intrathecal ketamine in dogs. Anesth. Analg., 85(1):
106110.
Imaizumi, T, Kocsis, JD, et al. (1998) Resistance to anoxic
injury in the dorsal columns of adult rat spinal cord following demyelination. Brain Res., 779(1/2): 292296.
Inoue, S, Kawaguchi, M, et al. (2002) Amplitudes and
intrapatient variability of myogenic motor evoked
119
potentials to transcranial electrical stimulation during
ketamine/N2O- and propofol/N2O-based anesthesia. J.
Neurosurg. Anesthesiol., 14(3): 213217.
Iwayama, K, Mori, K, et al. (1986) Changes of somatosensory evoked potential accompanying ischaemia and hypoxia in cats. Neurol. Res., 8(3): 157163.
Jaffe, RA and Rowe, MA (1996) A comparison of the local
anesthetic effects of meperidine, fentanyl, and sufentanil
on dorsal root axons. Anesth. Analg., 83(4): 776781.
James, MFM, Thornton, C, et al. (1982) Halothane anaesthesia changes the early components of the auditory
evoked response in man. Br. J. Anaesth., 54: 787.
Jantti, V, Sonkajarvi, E, et al. (1998) Single-sweep cortical
somatosensory evoked potentials: N2O and evoked
bursts in sevoflurane anaesthesia. Electroencephalogr.
Clin. Neurophysiol., 108(3): 320324.
Javel, E, Mouney, DF, et al. (1982) Auditory brainstem
responses during systemic infusion of lidocaine. Arch.
Otolaryngol., 108(2): 7176.
Jellinek, D, Jewkes, D, et al. (1991a) Noninvasive intraoperative monitoring of motor evoked potentials under
propofol anesthesia: effects of spinal surgery on the
amplitude and latency of motor evoked potentials. Neurosurgery, 29(4): 551557.
Jellinek, D, Platt, M, et al. (1991b) Effects of nitrous oxide
on motor evoked potentials recorded from skeletal muscle in patients under total anesthesia with intravenously
administered propofol. Neurosurgery, 29(4): 558562.
John, ER and Prichep, LS (2005) The anesthetic cascade: a
theory of how anesthesia suppresses consciousness.
Jones, SJ, Harrison, R, et al. (1996) Motor evoked potential monitoring during spinal surgery: responses of distal
limb muscles to transcranial cortical stimulation with
pulse trains. Electroencephalogr. Clin. Neurophysiol.,
100(5): 375383.
Jou, IM, Chern, T-C, et al. (2003a) Effects of desflurane
on spinal somatosensory-evoked potentials and conductive spinal cord evoked potential. Spine, 28(16):
18451850.
Jou, IM, Chu, KS, et al. (2003b) The effects of intrathecal
tramadol on spinal somatosensory-evoked potentials
and motor-evoked responses in rats. Anesth. Analg.,
96: 783788.
Kai, Y, Owen, JH, et al. (1993) Use of sciatic neurogenic
motor evoked potentials versus spinal potentials to predict early-onset neurologic deficits when intervention is
still possible during overdistraction. Spine, 18(9):
11341139.
Kaieda, R, Maekawa, T, et al. (1981) Effects of diazepam
on evoked electrospinogram and evoked electromyogram in man. Anesth. Analg., 60(4): 197200.
Kakinohana, M, Nakamura, S, et al. (2005) Emergence
from propofol anesthesia in a nonagenarian at a Bispectral Index of 52. Anesth. Analg., 101(1): 169170.
120
Kalkman, CJ, Boezeman, EH, et al. (1991a) Influence of
changes in arterial carbon dioxide tension on the electroencephalogram and posterior tibial nerve somatosensory cortical evoked potentials during alfentanil/nitrous
oxide anesthesia. Anesthesiology, 75(1): 6874.
Kalkman, CJ, Drummond, JC, et al. (1991b) Low concentrations of isoflurane abolish motor evoked responses
to transcranial electrical stimulation during nitrous
oxide/opioid anesthesia in humans. Anesth. Analg.,
73(4): 410415.
Kalkman, CJ, Ten Brink, SA, et al. (1991c) Variability of
somatosensory cortical evoked potentials during spinal
surgery. Effects of anesthetic technique and high-pass
digital filtering. Spine, 16(8): 924929.
Kalkman, CJ, Drummond, JC, et al. (1992a) Intraoperative
monitoring of tibialis anterior muscle motor evoked
partial neuromuscular blockade. Anesth. Analg., 75(4):
584589.
Kalkman, CJ, Drummond, JC, et al. (1992b) Effects of
propofol, etomidate, midazolam, and fentanyl on motor
evoked responses to transcranial electrical or magnetic
stimulation in humans. Anesthesiology, 76(4): 502509.
Kalkman, CJ, Been, HD, et al. (1993) Intraoperative monitoring of spinal cord function. A review. Acta Orthop.
Scand., 64(1): 114123.
Kalkman, CJ, Drummond, JC, et al. (1994) Effects of droperidol, pentobarbital, and ketamine on myogenic transcranial magnetic motor-evoked responses in humans.
Kammer, T, Rehberg, B, et al. (2002) Propofol and sevoflurane in subanesthetic concentrations act preferentially on the spinal cord: evidence from multimodal
electrophysiological assessment. Anesthesiology, 97(6):
14161425.
Kano, T and Shimoji, K (1974) The effects of ketamine and
neuroleptanalgesia on the evoked electrospinogram and
electromyogram in man. Anesthesiology, 40(3):
241246.
Kawaguchi, M and Furuya, H (2004) Intraoperative spinal
cord monitoring of motor function with myogenic motor
evoked potentials: a consideration in anesthesia. J.
Anesth., 18(1): 1828.
Kawaguchi, M, Sakamoto, T, et al. (1996) Intraoperative
myogenic motor evoked potentials induced by direct
electrical stimulation of the exposed motor cortex under
isoflurane and sevoflurane. Anesth. Analg., 82(3):
593599.
Kawaguchi, M, Sakamoto, T, et al. (2000) Low dose propofol as a supplement to ketamine-based anesthesia during intraoperative monitoring of motor-evoked
potentials. Spine, 25(8): 974979.
Kayama, Y (1974) Evoked potentials of the central visual
system during and after hypoxia in cats. Electroencephalogr. Clin. Neurophysiol., 36(6): 619628.
NTTI
Keller, BP, Haghighi, SS, et al. (1992) The effects of propofol anesthesia on transcortical electric evoked potentials in the rat. Neurosurgery, 30(4): 557560.
Kerz, T, Hennes, HJ, et al. (2001) Effects of propofol on
H-reflex in humans. Anesthesiology, 94(1): 3237.
Kimovec, MA, Koht, A, et al. (1990) Effects of sufentanil
on median nerve somatosensory evoked potentials. Br.
J. Anaesth., 65(2): 169172.
Klee, MR, Pierau, FK, et al. (1974) Temperature effects on
resting potential and spike parameters of cat motoneurons. Exp. Brain Res., 19(5): 478492.
Kobrine, AI, Evans, DE, et al. (1980) Relative vulnerability of the brain and spinal cord to ischemia. J. Neurol.
Sci., 45(1): 6572.
Kochs, E (1995) Electrophysiological monitoring and mild
hypothermia. J. Neurosurg. Anesthesiol., 7(3): 222228.
Kochs, E, Treede, RD, et al. (1986) Increase in somatosensory evoked potentials during anesthesia induction with
etomidate. Anaesthesist, 35(6): 359364.
Koht, A, Schutz, W, et al. (1988) Effects of etomidate,
midazolam, and thiopental on median nerve somatosensory evoked potentials and the additive effects of fentanyl and nitrous oxide. Anesth. Analg., 67(5): 435441.
Koscielniak-Nielsen, ZJ, Stens-Pedersen, HL, et al. (1998)
Midazolam-flumazenil versus propofol anaesthesia for
scoliosis surgery with wake-up tests. Acta Anaesthesiol.
Scand., 42(1): 111116.
Kothbauer, K, Schmid, UD, et al. (1993) The effect of
ketamine anesthetic induction on muscle responses to
transcranial magnetic cortex stimulation studied in
man. Neurosci. Lett., 154(1/2): 105108.
Kottenberg-Assenmacher, E, Armbruster, W, et al. (2003)
Hypothermia does not alter somatosensory evoked
potential amplitude and global cerebral oxygen extraction during marked sodium nitroprusside-induced arterial hypotension [see comment]. Anesthesiology, 98(5):
11121118.
Kraft, GH (1972) Effects of temperature and age on nerve
conduction velocity in the guinea pig. Arch. Phys.
Med. Rehabil., 53(7): 328332.
Ku, ASW, Hu, Y, et al. (2002) Effect of sevoflurane/nitrous
oxide versus propofol anaesthesia on somatosensory
evoked potential monitoring of the spinal cord during surgery to correct scoliosis. Br. J. Anaesth., 88(4): 502507.
Kumar, A, Tandon, OP, et al. (1994) Recovery from preoperative sedation with clonidine brainstem auditory
evoked response. Anaesthesia, 49(6): 533537.
Lam, AM, Sharar, SR, et al. (1994) Isoflurane compared
with nitrous oxide anaesthesia for intraoperative monitoring of somatosensory-evoked potentials. Can. J.
Anaesth., 41(4): 295300.
Lang, E, Erdmann, K, et al. (1993) Median nerve blockade
during diagnostic intravenous regional anesthesia as
measured by somatosensory evoked potentials [see
comment]. Anesth. Analg., 76(1): 118122.

Lang, EW, Beutler, AS, et al. (1996a) Myogenic motorevoked potential monitoring using partial neuromuscular blockade in surgery of the spine. Spine, 21(14):
16761686.
Lang, EW, Chesnut, RM, et al. (1996b) The utility of
motor-evoked potential monitoring during intramedullary surgery. Anesth. Analg., 83(6): 13371341.
Langeron, O, Lille, F, et al. (1997) Comparison of the effects
of ketamine-midazolam with those of fentanyl-midazolam
on cortical somatosensory evoked potentials during major
spine surgery. Br. J. Anaesth., 78(6): 701706.
Laschinger, JC, Owen, J, et al. (1988) Direct noninvasive
monitoring of spinal cord motor function during thoracic aortic occlusion: use of motor evoked potentials.
J. Vasc. Surg., 7(1): 161171.
Lee, VC (1994) Spinal and cortical evoked potential studies
in the ketamine-anesthetized rabbit: fentanyl exerts
component-specific, naloxone-reversible changes dependent on stimulus intensity. Anesth. Analg., 78(2):
280286.
Lee, WY, Hou, WY, et al. (1995) Intraoperative monitoring of motor function by magnetic motor evoked potentials. Neurosurgery, 36(3): 493500.
Levy, WJ, McCaffrey, M, et al. (1984) Motor evoked potentials from transcranial stimulation of the motor cortex in
cats. Neurosurgery, 15(2): 214227.
Liang, W-M, Zhang, J, et al. (2004) Effects of intravenous
and inhalational anesthetics on short-latency somatosensory evoked potentials. Chung-Hua i Hsueh Tsa Chih
[Chin. Med. J.], 84(6): 460463.
Liu, EHC, Wong, HK, et al. (2005) Effects of isoflurane
and propofol on cortical somatosensory evoked potentials during comparable depth of anaesthesia as guided
by bispectral index. Br. J. Anaesth., 94(2): 193197.
Lloyd-Thomas, AR, Cole, PV, et al. (1990) Quantitative
EEG and brainstem auditory evoked potentials: comparison of isoflurane with halothane using the cerebral
function analysing monitor. Br. J. Anaesth., 65(3):
306312.
Logginidou, HG, Li, B-H, et al. (2003) Propofol suppresses the cortical somatosensory evoked potential in
rats. Anesth. Analg., 97(6): 17841788.
Loughman, BA, Fennelly, ME, et al. (1995) The effects of
differing concentrations of bupivacaine on the epidural
somatosensory evoked potential after posterior tibial
nerve stimulation. Anesth. Analg., 81(1): 147151.
Loughnan, BA and Fennelly, ME (1995) Spinal cord monitoring. Anaesthesia, 50(2): 101102.
Loughnan, BA, Anderson, SK, et al. (1989) Effects of halothane on motor evoked potential recorded in the extradural space. Br. J. Anaesth., 63(5): 561564.
Loughnan, BA, Murdoch, LJ, et al. (1990) Effects of 2%
lignocaine on somatosensory evoked potentials recorded in the extradural space. Br. J. Anaesth., 65(5):
643647.
121
Lubicky, JP, Spadaro, JA, et al. (1989) Variability of
somatosensory cortical evoked potential monitoring
during spinal surgery. Spine, 14(8): 790798.
Lumenta, CB (1991) Effect of etomidate on motor evoked
potentials in monkeys [see comment]. Neurosurgery,
29(3): 480482.
Lund, C, Selmar, P, et al. (1987) Effect of epidural bupivacaine on somatosensory evoked potentials after dermatomal stimulation. Anesth. Analg., 66(1): 3438.
Lund, C, Hansen, OB, et al. (1989a) Effect of epidural
0.25% bupivacaine on somatosensory evoked potentials to dermatomal stimulation. Reg. Anesth., 14(2):
7277.
Lund, C, Hansen, OB, et al. (1989b) Effect of epidural clonidine on somatosensory evoked potentials to dermatomal stimulation. Eur. J. Anaesthesiol. Suppl., 6(3):
207213.
Lundberg, A (1948) Potassium and the differential thermosensitivity of membrane potential, spike and negative
after-potential in mammalian A and C fibers. Acta
Physiol. Scand. Suppl., 50: 167.
MacDonald, DB, Al Zayed, Z, et al. (2003) Monitoring
scoliosis surgery with combined multiple pulse transcranial electric motor and cortical somatosensory-evoked
potentials from the lower and upper extremities. Spine,
28(2): 194203.
MacDonald, DB, Al Zayed, Z, et al. (2005) Tibial somatosensory evoked potential intraoperative monitoring:
recommendations based on signal to noise ratio analysis
of popliteal fossa, optimized P37, standard P37, and P31
potentials. Clin. Neurophysiol., 116(8): 18581869.
Machida, M, Weinstein, SL, et al. (1985) Spinal
cord monitoring. Electrophysiological measures of sensory and motor function during spinal surgery. Spine,
10(5): 407413.
Machida, K, Shinomiya, K, et al. (1995) A new method of
multisegment motor pathway monitoring using muscle
potentials after train spinal stimulation. Spine, 20:
22402246.
MacKenzie, MA, Vingerhoets, DM, et al. (1995) Effect of
steady hypothermia and normothermia on multimodality evoked potentials in human poikilothermia. Arch.
Neurol., 52(1): 5258.
Maeda, S, Miyamoto, T, et al. (1989) Prevention of spinal
cord ischemia by monitoring spinal cord perfusion pressure and somatosensory evoked potentials. J. Cardiovasc.
Surg., 30(4): 565571.
Mahla, ME, Long, DM, et al. (1984) Detection of brachial
plexus dysfunction by somatosensory evoked potential
monitoring a report of two cases. Anesthesiology,
60(3): 248252.
Manninen, PH, Lam, AM, et al. (1985) The effects of isoflurane and isoflurane-nitrous oxide anesthesia on brainstem auditory evoked potentials in humans. Anesth.
Analg., 64(1): 4347.
122
Marsh, RR, Frewen, TC, et al. (1984) Resistance of the
auditory brainstem response to high barbiturate levels.
Otolaryngol. Head Neck Surg., 92(6): 685688.
Mauguie`re, F (2004) Chapter 3.1 Definitions and introductory remarks. In: CD Binnie, R Cooper, F Mauguie`re,
JW Osselton, PF Prior, and BM Tedman (Eds.), Clinical
neurophysiology, Volume 1 (Revised and Enlarged Edition) EMG, Nerve Conduction and Evoked Potentials,
Elsevier BV, Amsterdam, pp. 357366.
Maurette, P, Simeon, F, et al. (1988) Propofol anaesthesia
alters somatosensory evoked cortical potentials. Anaesthesia, 43(Suppl.): 4445.
Mavroudakis, N, Vandesteene, A, et al. (1994) Spinal and
brainstem SEPs and H reflex during enflurane anesthesia.
Electroencephalogr. Clin. Neurophysiol., 92(1): 8285.
May, DM, Jones, SJ, et al. (1996) Somatosensory evoked
potential monitoring in cervical surgery: identification of
pre- and intraoperative risk factors associated with neurological deterioration. J. Neurosurg., 85(4): 566573.
McPherson, RW, Szymanski, J, et al. (1984) Somatosensory evoked potential changes in position-related brainstem ischemia. Anesthesiology, 61(1): 8890.
McPherson, RW, Mahla, M, et al. (1985) Effects of enflurane, isoflurane, and nitrous oxide on somatosensory
evoked potentials during fentanyl anesthesia. Anesthesiology, 62(5): 626633.
McPherson, RW, Sell, B, et al. (1986) Effects of thiopental, fentanyl, and etomidate on upper extremity somatosensory evoked potentials in humans. Anesthesiology,
65(6): 584589.
Merton, PA, Hill, DK, et al. (1982) Scope of a technique for
electrical stimulation of human brain, spinal cord, and
muscle. Lancet, 2(8298): 597600.
Minahan, RE, Riley, LH, 3rd, et al. (2000) The effect of neuromuscular blockade on pedicle screw stimulation thresholds. Spine, 25(19): 25262530.
Morota, N, Deletis, V, et al. (1997) The role of motor
evoked potentials during surgery for intramedullary spinal cord tumors. Neurosurgery, 41(6): 13271336.
Mosko, SS, Pierce, S, et al. (1981) Normal brainstem auditory evoked potentials recorded in sleep
apneics during waking and as a function of arterial oxygen saturation during sleep. Electroencephalogr. Clin.
Nagao, S, Roccaforte, P, et al. (1978) The effects of isovolemic hemodilution and reinfusion of packed erythrocytes on somatosensory and visual evoked potentials.
J. Surg. Res., 25(6): 530537.
Nagle, KJ, Emerson, RG, et al. (1996) Intraoperative monitoring of motor evoked potentials: a review of 116
cases. Neurology, 47(4): 9991004.
Nelson, LE, Guo, TZ, et al. (2002) The sedative component of anesthesia is mediated by GABA(A) receptors
in an endogenous sleep pathway [see comment]. Nature
Neurosci., 5(10): 979984.
NTTI
Nelson, LE, Lu, J, et al. (2003) The alpha2adrenoceptor agonist dexmedetomidine converges on
an endogenous sleep-promoting pathway to exert its
sedative effects. Anesthesiology, 98(2): 428436.
Newlon, PG, Greenberg, RP, et al. (1983) Effects of therapeutic pentobarbital coma on multimodality evoked
potentials recorded from severely head-injured patients.
Newton, DE, Thornton, C, et al. (1989) Early cortical
auditory evoked response in anaesthesia: comparison
of the effects of nitrous oxide and isoflurane. Br. J.
Anaesth., 62(1): 6165.
North, RB, Drenger, B, et al. (1991) Monitoring of spinal
cord stimulation evoked potentials during thoracoabdominal aneurysm surgery. Neurosurgery, 28(2): 325330.
Nuwer, MR (1986) Evoked Potential Monitoring in the
Operating Room. Raven Press, New York.
Nuwer, MR (1988) Use of somatosensory evoked potentials for intraoperative monitoring of cerebral and spinal
cord function. Neurol. Clin., 6(4): 881897.
Ohara, A, Mashimo, T, et al. (1997) A comparative study
of the antinociceptive action of xenon and nitrous oxide
in rats. Anesth. Analg., 85(4): 931936.
Oka, Y and Miyamoto, T (1987) Prevention of spinal cord
injury after cross-clamping of the thoracic aorta. J. Cardiovasc. Surg., 28(4): 398404.
Oro, J and Haghighi, SS (1992) Effects of altering core
body temperature on somatosensory and motor evoked
potentials in rats. Spine, 17(5): 498503.
Owen, JH (1997) Applications of neurophysiological measures during surgery of the spine. In: JW Frymoyer
(Ed.) , The Adult Spine: Principles and Practice. Lippincott-Raven Publishers, Philadelphia, 2nd ed., pp.
673702.
Paisansathan, C, Koenig, HM, et al. (2003) Loss of SSEP
during sitting craniotomy. J. Neurosurg. Anesthesiol.,
15(4): 327329.
Pang, WW, Mok, MS, et al. (1998) Local anesthetic effect
of tramadol, metoclopramide, and lidocaine following
intradermal injection [see comment]. Reg. Anesth. Pain
Med., 23(6): 580583.
Pathak, KS, Amaddio, MD, et al. (1989) Effects of halothane, enflurane, and isoflurane in nitrous oxide on
multilevel somatosensory evoked potentials. Anesthesiology, 70(2): 207212.
Pechstein, U, Cedzich, C, et al. (1996) Transcranial highfrequency repetitive electrical stimulation for recording
myogenic motor evoked potentials with the patient
under general anesthesia. Neurosurgery, 39(2):
335343; discussion 343344.
Pechstein, U, Nadstawek, J, et al. (1998) Isoflurane plus
nitrous oxide versus propofol for recording of motor
evoked potentials after high frequency repetitive electrical stimulation. Electroencephalogr. Clin. Neurophysiol., 108(2): 175181.

Pereon, Y, Bernard, JM, et al. (1999) The effects of desflurane on the nervous system: from spinal cord to muscles. Anesth. Analg., 89(2): 490495.
Peterson, DO, Drummond, JC, et al. (1986) Effects of halothane, enflurane, isoflurane, and nitrous oxide on
somatosensory evoked potentials in humans. Anesthesiology, 65(1): 3540.
Pierelli, F, Rizzo, PA, et al. (1986) Early auditory evoked
potential changes during hypoxic hypoxia in the rabbit.
Exp. Neurol., 94(3): 479488.
Plourde, G (2006) Auditory evoked potentials. Best Pract.
Res. Clin. Anaesthesiol., 20(1): 129139.
Porkkala, T, Jantti, V, et al. (1998) Clonidine does not
attenuate median nerve somatosensory evoked potentials during isoflurane anesthesia. J. Clin. Monit. Comput., 14(3): 165170.
Prior, PF (1985) EEG monitoring and evoked potentials in
brain ischaemia. Br. J. Anaesth., 57(1): 6381.
Purdie, JA and Cullen, PM (1993) Brainstem auditory
evoked response during propofol anaesthesia in children. Anaesthesia, 48(3): 192195.
Raeder, JC (1996) Basis of anaesthesia what do we
know after 150 years? Acta Anaesthesiol. Scand.,
40(9): 10681072.
Rampil, IJ (1997) Electroencephalogram. In: MA Albin
(Ed.), Textbook of Neuroanesthesia with Neurosurgical
and Neuroscience Perspectives. McGraw-Hill, New
York, pp. 193220.
Rappaport, M, Ruiz Portillo, S, et al. (1994) Effects of
stimulus intensity and duration on posterior tibial nerve
somatosensory-evoked potential patterns obtained under
anesthesia. Spine, 19(13): 15251529.
Rehberg, B, Ruschner, R, et al. (1998) Concentrationdependent changes in the latency and amplitude
of somatosensory-evoked potentials by desflurane,
isoflurane and sevoflurane. Anasthesiol. Intensivmed.
Notfallmed. Schmerzther., 33(7): 425429.
Reynolds, PC, Antoine, JA, et al. (1991) Regional hypothermia affects somatosensory evoked potentials.
Anesth. Analg., 73(5): 653656.
Richardson, J, Jones, J, et al. (1998) The effect of thoracic
paravertebral blockade on intercostal somatosensory
evoked potentials. Anesth. Analg., 87(2): 373376.
Rodi, Z, Deletis, V, et al. (1996) Motor evoked potentials
during brain surgery. Pflugers Arch. Eur. J. Physiol.,
431(6 Suppl 2): R291R292.
Rosner, BS and Clark, DL (1973) Neurophysiologic effects of
general anesthetics. II. Sequential regional actions in the
brain. Anesthesiology, 39: 5967.
Rundshagen, I, Kochs, E, et al. (1995) Surgical stimulation
increases median nerve somatosensory evoked
responses during isoflurane-nitrous oxide anaesthesia.
Br. J. Anaesth., 75(5): 598602.
Russ, W and Fraedrich, G (1984) Intraoperative detection of
cerebral ischemia with somatosensory cortical evoked
123
potentials during carotid endarterectomy presentation
of a new method. Thorac. Cardiovasc. Surg., 32(2):
124126.
Russ, W, Luben, V, et al. (1982) Der EinfluB der Neuroleptanalgesie auf das visuelle evozierte Potential
(VEP) des Menschen. Anaesthesist, 31: 575578.
Russ, W, Thiel, A, et al. (1986) Somatosensory evoked
potentials under thiopental and etomidate. Anaesthesist,
35(11): 679685.
Russ, W, Sticher, J, et al. (1987) Effects of hypothermia on
somatosensory evoked responses in man. Br. J.
Anaesth., 59(12): 14841491.
Russell, GB, Schwentker, MC, et al. (1994) Preservation of
neurogenic motor-evoked potentials during isoflurane
electroencephalographic burst suppression in rats. Spine,
19(23): 26322636.
Rytky, S, Huotari, AM, et al. (1999) Tibial nerve somatosensory evoked potentials during EEG suppression in
sevoflurane anaesthesia. Clin. Neurophysiol., 110(9):
16551658.
Sainz, M, Martinez, F, et al. (1987) Brainstem and middle
latency auditory evoked responses in rabbits with halothane anaesthesia. Acta Otolaryngol., 103(5/6):
613619.
Sakamoto, T, Kawaguchi, M, et al. (2001) Suppressive
effect of nitrous oxide on motor evoked potentials can
be reversed by train stimulation in rabbits under ketamine/fentanyl anaesthesia, but not with additional
propofol. Br. J. Anaesth., 86(3): 395402.
Salzman, SK, Beckman, AL, et al. (1986) Effects
of halothane on intraoperative scalp-recorded somatosensory evoked potentials to posterior tibial nerve
stimulation in man. Electroencephalogr. Clin. Neurophysiol., 65(1): 3645.
Samra, SK and Sorkin, LS (1991) Enhancement of somatosensory evoked potentials by etomidate in cats: an
investigation of its site of action. Anesthesiology,
74(3): 499503.
Samra, SK, Lilly, DJ, et al. (1984) Fentanyl anesthesia and
human brainstem auditory evoked potentials. Anesthesiology, 61(3): 261265.
Samra, SK, Vanderzant, CW, et al. (1987) Differential
effects of isoflurane on human median nerve somatosensory evoked potentials. Anesthesiology, 66(1):
2935.
Savoia, G, Esposito, C, et al. (1988) Propofol infusion and
auditory evoked potentials. Anaesthesia, 43(Suppl.):
4649.
Scheepstra, GL, De Lange, JJ, et al. (1989) Median nerve
evoked potentials during propofol anaesthesia. Br. J.
Anaesth., 62(1): 9294.
Scheufler, K-M and Zentner, J (2002) Total intravenous
anesthesia for intraoperative monitoring of the motor
pathways: an integral view combining clinical and
experimental data. J. Neurosurg., 96(3): 571579.
124
Schmid, UD, Boll, J, et al. (1992) Influence of some anesthetic agents on muscle responses to transcranial magnetic cortex stimulation: a pilot study in humans.
Schmidt, JF and Chraemmer-Jorgensen, B (1986) Auditory
evoked potentials during isoflurane anaesthesia. Acta
Anaesthesiol. Scand., 30(5): 378380.
Schonle, PW, Isenberg, C, et al. (1989) Changes of transcranially evoked motor responses in man by midazolam, a short acting benzodiazepine. Neuroscience Lett.,
101(3): 321324.
Schubert, A, Licina, MG, et al. (1990) The effect
of ketamine on human somatosensory evoked potentials
and its modification by nitrous oxide [erratum appears
in Anesthesiology 1990 Jun; 72(6):1104]. Anesthesiology, 72(1): 3339.
Schubert, A, Licina, MG, et al. (1992) Systemic lidocaine
and human somatosensory-evoked potentials during
sufentanil-isoflurane anaesthesia. Can. J. Anaesth.,
39(6): 569575.
Schwender, D, Klasing, S, et al. (1993) Mid-latency auditory evoked potentials during ketamine anaesthesia in
humans. Br. J. Anaesth., 71(5): 629632.
Schwender, D, Klasing, S, et al. (1996) Midlatency auditory
evoked potentials during anaesthesia with increasing
endexpiratory concentrations of desflurane. Acta Anaesthesiol. Scand., 40(2): 171176.
Schwentker, MC, Russell, GB, et al. (1995) Myogenic
response distortion of neurogenic motor evoked potential morphology. Anesthesiology, 83(3): 616619.
Sebel, PS, Ingram, DA, et al. (1986) Evoked potentials
during isoflurane anaesthesia. Br. J. Anaesth., 58(6):
580585.
Sebel, PS, Erwin, CW, et al. (1987) Effects of halothane
and enflurane on far and near field somatosensory
evoked potentials. Br. J. Anaesth., 59(12): 14921496.
Seyal, M and Mull, B (2002) Mechanisms of signal change
during intraoperative somatosensory evoked potential
monitoring of the spinal cord. J. Clin. Neurophysiol.,
19(5): 409415.
Shapiro, SM (2002) Somatosensory and brainstem auditory
evoked potentials in the Gunn rat model of acute bilirubin neurotoxicity. Pediatr. Res., 52(6): 844849.
Sharpe, RM, Brosnan, S, et al. (1997a) The effect of sevoflurane on the audiotry evoked response, spectral edge
and median frequency in man. Br. J. Anaesth., 78:
282285.
Sharpe, RM, Nathwani, D, et al. (1997b) Auditory evoked
response, median frequency and 95% spectral edge during anaesthesia with desflurane and nitrous oxide [see
comment]. Br. J. Anaesth., 78(3): 282285.
Shields, CB, Paloheimo, MPJ, et al. (1990) Intraoperative
use of transcranial magnetic motor evoked potentials.
In: S Chokroverty (Ed.), Magnetic Stimulation in Clinical
Neurophysiology. Butterworths, London, pp. 173184.
NTTI
Shimoji, K and Kano, T (1973) Evoked electrospinogram:
interpretation of origin and effects of anesthetics. In:
MI Phillips (Ed.), Brain Unit Activity During Behavior.
Charles C. Thomas, Springfield, pp. 171190.
Shimoji, K, Maruyama, Y, et al. (1984) The effects of
anesthetics on somatosensory evoked potentials from
the brain and spinal cord in man. In: QJ Gomez, LM
Egay and MF De La Cruz Odi (Eds.), Anaesthesia
Safety for All. Elsevier Science Publishers B.V., New
York, pp. 159164.
Skinner, S, Achenbach, J, et al. (2003) Detailed analysis of
persistent loss of transcranial electrical motor evoked
potentials (TceMEP) during spinal deformity correction
surgery. In: Proceedings of the 14th Annual Meeting of
the American Society of Neurophysiological Monitoring. Las Vegas, NV.
Sloan, TB (1985) Lecture 211: Clinical Monitoring of the
Brain and Spinal Cord. American Society of Anesthesiology: Refresher Courses.
Sloan, TB (1994) Nondepolarizing neuromuscular blockade does not alter sensory evoked potentials. J. Clin.
Monit., 10(1): 410.
Sloan, TB (1997) Evoked Potentials. In: MA Albin (Ed.), Textbook of Neuroanesthesia with Neurosurgical and Neuroscience Perspectives. McGraw-Hill, New York, pp. 221276.
Sloan, TB and Angell, D (1993) Differential effect of isoflurane on motor evoked potentials elicited by transcortical
electric or magnetic stimulation. In: SS Jones, S Boyd, M
Hetreed and NJ Smith (Eds.), Handbook of Spinal Cord
Monitoring. Kluwer Academic Publishers, Hingham,
MA, pp. 362367.
Sloan, TB and Erian, R (1993a) Effect of atracuriuminduced neuromuscular block on cortical motor-evoked
potentials. Anesth. Analg., 76(5): 979984.
Sloan, TB and Erian, R (1993b) Effect of vecuroniuminduced neuromuscular blockade on cortical motor
evoked potentials. Anesthesiology, 78(5): 966973.
Sloan, TB and Koht, A (1985) Depression of cortical
somatosensory evoked potentials by nitrous oxide. Br.
J. Anaesth., 57(9): 849852.
Sloan, TB and Levin, D (1993) Etomidate amplifies and
depresses transcranial motor evoked potentials in the
monkey. J. Neurosurg. Anesthesiol., 5: 299.
Sloan, TB, Ronai, AK, et al. (1988) Improvement of
intraoperative somatosensory evoked potentials by etomidate. Anesth. Analg., 67(6): 582585.
Sloan, TB, Fugina, ML, et al. (1990) Effects of midazolam
on median nerve somatosensory evoked potentials. Br.
J. Anaesth., 64(5): 590593.
Sloan, TB, Rogers, JNR, et al. (1995) MAC fractions of
nitrous oxide and isoflurane are not electrophysiologically additive in the ketamine anesthetized baboon. J.
Sohmer, H, Gafni, M, et al. (1982) Auditory nerve-brainstem
potentials in man and cat under hypoxic and hypercapnic

conditions. Electroencephalogr. Clin. Neurophysiol.,
53(5): 506512.
Sohmer, H, Freeman, S, et al. (1986) The depression of the
auditory nerve-brainstem evoked response in hypoxaemia mechanism and site of effect. Electroencephalogr. Clin. Neurophysiol., 64(4): 334338.
Sohmer, H, Gold, S, et al. (1989) Effects of hypothermia
on auditory brainstem and somatosensory evoked
responses. A model of a synaptic and axonal lesion.
Electroencephalogr. Clin. Neurophysiol., 74(1):
5057.
Starr, A and Achor, LJ (1979) Anatomical and physiological origins of auditory brainstem responses. In: D Lehmann and E Callaway (Eds.), Human Evoked
Potentials: Applications and Problems. Plenum Press,
New York, pp. 415429.
Stephen, JP, Sullivan, MR, et al. (1996) Cotrel-dubousset
instrumentation in children using simultaneous motor
and somatosensory evoked potential monitoring. Spine,
21(21): 24502457.
Steriade, M (2000) Brain Electrical Activity and Sensory
Processing During Waking and Sleep Stated. W.B.
Stinson, LW, Jr., Murray, MJ, et al. (1994) A computercontrolled, closed-loop infusion system for infusing
muscle relaxants: its use during motor-evoked potential
monitoring. J. Cardiothorac. Vasc. Anesth., 8(1): 4044.
Stockard, JJ, Sharbrough, FW, et al. (1978) Effects of
hypothermia on the human brainstem auditory response.
Ann. Neurol., 3(4): 368370.
Stone, JL, Ghaly, RF, et al. (1992) A comparative analysis of
enflurane anesthesia on primate motor and somatosensory
evoked potentials. Electroencephalogr. Clin. Neurophysiol., 84(2): 180187.
Streinzer, W, Gilly, H, Redl, G, Draxler, V, Zrunek, M and
Hofler, H (1986) Differences in the neuromuscular blockade of the larynx and thenar muscles following relaxation
with vecuronium. Laryngol. Rhinol. Octol., 65(11):
628631.
Svensson, P, Arendt-Nielsen, L, et al. (1993) Oral mucosal
analgesia quantitatively assessed by argon laser-induced
thresholds and single-evoked vertex potentials. Anesth. Pain
Control Dent., 2(3): 154161.
Symon, L and Murota, T (1989) Intraoperative monitoring
of somatosensory evoked potentials during intracranial
vascular surgery. In: JE Desmedt (Ed.), Neuromonitoring for Surgery. Elsevier, Amsterdam, pp. 263302.
Symon, L, Wang, AD, et al. (1984) Perioperative use of
somatosensory evoked responses in aneurysm surgery.
J. Neurosurg., 60(2): 269275.
Tabaraud, F, Boulesteix, JM, et al. (1993) Monitoring of
the motor pathway during spinal surgery. Spine, 18(5):
546550.
Takaki, O, Kuro, M, et al. (1992) Effects of hypothermia
with cardiopulmonary bypass on posterior tibial nerve
125
somatosensory evoked potentials in man. Masui Japanese J. Anesthesiol., 41(7): 11191124.
Taniguchi, M, Schram, J, et al. (1991) Recording of myogenic motor evoked potential (mMEP) under general
anesthesia. In: J Schramm and AR Mller (Eds.),
Intraoperative
Neurophysiological
Monitoring.
Springer Verlag, Berlin, pp. 7287.
Taniguchi, M, Cedzich, C, et al. (1993a) Modification of
cortical stimulation for motor evoked potentials under
general anesthesia: technical description. Neurosurgery,
32(2): 219226.
Taniguchi, M, Nadstawek, J, et al. (1993b) Effects of four
intravenous anesthetic agents on motor evoked potentials elicited by magnetic transcranial stimulation. Neurosurgery, 33(3): 407415; discussion 415.
Taylor, BA, Fennelly, ME, et al. (1993) Temporal summation the key to motor evoked potential spinal cord
monitoring in humans. J. Neurol. Neurosurg. Psychiatr., 56(1): 104106.
Thornton, C (1991) Evoked potentials in anaesthesia. Eur.
J. Anaesthesiol. Suppl., 8(2): 89107.
Thornton, C, Catley, DM, et al. (1983) Enflurane anaesthesia causes graded changes in the brainstem and early
cortical auditory evoked response in man. Br. J.
Anaesth., 55(6): 479486.
Thornton, C, Heneghan, CP, et al. (1984) Effects of halothane or enflurane with controlled ventilation on
auditory evoked potentials. Br. J. Anaesth., 56(4):
315323.
Thornton, C, Konieczko, KM, et al. (1989) Effect of propofol on the auditory evoked response and oesophageal
contractility. Br. J. Anaesth., 63(4): 411417.
Thornton, C, Creagh-Barry, P, et al. (1992) Somatosensory
and auditory evoked responses recorded simultaneously:
differential effects of nitrous oxide and isoflurane [see
comment]. Br. J. Anaesth., 68(5): 508514.
Thornton, C, Lucas, MA, et al. (1999) Effects of
dexmedetomidine on isoflurane requirements in healthy
volunteers. 2. Auditory and somatosensory evoked
responses. Br. J. Anaesth., 83(3): 381386.
Ting, CH, Angel, A, et al. (2003) Neuronal network modelling of the effects of anaesthetic agents on somatosensory pathways. Biol. Cybern., 88(2): 99107.
Tooley, MA, Greenslade, GL, et al. (1996) Concentrationrelated effects of propofol on the auditory evoked
response. Br. J. Anaesth., 77(6): 720726.
Ubags, LH, Kalkman, CJ, et al. (1996) The use of a circumferential cathode improves amplitude of intraoperative electrical transcranial myogenic motor evoked
responses. Anesth. Analg., 82(5): 10111014.
Ubags, LH, Kalkman, CJ, et al. (1997) The use of ketamine or
etomidate to supplement sufentanil/N2O anesthesia does
not disrupt monitoring of myogenic transcranial motor
evoked responses. J. Neurosurg. Anesthesiol., 9(3):
228233.
126
Ubags, LH, Kalkman, CJ, et al. (1998) Influence of isoflurane on myogenic motor evoked potentials to single and
multiple transcranial stimuli during nitrous oxide/opioid
anesthesia. Neurosurgery, 43(1): 9094; discussion 945.
Van Dongen, EP, Ter Beek, HT, et al. (1999a) Effect of
nitrous oxide on myogenic motor potentials evoked by
a six pulse train of transcranial electrical stimuli: a possible monitor for aortic surgery. Br. J. Anaesth., 82(3):
323328.
Van Dongen, EP, Ter Beek, HT, et al. (1999b) Withinpatient variability of myogenic motor-evoked potentials
to multipulse transcranial electrical stimulation during
two levels of partial neuromuscular blockade in aortic
surgery. Anesth. Analg., 88(1): 2227.
Van Dongen, EP, Ter Beek, HT, et al. (1999c) The influence of
nitrous oxide to supplement fentanyl/low-dose propofol
anesthesia on transcranial myogenic motor-evoked potentials during thoracic aortic surgery. J. Cardiothorac. Vasc.
Anesth., 13(1): 3034.
Vaugha, DJ, Thornton, C, et al. (2001) Effects of different
concentrations of sevoflurane and desflurane on subcortical somatosensory evoked responses in anaesthetized,
non-stimulated patients. Br. J. Anaesth., 86(1): 5962.
Velasco, M, Velasco, F, et al. (1984) Effect of fentanyl
and naloxone on human somatic and auditory-evoked
potential components. Neuropharmacology, 23(3):
359366.
Watt, JW, Fraser, MH, et al. (1996) Total i.v. anaesthesia
for transcranial magnetic evoked potential spinal cord
monitoring. Br. J. Anaesth., 76(6): 870871.
Waxman, SG, Ransom, BR, et al. (1991) Non-synaptic
mechanisms of Ca(2)-mediated injury in CNS white
matter. Trends Neurosci., 14(10): 461468.
Weight, FF and Erulkar, SD (1976) Synaptic transmission
and effects of temperature at the squid giant synapse.
Nature, 261(5562): 720722.
Wiedemayer, H, Fauser, B, et al. (2002) The impact of
neurophysiological intraoperative monitoring on surgical decisions: a critical analysis of 423 cases. J. Neurosurg., 96(2): 255262.
Winters, WD, Mori, K, et al. (1967) The neurophysiology of
anesthesia. Anesthesiology, 28(1): 6580.
Witzmann, A and Reisecker, F (1989) Somatosensory and
auditory evoked potentials monitoring in tumor removal
and brainstem surgery. In: JE Desmedt (Ed.), Neuromonitoring for Surgery. Elsevier, Amsterdam, pp. 219241.
Woodforth, IJ, Hicks, RG, et al. (1996) Variability of
motor-evoked potentials recorded during nitrous oxide
anesthesia from the tibialis anterior muscle after transcranial electrical stimulation. Anesth. Analg., 82(4):
744749.
Yamada, T, Muroga, T, et al. (1981) Tourniquet-induced
ischemia and somatosensory evoked potentials. Neurology, 31(12): 15241529.
NTTI
Yamada, H, Transfeldt, EE, et al. (1994) The effects of
volatile anesthetics on the relative amplitudes and
latencies of spinal and muscle potentials evoked by
transcranial magnetic stimulation. Spine, 19(13):
15121517.
Yang, LH, Lin, SM, et al. (1994) Intraoperative
transcranial electrical motor evoked potential monitoring during spinal surgery under intravenous ketamine
or etomidate anaesthesia. Acta Neurochir., 127(3/4):
191198.
Yeoman, RR, Moreno, L, et al. (1980) Enflurane effects on
acoustic and photic evoked responses. Neuropharmacology, 19(5): 481489.
York, DH, Pulliam, MW, et al. (1981) Relationship
between visual evoked potentials and intracranial pressure. J. Neurosurg., 55(6): 909916.
Young, W and Sakatani, K (1990) Neurophysiological
mechanisms of somatosensory-evoked potential changes.
In: SK Salzman (Ed.), Neural Monitoring. Humana
Press, Clinton, NJ, 127142.
Young, SS, et al. (1994) Magnetic motor evoked potentials
during methohexital anesthesia in the dog. Neurosurgery, 34(3): 490495.
Zeitlhofer, J, Steiner, M, et al. (1990) The influence of
temperature on somatosensory-evoked potentials during cardiopulmonary bypass. Eur. Neurol., 30(5):
284290.
Zentner, J (1989) Noninvasive motor evoked potential
monitoring during neurosurgical operations on the spinal cord. Neurosurgery, 24(5): 709712.
Zentner, J (1991a) Motor evoked potential monitoring during neurosurgical operations on the spinal cord. Neurosurg. Rev., 14(1): 2936.
Zentner, J (1991b) Motor evoked potential monitoring in
operations of the brainstem and posterior fossa. In: J
Schramm and AR Mller (Eds.), Intraop Neurophysiol
Monitoring. Springer-Verlag, Berlin, pp. 95105.
Zentner, J and Ebner, A (1989) Nitrous oxide suppresses
the electromyographic response evoked by electrical
stimulation of the motor cortex. Neurosurgery, 24(1):
6062.
Zentner, J, Kiss, I, et al. (1989) Influence of anesthetics
nitrous oxide in particular on electromyographic
response evoked by transcranial electrical stimulation
of the cortex. Neurosurgery, 24(2): 253256.
Zentner, J, Albrecht, T, et al. (1991) Propofol increases
amplitudes of SEP. Funct. Neurol., 6(4): 411412.
Zentner, J, Albrecht, T, et al. (1992) Influence of
halothane, enflurane, and isoflurane on motor evoked
potentials [see comment]. Neurosurgery, 31(2):
298305.
Zhou, HH and Zhu, C (2000) Comparison of isoflurane
effects on motor evoked potential and F wave. Anesthesiology, 93(1): 3238.
SECTION II
CNP Techniques Used in Monitoring Neural

Function in Surgery

M.R. Nuwer (Ed.)
128
CHAPTER 6
Electroencephalography used in monitoring neural

function during surgery
A.C. Van Huffelen*
Department of Clinical Neurophysiology (F.02.230), University Medical Center Utrecht, 3508 GA Utrecht, The Netherlands
6.1. Introduction
The main objective of intraoperative monitoring is to
prevent new neurologic impairment by identifying it
sufficiently early to allow prompt correction of its
cause (Burke et al., 1999).
A further advantage of monitoring is precise registration of adverse effects occurring during surgery
enabling retrospective analysis. This may not only
be of scientific but also of medicolegal importance.
Electroencephalography (EEG) is very sensitive to
impairment of oxygenation and circulation. Circulatory impairment may occur during carotid endarterectomy (CEA) (carotid clamping), cardiothoracic
surgery (short periods of circulatory arrest during
implantation of cardioverter defibrillator or prolonged
circulatory arrest with profound hypothermia), and
intracranial vascular surgery (aneurysm and bypass
surgery). Systemic hypotension may also give rise to
deterioration of the EEG during any form of surgery.
One should, however, realize that the EEG changes
caused by ischemia during surgery are different from
those occurring in awake subjects (Jonkman et al.,
1986). The interaction between anesthetic drugs (see
Chapter 4) and the effects of ischemia complicate
EEG monitoring. A further complication may be introduced with profound hypothermia.
For intraoperative monitoring of the effects of
hypoxia-ischemia to the brain, several techniques are
available. There are techniques based on the measurement of circulation (regional cerebral blood flow
[rCBF], stump pressure measurement, and transcranial
Doppler sonography [TCD]). Disorders of oxygenation
*
Correspondence to: A.C. Van Huffelen, M.D., Ph.D.,

Department of Clinical Neurophysiology (F.02.230), University Medical Center Utrecht, P.O. Box 85500, 3508
GA Utrecht, The Netherlands.
Tel.: 31-30-2506856; fax: 31-30-2505494.
E-mail: a.c.vanhuffelen@umcutrecht.nl (A.C. Van Huffelen).
may be detected with measurement of the oxygen saturation in the jugular vein (SjvO2) and near infrared spectroscopy (NIRS). Disordered brain function as a
consequence of hypoxia-ischemia may become apparent with EEG and somatosensory evoked potential
(SEP) monitoring. These latter two clinical neurophysiological (CNP) techniques are noninvasive, continuous,
very sensitive, and applicable to all patients, the EEG
covering the function of all cortical brain regions and
the SEP restricted to the somatosensory projections to
the cortex.
6.2. Anesthesia and surgery
During surgery, a close cooperation between the surgical, the anesthetic, and the CNP team is of major
importance. Continuous communication between the
anesthetic team and the CNP team is essential for
proper monitoring. The anesthetist should continuously provide data on arterial blood pressure (ABP),
oxygen saturation, and pCO2 to the CNP team. It is
recommended (Burke et al., 1999) that each institution
develops its own anesthetic regimens in close cooperation with all members of the anesthetic and CNP
teams. No single anesthetic regimen can be considered
superior to others.
In general, EEG monitoring demands a continuous,
rather superficial, level of anesthesia. Therefore, some
groups prefer the use of a local cervical plexus block
allowing such a superficial level of anesthesia. The
anesthetist should provide a stable depth of anesthesia
together with stable pO2 and pCO2. At the moments
when EEG monitoring is critical (e.g., carotid clamping), no change in anesthetic regimen should be introduced (e.g., no bolus administration of centrally active
drugs). Any change in the anesthetic regimen should
be communicated to the EEG monitoring team. These
data should also be clearly documented in a way
enabling reviewing of the monitored data. It is of
CNP TECHNIQUES USED IN MONITORING NEURAL FUNCTION IN SURGERY
129
utmost importance that the neurophysiology team is

familiar with the different effects of the anesthetic
agents used at their institution (Laman et al., 2001)
(for details Chapter 4).
There should also be a close cooperation between
the surgical and the CNP team. The surgeon should
indicate the events during surgery (clamping, removal
of clamp). Because surgical diathermy can block the
EEG amplifiers for some time, the surgeon should
preferably not use diathermy during critical stages of
monitoring (e.g., test clamping during CEA).
to maintain low (<5 kO) electrode contact impedance

to minimize electrostatic effect. The electrode leads
must be flexible and light. The electrode box may
be fixed to the operation table or to the recording
apparatus by means of an overhead arm.
6.3. EEG general equipment requirements
One would compare right hemisphere to left hemi-
EEG monitoring requires the continuous observation

of the EEG by an appropriately qualified clinical
neurophysiologist and an experienced neurophysiology technician. The equipment used in the operating
room (OR) has to conform to OR safety specifications. There should be an adequate common mode
rejection to eliminate 50 or 60 Hz line interferences
(at least 85 dB). There should also be automatic artifact rejection to minimize interference due to surgical
diathermy. Special attention should be given to electrical safety of the patient. Any leakage of current
through the patient exceeding 10 mA should be prevented. This implies that the patient should be properly electrically grounded at only one site. The
diathermy return plate should be positioned as close
as possible to the operating site. The application of
defibrillators and fibrillators may pose similar problems (introduction of implantable cardioverter defibrillator). The same ground is used for diathermy and
other OR equipment. Neurophysiology equipment
should have optical isolation of each patient contact.
Biomedical engineers should check the equipment
for proper grounding and for leakage current on a
regular basis (Burke et al., 1999).
6.4.2. Montages
Many different electrode montages for EEG monitoring
of ischemia have been applied. However, all authors so
far have used symmetrical montages (Fig. 1). Some
basic ideas behind these different choices were:
sphere derivations for the detection of asymmetries.
Midline electrodes (Fz, Cz) were considered far
distance reference electrodes enabling the detection
of asymmetries (Chiappa et al., 1979; Minicucci
et al., 2000; Laman et al., 2001, 2005).
Long interelectrode distance derivations were considered appropriate because one expected general
changes due to ischemia in anesthetized patients
Fp2-O2 (Chiappa et al., 1979), F4-O2 (Myers et al.,
1977), F4-P4 (Pronk and Simons, 1982; Krul et al.,
1989; Kearse et al., 1993; Nuwer et al., 1993), and
F8-T6 (Modica et al., 1992).
Some authors used symmetrical bipolar serial montages with short interelectrode distances (double
banana or parts of it) (Van Putten et al., 2004).
6.4. Electrodes and montages

6.4.1. Electrodes
Either needle or disk electrodes may be used. For both
types of electrodes, it is important that they are sufficiently anchored to the scalp with adhesive tape. Disk
electrodes attached with collodium are to be preferred
because they are mechanically secure and have a lower
contact impedance and a better frequency response
than needles (Burke et al., 1999). In the OR, it is especially important to use only one type of electrodes and
Fig. 1. Electrode positions found in the literature. The area

of the circle on each electrode position is proportional to
the frequency in which that electrode position was used in
the literature. Reproduced from Laman et al. (2001) J. Clin.
Neurophysiol. with permission by Lippincott, Williams and
Wilkins.
130
Some authors (Cucchiara et al., 1979) preferred
derivations including the posterior brain half electrodes (C4, T4, P4, T6, O2) which are essential for
the study of the mu and alpha rhythms. These
regions are very sensitive to hypoxic-ischemic
EEG changes in conscious patients (Jonkman
et al., 1986; Kraaier et al., 1988).
A study aiming at the detection of the optimal
electrode derivations for the detection of ischemia
(indicating that a shunt was needed) was undertaken
by Laman et al. (2001).
Starting with a 16-channel montage with Cz as a
common reference all 136 possible derivations were
computed: 16 with common reference Cz and 120
bipolar derivations. The derivations with the highest
discriminating power between shunt and no-shunt
groups were considered optimal. A ranking of all electrode derivations according to the area under the
receiver operating characteristic (ROC) curve was
determined. The results were somewhat different for
the two anesthetic regimens applied, but both had
some rather unexpected outcomes: the derivations
with the highest discriminating power were found unilaterally, ipsilateral to the side of carotid clamping, and
in the anterior half of that ipsilateral hemisphere, thus
in the anterior ipsilateral quadrant (Fig. 2). The optimal derivations included derivations with Cz as a common reference and short- and long-distance bipolar
derivations. The preference for F4-F8-C4-T4 (when
the right carotid artery was clamped) in a Cz common
reference montage was corroborated in a second study
(Laman et al., 2005).
A.C. VAN HUFFELEN
6.5. EEG interpretation

6.5.1. Ischemia
The interpretation of the complex EEG changes during intraoperative monitoring of ischemia is highly
subjective. Nevertheless, interobserver agreement concerning the necessity of shunting during CEA surgery
appears to be rather high (0.70.77, Laman et al.,
2001). The EEG changes during intraoperative monitoring are different from those occurring in awake subjects
(Kraaier et al., 1988) (Fig. 3). They are related to anesthesia (anesthetic agents used and depth of anesthesia)
and impairment of circulation (severity and abruptness).
Both anesthetic and circulatory effects may develop
bilaterally. For anesthetic agents, this is self-evident.
For circulatory effects, one should realize that unilateral
clamping of a carotid artery may give rise to a momentary redistribution of flow over the circle of Willis leading to bilateral EEG changes (Visser et al., 1999; Vriens
et al., 2001). Overall, in about one-third of the patients, a
bilateral EEG effect is seen (Blume et al., 1986) (Fig. 4).
Such bilateral changes occur more often in patients with
a stenosis or occlusion of the contralateral carotid
artery.
The neurophysiologist should have a priori knowledge of the EEG modifications due to anesthesia,
especially concerning the anesthetic agents used in
his institution. In general, these drugs cause an
increase in slow activity, but also an increase in fast
activity especially over the anterior head regions
(Chapter 4). It appears, however, that these effects
are different for some anesthetic agents (Laman
Fig. 2. Optimal electrode positions found by receiver operating characteristic (ROC) curves. A: Isoflurane anesthesia and
B: propofol anesthesia. The area of the circle on each electrode position is proportional to the frequency in which that electrode position was involved in the 20% highest-ranking derivations in each frequency band. Reproduced from Laman et al.
(2001) with permission by Lippincott, Williams and Wilkins.
Fig. 3. Subtraction spectrum (difference between hypobaric hypoxic (C) and baseline (B) condition averaged over
21 subjects. Pressure within low-pressure chamber:
46.5 kPa (corresponding to height of 6,096 m) derivation
P4-O2. Alert, eyes closed. Increase in the relative power
of low frequencies, decrease in the relative alpha and beta
power. () p < 0.01 and () p < 0.05. Reprinted from
Kraaier et al. (1988) with permission from the International
Federation of Clinical Neurophysiology.
et al., 2001) (Fig. 5). Deeper levels of anesthesia

causing a burst-suppression pattern or even complete
disappearance of all EEG activity preclude appropriate EEG monitoring in CEA.
Impaired circulation leads to an increase in slow,
especially slow delta activity. In contrast, there is
always a decrease in alpha and beta activity. Thus,
both deepening of level of anesthesia and circulatory
impairment may lead to an increase in slow activity.
However, anesthesia and ischemia have opposite
effects on faster (alpha and beta) activity.
131
One should realize that the EEG effects of circulatory impairment become only apparent with some
delay. With complete circulatory arrest, the EEG
changes develop after some 15 s (Visser et al.,
2001) (Fig. 6). The first changes consist of an initial
increase of power in the alpha range and a decrease
of power in the beta range. After 15 s, the power in
the alpha range starts to decrease, whereas the beta
range power continues to decrease. At the same
moment, the power in the delta1 range starts to
increase, whereas in the delta2 range, the power starts
to decrease. After about 30 s, the EEG shows electrocerebral silence (ECS). When there is a partial
impairment of circulation (unilateral carotid clamping), the development of EEG changes may take longer, but is always present within 2 min (Laman et al.,
2005). This rather slow EEG development demands
an adequate time basis for assessment: at least two
times that of routine EEG recording (Chiappa et al.,
1979).
Several authors have used sets of criteria for the
interpretation of ischemic EEG changes during monitoring. The basic principle is a decrease of fast
(alpha and beta) activity and an increase of slow
(delta and theta) activity. This has been quantified
into 50% increase of slow activity (Nuwer et al.,
1993; Burke et al., 1999; Pinkerton, 2002). A distinction in major and moderate EEG effects has been
made by Blume et al. (1986) and used by Minicucci
Fig. 4. The effect of clamping of the right internal carotid artery (ICA). Eight EEG channels are shown and transcranial
Doppler envelope for the ipsilateral middle cerebral artery. Reproduced from Laman (2004) thesis.
132
A.C. VAN HUFFELEN
Positive and negative Z-scores > 5

ISOFLURANE
PROPOFOL
100
100
Z>5
80
60
60
40
40
Percentage of shunt patients
Percentage of shunt patients
Z>5
80
20
0
20
40
20
0
20
40
60
60
80
80
Z < 5
100
Z < 5
100
delta theta alpha beta
delta theta alpha beta
Fig. 5. Percentage of shunt patients with large clamping induced power changes with high Z-scores (>5) or (<5) in the
optimal derivations in both anesthetic regimens showing different behavior in the frequency bands in each anesthetic regimen. Reproduced from Laman et al. (2001) with permission by Lippincott, Williams and Wilkins.
et al. (2000) and Cursi et al. (2005). As major EEG

effects were seen: attenuation of all activity by at
least 75% and/or 100% or more increase of <1 Hz
delta activity. As moderate EEG effects were considered: attenuation of all nondelta activity to 50% of
the preclamp condition and persistent increase of
delta activity >1 Hz. It should be noticed that these
criteria were both applied to unilateral and bilateral
changes.
It was found (Blume and Sharbrough, 2005) that
only major EEG effect demanded introduction of a
shunt. In some patients, Visser et al. (1999) found a
temporary increase of beta activity during carotid
clamping, which was interpreted as a sign of arousal.
Debatisse et al. (2005) found an increase of fast beta
activity in combined electrocorticography (ECoG)
and scalp EEG recording during temporary clipping
of arteries during aneurysm surgery. This effect was,
however, temporary and only visible in the ECoG.
6.5.2. Hypothermia
The EEG may be used for monitoring brain function
during cardiovascular surgery (involving extracorporeal circulation and hypothermia). Hypothermia is
induced to protect brain function during periods of
prolonged circulatory arrest. The state of ECS
induced by profound hypothermia is considered a
state of cerebral inactivity protecting the brain
against the effects of hypoxia-ischemia. This procedure is used both in children (Akiyama et al., 2001)
and adults (Mizrahi et al., 1989). During cooling,
the EEG shows a characteristic evolution. Initially,
there is a continuous change consisting of a progressive depression of amplitude and overall slowing of
background activity. The amplitude attenuation is
most pronounced for the beta band activities in the
anterior head regions (Akiyama et al., 2001). Then,
a discontinuous pattern, a suppression-burst pattern,
power difference (dB)
133
10
5
0
5
delta 1 factor
10
5
0
5
10
delta 2 factor
15
5
0
5
10
alpha factor
15
5
0
5
10
15
beta factor
5
17
29
time after circ. arrest
15
27
39
51
63
75
87
99
111
time after restoration of circulation (s)
Fig. 6. Spectral changes for 16 channels and 4 frequency ranges (delta1 (0.00.5 Hz); delta2 (1.53.0 Hz); alpha
(7.59.5 Hz); beta (15.020.0 Hz)) during circulatory arrest (ventricle fibrillation) and after restoration of circulation
(defibrillation), averaged over 54 patients. Reproduced from Visser et al. (2001) with permission by Lippincott, Williams
and Wilkins.
develops. This is succeeded by a pattern of periodic

slow wave activity of moderate voltage and long
duration alternating with progressively longer periods
of suppression. This pattern eventually develops into
ECS.
These EEG changes may be assessed visually,
studying the raw EEG (Mizrahi et al., 1989) or compressed spectral array, or quantitatively using the spectral edge frequency (SEF95) (Akiyama et al., 2001).
It should be realized that three different factors,
viz. progressive hypothermia, increasing depth of
anesthesia, and decrease in perfusion/circulation

may give rise to comparable EEG changes. On the
basis of progressive EEG changes, it is impossible
to disentangle the effects of these three factors.
One should also notice that the more pronounced
the EEG changes are, the less sensitive the EEG
becomes for disturbances of one of these contributing
factors (e.g., unilateral or local ischemia). At a level
of profound hypothermia with ECS, the use of SEPs
might be helpful (Guerit, 1998; Florence et al.,
2004).
134
A.C. VAN HUFFELEN
6.6. Processed EEG monitoring

6.6.1. Processed EEG monitoring techniques
The complexity and slowness of the EEG changes
during the intraoperative development of hypoxiaischemia as it is compromised by anesthetic agents
and possibly hypothermia and the subjectivity of its
interpretation have led to the use of several ways of
signal processing of the EEG. The ultimate goal of
these methods is to make intraoperative EEG monitoring a simple, objective, and reliable technique.
In the early seventies, amplitude-integrated EEG
(cerebral function monitor [CFM]) was developed
(Prior et al., 1971). This technique amplifies and filters the EEG signal (bandwidth 216 Hz), rectifies
the signal, and plots the envelope of the amplitude
distribution semilogarithmically (Fig. 7).
Nowadays, nearly all the techniques used start
with Fourier analysis of the EEG signal. Before this
analysis is applied, the EEG signal is generally filtered with cutoff values of 0.5 or 1 Hz for the high
pass filter and 35 or 70 Hz for the low pass filter.
Some authors have studied low delta activity (filter
settings 0.05 or 0.1 Hz) (Visser et al., 2001) and
others high beta frequency (filter settings 100 Hz)
for specific purposes (Debatisse et al., 2005). Factor

analysis (Visser et al., 2001) has shown that different
effects are present in the subdelta, delta, alpha, and
beta ranges (Fig. 8). The sampling rate used for digitization of the signal varies from 256 to 1,024 Hz at
least at 12-bit resolution. In view of the activities
studied, a sampling rate of 128 Hz can be considered as adequate. The precision of the frequency
analysis is determined by the epoch length used.
Many authors used a 2-s epoch length (0.5 Hz precision), others 4 s (0.25 Hz), 8 s (0.125 Hz), or 10 s
(0.1 Hz). With respect to the parameters that have
been used, such as broad band parameters or SEF
parameters a 0.5 Hz precision may be considered
adequate.
The spectral power is generally integrated within
some broad bands: delta (0.54 Hz), theta (48 Hz),
alpha (812, 813 or 815 Hz), and beta activity up
to 20, 25, or 30 Hz). There are no clear indications
that some limits are better than others. Factor analysis performed by Visser et al. (2001) revealed after
rotation four factors with eigen values above 0.71:
the delta1 band (0.00.5 Hz), the delta2 band (1.5
3.0 Hz), the alpha1 band (7.59.5 Hz), and the beta1
band (15.020 Hz) that should be considered separate
EEG patient 1
A
(Fig. 7 continued)
CFM
T C
135
DSAR
DSAL
B
MDF
SEF90
BSI
0.25
0.20
0.15
0.10
0.05
0
10
15
20
25
30
35
40
45
(Fig. 7 continued)
136
A.C. VAN HUFFELEN
EEG patient 2
D
CFM T
DSAR
DSAL
E
(Fig. 7 continued)

MDF T
137
SEF90
BSI
0.25
BSI
0.20
0.15
0.10
0.05
0
10
20
30
40
50
60
Fig. 7. EEG monitoring of carotid endarterectomy. Comparison of EEG signals processed according to several techniques,
all applied to the same EEG signal: cerebral function monitor (CFM), density-modulated spectral array (DSA), main dominant frequency (MDF) in the 815 Hz range, SEF 90% (SEF90), brain symmetry index (BSI). EEG changes during test
clamp (T), clamping (C) for introduction of shunt, and clamping for its removal (R) are indicated. Right side channels:
black lines, left side channels: gray lines. Patient 1: ipsilateral decrease of fast activity during clamping of right carotid
artery.
(a) EEG: ipsilateral decrease of fast activity
(b) Upper part: CFM: no clear changes
Lower part: DSA: bilateral decrease of theta and alpha activity
(c) Upper part: MDF: bilateral decrease
Middle part: SEF90: asymmetrical changes
Lower part: BSI: clear asymmetries
Patient 2: bilateral slowing and attenuation of all activity during clamping.
(d) EEG: attenuation of all activity
(e) Upper part: CFM: decrease of activity during longer clamp periods (C, R)
Lower part: DSA: bilateral decrease of all activity above delta band
(f) Upper part: MDF: bilateral decrease
Middle part: SEF90: bilateral decrease
Lower part: BSI: no clear asymmetries
138
A.C. VAN HUFFELEN

1.00
0.80
Factor loading
0.60
0.40
0.20
0.00
0.20
0.40
0.60
Factor 1
Factor 2
10
15
20
25
Frequency
Fig. 8. Factor analysis to study the log power differences of the spectral EEG changes during carotid clamping compared with
baseline (94 patients). Unrotated factor loadings for two factors (first explaining 42% of variance, second 22% of variance).
Reprinted from Visser et al. (1999) with permission from the International Federation of Clinical Neurophysiology.
factors. Thus, the bands used by most authors contain

the delta2, alpha1, and beta1 band, which all show a
decrease of power after circulatory arrest (only the
alpha1 band shows a short increase in power directly
after arrest) (Fig. 6).
The following approaches may be distinguished.
Right to left comparison will only give reliable

results in patients with asymmetries due to unilateral
carotid clamping (about two thirds of patients). Event
to baseline comparison will be applicable both to
carotid and cardiovascular surgery.
Representation of the whole power spectrum. This
6.6.2. Display of (processed) EEG
may be represented as a condensed spectral array

as a density-modulated spectral array (DSA)
(Minicucci et al., 2000) (Fig. 7).
Broad band parameters for the delta, theta, alpha,
and beta bands. The power in these bands may be
used absolutely and relative to the total power of
the spectrum (Laman et al., 2001, 2005). Also some
band quotient (e.g., (alpha beta)/(delta theta))
have been used. Unfortunately, not all authors have
used the same frequencies for their band limits.
Spectral edge frequencies, which are the frequencies below which a certain amount of energy (e.g.,
90%) of the spectrum is present (Fig. 7). One may
discern SEF50, which is the median frequency and
SEF90, SEF95, SEF98. These parameters have been
applied to the whole spectrum (Laman et al., 2005),
but also to certain bands (e.g., alpha band) (Minicucci et al., 2000) (Fig. 7).
These sets of parameters have been used in two
different ways: (1) to compare right (R) to left (L),
generally (R L)/(R L) (Van Putten et al.,
2004) (Fig. 7) and (2) to compare baseline (B) to
event (clamping, C), C B or (C B)/B (Minicucci
et al., 2000; Laman et al., 2001).
The unprocessed EEG should always be available for

simultaneous reviewing throughout EEG monitoring,
although for sake of simplicity, a selection of channels may be made. This enables detection of artifacts
(e.g., due to surgical diathermy, ECG, or movement)
interfering with the EEG signal to be processed and
contributing to erroneous interpretation when only
the processed EEG would be assessed.
Data compression enabling an overview may be
obtained in several ways. Huge compression of the
time scale is used in the CFM: 6 cm contains the
EEG amplitude information over 1 h (Prior and Maynard, 1986) (Fig. 7). EEG frequency spectra averaged
over periods of 0.5 min may be represented as compressed spectral array allowing the detection of gradual trends. The subsequent spectra are plotted one
above the other. The composite of many spectra is
made simpler by hidden line suppression: those parts
of the spectra which fall behind preceding peaks are
not written.
In the DSA, gray scales or color scales are used
to identify the amplitude ranges of the spectrum.
This method allows a considerable time compression
and no spectral information is hidden. The
amplitude resolution, however, is limited by the

number of gray or color scale steps that can be produced (Fig. 7).
Median frequency is identical to SEF50 or center
of gravity of the spectrum. For each epoch, the
median frequency may be plotted as a function of
time. Spectral edge frequencies (e.g., SEF90 is the
90th percentile of the frequency distribution) may
be presented equally as line diagrams (Fig. 7).
To obtain power band values, the energy within
each band is summated. These data may be presented
as absolute or relative (relative to the total spectral
power) data. This information may be presented with
lines; mostly, the area under the curve is shaded or
colored. Also, a presentation with histograms or a
numerical presentation may be provided. In many
commercially available equipments, combinations
of all possible presentations can be chosen.
Numbers may also be presented for all these
parameters. Some commercially available equipment
also allows the setting of threshold values, although
evidence for the application of such thresholds is still
lacking. Transgression of such threshold may result
in visual or acoustic alarm signs.
Acknowledgments
The physicists M. Hersevoort, G.J. Huiskamp,
M.J.A.M. Van Putten, and C. Rizzo performed the
EEG data processing resulting in Fig. 8.
References
Akiyama, T, Kobayashi, K, Nakahori, T, Yoshinaga, H,
Ogino, T, Ohtsuka, Y, Takeuchi, M, Morita, K, Sano, S
and Oka, E (2001) Electroencephalographic changes and
their regional differences during pediatric cardiovascular
surgery with hypothermia. Brain Dev., 23(2): 115121.
Blume, WT and Sharbrough, FW (2005) EEG monitoring
during carotid endarterectomy and open heart surgery.
In: E Niedermeijer and F Lopes Da Silva (Eds.),
Electroencephalography. Basic Principles, Clinical
Applications, and Related Fields. Lippincott, Williams
and Wilkins, Philadelphia, pp. 815827.
Blume, WT, Ferguson, GG and McNeill, DK (1986) Significance of EEG changes at carotid endarterectomy.
Stroke, 17: 891897.
Burke, D, Nuwer, MR, Daube, J, Fischer, C, Schramm, J, Yingling, CD and Jones, SJ (1999) Intraoperative monitoring.
In: G Deuschl and A Eisen (Eds.), Recommendations for
the Practice of Clinical Neurophysiology: Guidelines of
the International Federation of Clinical Neurophysiology.
139
Electroenceph. Clin. Neurophysiol., Suppl. 52. Elsevier

Amsterdam, pp. 133148.
Chiappa, KH, Burke, SR and Young, RR (1979) Results of
electroencephalographic monitoring during 367 carotid
endarterectomies. Use of a dedicated minicomputer.
Stroke, 10(4): 381388.
Cucchiara, RF, Sharbrough, FW, Messick, JM and Tinker,
JH (1979) An electroencephalographic filter-processor
as an indicator of cerebral ischemia during carotid endarterectomy. Anesthesiology, 51: 7779.
Cursi, M, Meraviglia, MV, Fanelli, GF, Chiesa, R, Tirelli,
A, Comi, G and Minicucci, F (2005) Electroencephalographic background desynchronization during cerebral
blood flow reduction. Clin. Neurophysiol., 116(11):
25772585.
Debatisse, D, Pralong, E, Dehdashti, AR and Regli, L
(2005) Simultaneous multilobar electrocorticography
(mEcoG) and scalp electroencephalography (scalp
EEG) during intracranial vascular surgery: a new
approach in neuromonitoring. Clin. Neurophysiol., 116
(12): 27342740.
Florence, G, Guerit, JM and Gueguen, B (2004) Electroencephalography (EEG) and somatosensory evoked potentials (SEP) to prevent cerebral ischaemia in the
operating room. Neurophysiol. Clin., 34(1): 1732.
Guerit, JM (1998) Neuromonitoring in the operating room:
why, when, and how to monitor? Electroencephalogr.
Jonkman, EJ, Van Huffelen, AC and Pfurtscheller, G
(1986) Quantitative EEG in cerebral ischemia. FH
Lopes Da Silva and W Storm Van Leeuwen (Eds.),
Clinical Applications of Computer Analysis EEG and
other Neurophysiological Signals. Elsevier, Amsterdam,
pp. 205237.
Kearse, LA, Jr., Martin, D, McPeck, K and
Lopez-Bresnahan, M (1993) Computer-derived density
spectral array in detection of mild analog electroencephalographic ischemic pattern changes during carotid endarterectomy. J. Neurosurg., 78(6): 884890.
Kraaier, V, Van Huffelen, AC and Wieneke, GH (1988)
Quantitative EEG changes due to hypobaric hypoxia
in normal subjects. Electroencephalogr. Clin. Neurophysiol., 69: 303312.
Krul, JMJ, Ackerstaff, RGA, Eikelboom, BC and
Vermeulen, FEE (1989) Stroke-related EEG changes
during carotid surgery. Eur. J. Vasc. Surg., 3: 423428.
Laman, DM (2004) Perioperative monitoring of brain function and cerebral circulation in carotid endarterectomy.
Thesis, University of Utrecht, 134 pp.
Laman, DM, Van Der Reijden, CS, Wieneke, GH, Van
Duijn, H and Van Huffelen, AC (2001) EEG evidence
for shunt requirement during carotid endarterectomy.
Optimal EEG derivations with respect to frequency
bands and anesthetic regimen. J. Clin. Neurophysiol.,
18(4): 353363.
140
Laman, DM, Wieneke, GH, Van Duijn, H, Veldhuizen, RJ
and Van Huffelen, AC (2005) QEEG changes during
carotid clamping in carotid endarterectomy: spectral
edge frequency parameters and relative band power
parameters. J. Clin. Neurophysiol., 22: 244252.
Myers, RR, Stockard, JJ and Saidman, LJ (1977) Monitoring of
cerebral perfusion during anesthesia by time-compressed
Fourier analysis of the electroencephalogram. Stroke, 8:
331337.
Minicucci, F, Cursi, M, Fornara, C, Rizzo, C, Chiesa, R, Tirelli, A, Fanelli, G, Meraviglia, MV, Giacomotti, L and
Comi, G (2000) Computer-assisted EEG monitoring during carotid endarterectomy. J. Clin. Neurophysiol., 17(1):
101107.
Mizrahi, EM, Patel, VM, Crawford, ES, Coselli, JS and
Hess, KR (1989) Hypothermic-induced electrocerebral
silence, prolonged circulatory arrest, and cerebral protection during cardiovascular surgery. Electroencephalogr. Clin. Neurophysiol., 72: 8185.
Modica, PA, Tempelhoff, R, Rich, KM and Grubb, RL, Jr.
(1992) Computerized electroencephalographic monitoring and selective shunting: influence on intraoperative
administration of phenylephrine and myocardial infarction after general anesthesia for carotid endarterectomy.
Neurosurgery, 30: 842846.
Nuwer, MR, Daube, J, Fischer, C, Schramm, J and Yingling, CD (1993) Neuromonitoring during surgery.
Report of an IFCN committee. Electroencephalor. Clin.
Pinkerton, JA, Jr. (2002) EEG as a criterion for shunt need in
carotid endarterectomy. Ann. Vasc. Surg., 16: 756761.
A.C. VAN HUFFELEN

Prior, PF and Maynard, DE (1986) Monitoring Cerebral
Function. Long-term Monitoring of EEG and Evoked
Potentials, Elsevier Science Publishers, Amsterdam,
449 pp.
Prior, PF, Maynard, DE, Sheaff, PC, Simpson, BR, Strunin,
L, Weaver, EJM and Scott, DF (1971) Monitoring cerebral function: clinical experience with new device for continuous recording of electrical activity of brain. Br. Med.
J., 2: 736738.
Pronk, RAF and Simons, AJR (1982) Automatic recognition of abnormal EEG activity during open heart and
carotid surgery. In: PA Buser, WA Cobb and T Okuma
(Eds.), Kyoto Symposia. Electroenceph. Clin. Neurophysiol., Suppl. 36. Elsevier, Amsterdam, pp. 590602.
Van Putten, MJAM, Peters, JM, Mulder, SM, De Haas, JAM,
Bruijninckx, CMA and Tavy, DLJ (2004) A brain symmetry index (BSI) for online EEG monitoring in carotid
endarterectomy. Clin. Neurophysiol., 115(5): 11891194.
Visser, GH, Wieneke, GH and Van Huffelen, AC (1999)
Carotid endarterectomy monitoring: patterns of spectral
EEG changes due to carotid artery clamping. Electroencephalogr. Clin. Neurophysiol., 110: 286294.
Visser, GH, Wieneke, GH, Van Huffelen, AC, De Vries,
JW and Bakker, PFA (2001) The development of spectral EEG changes during short periods of circulatory
arrest. J. Clin. Neurophysiol., 18(2): 169177.
Vriens, EM, Wieneke, GH, Hillen, B, Eikelboom, BC, Van
Huffelen, AC and Visser, GH (2001) Flow redistribution
in the major cerebral arteries after carotid endarterectomy: a study with transcranial Doppler scan. J. Vasc.
Surg., 33: 139147.

M.R. Nuwer (Ed.)
141
CHAPTER 7
Electrocorticography
Gregory A. Worrell*, Matthew Stead and Gregory D. Cascino
Department of Neurology, Division of Epilepsy and Electroencephalography, Mayo Clinic, Rochester, MN 55905, USA
7.1. Introduction
This chapter describes intraoperative electroencephalography (EEG) recorded directly from human brain,
that is, what is commonly called intraoperative electrocorticography (ECoG). To provide some context
for the clinical use of EcoG, this chapter will briefly
discuss the most common applications of intraoperative ECoG, but will not attempt to critically evaluate
the specific applications, which are addressed in separate chapters. We will describe the technical aspects
of modern ECoG recordings in humans, and will
include some possible pitfalls and limitations of
ECoG. The material presented here largely reflects
the current practice of ECoG, and is not intended to
cover all techniques and protocols.
The history of human EEG dates back to Hans Berger
(Berger, 1935) who obtained the first human recording
in a patient with a skull defect (Ajmone Marsan,
1998). The modern practice of ECoG descends directly
from the seminal work of Penfield and Jasper (Penfield
and Jasper, 1954), which was consistent with the current
use of the term ECoG today describing EEG recordings
obtained directly from brain.
7.2. The physiologic range of human EEG
The EEG recorded from human brain shows neuronal
oscillations extending well beyond what was first
reported by Berger (1935), and beyond what is commonly recorded in clinical practice (0.570 Hz) (Daly
and Pedley, 1990; Niedermeyer and Lopes da Silva,
2005; Quesney and Niedermeyer, 2005) (Fig. 1).
Correspondence to: Dr. Gregory A. Worrell, Department

of Neurology, Division of Epilepsy and EEG, Mayo
Clinic, 200 First St. SW, Rochester, MN 55905, USA.
Tel.: 1-507-284-1588; fax: 1-507-284-4795.
E-mail: worrell.gregory@mayo.edu (G.A. Worrell).
There is an increasing awareness of the physiologic,

and possibly clinically relevant, bandwidth of human
intracranial EEG. The presence of low-frequency
oscillations extending from 0.5 Hz down to 0.01 Hz,
as well as spontaneous direct current (DC) voltage fluctuations, in human EEG recordings are well established
and detectable from both scalp and intracranial EEG
recordings (Casper, 1993; Vanhatalo et al., 2005). Similarly the high-frequency end of the EEG spectrum,
40 Hz up to 600 Hz, is an active area of research and
clinical interest. In contrast to low-frequency oscillations, by virtue of the low amplitude of high-frequency
activity, the study of high-frequency EEG is almost
exclusively from intracranial recordings. Recent studies
using chronic intracranial EEG (i EEG) recordings
report gamma oscillations (4080 Hz) that may
play a role in learning and memory, ripple oscillations
(100200 Hz) that may be important for memory
consolidation, as well as somatosensory evoked potential oscillations (600 Hz). In addition to their role in
normal brain function, high-frequency oscillations
(>60 Hz) have been described at seizure onset
(Fig. 2A and 2B) (Allen et al., 1992; Fisher et al.,
1992; Alarcon et al., 1995; Bragin et al., 1999a; Bragin
et al., 2002; Grenier et al., 2003a,b; Worrell et al.,
2004; Jirsch et al., 2006), and more recently, interictal
high-frequency epileptiform oscillations (Fig. 3) have
been described in human epileptogenic brain at times
temporally remote from seizure onset (Bragin et al.,
1999b; Traub et al., 2001; Worrell et al., 2004). There
is now accumulating evidence that high-frequency
oscillations may have a fundamental role in the generation of focal seizures (Traub et al., 2001; Bragin et al.,
2002; Grenier et al., 2003a; Worrell et al., 2004) and
may be a useful electrophysiological signature of
epileptogenic brain.
Nonetheless, the majority of human EEG recordings from intracranial subdural and/or depth electrodes
report a limited dynamic range (0.570 Hz). The
clinical focus on such a limited frequency bandwidth
142
G.A. WORRELL ET AL.
Physiologic Range of Human Brain Activity
0.5
103
Fast
Ripples
Ripples
Berger Bands
ULF
Gamma
#15
Single
Units
30Hz
104
Frequency (Hz)
Fig. 1. The physiologic range of human brain activity. The timescales of electrophysiologic activity from human
cortex spans over 67 orders of magnitude. From DC potentials, the Berger bands (delta (14 Hz), theta (>4 to <8 Hz),
alpha (813 Hz), beta (>1330 Hz)), gamma oscillations (>3080 Hz), ripple oscillations (>80200 Hz), fast ripples
(250700 Hz), to unit and multiunit activity. Upper right: X-ray after implantation of an 8 8 grid over the convexity
and a single eight-contact strip (four mastoid reference electrodes are also apparent). Middle: Digital photograph of
implanted grid. The cortex is visible throughout the transparent grid substrate. Upper left: Schematic of a subdural grid.
A dipole generator located in the crown of a gyrus oriented such that it produced a large local field at the subdural electrodes (light gray lines in cortex). The dipole generator in a sulcus oriented parallel to the grid and producing a diffuse weak
field at the grid (dark gray). The field lines that continuously extend beyond the brain have been omitted in this drawing.
may partly reflect the fact that when using standard

clinical reviewing parameters (e.g., 10-s page and
50 mV/mm) oscillations with frequencies above
40 Hz are low amplitude and largely obscured by
higher amplitude cerebral activity in the classic Berger
bands (0.530 Hz) (Niedermeyer and Lopes da Silva,
2005). Similarly, very slow oscillations are not apparent with common EEG reviewing parameters because
the period of oscillation is longer than the typical 10-s
viewing page. The clinical usefulness of recording
from the entire physiologic bandwidth is not defined
at this time, but is an area of active investigation.
7.3. Applications and limitations of ECoG
The primary role of ECoG in most centers is currently as part of epilepsy surgery and stimulation
mapping to localize eloquent cortex. The practice
at many centers is for patients with intractable and
poorly localized partial epilepsy to undergo long-term

intracranial monitoring in an effort to localize the
region of seizure onset. In these patients, the role
of intraoperative ECoG is generally limited. However, there is a role for ECoG recordings at the time
of subdural and depth electrode placement. Often
the placement of the recording electrodes is performed using stereotactic imaging techniques to
guide the placement electrodes to the hypothesized
region of epileptogenic brain (Fig. 4). The ECoG
recording at the time of electrode implantation
provides assurance that the electrodes are recording
at the time of surgery, and will often show marked
epileptiform abnormalities. Occasionally, the subdural
electrodes may be repositioned, or additional electrodes added, because of the presence of epileptiform
activity at the margin of the recording electrodes.
In patients undergoing epilepsy surgery that do
not require long-term intracranial monitoring, that
143
Fig. 2. A: Referential montage. Clinical ECoG recorded with 1.0-Hz high-pass filter and 70-Hz low-pass filter and digitized
at 250 Hz. A focal seizure demonstrating a characteristic attenuation or decrement at onset, and as the seizure spreads
spatially there is a characteristic increase in the amplitude of rhythmic activity and a decrease in the frequency. B: Referential montage. DC-1,000 Hz bandwidth using a 5,000 Hz digitization rate. The attenuation at seizure onset is demonstrated
to be a high-frequency oscillation (150 Hz) that replaces the interictal background.
144
G.A. WORRELL ET AL.
Fig. 3. Bipolar montage. Interictal ECoG recorded from a frontal convexity grid (same patient as Fig. 1). A focal highfrequency epileptiform oscillation is indicated with the upper dashed line. At an adjacent location, an epileptiform sharp
wave is indicated with the lower dashed line.
Intraoperative Stereotactic Imaging

Subtraction Ictal SPECT Co-registered to MRI
Ictal SPECT
Interictal
Subtraction
MRI, SPECT Focus & CT Electrodes

Fig. 4. Intraoperative stereotactic imaging. The top row of images demonstrates construction of subtraction ictal SPECT
co-registered to MRI (SISCOM). The lower row of images illustrates intraoperative stereotactic placement of a grid over
a SISCOM abnormality. The last image demonstrates the use of high-resolution digital photography to create a
co-registered map of subdural electrodes onto cortex. The use of digital photographs for surgical planning is often useful.
is, the epileptogenic zone has been adequately localized from non-invasive methods, ECoG can be used
at the time of resective surgery. However, exactly
how intraoperative ECoG is used in these patients
varies between institutions, and remains an area of
considerable controversy.
It is important to realize the significant limitations generally associated with ECoG. The sulcated
structure of human brain leaves 2/3 of the cortical
mantle inaccessible to subdural electrodes (Fig. 1).
Subdural electrodes are only in direct contact with
the crowns of gyri, and primarily record local activity. Cortical generators that are deep within the sulci
are difficult to access. In addition to being farther
away from the subdural electrodes placed over the
cortex, a generator within a cortical sulcus produces
a dipole and potential field that is likely oriented parallel to the subdural surface, thus yielding a much
smaller contribution to the local potential field
(Fig. 1). In some cases, it is possible to record from
the gray matter within a sulcus by careful placement
of depth or strip electrodes.
7.4. ECoG recording methods
Some of the earliest ECoG recordings were obtained
using cotton wick electrodes and recorded using a
16-channel paper strip chart electroencephalograph.
The common filter settings and a 3050 mV/mm sensitivity provided ECoG recordings from 1 to 40 Hz.
Cotton wick electrodes and paper strip chart EEG has
been replaced by use of metallic electrodes and digital EEG acquisition systems. The move to digital
acquisition of EEG does not change the basic principles of recording EEG, but has yielded significant
improvement in the bandwidth that is routinely recorded, and opened up the entire field of quantitative
EEG analysis (Niedermeyer and Lopes da Silva,
2005). The basics of EEG recording, modern electronics, biomedical safety, and the technologic advances
associated with digital EEG are well described elsewhere (A. Kamp, Gert Pfurtscheller, G. Edlinger and
F. Lopes da Silva, Technological Basis of EEG
Recording, 5th Edition; Daube, 1996) and will not be
repeated in detail here.
7.4.1. Safety
All equipment used for recording ECoG must meet
biomedical safety requirements for medical equipments, and the equipment should be periodically
145
tested to assure compliance. The safety requirements

are especially important for ECoG given that the
recording electrodes are in direct contact with the
brain and the patient is commonly connected to multiple medical machines. Whenever patients are
connected to multiple medical instruments, the possibility of ground loops and leakage currents to flow
between instruments and to the patient must be prevented. To prevent putting the patient at risk, the
ECoG acquisition system must be isolated from other
electrical equipment and have isolation or current
limiting circuits. The patient should never be allowed
to be directly grounded to earth ground, or to come
into contact with grounded objects (Daube, 1996).
7.4.2. Electrodes
Modern ECoG recordings are generally obtained
using subdural and depth metallic electrodes, most
commonly platinum, platinum iridium alloys, or
stainless steel electrodes that are implanted under
general anesthesia. The platinumiridium electrodes
are inert and desirable for stimulation studies, and
are MRI compatible. The subdural electrodes are
commonly embedded in a thin flexible silastic substrate that will deform to the contour of the brain
and allow visual inspection of the underlying cortex.
Additionally, the visually transparent substrate makes
possible construction of high-fidelity co-registration
maps (Fig. 4) (Wellmer et al., 2002).
The electrode impedance describes the frequencydependent relation between voltage and current, and
is primarily determined by the electrode material
and the electrodebrain interface. When using most
subdural or depth electrodes, commonly with 1
10 mm2 surface area, the impedance is typically of
order 1 kO. The impedance is very sensitive to the
integrity of the contact between the electrode and
brain. The most common cause of unsatisfactorily
high impedance is that the electrode is not making
good contact with the brain.
7.4.3. Co-registration imaging and stereotactic
electrode placement
Given that the depth and subdural electrodes record
primarily local cerebral activity, the placement must
be accurate. The development of advanced imaging
methods has made routine the fusion of functional (e.g., single photon computed tomography
(SPECT) and positron emission tomography (PET))
146
and structural (e.g., magnetic resonance imaging

(MRI)) imaging scans that can be used with both
framed and frameless stereotactic approaches to place
electrodes (Fig. 4) (So, 2000). These techniques allow
accurate placement of electrodes into areas of functional or structural abnormality.
Grid and strip electrodes are placed through a craniotomy by direct visualization and stereotactic imaging based on the area of brain region of interest. Strip
electrodes can placed through cranial burr holes and
slipped into the subdural space. However, given the
strip is advanced through the burr hole without visual
guidance, on occasion the postoperative CT scan may
show that the strip is off target. Depth electrodes are
most commonly placed using stereotactic guidance
systems. The structures routinely recorded from
include temporal, frontal, parietal and occipital neocortex, interhemispheric, subfrontal, and inferior mesial
temporal sites using subdural strips and grid electrodes
(Fig. 1). To record from deeper mesial structures (e.g.,
amygdala, hippocampus, medial frontal lobe), depth
electrodes can be used and are commonly placed using
stereotactic imaging (Fig. 4). To record from the deep
mesial temporal structures, multiple depth electrodes
can be placed via temporal burr holes or a depth electrode placed posteriorly through the long axis of the hippocampus. Alternatively, depth electrodes can be
placed through a larger craniomotomy.
Unfortunately, it remains challenging to record
directly from cortex lying within sulci. In some situations, it is possible to place a depth electrode within
gray matter lying within a sulcus. However, the
amount of gray matter sampled is limited due to the
thickness of the cortical mantle, generally 35 mm.
7.4.4. Digital EEG acquisition system
To obtain the ECoG activity recorded from human
brain within the commonly used clinical 0.570-Hz
bandwidth, a sampling rate of at least 250 Hz
(45 times the low-pass cutoff frequency) is preferred. With the high-pass filters (passing frequency
above the filter number) set at 0.5 Hz, the difficulties
associated with DC drift and amplifier saturation
(blocking) from large DC offsets is avoided. Lowpass filters are commonly set at 70100 Hz. It is
common to use a notch filter at 60 Hz (50 Hz in Europe)
to filter line noise. It must be realized that filters
alter the observed waveforms recorded. For example,
reducing the low-pass filter from 70 to 30 Hz will
G.A. WORRELL ET AL.
visibly alter the sharpness of spikes. It is always a good

idea to view the data without the digital filters on to be
sure that artifacts and cerebral activity are not
misinterpreted.
It is useful to have ECG recording as part of the
ECoG recording to help differentiate ECG and pulse
artifacts.
7.4.5. Reference and ground electrodes
There are no truly inactive references for EEG, and
we routinely use cranial ground and reference electrode sites, and less commonly have employed intracranial epidural reference electrodes. The choice of
an extracranial reference, such as mastoid, provides
a quiet reference given that voltage of ECoG activity
is 45 times the amplitude of scalp-recorded activity.
Alternative reference choices include average reference from the intracranial electrodes or epidural strip
electrodes, although this is not routinely available
in hardware form and so, when used, is constructed
from the composite digital signals, which themselves
were recorded against a single fixed reference. Our
current practice is to use either mastoid ground and reference electrodes, or suture wires placed within scalp
(vertex) if the patient is to subsequently undergo
chronic intracranial monitoring to record their habitual
seizures. The scalp reference can introduce additional
artifact such as movement and muscle.
It is often useful to have additional ground and
reference electrodes in place to allow for rapid troubleshooting. For example, an additional 23 mastoid
electrodes (Fig. 1) placed prior to ECoG can be very
useful when troubleshooting a flawed intraoperative
ECoG recording, allowing for quick changes in
reference and ground electrodes.
7.4.6. Acquisition montage and reviewing montage
The ECoG recording is acquired using a specified
ground and reference electrode discussed above.
Modern digital acquisition systems make it is easy
to view the ECoG in different montages. Similar to
scalp EEG, it is often useful to view the ECoG in a
bipolar montage to remove more distributed activity
and to help localize focal activity. However, with
the typical depth and subdural electrode spacing
(0.51.0 cm electrode spacing) and an extracranial
reference, there is often good spatial localization with
a referential recording.
7.4.7. ECoG activity

Physiologic activity recorded from subdural and
depth electrodes can appear sharply contoured compared to scalp-recorded EEG. This has less to do with
any filtering effect of the dura/skull/scalp, but primarily related to the proximity of the electrode to
the cerebral generators. The morphology of epileptiform spikes and sharp waves are similar to scalp activity. The ECoG epileptiform spikes are more sharply
contoured, and many are not apparent on concurrent
scalp-recorded EEG (Tao et al., 2005). Seizures
recorded from depth and subdural electrodes can be
significantly different from those recorded from the
scalp. The commonly described attenuation or decrement at seizure onset is often related to a highfrequency discharge at seizure onset (Fig. 2), which
is of too low an amplitude and high a frequency to be
detected in most scalp EEG recordings.
147
to the connectors at the clinical tails or at the headbox. A commonly encountered difficulty in the OR
is related to a high-impedance electrode, which is usually from poor contact between the electrode and cortex. This can occur when the electrode is lying over
a sulcus, blood vessel, or when the substrate of the
subdural grid buckles lifting the electrodes off the
surface.
7.5.2. Electrode artifact
Even when all channels are recording, it is common
to see artifacts associated with the OR environment.
Sometimes it is useful to ask for a pause in the activity of the OR team, such that all movement, cautery,
etc. can be eliminated. With this procedure, noncerebral artifacts such as ECG and pulse artifact
can be investigated. In the OR, machine artifacts
are common as well, for example the ventilator.
7.4.8. Anesthesia effect
7.5.3. Complications
General anesthesia is associated with increased fast

activity, >30 Hz, and intermixed slow wave activity.
Anesthesia can markedly reduce or eliminate epileptiform activity. In some centers, pharmacologic
activation of epileptiform activity is utilized. Anesthesia is discontinued and an activating agent is administered, such as Brevital, fentanyl, and alfentanil
(Cascino, 1998). The clinical utility of this procedure
remains controversial.
The primary complications and risks associated with

ECoG recording are those that accompany any craniotomy (Korinek et al., 2005) and the placement of
depth, strip, and grid electrodes. The placement
of depth electrodes always carries the risk of hemorrhage, but in skilled hands this risk is low, 1% (Van
Buren, 1987). Conventional or MR cerebral angiography is used to identify vascular structures in some
centers to help plan placement of depth electrodes
when using burr holes. Extended periods of ECoG
recording presumably slightly increase the risk of
infection, but overall the infection rate likely remains
low. For chronic intracranial monitoring, complication risks are considerably higher with one retrospective study showing that 19/99 patients (19%) had
complications, including 2 patients (2%) with permanent sequelae (Wiggins et al., 1999; Hamer et al.,
2002). As is often the case, the practice varies
between centers, and there is currently a lack of data
to identify best practice.
Rarely during ECoG, but not uncommonly, when
electrical stimulation mapping is performed, electrographic and clinical seizures can occur. If the patient
is awake and undergoing ECoG as part of stimulation
mapping and resective surgery, a clinical seizure is dangerous and should be aggressively aborted. Perhaps
the best approach is to directly apply 50 cm3 of
cool saline over the seizing cortex. In most cases,
7.5. Troubleshooting common problems

The majority of the technical difficulties encountered
on a daily basis in the OR are easily solved. However, the OR is a very challenging environment with
multiple sources of electrical noise, and can be intimidating for troubleshooting because of the rigid
time pressures surrounding surgery and anesthesia.
7.5.1. Channels not recording
If all channels are not recording, this likely is a problem with the connection to the acquisition system,
headbox, connectors to the clinical electrode tails,
or possibly a problem with the reference electrode.
If, however, all but a few channels are recording,
then it is unlikely to be related to the reference electrode or acquisition systems, but could still be related
148
this will abruptly abort electrographic seizure activity.

The attending anesthesiologist should be aware that
stimulation studies are underway and have lorazepam
and fos-phenytoin, or an alternative for patients allergic
to phenytoin products.
7.6. Conclusions
The clinical role of ECoG has appropriately declined
with the rapid improvement in non-invasive functional (SPECT and PET) and structural (MRI) imaging modalities. Additionally, the use of chronic video
and intracranial monitoring to record a patients
habitual seizures when the non-invasive studies are
not sufficient to localize the region of epileptogenic
brain has diminished the role for ECoG. Nonetheless,
there remains a role for ECoG in epilepsy surgery
and as part of functional mapping. The future will
determine if current research will develop new roles
for ECoG, or if the continued advances in noninvasive functional imaging methods will ultimately
render ECoG obsolete.
References
Ajmone Marsan, C (1998) Electrocorticography. Historical
comments on its development and the evolution of its
practice. In: Electrocorticography: Current Trends and
Future Perspectives. Elsevier, Amsterdam.
Alarcon, G, Binnie, CD, et al. (1995) Power spectrum and
intracranial EEG patterns at seizure onset in partial epilepsy. Electroencephalogr. Clin. Neurophysiol. 94(5):
326337.
Allen, PJ, Fish, DR, et al. (1992) Very high-frequency
rhythmic activity during SEEG suppression in frontal
lobe epilepsy. Electroencephalogr. Clin. Neurophysiol.,
82(2): 155159.
Berger, H (1935) On the electroencephalogram of man. Tenth
report. In: P. Gloor (Ed.), Hans Berger on the Electroencephalogram of Man. Suppl. 28. Elsevier, Amsterdam.
Bragin, A, Engel, J, Jr., et al. (1999a) High-frequency oscillations in human brain. Hippocampus, 9(2): 137142.
Bragin, A, Engel, J, Jr., et al. (1999b) Hippocampal
and entorhinal cortex high-frequency oscillations (100
500 Hz) in human epileptic brain and in kainic acid
treated rats with chronic seizures. Epilepsia, 40(2):
127137.
Bragin, A, Mody, I, et al. (2002) Local generation of fast ripples in epileptic brain. J. Neurosci., 22(5): 20122021.
Cascino, G (1998) Pharmacological activation. In: Electrocorticogrpahy: Current Trends and Future Perspectives. Elsevier, Amsterdam.
G.A. WORRELL ET AL.

Casper, H (1993) DC potentials of the brain. In: W
Haschke, EJ Speckmann and AI Roitbak (Eds.), Slow
Potential Changes in the Brain. Birkhauser, Boston.
Daly, DD and Pedley, TA (1990) Current Practice
of Clinical Electroencephalography. Lippincott-Raven,
Philadelphia.
Daube, J (Ed.) (1996) Clinical Neurophysiology. Contemporary Neurology Series. F.A. Davis. Philadelphia.
Fisher, RS, Webber, WR, et al. (1992) High-frequency
EEG activity at the start of seizures. J. Clin. Neurophysiol., 9(3): 441448.
Grenier, F, Timofeev, I, et al. (2003a) Spontaneous field
potentials influence the activity of neocortical neurons
during paroxysmal activities in vivo. Neuroscience,
119(1): 277291.
Grenier, F, Timofeev, I, et al. (2003b) Neocortical very fast
oscillations (ripples, 80200 Hz) during seizures: intracellular correlates. J. Neurophysiol., 89(2): 841852.
Hamer, HM, Morris, HH, et al. (2002) Complications of
invasive video-EEG monitoring with subdural grid electrodes. Neurology, 58(1): 97103.
Jirsch, JD, Urrestarazu, E, et al. (2006) High-frequency
oscillations during human focal seizures. Brain, 129(6):
15931608.
Korinek, AM, Golmard, JL, et al. (2005) Risk factors for
neurosurgical site infections after craniotomy: a prospective multicenter study of 2944 patients. The French
Study Group of Neurosurgical Infections, the SEHP,
and the C-CLIN Paris-Nord. Service Epidemiologie
Hygie`ne et Prevention. Br. J. Neurosurg., 19(2):
155162.
Niedermeyer, E and Lopes da Silva, FL (2005) Electroencephalography: Basic Principals, Clinical Applications,
and Related Fields. Lippincott and Wilkins,
Philadelphia.
Penfield, W and Jasper, HH (1954) Epilepsy and the Functional Anatomy of the Human Brain. Little Brown &
Co., Boston.
Quesney, LF and Niedermeyer, E (2005) Electrocorticography. In: E Niedermeyer and F Lopes da Silva (Eds.),
Electroencephalography: Basic Principals, Clinical
Applications, and Related Fields. Lippincott and Wilkins, Philadelphia.
So, EL (2000) Integration of EEG, MRI, and SPECT in
localizing the seizure focus for epilepsy surgery. Epilepsia, 41(Suppl. 3): S48S54.
Tao, JX, Ray, A, et al. (2005) Intracranial EEG substrates of
scalp EEG interictal spikes. Epilepsia, 46(5): 669676.
Traub, RD, Whittington, MA, et al. (2001) A possible role
for gap junctions in generation of very fast EEG oscillations preceding the onset of, and perhaps initiating, seizures. Epilepsia, 42(2): 153170.
Van Buren, JM (1987) Complications of surgical procedures in the diagnosis and treatment of epilepsy.

In: JJ Engel (Ed.), Surgical Treatment of the Epilepsies. Raven Press, New York, pp. 465475.
Vanhatalo, S, Voipio, J, et al. (2005) Infraslow EEG
activity. In: E Niedermeyer and F Lopes da Silva
(Eds.), Electroencephalography: Basic Principals,
Clinical Applications, and Related Fields. Lippincott
and Wilkins, Philadelphia, pp. 489493.
Wellmer, J, Von Oertzen, J, et al. (2002) Digital photography
and 3D MRI-based multimodal imaging for individualized
149
planning of resective neocortical epilepsy surgery. Epilepsia, 43(12): 15431550.

Wiggins, GC, Elisevich, K, et al. (1999) Morbidity and infection in combined subdural grid and strip electrode investigation for intractable epilepsy. Epilepsy Res., 37(1):
7380.
Worrell, GA, Parish, L, et al. (2004) High-frequency oscillations and seizure generation in neocortical epilepsy.
Brain, 127(Pt 7): 14961506.

M.R. Nuwer (Ed.)
150
CHAPTER 8
Direct cortical stimulation to localize sensory,

motor and language function
Lara M. Schrader*
Department of Neurology, David Geffen School of Medicine, UCLA, Los Angeles, CA 90403, USA
8.1. Introduction
The goal of brain surgery is to remove as much pathological tissue as possible, yet preserve brain function.
The term eloquent cortex refers to areas of the cerebral cortex that, if removed, could result in significant
neurological functional deficit. Brain surgery in or near
eloquent cortex requires understanding of the anatomic
localization of eloquent brain function in these regions
(i.e., brain mapping) prior to resection of brain tissue.
While a number of techniques for brain mapping exist,
none have replaced the gold standard for brain mapping, direct cortical electrical stimulation. Brain mapping
with direct cortical electrical stimulation essentially
involves assessing whether electrical stimulation to a
small region of cortical surface results in a neurological
change in the patient, such as a movement or speech
arrest. If the cortical area stimulated produces a neurological response, then that brain region is considered
important for that particular function.
While there are general broad brain regions
known to be responsible for specific brain functions
that are common to all normal individuals (e.g., dominant hemisphere perisylvian regions are important
for language function), in an individual patient, the
function is actually localized to a more specific and
well-circumscribed area within that broader anatomical region (Ojemann, 1979; Uematsu et al., 1992).
There is considerable interindividual variability in
the specific anatomic locations of these smaller more
discrete regions of eloquent cortex (Ojemann, 1979;
Uematsu et al., 1992; Branco et al., 2003), making
it important to understand the specific anatomical
localization of eloquent function for each individual
patient. Without an understanding of the location of
*
Correspondence to: Lara M. Schrader, M.D., 20 Old Stone
Crossing, West Hartford, CT 06117, USA.
Tel.: +1-310-804-6673; fax: +1-310-267-1157.
E-mail: laraschrader@hotmail.com (L.M. Schrader).
specific regions of brain function in the individual

patient, the margins of a surgical resection would
be unnecessarily limited by the anatomic landmarks
that define the broad regions that are responsible
for brain function in the general normal population.
Accurate brain mapping allows neurosurgery to be
tailored to the individual patient and allow the maximal resection of pathological tissue while preserving
function (Duffau et al., 1999, 2005). Furthermore,
structural lesions such as brain tumors can markedly
distort normal anatomical landmarks that can serve
as guides for typical functional localizations, resulting in further necessity for brain mapping techniques.
Recently, perfusion-dependent brain mapping techniques, such as functional magnetic resonance imaging
(MRI), positron emission tomography, and optical
imaging of intrinsic signals, have become increasingly
used for neurosurgical guidance (Pouratian et al.,
2003). Some of the advantages of these mapping techniques are that they are noninvasive and can be done
preoperatively. The data from perfusion-dependent
functional mapping techniques can be incorporated
into intraoperative stereotactic MRI-guided systems
(known as functional neuronavigation), and this
combined approach can be very useful in guiding the
surgeon toward the best operative approach. However,
perfusion-dependent mapping techniques have limitations with regard to spatial specificity, sensitivity,
and reliability (Herholz et al., 1997; Fandino et al.,
1999; Pouratian et al., 2003; Roux et al., 2003).
In addition, after removal of tissue, shifting of brain
contents within the cranial vault can occur, rendering
the stereotactic MRI guidance systems less exacting
in their localizing abilities (Duffau, 2000). Recently,
functional neuronavigation techniques have been used
in combination with intraoperative direct stimulation
mapping, and this dual approach has been very helpful
in guiding the neurosurgical approach (Duffau, 2000;
Kamada et al., 2005).
The reliability of electrocortical stimulation mapping for predicting postoperative functional outcomes
depends on whether the stimulation used is sufficient
to alter cortical activity, the effects of stimulation
are focal, and the appropriate task is chosen to test
the function of the cortical region stimulated. Direct
cortical stimulation for brain mapping can be done
extraoperatively using implanted subdural electrodes,
as is often done in patients undergoing an epilepsy
surgery evaluation, or intraoperatively. Both techniques generally employ the same principles toward
mapping brain function but differ somewhat with
regard to their benefits and limitations. This chapter
focuses on the technical aspects of intraoperative
brain mapping using direct cortical stimulation.
8.2. Concurrent electrocorticography
Functional brain mapping with direct cortical stimulation should always be performed using concurrent
electrocorticography (ECoG). The ECoG is typically interpreted by a clinical neurophysiologist
who is in the operating room with the neurosurgeon.
This enables quick communication between the
151
neurosurgeon and clinical neurophysiologist of any

relevant technical problems or neurophysiological
findings.
The ECoG serves two main purposes. One is to
detect stimulation artifact while stimulating the
cortex (Fig. 1). The presence of stimulation artifact
verifies that the cortex is indeed being electrically
stimulated, and it also verifies that the ECoG is
recording properly. The other purpose of the ECoG
is to detect the occurrence of stimulation-induced
afterdischarges and seizures (Figs. 1 and 2). The
occurrence of afterdischarges or seizures, following
a train of stimulation, raises the possibility that the
physiological effect of that train of stimulation may
not have been well localized and may have propagated to a wider cortical region (Haglund et al.,
1992) or more distant brain sites. Thus, a neurological effect seen with stimulation associated with afterdischarges or seizures may not be related to the
cortical region stimulated, but rather, it may be
related to propagation of hypersynchronous neuronal
activity to more distant sites. Thus, functional data
from stimulations associated with afterdischarges or
seizures is not helpful in mapping brain function.
R
C
1
2
3
4
5
6
7
8
Recording Stimulus arifact stops.

continues. Afterdischarges begin.
Recording begins.
Stimulus arifact seen.
Fig. 1. This eight-channel electrocorticography tracing done using a 1 8 electrode strip demonstrates stimulus artifact
and afterdischarges. For this recording, the stimulating electrodes were held nearest to electrodes 7 and 8. The recording
actually begins near the middle of the display (solid arrow). Stimulus artifact can be seen in all eight channels, with maximum stimulus artifact seen at channels 7 and 8. The recording continues left to right until the far right end of the display
and resumes at the far left end of the display (open arrow) where stimulus artifact continues. At the stippled arrow, the stimulus artifact abruptly ceases and afterdischarges are seen at channels 7 and 8. The recording settings for this ECoG are: lowfrequency filter 0.5 Hz, high-frequency filter 100 Hz, and sensitivity 200 mV/mm. Bipolar stimulation was administered at a
rate of 60 Hz and an intensity of 14 mA.
152
L.M. SCHRADER
11-01 1
10 s
1 mV
12-01 2
10 s
500 V
13-01 3
10 s
500 V
14-01 4
10 s
500 V
15-01 5
10 s
50 mV
16-01 6
10 s
500 V
17-01
10 s
50 mV
18-01
10 s
50 mV
Fig. 2. This eight-channel recording demonstrates the end of an afterdischarge. The channels are numbered 1 through 8,
with the top channel being channel 1. Because of excessive technical artifact present in channels 5, 7, and 8, the amplification of these channels was adjusted to make these channels appear to be almost flat and the remaining channels
more readable. There is an ongoing afterdischarge as this 10-s epoch begins (the unit of time between each faint
gray line is 1 s). The beginning of this afterdischarge as well as the stimulus artifact from the train of stimulation
that produced it occurred prior to the start of the epoch shown here. The afterdischarge has maximal amplitude at channel
6 but can also be seen at channel 4. It is likely that the afterdischarge would have also been seen in some of the neighboring channels that were reduced in amplification for technical reasons. The afterdischarge ends 7 s into the displayed
record.
ECoG is typically done using saline-soaked disk

electrodes placed on the cortical surface. Typically,
the electrodes are embedded in thin silastic sheets,
referred to as strips or grids. A separate insulated
wire runs from each electrode contact to a connector
plug at the end of the electrode. To ensure the best
recording, saline-soaked sterile cloth is laid over the
grid/strip to gently weigh down the electrodes and
provide adequate contact with the cortical surface.
Typically four to eight electrodes are sufficient, since
the exposed cortex is of limited area for most craniotomies. A low frequency filter of 0.5 Hz and
high-frequency filter of 70100 Hz are used to ensure
adequate recording of afterdischarges and background activity. A sensitivity setting of 30200 mV/mm
is commonly used.
8.3. Electrical stimulation

Testing of patients for motor, sensory, and/or language function is done during trains of electrical stimulation applied to focal regions of exposed cortex
during neurosurgery. The neurosurgeon moves the
sterile stimulating electrodes from one brain region
to the next throughout testing to map out the function
of the exposed cortical surface. The electrical stimulation may either produce neurological symptoms,
such as movement or paresthesias, or inhibit neurological function, such as slowing movement or creating speech arrest. If stimulation of a focal brain
region produces such a neurological finding reliably,
without afterdischarges or seizures, it is then concluded that the region of cortex stimulated is important
for that brain function. At one center, the criteria for

considering an area eloquent is that the functional
response occurs during at least three separate stimulation trials to a single cortical region (Pouratian
et al., 2004).
Bipolar electrical stimulation is preferred over
unipolar stimulation (which uses a more distant reference electrode) because it appears to provide more
focal stimulation. In models comparing bipolar and
unipolar stimulation, the electrical field distributions
induced in the cerebral cortex are more focal with
bipolar stimulation (Nathan et al., 1993). Bipolar
stimulation in the operating room for functional
mapping is performed using a sterile two-pronged
wand with spherical or semi-spherical electrodes of
12 cm diameter at the end of each prong. The electrodes are usually separated from each other by
35 mm. This wand is held by the neurosurgeon,
who can move the wands stimulating electrodes from
one cortical location to the next, as needed, to test the
function of different brain regions. Current is generated between the two electrodes when a voltage difference is imposed. A constant current stimulator should
be used rather than a constant voltage stimulator; constant current stimulators deliver a reliable amount of
current to the tissue each time. This is important since
the potential for injury relates to the amount of current applied. The amount of current flow applied with
constant voltage stimulators, on the other hand, can
vary depending upon the impedence of the tissue
between the two stimulating electrodes (Barry,
1991). Biphasic square wave pulses of equal total
charge are used since monophasic pulses are more
likely to cause tissue injury (Mortimer et al., 1970;
Pudenz et al., 1975a).
Single pulses of stimulation are generally of
0.10.3 ms in total duration, since pulse durations
under 0.3 ms tend to excite cell bodies, while those
of 1 ms or more tend to excite axons. Pulses are
typically delivered at a frequency of 5060 Hz in a
215-s train. For assessing motor or sensory function,
a 23-s train is usually sufficient. For assessing
language function, longer train durations of 57 s are
typically needed. Even longer train durations may be
used, if necessary. At UCLA, we use 5-s trains for
routine motor and language mapping, although we
interrupt the train early once a clear motor response
is seen. Occasionally, we need to prolong the stimulation train to 710 s for language mapping, depending
on the complexity of the task and the speed with
which the patient is responding to the task.
153
There is some debate as to the best method for

selecting current intensity. When discussing current
intensity, the concepts of mapping thresholds and
afterdischarge thresholds must be introduced. The
mapping threshold is the lowest level of stimulation
that is required to produce a functional change (such
as a movement) in the patient. The afterdischarge
threshold is the lowest level of stimulation that
produces afterdischarges.
With regard to the current intensity used in intraoperative stimulation, there are generally two
approaches currently in use. At many centers, an
incremental increase method is used in which the
intensity of current stimulation is incrementally
increased at each cortical site in order to ensure
absence of eloquent function at maximal stimulation
intensities. A standard protocol for this method begins
with stimulation intensities at 24 mA that are then
increased in 2 mA increments to a maximum intensity
of 16 mA or until the mapping threshold or afterdischarge threshold is reached. If afterdischarges
occur, the intensity is turned down by 2 mA and
function is tested at this lowered intensity. This is
done for each cortical site tested (Pouratian et al.,
2004). At other centers, a single intensity method is
used for mapping the entire exposed cortex, and the
current level used is set just below the afterdischarge
threshold for the entire exposed cortex. A standard
protocol for this method begins with a 2-mA current
applied for 4 s to cortex adjacent an ECoG electrode.
Stimulation is then repeated at increasing current
intensities until afterdischarges are evoked, the patient
reports a response, or an arbitrary upper limit (such as
10 mA) is reached. The current is then reduced 2 mA
below this afterdischarge threshold, and the afterdischarge threshold at a neighboring electrode is then
determined. This is repeated until all electrodes are
sampled, which requires 510 min. Then, a current
intensity that is below the previously obtained discharge thresholds is selected for mapping the entire
exposed cortical surface (Ojemann, 1997). At present,
it is unsettled which of these two methods is best, as
there have been no head-to-head studies comparing
the postoperative functional outcomes of patients
using these two methodologies.
There are several theoretical advantages and disadvantages to both methods. There is convincing
evidence from two centers that both mapping thresholds and afterdischarge thresholds vary from one
region of cortex to another within an individual
(Lesser et al., 1984; Pouratian et al., 2004). This is
154
even true for adjacent cortical sites. Importantly, it

appears that mapping thresholds are often equal to,
or greater than, afterdischarge thresholds in adjacent
cortical regions (Pouratian et al., 2004). Thus, it
would seem that the most reliable way to map out
function is to use the incremental increase method
to ensure that the mapping threshold is reached for
each cortical site, and thus ensure that no areas of
eloquent cortex are missed. Using the single intensity
method may miss the mapping threshold in some
regions tested and thus give the false impression
that those regions are not important for eloquent
function. Therefore, the incremental increase method
is the preferred method at some institutions. A potential disadvantage to the incremental increase method
is the concern that this method requires more time
and the resulting extension in surgery time poses an
increased risk to the patient. However, this has not
been shown to be the case since, in most cases,
mapping using the incremental increase method can
be done in 3045 min (Pouratian et al., 2004).
Another concern about the incremental increase
method is that maximizing current may falsely identify a higher number of cortical regions as eloquent
and thus unduly limit the resection of pathological
tissue. Maximizing current may result in unintentional stimulation of a broader cortical area, and this
could result in less-specific cortical mapping. While
optical imaging during cortical stimulation does
reveal a larger area of activation with higher stimulus
intensities, the activation nevertheless remains localized just around the stimulating electrodes and does
not activate more distant sites (Haglund et al.,
1993). Lastly, another potential disadvantage of the
incremental increase method is the increased risk of
provoking afterdischarges or seizures. The occurrence
of afterdischarges or seizures suggests that the effects
of stimulation were less well-localized (Haglund
et al., 1992) and thus the preceding stimulation had
less reliable mapping value.
8.4. Localizing value of stimulation
In the absence of afterdischarges or stimulationevoked seizures, the physiological effects of bipolar
stimulation are more focal. Modeling of current density has demonstrated uniform current distribution
beneath the stimulating electrodes with a smooth and
rapid drop-off with increasing distance (Nathan et al.,
1993). Using this model, a 10 mA stimulating current produces a peak current density of 0.05 A/cm2
L.M. SCHRADER
in the region immediately beneath the electrodes,

which declines rapidly to 0.02 A/cm2 at a distance of
5 mm from the stimulating electrodes. Since current
density is inversely proportional to the square of the
distance from the electrodes, current density continues
to drop off very rapidly with increasing distance from
the electrodes.
Optical imaging in primates and humans has confirmed that the physiological effects of stimulation
are confined to the tissue between the stimulating
electrodes (Haglund et al., 1992, 1993). When stimulation current intensity was the only parameter
varied, a graded optical imaging response was seen
with increased area of activation and intensity of activation (Haglund et al., 1993). Even with increasing
stimulation intensity, the cortical activation nonetheless remained localized just around the stimulating
electrodes, and no activation of sites further away
or incongruent areas of activation occurred (Haglund
et al., 1993). Increasing stimulus train durations similarly produced a graded response, again without activation of local areas, other than those closely
associated with the stimulating electrodes (Haglund
et al., 1993). The intensity and spread of optical
changes induced by stimulation that provokes afterdischarges is greater for stimulation followed by long
(1216-s), compared to short (<4-s), afterdischarges
(Haglund et al., 1992). This fact supports the notion
that the presence of afterdischarges renders the preceding train of stimulation less reliable in determining
the function of the stimulated brain region.
Mapping studies of motor and language function
also suggest that the effects of intraoperative cortical
stimulation are well localized. Moving the stimulating electrodes 0.52 cm along the same gyrus can
yield very different effects on neurological function
(Ojemann and Whitaker, 1978; Ojemann, 1983).
A site that has no disruption of language function
with cortical stimulation can be within 2 cm of, and
along the same gyrus as, a site associated with
consistent language disruptions (Ojemann, 1979).
Stimulation mapping has revealed that language
localization is restricted to small regions of 12 cm2
(Ojemann, 1983, 1991). Postoperative outcome studies have revealed that if the margins of a cortical
resections are within 1.52.0 cm of sites identified
by stimulation mapping as consistently involved in
language function, significant increases in postoperative language deficits occur. If the surgical margins
are >2 cm from such sites, no permanent language
deficits occur (Ojemann, 1983, 1991; Haglund
et al., 1994). It appears that, in the absence of afterdischarges or seizures, the neurophysiological effects
of stimulation remain localized to the cortical region
beneath the stimulating electrodes.
There remain some limitations to the localizing
value of cortical stimulation. While there is substantial evidence that the brain regions stimulated are
typically focal, it always remains a possibility that
some behavioral effects of cortical stimulation could,
at least at times, be secondary to propagation from
the cells beneath the electrodes to other cells at more
distant sites. Another limitation is that there is no
way to know the function of brain tissue between
sites that have been stimulated. Stimulation mapping
is also difficult beyond the edges of the craniotomy.
Thus, the exposure should be generous enough to
include brain regions that are likely to be of functional
importance (Ojemann, 1997).
8.5. Anesthetic protocol
Mapping of the primary motor cortex can be done
under general anesthesia as long as the patient is
not paralyzed. For language mapping protocols that
require patient participation, an asleepawakeasleep
technique can be used, as long as it is anticipated that
the patient will be able to cooperate in remaining still
(while awake) in the operating room during language
testing (Huncke et al., 1998). For the awakeasleep
awake protocol, general anesthesia with propofol or
sodium thiopental using endotracheal intubation is
initially induced. Patients are hyperventilated during
dural opening. Operative sites are infiltrated with
0.5% bupivacaine with epinephrine to provide local
anesthesia. The endotracheal tubes are modified
by attaching a fine catheter with multiple holes
(punctured with a 25-gauge needle) that allows for
the spraying of local anesthetic via the catheter to
the pharynx, larynx, and trachea, in anticipation
of the intraoperative extubation. Patients are then
awakened and extubated for functional mapping.
After mapping, patients again undergo intubation
with the aid of a fiber-optic laryngoscope or tube
changer and undergo general anesthesia for the
remainder of the operation.
8.6. Safety
Cortical stimulation raises several safety issues.
Neural damage may occur due to charge transfer
across the electrodetissue interface, which may
155
produce pH changes, oxidation, and deposition of

metal ions from electrodes (Agnew and McCreery,
1987). Damage may also occur from the passage of
current through the tissue itself, which can result in
neural hyperactivity and tissue heating (Agnew and
McCreery, 1987).
8.6.1. Charge density
Animal studies of cortical stimulation have concluded that charge density delivered to brain tissue
is an important factor in determining the safety of a
cortical electrical stimulation protocol. Such animal
studies have examined the effects of continuous stimulation of a single brain region for anywhere from
0.5 to 205 h (Mortimer et al., 1970; Pudenz et al.,
1975b; Babb et al., 1977; Dauth et al., 1977). These
studies have suggested that the upper limit of the
range of charge densities that would be considered
safe is 40 mCi/cm2/phase (Agnew and McCreery,
1987).
Because the stimulation parameters used in animal
studies are very different from those used for intraoperative mapping (short 37-s trains of stimulation
over several minutes for a single cortical region),
the safety guideline based on animal research does
not extrapolate directly to safe limits of charge density for intraoperative mapping in humans. Intermittent stimulation, such as that used in intraoperative
functional mapping, is less damaging than continuous
stimulation (Dauth et al., 1977; Pudenz et al., 1977).
Several reasons for this have been proposed (Dauth
et al., 1977). It has been suggested that stimulation
may produce toxic substances that may accumulate
to toxic levels over time if stimulation proceeds uninterrupted. Intermittent stimulation may allow the
opportunity for such substances to be periodically
removed and thus not reach toxic levels. Another
hypothesis is that intermittent stimulation may prevent electrode polarization to such an extent that it
prevents deposition of metal ions. Lastly, compared
to animals, the thicker pia-arachnoid, and larger
volume of subarachnoid fluid in humans, may reduce
risk of injury by producing greater shunting of
current (Pudenz et al., 1975b).
Intraoperative cortical stimulation for functional
mapping in humans employs stimulation parameters
that result in higher charge densities than the upper
recommended limit based on animal studies. In fact,
intraoperative cortical stimulation performed at several centers utilizing several different methodologies
156
produces charge densities of 159796 mCi/cm2/phase

for peak currents of 25 mA (Gordon et al., 1990).
Such higher charge densities appear to be safe and
tolerated in this setting because the stimulations are
intermittent and involve only several brief successive
bursts of stimulation to each brain region tested.
Despite these high charge densities, there is abundant evidence that the stimulation parameters used in
human intraoperative functional mapping are safe.
Histological examinations reveal that no brain injury
occurs. In a study of three patients who underwent
extraoperative functional mapping with electrical
stimulation through implanted subdural grid electrodes, no histopathological abnormalities attributable
to stimulation were seen, when comparing tissue that
received electrical stimulation to tissue that did not
(Gordon et al., 1990). In this study, the maximum
charge density was 5257 mCi/cm2/phase. In 2 of
the 3 patients, 1 electrode site was used for each
patient as a common reference for stimulation,
receiving over 251 stimulation trials. Most of these
stimulation trials consisted of 25-s trains of stimulation at currents of 12.515.0 mA, and these mostly
occurred 1 day prior to resection. In this study, the
stimulating electrodes were subdural grid electrodes,
which have a larger surface area (and thus produce
a smaller charge density) than the stimulating wand
electrodes used in intraoperative mapping. Thus,
charge densities during intraoperative mapping that
uses wand-stimulating electrodes tends to be higher
than charge densities during extraoperative mapping
through subdural strip/grid electrodes. Ojemann has
also reported that there have been no histopathological changes in resected tissue that has undergone
intraoperative stimulation (Ojemann, 1997). Aside
from histopathological evidence, the fact that there
is no decline in patients neurological function
noticed during intraoperative testing between bursts
of stimulation in the operating room is further evidence that stimulation is not damaging to brain function. Perhaps the most compelling evidence that
intraoperative stimulation mapping is not damaging
to cortical tissue comes from the studies that have
assessed postoperative neurological function in
patients who did or did not undergo stimulation
mapping. In a study of patients with low-grade gliomas, patients with and without intraoperative electrical stimulation mapping were compared (Duffau
et al., 2005). The use of functional mapping with cortical stimulation led to a decrease risk of neurological
sequalae, compared to patients operated on without
L.M. SCHRADER
mapping. In this study, the use of intraoperative electrical stimulation had further benefits that included
expanding the pool of patients with gliomas in eloquent cortex that could undergo surgery, improving
the quality of tumor resection and improving survival. Therefore, despite the high charge densities
used in intraoperative functional mapping with cortical stimulation, the risks appear minimal compared to
the substantial benefits of preserved brain function
seen in patients who undergo this procedure.
8.6.2. Stimulating electrodes
With chronic cortical stimulation, there is a risk of
metal ion deposition into cerebral tissue when the
stimulating electrodes are made of certain types of
metals, such as stainless steel (Mortimer et al.,
1970). While this is generally considered less of an
issue with the brief bursts of stimulation that occur
over 0.51 h of testing in the operating room, the
stimulating electrodes are nonetheless typically made
of noble metals, such as platinum, to safeguard
against any such risk.
8.6.3. Risk of stimulation-provoked seizure
There is always a risk that electrical stimulation may
provoke a seizure in the operating room. In animal
studies of cortical stimulation, seizures occurred in
most unanesthetized animals (Agnew and McCreery,
1987). However, the occurrence of seizures was not
correlated with neural damage. Neural damage was
instead correlated with charge density. Thus, the
occurrence of a stimulation-provoked seizure is
thought to have no negative long-term consequences
to the brain. Stimulation-provoked seizures during
intraoperative monitoring are usually simple partial
seizures, or less frequently, complex partial seizures
that are self-limited. Stimulation-provoked seizures
are typically managed easily by the anesthesiology,
surgery, and neurophysiology operating room teams
with intravenous agents, such as short-acting barbiturates. Another option for disrupting intraoperative
stimulation-induced seizures is the application of cold
Ringers lactate solution. In a recent study (Sartorius
and Berger, 1998), cold Ringers lactate solution was
applied directly to the cortex in 22 patients with
stimulation-induced seizures that occurred during
intraoperative brain mapping procedures. The irrigation rapidly and reliably terminated these simple
partial seizures. If the seizure is associated with an
alteration of consciousness, it may create a temporary pause in the cortical mapping procedure while
the patient recovers back to their functional baseline. To decrease the risk of seizures, it is helpful to
not increase the intensity above the afterdischarge
threshold.
8.6.4. Kindling
While kindling can occur under certain circumstances with high-frequency stimulations, no change
suggestive of kindling has occurred in human
patients undergoing stimulation mapping. In examining afterdischarge thresholds from one day to
another, for patients with implanted subdural electrodes who were stimulated over successive several
days, there was no progressive decrease in afterdischarge threshold, an expected result in the process
of kindling (Lesser et al., 1984, 1987). This is not
surprising given the fact that the neocortex is difficult
to kindle, that the stimulation parameters used to produce kindling in animal models are very different
that those used for intraoperative mapping, and that
kindling is more difficult in primates than in other
animals (Lesser et al., 1998).
8.7. Mapping of specific functions
8.7.1. Motor
Classic teaching holds that the primary motor and
sensory cortices consist of a narrow strip both anterior
and posterior to the Rolandic fissure, respectively.
However, it has been demonstrated that such a
straightforward delineation of functional cortex is
an oversimplification (Penfield and Jasper, 1954),
and that there is great interindividual variability in specific localization of the human cortical motor map
(Uematsu et al., 1992; Branco et al., 2003). Uematsu
and colleagues examined mapping results done with
stimulation via subdural electrode grids in 35 patients
with epilepsy (Uematsu et al., 1992). These authors
found that two-thirds of the primary motor responses were located within the 10 mm strip anterior to
the Rolandic fissure, and the remaining one-third were
>10 mm anterior to the Rolandic fissure, or were posterior to it. Furthermore, in the patients with brain
lesions, less than one-third (28.1%) of the responses
were within the 10-mm narrow anterior strip. This
finding indicates that primary motor cortex may
extend beyond the gyrus immediately anterior to
157
the Rolandic fissure. In addition, while there was a

general trend toward body part representations following the classic homunculus, there was considerable
intermixing of body part representation. Lastly, there
was also significant interindividual variability in
motor maps.
Motor mapping can be done under general anesthesia in the absence of paralytic agents or with the
patient awake. In an awake patient, current intensities
of 26 mA are usually sufficient to produce a
response. In the setting of general anesthesia, higher
stimulus currents are required to produce responses,
and the tongue motor cortex cannot usually be identified (Ojemann, 1997). Testing of motor function is
typically done by assessment for involuntary contralateral tongue, face, or arm movements during
cortical stimulation in an unparalyzed patient. This
requires the use of an assistant who is closely observing the patient. Electromyography can also be used to
detect motor responses. Movements produced by
stimulation may be jerking or tonic contraction that
may begin immediately upon stimulation, or take a
couple seconds to develop. Regions of cortex associated with producing movements should not be
removed in order to preserve motor function.
It should also be noted that stimulation of certain
cortical regions can also produce inhibition of movement (Penfield and Jasper, 1954). For example, in an
awake patient who is asked to wiggle their fingers,
stimulation of some cortical regions may slow or
arrest this movement. This type of motor inhibition
generally occurs with stimulation of the supplementary motor cortex and other regions that may assume
an integrative role in motor function. Supplementary
cortex and cortex associated with inhibition of movement can typically be surgically removed without
significantly impairing postoperative motor function
(Lesser et al., 1998; Fontaine et al., 2002). However,
safe resection in the region of the supplementary
motor area requires intraoperative mapping of the
primary motor cortex and its subcortical fibers
that are posterior and inferior, respectively, to the
supplementary motor area (Rostomily et al., 1991).
8.7.2. Sensory
Sensory mapping is generally less reliable than motor
mapping because it relies on the subjective report of
the patient. Stimulation of somatosensory cortex can
result in patient-reported paresthesias localized to
contralateral discrete body regions. Occasionally,
158
sensation is felt on both sides of the face or tongue

and usually on both sides of the throat (Penfield
and Jasper, 1954). Like the organization of the
primary motor cortex, a strict localization of the primary somatosensory cortex to the postcentral gyrus is
an oversimplification. Penfield and Jasper (1954)
found that about 75% of the stimulations that produced sensation were postcentral in position and
25% were precentral. The character of the sensation
was the same at either location except that the desire
to move an extremity almost always came from stimulation of the precentral area. It should be noted,
however, that removal of the precentral gyrus does
not result in any lasting sensory deficit (Penfield
and Jasper, 1954).
Electrical stimulation of the primary visual cortex
produces the appearance, in the contralateral visual
field, of lights, shadows, colors, and simple shapes
that are often reported to be moving (Penfield and
Jasper, 1954). Stimulation of the visual association
cortex can lead to formed hallucinations (Penfield
and Jasper, 1954).
8.7.3. Language
Areas of cortex important for language are small
well-localized brain regions, and the specific anatomic localization of these small brain regions vary
from person to person (Ojemann, 1979). In a series
of 117 patients undergoing intraoperative mapping,
Ojemann and colleagues found that most patients
had essential sites for language with surface areas
of 2 cm2 or less. Only 16% had any sites as large as
6 cm2 (Ojemann, 1989). Some sites had sharp boundaries whereas others had surrounding areas of less
consistent language disruptions with stimulation,
suggesting a more gradual transition for some sites
from essential language cortex to cortex not essential
for language function. Most patients showed several
essential perisylvian areas for language function.
In the 117 patients, two-thirds had 2 sites, and onequarter had 3 sites, separated by cortex unrelated to
language. Usually, there was one frontal and one or
more temporoparietal sites. However, multiple sites
were sometimes found in the frontal lobe. This considerable interindividual variability of language localization makes functional mapping extremely important
in helping to maximize the margins of a resection,
yet also preserve function.
Perisylvian sites identified as important to language by intraoperative stimulation can predict the
L.M. SCHRADER
language deficits of a resection (Ojemann and Dodrill,

1985; Haglund et al., 1994). On the other hand, there
are brain regions in the dominant hemisphere outside
the perisylvian region that are associated with
stimulation-evoked language disturbances, yet their
resection would typically not cause a permanent language deficit (Ojemann, 1997). Nonetheless, resection
of such regions can cause a transient language deficit.
These regions include the supplementary motor area
in the superior frontal lobe (Fried et al., 1991;
Rostomily et al., 1991; Ojemann, 1997) and the basal
temporal cortex (Luders et al., 1986; Ojemann, 1997).
With regard to language mapping, Ojemann stresses that a completely negative stimulation mapping
does not provide security that resection of those sites
will not be associated with a language deficit. The
area covered by stimulation mapping should include
areas in which language is likely to be found, as well
as the area of the planned resection (Ojemann,
1993). Thus, it is important that areas essential for
language be identified in order to assure that a negative mapping result near the resection site will not
result in postoperative deficits.
Prior to testing with cortical stimulation, it is
important to first establish a steady baseline of accurate language performance. Without this, it is difficult to ascertain if language disruption during
stimulation is due to the stimulation itself, or merely
the result of an inconsistent performance at baseline.
The most commonly used task in language
mapping is visual object naming. For this task,
patients are asked to name objects presented in the
form of line drawings from the Boston naming test,
a standardized test of object naming with wellestablished normative ranges (Kaplan et al., 1976).
It is helpful to choose pictures of objects that a subject can name quickly without difficulty in preoperative testing (Ojemann, 1979). To establish a
reliable baseline of language performance for this
task, one center requires the patient to perform accurately on a visual object-naming task for greater
than 1 min with stimuli being presented at 5-s intervals (Pouratian et al., 2004). Naming is a good
task to use for surveying areas of cortex important
for language because naming deficits are present in
most aphasic syndromes (Ojemann, 1993). For this
reason, it is one of the most commonly used tasks
in cortical stimulation language mapping. If a cortical region stimulated is important for language,
patients may exhibit speech arrest or anomia during
this naming task.
However, stimulation of some cortical language

sites may not be associated with naming deficits
(Ojemann and Mateer, 1979). Occasionally, it may
be advisable to assess stimulation effects on other
language functions before concluding that a cortical
site can be safely resected (Ojemann, 1993). Several
other language tasks can be used. Sentence reading
has been useful for detecting cortical regions important for language that were not associated with naming difficulties upon stimulation (Ojemann, 1983).
For the Word Generation task, patients are asked to
generate lists of words beginning with a certain letter
or belonging to a certain category (e.g., animals).
Auditory Responsive Naming requires patients to
name objects that are described to them. For example, the appropriate patient response to hearing the
phrase tall pink bird would be flamingo. In functional imaging studies, the word generation and auditory responsive naming tasks activate frontal
language areas (Cuenod et al., 1995; Bookheimer
et al., 1998). Visual Responsive Naming is similar
to auditory responsive naming except instead of asking the patient to name the object after hearing the
phrase tall pink bird, the patient would be shown
these words written on a flash card. The Sentence
Comprehension task involves presenting patients
with a pair of sentences that are identical except for
one word that is replaced with either a synonym or
different word. Patients are instructed to listen to
each pair of sentences and decide whether the sentences have the same literal meaning. This task
robustly activates perisylvian regions involved in
semantic processing (Dapretto and Bookheimer,
1999). Other tasks useful in language mapping have
been described (Ojemann, 1983).
8.8. Special considerations

8.8.1. Subcortical stimulation for functional mapping
Intraoperative direct subcortical stimulation for identification of subcortical primary motor and language
pathways has been described as a useful technique
for preserving eloquent function (Berger et al.,
1990; Berger, 1995; Skirboll et al., 1996; Duffau,
2000; Duffau et al., 2002; Keles et al., 2004). Interestingly, mapping of the internal capsule has been
described in a case report (Duffau, 2000). Subcortical
motor mapping was done in a manner similar to that
of cortical mapping except for the use of a longer
159
total pulse duration, 1 ms. Biphasic square wave

pulses were administered through a bipolar electrode
at 60 Hz using a 16-mA current. Direct stimulations
were performed to detect the subcortical pyramidal
pathways. The technique was sensitive enough to
map the somatotopic organization of the internal capsule. The patient had a brief postoperative left hemiparesis with complete recovery in 10 days. In 30
patients with cortico-subcortical low-grade gliomas
in the dominant hemisphere, using the same stimulation protocol as in the above case report, language
cortical sites and subcortical pathways were clearly
identified and preserved in all 30 cases (Duffau
et al., 2002). Most patients experienced a transient
dysphasia after resection with complete resolution
within 3 months. Interestingly, in all patients, there
were specific and reproducible language disturbances
that were associated with stimulation-specific white
matter tracts. The subcallosal fasciculus stimulation
produced initiation disorders. Periventricular white
matter stimulation produced dysarthria, and stimulation of arcuate fasciculus and insular connections
produced anomia.
8.8.2. Pediatric population
Due to the decreased myelination of the central nervous system of pediatric patients, higher levels of
stimulation are usually needed to elicit responses to
functional mapping (Riviello et al., 2001; Signorelli
et al., 2004). Myelinated fibers are easier to stimulate
than unmyelinated fibers since all of the current
flows out of very tiny membrane areas restricted to
the nodes of Ranvier. Since the current flow is very
concentrated in myelinated fibers, small currents
can produce significant local depolarization (Barry,
1991).
The challenges that arise in the pediatric population are highlighted by one study that used typical
adult stimulation parameters for mapping motor cortex. In this study, motor responses to stimulation
were unobtainable in children younger than 1 year
of age and present in just 18% of those 45 years
old and in 51% in those 89 years old (Duchowny
and Jayakar, 1993). Thus, absence of a functional
response to cortical stimulation in young children
did not ensure the absence of eloquent cortical
function in areas tested. Because of this, in children,
it is often helpful to stimulate a larger sample of
cortical regions because discovery of a positive
motor response in one region can help reassure that
160
other regions that were tested, which did not produce a functional response, were indeed not eloquent
cortices.
Higher intensities are required to activate motor
responses in children, compared to adults, who
typically have mapping thresholds of <6.5 mA and
afterdischarge thresholds of <12 mA (Resnick
et al., 1988; Duchowny and Jayakar, 1993). In a
study of five children, aged 6 months to 9 years,
motor mapping was done using a standard adult stimulation protocol: 0.3 ms biphasic pulses at 50 Hz
with 35-s train durations (Resnick et al., 1988).
Higher stimulus intensities were required to reach
mapping and afterdischarge thresholds in children,
compared to typical values found in adults (Resnick
et al., 1988). The highest motor mapping thresholds
occurred in the youngest children and afterdischarges
were rarely elicited.
The need for higher intensities of stimulation
raises safety concerns. To help balance the need for
increased stimulation levels to provoke functional
responses, a dual paradigm stimulation protocol was
developed that involves incremental increases in both
stimulus intensity and pulse durations (Jayakar et al.,
1992). Such a paradigm involves a longer pulse
duration which improves the ability to obtain functional responses to stimulation in children (Riviello
et al., 2001; Signorelli et al., 2004). In a series of
17 patients ranging in age from 3 to 16 years, a protocol utilized 1 ms biphasic pulses at 50 Hz to perform cortical and subcortical stimulation under
general anesthesia. Results indicated that it was possible to locate the motor strip in 15 out of 17 patients,
including all children aged 5 years and younger
(Signorelli et al., 2004). In the two cases in which
stimulation failed to produce a motor response, the
reason is thought to be because, in both cases, the
posterior location of the tumor resulted in a craniotomy posterior to the motor strip. Thus, longer pulse
durations appear to greatly increase the yield of
mapping with cortical electrical stimulation in the
pediatric population.
A typical protocol in children includes squarewave biphasic pulses, a rate of 50 Hz, and 0.5
1.0 ms total pulse duration. The range of stimulation
current is 0.515 mA (Riviello et al., 2001; Signorelli
et al., 2004). When a given cortical region is stimulated, stimulation starts at a low current level and is
incrementally increased until the onset of afterdischarges, a functional response, or a seizure. The
interval between successive trains of stimuli should
L.M. SCHRADER
be at least 1 min and longer if a seizure occurred

(Riviello et al., 2001; Signorelli et al., 2004).
While mapping of the motor cortex can be done in
an anesthetized, unparalyzed patient, somatosensory
and language mapping require an awake, cooperative
patient. Thus, a child needs to be old/mature enough
to understand the procedure, and cooperate with testing. In general, children older than 1012 years can
cooperate with such mapping during an awake craniotomy (Soriano et al., 2000).
References
Agnew, WF and McCreery, DB (1987) Considerations for
safety in the use of extracranial stimulation for motor
evoked potentials. Neurosurgery, 20: 143147.
Babb, TL, Soper, HV, Lieb, JP, Brown, WJ, Ottino, CA and
Crandall, PH (1977) Electrophysiological studies of longterm electrical stimulation of the cerebellum in monkeys.
J. Neurosurg., 47: 353365.
Barry, DT (1991) AAEM minimonograph #36: basic concepts of electricity and electronics in clinical electromyography. Muscle Nerve, 14: 937946.
Berger, MS (1995) Functional mapping-guided resection of
low-grade gliomas. Clin. Neurosurg., 42: 437452.
Berger, MS, Ojemann, GA and Lettich, E (1990) Neurophysiological monitoring during astrocytoma surgery.
Neurosurg. Clin. N. Am., 1: 6580.
Bookheimer, SY, Zeffiro, TA, Blaxton, TA, Gaillard, WD,
Malow, B and Theodore, WH (1998) Regional cerebral
blood flow during auditory responsive naming: evidence
for cross-modality neural activation. Neuroreport, 9:
24092413.
Branco, DM, Coelho, TM, Branco, BM, Schmidt, L, Calcagnotto, ME, Portuguez, M, Neto, EP, Paglioli, E, Palmini, A, Lima, JV and Da Costa, JC (2003) Functional
variability of the human cortical motor map: electrical
stimulation findings in perirolandic epilepsy surgery.
J. Clin. Neurophysiol., 20: 1725.
Cuenod, CA, Bookheimer, SY, Hertz-Pannier, L, Zeffiro,
TA, Theodore, WH and Le Bihan, D (1995) Functional
MRI during word generation, using conventional equipment: a potential tool for language localization in the
clinical environment. Neurology, 45: 18211827.
Dapretto, M and Bookheimer, SY (1999) Form and content: dissociating syntax and semantics in sentence comprehension. Neuron, 24: 427432.
Dauth, GW, Defendini, R, Gilman, S, Tennyson, VM and
Kremzner, L (1977) Long-term surface stimulation of
the cerebellum in the monkey. I. Light microscopic,
electrophysiologic, and clinical observations. Surg.
Neurol., 7: 377384.
Duchowny, M and Jayakar, P (1993) Functional cortical
mapping in children. In: O Devinsky, A Beric and

M Dogali (Eds.), Electrical and Magnetic Stimulation of the Brain and Spinal Cord. Raven Press,
Duffau, H (2000) Intraoperative direct subcortical stimulation for identification of the internal capsule, combined
with an image-guided stereotactic system during surgery for basal ganglia lesions. Surg. Neurol., 53:
250254.
Duffau, H, Capelle, L, Sichez, J, Faillot, T, Abdennour, L,
Law Koune, JD, Dadoun, S, Bitar, A, Arthuis, F, Van
Effenterre, R and Fohanno, D (1999) Intra-operative
direct electrical stimulations of the central nervous system: the Salpetriere experience with 60 patients. Acta
Neurochir. (Wien), 141: 11571167.
Duffau, H, Capelle, L, Sichez, N, Denvil, D, Lopes, M, Sichez,
JP, Bitar, A and Fohanno, D (2002) Intraoperative mapping
of the subcortical language pathways using direct stimulations. An anatomo-functional study. Brain, 125: 199214.
Duffau, H, Lopes, M, Arthuis, F, Bitar, A, Sichez, JP, Van
Effenterre, R and Capelle, L (2005) Contribution of
intraoperative electrical stimulations in surgery of low
grade gliomas: a comparative study between two series
without (198596) and with (19962003) functional
mapping in the same institution. J. Neurol. Neurosurg.
Psychiatry, 76: 845851.
Fandino, J, Kollias, SS, Wieser, HG, Valavanis, A and
Yonekawa, Y (1999) Intraoperative validation of functional magnetic resonance imaging and cortical reorganization patterns in patients with brain tumors
involving the primary motor cortex. J. Neurosurg., 91:
238250.
Fontaine, D, Capelle, L and Duffau, H (2002) Somatotopy
of the supplementary motor area: evidence from correlation of the extent of surgical resection with the clinical
patterns of deficit. Neurosurgery, 50: 297303; discussion 303305.
Fried, I, Katz, A, McCarthy, G, Sass, KJ, Williamson, P, Spencer, SS and Spencer, DD (1991) Functional organization of
human supplementary motor cortex studied by electrical
stimulation. J. Neurosci., 11: 36563666.
Gordon, B, Lesser, RP, Rance, NE, Hart, J, Jr., Webber, R,
Uematsu, S and Fisher, RS (1990) Parameters for direct
cortical electrical stimulation in the human: histopathologic confirmation. Electroencephalogr. Clin. Neurophysiol., 75: 371377.
Haglund, MM, Ojemann, GA and Hochman, DW (1992)
Optical imaging of epileptiform and functional activity
in human cerebral cortex. Nature, 358: 668671.
Haglund, MM, Ojemann, GA and Blasdel, GG (1993) Optical imaging of bipolar cortical stimulation. J. Neurosurg., 78: 785793.
Haglund, MM, Berger, MS, Shamseldin, M, Lettich, E and
Ojemann, GA (1994) Cortical localization of temporal
lobe language sites in patients with gliomas. Neurosurgery, 34: 567576; discussion 576.
161
Herholz, K, Reulen, HJ, Von Stockhausen, HM, Thiel, A,

Ilmberger, J, Kessler, J, Eisner, W, Yousry, TA and Heiss,
WD (1997) Preoperative activation and intraoperative
stimulation of language-related areas in patients with glioma. Neurosurgery, 41: 125360; discussion 12601262.
Huncke, K, Van De Wiele, B, Fried, I and Rubinstein, EH
(1998) The asleep-awake-asleep anesthetic technique
for intraoperative language mapping. Neurosurgery,
42: 13121316; discussion 13161317.
Jayakar, P, Alvarez, LA, Duchowny, MS and Resnick, TJ
(1992) A safe and effective paradigm to functionally
map the cortex in childhood. J. Clin. Neurophysiol., 9:
288293.
Kamada, K, Todo, T, Masutani, Y, Aoki, S, Ino, K,
Takano, T, Kirino, T, Kawahara, N and Morita, A
(2005) Combined use of tractography-integrated functional neuronavigation and direct fiber stimulation.
J. Neurosurg., 102: 664672.
Kaplan, E, Goodglass, H and Weintraub, S (1976) Boston
Naming Test. Lea and Febiger, Philadelphia.
Keles, GE, Lundin, DA, Lamborn, KR, Chang, EF, Ojemann, G and Berger, MS (2004) Intraoperative subcortical
stimulation
mapping
for
hemispherical
perirolandic gliomas located within or adjacent to the
descending motor pathways: evaluation of morbidity
and assessment of functional outcome in 294 patients.
J. Neurosurg., 100: 369375.
Lesser, RP, Luders, H, Klem, G, Dinner, DS, Morris, HH
and Hahn, J (1984) Cortical afterdischarge and functional response thresholds: results of extraoperative testing. Epilepsia, 25: 615621.
Lesser, RP, Luders, H, Klem, G, Dinner, DS, Morris, HH,
Hahn, JF and Wyllie, E (1987) Extraoperative cortical
functional localization in patients with epilepsy. J. Clin.
Lesser, RP, Arroyo, S, Crone, N and Gordon, B (1998)
Motor and sensory mapping of the frontal and occipital
lobes. Epilepsia, 39(Suppl 4): S69S80.
Luders, H, Lesser, RP, Hahn, J, Dinner, DS, Morris, H,
Resor, S and Harrison, M (1986) Basal temporal language
area demonstrated by electrical stimulation. Neurology,
36: 505510.
Mortimer, JT, Shealy, CN and Wheeler, C (1970) Experimental nondestructive electrical stimulation of the brain
and spinal cord. J. Neurosurg., 32: 553559.
Nathan, SS, Sinha, SR, Gordon, B, Lesser, RP and Thakor,
NV (1993) Determination of current density distributions
generated by electrical stimulation of the human cerebral
cortex. Electroencephalogr. Clin. Neurophysiol., 86:
183192.
Ojemann, GA (1979) Individual variability in cortical
localization of language. J. Neurosurg., 50: 164169.
Ojemann, GA (1983) Brain organization for language from
the perspective of electrical stimulation mapping.
Behav. Brain Sci., 2: 189230.
162
Ojemann, GA (1989) Some brain mechanisms for reading.
In: C Von Euler, I Lundberg and G Lennerstrand (Eds.),
Brain and Reading. Mac-Millan, New York, pp. 4759.
Ojemann, GA (1991) Cortical organization of language.
J. Neurosci., 11: 22812287.
Ojemann, GA (1993) Functional mapping of cortical language areas in adults. Intraoperative approaches. Adv.
Neurol., 63: 155163.
Ojemann, GA (1997) Intraoperative Methods. In: J Engel
and TA Pedley (Eds.), Epilepsy: A Comprehensive Textbook. Lippincott-Raven Publishers, Philadelphia, pp.
17771783.
Ojemann, GA and Dodrill, CB (1985) Verbal memory deficits
after left temporal lobectomy for epilepsy. Mechanism and
intraoperative prediction. J. Neurosurg., 62: 101107.
Ojemann, G and Mateer, C (1979) Human language cortex:
localization of memory, syntax, and sequential motorphoneme identification systems. Science, 205: 14011403.
Ojemann, GA and Whitaker, HA (1978) Language localization and variability. Brain Lang., 6: 239260.
Penfield, W and Jasper, H (1954) Epilepsy and the Functional Anatomy of the Human Brain. Little, Brown, Boston, p. 896.
Pouratian, N, Sheth, S, Bookheimer, SY, Martin, NA and
Toga, AW (2003) Applications and limitations of
perfusion-dependent functional brain mapping for neurosurgical guidance. Neurosurg. Focus, 15: E2.
Pouratian, N, Cannestra, AF, Bookheimer, SY, Martin, NA
and Toga, AW (2004) Variability of intraoperative electrocortical stimulation mapping parameters across and
within individuals. J. Neurosurg., 101: 458466.
Pudenz, RH, Bullara, LA, Dru, D and Talalla, A (1975a)
Electrical stimulation of the brain. II. Effects on the
blood-brain barrier. Surg. Neurol., 4: 265270.
Pudenz, RH, Bullara, LA, Jacques, S and Hambrecht, FT
(1975b) Electrical stimulation of the brain. III. The neural damage model. Surg. Neurol., 4: 389400.
Pudenz, RH, Agnew, WF and Bullara, LA (1977) Effects
of electrical stimulation of brain. Brain Behav. Evol.,
14: 103125.
L.M. SCHRADER
Resnick, TJ, Alvarez, L and Duchowny, MS (1988) Cortical stimulation thresholds in children being evaluated
for resective surgery. Epilepsia, 29: 651652.
Riviello, JJ, Kull, L, Troup, C and Holmes, GL (2001) Cortical stimulation in children: techniques and precautions. Techniq. Neurosurg., 7: 1218.
Rostomily, RC, Berger, MS, Ojemann, GA and Lettich, E
(1991) Postoperative deficits and functional recovery
following removal of tumors involving the dominant
hemisphere supplementary motor area. J. Neurosurg.,
75: 6268.
Roux, FE, Boulanouar, K, Lotterie, JA, Mejdoubi, M,
LeSage, JP and Berry, I (2003) Language functional
magnetic resonance imaging in preoperative assessment
of language areas: correlation with direct cortical stimulation. Neurosurgery, 52: 133545; discussion 1345
1347.
Sartorius, CJ and Berger, MS (1998) Rapid termination of
intraoperative stimulation-evoked seizures with application of cold Ringers lactate to the cortex. Technical note.
J. Neurosurg., 88: 349351.
Signorelli, F, Guyotat, J, Mottolese, C, Schneider, F,
DAcunzi, G and Isnard, J (2004) Intraoperative electrical stimulation mapping as an aid for surgery of intracranial lesions involving motor areas in children.
Childs Nerv. Syst., 20: 420426.
Skirboll, SS, Ojemann, GA, Berger, MS, Lettich, E and
Winn, HR (1996) Functional cortex and subcortical
white matter located within gliomas. Neurosurgery,
38: 678684; discussion 684685.
Soriano, SG, Eldredge, EA, Wang, FK, Kull, L, Madsen,
JR, Black, PM, Riviello, JJ and Rockoff, MA (2000)
The effect of propofol on intraoperative electrocorticography and cortical stimulation during awake craniotomies in children. Paediatr. Anaesth., 10: 2934.
Uematsu, S, Lesser, R, Fisher, RS, Gordon, B, Hara, K,
Krauss, GL, Vining, EP and Webber, RW (1992) Motor
and sensory cortex in humans: topography studied with
chronic subdural stimulation. Neurosurgery, 31: 5971;
discussion 7172.

M.R. Nuwer (Ed.)
163
CHAPTER 9
Intraoperative approaches for deep brain stimulation

and targeting
Allen S. Mandir* and Robert Minahan
Department of Neurology, Georgetown University, Washington, DC 20007, USA
9.1. Introduction
Intraoperative procedures for extrapyramidal disorders have expanded in methodology and patient diagnoses. In addition, imminent further expansion is
seen for a variety of basal ganglia disorders that
involve both motor and nonmotor dysfunction (Bittar
et al., 2005a,b; Kosel et al., 2006; Mink et al., 2006).
Accompanying this proliferation in applications are
new requirements for coordinating teams in and out
of the operating theater to maximize benefits and
avoid complications. This chapter focuses upon those
neurophysiologic techniques applied intraoperatively,
which include high- and low-technology approaches
alike. In addition, a brief discussion of appropriate
perioperative issues is also included.
As is standard for optimizing any neurophysiologic study, preparation and setup is essential for
intraoperative therapeutic approaches to movement
disorder surgeries. This process starts outside the surgical theater and includes proper patient selection
and management. These first steps should increase
therapeutic yield, reduce complications and allow
for improved conditions during surgery. Correct
diagnosis, stage of disease, preexisting other conditions, and the ability for patient cooperation/cognition during awake procedures are often crucial
selection factors. Furthermore, realistic expectations
will serve to improve patient satisfaction as well as
to help with patient self-selection for procedures.
Proper postoperative management is the essential follow through for ideal patient therapeutic benefit and
Correspondence to: Allen S. Mandir, M.D., Ph.D., Department of Neurology, Georgetown University, 3800 Reservoir Rd, NW, Washington, DC 20007, USA.
Tel.: +1-202-444-7554; fax: +1-02-318-9146.
E-mail: allen_mandir@yahoo.com (A.S. Mandir).
includes what are now well-established programming

techniques as well as medical management (Volkmann and Benecke, 2002; Okun et al., 2005; Voon
et al., 2005).
For ideal results, these critical pre- and postoperative stages require strong integration to the intraoperative procedure. The neurologist performing the
intraoperative procedure should be clinically versed
with the movement disorder being treated, be able
to integrate the clinical information from patients
healthcare notes, and pass on critical information
gathered during the procedure for postoperative care.
9.2. Patient selection
The wide variety of movement disorders that are potential candidates for surgical procedures (e.g., Parkinsons
disease [PD], dystonia, multiple sclerosis, Tourettes)
necessitates limiting specific discussions behind patient
selection, but general principles apply to most procedures including those related to candidates for nonmovement disorders (Defer et al., 1999; Volkmann
and Benecke, 2002; Mink et al., 2006). Foremost, these
procedures are typically elective in nature, thus allowing for the obvious standard medical preoperative
evaluation (e.g., anticoagulation status, cardiac and pulmonary function). Beyond this, however, an awake and
cooperative patient markedly enhances safety and efficacy in the operating room (Kumar and Lange, 2003).
As with other procedures, a careful assessment of risk
to benefit should be performed for patients individually.
Those patients with preexisting dementia are at
increased risk for complications in the operating room
as well as in postoperative care (Volkmann and Benecke, 2002; Okun et al., 2005; Lang et al., 2006; Voon
et al., 2006). In addition, many procedures for movement disorders may exacerbate cognitive dysfunction
(Smeding et al., 2006). Outside of proper diagnoses,
proper staging of a patient within their disease also
164
impacts on the need for surgery as well as assessment of

risk to benefit ratio. For instance, patient response to
medical therapy may predict benefits of surgical intervention (Welter et al., 2002).
9.3. Target and procedure selection
Prior to the operative procedure, an informed decision
will be made as to the anatomical structure to be targeted and the type of operative procedure (e.g., deep
brain stimulation [DBS] implantation vs. lesion procedure or unilateral vs. bilateral procedures). These decisions go hand in hand with patient selection (Lang
et al., 2006). The anatomical structure to be targeted
clearly depends upon the patients diagnosis but interestingly, seemingly disparate disorders will share a
common anatomical target (e.g., Huntingtons disease
and PD). As well, there may be several targets from
which to choose within a single disorder (e.g., globus
pallidus, subthalamic nucleus (STN). and thalamus
for PD). Furthermore, the overlap of basal ganglia disorders for motor and nonmotor disease suggest common targets may exist for surgical intervention
(Handforth et al., 2006; Kosel et al., 2006). The decision whether to introduce a permanent lesion or
implantation of stimulating electrodes may be considered, though for the majority of movement disorder
cases, DBS surgery is preferred due to its reversible
nature (Tasker, 1998; Blomstedt and Hariz, 2006).
For all of these procedures, stereotactic technique is
the procedure of choice for the best first approximation
to localizing the anatomical target (Vitek et al., 1998).
There are different types of stereotactic frames and
frameless stereotactic procedures and each requires
preoperative imaging either individually or with a
combination of ventriculography, CT, and magnetic
resonance imaging (with the former being replaced
by the noninvasive imaging of CT and MRI).
In selecting an anatomical target, either direct or
indirect approaches are applied (Zonenshayn et al.,
2000; Starr et al., 2002; Andrade-Souza et al., 2005).
Direct approaches require an image of acceptable resolution to identify the desired anatomical structure
and/or neighboring structures to choose the approach
and target coordinates. Indirect targeting utilizes
identifiable landmarks of anterior and posterior commisures and applying fixed stereotactic coordinates
relative to these standards for target localization. To
aid in the approach and selecting coordinates of the targeted structure, imaging software packages exist that
may have the added advantage of superimposing
A.S. MANDIR
labeled anatomical atlases on patient brain scans.

Though targets for movement disorder procedures
are within deep brain structures, the trajectory
approaching these targets must transverse cortical
structures. In principle, the motor cortex is avoided to
minimize pyramidal lesions and anecdotally, the bases
of deep sulci as well as the ventricles are avoided to
reduce risks of intraoperative hemorrhage.
9.4. Intraoperative equipment
9.4.1. Clinical exam
This low-technology technique provides instrumental information during physiologic mapping and test
programming of DBS or test lesions in final target
selection. A knowledgeable movement disorder neurologist provides clinical interpretation during these
portions to assist in optimizing target localization.
The tools of the interpreting neurologist will include
knowledge of the movement disorder, other existing
neurologic disorders, and normal neurologic function
of afferent, efferent pyramidal, and extrapyramidal
pathways. An appropriate exam at baseline or even
serial baseline exams over time preoperatively may
assist in interpretations intraoperatively.
The expertise of the examiner should include an
appreciation for the inherent inter- and intrapatient
variability of movement disorder signs as well as a
knowledge of anatomy and characteristics of neuronalrelated activity of structures along the trajectory of
microelectrode recordings. Certain key portions of
movement disorder procedures including microelectrode recordings and stimulations may require keen
physical examination, detection of subtle signs, integration of symptoms, and cohesive interpretation by the
neurologist. The clinical decisions to adjust electrode
trajectories, interpret a physiologic map, and accept or
reject final targets may be heavily dependent upon an
experienced clinical interpretation to avoid serious side
effects and improve final outcomes.
9.4.2. Stereotactic frame and microdrive
A variety of stereotactic frame and frameless systems
are available, the details of which are beyond scope
of this chapter. However, if a standard frame is placed,
two general caveats to keep in mind are that this hardware will inevitably become uncomfortable with time
and may obstruct patient visual fields. Patient discomfort can reduce cooperation, concentration, and
patience during physical examinations and test programming. The latter concern with visual fields may
make testing for diplopia more difficult, for instance,
during stimulation testing for STN DBS electrode
implantation.
Incorporated into the stereotactic frame or frameless systems is a mechanism by which the recording,
microstimulating, and permanent deep brain stimulating electrodes may be advanced. The electrode is
advanced either manually or via a motorized drive
and the electrode depth is recorded at each point
along with microelectrode recording results and stimulation characteristics to create an intraoperative
physiologic map. One advantage of manual electrode
advancement is that it avoids electrical noise often
generated by the stepper motor.
9.4.3. Recording electrodes
Electrode recording is employed to provide a physiological map to refine anatomical targeting in a variety
of movement disorder procedures. Electrodes of differing characteristics have been employed, but classical
recordings utilize microelectrodes that are capable of
isolating extracellular action potentials of a single neuron (Mandir et al., 1997). That is, these electrodes have
a characteristic impedance to narrow the recording field
of view of only a few neuronal extracellular action
potentials at a time. These characteristic microelectrodes (typically around 1 MO impedance at 1 kHz)
best matches brain parenchyma corresponding to an
exposed tip diameter of about 5 mm. However, practical
issues of signal-to-noise ratios in the inherently noisy
operating theater is typically improved by lowering
this impedance (>500 kO) with a trade-off of slightly
larger field of view. Even lower impedance semimicroelectrodes are employed which typically make
single unit isolation more difficult, but in practice may
still allow for detecting characteristic firing patterns
within deep brain nuclei (Favre et al., 1996; Slavin
and Holsapple, 2004; Gross et al., 2006).
Details about microelectrode manufacture can be
found elsewhere (Mandir et al., 1997), but most often
tungsten or platinumiridium are the metals of choice;
the latter composition being more resilient to degradation and metal deposition with microstimulation.
Dependent upon the structure being targeted, individual electrodes may be advanced down trajectories
one at a time, or as in some centers, up to five electrodes are advanced in parallel tracts simultaneously.
Advantages may be touted for either single or
165
simultaneous multielectrode approaches. In our center, we elect to use single electrodes in part because
the majority of procedures target STN and we find they
require an average of less than two passes. Increased
number of electrode tracts may increase the complication rate of a procedure, though, in general, microelectrode passes provide a relativity small risk (Binder
et al., 2005; Gorgulho et al., 2005; Hamani et al.,
2005). In addition, single electrode passes in the efficient hands of experienced teams tend to negate a
potential time advantage of utilizing simultaneous
multiple electrode recordings.
For microstimulation techniques, the same electrode
employed for microrecording may be attached to a
biphasic microstimulator with the caveat that stimulation often degrades the electrode tip. Biphasic stimulation helps to lessen this effect and current is delivered
at minimal levels starting from 10 mA and held less
than 100 mA, with caution because higher currents also
may cause damage to brain parenchyma (Mandir et al.,
1997; Slavin and Holsapple, 2004). Square wave
pulses of 100300 ms at 300 Hz are typically delivered
for a total period lasting less than 10 s (Mandir et al.,
1997). For macrostimulation, a larger electrode (often
the recording electrode cannula or permanent deep
brain electrodes) is used and larger currents applied.
Precision of macrostimulation is thus lower than
microstimulation, but with the goal of determining if
electrode location is within a therapeutic region and
outside of a region that would provide unacceptable
side effects. In assessing the motor effects produced
by this stimulation, electromyography (EMG) and
accelerometry may be recorded intraoperatively in an
attempt to provide better correlation.
Some centers employ impedance measures at least
as an adjunct in deep structure localization (Heilbrun
et al., 1997; Limonadi et al., 1999; Siemionow et al.,
2000). By stimulating at a defined frequency and
constant current, a voltage can be measured and thus
impedance calculated. The general differences
between white matter and gray parenchyma produce
relative differences in impedance measurements,
though interpatient variability is often high.
9.4.5. Amplifiers, filters, and discriminators
Electrode recording signals are amplified and filtered
classically with a preamplifier near the recording
166
electrode to minimize external noise and amplified

up to about 10,000 times (Mandir et al., 1997; Slavin
and Holsapple, 2004). Filter settings (range: lowfrequency filter: 1005 kHz; high-frequency filter:
1020 kHz) are used to optimize identification of
extracellular action potentials for single unit discrimination. If isolation of single units is sought, a window
discriminator can be employed to trigger display upon
a computer screen or oscilloscope only when an action
potential of a specified size is encountered. The nature
of all-or-none neuronal firing allows for this technique
and since extracellular action potentials for a neuron
will maintain consistent morphology as well as size,
even neurons that produce action potentials of similar
amplitudes at the recording electrode may be discriminated. In addition to visual display, audio amplification of electrode recordings is employed to identify
neuronal-related firing activity. Throughout the electrode recording tract, each anatomical structure will
demonstrate their unique characteristic firing patterns
in terms of neuronal density and firing patterns (Vitek
et al., 1998; Starr et al., 2003). With experience, these
characteristics are perceived quickly by audio and/or
visual discrimination.
There is a variety of potential techniques to analyze
extracellular action potentials from isolated or populations of neurons. The more robust of these analyses
requires single unit discrimination with population
analysis reflecting information seen in impedance
measures, mentioned above. The simplest of these
analyses is to determine isolated neuronal firing rates
and interspike intervals. Neuronal firing rate is the
average activity of a neuronal over time. The interspike interval represents the interval between two successive spikes. Thus, neurons can have identical
neuronal firing rates with quite different interspike
intervals. A further characterization of interspike intervals is used to highlight those neurons that fire in a
pattern of activity demonstrating short interspike intervals or bursts (Wyler et al., 1975; Favre et al., 1999).
A percent ratio of neurons with interspike intervals of a
defined short latency over total firing over a defined
period is defined as the Burst Index. These types of
quantification of neuronal populations help to objectify descriptions of neurons that are characteristic of
populations within a structure (e.g., globus pallidus
external segment). Even more complicated analyses
have been performed post hoc from data collected during movement disorder procedures. Computing and
man power limitations in part prohibited real-time
analysis of more complexity, but as computing power
A.S. MANDIR
is increasing, the potential for real-time analysis exists.

However, despite this potential, it is not clear that any
current analyses will be particularly helpful in the
operating theater in a clinical sense. Furthermore, typically, populations within a structure are variable and
the time required for single unit isolation for many of
the more involved techniques such as cross-correlation
analysis may still make real-time analysis non desirable. Also, it is not known if the resultant analysis will
truly add information that would influence a clinical
decision as the experienced examiner is typically proficient at identifying structure populations of interest
and is able to adjust to changes in electrical noise.
The hope will be though that as advances are made in
understanding the physiologic behavior of neuronal
activity, newer analyses or insight will allow real-time
information for even further refinement of a physiologic map.
9.4.6. Anatomical atlases
Reference to a stereotaxic atlas during movement disorder procedures may help interpret the physiologic
map that is generated. This in turn assists choosing
the directive changes if additional passes are required
and in determining the location of final target. A variety of formats are available for the neurologist in the
operating room including printed atlases (Schaltenbrand and Bailey, 1959) or camera lucida drawings
on Plexiglas (Vitek et al., 1998). The latter format
allows one to draw representations of physiologically
labeled trajectories that can be overlaid on the atlas.
To best fit these physiologic mappings, they can be
manually moved within the plane or along different
parasagittal planes to help correlate with expected
anatomy. Computerized packages now exist that further offer overlaying the patients radiographic images
correlated to standard atlases and can help in plotting
the produced physiologic map onto these fused
images.
9.5. Intraoperative procedures
9.5.1. Anesthesia
Once an appropriate patient is screened for and consents to a surgical procedure for a movement disorder,
the stereotactic frame or fiducial pins are placed and
imaging performed for target selection, as discussed
above. Ideally, the movement disorders patient is fully
awake during surgery, but discretion may be used for
the anesthesia team to deliver analgesics, amnesic,

and/or sedative agents that may assist during frame
placement and drilling of burr holes along with local
anesthetics. There are patients who may require general anesthesia during movement disorder procedures,
for example, severely dystonic patients that would not
tolerate positioning on the operating table and stereotactic frame during microelectrode recordings. It is
thus important to keep in mind that when the patient
is under general anesthesia that, not only will patient
interaction be affected, but native neuronal activity
will also be affected by the agent used. Indeed, the
majority of movement disorders dissipate with sleep.
Dependent upon the patient, it may be possible to
lighten anesthesia to allow at least some periods of
patient interaction and lessen suppression of neuronal
activity. We have been able to perform intermittent
awake recordings during some dystonic patients in this
manner with the anesthesiologists titrating intravenous
propofol. However, the residual effects of these agents
upon neuronal activity even when patients are awake
enough to interact with the examiner are unknown.
Finally, patient positioning considerations include
patient comfort, especially for those patients with
rigidity, as well as practical issues of accessibility to
the patient for performing examinations of their face
and limbs.
9.5.2. Recordings
If microelectrode recordings are employed, the quality
of the recordings is evaluated at a point at the beginning of the electrode trajectory. One should note that
electrodes may mature with time upon entering
parenchyma; perhaps, as fluid enters small defects
within the glass coatings of the electrode impedance
is reduced. As such, presoaking microelectrodes in
sterile isotonic saline anecdotally may accelerate this
process. Typically, at the beginning of recording, the
167
level of extraneous noise sources and electrode impedance characteristics are evident. An abundance of
60 Hz noise prompts systematic unplugging of nonessential electrical devices that include the operative
bed, electrocautery devices, and florescent lighting.
High-impedance electrodes will increase the effect of
extraneous operating or device noise and a test of electrode impedance may reveal this condition. Since
impedance testing requires introduction of a current,
this test alone may alter microelectrode characteristics, but in the case of high-impedance electrodes
may do so in the neurologists favor.
The operating theater should be free of extraneous
noise, including asking the patient not to speak, in
order to appreciate the auditory signature of structures
that are being traversed. Focus is on the boundaries of
structures as well as on somatotopy that can be used to
refine a physiologic map to improve the predicted
radiographic anatomical map. It is important to keep
in mind that along with more apparent foreground
neuronal activity, relative background activity in
microelectrode recordings can also help identify structures. The background activity simply represents the
summation of neuronal activity more distant from the
microelectrode. A larger background, for example, is
seen with STN as compared to thalamus, since the density of active neurons is higher within STN (Fig. 1).
Conversely, white matter tracts which are devoid of
neuronal soma demonstrate inherently low background activity. White matter tracts are often key landmarks in physiologic mapping of commonly targeted
structures in PD. For instance, optic tract defines the
inferior border of Gpi and zona incerta surrounds the
STN. Thus, changes in background activity when
encountering these structures produced by the lack of
neuronal soma depolarization helps define boundaries
of structures. Other characteristics of neuronal-related
activity that can help identify a structure include neuronal density, as well as firing frequencies and firing
Fig. 1. Digitized representative neuronal microelectrode recordings of A: thalamus and B: subthalamic nucleus (STN)
from intraoperative recordings performed during physiologic mapping for deep brain stimulator electrode placement in
Parkinsons disease. Note that not only the pattern and density of the large action potentials are distinct between the two
anatomical areas, but also the baseline is larger in the STN. This latter difference is also a result of the higher density of
neuronal activity in the STN (time bases and amplitude sensitivities are the same for each trace).
168
patterns encountered. For examples, see reference

Starr et al. (2003) and the accompanying chapter on
DBS in this book.
In addition, the advantage of having a patient who is
awake during neuronal recordings allows for correlation of neuronal activity with interactive patient examination. It is helpful to have the patients limbs
accessible for movement by the examiner as well as
visualize limbs at repose for tremor. For instance, neuron activity may be related to movement of a limb, and
whether this is related to both passive (examiner
moves limb) and active (patient moves limb) can be
determined. For most deep structures, neuronal activity is most pronounced with movements of more proximal joints and of ballistic types of movements. It is
helpful for the examiner to move limbs contralateral
to the side of the deep structure being recorded in a
nonrhythmic manner. This will help verify that
neuronal-related activity is time locked to the movement and not to a sinusoidal noise pattern. Similarly,
contralateral sensory stimulation as performed during
thalamic procedures, applied in a randomly applied
pattern may demonstrate neuronal activity related to
either light touch or deep palpation. When visual sensation is related to neuronal recordings, localization
to optic tract as in trajectories through the globus pallidus internal segment may be demonstrated. This is
accomplished by darkening the operating room and
shining a flashlight in the patients eyes. As with
movement correlations to neuronal recordings, presenting the light in a nonrhythmic pattern will help distinguish a true response from patterned noise. We have
performed this latter technique even in patients who
require general anesthesia during globus pallidus trajectory mapping. Since the optic tract is a white matter
structure, the resultant evoked response is essentially
recordings devoid of somatic depolarization and represents axonal action potentials. As such, the examiner
should be attune to a small response that is often better
appreciated by sound than visualization. Though the
response is small, the signal-to-noise ratio is assisted
as the small amplitude response of optic tract axons
occurs within the low background activity inherent to
white matter.
9.5.3. Stimulation
Micro- and macrostimulation techniques during
movement disorder procedures provide additional
information for physiologic mapping. Both techniques may transiently produce beneficial or adverse
A.S. MANDIR
patient signs or symptoms to assist in localizing the

anatomical position of the electrode. Since injected
current intensity as well as the size of the electrode
will determine current spread, these parameters need
to be considered during stimulation. Microstimulation is performed using the same electrode that is
used for microelectrode recordings with the caveat
that following current delivery, the shape of the electrode may be altered with attendant lower electrode
impedance, as mentioned above. Microstimulations
are usually reserved for the target structure and performed at discrete intervals of 2 mm. The minimal
current is used to induce a response, starting at
10 mA with maximum currents less than 100 mA.
For many movement disorder surgeries, microstimulation may affect motor performance including bradykinesia and tremor in the awake patient. Physical
examination may also reveal improvement in tone.
Adverse effects may be seen and may also help in
localization, including muscle contractions (e.g.,
internal capsule), paresthesias (e.g., medial lemniscus
in STN procedures or Vac in thalamic procedures), or
forced eye movements (e.g., oculomotor area or substantia nigra pars reticulata). As with recording techniques, it is often helpful to turn the microstimulator
on and off in a nonrhythmic pattern. This can help,
for instance, when attempting to distinguish stimulusinduced muscle contractions from inherent patient
tremor. Likewise, it is helpful to have the patient
relax his muscles during observation for microstimulation, including resting his tongue on the bottom
of the mouth with a slack jaw to better examine for
microstimulation-induced tongue contractions.
Macrostimulation electrodes effect larger areas of
anatomy than microstimulation and thus are less
refined in discrimination; however, the same general
principles hold true for evaluating current intensities
for beneficial and adverse side effects. For DBS procedures, test programming of the permanent deep brain
electrodes may also be performed. As multiple contacts are provided in the DBS electrodes, different permutations of contacts serving as anode and cathode
are used in test programming at different rates and
intensities. It should be kept in mind that the ideal
pulse width and rate for test programming varies
among anatomical targets (Volkmann et al., 2006).
Ideally, stimulation results in symptomatic improvement without adverse effects, or adverse effects are
only seen at higher stimulation intensities. However,
especially in the case of macrostimulation alone, little or no improvement may be seen when the device
is turned on or off intraoperatively (Baker et al.,

2004). This may be attributable to lesioning effect,
that is, essentially a small subthalamotomy that
occurs with introducing the DBS electrodes which
have discernible girth. While in other cases, patient
fatigue or inability to fully cooperate may mask some
of these effects (Baker et al., 2004).
Typically, low-intensity stimulations that produce
side effects of motor contractions or visual phosphenes
are predictive of poor localization that requires redirection for final target localization (Baker et al.,
2004). However, we find that other side effects such
as transient paresthesias experienced by patients when
testing STN DBS lead placement, when current intensity is quickly ramped up and that do not persist, are
not contraindications to electrode placement. We,
and others, have found that many subjective and more
ill-defined side effects not related to autonomic phenomena that occur in the operating room also are
not persistent during long term DBS postoperatively
(Pollak et al., 2002).
9.5.4. Creating a physiologic map
Utilizing the data from electrode recording and stimulation, a physiologic map is produced during the
intraoperative procedure. This information is compared to the anatomical map from radiographic imaging and/or stereotactic atlas and can be instructive in
determining if additional recording passes are necessary. If more recordings are required, this information will assist in predicting the new best trajectory.
For example, consider a hypothetical electrode pass
to STN which would encounter areas of thalamus
and zona incerta above the STN and substantia nigra
pars reticulate below it. A recording pass may exhibit
neuronal-related recordings in thalamus that are characteristic of Vop and a swatch of thalamus neuronalrelated activity that extends more inferior than
expected. Accompanying these findings are a smaller
than expected swatch of zona incerta and STN
recordings, even though good passive movement correlated neuronal-related activity may be seen in STN.
Microstimulation within the target structure produces
effects of paresthesias. All of these clues predict a
trajectory more posterior than would be ideal. Vop
thalamus is situated posterior to Vim and Vac and
the relative representations of STN, and surrounding
zona inserta are different than what would be
expected with an ideal pass. Finally, the microstimulation findings also suggest a posterior pass.
169
9.5.5. Troubleshooting
The bane of every microelectrode recording setup is
external electrical noise with poor signal-to-noise ratios
which limits recording sensitivity. The techniques discussed above regarding removing potential sources of
operating room noise as well as paying attention to
microelectrode characteristics of impedance should be
first addressed. Other assistive techniques include
checking the grounding lead of the microelectrode and
other leads that may be on the patient including electrocardiogram (ECG) and electrocautery grounding pads.
A 60 Hz notch filter is seldom the preferred solution
and usually has little effect on recordings. Since audio
amplification is used during microelectrode recordings,
if external speakers are used, there is the potential for
audio feedback if the speakers are directed toward the
microelectrode leads. Adjusting speaker position or
using audio headsets will obviate the problem. Typically, it will be noticed that when the patient speaks, this
will be amplified through the recording system as well.
Even with excellent electrical characteristics,
microelectrode recordings may provide unexpected
physiologic maps. For example, this may result during
STN mapping when none of the expected thalamic
activity is experienced. This may represent an anterior
approach that skirts the border of thalamus and STN
neuronal-related activity may not be seen until a point
of entering STN. Furthermore, many structures have
patches of relatively silent areas that may mislead the
examiner into believing the inferior edge of the structure has been reached. Alternatively, skirting the edge
of a structure may show a patchy recording tract when
the microelectrode comes into contact with firing neurons. One should also keep in mind that somatotopy is
not typically adhered to within recording structures of
movement disorder surgeries. As well, inherent concentric patterns of somatotopy within structures may
prove confusing when traversing a structure. In general, when a pass is not anatomically ideal, observed
microelectrode recording and stimulation findings
provide key information to allow corrective directions
for a new stereotactic trajectory, as discussed above.
9.6. Summary
There is still debate about which technologies are
necessary for adequate physiologic mapping in
movement disorders. Each of the technologies, listed
above, may provide useful information to refine anatomic localization. The need for physiologic mapping
170
belies the resolution limitations of radiographic

imaging, the errors in stereotactic localization, and
the clear fact that anatomy does not always predict
physiology. Indeed, the ideal anatomic locale to produce maximal clinical benefit for thalamus procedures and STN procedures continue to be refined
(Tavella et al., 2002; Rodriguez-Oroz et al., 2005;
Plaha et al., 2006). The usefulness of microelectrode
recordings in our and others experiences more than
compensates for the extra effort and time this may
require. Efficiency of experienced teams minimizes
the routine time required and typically those cases
that require more extensive microelectrode recording
mappings are those we find would likely have radiographic localization with poorer outcomes.
References
Andrade-Souza, YM, Schwalb, JM, Hamani, C, et al. (2005)
Comparison of three methods of targeting the subthalamic nucleus for chronic stimulation in Parkinsons disease. Neurosurgery, 56: 360368; discussion 360368.
Baker, K, Boulis, N, Rezai, A and Montgomery, EJ (2004)
Target selection using microelectrode recording. In: Z
Israel and K Burchiel (Eds.), Microelectrode Recording
in Movement Disorder Surgery. Thieme, New York,
pp. 138151.
Binder, DK, Rau, GM and Starr, PA (2005) Risk factors for
hemorrhage during microelectrode-guided deep brain
stimulator implantation for movement disorders. Neurosurgery, 56: 722732; discussion 722732.
Bittar, RG, Kar-Purkayastha, I, Owen, SL, et al. (2005a)
Deep brain stimulation for pain relief: a meta-analysis.
J. Clin. Neurosci., 12: 515519.
Bittar, RG, Burn, SC, Bain, PG, et al. (2005b) Deep brain
stimulation for movement disorders and pain. J. Clin.
Neurosci., 12: 457463.
Blomstedt, P and Hariz, MI (2006) Are complications less
common in deep brain stimulation than in ablative procedures for movement disorders? Stereotact. Funct.
Neurosurg., 84: 7281.
Defer, GL, Widner, H, Marie, RM, Remy, P and Levivier, M
(1999) Core assessment program for surgical interventional therapies in Parkinsons disease (CAPSIT-PD).
Mov. Disord., 14: 572584.
Favre, J, Taha, JM, Nguyen, TT, Gildenberg, PL and
Burchiel, KJ (1996) Pallidotomy: a survey of current
practice in North America. Neurosurgery, 39:
883890; discussion 890882.
Favre, J, Taha, JM, Baumann, T and Burchiel, KJ (1999)
Computer analysis of the tonic, phasic, and kinesthetic
activity of pallidal discharges in Parkinson patients. Surg.
Neurol., 51: 665672; discussion 672663.
A.S. MANDIR
Gorgulho, A, De Salles, AA, Frighetto, L and Behnke, E
(2005) Incidence of hemorrhage associated with electrophysiological studies performed using macroelectrodes
and microelectrodes in functional neurosurgery. J. Neurosurg., 102: 888896.
Gross, RE, Krack, P, Rodriguez-Oroz, MC, Rezai, AR and
Benabid, AL (2006) Electrophysiological mapping for
the implantation of deep brain stimulators for Parkinsons
disease and tremor. Mov. Disord., 21(Suppl. 14):
S259283.
Hamani, C, Richter, E, Schwalb, JM and Lozano, AM
(2005) Bilateral subthalamic nucleus stimulation for
Parkinsons disease: a systematic review of the clinical
literature. Neurosurgery, 56: 13131321; discussion
13211314.
Handforth, A, DeSalles, AA and Krahl, SE (2006) Deep
brain stimulation of the subthalamic nucleus as adjunct
treatment for refractory epilepsy. Epilepsia, 47:
12391241.
Heilbrun, MP, Koehler, S, McDonald, P and Faour, F (1997)
Optimal target localization for ventroposterolateral pallidotomy: the role of imaging, impedance measurement,
macrostimulation and microelectrode recording. Stereotact. Funct. Neurosurg., 69: 1927.
Kosel, M, Sturm, V, Frick, C, et al. (2006) Mood improvement after deep brain stimulation of the internal globus
pallidus for tardive dyskinesia in a patient suffering from
major depression. J. Psychiatr. Res., 41(9): 801803.
Kumar, R and Lange, A (2003) Patient selection for movement disorders surgery. In: D Tarsy, J Vitek and A
Lozano (Eds.), Surgical Treatment of Parkinsons Disease. Humana Press, Totowa, New Jersey, pp. 5367.
Lang, AE, Houeto, JL, Krack, P, et al. (2006) Deep brain stimulation: preoperative issues. Mov. Disord., 21(Suppl. 14):
S171196.
Limonadi, FM, Roberts, DW, Darcey, TM, Holtzheimer,
PE, 3rd and Ip, JT (1999) Utilization of impedance
measurements in pallidotomy using a monopolar electrode. Stereotact. Funct. Neurosurg., 72: 321.
Mandir, AS, Rowland, LH, Dougherty, PM and Lenz, FA
(1997) Microelectrode recording and stimulation techniques during stereotactic procedures in the thalamus
and pallidum. Adv. Neurol., 74: 159165.
Mink, JW, Walkup, J, Frey, KA, et al. (2006) Patient
selection and assessment recommendations for deep
brain stimulation in Tourette syndrome. Mov. Disord.,
21(11): 18311838.
Okun, MS, Tagliati, M, Pourfar, M, et al. (2005) Management of referred deep brain stimulation failures: a retrospective analysis from 2 movement disorders centers.
Arch. Neurol., 62: 12501255.
Plaha, P, Ben-Shlomo, Y, Patel, NK and Gill, SS (2006)
Stimulation of the caudal zona incerta is superior to
stimulation of the subthalamic nucleus in improving
contralateral parkinsonism. Brain, 129: 17321747.

Pollak, P, Krack, P, Fraix, V, et al. (2002) Intraoperative
micro- and macrostimulation of the subthalamic nucleus
in Parkinsons disease. Mov. Disord., 17(Suppl. 3):
S155161.
Rodriguez-Oroz, MC, Obeso, JA, Lang, AE, et al. (2005)
Bilateral deep brain stimulation in Parkinsons disease:
a multicentre study with 4 years follow-up. Brain,
128: 22402249.
Schaltenbrand, G and Bailey, P (1959) Introduction to
Stereotaxis with an Atlas of the Human Brain. Thieme,
Stuttgart.
Siemionow, V, Yue, GH, Barnett, GH, Sahgal, V and
Heilbrun, MP (2000) Measurement of tissue electrical
impedance confirms stereotactically localized internal
segment of the globus pallidus during surgery. J. Neurosci. Meth., 96: 113117.
Slavin, K and Holsapple, J (2004) Microelectrode techniques:
equipment, components and systems. In: Z Israel and
K Burchiel (Eds.), Miroelectrode Recording in Movement
Disorder Surgery. Thieme, New York, NY, pp. 1427.
Smeding, HM, Speelman, JD, Koning-Haanstra, M, et al.
(2006) Neuropsychological effects of bilateral STN
stimulation in Parkinson disease: a controlled study.
Neurology, 66: 18301836.
Starr, PA, Christine, CW, Theodosopoulos, PV, et al.
(2002) Implantation of deep brain stimulators into the
subthalamic nucleus: technical approach and magnetic
resonance imaging-verified lead locations. J. Neurosurg., 97: 370387.
Starr, PA, Theodosopoulos, PV and Turner, R (2003) Surgery
of the subthalamic nucleus: use of movement-related
neuronal activity for surgical navigation. Neurosurgery,
53: 11461149; discussion 1149.
Tasker, RR (1998) Deep brain stimulation is preferable to
thalamotomy for tremor suppression. Surg. Neurol.,
49: 145153; discussion 153144.
171
Tavella, A, Bergamasco, B, Bosticco, E, et al. (2002) Deep

brain stimulation of the subthalamic nucleus in
Parkinsons disease: long-term follow-up. Neurol. Sci.,
23(Suppl. 2): S111S112.
Vitek, JL, Bakay, RA, Hashimoto, T, et al. (1998) Microelectrode-guided pallidotomy: technical approach and
its application in medically intractable Parkinsons disease. J. Neurosurg., 88: 10271043.
Volkmann, J and Benecke, R (2002) Deep brain stimulation
for dystonia: patient selection and evaluation. Mov. Disord., 17(Suppl. 3): S112115.
Volkmann, J, Moro, E and Pahwa, R (2006) Basic algorithms for the programming of deep brain stimulation
in Parkinsons disease. Mov. Disord., 21(Suppl. 14):
S284289.
Voon, V, Saint-Cyr, J, Lozano, AM, Moro, E, Poon, YY and
Lang, AE (2005) Psychiatric symptoms in patients with
Parkinson disease presenting for deep brain stimulation
surgery. J. Neurosurg., 103: 246251.
Voon, V, Kubu, C, Krack, P, Houeto, JL and Troster, AI
(2006) Deep brain stimulation: neuropsychological and
neuropsychiatric issues. Mov. Disord., 21(Suppl. 14):
S305327.
Welter, ML, Houeto, JL, Tezenas du Montcel, S, et al.
(2002) Clinical predictive factors of subthalamic
stimulation in Parkinsons disease. Brain, 125:
575583.
Wyler, AR, Fetz, EE and Ward, AA, Jr. (1975) Firing
patterns of epileptic and normal neurons in the chronic
alumina focus in undrugged monkeys during different
behavioral states. Brain Res., 98: 120.
Zonenshayn, M, Rezai, AR, Mogilner, AY, Beric, A,
Sterio, D and Kelly, PJ (2000) Comparison of anatomic
and neurophysiological methods for subthalamic
nucleus targeting. Neurosurgery, 47: 282292; discussion 292284.

M.R. Nuwer (Ed.)
172
CHAPTER 10
Visual evoked potentials during surgery

Donald York*
Department of Neuroscience, St. Johns Mercy Medical Center, St. Louis, MO 63141, USA
10.1. Use of VEPs to predict visual function

The visual evoked potential (VEP) recorded from the
occipital scalp reflects activation of fibers arising
from the area 810 central to the macular region
of the retina (Borda, 1977; Fuller and Hutton, 1990).
VEPs are sensitive to vascular insults that affect the
visual pathway. Transient ischemic episodes involving
the carotid artery are associated with amplitude reductions of a flash-evoked VEP to the ipsilateral eye
(Vaughan and Katzman, 1964). Degeneration of the macula typically results in abnormalities in VEPs, although
diseases such as retinitis pigmentosa, which affect the
peripheral retina initially may have a normal VEP. A
VEP demonstrating a complete absence of an occipital
response, despite the presence of a normal electroretinogram (ERG), indicates damage of the optic nerve or conduction pathway (Feinsod and Auerbach, 1973).
VEPs have been used for many years to compare
preoperative and postoperative visual function. For
example, Mizota et al. (1988) used pattern VEPs and
ERGs to compare pre- and postoperative findings in
patients with orbital tumors. Obstruction of the light
pathway to the retina, due to an intensive vitreous hemorrhage, for example, has been shown to result in a
indecipherable flash-evoked VEP, which may be
reversed following surgery, and may accompany
increases in visual acuity (Kellner and Foerster, 1996).
In subjects with pituitary tumors, other investigators have seen significant clinical improvement when
comparing their vision before and after surgery.
Heinemann et al. (1998) and Pojda-Wilczek et al.
(2000) showed abnormalities in VEP including
decreased visual acuity and visual field defects in
patients before surgery. Following removal of pituitary tumors which had compressed patients optic
*
Correspondence to: Donald York, Ph.D., Department of
Neuroscience, St. Johns Mercy Medical Center, St. Louis,
MO 63141, USA.
E-mail: dyork01@earthlink.net (D. York).
tracts, results indicated significant improvements in

P100 amplitudes and decreased P100 latencies. No
improvement of vision occurred in the patients whose
visual pathway had been damaged during the operation, such as with the removal of meningiomas and
gliomas lying in the immediate vicinity of an optic
nerve or optic chiasm. Gokalp et al. (1992) reviewed
50 transsphenoidal operations of pituitary adenomas
and also concluded that, following surgery, both conventional neurophthalmological tests and VEP values
improved due to decompression of optic nerves by
tumor removal.
Comparisons of flash-evoked VEPs, light-emitting
diode, and pattern VEPs have demonstrated similarities in waveform features (Hughes et al., 1989;
Tsaloumas et al., 1994). There are also clinical studies
which have demonstrated that changes in flash-evoked
VEPs occur within seconds of surgical and/or medical
decompression of intracranial pressure (York et al.,
1981, 1984; Zaaroor et al., 1993; Sjostrom et al.,
1995). Further strengthening the decompression theory, studies have also shown that VEP changes were
reversible by filling the space occupied by excised
lesions (Christophis, 1998). Testing of visual pathway
function in severely injured patients, or even during
craniomaxillofacial reconstructions, using VEPs and
ERGs has been shown to be a reliable method of
verifying the presence or absence of visual pathway
function regardless of any pathology (Gellrich, 1999).
Thus, there is evidence to suggest that the visual system can be successfully evaluated with VEPs between
the preoperative and postoperative stages. More importantly, postoperative improvements in VEPs correlate
directly with improvements in visual function.
10.2. Flash-evoked VEP methods
Normative studies using flash-evoked VEPs have
been undertaken with a variety of scalp montages
and stimulus parameters. Halliday (1993) reported
normative data for 17 subjects demonstrating a
173
Table 1
Visual evoked potentials (VEPs): normative studies database
Study
V APRE 94
AABRE 92
AABRE 90
AABRE 89
AABRE88
HC88
HC2 87
HC187
Number of
subjects
Electrode site
23
25
22
19
17
16
15
13
02-A2
02-A2
02-A2
Cz-A1
Cz-A1
Cz-A1
Cz-A1
Cz-A1
N70
P100
Amplitude (V)
10.5
9.4
16.2
17.1
15.0
15.0
17.8
8.8

5.4
4.6
10.0
8.7
6.2
8.5
8.7
2.8
P100 at 107 15.3 ms and a N70 at 72.9 15.8 ms.

Table 1 shows flash-evoked VEP amplitude and
latency data for 150 normal subjects from 8 different
studies performed over 7 years.
Scalp-recording montages of O1, Oz, and O2
referenced to the earlobe have been successfully
used. Recordings from the vertex (Cz) referenced to
the earlobe have also been used (York et al., 1981)
based on early studies which demonstrated very large
well-defined VEP waveforms from vertex recordings
(Allison et al., 1977). Through the studies, there were
no shared preferences for electrode sites, nor any
shared relationship found regarding the consistency
or reproducibility of flash-evoked VEP waveforms.
Delivering the flash stimulus relatively slowly at 1
flash per 23 s is more successful than faster rates. Also,
limiting the total number of flashes to 1015 per trial
has the advantage of preventing amplitude decrement
which may occur with large numbers of stimuli and artifact averaging in the operating room. A gain of 20 mV/
div with filters set at 170 Hz is sufficient to capture
important frequency components in the VEP.
Cataracts do not result in major waveform alterations to flash-evoked VEPs, although visual acuity
Latency (ms)
71.7
72.0
75.9
79.6
78.8
81.8
77.4
84.5

7.5
9.0
7.4
10.9
7.4
6.5
8.7
5.8
Amplitude (V)
11.4
18.8
25.0
27.5
26.8
27.5
27.1
14.5

6.1
11.6
11.1
11.6
7.9
7.8
9.1
5.1
Latency (ms)
95.1 6.3
97.0 11.0
98.2 8.1
96.0 12.4
100.8 7.8
101.0 5.9
99.2 8.1
100.7 6.2
may be markedly reduced to only the perception of

light (Halliday, 1978).
10.3. Anesthetic effects on flash-evoked VEPs
Many investigators in the 1980s believed that VEPs
could not be successfully performed in the operating
room (Cedzich et al., 1988). This was probably due
to the marked variability in waveform amplitude and
latency both between trials for a given patient, as well
as across different patients. One explanation is that the
variability was primarily due to the former use of halogenated gases as a component of anesthesia (Cedzich
et al., 1987). Not surprisingly, in the 1990s, with the
widespread use of propofol in neurosurgical anesthesia, the use of VEPs in the operating room became less
problematic.
In a study comparing flash-evoked VEPs in spine
patients before and after anesthesia with propofol/
oxygen/fentanyl (Table 2), it can be seen that the
major effect of this anesthetic is to cause a decrease
in the amplitude of P100. This is predominantly seen
at the occipital recording electrode but not from the
vertex. A similar change in response is seen with
Table 2
Compare before and after anesthesia
Amplitude (mV)
P100
Unanesthetized
Anesthetized
*P < 0.0001; single eye stim.
N70
O2
Cz
O2
Cz
12.4 1.6*
4.6 1.2*
4.9 1.2
5.7 2.3
7.3 1.4*
4.2 1.7*
4.5 0.7
5.0 2.6
174
D. YORK
Table 3
VEP anesthetized: normative database
Amplitude (V)
N70
Mean
S.D.
P100
Cz
Oz
O2
Cz
Oz
O2
7.5
4.1
8.4
5.8
6.6
5.3
9.3
5.7
9.7
5.4
8.2
5.8
Latency (ms)
N70
Mean
S.D.
P100
Cz
Oz
O2
Cz
Oz
O2
68.7
12.0
67.2
10.6
71.8
11.3
100.4
14.9
101.1
10.6
99.8
9.8
n 18; stimulus flash goggles, 0.5 s, 10 stimuli/trial; >6 trials/patient; 1 trial/23 min.
the N70 amplitude; yet, even with a P100 amplitude

reduction compared to the awake state, the amplitude
remains sufficiently large to serve as a marker of
visual pathway function.
Table 3 shows the average amplitudes and latencies of flash-evoked VEPs in 18 anesthetized spine
patients who have normal visual function. The P100
amplitude obtained at O2 with monocular stimulation
varied from 2 to 8 mV in repeated trials within a single patient. The P100 latency obtained at O2 varied
from 2 to 10 ms in repeated trials within a single
patient. There was no significant effect of the number
of repeated trials on the P100 amplitude. The mean
amplitude at each electrode site could be established
within four trials in all patients evaluated, and it was
not different from the mean amplitude across all
trials within the same patient.
An incidental finding was that female patients had a
significantly larger P100 amplitude (12.6 5.7 mV)
obtained at Oz with monocular stimulation than male
patients (6.7 2.9 mV). This sex-difference has also
been reported for pattern-shift VEPs (Halliday, 1993).
10.4. Using flash-evoked VEPs to monitor in the
operating room
The use of VEPs in the operating room now poses new
challenges. The new objective is to determine whether a
flash-evoked VEP can be successfully used in the
operating room with prognostic value. Specifically, it
would be useful to know if macular vision is preserved

in spite of changes seen in amplitude or latency during
the course of an operative procedure.
The use of pattern shift as a visual stimulus cannot
be applied in the operating room because a subject,
under general anesthesia, cannot voluntarily fixate
on a distant point.
There have been many versions of delivering flashevoked stimuli to an unconscious anesthetized patient.
Early versions used a strobe light directed at the
patients eyes. Most of the current versions consist of
light-emitting diodes mounted to a pair of goggles that
are placed over the patients eyes. The technical difficulty with this method is to deliver a flash of appropriate brightness that will be able to penetrate the
eyelids. However, this may not be a major obstacle,
since the decrease in ambient light caused by a closed
eyelid likely causes the pupil to dilate.
Pupillary dilators have been found to facilitate the
delivery of the visual stimulus when the patient is
under anesthesia (Smith, 1975). A 30% increase in
VEP amplitude has been demonstrated with pupillary
dilation. Also, it is known that the VEP is not
changed by deliberate decreases in blood pressure
down to 80 mm Hg, although below that level, there
is a progressive fall in amplitude (Smith, 1975).
10.5. Types of cases to be monitored
Flash-evoked VEPs have been used to evaluate the
visual pathway and offer a prognostic indicator of
visual function. The efficacy of this prediction among

the types of surgery in which it has been used is questionable, and it is not clear whether macular vision and
peripheral vision can be assessed equally with flashevoked VEPs.
Tumors along the visual pathway including the
orbit, perisellar region, intraventicular area, and occipital cortex have been monitored with VEPs (Cedzich
et al., 1987). VEP has also been used to monitor optic
nerve function during intraorbital surgery (Harding
et al., 1990). Several studies have focused uniquely
on sellar and perisellar tumors (Lorenz and Renella,
1989; Oliushin et al., 1990; Gokalp et al., 1992;
Chacko et al., 1996). In one study, VEP monitoring
in transsphenoidal tumor removal has demonstrated a
significantly greater improvement in field defects than
in a control group who did not receive VEP monitoring, although the study found no difference in postoperative improvement in visual acuity between the two
groups (Chacko et al., 1996). VEPs have also been
used to monitor posterior fossa and occipital lobe
tumors (Christophis et al., 1994; Curatolo et al.,
2000). Flash-evoked VEPs have also been used
to monitor the eye during endoscopic sinus surgery
(Herzon and Zealear, 1994; Hussain et al., 1996).
Monitoring of VEPs has also been useful in identifying
the optic tract in pallidotomy cases for Parkinsons disease (Tobimatsu et al., 1997; Yokoyama et al., 1999).
Perhaps one of the lesser known uses for VEPs has
been in the identification of patients who awoke from
a long-duration spine surgery with compromised
vision. Of 290 neurosurgeons polled at the turn of
the century, 8.3% reported a case of blindness in a
spine surgery case (Cheng et al., 2000). However, in
a 20-year retrospective review of 14,102 spine surgeries in 2005, only 4 cases were identified (Chang
and Miller, 2005). Although VEP monitoring has not
been extensively used to monitor spine cases at present, the visual loss associated with these surgeries typically present like an ischemic optic neuropathy,
which may be related to patients duration in prone
positions, with significant blood loss and hypotension.
The etiology appears to involve an increase in intraocular pressure associated with hypotension in that position. Perhaps some of these cases could have been
prevented by early warning changes in VEPs.
10.6. Criteria for significant change
In the presence of stable intraoperative VEP potentials,
what specific criteria can be used to predict
175
postoperative deficits? Under propofol anesthesia, it

is relatively easy to obtain stable, well-defined flashevoked VEPs using the strobe-light protocol described
above. In cases involving transsphenoidal pituitary
tumor removal, when the P100 amplitude variability
was defined within 1.5 S.D. of the mean baseline
anesthetized values, it has been found that this was
not associated with any postoperative visual deficits.
However, there are few documented cases involving
postoperative compromised vision in which specific
intraoperative changes in VEPs have been described.
10.7. Conclusions on efficacy of VEPs in surgery
Can VEPs be successfully recorded in the anesthetized patient? The answer is yes, given appropriate
anesthesia considerations. The variability of amplitudes within a case is no greater than that seen when
monitoring electromyography (EMG) associated with
transcranial motor evoked potentials. Thus, the question is, what constitutes a significant change in VEP?
From this review, it appears that increases in VEP
latency may indicate optic nerve compression, but
that changes in P100 amplitude are generally not
found to be reliable proctors of outcome.
What intraoperative interventions can be utilized
if a significant change in VEPs occurs? The answers
to this question remain elusive. Some investigators
have concluded that flash-VEP monitoring is not specific for visual acuity and has not proven helpful as
an intraoperative warning system (Cedzich and
Schramm, 1990). However, these conclusions were
reached during a period when gas was the primary
choice for anesthesia.
Other investigators have certainly found monitoring of VEPs during surgery to be beneficial. Although
VEPs have intermittently been abolished without correlation with surgery outcomes, the absence of a previously normal VEP for more than 4 min during surgical
manipulation within the orbit did show a correlation
with postoperative impairment of vision (Harding
et al., 1990).
An increase in P100 latency was found to be an
indicator of optic nerve compression when used in
endoscopic sinus surgery, although changes in P100
amplitude were not found to be useful (Hussain et al.,
1996). In monitoring transsphenoidal pituitary surgery, patients who had received VEP monitoring had
a significantly greater improvement in field defects
than those who had not been monitored (Chacko
et al., 1996).
176
Using VEPs to identify the optic tract in pallidotomy cases has been of major benefit in lesion placements without visual compromise (Tobimatsu et al.,
1997; Yokoyama et al., 1999).
References
Allison, T, Matsumiya, Y, Goff, GD and Goff, WR (1977)
The scalp topography of human visual evoked potentials. Electroencephalogr. Clin. Neurophysiol., 42:
185197.
Borda, RP (1977) Visual evoked potentials to flash in the
clinical evaluation of optic pathways. In: JE Desmedt
(Ed.), Visual Evoked Potentials in Man: New Developments. Clarendon Press, Oxford, pp. 481489.
Cedzich, C and Schramm, J (1990) Monitoring of flash
visual evoked potentials during neurosur-gical operations. Int. Anesthesiol. Clin., 28: 165169.
Cedzich, C, Schramm, J and Fahlbusch, R (1987) Are
flash-evoked visual potentials useful for intra-operative
monitoring of visual pathway function? Neurosurgery,
21(5): 709715.
Cedzich, C, Schramm, J, Mengedoht, CF and Fahlbusch, R
(1988) Factors that limit the use of flash visual evoked
potentials for surgical monitoring. Electroencephalogr.
Chacko, AG, Babu, KS and Chandy, MJ (1996) Value of
visual evoked potential monitoring during transsphenoidal pituitary surgery. Br. J. Neurosurg., 10:
275278.
Chang, SH and Miller, NR (2005) The incidence of vision
loss due to perioperative ischemic optic neuropathy
associated with spine surgery. Spine, 30: 12991302.
Cheng, MA, Sigurdson, W, Tempelhoff, R and Lauryssen,
C (2000) Visual loss after spine surgery: a survey. Neurosurgery, 46: 625631.
Christophis, P (1998) Visual evoked potentials as an indicator of supratentorial and infratentorial herniation.
Zentralbl. Neurochir., 59: 256262.
Christophis, P, Klug, N and Csecsei, G (1994) Changes in
visually evoked potentials in space-occupying processes
of the posterior fossa on the problem of ascending
transtentorial herniation. Zentralbl. Neurochir., 55:
9195.
Curatolo, JM, Macdonell, RA, Berkovic, SF and Fabinyi, GC
(2000) Intraoperative monitoring tp preserve central visual
fields during occipital corticectomy for epilepsy. J. Clin.
Neurosci., 7: 234237.
Feinsod, M and Auerbach, E (1973) Electrophysiological
examination of the visual system in the acute phase after
head injury. Eur. Neurol., 9: 5664.
Fuller, DG and Hutton, WL (1990) Prediction of postoperative vision in eyes with severe trauma. Retina,
10(Suppl. 1): S2034.
D. YORK
Gellrich, NC (1999) Controversies and current status of therapy of optic nerve damage in craniofacial traumatology
and surgery. Mund Kiefer Gesichtschir., 3(4): 176194.
Gokalp, HZ, Egemen, N, Culcuoglu, A, Naderi, S and Zorlutuna, A (1992) The use of Nd:YAG laser in pituitary
surgery and evaluation of visual function by visual
evoked potential (VEP). Neurosurg. Rev., 15: 193197.
Halliday, AM (1978) New developments in clinical application of evoked potentials. In: WA Cobb and H van Duijn
(Eds.), Contemporary Clinical Neurophysiology. Elsevier,
Halliday, AM (1993) The visual evoked potential in
healthy subjects. In: AM Halliday (Ed.), Evoked Potentials in Clinical Testing. Churchill Livingstone, London,
2nd ed, pp. 320324.
Harding, GF, Bland, JD and Smith, VH (1990) Visual
evoked potential monitoring of optic nerve function
during surgery. J. Neurol. Neurosurg. Psychiatry, 53:
890895.
Heinemann, B, Weik, R, Steudel, WI and Ruprecht, KW
(1998) Neurophysiologic follow-up in patients with
hypophyseal adenoma operations. Ophthalmologie, 95:
2832.
Herzon, GD and Zealear, DL (1994) Intraoperative monitoring of the visual evoked potential during endoscopic sinus
surgery. Otolaryngol. Head Neck Surg., 111: 575579.
Hughes, JR, Fino, JJ and Hart, L (1989) The visual evoked
potentials to the light emitting diode compared to the
flash and pattern reversal stimulus. Int. J. Neurosci.,
47: 359366.
Hussain, SS, Laljee, HC, Horrocks, JM, Tec, H and Grace,
AR (1996) Monitoring of intra-operative visual evoked
potentials during functional endoscopic sinus surgery
(FESS) under general anaesthesia. J. Laryngol. Otol.,
110: 3136.
Kellner, U and Foerster, MH (1996) Falsely nonrecordable
flash visual evoked cortical potentials in a diabetic eye
with severe vitreous hemorrhage. Ger. J. Ophthalmol.,
5: 2325.
Lorenz, M and Renella, RR (1989) Intraoperative monitoring: visual evoked potentials in surgery of the sellar
region. Zentralbl. Neurochir., 50: 1215.
Mizota, A, Adachi-Usami, E, Kakisu, Y and Asanagi, K
(1988) Pre- and post-operative findings of pattern VECPs
and ERGs in patients of orbital tumors. Doc. Ophthalmol.,
69: 4149.
Oliushin, VE, Tigliev, GS, Vainshtein, LG, Merkin, VM
and Sultanov, MM (1990) Monitoring control of the
visual evoked potentials during the surgical treatment
of tumors of the chiasmal-sellar area. Zh. Vopr. Neirokhir. Im. N N Burdenko, 3: 2224.
Pojda-Wilczek, D, Pojda, SM, Hendryk, S, Herba, E,
Zatorska, B and Jochan, K (2000) Visual system function in patients after surgery for intracranial tumors.
Neurol. Neurochir. Pol., 34: 11731186.

Sjostrom, A, Uvebrant, P and Roos, A (1995) The lightflash-evoked response as a possible indicator of
increased intracranial pressure in hydrocephalus. Childs
Nerv. Syst., 11: 381387.
Smith, B (1975) Anesthesia for orbital sugery: observed
changes in visual evoked response at low blood pressures. Mod. Probl. Opthalmol., 14: 457459.
Tobimatsu, S, Shima, F, Ishido, K and Kato, M (1997) Visual
evoked potentials in the vicinity of the optic tract during
stereotactic pallidotomy. Electroencephalogr. Clin. Neurophysiol., 104: 274279.
Tsaloumas, MD, Good, PA, Burdon, MA and Misson, GP
(1994) Flash and pattern visual evoked potentials in the
diagnosis and monitoring of dysthyroid optic neuropathy.
Eye, 8(Pt 6): 638645.
177
Vaughan, HG and Katzman, R (1964) Evoked response in

visual disorders. Ann. N. Y. Acad. Sci., 112: 305319.
Yokoyama, T, Suigiyama, K, Nishizawa, S, Yobota, N,
Ohta, S, Yamanoto, S and Uemura, K (1999) Visual
evoked oscillatory responses of the human optic tract.
York, DH, Pulliam, MW and Rosenfeld, JG (1981) Relationship between visual evoked potentials and intracranial pressure. J. Neurosurg., 55: 909916.
York, DH, Legan, M, Brenner, S and Watts, C (1984) Further studies with a non-invasive method of intracranial
pressure estimation. Neurosurgery, 14: 246461.
Zaaroor, M, Pratt, H, Feinsod, M and Schacham, SE (1993)
Real-time monitoring of visual evoked potentials. Isr. J.
Med. Sci., 29: 1722.
Section II.1
Somatosensory Evoked Potentials

M.R. Nuwer (Ed.)
180
CHAPTER 11
Somatosensory evoked potential monitoring with

scalp and cervical recording
Marc R. Nuwer* and James W. Packwood
Department of Clinical Neurophysiology, UCLA School of Medicine and UCLA Medical Center,
Los Angeles, CA 90095, USA
The most common technique for spinal cord monitoring uses somatosensory evoked potentials (SEPs)
with stimulation at the ankles and recording over
the neck and scalp. A similar technique is used for
monitoring the intracranial lemniscal sensory system
as it traverses the brainstem and cerebral hemispheres. For cervical or intracranial procedures, wrist
stimulation often substitutes for or supplements the
ankle stimulation techniques.
Monitoring provides services beyond simply warning of complications. With monitoring, a surgeon can
feel reassured about spinal cord or lemniscal sensory
pathway integrity, and therefore, extend the surgical
procedure to a greater degree than would have been
done without monitoring. Some patients are eligible
to undergo procedures with monitoring when they
may have been denied surgery in the past because of
a high risk of neurologic complications. Patients and
families can be reassured that certain feared complications are screened for during surgery. As such, monitoring provides an added dimension to surgical cases
even when the SEP itself is unchanged throughout
the procedure.
11.1. Techniques
Anklewrist to scalpneck SEP monitoring techniques are adapted from SEPs as used commonly in
the outpatient testing. These techniques should be
reasonably familiar to technologists and neurophysiologists who conduct routine outpatient evoked
potential (EP) testing.
*
Correspondence to: Marc R. Nuwer, M.D., Ph.D., Department of Clinical Neurophysiology, UCLA School of Medicine, Reed Research Building, Room 1-194, 710 Westwood
Plaza, Los Angeles, CA 90095-6987, USA.
Tel.: +1-310-206-3093; fax: +1-310-267-1157.
E-mail: MRN@UCLA.edu (M.R. Nuwer).
Usually, these techniques require about 300 trials

(individual stimulations) to produce a well-defined
EP tracing suitable for measurement and comparison
to baseline. That could be accomplished ideally in
1 min. In the more realistic setting, recording is interrupted by electrocautery artifact, movement, muscle,
and other artifacts, resulting in many rejected trials or
the need to start over for a heavily contaminated
tracing. In this more realistic setting, a new EP can
be generated in several minutes except when the surgeon uses electrocautery or produces other artifacts
frequently or continuously. Table 1 summarizes common simple techniques for SEP monitoring.
Table 1
Summary of techniques for SEP monitoring in surgery
Stimulation
Lower extremity: posterior tibial nerve at the ankle
Upper extremity: median or ulnar nerve at the wrist
Intensity to cause a 12 cm movement
Stimulate at 5.1/s/nerve, adjust rate as needed
Recording
Thoracic monitoring with lower extremity stimulations
CSp5forehead
Cz0 forehead
C30 C40
Cervical or intracranial monitoring with upper extremity stimulations
Erbs ipsilateralforehead
Earforehead
Cz0 forehead
C30 C40
Filters 30 and 3,000 Hz, notch filter off
300 or more trials per EP
Criteria for alarm
50% drop in amplitude
510% increase in latency
SOMATOSENSORY EVOKED POTENTIALS
11.2. Stimulation
11.2.1. Sites
Lower extremity stimulation is delivered to the posterior tibial nerve at the ankle. The nerve is found
easily at the ankle where it passes superficially just
posterior to the medial malleolus. That site has the
advantages that it can be accessed easily if needed
during the operation. Stimulation there causes a relatively small movement when neuromuscular blockade is incomplete. The desired movement is flexion
of the foot and toes. Its scalp EPs are produced reasonably well in most patients. The ideal stimulation
site is posterior to or slightly above the level of the
malleolus, so as to stimulate both the medial and lateral plantar terminal nerve branches. Stimulation too
distally may catch only one terminal branch.
The alternate lower extremity site is the common
peroneal nerve. Like the posterior tibial nerve, this is
a mixed motor nerve, that is, with both sensory and
motor components. Its sensory components project
up the lemniscal system to the cerebral hemispheres.
The peroneal nerve is stimulated where it runs superficially over the fibula immediately distal to the fibular
head, which is just distal to the lateral knee. Stimulation there produces dorsiflexion at the ankle. The peroneal nerves sensory distribution the lateral lower
leg and dorsal foot has a smaller cortical representation than the posterior tibial nerves plantar sensory
distribution, because the plantar foot is a more sensitive location. Therefore, peroneal stimulation may
produce on average a smaller cortical EP. But patients
with a peripheral neuropathy may have a much smaller
posterior tibial cortical SEP because of the damage to
distal sensory fibers. The peroneal nerve may be more
normal in many such patients, because it is more proximal. It should be considered a suitable alternative
lower extremity site among patients in whom the posterior tibial site produces unsatisfactory EPs.
The best upper extremity stimulation site often is
the median nerve at the wrist. This nerve is found
in the middle of the volar wrist. This serves a sensory
distribution that includes at least the thumb, index,
and middle fingers, and the lateral two-thirds of the
palm. Stimulation there produces thumb abduction.
Its sensory areas are very well represented at the
cerebral cortex because of the importance of human
hand function. This produces an excellent, easily
obtained cortical EP in addition to well-defined
peaks at cervical, subcortical, and plexus levels. It
181
is well suited as a control channel in thoracolumbar

spine cases. It is used often as the primary SEP monitoring channel for upper cervical and intracranial
lemniscal sensory monitoring. However, it enters
the spinal cord too rostrally for detection of some distal cervical impairment.
For middle to lower cervical spinal cord monitoring, the ulnar nerve is a stimulation site more suitable
than the median nerve. The ulnar nerve is stimulated
at the medial volar wrist just above the boney ulnar
head. Stimulation there produces fifth finger and lateral hand movement. Its sensory distribution includes
the lateral hand and fifth finger, a sensory region
much smaller than that of the median nerve. As a
result, its cortical peak often is smaller than for the
median nerve technique.
Stimulation usually is delivered to both left and
right side nerves. Those stimuli are alternated rather
than delivered simultaneously. This allows simultaneous measuring of each lemniscal pathway separately. The latter is helpful in the rare situation in
which a significant change occurs only with
impairment of one half of the spinal cord (Lesser
et al., 1986; Molaie, 1986). Most modern operative
monitoring equipment allows for separate averaging
from each stimulation site.
For thoracic and lumbar cases, the posterior tibial
nerve technique usually is used. Median nerve stimulation also is used. The latter may detect systemic or
anesthetic-related changes in the cortical EPs, so they
are used as a kind of control. In those cases, the
median or ulnar nerve channels rarely detect an incidental brachial plexus impairment, for example, from
arm positioning.
For cervical cases, median or ulnar nerve pathways are monitored along with posterior tibial nerve
channels. This gives a good coverage of the many
different possible areas of impairment. The lower
extremity channels also cover in case of a high thoracic or very low cervical level injury.
11.2.2. Rate and intensity
Stimuli are delivered above the motor threshold sufficient to cause a 12-cm movement. This stimulation intensity can be determined prior to initiation
of neuromuscular blockade. Otherwise, a default
stimulus intensity value such as 20 mA is chosen.
The presence of the stimulus can be checked during
the case, even in the presence of neuromuscular
182
M.R. NUWER AND J.W. PACKWOOD
blockade, by noting the continuing presence of the

stimulus artifact in the recording channels.
Needle electrodes can be used for nerve stimulation. Two are placed along the course of the nerve.
This allows for secure location, and avoids the problems of slippage or drying of the conductive paste
during long cases. It also avoids changes in skin
resistance that can occur over many hours of long
cases. Some users prefer using traditional surface
disc electrodes or bar electrodes with two disc or
rectangular contacts. These would need to be secured
in place, but without pressure enough to cause skin
damage from prolonged pressure in long cases. For
surface electrodes, paste should be calcium-free to
avoid chemical skin burns from iontophoresis of
those ions into the skin.
Stimulation is delivered generally at approximately 5 stimuli/s (Nuwer and Dawson, 1984a). Faster stimulation would have been helpful to produce
EPs quicker in the operating room. However, faster
stimulation generally produces lower amplitude
EPs, making monitoring more difficult. Slower stimulation can produce larger EPs, but slows the speed
of creating new EP tracings. As illustrated in Fig. 1,
the best trade-off of these two factors is about 5 stimuli/s/nerve. For patients with particularly large EPs, a
EFFECTS OF STIMULUS RATE
Rate Amplitude
PERCENT OF MAXIMUM VALUE
100
Amplitude
25
1.1
3.1
5.1
7.1
9.1
11.3. Recording
Recording at multiple sites above the level of surgery
is helpful. In individual patients, it is difficult to predict in advance which specific sites will be most useful for monitoring. Some degree of flexibility of
technique is helpful. Availability of many recording
channels also helps greatly many modern systems
can track 1632 channels simultaneously.
Once monitoring under anesthesia is under way,
the monitoring neurophysiologist or technologist
can choose among the best recordable potentials
available for that particular patient. These optimal
recording sites and potentials can then be used for
the further monitoring procedure itself.
Near-field scalp recording channels are less
affected by background muscle and movement artifact. Far-field potentials recorded from the cervical
channel are less affected by changes in anesthetic concentrations. A combination of recording channels is
usually optimal, rather than monitoring few channels.
11.3.1. Lower extremity recording channels
75
50
slightly faster stimulation rate can be employed. For

patients with particularly small EPs, a slower stimulation rate is helpful. Stimulation rates avoid exact
multiples of 60 Hz (or 50 Hz), such as 5.00/s. That
helps to average out any residual 60 Hz (or 50 Hz)
artifact in the recording.
11.1
STIMULI PER (S)
Fig. 1. Effects of increasing the rate of stimulus presentation in one patient, for posterior tibial stimulation and
scalp recordings. As rate is increased, EP amplitude
decreases. The product rate amplitude helps compare
the advantageous increase in speed of testing and the disadvantageous loss of amplitude. In the patient, the rate
5.1/s appeared to be the best compromise between speed
and attenuation. (Reprinted from Nuwer and Dawson,
1984a, with permission from the International Federation
of Clinical Neurophysiology.)
With lower extremity stimulation, recordings are

taken from skin over the cervical spine and scalp vertex. These are designed to monitor far-field cervical
and brainstem activity, as well as near-field somatosensory cortical potentials.
The primary scalp active recording is made from
site Cz0 (Cz-prime). The prime mark indicates a
modified site located 2 cm posterior to the named
International 1020 system scalp site. This midline
site Cz0 often is the optimal placement for recording
for each of the left and right lower extremities.
Secondary scalp active recording sites are offcenter at C10 and C20 , or at C30 and C40 . These are
better recording sites than Cz0 in some patients.
Orientation of the active somatosensory cortical generator varies among patients, a point illustrated in
Fig. 2. Note that the dipole crosses the midline before
appearing at the scalp. This results in a paradoxical
localization of the primary cortical peak, which is
best seen over the scalp ipsilateral to the leg
stimulated.
183
is often over the second cervical spine (CSp2). These

are referenced to the forehead or shoulder. These channels are desirable because the recorded cervical and
brainstem peaks are less affected by anesthesia depth.
For cervical procedures, electrodes cannot be placed
over the cervical spine. Instead, subcortical peaks are
recorded from mastoid or ear electrodes.
A lumbar channel is used occasionally to ascertain
that the stimulus induced transmission arrived at the
lumbar spinal cord. For older or obese patients, these
lumbar potentials are difficult to record.
+
CZ
C4
C3
11.3.2. Upper extremity recording channels
CZ
C4
C3
B
Fig. 2. Hypothesized variations in the electrical field distribution of the early cortical potentials upon stimulation of
the posterior tibial nerve. These dipole orientation variations are based on the known variability of the location of
the leg area in the cortex in the interhemispheric fissure.
Notice how the positive peak generated in one hemisphere
can show up best at the scalp vertex as in (A), or on the
scalp overlying the other hemisphere as in (B). The latter
phenomenon has been called paradoxical localization.
(Reprinted from Seyal et al., 1983, with permission from
the International Federation of Clinical Neurophysiology.)
Several reference electrodes sites are in common

use. Many monitorists measure from Cz0 using a
forehead, ear, or mastoid reference site. For the offcenter scalp active sites, many users reference the
contralateral electrodes, that is, C10 C20 and C30
C40 . Short distances to the reference reduce noisiness
in channels due to distant sources. But short distances
also may reduce some peak amplitudes. The optimal
recording technique varies among patients.
The subcortical peaks are recorded from electrodes placed at the neck, ear, or mastoid. The neck site
With upper extremity stimulation, recordings are

taken from skin over the shoulder, cervical spine,
and scalp vertex. These are designed to monitor farfield cervical and brainstem activity, as well as
near-field somatosensory cortical potentials.
The primary scalp active recording sites are at C30
and C40 . Several reference electrode sites are in common use, including forehead, ear, or mastoid reference sites. Some users reference the contralateral
electrodes, that is, C30 C40 .
For some intracranial surgery, those sites may be
in the surgical field. In those cases, modified scalp
sites are needed. In some cases, subdural or direct
cortical electrodes are placed in the surgical field.
Keeping the latter electrodes in place throughout a
procedure requires tethering of the electrodes cable
with sutures or other ad hoc techniques.
The subcortical brainstem or cervical peaks are
recorded from electrodes placed at the neck, ear, or
mastoid. The neck site is often over the fifth cervical
spine (CSp5), directly over the median nerves cervical peak generator located in the spinal cord at the C5
level. These sites are referenced to the forehead or
contralateral shoulder. These channels are desirable
because the recorded cervical and brainstem peaks
are less affected by anesthesia depth, and to assure
monitoring of the peripheral portion of the pathway.
An Erbs point channel is sometimes used to ascertain that the stimulus-induced transmission arrived
at the proximal brachial plexus. Erbs point is located
above the clavicle, 2 cm lateral to the insertion of the
sternocleidomastoid muscle.
11.3.3. Filters
Recording filters are generally chosen at 30 Hz and
3 kHz. These filter settings are chosen to optimize
184
noise rejection while retaining the principal EP characteristics in a typical surgical setting. They also
minimize the ordinary background fluctuations in
EP due to minor changes in anesthetic depth (Nuwer
and Dawson, 1984a). Open filters (e.g., a low filter of
1 Hz) are more susceptible to variability due to background noise. The individual techniques used are
open to clinical choice by the neurophysiology monitoring team.
Figure 3 illustrates the effects of various filter settings. In this case, the optimal trade-off occurred for
a lower filter set at 30 Hz. Lower filter settings led to
a greater amount of background variability. Higher
settings of the low filter resulted in attenuation of
the desired EP.
Figure 4 shows tracings from scalp recordings
during a routine monitoring during surgery for scoliosis. This shows an example of stable recordings
from upper and lower extremity stimulation. Such
stable, low-noise, easily reproducible potentials are
desirable for spinal cord monitoring.
The 60 Hz (or 50 Hz) notch filter usually is kept
off. That filter can interact with the stimulus signal
itself, resulting in a sinusoidally decaying artifact.
The artifact has a 60 Hz (or 50 Hz) frequency, resulting in peaks at 16.6, 33, and 50 ms. Some users have
mistaken these artifacts for stable EPs, which do not
change during the monitoring period.
11.4. Peaks, generators, and alarm criteria

11.4.1. Lower extremity peaks
For posterior tibial nerve techniques, an N22 peak
appears as a negative polarity near-field potential
over the lumbar spinal cord at about 22 ms after the
stimulation. The lumbar spinal cord anatomically lies
around the T12 or L1 spine, so the potential is seen
best over those spines. See Fig. 5 for an illustration
of typical routine lower extremity SEP peaks and
their nomenclature.
A cervical peak is seen using far-field recording
from the neck region to a distant reference. The peak
often occurs around 30 ms. The generator is thought to
be the gracilis nucleus, one of the posterior column
nuclei at the cervicomedullary junction. The lemniscal
pathways synapse at that level. That peak may be followed by an opposite polarity trough that may represent
conduction up the brainstems medial lemniscus.
The P37 is generated at the primary somatosensory cortex. As seen in Fig. 2, it may show up at
the midline or off-center, paradoxically, on the side
of the leg stimulated. It may have a doubled negative
peak, which sometimes leads to peak-picking errors
by inexperienced users. When encountering the doubled peak morphology, a user might sometimes
choose the first, and at other times choose the second
EFFECTS OF VARIOUS FILTER SETTINGS
1.25 V
11K Hz
153K Hz
303K Hz
758K Hz
0.625 V
50
100 0
ms
50
100
Fig. 3. Effects of four different filter settings during intraoperative recordings in one patient, taken from a single scalp
channel (Cz channel). The four recordings were made simultaneously from the same data. Variability is greatest in the
1 Hz channel and is reduced by higher filter settings. Each pair of EPs is a typical set of two consecutive recordings.
The amplitude scale is doubled for the lower two EPs. (Reprinted from Nuwer and Dawson, 1984b, with permission from
Lippincott Williams and Wilkins.)
185
Normal Stable Somatosensory Evoked Potentials

Cervical-Fz
Cc-Fz
Ci-Cc
Cz-Fz
Cervical-Fz
Fig. 4. Normal stable SEPs. Note the good reproducibility for the tracings in each channel. The montage includes upper and
lower extremity tracings, collected separately for each limb. Cortical and subcortical (cervical) channels are used. (From
UCLA Department of Clinical Neurophysiology, with permission.)
peak, leading to an erroneous impression that the

latency has shifted. Watching the peak shape (morphology) can help to avoid peak-picking errors.
The primary measurements are the P37 peaks
amplitude and latency. Secondary measurements are
the amplitude and latency for the cervical peak.
Latencies should stay within about 510% of baseline values, for example, 34 ms of normal variation
for a 37-ms P37 peak. Amplitudes should stay within
50% of baseline values. Some users raise alarms at
30% decrease of the P37, since such a degree of
amplitude drop is uncommon among well-defined
P37 peaks. These latencies and amplitudes usually
are chosen after the patient has been under anesthesia
for 20 min, since there is a period of gradual amplitude loss and minor latency increase due solely to
the anesthetic itself. That effect takes 2030 min to
take effect after induction. Even during that time
period, though, monitoring needs to watch for early
deterioration due to positioning on the table or problems related to intubation. Anesthetic effect is distinguished from clinical pathway impairment by
observing the cervical peaks, since they are relatively

steady despite anesthesia. It is the P37 cortical peak
that is affected more greatly by the onset of anesthetic. This trick applies to spine surgery, but not to
intracranial surgery. The effects of anesthetics are
described in detail in another chapter.
11.4.2. Upper extremity peaks
For median or ulnar nerve techniques, an N9 peak
appears as a negative polarity near-field potential
over the proximal brachial plexus at about 9 ms after
the stimulation. The electrode best used to measure
this is at a place referred to as Erbs point, and the
peak itself sometimes is referred to as the Erbs point
peak. Fig. 6 illustrates typical routine upper extremity SEP peaks and their nomenclature.
A cervical peak is seen using far-field recording
from the neck region to a distant reference. The negative polarity peak often occurs around 13 ms. The
generator of this N13 is thought to be the cervical
spinal cord gray matter at the level where the nerve
186

POSTERIOR TIBIAL NERVE
N8
PF-K
N22
1.0 V/division
T12-Ic
C5Sp-Fz
Ci-Cc
Cz-Fz
P37
6 ms/division
Fig. 5. SEPs from posterior tibial nerve stimulation are

shown. Typical peaks N8, N22, and P37 are noted. A cervical peak was also found. These peaks have normal latencies
and amplitudes. (Reprinted from Nuwer et al., 1994, with
permission from the International Federation of Clinical
Neurophysiology.)
enters, synapses onto reflex arcs, and send axons rostrally into the posterior column. For the median
nerve, this is in the C5 spinal cord.
Shortly thereafter, a positive polarity peak is generated probably from the cuneate nucleus, one of the
posterior column nuclei at the cervicomedullary junction. This peak is referred to as the P14. The lemniscal pathways synapse at that level. That peak may be
followed by a negative polarity N18 that may represent conduction up the brainstems medial lemniscus
and into the thalamus. The N18 is a broad potential
that gradually decreases over about 10 ms. Both the
P14 and N18 are best seen as a far-field potential
seen broadly over the scalp using a non-cephalic reference. The N18 may be obscured by the N20 when
recording near that peaks active site.
The N20 is generated at the primary somatosensory cortex. This usually is a large well-defined peak
seen over the C30 or C40 scalp site. It is a near-field
potential, that is, its amplitude drops off when the
electrode is relocated even a short distance away
from the optimal recording site.
The primary measurements are the N20 peaks
amplitude and latency. Latencies should stay within
about 510% of baseline values, for example, 12 ms
Fig. 6. SEPs from median nerve stimulation are shown.

Typical peaks are shown in each of four recording channels. The test is normal. (Reprinted from Nuwer et al.,
1994, with permission from the International Federation
of Clinical Neurophysiology.)
of normal variation for a 20 ms N20 peak. Amplitudes

should stay within 50% of baseline values. Some users
raise alarms at 30% decrease of the N20, since such a
degree of amplitude drop is uncommon among welldefined N20 peaks. As with the lower extremity P37
peak, these latencies and amplitudes usually are chosen after the patient has been under anesthesia for
20 min. That effect takes 2030 min to take effect
after induction. The effects of anesthetics are
described in detail in another chapter.
11.5. Interpretation of changes: alarms and
considerations
Generally, monitoring teams use both latency and
amplitude criteria for raising an alarm. A 50% drop
in recorded potentials is generally considered to be
sufficient for raising an alarm (Nuwer, 1986; Nuwer
et al., 1995). Latency increases greater than 510%
also are cause for alarm. Latency measures need to
take into account temperature effects. Amplitude
measures need to take into account anesthetic effects
especially for cortical potentials. The initial period
187
Systemic factors can cause EP changes. Examples

are hypotension and hypoxia. Occasionally, such systemic changes are first detected by EP changes. A
patient with preexisting neurologic problems, such as
spinal cord compression, may be particularly sensitive
to otherwise minor degrees of hypotension. Surgical
problems causing EP attenuation include direct cord
trauma, excessive traction, blunt trauma, excessive
compression, stretching of the cord from spinal distraction, vascular insufficiency from compression, embolus, thrombus, or occasionally other clinical problems.
Anesthetic effects are common. Monitoring often
assesses not only the lower extremity SEPs, but also
the upper extremity SEPs even in a thoracic or lumbar procedure. This allows for an assessment of anesthetic effect. Usually, that affects all cortical
channels but causes little change in subcortical channels. See Fig. 7 for an example of an asymmetric
after anesthetic induction is accompanied by latency

and amplitude changes as the nervous system equilibrates over 2030 min. Experienced monitoring
teams learn how to factor out these extraneous causes
of variability and so minimize false alarms due to
such effects as boluses of medication, raised levels
of inhalation anesthetics, and decreasing temperature.
When changes occur, the monitoring team needs
to assess quickly whether these have a technical, systemic, or surgical cause.
Technical and systemic issues also can cause
change and need to be investigated. Technical problems can occur from the electrodes themselves, for
example, if they become disconnected. Equipment
can malfunction. A good stimulus artifact or motor
signs of acceptable stimulus help assure the adequacy
of a stimulus. Impedance checks and review of raw
input data can assure the recording systems integrity.
UNILATERAL CORTICAL EVOKED POTENTIAL LOSS

DUE TO ANESTHETIC DEPTH CHANGE
LEFT MEDIAN N.
CERVICAL
T
r
c
1
N20
N20
CORTICAL
2
N14
RIGHT MEDIAN N.
CERVICAL
P25
N14
P25
3
N20
CORTICAL
LEFT POSTERIOR TIBIAL N.

CERVICAL
5
P31
P25
P25
P31
N34
N45
N34
CORTICAL
N20
P18
P18
6
P37
RIGHT POSTERIOR TIBIAL N.

CERVICAL
P31
P31
N45
N34
CORTICAL
N34
N45
8
P37
BASELINE
P37
CHANGES
Fig. 7. The baseline testing shows a relatively attenuated left lower extremity cortical peak (left tracings). After an increase
in anesthetic depth (right tracings), that channel no longer shows a reliable EP. (The baseline is superimposed on the newly
acquired tracings at the right.) An anesthetic effect is the likely cause of the change, as suggested by both the preserved
subcortical peaks for the affected pathway and somewhat attenuated cortical peaks in all other pathways. (From UCLA
Department of Clinical Neurophysiology, with permission.)
188
anesthetic effect seen in the setting of a partial preexisting unilateral impairment in one pathway.
Occasional transient significant diminishing of
EPs can occur without placing the patient at significant risk for postoperative neurologic problems. Even
transient EP loss for a few minutes can occur without
substantially raising the risk of postoperative problems, especially if the patients EPs return shortly
thereafter to baseline levels of amplitude and latency.
If EPs abruptly or slowly diminish to beyond the criterion levels for raising an alarm, the patient is at risk
for postoperative new neurologic complications,
especially if the EPs remain attenuated and prolonged through the end of the case. The gravest situation is the complete loss of previously easily
detected EPs. Even in that circumstance, the risk of
new neurologic postoperative impairment may only
be about 5075% (Nuwer et al., 1995).
for many hour or days. In each delayed onset case,

the patient awoke from surgery initially with good
motor strength in the lower extremities. Motor function was lost several hours to several days later.
The scoliosis surgery multicenter outcome study
assessed the likelihood of false results. Those results
are shown in Table 2. False negative studies were
rare, and many of those were delayed onset. False
positive outcomes were more common. Equivocal
results are those with minor or transient changes.
Table 3 shows the sensitivity and specificity results
as determined in the same large study.
11.7. Conclusions
SEP monitoring can be conducted with ankle or wrist
stimulation and recording over the scalp and neck.
These techniques are familiar to neurophysiologists
and technologists who conduct routine outpatient
11.6. False results

Occasional false positive monitoring events are seen.
In a false positive event, SEPs show significant
changes but the patient develops no postoperative
deficits. Many false positive events may be true
physiologic clinical changes detected by monitoring.
In those cases, monitoring raises an alarm, the clinical team makes changes, and the patient has no neurological deficit because the interventions were
successful. Others may be false alarms due to technical or anesthetic issues.
False negative cases have been reported despite
stable EPs in some studies (Tamaki et al., 1984; Wilber et al., 1984; Ginsburg et al., 1985; Johnston et al.,
1986; Ben-David et al., 1987; Harper et al., 1988;
More et al., 1988). Some such patients had patches
of dysesthesia for several days or weeks postoperatively, possibly due to segmental spinal wiring. Rare
cases have had serious, long-lasting neurological
complications. Lesser and colleagues (Lesser et al.,
1986) reported six cases from four institutions. Each
suffered new postoperative neurologic deficits
despite stable intraoperative SEPs. In three of these
cases, the new impairment probably developed postoperatively, referred to as delayed onset deficits.
Such delayed onset cases may be due to latent effects
of vascular compromise or mechanical compression.
Some may have been exacerbated by patient movement after awakening. Alternately, some compression or ischemia may gradually result in permanent
impairment only if the condition remains in place
Table 2
Neurologic outcome prediction rates for SEP monitoring
in spinal surgery (Total Procedures: 51,263 (100%))
False-negative (FN) rate: neurologic postoperative deficits
despite stable SEPs:
Definite
34 (0.063%)
Equivocal
13 (0.025%)
Delayed onset
18 (0.035%)
Total
65 (0.127%)
False-positive rate: no neurologic deficits despite SEP
changes:
Definite
504 (0.983%)
Equivocal
270 (0.527%)
Total
774 (1.510%)
True-positive (TP) rate: neurologic deficits predicted by
SEP changes:
Definite
150 (0.293%)
Equivocal
67 (0.131%)
Total
217 (0.423%)
Neurologic deficits (FN plus TP):
Definite
184 (0.356%)
Equivocal
80 (0.156%)
Delayed onset
18 (0.035%)
Total
282 (0.550%)
True-negative rate: no neurologic deficit and stable SEPs:
Total
50,207 (97.94%)
These data are from the multicenter outcome study of SEP spinal
cord monitoring in scoliosis organized through the Scoliosis
Research Society (Nuwer et al., 1995). They were obtained from
153 US surgeon respondents. Note the very low rate of definite
false negative cases (0.063%).

Table 3
SEP monitoring validity measures
Sensitivity
417/451
Specificity
50,207/50,781
Positive predictive value
417/991
Negative predictive value
50,207/50,241
92%
98.9%
42%
99.93%
These outcome measures are from the multicenter outcome survey of

153 US surgeons organized through the Scoliosis Research Society
(Nuwer et al., 1995). The table shows a breakout of validity measures
from that survey. The very high negative predictive value here indicates the high reliability of the monitoring when the SEP remains
normal and stable. The outcome survey report (Nuwer et al., 1995)
discusses in detail these data and related assumptions.
EP testing. In the nearly 30 years since these techniques were developed, they have spread into common
use in surgery. They are appropriate for orthopedic,
neurosurgical, and cardiovascular procedures. False
negative cases are rare. False positive results occur
in a small portion of cases. With appropriate choices
of reference sites and filter settings, and with avoidance of excess inhalation anesthetic, suitable peaks
can be found in most patients.
References
Ben-David, B, Haller, G and Taylor, P (1987) Anterior spinal fusion complicated by paraplegia. A case report of a
false-negative somatosensory evoked potential. Spine,
12: 536539.
Ginsburg, HH, Shetter, AG and Raudzens, PA (1985) Postoperative paraplegia with preserved intraoperative
somatosensory evoked potentials. J. Neurosurg., 6:
296300.
Harper, CM, Daube, JR, Lichy, WJ and Klassen, RA
(1988) Lumbar radiculopathy after spinal fusion for scoliosis. Muscle Nerve, 11: 386391.
Johnston, CE II, Happel, LT, Jr., Norris, R, Burke, S, King,
AG and Roberts, JM (1986) Delayed paraplegia complicating sublaminar segmental spinal instrumentation. J.
Bone Joint Surg. Am., 68: 556563.
189
Lesser, RP, Raudzens, P, Lueders, H, Nuwer, MR, Goldie,
WD, Morris, HH, III, Dinner, DS, Klem, G, Hahn, JF,
Shetter, AG, Ginsburg, HH and Gurd, AR (1986) Postoperative neurological deficits may occur despite
unchanged intraoperative somatosensory evoked potentials. Ann. Neurol., 19: 2225.
Molaie, M (1986) False negative intraoperative somatosensory evoked potentials with simultaneous bilateral stimulation. Clin. Electroencephalogr., 17: 619.
More, RC, Nuwer, MR and Dawson, EG (1988) Cortical
evoked potential monitoring during spinal surgery: sensitivity, specificity, reliability, and criteria for alarm. J.
Spinal Disord., 1: 7580.
Operating Room, Raven Press, New York, 246 pp.
Nuwer, MR and Dawson, E (1984a) Intraoperative evoked
potential monitoring of the spinal cord: enhanced stability of cortical recordings. Electroencephalogr. Clin.
Nuwer, MR and Dawson, E (1984b) Intraoperative evoked
potential monitoring on the spinal cord: a restricted filter, scalp method during Harrington instrumentation for
scoliosis. Clin. Orthop., 183: 4250.
Nuwer, MR, Aminoff, M, Demedt, J, Eisen, AA, Goodin,
D, Matsuoka, S, Mauguie`re, F, Shibasaki, H, Sutherling,
W and Vibert, JF (1994) IFCN recommended standards
for short latency somatosensory evoked potentials:
report of an IFCN committee. Electroencephalogr. Clin.
Nuwer, MR, Dawson, EG, Carlson, LG, Kanim, LEA and
Sherman, JE (1995) Somatosensory evoked potential
spinal cord monitoring reduces neurologic deficits
after scoliosis surgery: results of a large multicenter
survey. Electroencephalogr. Clin. Neurophysiol., 96:
611.
Seyal, M, Emerson, RG and Pedley, TA (1983) Spinal and
early scalp-recorded components of the somatosensory
evoked potential following stimulation of the posterior
tibial nerve. Electroencephalogr. Clin. Neurophysiol.,
55: 320330.
Tamaki, T, Noguchi, T, Takano, H, Tsuji, H, Nakagawa, T,
Imai, K and Inoue, S (1984) Spinal cord monitoring as a
clinical utilization of the spinal evoked potentials. Clin.
Orthop., 184: 5864.
Wilber, RG, Thompson, GH, Shaffer, JW, Brown, RH and
Nash, CL, Jr. (1984) Postoperative neurological deficits
in segmental spinal instrumentation. J. Bone Joint Surg.
[Am.], 66A: 11781187.

M.R. Nuwer (Ed.)
190
CHAPTER 12
Somatosensory evoked potential monitoring

with dermatomal stimulation
Jefferson C. Slimp*
Department of Rehabilitation Medicine, University of Washington School of Medicine, Seattle, WA 98195, USA
12.1. Introduction
Dermatomal somatosensory evoked potentials (DSEPs)
are a means to assess segmental-level function of the
spinal cord in both the clinical and the operating room
settings. The technique is based on the concept that each
spinal nerve innervates a well-defined area of skin
called a dermatome. An underlying assumption of the
technique is that electrical stimulation of the skin may
be done so that the ascending afferent activity is mostly
confined to a single sensory nerve root. An evoked
potential may typically be recorded from the scalp, generated in the primary somatosensory cortex, or from the
spine reflecting activity in the posterior columns of
the spinal cord (Lee and Seyal, 1998). The quality of
the response indicates the integrity of the conduction
of the ascending pathway from the skin through the sensory root to the spinal cord and to somatosensory cortex.
Changes in a response, such as may occur in the
operating room, may indicate alterations in conduction
of a specific root (Cohen and Huizenga, 1988; Cohen
et al., 1991b; Terada et al., 1993; Jou, 2004; Tsai
et al., 2005). Localization of impairment to a specific
segmental level of the spinal cord or nerve root may
be strengthened by comparison of responses to adjacent
dermatomes (Slimp et al., 1992).
DSEPs were developed as a refinement of mixed
nerve SEPs (Green et al., 1983; Katifi and Sedgwick,
1986; Pop et al., 1988; Slimp et al., 1992). It has long
been recognized that nerve fibers from mixed nerves,
such as median, ulnar, or tibial nerves enter the spinal canal over several nerve roots. Consequently,
the SEPs from a mixed nerve do not represent a
*
Correspondence to: Jefferson C. Slimp, Ph.D., Neuromonitoring Program, Department of Rehabilitation Medicine, Box 356490, University of Washington School of
Medicine, Seattle, WA 98195, USA.
Tel.: +1-206-543-7066; fax: +1-206-685-3244.
E-mail: jcslimp@u.washington.edu (J.C. Slimp).
single segmental level of the spinal cord. To restrict

the input to a single segmental, SEPs were recorded
to stimulation of the skin along dermatomal patterns.
The concept of a dermatome and dermatomal maps
is a basic teaching point of a foundation course in neurophysiology. The mental image is created of a mammal
in the quadripedal position with irregular shaped stripes
organized in an orderly fashion from rostral to caudal,
much as a zebras stripes are organized, that represent
the skin segments innervated by each sensory root of
the spinal cord. While these texts may mention the possibility of overlap of dermatomes or that dermatomal
maps may vary, seldom does the discussion focus fully
on the historical data. The concept of the dermatome
dates to the mid-19th century (for review see Greenberg, 2003). Various methods, including dissection,
degeneration techniques, assessment of zone of remaining sensibility following root lesions, strychnine
application, electrical stimulation, ruptured disks, and
electrophysiological recordings have been applied to
determine dermatomal maps in humans and animals.
The first map in humans was by Sir Henry Head
(Head and Campbell, 1900) based on eruptions of herpes zoster. This was followed by Foersters map, that
was based on the remaining sensibility following
transaction of nerve roots for pain (Foerster, 1933).
Foersters method was similar to that in Sherringtons
studies of monkeys (Sherrington, 1893, 1898) and
was considered superior to herpes zoster outbreak
because there is no reason to expect that the virus
would affect every neuron of a ganglion or necessarily lead to an outbreak. Nevertheless, there was a
degree of similarity in the maps in that they showed
large, irregular-shaped dermatomes that in the peripheral extremity were not continuous with the midline.
Another map that is often quoted is that of Keegan
and Garret, who published their maps in 1948, based
on hypoesthesia related to ruptured disks (Keegan
and Garret, 1948). Their map, showing long narrow
bands stretching from the midline to the periphery, differed significantly from previous maps.
Several factors may account for the variability in
maps by different investigators. First is the variability
in dermatomal patterns between individuals in a
given study. For the same root, two individuals may
have significantly different dermatomal patterns to
the point of hardly overlapping. Each investigator
states or, at least, hints of individual variability in
dermatomal borders. Secondly, not only are borders
variable but there is clearly a large degree of overlap
of the borders of adjacent dermatomes. With each
clinical study and particularly with electrophysiological studies, there is no doubt that dermatomal
borders are not distinct but overlap with adjacent dermatomes (Dykes and Terzis, 1981). A third variability, which can confound an association of clinical
findings and anatomy, is the notion of prefixed or
postfixed nerve roots, in which the pattern of the spinal root contribution to the brachial or lumbar plexi
is shifted one or more levels rostral (prefixed) or
one level caudal (postfixed) than typically expected.
The frequency of occurrence has been estimated
at 1015% (Phillips and Park, 1991; Owen et al.,
1993). Another anatomical variant that may contribute to dermatomal variability is conjugated or anastomosis of nerve roots (Boyer et al., 1981; Maiuri and
Gambardella, 1989; Moriishi et al., 1989). These
variants may occur with regular frequency. A fourth
reason for variability in dermatomes comes from the
work of Denny-Brown and colleagues, who demonstrated in monkeys that physiological connections in
the spinal cord influence the size of a dermatome for
a given root through inhibitory spinal mechanisms
(Denny-Brown and Kirk, 1968; Kirk and DennyBrown, 1970; Denny-Brown et al., 1973). They used
the Sherrington technique of remaining sensibility to
isolate a nerve root by sectioning three roots above
and below the root. The sensate skin was labeled
the dermatome for that root. The dermatome was
stable over time. However, if after a time, additional
roots on each side of the isolated root were cut,
the dermatome increased in size by as much as
twofold. The effects seem to be mediated by the tract
of Lissauer that routes information rostrally and caudally in the dorsal horn of the spinal cord. Additionally, they showed that subcutaneous injection of
strychnine also enlarged the dermatomes. Together,
these observations suggest the existence of physiological inhibitory spinal mechanisms that influence
dermatomal size, implying that dermatomes may be
191
pliable physiological entities and not fixed anatomical

representations.
The variability in dermatomal maps from different
investigators, which likely represents a combination
of the above factors, impacts both DSEP technique
and interpretation. Selection of stimulation sites
should be done so as to minimize the inherent variability of dermatomes. Several investigators have
proposed stimulating at sites thought to be signature
areas (Katifi and Sedgwick, 1986; Slimp et al.,
1992). Signature areas are those areas which, when stimulated, would send an afferent volley to the brain
mainly or entirely through a single sensory root. Clinically, it is known that certain areas show greater sensory changes in the presence of injury (Nieuwenhuys,
1975). Physiological support for signature areas
comes from Sherringtons observations that nerve
fibers were more abundant in the center of the dermatome than at the edges (Sherrington, 1893), an observation that has been verified electrophysiologically in
cats and monkeys (Kuhn, 1953). Even if one assumes
appropriate segmental-level stimulation, one cannot
state with total certainty that a segmental level based
on a DSEP is associated with a corresponding anatomical level. Anatomical variants such as anastomoses
and prefixed and postfixed roots or the physiological
inhibitory mechanisms in the spinal cord impart a
statistical nature to the association.
DSEPs have been recorded clinically from all segmental levels from cervical to sacral levels. Most studies have focused on the cervical or lumbar levels
(Scarff et al., 1979; Scarff et al., 1981; Green et al.,
1983; Dvonch et al., 1984; Aminoff et al., 1985; Louis
et al., 1985; Katifi and Sedgwick, 1986, 1987;
Machida et al., 1986; Synek, 1986; Rodriquez et al.,
1987; Leblhuber et al., 1988; Pop et al., 1988; Schimsheimer et al., 1988; Debatisse et al., 1994; Righetti
et al., 1996; Yazicioglu et al., 1999). A few studies
have examined all spinal levels (Schramm et al.,
1980; Jorg et al., 1982; Toleikis et al., 1985; Synek,
1986; Toleikis and Sloan, 1987, 1988; Bamford,
1993) and only one study has dealt statistically with
cervical, thoracic, and lumbar levels (Slimp et al.,
1992). Technically speaking, recording DSEPs is no
more difficult than recording mixed nerve SEPs, but
there are a few considerations that must be kept in
mind. For example, DSEPs can be recorded routinely
in awake subjects, but difficulties can ensue when
patients become somnolent. Amplitudes may be
reduced or signals lost during sleep (Slimp et al.,
1992). This fact implies that the anesthetic conditions
192
J.C. SLIMP
of the operating room may impact DSEPs (see below).

Another consideration is the topographical organization of somatosensory cortex, which implies that
recording electrodes must be place laterally or medially to optimally record DSEPs to arm and leg, respectively, and in a midlateral position for trunk
stimulation (Slimp et al., 1992; Castillo and Papanicolaou, 2005). Along this line of thinking, another consideration is the paradoxical lateralization of
responses to lower leg or foot stimulation. The topographical focus for this part of anatomy is the mesial
area of primary somatosensory cortex. If the responding tissue should be located on the mesial wall of midline cortex, then the electrical dipole will be tilted such
that the effective resultant dipole orientation will not
be to the surface but will be displaced and oriented
towards the opposite cortex. The results are larger
amplitude responses recorded over the cortex ipsilateral to the stimulus rather than over the contralateral
cortex as normally expected, a phenomenon that
occurs about 8% of the time (Slimp et al., 1992).
The remainder of this chapter will focus on a
description of the DSEP technique, its application
in the operating room, and a discussion of factors that
may affect DSEP acquisition.
12.2. Description of DSEP technique
The technique described here is tailored for use in the
operating room. A full description of the clinical
technique used by our laboratory has been published

elsewhere (Slimp et al., 1992). The technique is similar to that of others (Jorg et al., 1982; Katifi and
Sedgwick, 1986; Toleikis and Sloan, 1987; Date
et al., 1988; Pop et al., 1988; Bamford, 1993).
12.2.1. Stimulation parameters
12.2.1.1. Stimulation sites
Intraoperative dermatomal stimulation is usually confined to the cervical and lumbosacral dermatomes.
Thoracic dermatomes may be stimulated but are generally of little concern because the consequences of
thoracic nerve root loss are minimal and, in fact, thoracic nerve root transactions are commonly part of
costotransversectomy procedures. Stimulation sites,
as mentioned previously, are at the signature areas
for the selected dermatome. The sites used by our
laboratory may be found in Table 1.
C4 dermatomal stimulation may cause an interfering shoulder shrug due to direct activation of the trapezius muscle and, thus, is rarely done in the
operating room. The consequence of C4 nerve root
loss is also not extreme. C5 dermatomal stimulation
also may cause a muscle twitch but it tends to be less
interfering. For most cervical applications, the
dermatomes of primary interest are C6, C7, and C8.
Stimulation of the thumb and fingers elicit no interfering muscle movement. One caution when applying
needle electrodes to the digits is to keep the electrodes
Table 1
Description of the cathode and anode locations for dermatomal somatosensory evoked potential (DSEP)
stimulation sites
Dermatome
Cathode location
Anode location
C4
C5
C6
C7
4 cm superior to midclavicle
5 cm distal and 3 cm anterior to the acromion process
On one side of the thumb between first and second phalangeal joint
On one side of the middle finger between first and second
phalangeal joint
On one side of little finger between first and second phalangeal joint
Lateral edge of hip on same line as the crease of the groin
3 cm lateral and 3 cm proximal to anode
3 cm lateral
3 cm distal
Opposite side of thumb
Opposite side of middle finger
C8
L2
L3
L4
L5
S1
S24
(Pudendal n.)
6 cm distal to inferior edge of patella and medial on the edge of

the tibia
3.5 cm posterior to web space of first and second toe
3.5 cm posterior to little to little toe on lateral edge of foot
Men: needle on one side of penis proximal to glans
Women: needle on one side of clitoris in labial fold
Opposite side of little finger

3 cm medial along the crease
10 cm proximal to superior
edge of patella
3 cm distal
3 cm distal
3 cm distal
Opposite side of penis
Opposite side of clitoris
in the subcutaneous tissue and do not penetrate too

deeply to avoid injuring the digital nerves which run
alongside each digit.
The same concern about interfering muscle twitch
may be expressed for L2, L3, and L4 dermatomal
stimulation, but it is less concerning than the proximal shoulder stimulation as the twitching of the leg
muscle usually causes no problem for the surgeons.
12.2.1.2. Stimulation electrodes
Electrodes may be either 0.5" subcutaneous needle
electrodes or surface electrodes. Needles have the
convenience of easy and quick application and a constant electrode-stimulating area. Surface electrodes,
either patch or disk electrodes for proximal arm and
leg stimulation (a larger size for arm and leg and a
smaller size for foot stimulation is advised) or ring
electrodes for finger stimulation, are also convenient
but maintaining a secure surface electrode connection
in the operating room can sometimes be problematic.
One concern with ring electrodes is setting them too
tight and causing strangulation of the digits.
12.2.1.3. Stimulator parameters
These are similar to those used in clinic. Stimulation
intensity, however, cannot be set as a function of sensory threshold, so a constant value of 3040 mA is
used as this corresponds to approximately three times
sensory threshold. The intensity may be increased,
if thought necessary, but too high an intensity may
cause spread to adjacent dermatomes or to underlying
mixed nerves. Some evidence from animal studies
suggests that it may be more efficient to use submaximal stimulation rather than supramaximal stimulation
(Tsai et al., 2005). The small size of the potential,
however, may make this approach untenable.
Stimulus duration is usually 0.2 or 0.3 ms. Long
pulse durations are not advised as it may cause stimulus spread to adjacent dermatomes or to underlying
mixed nerves. Stimulation rate should be kept below
5 Hz, preferably near 3 Hz. Too rapid a stimulation
rate could lead to reduction in response amplitude
because the synapses in the sensory pathway are
unable to follow the rapid stimulation pattern and fail
to conduct, thereby reducing the response.
12.2.2. Recording parameters
12.2.2.1. Recording sites
DSEPs are generally recorded from scalp electrodes,
reflecting the cortical response. Subcortical recordings,
193
either spinal or brainstem level, may be difficult

because of the small amplitude of the response and
should be done with caution. For example, a recent
paper reports only a 58% success rate in recording cervical level DSEPs (Tsai et al., 1997). Scalp recordings,
which are of reasonable amplitude, may be made at
the standard locations of C30 and C40 (2 cm posterior
to C3 and C4 of the International 1020 system) for
right and left arm stimulation, respectively, and Cz0
(2 cm posterior to Cz) for leg and foot stimulation.
A reference electrode is placed at Fz. A mastoid or ear
reference may be used but this lead usually picks
up more interfering background electromyographic
activity than Fz and offers no particular advantage to
signal acquisition.
12.2.2.2. Recording electrodes
Our laboratory uses 0.5" subcutaneous electrodes for
their ease of application and quick removal and their
relatively uniform impedance due to constant surface
area. Screw electrodes are an alternative to straight
needles but are more difficult to remove if rapid
removal is required in the instance of cardiac failure
or extubation. Collodion applied surface electrodes
may be used but are more difficult and time consuming to apply and offer no particular advantage over
needle electrodes.
12.2.2.3. Data acquisition parameters
Analog filters may be set with a bandpass of 10 or
30 Hz for low filters and 1,000 or 1,500 Hz for high
filters. Analysis time is best done at 100 ms for both
upper and lower extremity stimulation in order that the
full waveform may be visualized. The number of
sweeps per average may be set for 300 sweeps and
increased as necessary to achieve a clean average.
12.3. DSEP waveforms and the operating room
Typical waveforms generated in unanesthetized,
awake persons are shown in Fig. 1.
These traces are given to appreciate the full
response to dermatomal stimulation. Stimulation of
the thumb (C6), middle finger (C7), and little finger
(C8) elicits a response that has an initial negative
component at about 22 ms followed by a positive
wave at about 30 ms. The positive wave is indicated
in Fig. 1 by a dot below the trace. These initial components are followed by a second negativepositive
sequence, although this latter component may be
quite reduced under the influence of anesthesia, as
194
J.C. SLIMP
C4
L2
C5
L3
C6
L4
C7
L5
C8
S1
Pudendal
1 V
10 ms
Fig. 1. Cervical (C4, C5, C6, C7, C8), lumbosacral (L2, L3, L4, L5, S1), and pudendal dermatomal somatosensory evoked
potentials (DSEPs) from a normal, awake person. The dot under the traces indicates the initial positive wave. Recordings
were made from C30 or C40 for cervical stimulation and Cz0 for lumbosacral and pudendal stimulation.
may seen by comparing Fig. 1 records to the waveforms in the right panel of Fig. 4. It should be noted
that the waveform to finger or thumb stimulation is
similar to that elicited by stimulation of the median
or ulnar nerves.
Stimulation of dermatomes on the shoulder (C4,
C5) or legs (L2S1) elicits a different waveform that
has only an onset to an initial positive wave instead
of an initial negative component. The initial positive
wave is followed by a negativepositivenegative

sequence. Again the later positivenegative sequence
is usually reduced by the effect of anesthesia (compare Fig. 1 with Fig. 2). Of course as may be seen
in Fig. 1, the latencies of the responses increase with
increasing distance of the dermatome from the brain.
Atypical waveform presentations, termed paradoxical lateralization, can be found to stimulate
dermatomes on the lower leg or foot stimulation
240
Minutes
180
120
60
Start
L4
L5
S1
Fig. 2. Stable L4 (left panel), L5 (middle panel), and S1 (right panel) dermatomal somatosensory evoked potential (DSEPs)
recording across 4 h from the start of a minimally invasive posterior lumbar interbody fusion procedure. The arrow indicates the initial positive component. Recordings are with leads at Cz0 .
(Cruse et al., 1982; Lesser et al., 1987; Nagamine

et al., 1992; Slimp et al., 1992; Turazzini et al.,
1994; Baumgartner et al., 1998; Yamada, 2000).
When paradoxical lateralization is present, which
occurs about 8% of the time (Slimp et al., 1992),
assessment of side-to-side differences is compromised because of the asymmetry. For clinical purposes, it is important to use sufficient scalp
electrode arrays to identify this issue and to compensate for the asymmetries that paradoxical lateralization may imply. In the operating room, however,
the patients data is compared to itself and not to a
normative database, so identification of a waveform
as atypical due to paradoxical lateralization is not
as important. Consequently, it is generally sufficient
to use Cz 0 as the recording electrode. Lateral leads
only need to be replaced if no response is obtained
and one suspects that it may be present as a paradoxical presentation of the response.
Successful recording of DSEPs in the operating
room depends on meticulous recording technique.
DSEP responses are not as large as mixed nerve
responses even in the awake person. For example,
L5 or S1 DSEP amplitudes are on the order of 1/2
or less than tibial nerve SEP amplitudes in the awake,
unanesthetized person. Under anesthesia, SEP amplitudes may be reduced by 50% or more, thus rendering DSEP amplitudes in the range of 0.10.5 mV.
In order to resolve DSEPs, care must be taken to
reduce the noise to a level such that an average
may be made with a conventional 300 or 500 sweeps
per average. Even at that, signal-to-noise ratio may
be very low. Because of the variability of DSEP
responses, interpretation is difficult and may be further compromised if underlying noise is present and
variable and obscures the response. Noise sources
that most often need to be controlled are electrical
noise sources usually of 60-Hz activity and patient
electromyographic activity.
Electrical noise sources may include blood warmers, operating beds with electric motors, sequential
inflatable stockings, microscopes, C-arm fluoroscopes, and operating room lights. To reduce the
interference, these sources, such as the blood warmers or sequential stocking pumps, may be moved
away from the recording leads or rotated to minimize
interference. Other devices, such as the operating
room beds, may have to be turned off in order to collect data. Another way to minimize interference from
electrical sources is through careful routing of leads
to avoid close proximity to electrical sources and,
195
most importantly, to bundle the leads by knotting

them or taping them together so as to resemble a harness. Keeping the leads aligned and together will
lead to interfering electrical signals inducing electrical artifacts with similar amplitude on both the active
and reference electrodes. The interfering noise will
be reduced by common mode rejection, which is
the subtraction of the reference signal from the active
signal by the differential amplifiers and reduction of
the common electrical artifact.
Patient electromyographic activity is another
interference that may be controlled to a certain
degree. A patients muscles are relaxed by anesthesia
even without neuromuscular blockade but often not
completely. The residual spontaneous activity in
head and face muscles may cause a background interference that is difficult to reduce with averaging.
Electromyographic activity is not necessarily random
either with regard to its positive and negative voltages or with regard to its appearance over the time
that an evoked potential is averaged. Hence, it may
not be eliminated by the signal averaging process,
which assumes random occurrence of noise both in
the voltage domain and in the time domain. Reduction of electromyographic may be achieved with
neuromuscular blockade; however, if the monitoring
protocol calls for recording of any electromyographic
activity, such as spontaneous elctromyography (EMG)
or motor evoked potentials, this approach may not
be advisable. Anesthetic levels of either or both inhalational agents and narcotics may be increased, which
can reduce muscle activity. However, these changes
may compromise the integrity of DSEPs due to
anesthesia (see below). Increasing the number of
sweeps per average may help as it would for any noise
interference.
Successful recording of DSEPs in the operating
room also depends upon anesthetic conditions.
DSEPs in clinical studies are best recorded while
patients are awake. Somnolence clearly reduces
DSEP amplitudes (Slimp et al., 1992). From that
observation, it is not difficult to suspect that anesthetic conditions may have a profound effect on
DSEPs. It is commonly known that anesthetics have
a deleterious affect on evoked potential latencies
and amplitudes, and DSEPs are no exception (Nuwer,
1986; Herdmann et al., 1996; Sloan, 1998; Sloan and
Heyer, 2002). In fact, early papers commented that
DSEPs were adversely affected by inhalational
agents (Cohen et al., 1991a; Coscia et al., 1995).
Because of the effect on scalp recording, one author
196
chose to record from the cervical region instead of

the scalp (Tsai et al., 1997); however, as already
mentioned, the success rate for recordings was much
less with this approach. Earlier, anesthetic techniques used were balanced anesthesia using nitrous
oxide and narcotic (Herron et al., 1987; Cohen and
Huizenga, 1988; Cohen et al., 1991a). A combination
of nitrous oxide and propofol was used with cervical
recordings (Tsai et al., 2005). More recently, the
effect of desflurane, a newly derived inhalation
agent, was examined in rats and found to cause a
dose-related reduction in response amplitude even
in spinal cord potentials (Jou et al., 2003). Our own
experience has been with inhalational agents (isoflurane or sevoflurane) at a level of 1 MAC (minimum
alveolar concentration) or less. Our recordings have
been about as successful as others, meaning that from
10% to 20% of the time DSEPs are not obtainable or
are too inconsistent to be of value. It is intriguing,
however, to consider if DSEPs may be more reliably
recorded under total intravenous anesthesia with propofol and narcotic, conditions that are favorable for
recording motor evoked potentials.
Another aspect of recording evoked potentials in
the operating room that impacts DSEPs is the known
gradual reduction of amplitude of evoked potentials
which increases with increasing time under anesthesia (Lubicky et al., 1989; Kalkman et al., 1991;
Rappaport et al., 1992, 1994; Lyon et al., 2005).
Since DSEPs normally are lower amplitude than
mixed nerve SEPs, DSEPs may reach an undecipherable level at some point during a case. Increasing
stimulation intensity is tempting to do to restore lost
amplitude but is generally unsuccessful because sensory evoked potentials are normally elicited with
supramaximal stimulation. Another caution with
increasing stimulation with DSEPs is that a stronger
stimulus may recruit nerve fibers or mixed nerves that
do not represent the segmental level intended.
This phenomenon more than any other makes the
application of DSEPs in the operating the most suspect, simply because one cannot predict at the outset
whether the responses will remain stable.
If the aforementioned procedures are followed,
however DSEPs may be recorded in the operating
room. As an example, Fig. 2 shows a series of scalp
recordings to L4, L5, and S1 dermatomal stimulation
taken over 34 h in a single patient during a minimally invasive L4/L5 posterior lumbar interbody
fusion with a bone spacer and pedicle screw and
rod fixation. The responses show a clear initial
J.C. SLIMP
positive peak followed by a negative peak. There is

no clear later components, due to their suppression
by anesthesia, compared to awake recordings (see
Fig. 1). This particular example shows good stability
of the responses across time.
Such stability, however, is not always the case.
Fig. 3 shows examples of S1 DSEPs taken from three
different patients. The waterfall on the left, taken
over 12 h, is a series of stable recordings similar
to those in Fig. 2. The middle waterfall, however, is
an example of gradual fading of amplitude as the
case proceeded with some runs demonstrating indistinct positive components. As mentioned above,
fading of amplitudes is not atypical even for mixed
nerves but can be more problematic for DSEPs
because of their small size. If the variability is too
extreme, the warning criteria for a change in signal
must be revised from the conventional 50% reduction
in amplitude to an all or none criteria.
The right panel in Fig. 3 is a series of waveforms
in which there is no response. Our experience has
shown that about 10% of patients produce no
responses and another 510% of patients produce
variable responses. Others have reported higher success rates (Toleikis et al., 1993). Not surprisingly,
recordings from the cervical spine have a poorer
success rate (Tsai et al., 1997).
One other detail is that the success of recording
DSEPs may also depend on the dermatome stimulated. For example, in Fig. 4, very reasonable and stable DSEPs were recorded to C6 dermatomal
stimulation on the thumb (right waterfall) but in the
same patient no responses were recorded to C5 dermatomal stimulation on the shoulder (left waterfall).
Physiologically speaking, the distal extremities have
the greater innervation density and have the largest
representation in terms of area of somatosensory cortex receiving distal extremity projections (Kandell
et al., 1991). Thus, it stands to reason that stimulation
on the hands and feet may have the most success rate
in terms of eliciting a measurable response.
The primary advantage of DSEPs is their
segmental-level assessment, which is not feasible
with mixed nerve SEPs. Because of the variability
of dermatomes across individuals, the overlap of
adjacent dermatomes, and the physiological mechanisms in the spinal cord that influence dermatomes,
attention must be paid to stimulate in the signature
areas to optimize input to a single segmental level.
The ability to record DSEPs in the operating room
is not particularly more difficult than in a clinical
197
Start
1 V
10 ms
1.75 hr
STABLE
FADE
NO RESPONSE
Fig. 3. S1 dermatomal somatosensory evoked potentials (DSEPs) recorded with a Cz0 lead over the course of about 2 h
from three different patients undergoing minimally invasive lumbar interbody fusions or microdiscectomy. Stable responses
are shown (left panel), responses with amplitudes that faded over time (middle panel), and no response (right panel).
C5 DSEP - no response
C5 DSEPs - stable response

Start
4h
1 V
10 ms
Fig. 4. C5 and C6 dermatomal somatosensory evoked potentials (DSEPs) recorded from contralateral scalp in a patient
undergoing two-level cervical discectomy. No responses are present to C5 stimulation but clear and reproducible waveforms are recorded to C6 stimulation.
198
setting as long as one is aware of the small size of the

signal and that interfering noise or the effect of anesthesia may obscure the signal and confound interpretation. One disadvantage in doing DSEPs is that
collecting the data can be quite time consuming.
Most spine cases put more than one segmental level
at risk; so, it is not inconceivable that in addition to
any mixed nerve SEPs that would conventionally be
done, several DSEPs may be added to the protocol,
if nothing else but to bracket a segmental level of
interest in order to have control levels for comparison. This means that several minutes are devoted to
collecting multiple SEPs and DSEPs. Consequently,
assessments are clearly not immediate and by necessity may be delayed after the fact by several minutes.
DSEPs have been used in the operating room in
cases involving segmental manipulation or investigation of nerve roots. Examples are correction of spondylolisthesis (Cohen and Huizenga, 1988), lumbar
decompression (Herron et al., 1987; Owen et al.,
1991; Cohen et al., 1991a; Tsai et al., 1997; Naguszewski et al., 2001), intrapedicular fixation (Toleikis
et al., 1993), and assessment of nerve roots during
dorsal rhizotomy treatment for pain (Nemecek
et al., 2003) or during repair of brachial plexus injury
(Slimp, 2000). The consensus of these studies is that
recording DSEPs in the operating room can be done
fairly successfully as long as procedure is adhered to
and the patients circumstances are controlled well
enough to allow successful recording. In general, the literature is supportive of using DSEPs as a segmentallevel assessment in the operating room, but questions
have been raised about the sensitivity and specificity
of DSEPs (Owen et al., 1991, 1993; Jou, 2004). Some
have suggested that DSEPs may be used to determine
adequacy of decompression in the lumbar spine (Herron
et al., 1987; Cohen et al., 1991a), but there is some question if DSEPs can accurately predict outcome (Tsai
et al., 1997). Given the technical difficulty of DSEPs,
the physiological and anatomical basis for variability
in dermatomes, and the small body of literature on the
subject, it is fair to say that DSEPs are a modality in
need of further evaluation.
References
Aminoff, MJ, Goodin, DS, Barbaro, NM, Weinstein, PR
and Rosenblum, ML (1985) Dermatomal somatosensory
evoked potentials in unilateral lumbosacral radiculopathy. Ann. Neurol., 17: 171176.
J.C. SLIMP
Bamford, CR (1993) Dermatomal somatosensory evoked
potentials at the cervical, thoracic and lumbosacral
levels. Electroencephalogr. Clin. Neurophysiol., 88:
432434.
Baumgartner, U, Vogel, H, Ellrich, J, Gawehn, J, Stoeter, P
and Treede, RD (1998) Brain electrical source analysis
of primary cortical components of the tibial nerve
somatosensory evoked potential using regional sources.
Electroencephalogr. Clin. Neurophysiol., 108: 588599.
Boyer, P, Buchheit, F, Thiebaut, JB, Arrouf, L and Rihaoui,
SA (1981) [Anatomy of intradural anastomoses between
cervical nerve roots (authors translation)]. Neurochirurgie, 27: 191196.
Castillo, EM and Papanicolaou, AC (2005) Cortical representation of dermatomes: MEG-derived maps after tactile
stimulation. Neuroimage, 25: 727733.
Cohen, BA and Huizenga, BA (1988) Dermatomal monitoring for surgical correction of spondylolisthesis. A
case report. Spine, 13: 11251128.
Cohen, BA, Major, MR and Huizenga, BA (1991a)
Predictability of adequacy of spinal root decompression
using evoked potentials. Spine, 16: S379S384.
Cohen, BA, Major, MR and Huizenga, BA (1991b) Pudendal nerve evoked potential monitoring in procedures
involving low sacral fixation. Spine, 16: S375S378.
Coscia, MF, Trammell, TR, Popp, B, Gawande, SR, Fitzgerald, J and Scott, JR (1995) Effect of anesthetic variables on dermatomal somatosensory-evoked potential
monitoring in elective lumbar spinal surgery. J. Spinal
Disord., 8: 451456.
Cruse, R, Klem, G, Lesser, RP and Lueders, H (1982) Paradoxical lateralization of cortical potentials evoked by
stimulation of posterior tibial nerve. Arch. Neurol., 39:
222225.
Date, ES, Ortega, HR, Hall, K and Rappaport, M (1988)
Somatosensory evoked responses to dermatomal stimulation in cervical spinal cord injured and normal subjects. Clin. Electroencephalogr., 19: 144154.
Debatisse, D, Desfontaines, P, Selak, I, Maassen, D, Raket,
D, Hotermans, JM and Guerit, JM (1994) Diagnostic
and prognostic contribution of somatosensory evoked
potentials by truncular and dermatomal stimulation in
lumbosacral radiculopathy. Apropos of 120 cases
surgically-treated. Rev. Neurol. (Paris), 150: 222228.
Denny-Brown, D and Kirk, E (1968) Hyperesthesia from
spinal and root lesions. Trans. Am. Neurol. Assoc., 93:
116120.
Denny-Brown, D, Kirk, E and Yanagisawa, N (1973) The
tract of Lissauer in relation to sensory transmission in
the dorsal horn of spinal cord in Macaque monkey. J.
Comp. Neurol., 151: 175200.
Dvonch, V, Scarff, T, Bunch, WH, Smith, D, Boscardin, J,
Lebarge, H and Ibrahim, K (1984) Dermatomal somatosensory evoked potentials: their use in lumbar radiculopathy. Spine, 9: 291293.

Dykes, RW and Terzis, JK (1981) Spinal nerve distributions in the upper limb: the organization of the dermatome and afferent myotome. Philos. Trans. R. Soc.
Lond. B Biol. Sci., 293: 509554.
Foerster, O (1933) The dermatomes in man. Brain, 56:
139.
Green, J, Gildemeister, R and Hazelwood, C (1983) Dermatomally stimulated somatosensory cerebral evoked
potentials in the clinical diagnosis of lumbar disc disease. Clin. Electroencephalogr., 14: 152160.
Greenberg, SA (2003) The history of dermatome mapping.
Arch. Neurol., 60: 126131.
Head, H and Campbell, A (1900) The pathology of herpes
zoster and its bearing on sensory localization. Brain, 23:
353523.
Herdmann, J, Deletis, V, Edmonds, HL, Jr. and Morota, N
(1996) Spinal cord and nerve root monitoring in spine
surgery and related procedures. Spine, 21: 879885.
Herron, LD, Trippi, AC and Gonyeau, M (1987) Intraoperative use of dermatomal somatosensory-evoked
potentials in lumbar stenosis surgery. Spine, 12:
379383.
Jorg, J, Dullberg, W and Koeppen, S (1982) Diagnostic
value of segmental somatosensory evoked potentials in
cases with chronic progressive para- or tetraspastic syndromes. Adv. Neurol., 32: 347358.
Jou, IM (2004) The effects from lumbar nerve root transection in rats on spinal somatosensory and motor-evoked
potentials. Spine, 29: 147155.
Jou, IM, Chern, TC, Chen, TY and Tsai, YC (2003) Effects
of desflurane on spinal somatosensory-evoked potentials and conductive spinal cord evoked potential. Spine,
28: 18451850.
Kalkman, CJ, Ten Brink, SA, Been, HD and Bovill, JG
(1991) Variability of somatosensory cortical evoked
potentials during spinal surgery. Effects of anesthetic
technique and high-pass digital filtering. Spine, 16:
924929.
Kandell, ES, Schwartz, JH and Jessell, TM (1991) Principles of Neural Science. Appleton and Lange, New York,
3rd ed.
Katifi, HA and Sedgwick, EM (1986) Somatosensory
evoked potentials from posterior tibial nerve and
lumbo-sacral dermatomes. Electroencephalogr. Clin.
Katifi, HA and Sedgwick, EM (1987) Evaluation of the
dermatomal somatosensory evoked potential in the
diagnosis of lumbo-sacral root compression. J. Neurol.
Neurosurg. Psychiatr., 50: 12041210.
Keegan, JJ and Garrett, FD (1948) The segmental distribution of the cutaneous nerves in the limbs of man. Anat.
Rec., 102: 409437.
Kirk, E and Denny-Brown, D (1970) Functional variation
in dermatomes in the macaque monkey following dorsal
root lesions. J. Comp. Neurol., 139: 307320.
199
Kuhn, RA (1953) Organization of tactile dermatomes in cat
and monkey. J. Neurophysiol., 16: 169182.
Leblhuber, F, Reisecker, F, Boehm-Jurkovic, H, Witzmann,
A and Deisenhammer, E (1988) Diagnostic value of different electrophysiologic tests in cervical disk prolapse.
Neurology, 38: 18791881.
Lee, EK and Seyal, M (1998) Generators of short latency
human somatosensory-evoked potentials recorded over
the spine and scalp. J. Clin. Neurophysiol., 15:
227234.
Lesser, RP, Luders, H, Dinner, DS, Hahn, J, Morris, H, Wyllie, E and Resor, S (1987) The source of paradoxical lateralization of cortical evoked potentials to posterior
tibial nerve stimulation. Neurology, 37: 8288.
Louis, AA, Gupta, P and Perkash, I (1985) Localization of
sensory levels in traumatic quadriplegia by segmental
somatosensory evoked potentials. Electroencephalogr.
Clin. Neurophysiol., 62: 313316.
Lubicky, JP, Spadaro, JA, Yuan, HA, Fredrickson, BE and
Henderson, N (1989) Variability of somatosensory cortical evoked potential monitoring during spinal surgery.
Spine, 14: 790798.
Lyon, R, Feiner, J and Lieberman, JA (2005) Progressive
suppression of motor evoked potentials during general
anesthesia: the phenomenon of anesthetic fade. J.
Machida, M, Asai, T, Sato, K, Toriyama, S and Yamada, T
(1986) New approach for diagnosis in herniated lumbosacral disc. Dermatomal somatosensory evoked potentials
(DSSEPs). Spine, 11: 380384.
Maiuri, F and Gambardella, A (1989) Anomalies of the lumbosacral nerve roots. Neurol. Res., 11: 130135.
Moriishi, J, Otani, K, Tanaka, K and Inoue, S (1989) The
intersegmental anastomoses between spinal nerve roots.
Anat. Rec., 224: 110116.
Nagamine, T, Kaji, R, Suwazono, S, Hamano, T, Shibasaki,
H and Kimura, J (1992) Current source density mapping
of somatosensory evoked responses following median
and tibial nerve stimulation. Electroencephalogr. Clin.
Naguszewski, WK, Naguszewski, RK and Gose, EE (2001)
Dermatomal somatosensory evoked potential demonstration of nerve root decompression after VAX-D therapy. Neurol. Res., 23: 706714.
Nemecek, AN, Avellino, AM, Goodkin, R, Little, J and
Kliot, M (2003) Mapping dermatomes during selective
dorsal rhizotomy: case report and review of the literature. Surg. Neurol., 60: 292297; discussion 297.
Nieuwenhuys, R (1975) Bolks studies of segmental anatomy. Acta Morphol. Neerl. Scand., 13: 733.
Operating Room. Raven Press, New York.
Owen, JH, Padberg, AM, Spahr-Holland, L, Bridwell, KH,
Keppler, L and Steffee, AD (1991) Clinical correlation
between degenerative spine disease and dermatomal
200
somatosensory-evoked potentials in humans. Spine, 16:
S201S205.
Owen, JH, Bridwell, KH and Lenke, LG (1993) Innervation
pattern of dorsal roots and their effects on the specificity of dermatomal somatosensory evoked potentials.
Spine, 18: 748754.
Phillips, L and Park, T (1991) Electrophysiological
mapping of the segmental anatomy of the muscles of
the lower extremity. Muscle Nerve, 14: 12131219.
Pop, PH, Oepkes, CT, Notermans, SL, Vlek, NM and Stegeman, DF (1988) Dermatomal somatosensory evoked
potentials of the lumbar and cervical roots. Method
and normal values. Eur. Arch. Psychiatry Neurol. Sci.,
238: 2227.
Rappaport, M, Leonard, J and Ruiz Portillo, S (1992)
Effects of anesthesia and stimulus intensity on posterior
tibial nerve somatosensory evoked potentials. Clin.
Electroencephalogr., 23: 2430.
Rappaport, M, Ruiz Portillo, S, Ortiz, D, Fountain, SS and
Kula, TA, Jr. (1994) Effects of stimulus intensity and
duration on posterior tibial nerve somatosensory-evoked
potential patterns obtained under anesthesia. Spine, 19:
15251529.
Righetti, CA, Tosi, L and Zanette, G (1996) Dermatomal
somatosensory evoked potentials in the diagnosis of
lumbosacral radiculopathies. Ital. J. Neurol. Sci., 17:
193199.
Rodriquez, AA, Kanis, L, Rodriquez, AA and Lane, D
(1987) Somatosensory evoked potentials from dermatomal stimulation as an indicator of L5 and S1 radiculopathy. Arch. Phys. Med. Rehabil., 68: 366368.
Scarff, TB, Toleikis, JR, Bunch, WH and Parrish, S (1979)
Dermatomal somatosensory evoked potentials in children with myelomeningocele. Z. Kinderchir. Grenzgeb.,
28: 384387.
Scarff, TB, Dallmann, DE, Toleikis, JR and Bunch, WH
(1981) Dermatomal somatosensory evoked potentials in
the diagnosis of lumbar root entrapment. Surg. Forum,
32: 489491.
Schimsheimer, RJ, Ongerboer De Visser, BW, Bour, LJ,
Kropveld, D and Van Ammers, VC (1988) Digital nerve
somatosensory evoked potentials and flexor carpi radialis H reflexes in cervical disc protrusion and involvement of the sixth or seventh cervical root: relations to
clinical and myelographic findings. Electroencephalogr.
Schramm, J, Oettle, GJ and Pichert, T (1980) Clinical
application of segmental somatosensory evoked potentials (SEP) experience in patients with non-space
occupying lesions. In: C Barber (Ed.), Evoked Potentials. MTP Press, Lancaster, 455465.
Sherrington, CS (1893) Experiments in the examination of
the peripheral distribution of the fibers of the posterior
roots of some spinal nerves II. Philos. Trans. R. Soc.
Lond. B Biol. Sci., 184: 641763.
J.C. SLIMP
Sherrington, CS (1898) Experiments in the examination of
the peripheral distribution of the fibres of the posterior
roots of some spinal nerves II. Philos. Trans. R. Soc.
Lond. B Biol. Sci., 190: 45186.
Slimp, JC (2000) Intraoperative monitoring of nerve
repairs. Hand Clin., 16: 2536.
Slimp, JC, Rubner, DE, Snowden, ML and Stolov, WC
(1992) Dermatomal somatosensory evoked potentials:
cervical, thoracic, and lumbosacral levels. Electroencephalogr. Clin. Neurophysiol., 84: 5570.
recordings. J. Clin. Neurophysiol., 15: 217226.
Sloan, TB and Heyer, EJ (2002) Anesthesia for intraoperative neurophysiologic monitoring of the spinal cord. J.
Synek, VM (1986) Somatosensory evoked potentials after
stimulation of digital nerves in upper limbs: normative
data. Electroencephalogr. Clin. Neurophysiol., 65:
460463.
Terada, K, Larson, BJ, Owen, JH and Sugioka, Y (1993)
The effect of nerve root lesioning on various somatosensory evoked potentials in the hog. Spine, 18:
10901095.
Toleikis, JR and Sloan, TB (1987) Comparison of major
nerve and dermatomal somatosensory evoked potentials
in the evaluation of patients with spinal cord injury. In:
C Barber and T Blum (Eds.), Evoked Potentials III. The
Third International Evoked Potentials Symposium. Butterworth, Boston, pp. 309315.
Toleikis, JR and Sloan, TB (1988) A comparison of dermatomal and major nerve evoked responses with clinical
diagnosis on acute spinal injury. In: TB Ducker and
RH Brown (Eds.), Neurophysiology and Standards of
Spinal Cord Monitoring. Springer, New York, pp.
295301.
Toleikis, JR, Sloan, TB, Schrader, SC and Koht, A (1985)
Scalp distribution of dermatomal evoked potentials. In:
J Schramm and J Jones (Eds.), Spinal Cord Monitoring.
Springer, Berlin, pp. 5963.
Toleikis, JR, Carlvin, AO, Shapiro, DE and Schafer, MF
(1993) The use of dermatomal evoked responses during
surgical procedures that use intrapedicular fixation of
the lumbosacral spine. Spine, 18: 24012407.
Tsai, RY, Yang, RS, Nuwer, MR, Kanim, LE, Delamarter,
RB and Dawson, EG (1997) Intraoperative dermatomal
evoked potential monitoring fails to predict outcome
from lumbar decompression surgery. Spine, 22:
19701975.
Tsai, TM, Tsai, CL, Lin, TS, Lin, CC and Jou, IM (2005)
Value of dermatomal somatosensory evoked potentials
in detecting acute nerve root injury: an experimental
study with special emphasis on stimulus intensity.
Spine, 30: E540E546.
Turazzini, M, Zanette, G, Bongiovanni, LG, Bertolasi, L,
Polo, A and De Grandis, D (1994) Variability of tibial

nerve early cortical potentials in normal subjects. Electromyogr. Clin. Neurophysiol., 34: 229235.
Yamada, T (2000) Neuroanatomic substrates of lower
extremity somatosensory evoked potentials. J. Clin. Neurophysiol., 17: 269279.
201
Yazicioglu, K, Ozgul, A, Kalyon, TA, Gunduz, S, Arpacioglu,
O and Bilgic, F (1999) The diagnostic value of dermatomal
somatosensory evoked potentials in lumbosacral disc herniations: a critical approach. Electromyogr. Clin. Neurophysiol., 39: 175181.

M.R. Nuwer (Ed.)
202
CHAPTER 13
Monitoring spinal epidural potentials to

peripheral nerve stimulation
Steve Jones*
Department of Clinical Neurophysiology, National Hospital for Neurology and Neurosurgery, London WC1N 3BG, UK
13.1. Background
The first recordings of sensory evoked activity in the
human spinal cord were made in the 1930s (Gasser
and Graham, 1933). With only one or two intervening publications, it was not until the 1970s that a
number of groups in Japan started to exploit the
experimental and clinical possibilities of the evoked
spinal cord potentials (e.g., Shimoji et al., 1972;
Tsuyama et al., 1978). These studies involved inserting an electrode at the end of a flexible lead into the
epidural space. When located at the appropriate level,
the electrode was used to record localized segmental activity generated in the dorsal columns and
dorsal horn, following stimulation of the dorsal roots
or a nerve trunk in the upper or lower limb. It was the
realization that an electrode at cervical level could be
used to record conducted activity following stimulation of a nerve in the lower limb, or of the spinal
cord at a lower level, that led to the exploitation of
this technique in investigating and monitoring the
integrity of long spinal cord tracts during surgery
(Tamaki et al., 1981).
In Japan, it was generally the surgeons themselves
who were responsible for all aspects of the recording
procedure. This, I believe, is the reason why the
methodology most widely used in Japan today
involves both stimulating and recording electrodes
located in the dorsal epidural space. Elsewhere, the
involvement of neurophysiologists, accustomed to
methods of peripheral nerve stimulation used for
the clinical application of somatosensory evoked
potentials (SEPs), resulted in the importation of their
*
Correspondence to: Steve Jones, Ph.D., Department of

Clinical Neurophysiology, The National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N
3BG, UK.
Tel.: 44-20-7837-3611, ext. 4109; fax: 44-20-7713-7743.
E-mail: sjjones@ion.ucl.ac.uk (S. Jones).
methods into the operating theater. The epidural SEP

(ESEP) method appears to have been introduced concurrently and independently in the UK and Australia
(Jones et al., 1982; Macon and Poletti, 1982).
13.2. Applicability
In the UK, the ESEP method is used almost exclusively
by orthopedic surgeons concerned with the treatment
of spinal deformities, in particular kyphoscoliosis,
occurring in adolescents and young adults. Most of the
patients, therefore, are otherwise healthy and neurologically normal. A minority will have kyphoscoliosis of
neurogenic origin; when associated with peripheral
neuropathy or combined peripheral/central conditions
such as Friedreichs ataxia, the degeneration of large
diameter peripheral sensory fibers may be so extensive
that it is impossible to record a reliable ESEP. In these
cases, cortical SEPs may sometimes be better preserved, but this is not necessarily to be relied upon. In
cases of severe neuropathy, therefore, the Japanese
method of recording evoked spinal cord potentials to
stimulation of the spinal cord at a rostral or caudal level
is probably the only one which can be resorted to. In
Friedreichs ataxia, the fact that the sensory pathways
of the spinal cord, specifically the dorsal columns, are
also affected may thwart even this technique.
The fixation and instrumentation used for the
correction of scoliosis (Harrington rods and compression/distraction systems attached via sublaminar wires
or pedicle screws) are all applied from the dorsal side,
so the initial placement of an electrode in the dorsal
epidural space is straightforward. When the operation
also involves treatment of lumbar kyphosis, it is
clearly not possible to insert an electrode via the surgical field. However, a number of publications describe
methods for inserting the electrode percutaneously in
the anesthetic room (e.g., Anderson et al., 1990) which
can also be exploited in other contexts.
ESEPs are less frequently used for spinal cord

monitoring during neurosurgical procedures. This is
probably because the insertion of an electrode into
the epidural space can be difficult or undesirable in
patients with a narrow spinal canal (stenosis), or in
whom the canal is otherwise distorted by tumor, subluxation (displacement of one vertebral body relative
to the adjacent one), or fracture.
When performed at multiple levels, ESEPs offer
the opportunity of detecting not only the occurrence
but also the level of conduction failure in the spinal
cord. Multi-level recordings are not performed routinely, on account of the obstruction caused by the
presence of leads in the surgical field, but additional
electrodes can be rapidly inserted by the surgeon if
and when a problem is identified.
13.3. Recording electrodes
The recording electrode consists of a cylindrical
sleeve of a chemically inert conducting material
(stainless steel or platinum) whose diameter is the
same as that of the lead whose end it forms. The
diameter is usually around 1 mm and the length of
the sleeve is around 35 mm. A second sleeve,
at least 1 cm and no more than 5 cm along the same
lead, may be used in conjunction with the first sleeve
in order to obtain bipolar recordings. Alternatively,
for monopolar recordings, a reference electrode (for
example, a stainless steel needle) can be placed in
the body tissue outside the surgical zone, as close
as possible to the epidural electrode in order to avoid
excessive pickup of extraneous noise.
In most centers, disposable electrodes are now
generally preferred to reusable types, but there is no
reason, in principle, why an epidural electrode, properly constructed so there is no point of potential
weakness or seepage between the conductive sleeve
and the rest of the lead, should not be sterilized and
used on multiple occasions. When under the control
of the surgeon, the lead is introduced to the epidural
space at the upper end of the surgical field, after
exposure of the spine. It can be inserted directly
when the dura is exposed by a laminectomy. Alternatively, the surgeon may choose to make a small hole
in the interspinous ligament and insert the lead
directly. A third option is to use the needle of a catheter of the appropriate diameter to pierce the interspinous ligament. The needle is then withdrawn leaving
the catheter in place and the lead is introduced.
Secured by a suture, the rest of the lead can then be
203
routed out of the surgical field in the direction of

the head.
The natural angle of the dorsal processes means
that the lead will be introduced aslant, the electrode
tip pointing in the direction of the head. This is desirable, because on entry into the epidural space, it will
then naturally tend to proceed in a cephalad direction. Obviously, recordings of evoked sensory activity need to be made from a level above any that are
considered to be at risk during surgery. If the lead
is introduced, for example, at the T1/2 level, there
will usually be no obstruction to advancing the electrode to a low- or mid-cervical level comfortably
above those segments of the cord that are likely to
be at risk during surgery for kyphoscoliosis.
Critical for the successful placement of the electrode, however, is the material of the lead. The
advantage of making the lead as thin as possible is
obvious, but if the conducting core and insulating
sleeve are too flexible, this may result in unpredictable deviations as the lead is advanced. Small deviations from the midline are unimportant and probably
inevitable, but clearly the surgeon needs to be confident that the lead is not so flexible that the tip finishes
up altogether on one side of the cord or the other.
In some centers, the surgeon or anesthetist will be
confident to introduce the epidural electrode percutaneously in the anesthetic room, after induction of
anesthesia (e.g., Anderson et al., 1990). This has the
advantage that the ESEP technique can then be used
to monitor operations performed from the anterior
side of the spine.
13.4. Recording parameters
The ESEP recorded at low cervical level to stimulation of the posterior tibial nerve at the level of the
knee (popliteal fossa) has a latency of around 15 ms
(depending, of course, on the age and height of the
patient) and a duration of around 58 ms. It therefore
demands a recording epoch (window) of at least
25 ms starting at the time of stimulus delivery. Most
evoked potential (EP) recording machines today offer
a wide range of epochs including 30 ms which is
probably the ideal. There is usually no means of
varying the digital sampling rate independently of
the recording epoch. To obtain accurate representation of the briefest peaks in the response, a sampling
rate of at least 10 kHz is desirable; with an epoch
of 30 ms, this will almost certainly be comfortably
exceeded.
204
Since the ESEP largely comprises brief, polyphasic

potentials rather than broad, slow waves, it is not necessary to amplify the signals over the full frequency
range at the lower end of the spectrum. This is highly
advantageous, since it means that mains-related noise
frequencies of 50 or 60 Hz can be excluded by the
use of a high-pass filter at 100 or even 200 Hz. In
order faithfully to reproduce the high frequencies in
the waveform, the low-pass filter needs to be set no
lower than 2 kHz.
The number of responses that need to be averaged
in order to get an accurate measure of the signal
depends on the level of background noise. Two hundred may be adequate in many circumstances, with
the option of adding further responses to the average
when noise levels are relatively high.
13.5. Bipolar versus monopolar recording
The choice of monopolar (one epidural electrode
referred to a needle or some other electrode alongside
the spine) or bipolar epidural recording is probably
best left to personal preference, since there seems to
be an approximately equal weight of opinion on both
sides. Monopolar recordings have the advantage that
the individual components of the response can all be
distinguished and are not confused with one another.
This is perhaps of greatest value to neurophysiologists interested in exactly which tracts of the spinal
cord are responsible for each component. Bipolar
recordings are generally less noisy, because the close
proximity of the two electrodes means that a larger
proportion of the background noise is excluded by
common mode rejection. In principle, it should be
possible to construct a bipolar electrode pair in which
the separation is chosen so as to maximize the amplitude of the response. For example, a typical ESEP
waveform might consist of three negative and three
positive peaks, the consecutive peaks of opposite
polarity being about 0.5 ms apart. Their mean conduction velocity in the spinal cord is in the order of
60 m/s; so, in 0.5 ms, the waveform would travel
approximately 3 cm. In theory, then, a 3-cm separation between the two electrodes should result in the
largest recorded amplitude, since a peak of negativity
recorded by one electrode would be amplified relative to a peak of positivity recorded by the other.
Having the electrode separation less than 3 cm (as
is usually the case) will undoubtedly reduce the
amplitude of the response, but the components of
the waveform should still be distinguishable and the
S. JONES
background noise should be less. Separations of more

than 3 cm should result in the waveform becoming
more polyphasic, which is probably not desirable,
and noisier.
13.6. Stimulating electrodes
Many types of stimulating electrodes have been used
for SEP studies outside the operating theater, but
most are unsuitable for intraoperative use. In particular, stimulating electrodes that need to be moistened
with water or saline may be prone to drying out during a long procedure. The use of straps around the
limb is also strongly advised against, since this might
cause constriction of the blood circulation. Stick-on
type electrocardiographic (ECG) electrodes have the
advantage that their contact area of gel is largely protected from evaporation. They can be effectively
used to stimulate virtually any peripheral nerve that
does not run too deep in the limb muscles, without
the use of elastic straps. Needle electrodes should
be equally effective, although they should only be
inserted when the patient is finally positioned for
surgery and removed if repositioning is necessary.
13.7. Stimulation sites
The largest lower limb nerve running close enough to
the surface to be easily stimulated transcutaneously
is the posterior tibial nerve at the level of the popliteal
fossa. The best way of discovering the most effective
site for locating the stimulating cathode is to try it on
oneself. The posterior tibial nerve contains the majority of large diameter (therefore, fast-conducting
and with a low threshold for electrical excitation)
group II cutaneous sensory fibers deriving from the
sole of the foot, where the density of innervation is
probably higher than elsewhere in the limb. It also
contains large diameter proprioceptive fibers (group I
muscle spindle and Golgi tendon organ afferents) from
muscles responsible for flexion of the foot and the
toes. The general experience is that this is the best
nerve to stimulate for intraoperative monitoring of
ESEPs. Cortical SEPs, however, are usually better
produced by stimulation of the posterior tibial nerve
at the level of the ankle. The reason for this is
unclear, but it may be related to inhibitory interaction at cortical or subcortical level between the cutaneous projection and the muscle afferents projecting
to different regions of the cortex. ESEPs can be
recorded to stimulation of other nerves in the leg,
but generally it is only the posterior tibial nerve at

the popliteal fossa that gives rise to the large amplitude, fast-conducted responses which are most easily
measured.
In the literature it is often advised that the stimulating cathode for SEP recordings be placed proximal
to the anode, in order to avoid anodic conduction
block (i.e., failure of the nerve volley originating
at the cathode to progress past the anode on account
of hyperpolarization of the axonal membrane). In
practice, however, this possibility can be discounted.
It is certainly important to locate the cathode in as
close proximity to the nerve as possible. The anode
then needs to be reasonably close by, since if it is
too distant, the stimulus artifact conducted through
the body and detected by the recording electrodes
may be unmanageably large. The anode does not have
to be in close proximity to the nerve, although either
a distal or a proximal location may be found to be
convenient. If the anode and cathode are reversed,
the response will usually be found to be degraded,
but this is simply due to the sub-optimal location
of the cathode and has nothing to do with anodic
block.
13.8. Stimulation parameters
All the large diameter (groups I and II) sensory fibers
in the posterior tibial nerve can be activated by a
square-wave cathodic (negative) electrical pulse 0.1
or 0.2 ms in duration. The threshold is lowest for
the fibers of largest diameter; the earliest components
of the ESEP may, therefore, be elicited by pulses of
only a few milliamperes (mA), while it may require
25 mA or more in order to produce maximal amplitude of the later peaks.
One of the advantages of epidural as compared
with cortical SEPs is that they can be recorded without loss of amplitude or definition at stimulation rates
as high as 20 pulses/s, whereas the optimal rate for
cortical SEPs, trading some loss of amplitude for
higher speed of acquisition, is probably around
5 pulses/s. Even with stimulation at 50 pulses/s the
ESEP was found to be only minimally attenuated
(Jones et al., 1982). However, at the highest rates it
may be found that the ESEP gradually decreases in
size when stimuli are applied for long periods. For
that reason, it is recommended that the standard rate
be no higher than 20 pulses/s, and that a few seconds
of rest be allowed after the completion of each average of 200 or 500 responses. Another hazard that has
205
been noted is that very high stimulation rates have

caused mild burns to the skin. This is probably only
a significant factor if the stimulating electrode has
become dry (old stock?) or is otherwise defective.
A very useful facility offered by most, if not all, of
todays EP recording machines is the ability to record
separate responses to stimuli alternating between one
stimulus site and another. Consequently, it is possible
to record ESEPs separately but concurrently to stimulation of the left and right legs, each leg being stimulated at 10 or even 20 pulses/s. At higher rates, the
stimulus artifact associated with stimulation on one
side will occur before the response to the last stimulus on the other side has completed.
13.9. Intervention criteria
As with cortical SEPs, the figures of 50% amplitude
reduction and 10% latency reduction are applied as
rough guidelines for the degrees of change that are
considered to be significant. It should be remembered, however, that these are arbitrary figures which
are not based on statistical evidence. In retrospect,
it has been generally found that only those patients
whose responses diminish by rather more than 50%
as compared with the initial baseline figure have
a substantially increased risk of neurological complications. One problem is the slight tendency (probably
less marked for epidural as compared with cortical
SEPs) for amplitudes to decline steadily throughout
the course of surgery. There is also a tendency for
latencies to gradually increase on account of the
cooling effect on the spinal cord when the vertebral
column is surgically exposed. For these reasons, it
is not recommended that the 50% amplitude and
10% latency criteria be rigidly applied, literally or
metaphorically, as triggers for alarm bells.
Since ESEPs are far more immune to systemic
factors associated with changes in blood pressure
and/or anesthetic agents than cortical responses, any
relatively sudden changes that occur in amplitude or
(more rarely) latency can be confidently ascribed to
surgical events. One is far less worried about a
response which hovers around 50% amplitude as
compared with the initial baseline value, perhaps
occasionally dipping slightly below 50%, than one
in which a sudden reduction of, say, 30% occurs,
even though this may not cause the overall amplitude
to fall below the non-magical 50% criterion.
Another important difference of the epidural as
compared with the cortical SEP is that parts of the
206
response (paradoxically, the earlier rather than the

later peaks, see below) are due to postsynaptic spinal
cord axons. Since the effects of cord ischemia are
likely to impact first on the gray matter rather than
the axonal tracts, there are circumstances in which the
postsynaptic components may be affected before the
later, asynaptic peaks. We (Jones et al., 1983) reported
one case in which the initial component of the waveform was temporarily lost on one side after mild
trauma to the conus medullaris, suggesting perhaps a
mild ischemia causing impaired synaptic processes in
the dorsal horn at this level.
13.10. Generators of ESEPs
Although it is not part of my brief to cover the clinical aspects of ESEP monitoring, it is not possible
entirely to dissociate the practical issues of monitoring from those pertaining to the normal neurophysiology of the sensory pathways of the spinal cord. This
was the object of our study (Halonen et al., 1989)
which examined the responses recorded at various
levels of the cord posteriorly and anteriorly (the latter
using needle electrodes inserted in the intervertebral
discs) and stimulating various nerves in the lower
limbs at a range of stimulus intensities.
The study first examined the changing waveform
as the electrode was placed at different levels of the
dorsal epidural space. Plotting the latency of each
peak against distance along the spine revealed a complex pattern of conduction in which different peaks
traveled at different velocities. The relatively simple,
triphasic positive-negative-positive (PNP) waveform
recorded at lumbar and low thoracic levels appeared
to be delayed by about 1 ms at the level of the lumbar
enlargement, before splitting into two and eventually
three or even four negative peaks, overlapping in
time and conducted at different velocities ranging
from about 40 to 80 m/s (Fig. 1). In addition to the
peaks whose latency increased in the rostral direction, there was one whose latency was apparently
constant from low to high thoracic levels, probably
a far-field potentials due to the change in volume
conductor characteristics as the volley in the dorsal
roots entered the spinal cord. At thoracolumbar
levels, there was also a peak whose latency increased
in the caudal direction, perhaps a reflex efferent volley associated with the F-wave or H-reflex.
A second experiment looked at the lateralization
of the peaks by deliberately placing the recording
electrode toward the left or right side of the epidural
S. JONES
space, as well as on the midline. This revealed a tendency for all the peaks to be larger on the side ipsilateral to the stimulated limb, although the degree of
lateralization was greatest for the first and least for
the third component identified at T4 level.
Varying the stimulus intensity also had the effect of
dissociating the three negative peaks of the ESEP. It
was consistently found that the first peak had the lowest electrical excitation threshold and the third peak the
highest. As the intensity was further increased, all the
peaks recorded at upper thoracic level eventually
reached a plateau of maximal amplitude, while the
major negativity recorded at the level of the lumbar
enlargement apparently did not. Clearly, the relationship of threshold with stimulus intensity suggests that
the later components in the upper thoracic waveform
were due to peripheral sensory fibers of smaller diameter, which is also compatible with their longer latency
and (presumably) slower conduction velocity.
In addition to its lower threshold and faster conduction velocity, evidence was obtained for a postsynaptic origin of the first and possibly also the
second component of the ESEP. When the posterior
tibial nerve was stimulated at the knee with a relatively low stimulus intensity, a delay of 12 ms was
evident between the single negative peak recorded
at thoracolumbar level and the first peak recorded
more rostrally, extrapolated back to estimate its
latency at thoracolumbar level (Fig. 2). At this low
intensity, two or three peaks were recorded at higher
levels, all having a similar, fast conduction velocity
suggesting a repetitive volley in postsynaptic axons.
Also, when stimuli were given in pairs separated by
26 ms, a differential effect was sometimes noted
on the three components, all three being present at
longer intervals, but the first being relatively attenuated or absent at intervals of 4 ms or less. In order
to further establish which afferent fiber types were
responsible for the response, ESEPs were recorded
to stimulation of different nerve trunks in the lower
limb. All amplitudes were reduced and latencies, of
course, increased when the posterior tibial nerve
was stimulated at the level of the ankle rather than
the knee, consistent with a mean peripheral conduction velocity of around 50 m/s (Fig. 3). However, the
morphology of the response was also altered, the
initial component recorded at upper thoracic level
being relatively reduced as compared with the later
peaks. When the stimulus was delivered to the sural
nerve at the ankle, the response was further reduced
in amplitude, and the initial peak was apparently
207
1 2 3
T11
T1
T4
T12
T5
T6
L1-upper
T7
L1-lower
T8
L2
T9
4 V
2 V
L3
T10
10
20
30 ms
10
Pt
20
30
3
C7
Ps
T4
T5
T6
T7
T8
T9
T10
T11
T12
L1
F/H?
L2
P
L3
10
12
14
16
18
ms
Fig. 1. Epidural SEPs (ESEPs) recorded at multiple spinal levels following posterior tibial nerve stimulation at the popliteal
fossa (adapted from Halonen et al., 1989). The plot of peak latencies against spinal level (scaled according to the relative
size of the vertebrae) reveals a number of features as described in the text.
completely absent. Since the sural nerve contains

only cutaneous sensory fibers, while the posterior tibial nerve also contains muscle afferents which are
likely to be more numerous at the level of the knee
than the ankle, this led us to conclude that the initial,

fastest conducted component of the ESEP was likely
to be due to the projection of group I muscle spindle
and Golgi tendon organ afferents, while the third
208
S. JONES
Level:
Stimulus: 25 ma
cm
30
T2
67
44
38 m/s
T5
20
T9
10
T12
10
20
Stimulus: 7 ma
2 V
30 ms
10
cm
30
Level:
77
T2
T5
20
T9
10
15
20 ms
70 83 m/s
T12
10
20
1 V
30 ms
10
15
20 ms
Fig. 2. Epidural SEPs (ESEPs) simultaneously recorded at four levels following posterior tibial nerve stimulation at two
intensities (adapted from Halonen et al., 1989).
component was due to group II afferents of cutaneous origin. The second component with an intermediate conduction velocity was more difficult to
characterize.
ESEPs recorded at four spinal levels to stimulation of the posterior tibial and sural nerves at the
ankle, and the tibial nerve at the knee, showed a clear
pattern of differences (Fig. 3). The first, fastest conducted component was only clearly visible in the
response to tibial nerve stimulation at the knee, and

was conducted at around 80 m/s. Responses to stimulation of the same nerve at the ankle mainly consisted
of two components, conducted respectively at around
55 and 45 m/s, similar to components two and three
of the knee response. The sural nerve response consisted of multiple small peaks, all except possibly
the first apparently conducted at velocities of around
40 m/s.
Level:
209
Tibial (knee)
cm
30
T3
80
T6
57
45 m/s
20
T11
10
L3
15
10
20
Tibial (ankle)
30
T3
T6
56
45 m/s
20
T11
10
L3
1 V
25
20
30
Sural (ankle)
30
T3
T6
56
37 m/s
20
T11
10
L3
10
20
30
40
0.5 V
50 ms
20
25
30 ms
Fig. 3. Epidural SEPs (ESEPs) simultaneously recorded at four levels following stimulation of the posterior tibial nerve at
the knee and the ankle and the sural nerve at the ankle (adapted from Halonen et al., 1989).
210
The model to which these various lines of evidence all contribute is one in which the first component of the ESEP is generated in a postsynaptic,
fast-conducting sensory tract located relatively laterally in the spinal cord on the side ipsilateral to the
stimulus. The fact that this component is not elicited
by stimulation of the purely cutaneous sural nerve
suggests that the fibers concerned are muscle afferents of group I, and the tract is the dorsal spinocerebellar tract which has its synaptic origin in Clarkes
column, located in lamina VII at the base of the
dorsal horn. The third component, due to a more
medially located tract and due to fibers of slower
conduction velocity with no intervening synapse, is
almost certainly the dorsal columns which convey
mainly cutaneous activity.
13.11. Advantages and disadvantages of
epidural as compared with cortical SEPs
From a practical perspective, the advantages of
recording ESEPs to stimulation of the posterior tibial
nerve at the knee are that the activity of more than
one afferent spinal cord pathway can be distinguished, and that one of the major constituents of
the response is postsynaptic and may, therefore, provide a sensitive indicator of ischemia in the cord at
the level of the lumbar enlargement. Further advantages are the virtual immunity of ESEPs to anesthetic
agents and variations in blood pressure (unless this
becomes catastrophically low), and the high speed
of acquisition.
The most significant disadvantage of ESEPs is
the invasiveness of the method and the fact that the
technique cannot be applied in all circumstances
where spinal cord monitoring is indicated. The technique is generally found to be safe in patients whose
spinal canal is unobstructed; very rarely, however,
insertion of the electrode may cause rupture of the
dura and leakage of cerebrospinal fluid. One factor
which has been noted to impair the recording of
ESEPs is pooling of blood in the vicinity of the
recording electrode. Another factor is that responses
may be lost when the lead is inserted too far, such
that the recording tip deviates too far from the midline of the spinal cord. Both of these problems may
require the electrode to be temporarily removed and
reinserted, and this will invalidate the baseline
amplitude and latency values used to identify
subsequent changes. Finally, it should be noted that
the application of high stimulus intensities at fast
S. JONES
rates and for long periods may risk the occurrence

of a compartment syndrome. For this reason, the
use of muscle relaxants, at least in low concentrations, may be considered advisable.
In adolescent patients undergoing surgery for
kyphoscoliosis, it is often possible to record spinal
SEPs of good quality from the skin of the neck
at mid-to-high cervical level. Consequently, it
may be argued that the use of invasive epidural electrodes is not entirely necessary. Of course, it is possible to record a cervical response concurrently with
cortical potentials, but drawbacks to this are that
the optimal rate of stimulation for cortical SEPs is
much lower, and that whereas the cervical response
is best recorded to stimulation at the popliteal fossa,
cortical SEPs tend to be better defined and of larger
amplitude to stimulation of the same nerve at the
ankle.
References
Anderson, SK, Loughnan, BA and Hetreed, MA (1990) A
technique for monitoring evoked potentials during scoliosis and brachial plexus surgery. Ann. R. Coll. Surg.
Engl., 72(5): 321323.
Gasser, HS and Graham, HT (1933) Potentials recorded in
the spinal cord by stimulation of the dorsal roots. Am. J.
Physiol., 103: 303320.
Halonen, JP, Jones, SJ, Edgar, MA and Ransford, AO (1989)
Conduction properties of epidurally recorded spinal cord
potentials following lower limb stimulation in man. Electroencephalogr. Clin. Neurophysiol., 74(3): 161174.
Jones, SJ, Edgar, MA and Ransford, AO (1982) Sensory
nerve conduction in the human spinal cord: epidural
recordings made during scoliosis surgery. J. Neurol.
Neurosurg. Psychiatry, 45(5): 446451.
Jones, SJ, Edgar, MA, Ransford, AO and Thomas, NP (1983)
A system for the electrophysiological monitoring of the
spinal cord during operations for scoliosis. J. Bone Joint
Surg. Br., 65(2): 134139.
Macon, JB and Poletti, CE (1982) Conducted somatosensory evoked potentials during spinal surgery. Part 1:
control conduction velocity measurements. J. Neurosurg., 57(3): 349353.
Shimoji, K, Kano, T, Higashi, H, Morioka, T and Henschel,
EO (1972) Evoked spinal electrograms recorded from
epidural space in man. J. Appl. Physiol., 33: 468471.
Tamaki, T, Tsuji, H, Inoue, SI and Kobayashi, H (1981) The
prevention of iatrogenic spinal cord injury utilizing the
evoked spinal cord potential. In Orthop., 4: 313317.
Tsuyama, N, Tsuzuki, N, Kurokawa, T and Imai, T
(1978) Clinical application of spinal cord action
potential measurement. In Orthop., 2: 3946.

M.R. Nuwer (Ed.)
211
CHAPTER 14
Somatosensory evoked potentials for intraoperative

mapping of the sensorimotor cortex
Theodoros Kombos*
Department of Neurosurgery, Charite-Universitatsmedizin Berlin, D-12200 Berlin, Germany
14.1. Introduction
Modern neurophysiological imaging and monitoring techniques enable precise localization of lesions
and correlation of them to anatomical landmarks.
These techniques increase safety and efficacy and
reduce invasiveness. A more aggressive approach to
a brain tumor increases survival and quality of life
(Hirakawa et al., 1984; Laws et al., 1984; Ammirati
et al., 1987; Ciric et al., 1987). Radicality though,
is often limited by the proximity of functionally eloquent areas. MR imaging enables exact localization
of lesions in relation to the central sulcus (Berger
et al., 1990; Yousry et al., 1996), but the morphology
and function do not necessarily correlate. Although
functional imaging techniques such as positron emission tomography (PET) or fMRI do allow preoperative localization of eloquent areas, they are only
available in a few centers worldwide and cannot be
used routinely. Furthermore, image guidance alone
is limited by the individual variations in the functional organization of the brain. Thus, intraoperative
functional mapping and monitoring have proven to
be complementary for localizing functionally relevant areas and allowing maximal tumor resection
with minimal morbidity.
The method of somatosensory evoked potential
phase reversal (SEP-PR) was introduced by Goldring
et al. (Goldring, 1978; Goldring and Gregorie, 1984)
based on experience gained in epileptic surgery.
A number of studies have since described its application in tumor surgery (Lesser et al., 1979; Allen
et al., 1981; Allison, 1982, 1987; Desmedt and
Cheron, 1982; Grundy, 1983; Luders et al., 1983;
Amassian and Cracco, 1987; Aiba and Seki, 1988;
*
Correspondence to: Theodoros Kombos, Charite-Universitatsmedizin Berlin, Campus Benjamin Franklin, Hindenburgdamm 30, D-12200 Berlin, Germany.
E-mail: theodoros.kombos@charite.de (T. Kombos).
Allison et al., 1989; Nuwer, 1991; Firsching et al.,

1992; Cedzich et al., 1996).
SEP-PR does allow intraoperative localization of
the central sulcus but yields no functiona information.
Following the technique of SEP-PR, its application as
well as its limitations will be described.
14.2. Method
Phase reversal of somatosensory evoked potentials
is performed by stimulating the median or tibial nerve
contralateral to the lesion. The median nerve contralateral to the lesion is stimulated at the wrist using surface
electrodes placed 1.5 cm apart. For tibial nerve stimulation, the same electrode setup is used, but stimulation
electrodes are placed at the inner ankle.
A stimulus of 5.7 Hz is applied using a constant
voltage electrical stimulator. Current intensity varies
between 5 and 65 mA. Starting with 5 mA, the intensity is gradually increased until slight twitches of the
thumb (median nerve) or the foot (tibial nerve) are
obtained. Control of the proper function of the stimulator is achieved by using a stimulus level above
the stimulation threshold. This minimizes technical
problems during SEP-PR.
Cortical SEPs, and therefore also SEP-PRs, are
recorded by a strip electrode (a row of five or six electrodes embedded in silicon) or by a grid electrode
(2 5 or 3 5) placed on the cortex (Fig. 1). Recordings are performed by a filter bandpass of 100
1,500 Hz, but the filter bandpasses in intraoperative
monitoring are individually set for each operation
theater. The time base for median nerve SEP-PR is
set at 50 ms and for tibial nerve SEP-PR at 80 ms.
Grid or strip electrodes can be used. Using grid
electrodes allows SEPs from larger cortical areas to
be recorded simultaneously. A larger amplifier with
more recording channels would be needed though,
and the software setup for the recording would be
212
Fig. 1. Grid electrode for recording somatosensory evoked

potential phase reversal (SEP-PR). The black line indicates
the central sulcus.
more demanding. Strip electrodes, in contrast, need

up to eight recording channels, depending on their
size. Repositioning of the electrode though, might
be necessary in order to record clear SEP-PR.
Care must be taken for the placement of the
recording electrodes. The following factors must be
taken into consideration during electrode placement:
the electrodes must (1) cross the central sulcus,
(2) cover the hand or leg area of the sensorimotor
gyri, (3) make an angle of 15 with the sagittal direction, and (4) not cover the center of the lesion but
instead lie adjacent to the visible margins of the
tumor mass (Fig. 2). The location of the electrode
must be adjusted to obtain maximum peak amplitudes by moving it in a mediolateral and frontolateral
direction or by rotating it at angles of 15 .
Fig. 2. Diagram showing the correct positioning of the

recording grid. Optimal position must have a 15 angle to
the central sulcus.
T. KOMBOS
For median nerve SEP-PRs, the electrode must be

placed in a cortical area between 3 and 8 cm from the
midline. The largeness of this area is due to the large
cortical representation area of the hand. For tibial
nerve SEP-PR though, the cortical area is limited to
03 cm from the midline. Positioning the strip in the
interhemispheric space is not recommended as this
is technically very difficult and the correct position
cannot be verified.
The distance between the recording anode and cathode, especially in the operation theater, determines the
quality of the recorded potential: the shorter the distance, the better the potential. The recording of the cortical SEP can be performed in two ways. For the
monopolar setup, the cathode is placed on the forehead and every contact of the strip or grid is used as a
cathode (Fig. 3A). For the bipolar setup, neighboring
contacts of the strip are used alternatively as anodes or
as cathodes (Fig. 3B). For example, using a four-contact
strip, the recording setup would be: 12, 23, 34
(Fig. 3). The advantage of the bipolar setup is, as
mentioned before, the better quality of the recorded
potentials, but interpretation is more difficult and
advanced programming of the amplifier is needed.
The time spent on identifying the central fissure
after opening the dura did not usually exceed 5 min.
14.3. Discussion
Phase reversal of somatosensory evoked potentials is
based on the fact that the dipole of the afferent volley
changes from the postcentral to the (Nuwer, 1991).
A stimulus applied on a peripheral nerve is forwarded
through the somatosensory pathway in the contralateral postcentral gyrus. Here, the electrical afferent
volley can be recorded as SEP potential. In other
words, the electrical stimulus applied on a peripheral
nerve generates an electrical dipole on the postcentral
gyrus. The polarity of this dipole changes, however,
on the adjacent precentral gyrus (Fig. 4). Therefore,
a somatosensory potential (N20/P30; N negative,
P positive) can thus be recorded from the postcentral gyrus, and its mirror image (P0 20/N0 30) can
be recorded from the precentral gyrus (Wood et al.,
1988) (Fig. 5). For the tibial nerves, the recordings
read P40/N45 and N0 40/P0 45 respectively (Fig. 6).
The phase-reversal potential is quick and easy
to record, and it provides reliable identification of
the central sulcus until direct mapping can be done.
Many studies have described the application of
SEP-PR in tumor surgery in adults (Lesser et al.,
213
Fig. 3. A: Monopolar recording setup for somatosensory evoked potential phase reversal (SEP-PR). The cathode is placed
on the forehead and every contact of the strip or grid is used as a cathode. B: Bipolar setup for SEP-PR. Neighboring contacts of the strip are used alternatively as anodes or as cathodes.
1979; Woolsey et al., 1979; Allen et al., 1981; Allison,

1982, 1987; Desmedt and Cheron, 1982; Grundy, 1983;
Luders et al., 1983; Amassian and Cracco, 1987; Aiba
and Seki, 1988; Wood et al., 1988; Allison et al., 1989;
Nuwer, 1991; Firsching et al., 1992; Cedzich et al.,
1996). Hence, the technique is not only useful in adults
but also in older children, where maturation of motor
pathways is complete (Sala et al., 2002).
In some cases, the typical phase reversal at 20 ms
is questionable or missing, but characteristic late
waveform components can be recorded from the
electrode lying over the postcentral gyrus (Romstock
et al., 2002). This involves either one single negative
wave with a markedly high amplitude at 35 ms or a
polyphasic sequence of positive and negative peaks
between 25 and 45 ms (Fig. 7). Waveforms may be

more complex if the electrode is lying adjacent to
or directly over the central sulcus. As a result, a varying number of small peaks within the major waves
are seen, giving the impression of a transitional and
less smooth waveform (Romstock et al., 2002).
The most prominent succeeding components for
the median nerve are a postcentral negative wave
between 30 and 40 ms and a precentral N27 in recordings adjacent to the central sulcus. At more posterior
recordings sites, the pattern of the postcentral N35 is
not changed except for a gradual reduction in amplitude. In general, there is no systematic effect of the
tumor masses on SEP-PR despite the fact that some
of the patients had motor or sensory deficits. The
Fig. 4. Schematic explanation of somatosensory evoked

potential phase reversal (SEP-PR). The dipole of the afferent volley changes from the postcentral to the precentral
gyrus.
Fig. 5. Somatosensory evoked potential phase reversal

(SEP-PR) following stimulation of the median nerve.
214
Fig. 6. Somatosensory evoked potential phase reversal

(SEP-PR) following stimulation of the tibial nerve.
situation is more complicated though with larger

tumor sizes, severe preoperative neurological deficits,
or lesions invading the precentral and postcentral gyri.
Under such circumstances, reduced feasibility and
reliability of the SEP-PR technique must be expected.
Wood et al. (1988) have suggested that the large
positivenegative waveform at 25 and 35 ms may serve
T. KOMBOS
as an additional localizing criterion, as it is usually

recorded with the highest amplitude from the postcentral gyrus at a location 10 mm medial to the hand area.
SEP-PR is associated with a success rate of
over 90% for intraoperative localization of the central
sulcus (King and Schell, 1987; Wood et al., 1988). In
our series (Kombos et al., 2000; Suess et al., 2003),
SEP-PR was recorded as 97.14%, which corresponds
to the results of other series. The failure rate may be
accounted for by various factors: (1) tumor-related
shifting of the central sulcus (Cedzich et al., 1996),
(2) misplacement of the recording electrodes in relation
to the anatomical location of the sensorimotor cortex
also called off axis placement (Wood et al., 1988),
and (3) the influence of narcotic agents and brain edema
(Cedzich et al., 1996). Dural adhesions in particular
may have prevented correct placement of the electrodes
and thus resulted in failure-cases.
Babu and Chandy (1997) have reported that the SEPPR could be recorded in all patients with predominant
motor disturbance. But in patients with marked sensorimotor deficits, the attempt to record an SEP was futile.
In contrast, Sonoo et al. (1991) have reported on two
patients with major sensory deficits and localized
lesions of the postcentral gyrus in whom the scalp
recording of N20P20 and later waveforms were eliminated, and only a widespread frontal activity was
obtained. These examples show that perplexing findings must be expected in patients with tumors. About
10% of the patients will not show the classic N20P20
inversion, possibly as a result of three causes: (1) the
tumor desynchronizes the propagated afferent electrical
volleys along the thalamocortical pathway, (2) the mass
effect of the lesion distorts the spatiotemporal projection of cortical electrical dipoles to the brain surface,
and (3) the recording site may not be appropriate for
recording a potential generated in the hand area of the
postcentral gyrus.
While phase reversal of sensory evoked potentials
reliably identifies the central sulcus, it yields no
information about motor function. Thus, anatomical
identification of the central sulcus alone is not a sufficient safeguard against postoperative motor deficits.
Therefore, an additional method must be used to map
motor function.
References
Fig. 7. No somatosensory evoked potential phase reversal

(SEP-PR) is recorded; however, a single negative wave
with a markedly high amplitude at 35 ms from the electrodes 36 is recorded. Electrodes 1 and 2 show no SEP.
Aiba, T and Seki, Y (1988) Intraoperative identification of

the central sulcus: a practical method. Acta Neurochir.
(Wien), 42: 2226.

Allen, A, Starr, A and Nudleman, K (1981) Assessment of
sensory function in the operating room utilizing cerebral
evoked potentials: a study of fifty-six surgically
anesthetized patients. Clin. Neurosurg., 28: 457481.
Allison, T (1982) Scalp and cortical recordings of initial
somatosensory cortex activity to median nerve stimulation
in man. Ann. N. Y. Acad. Sci., 388: 671678.
Allison, T (1987) Localization of sensorimotor cortex in neurosurgery by recording of somatosensory evoked potentials. Yale J. Biol. Med., 60: 143150.
Allison, T, McCarthy, G, Wood, CC, Darcey, TM, Spencer,
DD and Williamson, PD (1989) Human cortical potentials evoked by stimulation of the median nerve.
I. Cytoarchitectonic areas generating short-latency activity. J. Neurophysiol., 62: 694710.
Amassian, VE and Cracco, RQ (1987) Human cerebral cortex responses to contralateral transcranial stimulation.
Ammirati, M, Vick, N, Liao, Y, Ciric, I and Mickhael, M
(1987) Effect of the extent of surgical resection on survival and quality of life in patients with supratentorial
glioblastomas and anaplastic astrocytomas. Neurosurgery, 21: 201206.
Babu, KS and Chandy, MJ (1997) Reliability ofsomatosensory evoked potentials in intraoperative localization of
the central sulcus in patients with perirolandic mass
lesions. Br. J. Neurosurg., 11: 411417.
Berger, MS, Cohen, WA and Ojemann, GA (1990) Correlation of motor cortex brain mapping data with magnetic
resonance imaging. J. Neurosurg., 72: 383387.
Cedzich, C, Taniguchi, M, Schafer, S and Schramm, J (1996)
Somatosensory evoked potential phase reversal and
direct motor cortex stimulation during surgery in and
around the central region. Neurosurgery, 38: 962970.
Ciric, I, Ammirati, M, Vick, N and Mickhael, M (1987)
Supratentorial gliomas: surgical considerations and
immediate postoperative results. Gross total resection
versus partial resection. Neurosurgery, 21: 2126.
Desmedt, JE and Cheron, G (1982) Somatosensory evoked
potentials in man: subcortical and cortical components
and their neural basis. Ann. N.Y. Acad. Sci., 388: 388411.
Firsching, R, Klug, N, Borner, U and Sanker, R (1992)
Lesions of the sensorimotor region: somatosensory
evoked potentials and ultrasound guided surgery. Acta
Neurochir. (Wien), 118: 8790.
Goldring, S (1978) A method for surgical management of
focal epilepsy, especially as it relates to children.
J. Neurosurg., 49: 344356.
Goldring, S and Gregorie, EM (1984) Surgical
management of epilepsy using epidural recordings to
localize the seizure focus. J. Neurosurg., 60: 457466.
Grundy, BL (1983) Intraoperative monitoring of sensoryevoked potentials. Anesthesiology, 58: 7287.
Hirakawa, K, Suzuki, K, Ueda, S, Nakawa, Y, Yoshino, E and
Ibayashi, N (1984) Multivariate analysis of factors
215
affecting postoperative survival in malignant astrocytomas. J. Neurooncol., 12: 331340.
King, RB and Schell, GR (1987) Cortical localization and
monitoring during cerebral operations. J. Neurosurg.,
67: 210219.
Kombos, Th, Suess, O, Funk, Th and Brock, M (2000)
Intraoperative mapping of the motor cortex during surgery in and around the motor cortex. Acta Neurochir.
(Wien), 142: 263268.
Laws, ER, Taylor, WF, Clifton, MP and Okazaki, H (1984)
Neurosurgical management of low grade astrocytoma of
the cerebral hemispheres. J. Neurosurg., 61: 665673.
Lesser, RP, Koehle, R and Luders, H (1979) Effect of stimulus intensity on short latency somatosensory evoked
potentials. Electroencephalogr. Clin. Neurophysiol.,
47: 377382.
Luders, H, Lesser, RP and Hahn, J (1983) Cortical somatosensory evoked potentials in response to hand stimulation. J. Neurosurg., 58: 885894.
Nuwer, MR (1991) Localization of motor cortex with
median nerve somatosensory evoked potentials. In:
J Schramm and A Mller (Eds.), Intraoperative
Neurophysiological Monitoring. Springer Verlag, Berlin
Heidelberg, pp. 6371.
Romstock, J, Fahlbusch, R, Ganslandt, O, Nimsky, C and
Strauss, C (2002) Localisation of the sensorimotor cortex during surgery for brain tumours: feasibility and
waveform patterns of somatosensory evoked potentials.
J. Neurol. Neurosurg. Psychiatry, 72: 221229.
Sala, F, Matevz, JK and Deletis, V (2002) Intraoperative
neurophysiological monitoring in pediatric neurosurgery: why, when, how? Childs Nerv. Syst., 18: 264287.
Sonoo, M, Shimpo, T and Takeda, K (1991) SEPs in two
patients with localized lesions of the postcentral gyrus.
, Suess, S, da Silva, C, Brock, M
Suess, O, Ciklatekerlio, O
and Kombos, Th (2003) Klinische Studie zur Anwendung der hochfrequenten monopolaren Kortexstimulation
berwachung
(MKS) fur die intraoperative Ortung und U
motorischer Hirnareale bei Eingriffe in der Nahe der
Zentralregion. Klin. Neurophysiol., 34: 127137.
Wood, C, Spencer, D, Allison, T, McCarthy, G,
Williamson, P and Goff, W (1988) Localisation of
human sensorimotor cortex during surgery by cortical
surface recording of somatosensory evoked potentials.
J. Neurosurg., 68: 99111.
Woolsey, CN, Erickson, TC and Gilson, WE (1979) Localization in somatic sensory and motor areas of human
cerebral cortex as determined by direct recording of
evoked potentials and electrical stimulation. J. Neurosurg., 51: 476506.
Yousry, TA, Schmid, UD, Schmidt, D, Hagen, T, Jassoy, A
and Reiser, MF (1996) The central sulcal vein: a landmark for identification of the central sulcus using functional magnetic imaging. J. Neurosurg., 85: 608617.
Section II.2
Motor Evoked Potentials

M.R. Nuwer (Ed.)
218
CHAPTER 15
Motor EP physiology, risks and specific

anesthetic effects
Henricus Louis Journee*
Department of Neurosurgery, University Medical Center Groningen, 9700 RB Groningen, The Netherlands
and
Department of Orthopedic Surgery, Sint Maartenskliniek, 6522 JV Nijmegen, The Netherlands
15.1. Introduction
The response to transcranial stimulation of the brain
is an important means of assessing motor pathways
in the anesthetized patient. This chapter gives an outline of the physical, neurophysiological, and anesthetic background to elucidate practical aspects in
neurophysiological monitoring in regard to its clinical use and safety aspects.
15.2. Neurophysiological circuitry of motor
evoked potentials
A simplified overview of the neural circuits involved
in motor evoked potential (MEP) is shown in Fig. 1.
The gray areas indicate:
(I) Cortical level. The upper motor neuron
receives synaptic connections from facilitating (light
colored) and inhibiting (black) neurons. Transcranial
electrical or magnetic stimulation depolarizes axons
in the cortical layer. The i indicates possible locations where horizontal-oriented axons of interneurons in the cortical layer generate action potentials
(APs) that are conducted directly, or relayed via
interneurons, to the upper motor neuron. This results
in indirect (I) waves in the corticospinal tract (CT) in
spinal epidural recording. Direct (D) waves are generated at the initial segment (IS) of the motor neuron,
its recurrent connection (not shown), or at the Ranvier nodes (Amassian, 2002). The D-waves are indicated by d at these locations. The arrows at i
*
Correspondence to: Henricus Louis Journee, M.D., Ph.D.,

Department of Neurosurgery, University Medical Center
Groningen (UMCG), P.O. Box 30.001, 9700 RB Groningen,
The Netherlands.
Tel.: 31-50-361-2440; fax: 31-50-361-1715.
E-mail: h.l.journee@nchir.umcg.nl (H.L. Journee).
and d indicate optimal electrical field direction for

axonal depolarization.
(II) Segmental level. The lower motor neuron
(aMN) receives monosynaptic connections from the
CT facilitating and inhibiting interneurons like a
Renshaw cell (RC). Motor responses can also
be conducted along indirect pathways (not shown)
like rubro-spinal, tecto-spinal, vestibulo-spinal, and
cerebello-spinal tracts in the lateral and anterior part of
the spinal cord. When stimulated transcranially, the
motor potentials are unidirectionally conducted to the
aMN. Therefore, transcranial electrical stimulationMEP (TES-MEP) can be utilized for monitoring of the
integrity of the anterior and lateral part of the spinal cord
exclusively. This selectivity is not available in spinal
stimulation since the sensory tracts of the dorsal columns
are stimulated as well. This results in antidrome conduction of APs to the aMN resulting in an mMEP, in the target muscle. Since imminent damage of the anterior
spinal cord can be missed, this monosynaptic response,
the so-called upper-H-reflex, should be considered
inappropriate for motor monitoring of the spinal cord
(Deletis, 2001; Minahan et al., 2001; Jones et al., 2003).
Also shown is a monosynaptic circuit of the
H-reflex (HR). This circuit shares also a part of
the sensory axons of the neural circuit of the upper
HR. Many more neural circuits on segmental and
higher levels can be involved when sensory nerve
fibers are stimulated. Examples are reciprocal inhibition, short- and long-loop reflexes, and the polysynaptic bulbo-cavernosous reflex (BCR).
(III) Neuromuscular level. Action potentials from
axons of the motor neurons are transferred to the
muscle fibers of the motor unit via neuromuscular
(NM) junctions. The APs may originate from motor
neurons as well as from stimulation of peripheral
nerves. Root or nerve stimulation is used in nerve
MOTOR EVOKED POTENTIALS
219
Transcranial stimulation
i
i
i
d
I
Cortex
IS
Upper motor neuron
RN
Spinal cord
stimulation
(upper HR)
Epidural
MEP
mMEP
Peripheral
stimulation
(HR)
NM
junction
RC
II
Lower motor neuron
III
Nerve/root
stimulation
(MR)
Segmental level
Muscle
nMEP
Fig. 1. Schematic presentation of the basic neurophysiological circuitry involved in motor responses. An explanation is
given in the text.
mapping of cranial nerves (Mller, 2002) down to

nerve roots in the conuscauda region (Vodusek and
Deletis, 2002), and in identification of branches in cervical, lumbar, and sacral plexuses. The evoked potentials are recorded from peripheral nerves (nMEPs) or
from muscles (mMEP). The muscle response that
directly results from nerve stimulation is defined as
the M-response (MR). MRs may be an unwanted side
effect from TES of peripheral axons. An example is
the MR in the orbicular oris muscle using TES. This
response can be distinguished from the multipulse
TES response by a 57 ms shorter latency time and
persistance at 1 pulse (Dong et al., 2005).
15.3. Stimulation techniques
15.3.1. Stimulation of axons
is useful to verify preferential locations along an axon

with lowest firing thresholds. These are at borders of
compartments with different conductivities like exit
places of roots at the dura and at bends in axons
(Maccabee et al., 1993). The cable model starts at
the IS. The relatively low stimulation threshold of
the IS may be explained by an unbalanced counteracting transmembrane current due to a missing neighbor myelin segment and a different transmembrane
current density. Due to their large internodal lengths,
large fibers are activated at lower stimulus currents
and have higher conduction velocities than smaller
ones (McNeal, 1976; Coburn, 1989). When myelin
sheets are affected in chronically compressed nerve
roots, higher electrical stimulus intensities are required
for activation which is found in testing of adequate
placement of pedicle screws (Holland et al., 1998).
15.3.2. Transcranial stimulation
A cable model is useful to describe the electrical

properties of an axon (McNeal, 1976). An activation
function (Rattay, 1987) expresses changes of the electrical field over distance over the axons and indicates
where axon membranes will be depolarized or hyperpolarized by the stimulus. The activation function
15.3.2.1. Comparison of transcranial magnetic

stimulation and TES
Transcranial magnetic stimulation (TMS) (Barker
et al., 1985) and TES (Merton and Morton, 1980)
are both widespread used to elicit MEPs to study
220
H.L. JOURNEE
electrode montage and stimulus intensity. One can

define small focal fields, as well as large fields down
to the foramen magnum (Rothwell et al., 1994). At
this depth, the latency time of D-waves is reduced by
about 1.8 ms when compared to cortical level. The
dependence of latency on stimulus intensity is typical
for D-waves and can be used to differentiate
from I-waves of which the latency times remain
unchanged since they originate from the cell
membrane of upper MNs.
TES is widely used for neuromonitoring. TMS is
more vulnerable for anesthetic agents since cortical
synaptic transmission contributes relatively more to
the CT and activates a relatively small fraction of CT
axons. As a consequence, D-waves and mMEP amplitudes will mostly be smaller than in TES. TMS can be
used for monitoring (Rohde et al., 2003; Hargreaves
and Watt, 2005). However, the anesthetic regime is
more restricted than in TES.
the physiology and pathophysiology of the corticospinal projections in man. The methods have in common that they administer energy from outside the
skull and generate electrical fields in the cortical
layer as well as in the CT. Epidural recording from
the surface of the spinal cord in man has shown that
both types of stimuli can excite CT neurons consistent with direct activation of the upper MNs in the
cortex. These can be observed as D-waves. These
usually can be followed by transsynaptically generated I-waves (Patton and Amassian, 1954; Kernell
and Wu, 1967; Boyd et al., 1986; Rothwell et al.,
1991; Edgley et al., 1997; Houlden et al., 1999).
The recording of D-waves is utilized in intraoperative monitoring with electrodes in the epidural space
of the spinal cord (Boyd et al., 1986; Hicks et al.,
1992; Kothbauer et al., 1997).
TES and TMS also show distinct differences. TMS
induces an electrical field by a fast-changing strong
magnetic field by means of a coil over the head. The
field activates neurons at a restricted penetration depth
of several centimeters. A small focal field can be
obtained from a figure of eight coils (Thielscher and
Kammer, 2002). The field is predominantly horizontally oriented. In contrast, in TES, one is able to define
the orientation and volume of the electrical field by the
15.3.3. Geometry of TES: intensity and D-waves,

somatotopic selectivity, and electrodes
The conversion into APs occurs at axons (Nowak and
Bullier, 1998a,b). Maximum depolarization is usually
seen at the negative side of the electrical field.
50
Cathode
145
140
130
40
145
k. 1.2 V/cm2
1
CSF
ventricles
140
k. 0.45 V/cm2
130
0
10
30
25
0.4
20
0.3
Brain
Side ventricle
level
35
Activation functions
a) With ventricles
k. 2.8 V/cm2
Anode
at 135
k. V/cm2
45
50
b) No ventricles
k. 0.37 V/cm2
135
0.2
SKULL
10
CSF layer
inner compact layer
spongeous layer
outer compact layer
scalp
Cortex surface
15
0.1
0
10
50
Radius in mm
(Fig. 2 continued)
221
Voltage distribution in the corticospinal tract

10
100 V
C3
5
C4
3v
2
100 V
1
0
V
3
2.5
2
1.5
1
0.5
0
0.5
1
1.5
10
10
0
5
10 10
0.15
cm
10
Electric field strength
V/cm
0.05
0.1
Activation function
V/cm2
0.1
7 cm
B
Fig. 2. A: 2D volume conductor finite element model of the head, showing three radial trajectories near the anode of which
the activation functions are shown at the right. (a) Upper graph are from the model with ventricles filled with CSF and
(b) the lower graph concerns the activation function of the trajectory through the anode (135 ). The AFs result from polynomial interpolation of data points. All activation functions are expressed in arbitrary values so that they can be compared
are expressed in gray intensity levels. The gray scale is adjusted to the jEj
value
with each other. Isoelectrical fields (jEj)
range within the area enclosed by the skull for optimal visualization of the condensation spots aside from the ventricles
and below the stimulating electrodes. B: 3D volume conductor finite element model of the head representing the voltage
distribution along the CT with 4 cm electrodes. The CT trajectory is derived from diffusion tensor imaging (DTI-MRI).
Plotted below are the electric field strength and activation function along the CT. The maximums of the activation function
are indicated by dashed lines at: 12 cm under the skull absolute maximum, at 23 cm, and in the 67 cm section which is
at about peduncle level. The TES electrodes are modeled at locations C3 and C4 at pulse voltages of 100 V and 100 V.
Electrical volume conductor models explain why the

lowest thresholds are at the anode. Most used are
simple geometric three-layer models of the head to
analyze the locations with lowest stimulus thresholds
for activation of CT axons (Suihko, 1998; Stecker,
2005). The places where APs along the CT are generated depend on the intensity of TES. Bends in the CT
are likely places where D-waves originate because

the latency time often shows stepwise decrements
during gradual increase of the intensity (Rothwell
et al., 1994). Instead of clear steps, we also encountered continuous decreasing latency times in some
patients. This high variability in the course of the
latency could be explained by effects from highly
222
conducting CSF in ventricles. Their size varies markedly between patients. Fig. 2A shows the electrical
field along the CT with and without the presence of
side ventricles containing CSF. The ventricle volume
increases with age (Jernigan et al., 2001; Keats and
Sistrom, 2001). The high-conducting ventricles drain
TES current. This causes spots of dense isopotential
lines. These agree with the peaks of the radial activation functions (a) in Fig. 2A. The presence of the
CSF affects the activation function up to a distance
of several centimeters away from the side ventricles.
For a distance smaller than 12 mm, the peak of the
activation function is higher than the maximum value
of the activation function without CSF. This would
imply that, in this model, at threshold level, APs will
be generated markedly below cortex level. The
course of an activation function along the CT down
to brainstem level obtained by a 3D model is given
in Fig. 2B where the actual course of the CT is
obtained from a diffusion tensor imaging (DTIMRI). The activation function show peaks at CT
locations at bends near the location of the (not modeled) side ventricles and at the level of the peduncles.
One can select specific motor areas in the brain for
monitoring in intracranial surgical procedures. Nearthreshold TES intensities are highly selective for
motor-cortical areas near the electrodes unless when
the patient has large ventricles with CSF. The somatotopic selectivity degrades at high intensities and large
ventricles. Higher intensities are useful when muscles
in the upper and lower extremities are to be monitored
simultaneously like in spinal surgery. Since the size of
ventricles may differ markedly between patients, one
cannot be sure about the actual depth where TES
threshold intensity APs will be generated in the corticospinal axons. This is an important argument in
MEP monitoring during supratentorial intracranial
surgical procedure to apply cortical instead of TES.
For Cz0 Fz montage and 100 ms TES pulse width,
threshold voltages are 56 23 V (mean S.D.; n
286) (Journee et al., 2004a). C3C4 electrode montages show lowest voltage thresholds for muscle
MEPs in the upper limb whereas in Cz0 Fz montages
show the lowest threshold voltages pertaining to the
lower extremities. The latter is useful for monitoring
procedures of the spine below cervical level and C3
C4 montages for surgical procedures at cervical
level. When electrodes are placed in close proximity
as, for example, at C1C2, the currents are forced to
flow in a more horizontal direction along the cortex
permitting more I-waves to be generated.
H.L. JOURNEE
15.3.4. Cortical stimulation

In supratentorial intracranial surgery, the dura
becomes often detached from the inner surface of
the skull (Kombos et al., 2000b) so that TES stimulation is often replaced by direct cortical stimulation of
a cortical electrode strip (Kombos et al., 2000b;
Neuloh et al., 2004). This method has a high somatotopic selectivity which means that precise placement is important. The strip can adequately be
placed after sensory mapping using phase reversal
to identify the Rolandic fissure, and subsequent
motor mapping anterior to the fissure to confirm
the electrode location with the lowest current threshold. This applies to MEPs of muscles in the contralateral face, upper, or lower extremity. Like in
TES, high-frequency pulse trains are applied unipolar by a current stimulator of 47 anodal monophasic pulses, pulse width 50500 ms, interpulse
interval (IPI) 24 ms. Threshold currents are in a
range of 525 mA. For cathodal electrode, one can
use location C3 or C4, whatever is at the contralateral
side. Sensitivity and specificity are over 95%
(Cedzich et al., 1996; Kombos et al., 2000a). Motor
responses are generated in a restricted number of muscles. This results in minimal movement. Therefore,
continuous monitoring with stimulus rates up to 2 Hz
is usually possible. One should be aware that during
monitoring, the electrode strip may shift up to 2 cm
over the cortex (Suess et al., 2002). The unipolar multipulse stimulation technique differs essentially from
the Penfield technique which applies pure cortical
bipolar and continuous stimulation with a frequency
of 5060 Hz, 1 ms pulse width, and only in cases when
the motor cortex is surgically exposed. Furthermore,
at such frequencies and train durations, seizures are
easily induced (Deletis, 2002).
15.3.5. Electrical aspects of transcranial
stimulation and electrodes
Figure 3 shows that the total impedance RT of TES
electrodes is composed of two local electrode impedances, Rlocal, that are connected in series with net
impedance of the head, Rnet. Local electrode resistances depend on the type of electrode. This may be
of a concern when a voltage stimulator is used. There
is a high linear correlation between the TES voltage
threshold Vth and RT when RT is higher than 460 O,
whereas no correlation is present below 460 O
(Journee et al., 2004a). The input impedance RT for
223
RT = 2Rlocal+ Rnet
The stimulation voltage drops when the internal

Rint
Rlocal
Rnet
Rlocal
TES stimulator
Fig. 3. Electrical circuit of TES using a voltage stimulator.

Rlocal: local electrode impedance of stimulus electrode
which depends on electrode type, Rnet: enclosed net impedance of the head, Rint: internal resistance of the TES
stimulator, and RT: total electrode impedance.
corkscrew electrodes is in a range of 380650 O,

which means that voltage thresholds may depend on
the local electrode impedance. The input impedance
also depends on the intensity of the TES pulse
(Merton and Morton, 1980). Linear impedances are
inaccurate under conditions of TES. A dynamic
impedance is derived from administered TES voltages
and currents. The impedance of corkscrew electrodes
(C3C4) decrease by 30250 O and approach asymptotic levels at 230280 O when TES intensity increases
from 0 to 100 V. This impedance reduction is from the
local electrode impedance and thus dependent on electrode type. The net head impedance Rnet is independent
of electrode types. It can be measured exclusively by a
tetrapolar recording technique as applied in contemporary impedance tomographic equipment (Tidswell
et al., 2001) applying needle electrodes near the TES
electrodes. When excluding the capacitive component
of the impedance, the net head impedance Rnet was
100200 O. The results agree with data from literature:
131 15 O (46 kHz), 148 16 O (26 kHz) (Seipel,
1970) and 130180 O (30 kHz) (Lifshitz, 1967) using
electrodes with large contact surfaces.
Two facts are important for monitoring with
voltage TES:
The local impedance of corkscrew electrodes con-
tribute 3070% to the total impedance. Changes

in the local electrode impedance may affect TES
thresholds strongly and may play a role in threshold
monitoring (Calancie et al., 1998, 2001).
resistance of the stimulator is high. The internal

resistor of the often used D185 voltage stimulator
of DigitimerTM is 120 O (Digitimer Isolated Multipulse Stimulator Model D185, 1999). This is almost
equal to the net head impedance. The delivered
voltage may be reduced by over 60% from the
defined voltage. Moreover, pulse waves may get
distorted (Journee et al., 2003). This is not important for monitoring when stimulus voltages are used
within a relative context. It is recommended to use
a voltage stimulator along with electrodes with
large contact surfaces to restrict the sensitivity to
variations in local electrode impedance.
Stimulus thresholds of current stimulators are nearly
independent on local electrode impedances. Any
needle electrode can be used. However, TES current
thresholds are sensitive to scalp edema and increase
due to current shunting along the good conducting
edema leaving less current for the transcranial route.
15.4. Anesthetic effects
This paragraph gives a brief outline on the effects of
anesthetic agents pertaining to monitoring of motor
tracts. The sections follow the scheme of Fig. 1.
15.4.1. Axonal conduction
The nervous system outside the gray areas concern
the propagation of APs along uninterrupted axons in
motor tracts or nerves. Since no synapses are
involved, the production of recorded waves will be
robust and are highly resistant to the depressing
effects of almost all anesthetic agents. High concentrations of isoflurane may decrease the D-wave
amplitude (Sloan and Heyer, 2002). However, this
dependence is not yet confirmed by other studies.
Conduction blocking agents are not used under
general anesthesia unless when used deliberately in
a provocative test. An example is the use of xylocaine to check the functional integrity of sensory
and motor pathways to warn the neuroradiologist
for impending neurological damage in the endovascular treatment of spinal vascular malformations
(Sala et al., 2001).
15.4.2. Neuromuscular junction
Neuromuscular junctions (gray area III in Fig. 1) are
involved in any modality where muscular potentials
224
are recorded. This concerns the recording of spontaneous muscular activity and responses from direct
nerve or root stimulation (MR and HR) from pedicle
screw testing (Lenke et al., 1995; Toleikis, 2002),
nerve mapping in the conus-cauda region and posterior fossa (Mller, 2002; Quinones-Hinojosa et al.,
2004; Ashram et al., 2005; Shils et al., 2005) to
mMEPs from TES. When no muscle relaxants are
used and when the NM junction is not affected by
pathology, each AP from an axon of a motor unit
always will be transferred to its muscle fibers when
it arrives after the refractory interval of a previous
AP. Therefore, the use of muscle relaxants should
best be avoided in mMEP monitoring.
Unless when considered as necessary to apply NM
blocking agents, the NM blockade level should
be maintained constant for acceptable monitoring. This
can be realized by feedback of MRs (T1) produced by
supramaximal stimulation of a peripheral motor nerve
using standards NM function measuring techniques.
Titration levels mostly used in monitoring are in a range
of 530% compared to baseline (Kalkman et al., 1992;
Stinson et al., 1994; Lang et al., 1996a,b; De Haan
et al., 1997; Ubags et al., 1999). One should realize that
the NM function may differ markedly between muscle
groups and, despite a stable relaxant level, vary
in time. This adds an uncontrollable variable in the
interpretation of MEPs that would reduce the specificity of muscle MEP monitoring.
On the other hand, for recording of epidural or neurogenic responses, complete or near-complete muscular blockade with a fast-acting relaxant like atracurium
could be considered to suppress interfering responses
from adjacent muscles.
15.4.3. Neural synaptic transmission

The synaptic transmission to upper and lower motor
neurons are indicated in the gray areas at cortical (I)
and spinal (II) levels in Fig. 1. The target of most anesthetic actions appears to be at neural synapses, especially the gamma amino butyric acid (GABA) and
N-methyl-D-aspartic acid (NMDA) receptors which
mediate ion channels (Na, Cl, Ca2). NMDA receptors are involved in facilitation and GABA receptors in
inhibition of neural membranes. Most anesthetics are
either NMDA antagonists (ketamine, nitrous oxide,
xenon) or GABA agonists (barbiturates, propofol, volatile agents) and therefore cause hyperpolarizing
effects on cortical and segmental levels.
H.L. JOURNEE
In awake patients, TMS by a single pulse can often

generate mMEPs without conditioning by a voluntary
slight muscle contraction. The membrane potential of
a motor neuron is sufficiently high to generate an AP
with 1 stimulus pulse (Fig. 4A), whereas muscle
responses are easily abolished in anesthetized conditions. High-frequency double and multipulse stimulation have to be used instead for successful monitoring
(Kalkman et al., 1995; Pechstein et al., 1996). The
membrane potential raises by summation of postsynaptic excitatory postsynaptic potentials (EPSPs) until
the threshold level is reached for generation of an
AP (Fig. 4B). Such summation applies to the upper
(Thomson et al., 1993) as well as to the aMN (Edgley
et al., 1997). Lengths of EPSPs are about 15 ms
(Thomson et al., 1993; Nowak and Bullier, 1998b).
Facilitation by multipulse stimulation has made
TES-mMEP monitoring successful in many patients
(Ubags et al., 1998; Lo et al., 2004). Total intravenous
anesthesia with propofol and remifentanil seems to be
particular well suited for intraoperative MEP monitoring. Besides its unique pharmacokinetic properties
which allows rapid and reliable emergence from anesthesia even after prolonged application, remifentanil
has a much wider dosage window with respect to
recording of myogenic MEP than all other opioids.
This allows a sometimes needed tapering of propofol
dosage in order to obtain successful recordings of
myogenic responses (Drover et al., 2004). It could be
worthwhile to guide the reduction of the dosage of
propofol by a monitor for the depth of sedation, such
as by the Bispectral Index, to ensure a still adequate
level of anesthesia. Ideally, both intravenous agents
will be administered with a target-controlled infusion (TCI) pump which maintains a selected plasma
concentration and thus may decrease the variability
of MEP amplitudes (Scheufler and Zentner, 2002;
Scheufler et al., 2005). Multipulse facilitation still
may not be sufficient to elicit APs (Fig. 4C) when
neurons are extremely hyperpolarized.
In children with immature corticospinal motor
pathways and in patients with preexisting neurological deficits and muscular diseases, successful recordings of MEP responses are sometimes difficult to
obtain and will be abolished by commonly used
dosages of anesthetics. Under these circumstances,
ketamine can be tried to replace propofol partly or
complete in order to improve recording conditions
(Inoue et al., 2002; Erb et al., 2005).
Although volatile agents up to 1 MAC may still
allow recording of MEP, they are not recommended
225
10 ms
HR test pulse
TES conditioning pulses

(arrival at MN)
Fig. 4. Artist rendering of temporal summation of EPSPs resulting from one or a train of high-frequency stimulation pulses
(bottom of each panel) on a membrane potential of a motor neuron. A: Relatively high membrane potential near threshold
level. Only one pulse is required to cross the gap (gray shaded) for generation of an action potential (AP), B: hyperpolarized
level requiring a TES train with four pulses for AP elicitation, C: blocked neural transmission due insufficient facilitation
by any many pulses at strong hyperpolarization, and D: generation of an AP using the summating effect of EPSPs from
conditioning TES together with a peripheral test pulse of an H-reflex. The shown stimulus pulses are translated to their
arrival at the aMN.
since they set unnecessary limits for transcranial MEP

monitoring (Pelosi et al., 2001; Chen, 2004). The same
holds true for nitrous oxide which allows a reduction of
the concentration of volatile anesthetic agents and has
some weak analgesic properties.
Nitrous oxide should not be used when H- and other
segmental reflexes are to be included in the protocol
for neurophysiological monitoring. In our experience,
when nitrous oxide (max 60%) is used in combination
with propofol (1.5 mg/kg/h) and remifentanil (0.2
0.4 mg/kg/h), HR responses gra-dually decrease and
often disappear after 2030 min while TES-mMEP
amplitudes were preserved, whereas HR and TESmMEP responses remained stable using even at doubled to tripled propofol levels combined with various
opioids without nitrous oxide.
15.5. Physiology
Once APs have been generated in axons by stimulation, their further processing is controlled by neural
circuits.
15.5.1. Epidural recording

Epidural recordings permit study of transsynaptic transmission in cortical neurons on cortical level (gray area I
in Fig. 1) (Patton and Amassian, 1954; Kernell and Wu,
1967). This permits study of interaction of anesthetic
agents on neurotransmitter receptors on cortical level,
the relationship between D- and I-waves and mMEPs
(Deletis et al., 2001), and the distribution of conduction
velocities along the CT (Edgley et al., 1997). Most
efficient stimulation can be expected at dense packing
of D- and I-waves at relatively high stimulation intensities and short IPIs. High stimulation intensities allow
optimal packing at ISI 2 ms and for low intensities,
ISI 4 ms is an optimal IPI (Novak et al., 2004).
D-waves have been used to monitor integrity of
CT fibers during spinal tumor removal, reconstructive surgery of the spine, and intracranial surgery
(Boyd et al., 1986; Hicks et al., 1992; Kothbauer
et al., 1997). D-waves are sensitive and specific
for impending lesions of the anterior spinal cord.
Immediate response changes can be expected from
226
mechanical impact on the CT. Ischemic changes may

be noticed with a delay up to 20 min (Dong et al.,
2002; Lips et al., 2002a,b). Since the amplitude of
D-waves significantly depend on the position of the
recording electrodes, D-wave recording are unreliable during corrective surgery of the spine and may
cause false positive results (Ulkatan et al., 2006).
TES is not necessarily specific for propagation of
APs along the CT alone. A few arguments can be given.
(1) High TES intensities allow to generate APs at brainstem level where vestibulo-spinal, reticulo-spinal,
cerebello-spinal, and rubro-spinal tracts origi-nate. This
agrees with a significant increase of D-wave amplitude
that we often notice when latencies were reduced by
over 1.51.6 ms. (2) One cannot exclude conduction of
motor potentials via cortico-rubro-spinal and corticovestibulo-spinal and cortico-bulbo-spinal pathways.
(3) Facilitation by 12 Hz multi-train TES in a
non-deeply anesthetized patients shows unaltered
D-wave amplitudes while mMEPs can be strongly
facilitated (Deletis, 2002). All these motor pathways
are located in the anterior and lateral part of the spinal
cord and probably will not affect the specificity
of TES.
15.5.2. H-reflex
The HR is a simple monosynaptic reflex that is primarily used to test the excitatory state of aMNs.
The amplitude of the HR, although non-linearly, is
proportionally related to a mean membrane potential
of a group of aMNs. Except for extreme hyperpolarization, it can bridge a wide gap between the membrane potential and threshold level for the generation
of APs.
Prior to the introduction of TES, the HR has been
used for monitoring of the motor system and still is
considered by some authors as alternative for motor
monitoring by TES because of its high sensitivity to
segmental ischemic changes or to supra-segmental
innervation (Leppanen et al., 1995; Leppanen,
2006). Like in TES-mMEP, the HR directly responds
to local ischemic changes of the anterior and lateral
spinal cord. The reflex is furthermore modulated by
unknown local neural activity and supra segmental
activity propagated via motor tracts. This activity
may be modulated by impact on higher motor tracts
due to surgical action. The activity conveyed from
different motor tracts can be inhibitory as well as
facilitatory. This may complicate the interpretation
of events.
H.L. JOURNEE
The aMN is the final common path for motor

outflow to peripheral muscles. Its membrane potential is modulated by all neural systems related to
movement. All contribute by excitatory post synaptic
potentials (EPSPs) and inhibitory post synaptic
potentials (IPSPs) to the excitatory state of the membrane potentials of aMNs. Each stimulus from whatever location may contribute to the membrane
potential with a characteristic response pattern. This
pattern can be retrieved by repeated measurements
where the conditioning time interval (CTI), which is
defined as time between the conditioning stimulus
and the subsequent test stimulus of the HR, is stepwise increased. Many spinal circuits on segmental
and higher levels have been dismantled by the HR
(Pierrot-Deseilligny and Burke, 2005).
15.5.3. TES-mMEP
The course of the aMN membrane potential from
multipulse TES as measured by the HR can help to
explain how mMEPs are generated and describe
how conditioning from various possibilities for stimulation can be optimized for conditioning facilitation.
HR studies with near-threshold single or double pulse
TMS and TES show arrival of the D-waves and
subsequent I-waves at CTI 5 to 1 ms. This is
sometimes followed by a second facilitation interval
at CTI 316 ms (Cowan et al., 1986; Taniguchi
and Schramm, 1991; Baldissera and Cavallari, 1993;
Petersen et al., 1998; Terao et al., 2000; Chan
et al., 2002). Above stimulus threshold level and
mild anesthesia, the situation becomes complicated
due to contribution of many neuronal systems that
propagate APs parallel to the CT. All EPSPs and
IPSPs from these systems will add to the simple
course of the aMN membrane potential in Fig. 4B.
This results in a complex course of the membrane
potential as illustrated by a HR function from
a patient (Fig. 5A). The HR is conditioned by TES
with a train of four pulses. The function has a threemodal course. The EPSPs of the D-waves contribute
to the flank of the first peak. These will fade away with
the duration EPSPs as shown by the white area at the
beginning of the curve. The curve progresses to a
maximum from additional EPSPs which result from
I-waves from the CT. The progressing time after TES
permit arrival of APs from an increasing variety of
neural circuits and pathways to affect the aMN membrane potential. The first peak may cause a subgroup
aMNs to fire which contributes to the first part of the
227
4
Facilitation
From D-wave
HR (mV)
Except from D-wave

MR
2
2 mV
HR
1
HR
reference
0
20
30
130
80
Conditioning-test interval (ms)
180
1.0
0.2 mV
5
50 ms
0.0
0
50
100
ITI (ms)
150
200
Fig. 5. A: H-reflex (HR) conditioning curve conditioned by TES of a scoliosis patient, B: M- (MR) and HR responses,
C: unconditioned CMAP from TES, and D: double train TES curve. The graph is shifted to the right for comparison
with (A). Responses are from the right gastrocnemius muscle at stimulation of the posterior tibial nerve. TES: CzFz,
PW 100 ms, 4 p/train, IPI 2 ms, and 110% threshold voltage. Note that the mMEP amplitude from TES is sufficiently
small not to cause erroneous interference with the HR.
compound muscle action potential (CMAP). Another

group of motor units are (second peak) likely to fire
about 20 ms later at the second peak and contribute
to a subsequent part of the CMAP. This can explain
why the CMAP (Fig. 5B) has longer duration when
compared to the width of the HR (Fig. 5C). A third
broad lowest peak shows a maximum at 110 ms.
15.5.4. Facilitation by conditioning
The principle of facilitation by conditioning stimulation is to apply a test stimulus when the aMN
membrane is maximum excitable after conditioning.
Conditioning as well as test stimuli recruit their

own motor neuron pools. These overlap partly. Overlap is an essential requirement for facilitation. The
size of the non-overlapping parts may affect the
accuracy in predicting facilitation effects from conditioning curves of the conditioning and test stimuli.
Moreover, the size of the overlapping group of aMNs
may vary in time, especially when a subgroup of
motor neurons already have fired.
There are many choices for conditioning as well as
for the test stimulation. Fig. 5A illustrates a powerful
technique with conditioning multipulse TES and test
stimulation by the HR. This technique results in firing
228
H.L. JOURNEE
of aMNs, even when multipulse TES cannot bridge the

potential difference to firing threshold. As shown in
Fig. 4D, the EPSPs from the HR stimulus come on top
of the plateau from four EPSPs and brings the aMN to
fire. Optimal choices for maximum membrane excitation are at the instants of the peaks at CTI 0, 20, and
130 ms. One can monitor TES responses by pairwise
comparison of conditioned with unconditioned HR
amplitudes. Theoretically, combination of these different stimulation techniques will combine their characteristics. The HR is selective for a few muscle groups like
for example the gastrocnemius and tibial muscle. Contraction of these muscles are not a problem to the surgeon. One can measure the modulation of HRs by
TES on levels that are subthreshold for TES-mMEP
responses. This theoretically opens a door to continuous
monitoring of motor tracts by TES that does not interfere with the surgical procedure.
One can also interchange the conditioning and test
stimuli by using peripheral nerve stimulation for conditioning and TES as test stimulation. The stimulation in the receptive field of the foot sole and
palmar region of the hand with pulse trains have been
utilized in clinical monitoring (Andersson and Ohlin,
1999; Erb et al., 2005). Similar results can also be
obtained from direct peripheral nerve stimulation.
Each method will have its own characteristics in
regard to facilitation and somatotopic range.
A simple method which requires just one stimulator is double train stimulation (DT-TES). Fig. 5A
can predict facilitation effects. The first peak cannot
be utilized since the first and second train cannot be
given at the same time. Maximum facilitation is
expected when the second TES train is given at the
second peak at 20 ms when the aMN membrane is

hyper-excitable and also at 90130 ms (third peak).
Fig. 5D shows the conditioning double train stimulation (DTS) curve using DT-TES. The DTS curve is
bimodal but can be tri-modal as well. Similar to the
CTI, the time interval between onset of the TES pulse
trains is defined as the intertrain interval (ITI). A second peak is at ITI 25 ms. This differs from the predicted location at 20 ms. The 5-ms difference may be
due to a partial overlap of active motor neurons and a
blurring effect from a pulse train length of 8 ms. ITI
20 ms is in the short intertrain interval (S-ITI) region
between 10 and 35 ms where in usually facilitation is
seen (Journee et al., 2004b). A second region with
facilitation at ITI 80160 ms corresponds with
the long intertrain interval (L-ITI) region for facilitation. Fig. 6 illustrates how facilitation effects can
change at different TES intensities. For DTS at ITI
20 ms (S-ITI region), the situation is complex.
Fig. 6B shows initially a high facilitation at relative
low stimulation intensities. The amplitudes from single train stimulation (STS) are then small (Fig. 6A)
which means that only a small number of the aMN
have fired. The amplitude nMEP from STS, and also
the fraction of firing aMNs, increases. The interval
time allows, for example, RCs to inhibit an increased
population of aMNs leaving a decreased population
that are available for firing during the second TES
train. Even when all remaining aMNs fire at second
TES, the amplitude of the response is forced to
decrease. Fig. 6C (L-ITI region) shows a high facilitation for all stimulus intensities. Obviously, the
motor neurons have now recovered from firing from
the first train.
TES intensity
100
125
150
175
5 mV
200
250
300
400 V
20 ms
Fig. 6. TES-mMEP response series of the anterior tibial muscle of another scoliosis patient at increasing TES
intensities (Cz0 Fz; PW 200 ms; IPI 2 ms; 4 p/train). A: Responses are from (A) single train stimulation, double train
stimulation at B: ITI 20 ms, and C: ITI 125 ms.
DTS is often successful in neurologically impaired

patients that otherwise cannot be monitored (Langeloo
et al., 2001) and also most useful when TES-mMEP
responses are small (Journee et al., 2004).
15.5.5. TES-mMEP in children
In children with immature corticospinal motor pathways, TES-mMEP responses are sometimes difficult
to obtain. Due to the relatively small diameters of axons
and the immature myelin, one can expect an increased
chronaxia. Increased TES threshold voltages are
described by Lieberman et al. (2006). Besides
increased TES threshold voltages (range 75130 V;
PW 100 ms), we also found increased chronaxia in
5 of 49 patients (TIVA; halogenated agents excluded)
who were monitored during surgery in the conus-cauda
region for dethetering or tumor removal. All five
patients were younger than 3.5 years (range 2 months
to 3.5 years) (Journee et al., 2006). Four values were
outside the 95% confidence interval for reference TES
threshold values obtained from the subgroup aged above
10 years (mean threshold: 53.6 V; S.D. 16.8 V;
n 36). In a group of 04 years, in three of seven
cases, tibial mMEP responses were absent, and weak
(mean 60 mV) in the other four at STS with 46 p/train.
Conditioning by DTS showed a marked facilitation
resulting in a mean amplitude of 1,550 mV. Long pulse
trains up to 10 p/train showed a less marked increase to
a mean mMEP amplitude of 220 mV. This increase
agrees with the magnitude of facilitation of Erb et al.
(2005) who applied palmar conditioning stimulation
in young children. Like in adults, an IPI of 2 ms appear
to yield optimal responses. It is concluded that immature motor pathways in children below 4 years of age
result in TES-mMEP responses with small or absent
amplitudes that can be enhanced markedly by DTS
for successful monitoring. Recommended are pulse
trains of 6 p/train and ITIs of ISI 20 or 125 ms.
One also may expect more efficient stimulation when
pulse widths are prolonged to 500 ms to adapt to the
proposed increased chronaxie.
15.6. Risks
Safety concerns in TES can be categorized as
(1) direct tissue damage, (2) kindling and seizures,
and (3) complications due to mechanical impact from
muscle contraction and from electrode placement.
Tissue damage due to direct physical impact only can
occur on places with the highest dissipation of electrical
229
energy. This can only be expected in small tissue

volumes near electrodes. In worst case, one might get
small (<2 mm2) superficial skin burns from surface
electrodes that are not properly attached to the skin or
dried out. Needle electrodes from TES can be considered as safe. When skin burns have an electrical origin,
monopolar cautery is a likely cause (Knickenbocker and
Neufeld, 1996) and precautions are similar to those that
apply to any intraoperatively used electrodes for physiological measurements in general. In TES, the currents
are sufficiently dispersed to attenuate the stimulus
charge density at the brain surface to a safe level (less
than 40 mCi/cm2.phase) to be sure not to cause damage
to neural tissue (Agnew and McCreery, 1987;
McCreery et al., 1990). According to our volume conduction models, skull defects will not introduce unsafe
conditions as long as the electrodes are separated from
the brain surface.
Kindling has indiscriminately been used as potential danger in transcranial stimulation. Kindling,
which refers to the induction of self-perpetuating epileptic foci that has been induced by daily repeated
electrical stimulation of 50 Hz for several seconds
in a experimental primate model, has not been
reported as a complication of TES. In contrast to
bipolar cortical stimulation, TES with a short train
of anodal high-frequency pulses has an extremely
low risk of inducing seizures. In a review of more
that 15,000 TES-MEP monitoring cases, McDonald
reports only five unpublished seizures all of which
were not related to TES (MacDonald, 2002). Furthermore, mentioned complications were 29 tongue or lip
lacerations, 2 scalp burns, 5 cardiac arrhythmias, and
1 mandibular fracture. Bite injuries due to jaw muscle contractions during TES are the most common
but still infrequent complication, having an estimated
incidence of about 0.2% (MacDonald, 2002). The
mechanism may involve both corticobulbar activation with pulse trains and direct muscle or trigeminal
nerve stimulation, because jaw-clenching also occurs
with single pulses. Thus, C3/4 TES might produce
stronger biting than C1/2 TES because the electrodes
are closer to facial motor cortex, jaw muscles, and
trigeminal nerves (MacDonald, 2006). Although
possible, no spinal epidural recording electrode complications or injuries resulting from TES-induced
movement were reported, as well as recognized
adverse neuropsychological effects. Nerve root or
spinal cord trauma, infection, and especially hematoma are concerns with any spinal epidural invasion.
For example, Rodi et al. reported a patient who
230
required emergency laminectomy to relieve cord compression from intraoperative hematoma caused by an
epidural anesthetic catheter (Rodi et al., 2003). Almost
all of these complications can be prevented by
adequate measures like using bite blocks and restriction of TES-induced movement. One should take
care of possible increased risk factors like epilepsy,
raised intracranial pressure, cardiac disease, proconvulsant medication, and cardiac pacemakers. When
in expert hands, the benefits of TES-MEP monitoring
convincingly outweighs associated risks.
15.7. Conclusions
The methodology for monitoring of the functional
integrity of motor pathways over the last decade has
progressed into a reliable, fast, and relatively simple
tool that can be easily used intraoperatively. The
progressing insights in the actual stimulated neural
structures in relation to electrical fields from transcranial and cortical stimulation offer perspectives for
mapping techniques for neurosurgery and spine
surgery, and solidify choices for placement of electrodes and transcranial stimulation paradigms for monitoring of epidural and mMEP responses. Intravenous
anesthetics offer improved compatibility with motor
monitoring, whereas developments in infusion techniques like TCI and anesthesia depth assessment may
help to further reduce variance in motor responses.
Facilitation techniques appear to offer a wide variety
of possibilities that not only will improve responses
with a poor amplitude. They also offer a selectivity
by which the monitorist might become equipped with
possibilities that closely approximate the clinical
monitoring problem and set a perspective on continuous motor monitoring with minimal movement utilizing facilitation at subthreshold level. This may further
reduce the already minimal risks to the patient and
improve medical quality and comfort for the surgeon.
Further neurophysiological research and development
are necessary to explore possibilities and assess their
value in clinical practice.
Acknowledgments
The author is indebted for their support in joint
projects to Dr. M. Sun, T. Rath, and D.L. Li (UPMC,
Pittsburgh, PA; TES models), Dr. V. Deletis (New
York, NY), H.E. Polak, and Dr. M. De Kleuver
(ISSAR Nijmegen, The Netherlands) (impedance
H.L. JOURNEE
and conditioning studies) and Dr. U. Beese (UMCG,

Groningen, The Netherlands) for their helpful
comments.
References
evoked potentials. Neurosurgery, 20(1): 143147.
Amassian, VE (2002) Animal and human motor system neurophysiology. In: V Deletis and JL Shills (Eds.). Neurophysiology in Neurosurgery. Academic Press, New York,
pp. 323.
Andersson, G and Ohlin, A (1999) Spatial facilitation of motor
evoked responses in monitoring during spinal surgery.
Ashram, YA, Jackler, RK, Pitts, LH and Yingling, CD
(2005) Intraoperative electrophysiologic identification
of the nervus intermedius. Otol. Neurotol., 26(2):
274279.
Baldissera, F and Cavallari, P (1993) Short-latency subliminal effects of transcranial magnetic stimulation on forearm motoneurones. Exp. Brain Res., 96(3): 513518.
Barker, AT, Jalinous, R and Freeston, IL (1985)
Non-invasive magnetic stimulation of the human motor
cortex [letter]. Lancet, 1: 11061107.
Boyd, SG, Rothwell, JC, Cowan, JM, Webb, PJ, Morley, T,
Asselman, P and Marsden, CD (1986) A method
of monitoring function in corticospinal pathways
during scoliosis surgery with a note on motor conduction velocities. J. Neurol. Neurosurg. Psychiatr., 49(3):
251257.
Calancie, B, Harris, W, Broton, JG, Alexeeva, N and
Green, BA (1998) Threshold-level multipulse transcranial electrical stimulation of motor cortex for intraoperative monitoring of spinal motor tracts: description
of method and comparison to somatosensory evoked
potential monitoring. J. Neurosurg., 88(3): 457470.
Calancie, B, Harris, W, Brindle, GF, Green, BA and Landy,
HJ (2001) Threshold-level repetitive transcranial electrical stimulation for intraoperative monitoring of central motor conduction. J. Neurosurg., 95(Suppl. 2):
161168.
Cedzich, C, Taniguchi, M, Schafer, S and Schramm, J
(1996) Somatosensory evoked potential phase reversal
and direct motor cortex stimulation during surgery in
and around the central region. Neurosurgery, 38(5):
962970.
Chan, JH, Lin, CS, Pierrot-Deseilligny, E and Burke, D (2002)
Excitability changes in human peripheral nerve axons in a
paradigm mimicking paired-pulse transcranial magnetic
stimulation. J. Physiol. Lond., 542(Pt. 3): 951961.
spinal surgery. J. Clin. Monit. Comput., 18(4): 303308.

Coburn, B (1989) Neural modelling in electrical stimulation. In: JR Bourne (Ed.), Critical Reviews in Biomedical Engineering CRC Press, Boca Raton, FL, 2nd ed.,
Vol. 17, pp. 133180.
Cowan, JM, Day, BL, Marsden, C and Rothwell, JC (1986)
The effect of percutaneous motor cortex stimulation on
H reflexes in muscles of the arm and leg in intact man.
J. Physiol. Lond., 377: 333347.
De Haan, P, Kalkman, CJ, De Mol, BA, Ubags, LH, Veldman, DJ and Jacobs, MJ (1997) Efficacy of transcranial
motor-evoked myogenic potentials to detect spinal cord
ischemia during operations for thoracoabdominal aneurysms. J. Thorac. Cardiovasc. Surg., 113(1): 87100.
Deletis, V (2001) The motor inaccuracy in neurogenic
motor evoked potentials. Clin. Neurophysiol., 112(8):
13651366.
Deletis, V (2002) Intraoperative neurophysiology and
methodologies used to monitor the functional integrity
of the motor system. In: V Deletis and JL Shills
(Eds.), Neurophysiology in Neurosurgery. Academic
Press, New York, pp. 2551.
Deletis, V, Rodi, Z and Amassian, VE (2001) Neurophysiological mechanisms underlying motorevoked potentials
in anesthetized humans. Part 2. Relationship between epidurally and muscle recorded MEPs in man. Clin. Neurophysiol., 112(3): 445452.
Digitimer Isolated Multipulse Stimulator Model D185
Operators Manual. Issue 4. 7201999. Digitimer Ltd.,
Hertfordshire, UK.
Dong, CCJ, MacDonald, DB and Janusz, MT (2002)
Intraoperative spinal cord monitoring during
descending thoracic and thoracoabdominal aneurysm
surgery. Ann. Thorac. Surg., 74(5): S1873S1876.
Dong, C, MacDonald, DB, Akagami, R, Westerberg, B,
Alkhani, A, Kanaan, I and Hassounah, M (2005) Intraoperative facial motor evoked potential monitoing with
transcranial electrical stimulation during skull base surgery. Clin. Neurophysiol., 116(3): 588596.
Drover, DR, Litalien, CL, Wellis, V, Shafer, SL and Hammer,
GB (2004) Determination of the pharmacodynamic interaction of propofol and remifentanyl during esophagogastroduodenoscopy in children. Anesthesiology, 100:
13821386.
Edgley, SA, Eyre, JA, Lemon, RN and Miller, S (1997)
Comparison of activation of corticospinal neurons and
spinal motor neurons by magnetic and electrical transcranial stimulation in the lumbosacral cord of the
anaesthetized monkey. Brain, 120(Pt. 5): 839853.
Erb, TO, Ryhult, SE, Duitmann, E, Hasler, C, Leutsch, J
and Frei, FJ (2005) Improvement of motor-evoked
potentials by ketamine and spatial facilitation during
spinal surgery in a young child. Anesth. Analg., 100:
16341636.
Hargreaves, SJ and Watt, JW (2005) Intravenous anaesthesia and repetitive transcranial magnetic stimulation
231
monitoring in spinal column surgery. Br. J. Anaesth.,
94(1): 7073.
Hicks, R, Burke, D, Stephen, J, Woodforth, I and Crawford, M (1992) Corticospinal volleys evoked by electrical stimulation of human motor cortex after withdrawal
of volatile anaesthetics. J. Physiol. Lond., 456:
393404.
Holland, NR, Lukaczyk, TA, Riley, LH, III and Kostuik, JP
(1998) Higher electrical stimulus intensities are required
to activate chronically compressed nerve roots. Implications for intraoperative electromyographic pedicle
screw testing. Spine, 23(2): 224227.
Houlden, DA, Schwartz, ML, Tator, CH, Ashby, P and
MacKay, WA (1999) Spinal cord-evoked potentials
and muscle responses evoked by transcranial magnetic
stimulation in 10 awake human subjects. J. Neurosci.,
19(5): 18551862.
Inoue, S, Kawaguchi, M, Kakimoto, M, Sakamoto, T, Kitaguchi, K, Furuya, H, Morimoto, T and Sakaki, T (2002)
Amplitudes and intrapatient variability of myogenic
motor evoked potentials to transcranial electrical stimulation during ketamine/N2O- and propofol/N2O-based
anesthesia. J. Neurosurg. Anesthesiol., 14(3): 213217.
Jernigan, TL, Archibald, SL, Fennema-Notestine, C,
Gamst, AC, Stout, JC, Bonner, J and Hesselink, JR
(2001) Effects of age on tissues and regions of the cerebrum and cerebellum. Neurobiol. Aging, 22(4):
581594.
Jones, SJ, Buonamassa, S and Crockard, HA (2003) Two
cases of quadriparesis following anterior cervical discectomy, with normal perioperative somatosensory
evoked potentials. J. Neurol. Neurosurg. Psychiatr., 74
(2): 273276.
Journee, HL, Shils, J, Bueno de Camargo, A, Novak, K and
Deletis, V (2003) Failure of Digitimers D-185 transcranial stimulator to deliver declared stimulus parameters.
Journee, HL, Polak, HE and De Kleuver, M (2004a) Influence of electrode impedance on threshold voltage for
transcranial electrical stimulation in motor evoked
potential monitoring. Med. Biol. Eng. Comput., 42(4):
557561.
Journee, HL, Polak, HE, De Kleuver, M, Langeloo, DD and
Postma, AA (2004b) Improved neuromonitoring during
spinal surgery using double-train transcranial electrical
stimulation. Med. Biol. Eng. Comput., 42(1): 110113.
Journee, HL, Hoving, EW and Mooij, JJA (2006) Stimulation
threshold-age relationship and improvement of muscle
potentials by preconditioning transcranial stimulation in
young children. Clin. Neurophysiol., 117(Suppl. 1): 235.
Kalkman, CJ, Drummond, JC, Kennelly, NA, Patel, NK and
Partridge, BL (1992) Intraoperative monitoring of tibialis
anterior muscle evoked responses to transcranial electrical
stimulation during partial neuromuscular blockade.
Anesth. Analg., 75: 584589.
232
Kalkman, CJ, Ubags, LH, Been, HD, Swaan, A and Drummond, JC (1995) Improved amplitude of myogenic motor
evoked responses after paired transcranial electrical stimulation during sufentanil/nitrous oxide anesthesia. Anesthesiology, 83(2): 270276.
Keats, TE and Sistrom, C (2001) Atlas of Radiologic Measurement. Mosby Inc, St. Louis, MO, 7th ed.
Kernell, D and Wu, CP (1967) Responses of the pyramidal tract to stimulation of the baboons motor cortex.
J. Physiol. Lond., 191: 653672.
Knickenbocker, GG and Neufeld, GR (1996) Electrotrauma in
the operating room: shock, electrocution and burns. In: N
Gravenstein and NN Kirby (Eds.), Complications in Anesthesiology. Lippincott-Raven Publishers, 2nd ed., Philadelphia, pp. 7991.
Kombos, T, Suess, O, Funk, T, Kern, BC and Brock, M
(2000a) Intra-operative mapping of the motor cortex
during surgery in and around the motor cortex. Acta
Neurochir., 142(3): 263268.
Kombos, T, Suess, O, Pietila, T and Brock, M (2000b) Subdural air limits the elicitation of compound muscle
action potentials by high-frequency transcranial electrical stimulation. Br. J. Neurosurg., 14(3): 240243.
Kothbauer, K, Deletis, V and Epstein, FJ (1997) Intraoperative spinal cord monitoring for intramedullary surgery: an essential adjunct [see comments]. Pediatr.
Neurosurg., 26(5): 247254.
Lang, EW, Beutler, AS, Chesnut, RM, Patel, PM, Kennelly, NA, Kalkman, CJ, Drummond, JC and Garfin,
SR (1996a) Myogenic motor-evoked potential monitoring using partial neuromuscular blockade in surgery of
the spine. Spine, 21(14): 16761686.
Lang, EW, Chesnut, RM, Beutler, AS, Kennelly, NA and
Renaudin, JW (1996b) The utility of motor-evoked
potential monitoring during intramedullary surgery.
Anesth. Analg., 83(6): 13371341.
Langeloo, DD, Journee, HL, Polak, B and De Kleuver, M
(2001) A new application of TCE-MEP: spinal cord
monitoring in patients with severe neuromuscular
weakness undergoing corrective spine surgery. J. Spinal
Disord., 14(5): 445448.
Lenke, LG, Padberg, AM, Russo, MH, Bridwell, KH and
Gelb, DE (1995) Triggered electromyographic threshold
for accuracy of pedicle screw placement. An animal model
and clinical correlation. Spine, 20(14): 15851591.
Leppanen, RE (2006) Intraoperative applications of the
H-reflex and F-response: a tutorial. J. Clin. Monit. Comput., 20(4): 267304.
Leppanen, R, Maguire, J, Wallace, S, Madigan, R and Draper,
V (1995) Intraoperative lower extremity reflex muscle
activity as an adjunct to conventional somatosensoryevoked potentials and descending neurogenic monitoring
in idiopathic scoliosis. Spine, 20: 18721877.
Lieberman, JA, Lyon, R, Feiner, J, Diab, M and Gregory, GA
(2006) The effect of age on motor evoked potentials in
H.L. JOURNEE
children under propofol/ isoflurane anesthesia. Anesth.
Analg., 103: 316321.
Lifshitz, K (1967) An investigation of electrode guarding and
frequency effects in rheoencephalography. In: H Lechner,
N Geyer, E Lugaresi, F Martin, K Lifshitz and S Markovich (Eds.), Rheoencephalography and Plethysmographical Methods. Excerpta Medica Foundation, Amsterdam,
pp. 1016.
Lips, J, De Haan, P, Bouma, GJ, Jacobs, MJ and Kalkman,
CJ (2002a) Delayed detection of motor pathway dysfunction after selective reduction of thoracic spinal cord blood
flow in pigs. J. Thorac. Cardiovasc. Surg., 123(3):
531538.
Lips, J, De Haan, P, De Jager, SW, Vanicky, I, Jacobs, MJ
and Kalkman, CJ (2002b) The role of transcranial motor
evoked potentials in predicting neurologic and histopathologic outcome after experimental spinal cord ischemia. Anesthesiology, 97(1): 183191.
Lo, YL, Dan, YF, Tan, YE, Nurjannah, S, Tan, SB, Tan,
CT and Raman, S (2004) Intra-operative monitoring
in scoliosis surgery with multi-pulse cortical stimuli
and desflurane anesthesia. Spinal Cord, 42(6):
342345.
Maccabee, PJ, Amassian, VE, Eberle, LP and Cracco, RQ
(1993) Magnetic coil stimulation of straight and bent
amphibian and mammalian peripheral nerve in vitro:
locus of excitation. J. Physiol. Lond., 460: 201219.
MacDonald, DB (2002) Safety of intraoperative transcranial
electrical stimulation motor evoked potential monitoring.
J. Clin. Neurophysiol., 19(5): 416429.
MacDonald, DB (2006) Intraoperative motor evoked potential monitoring: overview and update. J. Clin. Monit.
Comput., 20(5): 347377.
McCreery, DB, Agnew, WF, Yuen, TG and Bullara, L (1990)
Charge density and charge per phase as cofactors in neural injury induced by electrical stimulation. IEEE Trans.
Biomed. Eng., 37(10): 9961001.
McNeal, DR (1976) Analysis of a model for excitation of
myelinated nerve. IEEE Trans. Biomed. Eng., 23:
329337.
Merton, PA and Morton, HB (1980) Stimulation of the
cerebral cortex in the intact human subject. Nature,
285(5762): 227.
Minahan, RE, Sepkuty, JP, Lesser, RP, Sponseller, PD
and Kostuik, JP (2001) Anterior spinal cord injury
with preserved neurogenic motor evoked potentials.
Mller, AR (2002) Monitoring and mapping the cranial nerves
and the brainstem. In: V Deletis and JL Shills (Eds.), Neurophysiology in Neurosurgery. Academic Press, New
York, pp. 291318.
Mller, AR (2002) Monitoring and mapping the cranial
nerves and the brainstem. In: V Deletis and JL Shills

Neuloh, G, Pechstein, U, Cedzich, C and Schramm, J (2004)
Motor evoked potential monitoring with supratentorial
surgery. Neurosurgery, 54(5): 10611070.
Novak, K, De Camargo, AB, Neuwirth, M, Kothbauer, K,
Amassian, VE and Deletis, V (2004) The refractory
period of fast conducting corticospinal tract axons in
man and its implications for intraoperative monitoring
of motor evoked potentials. Clin. Neurophysiol., 115(8):
19311941.
Nowak, LG and Bullier, J (1998a) Axons, but not cell bodies, are activated by electrical stimulation in cortical
gray matter. I. Evidence from chronaxie measurements.
Exp. Brain Res., 118(4): 477488.
Nowak, LG and Bullier, J (1998b) Axons, but not cell bodies, are activated by electrical stimulation in cortical
gray matter. II. Evidence from selective inactivation of
cell bodies and axon initial segments. Exp. Brain Res.,
118(4): 489500.
Patton, HD and Amassian, VE (1954) Single- and multipleunit analysis of cortical stage of pyramidal tract activation. J. Neurophysiol., 17: 345363.
Pechstein, U, Cedzich, C, Nadstawek, J and Schramm, J
(1996) Transcranial high-frequency repetitive electrical
stimulation for recording myogenic motor evoked
potentials with the patient under general anesthesia.
Pelosi, L, Stevenson, M, Hobbs, GJ, Jardine, A and Webb,
JK (2001) Intraoperative motor evoked potentials to
transcranial electrical stimulation during two anaesthetic
regimens. Clin. Neurophysiol., 112(6): 10761087.
Petersen, N, Christensen, LO and Nielsen, J (1998) The
effect of transcranial magnetic stimulation on the soleus
H reflex during human walking. J. Physiol. Lond., 513
(Pt. 2): 599610.
Pierrot-Deseilligny, E and Burke, D (2005) The Circuitry of
the Human Spinal Cord. Its Role in Motor Control and
Movement Disorders. Cambridge University Press,
Cambridge.
Quinones-Hinojosa, A, Gadkary, CA, Gulati, M, Von
Koch, CS, Lyon, R, Weinstein, PR and Yingling, CD
(2004) Neurophysiological monitoring for safe surgical
tethered cord syndrome release in adults. Surg. Neurol.,
62(2): 127133.
Rattay, F (1987) Ways to approximate current-distance
relations for electrically stimulated fibers. J. Theor.
Biol., 125(3): 339349.
Rodi, Z, Straus, I, Denic, K, Deletis, V and Vodusek, DB
(2003) Transient paraplegia revealed by intraoperative
neurophysiological monitoring: was it caused by the
epidural anesthetic or an epidural hematoma? Anesth.
Analg., 96(6): 17851788.
Rohde, V, Krombach, GA, Baumert, JH, KreitschmannAndermahr, I, Weinzierl, M and Gilsbach, JM (2003)
Measurement of motor evoked potentials following
repetitive magnetic motor cortex stimulation during
233
isoflurane or propofol anaesthesia. Br. J. Anaesth.,
91(4): 487492.
Rothwell, JC, Thompson, PD, Day, BJ, Boyd, SG and
Marsden, CD (1991) Stimulation of the human motor
cortex through the scalp. Exp. Physiol., 76: 159200.
Rothwell, J, Burke, D, Hicks, R, Stephen, J, Woodforth, I
and Crawford, M (1994) Transcranial electrical stimulation of the motor cortex in man: further evidence
for the site of activation. J. Physiol. Lond., 481(Pt. 1):
243250.
Sala, F, Niimi, Y, Berenstein, A and Deletis, V (2001)
Neuroprotective role of neurophysiological monitoring
during endovascular procedures in the spinal cord.
Ann. N.Y. Acad. Sci., 939: 126136.
Scheufler, KM and Zentner, J (2002) Total intravenous
anesthesia for intraoperative monitoring of the motor pathways: an integral view combining clinical and experimental data. J. Neurosurg., 96: 571579.
Scheufler, KM, Reinacher, PC, Blumrich, W, Zentner, J
and Priebe, HJ (2005) The modifying effects of stimulation pattern and propofol plasma concentration on
motor-evoked potentials. Anesth. Analg., 100(2):
440447.
Seipel, JH (1970) The influence of electrode size and material on the rheoencephalogram. Ann. N.Y. Acad. Sci.,
170: 604621.
Shils, J., Martin, C. and Deletis, V. (2005) Intraoperative
monitoring for surgeries at the jugular foramen. Oper.
Tech. Neurosurg., 8(1): 4249.
Sloan, TB and Heyer, EJ (2002) Anesthesia for intraoperative neurophysiologic monitoring of the spinal cord.
Stecker, MM (2005) Transcranial electric stimulation of
motor pathways: a theoretical analysis. Comput. Biol.
Med., 35(2): 133155.
Stinson, LW, Murray, MJ, Jones, KA, Assef, SJ, Burke,
MJ, Behrens, TL and Lennon, RL (1994) A computer
controlled, closed-loop infusion system for infusing
muscle relaxants: its use during motor-evoked potential
monitoring. J. Cardiothorac. Vasc. Anesth., 8: 4044.
Suess, O, Kombos, T, Ciklatekerlio, O, Stendel, R, Suess, S
and Brock, M (2002) Impact of brain shift on intraoperative neurophysiological monitoring with cortical
strip electrodes. Acta Neurochir., 12(144): 12791289.
Suihko, V (1998) Modeling direct activation of corticospinal axons using transcranial electrical stimulation.
Electroencephalogr. Clin. Neurophysiol., 109(3):
238244.
Taniguchi, M and Schramm, J (1991) Motor evoked potentials
facilitated by an additional peripheral nerve stimulation.
Electroencephalogr. Clin. Neurophysiol. Suppl., 43:
202211.
Terao, Y, Ugawa, Y, Hanajima, R, Machii, K,
Furubayashi, T, Mochizuki, H, Enomoto, H, Shiio, Y,
Uesugi, I, Iwata, NK and Kanazawa, I (2000)
234
Predominant activation of I1-waves from the leg motor
area by transcranial magnetic stimulation. Brain Res.,
859(1): 137146.
Thielscher, A and Kammer, T (2002) Linking physics with
physiology in TMS: a sphere field model to determine
the cortical stimulation site in TMS. Neuroimage, 17(3):
11171130.
Thomson, AM, Deuchars, J and West, DC (1993) Single
axon excitatory postsynaptic potentials in neocortical
interneurons exhibit pronounced paired pulse facilitation.
Neuroscience, 54(2): 347360.
Tidswell, T, Gibson, A, Bayford, RH and Holder, DS (2001)
Three-dimensional electrical impedance tomography of
human brain activity. Neuroimage, 13(2): 283294.
Toleikis, JR (2002) Neurophysiological monitoring during
pedicle screw placement. In: V Deletis and JL Shills
H.L. JOURNEE
Ubags, LH, Kalkman, CJ and Been, HD (1998) Influence
of isoflurane on myogenic motor evoked potentials to
single and multiple transcranial stimuli during nitrous
oxide/opioid anesthesia. Neurosurgery, 43(1): 9094.
Ubags, LH, Kalkman, CJ, Been, HD, Koelman, JH and De
Ongerboer de Visser, B (1999) A comparison of myogenic motor evoked responses to electrical and magnetic transcranial stimulation during nitrous oxide/
opioid anesthesia. Anesth. Analg., 88(3): 568572.
Ulkatan, S, Neuwirth, M, Bitan, F, Minardi, C, Kokoszka,
A and Deletis, V (2006) Monitoring of scoliosis surgery
with epidurally recorded motor evoked potentials
(D wave) revealed false results. Clin. Neurophysiol.,
117(9): 20932101.
Vodusek, DB and Deletis, V (2002) Intraoperative neurophysiological monitoring of the sacral nervous system.
In: V Deletis and JL Shills (Eds.), Neurophysiology in
Neurosurgery. Academic Press, New York, pp. 197217.

M.R. Nuwer (Ed.)
235
CHAPTER 16
Corticospinal tract monitoring with D- and I-waves from

the spinal cord and muscle MEPs from limb muscles
Vedran Deletisa,* and Francesco Salab
a
Center for Endovascular Surgery and Intraoperative Neurophysiology, Institute for Neurology and Neurosurgery,
Roosevelt Hospital, New York, NY 10019, USA
Department of Neurological and Visual Sciences, Section of Neurosurgery, University Hospital, 37100 Verona, Italy
16.1. Introduction
16.1.1. Intraoperative neurophysiology of the
corticospinal tract
Intraoperative monitoring (IOM) of the functional
integrity of the corticospinal tract (CT) has specific
requirements due to the pharmacologic influence of
applied anesthetics and the limitations inherent in surgery on an unconscious patient. These factors present
advantages and disadvantages for both IOM and the
continued physiological exploration of the CT. With
this in mind, several IOM methodologies have been
developed, which give us an unique opportunity to
explore the physiology of the CT, especially within
the spinal cord. The application of these methodologies expands on what is currently possible during
IOM of the functional integrity of the motor system.
The experience in intraoperative neurophysiology
of the CT comes from different sources.
When we are facing an existing neurological deficit
in a patient coming to surgery, IOM teaches us the

essential elements of maintaining the functional
status of the motor system
When an acute injury to the motor system occurs
during surgery, we can identify the pathophysiology of such event through IOM and learn how to
prevent this from happening in the future. A special
value has transient, that is, reversible, injury
The knowledge we gained from experimental work
done in animals, especially that done in primates
*
Correspondence to: Vedran Deletis, M.D., Ph.D., Institute

for Neurology and Neurosurgery, St. Lukes Roosevelt
Hospital, 1000 Tenth Avenue, Suite 11G-78, New York,
NY 10019, USA.
Tel.: 1-212-636-3281; fax: 1-212-636-3159.
E-mail: vdeletis@chpnet.org (V. Deletis).
Neurophysiological data collected intraoperatively
from neurologically intact patients.

16.2. Methodology of electrical stimulation of the
brain with recording activity of the CT
16.2.1. Methodologies for intraoperative
electrical stimulation of the CT
The strong pulse(s) of electrical current applied on the
scalp can overcome the impedance barrier of bone and
soft tissue and stimulate the brain. In Fig. 1, we present
schematics of the basic methodology to intraoperatively
monitor motor evoked potentials (MEPs) from the spinal
cords epi- or subdural space, or limb muscles by using
single and multipulse stimulating techniques applied
transcranially or over the exposed motor cortex respectively. The single pulse stimulating technique is used
to elicit D- and I-waves, while the multipulse stimulating
technique is used to elicite muscle MEPs. This schema
will help readers to better understand the methodology
of intraoperatively eliciting and recording of MEPs.
In Fig. 2, details of the positioning of stimulating
electrodes over the scalp use to elicite MEPs are
shown. The electrode placement on the skull is based
on the International 10/20 EEG system. Note that
instead of Cz, the Cz electrode is placed 1 cm
behind the typical Cz point because it corresponds
better to the primary motor cortex of lower extremities. We prefer to place a full set of stimulating electrodes for transcranial electrical stimulation (TES)
because sometimes which montage will provide better muscle MEPs without unnecessary muscle twitching is unpredictable. For eliciting D-wave muscle
twitching which can be disturbing to the surgeon
is not an issue because a single stimulus applied
transcranially over the exposed brain does not elicit
236
Fig. 1. Intraoperative methodology for eliciting and

recording motor evoked potentials (MEPs) from the spinal
cord and limb muscles. Top: illustration of electrode positions for transcranial (A) and direct (B) electrical stimulation of the motor cortex according to the International
1020 EEG system. Middle: schematic diagram of the
positions of the catheter electrodes (each with three recording surfaces) places cranial to the tumor (control electrode)
and caudal to the tumor to monitor the incoming signal
passing through the site of surgery. In the middle are Dand I-wave records rostral and caudal to the pathology.
Please note the peak latency difference between cranial
and caudal recordings of the D- and I-waves which are
marked with vertical lines. Bottom: recording of muscle
MEPs from the thenar, tibialis anterior, and abductor hallucis muscles after eliciting them with a short train of electrical pulses applied transcranially. (Modified from Deletis
et al., 2001b with permission from the International Federation of Clinical Neurophysiology.)
significant muscle twitching, and in most patients

twitching might be completely absent.
For transcranial stimulation, cork screw-like electrodes (CS electrodes; Viasis, Neurocare, Madison,
WI) are preferable due to their secure placement
V. DELETIS AND F. SALA
Fig. 2. Top: actual position of electrode inserted in the

scalp for TES (front row: C3, C1, Cz, C2, and C4, and
6 cm (Cz 6 cm). Electrodes positioned posteriorly
are for recordings of the somatosensory evoked potentials.
The most used stimulation montage is C1/C2. Bottom:
insert, cork screw electrode, enlarged with a scale in mm.
and low impedance below 1 kO (Journee et al.,

2004). Alternatively, EEG needle electrodes may be
used. We do not recommend the use of EEG cup
electrodes fixed with collodium because they are
impractical and their placement is time consuming.
The only exception is for young children in whom
the fontanel still exists. Since the CS electrodes can
penetrate the fontanel during placement, the use of
EEG cup electrodes is preferable.
The brain depth where electrical current activates
CT depends on a few factors: (a) the type of stimulation, whether TES or direct cortical stimulation
(DCS), (b) the intensity of the applied current, and
(c) the position of stimulating electrodes. A very
strong current applied transcranially can penetrate
deep into the brain to the brainstem (Rothwell
et al., 1994). This is not the case after direct stimulation of the brain because less current is required
to activate CT (usually not more than 25 mA).
In Fig. 3, graphic presentations of different ways of
237
Fig. 3. Top left: current flow during TES and direct brain stimulation via grid electrode are presented schematically. During
strong TES, current penetrates deep in the brain, activating both CTs. During direct brain stimulation, using a grid electrode
current flow is restricted to a single corticospinal tract if one uses low current intensity and activates only restricted motoneuron pools from selective cortical areas (upper or lower extremities depending on the position of the stimulating electrode). Top right: difference in amplitude and latencies of the D-waves record epidurally over the upper thoracic spinal
cord in a patient undergoing surgery for a spinal cord tumor. Note the 1.9 ms difference between latencies of the
D-waves when elicited with low intensity of current and stimulating montage C1/C2 versus high intensity of current and
montage C3/C4. Note the higher amplitude of the D-wave when more axons of the CT are recruited and current penetrates
deep in the brain (C3/C4 montage and 240 mA stimulating current). Bottom: D- and I-waves recorded after single electrical
stimulus delivered transcranially (Cz anode/6 cm anterior cathode) in a 14-year-old patient with idiopathic scoliosis. As a
result of increasing the intensity of the stimulus, the electrical current activates the CT deeper within the brain and the
latency of the D-wave becomes shorter. As current becomes stronger, more I-waves are induced (100% corresponds to
750 V of stimulator output). Note that at the bottom, the three traces of D-waves have a double peak as a result of CT activation at different depths within the brain. (Modified from Deletis, 2002 with permission from Elsevier.)
stimulating CT at the different depths of the brain are

shown.
16.2.3. Recording of the D- and I-waves from

epi- or subdural space of the spinal cord
16.2.2. Methodologies for recording MEPs

There are essentially two intraoperative methods that
give us fast feedback (practically online) on the
functional integrity of the CTs fast neurons: (1) eliciting D- and I-waves using a single pulse stimulation,
either transcranially (TES) or over the exposed motor
cortex, while recording them from the epi(sub)dural
space of the spinal cord, and (2) eliciting muscle MEPs
after TES or direct multipulse electrical stimulation
of the motor cortex and recording them from limb
muscles.
This method is a direct clinical application of Patton

and Amassians discovery in the 1950s, in which the
electrically stimulated motor cortex in monkeys
generated a series of well-synchronized descending
volleys in the pyramidal tract (Patton and Amassian,
1954). The first conducted volley through fast neurons of CT (FNCT), following transsynaptic (indirect
or I-wave) volley, represents asynaptic activation of
the CT (direct or D-wave). This knowledge of CT
neurophysiology, collected in primates, can in most
cases be applied to humans. Boyd et al. (1986)
238
obtained the first recordings of the D-wave in

humans. This particular method of IOM of the CT
began to be employed during spine, spinal cord, and
brain surgery. We must be aware that even small
methodological details of recording D-waves are of
the utmost importance and should be followed in
order to achieve reliable results.
The recording catheter electrode we used was a
JX-330 (Arrow International, Reading, PA) because
of its optimal distance of 18 mm between each of
its three recording cylinders and optimal recording
surface of 12.3 mm2 (Fig. 4). For practical purposes,
any recording electrode designed for this purpose
would be appropriate and good recording can be
obtained. In our design, we optimized features of
recording electrodes, including a semirigid feature
for easy penetration in epi(sub)dural space. This electrode also has a double lumen with the ability to
inject saline in order to improve contact of electrodes
and surrounding tissue when the electrode is placed
percutaneously. Most epidural electrodes are disposable. If one uses a nondisposable type, extreme care
should be taken to insure the electrode is clean before
sterilization to improve its electrical properties. To
clean the electrode, we recommend one of the following procedures: immerse the electrode contacts in
saline and pass a 9 V DC current (regardless of polarity) through it until a bubble of gas cleans the contact
surface for a period of a few minutes, or use an ultrasound cleaner (Branson 1210, Branson Ultrasonics
Corporation, Danbury, CT) and submerse the electrode in the cleaner for 5 min. Both techniques will
remove any film or biological material remaining on
the contact surfaces, and decrease their impedance.
This maneuver will diminish the stimulus artifact,
which usually appears when contact surfaces have
high impedance. Because of the short latency of the
D-wave, a large stimulus artifact in an uncleaned electrode can pose an insurmountable obstacle for D-wave
recording. As it was presented in Fig. 1, during monitoring of the D-wave in patients with spinal cord
pathology, two catheter electrodes should be placed
at the caudal and rostral edge of laminectomy. The rostral electrode is the controlled electrode for nonsurgically induced changes in the D-wave, while the
caudal electrode monitors surgically induced changes
to the CT. The amplitude of the D-wave recorded over
the cervical spinal cord can be 60 mV or more, while
over the thoracic segments it may only be 10 mV. With
a stimulating rate of 2 Hz, it takes 24 averaged
responses to get a reliable D-wave. This results in an
update every second. Unfortunately, the maximal
stimulating rate in most of the commercially available
TES stimulators is 1 Hz.
The strong electric stimulus applied over C1/C2
points activates both CTs; therefore, the D-wave
represents activity from both of them. If C3 or C4
versus Cz-1 cm montage is used, the D-wave represents activity from predominantly one CT. D-wave
monitoring can be achieved in patients as young as
21 months old (Szelenyi et al., 2003).
16.2.3.1. Disadvantages of D-wave monitoring
Though there are enormous advantages in using
D-wave monitoring intraoperatively (see Section
16.5), there are also some disadvantages that must
be considered.
In about 20% of patients with intramedullary spinal
Fig. 4. Semirigid catheter electrode for recording MEPs

(D-wave) from the spinal cord, either epi- or subdurally;
the electrode passed through a 14-gauge Touhy needle
for percutaneous placement in the epidural space. To the
left (enlarged) are two openings marked with asterisks for
flushing the three cylindrical recording contacts (1, 2, and
3) through the injection site (top, right). The same electrode
can be placed in open surgical field, after laminectomy,
without the need for a Touhy needle. (Modified from
Deletis, 2002 with permission from Elsevier.)
cord tumors (IMSCT), or postradiation myelopathy, the D-wave was not present at the beginning
of surgery even if the muscle MEP was. These findings are consistent with previous reports suggesting
that radiation therapy seriously damages conductivity in the spinal cords long tracts (Scisciolo et al.,
1991). This phenomenon is probably due to the
desynchronization of the D-wave (descending
activity through the CT) (Deletis and Kothbauer,
1998). Therefore, this activity cannot be recorded
239
Fig. 5. A: Recording of a D-wave cranially (upper trace) and caudally (lower trace) to an intramedullary spinal cord tumor
(IMSCT). Note the well synchronized D-wave cranial in contrast to the desynchronized D-wave caudal to the tumor.
B: Very small epidurally recorded MEPs caudal to a high cervical intramedullary tumor (due to extreme desynchronization), despite large muscle MEPs recorded from a small hand muscle elicited after a short train of six stimuli were present
(to the right). (Modified from Deletis, 2002 with permission from Elsevier.)
caudally at the lesion site (tumor, myelopathy)

(Fig. 5). In this situation, when the D-wave is absent
at the beginning of surgery, the disappearance of
muscle MEPs during surgery cannot be used to
predict either permanent or transient lesions to the
motor system, primarily because in those patients,
specific information about CT is not obtainable.
Another disadvantage of the D-wave monitoring
method is that it cannot be applied caudally to the
Th 1011 spinal cord segments during surgery.
There is a lack of fast-conducting neurons of the
CTs used to generate D-waves of amplitudes sufficient for monitoring.
An article by Ulkatan et al. (2006) challenged the
use of D-wave methodology during scoliosis surgery because of a high percentage of false positive
and negative data (Fig. 6). Their explanation is
that the data of the D-wave amplitude increment
(in 4 patients out of 93) or decrement (21 out of 93)

comes from a new spatial relationship between the
epidural recording catheter and spinal cord after
surgical correction of the scoliosis, without parallel
changes of muscle MEPs. Therefore, the authors
expressed concern about using the D-wave as a
method to judge the functional integrity of the
CTs during monitoring of this specific pathology
(Ulkatan et al., 2006).
According to the report by MacDonald and Janusz
(2002), simultaneous monitoring during thoracoabdominal aneurysm surgery of D-wave, muscle MEPs,
and somatosensory evoked potentials (SEPs)
revealed that SEPs can better detect spinal cord segmental ischemia of the gray matter than D-wave can
in one patient. In other patients, however, muscle
MEPs were more sensitive to the ventral gray matter
ischemia than SEPs (MacDonald and Janusz, 2002).
240
Fig. 6. Upper: proposed mechanism of the D-wave amplitude changes (schematic drawings) in relation to the spinal cord
and epidural recording catheter A: before correction of spine deformity, B: after correction of spine deformity (the spinal
cord is now in closer proximity to the ERC than before it was corrected), and C: after correction of the spine deformity
(the spinal cord is now further from the ERC than before it was corrected). Bottom: the recordings from the patients epidural space (D-wave) and lower extremities muscles (muscle MEPs) during surgery for idiopathic scoliosis. Left bottom:
D-wave amplitude showed significant increment during correction of the scoliosis and it did not improve till the end of
the surgery. Muscle MEPs during and after correction of scoliosis showed no changes and patient did not have any motor
deficit postoperatively. (Note: muscle MEPs during correction has not been shown). Right bottom: the recordings from
another scoliotic patient. D-wave amplitude showed significant decrement during and at the end of the correction of the scoliosis. Muscle MEPs during and after correction of scoliosis showed no changes. CT corticospinal tract; ERC epidural
recording catheter; mMEP muscle MEP. (Modified from Ulkatan et al., 2006 with permission from the International
Federation of Clinical Neurophysiology.)
In the same report, MacDonald described false pos-
itive results of D-wave monitoring due to scalp

edema. It remains a question of interpretation
whether these results belong to the category of
real false positive data when the scalp edema can
be rectified by an increased intensity of stimulation
or the use of more lateral stimulating montage
(C3/C4) where scalp edema is less prominent.

Scalp edema can produce false positive results during muscle MEP monitoring as well.
Another relative disadvantage of D-wave monitoring
is the percutaneous placement of the epidural electrode when the epidural space is not surgically open,
which might be a burden for the anesthesiological
or surgical team. Some authors placed a long smalldiameter catheter subdurally a day before surgery.
According to their report, they did not have serious
complications such as bleeding, infection, etc.
(Tamaki et al., 1985, 1986).
Pronounced dural adhesion during reoperation, or
following with therapeutic irradiation of the spinal
cord, may prevent placement of epidural electrodes.
16.2.4. Recording muscle MEPs from limb muscles
Recordings of muscle MEPs could be done either
with a disposable EEG needle electrode or with surface electrodes. Both give decent amplitudes of muscle MEPs. The prefered muscles for monitoring the
MEPs of the upper extremities are: abductor digiti
minimi, first dorsal interosseus, abductor pollicis brevis as well as flexors and extensors of the forearm.
For the lower extremities, the tibialis anterior and
abductor hallucis brevis are appropriate muscles
because of the reach of CT innervations to the spinal
a-motoneurons (Jankowska et al., 1975).
16.3. Relationship between D- and I-waves,
recorded from the spinal cord and MEPs
recorded from the limb muscles
In order to understand the mechanism of muscle
MEP generation and its behavior during different
types of surgery, it is critical to understand the generation of D- and I-waves as elements in the chain of
events preceding the generation of muscle MEPs.
The electrical stimulation of the motor cortex or subcortical motor pathways activates the FNCT whose
activity can be easily recorded from an electrode
placed near the spinal cord as a synchronized activity
named D-wave (direct activation of CT) (Patton and
Amassian, 1954). If the CT fast conducting neurons
are activated indirectly through vertically oriented
neurons which end synaptically on CT neurons
this activation is called an I-wave (indirect activation of the CT). The synaptic connections involved in
I-wave generation, which are sensitive to anesthetics
and not very well synchronized, makes them unsuitable for intraoperative use.
The opposite is true for the D-wave. This is an
asynaptic activity of the CT (neurogram of the
CT). Stimulation of the CT occurs intracranially distal to the cortical motoneuron body. Recording is
done caudal to the surgical site, but above the synapses
241
of the CT with the a-motoneuron. Since no synapses

are involved between the stimulating site and the
recording site, the D-wave is very stable and reliable.
Therefore, we consider it the gold standard for measuring the functional integrity of the CT.
Generation of muscle MEPs during anesthesia
depends on the excitability of the a-motoneurons in
the anterior horns. Muscle MEPs can be generated
only if the resting potential of a-motoneurons reaches
the firing threshold, and thus, transmits this activity
via motor axons of the peripheral nerves and neuromuscular junctions to the muscle.
A very critical element in the chains of events
after stimulation of the CT is the constantly changing
excitability of a-motoneurons, which is influenced by
the depth of anesthesia. A critical depolarization
of the a-motoneurons can be achieved by Dand I-waves, by multiple D-waves (in the case of
multipulse stimulating technique, where multiple
D-waves are generated), or a combination of the
two. I-waves have an amplitude three times higher
than what is actually recorded via epidural electrodes. Their actual amplitude can be confirmed in
killed end recordings (Amassian and Deletis,
1999). Furthermore, even relatively light changes in
the anesthetic regimen might influence the generation of muscle MEPs by changing the amount of
I-waves in the descending drive impinging a-motoneurons. Multiple D-waves, generated by multipulse
stimulation techniques, represent a very efficient
solution for successfully eliciting muscle MEPs
because the firing threshold of a-motoneurons can
be reached either by a combination of the D- and
I-waves or by multiple D-waves.
16.3.1. Facilitation of the D-wave
Our previous work (Rodi et al., 1996) described
another neurophysiological phenomenon related to
the train stimulation technique. When three or more
electrical stimuli were applied to the motor cortex
during relatively light anesthesia, they not only elicited a matching number of D-waves, but also additional I-waves (Fig. 7A). We believe that these
additional I-waves are generated by facilitating the
effect of multiple stimuli on the corticomotoneuronal
synapses within the motor cortex. We termed this
phenomenon facilitation of I-waves. These epidurally recorded I-waves do not disappear after a muscle relaxant is administered, the facilitation of
I-waves is an important contributing factor in the
242
Fig. 7. The relationship between the number of transcranial electrical stimuli, ISI and the generation of D- and I-waves and
tibialis anterior (TA) muscle MEPs. A: Single and double stimuli applied transcranially elicit one and two D-waves only,
respectively. Three and four stimuli elicit three and four D-waves, respectively at a reduced amplitude with additional
I-waves and no muscle MEPs until a fifth stimulus is added. B: Increasing the ISI to 4 ms markedly increases the efficacy
of the stimuli in eliciting a TA response associated with the complete recovery of the D-wave amplitude at an ISI of 4 ms.
Train of five stimuli are needed with an ISI of 2 ms in order to elicit muscle MEPs in the tibial anterior muscle (A,5), while
with an ISI of 4 ms only three stimuli are needed to elicit muscle MEPs in the tibial anterior muscle (B,3). D D-wave;
I I-wave; PM paraspinal muscle artifact. (Reprinted from Deletis et al., 2001b with permission from the International
Federation of Clinical Neurophysiology.)
generation of additional descending volleys that end

on the spinal a-motoneuron. It is the number of
volleys arriving to the a-motoneuron, regardless of
identity as D- or I-waves that is essential for bringing
it above the firing threshold (Deletis et al., 2001b).
16.3.2. Recovery time of the D-wave
There is a frequency limit for the transmission of
descending volleys through the CT axons to the
a-motoneurons. This limitation can be easily tested
by applying two identical electrical stimuli transcranially with different interstimulus intervals (ISIs).
This test can show the recovery time of the second
D-wave response. Using this paradigm (conditioning
and test stimuli), a D-wave recovery curve can be
plotted relative to the amplitude and latency of the
second D-wave. We show that the optimal ISI for

the complete recovery of the second D-wave amplitude and latency is around 4 ms using stimulus duration of 500 ms of moderate intensity (Deletis et al.,
2001a). While with high stimulus intensity, complete
recovery happened even with ISI of 2 ms (Novak
et al., 2004). Because the a-motoneurons are optimally bombarded, when the train of equal stimuli
elicits D-waves of equal amplitudes, the optimal ISI
for muscle activation is expected to be 4 ms. Fig. 7
indicates that with an ISI of 4 ms, three stimuli are
sufficient to elicit muscle MEPs due to the complete
recovery of each consecutive D-wave. Comparatively,
using the identical stimulus intensity but decreasing
the ISI to 2 ms, five stimuli are needed to elicit muscle
MEPs which are of even smaller amplitude. This is due
to the incomplete recovery of the amplitude of each
consecutive D-wave. This rule applies only if a single

stimulus elicits a single D-wave.
16.3.3. Number of D- and I-waves
As previously stated, to allow for complete recovery
of the D-wave, the ISI in the multipulse train should
be 4 ms when using moderate intensity stimuli.
In situations where a single stimulus generates more
than a single D-wave, the optimal ISI should be set
long enough to allow the entire set of D- and
I-waves to recover. In turn, this will allow the next
set of D- and I-waves to fully develop. Therefore,
the second stimulus can generate the same pattern
of D- and I-waves (Fig. 8). Otherwise, the second
set of D- and I-waves could fall into the CT axon
refractory period resulting from the first set. This is
the case in Fig. 8A where a single stimulus generates
a single D- and multiple I-waves. In this case, only
two stimuli 8 ms apart are necessary to generate muscle MEPs (8D). If the ISI is shorter (e.g., 4.1 ms,
243
Fig. 8B), partial cancellation of D- and I-waves elicited by a second stimulus will occur. Consequently,
the total number of D- and I-waves will be insufficient to bring a-motoneurons to firing level and muscle MEPs will not be generated. This mechanism
could be important in a lightly anesthetized patient
as well as in patients with idiopathic scoliosis where
a single stimulus generates multiple I-waves.
16.3.4. Build up phenomenon
If the multipulse technique (in a nondeeply anesthetized
patient) with a repetition rate of 1 or 2 trains/s is performed, each consecutive muscle MEPs will have an
increased amplitude. In cases where the intensity of stimuli is just slightly above the threshold, the first few
trains will not generate muscle MEPs at all. At the same
time, the D-wave amplitudes will remain the same. This
phenomenon is called build up and is due most likely
to the activation of another system in generating of muscle MEPs than the FNCT.
Fig. 8. A: In this patient, a single stimulus delivered over the exposed motor hand area elicits a single D-wave and multiple
I-waves. The ISI should be long enough to prevent the second set of D and I, elicited by a second stimulus, from falling into
the CT axon refractory period resulting from the previous waves (as is the case in trace (B). When the ISI is 5.9 ms (C) and
8 ms (D), this will not occur, resulting in a sufficient numbers of D- and I-waves to elicit MEPs. The stimulus is marked by
arrow and D-wave by asterisk. (Reprinted from Deletis et al., 2001b with permission from the International Federation of
Clinical Neurophysiology.)
244
Fig. 9. Shows mapping of the CT by D-wave collision technique (see text for explanation). S1 transcranial electrical
stimulation (TES); S2 spinal cord electrical stimulation; D1 control D-wave (TES only); D2 D-wave after combined
stimulation of the brain and spinal cord; R the cranial and caudal electrodes for recording the D-wave in the spinal epidural space. Below left: negative mapping results (D1 D2). Below right: positive mapping results (D2 wave amplitude
significantly diminished after collision). Insert: hand held stimulating probe over the exposed spinal cord. (Modified from
Deletis and De Camargo, 2001.)
16.4. The D-wave collision technique

This is a novel technique that involves simultaneous
TES of the motor cortex with concurrent stimulation
of the CT from the surgically exposed spinal cord
caudal to a lesion site (Fig. 9)
This newly developed technique has allowed us to:
(a) intraoperatively map the anatomical position of the
CT within the surgically exposed spinal cord and (b)
semiquantitatively determine the number of unaffected,
desynchronized, and blocked (nonconducting) FNCT.
16.4.1. Mapping of CT intraoperatively
The application of the D-wave collision technique
has allowed us to find the anatomical position of
the CT within the surgically exposed spinal cord.
In 18 patients undergoing surgery for thoracic
IMSCTs, we elicited D-waves by TES and recorded
them cranially and caudally in relation to the spinal
cord tumor. Simultaneous with TES, we stimulated
the surgically exposed spinal cord (caudal to the tumor
site) with a miniature bipolar hand held probe
(#5522.010 INOMED, GmbH, Germany). The tips of
the probe were 1 mm apart and delivered constant current stimuli up to 2.5 mA in intensity, 0.5 ms in duration, and 1 Hz as the repetition rate. Whenever the
stimulating probe was in a close proximity to the

CT, the D-wave elicited by TES collided with the
anti-D-wave elicited by the spinal cord stimulation.
This collision resulted in the diminished amplitude
of the D-wave recorded cranially to the lesion (see
Fig. 9) (Deletis and De Camargo, 2001).
Preliminary experience with this method indicates
an impressive ability to selectively map the spinal
cord for the CT. Using this method, the CT can be
localized within 1 mm. This is in concordance with
the other CT mapping techniques, which show the
same degree of selectivity and can be used to stimulate cranial nerves and motor nuclei within the brainstem (Deletis et al., 2000).
16.4.2. Calculations on the number of fast
neurons in the CT
Using the D-wave collision technique one can semiquantitatively determine the number of unaffected,
desynchronized, and blocked (nonconducting) FNCT,
Fig. 10.
Pathology of the spinal cord may affect FNCT in
different ways: some of the fibers passing through
the lesion site may not be affected at all while some
of them may become desynchronized. Therefore, each
fiber may conduct D- and I-waves with a different
245
Fig. 10. Schematic of semiquantitative calculation of unaffected, desynchronized, and blocked fast CTs neurons (see text
for explanations). S1 transcranial electrical stimulation; S2 stimulation of the spinal cord caudal to the lesion site;
R1 recording of the D-wave cranial to the spinal cord lesion; R2 recording of the D-wave caudal to the spinal cord
lesion. (Modified from Deletis, 2006 with permission from the International Federation of Clinical Neurophysiology.)
conduction velocity. The third group of FNCT

becomes blocked by the lesion and is nonfunctional
and nonconducting (Fig. 10A). D-waves from unaffected FNCT can be easily recorded caudally to the spinal cord lesion while other waves from desynchronized
or blocked FNCT cannot be recorded. However, even
desynchronized FNCT can elicit muscle MEPs
(Fig. 5) indicating that these fibers remain functional.
Before collision takes place (TES alone) (Fig. 10B)
the amplitude of the D-wave recorded cranially to the
lesion semiquantitatively represents the total amount
of FNCT approaching the lesion. The amplitude of
the D-wave recorded caudally to the lesion represents
the amount of unaffected axons of the FNCT transverse lesion. The amplitude difference between the
cranial (R1) and caudal (R2) D-waves represents the
total amount of blocked desynchronized FNCT
which does not contribute to the D-wave amplitude
caudal to the lesion and thus is not recorded.
After collision takes place (TES together with spinal cord stimulation) (Fig. 10C), the amplitude of the
cranially recorded D-wave (R1) represents the
amount of blocked (nonconducting) FNCT since
stimulation of the spinal cord caudal to the lesion
activates both unaffected and desynchronized FNCT,
but not blocked fibers since they can not conduct
through the lesion. Activation of desynchronized
and unaffected fibers by caudal stimulation of the
spinal cord produces a volley that travels as an
anti-D-wave moving antidromically. This wave
collides with the descending D-wave elicited by
TES and diminishes its amplitude.
By subtracting the D-wave amplitude difference
in Fig. 10B (R1R2) from the D-wave amplitude in
Fig. 10C (R1) one can semiquantitatively determine
the amount of desynchronized but still functioning FNCT: [(nonconducting desynchronized)]
nonconducting desynchronized.
246
Furthermore, the D-wave collision technique can

be successfully employed to intraoperatively evaluate the exact levels and extension of the spinal cord
lesion, while determining the upper level of the last
nonaffected segment of the spinal cord (see Fig. 11).
A collision with diminution of the D-wave, recorded
cranially, but not caudally, can be obtained only with
the stimulation of spinal cord CTs up to the lesion
level. The technique can be judged against killed
end potential recordings of the D-wave as it
was performed in the same paraplegic patient
(Fig. 11).
16.5. Importance of simultaneous intraoperative

recordings of the D- and I-waves and muscle
MEPs for the preservation of the functional
integrity of the motor tracts
The importance of using combined methodologies to
intraoperatively monitor the functional integrity of
the motor system is
to prevent intraoperatively induced neurological
injury to the motor system

to predict transient paraplegia and differentiate it
from permanent paraplegia
Fig. 11. Upper: 12-year-old boy undergoing surgery for intramedullary anaplastic astrocytoma after having been paraplegic
for 4 days prior to surgery. D-wave collision technique has been used to depict the exact level of nonfunctioning CTs. Note
the decrement of the D-wave at the nonaffected left and right CT (marked with an asterisk). The collision could not be
obtained caudally to the level of injury (market with a dot). Lower: the identical level of injured CTs corresponds with
the appearance of injury potential (marked with an asterisk).
247
to better understand the mechanism of injury

to document the exact moment of injury
to use monitoring records as a significant educa-
tional tool
Combining muscle MEPs and D-wave monitoring is extremely important in IMSCT surgeries
because of the risk of selectively injuring the CT
and consequently inducing a permanent motor deficit. Monitoring muscle MEPs only may suffice for
spine surgery because injury to the cord is expected
to be diffused and not selectively limited to the CT
or other descending motor systems. Therefore, in
spine surgery, mMEP monitoring would likely
reflect an injury to the CT. Nevertheless, we very
much encourage monitoring the functional integrity of the motor system using both methods whenever possible except for IMSCT surgeries when we
consider necessary.
Figure 12 illustrates an example of intraoperative
neurophysiological criterion for transient paraplegia
(disappearance of muscle MEPs from the lower
extremities, with preserved 50% amplitude of the
D-wave at the end of surgery). Surgeons respecting
this criterion will continue with the surgery despite
the disappearance of muscle MEPs, thus achieving
the surgical goal of gross total IMSCT resection. This
is a critical during resection of spinal cord ependymomas because complete resections result in complete recoveries.
Table 1 was based on the combined techniques
of the D-wave and muscle MEPs monitoring, and
on the postoperative motor outcome of 93 patients
who underwent surgery for IMSCTs (Kothbauer
et al., 1998). During this type of surgery, muscle
MEPs usually disappear before any significant
change is recorded in the D-wave. The loss of muscle MEPs alerts the surgeon to use caution, but also
offers a window of opportunity to proceed with the
surgery until the D-wave declines to 50% of its
baseline, at which time, the decision must be made
to aggressively remove the tumor or not. Using a
50% drop in the D-wave as the criterion to halt
the surgery, not one single patient of the 93 had a
permanent motor deficit 38% had transient motor
deficits that were predicted by the loss of muscle
MEPs.
When injury to the nervous system happens
intraoperatively, it is very important to document
the exact time of injury along with all the circumstances surrounding this event. We have introduced
Fig. 12. Muscle MEPs recorded from right and left TA

muscle (left) and D-wave recorded epidurally over the lower
cervical spinal cord (right). During surgery for a Th4Th6
ependymoma, we observed the progressive loss of the left
(time 14:55) and right (time 15:38) tibialis anterior muscles
(left panel). Meanwhile, the D-wave amplitude decreased
by 30% of its initial value and surgery was not stopped.
However, at 16:00, when the attempt was made to remove
the last piece of tumor adherent to the left anterolateral corticospinal tract, we observed a dramatic drop of D-wave
amplitude and surgery was transiently stopped. Corrective
measures were taken and total tumor removal was finally
achieved 20 min later more or less when the D-wave amplitude recovered to 7.90 mV. At finishing, D-wave amplitude
had recovered and it was 11.32 mV (86% of its opening
baseline amplitude). This neurophysiological data indicates
postoperative transient paraplegia. The patient completely
recovered within a week. (Reprinted from Sala et al., 2004
with permission from Springer, New York.)
a method of simultaneously recording the video

stream from the operating microscope together with
the neurophysiological data. The postfact analysis
using frame by frame as presented in Fig. 13, shows
the usefulness of this method. Todays modern
Table 1
Principles of motor evoked potential (MEP) interpretation (from Deletis, 1999)
D-wave
Muscle MEP a
Motor status
Unchanged or
3050% decrease
Unchanged or
3050% decrease
>50% decrease
Preserved
Unchanged
Lost uni- or
bilaterally
Lost bilaterally
Transient
motor deficit
Long-term
motor deficit
In the tibialis anterior muscle(s).
248
technology enables us to simultaneously record on

DVD or similar electronic media a composite video
consisting of a video stream from the operating
microscope, neurophysiological traces, and anesthesiological data. These comprehensive recordings are
very useful for solving previously unresolved events
occurring during (neuro) surgical procedures. The
use of frame by frame off-line analysis via composite
video recordings, used in the detection of the exact
moment of surgically induced injury to the spinal
cord, is presented in Fig. 13.
16.6. Transient paraplegia and its clinical
significance
Fig. 13. Three consecutive frames from the composite video

taken during surgery for intramedullary spinal cord tumor are
shown (starting from the top to the bottom). Each frame consists of an upper part with an image from the operative microscope and a lower part with continuous recordings of MEPs
from the right and left tibialis anterior muscles (R.TA, L.
TA), as well as the right and left abductor hallucis muscles
(R.AHB, L.AHB). In the first frame, the surgeon had just
started to take the part of the tumor adherent to the lateral (left)
side of the tumor cavity using an ultrasonic aspirator. In the
second and third frames, the exact moment is shown when
the surgeons actions damages the spinal cord resulting in
the loss of the left tibialis anterior muscle MEPs. Fortunately,
the D-wave, which had been simultaneously recorded,
remained stable (not shown). Disappearance of MEPs in one
leg with preserved D-wave postoperatively resulted in transient left leg monoplegia. The patient recovered completely
in two weeks (see also Table 1). Arrow in each frame indicates
tip of ultrasonic aspirator probe.
The disappearance of D-wave during surgery of IMSCT

is a neurophysiological marker for plegia. The illustration in Fig. 14 shows the importance of the D-wave
monitoring and its predictor value. Fortunately, the
D-wave usually does not suddenly disappear but
gradually, giving the surgeon the opportunity to
modify his strategy before the injury becomes irreversible. The only exception (sudden and irreversible disappearance of the D-wave) is likely related to a vascular
insult that can occur between the spinal cord and the
tumor in the vascular territory of the anterior spinal
artery if bipolar coagulation is used on the ventral cleavage plane.
A clinical phenomenon that we observe more
commonly is the so-called transient paraplegia.
The definition of transient paraplegia came into
our practice with consistent observation of patients
operated on for IMSCT acquiring severe motor deficits (not always necessarily a plegia) and then
invariably recovering a few days or weeks after surgery. These patients peculiarity presented the loss
of one or more mMEPs from the involved limbs intraoperatively but preservated a D-wave above the 50%
of its baseline amplitude.
These patients are likely those in whom IOM
prevented an irreversible injury to the spinal cord.
One possible explanation (Deletis and Kothbauer,
1998) is that, unlike from the D-wave, muscle
MEPs are generated by the combined action of
CT neurons as well as those of the propriospinal
and other descending tracts within the spinal cord.
Lesioning this non-CT supportive motor system
can result in early postoperative transient motor
deficits, but, in the long term, will be functionally
compensated for by the CTs. This group of patients
with reversible motor deficit is likely the one that
249
Fig. 14. Left: stable D-wave through an intramedulary spinal cord tumor removal. The presence of epidural MEPs at the end
of the procedure predicts a long-term preserved motor function postoperatively. Right: rapid loss of D-wave during intramedullary spinal cord tumor removal. In this instance, the patient became permanently quadriplegic after surgery. (Modified from
Morota et al., 1997 with permission from Lippincott Williams and Wilkins.)
may account for the benefits of IOM in IMSCT

surgery. As suggested by our recent clinical experiences (Sala et al., 2006), it seems that the benefit of
IOM becomes evident only a few months after surgery
for IMSCT, while early neurological outcomes
are similar in monitored and nonmonitored patients.
These clinical data likely reflect the transient paraplegia phenomenon. Monitored patients, where the
D-wave has been preserved at the end of the procedure,
retain the potentiality for clinical recovery. Accordingly, we have learned that the preservation of the
D-wave represents a very strong prognostic indicator
of a good long-term motor outcome and we are not
aware, so far, of a single exception to this rule.
16.7. Conclusion
Intraoperative neurophysiology of the spinal cord
motor pathways has dramatically evolved over the past
decade, offering unique opportunities in applying basic
neurophysiological principles of the spinal cord to clinical practice. Through IOM, we have learned some of
the neurophysiological mechanisms behind the changes
in the postoperative neurological outcome of patients
operated on the spine or with spinal cord pathologies.
Similarly, clinical importance of the transient paraplegia phenomenon has provided new insights into
the basic pathophysiology of the spinal cord.
Future goals should include the possibility to

quantify D-waves and gather semiquantitative information about fast conducting neurons in each CTs.
Furthermore, we should critically evaluate the roles
of more quantitative muscle MEP criteria (based
on threshold, amplitude, and waveform) for spinal
procedures other than IMSCT surgery.
Neurophysiological criteria to warn the surgeon
to take corrective measures, prevent neurological deficits, and predict the prognosis have been well established
for spine and spinal cord surgery. In contrast, such
criteria are far less standardized for MEP monitoring
during brain surgery involving motor areas (Neuloh
et al., 2004). One of the limitations in this field is the
existence of only one anecdotal report of D-wave recordings during these procedures (Yamamoto et al., 2004),
which mean that data on specific CT recording is very
scarce. Reaching the same level of reliability for MEPs
for brain surgery as we have already achieved for spinal
cord surgery will be a challenge for future research.
References
Amassian, VE and Deletis, V (1999) Relationships between
animal and human corticospinal responses. In: W Paulus,
M Hallett, PM Rossini and JC Rothwell (Eds.), Transcranial Magnetic Stimulation. Electroencephalogr. Clin.
Neurophysiol. Suppl. 51. Elsevier, Amsterdam, pp. 7992.
250
Boyd, SG, Rothwell, JC, Cowan, JMA, Webb, PJ, Morley,
T, Asselman, P and Marsden, CD (1986) A method of
monitoring function in corticospinal pathways during
scoliosis surgery with a note on motor conduction
velocities. J. Neurol. Neurosurg. Psychiatr., 49:
251257.
Deletis, V (1999) Intraoperative neurophysiological monitoring. In: D McLone (Ed.), Pediatric Neurosurgery:
Surgery of the Developing Nervous System. 4th ed.
W.B. Saunders, Comp Publ, Philadelphia, pp. 12041213.
methodology used to monitor the functional integrity
of the motor system. In: V Deletis and J Shils
(Eds.), Neurophysiology in Neurosurgery. A Modern
Intraoperative Approach. Academic Press, California,
pp. 2551.
Deletis, V (2006) Intraoperative neurophysiology of the
corticospinal tract of the spinal cord. Functional Neuroscience: Evoked Potentials and Related Techniques.
Suppl. To Clinical Neyrophysiology, Vol. 59. Elsevier,
Deletis, V and De Camargo, AB (2001) Interventional
neurophysiological mapping during spinal cord procedures. Stereotact. Funct. Neurosurg., 77: 2528.
Deletis, V and Kothbauer, K (1998) Intraoperative neurophysiology of the corticospinal tract. In: E Stalberg,
HSE Sharma and YHS Olsson (Eds.), Spinal Cord Monitoring. Springer, Vienna, pp. 421444.
Deletis, V, Sala, F and Morota, N (2000) Intraoperative
neurophysiological monitoring and mapping during
brainstem surgery: a modern approach. Oper. Tech.
Neurosurg., 2(2): 109113.
Deletis, V, Isgum, V and Amassian, V (2001a) Neurophysiological mechanisms underlying motor evoked potentials (MEPs) in anesthetized humans. Part 1. Recovery
time of corticospinal tract direct waves elicited by pairs
of transcranial stimuli. Clin. Neurophysiol., 112:
438444.
Deletis, V, Rodi, Z and Amassian, V (2001b) Neurophysiological mechanisms underlying motor evoked potentials
(MEPs) elicited by a train of electrical stimuli. Part 2.
Relationship between epidurally and muscle recorded
MEPs in man. Clin. Neurophysiol., 112: 445452.
Jankowska, E, Padel, Y and Tanaka, R (1975) Projections of
pyramidal tract to cells a-motoneurons innervating hind
limb muscles in the monkey. J. Physiol. (Lond.), 249:
637667.
Journee, HL, Polak, HE and De Kleuver, M (2004) Influence
of electrode impedance on threshold voltage for transcranial electrical stimulation in motor evoked potential monitoring. Med. Biol. Eng. Comput., 42: 557561.
Kothbauer, K, Deletis, V and Epstein, FJ (1998) Motor
evoked potential monitoring for intramedulary spinal
cord tumor surgery: correlation of clinical and

neurophysiological data in a series of 100 consecutive
procedures. Neurosurg. Focus, 4(5): 19.
MacDonald, DB and Janusz, M (2002) An approach to
intraoperative neurophysiological monitoring of thoracoabdominal aneurysm surgery. J. Clin. Neurophysiol.,
19(1): 4354.
Morota, N, Deletis, V, Shlomi, C, Kofler, M, Cohen, H and
Epstein, F (1997) The role of motor evoked potentials
(MEPs) during surgery of intramedullary spinal cord
tumors. Neurosurgery, 41: 13271366.
Neuloh, G, Pechstein, U, Cedzich, C and Schramm, J
(2004) Motor evoked potential monitoring with supratentorial surgery. Neurosurgery, 54(5): 10611070.
Novak, K, De Camargo, AB, Neuwirth, M, Kothbauer, K,
Amassian, V and Deletis, V (2004) The refractory
period of fast conducting corticospinal axons in man
and its implication fro intraoperative monitoring of
motor evoked potentials. Clin. Neurophysiol., 115:
19311941.
Patton, HD and Amassian, VE (1954) Single and multiple
unit analysis of cortical state of pyramidal tract activation. J. Neurophysiol., 17: 345363.
Rodi, Z, Deletis, V, Morota, N and Vodusek, DB (1996)
Motor evoked potentials during brain surgery. Pflugers
Arch., 431: R291R292.
Rothwell, J, Burke, D, Hicks, R, Stephen, J, Woodforth, I
and Crawford, M (1994) Transcranial electrical stimulation of the motor cortex in man: further evidence for
the site of activation. J. Physiol. (Lond.), 481:
243250.
Sala, F, Lanteri, P and Bricolo, A (2004) Intraoperative
neurophysiological monitoring of motor evoked potentials during brainstem and spinal cord surgery. In: JD
Pickard (Editor in Chief), VV Dolenc, J Lobo
Antunes, HJ Reulen, M Sindou, AJ Strong, N de
Tribolet, CAF Tulleken and M Vapalahti (Eds.),
Advanced and Technical Standards in Neurosurgery.
Vol. 29, pp. 133169.
Sala, F, Palandri, G, Lanteri, P, Deletis, V, Facioli, F and
Bricolo, A (2006) Motor evoked potential monitoring
improves outcome after surgery for intramedullary spinal cord tumor: a historical control study. Neurosurgery,
58: 11291143.
Scisciolo, G, Bartelli, M, Magrini, S, Biti, GP, Guidi, L and
Pinto, F (1991) Long term nervous system damage from
radiation of the spinal cord: an electrophysiological
study. J. Neurol., 238: 915.
Szelenyi, A, De Camargo, AB and Deletis, V (2003)
Neurophysiological evaluation of the corticospinal tract
by D wave recordings in young children. Childs Nerv.
Syst., 19: 3034.
Tamaki, T, Takano, H and Takakuwa, K (1985) Spinal cord
monitoring: basic principles and experimental aspects.
Cent. Nerv. Syst. Trauma, 2: 137149.

Tamaki, T, Takano, H and Nakagawa, T (1986) Evoked spinal cord potentials elicited by spinal cord stimulation and
its use in spinal cord monitoring. In: RQ Cracco and I
Bodis-Wollner (Eds.), Evoked Potentials. Alan Liss,
Ulkatan, S, Neuwirth, M, Bitan, F, Minardi, C, Kokoszka,
A and Deletis, V (2006) Monitoring of scoliosis surgery
with epidurally recorded motor evoked potentials
251
(D wave) revealed false results. Clin. Neurophysiol.,
117(9): 20932101.
Yamamoto, T, Katayama, Y, Nagaoka, T, Kobayashi, K
and Fukaya, C (2004) Intraoperative monitoring of
the corticospinal motor evoked potential (D-wave): clinical index for postoperative motor function and functional recovery. Neurol. Med. Chir. (Tokyo), 44:
169180.

M.R. Nuwer (Ed.)
252
CHAPTER 17
Recording MEPs to transcranial electrical stimulation

and SEPs to peripheral nerve stimulation
simultaneously from the spinal cord
David Burke*
Institute of Clinical Neurosciences, Royal Prince Alfred Hospital and University of Sydney, Sydney, NSW 2006, Australia
Perhaps the most common indication for spinal cord

monitoring is orthopedic surgery to correct scoliosis,
but even in a practice restricted to intraoperative
monitoring during surgery to correct spinal deformity, the clinical neurophysiologist will need to have
a number of techniques at his or her disposal. The
majority of patients are neurologically healthy adolescents, often female, who present with a worsening
deformity for which surgery is essentially cosmetic.
However, preexisting neurological disease may also
precipitate scoliosis. For example, cerebral palsy
and muscular dystrophy can present different challenges to the clinical neurophysiologist, with defective corticospinal function in cerebral palsy and loss
of compound muscle action potentials (CMAPs) in
muscular dystrophy. In addition, the operative procedures may involve approaches which are posterior
to the cord, anterior, or combined anterior and posterior approaches. Reoperation after previous surgery
may render the insertion of epidural leads for monitoring difficult. Other monitoring techniques may be
required in these situations.
When surgery involves a patient with a preexisting neurological deficit, or when it is for a neurosurgical indication and involves direct surgery on the
spinal cord rather than the spine, the risk of exacerbating the preexisting deficit is considerable, higher
than in operations on neurologically intact patients.
The onus on the monitoring team is then to use the
most reliable technique possible, given the circumstances, for the early detection of adverse changes.
*
Correspondence to: David Burke, M.D., D.Sc., Office of

Research and Development, Level 2, Medical Foundation
Building K-25, University of Sydney, Sydney, NSW
2006, Australia.
Tel.: 61-2-9036-3091; fax: 61-2-9036-3092.
E-mail: d.burke@med.usyd.edu.au (D. Burke).
The most sensitive monitoring procedures are those

that allow more than one modality of function to be
monitored, in as close to real time as possible, and at
more than one level of the neuraxis. This chapter will
focus on one such technique developed for scoliosis surgery, highlighting its advantages and disadvantages and
its current place in the monitoring armamentarium.
17.1. Methodological considerations and
underlying rationale
The technique relies on single-shock transcranial
electrical stimulation of the motor cortex to produce
a descending corticospinal volley and simultaneous
stimulation of the tibial nerves in the popliteal fossae
to produce an ascending somatosensory volley, both
of which are then recorded from the spinal cord using
epidural leads. Routinely, two epidural recordings
are made using separate leads: the upper positioned
over the cervical enlargement, and the lower over
the lumbar cord. The corticospinal and somatosensory volleys can be distinguished by their latencies
(shorter for the corticospinal volley) and by the direction of propagation between the two leads: downwards for the corticospinal volley, that is, earlier at
the upper level and later at the lower, and the reverse
for the somatosensory volley (Fig. 1).
17.1.1. Cortical stimulation
The motor cortex is stimulated through a pair of spiral subdermal needle electrodes. While current
density is theoretically higher at the tip of needle
electrodes than with surface electrodes, this author
has found needle electrodes more convenient and
has noted no evidence of scalp injury.
In our original studies, the anode was positioned
at the vertex, and the cathode was positioned laterally
MOTER EVOKED POTENTIALS
253
Motor Cortex
Stimulating Electrodes
anode
cathode
Epidural Recording
high-thoracic
low-thoracic
Tibial Nerve
Stimulation
7 cm
earth
A
Simultaneous Stimulation of Cortex and Tibial Nerves
high-thoracic
low-thoracic
Corticospinal
Somatosensory
5 V
3 ms
Fig. 1. A: The technique used for recording simultaneously descending corticospinal volleys in response to anodal electrical stimulation of the motor cortex and ascending somatosensory volleys in response to stimulation of the tibial nerves in
the popliteal fossae. The stimuli were delivered simultaneously, and the evoked volleys were recorded at two levels from
the spinal cord, as shown in the lower traces. The descending corticospinal volley had a shorter latency and propagated
down the spinal cord, while the ascending somatosensory volley had a longer latency and propagated up the spinal cord.
B: The traces are duplicate averages of 10 sweeps. Negativity for the corticospinal volley is shown as an upward deflection,
and negativity for the somatosensory volley is shown as a downward deflection reflecting the fact that the volleys approach
the bipolar recording electrodes from opposite directions. Reprinted from Burke and Hicks (1998), with kind permission of
Springer Science and Business Media.
over one hand area of the motor cortex (Burke et al.,

1990, 1992, 1995; Hicks et al., 1992). It was confirmed
that effective activation of the motor cortex occurs
under the anode with transcranial electrical stimulation
(Phillips and Porter, 1977; Rothwell et al., 1991). In
addition, it was found that this orientation produced
larger corticospinal volleys for less stimulus artifact
than any other. For this reason, we do not advocate a
midline frontal cathode, as might seem optimal for midline leg-area cortex. It is conceivable that, with weak stimuli, the preferred orientation might favor one leg area
and one corticospinal projection over the other, but at
the intensities required to produce reliable potentials,
both projections contribute roughly equally to the corticospinal volley recorded from the spinal cord. Over
more recent years, particularly when it was intended to
record, CMAPs from limb muscles, in response to
repetitive transcranial stimulation, we have placed the
anode over one hand area (C3 or C4) and the cathode
over the other (Bartley et al., 2002). This orientation
provides good CMAPs to brief trains of transcranial
stimuli and satisfactory epidural volleys to single stimuli, but the latter poses a slightly greater stimulus
artifact.
The transcranial stimuli consist of square-wave
electrical pulses of 50-ms duration, delivered at a rate
of less than once per 3 s, from a commercial stimulator
(Digitimer D185, London, UK). [Originally capacitively coupled stimuli (time constant 50 ms) were
delivered from a Digitimer D180A, but there seems
to be little reason to continue with this device, which
was designed to minimize discomfort in conscious
patients. Moreover, the D180A cannot deliver trains
of stimuli, as is optimal for CMAP recordings.] The
stimulus intensity is increased to produce a large, simple D-wave, as in the lower panel of Fig. 1, and this
usually requires an intensity of 250350 V. D-wave
activity typically becomes discernible at 150 V,
and at higher intensity, the volley becomes more complex as the D-wave decomposes into a number of components and I-waves become more intrusive (Fig. 3,
see below).
254
17.1.2. Tibial nerve stimulation

Both tibial nerves are stimulated at the same time as the
motor cortex. The tibial nerve stimulation requires the
use of the two stimulators on the recording electromyography (EMG) machine. Synchronization with the transcranial stimulation is achieved by triggering the EMG
machine externally from the transcranial stimulator.
In an anesthetized paralyzed patient, the intensity of
tibial nerve stimulation cannot be graded according to
EMG recordings or sensory reports, but this is not necessary anyway. We increase the stimulus intensity to
near-maximum to ensure that a maximal afferent volley enters the cord. In all likelihood, that volley
involves a major peroneal component in addition to
the tibial nerve input. Stimulation of both tibial nerves
simultaneously, rather than separately, prevents the
identification of unilateral pathology. A unilateral deficit that reduced the recorded volley by <50% might
go undetected if a decrement of 50% is the criterion
for abnormality. With our current technique, we
believe that dysfunction can be detected before this
level is reached. In addition, the bilateral input produces larger somatosensory potentials, and this is
important if only 1020 sweeps are being averaged,
as is our practice. These conditions are not optimal
for recording somatosensory volleys, but this is largely
offset by the greater discrimination that comes from
recording corticospinal volleys in the same sweeps.
17.1.3. Epidural recording
The recording leads used are bipolar cardiac pacing
leads with an interelectrode distance of 23 cm. Bipolar leads eliminate slower, less synchronized activity,
and reduce the size of the descending volley slightly,
but are more convenient and are less prone to artifact
for the surgeon than a monopolar lead with a remote
reference. Cardiac pacing leads are sufficiently rigid to
prevent flexion back on themselves during insertion.
Negativity at the rostral electrode of the bipolar
recording is usually displayed upwards so that the corticospinal volley appears upright and the somatosensory volley, which reaches the caudal electrode of
each pair first, appears inverted (Fig. 1, lower panel).
The leads are inserted into epidural space by the surgeon once the spine has been exposed. The upper lead is
passed up the epidural space to overlie the cervical
enlargement. The lower electrode is positioned to overlie the low-thoracic cord. Both are secured in place by
sutures. The electrodes intrude minimally into the operative field, but may still be inadvertently displaced. This
occurs more often with the lower electrode, but its
D. BURKE
removal is recognized quickly because there is loss of

both the motor volley and the somatosensory volley at
only one of the two electrodes. Preservation of the
somatosensory volley in the low-cervical recording
when it is absent in the low-thoracic recording indicates
a problem with the lower electrode.
17.1.4. Position of the epidural leads
The electrodes lie posterior to the cord and effectively
record the activity of both posterior columns and both
lateral columns without clear side differentiation. The
exact level for the upper electrode is not critical: the volley is quite large, usually >10 mV, and is easily identified in single sweeps. The site of the lower electrode is
critical: if too low it will overlie the conus medullaris
and cauda equina (not the lumbosacral cord) and the
bipolar recording will then not detect the descending
corticospinal volley. On a few occasions, we have averaged multiple sweeps at this site, but have failed to
detect a ventral root volley, presumably because, in an
anesthetized patient, few, if any, motoneurones can be
discharged by even a strong corticospinal volley.
17.1.5. Earthing
The earth lead is generally placed on the patients
buttock or thigh.
17.1.6. Filtering
To obtain reliable recordings, the signal is filtered
using a bandpass of 500 Hz5 kHz (Burke et al.,
1990, 1992). The 500-Hz high-pass filter distorts
the potentials, reducing their amplitude and altering
their latency (Fig. 2). These effects are greatest on
standing waves (the major component of the somatosensory potential at the low-thoracic level) and less
on traveling waves (the D- and I-waves, and the
somatosensory volley at the low-cervical region).
However, these distortions do not interfere with the
ability to identify the potentials and do not impact
monitoring. More importantly, the heavy filtering stabilizes the trace and minimizes stimulus artifacts
(Fig. 2), allowing continuous recording, with interruptions only on account of excessive operative artifact
or the use of the diathermy (Burke et al., 1990, 1992).
17.2. The recorded volleys
17.2.1. Size and morphology of potentials at the
two levels
The amplitude of the maximal D-wave is commonly
1030 mV at the low-cervical level and some

Descending Motor
255
Ascending Sensory
Cortical
stimulus
high thoracic
75 V
low thoracic
500 Hz-10 kHz
20 V
2 ms
150 V
225 V
10 V
2 ms
300 V
20 V
2 ms
200 Hz-2 kHz
375 V
450 V
20 Hz-2 kHz
50 V
2 ms
VS 19F, 600 V, isoflurane 2%
Fig. 2. The effects of different bandpass filters. Recordings

as in Fig. 1, except that the polarity is inverted. Note that,
to obtain a recording, the input amplification had to be
reduced when the 20 Hz high-pass filter was used. (Duplicate averages of 25 responses; female aged 19; transcranial
stimulus intensity 600 V; isoflurane 2%.) From Burke et al.
(1992), with permission from International Federation of
Clinical Neurophysiology.
610 mV at the low-thoracic level. Averaging is not

necessary to define these potentials, but 1020 sweeps
are still routinely averaged because the D-wave is
smaller at the more caudal site and because ascending somatosensory volley at the low-cervical level is
rarely more than a few microvolts, at the most (as in
Figs. 1 and 2).
17.2.2. Effects of stimulus intensity on the
corticospinal volley
Transcranial electrical stimulation at an intensity of
150 V produces a small D-wave that appears at
the low-cervical level with a latency of 34 ms
and, at the low-thoracic level, 3 ms later. This potential grows with stimulus intensity, but then becomes
Corticospinal
volley
Somatosensory
volley
Fig. 3. The effect of increasing stimulus intensity (time

constant 50 ms) on the corticospinal volley recorded at the
low-cervical site. Electrode configuration: anode at vertex,
cathode 7 cm lateral to vertex. Duplicate averages of
25 sweeps are superimposed for each stimulus intensity.
The low-amplitude polyphasic deflection to the right of
each trace represents the ascending somatosensory volley
set up by bilateral stimulation of the tibial nerves in the
popliteal fossae. The cortical and somatosensory stimuli
were delivered simultaneously, at the onset of the sweeps.
The dotted vertical line indicates the D-wave. Note that this
peak is completely replaced at 450 V by an earlier peak
that began to appear at 225 V. The deflections after the
dotted vertical line are I-waves (except for the late deflections which are the ascending somatosensory volley). From
Burke et al. (1990), with permission.
quite complex as the D-wave decomposes into components arising at lower levels along the descending axon and as I-waves begin to appear after
the D-wave (Fig. 3). Modest increases in stimulus
intensity can cause the site of activation of some
low-threshold axons in the volley to become deeper.
With strong stimuli (>750 V), most corticospinal
axons will be activated at the decussation of the
pyramidal tract, where there is a bend in the axons
(Burke et al., 1990; Rothwell et al., 1994).
The variability of small components of the corticospinal volley is too high for them to be valuable
256
D. BURKE
for monitoring (Burke et al., 1995). For monitoring

purposes, the optimal stimulus is modest, adjusted so
that it produces a large, simple D-wave (as in Figs. 1
and 2), without evidence of components from stimulation of corticospinal axons deep in the brain or
brainstem. The necessary voltages are usually
250350 V, though this can vary with the choice of
stimulus waveform, stimulus duration, and other variables including the age of a patient, thickness of the
skull, and the type and impedance of electrodes used.
I-waves are not useful for monitoring using the
technique described here. I-waves can be controlled
by the use of a modest stimulus intensity and adjustment of the depth of anesthesia. This will usually
result in a corticospinal volley consisting of only a
large, simple D-wave (as in Fig. 1).
17.2.3. Effects of anesthesia
A major advantage of the present technique for
recording corticospinal volleys is that the system is
activated directly at the nodes of Ranvier close to
the cell body to produce the D-wave (Amassian
et al., 1987; Rothwell et al., 1991), and the descending volley is recorded from the same axons. In other
words, the technique involves the direct stimulation
of, and recording from, the same axons; there is no
intervening synapse. Accordingly, the recording is
unaffected by changes in the excitability of the motor
cortex in contrast to the responses to transcranial
magnetic stimulation (TMS) or the spinal motoneurone
pool (contrast with CMAPs).
Vertex Anodal Stimulus 75 V
Being unconscious depresses the excitability of

the motor cortex and the spinal motoneurone pool,
and all anesthetic agents, even nitrous oxide, depress
excitability even further. Intravenous and volatile
anesthetics (specifically thiopentone and propofol;
sevoflurane and isoflurane) seem to have similar
depressant effects on I-waves in the corticospinal
volleys produced by transcranial electrical stimulation when given in equipotent anesthetic concentrations (Hicks et al., 1992; Woodforth et al., 1999).
The I-wave depression contributes to the depresssion
of the motor evoked potential (MEP) to TMS.
It is commonly stated that anesthetics have no
affect on the D-wave. However, volatile anesthetics,
such as isoflurane, ethrane, and sevoflurane, can
reduce the size of the D-wave an effect that is particularly obvious for liminal D-waves (Fig. 4; Hicks
et al., 1992). This is because they probably block
persistent and transient Na currents at the node of
Ranvier (Burke et al., 2000), thereby increasing the
stimulus required to produce the D-wave by 30%.
In practice, this does not constitute a limitation
because merely increasing stimulus strength will
compensate for the depressant action. The depressant
effect on I-waves cannot be overcome as readily by
increasing stimulus strength, and as a result, the corticospinal D-wave can be considered relatively
immune to the effects of anesthetics when compared
with I-waves and CMAPs.
Unlike the situation when recording CMAPs, full
muscle relaxation is preferable for epidural recordings, and this contributes to artifact-free recordings.
Stimulus 375 V
ethrane
2%
isoflurane
2%
1%
1%
0.2%
0%
10 V
2 ms
0%
AR 15 m
LP 14f
Fig. 4. The effects of withdrawing inhalational anesthetics on liminal D-waves and on I-waves recorded at the low-cervical
site. For each subject, the intensity of the transcranial stimulus was constant throughout (75 V left; 375 V right) and was
such that when the end-tidal concentration of ethrane (left) or isoflurane (right) was 2%, only a small D-wave was recorded.
Note the appearance of I-waves on withdrawal of isoflurane on the right. (Negativity upwards; duplicate averages of
25 responses.) From Burke et al. (1992), with permission from International Federation of Clinical Neurophysiology.
Stimulus-induced contraction of paraspinal muscles

can produce a stimulus-locked artifact of slow onset
and waveform in the epidural recording when the
degree of muscle relaxation is too light.
17.2.4. Reproducibility of potentials and criteria
for abnormality
Large, simple D-waves, such as in Fig. 1, are highly
reproducible provided that anesthesia, blood pressure,
and temperature remain stable (Burke et al., 1995).
Large, simple D-waves of amplitude 1030 mV
have a coefficient of variation (standard deviation
divided by mean amplitude) of <0.08, but smaller
waves in the corticospinal volley (I-waves and subcortically originating components of the D-wave) have
coefficients of >0.2 (Burke et al., 1995). This
implies that, under near-perfect conditions, a 20%
decrease in amplitude in single trials would be outside
2.5 SD of the normal mean for a large, simple D-wave.
These data apply to the low-cervical recording in
Fig. 1. The D-wave at the low-thoracic site is smaller
and is presumably more variable. In addition, the
ascending somatosensory potentials must be defined
in the same sweeps as the descending corticospinal
volley, as in Fig. 1. Therefore, it is recommended
that 1020 trials be averaged. A 20% decrease in
amplitude of the averaged D-wave can be detected
by eye. Further recordings are then undertaken to
confirm the changes and determine whether they
involve the two volleys at the two levels. If the
changes form a coherent pattern, the surgeon should
be notified that something could be amiss.
Artifactual changes in the potentials are readily
detected with this recording technique, and this
reduces the number of false-positive warnings given
to the surgeon, an issue that affects surgical enthusiasm for the value of monitoring. For example, the
loss of the volleys from the lower electrode can occur
because of inadvertent dislodging of the electrode
during surgery. This can be identified as clearly artifactual because the preservation of the somatosensory volley at the upper electrode would otherwise
be impossible. Loss of one modality at both electrodes could indicate selective involvement of, for
example, corticospinal pathways before somatosensory, but should raise concerns about the stimulator,
its leads, and electrodes. Rarely, selective loss of
one of the two modalities has occurred as a result
of spinal dysfunction, but generally, there are also
less prominent changes in the other volley.
257
17.2.5. Real-time monitoring

This is important if dysfunction is to be reversed. To
detect and then confirm deterioration requires two
averages of 1020 sweeps for the D-wave (and the
somatosensory volley is usually identifiable with this
number). With one trial every 3 s, this implies
1 min to detect an abnormality and a further minute
to confirm it before alerting the surgeon. It usually
takes longer than 2 min to convince the surgeon that
something is amiss.
A further advantage of near-real-time monitoring
is that it allows the surgeon to use the monitoring
as a guide to the extent of safe correction of scoliosis.
Similarly, in neurosurgical operations on the cord,
it is reassuring when delicate surgical dissections
produce few changes, if any, in the recorded volleys.
17.3. The advantages and disadvantages of
epidural recordings of corticospinal and
somatosensory volleys simultaneously
17.3.1. Advantages
The recordings involve two different modalities,

sensory and motor, and therefore monitor a large

volume of the spinal cord.
The recording of two modalities at two levels
greatly reduces false-positive reports to the surgeon.
The recordings provide information close to real
time.
The potentials are relatively resistant to the depressant effects of anesthesia.
The operation can be performed under complete
muscle relaxation.
The amplitude of the D-wave has a very low trialto-trial variability. A 20% loss of amplitude at the
low-cervical electrode does not normally occur
unless there has been a change in anesthetic conditions and/or electrode displacement.
Corticospinal and somatosensory volleys can be
recorded using epidural leads in virtually all neurologically intact patients, except when there are technical problems or scarring due to a previous
operation which prevent the insertion of leads.
Recordings can often be obtained in patients with
preexisting pathology.
A surgeon, confident in the monitoring technique,
can use the preservation of the potentials as an
indication that further intervention can be
attempted either further spinal cord dissection
or further correction of the scoliosis.
258
17.3.2. Disadvantages
The monitoring procedure is invasive.
The insertion of recording electrodes into the epidu-

ral space is usually appropriate only when there is

a posterior approach to the cord or posterior fossa.
The technique is not suitable when surgery is on the
distal cord, cauda equina, or nerve roots.
Epidural recordings do not allow identification of
unilateral dysfunction.
The caudal epidural lead is often dislodged during
spinal surgery and then must be replaced. [The
insertion and reinsertion of epidural leads carry a
risk of epidural hemorrhage, but this is low.]
Epidural leads have a metalmaterial interface, and
it is difficult to guarantee their sterility. Leads
should be one-use-only and then disposed, and
this increases cost.
17.4. Summary of protocol for epidural recordings

of corticospinal and somatosensory volleys
simultaneously (modified from Burke, 2005)
1. Insert spiral subdermal needle electrodes into the
scalp at C3 and C4 for stimulation. [Note that there
may be less stimulus artifact with the anode at
the vertex and cathode 7 cm to one side, but
C3C4 allows for a change to CMAP recordings.]
2. Trigger the EMG-recording machine from the
transcranial stimulator.
3. The surgeon inserts bipolar cardiac pacing electrodes at the high- and low-thoracic levels,
threads them up the epidural space to overlie the
cervical and lumbar enlargements, and then
secures them by sutures.
4. Filter recorded activity 500 Hz5 kHz, amplify to
10 mV per division, sweep duration 30 ms, average 1020 sweeps. Enable the artifact reject
facility.
5. Use single 50 ms stimuli from the Digitimer D185
stimulator, repeated about once every 3 s, and
increase stimulus intensity as necessary, usually
to 250350 V, to produce a large simple D-wave.
The D-wave should appear at the low-cervical
electrodes at about 34 ms and at the low-thoracic
electrodes after a further 34 ms (Fig. 1).
6. Increase the intensity of the stimuli delivered by
the stimulators on the recording machine to the
tibial nerves in the popliteal fossae, so that
somatosensory volleys can be recorded in the
same traces. These stimuli will be delivered
simultaneously with the transcranial stimulus,
D. BURKE
once every 3 s. A large standing wave should

appear in the low-thoracic recording at 1113 ms
and the smaller traveling wave should reach the
low-cervical level 4 ms later (Fig. 1).
17.5. Place of the technique in the
monitoring armamentarium
The technique described here was developed for
scoliosis surgery, and is one of the most reliable
and sensitive techniques available for spinal cord
monitoring. However, for routine scoliosis surgery
on patients with a normal nervous system, the risks
of inadvertent injury are low, though very real.
Recordings of CMAPs to brief trains of transcranial
electrical stimuli are more convenient, can be used
for a wider range of operative approaches, and do
not intrude into the operative field. As a result, the
present technique has now been relegated to a secondary role in monitoring for scoliosis surgery,
despite the greater variability of the CMAPs and
the care that must be taken with anesthesia and muscle relaxation with CMAP recordings (e.g., Bartley
et al., 2002).
CMAP recordings may not be obtainable even in
patients with no preexisting neurological disease,
though this is often because of anesthetic considerations. It is common not to be able to record reproducible CMAPs in patients with neurological
diseases (e.g., Bartley et al., 2002). In these patients
and in those undergoing neurosurgical procedures,
the clinical neurophysiologist must have other techniques in the armamentarium in order to be able to
offer a monitoring service.
References
Amassian, VE, Stewart, M, Quirk, GJ and Rosenthal, JL
(1987) Physiological basis of motor effects of a transient stimulus to cerebral cortex. Neurosurgery, 20:
7493.
Bartley, K, Woodforth, IJ, Stephen, JPH and Burke, D
(2002) Corticospinal volleys and compound muscle
action potentials produced by repetitive transcranial
stimulation during spinal surgery. Clin. Neurophysiol.,
113: 7890.
Burke, D (2005) Intraoperative monitoring of corticospinal
function using transcranial stimulation of the motor
cortex. In: M Hallett and S Chokroverty (Eds.), Transcranial Magnetic Stimulation. Elsevier ButterworthHeinemann, Philadelphia, 2nd ed., pp. 365379.

Burke, D and Hicks, RG (1998) Corticospinal volleys
evoked by transcranial electrical and magnetic stimulation. In: E Stalberg, HS Sharma and Y Olsson (Eds.),
Spinal Cord Monitoring. Springer, Vienna, New York,
pp. 445461.
Burke, D, Hicks, RG and Stephen, JPH (1990) Corticospinal volleys evoked by anodal and cathodal stimulation of the human motor cortex. J. Physiol. (Lond.),
425: 283299.
Burke, D, Hicks, R, Stephen, J, Woodforth, I and Crawford, M
(1992) Assessment of corticospinal and somatosensory
conduction simultaneously during scoliosis surgery.
Burke, D, Hicks, R, Stephen, J, Woodforth, I and Crawford, M
(1995) Trial-to-trial variability of corticospinal volleys in
human subjects. Electroencephalogr. Clin. Neurophysiol.,
97: 231237.
Burke, D, Bartley, K, Woodforth, IJ, Yakoubi, A and
Stephen, JPH (2000) The effects of a volatile
259
anesthetic on the excitability of human corticospinal
axons. Brain, 123: 9921000.
Hicks, R, Burke, D, Stephen, J, Woodforth, I and Crawford, M
(1992) Corticospinal volleys evoked by electrical stimulation of human motor cortex after withdrawal of volatile
anaesthetics. J. Physiol. (Lond.), 456: 393404.
Phillips, CG and Porter, R (1977) Corticospinal Neurones.
Their Role in Movement. Academic Press, London.
Rothwell, JC, Thompson, PD, Day, BL, Boyd, S and Marsden, CD (1991) Stimulation of the human motor cortex
through the scalp. Exp. Physiol., 76: 159200.
Rothwell, J, Burke, D, Hicks, R, Stephen, J, Woodforth, I and
Crawford, M (1994) Transcranial electrical stimulation of
the motor cortex in man: further evidence for the site of
activation. J. Physiol. (Lond.), 481: 243250.
Woodforth, IJ, Hicks, RG, Crawford, MR, Stephen, JPH
and Burke, D (1999) Depression of I waves in corticospinal volleys by sevoflurane, thiopental and propofol. Anesth. Analg., 89: 11821187.

M.R. Nuwer (Ed.)
260
CHAPTER 18
Transcranial electrical MEP with muscle recording

Anil Mendiratta* and Ronald G. Emerson
Department of Neurology, Columbia University College of Physicians and Surgeons, New York, NY 10032, USA
18.1. Introduction
Activation of human cortex by direct electrical stimulation was first demonstrated by Penfield in 1939
(Penfield and Boldrey, 1939). Forty years later,
Merton and Morton devised an electrical stimulator
suitable for use through the intact skull; it generated
brief (time constant <10 ms) high-voltage (up to
2,000 V) shocks that when applied to the cranium
elicited twitches in the contralateral limbs (Merton and
Morton, 1980). Not long thereafter, the technique was
adapted to intraoperative monitoring of motor function.
18.2. Rationale
Somatosensory evoked potential (SEP) monitoring is
well established as a safe and effective technique for
intraoperative evaluation of spinal cord integrity. It
has been shown to reduce postoperative neurological
deficits associated with spine surgery (Nuwer et al.,
1995). However, as a measure of motor tracts function, SEPs serve only a surrogate, as they are
mediated primarily by the dorsal column pathways
within the spinal cord and may be preserved despite
damage to the descending motor pathways (Ginsburg
et al., 1985; Emerson, 1988; Hilibrand et al., 2004).
Historically, the wake-up test was used to
confirm the functional integrity of the motor tracts during spine surgery (Vauzelle et al., 1973). This test
entails reversal of anesthesia and neuromuscular
blockade, requires a voluntary response from the
patient, and provides only a single snapshot of spinal
cord function. Motor evoked potential (MEP) monitoring, in contrast, allows ongoing assessment of motor
tract function during the operative procedure.
*
Correspondence to: Anil Mendiratta, M.D., Department of
Neurology, Columbia University College of Physicians and
Surgeons, New York, NY 10032, USA.
Tel.: +1-212-305-1742; fax: +1-212-305-1450.
E-mail: am441@columbia.edu (A. Mendiratta).
As SEPs and MEPs are mediated by different

neural pathways within distinct vascular territories,
monitoring of both can provide a more complete
assessment of spinal function. Furthermore, concurrent monitoring of both SEPs and MEPs provides an
important measure of redundancy, as intraoperative
injury often compromises both sensory and motor
pathways (Nagle et al., 1996). Technical factors such
as electrical interference and the effects of anesthetic
drugs may occasionally limit the utility of either test
in an individual patient, leaving the remaining monitoring modality to signal the occurrence of injury.
Transcranial stimulation activates spinal motor
pathways selectively, because intervening thalamic
synapses prevent antidromic firing of spinal sensory
tracts. Transcranial electrical stimulation activates cortical pyramidal cells both directly and indirectly via cortical interneurons, to produce D- and
I-waves, respectively (Amassian et al., 1987; Day
et al., 1987). These may both be recorded from
epidural electrodes. D- and I-waves summate in the
anterior horn to produce transsynaptic activation of
the spinal motor neurons and the resultant compound
muscle action potential (CMAP) (Day et al., 1987;
Taylor et al., 1993).
Epidural D-wave monitoring may be employed
along with, or as an alternative to, monitoring of
CMAPs. D-wave monitoring is discussed separately,
in Chapter 16. Each technique can offer certain
advantages; advantages of CMAP monitoring include
Evoked CMAPs depend on both the upper and
lower motor pathways, including the spinal motor

neurons and roots, whereas D-waves measure only
the integrity of the descending corticospinal tract.
Unilateral motor system injury may be detected by
a unilateral loss of CMAPs. D-wave monitoring
does not permit detection of unilateral spinal cord
injury.
Transcranially evoked CMAPs can be measured
from muscles innervated by the lowest sacral
roots, including the external anal sphincter. As such,

they can detect injury to the most caudal regions of
the spinal cord (Inoue et al., 2002; Haghighi and
Zhang, 2004). D-waves are often difficult to record
below the mid-thoracic cord. Their voltage normally diminishes caudally, limiting the utility
D-wave monitoring in cases in which the lower
spinal cord is at risk (MacDonald and Janusz, 2002).
Placement of recording electrodes over or in muscles is simple and entails little risk. In contrast,
epidural recording require that special electrodes
be placed within the operative field or percutaneously. These may involve greater risk.
Potential limitations of the CMAP recording technique include
Sequential, identical transcortical electrical stimuli
produce CMAPs differing from one to the next,

often considerably, in waveform amplitude and
morphology. This intertrial variability may confound interpretation. The variability of CMAP
response is likely explained by activation of only
a small number of different, low-threshold spinal
motor neurons by each descending corticospinal
volley (Woodforth et al., 1996; Van Dongen et al.,
1999b) (see Fig. 1).
CMAP recordings are more sensitive to halogenated anesthetics than D-waves. CMAPs are dependent on transsynaptic activation spinal motor
neurons; these agents may depress both synaptic
transmission and/or spinal motor neuron function
(Sloan and Heyer, 2002).
CMAP recordings are incompatible with complete
neuromuscular blockade. Paralysis must be either
avoided entirely (Kothbauer, 2002; MacDonald and
Janusz, 2002), or maintained at a constant, subparalytic level. Maintenance of a constant level of
blockage introduces some complexity into anesthetic management, and limited or no blockade
may result in electromyographic (EMG) artifact
which can compromise SEP recordings.
Since complete muscle paralysis is not possible,
MEP monitoring using CMAPs is generally accompanied by some movement of the patient. As such,
MEP monitoring must be performed intermittently,
rather than continuously, during procedures
where movement would interfere with surgery. In
contrast, D-waves, may be recorded with complete
paralysis and are fully compatible with continuous
monitoring in these cases.
261
18.3. Technique
Either specially designed capacitive-coupled constant
voltage stimulators or standard constant-current stimulators may be used. While both types are satisfactory, constant voltage stimulators produce more rapid
charge delivery (1 c/s vs. 0.1 c/s) requiring somewhat lower total charge for equivalent stimulation
(Hausmann et al., 2002).
Various stimulating electrode types are currently in use, including needle, corkscrew, and
surface cup electrodes (Legatt, 2002). Cup electrodes require attachment with collodion, which
can be time-consuming and thus impractical in
the operating room. Corkscrew electrodes offer
stable placement and relatively low impedances
but cause some scalp trauma. In our experience,
standard electroencephalograph (EEG)-type needle
electrodes are ideal. Secured with appropriate adhesives, they provide stable contact, and placement is
relatively simple, secure, and only essentially nontraumatic.
Corticospinal volleys are elicited most effective
by anodal transcranial stimulation (Burke et al.,
1990). CMAPs can be recorded in both upper extremities and lower extremities using a single pair
of electrodes placed at C3 and C4 (Bartley et al.,
2002). Electrode polarity is alternately reversed,
producing maximal CMAP amplitudes in the limbs
contralateral to the anode (see Fig. 1). Low intensity anodal stimulation will selectively elicit
CMAPs in the contralateral upper extremity; only
at greater intensities are CMAPs elicited bilaterally in upper and lower extremities. In that case,
deeper penetration of the electrical stimulation
likely activates the corticospinal tracts at the
level of the internal capsule (Deletis, 2002). Alternatively, lower extremity muscles may be activated
with the anode at Cz and with cathodes variously
placed at C3/C4, C1/C2, several centimeters anterior
to Cz, or around the cranial base (A1, A2, Fpz)
(Ubags et al., 1996; Deletis, 2002). Two large (3 cm
long, 0.4 mm diameter) interconnected stainless
steel needle electrodes at Cz (anode) along with a
large ground strip-type cathode over the forehead has been proposed by Journee and colleagues,
who have noted a strong correlation between
voltage thresholds required to elicit MEPs and electrode impedances (Journee et al., 2004a) (Fig. 2A
and B).
262
A. MENDIRATTA AND R.G. EMERSON
Fig. 1. A and B: Consecutive transcranially elicited compound muscle action potential (CMAP) responses. Note intertrial
variability.
CMAPs are typically monitored from distal limb

muscles, as these have richer corticospinal tract
innervation than proximal muscles (Jankowska
et al., 1975). Commonly used muscles include the
abductor pollicus brevis, adductor hallucis brevis,
and tibialis anterior. In our experience, CMAPs can
generally be monitored easily in proximal muscle
as well, including deltoid, biceps, and quadriceps.
Standard EEG-type needle electrodes are well suited
for recording.
It may be helpful to monitor CMAPs from other
muscles as well. Dong and colleagues recently demonstrated that the transcranially evoked facial nerve
CMAPs can be monitored reliably should the proximal
facial nerve be inaccessible to direct stimulation during skull-based surgery (Dong et al., 2005). Inoue
and colleagues reported successful recordings from
the external anal sphincter in 73% of patients undergoing elective spine surgery using C3/C4 stimulation
(Inoue et al., 2002).
Trains of rapidly delivered transcranial stimuli,
rather than a stimulus, greatly facilitate CMAP generation (Jones et al., 1996; Rodi et al., 1996). In
one study recording from both epidural space and
muscles, D-waves alone were elicited by single or
dual pulses, but D-wave, I-waves, and CMAPs were
generated by train of 3 or 4 pulses (Rodi et al.,
1996). Furthermore, as the interpulse interval was
reduced from 10 to 1 ms, the CMAP amplitude

increased and latency decreased progressively
(Jones et al., 1996). These findings are best
explained by temporal summation of multiple Dand I-waves inputs be spinal motor neurons (Taylor
et al., 1993).
Attempts have been made to enhance CMAPs that
were too small or inconsistent to be reliably monitored.
Preconditioning with a transcranial pulse train
(4 pulses, 2 ms interstimulus interval, 100 ms pulse
width) 1035 ms prior to the stimulation pulse has been
used to increase the amplitude of small (<100 mV)
CMAPs (Journee et al., 2004b). Spatial facilitation,
with a train of cutaneous stimuli applied to the foot
(within the sensory receptive field of the withdrawal
reflex of the tibialis anterior muscle) 50100 ms prior
to a single pulse transcranial stimulus, increased CMAP
amplitude and also made CMAPs recordable in some
patients in whom they had been otherwise absent
(Andersson and Ohlin, 1999). As only single transcranial pulses were needed and facilitation was limited to
only a few leg muscles, this technique may produce less
overall muscle contraction and movement than the standard multi-pulse technique.
Importantly, stimulus parameters required for
effective transcranial electrical stimulation vary significantly between patients. Stimulus intensity and
duration, as well as the number of pulses and interpulse interval, must be adjusted individually to obtain
the best response in a given patient. Adjustment of
electrode position, including placement of anode or
cathode electrodes centimeters from the standard
locations, may be helpful.
Typical stimulation parameters, which depend
significantly on level of paralytics and anesthetics,
are listed in Table 1. Stimulus intensity is initially
set at a low level, and gradually titrated to elicit a
reproducible response. Anodecathode polarity is
alternately reversed, as elicited responses are maximal contralateral to the anode.
18.4. Safety
Transcranial MEP stimulation has a favorable safety
profile. In a large, retrospective review of adverse
events in more than 15,000 published and unpublished transcranial electrical MEP monitored patients,
the most common adverse event was bite injury
(29 patients) (MacDonald, 2002). Others included
cardiac arrhythmia (five patients, some possibly coincidental), seizures (five patients, some possibly
263
coincidental), minor scalp burns (two patients),

mandibular fracture (one patient), and intraoperative
awareness (one patient). There were no reported neuropsychological or endocrine effects. Relative contraindications to transcranial electrical stimulation,
which should be considered on an individual basis,
include intracranial foreign bodies (including electrodes, shunts, clips), skull defects, cardiac pacemakers
and other implanted biomedical devices, cardiac disease/arrhythmia, and elevated intracranial pressure.
Some consider epilepsy to be a relative contraindication; we believe that most patients with epilepsy can
safely undergo transcranial stimulation as the risk of
seizure appears to be extremely low. Concurrent
EEG monitoring may be desirable in these cases.
18.5. Anesthesia
Successful MEP monitoring required careful
coordination and communication between the neurophysiologist and the anesthesiologist, as MEPs are
affected substantially by anesthetic and neuromuscular blocking agents. A detailed discussion of anesthetics is provided in Chapter 16, but a few key
points deserve consideration here.
Complete neuromuscular blockage is incompatible
with CMAP monitoring. Some favor complete avoidance of paralytic agent, arguing that partial paralysis
is inherently variable and therefore introduces an
uncontrollable variable in the interpretation of CMAP
responses (Kothbauer, 2002). On the other hand,
patient movement produced by transcranial stimulation can be distracting and may at time interfere with
surgery; when the operating microscope is in use during neurosurgery, MEPs have been likened by a surgeon at our institution to an 8.0 earthquake on the
Richter scale! In our experience and that of others
(Lang et al., 1996), effective monitoring of CMAPs
is possible with controlled partial neuromuscular
blockage using a continuous infusion of a nondepolarizing neuromuscular blocker, such as vecuronium (Sloan and Heyer, 2002). Typically, the infusion
of paralytic is titrated to allow one or two twitches to
train of four stimulation, that is, four peripheral
motor nerve stimuli delivered at a rate of 2 Hz (Sloan
and Heyer, 2002). An alternative technique entails
maintaining the amplitude of the peripherally elicited
M-wave response to between 5% and 50% of the baseline (pre-neuromuscular blockade) response (Sloan
and Heyer, 2002). It is essential that the level of neuromuscular blockade remain constant, as changes will
264
result in large variation MEPs. For this reason, the

paralytic agent must be delivered using a continuous
infusion rather than intermittent boluses.
In addition to limiting the degree of patient movement produced by stimulation, partial neuromuscular blockade greatly facilitates SEP monitoring.
Specifically, the use of a paralytic allows effective
recording of the subcortical components of the

SEP; these have the important characteristic of being
highly resilient to the effects of most anesthetics. As
discussed earlier, we believe that dual monitoring of
MEPs and SEPs provides a more complete assessment of spinal cord functioning, and provides an
important level of redundancy to the monitoring
(Fig. 2 continued)
265
Fig. 2. A and B: Consecutive transcranially elicited compound muscle action potential (CMAP) responses, with reversal of
anodecathode polarity with each trial. Note higher amplitude responses contralateral to the anode in each trial.
procedure. For these reasons, we believe these benefits

outweigh the potential variability in CMAP responses
introduced by partial neuromuscular blockade (see
Fig. 3).
Potent inhalational anesthetics (e.g., isoflurane)
and nitrous oxide interfere with CMAP MEPs;
D-waves, in contrast, are much more resilient. This

likely reflects the dependence of CMAP generation
on cortical excitability as well as synaptic transmission in the anterior horn, while D-waves depend only
on the direct electrical activation of cortical pyramidal cells (Haghighi et al., 1990; Kalkman et al.,
266
Table 1
Typical parameters for recording compound muscle
action potentials (CMAPs) elicited by transcranial
electrical stimulation
Stimulus intensity
Pulse duration
Number of pulses
Interpulse interval
Recording time sweep
Filter settings
50200 mA (constant current)

100800 V (constant voltage)
50500 ms (constant current)
50 ms (constant voltage)
17
14 ms
100 ms
301,000 Hz
1991). In contrast, total intravenous anesthesia using

propofol and opiate attenuates CMAPs to a much
lesser degree, facilitating MEP monitoring with
CMAPs (Pechstein et al., 1998; Scheufler and Zentner, 2002) (see Fig. 4). Nonetheless, CMAPs may
be monitorable, albeit with generally higher stimulation intensities, when moderate doses of inhalational
agents are employed (Van Dongen et al., 1999a;
Pelosi et al., 2001) (Fig. 4A and B).
The term anesthetic fade has been used describe
attenuation of MEP over the course of a surgical procedure, without an identifiable systemic, anesthetic, or
surgical cause. Lyon and colleagues studied 62 patients
who underwent surgery using either desflurane/nitrous
oxide/narcotic or desflurane/propofol/narcotic anesthetic protocols and observed that the threshold voltage
required to elicit MEPs was significantly higher by the
end of each case (Lyon et al., 2005). They further noted
that this increase in voltage correlated with the duration
of the procedure ( p < 0.01) and was most marked in
patients with preexisting myelopathies. While this
phenomenon has not been studied systematically, it
should be considered when interpreting decrements in
responses, particularly during lengthy procedures.
18.5.1. Interpretation
In contrast to SEP monitoring where identical
responses are elicited by consecutive stimulations,
CMAPs elicited by transcranial stimulation demonstrate significant intertrial variability (Woodforth
et al., 1996; Van Dongen et al., 1999b). As discussed
earlier, this is attributed to activation of different
pyramidal cells and spinal motor neurons with each
stimulus. This intertrial variability makes the interpretation of CMAP responses somewhat more
complicated than the interpretation of SEPs, and different approaches are employed.
One approach is simply to interpret the CMAP
MEP as an all-or-none phenomenon. Reviewing 100
consecutive patients undergoing surgery for intramedullary spinal cord tumors, Kothbauer and colleagues observed that only the presence or absence of
CMAP responses correlated with clinical outcome
(Kothbauer et al., 1998; Kothbauer, 2002). Specifically, presence or responses correlated with absence
of motor deficit in all cases, and complete loss of
responses correlated with postoperative motor
impairment with 91% specificity. No paralytics were
used; the authors monitored both D-waves and
CMAP responses simultaneously and found that
intraoperative changes were evident in all cases with
postoperative motor deficits.
Another approach is to establish a criterion, for
example, a percentage reduction of amplitude, as
threshold for informing the surgeon of significant
changes (Jacobs et al., 1999; Pelosi et al., 2002).
An alternative approach, and one which we favor
for SEP as well as for MEP monitoring, is to inform
the surgeon of any change, particularly of amplitude
or morphology, unexplained by systemic or anesthetic effects that exceeds the baseline variability
for that patient (Mller, 1995) (see Fig. 5). Although
this approach is somewhat subjective, we believe that
it avoids application of arbitrary and simplistic criteria, and enables the surgeon to address potential
causes for the change or to elect a wait and see
approach.
If averaging is employed to improving the recording quality, one must be cognizant that in contrast to
during SEPs, the averaging process alters the CMAP
MEP signal as well as the noise. When recording
short latency SEPs, it is assumed that sequential stimuli elicit identical physiological responses and that
the averaging process causes accompanying noise to
attenuate in a systematic manner. In the case of
CMAP MEPs, the signal is not invariant from one
stimulus to the next, and so averaging will both
diminish accompanying noise and produce a signal
waveform that is a composite of several somewhat
different waveform morphologies. Since several
responses must be acquired to produce an average,
this will necessarily increase the time required to
detect a change and provide feedback to the surgeon.
Another approach is to employ threshold-level
monitoring (Calancie et al., 1998, 2001). Once stable
anesthesia and neuromuscular blockade are achieved,

Fig. 3. Motor evoked potential (MEP) and somatosensory evoked potential (SEP) responses during partial neuromuscular blockade. Note that the paralysis allows the
subcortical components of the SEP to be easily discerned.
267
268
Fig. 4. A: Shows responses generated in a patient maintained on isoflurane. B: Shows responses generated in the same patient after withdrawal of isoflurane, on propofol and remifentanil only.
269
Fig. 5. Note the change in the compound muscle action potential (CMAP) responses, with a significant reduction in
amplitude compared to the baseline.
270
the minimal voltage threshold required to elicit a

CMAP response in each monitored muscle is determined. Subsequent increases in threshold voltage,
unexplained by technical, systemic, or anesthetic factors, are used as a guide for informing the surgeon of
potential motor tract injury. Utilizing this technique,
Calancie and colleagues have reported a high degree
of accuracy in predicting postoperative motor system
dysfunction.
As we have discussed the significant effects of
anesthetics, as well as systemic effects, we advocate
the use of control recordings. For example, when
patients are undergoing thoracolumbar procedures,
the use of upper extremity recordings allows easier
delineation of a systemic or anesthetic effect rather
than a focal injury at the operative site, as these will
be affected only in the former circumstance, not the
latter.
Transcranially evoked CMAP MEPs are substantially more difficult to record in patients with underlying neurological abnormalities. In a series on 82
patients, Bartley and colleagues found that satisfactory CMAP responses were obtainable in 84% of
neurologically normal patients, but in only 41% of
neurologically impaired patients (Bartley et al.,
2002). This is consistent with our experience that
CMAP responses are relatively easily obtained in
nearly all neurologically normal patients, but are particularly difficult to obtain in patients with underlying motor system dysfunction, such as myelopathy
or cerebral palsy.
References
Amassian, VE, Stewart, M, Quirk, GJ and Rosenthal, JL
(1987) Physiological basis of motor effects of
a transient stimulus to cerebral cortex. Neurosurgery,
20: 7493.
Andersson, G and Ohlin, A (1999) Spatial facilitation of
motor evoked responses in monitoring during spinal
surgery. Clin. Neurophysiol., 110: 720724.
Bartley, K, Woodforth, IJ, Stephen, JP and Burke, D (2002)
Corticospinal volleys and compound muscle action
potentials produced by repetitive transcranial stimulation
during spinal surgery. Clin. Neurophysiol., 113: 7890.
Burke, D, Hicks, RG and Stephen, JP (1990) Corticospinal
volleys evoked by anodal and cathodal stimulation of the
human motor cortex. J. Physiol. (Lond.), 425: 283299.
Green, BA (1998) Threshold-level multipulse transcranial electrical stimulation of motor cortex for

intraoperative monitoring of spinal motor tracts:
description of method and comparison to somatosensory
evoked potential monitoring. J. Neurosurg., 88:
457470.
HJ (2001) Threshold-level repetitive transcranial electrical stimulation for intraoperative monitoring of central motor conduction. J. Neurosurg., 95: 161168.
Day, BL, Rothwell, JC, Thompson, PD, Dick, JP, Cowan,
JM, Berardelli, A and Marsden, CD (1987) Motor cortex stimulation in intact man. 2. Multiple descending
volleys. Brain, 110(Pt. 5): 11911209.
of the motor system. In: V. Deletis and J. Shils (Eds.),
Neurophysiology in Neurosurgery: A Modern Intraoperative Approach. Academic Press, Amsterdam.
Dong, CC, MacDonald, DB, Akagami, R, Westerberg, B,
Alkhani, A, Kanaan, I and Hassounah, M (2005) Intraoperative facial motor evoked potential monitoring with
transcranial electrical stimulation during skull base surgery. Clin. Neurophysiol., 116: 588596.
Emerson, R (1988) Anatomic and physiologic bases of posterior tibial nerve somatosensory evoked potentials.
Neurol. Clin., 6: 735749.
somatosensory evoked potentials. Case report. J. Neurosurg., 63: 296300.
Haghighi, SS and Zhang, R (2004) Activation of
the external anal and urethral sphincter muscles by
repetitive transcranial cortical stimulation during spine
surgery. J. Clin. Monit. Comput., 18: 15.
Haghighi, SS, Madsen, R, Green, KD, Oro, JJ and Kracke,
GR (1990) Suppression of motor evoked potentials by
inhalation anesthetics. J. Neurosurg. Anesthesiol., 2:
7378.
Hausmann, ON, Min, K, Boos, N, Ruetsch, YA, Erni, T
and Curt, A (2002) Transcranial electrical stimulation:
significance of fast versus slow charge delivery for
intra-operative monitoring. Clin. Neurophysiol., 113:
15321535.
Hilibrand, AS, Schwartz, DM, Sethuraman, V, Vaccaro,
AR and Albert, TJ (2004) Comparison of transcranial
electric motor and somatosensory evoked potential
monitoring during cervical spine surgery. J. Bone Joint
Surg. Am., 86A: 12481253.
Inoue, S, Kawaguchi, M, Takashi, S, Kakimoto, M, Sakamoto, T, Kitaguchi, K, Furuya, H, Morimoto, T and
Sakaki, T (2002) Intraoperative monitoring of myogenic
motor-evoked potentials from the external anal sphincter muscle to transcranial electrical stimulation. Spine,
27: E454E459.
Jacobs, MJ, Meylaerts, SA, De Haan, P, De Mol, BA and
Kalkman, CJ (1999) Strategies to prevent neurologic

deficit based on motor-evoked potentials in type I and II
thoracoabdominal aortic aneurysm repair. J. Vasc. Surg.,
29: 4857; Discussion 579.
Jankowska, E, Padel, Y and Tanaka, R (1975) Projections
of pyramidal tract cells to alpha-motoneurones innervating hind-limb muscles in the monkey. J. Physiol.
(Lond.), 249: 637667.
Jones, SJ, Harrison, R, Koh, KF, Mendoza, N and Crockard, HA
(1996) Motor evoked potential monitoring during spinal
surgery: responses of distal limb muscles to transcranial
cortical stimulation with pulse trains. Electroencephalogr.
Journee, HL, Polak, HE and De Kleuver, M (2004a) Influence of electrode impedance on threshold voltage for
transcranial electrical stimulation in motor evoked
potential monitoring. Med. Biol. Eng. Comput., 42:
557561.
Journee, HL, Polak, HE, De Kleuver, M, Langeloo, DD and
Postma, AA (2004b) Improved neuromonitoring during
spinal surgery using double-train transcranial electrical
stimulation. Med. Biol. Eng. Comput., 42: 110113.
Kalkman, CJ, Drummond, JC and Ribberink, AA (1991)
Low concentrations of isoflurane abolish motor evoked
nitrous oxide/opioid anesthesia in humans. Anesth.
Analg., 73: 410415.
Kothbauer, K (2002) Motor evoked potential monitoring
for intramedullary spinal cord tumor surgery. In: V
Deletis and J Shils (Eds.), Neurophysiology in Neurosurgery: A Modern Intraoperative Approach. Academic
Press: Amsterdam.
Kothbauer, K, Deletis, V and Epstein, F (1998) Motorevoked potential monitoring for intramedullary spinal
cord tumor surgery: correlation of clinical and neurophysiological data in a series of 100 consecutive procedures. Neurosurg. Focus, 4, Article 1.
Lang, EW, Beutler, AS, Chesnut, RM, Patel, PM, Kennelly, NA,
Kalkman, CJ, Drummond, JC and Garfin, SR (1996)
Myogenic motor-evoked potential monitoring using partial
neuromuscular blockade in surgery of the spine. Spine, 21:
16761686.
Legatt, AD (2002) Current practice of motor evoked potential monitoring: results of a survey. J. Clin. Neurophysiol., 19: 454460.
anesthesia: the phenomenon of anesthetic fade.
MacDonald, DB (2002) Safety of intraoperative transcranial electrical stimulation motor evoked potential monitoring. J. Clin. Neurophysiol., 19: 416429.
MacDonald, DB and Janusz, M (2002) An approach to
intraoperative neurophysiologic monitoring of thoracoabdominal aneurysm surgery. J. Clin. Neurophysiol.,
19: 4354.
271
Merton, PA and Morton, HB (1980) Stimulation of the
cerebral cortex in the intact human subject. Nature,
285: 227.
Mller, AR (1995) Intraoperative neurophysiologic monitoring. Am. J. Otol., 16: 115117.
Nagle, KJ, Emerson, RG, Adams, DC, Heyer, EJ, Roye,
DP, Schwab, FJ, Weidenbaum, M, Mccormick, P,
Pile-Spellman, J, Stein, BM, Farcy, JP, Gallo, EJ, Dowling, KC and Turner, CA (1996) Intraoperative monitoring of motor evoked potentials: a review of 116 cases.
Neurology, 47: 9991004.
Nuwer, MR, Dawson, EG, Carlson, LG, Kanim, LE and
spinal cord monitoring reduces neurologic deficits after
scoliosis surgery: results of a large multicenter survey.
Pechstein, U, Nadstawek, J, Zentner, J and Schramm, J (1998)
Isoflurane plus nitrous oxide versus propofol for
recording of motor evoked potentials after high frequency repetitive electrical stimulation. Electroencephalogr. Clin. Neurophysiol., 108: 175181.
transcranial electrical stimulation during two anaesthetic regimens. Clin. Neurophysiol., 112: 10761087.
Pelosi, L, Lamb, J, Grevitt, M, Mehdian, SM, Webb, JK and
Blumhardt, LD (2002) Combined monitoring of motor
and somatosensory evoked potentials in orthopaedic
spinal surgery. Clin. Neurophysiol., 113: 10821091.
Penfield, W and Boldrey, E (1939) Somatic motor and sensory representation in the cerebral cortex of man as
studied by electrical stimulation. Brain, 60: 389.
Motor evoked potentials during brain surgery. Pflugers
Arch, 431: R291R292.
Scheufler, KM and Zentner, J (2002) Total intravenous
anesthesia for intraoperative monitoring of the motor
pathways: an integral view combining clinical and
experimental data. J. Neurosurg., 96: 571579.
Sloan, TB and Heyer, EJ (2002) Anesthesia for intraoperative neurophysiologic monitoring of the spinal cord.
Taylor, BA, Fennelly, ME, Taylor, A and Farrell, J (1993)
Temporal summationthe key to motor evoked potential
spinal cord monitoring in humans. J. Neurol. Neurosurg. Psychiatry, 56: 104106.
Ubags, LH, Kalkman, CJ, Been, HD and Drummond, JC
(1996) The use of a circumferential cathode improves
amplitude of intraoperative electrical transcranial myogenic motor evoked responses. Anesth. Analg., 82:
10111014.
Van Dongen, EP, Ter Beek, HT, Schepens, MA, Morshuis,
WJ, De Boer, A, Aarts, LP and Boezeman, EH (1999a)
Effect of nitrous oxide on myogenic motor potentials
evoked by a six pulse train of transcranial electrical
272
stimuli: a possible monitor for aortic surgery. Br. J.
Anaesth., 82: 323328.
Van Dongen, EP, Ter Beek, HT, Schepens, MA, Morshuis,
WJ, De Boer, A, Aarts, LP and Boezeman, EH (1999b)
Within patient variability of lower extremity muscle
responses to transcranial electrical stimulation with pulse
trains in aortic surgery. Clin. Neurophysiol., 110:
11441148.

Vauzelle, C, Stagnara, P and Jouvinroux, P (1973) Functional monitoring of spinal cord activity during spinal
surgery. Clin. Orthop. Relat. Res., 173178.
Woodforth, IJ, Hicks, RG, Crawford, MR, Stephen, JP and
Burke, DJ (1996) Variability of motor-evoked potentials
recorded during nitrous oxide anesthesia from the tibialis anterior muscle after transcranial electrical stimulation. Anesth. Analg., 82: 744749.

M.R. Nuwer (Ed.)
273
CHAPTER 19
Neurogenic mixed evoked potentials

Robert E. Minahan* and Allen S. Mandir
19.1. Introduction
Neurogenic mixed evoked potential (NMEP) refers to
an evoked potential of averaged responses recorded
over lower extremity peripheral nerves elicited with spinal stimulation rostral to the surgical field. NMEP initially stood for Neurogenic Motor Evoked Potentials as
this test was intended to monitor spinal motor pathways.
However, classically described NMEPs are now
thought to represent primarily a dorsal columnmediated sensory response (Su et al., 1992; Rose,
1998; Toleikis et al., 2000; Minahan et al., 2001). This
response may or may not contain a relatively minor
motor contribution depending on testing conditions
and methods (Toleikis et al., 2000; Pereon et al., 2002).
Thus, the term neurogenic motor evoked potential
is no longer appropriate and in response to this, some
have labeled these as spinally elicited peripheral nerve
responses (Toleikis et al., 2000; Minahan et al., 2001),
descending neurogenic evoked potentials (Leppanen,
2004), descending spinal cord evoked potentials
(Su et al., 1992; Haghighi et al., 1994), or neurogenic mixed evoked potentials (Pereon et al., 1999b;
Delecrin et al., 2000). The initials NMEP are ingrained
in the neuromonitoring lexicon and thus will be used
for this chapter referring to Neurogenic Mixed Evoked
Potential and not Neurogenic Motor Evoked Potential.
However, two caveats are needed. First, under some
conditions, the motor component is absent and, therefore, the signal is not truly mixed. Second, confusion
may result if this same name is applied to transcranial
motor evoked potentials (TcMEPs) recorded from
nerves (Burkholder et al., 2003).
NMEPs were introduced to meet a recognized need
to monitor spinal motor pathways in addition to sensory
*
Correspondence to: Robert E. Minahan, M.D., Department

of Neurology, Georgetown University, 3800 Reservoir Rd
Tel.: 1-202-444-8290; fax: 1-202-318-9146.
E-mail: rminahan@yahoo.com (R.E. Minahan).
pathways. In the 1980s, a number of investigators put

forward methods to elicit MEPs. Levy, in 1984,
reported a transcranial stimulation method (Levy
et al., 1984) and Machida, in 1985, reported a spinal
stimulation method using epidural stimulation
(Machida et al., 1985). In 1987, Owen et al. reported a
variation of Machidas technique using transspinal
stimulation with recording from peripheral nerves. This
method had the advantage of being non-invasive, easy
to obtain, and insensitive to usual anesthetic regimens.
As a result, this form of classic NMEP (as
it is described throughout this chapter) became a frequently used spinal cord monitoring tool by the early
1990s (Owen et al., 1988, 1991; Owen, 1990) and for
a time was even put forward as the standard of care
(Padberg et al., 1998).
The use of NMEPs subsequently waned with the realization that they may not reliably assess the anterior spinal cord and with the wider adoption of TcMEPs.
However, these signals and related testing may still be
a useful part of the neuromonitoring armamentarium;
so a discussion of the technique is productive. In addition
to description of classic NMEP testing, alterations to the
method that may produce a true MEP are discussed.
19.2. Technique
19.2.1. Stimulation methods
Spinal stimulation for NMEPs must be performed at a
level rostral to any potential site of spinal injury. This
can be achieved by needle electrodes placed against
the cervical lamina, needle electrodes placed in the
rostral surgical field, or epidural electrodes (WilsonHolden et al., 2000).
19.2.1.1. Method 1: percutaneous cervical lamina
stimulation
Cervical lamina needles are the simplest stimulating
method as they are relatively non-invasive and do
not require surgeon participation. Monopolar EMG
274
electrodes (insulated except at the tip) are typically

used. Many varieties of these electrodes are available
and the length of the needle shaft should be long
enough so that the needle tip abuts the bony cervical
lamina. Needle insertion is performed in a direction
slightly caudal to the perpendicular to avoid entering
the spinal canal. For similar reasons, this method
should not be used in patients with prior cervical
laminectomy.
Initial needle placement is first attempted on either
side of the C5 cervical level. If a satisfactory response
is not obtained, a number of stimulus adjustments may
improve the signal. These include reversing the delivered polarity, placement of one needle (usually the
anode) at a slightly higher cervical level, placement
of both electrodes at different levels on the same side,
or switching to alternate stimulation methods as
described below.
19.2.1.2. Method 2: stimulation within the
surgical field
Short, sturdy stimulating needles (JO5 electrodes, The
Electrode Store, P.O. Box 188, Enumclaw, WA 98022,
#JO5T) may be placed by the surgeon at the rostral end
of the operative field. With a posterior approach, the
surgeon can remove the tips of two sequential spinous
processes and insert the electrodes at these sites. With
an anterior approach, the surgeon inserts the electrodes
into sequential intervertebral spaces.
19.2.1.3. Method 3: epidural spinal stimulation
An epidural cathode can be placed by the surgeon at
the time of surgery (Pereon et al., 2002) with the
anode placed in the adjacent interspinous ligament.
Alternatively, this electrode may be placed under
guidance prior to surgery (Mochida et al., 1997).
Using laminar or JO5 electrode stimulation, stimulus
current penetration to the spinal cord may vary from
patient to patient. Thus, stimulation parameters should
be adjusted to the minimum stimulus that produces
a maximal response. Typical stimulation parameters
include a stimulation rate 4.7 Hz, intensity 250
400 V, and duration 0.31 ms.
Epidural stimulation provides a more reliable and
consistent delivery method (Wilson-Holden et al.,
2000). Parameters as described by Pereon et al.
(2002) are a stimulation rate 4.2 Hz, intensity 20
50 mA, and duration 1 ms. As with other methods,
R.E. MINAHAN AND A.S. MANDIR
the stimulus intensity range reflects the minimum

intensity that yields a maximal response.
19.2.3. Recording methods and parameters
Recording electrodes are typically subdermal EEG
electrodes placed over the posterior tibial nerve at
the popliteal fossae, or less commonly at the ankle.
The interelectrode distance is 3 cm. Filter bandpass settings are 302,000 Hz. Under usual conditions, less than 100 averages are needed for a robust
signal-to-noise ratio.
19.2.4. Anesthetic parameters
Classically described NMEPs are insensitive to commonly used anesthetic agents (Owen, 1993, 1999; Bernard et al., 1996). Neuromuscular blockade is typically
kept at very high levels to limit patient movement with
stimulation and to prevent recording of the highly variable myogenic component that may contaminate the
neurogenic signal (Schwentker et al., 1995).
Alterations of the classic NMEP to allow identification of motor components require highly specific
anesthetic regimens which are described below.
19.3. Signal interpretation
Signal morphology may range between simple bi/triphasic morphology (Owen et al., 1988) to highly polyphasic signals. Typically, when recording at the
popliteal fossa, the signals are oligophasic, whereas
when recording at the ankle, they are polyphasic
(Fig. 1). A loss of 80% of the baseline greatest peakto-peak amplitude is typically considered a significant
change indicating injury (Padberg et al., 1998; Owen,
1999) though a degree of amplitude loss as low as
60% (Owen, 1993; Pereon et al., 1998) or change in
morphology of the signal have also been proposed
(Phillips et al., 1995; Pereon et al., 1998). However,
like any test in neuromonitoring, interpretation of
changes will be altered based on initial signal quality
and variability, surgical activity correlating to signal
changes, and the confounding factors noted below.
Especially when not employing epidural stimulation, NMEP signal interpretation may suffer from variable stimulation efficiency. This often results from
movement of the stimulating needles and it is usually
a simple matter to return the electrodes to their initial
position. Any metal introduced into the surgical field
can cause degradation of the recorded signal by
275
N:
Left popliteal fossa
N:
Right popliteal fossa
1 uV
NR:
Normal
Amp 5
38
5 ms
NR:
Normal
1 uV
Amp 6
38
5 ms
A
N:
Left tibial n. at ankle
63 NR:
6 ms Normal
1 uV
N:
Right tibial n. at ankle
Amp 3
63 NR:
6 ms Normal
1 uV
Amp 4
B
Fig. 1. Morphology on NMEP signals may be simple A: or complex and polyphasic (B).
reducing stimulation efficiency from shunting of current (Schwartz et al., 1996). This often leads to a difficult interpretative scenario during spinal deformity
surgery as introduction of metal rods into the field
often corresponds to the portion of the procedure with
highest risk. Thus, signals may be lost at a time when
they are needed most and distinguishing the cause of
signal degradation between stimulus shunting from
surgical injury may be difficult or impossible. To help
distinguish these effects, it is extremely important
that signals are obtained after introduction of rods but
prior to application of corrective forces. When stimulus shunting does occur, signals can often be reestablished by adjusting stimulating electrode polarity or
position as described in the section of cervical laminar
stimulation methods. However, this process may require multiple adjustments of electrodes and even then
may not be successful, potentially leading to uncertainty and delay in corrective action.
In some cases, muscles adjacent to the NMEP
recording needles may be activated. In classic NMEP
recordings, these myogenic responses should be
avoided as they are often highly variable compared to
the neurogenic response (Owen, 1993) and emergence
of this myogenic component may serve to contaminate
the target signal (Schwentker et al., 1995). These myogenic signals may result from a motor component of
the transmitted signal, from dorsal column-mediated
activity that activates an H-reflex pathway or a combination of these two (Mochida et al., 1995).
In their favor, NMEPs may remain robust in the
face of high anesthetic gases (Owen, 1993; Bernard
et al., 1996) or organic brain dysfunction, situations
where somatosensory evoked potentials (SEPs) are
often altered or absent.
19.4. Research related to NMEPs
19.4.1. Experimental performance in animals
Conclusions from a number of initial animal studies
of NMEPs were very encouraging that NMEPs reliably assessed motor pathways. These studies reported
that NMEPs were (1) lost earlier than SEPs to direct
spinal cord compression (Owen et al., 1988), (2) reliably absent when paraplegia occurred due to spinal
ischemia (Owen et al., 1988), (3) more sensitive to
spinal distraction than SEPs (Owen et al., 1988,
1990a,b), and (4) more sensitive but less specific in
identifying ischemic cord injury (Kai et al., 1995).
For the most part, these studies showed near perfect correlation of NMEP loss and SEP preservation
with pure motor injury and SEP loss with NMEP
preservation to pure sensory injury. However, by
the early 1990s, it was increasingly felt that the
NMEP contained a significant sensory component
276
(Su et al., 1992; Kai et al., 1993). It was then proposed that the early and largest portion of the NMEP
signal is the motor component which was more sensitive to distraction than the later and smaller sensory
component. This smaller, later sensory contribution
was less sensitive to distraction and changed in correlation to the SEP (Kai et al., 1993). However,
subsequent and more convincing lesioning and collision studies described below suggest just the opposite
conclusion that if a motor component is contained
within the NMEP response, it is a small and late
component of the summated signal.
19.4.2. Experimental evidence localizing the tracts
assessed by NMEPs
Spinal cord lesioning experiments by Owen et al.
(1989) were reported to show reliable loss of NMEP
signals with anterior cordotomy or ventral rhizotomy
while NMEPs were reliably preserved during dorsal
rhizotomy that obliterated lower extremity SEPs.
However, soon after the publication of the above
report, a nearly opposite finding was made by Su
et al. (1992). They noted complete loss of the NMEP
signal with both dorsal column sectioning and with
dorsal rhizotomy and these latter findings were confirmed in at least two subsequent independent studies
(Haghighi et al., 1994; Mochida et al., 1995).
It is possible to remove any sensory component of
the NMEP by using collision techniques (Rose,
1998). Collision studies utilize a supramaximal stimulus delivered to the tibial nerve at the popliteal fossa
prior to spinal cord stimulation at the spinal level.
The interval between stimuli is timed such that the
ascending sensory volley meets the descending
motorsensory volley within the spinal cord. Ascending and descending sensory volleys collide with termination of all sensory transmission in axons
projecting to the tibial nerve. The descending orthodromic spinal motor volley will not collide with the
peripherally elicited antidromic motor volley because
the peripheral nerve motor activity will terminate at
the anterior horn cell body. Thus, any descending
motor volley is free to proceed to the anterior horn
cell and then into the lower extremities assuming that
this cell is not in its absolute or relative refractory
period. The propagated motor signal is then recorded
over the tibial nerve below the site of the initial colliding stimulation.
Initial reports of collision studies by Owen (1993)
suggested that the primary portion of the NMEP
response is motor and comprises the earliest portion

of the signal. However, these finding are suspect for
a number a reasons. That the motor component
should occur earlier is a highly unlikely result given
the motor pathway would encounter a synaptic delay
not present in sensory pathways at these levels.
Furthermore, the fastest peripheral sensory fibers
demonstrate faster conduction than motor fibers
(Dawson, 1956). In addition, an inadequate collision
technique was used in these studies, with the initial
colliding stimulus delivered below the final recording
site which will inevitably lead to incomplete collisions. Subsequent collision study reports in other
laboratories demonstrate that the only reliable component of the classic NMEP is sensory (Leppanen
et al., 1998; Toleikis et al., 2000) and that even with
modifications to allow a motor component to be
identified, the largest and earliest component is sensory (Pereon et al., 1995, 2002). Fig. 2 shows an
example of a classic NMEP response and its alteration after a right tibial nerve collision stimulus at
the popliteal fossa with NMEP recording at the ankle.
Criticism of collision studies includes the point that
Renshaw cell activation will occur due to activation of
ascending peripheral nerve and may suppress the
motor component of the NMEP. However, the inhibitory effect of Renshaw cell activation is both weak
(Van, 1981; Lindsay and Binder, 1991; Maltenfort
et al., 1998) and brief (Van, 1981; Lindsay and Binder,
1991). Thus, the consistent absence of a significant
motor component through a range of distal stimulation
to spinal stimulation interstimulus intervals (Toleikis
et al., 2000) suggests that Renshaw inhibition is not a
significant factor. In addition, NMEP signals using
special techniques to allow a maximal motor component do not show major changes in the late motor polyphasic component with or without the collision
stimulus (Pereon et al., 2002).
A final clinical indicator suggesting that NMEP
responses reflect dorsal column pathways is their
insensitivity to anesthetic agents, which has even
been touted as an advantage of NMEP testing over
SEPs (Owen, 1993; Bernard et al., 1996). However,
such robust insensitivity to anesthetic effects suggests a pathway lacking synapses which are readily
depressed in the presence of anesthetic agents. This
implies dorsal column transmission as synapses do
not exist at this level of the afferent sensory pathways
compared with motor pathways that include synaptic
transmission at the anterior horn cell (Zhou et al.,
1997; Pereon et al., 1999a; Zhou and Zhu, 2000).
277
Right ankle
Left ankle
N:
7
N:
7
22 NR:
0
8 ms Normal
2 V Amp 3
0
35 NR:
8 ms Normal
2 V Amp 3
N:
N:
22 NR:
0
8 ms Normal
2 V Amp 4
0
35 NR:
8 ms Normal
2 V Amp 4
No
collision
Collision
stimulation
(Supramaximal,
right tibial nerve)
Stimulus artifact
Fig. 2. NMEP collision study. The top two traces show the classic NMEP signals in the left and right tibial nerves without
colliding stimulus delivered. The bottom two traces show the unaffected left NMEP signal and nearly obliterated right
NMEP signal after collision.
19.4.3. Clinical performance of NMEPs

in humans
Four large series assessing classic NMEPs show that
they can be reliably recorded and their persistence
accurately predicts postoperative spinal cord function
in the vast majority of cases (Owen, 1993; Padberg
et al., 1998; Pereon et al., 1998; Lubitz et al., 1999).
These studies represent a total of 1,971 cases monitored with NMEPs and combining results show a true
positive rate of 0.6%, a false positive rate of 1.3%,
and a false negative rate of 0%, and all remaining cases
being true negative (98.1%). Note that the SEP false
negative rate was also 0% in this series. The primary
question in assessing NMEPs is whether they are superior or inferior to SEPs in identifying spinal cord
insults and whether they should be combined with
SEPs for optimal monitoring.
An early report strongly suggested that NMEPs are
superior to SEPs in identifying postoperative motor
injury in patients undergoing intramedullary spinal
procedures or vascular procedures that place the spinal
cord at risk (Owen et al., 1991). In this report, 7 of 16
patients undergoing intramedullary procedures had
postoperative deficits. In five of these patients, there
was paralysis associated with NMEP loss and intact
SEP signals while in the other two, there was postoperative sensory disturbance with intact NMEPs
intraoperatively and loss of SEPs. For vascular procedures in this report, there were two cases of postoperative paralysis, and in these cases, the NMEP was lost
while SEPs were preserved. This strict and consistent
correlation of NMEPs to motor injury and SEPs to sensory injury is peculiar in light of what we now know to
be the tracts assessed by the classic NMEP and unfortunately specific case details and tracings are lacking
for further inspection in this report. However, this phenomenon is supported by at least one well-documented
case report in which SEPs are preserved and NMEPs
are lost (Mustain and Kendig, 1991). These reports
suggest a higher sensitivity for motor injury with
NMEPs as compared to SEPs. Likely, this finding
reflects a higher sensitivity of NMEP to dorsal column
dysfunction compared to SEPs (Pereon et al., 1998).
However, it is clear that both classic NMEPs and SEPs
may remain unchanged in the face of postoperative
paraplegia. Therefore, both are imperfect measures of
isolated anterior spinal injury (Minahan et al., 2001).
In the authors experience of 500 NMEP cases,
we have seen one case with motor injury where
NMEPs were lost and SEPs were preserved and one
case with motor injury where SEPs were lost and
NMEPs were preserved. In every other instance, disassociation of NMEPs and SEPs represented a probable
false positive result in the signal that was degraded.
Other reports of a disassociation between SEPs and
NMEPs do not clearly distinguish between true positive and false positive findings as patients awakened
without neural deficits (Pereon et al., 1998; WilsonHolden et al., 2002).
The specificity of NMEPs appears to be somewhat
worse than that of SEPs with false positive findings
of 1.4% and 27.1% in patients with idiopathic scoliosis and preoperative spinal pathology, respectively,
while the false positive of SEPs in these same groups
was 0% and 12.7%, respectively (Wilson-Holden
et al., 1999). NMEP specificity may also suffer, if
the monitoring team is not familiar with the technical
278
demands of the technique (Owen, 1990; Padberg

et al., 1998). In addition, the rate at which NMEP
signals are artifactually lost with placement of instrumentation (Schwartz et al., 1996) is not known and
likely varies based on the monitoring teams awareness of this phenomenon, their repertoire of remedies
for it, and their ability to distinguish the initial insertion of metal in the field from the application of
corrective forces.
Finally, complications as a result of both NMEPs
and SEPs are low. NMEP stimulation often leads to
patient movement, but with full neuromuscular
blockade, this is minimal. One report describes profound transient hypotension correlated to NMEP
stimulation at the cervical lamina which did not
recur when stimulating needles were placed in the
surgical field (Hays and Schwengel, 1999). The
authors have also encountered recurrent transient
hypertension in association with NMEP stimulation.
In this case, NMEP monitoring was stopped for the
remainder of the procedure and there were no
adverse sequela (unpublished data). Hemodynamic
effects due to NMEPs likely reflect brainstem or
sympathetic trunk activation. No adverse sequela
was seen in the cases described above, but potential
for more severe results are present, especially if the
causal role of NMEP stimulation is not immediately
recognized.
19.5. Spinal stimulation techniques allowing
motor tract assessment
Using neurogenic recordings, the sensory component
of the standard NMEP overwhelms any motor contribution. As a result, classically described NMEPs are
not a reliable test of motor tracts. However, a number
of techniques exist whereby the spinal cord is stimulated and motor signals are recorded in the lower
extremities allowing motor tract evaluation of the
intervening segment. One such technique utilizes
the previously described collision techniques (Pereon
et al., 2002), while others utilize muscle recording in
the lower extremities (Machida et al., 1985, 1988a,b,
1989, 1990; Taylor et al., 1993; Mochida et al., 1995,
1997).
19.5.1. Collision techniques to allow only motor
components to be recorded
A collision technique may be used to remove the
sensory component and reveal underlying motor
contribution to the signal. Unlike standard NMEPs,

this requires epidural stimulation and propofol-based
anesthetic in order to minimize anesthetic inhibition
at the anterior horn cell, thus allowing motor transmission. As described above for collision techniques,
an ascending supramaximal stimulus is delivered at
the tibial nerve at the popliteal fossa prior to epidural stimulation at the cervical level. The interval
between stimuli is timed such that the ascending sensory volley meets the descending motorsensory
volley at the level of the spinal cord. Ascending
and descending sensory volleys collide with termination of all sensory transmission while any descending motor volley is free to proceed to the anterior
horn cell and then into the lower extremities where
an averaged response is recorded over the tibial
nerve.
True neurogenic MEPs, as these signals might be
called, have successfully been used for motor tract
monitoring purposes (Pereon et al., 2002). Use of
epidural stimulation adds a more invasive aspect to
the monitoring and may not be available to all neuromonitoring teams. If instead a transspinal stimulation
method is applied for this purpose, the signal is likely
to be highly susceptible to the stimulus shunting problems described above in light of the small signals
produced with the collision technique and the potentially higher stimulus requirements for motor tracts
(Haghighi et al., 1994), but this has not yet been
reported.
19.5.2. Myogenic recordings after spinal
stimulation
Myogenic MEPs following spinal stimulation have
also been used as an indicator of motor tract function.
Epidural or subarachnoid stimulating electrodes are
typically used with recording made from muscles of
the lower extremities. Some use a single stimulus
for elicitation (Machida et al., 1985, 1988a,b, 1989,
1990; Adams et al., 1993; Nagle et al., 1996) and
an averaged signal may be recorded in the case of
low-amplitude signals or partial neuromuscular
blockade (Adams et al., 1993; Nagle et al., 1996).
The origin of these signals may also be open to
debate as the generated muscle responses could be produced in part by a dorsal column-mediated H-reflex
pathway. In fact, Mochida et al. (1995) reported that
with single stimulus-evoked muscle responses there
was a remarkable reduction in amplitudes following
dorsal column lesioning. In addition, single stimuli
used to elicit muscle responses may yield unreliable

results due to poor synaptic transmission at the anterior
horn cell under conditions of general anesthesia
(Machida et al., 1988a; Taylor et al., 1993; Haghighi
et al., 1994). To address these concerns, the efficiency
of motor synaptic activity and hence, the amplitude
and reliability of these myogenic signals may be
greatly improved by giving a pair (Taylor et al.,
1993) or train (Mochida et al., 1995) of stimuli with
an interstimulus interval of 12 ms. After a train of
stimuli, dorsal column lesioning has no effect on the
recorded compound motor action potential amplitude
(Mochida et al., 1995).
Further research is needed to compare the clinical
performance of myogenic MEPs elicited by spinal stimulation to those elicited by transcranial stimulation.
19.6. Recommendations and conclusions
Both the classic NMEP and its derivations remain
viable options for neuromonitoring. The classic
NMEP is primarily a test of dorsal columns and does
not reliably detect isolated motor injury. In fact,
given the uncertainty of an identifiable motor component, we like to think of classic NMEPs as Neurogenic Mixed Evoked, Potentially.
As a sensory evoked potential, the classic NMEP
is possibly more sensitive than SEPs, but is probably less specific and has a slightly higher risk of
complications which in extreme cases could be
severe. That being said, the NMEP and SEPs may
be used as complementary tests. For this purpose,
the SEP is less plagued by the occasional technical
difficulties encountered with NMEPs and may be
used to assess upper extremity and intracranial pathways in addition to lower levels. The NMEP, on the
other hand, is insensitive to anesthetic agents and
less affected by organic brain dysfunction than is
the cortical SEP.
The various types of MEPs that utilize spinal stimulation are likely to be reliable tests for motor function and are potential alternatives to TcMEP testing.
However, recommendations as to the use of transcranial versus spinally elicited motor evoked potentials
await direct comparison of these methods.
References
Adams, DC, Emerson, RG, Heyer, EJ, McCormick, PC,
Carmel, PW, Stein, BM, Farcy, JP and Gallo, EJ
(1993) Monitoring of intraoperative motor-evoked
279
potentials under conditions of controlled neuromuscular blockade. Anesth. Analg., 77(5): 913918.
Bernard, JM, Pereon, Y, Fayet, G and Guiheneuc, P (1996)
Effects of isoflurane and desflurane on neurogenic
motor- and somatosensory-evoked potential monitoring
for scoliosis surgery. Anesthesiology, 85(5): 10131019.
Burkholder, LM, Houlden, DA, Midha, R, Weiss, E and
Vennettilli, M (2003) Neurogenic motor evoked potentials: role in brachial plexus surgery. Case report.
J. Neurosurg., 98(3): 607610.
Dawson, GD (1956) The relative excitability and conduction
velocity of sensory and motor nerve fibres in man.
J. Physiol., 131: 436451.
Delecrin, J, Nguyen The Tich, S, Passuti, N and Pereon, Y
(2000) Neurogenic mixed evoked potential monitoring
during scoliosis surgery: retrospective analysis of 149
cases. Rev. Chir. Orthop. Reparatrice Appar. Mot., 86
(1): 4653.
Haghighi, SS, York, DH, Gaines, RW and Oro, JJ (1994)
Monitoring of motor tracts with spinal cord stimulation.
Spine, 19(13): 15181524.
Hays, SR and Schwengel, DA (1999) Transient hypotension
as a complication of monitoring transcervical motor
evoked potentials. Anesthesiology, 90(1): 314317.
Kai, Y, Owen, JH, Lenke, LG, Bridwell, KH, Oakley, DM
and Sugioka, Y (1993) Use of sciatic neurogenic motor
evoked potentials versus spinal potentials to predict
early-onset neurologic deficits when intervention is still
possible during overdistraction. Spine, 18(9):
11341139.
Kai, Y, Owen, JH, Allen, BT, Dobras, M and Davis, C
(1995) Relationship between evoked potentials and clinical status in spinal cord ischemia. Spine, 20(3): 291296.
Leppanen, RE (2004) Faces of spine care. From the electrodiagnostics lab. Descending neurogenic evoked
potentials. Spine J., 4(6): 713716.
Leppanen, R, Madigan, R, Sears, C, McGuire, J, Wallace, S
and Captain, J (1998) Intraoperative collision studies
demonstrate descending spinal cord stimulated evoked
potentials and ascending somatosensory evoked potentials are mediated through common pathways. Clin. Neurophysiol, 110[1999], 2265. Ref Type: Abstract.
Levy, WJ, York, DH, McCaffrey, M and Tanzer, F (1984)
Motor evoked potentials from transcranial stimulation
of the motor cortex in humans. Neurosurgery, 15(3):
287302.
Lindsay, AD and Binder, MD (1991) Distribution of effective
synaptic currents underlying recurrent inhibition in cat triceps surae motoneurons. J. Neurophysiol., 65(2): 168177.
Lubitz, SE, Keith, RW and Crawford, AH (1999) Intraoperative experience with neuromotor evoked potentials. A review of 60 consecutive cases. Spine, 24(19):
20302033.
Machida, M, Weinstein, SL, Yamada, T and Kimura, J
(1985) Spinal cord monitoring. Electrophysiological
280
measures of sensory and motor function during spinal
surgery. Spine, 10(5): 407413.
Machida, M, Weinstein, SL, Yamada, T, Kimura, J, Itagaki,
T and Usui, T (1988a) Monitoring of motor action potentials after stimulation of the spinal cord. J. Bone Joint
Surg. Am., 70(6): 911918.
Machida, M, Weinstein, SL, Yamada, T, Kimura, J and Toriyama, S (1988b) Dissociation of muscle action potentials
and spinal somatosensory evoked potentials after ischemic damage of spinal cord. Spine, 13(10): 11191124.
Machida, M, Weinstein, SL, Imamura, Y, Usui, T,
Yamada, T, Kimura, J and Toriyama, S (1989) Compound muscle action potentials and spinal evoked
potentials in experimental spine maneuver. Spine, 14
(7): 687691.
Machida, M, Yamada, T, Ross, M, Kimura, J and
Hitchon, P (1990) Effect of spinal cord ischemia on
compound muscle action potentials and spinal evoked
potentials following spinal cord stimulation in the
dog. J. Spinal Disord., 3(4): 345352.
Maltenfort, MG, Heckman, CJ and Rymer, WZ (1998)
Decorrelating actions of Renshaw interneurons on the
firing of spinal motoneurons within a motor nucleus: a
simulation study. J. Neurophysiol., 80(1): 309323.
Minahan, RE, Sepkuty, JP, Lesser, RP, Sponseller, PD and
Kostuik, JP (2001) Anterior spinal cord injury with preserved neurogenic motor evoked potentials. Clin. Neurophysiol., 112(8): 14421450.
Mochida, K, Shinomiya, K, Komori, H and Furuya, K
(1995) A new method of multisegment motor pathway
monitoring using muscle potentials after train spinal
stimulation. Spine, 20(20): 22402246.
Mochida, K, Komori, H, Okawa, A and Shinomiya, K
(1997) Evaluation of motor function during thoracic
and thoracolumbar spinal surgery based on motorevoked potentials using train spinal stimulation. Spine,
22(12): 13851393.
Mustain, WD and Kendig, RJ (1991) Dissociation of neurogenic motor and somatosensory evoked potentials. A
case report. Spine, 16(7): 851853.
Nagle, KJ, Emerson, RG, Adams, DC, Heyer, EJ, Roye, DP,
Schwab, FJ, Weidenbaum, M, McCormick, P, Pile-Spellman, J, Stein, BM, Farcy, JP, Gallo, EJ, Dowling, KC and
Turner, CA (1996) Intraoperative monitoring of motor
evoked potentials: a review of 116 cases. Neurology, 47
(4): 9991004.
Owen, JH (1990) Motor evoked potentials. In: SK Salzman
(Ed.), Neural Monitoring. The Humana Press, Clifton,
NJ, pp. 219241.
Owen, JH (1993) Intraoperative stimulation of the spinal
cord for prevention of spinal cord injury. Adv. Neurol.,
63: 271288.
Owen, JH (1999) The application of intraoperative monitoring during surgery for spinal deformity. Spine, 24
(24): 26492662.

Owen, JH, Laschinger, J, Bridwell, K, Shimon, S, Nielsen, C,
Dunlap, J and Kain, C (1988) Sensitivity and specificity of
somatosensory and neurogenic-motor evoked potentials in
animals and humans. Spine, 13(10): 11111118.
Owen, JH, Jenny, AB, Naito, M, Weber, K, Bridwell, KH
and McGhee, R (1989) Effects of spinal cord lesioning
on somatosensory and neurogenic-motor evoked potentials. Spine, 14(7): 673682.
Owen, JH, Naito, M and Bridwell, KH (1990a)
Relationship among level of distraction, evoked potentials, spinal cord ischemia and integrity, and clinical status in animals. Spine, 15(9): 852857.
Owen, JH, Naito, M, Bridwell, KH and Oakley, DM
(1990b) Relationship between duration of spinal cord
ischemia and postoperative neurologic deficits in animals. Spine, 15(7): 618622.
Owen, JH, Bridwell, KH, Grubb, R, Jenny, A, Allen, B, Padberg, AM and Shimon, SM (1991) The clinical application
of neurogenic motor evoked potentials to monitor spinal
cord function during surgery. Spine, 16(8 Suppl.):
S385S390.
Padberg, AM, Wilson-Holden, TJ, Lenke, LG and Bridwell, KH (1998) Somatosensory- and motor-evoked
potential monitoring without a wake-up test during idiopathic scoliosis surgery. An accepted standard of care
[see comments]. Spine, 23(12): 13921400.
Pereon, Y, Bernard, JM, Delecrin, J and Passuti, N (1995)
Could neurogenic motor evoked potentials be used to
monitor motor and somatosensory pathways during scoliosis surgery? [letter; comment]. Muscle Nerve, 18(10):
12141215.
Pereon, Y, Bernard, JM, Fayet, G, Delecrin, J, Passuti, N
and Guiheneuc, P (1998) Usefulness of neurogenic
motor evoked potentials for spinal cord monitoring:
findings in 112 consecutive patients undergoing surgery
for spinal deformity. Electroencephalogr. Clin. Neurophysiol., 108(1): 1723.
Pereon, Y, Bernard, JM, Nguyen, ST, Genet, R, Petitfaux,
F and Guiheneuc, P (1999a) The effects of desflurane
on the nervous system: from spinal cord to muscles.
Anesth. Analg., 89(2): 490495.
Pereon, Y, Delecrin, J, Nguyeni The Tich, SN, BertrandVasseur, A and Passuti, N (1999b) Successful monitoring of neurogenic mixed evoked potentials elicited by
anterior spinal cord stimulation through thoracoscopy
during spine surgery. Spine, 24(19): 20259.
Pereon, Y, Nguyen The Tich, Y, Delecrin, J, Pham, DC,
Bodin, J, Drouet, JC and Passuti, N (2002) Combined
spinal cord monitoring using neurogenic mixed evoked
potentials and collision techniques. Spine, 27(14):
15711576.
Phillips, LH, Blanco, JS and Sussman, MD (1995) Direct
spinal stimulation for intraoperative monitoring during
scoliosis surgery [see comments]. Muscle Nerve, 18
(3): 319325.

Rose, RD (1998) Removing the antidromically driven sensory component from cervically evoked motor potentials.
Med. Hypotheses, 50(2): 147154.
Schwartz, DM, Drummond, DS and Ecker, ML (1996)
Influence of rigid spinal instrumentation on the neurogenic motor evoked potential. J. Spinal Disord., 9(5):
439445.
Schwentker, MC, Russell, GB, Rodichok, LD, Segal, LS,
Schwentker, EP and Blackburn, TW (1995) Myogenic
response distortion of neurogenic motor evoked potential morphology. Anesthesiology, 83(3): 616619.
Su, CF, Haghighi, SS, Oro, JJ and Gaines, RW (1992)
Backfiring in spinal cord monitoring. High thoracic
spinal cord stimulation evokes sciatic response by antidromic sensory pathway conduction, not motor tract
conduction. Spine, 17(5): 504508.
Taylor, BA, Fennelly, ME, Taylor, A and Farrell, J (1993)
Temporal summation the key to motor evoked potential spinal cord monitoring in humans. J. Neurol. Neurosurg. Psychiatry, 56(1): 104106.
Toleikis, JR, Skelly, JP, Carlvin, AO and Burkus, JK (2000)
Spinally elicited peripheral nerve responses are sensory
rather than motor. Clin. Neurophysiol., 111(4): 736742.
281
Van, KL (1981) Autogenetic recurrent inhibition of individual spinal motoneurones of the cat. Neurosci. Lett.,
21(3): 297300.
Wilson-Holden, TJ, Padberg, AM, Lenke, LG, Larson, BJ,
Bridwell, KH and Bassett, GS (1999) Efficacy of intraoperative monitoring for pediatric patients with spinal
cord pathology undergoing spinal deformity surgery.
Spine, 24(16): 16851692.
Wilson-Holden, TJ, Padberg, AM, Parkinson, JD, Bridwell,
KH, Lenke, LG and Bassett, GS (2000) A prospective
comparison of neurogenic mixed evoked potential stimulation methods: utility of epidural elicitation during
posterior spinal surgery. Spine, 25(18): 23642371.
Wilson-Holden, TJ, Van Sickle, D and Lenke, LG (2002) The
benefit of neurogenic mixed evoked potentials for intraoperative spinal cord monitoring during correction of
severe scoliosis: a case study. Spine, 27(10): E258E265.
Zhou, HH, Mehta, M and Leis, AA (1997) Spinal cord
motoneuron excitability during isoflurane and nitrous
oxide anesthesia. Anesthesiology, 86(2): 302307.

M.R. Nuwer (Ed.)
282
CHAPTER 20
Magnetic cortical stimulation techniques

Laverne D. Guginoa,*, Linda S. Aglioa, Harvey L. Edmondsb,1 and
Andres A. Gonzalezc,2
a
Department of Anesthesiology, Perioperative and Pain Medicine, Brigham and Womens Hospital, Harvard Medical School,
Boston, MA 02115, USA
b
Surgical Monitoring Associates Inc., Louisville, KY 40207, USA
Department of Neurology, Keck School of Medicine, University of Southern California, Los Angeles, CA 90089, USA
20.1. Introduction and historical perspectives

In 1985, Barker and colleagues introduced a practical
noninvasive stimulation technique for harnessing
pulsed magnetic fields (Barker et al., 1985). Since that
time, research has focused on both understanding the
effects of this form of stimulation on the central nervous
system, as well as the development of new clinical
applications for this technique (Barker et al., 1986).
This chapter will review the use of transcranial magnetic stimulation (TCMS) for monitoring the functional
integrity of the descending motor systems during surgery, as well as a brief discussion of a potential role in
the pre- and postoperative period in conscious patients.
Intraoperative use of somatosensory evoked potentials (SEPs) for monitoring central nervous system function has been a popular technique for over 25 years
(Gugino and Chabot, 1990). A number of studies have
shown that monitoring SEPs results in acceptable levels
of postoperative outcome prediction, as well as possibly
improving surgical outcomes (Brown et al., 1984; John
et al., 1988; Nuwer et al., 1995; McCaffrey, 1997; York,
1997). In spite of these studies, the appearance of several case reports describing new postoperative motor
deficits with unchanged intraoperative SEP responses
*
Correspondence to: Laverne D. Gugino, M.D., Ph.D., Department of Anesthesiology, Brigham and Womens Hospital,
Harvard University, 75 Francis St., Boston, MA 02115, USA.
Tel.: 1-617-732-8222; fax: 1-617-277-2192.
E-mail: vern@zeus.bwh.harvard.edu (L.D. Gugino).
1
Present Address: Department of Anesthesia and Perioperative Medicine, University of Louisville School of Medicine,
Louisville, KY 40207-3633, USA.
2
Present Address: Healthcare Consultation Center II, 1520
San Pablo St., Suite 3000, Los Angeles, California 900334606.
served as the stimulus for finding a direct motor system

monitor (Ginsburg et al., 1985; Chatrian et al., 1988;
Crawford et al., 1988; Zornow et al., 1990). The false
negatives presumably occur because SEP monitoring,
particularly during spinal cord procedures, assesses the
dorsal column pathway, which is anatomically
restricted to the posterior columns (Barker et al.,
1985). Descending motor paths, however, are located
in the lateral and anterior spinal cord white columns
(McCaffrey, 1997). Therefore, for SEPs to adequately
predict motor deficits, the spinal cord injury must affect
both lateral and anterior as well as posterior spinal
regions. The false negative rate can be expected to correlate with the frequency that surgically induced injury
is restricted to the anterior and/or lateral spinal cord
regions (Barker et al., 1985).
Two stimulation approaches have been developed
for selectively exciting descending motor pathways.
They are transcranial electrical stimulation (TCES)
and TCMS. Both excite corticospinal tract (CST) neurons of origin using a transcranial technique. Levy and
colleagues were the first to show that TCES could be
used for eliciting averaged spinal cord and peripheral
nerve responses (Levy and York, 1983; Levy et al.,
1984; Levy, 1987). Their stimulation protocol
involved a comb-like scalp electrode consisting of
numerous gold plated electrodes as the anode with a
single cathode fixed to the roof of the mouth (Levy
et al., 1984). Zentner used a single pulse (i.e., nonaveraging technique) with the anode placed at Cz and the
cathode 6 cm anterior along the scalp midline (Zentner
et al., 1988; Zentner, 1989a; Zentner and Rieder,
1990). These two groups of investigators generated
significant interest in the use of transcranial stimulation for monitoring descending motor systems during
spinal cord surgery.
283
The use of magnetic stimulation for exciting

central nervous system structures is a more recent
advance (Barker et al., 1985). Although Michael
Faraday described the phenomenon of electromagnetic induction in the early 1800s, it remained for
Barker and colleagues, in 1985, to rekindle interest
in magnetic stimulation. They demonstrated the
ability of pulsed magnetic fields to produce transcranial motor cortex stimulation with acquisition
of myogenic responses in a relatively painless fashion (Cushing, 1909; Barker et al., 1985; Gugino
et al., 1998). Since Barkers work, numerous
refinements in equipment have led to studies
exploring the many potential uses for TCMS in
humans. Dr. Edmonds and his group are credited
with being the first to demonstrate spinal cord
motor function monitoring using TCMS in anesthetized patients (Edmonds et al., 1989; Glassman
et al., 1996). This chapter begins by reviewing
the basic principles underlying the use of TCMS
for exciting the central nervous system. This is followed by a brief discussion of the anatomy, physiology, and anesthetic considerations important to
the use of TCMS for monitoring descending motor
system function. After establishing this foundation,
we will review the clinical experience using TCMS
for monitoring the functional integrity of the CST

anterior horn cell pathway during spinal cord surgery.
20.2. A primer on transcranial magnetic
stimulators
The magnetic stimulator consists of a current source
and a stimulating coil (Cadwell, 1991). The current
source (Fig. 1) produces, stores, and releases on
demand a large current pulse which flows through
the stimulating coil. Current sources typically
include: a power supply, large capacities for storing
the electrical energy produced by the power supply,
and a high current capacity switch for transferring
the stored electrical energy to the stimulating coil
(Cadwell, 1991) (Fig. 2).
The stimulating coil is typically constructed of
copper wire turns, which are insulated with a durable
material protecting both the patient and operator.
When a large current pulse is passed through the coil,
a magnetic field pulse is produced which surrounds
the coil windings (Barker et al., 1985; Cadwell,
1991). If the insulated coil is centered on the scalp
vertex and parallel to the ground, the induced magnetic field pulse will pass through the scalp, skull,
and cortex. It can be shown that magnetic field
TYPICAL MAGNETIC STIMULATOR BLOCK DIAGRAM

Magnetic Pulse Generation
Energy storage capacitor charged to a

high voltage
110240 V
400 W
Input
10 MW
1 ms
output
Energy storage capacitor is discharged

in less than 1ms
High current discharge pulse
channeled into a coil
Coil current produces a brief
magnetic pulse
Magnetic pulse produces eddy
currents in the tissue
If of sufficient value eddy currents
stimulate nerves
High Voltage
Charging
Circuitry
Energy
Storage
Capacitor
High Power
Discharge
Switch
Control and Safety Electronics
Fig. 1. Illustration of major components of a magnetic stimulation current source (reprinted with permission from Magstim,
Ltd., Wales UK (1994).
284
L.D. GUGINO ET AL.

COIL CURRENT AND MAGNETIC FIELD
Stimulating Coil Winding

Typically 520 Copper Turns
Magnetic Field
15 Tesla Peak
Discharge Current
~8,000 Amps Peak
Fig. 2. Diagram of the round magnetic stimulating coil. The right hand rule determines direction of the induced magnetic
field (reprinted with permission from Magstim Ltd., Wales UK, 1994).
intensity decreases as the inverse square or cube of

the distance from the plane of the coil, depending
on whether the field is measured at a distance less
than or greater than 1 cm diameter beneath the coil
windings (Jackson, 1963). Importantly, the attenuation is independent of the scalp or skull electrical
properties (Jackson, 1963; Jalinous, 1991).
When the magnetic pulse passes through a tissue capable of producing a current, an electrical field is produced
in that tissue which, as a first approximation, is parallel to
the coil windings (Jackson, 1963). [N.B. Roth et al. and
investigators demonstrated that the electric field actually
bends close to the inner skull due to static charge repulsion (Roth et al., 1991a,b).] The electric field induces a
current whose intracranial path follows the electric field
and is, therefore, parallel to the current pulse within the
coil (Barker et al., 1985). The intensity of the induced current within the tissue is proportional to the induced electric field intensity and tissue conductance (Claus, 1991).
Interestingly, the induced tissue current direction is opposite to that within the coil (Cadwell, 1991).
Examples of stimulating coil geometries available
are shown in Fig. 3. Note that the early coil geometries were of three basic shapes: (1) a tear-drop coil,
(2) a round coil, and (3) a figure eight coil. Fig. 4
shows the measured and calculated magnetic field
strength as a function of distance from the outer coil
A
C
D
F
E
Fig. 3. Radiographs of commercially available magnetic

stimulating coils used for studies on the induced magnetic
and electrical fields discussed in the text. Starting with
the left side of the upper row, are two tightly wound teardrop shaped coils (B) and (C) having diameters of 9 and
5 cm, respectively. The tightly wound round coil (A) has
a diameter of 9 cm. In the lower row of coils, starting at
the left are two spiral coils having diameters of 14 and
67 cm, respectively (D) and (E). Finally, the lower right
coil is the figure eight coil (butterfly coil, or twin oval
coil), with each component wing or oval having a diameter
of 4 cm (F). (Reprinted from Cohen et al., 1990 with permission from the International Federation of Clinical
Neurophysiology.)
Magnetic field (normalized)

1.0
Coil A
Corr .98
0.8
1.0
Coil B
Corr .989
0.8
1.0
0.6
0.6
0.4
0.4
0.4
0.2
0.2
0.2
0.0
0.0
0.0
0.2
0.2
0.2
0.4
0.4
2
4
6
Distance (cm)
Coil D
Corr .99
1.0
0.8
0.4
0
2
4
6
Distance (cm)
1.0
Coil E
Corr .977
0.8
1.0
0.6
0.4
0.4
0.4
0.2
0.2
0.2
0.0
0.0
0.2
0.2
0.2
0.4
2
4
6
Distance (cm)
0.6
0.0
2
4
6
Distance (cm)
Coil F
Corr .99
0.8
0.6
0.4
Coil C
Corr .999
0.8
0.6
Magnetic field (normalized)
285
0.4
0
2
4
6
Distance (cm)
2
4
6
Distance (cm)
Calculated
measured
Fig. 4. Plots of the induced magnetic field intensity as a function of distance from the center of each coil (as shown in
Fig. 3) along the coils diameter. The vertical axis represents the calculated (solid lines) and experimentally measured
derived field intensities (broken lines) as a function of location along each coils diameter. The vertical intensity axis is normalized to the maximum obtained magnetic field intensity. Note the excellent agreement between the values determined
from the mathematically modeled calculations and experimentally derived values. (Reprinted from Cohen, 1990 with permission from the International Federation of Clinical Neurophysiology.)
edge toward the center of each coil (Cohen et al., 1990;

Cohen and Cuffin, 1991). Note that the tear-drop and
round coil geometries (coils A through E in Fig. 4)
both generate the most intense magnetic field beneath
the coil windings. As the distance toward the center of
each coil increases, the magnetic field decreases
toward zero. This is not true for the figure eight coil
(coil F in Fig. 4). In this case, the current is directed
through the outer wings of the coil such that the current
flows in the same direction where the coils come in
juxtaposition (i.e., center of the coil). This has the
effect of doubling the current pulse flowing in the center of the figure eight coil. The resulting induced magnetic field also peaks beneath the coils center (Cohen
et al., 1990; Cohen and Cuffin, 1991).
If one moves a voltage sensing probe in a plane parallel to the bottom surface of each of the coils shown in
Fig. 3, the induced voltage distribution as a function of
loci 1 cm below each coil can be determined (Cohen

et al., 1990). The voltage distributions are shown using
pseudo-three-dimensional plots in Fig. 5. Note that, on
the one hand, the induced voltage peak beneath the
coils winding decreases in strength toward the center
of the round and tear-drop geometries. On the other
hand, a voltage peak is generated beneath the center
of the figure eight. This phenomenon is responsible
for the focal transcranial excitation of the cortex, a
characteristic feature of this coil, which is useful for
mapping functional cortical regions (Cohen et al.,
1990; Cadwell, 1991; Claus, 1991; Cohen and Cuffin,
1991; Roth et al., 1991; Yanokuchi and Cohen, 1991;
Amassian et al., 1992; Maccabee et al., 1993).
Figure 6 shows the calculated induced cortical
current eddies for the round coil placed flat on the scalp
vertex. Fig. 7 shows the same for the figure eight placed
flat on the scalp vertex (Cohen and Cuffin, 1991).
286
L.D. GUGINO ET AL.
800
800
Coil A
Coil B
E(x,y)
(V/m)
E(x,y)
(V/m)
400
400
8 8
0
x (cm)
800
y (cm) 0
8
8
0 y (cm) 0
8
0
x (cm)
8
0 y (cm)
8 8
Coil F
8
0
8
8
y (cm)
E(x,y)
(V/m)
400
400
0
x (cm)
800
Coil E
E(x,y)
(V/m)
400
0
8
0
x (cm)
800
Coil D
E(x,y)
(V/m)
Coil C
E(x,y)
(V/m)
400
8
0
8
800
0
x (cm)
8 8
y (cm)
8
0
8
0
x (cm)
8 8
y (cm)
Fig. 5. Plots of the induced electrical field intensity distribution in an infinite homogeneous plane 1 cm below the surface of
coils (A) through (F) as shown in Fig. 3. The electrical field intensities within the defined plane are based on calculated
values derived from a mathematical model. The vertical axis represents the calculated electrical field intensity (in V/m)
as a function of location within the x, y plane beneath each coil. The dimensions of the x and y coordinates of each plane
are given in centimeters. It should be noted that the greatest electrical field intensity occurs beneath the windings of coils
(A) through (E). The electrical field intensity demonstrates an attenuation toward the center and beyond the windings of
coils (A) through (E). Tightly wound coils produce more intense electrical fields (AC) than the spiral wound coils (D)
and (E). Coil (F) (i.e., the butterfly coil) produces the most intense electrical field beneath the junction of the two component coil wings, as well as an induced electrical field with the greatest focality. (Reprinted from Cohen, 1990 with permission from the International Federation of Clinical Neurophysiology.)
In B of each figure, the length of the current vectors

represents the current intensity with the arrow
showing the current direction within the cortex parallel to the coil windings. The figure eight coil, in line
with the focal voltage peak, induces a focal area of
intense current density beneath the body of the coil.
Fig. 8 is a summary diagram comparing the time
course of each event beginning with a current pulse
passing through the coil to the induced stimulating
electric charge within the cortex. Note that the time
course of the time rate of change of the magnetic
field pulse best predicts the induced electric field
time course. The induced electric field is a major
determinant of the induced stimulating current
(Gugino et al., 1998).
Figure 9 shows the expected induced electric field
attenuation within a reference frame as a function of
increasing separation distance of the reference from
beneath the plane of the figure eight coil (Cohen

et al., 1990). Note the marked attenuation of the
induced electric field as the reference plane distance
beneath the coil increases from 1 to 4 cm (Cohen
and Cuffin, 1991), as mentioned previously. This
attenuation, as discussed, is a consequence of the
induced magnetic fields which attenuate as a function
of the inverse square to inverse cube of the coil
radius, depending if the reference plane separation
distance is less or greater than the diameter of the stimulating coil (Jackson, 1963). These results suggest
that the early stimulating coils with diameters of 7
8 cm induced, at best, weak intracortical electric
fields that had their maximum stimulating effect in
the superficial cortical layers. A corollary of this
property of magnetic stimulating coils is that the
greater the diameter of the stimulating coil, the more
intense the induced electric field deep within the
287
COIL
SCALP
SKULL
BRAIN
B
B
Fig. 6. The calculated current eddies induced by TCMS
using a round coil placed over three concentric spheres in
which the outer layer represents the scalp, the middle layer
of the skull with the inner spherical volume representing
cortical tissue. Each volume is characterized as homogeneous, having similar conductivity for each layer of the
head and cortex. A: The current eddies of each layer are
depicted in which the magnitude and direction of the
induced current densities are represented quantitatively by
the width of each current loop. B: The calculated induced
current density distribution in a plane 2 cm below the surface of the sphere is depicted. The magnitude and direction
are represented by the length and direction of each arrow,
respectively. It should be noted that the current density at
the center of the coil in (B) (represented by the circle) is
zero as predicted from the electrical field distributions for
coil (A) as shown in Fig. 5. (Reprinted with permission
from Cohen and Cuffin, 1991, Clin. Neurophysiol.)
cortex (Jalinous, 1991; Cohen and Cuffin, 1991).

This latter phenomenon is important as the two most
useful stimulating coils for monitoring have diameters almost twice that of the early stimulating coils
(Beradelli et al., 1991; Kraus et al., 1992; Aglio
et al., 2002a).
Magnetic stimulation differs from electric stimulation in a number of ways. Magnetic stimulators generate coil heat and have a characteristic noise
Fig. 7. Diagram of the calculated TCMS-induced current

density distribution for the double circular coil placed with
a parallel orientation above the spherical model. A: and B:
Diagrams are constructed using the same format as shown
in Fig. 6. The double circular coil is similar to coil (F) as
shown in Fig. 3, which has a double oval or D-shaped, winding construction. The double circular coil configuration
induces a current density distribution characterized as having
the greatest focality of all the coils that are shown in Fig. 3.
(Reprinted with permission from Cohen and Cuffin, 1991,
J. Clin. Neurophysiol.)
(Cadwell, 1991). Coil heating is due to the enormous

current pulses (up to 8 A) that pass through the coil
in short periods of time (Cadwell, 1991). Most coils
have imbedded thermistors that shut the stimulators
down when a threshold coil temperature is reached
(Gugino et al., 1998). This may limit the time period
a transcranial stimulator can be used. In general,
stimuli separated by 30 s to 1 min usually do not
lead to coil overheating and subsequent automatic
shutdown (Beradelli et al., 1991). The noise caused
by magnetic stimulation is due to a mechanical pressure increase in the center of the coil when a current
pulse passes through. This pressure is due to the magnetic field repulsion on opposite sides of the coil that
tends to cause a small degree of coil enlargement.
When the magnetic field subsides, after each stimulation, elastic properties of the coil cause it to snap
288
L.D. GUGINO ET AL.

FROM COIL CURRENT TO INDUCED CHARGE
~8,000 Amps
~300V/m
Induced Electric
Field
Stimulating Coil
Current
0.5
0.5
~2 Tesla
~20mA/cm2
Induced Tissue
Current
Magnetic Field
Pulse
0.5
~30 kT/s
1.0
1.0
0.5
1.0
1.0
~1C/cm3
Rate of change of
Magnetic Field
0.5
1.0
Time (ms)
Induced Charge
Density
0.5
1.0
Time (ms)
Fig. 8. Diagram of the time course for the sequence of events occurring from the passage of a current pulse through the coil
to the induction of a cortical stimulating current eddy. The horizontal axis for each plot represents time in milliseconds. The
vertical axis is left dimensionless. The lower left plot represents the time rate of change of the induced magnetic field,
which has major significance with respect to transcranial stimulation efficacy. Data for the figure were compiled from several plots. Reprinted with permission from Magstim Ltd., Wales UK (1994).
back to its initial dimensions. This latter effect causes

the characteristic noise heard during magnetic stimulation (Cadwell, 1991).
Three types of coil current pulses have been used
for generating magnetic stimuli. They include (1) a
monophasic, (2) a biphasic, and (3) a polyphasic
pulse. Fig. 10 shows the advantages and disadvantages for each current pulse shape (Gugino et al.,
1998). Transcranial magnetic stimulators for intraoperative monitoring have made use of the monophasic and polyphasic pulses. As will be discussed in
Section 20.3, current direction of flow within the cortex will determine which side of the brain is most
effectively stimulated (Beradelli et al., 1991). Polyphasic coil current pulses stimulate both motor cortices during a single stimulus. Monophasic pulses
required a device for changing the coil current
polarity in order to stimulate both motor cortices,
sequentially.
20.3. Anatomic and physiological considerations
for motor system excitation
A predominant descending motor path originating in
the motor cortex which is excited by TCMS is the
CST (Boyd et al., 1986; Day et al., 1986; 1988;

Burke et al., 1990; Edgley et al., 1990; Rothwell et al.,
1991; Fujiki et al., 1996). Amassian studied the activation of the cells of origin of the CST using cortical surface anodal electrical stimulation (Amassian et al.,
1987b,c, 1991). Recording from the CST at the
medullary pyramids and spinal cord demonstrated a
train of descending waves (Boyd et al., 1986; Burke
et al., 1990; Edgley et al., 1990; Deletis, 1993; Fujiki
et al., 1996). The first wave in the train, called the
D wave, was shown to be elicited by direct activation of either the axon hillock region or the first node
of Ranvier of CST cells (Amassian et al., 1987b,c).
The waves following the D wave were referred to
as I waves, as they were believed to be generated
by recurrent cortical interneuronal excitation of
CST neurons (Amassian et al., 1987c). Similar descending CST activity has been recorded from the
human spinal cord epidural space with TCES and
TCMS (Boyd et al., 1986; Amassian and Cracco,
1987; Deletis, 1993; Fujiki et al., 1996; Burke and
Hicks, 1998).
The neuronal cell bodies of the CST axons are
located in widespread areas of the cortex including
premotor, motor, and somatosensory cortices
289
Z = 1 cm
E (x,y) (V/m)
800
Z = 2 cm
800
400
400
8
0
8
8 8
8
0
8
Z = 3 cm
800
E (x,y) (V/m)
8 8
Z = 4 cm
800
400
400
8
0
8
0
x (cm)
8 8
0
m)
y (c
8
0
8
0
x (cm)
8 8
0
m)
y (c
Fig. 9. Magnetic and induced electrical fields attenuate with axial distance (i.e., a line normal to the coils plane) from
magnetic coils. The decrease in the induced electrical field intensity distribution with distance is shown for the butterfly
coil ((F) as shown in Fig. 3). The z value, in centimeters, represents the distance of each plane below the flat surface of
the butterfly coil for which an induced electrical field distribution was calculated. It should be noted that with increasing
distance from the coil, there is a decrease in both the intensity of the central peak field and the focality of the induced field.
(Reprinted from Cohen et al. (1990) with permission from Elsevier Science.)
Magnetic
Field Strength
Rise Time
Monophasic
For: More accurate than biphasic,
lower noise,lowerheat
Resistance
Heating
in the coil
Time
Against: Not easy to obtain

bilateral cortical responses
Biphasic
For: Short Pulse, suited to bilateral
cortical stimulation
Period (1/Freq.)
Time
Against: Higher noise and heat,

less accurate than monophasic
Increased discharge
Click Noise
Time
Polyphasic
For: Suited to bilateral cortical
stimulation
Against: Highest noise and heat;
less accurate than monophasic
Fig. 10. Time course of the magnetic stimulating coils current pulse for three different types of current sources. The
advantages of each type are listed to the right of the figure. (Reprinted (1994) permission from Magstim, Ltd., Wales, UK.)
290
(Lawrence and Kuypers, 1968: Lawrence and Hopkins,

1972). TCMS or TCES can be expected to excite
many of these CST neurons. Lawrence, however,
has demonstrated using selective lesions of these cortical areas that the spinal cord terminals of CST cells
originating in premotor and motor cortex tend to pass
anteriorally toward spinal cord anterior horn cell columns within the spinal cord gray area (Lawrence and
Kuypers, 1968; Gilman and Marco, 1971). Brouwer
and Ashby have emphasized that characteristic of
the CST descending cortical motor path is the projection of the cortical neurons to spinal anterior horn
cells in the spinal cord gray matter (Brouwer and
Ashby, 1990). In primates, the CST axons have a
strong monosynaptic input to contralateral anterior
horn cells supplying the distal small muscles of the
upper and lower limbs. Brouwer and Ashby experimentally showed that TCMS activates the same descending pathway (i.e., CST path) in humans based on
the relative amplitudes of several upper and lower
limb muscle responses (Brouwer and Ashby, 1990).
In addition, they showed that TCMS activates the
larger diameter CST neurons and these neurons, in
turn, project in a monosynaptic fashion to the same
muscles in the baboon and monkey using cortical
surface anodal electrical stimulation. Phillips and
colleagues, using intracellular recordings, demonstrated that a train of CST pulses (as occurs during
cortical anodal surface stimulation) leads to a significant posttetanic potentiation of anterior horn cell
postsynaptic excitatory potentials (Phillips, 1969).
This property of the CST synaptic input onto anterior
horn cells led to more efficacious excitation of these
cells when compared to Group 1a (i.e., stretch reflex
sensory input) monosynaptic train input onto the
same cells.
Initial studies concerning the mechanism of
TCMS-induced excitation of CST neurons suggested
that it differed from the mechanism of TCES-induced
excitation (Day et al., 1986, 1988; Roth et al., 1991;
Rothwell et al., 1991). Rothwell and Day, using single fiber recordings, independently demonstrated that
CST excitation by TCMS was probably mediated
through cortical interneuronal excitation of CST neurons (see Fig. 11) (Day et al., 1987a,b, 1989;
Rothwell, 1991). TCMS at all stimulus intensities
produced single muscle fiber excitation at a longer
latency than seen using TCES. Investigators reasoned
that, because TCMS-induced current loops were parallel to the inner cranial surface they would likewise
parallel the surface of the cortical gyri caps (see
L.D. GUGINO ET AL.

P0
P 1P 2P 3
Anodal 55%
Clockwise
magnetic 35%
Anticlockwise
magnetic 35%
10
15
20
25
30
35
Time after stimulus (ms)
40
Fig. 11. Poststimulus histograms showing the timing of

responses from recordings taken from a first dorsal interosseus motor unit driven either by an anodal TCES (at 55% of
full strength) at the vertex (top plot) or by TCMS applied
with a vertex-centered round magnetic coil (middle and
lower plots). The middle plot was obtained using a clockwise
coil current at 35% of full power, whereas, the lowest plot
represents the response latencies acquired with a counterclockwise coil current. The horizontal axis of each histogram
represents time in milliseconds after 100 of the 3 types of stimuli were applied to the scalp. The vertical axis represents
the number of responses that responded with a latency as
shown along the horizontal time axis. It should be noted that
anodal TCES produced the shortest latency responses when
compared with both TCMS stimuli. Data of this type were
used to hypothesize that TCMS activates corticospinal tract
(CST) neurons indirectly (i.e., through interneuronal synaptic excitation). (Reprinted with permission from Day et al.,
1989 with permission from Blackwell Publishing.)
Section 20.2). The current loops would, therefore,

be oriented along the axis of cortical interneurons,
an orientation appropriate for cortical interneuron
excitation. The induced current loops, however,
would be orthogonal to CST neurons which are radially oriented within the cortical gyral caps. This current loop orientation is much less efficient for direct
stimulation of CST cortical neurons. Day centered
the round magnetic stimulating coil on the vertex
(Day et al., 1987b, 1989). Amassian later demonstrated that direct TCMS activation of CST neurons
was possible with appropriate tilting of the stimulating coil on the scalp (Amassian et al., 1987b,c,
1991, 1992). The change in coil orientation on the
scalp presumably led to induced current loops which
paralleled the orientation of CST neurons (i.e., the
neuronal axons) within the cortex. Jalinous has
shown that large diameter stimulating coils produce
electric fields of greater intensity as a function of
cortical depth than smaller coils (Jalinous, 1991)
(see Section 20.2). A scalp vertex orientation of a
large round coil should, theoretically, cause direct
excitation of CST cells located within the anterior
bank of the central sulcus. At this location, these
cells are oriented in a radial direction to both the cortical surface and induced current loops.
Amassian and colleagues (Amassian et al.,
1987c) studied the site of action potential initiation
in monkey cortical CST neurons using surface
anodal electrical stimulation. They recorded the
CST epidural responses at the pyramids and lateral
columns of the spinal cord. By studying the occurrence and latencies of both D and I waves, they
believed that cortical surface anodal electrical
stimulation initiated CST action potentials at the
first or second node of Ranvier. Cathodal stimulation was believed to initiate CST I waves through
synaptic activation by excitatory cortical interneurons because appropriate tilting of a circular coil on
the scalp could lead to direct excitation of motor
cortex CST neurons. Amassian (Amassian et al.,
1990, 1992) reasoned that TCMS caused CST activation by exciting axons within the subcortical
white matter. Amassian used a second approach
for studying the location along axons where action
potentials were initiated (Amassian et al., 1992). In
these studies, a long peripheral nerve was
immersed in a saline-filled plastic model of the
human skull. Figure eight and circular coils were
placed along the outside of the plastic model
(Fig. 12). The peripheral nerve was arranged
within the skull with a bend between the site of
magnetic stimulation and the recording electrodes
placed distal to the site of stimulation. Bends in
the nerve were placed in order to mimic the upper
291
ANT.
89
ELECTRICAL
POST.
90
ELECTRICAL
+
90
90
2 ms
Fig. 12. Experimental protocol for determining the TCMS

site of induced excitation of a phrenic nerve immersed in an
inverted saline-filled plastic model of the skull. The round
magnetic coil is used for eliciting an action potential at a site
along the phrenic nerve. This site is determined by comparing
the latencies of TCMS-acquired action potentials (recorded
in air after the nerve emerges from the saline-conducting
media with the latencies of direct electrical excitation in
which the cathode is moved along the nerve until the latencies
of both stimulation modalities match. Electrical stimulation
initiates action potentials beneath the cathode. It should be
noted that the skull is inverted so that the mediolateral axis
is oriented upwards, the opposite direction to that seen in
erect volunteers. The site of the coils minimal field gradient
should represent approximately the virtual cathode (i.e.,
site for action potential initiation). Thus, a virtual anodeto-cathode orientation for the round magnetic coil can be
predicted as a function of the polarity of the monophasic current pulse fed through the coil. The upper row represents three
trials in which the TCMS-induced virtual cathode is closest
to the recording electrode (left), the virtual anode is closest
to the recording electrode (middle), and the coil is placed
symmetrically, parallel to the vertex (right). The middle
row shows the responses to TCMS stimulation (at the percentage of full power indicated to the left of each trace) for
the trial indicated in the diagrams above. The bottom row
shows the responses obtained with the recording electrode
using electrical stimulation with a cathode at the bend in
the nerve within the skull (left lower trace) and at two locations further distal to the recording electrode and bend (middle lower trace). The rightmost TCMS-induced trials in the
second and third rows show the absence of TCMS initiated
action potentials using either coil current pulse polarity when
the round coil is symmetrically tangent to the vertex. There is
a latency match between direct electrical and magnetic stimulation when the electrical stimulating cathode is placed at
the bend in the nerve (left response column). This suggests
that TCMS excites intracranial axons at the sites where the
axon changes direction as opposed to the site along an axon
where the maximum negative spatial derivative of the
induced electrical field exists. This figure represents a model
for TCMS excitation of CST neurons projecting to the lower
limbs anterior horn cells spatial cord gray matter. (Reprinted
from Amassian et al., 1992 with permission from the International Federation of Clinical Neurophysiology.)
292
limb and lower limb CST axonal trajectories after

they emerged from the cortex en route to the internal capsule. The results of this study suggest that
action potential initiation occurred at bends in the
peripheral nerve within the skull model as opposed
to the maximum negative spatial derivative of the
magnetically induced electrical field. Amassian
reasoned that magnetically induced current flowing
within the axons would tend to exit at the axonal
bend, leading to a situation favorable for axonal
excitation (Amassian et al., 1991).
Epstein and his group (Epstein et al., 1990) used
two figure eight coils of different sizes for studying
the location of TCMS-induced direct excitation of
CST neurons in humans. A small magnetic coil
produces a more intense magnetic field close to
the coil surface than is the case with a larger coil
(see Section 20.2). The magnetic field attenuates
faster with distance, however, from the smaller
coil. Because CST neurons have the same threshold for activation, the locus for CST action potential initiation is the same for both types of coil.
The magnetic coil current intensities, however,
are greater for threshold impulse initiation when
using the smaller coil. Epstein determined the
depth within the cortex, where CST impulse initiation occurred by knowing the degree of the induced
electric field attenuation with distance and the
threshold for producing a thumb twitch. Their
results were consistent with CST activation deep
within the cortex located at the junction of the cortex and underlying white matter. They hypothesized that impulse initiation occurred near the
axon hillock region or the CST basal dendrites.
Recent studies comparing the latencies of the CST
epidural responses along the spinal cord in humans
have shown that both TCMS and TCES cause direct
activation of CST neurons (Boyd et al., 1986; Edgley
et al., 1990; Hicks et al., 1993; Burke et al., 2000).
As the intensity of TCMS was increased, the latency
of the CST epidural D response decreased, but was
still consistent with activation within the deep layer
of the cortex (Burke et al., 2000). In addition, I
waves appeared when the TCMS intensity was
increased (Fig. 13) (Burke et al., 2000). Increasing
TCES intensity produced an abrupt decrease in the
spinal cord epidural D-wave latency, which was
shown to be consistent with activation of CST axons
at a midbrain level (Burke et al., 2000). A shift in
CST axon location of impulse initiation with increasing TCES intensities could explain the earlier results
L.D. GUGINO ET AL.
Magnetic
stimulation
20 v
D
Anodal
stimulation
Cathodal
stimulation
5 ms
I
Anodal/GOI
Fig. 13. Examples of CST responses acquired from the

human epidural space at an upper cervical location. The
uppermost response was TCMS-induced at 100% of full
power. The arrow points to the CST D wavelet. The second
and third responses were induced by scalp surface anodal
and cathodal TCES, respectively. It should be noted that
the anodal TCES trial produced a D CST wavelet with the
same latency as TCMS. Cathodal TCES produces a smaller
D wavelet. The wavelets following the D wavelets in the
upper three trials are I wavelets (see text). The lowest
response in this figure was obtained from anodal TCES during general anesthesia. It should be noted that the D wavelet
(due to direct CST neuronal excitation) is still present,
despite the fact that the patient is anesthetized, whereas, the
I waves, produced by transsynaptic excitation of CST neurons, succumb to the depressant effects of general anesthesia.
Results depicted in this figure do not support the earlier
hypothesis that TCMS excites CST neurons only through a
transsynaptic mechanism (see text). The general oral tracheal inhalational anesthesia was 0.5% isoflurane, 50.0%
nitrous oxide by volume in oxygen. Negative voltage excursions are up in this figure. Voltage and time calibration are as
shown in the figure. (Reprinted from Fujiki et al., 1996.)
discussed above, where TCES caused an earlier activation of peripheral single muscle fibers than that
seen with TCMS (Burke et al., 1990; Edgley et al.,
1990). Thus, direct activation of CST neurons by
TCMS was compatible with the earlier single muscle
fiber studies if TCES caused earlier activation
because of a shift in CST impulse initiation closer
to the spinal cord.
20.4. Anesthetic considerations for monitoring
descending motor systems
Selective monitoring of spinal cord motor function
involves acquisition of TCMS-induced epidural and/
or myogenic responses (i.e., muscle compound action
potentials). Because spinal cord surgery is usually
performed in anesthetized patients, it is important to
discuss the effect of general anesthetic agents on these

responses. Tung et al. (1988) reported that general anesthesia using isoflurane-obliterated single pulse TCMSinduced myogenic responses. Since that report, several
research groups have searched for anesthetic agents
that permit TCMS-induced CST responses (Zentner
et al., 1989; Zentner and Ebner 1989; Nadstawek
et al., 1993). General anesthesia is expected to fulfill
the following requirements for spinal cord surgery:
loss of consciousness, amnesia, analgesia, absence
of surgically induced movements, homeostasis of
vital organ function, and expeditious postoperative
arousal, allowing early neurologic assessment (Stanski,
1994).
Early studies concerning the depth of anesthesia
using inhalational agents suggested that the loss of
function seen with increasing concentration of inhalational agents followed an orderly sequence.
Increasing anesthetic concentrations produce analgesia and amnesia followed by loss of consciousness
(Thornton and Jones, 1993; Stanski, 1994). Much
larger concentrations were required for depression
of surgically induced movements. This lead to the
concept of a minimum alveolar concentration in
which half of the population of patients would move
in response to a surgical stimulus. An analogous concept was developed for intravenous agents (Schwinn
et al., 1994). By achieving a minimum alveolar concentration, the anesthesiologist was usually assured
of analgesia, amnesia, and hypnosis in the patient.
The introduction of muscle relaxants allowed for
the use of lower concentrations of general anesthetic
agents leading to an increased incidence of patient
awareness under anesthesia (Sloan, 1990; Heneghan,
1993; Stanski, 1994; Stinson et al., 1994). Recent
studies suggest that general anesthetics inhibit surgically induced movements by depressing spinal cord
interneural circuits (Rampil and Laster, 1992; Rampil, 1993, 1994; Rampil et al., 1993). The depression
of the spinal gray matter in part explains the depressive effects of most general anesthesia agents on single pulse TCMS-induced myogenic responses (Zhou
et al., 1997; Pereon et al., 1999). Studies concerning
anesthetic effects on CST responses have also
showed minimal effects on the D wave, which is
thought to be elicited by direct activation of CST
cells of origin in the cortex. Occasional I waves are
reported when lower concentrations of inhalational
agents are used (Burke et al., 1990; Deletis, 1993).
Thus, a second site of anesthetic agent depression
of TCMS-induced myogenic responses is at the
293
cortical interneuronal level (Sloan and Levin, 1991;

Woodforth et al., 1999; Sloan and Heyer, 2002).
Amassian (Amassian et al., 1987a) has demonstrated
the necessity of trains of CST volleys (i.e., both D
and I waves) for producing myogenic responses.
It is reasonable to assume that under anesthesia, the
lack of I waves leads to loss of temporal summation
at the spinal cord interneuron and anterior horn cell
level, precluding the consistent acquisition of myogenic responses (Woodforth et al., 1999; Zhou and
Zhea, 2000; Sloan and Heyer, 2002; Kawaguchi and
Furuya, 2004). Therefore, some clinicians rely on
monitoring the epidural TCMS-induced D wave
because it is minimally affected by general anesthetic
agents (Burke et al., 1990; Deletis, 1993).
Myogenic responses should be monitored during
surgery on the descending aorta. The use of epidural
CST responses in this type of surgery leads to a high
false negative rate for predicting new postoperative
motor deficits (Elmore et al., 1991; DeHann et al.,
1996). Recent clinical experience has suggested a
few anesthetic techniques that permit acquisition of
single pulse TCMS-induced myogenic responses.
A nitrous oxide narcotic technique has been successfully used: nitrous oxide is used at a 50% by volume
concentration ratio in oxygen (Zentner, 1989a; Aglio
et al., 2002b). Narcotics in high concentrations minimally depress TCMS-induced myogenic responses
and have been used successfully as the main anesthetic when monitoring myogenic responses during
descending aortic surgery (Gugino et al., 1992). This
technique, however, does not permit early arousal
after surgery. Lower concentrations of narcotics (i.e.,
for analgesia) can be successfully combined with ketamine, etomidate, methohexital, and propofol in
oxygen for spinal cord surgery requiring early postoperative arousal (Ghaly et al., 1990a,b,c; 1991a,b,c;
Losasso et al., 1991; Kalkman et al., 1992; McPherson,
1994; Ubags et al., 1997; Ghaly et al., 2000; Aglio
et al., 2002a). In this regard, etomidate, in particular,
has been a useful hypnotic agent for the successful
use of the single pulse TCMS technique because of
its well-known disinhibitory effects of the cerebral
cortex (Ghaly et al., 1990b; Losasso et al., 1991; Ubags
et al., 1997). This increase in excitability is seen as an
amplitude increase in the cortical SEPs from both
upper and lower limbs (Ghaly et al., 1990b; Losasso
et al., 1991; Ubags et al., 1997). The same phenomenon may also affect the motor cortex in the frontal
lobes. Ketamine and methohexital also have been
shown to have permissive effects for acquiring
294
myogenic responses using the single pulse TCMS

techniques (Ghaly et al., 2000).
Each of these techniques requires the use of
muscle relaxants. TCMS-induced myogenic responses
can be acquired with controlled muscle relaxantinduced myoneural junction blockade of 5070%
(Sloan, 1990). An automated train-of-four monitor
can determine the degree of muscle relaxation. The
potent inhalational agents, barbiturates (except for
methohexital), and benzodiazepines should be avoided
when monitoring TCMS-induced myogenic responses
when using single pulse TCMS (Tung et al., 1988;
Haghighi et al., 1990; Kalkman et al., 1992;
McPherson, 1994; Yamada et al., 1994).
As will be discussed in Section 20.6, the use of the
train TCMS increases the list of permissive anesthetic
techniques. The probable reason for this is that train
TCMS produce, in general, repetitive D waves, which
are minimally depressed by general anesthetic agents.
This attribute, in effect, removes one anatomic region
of anesthetic depression the cortical interneurons
responsible for producing I waves, which is necessary
for adequate excitation of anterior horn cells using the
single pulse TCMS technique. Intravenous propofol or
methohexital, narcotic, and nitrous oxide in oxygen at
greater than 50% by volume and potent inhalational
agents using less than 0.5 MAC concentrations have
been used successfully for acquiring responses with
train magnetic and/or electrical stimulation for TCES
(Kalkman et al., 1993; Taniguichi et al., 1993; Pechstein
et al., 1996; Watt et al., 1996; Lotto et al., 2004). These
anesthetic techniques, suitable for train TCMS, presumably produce more depression of the cortical
interneurons than the anesthetic techniques which permit motor response acquisition using single pulse
TCMS.
20.5. Intraoperative monitoring with single pulse
TCMS techniques
The early experience of monitoring spinal cord motor
function with TCMS use single pulse magnetic stimulation. As discussed in the previous section, successful monitoring of myogenic responses requires
repetitive activation of spinal cord anterior horn
cells as afforded by the D and repetitive I wave
volleys generated by single transcranial stimulation.
As potent general anesthetics tend to suppress the
cortical synapses producing the CST I waves, two
modifications were necessary for successful use of
the single pulse technique. The first modification
L.D. GUGINO ET AL.
was the design of an improved stimulation coil for

producing TCMS excitation of the motor cortex.
The improved coil is called the cap coil (Cadwell Laboratories, Inc., Kennewick, WA), shown in
Fig. 14. Several characteristics of this coil make it
efficacious for TCMS. The first characteristic is the
increased size of the coil due to an increase in the
wire turns of the coils (Kraus et al., 1992). The larger
diameter (17 cm) induces more intensive magnetic
fields (Jalinous, 1991). This leads to induction of
greater electric fields deeper in the motor cortex.
This is useful because it leads to an electric field that
parallels CST neurons located in the anterior bank of
the central sulcus. These CST neurons excite anterior
horn cells activating distal upper extremity muscles,
as well as CST neurons located within the medial
aspects of motor cortex. The latter neurons excite
distal lower extremity muscles. Intracortical currents,
which flow parallel to the long axis of CST neurons,
are the most efficacious for producing excitation
(Amassian et al., 1991). The anterior horn cells
that supply distal extremity muscles typically receive
monosynaptic CST synaptic excitation. The
decreased requirement for spinal cord interneuron
excitation of these anterior horn cells may lead to a
decreased general anesthetic depression, as fewer
synapses are involved in the CST anterior horn call
pathway.
The second characteristic is the saddle shape of the
cap coil (Fig. 14). This shape brings the wire turns
of the coil closer to the scalp as compared to the rigid
planar round coil. The induced magnetic field pulse
coupling to the underlying motor cortex is increased.
Figure 15A shows the induced electric field
strength as a junction of location along a reference
plane 1 cm below the cap coil. Compared to the
electric field distribution of the tear-drop and round
coils (Fig. 4AE), the cap coil can be seen to have
a similar volcano appearance. However, there is a
fore and aft peak that is 1520% above the remainder
of the peak on top of the activation area. This
induced voltage boost is a manifestation of the
improved magnetic field coupling to the underlying
motor cortex afforded by the saddle shape. Fig. 15B
shows the electric field as a function of increasing
depths from the surface of a three-shell concentric
spherical model of the head and brain for the cap
coil, a 7 cm planar round coil, and the 5 cm figure
eight coil. The cap coil induces the greatest electric
field intensity at all depths within the model (Kraus
et al., 1992).
295
z
y
y
x
C
Fig. 14. A: Prototype of the cap coil (17 cm in diameter). B: Line segment representation of the cap coil showing saddlelike design of this coil. C: Radiograph of the cap coil placed on a volunteers head showing the potential for ease of secure
placement important during monitoring sessions. (Reprinted from Kraus et al., 1992 with permission from Lippincott,
Williams and Wilkins.)
E (V/m)
200
250
E
(V/m)
200
150
100
Cap coil
100
0
20
0
0
x (cm)
20
20
20
y
(cm)
Round coil
Figure-of-eight coil
50
0
10
20
30
40 50 60
Depth (mm)
70
80
90
Fig. 15. A: The induced electric field distribution within a homogeneous plane 1 cm below the lowest edges of the cap
coil. Note the peaks of accentuated field intensity caused by the curved edges of the cap coil, which approach the reference
plane at a reduced distance compared with the remainder of the coil. B: A comparison of the attenuation of the induced
electric fields beneath the cap coil (17 cm in diameter), the planar round coil (9 cm in diameter), and the figure of eight
(5 cm per wing) coil. Note the faster rate of field attenuation with axial distance for the smaller coils. See text for further
information (Reprinted from Kraus et al., 1992 with permission from Lippincott, Williams and Wilkins.)
296
L.D. GUGINO ET AL.
compound muscle action potential (CMAP) amplitude for each of the muscles studied. Fig. 17 shows
the same mapping data redrawn to scale on a diagram
of the scalp showing the locations where a CMAP of
greater than 50 mV was acquired for each muscle.
Fig. 18 shows the cap coil superimposed on the
motor map at four different positions. As shown
in Fig. 18, responses from all four limbs occur
when the caps coil overlies all four optimum scalp
loci. This, in general, occurs when the anterior edge
of the cap coil is located 2 cm above the nasion
allowing the caps posterior edge to overlie all four
scalp loci.
Figure 19 shows the change in MEP amplitude
and response area (the product of voltage and time
A second benefit of the cap coils saddle shape is

the ease of attaching the coil to the patients heads
during monitoring sessions. It is important to know
how to place the cap coil on a patients head in order
to maximize the amplitude of acquired motor evoked
potentials (MEPs). To determine the optimum placement of the cap coil with respect to the underlying
motor cortex, the cortex of five volunteers was
mapped using a figure eight stimulating coil (Aglio
et al., 2002a). Fig. 16 shows the typical resulting
amplitude distributions for the left and right adductor
digiti minimi and tibialis anterior muscles. The distributions are plotted as a pseudo-three-dimensional
plot where the xy plane represents stimulated locations along the scalp with the z axis showing the
2000
(V)
*
Cz
1000
12 10 8 6 4 2 0
Left
10 12
3
1
10
5 Anterior
Posterior
Right
A
RADM RTA LTA LADM
RADM
5
RTA
45
45
85
85
1000
V
LADM
ms
5
LTA
45
85
45
85
B
Fig. 16. A: An example of the topographical distribution of compound muscle action potential (CMAP) amplitudes
acquired using focal transcranial magnetic stimulation (TMS). The distribution is displayed as a pseudo-three-dimensional
plot of four bar graphs, one for each of the muscles used for locating the motor cortex optima. The location on the scalp
where focal TMS-elicited CMAP responses is indicated by bars plotted on the 1 cm 1 cm grid that is referenced to the
vertex and centered in the graph. The amplitude scale is given in mV. Each group of bars representing CMAP amplitudes
was derived as a function of the scalp location used for stimulating the left and right abductor digiti minimi (LADM and
RADM) and left and right tibialis anterior muscles (LTA and RTA). Gray scale-coding of the bar graphs is used to identify
each of the four muscles studied according to the gray scale below. B: Below the pseudo-three-dimensional plot are examples of CMAP responses acquired from the four muscles when local TMS was applied at the scalp locations resulting in the
largest response amplitude for each muscle. These scalp locations are marked by asterisks within each scalp optima indicated on the plot in (A). Each set of three consecutively acquired responses is shown to indicate the characteristic stability
of the replicated responses when focal stimulation was applied to the scalp locations indicated. The voltage calibration for
responses was (1) left and right ADM: 3000 mV and (2) LTA and RTA: 1000 mV. Time sweep calibration is 10 ms. All
traces begin at 8 ms for artifact elimination. Negativity is up for all responses. (Reprinted from Aglio et al., 2002a with permission from Clin. Neurophysiol.)
297
Nose
* *
Cz
Inion
RADM RTA
LTA LADM
Fig. 17. A map of the scalp showing the location of the

four muscles scalp optima is displayed. The data for producing this map were taken from the pseudo-three-dimensional plot shown in Fig. 16. This alternate format was
constructed to show the scalp motor optimal locations from
the viewers perspective of looking down on a twodimensional model of the scalp surface. The anterior aspect
of the scalp is at the top of the map. The 1 cm 1 cm grid
drawn on the subjects scalp during the mapping study is
drawn on the scalp map centered at the vertex (C2). Identification of each muscle optimum location uses the same
gray scale scheme used for the construction of the
pseudo-three-dimensional plot in Fig. 16. Asterisks mark
the location within each map optima where TMS resulted
in compound muscle action potentials (CMAPs) of maximum amplitude. (Reprinted from Aglio et al., 2002a with
permission from Clin. Neurophysiol.)
duration of the MEP) as a function of stimulation

strength. In general, as the stimulus intensity increases,
a hyperbolic relationship is seen in which MEP amplitude and response areas approach a plateau in the
awake volunteer. Take-off latency, however, shows a
modest decrease (Fig. 20).
Figure 21 shows in diagrammatic form, the monitoring protocol used for acquiring intraoperative
MEPs. All monitored procedures (N 27) were carried out on posterior spinal cord procedures (Aglio
et al., 2002b). Scalp recording electrodes were placed

at the internationally 1020 defined C30 and C40 for
recording SEP electrical stimulation. MEPs were
acquired from the adductor digiti minimi and tibialis anterior muscles using silversilver chloride
stick-on skin electrodes placed over the muscles
of interest in a muscle belly-tendon fashion. As
discussed above (see Sections 20.3 and 20.4), the
use of a single pulse TCMS requires efficacious
excitation of both the CST cells of origin, and the
cortical interneurons responsible for re-excitation
of the same population of cortical neurons. In this
way, a volley of descending CST activity is
obtained for producing temporal summation of
CST synaptic activation of anterior horn cells.
The second concern for anesthetic depression is
the CST-anterior horn cell synapse. In our experience, the CST anterior horn cell pathway was least
suppressed using one of three anesthetic techniques
(see Table 1). Etomidate is a hypnotic agent felt to
cause disinhibition of the cerebral cortex. Thus, the
excitability of cortical interneurons required for
repetitive reactivation of the CST neurons is better
maintained using this agent. It is given as a constant infusion along with a narcotic infusion and
nitrous oxide in oxygen at or less than 50% by volume. Muscle relaxation is maintained relatively
constant at a 50% block with a relaxant infusion
controlled by a train of four monitor. A second
anesthetic technique was a propofol total intravenous anesthetic (TIVA) with propofol infused at
75100 mg/kg/min supplemented with ketamine
(40 mg bolus once an hour). No nitrous oxide was
used for this technique. These two anesthetic techniques were developed in order to assure patient
loss of consciousness and a rapid patient emergence at the end of the procedure.
In procedures where immediate postprocedural
emergence was not required (i.e., aortic surgery), a
high-dose narcotic and an etomidate infusion technique was administered with oxygen. Table 1 also
shows the percent of patients in whom responses
suitable for monitoring were not obtained along with
the percentage of monitoring failures who had peri
operative motor deficits. Although the number of
patients within each anesthetic category is small,
the anesthetic most permissive for acquiring MEPs
is the high-dose narcotic in oxygen technique.
Figure 22 shows the distribution, in bar graph
form, of the tibialis anterior MEP frequency, baseline
to peak amplitude, maximum peak-to-trough
298
L.D. GUGINO ET AL.
RADM
LADM
48
88 ms
48
88 ms
RTA
LTA
48
88 ms
48
88 ms
RADM
LADM
48
88 ms
48
88 ms
LTA
48
88 ms
RTA
48
88 ms
RADM
LADM
48
88 ms
48
88 ms
RTA
LTA
48
88 ms
48
88 ms
RADM
LADM
8
48
88 ms
48
88 ms
48
LTA
RTA
48
B
C
A
2000
V
D
Cz
RADM RTA
88 ms
LTA LADM
20
ms
88 ms
Left TA
Right ADM
4000
4000
3000
3000
V
mV
Left ADM
299
2000
30
40 50 60 70 80
Stimulus intensity
90 100
30
0
20
30
Left TA
50
45
40
35
30
25
20
15
10
5
0
20
40 50 60 70 80 90 100
Stimulus intensity
40 50 60 70 80
Stimulus intensity
Right ADM
Area under the curve

(thousands)
Area under the curve

(thousands)
Left ADM
2000
1000
1000
0
20
Right TA
50
45
40
35
30
25
20
15
10
5
0
20
30
90 100
Right TA
40 50 60 70 80
Stimulus intensity
90 100
Fig. 19. Responses of the amplitude and area of the compound muscle action potentials (CMAPs) to increasing stimulus
intensity. The cap coil was placed anterior on the scalp 2 cm above the nasion in all five awake volunteers. Stimulus intensity along the x axis increased from 20% to 100% of the MES-10 power. Parts A: and B: represents the amplitude response
for the left ADM and TA, and right ADM and TA CMAPs, respectively. Parts C: and D: represent the area under the curve
response for the same CMAPs, respectively. Error bars represent the standard error the mean. (Reprinted from Aglio et al.,
2002a with permission from Clin. Neurophysiol.)
amplitude, and area under the response for 17 of the

patients anesthetized with the etomidate, narcotic
nitrous technique (Aglio et al., 2002a). The data were
collected 20 min after anesthetic induction, but
before surgical manipulation of the spinal column
began. Note that latency appears to have a Gaussian
distribution, whereas amplitude or response areas do
not. Logarithmic transformation produces a near
Gaussian distribution for these MEP parameters.

Fig. 23 shows, for comparison, the distribution for
the lower limb cortical SEPs.
Tables 2 and 3 show descriptive statistics for
the MEPs and SEPs acquired prior to surgical manipulation of the spinal column. Based on these values,
it was shown that a change of 40.6% decrease in
amplitude and an 8.6% increase in latency would
Fig. 18. Demonstration of compound muscle action potential (CMAP) amplitude for left and right ADM and TA as a
function of cap locations for the same volunteer whose data are shown in Figs. 16 and 17. In the center are depicted
four cap positions on the scalp. A: anterior with cap rim 2 cm above the nasion; B: cap coil centered on vertex; C: anterior rim at the vertex, and D: cap anterior rim 3 cm posterior to the vertex. The cap coil locations are depicted as a coil
of wire superimposed on replicas of the scalp model shown in Fig. 17, which identify the scalp areas where focal TMS
produced CMAPs for the left and right ADM and TA muscles. The grid has been removed for purposes of clarity. On
the lower left of the figure is a lateral view of the head that shows the relative position on the head of the cap coil placements (A) through (D) from this perspective. The gray scale shown at the bottom right of the figure is used for identifying the scalp muscle focus defined by the figure eight coil. To the left and right of each scalp model are examples
of CMAPs obtained from each of the four muscles when simulated at 80% full power of the MES-10. The vertical calibration bar is equal to 2000 mV. Negativity is up in all the traces. Sweep is shown in ms for each set of three
responses. In each set of responses, the sweep begins at 8 ms for artifact elimination. (Reprinted from Aglio et al.,
2002a with permission from Clin. Neurophysiol.)
L.D. GUGINO ET AL.
Latency (ms)
300
40
38
36
34
32
30
28
26
24
22
20
50
60
70
90
80
100
Stimulus Intensity
Fig. 20. Relationship of mean latency across volunteers of the left and right ADM and TA compound muscle action potentials (CMAPs) as a function of stimulus intensity. Stimulus intensity axis is represented as percent of maximum MES-10
power. Latency axis is in ms. Error bars represent the standard error of the mean. (Reprinted from Aglio et al., 2002a with
permission from Clin. Neurophysiol.)
SURGERY
To Evoked Potential
Recording Instrument
CAP COIL
T4 MUSCLE
RELAXANT MONITOR
Intra-venous
Anaesthesia
Descending Volley
Sciatic Nerve
Deep Peroneal Nerve
Tibialis Anterior
OPERATING BED
Fig. 21. Diagrammatic representation of the monitoring protocol used for acquiring motor evoked potentials (MEPs) during
spinal cord procedures. A cap-shaped magnetic stimulating coil is secured to the patients head for providing a single transcranial magnetic stimulus to the underlying cortex. MEPs were acquired from the tibialis anterior muscle for monitoring
spinal cord motor function during procedures involving the thoracic and lumbar spine (N 26). MEPs were acquired from
the first dorsal interossei muscles for a procedure on the cervical cord (not shown). The degree of muscle relaxation was
assessed with an automated train of four monitors placed on a muscle of the upper limb. The protocol for SEP acquisition
is omitted for ease of visualization. (Reprinted from Aglio et al., 2002b with permission from Clin. Neurophysiol.)
Table 1
Single pulse technique: distribution of anesthetic techniques and response failures for monitoring motor evoked
potentials (MEPs)a
Type of procedure
Number of
cases
Type of anesthetic
% Nonmonitorable
Vascular
Spinal column
Spinal cord
2
4
21
High-dose narcotic, etomidate

Ketamine, propofol, narcotics
Etomidate. narcotics, nitrous
oxide
0
50 (2)b
19 (4)
0
100 (2)
25 (1)
Total
27
22 (6)
50 (3)
% Preoperative deficits/
nonmonitorable
One of the 27 patients had the upper limb monitored. The other 26 patients had lower limb monitoring (reprinted from Aglio et al. (2002b)
with permission from Clin. Neurophysiol.)
b
Numbers in parentheses indicate number of patients.
Distribution of MEP Response Descriptors

MEP Peak Latency
MEP Log [Peak Latency]

50
Frequency
Frequency
60
50
40
30
20
10
0
<16 20
24
28 32
ms
36
40 >40
MEP [Peak to Trough

Amplitude]
100
30
20
10
0
<1.38 1.4 1.44 1.48 1.52 1.56 1.6 >1.6
Log [ms]
MEP Log [Peak to Trough
Amplitude]
50
40
Frequency
80
Frequency
40
60
40
20
30
20
10
0
0
0
0.2 0.4 0.6 0.8 1.0 1.2 >1.2

mV 103
<1 1.4 1.8 2.2 2.6

Log [mV]
MEP Area
100
40
Frequency
50
Frequency
60
40
20
30
20
10
0
1.6
0
0
1.5 3.0 4.5 6.0 7.5 9.0 >9.0

mV/ms 103
120
100
80
60
40
20
0
2.4 2.8 3.2 3.6

Log [mV]
<4
MEP Log [Base of Peak

Amplitude]
40
Frequency
Frequency
MEP Base to Peak

Amplitude
3.4 >3.4
MEP Log [Area]
120
80
30
20
10
0
0 0.15 0.3 0.45 0.6 0.75 0.9 <1.05

mV 10 3
0.9 1.2 1.5 1.8 2.1 2.4 2.7 <3
Log [mV]
Fig. 22. Distribution of TCMS-induced MEP response descriptors acquired from 17 patients, 20 min after induction of an
etomidate, narcotic, nitrous oxide general anesthetic. Myoneural junction block was maintained at 50% during acquisition
of the MEPs. The left column of histograms shows the distribution of nontransformed values for latency, first peak-totrough MEP amplitude, and baseline to first peak amplitude for the MEP responses. The corresponding histograms along
the right hand column show the same distribution after logarithmic transformation. Note the transformation of latency
values adds little to producing a Gaussian distribution as this parameter already has a normal distribution. On the other
hand, peak-to-trough amplitude incompletely approaches a normal distribution with logarithmic transformation. (Reprinted
from Aglio et al., 2002b with permission from Clin. Neurophysiol.)
302
L.D. GUGINO ET AL.

Distribution of SEP Response Descriptors
SEP Peak Latency
SEP Log [Peak Latency]
50
50
40
40
30
30
20
20
10
10
0
<36 38
40
42 44
ms
46
48 >48
<1.56 1.58 1.6 1.62 1.64 1.66 1.68 >1.68

Log (ms)
SEP Peak-Trough
Amplitude
SEP Log [Peak-Trough

Amplitude]
80
70
60
50
40
30
20
10
0
60
50
40
30
20
10
0
<0.7 1.4
2.1
2.8
3.5
4.2
4.9
>7
mV
<0.19 0
Log (mV)
SEP Area
SEP Log [Area]
60
60
50
50
40
40
30
30
20
20
10
10
0
0
0.8
0.06 0.18 0.3 0.48 0.82 >0.82
12
18
24
mV/ms
30
36 up to
36
<0.6 0.8
1.2
1.4
1.6
1.8 >1.8
Log (mV/ms)
Fig. 23. Distribution of cortical SEP response descriptors acquired from 17 patients, 20 min after induction an etomidate, narcotic, nitrous oxide general anesthetic. The left column of histograms shows the distribution of nontransformed values for peak
latency, first peak-to-trough cortical SEP amplitude, and SEP response area. The corresponding histograms along the right hand
column show the same distributions after logarithmic transformation. Note that transformation of latency values adds little to producing a Gaussian distribution as this parameter already has a normal distribution. On the other hand, peak-to-trough amplitude
and area incompletely approach a normal distribution with logarithmic transformation (unpublished observations).
define a significant change from baselines for the

SEPs (Aglio et al., 2002b). The variability of MEP
parameters is two to three times that for averaged
SEPs cortical responses. Similar calculations for
determining a statistically significant change in

MEPs would be a complete loss of MEPs over two
sequential updates during a surgical procedure (Aglio
et al., 2002b).
303
Table 2
Means and coefficients of variation for motor evoked potential (MEP) response characteristics (n 17)
Mean S.D.a
Coefficient of variationb (%)
Range of coefficient of variationc (%)
Required changed (%)
Required change after log
transformationd (%)
Take-off latency (ms)
Amplitude (mV)
Area (mV/ms)
29.6 3.1
11.2
1.2 53.6
20.2
44.1
264.0 152.8
53.6
17.0 99.2
113.4
54.4
1789.1 1025.2
58.8
26.3 105.7
112.3
41.5
Mean and S.D. are average mean and S.D. values of 17 patients.
The coefficient of variation is the average coefficient of variation of 17 patients.
c
Range of coefficient of variation is the minimum and maximum value of CV seen in 17 patients.
d
Required change is the percent change necessary before an individual value would be significantly different from the preoperative control
[p < 0.05]. It is calculated as follows: Required change [preoperative S.D. 100 1.96]/[preoperative mean] (reprinted from Aglio et al.
(2000b) with permission from Clin. Neurophysiol.)
b
Figure 24 shows an example of MEP monitoring

for a repair of a coarctation of the aorta (Gugino
et al., 1992). The patient experienced re-stenosis of
previous aortic coarctation repair. During the application of the aortic cross-clamp, the tibialis anterior
MEPs disappeared. A distal left atrial to femoral
artery bypass was not done. The MEPs were lost
within minutes after applying the aortic cross-clamp.
Note that the SEPs were unaffected. The aortic crossclamp was then removed. The surgeon then elected
to place a bypass graft across the coarctation rather
than attempt a primary repair. A side-biting clamp
was used to facilitate placement of the bypass graft.
This allowed some aortic flow to reach the lower
extremity during the graft placement. Fig. 24 shows
that during the placement of the side-biting clamp,
the MEPs, were attenuated to half of the baseline
MEP amplitudes. The distal pressure had decreased

from a mean of 9045 Torr. With removal of the
side-biting clamp, after completion of the bypass
graft, the distal pressure normalized to baseline
values as did the MEP amplitude. The patient
remained intubated in the intensive care unit. The
patient had a normal motor examination after extubation the next day.
Table 4 shows outcome data for the patients monitored with single pulse TCMS (Aglio et al., 2002b).
The state of the MEPs at the end of the procedure
were compared to the motor function exam after the
procedure. Note the incidence of one false positive
and no false negatives using the criterion discussed
above for predicting postoperative motor function
with single pulse TCMS monitoring (Aglio et al.,
2002b).
Table 3
Means and coefficients of variation for somatosensory evoked potential (SEP) response characteristics (n 17)
Mean S.D.a
Coefficient of variationb (%)
Range of coefficient of variationc
Required changed (%)
a
Peak latency (ms)
Amplitude (mV)
Area (mV/ms)
42.5 1.9
4.4
0.5 10.5
8.6
1.5 0.3
25.6
6.6 57.2
40.6
15.0 3.6
29.9
8.3 67.1
46.5
Mean and S.D. are average mean and S.D. values of 17 patients.
The coefficient of variation is the average coefficient of variation of 17 patients.
c
Range of coefficient of variation is the minimum and maximum value of CV seen in 17 patients.
d
Required change is the percent change necessary before an individual value would be significantly different from the preoperative control
[p < 0.05]. It is calculated as follows: required change [preoperative S.D. 100 1.96]/[preoperative mean] (reprinted from Aglio et al.
(2000b) with permission from Clin. Neurophysiol.)
b
304
L.D. GUGINO ET AL.
AWAKE
LAT xxx ms
AMP 1.0 uV
STIM: R Post Tib Nerve; Record: SEPCX
2.0 uV
LAT 38.5 msec

ANESTHETIZED AMP 1.9 uV
LATENCY (change from Baseline in ms)

AMPLITUDE (% change from Baseline)
LAT 38.0 ms
AMP 430.9 uV
STIM: Transcranial Magnetic; Record R CMAP Ant Tib Mus

100%
7
1000.0 uV
7
100%
0
TIME (s)
TCM
R SEPCX
R LATM
S LPTN
R RATM
S RPTN
85
ELAPSED TIME
0.0
2.4
3.3
SEPCX
MOTOR
COMPARED
MONITORING
TO AWAKE
BEGUN
BASELINE
SEPCX
AORTIC
COMPARED
CROSS
TO ANESTH.
CLAMP
BASELINE
ON
AORTIC
CROSS
CLAMP
OFF
AORTIC
SIDE
CLAMPS
PLACED
4.4
5.3 hrs
AORTIC SIDE
CLAMPS
REMOVED
Fig. 24. Monitoring SEPs and TCMS-induced CMAPs during repair of a coarctation of the aorta. The upper left hand of the
figure shows conscious and anesthetized baselines for the scalp somatosensory evoked potentials acquired by stimulating
the right posterior tibial nerve at the ankle. Below is an anesthetized baseline of the TCMS-elicited CMAP acquired from
the right tibialis anterior muscle. To the right of the baselines are trajectories showing the response latency and amplitude
changes during the operation. The latency and amplitude changes are depicted by solid and broken lines, respectively. Important events during the operation are shown below the anterior tibialis CMAP trajectory. After loss of the MEP response, the
aortic cross-clamp was removed, leading to an immediate restoration of tibialis anterior CMAPs. Note the lack of change in
the monitored SEPs. SEPs typically require an ischemic challenge of 1015 min in order to undergo changes suggestive of
pending damage to the spinal cord. Changes in lower limb peripheral amplitude loss was attributed to aortic cross-clampinduced ischemia of the spinal cord gray matter. Of interest, aortic cross-clamp-induced ischemia can lead to a central cord
syndrome, reflecting necrosis of the spinal cord gray matter. Because of the rapid loss of TCMS-induced lower limb myogenic
responses during the aortic test clamp period, the surgeon elected to perform a bypass procedure in which a graft was used to
bypass the aortic constriction. Side-biting clamps were used to isolate the anastomoses sites above and below the coarctation.
The femoral blood pressure was reduced with the decrease of these clamps. Concomitant to the distal blood pressure decrease,
the tibialis anterior CMAPs showed a partial loss of amplitude as seen in the response sequences at the right, as well as by the
lower trajectory in the figure. Removal of the partial aortic clamp was accompanied by restoration in distal blood pressure as
well as the monitored CMAP response amplitudes. The patient has an uneventful postoperative course on recovering from
anesthesia. The figure displays the acquired responses from the right side only to simplify the case presentation. Responses
from the left side during the operation showed similar changes. Negativity up in all response sequences. Voltage and time calibration shown at the right and left side of the figure. (Reprinted from Gugino et al., 1992 with permission.)
20.6. Train magnetic transcranial stimulation

The introduction of train stimuli was a major
improvement for intraoperative noninvasive excitation of the cortex (Kalkman et al., 1993; Taniguichi
et al., 1993; Pechstein et al., 1996). The use of stimulating trains in which the train pulses are set to an
appropriate interpulse interval (i.e., usually from 2.0
to 5.0 ms) can produce repetitive direct excitation
of cortical neurons. This eliminates the reliance on

intracortical interneurons for producing the repetitive
synaptic re-excitation of upper motor neurons. As
discussed in a previous section, volleys of descending CST activity are required for successful excitation of lower motor neurons, particularly during
general anesthesia. We have had experience with a
magnetic train stimulator capable of producing four
monophasic pulses at a frequency up to 1,000 Hz
305
Table 4
Motor evoked potential (MEP) outcome prediction
n
Postoperative changes
True positives
True negatives
False positives
False negatives
Sensitivity
Specificity
21
3b
3
17
1
0
100%
94%
75%
100%
Sensitivity (true positive)/{true positive false negative);

specificity (true negative)/(true negative false positive); positive predictive value ((true positive)/(true positive false positive); negative predictive value (true negative)/(true negative
false negative) (reprinted from Aglio et al. (2002b) with permission from Clin. Neurophysiol.)
b
The fourth patient did not survive the surgical procedure (patient
3, see Table 1).
a
(Quadropulse, Magstim Ltd., Wales, UK). It has

the capability for reversing the current polarity
within the stimulating coil. This feature allows for
reversing the induced intracortical currents polarity
for more efficient unilateral motor cortex excitation,
particularly of the upper limb area (Gugino et al.,
1997).
Although several stimulating coil geometries
have been tried, the currently used coil for acquiring
lower limb responses is an oversized figure eight
design, which has a bend along the body of the coil.
The coil is referred to as the cone coil (Magstim,
Ltd., Wales, UK). As was the case for the cap coil,
this geometry serves to bring the induced magnetic
flux from the coil wings as well as the center of
the coil closer to the scalp (see Fig. 25). This
improves the coupling of the magnetic flux with
the underlying cortex. A curved round coil has also
been used for procedures in which only the upper
limb MEPs was monitored.
Both coils are fixed to the scalp with the posterior
edge overlying the vertex such that the lateral and
longitudinal midpoint of the coil straddles the vertex.
Fine coil adjustment involved an anterior or posterior
displacement from the initial position for optimizing
the recorded responses. Bite blocks are placed in
order to prevent potential tongue lacerations from
possible masseter contractions.
MEPs are recorded from lower extremity muscles

with the same techniques used for single pulse TCMS.
Fig. 26 shows the appearance of bilateral tibialis anterior myogenic responses during a propofol-based
anesthetic in which the stimulating trains increase
from 2 to 4 pulses. The responses appear at a train of
3 pulses with a minor increase in response amplitude
at 4 pulses (Hufnagel et al., 1990). Fig. 27 shows the
effect of stimulating the motor cortex with the curved
round coil using trains differing in coil current polarity. When the coil current induced an anteriorly
directed cortical current through the right hemisphere,
a left adductor digiti minimi response was obtained.
Reversing the polarity of the coil current produced a
contralateral response. Fig. 28 shows the use of the
cone coil for acquiring bilateral tibialis anterior
responses. A posterior to anterior orientation of
induced cortical current lead to bilateral lower
extremity response (Gugino et al., 1997).
Figure 29 shows the motor responses from the
tibialis anterior muscle during a dorsal root entry
zone thermal ablation for chronic pain. The responses
were acquired with the train technique described
above under a propofol TIVA (Gugino et al., 1997).
Right-sided motor response attenuation occurred
early and persisted to the end of the procedure. The
patient emerged from anesthesia with a new rightside lower extremity weakness.
Table 5 demonstrates the success rate for acquiring
responses as a function of anesthetic technique. Note
that the majority of procedures were monitored during
a propofol-based anesthetic. There were a lower proportion of failed monitoring due to preoperative deficits than was the case for the single shock technique.
Table 6 shows statistical descriptors of the
responses acquired with the train technique for both
upper and lower extremity responses. Note that the
smaller coefficient of variation (compared to the single shock technique [see Table 2]) decreased the
change necessary for defining a significant statistical
change during monitoring. The data for this table
were collected in the same fashion as that for
Table 2.
Finally, Table 7 shows the relationship for the status of the last MEP and motor outcomes for 54 cases
in which motor function was monitored with train
TCMS. Based on the results in Table 6, a significant
MEP amplitude change was defined as an 80% attenuation from baseline values. As was the case for single pulse TCMS monitoring, there was a conspicuous
absence of false negative outcomes.
306
L.D. GUGINO ET AL.
Magnetic field flow
Coil cross-section
B Coil winding
Induced electric field distribution

Maximum
induced
electric
field
Maximum
induced
electric
field
Minimum
induced
electric
field
Planar round coil
Minimum
induced
electric
field
Double cone coil
C
Fig. 25. A: The double cone coil. B: Shows a cross-section of the coil at a segment of each component circular coil and the
region where the circular coils converge. C: The induced magnetic field around these regions of the coil are also shown.
(Reprinted with permission, 1996, Magstim Ltd., Wales, UK.)
20.7. Comparisons of electrical and magnetic

transcranial stimulation
Magnetic stimulation is a less popular technique
compared to electrical techniques for monitoring descending motor function during surgery. There are
several reasons for this. First, the coils are large and
bulky and, in general, preclude monitoring supraspinal
intracranial procedures. Train magnetic stimulators
are three to six times more expensive than electrical

train stimulation, in part due to the requirement of
expensive capacitors for energy storage. Cadwell
has discussed the theoretical reasons of why magnetic stimulators are less efficient than electrical
transcranial stimulation (Cadwell, 1991). For these
latter two reasons, the cost for producing equivalent
stimulation strength is greater for magnetic than electrical stimulators. Finally, magnetic stimulators produce
307
BRIGHAM AND WOMENS HOSPITAL

ANESTHESIOLOGY DEPARTMENT
NEUROMONITORING UNIT
Patient: F
Sex: Male
Test Date: 06-01-94
Anesthesia: Propofol/Fentanyl/O2
Stimulus: Magstim quadroPulse
Coil: Double cone
Recording: Tibialis anterior
Stimulus Artifact
200.00
V/ div
(Sto)
cz
Power Level: 90%

Inter-Pulse Spacing: 2.5 ms
Coil Position: 2 cm Posterior of Cz
2cm
Posterior
Left
Right
Three Pulse
0
200.00
V/ div
(Sto)
80ms
40
200.00
V/ div
(Sto)
1
Two Pulse
0
40
Four Pulse
80ms
40
80ms
Fig. 26. TCMS-induced myogenic responses obtained using a two (lower left), three (upper right), and four magnetic pulse
train. There is a stimulus artifact to the left of each trace. The left and right anterior tibialis CMAPs appeared with only the
three and four magnetic pulse trains. The TCMS train consisted of monophasic pulses with an interpulse interval of 2.5 ms.
These responses were obtained from a patient under a propofol-based total intravenous anesthetic (TIVA). The horizontal
axis is measured in milliseconds. The vertical axis is measured in microvolts per division. Negativity is up. (Reprinted from
Gugino, 1997 with permission from Elsevier.)
a characteristic noise as well as a tendency to overheat

particularly if the stimulation rate is greater than 1
stimulus every 30 s.
Figure 30 shows a comparison of myogenic
response amplitude for lower extremity motor
responses using single electrical or magnetic pulse
stimulation techniques (Ubags et al., 1999). Note
that, in general, TCES-induced myogenic responses
tend to be larger (Fig. 31). Our group compared magnetic to electrical train stimulation techniques in
patients randomized to an intravenous methohexital
or propofol anesthetic technique (unpublished observations). The experimental protocol is shown in
Fig. 32. A Quadropulse magnetic train stimulator
was used to produce a four pulse train with the technique described in Section 20.6. A curved round coil
was used for eliciting upper extremity (first dorsal

interossei) responses. The cone coil was used for
acquiring anterior tibialis responses. The Quadropulse coil current output was then lead into a magnetic to an electrical stimulus adapter. The adapter
converts a magnetic train surrounding the adapters
primary coil into an electrical train by taking the
train voltage drop off of a secondary coil within the
adapter. The dimensions of the secondary coil were
designed to produce, at full Quadropulse output,
an electrical train amplitude of 500 V. Figs. 33 and
34 show examples of the typical results seen when
comparing the upper and lower extremity myogenic
responses acquired using electrical and magnetic
stimulation. While the response amplitudes are similar, the electrical technique did a better job of
308
L.D. GUGINO ET AL.

Left
Right
Left
Right
20 ms
Anterior
07:11:00
07:54:43
07:11:10
07:54:52
2 mV
07:12:59
cz
07:13:10
07:14:56
Circular
130 mm
Coil
Posterior
Patient: H. (Female)
Operation: Ant. Discectomy
Test Date: 02-22-95
Anesthesia:
Propofol/Fentanyl/O2
Muscle Relaxant: Vecuronium
(T1 60-70%)
Stimulus: Magstim
QuadroPulse
Coil: Circular Coil S/N 8377
Recording: First Dorsal Int.
Power Level: 80%
Coil Position: Post Rim at CZ
07:15:06
07:16:55
07:17:05
07:18:53
07:19:02
2 mV
07:56:49
07:56:59
07:58:50
07:59:00
08:00:55
08:01:05
08:03:07
08:03:17
07:21:02
08:05:11
07:21:12
08:05:21
Fig. 27. Example of upper limb myogenic responses obtained with a four-pulse TCMS train during anesthesia. A curved
round coil centered left to right on the scalp with the posterior overlying the vertex was used for obtaining these responses.
The interpulse interval was 2.5 ms with a four-monophasic-pulse train. Time and voltage calibration bars are shown at
upper right of figure. Note the two different display gains for left and right responses. Each set of responses represents
and alternating left and right first dorsal interossei CMAP obtained with reversal of the stimulating coil current direction.
Response negativity up. (Reprinted from Gugino, 1997 with permission from Elsevier.)
isolating excitation to one hemisphere for the upper

extremity responses (Fig. 33). Both techniques produced bilateral lower extremity response despite
stimulation current polarity reversal. For the lower
extremity, however, the magnetic technique produced an obvious asymmetry when comparing the
left- to right-side response (Fig. 34). Tables 8 and 9
show the results obtained for seven trials involving
the upper extremity, as well as two electrical and
one magnetic trial for the lower extremities. Five
responses were collected for each stimulus polarity
and stimulation technique for computing the results
shown in the two tables. Although the numbers
are small, it is apparent that while the take-off
latencies are similar, the electrical technique
produced larger amplitude responses with smaller
variability (i.e., smaller coefficient of variation)
compared to magnetic-induced responses. Thus,
preliminary data supports those suggesting that
electrical transcranial stimulation technique is superior to the magnetic technique for monitoring the
descending motor systems intraoperatively in anesthetized patients.

20.8. Contraindication for TCMS in conscious and
anesthetized patients
Both single pulse and train TCMS have been shown
to minimally affect cognitive function or stress hormone levels in conscious patients (Kandler, 1990;
Krain et al., 1990; Pascual-Leone et al., 1993, 1994;
Chokroverty et al., 1995). Single pulse TCMS has
been reported to cause seizures in conscious patients
with a seizure history (Homberg and Netz, 1989;
Hufnagel et al., 1990; Classen et al., 1995). Train
TCMS, on the other hand, has been reported to elicit
seizures in conscious patients with or without a seizure history (Pascual-Leone et al., 1993, 1994).
For anesthetized patients, the authors know of
only one case of TCMS seizure induction. The procedure involved a 46-year-old female undergoing a
laminectomy for a thoracic tumor excision. The
309
Left
Right
Left
Right
Train of four 2.5 ms inter: pulse
100 mm double cone coil
Two hours
20 ms
500 V
Later
14:19:51
12:29:27
14:24:19
12:35:31
14:28:26
12:39:31
Coil not positioned

in time
Coil not positioned

in time
14:31:57
12:42:35
12:45:58
14:36:38
12:48:40
14:40:03
12:51:26
14:43:30
12:54:15
14:46:56
(ms)
105
Patient: (Male)
Operation: lumbar laminectomy for
intradural tumour
Test date: 06-01-94
Anaesthesia: propofol/fentanyl/O2
Muscle relaxant: vecuronium
(ms)
105
(ms)
(ms)
105
(ms)
105
Stimulus: Magstim QuadroPulse

Coil: double cone coil
Recording: tibialis anterior
Power level: 80%
Inter-pulse spacing: 2.5 ms
Coil position: 2cm posterior of CZ
Fig. 28. Stability of myogenic response sequence from left and right tibialis anterior muscles is illustrated using the double
cone coil and a TCMS train stimuli during a monitored case. The sequence on the right was obtained 45 min after
the sequence illustrated on the left and demonstrates improved response variability using TMS trains compared with
single-pulse techniques. Voltage and amplitude calibrations shown at the top of the left response columns. Negativity up.
(Reprinted from Gugino, 1997 with permission from Elsevier.)
310
L.D. GUGINO ET AL.
Left
Post lesioning
Lesioning
Pre lesioning
Left
Right
Left
Right
08:36:13
09:14:48
10:29:21
08:37:51
09:17:18
10:31:21
08:38:51
09:19:48
10:33:21
08:39:51
09:22:18
10:35:21
08:40:51
09:24:48
10:37:21
09:27:18
10:39:21
08:41:51
08:42:51
09:29:48
08:44:07
08:44:59
08:45:59
10:41:21
09:32:18
10:43:21
09:34:48
10:45:21
09:37:18
10:47:21
09:39:48
10:49:21
08:46:59
0.5 mV
20 ms
Right
0.5 mV
ANTERIOR
BRIGHAM AND WOMEN'S HOSPITAL

ANESTHESIOLOGY DEPARTMENT
NEUROMONITORING UNIT
100 mm
Double
Cone Coil
Posterior
Patient: S. (Male)
Operation: Thoracic DREZ
Test Date: 01-12-95
Anesthesia: Propofol/Fentanyl/O2
Muscle Relaxant: Vecuronium (T1 60-70%)
Stimulus: Magstim Quadro Pulse
Coil: Double Cone Coil
Recording: Rt. and Lt. Ant. Tib.
Power Level: 90%
Coil Position: Centered Over CZ
Fig. 29. Three left and right tibialis anterior CMAP response sequences are depicted in the columns in the upper half of the
figure. They were elicited during a right-sided dorsal root entry zone thermal ablation of the mid-thoracic spinal cord for
chronic pain. The time during the operation at which each TCMS-induced trace was acquired is shown to the left of the
response sequences. A train of four TCMS pulses was used for eliciting the responses shown. The shock artifacts resulting
from the TCMS train are shown to the left of each response. Relevant information concerning the anesthetic stimulation and
recording techniques are shown at the lower right of the figure. The cartoon at the lower left represents the placement of the
magnetic stimulation coil on the patients head. Note the right tibialis anterior CMAP attenuation during the initial postoperatively. Negativity up for all displayed responses. Time and voltage calibration bars are displayed at the bottom of the
first sequence column at left. (Reprinted from Gugino, 1997 with permission from Elsevier.)
patient was anesthetized with a methohexital-based

total intravenous technique.
The patient had a history of seizures with the last
occurring 10 years prior to her hospital admission.
She had not taken anti-seizure medication for over

8 years prior to admission. The patient emerged from
anesthesia with a generalized seizure, which was treated successfully with intravenous benzodiazepines.
Table 5
Four pulse transcranial magnetic stimulation (TCMS) technique
Operation
No.
Anesthetic
Failed responses
Preoperative deficit/failed
responses
Vascular
Spinal column/cord
3
6
49
58
High-dose fentanyl etomidate

Ketamine propofol fentanyl
Propofol fentanyl
0
33 (2)a
14 (7)
16 (9)
0
100 (2)
57 (4)
67 (6)
Total
a
Number in parenthesis equals number of patients.

Table organized using format for Table 1 (Unpublished observations).
311
Table 6
Four-pulse transcranial magnetic stimulation (TCMS) technique: motor evoked potential (MEP) response
characteristics
Upper (n 25)
Take-off latency (ms)

Peak-to-peak amplitude (mV)
Area (mV ms)
Lower (n 17)
Mean
CV%
% Change
needed
Mean
CV%
% Change
needed
23.48
1388.2
6661.1
4.86
31.98
33.98
9.52
62.68
66.59
34.64
555.74
3641.0
3.52
32.61
36.69
6.89
63.91
71.91
Stimulation: quadropulse TCMS (100% output); coil: curved round coil (upper) and double cone coil (lower); MEP recording site: upper
limb ! first dorsal interossei; lower limb ! tibialis anterior; anesthetic: propofol, sufentanil, and oxygen; muscle relaxation: 4060%
(unpublished observations).
Table 7
Four-pulse transcranial magnetic stimulation (TCMS)
technique
Prediction
Cases monitored (n 54)
True positives
True negatives
False negatives
False negatives
2
51
1
0
Train TCMS predictions of motor outcomes in 51 patients (see

text) (Unpublished observations).
It is instructive to note that even with the use of methohexital, an anesthetic known to support seizure activity during anesthesia, this patient only demonstrated
seizures after emergence from anesthesia. This suggests that the characteristic anesthetic depression of
the central nervous system may be responsible for the
rare incidence of intraoperative seizure activity during
motor system monitoring. Intracranial metallic
devices (i.e., aneurysm clips, deep brain stimulators)
as well as electronic devices within a patients body
(i.e., pacemaker) represent an absolute contraindication to the use of magnetic transcranial stimulation
(Kandler, 1990; Krain et al., 1990; Pascual-Leone
et al., 1993; Chokroverty et al., 1995; Classen et al.,
1995).
It should be pointed out that TCMS has yet to be
approved by the FDA for routine clinical use. Therefore, the use of these stimulators requires an FDA
Investigational Device Exemption.
tc-MEP amplitude (mV)
1200
1000
800
600
400
200
0
Fig. 30. Comparison of lower limb motor evoked potentials (MEPs) using TCES and TCMS. Note that the TCES
derived response is larger than the response obtained with
TCMS. Calibration mark on right side of figure. Negativity
up. (Reprinted with permission from Ubags et al., 1999,
Anesth. Analg.)
Electrical
stimulation
Magnetic
stimulation
Fig. 31. Comparison of lower limb motor evoked potentials (MEPs) acquired with TCES (left) and TCMS (right)
using box plots. (Reprinted with permission from Ubags
et al., 1999, Anesth. Analg.)
312
L.D. GUGINO ET AL.

Magstim
QuadroPulse
Magnetic to Electrical
Stimulus Adapter
Secondary Coil
Voltage & Current
Limiting Devices
Primary Coil
To the Patient
Fig. 32. Diagram of the modification to the Quadropulse used for producing both train TCMS and TCES. Train TCMS is
produced without the magnetic to electrical stimulus adapter in the center of the figure. In order to produce train TCES, the
voltage drop is taken across the secondary coil of the adapter and applied between two scalp needle electrodes. Reprinted
with permission from Magstim Ltd., Wales, UK (1997).
MAGNETIC STIMULATION
ELECTRICAL STIMULATION
Anterior
3 mV
0 ms
80 ms
C7
C3-Electrical
Stimulation
Electrode
Circular
130 mm
Coil
Posterior
Patient: (Male)
Operation: C7-8 Foraminotomy
Anesthesia: Methohexital/Remifentanyl/O2
Muscle Relaxant: Vecuronium
Coil: Bent round coil
Magnetic coil placement: Posterior edge centered over Cz
Electrical Stimulation electrode placement: C3-C4
Stimulus: TCMS- Magstim Quadropulse
TCES- Magstim Quadropulse
C4-Electrical
+ Adapter
Stimulation Current polarity: Reversed sequentially
Electrode Power level: 100%
Inter Stimulus Pulse Intervel: 2.5 ms
Recording: Left and Right First Dorsal Interosse Muscle
Fig. 33. Example of typical upper limb motor evoked potentials (MEPs) acquired with train TCES (left side of figure) and
train TCMS (right side of figure). The curved round coil was used for eliciting magnetic motor responses whereas the
adapter (see Fig. 32) was used for stimulating across the C30 and C40 scalp needle electrodes. Negativity is up. Calibration
bar to the left for both response types (unpublished observations).
313
LOWER LIMB
MAGNETIC
ELECTRICAL
Right
Left
AL
AR
NR
R
AL
AR
AL
NR
R
NR
AR
AL
R
NR
R
AR
1000 V
10 msec
Right
Left
1000 V
1000 V
10 msec
10 msec
1000 V
10 msec
Anterior
NR = Not Reversed
R = Reversed
AR = Anode right
AL = Anode left
C3-Electrical
Stimulation
Electrode
()
100 mm
Double
Cone Coil
Posterior
C4-Electrical
Stimulation
Electrode
(+)
Patient: Female
Operation: T12-L1 Laminectomy
Excision spinal cord Tumor
Test Date: 02/16/99
Stimulus: Magstim Quadropulse 100%
Electrical 500 V
Coil: Double Cone Coil
Recording: Tibialis Anterior
Interpulse Interval: 2.5 msec
Coil position: Centered Over Cz
Fig. 34. Example of lower limb motor evoked potentials (MEPs) produced by TCES and TCMS. The figure uses the same
format as used for Fig. 33. The cone coil was used for acquiring train TCMS evoked tibialis anterior myogenic responses.
The adapter (see Fig. 32) was used for applying train TCES across C3 and C4 scalp needle electrodes. Voltage and time
calibration bars at the bottom of each response column. Negativity is up (unpublished observations).
20.9. Conclusion
Although experience has demonstrated that transcranial magnetic single pulse and train stimuli are capable of producing myogenic responses suitable for
intraoperatively monitoring the functional integrity

of descending motor paths, there is little doubt that
this application is better handled by transcranial electrical techniques. However, because magnetic stimulation is relatively painless compared to electrical
Table 8
Comparison of transcranial electrical stimulation (TCES)- and transcranial magnetic stimulation (TCMS)-induced
characteristics upper limb
TOL
PPA
AREA
TCES (N 7)
TCMS (N 7)
Anodal
Anterior directed current
Mean S.D.
CV%
Mean S.D.
CV%
Mean S.D.
CV%
22.8 0.68
3
3370.4 404
12
15,881 2,382
15
Mean S.D.
CV%
Mean S.D.
CV%
Mean S.D.
CV%
22.5 0.68
3
2973 743
25
13276.7 4115
31
Anesthesia: brevital propofol total intravenous anesthetic (TIVA); muscle relaxation: 4060%; TCMS 100%, interstimulus interval (ISI)
2.5 ms; TCES 500 V, ISI 2.5 ms (unpublished observations).
314
L.D. GUGINO ET AL.
Table 9
Comparison of transcranial electrical stimulation (TCES)- and transcranial magnetic stimulation (TCMS)-induced
characteristics lower limb
TOL
PPA
Area
TCES (N 2)
TCMS (N 1)
Anodal
Anterior directed current
Mean S.D.
CV%
Mean S.D.
CV%
Mean S.D.
CV%
30.3 0.90
3
1210.8 133
11
8840 1502
17
Mean S.D.
CV%
Mean S.D.
CV%
Mean S.D.
CV%
30.3 0.45
1.5
1050 346
33
6147 1844
30
Anesthesia: brevital propofol total intravenous anesthetic (TIVA); muscle relaxation: 4060%; TCMS 100%, interstimulus interval (ISI)
2.5 ms; TCES 500V, ISI 2.5 ms (Unpublished observations).
techniques, it is the stimulation technique of choice

for studying motor system function in conscious
patients (Barker et al., 1985, 1986; Pascual-Leone
et al., 1994; Gugino et al., 1998).
References
Aglio, LS, Kraus, KH, Desai, S, et al. (2002a) Efficacious
use of a cap shaped coil for transcranial magnetic stimulation of descending motor paths. Clin. Electroencephalogr., 33: 2129.
Aglio, LS, Romero, R, Desai, S, et al. (2002b) The use of
transcranial magnetic stimulation for monitoring descending spinal cord motor function. Clin. Electroencephalogr. Neurophysiol., 33(1): 3041.
Amassian, VE and Cracco, RQ (1987) Human cerebral cortical responses in contralateral transcranial stimulation.
Amassian, VE, Stewart, M, Quirk, GJ, et al. (1987a) Physiological basis of motor effect on a transient stimulation
to cerebral cortex. Neurosurgery, 20: 7493.
Amassian, VE, Stewart, M, Quirk, GJ, et al. (1987b) Pathological basis of motor effect on a transient stimulation
to cerebral cortex. Neurosurgery, 20: 7493.
Amassian, VE, Quirk, GJ and Sewart, M (1987c) Magnetic
coil versus electrical stimulation of monkey motor cortex. J. Physiol. (Lond.), 394: 119.
Amassian, VE, Quirk, GJ and Stewart, M (1990) A
comparison of corticospinal activation by magnetic
coil and electrical stimulation of monkey motor
cortex. Electroencephalogr. Clin. Neurophysiol., 77
(5): 390401.
Amassian, VE, Cracco, RO, Maccabee, PJ et al. (1991)
Matching focal and non-focal magnetic coil stimulation
to properties of human nervous system: mapping motor
unit fields in motor cortex contrasted with alternating
sequential digit movements by pre-motor SMA
stimulation. In: WJ Levy, AT Cracco and AT Barker

(Eds.), Magnetic Motor Stimulation Basic Principles
and Clinical Experience. Electroencephalography and
Clinical Neurophysiology (Supplement 43). Elsevier
Publishing Company, Amsterdam, pp. 328.
Amassian, VE, Eberle, I, Maccabee, PJ, et al. (1992) Modeling magnetic coil excitation of human cerebral cortex
with a peripheral nerve immersed in brain-shaped
volume conductor: the significance of fiber bending in
excitation. In: WJ Levy, AT Cracco, et al. (Eds.), Magnetic Motor Stimulation: Basic Principles and Clinical
Experience. Electroencephalography and Clinical Neurophysiology, Suppl. 85. Elsevier Publishing, Amsterdam, pp. 291301.
Barker, AT, Jalinous, R and Freeston, IL (1985) Noninvasive magnetic stimulation of the human motor cortex.
Lancet, 1: 11061107.
Barker, AT, Jalinous, R, Freeson, IL, et al. (1986) Clinical
evaluation of conduction time measurements in central
motor pathways using magnetic stimulation of human
brain. Lancet, 1: 13251326.
Beradelli, A, Inghiller, M, Cracco, G, et al. (1991)
Corticospinal potentials after electrical and magnetic
stimulation in man. In: WJ Levyan, AT Cracco and
AT Barker (Eds.), Magnetic Motor Stimulation: Basic
Principles and Clinical Experience. Electroencephaologr Clin Neurophysiol, Supplement 43. Elsevier,
Boyd, S, Rothwell, JC, Cowan, JMA, et al. (1986)
A method of monitoring function in cortical pathways
during scoliosis surgery with a note on motor conduction velocities. J. Neurol. Neurosurg. Psychiatry, 69:
251257.
Brouwer, B and Ashby, P (1990) Corticospinal projections to upper and lower limb spinal motoneurons in
man. Electroencephalogr. Clin. Neurophysiol., 76:
509519.

Brown, RH, Nash, CL, Jr. and Berilla, JA (1984) Cortical
evoked potential monitoring: a system for intraoperative
monitoring of spinal cord function. Spine, 9: 256261.
Burke, D and Hicks, R (1998) Surgical monitoring of
motor pathways. J. Neurophysiol., 15: 194205.
Burke, D, Hicks, RG and Stephen, JPH (1990) Corticospinal volleys evoked by anodal and cathodal stimulation of the human motor cortex. J. Physiol. (Lond.),
425: 283299.
Burke, D, Bartley, K, Woodforth, I, et al. (2000) The effects of
a volatile anesthetic on the excitability of human corticospinal axons. Brain, 123: 9921000.
Cadwell, J (1991) Optimizing magnetic stimulator design.
In: WJ Levy, AT Cracco and AT Barker (Eds.), Motor
Stimulaton: Magnetic Principles and Clinical Experience. Electroencephalogr Clin Neurophysiol Supplement 43. Elsevier Science Publishers, Amsterdam,
pp. 238248.
Chatrian, GE, Berger, MS and Wirch, AL (1988) Discrepancy between intraoperative SSEPs and postoperative
function. Case Report. J. Neurosurg., 69: 450454.
Chokroverty, S, Hening, W, Wright, D, et al. (1995) Magnetic brain stimulation: safety studies. Electroencephalogr. Clin. Neurophysiol., 97(1): 3642.
Classen, J, Witte, OW, Schlaug, G, et al. (1995) Epileptic
seizures triggered directly by focal transcranial magnetic stimulation. Electroencephalogr. Clin. Neurophysiol., 94(1): 1925.
Claus, D (1991) Magnetic stimulation: technical aspects.
Electroencephalogr. Clin. Neuorphysiol., 43: 249254.
Cohen, D and Cuffin, BN (1991) Developing a more focal
magnetic stimulator. Part I. Some basic principles.
Cohen, LG, Roth, BJ, Nilsson, J, et al. (1990) Effects of coil
design on delivery of focal magnetic stimulation. Technical considerations. Electroencephalogr. Clin. Neurophysiol. Suppl., 75(4): 350357.
Crawford, ES, Mizrahi, EM, Hess, KR, et al. (1988) The
impact of distal aortic perfusion and somatosensory
evoked potential monitoring on prevention of paraplegia
after aortic aneurysm operation. J. Thorac. Cardiovasc.
Surg., 95: 357367.
Cushing, H (1909) A note upon the faradic stimulation of
the post-cerebral gyrus in conscious patients. Brain,
32: 4453.
Day, BL, Dick, JPR, Marsden, CD, et al. (1986) Differences
between electrical and magnetic stimulation of the human
brain. J. Physiol. (Lond), 378: 36.
Day, BL, Maertens de Noordhout, A, Marsden, CD, et al.
(1987a) A comparison of the effects of cathodal and anodal
stimulation of the human cortex through the intact scalp.
J. Physiol. (Lond), 394: 118.
Day, BL, Dressler, D, Maertens de Noordhout, A, et al.
(1987b) Different sites of action of electrical and
315
magnetic simulation of the human brain. Neurosci. Lett.,
75(1): 101106(MEDLINE) related records.
(1988) Differential effect of cutaneous stimuli on
responses to electrical or magnetic stimulation of the
human brain. J. Physiol. (Lond), 399: 68.
(1989) Electric and magnetic stimulation of the human
motor cortex: surface EMG and single motor unit
responses. J. Physiol. (Lond), 412: 449473.
DeHann, P, Kalkman, CJ, Ubags, LH, et al. (1996)
A comparison of the sensitivity of epidural and myogenic transcranial motor evoked responses in the detection of acute spinal cord ischemia in the rabbit. Anesth.
Analg., 83: 10221027.
Deletis, V (1993) Intraoperative monitoring of the functional integrity of the motor pathways. In: O Devinksi,
A Berie and M Dogali (Eds.), Electrical and Magnetic
Stimulation of the Brain and Spinal Cord. Advances in
Neurology. Raven Press, New York, Chapter 17,
Vol. 6, pp. 201214.
Edgley, SA, Eyre, JA, Lemon, RN, et al. (1990) Excitation
of the corticospinal tract by electromagnetic and electrical stimulation of the scalp in the macque monkey.
J. Physiol. (Lond.), 425: 301320.
Edmonds, HL, Poloheimo, MP, Backman, MH, et al. (1989)
Transcranial magnetic motor evoked potentials
(tcMMEP) for functional monitoring of motor pathways
during scoliosis. Spine, 14: 683686.
Elmore, JR, Glaviczki, P, Harper, CM, et al. (1991) Failure
of motor evoked potentials to predict neurological outcome in experimental thoracic aortic occlusions.
J. Vasc. Surg., 14(2): 131139.
Epstein, CM, Schwartzberg, DA, Dandy, KE, et al. (1990)
Localizing the site of magnetic brain stimulation in
humans. Neurology, 40: 666670.
Fujiki, M, Isono, M, Hori, S, et al. (1996) Corticospinal direct
response to transcranial magnetic stimulation in humans.
Ghaly, RF, Stone, JL, Levy, WJ, et al. (1990a) The effects
of nitrous oxide on transcranial magnetic induced
electromyographic responses in the monkey. J. Neuurosurg. Anesth., 2: 175181.
Ghaly, RF, Stone, JL, Levy, WJ, et al. (1990b) The effect of
etomidate on transcranial magnetic-induced motor
evoked potentials in the monkey. Neurosurgery, 20: 184.
Ghaly, RF, Stone, JL, Levy, WJ, et al. (1990c) The effect
of etomidate on motor evoked potentials induced by
transcranial magnetic stimulation in the monkey. Neurosurgery, 27: 936942.
Ghaly, RF, Stone, JL, Levy, WJ, et al. (1991a) The effect
of etomidate or midazolam hypnotic doses on motor
evoked potentials in the monkey. J. Neurosurg. Anesth.,
3: 2027.
316
Ghaly, RF, Stone, JL, Levy, WJ, et al. (1991b) The effect
of neuroleptanalgesia (Droperidol-Fentanyl) on motor
potentials evoked by transcranial magnetic stimulation
in the monkey. J. Neurosurg. Anesth., 3: 117123.
Ghaly, RF, Stone, JL, Levy, WJ, et al. (1991c) The effect
of an anesthetic induction dose of midazolam on motor
evoked potentials evoked by transcranial magnetic stimulation in the monkey. J. Neuroanesth., 20: 183.
Ghaly, RF, Stone, JL, Aldrete, A, et al. (2000) Motor
evoked potentials in primates, anesthetic considerations.
J. Neuroanesth., 182: 199.
Gilman, S and Marco, TA (1971) Effects of medullary pyramidotomy in the monkey. Clinical and electromyography
abnormalities. Brain, 94: 495514.
somatosensory evoked potentials: a case report. J. Neurosurg., 63: 296300.
Glassman, SD, Zhang, YP, Shields, CB, et al. (1996)
Transcranial magnetic motor evoked potentials in scoliosis surgery. Orthopedics, 18: 10171023.
Gugino, V and Chabot, RJ (1990) Somatosensory evoked
potentials. In: L Kearse (Ed.), International Anesthesiology
Clinics. Little Brown and Company, Boston. Vol. 28(3),
pp. 154164.
Gugino, LD, Kraus, K, Heino, R, et al. (1992) Peripheral
ischemia as a complicating factor during somatosensory
and motor evoked potential monitoring of aortic surgery.
J. Cardiovasc. Anesth., 6(6): 715719.
Gugino, LD, Aglio, LS, Segal, ME, et al. (1997) Transcranial Magnetic Stimulation for Monitoring Spinal Card
Motor Paths. Semin. Spine Surg., 9(4): 315336.
Gugino, LD, Potts, G, Aglio, LS, et al. (1998) Localization of
eloquent cortex using transcranial magnetic stimulation.
In: E Alexander and RJ Maciunas (Eds.), Advanced Neurosurgical Navigation. Thieme Publishers, New York.
Haghighi, SS, Madsen, R, Green, D, et al. (1990) Suppression of motor evoked potentials by inhalational anesthetics. J. Neurosurg. Anesth., 2: 7578.
Heneghan, C (1993) Clinical and medicolegal aspects of
conscious awareness during anesthesia. In: J Gareth
(Ed.), International Anesthesiology Clinics Depth of Anesthesia. Little Brown and Company, Boston, Vol. 31(4),
pp. 112.
Hicks, DBR, Gandevia, SC, Stephen, J, et al. (1993) Direct
comparison of corticospinal volleys in human subjects
to transcranial magnetic and electrical stimulation.
J. Physiol. (Lond), 470: 383392.
Homberg, V and Netz, J (1989) Generalized seizures
induced by transcranial magnetic stimulation of motor
cortex (letter). Lancet, 2: 1223.
Hufnagel, A, Elger, CE, Durwen, HF, et al. (1990) Activation of the epileptic focus by transcranial magnetic stimulation of the human brain. Ann. Neurol., 27: 4960.
L.D. GUGINO ET AL.

Jackson, JD (1963) Magnetostatics. In: JD Wiley (Ed.),
Classical Electrodynamics. John Wiley and Sons, New
York, pp. 132168.
Jalinous, R (1991) Technical and practical aspects of magnetic nerve stimulation. J. Clin. Neurophysiol., 8(1):
1025.
John, ER, Chabot, RJ, Prichep, LS, et al. (1988) Intraoperative monitoring during neurosurgical and neuroradiological procedures. J. Clin. Neurophysiol., 6: 125158.
Kalkman, CJ, Drummond, JC, Patel, PM, et al. (1992)
Effects of droperidol, pentobarbital and ketamine on
myogenic transcranial motor evoked responses in
humans. Anesthesiology, 77(3A): A163.
Kalkman, CJ, Been, HD, Ubags, JH, et al. (1993) Improved
amplitude of intraoperative myogenic evoked responses
after paired and transcranial electrical stimulation.
Anesthesiology, 79(3A): A176.
Kandler, R (1990) Safety of transcranial magnetic stimulation (letter). Lancet, 335: 469470.
Kawaguchi, M and Furuya, H (2004) Intraoperative spinal
cord monitoring of motor function with myogenic
motor evoked potentials: a consideration in anesthesia.
J. Anesth., 18: 1828.
Krain, L, Kimura, J, Yamada, T, et al. (1990) Consequences or cortical magnetoelectric stimulation. In:
S Chokroverty (Ed.), Magnetic Stimulation in Clinical
Neurophysiology. Buttersworth Publishing Company,
Boston, pp. 157163.
Kraus, KH, Gugino, LD, Levy, WJ, et al. (1992) The use of
a cap shaped coil for transcranial magnetic stimulation
of the motor cortex. J. Clin. Neurophysiol., 10(3):
353362.
Lawrence, DG and Hopkins, DH (1972) Development
aspects of pyramidal control in the rhesus monkey.
Brain Res., 40: 117118.
Lawrence, DG and Kuypers, HG (1968) The functional organization of the motor systems in the monkey. I.
The effects of bilateral pyramidal lesions. Brain, 91(1):
114.
Levy, WJ (1987) Transcranial stimulation of the motor cortex to produce motor evoked potentials. Med. Instrum.,
21: 248254.
Levy, WJ and York, DH (1983) Evoked potentials from the
motor tracts in humans. Neurosurgery, 12: 422429.
Levy, WJ, York, DH, McCaffrey, M, et al. (1984) Motor
evoked potentials from transcranial electrical stimulation of the motor cortex in humans. Neurosurgery, 15:
287302.
Losasso, TJ, Boudreaux, JB, Muzzi, DA, et al. (1991) The
effect of anesthetic agents on transcranial magnetic motor
evoked potentials (TMEP) in neurosurgical patients.
J. Neurosurg. Anesthesiol., 3(3): 200 (Abstract).
Lotto, M, Banoub, N and Schubert, A (2004) Effects and
anesthetic agents on physiological changes on

intraoperative motor evoked potentials. J. Neurosurg.
Anesth., 16(1): 3242.
Maccabee, PJ, Amassian, VE, Eberle, LP, et al. (1993)
Magnetic coil distribution of straight and bent amphibian and mammalian peripheral nerve in vitro: locus of
excitation. J. Physiol. (Lond.), 460: 201219.
McCaffrey, M (1997) Somatosensory evoked potential
monitoring during spinal surgery. In: J Owens and L
Lenke (Eds.), Seminars in Spine Surgery. WB Saunders
Company, Philadelphia, Pennsylvania, Vol. 4, pp.
309314.
McPherson, RW (1994) General anesthetic considerations in
intraoperative monitoring: effects of anesthetic agents
and neuromuscular blockade on evoked potentials, EEG
and cerebral blood flow. In: CM Loftus (Ed.), Intraperative Monitoring Techniques in Neurosurgery. McGraw
Hill Publishers, New York, pp. 97106.
Nadstawek, J, Rechstein, U, Taniguchi, M, et al. (1993)
Repetitive transcranial electrical stimulation for myogenic motor evoked potentials under balanced anesthesia (BA) with isoflurane, nitrous oxide, and alfentanil
versus total intravenous anesthesia (TIVA) with propofol and alfentanil. Anesthesiology, 79(3A): A461.
Nuwer, MR, Dawson, EG, Carlsen, LG, et al. (1995)
Somatosensory evoked potential spinal cord monitoring
reduces neurologic deficits are scoliosis surgery: results
of a large multicenter survey. Electroencephalogr. Clin.
Pascual-Leone, A, Houser, CM, Reese, K, et al. (1993)
Safety of rapid-rate transcranial magnetic stimulation
in normal volunteers. In: WJ Levy and WJ Cracco
(Eds.), Magnetic Motor Stimulation: Basic Principles
and Clinical Experience. Electroencephalography and
Clinical Neurophysiology, Suppl. 43. Elsevier Publishing Company, Amsterdam, pp. 120130.
Pascual-Leone, A, Sole, A, Wasserman, EM and Hallett, M
(1994) Responses to rapid-rate transcranial magnetic
stimulation of the human motor cortex. Brain, 117:
568572.
Pechstein, U, Cedzich, C, Nadstawek, J, et al. (1996) Cranial high frequency repetitive electric stimulation for
recording myogenic motor evoked potentials with the
patient under general anesthesia. Neurosurgery, 39:
335344.
Pereon, Y, Bernard, JM, Nguyen, S, et al. (1999) The
effects of desflurane on the nervous system: from spinal
cord to muscles. Anesth. Analg., 89: 490495.
Phillips, CG (1969) Motor apparatus of the baboons hand.
Proc. R. Soc. Lond. B Biol. Sci., 173(31): 141174.
Rampil, IJ (1993) Is MAC testing a spinal reflex?. Anesthesiology, 79(3A): A422.
Rampil, IJ (1994) F waves a nonsynaptic but sensitive
indicator of anesthetic effect in rats. Anesth. Analg., 78:
S1S503.
317
Rampil, IJ and Laster, MJ (1992) No correlation between
quantitative electroencephalographic measurements
and movement response to noxious stimuli during isoflurane anesthesia in rats. Anesthesiology, 77:
920925.
Rampil, IJ, Mason, P and Singh, H (1993) Anesthetic
potency (MAC) is independent of forebrain structures
in the rat. Anesthesiology, 78: 707712.
Roth, BJ, Cohen, LG and Hallet, M (1991a) The electric
field induced during magnetic stimulation. Electroencephalogr. Clin. Neurophysiol. Suppl., 43: 268278.
Roth, BJ, Saypol, JM, Hallett, M, et al. (1991b) A theoretical
calculation of the electrical field induced in the cortex by
magnetic stimulation. Electroencephalogr. Clin. Neurophysiol., 81: 4756.
Rothwell, JC (1991) Physiological studies of electric and
magnetic stimulation of the human brain. In: WJ Levy,
RW Cracco and AT Barker (Eds.), Magnetic Motor
Stimulation: Basics Principles and Clinical Experience.
Electroencephalography and Clinical Neurophysiology
(Supplement 43). Elsevier Publishing, Amsterdam,
pp. 2935.
Rothwell, JC, Thompson, PD, Day, BL, et al. (1991) Stimulation of the human motor cortex through the scalp.
Exp. Physiol., 76: 159200.
Schwinn, DA, Watkins, W and Leslie, JB (1994) Basic
principles of pharmacology related to anesthesia. In:
RD Miller (Ed.), Anesthesia. Churchill Livingstone,
New York, 4th ed., p. 43.
Sloan, TB (1990) The vecuronium alters cortical magnetic
motor evoked potentials. J. Neurosurg. Anaesthesiol.,
2(3): 251.
Sloan, T and Heyer, E (2002) Anesthesia for intraoperative
neurophysiologic monitoring of the spinal cord.
Sloan, TB and Levin, D (1991) Effect of enflurane, halothane, isoflurane and nitrous oxide on cortical magnetic
motor evoked potentials. J. Neurosurg. Anesthesiol.,
3(3): 201(Abstract).
Stanski, D (1994) Monitoring depth of anesthesia. In: RD
Miller (Ed.), Anesthesia. Churchill Livingstone, New
York, 4th ed., p. 1127.
Stinson, W, Murray, MJ and Jones, KA (1994) A computercontrolled, closed-loop infusion system for infusing
muscle relaxants: its use during motor evoked potential
monitoring. J. Cardiothorac. Vasc. Anesth., 8: 4044.
Taniguichi, M, Cedzich, C and Schramm, J (1993) Modification of cortical stimulation for motor evoked potentials under general anesthesia: technical description.
Thornton, C and Jones, JG (1993) Evaluating depth of anesthesia: review of methods. In: J Gareth (Ed.), International
Anesthesiology Clinics Depth of Anesthesia. Little Brown
and Company, Boston, Vol, 13(4), pp. 6768.
318
Tung, HL, Drummond, JC and Bickford, RG (1988) The
effects of anesthetic and sedative agents on magnetic motor
evoked responses. Anesthesiology, 69: 313A(Abstract).
Ubags, LH, Kalkman, CJ, Been, HD, et al. (1997) The use of
ketamine on etomidate to supplement sufentanil/N20
anesthesia does not disrupt monitoring of myogenic transcranial motor evoked responses. J. Neurosurg. Anesthesiol., 9(3): 228233.
Ubags, LH, Kalkman, CJ, Been, HD and Kaelman, JH (1999)
A comparison of myogenic motor evoked responses to
electrical and magnetic transcranial stimulation during
nitrous oxide/opioid anesthesia. Anesth. Analg., 88:
568572.
Watt, JWH, Fraser, MH, Soni, BM, et al. (1996) Total IV
anaesthesia for transcranial magnetic evoked potential
spinal cord monitoring. Br. J. Anaesth., 76: 870871.
Woodforth, IJ, Hicks, RG, Crawford, MR, Stephen, JPH
and Burke, D (1999) Depression of T waves in corticospinal volleys by sevoflurane, thiopental and propofol. Anesth. Analg., 89: 11821187.
Yamada, H, Torres, F, Tarnaki, T, et al. (1994) Effects of
halothane, isoflurane, and enflurane on motor potentials
evoked by transcranial magnetic stimulation in cats.
Anesth. Analg., 78: S1S492.
Yanokuchi, K and Cohen, D (1991) Developing a more
focal magnetic stimulator. Part II. Fabricating coils
and measuring induced current distributions. J. Clin.
York, D (1997) A review of neurophysiology monitoring of
spinal cord function in relation to changes in the marketplace. In: J Owens (Ed.), Seminars in Spine Surgery.
W.B. Saunders Company, Philadelphia, Pennsylvania,
Vol. 9(4), pp. 295301.
L.D. GUGINO ET AL.

Zentner, J (1989a) Influence of anesthetics on the electromyographic response evoked by transcranial electrical
cortex stimulation. Funct. Neurol., 4: 299300.
Zentner, J (1989b) Noninvasive motor evoked potential
monitoring during neurosurgical operations on the
spinal cord. Neurosurgery, 23: 709712.
Zentner, J and Ebner, A (1989) Nitrous oxide suppresses
the electromyographic response evoked by electrical
stimulation of the motor cortex. Neurosurgery, 24:
6062.
Zentner, J and Rieder, G (1990) Diagnostic significance of
motor evoked potentials in space-occupying lesions of
the brainstem and spinal cord. Eur. Arch. Psychiatry
Clin. Neurol. Sci., 239: 285289.
Zentner, J, Schumacher, M and Bien, S (1988) Motor
evoked potentials during interventional neuroradiology.
Neuroradiology, 30: 252255.
Zentner, J, Kiss, I and Ebner, A (1989) Influence of
anesthetics nitrous oxide in particular an
electromyographic response evoked by transcranial
electrical stimulation of the cortex. Neurosurgery,
24: 253256.
Zhou, H and Zhea, C (2000) Comparison of isoflurane
effects on motor evoked potential and F wave. Anesthesiology, 93: 3238.
Zhou, HH, Mehta, M and Leis, A (1997) Spinal
cord motoneuron excitability during isoflurane and
nitrous oxide anesthesia. Anesthesiology, 86:
302307.
Zornow, MH, Grafe, MR, Tybor, C, et al. (1990) Preservation of evoked potentials in a case of anterior spinal
aneurysm syndrome. Electroencephalogr. Clin. Neurophysiol., 77: 137139.

M.R. Nuwer (Ed.)
319
CHAPTER 21
Mapping the corticospinal tract

Charles D. Yingling*
Department of Otolaryngology/Head and Neck Surgery, Stanford School of Medicine and
California Neuromonitoring Services, San Francisco, CA 94127, USA
21.1. Anatomy of corticospinal tract

The corticospinal tract (CST), or the pyramidal tract,
is generally considered to be the primary pathway
mediating voluntary movements, particularly fine
motor control of distal musculature. In addition to
providing direct control of distal muscles, the CST
also regulates subcortical motor centers through a
combination of excitatory and inhibitory influences.
In humans, it has become the most important motor
pathway, and damage can have devastating effects
on quality of life. Thus, for decades, surgeons have
sought ways to identify and avoid the CST during
operations on the brain or on the spinal cord near this
critical pathway.
21.1.1. Cortical anatomy
It is well known that the CST primarily originates in the
precentral gyrus (Brodmanns area 4). It is often less
appreciated that phylogenetically, the CST initially
arose from the postcentral (sensory) gyrus and projected
primarily to sensory areas in the dorsal horns of the
spinal cord. In humans, about 2/3 of the CST now arises
from the precentral region (including Brodmanns
areas 4 and 6), but about 1/3 still originates in the postcentral areas of the anterior parietal lobe (Kahle and
Frotscher, 2003). Little is known about the functional
significance of this parietal contribution to the CST,
but it probably plays a role in regulating the sensory
consequences of voluntary movement.
The organization of the cortical motor area in the
precentral gyrus was established in the 1930s by
the pioneering work of Penfield and Boldrey (1937),

working in collaboration with Herbert Jasper, one of
the first clinical neurophysiologists to play a major
role in the operating room. Their well-known motor
homunculus still forms the basis for our understanding of cortical motor anatomy, and is based on functional, rather than anatomical topography, being
devoted roughly 1/3 to the face, 1/3 to the hand, and
1/3 to the entire rest of the body. The face is represented most laterally, with the hand more medial and
the trunk and leg area folding around the central fissure
so that the foot area lies on the medial surface of each
hemisphere. Microelectrode studies in animals have
shown that there are in fact multiple representations
of the body in both motor and sensory cortices; however, at the relatively gross level of intraoperative
stimulation and recording, the original conceptualization of a single homunculus is probably sufficient.
The CST consists of the axons of pyramidal cells
located in layer 5 of the cerebral cortex. While
it is often thought that the CST originates from the
so-called giant Betz cells, in fact these contribute only
some 23% of CST axons, with the remainder coming
from smaller pyramidal cells, about 40% of which are
unmyelinated. Since axonal conduction velocity is a
function of both size and myelination, there is a wide
variation of conduction speeds within the CST. While
gross movements may be mediated by a range of axons
of different sizes, it is likely that electrophysiological
recordings are dominated by the fastest conducting
cells, which would produce the most synchronous and
thus largest amplitude responses.
21.1.2. Internal capsule
Correspondence to: Charles D. Yingling, Ph.D., Stanford

School of Medicine and Surgical Neuromonitoring
Services, 1001 Bridgeway #434, Sausalito, CA 94965,
USA.
Tel.: 1-415-332-6763; fax: 1-415-665-1620.
E-mail: cy@brainmon.com (C.D. Yingling).
The descending axons from the cortical motor areas,

along with the ascending thalamocortical fibers,
together form the fan-shaped corona radiata, which
more deeply narrows to become the internal capsule.
In the horizontal section, the internal capsule forms
320
an angle with an anterior limb, which lies between

the head of the caudate nucleus medially and the pallidum and putamen laterally, and a posterior limb
which is bordered medially by the thalamus and laterally by the pallidum and putamen. The genu of
the internal capsule lies between the two limbs. The
CST lies mainly posterior to the genu, and maintains
a somatotopic organization with the head fibers anterior, and hand, trunk, and leg progressively more
posterior. Corticorubro and corticotegmental fibers
are mixed with those of the CST in this area, with
the thalamocortical fibers more posterior in the
caudal portion of the internal capsule.
21.1.3. Brainstem
At the transition to the midbrain, the fibers of the
CST together with the corticopontine tracts form
the cerebral peduncles, which occupy the ventrolateral parts of the mesencephalon. Fibers from the
parietal cortex lie the most laterally, followed in a
somatotopic sequence by CST projections to the
lower limb, trunk, upper limb, and most medially
corticobulbar fibers for the face region. As they pass
lower into the ventral pons, the bundles rotate so that
the corticobulbar fibers lie dorsally, with cervical,
thoracic, lumbar, and sacral projections at progressively more ventral locations.
In the medulla, the corticonuclear fibers terminate
on the various cranial nerve nuclei. Note that some
of these projections are bilateral (oculomotor nucleus
III, motor trigeminal nucleus V, caudal part of facial
nucleus VII projecting to forehead muscles, and
nucleus ambiguus X); others cross to the contralateral nuclei (abducens VI, rostral part of facial VII
projecting to nonforehead muscles, and hypoglossal
XII); and the ipsilateral projection to the trochlear
nucleus (IV) is uncrossed. Also at the level of the
medulla, the fibers of the CST partially cross to the
contralateral side in the decussation of the pyramids.
From 70 to 90% of the fibers cross, those to the
upper limb crossing dorsally to those for the lower
limb.
C.D. YINGLING
and cross to the contralateral side only at the segmental level where they terminate. The ventral tract is
more variable in size and may be asymmetrical or
completely absent, and only extends as far as the
cervical or the thoracic region.
Only a small percentage of the CST fibers directly
synapse on anterior horn motor cells, with the majority terminating on interneurons in the intermediate
zone between the dorsal and ventral horns. Within
the anterior horn, distal portions of the extremities
are represented most laterally, which is also where
most of the fibers that reach the anterior horn terminate. The classical direct pathway from the motor
cortex to alpha motor neurons is thus a relatively
small portion of the total CST, and is primarily
involved in fine movements of distal extremities.
Finally, the CST neurons primarily activate neurons supplying flexor muscles and inhibit those
supplying extensors. Those CST fibers originating
in the parietal lobe end in the dorsal column nuclei
(nucleus gracilis and nucleus cuneatus) and the
substantia gelatinosa of the dorsal horn, where they
presumably regulate sensory inputs.
It should be apparent from this brief review that
the CST is much more complex than the classical
picture of upper motor neurons arising in the precentral motor cortex and terminating contralaterally
on lower motor neurons in the ventral horn of the
spinal cord. There are polysynaptic as well as monosynaptic connections, inhibitory as well as excitatory
connections, ipsilateral as well as contralateral
projections, and projections to sensory as well as to
motor systems. Furthermore, there are numerous
other descending tracts within the brainstem and
spinal cord (vestibulospinal, reticulospinal, tegmentospinal, rubrospinal, tectospinal) which often run in
close proximity to the fibers of the CST proper and
are thus likely to be coactivated with the relatively
gross stimulation techniques available during surgery. It is good to keep this complexity in mind when
attempting to interpret the sometimes puzzling results
of intraoperative stimulation and recording.
21.1.4. Spinal cord
21.2. Mapping at the cortical level
The crossed pyramidal fibers form the lateral CST, in

which fibers destined for the cervical region (more
than half the total) run most medially, with thoracic,
lumbar, and sacral fibers found progressively more
laterally. The uncrossed fibers form the ventral CST
Mapping the sensory and motor systems at the cortical level is covered in detail in other chapters in
this volume (Chapters 8 and 14) and thus will only
be considered briefly here before concentrating on
subcortical and spinal mapping.
While the location of the central sulcus can be

inferred from anatomical MRI scans, and more
directly visualized with functional MRI, the variability from one individual to another in the location of
the central sulcus is such that direct intraoperative
localization is highly preferable. This can be done
in two complementary ways (assuming for this
discussion a procedure performed under general
anesthesia): mapping the polarity reversal of the
somatosensory evoked potential (SEP), and direct
electrical stimulation of the exposed cortical surface
to elicit motor responses.
21.2.1. SEP polarity reversal
The earliest cortical component of the SEP to median
or ulnar nerve stimulation is generated by pyramidal
cells in the postcentral gyrus that lie in the depths
of the central sulcus and are thus oriented parallel
to the cortical surface (Desmedt and Cheron, 1982).
At the peak of the earliest response (known as
N20), the electrical dipole created by currents from
the pyramidal cells is negative posteriorly and positive anteriorly. Thus, a strip electrode placed perpendicular to the central sulcus, and over the hand
cortical area, will record positive potentials anteriorly, and at the same latency negative potentials
posteriorly (Wood et al., 1988). The central sulcus
will lie between the two closest contacts producing
opposite polarity responses. It is important to assess
this polarity reversal at the latency of the earliest cortical peak, since there are responses of slightly longer
latency (P23) which are generated in the precentral
gyrus by pyramidal cells oriented perpendicularly to
the cortical surface and thus do not exhibit polarity
reversals. The potentials recorded at the cortical
surface (and scalp) represent a spatiotemporal summation of these and other active sites, and can complicate interpretation if care is not taken to correctly
identify the earliest cortical response.
21.2.2. Stimulation mapping of the cortical surface
21.2.2.1. Stimulation with prolonged trains
This basic technique for mapping the motor cortex is little changed from the original method of Penfield and
Boldrey (1937), and has been extensively described in
the modern literature by Ojemann, Berger, and others
(Berger, 1995; Berger and Ojemann, 1999; Duffau
et al., 2003). Briefly, the exposed cortex is stimulated
with a bipolar electrode, typically using 1 s trains at
321
50 Hz with a 1-ms pulse width and biphasic pulses at

intensities ranging from 3 to 8 mA, and assessing any
resulting contralateral motor responses. Originally, the
patient was observed for gross motor responses and
the location of such responses reported to the surgeon,
who could then mark responsive sites with sterile tags.
Yingling et al. (1999) demonstrated that addition of
electromyographic (EMG) recordings increases the
sensitivity of the technique, as many times stimulation
elicits clear EMG responses without any visible movement. (Since it is not technically feasible to record from
every contralateral muscle, movements are occasionally seen without corresponding EMG activity, but
much less frequently).
While this technique is easily implemented and is
of undoubted utility, it is not without its drawbacks.
First, the stimulation tends to be epileptogenic, with
intraoperative seizures elicited in up to 27% of cases
(Yingling et al., 1999) despite use of anticonvulsants
and even in patients with no preoperative seizure
history. While these can be readily terminated with
direct irrigation of the cortex with cold Ringers
solution (Sartorius and Berger, 1998), it is better to
avoid intraoperative seizures entirely. Second, the
technique may be less sensitive in pediatric patients
due to the relative inexcitability of the motor cortex
(Sala et al., 2002). Finally, this technique is a
mapping and not a monitoring technique, and thus
does not continuously assess the function status of
the motor pathways.
21.2.2.2. Stimulation with brief pulse trains and

time-locked EMG recording
In contrast to intermittent mapping, continuous
monitoring of the motor pathways can be achieved
by either transcranial or direct cortical multipulse
electrical stimulation (Sala et al., 2002). Transcranial
stimulation is simply the now well-established technique of transcranial electrical motor evoked potentials (tceMEP), which is covered in detail elsewhere
in this volume (Chapters 15, 16, 17, and 18). While
this technique can provide relatively continuous
monitoring of motor pathways, its use may be limited
due to excessive movement and/or inability to place
the surface stimulating electrodes over the appropriate cortical area (which is often within the zone of
the craniotomy for surgery near the central sulcus).
While suboptimal electrode placements can sometimes be used to elicit MEPs at higher stimulus
levels, such higher currents can activate neurons or
322
fiber tracts at deeper levels and potentially bypass the

region of concern. (This is less of an issue for monitoring spinal cord function, but in such cases there
are usually no constraints on scalp electrode
placement.)
The alternate to transcranial stimulation is direct
cortical stimulation once the motor cortical area has
been identified, with time-locked recording of muscle
responses. D-waves from the spinal cord can also be
recorded, but this requires the additional insertion of
an epidural spinal electrode (Yamamoto et al.,
2004).While multipulse trains can be delivered with
a handheld electrode for mapping the motor cortex,
and have the advantage of a lower likelihood of
inducing seizures (Cedzich et al., 1996), the real
advantage of this technique is in delivering stimulation via strip electrodes left in place over the motor
cortex. With this method, the motor pathways can
be monitored continuously during surgery near the
motor cortex and/or subcortical motor pathways.
We have been able to utilize this technique with automatic delivery of pulse trains every 2 s, giving almost
instantaneous feedback concerning the integrity of
the CST (Yingling and Heit, unpublished observations). Further discussion and details concerning this
method can be found in Sala et al. (2002).
21.3. Mapping the internal capsule and brainstem
21.3.1. Subcortical stimulation with
prolonged trains
The original Penfield and Boldrey (1937) technique
of stimulation with prolonged 50 Hz trains while
looking for movement (or EMG responses) has been
similarly applied to subcortical stimulation (Skirboll
et al., 1996; Berger and Ojemann, 1999; Yingling
et al., 1999; Duffau et al., 2003; Keles et al., 2004).
Since the cortex is not being directly activated, there
is very little risk of inducing seizures with deep stimulation, although it is still theoretically possible as
thalamocortical or corticocortical fibers may also be
activated and secondarily induce cortical seizure
activity. Just below the cortical level, the topography
of the descending CST mirrors that of the cortical
homunculus and thus responses tend to be restricted
to a specific part of the body. However, the deeper
the dissection proceeds, the closer the fibers projecting to different limbs becomes, and the results
of stimulation will be less predictable, particularly
if the normal topography has been significantly
C.D. YINGLING
distorted by pathological processes. In this situation,

the use of multichannel EMG recordings is most
valuable, as it is easier to scan multiple traces on
a single display screen than to observe the entire
contralateral body simultaneously (Yingling et al.,
1999). In a large series reported by Keles et al.
(2004), patients whose subcortical pathways were
identified with stimulation mapping were more prone
to develop additional temporary or permanent motor
deficits than those in whom subcortical pathways
could not be identified, implying that the dissection
had proceeded to the point where the CST was compromised. This result suggests that a more sensitive
method for detecting CST fibers before they are
disrupted is desirable.
A major unresolved issue is the relationship
between the intensity of stimulation and the distance
to the portion of the CST being stimulated. The stimulus intensity described in several published reports
(Berger, 1995; Berger and Ojemann, 1999; Yingling
et al., 1999; Duffau et al., 2003; Keles et al., 2004)
has ranged from 0.5 to 20 mA. (The figure of
20 mA probably represents a maximum current at
any given instant of 10 mA, due to a confusion in
terminology concerning the biphasic output of the
Ojemann cortical stimulator produced by Radionics.)
This wide range of current is probably due, at least
in part, to different levels of general anesthesia.
Although subcortical stimulation may activate the
CST directly, rather than secondarily via activation
of cortical networks depressed by general anesthesia,
a movement or EMG response still requires, at minimum, synaptic transmission onto the lower motor
neuron in the ventral horn of the spinal cord, which
as part of the central nervous system is also subject
to the depressant effects of anesthetic agents (Rampil
and King, 1996).
A second consideration concerning the relationship between threshold and distance to the CST is
the different effects of monopolar versus bipolar
stimulation. Schekutiev and Schmid (1996) convincingly demonstrated that bipolar stimulation with a
concentric electrode was more specific than monopolar stimulation with a focal cathode and remote
anode, using the same stimulus parameters. Haglund
et al. (1993), using optical imaging techniques in
exposed monkey cortex, demonstrated high spatial
specificity for bipolar stimulation of the cortex at
intensities of up to 4 threshold, consistent with
the observations of Ojemann (1991) that disruption
of language by cortical stimulation is limited to small
regions of cortex. On the contrary, Kombos et al.

(1999) claimed that monopolar stimulation produced
better localization in primary motor cortex, while
bipolar stimulation was better for localizing functional
areas in premotor cortex. However, these results are
not directly comparable, since these authors used
different stimulus parameters for monopolar versus
bipolar stimulation. While bipolar stimulation will
produce less spread of current at the same intensity, it
is also more sensitive to the orientation of the bipolar
tips with respect to the axis of the fibers being
stimulated.
A new MRI imaging technique, diffusion-tensor
imaging (DTI), may provide a means of resolving
these issues. DTI is based on the fact that in white
matter, water diffuses primarily along the longitudinal axis of fiber tracts. Berman et al. (2004) used
DTI to visualize the subcortical course of fibers
originating in cortical motor areas identified by intraoperative stimulation (Fig. 1). Combined with imageguided surgical navigation technology, it may soon
become possible to directly visualize subcortical
CST fiber tracts in relation to the precise location
Fig. 1. Three-dimensional diffusion-tensor imaging (DTI)

MRI showing corticospinal tracts (CSTs) adjacent to a precentral glioma. The image shows the fiber tracts originating
from two cortical sites that produced wrist and shoulder
movements with intraoperative stimulation, and can be
seen twisting around one another as they descend through
the internal capsule (reprinted from Berman et al., 2004
with permission from the American Association of
Neurological Surgeons).
323
of monopolar or bipolar stimulators and thus determine the relationship between current intensity and
distance to activated CST fibers.
21.3.2. Subcortical stimulation with brief pulse
trains and time-locked recording
Recently, Yamamoto et al. (2006) presented results
from both cortical and subcortical stimulation with
single pulses delivered at 2 Hz, while recording
D-waves from an epidural electrode in the cervical
region. The cortical stimulation was used to monitor
the status of the CST by looking for a decrement in
the amplitude of the D-wave, which is thought to
be linearly related to the number of CST axons activated. In contrast, subcortical stimulation was used
for subcortical mapping in an effort to determine
the distance from the site of stimulation to the CST.
The data presented suggested that a 20-mA threshold
for subcortical stimulation indicated a distance of
15 mm to the CST, a 10-mA threshold corresponding
to 10 mm, and a 5-mA threshold to a distance of
5 mm. Extrapolating this near-linear trend (at least
at lower stimulus levels) gives a convenient value
of 1 mA/mm!
At first glance, this result seems to violate the
inverse square law, where a doubling of the distance
should increase the threshold by a factor of 4, not 2.
However, the inverse square law is strictly valid only
in the case of a homogeneous and isotropically conducting medium, which may be encountered in computer simulations but rarely if ever in biological
systems! In fact, the nature of current spread through
tissue composed primarily of similarly oriented
axons, most with insulating myelin sheaths, and
interspersed with extracellular fluid, glial cells, and
perhaps infiltrating neoplastic tissue, is very poorly
understood and there is little experimental data
and virtually no realistic mathematical models to
serve as guidelines for the clinical neurophysiologist.
Further experimental investigations into the relationships among stimulus parameters, type of electrodes,
type and depth of anesthesia, and distance between
electrodes and the CST are clearly necessary to
resolve these issues. The best that can be suggested
at this point is to maintain as constant a depth of
anesthesia as possible, perhaps using an EEG-based
measure of anesthetic level (discussion of which is
beyond the scope of this chapter) or alternately monitoring the H-reflex as a measure of spinal cord excitability (Leppanen, 2006); stimulate frequently with a
324
C.D. YINGLING
bipolar electrode during subcortical dissection near

motor tracts; and consider halting dissection when
stimulation produces consistent motor responses.
21.3.3. Stimulation of the CST within the
brainstem
Surgery within the brainstem presents particular challenges to the neurosurgeon, given the density of critical structures. For many years, brainstem lesions
were often considered inoperable, but with modern
microsurgical techniques combined with neurophysiological mapping, many lesions in this region are
now amenable to surgical intervention.
Much of the morbidity in approaching lesions
within the brainstem is due to the problems encountered in approaching this region, which depending
on the location often involves working around multiple cranial nerves and/or their nuclei. Before even
addressing the issues of CST mapping, it is vital to
identify a surgical approach to intrinsic brainstem
lesions that in itself produces minimal morbidity.
These issues are beyond the scope of this chapter;
discussions of methods for monitoring cranial nerves
can be found in Yingling and Ashram (2005) and in
chapters of this volume by Lopez and by Ashram
and Yingling (see Chapters 25 and 26). Techniques
for mapping the floor of the fourth ventricle have
been described by Morota and Deletis (2006) and in
this volume by Lopez (see Chapter 23). Duffau
and Sichez (1998) have described a technique for
mapping the lamina quadrigemina in a patient with
a tectal cavernoma.
Depending on the approach, the CST may be
encountered only within the cavity deep to the lesion
(e.g., in the case of a dorsal approach through the
floor of the fourth ventricle), or relatively early
(e.g., in the case of a lesion in the ventral pons).
Thus, CST mapping may be carried out relatively
early or late in the procedure, and often in combination with other techniques to identify a safe approach
and/or other functional systems within the surgical
field. As opposed to mapping, the functional status
of the CST should also be monitored during such
cases by means of motor evoked potentials to transcranial stimulation, with recording of D-waves from
an epidural electrode and/or direct muscle responses.
Given the close proximity of CST fibers running to
all parts of the body and even bilaterally within the
brainstem, muscle responses should be monitored from
all four extremities as a complement to CST mapping
Fig. 2. Axial T2-weighted MRI scan showing cavernous

malformation in the left cerebral peduncle (reprinted from
Quinones-Hinojosa et al., 2005 with permission from
Lippincott, Williams and Wilkins).
techniques, which can utilize the same recording

channels synchronized to a different stimulator.
Quinones-Hinojosa et al. (2005) have recently
described an elegant use of both mapping and monitoring techniques in a patient with a cavernous
malformation in the left cerebral peduncle (Fig. 2).
They used a Kartush bipolar stimulator probe
(Medtronic-Xomed) in conjunction with an Ojemann
stimulator (Radionics), together with multichannel
EMG recording, to locate the CST which was
found to lie lateral to the lesion (Fig. 3). They were
able to identify a region over the lesion containing
no motor fibers, thus allowing a safe entry and total
resection. Transcranial MEPs were also monitored
during the resection (Fig. 4). The patients motor
function was improved upon discharge 2 weeks
after the procedure, compared to her presurgical
status. A similar procedure was described by
Deletis et al. (2000), except that rather than EMG
responses, D-waves were recorded from an epidural
electrode in response to stimulation of the cerebral
peduncles.
To sum up this section, CST mapping within the
brainstem, while conceptually similar to subcortical
mapping for supratentorial lesions, presents special
challenges both in determining a safe approach and
in performing precise mapping that is both sensitive
325
A
SC
A
PC
Leg
Trunk
Arm
Face
RHand-RHand2
RHand-RHand2
RTA-TA2
RTA-TA2
REH-EH2
REH-EH2
RFT-RFT2
RFT-RFT2
10 ms/Div
10 ms/Div
Fig. 3. A: Intraoperative photograph of left cerebral peduncle with number 1 indicating location of corticospinal tract
(CST) identified with electrical stimulation. SCA, superior cerebellar artery; PCA, posterior cerebral artery. B: Schematic
representation of relationship between the CST and the cavernous malformation. C: and D: EMG tracings showing responses
elicited by stimulation at the sites indicated, conforming to the expected somatotopy of the CST. RTA, right tibialis anterior;
REHL, right extensor hallucis longus; RFT, right intrinsic foot muscles (reprinted from Quinones-Hinojosa et al., 2005 with
permission from Lippincott, Williams and Wilkins).
R MEP-Average
R MEP-Average
RLO-
RLO-LO2
RTA-
RTA-TA2
REH-
REH-EH2
RFT-R
RFT-RFT2
10 ms
10 ms/Div
Fig. 4. Transcranial electrical motor evoked potentials (tceMEPs) elicited during resection of the cavernous malformation.
A: Normal polyphasic baseline responses, B: transiently increased latency and simplified biphasic waveform associated
with manipulation of cavernous malformation. Traces as in Fig. 3 (reprinted from Quinones-Hinojosa et al., 2005 with permission from Lippincott, Williams and Wilkins).
326
and specific, as well as technically feasible given the

limited surgical exposure that is possible. Again, the
issue of monopolar (more sensitive?) versus bipolar
(more specific?) stimulation will require further
careful study.
21.4. Mapping the CST within the spinal cord
Intramedullary spinal cord lesions present a similar,
although less complex, problem as intrinsic brainstem lesions: before mapping can be utilized, a safe
approach must first be determined. Since most intramedullary lesions are approached via a midline
myelotomy after an appropriate laminectomy, the
complexity of approaching brainstem lesions is
simplified to the single issue of determining where
the midline is. However, the physiological midline
does not always correspond to the anatomical midline, particularly when the anatomy is distorted by
pathological processes. Thus, physiological methods
of mapping the midline of the dorsal columns play
a key role in approaching intramedullary lesions
before the issue of CST mapping arises.
21.4.1. Mapping the dorsal column midline
The dorsal columns are unique in that, unlike other
sensory or motor systems within the spinal cord,
there are no synapses between the peripheral nerves
and the medulla. The cell bodies in the dorsal root
ganglia send bipolar axonal processes both peripherally and centrally, so that a single sensory axon originating in a toe, for example, may be 2 m long or
more in tall individuals. The significance of this for
midline determination is that axons, unlike chemical
synapses, conduct equally well in either direction.
Thus, methods utilizing both orthodromic and antidromic conduction have been described.
Within the dorsal columns, fibers from progressively more rostral segmental levels lie laterally to
those originating more caudally. Thus, at high cervical levels fibers from the L upper extremity, L lower
extremity, R lower extremity, and R upper extremity
would be encountered as one moved progressively
from L to R across the dorsal columns. (Unlike some
other sensory pathways, the dorsal column/medial
lemniscal system does not cross to the contralateral
side until after the first synapse in the dorsal column
nuclei.) Thus, one technique for locating the midline
is to stimulate each of the four extremities in turn
(or just the lower extremities for thoracic lesions)
C.D. YINGLING
and record the resulting action potentials from the

surface of the dorsal columns. This could theoretically be done with a single recording electrode, progressively moved from one side to the other; but
this method is painfully slow, since at each recording
location each extremity must be stimulated, and
usually signal averaging is required to obtain an
adequate signal/noise ratio.
A much faster method is to record simultaneously
from an array of small electrodes placed perpendicularly to the dorsal columns from one side to the other.
Krzan (2002) has described the fabrication of such an
electrode, and demonstrated its ability to accurately
identify the dorsal column midline in 55 of 65 cases.
(Reproducible recordings could not be obtained in
the other 10 cases for a variety of anatomical, physiological, and technical reasons). The responses
ranged from <1 to over 40 mV, and could be
obtained in 100 sweeps per average at a stimulation
rate of 13.3 s1, allowing collection of two sets of
averages in well under a minute.
The alternate approach is to stimulate the dorsal
columns and record peripheral sensory nerve action
potential (SNAP) responses, taking advantage of
antidromic conduction. In the context of placement
of dorsal column stimulating electrodes for alleviation of chronic deafferentation pain, Yingling and
Hosobuchi (1986) demonstrated that antidromic
recordings produced dramatically larger signal/noise
ratios than orthodromic recordings obtained from
the same electrodes by reversing the stimulation
and recording connections. Essentially the same technique can be applied to locating the dorsal column
midline to determine the safest location for myelotomy. A standard 1 mm ball-tip stimulating probe
(Medtronic-Xomed) is used to stimulate the exposed
dorsal columns, and progressively moved from one
side to the other while compound nerve action
potentials are recorded from electrodes over peripheral nerves. These can be the same electrodes used
for SEP stimulation, but it is simpler and faster
to place additional electrodes over the desired
nerves to avoid the necessity of replugging, with its
attendant potential for error.
Quinones-Hinojosa et al. (2002) described the
use of this technique in two pediatric patients with
intramedullary juvenile pilocytic astrocytomas
(Fig. 5A and C). The midline could be clearly identified by the transition from responses obtained on the
L to responses from the R as the midline was crossed,
together with a region of absent or smaller responses
327
An alternate method to recording antidromic sensory nerve potentials is to record compound muscle
action potentials (CMAP) from the same EMG
electrodes that are used for tcMEP recording. This
method is based on synaptic connections between
sensory and motor neurons normally subserving segmental reflexes, but in this case activated antidromically rather than orthodromically by stimulation of
the dorsal columns during mapping prior to myelotomy. Since CMAPs recorded directly from the
muscle have larger amplitudes than SNAPs obtained
from electrodes near, but not in, the sensory nerves,
this method provides almost instantaneous results
with little or no signal averaging necessary (Gardi
and Yingling, unpublished observations).
It is ironic that antidromic recordings were found
to be the actual mechanism underlying the so-called
neurogenic motor evoked potential technique
described by Owen et al. (1988), which was subsequently shown to be not a valid method for monitoring spinal motor pathways (Toleikis et al., 2000;
Minahan et al., 2001) and has been replaced by
tcMEP recordings in current practice. In this case,
however, the ability to obtain antidromic recordings
after stimulation of the dorsal columns has become
the basis for a different and valuable technique for
identifying the site where a myelotomy can be
performed safely.
Fig. 5. A: and C: Preoperative T1-weighted MRI scans
showing contrast-enhancing mass extending from the cervicomedullary junction to C4. B: and D: Postoperative scans
showing almost total resection of lesion; patient had
improved neurological function and no new deficits (reprinted from Quinones-Hinojosa et al., 2005 with permission from Lippincott, Williams and Wilkins).
corresponding to the midline septum (Fig. 6).

Mapping was carried out at several rostral/caudal
levels. It is significant that in this case, the physiological midline described an S curve due to distortion of normal anatomy by the tumor, and a straight
midline incision would have transected the R dorsal
column. By following the curved physiological
midline obtained by mapping, the tumor was exposed
by a myelotomy which avoided this complication,
and although SEPs were temporarily compromised
during the procedure, dorsal column function
returned to normal with no residual deficits.
21.4.2. Mapping the CST within the spinal cord

Having achieved safe entry into the spinal cord after
midline mapping, stimulation to identify the limits
for safe resection can proceed along the same lines
described above. During this phase of the procedure,
SEPs will be of less value (and may be transiently
depressed or lost even after an optimal myelotomy),
but tcMEP monitoring should be performed continuously, preferably with both D-wave and EMG recording to detect any compromise to the CST, as
described elsewhere in this volume (Chapters 16
and 45). In addition to continuous monitoring, it is
also useful to perform repeated mapping by intraspinal stimulation, in order to define the borders of
the CST before they are damaged. Again, a small
bipolar stimulating electrode such as the Kartush
probe described above (Section 21.3.3) is able to
deliver focal stimulation within the tight confines of
the spinal cord.
In the case described by Quinones-Hinojosa et al.
(2002) mentioned earlier in Section 4.1, free-running
328
C.D. YINGLING
A
Septum
B
Septum
C
Fig. 6. Dorsal column midline mapping performed with antidromic sensory nerve recordings (see text for details). The stimulating electrode was systematically moved from L to R, eliciting responses from L (upper traces) and R (lower traces) ulnar
nerves; similar traces were obtained from posterior tibial nerves. This mapping was repeated at several levels, and the myelotomy performed at the physiological midline, which was not straight and did not correspond to the anatomical midline; sensory
function was intact at the time of discharge (reprinted from Quinones-Hinojosa et al., 2002 with permission from Lippincott,
Williams and Wilkins).
EMG was continuously recorded from the same

muscles used to obtain tcMEPs. When tcMEP
recordings from the R side decreased during tumor
resection, dissection was temporarily halted and
intramedullary stimulation was applied at the tumor
capsule, beginning at 1 mA, which had not produced
responses earlier. However, after the decline in
tcMEP, the same level of stimulation produced
robust EMG responses on the R, even when the
intensity was lowered to 0.75 mA (Fig. 7). This
prompted the surgeon to pursue a less aggressive
resection of this tumor to avoid compromising functional tracts. While ependymomas are relatively well
encapsulated, making identification of the tumor border less difficult, in these cases of juvenile pilocytic
astrocytomas the ventral aspect of the tumor was
visually indistinguishable from functional neural
tissue, which could only be identified by the low
threshold response to stimulation. The R arm and
hand weakness with which she presented was no
worse postoperatively, and had continued to improve
at 4 month and one year followup; repeat MRI
revealed an almost total resection of the tumor
(Fig. 5B and D).
In a novel new method, Deletis (2006) recently
described a method for identifying the CST within
the spinal cord by use of a collision technique.
D-waves were elicited by transcranial stimulation,
and recorded from epidural electrodes both rostral
and caudal to the tumor. When the CST was stimulated with the proper timing in relation to the
transcranial stimulation, the antidromic (ascending)
activity collided with the descending D-wave,

reducing its amplitude by >50% (Fig. 8). This technique was claimed to be able to localize the CST
within 1 mm.
21.5. Conclusions and recommendations
It should be clear from this review that mapping the
CST below the cortical level is still very much a
work in progress, with no consensus on the optimal
methodology and a relatively sparse literature
compared to the extensive number of publications
concerning cortical mapping. This is obviously an
area of great interest and importance, and will provide fertile ground for further investigations in the
coming years. I will thus limit these closing comments to a few general issues.
First, one cannot overemphasize the distinction
between mapping and monitoring, to which I have
referred repeatedly. Monitoring techniques such as
tcMEP should always be carried out during surgery
near the CST at any level, as they can be done more
or less continuously and thus may provide the first
indication of a potential problem. In contrast,
mapping techniques are inherently intermittent and
generally require the surgeon to temporarily halt dissection in order to map the location of the CST. (It is,
however, interesting to speculate about the potential
utility of constantly electrified surgical instruments
such as ultrasonic aspirators or even routine suction
devices, which might allow mapping to become an
integral part of surgical dissection.)
329
15:43:38
16:04:05
Free
L UE
100 V
Amp 5
Free
1s
100 V
Amp 5
Free
Free
100 V
Amp 6
100
Amp 6
Free
Free
100 V
Amp 7
100
Amp 7
Free
Free
100 V
Amp 8
100
Amp 8
1s
L LE
1s
R UE
1s
1s
R LE
1s
1s
1s
Fig. 7. Responses in right upper extremity to direct intramedullary stimulation at 1.0 mA (left) and 0.75 mA (right) at the
ventral margin of the tumor (same patient as Figs. 5 and 6). Dissection was halted in the regions where low threshold EMG
responses were obtained, and no new motor deficits were seen postoperatively (modified from Quinones-Hinojosa et al.,
2002 with permission from Lippincott, Williams and Wilkins).
Second, since much of the morbidity in attempting surgery near the CST at brainstem or spinal
levels is a consequence of damage to structures
encountered during the surgical approach, CST
mapping must often be used in conjunction with

other techniques such as cranial nerve monitoring,
mapping the floor of the fourth ventricle, or locating
the dorsal column midline in order to maximize the
TES
C2 C1
Collision site
S1
S2 + S1
D1
S2
S1
S2 + S1
D2
D1
mm
S1
D2
50
50
0
10
15
20
25
30
35 40 ms
Negative mapping-no collision, D1 = D2
10
15
20
25
30
35 40 ms
Positive mapping-collision occurs, D2 = 42% of D1
Fig. 8. Mapping of CST by D-wave collision technique (see text for details). S1, transcranial electrical stimulation; S2,
spinal cord electrical stimulation; D1, control D-wave (transcranial stimulation only); D2, D-wave after combined brain
and spinal cord stimulation. Bottom left, negative mapping result (D1 D2); bottom right, positive mapping result after
stimulation of CST within spinal cord (D2 < D1 after collision). Inset shows bipolar stimulating probe over exposed spinal
cord (modified from Deletis, 2006).
330
safety of the approach to lesions near the CST.

Depending on the level and direction of approach,
actual mapping of the CST may occur relatively
early or relatively later in the procedure, whereas
tcMEP monitoring can be carried out throughout
the case.
Third, basic technical issues such as the relative
sensitivity and specificity of monopolar versus bipolar stimulation, the optimal pulse sequences to be
used, and some quantification of the distance to the
CST based on mapping parameters must be resolved.
Much of this work might be performed more effectively utilizing large animal models.
Finally, it should be pointed out that useful motor
function involves many more systems, both cortical
and subcortical, than simply the CST. Cognitive
planning, sensorimotor integration, and modulatory
influences from multiple systems all play a role and
should be preserved to the extent possible; neurophysiological methods for assessing such functionally important contributions do not exist with any
presently available technology. In this context, it is
interesting to note the recent report from Vitaz
et al. (2003), who were able to perform more effective motor mapping in patients operated on under
conscious sedation than under general anesthesia.
Awake patients are also able to perform more complex motor tasks and thus permit evaluation of higher
level motor function than simple movements.
Surgery under conscious sedation has long been carried out when language function is at risk, so the
technical obstacles are not insurmountable. For these
reasons, some surgeons (H. Duffau, personal communication) are now performing more supratentorial
procedures under local anesthesia, in order to better
assess motor competency during surgery and attempt
to avoid apraxias as well as frank paralysis when
operating near motor systems.
References
Berger, MS (1995) Functional mapping-guided resection of
low-grade gliomas. Clin. Neurosurg., 42: 437452.
Berger, MS and Ojemann, GA (1999) Techniques for functional brain mapping during glioma surgery. In: MS
Berger and CB Wilson (Eds.), The Gliomas. Saunders,
Philadelphia, pp. 421435.
Berman, JI, Berger, MS, Mukherjee, P and Henry, RG (2004)
Diffusion-tensor imaging-guided tracking of fibers of the
pyramidal tract combined with intraoperative cortical stimulation in patients with gliomas. J. Neurosurg., 101: 6672.
C.D. YINGLING
Cedzich, C, Tanaguchi, M, Schafer, S and Schramm, J
(1996) Somatosensory evoked potential phase reversal
and direct motor cortex stimulation during surgery in
and around the central region. Neurosurgery, 38:
962970.
Deletis, V (2006) Intraoperative neurophysiology of the
corticospinal tract of the spinal cord. In: C Barber,
S Tsuji, S Tobimatsu, T Uozumi, N Akamatsu and
E Eisen (Eds.), Suppl. Clin. Neurophysiol, Suppl. 59,
Elsevier, Amsterdam, pp. 107112.
Deletis, V, Sala, F and Morota, N (2000) Intraoperative
neurophysiological monitoring and mapping during
brainstem surgery: a modern approach. Oper. Tech.
Desmedt, JE and Cheron, G (1982) Somatosensory evoked
potentials in man: subcortical and cortical components
and their neural basis. Ann. N. Y. Acad. Sci., 388: 388411.
Duffau, H and Sichez, J-P (1998) Intraoperative direct electrical stimulation of the lamina quadrigemina in a case
of a deep tectal cavernoma. Acta Neurochir. (Wien),
140: 13091312.
Duffau, H, Capelle, L, Denvil, D, Sichez, N, Gatignol, P,
Taillandier, L, Lopes, M, Mitchell, M-C, Roche, S,
Muller, J-C, Bitar, A, Sichez, J-P and Van Effenterre,
R (2003) Usefulness of intraoperative electrical subcortical mapping during surgery for low-grade gliomas
located within eloquent brain regions: functional results
in a consecutive series of 103 patients. J. Neurosurg.,
98: 764778.
Haglund, MM, Ojemann, GA and Blasdel, GG (1993)
Optical imaging of bipolar cortical stimulation. J. Neurosurg. 78: 785793.
Kahle, W and Frotscher, M (2003) Color Atlas and Textbook of Human Anatomy. Nervous System and Sensory
Organs. Thieme Medical Publishers, Stuttgart, 5th ed.,
Vol. 3, 420 pp.
Keles, GE, Lundin, DA, Lamborn, KR, Chang, EF,
Ojemann, G and Berger, MS (2004) Intraoperative
subcortical stimulation mapping for hemispheric
perirolandic gliomas located within or adjacent to the
descending motor pathways: evaluation of morbidity
and assessment of functional outcome in 294 patients.
J. Neurosurg., 100: 369375.
Kombos, T, Suess, O, Kern, B-C, Funk, T, Hoell, T,
Kopetsch, O and Brock, M (1999) Comparison between
monopolar and bipolar electrical stimulation of the
motor cortex. Acta Neurochir. (Wien), 141: 12951301.
Krzan, JM (2002) Intraoperative neurophysiological
mapping of the spinal cords dorsal columns. In:
V Deletis and JL Shils (Eds.), Neurophysiology in
Neurosurgery: A Modern Intraoperative Approach.
Academic Press, San Diego and London, pp. 153165.
Leppanen, R (2006) Intraoperative applications of the
H-reflex and F-response: a tutorial. J. Clin. Mon. Comput., 20: 267304.

Kostuik, JP (2001) Anterior spinal cord injury with preserved neurogenic motor evoked potentials. Clin. Neurophysiol., 112: 14421450.
Morota, N and Deletis, V (2006) The importance of brainstem mapping in brainstem surgical anatomy before the
fourth ventricle and implication for intraoperative
neurophysiological mapping. Acta Neurochir. (Wien),
148: 499509.
Ojemann, GA (1991) Cortical organization of language.
J. Neurosci., 11: 22812287.
Owen, JH, Laschinger, J, Bridwell, K, Shimon, S, Nielsen, C,
Dunlap, J and Kain, C (1988) Sensitivity and specificity
of somatosensory and neurogenic-motor evoked potentials in animals and humans. Spine, 13: 11111118.
Penfield, W and Boldrey, E (1937) Somatic motor and sensory representation in the cerebral cortex of man as
studied by electrical stimulation. Brain, 60: 389443.
Quinones-Hinojosa, A, Gulati, M, Lyon, R, Gupta, N and Yingling, CD (2002) Spinal cord mapping as an adjunct for
resection of intramedullary tumors: surgical technique
with case illustrations. Neurosurgery, 51: 11991206.
Quinones-Hinojosa, A, Lyon, R, Du, R and Lawton, MT
(2005) Intraoperative motor mapping of the cerebral
peduncle during resection of a midbrain cavernous malformation: technical case report. Neurosurgery, 56(ONS
Suppl. 2): ONS-439.
Rampil, IJ and King, BS (1996) Volatile anesthetics depress
spinal motor neurons. Anesthesiology, 85: 129134.
Sala, F, Krzan, MJ and Deletis, V (2002) Intraoperative
neurophysiological monitoring in pediatric neurosurgery:
why, when, how? Childs Nerv. Syst., 18: 264287.
Sartorius, CJ and Berger, MS (1998) Rapid termination of
intraoperative stimulation-evoked seizures with application of cold Ringers lactate to the cortex. Technical
note. J. Neurosurg., 88: 349351.
Schekutiev, G and Schmid, UD (1996) Coaxial insulated
bipolar electrode for monopolar and bipolar mapping
of neural tissue: technical note with emphasis on the
principles of intra-operative stimulation. Acta Neurochir. (Wien), 138: 470474.
331
Skirboll, SS, Ojemann, GA, Berger, MS, Lettich, E and
Winn, HR (1996) Functional cortex and subcortical
white matter located within gliomas. Neurosurgery,
38: 678685.
Toleikis, JR, Skelly, JP, Carlvin, AO and Burkus, JK
(2000) Spinally elicited peripheral nerve responses are
sensory rather than motor. Clin. Neurophysiol., 111:
736742.
Vitaz, TW, Marx, W, Victor, JD and Gutin, PH (2003)
Comparison of conscious sedation and general anesthesia for motor mapping and resection of tumors located
near motor cortex. Neurosurg. Focus, 15: E8.
Wood, CC, Spencer, DD, Allison, T, McCarthy, G,
Williamson, PD and Goff, WR (1988) Localization of
human sensorimotor cortex during surgery by cortical
surface recording of somatosensory evoked potentials.
and Fukuya, C (2004) Intraoperative monitoring of the
corticospinal motor evoked potential (D-wave): clinical
index for postoperative motor function and functional
recovery. Neurol. Med. Chir. (Tokyo), 44: 170182.
Yamamoto, T, Fukaya, C and Katayama, Y (2006) Monitoring neurosurgical procedure using D-wave technique
Japan experience. In: Paper Presented at the Vth
International Symposium on Intraoperative Neurophysiological Monitoring in Neurosurgery, November 16,
2006, New York, NY. (Available on-line at www.neurophysiology.org).
Yingling, CD and Ashram, Y (2005) Intraoperative monitoring of cranial nerves in skull base surgery. In:
RK Jackler and DE Brackmann (Eds.), Neurotology.
Elsevier Mosby, Philadelphia, 2nd ed., pp. 958993.
Yingling, CD and Hosobuchi, Y (1986) Use of antidromic
evoked potentials in placement of dorsal cord disc electrodes. Appl. Neurophysiol., 49: 3641.
Yingling, CD, Ojemann, S, Dodson, B, Harrington, MJ and
Berger, MS (1999) Identification of motor pathways
during tumor surgery facilitated by multichannel
electromyographic recording. J. Neurosurg., 91:
922927.
Section II.3
Brainstem and Auditory Evoked Potentials

M.R. Nuwer (Ed.)
334
CHAPTER 22
BAEPs in surgery
Alan D. Legatt*
Departments of Neurology and Neuroscience, Montefiore Medical Center, and the Albert Einstein College of Medicine,
Bronx, NY 10467, USA
22.1. Introduction
Following the delivery of a brief acoustic stimulus,
such as a click or tone pip, a series of electrical
signals with latencies as long as hundreds of milliseconds can be recorded from human subjects. The earliest signals, which originate from the cochlea,
constitute the electrocochleogram (ECochG). The
ECochG was originally recorded from needle electrodes inserted through the tympanic membrane to a
location in the middle ear, in close proximity to the
cochlea (Ruben et al., 1961). It can also be recorded
from extratympanic recording sites, such as the external ear canal, but at substantially smaller amplitude;
signal averaging is used to record these volumeconducted signals with an adequate signal-to-noise
ratio. The longer-latency signals, called auditory
evoked potentials (AEPs), are most often recorded
from the skin surface. They can be further divided into
short-latency AEPs, with latencies of under 10 ms;
long-latency AEPs, with latencies over 50 ms; and
middle-latency AEPs, with latencies of 1050 ms.
The long-latency AEPs, which are predominantly
generated within the cerebral cortex, including cortical association areas, are profoundly changed by
whether the subject is attending to the acoustic stimuli
and extracting information from them, for example, by
performing discrimination tasks based on stimulus
characteristics or recognizing that novel stimuli have
been presented (Linden, 2005). The long-latency
AEPs are suppressed by surgical anesthesia, and are
not useful for intraoperative monitoring.
The middle-latency AEPs are most likely generated within the cerebral cortex, including primary
*
Correspondence to: Alan D. Legatt, M.D., Ph.D., Department

of Neurology, Montefiore Medical Center, 111 East 210th
Street, Bronx, NY 10467, USA.
Tel.: 1-718-920-6530; fax: 1-718-920-8509.
E-mail: legatt@aecom.yu.edu (A.D. Legatt).
auditory cortex and surrounding areas (Yvert et al.,

2001). They are also markedly affected by surgical
anesthesia, so much so that anesthetic-related variability impedes their use in continuous intraoperative
monitoring for focal neurologic dysfunction. However, this anesthetic sensitivity has led to application
of the middle-latency AEPs as an indicator of the
depth of anesthesia (Schneider et al., 2005).
The short-latency AEPs, most often called brainstem auditory evoked potentials (BAEPs) (Fig. 1),
are the AEPs most often used for clinical diagnostic
purposes because they are relatively easy to record
and their waveforms and latencies are highly consistent across normal subjects. They are almost identical
in the waking and sleeping states (Campbell et al.,
1992). Sedation (Loughnan et al., 1987) and surgical
anesthesia (Stockard et al., 1992; Legatt, 2002;
Banoub et al., 2003) produce only minor changes in
the BAEPs (Fig. 2). Some of the intraoperative BAEP
changes are attributable to changes in body temperature rather than anesthetic effects (Markand et al.,
1987; Litscher, 1995; Rodriguez et al., 1999). Because
of their resistance to anesthetic agents, BAEPs can be
used for intraoperative monitoring of the ears and the
infratentorial auditory pathways. The ECochG can
also be used for intraoperative monitoring. However,
since the ECochG is generated in the inner ear, it
may not detect dysfunction within the more proximal
portion of the auditory pathway, that is, the eighth
nerve or the brainstem auditory pathways.
Extraoperative diagnostic BAEP tests are evaluated
by comparing the BAEPs to those of a control normal
population. In order for this comparison to be valid,
the techniques used and the state of the subjects in
the patients study and the control studies must be
identical. This is difficult to ensure during intraoperative monitoring, where the anesthetic regimen, the
body temperature, the delivered stimulus intensity,
and other factors may differ from patient to patient.
Therefore, during intraoperative BAEP monitoring,
BRAINSTEM AND AUDITORY EVOKED POTENTIALS
III
[ISOFLURANE]A = 2.0% BP 94/55 TNP = 35.9 C
IV
II
335
VI
FP1-C3
VII
Cz-Ai
C3-O1
IN
VN
Cz-Ac
FP2-C4
C4-O2
0.2 V
1 ms
Ac-Ai
Fig. 1. Brainstem auditory evoked potentials (BAEPs)

recorded simultaneously from three different recording
electrode linkages following monaural stimulation in a
normal subject. The vertical dashed lines indicate the
peak latencies of waves IV and V in the Cz-Ai waveforms;
these peaks are more widely separated in the Cz-Ac waveforms. (Reprinted from Legatt, 2005, with permission from
Elsevier.)
T3-CZ
CZ-T4
EMG/ECG
15 v
1s
II III
IV V
each patient serves as his/her own control; BAEPs

recorded at a time when elements of the auditory pathways are at risk are compared to those recorded earlier
during the same operation (Legatt, 1991).
22.2. Components and sources
BAER
CZ +
0.25 V
60 dBHL
click
10
12
ms
22.2.1. The ECochG

The ECochG includes the cochlear microphonic and
the eighth nerve action potential. The cochlear
microphonic is generated in the receptor cells
the hair cells within the cochlea. It is so named
because its waveshape approximates the sound pressure waveform of the acoustic stimulus. Therefore,
if the waveshape of the acoustic stimulus is reversed
in polarity, the cochlear microphonic is also inverted.
The eighth nerve compound action potential is
generated by depolarization within the distal
(cochlear) ends of the auditory nerve axons, which
have been activated by excitatory synaptic input from
the cochlear hair cells. It is recorded as a phasic negativity in the middle ear or extratympanic recording
site irrespective of the stimulus polarity. Sounds in
the intensity range typically used for BAEP recordings typically elicit more than one volley within the
auditory nerve, producing the N1 and N2 components of the eighth nerve compound action potential,
and sometimes an N3 as well.
VI

recorded during surgery in a patient anesthetized with isoflurane at a concentration sufficient to render the electroencephalogram isoelectric. The component amplitudes were
reduced somewhat, but the latencies of waves I through V
were not significantly different from those recorded in this
patient in the unanesthetized state. (Reprinted from Stockard et al., 1992, with permission from Elsevier.)
22.2.2. BAEPs
BAEPs are most often recorded between electrodes
on the surface of the head, including an electrode at
the vertex (position Cz of the International 10
20 system) and electrodes at both earlobes (labeled
Ai ipsilateral to the stimulated ear and Ac contralateral to it) or at both mastoids (labeled Mi
and Mc). Recordings should include, at the minimum, a vertex-to-ipsilateral-ear (Cz-Ai) recording
channel. The vertex-positive peaks in this channel
are typically labeled with Roman numerals according
336
to the convention of Jewett and Williston (1971).

Other channels, such as Cz-Ac, may help in the
identification of components (Legatt, 2005) (Fig. 1).
Most of the BAEP components are recorded
from the skin surface as far-field potentials, which mean
that small displacements of the recording electrodes do
not significantly alter the BAEP waveform. The exceptions to this are wave I and part of wave II, which are
generated in the distal auditory nerve (see below) and
are therefore recorded as near-field potentials in the
vicinity of the stimulated ear. Changes in the location
of the Ai/Mi recording electrode can substantially alter
these components, and can therefore be used in an
attempt to improve their recording during intraoperative
BAEP monitoring.
Wave I of the BAEP arises from the first volley of
action potentials in the auditory nerve in the most
distal portion of the nerve (Legatt et al., 1988).
It represents the same electrical phenomenon as the
N1 component of the eighth nerve compound action
potential in the ECochG, as confirmed by simultaneous BAEP and ECochG recordings (Gersdorff,
1982). This produces a skin-surface negativity in
a circumscribed area around the stimulated ear
(Hughes and Fino, 1985; Grandori, 1986); the negativity at Ai appears as a positive peak in the Cz-Ai
recording. Since wave I is a near-field potential
around Ai, repositioning of the Ai recording electrode can substantially alter it, and alternate Ai electrode positions can be used to obtain a clearer wave I.
Because wave I arises from the most distal portion of
the auditory nerve, it may persist after the nerve is
sectioned at a more proximal location, such as during
surgery for eighth nerve tumors (Raudzens and
Shetter, 1982; Legatt et al., 1986) (Fig. 3).
Although some investigators have postulated a
one-peak-to-one-generator correspondence, research
has shown that Jewett and Williston (1971) were correct in their assertion that most of the BAEP components are composites of contributions of multiple
generators. The complexity of the generators of
human BAEPs (Fig. 4) derives in part from the pattern of connections within the auditory pathways,
with ascending fibers both synapsing in and bypassing various relay nuclei (Strominger and Strominger,
1971; Strominger, 1973; Strominger et al., 1977).
It also reflects the presence of at least two bursts of
activity in the auditory nerve (corresponding to the
N1 and N2 components of the eighth nerve compound action potentials in the ECochG), which can
drive the more rostral pathways. Because of both of
A.D. LEGATT
0.1 V
1 ms
Fig. 3. Intraoperative brainstem auditory evoked potentials

(BAEPs) to left ear stimulation recorded during surgery for
left acoustic neuromas in two different patients, showing
persistence of wave I (arrows) after transection of the intracranial eighth nerve. The nerves were intentionally sacrificed to permit total resection of the tumors. (Reprinted
from Legatt, 2002, with permission from Lippincott,
these factors, several different structures within the

infratentorial auditory pathways may be active and
generating field potentials simultaneously.
Wave II originates, in part, in neural activity that
began as the N1 component of the eighth nerve compound action potential and has propagated from the
distal auditory nerve to its proximal end and to the
cochlear nucleus. However, the activity at this point
in the auditory pathway occurs simultaneously with
the second auditory nerve volley, the N2 component
of the eighth nerve compound action potential, in
the distal nerve (Gersdorff, 1982). The latter contributes to the scalp-recorded BAEP in the same manner as the N1 component did when it was at
the same location. This can cause persistence of a
wave II in cases where the proximal eighth nerve
has been destroyed (Legatt, 2005). With regard to
the more proximal generator of wave II, the relative
contribution of activity in auditory nerve fibers
within the proximal nerve and of activity in cochlear
nucleus neurons has been a subject of controversy
(Legatt et al., 1988). This proximal generator is the
major determinant of the scalp topography of this
BAEP component over the dorsal part of the head
(Hughes and Fino, 1985).
Wave III predominantly originates in the caudal pontine tegmentum, including the region of the superior olivary complex, though a contribution from continued
activity at the level of the cochlear nucleus cannot be
337
AC
AC
VII?
VII?
V,VI
AR
MGN
IC
L V
L IV
IC
V L
IV L
BIC
MGN
AR
V
I II
I, II
IN,IIN
Cochlea
V,VI
V, VI?
VII?,SN
BIC
C
N
8th Nerv
e
II
S
O
C
III
II,III?
III
S
O
C
III IV
IN IIN
VI
VII
SN
Fig. 4. Diagram showing the probable generators of the human brainstem auditory evoked potentials (BAEPs). SN, slow
negativity after wave V; AC, auditory cortex; AR, auditory radiations; BIC, brachium of the inferior colliculus; CN,
cochlear nucleus; IC, inferior colliculus; LL, lateral lemniscus; MGN, medial geniculate nucleus; SOC, superior olivary
complex. (Reprinted from Legatt et al., 1988, with permission from Elsevier.)
ruled out (Legatt, 2005). Ascending projections from

the cochlear nucleus are bilateral, so wave III may
receive contributions from brainstem auditory structures both ipsilateral and contralateral to the stimulated ear. In patients with asymmetrical lesions of the
brainstem, wave III abnormalities are usually most
pronounced following stimulation of the ear ipsilateral
to the lesion (Brown et al., 1981; Oh et al., 1981;
Faught and Oh, 1985), though occasionally they are
more pronounced following contralateral stimulation
(Stockard and Rossiter, 1977).
Waves IV and V are often fused into a IVV
complex, and their anatomical generators are most
likely in close anatomical proximity or overlapping,
since they are usually either both affected or both
unaffected by brainstem lesions (Starr and Hamilton,
1976; Stockard and Rossiter, 1977; Stockard et al.,
1977). They may, however, be differentially affected
(Stockard and Rossiter, 1977; Legatt et al., 1988; Hirsch
et al., 1996) by intraoperative brainstem damage
(Fig. 5). Wave IV appears to reflect activity predominantly in ascending auditory fibers within the dorsal
and rostral pons, just caudal to the inferior colliculus,

while wave V predominantly reflects activity at
the level of the inferior colliculus, perhaps including
activity in the rostral portion of the lateral lemniscus
as it terminates in the inferior colliculus (Legatt,
2005). As is the case with wave III, wave V abnormalities due to unilateral brainstem lesions are usually most
pronounced following stimulation of the ear ipsilateral
to the lesion (Brown et al., 1981; Oh et al., 1981; Faught
and Oh, 1985; York, 1986; Scaioli et al., 1988), though
there are exceptions (Zanette et al., 1990; Fischer et al.,
1995).
Waves VI and VII are absent in some normal subjects. While they may in part reflect activity in more
rostral structures such as the medial geniculate
nucleus, they also receive contributions from activity
in the inferior colliculus (Legatt, 2005); the latter
generator may cause persistence of these waves in
patients with auditory pathway damage rostral to
the inferior colliculus. Therefore, BAEPs cannot be
used to assess or monitor the auditory pathways
rostral to the mesencephalon.
338
A.D. LEGATT
BAEPs to left ear stimulation
SEPs to left median SEPs to right median

nerve stimulation
nerve stimulation
18:30, aneurysm in view
19:30, dissecting around aneurysm
19:40, aneurysm ruptured
19:45, clip placed on basilar artery
19:55, clip taken off basilar artery

?
20:20, closing
20:40, closing
0.25 V
2 V
1 ms
2 V
8 ms
8 ms
Fig. 5. Intraoperative brainstem auditory evoked potentials (BAEPs) to left ear stimulation and cortical somatosensory
evoked potentials (SEPs) to stimulation of both median nerves recorded during surgery for a basilar artery aneurysm.
The cortical SEP to left median nerve stimulation disappeared when the aneurysm ruptured, most likely reflecting a loss
of perfusion pressure within the brainstem. The cortical SEP to right median nerve was still present at that point, but subsequently disappeared when the basilar artery was clipped in order to control the bleeding. A BAEP run was not obtained
between aneurysm rupture and basilar artery clipping, but following the clipping BAEP wave V (arrow) became delayed
and attenuated, and eventually disappeared, while BAEP wave IV (triangle) persisted. The patient suffered a brainstem
infarct. In the SEP waveforms, cortical negativity is shown as an upward deflection. The first 8 ms of each SEP waveform
was cropped off to remove the large stimulus artifacts. (Reprinted from Legatt, 2002, with permission from Lippincott,
22.3. Recording techniques

22.3.1. Stimulation
BAEPs are most commonly elicited by brief acoustic
click stimuli, produced by delivering trains of 100 ms
duration electrical square pulses to the acoustic transducer; brief tone pips can also be used. Since the
responses to rarefaction and compression clicks may
differ (Emerson et al., 1982; Schwartz et al., 1990),
extraoperative diagnostic BAEP studies typically
employ a single click polarity. Summation of the
responses to alternating click polarities are useful to
cancel a large stimulus artifact and/or the cochlear
microphonic, and are often employed during intraoperative BAEP monitoring. The same click polarity
(single or alternating) should be used throughout the
operation.
Since headphones are impractical for intraoperative monitoring, the acoustic stimuli are most often
delivered using ear inserts, incorporating foam cylinders that can be compressed and then gradually
expand to achieve a tight fit with the ear canal. The
foam can be covered with a thin layer of metal foil,
to serve as a near-field electrode for recording of
the ECochG and to record BAEP wave I at higher
amplitude. The ear insert can be held in place with
bone wax or cotton padding, secured with nonporous

tape or an adhesive waterproof dressing pad (Little
et al., 1983; Legatt, 1991). This is to prevent fluids
such as blood, skin prep, or irrigation fluid from
entering the ear canal, where they might interfere
with transmission of sound to the inner ear.
The ear insert is connected to the acoustic transducer via a length of plastic tubing. Care should be
taken to ensure that this tubing remains in place during positioning of the patient and is not kinked. The
time required for the acoustic signal to propagate
through the tubing typically prolongs the latencies of
all BAEP components by approximately 0.91.0 ms.
This causes no problems in the evaluation of the
BAEPs, since each patient serves as his own control
and the acoustic propagation delay is constant. Moreover, the delay helps to prevent obscuration of wave I
by the electrical stimulus artifact, because (1) it prolongs the latency of wave I, helping to separate it in
time from the electrical stimulus artifact (which
remains simultaneous with the activation of the acoustic transducer), and (2) it permits increasing the
distance between the acoustic transducer and
the recording electrodes, thus reducing the amplitude
of the electrical stimulus artifact.
In a clinical diagnostic BAEP recording, the stimulus intensity must be equal to that which was used
to obtain the normative data. The stimulus intensity
delivered during intraoperative BAEP monitoring
may be difficult to control precisely due to variability
in the positioning of the ear insert, but this again does
not cause a problem since each patient serves as his
or her own control. The intensity setting chosen
should be loud enough to produce a clear BAEP but
not loud enough to cause ear damage.
A stimulus rate of 10/s is typical, but a rate of
exactly 10 Hz or another submultiple of the power
line frequency should be avoided. If a line-frequency
or harmonic artifact appears in the averaged BAEPs
and examination of the raw data does not show
increased line frequency artifact, the stimulus rate
should be adjusted slightly, because timing circuits
can drift.
As is the case with extraoperative diagnostic
studies, each ear should be stimulated separately;
binaural stimulation should not be employed. Contralateral white noise masking is typically used during
extraoperative diagnostic BAEP studies to prevent
acoustic crosstalk air or bone conduction of the
acoustic stimulus to the nonstimulated ear. Most
current intraoperative monitoring equipment can
339
deliver interleaved left- and right-sided stimuli and

sort the responses into separate averages, in effect
acquiring averaged BAEPs to left-ear and to rightear stimulation simultaneously. This has the advantage of reducing the time during which one of these
BAEPs is not being examined. Contralateral noise
masking cannot be used when interleaved stimuli
are delivered, however. This does not pose a major
problem for intraoperative BAEP monitoring, since
the major reason for white noise masking is to prevent the appearance of a BAEP when a deaf ear is
stimulated during a diagnostic BAEP study; it will
not prevent recognition of auditory pathway compromise when a functioning ear is stimulated. Also,
the magnitude of acoustic crosstalk is less with earinsert transducers than with headphones (Roeser and
Clark, 2000).
22.3.2. Recording
Recording electrodes are placed at the vertex (Cz)
and at each ear or mastoid. Cup electrodes or needle
electrodes can be utilized. Duplicate or backup
electrodes are useful because it is usually difficult
or impossible to replace an electrode that becomes
unusable in the middle of an operation, when the
patient is positioned and draped. Electrodes and part
of the adjacent connecting wires should be attached
to the patient securely, to prevent dislodgement.
The electrodes should be oriented so the connecting
wires are directed away from the surgical field
(Legatt, 1991). Modern evoked potential recording
systems typically have enough channels to permit
recording of both the Cz-Ai and the Cz-Ac waveforms. Wave V is identifiable in both of these recording linkages, and recording both of them provides a
measure of redundancy should one of the ear electrodes become unusable. An Ac-Ai recording channel
can improve the detection of wave I; although this
component is picked up mainly as a near-field
negativity at the Ai electrode, the horizontal orientation of its dipole projects a small positivity to the
contralateral ear (Legatt, 2005).
If the auditory nerve is at risk, such as during
resection of an eighth nerve tumor, and the surgical
exposure permits it, an electrode can be placed on
the proximal eighth nerve to record a near-field compound action potential (Mller and Jannetta, 1983;
Legatt, 1991) (Fig. 6). This signal is typically much
larger than the far-field BAEP, permitting signal
averaging using fewer epochs and thus providing
340
A.D. LEGATT
22.4. Intraoperative use of BAEPs and

the ECochG
1 V
8th NCZ
CZ+ up
0.08 V
M1CZ
1 2 3 4 5 6 7 8 9 10
ms

recorded from a platinum pad electrode placed on the intracranial eighth nerve (top) compared to surface-recorded
BAEPs (bottom) during surgery in a 52-year-old woman
with a left-sided intracanalicular acoustic neuroma. Note
the voltage calibrations; the near-field response is considerably larger. The patient had transient BAEP changes during the resection but BAEPs were at baseline at its end, and
postoperative hearing was normal. (Courtesy of Dr. Timothy A. Pedley.)
more frequent assessment of the cochlea and of the

auditory nerve distal to the near-field electrode.
A typical analog filter bandpass for BAEP recording is 100 or 150 Hz to 3,000 Hz (3 dB points).
While line-frequency (e.g., 60 Hz) notch filters
should not be used for somatosensory evoked potential (SEP) recordings because they can cause a
ringing oscillatory artifact (Yamada, 1988), they
can be used for BAEP recordings. The analog gain
depends on the input window of the analog-to-digital
converter; a value of 100,000 is typical.
An averaging epoch duration of 10 ms is often
used for extraoperative diagnostic BAEP recordings
in adults. A longer epoch duration, typically 15 ms,
is preferable during intraoperative BAEP monitoring
because component latencies can be prolonged by
preexisting pathology, hypothermia, or intraoperative
compromise of the auditory system. The choice of
the number of sweeps per average will depend on
the signal-to-noise ratio of the raw data. A value of
1,000 sweeps per average is typical, but more may
be required if the raw data are noisy and the BAEPs
are small. Near-field recording from the proximal
eighth nerve (Fig. 6) may permit averaging using a
much smaller number of sweeps.
22.4.1. Surgical procedures

As noted above, BAEPs can be used to monitor
acoustically evoked activity within the auditory pathways up through the level of the mesencephalon;
they are not useful for monitoring the pathways rostral to this. They are most often used to monitor
surgery for eighth nerve tumors such as vestibular
schwannomas (formerly called acoustic neuromas)
and for tumors or vascular abnormalities within the
posterior fossa, both extra-axial and within the substance of the brainstem. BAEP monitoring can help
to avoid excessive eighth nerve stretch from cerebellar retraction, which can cause hearing loss, during
surgery in the cerebellopontine angle.
Intraoperative monitoring of the ECochG has also
been used during cerebellopontine angle surgery.
It may be a useful adjunct when combined with
BAEP monitoring, because it requires less averaging
(fewer epochs) than the scalp BAEPs and thus may
contribute to more rapid recognition of cochlear dysfunction; also, it may detect cochlear dysfunction that
does not cause BAEP changes (Ojemann et al., 1984;
Levine et al., 1994; Schlake et al., 2001). However,
the ECochG may not detect eighth nerve damage that
spares its distal end and the cochlea, and some
patients in whom the ECochG is preserved are deaf
postoperatively (Symon et al., 1988). Therefore,
ECochG monitoring by itself is not sufficient for
posterior fossa surgery.
22.4.2. Analysis and interpretation of BAEPs
Because BAEP component amplitudes are highly variable across a population of normal human subjects,
extraoperative diagnostic BAEP studies are typically
evaluated based on latency criteria examination of
absolute component latencies, interpeak intervals
(the IV interpeak interval is often called the central
transmission time), and the rightleft differences of
these measures. The sole exception to this is the
IVV:I amplitude ratio, which may demonstrate abnormality in subjects in whom the latencies are normal.
The interpretation of these BAEP studies is also based
on the most reliable peaks waves I, III, and V
(American Clinical Neurophysiology Society, 2006).
In contrast to the large intersubject variation in
component amplitudes, amplitudes on repeated testing
in the same subject are usually quite consistent, if the

recording techniques are not altered. Moreover,
intraoperative compromise of neural pathways may
cause amplitude changes earlier than, or in the
absence of, latency changes. Therefore, both the
amplitudes and the latencies of the BAEP components
are used in the interpretation of intraoperative BAEPs
monitoring data. The amplitudes of the vertexpositive peaks of waves I, III, and V are measured
with respect to the troughs that follow them. Typical
threshold criteria for the identification of an adverse
BAEP change are a 50% decrease in the amplitude
of a component (most often of wave V), or a 1-ms
increase in the absolute latency of wave V or in the
IV interpeak interval.
A IVV complex with a dominant wave IV and a
poorly identifiable wave V is a normal variant pattern.
If this occurs, the latency and amplitude of wave IV
can be followed. The Cz-Ac recording channel may
also be useful in obtaining a clearer wave V, since
the peaks of waves IV and V are often more separated
and better differentiated in this recording channel than
in the Cz-Ai recording channel (Legatt, 2005).
341
22.5. Causes of intraoperative BAEP changes

V
Intraoperative BAEP changes can be classified into

three categories: true positive changes, which
reflect compromise of the structures that the monitoring is intended to safeguard, changes produced by
other physiologic mechanisms such as anesthetic
effects or hypothermia, and changes due to technical
problems (Legatt, 2002). These categories will be
considered separately.
22.5.1. Technical problems
As with any evoked potential, signals may be lost
due to equipment malfunction, dislodged electrodes,
disconnected or broken wires, and operator error
(the use of incorrect protocols or settings). Additional
technical problems specific to BAEPs include dislodging or kinking of the plastic tubing through which
the acoustic stimuli reach the ear and the entry of
liquids into the ear canal.
Artifacts (Fig. 7) can obscure the BAEPs and prevent
their identification and measurement (Legatt, 1995).
Automatic artifact rejection and avoidance of a stimulus
repetition rate that is a submultiple of the line frequency
1 ms
D
Fig. 7. Single, unaveraged data epochs recorded during
intraoperative brainstem auditory evoked potential (BAEP)
monitoring. A: An electrical stimulus artifact is present
at the beginning of the epoch. The BAEP, which is less
than 1 mV in amplitude, is not visible in this single epoch
of raw data. B: Bipolar cautery produces a large artifact
that triggers automatic artifact rejection and also causes
clipping during analog-to-digital conversion. C: The
cavitational ultrasonic surgical aspirator (CUSA) device
produces a very high frequency but low voltage artifact
that completely obliterates the neurophysiologic data but
does not trigger automatic artifact rejection. D: The light
source of the operating microscope produces a repetitive
sharply contoured artifact that recurs at a harmonic of
the line frequency but is composed of higher frequencies
that would not be removed by a line-frequency notch
filter. In AC, the horizontal dotted lines show the
input window of the analog-to-digital converter and the
threshold for automatic artifact rejection. In D, the amplifier gain was reduced to show the light source artifacts
in their entirety. Voltage calibration bar 10 mV in AC,
and 40 mV in D. (Reprinted from Legatt, 2002, with permission from Lippincott, Williams and Wilkins.)
342
can help to reduce artifact in the averaged BAEPs.

A line frequency notch filer can also be employed.
However, some equipment, notably the light sources
of operating microscopes, may generate higher harmonics of the line frequency (Fig. 7D) that are not removed
by filtering and are more difficult to remove by signal
averaging than a pure 50 or 60 Hz sine wave.
Electrocautery artifact is typically large enough to
trigger artifact rejection (Fig. 7B). However, monopolar
electrocautery may saturate the amplifier input stages
for a considerable time after the cautery current stops.
Since DC and low frequencies are removed by analog
filtering, the voltages at the inputs to the analog-todigital converter rapidly return to near zero. The
subsequent data epochs, which do not contain any
evoked potential, will not be rejected as artifacts, and
their incorporation into the average will cause an apparent amplitude attenuation of the BAEPs. This problem
can be avoided by manually pausing the averager until
examination of the raw data shows that the amplifiers
have recovered (Legatt, 1991). Some commercial
evoked potential recording systems suspend averaging
when they detect an amplifier saturation condition.
The high-frequency artifact caused by cavitational
ultrasonic surgical aspirator (CUSA) devices may
also completely obliterate the neural signals without
triggering automatic artifact rejection (Fig. 7C), causing an apparent amplitude attenuation of the BAEPs.
Therefore, averaging should be manually paused
during CUSA use (Legatt, 1991).
A.D. LEGATT
is lowered enough, the BAEPs may disappear

completely; the longer-latency components usually
disappear before wave I does. The degree of hypothermia required to obliterate the BAEPs varies in
different reports, ranging from 25 C (Stockard
et al., 1978a, 1978b) to 20 C (Markand et al.,
1987; Hett et al., 1995), though Rosenblum et al.
(1985) reported recording interpretable BAEPs at
temperatures as low as 14 C. These temperatures
are most often encountered during hypothermic cardioplegia, but body temperature decreases of a lesser
degree are commonly seen in anesthetized patients
and may produce mild BAEP latency prolongations.
Localized hypothermia within the surgical field,
caused by irrigation with cold fluids, can also cause
BAEP alterations in the absence of tissue damage
(Fig. 8). Irrigation with cold fluids is also undesirable during posterior fossa surgery because it can
produce neurotonic facial EMG discharges resembling those caused by facial nerve injury (Kartush
and Bouchard, 1992).
Drilling of bone, such as the roof of the internal
auditory canal during resection of an eighth nerve
tumor, will produce high noise levels in both ears
via bone conduction and can alter the BAEPs due
to acoustic masking (Levine et al., 1994). This should
be kept in mind when evaluating BAEPs acquired
during drilling of bone. Alternatively, BAEP averaging
can be paused during drilling.
22.5.3. Localized auditory system dysfunction
22.5.2. Physiologic effects

Anesthetic agents in the usual concentrations produce
only minimal changes in BAEP amplitudes and
latencies (see Legatt, 2002 for more details), but
BAEPs do change in response to hypothermia. Both
the interpeak intervals and the latency of wave I
progressively increase as the patient is cooled.
Component latencies and interpeak intervals increase
by about 7% for each 1 C drop in temperature, and
at 26 C are about double their values at normal body
temperatures (Markand et al., 1987). The latency
changes are reversible upon rewarming, but there is
hysteresis latencies at the same temperature may
differ between the cooling and rewarming phases
(Markand et al., 1990).
BAEP component amplitudes may show an initial
increase as the core temperature is lowered to the
2530 C range, but then decrease as the patient is
cooled further (Kusakari et al., 1984; Markand et al.,
1987; Rodriguez et al., 1995). If the temperature
Auditory system dysfunction and damage can be

caused by mechanical forces such as compression
or traction, by thermal injury from cauterization, or
by ischemia due to compromise of the vascular supply to the tissue. Some changes may be irreversible,
but detection of these may, nonetheless, improve
the surgical outcome, since notification about the
BEP changes may prevent the surgeons from damaging more tissue in the area. The pattern of BAEP
changes that occur depends on the location of the
dysfunction (Legatt, 2002).
Cochlear dysfunction will cause delay and attenuation of wave I (and the proximal eighth nerve compound action potential, if this is being monitored).
As wave I becomes delayed, the latencies of later
components increase in parallel, with little change
in the interpeak intervals. Cochlear dysfunction may
also decrease the amplitude of wave I to the point
where it is no longer identifiable, resulting in a
BAEP waveform with a delayed wave V, a delayed
343
09:27, 36.6 C, starting operation

10:38, 37.0 C, opening dura
11:02, 37.3 C, retracting cerebellum
11:45, 37.3 C, resecting tumor
13:05, 37.4 C, resecting tumor
13:34, 37.6 C, irrigating

14:23, 37.3 C, stopped irrigating
15:07, 37.4 C, closing
0.25 V
1 ms
Fig. 8. Intraoperative brainstem auditory evoked potentials (BAEPs) to right ear stimulation during surgery for a meningioma in the right cerebellopontine angle. The BAEPs were stable during cerebellar retraction and tumor resection. After the
tumor had been removed, copious irrigation of the surgical field with cold fluids produced a transient prolongation of the I
III interpeak interval, reflecting slowing of the conduction velocity within the eighth nerve due to local cooling. The peaks
latencies of waves I, III, and V are marked by the small diamonds, and the clock times, esophageal temperatures, and surgical procedures corresponding to each of the BAEP waveforms are noted at the right. (Reprinted from Legatt, 2002, with
permission from Lippincott, Williams and Wilkins.)
or absent wave III, and an absent wave I. With more

severe cochlear dysfunction, all BAEP components
are lost (Fig. 9).
Dysfunction of the eighth nerve proximal to its
cochlear end will cause a prolongation of the IIII
interpeak interval, attenuation of waves III and V, or
both. The latencies of waves III and V increase in
parallel, with relatively little change in the IIIV
interpeak interval unless the auditory pathways within
the brainstem are also affected (Figs. 8 and 10).
If the damage is severe enough, waves III and V will
be lost (Fig. 3). Changes such as these have been
correlated with dissection of cerebellopontine angle
tumors off the eighth nerve (Levine et al., 1984).
Wave I may also become delayed or disappear if
there is concurrent cochlear dysfunction due to
compromise of the internal auditory artery (see
below); according to Sekiya et al. (1985), obliteration
of wave I during manipulation of the eighth nerve is

always due to such vascular compromise. If the
cochlea is unaffected, however, and the damage to
the eighth nerve is all proximal to its cochlear end,
wave I may persist, even if the eighth nerve is
completely transected (Fig. 3). Wave II could also
persist in this situation due to its contribution from
the distal eighth nerve.
Damage to the lower pons, around the area of
the cochlear nucleus or the superior olivary complex,
will also delay waves III and V or cause them both
to be lost; wave I will be preserved if the ear and
eighth nerve are intact. Damage to the brainstem
that is entirely rostral to the lower pons, but at or
below the level of the mesencephalon will affect
wave V, but not waves I or III. Changes in wave IV tend
to parallel those in wave V, though occasionally these
components may be differentially affected (Fig. 5).
344
A.D. LEGATT
0.2 V
I
2 ms
Fig. 9. Consecutive brainstem auditory evoked potentials (BAEPs) to left ear stimulation (earliest waveform at the top)
recorded in a patient undergoing surgery for a left acoustic neuroma. A clear wave I and a poorly formed wave V were
initially present and were stable during the initial dissection, but all BAEP components disappeared simultaneously
during dissection within the internal auditory canal and remained absent through the end of the operation. This was most
likely due to interruption of the blood supply to the cochlea via the internal auditory artery. (Reprinted from Legatt, 2002,
with permission from Lippincott, Williams and Wilkins.)
Brainstem compromise can also cause an increase in the

amplitude of wave I, reflecting dysfunction of descending inhibitory pathways within the auditory system
(Musiek, 1986; Legatt et al., 1988).
Intraoperative loss of wave V does not rule out the
possibility of preserved postoperative hearing, even if
wave V remains absent through the end of the operation (Levine et al., 1994; Harner et al., 1996). This
may be due to temporal dispersion without conduction block of activity in the ascending auditory pathways. It may also reflect damage that affects only a
portion of the brainstem auditory pathways; BAEPs
reflect activity in a subset of the brainstem auditory
system that subserves sound localization, and hearing

can be audiometrically normal in patients with
grossly abnormal BAEPs (Legatt et al., 1988).
Several specific surgical situations that can cause
true positive BAEP changes are discussed in the
following sections. Further details can be found in
Legatt (2002).
22.5.3.1. Direct mechanical damage to the cochlea
or labyrinth
During drilling of the temporal bone, the membranous
labyrinth may be inadvertently entered. This generally
leads to deafness and a loss of all BAEP components

1.68
4.20
2.52
345
6.06
1.86
0.2 V
2 ms
3.54
1.83
2.13
5.37
7.50 ms
Fig. 10. Intraoperative brainstem auditory evoked potentials (BAEPs) to right ear stimulation recorded during surgery for a
right acoustic neuroma, showing two runs recorded before (top) and after (bottom) retraction of the cerebellum. The most
prominent change in the BAEPs was an increase in the IIII interpeak interval of more than 1 ms, reflecting stretching of
the eighth nerve. The smaller change in the IIIV interpeak interval may reflect effects of the retraction on the brainstem.
(Reprinted from Legatt, 2002, with permission from Lippincott, Williams and Wilkins.)
to stimulation of the affected ear. However, hearing can

sometimes be preserved when a semicircular canal is
entered during drilling if it is immediately closed
with bone wax (Molony et al., 1992; Ojemann, 2001).
22.5.3.2. Cochlear ischemia or infarction
The cochlea receives its blood supply from the
intracranial circulation via the internal auditory
artery, which is usually a branch of the anterior
inferior cerebellar artery and passes through the
internal auditory canal alongside the eighth nerve
(Kim et al., 1990). Damage to this artery will cause
cochlea ischemia or infarction, affecting wave I and
all subsequent BAEP components. The effects occur
rapidly, though complete recovery can occur if
flow is restored within several minutes (Perlman
et al., 1959).
Damage to the internal auditory artery probably
accounts for most cases of sudden loss of all BAEP
components, including wave I (Fig. 9), during
surgery for cerebellopontine angle tumors (Nadol
et al., 1987). Although BAEP changes due to compression of this artery may be reversible (Perlman
et al., 1959), they often reflect obstruction, disruption, or coagulation of the artery during tumor
resection within the internal auditory canal, which

causes irreversible cochlear damage and postoperative deafness. Levine et al. (1994) reported that,
when monitoring the ECochG in conjunction with
the BAEPs during surgery for cerebellopontine angle
tumors, if the eighth nerve compound action potential
suddenly disappeared and remained absent for more
than 15 min, hearing was never preserved.
22.5.3.3. Eighth nerve stretch due to cerebellar
retraction
Retraction of the cerebellum to gain access to the
cerebellopontine angle also moves the brainstem
away from the internal auditory meatus and stretches
the eighth nerve, which may cause hearing loss.
Intraoperative BAEP monitoring may serve to notify
the surgeons when the eighth nerve is being stretched
and the retraction needs to be reduced or readjusted
(Mller and Jannetta, 1983) (Fig. 10).
22.5.3.4. Direct mechanical or thermal damage
to the eighth nerve
The eighth nerve can be directly traumatized during
posterior fossa surgery, such as for cerebellopontine
angle tumors, either mechanically or by heat from
346
tissue cauterization. Pressure on the nerve is a potentially reversible cause of nerve dysfunction. BAEP
changes during manipulation of the eighth nerve
may also reflect vasospasm within the nerve (Levine
et al., 1984; Nadol et al., 1987).
22.5.3.5. Distal avulsion due to traction on
the eighth nerve
Complete resection of eighth nerve tumors may
require scraping fragments of tumor off the nerve,
which may cause BAEP changes either due to direct
auditory nerve damage or to the effects of the traction on the nerve. At its distal (cochlear) end, the
auditory nerve breaks up into fine fascicles that enter
the bony modiolus. These fascicles are mechanically
fragile, and may be avulsed if the traction on the
nerve is from the ear towards the brainstem, whereas
traction on the nerve from the brainstem towards the
ear is relatively benign. Excessive cerebellar retraction is another potential cause of distal eighth nerve
avulsion (Sekiya and Mller, 1987).
22.5.3.6. Brainstem damage during posterior
fossa vascular surgery
BAEP changes during posterior fossa vascular surgery
may reflect ischemia or infarction to the brainstem auditory pathways due to clipping or compression of arteries
perfusing the brainstem (Fig. 5); also, in the event of
aneurysm rupture, they may reflect loss of perfusion
pressure within the posterior circulation, effectively a
steal phenomenon (note the changes in the SEP to left
median nerve stimulation in Fig. 5) (Legatt, 2002).
However, BAEPs may also remain unchanged in
patients who suffer brainstem damage that anatomically
spares the auditory pathways (Little et al., 1987). While
preservation of the BAEPs does not guarantee a good
outcome, patients with significant BAEP changes that
persist to the end of the operation almost always have
new postoperative neurologic deficits (Little et al.,
1983; Manninen et al., 1994).
22.5.3.7. Brainstem damage during tumor surgery
Mechanisms of brainstem damage during posterior
fossa tumor surgery include compression, ischemia/
infarction due to compromise of the blood supply,
direct mechanical damage from dissection (including
CUSA), and thermal injury from cautery or laser use.
The presence or absence of BAEP changes, and their
pattern if present, will depend on the areas of the brainstem that are involved. As in the case of vascular surgery, brainstem damage that spares the auditory
A.D. LEGATT
pathways may leave the BAEPs unchanged, even in

patients who suffer significant neurologic morbidity.
22.5.3.8. BAEP changes with closing of the dura
Adverse BAEP changes can occur at or after closure of
the dura, even if BAEPs were stable prior to that point
(Ojemann et al., 1984; Mller and Mller, 1989; Wahlig
et al., 1999). These may reflect shifts of posterior fossa
contents with compression of auditory pathway structures and, if untreated, may correspond to significant
postoperative hearing loss (Mller and Mller, 1989).
22.6. Reaction to BAEP changes
A variety of responses are possible when significant
BAEP changes are reported to the surgeons, including
readjustment of retraction, measures to reverse vasospasm within the eighth nerve, and modification
of the dissection techniques. Accurate localization of
the area of dysfunction that is causing the evoked
potential changes will permit selection of the most
appropriate responses. During surgery for cerebellopontine angle tumors, BAEP and ECochG findings
that indicate that hearing has been irreversibly lost
(disappearance of the eighth nerve compound action
potential for more than 15 min) may permit the surgeon to abandon attempts to preserve the eighth nerve,
and thus resect the remainder of the tumor more
quickly (Levine et al., 1994). It should be noted, however, that loss of wave V does not rule out the possibility of preserved postoperative hearing, even if wave V
remains absent through the end of the operation during
tumor (Levine et al., 1994) or microvascular decompression (Friedman et al., 1985; Radtke et al., 1989)
surgery. Neu et al. (1999) feel that patients with postoperative hearing despite loss of wave V may be at
high risk for a delayed postoperative hearing loss.
References
American Clinical Neurophysiology Society (2006) Guideline 9C: guidelines on short-latency auditory evoked
potentials. J. Clin. Neurophysiol., 23: 157167.
Banoub, M, Tetzlaff, JE and Schubert, A (2003) Pharmacologic and physiologic influences affecting sensory
evoked potentials: implications for perioperative monitoring. Anesthesiology, 99: 716737.
Brown, RH, Jr., Chiappa, KH and Brooks, E (1981) Brainstem auditory evoked responses in 22 patients with intrinsic brainstem lesions: implications for clinical
interpretations. Electroencephalogr. Clin. Neurophysiol.,
52: 38P.

Campbell, KB, Bell, I and Bastien, C (1992) Evoked
potential measures of information processing during
natural sleep. In: RJ Broughton and RD Ogilvie
(Eds.), Sleep, Arousal, and Performance. Birkhauser,
Boston, pp. 89116.
Emerson, RG, Brooks, EB, Parker, SW and Chiappa, KH
(1982) Effects of click polarity on brainstem auditory
evoked potentials in normal subjects and patients: unexpected sensitivity of wave V. Ann. N. Y. Acad. Sci., 388:
710721.
Faught, E and Oh, SJ (1985) Brainstem auditory evoked
responses in brainstem infarction. Stroke, 16: 701705.
Fischer, C, Bognar, L, Turjman, F and Lapras, C (1995)
Auditory evoked potentials in a patient with a unilateral
lesion of the inferior colliculus and medial geniculate
body. Electroencephalogr. Clin. Neurophysiol., 96:
261267.
Friedman, WA, Kaplan, BJ, Gravenstein, D and Rhoton, AL, Jr.
(1985) Intraoperative brainstem auditory evoked potentials
during posterior fossa microvascular decompression.
J. Neurosurg., 62: 552557.
Gersdorff, MCH (1982) Simultaneous recordings of human
auditory potentials: transtympanic electrocochleography
(ECoG) and brainstem-evoked responses (BER). Arch.
Otorhinolaryngol., 234: 1520.
Grandori, F (1986) Field analysis of auditory evoked brainstem potentials. Hear. Res., 21: 5158.
Harner, SG, Harper, CM, Beatty, CW, Litchy, WJ and Ebersold, MJ (1996) Far-field auditory brainstem response in
neurotologic surgery. Am. J. Otol., 17: 150153.
Hett, DA, Smith, DC, Pilkington, SN and Abbott, TR
(1995) Effect of temperature and cardiopulmonary
bypass on the auditory evoked response. Br. J. Anaesthiol., 75: 293296.
Hirsch, BE, Durrant, JD, Yetiser, S, Kamerer, DB and
Martin, WH (1996) Localizing retrocochlear hearing
loss. Am. J. Otol., 17: 537546.
Hughes, JR and Fino, JJ (1985) A review of generators of
the brainstem auditory evoked potential: contribution
of an experimental study. J. Clin. Neurophysiol., 2:
355381.
Jewett, DL and Williston, JS (1971) Auditory-evoked far
fields averaged from the scalp of humans. Brain, 94:
681696.
Kartush, JM and Bouchard, KR (1992) Intraoperative facial
nerve monitoring. Otology, neurotology, and skull base
surgery. In: JM Kartush and KR Bouchard (Eds.), Neuromonitoring in Otology and Head and Neck Surgery.
Raven Press, New York, pp. 99120.
Kim, HN, Kim, YH, Park, IY, Kim, GR and Chung, IH
(1990) Variability of the surgical anatomy of the neurovascular complex of the cerebellopontine angle. Ann.
Otol. Rhinol. Laryngol., 99: 288296.
Kusakari, J, Inamura, N, Sakurai, T and Kazutomo, K
(1984) Effect of hypothermia on brainstem auditory
347
evoked potentials in humans. Tohoku J. Exp. Med.,
143: 351359.
Legatt, AD (1991) Intraoperative neurophysiologic monitoring. In: EAM Frost (Ed.), Clinical Anesthesia in
Neurosurgery. Butterworth-Heinemann, Boston, MA,
pp. 63127.
Legatt, AD (1995) Intraoperative neurophysiologic monitoring: some technical considerations. Am. J. EEG
Technol., 35: 167200.
Legatt, AD (2002) Mechanisms of intraoperative brainstem
auditory evoked potential changes. J. Clin. Neurophysiol., 19: 396408.
Legatt, AD (2005) Brainstem auditory evoked potentials: methodology, interpretation, and clinical application. In: MJ Aminoff (Ed.), Electrodiagnosis in
Clinical Neurology. Churchill Livingstone, New York,
pp. 489523.
Legatt, AD, Pedley, TA, Emerson, RG, Stein, BM, Abramson, M, Dowling, K and Gallo, E (1986) Electrophysiological monitoring of seventh and eighth nerve function
during surgery for acoustic neuromas. Electroencephalogr. Clin. Neurophysiol., 64: 30P.
Legatt, AD, Arezzo, JC and Vaughan, HG, Jr. (1988) The
anatomic and physiologic bases of brainstem auditory
evoked potentials. Neurol. Clin., 6: 681704.
Levine, RA, Ojemann, RG, Montgomery, WW and
McGaffigan, PM (1984) Monitoring auditory evoked
potentials during acoustic neuroma surgery. Insights
into the mechanism of the hearing loss. Ann. Otol.
Rhinol. Laryngol., 93: 116123.
Levine, RA, Ronner, SF and Ojemann, RG (1994) Auditory
evoked potential and other neurophysiologic monitoring
techniques during tumor surgery in the cerebellopontine
angle. In: CM Loftus and VC Traynelis (Eds.), Intraoperative Monitoring Techniques in Neurosurgery.
McGraw-Hill, New York, pp. 175191.
Linden, DE (2005) The P300: where in the brain is it produced and what does it tell us? Neuroscientist, 11:
563576.
Litscher, G (1995) Continuous brainstem auditory evoked
potential monitoring during nocturnal sleep. Int. J. Neurosci., 82: 135142.
Little, JR, Lesser, RP, Luders, H and Furlan, AJ (1983)
Brainstem auditory evoked potentials in posterior circulation surgery. Neurosurgery, 12: 496502.
Little, JR, Lesser, RP and Luders, H (1987) Electrophysiological monitoring during basilar aneurysm operation.
Loughnan, BL, Sebel, PS, Thomas, D, Rutherfoord, CF and
Rogers, H (1987) Evoked potentials following diazepam
or fentanyl. Anaesthesia, 42: 195198.
Manninen, PH, Patterson, S, Lam, AM, Gelb, AW and
Nantau, WE (1994) Evoked potential monitoring during
posterior fossa aneurysm surgery: a comparison of two
modalities. Can. J. Anaesth., 41: 9297.
348
Markand, ON, Lee, BI, Warren, C, Stoelting, RK, King, RD,
Brown, JW and Mahomed, Y (1987) Effects of hypothermia on brainstem auditory evoked potentials in humans.
Ann. Neurol., 22: 507513.
Markand, ON, Warren, C, Mallik, GS and Williams, CJ
(1990) Temperature-dependent hysteresis in somatosensory and auditory evoked potentials. Electroencephalogr. Clin. Neurophysiol., 77: 425435.
Mller, AR and Jannetta, PJ (1983) Monitoring auditory
functions during cranial nerve microvascular decompression operations by direct recording from the eighth nerve.
J. Neurosurg., 59: 493499.
Mller, AR and Mller, MB (1989) Does intraoperative
monitoring of auditory evoked potentials reduce incidence of hearing loss as a complication of microvascular decompression of cranial nerves? Neurosurgery, 24:
257263.
Molony, TB, Kwartler, JA, House, WF and Hitselberger,
WE (1992) Extended middle fossa and retrolabyrinthine
approaches in acoustic neuroma surgery: case reports.
Am. J. Otol., 13: 360363.
Musiek, FE (1986) Neuroanatomy, neurophysiology, and
central auditory assessment. Part III: Corpus callosum
and efferent pathways. Ear Hear., 7: 349358.
Nadol, JB, Jr., Levine, R, Ojemann, RG, Martuza, RL,
Montgomery, WW and De Sandoval, PK (1987) Preservation of hearing in surgical removal of acoustic neuromas of the internal auditory canal and cerebellar pontine
angle. Laryngoscope, 97: 12871294.
Neu, M, Strauss, C, Romstock, J, Bischoff, B and Fahlbusch,
R (1999) The prognostic value of intraoperative BAEP
patterns in acoustic neurinoma surgery. Clin. Neurophysiol., 110: 19351941.
Oh, SJ, Kuba, T, Soyer, A, Choi, IS, Bonikowski, FP and
Vitek, J (1981) Lateralization of brainstem lesions by
brainstem auditory evoked potentials. Neurology, 31:
1418.
Ojemann, RG (2001) Retrosigmoid approach to acoustic
neuroma (vestibular schwannoma). Neurosurgery, 48:
553558.
Ojemann, RG, Levine, RA, Montgomery, WM and
McGaffigan, P (1984) Use of intraoperative auditory
evoked potentials to preserve hearing in unilateral
acoustic neuroma removal. J. Neurosurg., 61: 938948.
Perlman, HB, Kimura, R and Fernandez, C (1959) Experiments on temporary obstruction of the internal auditory
artery. Laryngoscope, 69: 591613.
Radtke, RA, Erwin, CW and Wilkins, RH (1989) Intraoperative brainstem auditory evoked potentials: significant decrease in postoperative morbidity. Neurology,
39: 187191.
Raudzens, PA and Shetter, AG (1982) Intraoperative monitoring of brainstem auditory evoked potentials. J. Neurosurg., 57: 341348.
A.D. LEGATT
Rodriguez, RA, Audenaert, SM, Austin, EH, III and
Edmonds, HL, Jr. (1995) Auditory evoked responses in
children during hypothermic cardiopulmonary bypass:
report of cases. J. Clin. Neurophysiol., 12: 168176.
Rodriguez, RA, Edmonds, HL, Jr., Auden, SM and Austin,
EH, III (1999) Auditory brainstem evoked responses
and temperature monitoring during pediatric cardiopulmonary bypass. Can. J. Anaesth., 46: 832839.
Roeser, RJ and Clark, JL (2000) Clinical masking. In:
RJ Roeser, M Valente and H Hosford-Dunn (Eds.), Audiology: Diagnosis. Thieme, New York, pp. 253279.
Rosenblum, SM, Ruth, RA and Gal, TJ (1985) Brainstem
auditory evoked potential monitoring during profound
hypothermia and circulatory arrest. Ann. Otol. Rhinol.
Laryngol., 94: 281283.
Ruben, RJ, Bordley, JE and Lieberman, AT (1961) Cochlear
potentials in man. Laryngoscope, 71: 11411164.
Scaioli, V, Savoiardo, M, Bussone, G and Rezzonico, M
(1988) Brainstem auditory evoked potentials (BAEPs)
and magnetic resonance imaging (MRI) in a case of facial
myokymia. Electroencephalogr. Clin. Neurophysiol., 71:
153156.
Schlake, HP, Milewski, C, Goldbrunner, RH, Kindgen, A,
Riemann, R, Helms, J and Roosen, K (2001) Combined
intra-operative monitoring of hearing by means of auditory
brainstem responses (ABR) and transtympanic electrocochleography (ECochG) during surgery of intra- and extrameatal acoustic neurinomas. Acta Neurochir. (Wien), 143:
985995.
Schneider, G, Hollweck, R, Ningler, M, Stockmanns, G and
Kochs, EF (2005) Detection of consciousness by electroencephalogram and auditory evoked potentials. Anesthesiology, 103: 934943.
Schwartz, DM, Morris, MD, Spydell, JD, Brink, CT, Grim,
MA and Schwartz, JA (1990) Influence of click polarity
on the brainstem auditory evoked response (BAER)
revisited. Electroencephalogr. Clin. Neurophysiol., 77:
445457.
Sekiya, T and Mller, AR (1987) Avulsion rupture of the
internal auditory artery during operations in the cerebellopontine angle: a study in monkeys. Neurosurgery, 21:
631637.
Sekiya, T, Iwabuchi, T, Kamata, S and Ishida, T (1985)
Deterioration of auditory evoked potentials during cerebellopontine angle manipulations. An interpretation
based on an experimental model in dogs. J. Neurosurg.,
63: 598607.
Starr, A and Hamilton, AE (1976) Correlation between
confirmed sites of neurological lesions and abnormalities of far-field auditory brainstem responses. Electroencephalogr. Clin. Neurophysiol., 41: 595608.
Stockard, JJ and Rossiter, VS (1977) Clinical and pathologic correlates of brainstem auditory response abnormalities. Neurology, 27: 316325.

Stockard, JJ, Stockard, JE and Sharbrough, FW (1977)
Detection and localization of occult lesions with brainstem auditory responses. Mayo Clin. Proc., 52: 761769.
Stockard, JJ, Sharbrough, FW and Tinker, JA (1978a)
Effects of hypothermia on the human brainstem auditory
response. Ann. Neurol., 3: 368370.
Stockard, JJ, Stockard, JE and Sharbrough, FW (1978b) Nonpathologic factors influencing brainstem auditory evoked
potentials. Am. J. EEG Technol., 18: 177209.
Stockard, JJ, Pope-Stockard, JE and Sharbrough, FW
(1992) Brainstem auditory evoked potentials in neurology: methodology, interpretation, and clinical application. In: MJ Aminoff (Ed.), Electrodiagnosis in
Clinical Neurology. Churchill Livingstone, New York,
pp. 503536.
Strominger, NL (1973) The origins, course and distribution
of the dorsal and intermediate acoustic striae in the
rhesus monkey. J. Comp. Neurol., 147: 209234.
Strominger, NL and Strominger, AI (1971) Ascending
brainstem projections of the anteroventral cochlear
nucleus in the rhesus monkey. J. Comp. Neurol., 143:
217242.
Strominger, NL, Nelson, LR and Dougherty, WJ (1977)
Second-order auditory pathways in the chimpanzee.
J. Comp. Neurol., 172: 349366.
349
Symon, L, Sabin, HI, Bentivoglio, P, Cheesman, AD,
Prasher, D and Barratt, H (1988) Intraoperative monitoring of the electrocochleogram and the preservation
of hearing during acoustic neuroma excision. Acta Neurochir. Suppl. (Wien), 42: 2730.
Wahlig, JB, Kaufmann, AM, Balzer, J, Lovely, TJ and
Jannetta, PJ (1999) Intraoperative loss of auditory function relieved by microvascular decompression of the
cochlear nerve. Can. J. Neurol. Sci., 26: 4447.
Yamada, T (1988) The anatomic and physiologic bases of
median nerve somatosensory evoked potentials. Neurol.
Clin., 6: 705733.
York, DH (1986) Correlation between a unilateral midbrainpontine lesion and abnormalities of brainstem
auditory evoked potential. Electroencephalogr. Clin.
Yvert, B, Crouzeix, A, Bertrand, O, Seither-Preisler, A and
Pantev, C (2001) Multiple supratemporal sources of magnetic and electric auditory evoked middle latency components in humans. Cereb. Cortex, 11: 411423.
Zanette, G, Carteri, A and Cusumano, S (1990) Reappearance
of brainstem auditory evoked potentials after surgical
treatment of a brainstem hemorrhage: contributions to the
question of wave generation. Electroencephalogr. Clin.

M.R. Nuwer (Ed.)
350
CHAPTER 23
Mapping the brainstem: floor of the fourth ventricle

Jaime R. Lopez*
Department of Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, CA 94305, USA
23.1. Introduction
Surgical treatment of brainstem lesions carries a
substantial risk of postoperative morbidity because of
the risk of injuring the tightly packed cranial nerve
nuclei (CNN) and neural tracts within the rhomboid
fossa and brainstem (Lang et al., 1991). Historically,
neurosurgeons considered this area to be a no mans
land with most lesions being inoperable (Baker,
1965; Katsuta et al., 1993). However, with the study
by Lassiter et al. (1971), surgical treatment of intraaxial brainstem lesions has been considered a realistic
option. In addition, continued improvement of technical imaging and microsurgical techniques has
allowed for increased success in the treatment of brainstem lesions (Heffez et al., 1990). The improvement in
magnetic resonance imaging (MRI) now allows better
definition of the pathological anatomy within the
brainstem and in many cases is also useful in helping
to predict the pathology of the lesion (Heffez et al.,
1990). Furthermore, the increased use of intraoperative neurophysiologic monitoring (IOM) with
somatosensory evoked potentials (SEPs), brainstem
auditory evoked potentials (BAEPs), and motor
evoked potentials (MEPs) has helped to reduce postoperative neurologic deficits (Strauss et al., 1994;
Steinberg et al., 2000). Unfortunately, these conventional neurophysiologic techniques are of limited use
when attempting to surgically resect intra-axial brainstem lesions because they cannot be used to identify
and map functional structures on the brainstem surface. This is of great importance because neurologic
deficits can occur from mechanical and/or vascular
injury to neural pathways and nuclei during surgical
*
Correspondence to: Jaime R. Lopez, M.D., Intraoperative

Neurophysiologic Monitoring Program, Department of
Neurology and Neurological Sciences, and Neurosurgery,
Stanford University School of Medicine, Room A343,
300 Pasteur Dr., Stanford, CA 94305, USA.
Tel.: 1-650-723-1975; fax: 1-650-725-5095.
E-mail: lopezjr@stanford.edu (J.R. Lopez).
dissection of surface structures prior to the resection

of intrinsic brainstem lesions (Strauss et al., 1994).
Thus, it is critical to be able to identify surface brainstem structures since the majority of intrinsic brainstem lesions are approached through the fourth
ventricle. Anatomical landmarks may not be discernible because they may be distorted by disease. Therefore, the use of neurophysiologic techniques to assist
in identifying neural structures, allowing the surgeons
to avoid these structures during surgery, has been
advocated (Eisner et al., 1995).
23.2. Floor of the fourth ventricle anatomical

landmarks
It is imperative to have an understanding of the anatomy of the rhomboid fossa in order to properly plan
the neurophysiologic techniques to use for mapping
and monitoring of the brainstem. Anatomical landmarks of this area have been reported by Lang et al.
(1991). They found that the striae medullares could
not be considered an anatomical landmark because of
its highly variable location on the surface. In addition,
in 14% of their sample, no stria was seen. They also
found that the broadest area of the rhomboid fossa is
more frequently not the midpoint of its length.
This is important because the facial colliculus is
normally located above the striae. However, if the
striae are not present, then visual identification of the
facial colliculus becomes very difficult (Strauss
et al., 1994). It is of critical importance to identify
the facial colliculus because injury to this region may
cause not only facial nerve dysfunction but also disturbance of the abducens nucleus, eye movements, and
vestibular function because of the different neural
pathways located in this area (Lang et al., 1991). The
abducens nucleus and the medial longitudinal fasciculus (MLF) are in close proximity to the facial colliculus. The parapontine reticular formation (PPRF) is
ventral and medial to the facial colliculus while the
vestibular nuclei are lateral (Strauss et al., 1994). The
ability to precisely identify functional neural structures is further complicated by the distortion and displacement produced by intrinsic brainstem space
occupying lesions (Strauss et al., 1994; Eisner et al.,
1995; Morota et al., 1996). Therefore, the need for a
method to identify functional structures on the surface
of the rhomboid fossa is vital. Such a technique, using
electrical stimulation of brainstem surface structures,
was first mentioned by Fahlbusch and Strauss in
1991 (Lang et al., 1991; Katsuta et al., 1993). Subsequently, similar techniques have been described by
other authors (Rusyniak et al., 1992; Katsuta et al.,
1993; Strauss et al., 1993; Eisner et al., 1995; Morota
et al., 1995, 1996; Chang et al., 1999).
23.3. Assessment of brainstem function
Routine, conventional neurophysiologic monitoring of
brainstem function typically requires the use of BAEPs,
SEPs, and MEPs. The combined multimodality use of
SEPs and MEPs allows for monitoring of a larger brainstem territory. However, the combined use of these
conventional modalities provides electrophysiologic
coverage of only 20% of the brainstem cross-sectional
area between the pontomedullary and pontomesencephalic junctions at any given level (Lang et al., 1991;
Strauss et al., 1994). In addition, these conventional
techniques are not useful in identifying surface structures prior to dissection. However, using techniques
similar to those used for stimulation of peripheral nerve
fibers, brainstem structures can be safely and reliably
activated; thus, allowing for identification of the neural
structures (Katsuta et al., 1993).
23.4. Brainstem stimulation
The purpose of electrical stimulation of brainstem
structures is to identify cranial motor nuclei and their
motor tracts. The nucleus of most interest is the facial
colliculus, not only because it lies above important
neural structures as mentioned previously, but also
because it lies only 0.25 mm beneath the ependyma.
The hypoglossal nucleus is also reported to be
situated 0.52.6 mm deep from the surface (Strauss
et al., 1994). Both of these nuclei lend themselves
to direct electrical stimulation and can be readily
obtained at relatively low stimulation intensities.
23.5. Stimulation techniques
There are several different stimulation parameters,
using both bipolar and monopolar techniques, reported
in the literature. The following will be a summary of
351
the several methods described. Strauss et al. (1993)

used both monopolar and bipolar as well as constant
current and constant voltage stimulation. Stimulation
intensities varied between 0.1 and 2 mA and 0.1 and
1 V. Stimulation duration was between 50 and
400 ms. Single rectangular stimuli with levels up to
10 Hz were used. When using monopolar stimulation,
a handheld probe which was insulated, except at the
tip, served as the cathode and a needle electrode placed
in a wound muscle acted as the anode stimulus. Pulse
duration was 100 ms for constant voltage stimulation.
Rusyniak et al. (1992) reported using monopolar, constant current stimulation with an intensity of 0.1 mA.
They applied the stimulation for 1.0 s. They estimated
that when using these stimulation parameters, the current
spread from a monopolar electrode was 0.5 mm.
In a separate article, Katsuta et al. (1993) reported
electrical stimulation of the facial colliculus using
bipolar silver-ball electrodes. Square wave pulses of
0.2 ms and an intensity of 1 mA, with a frequency
of 1 Hz were applied.
A combination of mapping and continuous monitoring was employed by Eisner et al. (1995). They used
stimulation forceps (GK 675, Aeskulap, Germany)
to apply constant current, square wave impulses at
0.13 mA with a repetition rate of 4.7 Hz.
Morota et al. (1995) reported using monopolar
stimulation and applying 12 s trains of stimuli of
0.2 ms duration delivered at 4 Hz. Distinct and stable
muscle responses were obtained at stimulation intensities of 1.52.0 mA. The intensity was then reduced and
the threshold level that was established (0.32.0 mA)
was used for mapping.
In 1999, Chang et al. reported using stimulation
intensities as high as 10.2 mA with stimulus duration
varying from 0.02 to 0.05 ms in patients undergoing
resection of pontine cavernous malformations. Stimulation was increased at 0.1 mA increments, after starting
at 0, until a threshold response was identified. Stimulation was then increased until a maximum compound
motor action potential (CMAP) or spread to other nuclei
was seen. No adverse outcomes were reported from
the higher stimulus intensities. However, it should
be noted that the higher stimulus intensities (above
4 mA) reported by the authors were applied to tissue
felt to be necrotic or to areas adjacent to the original
hemorrhage and not to normal appearing tissue.
23.6. Electromyographic recordings
The goal of brainstem stimulation is to activate motor
nuclei and/or pathways. Thus, recordings need to be
352
obtained from muscles corresponding to the motor

CNN and tracts. Because of the complex structures
of the rhomboid fossa, different sets of muscle
recordings have been reported by the different
groups. However, all groups reported electromyography (EMG) recordings from at least two facial nerve
innervated muscles. The following is a summary of
the recording techniques reported.
Rusyniak et al. (1992) reported EMG responses
from electrodes placed in the face and tongue. It
also appears that they visually inspected the face
for movement, since they state that gross motor
responses were noted.
Katsuta et al. (1993) report recording evoked
myographic responses from surface electrodes placed
on ipsilateral orbicularis oculi and oris muscles. The
EMG signal was amplified with a bandpass filter of
103,000 Hz. No averaging was required to obtain
the EMG response (Figs. 1 and 2).
A higher number of muscles were used for recordings
by Strauss et al. (1993). Monopolar stainless steel needles were used. For facial nerve recordings, the active
electrodes were placed in the orbicularis oris and oculi
muscles, and an indifferent electrode was placed at the
frontal midline. The genioglossus muscle was the site
for the active electrode for hypoglossal recordings.
Eisner et al. (1995) used custom-made, noninsulated,
steel-needle recording electrodes. Active and reference
electrodes were placed in muscles corresponding to
the CNNs being monitored and consisted of the following: superior rectus, medial rectus, superior oblique,
masseter, orbicularis oculi, orbicularis oris, trapezius,
velum palatinum, and lingual muscle. In addition to
recording-triggered EMG responses, they also monitored continuous EMG activity using auditory feedback
Fig. 1. Placement of orbicularis oris needle recording

electrodes.
PEZ
J.R. LO
Fig. 2. Needle electrodes placed in the orbicularis oculi

muscle, prior to being taped onto the skin.
to the surgeon via a loudspeaker. Recording parameters

are outlined in Table 1.
Morota et al. (1995) used custom-made, 30-gauge,
stainless steel, Teflon-coated wires with a 2-mm bare
hook tip, encased in a 27-gauge needle as their recording electrodes. After inserting the needle into the muscle, the needle was then withdrawn leaving the tip of
the electrode wire in the muscle. Recordings were
obtained from obricularis oris and oculi to record
facial nerve responses, posterior pharyngeal wall for
cranial nerves (CNs) IX and X activity, and lateral
tongue to monitor hypoglossal nerve responses. Four
EMG responses were recorded using epoch lengths
of 20 ms, amplified 10,000 times, filtered between
50 and 2,133 Hz and averaged.
Intramuscular hook-wire electrodes were also
used by Chang et al. (1999). These were and still
are commercially available (Xomed Surgical Products, Jacksonville, FL). Recordings were obtained
from the masseter, orbicularis oculi, and orbicularis
oris muscles. The active and reference electrodes
were placed in the same muscle and spaced a few
millimeters apart. In addition to recording electronically triggered compound action potentials, continuous monitoring of neurotonic discharges from the
muscles monitored was also performed.
Tables 1 and 2 summarize the different stimulation
and recording parameters reported in numerous studies.
The differences in the techniques highlight that
there is no single method which is clearly superior.
Adequate and reliable recordings can be obtained
using monopolar, bipolar, constant current, or constant
voltage stimulation. Furthermore, stimulation can be
either nonrecurrent with only single stimuli, or recurrent with frequencies reported as high as 10 Hz.
353
Table 1
Recording parameters
Setting
Continuous EMG
Time base
Amplitude gain (mV/unit)
High filter (Hz)
Low filter (Hz)
1000 msa
20
1500
100
Triggered EMG
200 ms per divisionb
50100
2000
100
20 msa
20100
1500
100
35 ms per divisionb
50200
1500
100
Eisner et al., 1995.

Chang et al., 1999.
23.7. Stimulation parameters and safety

Experimental studies on the effects of electrical stimulation on the brain indicate that certain variables are
important in determining neural damage (Agnew and
McCreery, 1987; McCreery et al., 1990; Harnack
et al., 2004). Current density (I/A amperes/cm2),
charge per phase (Q/phase I PD C/phase),
and charge density per phase (QD/phase J PD
C/cm2 phase), where I stimulus pulse amplitude
(ampere); PD stimulus pulse duration (sec); C
coulombs; and A surface area of the electrode
(cm2), are important stimulus parameters related to
neural damage (Agnew and McCreery, 1987).
Agnew and McCreery (1987) reported that neural
damage occurred in tissue immediately subjacent to
the subdural stimulating electrodes when the charge
density at the surface of the electrodes exceeded
40 mC/cm2 phase and the charge per phase was
greater than 0.4 mC per phase at a pulse frequency of
50 Hz. They also found that charge density, rather than
charge per phase, correlated more closely with the
severity of neural damage when using small electrodes.
Possible mechanisms of neural damage include charge
transfer across an electrodetissue interface such as
electrochemically produced toxic products or passage

of current through tissue causing neuronal hyperactivity or through power dissipation possibly leading to
thermal injury (Agnew and McCreery, 1987).
In the study by Chang et al. (1999), they subsequently reported using an average stimulation intensity of 3.4 mA and a pulse duration of 0.02 ms.
Based on these stimulation parameters and the surface area of the stimulating electrode, they calculated
the charge per phase to be 0.068 mC per phase and
the charge density per phase to be 17.4 mC/cm2
phase (Chang et al., 2000). Strauss et al. (1999)
reported a charge density of 0.59 mC/cm2 phase
using constant voltage stimulation and charge densities up to 0.0014 mC/cm2 phase when using constant current stimulation. The stimulus parameters
reported by both groups are well below the safety
limits. Clinically, patients seem to tolerate this type
of electrical stimulation well. There have been only
a few reports of ventricular arrhythmia in patients
with a known history of cardiac arrhythmia (Strauss
et al., 1999). Possible cardiovascular side effects
need to be monitored during brainstem stimulation
since the parasympathetic dorsal motor nucleus of
the vagus is lateral to the hypoglossal trigone
Table 2
Stimulation parameters
Stimulation
Intensity
Stimulation duration
Frequency
Rusyniak et al., 1992

Strauss et al., 1993
Strauss et al., 1993
Katsuta et al., 1993
Eisner et al., 1995
Morota et al., 1995
Chang et al., 1999
Monopolar
Both
Both
Bipolar
1.0 s
50400 ms
100 ms
0.2 ms
0.2 ms
0.020.05 ms
10 Hz
10 Hz
1.0 Hz
1.0 Hz
4.7 Hz
12 s trains at 4 Hz
1.1 Hz
Morota and Deletis, 2006
Monopolar
0.1 mA
0.12 mA
0.11 V
1 mA
0.13 mA
1.52.0 mA
0.110.2 mA
(average 3.4 mA)
2.0 mA for screening
0.2 ms
14 Hz
Monopolar
Monopolar
PEZ
J.R. LO
354
(Strauss et al., 1999). At our institution, no patient

has experienced side effects as a result of brainstem
stimulation. During the stimulation protocol, we
carefully monitor a patients blood pressure and heart
rate and to date have not encountered any alterations
in these parameters.
23.8. Structures stimulated
The fact that triggered EMG responses can be
obtained with relatively low-level stimulation intensities, as low as 0.1 mA and stimulus duration of
2050 ms (Chang et al., 1999), has been suggested
to indicate that the peripheral motor neuron is the site
of stimulation (Strauss et al., 1999). In their review
of the literature, Shannon et al. (1997) reported that
the time constant for stimulating myelinated central
nervous system axons ranges from 50 to 100 ms,
200 to 300 ms for large cortical neurons, and 1 to
10 ms for other neurons and their dendrites.
Based on the calculated estimates required to activate neural structures within the brainstem and
comparing those with motor threshold measured
stimulation intensities, Strauss et al. (1999) determined that the axons of the peripheral motor neurons
are the site of intraoperative electrical stimulation.
23.9. Anesthetic considerations
Different anesthetic protocols can be successfully used
during these procedures, but the most important aspect
is that muscle relaxants are not used and that the
patient not be under the influence of neuromuscular
blocking agents during the time that EMG monitoring
is to take place. These drugs are also likely to have
adverse effects on MEPs, which are frequently used
as part of a multimodality approach for monitoring
these cases. The typical anesthetic protocol at our institution is induction with a short-acting barbiturate, an
intravenous narcotic agent, and a short-acting neuromuscular blocking agent. Thereafter, anesthesia is usually maintained with a combination of isoflurane, with
a maximum end-tidal not exceeding 0.6%, and a maximum of 50% nitrous oxide. Very short-acting intravenous narcotics, such as remifentanyl, are used after
induction to supplement the inhaled agents.
23.10. Typical protocol at our institution
IOM techniques used, in addition to brainstem
mapping, include SEPs after bilateral median and
Table 3
Motor cranial nerves and their corresponding muscles
suitable for EMG monitoring
Cranial Nerve
Muscles
III
IV
V
VI
VII
Inferior rectus
Superior oblique
Masseter
Lateral rectus
Frontalis, orbicularis oris,
orbicularis oris, mentalis
Posterior pharyngeal
muscles
Crycothyroid or vocalis
Sternocleidomastoid
Tongue or genioglossus
IX
X
XI
XII
posterior nerve stimulation, bilateral BAEPs, and transcranial electrical motor evoked potentials (TcMEPs)
with muscle recordings obtained from all four limbs.
These techniques will allow monitoring of the intrinsic
brainstem auditory pathways as well as the somatosensory and motor tracts passing through the brainstem.
For mapping and monitoring of cranial motor nerve
nuclei, we use the technique similar to the one described
by Chang et al. (1999). Needle or hookwire electrode
pairs are placed in the muscles corresponding to the
CNN at risk for injury (Table 3). At a minimum, we will
monitor bilateral masseter (CN V), orbicularis oculi and
oris (CN VII), and trapezius (CN XI) (Fig. 3). The trapezius is used mainly as a control muscle and is useful in
Fig. 3. Needle recording electrodes placed into the orbicularis oris and masseter muscles. The needle electrodes were
later taped onto the skin in a similar fashion as the ones in
the orbicularis oris.
3 ms
50 V
3 ms
50 V
3 ms
100 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
Fig. 4. Stimulation of CN XI (1.1 mA and 0.05 ms duration) as a test of the stimulation and recording system prior
to brainstem stimulation. Traces 16 are EMG recordings
from the following muscles: left masseter, left orbicularis
oris, left trapezius, right masseter, and right orbicularis oris,
respectively. Trace 6 is recorded from the right posterior
pharynx (for CN IX/X activity).
differentiating neurotonic discharges from electrical

noise. In addition, if the glossopharyngeal nerve is
exposed, it can be used to test the stimulation set-up
before the actual stimulation of the brainstem. This will
ensure that the stimulation and recording capabilities of
the equipment and the electrodes are functioning properly (Fig. 4). With regard to monitoring of CNs III, IV,
VI, IX, and X, it is my opinion that placement of extraocular muscle and posterior pharyngeal electrodes
should be performed by physicians experienced in the
insertion of electrodes into these sites.
Baseline EMG recordings from appropriate muscles are made prior to the start of surgery, and continuous monitoring is performed until surgical closure.
At intervals during the surgery, when desired by
the operating surgeon, a small sterile monopolar
electrode probe (Xomed Surgical Products) is placed
on the brainstem in areas suspected to correspond to
nondiseased CNN, and a small electrical stimulus is
applied. Stimulation intensity begins at 0, and is
increased at 0.1-mA increments until a threshold
response is identified. Then, the stimulation is
increased until a maximum CMAP is seen or EMG
activity from another motor nucleus is triggered. We
will then use this maximal stimulation level to search
for excitable tissue in the surgical target area. By
observing the stimulus intensities required for a threshold response or a maximum CMAP, the approximate
locations of the CNN are determined. Over the years,
the stimulus intensities we have used range from 0.1
355
to 10.2 mA, with the stimulus duration varying from

0.02 to 0.1 ms. However, triggered EMG responses
are usually obtained with stimulus intensities below
3 mA and a stimulus duration of 0.020.1 ms. The
stimulation is presented as a single pulse with a rate
of 1.1 Hz. Such a slow stimulation frequency rate
reduces the overall number of stimuli to the brainstem,
but in order to acquire responses this technique
requires that the surgeon has the patience to hold the
stimulating electrode at each site of interest for a few
seconds. If the surgeon prefers to search for excitable
areas by quickly moving the probe across the brainstem, then the stimulation frequency should be
increased. We have found that stimulation frequencies
of 6.1 and 7.1 Hz work well in those situations.
23.11. Monopolar versus bipolar stimulation
Although we have no experience using bipolar stimulation at the brainstem, the literature suggests that either
technique, monopolar or bipolar, is useful and provides
adequate stimulation to activate the motor nuclei and
tracts. However, in one study bipolar electrodes were
reported to be difficult to use for mapping and that a
monopolar handheld probe defined the stimulation area
more easily (Strauss et al., 1993).
23.12. Constant current versus constant
voltage stimulation
There appears to be no significant differences in the
ability to easily and reliably obtain triggered EMG
responses when using either of these two techniques.
Strauss et al. (1993) reported that in cases using constant current, shunting effects were noted as a result
of the presence of cerebrospinal fluid but that a preliminary review of their data revealed no differences
between these two stimulation techniques. In our
experience, we have also found no appreciable differences between the two.
23.13. Special technical considerations
Simultaneous IOM of somatosensory, brainstem, and
motor evoked potentials should be performed in conjunction with continuous EMG monitoring. This may
present some difficulty if one is using older IOM
equipment, because of the limited number of channels. However, newer equipment typically has a sufficient number of electrode inputs and channels to be
356
PEZ
J.R. LO
Fig. 5. A: Monopolar stimulation on the floor of the fourth ventricle in a 46-year-old male with history of multiple episodes
of vertigo with new onset right facial numbness and right VI nerve paresis but no facial weakness. Stimulation at site in
B: localized the right facial nerve motor fibers immediately superficial to the site of the cavernous malformation. Resection
was not performed.
able to monitor the evoked potentials mentioned as

well as multiple, bilateral CNs.
With regard to the stimulating electrodes, there are
several commercially available monopolar electrodes
that function equally well. However, the one which
our surgeons prefer to use is a monopolar stimulating
probe with a ball end, attached by a small flexible wire
to the handle (Medtronic Xomed Surgical Products;
SKU 8225251) (Fig. 5A and B). Prior to stimulating
any neural structures, the stimulation system should
be tested to see if electrical current is being delivered.
A simple and easy method to do this is by stimulating a

muscle adjacent to the surgical site. The surgeon
should be able to visually confirm the presence of muscle contraction in relation to the stimulation. It is
important to remember that to reduce current spread
and not activate other more distant brainstem regions,
the surgeon should take special care to keep the area
to be stimulated clear of blood and fluids.
The types of recording electrodes that can be used
include surface disk or disposable electrodes commonly used for diagnostic nerve conduction studies,
Fig. 6. A: Placement of a pair of hookwire electrodes into the right genioglossus muscle. B: Note two small wire electrodes
which are still in the needle which is used to introduce the wires into the muscle. The wire electrodes taped above were
placed in the right posterior pharyngeal wall.
357
conventional subdermal needle electrodes, or intramuscular wire (also referred to as hookwire) electrodes.
We have found that the needle and hookwire electrodes
seem to provide the best responses without decay in the
signal characteristics over lengthy surgical cases. In
addition, when properly secured, these electrodes will
easily stay in place. The biggest technical problem we
have encountered is that the electrodes may inadvertently be pulled out during the positioning of the patient.
For this reason, we usually wait until the patient is fully
positioned and placed in the headholder device prior to
placing the electrodes into the masseter, orbicularis
oculi, orbicularis oris, trapezius, and genioglossus muscles (Fig. 6A and B). Unfortunately, this cannot be done
for electrodes inserted into the posterior pharyngeal
wall, which require placement with the aid of a laryngoscope. We usually place these electrodes with the assistance of the anesthesiologist who will hold the patients
mouth open, move the tongue to the side, and provide a
light source with the laryngoscope.
23.14. Interpretation of the electrophysiologic
data
Obviously, one cannot obtain preoperative or preincision data regarding the characteristic responses from
brainstem stimulation. Thus, each patient must serve
as his own control. If the brainstem lesion lies primarily on one side, one should stimulate the opposite
side and record the triggered EMG response. This
will provide an estimate of the stimulus intensity
needed to obtain the EMG signal, as well as, some
of the characteristics of the recorded potential, such
Fig. 7. Magnetic resonance imaging (MRI) showing the

location and extent of the cavernous malformation and
the associated hemorrhage.
4 ms Trig
50 V
Amp 1
1
N2
4 ms Trig
Amp 1
50 V
4 ms Trig
Amp 2
50 V
4 ms Trig
Amp 3
100 V
4 ms Trig
50 V
Amp 4
4 ms Trig
50 V
Amp 2
4 ms Trig
200 V Amp 3
4 ms Trig
50 V Amp 4
P1
Off
Fig. 8. A: Preresection monopolar brainstem stimulation at 0.2 mA, 0.05 ms duration, and 1.1 Hz using single stimuli.
Traces 14 correspond to triggered EMG recordings from the following muscles: left masseter, left orbicularis oris, right
masseter, and right orbicularis oris, respectively. B: Postresection monopolar brainstem stimulation, using the same stimulation parameters and the same muscles for EMG recordings.
IV
III
II
.
r
c
III
IV
I
N1
1
II
N1
N1 P1
2
N1 P1
P1
N1
N1
I
r
c
P1
V
III
IV
III
IV
II
P1
P1
1
III
I
5
II
II
N1
N1
I
3
N1
III
II
N1
P1
N1
IV
N1
N1
P1
P1
P1
8
5
P1
P1
P1
Fig. 9. A: Baseline brainstem auditory evoked potentials (BAEPs) after left and right ear stimulation (traces 1 and 5, respectively); cortical somatosensory evoked potentials (SEPs)
after left (traces 2 and 3) and right (traces 6 and 7) median nerve stimulation. Traces 4 and 8 were recorded from the C7 cervical spinal level. B: During resection, note loss of left
median nerve SEP (trace 2) without changes in left BAEP (trace 1), right BAEP (trace 3), or right median nerve SEP cortical recording (trace 4). C: After resection, note recovery of
left median nerve cortical SEP (traces 2 and 3).
N1
P1
N1
359
N2
N2
1
N2
P1
N2
P2
N2
N1
P1
P2
P2
P1
P2
P1
P2
P2
N2
P1
N1
P1
N2
N2
N2
N1
P1
P2
P1
N1
P1
N2
P2
N2
P1
N1
P2
P1
P2
P2
7
N2
P1
P2
8
Fig. 10. A: Baseline somatosensory evoked potentials (SEPs) after lower extremity stimulation. Traces 13: cortical SEPs
after left posterior tibial nerve stimulation; traces 57: cortical SEPs after right posterior nerve stimulation. B: Note smaller
cortical SEP amplitudes after leg stimulation, which is most pronounced after left leg stimulation (traces 57).
as latency and amplitude. Mapping should be done

prior to initiating the resection. Once an EMG
response is obtained, the stimulus intensity should
be adjusted in order to obtain a threshold and a maximal response. Latencies and amplitudes should be
recorded for both levels. Although it is unclear what
constitutes a significant or critical change in the triggered EMG potentials occurring during or following
resection, one study reported a patient developed a
transient VI and VII CN palsy after postresection
stimulation revealed an 80% amplitude reduction in
the EMG response when compared to the preresection baseline (Chang et al., 1999). In our experience,
we have found that patients do not develop new CN
deficits if the postresection triggered EMG responses

remain unaltered from baseline. However, if the
CMAP amplitudes are smaller or if higher stimulus
intensities are required in order to obtain the CMAP
after postresection stimulation, these patients have a
higher probability of developing a new CN deficit,
or a worsening of a preexisting CN deficits. In addition, in these patients, it cannot be ascertained whether
the deficit will be transient or permanent. However, in
our experience, in those patients where the CMAP cannot be registered after resection, the probability of a
new and permanent deficit is the highest. We have
not appreciated any significant changes in latencies,
other than those secondary to temperature. This is
PEZ
J.R. LO
360
c
1
20 ms
100 V
20 ms
100 V
20 ms
100 V
20 ms
100 V
20 ms
200 V
20 ms
200 V
Fig. 11. Transcranial electrical motor evoked potentials

(TcMEPs): traces 13: EMG recordings from left first
dorsal interosseus, left tibialis anterior, and left abductor
hallucis muscles, respectively. Traces 47 show EMG
recordings obtained from the same muscles on the right.
Note lack of left arm and leg muscle activity after stimulation. Baseline TcMEPs originally showed symmetrical
responses bilaterally.
activity of short duration which coincided with the

start and end of a surgical manipulation and was similar to the EMG bursts typically seen during facial
nerve monitoring. They also described pathological
spontaneous activity (PSA) and classified it as slight
PSA if it lasted for a few seconds or extreme PSA if
it persisted for several hours after the surgical manipulation had been terminated. As a result of these findings, the surgeons dissection strategy was guided
and modified based on the amount of audible EMG
produced by surgical manipulation. One of their
patients developed transient, followed by gradually
permanent EMG discharges in the lingual muscles.
This patient suffered from severe bilateral and permanent tongue weakness. In the remaining 15 patients,
the surgeons modified their dissection strategy by
reducing pressure or traction on tissue, or by changing
the direction and location of dissection based on the
occurrence of EMG discharges. They felt that this
resulted in EMG activity of shorter duration and lower
amplitude, and in these patients new postoperative
deficits were mild, transient, or absent.
23.15. Summary
The literature and electrophysiologic techniques
reviewed indicate that brainstem electrical stimulation
for localization of motor CNN is safe and reliable.
Brainstem stimulation is currently the only technique
available which can identify CN motor pathways in
the operative field and can guide the neurosurgeon in
where to place the incision on the brainstem. Although
there are no outcome studies using these techniques, it
is generally considered advantageous to utilize brainstem stimulation in these procedures, especially since
the floor of the fourth ventricle is often distorted.
Many neurosurgeons and neurophysiologists consider
its use to be indispensable.
23.16. Clinical examples
23.16.1. Case 1
important to keep in mind, especially if the surgery

is done under hypothermia.
The literature also does not provide clear criteria
on how to interpret neurotonic discharges. In our
practice, we use a strategy similar to that described
by Eisner et al. (1995). In addition to the triggered
EMG brainstem mapping, they used a standard audible and video-displayed EMG to monitor neurotonic
discharges. They defined contact activity as EMG
This case follows a 40-year-old male with history

of migraine headaches diagnosed with a pontine cavernous malformation after an MRI was obtained
(Fig. 7). He was followed over a period of two years
and was taken to surgery after he suffered two
hemorrhages, which resulted in mild right face and
left body numbness, mild right facial weakness, and
diplopia. IOM using multimodality recordings of
361
arm and leg (Fig. 11). After reducing traction and

altering slightly the direction of the dissection, there
was a gradual improvement in the SEPs and their
amplitudes returned to near baseline levels. By the
end of the case, the TcMEPs from the left side
remained absent. No significant neurotonic discharges were seen during the case and the stimulation
parameters and evoked EMG responses remained
unchanged from baseline (Fig. 8B).
Postoperatively, the patient awakened with a left
hemiplegia but no new or worsened CN abnormalities.
His weakness quickly began to improve and by one
week he displayed only mild left leg weakness. Four
months after surgery, he was neurologically normal.
bilateral median and posterior nerve SEPs, bilateral

BAEPs, TcMEPs, continuous EMG monitoring of
bilateral CNs V and VII, and brainstem mapping of
CN motor nuclei was performed.
Prior to resection, brainstem stimulation was performed which identified the brainstem trigeminal
motor pathways via triggered EMG responses from
the right masseter muscle (Fig. 8A). During the
resection, there was a loss of the cortical SEP after
left median nerve stimulation and a reduction in the
amplitudes of the cortical SEPs after posterior tibial
nerve stimulation, but most prominent after left leg
stimulation (Figs. 9 and 10). In addition, there was
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
r
c
1
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
3 ms
50 V
Fig. 12. EMG recordings after brainstem stimulation. Traces 15 correspond to the following muscles: left lateral rectus,
left orbicularis oris, right masseter, right lateral rectus, and right orbicularis oris, respectively. A: Preresection monopolar
brainstem stimulation at 0.3 mA, 0.05 ms duration, and 5.1 Hz using single stimuli. Compound motor action potential
(CMAP) recorded from the right masseter. B: Increasing the stimulation intensity to 0.7 mA while maintaining the other
stimulation parameters unchanged reveals triggered responses from the right masseter and right orbicularis oris. C: Slight
movement of the monopolar probe along the brainstem surface and away from the area of hemorrhage triggers a large
CMAP from the right masseter and smaller responses from the right lateral rectus and the right orbicularis oris. Stimulation
parameters were the same as those used in (B). D: Using the same stimulation parameters as in (B) and (C) and stimulating
inferiorly to the site at (C), reveal a small but isolated CMAP from the right orbicularis oris.
362
23.16.2. Case 2
A 15-year-old female was admitted after experiencing
a 2.6 2.9 cm pontine hemorrhage secondary to a
brainstem tumor. Initial physical examination revealed
multiple cranial neuropathies, including abnormal
extraocular movements, asymmetric pupils, and
absent gag and cough. She was also quadreparetic
and her muscle stretch reflexes were hyperactive.
IOM using multimodality recordings of bilateral
median and posterior nerve SEPs, bilateral BAEPs,
TcMEPs, continuous EMG monitoring of bilateral
CNs VI and VII, and right V, as well as, brainstem
mapping of these CN motor nuclei was performed
(Fig. 12). These stimulation examples demonstrate
that important CN motor pathways can be localized
even in patients with significant CN palsies. There
were no significant intraoperative changes in the sensory or motor evoked potentials. Postoperatively, the
patient exhibited no new neurological deficits.
References
Baker, GS (1965) Physiologic abnormalities encountered
after removal of brain tumors from the floor of the fourth
ventricle. J. Neurosurg., 23: 338343.
Chang, SD, Lopez, JR and Steinberg, GK (1999) Intraoperative electrical stimulation for identification of cranial
nerve nuclei. Muscle Nerve, 22: 15381543.
Chang, SD, Lopez, JR and Steinberg, GK (2000) Reply:
identification of cranial nerve nuclei. Muscle Nerve,
23: 1446.
Eisner, W, Schmid, UD, Reulen, HJ, Oeckler, R, OlteanuNerbe, V, Gall, C and Kothbauer, K (1995) The mapping
and continuous monitoring of the intrinsic motor nuclei
during brainstem surgery. Neurosurgery, 37: 255265.
Fahlbusch, R and Strauss, C (1991) Surgical significance of
cavernous hemangioma of the brainstem. Zentralbl.
Neurochir., 52(1): 2532.
Harnack, D, Winter, C, Meissne, W, Reum, T, Kupsch, A
and Morenstern, R (2004) The effects of electrode
material, charge density and stimulation duration on the
safety of high-frequency stimulation of the subthalamic
nucleus in rats. J. Neurosci. Methods, 138: 207216.
Heffez, DS, Zinreich, J and Long, DM (1990) Surgical
resection of intrinsic brainstem lesions: an overview.
PEZ
J.R. LO
Katsuta, T, Morioka, T, Fujii, K and Fukui, M (1993) Physiological localization of the facial colliculus during
direct surgery on an intrinsic brainstem lesion. Neurosurgery, 32: 861863.
Lang, J, Ohmachi, N and Lang Sen, J (1991) Anatomical
landmarks of the rhomboid fossa (floor of the 4th ventricle), its length and width. Acta Neurochir. (Wien),
113: 8490.
Lassiter, KRL, Alexander, E, Davis, CH and Kelly, DL
(1971) Surgical treatment of brainstem gliomas. J. Neurosurg., 34: 719725.
McCreery, DB, Agnew, WF, Yuen, TGH and Bullara, L
(1990) Charge density and charge per phase as cofactors
in neural injury induced by electrical stimulation. IEEE
Trans. Rehab. Eng., 37: 996999.
148: 499509.
Morota, N, Deltis, V, Epstein, FJ, Kofler, M, Abbott, R,
Lee, M and Ruskin, K (1995) Brainstem mapping: neurophysiologic localization of motor nuclei on the floor
of the fourth ventricle. Neurosurgery, 37: 922930.
Morota, N, Deltis, V, Lee, M and Epstein, FJ (1996)
Functional anatomic relationship between brainstem
tumors and cranial motor nuclei. Neurosurgery, 39:
787794.
Rusyniak, WG, Ireland, PD, Radley, MG, Pilcher, WH and
Okawara, SH (1992) Ultrasonographic and electrophysiological adjuncts to surgery within the brainstem: technical note. Neurosurgery, 31: 798801.
Shannon, RV, Moore, JK, McCreery, DB and Portillo, F
(1997) Threshold-distance measures from electrical
stimulation of human brainstem. IEEE Trans. Rehab.
Eng., 5: 7073.
Strauss, C, Romstock, J, Nimsky, C and Fahlbusch, R
(1993) Intraoperative identification of motor areas of
the rhomboid fossa using direct stimulation.
J. Neurosurg., 79: 393399.
Strauss, C, Romstock, J and Fahlbusch, R (1994) Intraoperative mapping of the floor of the IVth ventricle. In: CM
Loftus and VC Traynelis (Eds.), Intraoperative Monitoring Techniques in Neurosurgery. McGraw-Hill,
Strauss, C, Romstock, J and Fahlbusch, R (1999) Pericollicular approaches to the rhomboid fossa. Part II. Neurophysiological basis. J. Neurosurg., 91: 768775.
Steinberg, GK, Chang, SD, Gewirtz, RJ and Lopez, JR
(2000) Microsurgical resection of brainstem, thalamic,
and basal ganglia angiographically occult vascular malformations. Neurosurgery, 46: 260270.
Section II.4
Electromyographic, Reflex and Nerve Conduction Monitoring

M.R. Nuwer (Ed.)
364
CHAPTER 24
Intraoperative peripheral nerve

stimulation and recording
Brian A. Cruma,* and Jeffrey A. Strommenb
a
Department of Physical Medicine and Rehabilitation, Mayo Clinic College of Medicine, Rochester, MN 55905, USA
24.1. Introduction
Electrophysiological monitoring of peripheral nerves
during surgery is an extremely valuable procedure that
provides vital, real-time information to the surgical
team. Preoperative electrophysiological testing provides the surgeon with valuable data to assist with decision making; however, there is information that simply
cannot be garnered from these studies intraoperative
studies help bridge this gap. Monitoring of peripheral
nerves has been described and studied for nearly
40 years (Kline et al., 1969; Kline and Happel, 1993)
and has been the subject of several reviews (Terzis
et al., 1976; Brown and Veitch, 1994; Slimp, 2000;
Spinner and Kline, 2000; Harper, 2005). Intraoperative
monitoring (IOM) requires a skilled and knowledgeable
team (comprising physicians and technicians) who
ensure rapid and accurate acquisition and interpretation
of electrophysiological data. It also requires a good
working relationship with the entire surgical team.
Communication with the surgeon(s) is vital in assisting
with surgical decision making between sometimes quite
divergent surgical plans. Accurate and timely communication with the anesthesiologist is also important as
certain anesthetic agents can have detrimental effects
on electrophysiological studies. Having a peripheral
nerve IOM monitoring plan prior to surgery is ideal to
maximize the efficiency and utility of the studies.
The uses for intraoperative peripheral nerve monitoring include:
Identifying peripheral nerve.
Localizing preexisting disease processes along the
course of a nerve.
*
Correspondence to: Brian A. Crum, M.D., Department

of Neurology, Mayo Clinic College of Medicine, 200 First
Street SW, Rochester, MN 55905, USA.
Tel.: +1-507-538-7391; fax: +1-507-284-4795.
E-mail: crum.brian@mayo.edu (B.A. Crum).
Determining continuity across a preexisting lesion,
prior to detection by other electrophysiological

means.
Determining likelihood of nerve root avulsion.
The goal of this chapter is to provide a background
in the equipment, techniques, and interpretation
of intraoperative peripheral nerve stimulation and
recording. For the purpose of this review, the term
peripheral nerve will apply to nerve that is distal to
the spinal cord, usually also distal to the intervertebral
foramen (spinal root/nerve, plexus, peripheral nerve).
24.2. Nerve conduction study techniques
Nerve conduction studies are performed with stimulation and recording of motor, sensory, or mixed nerves,
yielding a nerve action potential (NAP). Potentials can
also be recorded over muscle [compound muscle
action potentials (CMAPs)], or centrally over spine
or scalp (EPs). Many factors will dictate which of
these (or what combination) are used, which will then
dictate the type of equipment necessary and the regimen of anesthesia that should be utilized.
Intraoperative peripheral nerve stimulation is given
directly to the surgically exposed nerve (Fig. 1).
A bipolar stimulator is held in place by the surgeon
with cathode directed toward the recording electrodes.
The interelectrode distance is usually 3 mm, though
this distance is dictated by the size of the nerve. A
larger nerve such as the sciatic demands a larger interelectrode distance, such as 7 mm. The proper orientation of the stimulating electrodes is important and
visual confirmation of correct placement by the IOM
physician or technician in the operating room is useful.
Monopolar stimulation with cathode on the nerve and
the anode some distance away can be done when the
proper bipolar orientation is impossible. This should
generally be avoided, however, as stimulation cannot
EMG, REFLEX AND NERVE CONDUCTION MONITORING
365
Fig. 2. Bipolar electrodes. Left: pointed tip electrodes

often used for stimulation. Right: hook electrodes often
used for recording.
Fig. 1. Intraoperative peripheral nerve bipolar stimulation

(bottom) and recording (top).
be as precisely focused, leading to current spread to

adjacent nerves and muscles, and potentially farther
down the nerve than desired. Bipolar stimulation
affords less current spread. A tripolar stimulating electrode can also be used (Kline and Happel, 1993; Tiel
et al., 1996). In this arrangement, a single cathode is
between two anodes, further focusing the site of stimulation while minimizing stimulus artifact. Special
stimulating electrodes, some with hooks or very small
pointed tips often need to be used (Fig. 2). Hook electrodes allow the nerve to be elevated out of the surgical
field to avoid contact with excessive fluid reducing
current spread. Our stimulating electrodes are made
of a silver solder alloy, and are gas sterilized or autoclaved after surgical procedures.
The amount of stimulation required to depolarize
the underlying axon is much less than that needed for
percutaneous stimulation. Excessive stimulation must
be avoided to limit current spread down the stimulated
nerve and to other nearby nerves and muscles and also,
to decrease the stimulus artifact which is amplified

with a shorter distance between the stimulating and
recording electrodes. A square-wave pulsed stimulus
with duration of 0.05 ms and intensity of only a few
milliamperes (15 mA) is usually sufficient for depolarization of the axonal membrane in a supramaximal
fashion. Of note, short duration stimulation of this type
is more likely to preferentially activate motor axons
(Veale et al., 1973). For constant voltage stimulation,
intensities of 2550 V are used (Nelson, 1988). It is
important to remember that in diseased nerves, higher
stimulus intensity may be needed to reach threshold. It
is our practice to increase the stimulus intensity up to
2025 mA, if no response is initially obtained. These
higher intensities, however, result in more stimulus
artifact and more likely evoke volume conducted
responses than true NAPs (see below). As the size of
the NAP is small (microvolts), averaging of one to five
responses is often helpful, though when a response is
present it is usually visible after the first stimulus
(Fig. 3).
Nerve action potential recording takes place along
the course of the stimulated nerve. The distance
between the cathode and the active, recording electrode should be at least 4 cm to minimize stimulus artifact. Recording is done with a bipolar arrangement
with an interelectrode distance of 35 mm. A wider
separation is used when there is a longer distance
between the cathode and recording electrode, allowing
the longer traveling wave to completely pass under the
active electrode before passing under the referential
electrode. The farther the separation, however, the
366
B.A. CRUM AND J.A. STROMMEN
TD1:EI1 1:Pair1
TD2:EI2 1:Pair1
T
r
c
N:
0.5 ms
2 V
NR:
Normal
Amp 1
NAP
Fig. 3. Nerve action potential with a peak latency of about 2 ms.
more likely extraneous noise and stimulus artifact will

be seen as different by the two recording electrodes
and, thus, not rejected or canceled out in common
mode. The nerve may need to be lifted out of the surgical field similar to the site of stimulation. The recording electrodes are held in place by the surgeon and
again visual confirmation of the correct orientation
(active electrode toward the cathode) is important.
The size of the recording electrodes should match the
size of the nerve being recorded from. A three-pronged
electrode with ground-active-referential arrangement
has also been used. The same electrodes can be used
for either stimulating or recording, depending on
whether they are connected to the amplifier or stimulator. We generally use the bipolar point electrodes for
stimulation and the curved hook electrodes for recording (Fig. 1). The ground electrode is a flat metal plate
that is placed under the patient and separate from the
ground used for cautery.
As the NAP waveforms contain frequencies in the
1 kHz range, using filter settings of 510 Hz for
the low frequency (high pass) filter and 23 kHz
for the high frequency (low pass) filter are appropriate.
Occasionally, it is helpful to increase the high frequency filter to better separate the NAP waveform
from the stimulus artifact (Tiel et al., 1996). The
NAP amplitude is usually less than 100 mV, so a gain
of 25 mV per division is utilized. The latency will
obviously depend on the length between stimulating
and recording electrodes. A rough guide is 1 ms/5 cm
distance (assuming 50 m/s conduction velocity).
A time base of 0.5 ms per division is reasonable,
increasing this for longer distances.
When the goal of intraoperative peripheral nerve
studies is to determine the degree of continuity or the
exact location of a peripheral nerve lesion, stimulation
and recording will need to be done on both sides of the
lesion. When assessing for continuity, it is important to
realize that the presence of only few large myelinated

axons can produce a response with relatively normal
conduction velocity, latency, and threshold; therefore,
it is most useful to assess the amplitude of the NAP to
determine the number of functioning axons across a
lesion. A NAP proves the existence of a large number
(over 4,000) functioning, medium-sized, myelinated
axons (Kline et al., 1969). When assessing continuity,
stimulation is usually done proximal to the lesion with
recording distally. When attempting to localize
lesions, recording electrodes are placed proximally
and the stimulating electrodes are moved proximal
to, then into, then distal to the suspected lesion, assessing for a change in morphology of the waveform
when a step is made into or across the lesion.
Recording of CMAPs has the advantage of amplification of responses as each axon innervates and
activates hundreds to thousands of muscle fibers.
Amplitudes are measured in millivolts, as opposed to
microvolts in NAP and EP recordings. CMAP recording is performed with surface (as in routine nerve
conduction studies), subcutaneous, or intramuscular
electrodes (Fig. 4). Subcutaneous recordings are done
with electroencephalographic (EEG) needles placed
above or into the muscle of interest. Fine, longer, intramuscular wires placed with the use of a hollow needle
can also be employed. Both intramuscular and subcutaneous recordings limit the size of the recording area,
thereby reducing extraneous noise in the recording.
Intramuscular recordings, though, do this to a much
greater degree only recording from a small part of a
muscle and, therefore, reflect activity in only a fraction
of axons that might be stimulated intraoperatively.
They also tend to introduce more noise than subcutaneous EEG electrodes. They are not suitable for
IOM when one is interested in the amount of innervation to a muscle. Their use is most often for deep muscles or those muscles that are smaller or difficult to
Fig. 4. Electrodes used for compound muscle action potential (CMAP) recording. Left: surface electrode. Middle:
EEG electrode for subcutaneous recording. Right: fine intramuscular wire electrode introduced in hollow needle which is
then removed with fine wire hooked into place in muscle.
selectively record from (i.e., rhomboids, laryngeal

muscles). Subcutaneous EEG recordings are favored
at our institution, Mayo Clinic College of Medicine,
because of the ease of placement and quiet recordings.
24.3. Technical problems
Given the importance of acquiring accurate information quickly as it may guide surgical decision
making, several potential technical problems should
be addressed. Low temperature of nerves is inevitable in IOM. Low temperatures lead to slowed
conduction velocities and higher amplitudes. As
increasing temperature is not an option during IOM,
this effect must simply be kept in mind. Most analysis is done by looking at the presence or the absence
of a potential, or the change in a potential at nearby
recording sites. These parameters are unlikely to be
significantly affected by the cool temperatures during
IOM. Peripheral ischemia from a blood pressure cuff
can impair nerve conduction studies and it is recommended that if a tourniquet is in place for 60 min in
the limb of study, it should be let down for at least
20 min before beginning IOM studies (Brown and
Veitch, 1994). The operating room is an electrically
hostile environment. Surgical instruments, beds,
machines, and lights can all contribute, especially in
the form of 60 Hz interference. Limiting fluorescent
lights and any electrical motors or cautery devices
during recording is helpful.
367
Stimulus artifact can be a challenge. Ensuring the

electrodes are lifted out of a wet field is important.
One must use the lowest intensity stimulus at the least
duration possible to achieve supramaximal stimulation. The distance between stimulation and recording
electrodes can be increased, if exposure in the surgical
field permits. One can also arrange the recording
electrodes in a monopolar fashion with G2 in adjacent
tissue perpendicular to G1. Muscle artifact can also
distort the NAP. This is usually caused by volume
conduction from activation of nearby muscles. Proper
orientation of the electrodes and grounding must be
assured. The placement of the stimulating and recording electrodes can be reversed, stimulating distally.
This can help to reduce the amount of activation of
nearby muscles innervated by more proximal branches
from the nerve being studied. Neuromuscular blocking
agents can be utilized to eliminate this artifact as well.
When a NAP is not obtained, or there is difficulty
getting the response, it is useful to record a NAP over
a normal portion of nerve. This acts as a control to
ensure functioning of the whole IOM system. If there
is no visible twitch in a downstream muscle and no
elicited electrical response, one must also ensure that
the stimulation is indeed occurring. The stimulating
electrodes then are placed on a nearby known functioning nerve or muscle to assess for twitch. Even
in the setting of neuromuscular blocking agents,
direct stimulation of muscle should result in a twitch.
Nerve action potential and CMAP recordings are
minimally affected by anesthesia. If CMAPs are being
recorded, neuromuscular blocking agents should be
minimized or not used at all. If, however, a neuromuscular blocking agent is used, a peripheral stimulator
with a train-of-four recording is done to monitor the
degree of block (amplitude change from the first to
the fourth stimulus). CMAP responses can still be
obtained if there is a less than 100% decrement over
the four successive supramaximal repetitive nerve stimulations (Holland, 2002). As mentioned above, neuromuscular blocking agents may be desirable for NAP
studies in which muscle artifact must be eliminated.
24.4. Application to peripheral nerve lesions
Intraoperative peripheral nerve monitoring can be
used to determine continuity across an injured segment of nerve or for precise localization of a peripheral nerve lesion. Following an axonal injury,
Wallerian degeneration will occur back to the level
of the nerve injury. Axonal regeneration then may
368
proceed from the lesion, moving distally at a rate of

about 1 mm per day. Following this, reinnervation
of the end organ must occur for functional recovery.
After axons reach a muscle, there may be a several
week to month delay before this reinnervation can
be detected. This is usually done first by electrophysiological means (nascent motor unit action potentials)
and then by visible voluntary contraction (Kline
et al., 1992). Depending on the distance from lesion
to end organ, therefore, a variable period of time
must then pass to see clinical and electrophysiological signs of reinnervation. By the time one concludes
there is no reinnervation by these grounds, the opportunity for a surgical approach to reinnervation (i.e.,
nerve grafting) may be lost as the best results are
obtained if surgery is performed in the first 6
12 months. In a proximal sciatic lesion, for example,
that much time or more would be required to conclude there is no reinnervation to the anterior compartment of the leg from a clinical or a standard
electrophysiologic measurement.
In this setting, therefore, only intraoperative assessment can make a timely determination of whether a
lesion is complete or incomplete. Visual inspection is
of great importance in an anatomically continuous
lesion as nerve thickening and neuroma formation
may indicate a complete lesion. There is, however,
no visual way to reliably determine functional continuity of axons, especially regenerating ones. Eliciting a
muscle twitch downstream from a site of peripheral
nerve stimulation indicates continuity of that section
of nerve, though a response from direct muscle stimulation must be excluded. As absent response, though,
could mean either a lesion in continuity (without muscle reinnervation) or a complete lesion. The presence
of a NAP across a lesion, therefore, remains the gold
standard to determine nerve continuity and, therefore,
the type of surgical approach. It is clear that a NAP
can be obtained in the setting of a lesion that appears
complete clinically and by routine, preoperative electrophysiological testing (Kline and Happel, 1993). In
addition, intraoperative CMAP recordings may not
show a response given the lack of time to reinnervate
the muscle. Only a few functioning axons are needed
to elicit a CMAP, which may not be functionally useful. For these reasons, CMAP recordings are usually
not helpful in determining continuity. Incomplete
lesions in continuity are treated with neurolysis and
are not grafted and have a high likelihood of a good
outcome. About 90% of those patients experience a
good outcome in the large experience of Kline (Kline
and Happel, 1993). Lesions with no NAP transmitted

are thought to have no chance of spontaneous reinnervation and are thus treated by grafting or nerve transfer
(Kline and Happel, 1993; Spinner and Kline, 2000).
A neurapraxic lesion that does not completely
block all conduction through a lesion will lead to
responses being recorded with both proximal and
distal stimulation (across a lesion). The response
with proximal stimulation may be longer in duration
(temporal dispersion), lower in amplitude (partial
conduction block), and arrive at the recording electrodes with a longer latency than would be expected
(focal slowing). Inching is a technique of stimulation
at short, incremental steps across a lesion making
assessments of the change in morphology and
latency of the evoked waveforms recorded distally
(Campbell, 1998). If one inches at 1 cm increments,
assuming a conduction velocity of 50 m/s, then the
onset of each successive waveform should be separated by 0.2 ms. As the main goal of inching is to
more precisely localize focal lesions, it is useful in
the intraoperative setting when this localization is
not provided by preoperative electrophysiologic
studies. Again, the intraoperative use of NAP is
superior to CMAP to localize lesions (Fig. 5).
An example of this application is in ulnar neuropathy at the elbow (UNE). Preoperative electrophysiological studies may confirm UNE, but not be
precisely localizing to the medial epicondyle/retroepicondylar groove or to the cubital tunnel (Kline
et al., 1992; Harper, 1996). Factors which contribute
to this include variability in anatomic location of
these two potential entrapment sites, selective
involvement of fascicles in the ulnar nerve, technical
difficulties with overstimulation (especially at the
below elbow site), and anastomotic nerve connections. The degree of intraoperative electrophysiological abnormalities is often more severe than expected
based on preoperative routine studies (Kline and
Happel, 1993). Intraoperative studies can even show
abnormalities when routine nerve conduction studies
across the elbow are normal (Kim et al., 2003). Additionally, compression may be at a more distal site than
expected (Campbell et al., 1988, 1992). Intraoperative ulnar nerve studies are, therefore, useful when
localization is needed to guide the type and site of
surgical intervention.
The ulnar nerve is usually easily exposed surgically. If possible, stimulation and recording should be
performed on a normal portion of the nerve to ensure
a working system. Then, recording proximally, the
369
AXONAL LOSS-PARTIAL
AXONAL LOSS-COMPLETE
CMAP
CMAP
Distal
CMAP
CMAP
Distal
3
4
4
in latency (ms)
3
Lesion
Proximal
DNAP
3
Lesion
DNAP
DNAP
2
1
1
in latency (ms)
DNAP
Proximal
Fig. 5. Comparison of compound muscle action potential (CMAP) and direct nerve action potential (DNAP) recordings
over a focal nerve lesion. The location of the lesion is more apparent (left) or only apparent (right) with direct NAP recordings. (Reprinted from Daube and Harper, 1989, with permission from Elsevier.)
stimulating electrodes are placed at intervals closer,

into, and distal to the area in question. An assessment
is made for changes in amplitude, latency, and morphology as stimulation moves into the abnormal segment of nerve. Based on intraoperative ulnar nerve
studies, the site of compression is most commonly at
or just proximal to the retroepicondylar groove/medial
epicondyle (Kline and Happel, 1993; Brown and
Veitch, 1994; Kim et al., 2003). In fact, in the experience from Louisiana State University over a 30-year
period, compressive ulnar neuropathies localized to
the epicondyle level in over 97% of cases (Kim et al.,
2003). Cubital tunnel localization was seen more frequently in other studies (Campbell et al., 1992; Campbell, 1998). Based on these findings, one should begin
stimulating across the epicondylar region and if no
clear abnormalities are seen, then assess the segment
of nerve across the cubital tunnel. If these sites are normal, it is important to realize that rarely the ulnar nerve
can be compressed at sites distal to the cubital tunnel
within or as exiting the flexor carpi ulnaris
(Campbell, 1989; Campbell et al., 1992).
24.5. Summary
Performing high-quality IOM of peripheral nerves is
a challenging endeavor. With the proper equipment,
training, and techniques, vital electrophysiological

information can be obtained quickly, interpreted,
and input given to the surgical team. Specifically,
this includes identification of peripheral nerve,
determination of continuity of peripheral axons,
and localization of peripheral nerve lesions. This
information can then be utilized to maximize the
effectiveness and safety of surgical intervention.
References
Brown, WF and Veitch, J (1994) AAEM minimonograph
#42: intraoperative monitoring of peripheral and cranial
nerves. Muscle Nerve, 17: 371377.
Campbell, WW (1989) AAEE case report #18: ulnar neuropathy in the distal forearm. Muscle Nerve, 12: 347352.
Campbell, WW (1998) The value of inching techniques in
the diagnosis of focal nerve lesions. Muscle Nerve, 21:
15541556.
Campbell, WW, Pridgeon, RM and Sahni, SK (1988) Entrapment neuropathy of the ulnar nerve at its point of exit
from the flexor carpi ulnaris muscle. Muscle Nerve, 11:
467470.
Campbell, WW, Pridgeon, RM and Sahni, SK (1992)
Short segment incremental studies in the evaluation of
ulnar neuropathy at the elbow. Muscle Nerve, 15:
10501054.
Daube, JR and Harper, CM (1998) Surgical monitoring
of cranial and peripheral nerves. In: JE Desmedt (Ed.),
370
Neuromonitoring in Surgery. Elsevier, Amsterdam,
p. 133.
Harper, CM (1996) Peripheral nervous system monitoring.
In: J Daube (Ed.), Clinical Neurophysiology. F.A.
Davis and Co, Philadelphia, pp. 465466.
Harper, CM (2005) Preoperative and intraoperative electrophysiologic assessment of brachial plexus injuries.
Hand Clin., 21: 3946.
Holland, NR (2002) Intraoperative electromyography.
Kim, DH, Han, K, Tiel, RL, Murovic, JA and Kline, DG
(2003) Surgical outcomes of 654 ulnar nerve lesions.
J. Neurosurg., 98: 9931004.
Kline, DG and Happel, LT (1993) A quarter centurys
experience with intraoperative nerve action potential
recording. Can. J. Neurol. Sci., 20: 310.
Kline, DG, Hackett, ER and May, PR (1969) Evaluation of
nerve injuries by evoked potentials and electromyography. J. Neurosurg., 31: 128136.

Kline, DG, Hudson, AR and Zager, E (1992) Selection and
preoperative work-up for peripheral nerve surgery. Clin.
Nelson, KR (1988) Use of peripheral nerve action potentials
for intraoperative monitoring. Neurol. Clin., 6: 917933.
Slimp, JC (2000) Intraoperative monitoring of nerve
repairs. Hand Clin., 16: 2536.
Spinner, RJ and Kline, DG (2000) Surgery for peripheral
nerve and brachial plexus injuries or other nerve
lesions. Muscle Nerve, 23: 680695.
Terzis, JK, Dykes, RW and Hakstian, RW (1976) Electrophysiological recordings in peripheral nerve surgery: a
review. J. Hand Surg., 1: 5266.
Tiel, RL, Happel, LT and Kline, DG (1996) Nerve action
potential recording method and equipment. Neurosurgery, 39: 103109.
Veale, JL, Mark, FR and Rees, S (1973) Differential sensitivity of motor and sensory fibres in human ulnar nerve.

M.R. Nuwer (Ed.)
371
CHAPTER 25
Intraoperative facial nerve monitoring

Yasmine A. Ashrama,* and Charles D. Yinglingb
a
Alexandria School of Medicine, University of Alexandria, Alexandria, Egypt
Stanford University School of Medicine, and Surgical Neuromonitoring Services, Stanford, CA 94305, USA
25.1. Introduction
25.1.1. Scope of this chapter
The use of intraoperative monitoring has significantly
improved postoperative facial nerve (FN) function and
has become an indispensable tool for facilitating surgery involving the FN. This tool, however, must be
properly applied and maintained, in order to maximize
its potential benefits and avoid inconvenient results.
To achieve this goal, a thorough understanding of monitoring techniques is required. This chapter will consider the technical aspects of FN monitoring. The
different techniques used to monitor the FN are discussed with emphasis on how to conduct the technique
and solve common problems to ensure successful monitoring. The practical aspects of instrumentation, electrode placement, stimulation parameters, types of
responses encountered, and artifact identification are
also discussed in detail. However, the use and outcome
of FN monitoring in specific clinical disorders is
beyond the scope of this chapter.
25.1.2. Iatrogenic FN injury
Several intraoperative events can pose an increased
risk of injury to the FN. The nerve could be directly
lacerated during dissection or drilling in close proximity to the nerve. The use of electrocautery, laser energy
or even a diamond drill producing excessive heat may
cause thermal injury to the nerve. Iatrogenic injury can
also occur by stretching the nerve during dissection
which can interrupt the myelin sheath, cause variable
degree of axonal damage, or compromise the blood
supply of the nerve. Another mechanism of FN injury
is compression of the nerve that can cause facial

paralyses by entrapment neuropathy (Monsell, 2005).
Many factors can increase the incidence of iatrogenic FN injury, including congenital anomalies, revision surgery, tumors, and severe inflammatory
processes. These factors may distort normal anatomical
relations, or cause the FN to be thinned and attenuated
thus making dissection around the nerve a more difficult
task (May et al., 2000). The lack of an epineural covering in the intracranial segment of the nerve is an additional factor that makes it more vulnerable to injury
where even mild stretching may produce significant
damage to the nerve. Whatever the mechanism, iatrogenic injury can produce variable effects ranging from
mild transient to severe irreversible damage to the FN.
A mild transient injury may result from interruption of
the myelin sheath covering the nerve leading to delayed
latency of the recorded compound muscle action
potential (CMAP) recorded from the facial muscles.
A more severe injury may be due to interruption of
the nerve axons and results in decrease of amplitude
of the recorded CMAP. The amount of axon loss can
be roughly estimated by comparing the amplitude of
the potential with a previous baseline value. With axonal damage, the axons distal to the injury will undergo
Wallerian degeneration in about 4 days from the time
of injury (Preston and Shapiro, 1998). However, intraoperatively, immediately after the injury, the distal
axons will still be responsive to electrical stimulation.
Therefore, when assessing the integrity of the FN after
an injurious event, it is essential to ensure that the stimulator is placed proximal to the site of injury.
25.2. Technical considerations for FN monitoring
25.2.1. Instrumentation
Correspondence to: Yasmine A. Ashram, M.D., D ABNM,

507 El Horreya Avenue, Fleming, Alexandria, Egypt.
Tel.: 203-585-0828; fax: 203-583-0785.
E-mail: ashramy@yahoo.com (Y.A. Ashram).
The instrumentation requirements for FN monitoring

are basically similar to those used for electrodiagnostic
purposes. The equipment should include an isolated
372
electrical stimulator that can be precisely controlled at

low levels, differential amplifiers of a good quality, filters to reduce noise, a multichannel oscilloscope to display the signal (at least four-channel), an audio monitor,
and the ability to permanently store and printout data.
However, to comply with the needs of the operating
room, there are certain modifications and different features that are required for intraoperative equipment,
especially those involving electrical safety, stimulation
parameters, data collection, and display.
Multiple channel systems are recommended as
they allow monitoring of multiple divisions of the
FN independently, as well as other cranial motor
nerves if needed. Even in cases where only the FN
is at risk, the extra channels can be hooked to the
contralateral facial muscles to provide a valuable
control for nonspecific increases in EMG activity
due to light anesthesia or other nonsurgical factors.
For more complex surgical cases, where auditory
brainstem response (ABR) may be monitored as well
as multiple cranial motor nerves, a multichannel system with averaging capabilities should be used. Such
systems are now commercially available from various manufacturers. These systems are also adaptable
for other types of surgical monitoring (i.e., spinal
surgery) by creating appropriate software templates.
Y.A. ASHRAM AND C.D. YINGLING
Fig. 1. Diagrammatic representation of electrode placement

for facial nerve (FN) monitoring. Paired needle electrodes
are placed in the ipsilateral orbicularis oculi (A) and in the
orbicularis oris muscles (B); needles should be advanced at
a relatively sharp angle because the muscle layer is thin
and superficial. Another paired needle electrode should be
hooked to the contralateral orbicularis oris (C) or the trapezius (D) even if the FN is the only nerve at risk to provide a
valuable control for nonspecific increases in EMG activity
due to light anesthesia or other nonsurgical factors. A monopolar needle electrode is placed on the shoulder as a signal
ground (E).

Needle electrodes are preferred to surface electrodes
for intraoperative monitoring as they are more specific, less prone to artifacts, and their application is less
time consuming than surface electrodes. A variety of
different styles of needle electrodes have been used
including standard EMG electrodes, subdermal EEG
electrodes, and a variety of custom designs. Probably,
the most popular recording electrodes for FN monitoring are platinum or stainless steel subdermal needles
that have a larger uninsulated surface than electrodes
designed for single fiber EMG, therefore are more
likely to detect activity arising anywhere in the desired
muscle. Regardless of the type of electrode used, it is
important to insert the electrodes correctly in the
desired muscle; this is relatively easy when monitoring
the FN as facial muscles are superficial and easy to
identify (Fig. 1). However, care should be taken in
obese patients to keep the needle in the muscle and
not to insert it in overlying fat.
After proper insertion, electrodes should be taped
in place to avoid accidental dislodgment, as it is usually unfeasible to reach their site of insertion after the
patient has been positioned and draped. Electrode

leads should be directed away from the surgical field;
they may also be loosely taped to the patients skin to
avoid any accidental movement of the wires into the
sterile field or pulling out of the electrode at the
insertion site. After the electrodes are secured in
place, they are connected to the respective amplifiers
which should be already switched on, as electrical
surges may develop from switching on the amplifiers
and can reach the patient if connected to the amplifier at that time. Impedance of the recording electrodes should always be checked and should be kept
below 5 kO (Nuwer et al., 1993). If impedance is
high, and/or the recording channel shows increased
noise levels, proper electrode insertion and connection to the amplifiers should be verified.
Various designs of stimulating electrodes have been
used during FN monitoring, each with its advantages
and disadvantages. Monopolar electrodes have a
single stimulating electrode in the hand piece, the tip

of which should be connected to the cathode of the
stimulator; the anodal return or reference is usually a
needle inserted at the periphery of the wound. Several
types of monopolar electrodes have been described,
the most commonly used ones are the Prass flush-tip;
ball tip, and flexible platinumiridium tip (Fig. 2).
With monopolar electrodes, current from the stimulating electrode travels a longer path through the body tissues in order to reach the anodal return which is
relatively far away, thus increasing the likelihood of
current spread. As a result, nerves in close proximity
to each other may be stimulated as a group. It may
therefore be more difficult to precisely locate an
individual nerve or accurately identify nerve fibers
within other tissues when monopolar stimulation is
used (Danke and Wiegand, 1994). On the other hand,
bipolar electrodes do not use a distant reference; thus
current spread is minimized leading to more accurate
localization of neural tissue. However, a major limitation to the use of bipolar electrodes is that the response
373
is greatly affected by variations in the orientation of

the bipolar tips in relation to the long axis of the nerve
(Kartush et al., 1985). Commercially monopolar electrodes are more widely used as they do not have this
disadvantage and if the intensity is near the threshold
level they can provide spatial resolution of less than
1 mm (Yingling and Ashram, 2005b). To overcome
the problem of probe orientation when using bipolar
electrodes, Danke and Wiegand (1994) suggested
using coaxial bipolar stimulation in which the stimulating electrode is entirely and symmetrically surrounded by the reference electrode in a coaxial
configuration.
All the previously described probes are used
entirely for stimulation, and thus dissection has to stop
each time stimulation is used. Kartush et al., (1985)
introduced a set of dissecting instruments that can be
interchangeably connected to the electrical stimulator,
allowing simultaneous dissection with continuous
stimulation. These stimulators are mainly useful for
removing parts of tumor closely adherent to a nerve.
25.2.4. Stimulation parameters used in FN
monitoring
Fig. 2. Different probes commonly used with monopolar

stimulators. A: Highly flexible-tip probe used for intracranial stimulation: The entire probe and the flexible wire
are insulated except for the 0.5 mm ball on the end in order
to achieve localized stimulation. B: Prass flush-tip probe:
the entire probe is insulated to a flush tip to prevent current
shunting. It could be bent so that the central portion of the
tip contacted the desired tissue. C: Ball-tip probe: insulated
except for 1 mm ball on the end; ball tip provides atraumatic contact to neural structures.
Despite the wide use of intraoperative electrical stimulation of the FN, stimulation criteria differ considerably among various centers and there is still no
consensus about the optimum criteria to be applied.
Although concerns have been raised about the safety
of direct electrical stimulation of the FN, the electrical stimulation parameters provided by commercial
equipment for FN monitoring are safe if properly utilized. Electrical stimulators in clinical use deliver
rectangular pulses, the amplitude and duration of
which vary within a wide range. Our preference is
to use monopolar constant voltage stimulation,
delivering pulses of 0.2 ms duration at a rate of
510/s. With these parameters, the threshold for an
evoked EMG response from normal nerves is usually
between 0.05 and 0.2 V, averaging about 0.1 V
(Yingling and Ashram, 2005a).
The issue of whether constant current or constant
voltage stimulators should be used has been a source
of continuing controversy. There is no literature that
demonstrates a clear advantage for either method.
Whether constant current or constant voltage is being
used in different centers is often determined by the
capability of the instrumentation, or with which
method the team has most experience and feels most
comfortable.
374
25.3. FN monitoring techniques

Several techniques for intraoperative FN monitoring
have made their way to clinical use, such as intraoperative EMG, recording of compound nerve action potential (CNAP), facial muscle F wave recording, and
video monitoring. EMG is by far the most widely used
technique, and will be the primary method discussed in
this chapter.
25.3.1. Intraoperative EMG monitoring
25.3.1.1. Principles of intraoperative EMG
There are two major ways in which intraoperative EMG
can be used to monitor the FN. The first is monitoring
mechanically evoked or spontaneous EMG activity.
Such activity is continuously recorded to detect thermal
mechanical or electrical irritation of the FN that can
result from intraoperative events. The second method
is electrical stimulation of the exposed FN and recording an evoked EMG response from the corresponding
facial muscles to locate and identify the FN.
The term mechanically evoked or spontaneous
EMG activity has been used to describe motor unit
activity elicited in the facial muscles in response to
surgical manipulation of the FN. Such activity has also
been called neurotonic discharges, which are high frequency bursts of motor unit action potential (MUAP)
activity generated in response to mechanical or metabolic irritation of the nerve that innervates the muscle
(Harper and Daube, 1998). Electrical stimulation, on
the other hand, produces a synchronous composite
electrical activity within the target muscle called the
CMAP. Measurements of the threshold, latency, and
amplitude of the CMAP are the basis of numerous
studies investigating the prognostic value of intraoperative FN monitoring.
When the FN is stimulated, its fibers are not all
equally excited at the same intensity of current. The
largest diameter fibers are recruited first as they have
lower internal resistance (Robinson, 1995). The minimal stimulus intensity required to produce a response
is known as threshold. Increasing the stimulus above
the threshold will cause recruitment of the smaller
diameter fibers (York, 1992), resulting in increase in
the amplitude of the CMAP due to activation of more
motor units until all muscle fibers have been activated
(suprathreshold stimulation). Amplitude of CMAP
reflects the total number of muscle fibers that contribute to the potential, provided that they fire synchronously. Intraoperatively, the amplitude of the CMAP
is usually measured from the most negative peak to

the most positive peak (peak-to-peak amplitude).
Decrease of CMAP amplitude primarily results from
fewer activated axons, however, it may also result
from partial conduction block (Campbell, 1998).
25.3.1.2. Anesthetic requirements for intraoperative
EMG monitoring
Intraoperative EMG responses are basically unaffected by routine concentrations of common anesthetics such as nitrous oxide, opiates, or halogenated
agents, so there are generally no constraints necessary
on anesthetic techniques, except avoiding the use of
muscle relaxants, since blockade of the neuromuscular
junction interferes with meaningful monitoring of
EMG activity (Blair et al., 1994). It is therefore recommended not to use any paralytic agent during FN monitoring. Short-acting agents such as succinylcholine
may be given at the time of intubation, but it should
be confirmed that such agents have cleared before
monitoring begins. While muscle relaxants can be useful to facilitate ventilation, prevent patient movement,
and allow use of less anesthetic agents, during FN
monitoring, the anesthesiologist must use other agents
to maintain an adequate depth of anesthesia to prevent
patient movement.
25.3.1.3. Use of electrical stimulation for FN
monitoring
Before using electrical stimulation for FN monitoring,
it is important to confirm that the stimulating system is
functioning, keeping in mind that the presence of a
stimulus artifact does not confirm a functioning system. It is possible to have a stimulus artifact with only
one lead (either the anodal return or the cathodal stimulator) connected. However, the absence of any artifact usually indicates an open circuit somewhere in
the system. To avoid any doubt, it is a good strategy
to confirm the operation of the entire system before
monitoring begins; a simple way to achieve this is by
placing the stimulating electrode directly on a muscle
and obtaining a direct muscular response, although
muscle requires higher stimulation levels than nerve.
Alternately, a different motor nerve can be stimulated;
for example, the spinal accessory nerve (XI) can often
be accessed during a retrosigmoid craniotomy and
used to confirm the integrity of the stimulator by
obtaining a CMAP from the trapezius muscle. Once
system function has been confirmed, electrical stimulation can be used to confirm the absence of a nerve
in a specific location. The various uses of electrical
stimulation during FN monitoring which go beyond

just localizing the course of the nerve will be discussed
in the following section.
25.3.1.3.1. Use of stimulation to identify and map
the course of the FN. FN identification is not a major
problem in cases with relatively undistorted anatomy
such as uncomplicated middle ear surgery, microvascular decompression, or vestibular neurectomy where
the anatomy of the FN and its relations among the
various cranial nerves are relatively constant. However, the presence of pathology, such as congenital
anomalies or a space-occupying lesion in the temporal bone or posterior fossa, may make identification
based on anatomical relationships difficult or impossible. If the nerve can be located at the beginning of
surgery, its location can be confirmed by an electrical
response before dissection begins. After establishing
a threshold for stimulation, the voltage is increased
to about three times the threshold (suprathreshold
stimulation) and the stimulator swept across the
exposed surface for dissection to confirm that there
are no FN fibers before beginning dissection. If the
nerve is visually difficult to identify at the beginning
of dissection, the only way to locate and trace the FN
is with electrical stimulation. A practical strategy is
to start mapping the accessible region with an intensity of 0.3 V until a response is obtained and the FN
is located. If no response is obtained, the search is
repeated at 0.5 and 1.0 V. If no response is obtained
at 1.0 V, it can be safely assumed that the FN is not
on the exposed surface and dissection can proceed
(Yingling and Ashram, 2005a).
25.3.1.3.2. Use of stimulation to differentiate
between the FN and other cranial nerves in the cerebellopontine angle. In larger tumors, cranial motor
nerves may be displaced in unexpected locations; it
may not be possible with visual inspection of the
surgical field to confirm which nerve is being
encountered. To gain more insight into anatomical
relationships, electrical stimulation can be used to
distinguish between the FN and other cranial nerves
such as the trigeminal, vestibulocochlear and to a
lesser extent the abducens and lower cranial nerves.
A tumor can push the FN rostrally so that it is
located very close to the trigeminal nerve. Stimulation
of the motor fibers of the trigeminal nerve produces
EMG responses in the masseter and temporalis muscles, and responses in these larger muscles can produce
a volume conducted response on FN monitoring
375
channels. However, these two nerves can be differentiated from one another despite overlap in the responding channels by their different onset latencies.
Stimulation of the trigeminal nerve produces EMG
responses that are of a considerably shorter latency
(34 ms to onset) than those to FN stimulation
(68 ms), allowing these nerves to be distinguished
(Mller, 1995).
Similarly, the vestibulocochlear nerve has an anatomically close relation to the FN in the cerebellopontine angle (CPA) and internal auditory canal,
and confirmation of the identity of the nerve is sometimes essential before manipulation. By electrical
stimulation with an intensity just above threshold, a
response is produced on the FN monitoring channels
only to direct FN stimulation. If the vestibulocochlear nerve is stimulated, no response is produced
in facial muscles below 0.7 V, much above the
threshold for FN stimulation (Ashram et al., 2005).
25.3.1.3.3. The use of stimulation before bipolar
electrocoagulation to confirm that the area to be
cauterized is free from FN fibers. Although bipolar
cautery produces less spread of electrical current
compared to monopolar cautery, neural injury can
still occur owing to spread of heat by convection or
conduction (Kartush and Bouchard, 1992). Therefore, the FN is at a significant risk of thermal injury
during electrocautery. Furthermore, bipolar devices
generate a high amplitude broad band noise that is
difficult to filter, thus rendering EMG monitoring
useless during a time when the FN is potentially at
high risk. In our experience, a practical way to deal
with this problem is to confirm that the nerve is not
in an area about to be cauterized, by using higher
levels of electrical stimulation (up to 1 V). If no
response is obtained at this level, it can be assumed
that electrocautery can proceed safely.
25.3.1.3.4. Assessment of functional status of the
FN and prediction of postoperative outcome by electrical stimulation. Another important utility of electrical stimulation is the determination of the
functional status of the FN at the end of dissection
and prediction of postoperative functional outcome.
Several groups of investigators have studied the prognostic value of intraoperative electrical stimulation in
predicting postoperative FN outcome using different
parameters based on measurement of the CMAP.
These methods include measurement of the threshold
of FN stimulation at the end of dissection proximal to
376
the tumor at the brainstem (Lalwani et al., 1994; Silverstein et al., 1994; Zeitouni et al., 1997), as well as
measurement of the peak-to-peak amplitude of the
CMAP at both the orbicularis oculi and orbicularis oris
muscles (Mandpe et al., 1998). Other methods based
on pre- and postresection comparisons of thresholds
(Prasad et al., 1993), or amplitudes (Ebersold et al.,
1992; Taha et al., 1995; Mandpe et al., 1998) have also
been suggested. Unfortunately, there is still no agreement about which method provides the highest degree
of accuracy and sensitivity. The best intraoperative
predictor of postoperative outcome remains to be
determined. A detailed discussion of the prognostic
significance of each of these methods is beyond the
scope of this chapter.
25.3.1.3.5. The use of stimulation to assess the integrity of the FN after an episode of spontaneous activity
and during dissection of large tumors. Electrical
stimulation can be used to evaluate the integrity of
the FN after the occurrence of an alarming episode
of spontaneous activity during dissection. This evaluation requires a baseline measurement of threshold
and amplitude proximal to the site of dissection. A
change of threshold or amplitude from baseline after
an episode of spontaneous activity is an indication
that manipulation has caused a change in the functional status of the nerve. Although proximal threshold and amplitude measurement at the beginning of
surgery may not always be possible to obtain in large
tumors, it should be attempted as they provide an
idea about the baseline values.
The use of electrical stimulation during dissection
of large CPA tumors is an important adjunct to spontaneous EMG to give information about the condition
of the nerve. In large tumors, axons of the FN can
become significantly stretched and partially damaged, and thus be less likely to respond than healthy
ones, producing little EMG when manipulated
(Mller, 1995). In such cases, the absence of spontaneous EMG activity should not be relied upon as an
indication of lack of mechanical irritation to the FN
during dissection. Rather, electrical stimulation of
the nerve should be used to assess its functional
status by comparing responses with a previously
recorded baseline (Yingling and Ashram, 2005a).
25.3.1.3.6. Assessment of the thickness of bone
over the FN during drilling. Electrical stimulation
can be used to estimate the thickness of the remaining bone under the burr end. This is important during
the last stages of drilling of the fallopian canal or the

internal auditory canal; a decreased threshold
(<0.5 V) implies a closer FN to the burr and drilling
should be done with marked caution to avoid injury
to the underlying FN.
25.3.1.4. Monitoring of spontaneous and
mechanically elicited activity
25.3.1.4.1. Patterns of mechanically evoked EMG
activity. Almost all patients exhibit some mechanically evoked facial EMG activity during tumor dissection, retraction, irrigation, or other intraoperative
events. Prass and Luders (1986) distinguished two
types of such EMG activity. They suggested that
the phasic burst pattern, characterized by short, relatively synchronous bursts of motor unit potentials,
corresponded to a single discharge of multiple FN
axons. This type of activity was associated with
direct mechanical nerve trauma, especially blunt
trauma, irrigation, or electrocautery. The second pattern, tonic or train activity, consisted of episodes of
prolonged asynchronous grouped motor unit discharges which could last up to several minutes. These
were most commonly associated with FN traction.
They further divided such train activity into higher
frequency trains (50100 Hz), called bomber potentials due to their sound characteristics, and lower
frequency discharges (150 Hz), which were more
irregular and had a sound resembling popping popcorn. More recently, Romstock et al. (2000) classified train activity into three distinct patterns A, B,
and C trains. A trains are characterized by a sinusoidal symmetrical sequence of high frequency and low
amplitude signals that have a sudden onset; B trains
are regular or irregular sequences of repeated spikes
or bursts with maximum intervals of 500 ms; and
C trains are characterized by continuous irregular
EMG responses that have many overlapping components. The authors suggested that the occurrence of
A trains poses an increased risk on the FN. Nakao
et al. (2001) classified train activity that occurred
during the last stage of tumor resection into an
irritable pattern with frequent EMG responses to
the slightest stimuli; a silent pattern with few or no
EMG responses; a stray pattern with persistent train
responses up to 20 min despite temporary discontinuance of surgical manipulation; and an ordinary
pattern related to mechanical stimulation of the nerve
but not easily elicited. They found that the occurrence of silent or stray EMG patterns may be associated with an increased risk of injury to the FN.
Fig. 3 shows representative samples of different

types of EMG patterns often encountered during FN
monitoring.
25.3.1.4.2. Recording and interpretation of spontaneous EMG. The occurrence of burst activity during
surgery is due to direct mechanical manipulation,
usually from blunt contact with the nerve. Their
occurrence is a warning that the FN was being
stimulated enough to result in depolarization and production of facial muscle EMG response but not necessarily enough to be injured (Prass et al., 1987).
Accordingly, the occurrence of small amplitude
bursts (<500 mV in amplitude) is not of major
377
concern and does not usually mean an increased risk

of injury to the FN (Magliulo and Zardo, 1997;
Romstock et al., 2000). The surgeon should not be
warned each time such activity occurs, although in
some situations their occurrence may be of utility to
the surgeon. Hearing such activity through the loud
speaker during dissection is of importance to the surgeon as it develops a high index of anticipation for
their occurrence, helping develop safer surgical techniques for dissection (Prass, 1996). They may also be
the first indicator to the proximity of the FN at the
beginning of dissection or during drilling of bone
over the nerve. During initial drilling, there is a layer
of bone overlying the FN but at the final stage of
Fig. 3. Examples of EMG activity often seen during facial nerve (FN) monitoring. A: Dense tonic (sustained) train activity,
having a sinusoidal pattern, usually occurring with nerve stretch. B: Lower frequency sustained activity called popcorn
activity. C: Burst activity which is a phasic (transient) activity that is usually associated with direct contact with the nerve.
Such events are not of major significance unless they are of large amplitude and occur during critical stages of dissection.
D: Burst activity superimposed on ongoing small-amplitude train. Such events overlapping on background activity should
be identified because they may pass unnoticed despite their significance (reprinted from Yingling and Ashram, 2005a with
permission from Lippincott, Williams and Wilkins).
378
drilling, the bone layer is markedly thinned out and

the possibility of burst activity being due to direct
contact with the nerve is increased. Therefore, bursts
occurring at this stage may be an indication of
impending direct contact with the FN and require
more cautious drilling to avoid serious impact with
the FN fibers. This can be confirmed as previously
mentioned by electrical stimulation and by noting a
decrease in threshold of FN stimulation from a previous baseline.
On the other hand, when burst activity of large
amplitude (500 mV or more) occurs during dissection
or final stages of drilling, the surgeon should be
promptly warned. He may decide to alter the technique or move to another area of dissection, as such
activity may indicate a degree of FN injury, the
extent of which varies according to the force and
number of impacts (Yingling and Ashram, 2005a)
Train activity during FN monitoring can occur due
to traction compression, contusion, heat spread by
electrocoagulation, or saline irrigation to the FN.
The neurophysiologist performing monitoring should
know when to warn the surgeon and not to alert the
surgeon each time such activity is displayed, as this
may interrupt rather than help surgery and the surgeon is likely to lose confidence in the monitoring
team. In general, there are several factors to be considered when interpreting train activity. First, it
should be interpreted in relation to a surgical event
whenever possible; for example, if train activity
occurs after saline irrigation there is no reason to
raise concern. However, if it occurs due to traction
or dissection in close proximity to the FN, the surgeon should be warned. It is worth noting here that
sometimes train activity starts with no apparent reason, or may be observed in baseline recordings and
decrease as the nerve is decompressed with opening
of the dura and draining of CSF. Although worrisome
to the surgeon, such activity does not usually pose a
significant risk to the FN.
A second point to consider is amplitude of train
activity. Small amplitude train activity does not usually mean an impending injury to the nerve; its
occurrence, however, does indicate the proximity of
the FN to the region of dissection and caution should
therefore be taken. It could also be considered a good
sign, since its presence implies an intact and responsive nerve (Prass et al., 1987). On the other hand,
when train activity of large amplitude (>500 mV)
occurs during dissection around the nerve, the surgeon should be warned immediately because such
activity may be associated with a degree of nerve

injury (Beck et al., 1991; Nakao et al., 2001). However, the absence of train activity of large amplitude
does not necessarily imply safer dissection, especially in large tumors in which FN fibers become significantly compressed. Furthermore, in some cases,
a change of pattern of EMG from a responsive to a
silent one can occur, in which a nerve that was readily responsive to various mechanical stimuli in early
stages of dissection becomes less responsive towards
the final stages of resection. Spontaneous activity
becomes less frequent and of low amplitude despite
significant surgical manipulation of the FN. This
decline in spontaneous activity may give a false
sense of security, causing more vigorous dissection
with the possibility of permanent damage to the
nerve. Therefore, such a change of pattern should
be cautiously interpreted and considered an ominous
sign indicating a certain degree of nerve injury has
already occurred.
A third point to consider when interpreting train
activity is the pattern of activity; certain patterns of
train activity have been suggested to pose a risk of
injury to the FN, such as the type A trains, stray pattern, and bomber-type high-frequency trains described
above. Fourth, ongoing EMG activity is often an indirect indicator of depth of anesthesia, which is of particular concern when no muscle relaxants are used.
A simultaneous increase in spontaneous EMG activity
on all channels is unlikely to result from localized dissection. When such a generalized increase occurs, the
anesthesiologist should be notified immediately to
increase the depth of anesthesia and avoid overt patient
movement.
In conclusion, the analysis of spontaneous EMG
activity is not based on a single criterion but rather
a combination of factors that together lead to a correct interpretation, including surgical orientation,
awareness of surgical events and anesthetic conditions, and understanding of waveform configuration
and morphology.
25.3.1.5. Limitations and pitfalls when using EMG
Despite the wide use of intraoperative EMG monitoring, it still has its limitations. A major drawback is
that EMG channels can easily pick up artifacts from
numerous operating room sources. It is important to
distinguish artifact from true EMG to avoid false
positive alarms which can be misleading to the surgeon. Such distinction may sometimes be difficult
and requires experienced monitoring personnel.
There are several causes of intraoperative artifact;

some artifacts are easily identified as linked to the
use of electrocautery equipment, ultrasonic aspirators, lasers, drills, etc. Other artifacts are more misleading, such as those due to contact between
surgical instruments made of different metals (bimetallic potentials); these may be associated with similar intraoperative events as those producing true
EMG responses. Some useful hints to distinguish
artifact from EMG is that artifacts tend to appear
simultaneously on several channels, which is unlikely
for EMG. Artifacts are usually higher in frequency
content than EMG, and thus have a crackly sound
while true EMG, has more of a popping sound on
the loud speaker. Another hint is that artifacts tend
379
to be regular in morphology while physiological

waves have a more variable morphology (Fig. 4).
Another problem with EMG is that FN stimulation
cannot be performed unless the nerve is accessible in
the surgical field. However, when removing large
tumors the FN is at risk of being traumatized before
it is visually apparent to the surgeon, and may not
be responsive to mechanical manipulation. In this situation, the FN location can be anticipated by using a
flexible-tip probe to stimulate within dissection
planes or even behind the tumor, out of the surgeons
view, without concern for damaging unseen vascular
or neural structures (Yingling and Gardi, 1992). The
importance of localizing the FN at the start of dissection is emphasized by the fact that a sharp cut to the
Fig. 4. Illustrative examples of artifacts that can be mistaken for EMG. A: Upper: bimetallic artifact produced by
contact of different metallic instruments in the surgical
field (may be confused with spike activity). Note the sharp
edge on waveforms with exponential decay. Lower: single
EMG spike superimposed on a background of lowamplitude EMG activity and no exponential decay.
B: Upper: regular sinusoidal artifact produced during drilling the internal auditory canal (IAC). Lower: irregular
EMG activity occurring while drilling the IAC. C: Upper
two traces: regular artifact represented on two time scales,
200 and 5 ms/cm, respectively. Bottom two traces: EMG
on the same two time scales. At 200 ms/cm, it can be difficult to differentiate between true EMG activity and artifact.
However, with the faster 5 ms/cm time base, trace 2 shows
that the artifact waveform is regular and synchronized
while trace 4 reveals the irregularities that characterize true
EMG activity (reprinted from Yingling and Ashram, 2005a
with permission from Lippincott, Williams and Wilkins).
380
FN will elicit no EMG activity and may pass unrecognized; furthermore, the distal segment of a transected nerve may still give a response to both
mechanical and electrical stimulation, giving a false
sense of security (Fig. 5).
A potentially misleading pitfall when performing
electrical stimulation is a volume conducted response
on a FN monitoring channel after stimulation of
another cranial nerve. For example, stimulation of
the trigeminal nerve may give a response on FN monitoring channels and may be mistaken for the FN;
however, the two nerves can be identified by their
different latencies as mentioned previously. This is
facilitated by the use of multichannel monitoring with
separate channels for recording EMG responses from
different cranial nerves (Fig. 6). Another example, is
stimulation of the nervus intermedius which may be
mistaken for a thinned attenuated FN filament in the
Fig. 5. Diagrammatic representation of site of stimulation

of an injured facial nerve (FN). Immediately after intraoperative injury to the FN, stimulation of the nerve proximal to the site of injury reveals no response (A) (or
reduced response if injury is incomplete), while stimulation
of the nerve distal to the site of injury results in a normal
EMG response (B), therefore it should be confirmed that
the stimulator is placed proximal to the site of injury when
assessing FN function. (a) Labyrinthine segment of the FN;
(b) tympanic segment of the FN proximal to the site of
injury; (c) mastoid segment of the FN distal to the site of
injury; (d) extracranial segment of the FN. Arrow points
to the site of FN injury.
CPA. Stimulation of the nervus intermedius can cause

volume conducted EMG activity in the FN monitoring
channels (at least the orbicularis oris), but the main
trunk of the FN may lie in an entirely different location
within the CPA. The nervus intermedius response
can be recognized by its electrophysiological features
(Ashram et al., 2005). Once the nervus intermedius is
recognized, it is crucial for the surgeon to locate the
main trunk of the FN by further stimulation mapping.
25.3.2. Other modalities used for FN monitoring
The earliest techniques of FN monitoring involved
stimulation of the FN intracranially and watching the
patients face under the drapes for visible contractions
(Krause, 1912; Olivecrona, 1940; Pool, 1966). These
old visual methods were revived with the introduction
of the new technology of video analysis to detect facial
movements (Filipo et al., 2000, 2002). Such techniques assessed facial movements by measuring the
shift along a line from labial commissure to zygomatic
bone since contraction of the zygomatic muscle moves
the labial commissure up and down. Filipo et al. (2002)
also compared video monitoring to EMG and found
that video monitoring, although less sensitive than
EMG, is a reliable method with the advantages of
being noninvasive, easy and quick to set up, and more
specific than EMG as it only responds to muscle contraction while EMG may produce many false positive
results unless the neurophysiologist is skilled at distinguishing artifacts from true EMG activity.
Another technique that has been used for FN monitoring is recording of the CNAP by stimulating the distal FN and recording the CNAP from the nerve within
the surgical field opposite to the normal direction of
impulse conduction (antidromic recording) using a
bipolar electrode (Richmond and Mahla, 1985). This
technique was further updated by Colletti et al.
(1996), using low-intensity stimulation of the mandibular branch of the FN and monopolar recording techniques with an intracranial electrode. Methods based
on CNAP rather than EMG recording have the advantage that they can be used even if muscle relaxants are
used. Another advantage is that these methods allow
actual continuous monitoring of FN function, in contrast to EMG, which provides information only when
the FN is mechanically or electrically stimulated.
On the other hand, the CNAP monitoring is less sensitive than EMG, and frequently requires averaging.
Also, direct nerve monitoring is technically more difficult than EMG; attaching electrodes to delicate neural
381
Fig. 6. Illustrative examples of volume conducted activity on facial nerve (FN) monitoring channels that can be mistaken
for FN responses. A: Upper trace shows response from temporalis muscle after stimulation of the trigeminal nerve; middle
trace is the same response (trigeminal nerve stimulation) recorded simultaneously on the orbicularis oculi channel. Note the
short latency of the two responses (3 ms). Lower trace shows a response to FN stimulation on orbicularis oculi channel from
the same patient (latency 6.2 ms). The trigeminal and FN responses can be differentiated by their latency and the use of
multiple channels. B: Responses on the orbicularis oris to stimulation of the nervus intermedius (upper) and FN (lower).
Nervus intermedius response can be recognized by its longer latency, smaller amplitude, and higher threshold for stimulation. C: A volume conducted response on the orbicularis oculi to stimulation of the abducens nerve (upper) compared to a
FN response recorded from the same muscle (lower). Note the lower amplitude and shorter latency of the response to abducens nerve stimulation; the onset of the response is not clearly demarcated from the stimulus artifact. The vertical lines indicate the point of stimulation.
structures can be difficult and even injurious to the

nerve. In addition, electrodes in the surgical field
may limit exposure (Colletti et al., 1997).
Recording of F-waves from the nasal muscle is
another method that made its way to clinical use in
FN monitoring (Wedekind and Klug, 1996). The
F-wave is a late muscle potential that is believed to
be due to recurrent discharge of a small percentage
(15%) of the motor neurons. It occurs due to antidromic spread of excitation that reaches the motor neurons and then again propagates orthodromically,
producing a delayed muscle contraction. The intraoperative recording of an F-wave should reflect the
integrity of the FN pathway from the axon hillock to
the motor end plate; a stable F-wave or reversible
changes that occur during surgery usually indicate
382
good postoperative FN function (Wedekind and Klug,

2001). Although the F-wave may be a promising tool
for intraoperative FN monitoring, its use has been limited by its sensitivity to anesthesia and the fact that it
may be normally absent in healthy adults.
Transcranial motor evoked potentials (tcMEP) is
a new technique now widely used in spinal surgery.
It is performed by electrical stimulation of the motor
cortex with recording of EMG responses from appropriate muscles. This technique has been successfully
adopted to monitor the FN by multipulse stimulation
of the contralateral motor cortex and recording of
facial MEPs from the orbicularis oris muscle on the
operative side. This method can provide an ongoing
assessment of FN function without the need of the
surgeon to stimulate the nerve, and could be especially useful in cases in which the proximal FN is
inaccessible to direct stimulation (Dong et al., 2005).
Another method under trial is the blink reflex,
which is elicited by stimulation of the supraorbital
nerve (trigeminal afferents) and consists of an early
ipsilateral EMG response (R1) and a later bilateral
response (R2), both mediated by the FN (Kimura,
1989). It is the R1 response that is recorded intraoperatively. A main problem with blink reflex, however, is the suppression of reflex activity by general
anesthetics. It is possible to overcome this problem
by greater reliance on intravenous anesthetics such
as propofol together with the use of repetitive train
stimulation. A chief advantage of the last two techniques is the ability to continuously monitor the FN
without the need for the surgeon to directly stimulate
the nerve. Further improvement in FN monitoring
techniques are likely to come about when such methods of continuous assessment of FN integrity become
more refined and integrated.
References
Ashram, YA, Pitts, L, Jackler, R and Yingling, CD (2005)
Intraoperative electrophysiological identification of the
nervus intermedius. Otol. Neurotol., 26: 274279.
Beck, DL, Atkins, JS Jr., Benecke, JE, Jr. and Brackmann,
DE (1991) Intraoperative facial nerve monitoring: prognostic aspects during acoustic tumor removal. Otolaryngol. Head Neck Surg., 104: 780782.
Blair, EA, Teeple, E Jr., Sutherland, RM, Shih, T and
Chen, D (1994) Effect of neuromuscular blockade on
facial nerve monitoring. Am. J. Otol., 15: 161167.
Campbell, WW (1998) Generalized peripheral neuropathies. In: Essentials of Electrodiagnostic Medicine.
Williams and Wilkins, Baltimore, MD, pp. 225253.

Colletti, V, Fiorino, FG, Policante, Z and Bruni, L (1996)
New perspectives in intraoperative facial nerve monitoring with antidromic potentials. Am. J. Otol., 17(5):
755762.
Colletti, V, Fiorino, F, Policante, Z and Bruni, L (1997)
Intraoperative monitoring of facial nerve antidromic
potentials during acoustic neuroma surgery. Acta. Otolaryngol. (Stockh), 117: 663669.
Danke, JA and Wiegand, DA (1994) Investigation of a
coaxial bipolar nerve stimulator for intraoperative
motor nerve monitoring. Laryngoscope, 104: 619622.
transcranial electrical stimulation during skull base surgery. Clin. Neurophysiol., 116: 588596.
Ebersold, M, Harner, SG and Beatty, CW (1992) Current
results of the retrosigmoid approach to acoustic neurinoma. J. Neurosurg., 76: 901909.
Filipo, R, De Seta, E and Bertoli, GA (2000) Intraoperative
video monitoring of the facial nerve. Am. J. Otol., 21:
119122.
Filipo, R, Pichi, B, Bertoli, GA and De Seta, E (2002)
Video-based system for intraoperative facial nerve monitoring: comparison with electromyography. Otol. Neurotol., 23: 594597.
Harper, CM and Daube, JR (1998) Facial nerve electromyography and other cranial nerve monitoring. J. Clin. Neurophysiol., 15: 206216.
Kartush, JM and Bouchard, KR (1992) Intraoperative facial
nerve monitoring in otology, neurotology, and skull
base surgery. In: Neuromonitoring in Otology and Head
and Neck Surgery. Raven Press, New York, pp. 99120.
Kartush, JM, Niparko, JK, Bledsoe, SC, Graham, MD and
Kemink, JL (1985) Intraoperative facial nerve monitoring: a comparison of stimulating electrodes. Laryngoscope, 95: 15361540.
Kimura, J (1989) The blink reflex. In: J Kimura (Ed.), Electrodiagnosis in Diseases of Nerve and Muscle: Principles
and Practice. F.A Davis Company, Philadelphia, 2nd ed.,
pp. 307331.
Krause, F (1912) Surgery of the Brain and Spinal Cord.
Rebman Company, New York, Vol. II.
Lalwani, AK, Butt, FY, Jackler, RK, Pitts, LH and Yingling,
CD (1994) Facial nerve outcome after acoustic neuroma
surgery: a study from the era of cranial nerve monitoring.
Otolaryngol. Head Neck Surg., 111: 561570.
Magliulo, G and Zardo, F (1997) Intra-operative facial nerve
monitoring. Its predictive value after skull base surgery.
J. Laryngol. Otol., 111: 715718.
Mandpe, AH, Mikulec, A, Jackler, RK, Pitts, LH and
Yingling, CD (1998) Comparison of responses amplitude
versus stimulation threshold in predicting early postoperative facial nerve function after acoustic neuroma resection.
Am. J. Otol., 19: 112117.

May, M, Schaitkin, BM, Weit, RJ and Gupta, P (2000)
Trauma to the facial nerve: external, surgical and iatrogenic. In: M May and BM Schaitkin (Eds.), The Facial
Nerve. Thieme, New York, 2nd ed., pp. 373382.
Mller, AR (1995) Monitoring of motor nerves. In: AR
Mller (Ed.) Intraoperative Neurophysiologic Monitoring. Harwood Academic Publishers, Sydney,
pp. 173216.
Monsell, EM (2005) Iatrogenic facial nerve injury: prevention and management. In: RK Jackler and DE Brackmann
(Eds.), Neurotology. Elsevier Mosby, St. Louis, 2nd ed.,
pp. 12701278.
Nakao, Y, Piccirillo, E, Falcioni, M, Taibah, A, Kobayashi,
T and Sanna, M (2001) Electromyographic evaluation
of facial nerve damage in acoustic neuroma surgery.
Otol. Neurotol., 22: 554557.
Nuwer, MR, Daube, J, Fischer, C, Schramm, J and
Yingling, CD (1993) Neuromonitoring during surgery.
Report of an IFCN committee. J. Electroencephalogr.
Olivecrona, H (1940) Acoustic tumors. J. Neurol. Psychiatry, 3: 141146.
Prasad, S, Hirsch, BE, Kamerer, DB, Durrant, J and
Sekhar, LN (1993) Facial nerve function following cerebellopontine angle surgery: prognostic value of intraoperative thresholds. Am. J. Otol., 14: 330333.
Prass, RL (1996) Iatrogenic facial nerve injury: the role of
facial nerve monitoring. Otolaryngol. Clin. North Am.,
29: 265275.
Prass, RL and Luders, H (1986) Acoustic (loudspeaker)
facial electromyographic monitoring: Part 1. Evoked
electromyographic activity during acoustic neuroma
resection. Neurosurgery, 19: 392400.
Prass, RL, Kinney, SE, Hardy, RW, Jr., Hahn, JF and
Luders, H (1987) Acoustic (loudspeaker) facial EMG
monitoring: II. Use of evoked EMG activity during
acoustic neuroma resection. Otolaryngol. Head Neck
Surg., 97: 541551.
Preston, DC and Shapiro, BE (1998) Fundamentals of
nerve conduction studies: basic nerve conduction studies.
In: DC Preston and BE Shapiro (Eds.), Electromyography
and Neuromuscular Disorders. Clinical Electrophysiologic Correlations. Butterworth-Heinemann, Boston,
MA, pp. 2544.
Pool, JL (1966) Suboccipital surgery for acoustic neurinomas: advantages and disadvantages. J. Neurosurg., 24:
483492.
383
Richmond, IL and Mahla, M (1985) Use of antidromic

recording to monitor facial nerve function intraoperatively. Neurosurgery, 16: 458462.
Robinson, AJ (1995) Physiology of muscle and nerve. In: AJ
Robinson and L Snyder-Mackler (Eds.), Clinical Electrophysiology: Electrotherapy and Electrophysiologic Testing. Williams and Wilkins, Baltimore, MD, pp. 110117.
Romstock, J, Strauss, C and Fahlbusch, R (2000) Continuous electromyography monitoring of motor cranial
nerves during cerebellopontine angle surgery. J. Neurosurg., 93: 586593.
Silverstein, H, Willcox, TO, Jr., Rosenberg, SI and Seidman, MD (1994) Prediction of facial nerve function following acoustic neuroma resection using intraoperative
facial nerve stimulation. Laryngoscope, 104: 539544.
Taha, JM, Tew, JM and Keith, RW (1995) Proximal to distal facial amplitude ratios as predictors of facial nerve
function after acoustic neuroma excision. J. Neurosurg.,
83: 994998.
Wedekind, C and Klug, N (1996) F-wave recordings from
nasal muscle for intraoperative monitoring of facial
nerve function. Zentralbl. Neurochir., 57: 184189.
Wedekind, C and Klug, N (2001) Recording nasal muscle
F waves and electromyographic activity of the facial
muscles: a comparison of two methods used for intraoperative monitoring of facial nerve function. J. Neurosurg., 95: 974978.
Yingling, CD and Ashram, YA (2005a) Intraoperative
monitoring of cranial nerves in skull base surgery. In:
RK Jackler and DE Brackmann (Eds.), Neurotology.
Elsevier Mosby, St. Louis, pp. 958993.
Yingling, CD and Ashram, YA (2005b) Intraoperative monitoring of cranial nerves in Neurotologic Surgery. In: CW
Cummings, PW Flint, LA Harker, BH Haughey, MA
Richardson, KH Robbins, DE Schuller and JR Thomas
(Eds.), Otolaryngology Head and Neck Surgery.
Elsevier Mosby, St. Louis, 4th ed., pp. 38773910.
Yingling, CD and Gardi, JN (1992) Intraoperative monitoring of facial and cochlear nerves during acoustic neuroma
surgery. Otolaryngol. Clin. North Am., 25: 413448.
York, D (1992) Basic principles of neurophysiologic recordings in the operating room. In: J Kartush and K Bouchard
(Eds.), Neuromonitoring in Otology and Head and Neck
Surgery. Raven Press, New York, pp. 119.
Zeitouni, AG, Hammerschlag, PE and Cohen, NL (1997)
Prognostic significance of intraoperative facial nerve
stimulus thresholds. Am. J. Otol., 18: 494497.

M.R. Nuwer (Ed.)
384
CHAPTER 26
Oculomotor and lower cranial nerve monitoring

Jaime R. Lopez*
Department of Neurology and Neurological Sciences, Stanford University School of Medicine, Stanford, CA 94305, USA
Intraoperative neurophysiologic monitoring (IOM)

of oculomotor and lower cranial nerve (CN) function is based on the same rationale and uses similar
stimulation and recording techniques as those
employed for assessing the functional integrity of
other motor nerves. The main difference when
compared to other motor and peripheral nerves is
in the placement of the recording electrodes and
the neural structures that are to be stimulated.
In this chapter, I will review the techniques used
to monitor oculomotor and lower CN function.
The extraocular muscles controlling eye movement
are innervated by the following CNs: oculomotor
(CN III), trochlear (CN IV), and abducens (CN
VI). For this discussion, the lower CNs are defined
as glossopharyngeal (CN IX), vagus (CN X), accessory (CN XI), and hypoglossal (CN XII). For excellent reviews of IOM of CN function not covered in
this chapter, I refer the reader to the corresponding
chapters in this book.
26.1. Rationale
Oculomotor and lower CN activity can be effectively
monitored by using electromyography (EMG) since
their function is either exclusively motor or has a motor
component. Although there are many practitioners
familiar with EMG, its application in the operating
room (OR) differs from that practiced in the neurodiagnostic laboratory. Patients in the OR are under anesthesia and, thus, muscle activity cannot be assessed under
volitional effort. Once the electrodes are placed, they
usually cannot be moved to sample other muscles.
*
Correspondence to: Jaime R. Lopez, M.D., Intraoperative

Neurophysiologic Monitoring Program, Department of
Neurology and Neurological Sciences, and Neurosurgery,
Room A343, Stanford University School of Medicine,
300 Pasteur Drive, Stanford, CA 94305, USA.
Tel.: 1-650-723-1975; fax: 1-650-725-5095.
E-mail: lopezjr@stanford.edu (J.R. Lopez).
In addition, the anesthetic regimen may also affect muscle function, and the amount of electrical noise generating recording artifact is usually much worse than that
seen in the outpatient laboratory setting. The OR also
presents a very dynamic situation, which requires that
the clinical neurophysiologist not only monitor and
report any changes in the neurophysiologic parameters
and how these relate to surgical events but also be aware
of any changes in the anesthetic regimen and the
patients blood pressure and temperature. All these
can influence the IOM results and need to be factored
into the interpretation of the findings.
26.2. Electromyography
The use of EMG as a monitoring technique is based
on the principle that EMG records muscle-fiber
generated electrical potentials. This can be accomplished by using surface, subcutaneous, or intramuscular
electrodes. Surface and subcutaneous electrodes are
generally considered less desirable because distant
motor unit potential (MUP) activity can be missed
and, in some cases, it may be difficult to identify
the specific muscle generating the EMG activity.
Using intramuscular electrodes improves the signalto-noise ratio and allows for near-field recordings
since the recording electrodes are placed as near as
possible to the neural generator (Schlake et al.,
2001b). Conventional subdermal EEG needle and
wire electrodes are the two main types of intramuscular electrodes currently used. The specific type
and length of the needle electrodes used depend on
the location and depth of the target muscle. In many
instances, subdermal EEG electrodes may not be
suitable for recording EMG activity, especially if
the length of the needle is too short to reach the
muscle. We find these electrodes to be adequate for
relatively superficial muscles, such as facial muscles,
trapezius and tongue.
Intramuscular wire electrodes, also known as
hookwire electrodes, have the advantage that they
Fig. 1. EMG paired hookwire electrodes, Medtronic

Xomed SKU 82-26326.
are thinner than needle electrodes and are easier to

be placed into smaller muscles (Fig. 1). The wire
electrodes have 23 mm noninsulated ends, which
are hooked onto the level of conventional hypodermic needles. These, in turn, are used to introduce
the wire into the muscle. The needle is then removed,
leaving the wires in the muscle. For deeply situated
muscles, longer hypodermic needles can be used as
introducers.
The external portion of the wire electrode is taped
to the skin to prevent dislodging. These wires are
usually very stable, even during positioning, and are
usually only dislodged if accidentally pulled out.
Furthermore, they provide excellent and stable
EMG signal recordings. Regardless of what type of
electrode is used, the recording parameters are the
same as those typically used for conventional outpatient needle EMG examination. Gain is set between
50 and 200 mV, low frequency filter between 20
and 30 Hz, and the high frequency filter at 10 kHz.
The sweep speed is usually set at 200 ms per division
for recording continuous EMG, and between 2 and
5 ms per division for recording electrically evoked
EMG potentials. EMG recordings can be made from
multiple muscles simultaneously with the activity
385
observed on an oscilloscope. Equally important is

monitoring the audio feedback, which can aid in distinguishing between EMG activity and electrode and/
or electrical artifact. The audio feedback also allows
the surgeon to hear the EMG activity on a real-time
basis, alerting the surgeon of surgical manipulation
as a possible cause of nerve stimulation. A discussion
of the different types of electrical artifacts and their
likely causes is beyond the scope of this review;
nonetheless, the reader is reminded that these are
commonly encountered problems in the OR and
should be ruled out prior to alerting the surgeon of
the presence of EMG activity.
There are several different types of EMG activity
that can be encountered during IOM. These include
MUPs, fibrillation potentials (which remain present
during anesthesia), and neurotonic discharges (Harper,
1998). Of these, the most important is the neurotonic
discharge (Daube and Harper, 1989). Neurotonic discharges can be distinguished from other types of
MUPs by the frequency and pattern of discharge,
and by the relationship of the activity to mechanical
or metabolic stimulation of the neural structure at risk
(Harper, 1998). Neurotonic discharges can have different discharge frequencies and have the appearance
and sound of high-frequency discharges similar to
end-plate spikes, myotonia and neuromyotonia. These
discharges are generated when the axon membrane is
depolarized, which can occur with mechanical, metabolic, or thermal stimulation, as well as ischemic or
traumatic nerve injury (Mller, 1994; Harper, 1998).
Unfortunately, severing of a CN may not produce
any EMG activity or neurotonic discharges (Mller,
1994) (Table 1).
Table 1
Characteristics of EMG activity observed during
intraoperative recordings
EMG activity
Fibrillation
potentials
MUPs
Neurotonic
discharges
Frequency (Hz)
15
1015
50200
Pattern
Regular and
continuous
Continuous and
semi-regular
Short bursts or
long trains
MUPs, motor unit potentials. Modified from Harper (1998) with

permission from the American Association of Electrodiagnostic
Medicine Annual Meeting 1998.
386
26.3. Anesthesia
Neuromuscular blocking (NMB) agents will affect
EMG activity and it is recommended that the patient
be free of the effects of NMB agents during the time
when EMG monitoring is being performed. Typically,
short-acting NMB agents can be used at the time of
anesthesia induction with their effects wearing-off
by the time the critical stages of surgery take place.
However, successful IOM of EMG activity has been
reported using a constant infusion of a short-acting
NMB agent titrated to a level of 50% reduction of the
baseline compound muscle action potentials (CMAPs)
recorded over the hypothenar muscle (Lennon et al.,
1992; Kizilay et al., 2003). Monitoring of motor CN
EMG activity can usually still be done in this situation
since NMB agents often have more effect on limb
rather than cranial muscles (Harper, 1998).
26.4. Electrical stimulation
The general concept of stimulating motor CNs is no
different than that of stimulating a peripheral nerve.
This can be achieved either with monopolar or bipolar
stimulation. Bipolar stimulation refers to the situation
when both the cathode and anode are directly on the
nerve. In contrast, monopolar stimulation occurs when
the cathode is directly on the nerve and the anode is
placed at a distance away from the nerve. An advantage of bipolar stimulation is that it provides a focal
stimulus, reducing current spread to adjacent neural
structures. However, current shunting can occur
which can lead to submaximal stimulation. Monopolar
stimulation reduces the probability of current shunting
but the chances of activating other neural structures,
via current spread, is increased.
In practice, both methods can yield good results.
However, the most widely used method is, arguably,
monopolar stimulation. This is especially true when
used for the stimulation of CN and brainstem structures, since the areas being stimulated are usually
small and can be located deep within the small surgical
opening. In these situations, it is easier to use one electrode to stimulate rather than trying to place two electrodes directly on the nerve. In addition, electrical
stimulation can be applied using either constant current or constant voltage. Mller (1994) has advocated
the use of constant-voltage stimulation when applying
electrical stimulation intracranially. The rationale is
that there is little variation in electrode impedance
and that shunting of current by cerebrospinal fluid
PEZ
J.R. LO
(CSF), irrigation fluids, and surrounding tissues plays

a much more important role. Therefore, constantvoltage stimulation is more likely to deliver the same
amount of current through the tissue regardless of
shunting. Unfortunately, not all commercially available IOM equipment has the capacity to deliver both
constant-current and constant-voltage stimulation, or
deliver low-level constant-voltage stimulation without
modifying the equipment. Furthermore, experience in
the OR has shown that both stimulation techniques
can be safely and effectively used and both can provide
excellent results.
26.5. Technique for monitoring oculomotor
function
Preserving neural function after a surgical procedure
where nervous tissue is at risk for injury is critically important. Thus, it is imperative to determine
which part of the nervous system may be at risk for
injury prior to surgery. In cases where oculomotor
function may be at risk, such as in certain skull-based,
brainstem, anterior fossa, cavernous sinus, and orbital
surgical procedures, IOM of the oculomotor system
is vital. Injury to one or more of the oculomotor nerves
can cause diplopia. If the diplopia cannot be corrected,
the patient will likely have to wear an eye patch. This
will render the patient monocular, which will cause
loss of stereoscopic vision and reduction of the peripheral visual field. Obviously, this can have a serious
adverse impact on a persons life. Therefore, preserving normal oculomotor function is of critical importance. The easiest and most effective means of doing
this is by monitoring EMG activity of some of the
extraocular muscles.
Although eye movements can be assessed and monitored noninvasively using surface electrodes placed
on the skin around the eyeball (Fukaya et al., 1999;
Sasaki et al., 2002), there appears to be a consensus
that the preferred method to obtain EMG recordings
from extraocular muscles is via intramuscular electrodes. Nonetheless, I will review both invasive and
noninvasive techniques.
Sekiya et al. (1993) reported their results obtained
from 18 patients. They electrically stimulated the
intracranial oculomotor nerves and were able to record
CMAP using a ring electrode that they had developed.
The custom ring electrode was placed by an ophthalmologist and was sutured into the muscle. Recording
electrodes were placed in the medial rectus (MR),
superior rectus (SR), superior oblique (SO), and lateral
rectus (LR) muscles. In two patients, they placed small

needle electrodes in the upper eyelid to record EMG
responses from the levator palpebrae superioris
(LPS), which is innervated by the superior ramus of
the oculomotor nerve.
Stimulation technique: This used either microsurgical bipolar forceps or a custom-made microbipolar
stimulator. Rectangular impulses of 100200 ms duration with a repetition rate of 12 Hz and a maximum
stimulus intensity of 7.0 mA were used for CN
stimulation.
Recording technique: Using the custom ring electrode, they were able to record from all muscles but
had difficulty distinguishing the EMG activity
between the SO and SR muscles.
They concluded that monitoring of MR EMG
activity provided a more reliable indicator of oculomotor nerve function than that from the SR.
In a separate report, Sekiya et al. (2000) described
their results from 12 patients. They employed and
compared two different methods for recording EMG
from extraocular muscles. Method #1 used a spring
electrode, which replaced the ring suture electrode
from their previous study (Sekiya et al., 1993). The
spring electrode was placed in the SR or MR muscle,
and a reference electrode was placed on the forehead.
In method #2, a needle electrode was inserted into
the space just beneath the roof of the orbit and above
the eyeball.
Stimulation technique: This used rectangular
impulses of 100 ms duration, at a rate of 23 Hz,
and a stimulus intensity of 1.0 mA.
They found that they could record equally robust
CMAPs using either recording method. They postulated that the recordings from method #2 likely represented the summed potentials from the SR and LPS
muscles. They also concluded that in certain situations, such as when an ophthalmologist is unavailable,
needle placement into the superior intraorbital space
might be a useful alternative to using the spring electrode. However, one could take this further and perhaps use this indirect recording technique from the
superior intraorbital space to monitor extraocular muscle activity instead of placing electrodes directly into
the SR or MR muscles.
Schlake et al. (2001b) reported their findings from
a study of 18 patients undergoing surgical resection
of various skull base tumors. They used a custommade single-shafted bipolar needle electrode to
record intramuscular EMG from selected extraocular
muscles. A novel aspect of their technique was that
387
they used ultrasound (US) to visualize the needle

tip within the muscle in the orbital space. They were
able to obtain EMG recordings from the MR, SR, and
LR muscles. They felt that there were a number of
advantages to their technique versus those using anatomical landmarks for needle placement. Their technique allowed for exact determination of needle
placement as well as a more controlled and safer
insertion of the electrode. They also felt that the use
of bipolar recordings improved noise rejection characteristics and spatial selectivity of the receptive
fields over conventional monopolar recordings.
Using this technique, they were able to show that
the CMAP latencies were similar to those reported
in other studies but that the amplitudes were on average 3- to 20-fold higher than those values calculated
from previous studies. These differences in amplitude, they postulated, were a result of achieving true
near-field responses.
Stimulation technique: Bipolar stimulation using
rectangular stimuli with a frequency of 30 Hz, an intensity of 0.5 mA, and a duration of 100 ms were applied.
Recording technique: After insertion of an eyelid
retractor, a conventional orbital US probe employing
an orbital US technique was used to place the needle
electrodes. The US probe emitted a focused sound
beam at a frequency of 10 mHz with its focal zone
coinciding with the posterior wall of the eye and
the anterior orbit. Once the electrodes were in place,
the eyelid retractor was removed and the electrodes
were fixed in position using adhesive tape (Fig. 2).
Superior rectus muscle
Lateral rectus muscle
Optic Nerve
10 mm
Inferior rectus muscle
Inferior oblique muscle
Fig. 2. Example of the anatomical position of the singleshafted bipolar needle into the lateral rectus (LR) muscle,
described by Schlake et al. (2001b). [Reprinted from Schlake
et al. (2001) with permission from Springer, Vienna, NewYork.]
PEZ
J.R. LO
388
Alberti et al. (2001) used image guided technology for anatomically correct intraorbital eye muscle
electrode placement in ten patients. They used a conventional neuronavigation system with the normal
placement of skin fiducials prior to brain imaging.
At least one skin marker was placed close to the
orbit, either at the frontal process or the zygomatic
bone or the supraorbital margin of the frontal bone.
Monopolar needle electrodes were inserted into the
LR, inferior rectus (IR), and SO muscles, using neuronavigation. The electrode was deemed in position
when the tip of the electrode was in the belly of
the muscle and the main part of the shaft was in the
orbital fat. Reference electrodes were placed on
the contralateral forehead and the ground was placed
on the nasion.
EMG recordings: Analysis time of 200 ms, amplification of 0.052.0 mV with a bandwidth of 10 Hz
10 kHz.
Stimulation techniques: Bipolar, constant-current
rectangular impulses with an intensity of between
0.05 and 5 mA, a duration of 0.2 ms, and a rate of 3
or 30 Hz. Stimulation was begun at an intensity of
0.05 mA for nuclear regions and 0.5 mA for peripheral
sites.
Noninvasive methods using surface recordings
have also been used to monitor oculomotor nerve
function. Instead of using EMG, this uses electrooculography (EOG). This is based on the electrical
potential difference between the cornea and the retina. Fukaya et al. (1999) studied eight patients with
skull-based tumors. Using this technique, they were
able to identify eye movements consistent with oculomotor and abducens nerve activity (Fig. 3).
Recording technique: Channel 1, active electrode
is placed on the right side of the eye and the reference on the left side. Channel 2, active electrode is
placed on the upper side of the eyeball and the reference on the lower side. Movement of the eyeball
toward the right induces a positive wave, while
movement toward the left generates a negative wave.
Stimulation technique: Monopolar stimulation
using rectangular impulses of 0.2 ms duration at a
rate of 3 Hz and a maximum intensity of 3 mA were
used.
In a separate study, Sasaki et al. (2002) questioned
the origins of the surface recorded evoked potentials
(EPs) generated by intracranial stimulation of the
oculomotor nerve. Using a canine model with the retina removed, they found that the surface recorded
EPs were almost identical to the simultaneously
OCULOMOTOR N. STlM.
A -B+
C -D +
ABDUCENS N. STlM.
D B
A -B+
C -D +
50 V
0
10
ms
Fig. 3. Example of electrooculographic responses after

intracranial stimulation of the left oculomotor and abducens
nerves. [Reprinted from Fukaya et al. (1999) with permission
from the American Association of Neurological Surgeons.]
recorded intramuscular bipolar EMG recordings.

They concluded that the surface recorded EPs elicited by electrical stimulation of the oculomotor
nerves originate from the extraocular muscles and
are not derived from the eye.
At our institution, we use hookwire electrodes to
record EMG activity from the different extraocular
muscles. A pair of wire electrodes is placed percutaneously into the IR, the SO, and the LR muscles for
monitoring of CNs III, IV, and VI, respectively.
The wire electrodes are placed via a hypodermic needle and great care is taken not to injure the globe. A
transeyelid technique is typically used, similar to that
described by Mller (1994) and Rivero et al. (1995)
(Fig. 4). To monitor CN III activity, the needle is
placed in the infraorbital space at approximately the
midpoint. The needle is aimed slightly away from
the eye and the globe is gently pushed up as the needle is introduced. The needle is then removed and the
wires are secured in place with strong adhesive tape.
A similar technique is used for placing the wires into
the SO and LR muscles. The needle is introduced
slightly medial to midline and directed medially into
the superior intraorbital space. For the LR, the needle
is placed lateral to the eye, again aiming it slightly
away from the globe and gently pushing the
eye medially. We have found this technique to generate good EMG potentials from the IR/inferior oblique
(IO) and LR muscles. However, obtaining adequate
EMG potentials from the SO muscle is much
more challenging. In our set-up, we also record from
the orbicularis oculi and use that muscle as a control
in order to differentiate activity between extraocular
389
Fig. 4. A: Schematic view from above of extraocular muscles: (1) superior oblique (SO); (2) medial rectus (MR); (3) superior rectus (SR)levator palpebralis complex; (4) lateral rectus (LR). B: Schematic lateral view of extraocular muscles:
(1) SO; (2) levator palpebralis; (3) SR; (4) LR; (5) inferior rectus (IR); (6) inferior oblique (IO). [Reprinted from Rivero
et al. (1995) with permission from John Wiley & Sons Inc.]
and facial muscle activity (Fig. 5 and 6). Little is

discussed in the literature regarding adverse effects
of needle placement into the extraocular muscles
other than subconjunctival hemorrhages (Rivero
et al., 1995). Inserting intramuscular electrodes into
extraocular muscles is no trivial matter, and it is
recommended that they be placed by an experienced
physician. In over 10 years of using the abovementioned recording technique, we have had only
one patient suffer a significant adverse event. This
consisted of a retrobulbar hemorrhage in a child,
which occurred shortly after placing the wire
electrodes into the LR. Fortunately, no intervention

was necessary, other than delaying surgery, and the
patient recovered completely without any visual
disturbance or lasting effects.
Stimulation techniques: Monopolar stimulation
with a duration of 0.05 ms, a rate between 3 and
5 Hz, and a stimulation intensity usually not exceeding
5 mA. Stimulation intensity begins at 0.1 mA and is
increased by 0.1 mA increments as necessary. We
have also used constant-voltage stimulation, using
similar settings and those described for constantcurrent stimulation. Stimulation is gradually increased
Fig. 5. A: and B: show the percutaneous placement of hookwire electrodes into the superior oblique (SO) muscle using a
25-gauge needle attached to a tuberculin syringe.
PEZ
J.R. LO
390
Fig. 6. The percutaneous placement of hookwire electrodes

into the region of the inferior rectus (IR)/inferior oblique
(IO) muscles.
by 0.1 V increments to a maximum level of 45 V.

Fig. 7 shows an example of neurotonic discharges
and a CMAP recorded from the IR/IO complex.
26.6. Monitoring of lower CN function
The lower CNs may be at risk for injury in certain
types of surgical procedures, especially those involving large tumors of the skull base. Surgery in the region
of the cerebellopontine angle, petroclival area, rhomboid fossa, brainstem, jugular foramen, and foramen
magnum may place CNs IXXII at risk for injury
(Schlake et al., 2001a). It is not only important to monitor the functional integrity of these nerves, but also to
be able to identify them within the surgical field. This
is particularly important since injury to one or more of
these CNs can lead to dysphagia, dysarthria, dysphonia, dysgeusia, and a possible increase in the risk of
aspiration (Schlake et al., 2001a). In addition, in certain situations, especially during resection of large
tumors, it may be difficult to clearly visually identify
CNs within the surgical fields. In these circumstances,
electrical stimulation of unidentified structures in the
surgical fields may help differentiate neural from nonneural structures. Lower CN function can be assessed
using EMG in a fashion similar to that described for
other cranial and peripheral motor nerves.
26.7. Glossopharyngeal nerve (CN IX)
The glossopharyngeal nerve contains motor, sensory,
and autonomic fibers. Although only a small component of it is motor, which supplies the stylopharyngeus
muscle, EMG monitoring can still be accomplished by

placing recording electrodes in the lateral aspect of
the soft palate (Mller, 1994; Schlake et al., 2001a).
However, it is recommended that monitoring of CN
IX function be done in conjunction with vagal nerve
monitoring, since this is the only way to clearly differentiate glossopharyngeal from vagally innervated
laryngeal muscle activity (DeMonte et al., 1994).
Stimulation technique: Schlake et al. (2001a) used
bipolar constant-current stimulation. They applied
square-wave pulses at a frequency of 30 Hz with a
duration of 100 ms and an intensity of either 0.5 or
0.05 mA for intracranial nerve or brainstem stimulation, respectively.
However, DeMonte et al. (1994) recommend
using a single stimulus instead of a train of stimuli.
They also recommend that the stimulus should be
initiated at the lowest intensity available, 0.05 mA,
or 0.20.5 V if using constant-voltage stimulation.
At our institution, we have obtained good results
using a single stimulus technique similar to that
described by DeMonte et al. (1994). We have also
found that using a 35-Hz stimulation mode yielded
good results without any noticeable differences in
adverse effects. Nonetheless, it is important to remain
cognizant of the possible side effects that may
result from electrical stimulation of CN IX. These
include bradycardia and hypotension, because of the
inhibitory effects exerted on the carotid sinus and
carotid body.
26.8. Vagus nerve (CN X)
The vagus nerve is also a mixed, motor, sensory, and
autonomic nerve. However, the motor component,
which innervates the vocal cords and laryngeal muscles, can be monitored using EMG. Several different
techniques have been described to monitor EMG
activity from the vocal cords. These include inserting
needle electrodes into the vocal cords by direct laryngoscopy, using surface electrodes embedded in the
wall or attached on the surface of an endotracheal
tube which are placed up against the vocal cords,
and inserting needle electrodes intraoperatively into
the vocalis muscles or thyroarytenoid through the cricothyroid ligament (Snyder and Hendricks, 2005;
Petro et al., 2006). Unfortunately, needle insertion into
the vocal cords carries the risk of vocal cord hemorrhage or edema and placement into the false cords
may also cause edema (DeMonte et al., 1994). For
these reasons and also the fact that it is difficult to
391
Fig. 7. A: EMG activity at a frequency of 1517 Hz recorded from the left inferior rectus (IR)/inferior oblique (IO) (trace 1)
muscles during resection of a recurrent clival chordoma in a 59-year-old female. B: Brainstem stimulation showing a
compound muscle action potential (CMAP) recorded from the left IR/IO (trace 1) muscle complex. Monopolar stimulation was performed at an intensity of 0.6 mA, pulse duration of 0.1 ms, and a frequency of 1.1 Hz.
place these electrodes by nonotolaryngologists, the

noninvasive techniques are generally preferred. However, one must take special care to insure that the
exposed wires embedded in the endotracheal tube are
placed up against the vocal cords and that they make

good contact. The best method to accomplish this is
to have the anesthesiologist directly visualize that the
exposed wires on the endotracheal tube are placed
PEZ
J.R. LO
392
against the vocal cords. In addition, good contact can

be demonstrated by showing low impedance for each
wire.
Other methods to detect vocal cord movement
using a monometric balloon between the cords and
video endoscopy to observe vocal cord movement
have also been reported (DeMonte et al., 1994;
Tamano et al., 2004).
At our institution, we have used both direct needle
insertion into the vocal cords and wire-embedded
endotracheal tubes to record EMG activity. We have
found the endotracheal tube method of recording surface EMG to yield adequate results, and now use this
technique almost exclusively (Fig. 8).
Stimulation techniques: Mller (1994) recommends electrical stimulation at a rate not to exceed
5 Hz. DeMonte et al. (1994), however, recommends
using a single stimulus and using the same technique
described for glossopharyngeal stimulation.
Romstock et al. (2000) used a concentric, bipolar,
constant-current stimulator. Stimulation intensity was
between 0.05 and 0.5 mA, with a pulse duration of
200 ms, and a stimulus frequency of 30 Hz.
Schlake et al. (2001a) used the same technique as
described for stimulation of the glossopharyngeal
nerve.
Thomusch et al. (2002) reported their results
obtained from a prospective, multicenter trial of surgery for benign goiter where they investigated if the
use of neuromonitoring reduced the rate of recurrent
laryngeal nerve (RLN) palsy. Electrical monopolar
stimulation was performed using a pulse duration of
200 ms, a frequency of 3 or 30 Hz, and an intensity
which varied from 0.05 to 5 mA.
Their results showed that IOM of RLN function
significantly reduced the rates of transient and permanent RLN palsy when compared to intraoperative
visual identification without IOM.
Tamano et al. (2004) stimulated the vagal trigone
during vascular malformation surgery in the medulla.
A single stimulus of 100 ms duration with a frequency of 1.0 Hz and an intensity of 0.83.0 mA
was used to elicit vocal cord movement.
Snyder and Hendricks (2005) used a Medtronic
Xomed Prass monopolar nerve stimulator wand set
at 1.0 mA. No other stimulation parameters were
reported.
Possible adverse effects from electrical stimulation of the vagus nerve are similar to those seen after
stimulation of the glossopharyngeal nerve and, thus,
care should be exercised when stimulating this nerve.
Fig. 8. A: Medtronic NIM EMG endotracheal tube.

B: View of the Medtronic NIM EMG endotracheal tube
after intubation and properly secured at midline.
Fig. 9 shows an example of a CMAP recorded from

the right vocal cord after stimulation of the right
vagus nerve.
26.9. Accessory nerve (CN XI)
EMG monitoring of the accessory nerve is relatively
simple since the spinal portion of the nerve supplies
the sternocleidomastoid and trapezius muscles. Both
of these muscles are relatively superficial and are
easily accessible. However, since the upper part of
the trapezius has mixed innervation from the C1
and C2 spinal roots, the recording electrodes should
be placed in the lower portion of the muscle (Schlake
et al., 2001a) (Fig. 10).
Fig. 9. Stimulation of the right vagus nerve in a 67-yearold patient undergoing microvascular decompression for
glossopharyngeal neuralgia. Note the well-defined surface
recorded compound muscle action potential (CMAP) from
the right vocal cord (trace 4) using a wire embedded endotracheal tube. Stimulation intensity was 0.4 mA with a
duration of 0.1 ms and a frequency of 3.1 Hz. Recordings
were done at a sweep speed of 3 ms per division and a gain
of 100 mV per division.
Stimulation technique: Similar to that described for

CNs IX and X. However, it should be initiated at low
levels since higher stimulus intensities are likely to
produce strong shoulder movement and head turning,
which can lead to excessive patient movement
(DeMonte et al., 1994). In addition, muscle and tendon
injuries may occur since electrical stimulation of the
nerve bypasses the feedback from muscle spindles
and tendon organs that normally limit the force of muscle contraction (Mller, 1994; Schlake et al., 2001a).
An example of neurotonic discharges recorded from
the right trapezius muscle is shown in Fig. 11.
393
Fig. 10. Example of a pair of needle electrodes placed into

the lower trapezius muscle.
nerve can cause dysarthria, and difficulty eating.

Needle or hookwire electrodes can be inserted into
the ipsilateral tongue or genioglossus muscles
(Romstock et al., 2000; Schlake et al., 2001a). However, some authors advocate against submental percutaneous placement into the genioglossus because
of possible injury to vascular and neural structures
in that area (DeMonte et al., 1994). We have experience recording from the tongue as well as the
genioglossus and have noticed no differences in adverse
effects. We have consistently obtained excellent EMG
recordings when recording from the genioglossus
(Fig. 12).
Stimulation technique: This is similar to that previously described for the other lower CNs.
26.11. Summary
26.10. Hypoglossal nerve (CN XII)

Muscles innervated by the hypoglossal nerve are also
easily monitored using EMG. It is important to preserve hypoglossal function because injury to this
IOM of oculomotor (oculomotor, trochlear, and abducens nerves) and lower CN (glossopharyngeal, accessory, vagus, and hypoglossal nerves) motor function
can be performed by monitoring EMG. Continuous
PEZ
J.R. LO
394
Fig. 11. Large amplitude neurotonic discharges recorded from the right trapezius muscle (CN XI, trace 5) during resection
of a right glomus jugulare tumor. In addition, note the small amplitude EMG activity recorded from the right soft palate
(CN IX, trace 3) without any concomitant EMG activity seen in the right vocal cord (CN X, trace 4).
monitoring of EMG activity and neurotonic discharges

should always be performed, and electrical stimulation
done when it is necessary to localize a nerve or assess
its functional integrity. The presence of an unchanged
CMAP amplitude at the end of a surgical procedure
provides supporting evidence that the axons within
that nerve segment, from the site of stimulation to the
muscle, are intact. Finally, with careful planning, an
understanding of the anatomy, and clear knowledge
of the technical aspects of electrical stimulation of
Fig. 12. Pair of hookwire electrodes being placed into the

right genioglossus with the aid of a hypodermic needle.
CNs and its possible adverse effects, IOM can be a safe

and important adjunct to surgery.
References
Alberti, O, Sure, U, Riegel, T and Bertalanffy, H (2001)
Image-guided placement of eye muscle electrodes for
intraoperative cranial nerve monitoring. Neurosurgery,
49: 660664.
Daube, JR and Harper, CM (1989) Surgical monitoring of cranial and peripheral nerves. In: JE Desmedt (Ed.), Neuromonitoring in Surgery. Elsevier, Amsterdam, pp. 115138.
DeMonte, F, Warf, P and Al-Mefty, O (1994) Intraoperative mapping of the lower cranial nerves during surgery of the jugular foramen and lower clivus. In: CM
Loftus and VC Traynelis (Eds.), Intraoperative Monitoring Techniques in Neurosurgery. McGraw-Hill,
Fukaya, C, Katayama, Y, Kasai, M, Kurihara, J and Yamamoto, T (1999) Intraoperative electrooculographic monitoring of oculomotor nerve function during skull base
Harper, CM (1998) Cranial nerve monitoring. In: 1998
American Association of Electrodiagnostic Medicine
Annual Meeting, Course F: Update in Intraoperative
Monitoring Syllabus, pp. 2331.
Kizilay, A, Aladag, I, Cokkeser, Y, Miman, MC, Ozturan,
O and Gulhas, N (2003) Effects of partial neuromuscular blockade on facial nerve monitorization in otologic
surgery. Acta Otolaryngol., 123: 321324.

Lennon, RL, Hosking, MP, Daube, JR and Welna, JO (1992)
Effect of partial neuromuscular blockade on intraoperative electromyography in patients undergoing resection
of acoustic neuromas. Anesth. Analg., 75: 729733.
Mller, AR (1994) Monitoring techniques in cavernous
sinus surgery. In: CM Loftus and VC Traynelis (Eds.),
Intraoperative Monitoring Techniques in Neurosurgery.
McGraw-Hill, New York, pp. 141155.
Petro, ML, Schweinfurth, JM and Petro, AB (2006) Transcricothyroid, intraoperative monitoring of the vagus
nerve. Arch. Otolaryngol. Head Neck Surg., 132:
624628.
Rivero, A, Crovetto, L, Lopez, L, Maselli, R and Nogues,
M (1995) Single fiber electromyography of extraocular
muscles: a sensitive method for the diagnosis of ocular
myasthenia gravis. Muscle Nerve, 18: 943947.
Romstock, J, Strauss, C and Fahlbusch, R (2000) Continuous electromyography monitoring of motor cranial
nerves during cerebellopontine angle surgery. J. Neurosurg., 93: 586593.
Sasaki, T, Suzuki, K, Matsumoto, M, Sato, T, Kodama, N
and Yago, K (2002) Origins of surface potentials evoked
by electrical stimulation of oculomotor nerves: are they
related to electrooculographic or electromyographic
events? J. Neurosurg., 97: 941944.
Schlake, H-P, Goldbrunner, RH, Milewski, C, Krauss, J,
Trautner, H, Behr, R, et al. (2001a) Intra-operative
395
electromyographic monitoring of the lower cranial motor

nerves (LCN IX-XII) in skull base surgery. Clin. Neurol.
Schlake, H-P, Goldbrunner, R, Siebert, M, Behr, R and
Roosen, K (2001b) Intra-operative electromyographic
monitoring of extra-ocular motor nerves (Nn III, VI)
in skull base surgery. Acta Neurochir. (Wien), 143:
251261.
Sekiya, T, Hatayama, T, Iwabuchi, T and Shuji, M (1993)
Intraoperative recordings of evoked extraocular muscle
activities to monitor ocular motor nerve function. Neurosurgery, 32: 227235.
Sekiya, T, Hatayama, T, Shimamura, N and Suzuki, S
(2000) Intraoperative electrophysiological monitoring
of oculomotor nuclei and their intramedullary tracts
during midbrain tumor surgery. Neurosurgery, 47:
11701177.
Snyder, SK and Hendricks, JC (2005) Intraoperative neurophysiology testing of the recurrent laryngeal nerve:
plaudits and pitfalls. Surgery, 138: 11831192.
Tamano, Y, Ujiie, H, Kawamata, T and Hori, T (2004)
Continuous laryngoscopic vocal cord monitoring for
vascular malformation surgery in the medulla oblongata: technical note. Neurosurgery, 54: 232235.
Thomusch, O, Sekulla, C, Walls, G, Machens, A and
Dralle, H (2002) Intraoperative neuromonitoring of
surgery for benign goiter. Am. J. Surg., 183: 673678.

M.R. Nuwer (Ed.)
396
CHAPTER 27
Intraoperative monitoring with free-running EMG

Jeffrey A. Strommena,* and Brian A. Crumb
a
b
27.1. Introduction
Intraoperative monitoring techniques are utilized
with the goal of preserving function and preventing
injury to vital neural structures at a time when clinical examination is not possible. The integrity of these
central and peripheral neural structures can be effectively monitored with multimodal techniques during
a wide variety of surgeries. Not only can these identify and locate nerves, warn of potential nerve injury,
and quantify the degree of nerve injury, they can also
serve to investigate the mechanism of injury as well
as potentially shorten the procedure and improve
future surgical techniques (Strommen and Daube,
2000). The ideal monitoring system should include
a mechanism to provide the surgeon continuous and
rapid feedback of reliable, easily interpreted data
while not interfering with the surgical procedure
and minimally impacting the anesthesia protocol. If
this feedback is provided in a rapid and reliable fashion, the surgeon or anesthesiologist can then take
appropriate action to prevent or reverse the potential
neurologic injury.
In this chapter, we will discuss the technique of
continuous free-running electromyography (EMG)
including the potential effects of physiologic variables, anesthesia, and technical pitfalls. The application of this technique to specific surgeries will be
discussed in greater detail in other chapters.
Needle EMG is well established for routine clinical
practice where it is used to record evidence of denervation in the way of fibrillation potentials and reinnervation by assessing morphologic changes in voluntary
*
Correspondence to: Jeffrey A. Strommen, M.D., Department of Physical Medicine and Rehabilitation, Mayo Clinic
College of Medicine, 200 First Street SW, Rochester,
MN 55905, USA.
Tel.: 1-507-255-4058; fax: 1-507-255-4641.
E-mail: strommen.jeffrey@mayo.edu (J.A. Strommen).
motor unit potentials. In that setting, one can establish

the presence of a process affecting fibers from the
motor unit in addition to providing information on
localization, severity, temporal profile, and prognosis
for recovery. The uses of similar procedures, with
minor modification, provide relatively noninvasive,
real-time information regarding the integrity of motor
axons during a variety of surgical procedures.
The primary purpose of free-running EMG in the
intraoperative setting is to either identify the nerve
at a time when the anatomy is disrupted, such as during tumor procedures or the spine, limbs, or posterior
fossa; or to protect peripheral nerves at times when
there is potential for peripheral axon traction or blunt
trauma, such as with hip or spine surgery. Any procedure that requires manipulation or dissection of a
nerve can be performed more safely and effectively
with intraoperative EMG monitoring.
27.2. Potential recognition
The potentials of interest are neurotonic discharges
and motor unit potentials. Neurotonic discharges are
spontaneous potentials which occur in response to
mechanical, thermal, or metabolic irritation of the
nerve that innervates a muscle. Motor unit potentials
may reflect either reflex activity of anterior horn cells
or incomplete relaxation with voluntary firing. The
morphology of individual motor unit potentials is generally not assessed with intraoperative monitoring as
they are used in routine EMG.
Neurotonic discharges are distinctive discharges of
a motor unit that appear as rapid, irregular bursts lasting several milliseconds or prolonged trains lasting
up to one minute. Each discharge may contain 110
individual motor unit potentials which discharge at
frequencies of 30200 Hz (Harper and Daube, 1998).
They are distinguished from motor unit potentials by
this burst pattern as well as by the relationship to thermal, mechanical, or metabolic irritation of the nerve
membrane. By contrast, motor unit potentials will

have a semi-rhythmic pattern, which may arise with
either irritation to the motor or sensory axon or voluntary activity due to incomplete muscle relaxation.
Neurotonic discharges may be noted with stretching,
compressing, rubbing, or manipulating a nerve,
providing a very sensitive means to inform the surgeon
of potential impending damage. Fig. 1 shows a variety
of neurotonic discharges ranging from fairly simple
discharges with few components to very long, complex discharges. Because of the morphologic heterogenicity, the neurophysiologist and surgeon learn to
recognize these, both visually and aurally, through
experience. Neurotonic discharges must be distinguished from a variety of other muscle activity as well
as artifactual waveforms. Muscle activity may include
motor unit potentials in patients not deeply anesthetized, fibrillation potentials in muscles that have been
partially denervated, myokymic discharges, muscle
end-plate noise and spikes, and complex repetitive discharges. Each of these types of activity is seen intraoperatively, but can be distinguished readily from
neurotonic discharges by the firing patterns and action
potential characteristic that are typically seen with
Neurotonic Discharges
Fig. 1. Neurotonic discharges recorded with free-running

EMG. [From Strommen and Daube (2000) by permission
of Mayo Foundation for Medical Education and Research.
All rights reserved.]
397
routine EMG. Artifactual activity is primarily related

to the electrically volatile environment of the operative
room. Sixty-cycle interference is very common requiring adequate grounding as well as proper equipment
shielding. Movement artifact can easily be misinterpreted as neurotonic discharges but can generally be
recognized as triangular waves that are very irregular
and have a characteristic popping sound. Interference from anesthesia equipment such as respirators,
warmers, and gas humidifiers as well as surgical
equipment such as cautery and cavitrons can generally
be easily recognized (Fig. 2).
27.3. Technique
Free-running EMG activity can be recorded with a
variety of electrodes. Surface electrodes are appropriately utilized in the preoperative diagnostic setting
and in some intraoperative circumstances where quantitation is the desired goal. In the intraoperative setting,
the primary usefulness of free-running EMG is in
accurate nerve localization and protection during a
variety of surgeries. In these circumstances, intramuscular electrodes are clearly advantageous in that they
can more accurately identify the specific muscle as
well as record deep electrical activity. With few exceptions, surface electrodes are generally inadequate for
this purpose. The exception may be the surface electrodes that can record laryngeal activity as is available
in either an embedded endotracheal tube or a postcricoid surface electrode (Figs. 3 and 4). In recent studies,
Pearlman et al. (2005) and Marcus et al. (2003) have
shown that these surface electrodes provide a reliable
means of monitoring the recurrent laryngeal nerve during thyroid and parathyroid procedures. Reported limitations of intramuscular needle electrode placement in
this setting include inexact needle placement with
false-negative results, needle fragmentation, and electrode displacement with potential vascular injury, laceration, vocal cord hematomas, and inadvertent
deflation of the endotracheal tube (Pearlman et al.,
2005). An alternative technique is intramuscular fine
wire electrodes placed directly in the vocal cords under
direct laryngoscopy. This technique certainly assures
proper placement, but could lead to the previously discussed complication and can become easily dislodged
during the procedure.
Most commonly, a variety of intramuscular electrodes are used. These may include monopolar
398
J.A. STROMMEN AND B.A. CRUM

ARTIFACTS DURING INTRAOPERATIVE
ELECTROMYOGRAPHIC MONITORING
Cautery
Cautery
and
electrode
Electrode
Wire
movement
Peripheral
nerve
stimulator
Cavitron
Fig. 2. Artifacts recorded with electromyography fine-wire electrodes during surgical monitoring. [Reprinted from Daube
and Harper (1989) with permission from Elsevier.]
Fig. 3. Postcricoid surface electrode (RLN Systems, Inc.,

Jefferson City, MO).
Fig. 4. Medtronic Xomed EMG endotracheal tube (Medtronic Inc., Minneapolis, MN).
bare-tipped electrodes of various sizes, subdermal

EEG electrodes, or fine wire hook electrodes
(Fig. 5). Until recently, in our practice, fine wire
hook electrodes have been the standard. These are
Teflon-coated nichrome wires (0.01 mm) which are
bared 23 mm at the end and have a hooked tip. A
26-gauge, hollow needle is placed percutaneously.
When the needle is in the appropriate position, it is
withdrawn with the hooked tip allowing the wire to
remain securely within the muscles. The advantage
of this recording electrode is the ease in identifying relatively deep muscles and avoiding any surgical interference since they can be taped directly onto the skin.
The disadvantages of fine wire needles include the
relative ease of dislodgement, somewhat variable
impedance, and the need to clearly be within the target
muscles before the needle can be withdrawn given that
the fine wire cannot be repositioned. Perfect placement
and stabilization is required to assure a high quality of
the recording throughout the surgical procedure. In
general, the electrodes are placed after the patient is
anesthetized. Although in difficult or deep muscles,
they can be placed prior to the surgical procedure when
Fig. 5. Variety of recording electrodes including standard

monopolar bare-tipped needles, surface disc electrode, subdermal EEG needles, and fine wire Nichrome wires.
399
the patient is awake and able to activate the appropriate

muscles. After placement, these electrodes are secured
to the skin by first taping the exiting electrode and then
looping the wire and applying a second strip of tape
(Fig. 6A and B). They are then connected to the amplifier through either a spring-loaded connector or a modified integrated circuit connector held in their sockets
by circuit pins (Fig. 6C and D). The leads from the
probes to the preamplifier must be kept as short as possible to reduce interference from external sources. The
circuit or spring-loaded connectors are then taped to
the skin to prevent movement or dislodgement. Once
the patient has been scrubbed and draped, there is no
longer any opportunity to replace or reconnect wires
that are dislodged. Thus, it is imperative to assess the
quality of the recording before the procedure proceeds.
Commercially available monopolar needle with
bare tips can also be used for intramuscular monitoring and are available in various sizes from 25 to
greater than 100 mm (Fig. 5). These will generally
provide excellent recording with limitations being
the cost and the risk of dislodgement. In our experience, these are more difficult to secure since they
do not conform as well to the skin, even when the
hub is bent to parallel the skin surface. Since they
are often raised off the skin surface, they are more
likely to be either caught or bumped by the surgeon
or his assistants, leading to potential dislodgement
or artifactual discharges which may be falsely confused as neurotonic discharges. These electrodes are
frequently used for intramuscular recording directly
in the surgical field where they are placed, and their
position is monitored by the surgeon. In most circumstances, this application is for recording evoked
EMG rather than for free-running EMG. Bare-tip
subdermal EEG electrodes are being used more frequently as they are relatively easy to place and
appear to have less impedance problems with very
clean recordings (Fig. 7). They can be secured to
the skin and given the low profile hub are generally
able to conform to the skin, thus keeping out of the
way of the surgeon and anesthesiologist. These are
ideal for monitoring superficial muscles such as
facial muscles during either parotid or posterior
fossa surgeries, and superficial muscles of the limbs.
The primary disadvantage is that they do not allow
identification of deeper muscles. These needles in
the subcutaneous position can be very helpful in
recording evoked potentials, but for identifying
neurotonic discharges they need to be placed within
the target muscles.
400
Fig. 6. Fine wire EMG electrode placement. A 26-gauge needle with a fine wire electrode in the hollow core is placed in
the muscle (A). The needle is withdrawn and the electrode is taped to the skin then looped and taped again (B). The wire is
connected to the amplifier through either a circuit connector (C) or with spring-loaded connector (D).
Fig. 7. Intramuscular EEG electrode placement in the orbicularis oculi, orbicularis oris, and masseter during posterior fossa
surgery.
27.3.2. Electrode placement

As in routine EMG, an active and reference electrode
must be placed in each channel. A single large lead
ground can be used for all channels. Dependent on
the surgical procedure and the number of channels
available, the recording and reference electrodes
can be placed in an individual muscle or, if a broader
field is desired, in separate muscles. In most settings,
a broader field is adequate. Examples of this would
include spine surgery where the goal may be to cover
all roots at risk with minimal channels or alternatively to identify specific root innervation rather than
individual muscles. For specific root identification,
the active electrode could be placed in a muscle such
as the tibialis anterior and a reference electrode in the
vastus lateralis. If the fascicles from the fourth lumbar root are irritated, a neurotonic discharge should
be identified in one or both of these muscles. With
this type of montage, one could continuously monitor
four root levels on each side with a standard eightchannel machine. If root specificity is not required
or there are limited channels, the electrodes could
be placed in separate muscles with different roots
such as the tibialis anterior referenced to the gastrocnemius and the vastus lateralis to the rectus femoris.
401
With this montage, multiple root levels can be continuously monitored with fewer channels but specific
root identification can be difficult. In procedures that
risk damage to cranial nerves, specificity is often
more important. In those situations, two wires or needles will often be placed in the same muscle 5 mm
apart. In small muscles, such as extraocular muscles,
two wires can be placed simultaneously through a
single 30-gauge needle as long as the tips are bared
in different areas to avoid shorting of the signal
(Fig. 8).
27.3.3. Recording parameters
The EMG recordings are made with standard gains of
50500 mV, low-frequency filter (LFF) 2030 Hz,
high-frequency filter (HFF) 10 kHz, and sweep speed
of 10200 ms per division. Both audio and visual
feedback are important for immediate recognition,
localization, and later review. An audio loudspeaker
allows the surgeon and neurophysiologist immediate
feedback as to potential nerve irritation or injury.
The visual display allows more precise localization
with verbal feedback to the surgeon when appropriate. EMG activity of interest can then be printed or
stored for later review.
Fig. 8. Lateral rectus monitoring with fine wire EMG placement during surgeries in the region of the orbit, cavernous sinus, or
petrous region of the temporal bone. Note that in this setting, two wires are placed in the muscle with a single hollow core needle.
402
27.4. Pitfalls
As in all intraoperative recordings, the neurophysiologist or a monitoring technician must be aware of all
possible technical and physiologic variables which
could lead to a false-positive or false-negative result.
Fortunately, free-running EMG is a simple, reliable
technique which does not interfere with the surgical
procedure, provides immediate feedback, and is minimally affected by physiologic variables. The primary
issues are generally related to the electrically volatile
environment of the operative suite. All potential electrical interference must be identified and minimized.
The impedance of all electrodes should be assessed
with the internal mechanism provided on most
machines, keeping these less than 50 kO. If the
impedance is high on all channels, the ground electrode may be at fault or, if it is high on an individual
channel, the electrodes for that channel may need
to be replaced. In our experience, the short EEG
needle electrodes have the best impedance followed
by monopolar needles and fine wire electrodes.
A general rule of baseline recording noise less than
20 mV at rest should ensure clean recordings. If
the gain is raised to compensate for noisy recording,
small neurotonic discharges may not be seen leading
to false-negative results. The interference from cautery
is best suppressed with a switch attached directly to
the amplifier but one must recognize that neurotonic
discharges will not be identifiable during cautery
and this may be at a time when the nerves are at risk.
Neuromuscular blockade will significantly attenuate motor activity and should be avoided as much as
possible. Neurotonic discharges can, however, still
be recorded with neuromuscular blocking agents
producing up to 75% blockade which corresponds
to a compound muscle action potential amplitude
(CMAP) decrement of less than 100% over four
successive supramaximal repetitive nerve stimulations (Holland, 2002). Inhalation agents or narcotic
anesthesia is preferred and generally will prevent
unwanted movement during the operation. At times,
additional agents such as fentanyl or midazolam may
need to be administered to reduce background muscle
contractions and associated motor unit potentials.
Blood pressure and temperature have minimal effect
on free-running EMG.
In addition, common reasons for a false-negative
result relate to either diseased nerves or a sharp transection. Reports and experience would indicate that
a sharp transection of the peripheral nerve may be
associated with minimal or no neurotonic discharges

(Nelson and Vasconez, 1995). Once the nerve is
transected, an evoked response can still be recorded
if the distal segment is stimulated. Similarly, if there
is mechanical irritation to the distal segment, neurotonic discharges can be elicited. Failure to recognize
this may mislead the neurophysiologist and surgeon
as to the continuity of the proximal nerve segment. If
a nerve transection is suspected, electrical stimulation
of the proximal segment will not elicit an evoked
potential. Additionally, damaged nerve has clearly
been shown to have a higher stimulation threshold
(Holland et al., 1998) which along with fewer axons
and greater scar tissue may lead to reduced neurotonic
discharges.
27.5. Application
Free-running EMG recordings have diverse applications for peripheral, spinal, and cranial procedures.
Recordings can be made from any somatic muscle,
including extraocular facial, laryngeal, intercostal,
abdominal, anal sphincter, and any limb muscles.
Any procedure that requires manipulation or dissection of a nerve can be performed more safely and effectively with intraoperative EMG monitoring. This
technique is most commonly used for cranial nerve
monitoring and spinal surgery. For cranial nerves,
facial nerve monitoring is the most common application. Other applications include CNs III, IV, and VI
during surgery in the region of the orbit, cavernous
sinus, or petrous portion of the temporal bone; the trigeminal nerve during posterior fossa tumor resection
or microvascular decompression surgery; the vagus
nerve with intraoperative recording of the vocalis or
cricothyroid during brainstem or laryngeal procedures; and CNs XI and XII during vascular or tumor
cases involving the lower brainstem or region of the
foramen magnum (Strommen and Daube, 2000;
Harper, 2004). For spinal monitoring, the spinal roots
of the cervical, lumbar, or sacral segments can be
easily monitored during decompression, deformity
correction, fusion, or during pedicle screw placement
(Owen et al., 1994; Beatty et al., 1995; Holland,
2002; Gunnarsson et al., 2004). Cranial, peripheral
nerve, or spinal tumors can utilize free-running EMG
for nerve protection or localization when the anatomy
has been altered by the tumor. The roots of the cauda
equina can be monitored during tumor resection in
this region or release of a tethered cord (Kothbauer
et al., 1994). Peripheral nerves such as the femoral or
sciatic nerves can also be protected during revision hip

arthroplasty (Brown et al., 2002) and periacetabular
osteotomy (Pring et al., 2002). Neurotonic discharges
tend to occur more often with cranial nerve manipulation than with peripheral nerve manipulation (Harner
et al., 1987). Although a discharge alone is not predictive of neurologic deficit, the amount of discharges
and, in particular, the long neurotonic discharge does
appear to correlate with postoperative functional deficit
(Harner et al., 1987). Long neurotonic discharges are,
however, also commonly seen during benign events
such as irrigation where the nerve is irritated by a combination of temperature and mechanical stimulation.
Given these factors, the neurophysiologist must be
aware of the surgical manipulation, either by means of
frequent communication, direct visualization, or video
feedback from the surgical field.
27.6. Summary
Multimodal intraoperative neurophysiologic monitoring has become a critical component to assure good
outcomes in a variety of surgeries. The use of freerunning EMG is a relatively simple monitoring technique that is very useful to identify and protect
peripheral nerve structures at risk. Essentially, any
procedure that requires manipulation or dissection
of a nerve can be performed more safely and effectively with intraoperative EMG monitoring. Understanding the technique of needle placement, the
variety of recording electrodes, the pitfalls, and the
interpretation of findings is critical to providing
rapid, accurate feedback to the surgeon. As with
other intraoperative techniques, gaining experience
in this modality is paramount to mastering the field
of intraoperative neurophysiologic monitoring.
References
Beatty, RM, McGuire, P, Moroney, JM and Holladay, FP
(1995) Continuous intraoperative electromyographic
recording during spinal surgery. J. Neurosurg., 82:
401405.
Brown, DM, McGinnes, WC and Mesghali, H (2002) Neurophysiologic intraoperative monitoring during revision total
hip arthroplasty. J. Bone Joint Surg., 84(Suppl. 2): 5661.
Daube, JR and Harper, CM (1989) Surgical monitoring
of cranial and peripheral nerves. In JE Desmedt (Ed.),
Neuromonitoring in Surgery. Elsevier Science, Amsterdam, The Netherlands, pp. 115138.
403
Gunnarsson, T, Krassioukov, AV, Sarjeant, R and Fehlings,

MG (2004) Real-time continuous intraoperative electromyographic and somatosensory evoked potential
recording in spinal surgery: correlation of clinical and
electrophysiologic findings in a prospective, consecutive series of 213 cases. Spine, 29(6): 677684.
Harner, S, Daube, J, Ebersold, M and Beatty, CW (1987)
Improved preservation of facial nerve function with
use of electrical monitoring during removal of acoustic
neuromas. Mayo Clin. Proc., 62: 92102.
Harper, CM (2004) Intraoperative cranial nerve monitoring. Muscle Nerve, 29(3): 339351.
Harper, C and Daube, J (1998) Facial nerve electromyography and other cranial nerve monitoring. J. Clin. Neurophysiol., 15(3): 206216.
Holland, NR (2002) Intraoperative electromyography.
Holland, NR, Lukaczyk, TA and Kostuik, JP (1998) A
comparison of the stimulus thresholds required to evoke
myogenic responses from normal and chronically compressed nerve roots: implications for intraoperative testing during transpedicular instrumentation. Spine, 23:
224227.
Kothbauer, K, Schmid, UD, Seiler, RW and Eisner, W
(1994) Intraoperative motor and sensory monitoring of
the cauda equina. Neurosurgery, 34(4): 702707.
Marcus, B, Edwards, B, Yoo, S, Byrne, A, Gupta, A,
Kandrevas, J, Bradford, C, Chepeha, DB and Teknos,
TN (2003) Recurrent laryngeal nerve monitoring in
thyroid and parathyroid surgery: the University of
Michigan experience. Laryngoscopy, 113: 356361.
Nelson, KR and Vasconez, HC (1995) Nerve transection
without neurotonic discharges during intraoperative
electromyographic monitoring. Muscle Nerve, 18:
236238.
Owen, JH, Kostuik, JP, Gornet, M, Petr, M, Skelly, J, Smoes,
C, Szymanski, J, Townes, J and Wolfe, F (1994) The use
of mechanically elicited electromyograms to protect
nerve roots during surgery for spinal degeneration. Spine,
19(15): 17041710.
Pearlman, RC, Isley, MR, Ruben, GD, Sandler, SC, Wiesbaum, B, Ali Khan, M, Greene, BS, Charles, V and
Shah, A (2005) Intraoperative monitoring of the recurrent laryngeal nerve using acoustic, free-run, and
evoked electromyography. J. Clin. Neurophysiol., 22:
148152.
Pring, ME, Trousdale, RT, Cabanela, ME and Harper, CM
(2002) Intraoperative electromyographic monitoring
during periacetabular osteotomy. Clin. Orthop. Relat.
Res., 400: 158164.
Strommen, JA and Daube, JR (2000) Electrophysiological
studies in the operating room. Cranial nerve monitoring.
In: WF Brown, CF Bolton and MJ Aminoff (Eds.), Neuromuscular Function and Disease. WB Saunders, Philadelphia, pp. 18191837.

M.R. Nuwer (Ed.)
404
CHAPTER 28
Intraoperative EMG during spinal pedicle

screw instrumentation
Jeffrey R. Balzera,b,*, Donald Crammonda, Miguel Habeycha,1 and
Robert J. Sclabassia,b,c,d
a
Department of Neurological Surgery, University of Pittsburgh, Pittsburgh, PA 15213, USA

b
Department of Neuroscience, University of Pittsburgh, Pittsburgh, PA 15213, USA

c
Department of Psychiatry, University of Pittsburgh, Pittsburgh, PA 15213, USA
Department of Electrical, Mechanical and Biomedical Engineering, University of Pittsburgh, Pittsburgh, PA 15213, USA
28.1. Introduction
The objective of performing intraoperative neurophysiological monitoring during procedures involving decompression and instrumentation of the
cervical, thoracic, and lumbosacral spine is to detect
insults to the central and peripheral nervous systems
and the subsequent prevention of iatrogenic neurological injury. This notion is predicated on the fact
that the modalities that we use in the operating room
to perform neuromonitoring are able to provide for
early detection of reversible injury to both the spinal
cord and spinal nerve roots. To this end, we have
developed and successfully implemented a number
of different monitoring modalities including somatosensory evoked potentials (SEPs), dermatomal sensory
evoked potentials (DSEPs), motor evoked potentials
(MEPs), and free-run and stimulus evoked electromyography (EMG). Each of these modalities has its
advantages and disadvantages as it relates to which
segment of the spinal cord it assays or whether it
monitors cord versus single nerve root function.
The selection of the modality(ies) to be used during
these procedures is based on the nature of the procedure
and the potential and systems at risk for injury. We and
*
Correspondence to: Jeffrey R. Balzer, Ph.D., Department

of Neurological Surgery, University of Pittsburgh Medical
Center, Suite B-400, 200 Lothrop Street, Pittsburgh, PA
15213, USA.
Tel.: 1-412-648-2570; fax: 1-412-383-8999.
E-mail: balzerjr@upmc.edu (J.R. Balzer).
1
Present Address: Department of Neurological surgery,
UPMC Presbyterian Hospital, University of Pittsburgh,
Pittsburgh, PA 15213-2582, USA.
others routinely adopt a multimodality approach to

intraoperative neuromonitoring to allow for complete
coverage of the neural axis during these procedures.
The use of SEP and MEP monitoring has been well
documented during these types of procedures for the
prevention of spinal cord injury while their usefulness
in detecting and preventing single root injury has been
scrutinized. In response to this perceived shortcoming,
we and others have developed and adopted intraoperative EMG protocols that allow for single nerve root
protection and identification. This chapter will review
and discuss the free-run and stimulus evoked EMG
literature in the context of other modalities and how
these EMG modalities are implemented, performed,
and interpreted during decompressive and instrumented
surgical procedures in the cervical, thoracic, and
lumbosacral spine.
28.2. Decompression and instrumentation
of the spine
Decompressive and instrumented surgical procedures
in the cervical, thoracic, and lumbosacral spine date
back to the late 1800s and early 1900s (Hadra,
1891; Albee, 1911; Hibbs, 1911; Knoeller and
Seifred, 2000). These procedures are now commonplace and extensively utilized for the treatment of
spinal stenosis, compression, degenerative disc disease, pseudarthrosis, congenital deformities, and
spine trauma. In contrast to earlier surgical constructs, current instrumentation procedures commonly utilize pedicle screws either in place of wires
and hooks or exclusively. The impetus behind the
increased usage of pedicle screws is a result of the
numerous advantages that screw affixation offers
over more traditional constructs such as hooks and

rods. The two most important advantages in using
pedicle screw fixation are improved deformity correction and the overall strength of the construct.
These advantages are primarily due to the fact that
pedicular screw placement allows for three-column
control over the spine. Other advantages include the
lack of a need to place instrumentation within the
spinal canal as is the case for some hook placement
and wires and that screw placement is independent
of bony integrity which makes their use possible
in cases such as trauma and destructive lesions of
the spine.
The benefits of using pedicle screws in the cervical,
thoracic, and lumbosacral spine have been tempered
by the potential for significant neurological injury
due to the close proximity of the spinal cord, spinal
nerve roots, and vascular structures to the pedicle. As
a consequence, a misplaced screw could lead to significant iatrogenic intraoperative injury to the patient. As
a result, surgeons have employed a number of techniques to ensure safe and accurate placement of pedicle
screws. These include detailed anatomic landmarks
used to determine pedicle location and screw trajectory, intraoperative imaging including neuronavigation, and neurophysiological monitoring. The
implementation of neuromonitoring techniques, particularly as these pertain to the use of free-run and
stimulus evoked EMG techniques has proven to allow
for the safe and successful placement of cervical, thoracic, and lumbosacral pedicle screws.
28.3. Intraoperative EMG
The earliest development and applications of intraoperative EMG were in surgical procedures performed in and around the facial nerve, particularly
during acoustic neuroma resection (Delgado et al.,
1979; Mller and Jannetta, 1985; Prass and Luders,
1986; Benecke et al., 1987). Activity, in the form of
compound muscle action potentials (CMAPs), was
recorded from the musculature innervated by the facial
nerve using free-run and stimulus evoked EMG
techniques. It was at this time that investigators began
to classify the amplitude and frequency of intraoperative free-run EMG that posed a threat to nerve function
as well as correlate these intraoperative spontaneous
discharges with outcome. The use of terms such
as asynchronous firing, waning discharges, bursting,
and neurotonic discharge were used in the operating
room as they relate to spontaneous EMG discharges.
405
Concomitantly, hand-held electrified probes were

first used for direct monopolar stimulation of the
nerve and recording of stimulus evoked EMG for
purposes of identification, continuity, firing threshold
acquisition, and prediction of neurological outcome.
Each of these EMG techniques, which were found to
be very easily performed in the operating room,
provided for near-instantaneous feedback to the surgical team concerning the health of the facial nerve.
These early techniques set the stage for the future
development, modification, and adoption of EMG
recording techniques for nerve root protection in the
cervical, thoracic, and lumbosacral spine along with
the use of evoked potentials.
28.4. Cervical spine procedures
Complications in the form of upper extremity paresis
are commonly associated with iatrogenic injury and
subsequent pathology of the fifth cervical (C5) nerve
root (Sakaura et al., 2003). While other nerve roots
(C68) can succumb to the same type of injury, the
report of paresis in these nerves is significantly lower
than in the C5 root. Typically nerve root morbidity
occurs subsequent to decompression for cervical
myelopathy via both anterior and posterior surgical
approaches. The incidence of postoperative C5
palsies has been reported to range from 0 to 30%
depending on the approach which is utilized (Epstein,
2001; Sakaura et al., 2003). C5 nerve root palsies are
most commonly attributed to direct nerve root injury
secondary to manipulation or traction or a segmental
spinal cord injury secondary to ischemia. In an
attempt to reduce the incidence of both nerve root
and spinal cord injury during these procedures,
several groups have supported the use of SEP and
MEP monitoring (Fan et al., 2002; Bose et al.,
2004; Hilibrand et al., 2004; Khan et al., 2006)
during these procedures. Additionally, these same
groups and others (Jimenez et al., 2005) have suggested that the addition of free-run EMG monitoring
can further reduce morbidity during these procedures
particularly, injury to individual nerve roots.
28.4.1. Free-run EMG during cervical procedures
In a very large series of retrospectively reviewing
508 patients undergoing anterior cervical corpectomies for myelopathy, Khan et al. (2006) found that
SEP recording was a very sensitive measure of spinal
406
cord perturbation and injury, particularly associated

with hypotension, but the modality lacked in its ability to predict isolated single nerve root injury, particularly the C5 nerve root. They concluded that in
addition to SEP recording, these procedures should
also include free-run EMG recording so as to be able
to sensitively detect manipulation and potential
injury to single cervical nerve roots.
Confirming their suspicion, Fan et al. (2002)
conducted a dual study looking both retro- and prospectively at the ability of intraoperative neurophysiological monitoring during cervical laminectomy to
detect iatrogenic C5 nerve root injury. The study
reviewed a total of 200 patients who underwent cervical laminectomy. The first 132 patients, who were
monitored using SEP, DSEP, and MEP techniques,
were reviewed retrospectively. In this group, six
patients awoke with unilateral C5 nerve root palsies
in the absence of any significant changes in any
evoked potential modality. Subsequent to these findings, a prospective study was applied to the remaining 68 patients in the group using not only the
techniques applied in the initial 132 patients but also
now with the addition of MEPs and free-run EMG
recordings specifically targeting the deltoid and
biceps muscle groups. In this group, intraoperative
C5 nerve root compromise was detected in two
patients as a sudden onset of sustained asynchronous
neurotonic discharge in the free-run EMG and eventually as a significant change (>75%) in the MEP
recordings. In both cases, additional decompressive
procedures were undertaken in an attempt to rectify
the neurophysiological changes observed. In both
cases, free-run EMG subsided and in one, MEP
recordings improved. Both patients awoke with C5
nerve root palsies which ultimately improved. No
C5 palsies were observed in the absence of significant free-run EMG discharge or MEP changes.
To further support the notion of using free-run
EMG in cervical procedures for the protection of
single nerve roots, Jimenez et al. (2005) reviewed
161 patients who underwent 171 various cervical
procedures. In any and all cases where spontaneous
free-run EMG was observed, the surgeon took the
appropriate operative action if warranted (Fig. 1).
This study utilized a retrospective cohort as a control
group, that is, one with no EMG monitoring. In this
group of 55 patients, 4 (7.3%) had postoperative
C5 nerve root palsies. In the prospective cohort of
106 patients, spontaneous free-run EMG events
resulted in a change in either positioning or operative
J.R. BALZER ET AL.
technique in three patients (2.8%). Of the three

patients who showed spontaneous free-run EMG
activity, only one (0.9%) experienced a postoperative
C5 nerve root palsy. Like the Fan et al. study, no
postoperative nerve root palsies were observed in
any patient who did not exhibit spontaneous freerun EMG activity. Taken together along with our current experience with free-run EMG during cervical
corpectomy and laminectomy procedures, it appears
that the addition of free-run EMG to a routine multimodality approach is effective at reducing single cervical nerve root morbidity during these procedures.
28.4.2. Stimulus evoked EMG during
cervical procedures
In addition to anterior and posterior decompressive
procedures for myelopathy, posterior instrumentation
using screws placed in the lateral mass and/or pedicles of the cervical spine is being utilized with higher
frequency for both stabilization and fusion purposes.
Along with these posterior instrumentation procedures comes the potential for morbidity related to
the misplacement of screws (Ludwig et al., 2000;
Reinhold et al., 2007). Screw placement in these
structures poses risk to several important anatomical
and vascular structures including the spinal cord,
cervical nerve roots, and the vertebral arteries
(Ebraheim et al., 1999). While intraoperative sensory
and MEP techniques can provide information
concerning cord function and brainstem ischemia
secondary to vertebral artery injury, the addition of
free-run EMG, as well as stimulus-triggered EMG
may serve to detect screw malpositioning and prevent individual nerve root injuries.
While the utility of stimulus-triggered EMG in
the cervical spine has been described in procedures
involving tumor resection and nerve root identification (Guo et al., 2006), little data exist describing
the use of stimulus-triggered EMG in activating lateral mass and pedicle screws for the confirmation
of proper placement. In the only paper to date,
Djurasovic et al. (2005) describe a technique which
utilizes the same principles of current flow in intact
pedicles and threshold values utilized in the lumbosacral spine in the cervical region.
In their investigation, 26 patients undergoing
posterior spine procedures involving lateral mass or
pedicle screw instrumentation were prospectively
studied to determine the correlation between screw
activation thresholds and the position of the screws
407
Fig. 1. Neurotonic discharge associated with nerve root manipulation using surgical instrument. Note that activity from
contralateral root is quiet during this period. [Reprinted from Jimenez et al. (2005) with permission from the American
Association of Neurological Surgeons.)
in the pedicle. One hundred and forty-seven screws

(122 lateral mass and 25 C-7 pedicle screws) were
electrically stimulated and their thresholds for activation of muscle EMG recorded. Postoperative CT
scans were obtained in each patient and screw position was independently evaluated. EMG threshold
data and CT scans were compared to assess accuracy
of EMG in predicting misplaced screws. The authors
found that a stimulation threshold of 15 mA or
greater provided a 99% positive predictive value that
the screw was within the lateral mass or pedicle.
Thresholds of 1015 mA provided a 13% predictive
value that the screw was positioned correctly and a
threshold value of <10 mA provided a predictive
value of 100% that the screw was not positioned correctly. The authors recommended that if thresholds
are <10 mA then the screw should be explored,
repositioned or possibly removed. Threshold values

of 1015 mA typically represented a properly placed
screw but should always be explored and thresholds
of 15 mA or greater reliably predict correct screw
position. While upper and lower limit threshold
values are defined in this study with a high degree
of confidence, future studies defining cervical screw
thresholds will need to be completed in order to more
accurately define what is currently a relatively large
range of unpredictable thresholds.
28.4.3. Muscle group selections and
electrode placement
Bipolar pairs of needle electrodes should be placed
subdermally over the belly of each muscle of interest.
Electrodes should be placed 1 cm apart with care
408
not to scissor the electrode shafts during placement.

Because upper extremity musculature receives innervation from multiple roots and that one root can
innervate multiple muscles, EMG should be recorded
bilaterally from multiple muscle groups to increase
specificity of root activation particularly in instances
where free-run EMG is being utilized and specific
nerve root identification may be requested. To this
end, the deltoid (C5), biceps (C56), extensor carpi
radialis (C6), triceps and flexor carpi radialis (C7),
and abductor pollicis brevis (C8, T1) muscle groups
should be considered for recording in each case
depending on the levels being decompressed and/or
instrumented.
28.4.4. Recording and stimulation parameters
CMAPs should be recorded using filter settings of 30
to 1 K (low- and high-filter settings, respectively)
and a gain of 500. A time base of 2550 ms should
be utilized to properly resolve the CMAP. Monopolar
stimulation should be delivered using an insulated
ball-tip probe or the like with the return electrode
(needle) placed in or around the site of incision.
Stimulation current should be slowly increased until
either a stimulus evoked EMG response is observed
or a maximum current is reached. It is always advisable to use a positive control when stimulating to
verify proper conduction given that the desired result
is a negative one.
28.5. Thoracic spine procedures
With increasing frequency, pedicle screws are being
used in entirety of the thoracic spine for procedures
including the correction of spinal deformities and
spinal fixation for thoracic and thoracolumbar fractures (Liljenqvist et al., 1997; Masferrer et al.,
1998). With an increase in utilization of instrumentation, particularly pedicle screws, comes an increase
in potential morbidity with the placement of these
devices to the spinal cord, spinal nerve roots, and
thoracic vasculature. This morbidity may occur for
a variety of reasons. First, the pedicle size, particularly diameter, is smaller in the thoracic spine than
in the lumbosacral spine, the second and probably
the most problematic is the variation in medial and
rostral angulation that is encountered in the thoracic
pedicles (Krag et al., 1988; Vaccaro et al., 1995b),
and lastly, major vascular structures as well as the
esophagus are at risk for screw placement that may
J.R. BALZER ET AL.
breach the anterior cortex of the vertebral body

(Vaccaro et al., 1995a).
Correct pedicle screw placement is of great concern in light of the high incidence of misplacement
evidenced in cadaveric studies. These studies have
shown that cortical breaches in thoracic pedicles during placement of screws can occur in as many as
41% of screws (Vaccaro et al., 1995a; Xu et al.,
1998; Cinotti et al., 1999). Clinically, incorrect screw
placement and cortical pedicle breaches generally
occur at a rate of greater than 8% (Liljenqvist et al.,
1997) with rates of 40% or higher being reported
(Roy-Camille et al., 1996; Merloz et al., 1998). Secondary to this high level of screw misplacement and
potential clinical morbidity, a variety of electrophysiological measures have been successfully utilized
(SEP and MEP) (Pelosi et al., 2002; MacDonald
et al., 2003) as well as nonelectrophysiological techniques (laminotomies, plain film X-rays, fluoroscopy, image-guided technology) (Kalfas et al., 1995;
Odgers et al., 1996; Ferrick et al., 1997; Xu et al.,
1999) with varying degrees of success have been
employed to aid in the guidance and confirmation of
accurate pedicle screw placement.
An additional electrophysiological modality,
namely stimulus evoked EMG, may further aid in
the reduction of iatrogenic nerve root injury secondary to screw misplacement in the thoracic spine
region. The utilization of this technique has proven
quite useful in surgical procedures involving screw
placement in the lumbosacral spine (Calancie et al.,
1994; Rose et al., 1997). In an attempt to validate
the use of this technique in the thoracic spine for
screw placement, several authors have investigated
the application of the stimulus evoked EMG
technique in animal studies (Danesh-Clough et al.,
2001; Lewis et al., 2001). In the study performed
by Danesh-Clough et al. (2001), a total of 91 screws
were placed in the lower thoracic pedicles (T812)
of sheep. The group recorded stimulus evoked
EMG in response to pedicle screw placement from
the abdominal and intercostal musculature and found
that using a stimulus evoked EMG threshold of
<10 V was highly effective in predicting pedicle
breaches in the lower thoracic spine. Their established threshold provided for a sensitivity of 94%
and specificity of 90% and supports the transition
of this technique to a clinical setting.
In an additional animal study, Lewis et al. (2001)
also examined the ability of stimulus evoked EMG
to improve the accuracy of thoracic pedicle screw
placement by using a porcine model. They recorded

from the intercostal musculature and found that the
threshold values that they obtained inconsistently
identified medial thoracic pedicle wall breaches and
concluded that this method could not differentiate
screws that were properly placed in the thoracic pedicle from screws that broke through the medial wall.
Subsequent to these animal studies, a number of
clinical studies have been performed in an attempt
to characterize stimulus evoked EMG threshold parameters in humans and determine the clinical utility
of the technique in general in instrumented thoracic
spine procedures (Reidy et al., 2001; Raynor et al.,
2002; Shi et al., 2003). Reidy et al. (2001) prospectively studied 17 patients while stimulating 95 thoracic pedicle screws. The technique they used was
one where, prior to screw insertion after a pilot hole
was created, a K-wire was placed into the pedicle
and electrified and thresholds obtained. Confirmatory
CT scans were performed postoperatively to establish
screw positions in the thoracic pedicles. In their
series, there were eight unrecognized breaches of
the pedicle. The authors used a 7.0-mA threshold as
their indication for a misplaced screw and found the
sensitivity of EMG to be 50% in detecting breaches
in pedicles with specificity at 83% using this value.
Overall, pedicle screws were properly placed in 90%
of patients and along with the EMG results, they
concluded that the stimulus evoked EMG did not significantly improve the reliability of screw placement.
In 2002, a much larger prospective investigation was
performed by Raynor et al. evaluating the sensitivity of
stimulus evoked thoracic EMG recording to assess
proper screw placement in thoracic pedicles. These
authors reviewed stimulus thresholds and screw placement in 92 consecutive patients representing a total of
409
677 thoracic screws placed. Stimulus evoked EMG

was recorded (and thresholds established) from the rectus abdominis muscles in response to stimulation of
every screw placed from T6 to T12. In their study, the
screw demonstrating the lowest threshold value in each
patient was removed and the pedicle hole was reinspected. Screws were separated into three groups:
Group A had thresholds >6.0 mA and were in the pedicle (n 650), Group B had thresholds <6.0 mA but
pedicles had intact medial walls (n 21), and Group C
had thresholds <6.0 mA and had medial pedicle wall
breaches (n 6) (Fig. 2). Overall, 3.9% (27 of 677)
of all screws in this study had thresholds <6.0 mA
but only 22% of those (6 of 27) had breaches in the
medial wall of the pedicle. Because of a lack of absolute threshold value consensus, each screw threshold
that was <6.0 mA (Groups B and C) was further
compared with an overall mean threshold value calculated by averaging all the mean stimulus evoked
EMG values for every screw tested within each
patient. The percent decrease from this mean was
calculated for each individual screw with a low
(<6.0 mA) threshold. Group B screws had a mean
decrease of 54% from the mean of all other screws
and Group C screws had a mean decrease of 68.9%
from the mean of all other screws tested. The authors
concluded that thresholds >6.0 mA indicated a properly placed screw but if values fell below <6.0 mA,
further assessment needed to be done. If a threshold
of <6.0 mA was obtained coupled with values that
were 6065% decreased from the overall mean
threshold then the surgeon should be alerted and the
medial wall of the pedicle inspected. These results
stand in contrast to a somewhat definitive cut-off
threshold value that has been established in the lumbosacral spine and has been used with great success
Group A: >6.0 mA threshold

and completely in the pedicle.
Group B: <6.0 mA threshold
and completely in the pedicle.
Group C: <6.0 mA threshold
and medial wall perforations.
A and B
Fig. 2. Screw placement groups used for threshold classification in the thoracic spine. [Reprinted from Raynor et al. (2002)
with permission from Lippincott, Williams and Wilkins.]
410
J.R. BALZER ET AL.
for at least 10 years now. While only a small percentage (5%) of screws fell into the <6.0 mA range, it
seems somewhat impractical in the intraoperative
environment to one, garner a threshold for every
screw tested and two, if any one screw is <6.0 mA
go on to calculate an average and subsequent deviation from that number. It is possible that with further
experience with thoracic screw stimulation and the
establishment of thresholds, a more comparable
cut-off value might be established.
In the only other study to date evaluating EMG
thresholds and thoracic pedicle screw placement, Shi
et al. (2003) retrospectively reviewed their experience
with the placement of 87 screws in 22 patients placed
at various levels in the thoracic spine ranging from T1
to T12. In all, five screws (5.7%) were misplaced in this
series as confirmed by postoperative CT scan. Six
screws (6.9%) were found to have stimulation thresholds of 11 mA. An example of stimulus evoked
thoracic EMG can be seen in Fig. 3. Of these six screws,
three were found to breach the pedicle on postoperative
CT scans. Of the 81 screws with thresholds >11 mA,
79 were found to be entirely within the bony column
of the pedicle (97.5% negative predictive value). These
results are more consistent with results in the lumbosacral spine both with respect to the threshold cut-off
values obtained as well as the consistency of the results.
Taken together, these three studies are inconsistent at
best and exemplify the need for stimulus evoked
EMG screw testing in many more cases as well as an
accepted protocol primarily as it relates to electrode
placement and stimulus parameters used for stimulus

evoked EMG recording.
28.5.1. Muscle group selections and
electrode placement
To effectively record stimulus evoked EMG in
response to the electrification of thoracic pedicle
screws, one would need to place electrodes over the
intercostal muscles (T26), the abdominal muscles
(T712), and in the case of T1, the flexor carpi
ulnaris. In practice, these muscles and more specifically their landmarks are not always easily identified
especially in morbidly obese patients. A lack of
proper electrode placement secondary to an inability
to palpate the muscles of interest could lead to high
false-negative rates due to a general lack of sensitivity and specificity. Several published reports have
recommended recording electrode montages for thoracic EMG recording. Raynor et al. (2002) suggest
using bipolar needle electrodes placed in each rectus
abdominis muscle lateral to the linea alba. This technique would require individual palpation and identification of individual musculature which, in most
cases, would be very difficult to achieve with a high
degree of specificity. Shi et al. (2003) provide a
detailed description of electrode placement for
recording thoracic EMG. For stimulation of screws
at T26, they record from intercostal muscles with
pairs of needle electrodes placed in the various intercostal spaces along the nipple line. For recording
Left
Abdominal wall
Quadriceps
Evolved EMG activity
Anterior tibialis
Right
Abdominal wall
Quadriceps
Anterior tibialis
Fig. 3. Stimulus evoked EMG recorded from bipolar pairs of needle electrodes placed over the abdominal musculature in
response to lower thoracic pedicle screw stimulation. Note that activity was unilateral and in response to the side of screw
stimulation. [Reprinted from Shi et al. (2003) with permission from Lippincott, Williams and Wilkins.]
411
It is always advisable to use a positive control when

stimulating to verify proper conduction given that the
desired result is a negative one. The threshold (intensity
needed to produce a repeatable, observable response)
should be recorded for each screw stimulated.
28.6. Lumbosacral spine procedures
Fig. 4. Placement of electrodes for recording stimulus

evoked EMG activity in response to lower thoracic (T712)
screw stimulation. [Reprinted from Shi et al. (2003) with
permission from Lippincott, Williams and Wilkins.]
stimulus evoked EMG in response to stimulation of

screws placed from T7 to T12, bipolar pairs of electrodes were placed in the abdominal musculature at a
point along the nipple line that equally divides the
distance between the lower margin of the tenth rib
and the iliac ridge (Fig. 4). Despite a detailed
description of electrode placement, the same challenges exist with regards to accurate placement of
electrodes as it relates to muscle specificity.
Like cervical studies, thoracic stimulus evoked EMG
should be recorded using filter settings of 30 to 1 K
(low- and high-filter settings, respectively) and a gain
of 500. A time base at least 50 ms should be utilized
to properly resolve the CMAP response. Monopolar
constant current stimulation (13 Hz, 0.2 ms duration)
should be delivered using an insulated ball-tip probe
or the like with the return electrode (needle) placed in
or around the site of incision. Stimulation current should
be slowly increased either until a stimulus evoked EMG
response is observed or a maximum current is reached.
The incidence of neurological injury during the insertion

of lumbar and sacral pedicle screws can range as high as
15% (Aebi et al., 1987; Whitecloud et al., 1989; West
et al., 1991; Blumenthal and Gill, 1993; Esses et al.,
1993; Lenke et al., 1995; Lonstein et al., 1999). The
iatrogenic injury that occurs during these procedures
typically results from a breach of the pedicle wall
resulting in screw penetration either into the spinal canal
and thecal sac or into the neural foramen potentially
injuring or irritating the lumbar and sacral nerve roots.
For these reasons, it is of great importance to be able
to place screws with great accuracy and assay nerve root
function during these procedures.
Pedicle screw misplacement has been reported by
several authors to occur with some frequency despite
the use of various intraoperative nonelectrophysiological methods (Weinstein et al., 1988; Gertzbein and
Robbins, 1990; Steinmann et al., 1993; Farber et al.,
1995; Castro et al., 1996; Ferrick et al., 1997; Laine
et al., 1997; Schwarzenbach et al., 1997; Isley
et al., 1997; Girardi et al., 1999). To this end, several
electrophysiological techniques have been utilized
with report of varying degrees of success and utility.
These modalities include SEPs and DSEPs.
28.6.1. SEPs during lumbosacral spine procedures
SEP monitoring has been reported to have intraoperative sensitivity and utility in lumbosacral procedures
in both animal (Jou, 2004; Tsai et al, 2005) and clinical studies (Balzer et al., 1998; Norcross-Nechay
et al., 1999; Weiss, 2001; Krassioukov et al., 2004;
Tsirikos et al., 2004). In fact, Norcross-Nechay
et al. (1999) in a study reviewing 90 patients reported
that acute, unilateral unresolved intraoperative
evoked potential deterioration was associated with
long-term ipsilateral weakness and that the findings
in their report support the need for monitoring SEPs
during surgery in all patients undergoing invasive
lumbar surgery. Additionally, Balzer et al. (1998)
reported the high sensitivity of SEP recordings
in detecting injury and subsequent neurological
sequelae during lumbosacral pedicle screw fusion
412
procedures. There are also reports in the literature that

have downplayed the utility of SEP monitoring during
lumbar fusion procedures (Gundanna et al., 2003;
Gunnarsson et al., 2004). Specifically, Gundanna et al.
(2003) retrospectively reviewed 186 consecutive
patients and found that none of the 186 had significant
SEP changes using 50% decrease in amplitude and
10% increase in latency as their alarm criteria. Despite
this lack of change, there were five patients with new
postoperative radiculopathies all of which were associated with malpositioned pedicle screws confirmed
using postoperative CT scanning or plain film radiographs. The authors concluded that the use of SEPs in
evaluating pedicle screw placement or detecting single
nerve root injury was of little utility and that methods
with greater sensitivity should be explored. The potential limitation to SEP recording in these cases arises
because stimulation of the posterior tibial nerve results
in the activation of multiple lumbosacral nerve roots
and the averaged SEP response may not change significantly secondary to a single root being damaged.
Despite this physiological shortcoming, we routinely
make SEP recording part of our multimodality approach
to monitoring of lumbosacral spine decompressions and
instrumentation procedures.
28.6.2. DSEPs during lumbosacral procedures
In an attempt to record evoked potentials which might
reflect the function of single lumbar and sacral nerve
roots, DSEPs have also been utilized intraoperatively
during lumbosacral spine procedures (Aminoff et al.,
1985; Herron et al., 1987; Owen et al., 1993; Toleikis
et al., 1993; Tsai et al., 1997). As with SEPs, the success
of DSEPs in assaying single nerve root function
during lumbosacral decompression has been mixed. In
two separate animal studies performed in rodents (Jou,
2004; Tsai et al., 2005), DSEPs were shown to be valuable in identifying and detecting acute single nerve root
injury. In an early clinical study performed by Herron
et al. (1987), the sensitivity of intraoperatively recorded
DSEPs was evaluated in patients with lumbar spinal
stenosis. In 30 patients, they found that DSEPs could
be routinely generated and in most patients found that
DSEP measurement pre- and postdecompression was
helpful in assessing the adequacy of the intraoperative
decompression. Likewise, Toleikis et al. (1993) studied
81 patients undergoing lumbosacral intrapedicular
fixation and was able to record repeatable responses
in all but one of these patients. Moreover, they found
that DSEP recording was sensitive to nerve root
J.R. BALZER ET AL.
compromise but pointed out that the prediction of postoperative deficits was dependent only on the presence
of the responses at the end of the procedure and not
on any changes that may have occurred during the
procedure.
In contrast to these findings, Tsai et al. (1997) found
that intraoperative DSEP recording was not only technically challenging but intraoperative changes that
were observed did not correlate to the patients clinical
outcomes. Specifically, useful DSEP recordings were
obtained in only a little over 50% of the patients. These
findings of inconsistency, unreliability, and insensitivity of intraoperative DSEP recordings are also in
agreement with other intraoperative investigations
using DSEP recording (Aminoff et al., 1985; Owen
et al., 1993).
Because of the reported potential shortcomings of
SEP and DSEP recording for the detection of single
nerve root injury during lumbosacral procedures,
EMG techniques have been developed which carry
a significantly higher degree of sensitivity and specificity with regard to the identification and prevention
of single nerve root iatrogenic injury during these
procedures. Other advantages of EMG recordings
include near-real-time recordings and instantaneous
intraoperative feedback as opposed to averaged
responses as well as the technical ease with which
the modality is implemented and interpreted. Each
of these, as well as other features of both free-run
and stimulus-triggered EMG during lumbosacral
spine procedures will be discussed in the following
sections.
28.6.3. Free-run EMG during lumbosacral
spine procedures
Free-run or spontaneous EMG recording is not a
new modality to be applied in the operating room
for single nerve root protection and has been used
extensively for the protection of cranial nerve function during a variety of intracranial procedures.
As mentioned previously, a large body of literature
exists concerning the successful application of this
EMG modality to procedures conducted on or near
the facial nerve (Mller and Jannetta, 1985, 1987;
Harner et al., 1987; Dickens and Graham, 1991).
This and other literature describing facial nerve
EMG monitoring has helped define the alarm criteria
that we currently use in the lumbosacral spine. Alarm
criteria for free-run EMG are defined by the amplitude and frequency of spontaneous firing patterns
that are observed and how this change in firing correlates with the surgical maneuver being performed at
that time, for example, nerve retraction, the patient
being under anesthetized, or saline irrigation in the
surgical site.
In the case of free-run EMG, baseline EMG (that
is recorded just after electrode placement) is defined
by the lack of any spontaneous activity or a quiet
recording. The key to utilizing free-run EMG successfully in the operating room is understanding the
significance of a situation where the EMG baseline
is no longer quiet. Spontaneous EMG activity can
occur in many forms and as a result of a number of
different manipulations. Identification of nonpathological spontaneous EMG versus spontaneous EMG
which has been initiated via a mechanical source
(traction, stretching, compression, or manipulation)
is crucial. Nonpathologic EMG firing patterns are
typically characterized by small amplitude, low frequency, or isolated discharges occurring at times
which do not correlate with surgical manipulation
of nerve roots. In these instances, while any and all
activity is made audible to surgical team, a high level
of alert is not established.
Pathologic spontaneous EMG activity has been
described by a number of groups (Hormes and
Chappuis, 1993; Owen et al., 1994; Beatty et al.,
1995; Holland and Kostuik, 1997; Balzer et al., 1998;
Holland, 1998; Bose et al., 2002). These studies have
413
correlated pathologic EMG activity with poor outcome

or have shown that corrective measures to lessen or
eliminate this type of EMG activity have resulted in
patients succumbing to no new neurological deficits.
In an attempt to correlate surgical maneuvers with
EMG discharge, Obi et al. (1999) methodically studied
the discharge of lumbar nerve roots in response to different manipulations such as tapping the root, hard
mechanical hits, and steady tensile force with retraction. In accordance with their results, it is generally
agreed that sustained and prolonged train EMG
activity which can last for many minutes (Fig. 5)
and bursting asynchronous EMG patterns termed
neurotonic discharges (Fig. 6) both can result in iatrogenic nerve root injury and portend poor neurological
outcomes. In either case, these types of EMG activities are immediately reported to the surgeon so that
corrective maneuvers may be taken to alleviate whatever might be causing these types of pathological discharges or in many cases, the surgeon will be asked to
stop whatever he is doing until the EMG activity normalizes to a quiet background.
Unlike SEP and DSEP monitoring which require
responses to be averaged and consequently take time
to collect and analyze and may not be sensitive to
single nerve root damage, free-run EMG is a powerful and simple tool that provides a near-instantaneous
assessment of individual nerve root irritation or
injury. This type of early-warning feedback in the
Fig. 5. Sustained high-frequency spontaneous EMG discharge noted in response to traction of the right S1 nerve root during
decompression. Columns 13 represent the right quadriceps, anterior tibialis, and gastrocnemius muscle groups, respectively.
414
J.R. BALZER ET AL.
Fig. 6. Neurotonic high-frequency bursting in response to traction of the left L5 nerve root during placement of interbody implant.
form of free-run EMG combined with other monitoring modalities such as SEPs provides the surgical
team with an extensive intraoperative understanding
of the patients electrophysiological status during
lumbosacral spine procedures. This multimodality
approach, as discussed above, continues to be invaluable in the prevention of iatrogenic injury during a
variety of spine procedures.
28.6.4. Stimulus evoked EMG during lumbosacral
procedures
As described earlier, several nonelectrophysiological
methods have been employed to prevent the misplacement of lumbosacral pedicle screws and
subsequent neurological morbidity. Despite these
tools, their employment has had limited success
based on the significant numbers of misplaced pedicle screws that have been reported in the literature.
To this end, several electrophysiological techniques
have been employed including SEP monitoring.
While the shortcomings of some of these techniques
have already been discussed, of all of the electrophysiological techniques utilized during instrumentation in the lumbosacral spine, stimulus evoked EMG
has been the most accurate in evaluating and predicting correctly placed pedicle screws (Calancie et al.,
1994; Glassman et al., 1995; Lenke et al., 1995;

Maguire et al., 1995; Clements et al., 1996; Darden
et al., 1998; Rose et al., 1997; Welch et al., 1997;
Balzer et al., 1998).
It was Rosen (1991) who first described the use of
electrification of the implanted pedicle screw using
an improvised stimulator in humans. While Rosen
did not directly measure stimulus evoked EMG activity from the lower extremities, he did use observational muscle twitching as his measure of whether
or not the screw was properly placed. Calancie
et al. (1992) advanced the notion of the use of pedicle screw stimulation for accurate pedicle screw
placement by recording muscle EMG in response to
pedicle screw stimulation using electrified instruments in a porcine animal model. The entire principle
of the pedicle screw stimulation technique is based
on the determination and evaluation of stimulus
evoked EMG thresholds subsequent to screw stimulation. If, after screw placement, the column of bone,
that is, the pedicle is intact, then a high resistance
pathway (i.e., bone) will exist between the medial
screw threads and the nerve root and/or the thecal
sac. The existence of this high resistance pathway
results in higher thresholds for EMG activation to
pedicle screw hole or pedicle screw stimulation. If,
on the other hand, the pedicle is breached then a
415
Fig. 7. Lumbar spine model and CT scan used to depict pedicle breach. Note that if the screw is electrically stimulated, current
will pass from the screw to the surrounding neural elements rather than encountering a high-resistance barrier such as bone.
low resistance pathway allows for current flow from

the screw to the root and results in low EMG activation
thresholds to screw stimulation (Fig. 7). After defining
the principle of using pedicle screw stimulation and
evoked-EMG thresholds to identify misplaced screws,
many authors, including Calancie et al. (1994) began
to systematically define stimulation and recording
protocols as well as the threshold ranges which defined
correctly versus incorrectly placed pedicle screws in
the lumbosacral spine (Glassman et al., 1995; Lenke
et al., 1995; Maguire et al., 1995; Young et al., 1995;
Clements et al., 1996; Rose et al., 1997; Welch et al.,
1997; Balzer et al., 1998; Darden et al., 1998; Toleikis

et al., 2000; Bose et al., 2002; Leppanen, 2005).
While each of these contributions have defined
different methods by which to activate pedicle screws
and record stimulus evoked EMG (i.e., constant current
versus constant voltage, stimulator type, stimulation
parameters, muscle group selection), different thresholds for defining proper versus improper placement,
and different means by which to interpret these thresholds, all have promoted the use of the technique as
an effective means by which to guide pedicle screw
placement in the lumbosacral spine (Fig. 8).
Fig. 8. Example of stimulus evoked EMG in response to stimulation of the right L5 pedicle screw at a low (5 V) intensity.
Note that activity is not only recorded from the right anterior tibialis muscle group but also from the quadriceps and
hamstrings suggesting current spread to the thecal sac and a pedicle breach.
416
The first comprehensive clinical study investigating the application and evaluation of this pedicle
screw stimulation technique was performed by
Calancie et al. (1994) where they evaluated 102
pedicle screws placed in 18 patients. It was in this
study that Calancie first described the use of constant
current stimulation delivered to a variety of surgical
instruments during various portions of the surgical
procedure, recording of stimulus evoked EMG
from multiple bilateral muscle groups, the use of a
searching stimulus intensity (7 mA), and the
recommendations that, if possible, EMG thresholds
to direct nerve root stimulation be established as a
baseline recording and that no pharmacological paralytic agents be used during testing (i.e., patients
have four of four twitches). Calancie et al. concluded
that the technique was sensitive and reliable in the
detection of perforations in the pedicle wall.
They went on to define that an EMG threshold of
10 mA in response to pedicle screw stimulation
was consistent with screw placement which was
entirely within the confines of the pedicle bone.
It was after the publication of this study that other
authors began to further evaluate the technique. In
1995, Maguire et al. evaluated stimulus evoked
EMG in 29 patients. Using both constant current
and constant voltage stimulation, EMG thresholds
were determined after electrification of 95 drill bits,
144 screws and 26 exposed nerve roots. In contrast
to the findings of Calancie et al., they found that an
EMG threshold of >6 mA (2 standard deviations
above the mean for direct lumbar nerve root
stimulation) was predictive of proper screw placement with a 98% degree of sensitivity. They also
state that they performed much of their screw testing
under partial paralysis and state that their goal was to
maintain muscle paralysis at 50% or less using a
continuous infusion technique. The last pertinent
point they make is that constant current seems to be
less variable than constant voltage which also was
an observation that Calancie et al. made in their
study.
In two additional publications in 1995, both of
which utilized constant current stimulation, Lenke
et al., in a combined animal and human study, and
Glassman et al., in the largest published series at that
time, again explored the application and sensitivity of
the pedicle screw stimulation technique in detecting
pedicle breaches during and after screw placement.
Lenke et al. (1995) studied 233 pedicle screws placed
in 54 patients and found that 93% of correctly placed
J.R. BALZER ET AL.
screws had stimulus evoked EMG thresholds >8.0 mA

under conditions where muscle relaxant was maintained at two to three twitches using train-of-four technology. If EMG thresholds were <4.0 mA, there was
a strong likelihood of a pedicle wall breach. In a prospective study, Glassman et al. (1995) studied 90
patients undergoing lumbosacral fusion procedures,
testing 512 screws. These investigators utilized postoperative CT scans to document screw placement. They
found that stimulation thresholds >15 mA provided
a 98% confidence that the screw was within the
pedicle. Stimulus evoked EMG thresholds between 10
and 15 mA were usually associated with adequately
placed screws (87% confidence level) although the
authors recommended screw exploration. Stimulation
thresholds <10 mA were associated with pedicle
breaches in most cases.
Clements et al. (1996) performed a prospective
study that evaluated 112 pedicle screws in 25
patients. Using a monopolar stimulator under conditions where the patient had no detectable pharmacological paralytics in their system (four of four
twitches), they used constant current stimulation
which was increased until a consistent EMG response
was obtained or to a maximum of 30 mA current
intensity. One hundred of the 112 pedicle screws that
were correctly placed had stimulus evoked EMG
thresholds of 11 mA. The remaining 12 screws
were found to have breached the pedicle wall and
all were found to have thresholds of <11 mA.
In a study performed at the University of Pittsburgh by myself and colleagues (Rose et al., 1997),
we prospectively studied 42 patients and 173 pedicle
screw insertions using a persistently electrified
pedicle stimulation technique. Rather than using
intermittent stimulation with electrified instruments,
this method persistently electrifies any and all instruments being used by the surgeon and provides continuous feedback via electrophysiologic monitoring
during preparation for, and placement of pedicle
screws. This method allows for the detection of
potential pedicle breaches during the dynamic
phases of the surgery such as probing of the pedicle
with a probe. Stimulation was performed using
constant-voltage square-wave pulses with intensities
ranging from 0.5 to 150 V under conditions of no
muscle relaxation. When possible, for each site that
a screw was to be placed, we determined voltage
thresholds for intact bone at the pedicle roof, spinal
nerve epineurium, and exposed cancellous bone
once the pedicle was unroofed. For bony thresholds,
stimulation intensity was set at 100 V and was gradually decreased until evoked EMG activity was
no longer observed. Subsequent search stimuli
(persistent stimulation) were set at a value between
the nerve and bone thresholds. If nerve thresholds
were not determined, then intensity was set 20%
below the bone threshold. Using this stimulus intensity, no evoked EMG activity was observed as long
as the pedicle was intact. Typically, the intensity
ranged from 20 to 40 V. We found that at each site,
bone thresholds were typically at least three times
as high as nerve root thresholds and, consequently,
persistent stimulation was set at least two times
as high as nerve threshold (just under bone threshold). We recorded nerve thresholds that ranged
from 3 to 15 V and bone thresholds ranging from
25 to 130 V. Using this technique, no false-positive
or false-negative findings were observed in this
study.
In the largest series to date, Toleikis et al. (2000)
performed a prospective study that included 662
patients and placement and testing of 3,409 pedicle
screws. Using multiple lower extremity muscle
recordings under condition of minimal muscle
relaxation, the authors utilized a hand-held insulated nasopharyngeal electrode to deliver constant
current stimulation to electrically activate screws.
Stimulation was begun at 0.0 mA and gradually
increased until a stimulus evoked EMG response
was observed or up to a maximum intensity of
50 mA. Of the 3,409 screws tested, 133 screws
had stimulus evoked EMG thresholds of 10 mA.
After inspection, 82 of these 133 screws were
not removed or redirected, 18 were redirected, and
33 removed and not replaced. Twenty-one screws
had EMG thresholds of 5 mA and of these, 19
were removed. Thirty-seven screws had EMG
threshold intensities between 5 and 7 mA. Of these,
22 were left in place and 15 removed. Toleikis et al.
found that even when stimulation threshold intensities were between 7 and 10 mA, close inspection
typically found a crack in the pedicle or one or
two threads of the screw breaching the medial wall
of the pedicle and over 75% of these screws were
left in place without redirection. The authors conclude that EMG threshold intensities of 10 mA
represent a high degree of confidence that the pedicle screw is properly placed. Under any other
circumstances (i.e., <10 mA), the screw placement
should be inspected and scrutinized.
417
28.6.5. Sources for discrepancies in thresholds and

false negatives
As evidenced from the literature review above, a
final consensus concerning an EMG threshold which
reflects accurate screw placement has yet to be
achieved, although a very narrow range of thresholds
exists for what is considered a properly placed screw
and activation thresholds <6 mA almost always indicate a pedicle breach. Thresholds indicative of proper
screw placement in the lumbosacral spine range from
6 to 15 mA with varying degrees of sensitivity. At
first glance, one might conclude that these studies
were performed using very similar protocols but this
was not the case. Numerous protocol differences
existed as well as intangible circumstances which
also could lead to the observed differences. Protocol
differences included the stimulus frequencies used
(range 15 Hz) and pulse durations (range 50
300 ms) utilized as well as the general stimulation
technique employed including when the assay was
conducted, that is, prior to and after screw placement
or only after the screw was placed, or whether or not
baseline nerve root thresholds were obtained.
False-negative results (and some of the discrepancies noted above) can occur secondary to a variety of
factors. Some technical factors influencing threshold
and consequently having the potential to result in
false-negative results include current shunting secondary to blood and irrigation fluid in the wound or
screw pilot hole (Skelly et al., 1999), the type of
screw being utilized for the fusion (Anderson et al.,
2002), and the degree of muscle relaxation present
when the threshold testing is performed. With regard
to confirming the degree of muscle relaxation, a
train-of-four test from the limb of interest as well
as a positive control (i.e., stimulating the thecal sac
directly and inspecting for a response) is highly
recommended. A major physiological variable that
can result in false-negative results pertains to the
health of the nerve at the level where the screw is
being placed. Several authors (Maguire et al., 1995;
Holland et al., 1998) have reported abnormally high
nerve root thresholds in response to direct stimulation
under conditions where chronic compression was
suspected. If nerve root activation thresholds are high
at baseline, then stimulus evoked EMG thresholds in
response to screw stimulation also can be significantly elevated. In this scenario, even modest stimulus intensities may not result in muscle EMG even
418
J.R. BALZER ET AL.
in the face of a screw that is malpositioned. A second

factor that may lead to elevated thresholds and, as a
consequence, the same false-negative result, is diabetes (Toleikis et al., 2000). In both cases, it is advisable to garner single nerve root thresholds whenever
possible to allow for adjustment of critical threshold
values dictating accurate screw placement.
28.6.6. Muscle group selections and electrode
placement
Bipolar pairs of needle electrodes should be placed
subdermally over the belly of each muscle of interest.
Electrodes should be placed 1 cm apart and affixed
with tape with care not to scissor the electrode shafts
during placement. We recommend never placing
needle electrode through compression stockings due
to a high probability of electrodes backing-out during the procedure. Because lower extremity musculature receives innervation from multiple roots and
that one root can innervate multiple muscles, EMG
should be recorded bilaterally from multiple muscle
groups to increase specificity of muscle activation.
As was discussed in the previous sections describing
electrophysiological techniques used in the cervical
and thoracic spine, muscle selection for stimulus
evoked EMG recording is dictated by the levels of
the lumbar and sacral spine that will be instrumented
during the surgical procedure. We routinely record from
five different muscle groups bilaterally utilizing the
adductor magnus (L24), rectus femoris (L34), tibialis
anterior (L45), biceps femoris (L5S1), and triceps
surae (S12) muscle groups (Table 1). Additional
Table 1
Muscle group options for recording stimulus evoked
EMG during lumbosacral fusion
L24
L34
L45
L5S1
S12
S25
Adductor magnus
Rectus femoris* (quads: extension)
Vastus lateralis (abduction)
Tibialis anterior* (dorsiflexion)
Biceps femoris* (hamstrings)
Gluteus maximus (S1, L5)
Triceps surae (gastrocnemius* and soleus:
plantarflexion)
Perianal musculature*
We typically (*) record from at least four muscle groups per limb
and often add the perianal musculature depending on the levels
being decompressed and fused.
muscle groups should be considered (e.g., anal sphincter) for recording in each case depending on the levels
being decompressed and/or the potential for increased
morbidity such as a redo fusion.
28.6.7. Recording and stimulation parameters
CMAPs should be recorded using filter settings of 30
to 1 K (low- and high-filter settings, respectively) and
a gain of 500. A time base of 50 ms should be utilized
to properly resolve the lower extremity CMAP. For
direct screw stimulation, monopolar, cathodal, constant
voltage stimulation is delivered using an insulated
ball-tip probe or the like with the return electrode
(needle) placed in or around the site of incision. For
direct nerve root stimulation, a Prass monopolar
stimulator is utilized, also utilizing constant voltage
stimulation. At our institution, we also utilize electrified
instruments throughout the procedure to dynamically
assess pedicle integrity prior to screw placement and
detect a breach before the screw is placed. Stimulation
current, delivered at a 5-Hz rate and 200 ms duration,
should be slowly increased either until a stimulus
evoked EMG response is observed or a maximum
current is reached. It is always advisable to use a
positive control when stimulating to verify proper conduction given that the desired result is a negative one.
28.7. Anesthetic considerations
As had been elucidated throughout, anesthetic considerations during decompressive and fusion procedures
in the cervical, thoracic, and lumbosacral spine are
critical particularly as they relate to the use of pharmacological paralytic agents. Because both spontaneous
and stimulus evoked EMG are essential components
to the monitoring approach in all of the procedures
discussed in this chapter, the level of pharmacological
paralytic needs to be closely regulated and measured.
Additionally, SEP recording also is an integral component to the overall multimodality neuromonitoring
approach. It is our recommendation that a typical
balanced anesthetic technique be utilized except
for continued administration of paralytic agents. We
recommend that anesthesia be induced with thiopentalsodium, etomidate, or propofol and intubation be
performed under the influence of short-acting muscle
relaxants such as succinylcholine or rocuronium. No
further muscle relaxant should be administered until
all instrumentation is complete and closure begins.
In certain circumstances, additional short-acting

agents can be used to aid with exposure of the spine
but should be terminated and four of four twitches
should be obtainable upon commencement of decompression and/or instrumentation and any time EMG is
being recorded. We, like Calancie et al. (1994), highly
recommend recording EMG only when four of four
twitches are obtainable because of the potential
confounds with regard to amplitude and threshold
determinations if the patient is partially relaxed.
We would further advocate the measurement and
documentation of the number of twitches generated
in response to train-of-four stimulation and recommend
that the train-of-four testing occurs from the limb
where EMG activity is being recorded. Anesthesia
should be maintained with any of the halogenated
agents and nitrous oxide with care not to exceed a
combined total of 1 minimum alveolar concentration
(MAC). Fentanyl or propofol infusions can be used
to augment the anesthetic regimen. The levels of
these gases can be manipulated depending on the
point in the surgery. For example, during dynamic stimulus evoked EMG testing (i.e., placement of instrumentation), we typically are not continuously collecting
SEP data so an increase in volatile inhalational agents
is acceptable especially in the face of our request
for no muscle relaxation at this time.
28.8. Conclusions
Decompression and instrumentation of the cervical,
thoracic, and lumbosacral spine can carry a significant risk of neurological injury due to a variety of
circumstances including vascular injury, spinal cord
compression, and contusion and direct nerve root
injury. Each of these iatrogenic events can result
in new neurological deficits in this patient population. In an attempt to reduce the incidence of
these injuries, the implementation of a multimodality
intraoperative neuromonitoring approach has been
adopted. The use of SEP and MEP monitoring has
proven to be quite useful in reducing morbidity associated with spinal cord iatrogenic injury. Additionally, it is clear that real-time EMG monitoring in
the form of free-run and stimulus evoked EMG also
can prevent neurological complications. Using freerun and stimulus evoked EMG techniques in the
lumbosacral spine has been shown to significantly
reduce morbidity and increase the accuracy of
pedicle screw placement. These same modalities are
also gaining popularity in the cervical and thoracic
419
spine and hold promise for the same effective application of the technique although further studies need
to be performed in order to adopt consistent threshold
ranges. The utilization of the above-discussed EMG
methods, along with the more traditional evoked
potential techniques, in spinal decompressive and
instrumented procedures offers the surgeon a comprehensive intraoperative assay of the entire neural
axis. Moreover, this neuromonitoring approach to
the spinal cord and spinal nerve roots has clearly
resulted in decreased surgical morbidity and better
outcomes in this patient population.
References
Aebi, M, Etter, C, Kehl, T and Thalgott, J (1987)
Stabilization of the lower thoracic and lumbar spine
with the internal spinal skeletal fixation system. Indications, techniques, and first results of treatment. Spine,
12: 544551.
Albee, FA (1911) Transplantation of a portion of the tibia
into the spine for Potts disease. JAMA, 57: 885886.
Aminoff, MJ, Goodin, DS, Barbaro, NM, Weinstein, PR
and Rosenblum, ML (1985) y somatosensory evoked
potentials in unilateral lumbosacral radiculopathy. Ann.
Neurol., 17: 171176.
Anderson, DG, Wierzbowski, LR, Schwartz, DM, Hilibrand, AS, Vaccaro, AR and Albert, TJ (2002) Pedicle
screws with high electrical resistance: a potential source
of error with stimulus-evoked EMG. Spine, 27:
15771581.
Balzer, JR, Rose, RD, Welch, WC and Sclabassi, RJ (1998)
Simultaneous somatosensory evoked potential and
electromyographic recording during lumbosacral
decompression and instrumentation. Neurosurgery, 42:
13181324.
Beatty, RM, McGuire, P, Moroney, JM and Holladay, FP
recording during spinal surgery. J. Neurosurg., 82:
401405.
Benecke, JE, Jr., Calder, HB and Chadwick, G (1987)
Facial nerve monitoring during acoustic neuroma
removal. Laryngoscope, 97: 697700.
Blumenthal, S and Gill, K (1993) Complications of the
Wiltse pedicle screw fixation system. Spine, 18:
18671871.
Bose, B, Wierzbowski, LR and Sestokas, AK (2002) Neurophysiologic monitoring of spinal nerve root function
during instrumented posterior lumbar surgery. Spine,
27: 14441450.
Bose, B, Sestokas, AK and Schwartz, DM (2004) Neurophysiological monitoring of spinal cord function during
instrumented anterior cervical fusion. Spine J., 4:
202207.
420
Calancie, B, Lebwohl, N, Madsen, P and Klose, KJ (1992)
Intraoperative evoked EMG monitoring in an animal
model: a new technique for evaluating pedicle screw
placement. Spine, 17: 12291235.
Calancie, B, Madsen, P and Lebwohl, N (1994) Stimulusevoked EMG monitoring during transpedicular lumbosacral spine instrumentation: initial clinical results.
Spine, 19: 27802786.
Castro, WHM, Halm, H, Jerosch, J, Malms, J, Steinbeck, J
and Blasius, S (1996) Accuracy of pedicle screw placement in lumbar vertebrae. Spine, 21: 13201324.
Cinotti, G, Gumina, S, Ripani, M and Postacchini, F (1999)
Pedicle instrumentation in the thoracic spine. A morphometric and cadaveric study for placement of screws.
Spine, 24: 114119.
Clements, DH, Morledge, DE, Martin, WE and Betz, RR
(1996) Evoked and spontaneous electromyography to
evaluate lumbosacral pedicle screw placement. Spine,
21: 600604.
Danesh-Clough, T, Taylor, P, Hodgson, B and Walton, M
(2001) The use of evoked EMG in detecting misplaced
thoracolumbar pedicle screws. Spine, 26: 13131316.
Darden, BV, II, Owen, JH, Hatley, MK, Kostuik, J and
Tooke, SM (1998) A comparison of impedance and
electromyogram measurements in detecting the presence of pedicle wall breakthrough. Spine, 23:
256262.
Delgado, TE, Buchheit, WA, Rosenholtz, HR and Chrissian, S
(1979) Intraoperative monitoring of facial muscle evoked
responses obtained by intracranial stimulation of the facial
nerve: a more accurate technique for facial nerve dissection. Neurosurgery, 4: 418421.
Dickens, JRE and Graham, SS (1991) A comparison of
facial nerve monitoring systems in cerebellopontine
angle surgery. Am. J. Otol., 12: 16.
Djurasovic, M, Dimar, JR, II, Glassman, SD, Edmonds, HL
and Carreon, LY (2005) A prospective analysis of intraoperative electromyographic monitoring of posterior cervical screw fixation. J. Spinal Disord. Tech., 18: 515518.
Ebraheim, NA, Xu, R, Stanescu, S and Yeasting, RA
(1999) Anatomic relationship of the cervical nerves to
the lateral masses. Am. J. Orthop., 28: 3942.
Epstein, N (2001) Anterior approaches to cervical spondylosis and ossification of the posterior longitudinal ligament: review of operative technique and assessment of
65 multilevel circumferential procedures. Surg. Neurol.,
55: 313324.
Esses, S, Sachs, B and Dreyzin, V (1993) Complications
associated with the technique of pedicle screw fixation.
Spine, 18: 22312239.
Fan, D, Schwartz, DM, Vaccaro, AR, Hilibrand, AS and
Albert, TJ (2002) Intraoperative neurophysiologic
detection of iatrogenic C5 nerve root injury during laminectomy for cervical compression myelopathy. Spine,
27: 24992502.
J.R. BALZER ET AL.

Farber, GL, Place, HM, Mazur, RA, Jones, DEC and
Damiano, TR (1995) Accuracy of pedicle screw placement in lumbar fusions by plain radiographs and computed tomography. Spine, 20: 14941499.
Ferrick, MR, Kowalski, JM and Simmons, ED (1997) Reliability of roentgenogram evaluation of pedicle screw
position. Spine, 22: 12491253.
Gertzbein, SD and Robbins, SD (1990) Accuracy of pedicle
screw placement in vivo. Spine, 15: 1114.
Girardi, FP, Cammisa, FP, Jr., Sandhu, HS and Alvarez, L
(1999) The placement of lumbar pedicle screws using
computerized stereotactic guidance. J. Bone Joint Surg.
Br., 81: 825829.
Glassman, SD, Dimar, JR, Puno, RM, Johnson, JR, Shields,
CB and Linden, RD (1995) A prospective analysis of
intraoperative electromyographic monitoring of pedicle
screw placement with computed tomographic scan confirmation. Spine, 20: 13751379.
Gundanna, M, Eskenazi, M, Bendo, J, Spivak, J and
Moskovich, R (2003) Somatosensory evoked potential
monitoring of lumbar pedicle screw placement for in situ
posterior spinal fusion. Spine J., 3: 370376.
recordings in spinal surgery: correlation of clinical and
electrophysiologic findings in a prospective, consecutive series of 213 cases. Spine, 29: 677684.
Guo, L, Quinones-Hinojosa, A, Yingling, CD and Weinstein, PR
(2006) Continuous EMG recordings and intraoperative
electrical stimulation for identification and protection
of cervical nerve roots during foraminal tumor surgery.
J. Spinal Disord. Tech., 19: 3742.
Hadra, BE (1891) Wiring the spinous processes in Potts diseases. Trans. Am. Orthop. Assoc., 4: 206210.
Harner, SG, Daube, JR and Ebersold, MJ (1987) Improved
preservation of facial nerve function with the use of
electrical monitoring during removal of acoustic neuromas. Mayo Clin. Proc., 62: 92102.
Herron, LD, Trippi, AC and Gonyeau, M (1987) Intraoperative use of dermatomal somatosensory-evoked potentials
in lumbar stenosis surgery. Spine, 12: 379383.
Hibbs, RA (1911) An operation for progressive spinal
deformities. N.Y. Med. J., 93: 10131016.
Hilibrand, AS, Schwartz, DM, Sethuraman, V, Vaccaro,
AR and Albert, TJ (2004) Comparison of transcranial
electric motor and somatosensory evoked potential monitoring during cervical spine surgery. J. Bone Joint Surg.
Am., 86(A): 12481253.
Holland, NR (1998) Intraoperative electromyography during
thoracolumbar spinal surgery. Spine, 23: 19151922.
Holland, NR and Kostuik, JP (1997) Continuous electromyographic monitoring to detect nerve root injury during thoracolumbar scoliosis surgery. Spine, 22:
25472550.

Holland, NR, Lukaczyk, TA, Riley, LH III and Kostuik, JP
(1998) Higher electrical stimulus intensities are required
to activate chronically compressed nerve roots. Implications for intraoperative electromyographic pedicle
screw testing. Spine, 23: 224227.
Hormes, JT and Chappuis, JL (1993) Monitoring of lumbosacral nerve roots during spinal instrumentation. Spine,
18: 20592062.
Isley, MR, Pearlman, RC and Wadsworth, J (1997) Recent
advances in intraoperative monitoring of spinal cord
function: pedicle screw stimulation techniques. Am. J.
END Tech., 37: 93126.
Jimenez, JC, Sani, S, Braverman, B, Deutsch, H and
Ratliff, JK (2005) Palsies of the fifth cervical nerve root
after cervical decompression: prevention using continuous intraoperative electromyography monitoring. J. Neurosurg. Spine, 3: 9297.
Jou, IM (2004) The effects from lumbar nerve root transection in rats on spinal somatosensory and motor-evoked
potentials. Spine, 29: 147155.
Kalfas, IH, Kormos, DW, Murphy, MA, McKenzie, RL,
Barnett, GH, Bell, GR, Steiner, CP, Trimble, MB and
Weisenberger, JP (1995) Application of frameless
stereotaxy to pedicle screw fixation of the spine. J. Neurosurg., 83: 641647.
Khan, MH, Smith, PN, Balzer, JR, Crammond, D, Welch,
WC, Gerszten, P, Sclabassi, RJ, Kang, JD and Donaldson,
WF (2006) Intraoperative somatosensory evoked potential
monitoring during cervical spine corpectomy surgery:
experience with 508 cases. Spine, 31: E105E113.
Knoeller, SM and Seifred, C (2000) Historical perspective:
history of spinal surgery. Spine, 25: 28382843.
Krag, MH, Weaver, DL, Beynnon, BD and Haugh, LD
(1988) Morphometry of the thoracic and lumbar spine
related to transpedicular screw placement for surgical
spinal fixation. Spine, 13: 2732.
Krassioukov, AV, Sarjeant, R, Arkia, H and Fehlings, MG
(2004) Multimodality intraoperative monitoring during
complex lumbosacral procedures: indications, techniques, and long-term follow-up review of 61 consecutive
cases. J. Neurosurg. Spine, 1: 243253.
Laine, T, Schlenzka, D, Makitalo, K, Tallroth, K, Nolte, LP
and Visarius, H (1997) Improved accuracy of pedicle
screw insertion with computer-assisted surgery. A prospective clinical trial of 30 patients. Spine, 22: 15541558.
Gelb, DE (1995) Triggered electromyographic threshold
for accuracy of pedicle screw placement: an animal
model and clinical correlation. Spine, 20: 15851591.
Leppanen, RE (2005) Intraoperative monitoring of segmental spinal nerve root function with free-run and
electrically-triggered electromyography and spinal cord
function with reflexes and F-responses. A positions statement by the American Society of Neurophysiological
Monitoring. J. Clin. Monit. Comput., 19: 437461.
421
Lewis, SJ, Lenke, LG, Raynor, B, Long, J, Bridwell, KH

and Padberg, AM (2001) Triggered electromyographic
threshold for accuracy of thoracic pedicle screw placement in a porcine model. Spine, 26: 24852489.
Liljenqvist, UR, Halm, HF and Link, TM (1997) Pedicle
screw instrumentation of the thoracic spine in idiopathic
scoliosis. Spine, 22: 22392245.
Lonstein, J, Denis, F, Perra, J, Pinto, MR, Smith, MD and
Perra, JH (1999) Complications associated with pedicle
screws. J. Bone Joint Surg. Am., 81: 15191528.
Ludwig, SC, Kramer, DL, Balderston, RA, Vaccaro, AR,
Foley, KF and Albert, TJ (2000) Placement of pedicle
screws in the human cadaveric spine: comparative accuracy of three techniques. Spine, 25: 16551657.
MacDonald, DB, Al Zayed, Z, Khoudeir, I and Stigsby, B
(2003) Monitoring scoliosis surgery with combined
multiple pulse transcranial electric motor and cortical
somatosensory-evoked potentials from the lower and
upper extremities. Spine, 28: 194203.
Maguire, J, Wallace, S, Madiga, R, Leppanen, R and
Draper, V (1995) Evaluation of intrapedicular screw
position using intraoperative evoked electromyography.
Spine, 20: 10681074.
Masferrer, R, Gomez, CH, Karahalios, DG and Sonntag,
VK (1998) Efficacy of pedicle screw fixation in the
treatment of spinal instability and failed back surgery:
a 5-year review. J. Neurosurg., 89: 371377.
Merloz, P, Tonetti, J, Pittet, L, Coulomb, M, Levallee, S
and Sautot, P (1998) Pedicle screw placement using
image guided techniques. Clin. Orthop., 354: 3948.
Mller, AR and Jannetta, PJ (1985) Monitoring of facial
nerve function during removal of acoustic tumor. Am.
J. Otol., (Suppl.): 2729.
Mller, AR and Jannetta, PJ (1987) Monitoring facial EMG
responses during microvascular decompression operations for hemifacial spasm. J. Neurosurg., 66: 681685.
Norcross-Nechay, K, Mathew, T, Simmons, JW and Hadjipaviou, A (1999) Intraoperative somatosensory evoked
potential findings in acute and chronic spinal canal
compromise. Spine, 24: 10291033.
Obi, T, Mochizuki, M, Isobe, K, Mizoguchi, K, Takatsu, M
and Nishimura, Y (1999) Mechanically elicited nerve
root discharge: mechanical irritation and waveform.
Acta Neurol. Scand., 100: 185188.
Odgers, CJ, Vaccaro, AR, Pollack, ME and Cotler, JM
(1996) Accuracy of pedicle screw placement with the
assistance of lateral plain film radiography. J. Spinal
Disord., 9: 334338.
Owen, JH, Bridwell, KH and Lenke, LG (1993) Innervation
pattern of dorsal roots and their effects on the specificity of dermatomal somatosensory evoked potentials.
Spine, 18: 748754.
Owen, JH, Kostuik, JP, Gornet, M, Petr, M, Skelly, J, Smoes,
C, Szymanski, J, Townes, J and Wolfe, F (1994) The use
of mechanically elicited electromyograms to protect
422
nerve roots during surgery for spinal degeneration. Spine,
15: 17041710.
Pelosi, L, Lamb, J, Grevitt, M, Mehdian, SM, Webb, JK and
Blumhardt, LD (2002) Combined monitoring of motor
and somatosensory evoked potentials in orthopaedic spinal
surgery. Clin. Neurophysiol., 113: 10821091.
Prass, RL and Luders, H (1986) Acoustic (loudspeaker)
facial electromyographic monitoring: part I. Evoked
electromyographic activity during acoustic neuroma resection. Neurosurgery, 19: 392400.
Raynor, BL, Lenke, LG, Kim, Y, Hanson, DS, WilsonHolden, TJ, Bridwell, KH and Padberg, AM (2002) Can
triggered electromyograph thresholds predict safe thoracic pedicle screw placement? Spine, 27: 20302035.
Reidy, DP, Houlden, D, Nolan, PC, Kim, M and
Finkelstein, JA (2001) Evaluation of electromyographic
monitoring during insertion of thoracic pedicle screws.
J. Bone Joint Surg. Br., 83: 10091014.
Reinhold, M, Magerl, F, Rieger, M and Blauth, M (2007)
Cervical pedicle screw placement: feasibility and accuracy
of two new insertion techniques based on morphometric
data. Eur. Spine J., 16: 4756.
Rose, RD, Balzer, JR, Welch, WC and Jacobs, GB (1997)
Persistently electrified pedicle stimulation instruments
in spinal instrumentation. Technique and protocol
development. Spine, 22: 334343.
Rosen, CD (1991) Letter to the editor. Spine, 16: 599.
Roy-Camille, R, Saillant, G and Mazel, C (1996) Plating of
thoracic, thoracolumbar and lumbar injuries with pedicle screw plates. Orthop. Clin. North Am., 17: 147159.
Sakaura, H, Hosono, N, Mukai, Y, Ishii, T and Yoshikawa,
H (2003) C5 palsy after decompression surgery for cervical myelopathy. Spine, 28: 24472451.
Schwarzenbach, O, Berlemann, U, Jost, B, Visarius, H,
Arm, E, Langlotz, F, Nolte, LP and Ozdoba, C (1997)
Accuracy of computer-assisted pedicle screw placement.
Spine, 22: 452458.
Shi, YB, Binette, M, Martin, WH, Pearson, JM and Hart,
RA (2003) Electrical stimulation for intraoperative
evaluation of thoracic pedicle screw placement. Spine,
28: 595601.
Skelly, JP, Toleikis, JR and Carlvin, AO (1999) Pedicle
screw stimulation in a fluid environment. In: Proceeding of the 10th Annual Meeting of the American Society
of Neurophysiological Monitoring, Denver, CO.
Steinmann, JC, Herkowitz, HN, El-Kommos, H and Wesolowski, DP (1993) Spinal pedicle fixation: confirmation
of an image-based technique for screw placement.
Spine, 18: 18561861.
Toleikis, JR, Carlvin, AO, Shapiro, DE and Schafer, MF
(1993) The use of dermatomal evoked responses during
surgical procedures that use intrapedicular fixation of
the lumbosacral spine. Spine, 18: 24012407.
Toleikis, JR, Skelly, JP, Carlvin, AO, Toleikis, SC, Bernard, TN,
Burkus, JK, Burr, ME, Dorchak, JD, Goldman, MS and
J.R. BALZER ET AL.

Walsh, TR (2000) The usefulness of electrical stimulation
for assessing pedicle screw placements. J. Spinal Disord.,
13: 283289.
Tsai, RY, Yang, RS, Nuwer, MR, Kanim, LE, Delamarter, RB
and Dawson, EG (1997) Intraoperative dermatomal
evoked potential monitoring fails to predict outcome from
lumbar decompression surgery. Spine, 22: 19701975.
Tsai, TM, Tsai, CL, Lin, TS, Lin, CC and Jou, IM (2005)
Value of dermatomal somatosensory evoked potentials in
detecting acute nerve root injury: an experimental study
with special emphasis on stimulus intensity. Spine, 30:
E540E546.
Tsirikos, AI, Aderinto, J, Tucker, SK and Noordeen, HH
(2004) Spinal cord monitoring using intraoperative
somatosensory evoked potentials for spinal trauma.
Vaccaro, A, Rizzolo, S, Balderston, R, Allardyce, TJ,
Garfin, SR, Dolinskas, C and An, HS (1995a) Placement of thoracic screws in the thoracic spine: II. An
anatomical and radiographic assessment. J. Bone Joint
Surg. Am., 77: 12001206.
Vaccaro, A, Rizzolo, S, Allardyce, TJ, Ramsey, M, Salvo,
J, Balderston, R and Cotler, JM (1995b) Placement of
thoracic screws in the thoracic spine: I. Morphometric
analysis of the thoracic vertebrae. J. Bone Joint Surg.
Am., 77: 11931199.
Weinstein, JN, Spratt, KF, Spengler, D, Brick, C and Reid, S
(1988) Spinal pedicle fixation: reliability and validity of
roentgenogram-based assessment and surgical factors
on successful screw placement. Spine, 13: 10121018.
Weiss, DS (2001) Spinal cord and nerve root monitoring
during surgical treatment of lumbar stenosis. Clin.
Orthop. Relat. Res., 384: 82100.
Welch, WC, Rose, RD, Balzer, JR and Jacobs, GB (1997)
Evaluation with evoked and spontaneous electromyography during lumbar instrumentation: a prospective
study. J. Neurosurg., 87: 397402.
West, J, Oglivie, J and Bradford, D (1991) Complications
of the variable screw plate pedicle screw fixation. Spine,
16: 576579.
Whitecloud, T, III, Butler, J, Cohen, J and Candelora, PD
(1989) Complications of the variable spinal plating system. Spine, 14: 472476.
Xu, R, Ebraheim, NA, Ou, Y and Yeasting, RA (1998)
Anatomic considerations of pedicle screw placement
in the thoracic spine. Roy-Camille technique versus
open-lamina technique. Spine, 23: 10651068.
Xu, R, Ebraheim, NA, Shepard, MD and Yeasting, RA
(1999) Thoracic pedicle screw placement guided by
computed tomographic measurements. J. Spinal Disord., 12: 222226.
Young, WF, Morledge, DE, Martin, WH and Park, KB
(1995) Intraoperative stimulation of pedicle screws: a
new method for verification of screw placement. Surg.
Neurol., 44: 544547.

M.R. Nuwer (Ed.)
423
CHAPTER 29
Sacral roots and nerves, and monitoring

for neuro-urologic procedures
David B. Voduseka,* and Vedran Deletisb
a
Division of Neurology, Institute of Clinical Neurophysiology, University Medical Centre, SI 1525 Ljubljana, Slovenia
b
Center for Endovascular Surgery and Intraoperative Neurophysiology, Institute for Neurology and Neurosurgery,
Roosevelt Hospital, New York, NY 10019, USA
29.1. Introduction
The functions involving the genitourinary and the
anorectal systems are uniquely controlled by the
complex interaction of the vegetative and the somatic
nervous system. Insofar as it is the sacral parasympathetic and somatic systems that comprise the most
important peripheral nervous structures controlling
these functions, they may also be referred to as
sacral functions. The functions themselves (micturition, defecation, erection, etc.) can be tested by measuring several different physiological parameters
(urodynamics, anorectal functional testing, measurements of the sexual response), which provide for better diagnosis of dysfunction. The awareness that such
dysfunction may also be a consequence of damage to
neural structures has greatly increased, also because
of the possibility to better define the various lesions
to the nervous system by electrophysiological methods. These methods by and large document, however,
only the function of somatic sacral nervous system
and its central pathways (Vodusek, 2006). Nevertheless, such information is clinically relevant since:
(1) the somatic nervous system plays a part in all
sacral functions, and (2) the somatic and the parasympathetic sacral systems are closely anatomically
related and information on the somatic system may
therefore be a relevant indicator of the overall neurogenic lesion in several clinical situations. Methods to
identify autonomic nerves intraoperatively have also
been developed.
*
Correspondence to: Prof. David B. Vodusek, M.D., Ph.D.,

Division of Neurology, University Medical Center, Faculty
of Medicine, University of Ljubljana, Zaloska 7,
1525 Ljubljana, Slovenia.
Tel.: 386-41673050; fax: 386-15222293.
E-mail: david.vodusek@kclj.si (D.B. Vodusek).
The common denominator of lower urinary tract,

anorectal, and sexual functions is that a large part of
their efferent and afferent neural pathways travel at least
partly in close vicinity. They share common spinal
cord regulatory segments (the upper lumbar segments
sympathetic efferents; the middle and lower sacral
segments parasympathetic and somatic efferents)
(Fig. 1). Even the long pathways connecting the relevant spinal cord segments with higher levels of the central nervous system are situated close together in the
lateral spinal cord. Thus, it is not uncommon that in
lesions to the nervous system and particularly clearly
in lesions affecting the spinal cord, cauda equina, the
sacral plexus, and the pudendal nerve, the sacral functions are affected together. Dysfunctions may arise not
only from disease or trauma, but also from inadvertent
lesions to the above-named structures during invasive
procedures, particularly several surgeries involving the
spinal canal and the pelvic organs. Functional consequences of lesions to sacral functions neurocontrol
include disturbing sensory phenomena like pain, dysesthesiae, urgency, and frequency; loss of genital sensation, sensation of bladder or rectal fullness, as well as
motor deficits: retention, obstipation, soiling, incontinence, and erectile dysfunction. Neurogenic damage
may furthermore cause loss of coordination between
detrusor and sphincter function, leading to high bladder
pressures and upper urinary tract dysfunction. Incontinence may particularly in motor disabled persons
lead to problems with hygiene, skin problems, infections, and decubiti. While neurogenic sexual dysfunction may not be life-threatening, it can be extremely
psychologically disruptive and lead to severe emotional
and interpersonal problems. Each sacral dysfunction, in
short, causes significant deterioration of quality of life.
This is particularly tragic if it arises as a consequence
of an inadvertent intraoperative lesion.
D.B. VODUSEK AND V. DELETIS
424
Fig. 1. Anatomical drawings of pudendal nerve and vegetative fibers in the prostatic plexus.
A whole array of clinical neurophysiological diagnostic methods has been modified for use in the anogenital area, including electromyographic (EMG)
methods, reflex, conduction, and evoked potential
studies (Vodusek et al., 2005; Vodusek, 2006). These
methods are routinely employed in uro-neurological/
neuro-urological laboratories for diagnostics and
follow-up in patients with (suspected) neurogenic
sacral dysfunction. In addition, research has been conducted to establish some of these neurophysiological
methods in the operating room, to help the surgeon
identify particular sacral nervous structures, and/or
monitor the function of the sacral neuromuscular system during surgery (Deletis et al., 1992; Deletis and
Vodusek, 1997; Huang et al., 1997; Rodi and Vodusek,
2001). In addition, monitoring pressure responses of
pelvic structures and penile circumference (Pang and
Casey, 1983; Klotz and Herschorn, 1998; Schaan
et al., 2004) on intraoperative electrical stimulation
has also been proposed as helpful procedures to identify
relevant neural tissue to be preserved during surgery.
29.2. Clinical neurophysiological tests
in diagnostics
Function of pelvic organs relies on neural control, and
clinical neurophysiological tests have been introduced
to support and supplement clinical evaluation in
patients with suspected neurogenic dysfunction. The
tests comprise electrophysiologic methods of testing

conduction through motor and sensory pathways (both
peripheral and central) and EMG methods. Traditionally, in testing both the lower urinary tract and anorectal
function, the EMG signal obtained from sphincter muscles has been used to delineate the sphincter activity
patterns in relationship to micturition of defecation.
In addition, EMG methods have been used to distinguish between normal and neuropathic sphincter (pelvic
floor) muscles. Conduction tests have been introduced
to evaluate the integrity of different reflex pathways
(sacral reflexes), the individual motor (pudendal nerve
terminal latency, MEP) and sensory pathways (penile
sensory neurography, SEP). Autonomic tests have also
been introduced (sympathetic skin response, corpus
cavernosum EMG the latter is still controversial as
to its validity).
In the diagnostic situation, results of electrophysiologic testing have to be compared to normative
values values obtained from healthy subjects. Individual tests have a certain specificity and sensitivity
(differing for individual diagnostic categories). The tests
of conduction have been found helpful in diagnosing
demyelinization, but to be relatively insensitive to axonal lesions. Amplitudes of responses vary widely in the
control population (particularly due to technical reasons), and are therefore not sensitive parameters of an
axonal lesion in patients; conduction may remain normal
in partial axonal lesions, and is thus not a specific
parameter. Thus, in the diagnostic situation, the ability

of the concentric needle EMG to detect abnormal spontaneous activity as an indicator of denervation, and
changes of motor unit potentials indicating reinnervation, has been found to be particularly helpful for diagnosing neuronopathic and axonal lower motor neuron
lesions. EMG and recording the bulbocavernosus reflex
(indicating the patency of the lower sacral reflex arc)
have been proposed as the basic battery of tests for evaluation of patients with sacral dysfunctions and suspected neurogenic involvement (Vodusek et al., 2005).
From conduction tests, only recording the sacral reflex
and the SEP to dorsal penile/clitoral nerve stimulation
have been suggested since they have been validated by
extensive clinical studies. They may be of value in
selected patients with suspected peripheral (i.e., bulbocavernosus reflex testing) and central nervous system
(i.e., SEP testing) lesions (Lundberg et al., 2000;
Vodusek, 2006). The other neurophysiological tests
have been suggested as useful in research. The corpus
cavernosum EMG is the most controversial of the tests
so far described. It is not yet well clarified whether the
signal really originates from penile smooth muscle; validation of the method would offer a most important
source of information on penile innervation status
necessary for erection.
Clinical neurophysiological methods have also been
taken out of the clinical neurophysiological laboratory to help with some particular interventions related
to the sacral neuromuscular system. These applications
of clinical neurophysiological methods may not yet be
widely spread or fully established. They, however,
show promise in at least three particular areas: botulinum injection therapy, application of electrical stimulation devices for treatment of sacral dysfunction, and
intraoperative monitoring (Vodusek, 2001).
29.3. Intraoperative clinical neurophysiology
With appropriate modifications of methods, it is technically feasible to record in anesthetized subjects
dorsal root action potentials to pudendal nerve stimulation, pudendal somatosensory evoked potentials
over the conus, the spinal cord and over the scalp,
sphincter muscle EMG responses to sacral ventral root
stimulation and motor cortex stimulation, the bulbocavernosus reflex, and also some other parameters such
as intravesical, intraurethral, intrarectal, and intracavernosal pressures, and the penile circumference (see
Rodi and Vodusek, Chapter 53 in this volume) (Fig. 2).
These techniques have been proposed as useful in
the surgical theater for two different purposes: for
425
intraoperative identification of relevant neural structures (mapping), and/or for (continuous) monitoring of functional integrity of particular neural
structures during surgery.
29.3.1. Intraoperative identification of neural tissue
(mapping)
Identification of relevant sacral structures by neurophysiological techniques intraoperatively started with
EMG recordings from the anal sphincter in patients
with spinal dysraphism, in whom unidentified tissue
was electrically stimulated and if inducing response
in the anal sphincter preserved. Thus, it was reported
that such use of neurophysiologic techniques allowed
spinal operations in 10 patients without changes in
their neurological or urological function (James
et al., 1979). Intraoperative urodynamic measurements
of intravesical pressure have been proposed as useful
indicators of parasympathetic fibers within the cauda
equina (identified by bladder pressure rises after electrical stimulation). Thus, in patients with tethered cord
syndrome and cauda equina tumors, bladder function
could be preserved by identification of relevant neural
tissue (Schaan et al., 2004). The same objective can be
achieved by measuring anal sphincter pressure
response (Pang and Casey, 1983).
Nerve sparing modifications trying to preserve
autonomic nerves were furthermore introduced into
various pelvic surgeries thus preserving innervation of
pelvic organs and their function. It is well known that
(particularly radical) surgery for uterine, rectal, and
prostate pathology causes a very high percentage of
lower urinary tract, anorectal, and sexual dysfunction.
Some nerve sparing techniques rely on purely anatomical approaches (Sakuragi et al., 2005). In urology, it has
been claimed that outcomes of prostate surgery regarding sexual function should be improved by intraoperative identification of cavernosal nerves by electrical
stimulation. As proof of cavernosal nerve depolarization, visual assessment of penile tumescence was introduced first (Lue et al., 1995), then pressure recordings
made from corpora cavernosa (Michl et al., 1999; Rehman et al., 1999), then penile girth measurement (Klotz
and Herschorn, 1998). Intraoperative cavernous nerve
stimulation with penile tumescence monitoring during
radical prostatectomy was shown to help preserve erectile function with no associated adverse events (Klotz
et al., 2003) (Fig. 3).
The added usefulness of intraoperative identification
of nerves by electrical stimulation and physiological
monitoring has been pointed out also for cystectomies
426
AFFERENT EVENTS
EFFERENT EVENTS
Pudendal SEPs
(Traveling Waves)
1.
Stimulation
1 V
100 ms
Anal M-Wave
4.
Pudendal DRAP
100 V
5 ms
2.
Anal MEP
50 V
4 ms
Pudendal SEPs
(Stationary Wave)
100 V
20 ms
5.
BCR
3.
6.
10 V
10 ms
70 V
50 ms
Fig. 2. Neurophysiological events used to intraoperatively monitor the sacral nervous system. To the left are afferents
events after stimulation of the dorsal penile/clitoral nerves and recording over the spinal cord: (1) Pudendal SEPs, traveling
waves, (2) pudendal DRAP, and (3) pudendal SEPs, stationary waves recorded over the conus. To the right are efferent
events: (4) anal M-wave recorded from the anal sphincter after stimulation of the S1S3 ventral roots, (5) anal MEPs
recorded from the anal sphincter after transcranial electrical stimulation of the motor cortex, and (6) bulbocavernosus reflex
obtained from the anal sphincter muscle after electrical stimulation of the dorsal penile/clitoral nerves. [Reprinted with permission from Deletis et al. (2001).]
Fig. 3. (Left) Schematic anatomical positional drawings of dorsal nerve of penis and clitoris and plexus cavernosus penis in
the relationship with the prostate. (Middle) Stimulating electrodes are placed on both sides of prostate after endopelvic fascia is opened with a tumescence sensor placed at the base of penis. (Right) The drawing at A shows how the surgeon
would normally visualize or estimate the location of the nerve to be near the 5 and 7 o0 clock positions on the prostate.
He would normally remove the tissue above the expected nerve location. In the drawing on the right (B), using the
CaverMap Surgical Aid, the surgeon discovered that the nerve was close to the 3:30 and 8:30 positions. This discovery
caused the surgeon to change his dissection and consequently preserve the nerve as measured by stimulation at the end of
the surgery. [Modified from Klotz et al. (2003) with permission from Blue Torch Medical Technologies, Inc.]
(Spitz et al., 1999), in gynecological surgery (Possover

et al., 2004), and in rectal cancer resections (Hanna
et al., 2002). The mentioned approaches of refined identification of autonomic nerves were suggested as inherently rational for improved functional outcomes
(particularly regarding erection), but practical outcomes
in patients as reported by different authors vary significantly (cf. Klotz, 2004). Sensitivity of a marketed
device (CaverMap) to neurophysiologically identify
(map) cavernosal nerves was reported to be 88%, and
specificity 54% (Walsh et al., 2001), and thus concerns
are raised as to the specificity of the tumescence
response.
So far, no method has been introduced to continuously monitor functional integrity of the autonomic
nerves.
29.3.2. Intraoperative identification of pudendal
afferents in sacral roots (Fig. 4)
During the past two decades, increasing numbers of
neurosurgeons have included sacral roots in the rhizotomy procedure for treatment of spasticity of lower
R2
R2
()
Stim
(+)
R1
R1
()
Stim (+)
R1
Fig. 4. To the lower left, position of the stimulating electrodes over the clitoris and labia majora for the stimulation
of the dorsal clitoral nerves. To the lower right, position of
the stimulating electrode for stimulating the dorsal penile
nerves. R recording BCR from anal sphincter. To the
upper right, schematics of recording DRAPs with a hand
held hook electrode over the exposed dorsal sacral roots
of the cauda equina. To the upper left, intraoperative picture of DRAP recordings. [Reprinted from Vodusek and
Deletis (2002) with permission from Elsevier.]
427
extremities. S1 roots have always been a potential

target for rhizotomy, but the S2 dorsal roots have
been mostly excluded from the procedure for fear
of inducing lower urinary tract and sexual dysfunction. Sparing the S2 roots from rhizotomy, however,
may allow the remaining abnormal reflexive circuits to cause muscle spasticity in lower extremities
(plantar spasticity). Thus, it has been demonstrated that children who underwent L2S2 rhizotomies had an 81% greater reduction in plantar
flexors spasticity compared to children who underwent only L2S1 rhizotomies (Lang et al., 1994).
As more sacral roots are included in rhizotomies,
the potential risk for postoperative complications
related to sacral dysfunction, however, increases.
The reported incidence of bladder dysfunction after
selective dorsal rhizotomies may be up to 24%, and
there seems to be a tendency among neurosurgeons
to underreport the occurrence of such complications
(McDonald, 1991). Because the median age of
patients undergoing rhizotomy is typically between
3 and 4 years, not all symptoms are effectively communicated. Possible (consecutive) sexual complications cannot be evaluated at all. Techniques to
protect afferent fibers from the anogenital region
and the pelvic organs should decrease complications
of rhizotomies. As direct electrical stimulation of
bladder and bowel afferents is technically difficult,
stimulation of the cutaneous branches of the pudendal nerve was proposed as a practical alternative.
Intraoperative pudendal afferent mapping of the
sacral roots was successfully introduced (Deletis
et al., 1992, 2001). If any sensory action potential
on dorsal penile/clitoral or anal mucosa stimulation
was recorded on a root identified as causing spasticity, the root was subdivided and the action potentials
on each rootlet were tested again. The rootlets with
significant pudendal afferent activity were spared
from rhizotomy (Deletis et al., 1992). A report on
114 children with debilitating spastic cerebral palsy
revealed successful afferent mapping performed in
105 children undergoing selective dorsal rhizotomies.
Although some contribution of pudendal afferent
activity was frequently recorded from S1 roots, it
was never the major component. On the contrary,
S2 roots carried more than 2/3 of pudendal afferent
fibers in 50% of the patients. Overall 32% of the
S2 rootlets were cut in children who underwent
pudendal nerve mapping, as opposed to 65% being
cut in children who were operated on before the
introduction of this technique. Five percent of the
428
patients had difficulty voiding during the immediate

postoperative period, mostly complaining of dysuria.
All of the children began voiding spontaneously
within 12 weeks after surgery. Thus, none of the
105 patients who underwent successful sacral root
mapping developed long-term bowel or bladder complications (Huang et al., 1997). This compares favorably to the 24% incidence of transient bladder
dysfunction with one patient left with permanent disability requiring intermittent catheterization in children operated before introduction of the mapping
procedure (Abbott, 1992). The results also suggest
that the root distribution of afferent fibers which are
important for the control of voluntary micturition
may be similar to the distribution of mucocutaneous
afferent fibers from the pudendal nerve. Furthermore,
preserved sensation of genitals is proposed to be
important for normal sexual development and for
the normal sexual response. In conclusion, selective
S2 rhizotomy proved to be a safe procedure without
an associated increase in residual spasticity while
preserving bowel and bladder function by performing
pudendal afferent mapping (Lang et al., 1994).
It is suggested that the mapping of pudendal afferents in the dorsal roots should be employed whenever
these roots are considered for rhizotomy in children
with cerebral palsy without urinary retention. Preoperative neuro-urological investigation of the children
should help in making appropriate decisions, as in children with cerebral palsy with hyperreflexive detrusor
dysfunction, sacral rhizotomy may be considered to
alleviate the problem. In any case, intraoperative
Patient: W.S. - 4 y.o.
Dg.: Cerebral palsy
Right
mapping of sacral afferents should make selective

surgical approaches possible (Figs. 5 and 6).
For other neurosurgical procedures in the lumbosacral spinal canal (e.g., the release of a tethered cord
or the removal of a tumor), other authors have proposed, and in small series have performed, the identification of motor and sensory nerve roots. This has
been achieved through continuous monitoring of
EMG activity in muscles innervated by lumbosacral
segments, and monitoring of tibial nerve somatosensory evoked potentials (Kothbauer et al., 1994).
Kothbauer et al. (1994) claim that intraoperative
recordings both saved operating time by allowing
the surgeon more rapid and decisive preparation than
would be possible on an anatomical basis alone, and
gave the impression that the procedures were safer.
29.3.3. Intraoperative monitoring of sacral
nervous structures
Several neurophysiological techniques have been
introduced to provide intraoperative monitoring of
selected sacral structures, for instance, root stimulation and recording of the anal sphincter pressure
(Pang and Casey, 1983), or recordings of neurotonic
discharges in response to nerve irritation (Kothbauer
et al., 1994). It is questionable how specific is increment in the anal pressure as a result of stimulation of
the motor fibers (roots) of the pudendal nerves. Stimulation of the gluteal and other adjacent muscles can
give false-positive response. The bulbocavernosus
reflex response (BCR) is reflecting the integrity of
Patient: M.G. - 4 y.o.
Dg.: Cerebral palsy
Left
S1
S2
S2
S3
S
Left
Right
S1
S3
40 V
S
5 ms
40 V
S
S
5 ms
Fig. 5. Example of extreme asymmetry of pudendal afferents from the pudendal nerves. To the right, all afferents from
right and left penile nerves in this 4-year-old cerebral palsy patient enter to the spinal cord via a single left S2 root. Each
trace is an average of 100 responses twice recorded to show reproducibility. To the left, example of extreme asymmetry of
pudendal afferents from anal mucosa. All afferents from anal mucosa in this 4-year-old cerebral palsy patient enter to the
spinal cord via a single left S3 root. Each trace is an average of 100 responses twice recorded to show reproducibility.
[Reproduced from Deletis et al. (2001) with permission.]
429
Mapping of pundendal afferents
Right
Left
Right
Right
Left
Left
S1
S1
S1
S2
S2
S2
S3
S3
S
40 V
5 ms
C
Patient: L.,.M., 6y
Dg: Cerebral palsy
Patient : R.S., 4y
Dg: Cerebral palsy
Left
Right
Left
Right
S3
S
S
Patient: Z.M., 4y
Dg: Cerebral palsy
Left
Right
S1
S1
S1
S2
S2
S2
S3
S
Patient: M.G., 4y
Dg: Cerebral palsy
Patient: L.,M., 3y
Dg: Cerebral palsy
Patient: W.A., 4 y
Dg: Cerebral palsy
S3
S3
S
Fig. 6. Six characteristic examples of DRAP showing the entry of a variety of pudendal nerve fibers to the spinal cord via S1S3
sacral roots. A: Symmetrical distribution of DRAPs confined to one level (S2) or three levels (D). Asymmetrical distribution of
DRAPS confined to the side (B), only one root (C or F), or all roots except right S1 (E). Recordings were obtained after electrical
stimulation of bilateral penile/clitoral nerves. [Reprinted from Vodusek and Deletis (2002) with permission from Elsevier.]
the whole lower sacral reflex arc, and it seems an

ideal test for intraoperative monitoring in surgeries
putting cauda equina, conus medullaris, the sacral
plexus, and the pudendal nerves at risk (Fig. 7).
It is also a practical intraoperative technique because
both stimulation and recording are performed outside

the surgical field. The feasibility of intraoperative
recording of BCR was demonstrated in a group of
119 patients, both adults and children, both males
and females (Deletis and Vodusek, 1997). The
Iso = 0% 11:58
Iso = 0.8%
Iso = 1.25%
17:51
N20 = ON 17:52
17:53
Relax. = ON
10:41
17:54
Iso = off
17:55
10:42
17:56
12 : 54
75 V
10 ms
200 V
20 ms
10:43
50 V
20 ms
Fig. 7. To the left, the influence of isoflurane (Iso) on the BCR. Note that the response was almost completely blocked
when the concentration of 1.25% was administered for 15 min and did not recover until almost 30 min after the isoflurane
was discontinued (Iso off). To the middle, influence of nitrous oxide (N2O) on the amplitude of BCR. Nitrous oxide (60%)
was administered at 17:52, resulting in a significant decrease in response amplitude within 2 min. To the left, influence of
muscle relaxants on the amplitude of BCR. The response disappeared 1 min after muscle relaxants were administered
(10:41). [Modified from Deletis and Vodusek (1997) with permission from Lippincott, Williams and Wilkins.]
430
technique requires at least some preoperative preservation of the lower sacral reflex arc, which it is
advisable to test before surgery. Monitoring of BCR
is technically easily applicable in males, but has still
some technical problems related to less satisfactory
stimulation technique in females (Rodi and Vodusek,
2001). Intraoperative anesthesiological regime for successful monitoring of the BCR should avoid isoflurane
or similar kinds of inhalatory anesthetics, nitrous
oxide, and muscle relaxants (Fig. 8). It is suggested
to use propofol/fentanyl anesthesia with muscle relaxants for intubation purpose only. It has so far been
applied in more than 250 patients at risk for conus or
cauda equina damage, with no false-positive or falsenegative results (Deletis and Rodi, personal communication). Sala et al. (2000) reported, however, that some
patients who lost BCR during surgery did not have
urinary or bowel complications postoperatively.
Stimulation of the mixed sacral roots (motor and
sensory) within cauda equina (supplying fibers to
the pudendal nerves) can identify the nerve fibers
100 ms
50 V
50 V
Fig. 9. Stimulation of the right mixed sacral root resulting

in two responses (arrows pointed downwards) in the right
anal sphincter muscle; one, early as a result of direct activation of the efferent fibers and late one as a result of activation
of the afferents fibers, corresponding to the bulbocavernosus
reflex.
being depolarized. If a motor root is stimulated selectively, only a short latency muscle response the
compound muscle action potential (CMAP) from the
anal sphincter can be obtained. Stimulation of sensory root results in a reflex response corresponding
to the bulbocavernosus reflex (but with a shorter
latency than the reflex obtained on penis/clitoris!).
Stimulating both the sensory and motor root together
gives a complex response. This is a useful method
to distinguish between motor and sensory roots of
the pudendal nerves (Fig. 9).
Pudendal somatosensory evoked potentials is
another method which has been introduced to intraoperative monitoring, and shown to be easily applicable also under the regime of anesthesia (Cohen et al.,
1991; Vodusek, 1993). It was suggested as useful in
procedures involving the spine or spinal canal below
the S1 level. A large series of 154 patients revealed
one case of a false-positive response, and no falsenegative response (Cohen et al., 1991).
10 min
29.4. Comment
Fig. 8. Continuous monitoring of the BCR for a period of

10 min showing the stability of the BCRs appearance.
[Reprinted from Vodusek and Deletis (2002) with permission from Elsevier.]
Sexual, anorectal, and lower urinary tract functions

depend on neural control, and are paramount to quality of life. In the last decades, surgeons in different
specialities who perform interventions which may
jeopardize these sacral functions have begun to realize that preservation of relevant neural structures is
desirable and possible.
Neurophysiology has to offer the surgeon techniques to accomplish two goals: first, identification
of structures as functional nervous tissue, and their

distinction from other tissue; and second, continuous
monitoring of function of selected relevant nervous
structures. Studies demonstrating the feasibility of
achieving these two goals in the pelvis and for the
sacral nervous system have made surgeons more
sensitive of their roles in preserving sacral functions
(Possover et al., 2004).
In the most basic approach, the surgeon is simply
paying more attention to neuroanatomy in the pelvis,
depending on accumulated knowledge of previous
research demonstrating important neural tissue in
parts of the surgical field which have previously been
ignored and resected (Sakuragi et al., 2005). If a
physiological parameter can be recorded (such as an
electrophysiological response, or intraorgan pressure
change, or penile girth change), intraoperative electrical stimulation can be used to better define the relevant neural structures (Klotz and Herschorn, 1998;
Michl et al., 1999; Possover et al., 2004; Schaan
et al., 2004). The technical requirements and possible
pitfalls of intraoperative recording of electrophysiological responses on one hand, and other physiological responses on the other, differ; it is probably fair
to say that experience and expertise for the former
is widespread, and for the later as yet limited. Therefore, more experience and possibly further technical developments might be required before the
mapping of autonomic nerves will really establish
itself as valid. Furthermore, practical outcomes in
patients have been so far controversially reported
(cf. Walsh et al., 2001; Klotz, 2004). Thus, further
research and experience is needed before the
described methods will be more widely accepted as
possibly useful in prostatectomies (Klotz, 2004),
cystectomies (Spitz et al., 1999), in gynecological
surgery (Possover et al., 2004), and in rectal cancer
resections (Hanna et al., 2002). It also needs to be
stressed that so far no method has been introduced
to continuously monitor functional integrity of
autonomic nerves.
In the case of surgeries endangering the proximal
sacral plexus, cauda equina, and spinal cord, a battery of tests possibly would need to be applied, and
a whole set of structures assessed bilaterally. Both
the lower sacral segments and sphincter muscles,
and also the upper sacral segments and the lumbar
segments would need to be included. Also, all these
segments and several functional modalities would
need to be monitored more or less simultaneously.
The appropriate set-up for each surgical situation
431
would need to be selected on the basis of anatomical

and physiological considerations, certainly seeking a
compromise between the possible and the realistically
necessary.
Intraoperative monitoring of the sacral nervous
system might be particularly helpful in some selected
patient groups. These would include patients in
whom sacral function may be one of the few spared
functions, for instance, patients with advanced
neuromuscular diseases who have heavily compromised motor function of lower limbs but no sacral
dysfunction.
Recording of dorsal root action potentials (to
identify sacral roots carrying genitourinary afferents) has been shown although in a retrospective
comparison to decrease postoperative voiding disturbances (Deletis et al., 1992; Huang et al., 1997).
Similar claims are made for preserving sacral functions for nerve sparing techniques in radical pelvic
surgery (Rehman et al., 1999; Klotz et al., 2003;
Possover et al., 2004). These approaches should
relatively easily be assessed by further research,
since particular types of surgeries are performed in
a large number of patients in single centers and the
percentage of dysfunction after classical surgery
is quite high.
Also, in other instances of monitoring of sacral
nervous system, there is an impression that surgeries
accompanied by such monitoring were easier for the
surgeon and safer for the patient (Schaan et al.,
2004). For the time being, there is anecdotal data
that difficult surgical decisions which relied on
the results of intraoperative neurophysiological
measurements did not result in unexpected neurological deficits. Although the techniques described
clearly provide the surgeon with additional information about nerve location and function, it will
be important in the future that the value of these
techniques be further carefully defined and documented. As suggested by Daube (1991), it will be
necessary to demonstrate whether these techniques
save operative time, anesthesia time, and improve
outcomes. Importantly, the outcomes will have to
include quality of life measures. To demonstrate
the value of intraoperative physiological testing to
identify and preserve sacral nervous tissue, randomized studies matching patients undergoing surgery
with and without such monitoring techniques will
have to be performed. This may be difficult for
different reasons, not least because for those using
the techniques currently, they seem to be intuitively
432
and obviously contributing to the quality of patient

care, and omitting them would mean unnecessarily
omitting relevant information.
Intraoperative electrophysiological stimulation
and recordings have been introduced to preserve
sacral functions, but as a byproduct relevant
basic knowledge has been gathered as well. To give
an example, systematic recordings of penile/clitoral
afferent activity in S1S2 and S3 roots bilaterally
(and in some children also S4 and S5 roots) provided
new neuroanatomical data. Most subjects showed
evidence of pudendal afferents in S2 and S3 roots
bilaterally; about half of them also showed evidence
of some afferent activity in S1 (either unilaterally
or bilaterally). In slightly more than half of the
subjects, the root potentials were symmetrical, but
the pudendal afferents of many were irregularly
distributed across the sacral roots. In a few of the
children, the afferent activity was confined to a
single root (either S2 or S3 root) (Huang et al.,
1997) (Fig. 5). Similar findings that all afferents
from anal mucosa have been conveyed by single
sacral root have been reported by Deletis et al.
(2001). The finding of asymmetrical distribution of
fibers which may be confined to a single root is consistent with the finding of Junemann et al. (1987)
who found that the majority of motor fibers for the
urethral sphincter may be carried by a single variable lower sacral root. Therefore, for some patients
undergoing surgery in the area of cauda equina,
the sacrifice of one single root may be of dire functional consequences. These data (and other: cf.
Hamdy et al., 1999; Turnbull et al., 1999) indicate
that unilateral versus bilateral monitoring of sacral
neural structures will have to be one of the points
clarified by further research.
29.5. Conclusion
There is growing awareness that surgically induced
lesions to nervous structures are avoidable with the
aid of intraoperative monitoring of their functional
integrity. The lost functional integrity of sacral nervous structures may result in disorders of micturition,
defecation, and sexual function. Electrophysiological
and other functional diagnostic methods as applied to
the sacral neuromuscular system are used for identification of relevant neural structures, and for intraoperative monitoring of the integrity of nervous
system. Further developments are expected, as the
awareness of the significance of sacral neurocontrol
is increasing. It has to be recognized that the abovementioned techniques have so far not been used
enough to gather high level proof of the value of
sacral nervous system monitoring during surgical
interventions. On account of experience so far, it
is proposed that electrophysiological techniques
are valid and valuable safeguards against inadvertent
lesions of those nervous structures, whose lesioning
would necessarily lead to some (neurogenic) dysfunction of micturition, defecation, or the sexual
response. Further studies are required to clarify these
issues.
References
Abbott, R (1992) Complications with selective posterior
rhizotomy. Pediatr. Neurosurg., 18(1): 4347.
Cohen, BA, Major, MR, et al. (1991) Pudendal nerve
evoked potential monitoring in procedures involving
low sacral fixation. Spine, 16(Suppl. 8): S375S378.
Daube, JR (1991) Intraoperative monitoring of cranial nerves.
In: J Schramm and AR Mller (Eds.), Intraoperative
Neurophysiologic Monitoring in Neurosurgery. Springer
Verlag, Berlin, pp. 246267.
Deletis, V and Vodusek, DB (1997) Intraoperative recording of the bulbocavernosus reflex. Neurosurgery,
40(1): 8892; discussion 9293.
Deletis, V, Vodusek, DB, et al. (1992) Intraoperative monitoring of the dorsal sacral roots: minimizing the risk of
iatrogenic micturition disorders. Neurosurgery, 30(1):
7275.
Deletis, V, Krzan, M, et al. (2001) Functional anatomical
asymmetry of pudendal nerve sensory fibers. In: HP
Bruch, F Kockeling, R Bouchard and C Schug-Pab
(Eds.), New Aspects of High Technology in Medicine.
Monduzzi Editore, Bologna, pp. 153158.
Hamdy, S, Enck, P, et al. (1999) Laterality effects of
human pudendal nerve stimulation on corticoanal pathways: evidence for functional asymmetry. Gut, 45(1):
5863.
Hanna, NN, Guillem, J, et al. (2002) Intraoperative parasympathetic nerve stimulation with tumescence monitoring during total mesorectal excision for rectal
cancer. J. Am. Coll. Surg., 195(4): 506512.
Huang, JC, Deletis, V, et al. (1997) Preservation of pudendal afferents in sacral rhizotomies. Neurosurgery, 41(2):
411415.
James, HE, Mulcahy, JJ, et al. (1979) Use of anal sphincter
electromyography during operations on the conus
medullaris and sacral nerve roots. Neurosurgery, 4(6):
521523.
Junemann, KP, Schmidt, RA, et al. (1987) Neuroanatomy
and clinical significance of the external urethral sphincter. Urol. Int., 42(2): 132136.

Klotz, L (2004) Cavernosal nerve mapping: current data
and applications. BJU Int., 93(1): 913.
Klotz, L and Herschorn, S (1998) Early experience with
intraoperative cavernous nerve stimulation with penile
tumescence monitoring to improve nerve sparing during
radical prostatectomy. Urology, 52(4): 537542.
Klotz, L, Jewett, M, et al. (2003) A randomized phase 3 study
of intra-operative cavernous nerve stimulation with penile
tumescence monitoring (CaverMap Surgical Aid) to
improve nerve sparing during radical prostatectomy
http://www.bluetorchmed.com/surgeons/index.html.
Kothbauer, K, Schmid, UD, et al. (1994) Intraoperative
motor and sensory monitoring of the cauda equina. Neurosurgery, 34(4): 702707; discussion 707.
Lang, FF, Deletis, V, et al. (1994) Inclusion of the S2
dorsal rootlets in functional posterior rhizotomy for
spasticity in children with cerebral palsy. Neurosurgery,
34(5): 847853; discussion 853.
Lue, TF, Gleason, CA, et al. (1995) Intraoperative electrostimulation of the cavernous nerve: technique, results
and limitations. J. Urol., 154(4): 14261428.
Lundberg, PO, Brackett, NL, et al. (2000) Neurological disorders: erectile and ejaculatory dysfunction. In: A Jardin,
G Wagner, S Khoury, et al. (Eds.), Erectile Dysfunction.
Health Publication, Plymouth, UK, pp. 593645.
McDonald, C (1991) Selective dorsal rhizotomy. Phys.
Med. Rehabil. Clin. N. Am., 2: 891915.
Michl, U, Dietz, R, et al. (1999) Is intraoperative electrostimulation of erectile nerves possible? J. Urol.,
162(5): 16101613.
Pang, D and Casey, K (1983) Use of an anal sphincter pressure monitor during operations on the sacral spinal cord
and nerve roots. Neurosurgery, 13(5): 562568.
Possover, M, Rhiem, K, et al. (2004) The laparoscopic
neuro-navigation LANN: from a functional cartography of the pelvic autonomous neurosystem to a
new field of laparoscopic surgery. Minim. Invasive
Ther. Allied Technol., 13(5/6): 362367.
Rehman, J, Christ, GJ, et al. (1999) Intraoperative electrical stimulation of cavernosal nerves with monitoring of intracorporeal pressure in patients undergoing
nerve sparing radical prostatectomy. BJU Int., 84(3):
305310.
433
Rodi, Z and Vodusek, DB (2001) Intraoperative monitoring

of the bulbocavernosus reflex: the method and its problems. Clin. Neurophysiol., 112(5): 879883.
Sakuragi, N, Todo, Y, et al. (2005) A systematic nervesparing radical hysterectomy technique in invasive cervical cancer for preserving postsurgical bladder function.
Int. J. Gynecol. Cancer, 15(2): 389397.
Sala, F, Chang, D, et al. (2000) Reliability of neurophysiological monitoring of the lumbosacral nervous system during
tethered cord release. Childs Nerv. Syst., 116(6): 374.
Schaan, M, Boszczyk, B, et al. (2004) Intraoperative urodynamics in spinal cord surgery: a study of feasibility.
Eur. Spine J., 13(1): 3943.
Spitz, A, Stein, JP, et al. (1999) Orthotopic urinary diversion
with preservation of erectile and ejaculatory function
in men requiring radical cystectomy for nonurothelial
malignancy: a new technique. J. Urol., 161(6): 17611764.
Turnbull, GK, Hamdy, S, et al. (1999) The cortical topography of human anorectal musculature. Gastroenterology, 117(1): 3239.
Vodusek, DB (1993) Intraoperative bulbocavernosus reflex
monitoring: decreasing the risk of postoperative sacral
dysfunction. Neurourol. Urodyn., 12: 425427.
Vodusek, DB (2001) Interventional neurophysiology of the
sacral nervous system. Neurophysiol. Clin., 31(4):
239246.
Vodusek, DB (2006) Pelvic floor conduction studies. In:
J Kimura (Ed.), Peripheral Nerve Diseases. Elsevier,
Edinburgh, London, pp. 295310.
Vodusek, DB and Deletis, V (2002) Intraoperative neurophysiologial monitoring of the sacral nervous system.
In: V Deletis and JL Shils (Eds.), Neurophysiology in
Neurosurgery. A Modern Intraoperative Approach.
Academic Press, San Diego, CA, pp. 197217.
Vodusek, DB, Amarenco, G, et al. (2004) Committee 8: clinical neurophysiology. In: 3rd International Consultation
on Incontinence, June 2629, 2004. Health Publication
Ltd., Paris.
Walsh, PC, Marschke, P, et al. (2001) Efficacy of firstgeneration Cavermap to verify location and function
of cavernous nerves during radical prostatectomy: a
multi-institutional evaluation by experienced surgeons.
Urology, 57(3): 491494.

M.R. Nuwer (Ed.)
434
CHAPTER 30
Motor evoked potentials from external anal and

urethral sphincter muscles by transcranial cortical
stimulation during surgery
Siavash S. Haghighi*
Department of Clinical Neurodiagnostic, Sharp Memorial Hospital, San Diego, CA 92123, USA
30.1. Introduction
Bowel function is maintained through voluntary contraction of anorectal musculature, which is under
descending control from the cerebral cortex (Turnbull
et al., 1999). Cortical mapping studies demonstrate
the anal response to be bilaterally represented on
the superior motor cortex of both cerebral hemispheres
(Penfield and Rasmussen, 1957; Woolsey et al., 1979).
The external anal sphincter (EAS) muscle differs from
the muscles in the limbs both physiologically and anatomically. There is a lesser degree of voluntary control
over the EAS muscle than over other skeletal muscles
(Turnbull et al., 1999). The EAS and the external
urethral sphincter (EUS) are driven from lower motor
neurons residing in the pudendal nerve. The pudendal
nerve arises from the 2nd, 3rd, and 4th sacral nerves
(James et al., 1979).
The technique of the repetitive transcranial electrical stimulation (rTES) has made it possible to examine
descending cortico-motoneuronal pathways at different segments (Dong et al., 2002; Pelosi et al., 2002;
Sala et al., 2002). Recordings of the motor evoked
potentials (MEPs) have been reported at different neuroaxial levels and various muscles (Burke and Hicks,
1998; Di Lazzaro et al., 1999; Inoue et al., 2002). Only
a few reports have been published concerning
responses in the EAS and EUS muscles (Ertekin
et al., 1990; Inoue et al., 2002; Brostrom et al., 2003;
Haghighi and Zhang, 2004). Both EAS and EUS muscles are small in size, with unique innervations.
Following electro-cortical stimulation, a strong MEP

response from surrounding muscles such as gluteus
can potentially contaminate MEPs from EAS or EUS
muscles. Therefore, proper placement of electrodes is
essential for optimal myographic recordings (Brostrom et al., 2003). This chapter is designed to specifically address stimulation and recording techniques
and validate the use of the ring electrodes in rTES studies of the EUS. This technical report is useful since
complex spinal instrumentation can potentially endanger sacral nerve roots leading to bowel and bladder
dysfunction after surgery (Inoue et al., 2002).
30.2. Technical considerations
30.2.1. Anesthesia consideration
Total intravenous anesthesia is induced using propofol,
fentanyl, and a nondepolarizing muscle relaxant. The
patient is initially paralyzed with a single injection of
atracurium besylate to facilitate intubation. Anesthesia
is maintained with sufentanil (0.20.4 mg/kg/h), propofol (2,6-diisopropylphenol; Diprivan, Zeneca Ltd.,
Macclesfield, Cheshire, UK), and 5060% N2O in O2.
Propofol infusion does not exceed 150 mg/kg/min.
Vital parameters such as blood pressure, end-tidal
CO2, and temperature are continuously monitored.
End-tidal CO2 is maintained at 3.74.2 kPa and body
temperature is maintained around 37 C with a warming air blanket and heating mattress. No volatile inhalation agents are used at any time during the surgery. No
muscle relaxation is present at the time of MEP testing.
Correspondence to: Siavash S. Haghighi, D.V.M., Ph.D,

DABNM, Department of Clinical Neurodiagnostic, Sharp
Memorial Hospital, 7901 Frost Street, San Diego, CA
92123, USA.
Tel.: +1-858-541-3459; fax: +1-858-541-3466.
E-mail: Siavash.haghighi@sharp.com (S.S. Haghighi).
30.2.2. Cortical stimulation technique

During cortical stimulation, anode and cathode are
placed over the excitable tissue and the induced
current passes between these two electrodes. The
impedance of the tissue underlying the stimulating

electrode determines the path of electrical current
flow. The electrical cortical stimulation is either constant voltage or constant current and is delivered in
the form of a square-wave pulse. For brain stimulation, we used the Digitimer D185 (Isolated Multipulse stimulator, Digitimer Ltd., Welwyn Garden
City, UK) which uses constant voltage square wave
with rise time of 10 ms and duration of 50 ms. This
stimulator is capable of delivering electrical pulses
up to 1,200 V with switches for selecting the number
of pulses and interstimulus interval (ISI). The Digitimer D185 is triggered via an evoked potential system. The rTES is performed using two corkscrew
electrodes, which are applied to C4 (anode) and C3
(cathode) locations (International 10-20 system of
electrode placement). Dislodgement of these stimulating electrodes is rare and they do not interfere with
the anesthesiologist access to the head (Burke and
Hicks, 1998). Transcranial stimulation is delivered
with a train-of-five square-wave pulses (50 ms time
constant) while maintaining the ISI at 1 ms
(1,000 Hz). The lowest electrical intensity of the
rTES producing a robust muscle response is defined
as the threshold stimulation. The neuromonitoring is
performed throughout the entire surgical procedure
up to the closure time. During MEP testing, strong
contraction occurs in all muscle groups including
head and neck musculature. Before initiating cortical
stimulation, we recommend inserting an oral airway
or bite block to prevent tongue and tooth damage.
The Cascade multimodality averager (Cascade,
Cadwell Labs, Kennewick, WA) is used to record
MEPs. The machine is set at bandpass of 10 Hz
3 kHz with a sweep time of 150 ms. Amplification
is at 50100 mV per division for muscle groups.
30.2.3. Recording technique
30.2.3.1. Electrodes
For determining the EAS and tibialis anterior (TA)
responses, we used two disposable subdermal stainless steel electroencephalography (EEG) needles
(Viasys Healthcare, Madison, WI) placed subcutaneously into the muscle. To record EMG activity
from the EUS, a custom-made urethral ring electrode
(NeuroSupplies Inc., Waterford, CT) is used. The
ring electrode is carefully pulled over a size 14 Foley
catheter. Position of the ring is 23 cm below the balloon for females, and 25 cm below the balloon for
males. The mounted electrode should be placed on
435
a sterile sheet before using. For disinfection, follow

the recommendations and instructions of the disinfectant manufacturer. Ground electrode is placed on the
shoulder region.
30.2.3.2. Recording location
The muscle action potentials are routinely recorded
from the EAS, EUS, and TA muscles. The TA
response serves as the control recording to insure
motor cortex activation for at least two different muscle groups. The EAS electrodes are placed after
patient is positioned prone on the operating table.
The location of these electrodes are at 3/9 or 12/6
points relative to anal opening. The electrodes are
secured using adhesive tape and adequate separation
of the leads by placing a rolled sponge in between
the electrodes. Recording impedances is kept below
5 kO. Timebase is set at 150 ms, and the bandpass
filters are set at 10 and 3,000 Hz. The latency and
amplitude of evoked muscle responses are measured
using auto cursors. The onset latency is defined as
the time between the stimulus artifact and the beginning of the muscle response. Amplitude is defined as
the largest peak-to-peak deflection of the muscle
response (tallest peak to the lowest trough). The
amplitude and onset latency are calculated as the
mean of three sequential stimulation responses per
target muscle.
30.2.4. Normative data
Table 1 depicts the age, gender, and diagnosis of our
patients at the time of admission. All subjects were
patients undergoing spine surgery. None had neoplastic disease or previous surgery on any intrapelvic
organ. All had normal bowel and bladder function
at the time of study and no patient reported a history
of neurological or other illnesses. An informed consent
was obtained from each patient prior to the surgery.
Thus far, no neurological squeal occurred during or
after the rTES testing. No scalp burns have been
observed at the stimulation site. Table 2 summarizes
latency and amplitude of the EAS, the EUS, and
the TA muscles. Representative waveforms of these
muscles are demonstrated in Figs. 13. The EAS and
EUS have similar onset latency (Table 2). However,
for EAS and EUS muscles, it is significantly shorter
than TA muscle (p < 0.05). Amplitude for EAS
and EUS muscles are significantly smaller than TA
muscle (p < 0.05). In our patient group (n 23),
the average intensity to evoke an EAS response was
436
S.S. HAGHIGHI
Table 1
Patient data, clinical, and surgical interventions
Number
Age (years)
Sex
Diagnosis
Procedure
Preoperative
myelopathy
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
60
18
53
76
42
51
42
49
23
32
35
70
19
58
62
41
45
19
45
21
68
49
61
M
M
M
M
M
M
M
M
F
M
F
M
F
F
M
M
M
F
F
F
M
M
M
Spinal stenosis
Thoracic bust fracture
Spinal instability
Spinal instability
Cervical myelopathy
Cervical stenosis
Lipomeningocele
Tethered cord
Thoracic kyphosis
Spinal instability
Spinal instability
Spinal lipoma
Spinal deformity
Cervical instability
Lumbar instability
Lumbar instability
Spinal stenosis
Thoracic scoliosis
Spinal stenosis
Spinal deformity
Herniated cervical disc
Herniated cervical disc
Spinal instability
Decompression
Decompression
Decompression and fusion
Instrumentation
ACDF
ACDF
Decompression
Tethered release
Instrumentation
Instrumentation
Instrumentation
Decompression
Spinal instrumentation
Occipito-cervical fusion
Decompression
Instrumentation
Decompression
Instrumentation
ACDF
ACDF
Instrumentation
N
Y
N
N
Y
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
ACDF, anterior cervical discectomy and fusion. All patients had normal bowel and bladder function at the time of study and none reported
any history of neurological or other illnesses.
789 78 V compared with 831 11 V for EUS

muscle (range 6501,100 V). The TA muscle latency
was stable with good trial-to-trial reproducibility in
all cases while the latencies for EAS and EUS were
not always clear. We have observed intraindividual
variability in amplitude at each stimulation trial with
fluctuation up to 40% between responses.
30.3. Discussion
The preliminary data show the possibility of activation of the EAS and EUS muscle by transcranial cortical stimulation under anesthesia. We utilized the
C3C4 electrode montage for cortical stimulation.
This electrode placement appears to be more effective than CzFz location similar to findings by Inoue
et al. (2002). They postulated that the direction of
current flow from C3 to C4 was more effective to
activate the cortical representation of the anal region
located in the medial aspect of the superior motor
cortex. The EAS and EUS latency in our subjects

was similar to the previous studies (Ertekin et al.,
1990; Inoue et al., 2002). According to a study by
Ertekin et al. (1990), the latency of the EAS was longer compared to the distal limb muscles. They postulated that since EAS motoneurons were smaller than
the usual motoneurons and therefore peripheral conduction velocity was slower in accordance with the
size principle. The longer latencies of the responses
in the EAS may therefore be explained by a slower
transmission through the central nervous system. In
our study, the EAS latency was around 19 ms. This
latency is significantly shorter than the distal limb
muscles. It has been reported that motor latency after
lumbo-sacral root stimulation to the EAS is much
shorter than distal limb muscles (Herdmann et al.,
1991; Loening-Baucke et al., 1994). Thus, short
peripheral conduction time for the EAS and EUS muscles could contribute to producing such short motor
latency. In the TA muscle, the primary motor response
437
(23)LTA-LTA
(24)LTA-LTA
(7)ANA-ANA
LTA-LTA
BL(20)LTA-LTA
(8)ANA-ANA
(Live)ANA-ANA
(23)RTA-RTA
(24)RTA-RTA
RTA-RTA
(23)ANA-ANA
(24)ANA-ANA
ANA-ANA
10 ms/Div
100 V/Div
15 ms/Div
Fig. 1. Simultaneous recordings of the muscle responses

from tibialis anterior (TA), and external anal sphincter
(EAS) muscles after repetitive transcranial electrical stimulation (rTES) of the motor cortex in a patient during spine
surgery. The cortical stimulation was delivered using corkscrew electrodes at C3/C4 montage (International 10-20
electrode system). Recording electrodes were conventional
EEG needle electrodes placed precutaneously into the EAS
and TA muscles. This patient was under general total intravenous anesthesia with propofol and narcotics infusion, and
nitrous oxide.
Fig. 2. The external anal sphincter (EAS) muscle response

to the repetitive transcranial electrical stimulation (rTES)
of the motor cortex. The cortical stimulation was delivered
using corkscrew electrodes at C3/C4 montage (International 10-20 electrode system). Recording electrodes were
conventional EEG needle electrodes placed precutaneously
into the EAS muscles.
N 45
(1)URl-URl
N 45
(1)URl-URl
37
corresponds to a fast pyramidal tract D-wave, and is

probably transmitted monosynaptically to the motoneuron (Ertekin et al., 1990). They also reported a later
and more obscure response with a latency of 100 ms.
Ertekin et al. (1990) recorded late EAS responses at the
low strength transcranial cortical stimulation. The
excitability threshold for these long latency responses
was surprisingly low. Results indicated that the EAS
motoneurons are easier to excite than other motoneurons, probably because they were under a continuous
facilitatory influence at the spinal level. The long
latency responses may be transmitted by polysynaptic
pathways with relays in the brainstem or the spinal
cord, which in turn may be excited directly from the
motor cortex. However, no evidence was given that
cortical neurons may be excited at this low stimulus
strength. The latency of the responses gradually
became shorter with increasing stimulation strength.
According to Hostege and Tan (1987), the neural
P37
20 V/Div
URl-URl
15 ms/Div
Fig. 3. The external urethral sphincter (EUS) muscle

response to the repetitive transcranial electrical stimulation
(rTES) of motor cortex under anesthesia. The cortical stimulation was delivered using corkscrew electrodes at C3/C4
montage (International 10-20 electrode system). A ring
electrode mounted on a Foley catheter and was placed
against the urethral sphincter muscle. This electrode can
be used in men and women.
438
S.S. HAGHIGHI
Table 2
Latency and amplitude of the responses in external anal
sphincter (EAS), external urethral sphincter (EUS), and
tibialis anterior (TA) muscles following transcranial
electrical cortical stimulation under total intravenous
propofol/narcotic anesthesia
Muscle (n)
Latency (ms)
Amplitude (mV)
EAS (13)
EUS (10)
TA (46)
20.2 3.5
19.9 1.8
26.7 3.2
46.1 17.7
41.7 11.6
460.8 245
pathways exerting inhibitory and facilitatory influences on the EAS motoneurons were located in the
same descending tract of the spinal cord. Transcranial
stimulation may provide a reliable method for demonstrating lesions to these pathways.
Our study demonstrated an objective monitoring
method for bowel and bladder function using the
rTES. The EAS and EUS muscles are recordable
with some variability in amplitude. In our experience, no significant shift in the EAS or EUS latency
occurred during surgery with any postoperative
bowel or bladder complications.
The intraoperative use of rTES to evoke myogenic MEPs can only be performed under intravenous-based anesthesia (Calancie et al., 1991;
Hicks et al., 1992). This fact presents a challenge to
anesthesiologists to maintain adequate depth of surgical anesthesia without usage of neuromuscular
blocking agents.
References
Brostrom, S, Jennum, P and Gunar, L (2003) Motor evoked
potentials from the striated urethral sphincter: a comparison of concentric needle and surface electrodes. Neurourol. Urodyn., 22: 123129.
Burke, D and Hicks, RG (1998) Surgical monitoring of
motor pathways. J. Clin. Neurophysiol., 15: 194205.
Calancie, B, Klose, KJ, Bajer, S and Green, BA (1991) Isoflorane-induced attenuation of motor evoked potentials
caused by electrical motor cortex stimulation during
Di Lazzaro, V, Oliviero, A, Profice, P, Ferrara, L, Saturno,
E, Pilato, F and Tonali, P (1999) The diagnostic value
of motor evoked potentials. Clin. Neurophysiol., 110:
12971307.
Dong, CC, MacDonald, DB and Janusz, MT (2002) Intraoperative spinal cord monitoring during descending
thoracic and thoracoabdominal aneurysm surgery. Ann.

Thorac. Surg., 74: S1873S1876.
Ertekin, C, Hansen, MV, Larsson, LE and Sjodahl, R
(1990) Examination of the descending pathway to the
external anal sphincter and pelvic floor muscles by
transcranial cortical stimulation. Electroencephalogr.
Haghighi, SS and Zhang, R (2004) Activation of the external anal and urethral sphincter muscles by repetitive
transcranial cortical stimulation during spine surgery.
J. Clin. Monit. Comput., 18: 14.
Herdmann, J, Bielefeldt, K and Enck, P (1991) Quantification of motor pathways to the pelvic floor in humans.
Am. J. Physiol., 260: G720G723.
Hicks, R, Burke, D, Stephen, J, Woodforth, I and Crawford, M (1992) Corticospinal volleys evoked by electrical stimulation of human motor cortex after withdrawal
of volatile anaesthetics. J. Physiol., 456: 393404.
Hostege, G and Tan, J (1987) Supraspinal control of motoneurons innervating the striated muscles of the pelvic
floor including urethral and anal sphincters in the cat.
Brain, 110: 13231344.
Inoue, S, Kawaguchi, M, Takashi, S, Takashi, S, Kakimoto,
M, Sakamoto, T, Kitaguchi, K, Furuya, H, Morimoto, T
and Sakaki, T (2002) Intraoperative monitoring of myogenic motor evoked potentials from the external anal
sphincter muscle to transcranial electrical stimulation.
Spine, 27: E454E459.
James, HE, Mulcahy, JJ, Walsh, JW and Kaplan, GW
(1979) Use of anal sphincter electromyography during
operations on the conus medullaris and sacral nerve
roots. Neurosurgery, 4: 521523.
Loening-Baucke, V, Read, NW, Yamada, T and Barker,
AT (1994) Evaluation of the motor and sensory components of the pudendal nerve. Electroencephalogr. Clin.
Pelosi, L, Lamb, J, Grevitt, M, Mehdian, SM, Webb, JK
and Blumhardt, LD (2002) Combined monitoring of
motor and somatosensory evoked potentials in orthopaedic spinal surgery. Clin. Neurophysiol., 113:
10821091.
Penfield, W and Rasmussen, T (1957) The Cerebral Cortex
of Man. Macmillan, New York, NY, 209212.
why, when, how? Childs Nerv. Syst., 18: 264287.
Turnbull, GK, Hamdy, S, Aziz, Q, Singh, KD and
Thompson, DG (1999) The cortical topography of
human anorectal musculature. Gastroenterology, 117:
3239.
Woolsey, CN, Erickson, TC and Gilson, WE (1979) Localization in somatic sensory and motor areas of human cerebral
cortex as determined by direct recording of evoked potentials and electrical stimulation. J. Neurosurg., 51: 476506.

M.R. Nuwer (Ed.)
439
CHAPTER 31
Selective dorsal rhizotomy

Charles D. Yingling*
Department of Otolaryngology/Head and Neck Surgery, Stanford School of Medicine and California Neuromonitoring Services,
San Francisco, CA 94127, USA
31.1. Introduction
Cerebral palsy primarily manifests as a disruption of
motor function characterized by increased limb rigidity,
primarily in the lower extremities. Sherrington (1898)
first reported that spasticity could be relieved by segmental dorsal rhizotomy in decerebrate animals, indicating that the underlying mechanism was hyperactive
reflexes due to loss of descending inhibitory inputs.
Foerster (1913) first applied this technique to a series
of spastic patients, severing multiple adjacent sensory
roots, typically L2-S2. He soon modified his technique
to preserve the L3 or L4 root that was shown to be most
associated with quadriceps function by intraoperative
electrical stimulation, thus foreshadowing the later
development of more sophisticated neurophysiological
methods. This report of partial dorsal rhizotomy constituted the entire literature in the field for over 50 years.
The obvious issue is that while such extensive sectioning of dorsal roots clearly interrupted the reflex arc
and reduced spasticity, consistent with Sherringtons
observations, it also produced dense sensory loss.
In an effort to provide relief from spasticity while
preserving at least partial sensory function, Gros
et al. (1967) introduced the technique of partial rhizotomy at each segmental level, sectioning four of
every five rootlets at the level of the dorsal root entry
zone. This technique, while partial, was nonselective
except that the rootlets bearing the largest radicular
arteries were preserved in an effort to prevent conus
ischemia. His patients did well in terms of reduced
spasticity, but a 40% incidence of sensory disturbance and 16% incidence of bladder dysfunction
was noted. Clearly, a more sophisticated method was
*
Correspondence to: Charles D. Yingling, Ph.D., Stanford

School of Medicine and California Neuromonitoring Services, 1001 Bridgeway #434, Sausalito, CA 94965, USA.
Tel.: 1-415-332-6763; fax: 1-415-665-1620.
E-mail: cy@brainmon.com (C.D. Yingling).
needed in order to provide the desired clinical result

while minimizing side effects. More than another
decade passed before modern neurophysiological
techniques were introduced during this procedure.
31.2. Selective dorsal rhizotomy: introduction of
neurophysiological techniques
The first application of modern intraoperative neurophysiological recordings during dorsal rhizotomy
was the seminal paper of Fasano et al. (1979), which
introduced most of the techniques still in use today.
They electrically stimulated lumbar and sacral dorsal
roots or rootlets with two electrodes 1 cm apart, and
recorded electromyographic (EMG) responses from
(unspecified) target muscles. Beginning at 1 Hz, they
first established a threshold for each dorsal root,
and then progressively increased the stimulation
rate from 1 to 50 Hz. They noted three main features
in the responses of patients with cerebral palsy.
First, some circuits showed normal inhibitory activity; low frequency stimulation produced a response
to each stimulus, but when the stimulus frequency
was increased to 3050 Hz, responses were seen
only to the first stimulus in each train. Second, in
most of the circuits tested inhibition was lacking
and a response was seen to each stimulus, even at
higher frequencies; sometimes an increased response
frequency with an interference pattern or afterdischarges were also seen. In addition, roots lacking
inhibition also sometimes exhibited abnormal diffusion, manifested as (a) simultaneous activation of
agonist and antagonist muscles, (b) activation of
muscles innervated from other segmental levels,
or even (c) activation of muscles contralateral to the
side stimulated. Such diffusion to other groups was
never found in roots with normal inhibition. Third,
they found an excess of inhibition in some cases,
demonstrated by failure of responses to some stimuli
even at rates as low as 2 Hz or by lengthening of the
440
refractory period to paired stimuli. When a mixed

response was obtained from stimulation of an entire
root (i.e., inhibition in one muscle and lack of inhibition in another), the root was separated into individual rootlets which were independently tested to
identify the abnormal ones.
Fasano et al. (1979) did not explicitly state their criteria for deciding which roots or rootlets to spare, but in
75% of cases only 2550% of rootlets were cut; in 23%
of cases, less than 25%; and in only 2% of cases were
more than 50% of rootlets cut. Sensory deficits were
minimal and transient, and permanent improvements
in motor function were noted. Of a group of 80 patients
ranging in age from 5 to 25, only 2% had poor results,
27% showed slight improvement, 52% had good
results, and 19% had very good results in reduction of
spasticity. Voluntary motility of lower limbs was
unchanged in 14%, moderately improved in 32%,
markedly improved in 49%, and normal in 5%. Exteroceptive and proprioceptive sensations and sphincter
control were preserved. Interestingly, improvements
were also noted at segmental levels not involved in the
procedure, for example, improvements in skilled
movements of the hand and facilitation of speech and
swallowing ability.
Fasano et al.s technique was soon picked up by
the South African neurosurgeon, Warwick Peacock,
who subsequently moved to California (UCLA and
later UCSF) and has been largely responsible for
popularizing the technique in the USA. In a series
of publications (Peacock and Arens, 1982; other references in Staudt et al., 1995), Peacock and his colleagues extended the observations of Fasano et al.
(1979) and described a larger variety of response patterns, offering specific criteria for deciding which
rootlets should be cut or spared. While other grading
schemes have been proposed, most are in effect
variants of the scheme proposed by Peacock and
colleagues (see Section 31.4 below for more details).
31.3. Surgical approaches
Several different surgical approaches have been
described. Peacock and colleagues, as well as others,
utilized an extended laminectomy or laminotomy
from L2-L5 or S1 (Peacock and Arens, 1982; Harper
and Nelson, 1992). Some groups have extended
this even further, for example, L1-S2 (Mittal et al.,
2001). The advantage of such an extended exposure
is that it facilitates visualization of the entire cauda
equina and identification of segmental levels. On
C.D. YINGLING
the other hand, an extended exposure prolongs the

surgery and may increase the risk of subsequent spinal instability (Peter et al., 1990). As a result, some
groups have utilized a more limited exposure. The
original procedure of Fasano et al. (1979) utilized a
T12-L1 laminectomy to expose the conus medullaris;
a similar approach has been advocated by others
(Barolat, 1991; Park et al., 1993). Conversely,
Lazareff et al. (1999) advocated a limited exposure
from L4 or L5-S1, which obviously also limits the
levels at which rhizotomy can be performed to those
below the rostral extent of the exposure. Lazareff
et al. argued that satisfactory results were obtained
even with selective rhizotomy limited to the L5 and
S1 levels.
A different technique entirely has been advocated
by Sindou et al. (1987). Instead of sectioning the
dorsal roots directly, they performed dorsal root entry
zone (DREZ) lesions after a more rostral laminectomy (T10-L2) for lower extremities or cervical hemilaminectomy (C4C7) for irreducible flexion of the
upper extremity in hemiplegic patients. Since this is
a completely different method than the technique of
selective rhizotomy utilized in the bulk of the literature cited here, it will not be considered further in
this review.
31.4. Neurophysiological techniques and criteria
31.4.1. Peacocks criteria
Since the original report of Fasano et al. (1979), there
have been many attempts to clarify the criteria used
to determine which roots or rootlets should be sectioned. Fasano proposed a straightforward criterion:
there was either a single response to a 50 Hz train
(normal), or an extended response (abnormal). Peacock and colleagues originally followed Fasanos
lead but soon relaxed the definition of normal after
finding that the strict criterion of a single response
was overly simplified and potentially led to section of
more rootlets than necessary to achieve good clinical
outcomes. They identified at least eight different patterns, and gave each descriptive names that have proven
useful in communication between the surgeon and
neurophysiologist: decremental, squared, decrementalsquared, incremental, multiphasic, clonic, irregular,
and sustained (see Fig. 1). The first of these roughly
corresponds to Fasanos normal, although rather
than requiring only a single response to 50 Hz train
stimulation, a decrementing response only requires
Decremental
441
spread to multiple segmental levels or involve bilateral

responses, and would be candidates for sectioning.
31.4.2. Phillips and Parks grading scheme
Squared
Decremental - Squared
Incremental
Multiphasic
Clonic
Irregular
Sustained
Stimulus 1 s, train
Fig. 1. Representative examples of the variety of responses

elicited by 50 Hz train stimulation for 1 s. [Reprinted from
Staudt et al. (1995) with permission from the International
Federation of Clinical Neurophysiology.]
that the response amplitude continues to decline

throughout the duration of the stimulus train. The
squared and decremental-squared categories appear
to represent what Fasano et al. would have termed
lack of inhibition and thus abnormal; however,
these are considered within the normal range of
variation in Peacocks scheme, and would thus be
spared. The last five categories, which involve either
increasing amplitude of the response during the
stimulus train, more complex patterns of alternating
excitation and inhibition, or afterdischarges after the
end of the stimulus, are all termed abnormal by
Peacock and colleagues, as are any responses that
Phillips and Park (1989) proposed a five-level grading scheme to facilitate surgical decision making.
Referring to the response elicited by a 1-s 50 Hz train
to each dorsal rootlet, each rootlet is graded as follows: 0 no sustained discharge; 1 sustained
response only in muscles innervated by the segmental
level stimulated; 2 sustained response as in 1
but also seen in an additional segmental level;
3 sustained response seen in multiple ipsilateral
segmental levels; 4 sustained response with
spread to the contralateral leg (see Fig. 2). Note that
this scheme does not distinguish among different
categories of sustained response, as does Peacocks,
but relies primarily on the anatomical distribution
of the responses. Typically, nearly all rootlets producing 3 or 4 responses are sectioned, although
at least one rootlet may be spared at each level
(Harper and Nelson, 1992).
Mittal et al. (2001) have studied the reliability of
responses to repeated train stimulation, using the
grading scheme of Phillips and Park. They found that
93% of dorsal rootlets had either no change or a
change of only one grade between repeated stimulation runs. They proposed an algorithm for determining which rootlets to section (Fig. 3), in which
whole roots are first tested, and if the response is
grade 0, 1, or 2, the root is not subdivided and
they move on to the next level. If the entire root produces a grade 3 or 4 response, then it is subdivided and each rootlet tested individually. Only the
rootlets that consistently produce a 4 response are
sectioned. Overall, this produced a lesion rate of 51%.
Harper and Nelson (1992) also noted that they
may section rootlets showing only 1 or 2 responses if they innervate muscles contributing to
the patients spasticity, and all rootlets at that level
have a similar grade. They also note that the contraction strength perceived by the electrophysiologist
may also be used to distinguish rootlets of the same
grade. Here, as in the descriptive terms employed
in Peacocks scheme, a certain amount of subjectivity
is always present. This becomes even more apparent
when one attempts to compare the numerical grading
of Phillips and Park with the descriptive terminology
employed by Peacock. For example, is a rootlet producing a strongly incrementing response in only one
442
C.D. YINGLING
Left
Right
Left
Right
quadriceps
hamstrings
gastrocnemius
deltoid or biceps
0.25 mV
0.5 s
Left
Right
Left
Right
Left
Right
Fig. 2. Eight-channel recordings illustrating the five category grading scheme proposed by Phillips and Park (1989). Top
left, unsustained response in R gastrocnemius after stimulation of R S2 dorsal root, assigned grade 0. Top right, sustained
response in R quadriceps after stimulation of R L3 dorsal root, assigned grade 1. Bottom left, sustained response in adjacent segmental level after stimulation of R S2 dorsal root, assigned grade 2. Bottom center, sustained activity spreading to
multiple ipsilateral myotomes after stimulation of L L5 dorsal root, assigned grade 3. Bottom right, augmenting response
spreading to contralateral lower extremity after stimulation of L S1 dorsal root, assigned maximally abnormal grade 4.
[Reprinted from Mittal et al. (2001) with permission from the American Association of Neurological Surgeons.]
Intraoperative Electrophysiological Monitoring (EMG and PT)

Grading of Motor Responses to Stimulation
Roots Assigned Grades 0, 1+, or 2+
Roots Assigned Grades 3+ or 4+
Do Not Divide into Rootlets
Divide into Rootlets
Do Not Section
Stimulation of Rootlets
Rootlets Assigned Grade 3+
Rootlets Assigned Grade 4+
Second Stimulation at Same Current
Section Rootlet
Grade 3+ Response
Grade 4+ Response
Do Not Section
Section Rootlet
Fig. 3. Algorithm proposed by Mittal et al. (2001) for determining which rootlets to section. In this scheme, whole dorsal
roots are first tested and if the whole root response falls in grades 0, 1, or 2, the root is left intact and not subdivided. If
the whole root is graded 3 or 4, then it is subdivided into rootlets which are individually tested. In Mittal et al.s method,
only rootlets consistently showing grade 4 are sectioned. Other groups would also section rootlets exhibiting grade 3;
see text for further details of selection criteria. [Reprinted from Mittal et al. (2001) with permission from the American
Association of Neurological Surgeons.]
muscle a better or worse candidate for section than

one producing a squared response at multiple segmental levels, or bilaterally? Unfortunately, such
questions do not admit of easy experimental
approaches, given that 50 or more rootlets may be
tested in a given procedure and only the clinical outcome is available as a dependent variable.
443
RIGHT
LEFT
S1
S2
S3
40 V
S
5 ms
31.4.3. Preservation of bowel/bladder function

31.4.3.1. Anal sphincter recordings
A major concern during any rhizotomy procedure,
particularly when excessive ankle or toe flexion is
involved and thus sacral roots are particularly targeted, is loss of bowel and/or bladder function. While
ventral roots are always preserved, and thus motor
control of sphincter function should not be disturbed,
interruption of sensory input from bowel or bladder
can also have adverse consequences for control. Two
distinct methods have been utilized in an effort to
minimize or avoid this complication: anal sphincter
recordings and mapping of pudendal afferents.
Peacock and colleagues (Staudt et al., 1995)
utilized EMG recordings from the anal sphincter,
and always spared rootlets at S2 that produced
sphincter activity. At S1, rootlets showing sphincter
responses were usually spared, unless the rootlet
showed strong indications of abnormal activity by
other criteria. Sphincter responses at lumbar levels
were ignored. Ojemann et al. (1997) reported that
sphincter responses were seen in the overwhelming
majority of dorsal roots from L2 and down, and
argued that this lack of specificity rendered intraoperative EMG recordings unreliable in identification
of the critical S3-S5 rootlets which are most important
to preserve. They proposed exposure of the conus
medullaris for positive identification of lower sacral
roots, and also suggested that intraoperative measures of bladder and rectal pressures might be useful.
31.4.3.2. Pudendal afferent mapping
A different approach to this problem has been proposed by Huang et al. (1997). Rather than relying
on reflex responses as measured by EMG, they identified pudendal afferents by electrical stimulation of
the dorsal penile or clitoral nerve, recording dorsal
root action potentials (DRAPs) from each S1-S3 root
with handheld hook electrodes. Rootlets showing significant pudendal afferent activities were spared, and
no S3 rootlets were excised (Fig. 4). They found
that only 44% of patients showed a symmetrical
Fig. 4. Dorsal root action potentials (DRAP) after stimulation of the pudendal nerves in one patient, showing pudendal afferents to be present in only the R S2 and (to a lesser
extent) R S1 roots. This distribution across two sacral
levels was the most common pattern (57% of patients),
but other individuals showed a range of patterns from a single unilateral level to multiple bilateral levels. Rootlets
showing significant pudendal DRAP responses were spared
to avoid bowel, bladder, or sexual dysfunction. [Reprinted
from Huang et al. (1997) with permission from Lippincott,
Williams and Wilkins.]
distribution of pudendal afferents. Overall, S1 roots

contributed 4%, S2 roots 60.5%, and S3 roots
35.5% of the overall pudendal afferent activity. Note,
however, that sensory fibers supplying anal and urethral sphincters may run with other branches of the
pudendal nerve than the one stimulated in this study,
so this technique may not provide an infallible guide
to avoid bowel/bladder dysfunction.
Given the low incidence of pudendal afferents
from levels above S2, it might be argued that limiting
the rhizotomy procedure to L2 through S1 would
obviate the need for either anal sphincter EMG or
pudendal afferent recordings. However, Lang et al.
(1994) followed the results from 85 children, of
whom 13 had only L2-S1 rhizotomy while the
remaining 72 had rhizotomy extending from L2-S2.
They noted that the addition of S2 roots resulted in
an 81% reduction in the number of legs with residual
plantar flexor spasticity. Thus, unless plantar flexor
spasticity is not an issue in a specific patient, the
procedure should include the S2 level and some
method of avoiding section of rootlets contributing
to sphincter function should be employed.
To my knowledge, there have been no studies
directly comparing pudendal afferent recordings with
anal sphincter EMG recordings, so it is not clear
whether the two methods would produce comparable
results and/or identify the same rootlets to be spared.
The method of anal sphincter recording is technically
easier, as it involves the same procedure already in
use to identify rootlets to be sectioned or spared at
444
other levels, with the simple addition of sphincter

electrodes. In contrast, pudendal afferent mapping
involves applying additional stimulating electrodes
during setup, and switching from stimulating to
recording electrodes in the surgical field, with signal
averaging to identify the small amplitude pudendal
afferents. Thus, unless a clear improvement in outcome can be demonstrated with the use of pudendal
afferent recording, using anal sphincter EMG recordings would be the method of choice due to its relative
simplicity. In this case, sphincter activity elicited
by stimulation of lumbar roots would generally be
ignored.
31.4.4. Reliability of neurophysiological criteria
As will be apparent from this review, there has been
considerable variation in specific techniques used by
different groups performing selective dorsal rhizotomy, and not surprisingly, much variation in the
results obtained. Steinbok et al. (1994) compared 32
spastic patients undergoing selective rhizotomy with
five nonspastic control children undergoing lumbar
laminectomy for untethering of the spinal cord. They
found that 99.7% (!) of the stimulated roots showed
sustained responses to a 50 Hz train in the spastic
group, but so did 70% of the stimulated roots in the
nonspastic controls. The controls, however, never
showed incrementing responses, which were seen in
31% of the spastic group, who also exhibited a 73%
incidence of bilateral spread, 47% including upper
extremity muscles as well. They proposed using the
extent of contralateral and suprasegmental spread as
the primary criterion for selection of rootlets for sectioning. Cohen and Webster (1991) similarly found
sustained responses in all rootlets tested in a nonspastic control subject. Both Cohen and Webster (1991)
and Steinbok et al. (1994) stressed the importance
of not defining abnormal rootlets by absolute criteria,
but rather using the patients clinical presentation as
a guide for which levels to emphasize, and then
basing decisions on relative severity rather than
fixed criteria.
Steinbok and Kestle (1996) surveyed 16 centers
using intraoperative electrophysiologic responses to
guide the extent of dorsal root section. There was significant agreement on some technical details (unipolar stimulation <4 cm from the root exit foramen,
separation of each root into three to eight rootlets,
tetanic stimulation at 50 Hz at 100% of threshold,
multiple muscle recording in each limb, and use of
C.D. YINGLING
contralateral spread as an indication of abnormality).

However, there was more variation (<11 of 16
concurrence) in the use of constant voltage versus
constant current, location of cathode versus anode,
definition of threshold, type of recording electrodes,
and relative importance of electrophysiologic versus
clinical findings in determining how much of each
dorsal root to cut.
Given this variation in techniques, it is perhaps not
surprising that questions have been raised concerning
the reliability and thus validity of electrophysiological
criteria. Weiss and Schiff (1993), using a special
clamp designed to maintain a constant pressure
between stimulating electrodes and dorsal roots, found
substantial changes in threshold and/or response patterns with repeated trials over a few minute period.
These results have been challenged by Peacock et al.
(1994), who pointed out numerous technical and artifactual problems with their data. Logigian et al.
(2001) measured H-reflexes elicited by tibial nerve
stimulation during one rhizotomy and two operations
for tethered cord, and found that even gentle retraction
of dorsal roots depressed the H-reflex. They suggested
that coactivation of ventral roots by current spread
during stimulation of dorsal roots could produce nonreflexive responses showing incrementing or clonic
patterns, thus invalidating the underlying physiological concept of hyperactive reflex loops. However,
such coactivation was only seen at relatively high
stimulus intensities (>14 mA). Sacco et al. (2000)
stimulated dorsal roots both proximally (near the
conus) and distally (near the foramina) and argued that
for a true reflex response, the latency should increase
with more distal stimulation. Using this criterion, they
found that only 16% of dorsal roots revealed a reflex
response which could be evaluated for selective rhizotomy. For the remaining 84%, they nonselectively
cut an average of 60% of each root, with good results.
Many of these critical studies were marred by technical concerns, reported on relatively small numbers
of patients, or used overly stringent criteria for normality. In a larger recent study, Mittal et al. (2001)
studied 77 patients undergoing selective rhizotomy
at a single institution, and performed repeated train
stimulation at 24 threshold in a total of 752 roots.
Their results were more encouraging: using the
grading scheme of Phillips and Park (1989), more
than 93% of dorsal roots had either no change or a
difference of only one grade between the two trials.
Furthermore, the vast majority of dorsal roots
assigned a grade of 4 at the first trial maintained
the same maximally abnormal electrophysiological

response during the second stimulation run.
31.5. Conclusions and recommendations
Technical concerns notwithstanding, there is a consensus among most surgeons performing selective
dorsal rhizotomies that neurophysiological monitoring is an integral part of the procedure. The present
author has monitored several hundred such procedures with several different surgeons, including
Warwick Peacock during his years at UCSF. The following procedure is based primarily on this experience, as well as the literature cited above.
The key to successful utilization of intraoperative
monitoring to select rootlets for sectioning is to recognize that individual patients differ greatly in the extent
to which they demonstrate the different EMG patterns
to train stimulation described above. In some patients,
most responses will be in the normal range and the
least normal will be candidates for section. Other
patients will exhibit almost all abnormal responses,
and the least abnormal will be the ones to be left intact.
Finally, some patients will exhibit the complete range
of responses described above, and in these patients
the decisions are more straightforward. In other words,
there is no fixed set of criteria that will work for all
patients. Rather, the criteria must be adjusted in the
light of each individuals characteristic pattern of
responses to train stimulation.
Our typical setup involves recording with twisted
pair needle electrodes from the following muscles
in each leg: quadriceps (vastus medialisvastus lateralis), adductor magnus, tibialis anterior, biceps
femoris, gastrocnemius, and intrinsic toe flexors, as
well as the external anal sphincter. Most of the leads
are placed while the patient is supine, before positioning prone for the procedure. Since flexion at
the knee is a common occurrence with stimulation,
the leads are all brought together at the knee before
being led to the amplifier inputs; this allows free
movement of the leg without dislodging the electrodes. After placement in the prone position, the final
electrodes (i.e., anal sphincter, biceps femoris) are
placed, connected to the amplifiers, and a tap test
is performed by watching free-running EMG while
tapping each electrode in turn to elicit movement
artifacts. This confirms that all electrodes are
connected correctly and that all components of the
recording system, both hardware and software, are
intact from electrode to display.
445
The patient is positioned prone so that the feet are at

the bottom of the operating table, and the neurophysiologist sits at the foot of the table so as to be able to palpate the muscles of the legs. A transparent drape is
used over the legs, so the surgeon can see movements
elicited by stimulation. This drape is tented up to IV
poles placed at each side so the legs are accessible to
the neurophysiologist during the procedure.
General anesthesia can be maintained with any
common agents, as long as no neuromuscular blockade remains by the time stimulation is necessary.
A relatively light plane of surgical anesthesia is
desirable, so as not to overly suppress reflex
responses. Ankle clonus is a good indicator of appropriate anesthetic depth; almost all patients with spasticity secondary to cerebral palsy will exhibit marked
clonus, and its presence under anesthesia indicates
conditions that will be favorable for neurophysiological monitoring. Spontaneous EMG activity should
be quiet, although clonic EMG will of course be seen
when testing the ankle reflexes.
Regardless of the surgical exposure, monitoring
begins in earnest after completion of the laminectomy. We have typically used an exposure from L2
through S2, to allow visualization of the nerves as
they exit the foramina at these levels. The first step
is to confirm the segmental levels by stimulation.
The S1 root is usually the largest, and stimulation
at 1 per second will produce responses that are largest in biceps femoris and gastrocnemius muscles,
although smaller responses may be seen in muscles
innervated from higher and lower levels (i.e., tibialis
anterior or toe flexors). It should be noted that many
patients who are candidates for selective dorsal rhizotomy may exhibit anomalous innervation patterns;
in one of the first such cases I monitored, the gastrocnemius was primarily innervated from L5 on the L
but from S3 on the R!
Stimulation is delivered through specially designed
nerve hooks (Aesculap) which are insulated except for
the distal portion, and connected to the monitoring
equipment through wires which plug into the end of
the handles. Different colored wires are used so that
the polarity of the stimulation can be controlled; the
surgeon holds the stimulating probes so the cathode
is proximal and anode distal, to avoid the possibility
of anodal block when stimulating dorsal roots. (When
testing ventral roots, the stimulators should be
reversed so that the cathode is distal, that is, in the
direction of orthodromic conduction.) Stimulation
thresholds are typically much lower for ventral roots
446
compared to dorsal roots at the same level. Ventral roots

will often exhibit thresholds as low as 0.1 mA or 0.1 V,
while dorsal root thresholds are usually at least three
times higher, and often as high as several volts or milliamperes. Constant voltage stimulation is preferred if
available, since it is less prone to shifting thresholds
due to shunting from cerebrospinal fluid (CSF) or
saline. However, most commercial monitoring systems
have only constant current stimulators; fortunately,
the typical impedance of the electrode/ tissue interface
is on the order of 1,000 O, so 1 V 1 mA by
Ohms law, thus the levels are roughly comparable with
either type of simulator.
Filters are typically set at 301000 Hz (higher frequencies which are appropriate for single-fiber EMG
are not needed for compound muscle action potentials, and a lower cutoff produces less noise in the
OR). A time base of 50100 ms total sweep time will
allow easy visualization of evoked CMAP responses
for threshold establishment. However, depending on
the particular system used, it may be preferable to
use one recording window with a single time base
for both threshold and train stimulation. In this case,
a total time base of 2 s is used. This allows for
200 ms prestimulus activity to be displayed, a 1-s
train, and then 800 ms of posttrain recordings to
check for any afterdischarges or sustained activity.
After confirmation of segmental levels, testing of
dorsal roots begins, usually starting at L2 and proceeding down to S2 on one side before moving to
the other. If enough display channels are available,
it is useful to display EMG from all muscles on both
sides in order to easily detect any bilateral responses.
If channels are limited, then a sampling of contralateral levels can be utilized, combining muscles innervated from different levels to obtain maximum
coverage, for example, quadricepstibialis anterior
(L4 and L5) in one channel and gastrocnemiustoe
flexors (S1 and S2) in another.
At each level, the entire dorsal root is stimulated
first at 1 per second, and the channels responding
are noted. The root is then separated into (typically)
three to eight rootlets, and each one is tested separately. The threshold for each rootlet is first established
with stimulation at 1 per second. Since it is not feasible
to record from all muscles innervated from L2-S3,
movement may be perceived before EMG responses
are noted; the threshold is thus based on a clear EMG
response, typically 200 mV or more in amplitude. It is
important to be able to switch easily between single
stimuli, delivered at 1 per second for threshold testing,
C.D. YINGLING
and 50 Hz trains of 1 s duration so that once a threshold

is established, a train can be delivered at the same level
without significant delay.
It is not uncommon for the dissection of a given
root into its component rootlets to cause some irritation and thus a high level of EMG activity in muscles
innervated by that level. If the ongoing EMG activity
is large enough to interfere with the assessment of
stimulus-evoked activity, then a brief pause and irrigation of the field with warm (body temperature,
not room temperature) saline will usually quiet the
recordings enough to proceed.
After the threshold is determined, a train at 50 Hz
is delivered at the same stimulus level and the resulting EMG activity reported. It is usually helpful
to repeat the train stimulus two or three times to
confirm the replicability of the response. As noted
above, a 2-s window is used during train stimulation
to allow the entire response period to be seen as well
as 200 ms prestimulus and 800 ms poststimulus
activity. The response amplitude is sometimes as
low as 100200 mV, but is often much larger, and
may reach several thousand microvolts, particularly
if an augmenting pattern is seen. It is useful to be
able to change display gains even after a trace has
been collected, in the event of an unusually large or
small response. An initial gain of 1,000 mV per division is typically a good starting place; with lower
gains, a large response will frequently overlap the
channels above and below, making identification of
the responding muscles difficult.
Figures 516 show examples of multichannel
traces recorded from a single patient. This case is
somewhat atypical in that virtually the entire range
of possible responses can be seen from stimulation
of various rootlets. In such cases, the decisions on
which roots to transect versus spare are relatively
straightforward; rootlets eliciting single twitches, a
large twitch followed by constant activity at a lower
level, or a constant level response are spared. In contrast, rootlets exhibiting augmenting, clonic, widely
diffuse, or bilateral responses are generally transected. In many cases, the good rootlets will elicit
a barely noticeable movement to train stimulation,
while stimulation of the bad rootlets will cause
the entire leg to flex vigorously and sometimes even
end up hanging off the OR table. (This is one of the
reasons to have the drapes tented up at the legs and
the neurophysiologist sitting at the feet!)
During this phase of the procedure, it is helpful to
have an additional person available to keep a log of
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 5. Multichannel recordings illustrating a single twitch at the onset of a 1-s 50 Hz train to an R S1 rootlet (0.8 mA,
1000 mV per division). This decrementing pattern would be considered normal by any published criteria. This and all
subsequent figures are from a single subject, and illustrate the variety of response patterns that may be obtained in one individual. In all these figures, the time base is 200 ms per division (2 s total), with the onset of the train delayed 200 ms from
the beginning of the traces.
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 6. Decrementing-squared pattern to stimulation of R S1 rootlet (0.3 mA, 500 mV per division), with an initial twitch
followed by much lower level sustained activity. Despite slight spread to other segmental levels, this is still considered a
normal pattern.
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 7. Mixed pattern after stimulation of R L5 rootlet (0.5 mA, 1,000 mV per division), showing decrementing response in
ipsilateral adductor, squared pattern in tibialis anterior, and a mildly multiphasic pattern in the hamstring. Note that the later
activity never reaches the amplitude of the initial response; this is also considered a normal pattern.
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 8. Mixed squared/clonic pattern with spread to several segmental levels after stimulation of L L5 rootlet (0.5 mA,
5,000 mV per division). With large amplitude activity at multiple levels, this rootlet might be a candidate for sectioning,
depending on whether a large proportion of rootlets exhibited a more severely abnormal pattern.
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 9. Decrementing-squared pattern to stimulation of R L5 rootlet (0.8 mA/200 mV per division) with spread to several
segmental levels, including significant anal sphincter activity. Despite its similarity to the response pattern seen in Fig. 8,
this rootlet would probably be spared due to the low amplitude and significant sphincter activity.
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 10. Weakly multiphasic response to stimulation of R L4 rootlet (1.7 mA, 2,000 mV per division), showing initial augmenting/clonic pattern with burst of late activity following a silent period during the middle of the stimulus train. Since
there is no significant sphincter activity, this rootlet might be a candidate for sectioning, again depending on the pattern
seen in other rootlets.
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 11. Strongly multiphasic response, limited to initial 500 ms of stimulus train (R L3 rootlet, 1.8 mA, 2,000 mV per division). Note significant augmentation of amplitude after the initial response; again, given the lack of sphincter activity, this
rootlet would be a candidate for sectioning.
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 12. Augmenting response with significant spread to multiple segmental levels (R S1 rootlet, 0.4 mA, 2,000 mV per
division). This rootlet would typically be sectioned.
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 13. Late augmenting response, with dramatic increase in amplitude near end of stimulus train (R L4 rootlet, 1.0 mA,
1,000 mV per division). Despite activity seen in sphincter, this rootlet would be sectioned since sphincter activity elicited
from lumbar levels is usually ignored.
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 14. Strongly augmenting response, with spread to multiple segmental levels (R S1 rootlet, 0.7 mA, 1,000 mV per
division). Note initial small and focal response, followed by period of silence before augmentation. This rootlet would
be sectioned, since the response seen in the sphincter is small and delayed and the rootlet is otherwise quite abnormal.
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 15. Extremely augmenting response to stimulation of L S1 rootlet (3.7 mA, 2,000 mV per division). Note that at this
display gain, the small initial response cannot even be seen, and the later response exceeds 10,000 mV. The response is
truncated at 400 ms into the stimulus train because the extreme movement caused the rootlet to lose contact with the stimulating
electrodes. This rootlet was sectioned despite the small sphincter response (<10% of the amplitude seen in other channels).
IPSI
QUAD
IPSI
ADD
IPSI
TIB ANT
IPSI
HAMSTR
IPSI
GASTROC
IPSI
FOOT
CONTRA
QUAD
CONTRA
TIB ANT
CONTRA
GASTROC
SPHINCTER
Fig. 16. Bilateral, augmenting, and diffuse (BAD) response after stimulation of R L5 rootlet (1.9 mA, 2,000 mV per division). Note spread of response to all ipsilateral channels as well as the contralateral gastrocnemius. Again, the response is
cut short after 450 ms, due to excessive patient movement. This abnormal rootlet was sectioned.
the responses, since the neurophysiologist is usually

busy with the recording system. This can be a nurse
or physical therapist, but it is helpful if the person
logging the responses is familiar with both the procedure and the individual patients clinical presentation. We use previously prepared log sheets which
have, for each segmental level on each side, spaces
for up to eight rootlets to have at least three parameters recorded: the threshold, the classification of
the response using a predetermined coding scheme,
and whether the rootlet was preserved or transected.
Our categories (following those developed by Peacock and associates) are as follows: decrementing
(DC), decrementing-squared (DS), squared (S), augmenting (A), clonic (C), multiphasic (M), irregular
(I), afterdischarge or sustained (F), diffuse (D), and
bilateral (B). Note that some roots may exhibit a
combination of two or more of these patterns, for
example, bilateral, augmenting, and diffuse (BAD).
In a case such the one illustrated here, the criteria
for which rootlets to transect versus spare are relatively straightforward. Rootlets exhibiting decrementing, decrementing-squared, or squared patterns
are preserved; the rest are transected. The exception
to this is that rootlets at sacral levels showing significant sphincter activity are usually spared; sphincter
activity elicited by stimulation at lumbar levels is
ignored. However, many if not most cases do not
present such clear-cut distinctions. Many patients
will exhibit very few responses in the three normal
categories, so the problem becomes determining
the worst of the worst. Other patients will exhibit
few strikingly abnormal responses, so the problem
becomes determining the worst of the best. In other
words, given the variability among individuals, absolute criteria are not appropriate and the clinical
decisions must be made based on a sliding criterion
that takes the range of variation in an individual
patient into account.
In general, we aim for transection of 4050% of
individual rootlets. Several other factors may enter
into the decision process. For example, if an individuals clinical presentation is primarily scissoring,
then more emphasis would be placed on identifying
the worst lumbar roots, which innervate the adductors.
On the other hand, individuals who present primarily
as toe walkers would have more emphasis placed on
sacral roots innervating the gastrocnemius muscles.
Unless otherwise indicated, we tend to be more conservative at L4 and S1, to minimize possible effects on
knee extension and sphincter function, and will only
453
sacrifice rootlets at S2 if there is a clearly abnormal

pattern with no evidence of sphincter responses.
Given the overlapping sensory distribution of
dermatomes, it is theoretically possible to sacrifice
an entire root without producing a dense sensory loss
as long as the roots one level rostrally and caudally
are relatively preserved. In practice, this extreme is
rarely performed, and typically at least 2030% of
each root is left intact in an effort to minimize sensory
loss. Finally, intraoperative assessment of ankle clonus
is usually an immediate indicator of the effectiveness
of the procedure; typically, clonus will be dramatically
reduced or absent after selective rhizotomy on one
side, but still present on the unoperated side. Persistence of clonic reflexes on the operated side after unilateral rhizotomy suggests that the remaining rootlets
should be retested to determine if there are more
potential candidates for transection.
In summary, intraoperative testing of individual
rootlets is an appropriate and integral part of the procedure of selective dorsal rhizotomy. Successful
application of this technique demands a high degree
of cooperation between the surgeon, anesthesiologist,
and neurophysiologist, and making clinical decisions
based on the overall context of an individuals pathophysiology, rather than relying on rigidly fixed criteria. Such a flexible approach will help achieve the
goal of reducing spasticity without causing significant sensory or motor deficits.
Acknowledgment
I would like to thank Peter P. Sun, MD, Childrens
Hospital, Oakland, CA, for the data represented in
Figures 516.
References
Barolat, G (1991) Dorsal selective rhizotomy through a
limited exposure of the cauda equina at L-1.
J. Neurosurg., 75: 804807.
Cohen, AR and Webster, HC (1991) How selective is selective posterior rhizotomy? Surg. Neurol., 35: 267272.
Fasano, VA, Barolat-Romana, G, Zeme, S and Sguazzi, A
(1979) Electrophysiological assessment of spinal circuits in spasticity by direct dorsal root stimulation. Neurosurgery, 4: 146151.
Foerster, O (1913) On the indications and results of the
excision of posterior spinal nerve roots in men. Surg.
Gynecol. Obstet., 16: 463475.
Gros, C, Ouaknine, G, Vlahovitch, B and Frerebeau, P
(1967) La radicotomie selective posterieure dans le
454
traitements neuro-chirurgical de l0 hypertoie pyramidale.
Neurochirurgie, 13: 505518.
Harper, CM and Nelson, KR (1992) Intraoperative electrophysiological monitoring in children. J. Clin. Neurophysiol., 9: 342356.
Huang, JC, Deletis, V, Vodusek, D and Abbott, R (1997)
Preservation of pudendal afferents in sacral rhizotomies.
Lang, FF, Deletis, V, Cohen, H, Velasquez, L and Abbott,
R (1994) Inclusion of the S2 dorsal rootlets functional
posterior rhizotomy for spasticity in children with cerebral palsy. Neurosurgery, 34: 847853.
Lazareff, JA, Garcia-Mendez, MA, De Rosa, R and Olmstead,
C (1999) Limited (L4-S1, L5-S1) selective dorsal rhizotomy for reducing spasticity in cerebral palsy. Acta Neurochir. (Wien), 141: 743752.
Logigian, EL, Soriano, SG, Herrmann, DN and Madsen, JR
(2001) Gentle dorsal root retraction and dissection can
cause areflexia: implications for intraoperative monitoring during selective partial dorsal rhizotomy. Muscle
Nerve, 24: 13521358.
Mittal, S, Farmer, JP, Poulin, C and Silver, K (2001) Reliability of intraoperative electrophysiological monitoring
in selective posterior rhizotomy. J. Neurosurg., 95:
6775.
Ojemann, JG, Park, TS, Komanetsky, BS, Day, RAA and
Kaufman, BA (1997) Lack of specificity in electrophysiological identification of lower sacral roots during
selective dorsal rhizotomy. J. Neurosurg., 86: 2833.
Park, TS, Gaffney, PE, Kaufman, BA and Molleston, MC
(1993) Selective lumbosacral dorsal rhizotomy immediately caudal to the conus medullaris for cerebral palsy
spasticity. Neurosurgery, 33: 929934.
Peacock, WJ and Arens, LJ (1982) Selective posterior rhizotomy for the relief of spasticity in cerebral palsy. Sa
Mediese Tydskrif Deel (South Afr. Med. J.), 62:
119124.
C.D. YINGLING
Peacock, WJ, Nuwer, MR and Staudt, LA (1994) Dorsal
rhizotomy: to monitor or not to monitor? J. Neurosurg.,
80: 769772.
Peter, JC, Hoffman, EB, Arens, LJ and Peacock, WJ (1990)
Incidence of spinal deformity in children after multiple
level laminectomy for selective posterior rhizotomy.
Childs Nerv. Syst., 6: 3032.
Phillips, LH and Park, TS (1989) Electrophysiological
studies of selective posterior rhizotomy patients. In:
TS Park, LH Phillips and WJ Peacock (Eds.), Management of Spasticity in Cerebral Palsy and Spinal Cord
Injury. Neurosurgery State of the Art Reviews. Hanley
& Belfus, Philadelphia. Vol. 4, pp. 459469.
Sacco, DJ, Tylkowski, CM and Warf, BC (2000) Nonselective partial dorsal rhizotomy: a clinical experience
with 1-year follow-up. Pediatr. Neurosurg., 32: 114118.
Sherrington, CS (1898) Decerebrate rigidity and reflex
coordination of movements. J. Physiol., 22: 319322.
Sindou, M, Mifsud, JJ, Rosati, C and Boisson, D (1987)
Microsurgical selective posterior rhizotomy in the
dorsal root entry zone for treatment of limb spasticity.
Acta Neurochir. Suppl. (Wien), 39: 99102.
Staudt, LA, Nuwer, MR and Peacock, WJ (1995) Intraoperative monitoring during selective posterior rhizotomy:
technique and patient outcome. Electroencephalogr.
Steinbok, P and Kestle, JR (1996) Variation between centers in electrophysiologic techniques used in lumbosacral selective dorsal rhizotomy for spastic cerebral
palsy. Pediatr. Neurosurg., 25: 233239.
Steinbok, P, Keyes, R, Langill, L and Cochrane, DD (1994)
The validity of electrophysiological criteria used
in selective functional posterior rhizotomy for treatment of spastic cerebral palsy. J. Neurosurg., 81:
354361.
Weiss, IP and Schiff, SJ (1993) Reflex variability in selective dorsal rhizotomy. J. Neurosurg., 79: 346353.

M.R. Nuwer (Ed.)
455
CHAPTER 32
Nerve root assessment with SEP and MEP

Brian A. Cruma,* and Jeffrey A. Strommenb
a
32.1. Introduction
The assessment of nerve root integrity is an important
function of intraoperative neurophysiologic monitoring (IOM), especially in the evaluation and reconstruction of brachial plexus injuries. Whether a particular
nerve root is functionally connected to the spinal cord
is vital information in the surgical decision-making
process. The presence or absence of this functional
connection will often dictate between different, often
quite disparate, procedures. To make this determination, somatosensory and motor evoked potentials are
performed. With close attention to potential technical
problems, this electrophysiologic information can be
quickly and accurately obtained.
32.2. Techniques
32.2.1. Somatosensory evoked potentials
Responses from stimulation of peripheral nerve can be
recorded from the spinal cord and cerebral cortex as
somatosensory evoked potentials (SEPs). Although
stimulation of a peripheral nerve will depolarize both
motor and sensory axons, selective SEP recordings
ensure assessment of only the large fiber/dorsal column
pathway. As the goal for intraoperative SEPs in most
cases is to assess nerve root integrity, stimulation is
given as close to the intervertebral foramen as possible
(Fig. 1). The cathode is directed proximally. Recording
is from the cervical spine level via either a nasopharyngeal electrode or a needle electrode placed directly on
the lamina in the lower cervical spine. Scalp EEG
Correspondence to: Brian A. Crum, M.D., Department of

Neurology, W8A, Mayo Clinic College of Medicine,
200 First Street SW, Rochester, MN 55905, USA.
Tel.: 1-507-538-7391; fax: 1-507-284-4795.
E-mail: crum.brian@mayo.edu (B.A. Crum).
electrodes are placed at C30 and C40 with Fz as a reference (International 10-20 system). As these responses
are very small in amplitude, many stimuli must be averaged, typically at least 2050, stimulating at 1.11.9 Hz.
It is very important to ensure that the individual holding
the stimulator on the nerve maintain their position until
the stimulation is finished. Stimulus intensity is typically between 10 and 20 mA. The sweep speed is set at
23 ms per division with a sensitivity of 25 mV per
division. Filters are 3 kHz for the high frequency and
30 Hz for the low frequency (Fig. 2A and B).
32.2.2. Motor evoked potentials

Motor responses from transcranial electrical stimulation
(motor evoked potentials, MEPs) can be recorded peripherally at or just distal to the intervertebral foramen at the
same location as stimulation for SEPs (Fig. 2). Anodal
stimulation is given with a short duration (0.05 ms), rapid
rise time stimulus using subcutaneously placed EEG
electrodes at C3 and C4. Several (25) stimuli with an
interstimulus interval of 1 ms are given with intensity
of 200600 V (MultiPulse Cortical Stimulator D185,
Digitmer, Ltd). Direct nerve recording is done with bipolar hook electrodes, usually placed onto the spinal nerve
as close to the exit from the intervertebral foramen as
possible. The sweep speed is 5 ms per division and sensitivity 10 mV per division. Filter settings are 10 kHz for
the high frequency and 2 Hz for the low frequency. An
example of our set-up and protocol for using MEPs during brachial plexus reconstructive surgery is given in
Table 1. Again, good communication with the surgeon holding the recording electrodes on the nerve
is important to ensure the correct placement during the
stimuli.
Motor evoked potentials can also be recorded
from the muscle using a surface, subcutaneous, or
intramuscular electrode though the size of a response
may have more to do with the distance between any
456
may be the result of only minimal reinnervation that

will not result in meaningful recovery of function.
For all of these reasons, recording MEPs over the spinal nerve, not muscle, is ideal when assessing for
integrity of the ventral roots.
32.3. Technical issues
Fig. 1. Bipolar stimulator held by surgeon on C6 spinal

nerve just outside exit from intervertebral foramen.
depolarizing muscle fibers and the recording electrodes and less to do with the actual number of functioning axons. Recording MEP over the end-organ
(muscle) has the limitation of not being able to determine whether a lesion is pre- or postganglionic. This
is vital information for operative planning in reconstructive surgeries. An absent MEP response recording over a muscle could indicate either complete
axonal disruption (no hope for recovery spontaneously) or regenerating axons just not all the way to
the muscle (some hope for spontaneous recovery).
If there is nerve regeneration proximally, this will
only be detected with spinal nerve or nerve-to-nerve
recordings. In addition, due to the amplification of
the MEP response over muscle, a small number of
motor axons can elicit a response which may not be
of functional significance. A small response, therefore,
In both SEP and MEP recordings, it is desirable to

have a normal control to confirm the correct functioning of all equipment and reliability of the study.
This would ideally entail study of a spinal nerve
known to be functioning by clinical examination,
radiologic studies, or preoperative electrophysiological studies with the expectation of SEP and MEP
responses. No response would indicate a technical
problem which would need to be remedied before
further testing is undertaken. Given the limited surgical exposure in most cases, however, this normal
control neurophysiological assessment may be difficult or impossible to accomplish. As an alternative,
the unaffected contralateral limb can be tested with
stimulation of the median nerve at the wrist checking for SEP responses over the cervical spine and
scalp. A MEP can be recorded over a muscle in the
contralateral limb [biceps or abductor digit minimi
(ADM)]. The recording electrodes over ADM can
also be used to record a compound muscle action
potential (CMAP) with peripheral (wrist) stimulation
in the monitoring of neuromuscular blockade.
Muscle artifact from neck and proximal arm muscles is a major technical challenge. The cervical SEP
potentials can be less reliable given the presence of
muscle artifact. When there is concern about a false
positive MEP response (i.e., a volume conducted muscle response), recording of a spinal nerve from a
known avulsed ventral root is helpful. A response there
confirms a likely false-positive response as a result of
volume conduction. If this is not possible, neuromuscular blocking agents can also be used to make this
distinction. If the waveform in question disappears
or decreases in size, a volume conducted response is
most likely because nerve action potentials (NAPs)
are not affected by neuromuscular blockade.
Excessive stimulus artifact can also be a problem
given the short distance between stimulating and
recording electrodes. Occasionally, one must move
the recording (for MEP) or stimulating (for SEP) electrodes distally onto brachial plexus elements to obviate

TD1:EI1 1:Pair1
TD2:EI2 1:Pair1
T
r
c
Rate:
Rate:
457
1.9 Hz
1.9 Hz
Lev: 11.0 mA
Lev: 0.0 mA
N:
46 NR:
4 ms Normal
0.5 V
Amp 1
N:
2
C5 root stimulation
Scalp/Neck recording
46 NR:
4 ms Normal
1 V Amp 2
A
TD1:EI1 1:Pair1
TD2:EI2 1:Pair1
T
r
c
Rate:
Rate:
1.9 Hz
1.9 Hz
Lev: 11.0 mA
Lev: 0.0 mA
48 NR:
4 ms Normal
Amp 1
0.5 V
N:
N:
2
48 NR:
4 ms Normal
Amp 2
1 V
C6 root stimulation
Scalp/Neck recording
B
Slope: Pos
Slope: Pos
TD1:Ext
TD2:Ext
T
r
c
Transcortical stimulation (192V)

N:
C5 root recording
4 ms
20 V
2 NR:
Normal
Amp 1
C
TD1:Ext
TD2:Ext
T
r
c
Slope: Pos
Slope: Pos
C6 root recording
N:
4 ms
10 V
2 NR:
Normal
Amp 1
D
Slope: Pos
Slope: Pos
TD1:Ext
TD2:Ext
T
r
c
P
N:
2 ms
10 V
O
C7 root recording
Rate:
Rate:
TD1:EI1 1:Pair1
TD2:EI2 1:Pair1
T
r
c
1.9 Hz
1.9 Hz
Lev: 18.8 mA
Lev: 0.0 mA
8 NR:
N:
Normal
0.5 ms
2 V
Amp 1
2 NR:
Normal
Amp 1
NAP
Fig. 2. AF: Recordings during brachial plexus exploration.
458
Table 1
Hooks held completely away from surrounding tissue or
Brachial plexus surgery monitoring with motor evoked

potentials
Display settings
Sensitivity: 1020 mV
Sweep: 5 ms
One channel recording
MEP monitoring should not be used for patients with implanted

medical devices:
Pacemaker
Pumps
Cochlear implants
fluid
3.
Contralateral normal arm recording with surface ECG

electrodes: biceps, extensor digitorum communis (EDC), or
hand
Transcranial cerebral stimulation is contraindicated on patients

with:
Epilepsy
Seizures
Skull fracture
Skull defect
Major head trauma
Stroke
Other intracranial disease
Aneurysm clips
Retained metal fragments
To test validity of an apparent root recorded MEP:
Before anesthesia, the anesthesiologist and neurophysiologist

should agree on the type and depth of anesthesia and neuromuscular block.
Anesthesia:
MEP, motor evoked potential.
1.
2.
Maintenance during MEP recordings
Narcotic agent (e.g., alfentanil, fentanyl, sufentanil,

remifentanil)
continuous infusion, short-acting neuromuscular block

(e.g., atracurium, vecuronium)
Inhalational agents: mean alveolar anesthetic concentration
(MAC) 0.150.4 primarily for amnestic effect
Insert soft bite block between teeth for transcranial stimulation
Set-up:
Digitimer stimulation
Nonrecurrent
C3-C4 (C4-C3) test both polarities
25 pulses at the rate of 24 ms interpulse interval as needed
for maximal response
Find maximal response with lowest stimulus voltage (up to

1,000 V at 0.05 ms, but 200600 V is usually enough)
Recordings
1.
Insert EEG needles before skin preparation:
Rhomboid and infraspinatus, far medial, bent monopolar

electrode
Trapezius, upper near midline

Two (or more) channel recording
2.
recording (should change)
Reverse recording electrode polarity (MEP waveform

should invert)
Stimulate at 200 V above maximal (MEP should not continue to increase)
Test adjacent nerve roots for volume conduction spread (should

be no MEP from clearly avulsed roots)
Increase or decrease neuromuscular block (MEP should not

change)
Induction
Midazolam or similar agent

Short-acting neuromuscular block
With propofol or similar agent

Neuromuscular blockade usually not necessary but if,
3.
Compare MEP for latency change with proximal and distal
Spinal nerve recording:
Bipolar hook electrodes

Minimum of 3 mm between leads
this problem. The polarity of the MEP stimulus can

also be alternated and several stimuli averaged in an
attempt to reduce the stimulus artifact by phase
cancellation.
Anesthesia can have a major detrimental impact
on IOM, especially the use of inhalational agents that
suppress cortical excitability. For SEPs, this negatively
impacts scalp recordings more than cervical spine or
nasopharyngeal recordings. These agents reduce the
effectiveness of transcranial electrical stimulation in
initiating a MEP. Low systemic blood pressure can
reduce SEP and MEP amplitudes. Neuromuscular
blocking agents may be desirable for SEP or MEP
studies in which muscle artifact must be eliminated.
Our preferred anesthetic regimen is intravenous
narcotic and propofol with use of medium-acting
neuromuscular blocking agents. Once adequate responses are obtained, low-level halogenated agents can
be used.
Accurate communication with the surgical team is
vital, as mentioned above. The IOM physician and
technicians cannot always see into the operative field
during the SEP or MEP studies. Before beginning
the stimuli, it must be confirmed that the surgeon is
holding the electrodes in the proper place with the
proper orientation and that they continue to hold
their position until the stimuli are finished. In the
hectic environment of the operating room, often with
multiple surgeons, this can be a challenge. Communication with the anesthesia team is also important,
especially as it relates to the use of inhalational
agents and the use of short-acting neuromuscular
blocking agents during the procedure. This, too, can
be challenging with teams of anesthesiologists/nurse
anesthetists or rotating anesthesia staff during these
long procedures.
In children, a few points must be kept in mind.
Generally, children are more sensitive to the cortical
suppressant effects of inhalational anesthesia and there
is more variability in this sensitivity. The distances
between stimulation and recording electrodes are
obviously shorter, making stimulus and muscle artifact
more of a factor.
32.4. Interpretation
The presence of a central SEP response (scalp or cervical spine) after spinal nerve stimulation indicates
the continuity of the dorsal root in cases where avulsion is questioned (Hashimoto et al., 1999; Oberle
et al., 2002). Lack of a response argues for dorsal
nerve root avulsion or disruption, especially when
NAPs can be recorded from the corresponding spinal
nerve or plexus element. In a pure preganglionic
lesion affecting the dorsal roots, the cell body and
peripheral axon are still intact and a peripheral
NAP would be expected, usually with normal conduction velocity. A MEP response indicates continuity of the ventral root while absence suggests root
avulsion (Carvalho et al., 1997). It is important to
realize that the integrity of the dorsal root does not
guarantee integrity of the ventral root, and vice versa
(Oberle et al., 2002). A mismatch (partially avulsed
dorsal or ventral roots) was noted in 11% of roots
studied by laminectomy (Carvalho et al., 1997). In
most of these instances, the ventral root was avulsed
with an intact dorsal root. A combination of these
two IOM techniques (SEP and MEP), therefore,
may be ideal (Burkholder et al., 2003; Harper,
2005). It is important to realize that from a surgical
reconstruction perspective, the continuity of the ventral roots is the most functionally important to determine. Functioning ventral roots can be used as the
proximal stump for nerve grafting. The loss of the
ventral roots as a grafting vehicle will necessitate
other transposition procedures utilizing nerve and/or
nerve muscle transfers from other territories (Spinner
and Kline, 2000; Shin et al., 2005).
459
32.5. Application to brachial plexus

reconstruction
Given the complexity of brachial plexus injuries, in
terms of anatomy and type of injury, intraoperative
electrophysiologic monitoring is essential to enhance
clinical outcome. Brachial plexus injuries are typically
classified as preganglionic, postganglionic, or a combination of both. The continuity of the spinal nerve is
likely the most important factor in surgical planning
since root avulsion is nonreversible with no chance for
recovery spontaneously or with primary anastomosis
or grafting procedures. In a suspected postganglionic
injury, IOM is also essential to determine the presence
of peripheral axonal continuity at a time when CMAPs
or voluntary motor unit potentials cannot be detected
due to lack of distal axonal regeneration.
Preoperative evaluation is important to determine
the degree of both vertical (root or plexus level) and
horizontal (i.e., preganglionic versus postganglionic)
involvement; however, available methods are not
always adequate to make this determination. Factors
suggestive of preganglionic injury include: winging of
the scapula due to serratus anterior weakness, presence
of Horner syndrome, or pseudomeningocele seen with
myelography or MRI. Unfortunately, myelography
and MRI may not be able to clearly identify root avulsion in some cases (Carvalho et al., 1997). The presence
of sensory nerve action potentials on routine nerve conduction studies in a flail limb is suggestive of a preganglionic lesion; however, the absence of these sensory
responses may represent either a postganglionic process
or a mixed process with additional root avulsion. The
presence of prominent fibrillation potentials in cervical
paraspinal muscles also suggests root avulsion but cannot determine the precise root level. Intraoperatively,
after surgical exposure, anatomic continuity of the
nerve root can generally be determined although if an
avulsed root remains intradural or, as is frequently the
case, is very scarred, anatomic continuity cannot be
determined. Additionally, even with full exposure (with
laminectomy), nerve roots in anatomic continuity may
not conduct electrically (Turkof et al., 1997; Oberle
et al., 2002). The functional integrity of the axons, then,
cannot be precisely determined without electrophysiological techniques.
The following case represents the use of these IOM
techniques. A 42-year-old man presented with a flail
left arm after a snowmobile accident. Six months after
the injury, there was complete loss of motor and sensory function in the arm. On clinical examination,
460

10 V
SEP
5 ms
CMAP
5 mv
10 ms
Root
Trunk
Cord
Nerve
Sensory
Muscle
R1
R2
S
20 V
2 V
4 ms
2 V
1 ms
4 ms
MEP
NAP (R2)
NAP (R1)
A
10 V
SEP
5 ms
Root
Trunk
Cord
Nerve
Sensory
Muscle
S
R
2 V
20 V
1 ms
4 ms
CMAP
NAP (R1)
MEP
5 mv
10 ms
B
10V
SEP
CMAP
5 ms
5 mv
10 ms
Root
Trunk
Cord
Nerve
Sensory
Muscle
S1
20 V
MEP
R1
R2
2 V
50 V
4 ms
4 ms
NAP (R1)
4 ms
NAP (R2)
C
(Fig. 3 continued)
461
10V
SEP
5 ms
Root
Cord
Trunk
Nerve
Sensory
Muscle
R
2 V
20 V
MEP
1 ms
4 ms
CMAP
5 mv
NAP
10 ms
D
10 V
SEP
5 ms
Root
Trunk
Nerve
Cord
Sensory
Muscle
S
20 V
MEP
R
2 V
4 ms
4 ms
CMAP
NAP (R1)
5 mv
10 ms
E
Fig. 3. IOM findings with various patterns of brachial plexus injuries. A: Complete ventral and dorsal avulsion. B: Ventral
root disruption with preservation of sensory fibers. C: Complete postganglionic lesion. D: Incomplete postganglionic lesion.
E: Mixed preganglionic and postganglionic lesions.
there was contraction in only the rhomboid, trapezius,

and serratus anterior with diffuse sensory loss
throughout the limb. Reflexes were absent. A Tinels
sign was present in the supraclavicular region radiating to the thumb. There was no evidence of Horner
syndrome. Routine nerve conduction studies revealed
absence of median and ulnar motor responses and a

low-amplitude median sensory nerve action potential.
The lateral antebrachial sensory response was absent.
Needle examination showed dense fibrillation potentials with no motor unit potentials activated in all limb
muscles plus infraspinatus. Rhomboids were normal as
462
were the mid-cervical paraspinals. Prominent fibrillation potentials were noted in low cervical paraspinals.
These findings were consistent with a very severe left
pan-brachial plexopathy with mixed preganglionic
and postganglionic injury and probable complete root
avulsion affecting the lower segments (C8, T1, and
possibly C7). There was likely at least partial preservation of the C5 root although this was difficult to predict
with absolute certainty because in some cases the
rhomboids may be innervated by C4. A CT myelogram
was consistent with left C8 and T1 nerve root avulsions. It was felt that he likely had intact outflow from
C5 and C6. The continuity of C7 was uncertain but
given the serratus anterior activation, this was likely
to be intact. Reconstructive surgery was indicated with
IOM to determine root continuity and assist with surgical planning.
After surgical exposure, visual inspection suggested that the C5, C6, and C7 roots were intact with
probable significant postganglionic injury. C8 and
T1 roots were visibly avulsed. Intraoperative SEP
testing with stimulation of the C5 and C6 roots
showed reproducible responses (Fig. 2A and B).
Motor evoked potentials were also present on C5,
C6, and C7 roots while under a short-acting paralytic
agent (Fig. 2CE). This confirmed that there was both
motor and sensory root integrity of C5 and C6 and at
least motor root integrity of C7. Attention then turned
to determining whether there was axonal continuity
across the injured brachial plexus. No reproducible
response could be recorded across the upper or middle
trunk segments but a NAP was present stimulating and
recording proximal to this level (Fig. 2F). Based on
these findings, the following grafting procedures were
performed: C5 to axillary nerve, C6 to musculocutaneous nerve, and C7 to radial nerve. Also performed
was a transposition of the contralateral C7 root to the
left median nerve via a vascularized ulnar nerve graft.
In this particular case, IOM confirmed integrity of the
C5-C7 roots which were then used as grafting vehicles, allowing the contralateral C7 root to be used in
an attempt to achieve more distal reinnervation. It also
helped to define the proximal extent of the nerve
injury.
Figure 3 shows examples of various injuries with
the expected IOM findings. Fig. 3A and B both have
ventral root avulsion, eliminating a grafting procedure, even with the presence of a SEP in Fig. 3B.
A complete postganglionic injury is depicted in
Fig. 3C. The SEP and MEP are present whereas a

NAP across the plexus is not present. Stimulation
and recording in the proximal plexus yields a NAP
identifying the proximal extent of the lesion. In this
setting, nerve grafting/transfers or end-to-end anastomosis would be appropriate. If there is a NAP present
across elements of the brachial plexus (Fig. 3D), only
neurolysis of that segment would be done. A mixed
lesion with sensory root avulsion and severe postganglionic injury is shown in Fig. 3E. In this case, the
utility of performing both MEP and SEP studies is
demonstrated as absence of SEP without testing of
MEP could be interpreted as a low likelihood of a
successful grafting procedure when, in reality, that
is likely to be most beneficial.
32.6. Summary
In performing intraoperative SEP and MEP studies,
there are several technical challenges that must be monitored for and overcome in order to acquire reliable
electrophysiological data. This data is indispensable in
the timely assessment of the functional continuity of
nerve roots in the surgical approach to brachial plexus
reconstruction.
References
Vennettilli, M (2003) Neurogenic motor evoked potentials: role in brachial plexus surgery. Case report. Neurosurgery, 98: 607610.
Carvalho, GA, Nikkhah, G, Matthies, C, Penkert, G and
Samii, M (1997) Diagnosis of root avulsions in traumatic brachial plexus injuries: value of computerized
tomography, myelography and magnetic resonance
imaging. J. Neurosurg., 86: 6976.
Harper, CM (2005) Preoperative and intraoperative electrophysiologic assessment of brachial plexus injuries.
Hand Clin., 21: 3946.
Hashimoto, T, Mitomo, M, Hirabuki, N, Miura, T,
Kawai, R, Nakamura, H, Kawai, H, Ono, K and
Kozuka, T (1999) Nerve root avulsion of birth palsy:
comparison of myelography with CT myelography
and somatosensory evoked potential. Radiology, 178:
841845.
Oberle, J, Antoniadis, G, Kast, E and Richter, HP (2002)
Evaluation of traumatic cervical nerve root injuries by
intraoperative evoked potentials. Neurosurgery, 51:
11821190.

Shin, AY, Spinner, RJ, Steinmann, SP and Bishop, AT
(2005) Adult traumatic brachial plexus injuries. J. Am.
Acad. Orthop. Surg., 13: 382396.
463
Turkof, E, Millesi, H, Turkof, R, Pfundner, P and Mayr, N

(1997) Intraoperative electroneurodiagnostics (transcranial electrical motor evoked potentials) to evaluate the
functional status of anterior spinal roots and spinal
nerves during brachial plexus surgery. Plast. Reconstr.
Surg., 99: 16321641.

M.R. Nuwer (Ed.)
464
CHAPTER 33
Technological advances in intraoperative

neurophysiological monitoring
Robert J. Sclabassia,b,c,d,*, Jeffrey Balzera,b,d
Donald Crammonda,b and Miguel Habeycha,b,1
a
Center for Clinical Neurophysiology, University of Pittsburgh, Pittsburgh, PA 15213, USA
Department of Electrical Engineering, University of Pittsburgh, Pittsburgh, PA 15213, USA

d
33.1. Introduction
Intraoperative neurophysiological monitoring provides
a real-time control loop around a system composed of
the surgeon, patient, and anesthesiologist. The goals of
this control loop are both the reduction of morbidity
and a dynamic assessment of the structurefunction
relationships of the patients nervous system. This is
accomplished by making specific and sensitive measurements that reflect the interactions between the
surgeons operative manipulations and the functioning
of the patients central nervous system (CNS). This
requires obtaining real-time measurements of CNS
function that can be closely correlated with operative
manipulations. To achieve these measurements within
a time frame that is of value to the progress of the
operation, it is highly advantageous to acquire, process, and display multimodality data both rapidly and
simultaneously. In addition, the paucity of trained
individuals to interpret this real-time data has led to
the development of powerful internet-based remote
viewing and communication facilities as a means for
overcoming this lack of qualified neurophysiologists.
Many technical advances have led to the ability
to acquire multimodality data simultaneously with
appropriate acquisition parameters for each data type,
*
Correspondence to: Robert J. Sclabassi, MD., Ph.D.,
Department of Neurological Surgery, School of Medicine,
University of Pittsburgh, Pittsburgh, PA 15213, USA.
Tel.: 1-412-692-5093.
E-mail: sclabassirj@upmc.edu (R.J. Sclabassi).
1
Present Address: Department of Neurological Surgery,
to digitally filter this data with filter settings which

are appropriate for each data type, to display the
data in flexible displays based on modern graphical
user interface technologies, and to support remote
internet-based viewing and interpretation of the data
as well as continuously available communications
between interpreting physician/neurophysiologist and
a locally positioned neurotechnologist.
This chapter briefly describes the technical
advances which have occurred that allow data to be
acquired rapidly from multiple data types simultaneously, including modern stimulation, display, and
filtering techniques. Particular attention will be paid
to technology which supports remote viewing of data
and communication between remote neurophysiologists and local neurotechnologists.
33.2. Multimodality neurophysiological
measures
Neurophysiological measures routinely utilized in
monitoring can provide a functional map of much of
the entire neuroaxis when acquired and viewed either
simultaneously or nearly simultaneously. Being able
to acquire multimodality data represents an important
technical advance; however, being able to do so
requires an understanding of the relationships between
the physiological generators of the various neurophysiological measures which are utilized and by implication of the requirements which these relationships
place on the data acquisition parameters.
For example, during an anterior cord decompression and fusion, it is reasonable to simultaneously
acquire bilateral median (MSPs) and posterior
tibial nerve evoked potentials (TSPs) as well as
continuous electromyograms (EMGs); during clipping of a basilar tip aneurysm, it is reasonable to

simultaneously monitor auditory brainstem potentials (BAPs), bilateral median nerve somatosensory
evoked potentials (MSPs), and digital EEGs; during
resection of an acoustic neuroma to simultaneously
monitor BAPs, MSPs, continuous EMGs from muscles innervated by various cranial nerves, and
evoked EMGs (CN MEPs) from direct stimulation
of cranial nerve VII; or during resection of a spinal
cord tumor to simultaneously acquired bilateral posterior tibial nerve evoked potentials, transcranially
evoked descending activity (TCr MEPs), and continuous EMGs. These are just representative examples of how multimodality data acquisition may be
utilized to enhance the efficacy of intraoperative
neurophysiological monitoring; but they provide
examples which clarify the issues in both neurophysiology and technology. (See Table 1 for definitions of above abbreviations.)
465
Figure 1 is an example of multimodality data

being acquired during the resection of an acoustic
neuroma. This figure demonstrates three different
data types being simultaneous acquired. These
are MSPs, BAPs, and continuous EMGs from three
different cranial nerves.
33.2.1. Relationships between neurophysiological
variables
An understanding of the fundamental similarities and
differences of the electrical characteristics of neurophysiological signals, including number of channels,
appropriate recording sites, and signal bandwidth to
maximize the independent information recorded is
necessary. An extensive review of all monitoring
modalities is presented in Sclabassi et al. (2006).
The recording and stimulating parameters for the
EEGs, MSPs, TSPs, BAPs, EMGs, and MEPs are
summarized in Table 1.
Table 1
Examples of modalities and modality parameters which may be simultaneously acquired and displayed
Modalities
Stimulation sites
Recording sites
Observation
intervals
Stimulus rates
Delay
Bandwidth/
sampling rate
MSPs
or
USPs
MS and MD
P3/F3, P4/F4
Cv2/Fz, EP
100 ms
3.43 Hz (SM)
100 ms (BS)
2.45 Hz (BM)
200 ms (BM)
TSPs
or
PSPs
TS and TD
3.43 Hz (SM)
100 ms (BS)
BAPs
AS or AD
MEPs
1300/
2,000 Hz
1001 kHz/
4 kHz
1300/
2,000 Hz
1001 kHz/
4 kHz
1001 kHz/
8 kHz
31 kHz/8 kHz
31 kHz/8 kHz
31 kHz/8 kHz
1100 Hz/
500 Hz
11 kHz/8 kHz
US and UD
Pz/Fz, P3/P4
Cv2/Fz, L1/L2
100 ms
PS and PD
200 ms (BM)
EEGs
TCr
CNs
PedS
None
Cz/A1, Cz/A2
Cz/Cv2
apmgs
apmgs
apmgs
10/20 montage
EMGs
None
apmgs
20 ms
17.5 Hz
None
50 ms
None
14 s
Burst
5.1 Hz cont.
5.1 Hz cont
None
None
50 ms1 s
None
None
Three letter extensions are used for evoked potential (EPs) modalities to allow unique identification of data types (note figures in text).
MSP, median nerve EPs; MS, left median nerve; MD, right median nerve; USP, ulnar nerve EPs; US, left ulnar nerve; UD, right ulnar
nerve; TSP, posterior tibial nerve EPs; TS, left tibial nerve; TD, right tibial nerve; PSP, common peroneal nerve EPs; PS, left peroneal
nerve; PD, right peroneal nerve; BAPs, brainstem auditory evoked potentials; AS, left ear; AD, right ear; no restrictions on stimulation
MEPs; TCr, transcranial electrical stimulation; C3/C4 (anode/cathode), right muscle groups; C4/C3 (anode/cathode), left muscle groups;
stimulus is 250 Hz burst lasting 20 ms (five pulses); 200 ms width, 150 V; CNs, all cranial nervesrecording from appropriate muscle
groups (apmgs), stimulus is 5.1 Hz, continuously; intensity is 0.110 V; PedS, pedicle screws recording from appropriate muscle groups
(apmgs), stimulus is 5.1 Hz, continuously, intensity is 520 V; SM, single modality stimulation; BS, both sides, for example, MS/MD with
100 ms delay between; BM, two modality stimulation; BS, both sides, for example, MS/MD than TS/TD; MEPsTCrmust be synchronized with MSPs/TSPs.
466
R.J. SCLABASSI ET AL.
33.3. Technological features supporting

data acquisition
Much of the material related to acquiring and displaying
data is discussed elsewhere in this volume; thus we shall
discuss these issues from the perspective of multimodality acquisition and display. To make this discussion
concrete, we shall frame it in the context of the system
which we utilize (NeuroNet#; Computational Diagnostics, Inc.) which is a system developed in our group
to support multimodality data acquisition and remote
professional supervision and which is a distributed computing system based on workstation and network technology, which provides user-transparent, shared file
systems and powerful interprocess communication protocols to facilitate sharing and consulting on real-time
multidimensional neurophysiological data (Sclabassi
et al., 1987, 1996; Krieger et al., 1988, 1991; Simon
et al., 1995). Other systems also have these properties,
thus this discussion is not intended as an endorsement
of this particular system but is purely a convenient
didactic approach to the discussion of the technology.
The concept behind this system is that each data
acquisition node or viewing node is analogous to a
neuron in a network (Fig. 3), and has an important feature in that the communication between these nodes is
integral to the functioning of the system. All acquisition software is integrated; that is, there is no concept
of separate packages for each type of data being collected. This provides maximum capability for collecting and analyzing combinations of different data types.
The EEG capabilities include compressed spectral
arrays on all available channels, digital filtering of
EEG, and real-time spectral computations on the
incoming data with arbitrary length spectral averages.
The EMG capabilities include acquiring spontaneous
and evoked data (MEPs); while the evoked potential acquisition capabilities include simultaneous
mixed modalities with appropriate sampling rates,
467
observation intervals, digital filtering, and noise estimation. The system provides real-time remote viewing
of all acquired data, multiway audio or text communication across the network, and unified user interfaces
for local and remote systems, requiring familiarity
with only one user interface (Krieger et al., 1988,
1991; Sclabassi et al., 1987, 1996). Some versions of
the system have also included real-time video streams
for operating microscopes or endoscopes (Sclabassi
et al., 1991, Nardi et al., 1993; Simon et al., 1995).
This system permits simultaneous data collection
and on-screen viewing of multiple modalities; each with
user-determined observation intervals and stimulus
rates which can be independently displayed and processed in real time on any other system on the network.
All data manipulations are handled by calls to a
common data file library (Neuro Data File or NDF),
a file system developed for use in this NeuroNet system. NDF uses a specific convenient support structure
(Neuro Data Structure or NDS). NDS supports the concept of a case abstraction, that is a logical grouping
of all data pertaining to a single patient. Different data
streams are identified and managed by a channel
manager structure. Data types (Table 1) are defined
for classes of neurophysiologic, physiologic, and
anesthesiologic data. The channel manager contains
all pertinent information for each data type in its
header portion, and handles variable length records.
This system uses an extensive package for evoked
potential data collection and presentation. All modalities may be collected individually or mixed simultaneously. Data trending over time is flexible in that
each channel of each modality may be independently
displayed and controlled. Examples of processing
available for all signals include digital filtering, standard averaging, odd/even averaging, noise estimation, and peak marking (both time and amplitude).
The user may enter comments at any time during data
collection and may store and retrieve comments from a
Fig. 1. Multimodality data acquired during resection of an acoustic neuroma displayed in NeuroDisplay. Twelve channels
of data are shown with different sensitivities and observation intervals dependent on the data type. Channels 7, 8, and 9 (top
three) are BAP data acquired from left ear stimulation with an observation interval of 12 ms, stimulus rate of 17.4 Hz, and
number of stimuli being 512. Channels 10, 11, and 12 (second group of three) are MSPs to MS and MD stimulation 100 ms
apart. The observation interval in this case is 200 ms, the stimulus rate is 2.45 Hz, and 128 stimuli are averaged. The light
traces in the top six channels are baseline data; while the dark traces are current data. Channels 1 through 6 are continuous
cranial nerve EMGs, with Channels 1 through 3 (third group of three) being from the facial nerve and the last three from
cranial nerves VI, IX, and X, respectively. Irritation activity is shown on the mentalis branch of CN VII. The observation
interval in all six channels is 1,000 ms. Baselines may also be displayed for continuous data if desired. Sensitivities for each
channel are noted on the left side of the display.
468
list of predefined comments for quick annotation of a

currently collected record through a pop-up window.
These comments are automatically appended to the data
record and appear remotely in (Figs. 5 and 6) windows.
Baseline data may be displayed for any waveform (both
in the real-time displays and in trended displays). The
baselines may be retrieved from any channel of any data
file, thereby permitting the inclusion of baselines from
preoperative studies. Artifact rejection is also fully user
controlled. The user may define two time windows per
channel for artifact rejection. Furthermore, the amplitude rejection criterion is user-settable along with a
spike allowance parameter which permits a percentage of the data to exceed the artifact rejection limits
without throwing away the trial. The data acquisition
nodes provide the additional flexibility of being able to
operate independently of the network. Whether networked or standing alone, the acquisition nodes provide
complete user-control over every acquisition parameter.
33.3.1. Technical considerations
The acquisition of multimodality data imposes great
flexibility requirements on the electronics of the system. Since multiple input data channels can be reformatted to construct multiple data display channels
(traces), all input data channels are required to have
the same input impedance, anti-aliasing analog filter
characteristics, signal gain, and sampling frequency.
All specific sampling, filtering, and sensitivity adjustment are performed digitally when the data are reformatted for the desired signal channels.
33.3.1.1. Signal acquisition
Signal acquisition from the perspective that multimodality data places on a system is slightly different
than when systems are conceived of as acquiring
single modality data. These differences in approach
come from the required flexibility of combining data
acquired from different electrodes in multiple combinations to be consistent with different data types. For
example, P3/F3 could be both utilized for recording
MSPs to right median nerve stimulation and an
EEG channel at the same time.
Also P3/P4 could be utilized at the same time as
a component of the montage for recording TSPs.
33.3.1.2. Analog filtering
Unwanted activity (noise) originates from both the
signal recorded from the subject and electrical
devices in the immediate neighborhood of the
recording equipment. Since the aim of evoked potential recording is to ensure a large, clear response with
the least possible noise contamination (i.e., the best
signal-to-noise ratio possible), the elimination of
these unwanted signal components is essential. This
elimination is accomplished partially through the use
of analog filtering techniques combined with averaging and digital filtering. A source of potential noise is
the relationship between the sampling frequency and
the frequency contents of the neurophysiologic signals. The analog input filters must bandlimit the neurophysiologic signals to less than half the sampling
frequency otherwise signal distortions are created by
the aliasing of high frequency components into the
low frequency spectrum of the signal.
The frequency response characteristics are defined
by the high-pass cutoff point (i.e., the frequency
above which the amplifier passes the frequency components of the signal essentially unattenuated), the
low-pass cutoff point (i.e., the frequency below
which the amplifier passes the frequency components
of the signal essentially unattenuated), and the rate of
attenuation occurring below and above these cutoff
points, respectively. Care must be given to providing
minimum phase shift through this filtering process as
the phase shifts for different frequencies will also
introduce signal distortion. Analog filters are provided
at preset values in all preamplifiers to act as antialiasing filters. In many systems, analog filters are also
provided with selectable values at the amplifier stage,
even though this feature is probably not useful in the
context of multimodality testing where digital filtering
is more useful. Care must be taken to ensure that the
analog filtering built into the preamplifiers (or in the
first stage of the amplification) is not so tight as to provide apparent noise-free data at the expense of significant signal distortion.
33.3.1.3. Signal amplification
Neurophysiologic signals are most commonly amplified using differential amplifiers, that is, amplifiers in
which two input channels to the amplifier are differenced. Differencing has the effect of eliminating
identical (in-phase) signal components which might
be present at each recording electrode (presumably
noise), and retaining the signals which are different
(out-of-phase) and presumably produced by different
physiological generators appropriate for each data
type. Important functional specifications for these
amplifiers are the input impedance, common mode
rejection ratio, sensitivity, gain, noise figure, frequency
response, and output impedance. The input and

output impedances determine, respectively, the effectiveness with which an amplifier picks up electrical
activity from the electrodes at the scalp and passes
it on to other components. The input impedance of
the amplifier must be great compared to the impedance between the recording electrodes feeding the
amplifier (greater than 100 times larger). The input
impedances of modern commercial amplifiers are usually greater than 100 MO, and the greater this impedance is the more fidelity is captured in the recorded
differential signal. The gain specifies the factor by
which an amplifier multiplies the voltage at the input
to produce the output voltage. Amplifiers typically
have had adjustable gains, with ranges between
1,000 and 500,000; however, with the desire to have
all signals as common as possible in characteristics
prior to reformatting, a single fixed gain may be used
on all amplifiers The availability of 1618-bit A/D
converters allows all neurophysiologically interesting
signals to be acquired with the gain being on the order
of 80 dB.
33.3.1.4. Common mode rejection ratio
Common mode rejection ratio (CMRR) governs the
amplifier efficiency in discriminating between the
local potentials of interest and other, usually larger,
interference potentials, picked up at both input electrodes (e.g., 60 Hz). Apart from brain activity, bioelectric activity originating from muscles in the
head and neck, from the eye and heart are present
and as previously mentioned, are likely to be much
greater than the evoked potentials of interest. Also
present at the scalp will be relatively large induced
voltages of extraneous origin arising from other electrical equipment. Amplifiers increase all signals presented across the input leads irrespective of their
source by a factor of their gain. With differential
amplification, electrical activity from one pair of electrodes connected to the amplifier is compared directly
to the activity presented at the other electrode, thus
only potential differences between the two input electrodes are amplified. Any potentials which are picked
up equally at both electrodes (common mode or inphase signals) are canceled out and only the voltages
developed between the two electrodes (out-of-phase
signals) are separated and preferentially amplified.
The effectiveness with which a differential amplifier
rejects in-phase signals compared to its ability to
amplify out-of-phase signals is called the CMRR.
Differential amplifiers used in neurophysiologic
469
investigations typically have CMRRs of greater that

80 dB; that is, an in-phase signal has to be 10,000 times
greater than an out-of-phase signal in order for both to
be recorded as the same size signal. For efficient rejection of in-phase signals, it is extremely important that
the electrode impedances of a pair of electrodes should
not only be as low as possible, but as similar as possible for both electrodes of the input pair; since any
impedance inequalities will produce amplitude differences in the in-phase activity that will be amplified
along with the desired signal. Thus, the observed signal would consist of the desired signal and the component of the in-phase signal due to the impedance
imbalance.
33.3.1.5. Sensitivity
The sensitivity of an amplifier specifies the range of
input voltages which it will amplify without distortion; that is, it refers to the actual voltage relationship
between the amplifier input and output. The amplifiers working range is specified by the minimum
input voltage which will produce a specified output
voltage suitable for interfacing with other equipment,
and the largest output voltage producing an undistorted output voltage. The noise figure for the amplifiers specifies the magnitude of noise inherent in the
amplifier itself, independent of all other factors.
33.3.1.6. Analog-to-digital conversion
In order to process data digitally, the analog signals
must be sampled using an analog-to-digital (A/D) converter, which has a maximum peak-to-peak input voltage range (e.g., 10 V) and which must be sized with
respect to the characteristics of the amplifiers to provide maximum sensitivity. The sampling rate is determined by the frequency content of the signal being
measured and the Nyquist sampling criteria, which
specifies that the sampling rate must be at least two
times greater than the maximum frequency content of
the signal. The sampling of a signal at greater than
the Nyquist rate avoids signal distortion produced by
aliasing due to sampling the signal at too low a sampling rate as mentioned in the discussion on antialiasing filters. It is best, in the multimodality situation
to sample at higher sampling rates (four to six times the
Nyquist rate) and then decimate the data to the desired
sampling rate for each combined data channel while
the high frequency sampled data is stored for possible
recombination to produce other display traces.
Modern A/D converters have a 1618-bit accuracy,
sample and hold circuits for accurate conversion, and
470
single-channel throughput rates of 40 kHz. The details

of these data acquisition elements are not always
important to the user; however, it is worthwhile to have
some idea that they exist and what their limitations are
when attempting to acquire and interpret data at the
extremes; for example, auditory brainstem responses
which tend to be comparatively small in amplitude
and rapidly occurring.
33.3.1.7. Digital filtering
Digital filters enhance the extraction of the signal
from the noise. Digital filters have several significant advantages including the ability to introduce
zero or constant phase shift (important in assessing
latencies in different components), and flexibility in
implementation (multiple filtering routines can be
utilized dependent on the nature of the data). These
calculations are implemented either as convolutions
(Krieger et al., 1991), regressions (Krieger and Sclabassi, 1994), or as manipulations on Fourier transforms (Sclabassi and Harper, 1973).
33.3.1.8. Stimulus timing
Multimodality data requires integrated timing capabilities for multiple data types. Most systems will
include multiple programmable real-time clocks used
to time interstimulus intervals, and delays between
the various stimulus channels, which normally have
a time resolution of at least 10 ms. For most flexible
use, multimodality systems require a separate timing
clock for each independent triggered data type.
33.3.1.9. Multimodality signal averaging
Evoked potentials are typically a fraction of the size of
the spontaneous brain activity appearing in the background EEG, and about one thousandth the size of
the other physiological and extraneous potentials with
which they are intermixed. The most effective method
for extracting the signal of interest from the noise, after
amplifying the signal with differential amplifiers, is to
use signal averaging, which is in effect a crosscorrelation between a point-process defined by the
occurrence of the stimuli and the recorded evoked
activity (i.e., an optimal filter). In averaging, the signal
component at each point is coherent and adds directly,
while the background and noise components tend to be
statistically independent and summate in a more-orless RMS fashion.
The usefulness of averaging as a signal extraction
technique is dependent on the assumption that the
observed data is stationary, that is, the data is not
changing rapidly. This assumption reenforces the

need to acquire data as rapidly as possible for any
single average. We have been investigating techniques for estimating time-varying evoked potentials
(Krieger and Sclabassi, 1994). In addition to the classical averaging techniques, a number of modified
averaging techniques have been developed and found
to be useful. These include odd/even averaging
(where two responses are computed for each data
channel, one from the even number stimuli, the other
from the odd numbered stimuli), moving averaging
(which allows a sliding average to be computed),
averaging to bursting trains (which allows high frequency response properties to be studied and which
are also useful in producing TCr MEPs at low stimulus intensities), random train stimulation (which
allows nonlinear properties and system interactions
to be characterized) (Sclabassi et al., 1977), and
noise estimation by plus/minus averaging (which
allows the residual noise on a response to be estimated and is extremely useful when a patient has significant pathology).
33.3.2. User interfaces
All user interfaces in our NeuroNet system are based
on X-windows and Motif and allow for the manipulation and presentation of all data types. The baseline
responses are displayed as a background display on
the computer monitor so that differences may be
automatically calculated and displayed. A waterfall
display window (NeuroView) is used to follow the
patterns of the change over a period of time during
the case with cascaded data. New responses are automatically updated to this display, whether locally or
remotely, as they are saved from the current data display (NeuroDisplay). Thus, the waterfall display provides a comparative record of the patients data and
facilitates the process of identifying significant
changes in activity either locally or remotely.
33.3.2.1. Data display
This again is discussed specifically for the system we
use as a didactic convenience and example. Other
systems are available with their own approaches to
address such issues. NeuroDisplay is an oscilloscope
style display, which allows current physiological data
to be reviewed on the local data acquisition machine
(see Fig. 1). It uses an X-Windows-based program,
and all the graphics programming has been coded
using Xt Intrinsics, XLib, and Motif convenience
functions. NeuroDisplay consists of a display area,

for display waveforms, and a menu. The options
available from the menu allow the user to control
how the display area presents data. NeuroDisplay is
a user interface screen and constitutes the front-end
to NeuroNet.
NeuroView is used to view data being acquired
either locally (Fig. 2) or across the network
(Fig. 5). NeuroView is also an X-Windows-based
program, and all the graphics programming has been
coded using Xt Intrinsics, XLib, and Motif convenience functions. NeuroView relies on NNCP to support network transport, and NDF to support data
access. NeuroView has five distinct components:
the Application Shell, the Spawner, the Help File
Reader, the Application Thread, and the Display
Shell. All of these components, except the Help File
Reader, rely on a context structure modeled after a
process context, which contains all the information
needed to characterize the state of a display shell.
The Application Shell consists of the user interface,
callback functions to set elements of the context
structure, the interface to the Spawner, and the interface to the Help File Reader.
Neurophysiologic data is captured in an observation window and displayed in a window of the same
length of time. However, many measures produced
by the same stimulus occur at different latencies with
respect to that stimulus. Thus, useful methods of data
display included in the software are being able to tailor the observation window to the expected latency of
the interesting component of the response. For example, short-latency potentials typically occur within
20 ms of the stimulus; while intermediate latency
potentials occur within the first 100 ms of the stimulus presentation. Thus, the optimal observation of
these different components is facilitated by the capability to specify different observation windows for
different waves of interest, being observed at different regions of time after the stimulus presentation.
In addition, simple and easy display facilities are
provided for presenting digitally filtered data, spectra, and noise estimates along with the primary data,
as are capabilities for identifying, tracking, and comparing current data against baseline data.
33.3.3. Remote monitoring system
The NeuroNet Remote Monitoring System
(NNRMS) is used by neurophysiologists to remotely
monitoring surgical cases in near real time. It
471
interacts with the Neuro processes previously

described to automatically create remote monitoring
display windows (NeuroView windows in Figs. 5
and 6) for each modality of each case. This system
can also be used to review archived data, or monitor
an ongoing case by manual selection. Also lists of
comments annotated to the data records are automatically produced (Fig. 6C) to facilitate the remote
viewer understanding the case context.
33.3.3.1. Network structure
A fundamental and unique feature of the NeuroNet
system is its ability to support multiple mobile instrumentation carts and remote-viewer computers
(Fig. 3). All data acquired at any one of the instrumentation carts may be viewed at any other cart or computer (configured to be a node) connected to the
network. Thus, one neurophysiologist can monitor
several procedures at the same time. The system has
remote monitoring built-in to provide shared data as
well as instant typed (Fig. 4) or audio communications
between users. In addition, when on the network,
NeuroNet can automatically provide data backup as
the data is collected by the instrument carts. The facilities are designed to work across any intranet/internet
architecture. Therefore, virtually any site can be
connected in some form to another site. Data can be
displayed and analyzed anywhere within the configured network and performance is even supported
through low bandwidth modem connectivity.
33.3.3.2. NeuroNet Communication Protocol (NNCP)
The NeuroNet Communication Protocol level encapsulates the communication control structure. This level
provides global naming and location information.
Remote data access is provided by a robust software
system organized in two layers which are supported
by distributed daemons: a Parallel Virtual Machine
(PVM) Daemon (pvmd3); and an Information Services
Daemon (ISD). PVM enables a collection of heterogeneous computers to be used as a coherent and flexible
concurrent computational resource (Geist et al.,
1997). PVM provides global naming services, dynamic process groups, message passing, multicasting,
and global synchronization functions. A PVM daemon
(pvmd3) runs on each NeuroNet machine. In addition,
there is a single instance of the PVM group server
daemon on the network.
In the NNRMS context, the Display Shell and the
File reader are processes in the virtual machine that
472
exchange NeuroNet data. A typical NNRMS configuration is illustrated in Fig. 3. A collection of computers
accessible to each other through a TCP/IP network
connection are folded into a PVM virtual machine by
a master PVM node which is running the Spawner process. These nodes may be located in different hospitals
(or anywhere in the world, for that matter, as long as
they are accessible through the internet).
An ISD runs on each NeuroNet machine. This deamon service requests lists of active cases. It utilizes the
message transport services provided by PVM to
receive requests and service. Thus, the ISD is the
server process in a client server architecture, where
the applications are the clients. Each ISD maintains a
list of both historical and active cases on the machine
where it is running. Active cases are defined as those
which have collected and saved data within the past

hour.
Multiple applications may be run on each node.
The applications use case listings and case data,
fetched by the ISDs and transported by the pvmd3s,
to generate data displays. They utilize the dynamic
group services provided by PVM to identify server
processes (ISDs) and the PVM message transport services to send requests and receive responses.
Three main components constitute NNRMS: the
Spawner, the Display Shell, and the File Reader.
The Spawner is responsible for monitoring the network of computer node configures in the NeuroNet
system. Towards this end, the NNRMS maintains a
complete list of computer nodes which can register
with a location broker Daemon. On each network
(Fig. 2 continued)
473
Fig. 2. Local waterfall displays (NeuroView) of data from case in Fig. 1 providing cascaded history of the case for local
viewing. Neurotechnologists have complete control over size of the windows, positions on the computer monitor, number
of channels for each modality displayed, numbers of traces displayed, colors of traces and background, observation interval,
sensitivity, numbers of traces, and digital filtering of data. These may be set the same as in the Neurodisplay window (automatically) or may send them to the preferences of the neurotechnologists in the operating room. A: presents BAP data from
this case as displayed in a cascade display. The light trace at the bottom of the figure is baseline data, which may be different from that used in NeuroDisplay (Fig. 1). Fifteen traces are stacked and the observation filter is as in Fig. 1. This data
is also being digitally filtered. B: presents the MSP data being collected. In this figure, the observation interval for the data
is different with each frame having an observation interval of 100 ms. Baseline date for the MS responses (frame 1) and the
MD responses (frame 2) is again the light trace at the bottom of the figure. Frame 3 presents the subcortical data for both
MS and MD stimulation. The attached note notes irritation activity occurring on CN X. C: presents the continuous EMG
traces also being acquired simultaneously again with a 1,000 ms observation interval. In this figure, only the traces from the
three branches of CN VII are shown. Frame 3 demonstrates irritation activity being noted from the mentalis branch of CN
VII. Also noted the attached note noting irritation EMG activity on the mentalis branch. D: presents data not shown in
Fig. 1; namely evoked EMGs (MEPs) obtained over simultaneously by electrical stimulation through the tumor. In this figure,
all six EMG channels from which continuous EMGs are also being recorded are shown. However, the observation interval is
now 50 ms, synchronized to the occurrence of the stimulus pulse. The stimulus rate used in this situation was 5.3 Hz. Note the
MEP in frame evoked by stimulation of the mentalis branch of CN VII.
474
Controller
Master node
Display Shell
Display Shell
Data Reader
NeuroNet Workstation
NeuroNet Cart
Display Shell
Data Reader
Data Reader
NeuroNet Cart
NeuroNet Cart
Display Shell
NeuroNet Workstation
Fig. 3. NeuroNet Remote Monitoring System (NNRMS). The Master node executes a Spawner process based on lists of
system allowed into the network. This Spawner process activates all communication processes between nodes when they
are attached to the internet. The figure shows three NeuroNet data acquisition carts entered into the network, an instance
of the File Reader process is executing on each of these nodes. Two nodes in this example are workstations (installed in
the neurophysiologists office) and they execute only an instance of the Display Shell. The NeuroNet carts can also run
a Display Shell so that they can also function as viewing nodes for other cases as well as for the data being collected locally.
Thus, many data acquisition systems have multiple display screens which allow neurophysiologists to be in one operating
room and be interacting with cases in other rooms.
monitoring pass, the Spawner compares its list of

nodes with the list of nodes registered with the location broker. Then, it attempts to connect to the unregistered nodes to fold them into the NNRMS.
The connection between the Spawner and an unregistered node is made through Secure SHell public/
private authentication. The Spawner is an entry point

where the ISD can notify NeuroView that a remote
process is active. The Spawner is also responsible for
starting Display shells on user requests. The Application Thread is a routine responsible for waiting for data
from an acquisition process or a file and updating the
475
Fig. 4. Example of communication facilities provided within the NeuroNet system. A: is a communication control window
which is constructed to provide a limited set of nodes which an individual can communicate with. In our practice, we segregate the systems by place in a call rotation. All communication facilities may be started by all neurophysiologists and all neurotechnologists. B: presents a list of nodes entered into the list for second call at that particular time. This list will expand or
contract depending on what systems are on the network at any given time. The authors of this chapter, who are all neurophysiologists in our group, are all available on-line, as are a number of support personnel. In this list, only one data acquisition
node (nnstclair1) is actually listed. C: presents the real-time communication going on between one of our neurophysiologists
and a neurotechnologist using that particular data acquisition node during a carotid endarterectomy. The top half of the screen
are the comments from the neurophysiologists. The bottom half are the comments from the neurotechnologists, and the line in
the middle identifies the NeuroNet data acquisition cart (stclair1) and the call rotation (neuro_rmon2).
476
associated Display shell as necessary. The Help

Reader allows the user to obtain online Help about
NeuroView. It consists of an index of topics and a
scrollable area for displaying the help file.
The Display Shell is the component of the NNRMS
that the user interacts with most frequently. The shell
is composed of a drawing area, widgets the user utilizes
to select a style of information display, and all the functions actually required to perform the display. The shell
is responsible for managing the monitoring display
windows. As each case comes online, the Display Shell

creates a new window in which the neurophysiological
data being acquired is displayed. The Display Shell contacts the File Reader on the node with the new case and
requests the data as it is acquired. Examples of remote
NeuroView windows are shown in Fig. 5 for the case
in Figs. 1 and 2, and Fig. 6 for the communication facilities presented in Fig. 4.
For each monitoring window, a case context is
created. The context data structure contains all the
(Fig. 5 continued)
477
Fig. 5. Remote data as it appeared on neurophysiologists node in NeuroView from the case whose data is presented in
Fig. 1 (NeuroDisplay) and Fig. 2 (NeuroViewlocal). Note some preferential differences between displays. A: The BAPs
are displayed with a 20-ms observation interval and only 10 epochs displayed in the waterfall. B: The MSPs are displayed
with a 200-ms observation interval for each frame allowing the right and left stimulus effects to be viewed from each
hemisphere. C: presents irritation activity again on the mentalis branch of CN VIII; while D: presents evoked activity
predominately from the left CN X and weaker from CN II mentalis and oris branches.
478
(Fig. 6 continued)
479
Fig. 6. Remote data from carotid endarterectomy being discussed in NeuroComm window in Fig. 4C. A: presents bilateral
MSP data; while B: presents bilateral EEG data with both the continuous EEG and power spectrum being viewed remotely.
C: presents another communication tool to facilitate the remote neurophysiologists being aware of all aspects of the cases he
is responsible for. What is displayed in this Comments window are all appended comments by the neurotechnologists for
the cases for which that individual is responsible. Picking a particular row, for example the third from the bottom, the number 12 refers to the 12th epoch, the time, the data acquisition node, and the comment (reference Fig. 6A for comparison).
This window is automatically updated and each time a commented is appended to any data epoch for which that neurophysiologists is responsible.
information necessary to update the Display shell with

information. This information is stored in a stack of
structures, one structure per display shell. The context
of a shell is similar to the context of a process; when
one shell has an operation affecting it, the other shells
have their states stored and are unchanging. By accessing various elements on the stack, NeuroView quickly
and efficiently updates the real-time displays, each
with different parameters, utilizing the same routines.
NeuroView data may also be accessed across phone
lines, via phone dial up, for remote viewing on PCs.
For example, all the members of the center for clinical
neurophysiology (CCN) have the capability of accessing any activity on NeuroNet from home, allowing
them to consult on cases late at night.
The Data Reader is a process that resides on the
computer node where the Neuro process is executed.
It acts as a disk server that accepts calls requesting
NeuroNet data, reads information from disk, and
returns it to the calling process. This is a read-only
process, and does not return patient identifiable

information.
33.4. Summary
The commonly accepted principal goal of intraoperative monitoring is to prevent morbidity, and at a certain level this is true; however, the more fundamental
goal of intraoperative monitoring is to provide the
surgical team with information that allows them to
accomplish the desired operative objective with as
optimal a surgical strategy as possible, while having
a clear idea of what surgical morbidity is being
induced along the way.
We conceive of monitoring as placing a real-time
feedback control loop around a dynamic, changing
system comprised of the surgeon and the patient.
This requires a strong commitment to the concept
that the central nervous system of the patient is
highly sensitive to the operative manipulations of
480
the surgeon and that appropriately observed variables

may predict, and if appropriately interpreted prevent,
lesions about to develop. This information permits
the surgeon to dynamically modify his approach to
the operation and thereby minimize the degree of
morbidity induced in the patient.
Stringent time constraints exist in intraoperative
monitoring of neurophysiologic function, and damage to the central nervous system may occur rapidly.
This constraint has inspired the development of
methods for extracting and analyzing evoked potential, EMG, and EEG waveforms rapidly and efficiently. A corollary of the increased sensitivity
required to decrease the monitoring time is a higher
rate of individually false-positive measures. These
are usually, rapidly identified as such and produce
no disruption in the flow of the case.
In support of the intraoperative monitoring of
these measures, we have developed a distributed
computer system, NeuroNet, specifically configured
to support the considerations discussed in this chapter. This system provides both off-line and real-time
signal processing and data review capabilities, and
it addresses many of the problems associated with
the acquisition, processing, and display of multivariate neurophysiologic data in these complex cases.
It cannot be emphasized enough that the measures
utilized be both specific to the neural tissue being
manipulated and sensitive to changes in the functioning of the neural tissue produced by the surgical
manipulations. These measures must be obtained,
displayed, and interpreted simultaneously permitting
a multidimensional assessment of the integrity of
the neural structures at risk.
References
Geist, A, Beguelin, A, Dongarra, J, Jiang, W, Manchek, R
and Sunderam, V (1997) PVM: parallel virtual machine.
In: J Kowalik (Ed.), Scientific and Engineering Computation. MIT Press, Cambridge, MA.

Krieger, D and Sclabassi, RJ (1994) Time-varying evoked
potentials. J. Med. Eng. Technol., 18(3): 96100.
Krieger, DN, Lofink, RM, Doyle, EL, Burk, G and
Sclabassi, RJ (1988) NeuroNet: implementation of an
integrated clinical neurophysiology system. Med.
Instrum., 21(6): 296303.
Krieger, DN, Burk, G and Sclabassi, RJ (1991) NeuroNet:
a distributed real-time system for monitoring neurophysiological function in the medical environment.
Computer, 24(3): 4555.
Nardi, BA, Schwarz, H, Kuchinsky, A, Leichner, R, Whittaker, S and Sclabassi, RJ (1993) Turning Away from
Talking Heads: The Use of Video as Data in Neurosurgery. Proc. Interchi 93, AACM, New York,
pp. 327334.
Sclabassi, RJ and Harper, RM (1973) Laboratory computers
in neurophysiology. Proc. IEEE, 61(11): 16021614.
Sclabassi, RJ, Risch, HA, Hinman, C, Kroin, JS, Enns, NF
and Namerow, NS (1977) Complex pattern evoked
somatosensory responses in the study of multiple sclerosis. Proc. IEEE, 65(5): 626633.
Sclabassi, RJ, Lofink, RM and Doyle, EL (1987) NeuroNet: a distributed microprocessor network for clinical
neurophysiology. In: MJ Geisow and AN Barret
(Eds.), Microcomputers in Medicine. Elsevier, New
York.
Sclabassi, RJ, Leichner, R, Kuchinsky, A, Krieger, DN and
Prince, F (1991) The multimedia medical monitoring
diagnosis and consultation project. In: Proceedings of
the Hawaii International Conference on System Science
(HICSS-24): DSS and Knowledge-Based Systems,
IEEE CS Press, Los Alamitos, California. Vol. 3,
pp. 717728.
Sclabassi, RJ, Krieger, D, Simon, R, Lofink, R, Gross, G
and DeLauder, DM (1996) NeuroNet: collaborative
intraoperative guidance and control. IEEE Comput.
Graph. Appl., 16(1): 3945.
Sclabassi, RJ, Balzer, JR, Crammond, D and Habeych, ME
(2006) Neurophysiological monitoring: a tool for neurosurgery. In: LN Shekhar and RG Fessler (Eds.), Atlas
of Neurosurgical TechniquesBrain. Thieme, New
York, Chapter 3.
Simon, R, Krieger, D, Znati, T, Lofink, R and Sclabassi, RJ
(1995) Multimedia MedNet: a medial collaboration and
consultation system. Computer, 28(5): 6573.
SECTION III
Disorders Monitored during Surgery
Section III.1
Cerebral Neurosurgery

M.R. Nuwer (Ed.)
484
CHAPTER 34
Neurophysiology during epilepsy surgery

John M. Stern* and Jerome Engel, Jr.
Department of Neurology, Geffen School of Medicine, UCLA, Los Angeles, CA 90095, USA
For many patients with medication-refractory epilepsy, resective epilepsy surgery provides the greatest
chance of becoming seizure-free (Engel et al., 2003).
Such surgery is possible only when the epileptogenic
region, that is, the cerebral region necessary and
sufficient to generate the habitual epileptic seizures,
has been localized. Indeed, the likelihood of complete freedom from seizures following resection
depends upon the accuracy of the localization. However, no single diagnostic test can localize this region
and, ultimately, the epileptogenic region is approximated through a synthesis of results from complementary tests that assess for structural or functional
abnormality.
Electroencephalography (EEG) has been a standard functional test within the testing battery for most
of the history of epilepsy surgery. Reports of its use for
epilepsy surgery date to 1939 when a combination of
scalp and epidural recordings of both interictal and
ictal EEG were used to guide a temporal lobe resection
(Penfield, 1939; Almeida et al., 2005). The current
evaluation for epilepsy surgery still includes both
scalp and invasive EEG monitoring within the armamentarium of important tests. In its current sense,
invasive EEG monitoring is the extraoperative recordings of EEG through intraoperatively placed electrodes. These electrodes may be depth electrodes that
penetrate the cerebral parenchyma or subdural electrodes that are organized in strips or grids. The grid and
strip recordings essentially are extraoperative versions
of intraoperative electrocorticography (ECoG).
The percentage of patients who require invasive
EEG monitoring has decreased steadily over the past
25 years mostly due to the major advances in brain
imaging. The standard epilepsy surgery evaluation
*
Correspondence to: John M. Stern, M.D., UCLA Department of Neurology, 710 Westwood Plaza, Suite 1250,
Tel.: 1-310-825-5745; fax: 1-310-206-8461.
E-mail: jstern@ucla.edu (J.M. Stern).
continues to require ictal scalp EEG because EEG,

unlike the imaging modalities, has specificity for
epileptic abnormality. The patients who still require
invasive EEG monitoring usually are ones with normal
imaging, imaging demonstrating either a large lesion
or multiple lesions, or more than one ictal onset zone
on scalp EEG. However, those patients whose magnetic resonance imaging (MRI) or positron emission
tomographic (PET) scan demonstrate concordant
imaging and ictal scalp EEG abnormalities often are
eligible for resective surgery without invasive ictal
recordings. As such, imaging has become a critical factor in the evaluation of increasingly more patients and
invasive EEG monitoring has been required less often.
In contrast to the decreasing use of extraoperative
invasive EEG, intraoperative EEG (ECoG) remains a
commonly used and useful tool (Fig. 1). ECoG typically serves to refine the resection boundaries during
epilepsy surgery or other cerebral resections of
patients who have experienced focal seizures. Even
when the overall location of the resection is determined by the imaging abnormality, as is the case
when the imaging and the scalp EEG are concordant
and invasive EEG monitoring is not required, the
imaging abnormality is only an approximation of
the epileptogenic region. The actual boundaries
of the resection are determined by a combination of
the normal function of the surrounding tissue, the
anatomic extent of the epileptic functional abnormality, and knowledge of outcomes following resection
of abnormalities with similar location and pathologic
changes. For example, numerous case series have
shown that the margins of anterior temporal lobe
resections for mesial temporal lobe epilepsy may be
determined based upon knowledge of the common
pathophysiological and anatomical substrates of the
disorder and without supplementary, functional information from ECoG (Doyle and Spencer, 1997). For
patients with very frequent seizures during either
scalp or invasive EEG monitoring, intraoperative
CEREBRAL NEUROSURGERY
485
Fig. 1. ECoG obtained during epilepsy surgery. No epileptiform discharges are present. The recurring higher amplitude
slow wave that is most prominent in the ninth channel is pulse artifact generated by the placement of one electrode over
an artery (1 s between solid lines, low-frequency filter (LFF) 1 Hz, high-frequency filter (HFF) 50 Hz).
ictal ECoG during the epilepsy surgery may be useful

to demarcate the electrographic ictal onset zone more
precisely. Interictal ECoG is much more commonly
obtained. Its utility is in demarcating the epileptic
irritative zone, the region producing interictal discharges that usually includes the epileptogenic zone
and extends beyond it.
Overall, this chapter will discuss the uses and limits of ECoG in the context of epilepsy surgery, and for
the purpose of improving the accuracy of localizing
the epileptogenic region. Discussion of the mapping
of normal function to plan a resection, even for epilepsy surgery, is contained in Chapter 8 and portions
of other chapters in Section II.
34.1. Technical aspects
As a functional test, EEG is sensitive to the functional
changes of the cerebrum that many medications produce.
This is true for both scalp EEG and ECoG and it is especially important for ECoG because of the anesthetics that
are necessary to obtain an ECoG. The anesthetic effect
on the ECoG depends upon the drug and its dose and
may be either an increase or decrease in amplitude, an
increase or decrease in background frequencies, or an
increase or decrease in epileptiform discharges (Sloan,
1998). The effect on epileptiform discharges is particularly important because the mapping of spontaneous epileptiform discharges is the goal of ECoG. A reduction in
discharge frequency obviously interferes with this goal.
The production of increased discharges also is problematic because of the possibility that such discharges may
be produced in patients without epilepsy. Therefore, they
may not be as reliable a marker of the true irritative zone.
Avoiding anesthetic effects on the ECoG is accomplished by the choice of anesthetic and minimizing the
anesthetic dose. In its strictest sense, minimizing the

anesthetic dose is the discontinuation of anesthesia
during the portion of the operation when the ECoG
is obtained. In such cases, the effects of the anesthetic
on the ECoG are clearly less important than the anesthetics clearance time. Propofol produces considerable suppression of EEG activity when at high doses
and may be used to treat status epilepticus. Although
initial reports indicated that it may produce epileptiform discharges, at least three subsequent prospective
series of patients (including one of patients only on
propofol) did not observe this to occur (Ebrahim
et al., 1994; Samra et al., 1995; Cheng et al., 1996).
However, more important for propofol use in the context of ECoG is its short half-life. Its use for such surgeries is well demonstrated in a retrospective case
series of 332 patients (Skucas and Artru, 2006). These
patients were anesthetized with propofol for craniotomies involving cerebral resections that included
planned awake assessments before resection and after
temporary discontinuation of the propofol. The propofol discontinuation procedure was well tolerated overall, as compared to 129 patients from the same time
period and hospital, who were maintained with general anesthesia throughout their craniotomies. A series
of 12 children found that spontaneous epileptiform
discharges were present on ECoG during the period
that propofol was discontinued, which more directly
supports propofol discontinuation prior to ECoG
(Soriano et al., 2000).
Physical or emotional discomfort is common
when sedation is discontinued for awake craniotomies, so a low level of anesthesia often is continued.
Nitrous oxide produces mild to moderate slowing of
ECoG background activity with an increase in theta
and delta frequency range activity and a decrease in
486
beta frequency range activity. However, it does not

significantly change the frequency or location of epileptiform discharges, as was observed in a study of
18 patients whose ECoGs on and off nitrous oxide
were reviewed blindly (Hosain et al., 1996). Individual
patients had differing responses, with either increases
or decreases in discharge frequency (Artru et al.,
1990; Hosain et al., 1996).
Opioid narcotics have clear potential to produce
epileptiform discharges on ECoG and the discharges
may have misleading localizations. This is well
demonstrated in a report of fentanyl induction of anesthesia producing seizures in eight of nine patients who
had completed bitemporal epidural EEG monitoring
(Tempelhoff et al., 1992). The EEG lateralization for
four of the eight seizures was contralateral to the spontaneous seizures. Fentanyl also produces increases in
interictal epileptiform discharge frequencies, as does
alfentanyl (Cascino et al., 1993; Manninen et al.,
1999). This increase has been considered helpful to
some authors because the discharge locations match
the locations of maximum, extraoperative discharge
production for limbic epilepsy. One series of 22
patients found that remifentanil also produces an
increase in epileptiform discharge frequency and does
so in the context of suppression of background activity, thereby making the discharges more apparent
(Wass et al., 2001). The significance of this and the
effects of fentanyl and alfentanyl on prognosis following epilepsy surgery by improving the identification of
the epileptogenic region has not been determined.
Remifentanil at a lower dose as a continuous infusion
does not have a significant effect on discharge frequency (Herrick et al., 2002).
Volatile anesthetics also may produce epileptiform discharges. Specifically, halothane, isoflurane,
and enflurane have been reported to produce seizures; however, much of the evidence arises from
older case reports (Smith et al., 1966; Poulton and
Ellingson, 1984; Hymes, 1985). In a more recent
report, which is a case series including 21 children,
ECoG under isoflurane anesthesia was compared to
the extraoperative grid monitoring performed without
sedation and the discharge frequency was lower with
isoflurane (Asano et al., 2004). This effect also was
observed in a series of four patients (Ito et al.,
1988). Lower doses of isoflurane were found not to
affect the epileptiform discharge frequency in a
series of 15 patients (Fiol et al., 1993). Enflurane also
has been observed to produce an increase in epileptiform discharge frequency (Ito et al., 1988). For both
J.M. STERN AND J. ENGEL JR.
anesthetics, the discharge locations were not affected

by the anesthesia. Sevoflurane has been assessed
more systematically through at least two case series.
One series included 10 patients with temporal lobe
epilepsy who received sevoflurane in combination
with fentanyl during ECoG for epilepsy surgery
(Endo et al., 2002). A greater than 50% decrease in
spontaneous epileptiform discharges was observed,
which was undesirable because of the ECoGs purpose. It is not known whether an interaction with
fentanyl is a factor in this effect. Another series
included 12 patients who received, at different times,
sevoflurane and isoflurane (Watts et al., 1999).
Sevoflurane was associated with a larger number of
epileptiform discharges, but it is unclear whether this
was due to a decrease in discharges from isoflurane.
34.2. Mapping of epileptic abnormality
The identification of epileptic abnormality during
ECoG is compromised by anesthetic effects described
above, the possibility that the resection performed during or before the ECoG is affecting the ECoG results,
the limits to the duration of recording that are imposed
by the practical aspects of the operation, the uncertainty for whether visible epileptic abnormality could
be propagated from a region that is not visible, and
the imprecise relationship between the irritative zone
and the epileptogenic zone (Kuruvilla and Flink,
2003). Despite these limitations, ECoG may provide
critical real-time functional information that commonly leads to modification of the resection (Figs. 2
and 3). This is demonstrated by its use for more than
60 years and at a substantial majority of surgical epilepsy centers. However, a controlled trial of its application has not been conducted and the best evidence
for ECoG utility is found in case series.
Epileptiform discharges that are observed at the
margins of a cerebral resection, especially a resection
that has been performed during the same operation as
the ECoG, commonly are viewed as possible injury
potentials or injury spikes. As such, they may be
due to the acute trauma of the resection and not
related to the epileptic abnormality that the resection
is attempting to eliminate. Moreover, these discharges are not believed to indicate the creation of
a new epileptic abnormality as they are an acute reaction. Evidence for the possibility of injury spikes in
the absence of epilepsy has been reported in a series
of seven patients without histories of epilepsy who
underwent resection of neocortical tumors (Schwartz
487
Fig. 2. ECoG including a 45 electrode grid (first 20 channels) and a 4-electrode strip. Three interictal epileptiform discharges occur across the strips channels without simultaneous appearance in the grid (1 s between solid lines, lowfrequency filter (LFF) 1 Hz, high-frequency filter (HFF) 50 Hz).
et al., 2000). Preresection ECoG identified epileptiform abnormality in two patients, which was not
present after resection. Two other patients had epileptiform discharges only after resection, suggesting
the occurrence of injury spikes. Neither of these

patients developed seizures. This series also demonstrates the limits of the specificity of epileptiform
discharges for epilepsy.
Fig. 3. ECoG including a 45 electrode grid (first 20 channels) and a 4-electrode strip. Multiple interictal epileptiform discharges are present broadly across the grids electrodes with best representation in the first four channels. The grids discharges are independent to and distant from the strips interictal epileptiform discharges, which may have prognostic
significance for seizure control following resection (1 s between solid lines, low-frequency filter (LFF) 1 Hz, highfrequency filter (HFF) 50 Hz).
488
Although limited in specificity, multiple case series

have demonstrated the value of epileptiform discharges on ECoG for prognostication of seizure freedom and defining the boundaries of epileptic
abnormality. One series includes 25 patients with
benign brain tumors and epilepsy who underwent a
uniform protocol that included ECoG to determine
whether to extend the resection beyond the tumor
and include the neighboring gyrus (Mikuni et al.,
2006). The five patients without adjacent epileptiform
abnormality underwent a lesionectomy and became
seizure-free. Of the remaining 20 patients, 18 had
adjacent ECoG abnormality and underwent resection
of the cortical margins with resultant seizure freedom
for 17 of them. Cortical dysplasia was identified in
neighboring tissue for seven of these patients. The
remaining two patients had distant ECoG abnormality.
Another series including uniform patient population
further supports the benefits of ECoG through a collection of 12 patients with gangliogliomas (Pilcher
et al., 1993). Resection was extended to include
regions identified by ECoG to be epileptic and the seizure control rate was larger than the expected (historical) control. Because neither study included controls,
the actual value of the ECoG remains unclear.
The observational results in these two reports contrast strongly with the results of a comparison study.
A series of 36 patients with cerebral tumors and epilepsy agreed to have preresection and postresection
ECoG without tailoring of the resection based upon
the ECoG findings (Tran et al., 1997). When the
ECoG results were compared to whether the patient
became seizure-free following resection, there was
no significant correlation between seizure control and
the presence of epileptiform discharges at the boundaries of the resection. However, this result may be
partly due to the extent of the marginal discharges.
A series of 22 patients with lesional (mixed pathology)
frontal lobe epilepsy were found to have the largest
likelihood of postresection seizure control when the
entire lesion was removed and the preresection epileptiform discharges were limited to within two gyri of
the lesion (Wennberg et al., 1999). A similar finding
was observed in a series of 60 patients with nontumoral frontal lobe epilepsy, in which epileptiform discharges that were three or more gyri away from the
resection site indicated a low likelihood of seizure
freedom (Wennberg et al., 1998).
Although mesial temporal lobe resections commonly are done without ECoG, a center that routinely uses ECoG to determine limits reviewed 140
consecutive patients to determine whether postoperative seizure freedom was associated with either the
presence of postresection epileptiform discharges or
the extent of resection (McKhann et al., 2000). The
result was that postresection epileptiform discharges
were significantly associated with incomplete seizure
control, but the size of the hippocampal resection
(less than or greater than 2.5 cm) was not associated
with this outcome. This suggested to the authors that
ECoG allowed some patients to have a more limited
resection. Similar tailoring of the temporal resection
without compromising seizure control rates also was
observed to be possible in a series of 24 patients at
a different center (Kanner et al., 1995). Whether tailoring the resection truly was beneficial would
require a controlled study. However, a comparison
study contrasts strongly with these results. A series
of 47 patients with mesial temporal lobe epilepsy
had preresection and postresection ECoG during an
operation that included an anatomically defined anterior temporal lobe resection (Tran et al., 1995). Neither the presence of epileptiform discharges outside
of the resection region before or after resection was
associated with seizure control postoperatively.
A similar finding was observed in series of 87 and
29 patients at different centers (Kanazawa et al.,
1996; Schwartz et al., 1997). Essentially, the tailoring of temporal lobe resections has not been shown
to be clearly indicated.
For less structurally discrete pathological substrates for epilepsy, resective approaches require
even greater individualization, which ECoG may
provide. The utility of ECoG has been described in
case reports for large porencephalic cysts, subcortical
ectopic gray matter, schizencephaly, and Sturge
Weber syndrome (Maehara et al., 1997; Hata et al.,
1998; Morino et al., 2004; Iida et al., 2005). These
reports further the support for ECoG as a valuable
tool, but do not add to the evidence-based understanding of ECoGs role and limits. As epilepsy
surgery continues to be studied in increasingly
sophisticated clinical investigations, ECoG also
should be more systematically investigated to determine its true benefits and best applications.
References
Almeida, AN, Martinez, V, et al. (2005) The first case of
invasive EEG monitoring for the surgical treatment of
epilepsy: historical significance and context. Epilepsia,
46(7): 10821085.
Artru, AA, Lettich, E, et al. (1990) Nitrous oxide: suppression of focal epileptiform activity during inhalation, and
spreading of seizure activity following withdrawal.
J. Neurosurg. Anesthesiol., 2(3): 189193.
Asano, E, Benedek, K, et al. (2004) Is intraoperative electrocorticography reliable in children with intractable
neocortical epilepsy? Epilepsia, 45(9): 10911099.
Cascino, GD, So, EL, et al. (1993) Alfentanil-induced epileptiform activity in patients with partial epilepsy.
Cheng, MA, Tempelhoff, R, et al. (1996) Large-dose propofol
alone in adult epileptic patients: electrocorticographic
results. Anesth. Analg., 83(1): 169174.
Doyle, WK and Spencer, DD (1997) Anterior temporal
resections. In: J Engel Jr. and T Pedley (Eds.), Epilepsy: A Comprehensive Textbook. Lippincott-Raven
Publishers, Philadelphia. Vol. 2, pp. 18071817.
Ebrahim, ZY, Schubert, A, et al. (1994) The effect of propofol on the electroencephalogram of patients with epilepsy. Anesth. Analg., 78(2): 275279.
Endo, T, Sato, K, et al. (2002) Effects of sevoflurane on
electrocorticography in patients with intractable temporal
lobe epilepsy. J. Neurosurg. Anesthesiol., 14(1): 5962.
Engel, J, Jr., Wiebe, S, et al. (2003) Practice parameter:
temporal lobe and localized neocortical resections for
epilepsy. Epilepsia, 44(6): 741751.
Fiol, ME, Boening, JA, et al. (1993) Effect of isoflurane
(Forane) on intraoperative electrocorticogram. Epilepsia, 34(5): 897900.
Hata, D, Isu, T, et al. (1998) Intraoperative electrocorticography and successful focus resection in a case of
Sturge-Weber syndrome. Seizure, 7(6): 505508.
Herrick, IA, Craen, RA, et al. (2002) Sedative doses of
remifentanil have minimal effect on ECoG spike activity during awake epilepsy surgery. J. Neurosurg.
Anesthesiol., 14(1): 5558.
Hosain, S, Nagarajan, L, et al. (1996) Effects of nitrous
oxide on electrocorticography during epilepsy surgery.
Hymes, JA (1985) Seizure activity during isoflurane anesthesia. Anesth. Analg., 64(3): 367368.
Iida, K, Otsubo, H, et al. (2005) Cortical resection with electrocorticography for intractable porencephaly-related partial epilepsy. Epilepsia, 46(1): 7683.
Ito, BM, Sato, S, et al. (1988) Effect of isoflurane and
enflurane on the electrocorticogram of epileptic
patients. Neurology, 38(6): 924928.
Kanazawa, O, Blume, WT, et al. (1996) Significance of
spikes at temporal lobe electrocorticography. Epilepsia,
37(1): 5055.
Kanner, AM, Kaydanova, Y, et al. (1995) Tailored anterior
temporal lobectomy. Relation between extent of resection
of mesial structures and postsurgical seizure outcome.
Arch. Neurol., 52(2): 173178.
489
Kuruvilla, A and Flink, R (2003) Intraoperative electrocorticography in epilepsy surgery: useful or not? Seizure,
12(8): 577584.
Maehara, T, Shimizu, H, et al. (1997) Surgical treatment of
epilepsy from schizencephaly with fused lips. Surg.
Neurol., 48(5): 507510.
Manninen, PH, Burke, SJ, et al. (1999) Intraoperative localization of an epileptogenic focus with alfentanil and
fentanyl. Anesth. Analg., 88(5): 11011106.
McKhann, GM, 2nd, Schoenfeld-McNeill, J, et al. (2000)
Intraoperative hippocampal electrocorticography to predict the extent of hippocampal resection in temporal
lobe epilepsy surgery. J. Neurosurg., 93(1): 4452.
Mikuni, N, Ikeda, A, et al. (2006) A step-by-step resection
guided by electrocorticography for nonmalignant brain
tumors associated with long-term intractable epilepsy.
Epilepsy Behav., 8(3): 560564.
Morino, M, Ishibashi, K, et al. (2004) Surgical treatment of
temporal lobe epilepsy associated with subcortical
ectopic gray matter under the guidance of intraoperative
electrocorticography. Seizure, 13(7): 470474.
Penfield, W (1939) The epilepsies: with a note on radical
therapy. N. Engl. J. Med., 221: 209218.
Pilcher, WH, Silbergeld, DL, et al. (1993) Intraoperative
electrocorticography during tumor resection: impact on
seizure outcome in patients with gangliogliomas.
J. Neurosurg., 78(6): 891902.
Poulton, TJ and Ellingson, RJ (1984) Seizure associated
with induction of anesthesia with isoflurane. Anesthesiology, 61(4): 471476.
Samra, SK, Sneyd, JR, et al. (1995) Effects of propofol
sedation on seizures and intracranially recorded epileptiform activity in patients with partial epilepsy. Anesthesiology, 82(4): 843851.
Schwartz, TH, Bazil, CW, et al. (1997) The predictive
value of intraoperative electrocorticography in resections for limbic epilepsy associated with mesial temporal sclerosis. Neurosurgery, 40(2): 302309; discussion
309311.
Schwartz, TH, Bazil, CW, et al. (2000) Do reactive postresection injury spikes exist? Epilepsia, 41(11):
14631468.
Skucas, AP and Artru, AA (2006) Anesthetic complications
of awake craniotomies for epilepsy surgery. Anesth.
Analg., 102(3): 882887.
recordings. J. Clin. Neurophysiol., 15(3): 217226.
Smith, PA, MacDonald, TR, et al. (1966) Convulsions
associated with halothane anesthesia. Anaesthesia, 21:
229233.
Soriano, SG, Eldredge, EA, et al. (2000) The effect of propofol
on intraoperative electrocorticography and cortical stimulation during awake craniotomies in children. Paediatr.
Anaesth., 10(1): 2934.
490
Tempelhoff, R, Modica, PA, et al. (1992) Fentanyl-induced
electrocorticographic seizures in patients with complex
partial epilepsy. J. Neurosurg., 77(2): 201208.
Tran, TA, Spencer, SS, et al. (1995) Significance of spikes
recorded on electrocorticography in nonlesional medial
temporal lobe epilepsy. Ann. Neurol., 38(5): 763770.
Tran, TA, Spencer, SS, et al. (1997) Significance of spikes
recorded on intraoperative electrocorticography in patients
with brain tumor and epilepsy. Epilepsia, 38(10):
11321139.
Wass, CT, Grady, RE, et al. (2001) The effects of remifentanil on epileptiform discharges during intraoperative

electrocorticography in patients undergoing epilepsy
surgery. Epilepsia, 42(10): 13401344.
Watts, AD, Herrick, IA, et al. (1999) The effect
of sevoflurane and isoflurane anesthesia on interictal
spike activity among patients with refractory epilepsy. Anesth. Analg., 89(5): 12751281.
Wennberg, R, Quesney, LF, et al. (1998) Electrocorticography and outcome in frontal lobe epilepsy. Electroencephalogr. Clin. Neurophysiol., 106(4): 357368.
Wennberg, R, Quesney, LF, et al. (1999) Role of electrocorticography at surgery for lesion-related frontal lobe
epilepsy. Can. J. Neurol. Sci., 26(1): 3339.

M.R. Nuwer (Ed.)
491
CHAPTER 35
Intraoperative neurophysiology during surgery

for cerebral tumors
Hugues Duffau*
Department of Neurosurgery, Hopital Gui de Chauliac, CHU Montpellier, 80 avenue Augustin Fliche,
34295 Montpellier, France
35.1. Introduction
Extensive surgical removal of cerebral tumors, when
possible, currently appears to be the first treatment to
influence the natural history of this disease (Berger
et al., 1994; Keles et al., 2001; Lacroix et al., 2001;
Laws et al., 2003; Claus et al., 2005; Duffau et al.,
2005a).The antagonist goal of surgery is to maximize
the quality of resection, on the one hand, and on
the other, to minimize the operative risk (Duffau,
2006). However, due to the frequent location of supratentorial tumors near or within the so-called eloquent areas (Duffau and Capelle, 2004), and due to
their infiltrative feature (i.e., poorly demarcated, especially with regard to the gliomas), it was believed, for a
long time, that the chances of performing a maximal
tumor removal were low, while the risk to generate
postoperative sequelae was high; many surgical series
have reported a rate between 15% and 25% of permanent and severe deficit, following the removal of
intra-axial tumors (Fadul et al., 1988; Apuzzo, 1993;
Tandon et al., 1993; Cabantog and Bernstein, 1994;
Cedzich et al., 1996; Sawaya et al., 1998; Vives and
Piepmeier, 1999; Brell et al., 2000). As a consequence,
the indications of surgical resection of brain tumors are
still matter of debate (Whittle, 2004).
In order to optimize the benefit-to-risk ratio of the
surgery, the use of functional mapping methods was
proposed in the last decade, due to the considerable
interindividual anatomofunctional variability demonstrated in healthy volunteers (Tzourio-Mazoyer et al.,
2004), and its usage increased in cases of brain lesion
*
Correspondence to: Hugues Duffau, M.D., Ph.D., Department of Neurosurgery, Hopital Gui de Chauliac, CHU
Montpellier, 80 Avenue Augustin Fliche, 34295 Montpellier, France.
Tel.: 33-4-67-33-66-12; fax: 33-4-67-33-75-57.
E-mail: h-duffau@chu-montpellier.fr (H. Duffau).
(Duffau et al., 2003a). The goal is to study the

organization of the cortical eloquent areas for each
patient, then to tailor the resection according to both
oncological and functional boundaries (Duffau,
2006). The development of preoperative noninvasive
techniques of functional neuroimaging, including positron emission tomography (PET) (Meyer et al., 2003),
functional magnetic resonance imagning (fMRI)
(Vlieger et al., 2004), and magnetoencephalography
(MEG)/EEG (Orrison, 1999; Schiffbauer et al.,
2002), has allowed researchers to perform cortical
mapping of the whole brain. Such individual data have
enabled an estimation of the location of the eloquent
areas in relation to the glioma in order to better select
the surgical indications. This depends on the feasibility
of tumor resection (Ganslandt et al., 2004), as well as
the best surgical approach and the limits of resection,
which both aim to avoid any damage to the functional
cortex, and thus decrease the surgical risk. Nevertheless, accumulating evidence indicates that the reliability of these cortical mapping methods is still not
optimal, despite constant effort for its improvement
(sensitivity ranging from 82% to 100% for the identification of sensorimotor sites and from 66% to 100% for
the idenfication of the language areas) (Roux et al.,
2003a). In particular, it seems that the BOLD response
in the vicinity of brain tumors does not reflect the neuronal signal as accurately as it does in healthy tissue
(Holodny et al, 2000). This is likely due to an alteration
of neurovascular and metabolic coupling (Aubert
et al., 2002). Furthermore, these methods are not able
to differentiate the areas essential for the function,
which should be surgically preserved, from the modulatory areas, which can be functionally compensated
and so potentially resected without permanent deficit
(Duffau et al., 2003a).
Intraoperatively, the integration of PET, fMRI,
MEG, or even multimodal imaging into frameless
492
stereotactic surgery has been used extensively and

they are often referred to as functional neuronavigation (Nimsky et al., 2004). However, despite an accuracy evaluated approximately between 1 and 4 mm, it
is important to be careful with image-guided surgery
for voluminous tumors. The caution is necessary
because of the high risk of intraoperative brain shift,
due to surgical retraction, mass effect, gravity, and
the extent of the resection or cerebrospinal fluid leakage (Reinges et al., 2004). The following are the reasons why several technical improvements have been
recently proposed to reduce the effects of this brain
shift: (1) the combination with data from intraoperative ultrasound, which produces real-time imaging
(Keles et al., 2003); (2) the use of mathematical models and computational methods based on data from
intraoperative ultrasound or digital images that track
cortical displacement and movements of known landmarks (Platenik et al., 2002); (3) the recent development of intraoperative MRI which has permitted the
acquisition of high-quality, multiplanar, and real-time
imaging during surgery (Keles, 2004).
As a consequence, despite the development of
functional neuroimaging, but due to its current limitations, the additional use of invasive electrophysiological investigations has been advocated for surgery
of tumors located near, or within, eloquent structures
(Berger et al., 1994; Duffau et al, 2005a).
The goal of this chapter is to review how these
methods of intraoperative testing and monitoring
(IOM) can help in surgery for cerebral tumors,
namely, their contribution regarding both functional
and oncological results. The future developments
of the methods and their fundamental applications
will also be considered.
35.2. IOM methodology: advantages and
limitations
First, the technique of somatosensory evoked potentials (SEPs) was used extensively for the resection
of tumors located within the central region (Cedzich
et al., 1996; Romstock et al., 2002). Nonetheless, its
reliability is not optimal, since the accurate localization of the central sulcus is reported with a frequency
of only between 91% and 94% (Cedzich et al., 1996;
Romstock et al., 2002). In a recent review, the
overall sensitivity and negative predictive value of
SEP monitoring was 79% and 96%, respectively
(Wiedemayer et al., 2004). Moreover, phase reversal
recording identifies only the central sulcus itself, but
H. DUFFAU
offers no direct information on the particular distribution of motor function on the adjacent exposed
cerebral structures. Also, whereas the method of
motor evoked potentials (MEPs) improved (Kombos
et al., 2001; Neuloh and Schramm, 2004) when
recording compound muscle action potentials, only
the monitored muscles can be controlled; that is, it
is difficult to detect, and therefore avoid, motor deficits in nonmonitored muscles. Finally, intraoperative
evoked potentials cannot currently be used to map
language, memory, or other higher functions quickly
and reliably.
Other authors are also prone to use extraoperative electrophysiological recordings and stimulations
through the implantation of subdural grids (Schaffler
et al., 1996; Ikeda et al., 2002). Using this method,
the patient is in his room under optimal conditions,
to perform the tasks. This point is particularly important for children (Chitoku et al., 2001). Moreover,
recent advances in the interpretation of the electrophysiological signal, such as electrocorticographic
spectral analysis evaluating the event-related synchronization in specific band of frequency (Crone,
2000), have allowed a better understanding of the
organization of the functional cortex, and a study of
the connectivity, in particular via the recording of
cortico-cortical evoked potential (Matsumoto et al.,
2004). However, extraoperative electrophysiological
mapping usually used grids with 1-m-spaced electrodes, thus with a limited accuracy. Also, it is necessary to perform two surgical procedures: one to
implant grids, and another to resect the lesion. Finally,
there is still a risk of infectious complications due to
the presence of subdural grids over several days
(Onal et al., 2003).
Taking into account the advantages and limits of
these different mapping techniques, more and more
neurosurgeons advocate the additional use of intraoperative direct electrical stimulation (DES) under
general or local, anesthesia during surgery of tumors
in eloquent areas (Berger et al., 1990, 1994; Duffau
et al., 1999, 2005a; Taylor and Bernstein, 1999;
Danks et al., 2000; Meyer et al., 2001; Signorelli
et al., 2001; Whittle et al., 2003; Peraud et al.,
2004; Picht et al., 2006). A bipolar electrode with
tips, spaced 5 mm apart and delivering a biphasic
current (pulse frequency 60 Hz, pulse duration 1 ms,
and amplitude from 6 to 18 mA under general anesthesia or from 2 to 6 mA under local anesthesia), is
applied to the cortex. DES allows the mapping of
motor function (possibly under general anesthesia,
by inducing unvoluntary motor response, if the

stimulation is applied at the level of a motor site).
It allows somatosensory function by eliciting dysesthesia described by the patient himself intraoperatively, and also the mapping of cognitive functions.
Examples of the functions that can be mapped include
language (spontaneaous speech, oject naming, comprehension, reading, writing, bilinguism, etc. . . .), calculation, memory, or even visualspatial processing,
performed in these cases on awake patients by generating transient disturbances when the electrical stimulation is applied at the level of a functional epicenter
(Ojemann et al., 1989). It is important that a speech therapist be present in the operative room, in order to
interpret the kind of disorders induced by DES, for
instance speech arrest, anarthria, speech apraxia,
phonological disturbances, semantic paraphasia, perseveration, anomia, dysculia, and others. Thus, DES
is able to identify, in real time, the cortical sites
essential for the function (i.e., to be imperatively preserved), following the dura-matter opening, but
before the beginning of the resection, in order to
select the best surgical approach and to define the
cortical limits of the glioma removal (Duffau et al.,
2005a).
In addition, it is possible to use DES to identify
and to preserve not only eloquent cortical sites but
also subcortical pathways that are indispensable for
sensorimotor, language, and other cognitive functions
(Duffau et al., 2003b; Duffau, 2005a). DES allows
the study of the anatomicalfunctional connectivity
by directly stimulating the white matter tracts very
regularly all along the resection, and by eliciting
functional response when in contact with the essential eloquent fibers, following a similar principle as
the one described at the cortical level. Thus, except
for motor mapping, it is important that the patient
remains awake during the entire duration of the
glioma removal, and not exclusively for the initial
stage of cortical simulation. This is the reason why
intraoperative cortico-subcortical stimulation is time
consuming, with a limitation of the number of tasks
that can be used during surgery. The progressive
tiredness of the patient is a hindrance.
In conclusion, DES represents an accurate, reliable, and safe technique of online detection of the
cortical and subcortical regions essential for function, at each place and each moment of the resection. Consequently, in all cases, a functional
disturbance induced by stimulation, with reproducibility, must interrupt resection at this level, to
493
safeguard both cortical and subcortical structures.

The tumor removal is then performed according to
functional boundaries, in order to optimize the quality of tumor removal while minimizing the risk of
permanent postoperative deficit (Duffau et al.,
2005a, 2005b).
35.3. Contribution of IOM in brain tumor surgery
35.3.1. New insights into the pathophysiology of
functional areas invaded by tumors
IOM have been used to better understand the pathophysiology of brain areas regularly affected by cerebral tumors, in order to better plan resection of these
eloquent regions.
35.3.1.1. Anatomicalfunctional organization of the
supplementary motor area
The supplementary motor area (SMA), namely the
frontomesial area located in front of the primary
motor area of the inferior limb, is a region involved
in the planning of movement (Freund, 1996). Its
resection induces the classical SMA syndrome
(Laplane et al., 1977), which is characterized by a
complete akinesia and even a mutism (in cases of
lesions of the left dominant SMA), and which occurs
30 min following the end of the resection, as
observed in awake patients (Duffau et al., 2001a).
Around the tenth day following surgery, this syndrome suddenly and spontaneously resolves, though
some rehabilitation is often needed during 1
3 months to provide for a complete functional recovery. Interestingly, using IOM combined with fMRI, it
was shown that the occurrence of this syndrome is
not associated with the volume of the frontal resection, but it is directly related to the removal of the
specific structure called the SMA-proper (Krainik
et al., 2001, 2003). Moreover, the existence of a
somatotopy within the SMA-proper was demonstrated namely, from anterior to posterior: the representation of language (at least in the dominant
hemisphere), the one of the face, then the superior
limb, then the inferior limb (immediately in front of
the paracentral lobule) (Fontaine et al., 2002). It is
now possible to predict, before surgery, if a SMA
syndrome will occur postoperatively, the severity
and pattern of this transient postoperative deficit
(e.g., only mutism, or mutism and akinesia of the
superior limb, or akinesia of the entire hemibody).
This has an important impact concerning the
494
information of the patient and his family, and regarding the planning of specific rehabilitation.
35.3.1.2. Role of dominant insula in language
The insular lobe also represents a structure frequently
invaded by tumors, especially low-grade gliomas
(Duffau and Capelle, 2004). Although poorly studied
for a long time for technical reasons, it seems that the
insula is an anatomical, cytoarchitectonic and functional interface between the allocortex and the neocortex (Augustine, 1996). Recent developments in
functional mapping methods have enabled a better
understanding of the implication that this multimodal
lobe has on many functions, in particular, on language (Ackermann and Riecker, 2004). In brain
tumor surgery, IOM has shown a participation of
the anterior insular cortex in the dominant hemisphere during language tasks, as reported in healthy
volunteers using neurofunctional imaging (Wise
et al., 1999). More specifically, DES has elicited
articulatory disturbances when applied on the insular
cortex (Duffau et al., 2000a, 2002a; Duffau and
Fontaine, 2005). This finding supports the role of this
structure in the complex planning of speech, as previously suggested in stroke studies (Dronkers, 1996).
These data have important implications for the selection of surgical indications, since in left-dominant,
frontotemporalinsular gliomas, resection has a large
risk of being incomplete (Duffau et al., 2006).
35.3.1.3. Role of premotor cortex in language
Another region frequently involved by tumors is
the premotor cortex. Although many studies have
allowed for a better clarification of the implication
of this structure in motor function (Picard and Strick,
2001), its participation in language remains poorly
understood. Interestingly, it has been demonstrated
using IOM that DES of the dominant dorsal premotor
area (namely the structure lateral to the SMA, in
front of the primary motor area of the hand), induced
anomia when stimulated. On the other hand, stimulation of the dominant ventral premotor cortex regularly elicited articulatory disorders. It is imperative,
therefore, that this area be preserved, to avoid postoperative anarthria (Duffau et al., 2003c).
35.3.1.4. Role of angular gyrus in calculation
The supramarginal and angular gyri, in the left hemisphere, are known to participate in complex cognitive
functions, such as calculation (Dehaene et al., 2004).
In patients harboring left posterior parietal gliomas,
H. DUFFAU
both multiplication and subtraction have been tasked

during electrical mapping (Duffau et al., 2002b).
Interestingly, it was found that functional epicenters
more involved in arithmetic facts such as rote multiplication, with tables learnt by heart, were located
immediately above the posterior end of the sylvian
fissure, thus very close to the language sites. On the
other hand, actual calculation such as subtraction
recruited functional sites located in the superior
part of the angular gyrus, immediately below the
intraparietal sulcus, close to the areas involved in
working memory (Duffau et al., 2002b). Despite a
transient dyscalculia following surgery, all patients
recovered.
35.3.1.5. Functional organization of Wernickes area
Concerning tumors located within dominant temporal
posterior areas, a task adapted to test comprehension
during intraoperative mapping has been developed
(Gatignol et al., 2004). A triad of picture is shown,
and the patient is asked to pair them by naming
two pictures with conceptual links for instance, a
pyramid and a palm tree. Interestingly, several sites
within the posterior part of the superior temporal
gyrus regularly elicited specific anomia without any
comprehension disturbances when stimulated, though
other sites within the same gyrus elicited only comprehension disorders with the preservation of the
ability to name. Moreover, DES generated phonological disturbances at the end of the resection. These
results supported the complexity of the functional
organization of Wernickes area previously studied
using extraoperative stimulations (Boatman et al.,
2000), with its participation with but also possible
dissociations among comprehension, naming, and
phonological processings (Gatignol et al., 2004).
35.3.1.6. Functional organization of Brocas area
Using both preoperative neurofunctional imaging and
IOM, it was shown that Brocas area was, in fact
implied in several language processings. Its posterior
part, the pars opercularis, was more involved in phonological processing; its superior part, pars triangularis,
was implicated in syntactic processing. Its anterior
part, pars orbitaris, was more involved in a large
semantic network underlied by the inferior frontooccipital fasciculus (Fig. 1; Duffau et al., 2005b).
35.3.1.7. Frontal eye fields
In patients with precentral tumors, the functional anatomy of the frontal eye field has also been studied,
Fig. 1. A: Preoperative anatomical magnetic resonance imaging (MRI) showing a low-grade glioma which involves the left
inferior fontal gyrus, namely the Brocas area, in front of the rolandic operculum the arrow marks the central sulcus
(from Duffau et al., 2002c). B: Intraoperative views before resection of the tumor (star), which is delineated by letter tags.
Electrical cortical mapping shows a reshaping of the eloquent maps, with a recruitment of perilesional language sites, that is
the ventral premotor cortex (tags 2325), dorsolateral prefrontal cortex (tag 29), and pars orbitaris of the inferior frontal
gyrus (tags 30 and 31). This language remapping induced by the slow-growing low-grade glioma has allowed the compensation of the Brocas area. Therefore, a resection of this tumor can be envisoned despite its location in classical inoperable region. The arrow shows the lateral part of the central sulcus. A anterior, P posterior. C: Intraoperative views
after resection of the tumor. Electrical subcortical mapping has enabled to study the individual anatomicalfunctional language connectivity. Indeed, deeply and anteriorly, the anterior part of the inferior fronto-occipital fasciculus was identified,
by eliciting systematic semantic paraphasia during each stimulation (tag 50) (from Duffau et al., 2005b). More posteriorly,
the insulo-frontal connections and the anterior part of the arcuate fasciculus were detected, by inducing speech production
disorders (phonemic paraphasia or articulatory disturbances) (tag 40) (from Duffau et al., 2002c). It is worth noting that in
the depth, the resection was continued up to the contact of these language pathways, in order to optimize the quality of
resection while preserving the eloquent white matter tracts. The arrow shows the lateral part of the central sulcus. A anterior, P posterior. D: postsurgical control MRI, showing a complete glioma removal, performed according to individual
cortico-subcortical functional boundaries given by intraoperative testing and monitoring (IOM), in a patient with a normal
neurological and neuropsychological status. In conclusion, the resection was adapted to the cortical plasticity and the
language connectivity, with the goal to optimize the benefit/risk ratio of the surgery.
496
using IOM (Lobel et al., 2001; Blanke and Seeck,

2003). This is a region located laterally to, and in front
of, the primary motor area of the face, and it is
involved in the regulation of the voluntary and unvoluntary ocular saccades (Milea et al., 2002). The resection of this area is possible, inducing a transient deficit,
which completely recovers within weeks.
35.3.1.8. Role of the right supramarginal gyrus and
posterior temporal areas in spatial awareness
Recently, a line bisection task was used during awake
surgery in patients with tumors that involved the
right parietotemporal junction. The aim was to map
the areas involved in the spatial awareness (Thiebaut
de Schotten et al., 2005). A significant rightward
deviation was observed during the stimulation of
the anteroinferior part of the supramarginal gyrus
and the caudal part of the superior temporal gyrus.
In other words, a transient and reproducible leftneglect was elicited by electrical inactivation of
cortical sites essential for visualspatial integration,
at the level of the right parietotemporal junction.
These eloquent areas were preserved, and the patients
showed no signs of neglect, when examined a few
days after surgery. These findings demonstrate that
the supramarginal gyrus and the caudal part of the
superior temporal gyrus, at least in the right hemisphere, are critical to the symmetrical processing of
the visual scene in humans (Thiebaut de Schotten
et al., 2005).
Interestingly, other anatomicalfunctional correlations have also been made IOM in patients undergoing operations for brain lesions, in particular,
concerning writing (Roux et al., 2003b), reading
(Roux et al., 2004), bilinguism (Roux and Tremoulet,
2002; Lucas et al., 2004), or even control of micturition (Duffau and Capelle, 2005).
In summary, IOM enables the neurosurgeon to
adapt his surgical strategy to the knowledge of the
functional organization of the brain, and to apply it
to each patient. This isespecially important regarding
the selection of functional tasks, in order to optimize
the reliability of the intraoperative mapping and the
precise boundaries of a resection.
35.3.2. Study of the individual anatomical
functional connectivity
Tumors, especially gliomas, invade both subcortical
and cortical structures, and moreover, show an infiltrative progression along the fibers (Mandonnet
H. DUFFAU
et al., 2006). Because of this, definitive deficits

may occur because of surgical damage to pathways
running in the white matter (Coenen et al., 2001;
Lang et al., 2001). It is now necessary to incorporate
methods of subcortical mapping in addition to the
IOM techniques, allowing the study of the sole cortical functional organization. As a consequence, in
order to avoid postoperative sequelae, the study of
individual anatomicalfunctional connectivity underlying the networks between the eloquent areas is also
mandatory (Duffau, 2005b, 2006).
35.3.2.1. Subcortical motor pathways
For patients harboring a precentral tumor, after detection and preservation of the primary motor cortical
areas using IOM, it is also possible to detect the
corresponding descending motor pathways, using
subcortical DES, and their somatotopy that is the
different fibers of the corona radiata, with the pyramidal bundles of the lower limb medially, and laterally, with the bundles of the upper limb and face
(Duffau et al., 2003b). As at the cortical level, these
subcortical motor fibers constituted the posterior
and deep functional limits of the resection.
The pyramidal pathways may also be identified
at the level of the posterior limb of the internal
capsule. In particular, they can be identified in cases
of frontotemporalinsular glioma, in which the deep
boundaries of the resection were given when subcortical
stimulations induced motor responses in the inferior
part of the corona radiata up to the superior part of the
mesencephalic peduncles (Duffau et al., 2000a, 2002a).
35.3.2.2. Subcortical somatosensory pathways
Similarly, the thalamocortical somatosensory pathways and their somatotopy can be identified by subcortical DES, which induces dysesthesias in awake
patients, in cases of retrocentral tumors (Duffau
et al., 2003b).
35.3.2.3. Subcortical visual pathways
Subcortical visual pathways can also be mapped in
cases of patients undergoing operations in awake conditions for temporal tumors, with induction of a transient
shadow in the controlateral visual field during stimulation of the posterosuperior and deep part of the surgical cavity, sometimes also with metamorphopsia (i.e.,
visual illusion) (Duffau et al., 2004a). Thus, if resection
is stopped at this level, patients keep only a residual
quadrantanopsia, without any consequence to their
quality of life.
35.3.2.4. Subcortical language pathways

Interestingly, it is also possible to map the individual
anatomicalfunctional connectivity of language using
IOM.
Frequently, in patients with left dominant precentral gliomas, after identification of the motor and
language cortical sites within the prerolandic and
inferior frontal gyri (Brocas area), IOM has enabled
detection of different language pathways (Duffau
et al., 2002c). Medially, the fasciculus subcallosal
medialis (running from the SMA and cingulate gyrus
to the head of the caudate nucleus) was identified, by
eliciting transient transcortical motor aphasia during
its stimulation (this tract is involved in the initiation
of language). Posteriorly, the fibers coming from
the premotor ventral cortex have been detected,
inducing anarthria when stimulated; this pathway is
implied in speech production (Duffau et al., 2003c).
More laterally, the operculo-insular connections have
also been detected, by generating a complete speech
arrest during stimulation; these connections are
involved in speech planning (Duffau et al., 2002c).
In all cases, these language bundles have systematically constituted the subcortical functional limits of
the resection.
In addition to these locoregional language pathways, IOM is able to detect at least a part of longdistance association pathways, such as the medial
superior longitudinal fasciculus namely, the arcuate fasciculus (AF). In patients with tumors involving
the left insula and/or left inferior frontal gyrus, it was
possible to identify the anterior part of AF, located
within the anterior floor of the external capsule
(under the superior part of the insula), and also under
the posterior part of Brocas area (BA 44 and 45).
Stimulation has elicited transient symptoms classically observed in conduction aphasia, associating
phonemic paraphasia and repetition disorders. In the
same way, the AF can be detected at the level of its
posterosuperior loop, located under the supramarginal gyrus, in patients operated on for left posterior
parietal tumors. The same symptoms associating phonemic paraphasias and repetition disorders have been
induced during stimulation. Again, AF was detected
in posterior temporal tumors, with the posterior part
of its posterior funiculus corresponding to the anterior functional limit of the resection. Finally, the
anterior part of the anterior funiculus of AF has also
been used as the posterior functional boundary of left
dominant anterior and mid-temporal tumor (Duffau
et al., 2002c).
497
In addition to the AF, a lateral part of the superior

longitudinal fasciculus was also identified. In particular, in patients harboring a left retrocentral sus-sylvian
tumor, it was possible to detect not only the language
cortical sites at the level of both the ventral premotor
cortex, in front of the tumor and of the supramarginal
gyrus behind it, but also a frontalparietal subcortical
network, inducing speech apraxia when stimulated
(Duffau et al., 2003d). This operculo-opercular loop
might underly the anatomicalfunctional connectivity
of the working memory circuit. Interestingly, this
loop corresponds to the anterior segment of an indirect pathway of the classical superior longitudinal fasciculus, which runs parallel and lateral to the direct
pathway described above (i.e., the AF) by connecting
Brocas territory with the inferior parietal lobe.
This was recently demonstrated by diffusion tensor
imaging (DTI; Catani et al., 2005). It seems that
this tract might be involved in the vocalization of
semantic content.
More recently, in addition to this dorsal phonological root, arguments supporting the likely role
of the inferior fronto-occipital fasciculus (IFOF) in
the semantic system (i.e., ventral semantic root)
were provided using IOM (Duffau et al., 2005b).
In patients harboring a tumor that is invading the
frontal structures immediately in front of, and above,
Brocas area (namely, the pars orbitaris of the left
inferior frontal gyrus BA 47 and the dorsolateral prefrontal area BA 46/9), the anterior part
of the IFOF has been identified under these regions
by eliciting semantic paraphasias during subcortical
DES. In the same way, IFOF has also been detected,
by inducing the same symptoms (i.e., semantic paraphasias) when stimulated, in its intermediate part
located within the anterior floor of the internal capsule (in front of, and inferior to, the AF, and behind,
and superior to, the uncinate fasciculus), in patients
undergoing a left insular tumor operation. Again,
IFOF has been detected in left temporal tumors, by
eliciting semantic disturbances when stimulated,
and has constituted the deep limit of the resection
(above the roof of the temporal horn of the ventricle).
Finally, intraoperative DES has also allowed the
mapping of the deep gray nuclei, sometimes invaded
by tumors. Indeed, stimulation of the head of the
dominant caudate in patients harboring a frontomesial glioma coming in contact with the striatum at
such depth has generated, with a high reproducibility, perseverations; for example, patients repeatedly name the previous item when presented with
498
the next item (Gil Robles et al., 2005). These results

give further testament to the inhibitory role of the
caudate in the control of cognition. It has also been
possible to map the lateral part of the dominant lentiform nucleus, at the end of the resection of insular
tumor. Lentiform stimulation has elicited anarthria,
supporting the apparent role of this structure in the
planning of articulation, along with the insula and
ventral premotor cortex (Duffau et al., 2001b).
35.3.2.5. Subcortical pathway involved in spatial
awareness
Using the line bisection task during an awake surgery
for a patient with a tumor located within the right
parietotemporal junction (as previously described at
the cortical level), the white matter tracts implied
in spatial processing was also mapped by IOM, in
order to avoid postoperative left-neglect (Thiebaut
de Schotten et al., 2005). During stimulation of
the superior fronto-occipital fasciculus, a significant
rightward deviation of more than 30% was observed,
that is, illustrating more significance than during
the cortical stimulation. As a consequence, it seems
that this parietalfrontal pathway subserves spatial
awareness, and that a lesion at its level may generate
a permanent left-neglect. These results suggest that
damage to resctricted regions of white matter can
cause the dysfunction of large-scale cognitive networks. Also, this illustration shows that it is possible
to adapt the intraoperative testing to each patient
with the goal to map the subcortical pathway underlying cognitive functions other than language.
35.3.2.6. Corpus callosum
Interestingly, whereas DES of the interhemispheric
white pathways has been performed, no functional
response has been elicited by the stimulation of the
corpus callosum. Such results have allowed the
resection of gliomas involving this structure without
any change in quality of life (Duffau et al., 2004b).
In summary, IOM allows the neurosurgeon to
improve his knowledge of the anatomicalfunctional
connectivity, thus integrating, more easily and more
systematically, the concept of subcortical mapping
in his surgical strategy: first, because the gliomas
involve both cortical and subcortical structures, and
second, because lessons from stoke studies have
shown that the white matter lesions elicited more
severe and more permanent neurological deficits than
cortical damage.
H. DUFFAU
35.3.3. Study of the individual cerebral plasticity

Cerebral plasticity could be defined as the continuous
processing that allows short, middle, and long-term
remodeling of the neuronalsynaptic organization,
in order to optimize the functioning of the networks
in the brain. The behavioral consequences of such
cerebral phenomena have been analyzed in the last
decade, both in physiologyontogeny and learning
(Karni et al., 1998; Johnson, 2003) and in pathology (Chen et al., 2002; Kolb, 2003). In particular, the
ability to recover after a lesion of the nervous system,
and the patterns of map reorganization within eloquent area and/or within distributed networks, allowing such compensation, have been extensively
studied notably in stroke (Rijntjes and Weiller,
2002; Rossini et al., 2003). In the same way, in
neurooncology, IOM can also enable the study of
the dynamic reorganization of the eloquent maps
induced by tumors, and the surgical applications for
each patient (Duffau, 2005b).
35.3.3.1. Study of brain plasticity in cerebral tumors
35.3.3.1.1. Preoperative plasticity. It could seem
surprising that numerous brain tumor patients, especially low-grade gliomas (LGG) revealed by seizures,
have no neurological deficit, despite frequent tumor
invasion of eloquent structures. This means that these
slow-growing lesions have likely induced progressive
functional brain reshaping, as suggested by preoperative neurofunctional imaging. Interestingly, the patterns of reorganization may differ between patients,
a notion very important to a neurosurgeon who aims
to optimize the indication of surgery as well as surgical planning (Duffau, 2005b). Despite the limitations
of the preoperative neurofunctional imaging previously described, these methods have shown that three
kinds of preoperative functional redistribution are
possible in patients without deficits. In the first
redistribution, function still persists within the tumor,
due to the infiltrative feature of gliomas, thus limiting a surgeons chances of performing a good
resection without inducing postoperative sequelae
(Schiffbauer et al., 2001). In the second one, eloquent
areas are redistributed around the tumor (Thiel et al.,
2001), thereby giving a surgeon reasonable chances
to performing a near-total resection, despite a likely,
immediate transient deficit. With the expectation for
secondary recovery, however, is merely weeks to
months. In the third redistribution, there is already a
preoperative compensation by remote areas within

the lesional hemisphere and/or by the contralateral
homologues (Holodny et al., 2002; Baciu et al.,
2003); consequently, the chances of a surgeon
performing a total resection for this kind of glioma
are very high, with only slight, and transient, deficits.
35.3.3.1.2. Intraoperative plasticity. IOM before
any resection has allowed the confirmation of the existence of a functional reshaping induced by brain lesions,
concerning both remapping of the motor homonculus
(Branco et al., 2003) and the language network. When
a tumor is located in the so-called eloquent regions,
such as Brocas area or the left dominant insula, for
instance, cortical mapping before resection has
frequently shown perilesional reorganization of the
functional sites (Duffau, 2005b).
Moreover, acute reorganization of functional
maps was equally observed during the resection,
likely due to the surgical act itself, which can generate a locoregional hyperexcitability (Duffau, 2001a),
as already demonstrated in head injury.
In several patients harboring a frontal lesion,
though stimulations of the precentral gyrus induced
motor responses at only a limited number of cortical
sites before the resection, an acute unmasking of
redundant motor sites located within the same
precentral gyrus (and eliciting the same movements
as the previous adjacent sites when stimulated)
was observed immediately following lesion removal
(Duffau, 2001a). Acute unmasking of redundant
somatosensory sites was also regularly observed
within the retrocentral gyrus in parietal glioma
patients (Duffau and Capelle, 2001). Furthermore, it
was equally possible to detect a redistribution within
a larger network that involves the whole rolandic
region, that is, with the unmasking of functional
homologues located in the precentral gyrus for the
first cortical representation and in the retrocentral
gyrus for its redundancy (or vice versa) (Duffau
et al., 2000b). Finally, IOM also has prognostic value
concerning postoperative recovery (Duffau, 2001b).
35.3.3.1.3. Postoperative plasticity. The mechanisms of such plasticity induced by surgical resection
within eloquent areas were also studied with postoperative neuroimaging once the patient had recovered
all preoperative functional status (Kamada et al.,
2004). In particular, several patients were examined
following resection of gliomas involving the SMA,
which had elicited a transient postsurgical SMA
499
syndrome. In comparison to the preoperative imaging, functional MRI showed, in these cases, the
occurrence of activations of the SMA and premotor
cortex contralateral to the lesion. This indicates the
likelihood that the contrahemispheric homologues
participated in the postsurgical functional compensation and recovery (Krainik et al., 2004).
35.3.3.2. Surgical use of brain plasticity
Such pre-, intra-, and postoperative brain plasticity
mechanisms, studied using both functional neuroimaging and IOM, are very useful to know for the neurosurgeons, since it is possible to apply this brain
dynamic potential to extend the indications and the
limits of tumor resection located in the eloquent areas
without inducing a definitive neurological deficit
(Duffau, 2005b) (Fig. 2).
It was possible to totally remove the SMA, despite
the occurrence of an immediate postsurgical transient
SMA syndrome (with akinesia and with or without
mutism) which resolved in some weeks due to the
recruitment of the contralateral homologues (Krainik
et al., 2004). Moreover, the resection of the insular
lobe could be performed without any permanent deficit, even in the dominant hemisphere. In particular,
this was due to the compensation of the frontal and
temporal operculae, and also heavily influenced by
the left putamen (Duffau et al., 2001b). In the same
way, it was regularly possible to remove gliomas
involving the pars opercularis and pars triangularis
of the left inferior frontal gyrus, namely Brocas area.
This was done, without generating any aphasia,
because of a perilesional reorganization of language
areas; in particular, in the ventral premotor cortex
behind the lesion and in the pars orbitaris of the
inferior frontal gyrus in front of the glioma (Duffau,
2005b) (Fig. 1). The removal of the primary sensorimotor area of the face was also performed without
eliciting permanent postoperative central facial
palsy, but only within the nondominant hemisphere
(LeRoux et al., 1991). In these cases, the subcortical
motor sites, which represented the deep functional
boundaries of the resection, corresponded to the
pyramidal pathways of the upper limb, running under
the representation of the face previously removed
(Duffau et al., 2003a). Concerning subcortical structures, compensation was equally possible, especially
in gliomas involving the right frontotemporalinsular
structures and the striatum, with a total resection
without permanent palsy and without movement
disorders (Duffau et al., 2002d).
500
H. DUFFAU
Fig. 2. A: Preoperative anatomical magnetic resonance imaging (MRI) showing a low-grade glioma which involves the left
inferior parietal gyrus, namely the supramarginal gyrus. B (left): Intraoperative views before resection of the tumor, which
is delineated by letter tags. Electrical cortical mapping shows a reshaping of the eloquent maps, with a recruitment of perilesional language sites, that is the rolandic operculum (tags 20 and 21), angular gyrus (tags 30 and 32), and posterior part of
the superior temporal gyrus (tags 40 and 50). This language remapping induced by the slow-growing low-grade glioma has
allowed the compensation of the entire supramarginal gyrus. Therefore, a resection of this tumor can be envisoned despite
its location in classical inoperable region. The arrow shows the lateral part of the retrocentral sulcus. A anterior, P
posterior. C (right): Intraoperative views after resection of the tumor. Electrical subcortical mapping has enabled to study
the individual anatomicalfunctional language connectivity. Indeed, deeply and posteriorly, the posterosuperior loop of
the arcuate fasciculus was identified, by eliciting systematic phonological paraphasia during each stimulation (tags 40,
47, and 49). More posteriorly, the lateral part of the superior longitudinal fasciculus was detected, by inducing speech
apraxia (articulatory disturbances) (tags 45 and 46). It is worth noting that in the depth, the resection was continued up
to the contact of these language pathways, in order to optimize the quality of resection while preserving the eloquent white
matter tracts. The arrow shows the lateral part of the retrocentral sulcus. A anterior, P posterior. D: Postsurgical control
MRI, showing a subtotal glioma removal, performed according to individual cortico-subcortical functional boundaries given
by intraoperative testing and monitoring (IOM), in a patient with a normal neurological and neuropsychological status.
Again, the resection was adapted to the cortical plasticity and the language connectivity, with the goal to optimize the
benefit/risk ratio of the surgery.
Finally, based on the data provided by IOM, the surgical use of long-term functional reshaping induced by
a first surgical act was also suggested (Duffau et al.,
2002e). It seems that the intraoperative acute unmasking of redundancies described during a surgical resection could have an important functional role. That is,
the latent networks that were disinhibited during
tumor removal might be reinforced, and then lead to
a durable remapping. Interestingly, the functional
reorganization generated by the first surgery could be
useful in considering an eventual reoperation, with a
possibility to extend the tumor removal without eliciting sequelae. For instance, in cases of precentral glioma, homologues generating the same motor
responses as previous sites, also localized within the
precentral gyrus, were identified by stimulation in
some patients, after incomplete resection of the tumor,
because of invasion of the primary motor area of the
hand. Because of a recurrence, years later, a new surgery was performed which used intraoperative electrical mapping. Stimulation showed a motor map
reshaping, with essential areas now corresponding to
the unmasked sites during the first procedure, thus
enbling a total gliomal removal, without deficit (Duffau et al., 2002e). These long-term plasticity phenomena, induced by surgical resection, or possibly by
continued glioma growth, have also been used to
extend tumor removal in other eloquent regions during
a second surgery in those cases for which total resection was not initially possible. In particular, removal
of primary somatosensory areas and language sites
has been done without permanent deficit in reoperated
patients (Duffau et al., 2003a).
In summary, IOM has allowed the neurosurgeon
to gain better knowledge of these plasticity phenomena, and their variability among patients, in order to
integrate this potential in surgical indications and
dynamic surgical planning. In other words, the extent
of resection, and the number of surgical acts necessary to perform this resection should be adapted to
the individual potential of functional compensation,
thus to its limits.
35.4. Improvement of the functional and
oncological results of the surgery in brain tumors
35.4.1. Functional results
The integration of the study of individual functional
mapping and connectivity provided by IOM in the
surgical decision and planning has permitted the
501
extension of the indications of surgery for tumors

located in areas once considered inoperable
Moreover, despite a frequent, though transient,
immediate worsening of postoperative function (due
to attempts at performing maximal tumor removal
according to both cortical and subcortical functional
limits using IOM), in a delay of 3 months following
the surgery, more than 95% of patients recovered a
normal neurological examination. Some of the recoveries even posed possible improvement, comparedwith their preoperative status. Also, 80% of patients
with preoperative chronic epilepsia prsented with a
significant decrease of seizures (Duffau et al.,
2002a). All of these patients returned to a normal
socio-professional life. It is worth noting that the
rate of <5% of permanent postoperative sequelae is
reproducible among the teams using IOM worlwide
(for a review, see Duffau et al., 2005a).
Interestingly, as a comparison, for those series
which did not use IOM, the rate of sequelae ranged
from 13% to 27.5%, with a mean of around 19%
(Fadul et al., 1988; Apuzzo, 1993; Tandon et al.,
1993; Cabantog and Bernstein, 1994; Cedzich et al.,
1996; Sawaya et al., 1998; Vives and Piepmeier,
1999; Brell et al., 2000).
35.4.2. Neurooncological results
Since IOM allows identification of the cortical and
subcortical eloquent structures individually, it seems
logical to perform a resection according to functional
boundaries. The recommendation has come up of
continuing the resection until the functional structures are detected by DES, and not before, in order
to optimize the quality of resection, without increasing the risk to generate postoperative permanent
deficit (Duffau et al., 2005a). This surgical strategy
enables a significative improvement on the quality
of tumor removal, despite a higher number of surgeries within critical areas, and a parallel decrease
in the rate of sequelae. In a recent study comparing
LGG resected with or without IOM in the same institution, it was demonstrated that: (1) 62% of gliomas
selected for surgery were located within eloquent
area with the use of IOM, instead of only 35%
without IOM; (2) only 37% of resections were subtotal and 6% were total without IOM, whereas
50.8% of resections were subtotal and 25.4% subtotal
with the use of IOM (P < 0.001) (Duffau et al.,
2005a).
502
Moreover, while extensive surgery is still controversial in neurooncology, especially with regard to
LGG, current surgical results support the positive
impact of such a maximalist treatment strategy,
that is with benefits on the natural progress of the
tumors that seem to be directly related to the quality
of resection. Indeed, it was recently shown in a consecutive series of LGG operated on according to
functional boundaries using IOM that the mortality
rate was 20.6% in cases of partial resection, while
8% in cases of subtotal resection and 0% in cases
of total resection (follow-up 48 months) (P 0.02)
35.5. Conclusions and perspectives
Brain tumor surgery may now benefit from important
technical developments in the field of functional
mapping, using complementary noninvasive methods
of neuroimaging and IOM. Such recent advances have
enabled a better understanding of the organization of
the brain eloquent areas for each patient, in order to
integrate the notion of interindividual anatomical
functional variability in the surgical strategy. Furthermore, intraoperative real-time subcortical stimulation,
in association with cortical mapping, gives an unique
opportunity to investigate effective connectivity,
since it allows online correlations between discrete
and transient virtual lesions which can be performed
at each place of a distributed network (each cortical
and subcortical sites being perfectly identified
anatomically using 3D MRI) and their functional consequences (accurately analyzed by a speech therapist
all along the surgical procedure). A combination
of these intraoperative anatomicalfunctional data
with those provided by DTI (subcortical anatomical
informations), MEG (temporal data), fMRI, and PET
(perioperative functional data) coud enable one to
elaborate individual and predictive models of the
functioning of neuronalsynaptic circuits, that is to
open a new door to connectionism (Duffau, 2006). It
is worth noting that such correlations between IOM,
which remains the gold-standard with regard to functional brain mapping, can also enable to validate the
noninvasive method of neuroimaging, especially the
new technique of DTI.
Moreover, such connectionist models (Marrelec
et al., 2006) may lead to improved knowledge of
the dynamic potential of spatialtemporal reorganization of the parallel and interactive networks, such
H. DUFFAU
as the mechanisms of brain plasticity, thought to play

a major role in functional compensation in slowgrowing tumors and in their surgical resections. In
this way, individual plastic potential could be better
understood using repeated intraoperative mappings
combined to postsurgical neurofunctional imaging,
then possibly guiding specific postoperative rehabilitation programs to optimize the quality of functional
recovery.
In conclusion, IOM allows the neurosurgeon to
perform tumor resection according to functional
boundaries, thereby leading to the optimization of
the ratio of benefit/risk of surgical removal in cerebral tumors, via: (1) the extension of surgical indications, in particular in eloquent brain areas; (2) a
better quality of tumor resection with a greater
neurooncological impact; and (3) a minimization of
the risk of postoperative sequelae, with preservation
of the quality of life.
References
Ackermann, H and Riecker, A (2004) The contribution of
the insula to motor aspects of speech production: a
review and a hypothesis. Brain Lang., 89: 320328.
Apuzzo, MLJ (1993) Brain surgery. In: Complication
Avoidance and Management. Churchill Livingstone,
New York. Vol. 1, pp. 379390.
Aubert, A, Costalat, R, Duffau, H and Benali, H (2002)
Modeling of pathophysiological coupling between brain
electrical activation, energy metabolism and hemodynamics: insights for the interpretation of intracerebral
tumor imaging. Acta Biotheor., 50: 281295.
Augustine, JR (1996) Circuitry and functional aspects of
the insular lobe in primates including humans. Brain
Res. Rev., 22: 229244.
Baciu, M, Le Bas, JF, Segebarth, C and Benabid, AL
(2003) Presurgical fMRI evaluation of cerebral reorganization and motor deficit in patients with tumors
and vascular malformations. Eur. J. Radiol., 46:
139146.
Berger, MS, Ojemann, GA and Lettich, E (1990) Neurophysiological monitoring during astrocytoma surgery.
Neurosurg. Clin. N. Am., 1: 6570.
Berger, MS, Deliganis, AV, Dobbins, J and Keles, GE
(1994) The effect of extent of resection on recurrence
in patients with low grade cerebral hemisphere gliomas.
Cancer, 74: 17841791.
Blanke, O and Seeck, M (2003) Direction of saccadic and
smooth eye movements induced by electrical stimulation of the human frontal eye field: effect of orbital
position. Exp. Brain Res., 150: 174183.
Boatman, D, Gordon, B, Hart, J, Selnes, O, Miglioretti, D
and Lenz, F (2000) Transcortical sensory aphasia: revisited and revised. Brain, 123: 16341642.
Branco, DM, Coelho, TM, Branco, BM, Schmidt, L,
Calcagnotto, ME, Portuguez, M, Neto, EP, Paglioli, E,
Palmini, A, Lima, JV and Da Costa, JC (2003) Functional variability of the human cortical motor map: electrical stimulation findings in perirolandic epilepsy
surgery. J. Clin. Neurophysiol., 20: 1725.
Brell, M, Ibanez, J, Caral, L and Ferrer, E (2000) Factors
influencing surgical complications of intra-axial brain
tumours. Acta Neurochir. (Wien), 142: 739750.
Cabantog, AM and Bernstein, M (1994) Complications of
first craniotomy for intra-axial brain tumour. Can. J.
Neurol. Sci., 21: 213218.
Catani, M, Jones, DK and Ffytche, DH (2005) Perisylvian
language networks of the human brain. Ann. Neurol.,
57: 816.
Cedzich, C, Taniguchi, M, Schafer, S and Schramm, J (1996)
Somatosensory evoked potential phase reversal and
direct motor cortex stimulation during surgery in and
around the central region. Technical application. Neurosurgery, 38: 962971.
Chen, R, Cohen, LG and Hallett, M (2002) Nervous system
reorganization following injury. Neuroscience, 111:
761773.
Chitoku, S, Otsubo, H, Harada, Y, Jay, V, Rutka, JT,
Weiss, SK, Abdoll, M and Snead, OC, 3rd (2001) Extraoperative cortical stimulation of motor function in children. Pediatr. Neurol., 24: 344350.
Claus, EB, Horlacher, A, Hsu, L, Schwartz, RB, DelloIacono, D, Talos, F, Jolesz, FA and Black, PM
(2005) Survival rates in patients with low-grade glioma after intraoperative magnetic resonance image
guidance. Cancer, 103: 12271233.
Coenen, VA, Krings, T, Mayfrank, L, Polin, RS, Reinges,
MH, Thron, A and Gilsbach, JM (2001) Three-dimensional visualization of the pyramidal tract in a neuronavigation system during brain tumor surgery: first
experiences and technical note. Neurosurgery, 49:
8692.
Crone, NE (2000) Functional mapping with ECoG spectral
analysis. Adv. Neurol., 84: 343351.
Danks, RA, Aglio, LS, Gugino, LD and Black, PM (2000)
Craniotomy under local anesthesia and monitored conscious sedation for the resection of tumors involving
eloquent cortex. J. Neurooncol., 49: 131139.
Dehaene, S, Molko, N, Cohen, L and Wilson, AJ (2004)
Arithmetic and the brain. Curr. Opin. Neurobiol., 14:
218224.
Dronkers, NF (1996) A new brain region for coordinating
speech articulation. Nature, 384: 159161.
Duffau, H (2001a) Acute functional reorganisation of the
human motor cortex during resection of central lesions:
503
a study using intraoperative brain mapping. J. Neurol.
Neurosurg. Psychiatry, 70: 506513.
Duffau, H (2001b) Recovery from complete hemiplegia
following resection of a retrocentral metastasis: the
prognostic value of intraoperative cortical stimulation.
J. Neurosurg., 95: 10501052.
Duffau, H (2005a) Intraoperative cortico-subcortical stimulations in surgery of low-grade gliomas. Expert Rev.
Neurother., 5: 473485.
Duffau, H (2005b) Lessons from brain mapping in surgery
for low-grade glioma: insights into associations between
tumour and brain plasticity. Lancet Neurol., 4: 476486.
Duffau, H (2006) New concepts in surgery of WHO grade
II gliomas: functional brain mapping, connectionism
and plasticity. A review. J. Neurooncol., 79: 77115.
Duffau, H and Capelle, L (2001) Functional recuperation
following lesions of the primary somatosensory fields.
Study of compensatory mechanisms. Neurochirurgie,
47: 557563.
Duffau, L and Capelle, L (2004) Preferential brain locations of low-grade gliomas. Cancer, 100: 26222626.
Duffau, H and Capelle, L (2005) Incontinence after brain
glioma surgery: new insights into the cortical control
of micturition and continence. Case report. J. Neurosurg., 102: 148151.
Duffau, H and Fontaine, D (2005) Successful resection of a
left insular cavernous angioma using neuronavigation
and intraoperative language mapping. Acta Neurochir.
(Wien), 147: 205208.
Duffau, H, Capelle, L, Sichez, JP, Faillot, T, Abdennour, L,
Law Koune, JD, Dadoun, S, Bitar, A, Arthuis, F, Van
Effenterre, R and Fohanno, D (1999) Intraoperative
direct electrical stimulations of the central nervous system: the Salpetrie`re experience with 60 patients. Acta
Neurochir. (Wien), 141: 11571167.
Duffau, H, Capelle, L, Lopes, M, Faillot, T, Sichez, JP and
Fohanno, D (2000a) The insular lobe: physiopathological and surgical considerations. Neurosurgery, 47:
801810.
Duffau, H, Sichez, JP and Lehericy, S (2000b) Intraoperative unmasking of brain redundant motor sites during
resection of a precentral angioma: evidence using direct
cortical stimulation. Ann. Neurol., 47: 132135.
Duffau, H, Lopes, M, Denvil, D and Capelle, L (2001a)
Delayed onset of the supplementary motor area syndrome after surgical resection of the mesial frontal lobe:
a time course study using intraoperative mapping in an
awake patient. Stereotact. Funct. Neurosurg., 76:
7482.
Duffau, H, Bauchet, L, Lehericy, S and Capelle, L (2001b)
Functional compensation of the left dominant insula for
language. Neuroreport, 12: 21592163.
Duffau, H, Capelle, L, Lopes, M, Bitar, A, Sichez, JP and
Van Effenterre, R (2002a) Medically intractable
504
epilepsy from insular low-grade gliomas: improvement
after an extended lesionectomy. Acta Neurochir.
(Wien), 144: 563572.
Duffau, H, Denvil, D, Lopes, M, Gasparini, F, Cohen, L,
Capelle, L and Van Effenterre, R (2002b) Intraoperative
mapping of the cortical areas involved in multiplication
and subtraction: an electrostimulation study in a patient
with a left parietal glioma. J. Neurol. Neurosurg. Psychiatry, 73: 733738.
Duffau, H, Capelle, L, Sichez, N, Denvil, D, Lopes, M,
Sichez, JP, Bitar, A and Fohanno, D (2002c) Intraoperative mapping of the subcortical language pathways
using direct stimulations. An anatomo-functional study.
Brain, 125: 199214.
Duffau, H, Denvil, D and Capelle, L (2002d) Absence of
movement disorders after surgical resection of glioma
invading the right striatum. J. Neurosurg., 97: 363369.
Duffau, H, Denvil, D and Capelle, L (2002e) Long term
reshaping of language, sensory, and motor maps after glioma resection: a new parameter to integrate in the surgical
strategy. J. Neurol. Neurosurg. Psychiatry, 72: 511516.
Duffau, H, Capelle, L, Denvil, D, Sichez, N, Gatignol, P, Lopes,
M, Mitchell, MC, Sichez, JP and Van Effenterre, R
(2003a) Functional recovery after surgical resection of
low-grade gliomas in eloquent brain: hypothesis of
brain compensation. J. Neurol. Neurosurg. Psychiatry,
74: 901907.
Duffau, H, Capelle, L, Denvil, D, Sichez, N, Gatignol, P,
Taillandier, L, Lopes, M, Mitchell, MC, Roche, S,
Muller, JC, Bitar, A, Sichez, JP and Van Effenterre, R
(2003b) Usefulness of intraoperative electrical subcortical mapping during surgery for low-grade gliomas
located within eloquent brain regions: functional results
in a consecutive series of 103 patients. J. Neurosurg.,
98: 764778.
Duffau, H, Capelle, L, Denvil, D, Gatignol, P, Sichez, N,
Lopes, M, Sichez, JP and Van Effenterre, R (2003c)
The role of dominant premotor cortex in language: a
study using intraoperative functional mapping in awake
patients. Neuroimage, 20: 19031914.
Duffau, H, Gatignol, P, Denvil, D, Lopes, M and Capelle,
L (2003d) The articulatory loop: study of the subcortical
connectivity by electrostimulation. Neuroreport, 14:
20052008.
Duffau, H, Velut, S, Mitchell, MC, Gatignol, P and
Capelle, L (2004a) Intra-operative mapping of the subcortical visual pathways using direct electrical stimulations. Acta Neurochir. (Wien), 146: 265269.
Duffau, H, Khalil, I, Gatignol, P, Denvil, D and Capelle, L
(2004b) Surgical removal of corpus callosum infiltrated
by low-grade glioma: functional outcome and oncological considerations. J. Neurosurg., 100: 431437.
Duffau, H, Lopes, M, Arthuis, F, Bitar, A, Sichez, JP, Van
Effenterre, R and Capelle, L (2005a) Contribution of
H. DUFFAU
intraoperative electrical stimulations in surgery of low
grade gliomas: a comparative study between two series
without (198596) and with (19962003) functional
mapping in the same institution. J. Neurol. Neurosurg.
Duffau, H, Gatignol, P, Mandonnet, E, Peruzzi, P, TzourioMazoyer, N and Capelle, L (2005b) New insights into
the anatomo-functional connectivity of the semantic
system: a study using cortico-subcortical electrostimulations. Brain, 128: 797810.
Duffau, H, Taillandier, L, Gatignol, P and Capelle, L
(2006) The insular lobe and brain plasticity: lessons
from tumor surgery. Clin. Neurol. Neurosurg., 108(6):
543548.
Fadul, C, Wood, J, Thaler, H, Galicich, J, Patterson, RH,
Jr., and Posner, JB (1988) Morbidity and mortality of
craniotomy for excision of supratentorial gliomas. Neurology, 38: 13741379.
Fontaine, D, Capelle, L and Duffau, H (2002) Somatotopy
of the supplementary motor area: evidence from correlation of the extent of surgical resection with the clinical
patterns of deficit. Neurosurgery, 50: 297303.
Freund, H (1996) Supplementary Sensorimotor Area. Historical overview. Adv. Neurol., 70: 1727.
Ganslandt, O, Buchfelder, M, Hastreiter, P, Grummich, P,
Falbusch, R and Nimsky, C (2004) Magnetic source imaging supports clinical decision making in glioma patients.
Clin. Neurol. Neurosurg., 107: 2026.
Gatignol, P, Capelle, L, Le Bihan, R and Duffau, H (2004)
Double dissociation between picture naming and comprehension: an electrostimulation study. Neuroreport,
15: 191195.
Gil Robles, S, Gatignol, P, Capelle, L, Mitchell, MC and
Duffau, H (2005) The role of dominant striatum in language: a study using intraoperative electrical stimulations. J. Neurol. Neurosurg. Psychiatry, 76: 940946.
Holodny, AI, Schulder, M, Liu, WC, Wolko, J, Maldjian,
JA and Kalnin, AJ (2000) The effect of brain tumors
on BOLD functional MR imaging activation in the adjacent motor cortex: implications for image-guided neurosurgery. Am. J. Neuroradiol., 21: 14151422.
Holodny, AI, Schulder, M, Ybasco, A and Liu, WC (2002)
Translocation of Brocas area to the contralateral hemisphere as the result of the growth of a left inferior frontal glioma. J. Comput. Assist. Tomogr., 26: 941943.
Ikeda, A, Miyamoto, S and Shibasaki, H (2002) Cortical
motor mapping in epilepsy patients: information from
subdural electrodes in presurgical evaluation. Epilepsia,
43: 5660.
Johnson, MH (2003) Development of human brain functions. Biol. Psychiatry, 54: 13121316.
Kamada, K, Sawamura, Y, Takeuchi, F, Houkin, K, Kawaguchi, H, Iwasaki, Y and Kuriki, S (2004) Gradual recovery
from dyslexia and related serial magnetoencephalographic
changes in the lexicosemantic centers after resection of a
mesial temporal astrocytoma. Case report. J. Neurosurg.,
100: 11011106.
Karni, A, Meyer, G, Rey-Hipolito, C, Jezzard, P, Adams,
MM, Turner, R and Ungerleider, LG (1998) The acquisition of skilled motor performance: fast and slow
experience-driven changes in primary motor cortex.
Proc. Natl. Acad. Sci. USA, 95: 861868.
Keles, GE (2004) Intracranial neuronavigation with intraoperative magnetic resonance imaging. Curr. Opin.
Neurol., 17: 497500.
Keles, GE, Lamborn, KR and Berger, MS (2001) Low-grade
hemispheric gliomas in adults: a critical review of extent
of resection as a factor influencing outcome. J. Neurosurg.,
95: 735745.
Keles, GE, Lamborn, KR and Berger, MS (2003) Coregistration accuracy and detection of brain shift using
intraoperative sononavigation during resection of hemispheric tumors. Neurosurgery, 53: 556562.
Kolb, B (2003) Overview of cortical plasticity and recovery from brain injury. Phys. Med. Rehabil. Clin. N.
Am., 14: S7S25.
Kombos, T, Suess, O, Ciklatekerlio, O and Brock, M
(2001) Monitoring of intraoperative motor evoked
potentials to increase the safety of surgery in and
around the motor cortex. J. Neurosurg., 95: 608614.
Krainik, A, Lehericy, S, Duffau, H, Vlaicu, M, Poupon, F,
Capelle, L, Cornu, P, Clemenceau, S, Sahel, M, Valery,
CA, Boch, AL, Mangin, JF, Bihan, DL and Marsault, C
(2001) Role of the supplementary motor area in motor
deficit following medial frontal lobe surgery. Neurology, 57: 871878.
Krainik, A, Lehericy, S, Duffau, H, Capelle, L, Chainay, H,
Cornu, P, Cohen, L, Boch, AL, Mangin, JF, Le Bihan,
D and Marsault, C (2003) Postoperative speech disorder
after medial frontal surgery: role of the supplementary
motor area. Neurology, 60: 587594.
Krainik, A, Duffau, H, Capelle, L, Cornu, P, Boch, AL,
Mangin, JF, Le Bihan, D, Marsault, C, Chiras, J and
Lehericy, S (2004) Role of the healthy hemisphere in
recovery after resection of the supplementary motor
area. Neurology, 62: 13231332.
Lacroix, M, Abi-Said, D, Fourney, DR, Gokaslan, ZL, Shi,
W, DeMonte, F, Lang, FF, McCutcheon, IE, Hassenbusch, SJ, Holland, E, Hess, K, Michael, C, Miller, D
and Sawaya, R (2001) A multivariate analysis of 416
patients with glioblastoma multiforme: prognosis, extent
of resection, and survival. J. Neurosurg., 95: 190198.
Lang, FF, Olansen, NE, DeMonte, F, Gokaslan, ZL, Holland, EC, Kalhorn, C and Sawaya, R (2001) Surgical
resection of intrinsic insular tumors: complication
avoidance. J. Neurosurg., 95: 638650.
Laplane, D, Talairach, J, Meininger, V, Bancaud, J and
Orgogozo, JM (1977) Clinical consequences of
505
corticectomies involving the supplementary motor area
in man. J. Neurol. Sci., 34: 301314.
Laws, ER, Parney, IF, Huang, W, Anderson, F, Morris, AM,
Asher, A, Lillehei, KO, Bernstein, M, Brem, H, Sloan, A,
Berger, MS and Chang, S Glioma outcomes investigators
(2003) Survival following surgery and prognostic factors
for recently diagnosed malignant glioma: data from the
glioma outcomes project. J. Neurosurg., 99: 467473.
LeRoux, PD, Berger, MS, Haglund, MM, Pilcher, WH and
Ojemann, GA (1991) Resection of intrinsic tumors from
nondominant face motor cortex using stimulation
mapping: report of two cases. Surg. Neurol., 36: 4448.
Lobel, E, Kahane, P, Leonards, U, Grosbras, M, Lehericy, S,
Le Bihan, D and Berthoz, A (2001) Localization of human
frontal eye fields: anatomical and functional findings of
functional magnetic resonance imaging and intracerebral
electrical stimulation. J. Neurosurg., 95: 804815.
Lucas, TH, McKhann, GM and Ojemann, GA (2004) Functional separation of languages in the bilingual brain: a
comparison of electrical stimulation language mapping
in 25 bilingual patients and 117 monolingual control
patients. J. Neurosurg., 101: 449457.
Mandonnet, E, Capelle, L and Duffau, H (2006) Extension
of paralimbic low-grade gliomas: towards an anatomical
classification based on white matter invasion patterns.
J. Neurooncol., 78: 179185.
Marrelec, G, Krainik, A, Duffau, H, Pelegrini-Issac, M,
Lehericy, S, Doyon, J and Benali, H (2006) Partial correlation for functional brain interactivity investigation
in functional MRI. NeuroImage, 32: 228237.
Matsumoto, R, Nair, DR, LaPresto, E, Najm, I, Bingaman,
W, Shibasaki, H and Luders, HO (2004) Functional
connectivity in the human language system: a corticocortical evoked potential study. Brain, 127: 23162330.
Meyer, FB, Bates, LM, Goerss, BS, Friedman, JA, Windschitl, WL, Duffy, JR, Perkins, WJ and ONeill, BP
(2001) Awake craniotomy for aggressive resection of
primary gliomas located in eloquent brain. Mayo Clin.
Proc., 76: 677687.
Meyer, PT, Sturz, L, Schreckenberger, M, Spetzger, U,
Meyer, GF, Setani, KS, Sabri, O and Buell, U (2003) Preoperative mapping of cortical language areas in adult brain
tumor patients using PET and individual non-normalised
SPM analyses. Eur. J. Nucl. Med. Mol. Imaging, 30:
951960.
Milea, D, Lobel, E, Lehericy, S, Duffau, H, Rivaud-Pechoux,
S, Berthoz, A and Pierrot-Deseilligny, C (2002) Intraoperative frontal eye field stimulation elicits ocular deviation and saccade suppression. Neuroreport, 13:
13591364.
Neuloh, G and Schramm, J (2004) Motor evoked potential
monitoring for the surgery of brain tumours and vascular malformations. Adv. Tech. Stand. Neurosurg., 29:
171228.
506
Nimsky, C, Ganslandt, O and Fahlbusch, R (2004) Functional neuronavigation and intraoperative MRI. Adv.
Tech. Stand. Neurosurg., 29: 229263.
Ojemann, G, Lettich, E and Berger, M (1989) Cortical language localization in left, dominant hemisphere. An
electrical stimulation mapping investigation in 117
Onal, C, Otsubo, H, Araki, T, Chitoku, S, Ochi, A, Weiss, S,
Elliott, I, Snead, OC, 3rd, Rutka, JT and Logan, W (2003)
Complications of invasive subdural grid monitoring in
children with epilepsy. J. Neurosurg., 98: 10171026.
Orrison, WW, Jr. (1999) Magnetic source imaging in stereotactic and functional neurosurgery. Stereotact. Funct.
Peraud, A, Ilmberger, J and Reulen, HJ (2004) Surgical
resection of gliomas WHO grade II and III located in
the opercular region. Acta Neurochir. (Wien), 146:
917.
Picard, N and Strick, PL (2001) Imaging the premotor
areas. Curr. Opin. Neurobiol., 11: 663672.
Picht, T, Kombos, T, Gramm, HJ, Brock, M and Suess, O
(2006) Multimodal protocol for awake craniotomy in
language cortex tumour surgery. Acta Neurochir.
(Wien), 148: 127138.
Platenik, LA, Miga, MI, Roberts, DW, Lunn, KE, Kennedy,
FE, Hartov, A and Paulsen, KD (2002) In vivo quantification of retraction deformation modeling for updated
image-guidance during neurosurgery. IEEE Trans.
Biomed. Eng., 49: 823835.
Reinges, MH, Nguyen, HH, Krings, T, Hutter, BO, Rohde,
V and Gilsbach, JM (2004) Course of brain shift during
microsurgical resection of supratentorial cerebral
lesions: limits of conventional neuronavigation. Acta
Rijntjes, M and Weiller, C (2002) Recovery of motor and
language abilities after stroke: the contribution of functional imaging. Prog. Neurobiol., 66: 109122.
Romstock, J, Fahlbusch, R, Ganslandt, O, Nimsky, C and
Strauss, C (2002) Localisation of the sensorimotor cortex during surgery for brain tumours: feasibility and
waveform patterns of somatosensory evoked potentials.
Rossini, PM, Calautti, C, Pauri, F and Baron, JC (2003)
Post-stroke plastic reorganisation in the adult brain.
Lancet Neurol., 2: 493502.
Roux, FE and Tremoulet, M (2002) Organization of language areas in bilingual patients: a cortical stimulation
Roux, FE, Boulanouar, K, Lotterie, JA, Mejdoubi, M,
LeSage, JP and Berry, I (2003a) Language functional
magnetic resonance imaging in preoperative assessment
of language areas: correlation with direct cortical stimulation. Neurosurgery, 52: 13351345.
Roux, FE, Boetto, S, Sacko, O, Chollet, F and Tremoulet,
M (2003b) Writing, calculating, and finger recognition
H. DUFFAU
in the region of the angular gyrus: a cortical stimulation
study of Gerstmann syndrome. J. Neurosurg., 99:
716727.
Roux, FE, Lubrano, V, Lauwers-Cances, V, Tremoulet, M,
Mascott, CR and Demonet, JF (2004) Intra-operative
mapping of cortical areas involved in reading in monoand bilingual patients. Brain, 127: 17961810.
Sawaya, R, Hammoud, M, Schoppa, D, Hess, KR, Wu, SZ,
Shi, WM and Wildrick, DM (1998) Neurological outcomes in a modern series of 400 craniotomies for treatment of parenchymal tumors. Neurosurgery, 42:
10441056.
Schaffler, L, Luders, HO and Beck, GJ (1996) Quantitative
comparison of language deficits produced by extraoperative electrical stimulation of Brocas, Wernickes, and
basal temporal language areas. Epilepsia, 37: 463475.
Schiffbauer, H, Ferrari, P, Rowley, HA, Berger, MS and
Roberts, TP (2001) Functional activity within brain
tumors: a magnetic source imaging study. Neurosurgery, 49: 13131320.
Schiffbauer, H, Berger, MS, Ferrari, P, Freudenstein, D,
Rowley, HA and Roberts, TPL (2002) Preoperative
magnetic source imaging for brain tumor surgery: a
quantitative comparison with intraoperative sensory
and motor mapping. J. Neurosurg., 97: 13331342.
Signorelli, F, Guyotat, J, Isnard, J, Schneider, F, Mohammedi, R and Bret, P (2001) The value of cortical stimulations applied to the surgery of malignant gliomas in
language areas. Neurol. Sci., 22: 310.
Tandon, P, Mahapatra, AK and Khosla, A (1993) Operations on gliomas involving speech centres. Acta Neurochir. Suppl. (Wien), 56: 6771.
Taylor, MD and Bernstein, M (1999) Awake craniotomy
with brain mapping as a routine surgical approach to
treating patients with supratentorial intraaxial tumors:
a prospective trial of 200 cases. J. Neurosurg., 90:
3541.
Thiebaut de Schotten, M, Urbanski, M, Duffau, H, Volle,
E, Levy, R, Dubois, B and Bartolomeo, P (2005)
Direct evidence for a parietal-frontal pathway subserving spatial awareness in humans. Science, 309:
22262228.
Thiel, A, Herholz, K, Koyuncu, A, Ghaemi, M, Kracht,
LW, Habedank, B and Heiss, WD (2001) Plasticity of
language networks in patients with brain tumors: a positron emission tomography activation study. Ann.
Neurol., 50: 620629.
Tzourio-Mazoyer, N, Josse, G, Crivello, F and Mazoyer, B
(2004) Interindividual variability in the hemispheric
organization for speech. NeuroImage, 21: 422435.
Vives, KP and Piepmeier, JM (1999) Complications and
expected outcome of glioma surgery. J. Neurooncol.,
42: 289302.
Vlieger, EJ, Majoie, CB, Leenstra, S and Den Heeten, GJ
(2004) Functional magnetic resonance imaging for
neurosurgical planning in neurooncology. Eur. Radiol.,
14: 11431153.
Whittle, IR (2004) The dilemma of low grade glioma.
Whittle, IR, Borthwick, S and Haq, N (2003) Brain dysfunction following awake craniotomy, brain mapping
and resection of glioma. Br. J. Neurosurg., 17:
130137.
507
Wiedemayer, H, Sandalcioglu, IE, Armbruster, W, Regel, J,
Schaefer, H and Stolke, D (2004) False negative findings
in intraoperative SEP monitoring: analysis of 658 consecutive neurosurgical cases and review of published
reports. J. Neurol. Neurosurg. Psychiatry, 75: 280286.
Wise, RJ, Greene, J, Buchel, C and Scott, SK (1999) Brain
regions involved in articulation. Lancet, 353:
10571061.

M.R. Nuwer (Ed.)
508
CHAPTER 36
Movement disorders
Jeffrey W. Cozzens*
Department of Neurosurgery, Feinberg School of Medicine, Northwestern University, Evanston Hospital, Evanston, IL 60201, USA
Functional neurosurgery is a branch of neurosurgery

that deals with abnormalities of neurophysiology.
Surgery for movement disorders represents one of
its purest forms. Unlike neurosurgery for epilepsy
where the best results are obtained when there is an
associated anatomical abnormality, neurosurgery for
movement disorders has the best results when the
brain appears to be normal on neuroimaging studies.
Thus, without any definable lesions, the surgical
team must rely solely upon neurophysiology and the
clinical response to treatment. Also, because the surgical team is operating on a brain without anatomic
lesions, they must rely heavily upon the experience
of others. For this reason, more than just about any
other subspecialty of neurosurgery, it is important
to review what others have learned.
36.1. History
The history of surgery for movement disorders
closely follows the history of surgery for Parkinsons
disease and dates back to 1909 and Victor Horsley
who attempted to relieve some of the symptoms of
this disease by ablating certain areas of the sensorimotor cortex (Horsley, 1909). Patients often experienced a reduction of their tremor with an associated
impairment of voluntary function. A large number
of individuals who survived the influenza epidemic
of 19181920 later developed Parkinsons disease,
and because medical treatment for Parkinsons disease was so limited at this time, there was a great
need for a surgical treatment. An operation to lesion
the corticospinal tracts was popularized by Bucy in
1939 (Bucy and Case, 1939). At the same time,
experimental neurophysiological research began to
*
Correspondence to: Jeffrey W. Cozzens, M.D., Department of Neurosurgery, Feinberg School of Medicine,
Northwestern University, Evanston Hospital, Evanston,
IL 60201, USA.
Tel.: 1-847-570-1440.
E-mail: cozzens@northwestern.edu (J.W. Cozzens).
identify the extrapyramidal system as being important in regulating movement and its role in movement
disorders such as Parkinsons disease. Russell
Meyers was the first to operate on the basal ganglia
in 1939 (Meyers, 1942) through an open craniotomy,
but with a mortality rate of 15.7% (Meyers, 1958).
Less invasive procedures were sought, and the idea
of stereotactic surgery was reexplored. [There is
often discussion as to whether the word should be
stereotaxic or stereotactic. Both words have been
used in the past and the current trend is to use the latter term which has its Latin root in the word tactus
(to touch). The American and World Societies for
Stereotactic and Functional Neurosurgery accepted
the word stereotactic when they were founded in
1973 (Gildenberg, 1993).]
Stereotactic neurosurgery was first conceived by
Horsley and Clark in 1906 (Horsley and Clarke,
1908) as a method to introduce instruments, and
devices into the brain through a small opening. The
brain is unique in that it is held immobile in a protective case which makes it ideal for applying a threedimensional Cartesian coordinate system for locating
any structure. Stereotactic surgery utilizes a reference
system usually a frame which is secured to the
patients skull which is then matched to the coordinates of structures which are mapped on an atlas
or image of the patients brain. This technique, which
is currently used extensively by neurosurgeons,
began in the early 1950s. What makes this technique
so successful is the ability to use it with CT and/
or MRI images of the brain which provides an
individual atlas of each patients brain.
For two decades in the 1950s and 1960s, stereotactic surgery for Parkinsons disease became widely
used and accepted by the public. As the understanding of neurophysiology and the pathoneurophysiology of Parkinsons disease became more developed,
various targets in the basal ganglia were ablated with
such techniques as alcohol, heat, and cold (Clower,
2001, 2002) with varying degrees of success. Some
targets were found accidentally when excellent

clinical outcomes were noted in individuals who
had received unintended lesions in the wrong area
of the brain or who had infarctions as a complication
of brain surgery (Gildenberg, 2000). Microelectrode
recording of deep brain neurons was first used to help
with targeting in 1961 (Albe-Fessard et al., 1961).
This enthusiasm for surgery came to a halt in 1968
when levodopa was introduced for the treatment of
Parkinsons disease. When the first results of the
treatment of Parkinsons disease with levodopa were
known, most neurologists were convinced that surgery would never again be needed. This belief lasted
only for about a decade when it became apparent that
some patients were intolerant or resistant to treatment
with levodopa. At the same time, there was a revolution in neuroimaging with the development of CT
imaging and later MRI, and suddenly surgery for
Parkinsons disease and other movement disorders
was reborn (Gildenberg, 2000). In the early 1990s,
Laitinen and colleagues reintroduced unilateral posteroventral pallidotomy for the treatment of intractable Parkinsons disease (Laitinen et al., 1992a,b).
Sigfried and Lippitz (1994) reported on pallidal stimulation for Parkinsons disease in 1994, and in 1995
Limousin and colleagues reported on their first series
of Parkinsons disease patients treated with deep
brain high-frequency electrical stimulation in bilateral
subthalamic nuclei (Limousin et al., 1995a,b). Since
then, there has been an explosion of papers detailing
the technique of deep brain stimulation (DBS).
In recent years, surgery for other movement disorders has become widely accepted. This includes DBS
for Tourettes syndrome, essential tremor, and dystonia (Diederich et al., 2005; Foote et al., 2005, 2006;
Houeto et al., 2005; Marks, 2005; Vidailhet and
Pollak, 2005; Vidailhet et al., 2005; Pahwa et al.,
2006). The use of DBS to treat dystonia is a rapidly
evolving area, and preliminary evidence suggests that
primary dystonia linked to genetic mutation, other
primary dystonias, and tardive dystonic syndromes
respond most dramatically to treatment with DBS,
whereas secondary dystonia tends to be less responsive (Marks, 2005).
36.2. General principles
The most common movement disorder is the Parkinsons syndrome which is characterized by akinesia
(lack of movement and movement initiation), bradykinesia (slowness of movement), muscle rigidity,
509
and a resting tremor. Normal movement is initiated

by neurons in the cerebral cortex, and is modulated
by neurons in the basal ganglia and thalamus. There
are extensive interactions between the various
nuclei in the basal ganglia and thalamus with some
pathways resulting in inhibition of movement and
other pathways resulting in facilitation of movement.
If there is an imbalance between inhibition and facilitation, movement initiated by the cortical neurons is
either dampened and suppressed, or exaggerated and
enhanced.
The symptoms of Parkinsons disease results from
increased basal ganglia output which ultimately leads
to hypoactivity. The output of the basal ganglia is
inhibitory on the various nuclei of the thalamus and
leads to decreased movement. In idiopathic Parkinsons
disease, the increased output of the basal ganglia results
from decreased dopamine production in the substantia
nigra compacta.
Surgery for Parkinsons disease is successful
because it attempts to restore a balance in a neurophysiological circuit that has become out of balance
(Betchen and Kaplitt, 2003). A simplified diagram
of the basal ganglia circuits in normal individuals
and those with Parkinsons disease is seen in Fig. 1.
One can see that in Parkinsons disease, the primary
defect is the lack of output from the substantia nigra
which leads to increased activity of the subthalamic
nucleus (STN) and globus pallidus interna (GPi)
which ultimately leads to increased basal ganglion
inhibition of the thalamus (Wichmann et al., 2000).
Ideally, one would like to correct the initial fault by
increasing the output of the substantia nigra (Lang
and Obeso, 2004). One can also achieve the same
end by reducing the activity of the STN or the GPi
which will restore the normal basal ganglia outflow
to the thalamus, and, therefore, restore normal
movements.
It is very easy to decrease neural activity with
surgery. This can be accomplished with ablative or
destructive agents such as chemical (alcohol injection)
or thermal (hot or cold) lesioning agents. Normal neural activity can also be decreased with high-frequency
electrical stimulation. This high-frequency electrical
stimulation is accomplished through implanted electrodes which modulate, and disrupt normal neural
activity. For movement disorders, these electrodes
are passed into deep brain structures and the process
is commonly called DBS.
On the other hand, increased neural activity is
very difficult to achieve surgically. This usually
510
J.W. COZZENS
Fig. 1. A simplified diagram of basal gangliathalamic circuitry under normal conditions and in Parkinsons disease. Inhibitory connections are shown with solid black arrows and excitatory connections are shown with gray arrows. The strength of
the connection is shown by the thickness of the arrow. In Parkinsons disease, the basal ganglia output to the thalamus is
increased, influenced in part by overactivity of the subthalamic nucleus (STN). GPe, external segment of the globuspallidus;
GPi, internal segment of the globus pallidus; SNr, substantia nigra, pars reticulata; SNc, substantia nigra, pars compacta;
STN, subthalamic nucleus; VL, ventrolateral thalamus; PPN, pedunculopontine nucleus; and CM/Pf, centromedian/parafascicular nucleus of the thalamus.
requires transplantation or implantation of new, and

active neurons into the brain. There is some hope that
the implantation of drug delivery devices directly
into the brain may also achieve the same objective.
At present, there is no practical method to enhance
neural activity with implanted electrodes and lowfrequency stimulation.
Ablation of various nuclei in the brain has been
the main surgical treatment of Parkinsons disease
from 1950 until 1997. In recent years, however,
ablation has given way to modulation through
DBS. There is still a place for ablation in the surgical treatment of Parkinsons disease but in recent
years it is rarely performed. Ablation of the GPi
(pallidotomy) will treat the major symptoms of Parkinsons disease but this is an all-or-nothing type of
treatment (Alkhani and Lozano, 2001; Palur et al.,
2002; Parkin et al., 2002). If the lesion is too small,
it will be ineffective and it is difficult to relesion
the area. If the lesion is too large, it cannot be
undone. There is a relatively high risk of dysarthria,
dysphonia, and dysphagia in patients undergoing
bilateral pallidotomy so it is rarely performed on
both sides (Hariz, 2000). Pallidotomy is, however,
less expensive than DBS since there are no other
devices required and this may be important when
cost issues are of primary importance. Since the
lesion is permanent and fixed, there is no extensive

follow-up required with the neurologist as there is
with DBS for programming of the stimulator.
Recent studies have shown that patients who fail
to achieve satisfactory results with pallidotomy
may benefit by subsequent surgery for DBS (Mogilner et al., 2002).
Deep brain stimulation of the STN or GPi also
treats the major symptoms of Parkinsons disease
and has several advantages over ablation. There is a
good argument that it treats more of these symptoms
and more effectively than pallidotomy. The output
of the stimulator can be programmed to the specific
needs of the patient, maximizing symptom relief,
and minimizing any side effects. This programming
is quite complex, however, and requires extensive
follow-up. The stimulator and electrode are expensive, and the stimulator needs to be replaced every
35 years when the battery runs out. Like pallidotomy, it is thought that DBS does not alter the
ultimate course of degeneration of the dopamineproducing neurons in the substantia nigra compacta
or the mental deterioration seen in Parkinsons
disease. Because DBS is the predominant surgical
procedure in recent years for Parkinsons disease,
it is worthwhile to look more extensively at how this
procedure is performed.
36.3. Patient selection

As with any surgical technique, good patient selection is essential for good results (Lang and Widner,
2002; Lopiano et al., 2002; Welter et al., 2002;
Pahwa et al., 2005). The criteria for selecting patients
who would benefit from DBS are listed in Table 1.
The most important factor is that the individual
should have idiopathic Parkinsons disease. The
Parkinsons syndrome of bradykinesia, and resting
tremor can result from a number of different insults
to the brain which all have the common end result
of inhibiting neural activity in the thalamus. In idiopathic Parkinsons disease, the syndrome results from
the degeneration of dopaminergic secreting neurons
in the substantia nigra, but other etiologies can cause
this syndrome as well, such as, small lacunar infarcts
from cerebral vascular disease, encephalitis, exposure to toxins or neuroleptics (Bakay and Vitek,
2000). Surgery for Parkinsons disease seems to
be most effective in patients who have idiopathic
Parkinsons disease, and who have a favorable
response (i.e., improvement of bradykinesia/akinesia)
with levodopa (Welter et al., 2002; Kleiner-Fisman
et al., 2003). This group includes most of the population diagnosed with Parkinsons disease (Lang and
Widner, 2002).
Unfortunately, as treatment with levodopa continues over time, most patients will begin to have
complications of levodopa therapy which include
motor fluctuations, and severe dyskinesias. These
problems can sometimes be corrected by smoothing
Table 1
Patient selection criteria for deep brain stimulation
(DBS)
Patient selection:
Idiopathic Parkinsons disease
Responsive to levodopa
Some periods of good on time with or without
dyskinesia
Predictors of good outcome:
Severe dyskinesias
Motor fluctuations
Exclusion criteria:
Dementia
Severe medical illness
511
out the dosage of levodopa over the day with smaller

and more frequent dosing. In many cases, however,
there can be wide fluctuations of motor function
between no movement (akinesia or bradykinesia)
and too much movement (dyskinesia). One of the
effects of surgery is that ablation or modulation of
the basal ganglia will mimic the effects of the natural
production of dopamine. Therefore, normal movement is restored with a reduced need for levodopa
and resulting motor fluctuations and dyskinesia.
Individuals with severe medical illness or severe
dementia should not have this surgery. The surgical
procedures can last up to 7 h under local anesthesia,
thus the patient should be able to tolerate a long operation. One must also weigh the risks and the benefits
of surgery. If someone is too sick or too demented to
appreciate the benefits of this surgery, they should
not undergo the procedure.
36.4. Surgical technique
At the time of writing, there is still some controversy
regarding surgical technique. These issues are outlined in Table 2. The first issue is whether the targeting should be based off of ventriculography or MRI.
Ventriculography is an old technique where contrast
is injected into the ventricles and imaged with radiographs. When performed with the stereotactic frame
secured to the head, measurements can be made from
the films, and a trajectory can be calculated for
placement of electrodes. This technique is very accurate, but can be poorly tolerated by the patient. The
alternative is to target with the frame in place using
the MRI. This is the technique used in most medical
centers in the United States. The advantage of this
technique is that it is less invasive, well tolerated,
Table 2
Controversies in DBS surgery 2006
Stereotactic surgery
Ventriculography
MRI
Localization
With microelectrode recording
Without microelectrode recording
Targets
GPi versus STN
Stimulation versus ablation
DBS, deep brain stimulation; STN, subthalamic nucleus; GPi,
globus pallidus interna.
512
widely available, and the MRI computer can be used

to make any measurements.
The second controversy is over the use of microelectrode recording to localize the target. Some
targets such as the GPi are relatively large, and electrodes can be placed in them without too much difficulty (Loher et al., 2002). The most commonly used
target, in recent years, is the STN which is no more
than 8 mm in greatest diameter and can be very difficult to hit (Littlechild et al., 2003; Richter et al.,
2004). It is known that the best results in DBS
surgery for Parkinsons disease are achieved not only
when the stimulating electrode is placed in the
greatest portion of the STN, but also when this is
the posteriorlateral portion of the STN (Zonenshayn
et al., 2004). For this type of precision, it is extremely
important to characterize the location and extent of
the STN with multiple passes of a microelectrode
prior to placement of the stimulating electrode. This
painstaking procedure adds considerable time, and
some increased risk to the operation, but most
surgeons feel that the benefits outweigh the risks
(Benazzouz et al., 2002; Cuny et al., 2002; Starr
et al., 2002; Sterio et al., 2002; Nowinski et al., 2003;
Priori et al., 2003).
Various treatment centers also differ over the
selection of targets. The two most widely used targets, in recent years, are the GPi and the STN. The
GPi is larger than the STN and both have distinct
microelectrode recording characteristics. It is felt by
most investigators that stimulation of the STN
improves most of the symptoms of Parkinsons disease including tremor, dyskinesia, motor fluctuations,
rigidity, and bradykinesia (Kumar et al., 1998;
Burchiel et al., 1999; Krause et al., 2001; Peppe
et al., 2004; Anderson et al., 2005; Okun and Foote,
2005). It is generally understood that stimulation of
the GPi is very helpful when the main problem is
dyskinesia, and is less effective than the STN for
the other symptoms. There seems to be more cognitive and behavioral complications with STN stimulation over GPi stimulation (Anderson et al., 2005).
Recent reports, however, have led to some doubt
about the superiority of one site over the other
(Anderson et al., 2005; Okun and Foote, 2005).
In our institution, our technique is the same as that
in the vast majority of centers in the United States
(Ondo and Bronte-Stewart, 2005). We use MRI for
direct targeting, we use micro electrode recording,
and we target the STN. We also prefer DBS rather
than ablation.
J.W. COZZENS
The patients are admitted to the hospital the day

before surgery so that their medication can be withdrawn in a setting where they can receive complete
care. Removal of the Parkinsons disease drugs
will usually result in return of bradykinesia which,
although normally disabling, allows the patient to
remain still in the operating room.
The next morning, the patient is taken to the holding area of the operating room where the stereotactic
base frame is fixed to his head. This frame is an aluminum ring that fits around the patients head and is
fixed to the head with screws that are advanced into
the outer table of the skull with local anesthesia. This
base ring will become the stable reference plane upon
which electrodes are mounted and advanced into target points in the brain. The patient is then brought to
the MRI scanner, and a localizing ring is mounted on
the base ring. The MRI is then obtained with these
rings in place so that the structures inside the brain
can be referenced to the base ring with Cartesian X,
Y, and Z coordinates. When the MRI is obtained,
careful measurements are made of the deepest structures in the brain including the anterior commissure
(AC), the posterior commissure (PC), and the red
nuclei. The line between the AC and PC defines the
ACPC plane which is the reference plane upon
which most stereotactic atlases of the brain are based.
The STN is rarely actually seen when the targeting
MRI is obtained, but its position can be determined
from the position of the ACPC line. From these
measurements, the X, Y, and Z coordinates of the
STN are obtained. We usually make these calculations utilizing the MRI computer terminal but there
are commercially available computer programs
which will assist the surgeon in these calculations
(Fig. 2). In our institution, we find that the procedure
is better tolerated if only one side is implanted during
a hospital stay. These patients are usually elderly and
frail, and can only tolerate a limited amount of time
in the stereotactic frame on the operating table under
local anesthesia.
Some centers have found that it is advantageous to
obtain the MRI scan without the head frame several
days prior to the procedure. All of the targeting and
planning is then performed prior to the day of surgery
and admission. On the day of surgery, the patient has
the stereotactic ring fixed to the skull and then has a
CT scan of his head with the ring in place. The new
CT is then merged with the prior MRI image to obtain
the stereotactic coordinates. With this method, less
time is spent with the patient in the head ring. The
Fig. 2. Computer simulation of electrode placement. This

image is taken directly from a Stealth frameless stereotactic
case targeting the right STN. There are four MRI views including a coronal view (upper left), a sagittal view (upper right), an
axial view (lower left), and a trajectory view in the plane of the
electrode path (lower right). An outline of the deep brain nuclei
taken from the Schaltenbrand and Wharen atlas is displayed by
the computer over the target areas.
disadvantage is that another level of error is added, that

is, the inherent level of accuracy of the MRI targeting
plus the inherent level of accuracy of the program that
merges the two images. Most centers that use this
method do not feel that this is a problem, however.
After targeting, the patient is brought into the
operating room where the scalp behind the forehead
is shaved, prepped, and draped. An incision is made
in the scalp and a burr hole is made in the skull.
A microelectrode is mounted on a microdrive and
placed on the stereotactic frame. With this device,
the microelectrode is slowly advanced through the
burr hole and into the brain using the trajectory determined by stereotactic planning (Fig. 3). As it is
advanced, recordings are made of the neural activity
at specified locations and depths. The STN has a
characteristic neurophysiological signature which is
easily recorded and characterized with microelectrode recordings. The results of these recordings are
plotted on images obtained from the Schaltenbrand
and Wahren (1977) brain atlas.
In our institution, the trajectories for these electrode tracks are planned so that the microelectrode
513
Fig. 3. Patient undergoing microelectrode recording with the

stereotactic frame mounted on his head.
will pass primarily through white matter before it

encounters the STN. To achieve this, we like to have
the electrode approach the STN at a 55 sagittal
angle to the ACPC line. We also plan a coronal
angle from the coronal MRI that will pass outside
the ventricle to avoid passing through the lateral ventricle with its ependymal blood vessels.
Usually several passes are made with the microelectrode in the region of the STN to fully characterize the location and extent of this nucleus. If cells are
recorded several millimeters short of the target with
the characteristics of thalamic neurons and cells are
recorded that are typical of STN cells in the immediate neighborhood of the target, then the next electrode pass is made 1.52.0 mm anterior. Similarly,
if an STN track also has cells typical of caudate
nucleus above the target, the next track is made
1.52.0 mm posterior. Tracks are also made both lateral and medial in the STN to define these borders.
This precise characterization of the STN is accomplished in a methodical manner so that the stimulating electrode can be placed to achieve the greatest
benefit with the fewest adverse effects (Fig. 4).
514
J.W. COZZENS
Fig. 4. Operating room with patient undergoing microelectrode recording of the STN.
We are well aware that there are many centers that

will stop microelectrode recording after cells with
the characteristic STN firing pattern are identified.
We are uncomfortable with this technique since it
does not fully characterize the location and extent
of the STN. We feel that this technique can also lead
to errors such as misplacement of the stimulating
electrode into the red nucleus which can sometimes
have cells with firing patterns similar to the STN.
We feel that three microelectrode passes through
the STN are usually required to find the best location
for the stimulating electrode.
When this mapping is complete, the microelectrode is removed and replaced with the stimulating
electrode. It is important that the stereotactic device
be reliable and accurate enough that the contact
points of the stimulating electrode will ultimately
be in the same optimal location that was defined by
the recordings of the microelectrode. This is not a
trivial point and great care must be made to ensure
that the device that guides and places these electrodes
is reliable and reproducible for both electrodes.
The stimulating electrode is then advanced into
the largest area of the STN. When the stimulating
electrode is in position, it is connected to a pulse
generator and several test stimulations are performed
(Fig. 5). During these stimulations, the patient is
tested for any beneficial changes in movement and
also for unwanted side effects. When the position of
the electrode is found to be satisfactory, it is locked
into place and the wound is closed. The stereotactic

frame is then removed.
The second stage of the operation for placement
and connection of the pulse generator immediately
follows the first stage in most cases. This is performed under general anesthesia and is much like
the operation for implantation of a cardiac pacemaker. Postoperatively, the patient usually stays in
the hospital for 23 days. There is normally a little
localized brain edema, and swelling that surrounds
the newly implanted electrode which makes it impossible to accurately program the pulse generator.
Therefore, the generator is not activated and programmed for 23 weeks after the surgery when this
reactive edema subsides.
36.5. Results
Patients with Parkinsons disease are usually evaluated
using several rating scales. The most widely used
scale is the unified Parkinsons disease rating scale
(UPDRS) (Fahn and Elton, 1987), which provides a
quantitative assessment of mental and motor function
in patients with Parkinsons disease. The UPDRS
consists of six parts which evaluate (i) mental status,
(ii) activities of daily living, (iii) motor function,
(iv) complications of therapy, (v) severity of Parkinsons disease using the Hoen and Yahr (H and Y)
staging (Hoehn and Yahr, 1967), and (vi) disability
using the Schwab and England scale.
515
Fig. 5. Intraoperative testing of the deep brain stimulator electrode array.
Patients are evaluated both in the off and the

on state. These states refer not to the deep brain
stimulator, but to the medication state of the patient.
The off time for a patient is at least 12 h after the
last dose of levodopa, typically in the morning after
omitting the bedtime dose, and after 1 h of rising to
avoid the potentially confounding effects of sleep.
The on state is typically 3060 min after the
patients usual dose of levodopa when both the physician and patient agree that maximal therapeutic
effect is present.
The results of DBS surgery for Parkinsons disease are typically reported by averaging the combined UPDRS scores of patients before surgery both
in the on-drug state, and the off-drug state and comparing these numbers with averages in both states
after surgery. It has been found that some parts of
the UPDRS are affected more by DBS than other
parts. Table 3 summarizes the experience of numerous investigators with DBS of bilateral STN. The
two most commonly reported scores are the improvements in both bradykinesia and dyskinesia. From this
table, one can see that there are major improvements
in both areas. It has been our experience that almost
all patients with electrodes in the STN will have
some improvement in their symptoms including bradykinesia, dyskinesia, and quality of life (Tavella
et al., 2002; Betchen and Kaplitt, 2003; Lezcano
et al., 2004). We also find that the majority of these
patients will have a significant improvement in these

three areas.
Also significant is the number of patients who can
either have a reduction in their anti-Parkinsons medications, or complete cessation (Thobois et al., 2002;
Kleiner-Fisman et al., 2003; Rodriguez-Oroz et al.,
2004). Reports of the percentage of patients who no
longer take levodopa at one year after surgery ranges
from 29% to 53% (Molinuevo et al., 2000; Tavella
et al., 2002; Krack et al., 2003). The reduction of
medication also lowers the overall cost of treatment
of this disease. A number of investigators have noted
that the initial cost of surgery for DBS is more than
offset by the savings in medication costs and overall
care (Spottke et al., 2002; Eskandar et al., 2003;
McIntosh et al., 2003; Charles et al., 2004).
Despite the numerous publications in the past nine
years that usually describe good results with DBS
surgery for Parkinsons Disease, there are very few
double-blind and/or randomized-controlled studies
published (Lang, 2000; Stowe et al., 2003; Goetz
et al., 2005; Hamani et al., 2005). There are also no
published reports of the percentage of patients who
achieve a good outcome versus a bad outcome
with surgery. This is probably due to the fact there
is no established criteria for what should be considered a good outcome. The next phase of outcome
analysis for movement disorders surgery will come
after these criteria are established.
516
J.W. COZZENS
Table 3
Summary of results of DBS of bilateral STN
Reference
Follow-up
(months)
Percent improvement in
off-med motor score (%)
Percent improvement in
dyskinesia disability (%)
Limousin et al., 1998

Kumar et al., 1998
Moro et al., 1999
Pinter et al., 1999
Burchiel et al., 1999
Houeto et al., 2000
Fraix et al., 2000
Molinuevo et al., 2000
Volkmann et al., 2001
The Deep Brain Stimulation for Parkinsons
Disease Study Group, 2001
Starr et al., 2002
Simuni et al., 2002
Welter et al., 2002
Tavella et al., 2002
Kleiner-Fisman et al., 2003
Krack et al., 2003
12
6
12
12
12
6
12
6
12
60
58
36
55
44
67
68
66
67
60
83
90
n/a
n/a
77
93
n/a
83
6
12
12
6
12
12
12
60
24
12
12
51
45
47
65
58
51
66
54
42
64
48
67
n/a
64
69
80
46
68
65
n/a
53
62
Kleiner-Fisman et al., 2004

Zonenshayn et al., 2004
Anderson et al., 2005
DBS, deep brain stimulation; STN, subthalamic nucleus; n/a, not available.
36.6. Other movement disorders

Deep brain stimulation has also been shown to be effective in the treatment of tremor and primary dystonia. In
a prospective, blinded, multicenter study, 22 patients
with primary generalized dystonia underwent bilateral
implantation of deep brain stimulator electrodes into
the GPi. In this study, there was a significant improvement in both dystonia and disability scores that was confirmed with a double-blind crossover assessment done
at three months (Vidailhet et al., 2005). Other movement disorders, such as Tourette syndrome and tardive
dyskinesias, have also shown significant response to
DBS surgery (Houeto et al., 2005; Lenders et al.,
2005). Deep brain stimulation has also been reported
to be effective in the treatment of obsessive-compulsive
disorder (Rauch et al., 2006).
36.7. Adverse events
Complications of DBS can be divided between
those complications specifically associated with the
procedure, those associated with the hardware, and

those associated with the stimulation and programming. One should also remember that there are risks
of injury and disability associated with both the natural
history and medical treatment of Parkinsons disease.
When reviewing the risk of complications associated with the surgical procedure, one must understand that there is an inherent danger with all
neurosurgical procedures and particularly with DBS
surgery for Parkinsons disease. This surgery requires
the blind placement of a series of very sharp electrodes deep into the brain followed by placement of
a blunt electrode. These electrodes can easily pass
through vascular structures which could cause significant hemorrhage and stroke. Despite this theoretical
danger, the actual rate of morbidity and mortality is
quite low. Most investigators report a mortality rate
of less than 1% with this surgery, and the risk of
major hemorrhage and stroke is 38% per patient
and 24% per procedure (The Deep Brain Stimulation for Parkinsons Disease Study Group, 2001;
Pollak et al., 2002; Binder et al., 2005) (most patients
will have two procedures one electrode placed per

side). Other surgical complications include nonhemorrhagic stroke, seizures, infection, confusion, and
pulmonary embolus for a total of 810% (Hariz,
2002; Pilitsis et al., 2005).
Deep brain stimulation also involves implantation of
a significant amount of hardware both in the brain and in
the subcutaneous fat of the chest wall. Like all foreign
objects implanted in the human body, these devices
can harbor and shield bacteria from the natural host
defense mechanisms. Therefore, if an infection develops, the devices usually need to be removed and the
infection treated with six weeks of intravenous antibiotics. The reported infection rates for DBS have been
as high as 12% (Kleiner-Fisman et al., 2003), therefore,
most centers will treat the patients prophylactically with
preoperative and postoperative antibiotics. Occasionally, in elderly and debilitated patients with thin skin
and subcutaneous tissues, there is a significant risk of
erosion of the electrode lead through the skin with
resulting infection. Other hardware complications
include electrode lead fracture or breakage, migration
of leads or pulse generator, and lead disconnection
(Kondziolka et al., 2002; Joint et al., 2002). If the wire
failure is in the electrode which extends into the brain,
then the electrode will need to be replaced, usually with
stereotactic guidance and microelectrode recordings.
Complications related to stimulation and programming usually have to do with placement of the electrode and current spread. The STN is quite small
and often the stimulating current will extend beyond
the nucleus to adjacent areas. Some of these events
can include speech slurring or arrest, dysesthesia,
ocular deviation, weakness, and sudden and profound
sadness or depression. Many of these side effects of
stimulation can be picked up in the operating room
when the stimulating electrode is first tested. If these
adverse effects are noted before the electrode is
secured to the skull, the electrode can be moved to
a new location. If they develop later after surgery, they
can sometimes be reduced with variations in programming of the pulse generator (Krack et al., 2002;
Kumar, 2002). If they cannot be avoided, then the
patient may need to return to surgery for repositioning
of the stimulating electrode.
36.8. Conclusions
Stereotactic surgery for bilateral placement of deep
brain stimulator electrode arrays in the STN for
treatment of the symptoms of Parkinsons disease
517
is a very effective and safe alternative to medical

therapy when that therapy becomes ineffective and
poorly tolerated. The procedure is not without
risk but in most cases will result in a significant
improvement in movement and a reduction in dyskinesia. The risks of the procedure should be carefully
weighed against the risks of not having the procedure
and the benefits of the procedure. Most patients
undergoing this surgery will experience a significant
improvement in activities of daily living and quality
of life (Lezcano et al., 2004).
References
Albe-Fessard, D, Arfel, G and Guiot, G (1961) Identification
et delimitation precise de certaines structures souscorticals
de lhomme par lelectro-physiologie. C. R. Acad. Sci.
(Paris), 243: 24122414.
Alkhani, A and Lozano, AM (2001) Pallidotomy for
Parkinson disease: a review of contemporary literature.
Anderson, VC, Burchiel, KJ, Hogarth, P, Favre, J and
Hammerstad, JP (2005) Pallidal vs subthalamic nucleus
deep brain stimulation in Parkinson disease. Arch.
Neurol., 62: 554560.
Bakay, R and Vitek, J (2000) Rational basis for pallidotomy
in the treatment of Parkinsons disease. In: A Lozano
(Ed.), Movement Disorder Surgery. Karger, Basel.
Benazzouz, A, Breit, S, Koudsie, A, Pollak, P, Krack, P and
Benabid, AL (2002) Intraoperative microrecordings of
the subthalamic nucleus in Parkinsons disease. Mov.
Disord., 17(Suppl. 3): S145S149.
Betchen, SA and Kaplitt, M (2003) Future and current surgical therapies in Parkinsons disease. Curr. Opin.
Neurol., 16: 487493.
Binder, DK, Rau, GM and Starr, PA (2005) Risk factors for
hemorrhage during microelectrode-guided deep brain
stimulator implantation for movement disorders. Neurosurgery, 56: 722732; discussion 722732.
Bucy, PC and Case, TJ (1939) Tremor: physiologic mechanism and abolition by surgical means. Arch. Neurol.
Burchiel, KJ, Anderson, VC, Favre, J and Hammerstad, JP
(1999) Comparison of pallidal and subthalamic nucleus
deep brain stimulation for advanced Parkinsons
disease: results of a randomized, blinded pilot study.
Neurosurgery, 45: 13751382; discussion 13821384.
Charles, PD, Padaliya, BB, Newman, WJ, Gill, CE, Covington, CD, Fang, JY, So, SA, Tramontana, MG, Konrad, PE
and Davis, TL (2004) Deep brain stimulation of the subthalamic nucleus reduces antiparkinsonian medication
costs. Parkinsonism Relat. Disord., 10: 475479.
Clower, WT (2001) Lesions as therapy: rigidity and Parkinsons disease. J. Hist. Neurosci., 10: 93106.
518
Clower, WT (2002) Lesions as therapy: surgical intervention in Parkinsons disease prior to L-DOPA. J. Hist.
Neurosci., 11: 375391.
Cuny, E, Guehl, D, Burbaud, P, Gross, C, Dousset, V and
Rougier, A (2002) Lack of agreement between direct
magnetic resonance imaging and statistical determination of a subthalamic target: the role of electrophysiological guidance. J. Neurosurg., 97: 591597.
Diederich, NJ, Kalteis, K, Stamenkovic, M, Pieri, V and
Alesch, F (2005) Efficient internal pallidal stimulation
in Gilles de la Tourette syndrome: a case report. Mov.
Disord., 20: 14961499.
Eskandar, EN, Flaherty, A, Cosgrove, GR, Shinobu, LA
and Barker, FG, II (2003) Surgery for Parkinson disease
in the United States, 1996 to 2000: practice patterns,
short-term outcomes, and hospital charges in a nationwide sample. J. Neurosurg., 99: 863871.
Fahn, S and Elton, RL and Committee MotUD (1987)
Unified Parkinsons disease rating scale. In: S Fahn,
CD Marsden, DB Calne and M Goldstein (Eds.), Recent
Developments in Parkinsons Disease. Macmillan Health
Care Information, Florham Park, NJ.
Foote, KD, Sanchez, JC and Okun, MS (2005) Staged deep
brain stimulation for refractory craniofacial dystonia
with blepharospasm: case report and physiology. Neurosurgery, 56: E415; discussion E415.
Foote, KD, Seignourel, P, Fernandez, HH, Romrell, J,
Whidden, E, Jacobson, C, Rodriguez, RL and Okun,
MS (2006) Dual electrode thalamic deep brain stimulation for the treatment of posttraumatic and multiple
sclerosis tremor. Neurosurgery, 58(ONS Suppl. 2):
ONS-280285; discussion ONS-285286.
Fraix, V, Pollak, P, Van Blercom, N, Xie, J, Krack, P,
Koudsie, A and Benabid, AL (2000) Effect of subthalamic nucleus stimulation on levodopa-induced
dyskinesia in Parkinsons disease. Neurology, 55:
19211923.
Gildenberg, PL (1993) Stereotaxic versus stereotactic.
Gildenberg, PL (2000) Fifty years of stereotactic and functional neurosurgery. In: DL Barrow, D Kondziolka,
ER Laws Jr. and VC Traynelis (Eds.), Fifty Years of
Neurosurgery. Lippincott, Williams & Wilkins,
Philadelphia.
Goetz, CG, Poewe, W, Rascol, O and Sampaio, C (2005)
Evidence-based medical review update: pharmacological and surgical treatments of Parkinsons disease:
2001 to 2004. Mov. Disord., 20: 523539.
Hamani, C, Richter, E, Schwalb, JM and Lozano, AM
(2005) Bilateral subthalamic nucleus stimulation for
Parkinsons disease: a systematic review of the clinical
literature. Neurosurgery, 56: 13131321; discussion
13211324.
Hariz, MI (2000) Complications of movement disorder surgery and how to avoid them. In: AM Lozano (Ed.), Movement Disorder Surgery. Karger, Basel.
J.W. COZZENS
Hariz, MI (2002) Safety and risk of microelectrode recording in surgery for movement disorders. Stereotact.
Funct. Neurosurg., 78(3/4): 146157.
Hoehn, MM and Yahr, MD (1967) Parkinsonism: onset,
progression and mortality. Neurology, 17: 427442.
Horsley, V (1909) The Linacre Lecture on the function of
the so-called motor area of the brain. Br. Med. J., 2:
125132.
Horsley, V and Clarke, RH (1908) The structure and functions of the cerebellum examined by a new method.
Brain, 31: 45124.
Houeto, JL, Damier, P, Bejjani, PB, Staedler, C, Bonnet,
AM, Arnulf, I, Pidoux, B, Dormont, D, Cornu, P and
Agid, Y (2000) Subthalamic stimulation in Parkinson
disease: a multidisciplinary approach. Arch. Neurol.,
57: 461465.
Houeto, JL, Karachi, C, Mallet, L, Pillon, B, Yelnik, J,
Mesnage, V, Welter, ML, Navarro, S, Pelissolo, A, Damier,
P, Pidoux, B, Dormont, D, Cornu, P and Agid, Y (2005)
Tourettes syndrome and deep brain stimulation. J. Neurol.
Neurosurg. Psychiatry, 76: 992995.
Joint, C, Nandi, D, Parkin, S, Gregory, R and Aziz, T (2002)
Hardware-related problems of deep brain stimulation.
Mov. Disord., 17(Suppl. 3): S175S180.
Kleiner-Fisman, G, Fisman, DN, Sime, E, Saint-Cyr, JA,
Lozano, AM and Lang, AE (2003) Long-term follow
up of bilateral deep brain stimulation of the subthalamic
nucleus in patients with advanced Parkinson disease.
J. Neurosurg., 99: 489495.
Kleiner-Fisman, G, Fisman, DN, Zamir, O, Dostrovsky,
JO, Sime, E, Saint-Cyr, JA, Lozano, AM and Lang,
AE (2004) Subthalamic nucleus deep brain stimulation
for Parkinsons disease after successful pallidotomy:
clinical and electrophysiological observations. Mov.
Disord., 19: 12091214.
Kondziolka, D, Whiting, D, Germanwala, A and Oh, M
(2002) Hardware-related complications after placement
of thalamic deep brain stimulator systems. Stereotact.
Funct. Neurosurg., 79: 228233.
Krack, P, Batir, A, Van Blercom, N, Chabardes, S, Fraix, V,
Ardouin, C, Koudsie, A, Limousin, PD, Benazzouz, A,
LeBas, JF, Benabid, AL and Pollak, P (2003) Five-year
follow-up of bilateral stimulation of the subthalamic
nucleus in advanced Parkinsons disease. N. Engl.
J. Med., 349: 19251934.
Krack, P, Fraix, V, Mendes, A, Benabid, AL and Pollak, P
(2002) Postoperative management of subthalamic
nucleus stimulation for Parkinsons disease. Mov. Disord., 17(Suppl. 3): S188S197.
Krause, M, Fogel, W, Heck, A, Hacke, W, Bonsanto, M,
Trenkwalder, C and Tronnier, V (2001) Deep brain
stimulation for the treatment of Parkinsons disease:
subthalamic nucleus versus globus pallidus internus.
Kumar, R (2002) Methods for programming and patient management with deep brain stimulation of the globus pallidus
for the treatment of advanced Parkinsons disease and
dystonia. Mov. Disord., 17(Suppl. 3): S198S207.
Kumar, R, Lozano, AM, Montgomery, E and Lang, AE
(1998) Pallidotomy and deep brain stimulation of the
pallidum and subthalamic nucleus in advanced Parkinsons disease. Mov. Disord., 13(Suppl. 1): 7382.
Laitinen, LV, Bergenheim, AT and Hariz, MI (1992a)
Leksells posteroventral pallidotomy in the treatment
of Parkinsons disease. J. Neurosurg., 76: 5361.
Laitinen, LV, Bergenheim, AT and Hariz, MI (1992b)
Ventroposterolateral pallidotomy can abolish all parkinsonian symptoms. Stereotact. Funct. Neurosurg., 58:
1421.
Lang, AE (2000) Surgery for Parkinson disease: a critical
evaluation of the state of the art. Arch. Neurol., 57:
11181125.
Lang, AE and Obeso, JA (2004) Challenges in Parkinsons
disease: restoration of the nigrostriatal dopamine system
is not enough. Lancet Neurol., 3: 309316.
Lang, AE and Widner, H (2002) Deep brain stimulation for
Parkinsons disease: patient selection and evaluation.
Mov. Disord., 17(Suppl. 3): S94S101.
Lenders, MW, Buschman, HP, Vergouwen, MD, Steur, EN,
Kolling, P and Hariz, M (2005) Long term results of unilateral posteroventral pallidotomy for antipsychotic drug
induced tardive dyskinesia. J. Neurol. Neurosurg. Psychiatry, 76: 1039.
Lezcano, E, Gomez-Esteban, JC, Zarranz, JJ, Lambarri, I,
Madoz, P, Bilbao, G, Pomposo, I and Garibi, J (2004)
Improvement in quality of life in patients with advanced
Parkinsons disease following bilateral deep-brain stimulation in subthalamic nucleus. Eur. J. Neurol., 11:
451454.
Limousin, P, Pollak, P, Benazzouz, A, Hoffmann, D,
Broussolle, E, Perret, JE and Benabid, AL (1995a)
Bilateral subthalamic nucleus stimulation for severe
Parkinsons disease. Mov. Disord., 10: 672674.
Limousin, P, Pollak, P, Benazzouz, A, Hoffmann, D, Le
Bas, JF, Broussolle, E, Perret, JE and Benabid, AL
(1995b) Effect of parkinsonian signs and symptoms of
bilateral subthalamic nucleus stimulation. Lancet, 345:
9195.
Limousin, P, Krack, P, Pollak, P, Benazzouz, A, Ardouin,
C, Hoffmann, D and Benabid, AL (1998) Electrical
stimulation of the subthalamic nucleus in advanced
Parkinsons disease. N. Engl. J. Med., 339(16):
11051111.
Littlechild, P, Varma, TR, Eldridge, PR, Fox, S, Forster, A,
Fletcher, N, Steiger, M, Byrne, P, Tyler, K and
Flintham, S (2003) Variability in position of the subthalamic nucleus targeted by magnetic resonance imaging
and microelectrode recordings as compared to atlas
co-ordinates. Stereotact. Funct. Neurosurg., 80: 8287.
Loher, TJ, Burgunder, JM, Pohle, T, Weber, S,
Sommerhalder, R and Krauss, JK (2002) Long-term
519
pallidal deep brain stimulation in patients with
advanced Parkinson disease: 1-year follow-up study.
J. Neurosurg., 96: 844853.
Lopiano, L, Rizzone, M, Bergamasco, B, Tavella, A,
Torre, E, Perozzo, P and Lanotte, M (2002) Deep
brain stimulation of the subthalamic nucleus in PD:
an analysis of the exclusion causes. J. Neurol. Sci.,
195: 167170.
Marks, WJ (2005) Deep brain stimulation for dystonia.
Curr. Treat. Options Neurol., 7: 237243.
McIntosh, E, Gray, A and Aziz, T (2003) Estimating the
costs of surgical innovations: the case for subthalamic
nucleus stimulation in the treatment of advanced Parkinsons disease. Mov. Disord., 18: 993999.
Meyers, R (1942) The modification of alternating
tremors, rigidity and festination by surgery of the basal
ganglia. Res. Publ. Assoc. Res. Nerv. Ment. Dis., 12:
602665.
Meyers, R (1958) Historical background and personal
experiences in the surgical relief of hyperkinesia and
hypertonus. In: W Fields (Ed.), Pathogenesis and
Treatment of Parkinsonism. Thomas, Springfield, IL.
Mogilner, AY, Sterio, D, Rezai, AR, Zonenshayn, M,
Kelly, PJ and Beric, A (2002) Subthalamic nucleus
stimulation in patients with a prior pallidotomy. J. Neurosurg., 96: 660665.
Molinuevo, JL, Valldeoriola, F, Tolosa, E, Rumia, J, VallsSole, J, Roldan, H and Ferrer, E (2000) Levodopa withdrawal after bilateral subthalamic nucleus stimulation in
advanced Parkinson disease. Arch. Neurol., 57:
983988.
Moro, E, Scerrati, M, Romito, LM, Roselli, R, Tonali, P
and Albanese, A (1999) Chronic subthalamic nucleus
stimulation reduces medication requirements in Parkinsons disease. Neurology, 53(1): 8590.
Nowinski, WL, Belov, D and Benabid, AL (2003) An algorithm for rapid calculation of a probabilistic functional
atlas of subcortical structures from electrophysiological
data collected during functional neurosurgery procedures. Neuroimage, 18: 143155.
Okun, MS and Foote, KD (2005) Subthalamic nucleus vs
globus pallidus interna deep brain stimulation, the
rematch: will pallidal deep brain stimulation make a triumphant return? Arch. Neurol., 62: 533536.
Ondo, WG and Bronte-Stewart, H (2005) The North American survey of placement and adjustment strategies for
deep brain stimulation. Stereotact. Funct. Neurosurg.,
83: 142147.
Pahwa, R, Wilkinson, SB, Overman, J and Lyons, KE
(2005) Preoperative clinical predictors of response to
bilateral subthalamic stimulation in patients with Parkinsons disease. Stereotact. Funct. Neurosurg., 83:
8083.
Pahwa, R, Lyons, KE, Wilkinson, SB, Simpson, RK, Jr.,
Ondo, WG, Tarsy, D, Norregaard, T, Hubble, JP, Smith,
520
DA, Hauser, RA and Jankovic, J (2006) Long-term evaluation of deep brain stimulation of the thalamus. J. Neurosurg., 104: 506512.
Palur, RS, Berk, C, Schulzer, M and Honey, CR (2002) A
metaanalysis comparing the results of pallidotomy performed using microelectrode recording or macroelectrode stimulation. J. Neurosurg., 96: 10581062.
Parkin, SG, Gregory, RP, Scott, R, Bain, P, Silburn, P,
Hall, B, Boyle, R, Joint, C and Aziz, TZ (2002) Unilateral and bilateral pallidotomy for idiopathic Parkinsons
disease: a case series of 115 patients. Mov. Disord., 17:
682692.
Peppe, A, Pierantozzi, M, Bassi, A, Altibrandi, MG, Brusa,
L, Stefani, A, Stanzione, P and Mazzone, P (2004)
Stimulation of the subthalamic nucleus compared with
the globus pallidus internus in patients with Parkinson
disease. J. Neurosurg., 101: 195200.
Pilitsis, JG, Rezai, AR, Boulis, NM, Henderson, JM,
Busch, RM and Kubu, CS (2005) A preliminary study
of transient confusional states following bilateral subthalamic stimulation for Parkinsons disease. Stereotact.
Pinter, MM, Alesch, F, Murg, M, Seiwald, M, Helscher,
RJ and Binder, H (1999) Deep brain stimulation
of the subthalamic nucleus for control of extrapyramidal features in advanced idiopathic Parkinsons
disease: one year follow-up. J. Neural Transm., 106:
693709.
Pollak, P, Fraix, V, Krack, P, Moro, E, Mendes, A,
Chabardes, S, Koudsie, A and Benabid, AL (2002)
Treatment results: Parkinsons disease. Mov. Disord.,
17(Suppl. 3): S75S83.
Priori, A, Egidi, M, Pesenti, A, Rohr, M, Rampini, P, Locatelli, M, Tamma, F, Caputo, E, Chiesa, V and Barbieri,
S (2003) Do intraoperative microrecordings improve
subthalamic nucleus targeting in stereotactic neurosurgery for Parkinsons disease? J. Neurosurg. Sci.,
47: 5660.
Rauch, SL, Dougherty, DD, Malone, D, Rezai, A, Friehs,
G, Fischman, AJ, Alpert, NM, Haber, SN, Stypulkowski, PH, Rise, MT, Rasmussen, SA and Greenberg,
BD (2006) A functional neuroimaging investigation of
deep brain stimulation in patients with obsessivecompulsive disorder. J. Neurosurg., 104: 558565.
Richter, EO, Hoque, T, Halliday, W, Lozano, AM and
Saint-Cyr, JA (2004) Determining the position and size
of the subthalamic nucleus based on magnetic resonance imaging results in patients with advanced Parkinson disease. J. Neurosurg., 100: 541546.
Rodriguez-Oroz, MC, Zamarbide, I, Guridi, J, Palmero,
MR and Obeso, JA (2004) Efficacy of deep brain stimulation of the subthalamic nucleus in Parkinsons disease 4 years after surgery: double blind and open label
evaluation. J. Neurol. Neurosurg. Psychiatry, 75:
13821385.
J.W. COZZENS
Schaltenbrand, G and Wahren, W (1977) Atlas for Stereotaxy
of the Human Brain; With an Accompanying Guide. Year
Book Medical Publishers, Inc., Chicago, IL.
Siegfried, J and Lippitz, B (1994) Bilateral chronic electrostimulation of ventroposterolateral pallidum: a new therapeutic approach for alleviating all parkinsonian
symptoms. Neurosurgery, 35: 11261129; discussion
11291130.
Simuni, T, Jaggi, JL, Mulholland, H, Hurtig, HI, Colcher,
A, Siderowf, AD, Ravina, B, Skolnick, BE, Goldstein,
R, Stern, MB and Baltuch, GH (2002) Bilateral stimulation of the subthalamic nucleus in patients with
Parkinson disease: a study of efficacy and safety.
J. Neurosurg., 96: 666672.
Spottke, EA, Volkmann, J, Lorenz, D, Krack, P, Smala,
AM, Sturm, V, Gerstner, A, Berger, K, Hellwig, D,
Deuschl, G, Freund, HJ, Oertel, WH and Dodel, RC
(2002) Evaluation of healthcare utilization and health
status of patients with Parkinsons disease treated with
deep brain stimulation of the subthalamic nucleus.
J. Neurol., 249: 759766.
Starr, PA, Christine, CW, Theodosopoulos, PV, Lindsey,
N, Byrd, D, Mosley, A and Marks, WJ, Jr. (2002)
Implantation of deep brain stimulators into the subthalamic nucleus: technical approach and magnetic resonance imaging-verified lead locations. J. Neurosurg.,
97: 370387.
Sterio, D, Zonenshayn, M, Mogilner, AY, Rezai, AR,
Kiprovski, K, Kelly, PJ and Beric, A (2002) Neurophysiological refinement of subthalamic nucleus targeting.
Neurosurgery, 50: 5867; discussion 6759.
Stowe, RL, Wheatley, K, Clarke, CE, Ives, NJ, Hills, RK,
Williams, AC, Daniels, JP and Gray, R (2003) Surgery
for Parkinsons disease: lack of reliable clinical trial
evidence. J. Neurol. Neurosurg. Psychiatry, 74:
519521.
Tavella, A, Bergamasco, B, Bosticco, E, Lanotte, M, Perozzo, P, Rizzone, M, Torre, E and Lopiano, L (2002)
Deep brain stimulation of the subthalamic nucleus in
Parkinsons disease: long-term follow-up. Neurol. Sci.,
23(Suppl. 2): S111S112.
The Deep Brain Stimulation for Parkinsons Disease Study
Group (2001) Deep-brain stimulation of the subthalamic
nucleus or the pars interna of the globus pallidus in Parkinsons disease. N. Engl. J. Med., 345: 956963.
Thobois, S, Mertens, P, Guenot, M, Hermier, M, Mollion,
H, Bouvard, M, Chazot, G, Broussolle, E and Sindou,
M (2002) Subthalamic nucleus stimulation in Parkinsons disease: clinical evaluation of 18 patients.
J. Neurol., 249: 529534.
Vidailhet, M and Pollak, P (2005) Deep brain stimulation
for dystonia: make the lame walk. Ann. Neurol., 57:
613614.
Vidailhet, M, Vercueil, L, Houeto, JL, Krystkowiak, P,
Benabid, AL, Cornu, P, Lagrange, C, Tezenas du
Montcel, S, Dormont, D, Grand, S, Blond, S, Detante,
O, Pillon, B, Ardouin, C, Agid, Y, Destee, A and Pollak, P (2005) Bilateral deep-brain stimulation of the globus pallidus in primary generalized dystonia. N. Engl. J.
Med., 352: 459467.
Volkmann, J, Allert, N, Voges, J, Weiss, PH, Freund, HJ
and Sturm, V (2001) Safety and efficacy of pallidal or
subthalamic nucleus stimulation in advanced PD. Neurology, 56: 548551.
Welter, ML, Houeto, JL, Tezenas du Montcel, S, Mesnage,
V, Bonnet, AM, Pillon, B, Arnulf, I, Pidoux, B,
Dormont, D, Cornu, P and Agid, Y (2002) Clinical
521
predictive factors of subthalamic stimulation in Parkinsons disease. Brain, 125: 575583.
Wichmann, T, DeLong, MR and Vitek, JL (2000) Pathophysiological considerations in basal ganglia surgery:
role of the basal ganglia in hypokinetic and hyperkinetic
movement disorders. In: AM Lozano (Ed.), Movement
Disorder Surgery. Karger, Basel.
Zonenshayn, M, Sterio, D, Kelly, PJ, Rezai, AR and Beric,
A (2004) Location of the active contact within the
subthalamic nucleus (STN) in the treatment of
idiopathic Parkinsons disease. Surg. Neurol., 62:
216225; discussion 225226.

M.R. Nuwer (Ed.)
522
CHAPTER 37
Mapping and monitoring for brainstem lesions

Georg Neuloh*, Christian Strauss and Johannes Schramm
Neurochirurgische Universitatsklinik, D-53105 Bonn, Germany
37.1. Brainstem mapping
37.1.2. Stimulation technique
37.1.1. Introduction
Various stimulation techniques can be used (Strauss

et al., 1993; Eisner et al., 1995; Morota et al., 1995;
Chang et al., 1999). Constant voltage, constant current, bipolar or monopolar techniques have all been
successfully employed. Constant current bipolar
stimulation with rectangular monophasic impulses
(a stimulus duration of 100400 ms, a stimulus frequency not exceeding 10 Hz and intensities up to
25 mA) has proved reliable and safe and has been
used by our group since 1996 (Strauss et al., 1999).
Following surgical exposure of the rhomboid fossa,
a hand-held stimulation probe is moved on the surface of the rhomboid fossa under constant electrical
stimulation using rather high stimulation intensities
(2 mA). The stimulation tip should have a rather
planar area in order to avoid injury to the ependyma
and underlying tissue (Fig. 3) (Strauss et al., 1993).
The short distance, for example of 0.25 mm, from
the ependyma to the ascending fibers of the facial
nerve within the facial colliculus has to be considered
(Strauss and Fahlbusch, 1997; Strauss et al., 1997).
Supramaximal stimulation considerably shortens the
period of mapping and easily identifies structures of
the facial colliculus within several seconds, but by
nature is not specific. With the area of interest identified, stimulation intensity is then turned down to
threshold levels, which in a normal anatomical situation usually measures around 0.05 mA, when the colliculus is the target. With the threshold technique, the
area of the shortest distance between facial nerve
fibers and the ependyma can be selectively identified
(Strauss et al., 1999). Usually the area of shortest distance can be identified paramedian to the median sulcus, where the ascending fibers of the facial nerve
come closest to the surface (0.25 mm) (Strauss
et al., 1997). With marginal variation of the stimulation intensity, a road map of the facial nerve fibers
can be drawn (Fig. 1). With the same technique,
other motor nerve nuclei (V, VI, IX/X, and XII) can
Surgical treatment of brainstem lesions is limited by

dense concentration of functionally important neural
pathways, nuclei, and fibers in pons variolii and
medulla oblongata and the lack of reliable visible
anatomical landmarks (Lang et al., 1991). Since most
brainstem lesions are approached via the IVth ventricle, the presence of functionally intact brain tissue
between ependyma and lesion poses already a potential risk due to the surgical corridor itself. In pontine
lesions, facial and abducens nerve deficits as well
as conjugated gaze palsies can affect the quality of
life. Deficits of lower cranial nerve nuclei (nucleus
n.hypoglossii, nucleus ambiguus, nucleus dorsalis
n.vagi as the parasympathetic motor nucleus of the
vagal nerve) are severe life-threatening complications following surgery of medullary lesions. Careful
anatomical studies have defined safe entry zones into
the brainstem above and below the facial colliculus
measuring several millimeters (Fig. 1) (Kyoshima
et al., 1993; Bogucki et al., 1997a,b, 2000; Strauss
et al., 1997). The effect of space occupying intrinsic
lesions on the superficial anatomy of the rhomboid
fossa, however, limits the value of these morphometric investigations (Strauss et al., 1993, 1997).
Direct electrical stimulation during surgery has
emerged as a reliable and safe technique for identification of superficially located nuclei and fibers in
order to define safe entry zones into pons and
medulla (Figs. 13) (Strauss et al., 1993, 1999;
Eisner et al., 1995; Morota et al., 1995, 1996; Chang
et al., 1999; Morota and Deletis, 2006).
*
Correspondence to: Georg Neuloh, M.D., Neurochirurgische Universitatsklinik, D-53105 Bonn, Germany.
Tel.: 49-228-287-6521; fax: 49-228-287-4758.
E-mail: georg.neuloh@ukb.uni-bonn.de (G. Neuloh).
IV
14
mm
VI
VII
9
mm
XII
Fig. 1. Anatomical schematic drawing of superficially located motor nuclei and fibers of the floor of the rhomboid fossa
(left side) with corresponding operative site (right side). The nuclei and fibers were mapped using electrical stimulation and
indicated by black silk sutures.
Fig. 2. Selective stimulation of the VIth nerve nucleus. Multiple EMG traces of various target muscles demonstrate selective stimulation effect.
Fig. 3. Supracollicular approach to a recurrent symptomatic pontine cavernous hemangioma. The patient had been operated
upon for this lesion several years ago using an approach above the right facial colliculus. Following localization with the
hand-held stimulator (upper left), the hematoma was released (upper right) and the vascular malformation subsequently
removed (lower left and right).
524
be localized. Stimulation intensity has to be adjusted

accordingly, since these nuclei are located further
away from the ependyma surface. For identification
of the nucleus of the hypoglossal nerve, stimulation
intensity of 1 mA and more have to be employed.
It has also to be taken into consideration that higher
stimulation intensities are less specific and may
result in diffuse stimulation effects. We have never
employed stimulation intensities above 5 mA, except
for one patient who was operated upon for a primary
hematoma of the pons. In this patient, we used up
to 10 mA stimulation intensity. Stimulation intensities may vary according to the underlying disease.
In tumors infiltrating the floor of the fourth ventricle,
higher intensities may be required when stimulation
is applied through the tumor tissue, in order to guide
the extent of removal, for example, in primitive neuroectodermal tumor (PNET) and ependymomas.
37.1.3. Recording technique
Recording electrodes are placed into the target muscles. We use a pseudobipolar recording setup with
two electrodes for each muscle. These electrodes
are placed 510 mm apart and are secured by tapes
in order to avoid dislocation (Strauss et al., 1999;
Romstock et al., 2000). For the lateral rectus muscle,
we use insulated electrodes to avoid unintended
recordings from the orbicularis oris muscle. For
lower cranial nerve monitoring, electrodes are placed
into the soft palate and into the pharyngeal wall. For
vocal cord monitoring, we have directly placed electrodes into the vocalis muscle under laryngoscopic
control, but lately have exclusively relied on endotracheal tube electrodes. For hypoglossal nucleus localization, we apply two electrodes into the tip of the
tongue. For practicability, it is sufficient to use one
pair of electrodes for hypoglossal nuclei, since the
nuclei of both sides are located extremely close
together (0.6 mm). The distance of the upper pole
of the hypoglossal nerve nuclei from the ependyma
measures 0.55 mm. This results in a stimulation
threshold of 12 mA. This stimulus intensity evokes
responses from both nuclei. Other than that we record
activity separately from both sides. For the trigeminal
motor nucleus, we apply electrodes into the masseter
muscle. With a standard two-channel recording setup
and referential recordings, motor responses from the
orbicularis oris muscle may prove difficult to differentiate. We therefore routinely record facial nerve
activity from all three branches of the facial nerve.
G. NEULOH ET AL.
Multichannel recordings are required for most brainstem lesions in order to simultaneously record activity
from all possible target muscles. This ensures selectivity of responses and proved time saving (Fig. 2).
We therefore do not advocate the standard twochannel recording procedure techniques as for example in acoustic neuroma surgery, although in selected
cases reduction to two channels may be sufficient
(Fig. 3).
37.1.4. Anesthesia
The anesthetic regimen is based on total intravenous
anesthesia (TIVA). Following induction with midazolam, nitrous oxide and a short-acting muscle relaxant
(atracurium besilate) anesthesia was maintained with
propofol (612 mg/kg/h) and alfentanil (60 mg/kg/h).
Muscle relaxing agents were not given until the very
end of the surgical procedure (Strauss et al., 1999;
Romstock et al., 2000). With respect to cost management, volatile anesthetic regimens have also proved
to have no negative influence on the reliability of
stimulation (Ruskin et al., 1994; Morota et al., 1995).
37.1.5. Safety aspects
Cardiovascular side effects of electrical stimulation
are of particular concern, since the parasympathetic
motor nucleus dorsalis n.vagi is located directly lateral to the trigonum hypoglossi. With the stimulation
parameters outlined above, a short run of asymptomatic ventricular extrasystoles was observed in a single
patient of a series of more than 100 cases immediately following stimulation. Other than that measurable effects regarding cardiac arrhythmia or blood
pressure were never observed since introduction of
this method in 1991.
Possible brain damage with respect to electrical
stimulation is mainly caused by an imbalance of
the bloodbrain barrier. Damage of the barrier is
dependent on the applied charge density (Coulomb
(mC)/cm2 phase) under the stimulation electrode
(Agnew and McCreery, 1987; Gordon et al., 1990).
From histology of epileptogenic tissue, which was
chronically stimulated prior to resection, it is known
that up to a charge density of 57 mC/cm2 phase no
histological changes occur (Agnew and McCreery,
1987; Gordon et al., 1990). For constant voltage stimulation, the electrode surface of 0.0043 cm2 produces a
charge density of 0.59 mC/cm2 phase (Mller and
Jannetta, 1984). Constant current stimulation produces
charge densities up to 0.0014 mC/cm2 phase. For

cortical stimulation in surgery of centrally located
tumors and in epilepsy surgery, charge densities up to
several 100 mC/pulse/cm2 are considered safe (Cedzich et al., 1998). Limitation of electrical stimulation
to 1 V (constant voltage technique) and 2 mA (constant current technique) has no noticeable adverse side
effects regarding the bloodbrain barrier (Strauss
et al., 1999). The limitation of the stimulus frequency
to 10 Hz is based on animal experiments in which
changes of cerebral blood flow, blood pressure, and
intracranial pressure were provoked by using stereotactically implanted electrodes into the dorsal and
rostral areas of the formatio reticularis. Maximal
effects were seen with 4050 Hz, although when using
high stimulus intensities (50 mA) effects were
observed with as low as 2 Hz with limitation exclusively to the time period of stimulation (Iadecola
et al., 1983).
37.1.6. Site of stimulation
Precise location of the stimulation site is a prerequisite for the clinical use of this technique. Since
peripheral electromyographic (EMG) responses are
obtained with stimulation intensities as low as
0.1 mA and a stimulation duration of 100 ms, the
peripheral motor neuron seems most likely the site
of stimulation. These parameters are similar to those
for identification of the facial nerve during cerebellopontine angle surgery (Mller and Jannetta, 1984;
Romstock et al., 2000). These parameters are inadequate for stimulation of the first motor neuron
(Cedzich et al., 1998). For direct cortical stimulation
of the pyramidal tract, the stimulation intensity usually varies between 2 and 25 mA with frequencies
above 4050 up to 500 Hz using a stimulation duration between 100 and 400 ms (Cedzich et al., 1998).
A further peripheral stimulation toward the direction
of root exit zones is unlikely, since stimulation would
include other peripheral motor neurons of the
areas of the cranial nerves. This was excluded by
simultaneous bilateral multichannel EMG recordings
demonstrating selective responses on the side and site
of stimulation (Fig. 2). The selectivity of responses
also excludes diffuse brainstem stimulation. The
site can be directly calculated with the equation
(I K d2), using the actual motor threshold for
the facial nerve fibers and the fibers of the nucleus
n.hypoglossii (Strauss et al., 1997, 1999) and the anatomically measured distances between the fibers and
525
the ependyma surface (Strauss and Fahlbusch, 1997;

Strauss et al., 1997). The clinical observation by
Morota et al. using high dosage of volatile anesthetic
without negative effect on the stimulation results also
point to the peripheral motor neurons as the site of
stimulation, since volatile anesthetics have a negative
effect on the excitability of the first motor neuron,
where as its influence regarding the excitability
of the peripheral motor neuron can be neglected
(Ruskin et al., 1994; Morota et al., 1995; Cedzich
et al., 1998). Most likely, the axonal cone with an
electric threshold 10 times lower than the axons
threshold represents the presumed site of stimulation
(Strauss et al., 1999).
37.1.7. Clinical value and review of literature
Since the technique of intraoperative localization has
rapidly evolved as a mandatory adjunct for brainstem
surgery, a prospective study on its clinical value has
never been published. Apart from brainstem
cavernous hemangiomas (Fig. 3), for which the
technique was originally designed (Fahlbusch et al.,
1991; Houtteville, 1995; Samii et al., 2001;
Sandalcioglu et al., 2002), brainstem tumors such as
gliomas and ependymomas are of particular interest
(Fig. 4), since the extent of surgical removal has been
identified as the major prognostic factor (Pollack
et al., 1993; Van Veelen-Vincent et al., 2002). Based
on our experience, we have evaluated all surgically
treated brainstem tumors between 1991 and 1999,
which had been operated upon using this technique
and compared their outcome with a matched (histology, location) series of patients undergoing surgery
without mapping. Functional outcome of electrically
localized motor nuclei and fibers (cranial nerves V,
VI, VII, IX/X, and XII) were investigated before,
immediately after surgery and on follow up after
one year (Fig. 5). The study population consisted of
29 patients in each group (7 PNET, 12 medial ependymomas, 10 gliomas of various growth patterns)
with an average age of 23.5 years in the mapping
group versus 29 years in the nonmonitored group,
tumor size averaged at 3.5 cm each, and 18 patients
in both groups underwent postoperative radiotherapy.
Six patients of the mapping group received chemotherapy, as compared to 7 non-mapped patients.
Before surgery, the absolute number of cranial nerve
deficits was comparable. After surgery, both groups
deteriorated with respect to their cranial nerve deficits, which was to be expected, but after one year
526
G. NEULOH ET AL.
Fig. 4. Large pilocytic astrocytoma on axial and sagittal T1-weighted MRI images with contrast enhancement, before (upper)
and after surgical removal (lower). The surgical corridor is clearly visible (lower left). Postoperative persistent morbidity was limited to a motor weakness of the trigeminal nerve, corresponding to the surgical corridor at the lateral edge of the rhomboid fossa.
mapped patients showed significantly better functional

results (w2 test: p 0.003). In those patients undergoing surgery without the application of intraoperative
mapping, postoperative cranial nerve deficits remained
fixed and did not recover over the observation period
(Fig. 5). Dorsally exophytic brainstem gliomas and

focal endophytic gliomas as well as low grade ependymomas seem to benefit from careful intraoperative
mapping (Fig. 4). These results, although based on a
study design implying a low evidence level, underline
Fig. 5. Total number and course of cranial nerve deficits (V, VI, VII, IX/X, and XII) as evaluated in a retrospective analysis of
58 patients with brainstem tumors (glioma, ependymoma, PNET) undergoing surgery with and without brainstem mapping.
the neurosurgical impact of this technique. Direct

electrical brainstem stimulation for mapping of the
floor of the IVth ventricle is the neurophysiologic
method with the fastest and widest acceptance in neurosurgery. Two years following its first description in
1993 (Strauss et al., 1993), this technique has been
stated as mandatory for surgical management of brainstem lesions (Bricolo and Turazzi, 1995). Based on
published experiences of several groups, the technique
has gained wide acceptance, because it is easy to
apply, safe, and provides reliable information on the
location of motor brainstem fibers and nuclei (Strauss
et al., 1993; Eisner et al., 1995; Morota et al., 1995,
1996; Chang et al., 1999; Strauss et al., 1999; Morota
and Deletis, 2006). It must be concluded that identification of superficially located motor nuclei prior to
microsurgical removal of brainstem lesions does not
exclude functional morbidity of identified structures
following surgery (Morota et al., 1996). The technique
of threshold stimulation is a method for localization,
but not for intraoperative monitoring. Continuous multichannel EMG recordings and analysis of these data
have yet failed to identify pathological EMG activity,
indicating damage to monitored structures. The presence of prolonged EMG activity (bursts and spikes)
has been associated with postoperative morbidity
(Grabb et al., 1997); however, in our own data, we
have not been able to reliably associate presence of
spikes and bursts with cranial motor nerve deficits
after surgery. A specific EMG pattern, comparable to
the A-train indicating pending damage to brainstem
nuclei has so far not been identified (Romstock et al.,
2000). At present, any prolonged EMG activity has to
be considered potentially harmful. Future research
should be directed toward detailed software-based
analysis of intraoperative EMG in order to reliably
apply this technique for continuous monitoring during
removal of brainstem lesions.
37.2. Brainstem motor-tract monitoring
37.2.1. Introduction
Continuous monitoring of sensory evoked responses
[somatosensory evoked potential (SEP) and BAEP,
rarely trigeminal sensory responses (Soustiel et al.,
1993)] during brainstem surgery was described early
(Albright and Sclabassi, 1985) and is still a commonly
applied technique for brainstem monitoring, although
reports on clinical applications remain anecdotal
(Gentili et al., 1985; Kohno et al., 1993; Wagner
527
et al., 1994; Pechstein et al., 1997; Cedzich et al.,

1999; Anderson et al., 2003). However, the tegmental
somatosensory and auditory pathways do not cover
an area of the brainstem large enough (Fahlbusch and
Strauss, 1991) to represent its general functional state
with a high sensitivity, although combined SEP and
BAEP monitoring has an increased sensitivity for, for
example, brainstem ischemia as compared with either
method alone (Manninen et al., 1994). Specifically,
the functional integrity of the corticospinal fibers cannot be validly monitored by sensory evoked potentials.
It must be remembered that the pyramidal tract courses
through the ventral cerebral peduncles, the fibrae longitudinales of the pars ventralis pontis, and the likewise
ventrally located decussatio pyramidorum of the
medulla oblongata, clearly separated from the bulbothalamic and spinothalamic fibers of the lemniscus
medialis and from the auditory pathways and nuclei
of the lateral lemniscus and the corpus trapezoideum,
also with a partially differing (ventromedial vs. lateral
perforators) blood supply, particularly at the pontine
level. The same is true for the corticonuclear fibers of
the mesencephalic and upper rhombencephalic pyramidal tract. Reliable monitoring of the motor pathways functional integrity under general anesthesia is
a more recent development (Deletis, 1993; Taniguchi
et al., 1993; Pechstein et al., 1996; Deletis and Kothbauer, 1998), and has been extensively explored in
the neurosurgical context for supratentorial brain and
spinal cord surgery (see Chapters 1518 and 21 in this
volume). The application of motor evoked potential
techniques for brainstem monitoring has been discussed more sporadically (Cedzich et al., 1999; Neuloh and Schramm, 2004b; Sala et al., 2004; Akagami
et al., 2005; Dong et al., 2005; Glasker et al., 2006).
Here, we will briefly summarize methodological and
clinical features of the technique which are important
for brainstem monitoring.
Whereas brainstem monitoring is of obvious importance in child neurosurgery, there are no truly childspecific aspects of brainstem motor monitoring except
for the more difficult elicitation of motor responses in
young children which require the most careful selection
of optimum stimulation and recording parameters in
order to obtain reliable motor responses.
37.2.2. Motor evoked potential stimulation
Transcranial motor evoked potential pulse-train stimulation (Pechstein et al., 1996; Neuloh and Schramm,
2004b), in general as applicable for posterior fossa
528
approaches, is described in detail elsewhere in this

volume and cannot be fully discussed here. The electrode montage is at sites about 12 cm anterior to the
C1-C2 or C3-C4 positions of the International 10
20 system. A C3/4-Cz 2 cm electrode montage
has proven optimal for specific corticobulbar tract
monitoring (see below). More lateral positions provide a more efficient stimulation of the corticospinal
and particularly of the corticobulbar tracts, but the
movement artifact typically associated with transcranial electrical stimulation may be more pronounced.
Constant current stimulation is preferred for a better
control of current distribution and a lesser twitching
artifact. Since upper extremity muscle motor evoked
potentials (MEPs) are typically recorded for brainstem monitoring, current intensities of 100 mA or
less are sufficient to obtain the desired just-above
threshold motor responses. Supramaximal stimulation is discouraged to avoid caudal bypassing of the
target pathways in particular with mesencephalic
procedures. If surgery is performed in a sitting or
semi-sitting position, subdural air collection after
durotomy may dramatically increase the electrical
impedance (Kombos et al., 2000). More lateral electrode positions are preferred over a very strong stimulation intensity to overcome the high apparent
motor threshold for the above reasons. Too lateral
electrode positions caudally to the target structures
may in theory occur in midbrain surgery, but the
typical approaches these procedures are usually not
performed in a sitting position. Successful MEP
monitoring may be impossible in up to 10% of cases
mainly due to the problem of subdural air collection.
The frequency of MEP stimulation depends on the
monitoring program implemented in a given setting.
Alternating only with SEPs, a stimulation rate of
about 2 per minute can be achieved, in critical situations MEP alone may be performed at a higher rate.
If BAEPs are recorded as well, MEPs may be
obtained only every few minutes (Dong et al.,
2005), although every attempt must be made to
increase the update speed since real-time monitoring
as required for useful surgery monitoring cannot be
realized at such a low stimulation rate.
37.2.2.1. Corticospinal tract mapping
With lesions displacing the corticospinal tract or distorting the local peduncular or pontine topography,
direct stimulation of the motor fibers via a hand-held
probe can be used, employing the multipulse MEP
technique to map the motor tract (Cedzich et al.,
G. NEULOH ET AL.
1998; Sala et al., 2004; Quinones-Hinojosa et al.,

2005) for identification of safe entry zones for the surgical approach. As opposed to cortex stimulation,
cathodal stimulation may be most effective (Cedzich
et al., 1998). It must be noted that this mapping procedure differs fundamentally from, and cannot replace
continuous motor monitoring which is rather intended
to pick up impending motor damage at unexpected
instances.
There are no aspects of anesthesia and safety with
MEP stimulation which are truly specific to brainstem monitoring; therefore, we refer the reader to
the respective chapters in this volume.
37.2.3. MEP recording and interpretation
In our experience, recording of MEPs from distal
upper limb muscles is adequate for brainstem corticospinal tract monitoring. We prefer subcutaneous
needle electrodes placed in a muscle-tendon fashion
in order to pick up compound muscle responses
(Neuloh and Schramm, 2004b). As it is well known
from MEP monitoring for pericentral tumors, motor
responses can be usefully recorded after direct cortical stimulation from facial muscles (Fig. 6, Neuloh
and Schramm, 2002, 2004b; Neuloh et al., 2004) to
avoid lesions of the corticonuclear projections with
lateral resection. Similarly, continuous corticobulbar
tract monitoring can be achieved during brainstem
surgery with transcranial stimulation and recording
Fig. 6. Facial muscle response to transcranial pulse-train

stimulation of the corticonuclear tract in comparison with
arm and leg muscle responses. Note the large stimulus artifact and the clearly discernible latency around 15 ms which
excludes peripheral cranial nerve stimulation.
529
trade-off) in (Dong et al., 2005). Since corticonuclear

MEP responses do not seem to represent all of the
target fibers, mild paresis despite preserved responses
as well as partially preserved function despite
irreversible loss may occur (Dong et al., 2005).
In spinal cord monitoring (see Chapter 16 in this
volume), D-wave monitoring via epidural electrodes
provides a semi-quantitative amplitude criterion for
assessment of corticospinal tract function, whereas
only the presence or absence of muscular motor
responses is useful parameter of motor function due
to the propriospinal supportive motor system
(Kothbauer et al., 1997; Deletis, 2002; Sala et al.,
2004). Although a typical median suboccipital craniotomy would allow for placement of cervical epidural
electrodes, brainstem surgery does not necessarily
require this additional recording technique due to the
above semi-quantitative relation between muscular
MEP responses and motor function comparable to
supratentorial surgery (Neuloh and Schramm, 2004b;
Neuloh et al., 2004; Akagami et al., 2005; Dong
et al., 2005).
from facial and lower cranial nerve target muscles

(Sala et al., 2004; Akagami et al., 2005; Dong
et al., 2005), possibly providing a solution for the
uncertainties associated with continuous cranial
nerve EMG monitoring for brainstem surgery as discussed above in part I of this chapter. Direct peripheral cranial nerve excitation can easily occur with
the transcranial stimulation technique and would lead
to false-negative monitoring results with brainstem
surgery. Therefore, a low suprathreshold stimulation
intensity is mandatory for this corticobulbar monitoring technique. A single pulse stimulation at the
intended current intensity for corticobulbar tract
excitation can exclude peripheral nerve stimulation if
it does not evoke a motor response, which would
require transsynaptic multipulse train stimulation of
the bulbar motoneuron as opposed to the peripheral
nerve. However, if a clearly discernible typical MEP
onset latency around 1216 ms can be measured, the
corticobulbar pathway must be involved and direct
peripheral cranial nerve excitation which yields motor
responses after 57 ms is very unlikely (Dong et al.,
2005).
The MEP amplitudes which are achieved with
transcranial stimulation are less stable than after
direct cortical stimulation. Therefore, the typical
50% amplitude criterion for significant MEP decrement which has proven valid in our experience
(Fig. 7, Neuloh and Schramm, 2002, 2004a,b; Neuloh
et al., 2004) and seems to apply also for brainstem
monitoring (Akagami et al., 2005) is not always
applicable. In particular for corticobulbar tract monitoring, where lower amplitudes and a lower signal-tonoise ratio is achieved, a somewhat tighter criterion
must be chosen, for example, the 35% criterion
which proved most valid (best sensitivity-specificity
Intraoperative
MEP Findings
37.2.4. Clinical applications

37.2.4.1. Indications for brainstem motor monitoring
It is obvious from the above anatomical considerations that corticospinal/corticonuclear monitoring is
mandatory with lesions within or in close adjacency
to the ventral or ventrolateral aspect of the brainstem.
This is true for both intra- and extraaxial lesions,
including resectable gliomas, ependymomas, large
petroclival and foramen-magnum meningiomas, epidermoids, cavernomas, arteriovenous vascular malformations, and very large vertebral or basilar artery
aneurysms (Figs. 8 and 9). The ventral approach
Permanent new
deficit
Motor Outcome
Transient new
deficit
No new deficit
Irreversible Loss
always
Irreversible
Deterioration
frequent
frequent
rare
Reversible Loss
rare*
frequent
frequent
Reversible
Deterioration
rare*
frequent
frequent
Unaltered
never
never**
always
*not yet observed in brainstem surgery

**exception: mild facial paresis
Fig. 7. Qualitative correlation of intraoperative MEP findings and motor outcome as typically observed during supratentorial
brain surgery and apparently applicable also for brainstem procedures. Modified with permission from Neuloh and Schramm
(2004b).
530
G. NEULOH ET AL.
Thenar MEPs
Right
Left
5 mV
OP Start
20 ms
Opening Dura
CSF Outflow
Dissecting
ependymoma
Preoperative MRI
Warning
500 V
End resection
OP End
Postoperative MRI
20 ms
Fig. 8. During surgery for an ependymoma of the cervicomedullary junction, resection was halted with MEP deterioration.
MEPs recovered partially, and there was no new motor deficit postoperatively. Follow-up imaging confirmed the intraoperative impression that complete tumor resection had been achieved at the point of MEP deterioration. Modified with permission from Neuloh and Schramm (2004b).
determines the risk for brainstem motor function as

much as a possible resection in adjacency to the
motor fibers. One major concern, for example, with
foramen magnum meningiomas is a possible affection of the vascular supply of the motor tract which
has very little collateralization in this area.
In our view, a targeted indication for any monitoring technique is desirable for reliable results rather
than the accumulation of all available methods in
all posterior fossa cases. We do not perform routine
MEP/SEP monitoring in, for example, typical cerebellopontine angle tumor surgery, even with large vestibular schwannomas, where BAEP and cranial nerve
mapping/monitoring provides adequate monitoring.
Likewise, surgery via a rhomboid fossa approach
for tegmental intrinsic tumors or cavernomas will
require MEP monitoring only if these lesions extend
deep into the brainstem. However, extended corticonuclear MEP monitoring (Dong et al., 2005) is a possible future routine application with such lesions, and
preserved SEPs may indirectly indicate preserved
functional integrity of the tegmentum.
37.2.4.2. Impact on surgery

In our experience, ischemia arises with inadequate
retraction, electrocoagulation, or manipulation of
vessels during dissection, and is a major cause for
new motor deficit in brainstem surgery. Venous
congestion can cause delayed deficit and must be
carefully avoided as well (Strauss et al., 2000).
Therefore, if MEPs deteriorate, dissection is temporarily halted, retractors are released and possibly
readjusted, the situs is checked and irrigation is
applied as well as papaverine to possibly spastic
vessels. With early intervention, MEP deterioration
proves reversible in most cases and no permanent
new deficit must be expected.
If MEPs deteriorate during resection of a brainstem tumor, the above measures should be considered to reverse the MEP change, before opting for
definite cessation of tumor resection close to the
motor pathways if MEP deterioration proves irreversible or reoccurs after some initial recovery. In turn,
very much like in supratentorial surgery, stable
recordings allow safe maximization of resection
531
Thenar MEPs
Median nerve SEPs
OP start
dissecting vertebral
artery loop
temporary clip on
brainstem perforator
Preoperative MRI
angiolysis
1 V
2 mV
closing
10 ms
Preoperative DSA
OP end
10 ms
Fig. 9. In a patient with chronic left hyperpathia and mild spasticity, a large vertebral artery loop compressing the medulla
oblongata was dissected from the brainstem and cushioned with a Teflon felt implant under SEP and MEP monitoring.
MEPs showed some mild instability throughout the dissection procedure but did not deteriorate further during temporary
clipping of a brainstem perforator, allowing for safe completion of this critical step of the procedure. The outcome was
uneventful. Reprinted from Neuloh and Schramm (2004b) with permission from Springer, Wien, New York.
without the fear of impending new deficit (Neuloh

and Schramm, 2004b; Neuloh et al., 2004).
The data available so far do not allow a definite
answer to the question whether motor monitoring
improves the outcome of brainstem surgery. The
multitude of factors involved including the impact
of other monitoring and mapping techniques typically applied at the same time would require a very
large number of cases in a possible prospectively
controlled study, although this might be the way to
go in the future. For the time being, it appears prudent enough to conclude from recent retrospectively
controlled data for spinal cord surgery (Sala et al.,
2006) that MEP monitoring can significantly
improve the functional and surgical outcome also in
brainstem surgery.
References
evoked potentials. Neurosurgery, 20(1): 143147.
Akagami, R, Dong, CC and Westerberg, BD (2005) Localized transcranial electrical motor evoked potentials
for monitoring cranial nerves in cranial base surgery.
Neurosurgery, 57(Suppl. 1): 7885; discussion 7885.
Albright, AL and Sclabassi, RJ (1985) Use of the
Cavitron ultrasonic surgical aspirator and evoked
potentials for the treatment of thalamic and brainstem tumors in children. Neurosurgery, 17(4):
564568.
Anderson, RC, Dowling, KC, Feldstein, NA and Emerson,
RG (2003) Chiari I malformation: potential role for
intraoperative electrophysiologic monitoring. J. Clin.
532
Bogucki, J, Gielecki, J and Czernicki, Z (1997a) The anatomical aspects of a surgical approach through the floor of
the fourth ventricle. Acta Neurochir. (Wien), 139(11):
10141019.
Bogucki, J, Gielecki, J and Czernicki, Z (1997b) Digitalimage analysis of the rhomboid fossa surface. Folia
Morphol. (Warsz), 56(3): 129135.
Bogucki, J, Czernicki, Z and Gielecki, J (2000) Cytoarchitectonic basis for safe entry into the brainstem. Acta
Neurochir. (Wien), 142(4): 383387.
Bricolo, A and Turazzi, S (1995) Surgery for gliomas and
other mass lesions of the brainstem. Adv. Tech. Stand.
Neurosurg., 22: 261341.
Cedzich, C, Pechstein, U, Schramm, J and Schafer, S (1998)
Electrophysiological considerations regarding electrical
stimulation of motor cortex and brainstem in humans.
Cedzich, C, Pechstein, U, Zentner, J and Van Roost, D
(1999) Minimally invasive stereotactically-guided extirpation of brainstem cavernoma with the aid of electrophysiological methods. Minim. Invasive Neurosurg.,
42(1): 4143.
Chang, SD, Lopez, JR and Steinberg, GK (1999) Intraoperative electrical stimulation for identification of cranial nerve nuclei. Muscle Nerve, 22(11): 15381543.
63: 201214.
of the motor system. In: V Deletis and J Shils (Eds.),
Neurophysiology in Neurosurgery.Academic Press,
London, pp. 2551.
Deletis, V and Kothbauer, K (1998) Intraoperative neurophysiology of the corticospinal tract. In: E Stalberg, HS Sharma
and Y Olsson (Eds.), Spinal Cord Monitoring. Springer,
Wien, New York, pp. 421444.
transcranial electrical stimulation during skull base surgery. Clin. Neurophysiol., 116(3): 588596.
Eisner, W, Schmid, UD, Reulen, HJ, Oeckler, R, OlteanuNerbe, V, Gall, C and Kothbauer, K (1995) The
mapping and continuous monitoring of the intrinsic
motor nuclei during brainstem surgery. Neurosurgery,
37(2): 255265.
Fahlbusch, R and Strauss, C (1991) Surgical significance of
cavernous hemangioma of the brainstem. Zentralbl.
Neurochir., 52(1): 2532.
Fahlbusch, R, Strauss, C and Huk, W (1991) Pontine-mesencephalic cavernomas: indications for surgery and
operative results. Acta Neurochir. Suppl. (Wien), 53:
3741.
G. NEULOH ET AL.
Gentili, F, Lougheed, WM, Yamashiro, K and Corrado, C
(1985) Monitoring of sensory evoked potentials during
surgery of skull base tumours. Can. J. Neurol. Sci., 12
(4): 336340.
Glasker, S, Pechstein, U, Vougioukas, VI and Van Velthoven, V (2006) Monitoring motor function during resection of tumours in the lower brainstem and fourth
ventricle. Childs Nerv. Syst., 22(10): 12881295.
cortical electrical stimulation in the human: histopathologic confirmation. Electroencephalogr. Clin. Neurophysiol., 75(5): 371377.
Grabb, PA, Albright, AL, Sclabassi, RJ and Pollack, IF
monitoring of cranial nerves during resection of fourth
ventricular tumors in children. J. Neurosurg., 86(1):
14.
Houtteville, JP (1995) The surgery of cavernomas both
supra-tentorial and infra-tentorial. Adv. Tech. Stand.
Neurosurg., 22: 185259.
Iadecola, C, Nakai, M, Arbit, E and Reis, DJ (1983) Global
cerebral vasodilatation elicited by focal electrical stimulation within the dorsal medullary reticular formation in
anesthetized rat. J. Cereb. Blood Flow Metab., 3(3):
270279.
Kohno, K, Matsui, S, Nishizaki, A, Takeda, S, Sadamoto,
K and Sakaki, S (1993) Successful total removal of
intramedullary hemangioblastoma from the medulla
oblongata. Surg. Neurol., 39(1): 2530.
Kombos, T, Suess, O, Pietila, T and Brock, M (2000) Subdural air limits the elicitation of compound muscle
action potentials by high-frequency transcranial electrical stimulation. Br. J. Neurosurg., 14(3): 240243.
Kothbauer, K, Deletis, V and Epstein, FJ (1997) Intraoperative spinal cord monitoring for intramedullary surgery: an essential adjunct. Pediatr. Neurosurg., 26(5):
247254.
Kyoshima, K, Kobayashi, S, Gibo, H and Kuroyanagi, T
(1993) A study of safe entry zones via the floor of the
fourth ventricle for brainstem lesions. Report of three
cases. J. Neurosurg., 78(6): 987993.
Lang, J, Jr., Ohmachi, N and Lang, J, Sr. (1991)
Anatomical landmarks of the rhomboid fossa (floor of
the 4th ventricle), its length and its width. Acta Neurochir. (Wien), 113(12): 8490.
modalities. Can. J. Anaesth., 41(2): 9297.
Mller, AR and Jannetta, PJ (1984) Preservation of facial
function during removal of acoustic neuromas. Use of
monopolar constant-voltage stimulation and EMG.
J. Neurosurg., 61(4): 757760.
148(5): 499509.
Morota, N, Deletis, V, Epstein, FJ, Kofler, M, Abbott, R,
Lee, M and Ruskin, K (1995) Brainstem mapping: neurophysiological localization of motor nuclei on the floor
of the fourth ventricle. Neurosurgery, 37(5): 922929;
discussion 929930.
Morota, N, Deletis, V, Lee, M and Epstein, FJ (1996) Functional anatomic relationship between brainstem tumors
and cranial motor nuclei. Neurosurgery, 39(4):
787793; discussion 793794.
Neuloh, G and Schramm, J (2002) Intraoperative neurophysiological mapping and monitoring for supratentorial procedures. In: V Deletis and JL Shils (Eds.), Neurophysiology
in Neurosurgery. Academic Press/Elsevier Science,
Neuloh, G and Schramm, J (2004a) Monitoring of motor
evoked potentials compared with somatosensory evoked
potentials and microvascular Doppler ultrasonography in
cerebral aneurysm surgery. J. Neurosurg., 100(3): 389399.
Neuloh, G and Schramm, J (2004b) Motor evoked potential
monitoring for the surgery of brain tumours and vascular malformations. Adv. Tech. Stand. Neurosurg., 29(5):
171228.
(2004) Motor evoked potential monitoring with supratentorial surgery. Neurosurgery, 54(5): 10611070; discussion 10701072.
Neurosurgery, 39(2): 335343; discussion 343344.
Pechstein, U, Zentner, J, Van Roost, D and Schramm, J
(1997) Surgical management of brainstem cavernomas.
Neurosurg. Rev., 20(2): 8793.
Pollack, IF, Hoffman, HJ, Humphreys, RP and Becker, L
(1993) The long-term outcome after surgical treatment
of dorsally exophytic brainstem gliomas. J. Neurosurg.,
78(6): 859863.
Quinones-Hinojosa, A, Lyon, R, Du, R and Lawton, MT
(2005) Intraoperative motor mapping of the cerebral peduncle during resection of a midbrain cavernous malformation: technical case report. Neurosurgery, 56(Suppl. 2):
E439; discussion E439.
Romstock, J, Strauss, C and Fahlbusch, R (2000) Continuous
electromyography monitoring of motor cranial nerves
during cerebellopontine angle surgery. J. Neurosurg.,
93(4): 586593.
Ruskin, KJ, Deletis, V and Morota, N (1994) Anesthesia
and intraoperative monitoring. J. Neurosurg., 80(5):
946947.
533
Sala, F, Lanteri, P and Bricolo, A (2004) Motor evoked
potential monitoring for spinal cord and brainstem surgery. Adv. Tech. Stand. Neurosurg., 29: 133169.
Sala, F, Palandri, G, Basso, E, Lanteri, P, Deletis, V, Faccioli, F and Bricolo, A (2006) Motor evoked potential
monitoring improves outcome after surgery for intramedullary spinal cord tumors: a historical control study.
Neurosurgery, 58(6): 11291143; discussion 1129
1143.
Samii, M, Eghbal, R, Carvalho, GA and Matthies, C (2001)
Surgical management of brainstem cavernomas. J. Neurosurg., 95(5): 825832.
Sandalcioglu, IE, Wiedemayer, H, Secer, S, Asgari, S and
Stolke, D (2002) Surgical removal of brainstem cavernous
malformations: surgical indications, technical considerations, and results. J. Neurol. Neurosurg. Psychiatry, 72
(3): 351355.
Soustiel, JF, Hafner, H, Chistyakov, AV, Guilburd, JN,
Zaaroor, M, Yussim, E and Feinsod, M (1993) Monitoring of brainstem trigeminal evoked potentials. Clinical
applications in posterior fossa surgery. Electroencephalogr. Clin. Neurophysiol., 88(4): 255260.
Strauss, C and Fahlbusch, R (1997) Anatomical aspects for
surgery within the floor of the IVth ventricle. Zentralbl.
Neurochir., 58(1): 712.
Strauss, C, Romstock, J, Nimsky, C and Fahlbusch, R
(1993) Intraoperative identification of motor areas of
the rhomboid fossa using direct stimulation.
J. Neurosurg., 79(3): 393399.
Strauss, C, Lutjen-Drecoll, E and Fahlbusch, R (1997)
Pericollicular surgical approaches to the rhomboid
fossa. Part I. Anatomical basis. J. Neurosurg., 87(6):
893899.
Strauss, C, Romstock, J and Fahlbusch, R (1999) Pericollicular approaches to the rhomboid fossa. Part II. Neurophysiological basis. J. Neurosurg., 91(5): 768775.
Strauss, C, Naraghi, R, Bischoff, B, Huk, WJ and Romstock, J (2000) Contralateral hearing loss as an effect
of venous congestion at the ipsilateral inferior colliculus
after microvascular decompression: report of a case.
J. Neurol. Neurosurg. Psychiatry, 69(5): 679682.
Taniguchi, M, Cedzich, C and Schramm, J (1993) Modification of cortical stimulation for motor evoked potentials under general anesthesia: technical description.
Van Veelen-Vincent, ML, Pierre-Kahn, A, Kalifa, C,
Sainte-Rose, C, Zerah, M, Thorne, J and Renier, D
(2002) Ependymoma in childhood: prognostic factors,
extent of surgery, and adjuvant therapy. J. Neurosurg.,
97(4): 827835.
Wagner, W, Peghini-Halbig, L, Maurer, JC and Perneczky, A
(1994) Intraoperative SEP monitoring in neurosurgery
around the brainstem and cervical spinal cord: differential
recording of subcortical components. J. Neurosurg., 81(2):
213220.

M.R. Nuwer (Ed.)
534
CHAPTER 38
Cranial base surgery

Andrew F. Cannestraa, Rinaldo F. Canalisb, Eduardo H. Rubinsteinc,* and
Donald P. Beckera
a
Department of Surgery/Neurosurgery, University of California, Los Angeles, CA 90095, USA
Department of Surgery/Head and Neck Surgery, University of California, Los Angeles, CA 90095, USA
c
Department of Anesthesiology, University of California, Los Angeles, CA 90095, USA
38.1. Introduction
38.1.1. Relationship between neurophysiological
monitoring and clinical outcome during skull
base surgery
Systematic neuromonitoring in skull base surgery has
been used for the past 20 years with most reports supporting the utilization of the techniques that we are
going to discuss. This experience justifies the development of a team of neurotologists, neurosurgeons,
neuroanesthesiologists, and clinical neurophysiologists to assure the utilization of the most current
and appropriate neuroprotective modalities while
sharing the status of the monitored parameters.
Preoperatively, one must consider the diagnostic
possibilities, choose the best surgical approach, and
the possibility of combined surgical interventions.
Resection of these lesions is technically demanding
and paramount to a successful outcome is the protection of vascular and neural structures at risk during
surgical manipulation.
The preservation of neurovascular elements requires
extensive in-depth knowledge of both the normal and
pathologic skull base anatomy and anticipation of the
distortion of normal structures by the lesion. The gold
standard regarding the protection of neural structures
is their visual recognition during the procedure, but
many times the pathology involves them, making their
identification difficult. In these conditions, specific
monitoring will effectively decrease the incidence of
postoperative complications.
*
Correspondence to: Eduardo H. Rubinstein, M.D., Ph.D.,

Department of Anesthesiology, University of California,
Los Angeles, CA 90095-1778, USA.
Tel.: +1-310-825-6761.
E-mail: erubinstein@mednet.ucla.edu (E.H. Rubinstein).
We will first discuss the rationale, from the surgical point of view, for the use of the monitoring methods utilized during skull base surgery and then, we
will present the specific aspects of neural and vascular monitoring that are very helpful during the surgical approach to the posterior, middle, anterior fossae,
and craniocervical junction.
38.2. General aspects of neurological monitoring
Since many pathological conditions are approached
via the base of the skull, several cranial nerves and
brainstem structures are commonly placed at risk.
The monitoring techniques utilized for cerebral and
nerve protection include
electromyography (EMG), both the spontaneous
and the evoked activity;

evoked potentials (EPs), including the visual, audi-
tory, somatosensory, and motor modalities;

electroencephalography (EEG), recorded using one
or very few channels with scalp electrodes, mostly

for the estimation of anesthetic depth or cerebral
ischemia; and
ultrasonic Doppler, used within the 20 MHz range
to identify the course of blood vessels near the
exploring probe.
38.2.1. Electromyography
The activation of brainstem or spinal cord motoneurons generates action potentials in the specific motor
nerve that innervates its particular muscle group.
Subsequent neuromuscular transmitter release initiates the muscle contraction that is preceded by muscle action potentials that generate electrical vectors
that are recorded from the muscle with small needle
or surface electrodes.
Accessible muscle groups innervated by a specific

cranial nerve are selected for recording. Bipolar
needles are preferred during surgery because they
provide a stable and reliable contact with the muscle.
A large ground electrode is placed in close proximity
to the monitored sites to minimize electromagnetic
noise because of better common mode rejection.
The position of this reference will affect the size of
the stimulus artifact on the evoked EMG recording.
This usually does not constitute a problem if a short
stimulus duration, for example, 100200 ms, is used.
38.2.1.1. Spontaneous EMG activity
The electrical activity is maximum in the awake subject performing a voluntary contraction of a desired
muscle group; however, even in the anesthetized,
unparalyzed patient, there is some muscle tone
(spontaneous EMG activity) that varies in voltage
according to the anesthetic depth. Thus, the baseline
tone will be high in the lightly anesthetized subject
and very low or absent, during deep anesthesia (and
obviously when muscle relaxants are used).
This spontaneous activity is of significant clinical
value in many conditions. For example, the facial
nerve is sensitized by the mechanical (or chemical)
effects of the surrounding pathology, and it generates
action potentials during surgical manipulation
near the axons. The affected nerve fibers behave as
mechanoreceptors because the axonal sodium
channels may be modified if the perineurum has been
damaged (by the pathology) and they will open
more readily to a stretch that normally should have
no effect. This is expressed by dissection-related
EMG activity, the so-called neurotonic discharges
(Harner et al., 1986) Their presence indicates that
the manipulation is very close to the nerve and that
the former is already negatively modified by the
pathology and more prone to immediate postoperative dysfunction. However, if the connectivity with
the motoneurons and the muscle is not interrupted,
a significant recovery should be expected.
38.2.1.2. Evoked EMG activity
In contrast to the spontaneous activity, the evoked
EMG is initiated with low current electrical stimulation in the area where the course of the nerve is
expected. Typically, the stimulator is a battery operated or optically isolated unit and it delivers currents
that should not exceed 4 mA. A long, single sterile
electrode operated by the surgeon explores the surgical field, while a large electrode outside the sterile
535
area, constitutes the reference. Stimuli with a duration of 100200 ms are delivered at a frequency of
1 per second with the unipolar electrode as the negative lead. This configuration may be better than the
bipolar because even if the stimulus artifact is larger,
it makes for a much easier exploration and better
electrical contact (Dankle and Wiegand, 1994;
Mermelstein et al., 1996).
Each stimulus triggers the initiation of the sweep
on the screen of the recording apparatus and the
immediate display of the stimulus artifact and after
a certain latency a biphasic (or more complex) compound action potential is recorded. This latency
varies with the nerve conduction velocity and the distance between the stimulated site and the recording
electrodes in the muscle.
When the evoked response is detected, it is important to decrease the current intensity while exploring
with the electrode in an effort to obtain the highest
signal voltage with the minimal possible current,
because this will certify that the searching tip is
closest to the nerve.
False-positive results, that is, the recording of an
evoked EMG when the desired nerve is not being stimulated, are infrequent. There are two potential
sources of error: (a) the nerve is stimulated effectively, but its connection with the brainstem has been
severed. This possibility could be checked by direct
application of the electrode to the brainstem area in
which the nerve originates. (b) The specific nerve is
not stimulated but an adjacent one, which will trigger
evoked activity in its muscle group that, in turn, will
be picked up by the electrodes on the muscle group
of interest. This occurs because of the volume conduction of the generated vector, from the wrong
muscle to the correct electrodes. Typically, this
may take place during posterior fossa surgery while
exploring for the facial and stimulating instead the
trigeminal nerve with activation of the masseter muscle. The potentials evoked from the latter will be
picked up by the electrodes on the orbicularis muscles leading to the incorrect conclusion that the facial
nerve had been stimulated. This event can be differentiated by the shorter latency of the EMG response
triggered from the trigeminal (4 ms) as opposed to
6 ms from the facial, as it is observed in normal size
adult subjects.
Another complex situation is that of differentiating among oculomotor nerves within the cavernous
sinus. It is possible to orient various electrodes
around the eye globe placing fine wires within each
536
oculomotor muscle (Sekiya et al., 2000; Schlake

et al., 2001a) and triggering a selective muscle potential when the specific nerve is stimulated. However,
this may be simplified by placing just two periorbital
surface electrodes that may be sufficient to identify
a certain structure as an oculomotor nerve, once an
evoked EMG is observed. Moreover, the configuration of this compound action potential will vary
depending on which oculomotor muscle is responding, because the direction of the EMG electrical
vector from each one, is different with respect to
the pickup electrodes.
The analysis of the evoked EMG is straightforward, either there is a signal or not. However, the
lack of a muscular response may be due not only
because the stimulating probe is far away from the
nerve but also because of the following reasons:
(a) low stimulating current, which is tested by
increasing the stimulus up to the maximal acceptable
current; (b) technical problem leading to failure to
stimulate and this associates with a lack of stimulus
artifact or the inability to activate an adjacent muscle
directly touched with the probe; (c) the probe is stimulating an identifiable segment of the nerve but the
more peripheral segment of it has been severed
during the surgical approach; (d) the patient has not
recovered from the previous administration of a
long-acting muscle relaxant (Blair et al., 1994).
Regarding this last consideration, it has been argued
that partial muscle relaxation characterized by a
one-twitch response, using a blockade monitor,
should be adequate for nerve testing while preventing
a possible excessive and dangerous movement from a
lightly anesthetized patient (Adams et al., 1994). Our
opinion is that sometimes the identification of the
nerve with evoked EMG is not very easy and having
the patient partially paralyzed may add to those difficulties especially if the muscle paralysis is more
complete than expected.
38.2.1.3. Visual versus aural monitoring of the EMG
The spontaneous or evoked signals are displayed
on the screen of the recording-stimulating system and
the anesthesiologist or neurophysiologist can indicate
orally (yes, no) for the presence of an EMG pattern.
However, the auditory signal corresponding to the
visual display should also be available to the surgeon.
This auditory feedback is especially useful when the
stimulating probe is used to dissect a tumor that is
adjacent to the nerve and also to control the dissection when neurotonic discharges are triggered.
A.F. CANNESTRA ET AL.
Moreover, the sound of an evoked EMG is also characteristic and its complex lower frequency pattern
(trac) is easily differentiated from the brief, high
pitch clic of the stimulus artifact alone.
The previous techniques are very adequate to
identify and protect most of the cranial nerves that
have a motor output to specific muscle groups. The
others: olfactory, optic, and cochleovestibular are
monitored using EPs triggered by their physiological
sensory stimuli.
38.2.2. Evoked potentials
38.2.2.1. Olfactory evoked potentials
Chemosensory EPs have been described in awake
volunteers exposed to hydrogen sulfide nasal stimulation; however, the practical intraoperative use of
this modality has not been established (Sato et al.,
1996). Electrical stimulation of the nasal mucosa will
activate both chemoreceptor and trigeminal afferents,
thus complicating the selectivity of such test. Unless
the pathology in the area of the nasal mucosa or
cribiform plate is very small and unilateral, it is
very difficult to prevent the loss of olfaction during
surgery in the olfactory groove.
38.2.2.2. Visual evoked potentials
This evoked activity is very sensitive to the level of
anesthesia and this should be as light as possible,
using preferably intravenous agents (Domino et al.,
1963). The recording is very variable because the
evoked waves have a very long latency 60
120 ms and represent events originating only in
the occipital cortex. Moreover, the potentials are
generated by the flash activation of the retina that
produces a slow action potential that is conducted
equally slowly towards the primary visual cortex.
These characteristics make the synthesis of the
evoked activity less well defined and prone to marked
variability in the anesthetized subject. It is possible
that direct optic nerve stimulation from the operative
site may be able to generate a sharper waveform;
however, there are no reports on the use of such a
technique, except in the experimental setting (Jones,
1997; Ng and North, 2001). The use of muscle relaxants, when appropriate, will improve the quality of
the recording. The retina is activated physiologically
using goggles that support a group of light emitting
diodes (LEDs). This arrangement generates a red
flash that is effective through the closed eyelids. This
setup offers adequate corneal protection with the
disadvantage of the bulky arrangement in front of

the eyes. More direct visual activation using LEDs
mounted on corneal lenses offers a better profile for
the setup at the risk of potential corneal damage.
38.2.2.3. Brainstem evoked potentials
This is also a well-recognized technique and the
recorded waves correspond to the activation of
clearly defined anatomical structures that belong to
the initial neuronal nuclei of the auditory pathway
(Stockard et al., 1980). The most important objective
with this monitoring modality is to follow the function of the cochlea and cochlear nerve to facilitate
hearing preservation (Harper et al., 1992). The intraoperative recording is relatively straightforward,
because this evoked activity is not modified by depth
of anesthesia. However, mobilization of the ear or ear
canal to improve the surgical approach may decrease
the tightness of the click generating ear insert with
some loss of the sound pressure delivered with each
click. This could affect the recording and decrease
its reliability.
Cochlear potentials recorded intracranially in the
region of the cochlear nerve, can also effectively
monitor the integrity of the desired structures and
they have the advantage that are faster to obtain
and do not require the prolonged averaging times
characteristic of the brainstem evoked potential
(BSEP). The quality and ease of recording is a function of the design of the stimulating and the recording electrodes (Mller et al., 1994).
BSEPs are used also to monitor the well being of
the brainstem, that may be affected by mechanical
displacement during tumor dissection or placement
of a retractor, and for this use, attention is placed
on changes in the wave III to V latency and the voltage of those waves, provided that waves I and II
amplitudes and latencies have not significantly
changed.
38.2.2.4. Somatosensory evoked potentials
The most important role of this modality is during
procedures in which brainstem displacement or compression may be expected, for example, posterior
fossa surgery resulting in transient ischemia. There
are other indicators of these events, for example, bradycardia or a change in later waves of the BSEP, but
it is convenient in difficult cases, to take advantage
of a multimodality approach, to check the reliability
of the observed changes in any one parameter (Carter
and Butt, 2001).
537
38.2.2.5. Motor evoked potentials

This technique is seldom applied to skull base surgery
because it is mostly helpful in identifying the conduction along the corticospinal tract and the excitability of
the spinal motor neurons. However, it could be applied
during posterior fossa surgery for large tumors. In those
cases, surgical dissection may lead to the unplanned disconnection of a neural structure, typically the facial
nerve, from the brainstem neurons. MEP could be used
to confirm the continuity of the motor neurons with their
specific nerve (Akagami et al., 2005).
38.2.3. Electroencephalography
Recording the EEG may be a byproduct of monitoring EPs, because the electrodes are already in place,
and this is a simple and safe procedure that is very convenient to continuously assess the anesthetic depth
(Hosick et al., 1971). This is important to assure that
the patient is not too lightly anesthetized, because this
may lead to high spontaneous EMG activity that may
be erroneously labeled as neurotonic discharge.
Moreover, if pharmacological cerebral protection is
required, the setup is already in place to achieve a
burst-suppression pattern on the EEG using propofol
or thiopental and if the electrodes are placed at the
appropriate scalp location, it will be possible to identify those periods of cerebral ischemia that may result
from excessive retraction, especially during posterior
fossa surgery. The transition into an EEG pattern of
ischemia will be associated with other modifications,
triggered by mechanically induced inadequate brainstem blood flow, such as cardiovascular changes, for
example, bradycardia, and alterations of the BSEP
and somatosensory evoked potential (SEP).
38.2.4. Cardiovascular monitoring
This non-neural topic is briefly discussed here
because cardiovascular changes triggered during skull
base surgery are often indicators of potentially harmful effects on the neural elements. The vagal and sympathetic controlling areas in the medulla are frequently
stimulated by mechanical displacement or a change in
the composition of the cerebrospinal fluid (CSF). As a
consequence, classical cardiovascular signs such as
bradycardia or hypertension are triggered and recognized by the anesthesiologist. These changes should
be treated by first halting surgical manipulation and
minimizing pharmacological intervention. Typically,
an anticholinergic agent, for example, atropine,
538
should never be used to treat bradycardia because it

will block the display of this sign if the triggering
maneuver is repeated during the procedure.
Prolonged irrigation of the IVth ventricle with a
solution with low pH, such as normal saline (pH 5.0)
can trigger chemoreceptive responses from the adjacent brainstem region manifested by increased spontaneous respiration in an unparalyzed well-ventilated
patient with a Paco2 below 35 mm Hg, and sometimes
tachycardia and hypertension. The recommendation
is to use a solution with a balanced pH (7.4) such as
Plasmalyte. Changes in ventilation are expected
when the surgery is close to the lower brainstem,
together with cardiovascular changes. In the past,
patients were allowed to breathe spontaneously and a
modification in the ongoing pattern of respiration
was considered an indication of surgical stimulation
of the brainstem; however, presently few anesthesiologists recommend this technique (Manninem, 1995).
Most anesthetized patients are mechanically ventilated
to a Paco2 of 2832 mm Hg and in these conditions
spontaneous ventilation is unlikely.
Neural and vascular structures are tightly arranged
in the base of the skull and visual recognition of those
elements is the desired criteria for their preservation.
When the vascular elements are covered or distorted
by the pathology, it is convenient to anticipate the
closeness of the vessels using simple Doppler ultrasonography. A long and slender probe with a small crystal activated at 20 MHz can sense the velocity of the
red cells in arteries and veins at a depth of not more
than 5 mm. The use of probes with lower frequencies
is not indicated because these may illuminate vessels that are much deeper than the desired ones, and
it will confuse their identification. The sound of the
recorded Doppler shift will clearly define the high
pitch pulsatile characteristic of arterial flow or the
low pitch continuous rumble of venous flow. This
is helpful to identify the nature and location of a vascular structure but it will also reflect changes in the blood
flow velocity. Typically, a change from the ongoing
sound pitch characteristic of arterial pulsatile flow into
a higher frequency pattern may result from compression or vasospasm of the artery being tested.
38.2.4.1.1. The sitting position. The use of this

option for posterior fossa surgery has decreased,
because of two major problems: (a) cerebral hypoperfusion, secondary to the decreased systemic arterial
pressure resulting from the gravitational blood volume
redistribution and the added negative pressure gradient
between the lower heart and the higher head, and (b) air
embolism, because the venous pressure gradient
between the operative site and the heart is also negative, and consequently, instead of blood draining out
from an open venous structure, air will be drawn in.
Venous air embolism is characterized by pulmonary
hypertension, increased alveolar (endtidal)arterial
Pco2 gradient, relative hypoxemia, bronchoconstriction, hypotension, and bradycardia. Moreover, the flow
of air bubbles from the right to the left atrium via an
interatrial septal defect has been reported as a complication which may lead to an embolic stroke (Black
et al., 1990). To diagnose and treat during this situation, it is customary to use a precordial 4 MHz Doppler
and a central line with the tip placed at the junction of
the superior vena cava with the right atrium. The
described risks may be offset by the improvement of
blood and CSF drainage and a more clear anatomical
view that may be obtained in the sitting position
(Leonard and Cunningham, 2002). Another infrequent
complication in the sitting patient is pneumoencephalus, caused by the migration of outside air toward the
cortical convexity along the subdural space. This is
detected intraoperatively because the voltages of the
SEP, BSEP, and EEG rapidly decrease without a
change in their latency or pattern. This results from
the interposition of air between the EEG generating
cortex and the recording electrodes on the scalp
(Schubert et al., 1986; Watanabe et al., 1989a).
38.2.4.1. Positioning and cardiovascular monitoring

Cardiovascular stability is markedly influenced by
the position of the patient during surgery. This is
very evident when the lower extremities are placed
below the level of the heart causing a decrease in
venous return and cardiac output.
38.2.4.2. Communication effectiveness and the team

approach during monitoring
We have mentioned that it is important to develop a
team of specialists to facilitate the safest approach
to these complex surgical procedures. The quality
and reliability of the neuromonitoring techniques
38.2.4.1.2. The lateral position. The risks previously discussed are minimized (but not completely
avoided) when using this position. However, excessive neck torsion should be avoided to prevent interference with venous drainage and stretching of the
brachial plexus. The shoulder over the edge lateral
alternative allows for a good exposure without these
risks (Huncke et al., 1998).
should be assured from the start and also the methodology used by the team to communicate among its
members. For instance, the output of some recording
modalities such as the EMG should be available as a
continuous sound signal together with the vocal confirmation by the neurophysiologist. In this manner,
the surgeon will be able to guide the exploring probe
or the dissection toward the safest direction.
38.3. Relationship between surgical approach and
monitoring requirements
The following section identifies the various structures that are at risk depending on the anatomical
site, the selected surgical route, and the desire to
obtain the largest possible surgical exposure.
38.3.1. The posterior fossa
38.3.1.1. Retrosigmoid, presigmoid, and
translabyrinthine approaches
A lesion within the cerebellopontine (CP) angle requires
cranial nerve, fiber tract, and vascular mapping. Specifically, cranial nerves V, VII, and VIII must be monitored and with larger lesions that require significant
retraction on the cerebellum and brainstem, recording
the BSEP and SEP is very important. During these
approaches, the first structures encountered are the
transverse and sigmoid sinuses. After dural opening
and cerebellar retraction, the lower cranial nerves are
often seen first followed by the VII and VIII nerve
complex often associated with a loop of anterior
inferior cerebellar artery (AICA). The trigeminal and
the trochlear nerves are located superiorly under the
tentorium.
When pathology is present, the normal neurovascular structures are displaced by the lesion. Regardless of the specific surgical strategy to its access,
the facial nerve may be at risk from the start of the
dissection. Therefore, immediate electrical testing of
the surface of the lesion is performed. During this
initial testing, a high current (up to 24 mA) may
be used. If an evoked EMG is obtained, it will give
an indication of the general orientation of the VII
nerve. However, the nerve may not be a discrete fiber
bundle due to the splaying of its fibers caused by
their progressive displacement by the growing
mass. In these cases, there is an area of positive
responses as opposed to a more discrete location
from where low threshold EMG potentials may be
triggered. Bipolar stimulation may be more precise
539
and will give a smaller stimulus artifact on the

recording, while with monopolar stimulation the
contact with the nerve will be easier to achieve.
When the VIIth nerve has been stretched by the
lesion, the incidence of neurotonic discharges is
very high. Continuous auditory feedback of the spontaneous EMG activity is very important in these situations. Prolonged discharges after minimal dissection
or a dive bomber firing pattern are associated with
a poor outcome. It is important to frequently test for
the integrity of the connection of the nerve with its
nucleus, using direct electrical stimulation of the
brainstem, or motor evoked potentials (MEPs) or stimulation of the nerve with recording of both the orthodromic M and the antidromic F wave of the evoked
EMG. The latter will disappear if there is a loss of
the central connection with the nerve. This test is also
practical during nerve decompression for hemifacial
spasm. In this case, the F wave is triggered using a
low intensity stimulus as opposed to the high voltage
required for all other cases. The disappearance of the
low threshold F wave is considered a good sign
indicating that the nerve decompression was successful (Andrews and Rubinstein, 1999).
Another very important consideration during posterior fossa surgery is the protection of the cochlea
and the cochlear nerve. BSEP monitoring is the most
common method used and it should be recorded
from both the tumor and the normal sides. A common change during the approach to the pathology is
the prolongation of the wave I latency, which corresponds to the effect of drilling the temporal bone on
the cochlear sensitivity to the stimulus. A transient
hearing defect may result from the mechanical
vibration that may even be apparent on the contralateral ear (Da Cruz et al., 1997). The mechanisms
responsible for the changes secondary to drilling
may be vibrational or vascular and are usually reversible. Also, the BSEP may be transiently changed
during tumor dissection, especially the voltage of
the waves IIIV as the result of retractor ischemia
or it may follow the displacement of the brainstem
causing reorientation of the evoked electrical
vectors, and consequently it will rapidly reverse with
repositioning (Grundy et al., 1981).
38.3.1.1.1. The retrosigmoid approach (Fig. 1). This
technique is selected to reduce the risk of hearing
loss while preserving facial nerve function because
it allows clear visualization of the brainstem and cranial nerve/brainstem junction. The internal acoustic
540
Dura
Transverse sinus
Cerebellum
limited to unilateral off midline lesions, or lesions that

at least do not cross the midline. The typical lesion
for this approach is a moderately large acoustic neuroma, with preserved facial and auditory functions.
Tentorium
CN III
Petrosal vein
CN V
CN VII
CN VIII
Anterior inferior
cerebellar artery
Flocculus
Choroid plexus
CN IX, X, XI
Posterior inferior
cerebellar artery
Fig. 1. Suboccipital retrosigmoid approach to the posterior

fossa. Wide exposure of the cerebellopontine angle and neurovascular structures with less risk of hearing loss (reprinted
from Samii et al., 1995 with permission from Elsevier).
meatus is drilled to permit complete resection of the

tumor and direct recognition of the nerves from the
fundus to the brainstem.
The trajectory of the retrosigmoid approach will
allow a clear view of lesions visible within a line drawn
from the sigmoid sinus to the internal auditory canal
(IAC) and beyond it, to the clivus. Thus, it should be
38.3.1.1.2. The presigmoid approach (Fig. 2). The

previous considerations are also applicable to this
approach. Its trajectory allows a clear view of lesions
visible within a line drawn from the sigmoid sinus to
the IAC and beyond to the clivus. Unlike the retrosigmoid approach, it can be used for tumors extending past the midline. The typical lesion for this
technique is a petroclival meningioma.
With this approach, there is an increased risk to
hearing and facial function because it requires drilling over the facial canal and labyrinth. Also, the
extended access will increase the risk to the cranial
nerves traveling near the midline (i.e., IV and VI).
During the dissection, several vascular structures
are at an increased risk: the sigmoid and petrosal
sinuses, the vein of Labbe, and the basilar artery.
Particular attention should be paid to the vein of
Labbe because damage to this structure will result
in disruption to the drainage of the dominant temporal lobe and severe cortical complications.
38.3.1.1.3. The translabyrinthine approach. The
translabyrinthine approach allows access to the CP
Dura
CN III
CN IV
Superior
petrosal sinus
Superior
cerebellar artery
Petrosal vein
Vein of Labb
CN VI
CN VII
Tentorium
CN VIII
Anterior inferior
cerebellar artery
Cerebellum
CN IX
CN X
CN XI
CN V
Flocculus
Sigmoid sinus
Fig. 2. Suboccipital presigmoid approach to the posterior fossa. Exposure of the cerebellopontine angle with less cerebellar
retraction and visualization past the midline. High risk of hearing loss (reprinted from Samii et al., 1995 with permission
from Elsevier).
541
angle. As with the retrosigmoid and presigmoid

approaches, the translabyrinthine is chosen to preserve facial nerve function. However, because it
requires drilling through the inner ear, it is selected
only when there is no serviceable hearing on the
operated site. The approach allows early visualization of the seventh nerve at the tympanic segment
and from this site, its dissection to the brainstem
junction.
38.3.2. The middle fossa
38.3.2.1. Subtemporal/transtentorial and
transtentorial/interhemispheric approaches
The concave shape of this region and the presence
of the tentorium restrict the visualization of medially located lesions, therefore the surgical strategies
used in this area are designed to offset this limitation. In some cases, for example, an intracannalicular acoustic neuroma, the exposure of the lateral
tentorium is sufficient, requiring modest temporal
lobe displacement; however, with lesions extending
to the midline, the techniques that are used not
only improve exposure but reduce temporal lobe
retraction.
38.3.2.1.1. The subtemporal approach/transtentorial extension (Fig. 3). The subtemporal approach is
often chosen for the excision of intracanalicular
acoustic neuromas. In this situation, the distortion
Geniculate ganglion
Lateral semicircular canal
Greater superficial
petrosal nerve
Cochlea
Cochlear
nerve
Facial nerve
Superior
vestibular
nerve
Inferior
vestibular
nerve
Posterior
semicircular
canal
Superior
semicircular
canal
Fig. 3. Subtemporal approach to the middle fossa. Exposure of small lesions within the range of the porus acusticus
(reprinted from Samii et al., 1995 with permission from
Elsevier).
of the anatomical landmarks is usually limited.

Nevertheless, identification of the cochlear and facial
nerves within the auditory canal during drilling and
tumor removal is critical for their functional
preservation.
The subtemporal approach is often combined with
posterior fossa approaches to provide middle and
posterior fossa access. The standard craniotomy is
extended inferiorly until there is a tangential view
along the base of the middle fossa. This will
minimize the amount of superior retraction of the
temporal lobe.
During the middle fossa approach, it is critical to
identify and protect the vein of Labbe which drains
the temporal lobe. Damage to this structure may
result in temporal lobe infarction with catastrophic
results if the dominant hemisphere is involved.
This approach is also used for lesions extending
from the mid-basilar region to the posterior clinoid.
Its advantage is that with proper brain relaxation,
the surgeon could have access from the mid-clival
region to the posterior clinoid and even the pituitary
stalk. This approach may be extended by dividing
the tentorium to allow access to the inferior clivus
in situations such as basilar artery aneurysm or posterior clival meningiomas. This maneuver increases the
risk of injury to the IV, V, and VI cranial nerves and
their selective monitoring is essential.
38.3.2.1.2. The transtentorial/interhemispheric
approach. The division of the tentorium may also
be combined with an interhemispheric approach
which is chosen when the lesion is directly under this
structure, for example, lesions of the quadrigeminal
plate. The surgical manipulations of the tentorium
may be followed by a transient but very marked
hypertension. The interhemispheric approach can
place the more superficial cortex at risk and thus
EEG, EPs (motor or visual depending on the
approach) should be monitored. Once the subtemporal tentorium and the inferior petrosal sinus is
ligated (which can cause significant bleeding), the
trigeminal nerve is visualized traveling to Meckels
cave. In the interhemispheric/transtentorial approach,
the venous anatomy often dictates the extent of the
resection that increases the risk to the superior sagittal sinus, straight sinus, basal veins of Rosenthal,
cortical venous anatomy, and brainstem draining
veins. Therefore, this method requires brainstem
monitoring of BSEP, SEP, and EEG to verify
functional integrity.
542
38.3.3. The anterior fossa

38.3.3.1. Bifrontal, transnasal, transcilliar
approaches and lateral rhinotomy
When a patient presents with a mass in the anterior
fossa, the surgeon must consider the angle of access
required because this region may be approached from
two trajectories: a superior bifrontal pteryonal and
keyhole craniotomies and an inferior transnasal
and lateral rhinotomy. Superior approach craniotomies allow significant access and visualization of
vital structures including the optic nerves/chiasm,
oculomotor nerves (III, IV, and VI), and vessels of
the anterior circulation. Inferior trajectories can
access midline and off midline structures at the base
of the skull without the need for craniotomy. As with
a superior approach, a CSF leak is a significant concern but they provide excellent access to the medial
temporal region, cavernous sinus, planum spenoidale,
sella, and parasellar region. The pituitary gland,
carotid arteries/anterior circulation, and the optic
nerves are also at risk.
38.3.3.1.1. The bifrontal approach. The bifrontal
approach is very commonly used for the resection
of meningiomas of the olfactory bulb and planum
sphenoidale. A frequent complication is anosmia
and loss of taste resulting from either retractor
ischemia of the olfactory bulb or direct damage to
the cribiform plate. Presently, there are no practical
neurophysiological techniques to monitor olfactory
function, therefore, its preservation is based on
visualization and careful dissection of the nerves.
As the dissection moves posteriorly, several structures are at risk but they can be selectively monitored: the optic nerves and chiasm with visual
evoked potential (VEP), the carotid and anterior
communicating arteries with Doppler ultrasonography, the pituitary stalk (only indirectly by the
development of diabetes insipidus), and the hypothalamus by the sudden display of marked cardiovascular responses.
38.3.3.1.2. Keyhole approaches (transcilliar). The
keyhole approach is gaining in popularity due to the
current trend toward minimally invasive procedures.
It is similar to a unilateral subfrontal approach and
has comparable risks. However, since the dissection
is mostly limited to the side of entrance, the incidence of olfactory dysfunction is decreased. On the
other hand, because the approach is extended
laterally, the III and IV nerves are placed at risk

and should be monitored. In common with the bifrontal approach, the carotid and anterior communicating
arteries should be identified with Doppler ultrasound
and visualized.
38.3.3.1.3. The transnasal, transphenoidal (TNTS)
approach. The most common approach is directed
along the midline and if there is an unintended lateral
deviation, the carotid artery and optic nerve will be
put at risk. In patients with larger tumors and the preoperative evidence of a visual field defect, monitoring VEP is very important with all the difficulties
involved in obtaining reliable recordings as discussed
before. When the lesions extend laterally, the oculomotor nerves running within the cavernous sinus
may be damaged and their monitoring, either specific
or unselected, should be available. With lesions
involving the clivus, a wider approach may be
required using unilateral or bilateral medial maxillectomies. During extensive resections of the clivus, the
basilar artery may be exposed.
38.3.3.1.4. Lateral rhinotomy. The lateral rhinotomy is most commonly used for the excision of
esthesioneuroblastomas. These may have a superior
extension through the cribiform plate or less frequently a lateral extension into the cavernous sinus.
In the latter case, cranial nerves III, IV, and VI
should be monitored. A long Doppler probe should
also be available to identify the carotid artery.
38.3.4. The foramen magnum and craniocervical
junction
In this region, the lesions are classified according to
their location and this will dictate the surgical
strategy.
Posterior lesions: They can be accessed by a midline or paramedian suboccipital approach or a high
cervical laminectomy and they may require retraction
of the brainstem and spinal cord and their specific
monitoring is critical: BSEP and SEP.
Anterior lesions: Most of those are approached by
either a transoral or far-lateral technique. A transoral
approach is frequently used for pathology of the
odontoid process. This may require only SEP and
possibly motor EP. Although, the transoral route
requires very little retraction to obtain access to the
lesion, it is less frequently used due to the risk of
CSF leak.
543
Anterior inferior
cerebellar artery
CN VII, VIII
CN IX, X, XI
Accessory nerve
Posterior inferior
cerebellar artery
Vertebral artery
Fig. 4. Lateral approach of the craniocervical junction. Wide exposure of anterior and lateral lesions. Multimodality neurophysiological monitoring is of great importance (reprinted from Samii et al., 1995 with permission from Elsevier).
The far-lateral (Fig. 4) approach may be chosen

for anterior and lateral lesions. This technique is very
complex as it involves wide exposure of lateral
skull base structures. The vertebral artery is initially
encountered and it should be checked with Doppler
ultrasonography and visualization. Thereafter, exposure of the occipital condyle may require identification of the hypoglossal nerve that could include
evoked EMG. During the intradural phase of the
procedure, the lower cranial nerves at the jugular
foramen: IX, X, and XI, are at significant risk and
they should be selectively monitored. It is important
to note that the nerves in this location are extremely
sensitive to mechanical displacement. The most
important nerve to monitor is the vagus which is
achieved by recording the laryngeal evoked EMG
with endotracheal tube electrodes (Mermelstein
et al., 1996). Alternatively, this activity may be
followed with retrolaryngeal or with externally
placed intramuscular electrodes.
Finally, when the brainstem is being decompressed,
perforating branches of the vertebral and basilar
arteries need to be identified because vertebro-basilar
insufficiency may result from compression or inadvertent coagulation, with potentially devastating
consequences.
evaluation of its usefulness may show the advantages

of the capability to identify the borders of a lesion
and thus allow for a complete resection of the
pathological tissue with preservation of the normal
structures.
Intraoperative brainstem electrical techniques can
provide surface stimulation maps or real-time tract
monitoring Topical electrical stimulation mapping
(ESM) can reliably identify the peduncular corticospinal tracts and the IVth ventricular motor nuclei (Morota and Deletis, 2006). However, brainstem ESM is
limited to anatomic structures easily stimulated and
the ability to record subsequent tract output (i.e., EMG
or evoked responses). However, tract monitoring does
not provide useful maps to prevent iatrogenic injury
because any detected change will result from damage
already present in the tested structure.
Optical intrinsic signal (OIS) imaging detects
functional activity of the neural tissue based on the
reflectance of visible light. The signal contains contributions from physiologic processes that undergo
reflectance/absorbance changes (within the visible
spectrum) during normal activation. This technique
has been used extensively for supratentorial cortical
mapping (Cannestra et al., 2004) allowing visualization and protection of those structures prior to structural or functional damage.
38.4. New technology

Our discussion regarding neurovascular protection is
based on the presently available and established
techniques. There are other experimental methods
that may be considered.
Intraoperative scanning, typically MRI, requires a
very sophisticated setup. However, the long-term
38.5. Conclusions
We discussed the relationships between the anatomical aspects of the surgical approach during skull
base surgery and the requirements for neurophysiological and cardiovascular monitoring that should maximize
the protection of neural structures during the procedure.
544
Many of the presently used electrophysiological

techniques were described early on in the experimental setting, for example, brainstem EPs in 1970
(Jewett, 1970; Jewett et al., 1970) but their more
extensive use started around 1982 (Grundy et al.,
1982; Raudzens and Shetter, 1982; Walser et al.,
1982; Zapulla et al., 1984; Friedman et al., 1985;
Piatt et al., 1985; Kalmanchey et al., 1986;
Silverstein et al., 1986; Little et al., 1987; Schramm
et al., 1988; Mller and Mller, 1989; Radtke et al.,
1989; Watanabe et al., 1989b; Salzman, 1990;
Fischer et al., 1992; Kawaguchi et al., 1995; Colleti
and Fiorino, 1998; Schlake et al., 2001b). Occasionally, the techniques utilized were not sufficient to
prevent nerve damage (Piatt et al., 1985) and that
prompted many reevaluations of the criteria used to
define significant intraoperative changes. We have
discussed the difficulties inherent to the surgical
manipulation and dissection when tumor pathology
distorts the anatomical landmarks; however, even in
situations when that is not the case, for example,
decompressive procedures for trigeminal neuralgia
or hemifacial spasm, there are complications that
may be prevented by specific neuromonitoring
(Mller and Mller, 1989).
The positive correlation of neurophysiological
monitoring and the long-term outcomes of skull base
surgery has been reported consistently in the past
10 years (OMalley and Janecka, 1995; McElveen
et al., 2000; DeMonte, 2001; Mann et al., 2002)
and on the basis of those results is presently possible
to counsel the patients in anticipation of surgery
(Harsha and Backous, 2005), presenting the various
surgical strategies and their outcome characteristics
as well as the modalities of neural monitoring to be
utilized for the procedure.
References
Adams, DC, Emerson, RG, Heyer, EJ, McCormick, PC,
Carmel, PW and Stein, BM (1994) Monitoring of
intraoperative motor-evoked potentials under conditions
of controlled neuromuscular blockade. Anesth. Analg.,
77: 913918.
Akagami, R, Dong, CC and Westerberg, BD (2005) Localized transcranial motor evoked potentials for monitoring
cranial nerves during cranial base surgery. Neurosurgery, 57: 7885.
Andrews, JC and Rubinstein, EH (1999) Examination and
diagnostic techniques in facial nerve disorders. In: RF
Canalis and PR Lambert (Eds.), The Ear. A

Comprehensive Textbook of Otology. Lippincott Williams
& Wilkins, Philadelphia, Chapter 44, pp. 705717.
Black, S, Muzzi, DA, Nishimura, RA and Cucchiara, RF
(1990) Preoperative and intraoperative echocardiography to detect right-to-left shunts in patients undergoing
neurosurgical procedures in the sitting position.
Blair, EA, Teeple, E, Jr., Sutherland, RM, Shih, T and
Chen, D (1994) Effect of neuromuscular blockade on
facial nerve monitoring. Am. J. Otol., 15: 161167.
Cannestra, AF, Pouratian, N, Forage, J, Bookheimer, SY,
Martin, NA and Toga, AW (2004) Functional MRI
and optical imaging for dominant hemisphere perisylvian arteriovenous malformations. Neurosurgery,
55(4): 804814.
Carter, BG and Butt, W (2001) Review of the use of
somatosensory evoked potentials in the prediction of
outcome after severe brain injury. Crit. Care Med., 29:
178186.
Colleti, V and Fiorino, FG (1998) Advances in monitoring
of seventh and eight cranial nerve function during posterior fossa surgery. Am. J. Otol., 19: 503512.
Da Cruz, MJ, Fagan, P, Atlas, M and McNeill, C (1997)
Drill-induced hearing loss in the non operated ear. Otolaryngol. Head Neck Surg., 177: 555558.
Dankle, JA and Wiegand, DA (1994) Investigation of a
coaxial bipolar nerve stimulator for intraoperative
motor nerve monitoring. Laryngoscope, 104: 619622.
DeMonte, E (2001) Functional outcomes in skull base
surgery. What is acceptable? Clin. Neurosurg., 48:
340350.
Domino, EF, Corssen, G and Sweet, RB (1963) Effects of various general anesthetics on the visually evoked response in
man. Anesth. Analg., 42: 735747.
Fischer, G, Fischer, C and Remond, J (1992) Hearing preservation in acoustic neuroma surgery. J. Neurosurg., 76:
910917.
Friedman, WA, Kaplan, BJ, Gravenstein, D and Rhoton,
AL, Jr. (1985) Intraoperative brainstem auditory evoked
potentials during posterior fossa microvascular decompression. J. Neurosurg., 62: 552557.
Grundy, BL, Lina, A, Procopio, PT and Jannetta, PJ (1981)
Reversible evoked potential changes with retraction of
the eighth cranial nerve. Anesth. Analg., 60: 835838.
Grundy, BL, Jannetta, PL, Procopio, PT, Lina, A, Boston,
JR and Doyle, E (1982) Intraoperative monitoring of
brainstem auditory evoked potentials. J. Neurosurg.,
57: 674681.
Harner, SG, Daube, JR and Ebersold, MJ (1986) Electrophysiological monitoring of the facial nerve during temporal bone surgery. Laryngoscope, 96: 6569.
Harper, CM, Harner, SG, Slavit, DH, Litchy, WJ, Daube,
JR, Beatty, CW and Ebersold, MJ (1992) Effect of
BAEP monitoring on hearing preservation during
acoustic neuroma resection. Neurology, 42: 15511553.
Harsha, WJ and Backous, DD (2005) Counseling patients
on surgical options for treating acoustic neuroma. Otolaryngol. Clin. North Am., 38: 643652.
Hosick, EC, Clark, DL, Adam, N and Rosner, BS (1971)
Neurophysiological effects of different anesthetics in
unconscious man. J. Appl. Physiol., 31: 884891.
Huncke, K, Van De Wiele, B, Fried, I and Rubinstein, EH
(1998) The asleep-awake-asleep anesthetic technique
for intraoperative language mapping. Neurosurgery,
42: 13121317.
Jewett, DL (1970) Volume-conducted potentials in response
to auditory stimuli as detected by averaging in the cat.
Jewett, DL, Romano, MN and Williston, JS (1970) Human
auditory evoked potentials: possible brainstem components detected on the scalp. Science, 167: 15171518.
Jones, NS (1997) Visual evoked potentials in endoscopic
and anterior skull base surgery: a review. J. Laryngol.
Otol., 111: 513516.
Kalmanchey, R, Avila, A and Symon, L (1986) The use of
brainstem auditory evoked potentials during posterior
fossa surgery as a monitor of brainstem function. Acta
Neurochir., 82: 128136.
Kawaguchi, M, Ohnishi, H, Sakamoto, T, Shimizu, K,
Touho, H, Monobe, T and Karasawa, J (1995) Intraoperative electrophysiological monitoring of cranial motor
nerves in skull base surgery. Surg. Neurol., 43: 177181.
Leonard, IE and Cunningham, AJ (2002) The sitting position in neurosurgery not yet obsolete. Br. J. Anaesth.,
88: 13.
Mann, WJ, Maurer, J and Marangos, N (2002) Neural conservation in skull base surgery. Otorlaryngol. Clin.
North Am., 35: 411424.
Manninem, PH (1995) Spontaneous ventilation is an useful
monitor of brainstem function during posterior fossa
surgery. J. Neurosurg. Anesthesiol., 7: 6365.
McElveen, JT, Jr., Belmonte, RG, Fukushima, T and Bullard, DE (2000) A review of facial nerve outcome in
100 consecutive cases of acoustic tumor surgery. Laryngoscope, 110: 16671772.
Mermelstein, M, Nonweiler, R and Rubinstein, EH (1996)
Intraoperative identification of laryngeal nerves with laryngeal electromyography. Laryngoscope, 106: 752756.
Mller, AR and Mller, AB (1989) Does intraoperative monitoring of auditory evoked potentials reduce incidence of
hearing loss as a complication of microvascular decompression of cranial nerves? Electroencephalogr. Clin.
Mller, AR, Jannetta, PJ and Jho, HD (1994) Click evoked
responses from the cochlear nucleus: a study in humans.
545
148: 499509.
Ng, YT and North, KN (2001) Visual evoked potentials in
the assessment of optic gliomas. Pediatr. Neurol., 24:
4448.
OMalley, BW and Janecka, JP (1995) Evolution of outcomes in cranial base surgery. Semin. Surg. Oncol.,
136: 111116.
Piatt, JH, Jr., Radtke, RA and Erwin, CA (1985) Limitations
of brainstem auditory evoked potentials for intraoperative monitoring during a posterior fossa operation: case
report and technical note. Neurosurgery, 16: 818821.
39: 187191.
Raudzens, PA and Shetter, AG (1982) Intraoperative monitoring of brainstem auditory evoked potentials. J. Neurosurg., 57: 341348.
Salzman, SK (1990) Neural Monitoring. The Prevention of
Intraoperative Injury. Humana Press, Clifton, NJ.
Samii, M, Cheatham, ML and Becker, DP (1995) Atlas of
Cranial Base Surgery. WB Saunders, Philadelphia.
Sato, M, Kodama, N and Sasaki, T (1996) Olfactory
evoked potentials: experimental and clinical studies.
J. Neurosurg., 85: 11221126.
Schlake, HP, Goldbrunner, RH, Milewski, C, Krauss, J,
Trautner, H, Behr, R, Srensen, N, Helms, J and Roosen, K (2001a) Intraoperative electromyographic monitoring of lower cranial motor nerves (IX-XII) in skull
base surgery. Clin. Neurol. Neurosurg., 103: 7282.
Schlake, HP, Goldbrunner, RH, Siebert, M, Behr, R and
Roosen, K (2001b) Intra-operative monitoring of
extra-ocular motor nerves (III, IV, VI) in skull base
surgery. Acta Neurochir. (Wien), 143: 251261.
Schramm, J, Mokrusch, T, Fahlbusch, R and Hochstetter, A
(1988) Detailed analysis of intraoperative changes monitoring brainstem acoustic evoked potentials. Neurosurgery, 22: 694702.
Schubert, A, Zornow, MH, Drummond, JC and Luersen,
TG (1986) Loss of cortical evoked responses due to
intracranial gas during posterior fossa craniectomy in
the seated position. Anesth. Analg., 65: 203206.
Sekiya, T, Hatayama, T, Shimamura, N and Suzuki, S
(2000) Intraoperative electrophysiological monitoring
of oculomotor nuclei and their intramedullary tracts
during midbrain tumor surgery. Neurosurgery, 47:
11701176.
Silverstein, H, Daniel, A, Norrell, H and Haberkamp, T
(1986) Hearing preservation after acoustic neuroma surgery with intraoperative direct eight cranial nerve
546
monitoring. Part II: Classification of results. Otolaryngol.
Head Neck Surg., 95: 285291.
Stockard, JJ, Stockard, JE and Sharbrough, FW (1980)
Brainstem auditory evoked potentials in neurology:
methodology, interpretation, clinical application. In:
MJ Aminoff (Ed.), Electrodiagnosis in Clinical Neurology. Churchill Livingstone, New York, pp. 370413.
Walser, H, Yasargil, MG and Curcic, M (1982) Auditory
brainstem responses in patients with posterior fossa
tumors. Surg. Neurol., 18: 405415.
Watanabe, E, Schramm, J and Schneider, W (1989a) Effect
of a subdural air collection on the sensory evoked

potential during surgery in the sitting position. Electroencephalogr. Clin. Neurophysiol., 74: 194201.
Watanabe, E, Schramm, J, Strauss, C and Fahlbusch, R
(1989b) Neurophysiological monitoring in posterior
fossa surgery. BAEP waves I and V and preservation
of hearing. Acta Neurochir. (Wien), 98: 118128.
Zapulla, RA, Malis, LI, Greenblatt, E and Karmel, BZ
(1984) Utility of brainstem auditory evoked potentials
in the diagnosis and surgical treatment of tumors of
the cerebellopontine angle. In: RH Nodar and C
Barber (Eds.), Evoked Potentials II. Butterworth, Boston,
pp. 194202.

M.R. Nuwer (Ed.)
547
CHAPTER 39
Neurophysiology during microvascular facial

nerve decompression
Aage R. Mller*
School of Behavioral and Brain Sciences, University of Texas at Dallas, Richardson, TX 75080, USA
39.1. Introduction
Microvascular decompression (MVD) of the facial
nerve root eliminates the symptoms of hemifacial spasm
(HFS) in 8595% of patients (Barker et al., 1995), and
recurrences are rare. Intraoperative neurophysiologic
recordings can guide the surgeon in identifying the anatomical location of the offending blood vessel, or vessels. Intraoperative neurophysiologic recordings can
also provide evidence that the goal of the operation has
been accomplishment and it can increase the efficacy
of the operation (Mller and Jannetta, 1987).
Over time, the spasm spreads downwards in the face

affecting more and more muscles and the attacks occur
more frequently. After 1520 years, all facial muscles,
including the platysma, may be involved, but the muscles of the forehead are seldom involved. Many
patients with HFS experience synkinesis, but weakness is rarely observed. HFS is different from other
hyperactive disorders of facial muscles such as facial
myokymia and blepharospasm. Blepharospasm is
characterized by bilateral contractions of facial
muscles, mostly the muscles around the eyes.
39.2. Hemifacial spasm
39.2.2. Electrophysiological signs of HFS
HFS is a rare disorder that has an incidence of 0.74 per

100,000 in Caucasian men and 0.81 per 100,000 in
Caucasian women (Auger and Whisnant, 1990), thus
less than trigeminal neuralgia (TGN) that also can be
cured by MVD of the trigeminal nerve root. TGN is
characterized by short periods of excruciating pain in
specific areas of the face. Glossopharyngeal neuralgia
(GPN) and certain forms of tinnitus and a vestibular
disorder (disabling positional vertigo, DPV) can also
be cured by MVD of the respective cranial nerve. It
was previously assumed that only major arteries could
cause these vascular compression symptoms, but now
there is evidence that veins and arteries of a very small
caliber that are in close contact with a cranial nerve
root can also be the cause of HFS.
The abnormal muscle response and synkinesis of the

blink reflex are signs of HFS. The abnormal muscle
response (also known as the lateral spread response)
can be demonstrated by stimulating one branch of the
facial nerve electrically while recording electromyographic (EMG) potentials from muscles that are innervated by other branches of the facial nerve (Esslen,
1957; Nielsen, 1984a). The amplitude of the abnormal
muscle response is only 510% of that of the direct
muscle response (M-response) indicating that a small
fraction of the total number of motor units are activated; The M-response is assumed to involve most of
the motor units of the muscle when the facial nerve is
stimulated at a supramaximal strength. The abnormal
muscle response is similar to the F-response. The blink
reflex typically shows synkinesis when elicited by
electrical stimulation of the supraorbital nerve on the
spasm side (Nielsen, 1984b).
39.2.1. Clinical signs of HFS

HFS typically begins with involuntary contractions of
the orbicularis oculi muscles on one side of the face.
39.2.3. Treatment of HFS

*
Correspondence to: Aage R. Mller, Ph.D., School of

Behavioral and Brain Sciences, University of Texas at
Dallas, GR 41, P.O. Box 830688, Richardson, TX 750830688, USA.
Tel.: 1-972-883-2313.
E-mail: amoller@utdallas.edu (A.R. Mller).
Moving blood vessels off the facial nerve can cure

HFS with a success rate of 85% (Barker et al.,
1995). Clinical experience and observations during
MVD operations have shown that the offending vessel
is most often located near the root exit zone (REZ) of
548
the facial nerve, but vessels that are in contact with

other parts of the intracranial portion of the facial
nerve may also be involved. While it was believed earlier that the offending vessels were arteries, it has been
shown that moving arterioles, and even veins off the
intracranial portion of the facial nerve, can effectively
cure the disorder. Intraoperative monitoring of the
abnormal muscle response can increase the success
rate of the MVD operation (Mller and Jannetta,
1987; Mller, 2006a).
Partial destruction of the temporal branch of the
facial nerve has been used to treat HFS (Fisch and
Esslen, 1972) and injections of botulinum toxin
(Botox) into facial muscles is also used (Marion,
1997). Partial section of the facial nerve has severe
side effects and treatments with botulinum toxin must
be repeated every 34 month, and it has long-term side
effects (Mller, 2006b). Unlike TGN and GPN, no
effective medical treatment has been found for treating
HFS. The most common current treatment of HFS is
MVD of the intracranial portion of the facial nerve.
39.2.4. Pathophysiology of HFS
Conflicting hypotheses have been presented regarding
the pathophysiology of HFS and the anatomical location of the physiological abnormality that causes the
symptoms of spasm and synkinesis that are characteristic for the disorder. One hypothesis states that the symptoms and signs are caused by ephaptic transmission at
the location of the vascular contact (Gardner, 1966;
Nielsen, 1984a). Another hypothesis postulates that
the signs and symptoms are caused by hyperactivity of
the facial motonucleus (Ferguson, 1978). While the
hypothesis about ephaptic transmission was favored
for many years, the nucleus hypothesis was supported
by the results of electrophysiological recordings made
during MVD operations (Mller, 1993). In addition,
clinical studies of patients with HFS (Esteban and
Molina-Negro, 1986) have provided support for the
hypothesis that changes in the function of the facial
motonucleus (hyperactivity and increased crossconnections between motoneurons) are the cause of
the spasm and the synkinesis. There are indications that
these changes are brought about by expression of neural
plasticity (or the kindling phenomenon) (Mller, 1993).
39.2.4.1. Anatomical location of the physiological
abnormality
Intraoperative studies of patients undergoing MVD
operations for HFS using recordings of the abnormal
A.R. MLLER
muscle response have contributed to understanding

the pathophysiology of HFS. The activity that is elicited in one branch of the facial nerve by electrical
stimulation propagates in both directions from the
location of the stimulation. The orthodromic activity
elicits a (direct) muscle contraction in the muscles
that are innervated by the branch of the facial nerve
that is stimulated. The antidromic activity elicits the
abnormal muscle contraction by somehow crossing
over to other branches of the facial nerve causing
contraction of muscles that are innervated by these
branches. The anatomical location where this crossover occurs is most likely also the location of the
abnormality that causes the synkinesis that is a part
of the signs of HFS, and it is also probably the site
where the neural activity that causes the spasm is
generated.
It has been hypothesized that the anatomical location of this cross-transmission occurs in the facial
nerve at the location where the blood vessel is compressing the nerve root by ephaptic transmission
(Gardner, 1966). The other possibility is that the
cross-transmission occurs in the facial motonucleus
(Ferguson, 1978).
Measurements of neural conduction times made
during operations for HFS revealed that the latency
of the abnormal muscle response was longer than that
which could be explained by cross-transmission at
the site of the vascular contact (Mller and Jannetta,
1984). In rare instances, when the facial nerve has
been injured, signs of ephaptic transmission lasting
a few minutes have been observed (Mller, 1987).
These observations supported the hypothesis that
the cross-transmission occurred in the facial motonucleus where the antidromically elicited neural activity
in the facial nerve reached the facial nucleus and fired
motoneurons. The cross-transmission between motoneurons was supposed to occur through interneurons
in the facial motonucleus and thereby activating motoneurons that innervate nerve fibers of other branches of
the facial nerve.
These conclusions were supported by clinical
observations that did not favor the ephaptic hypothesis
(Esteban and Molina-Negro, 1986) because it seems
unlikely that a sufficiently large portion of axons in
the facial nerve would be denuded and be in contact
with each other to account for the massive contractions
of facial muscles that occur in late stages of the
disorder.
The fact that the facial nucleus is hyperactive in
HFS is also supported by studies of the blink reflex.
It can be elicited during general anesthesia on the

spasm side, but is suppressed on the normal side
(Mller and Jannetta, 1986). The blink reflex has
two components, R1 and R2, when elicited by stimulation of the supraorbital nerve in awake individuals,
and when the R2 component is bilateral. The R1
component can normally only be recorded from the
muscles around the eye, but, in patients with HFS,
it can often be recorded from other facial muscles
including the mentalis muscle (Nielsen, 1984b).
In anesthetized patients, the blink reflex is usually
suppressed. In patients with HFS, it can be elicited
from the supraorbital nerve on the side of the spasm
but only the R1 component is present. The fact that
the R1 response can be obtained on the spasm side,
despite the suppression from general anesthesia, has
been taken as a support for the hypothesis that the
facial motonucleus is hyperactive in HFS.
This fact, however, assumes that hyperactivity in
the facial motonucleus compensates for the suppression from anesthesia.
Since moving small arteries, and even veins, can
cure HFS, it seems unlikely that the symptoms of
HFS are primarily caused by an affect on the facial
nerve. It seems more likely that the symptoms and
signs of HFS are caused by changes in the function of
the facial motonucleus brought about by expression
of neural plasticity, elicited by abnormal neural activity that is generated by the close contact with a blood
vessel and conducted antidromically to the facial
motonucleus. This hypothesis is supported by animal
experiments which showed that hyperactivity and
cross-communication between branches of the facial
nerve could be induced by daily electrical stimulation
of the peripheral part of the facial nerve in rats (Sen
and Mller, 1987; Kuroki and Mller, 1994). The
offending blood vessel may cause slight injury to the
facial nerve; injuries to the facial nerve can influence
the function of the facial motonucleus has been confirmed in other studies (Tetzlaff et al., 1986; Mller,
2006b).
Intraoperative studies during MVD operations for
HFS have shown that the abnormal muscle response
disappears when the offending vessel is lifted off of
the facial nerve (Fig. 1). This observation is the basis
of the use of monitoring the abnormal muscle activity
as a guide for MVD operations, and it indicates that
the activity that is presumably generated by the close
contact with a blood vessel is necessary to maintain
the hyperactivity of the facial motonucleus (Mller,
1993).
549
39.2.5. Cause of HFS

The fact that moving a blood vessel off of the intracranial portion of the facial nerve can cure HFS does not
mean that the close contact between a vessel and the
facial nerve causes the symptoms of HFS. The fact that
close contact between the facial nerve root and blood
vessels is common (Sunderland, 1948) means that the
presence of another factor must be necessary in order
that symptoms become manifest (Mller, 1993). Since
both (or all) factors must be present for symptoms to
occur, it is sufficient to remove one factor to cure
HFS. Therefore, moving a vessel away from the facial
nerve root is an effective cure, although the close
contact with a blood vessel is not the cause of the
symptoms and signs of HFS.
39.3. Intraoperative monitoring in MVD
operations
Intraoperative neurophysiologic monitoring during
MVD operations for HFS has two purposes: one, to
assist the surgeon in the operation with the goal of
achieving the therapeutic goal, and the other is reducing the risk of injury to neural structures that may be
manipulated in the operation (mainly the facial and
the auditory nerves).
39.3.1. Use of the abnormal muscle response
The fact that the abnormal muscle response disappears
after the offending vessel has been moved off of the
facial nerve root (Mller and Jannetta, 1987; Haines
and Torres, 1991) (Fig. 1) makes monitoring of the
abnormal muscle responses a valuable tool to find the
offending vessel and to ensure that it has been moved
away from the facial nerve. In many cases, this can
be done by visual inspection. In some patients, there
are several vessels in contact with the nerve root and
usually only one of these needs to be moved to effectively treat the patient. In some cases, moving a vessel
off of the facial nerve does not make the abnormal
muscle response disappear, and that is interpreted to
mean that the vessel was not involved in the symptoms
of HFS, and therefore another vessel has to be found in
order to make the abnormal muscle response disappear. In some situations, visual inspection fails to
uncover any vessel that is in contact with the facial
nerve, but more extensive exploration may identify a
small artery or a vein. Recordings of the abnormal
muscle response can determine where a small vessel
550
A.R. MLLER
VESSEL BACK ON
100 V
VESSEL
OFF
10
20
TIME IN ms
30
40
10
20
TIME IN ms
30
40
10
20
30
40
TIME IN ms
Fig. 1. Electromyographic (EMG) recordings from a patient undergoing microvascular decompression (MVD) to relieve
hemifacial spasm (HFS). Consecutive recordings (beginning at top) from the mentalis muscle in response to electrical stimulation of the zygomatic branch of the facial nerve are shown. As indicated, the recordings in the middle of the left hand
column were made when the vessel was lifted off the nerve. The middle and right hand had columns show recordings
obtained after that the offending vessel fell back onto the facial nerve. (Reprinted from Mller and Jannetta (1985a) with
permission from Lippincott Williams and Wilkins).
is implicated in the disorder. In a few cases, more than

one vessel must be moved off of the facial nerve before
the abnormal muscle response disappears totally.
Some investigators have stated that it is not necessary that the abnormal muscle response disappear, to
ensure freedom of symptoms postoperatively (Hatem
et al., 2001). However, I have found that the best
results are obtained when the operation results in a
total disappearance of the abnormal muscle response.
To test if the abnormal muscle response has totally
disappeared, the rate of stimulation must be increased
to 50 pps for a few seconds. If that does not bring
back the abnormal muscle response, it is a strong
indication that the patient will be free of symptoms
immediately after the operation, or within a few days.
This means that monitoring of the abnormal muscle

response is important in MVD operations for HFS,
and it can increase the success rate and eliminate
the need for reoperations. Use of such monitoring
can increase the success rate of the operation from
85% to 97% (Mller and Jannetta, 1987; Haines
and Torres, 1991).
The blink reflex response also disappears when
the offending blood vessel is moved off of the facial
nerve. Monitoring of that may serve a similar purpose as monitoring of the abnormal muscle response
in MVD operations for HFS. However, there does not
seem to be any specific advantages in the use of the
blink reflex over monitoring of the abnormal muscle
response.
39.3.1.1. Practical aspects of monitoring the

abnormal muscle response
Monitoring of the abnormal muscle response during
operations for HFS is best done using needle electrodes for stimulation, as well as for recordings.
Two electrodes are placed in the mentalis muscle
for recording, and two electrodes are placed near
the temporal and/or the zygomatic branch of the
facial nerve that innervates the orbicularis oculi
muscles (Fig. 2). The electrodes should be placed at
a distance of 1 cm, with the negative electrode closest to the ear. The direct muscle response can be
recorded from the orbicularis oculi muscles. Stimulating and recording electrodes should be inserted
after the patient has been anesthetized. With the stimulator set to deliver 0.1 ms impulse rate at a rate of
5 pps and at 20 V (25 mA), the stimulating electrodes are placed while observing the EMG responses
that are recorded from the mentalis and the orbicularis oculi muscles to find the best location for the
stimulating electrodes. Normally, the best placement
is along a line between the corner of the eye and
ear canal, but there is considerable individual variation in the anatomical locations of the branches of
the facial nerve.
When all the electrodes are in place, the stimulus
strength should be lowered to find the threshold for
eliciting the abnormal muscle response. A stimulus
Fig. 2. Electrode placement for recording the abnormal

muscle response and the direct muscle response. [Reprinted
from Mller and Jannetta (1985b) with permission from
Birkhauser Verlag, Basel.]
551
level of 2030% above threshold will usually provide

a stable abnormal muscle response. A stimulus repetition rate of 12 pps is suitable for monitoring
purposes.
Amplifiers for the EMG potentials should have filter settings at 103,000 Hz, and the recorded EMG
potentials should be made audible by using a similar
technique with suppression of the stimulus artifact as
was used with intraoperative monitoring of the function of the facial nerve during removal of vestibular
schwannoma (Mller, 2006a).
The abnormal muscle response has a latency of
10 ms and the initial EMG response is likely to be
followed by trains of EMG activity (Figs. 1 and 3).
These later components are less stable than the initial
10 ms response. The late responses have little importance for monitoring, and they may disappear totally
during the operation, and only the 10 ms component
may be present. Similar EMG potentials can be
recorded from the orbicularis oculi muscles in
response to stimulation of the marginal mandibular
branch. But recordings of the responses from the mentalis muscle elicited by stimulation of the temporal or
zygomatic branch of the facial nerve are usually more
stable than those recorded from the orbicularis oculi, in
response to stimulation of the marginal mandibular
branch of the facial nerve.
39.3.1.2. Interpretation of recordings of the
abnormal muscle response
Disappearance of the abnormal muscle response is a
sign that the offending vessel is moved off the facial
nerve. However, the abnormal muscle response may
disappear before access to the facial nerve root is
gained. One reason for that may be that opening the
dura or that retraction of the cerebellum may have
moved the offending vessel off the facial nerve. This
could occur if the offending vessel is in loose contact
with the facial nerve. Unknown factors may also cause
the abnormal muscle response to disappear. If the
abnormal muscle response disappears while the
offending vessel is still in contact with the facial
nerve, the abnormal muscle response can be brought
back by increasing the rate of stimulation for a short
period of time (Fig. 4). The increase in the rate of
stimulation may have to be repeated several times
during an operation. This often occurs in patients
who have had their HFS for a short time. In patients
who have had their symptoms for many years, the
offending vessel is often held tight in contact with
the facial nerve by arachnoidal bands. In such cases,
552
A.R. MLLER
M. ORB. OCULI
M. MENTALIS
200 V
50
100
TIME IN ms
150
200
50
100
TIME IN ms
150
200
Fig. 3. Electromyographic (EMG) response from the orbicularis oculi (left hand records) and mentalis muscles (right hand
records) in response to stimulation of the zygomatic branch of the facial nerve recorded from a patient undergoing microvascular decompression (MVD) to relieve hemifacial spasm (HFS). The recordings were obtained after the patient was
anesthetized, but before the operation was begun. [Reprinted from Mller and Jannetta (1986a) with permission from
Elsevier.]
the abnormal muscle response remains stable until the

offending vessel is moved off of the nerve.
When a vessel is moved away from the facial
nerve and the abnormal muscle response disappears,
it is important to increase the stimulus rate to make
sure that the abnormal muscle response has disappeared totally. If not, most likely, another vessel is
in contact with the facial nerve. An operation for
HFS should not be regarded complete until the
abnormal muscle response has disappeared and that
increasing the stimulus rate cannot bring it back.
Disappearance of the abnormal muscle response
does not only occur because of removing the effect
of a blood vessel on the facial nerve, but it can
also occur because of injury to the facial nerve.
Recording of the abnormal muscle response thus
does not allow distinguishing between the beneficial

effect of moving a blood vessel off of the facial nerve
and injury to the facial nerve, thus a serious complication to the operation. The function of the facial
nerve must therefore be monitored by stimulating
the nerve electrically by a handheld electrode, as is
done in other operations where the facial nerve is at
risk of being injured.
39.3.2. Requirements of the surgeon and the
physiologist
Like other forms of intraoperative neurophysiologic
monitoring, it is important that the physiologist who
is responsible for the monitoring understands the
physiology upon which the monitoring is based and
553
39.3.3. Benefits from monitoring the abnormal

muscle response
Sti Temp. NVII

M. Mentalis
5pps
50pps
5pps
0.5 mV
10
15
20 ms
Fig. 4. Recordings of the abnormal muscle response

obtained before the offending vessel was moved off the
facial nerve. Consecutive recordings (beginning at top)
from the mentalis muscle in response to electrical stimulation of the temporal branch of the facial nerve are shown.
The effect on the abnormal muscle response from increasing the stimulus rate from 5 to 50 pps for a short period
of stimulation is shown in a patient in whom the abnormal
muscle response was very small when recorded in the
beginning of the operation. [Reprinted from Mller and
Jannetta (1986b) with permission from Elsevier.]
is trained in the practical aspects of the monitoring.

The physiologist must interpret the results and the
surgeon must understand the meaning of the results
and know how to act in the different situations that
may result. This means that the physiologist must
be well trained, not only in the practical aspects of
the monitoring, but he/she must also understand the
theoretical basis for the particular recordings that
are made, and for interpreting the results. The surgeon must also be willing to follow the suggestions
made by the physiologist.
Monitoring of the abnormal muscle contraction can

increase the success rate of MVD operations for HFS
by helping the surgeon identify the vessel that is
involved in causing the symptoms (Mller and Jannetta,
1987). As a result, it can also reduce the operating time
and reduce the amount of dissections that are needed,
and thereby reduce the risk of complications. The benefits of monitoring the abnormal muscle contraction in
MVD operations for HFS (Mller and Jannetta, 1987)
have been confirmed by other investigators (Haines
and Torres, 1991) (Fig. 5).
The benefit of monitoring the abnormal muscle
response in MVD operations for HFS depends on the
experience of the surgeon. The less experienced surgeon may benefit from such monitoring in all steps in
the process of moving blood vessels off of the facial
nerve root; the more experienced surgeon may benefit
only in difficult cases, and the very experienced surgeon may have limited benefit from monitoring of
the abnormal muscle response. Indeed, some very
experienced surgeons have reported good outcome of
MVD operations without the use of monitoring of the
abnormal muscle response (Hatem et al., 2001).
The difference in benefit from such monitoring is
not only related to the experience of the surgeons, but
it is also related to the way the abnormal muscle
response is monitored, and how the recorded EMG
potentials are interpreted. While it is true that patients
in whom the abnormal muscle response is present at
the end of the operation may experience a decrease
and eventual disappearance of the symptoms of HFS,
it is my experience that absence of the abnormal muscle response at the end of the operation provides the
best possibility of immediate (and permanent) relief
of symptoms. In this connection, absence means that
it is not possible to bring back the abnormal muscle
response by increasing the rate of stimulation, a test
that is necessary to make sure the therapeutic goal of
the operation has been achieved.
39.3.4. Monitoring the facial and the auditory
nerves to reduce the risk of injury
While monitoring of the abnormal muscle response is
done to assist the surgeon in achieving the therapeutic goal of the operation, it is also important to
include monitoring for reducing the risk of complications in these operations. Injury to the facial nerve
554
A.R. MLLER
2 mV
10
20
30
40
50 ms
2 mV
10
20
30
40
50 ms
Fig. 5. Electromyographic (EMG) recordings of the abnormal muscle response in a patient who was undergoing microvascular decompression (MVD) to relieve hemifacial spasm (HFS). Consecutive recordings (beginning at top) from the mentalis muscle are shown. A: Recordings done before the dura was opened. B: Recordings obtained after the dura was opened
showing only the initial component of the abnormal muscle response. The recordings in the middle of this column were
obtained when the vessel was moved off the nerve after which the abnormal muscle response became absent. [Reprinted
from Mller and Jannetta (1987), with permission from the Journal of Neurosurgery.]
can occur because of surgical manipulations and by

electrocoagulation. It is therefore important to monitor
the function of the facial nerve during these operations
by stimulating it using a handheld stimulating electrode in the same way as is done, for example, in
operations for vestibular schwannoma (Mller,
2006a) (see Chapter 41). The facial EMG responses
for such monitoring can be recorded using the same
types of electrodes and amplifiers that are used for
monitoring the direct muscle response in connection

with monitoring of the abnormal muscle contraction.
The auditory nerve is also at risk in MVD operations
for HFS. Before intraoperative monitoring of auditory
evoked potentials was introduced, hearing loss was a
common complication to MVD operations for HFS
(Mller and Mller, 1989). It is therefore important to
monitor the function of the auditory nerve in these
operations (Mller, 2006a) (see Chapter 41).
References
Auger, RG and Whisnant, JP (1990) Hemifacial spasm in
Rochester and Olmsted County, Minnesota, 1960 to
1984. Arch. Neurol., 41: 12331234.
Barker, FG, Jannetta, PJ, Bissonette, DJ, Shields, PT and Larkins, MV (1995) Microvascular decompression for hemifacial spasm. J. Neurosurg., 82: 201210.
Esslen, E (1957) Der Spasmus facialis eine Parabiosserscheinung: elektrophysiologische Untersuchnungen
zum Enstehungsmechanismus des Facialisspasmus.
Dtsch. Z. Nervenheilk., 176: 149172.
Esteban, A and Molina-Negro, P (1986) Primary hemifacial
spasm: a neurophysiological study. J. Neurol. Neurosurg. Psychiatry, 49: 5863.
Ferguson, JH (1978) Hemifacial spasm and the facial
nucleus. Ann. Neurol., 4: 97103.
Fisch, U and Esslen, E (1972) The surgical treatment of
facial hyperkinesis. Arch. Otolaryngol. Head Neck
Surg., 5: 400405.
Gardner, W (1966) Crosstalk the paradoxical transmission of a nerve impulse. Arch. Neurol., 14: 149156.
Haines, SJ and Torres, F (1991) Intraoperative monitoring of
the facial nerve during decompressive surgery for hemifacial spasm. J. Neurosurg., 41: 254257.
Hatem, J, Sindou, M and Vial, C (2001) Intraoperative monitoring of facial EMG responses during microvascular
decompression for hemifacial spasm. Prognostic value
for long-term outcome: a study in a 33-patient series. Br.
J. Neurosurg., 15: 496499.
Kuroki, A and Mller, AR (1994) Facial nerve demyelination and vascular compression are both needed to
induce facial hyperactivity: a study in rats. Acta Neurochir. (Wien), 126: 149157.
Marion, M-H (1997) Hemifacial spasm: treatment with
botulinum toxin (long term results). In: M Sindou, Y
Keravel and AR Mller (Eds.), Hemifacial Spasm.
A Multidisciplinary Approach.Springer, Wien, pp.
141144.
Mller, AR (1987) Hemifacial spasm: ephaptic transmission or hyperexcitability of the facial motor nucleus?
Exp. Neurol., 98: 110119.
Mller, AR and Jannetta, PJ (1985a) Microvascular
decompression in hemifacial spasm: intraoperative
electrophysiological observations. Neurosurgery, 16:
612618.
555
Mller, AR and Jannetta, PJ (1985b) Synkinesis in hemifacial
spasm: results of recording intracranially from the facial
nerve. Experientia, 41: 415417.
Mller, AR (1993) Cranial nerve dysfunction syndromes:
pathophysiology of microvascular compression. In: DL
Barrow (Ed.), Neurosurgical Topics Book 13, Surgery
of Cranial Nerves of the Posterior Fossa.American
Association of Neurological Surgeons, Park Ridge, IL,
Chapter 2.
Mller, AR (2006a) Intraoperative Neurophysiologic Monitoring. Humana Press Inc., Totowa, New Jersey, 2nd ed.
Mller, AR (2006b) Neural Plasticity and Disorders of the
Nervous System. Cambridge University Press, Cambridge.
Mller, AR and Jannetta, PJ (1984) On the origin of synkinesis in hemifacial spasm: results of intracranial
recordings. J. Neurosurg., 61: 569576.
Mller, AR and Jannetta, PJ (1986a) Blink reflex in
patients with hemifacial spasm: observations during
microvascular decompression operations. J. Neurol.
Sci., 72: 171182.
Mller, AR and Jannetta, PJ (1986b) Physiological abnormalities in hemifacial spasm studied during microvascular
decompression operations. Exp. Neurol., 93: 584600.
Mller, AR and Jannetta, PJ (1987) Monitoring facial EMG
during microvascular decompression operations for
hemifacial spasm. J. Neurosurg., 66: 681685.
Mller, AR and Mller, MB (1989) Does intraoperative
monitoring of auditory evoked potentials reduce incidence of hearing loss as a complication of microvascular decompression of cranial nerves? Neurosurgery, 24:
257263.
Nielsen, V (1984a) Pathophysiological aspects of hemifacial
spasm. Part I. Evidence of ectopic excitation and ephaptic
transmission. Neurology, 34: 418426.
Nielsen, VK (1984b) Pathophysiology of hemifacial spasm:
II. Lateral spread of the supraorbital nerve reflex. Neurology, 34: 427431.
Sen, CN and Mller, AR (1987) Signs of hemifacial spasm
created by chronic periodic stimulation of the facial
nerve in the rat. Exp. Neurol., 98: 336349.
Sunderland, S (1948) Microvascular relations and anomalies at the base of the brain. J. Neurol. Neurosurg. Psychiatry, 11: 243257.
Tetzlaff, W, Graeber, MB and Kreutzberg, GW (1986)
Reaction on motoneurons and their microenvironment
to axotomy. Exp. Brain Res., 3: 38.

M.R. Nuwer (Ed.)
556
CHAPTER 40
Neurophysiology during middle ear, mastoid and

parotid surgery
Tina C. Huang and Anil K. Lalwani*
Department of Otolaryngology, New York University School of Medicine, New York, NY 10016, USA
Intraoperative monitoring during middle ear, mastoid,

and parotid surgery with the equipment and techniques
currently used is a fairly recent phenomenon. Techniques for auditory monitoring were discovered prior
to techniques for facial nerve monitoring, but the popularity of facial nerve monitoring has surpassed auditory
monitoring for middle ear and mastoid surgery. Facial
nerve monitoring during parotid gland surgery started
at about the same time that it began being used for
otologic surgery. Some of the goals of intraoperative
facial nerve monitoring include (1) early identification
of the facial nerve with the use of stimulating probes/
tools, (2) warning the surgeon of unexpected facial
nerve anatomy, (3) mapping the course of the facial
nerve using stimulation, (4) reducing mechanical trauma
to the nerve, and (5) evaluation and prognosis of nerve
function at the conclusion of surgery. The goals of auditory monitoring include (1) avoidance of permanent sensorineural hearing loss and (2) evaluating hearing
function prior to the conclusion of surgery. This chapter
will discuss some outcomes using intraoperative monitoring as well as briefly review techniques and pitfalls
specific to middle ear, mastoid, and parotid surgery.
40.1. Facial nerve monitoring in parotid surgery
40.1.1. Clinical outcome studies
One of the most critical parts of parotid gland surgery
with facial nerve preservation is the identification of
the facial nerve between the superficial and deep lobe
of the parotid. The most commonly used technique
for finding the nerve is to locate the main trunk of
*
Correspondence to: Anil K. Lalwani, M.D., Department of

Otolaryngology, New York University School of Medicine,
550 First Avenue, NBV 5E5-10, New York, NY 10016,
USA.
Tel.: 1 212-263-6344; fax: 1 212-263-8257.
E-mail: anil.lalwani@nyumc.org (A.K. Lalwani).
the nerve as it exits the stylomastoid foramen and then

to follow the trunk distally to its branches. While the
main trunk is in a fairly predictable location, factors
such as prior surgery and alteration of anatomy by
inflammation or tumor growth may make its identification more difficult. Published reports of the incidence of permanent facial paresis after parotidectomy
may be as high as 5.6% while the incidence of
temporary paresis may range up to 71%.
The use of electrical stimulation to aid in the identification of the nerve was published as early as the
1950s. Parsons (1968) described the first portable unit
for facial nerve stimulation. His device consisted of a
pulse generator powered by batteries, a unipolar search
probe, and a ground plate electrode. He used the stimulator to identify the main trunk and trace out the distal
branches using the stimulator at a lower intensity and,
at the conclusion of the surgery, confirm the integrity
of the nerve. He found that if the facial muscles moved
at a low setting, the patient would have normal nerve
function whereas if the nerve required higher stimulation, there would be temporary weakness and, if the
nerve had no response to any stimulus, reexploration
would be in order.
Continuous passive facial nerve monitoring began
in the late 1980s. The early electromyographic (EMG)
monitoring systems employed both a viewing screen
showing the electromyogram as well as an audible
signal when a signal was recorded. Typically, the
earlier reports also mandate an assistant trained in
electroneurophysiology to monitor the EMG. Rea
(1990) used active electrical stimulation paired with
continuous EMG monitoring of the facial musculature. He used two types of stimulator devices hooked
to a hemostat which were used to dissect the facial
nerve. While no objective measures were reported,
he found the dissections subjectively easier and was
able to perform the operations with greater confidence. Anon et al. (1991) also found the use of
continuous EMG monitoring with stimulator helpful

in identifying and tracing out the facial nerve. They
felt that they were able to detect artifactual stimulation from true neural stimulation both from the
waveform and the quality of the audible alarm and
no longer required the use of an assistant to monitor
the monitor. They reported on a series of 40 patients
and had only two incidences of permanent palsy of a
branch of the facial nerve in patients with malignancy that involved those branches and two temporary palsies. They also noted that spontaneous
activity may represent traction on the nerve, a lightening of anesthesia, or thermal irritation from cold
irrigation (Anon et al., 1994).
Doikov et al. (2001) performed a retrospective
review of 15 patients who underwent parotid gland
surgery and found no cases of either temporary or
permanent facial paresis. A prospective study by
Brennan et al. of 44 nerve-sparing parotid surgeries
using continuous EMG monitoring found an incidence of temporary paresis of 15.9% and a 0% incidence of permanent paresis (Brennan et al., 2001).
They noted a higher incidence of temporary paresis
after total parotidectomy (23%) than after superficial
parotidectomy (12.9%). In addition, they noted that
one patient with total postoperative paralysis required
a higher level of stimulation at the nerve trunk, but
found no predictive value at the distal branches.
Dulguerov et al. (1999), using either a custom
mechanical transducer placed in the mouth between
the teeth and cheek and a commercially available
EMG monitoring device, reported a 27% incidence
of temporary paresis and a 4% incidence of permanent paresis with no temporary nerve deficits in
patients undergoing resection of benign tumors.
None of the previous studies have compared the
results of parotid surgeries performed with and without facial nerve monitoring so, even though all the
authors advocate its use, it is difficult to judge the
utility of monitoring. Witt (1998) performed a retrospective review of 53 consecutive parotid surgeries
with the initial 33 surgeries performed without a
monitor and the last 20 with monitoring. He found
no statistically different results with the facial nerve
monitor. There was an overall temporary paresis rate
of 17% with 15% of the unmonitored patients and
20% of the monitored patients developing paresis.
There were no cases of permanent facial nerve
injury. However, his study excluded patients deemed
at high-risk including those with preoperative
facial nerve dysfunction, multiple parotid lesions,
557
neck metastasis, fixed tumors, deep-lobe tumors, or

revision cases. Both Brennan et al. and Dulguerov
et al. included total parotidectomy cases within their
series. Another retrospective study by Terrell et al.
(1997) also compared monitored and unmonitored
cases. Their study also excluded revision cases, but
included patients with deep-lobe tumors. Their overall rate of temporary facial paresis was 53.5%. The
62.3% rate of early paresis in unmonitored patients
was statistically higher than the 43.6% in monitored
patients. While there was no difference in the rate
of permanent facial paresis between the two groups,
only the patients undergoing a simple parotidectomy
were included in the analysis.
The studies done so far have not shown conclusively that facial nerve monitoring during parotid
surgery is beneficial. However, one must keep in
mind that most of these studies include patients with
benign tumors who are undergoing uncomplicated
parotidectomies. The risks of facial nerve injury
have been shown in some studies to be higher for
patients with parotid gland malignancies and revision
surgeries.
40.1.2. Practice patterns
Anonymous questionnaires were sent to members of
the American Academy of Otolaryngology-Head
and Neck Surgery in 2004 with a response rate of
49.3% (Lowry et al., 2005). The majority of physicians performed less than 20 parotidectomies per
year with only 6.7% of respondents stating they performed no parotid surgery. 60% of surgeons used
facial nerve monitoring some or all of the time during parotid gland surgery and, though a minority
had used it during training; they were 5.6 times more
likely to use it if they had used it during training.
In addition, fellowship trained physicians were
significantly more likely to perform more than 10
or 20 parotidectomies per year and those surgeons
performing more than 10 or 20 parotidectomies per
year were significantly more likely to use the facial
nerve monitor. Also, those performing more than 10
parotidectomies per year were less likely to have a
history of self-reported inadvertent facial nerve
injury and those who used the facial nerve monitor
were less likely to have a history of parotid gland
surgery-related lawsuits. Interestingly, no significant
associations were noted between fellowship training
and the use of monitoring. The majority of those
using monitoring used passive monitoring.
558
40.1.3. Possible indications

While previous parotid surgery is a widely accepted
risk factor for increased facial nerve injury, other risk
factors include (1) more extensive surgery, (2) malignant tumors, (3) larger lesion size, and (4) inflammatory conditions. Dulguerov et al. (1999) found that
total parotidectomy, malignancy, increased tumor
size, infections, and duration of surgery significantly
correlated with a worse postoperative facial nerve outcome. Terrell et al. (1997) showed that increased age,
increased operative time, and lack of monitoring correlated significantly with increased temporary paresis.
A study comparing patients undergoing monitored
and unmonitored surgery for recurrent pleomorphic
adenoma of the parotid found that while all their
patients suffered immediate postoperative palsy, a significant increase in the degree of postoperative facial
nerve palsy was noted in the unmonitored group
(Makeieff et al., 2005). The degree of palsy also correlated with the facial nerve recovery time and monitoring significantly decreased operative time. With these
results in mind, one should strongly consider the use
of facial nerve monitoring in patients at increased
risk of facial nerve injury.
40.2. Facial nerve monitoring during middle ear
and mastoid surgery
40.2.1. Clinical outcome studies
The entire intratemporal facial nerve may be at risk
during middle ear and mastoid surgery. The vertical
segment of the facial nerve should be identified when
opening a facial recess tract and the tympanic portion
should be identified during any middle ear surgery.
The facial nerve has been noted to have areas of dehiscence in temporal bone studies and the bony fallopian
canal can be eroded by cholesteatoma or granulation
tissue. In addition, prior surgery can lead to iatrogenic
dehiscence and distortion of the usual landmarks. The
incidence of iatrogenic facial nerve injury during middle ear and mastoid surgery ranges between 0.6% and
3.7% while the incidence during revision surgery
ranges from 4% to 10% (Wilson et al., 2003).
Along with passive facial nerve monitoring during
parotid gland surgery, facial nerve monitoring during
otologic surgery began to be popularized in the late
1980s. An EMG monitoring device was, and continues
to be, the most common device for both types of surgery. Several of the initial papers simply describe
T.C. HUANG AND A.K. LALWANI
new devices used for otologic monitoring (Metson

et al., 1988; Beck and Benecke, 1990). These EMG
monitors utilized needle electrodes with a monitor
showing the continuous EMG activity as well as an
audible alarm indicating a neural stimulus. These
devices were all used in conjunction with a stimulating
probe for active identification of the nerve. One of the
benefits of the new devices was their simplified design
which could eliminate the need for an additional technician to aid in monitoring. However, many institutions continued to utilize an electrophysiologist.
Leonetti et al. (1990) described four difficult otologic cases where they felt that EMG monitoring aided
in the identification of the facial nerve and possibly
averted injury. While they utilized monitoring in a
minority (12%) of their major otologic procedures,
they suggested that monitoring would be helpful in
patients who had a higher risk of facial nerve injury
based on clinical and radiographic findings. However,
they warned that monitoring did not prevent facial
nerve injury and that it was only a supplement to a
surgeons knowledge. Noss et al. (2001) studied 262
middle ear and mastoid surgeries of which 32% were
revision surgeries. They found facial nerve dehiscence
in 10% of the primary surgeries and in 20% of the revision surgeries. 61% of the cases overall had a stimulation event during surgery with 39% of the first
events being a mechanical stimulation. However,
the ears that had dehiscent facial nerves had a stimulation event rate of 89%. They suggested that the electrical stimulation threshold was a reliable marker for
true dehiscence and that a threshold of less than 1 V
indicated a nerve that was electrophysiologically
dehiscent. When they used electrophysiological criteria, the rate of dehiscence increased to 53% for primary surgeries and 96% for revision surgeries. They
concluded that monitoring helped avoid facial nerve
injury in 12% of their cases.
The group at the Silverstein Institute wrote several
articles regarding the use of a mechanical-pressure
detection system for facial nerve monitoring.
A strain-gauge sensor was placed in the ipsilateral corner of the mouth and facial contraction was alerted
with an audible signal. A monopolar stimulus probe
was used to facilitate nerve identification. They also
developed a device to send current through the drill
and microinstruments in order to stimulate as they
were working. They found that the stimulation threshold predicted the depth of bony covering over the
facial nerve and that 1.0 mA of current was approximate to 1.0 mm of bony covering. In addition, they
comparison studied a mechanical-pressure device, the

Silverstein facial nerve monitor/stimulator, with an
EMG device, the Brackmann EMG system. They
found that the EMG required a significantly lower
stimulation to activate the audible alarm and that the
EMG was significantly more sensitive when an
exposed facial nerve trunk was stimulated. Their group
routinely uses a mechanical-pressure device for all
otologic surgeries and uses both a mechanical-pressure
device as well as an EMG device during neurotologic
surgery. They concluded that a mechanical-pressure
device was suitable for otologic surgery because,
though the EMG was more sensitive, it generated a
great number of false-positive alarms, was more difficult to set up, was more expensive, and more invasive
(Silverstein et al., 1988a,b; Silverstein and Rosenberg,
1991; Bendet et al., 1999).
40.2.2. Cost of monitoring
Wilson et al. (2003) published a cost analysis of
intraoperative facial nerve monitoring during middle
ear and mastoid surgery. Their primary outcome was
incremental cost per incremental quality-adjusted
life-year (QALY) saved. They found that the use of
monitoring during all otologic surgeries was associated with the lowest cost and highest efficacy, followed by monitoring during revision surgeries only.
While the use of monitoring added $222.73 to
$528.00 to each surgery, this was offset by the reduction in costs associated with the management of facial
palsy. Note that their costs included a neurophysiologist monitoring the monitor. The use of an additional
neurophysiologist, instead of the surgeon alone, to
aid in monitoring the EMG activity is advocated by
several groups while other groups find it unnecessary
(Edwards and Kileny, 2005).
40.2.3. Practice patterns
Several surveys regarding usage of intraoperative
facial nerve monitoring have been done. Roland and
Meyerhoff (1994) sent questionnaires to all members
of the American Otological and American Neurotology Societies asking them to choose between two
statements: (1) standard of care requires electrophysiological facial nerve monitoring for all tympanomastoid surgery, or (2) electrophysiological monitoring
of the facial nerve should be reserved for neurotologic procedures, and those tympanomastoid procedures in which the facial nerve is thought to be at
559
high risk. 64% responded and of those, 95% chose

statement #2, 4% chose statement #1, and 1% wrote
comments but did not choose a statement. A questionnaire regarding usage of facial nerve monitoring
was sent to all members of the E.A.R. Foundation
Alumni with a response rate of 87%. They found that
71% used it in aural atresia cases, 59% used it in
severe chronic ear cases, 38% used it in trauma cases,
21% used it in routine chronic ear cases, and 5% used
it during stapedectomies (Anonymous, 1996). 67%
did not consider monitoring to be standard of care
in their community for chronic ear surgery. Interestingly, 63% somewhat or strongly agreed that it was
a good tool for resident teaching. Olds et al. (1997)
sent a questionnaire to 16 graduates of the Ear
Research Foundation, of whom 15 were practicing
neurotologists. 100% used facial nerve monitoring
for chronic ear surgery, trauma surgery, and surgery
of the facial nerve, while 86% used it during stapes
surgery. 66% did not feel that monitoring was standard of care in their community and 93% strongly
or somewhat agreed that monitoring was useful for
teaching.
While the previous surveys were of otolaryngologists specializing in otologic and neurotologic procedures, Greenberg et al. (2002) conducted a survey of
randomly selected members of the American Academy of Otolaryngology-Head and Neck Surgery. Their
response rate was 45.2%. 76% of members reported
having access to a facial nerve monitor, but only 54%
of surgeons who performed some otologic surgery
used monitoring at least some of the time. Only 32%
thought monitoring should be required for chronic
ear surgery with 4% responding that it was required
for stapedectomy. A significantly higher percentage
of members describing themselves as otologists
used monitoring and, among those who performed
more than 10 otologic surgeries per month, 100% used
monitoring at least some of the time. Also, those in
academic practice and those who finished their training recently were significantly more likely to use monitoring. Again, a majority, 78%, felt that monitoring
should be used during resident training. A survey of
members of the British Association of Otolaryngologists/Head and Neck Surgeons was also reported
(Saravanappa et al., 2003). They had a 49% response
rate with 91% reporting that they performed some otologic surgery. 63% of those who regularly performed
mastoidectomies used monitoring occasionally or routinely and 73% used it occasionally or routinely for
combined approach tympanoplasty. However, 80%
560
of otolaryngologists reported never using facial nerve

monitoring during other types of middle ear surgery.
40.2.4. Possible indications
As is highlighted in the previous section regarding
practice patterns of intraoperative facial nerve monitoring, there is no consensus regarding its routine use
during middle ear and mastoid surgery. Most surgeons agree that monitoring is useful in cases where
the facial nerve may be a higher risk, that is, revision
surgery, extensive disease on clinical or radiological
examination, and radiographic anatomic anomalies.
However, it appears that the majority of otolaryngologists do not feel that monitoring is standard of care
during routine otologic surgery.
There is also debate regarding its utility during resident training. While a majority of surgeons in the
surveys regarded it as a good tool during training, other
surgeons disagreed (Roland and Meyerhoff, 1993;
Silverstein et al., 1994). A study of monitoring in a resident training program found that the supervising faculty
felt that monitoring positively affected the outcome
of 0.4% of the surgeries (Pensak et al., 1994). At our
institution, New York University School of Medicine,
which is a resident training program, all otologic surgeries are performed with the facial nerve monitor.
40.2.5. Possible complications
Haenggeli et al. (1999) reported a case series of three
patients who sustained facial burns from the electrode
sites using the Neurosign 100 device. They attributed
one of the cases to a short circuit caused by the electrode contacting the wet surgical draping. The other
two cases may have been due to a faulty connection
within the device. The manufacturer has since created
a virtual ground to prevent current leaks to the patient.
This appears to be the only report of a complication of
intraoperative facial nerve monitoring.
40.3. Auditory monitoring during middle ear and
mastoid surgery
40.3.1. Auditory brainstem evoked response
Auditory brainstem response (ABR) evolved during
the 1930s when researchers noted a change in the pattern of electroencephalograms (EEGs) with auditory
stimuli. It was not until computer technology was
able to average multiple responses did modern ABR
monitoring begin. Now, ABR is likely the most

widely used type of surgical auditory monitoring.
ABR can be evoked using either a series of clicks or
tone bursts as the stimuli and consists of five waves
corresponding to different locations from distal to
proximal along the auditory pathway. Wave V typically is the most robust and most frequently analyzed.
It is thought to originate from auditory structures near
the inferior colliculus. Gerull et al. (1978) were one of
the first to report its using during tympanoplasty. They
found that an increase in the latency of the brainstem
potentials correlated with a conductive hearing loss
and a decrease in latency correlated with an improvement in hearing. Jansen and Janssen (1980) reported
on their preliminary use of ABR during tympanoplasty
and ossicular chain reconstruction (OCR) and found
that they were not able to accurately predict the
amount of hearing improvement. In 1989, Von Scheel
et al. (1989) reported the use of ABR during OCR.
However, instead of using auditory stimuli, they used
a piezoelectric transducer placed on the ossicles. They
determined that changes in wave V latency approximate changes in hearing loss within 510 dB.
40.3.2. Electrocochleography
The cochlear microphonic (CM), an alternating current electrical potential generated by the cochlear hair
cells, was first discovered in cats in the 1930s. However, it took another decade before it was demonstrated
in humans. Ruben et al. (1959) attempted to record
cochlear microphonics during chronic ear surgery by
placing a recording electrode on or near the round window membrane and using undirected auditory stimuli
and were able to record CM on half of their patients.
A follow-up study used clicks and tones directed to
the ear through a tube placed above the external ear
canal (Ruben et al., 1960). They were able to record
not only a CM, but also the action potential (AP) of
the VIII nerve which consisted of a first slow potential,
N1, and a second slow potential, N2. They discovered
an increase in CM amplitude after stapes footplate
mobilization. Further studies confirmed the increase
in CM amplitude after stapes mobilization without an
increase in the action potential waveforms (Bordley
et al., 1964). They reported a correlation between a
lack of improvement in CM amplitude with a lack of
improvement in hearing.
Modern electrocochleography (ECochG) consists
of three different components. The first component
is the cochlear microphonic. Some researchers use
alternating polarity stimuli to cancel out the CM

because the CM continues as long as auditory stimulus
continues and can overwhelm the remainder of the
ECochG waveform. The second component is the
summating potential (SP) which is a direct current distortion response reflecting the motion of the basilar
membrane. The size of the SP increases with alteration
of the basilar membrane vibration. The action potential is generated by the distal end of the VIII nerve.
The amplitude reflects the number of nerve fibers
synchronously firing. A decrease in latency typically
corresponds with an improvement in function.
Gibson (1992) found no changes in the ECochG
waveform with perforation of either the oval or round
window. The waveform changed only with suctioning
of perilymph and this change was reversible if the perilymph was replaced with an increase in intrathoracic
pressure. Filipo et al. (1993) studied the use of ECochG
during stapes surgery and found that while amplitude
increased and latency decreased after removal of the
stapes suprastructure, the results were more varied after
footplate mobilization. Wazen measured the sound
threshold at which the AP waveform disappeared during stapedectomy. He found a positive significant correlation between the changes in pure tone average
audiometry and the changes in ECochG threshold in
one study, but in a follow-up study, was unable to demonstrate a significant correlation between the improvement in ECochG threshold and degree of hearing
improvement (Wazen, 1994; Wazen et al., 1997).
40.4. Methods
40.4.1. Anesthesia considerations
Good communication with the anesthesia team is
essential. Since muscle relaxants and paralytic agents
will affect the effectiveness of either EMG or
pressure-transducer systems, these agents should be
avoided during surgery (Edwards and Kileny, 2005).
A short-acting paralytic agent may be used for the
induction of anesthesia, but one must be certain that
the medication is cleared from the patient before
beginning facial nerve identification and dissection.
40.4.2. Surgical draping
The use of a clear plastic drape over the patients face
during parotid surgery is advocated by many head and
neck surgeons. This allows for visual monitoring of
facial muscle movement as an adjunct to the use of
561
facial nerve monitoring. Visualization of the face

may also be useful during otologic surgery. However,
one must recognize that a stimulus which causes an
EMG waveform may not activate visible muscle
activity.
40.4.3. Use of local anesthetic
While many surgeons use lidocaine mixed with epinephrine to inject the incision site to achieve vasoconstriction, we recommend the use of dilute
epinephrine alone, particularly during otologic surgery. A local injection of lidocaine, or any local
anesthetic, can cause temporary paralysis of the
facial nerve since the anesthetic can infiltrate medially along the bony external auditory canal into the
middle ear to the facial nerve or through the soft tissues at the mastoid tip to the nerve as it exits the stylomastoid foramen. This will result in false-negative
feedback during facial nerve monitoring and may
lead to unrecognized injury of the nerve. Jones and
Mellert (1991) studied the use of a dilute epinephrine
(1:100,000) solution without a local anesthetic agent
during monitored otologic surgery and found no
adverse events associated with the epinephrine only
injection. We also use a 1:100,000 dilution of epinephrine during our otologic procedures and, to further decrease the risk of cardiovascular effects on
children, the elderly, or patients with preexisting cardiac disease, further dilute the epinephrine to
1:200,000. One must be careful that the epinephrine
solution is not injected intravenously by withdrawing
on the needle and confirming no flashback of blood
prior to subcutaneous injection.
40.4.4. Facial nerve monitoring
It appears that the majority of surgeons use commercially available EMG monitors, such as the Medtronic NIM-Response 2.0, the Inomed Neurosign 400,
or the WR Brackman II (Figs. 13). These monitors
are all equipped with multiple recording channels.
Most surgeons utilize two channels which require
six subdermal electrodes. One ground electrode is
placed either on the forehead or sternal notch, one
negative (anode) electrode is placed on the contralateral shoulder, two active electrodes are placed into
the ipsilateral orbicularis oculi muscle, and two
active electrodes are placed into the ipsilateral orbicularis oris muscle (Fig. 4). The electrodes are often
color-coded to aid in placement. The orbicularis oculi
562
Fig. 3. WR Brackman II (WR Medical Electronics Co.,

Stillwater, MN).
Fig. 1. Medtronic NIM-Response 2.0 (Medtronic Inc.,

Minneapolis, MN).
and oris muscles are generally chosen because of

their importance in eye protection and maintenance
of oral competency. The older EMG models used single electrodes while the newer models may have
paired active electrodes. If single active electrodes
are used, it was recommended that the electrodes be
placed at least 2 cm apart to avoid possible signal cancellation (Terrell et al., 1997; Doikov et al., 2001).
However, the needles should also not be too widely
spaced in order to reduce background noise. The Medtronic device has a feature to check electrode impedances. We recommend that the impedance values be
less than 5 kO in order to assure the integrity of the
electrode wiring as well as to reduce noise from other
electrical devices within the operating room. The
Silverstein monitor uses a mechanical-pressure system
Fig. 2. Inomed Neurosign 400 (Inomed Medizintechnik

GmbH, Teningen, Germany).
and has a strain gauge sensor which is placed in the

ipsilateral corner of the mouth (Fig. 5).
The nerve stimulators are typically insulated flush
with the tip in order to decrease the possibility of current leak through fluid around the nerve. Therefore,
prior to stimulation, any blood or irrigation fluid
should be removed from the operative field. While
the stimulator current can be varied from 0.0 to at
least 10.0 mA, the stimulator should be turned to
the lowest intensity possible to elicit a response.
The initial attempt at stimulation may need to be
set at a higher level, up to 0.5 mA if stimulating
through bone or significant soft tissue, but should
then be titrated to the lowest possible setting.
The use of an assistant to monitor the EMG readings is controversial. While the earlier studies advocate the use of an assistant, more recent studies
typically do not utilize an assistant. In our institution,
Fig. 4. Two active electrodes are placed in the ipsilateral

orbicularis oculi and orbicularis oris muscle.
563
the monitors to either temporarily disable or will create

a significant amount of electrical artifact both audibly
and on the waveform. While the monitors are designed
to minimize and/or mute the effect of cautery, it can
be distracting. Mechanical artifact from two metal
instruments touching each other can also cause an
audible signal. A waveform will not be generated on
the EMG, but the audible signal itself may be difficult
to distinguish from a true stimulus. Thermal irritation
from cold irrigation fluid can also stimulate the facial
nerve so using warm irrigation fluid is beneficial
(Edwards and Kileny, 2005).
40.5.2. Prevention of nerve injury
Fig. 5. The Silverstein monitor (WR Medical Electronics
Co., Stillwater, MN).
the surgeons perform the monitoring and, after regular use, do not find it difficult to distinguish between
true stimuli and artifact.
monitoring
Needle electrodes are placed at the vertex (positive
electrode), the contralateral mastoid (negative electrode), and the forehead (ground). Ideally, the auditory stimulus should be placed within the ear canal.
However, this may be impossible during middle ear
surgery. The speaker can be attached to the microscope lens.
A ball electrode should, ideally, be placed onto the
round window membrane. If the membrane is not
visible, it should be placed into the niche or as close
as possible to the round window. The electrode can
also be placed onto the stapes footplate. As with
ABR, the sound transmitter may be placed either
within the ear canal or onto the microscope.
Almost every author stresses that intraoperative

facial nerve monitor is not a substitute for knowledge
of the anatomy of the course of the facial nerve, technical skill, and close supervision of trainees. The
monitor will not prevent a facial nerve injury. One
should not rely on the monitor to alert to the presence
of a facial nerve at risk for injury. One should always
be cognizant of the risks of facial nerve injury with
thorough preoperative preparation.
40.6. Limitations of auditory monitoring
Artifact is also a problem with ABR monitoring. While
filters can decrease the amount of electrical noise in
the waveform, noise may still distort the wave. The
use of skin electrodes should be avoided as they have
higher impedance and less stability which can lead to
an increase in noise. Actual noise such as from the drill
or even suction can also distort the ABR. Note that the
placement of the speaker is critical. A change in the
distance between the source of the sound and the middle ear will change the ABR. So, if the speaker is not
fixed in position, care should be taken to place it at
the same distance every time.
40.5. Limitations of facial nerve monitoring

40.5.1. Artifact
One of the major reasons cited against use of the facial
nerve monitor is the amount of false-positive alarms
generated by electrical and mechanical artifact. The
use of both monopolar and bipolar cautery will cause
Some surgeons feel that ECochG is more useful for

intraoperative monitoring because it does not require
time to average the responses to multiple auditory stimuli. In addition, the potentials are larger and require
fewer repetitions or less averaging and noise may be
less of a concern (Wazen, 1994). Conversely, ECochG
only indicates the activity of the distal cochlear nerve
564
while an intact ABR demonstrates a complete auditory pathway.

40.7. Conclusion
While both auditory and facial nerve monitoring
accomplish many of the goals stated at the beginning
of the chapter, both types of monitoring are not yet
ideal. Neither type of monitoring is considered
standard of care and, while the use of facial nerve monitoring may be increasing among surgeons, there continues to be considerable resistance toward making it
a standard of care. The difficulty of delivering a constant auditory stimulus to the ear while simultaneously
working through the ear canal must be overcome for
auditory monitoring. The noise and artifact during
both types of monitoring must also be overcome. Also,
facial nerve monitoring requires the practitioner to be
familiar with the device and the different types of audible signals produced in order to use it effectively.
Finally, we concur with the majority of authors that
facial nerve monitoring is no substitute for knowledge
of facial nerve anatomy and good surgical technique.
The use of a facial nerve monitor will not prevent
facial nerve injury just as the use of auditory monitoring will not prevent a poor hearing outcome.
References
Anon, JB, Lipman, SP, Guelcher, RT, Sibly, DA and Thumfart, W (1991) Monitoring the facial nerve during parotidectomy. Arch. Otolaryngol. Head Neck Surg., 117: 1420.
Anon, JB, Lipman, SP, Thumfart, W and Sibly, DA (1994)
Parotidectomy with the nerve integrity monitor II. Eur.
Arch. Otorhinolaryngol., (Suppl.): S383S386.
Anonymous (1996) Facial nerve monitoring: an EAR
Foundation Alumni study. Am. J. Otol., 17: 162164.
Beck, DL and Benecke, JE (1990) Intraoperative facial
nerve monitoring: technical aspects. Otolaryngol. Head
Neck Surg., 102: 270272.
Bendet, E, Rosenberg, SI, Willcox, TO, Gordon, M and
Silverstein, H (1999) Intraoperative facial nerve monitoring:
a comparison between electromyography and mechanicalpressure monitoring techniques. Am. J. Otol., 20: 793799.
Bordley, JE, Ruben, RJ and Leiberman, AT (1964) Human
cochlear potentials. Laryngoscope, 74: 463479.
Brennan, J, Moore, EJ and Shuler, KJ (2001) Prospective
analysis of the efficacy of continuous intraoperative
nerve monitoring during thyroidectomy, parathyroidectomy, and parotidectomy. Otolaryngol. Head Neck
Surg., 124: 537543.
Doikov, IY, Konsulov, SS, Dimov, RS, Deenitchin, GP and
Yovchev, IP (2001) Stimulation electromyography as a

method of intraoperative localization and identification
of the facial nerve during parotidectomy: review of 15
consecutive parotid surgeries. Folia Med., 43: 2326.
Dulguerov, P, Marchal, F and Lehmann, W (1999) Postparotidectomy facial nerve paralysis: possible etiologic
factors and results with routine facial nerve monitoring.
Laryngoscope, 109: 754762.
Edwards, BM and Kileny, PR (2005) Intraoperative neurophysiologic monitoring: indications and techniques for
common procedures in otolaryngology-head and neck
surgery. Otolaryngol. Clin. North Am., 38: 631642.
Filipo, R, Bertoli, GA, De Seta, E, Cordier, A and Barbara,
M (1993) First experiences with intraoperative ECochG
monitoring during stapedotomy. Rev. Laryngol. Otol.
Rhinol., 114: 161163.
Gerull, G, Giesen, M and Mrowinski, D (1978) Brainstem
audiometry during tympanoplasty. Clin. Otolaryngol.,
3: 503510.
Gibson, WPR (1992) Electrocochleography in the diagnosis of perilymphatic fistula: intraoperative observations
and assessment of a new diagnosis office procedure.
Am. J. Otol., 13: 146151.
Greenberg, JS, Manolidis, S, Stewart, MG and Kahn, JB
(2002) Facial nerve monitoring in chronic ear surgery:
US practice patterns. Otolaryngol. Head Neck Surg.,
126: 108114.
Haenggeli, A, Richter, M, Lehmann, W and Dulguerov, P
(1999) A complication of intraoperative facial nerve
monitoring: facial skin burns. Am. J. Otol., 20:
679682.
Jansen, C and Janssen, T (1980) Brainstem audiometry in
tympanoplasty. J. Laryngol. Otol., 95: 391398.
Jones, RO and Mellert, TK (1991) Use of dilute epinephrine as an aid in facial nerve monitoring. Am. J. Otol.,
12: 446449.
Leonetti, JP, Matz, GJ, Smith, PG and Beck, DL (1990)
Facial nerve monitoring in otologic surgery: clinical
indications and intraoperative technique. Ann. Otol.
Lowry, TR, Gal, TJ and Brennan, JA (2005) Patterns of use
of facial nerve monitoring during parotid gland surgery.
Makeieff, M, Venail, F, Cartier, C, Garrel, R, Crampette, L
and Guerrier, B (2005) Continuous facial nerve monitoring
during pleomorphic adenoma recurrence surgery. Laryngoscope, 115: 13101314.
Metson, F, Thornton, A, Nadol, JB and Fee, WE (1988) A
new design for intraoperative facial nerve monitoring.
Noss, RS, Lalwani, AK and Yingling, CD (2001) Facial
nerve monitoring in middle ear and mastoid surgery.
Olds, MJ, Rowan, PT, Isaacson, JE and Silverstein, H (1997)
Facial nerve monitoring among graduates of the Ear
Research Foundation. Am. J. Otol., 18: 507511.
Parsons, RC (1968) Electrical nerve stimulation at surgery.
Pensak, ML, Willging, JP and Keith, RW (1994) Intraoperative facial nerve monitoring in chronic ear surgery: a resident training experience. Am. J. Otol., 15: 108110.
Rea, JL (1990) Use of a hemostat/stimulator probe and
dedicated nerve locator/monitor for parotid surgery.
Ear Nose Throat J., 69: 566573.
Roland, PS and Meyerhoff, WL (1993) Intraoperative
facial nerve monitoring: what is its appropriate role?.
Am. J. Otol., 14: I.
Roland, PS and Meyerhoff, WL (1994) Intraoperative
electrophysiological monitoring of the facial nerve: is
it standard of practice? Am. J. Otolaryngol., 15:
267270.
Ruben, RJ, Knickerbocker, GG, Sekula, J, Nager, GT and
Bordley, JE (1959) Cochlear microphonics in man.
Ruben, RJ, Sekula, J, Bordley, JE, Knickerbocker, GG,
Nager, GT and Fisch, U (1960) Human cochlea
responses to sound stimuli. Ann. Otol. Rhinol. Laryngol., 69: 459479.
Saravanappa, N, Balfour, A and Bowdler, DA (2003) Use
of laser, otoendoscopy and facial nerve monitoring in
otological surgery: United Kingdom survey. J. Laryngol. Otol., 117: 751755.
Silverstein, H and Rosenberg, S (1991) Intraoperative
facial nerve monitoring. Otolaryngol. Clin. North Am.,
24: 709725.
Silverstein, H, Smouha, E and Jones, R (1988a) Routine identification of the facial nerve using electrical stimulation
565
during otological and neurotological surgery. Laryngoscope, 98: 726730.
Silverstein, H, Smouha, EE and Jones, R (1988b) Routine
intraoperative facial nerve monitoring during otologic
surgery. Am. J. Otol., 9: 269275.
Silverstein, H, Rosenberg, SI, Willcox, TO, Jr. and Gordon,
MA (1994) Intraoperative facial nerve monitoring: what
is its appropriate role? [Comment]. Am. J. Otol., 15:
121122.
Terrell, JE, Kileny, PR, Yian, C, Esclamado, RM, Bradford,
CR, Pillsbury, MS and Wolf, GT (1997) Clinical outcome
of continuous facial nerve monitoring during primary
parotidectomy. Arch. Otolaryngol. Head Neck Surg.,
157: 10811087.
Von Scheel, J, Gerull, G, Mrowinski, D and Thoma, J
(1989) Auditory brainstem response monitoring during
middle ear surgery. Ann. Otol. Rhinol. Laryngol., 98:
605610.
Wazen, JJ (1994) Intraoperative monitoring of auditory
function: experimental observations and new applications. Laryngoscope, 104: 446455.
Wazen, JJ, Emerson, R and Foyt, D (1997) Intraoperative electrocochleography in stapedectomy and
ossicular reconstruction. Am. J. Otol., 18: 707713.
Wilson, L, Lin, E and Lalwani, A (2003) Costeffectiveness of intraoperative facial nerve monitoring
in middle ear or mastoid surgery. Laryngoscope, 113:
17361745.
Witt, RL (1998) Facial nerve monitoring in parotid surgery: the standard of care? Otolaryngol. Head Neck
Surg., 119: 468470.

M.R. Nuwer (Ed.)
566
CHAPTER 41
Monitoring during surgery around the acoustic

and vestibular nerves
C. Matthies*
Section of Functional Neurosurgery, Department of Neurosurgery, Julius-Maximilians-University of Wurzburg,
D-97080 Wurzburg, Germany
41.1. Introduction
Surgery of the cerebellopontine angle (CPA) has
been revolutionized by the introduction of neurophysiological monitoring techniques. The foregoing
history of successful surgery of this delicate region
is relatively short: the implementation of the microscope in the beginning of the 1960s and its systematic and practical surgical adaptation by Rand and
Kurze (1968) had brought about a significant reduction of morbiditiy, specifically of hemiparesis and
motor cranial nerve deficits. The unexpected
evidence of functional deficit despite reliable anatomical nerve preservation in numerous cases, however, necessitated an additional tool for functional
intraoperative nerve control, neurophysiological
monitoring (Cohen et al., 1979). After the first enthusiastic reports, especially with intraoperative electromyography recording, a certain drawback from
regular neuromonitoring was to be observed because
of frequent difficulties with artifacts and no chance
of any continuous monitoring at all. In those
operating theaters where general electrophysiological
monitoring worked reasonably well, mostly auditory
brainstem responses (ABR) still posed a problem
with artifacts, deformation, and limited reliability.
In order to record ABR under operating conditions
the most refined technique in electrode placement, in
cable guidance and powerful hardware are essential.
While the standard methods of ABR have been
described, this chapter focuses on its practical application and modification, on techniques for increasing
*
Correspondence to: Cordula Matthies, M.D., Ph.D.,

Section of Functional Neurosurgery, Department of
Neurosurgery, Julius-Maximilians-University of Wurzburg,
Josef-Schneider-Str. 11, D-97080 Wurzburg, Germany.
Tel.: 49-931-201-24808; fax: 49-931-201-24935.
E-mail: matthies.c@nch.uni-wuerzburg.de (C. Matthies).
speed of recordings and, thereby, online information

and for improvement of reliability. Especially this
chapter will introduce the philosophy of functional
microsurgery of the CPA. The subsequent subdivisions are the main topics:
Monitoring of the auditory pathway
Monitoring of the auditory pathway in CPA lesions
Monitoring in schwannoma surgery
Direct monitoring techniques

Identification of cochlear versus vestibular nerves
Monitoring in surgery of CPA meningiomas
Monitoring in surgery of epidermoids
Monitoring in vascular compression syndromes
Examples on CPA monitoring
Functional microsurgery of CPA lesions
References

41.2. Monitoring of the auditory pathway

As a standard, conventional ABR recording is recommended simultaneously for both sides. Even in case
of unilateral complete deafness, bilateral recording
is useful in order to detect activity of crossing parts
of the pathway. In case of bilateral hearing, separate
right- and left-sided acoustic stimulation is performed with bilateral recording; this offers ideal
conditions for side comparison, it facilitates the peak
analysis and the detection of false positive ipsilateral
responses and of brainstem disturbances in very large
brainstem compressive lesions.
Furthermore, multimodality monitoring of the
eighth nerve is to be recommended for all retrocochlear lesions. This includes the conventional recording ipsilateral and contralateral to the lesion side and
near-field potentials such as electrocochleography
and direct nerve recording or brainstem recording
(Mller and Jannetta, 1983, 1984; Yingling and Gardi,
1992; Wazen, 1994; Matthies and Samii, 1997a,b).
567
41.2.1. Conventional ABR recording

Stimulation for conventional ABR recording is
performed by 90100 dB click stimuli to the
involved ear and simultaneous white noise masking
at 6070 dB to the contralateral ear. Stimulation is
delivered at a frequency of 1120 Hz, at a stimulus
duration of 100200 ms. In case of earphones with air
tube conduction tubes, a period of 0.51 ms needs to
be subtracted from the recorded latencies, depending
on the exact tube length.
Recording is performed by monopolar needle
electrodes at A2 and A1 toward the vertex reference
at Cz.
For baseline measurements two recordings are
performed at condensation and two further at rarefaction click stimuli, on the lesion side and on the
healthy side. Hereupon a decision is taken for either
type of stimulation during continuous monitoring,
taking into account a clear formation of the peaks,
especially I, III, and V.
41.2.1.1. Near-field recording techniques
41.2.1.1.1. Electrocochleography. Electrocochleography (ECochG) may be performed by a transtympanal penetrating needle electrode or by a noninvasive
tympanal surface electrode on the involved side.
Noninvasive technique of electrocochleography:
Under visual control by otoscopy a little ball electrode with some electrode contact jelly is placed on
the tympanon. It is important to restrict the size of
the electrode and the amount of surface jelly in order

to contain the flexibility of the tympanon and to
allow the simultaneous insertion of the earphone.
This near-field recording under the same stimulation conditions as above provides a cochlear
microphonic and a large peak I and due to the
increasing distance of the generating anatomic structures smaller peaks II, III, IV, and V (Fig. 1).
Brainstem recording is performed either by a little
ball electrode placed at the lateral recess and covered
with some cotton or by a ball electrode mounted to
the cerebellar retractor. The latter technique may be
performed in a bipolar fashion or with the Cz reference used generally for ABR. The same near-field
electrode may be changed in position in order to be
set closer to or directly on the auditory nerve during
the microsurgical process. Recording produces a nearfield ABR containing components III, IV, and V. It
may also show components I and II in an inverted
mode (Fig. 1).
Multimodality monitoring of auditory function is
enabled by a combination of bilateral conventional
ABR recording, noninvasive ECochG, and near-field
brainstem recording (Fig. 1). Conventional ABR and
ECochG are set up before the start of surgery; brainstem
recording electrodes are placed after dura opening
(Fig. 2).
By bilateral conventional ABR recording ispilateral and contralateral peaks are compared, especially components III to V to confirm the correct
analysis and to decide on the monitoring conditions.
Average: rechts
5
A2/Cz
ABR ipsilateral
C2: 09 ms
ECochG
500nV/Div
A1/Cz
ABR contralateral
Co/Cz
C2
Hi/Cz
Near-filed
Brainstem response
AEP
11
Fig. 1. Multimodality recording.
15
19
568
In case of no components on the lesion side, but very

good hearing function, recording will be tried by
near-field techniques. The identification of conventional components I and II is facilitated by relating
these recordings to ECochG. ECochG latency of
C. MATTHIES
component I is confirmed while the amplitude is

multifold larger.
For the ECochG recording 15 sweeps are sufficient, for the conventional ABR a clear response must
be visible at 2060 sweeps; for the computerized
A near-field electrode is
mounted to the tip of the
retractior
The retractor with the nearfield electrode is in the place

and tumor resection is started.
Clear recording is obtained
(see Fig.)
Intramental nerve stimulation

and recording is performed at
the lateral exposed parts of
VIIth and VIIIth nerves
(Fig. 2 continued)
569
Direct nerve recording at

tumorous nerve parts
The cochlear nerve is

preserved in continuity
Fig. 2. Near-field recording in vestibular schwannoma surgery.
monitoring process averaging is in general set at for

200256 sweeps. Near-field ABR delivers clear
responses at 520 averages. In case of increasing
de-synchronization during surgery a higher number
of measurements might become necessary.
41.3. Monitoring of the auditory pathway in
CPA lesions
Space occupying lesions of the CPA are operated
under multifunctional neurophysiological monitoring.
Approach: Small lesions within the CPA cistern
are monitored by ABR and motor cranial nerve monitoring, larger lesions with brainstem contact or compression need additional neurophysiological control
of the long tracts. Only very small lesions confined
to the internal auditory canal or partially growing
out of the internal auditory meatus are approached
by the middle fossa approach while the majority
of lesions are operated via the suboccipital route.

The suboccipital approaches are the only option for
lesions of any size, if functional nerve preservation
is aimed for. Vestibular schwannomas, meningiomas
of the petrous bone, and tentorium are operated by
the lateral suboccipital approach. Vascular compression lesions and other pathologies are also best
controlled and treated by the lateral suboccipital
approach. Brainstem lesions are mostly best
approached by the midline suboccipital route.
Position: While many centers apply a supine or
park bench position, there are definite technical
microsurgical advantages in the semi-sitting position,
especially for large lesions.
Technical measures: After setup of conventional
and ECochG recording, reliable registration needs
to be checked during and after sterile draping of
the patient in order to detect and counteract
electrode slipping, dislocation, or disconnection.
570
At craniotomy, there should not occur any deterioration, except for extreme fluid effusion at the
mastoid.
schwannoma surgery
For the acoustic/vestibular schwannoma, a standardized protocol for microsurgery accompanied by fast
multimodal monitoring is described.
The subsequent microsurgical steps in schwannoma resection are correlated with ABR:

Baseline (after patient positioning)

Second baseline after craniotomy/craniectomy
Cerebrospinal fluid release
Retraction of cerebellum
Arachnoid dissection and exposure of nerves and
tumor
Bone removal of posterior wall of the internal auditory canal (conservation of labyrinthine structures
Intrameatal tumor dissection
Tumor mobilization at the fundus of the IAC
Volume reduction (by 50%) of space occupying
lesion in the CPA
Stretching of tumor nerve bundle
Direct dissection at the tumor nerve plane
Complete resection of space occupying lesion
Measures for hemostasis
Sealing of the internal auditory canal
Retractor removal
Dura closure
Technical performance: At insertion of the retractor, there is the option for insertion of near-field
recording electrodes; these may be placed at the cerebellopontine junction or at the lateral recess or stay
mounted to the tip of the retractor. Immediately, a
baseline near-field brainstem recording is available.
Wave V may appear slightly earlier and broader than
by conventional recording.
Surgical procedure and related ABR changes:
Vestibular schwannomas of any size and extension
are approached by a craniotomy within the borders of
the transverse and sigmoid sinuses. After dura opening
and cerebrospinal fluid (CSF) drainage from the cisterna magna an amplitude reduction of ABR component II is frequently observed. This will improve
immediately at irrigation with Ringer solution. At
cerebellar retraction deterioration of amplitudes and
latencies of III, IV, and V are observed. If this
C. MATTHIES
continues after retractor fixation, some loosening is

necessary and some early tumor enucleation for the
reduction of brainstem compression.
Thereafter, the internal auditory canal is opened
by a diamond drill. The drilling noise causes a masking of the operated ear, and therefore no reliable
ABR responses can be obtained simultaneously to
any drilling, and short breaks for functional control
of ABR are essential. Besides ABR, ECochG shall
be observed, allowing a very fast feedback and
immediate information on any deterioration. In case
of opening of a semicircular canal a steep amplitude
loss of wave I will occur. This is best to be recognized in the ECochG; mostly this is followed by a
general of all ABR components. If drilling causes
only a fenestration of the bone layer of the semicircular canal, but the lining is left intact, there will be
ABR and ECochG deterioration, but no loss. As the
only remedy, a sealing of the semicircular canal with
microsurgical fibrin plaster can be tried. In case of
hearing preservation there is a significant drop of
quality of hearing to be expected.
If the dorsal wall of the IAC is successfully
opened with stable ABR, intrameatal tumor reduction
and nerve mobilization at the lateral end in the fundus are the next and risky steps. During this period
special attention must be paid to components I and II.
Repeated amplitude reductions by 50% with intermittent recovery are nonfavorable signs; then some longer
period for recovery to the original state is indicated.
If this does not develop, the implantation of a small
fibrin sponge soaked with nimodipine or papaverine
solution under or on the nerve bundle may be helpful.
If the recovery of the cochlear and nerve potentials (I and II) remains scarce, a change to a different
site of action may be wise.
The extrameatal tumor portion is enucleated as
far as possible in order to reduce the compression
to brainstem and nerves. This is accomplished at first
by the resection of the inner tumor parts by the
platelet-shaped knife and by the CUSA. Then the
tumor capsule may be pulled away from the nerves,
and the tumor is dissected off the nerves within the
arachnoid plane. Throughout tumor reduction conservations of waves I and V are the first goals. However,
the sole observation of waves I and V is not sufficient. Before a significant deterioration of I or V will
be detected, other critical changes are to be identified: whenever a certain stretch toward one direction
is exerted, an amplitude decrease and latency deterioration might develop of a single ABR component.
On stretching medially away from the fundus and

toward the brainstem, wave I (and II) may decrease;
on leaving the tissue loose, an immediate recovery
should be observed. On stretching laterally toward
the meatus and away from the brainstem, III may
decrease. On stretching backwards or downwards,
III and V may deteriorate. On pulling upwards very
often a certain improvement of all components is
observed. If these slight deteriorations are overlooked
or not detected because of too rough recording
setting, a seemingly acute drop or loss of V may
occur with no remaining chance to make it reversible. If small initial changes are counteracted, definite component loss can frequently be prevented.
Gross tumor reduction has to aim for component
preservation in order to provide optimal conditions
for the most difficult resection of small tumor remnants at the tumor nerve border. Here, repeated
breaks for component recovery may be necessary.
Nerve recovery may be supported by the application
of some nimodipine or papaverine directly onto the
nerves or on a fibrin sponge.
41.5. Direct recording techniques in schwannoma
surgery
Direct recording techniques do all use near-field
recording conditions and aim for receiving large
potentials in order to avoid the time-consuming process of averaging.
Compared with conventional recording much
larger peaks are registered. By noninvasive electrocochleography, tenfold larger potentials are obtained;
by near-field brainstem recording potentials may be
up to 100-fold larger. Moreover, these responses are
larger than the typical artifacts. If artifacts are
produced in the vicinity of the retractor tip and
electrode, by the cavitational ultrasonic surgical aspirator (CUSA) or bipolar forceps, those artifacts may
also be much larger than usual; but this is relevant
only for limited periods. The variability of waveforms may be larger, especially, if the retractor
position is changed very frequently. Nonetheless,
surgeons and neurophysiologists usually adapt fast
to this monitoring because of the speed of feedback
and the convincingly large amplitudes.
Two surgical periods are especially important for
the application of near-field conditions, the manipulations at the internal auditory canal, that is the drilling and the intrameatal dissection on the one hand,
and the extrameatal dissection at the tumor nerve
571
border at the other hand. These are the most critical

periods where significant and fast potential changes
can occur, which should not be disguised during
any averaging process. Those who are not yet familiar with near-field recording might use it during these
phases and simultaneously to the conventional
recording modes.
Opening of the internal auditory canal: This operative phase should be controlled by a special focus on
wave I, ideally by electrocochleography and ABR.
The drilling with the diamond drill needs to be interrupted at least every 15 s as the drilling provokes
a complete masking of the ear and abolishes any
reliable recording. Only during the breaks a fast
recording can be performed. Drilling is dangerous
by several factors, the exerted heat, the vibration,
the noise, and with regard to the integrity of the
labyrinthine structures. Continuous irrigation, sharp
drills and frequent breaks, and a thoughtful planning
of the opening zone are the measures at hand.
During the intrameatal tumor dissection, cochlear
function remains at high risk. Therefore, all manipulations have to aim for minimal stretch away from
cochlear nerve origin; any amplitude reduction of I
has to be followed by an immediate break and a
change of action. A sudden loss of wave I is most
likely followed by deafness, a serious amplitude
reduction indicates a significant hearing impairment
or even hearing loss.
41.5.1. Final tumor dissection at the tumor
nerve border
The final tumor dissection may be very critical for
the ABR, especially at the brainstem junction and
medial to the porus. Specifically, any traction backwards and downwards needs online observation of
ABR changes. Waves III, IV, and V are very sensitive and significant amplitude reductions and latency
increases have to be avoided or counteracted before
wave disappearance. Before those changes are visible
by conventional ABR, direct brainstem recording
will provide faster and more reliable information.
Furthermore, near-field recording techniques are
sensible tool and prerequisite for centers performing
auditory implants for hearing rehabilitation. Colletti
et al. (2000) describe their setup in threatened
hearing in the last hearing ear. They outline the possibility of partial tumor resection under monitoring
guidance, or in case of hearing loss the electrophysiological control whether a cochlea implant or
572
an auditory brainstem implant is the therapy of

choice. If direct cochlear nerve potential is preserved,
but ABR are lost, a CI is performed; if ABR and compound nerve action potential (CNAP) are lost, brainstem implant is indicated. Intraoperative electrical
testing of the cochlear nucleus is performed and, if
electric ABR can be obtained, the ABI placement can
be adjusted and the likelihood of some postoperative
hearing perception may be predicted (Colletti et al.,
2000; Frohne et al., 2000; Matthies et al., 2000).
41.6. Identification of cochlear versus
vestibular nerves
In most instances electrophysiological differentiation
between cochlear and vestibular nerves is not necessary, as the anatomical location at the dorsal part of
the tumor and the IAC is a clear indication of the
vestibular fibers. If only one vestibular nerve forms
the tumor origin or if the cochlear or facial nerves
are tumorous, the situation is different and direct
recording with a bipolar electrode can be helpful
(Debatisse et al., 2005; Ferber-Viart et al., 1997,
1999; Nguyen et al., 1999).
Differentiation of cochlear versus vestibular
nerve fibers: the direct recording technique is
applied by a bipolar forceps electrode (Fig. 2) which
is placed lightly onto the fibers in question. Auditory
stimulation is continued as for the ABR. The registration is compared with the conventional or with
the other near-field responses. Large I and II components like in the ECochG speak for auditory fibers.
Further controls are possible by testing vestibulo
muscular reflexes such as with the sternocleidomastoid muscle.
meningioma surgery
In planning meningioma surgery in the CPA two
aspects are relevant to diagnose and to analyze: the
site of origin and the type of tumor biology. The predominant site of tumor origin may be medial anterior
or dorsal lateral to the cranial nerves, a factor of
major importance for the difficulty of microsurgical
resection. In tumor resection dorsal and lateral to
the cranial nerves the surgeon has a wider space for
microsurgical actions, but has no direct sight onto
most of the nerves until the final part of tumor
removal. In tumor resection anterior to the nerves,
the space for microsurgical activity is limited as it
has to be performed between the nerves, far anterior
C. MATTHIES
with a higher risk of direct nerve irritation, but good

visibility of the nerves.
Depending on the type of biology, circumscribed
or infiltrative meningioma, the goal of surgery must
be defined beforehand. In well-defined meningioma,
complete tumor resection with functional nerve preservation and sometimes even functional hearing
improvement is possible. In case of en plaque tumor
extension with dural tail formation, intrameatal infiltration and/or osseous involvement of the petrous
bone only partial tumor resection might allow functional acoustic preservation. Under these circumstances, long-term preservation is often impossible.
The variety of tumor origins and related extensions necessitates considerable surgical flexibility.
Nonetheless, the subsequent microsurgical steps
may be standardized (in variable sequence) and correlated with ABR:


Cerebrospinal fluid release from cisterna magna
tumor
Volume reduction (by 50%) of space occupying
lesion in the CPA
Stretching of tumor nerve bundle
Direct dissection at the tumor nerve plane
Complete resection of space occupying lesion
And, in exceptional cases:
Bone removal of posterior wall of the internal audi

tory canal (conservation of labyrinthine structures)

Intrameatal tumor dissection
Tumor mobilization at the fundus of the IAC
Sealing of the internal auditory canal
Retractor removal
Dura closure
ABR monitoring in meningioma surgery may

show very sensitive or very stable potentials. If critical vascularization or preexisting overstretching of
the nerves are important in a case, ABR may be very
variable and unstable. If this aspect is neglected and
deterioration is not quickly counteracted, ABR loss
and hearing loss may be the result. It has been shown
quite clearly that the intermittent deterioration of
ABR with following recovery may nonetheless
result in postoperative deafness, depending on
the duration of these events during surgery (Nakamura et al., 2005). Therefore, prolonged critical
amplitude reduction of III and V must be prevented.
If ABR amplitudes and latencies are stable and
preserved throughout surgery, there is a good chance
of retained or improved postoperative hearing function. This is a fundamental difference to schwannoma
surgery where even preserved ABR is usually followed by a slight hearing deterioration. Therefore,
the trial of ABR preservation is worthwhile in meningiomas. If it is of an infiltrative type, surgical radicality should not be attempted because this will bring
about multiple cranial nerve lesions; only decompression of neural structures should be attempted. Bony
decompression of the internal auditory meatus and
partial tumor dissection at the intrameatal part are
to be recommended. Further dissection will be hazardous to the vascular nerve supply and cause early
or delayed nerve lesion. Near-field recording is very
helpful in this matter.
epidermoid and other tumor lesions
Monitoring and preserving auditory function in epidermoids of the CPA are realistic goals. From the
clinical point of view, epidermoid cysts may develop
to a considerable size and exert serious brainstem
compression without any focal symptoms. Rather
unspecific symptoms like dizziness and headaches
may occur; late symptoms relate to the VIIth and
VIIIth nerve bundle, the oculomotoric nerves and trigeminal nerves. Even in normal preoperative auditory function, there is a considerable risk of hearing
loss at surgery; as small parts of the epidermoid
may extend into the tiniest spaces, posterior and anterior to the nerves, a considerable amount of manipulation and phases of enhanced retraction may put the
auditory fibers at risk.
From the biological point of view, there are two
types of epidermoid behaving very differently at surgery. The most common type is of silver white
appearance and loose consistence and is easily separated into pieces and removed. Under these conditions, there may be very stable ABR monitoring as
there is very little direct dissection at the nerves necessary; then ABR preservation and hearing preservation are most likely. Also hearing improvement,
especially improvement of speech discrimination
may develop. The only risky aspect is the need for
strong retraction in order to remove small tumor
573
remnants. Retraction may be followed from very

abrupt amplitude loss of waves III and V and needs
immediate reaction by loosening the retractor and
by keeping retraction periods short.
The second less frequent epidermoid type is of
darker, more yellow color and lined by connective
tissue with vessels that show a tight junction with
inherent nerves and their vessels. The resection of
these epidermoids is necessary by sharp dissection
with a higher risk of vascular compromise, of bleeding, and of loss of function. Here, surgery has to be
performed with a lot of patience and time, frequent
retractor release, intermittent irrigation, and stepwise
dissection.
There is a variety of other CPA tumors such as lipomas, hamartomas, metastasis, especially from melanoma, vascular malformations like hamartomas and
AV lesions. Regarding metastasis, surgery should be
limited to biopsy or partial resection for brainstem
decompression and should be followed by adjuvant/
systemic therapy. Even the finest microsurgical dissection bears an extremely high risk of functional
nerve compromise despite nerve preservation. Monitoring has to support the surgeon during the biopsy in
order to minimize any adverse effects.
In rare lesions such as lipomas and hamartomas
their clinical and radiological image might imitate a
vestibular schwannoma. However, complete resection is impossible without functional nerve loss,
and, moreover, it is not necessary in view of the
nonproliferative nature. Surgery should be limited
to exploration and histological testing (Carvalho
et al., 2003).
41.9. Microsurgical decompression in vascular
compression syndrome
Vascular compression syndromes like trigeminal
neuralgia and hemifacial spasm are treated by the
identification of the critical vessel and by the decompression of the sensitive nerve part from it. For
approaching the nerve entry or exit zone at the brainstem considerable retraction may be necessary
exposing the nerves to stretch and leading to some
deterioration of conductivity. Furthermore, dissection
between the cranial nerves exerts some stress on the
vascular nerve supply. Therefore, cranial nerve monitoring is indispensable. The risk of deafness after
facial nerve decompression is reported at 5% and
after trigeminal nerve decompression somewhat
lower.
574
The subsequent microsurgical steps in vascular

decompression are correlated with ABR:


Cerebrospinal fluid release from cisterna magna
vessels
Dissection and exposure of the irritated nerve at its
brainstem junction
Mobilization of the compressive vessel
Insertion of Teflon piece between nerve zone and
vessel
Retractor removal
Dura closure
While the microsurgical steps have a lot in common with tumor resections, there are fundamental differences in ABR monitoring for tumor surgery and
vascular surgery. The important rate of hearing loss
and an unknown proportion of hearing deterioration
have in part to be attributed to the lack of a predisposing nerve stretch: in tumors a chronic stretch of the
related nerves has developed over the years; at the
start of surgery slight retraction is not immediately
reflected on the auditory nerve, except for very large
lesions. In non-tumorous lesions, minimal manipulations are followed by much more significant ABR
changes. Therefore, at the first feedback reports of
some ABR changes there is already a realistic risk of
permanent loss. Various measures are at hand to counteract fast ABR deterioration: The complete resection
of the arachnoid around the nerves is essential before
retraction and before decompression. Hereby, all the
structures become much looser and further preparation
much easier as the cerebellum falls away without
retraction. If vasospasm comes up, it will be recognized by amplitude drop of wave I, and this is best
treated by the application of local vasodilators. Some
more retraction is necessary for complete exposure
of the compressive vessels at the brainstem; this has
to be exerted slowly. Before implantation of any material for vascular decompression (Teflon), ABR must
be stable as the implantation may cause some further
deterioration. If implantation is followed by slight
amplitude reduction, this is observed carefully under
irrigation. If, in rare cases, the deterioration proceeds,
implant material might be removed and a smaller
piece has to be used. After Teflon implantation a
C. MATTHIES
further observation of at least 5 min is indicated,

before dural closure can be performed.
41.10. Examples on monitoring in CPA surgery
Multimodality control of the auditory pathway: In a
29-year-old female with a large brainstem compressive schwannoma hearing preservation is not the
primary goal, but tumor removal for brainstem
decompression and facial nerve conservation. The
initial ABR are extremely delayed and of reduced
amplitudes. Electrocochleography and direct brainstem recording provide much larger and clearer
responses (Figs. 1 and 2). Monitoring of the auditory
pathway, ipsilateral, and contralateral is mainly
performed for the safety of the brainstem, while a
realistic chance of hearing conservation is not
expected. In view of the reduced and delayed conventional responses, near-field potentials are
expected to give larger and faster information
(Fig. 3). The direct brainstem recording shows all
components clearly visible during the microsurgical
process in the CPA. ABR is preserved up to the end
of resection.
Conventional ABR monitoring with ABR loss and
delayed surgical reaction (Fig. 4): A 46-year-old
male is operated for a medium-sized CPA tumor.
Baseline ABR are preserved at the start of surgery.
During CSF release and retraction there is some
temporary amplitude reduction of waves IV and V.
Recovery is awaited and drilling is performed without any ABR problems. On tumor dissection there
is strong amplitude reduction of III, IV, and V. The
surgeon is warned and asked to wait, but he continues with dissection and stretching the tumor nerve
bundle in various directions. This is followed by a
serious amplitude reduction of I and V of over
50%. Still the surgeon does not perform any break.
He mobilizes the tumor by pulling it backwards and
downwards that causes complete disappearance of
V and shortly later also of I. When the surgeon now
stops his action, this is too late, there is no relevant
recovery to be observed. The patient has no useful
hearing after surgery. This example clearly demonstrates the necessity of immediate response to critical
neurophysiological information.
Conventional ABR monitoring with deterioration
(Fig. 5): A 33-year-old male presents with intermittent tinnitus due to a vestibular schwannoma on the
left with large intra- and extrameatal extension.
575
Zeit
III
I
[ms]
I
[V]
III
[ms]
III
[V]
V
[ms]
1.87
0.000
7.53
0.000
9.53
1.93
0.000
7.47
0.000
1.00
1.93
0.000
7.47
0.000
9.47
1.93
0.000
7.47
0.000
9.47
1.93
0.000
7.47
0.000
9.47
1.93
0.000
7.47
0.000
9.47
1.93
0.000
7.53
0.000
9.93
2.07
0.000
7.53
0.000
9.07
2.07
0.000
7.53
0.000
1.00
2.07
0.000
7.53
0.000
9.60
1.93
0.000
7.53
0.000
9.60
Retractor
electrode
11:10:07
III
Tumor
reduction
11:11:08
III
III
III
V
11:12:08
V
11:13:09
3 V/Div
11:14:09
III
V
11:15:10
V
III
11:16:10
III
V
11:17:11
III
III
11:18:11
11:19:12
V
III
Facial
and cochlear
Nerve
preservation
11:20:12
1
11
Latenz [ms]:
17
23
29.00
29
11
Tabelle:
Abs. Werte
Fig. 3. Large vestibular schwannoma, 29-year-old female.
Maneuvers
Monitoring
Warning
Start
CSF, Retractor, Amplitude
Drilling
Dissection, stretch
Stop!
Stretch down, amplitudes
Break!
Stretch down, no V
Stop!
Stretching down, no I
Stop!
Stop, too late, no recovery
200 nV
l
0
l
10
ms
Fig. 4. Auditory brainstem responses (ABR) monitoring in schwannoma resection complete ABR loss and hearing loss.
576
C. MATTHIES
ABR left
Baseline
Delayed ABR
CSF drainage
IAC drilling,
artifacts
#0 95
8:30
#1 95
8:31
#2 95
8:32
#3 95
8:33
#4 95
8:35
#5 95
8:36
#6 95
8:37
#7 95
8:37
#8 95
10:46
#9 95
10:47
#10 95
10:48
#11 95
10:49
#12 95
10:50
#13 95
10:51
#14 95
10:52
#15 95
10:54
#16 95
10:55
#17 95
10:56
#18 95
10:57
#19 95
10:58
#20 95
10:59
#21 95
11:00
#22 95
11:01
#23 95
11:02
#24 95
11:03
#25 95
11:04
#26 95
11:04
#27 95
11:05
#28 95
11:06
#29 95
11:07
#30 95
11:08
#31 95
11:09
#32 95
11:10
#33 95
11:11
#34 95
11:11
#35 95
11:11
#36 95
11:15
#37 95
11:19
#38 95
11:20
#39 95
11:20
183 V
10
12
14
ms
A
(Fig. 5 continued)
577
ABR left
Further drilling
Intermittent breaks
Deterioration
Intrameatal
Dissection
Splitting I and II
#40 95
11:21
#41 95
11:21
#42 95
11:22
#43 95
11:22
#44 95
11:23
#45 95
11:23
#46 95
11:24
#47 95
11:26
#48 95
11:26
#49 95
11:27
#50 95
11:27
#51 95
11:27
#52 95
11:28
#53 95
11:28
#54 95
11:29
#55 95
11:30
#56 95
11:30
#57 95
11:31
#58 95
11:31
#59 95
11:32
#60 95
11:32
#61 95
11:33
#62 95
11:33
#63 95
11:34
#64 95
11:35
#65 95
11:35
#66 95
11:36
#67 95
11:36
#68 95
11:38
#69 95
11:39
#70 95
11:39
#71 95
11:40
#72 95
11:40
#73 95
11:41
#74 95
11:41
#75 95
11:42
#76 95
11:42
#77 95
11:42
#78 95
11:43
#79 95
11:43
183 V
10
12
14
ms
(Fig. 5 continued)
578
C. MATTHIES
ABR left
Intrameatal dissection,
I and splitting
End of IAC dissection
I? II? V+
Tumor dissection CPA
Tumor enucleation
V preserved,
other components lost?
Desynchronization
I recovery?
#80 95
11:44
#81 95
11:45
#82 95
11:45
#83 95
11:46
#84 95
11:46
#85 95
11:47
#86 95
11:47
#87 95
11:48
#88 95
11:48
#89 95
11:49
#90 95
11:49
#91 95
11:50
#92 95
11:51
#93 95
11:51
#94 95
11:52
#95 95
11:53
#96 95
11:54
#97 95
11:55
#98 95
11:56
#99 95
11:56
#100 95
11:56
#101 95
11:57
#102 95
11:57
#103 95
11:58
#104 95
11:58
#105 95
11:59
#106 95
12:00
#107 95
12:00
#108 95
12:01
#109 95
12:02
#110 95
12:03
#111 95
12:03
#112 95
12:04
#113 95
12:05
#114 95
12:05
#115 95
12:06
#116 95
12:07
#117 95
12:08
#118 95
12:08
#119 95
12:09
183 V
10
12
14 ms
C
(Fig. 5 continued)
579
ABR left
I difficult to reproduce
No I
Retractor loosened
Nimodipine
I recovery?
Small variable I
Small V
I and V at small
amplitudes
#120 95
12:10
#121 95
12:10
#122 95
12:11
#123 95
12:12
#124 95
12:12
#125 95
12:13
#126 95
12:13
#127 95
12:14
#128 95
12:15
#129 95
12:15
#130 95
12:16
#131 95
12:16
#132 95
12:17
#133 95
12:17
#134 95
12:19
#135 95
12:19
#136 95
12:19
#137 95
12:19
#138 95
12:20
#139 95
12:20
#140 95
12:21
#141 95
12:21
#142 95
12:23
#143 95
12:23
#144 95
12:25
#145 95
12:26
#146 95
12:26
#147 95
12:27
#148 95
12:27
#149 95
12:28
#150 95
12:38
#151 95
12:33
#152 95
12:34
#153 95
12:35
#154 95
12:36
#155 95
12:36
#156 95
12:37
#157 95
12:38
#158 95
12:39
#159 95
12:40
183 V
10
12
14 ms
D
(Fig. 5 continued)
580
C. MATTHIES
ABR left
Tumor resection
complete
V preserved,
I unstable
#160 95
12.41
#161 95
12.43
#162 95
12.44
#163 95
12.44
183 V
10
12
14 ms
E
Fig. 5. Auditory brainstem responses (ABR) monitoring in versus resection, 33-year-old male ABR left.
ABR are preserved, but extremely delayed at the

start of surgery with wave III above 5 ms and V above
7 ms. After drilling of the IAC and during intrameatal
tumor resection amplitude reduction and splitting of
components I and II occur. Therefore, surgical action
is transferred to another site: At the beginning of extrameatal tumor reduction V is preserved, but I and II are
uncertain with variable latencies. The reproducibility
of I becomes worse during further tumor resection.
Intermittent breaks and nimodipine application via a
fibrin sponge placed underneath the nerve lead to some
V improvement. At final complete tumor resection V
is well preserved and reproducible, but I and III
are uncertain. The patient is receiving continuous
intravenous administration of plasma expander and
nimodipine, but postoperative control discloses
hearing loss at 90 dB.
This case illustrates an unsolved problem of limited reliability of ABR, only if one considers the surgery end point and argues that ABR component V
was present. It shows that ABR are undergoing a
dynamic process during surgery which is not at an
end point at the end of surgery. Furthermore, intermittent complete loss of some ABR components is
a critical sign that may indicate despite some
581
recovery for some time the development of final

hearing loss. In those cases, hearing may disappear
between hours to 2 weeks after surgery.
Case report on ABR preservation (Fig. 6): At the
start of the tumor resection ABR is well preserved.
During drilling there is some amplitude decrease of
I and V, at the break recovery is slow and incomplete. When drilling is finished, all ABR components
are small, but reproducible. At intrameatal dissection
with the micro-hook a serious decrease of II and III
occurs. After a break I and V show recovery, but on
dissection with stretch away from the brainstem, there
is again a serious amplitude drop of V which is soon
followed by the deterioration of I. Tumor resection
is finished and closure of the IAC is performed by a
piece of muscle. This causes a loss of V. On removal
of the muscle there is fast recovery of V; a smaller
piece of muscle is then implanted for the prevention
of CSF fistula. ABR remains preserved and correlates
with normal postoperative hearing function. Normal
auditory function in this patient is retained for meanwhile over 6 years.
ABR monitoring in epidermoid surgery (Figs. 7
and 8): A 10-year-old boy with an epidermoid of
the CPA suffers from intermittent severe headaches.
Fig. 6. Auditory brainstem responses (ABR) monitoring with hearing preservation.
582
In the supine position the left cerebellopontine angle is

exposed. A part of the epidermoid has been removed
between the trigeminal nerve and the facial and
vestibulcochlear nerve bundle. A major part of the
epidermoid is still situated anterior to the cranial nerves.
C. MATTHIES
is resected in a piecemeal fashion. Pieces are

removed between V and VII/VIII. At removal attention is to be paid not to have pieces drifted away to
the subarachnoid spaces. Touching and manipulation
anterior to the VIIIth nerve and at its entry zone
causes slight deterioration in amplitudes and latencies. Finally, the epidermoid is completely resected
with the preservation of nerves and vessels. All
ABR components are present at good amplitudes,
still at some increased latency.
ABR monitoring in microvascular decompression
of trigeminal neuralgia (Figs. 9 and 10): At exposure
of the CPA in the supine position V, VII, VIII, and
caudal nerves are covered by arachnoid which is
stepwise resected (Fig. 9). Fig. 10: ABR monitoring
is performed at 90 dB rarefaction click stimuli. At
CSF release and dissection latency I increases and
amplitude II decreases. The arachnoid is dissected
to loosen the nerves from each other. Only thereafter,
the cerebellum is more retracted. V latency increases
and components II and III decrease in amplitude. At
the insertion of a piece of Teflon for vascular decompression all components decrease, but after placement and retractor removal the whole ABR returns
to its original shape. It remains stable the following
minutes, so dura closure is performed. Immediately
after surgery the patient is pain free and all cranial
nerves are intact.
After complete resection of the epidermoid the brainstem,

its vessels and the cranials nerves are well preserved.
41.11. Functional microsurgery of the CPA lesions
Fig. 7. Epidermoid resection left cerebellopontine angle

(CPA) in a 10-year-old boy.
During the first phase of auditory pathway monitoring in skull base procedures, ABR recording and
intraoperative evaluation were focused on the presence and on the quality of component V. To date,
these criteria hold true with regard to the reliability
on postoperative hearing or deafness: If wave V is
lost, there is about 90% probability of complete
hearing loss (Matthies and Samii, 1997b); exceptions
may be identified for example in case of immense
fluid effusion of the mastoid and the middle ear during surgery. On the other hand, it could be shown
that a short-term temporary loss of V with the final
presence of V at the end of surgery was followed
by deafness in only 14% of cases. Both aspects, however, are the major causes for the concept of focusing
on other aspects in the goal of hearing preservation.
It took considerable time to prove that first of all
each component is of considerable importance and
that it should always be related to the ongoing
Clinical investigation and neurophysiological tests

give proof of normal cranial nerve function. Because
of the increasing intensity of pain and the brainstem
compressive aspect of the epidermoid, surgery is
agreed on. In the supine position, with the head rotated
to the right side and slightly elevated, a retrosigmoid
craniotomy is carried out. Monitoring consists of
median nerve SEP, ABR, facial and trigeminal EMG,
and MEP. At exposure of the left CPA nerves V, VII,
VIII, and caudal nerves are explored by the resection
of the arachnoid (Fig. 7). Components I and II
become later (Fig. 8), amplitude II is reduced. The
epidermoid extends anteromedial to the cranial
nerves from the tentorium to the caudal nerves,
which are significantly dislocated. The epidermoid
583
Baseline
CSF drainage
retractor
183 V
10
12
14 ms
A
(Fig. 8 continued)
584
C. MATTHIES
Dissection between
trigeminal and facial
nerves
Warning: V
Warning: I
- break
Recovery of I
Dissection at
Cochlear entry zone
Warning: I ,III
Break
183 V
Recovery
0
10
12
14 ms
B
(Fig. 8 continued)
585
Complete resection
Retractor removal
ABR preserved
183 V
10
12
14 ms
C
Fig. 8. Auditory brainstem responses (ABR) monitoring in epidermoid resection L, 10-year-old male.
586
C. MATTHIES
chance of recovery, if the surgeon reacts fast to critical changes, such as 30% and 50% amplitude reduction or 0.5 ms latency increase. These changes occur
long before a critical change or loss of V, except for
extreme microsurgical maneuvers with severe brainstem compression or cerebellar retraction. In conclusion, wave V may be used for the prediction of
postoperative function, the foregoing components I,
II, and III should be focused on during the monitoring process.
Regarding the type of ABR changes, primary
focus must be put on amplitude changes. Experimental studies correlating ABR changes, recovery and
immunohistochemical investigations demonstrate a
very reduced reliability of normalized ABR latencies
(Sekiya et al., 2002).
As mentioned above, monitoring I, II, and III precisely puts a higher challenge to the monitoring
setup. Here, also near-field recording can help tremendously (see before and below).
For increasing speed and sensitivity of ABR several centers tried out near-field and direct recording
techniques with various types of electrodes (Mller
and Jannetta, 1983, 1984; Mller et al., 1994; Cueva
Fig. 9. At exposure of the cerebellopontine angle (CPA) in

the supine position V, VII, VIII, and caudal nerves are covered by arachnoid which is stepwise resected.
microsurgical action (Matthies and Samii, 1997b;

Legatt, 2002; James and Husain, 2005; Nakamura
et al., 2005). Relevant aspects are those changes with
a higher chance of reversibility: especially waves II
and III are most sensitive to manipulation along the
nerve up to the brainstem entry, but they have a good
#0.95
11:51
ABR left
#1.95
11:52
#2.95
11:53
#3.95
12:05
58 V
10
12
14 ms
A
(Fig. 10 continued)
ABR right, ipsilateral
Arachnoid dissection,
vessel inspection
#0 95
11:41
Critical vessel identified
#1 95
11:42
Start
Retractor,
CSF release,
IV and V smaller
Recovery
Teflon implantation
warning: II amplitude
V latency
#2 95
11:43
#3 95
11:44
Recovery
#4 95
11:45
Manipulation to improve
Teflon position
Recovery
#6 95
11:47
Recovery
#7 95
11:48
Further arachnoid
Dissection, exposing
cranial nerves
#8 95
11:49
#32 95
14:44
#33 95
14:47
#34 95
14:50
#35 95
14:52
#36 95
14:55
#5 95
11:46
Arachnoid dissection,
Retraction,
III, IV, V smaller
#31 95
14:42
End of surgery,
ABR preserved
#37 95
14:57
#38 95
15:00
#39 95
15:03
#40 95
15:05
58 V
#9 95
12:07
58 V
0
0
10
12
10
12
14 ms
14 ms
B
587
Fig. 10. Trigeminal neuralgia R (60 g, male) auditory brainstem responses (ABR) left healthy side.
588
et al., 1998; Matthies and Samii, 1997a). Furthermore, direct nerve recording gives the opportunity
of differentiating between the cochlear and other
nerve fibers (Roberson et al., 1996; Nguyen et al.,
1999). To date, it is a minority that applies such techniques regularly while most would like to try them
and clearly list up all the advantages involved
(Schmerber et al., 2004).
Mller and Jannetta (1983, 1984) were among the
very first to apply and report direct recording techniques of the auditory nerve and to underline the
advantage of online information. Colletti and Fiorino
(1998) also advocate the use of immediate feedback
monitoring: The fundamental prerequisite for
obtaining optimal benefits from monitoring is the
use of techniques of direct and continuous electrophysiologic recording with instantaneous feedback
to the surgeon, such as CNAPs . . . .
Direct monitoring of the auditory pathway can be
performed close to the generators of components I,
II, and III, namely cochlea, auditory nerve, and
auditory nucleus. As the microsurgical activity is
centered between the cochlea and the nucleus, it is
logical to record their activity. Direct recording
techniques give larger and faster responses and may
thereby be more helpful in preventing critical maneuvers (Wazen, 1994; Matthies et al., 2000; Yingling
and Gardi, 1992).
Today all centers performing regular CPA surgery
and aiming for functional nerve preservation, apply
monitoring of the auditory pathway. The overall
preservation rate independent of any tumor size and
hearing quality varies between 40% and 50% in
schwannomas and is higher in all the other pathologies. Besides the preservation rate the quality of the
preserved hearing is of increasing interest. If refined
neuromonitoring is performed successfully, not only
rates of hearing preservation are improved, but there
is a chance for improving the quality of functional
outcome (Matthies and Samii, 2002). Regarding the
quality of the preserved hearing, up to one third of
cases of small schwannomas retain their preoperative
hearing level, and around 20% in medium-sized
schwannomas that are filling out the CPA cistern.
This opens a new dimension to CPA surgery and
documents that the neuromonitoring has to be used
by the surgeon just with the same ease and expertise
as the micro-instruments and the microscope.
C. MATTHIES
References
Carvalho, GA, Matthies, C, Osorio, E and Samii, M (2003)
Hamartomas of the internal auditory canal: report of
two cases. Neurosurgery, 52(4): 944948; discussion
948949.
Cohen, NL (1979) Acoustic neuroma surgery with emphasis
on preservation of hearing. Laryngoscope, 89(6 Pt 1):
886896.
Colletti, V and Fiorino, FG (1998) Advances in monitoring
of seventh and eighth cranial nerve function during posterior fossa surgery. Am. J. Otol., 19(4): 503512.
Colletti, V, Fiorino, FG, Carner, M, Giarbini, N, Sacchetto,
L and Cumer, G (2000) The retrosigmoid approach for
auditory brainstem implantation. Am. J. Otol., 21(6):
826836.
Cueva, RA, Morris, GF and Prioleau, GR (1998) Direct
cochlear nerve monitoring: first report on a new atraumatic, self-retaining electrode. Am. J. Otol., 19(2):
202207.
Debatisse, D, Pralong, E, Guerit, JM and Bisdorff, A
(2005) Recording click-evoked myogenic potentials
(CEMPs) with a setup for brainstem auditory evoked
potentials (BAEPs). Neurophysiol. Clin., 35(4):
109117.
Ferber-Viart, C, Duclaux, R, Colleaux, B and Dubreuil, C
(1997) Myogenic vestibular-evoked potentials in normal subjects: a comparison between responses obtained
from sternomastoid and trapezius muscles. Acta Otolaryngol., 117(4): 472481.
Ferber-Viart, C, Dubreuil, C and Duclaux, R (1999) Vestibular evoked myogenic potentials in humans: a
review. Acta Otolaryngol., 119(1): 615.
Frohne, C, Matthies, C, Lesinski-Schiedat, A, Illg, A, Rost,
U, Battmer, RD, Samii, M and Lenarz, Th (2000)
Auditory brainstem implant in the rehabilitation of
patients with neurofibromatosis II. J. Laryngol. Otol.,
114(Suppl. 27): 1114.
James, ML and Husain, AM (2005) Brainstem auditory
evoked potential monitoring: when is change in wave
V significant? Neurology, 22; 65(10): 15511555.
Legatt, AD (2002) Mechanisms of intraoperative brainstem
auditory evoked potential changes. J. Clin. Neurophysiol., 19(5): 396408.
Matthies, C and Samii, M (1997a) Direct brainstem recording of auditory evoked potentials during vestibular
schwannoma resection: nuclear BAEP recording. Technical note and preliminary results. J. Neurosurg., 86(6):
10571062.
Matthies, C and Samii, M (1997b) Management of vestibular schwannomas: the value of neurophysiology for
intraoperative monitoring of auditory function in 200
cases. Neurosurgery, 40(3): 459468.
Matthies, C and Samii, M (2002) Vestibular schwannomas
and auditory function: options in large T3 and T4
tumors? Neurochirurgie, 48(6): 461470.
Matthies, C, Thomas, S, Moshrefi, M, Lesinski-Schiedat,
A, Frohne, C, Battmer, RD, Lenarz, Th and Samii, M
(2000) Auditory brainstem implants: current neurosurgical experiences and perspective. J. Laryngol. Otol., 114
(Suppl. 27): 3236.
Mller, AR and Jannetta, PJ (1983) Monitoring auditory
functions during cranial nerve microvascular decompression operations by direct recording from the eighth
nerve. J. Neurosurg., 59(3): 493499.
Mller, AR and Jannetta, PJ (1984) Monitoring auditory nerve
potentials during operations in the cerebellopontine angle.
Otolaryngol. Head Neck Surg., 92(4): 434439.
Mller, AR, Colletti, V and Fiorino, FG (1994) Clickevoked responses from the exposed intracranial portion
of the eighth nerve during vestibular nerve section:
bipolar and monopolar recordings. Electroencephalogr.
Nakamura, M, Roser, F, Dormiani, M, Samii, M and
Matthies, C (2005) Intraoperative auditory brainstem
responses in patients with cerebellopontine angle
meningiomas involving the inner auditory canal: analysis of the predictive value of the responses. J. Neurosurg., 102(4): 637642.
Nguyen, BH, Javel, E and Levine, SC (1999) Physiologic
identification of eighth nerve subdivisions: direct
589
recordings with bipolar and monopolar electrodes. Am.
J. Otol., 20(4): 522534.
Rand, R and Kurze, T (1968) Case reports and technical
notes: preservation of vestibular, cochlear and facial
nerves during microsurgical removal of acoustic
tumours. J. Neurosurg., 28: 158.
Roberson, J, Senne, A, Brackmann, D, Hitselberger, WE
and Saunders, J (1996) Direct cochlear nerve action
potentials as an aid to hearing preservation in middle
fossa acoustic neuroma resection. Am. J. Otol., 17(4):
653657.
Schmerber, S, Lavieille, JP, Dumas, G and Herve, T (2004)
Intraoperative auditory monitoring in vestibular
schwannoma surgery: new trends. Acta Otolaryngol.,
124(1): 5361.
Sekiya, T, Shimamura, N, Yagihashi, A and Suzuki, S
(2002) Axonal injury in auditory nerve observed in
reversible latency changes of brainstem auditory
evoked potentials (BAEP) during cerebellopontine
angle manipulations in rats. Hear Res., 173(12):
9199.
Yingling, CD and Gardi, JN (1992) Intraoperative monitoring of facial and cochlear nerves during acoustic neuroma surgery. Otolaryngol. Clin. North Am., 25(2):
413448.
Wazen, JJ (1994) Intraoperative monitoring of auditory
function: experimental observations and new applications. Laryngoscope, 104(4): 446455.

M.R. Nuwer (Ed.)
590
CHAPTER 42
Monitoring neural function during surgery around the

glossopharyngeal, vagus and laryngeal nerves during neck
(thyroid, larynx, carotid) and chest procedures
Thomas M. Hemmerling*
Department of Anesthesiology, McGill University Health Center, Montreal General Hospital, and Institut de Genie Biomedical,
University of Montreal, Montreal, PQ H3G 1A4, Canada
Nerve injury is a common complication in different

types of surgery. This chapter focuses on injury to the
recurrent laryngeal nerve (RLN) and vagus nerve in
thyroid surgery and presents several techniques of
monitoring. In addition, nerve injury in different types
of surgery is briefly discussed, though definite nerve
monitoring techniques are practically not used or nonexistent: thoracic surgery, cardiac surgery, surgery for
esophagus cancer, anterior cervical spine surgery, and
surgery for the carotid artery all bear the possibility of
significant nerve damage. At last, the very rare complication of damage to the glossopharyngeal nerve during
surgery is discussed.
42.1. Nerve damage and neuromonitoring in
thyroid surgery
42.1.1. Nerve damage in thyroid surgery: save the
recurrent laryngeal nerve!
Thyroid surgery implies several complications of
which damage to the recurrent laryngeal nerve
(RLN) is one of the most feared and common. Damage to the nerve has widespread implications for the
patient reaching from transient paresis with clinical
symptoms such as hoarseness and disturbances of
the voice, to permanent damage including the need
for laryngeal surgery or tracheostomy in rare cases.
*
Correspondence to: Prof. Thomas M. Hemmerling, M.D.,
DEAA, Department of Anesthesiology, McGill University
and Institute of Biomedical Engineering, University of
Montreal, Montreal General Hospital, 1650 Cedar Avenue,
Montreal, PQ H3G 1A4, Canada.
Tel.: 1-514-934-1934, ext 43030; fax: 1-514-934-8249.
E-mail: thomashemmerling@hotmail.com
(T.M. Hemmerling).
It is not surprising, therefore, that RLN lesions are

an important medicolegal issue after thyroid surgery.
The need for monitoring systems that help to avoid
this complication is important to improve the safety
of the surgical procedure and the quality of life thereafter. The German surgeons Kocher and Billroth dictated as surgical guidelines in the early era of thyroid
surgery that nerve identification and nerve exposure
should be avoided. In 1911, Bier, and in 1938, Lahey
proposed the routine identification of the RLN
(Lahey, 1938) as a means to reduce the incidence
of nerve damage after thyroid surgery. One of the
basic principles to avoid nerve damage is the surgical
exposure of the nerve that can lower the incidence of
transient or permanent nerve damage to below 5% or
below 1%, respectively (Jatzko et al., 1994; Hermann
et al., 2002; Dralle et al., 2004).
Traditionally, the recurrent laryngeal nerve is
located by the palpation and identification of the
inferior thyroid artery, followed by gentle dissection
in the area directly caudad to the artery in the plane
of the tracheoesophageal groove (Fig. 1). Identification by this simple tactile method might be
impaired by previous operative procedures, anatomic
variation, or just by the mere presence of large,
multinodular goiters.
Electric nerve stimulation and recording of the
evoked response has the potential to significantly
help the surgeon to identify the RLN.
42.1.2. Intraoperative monitoring of the RLN
Intraoperative monitoring of the RLN has started in
Europe and was first described by Flisberg and
Lindholm (1969). Neuromonitoring has gained widespread use, especially in Germany, with growing
popularity in North America.
591
4
2
8
3
9
5
5
6
10
Fig. 1. Schematic illustration of the recurrent laryngeal nerve

and the superior laryngeal nerve. 1 external carotid artery,
2 superior thyroid artery, 3 common carotid artery, 4
internal jugular vein, 5 inferior thyroid artery, 6 recurrent
laryngeal nerve, 7, 8 internal, external branch of the
superior laryngeal nerve, and 9, 10 superior and inferior
parathyroid gland. Modified after Jonas and Bahr (2000) with
permission from Springer and Science Media.
42.1.2.1. Techniques of intraoperative monitoring of

the RLN
42.1.2.1.1. Laryngeal palpation after RLN
stimulation. A very simple technique consists of
stimulating the RLN and measuring the muscle
response by palpating the postcricoid region of
the larynx through the posterior hypopharyngeal
wall (James et al., 1985; Gavilan and Gavilan, 1986;
Echeverri and Flexon, 1998; Otto and Cochran,
2002) the contraction of the posterior cricoarytenoid muscle is sensed. However, most would consider this technique merely as an adjunct to
laryngeal electromyography (EMG) monitoring
(Randolph et al., 2004) since any quantitative evaluation of RLN function is impossible or at least subjective, depending on the sensitivity of the
surgeons finger tip (Fig. 2).
Fig. 2. Illustration of the palpating maneuver to sense contraction of the posterior cricoarytenoid muscle (PCA).The
finger tip glides behind the larynx and palpates the posterior plate of the cricoid cartilage to sense contraction of
PCA. Modified after Randolph et al. (2004) with permission from Elsevier.
42.1.2.1.2. The use of needle electrodes for intraoperative monitoring of the RLN. Initially, needle
electrodes were placed via direct vision either into
the vocal cords using either an endoscopic technique
or direct but blind insertion into the vocal cords
through the cricothyroid membrane.
42.1.2.1.3. Needle electrodes placed intramuscular
through the cricothyroid ligament into the vocal
cords. This technique can be applied by the surgeon
himself from the surgical field and is easy to perform
(Fig. 3) (Jonas and Bahr, 2000; Brauckhoff et al.,
2002; Hamelmann et al., 2002; Tschopp and Gottardo,
2002; Kunath et al., 2003).
However, placement of the needle electrode is
blind, hematoma creation of the vocal cords cannot
be visualized and the tip of the needle might not be
placed in the right place.
42.1.2.1.4. Endoscopic needle placement into the
vocal folds (intramuscular electromyography). This
technique allows the insertion of needle electrodes
(fine wire electrodes) under endoscopic vision
directly into the vocal folds (Fig. 4A and B)
592
T.M. HEMMERLING
laceration, vocal cord hematomas, or, in rare cases,

the puncture of the endotracheal cuff can occur.
Fig. 3. Intraoperative situs after thyroidectomy with lymphadenectomy using intraoperative nerve monitoring via
needle electrodes placed intramuscularly through the cricothyroid ligament into the vocal cords. Neurostimulation
of the right recurrent laryngeal nerve (RLN) 1 larynx,
2 trachea, 3 aortic arch, 4 right carotid artery,
A intralaryngeal electrode through the cricothyroid ligament, B stimulation electrode. Reprinted from Brauckhoff
et al. (2002) with permission from Elsevier.
(Lipton et al., 1988; Rice et al., 1991; Tschopp and

Probst, 1994). However, needle electrodes can potentially cause damage to the vocal folds or cords;
needle fragmentation, electrode displacement with
42.1.2.2. Intraoperative monitoring of the RLN using

surface EMG
42.1.2.2.1. The postcricoid surface electrode. The
postcricoid surface electrode is inserted using a
standard rigid laryngoscope into the upper pharynx,
sometimes under the guidance of heart beat recording that indicates correct position of the electrode.
It rests there attached toward the posterior part of
the larynx (Fig. 5; Pearlman et al., 2005) (Khan
et al., 1997; Rea and Khan, 1998; Marcus et al.,
2003).
42.1.2.2.2. Surface EMG of the vocal cords using
specialized endotracheal tubes. The EMG endotracheal tube consists of a low-pressure cuffed silicone
elastomer endotracheal tube with integrated bilateral
paired stainless-steel wire electrodes (Fig. 6; Pearlman et al., 2005) (Eisele, 1996; Barwell et al.,
1997; Lamade et al., 2000). The EMG endotracheal
tube interfaces with a monitor that continuously
tracks EMG activity and has a built-in pulse generator for electrically evoked EMG.
Wires
ANT.
Ventricular fold
(False cord)
Vocal fold
(True cord)
Ext. thyroarytenoid m.
Int. thyroarytenoid m.
Lat. cricoarytenoid m.
Ventricular fold
(False cord)
Epiglottis
Ventricular fold
(False cord)
Cricothyroid m.
Cricoid cartilage
FRONTAL SECTION
Vocal Fold (True cord)

POST.
Fig. 4. Illustration of endoscopic placement of needle electrodes into the vocal folds. A: Coronal section of the larynx at the
midmembranous cord level during electrode insertion. B: Endoscopic view of larynx during electrode insertion. Reprinted
from Lipton et al. (1988) with permission from Lippincott, Williams and Wilkins.
Fig. 5. Illustration of the postcricoid surface electrode

for recurrent laryngeal nerve (RLN) system monitoring
(RLN System, Jefferson City, MO, USA). Reprinted from
Pearlman et al. (2005) with permission from Lippincott,
William and Wilkins.
Fig. 6. The Medtronic Xomed (Metronic, Jacksonville, FL,

USA) electromyography (EMG) endotracheal tube. Reprinted from Pearlman et al. (2005) with permission from
Lippincott, William and Wilkins.
42.1.2.2.3. Surface electrode EMG monitoring. This

method is similar to the method where wires are built
into the special tube; however, this surface electrode
can be applied to different tubes (Fig. 7), including
double-lumen tubes (Hemmerling et al., 2000). Care
has to be taken not to lose contact between the tube
and the surface electrode.
In the following, pros and cons of the different
surface electrodes are outlined:
Postcricoid surface electrode
Pro
Can be changed in size and form
593
Fig. 7. Endotracheal tube and surface electrode attached

distally to the endotracheal tube about 2 cm above the cuff;
the surface electrode is then placed between the vocal cords
for electromyography (EMG) tracing. Reprinted from
Hemmerling et al. (2000) with kind permission from
Springer Science and Business Media.
Heart beat is used to confirm correct positioning

Low cost
Con
Sensitive to downwards pressure artifact
Placement might be prevented by tumors
No unilateral identification possible (!); gives only
global RLN result from both sides, possibly dominant from posterior muscles.
EMG Endotracheal tube
Pro
High amplitudes
Individual, unilateral cord identification possible
at low-stimulation currents
Con
Different sizes necessary
Higher cost
When other specialty tubes need to be placed, surface electrode is better
Surface EMG electrode
Pro
Same pros as EMG endotracheal tubes
Low costs
Can be applied to all sorts of specialty tubes
Con
Can get detached during long procedures since it
is only glued; great care has to be taken to secure it
properly
42.1.2.3. Glottic pressure monitoring
An older and quite elaborate form of intraoperative
monitoring measures vocal cord movement by means
of placing a cuff of a double-cuffed endotracheal tube
between the vocal cords (Fig. 8A and B) (Woltering
et al., 1984). Surface EMG-based methods have
widely replaced this method.
594
T.M. HEMMERLING
EKG
Pacer
Pressure
Vocal Cords
(+)
()
7 90
ASYNC
B
Fig. 8. Illustration of the glottic pressure monitoring. A: Double-cuffed endotracheal tube used to monitor vocal cord
motion. B: The monitoring system used to monitor vocal cord motion during electric stimulation of the recurrent laryngeal
nerve. Reprinted from Woltering et al. (1984) with permission from Excerpta Medica Inc.
42.1.2.3.1. Direct visualization using laryngeal

mask airway and fiberoptic video imaging of vocal
cord movement. This recently described technique
(Fig. 9; Eltzschig et al., 2002) replaces the endotracheal tube by a laryngeal mask airway; anesthesia
is performed under spontaneous respiration (Akhtar,

1991; Hobbiger et al., 1996; Eltzschig et al., 2002)
and whenever the RLN is stimulated, a fiberoptic
scope is inserted into the laryngeal mask airway
(LMA), and vocal cord movement is recorded.
Fig. 9. Video monitoring of the vocal cord movement with

laryngeal mask airway. A laryngeal mask airway showing
all three ventilation ports before the use of the flexible
laryngoscope. A: Laryngeal mask airway with placement
of the flexible laryngoscope in the middle port. B: Side
ports remain completely open for ventilation. Reprinted
from Eltzschig et al. (2002) and Hermann et al. (2004)
with permission. Copyright # 2002 American Medical
Association. All rights reserved.
42.1.2.4. Basic principles of intraoperative

monitoring of the RLN
Basis of all techniques is the electric stimulation of
the RLN or vagal nerve; this can be achieved either
transcutaneously using superficial electrodes, usually
performed before or after surgery as a base line
reading of RLN integrity, or directly during surgery
using different handheld stimulating electrodes. The
compound action potentials of the EMG evoked signal are recorded from muscles of the larynx, either
the vocalis muscle or other muscles of the larynx,
such as the arytenoids or posterior cricothyroid muscles. Therefore, the evoked signals are predominantly
derived from muscles responsible for opening the
vocal cords, and, when using the pharyngolaryngeal
electrode, the sole muscle responsible for closing of
the vocal cords. However, as with most superficial
electrodes and even needle electrodes, the compound
action potential might reflect signals from several
muscles since they are very closely positioned within
the larynx. These evoked signals can be visualized or
transferred into acoustic signals. In addition, some

centers add continuous spontaneous EMG into their
monitoring, thus providing a helpful warning signal
to the surgeon whenever his maneuvering of the thyroid gland puts strain to the RLN.
42.1.2.4.1. Evolution of intraoperative RLN monitoring. In the early stages of intraoperative monitoring of the RLN, mostly needle electrodes were
inserted under direct laryngoscopy into the arytenoid
muscles, mostly by the surgeon, in this case ENT surgeon, himself. During the last decade, less invasive
forms of monitoring have emerged in form of
superficial electrodes, inserted either into the pharynx, just behind the larynx, for monitoring the posterior part of the larynx, mainly the posticus muscle,
or surface electrodes, glued and attached to the endotracheal tube, as well as specially modified endotracheal tubes with built-in wires for recording of the
vocalis muscle EMG. Recently, the combination of
laryngeal mask airway and direct fiberoptic visualization of the vocal cords during stimulation of the
RLN has been proposed as a valid and easy-to-use
monitoring method.
In most centers, monitoring systems mostly provide acoustic signals for the surgeon, thus providing
help in identifying the RLN and testing its integrity
after the resection of the thyroid gland. More sophisticated analysis is available by using the visual
recording of the EMG, measuring parameters such
as height, form, or latency of the signal. However,
reliable criteria how to interpret and identify changes
in the form or latency of the EMG signal do not exist.
Some recent studies have looked at things like stimulation current threshold in order to relate especially
the occurrence of temporary RLN damage with for
example diminished amplitudes, changes in form
and latency of the RLN at the end of surgery.
42.1.2.5. Clinical usefulness of intraoperative
RLN monitoring
Basic function of any monitoring system of RLN
integrity consists of, firstly, aid to identify the nerve
during surgery and, secondly, the analysis of nerve
integrity before and after surgery.
Several studies have shown that nerve identification can be helped by using a stimulating probe. This
is especially true for teaching centers or novice surgeons who use the stimulating probe like an identifying probe thus probably helping their confidence
and saving time by more deliberate resection of the
595
thyroid gland. Alternative methods are purely visual,

digital, or microscopic identification of the nerve.
Many surgeons will admit that in some cases even
to the eye and touch of an experienced surgeon,
objectively identifying the nerve by stimulation and
acoustic signal detection, is helpful and saves time.
The final verdict of the usefulness of intraoperative monitoring of the RLN must be based on the
avoidance or limitation of temporary or permanent
damage to the RLN as verified by postoperative
laryngoscopic analysis. This verdict, however, is
limited by the absence of well-done prospective
randomized studies comparing postoperative outcome with or without nerve monitoring.
42.1.2.6. Technical pitfalls of intraoperative
RLN monitoring
It is very important to first stimulate a tissue other
than nerve tissue that could evoke a muscle response.
Thus, an acoustic base signal is formed, which is
easily distinguishable from any correct monitoring
(stimulation of RLN or vagus nerve detection of
the evoked muscle response by acoustication of the
EMG signal). This usually results in a very highpitched, sharp acoustic signal, being monophasic
and not all like the rhythmic machine-gun type signal
evoked by the RLN stimulation.
No signal means:
a.
no nervous structure is stimulated that could

result in laryngeal contraction, depending on the
muscle;
b. nervous structures are stimulated that do not
innervate the vocalis muscle or the posticus; and
c. RLN is damaged. However, in the latter case, a
falsely positive signal might arise whenever the
nerve damage occurs distal to the stimulation site.
A changed signal (either in latency, form, amplitude,
etc.) can arise by damage to the RLN, very often
either by continuous strain on the nerve or even by
continuous stimulation.
Several studies have shown the importance of the
degree of muscle relaxation during general anesthesia
and its influence on the signal (see below).
Ideally, intraoperative monitoring starts with a
first stimulation of the RLN, either preoperatively
by transcutaneous stimulation or directly during surgery at first site so to speak, followed by a control
after resection, again either directly in the operating
situs or immediately after surgery transcutaneously.
At present, it is only differentiated to have a
596
normal mainly acoustic signal (same frequency and

amplitude) or with more sophisticated devices normal shape, form, latency, and amplitude of the signal.
Any derivation from this is considered as changed
signal and can be due to various reasons, such as
operator connected (different stimulation site, change
from direct to transcutaneous stimulation, etc.) or
indicating nerve damage of various degrees.
42.1.2.7. Intraoperative RLN monitoring and
postoperative vocal cord function
The non-plus-ultra of diagnosing RLN damage is
postoperative function testing of the vocal cords.
Changes in vocal cords function have to take into
account a certain degree of damage due to endotracheal intubation; especially with surgery such as
thyroid surgery, where very often hyperextension
and prolonged duration of surgery occurs, the normal
prevalence of intubation-related damage of up to 6%
(Friedrich et al., 2000) might even be higher.
It is equally important that postoperative vocal
cord function testing is performed at a given time
after surgery, for example 3 days postoperatively
and not only when clinical impairment of vocal cord
function occurs. In addition, temporary damage to
the vocal cords might be missed if vocal cord laryngoscopic examination is delayed. Anesthesia- and
surgery-related causes, such as the length of ventilation, postoperative alertness, and pain, have an
impact on the availability and validity of vocal cord
function testing.
The use of surface electrode monitoring or
monitoring using the EMG endotracheal tube only
monitors the adducting laryngeal muscles. Should
therefore the RLN be separated into an anterior and
posterior branch prior its insertion to the larynx, an
isolated damage to the posterior branch might be left
unnoticed. This is one of the reasons why some surgeons advocate the additional use of the posterior
laryngeal electrode and intraoperative direct fiberoptic monitoring might be superior to isolated monitoring of the adducting laryngeal muscles.
It is equally important that the person performing
postoperative laryngoscopic diagnostics is aware of
the intraoperative and immediately postoperative
findings of the RLN monitoring. It is clear that any
positive findings in vocal cord function testing
should initiate more elaborate function testing, such
as EMG and stroboscopy in order to differentiate
between neurogenic and non-neurogenic origin of
the damage.
T.M. HEMMERLING
42.1.2.7.1. Falsely positive findings [modified

after Timmermann et al. (2004)]. (Intraoperative
monitoring RLN at the end of surgery indicates
normal RLN function; however, postoperative
function testing shows vocal cord damage.)
a. RLN damage more distal than stimulation site
b. Changes in signal form, shape, signal amplitude,
or latency; different stimulation site
c. Postoperative causes such edema, hematoma, or
surgical damage after the last stimulation
d. Damage due to endotracheal intubation or extubation, or prolonged ventilation
e. Damage to the posterior branch when posticus
integrity is not monitored during surgery.
It is highly unusual, however, to find a positive function testing during surgery and permanent RLN damage after surgery. Most lesions are transient and most
of them do not necessitate any further therapy.
42.1.2.7.2. Falsely negative findings [modified
after Timmermann et al. (2004)]. (Postoperative
function testing reveals normal RLN function despite
an impaired or absent signal during intraoperative
nerve testing.)
a.
Changes in signal form, shape, signal amplitude,

or latency; different stimulation site
b. Different degrees of muscle relaxation. This situation is rather unlikely at the end of surgery if
immediate extubation is thought after.
c. Transient damage that reverses before postoperative function testing.
d. The use of monopolar versus bipolar stimulation?
e. Increasing stimulation currents and highfrequency stimulation during long periods.
42.1.2.8. Nerve identification during surgery by
using intraoperative RLN monitoring
Several studies have proven the high frequency of
nerve identification independent which monitoring
system and technique is used; a success rate of up
to 100% can be reached. Whichever difficulty exists
concerning the analysis of the present studies, most
studies have shown a significant reduction of permanent nerve damage when intraoperative nerve monitoring was used. Most studies, however, have also
shown that the rate of transient nerve damage is
unchanged by intraoperative RLN monitoring. This
might be caused by either a lack of fine-tuning of
the present monitoring techniques which are simply
597
too brittle to detect a transient damage or the fact

that other than neurogenic reasons play an important
role in transient nerve damage (e.g., intubation,
extubation damage).
42.1.3. The key question: does intraoperative
monitoring of the RLN really reduce the
incidence of RLN damage?
The most elaborate study was performed in Germany
and published in 2004 (Dralle et al., 2004). In a prospective audit, a total of 16,448 consecutive thyroid
interventions were performed (29,998 RLN nerves
at risk) and three treatment forms were compared:
no nerve identification, visual nerve identification
and visual identification, and RLN monitoring using
evoked laryngeal EMG using direct intramuscular
needle monitoring through the cricothyroid ligament.
The study clearly showed that the visual RLN identification is the gold standard of thyroid surgery and
that RLN monitoring as an adjunct can help avoiding
RLN damage in selected cases (Table 1) (Dralle
et al., 2004).
There are, however, certain limitations to this
study; it was not a randomized study, surgeons could
use the technique they preferred, and there was significantly more RLN monitoring in difficult cases
or reoperation.
The authors conclude the following for thyroid
surgery and RLN monitoring:
1. Recurrent benign or malign thyroid disease is the
biggest risk factor for RLN damage.
2. In surgery for Graves disease and Hashimotos
goiter, RLN monitoring significantly reduces
RLN nerve damage.
3. Lobectomy is a significant risk factor for RLN

damage.
4. There is no difference of RLN damage with or
without RLN visualization in subtotal thyroidectomy.
5. High-volume, experienced surgeons have the
lowest risk of postoperative RLN damage.
This results in the risk factors outlined in Table 2
(Dralle et al., 2004).
The enormous number of patients needed to sufficiently power a study to find significant differences
on terms of RLN damage with or without RLN monitoring is a very limiting factor; however, this large,
multicenter study clearly gives the message that
though RLN damage carries a low overall risk in
thyroid surgery (<1%), RLN monitoring is helpful
in cases carrying a high risk of nerve damage and
might be especially helpful for beginners.
Table 2
Risk factors for thyroid surgery and RLN monitoring
Risk factor
Risk (OR)
Recurrent malign goiter

Recurrent benign goiter
Thyroid cancer
Lobectomy
No nerve identification
Low- or medium-volume hospital
Low-volume surgeons
6.7
4.7
2.0
1.8
1.4
1.3
1.2
RLN: recurrent laryngeal nerve.

Source: Results from Dralle et al. (2004) with permission from
Lippincott, Williams and Wilkins.
Table 1
Multivariate analysis with logistic regression of risk factors of permanent RLN paralysis
Variable
Thyroid disease
Immunogenic goiter vs. benign MNG
Recurrent benign goiter vs. benign MNG
Thyroid malignancy vs. benign MNG
Recurrent thyroid malignancy vs. benign MNG
Extent of resection lobectomy vs. subtotal resection
P value
0.9734
<0.0001
0.0002
<0.0001
<0.0001
Odds ratio*
95% CI
0.99
4.67
2.04
6.66
1.76
0.60361.6290
3.48466.2566
1.40992.9586
2.682816.5176
1.37662.2571
Note: Logistic regression coefficient, b; odds ratio <1 decreases the risk (e.g., subtotal resection decreases the risk of RLN paralysis compared with lobectomy, odds ratio >1 increases the risk; MNG: multinodular goiter; RLN: recurrent laryngeal nerve.
Source: Modified after Dralle et al. (2004) with permission from Elsevier.
598
T.M. HEMMERLING
42.1.4. How predictive is intraoperative

electrophysiological testing of RLN injury?
There is no question about the usefulness of intraoperative RLN monitoring for nerve identification.
Unfortunately, the correlation between RLN monitoring and postoperative vocal cord function testing is
less pronounced. Some studies have proposed stimulation thresholds and related parameters as predicting
outcome (Brennan et al., 2001). Hermann et al.
showed that monitoring RLN has a low sensitivity
to determine an abnormal response when postoperative paresis occurs (Hermann et al., 2004), especially
in cases where usually RLN monitoring is considered
as very helpful or needed such as thyroid cancer or
reoperations (Table 2) (Hermann et al., 2004). Interestingly, the predictive value in benign cases is better
(Table 3, 4) (Hermann et al., 2004).
However, as found in other studies, intact RLN

monitoring at the end of surgery predicts with up to
98% a normal postoperative function testing. Mostly,
intraoperative technical problems account for not
reaching 100%.
In addition, you would suspect that patients with
preoperative palsy should have a raised threshold
for stimulation or would exacerce no signal whatsoever. Hermann et al. found these nerves intraoperatively anatomically intact; there was even more
interestingly no change in the stimulation threshold
when either RLN or vagal stimulation occurred
(Fig. 10) (Hermann et al., 2004). However, these
findings are contradicted by Brennan et al.s findings
which showed that nerves requiring more than
0.5 mA of stimulation threshold are prone to postoperative palsy (Brennan et al., 2001). It is clear that
the verdict of the significance of the stimulation
Table 3
Sensitivity and negative predictive value of RLN monitoring in distinct surgical risk groups
Total
Benign disease, 1st operation
Benign disease, reoperation
Malign disease, 1st operation
Malign disease, reoperation
Nerves
at risk
Sensitivity
po paresis
Sensitivity
permanent palsy
Negative predictive
value (po)
Negative predictive
value (permanent)
475
179
157
70
30
57.1%
85.7%
57.1%
50%
25%
44.4%

40%
50%
96.6%
99.4%
95.9%
91.9%
94.7%
98.8%
100%
97.9%
96.6%
100%
NB: po: postoperative; RLN: recurrent laryngeal nerve.

Source: Reprinted from Hermann et al. (2004) with permission from Lippincott, William and Wilkins.
Table 4
Positive predictive value of RLN monitoring in distinct surgical groups
Total
Benign disease,
1st operation
Benign disease,
reoperation
Malign disease,
1st operation
Malign disease,
reoperation
Positive predictive
value, permanent
palsy (global)
Positive predictive
value, permanent
palsy (minus
technical problems)
87%
75%
25%
57.1%
72.7%
88.9%
50%
100%
57
62.5%
100%
26
100%
100%
Correct
recording
of nerve
function
Positive predictive
value, po paresis
(global)
428
166
62.5%
50%
141
Positive predictive
value, po paresis
(minus technical
problems)
NB: po: postoperative; RLN: recurrent laryngeal nerve.

Source: Reprinted from Hermann et al. (2004) with permission from Lippincott, William and Wilkins.
80
recurrent-benign
vagus-benign
vagus-benign
recurrent-postop.palsy
recurrent-preop.palsy
vagus-postop.palsy
vagus-preop.palsy
100
80
60
60
40
40
20
20
0
0.00
recurrent-benign
number of nerves at risk (%)
number of nerves at risk (%)
100
599
0
0.03
0.10 0.30 1.00

current (mA)
3.00
0.00
0.03
0.10 0.30 1.00

current (mA)
3.00
Fig. 10. Cumulative distribution of stimulation thresholds A: in patients who developed postoperative palsy and B: in
patients who presented with preoperative palsy. The stimulation threshold was determined before wound closure as outlined
in B. Shown are the cumulative distributions of those patients who developed postoperative palsy ( and , recurrent and
vagus nerve abbreviated as recurrent-postoperative palsy and vagus postoperative palsy, respectively) and those patients
who presented with preoperative palsy ( and , recurrent and vagus nerve abbreviated as recurrent-postoperative palsy
and vagus-postoperative palsy, respectively) and in whom an electric field response was nevertheless recorded. Control
patients (same data as in A, , recurrent nerve recurrent-benign; vagus-benign) are included for comparison.
Reprinted from Hermann et al. (2004) with permission from Lippincott, Williams and Wilkins.
threshold is still out and more studies should focus on

this issue since it might hold the answer why
transient RLN damage after surgery can as yet not
been predicted by intraoperative RLN monitoring.
42.1.5. The influence of muscle relaxation on
RLN monitoring
Most authors stress the importance of the absence of
muscle relaxation while RLN monitoring is used.
This might not always be easy to achieve, especially
in large goiters, with deviation of the trachea or complicated, prolonged cases. We have found (Hemmerling et al., 2000) that the combination of propofol
and remifentanil general anesthesia with high levels
of remifentanil allows best to avoid muscle relaxation. This is based on some studies showing that
even the insertion of the endotracheal tube is possible
using these two drugs without muscle relaxation
(Robinson et al., 1998; Erhan et al., 2003). Techniques
of spontaneous respiration together with LMW insertion, local analgesic techniques, and video-monitoring
of the RLN function have been described (Akhtar,
1991; Tanigawa et al., 1991; Hobbiger et al., 1996;
Eltzschig et al., 2002), though this is not applicable
for every patient (morbidly obese patients, very large

goiters with airway obstruction, other contraindications to the use of LMA such as a strong history for
reflux, etc.) nor would most anesthesiologists, especially in North America, feel very comfortable with
using LMAs in surgery where difficult airway management might suddenly occur. In addition, access to
the head in these cases might be hampered by surgical
draping; at present, the use of LMAs in thyroid surgery
is certainly no standard. Marusch et al. have shown
that RLN monitoring is possible even in the presence
of muscle relaxation (Fig. 11); however, the threshold
for nerve stimulation increases and the amplitude of
the evoked potentials decreases with increasing
degree of muscle relaxation (Marusch et al., 2005).
Therefore, a high degree of muscle relaxation during RLN monitoring is certainly not recommended.
However, usually and in Maruschs study, muscle
relaxation would be determined at hand muscles; muscles a high degree of muscle relaxation at the hand is
associated with a much less degree of muscle relaxation at the larynx (Hemmerling and Donati, 2003).
This might explain why even with moderate hand
relaxation reasonably good RLN monitoring can be
achieved. This means that moderate muscle relaxation
T.M. HEMMERLING
SAcP mVs
600
26
24
22
20
18
16
14
12
10
8
6
4
2
0
**
**
**
**
**
**
**
+
+
0
+
5
+
10
+
+
15
20
TW %
+
25
+
30
100
Maximal 4.64 9.89 10.89 14.50 13.35 20.78 22.74 15.36

Mean
1.27 1.92 2.68 3.51 4.18 5.08 5.55 8.45
+ Minimal 0.06 0.08 0.14 0.27 0.46 0.24 0.35 4.83
Fig. 11. Summed action potentials of the vocalis muscle at various degrees of relaxation (study subjects). **P < 0.001.
Reprinted from Marusch et al. (2005) with permission. Copyright # The Board of Management and Trustees of the British
Journal of Anaesthesia. Reproduced by permission of Oxford University Press/British Journal of Anaesthesia.
at the hand, for example 50% decrease of NMB,

equals depending on the patient 25% to 15%
of NMB at the larynx, not sufficient to make RLN
monitoring impossible. In order to safely identify the
RLN with as little current as possible higher stimulation currents and continuous stimulations might
cause transient RLN nerve damage and as early
as possible, as little muscle relaxation possible or at
best no muscle relaxation should be used.
42.1.6. Conclusion
Intraoperative monitoring of the RLN significantly
reduces the risk of permanent RLN damage after thyroid surgery.
Especially in surgeries with a high risk of nerve
damage, such as extensive subtotal resections, lobectomies, recidive thyroid surgery, or surgery with
anatomically difficult RLN, intraoperative RLN
monitoring should be used.
However, it needs to be remembered that the anatomic integrity of the RLN does not necessarily guarantee normal function after surgery, though it can
with a very high certainty indicate no permanent nerve
lesion, transient nerve impairment cannot be ruled out.
Especially in surgeries of both sides, the intraoperative testing of RLN integrity at the end of surgery of one side is very important; a positive signal
allows proceeding from unilateral surgery to bilateral
surgery whereas if a negative signal is obtained, care

should be taken not to damage the other side, including abandoning the procedure.
This means that bilateral surgery should start on
the side where the bigger thyroid gland is noted.
The following intraoperative RLN monitoring
should be performed as a standard:
1. Preoperative stimulation of the vagus nerve via
transcutaneous stimulation
2. Stimulation of the vagus nerve at the beginning of
surgery
3. Nerve identification using direct RLN stimulation
throughout surgery
4. For bilateral resection: Whenever RLN intact
(positive signal), proceed to next side; if signal
negative, review indication for (immediate) surgery of the either, consider two-stage surgery
5. Direct stimulation of vagus nerve at the end of
resection on both sides
6. Postoperative transcutaneous stimulation of the
vagus nerve, comparison to preoperating findings
An example of a very detailed intraoperative monitoring is given in Fig. 12 (Pearlman et al., 2005).
Postoperative function testing of the vocal cords
needs a cooperative patient, unimpaired by pain or
swelling. It is best performed 35 days after surgery.
Intraoperative RLN monitoring allows nerve
identification en route to visual identification that
601
Fig. 12. Example of a very sophisticated monitoring system. Electromyography (EMG) train activity and evoked compound muscle potentials to stimulation. Reprinted from Pearlman et al. (2005) with permission from Lippincott, William
and Wilkins.
remains a benchmark of thyroid surgery. Positive signal testing at the end of surgery almost always
predicts normal postoperative function testing. Negative RLN monitoring on one side should initiate revision of the plan of doing the other side immediately,
a two-stage surgical approach might be the safer
alternative if possible.
The most widely used intraoperative monitoring
system (surface electrode or special endotracheal
monitoring tube) only monitor the adducting laryngeal muscles; damage to the posterior branch of the
RLN might be unnoticed.
Future studies have to focus on a more detailed
analysis of EMG signals; more reliable signal detection and visualization might be useful, such as EMG
recording and comparison of form, shape, signal
height, and latency in terms of their predictive value
for either transient or permanent RLN damage. The

importance of using continuous spontaneous EMG
tracing as guidance for a surgical no touch technique should be evaluated in terms of decrease of
postoperative transient damage.
42.2. Intraoperative vagal/RLN nerve damage in
surgery of the neck
The same principles as described earlier apply to
other types of surgery where either the RLN or the
vagal nerve are at risk. Clinical complications of
nerve damage result in problems swallowing, and
different problems of airway functioning that might
in a worst case scenario lead to temporary or permanent tracheostomy. These surgeries include surgery
for benign or malign tumors in the cervical neck,
602
parapharyngeal space, or infratemporal fossa. In

addition, surgery of the cervical spine, especially
with an anterior approach can cause impairment
of RLN or vagal nerve function. However, in comparison to thyroid surgery, regular nerve monitoring
during surgery is not a standard of care.
The vagus nerve plays an important role in deglutition, airway protection, and voice production. It
innervates all laryngeal muscles and contributes to
the pharyngeal plexus.
Vagal nerve monitoring consists of endoscopically
inserting needle electrodes into the vocalis muscle
and using a standard stimulating probe for neural
identification. The nerve is at risk as the RLN during thyroid surgery due to direct surgical trauma,
stretching due to maneuvering in more distant
tissues pulling the nerve concomitantly and
thermal injury due to cautery. Leonetti et al. have
described this technique for neck and base skull surgery (Leonetti et al., 1996) and showed a significant
decrease in vocal cord dysfunction after surgery in
comparison to no monitoring. They compared retrospectively 18 patients undergoing cervical neck or
skull base surgery without monitoring where the
vagal nerve was anatomically preserved, of which
5 (22%) showed normal vocal cord movement at
1 month postoperative, 13 (72%) at 6 months postoperatively. In the following 11 patients where intraoperative vagal monitoring was performed, 7 patients
(63%) had normal vocal cord function at 1 month,
and all patients at 6 months postoperative.
Several other studies support the usefulness of
intraoperative vagal nerve monitoring in these types
of surgery. However, the frequency of these surgeries
is much less than thyroid surgery, explaining the lack
of sufficiently powered studies to demonstrate a significant reduction of morbidity after surgery with
monitoring.
42.2.1. Intraoperative nerve monitoring in anterior
cervical spine surgery
Interestingly, the incidence of RLN damage after anterior cervical spine surgery has been reported to be
higher than after thyroid surgery; some studies have
shown it to be around 2%, but others have found it to
be higher toward 20% (Cloward, 1962; Bulger et al.,
1985). One of the reasons for the fact that widespread
and routine RLN monitoring is not part of this type
of surgery is the multifactorial cause, ranging in
theory from traumatic division during surgery,
T.M. HEMMERLING
postoperative edema and scar formation, entrapment

of the nerve between retractors and ETT and overstretching of the trachea and the subsequent stretching
of the nerve during spine exposure (Jellish et al.,
1999). The fact that some of these causes can hardly
be avoided might explain the lack of routine monitoring. Jellish et al. applied therefore EMG monitoring
using the postcricoid laryngeal electrode as a surrogate
for recurrent LNS; monitoring as such did not diminish
the rate of RLN damage which was prospectively
38%, however, the damage clinically ressembled more
a pharyngolaryngeal discomfort syndrome, and was
probably more due to irritation of the ETT and prolonged intubation than direct nerve damage. Another
index for these factors being the reason of the symptoms
was the absence of any correlation of EMG activity to
these symptoms. Indeed, RLN monitoring might be less
helpful in these types of surgery since it does not change
the incidence of voice impairments and does not correlate with its incidence; these changes are most probably
due to ETT-related causes than actual RLN damage.
42.2.2. Cranial and cervical nerve risk in carotid
endarterectomy
Because of the anatomy, several cranial and cervical
nerves are also at risk when surgery of the carotid
artery is performed. As with other types of surgery
with life-threatening complications, such as esophagectomies, cardiac surgery (especially pediatric cardiac surgery) or thoracic surgery, risk of nerve
damage, though present and very often higher in frequency than for thyroid surgery, has not been a major
concern. This is not only due to the severity of the
surgery and the focus of surgical concern toward
the avoidance of major complications, such as
hemiplegia, but also due to the fact that the reason
for nerve damage is more multifactorial than
described for thyroid surgery. Direct damage by
transecting a given nerve is extremely rare since
operating site exposure is usually very good. Therefore, nerve damage is mostly due to retraction,
stretching, or clamping with the subsequent malperfusion of the nervous structures. Only recently, there
are more studies on the actual incidence of nerve
damage in carotid surgery with efficient monitoring
systems still to be assessed. Several retrospective
and prospective studies have shown that the incidence of cranial nerve injury is in the region of
312% (Rogers and Root, 1988; Forssell et al.,
1995) with a recent prospective study showing
12.5% of cranial nerve injury, including hypoglossal

nerve, RLN, superior laryngeal nerve, marginal mandibular, and greater auricular nerve injuries (Ballotta
et al., 1999).
42.3. Nerve injury in cardiac surgery
Topical cooling is a common feature of cardiac surgery; especially the topical application of iced slush
has been associated to phrenic nerve injury (Diehl
et al., 1994). It will result in elevated hemidiaphragm, a complication ranging from 10% to 85% in
some studies (Large et al., 1985; Dimopoulou et al.,
1998). Large et al. showed in their prospective study
of 36 consecutive patients undergoing cardiopulmonary bypass (Large et al., 1985) that 16 patients
(44%) had left diaphragmatic weakness or paralysis
in the early postoperative period, 4 patients (80%)
at 6 months and 2 patients (90%) at 1 year after surgery, respectively.
So, some clinicians have now abandoned the use
of ice slush and use cold water which might decrease
the incidence of this complication. Obviously, application of cold water cannot be avoided completely in
cardiac surgery, but care should be taken to avoid
direct application of ice onto the phrenic nerve.
603
Even in minimally invasive cardiac surgical

procedures, damage of the RLN might be an issue,
such as in video-assisted thoracoscopic ligation of
the patent ductus arteriosus. A study (Odegard et al.,
2000) has recently presented a new technique where
intramuscular monitoring of the vocal cords through
the cricothyroid ligament was helpful in identifying
and sparing the RLN (Fig. 13A and B).
42.4. Nerve injury during thoracic surgery
Nerve injury occurs during thoracic surgery in <1%
of the cases (Krasna and Forti, 2006). It is usually
due to either direct dissection of the RLN, long thoracic, sympathetic or vagus nerve, or stretching,
retractor action or stretching of adjacent structures.
Some treatment modalities exist with type I thyroplasty concomitant to thoracic operation as the latest
proposed procedure. One surgical group (Mom et al.,
2001) performed this type of surgery in 25 patients
(Fig. 14); 12 (86%) patients were fully satisfied with
the quality of voice. In addition, the frequent aspiration rate in patients with affected laryngeal function
after thoracic surgery [up to one third (Mom et al.,
2001)] can be avoided with this promising new
approach.
10
15
20
25 30
ms
35
40
45 50 ms
Fig. 13. Intraoperative recurrent laryngeal nerve (RLN) monitoring during video-assisted thoracoscopic surgery for patent
ductus arteriosus. A: Typical operative view using video-assisted thoracoscopic surgery. White arrow indicates surgical clip
on the patent ductus arteriosus. Black arrow indicates stimulating probe, the distal tip of which is located on the left recurrent laryngeal nerve. The recurrent laryngeal nerve is lying inferior and medial to the patent ductus arteriosus and in this
patient coursed close to its medial (behind in this view) surface. AO aorta; PA pulmonary artery. B: Representative
compound electromyograms obtained before (top tracing) and after (lower tracing) video-assisted thoracoscopic clipping
of the patent ductus arteriosus. Reprinted from Odegard et al. (2000) with permission from Elsevier.
604
T.M. HEMMERLING
during tonsillectomies but are extremely rare (Vories,

1999; Goins and Pitovski, 2004). Damage to the glossopharyngeal nerve most often affects the lingual
branch which is in close anatomic proximity to the
palatine tonsil. The clinical feature of nerve damage
is mostly a localized loss of taste function with or
without the occurrence of chronic tastes, mostly bitter or metallic.
References
Fig. 14. Vocal fold medialization: intraoperative view (left
side of the thyroid cartilage). The graft has been interposed
between the cartilage window and the remaining thyroid
lamina. Ant, anterior; Inf, inferior; TL, thyroid lamina.
Reprinted from Mom et al. (2001) with permission.
Copyright # 2001 American Association for Thoracic
Surgery.
42.5. Nerve injury in esophagus surgery

Open transhiatal, transthoracic, or three hole approach
are the three most common operating procedures for
esophagus cancer. If surgery for esophagus cancer is
associated with the dissection of the neck, RLN damage is possible, the risk is the greatest.
One study (Gockel et al., 2005) found that in 63
of 404 patients undergoing esophagectomy for carcinoma, RLN paralysis was diagnosed postoperatively,
47 of whom unilaterally and 16 bilaterally. In addition, these patients suffered most from consequent
respiratory complications. The incidence of postoperative pneumonia was significantly increased (33 of
63 as opposed to 90 of 341 patients without RLN
paralysis; P 0.027). All patients with bilateral
vocal cord paralysis developed pneumonia, against
17 of 47 patients with a unilateral lesion. In patients
with unilateral as well as with bilateral RLN paralysis, the duration of intensive care unit and postoperative in-hospital stay was significantly longer. So
RLN damage increases significantly morbidity and
length of stay in these patients. Minimally invasive
esophagectomy might be an alternative surgical
approach that has the potential to completely avoid
injury to the RLN (Bizekis et al., 2006).
42.6. Injury to the glossopharyngeal nerve
Injury to the glossopharyngeal nerve is extremely
rare; most of the case reports show nerve damage
Akhtar, TM (1991) Laryngeal mask airway and visualisation of vocal cords during thyroid surgery. Can. J.
Anaesth., 38: 140.
Ballotta, E, Da Giau, G, Renon, L, Narne, S, Saladini, M,
Abbruzzese, E and Meneghetti, G (1999) Cranial and
cervical nerve injuries after carotid endarterectomy: a
prospective study. Surgery, 125: 8591.
Barwell, J, Lytle, J, Page, R and Wilkins, D (1997) The
NIM-2 nerve integrity monitor in thyroid and parathyroid surgery. Br. J. Surg., 84: 854.
Bizekis, C, Kent, MS, Luketich, JD, Buenaventura, PO,
Landreneau, RJ, Schuchert, MJ and Alvelo-Rivera, M
(2006) Initial experience with minimally invasive Ivor
Lewis esophagectomy. Ann. Thorac. Surg., 82:
402406; discussion 406407.
Brauckhoff, M, Gimm, O, Thanh, PN, Brauckhoff, K,
Ukkat, J, Thomusch, O and Dralle, H (2002) First
experiences in intraoperative neurostimulation of the
recurrent laryngeal nerve during thyroid surgery of children and adolescents. J. Pediatr. Surg., 37: 14141418.
Brennan, J, Moore, EJ and Shuler, KJ (2001) Prospective
analysis of the efficacy of continuous intraoperative
nerve monitoring during thyroidectomy, parathyroidectomy, and parotidectomy. Otolaryngol. Head Neck
Surg., 124: 537543.
Bulger, RF, Rejowski, JE and Beatty, RA (1985) Vocal
cord paralysis associated with anterior cervical fusion:
considerations for prevention and treatment. J. Neurosurg., 62: 657661.
Cloward, RB (1962) New method of diagnoss and treatment of cervical disc disease. Clin. Neurosurg., 8:
93132.
Diehl, JL, Lofaso, F, Deleuze, P, Similowski, T, Lemaire, F
and Brochard, L (1994) Clinically relevant diaphragmatic dysfunction after cardiac operations. J. Thorac.
Cardiovasc. Surg., 107: 487498.
Dimopoulou, I, Daganou, M, Dafni, U, Karakatsani, A,
Khoury, M, Geroulanos, S and Jordanoglou, J (1998)
Phrenic nerve dysfunction after cardiac operations: electrophysiologic evaluation of risk factors. Chest, 113:
814.
Dralle, H, Sekulla, C, Haerting, J, Timmermann, W, Neumann, HJ, Kruse, E, Grond, S, Muhlig, HP, Richter, C,
Voss, M, Thomusch, O, Lippert, H, Gastinger, I,
Brauckhoff, M and Gimm, O (2004) Risk factors of
paralysis and functional outcome after recurrent laryngeal nerve monitoring in thyroid surgery. Surgery,
136: 13101322.
Echeverri, A and Flexon, PB (1998) Electrophysiologic
nerve stimulation for identifying the recurrent laryngeal
nerve in thyroid surgery: review of 70 consecutive thyroid surgeries. Am. Surg., 64: 328333.
Eisele, DW (1996) Intraoperative electrophysiologic monitoring of the recurrent laryngeal nerve. Laryngoscope,
106: 443449.
Eltzschig, HK, Posner, M and Moore, FD, Jr. (2002) The
use of readily available equipment in a simple method
for intraoperative monitoring of recurrent laryngeal
nerve function during thyroid surgery: initial experience
with more than 300 cases. Arch. Surg., 137: 452456;
discussion 456457.
Erhan, E, Ugur, G, Gunusen, I, Alper, I and Ozyar, B
(2003) Propofol not thiopental or etomidate with
remifentanil provides adequate intubating conditions in
the absence of neuromuscular blockade. Can. J.
Anaesth., 50: 108115.
Flisberg, K and Lindholm, T (1969) Electrical stimulation
of the human recurrent laryngeal nerve during thyroid
operation. Acta Otolaryngol. Suppl., 263: 6367.
Forssell, C, Kitzing, P and Bergqvist, D (1995) Cranial
nerve injuries after carotid artery surgery. A prospective
study of 663 operations. Eur. J. Vasc. Endovasc. Surg.,
10: 445449.
Friedrich, T, Hansch, U, Eichfeld, U, Steinert, M, Staemmler,
A and Schonfelder, M (2000) Recurrent laryngeal nerve
paralysis as intubation injury? Chirurg, 71: 539544.
Gavilan, J and Gavilan, C (1986) Recurrent laryngeal
nerve. Identification during thyroid and parathyroid surgery. Arch. Otolaryngol. Head Neck Surg., 112:
12861288.
Gockel, I, Kneist, W, Keilmann, A and Junginger, T (2005)
Recurrent laryngeal nerve paralysis (RLNP) following
esophagectomy for carcinoma. Eur. J. Surg. Oncol.,
31: 277281.
Goins, MR and Pitovski, DZ (2004) Posttonsillectomy taste
distortion: a significant complication. Laryngoscope,
114: 12061213.
Hamelmann, WH, Meyer, T, Timm, S and Timmermann,
W (2002) A critical estimation of intraoperative neuromonitoring (IONM) in thyroid surgery. Zentralbl. Chir.,
127: 409413.
Hemmerling, TM and Donati, F (2003) Neuromuscular
blockade at the larynx, the diaphragm and the corrugator supercilii muscle: a review. Can. J. Anaesth., 50:
779794.
Hemmerling, TM, Schurr, C, Dern, S, Schmidt, J, Braun, GG
and Klein, P (2000) Intraoperative electromyographic
605
recurrent laryngeal nerve identification as a routine measure. Chirurg, 71: 545550.
Hermann, M, Alk, G, Roka, R, Glaser, K and Freissmuth,
M (2002) Laryngeal recurrent nerve injury in surgery
for benign thyroid diseases: effect of nerve dissection
and impact of individual surgeon in more than 27,000
nerves at risk. Ann. Surg., 235: 261268.
Hermann, M, Hellebart, C and Freissmuth, M (2004) Neuromonitoring in thyroid surgery: prospective evaluation
of intraoperative electrophysiological responses for the
prediction of recurrent laryngeal nerve injury. Ann.
Surg., 240: 917.
Hobbiger, HE, Allen, JG, Greatorex, RG and Denny, NM
(1996) The laryngeal mask airway for thyroid and parathyroid surgery. Anaesthesia, 51: 972974.
James, AG, Crocker, S, Woltering, E, Ferrara, J and Farrar,
W (1985) A simple method for identifying and testing
the recurrent laryngeal nerve. Surg. Gynecol. Obstet.,
161: 185186.
Jatzko, GR, Lisborg, PH, Muller, MG and Wette, VM
(1994) Recurrent nerve palsy after thyroid operations
principal nerve identification and a literature review.
Surgery, 115: 139144.
Jellish, WS, Jensen, RL, Anderson, DE and Shea, JF (1999)
Intraoperative electromyographic assessment of recurrent laryngeal nerve stress and pharyngeal injury during
anterior cervical spine surgery with Caspar instrumentation. J. Neurosurg., 91: 170174.
Jonas, J and Bahr, R (2000) Intraoperative electromyographic identification of the recurrent laryngeal nerve.
Chirurg, 71: 534538.
Khan, A, Pearlman, RC, Bianchi, DA and Hauck, KW
(1997) Experience with two types of electromyography
monitoring electrodes during thyroid surgery. Am. J.
Otolaryngol., 18: 99102.
Krasna, MJ and Forti, G (2006) Nerve injury: injury to the
recurrent laryngeal, phrenic, vagus, long thoracic, and
sympathetic nerves during thoracic surgery. Thorac.
Surg. Clin., 16: 267275, vi.
Kunath, M, Marusch, F, Horschig, P and Gastinger, I
(2003) The value of intraoperative neuromonitoring in
thyroid surgery a prospective observational study
with 926 patients. Zentralbl. Chir., 128: 187190.
Lahey, FH (1938) Routine dissection and demonstration of
the recurrent laryngeal nerve in subtotal thyroidectomy.
Surg. Gynecol. Obstet., 66: 775777.
Lamade, W, Meyding-Lamade, U, Buchhold, C, Brauer, M,
Brandner, R, Uttenweiler, V, Motsch, J, Klar, E and
Herfarth, C (2000) First continuous nerve monitoring
in thyroid gland surgery. Chirurg, 71: 551557.
Large, SR, Heywood, LJ, Flower, CD, Cory-Pearce, R,
Wallwork, J and English, TA (1985) Incidence and
aetiology of a raised hemidiaphragm after cardiopulmonary bypass. Thorax, 40: 444447.
606
Leonetti, JP, Jellish, WS, Warf, P and Hudson, E (1996)
Intraoperative vagal nerve monitoring. Ear Nose Throat
J., 75: 489491, 495496.
Lipton, RJ, McCaffrey, TV and Litchy, WJ (1988) Intraoperative electrophysiologic monitoring of laryngeal muscle
during thyroid surgery. Laryngoscope, 98: 12921296.
Marcus, B, Edwards, B, Yoo, S, Byrne, A, Gupta, A, Kandrevas, J, Bradford, C, Chepeha, DB and Teknos, TN
(2003) Recurrent laryngeal nerve monitoring in thyroid
and parathyroid surgery: the University of Michigan
experience. Laryngoscope, 113: 356361.
Marusch, F, Hussock, J, Haring, G, Hachenberg, T and
Gastinger, I (2005) Influence of muscle relaxation on
neuromonitoring of the recurrent laryngeal nerve during
thyroid surgery. Br. J. Anaesth., 94: 596600.
Mom, T, Filaire, M, Advenier, D, Guichard, C, Naamee, A,
Escande, G, Llompart, X, Vallet, L, Gabrillargues, J,
Courtalhiac, C, Claise, B and Gilain, L (2001) Concomitant type I thyroplasty and thoracic operations for lung
cancer: preventing respiratory complications associated
with vagus or recurrent laryngeal nerve injury.
J. Thorac. Cardiovasc. Surg., 121: 642648.
Odegard, KC, Kirse, DJ, Del Nido, PJ, Laussen, PC, Casta,
A, Booke, J, Kenna, MA and McGowan, FX, Jr. (2000)
Intraoperative recurrent laryngeal nerve monitoring during video-assisted throracoscopic surgery for patent
ductus arteriosus. J. Cardiothorac. Vasc. Anesth., 14:
562564.
Otto, RA and Cochran, CS (2002) Sensitivity and specificity of intraoperative recurrent laryngeal nerve stimulation in predicting postoperative nerve paralysis. Ann.
Otol. Rhinol. Laryngol., 111: 10051007.
Pearlman, RC, Isley, MR, Ruben, GD, Sandler, SC, Weisbaum, B, Khan, MA, Greene, BS, Charles, V and Shah,
A (2005) Intraoperative monitoring of the recurrent
laryngeal nerve using acoustic, free-run, and evoked
electromyography. J. Clin. Neurophysiol., 22: 148152.
Randolph, GW, Kobler, JB and Wilkins, J (2004) Recurrent
laryngeal nerve identification and assessment during thyroid
surgery: laryngeal palpation. World J. Surg., 28: 755760.
T.M. HEMMERLING
Rea, JL and Khan, A (1998) Clinical evoked electromyography for recurrent laryngeal nerve preservation: use
of an endotracheal tube electrode and a postcricoid
surface electrode. Laryngoscope, 108: 14181420.
Rice, DH, Wetmore, SJ and Singer, M (1991) Recurrent
squamous cell carcinoma of the true vocal cord. Head
Neck, 13: 549552.
Robinson, DN, OBrien, K, Kumar, R and Morton, NS
(1998) Tracheal intubation without neuromuscular
blockade in children: a comparison of propofol combined either with alfentanil or remifentanil. Paediatr.
Anaesth., 8: 467471.
Rogers, W and Root, HD (1988) Cranial nerve injuries
after carotid artery endarterectomy. South. Med. J., 81:
10061009.
Tanigawa, K, Inoue, Y and Iwata, S (1991) Protection of
recurrent laryngeal nerve during neck surgery: a new
combination of neutracer, laryngeal mask airway, and
fiberoptic bronchoscope. Anesthesiology, 74: 966967.
Timmermann, W, Hamelmann, WH, Thomusch, O,
Sekulla, C, Grond, S, Neumann, HJ, Kruse, E, Muhlig,
HP, Richter, C, Voss, J and Dralle, H (2004) Effectiveness and results of intraoperative neuromonitoring in
thyroid surgery. Statement of the Interdisciplinary
Study Group on Intraoperative Neuromonitoring of
Thyroid Surgery. Chirurg, 75: 916922.
Tschopp, K and Probst, R (1994) New aspects in surgery of
the thyroid gland with intraoperative monitoring of the
recurrent laryngeal nerve. Laryngorhinootologie, 73:
568772.
Tschopp, KP and Gottardo, C (2002) Comparison of various methods of electromyographic monitoring of the
recurrent laryngeal nerve in thyroid surgery. Ann. Otol.
Vories, AA (1999) Dysgeusia associated with tonsillectomy. Otolaryngol. Head Neck Surg., 121: 303304.
Woltering, EA, Dumond, D, Ferrara, J, Farrar, WB and
James, AG (1984) A method for intraoperative identification of the recurrent laryngeal nerve. Am. J. Surg.,
148: 438440.
Section III.2
Spine Surgery

M.R. Nuwer (Ed.)
608
CHAPTER 43
Monitoring during the surgical treatment of scoliosis

Marc R. Nuwera,b,* and James W. Packwoodb
a
Department of Neurology, UCLA School of Medicine, Los Angeles, CA 90095, USA
Department of Clinical Neurophysiology, UCLA Medical Center, Los Angeles, CA 90095, USA
43.1. Scoliosis
Scoliosis is a relatively common problem in the general
community. A variety of types of scoliosis are recognized. These include idiopathic scoliosis (Fig. 1), which
can present at adult, adolescent, or congenital portions
of the life spectrum and degenerative scoliosis which
is common among the elderly. Other secondary causes
of scoliosis include many neuromuscular disorders.
Idiopathic scoliosis occurs in 24% of children older
than 10 years (Roach, 1999). The Scoliosis Research
Societys formal definition is a lateral spinal curvature
>10% accompanied by vertebral rotation. While the
exact cause is unclear, it is thought to be a dominant
multi-gene inherited predisposition with variable
expression. When both parents have scoliosis, a child
has a 50-fold excess risk for developing scoliosis. Only
about 10% of adolescents with scoliosis require medication, bracing, or surgery. The biggest risk factors
for progression are female gender, a large curve magnitude and skeletal deformity. Several orthopedic spinal
rating scales are used to predict the likelihood and magnitude of future progression. The degree of curvature is
assessed on radiographs using a method described by
Cobb, which measures the angle between lines perpendicular to the intervertebral disk space above and below
the curve (Fig. 2). Adams described the most common
screening test in which a patient stands and bends forward. The physician looks for a rib hump on one side.
Many authors suggest surgery when the degree of
curvature exceeds 50 , or 40 in a skeletally immature
patient or a patient who is progressing despite bracing
(Bridwell, 1999). Above that degree, untreated scoliosis
leads to further progression, pain, reduced pulmonary
function, and psychosocial effects (Nilsonne and
*
Correspondence to: Marc R. Nuwer, M.D., Ph.D., UCLA

Department of Neurology, Reed Bldg, Room 1190, 710
Westwood Plaza, Los Angeles, CA 90095, USA.
Tel.: 1-310-206-3093; fax: 1-310-267-1157.
E-mail: MRN@UCLA.edu (M.R. Nuwer).
Lundgren, 1968; Fowles et al., 1978; Nachemson,

1979; Pehrsson et al., 1992; Lonstein, 2006). For adolescent idiopathic scoliosis, progression may decrease
or stop as the girl passes through menarche or completes her adolescence, times when metabolism alters
and normal bone growth stops. In younger patients,
measurements of the individuals progression can help
determine who needs surgery and when. A combined
anterior and posterior approach is recommended when
the patients curvature exceeds 100 or for a hyperkyphotic curve >70 .
Scoliosis also can be caused by neuromuscular disorders, cerebral palsy, spinal tumors and dysraphisms, and
other congenital spinal anomalies. The neuromuscular
disorders associated with increased risk of scoliosis
include some of the muscular dystrophies, poliomyelitis
and some other anterior horn cell disorders, and spinocerebellar degenerative disorders. Some such disorders
make surgical neurophysiologic monitoring difficult,
because they also impair conduction along the same
pathways monitored. In those patients, even light anesthesia can abolish evoked potentials because of a synergistic adverse effect of the disorder and the drug effects.
The earlier that curvature is seen among these neuromuscular patients, the more likely that they will experience progression to severe scoliosis (James, 1956;
Garrett et al., 1961). Progression is seen even after skeletal maturity. In one study the curve progressed at a rate
of 1.4 annually, for curves >50 (Thometz and Simon,
1988). Wheelchair bound muscular dystrophy patients
can progress at 10 per year. The Duchennes muscular
dystrophy patients often also need fusion to the pelvis
for the correction of pelvic obliquity. For that reason,
25 curvature has been recommended as a criterion for
surgery (Cambridge and Drennan, 1987; Miller et al.,
1988), a lesser degree cut off than used for idiopathic
patients. Surgical therapy carries increased risk for
many neuromuscular scoliosis patients. The risk is due
to their generally compromised health for respiratory
function, participation in rehabilitation, seizure disorders, predisposition to urosepsis, and poor bone quality.
SPINE SURGERY
609
Fig. 1. Scoliosis patient. (From UCLA Clinical Neurophysiology Lab, with permission.)
Surgery often takes longer in these patients. Some neuromuscular patients are more predisposed to malignant
hyperthermia. Spastic patients do better with combined
anterior and posterior correction.
Neurological complications occur in 0.517% of
patients undergoing spinal corrective surgery for neuromuscular scoliosis (Herring and Wenger, 1982;
Weimann et al., 1983; Allen and Ferguson, 1986).
The specific risks depend on the degree of deformity,
amount of correction, type of instrumentation, and
whether the approach is anterior, posterior, or both.
Cerebral palsy patients with scoliosis often progress to a point where surgery is required. Kyphosis
is seen among these patients due to poor sitting posture. Nonsurgical approaches are clearly preferable
for correcting sitting posture. Surgery ought to be
reserved for patients older than 10 years, who failed
conservative management of sitting, with good nutrition and other factors (Banta et al., 1998).
Degenerative scoliosis is seen among 6% of older
patients, with an average onset in the seventh decade
(Vanderpool et al., 1969; Pritchett and Bortel, 1993).
As the percentage of the population over the age of
70 is increasing, the incidence of this problem continues to grow. Factors predicting progression include
an angle of 30 or more, lateral vertebral translations
of 6 mm or more and apical rotation of moderate or
severe degree (Pritchett and Bortel, 1993). The
Fig. 2. Determination of the Cobb angle from a radiograph.

Angles are measured from the intersections of lines drawn
perpendicular to the disk spaces above and below the curvature. (From Reamy and Slakey, 2001, with permission.)
deformity is most typical at the lumbar or thoracolumbar level. The patients often have a sagittal imbalance
with leaning forward, possibly with flexed hips and
knees, to compensate for the loss of lumbar lordosis
or even a frank kyphosis (Gupta, 2003).
Decompression alone is the surgical procedure
preferred for 10 30 of curvature in degenerative
scoliosis. Posterior fusion with segmental instrumentation is added for patients with 25 40 curvature.
Greater degrees are usually treated with both anterior
and posterior fusion.
43.2. Surgical and intraoperative monitoring
techniques
The traditional surgical spinal correction involves
partial straightening of the curvature and placement
of mechanical supports to keep the spine straighter.
The process of straightening the spinal column by
reducing the curvature is referred to as distraction.
Placement of mechanical supports risks spinal cord
610
damage. Many injuries arise from mechanically

stretching the spinal cord or from compressing vital
arteries. Occasionally the spinal cord can be damaged
from passing or inserting instrumentation through or
into the spinal canal, especially when the spinal cord
lies asymmetrically against the wall of the spinal
canal directly under the instrumentation site.
For many years, Harrington rods were used. Those
single or double rods were secured by hooks around
pedicles, and occasionally further secured by Luque
wires passed below lamina at several levels. Problems
with Harrington rods instrumentation included a
kyphosing effect on the lumbar spine and little control
in a sagittal plane (Bridwell, 1999). It also relied on
postoperative casting or bracing.
More recent techniques involve segmental instrumentation that secure hardware to pedicles, e.g., with
screws. The popular CotrellDubousset (CD) system

is the first complex posterior double rod instrumentation system to provide multiple hook fixation (Remes
et al., 2004). It provides for bilateral segmental
fixation of the spine, and selective distraction and
compression at different levels. This offers a way to
improve alignment of the spine in the coronal and sagittal planes. Control in an axial plane is controversial.
Alternatives include the Universal Spine System
(USS), which allows for rod translation instead of
rod rotation, the option to secure pedicle hooks with
fixation screws or use of transpedicular screws in the
lower thoracic and lumbar spine. Supplemental segmental wires still may be used. The placement of
transpedicular screws greatly improves secure hardware placement. Laminar hooks remain popular at
a thoracic level (Fig. 3).
(Fig. 3 continued)
SPINE SURGERY
611
Fig. 3. PA (A, C) and lateral (B, D) radiographs of a 35-year-old scoliosis patient before (A, B) and after (C, D) placement
of instrumentation. Posterior fusion resulted in an excellent correct with stability and excellent arthrodesis. (From Bradford
et al., 1999, with permission from Lippincott, Williams and Wilkins.)
An anterior approach can be used allowing for a

greater degree of distraction and correction. The
anterior approach is not usually needed for routine
idiopathic scoliosis. It is used more for severe, neuromuscular, or other causes of scoliosis (Bridwell,
1999). The anterior approach includes release of tight
anterior ligaments, removing the intervertebral disk,

and providing for a spinal fusion. Discectomy and
fusion with grafting are performed in some cases to
add a degree of long-term stability. Occasionally,
the pelvis is included in the instrumentation such as
in patients with pain due to excess lumbar curvature
612
or degenerative disease of the lower lumbar spine. In

the latter cases, sacral or iliac crest screws may be
used. Recovery takes 612 months.
43.3. Neurological complications of scoliosis
surgery
Neurological complications occur in 15% of cases.
Risk factors include combined anterior and posterior
surgery, severe rigid curves and kyphosis. Occasional
radiculopathy occurs when roots are damaged as a
result of the placement of hardware. Feared complications are paraparesis and paraplegia.
Mechanical injury to the spinal cord can happen in
several ways. Simple trauma or external cord compression can occur with some hardware passed through the
epidural space. In a more normal spinal anatomy, the
cord lies in the middle of the canal. In the scoliosis
patient, the cord often lies along the concave side of
the canal. In that case, the spinal cord takes a short
cut along that concave edge, and it would not be as
long as a comparable spinal cord in a normal person.
When the spinal column is partially straightened, the
relatively shorter spinal cord has difficulty accommodating itself in the relatively longer spinal column. It
is stretched. Modern segmental instrumentation has
less of a problem than Harrington rods.
Ischemic spinal cord injury is a devastating complication, especially when it involves an anterior spinal
cord thrombosis. That produces a central cord syndrome. Some such injuries traverse the entire cord
width, producing paraplegia. Other injuries are partial,
with damage especially to deeper cord tracts including
the lateral corticospinal motor pathway. It may relatively spare the posterior columns and anterior corticospinal tracts, which may be impaired in function
only in the initial days or hours. The mid-thoracic level
is the most common level for such impairment,
because that is a watershed vascular region lying halfway between the rostral and caudal feeding arteries.
Ischemic cord injury can have a variety of
contributing factors. One commonly cited cause is
compression of the major lumbar radicular artery,
known as the artery of Adamkiewicz. That is the
major caudal feeding vessel for cord perfusion. It lies
at an upper lumbar level. It can be compressed
mechanically by placement of hardware or by spinal
column straightening and rotation. Another possibility is compression or stretching of the anterior spinal
artery itself. Finally, the mechanical spinal column
rotations and straightening can compress minor
radicular feeding arteries that occur at each spinal

level. The exact cause of ischemic injury is difficult
to determine for individual patients. It is a
recognized and feared complication of scoliosis
correction surgery.
Ischemic damage may be more likely when blood
pressures are low because a partially compressed
vessel is more at risk for thrombosis. An interesting
additional factor toward ischemic complications is
atherosclerosis. Myers reported 37 cases of visual
loss after spinal scoliosis correction surgery that
occurred from ischemic optic neuropathy, retinal
artery occlusion, or cerebral ischemia (Myers et al.,
1997).
Delayed onset paraparesis or paraplegia occurs in
a small portion of patients during the hours or days
after scoliosis correction surgery. There are several
theories about the underlying etiology of these
injuries. One theory is the eventual decompensation
of a stretched spinal cord or vessel, which goes on
to damage axons mechanically or thrombose the
compromised vessel. Another is the relatively hypercoagulable state immediately after surgery, which
can lead preferentially to thrombosis of inflamed
or compressed vessels. Other known coagulation
complications are pulmonary emboli that occur in
up to 20% of patients, a rate that varies among
clinical series (Cain et al., 1995; Wood et al., 1997;
Dearborn et al., 1999). Good postoperative volume
support and clinical monitoring are considered
highly desirable to reduce the contribution of hypovolemia and hypotension to postoperative thromboses (Bradford et al., 1999).
43.4. Intraoperative monitoring during scoliosis
surgery
Monitoring the spinal cord has become a widely used
tool to safeguard neurological function during scoliosis correction. Several neurophysiologic techniques
are applied, especially various types of somatosensory evoked potentials (SEP) and motor evoked
potentials (MEP). The specific techniques themselves
are described in other chapters of this volume. SEP
and MEP monitor the whole length of the spinal
cord.
When potentials drop out or drop in amplitude
substantially, several factors need to be assessed rapidly. First, is there a technical problem (e.g., a loose
electrode)? Impedance testing, review of unaveraged
incoming signals, and the presence of a stimulus
SPINE SURGERY
artifact offers some reassurance of the systems

integrity. Second, is there an important systemic factor such as hypotension or change in the anesthetic
regimen? Third, was there a surgical intervention
recently that could have caused risk? The suspect
intervention may have occurred many minutes previously, not just during that tracings acquisition.
If the critical window for intervention is just
15 min, then the monitoring and surgical teams
should act within that window to identify and correct
any problems and to optimize clinical factors such as
blood pressure. After that time window, the risk of
long-term injury climbs. A complete loss of signals
with complete recovery within 15 min is of concern,
but usually is associated with no postoperative
deficits. This rule applies especially when both SEP
and MEP are monitored. On the other hand, very
slow partial recovery over hours has a much higher
risk of postoperative neurological deficits. Of
course, a complete loss with no recovery has a high
degree of prediction of postoperative neurological
deficits.
Electromyography monitors for any signs of acute
root injury, as may occur with hardware around
roots. H-reflexes test both peripheral S1 level sensory
and motor pathways along with the lumbar spinal
cord. Some users evaluate these peripheral modalities
for neurotonic discharges or dropout of reflexes that
might be caused by impending central injury.
What can the surgeon or anesthesiologist do if
the potentials change? There are several commonly
cited opportunities for intervention. The first is to
reduce the distraction, straightening or rotation
applied to the spinal column. Another option is
to remove wires, screws, or grafts that might have
been associated with deterioration. Sometimes the
solution could be to adjust retractors that might compress vessels or injure roots. Increasing the blood
pressure is often used, since some causes of injury
are substantially worsened by hypotension. In the
case where segmental arteries have been sacrificed,
one could shunt or reimplant vessels. Visual inspection can be helpful to identify a hemotoma, though
an imaging test may be needed immediately postoperatively for that. Adjusting a patients position on
the table is traditionally considered in cervical surgery, and might occasionally be appropriate in more
mid or caudal spinal surgery. Finally, the surgeon
may have the option to stop the procedure. If the
changes are associated with anesthetic effects, consideration should be given to re-establishing an
613
anesthetic regimen compatible with or more ideal

for monitoring.
The old version of intraoperative assessment,
Stagnaras wake-up test (Vauzelle et al., 1973), is
occasionally carried out to verify the clinical alarm
triggered by SEP or MEP changes. It is also done
when traditional neurophysiologic monitoring cannot
be accomplished. In this procedure, the patient is
awakened while still on the operating room table,
usually with the back still open and the patient intubated. Narcotics are used to relieve pain. When sufficiently awake, the patient is instructed loudly to
move each extremity, one at a time. If done well, this
proves the point that each extremitys power is still
preserved and the spinal cord must be grossly intact.
Problems can occur if the patient is too confused to
cooperate, or if the patient becomes agitated. In the
latter case, the patient might move on the table, extubate himself, knock out intravenous lines, or cause
other difficulties.
43.5. Outcome studies
An extensive literature exists about outcomes after
monitoring during scoliosis correction surgery and
related orthopedic procedures. Some salient, welldesigned studies are reviewed here.
Pelosi and colleagues (Pelosi et al., 2002) looked
at 126 operations conducted on 97 patients, mostly
for spinal deformities. They used posterior tibial
stimulation with recording from the scalp and neck,
as well as transcranial electrical MEPs with D-wave
and M-wave monitoring. Among their patients, they
had 16 cases in which they intervened or conducted
a wake-up test. In another eight cases, they reduced
the degree of correction. In six cases they gave fluids
and increased blood pressure. In two cases, they
removed instrumentation. Among their patients, they
had two cases of postoperative paraplegia, one of
monoplegia and the other of monoparesis. Overall,
they predicted all deficits in monitored patients. Six
SEP alarms and 10 MEP alarms, some in the same
cases, triggered interventions that led to normal neurological outcomes. They could perform only the
SEP technique in 18 surgeries. They could carry out
only the MEP technique in two patients. The loss of
both modalities more strongly urged prompt intervention for these surgeons. They concluded that
both monitoring modalities were useful in preventing
postoperative neurological deficits, and they too were
complementary.
614
Meyer and colleagues (Meyer et al., 1988) looked

at 150 monitored cases and 145 unmonitored surgical
procedures for spinal deformities. They used lower
extremity stimulation and scalp recorded SEPs.
MEPs were not used. They had one bad neurological
outcome among the monitored case, compared to 10
among the unmonitored cases. Instances of intervention from monitoring alarms included removal of
wires in one patient, removal of a rod in four
patients, and to cease further manipulation. They
concluded that monitoring substantially reduced the
number of postoperative neurological deficits.
Mostegl and colleagues (Mostegl et al., 1988)
looked at 127 operations in 97 patients with spinal
deformities. They monitored with posterior tibial
stimulation and scalp recorded SEP. MEPs were not
used. They had one case of postoperative paraplegia
and two of transient paraparesis. All postoperative
deficits were predicted by SEP changes. In two further cases, SEP alarms triggered interventions in
patients who went on to have no postoperative deficit. Interventions included raising the blood pressure
and stopping early.
Numerous other studies also have evaluated SEP
and MEP use for scoliosis correction, and found the
techniques useful to identify high-risk situations and
thereby to initiate prompt intervention to prevent
adverse outcomes.
43.6. Multicenter survey of spinal cord monitoring
outcome
A multicenter survey of spinal cord monitoring outcome assessed the key clinical issue about whether
monitoring reduces adverse outcomes (Nuwer et al.,
1995). The Scoliosis Research Society (SRS), a
group composed of surgeons with a special interest
in scoliosis, was used for the survey. Most SRS
members practice in the United States. Each member
must submit, annually, scoliosis surgical outcomes as
a condition of continuing membership. For that reason, the SRS is a valuable resource for studying
outcomes. SRS has repeatedly assessed neurological
deficits after scoliosis surgery. For the multicenter
outcome survey, SRS members and their clinical
neurophysiologist were questioned about their neurophysiologic monitoring. The 173 U.S. surgeons had
conducted 97,586 spinal surgery cases, among which
51,263 (53%) were done using SEP monitoring. Most
used posterior tibial stimulation with scalp and neck
recording. Common alarm criteria were a 50%

decreased amplitude or a 710% latency increase.
False negative outcomes are those in which the
SEP remained stable throughout the case but the
patient awoke with a new postoperative neurological
deficit. Among the 51,263 monitored patients, 65
(0.127%) had a false negative outcome. Among
those, 34 were considered definite adverse outcomes,
13 were equivocal (such as root injuries, minor, or
very transient deficits) and 18 were delayed onset
deficits. Some were reported separately in the literature on false negative SEP monitoring (Ginzburg
et al., 1985; Lesser et al., 1986; Ben-David et al.,
1987; Harper et al., 1988; More et al., 1988). Outcome rates are presented in Table 1.
True positive outcomes are those in which the
SEP correctly predicted a new postoperative deficit.
The surgical teams presumably attempted to intervene either unsuccessfully or with only a partial
success at mitigating but not eliminating the entire
deficit. Among the 51,263 monitored patients, 217
patients (0.42%) had true positive monitoring
outcome.
The neurological deficits rate is the sum of the false
negative and true positive rates. Among the 51,263
monitored patients, 282 (0.55%) had new postoperative neurological deficits. Of that, the rate of transient
deficits was 0.24%. The rate at which persistent deficits occurred was 0.31%. The majority of new deficits
were characterized as minor, such as radiculopathy,
sensory impairment without motor loss, postoperative
Table 1
Neurological outcome for somatosensory evoked
potentials (SEP) monitoring in scoliosis surgery
Total procedures
False negative (FN)
False positive
True positive (TP)
Neurological deficits
(FN plus TP)
True negative
Sensitivity
Specificity
51,263
65
774
217
282
(100%)
(0.127%)
(1.510%)
(0.423%)
(0.550%)
50,207
(97.94%)
92%
98.9%
99.93%
42%
Source: From the multicenter outcome survey for spinal cord monitoring (Nuwer et al., 1995).
SPINE SURGERY
615
Table 2
Comparison of neurological outcomes with or without
monitoring in scoliosis surgery
Postoperative neurological
deficits
Without
With
monitoring monitoring
(%)
(%)
All neurological deficits
0.55
Persistent neurological deficits 0.31
Major neurological deficits
0.24
0.72
0.46
0.61
Source: From the multicenter outcome survey for spinal cord monitoring (Nuwer et al., 1995).
bladder control problems, or other lesser degrees of

neurological deficits. The rate of major deficits (i.e.,
paraparesis or paraplegia) was 0.24%.
Persistent and major neurological deficits were
substantially less when monitoring was used. Table 2
compares deficits with and without monitoring. This
shows that persistent deficits dropped by one third
and major deficits dropped by about 60% among
the patients operated on with neurophysiologic
monitoring.
Geographical comparisons and comparisons
among different monitoring techniques failed to
show significantly better or worse outcomes for those
factors (Nuwer and Carlson, 1994). This suggests
that SEP monitoring can be done in a variety of
ways, and these various technical approaches are all
reasonable, satisfactory, and similarly accurate in
predicting neurological outcome and performing the
monitoring task.
43.7. Limitations and problems with monitoring
Does monitoring predict all deficits? No, monitoring
may miss some significant deficits. Hopefully, nearly
all major persistent deficits are recognized. Some
root injuries would be missed when the spinal cord
itself is monitored. Some delayed deficits have clinical onset hours after surgery, and monitoring only
determines the state of the spinal cord at the moment
of monitoring. A few major deficits may still be
undetected despite the best available monitoring.
More to the point, though, monitoring warns of many
neurological deficits that occur despite interventions
in response to the warning. This may occur because

there was nothing that could be done upon warning,
for example with an acute anterior spinal artery
thrombosis. Or, some deficits may occur despite
interventions that did mitigate, but did not eliminate the deficit. Surgeons and monitoring teams
must be realistic about the continuing possibility
of some postoperative neurological deficits despite
monitoring.
In some cases monitoring raises false alarms. In
the survey, this occurred in 12% of monitored cases.
Some of those were true positive cases where deficits
were totally prevented. Others were really false
alarms due to technical causes. Those remain an
annoyance to surgical procedures.
Problems with transcranial electrical MEP include
the need for total intravenous anesthetic, or at least
only small amounts of inhalation anesthetic. Some
anesthesiologists feel uncomfortable with those
restrictions, and in some patients higher levels of
inhalation anesthetic must be used. In those cases,
MEP may be impractical. Likewise, transcranial
electrical MEP requires reduced, controlled amounts
of neuromuscular junction blockade. Occasionally,
the patient might move on the table. At the time of
stimulation, the movements are noticeable but cannot
be tolerated during delicate interventions. So, MEPs
are often carried out only intermittently. MEP recording at caudal spinal level or cauda equina attempts to
get around this problem.
43.8. Conclusions
A variety of techniques are available to monitor the
spinal cord during scoliosis surgery. They offer the
opportunity of alarms for high-risk situations. Scoliosis correction surgery has an inherent risk of spinal
cord injury. Monitoring is associated with substantial
reductions in postoperative neurological deficits in
scoliosis surgery, though it does not prevent all
deficits. When monitoring does raise an alarm, a
variety of possible interventions are available for
consideration.
References
Allen, B and Ferguson, R (1986) Neurologic injuries with
the Galveston technique of L-rod instrumentation for
scoliosis. Spine, 11: 1417.
616
Banta, JV, Lubicky, JP and Lonstein, JE (1998) Resolution:
a 15 year old wit spastic quadriplegia and a 60 degree
scoliosis should have a posterior spinal fusion with
instrumentation. Dev. Med. Child Neurol., 40: 278283.
Ben-David, B, Haller, GS and Taylor, P (1987) Anterior
spinal fusion complicated by paraplegia: a case report
of a false-negative somatosensory evoked potential.
Spine, 12: 536539.
Bradford, DS, Tay, BKB and Hu, SS (1999) Adult scoliosis: surgical indications, operative management, complications, and outcomes. Spine, 24: 26172629.
Bridwell, KH (1999) Surgical treatment of idiopathic adolescent scoliosis. Spine, 24: 26072616.
Cambridge, W and Drennan, J (1987) Scoliosis associated
with Duchenne muscular dystrophy. J. Pediatr. Orthop.,
7: 436440.
Cain, JE, Jr., Major, MR, Lauerman, WC, West, JL, Wood,
KB and Fueredi, GA (1995) The morbidity of heparin
therapy after development of pulmonary embolus in
patients undergoing thoracolumbar or lumbar spinal
fusion. Spine, 20: 16001603.
Dearborn, J, Hu, S, Tribus, C and Bradford, D (1999)
Thromboembolic complications following major thorscolumbar spine surgery. Spine, 24: 14711476.
Fowles, JV, Drummond, DS, LEcuyer, S, Roy, L and Kassab, MT (1978) Untreated scoliosis in the adult. Clin.
Garrett, A, Perry, J and Nickel, V (1961) Paralytic scoliosis. Clin. Orthop., 21: 117124.
Ginzburg, HH, Shetter, AG and Raudzens, PA (1985) Postoperative paraplegia with preserved intraoperative
somatosensory evoked potentials. J. Neurosurg., 63:
296300.
Gupta, MC (2003) Degenerative scoliosis options for surgical management. Orthop. Clin. North Am., 34: 269279.
Harper, CM, Jr., Daube, JP, Litchy, WJ and Klassen, RA
(1988) Lumbar radiculopathy after spinal fusion for scoliosis. Muscle Nerve, 11: 386391.
Herring, J and Wenger, D (1982) Segmental spinal instrumentation: a preliminary report of 40 consecutive cases.
Spine, 7: 285298.
James, JI (1956) Paralytic scoliosis. J. Bone Joint Surg. Br.,
38: 660685.
Lesser, RP, Raudzens, P, Luders, H, Nuwer, MR, Goldie,
WD, Morris, HH, III, Dinner, DS, Klem, G, Hahn, JF,
Shetter, AG, Ginsburg, HH and Gurd, AR (1986) Postoperative neurological deficits may occur despite
unchanged intraoperative somatosensory evoked potentials. Ann. Neurol., 19: 2225.
Lonstein, JE (2006) Scoliosis: surgical versus nonsurgical
treatment. Clin. Orthop. Relat. Res., 443: 248259.
Meyer, PR, Jr., Colter, HB and Gireesan, GT (1988) Operative neurological complications resulting from thoracic
and lumbar spine internal fixation. Clin. Orthop. Rel.
Res., 237: 125131.

Miller, F, Moseley, C, Koreska, P and Levison, H (1988) Pulmonary function and scoliosis in Duchenne dystrophy. J.
Pediatr. Orthop., 8: 133137.
More, RC, Nuwer, MR and Dawson, EG (1988) Cortical
evoked potential monitoring during spinal surgery: sensitivity, specificity, reliability, and criteria for alarm. J.
Spinal Disord., 1: 7580.
Mostegl, A, Bauer, R and Eichenauer, M (1988) Intraoperative somatosensory potential monitoring. A clinical analysis of 127 surgical procedures. Spine, 13:
396400.
Myers, MA, Hamilton, SR, Bogosian, AJ, Smith, CH and
Wagner, TA (1997) Visual loss as a complication of spine
surgery: a review of 37 cases. Spine, 22: 13251329.
Nachemson, A (1979) Adult scoliosis and back pain. Spine,
4: 513517.
Nilsonne, U and Lundgren, KD (1968) Long term prognosis in idiopathic scolisosis. Acta Orthop. Scand., 39:
456465.
Nuwer, MR and Carlson, LG (1994) A multicentre survey
of spinal cord monitoring outcome. In: SJ Jones, S
Boyd, M Hetreed and NJ Smith (Eds.), Handbook of
spinal cord monitoring. Kluwer, London, pp. 7887.
Nuwer, MR, Dawson, EG, Carlson, LG, Kanim, LE and
spinal cord monitoring reduces neurologic deficits
after scoliosis surgery: results of a large multicenter
survey. Electroencephalogr. Clin. Neurophysiol., 96:
611.
Pehrsson, K, Larsson, S, Oden, A and Nachemson, A
(1992) Long-term follow-up of patients with untreated
scoliosis: a study of mortality, causes of death, and
symptoms. Spine, 17: 10911096.
Pelosi, L, Lamb, J, Grevitt, M, Mehdian, SMH, Webb, JK
and Blumhardt, LD (2002) Combined monitoring
of motor and somatosensory evoded potentials in
orthopaedic spinal surgery. Clin. Neurophysiol., 113:
10821091.
Pritchett, JW and Bortel, DT (1993) Degenerative symptomatic lumbar scoliosis. Spine, 18: 700703.
Reamy, BV and Slakey, JB (2001) Adolescent idiopathic
scoliosis: review and currecnt concepts. Am. Fam.
Phy., 64: 111116.
Remes, V, Helenius, I, Schlenzka, D, Yrjonen, T, Ylikoski,
M and Poussa, M (2004) Cotrel-Dubousset (CD) or Universal Spine System (USS) instrumentation in adolescent idiopathic scoliosis (AIS): comparison of midterm
clinical, functional, and radiologic outcomes. Spine,
29: 20242030.
Roach, JW (1999) Adolescent idiopathic scoliosis. Orthop.
Clin. North Am., 30: 353365.
Thometz, JG and Simon, SR (1988) Progression of scoliosis after skeletal maturity in institutionalized adults who
have cerebral palsy. J. Bone Joint Surg. Am., 70:
12901296.
SPINE SURGERY
Weimann, RL, Gibson, DA, Moseley, CF and Jones, DC
(1983) Surgical stabilization of the spine in Duchenne
muscular dystrophy. Spine, 8: 776780.
Wood, KB, Kos, PB, Abnet, JK and Ista, C (1997) Prevention of deep-vein thrombosis after major spinal surgery:
a comparison study of external devices. J. Spinal Disord., 10: 209214.
617
Vanderpool, DW, James, JI and Wynne-Davies, R (1969)
Scoliosis in the elderly. J. Bone Joint Surg. Am., 51:
446455.
Vauzelle, C, Stagnara, P and Jouvinroux, P (1973)
Functional monitoring of spinal cord activity
during spinal surgery. Clin. Orthop. Relat. Res., 93:
173178.

M.R. Nuwer (Ed.)
618
CHAPTER 44
Monitoring during spinal surgery for fractures

and extramedullary tumors
Frederick Vincenta, Michael O. Kellehera and Michael G. Fehlingsb,*
a
Division of Neurosurgery and Spinal Program, University of Toronto, Toronto Western Hospital, Toronto, ON M5T 2S8, Canada
b
Professor and Krembil Chair in Neural Repair and Regeneration, University of Toronto, Toronto, ON M5T 2S8, Canada
44.1. Introduction
The surgical treatment of complex spinal pathologies
such as fractures and tumors can be very challenging
because they are associated with an increased risk of
neurological injury to the patient compared with more
routine spinal procedures (Padberg and Thuet, 2006).
The extent of any resultant deficit can vary from transient dysfunction of a single root to complete permanent paraplegia. During the course of these complex
operations, the neural tissue is exposed to hazards at
different stages of the surgery, especially during the
decompression and stabilization stages of the operation. Often times before the onset of the surgery the
cord is already significantly compromised and ischemic, from either direct trauma or compressive forces
such as tumor, bone, disk, or hematoma. Patients with
preexisting cord compromise have a higher risk of having an intraoperative neurophysiological deterioration
(Thuet et al., 2005; Lee et al., 2006).
The purpose of monitoring is to assess the physiologic integrity of the spinal cord and nerve roots during surgical treatment and to alert the surgeon of any
deterioration before irreversible neural injury has
occurred. Early attempts at monitoring solely relied on
recording somatosensory evoked potentials (SEPs).
Reports of false negative outcomes when using only
SEP monitoring illustrated the need for multimodality
monitoring (Lesser et al., 1986; Ben-David et al.,
1987; Wiedemayer et al., 2004). Numerous neurophysiological monitoring methods are now available including continuous free running electromyography (EMG),
*
Correspondence to: Dr. Michael Fehlings, M.D., Ph.D.,

FRCSC, Division of Neurosurgery, Toronto Western Hospital, 399 Bathurst St., Toronto, ON M5T 2S8, Canada.
Tel.: 1-416-603-5627; fax: 1-416-603-5298.
E-mail: michael.fehlings@uhn.on.ca (M.G. Fehlings).
evoked electromyography (E-EMG), compound muscle

action potentials, rectal and urinary sphincter EMG,
motor evoked potentials (MEPs), SEP, and most recently
spinal cord mapping (Quinones-Hinojosa et al., 2002;
Shi et al., 2003; Kothbauer and Novak, 2004; Paradiso
et al., 2006). None of these tests individually provide a
global assessment of cord and root function on their
own. However when multiple modalities are monitored
each one adds selective information that allows the surgical team to assess neural function with enhanced precision. Each of these electrophysiological approaches has
advantages and disadvantages. Hence, the decision
regarding the optimal choice of approaches to monitor
needs to be individually tailored for each surgery,
depending on what level of the spine is undergoing surgery and what aspect of neural function is most at risk.
44.2. Fractures
The operative management of spinal fractures can be
summarized under three general principles:
Decompression of the neural elements
Reduction of malalignment or subluxation
Stabilization of the spine
The tests employed for a given fracture are dependent

on the spinal level, the approach and what neural structures are deemed to be most at risk during the procedure.
The surgical approach, whether anterior, posterior or
both, is determined by the type of fracture, the degree
of instability, and the presence of neural compression.
There are four phases during which monitoring the
cord/nerve root function during the surgical procedure
is critical. These are positioning, decompression, reduction maneuvers (including traction), and reconstruction.
Initially, positioning is critical because poor positioning can cause signal impairment secondary to entrapment neuropathy. Acute compression compromises
SPINE SURGERY
axoplasmic flow, which can reduce membrane excitability of peripheral nerve. This could cause false positive signal change not directly related to the surgical act.
Concerning the other phases, the cause of signal change
is directly related to the actual procedure being undertaken causing potential neural compromise. Electrophysiological recordings will show deterioration in the
event of direct injury such as mechanical compression
(e.g., neural impingement by instrumentation) and distraction maneuvers, or an indirect event such as ischemic compromise.
44.2.1. Positioning
The difficulties encountered during positioning are
best appreciated by contemplating the following
scenario. To position a patient who has suffered an
unstable lower cervical spine fracture for a posterior
cervical operation requires that the patient be transferred from his bed onto the operating table and then
be turned prone. In order to visualize the lower cervical
vertebrae with fluoroscopy the shoulders need to be
taped down. Given this hypothetical scenario, it is easy
to understand the dangers involved in positioning
alone and how easy it is for an iatrogenic injury to
occur with all these maneuvers. Schwartz et al. found
that 1.8% of patients undergoing anterior cervical
spine surgery had evidence of impending neurological
injury secondary to positioning (Schwartz et al., 2006).
The majority of their patients had evidence of impending brachial plexopathy, following shoulder taping and
the application of counter traction. They found similar
results in their pediatric scoliosis population [3% had
evidence of position-induced brachial plexopathy
(Schwartz et al., 2000)]. One of the most vulnerable
619
areas for positional peripheral nerve injuries is the

ulnar groove, especially when the patients arms are
tucked tightly at the side (Schwartz et al., 2006).
44.2.2. Decompression
Decompression of the spinal neural elements usually
requires removal of bone and ligamentous structures.
In a trauma situation the natural protective layers surrounding the cord can often be displaced or torn. Bony
spurs can lacerate the dura thereby potentially exposing the neural tissue to injury during the decompressive phase. Bleeding from the bony fracture can often
be profuse and may obscure optimal visualization of
the neural elements. For these reasons, intraoperative
spinal cord monitoring may be very useful to minimize
iatrogenic neurological injury and to assess the physiologic integrity of the spinal cord and nerve roots during the surgical decompression. Patients who are
operated on for cervical trauma are much more likely
to have a major intraoperative neurophysiological alert
than those who have other cervical pathologies (Lee
et al., 2006). Tsirikos et al. found that in patients who
had an incomplete spinal cord injury, more than 20%
recovery in signal amplitude at the conclusion of the
operation was associated with favorable neurological
outcome (Tsirikos et al., 2004).
44.2.3. Reduction
Reduction of subluxation and restoration of spinal
alignment is undertaken to restore the coronal and sagittal balance and to reduce the risk of secondary syrinx
formation. Reduction maneuvers such as distraction
can cause neural injury, either by causing a traction
Fig. 1. Case study 1: A traumatic T1011 fracture-dislocation in a 54-year-old with ankylosing spondylitis, A: computed
tomographic (CT) sagittal reconstruction, B: CT axial, C: sagittal T2WI MRI, D: axial T2WI MRI, E: postoperative lateral
view, F: antero-posterior (AP) view.
620
F. VINCENT ET AL.
Fig. 2. Somatosensory evoked potentials (SEPs) recorded from the upper limbs A: and lower limbs B: of the patient described in
case study 1. The solid line displays the baseline recording and the dotted line displays the recording at closure. Averages of up to
1,000 stimulation are presented in all channels with recordings of SEPs. There was no significant change compared to baseline.
injury or by compressing the neural structures if inadequate decompression has been carried out. DAlpa
et al. reported a transitory impairment of spinal cord
function as measured by the SEPs during spinal distraction and vertebral body fusion (DAlpa et al.,
1989). In spinal pathologies such as scoliosis surgery
where distraction techniques are used routinely, there
is good evidence to support the value of intraoperative
neurophysiological monitoring with combined SEP/
MEP monitoring without the need for a wake-up test
(Padberg et al., 1998; Padberg and Thuet, 2006).
Patients with spinal deformity secondary to fracturedislocation are at risk for cord ischemia following
acute spinal distraction. Distraction might reduce
blood vessel caliber and decrease spinal cord blood
volume (Lyon et al., 2005).
44.2.4. Instrumentation
Placement of hardware, with screws, plates, and cages
to internally stabilize the spine, is usually performed
after the cord, and roots have been decompressed.
There is a potential risk of injury to the exposed cord
during the placement of this hardware. The need to
assess individual nerve root function reliably has

become even more critical with the advent of transpedicular fixation (Padberg and Thuet, 2006). Both triggered and spontaneous EMG has been used in this
regard (Gunnarsson et al., 2004; Djurasovic et al.,
2005). The intraoperative monitoring (IOM) principles related to corrective surgery for spinal fractures
are illustrated by the case example shown in Figs. 13.
44.3. Tumor
The hazards that exist in relation to positioning,
decompression, and instrumentation in a trauma
patient also hold true for treating a patient with a spinal
tumor. As well as these problem tumors can distort and
compress the neural tissue, making it extremely difficult to discriminate normal and physiologically intact
structures. Intraoperative monitoring uses vary,
depending on what type of extramedullary tumor is
being operated on, whether the tumor is intradural
extramedullary or completely extradural. Likewise
IOM has different roles dependent on the type of
resection that is being undertaken, whether a complete
en bloc resection is being attempted or whether the
SPINE SURGERY
621
Fig. 3. Motor evoked potentials (MEPs) recorded from the left upper limbs A: right upper limbs B: left lower limb C: and
right lower limbs D: of the patient described in case study 1. The solid line displays the baseline recording and the dotted
line displays the recording at closure. There was no significant change compared to baseline.
goal is a palliative resection and stabilization of a spinal metastasis.

44.3.1. Intradural extramedullary
Direct stimulation of the spinal cord with EMG recording, or intramedullary spinal cord mapping, can be used
to identify the extent of the tumor margin at the interface with functional tracts (Quinones-Hinojosa et al.,
2004b). In cervical dumbbell and foraminal tumors,
anatomical relations can often be difficult to interpret
because the tumor may displace the nerve rootlets.
Direct electrical stimulation with compound muscle
action potential recordings at multiple sites can be used
to identify the location and course of the involved nerve
root (Guo et al., 2006). This provides real-time information regarding the functional status of the roots to predict postoperative outcome. Excision of a filum tumor
by sectioning of the filum terminale carries a significant
potential risk of injury to the neighboring motor and
sensory roots. Intraoperative neurophysiological monitoring techniques can help to minimize these adverse
neurological outcomes by direct stimulation of the
filum and the surrounding roots with EMG recording
(Quinones-Hinojosa et al., 2004a).
44.3.2. Spinal metastasis: palliative surgery
The main goals for operative treatment of spinal metastasis are to decompress the neural elements and to stabilize the spine in order to improve the patients pain. At
the time of presentation, a lot of these patients are in
severe pain and are beginning to lose neurological
function. Multimodality monitoring is employed to

allow the surgeon feedback on how the already compromised cord tolerates the decompression. During decompression blood loss from the tumor can be profuse which
obscures the operative field increasing the danger of an
iatrogenic injury to the spinal cord. Hypotension secondary to hypovolemia can be very detrimental to an
ischemic, compressed cord. The surgical approach is tailored to achieve the primary goals of decompression and
stabilization. In the thoracolumbar spine the posterolateral transpedicular approach is one such approach that
can be employed. This allows 360o decompression of
the spinal cord and also affords the opportunity to
re-enforce the anterior column (Wang et al., 2004).
44.3.3. En bloc spinal tumor resections
En bloc spinal resections are reserved for patients
who have tumors that have a potential of achieving
a cure or long-term survival and who may only have
limited adjunctive therapy options (Table 1). The
Table 1
Selected primary malignant and benign tumors
of vertebral column (Jacobs and Fehlings, 2006)
Benign
Malignant
Osteoid osteoma/
osteoblastoma
Aneurysmal bone cyst
Giant cell tumors
Myeloma/plasmacytoma
Chordoma
Osteosarcoma
Ewing sarcoma
Chondrosarcoma
622
F. VINCENT ET AL.
Fig. 4. Case study 2: (A) sagittal T2WI MRI shows an expansive intradural extramedullary invasive schwannoma located
at the conus medularis with distal extension to the proximal cauda equina. On axial T2WI and T1WI MRIs, there is significant expansion with erosion of the pedicles and scalloping of vertebral bodies (B and C).
surgical exposure is usually more extensive, as is the

associated surgical morbidity. In this type of surgery
it is often necessary to remove several vertebral bodies. The associated segmental vessels at these levels
are ligated. To determine if the spinal cord has sufficient anastomotic supply in this region, the surgeon
may initially clamp the vessel in a temporary reversible fashion (Apel et al., 1991; Leung et al., 2005). If
no change occurs in the evoked responses over a
period of 1520 min, then the vessel is ligated and
divided.
The principles used for IOM during the resection

of a spinal tumor are illustrated in Figs. 46.
44.4. How IOM helps, how it changes the
procedure
The functional outcome after spinal reconstructive
procedures is primarily dependant on the extent of
baseline neurological deficits but also related to surgical maneuvers that could potentially create further
damage to an already compromised spinal cord.
Fig. 5. Somatosensory evoked potentials (SEPs) recorded from the lower limbs of the patient described in case study 2.
A: Baseline responses. During tumor dissection from the distal conus/proximal cauda equina on the left side, there was
a reduction (black arrow) of the left lower limb evoked potentials (B). Based on this electrophysiological change, an
alteration in the microsurgical plan occurred. SEPs responses at the end of the case (C), and at closure (D). The solid line
displays the baseline recording and the dotted line displays the active trace.
SPINE SURGERY
Fig. 6. Motor evoked potentials (MEPs) recorded from the

upper A: and lower limbs B: of the patient described in
case study 2; the solid line displays the baseline recording
and the dotted line displays the recording at closure. The
MEPs were recorded from the first dorsal interosseous,
quadriceps, tibialis anterior, and gastrocnemius and plantar
foot muscles.
Intraoperative monitoring is a modality to help identify, at an early stage, potentially reversible sensory
or motor injury during surgery. Changes in axonal
conduction in motor and somatosensory tracts of
the cord have a significant correlation to the extent
of cord injury (initial cord deficit and potential further damage induced by surgery) and can be accurately monitored by combined recording techniques
of MEPs and SEPs, with the goal of assessing
the posttraumatic functional integrity of the spinal
cord.
If recorded SEPs deteriorate and reach the warning criteria and this degradation is not explained by
a change in perioperative physiological variables or
by anesthetic factors, then corrective action should
be undertaken immediately. Strategies include: stopping the surgical maneuvers for a few minutes and
waiting for data to improve, administering the Stagnara wake-up test (rarely done in our unit), removing
spinal instrumentation, increasing the cord perfusion
by hypertensive therapy, and administering high dose
steroids.
IOM is obligatory whenever neurological complications are expected as predicted by a known
623
mechanism (e.g., compression, distraction, ischemia,

inflammation, or a combination of these). It is therefore worthwhile to perform IOM when dealing with
spinal lesions. Spinal cord monitoring during intraspinal extramedullary tumor operations could be used
to minimize injury to the spinal cord and nerve roots
during the high-risk phase of microsurgical tumor
dissection. The predominant potential mechanism of
injury during extramedullary tumor dissection is
related to the mechanical manipulation of the tumors
and surrounding neural tissues place these critical
structures at risk. IOM can be used to assess the functional integrity of neural pathways during these highrisk procedures to minimize the risk of injury.
Neurological injury may follow even a technically
precise spinal surgery. The IOM of neurological
functions put at risk by the operation is a method utilized to correctly identify the topography of neural
structures and to avoid surgical insults by real-time
response-actions by the surgical team. It is only in
this circumstance that a spinal cord injury could be
avoided. Neurophysiological monitoring aids in the
intraoperative decision making and also as a tool
for the prediction of neurological outcome (Kothbauer et al., 1997).
44.5. What IOM cannot do, what are the
limitations?
Our team recently published a retrospective analysis
of a prospectively accrued series of 213 consecutive
patients who underwent intraoperative neurophysiological monitoring with EMG and SEPs during thoracolumbar spine surgery (Gunnarsson et al., 2004).
We observed that the intraoperative electromyographic activation has a high sensitivity (100%) for
the detection of a new postoperative neurological
deficit but a low specificity (23.7%). On the other
hand, the SEPs were found to have a low sensitivity
(28.6%) but high specificity (94.7%). We concluded
that a combined intraoperative neurophysiological
monitoring with EMG and SEPs is helpful for predicting and possibly preventing neurological injury
during thoracolumbar spine surgery.
During reconstructive procedures, transient cord
compromise may take place secondary to traction,
decompression, or reduction. These could result in
attenuation in evoked potentials, which could represent mild, reversible neurological impairment during
surgery without detectable postoperative clinical consequences. Thus, these attenuations in IOM are
624
eventually interpreted as false positive, when correlated with the clinical outcomes but have to be taken
seriously during the procedure.
The reliability of the SEPs is significantly influenced by the perioperative physiological and anesthetic variables (Clapcich et al., 2004). SEPs are not
a direct measure of motor tract function and should
not be used exclusively for this purpose. Falsenegative results with SEPs represent a failure of
SEPs to detect an iatrogenic lesion during surgery.
An isolated injury to the motor tracts with sparing
of sensory pathways allows maintenance of normal
SEPs and explains the false-negative result (Wiedemayer et al., 2004). This supports the use of IOM
by stimulating motor pathways, as an adjunct to SEPs
with the aim to increase accuracy and reproducibility
of our approach. Recording from the arms (used as a
test control) facilitates differentiation between systemic alterations and focal neurological compromise
(MacDonald et al., 2003).
A number of other potential limitations of IOM
must be carefully considered. For example, SEPs
are averaged responses and consequently there may
be a time delay before being aware of a significant
deterioration and communicated to the surgical team.
MEPs can be challenging to obtain because of the
variability of the recordings, and also of not being
able to use neuromuscular paralysis that can lead to
difficulties with patient ventilation and difficulty
with muscle retraction during the surgical exposure.
EMGs have a high sensitivity but a low specificity
for monitoring potential injury. It is important to recognize that the different modalities complement each
other in term of their sensitivity and specificity, and
should be used in combination to maximize the effectiveness of the monitoring protocol.
Choosing the right anesthetic agents for patients
undergoing spinal surgery, with motor evoked potential, is crucial. For instance the variability of the
evoked potential signal is related to and dependent on
the type of anesthesia. The protocol used in our center
implies moderating the amount of inhalation anesthetic used or the use of intravenous anesthetic agents.
The utilization of muscle relaxants needs to be
employed with control when MEPs or EMG are being
recorded, because muscle relaxants can significantly
alter or completely block the recordings. It is well
known that anesthetic agents have a predictable and
consistent effect on MEP responses (Lo et al., 2006;
Sekimoto et al., 2006). Anesthetic agents show a
dose-dependant decreasing effect on the amplitude of
F. VINCENT ET AL.
MEPs. At constant dose, this effect has a tendency to

increase with the duration of the procedure, so escalating threshold voltage is needed. The rise in threshold
voltage is larger in myelopathic patient (Lyon et al.,
2005). Lyon et al. demonstrated that the rate of rise is
inversely proportional to anesthetic duration (Lyon
et al., 2005).
Many patients undergoing surgery for trauma or
tumor have preexisting neurological compromise that
may mean that no baseline electrophysiological recordings can be obtained. Unfortunately, such patients
may be at particularly high risk for perioperative neurological compromise Indeed, Theut et al. reported
an increased incidence of postoperative neurological
deficits in patients with absent monitoring data despite functional sensation and motor strength (Thuet
et al., 2005).
44.6. Which methods should be applied?
The choice of what modalities to use is dependent on
the preference of the surgeon, the nature and localization of the pathology involved, and the technical
feasibility of the modality in the specific context.
A good way to select which tests to use is to choose
the modality most specific to the neural tissue at risk
during the procedure: MEPs for anterior cord, SEPs
for posterior cord, and EMGs for roots at risk,
respectively.
Combined application of SEPs after peripheral
nerve stimulation and motor evoked potentials after
transcranial stimulation has been routinely used to
assess the integrity of pathways in the posterior and
anterior spinal cord, respectively, in patients undergoing spinal surgery.
SEPs remain the standard technique for IOM. Newer
techniques such as EMG and MEP have been developed and are used as an adjunct to SEP monitoring.
Somatosensory evoked potentials are obtained with
electrical stimulation of a peripheral nerve. Peripheral, subcortical, and cortical recording sites are monitored for each nerve stimulated. We recommend the
use of a minimum of three cortical sites and one cervical recording site for spinal surgery. The cortical
recording electrodes are positioned over the sensory
cortex in reference to the homunculus part associated
with the stimulation site. The cervical recording site
is placed over a cervical vertebra rostral to the
SPINE SURGERY
surgical site. Peripheral recording sites are useful for

detecting peripheral ischemia or neural compression,
and also for serving as an internal control in the SEP
circuit.
Subcortical SEP traces have proved to be very
reliable and sensitive to surgical interventions while
remaining relatively resistant to the effects of volatile
anesthetic agents. The recorded responses are measured for both amplitude and latency. The surgical
team is alerted if the baseline responses demonstrate
more than a 10% increase in latency or a 50% decrease
in amplitude. Baseline responses are established for
each patient at the start of the procedure.
44.6.2. Transcranial motor evoked potentials
Because of the well-reported limitations of SEP monitoring, the recording of MEPs has been advocated.
Myogenic transcranial MEP offers the advantage of
assessing the entire length of the corticospinal tract
from the cortex to the distal extremity beyond the neuromuscular junction. MEPs are recorded directly in the
muscles whose function is being assessed. One advantage it has, over other monitoring techniques, is its
topographic value since the recording of different
muscles allow the precise mapping of the radicular territory involve by a spinal cord lesion. According to
anatomical somatotopy, it is logical to think that the
more muscles recorded, the better the sensitivity and
specificity to predict injury will be. And even a single
loss of a recording specific to one muscle should be
significant and alert the surgeon of a potential injury.
After stimulation of the motor cortex, the responses
can be recorded from the spinal tracts or as a compound
muscle action potential. The most significant issue
when recording MEPs is the anesthetic regimen. A total
intravenous anesthetic without neuromuscular blockade needs to be utilized to provide the optimal environment for recording MEPs. There are fewer consensuses
in MEP monitoring regarding what parameters are considered to represent a significant change. In our unit a
complete loss of MEPs is considered a significant
event.
625
are thought to represent ongoing compression or

stretch. Spontaneous EMG is very sensitive but
not very specific (Gunnarsson et al., 2004). As well
as recording spontaneous EMG activity a number of
other applications have utilized EMG recordings to
determine the proximity of nerve roots. Direct
nerve root testing with EMG recording aids in the
dissection of nerve roots off intradural tumors
especially in the conus region. Triggered EMG has
been used to aid in placement of pedicle screws.
The elicited or evoked electromyographic (EEMG) response is used to evaluate the integrity of
the pedicle wall. By stimulation of pedicle tap,
finder, or screw in the presence of a wall fracture,
the walls impedance would be decreased, which
would allow current to flow through the fracture
site and stimulate the nearby root. E-EMG can
also be used to identify eloquent neural tissue and
functional nerve fibers from nonfunctional tissue,
especially useful in situation where the anatomy is
distorted by a lesion (e.g., tumoral or traumatic).
The interpretation of E-EMG is based on the threshold intensity at which a response is elicited.
44.6.4. Sphincter monitoring
A urethral ring electrode can be used to record from
the external urethral sphincter. During complex neurosurgical procedures involving the conus medullaris
and cauda equina, the combined SEP and EMG monitoring of lower limb muscles, external anal (EAS)
and urethral sphincters (EUS) is a practical and reliable method for obtaining optimal electrophysiological feedback (Krassioukov et al., 2004). These
intraoperative adjunctive modalities positively influence decision making with regard to microsurgery
and reduce the risk of per operative neurological
complications. The presence of electrically evoked
EMG activity in the sphincters and lower limb
muscles altered the perioperative micro-dissection
plan in over 40% of complex lumbrosacral cases
(Krassioukov et al., 2004; Paradiso et al., 2006).
44.6.3. EMG
44.7. What to think about clinically when the

potentials change?
Spontaneous EMG recordings provide real time

data, sensitive to surgical manipulation or compression. Myotomes are selected for recording based on
the operative level and the nerve roots most at risk.
Burst or train activity are considered significant and
The surgical team needs to be well prepared for the

eventuality of electrophysiological changes. The team
must have a protocol in place if alerted to an IOM
event. The first question that must be asked is whether
the recording change is related to any change in
626
anesthesia or physiological variable? The second question is whether there is any technical issue involved
(i.e., loose or dislodged recording or stimulating electrode). If the changes cannot be explained by technical,
anesthesia, or physiological variables, the surgical
team needs to re-evaluate their recent maneuvers
(e.g., distraction, screw placement, and graft insertion)
that may have contributed to the electrophysiological
change. In the advent of changes in SEPs or MEPs then
concrete actions such as increasing blood pressure, giving steroids, releasing distraction, stopping any operative manipulation, and removing hardware must be
seriously considered.
The specific criteria for the evaluation of SEPs are
well accepted and widely used. These criteria are
only a warning and do not automatically indicate that
the neurological status will be impaired. A 10%
increase in latency or a 50% or larger diminution in
amplitude are the criteria used to determine if
recorded data have changed significantly. For the
detection of mechanical insults to the spinal cord,
the latency and amplitude seems to be equally sensitive. This is not the case for the detection of ischemic
insults where amplitude is more sensitive compared
with latency. If recorded SEPs deteriorate and reach
the warning criteria and this degradation is not
explained by a change in per operative variables,
then correction steps should be undertaken immediately. Lack of recovery of SEP amplitude at the conclusion of surgery is associated with a significant risk
of neurological compromise (Tsirikos et al., 2004).
Patients with a >20% rise in signal amplitude before
completion of the procedure demonstrated a better
neurological picture compared with their preoperative status. Martin et al. reported three cases where
SEPs were improved after decompression and reduction of traumatic spine fractures that was associated
with a clinical improvement (Martin et al., 1998).
It is known that long-term mechanical strain, such
as that exerted by pressure from tumors in the spinal
canal, leads to an increase of latencies of SEPs. Differences of latencies of the two sides correlate
closely to the location of the tumor and the defects
of neurological function (Fromme et al., 1990).
Tumor dissection and manipulations of the spinal
cord as they occur during surgery do, at best, cause
reversible decrease of amplitudes of SEPs.
During reduction of spinal fracture-dislocations,
and in particular with distraction, ischemia to the
cord may occur. If this is detected by a loss of MEPs,
then there is an imperative need to restore adequate
F. VINCENT ET AL.
perfusion to the spinal cord as soon as possible.

To avoid the dilemma of whether to remove distractive forces, one might opt to raise the blood pressure,
blood volume, and hemoglobin levels before correction (Lyon et al., 2005).
There is considerable variation in the literature
with regard to the most reliable criteria to detect an
intraoperative event using MEPs. In our unit a complete loss of signal is considered significant. Other
authors have suggested a change in the morphology
of the waveform from polyphasic to biphasic, or a
decrease of 50% in amplitude or an increase in
threshold voltage during the surgery (Jellinek et al.,
1991; Morota et al., 1997; Padberg and Thuet,
2006). Until more evidence comes from the literature
with regards to what the optimal criteria are for
determining significance for myogenic MEPs, most
centers rely on an present or absent criteria to
determine significance.
What is the implication of the duration of morphological changes, in waveform myogenic MEPs,
according to neurological outcome? For instance, is
prolonged reduction of intraoperative MEP amplitude
predictive for postoperative neurological dysfunction?
Lang et al. studied this issue and concluded that the
recovery of lost MEPs was not associated with postoperative motor deficits, whereas nonrecovery of lost
MEPs was (Lang et al., 1996).
For transcranial MEP, the threshold values that
warn of imminent neurological damage have not yet
been established. In the case of intramedullary tumors,
Quinones-Hinojasa et al. support the concept that
changes in pattern and duration of the MEP waveform
morphology correlated with the occurrence of postoperative motor deficits (Quinones-Hinojosa et al.,
2004b).
During a surgical procedure, anesthesia deepens or
accumulates provoking additional lower motor neuron
suppression and causing fading or disappearance of
muscle responses to the initially chosen stimulus
parameters. Options to solve this frequent observation
are to increase the stimulation intensity or the pulse
number. This allows maintaining responses in situation
of evolving systemic effects. Systemic alterations can
be identified by parallel upper- and lower-extremity
changes (MacDonald et al., 2003). In contrast, focal
changes in the electrophysiological recordings may
suggest focal neurological compromise. (Table 2
depicts a summary of criteria for the interpretation of
signal change during per operative monitoring, in accordance to the modality used.)
SPINE SURGERY
627
Table 2
Advantages and disadvantages of neurophysiological options for intraoperative monitoring
Modality
Advantages
Disadvantages
Applications
SEP
Recorded from skin or scalp.

Dorsal column monitoring
(posterior cord). High
specificity
Benchmark standard
monitoring test
MEP
Can be recorded from

epidural space, peripheral
nerves, or muscles. Direct
monitoring of motor
pathways (lateral and
anterior cord).
Improvement of
intraoperative MEPs is
predictive of favorable
outcome
Real-time feedback
monitoring of spinal roots.
Bilateral continuous
monitoring. Loudspeaker
allows immediate acoustic
feedback. High sensitivity
Identify eloquent neural
tissue and functional nerve
fibers from nonfunctional
tissue
Postoperative motor deficits

could occur with stable
SEPs. Recorded using
averaging (which induces
delay). Suboptimal
sensitivity
More difficult to record.
Difficult anesthetic
regimen (e.g., continuous
intravenous anesthesia).
Variability of the
recordings
EMG (spontaneous)
EMG (evoked)
High-risk cases, monitor

anterior cord function
Low specificity
Facilitates preservation of
nerve roots and anterior
horn cells during tumor
resection or spinal
instrumentation
May be time consuming and

requires careful controls to
minimize false positives
and false negatives
Identification of
functional neural tissue.
Prevention of nerve
root injury during
pedicle screw
placement
SEP somatosensory evoked potential; MEP motor evoked potential; EMG electromyography.
44.8. Team-based approach

It is our view that IOM can improve the outcomes
obtained following surgery for spine fractures or
tumors. We recommend that it be used for all cases
of spine surgery cases where neurological compromise
is a potential risk. The applicability and effectiveness
of IOM increase with its use and familiarity. The most
effective application of these techniques requires a
team-based approach involving the surgical, anesthetic, and monitoring personnel. The combined
electrophysiological exploration of motor and sensory
potentials proved to be the most useful tool for monitoring patients with significant neural cord damage.
MEPs and SEPs (as well as EMGs) complement each
other. Good practices, such as being consistent in its
use and keeping a good and constant communication
between the monitoring and surgical team, are key

for its success.
44.9. What new types of monitoring would be
desirable for the future? (Table 3)
Direct spinal cord/epidural recording of dermatomal
SEPs provides more accurate results, as it records
responses directly from the cord. These responses are
more resistant to anesthesia than cortical EPs, which
are amenable to signal amplitude depression and peak
latency increases, particularly as the result of inhalation anesthetic agents, systemic hypotension, or
hypothermia.
SEPs monitoring of thoracolumbar procedures
typically includes posterior tibial and peroneal nerve
recordings. Femoral nerve SEP monitoring would
628
F. VINCENT ET AL.
Table 3
Recommendations for choice of intraoperative electrophysiological monitoring modality for spinal
tumors or fractures
Location
Anterior approach
Posterior approach
Cervical
Thoracic
Conus medullaris and cauda
equina
Lumbar, sacral
UL and LL SEPs, MEPs, UL EMGs

UL and LL SEPs, MEPs
LL SEPs and EMG EAS and EUS MEP
(conus)
EMG, LL SEPs
UL and LL SEPs, MEPs UL EMGs

UL and LL SEPs, MEPs
LL SEPs and EMG EAS and EUS MEP
(conus)
EMG, LL SEPs
UL upper limbs; LL lower limbs. EAS external anal sphincters; EUS external urethral sphincters; SEPs somatosensory evoked
potentials; MEPs motor evoked potentials; EMG electromyography.
provide an effective tool to monitor the midlumbar

roots intraoperatively during surgical treatment of
thoracolumbar fractures (Robinson et al., 1993).
The simultaneous use of different recording techniques for MEPs could enhance their potential effectiveness. In addition to electrodes recording the
activity from limb muscle, one could record the direct
response in the spinal cord (i.e., D waves or epidural
electrode directly caudal to the surgical site). The
D-wave amplitude provides a relative measure of
the number of functioning fast-conducting fibers in
the corticospinal tracts. When muscle MEPs can be
recorded, the functional integrity of the motor pathways is preserved. If muscle MEPs are lost during
dissection, then at least a temporary motor dysfunction
is to be expected postoperatively. This is where the
usefulness of adding D-wave recording is most useful.
The lost of muscle MEP with fully preserved D-wave
amplitude is a predictor of temporary weakness. This
allows the procedure to continue safely and be completed. Because this pattern of neurophysiological
parameters characterizes a temporary motor deficit, it
allows the identification of impending damage to the
motor pathways before this damage becomes irreversible. This significantly influences surgical decision
making. Another added advantage of D-wave recording is that they have relatively less variability to per
operative variables such as anesthetic agents.
44.9.1. Case study 1
A 54-year-old male, with ankylosing spondylitis, sustained a traumatic T1011 fracture dislocation (Fig. 1)
along with an epidural hematoma, but did not suffer
any neurological deficit. In view of the marked instability, a stabilization procedure was recommended.
Considering the high-risk nature of the proposed
surgery, the combined recording of (upper and lower

limbs) MEPs and SEPs was undertaken.
Using fluoronavigation, pedicle screws were placed
at the level of T7, T8, T9, T11, T12, and L1 bilaterally.
Following the insertion and confirmation of the screw
positioning, a decompressive laminectomy was performed at T9 and T10. There was a significant extradural collection (hematoma) overlying the dura,
which was evacuated. Adequate decompression was
achieved. Intraoperative monitoring consisted of
MEPs and SEPs with stimulation of the ulnar and posterior tibial nerves. The ulnar evoked SEPs were used
as a reference in determining whether changes in posterior tibial nerve were caused by surgical or nonsurgical insults. MEPs were recorded as compound muscle
action potentials (from left and right first dorsal interosseous, tibialis anterior, gastrocnemius, and plantar
foot muscles). During the decompression and instrumented stabilization, no significant changes were
observed in either the SEP or the MEP responses.
Figs. 2 and 3 show the baseline and closing SEPs and
MEPs, respectively. Postoperatively the patient awoke
without any new neurological deficit.
44.9.2. Case study 2
A 55-year-old woman presented with severe low back
and leg pain with evidence of a large intradural expansive lesion, which appeared to be either arising from or
compressing the conus medullaris with distal extension to the proximal cauda equina (Fig. 4). In view of
the erosion of the pedicles and scalloping of the vertebral bodies coupled with the extensive decompression
that would be required it was felt that this patient
would require tumor excision with the reconstruction
of the thoracolumbar spine. The modalities used for
IOM were: upper and lower limbs MEPs, combined
SPINE SURGERY
SEP and EMG monitoring of lower limb muscles,

external anal (EAS) and urethral sphincters (EUS).
Under microscopic visualization, the dura was
opened in midline initially starting at L2 caudally
and then proceeding rostrally. The tumor, which
proved to be a schwannoma arising from an L3 root,
was visualized and was found to be in close apposition
with the sensory roots of the cauda equina. The roots of
the cauda equina particularly on the left side were
markedly adherent to the dura and required to be dissected off both the dura and the tumor. The motor
nerves of the cauda equina were identified using direct
stimulation.
When the tumor mass was dissected rostrally it was
possible to identify a plane of cleavage between the
tumor and the conus. During this part of the dissection
there was a significant change in the SEPs from the left
leg (Fig. 5). At this stage, the dissection of the tumor
from the conus was terminated and microsurgical resection proceeded in a different region. Partial recovery of
the SEPs (Fig. 5) occurred within a few minutes, and the
resection of the tumor was completed uneventfully.
Motor evoked potentials were stable throughout
(Fig. 6). One sensory nerve on the left side was exiting
from the tumor, and this was sacrificed. A complete
tumor excision was achieved. Following the procedure,
the patient was noted to have a new postoperative clinical deficit with the left lower limb weakness (power 3/5
psoas and biceps femoris muscles) and numbness. However, the patient recovered fully from the lower limb
weakness over the ensuing weeks.
This case illustrates how multimodality electrophysiological monitoring may positively influence
intraoperative decision making and minimize the risk
of an adverse outcome. The change in SEPs prompted
a change in operative approach which was associated
with partial electrophysiological improvement. While
the patient awoke with a new neurological deficit, this
rapidly resolved over several weeks.
44.10. Conclusion
The treatment of fractures and tumors of the spine
poses a great challenge to surgeons, given the potential
for causing severe neurological deficits. Multimodality monitoring provides the surgical team with a tool
that allows them to guide the surgery and modify the
surgical plan if needed. With increasing frequency of
use, the advantages and more importantly the limitations for each facet of IOM is becoming more clearly
defined.
629
Intraoperative monitoring should be recorded

throughout the surgical procedure. Baseline responses
should be obtained before the patient is positioned on
the operating table and monitoring continued right up
until the end of the procedure. These methods will help
capture any neurological insult caused by the
positioning.
SEP monitoring remains the standard test for IOM
despite its suboptimal sensitivity to detect all types of
cord injury. Myogenic MEPs appear to offer superior
sensitivity and specificity for accurately predicting
postoperative deficits in neurological motor function.
Unfortunately, the technique is not as user friendly as
SEP monitoring, both in terms of maintaining the
response and the anesthetic regimen. EMG monitoring
remains a very useful sensitive tool; however, it has a
limited specificity.
At present there is no all-encompassing test that can
adequately monitor all facets of spinal cord and root
function. Multimodality monitoring offers the most
comprehensive assessment of real-time intraoperative
function. Which set of tests are optimal to monitor
needs to be individually tailored for each surgery.
The use of intraoperative monitoring has become the
standard of care during complex spinal procedures.
It aids in intraoperative decision making, allows
the surgeon a second chance to per operatively avoid
a poor outcome by alerting him or her to the real time
consequences of his or her surgical maneuvers. MEPs
have a reliable power of predicting clinical outcome
and their use have altered the surgical approach in
that gross total resections are more readily attempted
as long as motor evoked potential data indicate the
intact functional integrity of the corticospinal tract.
Combined MEPs and SEPs provide independent
and complementary information and improve spinal
cord monitoring. Attention to such quantitative IOM
data may help to minimize postoperative motor deficits by avoiding or correcting excessive spinal cord
manipulation and modifying surgical technique during tumor resection. We also advocated the use of
free running and evoked EMGs when operating in
the cervical or lumbar areas.
References
Apel, DM, Marrero, G, King, J, Tolo, VT and Bassett, GS
(1991) Avoiding paraplegia during anterior spinal surgery. The role of somatosensory evoked potential monitoring with temporary occlusion of segmental spinal
arteries. Spine, 16: S365S370.
630
Ben-David, B, Haller, G and Taylor, P (1987) Anterior spinal fusion complicated by paraplegia. A case report of a
false-negative somatosensory-evoked potential. Spine,
12: 536539.
Clapcich, AJ, Emerson, RG, Roye, DP, Jr., Xie, H, Gallo, EJ,
Dowling, KC, Ramnath, B and Heyer, EJ (2004) The
effects of propofol, small-dose isoflurane, and nitrous
oxide on cortical somatosensory evoked potential and
bispectral index monitoring in adolescents undergoing
spinal fusion. Anesth. Analg., 99: 13341340.
DAlpa, F, Scandurra, L, Savuto, R, Russo, T, Grasso, A,
Denaro, E and Foti, L (1989) Pre-, intra- and postoperative somatosensory evoked potentials in cervical vertebral and spinal cord injuries. Minerva Anestesiol., 55:
115118.
Djurasovic, M, Dimar, JR, Glassman, SD, Edmonds, HL
and Carreon, LY (2005) A prospective analysis of
intraoperative electromyographic monitoring of posterior cervical screw fixation. J. Spinal Disord. Tech.,
18: 515518.
Fromme, K, Miltner, FO, Klawki, P and Friedrich, M
(1990) Spinal cord monitoring during intraspinal extramedullary tumor operations (peroneal nerve evoked
responses). Neurosurg. Rev., 13: 195199.
Guo, L, Quinones-Hinojosa, A, Yingling, CD and Weinstein, PR (2006) Continuous EMG recordings and
intraoperative electrical stimulation for identification
and protection of cervical nerve roots during foraminal
tumor surgery. J. Spinal Disord. Tech., 19: 3742.
Jacobs, WB and Fehlings, MG (2006) Primary vertebral column tumors. In: CA Dickman, MG Fehlings and ZL
Gokaslan (Eds.), Spinal Cord and Spinal Column
Tumours. Thieme, New York, p. 372.
Jellinek, D, Jewkes, D and Symon, L (1991) Noninvasive
intraoperative monitoring of motor evoked potentials
under propofol anesthesia: effects of spinal surgery on
the amplitude and latency of motor evoked potentials.
Kothbauer, K, Deletis, V and Epstein, FJ (1997) Intraoperative
spinal cord monitoring for intramedullary surgery: an
essential adjunct. Pediatr. Neurosurg., 26: 247254.
Kothbauer, KF and Novak, K (2004) Intraoperative monitoring for tethered cord surgery: an update. Neurosurg.
Focus, 16: E8.
complex lumbosacral procedures: indications, techniques, and long-term follow-up review of 61 consecutive cases. J. Neurosurg. Spine, 1: 243253.
F. VINCENT ET AL.
Lang, EW, Beutler, AS, Chesnut, RM, Patel, PM, Kennelly, NA, Kalkman, CJ, Drummond, JC and Garfin,
SR (1996) Myogenic motor-evoked potential monitoring using partial neuromuscular blockade in surgery of
the spine. Spine, 21: 16761686.
Lee, JY, Hilibrand, AS, Lim, MR, Zavatsky, J, Zeiller, S,
Schwartz, DM, Vaccaro, AR, Anderson, DG and Albert,
TJ (2006) Characterization of neurophysiologic alerts
during anterior cervical spine surgery. Spine, 31:
19161922.
Lesser, RP, Raudzens, P, Luders, H, Nuwer, MR, Goldie,
WD, Morris, HH, III, Dinner, DS, Klem, G, Hahn, JF
and Shetter, AG (1986) Postoperative neurological deficits may occur despite unchanged intraoperative
somatosensory evoked potentials. Ann. Neurol., 19:
2225.
Leung, YL, Grevitt, M, Henderson, L and Smith, J (2005)
Cord monitoring changes and segmental vessel ligation
in the at risk cord during anterior spinal deformity surgery. Spine, 30: 18701874.
Lo, YL, Dan, YF, Tan, YE, Nurjannah, S, Tan, SB, Tan,
CT and Raman, S (2006) Intraoperative motor-evoked
potential monitoring in scoliosis surgery: comparison
of desflurane/nitrous oxide with propofol total intravenous anesthetic regimens. J. Neurosurg. Anesthesiol.,
18: 211214.
anesthesia: the phenomenon of anesthetic fade.
MacDonald, DB, Al, ZZ, Khoudeir, I and Stigsby, B (2003)
Monitoring scoliosis surgery with combined multiple
pulse transcranial electric motor and cortical
Martin, F, Madany, M, De, PF, Argenson, C, Schlatterer, B,
Tinsi, L and Dolisi, C (1998) Amelioration of perioperative somatosensory evoked potentials in the surgery of
injuries of the spine. Rev. Chir. Orthop. Rep. Appar.
Mot., 84: 189193.
Morota, N, Deletis, V, Constantini, S, Kofler, M, Cohen, H
and Epstein, FJ (1997) The role of motor evoked potentials during surgery for intramedullary spinal cord
Padberg, AM and Thuet, ED (2006) Intraoperative electrophysiologic monitoring: considerations for complex
spinal surgery. Neurosurg. Clin. North Am., 17:
205226, v.
Padberg, AM, Wilson-Holden, TJ, Lenke, LG and Bridwell, KH (1998) Somatosensory- and motor-evoked
potential monitoring without a wake-up test during idiopathic scoliosis surgery. An accepted standard of care.
Spine, 23: 13921400.
Paradiso, G, Lee, GY, Sarjeant, R, Hoang, L, Massicotte,
EM and Fehlings, MG (2006) Multimodality
SPINE SURGERY
intraoperative neurophysiologic monitoring findings
during surgery for adult tethered cord syndrome: analysis of a series of 44 patients with long-term follow-up.
Spine, 31: 20952102.
Quinones-Hinojosa, A, Gulati, M, Lyon, R, Gupta, N and
Yingling, C (2002) Spinal cord mapping as an adjunct
for resection of intramedullary tumors: surgical technique
with case illustrations. Neurosurgery, 51: 11991206.
Quinones-Hinojosa, A, Gadkary, CA, Gulati, M, Von
Koch, CS, Lyon, R, Weinstein, PR and Yingling, CD
(2004a) Neurophysiological monitoring for safe surgical
tethered cord syndrome release in adults. Surg. Neurol.,
62: 127133.
Quinones-Hinojosa, A, Lyon, R, Ames, CP and Parsa, AT
(2004b) Neuromonitoring during surgery for metastatic
tumors to the spine: intraoperative interpretation and
management strategies. Neurosurg. Clin. North Am.,
15: 537547.
Robinson, LR, Slimp, JC, Anderson, PA and Stolov, WC
(1993) The efficacy of femoral nerve intraoperative
somatosensory evoked potentials during surgical treatment of thoracolumbar fractures. Spine, 18: 17931797.
Schwartz, DM, Drummond, DS, Hahn, M, Ecker, ML and
Dormans, JP (2000) Prevention of positional brachial
plexopathy during surgical correction of scoliosis. J. Spinal Disord., 13: 178182.
Schwartz, DM, Sestokas, AK, Hilibrand, AS, Vaccaro, AR,
Bose, B, Li, M and Albert, TJ (2006) Neurophysiological
identification of position-induced neurologic injury during anterior cervical spine surgery. J. Clin. Monit. Comput., 20: 437444.
Sekimoto, K, Nishikawa, K, Ishizeki, J, Kubo, K, Saito, S
and Goto, F (2006) The effects of volatile anesthetics
631
on intraoperative monitoring of myogenic motorevoked potentials to transcranial electrical stimulation
and on partial neuromuscular blockade during propofol/fentanyl/nitrous oxide anesthesia in humans. J. Neurosurg. Anesthesiol., 18: 106111.
Shi, YB, Binette, M, Martin, WH, Pearson, JM and Hart,
RA (2003) Electrical stimulation for intraoperative
evaluation of thoracic pedicle screw placement. Spine,
28: 595601.
Thuet, ED, Padberg, AM, Raynor, BL, Bridwell, KH,
Riew, KD, Taylor, BA and Lenke, LG (2005) Increased
risk of postoperative neurologic deficit for spinal surgery patients with unobtainable intraoperative evoked
potential data. Spine, 30: 20942103.
somatosensory evoked potentials for spinal trauma. J.
Spinal Disord. Tech., 17: 385394.
Wang, JC, Boland, P, Mitra, N, Yamada, Y, Lis, E,
Stubblefield, M and Bilsky, MH (2004) Single-stage
posterolateral transpedicular approach for resection
of epidural metastatic spine tumors involving the vertebral body with circumferential reconstruction:
results in 140 patients. Invited submission from the
Joint Section Meeting on Disorders of the Spine and
Peripheral Nerves, March 2004. J. Neurosurg. Spine,
1: 287298.
Wiedemayer, H, Sandalcioglu, IE, Armbruster, W, Regel,
J, Schaefer, H and Stolke, D (2004) False negative findings in intraoperative SEP monitoring: analysis of 658
consecutive neurosurgical cases and review of published reports. J. Neurol. Neurosurg. Psychiatry, 75:
280286.

M.R. Nuwer (Ed.)
632
CHAPTER 45
Intraoperative neurophysiological monitoring during

surgery for intramedullary spinal cord tumors
Francesco Salaa,* and Karl F. Kothbauerb
a
Department of Neurological and Visual Sciences, Section of Neurosurgery, University Hospital, 37100 Verona, Italy
b
Division of Neurosurgery, Kantonsspital Luzern, CH-6000 Luzern 16, Switzerland
45.1. Introduction
45.1.1. A brief history of spinal cord
tumor surgery
Neurosurgeons have a lot of respect for the spinal
cord. The removal of intramedullary tumors of the
spinal cord is considered challenging and believed
to carry a significant risk for surgical damage and
neurologic dysfunction.
The first successful resection of an intradural
tumor, a fibro-myxoma, is attributed to Victor
Horsley in 1887 (Gowers and Horsley, 1888). The
first successful resection of an intramedullary spinal
cord tumor was performed in 1907 by Anton Von
Eiselsberg in Vienna (Von Eiselsberg and Ranzi,
1913; Von Eiselsberg and Marburg, 1917). The first
report about such a resection appeared in 1911 by
Charles Elsberg in New York (Elsberg and Beer,
1911). Elsberg described a two-stage strategy for
the removal of these intramedullary tumors. At the
initial operation only a myelotomy would be performed, and the surgeon would then return one week
later to remove the extruded portion of the tumor.
After several reports advocating aggressive surgery
for the resection of intramedullary tumors, postoperative neurological deficits were significant. Thereafter,
many neurosurgeons recommended a conservative
strategy with biopsy, dural grafting, and radiation therapy regardless of histological diagnosis (Wood et al.,
1954). There was no way at that time to obtain functional information during surgery. Tumor resection
then was a purely anatomic operation. With the
Correspondence to: Francesco Sala, M.D., Department of

Neurological and Visual Sciences, Section of Neurosurgery,
University Hospital, Piazzale Stefani 1, 37100 Verona, Italy.
E-mail: francesco.sala@azosp.vr.it; and francescosala@yahoo.
com (F. Sala).
advent of microsurgery, imaging technology, and

intraoperative neurophysiology, the treatment strategy
for these intramedullary neoplasms has again become
more aggressive. This is particularly important as the
majority of intramedullary tumors are histologically
benign (Constantini et al., 2000) and the radical
removal results in a long-term survival with an acceptable morbidity (Greenwood, 1967; McCormick and
Stein, 1990; Epstein et al., 1993; Fischer and Brotchi,
1996; Constantini et al., 2000).
With the refinement of microsurgical techniques
(Jallo et al., 2001; Brotchi, 2002), the use of ultrasonic
aspiration (Epstein, 1984), the microsurgical laser
(Jallo et al., 2002), and intraoperative neurophysiological monitoring (Kothbauer et al., 1997), the microsurgical resection of intramedullary tumors is the primary
treatment modality for these neoplasms. Radiotherapy
should be indicated only for recurrent or malignant
intramedullary spinal cord tumors (ISCTs) (Fischer
et al., 2005).
45.1.2. Why monitoring?
In spite of the advances in diagnostic neuroimaging
and refinements of microsurgical techniques, ISCT
surgery remains challenging because of the intimate
relationship between tumor and long tracts of the spinal cord. Moreover, spinal cord vascularization is
extremely vulnerable due to the small caliber of feeding arteries and, especially at the thoracic level, the
lack of anastomoses between the pial network and central arteries. Therefore, any sustained mechanical traction or compression of the cord as well as an extensive
use of bipolar coagulation can potentially jeopardize
the functional integrity of the sensory and motor pathways (Jallo et al., 2001; Brotchi, 2002).
The rarity of this pathology (24% of central nervous system tumors) (Stein and McCormick, 1992)
accounts for the fact that only few neurosurgeons
SPINE SURGERY
633
have the opportunity, along their career, to build a

significant personal experience in this field. Nevertheless, technical skills, a deep knowledge of the
surgical anatomy of the spinal cord, and personal
experience in ISCT surgery not always prevent
injury to the cord during surgical manipulation.
The invaluable information that only intraoperative
neurophysiological monitoring can provide is the
online feedback on the level of tolerance of that
specific spinal cord during the removal of that specific tumor.
The goal of intraoperative monitoring (IOM) during ISCT surgery is not only to predict a postoperative neurological deficit but also to identify an
impending injury to the cord in time for corrective
measures to be taken. A strict collaboration between
surgeon, neurophysiologist, and anesthesiologist is
the key to critically analyze the neurophysiological
signals, provide a prompt feedback to the surgeon,
and take action to protect the spinal cord. This will
ultimately decrease surgical morbidity.
of IOM on the outcome following ISCT surgery will

be critically discussed. Finally, we will present new
perspectives on mapping sensory and motor pathways within the spinal cord in ISCT surgery.
45.1.3. How to monitor?
45.2.1. Clinical presentation
A matter of debate in ISCT surgery is which neurophysiological technique to use and which warning
criterion to adopt. The choice of these parameters
(techniques and warning criteria) does not simply
represent a neurophysiological dilemma but it should
be considered within the context of the oncological
aspects of this disease. The following questions must
be addressed when planning a spinal cord tumor
resection:
Intramedullary tumors may remain asymptomatic for

a long time or cause nonspecific complaints. Thus,
the history of symptoms prior to diagnosis has a
median of about nine months. High-grade neoplasms
have a shorter prodrome of about 4.5 months. The
most common symptom is pain, which is typically
pronounced at night.
Children can present with motor deficits which can
initially be seen as clumsiness, weakness, or frequent
falls. In young infants, this may manifest as motor
regression such as refusal to stand or crawl after having
learned to walk. Sensory symptoms, mostly dysesthesias, are found in 20% of patients.
Scoliosis can be a presenting complaint in up to one
third of patients. The scoliosis curve is not specific in
any direction. Children with scoliosis typically have
paraspinal pain which is the exception for idiopathic
scoliosis.
1) Is this tumor curable?

2) If yes, what is the price that this patient is willing
to pay to have the best chance of no tumor recurrence and no need for further therapies?
3) If not, is there a justification to charge the patient
with severe, even if transient, motor deficits?
Intraoperative neurophysiologists should be aware
of these issues because the level of reliability of their
technique will definitely have an impact not only on
the neurological but also on the oncological outcomes of these patients.
In this chapter we will review the principles of
ISCT surgery, the evolution of IOM strategies for
it, and the value of different IOM techniques at different stages during the surgical procedure. Surgical
dilemmas, their interaction with neurophysiological
and oncological aspects as well as the real impact
45.2. Treatment of intramedullary spinal cord

tumors
Intramedullary spinal cord tumors account for 24%
of all central nervous system tumors. They do, however, make up to 55% of all intradural neoplasms in
children. A review of the intramedullary spinal cord
database at a single institution between 1991 and
1998 yielded 294 cases in adults and children (Miller,
2000). The 294 tumors included 117 removed from
children under the age of 21 years, and 177 from
patients aged 21 years and above. Ependymomas predominate the adult category, while in children the
astrocytic tumors are most frequent. Only a small
minority of these tumors are malignant. The great
majority is histologically benign.
45.2.2. Diagnostic studies

Magnetic resonance imaging (MRI) is the imaging
modality of choice to identify an intramedullary
tumor (Brotchi et al., 1991). MRI scans should be performed before and after administration of the standard
intravenous contrast agent (gadolinium diethylenetriamine-pentacetic acid) and in multiple planes.
634
These images ideally demonstrate solid tumor components, cysts, and edema. Although MRI does not
provide the histological diagnosis, there are some typical patterns of appearance for intramedullary tumors.
Ependymomas tend to enhance brightly and homogenously. They are often associated with rostral and caudal cysts which contain hemosiderin, which appears
dark on T2-weighted images. These tumors are centrally located within the spinal cord. On the other
hand, astrocytomas and gangliogliomas have a heterogeneous enhancement pattern. These tumors are often
eccentrically located and result in an asymmetric
enlargement of the spinal cord.
Computed tomography with myelography is used
only for patients in whom MRI is contraindicated.
Plain radiographs are used to identify, quantify,
and follow spinal deformity, particularly scoliosis
associated with intramedullary tumors.
45.2.3. Surgery
45.2.3.1. Surgical instruments
The traditional method of suction and bipolar cautery
for the removal of intramedullary neoplasms is supplemented by specialized instruments which have
become essential for the microsurgical resection of
spinal cord tumors.
The Cavitron ultrasonic surgical aspirator (CUSA)
uses high-frequency ultrasound to fragment and then
suction tumor from the tip of this device. This allows
for efficient tumor debulking (Constantini and
Epstein, 1996). It has been our experience, however,
that using the CUSA within a tumor mass which is still
in situ, significant motor evoked potential (MEP)
changes may occur. We have thus increasingly used
the CUSA only as far away from the intact cord tissue
as possible, mostly to remove partly mobilized tumor
components.
A microsurgical laser, that is, a handheld instrument with a laser tip [Nd:YAG Contact Laser System (Photomedex, Montgomeryville, PA)] that is
used like a microinstrument, is an excellent surgical
tool for intramedullary surgery (Jallo et al., 2002). It
serves as a scalpel for the myelotomy, to demarcate
the glialtumor interface, and to remove small fragments of tumor tissue. The laser allows for precise cutting of both very soft and very firm tissues alike, thus
minimizing manipulation-related injury. It is also
very useful for lipomas. For the neurophysiologists,
it is important that the laser, unlike electric cautery,
does not cause an electric artifact which saturates
F. SALA AND K.F. KOTHBAUER
amplifiers. Therefore, it is possible to continue neurophysiologic recordings during laser dissection without
interference.
45.2.3.2. Surgical technique
Surgery for all spinal cord tumors is done with the
patient in prone position. A laminectomy or laminotomy is done to access the spinal canal. For cervical
or cervicothoracic tumors, the patients head is fixed
in a Mayfield headholder. The bone opening is done
in a way that permits repositioning of the laminae
and is made large enough to expose the solid component of the tumor. Usually cystic components do not
need to be exposed. This opening is planned with
X-ray control and verified with the intraoperative
ultrasound prior to dural opening (Theodotou and
Powers, 1986).
The dura is opened and the edges are retracted
with sutures. The myelotomy is most often done
through the midline, which may be difficult to identify when the cord is rotated or deformed by the
tumor, or in the dorsal root entry zone, if this permits
better exposure.
Ependymomas have a characteristically distinct
interface between their capsule and the cord tissue.
This most often permits dissection of the tumor away
from the cord and thus most often complete resection
(McCormick et al., 1990). Astrocytomas more frequently do not have such a useful cleavage plane
(Epstein et al., 1992). Resection thus remains most
often biologically subtotal, or near-total. The tumors
receive their blood supply from the anterior spinal
artery and therefore are attached to this area, and this
is also the area of critical hemostasis which must not
affect the parent vessel.
A frozen section biopsy at this stage provides
information about the presence of a malignant tumor,
which significantly influences the surgical strategy:
The poor prognosis of all malignant tumors does
not justify the risks of aggressive resection. The surgeon therefore will only debulk the tumor, with minimized risk for even transient neurologic dysfunction.
By contrast, tumor removal for low-grade astrocytomas and gangliogliomas begins often with an internal
debulking to reduce the tumor volume. Then using
the suction, contact laser, bipolar forceps, and dissectors, the tumor is gently removed from the surrounding spinal cord tissue.
Interestingly, the rostralcaudal length of the
tumor does not seem to influence the functional outcome after tumor resection. The removal of small
SPINE SURGERY
tumors with a wide transverse diameter is more difficult and hazardous than that of a long tumor. Hemangioblastomas (Roonprapunt et al., 2001), regardless
of size, in the spinal cord are often associated with
significant edema and syrinx formation. The resection of these lesions is similar to their intracranial
counterpart. The lesion must be resected in a circumferential fashion. The tumor surface can be coagulated to allow for the manipulation of the lesion.
Cavernomas are usually situated at the dorsolateral circumference of the cord. These are surrounded
by a gliotic plane, which permits the delineation from
the surrounding spinal cord tissue. These bleed little
during resection.
Intramedullary lipomas require a different strategy
than glial neoplasms because these are densely adherent to the cord, and total removal is dangerous and not
recommended. The microsurgical laser is by far the
most useful instrument for debulking.
In intramedullary surgery, hemostasis is obtained
with saline irrigation and local application of microfibrillar collagen (AviteneW, C. R. Bard, Inc., Murray
Hill, NJ). Bipolar coagulation is used as little as possible. The use of bipolar coagulation, furthermore,
disrupts all electrophysiological recordings for the
time the current is active. The dura is closed primarily in a watertight fashion with a running locked
suture. If an osteoplastic laminotomy was performed,
the laminae are secured with a nonabsorbable suture
or miniplates. The muscle fascia must be closed in
a watertight fashion.
45.2.4. Surgical outcome
The great majority of intrinsic spinal cord tumors are
benign. Particularly ependymoma resection provides
cure when it is complete (Epstein et al., 1993). Most
other low-grade astrocytic tumors apparently do not
recur when a near complete resection is accomplished
(Constantini et al., 2000). This coincides with the
experience that the resection of the part of the tumor
closest to normal tissue, that is, the last bits and pieces,
is the most hazardous for neurologic injury. Guided
with monitoring, this problem can therefore be
avoided as it is not oncologically essential to remove
these last bits and pieces.
Neurologically, the outcome of spinal cord surgery is better than the common expectation of a high
risk for permanent paralysis. There is evidence that
the preoperative neurologic status is a clear indicator
for postoperative outcome: patients with significant
635
deficits preoperatively have a higher risk of neurologic deterioration (Morota et al., 1997). Furthermore, an already established significant deficit is
unlikely to be reversible by surgery. This is the single
most important argument for early surgery.
On the downside, there is an up to one-third risk for
a transient motor deficit, reflected by intraoperative
MEP changes (Kothbauer et al., 1998). Permanent
motor deficits occur in <5% and paralysis is fortunately occurring in only <1% of patients.
45.2.5. Anesthesiological considerations
The anesthesia management that allows intraoperative monitoring particularly of MEPs consists of a
constant infusion of propofol (usually in a dose of
about 100150 mg/kg/min) and fentanyl (usually
around 1 mg/kg/h). The use of propofol for anesthesia
with MEP monitoring has been reported with various
stimulation techniques (Jellinek et al., 1991; Kalkman et al., 1992; Schmid et al., 1992; Fennelly
et al., 1993; Taniguchi et al., 1993b; Sloan, 2002).
Nitrous oxide not exceeding 50 vol.% can be used.
Bolus injections of both intravenous agents should
be avoided because this temporarily disrupts muscle
MEP recordings, which is particularly problematic
during the critical resection part of the operation.
Short-acting muscle relaxants are given for intubation but not thereafter.
Halogenated anesthetics should not be used
(Taniguchi et al., 1991, 1993a). These anesthetics
elevate muscle MEP stimulus thresholds and block
muscle MEPs in a dose-dependent fashion. Using
them would add an uncontrollable variable.
Partial myorelaxation is used by some (Lang
et al., 1996), but there is no convincing evidence that
its use improves management of anesthesia. This is
still controversial because many neurosurgeons are
still reluctant to accept even the possibility of slight
patient movement during surgery.
With the patient fully relaxed, muscle MEP monitoring is impossible. Controlled relaxation would
add another uncontrollable variable to the interpretation of MEP data. The specificity of muscle MEP
monitoring would suffer. On the other hand, it is still
unlikely that patient movement from stimulation
could be completely avoided. Therefore, this would
combine poor monitoring with poor relaxation.
Relaxation is not necessary and both surgeon and
anesthesiologist have to get used to working
without it.
636
45.3. The role of intraoperative neurophysiology

45.3.1. Evolution of IOM in ISCT surgery
Before the advent of intraoperative monitoring of the
functional integrity of the motor pathways, intrinsic
spinal cord surgery was a purely anatomical
endeavor. This naturally comes from the fact that surgery, and neurological surgery no less, is an anatomically defined branch of medicine.
The first ones to use evoked potentials to assess the
functional integrity of the spinal cord during surgery
were orthopedic surgeons (Nash et al., 1977; Engler
et al., 1978). At that time, somatosensory evoked
potentials (SEPs) were the only neurophysiological
method available for monitoring. However, the technology was slow and the results were often difficult
to interpret. A major problem was conceptual: SEPs
reflect the functional integrity of the sensory pathways.
Information for the more important motor pathways
was therefore only indirect. This may be acceptable
when external cord compression is the expected mechanism of injury. If the presumed mechanism of injury
to the cord is external overall compression, as may be
mostly the case for intradural extramedullary tumors
or extradural tumors, and extradural surgery for deformity, SEPs convey some degree of overall monitoring, encompassing all pathways only because an
external compression will likely affect all pathways
to a similar extent.
The resection of lesions within the substance of the
cord, however, carries a risk of selective damage to the
motor tract, which may not be reflected by changes in
SEP recordings (Ginsburg et al., 1985; Lesser et al.,
1986). The possibility of a dorsolateral corticospinal
tract insult occurring independently from an injury to
the dorsal column which would be reflected by the
SEPs is possibly due to the different location and
vascular supply of these structures.
ISCTs are rare and reports on the use of intraoperative neurophysiological techniques specifically
during the surgical removal of an ISCT remain scattered through the past 15 years. Interestingly enough,
nevertheless, ISCTs recently accounted for the most
eventful monitoring in a series of 423 neurosurgical
monitored cases. The first report on the use of MEP
in ISCT surgery is by Morota et al. (1997). Earlier
than that, SEPs were used only with the assumption
that changes in SEPs specifically represent spinal
cord dysfunction. However, Whittle et al. (1986)
concluded that spinal cord monitoring by means of
SEPs recording would appear to be useful during

extradural spinal surgery, but there are limitations
associated with this technique during some types of
intradural surgery. Kearse et al. (1993) reported good
sensitivity but poor specificity of SEPs; this high rate
of false-positive results (changes in SEPs without
changes in neurological outcome) suggests that, monitoring only SEPs, may unjustifiably stop surgery precluding a complete resection of the tumor. Recently,
Skinner et al. (2005) reported on a patient with a C7T1 intramedullary cavernous malformation with
unchanged posterior tibial nerve SEPs in spite of lost
transcranial MEPs from both feet at the end of surgery.
The patient woke up with a new paraparesis that would
not have been recognized by monitoring SEPs only.
Furthermore, change in or even loss of SEPs during intramedullary surgery is frequent, because of the
approach through the dorsal midline (Epstein and
Farmer, 1990). Loss of SEPs, however, does not correlate to the motor status. SEPs can even be recordable in nonambulatory patients prior to surgery
(Kothbauer et al., 1998). Then, there is the problem
of recording delay of SEPs: due to the need for signal
averaging, a time delay occurs and the identification
of injury can lag behind the progress of the surgery;
the injury is already potentially irreversible before it
is even detected.
Other authors (Koyanagi et al., 1993) have explored
the possibility to elicit spinal cord evoked potentials
(SCEPs) rostral to the level of the tumor after applying
an electrical stimulation to the spinal cord caudally to
the tumor. However, this stimulation is nonspecific
and does not allow recognizing which pathway is
injured when these SCEPs change during surgery.
The sensitivity of this method turned out to be
unsatisfactory.
Morota et al. (1997) introduced the use of D-wave
(epidural MEP) monitoring after transcranial electrical
stimulation and concluded that this appeared as a better predictor of functional outcome than the patients
preoperative motor status. This concept opened to the
possibility to have neurophysiological parameters as
major outcome predictors in spinal cord surgery.
MEP monitoring is based on concepts of the
motor system developed since the 1950s (Patton
and Amassian, 1954; Philips and Porter, 1964), when
a small but essential fiber population in the corticospinal tract was identified and found to give rise to a
recordable traveling wave, the D-wave. After
Mertons description of transcranial electrical motor
cortex stimulation in man (Merton and Morton, 1980),
SPINE SURGERY
this knowledge was applied in the operating room (Levy

et al., 1984; Boyd et al., 1986; Katayama et al., 1988a,
1988b; Burke et al., 1990).
Muscle recording techniques were introduced with
magnetic (Edmonds et al., 1989) and electric (Zentner
et al., 1989) motor cortex stimulation. Anesthesiainduced blocking of the alpha-motoneurons posed a
major problem, but the development of a multipulse
technique (Taniguchi et al., 1991, 1993a) resolved this
difficulty.
In the 1990s, the understanding of the neurophysiological concepts, interpretation of data and safety
increased, and the experience with practical monitoring
improved. Several series provided evidence that intraoperative MEP monitoring really works and is indeed
useful for the experienced neurosurgeon (Jones et al.,
1996; Pechstein et al., 1996; Rodi et al., 1996; Calancie
et al., 1998; Kothbauer et al., 1998).
45.3.2. Intraoperative neurophysiological
monitoring of SEPs and MEPs during ISCT surgery
We have developed our IOM strategy according to
the progress of neurophysiological techniques and
adapting the strategy to the steps of surgery. Initially, identification of the midline is necessary for
the incision of the dorsal median raphe` in order to
access the tumor without injuring the dorsal columns. Visual identification of the midline may be
difficult particularly when the cord anatomy is
distorted by the tumor. Krzan et al. developed a
technique (Krzan et al., 1996) to identify the physiologic midline. With very thin wire-electrodes
mounted on an array electrode which is placed
transversely over the cord, it is possible to record
the traveling SEP waves in the dorsal columns very
selectively. Taking advantage of the somatotopic
distribution of afferent fibers in the dorsal column,
with the most caudal fibers lying nearest to the midline, it is possible to determine the midline in the
following manner. SEP stimulation on the right posterior tibial nerve allows recording of a traveling
wave which will have its highest amplitude close
to the midline. The same will happen for the contralateral side and the physiologic midline therefore
is located between these two amplitude peaks
(Fig. 1). We do not use dorsal column mapping routinely, because in many cases the true midline can
be identified anatomically. Nevertheless, this
mapping technique can be useful whenever the
anatomical identification is not reliable.
637
Similarly, in the surgical treatment of syringomyelia, the placement of an intracavitary catheter to

drain the cyst into the subdural space raises the issue
of the safest entry zone for the catheter.
Given the dorsal approach to the spinal cord, SEPs
must be monitored continuously during the incision
of the dorsal midline. Bipolar coagulation can be used
to coagulate the veins on the surface of the cord and
usually this maneuver does not disrupt somatosensory
pathways. Then, the surgeon separates the dorsal columns to access the tumor. During the lateral displacement of the dorsal columns, SEPs often deteriorate,
usually first from the lower extremities because of
their somatotopic representation nearest to the midline
(Fig. 2). The SEP drop may be delayed because of the
need for averaging. Nevertheless, this should be recognized by the monitoring team and this information
should be passed to the surgeon. In many instances,
temporarily ceasing the retraction of the dorsal columns or irrigating the surgical field with warm saline
may be enough to facilitate SEP recovery. Sometimes
SEPs recover only later, at the end of the case, or not at
all. Late recovery is probably frequent as one may
detect only subtle sensory deficits or no deficits at all
postoperatively.
SEPs loss during the initial posterior myelotomy
may not correlate with a significant postoperative sensory deficit and also does not correlate to motor deficits. The preservation of SEPs is a good thing but
their loss during myelotomy is never used as the criterion to abandon surgery (Kothbauer et al., 1997;
Brotchi, 2002). After the dorsal surface of the tumor
has been exposed and the surgeon starts to work on
the cleavage plane between tumor and spinal cord,
attention must be paid to MEPs. The location of the
corticospinal tract in the anterolateral aspects of
the cord make them mechanically vulnerable while
detaching the tumor from the lateral aspects of the
cord.
For cervical tumors, we routinely monitor muscle
MEPs from all four extremities (abductor pollicis brevis, tibialis anterior, abductor hallucis) and, when
recordable, the D-wave from an epidural electrode
placed caudal to the tumor. Below the level of T1, only
muscles from lower extremities should be monitored
but, methodologically, we favor the adjunctive monitoring of an upper extremity muscle as a control
parameter to discriminate whether or not an intraoperative neurophysiological event is related to surgical maneuvers or general influences such as anesthesia
or cardiovascular factors. For the same reason,
638
SSEP-LPTN
SSEP-RPTN
10
20
30
40
50 ms
10
V
0
10
20
30
40
50 ms
Fig. 1. Dorsal column mapping in an 18-year-old patient with a syringomyelic cyst between the C2 and C7 segments of the
spinal cord (upper panel). Lower panels: placement of a miniature electrode over the exposed dorsal columns (middle
panel). Vertical bars on the electrode represent the position of the underlying exposed electrode surfaces. Sensory evoked
potentials after stimulation of the left (left panel) and right (right panel) posterior tibial nerves showing maximal amplitude
between electrodes 1 and 2. These data indicate that both dorsal columns have been dislocated to the extreme right side of
the cord. Using this information, the surgeon performed the myelotomy through the left side of the spinal cord and inserted
the shunt to drain the cyst. The patient did not experience postoperative sensory deficits. (Modified from Krzan, 2002 with
permission from Elsevier.)
whenever possible, we place an epidural electrode also

rostral to the tumor as a control recording.
During tumor removal, we periodically monitor
D-wave and myogenic motor evoked potentials
(mMEPs), sustaining the stimulation during the most
critical steps of the procedure. A schematic illustration
of MEP monitoring is presented in Fig. 3.
Based on our previous experience of several
hundreds of monitored ISCT surgeries, we are now
able to predict neurological outcome and therefore
to adopt intraoperative corrective measures based
on the criteria presented in Table 1.
If the recordings remain stable, there is no doubt
about the safety of the ongoing procedure and it
therefore can proceed quickly and efficiently. Thus,
stable recordings are a significant, important, and

useful information. If there is a loss of muscle MEPs,
the surgeon will stop resection and manipulation.
When judging mMEPs loss, we should consider that
the specificity of mMEP monitoring increases with
the number of muscles that are monitored. If, for
example, only the tibialis anterior muscle is monitored during surgery at the thoracic level, and this
mMEP is lost during tumor removal while the
D-wave is preserved, then a postoperative transient
paraplegia would be predicted. However, if we also
monitor the abductor hallucis, we may see that a
response from this muscle is still present.
Usually MEP changes occur toward the end of the
resection. Since most spinal cord tumors are resected
SPINE SURGERY
639
SEPs (posterior tibial nerve stimulation)

Right
Left
11:48
11:49
11:50
11:51
11:52
11:53
11:55
11:56
11:52
11:58
5 V
12:00
12:01
12:02
12:03
12:05
12:06
12:07
12:08
12:09
12:17
12:21
12:13
Myelotomy
50
100 ms
Fig. 2. SEP drop during myelotomy. Bilateral SEPs after posterior tibial nerve stimulation during removal of a spinal cord
tumor. Responses disappear during the initial myelotomy.
(From Deletis, 2001 with permission.)
in an inside-out and piecemeal fashion (with the

exception of ependymomas and hemangioblastomas),
direct manipulation or vascular compromise occurs
when the tumorcord interface is reached.
After mMEP loss, if recordings return right away,

the removal will be resumed at a different spot. If
they do not recover right away, the surgical field is
irrigated with warm saline solution. This dilutes the
intercellular potassium which may cause a block of
conduction, and generally clears out irritating blood
products and metabolites. If this does not work, then
increasing the mean arterial blood pressure by 10 to
15 mm Hg is considered. Still, even if muscle MEPs
do not return at this stage, the change only indicates a
transient motor deficit. If the D-wave amplitude is
still at baseline, it is justified to continue with the
resection.
Most of the times, muscle MEPs disappear first.
This may be preceded by an increase in threshold
for this particular muscle response. Decrement of
the D-wave amplitude usually followed changes in
mMEPs and occurs in a stepwise fashion. Sudden
decrease in D-wave amplitude, often coinciding with
sudden loss of muscle MEPs, is rare and believed to
be associated with some sort of vascular injury rather
than direct physical tissue manipulation.
Whenever the D-wave amplitude decreases and
reaches a level of 50% of its baseline value, then
the surgery will be terminated. If the resection is
not sufficiently completed at this stage to ensure a
good oncologic outcome, it may be necessary to
stage the procedure, that is to remove the rest of
the tumor in a second operation. This is of high
Epidural recording
Transcranial electrical
stimulation
C2
Cz
C4
D-wave
Single stimulus
0.5 ms
C1
C3
Muscle recording
6 cm
Train of 5 stimuli
4 ms
Fig. 3. Motor evoked potentials (MEPs) for spinal cord surgery. Left: schematic illustration of electrode positions for transcranial electrical stimulation of the motor cortex according to the International 10-20 EEG system. The site labeled 6 cm
is 6 cm anterior to Cz. Top right: schematic diagram of the position of the epidural catheter electrode placed caudal to the
lesion to monitor the incoming signal (D-wave) passing through the site of surgery. A single stimulus of 0.5 ms duration is
used. Bottom right: recording of mMEPs from the abductor pollicis brevis and tibialis anterior muscles after eliciting them
with a short train of electrical stimuli (multipulse technique), 4 ms apart. (Modified from Kothbauer et al., 2000 with permission from Blackwell Publishing.)
640
Table 1
Principles of combined MEP data interpretation and corrective measures
D-wave
Muscle MEPs
Corrective measures
Predicted outcome
Unchanged
Unchanged or
above 50%
Present
Present with minor
changes (decreased
amplitude or increased
threshold)
Lost uni- or bilaterally
None
Transiently move surgical
manipulation to a different
area; warm irrigation; correct
hypotension.
All the above, then transiently
stop surgery and/or improve
spinal cord blood flow (local
irrigation with papaverine). If
mMEPs do not reappear,
abandon surgery in selective
cases; as a rule surgery can
proceed.
Stop surgery immediately. If
D-wave does not recover,
abandon surgery.
All the above. If mMEPs do not
recover, abandon surgery.
Unchanged
Unchanged
Unchanged or
above 50%
Decreased >50%
Lost bilaterally
Unmonitorable
Lost bilaterally
Transient motor deficit

(affecting the involved
extremity)
Permanent motor deficit
Cannot differentiate
between transient and
permanent motor deficit
From Sala et al. (2006).
importance for spinal cord ependymomas where

complete resection is the main predictor for longterm progression free survival (Epstein et al., 1993).
MEPs recover frequently after ceasing the manipulation or at least with irrigation. Even a decline in
D-wave amplitude is potentially reversible with
warm irrigation and induced mild hypertension.
A particularly challenging situation occurs where
baseline mMEPs are present while the D-wave is
absent from the beginning. We have interpreted and
previously described this phenomenon which
occurs in 30% of the cases as a desynchronization of the D-wave (Morota et al., 1997; Kothbauer
et al., 1998). Desynchronization of the D-wave is more
common after irradiation of the spinal cord, when
ISCTs are associated to extended syringomyelic cysts,
or in the case of recurrent tumors, when the cord has
already sustained previous surgeries. It seems that,
under these circumstances, fast fibers of the corticospinal tracts conduct D-waves at different speeds and
therefore the recording epidural electrode will pickup only desynchronized descending volleys. When
the D-wave is desynchronized or the epidural electrode cannot be placed due to anatomical reasons
(e.g., dural scars due to previous surgery), monitoring
can rely only on mMEP. Unfortunately, mMEP loss

during tumor removal would not allow distinguishing
between a permanent and a transient postoperative
motor deficit. In this case, to continue surgery despite
mMEP disappearance exposes the patient to a significant risk of complete and permanent motor deficit.
Therefore, it is wise to err on the conservative side,
thus to leave residual tumor behind, which may be
accessible to a second surgical attempt rather than to
risk a significant and possibly permanent neurological
deficit.
To summarize, what we have found useful for
promoting the recovery of lost MEPs or deteriorated
SEPs during ISCT surgery can be easily recalled
using the acronym TIP: time, irrigation, papaverine/
pressure (blood pressure) (Sala et al., 2004).
The variable time is a critical one. We have consistently observed that if surgery is transiently stopped
immediately after mMEP has disappeared or D-wave
has significantly deteriorated, these potentials often
spontaneously recover. At this point, the spinal cord
is again able to sustain the further manipulation necessary to remove the remaining tumor. Conversely, to
ignore these events and continue or, even worse, speed
up the use of CUSA or any other cord manipulation
SPINE SURGERY
would likely transform a reversible injury into an irreversible one. We have, therefore, adopted a sort of
stop and go strategy that we did not use before the
neurophysiological feedback became available. As a
result, surgery can sometimes be transiently stopped
for half an hour or more, to allow mMEPs and/or
D-wave to recover; at that point further manipulation
of the cord is possible. Doing so, we attempt to adapt
the surgical strategy to the changes in the level of tolerance of the spinal cord along the procedure.
The mechanism behind the beneficial effect of
warm irrigation of the surgical field has not yet been
explained. Nevertheless, it is a common observation
that this application accelerates evoked potential
recovery. A possible explanation is that irrigation
affects the washout and dilution of extracellular potassium, which may accumulate with the disruption of
cell membranes and as a result of depolarization
(Young and Koreh, 1986; Young et al., 1989).
Local application of papaverine and increasing the
mean arterial pressure are both methods to improve
local perfusion to counteract an incipient ischemia.
Sometimes, MEPs are dramatically correlated with
blood pressure values and a sustained hypotension
may affect MEPs and unfavorably affect the outcome
(Sala et al., 1999).
Illustrative cases of MEP monitoring during ISCT
surgery are presented in Figs. 4-10.
Case 1: This is the case of a 40-year-old man who
presented with dysaesthesiae in the right arm and right
thoracic region, but no motor deficit. Preoperative spinal MRI disclosed a T3-T6 intramedullary spinal cord
tumor (Fig. 4). During tumor removal, we observed a
progressive deterioration of the D-wave together with
bilateral loss of tibialis anterior mMEPs (Fig. 5).
According to the criteria presented in Table 1, surgery
641
was transiently stopped and corrective measures were

taken. About 20 min later, D-wave recovered up to
70% of baseline values and we therefore encouraged
the surgeon to remove the residual tumor. By the end
of the procedure, mMEPs from tibialis anterior muscles also recovered (Fig. 6) and the patient had no significant motor impairment after surgery.
Case 2: A 11-year-old boy presented with a onemonth history of progressive left hemiparesis, right
cervicobrachialgia and urge urinary incontinence. Spinal MRI disclosed a C2-C4 ISCT (Fig. 7A). During
tumor removal, mMEPs were monitored throughout
the procedure, while an epidural catheter for monitoring of the D-wave was placed after the laminotomy.
At the dural opening, mMEPs from right and left extensor digitorum longus (EDL) and abductor hallucis
(ABH) were present. D-wave was also recordable caudally to the level of the lesion from an epidural catheter
placed at about C6. Baseline D-wave amplitude was
9.4 mV (Fig. 8). During tumor resection, mMEP from
the left ABH disappeared but D-wave amplitude was
not significantly decreased. By the time when almost
90% of the tumor was removed, D-wave amplitude
was still similar to baseline. Shortly thereafter, however, while attempting to remove the most ventral part
of the tumor, also the mMEP from the left EDL disappeared while the response from the right EDL was fluctuating but still present. At this point, surgery was only
transiently stopped and corrective measures were taken
such as warm irrigation, to favor mMEP recovery.
Nevertheless, since D-wave amplitude remained stable
and above the 50% threshold, the decision was made
to continue surgery and a total removal of the tumor
was achieved. Just before the epidural catheter was
removed, D-wave amplitude was 8.8 mV. By the end
of the procedure, left ABH mMEP recovered and
Fig. 4. Illustrative case 1 of MEP monitoring during ISCT surgery (reprinted from Sala et al., 2004 with permission from
Springer-Verlag Wien). Axial (left) and sagittal (right) contrast enhanced T1-weighted MR images of a T3-T6 spinal cord
ependymoma with associated syringomyelia.
642

mMEPs
RT
LT
D-wave
7.8 ms
11:26
11:28
Opening: 15.2 V
11:30
13:25
13:33
13:41
13:50
13:56
14:01
14:04
14:12
Myelotomy: 14.2 V
Removal: 15.1 V
h. 14.10
Lower pole: 10.4 V
14:32
14:38
14:44
h. 14.50
14:52
h. 16.00
White matter: 6.87 V
(decrease > 50%)
14:55
h. 15.40
15:00
15:24
15:38
15:39
100 V
15:40
Time
Irrigation
Papaverine
T.I.P.
10 ms
Total removal: 7.90 V

Closing: 10.57 V
Final: 11.32 V
Fig. 5. Motor evoked potential (MEP) monitoring during surgery for the spinal cord ependymoma. Muscle MEPs from
right (RT) and left (LT) tibialis anterior muscles, after TES (left panel), and D-wave monitoring from an epidural electrode
inserted caudally to the lesion (right panel). Same TES stimulation parameters are used for mMEPs and D-wave monitoring.
At 14.50, the LT mMEP is lost and D-wave has decreased to about 65% of initial values (10.4 vs. 15.2 mV). At 15.40, also
the RT mMEP disappeared, but D-wave amplitude was still above the 50% drop threshold. Therefore, the decision was
made to attempt the removal of the last piece of the tumor, which was adherent to the anterolateral spinal cord white matter.
Shortly thereafter, the D-wave amplitude dropped to <50% of initial values (6.87 vs. 15.2 mV). Surgery was therefore
stopped and corrective measures were taken. About 20 min later, after warm irrigation and local infusion of papaverine,
the D-wave recovered to more than 50% of initial values and it was possible to carefully proceed to a complete tumor
removal.
D-wave
mMEPs
A
RA C1/2 5ST 70MA OP
RA C1/2 7ST 50MA CL
LA C2/1 5ST 70MA OP
LA C2/1 7ST 50MA CL
RT C2/6 5ST 90MA OP
RT C2/6 7ST 120MA CL
LT C2/6 5ST 90MA OP
25 V
22 V
7.8
A
31 V
C2/6 90MA OP
16.22 V
7.6
C
29 V
77 V
C2/6 90MA CL
11.32 V
61 V
43 V
O
0.0
5.0
10.0
15.0
20.0
25.0
30.0 ms
LT C2/6 7ST 140MA CL

45 V
10
20
30
40
50
60
70
80
100 ms
Fig. 6. D-wave (left panel) and mMEPs from the bilateral abductor pollicis brevis (RA and LA) and tibialis anterior (RT
and LT) muscles (right panel), before (OP) and after (CL) tumor removal. The D-wave amplitude declined throughout the
critical part of the procedure but at the end recovered to about 70% of initial values. Muscle MEPs were preserved at the
end of the procedure. The patient had no significant postoperative motor deficits.
SPINE SURGERY
Fig. 7. Illustrative case 2 of MEP monitoring during ISCT

surgery (reprinted from Sala et al., 2006 with permission
from Lippincott, Williams and Wilkins). Preoperative spinal
MRI. Contrast-enhanced T1-weighted image showing a
C2-C4 intramedullary spinal cord tumor (A). Postoperative
spinal MRI at three months of follow-up. Contrast-enhanced
T1-weighted image showing no residual tumor (B).
only left EDL was still absent (Fig. 9). Immediately

postoperatively, the patient presented a transient
worsening of the left hemiparesis. At discharge, 10
days after surgery, he was able to walk with a cane
and to move upper and lower limbs against gravity.
At follow-up three months after surgery, his neurological examination was completely normal (Fig. 10) and
the MRI showed no evidence of residual disease or
recurrence (Fig. 7B).
45.3.3. How sensitive should warning signs be?
Besides the combined mMEP/D-wave monitoring protocol that we have been using over the past 10 years,
other MEP monitoring strategies have then appeared
in the literature.
Calancie et al. (1998, 2001) suggested thresholdlevel parameters during multipulse TES to assess intraoperative muscle MEPs changes. One problem with
threshold-level criteria is that threshold to elicit muscle
MEPs after TES is highly variable because muscle
MEPs are generated through a polysynaptic pathway
643
and are very sensitive to the effects of anesthesia.

Wide variation in amplitude and latency can therefore
be expected even in the neurologically intact patient
(Jones et al., 1996). This variability also explains the
lack of a linear correlation between intraoperative
changes in muscle MEPs amplitude, and/or latency,
and the motor outcome in spinal cord surgeries.
More recently, Quinones-Hinojosa et al. (2005)
have also suggested muscle MEP criteria only to predict postoperative outcome during ISCT surgery. In
their study, alterations in muscle MEP morphology
(from polyphasic to biphasic and from biphasic to
loss) correlated with motor grade loss in the immediate postoperative period, at discharge and at followup. This study, where the D-wave was not monitored,
introduces mMEPs waveform morphology as an
additional criterion to assess motor pathways integrity different from the all-or-none criterion proposed
by others (Zentner et al., 1989; Jones et al., 1996;
Kothbauer et al., 1998; Pelosi et al., 2001; Dong
et al., 2002; Sala et al., 2006). Eight patients, in this
study, had muscle MEP changes from polyphasic to
biphasic waves at the end of the procedure, and their
maximum motor grade worsening at discharge as
compared to preoperatively was from grade 5 to
grade 4. Changes in the mMEP morphology, therefore, correlated with only a mild degree of motor
impairment, which was mostly transient and the
relevance of which is highly questionable when the
patient has not even been mobilized postoperatively.
Furthermore, when these warning mMEP criteria were
used, the rate of total tumor removal in the entire
series was 57%. This rate compares unfavorably with
other series (Xu et al., 1996; Constantini et al., 2000;
Hanbali et al., 2002; Raco et al., 2005), especially if
we consider that ependymomas are amenable of gross
total removal and must be eradicated to achieve cure.
Further refinement of mMEPs criteria to predict
postoperative outcome and prevent neurological deficits is clearly desirable. However, the balance
between strict and loose warning criteria would
invariably reflect the balance between false-positive
and false-negative results. It has been claimed that
MEP monitoring may be too sensitive, actually stopping resection too early (Albright, 1998). From a
neurosurgical standpoint, the possibility of a mild,
often transient, motor impairment may be an acceptable price for the patient to pay, if this is rewarded by
a complete tumor removal that, in the case of spinal
cord ependymomas, cavernomas, and hemangioblastomas, means complete cure of the disease.
644

R EDL
L EDL
R ABH
L ABH
Time
D-wave
12.00 Dura opening

12.55 CUSA
13.20 L ABH mMEP loss
14.00 90% tumor out

14.05 L EDL mMEP loss
10 V
15.00 Tumor out
4.6
ms
30 V
Fig. 8. Motor evoked potential (MEP) monitoring during surgical removal of the tumor. Left panel: mMEP monitoring
showing progressive disappearance of the response from left abductor hallucis (L ABH) (time 13.20) and left extensor digitorum longus (L EDL) (time 14.05). Response from the right extensor digitorum longus (R EDL) fluctuated on and off
throughout the procedure but was still present at the end. The right abductor hallucis (R ABH) mMEP remained stable.
Response from the L ABH returned by the end of surgery (see Fig. 9). Right panel: D-wave monitoring showing a stable
D-wave amplitude during surgery, without critical drop below the 50% threshold. CUSA Cavitron ultrasonic surgical
aspirator.
R EDL opening
R EDL closing
L EDL opening
L EDL closing
R ABH opening
R ABH closing
L ABH opening
30 V
L ABH closing
0 10 20 30 40 50 60
ms
Fig. 9. Opening and closing mMEPs from upper and

lower extremities. At the end of surgery, all mMEPs from
the four extremities were present except the left EDL. At
discharge, the patient was ambulatory with assistance.
Opening response recorded at the beginning of the surgical procedure, before opening of the dura. Closing
response recorded at the end of the surgical procedure,
during muscle closure. R EDL right extensor digitorum longus; L EDL left extensor digitorum longus;
RABH right abductor hallucis; L ABH left abductor
hallucis.
Fig. 10. At 3-month follow-up, the child is neurologically

intact with no residual motor deficit.
SPINE SURGERY
In our opinion combining mMEPs and D-wave

monitoring, when available, remains the gold standard
for ISCT surgery at this time because it supports a more
aggressive surgery in an attempt to achieve a complete
tumor removal. Without D-wave monitoring, only a
less reliable prognostic evaluation is available, and if
only quantitative (threshold or waveform dependent)
muscle MEP criteria are used to stop surgery, this
appears to decrease the rate of transient motor deficits,
but decreases the rate of total tumor removal.
Very recently, Skinner et al. (2005) reported the use
of free-running EMG as a method to detect early motor
tract injury during ISCT surgery. Free-running EMG
is a routine method to monitor cranial nerves in posterior fossa surgery, but its use in ISCT surgery is a new
proposal. In this study, changes in free-running EMG
anticipated transcranial electrical MEP changes in three
cases. In other two patients, changes in free-running
EMG were the only intraoperative finding while MEPs
remained unchanged. Both patients presented a mild
postoperative worsening, which completely recovered
at follow-up. This report represents the first clue to the
possibility of using free-running EMG criteria to
improve the warning threshold and predictive value of
monitoring techniques during ISCT surgery. However,
the small number of patients (14 cases) does not allow
any conclusion at this time about the real value of
this approach. Furthermore, the fact that both patients
with seemingly falsely negative muscle MEPs, but
identified by free-running EMG, had a complete recovery at the follow-up, raises again the issue of how sensitive an ideal MEPs technique should be. Finally,
it should be recognized that, under some circumstances,
the warning provided by intraoperative MEP changes
remains of only documentary value: while the removal
of an astrocytoma can be terminated without jeopardizing the patients neurologic and oncologic outcome,
the resection of a hemangioblastoma is an all-or-none
enterprise. The lesion has to be entirely removed, no
matter what the MEP parameters indicate, or serious
bleeding and/or swelling would lead to a certain damage
of the spinal cord. This, of course, is a limitation
imposed by the nature of the tumor, rather than a shortcoming of the monitoring technique.
45.4. Does monitoring make a difference?
The real impact of neurophysiological monitoring on
the neurological outcome after ISCT surgery remains
debated. The report of control studies to validate the
efficacy of IOM techniques in neurosurgery remains
anecdotal and mostly limited to historical control
645
studies (Radtke et al., 1989). In fact, those neurosurgeons who operate with the assistance of IOM, and
believe in its efficacy to prevent neurological deficit,
would not accept a prospective randomized study
given the ethical and medicolegal concerns of designating a control group. In the field of ISCT surgery, the rarity of this pathology further limits the
chance for a prospective study on the role of IOM.
So far, therefore, support for MEP monitoring as
an essential component of spinal cord tumor surgery
has been based more on the lack of bad results than
on documented better outcomes of patients who have
benefited from IOM when compared with an unmonitored control group.
In a retrospective study on a small population of
spinal cord ependymomas, operated over a 37-year
period, neurological outcome in patients operated
with the aid of a microscope and IOM was compared
with that of patients operated before these tools were
available (Asazuma et al., 1999). The authors concluded that the microscope and IOM were indispensable for improving outcome. However, the kind of
monitoring used (SEPs or MEPs) was not specified
and there were no data to statistically support the
advantages of IOM versus those of microsurgery.
We recently tried to address the question concerning the real impact of IOM by comparing the neurological outcome of 50 patients operated on with the
assistance of IOM (SEPs, mMEPs, D-wave) with that
of 50 patients selected from 301 ISCTs previously operated on by the same team without IOM (Sala et al.,
2006). We matched the two groups by the preoperative
neurological status, tumor histology, location, and
extent of removal. Matching was blind to outcome. A
>50% SEP amplitude decrement influenced only myelotomy. Muscle MEP disappearance modified surgery
but >50% D-wave amplitude decrement was the major
indication to stop surgery. The postoperative to preoperative neurological status at discharge and at a follow-up
of at least three months was compared between the two
groups. Motor outcome was significantly better in the
monitored group at the follow-up, while at discharge
from hospital there was only a trend toward better outcome. Our interpretation was that early motor outcome
is similar due to transient motor deficits in IOM group,
which have not fully recovered by the time the patient
leaves the hospital.
A further consolidation of the invaluable role of
MEPs during ISCT surgery will also rely on MRI studies. Today, the question may rise whether or not the use
of intraoperative neuromonitoring results in a more
conservative approach because the deterioration of
646
the corticospinal tracts and elicit EMG activity from

upper and lower extremity muscles. By repeating the
stimulation at different thresholds, they were able to
follow and preserve the descending motor tracts adjacent to the tumor. At the end of tumor removal, the
tracts were restimulated to confirm their integrity.
Duffau et al. (1998) applied a similar method to two
more patients, further improving the accuracy of the
stimulation by reducing the distance between the tips
of the bipolar probe (from 1 to 0.2 mm). While in these
four anecdotal patients, no significant motor deficits
were observed after surgery, this methodology suffers
from the same limitations of the so-called neurogenic
and myogenic MEP methods where a motor response
is recorded from either peripheral nerves or limb muscles after stimulation of the spinal cord. Whether or not
this response represents a true stimulation of motor
pathways and not an antidromic potential from the
electrically stimulated dorsal column, remains
debated. Therefore, as already described by Minahan
et al. (2001) and Toleikis et al. (2000), the possibility
of false-negative results exists (namely, a postoperative motor deficit in spite of preserved positive spinal
cord motor mapping at the end of tumor resection).
Deletis and Bueno De Camargo (2001) have
recently proposed an alternative method based on
the so-called collision technique (Fig. 11).
the neurophysiological signals may stop surgery

before a radical removal is reached. Based on the subjective opinion of those neurosurgeons operating on
ISCT with the assistance of IOM, the opposite seems
to be true. Nevertheless, opinions need to be supported
by postoperative MRI data demonstrating that the rate
of residual tumor (especially for astrocytomas) is equal
or even smaller in monitored patients. The results of
our study did not answer this question because the
same rate of total removal in the monitored and not
monitored group was a prerequisite of the study.
45.4.1. New perspectives in neurophysiological
monitoring during ISCT surgery
Unlike brain surgery where mapping techniques have
been much more extensively applied than monitoring
techniques, the opposite seems to be true for ISCT
surgery where the use of mapping techniques is only
anecdotally reported.
Dorsal column mapping has been described previously. The possibility to map corticospinal tracts, within
the spinal cord, would also further enhance the safety of
surgery and likely increase the extent of resection.
So far, two methods have appeared in the literature.
Quinones-Hinojosa et al. (2002) described in two
patients the use of bipolar 50-Hz technique to map
TES
C2 C1
S1
S1
S1
S2+S1
D1
S2
S2+S1
D2
D2
D1
50 V
50V
0
10
15
20
25
30
35
Negative mapping - no collision, D1 = D2
40ms
10
15
20
25
30
35
40ms
Positve mapping - collision occurs, D2 = 42 % of D1
Fig. 11. Mapping of the CT corticospinal tract by a D-wave collision technique. S1 transcranial electrical stimulation
(TES); S2 spinal cord electrical stimulation (SpES); D1 control D-wave (TES only); D2 D-wave after combined
stimulation of the brain and spinal cord; R D-wave recording electrode in the spinal epidural space. Below left: negative
mapping results (D1 D2). Below right: positive mapping results (D-wave amplitude significantly diminished after collision). Reprinted from Deletis and Camargo (2001) with permission from Karger, Basel.
SPINE SURGERY
This method is based on the fact that the D-wave

descending in the corticospinal tracts after transcranial electrical stimulation can collide with and be
annihilated by an anti-D-wave elicited by electrical
stimulation of the exposed spinal cord. If the surgeons hand-held stimulating probe comes in close
proximity to the corticospinal tract (TES and handheld probe electrical stimulation of the spinal cord
are performed simultaneously), a D-wave decrement
or complete disappearance will occur during collision. Although still in the early stages of its development, this methodology proved highly successful in
identifying and localizing the corticospinal tract
within the spinal cord.
45.5. Conclusion
Only a decade ago, intraoperative neurophysiological
monitoring was considered of limited value during
ISCT surgery because motor and sensory deficits can
occur independently and SEPs were not reliable to provide information on motor pathways. Since then, our
neurophysiological techniques and knowledge on the
pathophysiology of these tumors have significantly
improved. The reliability of a combined SEPMEP
method has dramatically enhanced the value of monitoring. Different MEP methods and warning criteria
still exist (mMEP, D-wave, absence/presence criteria,
amplitude criteria, morphology criteria, . . .) but future
studies will likely clarify which method provides the
strongest reliability to support both an aggressive
tumor removal while preserving the long-term neurological outcome of the patient.
References
Albright, AL (1998) Intraoperative spinal cord monitoring
for intramedullary surgery: an essential adjunct?
Pediatr. Neurosurg., 29: 112.
Asazuma, T, Toyama, Y, Suzuki, N, Fujimura, Y and Hirabayshi, K (1999) Ependymomas of the spinal cord and
cauda equina: an analysis of 26 cases and a review of
the literature. Spinal Cord, 37: 753759.
Boyd, SG, Rothwell, JC, Cowan, JMA, Webb, PJ, Morley,
T, Asselman, P and Marsden, CD (1986) A method of
monitoring function in corticospinal pathways during
scoliosis surgery with a note on motor conduction velocities. J. Neurol. Neurosurg. Psychiatry, 49: 251257.
Brotchi, J (2002) Intrinsic spinal cord tumor resection.
Brotchi, J, Dewitte, O, Levivier, M, Baleriaux, D, Vandesteene, J, Raftopoulos, C, Flament-Durand, J and
647
Noterman, J (1991) A survey of 65 tumors within the
spinal cord: surgical results and the importance of preoperative magnetic resonance imaging. Neurosurgery,
29: 651656; discussion 656657.
Burke, D, Hicks, RG and Stephen, JPH (1990) Corticospinal volleys evoked by anodal and cathodal stimulation
of the human motor cortex. J. Physiol., 425: 283299.
Green, BA (1998) Threshold-level multipulse transcranial electrical stimulation of motor cortex for intraoperative monitoring of spinal motor tracts: description
of method and comparison to somatosensory evoked
potential monitoring. J. Neurosurg., 88: 457470.
HJ (2001) Threshold-level repetitive transcranial electrical stimulation for intraoperative monitoring of central motor conduction. J. Neurosurg. (Spine 1), 95:
161168.
Constantini, S and Epstein, FJ (1996) Ultrasonic dissection in
neurosurgery. In: RH Wilkins and SS Rengachary (Eds.),
Neurosurgery. McGraw-Hill, New York, pp. 607608.
Constantini, S, Miller, DC, Allen, JC, Rorke, LB, Freed, D
and Epstein, FJ (2000) Radical excision of intramedullary
spinal cord tumors: surgical morbidity and long-term
follow-up evaluation in 164 children and young adults.
J. Neurosurg. (Spine), 93: 183193.
Deletis, V (2001) Neuromonitoring. In: DG McLone (Ed.),
Pediatric Neurosurgery: Surgery of the Developing
Nervous System. Saunders, Philadelphia, pp. 12041213.
Deletis, V and Bueno De Camargo, A (2001) Interventional
neurophysiological mapping during spinal cord procedures. Stereotact. Funct. Neurosurg., 77: 2528.
Dong, CC, MacDonald, DB and Janusz, MT (2002) Intraoperative spinal cord monitoring during descending thoracic and thoracoabdominal aneurysm surgery. Ann.
Duffau, H, Capelle, L and Sichez, J (1998) Direct spinal
cord electrical stimulations during surgery of intramedullary tumoral and vascular lesions. Stereotact. Funct.
Neurosurg., 71: 180189.
Edmonds, HL, Paloheimo, MPJ, Backman, MH, Johnson,
JR, Holt, RT and Shields, CB (1989) Transcranial magnetic motor evoked potentials (tcMMEP) for functional
monitoring of motor pathways during scoliosis surgery.
Spine, 14: 683686.
Elsberg, CA and Beer, E (1911) The operability of intramedullary tumors of the spinal cord. A report of two
operations with remarks upon the extrusion of intraspinal tumors. Am. J. Med. Sci., 142: 636647.
Engler, GL, Spielholz, NI, Bernhard, WN, Danziger, F,
Merkin, H and Wolff, T (1978) Somatosensory evoked
potentials during Harrington instrumentation for scoliosis.
J. Bone Joint Surg., 60 A: 528532.
Epstein, F (1984) The Cavitron ultrasonic aspirator in
tumor surgery. Clin. Neurosurg., 31: 497505.
648
Epstein, FJ and Farmer, JP (1990) Pediatric spinal cord
tumor surgery. Neurosurg. Clin. N. Am., 1: 569590.
Epstein, FJ, Farmer, JP and Freed, D (1992) Adult intramedullary astrocytomas of the spinal cord. J. Neurosurg.,
77: 355359.
Epstein, FJ, Farmer, JP and Freed, D (1993) Adult intramedullary spinal cord ependymoma: the result of surgery in 38
Fennelly, ME, Taylor, BA and Hetreed, M (1993) Anaesthesia and the motor evoked potential. In: SJ Jones, S Boyd,
M Hetreed and NJ Smith (Eds.), Handbook of Spinal
Cord Monitoring. Proceedings of the Fifth International
Symposium on Spinal Cord Monitoring, London, UK,
June 25, 1992.Kluwer Academic Publishers, Dordrecht,
pp. 272276.
Fischer, G and Brotchi, J (1996) Intramedullary Spinal
Cord Tumors. Thieme, Stuttgart, pp. 115116.
Fischer, G, Brotchi, J and Mahla, K (2005) Surgical management of intramedullary spinal cord tumors in adults.
In: H Schmidek and D Roberts (Eds.), Schmidek &
Sweet Operative Neurosurgical Techniques: Indications, Methods, and Results. Saunders Elsevier, Philadelphia, pp. 19451954.
Ginsburg, HH, Shetter, AG and Raudzens, PA (1985) Postoperative paraplegia with preserved intraoperative somatosensory evoked potentials. J. Neurosurg., 63: 296300.
Gowers, WR and Horsley, V (1888) A case of tumour of
the spinal cord. Removal and recovery. Med. Chir.
Trans., 53: 377428.
Greenwood, J, Jr. (1967) Surgical removal of intramedullary tumors. J. Neurosurg., 26: 276282.
Hanbali, F, Fourney, DR, Marmor, E, Suki, D, Rhines, LD,
Weinberg, JS, McCutcheon, IE, Suk, I and Gokaslan,
ZL (2002) Spinal cord ependymoma: radical surgical
resection and outcome. Neurosurgery, 51: 11621174.
Jallo, GI, Kothbauer, KF and Epstein, FJ (2001) Intrinsic spinal cord tumor resection. Neurosurgery, 49: 11241128.
Jallo, GI, Kothbauer, KF and Epstein, FJ (2002) Contact laser
microsurgery. Childs Nerv. Syst., 18: 333336.
Jellinek, D, Jewkes, D and Symon, L (1991) Noninvasive
intraoperative monitoring of motor evoked potentials
under propofol anesthesia: effect of spinal surgery on
the amplitude and latency of motor evoked potentials.
Jones, SJ, Harrison, R, Koh, KF, Mendoza, N and Crockard, HA (1996) Motor evoked potential monitoring during spinal surgery: responses of distal limb muscles to
transcranial cortical stimulation with pulse trains. Electroencephalogr. Clin. Neurophysiol., 100: 375383.
Kalkman, CJ, Drummond, JC, Ribberink, AA, Patel, PM,
Sano, T and Bickford, RG (1992) Effects of propofol,
etomidate, midazolam and fentanyl on motor evoked
responses to transcranial electrical or magnetic stimulation in humans. Anesthesiology, 76: 502509.

Katayama, Y, Tsubokawa, T, Maemjima, S, Hirayama, T
and Yamamoto, T (1988a) Corticospinal direct response
in humans: identification of the motor cortex during intracranial surgery under general anesthesia. J. Neurol. Neurosurg. Psychiatry, 51: 5059.
Katayama, Y, Tsubokawa, T, Yamamoto, T and Maejima, S
(1988b) Spinal cord potentials to direct stimulation of the
exposed motor cortex in humans: comparison with data
from transcranial motor cortex stimulation. In: PM Rossini and CD Marsden (Eds.), Non-Invasive Stimulation
of Brain and Spinal Cord. Alan R. Liss, Inc., New York,
pp. 305311.
Kearse, LA, Jr., Lopez-Bresnahan, M, McPeck, K and
Tambe, V (1993) Loss of somatosensory evoked potentials during intramedullary spinal cord surgery predicts
postoperative neurologic deficits in motor function [corrected] [published erratum appears in J. Clin. Anesth.,
1993; 5(6): 529]. J. Clin. Anesth., 5: 392398.
Kothbauer, K, Deletis, V and Epstein, FJ (1997) Intraoperative spinal cord monitoring for intramedullary surgery: an
essential adjunct. Pediatr. Neurosurg., 26: 247254.
Kothbauer, KF, Deletis, V and Epstein, FJ (1998) Motorevoked potential monitoring for intramedullary spinal
cord tumor surgery: correlation of clinical and neurophysiological data in a series of 100 consecutive procedures. Neurosurgery, Focus 4: Article 1.
Koyanagi, I, Iwasaki, Y, Isu, T, Abe, H, Akino, M and
Kuroda, S (1993) Spinal cord evoked potential monitoring after spinal cord stimulation during surgery of spinal
cord tumors. Neurosurgery, 33: 451460.
Krzan, M, Deletis, V and Isgum, V (1996) Intraoperative
neurophysiological mapping of dorsal columns. A new
tool in the prevention of surgically induced sensory deficit? Electroencephalogr. Clin. Neurophysiol., 102: 37P
(abstract).
Lang, EW, Beutler, AS, Chesnut, FM, Patel, PM, Kennelly,
NA, Kalkman, CJ, Drummond, JC and Garfin, SR
(1996) Myogenic motor-evoked potential monitoring
using partial neuromuscular blockade in surgery of the
spine. Spine, 21: 16761686.
Lesser, RP, Raudzens, P, Luders, H, Nuwer, MR, Goldie, WD,
Morris, HH, Dinner, DS, Klem, G, Hahn, JF, Shetter, AG,
Ginsburg, HH and Gurd, AR (1986) Postoperative neurological deficits may occur despite unchanged intraoperative somatosensory evoked potentials. Ann. Neurol., 19:
2225.
Levy, WJ, York, DH, McCaffrey, M and Tanzer, F (1984)
Motor evoked potentials from transcranial stimulation of
the motor cortex in humans. Neurosurgery, 15: 287302.
McCormick, PC and Stein, BM (1990) Intramedullary
tumors in adults. Neurosurg. Clin. N. Am., 1: 609630.
McCormick, PC, Torres, R, Post, KD and Stein, BM (1990)
Intramedullary ependymoma of the spinal cord. J. Neurosurg., 72: 523532.
SPINE SURGERY
Merton, PA and Morton, HB (1980) Stimulation of the cerebral cortex in the intact human subject. Nature, 285: 227.
Miller, DC (2000) Surgical pathology of intramedullary spinal
cord neoplasms. J. Neurooncol., 47: 189194.
Kostuik, JP (2001) Anterior spinal cord injury with preserved neurogenic motor evoked potentials. Clin.
Morota, N, Deletis, V, Constantini, S, Kofler, M, Cohen, H
and Epstein, FJ (1997) The role of motor evoked potentials during surgery for intramedullary spinal cord
Nash, CL, Lorig, RA, Schatzinger, L and Brown, RH
(1977) Spinal cord monitoring during operative treatment of the spine. Clin. Orthop. Relat. Res., 126:
100105.
Patton, HD and Amassian, VE (1954) Single and multiple
unit analysis of cortical stage of pyramidal tract activation. J. Neurophysiol., 17: 345363.
transcranial electrical stimulation during two anaesthetic
regimens. Clin. Neurophysiol., 112: 10761087.
Philips, CG and Porter, R (1964) The pyramidal projection
to motoneurones of some muscle groups of the
baboons forelimb. In: JC Eccles and JP Schade (Eds.),
Progress in Brain Research. Elsevier, Amsterdam,
pp. 222243.
Quinones-Hinojosa, A, Gulati, M, Lyon, R, Gupta, N and
Yingling, C (2002) Spinal cord mapping as an adjunct
for resection of intramedullary tumors: surgical technique with case illustrations. Neurosurgery, 51:
11991207.
Quinones-Hinojosa, A, Lyon, R, Zada, G, Lamborn, KR,
Gupta, N, Parsa, AT, McDermott, MW and Weinstein,
PR (2005) Changes in transcranial motor evoked potentials during intramedullary spinal cord tumor resection
correlate with postoperative motor function. Neurosurgery, 56: 982993.
Raco, A, Esposito, V, Lenzi, J, Piccirilli, M, Delfini, R and
Cantore, G (2005) Long-term follow-up of intramedullary spinal cord tumors: a series of 202 cases. Neurosurgery, 56: 972981.
39: 187191.
Motor evoked potentials during brain surgery. Pfl
ugers
Arch., 431: R291R292.
649
Roonprapunt, C, Silvera, VM, Setton, A, Freed, D, Epstein,
FJ and Jallo, GI (2001) Surgical management of isolated
hemangioblastomas of the spinal cord. Neurosurgery,
49: 321328.
Sala, F, Niimi, Y, Krzan, MJ, Berenstein, A and Deletis, V
(1999) Embolization of a spinal arteriovenous malformation: correlation between motor evoked potentials
and angiographic findings: technical case report. Neurosurgery, 45: 932937; discussion 937938.
Sala, F, Palandri, G, Basso, E, Lanteri, P, Deletis, V,
Faccioli, F and Bricolo, A (2006) Motor evoked potential
monitoring improves outcome after surgery for intramedullary spinal cord tumors: a historical control study.
Schmid, UD, Boll, J, Liechti, S, Schmid, J and Hess, CW
(1992) Influence of some anesthetic agents on muscle
responses to transcranial magnetic cortex stimulation:
a pilot study in man. Neurosurgery, 30: 8592.
Skinner, S, Nagib, M, Bergman, T, Maxwell, R and
Msangi, G (2005) The initial use of free-running electromyography to detect early motor tract injury during
resection of intramedullary spinal cord lesions. Neurosurgery, 56: 299314.
Sloan, TB (2002) Intraoperative neurophysiology and anesthesia management. In: V Deletis and J Shils (Eds.), Neurophysiology in Neurosurgery: A Modern Intraoperative
Approach. Academic Press, San Diego, pp. 451474.
Stein, B and McCormick, P (1992) Intramedullary neoplasms and vascular malformation. Clin. Neurosurg.,
39: 361387.
Taniguchi, M, Schramm, J and Cedzich, C (1991) Recording of myogenic motor evoked potentials under general
anesthesia. In: J Schramm and AR Mller (Eds.),
Intraoperative Neurophysiologic Monitoring in Neurosurgery. Springer, Berlin, pp. 7287.
Taniguchi, M, Cedzich, C and Schramm, J (1993a) Modification of cortical stimulation for motor evoked potentials under general anesthesia: technical description.
Taniguchi, M, Nadstawek, J, Langenbach, U, Bremer, F and
Schramm, J (1993b) Effects of four intravenous anesthetic
agents on motor evoked potentials elicited by magnetic
transcranial stimulation. Neurosurgery, 33: 407415.
Theodotou, BC and Powers, SK (1986) Use of intraoperative ultrasound in decision making during spinal operations. Neurosurgery, 19: 205211.
Toleikis, JR, Skelly, JP, Carlvin, AO and Burkus, JK (2000)
Spinally elicited peripheral nerve responses are sensory
rather than motor. Clin. Neurophysiol., 111: 736742.
ber die chirurVon Eiselsberg, A and Ranzi, E (1913) U
gische Behandlung der Hirn- und Ruckenmarkstumoren. Arch. Klin. Chir., 102: 309468.
650
Von Eiselsberg, A and Marburg, O (1917) Zur Frage der
Operabilitat intramedullarer Ruckenmarkstumoren.
Arch. Psychiatr. Nervenkr., 59: 453461.
Whittle, I, Johnston, I and Besser, M (1986) Recording
of somatosensory evoked potentials for intraoperative
spinal cord monitoring. J. Neurosurg., 64: 601612.
Wood, E, Berne, A and Taveras, J (1954) The value of radiation therapy in the management of intrinsic tumors of the
spinal cord. Radiology, 63: 1124.
Xu, Q, Bao, W, Mao, R and Yang, G (1996) Aggressive
surgery for intramedullary tumor of cervical spinal cord.
Surg. Neurol., 46: 322328.

Young, W and Koreh, I (1986) Potassium and calcium
changes in injured spinal cords. Brain Res., 365:
4253.
Young, W, Rosenbluth, J, Wojak, JC, Sakatani, K and Kim,
H (1989) Extracellular potassium activity and axonal
conduction in spinal cord of the myelin-deficient mutant
rat. Exp. Neurol., 106: 4151.
Zentner, J, Kiss, I and Ebner, A (1989) Influence of
anesthetics nitrous oxide in particular on
electromyographic response evoked by transcranial
electrical stimulation of the cortex. Neurosurgery,
24: 253256.

M.R. Nuwer (Ed.)
651
CHAPTER 46
Neurophysiological monitoring during spinal

endovascular procedures
Yasunari Niimi*, Vedran Deletis and Alex Berenstein
Center for Endovascular Surgery and Intraoperative Neurophysiology, Institute for Neurology and Neurosurgery, Roosevelt Hospital 10G,
New York, NY 10019, USA
46.1. Introduction
All vascular lesions involving the spine, spinal cord,
and surrounding tissues are potential candidates for
endovascular embolization. The indications include
preoperative treatment to decrease vascularity and
therefore intraoperative blood loss, palliation for incurable diseases and curative therapy by embolization
alone. Various embolic agents are used depending on
the purpose of the treatment and the nature of the disease. In order to avoid neurological complications, it
is important to preserve the blood supply to the normal
spinal cord during embolization. For this purpose, it is
essential to superselectively catheterize the feeder to
the lesion and carefully analyze the vascular anatomy.
All spinal endovascular procedures are performed
under general anesthesia and neurophysiological monitoring in our institution. The role of neurophysiological
monitoring varies depending on the nature of the disease and procedures performed. We will discuss neurophysiological monitoring for endovascular treatment of
spine and spinal cord vascular lesions.
46.2. Vascular anatomy of the spine and spinal
cord (Fig. 1A and B)
The vascular supply to the spine and paraspinal musculature arises from the main trunk of the intercostal or
lumbar artery as well as the dorsospinal artery (Chiras
et al., 1979). Vascular supply to the spinal dura and
spinal cord is derived from the ventral division of the
dorsospinal artery. There are rich longitudinal and
*
Correspondence to: Yasunari Niimi, M.D., D.MSc.,

Center for Endovascular Surgery, INN, Roosevelt Hospital
Suite 10G, 1000 Tenth Avenue, New York, NY 10019,
USA.
Tel.: 1-212-636-3400; fax: 1-212-636-3296.
E-mail: yniimi@chpnet.org (Y. Niimi).
transverse anastomoses between the adjacent segmental arteries. Longitudinal anastomotic vessels connect
branches of the segmental arteries to adjacent branches
above and below. Transverse anastomotic vessels connect right and left segmental arteries across the midline. Both longitudinal and transverse anastomotic
vessels can be outside or inside the spinal canal. Nerve
roots and the spinal dura are supplied by the radicular
artery arising from each segmental artery. Progressing
caudally from the intercostal to the lumbar levels,
there is increasing obliquity of both the nerve roots
and the radicular arteries due to differences in the
growth rate between the spine and the spinal cord. If
a radicular artery supplies the anterior spinal artery
(ASA), it is called a radiculomedullary artery and if it
supplies the posterior spinal artery (PSA), it is called
a radiculopial artery. A radiculomedullary artery and
a radiculopial artery may have a common trunk.
There are 48 radiculomedullary arteries and 1020
radiculopial arteries at the end of spinal cord vascular
development. The ASA extends almost uninterrupted
from the medulla to the filum terminale. It may be focally
discontinuous, especially at the thoracic level. At the cervicomedullary junction it originates from the two vertebral arteries near the vertebro-basilar junction. Other
major radiculomedullary arteries arise, at the level of
the cervical enlargement, from vertebral, deep cervical,
or ascending cervical arteries. Additional sources of supply at the craniocervical junction include the ascending
pharyngeal artery and occipital artery. The thoracolumbar territory is supplied mainly by the arteria radicularis
anterior magna or artery of Adamkiewicz. This usually
rises from the 9th to the 12th intercostal artery, on the left
side in 80% of the cases. It gives off a small ascending
branch and a large descending branch which anastomoses with the posterior spinal arteries to configurate
the anastomotic basket surrounding the conus medullaris
(Lazorthes et al., 1971; Thron, 1988).
652
Y. NIIMI ET AL.
Fig. 1. Schematic illustration of spinal cord arteries (A) and veins (B). A: 1: radiculomedullary artery; 2: anterior radicular
artery; 3: posterior radicular artery; 4: anterior spinal artery; 5: posterior spinal artery; 6: transverse and longitudinal interconnections (vasa corona) giving radial perforating arteries; 7: sulcal arteries. B: 1: radicular vein; 2: anterior radicular vein;
3: posterior radicular vein; 4: anterior median vein; 5: posterior median vein; 6, 7: additional discontinuous longitudinal
veins. Modified from Thron (1988) with permission from Springer-Verlag, Vienna.
The paired posterior spinal arteries arise, at the

cervical level, either from the vertebral artery or from
the posteroinferior cerebellar artery, depending on
the origin of the posterior inferior cerebellar artery
(intradural or extradural). They are often ventral to
dorsal nerve roots and are called lateral spinal
arteries. Caudally, these paired posterior spinal axes
also receive radiculopial feeders from the vertebral,
the intercostal, and lumbar arteries, and are located
on the posterolateral surface of the cord dorsal to
the dorsal roots. Circumferential vessels from the
ASA anastomose with the PSAs through a complex
pial network, the so-called vasa corona (Herren and
Alexander, 1939) which supplies the peripheral rim
of the white matter and represents a functionally relevant dorso-ventral connection.
Intrinsic blood supply to the spinal cord is from
sulcal arteries ventrally and radial perforating arteries
from the circumferential pial network or vasa corona.

Each sulcal artery supplies one side (right or left) of
the ventral portion of the spinal cord. They anastomose with adjacent caudal and rostral sulcal arteries
within the ventral sulcus. They supply up to the
50% of the section at the level of the spinal cord
enlargement but only 1520% at the thoracic level.
Because their supply includes the anterior column
of the central gray matter, the anterior and lateral
corticospinal tracts (CT), and the anterior and lateral
spinothalamic tracts, the ASA accounts for vascularization of the structures involved in the propagation
of motor evoked potentials (MEPs) from their cortical generators to the a-motoneurons: anterior and lateral corticospinal tracts. In contrast, the PSAs supply
the posterior horns of the central gray matter and
the dorsal columns, and though still debated, these
posterior columns are usually considered the main
SPINE SURGERY
tracts for central propagation of somatosensory

evoked potentials (SEPs) after peripheral stimulation
(Cohen et al., 1981). There are rich axial and longitudinal anastomoses among these intrinsic arteries.
These anastomoses are richer in the thoracic level
and more potent adjacent to the spinal cord arteriovenous malformation.
The venous drainage of the spinal cord is characterized by rich intramedullary and extramedullary anastomoses. The extramedullary veins include the pial
venous networks, the longitudinal collectors, and the
radicular veins. The pial network consists of longitudinal veins collecting intrinsic venous perforators. Ventral and dorsal sulcal veins drain blood from the gray
matter. The longitudinal collector veins are seen on
the ventral and dorsal aspect of the spinal cord. At
the cervical and lumbar levels, these collectors are single, in the midline on the anterior and posterior surfaces of the spinal cord (the anterior and posterior
median spinal veins). At the thoracic level, three longitudinal channels may be seen. These longitudinal veins
draining to the radicular veins may be angiographically opacified by injection in a large radiculomedullary
artery (Lasjaunias et al., 2001).
The radicular veins can be along the ventral or
dorsal nerve root and pierce the dura to drain into
the epidural veins. They pierce the dura together with
the nerve roots in only 60% of the time and through a
separate hole for the remaining 40%. These radicular
veins lack valves but typically narrow at the dural
penetration to prevent retrograde venous flow (Lasjaunias et al., 2001).
Epidural venous plexus is small dorsally and large
ventrally. The ventral plexus drains the vertebral
bodies and appears as hexagonal channels anastomosing across the midline and with the caudal and
rostral systems, and continues from the sacrum to
the base of the skull. Flow in the epidural veins is
affected by gravity, position, and changes in the
abdominal and intrathoracic pressure, which in turn
affects the venous drainage of the spinal cord below
the heart level (Lasjaunias et al., 2001).
46.3. Spinal angiography
Because of the segmental arrangement during embryonic development, particularly in the cervical region,
several vessels must be angiographically evaluated to
delineate the vascular supply of the spine and the spinal cord. At the cervical level, the ascending cervical
artery, the vertebral artery, and the deep cervical
653
artery on both sides must be studied. Additionally,

at the C1C2 levels, the ascending pharyngeal and
occipital arteries should also be studied. For the thoracolumbar levels, angiographic evaluation of the
bilateral supreme intercostal, intercostal, and lumbar
arteries need to be studied. At the sacral level, bilateral lateral sacral and iliolumbar arteries arising from
the internal iliac artery as well as the median sacral
artery may supply the sacral nerve roots, spinal cord,
vertebrae, and parasacral musculature.
As previously mentioned, the spinal cord is
supplied by one anterior midline ASA and two posterolateral paramedian PSAs. Angiographically, the
radiculomedullary artery has a characteristic hairpin
configuration which continues to the ASA that
appears as a midline continuous longitudinal straight
vessel. The ascending limb of the ASA may be
opacified from a large radiculomedullary artery
(Fig. 2A). The ASA continues caudally to the filum
terminale and forms basket-shaped anastomoses with
bilateral PSAs at the level of the conus (Lazorthes
et al., 1971). The PSA appears as a relatively small
paramedian longitudinal straight vessel. The PSA
axis is smaller and discontinuous compared with the
ASA axis. The radiculopial artery also forms a hairpin configuration that has a more acute angle than
the radiculomedullary artery does because of its paramedian location (Fig. 2B and C).
Spinal angiography and the subsequent endovascular treatment are best performed under general
anesthesia. This not only provides the patients with
comfort but also allows for extended periods of
apnea (up to 40 s) and thus provides the opportunity
to obtain high-resolution images necessary to identify
small spinal cord arteries and to evaluate slow flow
lesions. For lower thoracic and lumbar lesions, glucagon may also be administered to limit bowel motion
that degrades the image quality.
It is important to identify spinal cord arteries and
differentiate them from surrounding extradural vessels supplying osteomuscular structures. In order to
identify the midline ASA and paramedian PSAs, it
is mandatory to perform spinal angiography in the
exact posterioranterior projection. This is sometimes impossible due to the distortion of the spinal
column and spinal cord from previous treatment or
pathology, either related or unrelated to the target
lesion. In such a case, oblique and lateral views
may be necessary to identify the spinal cord vessels.
We recently use three dimensional (3D) angiography
to assess the 3D structure of the normal and abnormal
654
Y. NIIMI ET AL.
Fig. 2. Typical angiographic appearance of spinal cord arteries in a 19-year-old female patient. A: Left T9 intercostal artery
injection in the posterioranterior (PA) view showing a radiculomedullary artery with ascending and descending limbs of
the anterior spinal artery (ASA, arrow heads). The descending limb reaches to the conus. There is retrograde opacification
of another contribution to the ASA from the right L1 level (small arrows). Compare with Fig. 1C. Notice hemivertebral
brush of the T11 vertebral body (large arrows). B: Left T11 intercostal artery injection in the PA view showing radiculopial artery with ascending and descending limbs of the left posterior spinal artery (PSA, arrow heads). There is anastomosis
to the right T11 intercostal artery with opacification of the right PSA (arrows). C: Right L1 lumbar artery injection in the
PA view showing a common trunk of radiculomedullary and radiculo-pial artery (arrows). This radiculomedullary artery is
opacified on the left T9 injection. Compare with Fig. 1A.
vessels and their relationship with the surrounding

bony structures (Prestigiacomo et al., 2003).
In general, pretherapeutic spinal angiography
should evaluate the vascular anatomy of both the
pathology and the surrounding normal spinal cord.
More details of our angiographic protocol will be discussed under each disease category.
46.4. Neurophysiological monitoring and
pharmacological provocative testing
Intraoperative neurophysiological monitoring has been
used to assess the functional integrity of neural pathways during endovascular procedures of spine and spinal cord vascular lesions under general anesthesia. We
started using SEPs in the mid-1980s based on the clinical evidence that SEPs were sensitive to compromizes
in the ASA circulation (Berenstein et al., 1984). Concerns about the reliability of SEPs in evaluating the
integrity of descending motor tracts during spine and
spinal cord surgery (Ginsburg et al., 1985; Lesser
et al., 1986), as well as during aortic surgery, have

then appeared in the literature (Takaki and Okumura,
1985; Zornow et al., 1990). The occurrence of motor
deficit despite intraoperatively unchanged SEPs is
explained by the limited ability of SEPs to assess the
functional integrity of CT (Minahan et al., 2001;
Pelosi et al., 2002; Jones et al., 2003). Therefore, we
introduced MEPs in addition to SEPs. One of the
major problems of previously reported technique in
monitoring MEPs was its invasiveness with the need
to place the recording electrodes through a burr hole
to elicit MEPs and to use epidurally inserted electrodes to record MEPs (Katayama et al., 1991). One
of the coauthors (V.D.) has established a protocol for
a reliable noninvasive technique of transcranial cortical stimulation along with recording from peripheral
muscles in the mid-1990 (Kothbauer et al., 1997,
1998a; Sala et al., 1999, 2001).
We now routinely use SEPs and MEPs monitoring
for all spinal angiography procedures with intent to
treat by embolization. These cases include spinal cord
SPINE SURGERY
655
arteriovenous malformations (AVMs), spinal dural

arteriovenous fistulas (AVFs), and spine tumors. This
monitoring is useful for the early detection of ischemia due to blockage of flow in the spinal cord artery
either by vasospasm or by a microcatheter itself, or
Time
TA left
during particle embolization. A case demonstrating

the high sensitivity of MEPs monitoring is shown in
Fig. 3 (Sala et al., 1999).
In the following sections we describe the protocol
currently used at the Institute for Neurology and
Time
TA left
11:32
12:18
10:54
10:56
10:57
13:10
13:11
10:58
11:01
11:03
11:04
11:31
13:12
100 V
20 ms
100 V
Time
B
TA left
20 ms
Time
TA left
13:32
13:33
14:32
13:34
14:39
13:40
14:46
13:42
14:48
13:43
14:49
13:44
14:50
100 V
100 V
20 ms
20 ms
Fig. 3. A 28-year-old male patient presented with progressive weakness and numbness of both lower extremities and urinary
retention. A: Left: Normal motor evoked potentials (MEPs) recorded from the left anterior tibialis (TA) muscle. Right: Left
T11 intercostal angiogram showing intramedullary arteriovenous malformation (AVM) (arrows) supplied by the anterior spinal
artery (ASA, arrow heads). B: During superselective catheterization of the ASA, MEPs from the TA muscle disappeared (left).
Complete flow arrest in the ASA (arrow heads) distal to the tip of the microcatheter (arrow) was noted (right). C: Following
quick particle embolization of the AVM, the microcatheter was removed with temporary partial improvement of MEPs and flow
in the ASA (not shown). This was followed by complete re-disappearance of MEPs from the left TA muscle (left). Left T11
angiogram demonstrating no opacification of the ASA due to severe vasospasm (right). D: This was treated by superselective
infusion of papaverine into the radiculomedullary artery with resultant complete recovery of MEPs (left). Angiogram showed
complete resolution of vasospasm with opacification of the normal ASA and minimal remaining opacification of the AVM
due to embolization (right). Modified from Sala et al. (1999) with permission from Lippincott, Williams and Wilkins.
656
Neurosurgery to perform multimodal neurophysiological monitoring during endovascular treatments.

46.4.1. Neurophysiological monitoring
The basic setup of the patient does not differ from
that occurring during standard neurosurgical procedures. The only difference is that the angiographic
machine rotates around the axis of the patient. Therefore, the careful placement of cables for evoked
potentials stimulation and recording is necessary to
avoid stretching of the wires or interference with
the angiographic steps.
During angiographic assessment and embolization, the patients are maintained under general
anesthesia using continuous infusion of propofol
(100150 mg/kg/min) and fentanyl (1 mg/kg/h). Short
acting muscle relaxant is used for intubation purpose
only. After anesthesia induction, no inhalational
anesthetics are used.
SEPs monitoring was performed in a conventional
method (Deletis and Engler, 1997). Briefly, SEPs
were elicited by stimulating the right and the left posterior tibial nerves at the ankle and median nerves at
the wrist with electric stimuli (40 mA, 0.2 ms duration, and 4.3 Hz repetition rate). Recordings are performed via corkscrew-like electrodes inserted
subcutaneously in the scalp (CS electrodes Nicolet,
Madison, WI), respectively, at C30 /C40 -Fz (median
nerve) and at Cz0 -Fz (tibial nerve) according to the
International 10/20 EEG system.
The MEPs are elicited with transcranial electrical
stimulation of the motor cortex using CS electrodes.
Short trains of 57 square-wave stimuli of 0.5 ms
duration each and interstimulus intervals of 4 ms
are applied at a repetition rate of 2 Hz and intensity
up to 200 mA, through electrodes placed at C1 and
C2 scalp sites, according to the International 10/20
EEG system. MEPs are then recorded via needle
electrodes inserted into the abductor pollicis brevis
(APB) for the upper extremities and anterior tibial
(TA) and abductor hallucis (AHB) muscles for the
lower extremities, bilaterally. Recording from the
upper extremity muscles were used as a control for
embolization of thoracic or lumbar lesions. This technique has been used at our institution to monitor over
100 spinal cord tumor cases and is described in detail
in the following references (Deletis and Kothbauer,
1998; Kothbauer et al., 1998a).
Whenever the lesion involves the lumbosacral segments of the spinal cord, we add the monitoring of the
Y. NIIMI ET AL.
bulbocavernosus reflex (BCR) to the battery of neurophysiologic tests (Deletis and Vodusek, 1997).
This oligosynaptic reflex allows the assessment of
the functional integrity of both the afferent and efferent fibers of the pudendal nerves together with the
reflex center located in the gray matter at S2S4 spinal segments. For the stimulation of the dorsal penile
nerve (pudendal afferents), two silver/silver chloride
disk electrodes are placed on the dorsal aspect of the
penis with the cathode proximal. In female patients,
the cathode is placed over the clitoris and the anode
over the labia major. Rectangular pulses of 0.2
0.5 ms duration are applied as a train of 5 stimuli
(interstimulus intervals of 4 ms) at a repetition rate
of 2.3 Hz. Stimulus intensities do not exceed
40 mA. Recordings are made from the anal sphincter
muscles using two pairs of intramuscular Tefloncoated hooked wire electrodes stripped 2 mm at the
tip inserted into the anal hemisphincters. The technical details are described elsewhere (Deletis and
Vodusek, 1997).
With regard to the feasibility of evoked potentials during endovascular procedures, results from
our series demonstrate that these potentials are easily elicitable in the majority of the patients, unless
severe neurological deficits have already compromised the functional integrity of neural pathways.
Monitorable SEPs often cannot be obtained in
patients who lose proprioception. In over 110 endovascular procedures in 87 patients who were treated
for spine and/or spinal cord vascular lesions
between 1996 and 1999, the monitorability of
evoked potentials was 80% for SEPs, 85% for the
BCR, and 92% for MEPs. Monitorability is defined
as the presence of a reliable response after the
induction of anesthesia but before any interventional procedures. There were no significant differences in monitorability between males and females
for MEPs and SEPs, while the BCR seemed more
difficult to elicit in females, most likely because
of technical difficulties in placing stimulating electrodes (Sala and Niimi, 2002). Absent BCR is so far
well correlated with symptoms of bladder, bowel,
or sexual dysfunction. Clinical usage of BCR monitoring for the spine and spinal cord embolization is
still under investigation.
46.4.2. Provocative testing
In addition to the careful angiographic analysis, pharmacological provocative testing is a method used to
SPINE SURGERY
identify the functional eloquence of the territory of a

catheterized vessel. This is done by estimating the
influence of pharmacological agents to the parameter
of neurophysiological signals. It was originally performed by clinically assessing neurological status of
an awake patient after the injection of a short-acting
anesthetic from a microcatheter which has been placed
in a feeding artery before embolization (Berenstein et al.,
1990). However, for the spine and spinal cord embolization procedures, we prefer general anesthesia to
control patient breathing in order to obtain highresolution images to identify small spinal cord vessels as well as for the comfort of the patient during
this potentially long procedure. In order to assess
the neurological status of the patient under general
anesthesia, we use neurophysiological monitoring
during pharmacological provocative testing.
Pharmacological provocative testing consists of
intra-arterial (IA) injection of short-acting barbiturates (e.g., sodium amytal) and lidocaine through a
microcatheter. A low dose of a short-acting barbiturate (e.g., sodium amytal) predominantly suppresses
neuronal activity (Doppman et al., 1986) as opposed
to a low dose of lidocaine, which predominantly suppresses axonal conduction in the central nervous system (Tanaka and Yamasaki, 1966). Therefore, both
agents theoretically should be used for the better reliability of provocative testing. In our series, lidocaine
caused more positive results than sodium amytal did.
The most likely reason for this observation is because
lidocaine blocks the nerve transmission through the
fibers traversing the anesthetized area, as opposed
to sodium amytal that blocks transmission within
neurons in the anesthetized area.
The motor pathway affected by the injection of
sodium amytal may not be detected because of a
limited number of muscles monitored with MEPs
technique. We, therefore, have added monitoring of
important muscles for specific cases, such as diaphragm for high-cervical cord lesions. For radicular
feeders to a malformation, we also place additional
electrodes to monitor MEPs from the muscles supplied by the nerve root at the level of the lesion.
Although we have not yet experienced positive
results from the provocative testing involving these
additional MEPs, the number is still small and further
investigation is necessary to assess their usefulness.
Provocative testing has the most important role for
the embolization of intradural spinal cord AVMs
among all spine and spinal cord vascular lesions,
because they are supplied by the spinal cord arteries
657
and embolization for these lesions carries the highest

risk of causing spinal cord ischemia. This test can be
performed for dural or extradural lesions, if the existence of a spinal cord artery distal to the tip of the
microcatheter cannot be confidently excluded before
embolization. Provocative testing does not need to
be performed if the operator is confident enough for
the safety of injection of the embolic agent from
the microcatheter placed for embolization. We perform less provocative testing for fistulous malformations than nidus type malformations. In cases with
previous interventions and an altered anatomy, we
find these testings of great value.
SEPs and MEPs are sensitive to the depth of anesthesia even without using muscle relaxant or inhalational anesthetics. Therefore, it is important to
obtain baseline recordings just before the provocative
testing. Before provocative testing, superselective
digital subtraction angiography (DSA) is performed to
study the normal and abnormal vascular anatomy
through the microcatheter placed as close as possible
to the AVM at the location intended for embolization.
Contrast material is then injected under road map
fluoroscopy, in order to determine the optimal force
for injection which will distribute the anesthetic
distal enough without creating reflux proximal to
the tip of the microcatheter. Provocative testing for
neuronal function follows this by the IA injection
of 50 mg of sodium amytal. If there are no changes
in SEPs or MEPs, this is then followed by the IA
injection of 2040 mg of lidocaine, depending on
the feeder size and the degree of shunting. If there
are still no changes in SEPs or MEPs, embolization
is performed using a liquid embolic agent such as
N-butyl cyanoacrylate (NBCA) from that catheter
position. If over 50% decrease in the amplitude of
SEPs and/or disappearance of MEPs occurred after
injection of sodium amytal or lidocaine, the provocative test is considered positive and liquid embolization from that catheter position is not performed. If
sodium amytal produced a positive result, the test is
considered positive and lidocaine is not injected. If
the lidocaine test is positive following a negative
amytal test, a second provocative testing in the same
vascular territory is performed by injecting only
lidocaine, after advancing the microcatheter further
distally or after protecting the normal territory by
placing a liquid coil as shown in Fig. 4 (Sala et al.,
1999).
If a provocative test is positive, it is to be assumed
that there is higher possibility of creating functional
658
Y. NIIMI ET AL.
Rt PTN CSEPs
baseline
before Xyl. I
Xylocaine test I
1 after Xyl. I
Rt PTN CSEPs
4 after Xyl. I
5 after Xyl. I
8 after Xyl. I
baseline
Coiling procedure
Xylocaine test II
1 after Xyl. II
3 after Xyl. II
4 after Xyl. II
6 after Xyl. II
Saline test
Xylocaine test IV
1 after Xyl. IV
4 after Xyl. IV
4 after saline
5 after saline
Xylocaine test III
Embolization
5 after Xyl. III

6 after Xyl. III
50 ms
glue injection
closing baseline
0.4 v
3 after Xyl. III

4 after Xyl. III
0.4 v
before coiling
coiling
5 after coiling
50 ms
Fig. 4. A 46-year-old male presented with progressive paraparesis, urinary, and fecal incontinence, following sudden onset
of lower back pain. A: Right L2 lumbar artery angiogram demonstrating an intramedullary arteriovenous malformation
(AVM) at the T12 level supplied by the posterior spinal artery (PSA, arrow heads). This vessel was superselectively catheterized for embolization. B: Superselective angiogram of the right PSA showing the AVM. No normal spinal cord supply is
identified on this study. The arrow indicates the tip of the microcatheter. Provocative testing from this catheter position was
positive with disappearance of somatosensory evoked potentials (SEPs) from the right posterior tibialis nerve (PTN). Repeat
testing was also positive and saline injection from the same catheter position did not cause any change in SEPs (see F).
C: The microcatheter was further advanced distally (arrow). Repeat superselective angiogram demonstrated distal portion
of the AVM as well as normal PSA on both sides (small arrow heads). There is also anastomotic opacification of the anterior spinal artery (ASA) (curved arrow) with the deviation of its proximal portion (large arrow heads) due to the AVM.
D: The normal right PSA was protected by placement of a microcoil. Nonsubtracted image demonstrating a microcoil
placed to protect the distal normal PSA (arrows). The large arrow indicates the tip of the microcatheter which was brought
back after the placement of the microcoil. The arrowheads indicate N-butyl cyanoacrylate (NBCA) cast due to prior embolization from the ASA. Repeat provocative testing from the catheter position in D was negative (see G) and embolization
was performed using NBCA from this catheter position. E: Postembolization control angiogram of the right T11 intercostal
artery demonstrating small residual nidus of the AVM. The right PSA distal to the AVM (arrows) is supplied by anastomotic vessels (arrow heads) from the left PSA. No change in SEPs was seen after the embolization (see G). F and G: Trace
of the provocative testing using Xylocaine and SEPs from the right PTN. The patient was neurologically unchanged after
the embolization. Modified from Sala et al. (2000) with permission from Interventional Neuroradiology.
SPINE SURGERY
damage to the spinal cord by embolization using a

liquid material from that catheter position. Therefore,
some counter-measurement should be undertaken in
such a situation. The best solution is to advance the
microcatheter closer to the nidus and repeat provocative testing. If the second provocative test is negative,
embolization can be performed using a liquid agent.
Another solution is to protect the normal territory using
a fiber or liquid coil. If, after blocking the normal territory, the repeated provocative test is negative, liquid
embolization can be performed (Sala et al., 2000). If
neither of the above is possible, embolization may still
be performed using particles, depending on the flow
dynamics in the feeder. If none of these alternatives
are possible, embolization from another feeder should
be considered (Niimi et al., 2004).
Pharmacological provocative testing under neurophysiological monitoring is a useful adjunct and
makes embolization of a spine and spinal cord vascular lesions under general anesthesia safer. However,
this test is in no meaning to replace careful angiographic analysis of vascular anatomy.
46.5. Clinical application of neurophysiological
monitoring
46.5.1. Vascular malformations
Vascular malformations can be simply classified into
dural/extradural and intradural lesions. This distinction is important because the risk of embolization
and the role of neurophysiological monitoring are
significantly different between these two categories.
46.5.1.1. Intradural vascular malformations
Intradural vascular malformations are supplied by spinal cord arteries and drained by spinal cord veins.
They are further classified into spinal cord AVMs,
AVFs, telangiectasias, and cavernous malformations.
Endovascular embolization is indicated and is the first
choice of treatment for spinal cord AVMs and AVFs.
Embolization is usually curative for simple AVMs
and AVFs, but palliative or rarely curative for complex or extensive AVMs. Palliative embolization is
targeted to occlude dangerous structures such as
aneurysms or high-flow fistulas and is performed to
decrease the risk of hemorrhage or to improve neurological symptoms. Endovascular treatment for this
disease group is considered as a potentially high-risk
procedure because embolization is performed through
the ASAs or PSAs that supply the normal spinal cord
659
as well as the lesion. A liquid embolic agent, such as

NBCA, is preferred for nidus AVMs and small fistulas
due to its ability to penetrate distally and cause permanent occlusion. For large fistulas and associated
aneurysms, coils are also effective. Particles are used
mainly for small AVMs or AVFs for which distal
catheterization through the feeder is difficult
(Djindjian et al., 1978; Berenstein et al., 1990; Biondi
et al., 1990; Niimi and Berenstein, 1999).
Angiographic identification of spinal cord vascular supply in the presence of a spinal cord AVM
may be difficult, despite magnification or additional
lateral and oblique views, due to their small sizes
and overlapping normal and pathological vessels. In
addition, supply to the normal spinal cord adjacent
to the malformation can be modified and unpredictable due to the hemodynamic changes caused by
the spinal cord AVM or the previous treatment.
Neurophysiological monitoring including provocative testing is most important in this disease
category because of the high-risk nature of the treatment. The main purposes of the monitoring are early
detection of spinal cord ischemia by continuous monitoring of SEPs, MEPs, and BCRs, and prediction of
the safety of embolization from a certain microcatheter position by performing provocative testing. Spinal cord ischemia due to compromised spinal cord
vascular supply during the embolization procedure
can occur not only by the injection of embolic agents
but also by the catheterization of a feeder, either due
to blockage of the flow by the catheter itself or
spasm, or dissection created by catheter manipulation
(Sala et al., 1999). An example of early detection of
compromised spinal cord vascular supply by MEP
monitoring and its treatment is demonstrated in
Fig. 3.
In a rare situation where vascular anatomy of the
lesion is so complicated that it is difficult to differentiate a feeder to the malformation and a normal spinal
cord artery, provocative testing can be used as an aid
for the analysis of the vascular anatomy. It should be
emphasized, however, that the availability of provocative testing does not decrease the importance of
precise angiographic analysis of the vascular anatomy of the malformation and surrounding normal
spinal cord. Worsening of MEPs and SEPs after
embolization usually correlates with clinical deterioration. Improvement of MEPs or SEPs after embolization, however, does not necessarily correlate with
clinical improvement for intradural AVFs and
AVMs (Sala et al., 2000). Further accumulation of
660
Y. NIIMI ET AL.
experience is needed to assess the role of neurophysiological monitoring for the prediction of clinical
outcome.
We are presenting our recently published data on
spinal cord arteriovenous malformation (SCAVM)
embolization using this monitoring technique (Niimi
et al., 2004). Eighty-four spinal cord angiography
procedures with intent to treat by embolization were
performed for 52 patients with a SCAVM since
1996. Endovascular embolization was performed in
the same setting as the angiography whenever feasible. During these procedures, 60 provocative testings
were performed.
46.5.1.1.1. Results. SEPs and MEPs monitoring
was attempted in all 84 procedures. Embolization
was performed in 48 procedures. In the rest of 36
procedures, embolization was not performed because
there was no feeder feasible for embolization, or
embolization was attempted but aborted because it
was impossible to catheterize distal enough to consider embolization or because provocative tests were
positive. Monitorable SEPs were obtained in 66.5%
and MEPs in 83.9%. BCR was monitorable in
76.7% of attempted cases. Total 60 provocative tests
were performed and 19 positive results (31.7% positive) were obtained. Forty-seven amytal tests were
performed with 4 (8.5%) positive results and 56 lidocaine tests were performed with 15 (26.8%) positive
results (Table 1). The summary of the positive tests
is shown in Table 2. The positive sodium amytal test
occurred after one injection in the PSA and two
injections in the ASA feeders resulting in loss of
MEPs. In one patient with a conus AVM, bilateral
BCRs and MEPs were lost without changes in SEPs
after the injection of sodium amytal into the ASA
feeder. The positive lidocaine tests included six

injections in the PSA (loss of MEPs in 2, SEPs in
1, MEPs and BCR in 2, and MEPs and SEPs in 1),
eight injections in the ASA (loss of MEPs in 8),
and one injection in the posterior inferior cerebellar
artery with loss of MEPs. Only one injection of an
anesthetic (sodium amytal or lidocaine) resulted in
loss of both MEPs and SEPs. Each positive test
resulted in either unilateral or bilateral loss of MEPs
or SEPs. It was unpredictable if either unilateral or
bilateral responses would be lost in the case of a positive result, based on the superselective angiogram
from the microcatheter. In one patient, injection of
lidocaine into an ASA feeder of the cervical spinal
cord AVM resulted in transient disappearance of
MEPs from both the upper extremities without
changes in the lower extremity MEPs.
After positive provocative testing and disappearance of MEPs or SEPs, they recovered to the baseline
value within 15 min for the initial testing in all
patients. Recovery of MEPs or SEPs tended to be
delayed up to 1 h after the second or the third injection of sodium amytal or lidocaine in the same
patient.
Out of 19 positive provocative tests, embolization
was aborted in 6 patients, in which two patients had
2 consecutive positive results on repeated provocative tests by advancing a microcatheter distally.
Embolization with NBCA was possible in five
patients by further advancing a microcatheter, in
which one patient had two previous consecutive positive results during the advancement of a microcatheter. One patient was embolized with diluted
particles and another patient was embolized with
coils. In one patient, we performed a clinical provocative test to further clarify a positive result in the
Table 1
Summary of provocative testing results
Sodium amytal
ASA
PSA
PICA
RA
Total
Lidocaine
No. of vessels
No. of positive
No. of vessels
No. of positive
32
13
1
1
47
3
1
0
0
4
37
15
1
3
56
8
6
1
0
15
ASA: anterior spinal artery; PSA: posterior spinal artery; PICA: posterior inferior cerebellar artery; RA: radicular artery.
Source: Modified from Niimi et al. (2004) with permission. Copyright # American Society of Neuroradiology.
SPINE SURGERY
661
Table 2
Summary of positive of provocative test results
Agent
Vessel studied
Changes
Sodium amytal
PSA
ASA
ASA
Lidocaine
PSA
Unilateral MEP
Unilateral MEP
Bilateral BCR
and unilateral MEP
Bilateral MEP
Bilateral MEP and SEP
Unilateral MEP
Unilateral SEP
Unilateral MEP and BCR
Unilateral MEP
Bilateral MEP
Unilateral MEP
ASA
PICA
Total
No. of procedures
1
2
1*
1
1
1
1
2
4
4
1
19
*
MEPs and SEPs were monitored from only one leg because of previous amputation of the other leg.
PSA, posterior spinal artery; ASA, anterior spinal artery; PICA, posterior inferior cerebellar artery; SEP, somatosensory evoked potential;
MEP, motor evoked potential.
provocative test using electrophysiological monitoring. We woke the patient from general anesthesia,
injected the same pharmacological agent (lidocaine),
and assessed for the development of new neurological signs (wake-up test). We felt that a wake-up test
was indicated in this particular case of a young
patient with progressive neurological deterioration,
because we were able to achieve an optimal catheter
position for embolization. This patient was safely
embolized with NBCA because there were no
changes in neurological signs by the wake-up test.
In one patient, a positive result was found to be due
to reflux of lidocaine to the ASA, which had a
common origin with the PSA feeder. This patient

was embolized with gentle antegrade NBCA injection without reflux. In one patient, a normal PSA
was found distally by advancing the microcatheter.
This case was embolized with NBCA after protecting
the normal territory by placement of liquid injectable
microcoils (Fig. 3 and Table 3) (Sala et al., 2000).
No patient showed worsening of symptoms after
the embolization performed based on the results of
the provocative tests, except one patient who developed transient worsening of spasticity after NBCA
embolization through an ASA feeder following a
negative provocative test.
Table 3
Actions taken after positive provocative tests
Aborted embolization
6*
Advanced catheter and embolization with NBCA

Distal vessel protection and embolization with NBCA
Embolization with particles
Embolization with coils
Embolization with NBCA after negative wake-up test
Embolization with NBCA with less forceful injection
5**
1
1
1
1
1
In two patients, the procedure was aborted after 2 consecutive positive tests during advancement of a microcatheter.
In one patient, embolization was performed by advancing a microcatheter from the proximal positions where 2 positive provocative tests
were obtained. NBCA: N-butyl cyanoacrylate.
**
662
46.5.1.1.2. Data analysis. It is noteworthy that one

injection of sodium amytal and five injections of
lidocaine in the PSA resulted in loss of MEPs.
In contrast, changes in SEPs occurred only in two
injections in the PSA, one of which resulted in
changes in both MEPs and SEPs. All positive results
from the injection of sodium amytal or lidocaine in
the ASA caused loss of MEPs without changes in
SEPs. Our observation suggests that the motor pathway can be affected by the superselective injection
of anesthetics through the PSA. This may be due to
the rich anastomosis between the ASA and the PSA
branches especially in the existence of an AVM.
There may also be hemodynamic shift of the watershed zone between the ASA and the PSA territories,
either due to the existence of the AVM or due to a
previous embolization procedure. Among six patients
in which MEPs were lost by provocative testing from
a PSA feeder, four patients had supply from both
ASA and PSA. One patient had supply to the AVM
from the contralateral PSA across the midline and
another patient had a common trunk between the
ASA and the PSA. Four patients angiographically
showed rich anastomosis between the ASA and the
PSA, and two patients had a previous embolization
procedure. Although there was no patient in whom
SEPs changed after the injection of anesthetics in
the ASA in this series, it is known to occur in our
experience (Berenstein et al., 1984, 2004a).
Our data suggest that both SEPs and MEPs should
be monitored regardless of whether the provocative
test is performed in the ASA or in the PSA.
The negative predictive value of our provocative
test is very high (97.6%) even if we consider one
case of transiently increased spasticity after embolization as a false negative. Using current techniques
of MEPs monitoring, subtle changes such as increase
in spasticity without a worsening of motor strength
cannot be reliably predicted. If a provocative test is
negative, however, we can consider that the malformation can be safely embolized from that catheter
position using a liquid embolic agent without causing
damage to the patient as shown in Fig. 4.
The numbers of true positive and false positive are
unknown, because we generally do not embolize the
malformation with a liquid embolic agent from that
catheter position if a provocative test is positive.
Therefore, sensitivity, specificity, accuracy, and positive predictive value cannot be calculated. Theoretically, false positive results can occur because of the
different distribution pattern of the liquid embolic
Y. NIIMI ET AL.
material from the anesthetics due to different viscosity and different injection force as well as progressively polymerizing nature of the embolic material
as opposed to persistently liquid nature of anesthetics. False negative results can also happen for
the same reasons, but it is very rare in our experience. This may be due to the relatively large doses
of sodium amytal (50 mg) and lidocaine (40 mg) to
anesthetize a small territory distal to the tip of the
microcatheter, as well as the tendency of the embolic
agent to penetrate less than sodium amytal or lidocaine because of its higher viscosity and progressively polymerizing nature. Therefore, we think that
this method of provocative testing tends to overestimate the risk of embolization, resulting in the high
negative predictive value of this test.
In conclusion, neurophysiological provocative
testing using sodium amytal and lidocaine is a useful
adjunctive test for embolization of spinal cord AVMs
under general anesthesia. Both MEPs and SEPs
should be monitored regardless if the provocative test
is performed in the ASA or PSA. If the test is negative, the AVM can be comfortably embolized using
a liquid embolic agent. If it is positive, we believe
that aggressive embolization with a liquid agent carries high risk of causing spinal cord damage, and that
the best possible solution should be considered based
on the careful angiographic analysis of the malformation and indication for the treatment.
46.5.1.1.3. Illustrative cases. Case 1 (Fig. 5): A
24-year-old female patient initially presented with
progressive weakness and numbness of the right
lower extremity since the age of 17. At the age of
23 years, she experienced spinal subarachnoid hemorrhage. Spinal angiography demonstrated a spinal
cord AVM extending from C5 to C7 (Fig. 5A). She
underwent two previous endovascular embolization
procedures for this spinal cord AVM with partial
occlusion of the nidus. At the second embolization,
the ASA feeder was embolized from the dorsocervical artery with significant decrease in the opacification of the nidus with the preservation of the ASA
axis (Fig. 5B). At the time of the third embolization
procedure, spontaneous thrombosis and disconnection of the ASA axis were discovered (Fig. 5C).
The ASA was superselectively catheterized to the
origin of this remaining feeder and DSA examination
was performed (Fig. 5D). Based on the angiographic
analysis, it could not be determined if this vessel
provided collateral supply to the normal spinal cord
SPINE SURGERY
663
Fig. 5. Case 1. A: Posterioranterior (PA) view of the right dorsocervical artery before embolization demonstrating a large
arteriovenous malformation (AVM) involving the C5C7 levels supplied by the anterior spinal artery (arrow). B: PA view
of the right dorsocervical artery angiogram after the second embolization showing decreased opacification of the nidus of
the malformation with preservation of the anterior spinal axis (arrowheads) with one indirect remaining feeder to the malformation (arrows). C: PA view of the right dorsocervical angiogram 11 months after the second embolization at the time of
the third embolization showed spontaneous occlusion of the anterior spinal axis (arrowhead) and the remaining indirect
feeder to the malformation (arrow). Compare with B. D: PA view of the superselective angiogram from the anterior spinal
artery just before the origin of the feeder demonstrating complete occlusion of the anterior spinal axis just distal to the origin of the feeder (arrows). The arrowhead indicates the tip of the microcatheter within the anterior spinal axis. Because the
provocative test result was negative, the malformation was embolized from this catheter position with N-butyl cyanoacrylate (NBCA) without causing any worsening of the symptoms. E: PA view of the right dorsocervical artery angiogram after
the third embolization showing occlusion of the collateral feeder by embolization. Arrow indicates occlusion of the anterior
spinal artery (ASA). F: PA view of the right vertebral artery angiogram after the third embolization. The anterior spinal
artery is opacified from above with minimal supply to the remaining nidus (arrowhead). The remaining nidus is mainly supplied by the vertebral artery branch. There is slow flow in the radiculomedullary artery from the right dorsocervical artery
(arrow). Modified from Niimi et al. (2004) with permission. Copyright # American Society of Neuroradiology.
664
Y. NIIMI ET AL.
Fig. 6. Case 2. A: Posterioranterior (PA) view of the right dorsocervical artery angiogram demonstrating a pial feeder
(arrow) to the arteriovenous malformation (AVM) originating from the radiculomedullary artery just before the origin of
the anterior spinal axis. B, C, and D: PA (B, C) and lateral (D) views of superselective angiogram of the pial feeder to
the malformation in early (B) and late (C, D) phases. Arrow indicates the tip of the microcatheter. The malformation is
draining to the anterior spinal vein (arrow heads in C). This was not embolized because of positive provocative testing, indicating existence of supply to the normal spinal cord. Modified from Niimi et al. (2004) with permission. Copyright #
American Society of Neuroradiology.
at the level of the occluded segment of ASA.

Based on the negative result of the provocative testing, this was embolized with NBCA (Fig. 5E, F)
without causing aggravation of her neurological
conditions.
Case 2 (Fig. 6): A 20-year-old male patient with a
cervical spinal cord AVM extending from C5 to C6
levels. He initially presented with cervical subarachnoid hemorrhage at the age of 10 and underwent 3
embolization procedures to decrease the size of the
nidus. He recently developed a small hematomyelia
without neurological deterioration. Follow-up angiogram demonstrated a remaining cervical spinal cord
AVM supplied by the ASA and the lateral spinal
artery from the vertebral artery and the ASA from
the dorsocervical artery (Fig. 6A). Following embolization from the ASA through the vertebral artery, the
feeder from the radiculomedullary artery from the
dorsocervical artery was superselectively catheterized. Superselective angiogram demonstrated a pial
feeder on the surface of the spinal cord supplying
the malformation (Fig. 6B, C, and D). It was unclear
if this vessel supplied functional spinal cord based on
the angiographic analysis. Because lidocaine injection from the microcatheter caused disappearance of
both upper extremity MEPs, the decision was made
not to embolize this feeder.
46.5.1.2. Dural or extradural lesions

Dural or extradural lesions include spinal dural
arteriovenous fistulas (SDAVFs), epidural or paraspinal
arteriovenous malformations (AVMs) or fistulas
(AVFs), and spine AVMs or AVFs. Endovascular
embolization is indicated in all of these lesions, as
a curative, palliative, or preoperative treatment. We prefer to use liquid adhesives, such as NBCA, as the
embolic agent for these lesions because of its ability to
penetrate into small vessels and its permanent occlusive
effect. For the permanent cure of an AVF, the liquid
embolic material should penetrate into the proximal
portion of the draining vein through the fistula site
(Niimi et al., 1997). Insufficient penetration of the
embolic material frequently results in recanalization of
the lesion due to rich collateral vessels. Coils may be
used as an adjunct agent in NBCA embolization to protect normal territory or may be used as a primary agent
for high flow AVFs. Particles are not utilized because
their occlusive effect tends to be temporary resulting
in a high rate of recanalization (Nichols et al., 1991).
The pretherapeutic angiographic protocol for
these lesions is similar to that for tumors except that
the evaluation of the venous drainage of the lesion
and the normal spinal cord is more important. The
angiograms of the bilateral segmental arteries should
be obtained at the level of the feeders as well as at
SPINE SURGERY
least two levels above and below the level of the

malformation. If there is intradural venous drainage
to the perimedullary veins, it is essential to evaluate
the circulation time of the ASA. For this purpose,
an angiogram of the dominant radiculomedullary
artery (frequently Adamkiewiczs artery) should be
obtained by injecting a large amount of contrast
material with a long imaging time (e.g., 10 ml of
contrast injected at a rate of 1 ml/s, imaging time
for 4060 s). If the circulation time of the ASA is
prolonged with no opacification of the spinal cord
venous drainage, this indicates the existence of spinal
cord venous hypertension and explains the most
likely etiology of the patients neurological deficits.
Spinal cord venous hypertension is the underlying
cause of almost all SDAVFs, many perimedullary
AVFs, and many epidural AVFs with intradural perimedullary venous drainage (Niimi and Berenstein,
1999; Berenstein et al., 2004a,b).
The primary role of neurophysiological monitoring for this disease group is to detect a masked spinal
cord artery originating from the same pedicle as the
feeder to the malformation. If there is an ASA or a
PSA originating from the same pedicle as the feeder,
endovascular embolization is contraindicated unless
very distal advancement of the microcatheter within
the feeding vessel can be obtained beyond the origin
of the spinal cord artery. Provocative testing should
be performed before embolization if there is any suspicion of the existence of an unidentified spinal cord
artery from the feeding vessel to the malformation.
The necessity of provocative testing in dural or extradural vascular malformations, however, is exceptional
because, compared with tumor cases, identification of
a spinal cord artery is easier due to less distortion of
the spine and spinal cord and less overlapping abnormal vascularity or metallic devices. Careful analysis
of the vascular anatomy of the lesion and the normal
spinal cord is far more important than provocative testing. Another role of neurophysiological monitoring is
early detection of possible spinal cord ischemia. If significant changes in SEPs or MEPs are detected during
embolization, the procedure should be suspended until
full recovery of SEPs and MEPs or terminated to minimize the permanent damage to the spinal cord.
For certain diseases in this category, neurophysiological monitoring has another promising but not yet
proven role; that is, the prediction of functional recovery after embolization. Significant improvement of
SEPs or MEPs after embolization is frequently observed
for the diseases symptomatic due to spinal cord venous
665
hypertension, including SDAVFs and spinal epidural

fistulas with intradural venous drainage. Reduction of
spinal cord venous hypertension after embolization
is sometimes associated with improvement of SEPs
or MEPs and correlated with immediate improvement
of neurological symptoms (Kothbauer et al., 1998b;
Sala and Niimi, 2002). However, improvement of SEPs
or MEPs after embolization is not always associated
with clinical improvement. The patient needs intensive
rehabilitation after the treatment to maximize functional
recovery. Clinical improvement typically occurs first
in motor functions, followed by sensory functions.
Improvement of bladder, bowel, and sexual dysfunctions tends to be delayed and less satisfactory, if at
all (Berenstein et al., 2004b). It should be noted that
the improvement of MEPs or SEPs immediately after
embolization does not guarantee complete cure of the
disease or long-lasting remission of the symptoms.
Symptomatic recurrence of the disease is typically
associated with deterioration of the neurophysiological
findings. Therefore, correlation of neurophysiological
improvement (SEPs and MEPs) with angiographical
cure of the lesion (i.e., penetration of the embolic
material into the venous side) is important to predict
permanent clinical improvement. Further accumulation
of cases with detailed analysis and long-term follow
up is necessary to establish the exact role of neurophysiological monitoring as a predictor of functional
recovery after endovascular treatment of these diseases.
A case of endovascular treatment of an epidural
fistula with intradural venous drainage is shown in
Fig. 7. This patient had improvement of MEPs and
SEPs immediately after the embolization with angiographic demonstration of complete cure of the lesion
and improvement of spinal cord venous hypertension.
The patient experienced immediate clinical improvement after the embolization and was neurologically
intact at the 3-month follow-up.
46.5.2. Tumors
Vascular tumors are classified as benign and malignant. Malignant tumors can be further classified as
primary and metastatic. For the purpose of endovascular embolization, it is also useful to differentiate
between intramedullary and extramedullary tumors.
Intramedullary tumors are supplied by the spinal cord
arteries and their embolization carries higher risk
than extramedullary tumors which are not supplied
by the spinal cord arteries. Hemangioblastoma is
practically the only intramedullary tumor that is a
666
Y. NIIMI ET AL.
AH Rt OP
30 V
AH Rt CL
50 ms
AH Lt OP
100 V
AH Lt CL
50 ms
RPTN OP
38.4
RPTN CL
36.9
LPTN OP
38.4
4 V
LPTN CL
36.9
50 ms
SPINE SURGERY
candidate for embolization. This tumor is benign but

usually hypervascular and is embolized primarily as
a preoperative procedure to decrease intraoperative blood loss. Indication for embolization is determined based on the angiographic assessment of the
feeders (ASA or PSA), their sizes, and vascularity
of the tumor. Angiographic assessment and endovascular treatment are performed in a similar manner
to that utilized for intradural spinal cord vascular
malformations.
Intradural extramedullary tumors such as meningiomas and neurinomas are not usually very vascular
and embolization is not indicated. In contrast, vascular tumors located in the extradural or paraspinal
compartments are frequently amenable to embolization as a preoperative or palliative measure. These
tumors include benign and malignant lesions. Common benign tumors include hemangiomas, giant
cell tumors, and aneurysmal bone cysts. Malignant
tumors include primary sarcomas, plasmacytomas,
hemangiopericytomas, and metastatic carcinomas
such as those from the breast, thyroid, kidney, and
stomach.
We usually use particles such as polyvinyl alcohol
(PVA) particles as the primary embolic agent. Coils
are used to protect normal territory from inadvertent
embolization. Liquid adhesive, such as NBCA, is
not routinely used for tumors because of its higher
667
potential risk of penetration into the spinal cord

artery. It may, however, be used in highly vascular
tumors to obtain a better occlusive effect. For palliative embolization of malignant tumors, ethanol may
also be used as an embolic agent, resulting in a
long-lasting effect due to its cytotoxicity.
Pretherapeutic angiographic study is important
to evaluate the vascularity of the lesion including
feeding arteries, draining veins, existence of arteriovenous shunting and associated aneurysms as well
as the extension of the lesion. It is also important to
determine if there is ASA or PSA contribution at or
near the level of the lesion. For this purpose, angiographic evaluation should include the assessment of
bilateral segmental arteries not only at the level of
the lesion but also at least two levels above and
below the lesion. In the case of distortion of the spine
or the spinal cord from the previous treatment, the
disease itself, or the existence of overlapping metallic
stabilization instruments or hypervascularity of the
tumor, oblique, and lateral views may be necessary
to identify spinal cord arteries (Djindjian and
Merland, 1981; Berenstein et al., 2004c).
The main purpose of neurophysiological monitoring for spinal or paraspinal tumor embolization is to
detect masked and unrecognized spinal cord arteries.
If existence of a spinal cord artery is suspected but
uncertain on the superselective angiography from the
Fig. 7. A 23-year-old female presented with progressive bilateral lower extremity weakness, numbness, and bladder and
bowel dysfunction. Magnetic resonance imaging (MRI) showed T2 high signal abnormality as an evidence of spinal cord
congestion due to venous hypertension (not shown). Spinal angiography demonstrated an epidural arteriovenous fistula
(AVF) draining to the intradural perimedullary veins causing venous hypertension of the spinal cord. A: Posterioranterior
(PA) view of the left T9 intercostal artery angiogram demonstrating an epidural AVF draining to the dilated perimedullary
veins (arrows). B and C: PA view of the early (B) and late (C) phases of the left L1 lumbar artery angiogram showing the
anterior spinal artery with delayed circulation time and the stagnation of contrast material (arrow heads in C). There is no
visualization of the draining vein of the spinal cord in the late venous phase (C). D: PA view of the left T9 intercostal artery
superselective angiogram showing the fistula to the epidural vein (large arrow). Arrow head indicates dural penetration of
the draining vein. Small long arrow indicates the tip of the microcatheter. The fistula was embolized from this catheter position using N-butyl cyanoacrylate (NBCA). E: Plain X-ray showing the NBCA cast penetrating to the proximal portion of the
draining vein. Compare with D. F and G: PA view of the early (F) and late (G) phases of the left L1 lumbar artery angiogram showing the anterior spinal artery with improvement of the circulation time. The late phase (G) shows visualization of
the dilated perimedullary veins draining the normal spinal cord. These veins used to drain the fistula before embolization.
Compare arrows in A and G. This demonstrates angiographic improvement of venous hypertension after embolization.
H: Somatosensory evoked potentials (SEPs) from the bilateral posterior tibialis nerve (PTN) before (OP) and after (CL)
embolization procedure, demonstrating significant improvement of the latency of the response. I: MEPs from the bilateral
abductor hallucis muscles (AH) before (OP) and after (CL) the embolization procedure, demonstrating significant improvement of the latency of the response. The patient neurologically improved immediately after the embolization. She was neurologically normal 3 months after the embolization. Follow-up MRI showed improvement of T2 high-signal abnormality
(not shown). This case is an example of correlation among clinical, angiographic, and neurophysiologic improvement of
the spinal cord venous hypertension. Modified from Sala and Niimi (2002) with permission from Elsevier.
668
Y. NIIMI ET AL.
Fig. 8. A 48-year-old male with a cervical spine hemangiopericytoma. Preoperative embolization was requested to decrease
blood loss during surgery. A: Posterioranterior (PA) view of the right dorsocervical artery angiogram demonstrating a
hypervascular tumor stain involving the right C6 hemivertebra. No definite spinal cord artery was identified on this study.
B: Postembolization control angiogram of the right dorsocervical artery demonstrating complete devascularization of the
tumor stain with preservation of the anterior spinal artery (arrows). Embolization was performed using polyvinyl alcohol
(PVA) particles with assistance of one provocative testing to confirm the absence of a spinal cord artery distal to the tip
of the microcatheter. C: Left dorsocervical artery angiogram before embolization demonstrating a hypervascular tumor stain
in the left C6 hemivertebra. D: Superselective angiogram from a branch of the left dorsocervical artery. The small arrow
indicates the tip of the microcatheter and the large arrow indicates the tip of the guiding catheter. No spinal cord artery
was identified on this study. E: Superselective angiogram of a branch of the left dorsocervical artery during embolization
using PVA particles. The tip of the microcatheter (medium arrow) is further advanced distally. Compare with D. There
is anastomotic opacification of the anterior spinal artery (small arrows). Compare with Fig. 8B. Provocative test could have
been performed to confirm the existence of the anterior spinal artery if there were any question. The large arrow indicates
the tip of the guiding catheter. F: Postembolization control angiogram of the left dorsocervical artery demonstrating complete devascularization of the tumor. The patient was operated 2 days later without significant blood loss and transfusion.
Modified from Sala and Niimi (2002) with permission from Elsevier.
SPINE SURGERY
feeder to the lesion before or during embolization, provocative testing may be performed by observing
changes in SEPs or MEPs. If any changes are noted
in either SEPs or MEPs, aggressive embolization from
that catheter position should be avoided. Also, if a significant change in SEPs or MEPs occurs during an
embolization procedure, spinal cord ischemia should
be suspected and the procedure should be terminated
to minimize the risk of permanent damage and maximize the possibility of recovery of the spinal cord.
Improvement of SEPs or MEPs after tumor embolization is sometimes observed in association with clinical
improvement. This phenomenon most often occurs in
a tumor with epidural extension and spinal cord compression and is probably due to decreased mass effect
secondary to tumor devascularization and shrinkage.
This improvement is an indicator of effective embolization either as a preoperative or as a palliative
treatment.
An example of preoperative embolization for a
cervical spine tumor is presented in Fig. 8.
46.6. Conclusions
Neurophysiological monitoring is feasible in the
great majority of patients undergoing endovascular
procedures for spine or spinal cord lesions. MEPs
and SEPs retain their own specificity to assess the
functional integrity of motor and sensory pathways,
respectively. Neurophysiological monitoring and
pharmacological provocative testing during endovascular procedures also offer a unique opportunity to
investigate the spinal cord hemodynamics and to
integrate functional and vascular anatomy.
References
Berenstein, A, Young, Y, Ransohoff, J, Benjamin, V and
Merkin, H (1984) Somatosensory evoked potentials during spinal angiography and therapeutic transvascular
embolization. J. Neurosurg., 60: 777785.
Berenstein, A, Choi, I, Neophytides, A and Benjamin, V
(1990) Endovascular treatment of spinal cord arteriovenous malformations (SCAVMs). AJNR Am. J. Neuroradiol., 10: 898.
Berenstein, A, Lasjaunias, P and Ter Brugge, K (2004a)
Spinal arteriovenous malformations. Surgical Neuroangiography. Springer-Verlag, Berlin Heidelberg, Vol. 2.2,
pp. 737847.
Berenstein, A, Lasjaunias, P and Ter Brugge, K (2004b) Spinal
dural arteriovenous fistulae. Surgical Neuroangiography.
Springer-Verlag, Berlin Heidelberg, Vol. 2.2, pp. 849872.
669
Berenstein, A, Lasjaunias, P and Ter Brugge, K (2004c)
Tumors of the spinal column and spinal cord. Surgical
Neuroangiography. Springer-Verlag, Berlin Heidelberg,
Vol. 2.2, pp. 873912.
Biondi, A, Merland, J, Reizine, D, Aymard, A, Hodes, J,
Lecoz, P and Rey, A (1990) Embolization with particles
in thoracic intramedullary arteriovenous malformations:
long-term angiographic and clinical results. Radiology,
177: 651658.
Chiras, J, Morvan, G and Merland, J (1979) The angiographic appearance of the normal intercostal and lumbar
arteries: analysis of the anatomic correlation of the lateral branches. J. Neuroradiol., 6: 169196.
Cohen, AR, Young, W and Ransohoff, J (1981) Intraspinal
localization of the somatosensory evoked potential.
Deletis, V and Engler, G (1997) Somatosensory evoked
potentials for spinal cord monitoring. In: K Bridwell
and Y. Olsson (Eds.), The Textbook of Spinal Surgery.
Lippincott-Raven, Philadelphia, pp. 8592.
Deletis, V and Kothbauer, K (1998) Intraoperative neurophysiology of the corticospinal tract. In: E Stalberg, H
Sharma and Y Olsson (Eds.), Spinal Cord Monitoring.
Springer, Wien, New York, pp. 421444.
Deletis, V and Vodusek, D (1997) Intraoperative recording
of the bulbocavernosus reflex. Neurosurgery, 40: 8893.
Djindjian, R and Merland, J (1981) Angiography of Spinal
Column and Spinal Cord Tumors. Georg Thieme
Verlag, Stuttgart.
Djindjian, R, Merland, J, Djindjian, M, et al. (1978) Embolization in the treatment of medullary arteriovenous malformations in 38 cases. Neuroradiology, 16: 428429.
Doppman, J, Girton, M and Oldfield, E (1986) Spinal
Wada test. Radiology, 161: 319321.
somatosensory evoked potentials. Case report. J. Neurosurg., 63: 296300.
Herren, R and Alexander, L (1939) Sulcal and intrinsic
blood vessels in human spinal cord. AMA Arch. Neurol.
cases of quadriparesis following anterior cervical discectomy, with normal perioperative somatosensory
evoked potentials. J. Neurol. Neurosurg. Psychiatry,
74(2): 273276.
Katayama, Y, Tsubokawa, T, Hirayama, T, Himi, K,
Koyama, S and Yamamoto, T (1991) Embolization of
intramedullary spinal arteriovenous malformation fed
by the anterior spinal artery with monitoring of the corticospinal motor evoked potential: Case report. Neurol.
Med. Chir., 31: 401405.
Kothbauer, K, Deletis, V and Epstein, F (1997) Intraoperative spinal cord monitoring for intramedullary surgery:
an essential adjunct. Pediatr. Neurosurg., 26: 247254.
670
Kothbauer, K, Deletis, V and Epstein, F (1998a) Motorevoked potential monitoring for intramedullary spinal
cord tumor surgery: correlation of clinical and neurophysiologic data in a series of 100 consecutive procedures.
Neurosurg. Focus, (electronic journal) 1998; 4(5): 19
http//www.aans.org/journals/online_j/may98/4-5-1.
Kothbauer, K, Pryor, J, Berenstein, A, Setton, A and Deletis,
V (1998b) Motor evoked potentials predicting early
recovery from paraparesis after embolization of a spinal
dural arteriovenous fistula. Interv. Neuroradiol., 4: 8184.
Lasjaunias, P, Berenstein, A and Ter Brugge, K (2001) Spinal and spinal cord arteries and veins. Surgical Neuroangiography. Springer-Verlag, Berlin Heidelberg,
Vol. 1, pp. 73164.
Lazorthes, G, Gouaze, A and Djindjian, R (1971) Arterial
vascularization of the spinal cord. J. Neurosurg., (35):
253260.
Lesser, R, Raudzens, P, Luders, H, Nuwer, M, Goldie, W,
Morris, HI, Dinner, D, Klem, G, Hahn, J and Shetter,
A (1986) Postoperative neurological deficits may occur
despite unchanged intraoperative somatosensory evoked
potentials. Ann. Neurol., 19: 2225.
Nichols, D, Rufenacht, D, Jack, CJ and Forbes, G (1991)
Embolization of spinal dural arteriovenous fistula with
polyvinyl alcohol particles. AJNR Am. J. Neuroradiol.,
13: 933940.
Niimi, Y and Berenstein, A (1999) Endovascular treatment
of spinal vascular malformations. Neurosurg. Clin. N.
Am., 10: 4771.
Niimi, Y, Berenstein, A, Setton, A and Neophytides, A
(1997) Embolization of spinal dural arteriovenous fistulae: results and follow-up. Neurosurgery, 40: 675683.
Niimi, Y, Sala, F, Deletis, V, Setton, A, Bueno De
Camargo, A and Berenstein, A (2004) Neurophysiologic monitoring and pharmacological testing for embolization of spinal cord AVMs. AJNR Am. J. Neuroradiol.,
25: 11311138.
Y. NIIMI ET AL.
Pelosi, L, Lamb, J, Grevitt, M, Mehdian, SM and Webb, JK
(2002) Combined monitoring of motor and somatosensory evoked potentials in orthopaedic spinal surgery.
Prestigiacomo, C, Niimi, Y, Setton, A and Berenstein, A
(2003) Three-dimensional rotational spinal angiography
in the evaluation and treatment of vascular malformations. AJNR Am. J. Neuroradiol., 24: 14291435.
Sala, F and Niimi, Y (2002) Neurophysiologic monitoring
during endovascular procedures on the spine and the
spinal cord. In: V Deletis and J Shils (Eds.), Neurophysiology in Neurosurgery. Academic Press, San Diego,
pp. 119151.
Sala, F, Niimi, Y, Krzan, M, Berenstein, A and Deletis, V
(1999) Embolization of a spinal arteriovenous malformations: correlation between motor evoked potentials
and angiographic findings: technical case report. Neurosurgery, 45: 932938.
Sala, F, Niimi, Y, Berenstein, A and Deletis, V (2000) Role
of multimodality intraoperative neurophysiologic monitoring during embolisation of a spinal cord arteriovenous malformation. A paradigmatic case. Interv.
Neuroradiol., 6: 223234.
Neuroprotective role of neurophysiologic monitoring
Ann. N.Y. Acad. Sci., 939: 126136.
Takaki, O and Okumura, F (1985) Application and limitation of somatosensory evoked potential monitoring during thoracic aortic aneurysm surgery: a case report.
Tanaka, K and Yamasaki, M (1966) Blocking of cortical
inhibitory synapses by intravenous lidocaine. Nature,
209: 207208.
Thron, AK (1988) Anatomy of the Spinal Cords Blood
Supply. Vascular Anatomy of the Spinal Cord,
Springer-Verlag, Wien, pp. 812.
Zornow, M, Grafe, M, Tybor, C and Swenson, M (1990)
Preservation of evoked potentials in a case of anterior
spinal artery syndrome. Electroencephalogr. Clin. Neurophysiol., 77: 137139.

M.R. Nuwer (Ed.)
671
CHAPTER 47
Monitoring the spinal cord during surgery

for cervical myelopathy
Steve Jones*
Department of Clinical Neurophysiology, The National Hospital for Neurology and Neurosurgery, London WCIN 3BG, UK
My objective in this chapter is to evaluate the evidence

for the effectiveness of spinal cord monitoring during
surgery performed for cervical spinal cord decompression due to skeletal deformity or degeneration rather
than tumor or arteriovascular abnormalities. This
includes the effects of spondylosis, stenosis, disk protrusion, and subluxation. Many of the conclusions
will also apply to other extrinsic causes of myelopathy,
and some authors in their studies (including our own,
May et al., 1996) have not treated these conditions
differentially.
47.1. Does spinal cord monitoring work?
There are two levels at which this question may be
asked, but only one at which it may satisfactorily be
answered. Clearly, the sole objective of spinal cord
monitoring is to prevent the avoidable occurrence of
iatrogenic neurological deficits. In order to determine
whether this is achieved in practice, it is absolutely
necessary to conduct carefully controlled, prospective
trials in which equal numbers of cases receiving identical surgical treatment of similar conditions are either
monitored or not monitored (or have the findings communicated to or withheld from the surgeon). Virtually
no such trials have been performed, perhaps not for
ethical reasons though these are easy enough to perceive. Should the conditions of such a trial be adhered
to, and the benefits of monitoring scientifically proven, both surgical and anesthetic methods would,
strictly speaking, have to undergo no further variation
which might increase or decrease the risks.
*
Correspondence to: Steve Jones, Ph.D., Department of

Clinical Neurophysiology, The National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N
3BG, UK.
Tel.: (44)-20-7837-3611, ext. 4109; fax: (44)-20-77137743.
E-mail: sjjones@ion.ucl.ac.uk (S. Jones).
It appears that only one study has attempted a controlled assessment of the benefits of spinal cord monitoring using noninvasive somatosensory evoked
potentials (SEPs) during cervical surgery. Epstein
et al. (1993) compared the morbidity rates associated
with 218 patients who were not monitored (operated
on between 1985 and 1989) with those in 100 monitored procedures performed between 1989 and 1991.
The incidence of quadriplegia was 3.7% in the former group and 0% in the latter, an encouraging trend
attributed in part to early somatosensory evoked
potential detection of vascular or mechanical
compromise . . . and to the immediate alteration of
anaesthetic or surgical technique. As the authors
acknowledge, however, there are other factors that
may have made some contribution. Even if the surgical and anesthetic methods were broadly similar, the
increasing experience of the surgeon may (even
should) have led to more risky procedures being
identified in the control group and avoided as far
as possible in the subsequent cases. In the control
group seven out of eight instances of postoperative
quadriplegia occurred after posterior cervical laminectomy, which was less frequently performed in
the monitored group. Another condition that changed
between the two groups was the proportion of
patients operated on by the senior author.
If we cannot scientifically determine whether spinal
cord monitoring prevents neurological deficits, we can
at least try to discover whether it successfully detects
the kind of lesions that are likely to occur at an early
and (it is hoped) reversible stage. This obviously
requires the results of monitoring to be classified
according to the neurological outcome, leading to the
four familiar categories of true negative (TN, no
SEP change, no additional deficit), true positive
(TP, SEP deterioration plus new neurological deficit),
false negative (FN, no SEP change but a new deficit), and false positive (FP, SEP deterioration but
672
no new deficit). Even here, however, difficulties of

interpretation are encountered because some authors
find it necessary to redefine these terms. Should a case
in which SEP deterioration is apparently due to an
identifiable problem which is then reversed with no
neurological consequences be regarded as a false positive? In the study already cited, Epstein et al. (1993)
chose not to label these as false-positive changes
but rather preferred to assume they reflected the true
occurrence of a neurologic injury that was reversed
successfully. In this series a substantial percentage,
65%, of cases experienced significant SEP deterioration at some stage. If even half of these represented
real, albeit temporary, neurological injury, it is hard
to reconcile the finding with the 3.7% incidence of
actual clinical deficits in the unmonitored cases.
Properly defined, we can never know whether a
false positive was in fact a false alarm. Again
properly defined, a zero incidence of FP means that,
however successful monitoring may have been at
detecting defects, it has not resulted in any being
completely averted (though some, of course, may
have been limited in their severity). One datum that
gives a useful measure of comparability between
studies is the ratio of false positives to true positives, FP/TP. In surveys of spinal cord monitoring
during surgery for kyphoscoliosis FP/TP has usually
been around 24 (e.g., Forbes et al., 1991; Nuwer
et al., 1995), and similar figures have been obtained
in more than one study of cervical surgery (e.g.,
May et al., 1996; Manninen, 1998). The ratio
depends not only on the inherent sensitivity of the
technique but also on the stringency of the intervention criteria chosen by the authors (statistically
speaking, the position on the receiver operating characteristic curve). Most studies appear to have chosen
criteria based on the smallest degree of change that
detects all or nearly all instances of neurological deficit, resulting in zero FN and similar FP/TP. Having
thereby optimized the discriminative power of the
technique, we can at least hold as an article of faith
that some FP may actually be those patients who
benefit the most from the performance of monitoring.
47.2. SEP changes in cervical spondylotic
radiculopathy and myelopathy
Cervical spondylosis is an extremely common condition in middle age, often resulting in radiculopathy
and/or myelopathy. Although other factors may contribute, myelopathy can occur because of disk
S. JONES
protrusion or compression by osteophytes as well as

by constriction of the spinal canal and buckling of
the ligamentum flavum. It is not clear to what extent
ischemia may also contribute.
In this chapter, I do not intend to review the literature on SEP changes in cervical spondylosis. In our
own experience (Yu and Jones, 1985), responses to
lower limb stimulation were more often affected than
those to upper limb stimulation. The upper limb
responses to ulnar nerve stimulation were more often
affected than median nerve SEPs do. However, the
diagnostic implications of this should be considered
quite separately from those for intraoperative monitoring. Whereas median nerve SEPs are surprisingly
immune to the effects of chronic cervical myelopathy, even when this results in sensory and motor deficits extending above the C6 segmental level, they
are certainly sensitive to the acute insults that occur
during surgery. The lower limb SEPs, while more
frequently abnormal at the commencement of surgery, are more often too severely degraded to be of
value in intraoperative monitoring. Epstein et al.
(1993) monitored both the upper and the lower limb
SEPs, which they believed to be essential but which
could have been a factor in their very high incidence
of significant SEP changes.
47.3. The anterior spinal artery and
MEP monitoring
Although posterior laminectomy and open door laminoplasty are performed in some circumstances, most
operations for cervical decompression are performed
from the anterior side of the neck (diskectomy and corpectomy), since it is from that side that disks and osteophytes impinge on the spinal cord. The position of the
anterior spinal artery implies that it is likely to be
involved in such compression. Of course, the cortical
SEPs that depend on conduction through the dorsal columns say nothing about more anterior cord tracts or the
spinal gray matter whose blood supply is obtained from
the anterior spinal artery. It seems somewhat fortunate,
therefore, that SEPs still successfully detect the majority of problems that arise. We have reported two
instances of purely motor deficits following anterior
cervical decompression, both of them unaccompanied
by significant SEP changes (Jones et al., 2003). The
one SEP false negative reported by Taunt et al.
(2005) was also a purely motor problem.
Fortunately, such occurrences are very rare and
do not suggest a pressing need for motor evoked
SPINE SURGERY
potential (MEP) monitoring in straightforward anterior cervical decompression. The first two reports of
MEP monitoring specifically during cervical surgery
both used transcranial stimulation and epidural recording of activity in the motor tracts of the cord at the
upper thoracic level (Kitagawa et al., 1989; Gokaslan
et al., 1997). With a combined total of just 36 patients,
it is only appropriate here to note that the technique is
feasible. Haghighi (2002) described the use of repetitive transcranial electrical stimulation to elicit muscle
responses from the upper and the lower limbs. The
46 patients included 12 receiving surgery at cervical
level, of whom 4 had myelopathy. Again, however,
the findings are too preliminary to indicate whether
MEP monitoring is likely to be found routinely useful.
The study of Bose et al. (2004), described in the literature review section below , is the first from which some
such indication may be drawn.
47.4. Methodological considerations
Whereas in surgery for kyphoscoliosis invasive techniques for recording evoked spinal cord potentials
and conducted spinal SEPs have been preferred in
some countries, mainly on account of their greater stability to anesthetic and other systemic factors (Burke
et al., 1999), in surgery for cervical myelopathy noninvasive SEP techniques are practically universal.
The main reason for this is likely to be the difficulty
or undesirability of inserting an electrode into a constricted spinal canal at high cervical level. The principles that govern intraoperative monitoring of cortical
and subcortical SEPs (Nuwer, 1998), therefore, apply
without any particular modification. It should be
stressed, however, that the level(s) of myelopathy
and surgical intervention will determine which nerves
may be used for stimulation. When the problem is at
C5 level or above, the median nerve will probably be
preferred. SEPs to median nerve stimulation depend
on input to the spinal cord via the C6 and C7 dorsal
roots (the contribution of muscle afferents entering
via C8 being relatively insignificant). However, one
should bear in mind that the C6 and C7 segments
of the spinal cord are displaced toward the head
by approximately one level with respect to the
corresponding vertebral bodies. For this reason,
median SEPs can only be relied on when surgery is
exclusively above the C6 vertebra (i.e., addressing
disks at C5/6 level and above). Ulnar nerve SEPs
may be used for surgery one segment lower, but for
operations at C7 level and below there is no viable
673
alternative to the lower limb stimulation. Ideally, it

may be argued, both the upper and the lower limb SEPs
should be recorded in every case (e.g., Epstein et al.,
1993), but in our experience this has not been found
necessary.
Whether the addition of subcortical recording channels contributes to the sensitivity and reliability of the
method is debatable. However, if there are sufficient
amplifiers available, recordings made from locations
over the brachial plexus will ensure the adequacy of
a peripheral nerve volley and, if referred to the scalp,
also reveal subcortical far-field potentials (P11 and
P13/14, inverted so as to appear with the same polarity
as the peripheral N9) generated between the dorsal roots
and the brainstem. In the study by Sebastian et al.
(1997) all the SEP changes occurring during anterior
operations for cervical myeloradiculopathy involved
the subcortical potentials, but without any instances of
postoperative neurological deterioration it is hard to
judge the significance of this atypical study.
47.4.1. Intervention criteria
The often-quoted criteria of 50% amplitude reduction
and 10% latency increase, beyond which cortical
SEPs are deemed to have deteriorated significantly,
are purely arbitrary and not formally evidence-based.
Indeed, in our own experience, it was only in those
cases where the cortical responses effectively became
flat that the risk of a neurological deficit was greater
than in the overall group (May et al., 1996). This is
not to say that lesser changes should not be heeded,
but the knowledge that there is a subcritical zone
between the threshold for significant SEP change
and that for neurological problems is obviously an
advantage for the surgeon, who may decide simply
to proceed more cautiously when subcritical SEP
deterioration occurs.
47.5. Review of the literature
Some of the findings of Epstein et al. (1993) have
already been critically summarized. Nevertheless,
their detailed description of the SEP changes that
occurred is extremely valuable. Among the 65/100
cases in whom SEPs deteriorated were 46 in which
the changes lasted only 24 min, and for practical
purposes were therefore (in my opinion) probably
negligible. Almost all the more persistent changes
occurred in patients with preexisting myelopathy,
and 17/19 recovered significantly before closing.
674
SEP improvements were also seen more frequently

in the lower limb responses and more often in terms
of latency than of amplitude. Like the instances of
deterioration, however, there was no association with
enhanced neurological outcomes. Without any neurological corroboration in either a positive or a negative
sense, it is hard to escape the conclusion that these
authors were overreacting to SEP changes that were
actually mostly of a random nature or due to nonphysiological factors.
Bouchard et al. (1996) reported 32 patients with
moderate to severe spondylotic myelopathy who underwent multilevel anterior decompression and fusion.
They did not report any in whom the upper and/or the
lower limb SEPs deteriorated, but 11 demonstrated
intraoperative SEP improvement (mostly a decrease in
latency) after decompression. All these patients had
rapid recovery of motor strength, bladder control, and
ambulatory capacity within days of surgery. Of the
remainder, whose SEPs remained stable, 5 had rapid
resolution of their symptoms, 15 improved over the
course of 68 weeks, and 1 failed to improve. However,
the motor improvement of this group at 8 weeks was
equal to that of the group those in whom SEPs improved
intraoperatively.
Out of the group of 31 patients reported by Dennis
et al. (1996), all receiving cervical decompression for
various conditions, 4 had a poor postoperative neurological outcome. In two of the latter, the SEPs to
median nerve stimulation disappeared permanently.
In a further two the latency of the cortical N20 was
increased by more than 20% as measured from the
N9. Latency increases of a lesser degree were seen in
6 of the 27 patients whose outcome was improved or
unchanged. In one of the six, the change occurred during positioning for cervical traction, and was subsequently reversed.
The 191 cervical operations in the study of May
et al. (1996) were for diverse pathology, but the 60
patients with cervical spondylosis formed the largest
single group. Surgery was performed at a single level
in 31% of procedures, at two levels in 22%, three
levels in 20%, four levels in 11%, and five or more
levels in 16%. Ninety-five procedures were performed
from the anterior side, 50 from the posterior side,
32 from both anterior and posterior sides on the
same occasion, and 14 from the extreme lateral side.
Effective monitoring was impossible in nine cases
owing to the absence of cortical SEPs. These patients
all had severe myelopathy and three of them deteriorated postoperatively. Following 182 successfully
S. JONES
monitored operations, 10 patients developed exacerbated neurological signs. A total of 33 showed SEP
changes (more than 50% amplitude reduction of
the median nerve cortical response). Eight of the
10 patients whose neurological condition deteriorated were among the 16 whose SEPs disappeared
completely and permanently, while only one had a partial SEP loss and one showed no change. The last case
experienced exacerbated weakness in the distribution
of the ulnar nerve, which was understandably missed
by median nerve SEP monitoring. In the authors
view, some of the false positives (including four
which occurred during head manipulation) were real
neurophysiological changes whose neurological consequences were minimized by an altered surgical
response, but of course this was not proven. The overall incidence of SEP deterioration was very much
higher in this series (17.3%) as compared with that
reported by Nuwer et al. (1995) in a multicenter survey
of operations for kyphoscoliosis (1.9%), reflecting the
increased risks associated with surgery at this level and
performed in patients, of whom 42% already had clinical signs of cervical myelopathy.
From this study it was also possible to identify clinical and surgical risk factors that were significantly
associated with SEP deterioration. The most important
of these was the severity of the preexisting myelopathy. The relative risk was four times higher in nonambulatory as compared with ambulatory patients, and
two times higher in ambulatory patients as compared
with those without neurological signs of myelopathy.
Among those without clear signs of myelopathy, risk
factors for SEP deterioration were the longitudinal
extent of surgery, surgery addressing the very high cervical region, and the use of instrumentation.
Describing an overall group of 309 patients, Manninen (1998) found that the incidence of SEP changes
was much lower in the 88 receiving cervical surgery
(1.2%) as compared with those having surgery at the
thoracic (18%) or lumbar (5.4%) level. Overall, intraoperative SEP deterioration occurred in 6%, while
2.7% of patients developed a new neurological deficit.
Although the individual figures are much lower than
those obtained by May et al. (1996) in relatively complex cervical cases and higher than those reported by
Nuwer et al. (1995) in surgery for kyphoscoliosis, the
ratio of false positive to true positive findings is
comparable in all three studies.
Two publications by Wiedemayer et al. (2002,
2004), presumably reporting mostly the same patient
group, did not clearly distinguish the SEP monitoring
SPINE SURGERY
findings in spinal as distinct from supra- and infratentorial surgery. In the earlier of these the authors
identified a category of true positive findings with
intervention, including not only patients who had a
postoperative deficit corresponding to the target of
monitoring but also those who had no neurological
deficit after corresponding intraoperative findings
or problems were identified and the surgeon reacted
to this event (Wiedemayer et al., 2002). Whether
the SEP decrement was real in neurophysiological
terms, whether the surgeons reaction was in fact
instrumental in preventing a deficit, and whether similar recovery might not have occurred spontaneously,
of course all remain unproven. Among 84 of 423
cases in whom the SEPs were believed to indicate a
real problem, the surgeon was reportedly able to
respond with corrective action in 42. Obviously there
will always be circumstances in which SEP changes
occur which are due to irreversible damage or lack
any obvious surgical or anesthetic cause which can
be reversed, so it is useful to know that there was at
least believed to be some possibility of corrective
intervention in half of the cases in whom SEP monitoring suggested a problem.
The later study of 658 cases (Wiedemayer et al.,
2004) was directed specifically at 27 in whom monitoring was classified as false negative. The overall
patient group included 158 with extramedullary and
51 with intramedullary spinal lesions, and the 6 cases
with FN SEP findings all had intra- or extramedullary tumors rather than myelopathy due to spondylosis or disk disease.
Kombos et al. (2003) summarized the SEP monitoring findings in 100 patients treated by an anterior
cervical approach. Although there were apparently
no cases of postoperative neurological deterioration,
a large percentage showed a SEP amplitude decrease
exceeding 40%, sometimes accompanied by a
latency increase of more than 10%. One significant
cause of reversible SEP changes was distraction of
the intervertebral space, seen in 14 cases. No deterioration of SEPs was seen during decompression of the
cord. Interestingly, by far the largest number of
changes (35) occurred during positioning of the
patient, before any surgical incision, and it was the
patients with chronic cervical myelopathy rather than
radicular deficits or neurological symptoms of acute
onset who were most at risk of this. Since most other
studies have not specifically addressed this stage of
the operation (one similar example is described by
Dennis et al., 1996), it may be the case that many
675
patients are inadvertently being placed at risk from

an early stage of surgery.
Hyperextension of the neck before and during
anterior decompression may result in stretching and
occlusion of small blood vessels, particularly those
supplied by the anterior spinal artery. This was suggested as a possible cause of a primarily motor deficit seen in two cases reported by Jones et al. (2003).
Intraoperative SEPs (one to median, one to ulnar
nerve stimulation) were unchanged in both cases,
and both patients awoke with quadriparesis but with
relatively spared proprioceptive sensation.
The largest study to date of MEP monitoring in cervical surgery is that of Bose et al. (2004). Using both
SEPs and repetitive transcranial stimulation to elicit
the upper and the lower limb muscle responses in 119
operations, they reported six neurophysiologic alerts
that prompted surgical and/or anesthetic intervention.
Three cases had postoperative motor deficits, but only
one of these was detected intraoperatively. One patient
who had exacerbated weakness of the affected deltoid
and biceps muscles had no measurable baseline
responses. Another with preexisting upper extremity
weakness developed quadriparesis postoperatively,
but MEP monitoring was unfortunately compromised
by excessive neuromuscular blockade and SEPs to
ulnar nerve stimulation showed no remarkable
changes. Interestingly, however, in view of the contrary findings of our own report of two cases (Jones
et al., 2003), in three of the alerts it was SEP deterioration that drew attention to hyperextension of the neck.
There would still appear to be considerable doubt as to
whether MEP monitoring will eventually be found to
be of any practical value in monitoring surgery for cervical myelopathy.
The most recent study by Taunt et al. (2005) is
especially valuable in redressing the natural tendency
for the literature to emphasize positive results at
the expense of null findings. They reviewed 163
patients who underwent anterior cervical diskectomy
and fusion (ACDF). Only one patient developed
a postoperative neurological deficit, in spite of no
SEP changes during surgery. The nature of the
deficit a deltoid palsy which resolved after
2 weeks was clearly indicative of a localized ventral root or horn lesion, which one would not expect
SEP monitoring to detect. In spite of three false
positives, which may possibly have represented real
neurophysiological changes although the surgeons
did not feel impelled to alter or reverse any
part of their procedure, the authors concluded that
676
intraoperative SSEP monitoring is not helpful to the

surgeon when performing routine ACDF (Taunt
et al., 2005).
47.6. Conclusions
The following conclusions are based on what I hope
has been a reasonably impartial assessment of the published evidence, though scientifically unsupported
impressions cannot be entirely suppressed.
Patients undergoing decompressive surgery for
longstanding cervical myelopathy are at a greater

risk of iatrogenic neurological impairment than
those receiving similar treatment for radicular
lesions or acute spinal problems.
SEP monitoring using noninvasive techniques is
capable of detecting the majority of problems that
arise during surgery due to direct cord compression
or ischemia.
Although the evidence that SEP monitoring actually helps prevent the exacerbation of cervical myelopathy is inconclusive, the opinion of most
surgeons/authors is that the information provided
is frequently helpful.
Median nerve SEPs provide an adequate safeguard
for surgery at the level of C5/6 and above, while ulnar
SEPs can be applied one level lower. SEPs to lower
limb stimulation are essential for surgery below C6/
7, but are more frequently degraded or unrecordable.
The adoption of quite stringent intervention criteria
(e.g., the complete and persistent loss of cortical
responses, or latency increases exceeding 20%) is
necessary in order to avoid a high rate of false
positives, and of apparent neurophysiological
improvements which are not reflected in the neurological outcome.
In addition to the decompressive stage of surgery,
SEPs can help identify problems occurring during
positioning of the patient on the operating table and
subsequent neck manipulation.
Riders to the above are that SEP monitoring is only
rarely informative in routine cases of single level
anterior cervical decompression, and that hyperextension of the neck may occasionally cause ischemia in the territory of the anterior spinal artery
which has no significant effect on the SEPs.
Electrophysiological monitoring of the cervical
motor pathways by transcranial electrical stimulation
is feasible, but there is no real evidence as to its
effectiveness.
S. JONES
References
Bose, B, Sestokas, AK and Schwartz, DM (2004) Neurophysiological monitoring of spinal cord function during
instrumented anterior cervical fusion. Spine J., 4:
202207.
Bouchard, JA, Bohlman, HH and Biro, C (1996) Intraoperative improvements of somatosensory evoked potentials: correlation to clinical outcome in surgery for
cervical spondylitic myelopathy. Spine, 21: 589594.
Burke, D, Nuwer, MR, Daube, J, Fischer, C, Schramm, J,
Yingling, CD and Jones, SJ (1999) Intraoperative
monitoring. The International Federation of Clinical
Neurophysiology. Electroencephalogr. Clin. Neurophysiol. Suppl. 52. Elsevier, Amsterdam, pp. 133148.
Dennis, GC, Dehkordi, O, Millis, RM, Cole, AN, Brown, DS
and Paul, OA (1996) Monitoring of median nerve
somatosensory evoked potentials during cervical spinal
cord decompression. J. Clin. Neurophysiol., 13: 5159.
Epstein, NE, Danto, J and Nardi, D (1993) Evaluation of
intraoperative somatosensory-evoked potential monitoring during 100 cervical operations. Spine, 18: 737747.
Forbes, HJ, Allen, PW, Waller, CS, Jones, SJ, Edgar, MA,
Webb, PJ and Ransford, AO (1991) Spinal cord monitoring in scoliosis surgery. Experience with 1168 cases.
J. Bone Joint Surg., 73B: 487491.
Gokaslan, ZL, Samudrala, S, Deletis, V, Wildrick, DM and
Cooper, PR (1997) Intraoperative monitoring of spinal
cord function using motor evoked potentials via transcutaneous epidural electrode during anterior cervical spinal
surgery. J. Spinal Disord., 10: 299303.
Haghighi, SS (2002) Monitoring of motor evoked potentials with high intensity repetitive transcranial electrical
stimulation during spinal surgery. J. Clin. Monit. Comput., 17: 301308.
cases of quadriparesis following anterior cervical discectomy, with normal perioperative somatosensory evoked
potentials. J. Neurol. Neurosurg. Psychiatry, 74: 273276.
Kitagawa, H, Itoh, T, Takano, H, Takakuwa, K, Yamamoto, N, Yamada, H and Tsuji, H (1989) Motor evoked
potential monitoring during upper cervical spine surgery. Spine, 14: 10781083.
Kombos, T, Suess, O, Da Silva, C, Ciklatekerlio, O, Nobis,
V and Brock, M (2003) Impact of somatosensory
evoked ptential monitoring on cervical surgery. J. Clin.
Manninen, PH (1998) Monitoring evoked potentials during
spinal surgery in one institution. Can. J. Anaesth., 45:
460465.
May, DM, Jones, SJ and Crockard, HA (1996) Somatosensory evoked potential monitoring in cervical surgery:
identification of pre- and intraoperative risk factors
associated with neurological deterioration. J. Neurosurg., 85: 566573.
SPINE SURGERY
Nuwer, MR (1998) Spinal cord monitoring with somatosensory techniques. J. Clin. Neurophysiol., 15: 183193.
Nuwer, MR, Dawson, EG, Carlson, LG, Kanim, LEA and
spinal cord monitoring reduces neurologic deficits after
scoliosis surgery. Electroencephalogr. Clin. Neurophysiol., 96: 611.
Sebastian, C, Raya, JP, Ortega, M, Olalla, E, Lemos, V and
Romero, R (1997) Intraoperative control by somatosensory evoked potentials in the treatment of cervical myeloradiculopathy. Eur. Spine J., 6: 316323.
Taunt, CJ, Sidhu, KS and Andrew, SA (2005) Somatosensory evoked potential monitoring during anterior cervical discectomy and fusion. Spine, 17: 11701172.
677
Wiedemayer, H, Fauser, B, Sandalcioglu, IE, Schafer, H and
Stolke, D (2002) The impact of neurophysiological intraoperative monitoring on surgical decisions: a critical analysis of 423 cases. J. Neurosurg., 96: 255262.
Wiedemayer, H, Sandalcioglu, IE, Armbruster, W, Regel,
J, Schaefer, H and Stolke, D (2004) False negative findings in intraoperative SEP monitoring: analysis of 658
consecutive neuro surgical cases and review of published reports. J. Neurol. Neurosurg. Psychiatry, 75:
280286.
Yu, YL and Jones, SJ (1985) Somatosensory evoked potentials in cervical spondylosis: correlation of median,
ulnar and posterior tibial nerve responses with clinical
and radiological findings. Brain, 108: 273300.

M.R. Nuwer (Ed.)
678
CHAPTER 48
Monitoring during lumbar stenosis and fusion surgery

Bret Balla and Paul M. Huddlestonb,*
a
b
Department of Neurologic Surgery, Mayo Clinic, Rochester, MN 55905, USA
Department of Orthopedic Surgery, Mayo Clinic, Rochester, MN 55905, USA
48.1. Introduction
Ninety percent of people suffer from back pain at some
point in their lives. In the majority of people, the pain
will resolve spontaneously without any intervention.
Other people have pain that may last somewhat longer
but responds to more conservative therapies including
medical management, physical medicine and rehabilitation, or interventional procedures such as epidural,
foraminal, or joint injections. However, for some, the
pain can be persistent and progressive in spite of any
or all of these measures and significantly compromise
their ability to participate in activities of daily living
and their overall quality of life. Frequently, these
patients are debilitated by their pain, dependent on
chronic narcotics, confined to sedentary lifestyles,
and suffer from depression. For these patients, spinal
corrective surgery can be a life-redeeming event.
However, any operative procedure has inherent
risks. Reported incidences of complication following
spinal surgery range from <1% to 33%. Of particular
concern during spinal surgery is injury to neural structures intimately associated with the spinal column
including the spinal cord, conus medullaris, the cauda
equina, nerve roots, and proximal spinal nerves. The
possibility of iatrogenic injury to any of these structures is always a possibility and of serious concern to
patients and surgeons. It would not be uncommon or
unheard of for an otherwise satisfactory clinical outcome following surgery to be completely lost in dealing with a patients numbness or weakness that was
not present preoperatively. The development of surgical techniques to improve the safety of thoracolumbar
Correspondence to: Paul M. Huddleston, M.D., Department of Orthopedic Surgery, Mayo Clinic, 200 First Street
SW, Rochester, MN 55905, USA.
Tel.: 1-507-284-8309; fax: 1-507-266-4234.
E-mail: huddleston.paul@mayo.edu (P.M. Huddleston).
spinal surgery remains an area of constant evolution

(Rose et al., 1997; Wilson-Holden et al., 1999).
The placement of spinal implants for the purposes of
manipulation or stabilization of the spine remains one of
the critical steps during many thoracolumbar operations, both in adults and in children, for elective and
traumatic cases (Holland, 1998). Metallic implants are
often placed in and around the vertebra to serve as an
anchor for spinal rods or in the fixation of vertebra
either in isolation or in combination with other implants
such as plates. These implants assist in obtaining and
maintaining a vertebral reduction during spinal fusion
procedures. Accurate placement of pedicle screws is
critical because of their immediate proximity to neural
elements. The continued use of these implants has
improved the surgical treatment of patients by decreasing the use of casting and bracing, increased fusion
rates, and increased the degree of deformity correction
possible as well as maintenance of that advantage.
For example, children undergoing spinal fusion for a
symptomatic spondylolisthesis no longer need be treated with a hip spica cast and 36 months of prolonged
bed rest as they may be instrumented and mobilized
immediately.
Surgeons have used a variety of techniques to assist
in the accurate placement of pedicle screws. These
include an intimate knowledge of anatomic landmarks, operative planning with pre- and intraoperative
radiography, and direct visualization of the vertebral
anatomy, when feasible. However, following the placement and obtaining secure fixation in the vertebra,
the surgeon is ultimately interested in preserving the
integrity of the nervous system. Historically, a primary
method used to assess the nervous system was the
intraoperative wake-up test. Patients were weaned
from anesthesia while still in the operating theater and
asked to perform basic movements to form a rudimentary physical examination during surgery. Although still
useful in specific clinical scenarios, the wake-up test
is undesirable for a variety of reasons. It is time
SPINE SURGERY
consuming, technically difficult to complete, uncomfortable for patients, and, because of the effects of
anesthesia and patient cooperation, can be unreliable.
A surrogate method for assessing the integrity of the
nervous system during surgery is the use of intraoperative electrophysiological monitoring (Balzer
et al., 1998; Bose et al., 2002). A variety of different
techniques are available such as electromyography
(EMG), somatosensory evoked potentials (SEPs), dermatomal SEP (DSEP), and pedicle screw testing, but
the basic principle remains the same: probes are used
to detect and monitor spontaneous and evoked electrical
signals as they are transmitted through the nervous
system. If, during or after a surgical maneuver, changes
in the monitoring are observed such as variations in
amplitude or delays in transmittance, the possibility of
compromise of neural pathways should be considered
and the surgical plan reassessed (Welch et al., 1997).
The decision to use intraoperative monitoring is
guided by several principles. First, monitoring signals
must be obtainable. The inability to obtain satisfactory
signals is not an infrequent obstacle in successful monitoring with reported failures of as high as 4%. Furthermore,
inability to obtain satisfactory monitoring has been associated with poorer outcomes. Second, monitoring should
guide or affect the operative course. For example, if
significant signal changes would not lead to an alteration
in the operative procedure, the utility of monitoring is lost.
Third, monitoring should be safe and effective. If
monitoring were too risky or the results too unreliable, it
should not be used. For example, direct spinal cord
monitoring could be very accurate and reliable, but the
risks of such monitoring might outweigh the benefits.
On the other hand, cutaneous electrodes may be very safe,
but are sometimes insufficiently accurate to be useful.
A variety of intraoperative monitoring techniques
have been described including SEP, DSEP, evoked
and spontaneous EMG, motor evoked potentials
(MEP), magnetoencephalography (MEG), or pedicle
screw stimulation. Of these, MEG has proven less useful in the operating room because of technical difficulties and anesthesia interference with monitoring
(DiCindio and Schwartz, 2005). DSEPs have been
heavily marketed but have not proven to be any better
than traditional monitoring techniques. The workhorse
techniques most commonly used in the operating room
are SEP, MEP/EMG, and pedicle screw stimulation.
Somatosensory evoked potentials evaluate the
transmission of signals from peripheral nerves through
the dorsal columns and up to the sensory cortex.
A signal is applied to a peripheral nerve, such as the
679
posterior tibial nerve, and the resultant transmission

is monitored at sites proximal and distal to the level
of surgery, typically in the skin overlying the spine
and scalp. Both the amplitude and the latency of the
signal are assessed and used to infer the integrity of
the associated neural pathways. For example, a deficiency in either the amplitude or the latency of a signal
could be suggestive of injury to intervening pathways.
One of the limitations of SEP is that it only monitors
sensory pathways and is not always predictive of postoperative motor deficits.
Motor evoked potentials do directly assess motor
pathways. They generally involve the placement of
electrodes over the motor cortex in the scalp. A stimulating current is applied and the resulting motor activity measured to assess the transmission of signals from
the motor cortex through the brainstem and spinal
cord and into the peripheral nerves terminating in
muscle electrical activity. Although axonal activity
can be detected in nerves, muscle activity is generally
measured to ensure detected signals that are truly motor
in nature. There are some drawbacks to MEP when
compared with SEP. They may be more frequently
affected by anesthetic agents like paralytics, and guidelines for interpretation are less well established. They
do assess the motor system, which, of a general rule,
is of more interest to the surgeon. They may also be
ableto provide feedback when adequate SEP monitoring is unobtainable (Wilson-Holden et al., 2000).
Another monitoring technique is direct pedicle
screw stimulation (Calancie et al., 1994; Clements
et al., 1996; Danesh-Clough et al., 2001). One of the
earliest descriptions of using direct pedicle screw stimulation for the verification of screw placement was by
Young et al. (1995) when they published a case report
describing application of a stimulating current to a
pedicle screw after placement. Monitoring electrodes
were placed in muscles innervated by adjacent spinal
nerves and an incrementally increasing voltage was
applied to the screw until an EMG response was noted.
Higher threshold voltages were suggestive of a greater
degree of insulation between the pedicle screw and the
nerve root and hence safer screw placement. A screw
that stimulated at an abnormally low voltage would
be worrisome for possible neural compromise. Alternatively, a tapped hole can be stimulated before the
placement of the pedicle screw giving similar information but before the placement of the screw. An interesting insight into the complexities and difficulties of
pedicle screw placement and monitoring intraoperatively was made by J.A. Finkelstein (2003) when he
680
compared and contrasted two studies evaluating the

placement of thoracolumbar pedicle screws. The studies largely used similar methods but resulted in contradicting conclusions and posed some interesting
questions. Do intraoperative monitoring changes predict patient outcomes? Should pedicles be stimulated
before or after screw placement? If signal changes
are observed when a screw is placed, is it too late
because the damage has already been done? Will monitoring affect overall patient outcomes or only the postoperative radiograph?
In spite of these questions, however, a few generalizations and guidelines can be made to assist a surgeon in interpreting monitoring results while placing
a pedicle screw. Whenever possible, direct nerve
stimulation should be undertaken as a positive control. Particularly in cases of radiculopathies and
severe stenosis, severe chronic nerve compression
can result in axonomesis. These nerves can have
falsely elevated stimulation values even to direct
probing. If the nerve is not tested as a positive control, and pedicle screw stimulation reveals an abnormal value, the surgeon cannot be sure whether the
findings are a result of chronic nerve injury or acute
injury secondary to the surgical procedure.
The utility of intraoperative monitoring lies in its
ability to provide real-time feedback on the integrity
of the neural pathways during surgery (Young et al.,
1995). For example, if during a critical step in the
procedure monitoring changes are observed, an
attempt at reversal can be made and frequently the
monitoring changes return to baseline. Intraoperative
monitoring can also provide useful information about
the successful decompression of a compressed nerve.
Ultimately, however, it should always be remembered that intraoperative monitoring is an adjuvant
technique, and its capabilities should not be overestimated. It is a screening test, subject to the principles
of sensitivity and specificity and its use should be evaluated based on objective results. A significant amount
of data has been published regarding the accuracy and
relevance of intraoperative monitoring during lumbar
spine surgery, leading to a wide variety of conclusions
(Reidy et al., 2001; Raynor et al., 2002). Some authors
have reported absolutely no benefit from intraoperative monitoring while other authors have reported
improvements in the incidence of misplaced screws
and in postoperative outcomes (Tsai et al., 1997). A
survey of some of the more relevant intraoperative
monitoring studies involving lumbar stenosis and
fusion surgeries can be seen in Table 1.
B. BALL AND P.M. HUDDLESTON
Several factors can limit the reliability and usefulness of monitoring techniques (Minahan et al., 2000).
First of all, monitoring is a technically demanding
modality with a high level of inter-operator variability. It requires a well-trained staff capable of obtaining consistent and reproducible results. Second, the
incidence of false positive tracings can result in
unnecessary screw revision and complication of the
surgical procedure. Third, accurate recordings frequently cannot be obtained intraoperatively for a
variety of reasons. Fourth, intraoperative monitoring
requires appropriate anesthesia. A variety of papers
have been published, suggesting specific anesthetic
regimens for patients undergoing monitored surgery,
not always in agreement. Fifth, the success of monitoring may be dependent on the type of injury to
the nerve. For example, monitoring may be efficient
at detecting blunt injury to a nerve but less accurate
at detecting sharp injuries. Sixth, overzealous direct
nerve stimulation should be avoided to minimize any
possibility of associated nerve injury. Finally, monitoring should not be used as a surrogate for good surgical technique, including adequate anatomic dissection
and identification of surgical landmarks, consistent
probing of pedicles during successive steps in order
to assure integrity and accurate placement of pedicle
screws, and adequate preoperative planning and coordination with radiographic findings. Finally, results
should be confirmed and surgical technique be continuously improved through appropriate postoperative
imaging.
Here are two cases from our practice that give
examples of how intraoperative monitoring has been
useful in guiding operative management.
48.1.1. Case 1
A very pleasant 76-year-old female with a history of
a right frontal grade 3 astrocytoma status post gross
total resection presented with a history of chronic
back pain that had originally been evaluated and
treated at an outside institution. Three years before
presentation, she had been diagnosed with stenosis
at the L45 interspace and had undergone an L45
laminectomy and instrumented arthrodesis of L3 to
the sacrum. Immediately postoperatively, the patient
had developed a novel back pain starting in her low
back and radiating to her left thigh.
Over the subsequent 3 years, the patient underwent multiple interventions for her pain including
epidural and facet injections, anesthetic injections
SPINE SURGERY
681
Table 1
Outcomes of nine published reports of intraoperative monitoring in lumbosacral spine surgery
Study
Methods
Conclusions
Useful
Thuet et al. (2005)
Retrospective review of 4,310

monitored patients.
Yes
Krassioukov et al.
(2004)
Retrospective review of 161

consecutive monitored cases.
Tsirikos et al. (2004)
patients with spinal fractures.
Gunnarsson et al.
(2004)

consecutive patients
monitored during
thoracolumbar spine
procedures.
consecutive lumbar spine
fusions monitored with SEPs.
Patients in which SEPs or EMG were

unobtainable intraoperative (1.37%)
had higher incidence of
postoperative deficit.
SEP and EMG provide effective
feedback guiding technique
during complex lumbar
procedures.
A loss of SEP signal >50% was 67%
sensitive and 71% specific for
predicting postoperative neurological
deficit.
EMG has high sensitivity (100%) but
low specificity (23.7%) while SEP
had lower sensitivity (28.6%) and
higher specificity (94.7).
Gundanna et al. (2003)
Iwasaki et al. (2003)

cases of intraoperative spinal
monitoring with SEP and
CMAP.
Norcross-Nechay et al.
(1999)
Retrospective study of 70
patients undergoing correction
lumbar stenosis.
Tsai et al. (1997)
33 consecutive patients
undergoing microdecompression of lumbosacral
radiculopathy.
SEP-monitored lumbosacral
procedures.
Stechison et al. (1995)
No cases of SEP changes were

observed, but 5 patients (2.7%) had
new postoperative radiculopathic
findings.
Predictive value of monitoring was
positive 75% and negative 99%
for new postoperative neurological
finding. 6.1% could not be
recorded.
9 of 12 patients with SEP changes were
successfully reversed intraoperative.
The 3 unreversed had new
postoperative weakness. 80% of
patients with normal SEP at
termination had improved pain
postoperative, while only 54% of
patients with abnormal SEP had
improved pain.
Improvement in DSEP latency
intraoperatively did not correlate
with patient outcomes on
postoperative evaluation.
12% of cases had significant
monitoring changes intraoperatively.
18% of these cases with changes had
new postoperative findings. No cases
without changes had new
postoperative findings.
Yes
Yes
Yes
No
Yes
Yes
No
Yes
Seven out of nine studies reported utility of intraoperative monitoring for lumbar spine surgery for guiding surgical technique and predicting patient outcomes. This likely reflects some publication bias, but does support the utility of monitoring in some situations. The majority
of the benefits were related to predicting postoperative outcomes and guidance of intraoperative technique. No studies investigating the
cost-effectiveness of monitoring were identified.
SEP: somatosensory evoked potentials; EMG: electromyography; DSEP: dermatomal SEP; CMAP: compound muscle action potential.
682
around her instrumentation, medical management,

and physical medicine and rehabilitation. The
implant injections provided some short-term relief,
but otherwise the patient had experienced no meaningful improvement.
Her physical examination was complicated by the
right frontal tumor resection which resulted in
subsequent mild left-sided upper and lower extremity
weakness and poor coordination. However, the
patient felt that these had remained stable since her
tumor resection. She had severe difficulty walking,
with only limited ambulation with a walker in a
severe stooped position. She had previously been
diagnosed with depression and was in a dire state of
deconditioning, both likely being secondary to her
chronic pain. Preoperative radiology including plain
anteroposterior and lateral films and a lumbar magnetic resonance imaging (MRI) revealed a low-set
screw in the left L3 pedicle (Fig. 1).
It was determined that the patients pain was a result
of two factors. First, the loose instrumentation was
causing significant back pain. Second, a left L3 radiculopathy was causing the left leg pain. The patient was
taken to the operating suite for a re-exploration, instrumentation removal, repeat decompression, and a left
L3 foraminotomy. Intraoperative EMG monitoring
was performed throughout the procedure.
After exploration and exposure of the implants,
direct screw stimulation was performed. The left
L3 stimulated at a low 11 mA. Following removal
of all the screws, the pedicle holes themselves were
stimulated with active EMG. The left L3 pedicle hole
stimulated the L3 nerve root at 5 mA with activation
of the thigh muscles. The revision laminectomy,
decompression hemiforaminotomy, and exploration
allowed direct visualization of the nerve root. Direct
probing of the nerve root itself stimulated at 3 mA.
Please see Fig. 2 for further details.
Postoperatively, the patients condition rapidly
improved, and she experienced essentially complete
resolution of her back and leg pain. She was transferred to an acute rehabilitation unit where she was
able to walk with a walker and participate in rehabilitation exercises without limitation, and was eventually dismissed home.
This case demonstrated the utility of direct screw
stimulation in identifying misplaced pedicle screws.
The preoperative imaging revealed a left L3 pedicle
screw that appeared to be abnormally placed. However, given the limitations of the plain radiographs
and artifact in the MRI, it was impossible to clearly
identify any hardware impingement on the exiting

adjacent nerve roots. However, when the L3 nerve
root stimulated at an abnormally low level through
the pedicle screw, it suggested that pedicle screw
impingement on the nerve root was contributing to
her pain. It also hints that if monitoring had been
used during her initial operation, the communication
could have been detected and revision made in a
more timely manner.
48.1.2. Case 2
A very pleasant 61-year-old female with a long history
of chronic back pain presented status post posterior
lumbar instrumented fusion at L5S1 at an outside
institution. She initially was seen 1 year ago with persistent low back pain that was not relieved by her prior
procedure. Lidocaine injections around her instrumentation yielded temporary relief, suggesting her pain
was referable to her implants that we subsequently
removed. She did well for the first several months until
she began to notice new, increasing left leg pain. The
pain started in the left buttock and radiated down into
the left thigh, and was exacerbated with weight
bearing. She also noticed left leg weakness and began
needing a cane for assistance in ambulation. She
denied any symptoms on the contralateral side. Physical examination was notable for mild left quadriceps
weakness and a diminished left patellar reflex.
Imaging revealed solid arthrodesis of L5S1 and
the development of degenerative joint disease at the
L34 and L45 levels, most pronounced at L45
which can be appreciated on her MRI findings in
Fig. 3. The MRI of the lumbar spine revealed a small
left lateral disk herniation and hypertophic changes at
the L45 facet joint, resulting in L4 foraminal stenosis.
It was determined that the patient had developed
adjacent segment disease resulting in left L3 and L4
radiculopathies. The patient was taken to the
operating suite for a posterior lumbar decompression
at L34 and L45.
Intraoperative monitoring was used throughout the
case. The left lumbar vertebrae were exposed using
sharp and blunt dissection. The neural threshold for
the left L3 was noted to be 14 mA. A left L3 laminofoaminotomy was performed, and the left L3 and L4
hemilaminectomies were completed. The decompression was continued laterally flush to the wall of the
pedicle until the nerve root was visualized and noted
to exit with abundant space through the L3 foramen.
After decompression, the L3 nerve readily stimulated
SPINE SURGERY
683
Fig. 1. Case 1, preoperative radiology. Case 1 presented with a history of chronic back pain radiating down into her left
anterior thigh. Radiographic evaluation of the patient during preoperative work-up revealed an inferiorly situated screw in
the left L3 pedicle. A: Plain anteroposterior flat plate film of the lumbar spine and instrumentation. B: Lateral lumbar spine
plain film of the lumbar spine. C: Sagittal magnetic resonance imaging (MRI) through the plane of the left pedicle screws
shows the distortion around the left L3 pedicle screw abutting the adjacent nerve.
at 5 mA. A similar procedure was completed at the

L4 level with the nerve root stimulating at 810 mA.
Postoperatively, the patient did very well with
complete resolution of her buttock and thigh pain.
After a short postoperative recovery, the patient was
dismissed home with relief of her preoperative
symptoms.
This case demonstrates how by giving feedback on

whether nerve function is lost or worsened, electrophysiological monitoring can change the course and
execution of the case. In this situation, safely exposing
the neural elements in an abnormally confined space
challenges the surgeon at several points during the
foraminal decompression. Free run EMG tracings
684
coupled with audible feedback temporally related to

the passage and placement of surgical instruments will
tell the surgeons that the tools they are using are
too big or their actions are too rough. An immediate change in the instrument size or technique allows
an intraoperative change to be made instead of trying
to correlate a new deficit with an intraoperative maneuver in hindsight. Although controversial, measured
improvement in the stimulation threshold in an exiting
nerve root may help the surgeon assess the level of
decompression and gives reassurance that an adequate
operation has been performed (Limbrick and Wright,
2005). In this case, the L3 root was stimulating at an
abnormally high level, suggestive of physical compromise of the neuronal pathways. As the posterior
T
r
c
1
decompression and foraminotomy were completed,

a measurable improvement in the stimulation threshold
was observed. This suggests two hypotheses; either
a better response was obtained because of a more
ideal placement of the stimulating probe as the surgeon could better visualize the neural elements, or
there exists a measurable improvement in neural function that may guide the operator in the depth and
breadth of the surgical technique. If the first hypothesis is true, then the improved signals serve as a surrogate for measuring the extent of the decompression
and the avoidance of any residual neural compression.
If the latter hypothesis is correct, the intraoperative
result is the same; a complete decompression. Although
such a measurable change in electrophysiological
7
8
A2
(Fig. 2 continued)
SPINE SURGERY
685
B2
Fig. 2. Case 1, intraoperative photographs and corresponding intraoperative monitoring findings. A: Sample of direct
pedicle screw stimulation after dissection and exposure of the instrumentation. Stimulation was detected at 11.7 mA.
B: Probing of the hole in the pedicle after screw removal stimulated at 4.7 mA.
(Fig. 2 continued)
C2
Fig. 2. Contd. C: Direct stimulation of the exposed nerve root after completion of the laminoframinotomy stimulated at
3.1 mA.
Fig. 3. Case 2, preoperative imaging. Axial A: and sagittal B: T2 weighted preoperative magnetic resonance imaging
(MRI) scans obtained during the work-up and evaluation of Case 2 who presented with a recurrence of low back pain radiating into the left leg after removal of painful instrumentation 1 year prior. The scans were suggestive of left L3 and L4
foraminal stenosis.
SPINE SURGERY
monitoring is not always observed, it does provide

two benefits: first, it demonstrates abnormal function
ofthe nerve before decompression thereby establishing
a baseline; and second, it objectifies any change,
if any, in nerve function as a direct result of the
decompression.
48.2. Summary
Intraoperative electrophysiological monitoring is a
useful adjuvant modality in guiding operative technique during lumbar spinal stenosis and fusion surgery. Although there is still some debate over its
specific utility, the majority of the data seems to suggest that when used appropriately, intraoperative
monitoring will lead to an increased surgeon and surgical assistant awareness of the consequences of their
actions and can lead to the avoidance of complications and improvement in patient outcomes with very
little risk.
References
Balzer, JR, Rose, RD, Welch, WC and Sclabassi, RJ (1998)
Simultaneous somatosensory evoked potential and electromyographic recordings during lumbosacral decompression and instrumentation. Neurosurgery, 42: 13181324.
Bose, B, Wierzbowski, LR and Sestokas, AK (2002) Neurophysiologic monitoring of spinal nerve root function
during instrumented posterior lumbar spine surgery.
Spine, 27: 14441450.
Calancie, B, Madsen, P and Lebwohl, N (1994) Stimulusevoked EMG monitoring during transpedicular lumbosacral spine instrumentation. Initial clinical results.
Spine, 19: 27802786.
Clements, DH, Morledge, DE, Martin, WH and Betz, RR
(1996) Evoked and spontaneous electromyography to
evaluate lumbosacral pedicle screw placement. Spine, 21:
600604.
Danesh-Clough, T, Taylor, P, Hodgson, B and Walton, M
(2001) The use of evoked EMG in detecting misplaced
thoracolumbar pedicle screws. Spine, 26: 13131316.
DiCindio, S and Schwartz, DM (2005) Anesthetic management for pediatric spinal fusion: implications of
advances in spinal cord monitoring. Anesthesiol. Clin.
N. Am., 23: 765787.
Finkelstein, JA (2003) Can triggered electromyograph
thresholds predict safe thoracic pedicle screw placement. Spine, 28: 960.
Gundanna, M, Eskenazi, M, Bendo, J, Spivak, J and Moskovich, R (2003) Somatosensory evoked potential monitoring of lumbar pedicle screw placement for in situ
posterior spinal fusion. Spine J., 3: 370376.
687
Holland, NR (1998) Intraoperative electromyography during thoracolumbar spinal surgery. Spine, 23:
19151922.
Iwasaki, H, Tamaki, T, Yoshida, M, Ando, M, Yamada, H,
Tsutsui, S and Takami, M (2003) Efficacy and limitations
of current methods of intraoperative spinal cord monitoring. J. Orthop. Sci., 8: 635642.
complex lumbosacral procedures: indications, techniques, and long-term follow-up review of 61 consecutive cases. J. Neurosurg. Spine, 1: 243253.
Limbrick, DD, Jr. and Wright, NM (2005) Verification of
nerve root decompression during minimally-invasive
lumbar microdiskectomy: a practical application of
surgeon-driven evoked EMG. Minim. Invasive Neurosurg., 48: 273277.
Minahan, RE, Riley, LH, 3rd, Lukaczyk, T, Cohen, DB and
Kostuik, JP (2000) The effect of neuromuscular blockade on pedicle screw stimulation thresholds. Spine, 25:
25262530.
Norcross-Nechay, K, Mathew, T, Simmons, JW and Hadjipavlou, A (1999) Intraoperative somatosensory evoked
potential findings in acute and chronic spinal canal
compromise. Spine, 24: 10291033.
Raynor, BL, Lenke, LG, Kim, Y, Hanson, DS, WilsonHolden, TJ, Bridwell, KH and Padberg, AM (2002)
Can triggered electromyograph thresholds predict safe
thoracic pedicle screw placement? Spine, 27:
20302035.
Reidy, DP, Houlden, D, Nolan, PC, Kim, M and
Finkelstein, JA (2001) Evaluation of electromyographic
monitoring during insertion of thoracic pedicle screws.
J. Bone Joint Surg. Br., 83: 10091014.
Rose, RD, Welch, WC, Balzer, JR and Jacobs, GB (1997)
Persistently electrified pedicle stimulation instruments
in spinal instrumentation. Technique and protocol
development. Spine, 22: 334343.
Stechison, MT, Panagis, SG and Reinhart, SS (1995)
Somatosensory evoked potential monitoring during spinal surgery. Acta Neurochir. (Wien), 135: 5661.
Thuet, ED, Padberg, AM, Raynor, BL, Bridwell, KH,
Riew, KD, Taylor, BA and Lenke, LG (2005) Increased
risk of postoperative neurologic deficit for spinal surgery patients with unobtainable intraoperative evoked
potential data. Spine, 30: 20942103.
Tsai, RY, Yang, RS, Nuwer, MR, Kanim, LE, Delamarter,
RB and Dawson, EG (1997) Intraoperative dermatomal
evoked potential monitoring fails to predict outcome
688
from lumbar decompression surgery. Spine, 22:
19701975.
somatosensory evoked potentials for spinal trauma.
Welch, WC, Rose, RD, Balzer, JR and Jacobs, GB (1997)
Evaluation with evoked and spontaneous electromyography during lumbar instrumentation: a prospective
Wilson-Holden, TJ, Padberg, AM, Lenke, LG, Larson, BJ,
Bridwell, GS and Bassett, GS (1999) Efficacy of

intraoperative monitoring for pediatric patients with spinal cord pathology undergoing spinal deformity surgery.
Spine, 24: 16851692.
Wilson-Holden, TJ, Padberg, AM, Parkinson, JD, Bridwell,
KH, Lenke, LG and Bassett, GS (2000) A prospective
comparison of neurogenic mixed evoked potential stimulation methods: utility of epidural elicitation during
posterior spinal surgery. Spine, 25: 23642371.
Young, WF, Morledge, DE, Martin, W and Park, KB (1995)
Intraoperative stimulation of pedicle screws: a new
method for verification of screw placement. Surg. Neurol.,
44: 544547.

M.R. Nuwer (Ed.)
689
CHAPTER 49
Surgery for tethered cord syndrome and other

cauda equina lesions
Aatif M. Husain*
Department of Medicine (Neurology), Duke University Medical Center and Neurodiagnostic Center,
Veterans Affairs Medical Center, Durham, NC 27710, USA
49.1. Introduction
49.2. Anatomy and pathology of TCS
Tethered cord syndrome (TCS) is a disorder that

results from the anchoring of the caudal end of the spinal cord to fixed, inelastic tissues. This anchoring produces a stretch-induced injury that results in impaired
oxidative metabolism in the lower spinal cord (Yamada
et al., 1981). Whereas TCS usually presents in children,
it can occur in adults and usually presents with motor,
sensory, and sphincter dysfunction involving the lower
limbs. Surgery to release the tethering of the spinal
cord is recommended to prevent further deterioration
and possibly reversal of the deficits.
Other types of lesions, such as intra- and extradural
tumors, bony lesions, congenital malformations,
degenerative changes, acquired infections, and trauma
can also produce symptoms similar to TCS. In these
cases, symptoms occur because of pressure effects on
the cauda equina or individual nerve roots. Often surgery is necessary in these cases to prevent progression
of disability and provide relief of symptoms.
Surgery for TCS and other cauda equina lesions
involves very careful dissection as often lower lumbar
and sacral nerve roots are embedded in the lesion and
can be damaged during surgery. Neurophysiologic
intraoperative monitoring (NIOM) can be helpful in
identifying these neural elements during surgery so
that they can be spared. This chapter will briefly
review the anatomy, pathology, and clinical features
of TCS and describe TCS surgery. Thereafter, techniques used for NIOM for surgeries for TCS and cauda
equina lesions and their utility will be discussed.
Tethering of the spinal cord occurs most often in association with spinal dysraphism and other congenital
disorders of the spinal cord (Yamada et al., 2004).
Spinal dysraphism results from impaired development
of the spinal cord during the first few weeks of gestation. During embryonic development, the nervous system develops from the neural tube. The neural tube
forms by a process known as primary neurulation
which begins on day 21 of gestation with proliferation
and folding of the edges of the neural ectoderm. Once
the edges of the neural ectoderm fuse, the neural tube is
formed. The caudal portion of the neural tube will form
the spinal cord down to the level of the S2 segment.
Proliferation of the caudal cell mass (CCM), in a process know as secondary neurulation, produces the
remainder of the spinal cord, conus medullaris, and
the filum terminale. By 67 weeks this process is
largely complete, and a layer of cutaneous ectoderm
overlies the newly formed spinal cord. Continued
intrauterine growth results in atrophy of the filum terminale and upward movement of the spinal cord in
relation to the vertebral column (McLone, 1999).
At the time of birth the terminal end of the spinal cord
lies at the level of L1/L2 due to disproportionately
greater growth of the vertebral column (Wilson and
Prince, 1989; Beek et al., 1996).
Open spinal dysraphism occurs when there is incomplete closure of the neural tube associated with defects
in the overlying skin. The resulting myelomeningocele
is evident at birth and often with intrauterine ultrasound.
Surgical correction of the myelomeningocele is undertaken shortly after birth and accompanying tethering
of the spinal cord released at the same time.
Occult spinal dysraphism refers to a group of conditions in which a covering of skin overlies the spinal
malformation. Lipomyelomeningocele, dermal sinus,
Correspondence to: Aatif M. Husain, M.D., Department of

Medicine (Neurology), Duke University Medical Center,
Box 3678, 202 Bell Building, Durham, NC 27710, USA.
Tel.: 1-(919)-684-8485; fax: 1-(919)-684-8955.
E-mail: aatif.husain@duke.edu (A.M. Husain).
690
and diastematomyelia occur because of derangement

of primary neurulation, whereas terminal myelocystocele and thickened filum terminale occur when there is
impairment of secondary neurulation (McLone, 1999).
Lipomyelomeningocele, also referred to as spinal
lipoma, is a mass of adipose tissue, often palpable
under the skin, which tethers the spinal cord. Dermal
sinus connects the surface skin to the underlying
spinal cord resulting in tethering. Often an accompanying epidermoid tumor may cause compression of
the cauda equina or lower spinal cord in addition to
TCS. Diastematomyelia, or split cord malformation,
refers to a condition in which boney or cartilaginous
septum splits the spinal cord into two hemicords. Both
hemicords may either have separate dural and arachnoid sheaths (type I) or share one sheath (type II). As
the name implies, terminal myelocystocele is a condition in which the lower spinal cord is ballooned into a
cystic structure. A thickened filum terminale occurs
when there is a failure of atrophy of CCM and the
filum terminale during intrauterine growth. There are
numerous other rare syndromes that can result in
TCS, and the reader is referred to more comprehensive reviews to learn more about these syndromes
(McLone, 1999; Muraszko, 1999; Zerah et al., 1999).
49.3. Symptoms of TCS and cauda equina lesions
As mentioned above, when TCS is associated with
open spinal dysraphism, diagnosis is easily established either prenatally or at birth. However, in cases
of occult spinal dysraphism associated with TCS and
other cauda equina lesions, diagnosis may be more
difficult but is no less important as early surgery
may minimize or reverse disabilities.
Clinical features of TCS are protean and involve
many organ systems. Neurological symptoms and signs
consist of: lower extremity weakness, gait impairment,
diminished or asymmetric reflexes, and atrophy. If the
lower spinal cord is involved, spasticity and hyperreflexia may occur. The sensory system is also involved
with patchy sensory loss and paresthesias in the legs
and saddle area (Pang and Wilberger, 1982; Sathi
et al., 1993; Warder and Oakes, 1994). Painless ulcers
may occur. Urogenital dysfunction is common and consists of urgency, frequency, incontinence, and recurrent
infections. Involvement of the anal sphincter results in
fecal incontinence and constipation. Musculoskeletal
abnormalities are also common and include: club feet,
scoliosis, asymmetric leg length, and hip subluxation.
Various types of occult spinal dysraphism are
A.M. HUSAIN
associated with cutaneous manifestations that suggest

their diagnosis. These include: subcutaneous lipoma,
dermal sinus, capillary hemangioma, caudal appendage, atretic meningocele, and hypertrichosis (Michelson and Ashwal, 2004).
The diagnostic testing modality of choice is magnetic resonance imaging (MRI) of the lower spine.
MRI often shows the caudal end of the spinal cord
to be lower than the L2 vertebral body (Fitz and
Harwood-Nash, 1975). The source of the tether, that
is a thickened filum terminale or other cauda equina
lesions, can often be visualized. Ultrasound (US) has
also been employed and is especially useful in neonates in whom spinal cord pulsations are diminished
in the setting of TCS. However, MRI has been shown
to have higher sensitivity than US in diagnosis of
occult spinal dysraphism (Hughes et al., 2003). Rarely,
a computed tomographic (CT) myelogram is needed
when there is a contraindication to MRI or if the latter
is equivocal (Roos et al., 1986).
49.4. Surgical considerations for TCS and cauda
equina lesions
Detailed methodology for spinal dysraphism and
cauda equina surgery is beyond the scope of this chapter, the reader is referred to standard neurosurgical
textbooks if this is needed (Muraszko, 1999; Sutton,
2005). Highlights of surgery as they relate to NIOM
will be presented here. The patient is positioned prone
and a skin incision is made overlying the spinal segments identified to be pathological on MRI. Fascia
are dissected until the underlying spinous processes
and laminae are reached. Laminectomy is performed
to expose the underlying dura. The dura is opened to
expose the lesion. Often lower lumbar and sacral nerve
roots are adherent to the lesion, and even with microsurgical dissection these cannot be easily identified.
It is during this dissection, when the surgeon is trying
to free the spinal cord and cauda equina from adhesions, that risk to the nerve roots is greatest and NIOM
is used. Once the cord and cauda equina are free from
adhesions and tethering of the cord released, the dura
is closed, often with the aid of a dural graft. Thereafter,
the muscle, fascia, and skin layers are closed.
The lower sacral nerve roots are very important to
preserve during dissection around the cauda equina
and when cutting the filum terminale. These roots
innervate the urinary bladder and the urethral and
anal sphincters and preserving their function is an
important goal of surgery. Motor control of the
SPINE SURGERY
detrusor muscle of the urinary bladder is exerted

mostly through the parasympathetic nervous system.
Preganglionic fibers lie in the cauda equina in the ventral rami arising from S2, S3, and S4 segments. The
fibers of these nerves traverse in the pelvic splanchnic
nerves through the vesical plexus and terminate on
the ganglia on the wall of the urinary bladder. From
there, short postganglionic fibers innervate the detrusor muscle. Activation of this parasympathetic system
causes detrusor muscle contraction, relaxation of the
internal uretheral sphincter and aids in voiding. Sympathetic nervous system innervation of the urinary
bladder does not appear to aid in emptying. Sympathetic supply originates from the upper lumbar segments of the spinal cord and supplies the inferior
mesenteric ganglion. Postganglionic sympathetic
fibers traverse the lumbar splanchnic nerves through
the vesical plexus to the detrusor muscle. Sympathetic
activation relaxes the detrusor muscle (Gilman and
Newman, 2002).
The external urethral and anal sphincters, however,
are under voluntary, not autonomic, control. The
pudendal nerve with contributions from S2, S3, and
S4 segments innervates these structures. The inferior
rectal and perineal nerves are branches of the pudendal
nerve that supply the external anal and urethral sphincters, respectively (Ryken and Menezes, 1994). Though
unlikely, it is possible that these two branches of the
pudendal nerve may be damaged differentially. These
anatomic considerations are important to consider
when designing NIOM paradigms for surgery for
TCS and cauda equina.
49.5. NIOM monitoring paradigms
Various types of NIOM can be performed in patients
undergoing TCS and other types of cauda equina surgery. These include monitoring motor and sensory
innervation of the lower extremities and sphincter
function. Often more than one monitoring modality
is used. Below is a description of the various modalities that have been used.
49.5.1. Motor pathway monitoring
One of the most common types of NIOM techniques
used in surgery for TCS and cauda equina lesions is
monitoring electromyographic (EMG) activity from
various lower extremity muscles (Shinomiya et al.,
1991; Legatt et al., 1992; Hormes and Chappuis,
1993; Kothbauer et al., 1994; Phillips and Jane, 1996;
691
Von Koch et al., 2002; Krassioukov et al., 2004;

Quinones-Hinojosa et al., 2004). The muscles monitored most frequently include the quadriceps femoris
(most often the rectus femoris, but sometimes also
the vastus lateralis) (L2, L3, and L4), tibialis anterior
(L4, L5), extensor hallucis longus (L5, S1), one of
the hamstrings (L5, S1, and S2), gluteus maximus
(L5, S1, and S2), and medial gastrocnemius (S1, S2)
muscles. Which one of these muscles is monitored is
determined in part by the spinal level of the surgery
and the number of channels available for monitoring.
It is important to consider monitoring the gluteus maximus muscle since it is a major hip extensor muscle and
plays a vital role in ambulation.
Subdermal monopolar needle electrodes are used
most often to monitor muscles during these types of
surgeries (Kothbauer et al., 1994; Phillips and Jane,
1996; Von Koch et al., 2002). However, surface
electroencephalography (EEG) cup electrodes have
also been used (Legatt et al., 1992). Needle electrodes
offer the advantage of quicker application and better
recording of neurotonic discharges (Emerson and
Adams, 2003).
Two types of EMG monitoring are useful in TCS
and cauda equina surgeries: continuous monitoring
for spontaneous activity and intermittent stimulation
of various structures in the operative field to elicit
compound muscle action potentials (CMAP). Modern
NIOM machines allow the EMG activity to be interfaced with a loudspeaker to allow easy identification
of the activity. With continuous EMG monitoring,
mechanical irritation of nerve roots during dissection
or irrigation produces brief bursts of spontaneous
activity indicative of the integrity of the nerves
(Fig. 1). Longer trains of neurotonic discharges suggest damage to the associated nerves (Emerson and
Adams, 2003) (Fig. 2). These discharges can be used
to alert the surgeon that neural elements may be
compromised.
In addition to continuous, free-running EMG, various structures within the surgical field can be stimulated to determine if they contain neural elements
(Fig. 3). Stimulation is usually carried out with a
bipolar stimulating electrode to ensure focal delivery
of the current. Bipolar cautery forceps can be used
for this purpose and are convenient as they allow
the surgeon to isolate the structure to be stimulated
(Legatt et al., 1992). Monopolar stimulators are usually not recommended since current delivery is not as
precise. However during dissections of large tumors
in or near the cauda equina, monopolar stimulators
692
A.M. HUSAIN
Fig. 1. Burst of spontaneous electromyographic (EMG) activity in the external anal sphincter channel. LQ left quadriceps femoris muscle, LAT left anterior tibialis muscle, LMG left medial gastrocnemius muscle, LH left hamstrings,
R right. Scale 100 ms/div and 1 mV/div.
are useful in searching for neural tissue in the tumor

mass.
Within the surgical field, nerve roots, spinal cord,
or structures that may contain neural elements (i.e.,
filum terminale) can be stimulated during dissection.
This helps isolate nerve roots from fibrous tissue and
determine when the filum terminale can be safely
cut. With a constant-voltage stimulus duration of
200 ms, the intensity needed to stimulate nerve roots
ranges from 0.05 to 1.0 V in children and 0.1 to 7 V
in adults (about 0.1 to 10 mA with a constant-current
stimulator) (Kothbauer et al., 1994; Von Koch et al.,
2002; Quinones-Hinojosa et al., 2004). If the filum terminale does not contain neural elements, it cannot be
stimulated at this intensity. Some investigators recommend increasing the intensity of the stimulus (up to
100 V) to stimulate the filum terminale. If the response
threshold between the roots and filum terminale is
1:100 (i.e., the filum terminale requires 100 times
higher stimulus intensity to stimulate than the roots),
it does not contain neural elements (Von Koch et al.,
2002; Quinones-Hinojosa et al., 2004). Responses
obtained at these high intensities represent current
spread to adjacent structures. Many investigators do
not use such high stimulus intensities to identify and
cut the filum terminale; rather, they advocate using
lower stimulation setting to prevent current spread
(Shinomiya et al., 1991).
Fig. 2. Neurotonic discharge in the left anterior tibialis muscle. LQ left quadriceps femoris muscle, LAT left anterior
tibialis muscle, LMG left medial gastrocnemius muscle, LH left hamstrings, R right. Scale 100 ms/div and 1 mV/div.
SPINE SURGERY
693
Fig. 3. Stimulated electromyographic (EMG) activity in a right hamstring muscle. LQ left quadriceps femoris muscle,
LAT left anterior tibialis muscle, LMG left medial gastrocnemius muscle, LH left hamstrings, R right. Scale
8 ms/div and 50 mV/div.
49.5.2. Sensory pathway monitoring

Sensory pathway monitoring with tibial somatosensory evoked potentials (SEP) has been used less consistently than motor pathway monitoring in TCS and
cauda equina surgery (Shinomiya et al., 1991; Legatt
et al., 1992; Hormes and Chappuis, 1993; Kothbauer
et al., 1994; Krassioukov et al., 2004). Various techniques have been used for SEP NIOM, with the commonest being stimulation of the tibial nerve along the
medial malleolus and recording off scalp and neck
electrodes (cortical and subcortical waveforms). These
techniques have been described in the section on SEP
monitoring.
As can be done with various types of spine surgeries, tibial SEP can be recorded with epidural electrodes. In one study, a catheter electrode was passed
upward into the epidural space from the surgical site
and tibial SEP recorded throughout TCS surgery
(Shinomiya et al., 1991). This type of recording allows
quicker data collection, but is more cumbersome for
the surgeon. Additionally, the electrode can move during surgery, distorting the response. Loss of cortical,
subcortical, or epidural responses after tibial nerve
stimulation is concerning for damage to the sensory
roots, once technical and systemic factors have been
excluded.
An innovative method of performing tibial SEP
uses double stimuli to elicit two cortical responses
(Nishijima et al., 1992).This method postulates that
the absolute refractory period after the first response

increases with neural compromise. Thus, if there is
damage to nerve during surgery, the second cortical
peak will be diminished in amplitude. The interval
needed between the two stimuli has not been determined, and this technique is not widely used.
Tibial SEP have also been used for identifying
sensory roots. Structures resembling sensory roots
can be stimulated with a hand-held probe by the surgeon and scalp responses averaged as is done with
tibial nerve SEP. The presence of a reproducible
response with scalp recording electrodes confirms
the presence of a sensory root (Kothbauer et al.,
1994; Kothbauer and Novak, 2004) (Fig. 4). This
process is time consuming and requires the surgeon
to hold the probe in one location while averaging is
completed.
SEP can also be obtained with stimulation of the
pudendal nerve branches. It has been suggested that
this monitoring is important to minimize the chance
of micturation problems and sexual dysfunction after
cauda equina surgery (Deletis et al., 1992; Kothbauer
and Novak, 2004). To obtain these responses, the
dorsal penile nerve is stimulated with the cathode and
anode on the dorsal surface of the penis, and the clitoral nerve is stimulated with the cathode on the clitoris
and anode on the adjacent labia. Recording is obtained
from various dorsal roots by the surgeon with a handheld electrode. This technique is especially important
694
A.M. HUSAIN
Fig. 5. Sample setup of needle electrodes in the anal

sphincter muscle. Note the presence of four electrodes,
one in each quadrant.
Fig. 4. Sample somatosensory evoked potentials (SEPs)

obtained after stimulation of the tibial nerve (A) and of a sensory nerve root in the lumbosacral region (B). Notice the
shorter latency of the response obtained after sensory nerve
root stimulation. Reproduced from Kothbauer et al. (1994)
with permission from Lippincott, Williams and Wilkins.
in selective dorsal rhizotomy (so that these fibers can

be spared); however, it may also be useful in identifying lower sacral roots during dissection around the
cauda equina.
49.5.3. Monitoring sphincter function
Preserving bowel and bladder sphincter function is
critical in surgeries involving the cauda equina.
Potentially, the external anal and external urethral
sphincters and the detrusor muscle can be monitored.
External anal sphincter monitoring is done most
commonly since this muscle is easily accessible. This
can be done in a number of different ways; however,
monitoring with monopolar needle electrodes (like
those used for EMG monitoring in the limbs) has been
reported most often (James et al., 1979; Shinomiya
et al., 1991; Legatt et al., 1992; Kothbauer et al.,
1994; Phillips and Jane, 1996; Von Koch et al., 2002;
Quinones-Hinojosa et al., 2004). Two monopolar needle electrodes (typically about 1 cm in length) are
inserted on either side of the external anal sphincter
and are referenced to each other. To provide more
redundancy, four electrodes, one in each quadrant,
can be used (Fig. 5). Only two are used at a given time,
however, if one or two become dislodged, the spare

ones can be used. Care must be taken to adequately
secure these electrodes so that they do not come out
during positioning and the procedure (Fig. 6). This
type of EMG monitoring of the external anal sphincter
is sensitive for free running and stimulated discharges.
Activity in this channel is similar to that seen in the
lower limbs, though occurs at a shorter latency.
Surface EMG electrodes can also be used to monitor the external anal sphincter (James et al., 1979;
Krassioukov et al., 2004). These plug electrodes consist of a sponge with two (or four) platinum leads
embedded on the surface (Fig. 7). After induction
of anesthesia, the plug electrode is inserted into the
Fig. 6. Sample setup of needle electrodes in the anal

sphincter muscle after they have been adequately secured.
SPINE SURGERY
695
Fig. 7. Anal sphincter plug electrode; this electrode

records surface electromyographic (EMG) activity. Reproduced from Krassioukov et al. (2004) with permission from
American Association of Neurological Surgeons.
Fig. 8. External urethral sphincter electrode (arrow); this

electrode records surface electromyographic (EMG) activity. Reproduced from Krassioukov et al. (2004) with permission from American Association of Neurological
Surgeons.
rectum so that the surface electrodes lie in contact

with the external anal sphincter. Like the needle electrodes, this electrode must be adequately secured to
prevent it form being dislodged. If the electrodes
make good contact with the sphincter, free running
and stimulated EMG can be recorded.
In the past, external anal sphincter pressure monitors were used to monitor function (Pang and Casey,
1983; Ikeda et al., 1986; Hellbusch and Nihsen,
1989). These devices consist of an air-filled balloon
connected to a pressure transducer. Stimulation of
the lower sacral nerve roots causes contraction of the
external anal sphincter, which in turn increases the
pressure in the balloon. This monitoring system requires more time and equipment than EMG monitoring
described above and is seldom used now.
The external urethral sphincter is harder to monitor than the external anal sphincter. Since this sphincter is not on the surface of the urethra, needle
electrodes cannot be inserted into it percutaneously.
Instead a ring electrode is attached 12 cm below
the balloon of a Foley catheter (Fig. 8). Once the
catheter is inserted into the bladder, the balloon is
inflated and gently pulled outward so that the electrode lies on the external urethral sphincter. This
electrode can record both free running and triggered
surface EMG activity from the sphincter. Care must
be taken to avoid slippage of the electrode during
the insertion of the catheter and pulling the catheter
too far down after it has been inflated as both
will result in the electrode in the distal urethra
(Krassioukov et al., 2004). Insertion of this electrode

requires special training and is often done by a urologist (Shinomiya et al., 1991; Krassioukov et al.,
2004).
Monitoring of the detrusor muscle function has been
advocated since it is supplied by the pelvic splanchnic
nerves, rather than the pudendal nerve, and is under
autonomic control. This is best done by monitoring
intravesical pressure of the urinary bladder. Before
initiating this type of monitoring the bladder capacity
must be determined with a preoperative cystometrogram. After induction of anesthesia, a Foley catheter
is inserted into the urinary bladder and attached to a
three-way flow adapter. This adapter is connected to
a flow transducer or manometer. The bladder is filled
to near capacity. Activation of the relevant sacral
nerve roots will cause contraction of the detrusor muscle, which in turn will increase the intravesical pressure of the urinary bladder. This will be transmitted
to the pressure transducer or manometer connected to
the Foley catheter (Ryken and Menezes, 1994). In this
manner, sacral nerve roots supplying the detrusor
muscle can be identified during surgery. A problem
encountered with this type of monitoring is that a
high-frequency stimulus over 1020 s is needed to
produce an increase in intravesical pressure. The pressure increase is also delayed for several seconds (Shinomiya et al., 1991). This increases the time needed
to provide the surgeon with feedback. The setup for
urinary bladder pressure monitoring is also cumbersome and time consuming.
696
49.6. Utility of NIOM in surgery for TCS

and other cauda equina lesions
As with many other types of procedures, outcomes of
patients undergoing surgery for TCS and other cauda
equina lesions with and without NIOM have not been
compared. To compound this problem further, surgical
outcomes data is often not readily available. Furthermore, variability in surgical technique and experience
and disease severity make comparisons between surgeons and centers difficult. Despite these concerns,
some studies provide useful insights.
One study of pediatric TCS surgery without the use
of NIOM reported improvement in pain, sensorimotor
function, and sphincter function in 100%, 43%, and
42% of 72 patients, respectively. Deterioration in the
same was noted in 0%, 9%, and 12% of the children
(Anderson, 1975). In an adult study of 21 patients not
using NIOM for TCS surgery, improvement in pain,
sensorimotor function, and sphincter function was
noted in 100%, 87%, and 38%, respectively. Worsening in these symptoms was seen in 0%, 0%, and 8%
(Quinones-Hinojosa et al., 2004). Similar findings
have been reported in various reviews (Lapsiwala
and Iskandar, 2004; George and Fagan, 2005). In one
report, a retrospective review of TCS surgery outcome
was conducted at one center with three neurosurgeons
performing the surgeries (Albright et al., 1999). Two
surgeons had no adverse neurological outcomes,
whereas the third had postoperative transient or permanent neurological deficits in 12% of patients. Interestingly, the first two surgeons used SEP NIOM in 7%
and 17% and EMG NIOM in 11% and 63% of cases,
whereas the third surgeon did not use SEP NIOM at
all and used EMG NIOM in 8%. Although this suggests that the use of NIOM is beneficial in reducing
morbidity, the study is confounded by a higher use of
the operating microscope by the first two surgeons
and a greater number of repeat untethering operations
for the third surgeon.
Surgical outcomes in series in which NIOM was
used have ranged from no neurological deterioration
to results similar to when no NIOM was used. In two
series reporting only EMG NIOM of lower limbs
and/or the external anal sphincter, no new neurological
deficits were observed (James et al., 1979; Von Koch
et al., 2002). In another study using SEP and EMG
NIOM only 6% of patients had new motor weakness
(Kothbauer et al., 1994). Several other studies note that
often surgical procedures were altered because of findings of NIOM (Legatt et al., 1992; Hormes and
A.M. HUSAIN
Chappuis, 1993; Phillips and Jane, 1996). NIOM prevented resection of tissue in over 50% of patients in
one series (Phillips and Jane, 1996). Although certainly not irrefutable, the data suggest that NIOM
may be helpful in surgery for TCS and cauda equina
lesions.
Although monitoring motor and sensory pathways
and external anal and external urethral sphincters and
detrusor muscles is possible, is it necessary? Monitoring of free running and stimulated EMG in lower
extremity muscles and the external anal sphincter
has been done consistently and appears to be useful.
It is argued that dissection in the region of the cauda
equina can result in trauma to nerves that would be
manifested as neurotonic discharges. Also, various
structures in the operative field can be stimulated to
determine the presence or the absence of neural
elements.
Tibial SEP monitoring is controversial. Several
studies have suggested that tibial SEP is important
as EMG NIOM does not protect the sensory nerve
roots (Shinomiya et al., 1991; Krassioukov et al.,
2004). Additionally, stimulation of the dorsal roots
can elicit a scalp recorded response, thus helping to
identify neural tissue (Kothbauer et al., 1994). Loss
of tibial SEP has also been noted in the lower spine
surgery after positioning, prompting rapid surgical
intervention (Hormes and Chappuis, 1993). Despite
these reports, there are compelling arguments against
using tibial SEP. Tibial SEP are widely used in monitoring spinal cord compromise, but in these cases a
gradual spinal cord dysfunction is detected by a
gradual deterioration of the responses. In cauda
equina surgeries, tibial SEP are lost when the nerve
is transected; a point at which it is too late for surgical intervention (McQuillan and Newberg, 1995).
Tibial SEP are also not sensitive in evaluating lumbosacral radiculopathies and often can provide false
localization (Aminoff et al., 1985). This most likely
occurs because tibial SEP are transmitted to the spinal cord through multiple lumbosacral segments;
compromise of any one segment as most commonly
occurs with dissection near the cauda equina will
not affect the tibial SEP (Von Koch et al., 2002).
Several authors have noted that tibial SEP do not
provide useful information in NIOM of surgeries
for TCS and cauda equine (Legatt et al., 1992; Von
Koch et al., 2002).
Whereas most published reports stress the importance of monitoring sphincter function, there is disagreement as to what should be monitored. Some
SPINE SURGERY
investigators suggest that since both the external anal

and external urethral sphincters are supplied by
branches of the pudendal nerve, monitoring one
(anal) is sufficient (James et al., 1979; Kothbauer
et al., 1994). Others have suggested that since two
different branches of the pudendal nerve (see above)
supply these sphincters, damage to one and not the
other is possible (Krassioukov et al., 2004). These
investigators note that in their series, spontaneous
activity was sometimes seen only in the external anal
sphincter and at other times only in the external urethral sphincter, implying potential for damaging one
while sparing the other sphincter. Consequently, they
suggest monitoring both external anal and external
urethral sphincters. The detrusor muscle is supplied
by the pelvic splanchnic nerves, and some have suggested monitoring its function in addition to the
sphincters (Shinomiya et al., 1991; Ryken and
Menezes, 1994).
A very low incidence of sphincter dysfunction has
been noted in almost all studies using any type of
sphincter monitoring. Interestingly, in a study in
which not only external anal and external urethral
sphincters, but also the detrusor muscle were monitored, 1 of 10 (10%) patients had worsening bladder
function (Shinomiya et al., 1991).
49.7. Proposed NIOM paradigm
The challenge for the neurophysiologist in TCS and
cauda equina surgeries is choosing monitoring which
is necessary and sufficient. Because various monitoring strategies are available does not mean they must
be used. The proposed paradigm is one used in the
authors practice.
NIOM is conducted with a 16 (sometimes 8)
channel machine. Two monopolar needle electrodes
are inserted into bilateral vastus lateralis, tibialis
anterior, medial gastrocnemius, gluteus maximus,
and semitendinosus muscles; a bipolar recording is
obtained from each muscle group. Additionally,
two to four needle electrodes are inserted into the
external anal sphincter. If four needles are used,
one is placed in each quadrant. Usually one channel
is used for external anal sphincter EMG monitoring.
If only eight channels are available, some of the
limb muscle channels are sacrificed in favor of the
external anal sphincter. The extra needle electrodes
in the external anal sphincter are for redundancy;
if one or two electrodes get dislodged, others can
be used instead.
697
Free-running and stimulated EMG is monitored

during the surgery. When the surgeon is dissecting
and only free-running EMG is being viewed, it is
best to set the time window of the NIOM machine
to 0.51.0 s full screen to better appreciate neurotonic discharges. However, during stimulation of
neural tissues, it is best to decrease the time window
to 100 ms full screen to separate the response from
the stimulus artifact. The surgeon is alerted whenever
neurotonic discharges are noted. The surgeon may be
able to hear them himself, if the EMG activity is
interfaced with a loudspeaker. When structures
within the surgical field are stimulated, the intensity
with which a response is elicited must be noted. Very
low intensity activated tissues are likely to be motor
nerve roots, whereas much greater intensity is needed
to stimulate the filum terminale.
49.8. Future direction
NIOM modalities continue to evolve. Studies that
assess adequacy of intervention in TCS by comparing
the spinal cord stimulation threshold before and after
untethering are under way. New stimulating probes,
which are easier to handle, deliver a more precise
stimulus, and can switch between monopolar and
bipolar modes are becoming available commercially
(Schekutiev and Schmid, 1996).
Although many neurophysiologists and surgeons
feel strongly that NIOM provides valuable information
during TCS and cauda equina surgery, studies comparing outcomes of these surgeries using and not using
NIOM have not been done. Furthermore, the extent
of monitoring should also be studied so that adequate
and necessary procedures are performed.
49.9. Conclusion
Complex surgeries in the cauda equina region and
surgeries for TCS have the potential to damage
peripheral nerves supplying the lower extremities
and interfere with bowel and bladder continence.
Though unequivocal evidence is lacking, many neurophysiologists and surgeons feel that NIOM can be
used to lessen the potential of damage to neural
structures during these surgeries. Many types of
NIOM have been reported for monitoring surgery
for TCS and cauda equina lesions. However, freerunning and stimulated EMG monitoring of muscles
of the lower extremities and external anal sphincter
appears to be adequate for most patients.
698
References
Albright, AL, Pollack, IF, Adelson, PD, et al. (1999) Outcome data and analysis in pediatric neurosurgery. Neurosurgery, 45: 101106.
Aminoff, MJ, Goodin, DS, Barbaro, NM, et al. (1985)
Dermatomal somatosensory evoked potentials in unilateral lumbosacral radiculopathy. Ann. Neurol., 17:
171176.
Anderson, FM (1975) Occult spinal dysraphism: a series of
73 cases. Pediatrics, 55: 826835.
Beek, FJ, De Vries, LS, Gerards, LJ, et al. (1996) Sonographic determination of the position of the conus
medullaris in premature and term infants. Neuroradiology, 38(Suppl. 1): S174S177.
Deletis, V, Vodusek, DB, Abbott, R, et al. (1992) Intraoperative monitoring of the dorsal sacral roots: minimizing the risk of iatrogenic micturition disorders.
Emerson, RG and Adams, DC (2003) Intraoperative monitoring. In: JS Ebersole and TA Pedley (Eds.), Current Practice of Clinical Electroencephalography. Lippincott,
Williams & Wilkins, Philadelphia, 3rd ed., pp. 936954.
Fitz, CR and Harwood-Nash, DC (1975) The tethered
conus. Am. J. Roentgenol. Radium Ther. Nucl. Med.,
125: 515523.
George, TM and Fagan, LH (2005) Adult tethered cord
syndrome in patients with postrepair myelomeningocele: an evidence-based outcome study. J. Neurosurg.,
102: 150156.
Gilman, S and Newman, SW (2002) Manter and Gatzs
Clinical Neuroanatomy and Neurophysiology. F. A.
Davis, Philadelphia, 10th ed.
Hellbusch, LC and Nihsen, BJ (1989) Rectal sphincter pressure monitoring device. Neurosurgery, 24: 775776.
Hormes, JT and Chappuis, JL (1993) Monitoring of lumbosacral nerve roots during spinal instrumentation. Spine,
18: 20592062.
Hughes, JA, De Bruyn, R, Patel, K, et al. (2003) Evaluation
of spinal ultrasound in spinal dysraphism. Clin. Radiol.,
58: 227233.
Ikeda, K, Kubota, T, Kashihara, K, et al. (1986) Anorectal
pressure monitoring during surgery on sacral lipomeningocele. Case report. J. Neurosurg., 64: 155156.
James, HE, Mulcahy, JJ, Walsh, JW, et al. (1979) Use of anal
sphincter electromyography during operations on the
conus medullaris and sacral nerve roots. Neurosurgery,
4: 521523.
Kothbauer, KF and Novak, K (2004) Intraoperative monitoring for tethered cord surgery: an update. Neurosurg.
Focus, 16: E8.
Kothbauer, K, Schmid, UD, Seiler, RW, et al. (1994)
Intraoperative motor and sensory monitoring of the
cauda equina. Neurosurgery, 34: 702707; discussion 7.
A.M. HUSAIN
Krassioukov, AV, Sarjeant, R, Arkia, H, et al. (2004) Multimodality intraoperative monitoring during complex lumbosacral procedures: indications, techniques, and longterm follow-up review of 61 consecutive cases. J. Neurosurg. Spine, 1: 243253.
Lapsiwala, SB and Iskandar, BJ (2004) The tethered cord
syndrome in adults with spina bifida occulta. Neurol.
Res., 26: 735740.
Legatt, AD, Schroeder, CE, Gill, B, et al. (1992) Electrical
stimulation and multichannel EMG recording for identification of functional neural tissue during cauda equina
surgery. Childs Nerv. Syst., 8: 185189.
McLone, DG (1999) Occult dysraphism and the tethered
spinal cord. In: M Choux, C Di Rocco, AD Hockley,
et al. (Eds.), Pediatric Neurosurgery.Churchill Livingstone, London, pp. 6177.
McQuillan, PM and Newberg, N (1995) Intraoperative
electromyography. In: GB Russell and LD Rodichok
(Eds.), Primer of Intraoperative Neurophysiologic Monitoring. Butterworth-Heinemann, Boston, pp. 171187.
Michelson, DJ and Ashwal, S (2004) Tethered cord syndrome in childhood: diagnostic features and relationship
to congenital anomalies. Neurol. Res., 26: 745753.
Muraszko, KM (1999) Spinal dysraphism in the adult and
pediatric populations. In: RG Grossman and CM Loftus
(Eds.), Principles of Neurosurgery. Lippincott-Raven
Publishers, Philadelphia, 2nd ed., pp. 5975.
Nishijima, Y, Okada, M, Yasuaki, Y, et al. (1992) Intraoperative monitoring for thoracolumbar or lumbar surgery with somatosensory evoked potentials after
double stimuli. Spine, 17: 13041308.
Pang, D and Casey, K (1983) Use of an anal sphincter pressure monitor during operations on the sacral spinal cord
and nerve roots. Neurosurgery, 13: 562568.
Pang, D and Wilberger, JE, Jr. (1982) Tethered cord syndrome in adults. J. Neurosurg., 57: 3247.
Phillips, LH, 2nd and Jane, JA (1996) Electrophysiologic
monitoring during tethered spinal cord release. Clin.
Neurosurg., 43: 163174.
Quinones-Hinojosa, A, Gadkary, CA, Gulati, M, et al.
(2004) Neurophysiological monitoring for safe surgical
tethered cord syndrome release in adults. Surg. Neurol,
62: 127133; discussion 3335.
Roos, RA, Vielvoye, GJ, Voormolen, JH, et al. (1986)
Magnetic resonance imaging in occult spinal dysraphism. Pediatr. Radiol., 16: 412416.
Ryken, TC and Menezes, AH (1994) Intraoperative electrical and manometric monitoring in lumbosacral surgery.
In: CM Loftus and VC Traynelis (Eds.), Intraoperative
Monitoring Techniques in Neurosurgery. McGraw-Hill,
Inc., New York, pp. 257268.
Sathi, S, Madsen, JR, Bauer, S, et al. (1993) Effect of
surgical repair on the neurologic function in infants with
lipomeningocele. Pediatr. Neurosurg., 19: 256259.
SPINE SURGERY
Schekutiev, G and Schmid, UD (1996) Coaxial insulated
bipolar electrode for monopolar and bipolar mapping
of neural tissue: technical note with emphasis on the
principles of intra-operative stimulation. Acta Neurochir. (Wien), 138: 470474.
Shinomiya, K, Fuchioka, M, Matsuoka, T, et al. (1991)
Intraoperative monitoring for tethered spinal cord syndrome. Spine, 16: 12901294.
Sutton, LN (2005) Spinal dysraphism. In: SS Rengachary
and RG Ellenbogen (Eds.), Principles of Neurosurgery.
Elsevier Mosby, Edinburgh, 2nd ed., pp. 99115.
Von Koch, CS, Quinones-Hinojosa, A, Gulati, M, et al.
(2002) Clinical outcome in children undergoing
tethered cord release utilizing intraoperative neurophysiological monitoring. Pediatr. Neurosurg., 37:
8186.
699
Warder, DE and Oakes, WJ (1994) Tethered cord syndrome: the low-lying and normally positioned conus.
Neurosurgery, 34: 597600; discussion.
Wilson, DA and Prince, JR (1989) John Caffey award. MR
imaging determination of the location of the normal
conus medullaris throughout childhood. AJR Am. J.
Roentgenol., 152: 10291032.
Yamada, S, Zinke, DE and Sanders, D (1981) Pathophysiology
of tethered cord syndrome. J. Neurosurg., 54: 494503.
Yamada, S, Won, DJ and Yamada, SM (2004) Pathophysiology of tethered cord syndrome: correlation with
symptomatology. Neurosurg. Focus, 16: E6.
Zerah, M, Pierre-Kahn, A and Catala, M (1999) Lumbosacral lipomas. In: M Choux, C Di Rocco, AD Hockley,
et al. (Eds.), Pediatric Neurosurgery. Churchill Livingstone, London, pp. 79100.

M.R. Nuwer (Ed.)
700
CHAPTER 50
Dorsal root entry zone procedures and other

surgeries for pain
Aatif M. Husain*
Department of Medicine (Neurology), Duke University Medical Center and Neurodiagnostic Center, Veterans Affairs Medical Center,
Durham, NC 27710, USA
50.1. Introduction
Pain is a common symptom that is most often treated
medically. However, when pain becomes chronic and
medical therapy fails, surgical treatment is considered.
Surgical treatment of pain has targeted all levels of the
nervous system, from the peripheral nerves to the cortex. In this chapter the use of neurophysiological
intraoperative monitoring (NIOM) will be discussed
in various types of pain surgery with an emphasis on
the dorsal root entry zone (DREZ) procedure. This
discussion will not include surgeries in which pain
relief is not the sole primary goal, that is spinal stenosis, disk herniation, etc. Additionally, microvascular
decompression surgery for various pain syndromes
will not be discussed here as it is covered elsewhere.
This chapter will start with an introduction to pain
and pain pathways in the nervous system. A summary of the various types of pain surgeries will then
be presented. This will be followed by a detailed discussion of NIOM, as it relates to the pain surgeries.
It should be noted that data on NIOM for pain surgery is sparse, and most reports on a given procedure
are from a few centers.
50.2. Anatomy of pain
Chronic pain is that which persists beyond 3 months.
Broadly, chronic pain can be divided into nociceptive
and neuropathic types. Nociceptive pain is due to an
external stimulus that typically causes tissue damage,
inflammation, and the release of pain-producing
substances such as prostaglandins and bradykinin
*
Correspondence to: Aatif M. Husain, M.D., Department of

Medicine (Neurology), Duke University Medical Center,
Box 3678, 202 Bell Building, Durham, NC 27710, USA.
Tel.: 1-(919)-684-8485; fax: 1-(919)-684-8955.
E-mail: aatif.husain@duke.edu (A.M. Husain)
(Romanelli and Esposito, 2004). Tissue damage

results in stimulation of nociceptive receptors and
pain perception. A common example of nociceptive
pain is tumor invasion or compression of tissues.
Neuropathic or nerve-mediated pain, on the other
hand, is due to an intrinsic dysfunction in the nervous
system at the level of individual nerves, spinal cord,
or brain. There is no evident ongoing tissue injury.
Dysesthesia, paresthesia, and allodynia are terms
used to describe neuropathic pain.
A brief discussion of pain pathways is presented
here as an introduction to surgical therapy for pain.
This discussion will aim to help the reader appreciate
why certain parts of the nervous system are targeted
in pain surgery. For more comprehensive reviews
on this subject, excellent recent reviews should
be consulted (Ovelman-Levitt, 1996; Willis and
Westlund, 1997; Romanelli and Esposito, 2004).
The pain pathway starts with pain receptors, or
nociceptors. These receptors are terminal nerve endings of Ad and C nerve fibers. Ad-fiber nociceptors
mediate sharp, pricking, and aching pain, whereas
C-fiber nocicpetors transmit a burning type of pain.
Cell bodies for these fibers lie in the dorsal root
ganglia. Most of the pain-transmitting nerve fibers
traverse the dorsal root to enter the dorsal horn of
the spinal cord gray matter. Some fibers, however,
travel in the ventral root to reach the dorsal horn.
Fibers entering the dorsal horn divide into two
branches, with one ascending and one descending,
for one to two segments of the spinal cord. This
network of ascending and descending nerve fibers
spans the length of the spinal cord, and is known as
Lissauers tract. The dorsal horn is divided into
laminae, with a distinct function for each lamina.
Laminae I, II, III, and V are the main sites of termination of nociceptive fibers. Ad-fibers synapse in laminae I and V, whereas C-fibers synapse in lamina II
SPINE SURGERY
(also known as the substantia gelatinosa). Visceral

nociceptors terminate in lamina III.
Most second-order neurons carrying pain impulses
project from the dorsal horn across the anterior white
commissure to ascend as the lateral spinothalamic
tract in the anterolateral aspect of the spinal cord.
Some fibers may not decussate and ascend in the
ventral spinothalamic tract. The lateral spinothalamic
tract has a topographic representation with fibers from
the lower extremities in the most postrolateral position, whereas fibers from higher (i.e., cervical) levels
occupying a more anteromedial position. In the
medulla, the lateral spinothalmic tract lies adjacent
to the ventral spinothalamic and spinotectal tracts,
together forming the spinal lemniscus. The spinal lemniscus lies between the spinal nucleus of the trigeminal nerve and the inferior olivary nucleus in the
medulla. It ascends through the posterior part of the
pons and the tegmentum of midbrain to the thalamus.
The spinothalamic pathway relays on third-order
neurons in several thalamic nuclei. Most fibers
relay in the ventral posterolateral nucleus. From this
nucleus, fibers project to the primary somatosensory
cortex and the insular cortex. The spinothalamic tract
also synapses on the ventral posterior inferior and
dorsomedial nuclei. From here, fibers relay to the
secondary somatosensory cortex and the cingulate
gyrus, respectively. Whereas the primary somatosensory cortex is responsible for pain perception, the
secondary somatosensory cortex, cigulate gyrus, and
insular cortex modulate pain perception and response
to painful stimuli.
In addition to the spinothalamic pathway, there
are other ascending pathways that participate in pain
transmission. The spinoreticular pathway is involved
with the affective, somatic, and autonomic response
to pain. It arises mostly from cells in laminae VII
and VIII and ascends mostly contralaterally, adjacent
to the lateral spinothalamic tract. This pathway
synapses on neurons in the medullary, pontine, and
mesencephalic reticular formation. The spinomesencephalic pathway arises mainly from cells in lamina I.
It ascends contralaterally along with the spinoreticular and lateral spinothalamic tracts and synapses on
neurons in the rostral mesencephalic reticular
formation. These two pathways are responsible for
mediating burning type of pain, whereas the spinothalamic pathway mediates the initial sharp,
pricking type of pain.
Two other pain pathways ascend in the dorsal
columns. One arises from laminae III and IV, and
701
has the same somatotopic distribution as the dorsal

column pathway. It synapses on the ventral posterolateral nucleus of the thalamus, and is thought to
have a modulatory role in pain perception. The other
pain pathway in the dorsal column arises from lamina
X and is located medially. It projects to the brainstem
reticular formation and then on to medial and intralaminar thalamic nuclei.
Facial sensation and pain are transmitted by the
ophthalmic, maxillary, and mandibular branches of
the trigeminal nerve. The cell bodies of these nerves
lie in the trigeminal or Gasserian ganglion. The Gasserian ganglion has a somatotopic organization, with
ophthalmic representation most posteriorly, and mandibular representation most anteriorly. Axons from
the Gasserian ganglion synapse on various brainstem
trigeminal nuclei, with the pain-transmitting fibers
synapsing mostly on neurons in the spinal nucleus.
The spinal nucleus has three parts: nuclei oralis,
interpolaris, and caudalis. The nucleus caudalis is
the inferior most part of the spinal nucleus, extending
from the level of the obex to the C2 root. It is this
part of the spinal nucleus that receives the facial pain
fibers, and consequently, is similar to laminae I and
II, of the dorsal horns. In fact, it is continuous with
those laminae in the cervical spinal cord. The nucleus
caudalis is also somatotopically organized with fibers
from the lips terminating in the rostral most part of
the nucleus and fibers from more peripheral parts of
the face synapsing at the caudal end. Most secondorder neurons from the nucleus caudalis ascend with
the contralateral spinothalamic tract to relay on the
ventral postromedial nucleus of the thalamus. Some
projections from the nucleus caudalis also go to the
posterior thalamus and the internal medullary lamina.
Third-order neurons project to the cortex in a manner
similar to the spinothalamic tract.
50.3. Classification of surgeries for pain
Surgical procedures aimed at alleviating pain can be
either ablative or augmentative. As the name implies,
ablative procedures involve selective destruction of a
part of the pain pathway. These are irreversible procedures. Augmentative procedures alter pain modulation and perception, and usually involve stimulation
of various parts of the pain pathway. Their reversibility makes augmentative procedures preferred over
ablative ones (Table 1). Below is a brief discussion
of the various types of ablative and augmentative
procedures that can be performed for pain relief.
702
A.M. HUSAIN
Table 1
Classification of various types of surgeries for pain
Ablative surgeries
Augmentative surgeries
Neurectomy
Nerve decompression
Dorsal root ganglionectomy
Rhizotomy
Sympathectomy
Dorsal root entry zone procedure
Cordotomy
Midline myelotomy
Commissural myelotomy
Mesencephalotomy
Hypophysectomy
Medial thalamotomy
Cingulotomy
Peripheral nerve stimulation

Spinal cord stimulation
Intraspinal analgesia
Deep brain stimulation
Motor cortex stimulation
For details on these procedures, standard neurosurgical texts may serve for further explanation.
50.3.1. Ablative surgeries
Neurectomy involves cutting the peripheral nerve subserving the area of pain. This can only be done with
pure sensory nerves, because cutting a mixed nerve
would result in undesired motor deficits. Since the
dorsal root ganglion and the cell bodies of the cut neurons survive in this procedure, there is regenerative
sprouting of the nerve endings. In some circumstances,
this can lead to recurrence of neuropathic pain.
Because of these limitations, neurectomy is reserved
for patients with pain related to malignancies. The limited life expectancy of these patients makes potential
long-term side effects less concerning.
Decompression of peripheral nerves is also done
to relieve pain. This is applicable in situations in
which the nerve is entrapped by thickened ligaments
or bone spurs. A common example of this type of
surgery is carpal tunnel release for median nerve
compression. This type of surgery is effective in
relieving nociceptive type of pain; however, if there
is an additional neuropathic pain component, it may
not resolve with this surgery (Brown, 2005).
Dorsal root ganglionectomy is the ablation, or
removal, of the dorsal root ganglion and the distal
parts of the sensory rootlets. This procedure is performed extradurally and pain-transmitting fibers traversing not only the dorsal root, but also the ventral
root are transected. Ganglia of cranial nerves, such
as the trigeminal and glossopharyngeal, can also be

targeted. It is particularly effective for occipital
neuralgia, chest wall pain, cervicofacial tumors, and
perineal pain. Mid-cervical and lumbar-level dorsal
root ganglionectomy is avoided, because it may
cause unacceptable proprioception difficulties.
Rhizotomy of the dorsal roots involves lesioning
the dorsal rootlets to interrupt pain pathways. Usually, some rootlets at each level are spared to preserve proprioception. Since this procedure involves
opening the dura and does not address pain fibers traversing the ventral root, it is done with less frequency
than dorsal root ganglionectomy (Brown, 2005).
Sympathectomy is a rare type of pain surgery in
which various sympathetic ganglia are excised or
ablated. Most often, ganglia at T1T4 are lesioned
for complex regional pain syndrome or other types
of pain syndromes arising from the viscera. This
procedure is often performed endoscopically. Sympathectomy is only effective in relieving sympathetically mediated pain.
The DREZ procedure is one of the most common
ablative pain procedures currently being performed.
As the name implies, the DREZ is ablated, either
with a radiofrequency (RF) electrode or with an incision. This procedure is effective for various types
of neuropathic pain, including phantom limb pain
(Nashold, 1996). It can also be performed for facial
pain by lesioning the nucleus caudalis. Potential
complications involve damage to adjacent tracts in
the spinal cord, that is, the corticospinal and dorsal
spinocerebellar tracts.
SPINE SURGERY
Cordotomy is a procedure in which the lateral spinothalamic and spinoreticular tracts, which lie in the
anterolateral aspect of the spinal cord, are lesioned.
Previously, this was done through an open approach;
but more recently, a percutaneous cervical approach
by using RF thermal lesions has come into use
(Kanpolat, 2004). Cordotomy is particularly useful
for unilateral, regional pain syndromes. Bilateral
cordotomies, especially in the cervical region, can
lead to respiratory depression. The ventral spinocerebellar tract lies postrolateral, and the lateral pyramidal tract lies posterior, to the lateral spinothalamic
tract, and can be damaged during lesioning, leading
to ataxia and weakness.
Midline myelotomy involves creating punctuate
midline lesions with fine forceps 1 mm to each side
of the posterior midline. The part of both dorsal columns between the blades of the forceps is crushed
(Stanton-Hicks and Salamon, 1997; Willis and
Westlund, 1997; Nauta et al., 2000). This results in
lesioning of the pain fibers located in the medial
portion of the dorsal column pathways. Visceral,
pelvic, and abdominal pain are often treated with this
procedure (Romanelli et al., 2004).
A commissural myelotomy is a procedure in
which the entire thickness of the spinal cord is longitudinally sectioned. This results in interruption of the
anterior spinal commissure and decussating spinothalamic pain pathways.
Mesencephalotomy is a procedure performed stereotactically, in which the lateral spinothalamic tract
in the dorsal midbrain tegmentum is lesioned. An RF
current is used to make the lesion under magnetic
resonance imaging (MRI) guidance. This procedure
is mostly used for unilateral upper chest and face
pain. Ocular motility disorders can result from this
procedure.
Hypophysectomy is the removal of the pituitary
gland, which can be accomplished through an open
surgical procedure or by stereotactic lesioning. The
mechanism by which this procedure leads to pain
relief is unclear (Ramirez and Levin, 1984). It is useful in patients with bone pain that has not responded
to hormonal or medical management. Hormone
replacement is required after this surgery.
Medial thalamotomy is also a stereotactic procedure in which lesions are made in the centromedian,
parafascicular, centrolateral, and dorsomedial nuclei.
These nuclei are involved in pain processing rather
than pain transmission (Romanelli et al., 2004).
Lesioning of the lateral thalamus where the pain
703
pathways (spinothalamic tract) relay produces unacceptable contralateral sensory loss. This procedure
is used for patients with nociceptive allodynic, shooting, or hyperpathic types of neuropathic pain in the
head, face, and neck region (Tasker, 1990).
Cingulotomy involves lesioning the anterior cingulate gyrus through a stereotactically guided procedure. Like medial thalamotomy, this results in
alteration in pain perception, because the cingulate
gyrus is involved in pain processing, not in pain
transmission. Bilateral cingulotomies can be performed for cancer and non-cancer-related bilateral
pain (Wilkinson et al., 1999). Emotional problems
are uncommon, though attention and spontaneity
may be reduced (Cohen et al., 2001).
50.3.2. Augmentative surgeries
Peripheral nerve stimulation (PNS) involves placing a
stimulating electrode on a peripheral nerve, with the
battery implanted nearby. This is usually done with an
open procedure and requires visualization of the nerve.
PNS is used when pain is localized to the distribution
of a peripheral nerve (Follett, 2004; North, 2004).
Spinal cord stimulation (SCS) is performed with
an epidural electrode which is inserted after a small
laminectomy or percutaneously. The electrode is
connected to a battery. SCS is particularly useful
for failed back syndrome and complex regional pain
syndrome (Burchiel et al., 1996; Kumar et al.,
1997). SCS has also been used effectively for chronic
angina pectoris (De Jongste et al., 1994).
Intraspinal analgesic administration is accomplished with a small catheter inserted, usually percutaneously, into the intrathecal space and connected
to a refillable electronic pump. The analgesic is
released in programmable doses and intervals from
the pump. Opioids such as morphine are commonly
used. This procedure is used most often for nociceptive, and sometimes for neuropathic, pain. Analgesics
can also be delivered intraventricularly for head and
neck pain (Follett, 2004).
Deep brain stimulation is performed with stereotactically-placed electrodes in various brain regions.
Stimulation of the periaqueductal gray and periventricular gray regions is thought to release endogenous
opioids, and is useful for nociceptive pain (Hosobuchi
et al., 1977). Ventral posterolateral and ventral posteromedial nuclei stimulation produces pain relief in
the contralateral body and face, respectively (Kaplitt
et al., 2004).
704
Motor cortex stimulation involves implanting epidural electrodes over the area of the motor cortex
corresponding to the region of pain. The exact mechanism by which motor cortex stimulation alleviates
pain is unknown (Hanajima et al., 2002). One possible
mechanism is that low-intensity motor cortex stimulation inhibits thalamic pain pathways (Brown, 2005).
This procedure has been used for stroke pain, postherpetic neuralgia, and brachial plexus avulsion pain.
50.4. NIOM for pain surgeries
As noted above data for NIOM in various types of
pain surgery, with the exception of the DREZ procedure, is sparse. Many procedures are stereotactically
guided and/or done on awake patients, decreasing
the role of NIOM. When NIOM is used for pain surgery, its purpose is twofold. As with other types of
NIOM, it is used to prevent damage to neural tissue
or tracts adjacent to the one being lesioned or
ablated. Additionally, NIOM can be used for functional localization of the tracts to be lesioned, thus
better targeting therapy. Of the surgeries for pain discussed above, only those in which NIOM has been
used will be described further. Additionally, selective
dorsal rhizotomy, thalmotomy, and deep brain stimulation, all procedures in which NIOM is commonly
used, will not be discussed here as they have been
covered in detail elsewhere in this text.
50.4.1. Trigeminal rhizotomy
Rhizotomy of dorsal spinal roots (selective dorsal
rhizotomy) is often performed for relief of spasticity
and is discussed elsewhere in this text. However, rhizotomy of the preganglionic trigeminal rootlets is a
procedure done to relieve trigeminal neuralgia or
other facial pain syndromes, in which NIOM can be
used. Consequently, it is discussed here.
Trigeminal rhizotomy is a percutaneous procedure in
which partial destruction of trigeminal preganglionic
rootlets leads to pain relief, with sparing of facial sensation. Since this procedure does not involve a craniotomy,
it is preferred in elderly patients or those who cannot tolerate the more invasive microvascular decompression
surgery. Pain relief often lasts for a few years, and the
procedure may need to be repeated. Consequently, for
younger patients, microvascular decompression surgery
may be preferred (Liu and Apfelbaum, 2004).
To perform a trigeminal rhizotomy, the patient is
awake, but given sedatives and neuroleptic analgesia.
A.M. HUSAIN
A cannula is introduced lateral to the oral commissure,

and advanced to the Gasserian ganglion, through the
foramen ovale, under fluoroscopic guidance. Once adequate positioning of the cannula in the trigeminal cistern is confirmed, the lesioning (RF) electrode is
advanced in the cannula. Lesions are made by heating
the electrode. Nociceptive (Ad- and C-) fibers are coagulated at lower temperature than thicker fibers do,
which subserve tactile sensation. Consequently, the
temperature of the lesioning electrode must be carefully monitored (Sweet and Wepsic, 1974; Sweet,
1976). Another method of performing this procedure
is with glycerol injection. Once the cannula has been
inserted, and its position verified, the volume of the trigeminal cistern is determined radiographically. Thereafter, an equal quantity of glycerol is injected into the
trigeminal cistern through the cannula (Lunsford and
Bennett, 1984). Yet another technique involves balloon
compression of the Gasserian ganglion after a larger
balloon catheter is introduced in a manner similar to
that described above. As the balloon is inflated, the
ganglion is destroyed (Mullan and Lichtor, 1983). This
can lead to bradyarrhythmia, and consequently, is done
under general anesthesia with external cardiac pacing
(Liu and Apfelbaum, 2004).
Monitoring the extent of the lesion can be done with
the RF thermocoagulation technique described above.
Several investigators have used clinical methods
for monitoring the location of the lesioning electrode (Sanders and Henny, 1992; Lee et al., 1997; Spendel et al., 1997). Once the cannula is in place, a
temperature-sensitive RF lesioning electrode is introduced. If this electrode is connected to a stimulator, it
can be used as a stimulating electrode. Another needle
is placed in the wound site, and acts as a reference
electrode. With the patient awake, a fast frequency
(5075 Hz) electrical stimulus at 0.11.0 V (0.05 ms
duration) intensity is delivered for 1 ms. The patient is
asked about paresthesias. The lesioning electrode can
be assumed to be adjacent to fibers subserving the part
of the face experiencing the paresthesia (i.e., V1, V2,
or V3). Similarly, with a slower frequency stimulus
(5 Hz) delivered at 0.72.0 V (0.05 ms duration) intensity, contraction of the masseter muscle is noted. The
RF lesioning electrode is positioned so that paresthesia
is noted in the painful area of the face, and motor activation is not seen. Thereafter, the electrode is connected to
the lesion generator. As the lesion is being made, the
presence or absence of paresthesia can be monitored.
Because percutaneous trigeminal rhizotomy with
RF thermocoagulation can be painful, intermittent
SPINE SURGERY
sedation, and anesthesia are used. This often limits the

ability of patients, many of whom are elderly, to report
paresthesias when the trigeminal nerve branches are
stimulated in the above protocol. Additionally, this protocol relies completely on the subjective assessment of
patients, which sometimes may not be reliable.
Because of these limitations, several authors have proposed NIOM protocols that can be used with the patient
either awake or asleep, and provide objective neurophysiological verification of trigeminal anatomy (Salar
et al., 1982; Macon and Poletti, 1987; Karol et al.,
1991; McGlone and Wells, 1991; Leandri and Gottlieb,
1996). Although all the protocols describe slightly different techniques, the basic principle is the same.
Branches of the trigeminal nerve are stimulated and
evoked potentials are recorded from either the scalp
or the vicinity of the Gasserian ganglion. The technique
is best described by Leandri and Gottlieb (1996), and is
further discussed in their literature.
Stimulating electrodes are 25 mm Teflon-coated
needles whose tips are bare. Two electrodes are inserted
close together into the supraorbital notch (Leandri et al.,
1989). These serve to stimulate the supraorbital nerve
(V1 distribution). Similar electrodes are also inserted
along the nasolabial fold to enter the infraorbital foramen to stimulate the infraorbital nerve (Leandri et al.,
1988). The mental nerve is stimulated by inserting the
needle electrodes along the lateral surface of the chin
into the mental foramen (Leandri et al., 1990). These
electrode pairs are connected to a stimulator and
sequentially stimulated at 5 Hz with 0.05 ms duration
and 35 mA intensity stimulus. The painful area of the
face can also be stimulated with a hand-held bipolar
stimulator. This stimulator must be time-locked to the
signal averager. The evoked potentials generated by this
latter type of stimulation are smaller, and recorded only
with the invasive recording electrode described below.
Recording electrodes are placed on the scalp at
the vertex and C7 spine for a one channel recording
that will display far-field trigeminal evoked potentials. Filter settings are the same as those used for
other somatosensory evoked potential (SEP) studies.
Approximately 100 responses are averaged to obtain
a reproducible response.
Trigeminal evoked potentials can also be recorded
with a specially designed thermorhizotomy cannula
which is Teflon-coated except for the 5 mm at the tip.
When recordings are to be made, the thermocouple
is replaced with a stylus which is also Teflon-coated
except for 1 mm at the tip. The stylus and the cannula make up a bipolar recording electrode. This
705
electrode can be advanced beyond the foramen ovale

to the clivus and further, for recording trigeminal
evoked potentials from different locations. Only
1020 responses need to be averaged with this electrode to obtain a reproducible response.
Scalp recorded trigeminal evoked potentials show a
series of waves with the first three, being most important. These three are labeled SW13, W13, and
MW13, depending on whether they were obtained after
stimulation of the supraorbital, infraorbital, or mental
nerves, respectively (Fig. 1). The latencies of these
responses range from 0.9 ms (for W1) to 3.5 ms (for
MW3). The first of these responses is thought to originate from distal part of the nerve stimulated, the second
from the preganglionic root, and the third between the
entry to the pons and the sensory nuclei (Leandri,
1998). Responses recorded from the invasive bipolar
electrode described above are usually either biphasic
or triphasic. When the response is of greatest amplitude, the electrode is nearest to the preganglionic
fibers of the nerve stimulated. The highest amplitude
response from stimulation of the supraorbital nerve is
usually obtained when the electrode is 35 mm past
the clivus. After stimulation of the infraorbital nerve,
the highest amplitude response is noted in the vicinity
of the clivus. Mental nerve stimulation results in the
highest amplitude response 29 mm before the clivus
(Karol et al., 1991) (Fig. 2).
The motor part of the trigeminal nerve can be monitored as well. Stimulation of the trigeminal nerve is
done with the invasive trigeminal electrode described
above. Alternatively, a monopolar stimulator can be
used with the cannula serving as the cathode, and a
separate electrode placed on the skin serving as the
anode. The stimulus pulse duration is 0.1 ms, intensity
is 0.11.5 mA, and it is delivered at a frequency of
3 Hz. The motor response is recorded with surface
cup electrodes placed on the ipsilateral masseter muscle and a nearby bony prominence. Alternatively, two
needle electrodes inserted into the masseter muscle
can also be used as recording electrodes (Fig. 3).
Before testing motor function, neuromuscular blocking agents must be discontinued.
NIOM and thermocoagulation is performed in five
steps (Leandri and Gottlieb, 1996). In the first step,
the baseline scalp trigeminal evoked potentials are
obtained after stimulation of the supraorbital, infraorbital, and mental nerves. In the second step, the invasive
electrode described above is positioned in the foramen
ovale. The three nerves are stimulated again, and
responses are recorded from the invasive electrode.
706
A.M. HUSAIN
Supraorbital
nerve
0.3 V/div
SW1
W3
W2
1 V/div
Infraorbital
nerve
Mental
nerve
0.5 V/div
W1
1 ms/div
MW1
Fig. 1. Supraorbital (SW13), infraorbital (W13), and mental (MW13) nerve evoked potentials recorded from the scalp
(derivation CzC7S). Reproduced from Leandri and Gottlieb (1996) with permission of the American Association of Neurological Surgeons.
5 V/div
Supraorbital
nerve
1.55 ms
5 V/div
Infraorbital
nerve
Mental
nerve
5 V/div
1.05 ms
2.30 ms
1 ms/div
Fig. 2. Supraorbital, infraorbital, and mental nerve evoked potentials recorded from an invasive bipolar electrode inserted
through the foramen ovale in the vicinity of the Gasserian ganglion (near the clivus). With the current placement the electrode
is closest to the fibers of the mental nerve. Further movement of the electrode would change the morphology of these responses.
Reproduced from Leandri and Gottlieb (1996) with permission of the American Association of Neurological Surgeons.
SPINE SURGERY
707
Fig. 3. Trigeminal motor evoked potential recorded from the masseter muscle. Two needle electrodes were used for recording the muscle evoked potential. Legend: 2 ms/div and 5 mV/div.
The electrode is slowly advanced to the clivus and

beyond to note the location of maximum amplitude of
the evoked potentials from stimulation of each of the
nerves. In the third step, the invasive electrode is positioned in the location in which the response from the
painful area of the face was of highest amplitude. Then,
that area of the face is stimulated with a hand-held
stimulator and evoked potentials again recorded from
the invasive electrode to confirm correct positioning.
The fourth step involves stimulating through the invasive electrode to make sure a motor response is not elicited, so as to avoid damage of motor fibers of the
trigeminal nerve. Assuming that motor activation is
not seen, a lesion is made. The final step is repeating
the scalp trigeminal evoked potentials after stimulation of all three nerves. After more than two lesions,
the W2 response starts to decrease in amplitude and
prolong in latency. A decrease of 2050% of baseline
is considered satisfactory to induce pain relief
(Drechsler and Neuhauser, 1986; Leandri and Gottlieb, 1996). The higher the temperature while making
the lesion, the more significant the degree of change of
the trigeminal evoked potential and subsequent sensory loss (Salar et al., 1982). Similar findings have
been reported in trigeminal rhizotomies performed

with glycerol and balloon compression (Bennett and
Lunsford, 1984; Landi et al., 1991).
Although these techniques can be performed
quickly, considerable expertise is required in
correct placement of the stimulating electrodes. Additionally, if one or more of the nerves being stimulated
has been damaged, either by disease (herpes zoster) or
by prior therapeutic interventions, evoked potentials
may be not be obtainable, even after stimulation with
high intensities. Anesthetic concerns are the same as
those for other types of SEP and motor evoked potential monitoring. If general anesthesia is used, high
doses of halogenated compounds can reduce the
amplitude of the evoked potentials. The use of neuromuscular blocking agents must be stopped if motor
stimulation is to be performed. Despite these concerns, trigeminal NIOM can provide valuable assistance in accurately targeting lesions.
50.4.2. DREZ procedure
The DREZ procedure is an ablative pain surgery in
which the DREZ of the spinal cord is lesioned.
708
Predominantly, laminae I and II of the dorsal horn

are targeted, because it is here that pain transmitting
Ad- and C-fibers synapse (Ovelman-Levitt, 1996).
This procedure was introduced for treatment of pain
associated with brachial plexus avulsions, yet it has
been found to be useful for lumbosacral root avulsions, paraplegia, phantom pain, and other types of
neuropathic pain syndromes (Nashold et al., 1976;
Nashold, 1984). Many types of facial pain, including
trigeminal neuralgia, post-herpetic pain, posttraumatic pain, multiple sclerosis, atypical facial pain,
and cancer-related facial pain, have been successfully
treated with DREZ lesions made in the nucleus caudalis (Rossitch et al., 1989; Delgado-Lopez et al.,
2003; Bullard and Nashold, 2004). Although the
nucleus caudalis DREZ procedure reduces pain in a
large majority of patients, postoperative ataxia and
hemiparesis from injury to adjacent tracts have limited its widespread use. The use of NIOM for the
DREZ procedure in the spinal cord will be discussed
first, followed by its use in the nucleus caudalis
DREZ.
50.4.2.1. Spinal cord DREZ procedure
The spinal cord DREZ procedure can be performed at
any level of the spinal cord, but it is most often done
on cervical and lumbar enlargements. The patient is
A.M. HUSAIN
placed in a prone position and a laminectomy,

corresponding to the dermatomal area of pain, is performed. The posterior aspect of the spinal cord is
exposed, and the site of pathology is then identified
through an operating microscope. The dorsolateral sulcus, the site of entry of dorsal rootlets into the dorsal
horn, is identified at the lateral margin of the dorsal column (Gorecki, 2004). Lesions are made with the DREZ
electrode, which has a diameter of 0.25 mm, and an
exposed tip of 2 mm (Rawlings et al., 1989). The electrode is inserted into the dorsolateral sulcus so that the
entire 2 mm tip is in the substance of the spinal cord.
Lesions are made by heating the DREZ electrode to
75 C for 15 s. Consecutive lesions are 1 mm apart;
10 lesions are made per spinal level. Once the lesions
are completed, the dura is closed (Nashold, 1996).
A variation of the DREZ procedure is known as the
microsurgical DREZotomy (Gorecki, 2004). In this
procedure, an incision is made along the lateral edge
of the dorsal rootlets and extended ventromedially
for 23 mm in the dorsal root using microcoagulation.
This destroys small nociceptive fibers, the cells in
laminae IIII, and the medial part of Lissauers tract.
Because of the proximity of the DREZ to the
dorsal column, lateral pyramidal, and dorsal spinocerebellar tracts, these tracts may be damaged during
the DREZ procedure (Fig. 4). Consequently,
Dorsal column pathway

Dorsal root
Dorsal spinocerebellar tract
Lateral corticospinal tract
Fig. 4. Schematic of one-half of cross-section of spinal cord at cervical level. The heavy arrow indicates the proper trajectory of the dorsal root entry zone (DREZ) electrode; deviation ventrally risks damage to the lateral corticospinal and dorsal
spinocerebellar tracts, whereas deviation dorsally risks damage to the dorsal column pathway.
SPINE SURGERY
ipsilateral weakness, ataxia, and sensory loss are

potential complications of this procedure. NIOM is
used in two ways to minimize these complications.
Various protocols can be used to identify the DREZ
physiologically in addition to the visual identification
done with the operating microscope. Additionally,
during lesioning nearby tracts can be physiologically
monitored to minimize the chance of inadvertent
injury. Several investigators have described their
techniques for localizing the DREZ and monitoring
function of nearby tracts during lesioning, and a summary of the various techniques will be presented.
Techniques for localizing the DREZ beyond simple visual inspection have been advocated by several
investigators (Campbell and Miles, 1984; Iacono
et al., 1992; Fazl et al., 1995; Tomas and Haninec,
2005). The tibial nerve is stimulated with surface or
needle electrodes according to commonly used protocols. The recording electrode is a hand-held concentric bipolar electrode. With the aid of the operating
microscope, this electrode is inserted into the spinal
cord in the region of the dorsal columns, DREZ, and
the dorsal spinocerebellar tract with small punctures.
High amplitude, polyphasic responses are obtained
from the dorsal column and smaller responses are
obtained from the dorsal spinocerebellar tract. No
response is obtained from the DREZ, thereby localizing the site for lesioning (Campbell and Miles, 1984).
A variation of the above technique places both the
recording and stimulating electrodes on the spinal cord.
The recording electrode is a strip electrode with two
contacts, and it is slipped under the dura, rostral to the
site of surgery (Tomas and Haninec, 2005). This electrode allows a bipolar recording because two electrodes
are present on the surface of the spinal cord. A small silver ball electrode can also be placed on the dorsal aspect
of the cord and referenced to a needle electrode in the
wound, for a referential recording (Fazl et al., 1995).
The latter approach may record more artifacts, due to
the distance between the two electrodes. The stimulating electrode can also be either bipolar or monopolar.
The bipolar electrode is similar to the bipolar recording
electrode, and provides surface stimulation of the spinal
cord (Tomas and Haninec, 2005). The monopolar electrode is typically either a cordotomy or DREZ lesioning
electrode, and its reference is placed elsewhere on the
body (Fazl et al., 1995). The advantage with the bipolar
electrode is its more focal delivery of stimulation. Stimulation is with a square-wave pulse, with a duration
of 0.1 ms, intensity of 0.10.2 mA, and frequency of
13 Hz. Typically, no more than 510 responses need
709
to be averaged to obtain a reproducible response. The

stimulating electrode is moved in 1 mm increments
from the dorsal column, over the DREZ to the dorsal
spinocerebellar tract. A triphasic response is seen when
the stimulating electrode is over the dorsal column
and the dorsal spinocerebellar tract, but not when it is
over the DREZ. This localization technique can be
repeated at different levels to confirm the location of
the DREZ, and lesions can be made accordingly.
NIOM has also been used for spinal cord level
localization after exposure to ensure appropriate
lesioning (Friedman and Nashold, 1984; Nashold
et al., 1985; Makachinas et al., 1988; Jeanmonod
et al., 1989, 1990; Jeanmonod and Sindou, 1991;
Sindou et al., 1994; Sharpe and Pearlstein, 1996).
Stimulating electrodes are placed over the median,
ulnar, radial, or tibial nerve, depending on the location
of the pain. The recording electrode consists of 24
small disks embedded in Silastic or a small silver ball
attached to a silver wire. The reference electrode is
attached to the retractor in the wound site or another
nearby location. The recording electrode is oriented longitudinally over the DREZs and the evoked potentials
are obtained simultaneously from all contacts of the
electrodes in a monopolar derivation. The site of the
maximum amplitude is the main root involved in sensory transmission of the nerve stimulated (Sharpe and
Pearlstein, 1996). Dorsal roots can also be stimulated,
and recordings made, from the same recording electrode (Nashold et al., 1985). Some problems may be
encountered with these techniques. If the painful area
has damaged or avulsed dorsal roots, evoked potentials
may be difficult or impossible to record from those most
important segments of the spinal cord. With dorsal root
stimulation and recording over the DREZ, the distance
between the stimulating and recording electrodes is
about 1 cm. This may result in the stimulus artifact contaminating the response. Additionally, with dorsal root
stimulation, it may not be possible to localize the spinal
level with accuracy because the stimulated level of the
dorsal root stimulated may not be clear.
Motor root stimulation has also been used to localize the spinal level (Jeanmonod and Sindou, 1991;
Sharpe and Pearlstein, 1996). The ventral root is
stimulated with a bipolar electrode (in the form of
hook electrodes or forceps). A stimulus of 0.05 ms
duration, 0.26 V intensity, and 2.5 Hz frequency is
used. Recordings can either be made with needle or
surface electrodes on muscles of interest, or by direct
observation of the muscles. Neuromuscular blocking
agents cannot be used during this type of NIOM.
710
NIOM can also be useful to ascertain if the lesioning electrode is too close to the lateral pyramidal tract.
Because this tract lies immediately ventral to the
dorsal root, if the DREZ electrode is placed too
deeply, it can damage the tract, producing ipsilateral
motor weakness. At the authors center, once the
DREZ electrode has been positioned by the surgeon,
a square-wave impulse of 0.05 ms duration is passed
through the electrode and the intensity increased up
to 1 V. Recording needle electrodes are placed on
proximal and distal muscles of the upper and lower
extremities (examples include: biceps brachii, triceps,
first dorsal indices, flexor digitorum superficialis,
vastus lateralis, tibialis anterior, medial gastrocnemius, and semitendinosus muscles). If a muscle
response is noted in any of the muscles at 1 V stimulation, the DREZ electrode is repositioned. If activation of muscles is not observed, a lesion is made.
Changes in evoked potentials before and after
DREZ lesions are made have been used to gauge adequacy of lesioning as well as possibility of postoperative complication (Prestor et al., 1989; Jeanmonod
and Sindou, 1991; Mauguie`re, 1999; Tomas and Haninec, 2005). When the median or tibial nerve is stimulated, and recordings are made over the DREZ with a
small silver ball electrode attached to a silver wire
and referenced to a distal silent reference (i.e., contralateral Erb point or knee), a complex waveform is seen
in normal individuals (Jeanmonod and Sindou, 1991).
The first wave is a positive potential (P9). This is
thought to be a far-field potential from the brachial
plexus. This is followed by a small negative peak,
N11, which is a near-field potential arising from presynaptic fibers in the dorsal root. A large negative peak,
N13, is also a near-field potentials arising form the
postsynaptic laminae IV and V of the dorsal root. The
tibial nerve equivalent of these waveforms is P17,
N21, and N24. If the recording electrode is moved to
lie over the dorsal column in the cervical region, and
the tibial nerve is stimulated, a series of positive and
negative peaks will be seen. After the DREZ procedure
or DREZotomy, the reduction in the amplitude of the
N13 (and N24), as recorded from the DREZ, will
depend on the number of roots lesioned (Mauguie`re,
1999). This amplitude reduction can be used to ensure
that not too many or too few spinal levels are lesioned.
The polyphasic response, recorded over the dorsal column, should not change with the lesioning. However, if
a 50% amplitude reduction, disappearance of the initial
negative components, or appearance of a new positive
component are noted in the dorsal column response,
A.M. HUSAIN
postoperative neurological deficit is likely (Prestor

et al., 1989; Tomas and Haninec, 2005).
Impedance measurements have also been used in the
DREZ procedures. Application of high-frequency alternating current through an electrode allows determination of impedance in electrically grounded human
tissue. Animal studies have revealed that different parts
of the nervous system have different impedance (Vieira
et al., 1988). For example, the impedance range of white
matter in the cat is between 1,500 and 1,800 O, whereas
in the DREZ it is between 2,200 and 3,100 O.
In humans, the impedance of the spinal cord is from
900 to 1,200 O. The impedance is lower in cases of
pathology. For example, after spinal root avulsion, the
impedance is typically below 1,000 O (Vieira et al.,
1988). While performing a DREZ procedure, the
impedance can be checked with the DREZ electrode,
each time it is inserted into the spinal cord. If avulsion
of spinals root has occurred, the margins of the abnormality can be mapped with impedance. The impedance
of the tissue also decreases after thermocoagulation, and
typically after a DREZ procedure, the impedance of the
DREZ is typically about 600 O (Bullard, 1996). However, studies describing how much the impedance must
be reduced for effective pain control have not yet been
reported.
Unequivocal data that NIOM aids in the DREZ procedure are not available. This type of surgery is performed in relatively few centers, and surgeons
performing these surgeries have either always used
some type of NIOM, or not used it all. Additionally,
outcome data of DREZ procedures is difficult to evaluate because it is highly dependent on surgical expertise,
as well as patient selection. One group of investigators
have, however, reported their experience without using
NIOM for the first 4 years, and then using NIOM to
identify the DREZ before lesioning for the next 6 years.
Whereas their complication rates were not different,
the frequency of pain relief was significantly higher
when NIOM was used; good pain relief was noted in
83% of patients when NIOM was used, compared to
33% when it was not used (Tomas and Haninec,
2005). Anecdotally, many surgeons feel that NIOM
provides valuable assistance in localizing spinal level,
identifying the DREZ, and/or preventing lesions from
extending to the lateral pyramidal tract. Thus, there
appears to be some justification in performing NIOM
for these procedures. Currently at the authors institution, NIOM for these procedures consists of stimulation
with the DREZ electrode, before lesioning with recordings obtained from muscles of the upper and lower
SPINE SURGERY
711
extremities. The surgeons feel that localization of the

spinal level, and identification of the DREZ, can be
done adequately by radiographic evaluation and visual
examination under the surgical microscope.
50.4.2.2. Nucleus caudalis DREZ procedure
The nucleus caudalis DREZ procedure is preformed
with the patient in the prone position, with the neck
slightly flexed (Nashold et al., 1996). A suboccipital
craniotomy and laminectomies of the upper cervical
vertebrae are performed, and dura opened. The C2
root is identified, and the point at which it enters the
spinal cord is noted; this is the DREZ. A thermocoagulation lesion is made at 75 C for 15 s. Subsequent
lesions are extended rostrally in the nucleus caudalis,
until the level of the obex or higher. Two rows of
lesions are typically made adjacent to each other, with
about 1520 lesions in each row having about 1520
lesions.
Rostrally, the dorsal spinocerebellar tract lies
immediately dorsolateral, and the dorsal column is
immediately dorsal, to the nucleus caudalis. Further
caudally, these tracts maintain their relation to the
nucleus caudalis, and the pyramidal tract lies immediately ventral to it (Fig. 5). Thus, the potential for
injury to these tracts, and resulting ataxia and weakness, is strong if the lesion is misplaced. To minimize
these complications, the DREZ electrode used for
making nucleus caudalis lesions has undergone
several modifications in the length and angulation of
the lesioning tip (Young et al., 1989; Nashold et al.,
1992, 1994). NIOM has also been used to localize
the nucleus caudalis and prevent damage to adjacent
tracts during lesioning (Bullard and Nashold, 1997;
Husain et al., 2002).
To localize the various parts of the nucleus caudalis (ophthalmic, maxillary, and mandibular), trigeminal evoked potentials can be used (Husain et al.,
2002). Needle electrode pairs are used to stimulate
the supraorbital, infraorbital, and mental nerves, as
described above (see section on trigeminal rhizotomy). Additionally, needle electrodes are placed at
the wrist to stimulate the median nerve. The DREZ
electrode is used as the recording electrode, and with
a special adapter, it is connected to the evoked
potentials machine. This electrode has an uninsulated
obex
b
a
c
Cross-section at
level of C2
C
e
e
ba
Coronal section of
medulla and upper
cervical cord
Cross-section at
level of
decussation of
corticospinal
tracts
D
e
Cross-section at
level of
decussation of
medial lemnisci
ba
d
c
Rostral most C2
rootlet
Fig. 5. Schematic of the lower brainstem demonstrating the nucleus caudalis and its relations. Legend: a nucleus caudalis; b spinal tract of the trigeminal nucleus; c corticospinal tract; d dorsal spinocerebellar tract; e dorsal column
pathway. Adapted from Husain et al. (2002) with permission of Lippincott, Williams and Wilkins.
712
A.M. HUSAIN
tip that serves as a monopolar electrode; a needle in

the wound site is used as the reference.
The supraorbital, infraorbital, and mental nerves
are stimulated with a square-wave pulse that has a
duration of 0.1 ms and an intensity of 5 mA.
The stimulation rate is 2 Hz. The median nerve
is stimulated as per standard protocol. All four
nerves are stimulated sequentially. The DREZ electrode is inserted just postromedial to the entry of the
rostralmost C2 rootlet for the first recording. Fifty
responses are averaged to obtain a reproducible
response. Once a recording from all four nerves is
obtained, the DREZ electrode is moved to a different
location, and stimulation and recording are repeated.
Depending on the amplitude of the response, the parts
of the nucleus caudalis subserving the supraorbital
(ophthalmic), infraorbital (maxillary), and mental
(mandibular) nerves can be localized. Additionally,
the dorsal column can be localized by noting the point
of highest amplitude of the median nerve response.
This data can be used to map the location of the
nucleus caudalis, and it can also help limit the number

of lesions made by better targeting lesions to the particular part of the nucleus caudalis subserving the area
of pain (Husain et al., 2002) (Fig. 6).
It should be cautioned that when one or more of
the nerves that need to be stimulated is damaged,
either by primary disease (as with herpes zoster) or
by iatrogenically (as with trigeminal rhizotomy), an
evoked potential from that nerve will be difficult to
obtain. Consequently, localization of the painful area
of the face on the nucleus caudalis may not be
achieved. In this circumstance, however, evoked
potentials form the other nerves can often be
obtained, and surrounding areas can be mapped, on
the nucleus caudalis. Thus, some useful information
can still be obtained with NIOM.
As with the spinal cord DREZ procedures, NIOM
can be used to determine if the DREZ electrode is too
close to the pyramidal tract (Husain et al., 2002).
Needle electrodes are placed in proximal and distal
ipsilateral upper and lower extremity muscles, as
7:MEDIAN
3:INFRAORBITAL
5:MENTAL
1:SUPRAORBITAL
16:12:05 17:17:16
16:11:24 17:18:43
16:10:54 17:18:17
16:10:24 17:15:51
3 ms
5 V
5 V
1 ms
1 ms
1 ms 20 V
5 V
/
/
ms
ms
L1/L2:
L1/L2:
/
L1/L2:
L1/L2:
/
ms
ms
Fig. 6. Supraorbital, infraorbital, mental, and median nerve evoked potentials recorded from the dorsal root entry zone
(DREZ) electrode. The amplitude of the response from various nerves can be used to localize within the nucleus caudalis
for more accurate lesioning. During recording of the first tracing, the electrode was closest to the fibers from the supraorbital nerve, whereas between the sixth and eighth tracing the electrode was closer to the dorsal column pathway.
SPINE SURGERY
described above (see section on spinal cord DREZ

procedures). Once the DREZ electrode is positioned
by the surgeon at a site for lesioning, a square-wave
impulse of 0.05 ms duration is passed through the
electrode, and the intensity increased up to 1 V. If a
motor response is noted, the electrode is too close
to pyramidal tract, and is repositioned. If no response
is seen, a lesion is made.
Data about whether NIOM helps improve efficacy,
and reduce morbidity, for nucleus caudalis DREZ is
even more limited than other types of DREZ procedures. One major limitation is that most studies reporting outcomes of this type of surgery have been
performed by surgeons trained at one center. In one
study, five patients underwent the type of NIOM
described above, and outcomes were compared to
prior studies in which NIOM was not used (Gorecki
et al., 1995; Nashold et al., 1996; Husain et al.,
2002). When NIOM was used, an average of 4 lesions
were made compared with 1520 in the other series.
The efficacy of the surgery in relieving pain was comparable, but fewer complications (ataxia and weakness) fewer when NIOM used (Husain et al., 2002).
Consequently, at the authors institution, the practice
for nucleus caudalis DREZ procedures is to use
NIOM, as described above, and make fewer, bettertargeted lesions.
50.4.3. Motor cortex stimulation
Motor cortex stimulation is a type of augmentative
pain surgery in which a small, bipolar electrical stimulator is placed epidurally over the motor cortex.
The stimulator delivers low-intensity stimulation to
the underlying brain. Although the exact mechanism
by which motor cortex stimulation alleviates pain
remains unknown, a leading hypothesis suggests that
thalamic hyperactivity associated with pain is inhibited with this type of stimulation (Brown and
Barbaro, 2003). Many different types of pain syndromes, including postherpetic neuralgia, brachial
plexus avulsion, post-stroke, phantom limb, and spinal
cord injury pain, have been treated with motor cortex
stimulation. Pain relief is noted in about 50% of
patients with limb or body pain, whereas better results
have been observed in patients with facial pain
(Meyerson et al., 1993; Nuti et al., 2005).
Although an epidural electrode can be implanted
surgically through a burr hole, a craniotomy is preferred, as it allows better localization of the sensorimotor cortex (Nguyen et al., 1997; Nuti et al.,
713
2005). Localization of the sensorimotor cortex can

be accomplished with neurophysiological techniques
(described below), or with neuroimaging. Intraoperative MRI, combined with neuronavigation software,
has greatly enhanced the surgeons ability to identify
the central sulcus, and consequently, the sensorimotor cortex (Roux et al., 2001).
NIOM is used for motor cortex stimulator implantation to help localize the sensorimotor cortex, and
not to prevent injury to neural tissue. The technique
used most consistently is recording of the contralateral median nerve SEP with a strip or grid electrode
placed epidurally (Nguyen et al., 1997; Yamamoto
et al., 1997; Carroll et al., 2000; Velasco et al.,
2002; Nuti et al., 2005) (Fig. 7). A phase reversal
(N20-P22) is seen when the electrodes straddle the
central sulcus, with the anterior electrode over the
motor cortex (P22) and the posterior one of
the sensory cortex (N20s) (Fig. 8). Details of the
use of median nerve SEPs in localization of the central sulcus are discussed elsewhere in this text.
Stimulation of the motor cortex to further identify
the optimal site for epidural stimulator placement has
also been described (Canavero and Bonicalzi, 1995;
Nguyen et al., 1997; Yamamoto et al., 1997; Carroll
et al., 2000; Velasco et al., 2002). Most investigators
have conducted motor stimulation in an awake
patient with the epidural electrode. However, alternative stimulators, such as the Ojemann cortical stimulator, can also be used. In an awake patient, the
motor response is in the form of a tonic muscle
Fig. 7. A 4 4 grid placed extradurally over the sensorimotor cortex for recording somatosensory evoked potentials evoked by stimulation of the contralateral median
nerve. Numbering on the grid corresponds to the numbers
of the channels in Fig. 8.
714
A.M. HUSAIN
P1
50 ms
50 V
P2
50 ms
50 V
P3
50 ms
50 V
P4
50 ms
50 V
P5
50 ms
50 V
P6
50 ms
50 V
P7
50 ms
50 V
P8
50 ms
50 V
EP-E
50 ms
5 V
112/0+.
112/0+.
112/0+.
112/0+.
112/0+.
112/0+.
112/0+.
112/0+.
108/5
Fig. 8. Somatosensory evoked potentials recorded from the first two rows of the 4 4 grid shown in Fig. 7. A high-amplitude
phase reversal (N20-P22) is noted between channels 8 and 7. Channel 8 depicts the N20 and consequently is over the somatosensory area of the hand, whereas channel 7 shows the P22 and is over the motor area. A phase reversal is also noted between
channels 4 and 3, however is of smaller amplitude than between 8 and 7, so is further away from the hand area. Channel 9 shows
the Erb point response ipsilateral to the side of stimulation, verifying adequate stimulation. Legend: 5 ms/div and 50 mV/div.
contraction, muscle spasm, involuntary movements,

or abnormal sensations. Muscle activation with cortical stimulation can also be monitored in a patient
who is asleep, if needle electrodes are inserted into
the muscles of interest in the face and limbs. However, if the patient is not awake, neuroleptic anesthesia or very low concentrations of halogenated gases
must be used to evoke the motor response.
The parameters used for motor stimulation have
varied, depending on the report. Most reports suggest
using a bipolar stimulus of 0.51.0 ms duration at an
intensity of up to 15 mA. The stimulation frequency
is 15130 Hz. When using motor stimulation for localization, caution must be used as excessive or highintensity stimulation may result in seizures (Fig. 9).
Whether motor stimulation helps with final localization remains unclear. Some authors have noted
that, in patients in whom a motor response was elicited, the outcome of the pain syndrome was better
(Nguyen et al., 1997; Carroll et al., 2000). Others
have suggested that this type of NIOM is unreliable.
In many cases, motor activation will be unsuccessful,
and in others, it may be misleading, that is, sensory perception despite motor cortex stimulation (Canavero
and Bonicalzi, 1995; Carroll et al., 2000; Velasco
et al., 2002).
The utility of NIOM in motor cortex stimulator

implantation remains uncertain. Whereas phase reversal with median nerve SEPs helps greatly in identifying the central sulcus, it is unclear if it adds
Fig. 9. Markings on the brain indicating the point of phase

reversal of the median somatosensory evoked potential
(N20-P22) point 2 (thin arrow), and location where
stimulation resulted in facial muscle contraction point
3 (thick arrow). Identification of point 2 (N20-P22) helps
the surgeon locate point 3, the site of implantation of the
epidural stimulating electrode.
SPINE SURGERY
to data visualized with intraoperative MRI and neuronavigation software. In many centers, including
the authors, neuronavigation software, median nerve
SEP phase reversal, and motor activation are performed to optimize placement of the motor cortex
stimulator.
50.4.4. Other surgeries for pain
Deep brain stimulation is another type of augmentative surgery which has been used to treat both
nociceptive and neuropathic types of pain. It is performed less often since other types of neurostimulation such as motor cortex and SCS have become
available. The target for stimulation is selected based
on type and location of the pain. Some common
targets are the ventral posterolateral and ventral posteromedial nuclei of the thalamus for neuropathic
body and face pain, respectively. Periaqueductal and
periventricular gray matter are also common targets
as their stimulation releases endogenous opiods.
Stimulation of the posteroinferior hypothalamus is
efficacious in the treatment of chronic headaches
(Wallace et al., 2004). The electrode used for deep
brain stimulation is stereotactically placed. The
patients are often awake so that electrical stimulation
of the target areas can be performed to help with localization. Neurophysiological methods such as intracellular and extracellular recording are also used in target
localization. These methods have been described elsewhere in this text and will not be discussed here.
There are numerous other types of pain surgeries.
Many are done with the patient awake. Special electrodes are used for electrical stimulation of various
neural structures near the site to be lesioned
(Schvarcz, 1975). This allows better localization of
the target and helps avoid injury to other adjacent
areas. Pain surgeries on the brain are usually performed stereotactically, allowing better target localization. In some cases, such as cordotomies, impedance
measurements are used (Kanpolat, 2004). The latter
procedure is done percutaneously under radiological
guidance, and impedance measurement helps determine the location of the electrode. Impedance of cerebrospinal fluid is lowest, becoming higher when the
electrode makes contact with the spinal cord, and
increasing further upon penetration. All these techniques are vital in helping the surgeon localize the target, however typically do not need involvement of a
neurophysiologist and consequently are not discussed
here.
715
50.5. Future direction

As surgery for pain evolves so will NIOM for these
surgeries. In the last decade, neurostimulation has
been the preferred method of operative treatment of
pain. Ablative therapies remain in the treatment algorithm, especially for those not responding to other
types of therapies. New techniques are being introduced to better localize targets. Recording of intracellular and extracellular potentials has been reported in
surgeries on the DREZ and may expand to other parts
of the nervous system as well (Falci et al., 2002).
Whether impedance measurements gain wider acceptance remains to be determined. Methods for selective
assessment of various tracts in the spinal cord need to
be developed so that complications can be minimized.
50.6. Conclusion
There are many types of surgeries for pain; all try to
interrupt or modify the sensation of pain. NIOM is
used for these procedures mostly to localize the site
selected for intervention. Once this site has been
selected, NIOM can be used to identify adjacent
tracts and prevent injury to them during surgery.
As pain surgeries evolve, NIOM will have to change
with them.
References
Bennett, MH and Lunsford, LD (1984) Percutaneous retrogasserian glycerol rhizotomy for tic douloureux:
Part 2. Results and implications of trigeminal evoked
potential studies. Neurosurgery, 14: 431435.
Brown, JA (2005) Pain management. In: SS Rengachary
and RG Ellenbogen (Eds.), Principles of Neurosurgery.
Elsevier Mosby, Edinburgh, 2nd ed., pp. 407414.
Brown, JA and Barbaro, NM (2003) Motor cortex stimulation for central and neuropathic pain: current status.
Pain, 104: 431435.
Bullard, DE (1996) The measurement of impedance in the
DREZ operation. In: BSJ Nashold and RD Pearlstein
(Eds.), The DREZ Operation. The American Association of Neurological Surgeons, Park Ridge, IL,
pp. 4146.
Bullard, DE and Nashold, BS, Jr. (1997) The caudalis
DREZ for facial pain. Stereotact. Funct. Neurosurg.,
68: 168174.
Bullard, DE and Nashold, BSJ (2004) Caudalis nucleus drosal
root entry zone procedure for the treatment of intractable
facial pain. In: HR Winn (Ed.), Youmans Neurological
Surgery. Saunders, Philadelphia, pp. 30853091.
716
Burchiel, KJ, Anderson, VC, Brown, FD, et al. (1996) Prospective, multicenter study of spinal cord stimulation
for relief of chronic back and extremity pain. Spine,
21: 27862794.
Campbell, JA and Miles, J (1984) Evoked potentials as an
aid to lesion making in the dorsal root entry zone. Neurosurgery, 15: 951952.
Canavero, S and Bonicalzi, V (1995) Cortical stimulation
for central pain. J. Neurosurg., 83: 1117.
Carroll, D, Joint, C, Maartens, N, et al. (2000) Motor cortex stimulation for chronic neuropathic pain: a preliminary study of 10 cases. Pain, 84: 431437.
Cohen, RA, Paul, R, Zawacki, TM, et al. (2001) Emotional
and personality changes following cingulotomy. Emotion, 1: 3850.
De Jongste, MJ, Hautvast, RW, Hillege, HL, et al. (1994)
Efficacy of spinal cord stimulation as adjuvant therapy
for intractable angina pectoris: a prospective, randomized clinical study. Working Group on Neurocardiology. J. Am. Coll. Cardiol., 23: 15921597.
Delgado-Lopez, P, Garcia-Salazar, F, Mateo-Sierra, O, et al.
(2003) Trigeminal nucleus caudalis dorsal root entry zone
radiofrequency thermocoagulation for invalidating facial
pain. Neurocirurgia (Astur), 14: 2532; discussion.
Drechsler, F and Neuhauser, B (1986) Somatosensory trigeminal evoked potentials in normal subjects and in
patients with trigeminal neuralgia before and after thermocoagulation of the ganglion Gasseri. Electromyogr.
Falci, S, Best, L, Bayles, R, et al. (2002) Dorsal root entry
zone microcoagulation for spinal cord injury-related
central pain: operative intramedullary electrophysiological guidance and clinical outcome. J. Neurosurg., 97:
193200.
Fazl, M, Houlden, DA and Kiss, Z (1995) Spinal cord
mapping with evoked responses for accurate localization of the dorsal root entry zone. J. Neurosurg., 82:
587591.
Follett, KA (2004) Neurosurgical management of intractable
pain. In: HR Winn (Ed.), Youmans Neurological Surgery.
Saunders, Philadelphia, 5th ed., pp. 30233032.
Friedman, AH and Nashold, BS, Jr. (1984) Dorsal root
entry zone lesions for the treatment of postherpetic neuralgia. Neurosurgery, 15: 969970.
Gorecki, JP (2004) Dorsal root entry zone and brainstem
ablative procedures. In: HR Winn (Ed.), Youmans Neurological Surgery. Saunders, Philadelphia, 5th ed., pp.
30453058.
Gorecki, JP, Nashold, BS, Jr., Rubin, L, et al. (1995) The
Duke experience with nucleus caudalis DREZ coagulation. Stereotact. Funct. Neurosurg., 65: 111116.
Hanajima, R, Ashby, P, Lang, AE, et al. (2002) Effects of
acute stimulation through contacts placed on the motor
cortex for chronic stimulation. Clin. Neurophysiol.,
113: 635641.
A.M. HUSAIN
Hosobuchi, Y, Adams, JE and Linchitz, R (1977) Pain
relief by electrical stimulation of the central gray matter
in humans and its reversal by naloxone. Science, 197:
183186.
Husain, AM, Elliott, SL and Gorecki, JP (2002) Neurophysiological monitoring for the nucleus caudalis dorsal
root entry zone operation. Neurosurgery, 50: 822827;
discussion 78.
Iacono, RP, Guthkelch, AN and Boswell, MV (1992)
Dorsal root entry zone stimulation for deafferentation
pain. Stereotact. Funct. Neurosurg., 59: 5661.
Jeanmonod, D and Sindou, M (1991) Somatosensory function following dorsal root entry zone lesions in patients
with neurogenic pain or spasticity. J. Neurosurg., 74:
916932.
Jeanmonod, D, Sindou, M and Mauguie`re, F (1989) Intraoperative spinal cord evoked potentials during cervical
and lumbo-sacral microsurgical DREZ-tomy (MDT)
for chronic pain and spasticity (preliminary data). Acta
Neurochir. Suppl. (Wien), 46: 5861.
Jeanmonod, D, Sindou, M and Mauguie`re, F (1990) Intraoperative electrophysiological recordings during microsurgical DREZ-tomies in man. Stereotact. Funct.
Neurosurg., 5455: 8085.
Kanpolat, Y (2004) The surgical treatment of chronic pain:
destructive therapies in the spinal cord. Neurosurg.
Clin. N. Am., 15: 307317.
Kaplitt, MG, Rezai, AR, Lozano, AM, et al. (2004) Deep
brain stimulation for chronic pain. In: HR Winn (Ed.),
Youmans Neurological Surgery. Saunders, Philadelphia,
5th ed., pp. 31193131.
Karol, EA, Sanz, OP and Rey, RD (1991) Sensory and
motor trigeminal evoked potentials to localize the position of trigeminal electrodes. Acta Neurochir. (Wien),
108: 110115.
Kumar, K, Nath, RK and Toth, C (1997) Spinal cord stimulation is effective in the management of reflex sympathetic dystrophy. Neurosurgery, 40: 503508;
discussion 89.
Landi, A, Zanusso, M, Curri, D, et al. (1991) Trigeminal
evoked potentials in patients undergoing percutaneous
microcompression of gasserian ganglion. Stereotact.
Leandri, M (1998) Generator sites of early scalp potentials
evoked from the three trigeminal branches. J. Neurosurg., 88: 718725.
Leandri, M and Gottlieb, A (1996) Trigeminal evoked
potential-monitored thermorhizotomy: a novel approach
for relief of trigeminal pain. J. Neurosurg., 84: 929939.
Leandri, M, Parodi, CI and Favale, E (1988) Normative
data on scalp responses evoked by infraorbital nerve
stimulation. Electroencephalogr. Clin. Neurophysiol.,
71: 415421.
Leandri, M, Parodi, CI and Favale, E (1989) Early scalp
responses evoked by stimulation of the supraorbital
SPINE SURGERY
nerve in man. Electroencephalogr. Clin. Neurophysiol.,
74: 367377.
Leandri, M, Parodi, CI, Rigardo, S, et al. (1990) Early
scalp responses evoked by stimulation of the mental
nerve in humans. Neurology, 40: 315320.
Lee, KH, Chang, JW, Park, YG, et al. (1997) Microvascular
decompression and percutaneous rhizotomy in trigeminal
neuralgia. Stereotact. Funct. Neurosurg, 68: 196199.
Liu, JK and Apfelbaum, RI (2004) Treatment of trigeminal
neuralgia. Neurosurg. Clin. N. Am., 15: 319334.
Lunsford, LD and Bennett, MH (1984) Percutaneous retrogasserian glycerol rhizotomy for tic douloureux: Part 1.
Technique and results in 112 patients. Neurosurgery,
14: 424430.
Macon, JB and Poletti, CE (1987) Human trigeminal root
evoked potentials during differential retrogasserian thermal and chemical rhizotomy. Pain, 31: 307316.
Makachinas, T, Ovelmen-Levitt, J and Nashold, BS, Jr.
(1988) Intraoperative somatosensory evoked potentials.
A localizing technique in the DREZ operation. Appl.
Mauguie`re, F (1999) Utility of somatosensory evoked
potentials (SEPs) in spinal cord lesions and functional
surgery of pain and spasticity. Electroencephalogr.
Clin. Neurophysiol. Suppl., 50: 3139.
McGlone, F and Wells, C (1991) Peroperative monitoring
during percutaneous thermocoagulation of the Gasserian ganglion for the treatment of trigeminal neuralgia.
Br. J. Neurosurg., 5: 3942.
Meyerson, BA, Lindblom, U, Linderoth, B, et al. (1993)
Motor cortex stimulation as treatment of trigeminal neuropathic pain. Acta Neurochir. Suppl. (Wien), 58: 150153.
Mullan, S and Lichtor, T (1983) Percutaneous microcompression of the trigeminal ganglion for trigeminal neuralgia. J. Neurosurg., 59: 10071012.
Nashold, BS, Jr. (1984) Current status of the DREZ operation: 1984. Neurosurgery, 15: 942944.
Nashold, BSJ (1996) Clinical applications of the DREZ
operation: general introduction. In: BSJ Nashold and
RD Pearlstein (Eds.), The DREZ Operation. The American Association of Neurological Surgeons, Park Ridge,
Illinois, pp. 4772.
Nashold, BSJ, Urban, B and Zorub, DS (1976) Pain relief
by focal destruction of the substantia gelatinosa of
Roland. Adv. Pain Res. Ther., 1: 959963.
Nashold, BS, Jr., Ovelmen-Levitt, J, Sharpe, R, et al.
(1985) Intraoperative evoked potentials recorded in
man directly from dorsal roots and spinal cord. J. Neurosurg., 62: 680693.
Nashold, BS, Jr., El-Naggar, A, Abdulhak, MM, et al.
(1992) Trigeminal nucleus caudalis dorsal root entry
zone: a new surgical approach. Stereotact. Funct. Neurosurg., 59: 4551.
Nashold, BS, Jr., El-Naggar, AO, Ovelmen-Levitt, J, et al.
(1994) A new design of radiofrequency lesion electrodes
717
for use in the caudalis nucleus DREZ operation.
Technical note. J. Neurosurg., 80: 11161120.
Nashold, BSJ, El-Naggar, AO and Gorecki, JP (1996)
The microsurgical trigeminal caudalis nucleus DREZ
procedure. In: BSJ Nashold and RD Pearlstein (Eds.),
The DREZ Operation. The American Association of
Neurological Surgeons, Park Ridge, Illinois, pp. 159188.
Nashold, JRB (1996) The surgical technique of the DREZ
operation. In: BSJ Nashold and RD Pearlstein (Eds.),
The DREZ Operation. The American Association of
Neurological Surgeons, Park Ridge, Illinois, pp. 7391.
Nauta, HJ, Soukup, VM, Fabian, RH, et al. (2000) Punctate midline myelotomy for the relief of visceral cancer
pain. J. Neurosurg., 92: 125130.
Nguyen, JP, Keravel, Y, Feve, A, et al. (1997) Treatment
of deafferentation pain by chronic stimulation of the
motor cortex: report of a series of 20 cases. Acta Neurochir. Suppl., 68: 5460.
North, RB (2004) Spinal cord and peripheral nerve stimulation for chronic, intractable pain. In: HR Winn (Ed.),
Youmans Neurological Surgery. Saunders, Philadelphia,
5th ed., pp. 31073118.
Nuti, C, Peyron, R, Garcia-Larrea, L, et al. (2005) Motor
cortex stimulation for refractory neuropathic pain: four
year outcome and predictors of efficacy. Pain, 118:
4352.
Ovelman-Levitt, J (1996) The neurobiology of the spinal
cord dorsal horn and pathophysiology of neuropathic
pain. In: BSJ Nashold and RD Pearlstein (Eds.), The
DREZ Operation. The American Association of Neurological Surgeons, Park Ridge, Illinois, pp. 1326.
Prestor, B, Zgur, T and Dolenc, VV (1989) Subpial spinal
evoked potentials in patients undergoing junctional
dorsal root entry zone coagulation for pain relief. Acta
Ramirez, LF and Levin, AB (1984) Pain relief after
hypophysectomy. Neurosurgery, 14: 499504.
Rawlings, CE, 3rd, El-Naggar, AO and Nashold, BS, Jr.
(1989) The DREZ procedure: an update on technique.
Romanelli, P and Esposito, V (2004) The functional anatomy of neuropathic pain. Neurosurg. Clin. N. Am., 15:
257268.
Romanelli, P, Esposito, V and Adler, J (2004) Ablative
procedures for chronic pain. Neurosurg. Clin. N. Am.,
15: 335342.
Rossitch, E, Jr., Zeidman, SM and Nashold, BS, Jr. (1989)
Nucleus caudalis DREZ for facial pain due to cancer.
Roux, FE, Ibarrola, D, Tremoulet, M, et al. (2001) Methodological and technical issues for integrating functional
magnetic resonance imaging data in a neuronavigational
system. Neurosurgery, 49: 11451156; discussion 5657.
Salar, G, Iob, I and Mingrino, S (1982) Somatosensory
evoked potentials before and after percutaneous
718
thermocoagulation of the Gasserian ganglion for trigeminal neuralgia. Adv. Neurol., 32: 359365.
Sanders, M and Henny, CP (1992) Results of selective percutaneous controlled radiofrequency lesion for treatment of trigeminal neuralgia in 240 patients. Clin. J.
Pain, 8: 2327.
Schvarcz, JR (1975) Stereotactic trigeminal tractotomy.
Confin. Neurol., 37: 7377.
Sharpe, R and Pearlstein, RD (1996) Recordings of
somatosensory evoked cord dorsum potentials and
electromyograms during the DREZ operation. In: BSJ
Nashold and RD Pearlstein (Eds.), The DREZ Operation. The American Association of Neurological Surgeons, Park Ridge, Illinois, pp. 2739.
Sindou, M, Turano, G, Pantieri, R, et al. (1994) Intraoperative monitoring of spinal cord SEPs during microsurgical DREZotomy (MDT) for pain, spasticity and
hyperactive bladder. Stereotact. Funct. Neurosurg., 62:
164170.
Spendel, MC, Deinsberger, R and Lanner, G (1997) Operative treatment of trigeminal neuralgia. Stereotact. Funct.
Neurosurg., 68: 187189.
Stanton-Hicks, M and Salamon, J (1997) Stimulation of the
central and peripheral nervous system for the control of
pain. J. Clin. Neurophysiol., 14: 4662.
Sweet, WH (1976) Controlled thermocoagulation of trigeminal ganglion and rootlets for differential destruction of pain fibers: facial pain other than trigeminal
neuralgia. Clin. Neurosurg., 23: 96102.
Sweet, WH and Wepsic, JG (1974) Controlled
thermocoagulation of trigeminal ganglion and rootlets
A.M. HUSAIN
for differential destruction of pain fibers. 1. Trigeminal
neuralgia. J. Neurosurg., 40: 143156.
Tasker, RR (1990) Thalamotomy. Neurosurg. Clin. N. Am.,
1: 841864.
Tomas, R and Haninec, P (2005) Dorsal root entry zone
(DREZ) localization using direct spinal cord stimulation
can improve results of the DREZ thermocoagulation procedure for intractable pain relief. Pain, 116: 159163.
Velasco, M, Velasco, F, Brito, F, et al. (2002) Motor cortex stimulation in the treatment of deafferentation pain.
I. Localization of the motor cortex. Stereotact. Funct.
Neurosurg., 79: 146167.
Vieira, JF, Shieff, C, Nashold, BS, Jr., et al. (1988) Impedance measurements of the spinal cord of man and animals. Appl. Neurophysiol., 51: 154163.
Wallace, BA, Ashkan, K and Benabid, AL (2004) Deep
brain stimulation for the treatment of chronic, intractable pain. Neurosurg. Clin. N. Am., 15: 343357.
Wilkinson, HA, Davidson, KM and Davidson, RI (1999)
Bilateral anterior cingulotomy for chronic noncancer
pain. Neurosurgery, 45: 11291134; discussion 3436.
Willis, WD and Westlund, KN (1997) Neuroanatomy of
the pain system and of the pathways that modulate pain.
Yamamoto, T, Katayama, Y, Hirayama, T, et al. (1997) Pharmacological classification of central post-stroke pain:
comparison with the results of chronic motor cortex stimulation therapy. Pain, 72: 512.
Young, JN, Nashold, BS, Jr. and Cosman, ER (1989) A
new insulated caudalis nucleus DREZ electrode. Technical note. J. Neurosurg., 70: 283284.
Section III.3
Peripheral Nerve Surgery

M.R. Nuwer (Ed.)
720
CHAPTER 51
Intraoperative testing and monitoring during

brachial plexus surgery
Huan Wanga, Allen T. Bishopb, Alexander Y. Shinb and Robert J. Spinnerb,*
a
Visiting Fellow, Mayo Clinic, and Department of Hand Surgery, Huashan Hospital, Fudan University,
Shanghai, Peoples Republic of China
b
Mayo Clinic, Brachial Plexus Clinic, Rochester, MN 55905, USA
Brachial plexus surgery can be challenging for a

number of reasons: intricate and somewhat variable
neuroanatomy, presence of scar or tumor obscuring
the anatomy, and limited proximal nerve resources
resulting in complex decision making. The preoperative assessment of the nerves is imperfect in cases of
injury or suspected entrapment, and even the location
of the nerve(s) in relation to tumor is suboptimal with
current state-of-the-art examinations and studies.
Operative assessment and intervention for brachial
plexus pathology can be optimized by intraoperative
electrophysiological techniques. These techniques
may be used in surgery of the brachial plexus or its
terminal branches, whether for injury, entrapment,
or tumor. In cases of injury, intraoperative electrophysiological testing allows surgeons to better judge
the function of nerve, the severity of injury, and the
potential for neural recovery. The intraoperative testing can help determine the severity of compression
(e.g., thoracic outlet syndrome) as well as assisting
in the avoidance of accidental injury to neighboring
nerves. Intraoperative electrophysiological testing
can assist the surgeon to preserve critical nerve function when resecting a tumor, and help select nonfunctional fascicles when a nerve biopsy is needed.
Transfer of a portion of normal nerve for muscle
reanimation is also assisted by such tests, by identifying appropriate motor fascicles. In short, intraoperative electrophysiology techniques play an important
role, assisting the surgeon in understanding of pathophysiology of nerve injury. Further, these studies are
*
Correspondence to: Robert J. Spinner, M.D., Mayo Clinic,

Brachial Plexus Clinic, Gonda 8S, 200 First Street SW,
Tel.: 1-507-284-2376; fax: 1-507-284-5206.
E-mail: spinner.robert@mayo.edu (R.J. Spinner).
helpful to preserve critical functions in incomplete

injuries or lesions requiring exploration, nerve
biopsy, or tumor resection. Finally, they are helpful
in appropriate fascicle selection when intra- or extraplexal nerve transfers are used for the reanimation of
paralyzed muscle.
51.1. Traumatic lesions
For brachial plexus injuries, there are a variety of
different techniques available to the surgeon. These
techniques include neurolysis, nerve grafting, nerve
transfers, and/or free functioning muscle transfer
depending on the pathology identified and defined at
surgery as well as the goals of the surgery (Spinner
and Kline, 2000; Kandenwein et al., 2005; Shin et al.,
2005). Because of the numerous different potential
lesions (or combinations of lesions) and corrective strategies, intraoperative decision making is often difficult.
As surgical techniques rely on knowledge of the type
and extent of neural pathology, it is imperative to determine as accurately as possible the extent of injury.
Intraoperative electrophysiological testing adds
useful information to the overall decision making in
brachial plexus trauma. This includes evaluation of
proximal nerve integrity with the central nervous
system (preganglionic lesion), and the assessment of
more distal lesions-in-continuity. In such cases, little
reliable information can be obtained from direct inspection of the external surface of the nerve unless dorsal
root ganglion and/or avulsed rootlets are seen and confirmed by histology. Histologic examination, while
helpful, demands sampling (or sacrifice) of a piece of
the nerve. Additionally, patients with a postganglionic
rupture may still have preganglionic injury.
Before surgery, an adequate history, physical
examination, electrophysiological testing, and imaging
PERIPHERAL NERVE SURGERY
studies are required to: (1) confirm the diagnosis of a

brachial plexus lesion and determine the roots involved,
(2) localize the level of injury (preganglionic vs. postganglionic), and (3) allow some determination of the
severity of nerve injury. However, there are limitations
to all of these examinations and studies.
Intraoperative electrophysiological testing is used
in such cases to provide information unobtainable
with noninvasive means, which can be integrated
with tests and imaging studies. For example, the
electrophysiological integrity of motor and sensory
pathways in individual spinal nerve ventral and
dorsal rami can best be evaluated at surgery. This
information, when correlated with preoperative computed tomography (CT) myelography, can be critical
when considering the use of nerve grafts from individual roots. In some cases, nerve root injury is not
detected by preoperative studies. For example, Kline
and Hudson (1995) found a nonfunctioning C7 nerve
root in 10% of cases by intraoperative electrophysiological testing, when preoperative testing would not
have predicted this. This observation attests to the
expendable nature of C7 due to the overlapping
innervation of muscles innervated by this nerve.
51.1.1. Distinguishing preganglionic from
postganglionic lesions
Standard brachial plexus exploration is performed at a
supraclavicular level, exposing neural elements proximally to an extraforaminal level; more proximal exposure of nerve roots is not typically obtained from this
exposure. The appearance of the neural elements may
be deceiving. At times, there may be little scarring
present at the level of the scalene muscles in patients
with preganglionic injury, due to the more proximal
nature of the injury. In contrast, in other cases of preganglionic injury, retracted avulsed dorsal root ganglia
may contribute to dense scarring. In cases of combined
pre- and postganglionic injury, the operative appearance may be more suggestive of an isolated postganglionic injury with a neuroma in continuity (see below);
sectioning of the nerve near the foramen however
may reveal lack of normal fascicular structure.
Several studies have documented the difficulty distinguishing pre- and postganglionic injury. One recent
study demonstrated that combining physical examination and both routine and paraspinal electromyography
(EMG) yielded an 80% diagnostic rate for intraforaminal lesion, 80% for extraforaminal lesion, and 67% for
combined lesions (Balakrishnan and Kadadi, 2004).
721
It should be noted that paraspinal fibrillations are not

specific for an affected nerve. CT-myelography and
magnetic resonance imaging (MRI) are only accurate
in 85% and 52% of the cases, respectively, as verified
by intradural exploration (Carvalho et al., 1997). Combined myelography and CT-myelography had 23%
false findings in judging ventral root lesions and 27%
false findings in judging dorsal root lesions when compared with findings on intradural inspection (Oberle
et al., 1998). Intraoperative evoked potential recording, on the other hand, has been shown to have 100%
sensitivity for ventral root avulsion with a negative
evoked motor action potential from neck muscles and
100% sensitivity for dorsal root avulsion with a negative somatosensory evoked potential (SEP) when compared with intradural exploration (Oberle et al., 2002).
The incidence of partial root avulsion at one or more
levels has been reported to be as high as 11% in
patients with brachial plexus root avulsions (Carvalho
et al., 1997).
Regardless, without intraoperative electrophysiology, it is difficult, if not impossible, to accurately
assess these nerve roots. Given the limitations with
individual techniques, we typically employ a combination of them, including SEPs, motor evoked potentials (MEPs), and nerve action potentials (NAPs)
during supraclavicular brachial plexus exposures to
distinguish pre- and postganglionic injuries (Kline
and Hudson, 1995; Turkof et al., 1997; Burkholder
et al., 2003; Hattori et al., 2004).
The distinction between a pre- and postganglionic
injury is clinically important in determination of
which reconstructive surgery to perform. Nerve grafting to specific roots with evidence of preganglionic
injury is not an option, for the obvious reason that continuity of the intraspinal rootlets with the spinal cord
has been disrupted. Reimplantation of avulsed nerve
roots has been reported (Carlstedt et al., 2000) but is
only being performed at a few centers. Many brachial
plexus surgeons feel that widespread clinical application of this technique is not yet warranted because of
lack of clinical data and risk of substantial complications (Thomeer et al., 2002). At present, other methods including nerve transfers (neurotization) alone or
combined with functioning free muscle transfer
should be considered for reconstruction of these preganglionic injuries (Fig. 1). It should be remembered
that each ventral ramus (nerve root) is assessed individually for avulsion (preganglionic injury) or rupture
(postganglionic injury), and a final reconstructive plan
then developed that makes use of available resources.
722
Fig. 1. This patient had evidence of a left C5 and C6

brachial plexopathy without evidence of recovery when
he was first evaluated at 3 months. He had denervation in
the rhomboids and midcervical paraspinal muscles. Tinels
sign was absent. CT myelogram suggested C5 and C6
avulsions. Rapid conductions on nerve action potentials
(NAP) recordings and absent motor evoked potentials
(MEPs) and somatosensory evoked potentials (SEPs) from
C5 (shown here) and C6 suggested preganglionic injuries.
Normal responses were obtained from C7. Correlating the
preoperative and intraoperative assessment, we concluded
that there was no functional integrity between the rootlets
and the spinal cord in C5 and C6 and elected to perform
nerve transfers for our reconstruction efforts. In an attempt
to restore shoulder abduction and elbow flexion, we
performed spinal accessory nerve to suprascapular nerve
transfer; triceps branch to anterior branch of axillary nerve;
ulnar nerve fascicle to biceps branch; median nerve fascicle
to brachialis branch. UT, upper trunk; P, phrenic nerve.
51.1.2. Assessment of neuromas in continuity

Intraoperative electrophysiology can be used to
evaluate neuromas in continuity. Evaluation will
determine the potential of spontaneous useful regeneration of the nerve. The assessment should provide
guidance on the choice between neurolysis or resection of the neuroma and nerve grafting. We believe that
the most useful intraoperative testing in this setting in
adults with brachial plexus lesions is NAP recording
(Kline and Happel, 1993) (Fig. 2).
The presence of a NAP has clinical significance.
Large operative series have demonstrated that 92%
of neuromas in continuity with a present NAP result
in good or better functional results with neurolysis
(Kline and Hudson, 1995). Recovery has been
demonstrated even in nerves with typically less
favorable outcomes such as those supplying distal
muscles (the lower trunk or the medial cord) when
NAPs are recorded. In such a circumstance, neuroma
H. WANG ET AL.
Fig. 2. This patient had a left C5 and C6 brachial plexopathy without evidence of reinnervation over 6 months.
In contrast to the patient in Fig. 1, examination of the rhomboids and cervical paraspinals was normal and a Tinels sign
was present at Erbs point. There was no sign of avulsion
based on magnetic resonance imaging (MRI). At surgery, a
bulbous neuroma (arrow) of the upper trunk (UT) was identified. Nerve action potential (NAP) recordings were performed across the lesion and were obtained, including from
C5 to posterior division of the upper trunk (illustrated here).
Based on this neurophysiologic assessment, neurolysis alone
was performed. The patient made good clinical recovery
over the ensuing 18 months. SSN, suprascapular nerve.
resection is ill-advised, likely resulting in a poorer

clinical result than neurolysis alone.
51.1.3. Guiding selection of nerve branch or
fascicles for transfer
Nerve transfer, sometimes termed neurotization,
refers to the use of all or a portion of a functioning
nerve to restore crucial missing function by transfer
to a specific distal motor or sensory target nerve
(Figs. 3 and 4). The development of new distal nerve
transfers, and the demonstration of results superior
to conventional grafting, has revolutionized brachial
723
Fig. 3. An innovative nerve transfer for shoulder abduction has been the introduction of the triceps branch to the axillary
nerve by Leechavengvongs et al. (2003). This technique moves the reconstruction closer to the distal end-organ and can be
accomplished safely in patients (AD, with permission, Mayo Foundation 2006). A: The exposure is done along the posterior aspect of the proximal arm. B: The axillary (anterior and posterior branches to the deltoid) and radial nerves (particularly the triceps branches) are identified. An expendable functioning triceps branch is selected. C: In order to facilitate direct
nerve repair, the triceps branch is sectioned distally and the anterior branch of the axillary nerve is sectioned proximally.
D: The direct nerve transfer can then be accomplished.
plexus surgery. Such nerve transfers may come from

extraplexal sources, or from functioning portions of
the plexus and/or its terminal branches. For several
decades, intercostal nerves have been the workhorse
of nerve transfers, most commonly to restore elbow
flexion. Other nerve transfers were then introduced,
including the transfer of cervical plexus nerves, the
spinal accessory nerve, and pectoral nerves. Other
more controversial extraplexal nerve donors in use
include the ipsilateral phrenic nerve, hypoglossal
nerve and even the opposite, normal side (contralateral) C7 root.
Recently, several new nerve transfers from terminal (distal) plexus branches have been described.
In most cases, these involve selection of a group of
fascicles which are dissected from a large mixed nerve
and transferred with direct coaptation to a specific
muscle motor nerve (Oberlin et al., 1994; Teboul
et al., 2004). The major advantage of this technique
is that it allows the nerve repair to be done close to
the target muscle, resulting in a shorter reinnervation
time. This feature can be exploited in patients who
present late after injury (915 months). These patients
are generally poor candidates for conventional nerve
724
H. WANG ET AL.
Fig. 4. This technique as described by Leechavengvongs et al. (2003) and illustrated in Fig. 3 is demonstrated in this patient
with a C5 and C6 lesion. The operative photograph was taken from an inferior to superior perspective compared to the Fig. 3
position. He underwent the same reconstructive strategy involving nerve transfers, as did the patient described in Fig. 1.
A: The anterior branch of the axillary nerve (A) is sectioned proximally. B: The sectioned axillary nerve (A) was brought
superficially and distally. The radial nerve (R) was exposed. The use of the nerve stimulator allowed us to select an expendable triceps branch (T). C: The triceps branch (T) has been sectioned and was brought near the axillary nerve (A). D: The
nerve ends have been coapted without tension (arrow).
grafting because of the long distance and time

required for nerve regeneration, together with ongoing
irreversible changes in paralyzed muscles following
the lower motor neuron injury.
In order to minimize the risk of causing a new
problem with motor weakness and/or permanent
sensory loss created by such use of a normal nerve,

intraoperative electric stimulation of the nerve is
required to select an expendable fascicle. A disposable hand-held or fine-tip monopolar stimulator is
used to test portions of the nerve or individual fascicles. The fascicle for transfer is chosen according to
the muscle contractions elicited. We typically use a

qualitative approach to select noncritical muscles.
Alternatively, one could perform compound muscle
action potential (CMAP) recordings with surface or
needle electrodes to take a more quantitative
approach. These techniques can be used for determining whether a nerve or a portion of it, such as
contralateral C7 (Holland and Belzberg, 1997), a triceps branch, or a fascicle from the ulnar or median
nerve, is expendable.
In some such transfers, a fascicle supplying a specific function is preferentially chosen. In the case of
nerve transfers for elbow flexion for example, fascicles supplying the wrist flexors from ulnar or median
nerve are directly transferred to the biceps and/or brachialis motor branches of the musculocutaneous
nerve. These transfers are synergistic and can be
learned independently with minimal effort. Such distal nerve transfers have improved outcomes for elbow
flexion reconstruction and for many, have become the
procedure of choice in C5 and C6 and even C5, C6,
and C7 lesions. In several series, this type of reconstruction has led to >90% recovery at M3 or better
for elbow flexion (Teboul et al., 2004). Improved
results can be achieved with double reinnervation
techniques by reinnervating both the biceps and brachialis muscles (Mackinnon et al., 2005).
Microvascular surgery techniques, such as freefunctioning muscle transfers, have also been applied
to brachial plexus reconstruction. This is a microsurgical
725
procedure, transferring a muscle such as the gracilis,

together with its vascular pedicle and motor nerve to
the upper extremity to restore function of one or more
denervated muscles. A common indication includes
restoration of distal function (grasp) when patients
are seen shortly after injury. Wrist and hand function
is seldom restored by nerve graft or nerve transfer procedures. Free muscle transfer also allows reconstruction of elbow flexion in late presentations when
irreversible muscle atrophy has occurred. We have
also used free muscle after nerve surgery when the
final result for a specific function has proved inadequate. Microvascular repair is performed and the muscle reinnervated by direct nerve transfer, most
commonly using 23 intercostal or spinal accessory
nerves. The donor nerve (or fascicle) is identified
and selected using a nerve stimulator (similar to the
method described above).
51.2. Entrapments
The use of NAP recordings in thoracic outlet syndrome surgery provides confirmation of brachial
plexus involvement close to the spine, involving C8
and T1 roots and the lower trunk. The severity of
the compression can be determined, and any functional deficit resulting from previous surgeries
demonstrated. Decrement of NAP amplitude and
NAP conduction velocity is found in most cases for
responses recorded from T1 to the lower trunk and
Fig. 5. GilliattSumner hand. A: Atrophy of the first dorsal web space. B: Atrophy of the thenar and hypothenar eminences.
726
H. WANG ET AL.
Fig. 6. This patient presented with progressive hand weakness and ulnar 2 digit numbness over the past 2 years and had a
GilliattSumner hand. A: Elongated right C7 transverse process (arrow). B: The lower trunk (LT) was compressed and
bowed over a fascial edge of the middle scalene muscle (arrow). SA, subclavian artery. C: The fascial band was released,
scalenotomies were performed, and the elongated transverse process was resected. The lower trunk (LT), C8, and T1 nerves
were decompressed. SA, subclavian artery.
C8 to the lower trunk, and occasionally for responses

recorded from C7 to the middle trunk. These findings
are particularly striking in patients who have neurogenic thoracic outlet syndrome with the typical
GilliattSumner hand (Figs. 5 and 6) (Kline and
Hudson, 1995; Tender et al., 2004). We do not typically use NAPs in routine cases of nerve compression, however, as the results do not change our
course of treatment, namely decompression.
51.3. Tumors/lesions in and around nerve
51.3.1. Neural tumors
Brachial plexus tumors are relatively rare. Reports of
large series have therefore come from centers with a
large referral practice (Ganju et al., 2001; Huang
et al., 2004). Benign nerve sheath tumors are most
common, including schwannomas and the less common neurofibromas. Malignant peripheral nerve
sheath tumors (MPNSTs) can occur in this region as
well, however. Resecting these neural tumors demand
special skill, expertise, and experience as the potential
for downgrading function is relatively high.
The path of nerve fibers across a tumor can be
mapped by intrafield stimulation at various places of
the tumor/nerve tissue mass and/or recording CMAPs
from a group of possible target muscles. A safe zone
free or relatively free of motor fibers can then be outlined and resection of the tumor carried out through
that area. In schwannomas, the tumor can often be dissected from the majority of functioning fascicles.
A single fascicle or two may be seen entering and exiting the tumor. NAP recordings across these entering
and exiting fascicles have demonstrated them to be
nonfunctional. Thus the tumor can be resected entirely

and safely (Kline and Hudson, 1995).
When a benign tumor involves additional fascicles
or is intermingled with the nerve, efforts should be
made to preserve fascicles that produce an NAP
recording or a distal EMG response; those that do
not produce a response may be sacrificed. In some
cases of nerve tumors with unfavorable fascicular or
entire neural involvement, monitoring may help predict postoperative outcome. In some tumors, resection
of the nerve may be necessary. Monitoring may help
predict overlap and ultimately, outcomes in these
cases (Fig. 7). When functioning fascicles are sacrificed and provide a necessary function, a spanning
nerve graft may then be applied. By applying these
techniques, good or excellent results may be achieved
in even large tumors, providing both pain relief and
functional preservation (Press et al., 1992; Kline and
Hudson, 1995).
51.3.2. Biopsy of neural lesions or tumors
Intraoperative monitoring may be employed to aid
safe selection of a fascicular biopsy when an
unknown lesion affects the brachial plexus. The use
of a portable stimulator or a hand-held fine probe
with sterile EMG needles is recommended. This
technique is similar to that described in the nerve
transfer selection. Here, the difference is that many
of the fascicles are nonfunctional due to the nature
of the underlying process.
In the past 5 years, we have performed fascicular
biopsy in more than 40 cases of brachial plexus
lesions (excluding obvious neurogenic tumors).
Extrapolated from techniques we have performed
727
Fig. 7. This 35-year-old woman presented to a head and neck surgeon with a painless neck mass. Open biopsy was done
without the benefit of preoperative imaging and was consistent with a schwannoma. The patient experienced new paresthesias after the biopsy, which was fortunately aborted once the extent and nature of the (brachial plexus) mass was discovered.
A: After the biopsy, T2-weighted MR image without fat saturation demonstrates a large dumbbell tumor extending out the
C7 neural foramen. The mass is displacing the thecal sac and has mass effect on the vertebral artery. It extends into the
supraclavicular fossa, displacing neighboring neural elements. Based on the appearance of this tumor, a two-stage procedure was planned and carried out. A posterior approach was first undertaken to decompress the cervical cord, resect the
posterior component of the tumor, and stabilize the cervical spine. The C7 nerve root was transected. Intraoperative electromyography (EMG) monitoring confirmed the overlap in the upper limb muscles by other neural elements and predicted
the lack of functional deficit with nerve resection. B: An anterior supraclavicular approach was then performed. The brachial plexus was then exposed and protected in vasoloops. The large mass had displaced the upper trunk (UT) and the lower
trunk (LT). At the distal pole, the entire cross-sectional area of the posterior division is involved in the tumor. C: The mass
and the involved middle trunk were resected. D: Resected specimen.
for other peripheral nerve lesions with indeterminate,

but severe and progressive neuropathies, we have
targeted fascicles localized by clinical examination
and/or high-resolution MRI imaging. In these cases,
we have had an 70% diagnostic rate. This type of
biopsy is particularly difficult as obtaining a fascicle
of sufficient length is challenging because of the
plexal interchange of individual fascicles. We have
successfully diagnosed and treated conditions such
as inflammatory lesions, vasculitis, perineurioma,
neurolymphoma, breast cancer (Fig. 8), and radiation
plexitis, among others, with this technique (Dyck
et al., 2003). Neurological downgrading was present
in 10% of patients, though some cases were
temporary. Paresthesias and neuropathic pain were

typically transient.
51.3.3. Tumors in the vicinity of nerves of the
brachial plexus
In selected cases, monitoring of the brachial plexus
during resection of tumors extrinsic to nerve may also
be helpful. It is helpful in certain cases of benign infiltrative processes such as desmoids tumors, malignant
processes, and especially in recurrent tumors (Fig. 9).
In these cases, scarring may impede definition of tissue planes. Electrophysiology facilitates identification,
mobilization, and protection of neighboring nerves
728
H. WANG ET AL.
Fig. 8. This 71-year-old woman presented with a 4-year history of painless weakness and numbness in the left upper limb.
Twenty-three years earlier, she had been diagnosed with breast cancer and had 14/17 positive lymph nodes. She underwent
left modified radical mastectomy and received chemotherapy. She also had a history of monoclonal gammopathy of undetermined significance (MGUS). On examination, she had a flail left shoulder and elbow with moderate weakness in the fingers and hand. She had percussion tenderness over the supraclavicular region radiating into C5 and C6 distributions.
Electromyography (EMG) showed evidence of a severe left brachial plexopathy, most severely affecting the upper trunk,
but involving all elements and miming conduction block in the lower trunk. Magnetic resonance imaging (MRI) and positron emission tomographic (PET) scans done at other institutions were thought to have been unchanged over the years.
A: Sagittal T1-weighted MR image with fat saturation post gadolinium shows enlarged nerves with peripheral enhancement
(arrow). B: PET scan shows increased uptake along the course of the brachial plexus. C: Exploration of the brachial plexus
was performed. A fascicular biopsy of a fibrotic upper trunk was undertaken. D: The resected fascicle (inset) demonstrated
metastatic adenocarcinoma, consistent with a primary breast cancer; estrogen and progesterone receptors were positive. She
received radiation therapy to the involved field followed by hormonal therapy and noted moderate improvement in function.
that might otherwise be difficult to identify or avoid.

Ultimately, this may enhance preservation of function
and extent of tumor resection. Similar monitoring
of nerves at risk is useful in other elective or reconstructive procedures as well (see peripheral nerve
Chapter 56).
729
Fig. 9. This 65-year-old woman presented with left shoulder and arm pain associated with an enlarging recurrent supraclavicular mass. She had undergone previous resection of a lipoma around the brachial plexus done through a partial claviculectomy. The mass initially was not encapsulated and was intertwined with the neural elements and subclavian artery.
A: Coronal T1-weighted image demonstrates a large encapsulated recurrent lipoma (*) displacing the brachial plexus and
the subclavian vessels. It extends between the scalene anterior and scalene middle. B: At reoperation, scarring was dense.
The mass was encountered early in the dissection. The neurovascular elements were identified, mobilized, and protected.
C: The mass could then be easily dissected. D: The supraclavicular and retroclavicular course of the brachial plexus and
subclavian artery (SA) is seen following tumor removal. The previous claviculectomy is seen. Despite intraoperative monitoring that was extremely helpful in the deep dissection, the patient experienced a neurapraxia of the musculocutaneous
nerve; elbow flexion was still strong through the preserved brachioradialis. At 1-year follow-up, she had complete relief
of her pain, was neurologically intact, and had no evidence of tumor recurrence on magnetic resonance imaging (MRI).
51.4. Conclusion
References
Although intraoperative monitoring is labor intensive, expensive, and time consuming, it provides critical additional information in brachial plexus lesions
that cannot be obtained by other methods or studies.
It is indispensable for posttraumatic reconstruction
for both diagnostic evaluation and surgical management. Electrophysiological monitoring provides a
greater margin of safety in tumor surgery that, at
least for us, makes it invaluable in our surgical
practice.
Balakrishnan, G and Kadadi, BK (2004) Clinical examination versus routine and paraspinal electromyographic
studies in predicting the site of lesion in brachial plexus
injury. J. Hand Surg. (Am.), 29(1): 140143.
J. Neurosurg., 98: 607610.
Carlstedt, T, Anand, P, Hallin, R, Misra, PV, Noren, G and
Seferlis, T (2000) Spinal nerve root repair and reimplantation of avulsed ventral roots into the spinal cord
730
after brachial plexus injury. J. Neurosurg., 93(Suppl. 2):
237247.
Carvalho, GA, Nikkhah, G, Matthies, C, Penkert, G and
Samii, M (1997) Diagnosis of root avulsions in traumatic brachial plexus injuries: value of computerized
tomography myelography and magnetic resonance
imaging. J. Neurosurg., 86(1): 6976.
Dyck, PJB, Amrami, KK, Spinner, RJ, Klein, CJ, Engelstad,
J and Dyck, PJ (2003) Fascicular biopsy of proximal
nerves in selected cases with MRI abnormality is diagnostically informative. J. Peripher. Nerv. Syst., 8: 14.
Ganju, A, Roosen, N, Kline, DG and Tiel, RL (2001) Outcomes in a consecutive series of 111 surgically treated
plexal tumors: a review of the experience at the Louisiana State University Health Sciences Center. J. Neurosurg., 95: 5160.
Hattori, Y, Doi, K, Dhawan, V, Ikeda, K, Kaneko, K and
Ohi, R (2004) Choline acetyltransferase activity and
evoked spinal cord potentials for diagnosis of brachial
plexus injury. J. Bone Joint Surg. Br., 86: 7073.
Holland, NR and Belzberg, AJ (1997) Intraoperative electrodiagnostic testing during cross-chest C7 nerve root transfer. Muscle Nerve, 20(7): 903905.
Huang, JH, Zaghloul, K and Zager, EL (2004) Surgical
management of brachial plexus region tumors. Surg.
Neurol., 61: 372378.
Kandenwein, JA, Kretschmer, T, Engelhardt, M, Richter,
HP and Antoniadis, G (2005) Surgical interventions
for traumatic lesions of the brachial plexus: a retrospective study of 134 cases. J. Neurosurg., 103: 614621.
Kline, DG and Happel, LT (1993) A quarters experience
with intraoperative nerve action potential recording.
Can. J. Neurol. Sci., 20: 310.
Kline, DG and Hudson, AR (1995) Nerve Injuries: Operative
Results for Major Nerve Injuries, Entrapments, and
Tumors. W.B. Saunders, Philadelphia.
Leechavengvongs, S, Witoonchart, K, Uerpairojkit, C and
Thuvasethakul, P (2003) Nerve transfer to deltoid muscle
using the nerve to the long head of the triceps, part II: a
report of 7 cases. J. Hand Surg., 28: 633638.
Mackinnon, SE, Novak, CB, Myckatyn, TM and Tung, TH
(2005) Results of reinnervation of the biceps and
H. WANG ET AL.
brachialis muscles with a double fascicular transfer for
elbow flexion. J. Hand Surg., 30: 978985.
Oberle, J, Antoniadis, G, Rath, SA, Seitz, K, Schneider, O,
Braun, V, Kahamba, JF and Richter, HP (1998) Radiological investigation and intra-operative evoked potentials for the diagnosis of nerve root avulsion:
evaluation of both modalities by intradural root
inspection. Acta Neurochir. (Wien), 140: 527531.
Oberle, J, Antoniadis, G, Kast, E and Richter, HP (2002)
Evaluation of traumatic cervical nerve root injuries by
intraoperative evoked potentials. Neurosurgery, 51:
11821190.
Oberlin, C, Beal, D, Leechavengvongs, S, Salon, A,
Dauge, MC and Sarcy, JJ (1994) Nerve transfer to
biceps muscle using a part of ulnar nerve for C5C6
avulsion of the brachial plexus: anatomical study and
report of four cases. J. Hand Surg. (Am.), 19:
232237.
Press, JM, Rayner, SL, Philip, M, Monga, TN and Katz, RT
(1992) Intraoperative monitoring of an unusual brachial
plexus tumor. Arch. Phys. Med. Rehabil., 73: 297299.
Shin, AY, Spinner, RJ, Steinmann, SP and Bishop, AT
(2005) Adult traumatic brachial plexus injuries. J. Am.
Acad. Orthop. Surg., 13(6): 382396.
Teboul, F, Kakkar, R, Ameur, N, Beaulieu, JY and Oberlin,
C (2004) Transfer of fascicles from the ulnar nerve to
the biceps in the treatment of upper brachial plexus
palsy. J. Bone Joint Surg. Am., 86: 14851490.
Tender, GC, Thomas, AJ, Thomas, N and Kline, DG
(2004) GilliattSumner hand revisited: a 25-year experience. Neurosurgery, 55: 883890.
Thomeer, RTWM, Malessy, MJA and Marani, E (2002)
Nerve root repair. J. Neurosurg., 96(Suppl. 1): 138139.
Surg., 99: 16321641.

M.R. Nuwer (Ed.)
731
CHAPTER 52

lumbosacral plexus surgery
Robert E. Minahan* and Allen S. Mandir
52.1. Introduction
This chapter explores neuromonitoring options for surgeries that place the lumbosacral plexus at risk. The
more specific techniques for intraoperative monitoring
of sacral and acetabular fractures are covered in separate chapters, so these will not be addressed here. In
addition, evaluation and monitoring of the brachial
plexus are also covered in a separate chapter, so techniques for peripheral nerve evaluation discussed there
will not be repeated. The remaining surgeries placing
the lumbosacral plexus at risk represent a relatively rare
monitoring situation. In addition, there is a paucity of
literature directly addressing the targeted group of monitored surgeries of the lumbosacral plexus.
Thus, the focus of this chapter will be on those
different monitoring techniques available, the
strengths and weaknesses of each of these techniques, and how they might be chosen and configured to optimally evaluate the lumbosacral plexus.
While the particular situation as applied to the lumbosacral plexus may be relatively rare, the need to
assess and configure the available neuromonitoring
tools to a rare or unique neuromonitoring situation
is not. To do this, the neurophysiologist must understand the anatomy at risk and the capability of the
monitoring tools available to him or her.
52.2. Anatomy
The lumbosacral plexus is divided into the lumbar
plexus and the sacral plexus. The lumbar plexus
(Fig. 1) arises from the first three lumbar roots and
Correspondence to: Robert E. Minahan, M.D., Department

of Neurology, Georgetown University, 3800 Reservoir
Road NW, Washington, DC 20007, USA.
Tel.: 1-202-444-8290; fax: 1-202-318-9146.
E-mail: rminahan@yahoo.com (R.E. Minahan).
the majority of the fourth lumbar root (anterior

divisions). Anterior root divisions innervate portions
of the lumbar plexus, supplying knee extensors and
anterolateral sensory branches. Posterior root divisions innervate hip adductors, posteromedial sensory,
and genital sensory areas. The lumbar plexus lies
underneath the psoas major muscle.
The sacral plexus (Fig. 2) is formed by the
remainder of the fourth lumbar root, the fifth lumbar
root, the first two sacral roots, and a portion of the
third sacral root. Anterior root divisions innervate
portions of the lumbar plexus, supplying foot and
toe dorsiflexors and anterior sensory branches. Posterior root divisions innervate hip extenders, knee
flexors, foot plantar flexors, and posterior sensory
areas. The lumbosacral trunk is formed from L4 to
L5 root contributions and runs along the medial edge
of the psoas muscle as it courses to join the main
portion of the sacral plexus lying posteriorly within
the pelvis.
As classically described above, the fourth lumbar
root contributes to both the lumbar and the sacral
plexuses; however, anatomic variation is common. In
a prefixed plexus, the third lumbar root contributes to
both the lumbar and sacral plexuses either with similar
contributions from the fourth root or with the fourth root
exclusively, supplying the sacral plexus. Similarly, a
plexus may be postfixed with the fifth lumbar root innervating both plexuses and the fourth root supplying only
the lumbar plexus. The nerve or nerves that supply both
plexuses is termed the nervus furcalis (Bannister et al.,
1995), and these variations will result in corresponding
alterations of the anatomy of each plexus.
Optimal monitoring depends on the specific areas
at risk in a given surgery and requires configuring
monitoring modalities appropriately. The distal muscles and nerves used in the different monitoring modalities described below should maximally evaluate
the areas at risk. When possible, both the anterior
732

From 12th thoracic
1st lumbar
Iliohypogastric
Ilioinguinal
2nd lumbar
Genitofemoral
3rd lumbar
Lat. femoral cutaneous
4th lumbar
To Psoas and
Iliacus
5th lumbar
Femoral
Accessory obturator
Obturator
Lumbosacral trunk
Fig. 1. The lumbar plexus (Gray, 1918).
and posterior divisions of the plexuses should be

assessed. For example, for a surgery that places the
lumbar plexus at risk, one might choose adductor
longus and/or gracilis to assess the posterior divisions
and psoas and/or quadriceps to assess the anterior
divisions. Table 1 outlines the most commonly
assessed components of the lumbosacral plexuses.
52.3. Electromyography
Electromyography (EMG) monitoring is a mainstay
of any surgery that places a peripheral nerve, plexus,
or root at risk. EMG monitoring of the lumbar or
sacral plexus can be used to assess all levels of these
plexuses if the appropriate muscles are chosen.
Table 1 lists the most common muscles used for this
purpose.
EMG monitoring of the lumbar and sacral
plexuses is done in much the same manner as
described for lumbosacral root monitoring in the
other chapters of this book and are not repeated here.
EMG activity occurs concomitantly with irritation and
the intensity of the EMG activity is a reasonable approximation of the degree of irritation. This, coupled with
the immediacy of EMG feedback, provides valuable

information at the time it is most needed. However,
EMG only reflects irritation and not injury or even
dysfunction of peripheral nerves, so attempts to
define sensitivity and specificity relative to eventual
nerve deficits are likely to yield poor results (Tonn
et al., 2000).
Since EMG only reflects irritation, it is expected
to be insensitive to certain compressive or stretch
insults if nerve ischemia precedes any neural insult
producing an EMG response. An instance of this
phenomenon is described in Section 52.4 case example. In addition, dysfunction of the nerve distal to the
site of surgery, neuromuscular junction dysfunction,
or muscle dysfunction may also limit or eliminate
the utility of EMG monitoring. Nerve, neuromuscular junction, or muscle dysfunction are not uncommon intraoperatively and may result from the
compression of nerve due to poor patient positioning,
excessive neuromuscular blockade or magnesium
administration, quadriplegic myopathy, or limb
ischemia.
In addition, even if an injury is initially irritative,
EMG silence may follow the initial irritation, and
733
4th Lumbar
5th Lumbar
1st Sacral
Superior gluleal
2nd Sacral
Inferior gluteal
Visceral br.
To Piriformis
3rd Sacral
Visceral br.
Sciatic
Comman
peroneal
4th Sacral
Tibial
Visceral br.
5th Sacral
To Quadratus femoris and

Inferior gemellus
To Obturator internus and
Superior gemellus
Post. fem. cutaneous
Perforating cutaneous
Coccygeal
Pudendal
To Levator ani, Coccygeus and
Sphineter ani externus
Fig. 2. The sacral plexus (Gray, 1918).
thus may be missed or seem insignificant. For

instance, sharp transection of nerves may yield no
or only brief EMG activity. An example is shown
below where the nerve to the levator scapulae muscle
was purposefully transected using surgical scissors.
A single EMG burst occurred lasting 300 ms,
and this was followed by EMG silence for
the remainder of the case. Such brief EMG activity
could completely be missed in some situations, if
accompanied by electrocautery or any other lapse in

EMG monitoring (Fig. 3).
While the limitations of EMG monitoring must be
recognized, EMG monitoring remains an important
part of peripheral nerve and nerve root monitoring
due to its sensitivity to irritative insults and the
immediacy of its findings. However, in light of these
limitations EMG will ideally be augmented with
other monitoring modalities as described below.
734
Table 1
Readily accessible muscles for monitoring the lumbar and sacral plexuses
Nerve (roots)
Division
Muscle
Lumbar plexus
Ilioinguinal/iliohypogastric (L1)
Obturator (L2, L3, and L4)
No. to iliopsoas (L2, L3, and ?L1)
Femoral (L2, L3, and L4)
Posterior
Posterior
Anterior
Anterior
Int. oblique abdominal muscles

Adductor magnus (upper), adductor longus, and gracilis
Psoas major
Quadriceps and sartorius
Sacral plexus
Superior gluteal (L4, L5, and S1)
Inferior gluteal (L5, S1, and S2)
Sciaticperoneal (L4, L5, and S1)
Sciatictibial (L5, S1, and S2)
No. to levator ani (S2, S3, and S4)
Anterior
Anterior
Anterior
Posterior
Posterior
Gluteus medius
Gluteus major
Tibialis anterior, peroneus longus, and others
Gastrocnemius, abductor hallucis, and others
Anal sphincter
52.4. Somatosensory evoked potentials

The primary method to assess the ongoing functionality of peripheral nerves is the somatosensory
evoked potential (SEP). Unlike EMG, SEPs are able
to interrogate the conductivity of proximal axons
mediating the volley of depolarization generated by
peripheral stimulation.
Case example: The utility of SEP monitoring is
shown below in the case of a 16-year-old patient
with history of pelvic obliquity and deformities
due to cloacal exstrophy. She underwent iliac
IOM RECORD
# 3 8 chn FREE RUN
osteotomies with removal of pelvic exostosis.

Neuromonitoring was performed with bilateral tibial
and peroneal nerve SEPs and bilateral tibialis anterior and gastrocnemius muscle EMG. Multiple
attempts were made to realign the pelvis into normal
anatomic position with loss of SEPs upon each
attempt. Of note, the monitored EMG was monitored
continuously through this time and remained quiet
throughout this portion of the procedure. A partial
correction was eventually performed with retained
signals and no neural deficits were present on
awakening (Fig. 4).
Cutting nerves to. .
Level or scapuli
18:35:38
SWITCH: STOP AVG: STOP

TD1: None
TD2: None
T
R
C
1
Off
Off
200 ms Free
100 v Amp 6
Off
Off
Fig. 3. Sharp transection of the nerve to levator scapulae results in a single burst of electromyography (EMG) activity lasting 300 ms and is followed by EMG silence.
735
Left tibial nerve SEP

Cerv5-Fpz
or at the level of the inguinal canal (DeLisa et al.,

1994). In addition, the lateral femoral cutaneous nerve
(DeLisa et al., 1994; Yamada et al., 1996; Seror, 2004)
and iliohypogastric (Rabie and Drory, 2005) can be stimulated peripherally and SEPs may be generated,
albeit responses are unlikely to be robust intraoperatively. Thus, careful evaluation at baseline would be
needed to determine if the signals generated would
be adequate for subsequent clinical judgments.
The posterior tibial nerve somatosensory projections are mediated by posterior divisions of the L5,
S1, and S2 roots, while the peroneal nerve projections are mediated by the L4, L5, and S1 anterior
divisions. The combination of these SEPs nicely
evaluates most of the sacral plexus. A proximal femoral nerve SEP will evaluate the anterior divisions of
the L2, L3, and L4 roots, while a saphenous nerve
SEP will be primarily mediated by the L4 root.
Most of the commonly used stimulation sites generate SEPs that are mediated by multiple elements of
the lumbosacral plexus. As a result, SEPs may have
imperfect sensitivity to discreet insults to the plexus
in a manner similar to its imperfect sensitivity to
individual root dysfunction (Aminoff, 2002). Despite
these limitations, SEP monitoring is recommended.
Baseline
Pelvic
correction
#1 14:06
Pelvic
correction
#2 14:22
Pelvic
correction
#3 15:07
52.5. Motor nerve conduction studies
1 V
5 ms
Fig. 4. Three attempts are made to correct a pelvic obliquity with presumed stretch of the sacral plexus or sciatic
nerve. With each attempt the somatosensory evoked potential (SEP) signal is lost (arrows) and then returns with
relaxation of the pelvis.
A number of peripheral nerves that project

through the lumbar or sacral plexus are accessible
for stimulation. The posterior tibial nerve and peroneal nerves are commonly used for neuromonitoring
of many types of procedures and remain excellent
choices for evaluation of the sacral plexus. These
more standard SEP nerves do not assess the lumbar
plexus. Femoral nerve stimulation can be used for
this purpose. The femoral nerve can be activated either
in the leg (saphenous nerve) (DeLisa et al., 1994)
Motor nerve conductions studies (MNCS) may be

the most valuable form of monitoring the lumbar
and sacral plexuses when electrical stimulation can
be achieved at proximal plexus elements. In this case,
electrical stimulation can be done specifically upon
neural elements of interest and the motor fibers distal
to this point can be assessed by recording the resultant compound motor action potential (CMAP). The
value of this technique lies in the ability to assess
more specific portions of the plexus, in the negligible
time it takes to acquire a response, and in the robust
and stable nature of CMAP responses.
The interpretation of MNCS is reasonably
straightforward because the ratio of an intraoperative
CMAP amplitude to a postexposure baseline is a reasonable approximation of the proportion of persisting
motor nerve fibers (Pugh et al., 1984). However, for
this relationship to hold true, a number of criteria
must be fulfilled. First, the level of neuromuscular
blockade must be held constant or a correction factor
must be used based on peripherally elicited CMAPs.
Second, stimulation must be supramaximal for the
736
target neural element. Third, stimulation must be

specific for the target neural element. That is, nearby
nerves are not being partially stimulated in addition
to the one of interest. Fourth, temperature should
not vary dramatically (Kimura, 1989). Finally, the
time at which to select a representative baseline
response must be chosen after full equilibration of
anesthesia. This is important, as awake CMAP amplitudes exceed those under anesthesia by 25%, and
setting baselines too early will yield spurious results
(Edmonds et al., 1988).
If proximal stimulation electrodes cannot be positioned and left securely in place, the surgeon may
wish to stimulate plexus elements using a hand-held
stimulator. In this case, the advantage of ongoing
monitoring is lost and stimulation may be less consistent from trial to trial, so direct CMAP comparisons
to baseline may be less reliable. A rough consistency
from trial to trial would still be expected, and the
presence of a CMAP implies at least partial neural
integrity distal to the point of stimulation. Likewise,
this stimulation method may be used to map or
identify plexus elements.
Finally, a similar technique with proximal electrodes used for recording rather than for stimulation
can be used to record mixed nerve conduction studies
where SEP stimulating electrodes are used for the
stimulus. This technique is likely to be susceptible
to small movements of recording electrodes; signals
are likely less robust and would require more time
than testing MNCS. However, these still may prove
useful in certain situations.
When reliable proximal stimulation is possible,
MNCS are recommended.
52.6. Motor evoked potentials
The use of motor evoked potentials (MEPs) for monitoring surgeries of peripheral nerves is largely
untested, though they may have a role in evaluation
of brachial plexus surgeries (Turkof et al., 1997;
Burkholder et al., 2003) and a similar role may be
present in the lumbar and sacral plexus levels.
Satcher et al. (2003) have reported encouraging
results in the closely related situation of monitoring
in cases of acetabular fracture repair; however, even
these authors state that they do not routinely employ
MEPs for these surgeries.
It is not clear, in a more general sense, that the use
of MEPs to evaluate mixed sensorimotor peripheral
nerves adds significant value when these same nerves
are already assessed using SEPs. Like SEPs, MEPs

are in general mediated by multiple nerve roots and
thus the sensitivity of MEPs to discreet plexus injuries may be similar to SEPs (Tsutsui et al., 2003).
Despite this, it is possible that MEPs may add some
additional sensitivity to monitoring these and other
surgeries of peripheral nerves. The key will be
whether the degree of increased sensitivity is outweighed by the negative aspects of the more restrictive anesthetic regimens associated with MEPs
(Calancie et al., 1991; Kawaguchi et al., 1998; Ubags
et al., 1998; Chen, 2004), slightly greater complication rates with MEPs (MacDonald, 2002), and a risk
of false positive findings in these more variable
signals may negate any benefit of marginally
increased sensitivity. Currently available data do not
adequately address this risk/benefit balance.
52.7. Hoffman reflex
The Hoffman reflex (H-reflex) is a monosynaptic
reflex most commonly obtained using stimulation of
the tibial nerve at the popliteal fossa and recording
from the gastrocnemius/soleus muscles (Fig. 5).
Specific techniques can be found in most standard
textbooks detailing nerve conduction studies
(Kimura, 1989; DeLisa et al., 1994). Under appropriate conditions, almost any nervemuscle pair can be
used to obtain an H-reflex, but only the gastrocnemius/soleus muscle response is reliably obtained in
the lower limbs of normal patients (Kimura, 1989).
The H-reflex is obtained with submaximal stimulation and diminishes in amplitude closer to supramaximal stimulus levels (Fig. 5) (Kimura, 1989). A loss
of >50% of the baseline amplitude can be used as
an indication of possible neural dysfunction
(Leis et al., 1996). However, the identification of
amplitude loss can only be usefully done under stable
anesthetic conditions as H-reflexes are sensitive
to standard volatile anesthetics (Zhou et al., 1998;
Leppanen, 2004).
Both the afferent and efferent limbs of the tibial
nerve to gastrocnemius response traverse the sacral
plexus at the S1 root level. When S1 projections of
the sacral plexus are placed at risk, H-reflex monitoring is recommended.
52.8. F-waves
F-waves may be elicited in any muscle following a supramaximal stimulation of its associated motor nerve.
737
52.9. Monitoring recommendations

100 ms
H-reflex
M response
5 mV
Stimulation: 48 mA, 1 ms
Timebase:10 ms/division
100 ms
H-reflex
M response
5 mV
100 ms
H-reflex
5 mV
M response
Fig. 5. H-reflexes are shown at various stimulation levels.

Note the maximal response at submaximal stimulation
levels and waning of the response as more motor
axons are directly activated by the external electrical
stimulus.
We suggest EMG and SEP monitoring of surgeries

that place the lumbosacral plexus at risk. The specific muscles and nerve stimulation sites chosen
depend on which elements of the plexus are at risk.
In addition, when proximal stimulation is possible
within the surgical field, motor nerve conduction
studies will be a valuable addition due to their
rapidity of acquisition and relative specificity to
the plexus element stimulated. Finally, H-reflexes
may be valuable when S1 root plexus elements are
at risk.
Motor evoked potentials and F-wave monitoring
may be helpful additions to this monitoring regimen,
but further investigation is needed to define whether
MEPs add sensitivity to the regimen and to define
the behavior and action criteria for F-waves in this
setting. We consider these two modalities as optional
additions to the monitoring regimen.
Our recommendations for monitoring of the lumbosacral plexus are largely based on the anatomy
and the current standard practices for monitoring at
the level of nerve root. Further refinement of these
recommendations is likely as more directly applicable research becomes available.
References
The F-wave is a late response resulting from a portion of

the antidromic motor depolarization resulting in reflection at the anterior horn cell of the spinal cord and
subsequent orthodromic motor depolarization which
can be recorded at the muscle. This response is purely
mediated by motor axons, and the amplitude is much less
than that of the directly elicited CMAP as only a small
proportion of motor nerve fibers are active in producing
the F-wave. This response is highly variable in amplitude, latency, and morphology. F-waves are not elicited
following every stimulation.
Although F-waves potentially allow rapid
acquisition and feedback, their intrinsic variability
may reduce their usefulness or at least make warning
criteria more complicated. Besides variation with
the level of neuromuscular blockade, F-waves may
be depressed by inhalational anesthetics (Rampil
and King, 1996; Zhou et al., 1998; Zhou and Zhu,
2000; Baars et al., 2005). Further study is needed
to more fully define F-wave usefulness within this
setting.
Aminoff, MJ (2002) Electrophysiological evaluation of

root and spinal cord disease. Semin. Neurol., 22(2):
197204.
Baars, JH, Kalisch, D, Herold, KF, Hadzidiakos, DA and
Rehberg, B (2005) Concentration-dependent suppression of F-waves by sevoflurane does not predict immobility to painful stimuli in humans. Br. J. Anaesth., 95
(6): 789797.
Bannister, LH, Gray, H and Williams, PL (1995) Grays
Anatomy. Churchill Livingstone, New York, NY, 38th ed.
J. Neurosurg., 98(3): 607610.
Calancie, B, Klose, KJ, Baier, S and Green, BA (1991)
Isoflurane-induced attenuation of motor evoked potentials caused by electrical motor cortex stimulation during surgery. J. Neurosurg., 74(6): 897904.
spinal surgery. J. Clin. Monit. Comput., 18(4): 303308.
DeLisa, J, Lee, H, Baran, E, Lai, K, Spielholz, N and
MacKenzie, K (1994) Somatosensory evoked potentials.
In: Manual of Nerve Conduction Velocity and Clinical
738
Neurophysiology, Raven Press, New York, NY, 3rd ed.,
pp. 195274.
Edmonds, HL, Paloheimo, M and Wauquier, A (1988)
Computerized EMG Monitoring. Applications in Anesthesia and Intensive Care. Malherbe Publishing Co.,
Schoutlaan, The Netherlands.
Gray, H (1918) Grays Anatomy of the Human Body. Lea
and Febiger, Philadelphia.
Kawaguchi, M, Inoue, S, Kakimoto, M, Kitaguchi, K, Furuya,
H, Morimoto, T and Sakaki, T (1998) The effect of
sevoflurane on myogenic motor-evoked potentials induced
by single and paired transcranial electrical stimulation of
the motor cortex during nitrous oxide/ketamine/fentanyl
Kimura, J (1989) Electrodiagnosis in Diseases of Nerve
and Muscle: Principles and Practice. F.A. Davis, Philadelphia, PA, 2nd ed.
Leis, AA, Zhou, HH, Mehta, M, Harkey, HL, III and
Paske, WC (1996) Behavior of the H-reflex in humans
following mechanical perturbation or injury to rostral
spinal cord. Muscle Nerve, 19(11): 13731382.
Leppanen, RE (2004) From the electrodiagnostics lab: where
transcranial stimulation, H-reflexes and F-responses monitor cord function intraoperatively. Spine J., 4(5): 601603.
MacDonald, DB (2002) Safety of intraoperative transcranial
electrical stimulation motor evoked potential monitoring.
Pugh, ND, Kay, B and Healy, TE (1984) Electromyography in anaesthesia. A comparison between two
methods. Anaesthesia, 39(6): 574577.
Rabie, M and Drory, VE (2005) A test for the evaluation of
the lateral cutaneous branch of the iliohypogastric nerve
using somatosensory evoked potentials. J. Neurol. Sci.,
238(12): 5963.
Rampil, IJ and King, BS (1996) Volatile anesthetics depress
spinal motor neurons. Anesthesiology, 85(1): 129134.
Satcher, RL, Noss, RS, Yingling, CD, Ressler, J and Ries,
M (2003) The use of motor-evoked potentials to

monitor sciatic nerve status during revision total hip
arthroplasty. J. Arthroplasty, 18(3): 329332.
Seror, P (2004) Somatosensory evoked potentials for the
electrodiagnosis of meralgia paresthetica. Muscle
Nerve, 29(2): 309312.
Tonn, JC, Schlake, HP, Goldbrunner, R, Milewski, C,
Helms, J and Roosen, K (2000) Acoustic neuroma surgery as an interdisciplinary approach: a neurosurgical
series of 508 patients. J. Neurol. Neurosurg. Psychiatry,
69(2): 161166.
Tsutsui, S, Tamaki, T, Yamada, H, Iwasaki, H and Takami,
M (2003) Relationships between the changes in compound muscle action potentials and selective injuries
to the spinal cord and spinal nerve roots. Clin. Neurophysiol., 114(8): 14311436.
Surg., 99(6): 16321641.
Ubags, LH, Kalkman, CJ and Been, HD (1998) Influence
of isoflurane on myogenic motor evoked potentials
to single and multiple transcranial stimuli during
nitrous oxide/opioid anesthesia. Neurosurgery,
43(1): 9094.
Yamada, T, Matsubara, M, Shiraishi, G, Yeh, M and Kawasaki, M (1996) Topographic analyses of somatosensory
evoked potentials following stimulation of tibial, sural
and lateral femoral cutaneous nerves. Electroencephalogr. Clin. Neurophysiol., 100(1): 3343.
Zhou, HH, Jin, TT, Qin, B and Turndorf, H (1998) Suppression of spinal cord motoneuron excitability correlates with surgical immobility during isoflurane

M.R. Nuwer (Ed.)
739
CHAPTER 53
Monitoring of neural structures involved

in micturition and sexual function
Zoran Rodi and David B. Vodusek*
Division of Neurology, Institute of Clinical Neurophysiology, University Medical Centre, SI-1525 Ljubljana, Slovenia
53.1. Introduction
A whole array of clinical neurophysiological methods has been modified for use in the anogenital area,
including electromyographic methods, reflex, and
evoked potential studies. These methods are already
routinely employed in some uroneurological/neurourological laboratories for diagnostic and follow-up
purposes in patients with (suspected) neurogenic
sacral dysfunction (Vodusek and Fowler, 2004).
Some of these neurophysiological methods have been
introduced in the operating room, to help the surgeon
identify particular sacral nervous structures, and/or
monitor the function of the sacral neuromuscular system during surgery. Following is the relevant functional
anatomy of ano-genito-perineal region, basic technical
aspects of stimulation and recording, and the overwiev
of neurophysiological and other physiological techniques introduced in intraoperative monitoring (IOM)
of neural structures involved in micturition, defecation
and sexual functions.
53.2. Functional anatomy
Functional anatomy of the genito-urinary and anogenital systems is highly complex but needs not be considered in detail as only the gross anatomy of the relevant
somatic nervous structures can be approached by those
available clinical neurophysiological methods that are
also applicable in the operating room.
Afferent fibers from the mucosa and skin from
the genito-perineal region mostly travel with the
pudendal nerves. The distally most accessible group
*
Correspondence to: Prof. David B. Vodusek, M.D., Ph.D.,

Division of Neurology, Institute of Clinical Neurophysiology, University Medical Centre, Zaloska 7, SI-1525
Ljubljana, Slovenia.
Tel.: 386-41673050; fax: 386-15222293.
E-mail: david.vodusek@kclj.si (D.B. Vodusek).
of sensory fibers is the dorsal nerves of penis (or clitoris). The sensory fibers from the genital, perineal,
and anal region enter through the dorsal spinal roots
S2S5 into the spinal cord, and synapse (through
interneurons) with sphincteric motor neurons. The
afferent information also ascends (the primary sensory neurons synapsing to higher order sensory
neurons at various levels) via the spinothalamic and
dorsal column tracts, the lemniscal system and thalamocortical tracts finally to the somatosensory cortex
at its interhemispheric location (Corcos, 2004;
Morrison et al., 2005).
The sphincteric lower motor neurons in the midventral spinal gray matter of the second to fourth
sacral spinal cord segments (the Onuf nucleus)
are under voluntary control from the motor cortex.
Somatic motor nerve fibers leave through the ventral
roots and the sacral plexus, combining into the
pudendal nerves; apart from that, direct branches
innervate the levator ani and the anal sphincter (see
Fig. 1 in Chapter 29) (Morrison et al., 2005).
53.3. Basic technical aspects of stimulation
In order to obtain bioelectrical signals useful for
monitoring purposes in the different segments of the
sacral neuromuscular system, it is necessary to depolarize the nervous system at particular segments.
Up to now only electrical stimulation has been appropriate for this purpose. Stimulation can be applied
either to the sensory part or to the motor part of the
system. At present, most IOM of the sacral system
has relied on responses evoked on stimulation of
the sensory system, apart from recording of anal
sphincter muscle responses on stimulation of ventral
spinal roots, or on stimulation of motor cortex (over
the scalp).
The appropriate peripheral sensory structures that
are available for stimulating purposes are the two
Z. RODI AND D.B. VODUSEK
740
+
G1R
G2R
G1L
G2L
G1R
G1L
G2
Fig. 1. Placement of superficial electrodes for stimulation

of dorsal nerve of penis/clitoris, and intramuscular wire
electrodes in anal sphincter for recording of bulbocavernosus reflex, or motor responses to stimulation of ventral
roots or motor areas of brain (MEP). Note that recording
may be bifocal (left: both electrodes in same hemisphincter
muscle) or monofocal (i.e., one electrode outside the muscle, at the neutral subcutaneous position). Monofocal
recording yields responses with higher amplitudes.
adjacent dorsal penile or clitoral nerves. These nerves

are stimulated by silver/silver chloride electrode
placed on the proximal dorsal surface of the penis
or clitoris (this electrode representing the cathode).
The other electrode (anode) is placed either distally
on the penis (12 cm apart from the proximal electrode) or on the adjacent labia (Fig. 1, see also
Fig. 4 in Chapter 29). Alternatively, especially for
male children, due to the short length of the penis,
the cathode is placed on the dorsal aspect and the
anode on the ventral aspect of the penis. The skin
must be prepared by cleaning and scrubbing gently
before placing electrodes in order to assure the current penetrating the skin and to avoid the stimulus
artifacts. The electrodes are secured appropriately

and then bandaged with a few layers of gauze in
order to prevent them from being displaced when
the patient is moved onto the operating table.
Electrode impedances should be kept below 5 kO.
As a general rule stimuli of 20 mA (up to 50 mA)
intensity at 0.2 ms duration are delivered in procedures involving stimulation on penis or clitoris (at
various frequencies for different measurements,
preferably 24 Hz, and up to 13.3 Hz).
Stimulation of ventral roots may be transcutaneous with the superficial electrodes placed over the
spine; cathode is placed at T12 and anode 20 cm caudally. Stimulation can also be performed directly at
the spinal roots, where hand-held sterile hook electrodes are placed under appropriate roots (or rootlets,
after the root is freed from neighboring roots and
lifted outside the spinal canal). The cathode hook
electrode should be placed at a position closer to
recording electrodes. Square wave pulses of
12 mA strength, 0.2 ms duration are delivered (at
stimulating rates up to 13.3 Hz) (Fig. 3).
Transcranial electrical stimulation is applied to
scalp using screw type electrodes at positions 2 cm
frontally to C3 and C4 points of the international
1020 electromyography (EEG) system of scalp electrode placement. In this case, anode may prove to be
better at one or the other side. Alternative placement
of stimulating electrodes is at points Fz (anode) and
6 cm dorsal.
53.4. Basic technical aspects of recording
Intraoperatively, bioelectrical activity from the
sacral neuromuscular system has up to now been
Fig. 2. Sterile hook wire recording electrodes are introduced into the muscle with sterile needles; these are immediately
carefully removed from the muscle the hooked wires remaining in place.
741
Fig. 3. Stimulation of ventral spinal roots and recording of anal sphincter motor responses. Note that the stimulating/recording procedure is side selective.
recorded from the dorsal spinal roots, the spinal

cord, the somatosensory cortex and the anal sphincter muscle.
Recordings from dorsal spinal roots [dorsal root
action potentials (DRAPs)] are obtained on stimulation of the dorsal nerves of penis or clitoris after
the root is freed from neighboring roots and lifted
outside the spinal canal by hand-held sterile bipolar
hook electrodes (see Fig. 4 in Chapter 29). In this
case, the electrode closer to point of stimulation is
the G1 (active) electrode. Epoch lengths of 50 ms
are used in these recordings.
Again, after penile or clitoral electrical stimulation, recordings of pudendal spinal somatosensory
evoked potentials (SEP) are obtained by spinal epidural electrodes inserted percutaneously in spinal canal
close to the spinal cord (in principle placed both
below and above the lesion). Directly, in the surgical
field, placed subdural electrodes can also be used.
Typically 100 responses are averaged together; epoch
lengths of 50 ms are used.
To obtain pudendal cerebral (cortical) somatosensory evoked potentials (CSEP) recording electrodes
are of the screw type, inserted in the scalp at
742
Iso = 0%
Iso = 0.8%
11:58
1 min
2 min
Iso = 1.25%
8 min
5 min
15 min
10 min
22 min
16 min
Iso = off
12:54
75 V
10 ms
100 V
20 ms
100 V
20 ms
Fig. 4. Influence of isoflurane (A), of bolus of 0.3 mg/kg muscle relaxant (B), and of bolus of propofol (C; 1.9 mg/kg) on
the bulbocavernosus reflex.
Cz2 cm (G1 or active electrode) and Fz (G2 or

reference electrode), according to the somatotopic
representation in the primary somatosensory cortex,
and related to the International 1020 EEG system
of scalp electrode placement. The active electrode
is placed in the midline as the sacral segments are
represented deep within the medial longitudinal
interhemispheric fissure. For CSEP, typically 100
200 traces are averaged together. Epoch lengths of
200 ms are used.
To record anal sphincter (EMG) responses either
surface type electrodes or hooked wire electrodes can
be used (Fig. 2). Patient should be free of muscle relaxant. Because of the close anatomical relationship
between the small sphincter muscle and neighboring
larger muscle groups (e.g., gluteal muscles inervated
by the lower lumbar and upper sacral segments)
recording must be selective (e.g., with intramuscular
hook wire electrodes) if the stimulation technique is

nonselective (such as in the case of ventral root stimulation or transcranial stimulation, as a consequence of
which neighboring muscles are also excited). In the
case when the stimulation procedure is more specific
[e.g., in the bulbocavernosus reflex response (BCR)
monitoring, where penile or clitoral stimulation is
applied], the recording may be obtained with properly
attached surface type electrodes.
Sterile hook wire recording electrodes are introduced into the left and right sides of the external anal
sphincter with sterile needles; these are immediately
carefully removed from the sphincter the hooked
wires remaining in place. The integrity of the electrodes and the proper intramuscular position can be
tested by passing a short train of 50 Hz current at
10 mA and observing sphincter contraction (even if
the patient has been given a muscle relaxant at the
time of electrode placement). The electrode impedances of these electrodes should be checked, though
clean recordings are usually still possible with highelectrode impedances.
When detecting with wire electrodes from the anal
sphincter, different montages may be used. Of a pair
of electrodes, the first (G1) may be placed in the
right and the second (G2) in the left hemisphincter,
or vice versa. With this montage it is not possible
to distinguish between body sides, and some cancellation of the signal may be expected on the differential amplifier. Another possibility is to insert a pair of
electrodes (G1 and G2) in each hemisphincter (i.e.,
bifocal detection), thus enabling the right-to-left differentiation but obtaining lower amplitudes of
responses due to the cancellation of the signal. Third
possibility is to use a separate active (G1) electrode
for each hemisphincter with the common reference
(G2) inserted subcutaneously over the coccyx (i.e.,
monofocal detection), thus differentiating the right
and the left hemisphincter with practically no signal
cancellation (cf. Fig. 1).
The epoch length used for anal sphincter EMG
recordings varies according to the type of response,
and is commonly either 50 or 100 ms. Either single
or few averaged responses can be obtained on stimulation but note that the patient should not be pharmacologically paralyzed during the recording procedure.
53.5. Specific sacral neuromuscular system
monitoring procedures
The techniques may be divided into those serving for
mapping, that is identifying the functional nervous
tissue, and distinguishing it from nonfunctional tissue
(e.g., filum terminale, scar tissue, and tumors), and
those serving for monitoring, that is continuously
checking for the functional integrity. Further, the
techniques may be divided into three groups, namely,
into those that test either motor or sensory structures,
or reflex arch. There may also be division of
techniques into strictly electrophysiological, and
those which include recording of other physiological
responses (e.g., intraorgan pressure and penile girth).
743
Technique employs an exposure of the cauda euqina

through a laminectomy (e.g., T12S2). The sacral
roots are identified by bone anatomy. The dorsal roots
are separated from ventral ones and lifted outside by
hand-held sterile bipolar hook electrodes utilized to
record the traveling DRAPs. The DRAPs in turn are
evoked by electrical stimulation of the penile or clitoral nerves. Commonly, 100 responses are averaged
together. Each average response is repeated to assess
its reliability. In standard setting, afferent activity
from the right and left dorsal roots of S1, S2, and S3
are recorded, occasionally also from the S4/5 dorsal
roots. Recording of DRAPs (see Figs. 4 and 6 in
Chapter 29) was reported as technically successful in
105 of 114 patients (92%) (Huang et al., 1997).
S1 roots contributed 4%, S2 roots 60.5%, and S3
roots 35.5% of the overall pudendal afferent activity,
which was in majority significantly asymmetrically
distributed (56%). In 18% of the patients, the pudendal afferent activity was confined to a single level,
and in 7.6% even to a single root.
A variation of the DRAPs was also performed
with the action potentials recorded after electrical
stimulation of the anal canal, thus representing an
anal afferent activity. Interesting is a conclusion that
penile clitoral afferents are carried mainly within the
S2 roots, and minor part within S3 roots, while it is
almost exactly the opposite with the anal mucosal
afferents (Deletis et al., 2000).
53.5.2. Pudendal spinal somatosensory evoked
responses
Pudendal spinal SEP is the possible way of monitoring
the peripheral part of sensory pathways. Recordings
were up to now performed in 4 children of both sexes,
2.57 years old, along with SEPs obtained from stimulation of individual tibial nerves at the ankle. Small
amplitude SEP (up to 2.5 mV) could be recorded with
subdurally placed electrodes both over the conus
region and over the thoracic spinal cord. The latencies
of the spinal SEP over the conus region ranged from 6
to 10 ms. The recordings were made as a pilot study,
and thus far only demonstrated the possibility to obtain
such recordings intraoperatively.
53.5.1. Pudendal dorsal root action potentials

Pudendal DRAPs is a sensory mapping technique. It
serves one purpose, and that is to define the amount
of afferent activity (elicited by penile/clitoral electrical stimulation) in individual sacral dorsal roots.
53.5.3. Pudendal cerebral (cortical) somatosensory

evoked potentials
Pudendal cerebral SEP monitors all length of somatosensory pathway from the peripheral nerve to the
744
primary somatosensory cortex, and in itself offers no

conclusion as to whether the harm was done to
peripheral or central nerves. Well-formed cerebral
SEP with amplitude of 0.50.7 mV could be recorded
on dorsal penile nerve stimulation throughout spinal
neurosurgical procedures in 2 adult male patients,
50 and 78 years old; stable P40 peaks were obtained.
The recordings were made as a pilot study, and thus
far only demonstrated the possibility to obtain such
recordings intraoperatively.
53.5.4. Distinguishing motor roots from filum
terminale
Release of tethered spinal cord by sectioning of the
filum terminale carries a risk of injuring neighboring
motor and sensory nerve roots involved in bowel and
bladder control. Therefore, intraoperative neurophysiological techniques have been developed to prevent
neurological complications postoperatively (Von Koch
et al., 2002). During the procedure ventral nerve roots
are identified and their electrical thresholds (i.e., for
eliciting the muscle response) obtained. Authors
reported that the mean stimulus intensity was 0.32 V,
and the range 0.051.0 V. These values were compared with electrical thresholds obtained by stimulation of the filum terminale. The mean was 26.1 V,
and the range 8100 V. In over 70% of patients, muscle activation via the filum required 100 times the
voltage needed to activate a motor root. A motor root
to filum threshold ratio of 1:100 was useful in identifying the filum.
53.5.5. Sphincter motor response monitoring
This may be done in two ways. One way to do it is
in the form of continuous EMG. The needle or wire
electrodes are inserted into muscle, tipically anal or
urethral sphincter, and interfaced to a loudspeaker to
detect any mechanical irritation associated with surgical dissection (Krassioukov et al., 2004; QuignonesHinojosa et al., 2004; Paradiso et al., 2005).
In effort to spare the cauda equina muscle responses
may be obtained to electrical stimulation of ventral
roots (James et al., 1979). Our experience with this
method is based on results obtained in 10 children of
both sexes, 2.59 years old. Anal sphincter muscle
EMG responses were recorded by stimulation of the
ventral spinal roots (L5, S1, S2, S3, and S4) to identify
the ventral roots carrying motor fibers to sphincter muscle. Recordings were obtained by surface conductive
rubber electrodes (applied para-anally), and intramuscular hooked wire electrodes (Fig. 3). In surface recordings, no unilateral responses could be identified and
responses were also obtained on stimulation of the
L5 and S1 roots. On the other hand, on stimulation of
L5 and S1 roots no adequate responses could be discerned from simultaneous recordings from intramuscular electrodes. The surface recorded responses were
recognized as nonspecific, derived from neighboring
muscles, most probably glutei (Vodusek and Zidar,
1988). The latency of surface recorded responses was
as a rule shorter than the latency of responses obtained
from intramuscular electrodes, which was between
5 and 8 ms.
In few patients, motor evoked potentials (MEP) of
anal sphincter have been tested. A train of 4 or 5 stimuli was applied to the scalp through screw electrodes,
and responses were recorded with wire electrodes.
Overall, a reproducible but low-amplitude muscle response was obtained, and stimulation intensity needed
was rather high, higher than that needed for MEP in
lower limbs.
53.5.6. Bulbocavernosus reflex monitoring
Technique employs recording in anal sphincter via
needle or wire electrodes, and penile/clitoral electrical stimulation (Deletis and Vodusek, 1997; Morota
and Nakagawa, 1998; Rodi and Vodusek, 2001). As
it is common with the IOM, patients are anesthetized
with propophol and phentanyl, and a short-acting
muscle relaxant is used at the time of induction of
anesthesia (Rodi and Vodusek, 2001). Often, nitrous
oxide is added (Deletis and Vodusek, 1997; Morota
and Nakagawa, 1998). Mostly, single electrical stimulus is not enough to elicit BCR, and double pulse
stimulus is used or even trains of up to 5 stimuli,
and with the interstimulus interval of 3 or 4 ms
(Deletis and Vodusek, 1997; Rodi and Vodusek,
2001).
In the first study (Deletis and Vodusek, 1997),
intraoperative recordings were performed in 119 neurosurgical patients. Clinically most patients had mild-tomoderate upper motor neuron deficit in the lower extremities. No patient was completely incontinent, but some
of them had minor urinary problems. The BCR was
recorded with the wire electrodes in anal sphincter muscle, and elicited through silver/silver chloride cup electrodes. Up to 20 single responses were averaged, the
optimum being averaging 4 consecutive responses.
The BCR was reliably recorded without habituation,
745
as has also been reported in awake subjects (Vodusek

et al., 1983). Interstimulus intervals of 2, 3, 4, and
5 ms and stimulation rates from 0.4 to 4.3 Hz were
tested. Optimal stimulating parameters were found to
be double pulses with an interstimulus interval of
3 ms, stimulating rate of 2.3 Hz, and intensity of
20 mA. With these parameters, it was possible to record
the BCR intraoperatively in all patients. Isoflurane and
nitrous oxide significantly suppressed the BCR, and
muscle relaxant completely abolished it (Fig. 4).
In the second study (Rodi and Vodusek, 2001),
intraoperative BCR was recorded in 65 patients,
53 men and 12 women, who underwent surgery for
lower thoracic and lumbar spinal trauma, with no
neurological deficit before or after surgery. Recording was with wire electrodes either bifocal or monofocal. No averaging was used. A single stimulus
elicited the BCR in 11/20 patients, pairs in 15/20,
trains of 3 in 17/20, and trains of 4 and 5 in 20/20
of patients. Furthermore, adding pulses to the stimulus added considerably to the amplitude of the
responses, appearently no difference being observed
between trains of 4 and 5 pulses (Fig. 5). Median
BCR amplitude with bifocal recording was 110 mV,
and with monofocal recording 200 mV. Interside
amplitude difference was up to 20-fold. With blind
preoperative placement of the detecting electrodes
2.5-cm deep to the skin, the BCR was detected in
13% of hemisphincters in women, and in 81% of
hemisphincters in men. In a separate group of 18 male
patients, intramuscular position of detecting electrodes was checked by passing through electrodes
tetanic current of 10 mA; muscle twitch was
observed in 35 of 36 hemisphincters (97%), and the
BCR was detected accordingly. Extended and not
yet published material of this study showed, that
there was late pronounced, and sustained habituation,
Number of
pulses
1000
53.5.7. Intraoperative recording of anal and

urethral sphincter pressure, and intravesical,
intrarectal, and intracavernosal pressure
Unrecognized nerve injury is believed to be the likely
cause of erectile dysfunction after radical prostatectomy (Rehman et al., 1999). To avoid this, the
mapping of cavernosal nerves, and monitoring their
functional status, was introduced, using electrical
stimulation of the nerves, and either observing penile
tumescence or measuring intracorporeal pressure
(ICP) as the response to stimulation (Lue et al.,
1995; Rehman et al., 1999; Klotz, 2004; Schiff and
Mulhall, 2004). Cavernosal nerves are stimulated
before resection on each side. Monophasic rectangular pulses of 10 mA are delivered at 20 Hz for 1 min.
Nerves are identified according to the greatest penile
tumescence or the greatest increase in ICP. This is
measured by inserting a cannulla into the corpus
cavernosum and connecting it to the pressure transducer (Lue et al., 1995). This technique has been
used in 14 patients, and was reported to show good
correlation of erections to mean increase in ICP during electrical stimulation. Namely, all patients with
normal erections before surgery and sustained ICP
2
Amp (mV)
sometimes following early moderate sensitization

(Fig. 6). Habituation was common, it was more pronounced with higher stimulation frequencies, and in
some cases even interfered with the monitorability
(Rodi, 1998). To avoid habituation, it is better to use
1 Hz stimulation frequency (or even less) than 2 Hz
or more. Additional bolus of propofol significantly
reduced BCR (Fig. 4). In the third study, BCR was
succesfully detected in 1 of 6 patients, but the data
on preoperative or postoperative clinical deficit are
not available (Morota and Nakagawa, 1998).
0
0
t (min)
Fig. 5. Influence of number of pulses in the train stimulus on the amplitude of the bulbocavernosus reflex in one patient.
Note variability of response amplitudes.
746
Fig. 6. Influence of stimulation frequency on habituation of bulbocavernosus reflex.
over 50 cm H20 (i.e., 38 mm Hg, resting is 4.5

7.5 mm Hg) throughout the surgery, reported postoperatively erections sufficient for sexual intercourse.
Those patients with ICP increases <30 cm H20 (i.e.,
22.5 mm Hg) reported partial or absent erections, with
recovery time of several months. On the other hand,
patients with preoperative erectile dysfunction, lost
erections completely and permanently regardless to
intraoperative ICP increase (Rehman et al., 1999).
Intraoperative monitoring of bladder function in
spinal cord surgery is a challenging task. In a study
catheters were introduced into bladder and rectum,
and connected to pressure transducers, to measure
intravesical and intraabdominal pressures; detrusor
pressure (Pdet) was calculated as the difference of
the two (Schaan et al., 2004). Exposed cauda equina
or spinal cord was stimulated electrically with
4.6 square pulses per second at the maximum intensity 45 mA. Pdet increased significantly after stimulation of cauda equina, whereas no response was
recorded after stimulation of the spinal cord. The
Pdet increases after caudal stimulation were between
35 and 70 cm H20 (i.e., 26.5 and 53 mm Hg, resting
values between 3.8 and 7.5 mm Hg). The stimulation
had to be applied continuously to achieve bladder
contraction with the latency between 2 and 10 s.
53.6. Conclusions
It is important to be aware that functions of holding
and passing urine, and stool, and sexual function
are not accessible to continuous intraoperative neurophysiological monitoring. Even nervous structures
that truly control those functions, that is autonomous
nerves, are out of reach of electrophysiology. Intraoperative neurophysiological monitoring can cover
somatic nerves of the lower sacral segments that
innervate ano-genito-perineal skin and mucosa, and
muscles of the pelvic floor, and sphincter muscles.
Because of the close anatomical and functional proximity of avtonomic and somatic nerves, the neurophysiological methods have the potential to guard
against dysfunction. Specific anatomical conditions
in ano-genito-perineal region demand modifications
in stimulation and recording. The identification of
autonomic nerves has been described but controversy persists as to the usefulness of methods. Apart
from, possibly, the pudendal DRAPS, the whole
area of intraoperative mapping and monitoring of
nervous structures associated with pelvic organ
innervation is open for further refinement of the techniques and analysis of the correlation with clinical
outcome.
References
Corcos, J (2004) Simplified anatomy of the vesico-urethral
functional unit. In: J Corcos and E Schick (Eds.), Textbook of the Neurogenic Bladder. Martin Dunitz, London,
New York, pp. 1116.
Deletis, V and Vodusek, DB (1997) Intraoperative recording of the bulbocavernosus reflex. Neurosurgery, 40(1):
8892.
Deletis, V, et al. (2000) Functional anatomical asymmetry of
pudendal nerve sensory fibres. In: HP Bruch, F Koeckerling, R Bouchard and C Schug-Pass (Eds.), New Aspects
of High Technology in Medicine. Hanover, Germany;
October 1620, 2000. Monduzzi Editore, International
Proceedings Division, pp. 153158.
Huang, JC, Deletis, V, Vodusek, DB and Abbott, R (1997)
Preservation of pudendal afferents in sacral rhizotomies.
James, HE, Mulcahy, JJ, Walsh, JW and Kaplan, GW
(1979) Use of anal sphincter electromyography during
operations on the conus medullaris and sacral nerve
roots. Neurosurgery, 4(6): 521523.
Klotz, L (2004) Cavernosal nerve mapping: current data
and applications. BJU Int., 93: 913.
Krassioukov, AV, et al. (2004) Multimodality intraoperative
monitoring during complex lumbosacral procedures:
indications, techniques, and long-term follow-up review
of consecutive cases. J. Neurosurg. (Spine 1), 3: 243253.
Lue, TF, et al. (1995) Intraoperative electrostimulation of
the cavernous nerve: technique, results and limitations.
J. Urol., 154: 14261428.
Morota, N and Nakagawa, H (1998) Intraoperative neurophysiological monitoring and mapping during surgery
of lumbosacral lesions (abstract). Proceedings of the
7th International Symposium on Spinal Cord Monitoring, Osaka, Japan, March 1820, p. 23.
Morrison, J, et al. (2005) Neural control. In: P Abrams, L
Cardozo, S Khoury and A Wein (Eds.), Incontinence,
Basics and Evaluation, Health Publications Ltd, Plymouth, UK. Vol. 1, pp. 363422.
747
Paradiso, G, et al. (2005) Multi-modality neurophysiological monitoring during surgery for adult tethered cord syndrome. J. Clin. Neurosci., 12(8): 935937.
Quignones-Hinojosa, A, et al. (2004) Neurophysiological
monitoring for safe surgical tethered cord syndrome
release in adults. Surg. Neurol., 62: 127135.
Rehman, J, et al. (1999) Intraopereative electrical stimulation of cavernosal stimulation of cavernosal nerves with
monitoring of intracorporeal pressure in patients undergoing nerve sparing radical prostatectomy. BJU Int., 84:
205210.
Rodi, Z (1998) Intraoperative Electrophysiological Monitoring of Bulbocavernosus Reflex in Patients with Injury
of Thoracolumbar Spine. University in Ljubljana,
Slovenia.
Rodi, Z and Vodusek, DB (2001) Intraoperative monitoring
of the bulbocavernosus reflex: the method and its problems. Clin. Neurophysiol., 112(5): 879883.
Schaan, M, et al. (2004) Intraoperative urodynamics in spinal cord surgery: a study of feasibility. Eur. Spine J.,
13: 3943.
Schiff, JD and Mulhall, JP (2004) Neuroprotective strategies in radical prostatectomy. BJU Int., 95: 1114.
Vodusek, DB and Fowler, CJ (2004) Pelvic floor clinical
neurophysiology. In: C Binnie, R Cooper, F Mauguie`re,
J Osselton, P Prior and B Tedman (Eds.), Clinical Neurophysiology. EMG, Nerve Conduction and Evoked
Potentials. Elsevier, Amsterdam, Vol. 1, pp. 281307.
Vodusek, DB and Zidar, J (1988) Perineal motor evoked
responses. International Continence Society 18th
Annual Meeting, Oslo, 13 September 1988. Neurourol.
Urodyn., 7(3): 236237.
Vodusek, DB, et al. (1983) Direct and reflex responses in
perineal muscles on electrical stimulation. J. Neurol.
Neurosurg. Psychiatry, 46(1): 6771.
Von Koch, CS, Quinones-Hinojosa, A, Gulati, M, Lyon, R,
Peacock, WJ and Yingling, CD (2002) Clinical outcome
in children undergoing tethered cord release utilizing
intraoperative neurophysiological monitoring. Pediatr.
Neurosurg., 37(2): 8186.

M.R. Nuwer (Ed.)
748
CHAPTER 54
Monitoring during total hip arthroplasty

Dominic Feea,* and Arman Sabetb
a
Department of Neurology, University of Kentucky Chandler Medical Center, Lexington, KY 40536, USA
b
Department of Neurology, Gold Coast Hospital, Southport, Queensland, QLD 4215, Australia
54.1. Introduction
The sciatic, femoral, obturator, and superior gluteal
nerves lie in close approximation to the hip joint.
This creates the possibility of nerve damage during
total hip arthroplasty/replacement (THA). In fact,
high rates of subclinical injury, assessed by electromyography (EMG), have been found, 877% (Weber
et al., 1976; Ahlgren et al., 1984; Abitbol et al., 1990;
Ramesh et al., 1996; Weale et al., 1996; Kenny et al.,
1999). However, except for intentional incision of
the gluteus medius as part of the surgical approach
(Abitbol et al., 1990; Ramesh et al., 1996; Kenny
et al., 1999; Siebenrock et al., 2000), clinically
identifiable weakness is rare, with an incidence of
0.153.0% for primary procedures (Johanson et al.,
1983; Stone et al., 1985; Edwards et al., 1987; Black
et al., 1991; Kennedy et al., 1991; Simmons et al.,
1991; DeHart and Riley, 1999; Farrell et al., 2005);
most often this weakness is in the peroneal portion
of the sciatic nerve. There are certain identifiable,
preoperative factors that are associated with higher
rates of weakness such as revision/re-do THA
(2.57.5%), hip developmental dysplasia (56%),
possibly excessive limb lengthening, and possibly
female gender (Amstutz et al., 1982; Johanson
et al., 1983; Edwards et al., 1987; Porter et al.,
1989; Schmalzried et al., 1991; Simmons et al.,
1991; Navarro et al., 1995; DeHart and Riley, 1999;
Eggli et al., 1999). At issue is whether intraoperative
monitoring (IOM) will allow for reduction in clinical
postoperative nerve palsy (PONP).
In 1976, Weber et al. wrote a seminal paper
assessing nerve injury following THA (Weber et al.,
*
Correspondence to: D. Fee, M.D., Department of Neurology, University of Kentucky Chandler Medical Center,
KY Clinic, Rm L-445, Lexington, KY 40536, USA.
Tel.: 1-859-323-6702-x251; fax: 1-859-323-5943.
E-mail: dbfee2@email.uky.edu (D. Fee).
1976). It highlights a paradox in all such studies.

The same surgeons utilizing the same surgical techniques that historically have a very low rate of clinical
deficits can demonstrate a high incidence of electrodiagnostic changes, suggesting injury; however, these
changes do not necessarily result in PONP. Retrospectively assessing THA cases over the prior 2 years, clinical PONP was seen in 14 of 2,072 THA (0.7%). They
then prospectively assessed EMG before and 1821
days after THA in 30 surgeries. There were EMG
changes in 21 of the 30 THA (70%), 6 sciatic, 8
obturator, 1 femoral, 5 obturator and sciatic, 1
femoral and sciatic. However, the only clinical
finding was subtle weakness on examination in two
patients, which was not appreciated by the patients
themselves.
54.2. Studies
There have been only 14 articles assessing IOM
during THA, all reported in the orthopedic literature.
All of these studies assessed techniques to minimize
PONP (Table 1) (Stone et al., 1985; Nercessian
et al., 1989; Porter et al., 1989; Black et al., 1991;
Kennedy et al., 1991; Rasmussen et al., 1994; Pereles
et al., 1996; Sutherland et al., 1996; Siebenrock et al.,
2000; Brown et al., 2002; Satcher et al., 2003;
Shiramizu et al., 2003). However, two studies utilize
highly unique techniques, video assessment of
mechanical strain and perineural pressure monitoring, and will not be further discussed (Slater et al.,
2000; Fleming et al., 2003). Early studies examined
somatosensory evoked potential (SEP) changes,
usually defined as P1 latency prolongation >10% of
baseline and/or P1 to N1 amplitude drop of >50%,
predominately in the peroneal portion of the sciatic
nerve. These studies reported SEP changes in
1191% of cases, some with multiple and/or persistent events despite repositioning of instruments and
the leg (Table 1) (Stone et al., 1985; Nercessian
749
Table 1
Studies assessing intraoperative monitoring during elective THA
Study
Stone et al.
(1985)
Nercessian
et al. (1989)
Porter et al.
(1989)
THA
cases
50
25
46
Intraoperative
technique
Cortical SEP
(peroneal nerve)
Cortical SEP
(peroneal nerve)
Cortical SEP
(peroneal nerve)
Intraoperative findings
Postoperative findings
12 events in 10 pts
(all reversible)
12 events in 8 pts
(all reversible)
66 events in 42 pts
(54 reversible,
12 persisted) 21 pts
had significant events
14 pts with events
(6 reversible,
8 persisted)
No clinical PONP
23
Regional SEP
(peroneal nerve)
Black et al.
(1991)
100
Cortical SEP
(peroneal nerve)
18 pts with events

(16 reversible,
2 persisted)
Rasmussen
et al. (1994)
290
Cortical SEP
(peroneal nerve)
31 pts with events

(25 reversible,
6 persisted)
Pereles et al.
(1996)
52
11 events in 8 pts
(all reversible)
Sutherland
et al. (1996)
44
Siebenrock
et al. (2000)
12
Brown et al.
(2002)
63
Satcher et al.
(2003)
27
Shiramizu
et al. (2003)
23
Cortical SEP
(peroneal and
tibial nerve)
Spontaneous EMG
(peroneal and
tibial nerve)
Spontaneous EMG
(superior gluteal
nerve)
Spontaneous EMG
and NAP
(peroneal and
tibial nerve)
Cortical MEP with
EMG recording
(peroneal and
tibial nerve)
Regional MEP
(peroneal nerve)
Kennedy et al.
(1991)
5 pts with events

(all reversible)
3 events in 1 pt
(all reversible)
21 events of EMG
changes, 4 events of
NAP changes, in 19 pts
(2 pts with both)
16 events in 15 pts
All with changes in

certain leg positions
(all reversible)
No clinical PONP (similar to studied,

unmonitored group, 2 of 35, 5.7%)
2 pts with PONP (4.3%, persistent SEP
changes in both) (similar to studied,
unmonitored group, 1 of 42, 2.4%)
No clinical PONP 4 pts with causalgia
(all had SEP changes) (another group
of 52 pts briefly mentioned; no PONP,
but 4 pts developed causalgia, all with
SEP changes)
2 pts with PONP (2.0%, persistent SEP
changes in both) additional pt with
numbness (similar incidence to
institutional rates, 2.6%)
8 pts with PONP (2.8%) (6 had persistent
SEP changes, 2 did not have changes)
(unchanged from unmonitored group,
13 of 485, 2.7%)
No clinical PONP
No clinical PONP, 1 pt with causalgia

(no EMG changes in the pt with
causalgia)
2 pts with PONP (16.7%) (1 with EMG
changes, the other without)
1 pt with PONP (1.6%) (in unmonitored
femoral nerve)
No clinical PONP (similar to historical

control, 3 of 182, 1.6%) (unchanged
from prospective unmonitored group,
0 of 113)
No clinical PONP
SEP: somatosensory evoked potential; pt: patient; PONP: postoperative nerve palsy; EMG: electromyography; NAP: near-nerve action
potential; MEP: motor evoked potential.
et al., 1989; Porter et al., 1989; Black et al., 1991;

Rasmussen et al., 1994; Pereles et al., 1996); however, none indicated a reduction in clinical PONP
with monitoring (Nercessian et al., 1989; Porter
et al., 1989; Black et al., 1991; Rasmussen et al.,
1994). The patients with PONP tended to have persistent SEP changes (Porter et al., 1989; Black
et al., 1991; Rasmussen et al., 1994). However, in
one study, 2 of 8 patients with PONP had no SEP
changes (Rasmussen et al., 1994). One study assessed
750
regional SEP, stimulation of the peroneal nerve at

the fibular head with direct recording on the proximal
sciatic nerve (Kennedy et al., 1991). No patients
developed PONP, but 4 of the 14 with regional
SEP changes, defined as amplitude loss >75% baseline or latency increase >10%, developed causalgia
(Kennedy et al., 1991).
Four studies assessed spontaneous EMG in THA
(Table 1) (Sutherland et al., 1996; Siebenrock et al.,
2000; Brown et al., 2002; Satcher et al., 2003).
Brown et al. combined it with near-nerve action
potential (NAP) recording (Brown et al., 2002), and
Satcher et al. used EMG to monitor nerve function
between motor evoked potentials (MEP) as well as
to assess the MEP response (Satcher et al., 2003).
Siebenrock et al. assessed the superior gluteal nerve
in an attempt to minimize damage to this nerve during the surgical approach (Siebenrock et al., 2000).
Three events were seen in one patient who developed
PONP; however, one patient had PONP of the
superior gluteal nerve without spontaneous EMG discharges during the surgery. The other three investigations assessed the peroneal and tibial nerves; no
PONP occurred in the monitored nerves (Sutherland
et al., 1996; Brown et al., 2002; Satcher et al.,
2003). One patient developed causalgia with no
spontaneous EMG discharges seen intraoperatively
(Sutherland et al., 1996), and another developed
PONP in the unmonitored femoral nerve (Brown
et al., 2002). Two studies have assessed MEP in
THA (Table 1); neither study had PONP (Satcher
et al., 2003; Shiramizu et al., 2003). Satcher et al. felt
that cortical-induced MEP with EMG recording,
allowed them to recognize certain positions to avoid
in surgery, such as hip flexion (Satcher et al.,
2003). PONP in an unmonitored surgical series by
the same surgeons before the study was 3 of 182
(1.6%) and was 0% (0 of 113) in an unmonitored
series after the study (Satcher et al., 2003). In opposition to the positioning recommended by Satcher
et al., Shiramizu et al., assessing sciatic nerveinduced MEP, felt that hip flexion with internal rotation was the preferred position to minimize MEP
changes (Shiramizu et al., 2003).
54.3. Conclusion
Of the 12 studies reviewed, only 2 suggested that
IOM during THA might reduce PONP (Nercessian
et al., 1989; Satcher et al., 2003). PONP following
D. FEE AND A. SABET
THA is rare, and the studies done to date do not

have enough power to detect a difference. Over
3,000 THA cases would need to be assessed to detect
a change in incidence from 3.0% to 0.1% (Porter
et al., 1989). It is also possible that IOM techniques
commonly utilized are neither specific nor sensitive
for nerve damage in THA. Nearly all the studies
had events in excess of PONP (Table 1); two studies
had false negative results with PONP in the nerve
assessed and no IOM changes (Rasmussen et al.,
1994; Siebenrock et al., 2000). For routine THA,
the value of IOM has not been convincingly demonstrated. IOM may have a role in selected circumstances such as high-risk THA, for example
revision/re-do or developmental dysplasia cases
(Black et al., 1991; DeHart and Riley, 1999), or to
allow a surgeon to assess his or her technique
(Satcher et al., 2003).
References
Abitbol, JJ, Gendron, D, Laurin, CA and Beaulieu, MA
(1990) Gluteal nerve damage following total hip arthroplasty. A prospective analysis. J. Arthroplasty, 5(4):
319322.
Ahlgren, SA, Elmqvist, D and Ljung, P (1984) Nerve
lesions after total hip replacement. Acta Orthop. Scand.,
55(2): 152155.
Amstutz, HC, Ma, SM, Jinnah, RH and Mai, L (1982)
Revision of aseptic loose total hip arthroplasties. Clin.
Black, DL, Reckling, FW and Porter, SS (1991) Somatosensory-evoked potential monitored during total hip
arthroplasty. Clin. Orthop. Relat. Res., 262: 170177.
Brown, DM, McGinnis, WC and Mesghali, H (2002) Neurophysiologic intraoperative monitoring during revision
total hip arthroplasty. J. Bone Joint Surg. Am., 84A
(Suppl. 2): 5661.
DeHart, MM and Riley, LH, Jr. (1999) Nerve injuries in total
hip arthroplasty. J. Am. Acad. Orthop. Surg., 7(2):
101111.
Edwards, BN, Tullos, HS and Noble, PC (1987)
Contributory factors and etiology of sciatic nerve palsy
in total hip arthroplasty. Clin. Orthop. Relat. Res., 218:
136141.
Eggli, S, Hankemayer, S and Muller, ME (1999) Nerve
palsy after leg lengthening in total replacement arthroplasty for developmental dysplasia of the hip. J. Bone
Joint Surg. Br., 81(5): 843845.
Farrell, CM, Springer, BD, Haidukewych, GJ and Morrey, BF
(2005) Motor nerve palsy following primary total hip
arthroplasty. J. Bone Joint Surg. Am., 87(12): 26192625.
Fleming, P, Lenehan, B, ORourke, S, McHugh, P, Kaar, K
and McCabe, JP (2003) Strain on the human sciatic nerve

in vivo during movement of the hip and knee. J. Bone
Joint Surg. Br., 85(3): 363365.
Johanson, NA, Pellicci, PM, Tsairis, P and Salvati, EA
(1983) Nerve injury in total hip arthroplasty. Clin.
Kennedy, WF, Byrne, TF, Majid, HA and Pavlak, LL (1991)
Sciatic nerve monitoring during revision total hip arthroplasty. Clin. Orthop. Relat. Res., 264: 223227.
Kenny, P, OBrien, CP, Synnott, K and Walsh, MG (1999)
Damage to the superior gluteal nerve after two different
approaches to the hip. J. Bone Joint Surg. Br., 81(6):
979981.
Navarro, RA, Schmalzried, TP, Amstutz, HC and Dorey, FJ
(1995) Surgical approach and nerve palsy in total hip
Nercessian, OA, Gonzalez, EG and Stinchfield, FE (1989)
The use of somatosensory evoked potential during revision or reoperation for total hip arthroplasty. Clin. Orthop.
Relat. Res., 243: 138142.
Pereles, TR, Stuchin, SA, Kastenbaum, DM, Beric, A,
Lacagnino, G and Kabir, H (1996) Surgical maneuvers
placing the sciatic nerve at risk during total hip arthroplasty as assessed by somatosensory evoked potential
monitoring. J. Arthroplasty, 11(4): 438444.
Porter, SS, Black, DL, Reckling, FW and Mason, J (1989)
Intraoperative cortical somatosensory evoked potentials
for detection of sciatic neuropathy during total hip
arthroplasty. J. Clin. Anesth., 1(3): 170176.
Ramesh, M, OByrne, JM, McCarthy, N, Jarvis, A, Mahalingham, K and Cashman, WF (1996) Damage to the
superior gluteal nerve after the Hardinge approach to
the hip. J. Bone Joint Surg. Br., 78(6): 903906.
Rasmussen, TJ, Black, DL, Bruce, RP and Reckling, FW
(1994) Efficacy of corticosomatosensory evoked potential monitoring in predicting and/or preventing sciatic
nerve palsy during total hip arthroplasty. J. Arthroplasty, 9(1): 5361.
Satcher, RL, Noss, RS, Yingling, CD, Ressler, J and Ries,
M (2003) The use of motor-evoked potentials to
751
monitor sciatic nerve status during revision total hip
Schmalzried, TP, Amstutz, HC and Dorey, FJ (1991) Nerve
palsy associated with total hip replacement. Risk factors
and prognosis. J. Bone Joint Surg. Am., 73(7): 10741080.
Shiramizu, K, Naito, M, Akiyoshi, Y and Yamaguchi, T
(2003) Acute effects of hip and knee positions on
motor-evoked potentials of the sciatic nerve in total
hip arthroplasty. Int. Orthop., 27(4): 211213.
Siebenrock, KA, Rosler, KM, Gonzalez, E and Ganz, R
(2000) Intraoperative electromyography of the superior
gluteal nerve during lateral approach to the hip for
arthroplasty: a prospective study of 12 patients.
J. Arthroplasty, 15(7): 867870.
Simmons, C, Jr., Izant, TH, Rothman, RH, Booth, RE, Jr.
and Balderston, RA (1991) Femoral neuropathy following total hip arthroplasty. Anatomic study, case reports,
and literature review. J. Arthroplasty, 6(Suppl.):
S57S66.
Slater, N, Singh, R, Senasinghe, N, Gore, R, Goroszeniuk,
T and James, D (2000) Pressure monitoring of the femoral nerve during total hip replacement: an explanation
for iatropathic palsy. J. Roy. Coll. Surg. Edinb., 45(4):
231233.
Stone, RG, Weeks, LE, Hajdu, M and Stinchfield, FE (1985)
Evaluation of sciatic nerve compromise during total hip
arthroplasty. Clin. Orthop. Relat. Res., 201: 2631.
Sutherland, CJ, Miller, DH and Owen, JH (1996) Use of
spontaneous electromyography during revision and
complex total hip arthroplasty. J. Arthroplasty, 11(2):
206209.
Weale, AE, Newman, P, Ferguson, IT and Bannister, GC
(1996) Nerve injury after posterior and direct lateral
approaches for hip replacement. A clinical and electrophysiological study. J. Bone Joint Surg. Br., 78(6):
899902.
Weber, ER, Daube, JR and Coventry, MB (1976) Peripheral
neuropathies associated with total hip arthroplasty.
J. Bone Joint Surg. Am., 58(1): 6669.

M.R. Nuwer (Ed.)
752
CHAPTER 55
Monitoring neural function during pelvic surgery

Berton R. Moed*
Department of Orthopedic Surgery, Saint Louis University School of Medicine, St Louis, MO 63110, USA
55.1. Introduction
Any discussion regarding pelvic surgery must begin
by differentiating between two major, functionally
separate, regions of bony pelvic anatomy: the pelvic
ring and the acetabulum. The pelvic ring is the part
of the axial skeleton formed by three bones (the
sacrum and the two innominate bones) and their
connecting ligamentous structures at the sacroiliac
joints and the pubic symphysis (Fig. 1). While the
anterior structures mainly resist rotational forces,
the posterior sacroiliac ligamentous complex is the
primary stabilizer of the pelvic ring (Tile, 2003).
Injuries to the posterior sacroiliac complex (type-C)
that destabilize the pelvic ring require surgery that
places particular neural structures at risk (Moed
et al., 2003). These neural structures include the
sacral nerve roots (mainly S1), as they traverse the
sacrum and exit the sacral foramina, the L5 nerve
root in its course along the sacral ala, and the lumbosacral plexus (Fig. 1). Therefore, nerve monitoring
during pelvic ring surgery has been directed toward
the intraoperative evaluation of lumbosacral nerve
root function.
The acetabulum is a highly specialized area of the
innominate bone, being the proximal portion of the
hip joint, and is an important structure in the transfer
of weight from the legs to the axial skeleton. Injury
to the acetabulum may be a component of a pelvic
ring injury. However, displaced fractures of the
acetabulum destabilize the hip joint, not the pelvic
ring (Fig. 2). Injuries to the acetabulum that result
in instability or incongruity of the hip joint require
surgery that places peripheral nerves, rather than the
*
Correspondence to: Berton R. Moed, M.D., Department of

Orthopedic Surgery, Saint Louis University School of
Medicine, 3635 Vista Avenue, 7th Floor, Desloge Towers,
St. Louis, MO 63110, USA.
Tel.:1-314-577-8850; fax: 1-314-268-5121.
E-mail: moedbr@aol.com (B.R. Moed).
lumbosacral nerve roots, at risk. Almost all major

peripheral nerves traversing the pelvis (including
the superior gluteal nerve, the pudendal nerve, the
obturator nerve, the femoral nerve, and the lateral
femoral cutaneous nerve) have been reported to have
been injured during acetabular fracture surgery
(Letournel and Judet, 1993). However, it is injury
to the sciatic nerve that is of the most functional
importance and has the greatest overall risk (Letournel
and Judet, 1993; Gruson and Moed, 2003; Schmeling
et al., 2003). Therefore, nerve monitoring during acetabular fracture surgery has been directed toward the
intraoperative evaluation of sciatic nerve function.
55.2. Pelvic ring surgery
As noted above, surgical instrumentation of the posterior pelvic ring places the sacral nerve roots, the
L5 nerve root, and the lumbosacral plexus at risk
for iatrogenic injury. The most commonly performed
surgical procedure involves reduction and fixation of a
fracture that has caused instability of the ring. However, similar risks are involved during surgery to treat
malunions and nonunions of the pelvic ring or for
arthrodesis of the sacroiliac joint. During surgery, the
neural structures may be injured mainly in three
different ways: stretch, compression, or laceration.
The nerve roots or lumbosacral plexus may sustain a stretch injury during reduction maneuvers used
to realign the displaced portion of the pelvic ring.
The nerve roots may also be compressed or crushed
by fracture fragments during these reduction maneuvers (Browner et al., 1987; Matta and Tornetta,
1996). The L5 nerve root is particularly at risk for
compression injury in fractures that involve the
sacral ala (Fig. 3); the sacral nerve roots are at similar risk during reduction maneuvers for fractures
through zone II or zone III (Denis et al., 1988) of
the sacrum (Fig. 3). These risks exist whether the
pelvic ring displacement is reduced by closed or open
means. Although reduction maneuvers do place the
753
S2
S1 and L5 NERVE ROOTS SACRUM SACROILIAC JOINT

INNOMINATE BONE
PELVIC
RING
LUMBOSACRAL
PLEXUS
ACETABULUM
and
HIP JOINT
PUBIC
SYMPHYSIS
Fig. 1. Drawing showing the pelvic ring, the hip joints, and the relationship of the neural structures to the bony pelvis.
neural structures at risk, it is the subsequent insertion

of implants that is most problematic.
Internal fixation of the posterior sacroiliac ligamentous complex places the L5 and sacral nerve
roots at risk for laceration or crush by the inserted
implants (Leggon et al., 2002; Moed et al., 2003).
Although a number of different fixation techniques
have been reported in the literature, iliosacral screw
fixation has emerged as the method of choice for posterior pelvic ring fracture fixation because of its biomechanical superiority (Matta and Saucedo, 1989;
Cole et al., 1996; Matta and Tornetta, 1996; Routt
Fig. 2. Anteroposterior radiograph of the pelvis showing a

posterior wall fracture of the acetabulum with a posterior
dislocation of the hip. The acetabulum has destabilized
the hip joint. However, the pelvic ring is not involved in
the fracture.
and Simonian, 1996; Moed et al., 2003). However,

misdirection of the guide wire, drill bit, or screw
may impale the nerve root (Fig. 4). Whether inserted
percutaneously in conjunction with closed reduction
or following open reduction, the iliosacral screw
has an ideal projected intraosseous course that brings
it close to the fifth lumbar and sacral nerve roots
and the spinal canal. Therefore, the implant must follow a narrow path in order to remain intraosseous
(Templeman et al., 1996a; Moed et al., 2003).
Templeman et al. (1996a) have shown that as little
as 4 of change in angulation of the surgeons hand
can cause a potentially disastrous alteration in the
trajectory of the implant.
The course of the iliosacral implant cannot be
directly visualized during the operative procedure.
Image intensification fluoroscopy by using multiple
views has become the standard procedure for determining implant position (Matta and Saucedo, 1989;
Matta and Tornetta, 1996; Routt and Simonian, 1996;
Routt et al., 1996; Moed et al., 2003). However, accurate screw placement using this method relies on
obtaining an anatomic reduction. Obtaining a less than
anatomic reduction of the displaced sacrum or sacroiliac joint decreases the safe area for implant insertion
even further (Reilly et al., 2003). In addition, the radiographic imaging of bony landmarks may occasionally be obscured or misinterpreted (Mostafavi and
Tornetta, 1996; Routt et al., 1996; Webb et al.,
2000). Therefore, various neural function monitoring
systems have been advocated to assist in avoiding
nerve root injury during pelvic fracture fixation.
754
B.R. MOED
Fig. 3. Drawings showing an anteroposterior view of the sacrum (including the L5 vertebrae, the L5/S1 disk space, and
L5S2 nerve roots) with a fracture through zone II. A: The zones of the sacrum (Denis et al., 1988) are shown on the right
and a displaced zone II fracture is depicted on the left. B: The zone II fracture has been reduced in slight malposition,
trapping, and compressing the nerve roots. C: The zone II fracture has been reduced in slight malposition. However, sufficient space remains available at the neural foramina, and the nerve roots have not been injured. D: The fracture from drawing (C) has now been fixed with a lag screw, resulting in compression of the malaligned fracture fragments, which in turn
compress and injure the nerve roots.
55.2.1. Nerve monitoring techniques

The technical aspects of the various methods of
intraoperative neural function monitoring are
described in other chapters. However, there are certain features that are unique to pelvic ring surgery
and bear mentioning. In general, three types of
intraoperative monitoring have been in use for pelvic
ring surgery: somatosensory evoked potential (SEP)
monitoring, free-running continuous electromyography (EMG), and stimulus-evoked electromyography
(SE-EMG).
To the best of the authors knowledge, the first
report describing intraoperative SEP monitoring for
pelvic ring surgery was provided by Vrahas et al.
(1992). Their monitoring technique was similar to that

described for spine surgery. However, they supplemented the standard common peroneal and tibial
nerve stimulating electrodes by adding pudendal nerve
stimulating electrodes. The value of pudendal nerve
SEP monitoring has not been established. Therefore,
its use is no longer recommended.
Free-running EMG monitoring relies on continuous monitoring of EMG activity in the muscles innervated by the neural structures of interest. For pelvic
ring surgery, monitored muscles mainly consist of
the tibialis anterior and the gastrocnemius. Depending upon the surgeons preference, vastas lateralis
and lateral hamstring muscle electrodes may be
added (Webb et al., 2000). An abnormal spontaneous
755
Fig. 4. Illustrations showing how misdirection of a screw may impale the nerve root. A: Drawing of a cross-section through
the posterior sacroiliac complex showing an iliosacral lag screw coursing through the S1 foramen. B: Computerized tomographic section from a patient in whom two iliosacral screws were inserted into the S1 vertebral body showing one of the
two screws passing through the anterior portion of the S1 foramen. Postoperatively, the patient had a functional deficit in
the S1 nerve distribution.
burst of neurotonic activity is thought to be indicative

that some type of mechanical insult has occurred to
the neural structure being monitored (i.e., the nerve
roots or lumbosacral plexus).
Stimulus-evoked electromyographic monitoring
is a relative newcomer to pelvic ring surgery. This
monitoring technique has been used successfully during operations on the facial and recurrent laryngeal
nerve (Mller and Jannetta, 1984; Gantz, 1985;
Benecke et al., 1987; Rice and Cone-Wesson, 1991;
Maloney et al., 1994) and during pedicle screw fixation of the lumbosacral spine both in animal models
and in clinical trials (Calancie et al., 1992, 1994;
Lenke et al., 1995; Maguire et al., 1995). Some differences in methodology do exist, depending on the specific application. For facial and recurrent laryngeal
nerve surgery, the objective is to locate but not injure
the nerve. In pelvic ring surgery, the objective is to
avoid all contact with the nerve, without its localization. However, in all instances, the monitoring technique has been essentially identical: electrodes for
EMG monitoring are placed in the muscle(s) innervated by the nerve in question and the potentially
offending agent (dissecting instrument or implant)
serves as the stimulating electrode (Mller and Jannetta, 1984; Benecke et al., 1987; Rice and Cone-Wesson, 1991; Calancie et al., 1992; Calancie et al., 1994;
Maguire et al., 1995; Moed et al., 1998a, 1998b, 1999).
The concept is also the same: to warn the surgeon that
the implant or instrument is close to the nerve. For pelvic ring surgery, bipolar needle recording electrodes
are placed in the gastrocnemius and tibialis anterior

muscles and the guide wire, drill bit or implant serves
as the stimulating (cathode) electrode (Moed et al.,
1998b). An EEG needle placed in the subcutaneous tissue serves as the reference (anode) electrode.
The location of this reference electrode is critical in
pelvic ring surgery. The anode must be placed in the
midline or on the side contralateral to implant insertion
(Moed et al., 1998b, 2006; Moed, 2003; Ricci et al.,
2003). It is a basic (but critical) principle of this monitoring technique that the reference lead be placed
appropriately, no closer to the side of drill-bit insertion
than the midline. Ipsilateral anode location will cause
an abnormally high-current threshold reading, meaning that the stimulating electrode (guide wire, drill
bit, or implant) is actually much closer to the neural
structure than indicated, thereby placing the structure
at risk. This technical error may lead to nerve
injury despite a satisfactory current threshold reading
(a false-negative result).
In pelvic ring surgery, the particular constantcurrent SE-EMG monitoring setup mainly follows
that recommended for spine surgery by Calancie
et al. (1994), using monopolar, monophasic squarewave stimuli of 0.2 ms duration delivered at 3 Hz.
However, for the pelvic ring a searching current of
50 mA is initially applied to the drill bit. Furthermore, a current threshold of >8 mA is recommended
as indicative of a safe distance between the drill bit
and neural structure (Moed et al., 1998b, 1999).
Therefore, a current threshold of 8 mA or less is an
756
indication for drill-bit redirection. After satisfactory

placement of a drill bit or guide wire, a cannulated
screw can be inserted over the drill bit or guide
wire, again using SE-EMG monitoring, by connecting the cathode lead to the shaft of the screwdriver.
A current threshold of >6 mA has been used as
the absolute minimum acceptable level for final screw
position (Moed et al., 1998b; Moed and Geer, 2006).
In an isolated report, Ricci et al. (2003) described their
technical difficulties using this method during pelvic
ring surgery. However, the applicability of SE-EMG
monitoring to pelvic ring surgery has been validated
in animals and humans and substantiated by finite
element modeling (Moed et al., 1998a,b, 1999,
2006).
55.2.2. Clinical outcomes without neural
function monitoring
Iatrogenic nerve injury rates have been reported to be
as low as 1% (Matta and Tornetta, 1996) to as high
as 18% (Webb et al., 2000). Matta and Tornetta
(1996) reported on 76 patients with 83 posterior injuries (69 unilateral and 7 bilateral). The vast majority
(64 of 83) were fixed using iliosacral screws. They
reported only one neural injury involving the L5
nerve root. Their findings are in contrast to other
investigators, reporting on smaller patient numbers
but possessing a similar level of surgical expertise.
Templeman et al. (1996b) found that 2 of their
30 patients sustained iatrogenic nerve injury involving the L5 nerve distribution resulting in foot drop
after the treatment of a displaced fracture through
the sacrum. One patient had been treated by open
reduction and iliosacral screw fixation, the other by
closed reduction and percutaneous iliosacral screw
insertion. In both cases, postoperative computed axial
tomography showed satisfactory intraosseous screw
location and patent neural foramina. The authors surmised that the nerve injury occurred during the
reduction maneuvers. They further stated that their
overall findings showed that the presence of neurological injury is the single most important predictor
of a compromised outcome in these pelvic fracture
patients. After reviewing these results, the authors
indicated that they would subsequently use intraoperative neural monitoring for these procedures.
Webb et al. (2000) retrospectively evaluated
22 unmonitored patients in their series. Of these
22 patients, 4 (18%) sustained neurological injury.
All four complained postoperatively of foot and/or
B.R. MOED
leg burning dysesthesias and one also had a foot

drop. Postoperative computed tomography (CT) showed
an errant extraosseous course of an iliosacral screw with
encroachment of the S1 tunnel in all four. Immediate
screw removal resulted in resolution of the burning dysesthesias. The one patient with the foot drop was
noted to have improvement at a 2-year follow-up
examination, with continued weakness but no longer
brace-dependent.
55.2.3. Clinical outcomes with neural
function monitoring
In general, the reported prevalence of iatrogenic neural injury during reduction and fixation of a posterior
pelvic ring injury incorporating various methods of
neural function monitoring approximates 0%. As
noted above, Vrahas et al. (1992) were perhaps the
first to report the use of intraoperative neural function monitoring during pelvic or acetabular fracture
surgery. They reported on 30 pelvic ring disruptions
in which there were no iatrogenic neural injuries.
They noted SEP changes during three distinct activities during the various surgical procedures: (1) during retraction of the L5 nerve root for exposure of
the anterior aspect of the sacroiliac joint, (2) reduction maneuvers for the displaced hemipelvis, and
(3) insertion of iliosacral screws. In each situation,
a prompt response resulted in a return to SEP baseline. These responses, respectively, consisted of:
(1) release of retraction of the nerve root, (2) release
of reduction forceps followed by slow reduction in
stages, and (3) redirection of the implant. It was presumed by the authors that these implants must have
been close to, but not have impaled, the nerve root
for redirection of iliosacral guide wires or implants
to be successful in avoiding iatrogenic neural injury.
They emphasized the value of SEP monitoring, especially in patients with a preoperative incomplete
traumatic neural injury. They also noted that SEP
changes could be labile, but that a rapid response to
remove the offending agent was required to prevent
neural injury.
Subsequently, Helfet et al. (1995) reported using
intraoperative SEP monitoring in a series of 28
patients with 30 unstable hemipelvic injuries. Their
retrospective analysis revealed that significant SEP
changes occurred during the manipulative reduction
in three. With expeditious response of the surgical
team, consisting of release of traction or removal of
instruments, SEPs returned to baseline and no patient
sustained an iatrogenic neural injury. They recommended that SEP monitoring should be used to avoid
neural stretch or compression injury during fracture
reduction and suggested that it could also be of value
during implant insertion.
In the prospective part of the study of Webb et al.
(2000), the preliminary results of which were initially
published in 1996, 29 patients were monitored intraoperatively using free-running, continuous EMG.
There were four intraoperative episodes of neurotonic discharge in three patients. These episodes all
occurred during the advancement of the tip of the
guide wire for the iliosacral screw just beyond the
S1 foramen. As a result of the burst of EMG activity,
the surgeon redirected the guide wire, resulting in
resolution of the EMG activity. All three patients
had normal postoperative neural function. All postoperative CT scans showed intraosseous screw position.
Overall, there were no iatrogenic neural injuries in
these 29 patients.
Moed et al. (1998b) utilized a combination of
nerve monitoring techniques in a prospective series
of 27 neurologically intact patients with 30 unstable
(type-C) disruptions of the pelvic ring in whom 51
iliosacral screws were inserted. In this series, SEP
monitoring was used during the surgical approach
(for the open procedures) and the reduction maneuvers and SE-EMG monitoring was used during
implant insertion for fracture fixation. In these
patients, leads were placed for both SEP monitoring
and SE-EMG. General anesthesia was induced using
narcotic agents, short-acting muscle relaxants, and
1.0% isoflurane. Baseline SEP responses were
recorded after positioning and updated throughout
the operative procedure. After reducing the fracture
and obtaining a normal SEP recording, neuromuscular blockade was reversed and kept at a minimum,
as indicated by a response of three or four twitches
to the train of four stimuli to allow the use of
SE-EMG monitoring during the insertion of iliosacral
screws. A current threshold of >8 mA to evoke an
electromyographic response was selected as the safe
zone for the drill bit. At the completion of iliosacral
screw insertion, a final run of SEPs was recorded.
In addition, observation for spontaneous electromyographic activity was added to the overall monitoring
regimen for the last 14 patients in this series. Based
on the SE-EMG current threshold criterion, intraoperative redirection of the drill bit was required
before screw insertion for 4 of the 51 screws. All
patients were neurologically intact postoperatively
757
with intraosseous screw location documented on

postoperative CT scans. Other findings included one
false-positive SEP and one false-positive spontaneous burst of EMG activity. The authors concluded
that SE-EMG monitoring provides reliable data and
should decrease the risk of iatrogenic nerve root
injury during pelvic ring surgery.
55.2.4. Clinical outcomes under special
circumstances
Limited space for the insertion of screws into the S1
body may exist because of anatomical variations. If
the shape of the sacrum is a variant of the normal
(sacral dysmorphism) the insertion of a single iliosacral S1 screw places the neural structures at increased
risk for injury (Routt et al., 1996, 1997). In addition,
size limitations may preclude the insertion of more
than one screw fixation into a normal S1 body (Routt
et al., 1996, 1997; Templeman et al., 1996a). Alternative posterior fixation constructs exist, using
plating techniques, transiliac bars, lumbopelvic fixation, and iliosacral screws inserted into the S2 body.
The S2 safe zone is smaller and more difficult to
understand fluoroscopically than that for the body
of S1 (Routt et al., 1997). Therefore, it follows that
the S2 iliosacral screw has been almost universally
avoided because of a perceived increased risk for
nerve root injury. In the clinical series of Van den
Bosch et al. (2002), a significantly greater number
of their patients with the second screw of a twoscrew construct inserted into S2 (6/31) sustained neurological injury as compared with those with both
screws in S1 (1/49). At issue in this study, and one
recognized by the authors, is that the vast majority
of the screws (and all of the S2 screws) were inserted
without the benefit of the lateral view of the sacrum.
Only the inlet and outlet fluoroscopic views were
used during the first 5 years of this 5.5 year study.
Ziran et al. (2002), as part of a larger study, inserted
31 screws into S2 without an adverse event using
a CT-guided technique. Intraoperative nerve monitoring was not used by any of these authors.
In contradistinction to the findings of Van Den
Bosch et al. (2002), Moed and Geer (2006) reported
no intraoperative iatrogenic neural injuries in a series
of 49 patients in whom S2 iliosacral screws were
inserted using intraoperative nerve monitoring. All
three fluoroscopic views (inlet, outlet, and lateral)
were employed by these authors. In addition, the
intraoperative nerve monitoring technique using
758
SE-EMG in combination with SEPs, as described

above, was utilized in all cases. The authors concluded
that, while perhaps not needed in all cases, the supplemental information provided by intraoperative nerve
monitoring may be important, considering the limitations of fluoroscopy in visualizing access to the
S2 body.
55.2.5. Limitations of monitoring in pelvic ring
surgery for the systems currently in use
Despite the risk for neural injury during pelvic ring
surgery and the reported successes for all of the
described monitoring techniques, monitoring of neural function during pelvic ring surgery has not gained
general acceptance. To a great degree, this lack of
acceptance is due to the reports of both false-positive
and false-negative results, especially with the use
of SEP and free-running EMG (Nash and Brown,
1989; Moed et al., 1998b). Some issues can be
directly attributed to the inherent limitations of the
monitoring method, and some to technical difficulties
experienced by the surgeons in setting up the monitoring system in an operating room environment.
Both SEP and free-running EMG monitoring
methods provide information, indicating that there
has been a change from normal to abnormal. Somatosensory evoked potentials are limited by the need for
a rapid response to an abnormal signal and the fact
that the sensory, not the motor, pathway is being
monitored. Free-running electromyographic monitoring relies on the observation of one or more spontaneous bursts of neurotonic activity, indicating that
there has been some mechanical insult to the nerve
root. Therefore, the hope is that the loss of signal
(SEP) or abnormal signal (free-running EMG) has
been caused by a situation (such as stretching of the
nerve by retractors or minor nerve root compression)
that is reversible. Laceration or crush of the neural
structures by a guide pin, drill bit, or implant cannot
be prevented using either of these after the fact monitoring techniques. In addition, it has been shown in
an animal model that both SEP and free-running
EMG monitoring fail to reliably indicate when an
implant has made contact with a sacral nerve root
(Moed et al., 1999). A likely explanation is that these
reactive monitoring techniques are not sensitive
enough to recognize damage to a relatively small
portion of the entire neural structure that is being
monitored. The S1 nerve root makes up only a
B.R. MOED
portion of the tibial nerve (the structure that is being

monitored) and only a part of the nerve root is being
damaged potentially a very small part when the
nerve root is impaled by the sharp implant. Therefore, it is possible that a more generalized insult to
a larger structure (such as stretch or compression of
the lumbosacral plexus during reduction of a displaced pelvic fracture) is amenable to continuous
electromyographic and SEP monitoring, while injury
to a nerve root from iliosacral implant insertion is
not. Stimulus-evoked EMG monitoring consistently
provides reliable data, indicating the proximity of
the implant to the nerve root (Moed et al., 1999).
The presence of a preoperative, traumatic neural
injury may severely limit the reliability of all
the monitoring methods. Free-running EMG and
SE-EMG monitoring require an intact preoperative
distal motor function in the nerve distribution of
interest. However, preoperative loss of tibialis anterior muscle motor activity does not preclude monitoring of gastrocnemius muscle for S1 function.
Similarly, a preoperative loss of gastrocnemius muscle motor activity does not preclude monitoring of
tibialis anterior muscle for L4/5 function. Although
SEP monitoring has been touted as having even
greater benefit in the presence of an incomplete nerve
injury (Vrahas et al., 1992), in this situation often a
reliable baseline signal cannot be obtained. Functional SEP monitoring requires that a consistent,
reliable baseline signal be obtained.
55.2.6. Recommendations for monitoring neural
function during pelvic ring surgery
Neural structures are at risk during pelvic ring surgery and numerous reports show the potential benefit
of intraoperative monitoring. However, the need to
monitor neural function during pelvic ring surgery
remains controversial pending a randomized prospective trial performed by experienced surgeons. Despite
this lack of definitive information, it is possible to
make a number of general recommendations.
55.2.6.1. Indications
Neural function monitoring appears to be of benefit
during the reduction of a displaced posterior pelvic
ring injury. Stimulus-evoked EMG monitoring is of
value during the insertion of any implants placed
close to the neural structures being monitored.
759
maneuvers, an alternative reduction method should

be considered, such as converting from a closed to
an open technique, or decompressing the neural
foramina. Stimulus-evoked EMG current thresholds
approaching 8 mA during guide wire or drill-bit
insertion warrant redirection of the device.
55.3. Acetabular fracture surgery
Fig. 5. Postoperative outlet pelvic radiograph in a 24-yearold woman with a dysmorphic S1 segment who was
involved in a motor vehicle accident sustaining a sacroiliac
joint dislocation on the right and a zone II sacral on the left.
Iliosacral screws were inserted into the S1 and S2 vertebral
bodies without incident using a combination of somatosensory evoked potential (SEP) and stimulus-evoked electromyograhy (EMG) monitoring.
Stimulus-evoked EMG monitoring is indicated when

iliosacral screws are inserted under less than ideal
circumstances. Such conditions exist when space
available for the implant is limited or the intraoperative radiographic imaging of osseous landmarks is
impaired. Examples of limited available space
include the insertion of screws into the S2 body or
into a dysmorphic S1 body (Fig. 5), and any time a
less than anatomic reduction is obtained (Reilly
et al., 2003). Intraoperative radiographic visualization of osseous landmarks may be obscured by excessive bowel gas, residual abdominal contrast material,
an obese body habitus, and malposition of the
fluoroscopy unit (Mostafavi and Tornetta, 1996).
55.2.6.2. Responding to abnormal intraoperative
signals
Intraoperative monitoring of neural function can be
of value only if the recognition of abnormal signals
results in an adjustment by the surgical team that limits or completely avoids an otherwise likely neural
injury. A significant change in SEP signals or a spontaneous burst of EMG activity during a reduction
maneuver requires an immediate reversal of the maneuver. If the abnormal signals recur with repeated
Damage to the sciatic nerve, whether it occurs at the

time of the initial fracture trauma or later during the
course of the patients subsequent treatment, is one
of the major complications encountered in acetabular
fracture management. Letournel and Judet (1993)
reported a 6% overall prevalence of postoperative iatrogenic nerve injury but noted that over one half of
these patients had not received an adequate preoperative physical examination, leaving the actual cause of
nerve injury in doubt. They stressed the need for a complete preoperative neurological evaluation. Aside from
the obvious medico-legal implications, preoperative
diagnosis of nerve injury has additional importance
because the previously injured nerve appears to be
at the greater risk for iatrogenic intraoperative injury
(Helfet et al., 1991; Vrahas et al., 1992). Patients with
posterior wall or column displacement are at the greatest risk for intraoperative iatrogenic nerve injury.
Intraoperative nerve injury is most commonly associated with surgical approaches that involve direct
exposure and retraction of the sciatic nerve (Vrahas
et al., 1992; Letournel and Judet 1993) (Fig. 6).
However, injury at the time of indirect reduction of posterior displacement using an anterior surgical approach
can also occur (Letournel and Judet 1993).
55.3.1. Nerve monitoring techniques
As noted above, the technical aspects of the various
methods of intraoperative neural function monitoring
are described in other chapters. However, there are
certain features that are unique to sciatic nerve
monitoring during acetabular fracture surgery. In
general, three types of intraoperative monitoring
have been used for acetabular fracture surgery:
SEP, free-running EMG, and motor evoked potential
(MEP) monitoring. The sciatic nerve consists of two
separate divisions (peroneal and tibial), which must
be individually monitored to allow detection of
incomplete lesions (Moed et al., 1992).
760
B.R. MOED
55.3.2. Clinical outcomes without and with neural

function monitoring
Fig. 6. A: Intraoperative photograph showing a posterior

approach to the hip (the patient is prone with his head
to the right) and a reduction clamp in appropriate position.
B: Intraoperative anteroposterior fluoroscopic image
showing the clamp position for the reduction of a transverse
fracture of the acetabulum. C: Drawing depicting the clamp
and the reduced fracture showing how the elbow of the clamp
can compress or stretch the sciatic nerve.
The rate of intraoperative sciatic nerve injury appears

to be inversely related to the experience of the
operating surgeon. Matta et al. (1986a) reported a
9% prevalence in a retrospective series, decreasing
to 5% in the subsequent prospective series (Matta
et al., 1986b), and further declining to 3% (Matta,
1996) without the aid of nerve monitoring. Other
authors with large numbers of acetabular fractures
operated on without nerve monitoring have reported
rates of injury of 1% or less (Middlebrooks et al.,
1997; Moed et al., 2002). These numbers compare
favorably with those reported using SEP monitoring
(Helfet et al., 1991; Vrahas et al., 1992; Baumgaertner et al., 1994; Helfet and Schmeling, 1994), which
range from 2% to 7%.
Haidukewych et al. (2002) reported on a retrospective, nonrandomized series of operatively treated acetabular fractures, comparing the results in monitored
versus unmonitored patients. In this study, a total of
140 unmonitored procedures and 112 monitored procedures were available for review and the decision
to use monitoring was at the discretion of the treating
surgeon. There were four iatrogenic sciatic palsies
(2.9%) in the unmonitored group and 10 iatrogenic
palsies (8.9%) in the monitored group (p 0.037). In
the monitored group, 76 patients were monitored with
SEPs, and 9 had iatrogenic injuries (11.8%). Thirty-six
patients were treated with a combination of SEP and
free-running EMG monitoring, and one had an iatrogenic nerve injury (2.8%). The difference in monitoring techniques was not statistically different. In 7 of
the 10 iatrogenic palsies in the monitored group, the
intraoperative monitoring was normal. An interesting
finding was that sciatic nerve injury was more common in anterior approaches in both groups. The
authors postulated that this situation was likely due to
reduction techniques for the posterior column performed with the hip flexed, placing the sciatic nerve
under tension. This is a situation (nerve stretch or compression) in which both SEP and free-running EMG
monitoring would be expected to be helpful. It was also
postulated by the authors (personal communication)
that the use of nerve monitoring had given the operating
surgeon a false sense of security, which resulted in a
more aggressive approach to fracture reduction and a
potentially greater insult to the sciatic nerve.
A small nonrandomized retrospective series using
MEP (as well as SEP) monitoring during acetabular
fracture surgery has been reported (Arrington et al.,

2000). Of the 12 patients studied, 4 were monitored
using only SEPs and 2 with only MEPs. In six
patients, both techniques were employed. In 10 neurologically intact patients with SEP monitoring,
6 had significant intraoperative SEP changes from
baseline. However, none of these changes correlated
with surgical activity, retractor placement, or traction. Therefore, no intraoperative maneuvers were
undertaken in response to these changes. One patient
awoke from anesthesia with a nerve deficit. In the
eight patients monitored with MEPs, three demonstrated MEP changes, all of which corresponded to
retractor placement or traction. Appropriate rapid
response to the abnormal signals was performed in
all three cases. However, the signals normalized in
only two of the three and all three patients awoke
from anesthesia with a peroneal nerve palsy.
55.3.3. Recommendations for monitoring neural
function during acetabular fracture surgery
Presently, intraoperative monitoring appears to be of
questionable benefit in preventing iatrogenic sciatic
nerve injury. There is no substitute for attention to
detail in the operating room. Careful patient positioning, cautious placement of retractors, limited traction
during fracture reduction, and maintaining the knee
flexed during posterior approaches (or limiting hip
flexion during anterior approaches) are all required
to minimize injury to the sciatic nerve.
55.4. Summary
Currently, intraoperative neural function monitoring
appears to be of value for pelvic ring surgery, but this
is not the case for acetabular fracture surgery. In the
future, the availability of real time, fluoroscopically
based intraoperative three-dimensional imaging may
obviate the need for monitoring during the insertion
of pelvic ring implants. In the interim, the performance of randomized prospective trials would be
helpful in further delineating the specific indications
for neural function monitoring during these operative
procedures.
References
Arrington, ED, Hochschild, DP, Steinagle, TJ, Mongan, PD
and Martin, SL (2000) Monitoring of somatosensory
761
and motor evoked potentials during open reduction
and internal fixation of pelvis and acetabular fractures.
Orthopedics, 23: 10811083.
Baumgaertner, MR, Wegner, D and Booke, J (1994) SSEP
monitoring during pelvic and acetabular fracture surgery. J. Orthop. Trauma, 8: 127133.
Benecke, JE, Jr., Calder, HP and Chadwick, G (1987)
Facial nerve monitoring during acoustic neuroma
removal. Laryngoscope, 97: 697700.
Browner, BD, Cole, JD, Graham, JM, Bondurant, FJ and
Nunchuck-Burns, SK (1987) Delayed posterior internal
fixation of unstable pelvic fractures. J. Trauma, 9:
9981005.
Calancie, B, Lebwohl, N, Madsen, P and Klose, KJ (1992)
Intraoperative evoked EMG monitoring in an animal
model: a new technique for evaluating pedicle screw
placement. Spine, 17: 12291235.
Calancie, B, Madsen, P and Lebwohl, N (1994) Stimulusevoked EMG monitoring during transpedicular lumbosacral spine instrumentation: initial clinical results.
Spine, 19: 27802786.
Cole, JD, Blum, DA and Ansel, LJ (1996) Outcome after
fixation of unstable posterior pelvic ring injuries. Clin.
Orthop., 329: 160179.
Denis, F, Davis, S and Comfort, T (1988) Sacral fractures:
an important problem. Clin. Orthop., 227: 6781.
Gantz, BJ (1985) Intraoperative facial nerve monitoring.
Am. J. Otol., 6 (Suppl. issue): 5861.
Gruson, KI and Moed, BR (2003) Injury of the femoral
nerve associated with acetabular fracture. J. Bone Joint
Surg., 85A: 428431.
Haidukewych, GJ, Scaduto, J, Herscovici, D, Sanders, RW
and DiPasquale, T (2002) Iatrogenic nerve injury in
acetabular fracture surgery: a comparison of monitored
and unmonitored procedures. J. Orthop. Trauma, 16:
297301.
Helfet, DL and Schmeling, GJ (1994) Somatosensory
evoked potential monitoring in the surgical treatment
of acute, displaced aetabular fractures: results of a prospective study. Clin. Orthop., 301: 213220.
Helfet, DL, Hissa, EA, Sergay, S and Mast, JW (1991)
Somatosensory evoked potential monitoring in the surgical management of acute aetabular fractures. J. Orthop.
Trauma, 5: 161166.
Helfet, DL, Koval, KJ, Hissa, EA, Patterson, S, DiPasquale,
T and Sanders, RW (1995) Intraoperative somatosensory
evoked potential monitoring during acute pelvic fracture
surgery. J. Orthop. Trauma, 9: 2834.
Leggon, RE, Meister, B and Lindsey, RW (2002) Inadvertent
sacral bar transfixation of the cauda equina. J. Orthop.
Trauma, 16: 127130.
Gelb, DE (1995) Triggered EMG stimulation threshold
for accuracy of pedicle screw placement: an animal
model and clinical correlation. Spine, 20: 15851591.
762
Letournel, E and Judet, R (1993) Fractures of the Acetabulum. Springer-Verlag, Berlin, 733 pp.
Maguire, J, Wallace, S, Madigan, R, Leppanen, R and
Draper, V (1995) Evaluation of intrapedunclar screw
position utilizing intraoperative evoked electromyography. Spine, 20: 10681074.
Maloney, RW, Murcek, BW, Steehle, KW, Sibly, D and
Maloney, RE (1994) A new method for intraoperative
recurrent laryngeal nerve monitoring. Ear Nose Throat
J., 73: 3033.
Matta, JM (1996) Fractures of the acetabulum: accuracy of
reduction and clinical results in patients managed operatively within three weeks after the injury. J. Bone Joint
Surg., 78A: 16321645.
Matta, JM and Saucedo, T (1989) Internal fixation of pelvic ring fractures. Clin. Orthop., 242: 8397.
Matta, JM and Tornetta, P, III (1996) Internal fixation of
unstable pelvic ring injuries. Clin. Orthop., 329: 129140.
Matta, JM, Anderson, LM, Epstein, HC and Hendricks, P
(1986a) Fractures of the acetabulum: early results of a
prospective study. Clin. Orthop., 205: 230240.
Matta, JM, Mehne, DK and Roffi, R (1986b) Fractures of
the acetabulum: a retrospective analysis. Clin. Orthop.,
205: 241250.
Middlebrooks, ES, Sims, SH, Kellam, JF and Bosse, MJ
(1997) Incidence of sciatic nerve injury in operatively
acetabular fractures without somstosensory evoked
potential monitoring. J. Orthop. Trauma, 11: 327329.
Moed, BR (2003) Significance of anode location
for stimulus-evoked electromyography during iliosacral
screw placement. J. Orthop. Trauma, 17: 597598.
Moed, BR and Geer, B (2006) S2 iliosacral screw fixation
for disruptions of the posterior pelvic ring: a report of
49 cases. J. Orthop. Trauma, 20(6): 378383.
Moed, BR, Maxey, JW and Minster, GJ (1992) Intraoperative somatosensory evoked potential monitoring of the
sciatic nerve: an animal model. J. Orthop. Trauma, 6:
5965.
Moed, BR, Anders, MJ, Ahmad, BK, Craig, JG and
Jacobson, GP (1998a) Intraoperative stimulus-evoked
electromyographic monitoring for placement of iliosacral implants: an animal model. J. Orthop. Trauma,
12: 8589.
Moed, BR, Ahmad, BK, Craig, JG, Jacobson, GP and
Anders, MJ (1998b) Intraoperative monitoring with
stimulus-evoked electromyography during placement
of iliosacral screws; an initial clinical study. J. Bone
Joint Surg., 80A: 537546.
Moed, BR, Hartman, MJ, Ahmad, BK, Cody, D and Craig,
JG (1999) Evaluation of intraoperative nervemonitoring during insertion of an iliosacral implant in
an animal model. J. Bone Joint Surg., 81A: 15291537.
Moed, BR, Carr, SEW and Watson, JT (2002) Results of
operative treatment of fractures of the posterior wall
of the acetabulum. J. Bone Joint Surg., 84A: 752758.
B.R. MOED
Moed, BR, Kellam, JF, McLaren, A and Tile, M (2003)
Internal fixation for the injured pelvic ring. In: M Tile,
DL Helfet and JF Kellam (Eds.), Fractures of the Pelvis
and Acetabulum. Lippincott, Williams & Wilkins, Philadelphia, pp. 217293.
Moed, BR, Kopec, MA and Barnett, DW (2006) A Finite
Element Study Of Electrode Location During Stimulus
Evoked Electromyographic Monitoring Of Iliosacral
Screw Insertion. Orthopaedic Research Society 52nd
Annual Scientific Meeting. Chicago, IL, March 1821,
2006.
Mller, AR and Jannetta, PJ (1984) Preservation of facial
function during removal of acoustic neuromas. J. Neurosurg., 61: 757760.
Mostafavi, HR and Tornetta, P, III (1996) Radiologic evaluation of the pelvis. Clin. Orthop., 329: 614.
Nash, CL, Jr. and Brown, RH (1989) Current concepts
review Spinal cord monitoring. J. Bone Joint Surg.,
71A: 627630.
Reilly, MC, Bono, CM, Litkouhi, B, Sirkin, M and Behrens, F (2003) The effect of sacral malreduction on the
safe placement of iliosacral screws. J. Orthop. Trauma,
17: 8894.
Ricci, WM, Padberg, AM and Borrelli, J (2003) The significance of anode location for stimulus-evoked electromyography during iliosacral screw placement. J. Orthop.
Trauma, 17: 9599.
Rice, DH and Cone-Wesson, B (1991) Intraoperative recurrent laryngeal nerve monitoring. Otolaryngol. Head
Neck Surg., 105: 372375.
Routt, MLC, Jr. and Simonian, PT (1996) Closed reduction
and percutaneous skeletal fixation of sacral fractures.
Clin. Orthop., 329: 121128.
Routt, MLC, Simonian, PT, Agnew, SG and Mann, FA
(1996) Radiographic recognition of the sacral alar
slope for optimal placement of iliosacral screws: a cadaveric and clinical study. J. Orthop. Trauma, 10: 171177.
Routt, MLC, Simonian, PT and Inaba, J (1997) Iliosacral
screw complications. Oper. Tech. Orthop., 7: 206220.
Schmeling, GJ, Perlewitz, TJ and Helfet, DL (2003) Early
complications of acetabular fractures. In: M Tile, DL Helfet and JF Kellam (Eds.), Fractures of the Pelvis and Acetabulum. Lippincott, Williams & Wilkins, Philadelphia,
pp. 729740.
Templeman, D, Schmidt, A, Freese, J and Weisman, I (1996a)
Proximity of iliosacral screws to neurovascular structures
after internal fixation. Clin. Orthop., 329: 194198.
Templeman, D, Goulet, J, Duwelius, PJ, Olson, S and
Davidson, M (1996b) Internal fixation of displaced fractures of the sacrum. Clin. Orthop., 329: 180185.
Tile, M (2003) Describing the injury: classification of
pelvic ring injuries. In: M Tile, DL Helfet and JF
Kellam (Eds.), Fractures of the Pelvis and Acetabulum. Lippincott, Williams & Wilkins, Philadelphia,
pp. 130167.

Van den Bosch, EW, Van Zwienen, CAM and Van Vugt, AB
(2002) Fluoroscopic position of sacroiliac screws in 88
patients. J. Trauma, 53: 4448.
Vrahas, M, Gordon, RG, Mears, DC, Krieger, D and Sclabassi, RJ (1992) Intraoperative somatosensory evoked
potential monitoring of pelvic and acetabular fractures.
J. Orthop. Trauma, 6: 5058.
Webb, LX, De Araujo, W, Donofrio, P, Santos, C,
Walker, FO, Olympio, MA and Haygood, T (2000)
763
Electromyography monitoring for percutaneous placement of iliosacral screws. J. Orthop. Trauma, 14:
245254.
Ziran, BH, Smith, WR, Towers, J and Morgan, SJ (2002)
Iliosacral screw fixation of the posterior pelvic ring
using local anaesthesia and computerized tomography.
J. Bone Joint Surg., 85B: 411418.

M.R. Nuwer (Ed.)
764
CHAPTER 56

peripheral nerve surgery
Huan Wanga and Robert J. Spinnerb,*
a
Visiting Fellow, Mayo Clinic, and Department of Hand Surgery, Huashan Hospital, Fudan University,
Shanghai, Peoples Republic of China
b
Mayo Clinic, Brachial Plexus Clinic, Rochester, MN 55905, USA
Intraoperative electrophysiology has become an

important neurological extension to surgeons. It
provides vital information: to guide intraoperative
decisions about the type of neural reconstruction to
perform; to localize the site of injury; and to avoid
unnecessary injury to nerve during procedures
directly involving or nearby nerve. In short, it allows
surgeons to assess, predict, and preserve nerve function thereby improving operative outcomes.
Intraoperative electrophysiology is valuable during
operations related to peripheral nerve injury, nerve
entrapment, tumor biopsy or resection, and other
reconstruction. In these situations, it is particularly
useful to assess neuromas in continuity, to evaluate
the site and degree of compression during nerve decompression, to prevent injury to affected or neighboring nerve(s) during tumor resection, or to minimize
neural complications from fascicular biopsy, fascicular nerve transfer or during high-risk procedures
around nerve.
56.1. Injury
56.1.1. Neuromas in continuity
The most frequently seen injury is one that leaves
peripheral nerve in continuity. Judgment of the severity of the lesion and potential for useful spontaneous
recovery is difficult yet critical to surgical decision
making and prognosis. Standard modes of assessment of nerve, including visualization and palpation
of the nerve, are unreliable and unpredictable in
*
Correspondence to: Robert J. Spinner, M.D., Mayo Clinic,

Brachial Plexus Clinic, Gonda 8S, 200 First Street SW,
Tel.: 1-507-284-2376; fax: 1-507-284-5206.
E-mail: spinner.robert@mayo.edu (R.J. Spinner).
determining the potential for meaningful recovery

in a given neuroma. As they say You cant judge a
book by its cover. So too, one cannot predict internal histology or neural recovery based on the external
appearance of an injured nerve.
At this time, the best developed and most reliable
method of intraoperative assessment involves electrophysiology, and in our opinion, rests on nerve action
potential (NAP) recordings (Kline and Hudson,
1995). NAP recordings yield reliable and reproducible
results for detecting the presence or absence of spontaneous nerve regeneration across a lesion in continuity,
months earlier than when standard electromyography
(EMG) would reveal recovery at a target muscle. Of
course, this benefit of NAP recordings is magnified
when the injury affects nerve(s) with distal targets.
The presence of a NAP across a lesion suggests preserved axonal function or meaningful regeneration
that frequently can be translated into significant clinical recovery. Experimental studies in primates have
revealed that 4,0005,000 myelinated axons are necessary to obtain such a response (Kline et al., 1969).
NAPs can be performed as early as 68 weeks following nerve injury. For these reasons, we prefer using
NAPs to other techniques such as compound muscle
action potentials (CMAPs).
Circumferential mobilization of nerve(s) across a
relatively long segment of nerve is completed to
allow ample exposure. When necessary, NAP recordings can be performed on a relatively short segment
(as little as 34 cm) of a nerve or part of a nerve
but may be difficult because of artifact. Recordings
should be first obtained on functioning, neighboring
nerves, if accessible in the operative field, or on the
affected nerve proximal to the lesion. Obtaining normal tracings allows the surgeon and electrophysiologist to test the system, as technical problems are
not uncommon. A careful dialogue between the surgeon, the technician, and the electrophysiologist is
necessary. Once these normal tracings are obtained,
one evaluates the pathologic segment of nerve. The
surgeon separates the electrodes and moves the distal
one into the zone of injury to see if a NAP response
is preserved. If a response is obtained, he or she
moves the distal electrode distal to the injury and
see if and how far distal a NAP can be obtained.
Muscle contraction is also looked for (Kline and
Hudson, 1995; Tiel et al., 1996). If a response is
not obtained at the level of the injury, one can inch
toward this zone injury; typically inching toward this
zone of injury allows a prediction of the location
where nerve is viable (and correlates where the neuroma will need to be resected to find healthy fascicles for grafting).
This same approach can be applied to part of a
nerve, such as individual or groups of fascicles.
Injury may not involve the entire cross section of a
nerve uniformly. In this situation, internal neurolysis
can be done followed by NAP recordings on different
fascicular groups. A split repair can be attempted:
preserving the functioning portion of nerve and
resecting and reconstructing the nonfunctioning
segment (Williams and Terzis, 1976; Kline and
Hudson, 1995).
Neurolysis based on NAP recordings across a lesion
in continuity has led to good functional recovery in
90% of cases. These data have been reproduced in
various large series of different nerves in the upper
and lower limbs (summarized in Kline and Hudson,
1995). In contrast, an absent NAP across a lesion
predicts failure which also has been substantiated
clinically over the years. Resection and reconstruction
of the lesion based on an absent NAP have been justified by the histological confirmation of neurotmetic
changes (Grade 4 Sunderland changes) (Kline and
Hudson, 1995; Burg et al., 2002).
NAP recordings thereby guide surgeons: when
present to perform neurolysis (Fig. 1); when absent
to perform other means of nerve reconstruction [typically with nerve grafts (Fig. 2), though occasional
with direct nerve repair (Fig. 3) or nerve transfer
(see below)]. Based on this simple algorithm,
surgeons can achieve predictable results based on
others experiences. Conversely, neurolysis alone
should not be performed in cases where a NAP is
absent in a nerve with an enlarged caliber, as a
crucial period for nerve regeneration is wasted by a
procedure that is doomed to fail (Kline and Hudson,
765
Fig. 1. This patient underwent exploration of the common

peroneal nerve 6 months following a knee dislocation with
ligamentous injury. A complete peroneal nerve lesion was
documented clinically and electrophysiologically. Despite
the extensive longitudinal injury (arrows) to the common
peroneal nerve (P) and large neuroma in continuity, to
our surprise, a nerve action potential (NAP) was obtained.
The deep and superficial peroneal nerve branches have
been identified at the inferior portion of the figure and
are protected in vasoloops. The patient went on to make
complete full neural recovery after neurolysis over the next
18 months. T tibial; S sciatic.
1995; Oberle et al., 1997); nerve grafting should

not be performed based on the scarred external
appearance of a neuroma in continuity in the face
of a preserved NAP.
56.1.2. Nerve transection
Surgeons may use several different techniques to
orient nerve ends accurately during early repair.
Commonly applied methods include using epineurial
vessel alignment, gross fascicular matching or
knowledge of nerve fascicle topography to help
Fig. 2. This patient presented with painless atrophy of the deltoid 4 months after shoulder arthroscopy done for impingement. A relatively minor, focal injury to the axillary nerve was encountered (arrow) (in contrast to Fig. 1). No nerve action
potential (NAP) recording could be obtained across the lesion. Short interpositional sural nerve grafts were placed. The
patient regained useful shoulder function over 2 years.
Fig. 3. This young woman presented 4 months after an arthroscopic shoulder stabilization procedure with a complete, painful axillary neuropathy. Preoperatively, the lesion was thought to be at an infraclavicular level. No scarring was seen. Nerve
action potentials (NAPs) could be obtained from the posterior cord to the infraclavicular level axillary nerve. No deltoid
contraction or compound muscle action potential (CMAP) could be obtained across the lesion. A: The axillary nerve was
then explored at the other side of the quadrangular space through a separate posterior arm incision. A capsular suture inadvertently encircled the axillary nerve (arrow). B: No NAP was obtained before or after the suture was removed. The focal
lesion was resected (inset). C: Because of redundancy in the nerve at this more distal site, a direct repair of the axillary
nerve (arrowhead) could be accomplished.
align the proximal and distal ends (Sunderland, 1945;

Jabaley et al., 1980). Others define motor and sensory axons with the aid of histochemical staining
using acetylcholinesterase and carbonic anhydrase
markers, respectively (Lang et al., 1991). In the
setting of acute nerve injuries, some use intraoperative fascicular stimulation of the proximal stump
in the awake patient to map out sensory bundles
by elicitating paresthesias or dysesthesias. The motor
(and by exclusion, sensory bundles) in the distal
end can be mapped if the transection is as fresh as
34 days in that muscle twitch can be seen when
the motor bundles are stimulated (Jabaley, 1991).
56.2. Entrapment
The exact site of peripheral nerve entrapment can
often be determined with clinical findings, electrodiagnostic and magnetic resonance imaging (MRI)
studies. Sometimes, the precise localization of the
entrapment cannot be determined preoperatively.
Additional information can be obtained from direct
nerve recordings across a shorter segment of nerve.
In such cases, slowing of conduction velocities can
be determined intraoperatively that may not have been
apparent on preoperative nerve conduction studies
across a longer segment. Such use of intraoperative
electrophysiology to conduct inching studies can help
define the exact location(s) of nerve entrapment and
facilitate more precise and thorough decompression.
For example, these studies have been employed by
some in operations for ulnar nerve entrapment at the
elbow, the second most common mononeuropathy
(Campbell et al., 1988; Oberle et al., 1994; Kim
et al., 2003). At the elbow region, the conduction
slowing nearly universally has been demonstrated to
occur at the level of the olecranon groove in cases
of primary ulnar nerve entrapment (Kline and
Hudson, 1995) rather than at other potential sites of
compression described at this anatomic region. For
patients undergoing operative intervention for thoracic outlet syndrome, intraoperative NAPs have
demonstrated slowing proximally affecting C8 and
T1 most commonly (Kline and Hudson, 1995;
Tender et al., 2004).
While we recognize the attributes of intraoperative electrophysiological testing of patients with
nerve compression syndromes, we do not routinely
utilize it as we do not believe its results alter the type
of operative management (namely nerve decompression) in the vast majority of our patients.
767
56.3. Tumors
Some surgeons utilize intraoperative NAPs in tumor
resection. They have demonstrated that NAP recordings across the individual fascicle(s) involved in
certain types of nerve sheath tumors are not present;
these findings suggest that these tumor fascicles are
nonfunctional compared with the remainder of the
nerve, which is often functioning normally.
We often use nerve stimulation by itself or together
with evoked EMG recordings to help us preserve
important motor function during the resection of primary (benign or malignant) or metastatic peripheral
nerve tumors. The use of a portable, fine-tipped monopolar stimulator at a low setting (e.g., 0.5 mA) allows
electrical mapping of the tumor surface to identify a safe
zone to remove the tumor. The focal stimulation of
fascicles produces an evoked response, whereas the
stimulation of those directly involving the tumor
produce no evoked response. Recording EMG activity
in distal muscles while individual or groups of fascicles
are stimulated during the dissection of the tumor, or
hearing discharges helps guide the path of dissection
and the extent of resection. The use of these techniques
allows surgeons to distinguish between functioning
and nonfunctioning tissue. The nerve tumor can then
be resected with as little damage to the normal nerve
fascicles as possible and maximal functional retention
of the involved limb is achieved. Other surgeons may
also employ somatosensory evoked potentials (SEPs)
to monitor sensory function during tumor resection
(Kwok et al., 2005).
For a schwannoma, the most common nerve sheath
tumor, typically there is a small single entering
and exiting non-functional fascicle. Resection of a
schwannoma at a fascicular level can often be accomplished without neurological deficit (Fig. 4). For nonplexiform neurofibromas, there may be several or more
fascicles involved in the tumor. Resection of the tumor
can still be achieved in many cases without deficit
(Figs. 5 and 6). In some cases, residual tumor may be
left behind. In other cases, interpositional grafting may
be necessary and deficits may be lessened over time.
For less cooperative neural lesions of a benign
(such as perineuriomas) or malignant nature (such
as malignant peripheral nerve sheath tumors),
intraoperative monitoring may be employed to preserve as much neural structure as possible during biopsy
(see below) or resection. For tumors that are in proximity to major nerves, intraoperative stimulation may
allow the detection and protection of these nerves.
Fig. 4. This 52-year-old woman presented with a slowly enlarging axillary mass. A: Paracoronal T1-weighted magnetic
resonance (MR) image demonstrates a mass (*) with its entering terminal branch being displaced around it. B: Axial
T2-weighted MR image with fat saturation shows the mass (*) with its speckled appearance typical of a neurogenic tumor.
C: At operation, the medial antebrachial cutaneous nerve is identified during the initial exposure. After proximal and distal
control of the median nerve, the nerve sheath tumor is mobilized. A safe zone () is identified to incise the epineurium
longitudinally. A wand-type stimulator can be used to map out fascicles. The functioning fascicles (arrows) are swept away
from the mass. D: The entering and exiting fascicles (arrowheads) to the tumor are identified. These are nonfunctional. The
well-encapsulated schwannoma is resected in toto. Neurological function is thereby preserved.
769
Fig. 5. This patient with neurofibromatosis type 1 (NF-1) presented with severe neuropathic pain from a large sciatic notch
tumor. Initial interpretation of magnetic resonance (MR) images performed at an outside facility suggested that the mass
was unresectable without significant neurological sequalae to the sciatic nerve due to its predicted direct involvement in
it. High-resolution axial T2 MR image with fat saturation suggested that the extensive neurogenic tumor displaced the nerve
[from Spinner et al. (2006) with permission from the American Association of Neurological Surgeons].
56.4. Intraoperative testing to avoid iatrogenic

injury to peripheral nerve
Selection and resection of a portion of a nerve can be
accomplished for diagnostic or therapeutic benefits.
Intraoperative electrophysiological monitoring is
helpful in decreasing the morbidity associated with
many operations directly involving peripheral nerve
or ones performed in the vicinity of nerves (where
they may be at risk).
56.4.1. Fascicular biopsy
Fascicular biopsy of a major proximal (mixed) nerve
can be performed successfully and safely in patients
with progressive neurological deficit of an idiopathic
nature (Dyck et al., 2003). Often these patients have
undergone previous nondiagnostic distal cutaneous
nerve biopsy (such as the sural or superficial peroneal nerve) and have received empiric medical or
surgical treatment without benefit. The localization
of the nerve fascicle(s) is selected based on clinical
and electrophysiological examinations and abnormalities (typically subtle) on high-resolution MRI.
Surgical accessibility is also taken into consideration.
A nonfunctioning fascicle is chosen using nerve
stimulation or evoked EMG techniques. The fascicle
is then submitted for pathologic investigation (Fig. 7).

In our experience with over 100 cases (excluding mass
lesions) in the past 5 years, a diagnosis of a pathologic
alteration has been established in over 80% of cases
with a permanent complication rate <5%. Various diagnoses have been established including inflammatory/
immune suggestive of vasculitis and inflammatory
demyelinating disorders, perineurioma, sarcoidosis,
lymphoma, metastatic adenocarcinomas, among others;
these diagnoses have had therapeutic implications.
A team of individuals with expertise in peripheral nerve
disorders, MRI of nerve, peripheral nerve surgery, and
pathology is needed.
56.4.2. Fascicular nerve transfer
We often utilize similar intraoperative techniques to
help us choose fascicles when performing nerve
transfer (neurotization). For example, when we try
to reconstruct a C5,6 brachial plexus lesion, we
may select the Oberlin transfer. This nerve transfer
is not only useful in cases of preganglionic injury in
which direct nerve grafting is not an option; it may
also be preferable to nerve grafting in cases where
the patient presents late or where more rapid reinnervation is desirable. An expendable fascicle or two of
the ulnar nerve is selected and is coapted directly to
770
H. WANG AND R.J. SPINNER
Fig. 6. Through a combined transabdominal and posterior transgluteal approach, the tumor demonstrated in Fig. 5 was
completely resected. Intraoperative electromyography (EMG) was utilized to ensure that the integrity of the lumbosacral
plexus was maintained. Neurological function was preserved. A: The anterior approach to the pelvis allowed mobilization
of the tumor (*) from the lumbosacral plexus (in vasoloops). B: The posterior approach to the buttock allowed exposure of
the sciatic notch and protection of the sciatic nerve (in vasoloops). A small portion of the tumor (*) can be seen penetrating
the gluteus maximus. C: The posterior approach allows the safe resection of the masses (* the largest mass). D: The large
sciatic notch dumbbell neurofibroma (*) is delivered through the posterior approach. The patient has had complete pain
relief which has been maintained at 2-year follow-up [from Spinner et al. (2006) with permission from the American
Association of Neurological Surgeons].
the biceps branch close to the end-organ (the Oberlin

transfer) (Fig. 8) (Liverneaux et al., 2006). A fascicle
to the flexor carpi ulnaris is chosen (as other wrist
flexors are available), and fascicles to more vital
structures such as hand intrinsics are preserved.
Either a disposable stimulator or a fine monopolar
stimulator can be utilized with an examiner looking
and feeling for distal muscular contraction. In addition, sterile percutaneous EMG needles can be placed
in individual ulnar-innervated muscles to help guide
fascicular selection without adding much additional
time to the operation. This type of nerve reconstruction has improved outcomes significantly. In some
series, 90% of patients may regain MRC 4 function.
The procedure can be done safely with little if any
permanent morbidity. Temporary paresthesias in the

ulnar nerve distribution are common and resolve
within several months. Similarly, we and others have
utilized this manner of nerve stimulation or evoked
EMG to help select other expendable fascicles
(from the median nerve), nerve branches (from the
triceps), or even nerves [a portion or the entire contralateral C7 (Holland and Belzberg, 1997)] for nerve
reconstruction.
56.5. Intraoperative monitoring of nerves when
performing other procedures around nerve
Some surgeons use a host of intraoperative electrophysiological techniques to help them identify
771
Fig. 7. This teenager presented with a 6-year history of a progressive, severe footdrop. She was found to have evidence of a
sciatic neuropathy, predominantly affecting the peroneal nerve division. A: T1-weighted image demonstrates the enlarged
and mass-like peroneal division (arrow) of the sciatic nerve at the level of the buttock. There is no fat between the fascicles
of the peroneal nerve. Contrast the unaffected tibial division (arrowhead) which is typically larger than the peroneal nerve.
B: At operation, a fascicular biopsy of the peroneal (P) division is targeted in the buttock region. The fascicle is tapered at
its proximal and distal ends and enlarged in its mid-portion. This fascicle is nonfunctional. The tibial (T) division is left
undisturbed. Inset, resected specimen. C: Pathology reveals pseudo-onion bulbs consistent with a diagnosis of a perineurioma. This diagnosis was confirmed with immunohistochemistry and electron microscopy.
changes in neural activity while performing certain

procedures in which nerve(s) may be at risk. This
may include elective or traumatic procedures in the
limbs, pelvis, spine, and head and neck region. For
example, orthopedists may use these techniques during open or arthroscopic procedures. Some have
reported their experience using these techniques for
half pin placement in external fixation of limb fractures, closed humeral nailing, primary or revision
arthroplasty at various joints (Sutherland et al.,
1996), acute limb deformity correction, acetabular
osteotomy and shoulder capsular shrinkage techniques (Makarov et al., 1996, 1997, 2003; Mills
et al., 2000; Brown et al., 2002; Esmail et al.,
2005). Specific chapters in this book will summarize
other uses of intraoperative monitoring in spine and

intracranial operations.
EMG and SEP have been utilized intraoperatively
to detect electrophysiological changes caused by surgical manipulation, suggesting the onset of neural
compromise. Ideally, these alterations may alert the
surgeon to a potential neural complication, and guide
intervention when the neurological deficit is reversible. For example, monitoring of evoked EMG potentials is also practiced since detecting changes of
CMAP amplitude enables serial real-time assessment
of nerve conduction. A decrease in motor potential
amplitude provides quantitative information about
possible nerve injury. This not only serves as surveillance of nerve at risk. It may also provide a cut-off
772
H. WANG AND R.J. SPINNER
Fig. 8. Oberlin transfer. A: An expendable fascicle of the ulnar nerve (UN) supplying the flexor carpi ulnaris is selected and
transferred to the biceps branch (B) to achieve elbow flexion in a patient with a C5,6 lesion. B: Artistic drawing of this
technique (with permission of Mayo Foundation).
line or a safety zone for surgical maneuver, as well as

evidence for approaches that can effectively prevent
nerve injury (Ljung et al., 1995; Katz et al., 2004).
While some surgeons have found intraoperative
electrophysiological monitoring useful in decreasing
iatrogenic nerve injury (Baumgaertner et al., 1994);
others have not appreciated a difference in their
practice (Middlebrooks et al., 1997; Haidukewych
et al., 2002). The application of intraoperative monitoring remains controversial and is largely dependent
on surgeons preference. Clearly large, prospective
studies need to be performed to measure potential
advantages.
56.6. Discussion
Intraoperative electrophysiology serves as a valuable
adjunct to surgery when it is properly applied. It thus
must be tailored to each patients clinical condition
and to the specific surgical risks. The choice of
technique must fulfill the goals of intraoperative
electrophysiological testing. Successful application
of electroneurologic testing also depends largely on
the techniques of intraoperative recordings. Both
the surgeon and electrophysiologist should have the
knowledge of the special features and limitations of
the methods, and the expertise to interpret the electrodiagnostic results.
While there are many advantages to using intraoperative electrophysiology routinely, we acknowledge
the disadvantages as well. Routine use of such

technology is labor intensive, adds time and expense
to the operation, and demands additional help of others
and equipment. False negatives should be ruled
out taking into consideration conditions such as tourniquet application, anesthesia status, neuromuscular
blockade, and low-wound temperature (Kline and Hudson, 1995). Unwanted current spread to adjacent nerves
can cause false positives in stimulus-triggered electromyography, while neurotonic discharges may be
recorded in free-running electromyographic monitoring
during cold water irrigation within the surgical field
even though no mechanical insult to the monitored
nerve is present (Holland, 2002).
Finally, intraoperative electrophysiological testing
should not be considered as a replacement but rather
as a complement to preoperative clinical, electrophysiological and imaging evaluations, and a thorough intraoperative morphological examination.
References
Baumgaertner, MR, Wegner, D and Booke, J (1994) SSEP
monitoring during pelvic and acetabular surgery. J.
Orthop. Trauma, 8: 127133.
Brown, DM, McGinnis, WC and Mesghali, H (2002) Neurophysiologic intraoperative monitoring during revision
total hip arthroplasty. J. Bone Joint Surg., 84A: 5661.
Burg, D, Infanger, M, Meuli-Simmen, C, Stallmach, T, Beer,
G, Amgwerd, S and Meyer, VE (2002) Methods,

indications and validation of intraoperative nerve conductivity testing. Handchir. Mikrochir. Plast. Chir., 34: 316.
Campbell, WW, Sahni, SK, Pridgeon, RM, Riaz, G and
Leshner, RT (1988) Intraoperative electroneurography:
management of ulnar neuropathy at the elbow. Muscle
Nerve, 11: 7581.
Dyck, PJB, Amrami, KK, Spinner, RJ, Klein, CJ, Engelstad,
J and Dyck, PJ (2003) Fascicular biopsy of proximal
nerves in selected cases with MRI abnormality is diagnostically informative. J. Peripher. Nerv. Syst., 8: 14.
Esmail, AN, Getz, CL, Schwartz, DM, Wierzbowski, L,
Ramsey, ML and Williams, GR, Jr. (2005) Axillary
nerve monitoring during arthroscopic shoulder stabilization. Arthroscopy, 21: 665671.
Haidukewych, GJ, Scaduto, J, Herscovici, D, Jr., Sanders,
RW and DiPasquale, T (2002) Iatrogenic nerve injury
in acetabular fracture surgery: a comparison of monitored and unmonitored procedures. J. Orthop. Trauma,
16(5): 297301.
Holland, NR (2002) Intraoperative electromyography. J.
Clin. Neurophysiol, 19(5): 444453.
Holland, NR and Belzberg, AJ (1997) Intraoperative electrodiagnostic testing during cross-chest C7 nerve root
transfer. Muscle Nerve, 20(7): 903905.
Jabaley, ME (1991) Electrical nerve stimulation in the
awake patient. In: RE Gelberman (Ed.) , Operative
Nerve Repair and Reconstruction. Lippincott, Philadelphia, PA, pp. 241257.
Jabaley, ME, Wallace, WH and Heckler, FR (1980) Internal topography of major nerves of the forearm and
hand: a current view. J. Hand Surg., 5: 118.
Katz, K, Attias, J, Weigl, D, Cizger, A and Bar-on, E
(2004) Monitoring of the sciatic nerve during hamstring
lengthening by evoked EMG. J. Bone Joint Surg. (Br.),
86: 10591061.
Kim, DH, Han, K, Tiel, RL, Murovic, JA and Kline, DG
(2003) Surgical outcomes of 654 ulnar nerve lesions.
J. Neurosurg, 98(5): 9931004.
Kline, DG and Hudson, AR (1995) Nerve Injuries: Operative Results for Major Nerve Injuries, Entrapments, and
Tumors. W.B. Saunders, Philadelphia, PA, 611 pp.
Kline, DG, Hackett, ER and May, PR (1969) Evaluation of
nerve injuries by evoked potentials and electromyography. J. Neurosurg., 31(2): 128136.
Kwok, K, Slimp, JC, Born, DE, Goodkin, R and Kliot, M
(2005) Evaluation and management of benign peripheral nerve tumors and masses. In: Textbook of NeuroOncology. W.B. Saunders/Elsevier Science, Philadelphia, PA, pp. 535563.
Lang, DH, Lister, GD and Jevans, AW (1991) Histological
and biochemical aids to nerve repair. In: RE Gelberman
(Ed.) , Operative Nerve Repair and Reconstruction.
Lippincott, Philadelphia, PA, pp. 259271.
Liverneaux, PA, Diaz, LC, Beaulieu, JY, Durand, S and
Oberlin, C (2006) Preliminary results of double nerve
773
transfer to restore elbow flexion in upper type brachial
plexus palsies. Plast. Reconstr. Surg., 117(3): 915919.
Ljung, P, Ahlmann, S, Knutson, K, Rosen, I and Rydholm,
U (1995) Intraoperative monitoring of ulnar nerve function during replacement of the rheumatoid elbow via the
lateral approach. Acta Orthop. Scand., 66: 132136.
Makarov, MR, Delgado, MR, Birch, JG and Samchukov,
ML (1996) Intraoperative SSEP monitoring during
external fixation procedures in the lower extremities.
J. Pediatr. Orthop., 16: 155160.
Makarov, MR, Delgado, MR, Birch, JG and Samchukov,
ML (1997) Monitoring peripheral nerve function during
external fixation of upper extremities. J. Pediatr.
Orthop., 17: 663667.
Makarov, MR, Samchukov, ML, Birch, JG, Johnston, CE,
Delgado, MR, Rampy, PL and Van Allen, EM (2003)
Acute deformity correction of lower extremities under
SSEP-monitoring control. J. Pediatr. Orthop., 23:
470477.
Middlebrooks, ES, Sims, SH, Kellam, JF and Bosse, MJ
(1997) Incidence of sciatic nerve injury in operatively
treated acetabular fractures without somatosensory evoked potential monitoring. J. Orthop. Trauma, 11: 327329.
Mills, WJ, Chapman, JR, Robinson, LR and Slimp, JC
(2000) Somatosensory evoked potential monitoring during closed humeral nailing: a preliminary report. J.
Orthop. Trauma, 14: 167170.
Oberle, JW, Rath, SA and Richter, HP (1994) Intraoperative electrically evoked nerve action potentials in
ulnar entrapment syndrome. Zentralbl. Neurochir., 55:
102109.
Oberle, JW, Antoniadis, G, Rath, SA and Richter, HP
(1997) Value of nerve action potentials in the surgical
management of traumatic nerve lesions. Neurosurgery,
41(6): 13371342.
Spinner, RJ, Endo, T, Amrami, KK, Dozois, EJ, BabovicVuksanovic, D and Sim, FH (2006) Resection of benign
sciatic notch dumbbell-shaped tumors. J. Neurosurg.,
105(6): 873880.
Sunderland, S (1945) The intraneural topography of the
radial, median, and ulnar nerves. Brain, 68: 243298.
Sutherland, CJ, Miller, DH and Owen, JH (1996) Use of
spontaneous electromyography during revision and complex total hip arthroplasty. J. Arthroplasty, 11(2):
206209.
Tender, GC, Thomas, AJ, Thomas, N and Kline, DG
(2004) Gilliatt-Sumner hand revisited: a 25-year experience. Neurosurgery, 55(4): 883890.
Tiel, RL, Happel, LT, Jr. and Kline, DG (1996) Nerve
action potential recording method and equipment. Neurosurgery, 39: 103109.
Williams, HB and Terzis, JK (1976) Single fascicular
recordings: an intraoperative diagnostic tool for the
management of peripheral nerve lesions. Plast.
Reconstr. Surg., 57(5): 562569.
Section III.4
Vascular Surgery

M.R. Nuwer (Ed.)
776
CHAPTER 57
Intraoperative monitoring during carotid endarterectomy

Jean-Michel Guerit*
Clinique Edith Cavell, Unite dExplorations Electrophysiologiques du Syste`me Nerveux, CHIREC, B-1050, Brussels, Belgium
57.1. Neurological and neuropsychological

outcome of carotid surgery
Stroke is the most feared complication of carotid artery
surgery. This is paradoxical as it is precisely what
carotid endarterectomy (CEA) aims at preventing. In
major, current trials of CEA, stroke rate ranges from
2% to 7.5% (Sila, 1998). Hence, while evidence-based
studies have established CEA as effective for recently
symptomatic patients with 7090% internal carotid
artery stenosis, CEA effectiveness may depend on
the actual risk in other situations. Thus, CEA is recommended in patients with 5069% symptomatic stenosis
provided that the rate of stroke/death is <6%, and in
6099% asymptomatic stenosis provided that the risk
is <3% (Chaturvedi et al., 2005). Similarly, there is
currently a debate over whether endovascular treatment of carotid artery stenosis should be preferred to
CEA. This debate continues because of the absence
of significant differences in the major risks of treatment, even if minor complication rates would slightly
favor endovascular treatments (Coward et al., 2005).
That is, any intraoperative monitoring (IOM) technique decreasing the stroke/death rate does extend
the range of CEA indications.
The neurological complications of carotid surgery
depend on both intra- and postoperative factors
(Table 1). Intraoperatively, two pathophysiological
mechanisms may be involved: emboli, which are dislodged from the plaque during its manipulation, and
hemodynamic disturbances consecutive to a drop in
systemic blood pressure (BP) or to carotid crossclamping (CCC) (Ackerstaff et al., 1995). Myocardial infarction is another intraoperative risk that is
linked to the high prevalence of ischemic heart
*
Correspondence to: Jean-Michel Guerit, M.D., Ph.D.,

Stress and Craniofacial Pain Clinic, Avenue Louise 390,
Tel.: 32-2-6480081; mobile: 32-477-390305.
E-mail: Jean-Michel.Guerit@chirec.be (J.-M. Guerit).
disease among patients with carotid atheroma (Moore

et al., 1995; Yeager et al., 1989). Two mechanisms
can be involved postoperatively: brain hemorrhage
and carotid restenosis (Sila, 1998; De Borst et al.,
2001). Each mechanism can be prevented, but at the
expense of an increased exposure to other risks. Thus,
the probability of embolism can be decreased by a
more careful manipulation of the carotid artery during
the operation and that of restenosis by angioplasty
(venous or synthetic patch). However, both procedures
increase the cross-clamping time, thereby increasing
the risk of stroke in cases of hypoperfusion. The ischemic consequences of a drop in systemic BP can be prevented by restoring a higher BP, which theoretically
increases the risk of myocardial infarction. The risk
of stroke due to CCC can be prevented by a shunt,
which increases the risk of embolism.
This implies that all measures aiming at restoring a
sufficient cerebral blood flow (CBF) should be applied
only if they are required. There is a strong argument
indicating that in the majority of cases they are not
required. Thus, based on our retrospective experience,
more than 85% of patients can undergo a 30-min
period of CCC without any neurological sequel and a
100% occlusion of the contralateral carotid artery
remains compatible in 50% of cases with a safe ipsilateral CCC for as long as 30 min (Guerit et al., 1997;
Witdoeckt et al., 1997). Moreover, the individual susceptibility to CCC or to a drop in BP is highly variable,
as it depends on the degree of ipsi- and contralateral
stenosis, the status of the circle of Willis, the overall
degree of atherosclerosis, the status of self-regulatory
mechanisms, etc. As we shall see, this variability
makes it virtually impossible to use individual preoperative variables to predict whether it is necessary to
shunt or to maintain BP above some predetermined
level. This explains the necessity to determine intraoperatively in each individual whether the current
hemodynamic situation is compatible with a sufficient
brain perfusion.
VASCULAR SURGERY
777
Table 1
Intra- and postoperative factors determining CEA morbidity and mortality, the ways to prevent those, and drawbacks
of preventative methods
Mechanism
Intraoperative
Embolism
Hemodynamic
Myocardial infarction
Postoperative
Carotid restenosis
Brain hemorrhage
Preventative methods
Drawbacks of preventative methods
Careful manipulation
Shunt avoidance
Shunt
"Blood pressure
#Blood pressure
"
"
"
"
"
Angioplasty
Avoid ischemia
Treat HTA
" Duration or surgery
Duration or surgery
Risk of brain hypoperfusion
Risk of embolism
Risk myocardial infarct
Risk of brain hypoperfusion
CEA, carotid endarterectomy.
Brain hemodynamic can be intraoperatively assessed

by transcranial Doppler, which only provides measurement of blood flow velocity in the sylvian artery. However, Doppler may not be applicable in 1015% of
cases and it just measures velocity, which is not always
correlated with CBF (Ackerstaff et al., 1998). Another
way to assess the integrity of the circle of Willis is to
measure stump pressure, that is, the arterial BP distal to
the more distal cross-clamp. According to the literature,
the acceptable limit for stump pressure varies from 25 to
50 mm Hg (Whitley and Cherry, 1996). However, both
transcranial Doppler and stump pressure provide a very
indirect view of the consequences of hemodynamic
variations, and none gives any direct idea of the brain
susceptibility to these variations. As opposed to these
methods, electroencephalography (EEG) and somatosensory evoked potentials (SEPs), directly assess brain
function. Moreover, they can detect the entry of the
brain in the ischemic penumbra zone (Astrup et al.,
1981) (see Chapter 61 on Cardiac Surgery) and thus
allow the detection of brain ischemia before it gives
rise to irreversible brain lesions.
57.2. EEG and SEP alterations that may
occur in CEA
If one just considers waveform morphology, the EEG
and SEP changes that may occur in CEA are largely
similar to those observed in cardiac surgery. These
will be considered in a first step. What makes the
specificity of IOM in CEA versus cardiac surgery is
their relationship with intraoperative events, which
will be considered in a second step.
57.2.1. EEG and SEP changes in CEA

Recent reviews can be found in Van Huffelen
(Chapter 6, this volume) and Florence et al. (2004).
EEG alterations first consist of an attenuation of fast
frequencies (alpha and beta) with a relative increase
of slow rhythms (theta and delta), followed by a
decrease of signal amplitude. These changes can be
either visually observed or quantified by frequency
analysis. SEP alterations first consist of a flattening
or a slowing of the frontal N30 and parietal P45,
initially not associated with any change in N20
latency, followed by the progressive disappearance
of earlier components (P27 and N20), together with
an N20 latency increase. Whatever the technique,
changes usually concern the ipsilateral hemisphere
(83% in our series) but can also be contralateral
(10%) or bilateral (7%). This implies that one should
never limit IOM to the ipsilateral hemisphere. This
means that bilateral EEG recordings should be performed and that both left and right median nerve
SEPs should be recorded (ideally, both recordings
should be simultaneously performed by alternating
pulses on the left and right median nerves).
There are still some controversies about what
changes should be considered as significant in terms
of potentially harmful brain ischemia. Regarding
EEG, a 95% spectral edge frequencies (SEF) decrease
of more than 50%, a 40%50% decrease of total power,
a shift of the relative power ratio toward slow frequencies, and a 6% change in the asymmetry index are
considered more reliable parameters. Regarding SEPs,
the most widely used quantitative criteria are an N20
J.-M. GUERIT
778
MILD
stim.
MODERATE
SEVERE
C-LE
compar.
Fpz-LE
C-LE
N20
N20
P27
Fpz-LE
C-LE
[P45]
P24
[FP]
follow-up
N20
N30
P14
N20
N20
N20
P24
1.25 v
P27
post-induct
P27
Fpz-LE
P14 N30
FP
P45
P14
N30
P14
100 ms
Fig. 1. Mild (left column), moderate (middle column), and severe (right column) somatosensory evoked potential (SEP)
changes suggestive of impaired perfusion and severe ischemia. Each set of two waveforms represents parietal (top curve)
and frontal (bottom curve) recordings, linked-ear (LE) reference. SEP baseline obtained immediately after induction
(bottom set of curves), altered SEPs (middle set of curves), and superimposition of the baseline and abnormal waveforms
(top set of curves). Mild abnormalities are characterized by a mere desynchronization of P45 and/or N30, moderate abnormalities by the disappearance of N30 and P27, and severe abnormalities by the disappearance of all activities following
N20, sometimes by the disappearance of N20 itself. Note a slight latency increase of N20 during severe alterations. The
lemniscal P14 remains unchanged and may be used as a quality control (adapted from Guerit et al., 1997).
amplitude decrease of more than 50%, and a more than

1 ms or 20% increase in the N13N20 (or P14N20)
interpeak latency (for a review, Florence et al., 2004).
In the absence of any gold standard, these criteria are
quite arbitrary. In addition, both the influence of anesthesia and of physiological factors like variations in
body temperature may complicate the use of such quantitative criteria. Therefore, we introduced qualitative
SEP criteria with a subdivision of SEP abnormalities
into mild, moderate, and severe alterations (Guerit
et al., 1997) (Fig. 1). A similar EEG classification can
be deduced from the proposal of Smith and Prior
(2003), based on International Federation of Clinical
Neurophysiology (IFCN) criteria of changes in EEG

suggestive of impaired brain perfusion and severe
ischemia (Burke et al., 1999). Both classifications are
summarized and compared in Table 2.
Rather than an absolute criteria, several authors stress
the dynamics of changes which follow CCC by 20 s
3 min for the EEG (McGrail, 1996; Ballotta et al.,
1997), 30 s10 min for SEPs (Guerit et al., 1997).
57.2.2. Relationships with intraoperative events
The aforementioned changes reflect a phenomenon,
that is, the entry of the brain in the ischemic
Table 2
Criteria of mild, moderate, and severe EEG and SEP changes suggestive of impaired brain perfusion (Guerit et al.,
1997; Smith and Prior, 2003)
D
EEG
SEPs
Mild
<50% decrease of fast activities (a b)

<50% increase of slow activities (y d)
>50% decrease of fast activities (a b)
>50% increase of slow activities (y d)
EEG loss over all frequency bands
Burst suppression
Desynchronization or disappearance of the frontal N30

and/or parietal P45
Desynchronization or disappearance of the parietal P27
(early warning) or P27 and P24 (urgent warning)
Disappearance of all activities following N20
Desynchronization or disappearance of N20
Moderate
Severe
EEG, electroencephalography; SEP, somatosensory evoked potential.
VASCULAR SURGERY
penumbra zone, but not the cause of this phenomenon. It is the cause and not the phenomenon itself,
which should determine the ad hoc strategy. This
reflects the general IOM rule that any EEG or SEP
change needs to be situated in the overall surgical
and anesthetic context.
As already mentioned in the introduction, the
main intraoperative determinants of brain ischemia
in CEA are hemodynamic disturbances (due to
CCC or a decrease in systemic BP) and embolism.
Based on the importance of SEP alterations and
their relationship with physiological parameters
and intraoperative events, we subdivided patients
undergoing CEA into six groups (Guerit et al.,
1997) (Table 3). Overall, IOM remained uneventful
in about 70% of the cases (N group) while 15% of
patients presented moderate-to-severe SEP alterations within 5 min after CCC, which justified shunt
installation (EC group). Among the remaining 15%
most patients presented changes that were consecutive to a drop in systemic BP (usually more than
30 mm Hg) outside (Po group) or inside (Pi group)
the CCC period. In only 2% of patients, moderateto-severe alterations occurred without identifiable
cause (changes not associated with any drop in
BP or changes occurring immediately after CCC
but not reversed by shunting). This latter group
(U group) was associated with the worse postoperative prognosis and is likely to include alterations of
embolic origin.
Note that in rare cases (in our series 3 out of about
1,800 cases, that is, <0.2% of cases) SEP alterations
appeared immediately after head positioning in
extension and external rotation, with immediate
recovery after repositioning the head in a more
neutral position. This implies that IOM should be
started as soon as possible, that is, immediately after
induction and before head positioning.
779
57.3. How IOM helps in and changes

the procedure
Based on the aforementioned distribution, the potential usefulness of IOM in CEA is threefold: (1) immediately after induction and before any surgical
maneuver to check whether head positioning is well
tolerated, (2) to decide whether a shunt should be
installed and, in the affirmative, whether the shunt is
well functioning, and (3) to decide whether BP is compatible with sufficient brain perfusion. In addition,
IOM might provide useful information in order to
decide for combined carotid and coronary bypass
grafting (CABG) surgery.
57.3.1. IOM as a guide to the decision to shunt
In principle, a shunt should be installed whenever
moderate-to-severe EEG and/or SEP changes occur
after CCC (Fig. 2). There is no consensus about
which delay abnormalities may be surely attributed
to CCC. According to the literature, EEG or SEP
changes are likely to occur within 20 s3 min after
CCC, but we have seen situations in which changes
were observed later on. Practically, the decision to
shunt may also depend on the expected crossclamping time, which, in turn, depends on surgical
strategy. If the surgeon limits himself to the endarterectomy in which case the cross-clamping time may
be as short as 10 min, the actual opportunity of a
shunt may be questioned. If he performs angioplasty,
the mean cross-clamping time is likely to be about
2030 min, in which case even later abnormalities
(within the 5 min after CCC) may require a shunt.
Whatever the preferred strategy, one should always
consider possible unexpected events. We consider it
wise to have a shunt ready on the operating table and
to keep the surgeon informed of the current brain status.
Table 3
Relationship between SEP changes and intraoperative events in CEA (adapted from Guerit et al., 1997)
Group
Features
N
EC
LC
Po
Pi
U
No changes
Moderate to severe changes early (<5 min) after CCC
Mild changes at the end of the CCC period
Mild-to-moderate changes consecutive to a drop in BP outside the CCC period
Mild-to-moderate changes consecutive to a drop in BP inside the CCC period
Changes of intraoperatively undetected etiology (most likely embolic)
69
15
1
9
4
2
SEP, somatosensory evoked potential; CEA, carotid endarterectomy; CCC, carotid cross-clamping; BP, blood pressure.
J.-M. GUERIT
780
09:26
N20
09:29
P27
P45
09:34
CCC : 09:35
09:38
09:39
09:41
Shunt open : 09:40
09:48
09:55
Right median nerve
Left median nerve
Fig. 2. Left CEA, severe SEP alterations occurring in the left hemisphere (right median nerve stimulation) 3 min after
CCC. Left median nerve SEPs remained unchanged at 09:39. Shunt installation (09:40) rapidly gave rise to SEP recovery.
At 09:48, right median nerve SEPs had fully recovered (the current SEP is superimposed to the baseline obtained at 09:29).
CEA, carotid endarterectomy; SEP, somatosensory evoked potential; CCC, carotid cross-clamping.
It is worth remembering that the necessity to shunt is

the exception rather than the rule. Several series
report that more than 85% of patients did not need
to be shunted. Only one out of our more than 1,500
patients who were not shunted presented a transient
aphasia on awakening, but was normal when released
from the hospital (Guerit et al., 1997).
Shunt installation must be associated with EEG or
SEP recovery. If this is not the case, either the EEG
or SEP changes do not actually reflect hemodynamic
changes but, for instance, embolism, either the shunt
is defective or wrongly placed. We had one patient in
whom moderate SEP changes did not recover after
shunt opening. After an unsuccessful attempt to
increase BP the surgeon removed the shunt and
noticed its defectiveness as an eccentric balloon
was actually pinning the duct against the artery wall.
Shunt replacement gave rise to almost immediate
recovery, which, however, intervened as long as 28
min after the initial CCC. The patient awoke with
transient (<24 h) and fully reversible hemiparesis.
This case is interesting for two reasons: (1) the
patient presented transient ischemic accident, which
confirms the specificity of found abnormalities, and
(2) even a 28 min period of altered SEPs remained
compatible with only transient deficits confirms that

the 23 min delay which is required to decide
whether to shunt on the basis of EEG or SEPs provides a sufficient safety margin.
57.3.2. IOM as a guide to blood-pressure
adjustment
Obviously, the absence of EEG or EP alterations is
likely to constitute a valuable criterion of a sufficient
brain perfusion, allowing maintenance of BP at its
current level. By contrast, it is more difficult to provide guidelines about what should be done in the case
of mild-to-moderate EEG or SEP changes occurring
after a drop in BP. Indeed, too high a BP may be
harmful for the myocardium and we know that the
brain may come through some minutes in ischemic
penumbra without irreversible lesions.
An example of a patient from the PI group is
shown in Fig. 3. This patient underwent a left transient ischemic accident after successful treatment
of hypertension. BP was higher than 200 mm Hg
at induction. A first episode of relative hypotension
(BP: 160 mm Hg) occurring 4 min after CCC was
followed by a small N20 latency increase and an
VASCULAR SURGERY
781
250
200
150
BP
Amplitude
100
50
0
40
ccc
50
60
70
Time (min)
80
90
100
Stop ccc
Fig. 3. BP-related changes in N20 amplitude during CCC in a hypertensive patient with a story of TIA. BP at induction was
220 mm Hg. During the CCC period, there was a systematic collapse of SEP amplitude whenever BP decreased under
180 mm Hg. After cross-clamp removal, BP could be decreased down to 140 mm Hg without any SEP change. No shunt
was used and the patient awoke without neurological deficit. BP, blood pressure; CCC, carotid cross-clamping; SEP,
somatosensory evoked potential.
amplitude decrease of the frontal N30. Since they were

not severe, and given the drop in BP, we first decided
to make a control recording before warning the surgeon and asking him to shunt. Meanwhile, the anesthesiologist decided to restore BP. This gave rise to full
SEP recovery. The same phenomenon occurred
10 min later, after the BP dropped to 180 mm Hg. This
patient was not shunted. A careful attempt to decrease
BP was performed under SEP monitoring at the end
of the operation. The removal of the carotid stenosis
allowed BP to be decreased down to 140 mm Hg
without any SEP alteration. This patient awoke
without neurological sequels.
We currently base our decisions to increase BP on
four criteria: (1) the fact that a drop in BP must
have occurred within the 3 min preceding the IOM
changes, (2) the severity of EEG or SEP changes,
(3) the previous patients cardiac (myocardial ischemia) and neurological (transient ischemic accident,
stroke) story, and (4) the EKG. Mild changes usually
do not justify increasing BP, except if there is a history of transient ischemic accident. If there is no such
history, mild changes only justify a control recording.
We suggest that moderate changes occurring after a
drop in BP justify restoring BP except if the patient

has a history of coronary disease or if abnormalities
suggestive of myocardial ischemia appear on the
EKG. If this is the case, BP should only be restored
after worsening has been demonstrated in a control
recording.
57.3.3. IOM changes as a contraindication of
combined CEA and CABG procedures
There is still debate on the issue of whether CEA and
CABG procedures should be combined. On the one
hand, significant carotid stenosis is one of the possible
causes of neurological complications in patients undergoing cardiac surgery. Indeed, the incidence of carotid
stenosis of 70% or greater ranges from 1.3% to 8.5%
in candidates for cardiac surgery (Ricotta et al., 1995).
On the other hand, myocardial infarction has been
reported to be the most frequent cause of perioperative
and late death after CEA (Moore et al., 1995). This
encourages combined procedures (Nakamura et al.,
2003). We studied the neurological outcomes of
33 patients in whom CEA was combined with CABG
or valvular replacement (Guerit et al., 1997). Among
J.-M. GUERIT
782
these 33 patients, 24 did not present any IOM changes

and 6 were shunted. Of the 24 non-shunted patients,
22 did not present any neurological sequels, one presented delayed awakening, and one presented major
neurological deficits with computed tomography (CT)
scan evidence of multiple lesions of embolic origin. Of
the six shunted patients of the EC group, three did not
present any sequels, one presented a postoperative confusion but was normal when released from the hospital,
and two presented a long-lasting hemiplegia or hemiparesis. Thus, neurological sequels were observed in half of
the six patients of the EC group who underwent combined procedures. This should be contrasted with
patients undergoing isolated CEA in whom only 2 of
24 shunted patients underwent sequels. Therefore, when
a combined procedure is envisaged, we suggest
performing CEA before thoracotomy and to postpone
the second step in all patients in whom a shunt was
required.
57.4. What to think clinically when the potentials
do change?
If one excludes technical failure, which can usually be
ruled out on the basis of preserved N13 and/or P14,
three factors can a priori explain IOM changes during
carotid surgery: (1) an increase or a decrease in the
level of anesthesia, (2) a decrease (or an increase) in
cerebral perfusion, which can be consecutive to a drop
(or the restoration) of systemic BP, CCC, or embolism,
and (3) normal individual variability.
Whatever technique is used (EEG or SEPs), it is
actually difficult to predict the nature of changes just
Table 4
Pathophysiological correlates and laterality of IOM
changes in carotid endarterectomy
D
Unilateral
Bilateral
Mild
# Systemic BP
CCC
Embolism
Moderate
## Systemic BP
CCC
Embolism
### Systemic BP
CCC
Embolism
" Level of anesthesia

# Systemic BP
CCC
Normal variability
## Systemic BP
CCC
Severe
### Systemic BP
CCC
IOM, intraoperative monitoring; BP, blood pressure; CCC, carotid

cross-clamping.
on the basis of the severity and laterality of changes.

Indeed, as is shown in Table 4, both a drop in
systemic BP and CCC can give rise to unilateral or
bilateral, mild, moderate, or severe changes. Nevertheless, bilateral changes are very unlikely to be
due to embolism (in contrast to what happens in
cardiac surgery) and unilateral changes are quite
unlikely to be due to an increase in the level of
anesthesia or normal individual variability. Based
on these principles, we propose two decision algorithms to be applied outside (Fig. 4A) and during
(Fig. 4B) the CCC period. Note, whether brain dysfunction revealed by IOM changes is reversible or
not depends less on the severity of the changes than
on their pathophysiology. Thus, moderate changes
due to embolism can be less reversible than severe
changes occurring after CCC. This means that the
decision to shunt should never been taken lightly.
In particular, mild abnormalities almost never
justify shunting and the possibility to raise BP should
always be considered when moderate-to-severe
abnormalities appear more than 3 min after CCC.
57.5. Relationships between IOM and pre- or
intraoperative parameters
Table 5 (adapted from Guerit et al., 2002) summarizes
the mean values of some pre- and intraoperative
parameters in the N, EC, PO, and PI groups. When
compared with the N group, the EC group was characterized by a lower degree of ipsilateral carotid stenosis
(78.7% vs. 86.6% in the N group), a higher degree
of contralateral stenosis (60.5% vs. 34.8%), a lower
stump pressure (34.9 vs. 53.7 mm Hg), and a higher
gradient between mean arterial BP and stump pressure
(60.0 vs. 41.7 mm Hg). However, the BP at induction
was not significantly different. The Pi group was also
characterized by a lower degree of ipsilateral stenosis
(71.3% vs. 86.6% in the N group), but did not significantly differ for the other parameters, including the
degree of contralateral stenosis. Finally, the mean BP
at induction was higher in the Po group (107 mm Hg)
than in the N group (91.7 mm Hg) and the EC group
(93.0 mm Hg) but did not differ from the mean BP
in the Pi group (97 mm Hg).
These observations could be expected. Indeed,
a lower degree of ipsilateral carotid stenosis implies a
higher deprivation of blood during CCC as there is
a higher degree of contralateral stenosis. A higher BP
at induction may reflect impairment of self-regulatory
mechanisms, in which case brain perfusion directly
VASCULAR SURGERY
783
Start
Control
no
Changes?
no
Moderate?
yes
yes
Mild?
Probable embolism
no
BP?
no
severe changes
no
Recovery?
yes
yes
Restore BP
yes
Unilateral?
yes
no
yes
anesthesia?
no
yes
BP?
yes
OK to BP?
no
no
New baseline
Warn surgeon
CCC
Probable embolism
no
no
Control
Changes?
yes
recovery?
yes
yes
Unilateral?
yes
no
no
no
Restore BP
Mild?
moderate to severe
<7 min?
anesthesia?
no
yes
BP?
yes
Shunt OK?
no
no
yes
yes
New baseline
Shunt
no
yes
recovery?
no
Shunt OK?
yes
Fig. 4. Proposed decision algorithms before (Fig. 4A) and during (Fig. 4B) the cross-clamping period. These algorithms are
based both on the severity of changes and on their relationship with intraoperative events.
J.-M. GUERIT
784
Table 5
Relationships between IOM events and the degrees of ipsi- and contralateral carotid stenosis, BP at induction, and
stump pressure (adapted from Guerit et al., 2002)
Group
Ipsilateral
stenosis (%)
N
EC
Po
Pi
141
30
18
8
86.6
78.7
89.2
71.3

11.4
17.8
7.2
20.7
Contralateral
stenosis (%)
34.8
60.5
37.1
31.9

Mean BP at
induction
36.7
44.1
36.1
39.6
91.7
93.0
107
97.3

18.1
15.5
11.2
23.6
Stump pressure
53.7
34.9
49.1
48.6

19.7
16.5
21.2
27.1
Gradient
41.7
60.0
44.9
57.7

21.2
24.1
23.5
15.3
IOM, intraoperative monitoring; BP, blood pressure.
patients with a stump pressure higher than 50 mm Hg.

Conversely, no IOM abnormalities were observed in
42% of patients with an ipsilateral stenosis lower
than 70%, in 49% of patients with a 100% occlusion
of the contralateral carotid artery, and in 44% of
patients with a stump pressure lower than 30 mm Hg.
Note, a shunt had to be installed in only 40% of
patients with a total occlusion of the contralateral
carotid (10% of these patients merely presented
BP-related changes recovering after BP restoration).
Whether the combined use of some of these parameters could allow us to safely predict the necessity
to shunt in individuals has not been demonstrated
and would require larger group studies.
100
80
80
60
40
20
0
% of patients
% of patients
100
40
20
100
100
80
80
60
40
20
0
<30
3050
>50
Stump pressure (mm Hg)
N-Group
EC-Group
60
<70 %
7090%
>90%
Ipsilateral stenosis (%)
% of patients
% of patients
depends on systemic BP, and it is thus more likely that

a drop in BP will give rise to hypoperfusion. A lower
stump pressure of a higher gradient is indicative of a
less efficient circle of Willis.
Does preoperative (the degree of ipsi- or contralateral stenosis) or intraoperative (the stump pressure or
the gradient between the mean stump pressure
and the mean systemic arterial BP) predict the necessity to shunt an individual in a simpler way to IOM?
The answer is no, at least when each parameters are
taken in isolation. Thus, in our series (Fig. 5), a shunt
was required in 5% of patients with an ipsilateral
stenosis higher than 90%, in 9% of patients with a
contralateral stenosis lower than 70%, and in 2% of
<70 % 7090% >90%

100%
Contralateral stenosis (%)
60
40
20
0
020
2040
4060
Gradient (mm Hg)
>60
Fig. 5. Distribution of degrees of ipsilateral and contralateral carotid stenosis, stump pressures, and gradients in patients
who did not present any SEP alteration (N group) and patients who had to be shunted, based to moderate-to-severe
SEP alterations appearing shortly after CCC (EC group). When taken in isolation, none of these parameters may predict
whether a shunt is required in individual patients. SEP, somatosensory evoked potential; CCC, carotid cross-clamping.
VASCULAR SURGERY
785
57.6. Respective advantages and drawbacks

of EEG and SEPs
Both EEG and SEPs correctly target these structures
that are at risk during CEA. Which of the two techniques should be preferred is still a matter of debate.
There are three ways to approach this problem: (1)
to consider the relationship between EEG/SEP
changes and CBF, (2) to compare the sensitivities
and specificities of EEG versus SEPs in carotid endarterectomy, and (3) to consider simultaneous EEG
and SEP recordings.
57.6.1. Relationship between EEG/SEP changes
and CBF
Fig. 6 summarizes the main EEG and SEP alterations
and their relationship with CBF. Theoretically, a
CBF decrease to 2022 ml/100 g/min is not associated with any change in brain function. In the
EEG, the relative decrease of fast versus slow frequencies is associated with a change in brain
function, and a global power decrease could be
observed for CBF values of 1022 ml/100 g/min.
The EEG would become isoelectric for CBF values
of 715 ml/100 g/min. Regarding SEPs, studies conducted in anesthetized baboons demonstrated an
amplitude decrease of cortical SEPs for CBF values
of 1620 ml/100 g/min, an increase in the central
conduction time for values of <15 ml/100 g/min,
and a disappearance of cortical SEPs for CBF values
of 1215 ml/100 g/min (reviewed by Florence et al.,
2004). Some authors consider that cortical SEPs
EEG : amplitude, slowing
SEPs: amplitude >50%
SEPs: CCT
SEPs
<15
715
Cortical SEP 0
1215
10
EEG
57.6.2. Sensitivity and specificity of EEG versus

SEPs in carotid endarterectomy
Florence et al. (2004) recently reviewed or attempted
to calculate the EEG and SEP sensitivities (defined
as the probability of a positive test result in patients
with cerebral ischemia) and specificities (defined as
the probability of a negative test result in patients
without ischemia) in all papers dealing with CEA
in IOM in which sufficient information could be
gained. The median values [ranges] of sensitivity
and specificity were, respectively, 44% [13100%]
and 100% [12100%] for visual analysis of the
EEG, 50% [5089%] and 100% [15100%] for digital EEG, and 54 [20100%] and 100% [43100%]
for SEPs. This analysis suggests that there is no clear
superiority of one technique over the other. It also
demonstrates a wide range of results, precluding
any definite conclusion. This variability is likely
due to methodological reasons, resulting from the
absence of a gold standard and the frequent use of
interactive strategies.
For the same reason, methodological differences
(criteria for patient selection, of EEG and/or SEP
abnormalities, duration of the CCC period, etc.)
make it almost impossible to make a valid comparison of the percentages of patients who are shunted
on the basis of one or the other technique. It actually
varies between 10% and 30%, irrespective of what
technique is used (personal communications).
1416
Isoelectric EEG
1022
disappear for CBF values 20% lower than those causing an isoelectric EEG (Prior, 1973; Nuwer, 1988).
Overall, even if the EEG could be a bit more
sensitive than SEPs, CBF relationships do not reveal
any clear superiority of one technique over the other.
15
20
25
CBF (ml/100 g/min)
Fig. 6. Relationship between CBF and EEG or SEP alterations (adapted from Florence et al., 2004). Although the
EEG could be a bit more sensitive, there is no clear superiority of one technique on the other. CBF, cerebral blood
flow; EEG, electroencephalography; SEP, somatosensory
evoked potential.
57.6.3. Simultaneous recording of EEG and SEPs

in carotid endarterectomies
The third and most straightforward way to compare
both techniques is to use these simultaneously in
the same patient. Only a few studies dealing with
limited series of patients directly compared EEG
and median nerve SEPs in CEA (Florence et al.,
2004; Table 6). Again, it did not demonstrate any
clear superiority of one technique over the other,
even if the SEPs appeared a bit more sensitive and
specific. More recently, Ragazzoni et al. (2000)
presented their data on 568 patients operated on
under simultaneous EEG and median nerve SEP
J.-M. GUERIT
786
Table 6
Sensitivity and specificity of SEPs and EEG when simultaneously recorded during CEA (reprinted from Florence et al.,
2004, with permission from Elsevier)
References
Methods
Sensitivity
Specificity
Comments
Pozzessere et al. (1987)
Visual
CCT/N20
Visual
CCT/N20
Visual
N20
Digital
CCT/N20
0%
100%
50%
100%
100%
100%
50%
100%
100%
100%
17%
18%
0%
100%
15%
100%
N 25 Selective shunting
Lam et al. (1991)

Kearse et al. (1992)
Fiori and Parenti (1995)
N 64 No shunt
N 53
N 240 Selective shunting
SEPs, somatosensory evoked potentials; EEG, electroencephalography; CEA, carotid endarterectomy
monitoring. They showed that the most reliable

component was the loss of cortical SEPs, which
was associated with and often preceded by a unilateral or bilateral complete loss of the EEG. For these
authors, SEPs appeared to be less sensitive but more
specific than the EEG for the detection of cerebral
ischemia.
Since 2000, we performed simultaneous EEG and
EP recording in about 200 procedures. Owing to our
previous experience, SEP alterations were considered
the gold standard against which EEG alterations were
compared. Two results should be underlined. First,
whenever a well-structured and continuous EEG with
clear fast (alpha and/or beta) activities was obtained
after induction, there was always a close correlation
between SEP and EEG changes. Although offline
analysis sometimes demonstrated earlier flattening
of the EEG, there was no single case in which online
examination of these changes would have led to
earlier shunt installation, or to shunt installation in a
patient who was not shunted, based on SEP criteria.
This is largely a result of the much greater reliability
of SEPs and their relative independence of the environmental noise. Note, none of these patients developed neurological sequels. Second, there was a
marked variability of the post-induction EEGs such
that fast activities were lost before CCC in about
50% of the cases in whom it was definitely impossible to identify any EEG flattening or slowing after
CCC (Figs. 7 and 8). These EEG alterations were
usually associated with the disappearance of the frontal N30/parietal P45, which precluded from identifying any mild SEP changes. This did not change the
procedure as mild SEP changes never constitute, in
our experience, a sufficient criterion to shunt.
Overall, both EEG and SEPs are valuable techniques and currently neither technique can be seen as
superior to the other. More comparative studies are
needed in larger groups of patients. Currently, we suggest each laboratory should choose the technique with
which it has the best experience.
57.7. Which surgeons expectancies can or cannot
be met?
Whatever technique is used (EEG or median-nerve
SEPs), the absence of any IOM change indicates to
the surgeon that a sufficient brain perfusion is
achieved in the sylvian artery territory. Provided that
a sufficient number of channels should be used, the
EEG may also provide some information on the territories of the anterior and posterior cerebral arteries
(even in the latter, it is quite unlikely to be at risk
in CEA, though it could be theoretically the case with
some atypical configurations of the circle of Willis).
By contrast, the EEG does not provide any straightforward information on the brainstem. Median nerve
SEPs assess the brainstem lemniscal pathways, but
do not provide assessment on the territories of the
anterior cerebral arteries whose assessment requires
stimulation of the posterior tibial nerves (PTN).
Moreover, SEPs do not directly assess the motor
pathways. That is, none of the classically used techniques provide whole-brain assessment, so that it
remains theoretically possible that some neurological
sequels corresponding to intraoperative brain damage
may remain undetected.
Fortunately, both EEG and SEPs correctly assess
the sylvian artery territory, which is the neural
structure at risk during CCC. It is also very likely that
VASCULAR SURGERY
787
Post-induction
2 after CCC
F3-A1
F3-A1
C3-A1
C3-A1
F4-A2
F4-A2
C4-A2
C4-A2
Fpz-A2
Fpz-A2
C6sp -
C6sp Signal hidden
Cerv.g.
100 ms
5 v
Signal hidden
Cerv.g.
100 ms
5 v
Left median nerve

N20
18.6
104/17+
C4
100 ms
5 v
P14
14.0 P24 xxx
22.3 xxx
xxx
Fpz
100 ms
5 v
C4
100 ms
5 v
N20
18.8
P14
14.0
112/10+
Fpz
100 ms
5 v
148/63+
P24 N30
22.026.1
P22
26.4
164/48+
Signal hidden
Cerv.dr.
100 ms
20 v
Cerv.dr.
100 ms
5 v
Right median nerve

N20
20.4
180/33+
120/0+
C3
100 ms
5 v
P14
14.9
P24
23.6
P27
28.5
N30
30.8
Fpz
100 ms
5 v
C3
100 ms
5 v
118/2+
Fpz
100 ms
5 v
P14
14.6
P24 P27
23.3 26.9
N30
29.4
190/23+
Fig. 7. Simultaneous recording of EEG (top) and SEPs (bottom) in a patient undergoing left CEA. Severe abnormalities of
the right median-nerve SEPs were observed 2 min after CCC. Left median nerve SEPs remained intact. As no fast activities
were present in the EEG immediately after induction, ipsilateral EEG changes were hardly identified 2 min after CCC.
EEG, electroencephalography; SEPs, somatosensory evoked potentials; CEA, carotid endarterectomy; CCC, carotid
cross-clamping.
Left median
C6sp
Right median
F3-A1
C3-A1
F4-A2
C4-A2
FpzA2
C6sp
C6sp
C4
Fpz
C3
Fpz
Fig. 8. Three examples of simultaneous EEG and SEP recordings obtained immediately after induction in CEA. All
patients were anesthetized with Propofol. None of these patients presented ischemic changes after CCC. EEG, electroencephalography; SEPs, somatosensory evoked potentials; CEA, carotid endarterectomy; CCC, carotid cross-clamping.
J.-M. GUERIT
788
median nerve SEPs also assess the boundary zones

between the anterior and sylvian artery territories
through the frontal N30 (Desmedt et al., 1987)
and that between the sylvian and posterior cerebral
artery territories through the parietal P45 (Fava
et al., 1992), which correspond to the last meadows
at risk during systemic hypoperfusion. In that respect,
IOM decreases the risks due to hypoperfusion. Indirectly, it also decreases the risk of embolism, for
two reasons. First, it allows avoiding shunting in
more than 85% of cases, thereby decreasing the risk
for embolism that shunting entails. Second, it allows
the surgeon to proceed without undue haste, which
implies that he has time to carefully clean up the
carotid artery after endarterectomy, which is another
way to prevent embolism. This can also give him time
to perform angioplasty in order to prevent restenosis.
The increase in safety may lead one to reconsider
the surgical indication in these patients with lower
degrees of stenosis and could be a reason to prefer
CEA to carotid stenting.
Of course, IOM may be unable to predict postoperative events such as delayed thrombosis of embolism, especially when an eventful procedure is
followed by CABG or other cardiac surgery. Moreover, IOM may be difficult, if not impossible, in
the presence of a preexisting stroke with EEG flattening or slowing over the sylvian area and an absence
of major deterioration of SEP cortical components.
This underlines the necessity to perform a preoperative recording, whenever possible, to evaluate IOM
feasibility. This is usually not a sufficient reason to
give up monitoring but the surgeon should then be
clearly informed of its limitations.
57.8. What new types of monitoring would be
desirable in the future?
From the neurophysiological standpoint, the abovementioned limitations of IOM can be easily solved
by combining techniques: EEG and SEPs to both
median and PTN stimulation. Motor evoked potentials (MEPs) obtained by magnetic or electrical
transcranial stimulation are another alternative, even
if their use may be limited as a result of the use of
muscle blockers. IOM should not be limited to
clinical neurophysiology but should also include
Doppler, near infra-red spectroscopy, and metabolic
measurement, in keeping with the concept of multimodality monitoring (Edmonds, 2002; Murkin,
2004) (see Chapter 61 on Cardiac Surgery). Finally,
postoperative complications may also be prevented

by pursuing monitoring during the ICU stay within
one or a few days after the operation. Note, such
IOM extension should aim not only at providing a
more exhaustive brain assessment but also at increasing redundancy and, thereby, specificity, and safety.
However, such an increase in IOM complexity
would be meaningless if not associated with the parallel developments of tools to facilitate its interpretation. Thus, it would be particularly demanding,
if not impossible, for a single neurophysiologist to
perform, simultaneously, visual analysis of EEG
and median and PTN SEPs, visual and auditory analysis of the transcranial Doppler, and to integrate all
this information with metabolic, anesthetic, and surgical events. The other problem is that of continuous
ICU monitoring, which faces the problem of the lack
of neurophysiological expertise of the intensivists.
Several systems are currently under development to
bypass these difficulties and provide systems (quantitative indices, automatic SEP analysis, artificial
intelligence, and expert systems) for automatic (but
human supervised) warning.
These requirements are not specific to CEA but
actually concern the whole domain of intraoperative
and ICU monitoring. To solve this paradoxical evolution toward more comprehensive and, at the same
time, simpler systems constitutes one of the major
challenges to be taken up by clinical neurophysiology in the near future.
References
Ackerstaff, RGA, Jansen, C, Moll, FL, Vermeulen, FE,
Hamerlijnck, RP and Mauser, HW (1995) The significance of microemboli detection by means of transcranial Doppler ultrasonographic monitoring in carotid
endarterectomy. J. Vasc. Surg., 21: 963969.
Ackerstaff, RGA, Chuck, JW and Van De Vlasakker, CJ
(1998) Monitoring of brain function during carotid endarterectomy: an analysis of contemporary methods.
J. Cardiothorac. Vasc. Anesth., 12: 341347.
Astrup, J, Siesjo, BK and Symon, L (1981) Thresholds in
cerebral ischemia the ischemic penumbra. Stroke,
12: 723725.
Ballotta, E, Dagiau, G, Saladini, M, Bottio, T, Abbruzzese,
E, Meneghetti, G and Guerra, M (1997) Results of
electroencephalographic monitoring during 369 consecutive carotid artery revascularizations. Eur. Neurol., 37:
4347.
Yingling, CD and Jones, SJ (1999) Intraoperative
VASCULAR SURGERY
monitoring. In: G Deuschl and A Eisen (Eds.), Recommendations for the Practice of Clinical Neurophysiology. Guidelines of the International Federation of
Clinical Neurophysiology. Elsevier, Amsterdam, (EEG
Suppl. 52), pp. 133148.
Chaturvedi, S, Bruno, A, Feasby, T, Holloway, R, Benavente, O, Cohen, SN, Cote, R, Hess, D, Saver, J,
Spence, JD, Stern, B and Wilterdink, J (2005) Carotid
endarterectomy an evidence-based review: report
of the Therapeutics and Technology Assessment Subcommittee of the American Academy of Neurology.
Neurology, 65: 794801.
Coward, LJ, Featherstone, RL and Brown, MM (2005)
Safety and efficacy of endovascular treatment of carotid
artery stenosis compared with carotid endarterectomy.
A Cochrane systematic review of the randomized evidence. Stroke, 36: 905911.
De Borst, GJ, Moll, FL, Van De Pavoordt, HD, Mauser,
HW, Kelder, JC and Ackerstaf, RG (2001) Stroke from
varotid endarterectomy: when and how to reduce perioperative stroke rate. Eur. J. Vasc. Endovasc. Surg.,
21: 484489.
Desmedt, JE, Nguyen, TH and Bourguet, M (1987) Bitmapped color imaging of human evoked potentials with
reference to the N20, P22, P27 and N30 somatosensory
response. Electroencephalogr. Clin. Neurophysiol., 68:
119.
Edmonds, HL, Jr. (2002) Multi-modality neurophysiologic
monitoring for cardiac surgery. Heart Surg. Forum, 5(3):
225228.
Fava, E, Bortolani, E, Ducati, A and Schieppati, M (1992)
Role of SEP in identifying patients requiring temporary
shunt during carotid endarterectomy. Electroencephalogr. Clin. Neurophysiol., 84: 426432.
Fiori, L and Parenti, G (1995) Electrophysiological monitoring for selective shunting during carotid endarterectomy.
J. Neurosurg. Anesth., 7: 168173.
Florence, G, Guerit, JM and Gueguen, B (2004) Electroencephalography (EEG) and somatosensory evoked potentials (SEP) to prevent cerebral ischaemia in the
operating room. Neurophysiol. Clin., 34: 1732.
Guerit, JM, Witdoeckt, C, De Tourtchaninoff, M, Ghariani,
S, Matta, A, Dion, R and Verhelst, R (1997) Somatosensory evoked potential monitoring in carotid surgery. I.
Relationships between quantitative SEP alterations and
intraoperative events. Electroencephalogr. Clin. Neurophysiol., 104: 459469.
Guerit, JM, Witdoeckt, C, Verhelst, R and Ghariani, S (2002)
Carotid endarterectomy monitoring. In: RC Reisin, MR
Nuwer, M Hallett and C Medina (Eds.), Advances in
Clinical Neurophysiology (Supplements to Clinical
Neurophysiology). Elsevier, Amsterdam, Vol. 54,
pp. 319324.
Kearse, LA, Brown, EN and McPeck, K (1992) Somatosensory evoked potentials sensitivity relative to
789
electroencephalography for cerebral ischemia during
carotid endarterectomy. Stroke, 23: 498505.
Lam, AM, Manninen, PH, Ferguson, GG and Nantau, W
(1991) Monitoring electrophysiologic function during
carotid endarterectomy: a comparison of somatosensory
evoked potentials and conventional electroencephalogram. Anesthesiology, 75: 1521.
McGrail, KM (1996) Intraoperative use of electroencephalography as an assessment of cerebral blood flow. Neurosurg. Clin. N. Am., 7: 685692.
Moore, WS, Barnett, HJM, Beebe, HG, Berstein, E, Brener,
BJ, Brott, T, Caplan, LR, Day, A, Goldstone, J, Hobson,
R, Kempczinski, RF, Matchar, DB, Mayberg, MR,
Nicolaides, AN, Norris, JW, Ricotta, JJ, Robertson,
JT, Rutherford, RB, Thomas, D, Toole, JF, Trout, HH
and Wiebers, DO (1995) Guidelines for carotid endarterectomy. A Multidisciplinary Statement from the Ad
Hoc Committee, American Heart Association. Circulation, 91: 566579.
Murkin, JM (2004) Perioperative multimodality monitoring: an overview. Semin. Cardiothorac. Vasc. Anesth.,
8: 167171.
Nakamura, Y, Kawachi, K, Imagawa, H, Hamada, Y,
Takano, S, Tsunooka, N, Sako, H and Kumon, Y
(2003) Combined carotid endarterectomy and cardiac
surgery for concomitant carotid and cardiac disease.
Ann. Thorac. Cardiovasc. Surg., 9: 180183.
Nuwer, MR (1988) Use of somatosensory evoked potentials for intraoperative monitoring of cerebral and spinal
cord function. Neurol. Clin. N. Am., 6: 881897.
Pozzessere, G, Valle, E, Santoro, A, Delfini, R, Rizzo, PA,
Cantore, GP and Morocutti, C (1987) Prognostic value
of early somatosensory evoked potentials during carotid
surgery: relationship with electroencephalogram, stump
pressure and clinical outcome. Acta Neurochir., 89:
2833.
Prior, PF (1973) The EEG in Acute Cerebral Anoxia.
Excerpta Medica, Amsterdam.
Ragazzoni, A, Chiaramonti, R, Zaccara, G, Ercolini, L and
Cecchi, M (2000) Simultaneous monitoring of multichannel somatosensory evoked potentials and electroencephalogram during carotid endarterectomy: a
comparison of the two methods to detect cerebral ischemia. Clin. Neurophysiol., 111: S138.
Ricotta, JJ, Faggioli, GL, Castilone, A and Hassett, JM
(1995) Risk factors for stroke after cardiac surgery. Buffalo Cardiac-Cerebral Study Group. J. Vasc. Surg., 2:
359364.
Sila, CA (1998) Neurologic complications of vascular surgery. Neurol. Clin. N. Am., 16: 920.
Smith, NJ and Prior, PF (2003) Neurophysiological monitoring during surgical operations. In: CD Binnie, R Cooper,
F Mauguie`re, JW Osselton, PF Prior and BM Tedman
(Eds.), Clinical Neurophysiology. Elsevier, Amsterdam.
Vol. 2, pp. 791903.
790
Whitley, D and Cherry, K (1996) Predictive value of
carotid artery stump pressure during carotid endarterectomy. Neurosurg. Clin. N. Am., 7: 723732.
Witdoeckt, C, Ghariani, S and Guerit, JM (1997) Somatosensory evoked potential monitoring in carotid surgery.
II. Comparison between qualitative and quantitative
J.-M. GUERIT
scoring systems. Electroencephalogr. Clin. Neurophysiol., 104: 328332.
Yeager, RA, Moneta, GL, McConnell, DB, Neuwelt, EA,
Taylor, LM, Jr. and Porter, JM (1989) Analysis of risk
factors for myocardial infarction following carotid endarterectomy. Arch. Surg., 128: 11421145.

M.R. Nuwer (Ed.)
791
CHAPTER 58
Monitoring during carotid balloon test occlusion

Dawn Eliashiv*, Vladimir Royter and Dolores Quinonez
Department of Clinical Neurophysiology, Cedars Sinai Medical Center,
David Geffen School of Medicine at UCLA, Los Angeles, CA 90048, USA
58.1. Introduction
Diagnostic and treatment approaches to a variety of
head and neck diseases often require the assessment
of functional capability of cerebral blood flow
(CBF). The reason is that not all patients can tolerate
abrupt occlusion of the carotid artery and in those
patients with no or low tolerance, acute or delayed
ischemic events may occur. Controversy exists on
how to evaluate the risk of procedures that might sacrifice or cause prolonged temporary occlusion of the
internal carotid artery (ICA). These procedures
include surgery for head and neck tumors, interventions for cerebral arteriovenous malformations
(AVM), carotid cavernous fistulas, arterial aneurysms, carotid endarterectomy (CEA), and a myriad
of other conditions.
Carotid artery balloon occlusion test (BOT) in conjunction with CBF analysis is believed to help identify
those patients who will not tolerate carotid occlusion.
Before the implementation of temporary test occlusion, the interventions involving ICA sacrifice were
associated with high morbidity and mortality rates.
Linskey et al. (1994) in a review of the literature noted
a cumulative stroke risk of 26%. Strokes tended to be
large and were associated with a high-mortality rate
of 12%.
First attempts to perform manual test compression
of the carotid artery to evaluate ischemic tolerance
were described by Matas (1911). The technique based
on endovascular balloon intervention initially introduced in 1970 by Serbinenko (1974) became by
now the most commonly used means to perform
BOT on ICA.
*
Correspondence to: Dawn Eliashiv, M.D., Department of

Clinical Neurophysiology, Cedars Sinai Medical Center,
David Geffen School of Medicine at UCLA, 8700 Beverly
Blvd., South Tower, Los Angeles, CA 90048, USA.
E-mail: EliashivD@cshs.org (D. Eliashiv).
BOT protocol with low-complications rate was

described by Mathis et al. (1995). Authors introduced
the balloon system utilizing nondetachable balloon
catheters. Once the balloon system is in place and a
digital road map is visualized the anticoagulation is
achieved with IV bolus of heparin. Proper balloon
inflation is crucial to create full damping of the arterial
pressure without overextension of the arterial lumen
and unnecessary endothelial damage. All patients must
have baseline neurological examination before BOT
followed by continuous neurological examination by
the same examiner throughout the procedure. If neurological symptoms or signs occur the procedure should
stop immediately. The time of occlusion varies in different studies from 15 min2 h; however, prolonged
occlusion should probably be avoided because of
increased rate of arterial damage found in animals
as directly related to increasing occlusion time according to MacDonald et al. (1994). Mathis et al. (1995)
demonstrated neurological complications rate of 1.6%
after BOT, which compares favorably with the 2.2%
average incidence of neurological complications from
other published reports.
Fox et al. (1987) and Anon et al. (1992) estimated
that the stroke risk associated with therapeutic
manipulation on ICA among the patients clinically
passed BOT is about 4.7% with zero mortality. These
results are way better than without BOT; however,
the risk is still significant.
During the last decades, several protocols were introduced to improve the detection of patients at risk for
ischemic strokes after the ICA sacrifice. In most of these
protocols, BOT is combined with various means to
directly or indirectly assess different parameters of the
CBF. Van Rooij et al. (2005) reported a study on 74 consecutive patients considered for therapeutic occlusion.
Authors hypothesized that local delay in venous phase
on cerebral angiogram may predict ischemic injury
upon permanent carotid artery occlusion. They performed cerebral angiograms during BOT to determine
792
the filling synchronicity of the cortical veins in

occluded and collateral vascular territories. Positive
predictive value of tolerance to carotid artery occlusion
after passing the angiographic test was as high as 98%.
The protocol has been demonstrated as safe and easy to
perform according to Van Rooij et al. (2005). Stable Xeenhanced computed tomography (CT) is another
modality widely used to predict patients tolerance to
carotid sacrifice was reported by Linskey et al. (1994)
and Johnson et al. (1991). The utilization of technetium-99m-hexamethylpropyleneamine oxime SPECT,
[15O]-labeled H2O positron emission tomography
(Brunberg et al., 1994), and perfusion magnetic resonance imaging (MRI) (Simonson et al., 1992) have also
been reported by Eckard et al. (1992) and Monsein et al.
(1991).
Kazuo and Takashi (1993), Hacke et al. (1981),
Cloughesy et al. (1993), Steed et al. (1990), and
Eckert et al. (1998) have all reported the various
common adjunctive methods studying indirect parameters of local CBF include evoked potential monitoring, electroencephalography, arterial stump blood
pressure measurement, and transcranial Doppler
ultrasonography. These studies have been shown as
significantly sensitive and specific in prediction of
brain tissue ischemic tolerance. Hypotensive challenge during BOT of the ICA is another fairly safe
and sensitive test, but its specificity needs further
investigation according to Standard et al. (1995).
Although the addition of CBF analysis to BOT
has been postulated as improving the detection of
patients at risk for strokes after ICA sacrifice, the
results of comparison of complications after sacrifice in patients having BOT alone versus BOT plus
are not in favor of the combination. The stroke risk
after passing clinical BOT alone as we mentioned is
about 4.7%. Surprisingly, combined data on 120
patients reported by AJNR (2001) who underwent
carotid occlusion after passing BOT combined with
any of adjunctive methods or induction of hypotension demonstrated 6.7% incidence of permanent
neurological deficit. The possible explanation is in
relatively small number of patients as well as wide
variations in patient selection, occlusion techniques,
and post-occlusion care. Considering available
information, it is impossible to prove superiority
for any particular method of BOT. Further prospective studies with standardized methodology enrolling larger number of patients are needed to establish
acceptable protocols conducting the highest benefit/
risk ratio.
D. ELIASHIV ET AL.
We have had extensive experience in our institution with the most common adjunctive procedure
consisting of electroencephalography (EEG) monitoring during BOT. Not only has EEG and evoked
potential monitoring become a widely used tool in
the operating room to safeguard the brain and spinal
cord during surgical procedures, it is commonly being
used during interventional neuroradiology procedures
to include BOT.
In the last two decades, there has been a major
shift from routine EEG and clinical evoked potential
studies to specialized, online operating room and
ICU monitoring (Spetzler et al., 1988; Razvi, 1992;
Nuwer, 1993).
The aim of neurophysiological monitoring during
BOT is the prompt identification of critical changes
that are suggestive of impending ischemia. It is used
as an adjunctive procedure in addition, not as a
replacement to vigilant constant clinical neurological
assessment of the patient for clinical signs. The neurologist should examine the patient at baseline and note
any baseline abnormalities on examination. The neurologist should pay attention to which particular vessel
is occluded and look for specific neurological deficits
that are to be expected from that particular territory.
Mild hemiparesis and/or homonymous hemianopsia
should be documented. As the literature relating to
neurophysiological monitoring during BOT is scant,
most of ours is derived from lessons learned during
EEG monitoring of carotid endarterectomies. These
are being applied in the neuroradiology suite. Realtime EEG monitoring is being applied during a variety
of endovascular procedures such as temporary balloon
test occlusion of the ICA, super selective amytal test
before embolization techniques used in the functional
evaluation of brain AVM.
Carotid balloon test occlusions with neurophysiological monitoring are used by interventional neuroradiologists as treatment or pretreatment evaluation
to assess the risk of a procedure where the ICA
may be sacrificed (Cloughesy et al., 1993).
Our experience in monitoring over 20 patients a
year undergoing temporary balloon test occlusion
demonstrates that the combination of online EEG or
evoked potential monitoring with frequent and
repeated clinical examinations provides a useful
approach in identifying patients at risk for stroke during carotid artery sacrifice.
There are presently no guidelines or published standards for real-time EEG or evoked potential monitoring during interventional neuroradiology procedures.
VASCULAR SURGERY
The goal of this chapter is to share our clinical experience, and provide useful guidelines that may be
used in this setting. Traditional EEG monitoring
techniques have been employed in the past as well
as the use of quantitative EEG and somatosensory
evoked potentials (SEPs) (Cloughesy et al., 1993).
While SEPs have been useful for monitoring cortical
areas of the rolandic area and have frequently been
used in carotid endarterectomy, this setting calls for
a more widespread coverage than just the central cortical areas, particularly for the determination of collateral circulation. Fourier-transformed quantitative
EEG monitoring techniques have been performed
utilizing very few electrodes. The evidence for the
utility of EEG monitoring as a useful adjunctive tool
in BOT is mainly supported through data from
EEG monitoring during carotid endarterectomies.
Sharbrough et al. (1973) have demonstrated the
utility of EEG monitoring during CEA procedures.
They found that EEG changes have a high sensitivity
in predicting ischemia and correlate highly with
CBF. Quantitative studies were later performed in
a large series by Sundt et al. (1981) which actually
determined the critical CBF required to maintain
a normal electroencephalogram (15 ml/100 g/min
of normal flow). These investigators studied 1,145
consecutive CEA with intraoperative CBF measurements and electroencephalograms followed by
postoperative EEGs and retinal artery pressure measurements. They concluded that the EEG has proved
to be a sensitive means of monitoring neurological
function and they never had a patient emerge from
anesthesia with a new deficit that was not predicted
by the EEG.
Cross-clamping of the carotid artery (analogous
to BOT) during CEA places patients at risk for developing ischemia during the process. In a study of 367
patients undergoing CEA, Chiappa et al. (1979)
showed the major EEG change indicative of ischemia
is the dropout of fast frequencies. This study demonstrates the various types of changes associated with
ischemia. The most significant change resulting from
ischemia according to this series was loss of superimposed faster frequencies. A significant change was
defined as a complete loss of all frequencies bilaterally
or unilaterally, a moderate change was defined as a
loss of anesthetic induced fast activity, and a mild
activity was defined as a gradual loss of amplitude
over 530 min following clamping of the carotid
artery. A loss of all EEG activity or suppression of
the ipsilateral hemisphere is highly likely due to
793
ischemia as a result of occlusion. According to these

authors, as a general rule, the loss of 7080% or more
over <1 min is treated as a significant change. Loss of
anesthetic-induced fast activity but preservation of
moderate-to-high voltage slowing, often rhythmic,
indicates a borderline or mild change but is not as significant as a loss of all activity. The EEG changes at
the time of balloon test occlusion are very comparable
to changes during cross-clamping of the carotid artery
during CEA monitoring, as mentioned above. We can
then take these lessons learned from investigators
(Sharbrough et al., 1973; Chiappa et al., 1979; Sundt
et al., 1981) who studied large populations of patients
undergoing carotid endarterectomies and apply them
to the balloon test occlusion process. In fact, the
only large-scale seminal series involving monitoring
carotid test occlusions with continuous large array
EEG and clinical examination was by Cloughesy
et al. (1993). In that series, 4/17 of consecutive cases
showed changes. One of the patients had clinical
changes with no EEG changes, one had attenuation
of alpha which was clinically deemed irrelevant and
insignificant but the other two had significant changes
similar to those described in ischemia. Hacke et al.
(1981) showed the utility of EEG during BOT but
used Fourier-transformed frequency analysis in two
patients utilizing only two channel recordings.
58.2. Balloon test occlusion protocol
Carotid balloon test occlusion is typically performed
with the patient lightly sedated for optimal neurological examinations and electrographical purposes. It is
necessary to use light levels of anesthesia to keep the
patient in the best awake state for effective testing
and monitoring purposes (Cloughesy et al., 1993).
A heavily sedated patient cannot undergo any testing
that is useful, and anesthetic agents may contaminate
the EEG with varying degrees of slowing or fast
activity that may lead to interpretation difficulties
during BTO. Communication between the anesthesiologist, neurophysiologist, and the EEG technologist
is crucial and highly recommended for the purpose of
selecting an appropriate anesthetic agent that will
achieve a desired state of analgesia with minimal
drug effect.
Diagnostic angiography performed before balloon
test occlusion involves the evaluation of both carotid
artery circulations. The angiogram is assessed for
vascular anatomy and secondary vascular abnormalities. The catheter is removed and replaced with a
794
balloon catheter. This will be used for the balloon

test occlusion. The selection of the balloon catheter
is based on the diagnostic angiogram. Usually the
internal carotid is occluded at high-cervical level,
C1 to C2 level. . . When the catheter is positioned,
heparin is administered intravenously. The balloon
is methodically inflated during fluoroscopy in order
to visualize the size of the artery. Arterial pressures
are monitored during the inflation of the balloon
while carefully examining the arterial waveform.
Occlusion is sustained by infusing a small amount
of contrast through the central catheter lumen. Once
arterial occlusion is sustained, the neurological
examination is carried out and the EEG is monitored
continuously during the 1530 min of BTO as
described below.
58.3. EEG monitoring
Electrodes are typically applied in the EEG laboratory ahead of time. Patients are asked to come in
about 1.5 h before the procedure for electrode application. The International 1020 system of electrode
placement is used and either gold cup electrodes are
applied with electrode paste or may be applied with
collodion. We find that securing the electrodes with
a bandage head wrap will keep the electrodes in
place when the patient has to be moved from the gurney to the angiography table. Electrode impedances
should be below 5 kO for optimal electrode integrity.
The head box may be attached with tape or other
means to the underside of the angiography table.
Care should be taken for head box stability, particularly with all the movement around the table. As in
the operating room, recording the EEG may be a challenging process. The same care that is applied to the
intraoperative process for carotid (CEA) or aneurysm
clipping monitoring is indicated in this environment
according to Kazuo and Takashi (1993). All settings
should conform to the EEG guidelines, the highfrequency filter should be set to 70 Hz and the use of
the 60 Hz filter is highly recommended. A lowfrequency filter of 1 Hz and a paper speed of 30/mm
per second for effective interpretation.
It is recommended to place the EEG instrumentation away from the table but close enough so that
communication between the team is optimal. The
use of digital video is encouraged, particularly for
after-the-fact review; it is a valuable approach that
improves the quality of the monitoring. We perform
continuous EEG monitoring with digital video
D. ELIASHIV ET AL.
utilizing a 32-channel biologic EEG system with a

24-channel montage. The use of digital video has
been gaining popularity in settings that involve a
multidisciplinary approach such as this. It is advantageous to collect data with video, as the technology is
readily available. The technologist may be able to
visualize and record subtle changes in the patient otherwise not obtained without the use of digital video.
Digital video has recently become an affordable tool
that can be used not only in long-term epilepsy monitoring and routine EEG studies but also in special
procedures like the balloon test occlusion. We recommend its use in this setting. For privacy issues
and HIPAA compliance, patients should be informed
in advance that they would be recorded on video.
Most hospitals have in place policy and procedures
for video recording of patients. A signed consent is
required and is similar to that used for the epilepsy
monitoring process.
A neurological examination was performed on the
patients by a qualified neurologist before the baseline
EEG and repeated through the evaluation process.
The baseline recording was obtained for a period of
10 min, we recommend at least 10 min of recording
during this time to ensure high-quality recordings
free of artifacts. This is necessary for optimal comparisons during the BOT. The patient was kept
lightly sedated to be able to maintain the best awake
state and be able to discern a posterior dominant
rhythm in the alpha range. A full compliment of electrode coverage very similar to the one utilized for
intraoperative monitoring for CEA was utilized.
The EEG was filtered with a bandpass of 0.3
70 Hz to exclude the possibility of artifacts. The
high-frequency filter may be reduced to 35 Hz particularly when excessive muscle obliterates the tracing.
A standard sensitivity of 7/mm was used and adjusted
for effective amplification. A standard double banana
montage with added transverse array was utilized
throughout the procedure.
The 10-min baseline was reviewed with emphasis
for preexisting abnormalities, as this may predict those
patients that may have less ischemic tolerance than
other patients without preexisting abnormalities do.
Patients with preexisting abnormalities have shown a
susceptibility to anesthesia at the time of BTO according to Sundt et al. (1981) in their study of the correlation of CBF and EEG changes during CEA, the EEG
showed good correlation with CBF. Critical flow factors and the type of anesthetic used particularly for
patients with preexisting abnormalities are of vital
VASCULAR SURGERY
importance and the reason for performing an adequate

baseline of at least 10 min. This allows enough time to
make an accurate assessment of the baseline before
BOT.
Continuous EEG was recorded before the test
occlusion. Traditional waveform analysis was utilized
for reviewing the waveforms during the baseline, and
post-occlusion. Repeated neurological examination
was performed beginning prior to the angiogram and
during the onset of BOT. The examination process is
repeated 20 or 30 min post-occlusion. Cloughesy
et al. (1993) reported that traditional neurological
examination was performed but limited to the supine
position, including cranial nerve, motor, sensory, coordination, and cognitive skills. We performed a similar
process for the examination lasting 510 min on average. The balloon test occlusion pass/fail criteria were
decided on the basis of an electrographical change during the BOT or changes in the clinical examination.
The patients baseline record is used as the control,
and this is compared with the EEG during and after
the occlusion. If negative findings during the BOT,
then a hypotensive challenge is carried out. If no
changes after 30 min, the test is discontinued.
Based on previous studies during BOT reported by
Cloughesy et al. (1993) and the vast literature in EEG
monitoring during CEA monitoring (Sharbrough
et al., 1973; Chiappa et al., 1979; Hacke et al., 1981;
Sundt et al., 1981), we have developed a grading system that is useful in the assessment of the EEG
changes associated with the balloon test occlusion
process (Table 1). Grading the EEG abnormality and
clinical examination is helpful to determine whether
to pass or fail a high-risk patient during the BOT process. A grade I classification is considered within
795
normal limits and described as no EEG changes during

the initial phase after BOT compared with baseline
associated with a normal neurological examination.
A grade II classification is considered a mild change
described as an intermittent increase of a preexisting
focal slowing with <20% change and no loss of superimposed focal fast activity. A grade III classification is
considered a moderate change and is associated with a
loss of superimposed fast activity with some slowing
in the delta range. A grade IV is considered a significant change characterized by the focal slowing of
varying degree, mostly high-amplitude delta frequencies. A critical change is characterized by a complete
loss of all frequencies or a severely attenuated
hemisphere.
The following case report in our series highlights
both the pitfalls of utilizing balloon occlusion testing
with and without adjunctive EEG monitoring and the
difficulties in interpretation of changes in patients
who have preexisting neurological deficits and/or
focal EEG slowing.
Sixty-five-year-old right-handed woman presented
with several year history of progressive visual loss
and right-sided headaches. MRI/MRA studies revealed a giant 2.5 cm aneurysm projecting off the
bifurcation of the right ICA. The patient underwent a
craniotomy and a decompression of the optic nerve
but safe clipping of the aneurysm could not be performed. The patient underwent a BOT as to determine
adequate collateral flow. At baseline, the background
activity showed 8 cps posterior dominant rhythm with
the right-sided focal slowing maximal over F4. Inflation of the balloon resulted in the prompt appearance
of delta slowing over the right hemisphere (grade IV)
with the loss of superimposed faster frequencies over
Table 1
EEG grading and clinical classification during balloon test occlusion of the carotid artery
Grade
EEG pattern
Clinical significance
I
II
Initial phase no EEG changes compared with baseline

Intermittent increase of preexisting focal slowing <20%
change, no loss of superimposed fast activity
Loss of superimposed focal fast activity with some slowing
in the delta range
Focal slowing of varying degree, mostly high-amplitude
delta frequencies
Complete loss of all frequencies or severely attenuated
hemisphere
WNL
Mild change (if examination WNL pass)
III
IV
V
EEG, electromyograph; WNL, within normal limits.
Moderate change (fail)

Significant change (fail)
Critical change (fail)
796
the right side (Fig. 1). Clinically, there was marked

weakness 3/5 of the left upper extremity and a left
central facial paresis.
The balloon was deflated and the clinical and
electrographic features resolved.
The patient subsequently underwent an external
carotid to the right middle cerebral artery bypass.
The patient was then subjected to a clinical BOT
without EEG. The patient developed left hemiparesis
and was administered intra-arterial tPA. The hemiparesis only partially resolved. Unfortunately, as the
patients collateral circulation did not suffice she
needed a repeat superficial temporal artery to the right
middle cerebral artery bypass surgery.
She subsequently had a repeat BOT approximately
with EEG 2 months later.
Baseline pre-inflation showed a right-sided breech
pattern as well as right-sided focal slowing in the
theta range (Fig. 2).
There were no clinical or electrographic changes
during the BOT, but there were fluctuations
D. ELIASHIV ET AL.
consistent with grade II. Of interest was some paradoxical increase in slowing with deflating of right
frontotemporal slowing possibly reflective of the
end of collateral circulation that was sustained during
the BOT run (Fig. 3).
The patient was able to tolerate occlusion of her
right ICA without further neurological sequalae.
58.4. Discussion
The physiological basis for EEG changes during
ischemia has been outlined above. Occlusion flows
of <10 ml/100 g/min always produced rapid changes
in the EEG and flows of <15 ml/100 g/min usually
caused EEG changes.
We performed a data query on 18 consecutive
BOT performed during 2005 in our institution.
We found that our data were very similar to the series
published by Cloughesy et al. (1993). All patients
underwent continuous online EEG monitoring with
digital video and clinical neurological examinations.
Fig. 1. Inflation of the balloon resulted in the prompt appearance of delta slowing over the right hemisphere (grade IV) with
the loss of superimposed faster frequencies over the right side.
VASCULAR SURGERY
797
Fig. 2. Baseline pre-inflation showed a right-sided breech pattern as well as right-sided focal slowing in the theta range.
One patient mentioned above underwent multiple

BOT. The discussed patient had a giant ICA aneurysm at the bifurcation of the medial cerebral artery
(MCA) and is an excellent example of the utility of
EEG with BOT. We had one additional patient with
loss of superimposed faster frequencies fail the
BOT. In one patient, the BOT was aborted because
of hemihypesthesia and the vessel was not occluded.
No EEG changes were noted. One patient had grade
II EEG changes noted during BOT consisting of mild
generalized paroxysmal slowing. The patient had no
associated changes on clinical examination and subsequently had sacrifice of his ICA with no sequalae.
Although our institution routinely involves a
hypotensive protocol if after 1015 min no clinical
or EEG signs are seen, we have not observed in our
small sample or in my more extensive experience
an added yield (unpublished data). Caution should
be utilized in the interpretation of EEG changes in
patients with preexisting focal slowing. We conclude
that neurophysiological monitoring is useful in

detecting changes associated during BOT.
58.5. Evoked potential monitoring
The use of SEPs is widely used in the operating room
to safeguard the brain and spinal cord during surgical
procedures. SEPs are short-latency components that
reflect activity in the spinal cord and brainstem pathways. Stimulation of the median nerves produces a
negative component at 20 ms, and a positive component at 2530 ms. The N20 component represents the
thalamocortical bundle and the arrival of the afferent
volley at the sensory cortex. Kazuo and Takashi
(1993) performed a retrospective study of 97 patients
with MCA or ICA aneurysms in whom median SEP
monitoring was carried out. All patients underwent
temporary vascular occlusion at the time of aneurysm
clipping. These investigators were able to clarify the
permissible time for temporary vascular occlusion
798
D. ELIASHIV ET AL.
Fig. 3. Some paradoxical increase in slowing with deflating of right frontotemporal slowing possibly reflective of the end
of collateral circulation that was sustained during the balloon occlusion test run.
during aneurysm surgery. They found that when

there is a gradual attenuation of the SEPs after temporary vascular occlusion, ischemic brain damage
does not occur if recirculation is established within
10 min after the absence of the SEP. In a study by
Ferbert et al. (1988), 28 patients presented with vertebrobasilar or basilar artery thrombosis. Brainstem
auditory evoked potentials (BAEPs) and SEPs were
recorded. These patients were monitored continuously during angiographic studies. They concluded
that the recording of the evoked potentials is a reliable tool in the diagnosis of basilar artery thrombosis.
Brainstem auditory evoked potentials are also
being applied in patients undergoing posterior fossa
or microvascular decompression surgeries. BAEPs
consist of a series of five to seven components of
very low amplitude thought to originate from sequential activation of the auditory pathway in the brainstem. The interpeak latency between wave I and
wave V represents brainstem transmission time,
which can be measured continuously. Both SEP and
BAEP are gaining popularity during interventional

neuroradiology due to the instant and reliable information they provide about the condition of the brainstem or posterior circulation during diagnostic or
therapeutic procedures. They provide the means of
determining whether a procedure is successful or
whether a procedure should be aborted.
We typically monitor patients with BAEPs during
interventional procedures such as thrombolytic therapy of the vertebrobasilar system. These are of help
to determine the effectiveness of the therapy and help
to safeguard brainstem circulation.
Evoked potential monitoring is typically carried
out using a Cadwell 16-channel evoked potential
machine. Brainstem auditory evoked responses are
obtained by stimulating each ear with rarefaction
clicks at an intensity of 85 dB nHL with a repetition
rate of 11.3/s. The contralateral ear is exposed to
white noise at 60 dB SPL. Recording montages are
ipsilateral ear to Cz and ipsilateral to contralateral
ears. Recording parameters are: LFF 100 Hz, HFF
VASCULAR SURGERY
3,000 Hz. The baseline recordings are obtained by

averaging two separate 2,000 trials to ensure reproducibility. During thee time the patient is monitored
we may reduce the number of trials to 500 as the
cycling period for 500 trials is 1 min. If the patient
is clinically stable with no changes we may increase
the number of trials. The clinical examination of the
patient is geared to the vascular territory tested that
is, visual fields, dysarthria. We utilize similar neurophysiological criteria for changes during evoked
potential intraoperative monitoring. We abort the balloon occlusion testing when there is either a 1 ms
prolongation of wave V, 10% increase either IV
IPL (interpeak latency), >50% decrease in amplitude
of any of the obligatory waveforms especially wave V
(James and Husain, 2005; Loiselle and Nuwer, 2005).
The duration of the study is similar to our protocol with
BOT for 1530 min.
References
AJNR (2001) Carotid artery balloon test occlusion. Am. J.
Neuroradiol., 22: S8S9.
Anon, VV, Aymard, A and Gobin, YP (1992) Balloon
occlusion of the internal carotid artery in 40 cases of
giant intracavernous aneurysm: technical aspects, cerebral monitoring and results. Neuroradiology, 34:
245251.
Brunberg, JA, Frey, KA, Horton, JA, Deveikis, JP, Ross,
DA and Koeppe, RA (1994) [15O]H2O position emission tomography determination of cerebral blood flow
during balloon test occlusion of the internal carotid
artery. AJNR Am. J. Neuroradiol., 15: 725732.
Chiappa, KH, Burke, SR and Young, RR (1979) Results of
electroencephalographic monitoring during 367 carotid
endarterectomies. Use of a dedicated minicomputer.
Stroke, 10(4): 381388.
Cloughesy, TF, Nuwer, MR, Hoch, D, Vinuela, F, Duckwiler, G and Martin, N (1993) Monitoring carotid test
occlusion with continuous EEG and clinical examination. J. Clin. Neurophysiol., 10(3): 363369.
Eckard, DA, Purdy, PD and Bonte, FJ (1992) Temporary
balloon occlusion of the carotid artery combined with
brain blood flow imaging as a test to predict tolerance
prior to permanent carotid sacrifice. AJNR Am. J. Neuroradiol., 13: 15651569.
Eckert, B, Thie, A, Carvajal, M, Groden, C and Zeumer, H
(1998) Predicting hemodynamic ischemia by transcranial
Doppler monitoring during therapeutic balloon occlusion
of the internal carotid artery. Am. J. Neuroradiol., 19(3):
577582.
Ferbert, A, Buchner, H, Bruckmann, H, Zeumer, H and
Hacke, W (1988) Evoked potentials in basilar artery
799
thrombosis correlation with clinical and angiographic
findings. Electro. enceph. Clin. Neurophysiol., 69:
136147.
Fox, AJ, Vinuela, F, Pelz, DM, Peerless, SJ, Ferguson, GG,
Drake, CG and Debrun, G (1987) Use of detachable balloons for proximal artery occlusion in the treatment of
unclippable cerebral aneurysms. J. Neurosurg., 66: 4046.
Hacke, W, Zeumer, H and Ringelstein, EB (1981) EEG controlled occlusion of the internal carotid artery during angiography. Neuroradiology, 22: 1922.
James, ML and Husain, AM (2005) Brainstem auditory
evoked potential monitoring when is change in wave
V significant? Neurology, 65: 15511555.
Johnson, DW, Stringer, WA, Marks, MP, Yonas, H, Good,
WF and Gur, D (1991) Stable Xenon CT cerebral blood
flow imaging: rationale for and role in clinical decision
making. Am. J. Neuroradiol., 12(2): 201213.
Kazuo, M and Takashi, Y (1993) Permissible temporary
occlusion time in aneurysm surgery as evaluated by
evoked potential monitoring. Neurosurgery, 33(3):
434440.
Linskey, ME, Jungreis, CA and Yonas, H (1994) Stroke risk
after abrupt carotid artery sacrifice: accuracy of preoperative assessment with balloon test occlusion of the internal carotid artery. AJNR Am. J. Neuroradiol., 15:
829834.
Loiselle, DL and Nuwer, MR (2005) When should we warn
the surgeon? Diagnosis-based warning for BAEP monitoring. Neurology, 65(10): 15221523.
MacDonald, JD, Gyorke, A, Jacobs, JM, Mohammad, SF,
Sunderland, PM and Reichman, MV (1994) Acute
phase vascular endothelial injury: a comparison of temporary arterial occlusion using an endovascular occlusive balloon catheter versus a temporary aneurysm clip
in a pig model. Neurosurgery, 34(5): 876881.
Matas, R (1911) Testing the efficiency of the collateral circulation as preliminary to the occlusion of the great surgical arteries. Ann. Surg., 53: 143.
Mathis, JM, Barr, JD, Jungreis, CA, Yonas, H, Sekhar, LN,
Vincent, D, Pentheny, SL and Horton, JA (1995) Temporary balloon test occlusion of the internal carotid artery:
experience in 500 cases. Am. J. Neuroradiol., 16: 749754.
Monsein, LH, Jeffrey, PJ and Van Heerden, BB (1991)
Assessing adequacy of collateral circulation during balloon test occlusion of the internal carotid artery with
99
mTc-HMPAO SPECT. AJNR Am. J. Neuroradiol.,
12: 10451051.
Nuwer, MR (1993) Intraoperative electroencephalography.
Serbinenko, FA (1974) Balloon catheterization and occlusion
of major cerebral vessels. J. Neurosurg., 41: 125145.
Sharbrough, F, Messick, J and Sundt, T, Jr. (1973) Correlation of continuous electroencephalograms with cerebral
blood flow measurements during carotid endarterectomy. Stroke, 4: 674683.
800
Simonson, TM, Ryals, TJ, Yuh, WTC, Farrar, GP, Rezai, K
and Hoffman, HT (1992) MR imaging and HMPAO
scintigraphy in conjunction with balloon test occlusion:
value in predicting sequelae after permanent carotid
occlusion. AJNR Am. J. Roentgenol., 159: 10631068.
Spetzler, RF, Hadley, MN, Rigamonte, D, Carter, LP, Raudzens, PA, Shedd, SA and Wilkerson, E (1988) Aneurysms
of the basilar artery treated with circulatory arrest, hypothermia and barbiturates cerebral protection. J. Neurosurg., 68: 868879.
Standard, SC, Ahuja, A, Guterman, LR, Chavis, TD, Gibbons, KJ, Barth, AP and Hopkins, LN (1995) Balloon test
occlusion of the internal carotid artery with hypotensive
challenge. Am. J. Neuroradiol., 16(7): 14531458.
D. ELIASHIV ET AL.
Steed, DL, Webster, MW and De Vries, EJ (1990) Clinical
observations on the effect of carotid artery occlusion on
cerebral blood flow mapped by xenon computed tomography and its correlation with carotid artery back pressure. J. Vasc. Surg., 11: 3844.
Sundt, T, Jr., Sharbrough, FW, Piepgras, DG, Kearns, TP,
Messick, JM, Jr. and OFallon, WM (1981) Correlation
of cerebral blood flow and electroencephalographic
changes during carotid endarterectomy. Mayo Clin.
Proc., 56: 533543.
Van Rooij, WJ, Sluzewski, M, Slob, MJ and Rinkel, GJ
(2005) Predictive value of angiographic testing for tolerance to therapeutic occlusion of the carotid artery.
Am. J. Neuroradiol., 26: 175178.

M.R. Nuwer (Ed.)
801
CHAPTER 59
Evoked potential monitoring during surgery

for intracranial aneurysms
Georg Neuloh* and Johannes Schramm
Department of Neurosurgery, Neurochirurgische Universitatsklinik, D-53105 Bonn, Germany
59.1. Introduction
This chapter focuses on evoked potential (EP) monitoring for cerebral aneurysm surgery and takes a look
at possible applications during endovascular procedures. It draws upon our own monitoring experience
during some 850 aneurysm surgeries, and upon the literature. It must be noted that all data available to date
are still descriptive in nature.
Monitoring of somatosensory EP (SEP) and brainstem auditory EP (BAEP) was introduced into cerebral
aneurysm surgery more than 20 years ago (Little et al.,
1983; Carter et al., 1984), and was first applied during
endovascular procedures at about the same time
(Hacke, 1985).
With growing experience from large case series,
monitoring of SEPs for posterior circulation aneurysms also BAEPs has become a standard adjunct
to the contemporary microsurgical approach in many
neurovascular centers. The more recent introduction
of reliable motor EP recordings during anesthesia
(Chapters 1618 in this volume) overcame some limitations of sensory EPs alone and has induced renewed
interest in neurophysiological monitoring methods
(Suzuki et al., 2003; Szelenyi et al., 2003; Neuloh
and Schramm, 2004a; Quinones-Hinojosa et al.,
2004; Horiuchi et al., 2005; Szelenyi et al., 2005).
A variety of other adjunctive techniques cannot be
adequately discussed in this chapter, such as microvascular Doppler-sonography (MVD) (Gilsbach and
Harders, 1989; Bailes et al., 1997; Firsching et al.,
2000; Stendel et al., 2000; Neuloh and Schramm,
2004a), direct measurements of brain tissue perfusion
Correspondence to: Georg Neuloh, M.D., Department of

Neurosurgery,
Neurochirurgische
Universitatsklinik,
D-53105 Bonn, Germany.
Tel.: 49-228-287-6521; fax: 49-228-287-4758;
E-mail: georg.neuloh@ukb.uni-bonn.de (G. Neuloh).
and oxygenation (Szelenyi et al., 2002), and processed

EEG (Nuwer, 1993).
59.2. Methods
This section summarizes some aspects of SEP, BAEP,
and motor evoked potential (MEP) recordings that are
relevant for aneurysm surgery monitoring. The techniques for stimulation and recording are described in
detail elsewhere in this volume (Chapters 1122) and
will therefore not be discussed here. In principle, these
standard techniques also apply for aneurysm monitoring. The skin incision for the typical frontotemporal
basal craniotomies does not interfere with an adequate
placement of stimulation and recording electrodes.
SEP monitoring relies on a close correlation with cortical cerebral perfusion (Symon, 1980). This correlation
appears to be weaker with ischemia at subcortical
levels, in particular in the brainstem (Branston et al.,
1984).
Both amplitude and latency of early cortical SEP
components (N20 for median, P40 for tibial nerve
SEPs) are suitable parameters. However, order and
speed of the changes depend on the dynamics of local
or global cerebral blood flow (CBF), which in turn
reflects on a multitude of factors including the cerebrovascular autoregulation and collateral perfusion and
systemic homeostasis parameters. Criteria that are
applied to assess intraoperative SEP changes are still
rather arbitrary, though substantiated by a long-term
clinical experience.
A 3050% attenuation of the amplitude of the primary cortical complex (N20/P40), and a 10% overall
latency increase (or a corresponding central conduction
time (CCT) prolongation by about 2 ms) is to be considered a significant impairment. In our experience,
combined latency increase and amplitude decrease or
802
amplitude attenuation alone are most frequently

observed in the intraoperative setting. Many papers
focus on latency (CCT) changes, as this is the major
parameter in conventional SEP diagnostics. Later SEP
components from association cortical areas are difficult
to assess and do not play a significant role in intraoperative monitoring. A further important parameter is the
speed at which SEP changes occur. Rapid SEP loss
within 2 min, for example, after temporary vessel
occlusion, seems to be an indicator of impending infarction, whereas slow and gradual amplitude attenuation is
more easily reversible (Mizoi and Yoshimoto, 1991).
59.2.2. Brainstem auditory evoked potentials
BAEPs have been employed alone or in combination
with SEPs for posterior fossa aneurysms. Less is
known about the exact correlation of BAEPs and
local CBF, but they seem to indicate the brainstems
general functional state more sensitively than SEPs
alone do, due to the extended course of the brainstem
auditory pathways (Manninen et al., 1994). Although
BAEPs are affected by decreased perfusion (Drake
et al., 1990), their sensitivity to detect ischemia
early enough for intraoperative monitoring purposes
still needs to be determined. The threshold for
complete BAEP loss is at very low CBF levels where
infarction is likely to occur (Kawahara et al., 1989;
Manninen et al., 1994).
A variety of inter-peak latency and amplitude differences have been proposed for intraoperative BAEP
assessment. In our experience, amplitude and latency
of the contralateral wave V are most easily monitored.
A criterion for deterioration of 50% amplitude
decrease and a 1 ms latency delay applies. This criterion fits with recent figures in the literature on BAEP
monitoring for hearing preservation during microvascular decompression procedures (Polo et al., 2004),
though in the latter case the auditory nerve is the target
structure for monitoring, rather than the brainstem
fiber tracts at risk during aneurysm surgery. Again,
sudden BAEP loss is a more urgent alarming sign than
gradual decrement or delay (Neu et al., 1999), which is
also more easily reversible. The typical averaging
cycle of 100150 s implies that a clear BAEP decline
should be reacted to when it is first noted.
59.2.3. Motor evoked potentials
Reliable intraoperative MEP monitoring under general
anesthesia requires the high-frequency train stimulation technique (see Chapters 1618 in this volume),
G. NEULOH AND J. SCHRAMM
which can be employed for direct cortical and transcranial stimulation (Taniguchi et al., 1993; Pechstein
et al., 1996). Both techniques have been explored for
spinal and, more recently, for intracranial tumor surgery (Deletis, 1993; Kothbauer et al., 1997; Kombos
et al., 2001; Zhou and Kelly, 2001; Neuloh et al.,
2004). Figure 7 in the Chapter 37 on brainstem
monitoring shows the typical semiquantitative correlation of intraoperative MEP readings and motor outcome, which has proven true in more than 600
surgeries for intracranial tumor and vascular surgery
in our department (Neuloh et al., 2004; Neuloh and
Schramm, 2004b), comparable with results from other
series in the literature (Kombos et al., 2001; Zhou and
Kelly, 2001). More recently, MEP monitoring has
been systematically explored during aneurysm surgery in a number of series (Suzuki et al., 2003; Szelenyi et al., 2003; Neuloh and Schramm, 2004a;
Quinones-Hinojosa et al., 2004; Horiuchi et al.,
2005; Szelenyi et al., 2005), and has been shown to
be far more sensitive for impending motor damage
as compared with SEPs (Neuloh and Schramm,
2004a; Horiuchi et al., 2005). Whereas direct cortical
stimulation has the advantage of a negligible stimulation artifact and stable motor responses, it only allows
for unilateral recordings, which is inadequate for midline lesions (Table 1) and requires the placement of a
subdural electrode over the motor cortex, which can
be difficult after subarachnoid hemorrhage (SAH)
and in particular with the typical frontotemporal basal
craniotomies for aneurysm surgery (Suzuki et al.,
2003; Horiuchi et al., 2005; Szelenyi et al., 2005).
Transcranial stimulation is independent from the craniotomy and allows for bilateral recordings, but can
be associated with a disturbing movement artifact,
and a too strong stimulation intensity may caudally
bypass the supratentorial target territory (Neuloh and
Schramm, 2004a; Quinones-Hinojosa et al., 2004;
Szelenyi et al., 2005). Therefore, the optimum MEP
stimulation parameters must be carefully chosen to
achieve good responses with a stimulation current as
low as possible (Table 2 and Fig. 1). In contrast to spinal surgery (Deletis, 2002; Sala et al., 2004) (see
Chapter 16 of this volume), MEP amplitude clearly
matters with intracranial, in particular supratentorial
procedures. However, all criteria applied to assess
MEP amplitude remain arbitrary to a certain extent,
as there are no experimental data on the correlation
of MEP amplitude and brain perfusion. In our experience, a 50% amplitude criterion pragmatically reflects
the threshold for clearly discernible MEP attenuation
beyond the typical level of spontaneous variation,
VASCULAR SURGERY
803
Table 1
Summarizes suggested EP recordings for specific intracranial aneurysm locations, as determined by the
territories of the target vessel, and of the vessels possibly affected during the surgical approach
Aneurysm
Suggested EP recordings
Anterior communicating artery
Bilateral tibial nerve SEPs

Unilateral (arm) MEPs: approach
Optional:
Contralateral recordings
Bilateral leg MEPs
Unilateral median nerve SEPs
Unilateral arm MEPs
Bilateral tibial nerve SEPs
Optional:
Bilateral leg MEPs
Unilateral median nerve SEPs
Medial cerebral artery (MCA)

Distal anterior cerebral artery (A2)
Pericallosal artery with interhemispheric approach
Iinternal carotid artery (ICA) including posterior communicating
artery, anterior choroidal artery, and ophthalmic artery
Complex ICA bifurcation aneurysms
Basilar tip, proximal posterior cerebral artery (PCA), and superior

cerebellar artery
Basilar/vertebral artery with proximal posterior inferior cerebellar

artery (PICA), anterior inferior cerebellar artery (AICA)
Distal PICA/AICA, distal PCA (P3/4), and distal MCA

(temporal M2/3)
Unilateral arm MEPs

Optional:
Contralateral recordings
Additional unilateral leg SEPs
Bilateral arm MEPs
Bilateral median nerve SEPs
Optional:
Bilateral BAEPs
Bilateral arm MEPs
Bilateral median nerve SEPs
Optional:
Bilateral BAEPs
No clearly suggested EP recordings
Optional:
SEPs/MEPs (approach)
MEP, motor evoked potential; EP, evoked potential; SEP, somatosensory evoked potential.
Recordings marked as optional appear useful but not indispensable, and may be recorded depending on the equipment and time
designated to patient setup for monitoring.
and is most often described in the literature (Suzuki

et al., 2003; Neuloh et al., 2004; Neuloh and
Schramm, 2004b; Horiuchi et al., 2005). A tighter criterion (Kombos et al., 2001) would in our experience lead to falsely normal responses, that is,
seemingly stable MEP recordings despite impending
new deficit, which is not observed in clinical series
applying the 50% threshold. However, if the spontaneous variability of MEP amplitudes is >50%, as it
occurs with careful transcranial stimulation, a tighter
criterion must be individually determined based on
the signal-to-noise ratio. A wider criterion (e.g., 30%
as frequently applied for SEPs) might misleadingly
classify MEP changes, which are unrelated to the
motor systems functional state as significant deterioration, and inadequate surgical maneuvers might be
the result. Isolated MEP latency increase without concomitant amplitude attenuation does rarely occur with
a poor signal-to-noise ratio, where initial small MEP
deflections may no longer be clearly discernible from
noise. With a better signal-to-noise ratio, a 1015%
latency increase can be considered a significant deterioration. As with sensory EPs, the speed at which
MEP loss occurs seems related to the severity of the
underlying event. Sudden complete MEP loss is more
often irreversible and heralding severe new paresis,
whereas gradual amplitude attenuation is more easily
reversible and indicates light and transient paresis.
804
Table 2
Summarizes the parameters suggested for direct cortical and transcranial motor cortex stimulation
and muscular MEP recordings
Direct cortical MEPs
Transcranial MEPs
Mode
Intensity
M1 hand knob according to SEP phase

reversal or MEP mapping via electrode
strip/grid
Monopolar anodal (), reference () at Fpz
Maximum 30 mA, slightly supra-threshold
Pulse width
Pulses in train
Interpulse ISI
Intertrain ISI
0.30.5 (1.0) ms
57
24 ms (250500 Hz)
10 (530) s
12 cm anterior to C1/C2 (better than

C3/C4 for supratentorial aneurysms),
needle/corkscrew electrodes
Anode ipsilateral ()
Maximum 220 mA, slightly
supra-threshold
0.30.5 (1.0) ms
57
24 ms (250500 Hz)
10 (530) s
For example, thenar, (hypothenar, forearm

flexor), anterior tibial muscle
Muscle-tendon, needle electrodes
303,000 Hz
Unaveraged waveforms
5080 ms
For example, thenar, (hypothenar, forearm

flexor), anterior tibial muscle
Muscle-tendon, needle electrodes
303,000 Hz
Unaveraged waveforms
5080 ms
20500 mV, depending on amplitude
20500 mV, depending on amplitude
Stimulation
Site
Recording
Sites
Mode
Filter
Averages
Time window
post-stimulus
Amplitude scale
EP, evoked potential; MEP, motor evoked potential; SEP, somatosensory evoked potential.
59.2.3.1. Specific considerations for aneurysm

surgery in children
Aneurysmal SAH does occur in children, though it
remains an uncommon finding (Lasjaunias et al.,
2005). In principle, the same considerations as for
adult patients do apply with neurophysiological monitoring for aneurysms in children (Sala et al., 2002).
Evoked potentials are less easily recorded in young
children, in particular under general anesthesia. It is
often difficult to obtain reliable MEPs in anesthetized
children below about 6 years of age, but there are
considerable interindividual differences.
sensitivity of SEPs for focal brainstem ischemia

(Manninen et al., 1994).
A small number of different parallel recordings
facilitate fast preparation for monitoring, and minimize technical faults and falsely positive responses
with unnecessary interruptions of the surgery. Table 1
summarizes the suggested recordings for specific
aneurysm locations, as detailed in the following section. Unilateral recordings may be complemented by
contralateral control recordings to reduce false positive results (Table 1).
59.3. Suggested recordings
59.3.1. Recordings for anterior circulation

aneurysms
The selection of recordings for a given operative

setting is determined both by the territory of the target
vessels and by the surgical approach. In general, SEP
recordings are suited to monitor cortical territories,
whereas MEPs are most useful for monitoring the subcortical motor pathways (Neuloh and Schramm,
2004a). BAEPs can be applied to improve the limited
Aneurysms of the internal carotid artery (ICA), including posterior communicating (Pcom), anterior choroidal (Achor), and ophthalmic aneurysms, require
monitoring of unilateral median nerve SEPs and MEPs
from upper extremity muscles. Arm muscle MEPs
seem to be representative for the deep pyramidal tract
as a whole which is at risk from the possible affection
VASCULAR SURGERY
805
80 mA
1
2
3
90 mA
100 mA
70 mA
200 V
Train frequency
Train count
10 ms
100 Hz
200 Hz
300 Hz
200 V
Current intensity
400 Hz
500 Hz
10 ms
Pulse width
0.5 ms
200 V
0.3 ms
6
200 V
0.1 ms
0.2 ms
10 ms
Polarity
+
10 ms
Current vs. voltage

0.7 ms
1 ms
10 ms
200 V
200 V
Current
Voltage
10 ms
Fig. 1. Influence of stimulation parameters on motor evoked potential (MEP) amplitude. The waveforms illustrate the influence of different MEP stimulation parameters on the amplitude, based on a train of 5 pulses with an inter-stimulus-interval
of 2 ms. The parameter constellation may be optimized to minimize stimulation current intensity in order to avoid bypassing the supratentorial target territory by distal excitation of the brainstem pyramidal tract. Modified with permission from
Neuloh and Schramm (2002).
of the lateral lenticulostriate perforating arteries that

supply the dorsal limb of the internal capsule. Complex, large ICA bifurcation aneurysms may require
additional unilateral tibial nerve SEPs if their size
and configuration suggest a possible involvement of
the proximal anterior cerebral artery (ACA).
For aneurysms of the medial cerebral artery
(MCA), unilaterally arm SEPs and MEPs are recorded.
However, with aneurysms distal to the M1/M2 bifurcation, inadvertent occlusion or insufficient collateralization of, for example, temporal MCA branches
may not be detected by EPs. New aphasia has been
observed despite SEP monitoring (Holland, 1998),
but is a rare event in our experience, if combined
SEP/MEP monitoring is supplemented by MVD readings (Neuloh and Schramm, 2004a).
Anterior cerebral artery (including anterior communicating artery, Acom) aneurysms require monitoring of tibial nerve SEPs. The perforating vessels that
are at risk during clipping of an Acom or distal ACA
aneurysms, including the recurrent artery of Heubner,
do not supply motor fibers, but the basal approach
may affect lateral lenticulostriate arteries, and (unilateral) MEP recordings are therefore advisable. In our
experience, arm MEPs are preferable for pyramidal

tract monitoring also for this application, since leg
MEPs typically require a much higher current intensity
for stimulation that is associated with a stronger muscle twitching artifact. It may nevertheless be worthwhile to prepare one leg muscle along with the arm
muscles, since lower extremity MEPs are sometimes
obtained at lower stimulation intensities. As opposed
to ICA/MCA aneurysms, unilateral SEP recordings
are clearly insufficient with midline lesions such as
Acom aneurysms where the cortical territories of both
ACAs are at risk.
59.3.2. Recordings for posterior circulation
aneurysms
dThese aneurysms are usually located close to the midline and require bilateral recordings. Aneurysms of the
basilar and vertebral artery as well as proximal aneurysms of their branches, including the proximal posterior cerebral artery (PCA), primarily require MEP
recordings, since direct or traction-related injury to the
pyramidal tract or affection of brainstem perforators
are the main risks in this context (Quinones-Hinojosa
806
et al., 2004). MEPs may be supplemented by SEPs and/

or BAEPs to increase the sensitivity of monitoring and
to provide some monitoring if MEP recordings fail.
For distal aneurysms, for example, of the posterior inferior cerebellar artery (PICA), MEP/SEP recordings may
help to detect retraction and manipulation-related complications, but cannot detect ischemia in the respective
distal territory.
The supratentorial posterior circulation distal to
the P1 segment of the PCA cannot be monitored by
SEP, MEP, or BAEP recordings. In theory, visually
evoked potentials (VEPs) could monitor this territory. However, early studies (Cedzich et al., 1987)
confirmed by recent results (Wiedemayer et al.,
2003) have shown that current flash evoked VEP
recordings is not reliable enough to serve as a useful
intraoperative monitoring tool.
59.4. Impact of EP monitoring on surgery
SEP monitoring alone alters the surgical procedure in
822% of surgeries (Schramm et al., 1994; Lopez
Med. n. SEPs
et al., 1999). Combined MEP/SEP monitoring altered

the surgical strategy in 20% of surgeries, mainly
MEPs (in 16%), occasionally SEPs (in 4%) in a more
recent series (Neuloh and Schramm, 2004a).
Definitive cessation of clipping and dissection is
usually not an option with aneurysms. If technical
reasons for EP changes are unlikely, significant
ischemia must be assumed, and dissection must be
temporarily halted until EPs recover, or specific measures are taken such as irrigation or papaverine for
mechanical vasospasm, or readjustment of retractors
to reverse, for example, traction or kinking of minor
vessels (Neuloh and Schramm, 2004a). In turn, stable
recordings allow for safe permanent/temporary occlusion, or extensive dissection for a better approach
(Fig. 2), whereas such surgical steps might be prematurely abandoned without the reassurance from functional monitoring (Schramm et al., 1990; Neuloh and
Schramm, 2004a). In addition to the immediate consequence of EP monitoring for one patient, there is also
some influence of monitoring on the surgeons individual learning curve, helping to identify possible
Thenar MEPs
Start OP
Dissecting
aneurysm
Carotid
occluded
Clipping
Aneurysm
1 V
2 mV
Aneurysm clipped
Carotid open
10 ms
10 ms
End OP
Fig. 2. Stable evoked potential (EP) recordings during temporary internal carotid artery (ICA) occlusion. The ICA was temporarily occluded during dissection and clipping of a large, proximal ICA aneurysm. Stable somatosensory EP (SEP) and
motor EP (MEP) recordings allowed the surgeon to safely complete the procedure, and the outcome was uneventful. With
permission from Neuloh and Schramm (2002).
VASCULAR SURGERY
faults and to find the optimum surgical solution

(Neuloh and Schramm, 2005).
59.4.1. Temporary vessel occlusion
The permissible occlusion time depends on the collateral vascularization of the territory at risk and may be
actively influenced by neuroprotective measures such
as hypothermia (Mooij et al., 1987) or electroencephalography (EEG)-burst-suppression barbiturate narcosis. It is impossible, though, to predict in the
individual patient when temporary clipping of a vessel
may lead to significant ischemia. The great advantage
of continuous EP measurements is their ability to
determine the point of impending neural damage.
In our experience (Schramm et al., 1990;
Schramm et al., 1994; Neuloh and Schramm, 2004a),
the occurrence and time delay of SEP deterioration
after vessel occlusion unpredictably ranges between
30 s and 45 min, and the duration varies widely without strict correlation with the duration of temporary
clipping, though long-lasting SEP loss and ischemia/new deficit seem to be more likely after protracted occlusion (Schramm et al., 1990; Neuloh
and Schramm, 2004a). MEPs are more sensitive and
tend to deteriorate earlier and recover later than
SEPs. In our recent series (Neuloh and Schramm,
2004a), MEPs vanished during temporary clipping
after 7 (210) min and recovered after 7.5 (1.530)
min. Longer MEP recovery times tended to be associated with longer occlusion times, but not with the
latency of clip removal after MEP deterioration.
In other series, MCA occlusion of 2 min after SEP
loss was tolerated without neurological sequalae,
whereas new deficit occurred with continued occlusion of more than 4 min (Buchthal et al., 1988).
Occlusion times under hypothermia (Mooij et al.,
1987; Buchthal et al., 1988) of 6.352 min was tolerated without new deficit, if reperfusion was achieved
within 24 min after SEP loss. The speed of SEP
deterioration/loss is an important indicator of the
permissible occlusion time: In one series (Momma
et al., 1987), full SEP recovery was achieved in 39/
42 patients with SEP loss at a mean delay of 8 min
after temporary clipping, despite a mean overall
occlusion time of 15.820.3 min, and though the mean
duration of continued occlusion after complete SEP
loss was 12 min. However, with rapid SEP loss within
15 min in three patients, SEPs did not recover and the
patients sustained persisting new hemiparesis. In
another series, if disappearance of SEPs took more
807
than 24 min, recovery within 20 min after clip

removal was to be expected, without lasting new
deficit (Momma et al., 1987; Symon et al., 1988).
In recent series (Suzuki et al., 2003; Horiuchi
et al., 2005), MEPs decreased or disappeared reversibly after temporary occlusion of the anterior choroidal artery (Achor) or ICA of only 10 s4 min
duration, predicting unimpaired outcome or transient
paresis, whereas irreversible MEP loss and permanent paresis was observed after 17 min temporary
clipping of the ICA in one case. Recovery of MEPs
occurred within 30 s28 min, and tended to be longer
with longer duration of temporary clipping. SEP
deterioration was only observed after MEP change/loss
had occurred. Similarly, in another series (QuinonesHinojosa et al., 2004), SEP deterioration occurred
25 min after MEP loss.
Immediate reversion of temporary clipping is possible in some instances of EP deterioration if vessel
occlusion was done only to allow for safe extensive
dissection, and provisional clipping of ruptured aneurysm instead of temporary vessel occlusion may
occasionally be performed. However, immediate
reperfusion is frequently not an option. Instead,
adverse monitoring results may trigger some speeding up of the surgery, increase in blood pressure, or
measures to reverse additional mechanical vasospasm. In contrast, stable recordings enable the surgeon to continue safely with the procedure without
undue haste for an optimum surgical result.
59.4.2. Inadvertent occlusion
Inadvertent occlusion of cerebral artery branches is
reflected by EPs if they monitor its territory (Fig. 3).
The time to EP change compares naturally to the above
safe periods for temporary clipping. However, accidental occlusion of major vessels is detected by immediate
careful inspection in the majority of cases, or even better by the application of MVD, clearly before EP deterioration occurs (Neuloh and Schramm, 2004a). In our
study comparing EP monitoring and MVD, it was
therefore concluded that MVD is superior to EPs for
the detection of inadvertent vessel occlusion. However,
the affection of minor vessels may go undetected by
inspection and MVD, and may be only detected by
EP recordings, with MEPs yielding again the highest
sensitivity as compared with SEPs (Neuloh and
Schramm, 2004a). Stable EP recordings do not exclude
accidental occlusion of, for example, temporal MCA
branches, and careful inspection is still mandatory.
808
M1 Bifurcation Aneurysm
Median nerve SEPs
Start OP
Clipping aneurysm
Warning
1 V
Inadvertent closure of
M2 branche detected
Readjusting clip
End OP
10 ms
Fig. 3. Somatosensory evoked potentials (SEPs) detect

inadvertent M2 occlusion. Median nerve SEPs deteriorated
shortly after clipping of a medial cerebral artery (MCA)
bifurcation aneurysm. Inspection revealed inadvertent
occlusion of one M2 branch. SEPs recovered fully within
3 min after readjustment of the clip. With permission from
Neuloh and Schramm (2002).
59.4.3. Permanent vessel occlusion

With intentional permanent complete or partial vessel
occlusion, EP monitoring helps to assess whether collateral flow is sufficient to avoid ischemia. Stable recordings allow for safe vessel occlusion (Neuloh and
Schramm, 2004a) if test occlusion under awake conditions is not available, for example, when intraoperatively deciding to trap an otherwise not eliminable
proximal ICA aneurysm, or when clipping a calcified
aneurysm with narrowing of the carrier vessel.
59.4.4. Brain retraction and mechanical
vasospasm
The basal approach for aneurysm surgery can be a
major mechanism of neural damage. Direct axonal
traction-related injury occurs, but impaired microperfusion and affection of minor perforating vessels
may be more frequent during the extensive frontal
retraction required for clipping of some aneurysms
(Fig. 4). Mechanical vasospasm, for example, when

dissecting in the sylvian fissure, and kinking of small
perforators are further possible reasons for ischemia
which may be particularly difficult to detect. SEP
monitoring may not reliably pick up such events
either, which do not lead to reduced perfusion in
large territories. In our previously evaluated series
on SEP monitoring alone (Schramm et al., 1990),
such events were hardly reflected by SEP changes
(in 2/13 cases). In contrast, in our more recent series
(Neuloh and Schramm, 2004a), MEP deterioration
indicated such potentially hazardous events in 15/19
detectable cases (SEPs in 3/19 cases), inducing measures such as readjustment of retractors, cessation of
perforator manipulation, or application of papaverine
and irrigation of spastic vessels to prevent lasting
new deficit (Figs. 5 and 6). Extensive brainstem or
cerebellar retraction for aneurysms of the posterior
circulation can lead to both SEP/BAEP and MEP
deterioration (Little et al., 1987; Schramm et al.,
1990; Quinones-Hinojosa et al., 2004), and we found
that new deficit may occur in such a constellation,
which remains transient if the retractor is readjusted
immediately as a reaction to MEP attenuation.
59.5. Impact of monitoring on outcome
There would be no reason to perform monitoring if
surgical/interventional outcome was not positively
affected. The true impact of the method on clinical
outcome and surgical results could only be clearly
assessed by prospectively controlled studies. A mere
retrospective comparison of series with and without
monitoring may be very misleading: transient slight
deficit was found in our prospective series focusing
on motor outcome (Neuloh and Schramm, 2004a) in
a higher proportion of cases (10%) than in most retrospective clinical series. Moreover, as detailed in
the previous section, the whole surgical strategy
including the decision whether to perform for certain
critical surgical steps is potentially influenced by the
availability and the results of functional monitoring.
Even with neurovascular emergency surgery, more
so with elective procedures, there are always different ways to proceed, and the availability of monitoring may enable the surgeon to opt for the fastest
and, on the long run, safest strategy for the patient
instead of side-stepping to an alternative, easier strategy, for example, incomplete clipping of an calcified
aneurysm instead of accepting some partial occlusion
of the adjacent vessel. Thus the surgical result in
VASCULAR SURGERY
809
Med. n. SEPs
Thenar MEPs
Start OP
Dissecting
aneurysm
1st clip
Change frontal
retractor
Further
dissection
Readjust retractor
Temporary
carotid occlusion
1 V
100 V
Final clip
Carotid open
10 ms
10 ms
Frontal edema/concussion
Transient hemiparesis
Fig. 4. Motor evoked potential (MEP) deterioration and stable somatosensory evoked potentials (SEPs) during internal
carotid artery (ICA) occlusion and frontal brain elevation. MEPs deteriorated during dissection of a large carotid aneurysm
under and frontal brain retraction and vanished after temporary occlusion of the ICA, but recovered partially after ICA
reperfusion. SEPs remained unchanged throughout the procedure. There was a transient postoperative hemiparesis, and
CCT showed frontobasal edema, possibly from extensive brain retraction. With permission from Neuloh and Schramm
(2004b).
terms of, for example, complete elimination of the

aneurysm may depend on monitoring.
59.6. Monitoring for endovascular procedures
Endovascular coil occlusion has become the procedure of primary choice in many instances of intracranial aneurysms (Molyneux et al., 2002, 2005). The
literature on EP monitoring during such procedures
is sparse, though monitoring had already been
described very early in this context (Hacke et al.,
1983; Hacke, 1985), and the >49% rate of permanent procedure-related morbidity (Winn, 2004; Ross
and Dhillon, 2005) may suggest some functional
monitoring. In principle, the same considerations as
for neurovascular surgery seem to apply, and similar
observations have been described (John et al., 1989;
Liu et al., 2003). In a recent study (Liu et al.,
2003), changes in SEP/BAEP or EEG recordings
were found in 26% of 35 endovascular procedures,
including aneurysm coil embolization and temporary
or permanent vessel occlusion. The management was

altered in 14% by monitoring results.
As opposed to surgery, direct axonal injury by
extensive brain retraction cannot occur during endovascular coil embolization. The affected vessels are
more easily determined on the angiogram, and monitoring can specifically be designed to the target vessels of super-selective catheterization. Unnoticed
occlusion of major vessels does not occur under
angiographic control. Therefore, the typical target
events for EP monitoring include catheter-induced
vasospasm, inadvertent occlusion of minor (perforating) arteries, for example, from stenting of a large
vessel, and intentional permanent occlusion of large
vessels (Liu et al., 2003). As with monitoring during
surgery, false-negative SEP responses have been
observed in some cases (Kuroda et al., 1991; Liu
et al., 2003), and MEP monitoring has proven useful
so far for endovascular management of spinal (Sala
et al., 2001; Niimi et al., 2004) and cerebral (Rohde
et al., 1999) AVMs.
810

SEPs
MEPs
flexor hypothenar median nerve
Dissecting
LSAs
around
aneurysm
Clip
Inspection/
Closing
MD OK
15 min.
Warning
Inspection:
LSAs OK
Irrigation
500 V
20 ms
2 V
20 ms
Second M1 (LSA) aneurysm, 7 mths after

SAH HH V.Persisting severe->light paresis
Fig. 5. Somatosensory evoked potential (SEP) and motor evoked potential (MEP) loss indicate subcortical stroke. MEPs
and SEPs deteriorated, and vanished successively after dissection of the lateral leticulostriate group of perforating arteries
for clipping of a proximal M1 aneurysm. Inspection was inconspicuous, but there was a dense hemiparesis postoperatively,
which resolved only partially during the further course, and CCT revealed a subcortical ischemia comprising the dorsal limb
of the internal capsule.
59.7. Staffing and equipment

General staffing issues for intraoperative monitoring
are discussed in Nuwer (2002). It appears mandatory
that the surgeon involved has himself some understanding of the method. In our department, monitoring
is primarily performed by well-trained, experienced
technicians according to a standardized program for a
given indication. A neurologist or neurosurgeon experienced in intraoperative clinical neurophysiology is
involved only if technical problems arise, or if specific
modifications, for example, of the MEP stimulation
are necessary to obtain the desired results.
Most certified monitoring systems available on the
market provide the stimulation and recording techniques required for monitoring during aneurysm surgery. A sufficient number of 48 recording channels
for bilateral SEP and MEP readings and two independent electrical stimulators that allow for highfrequency repetitive MEP stimulation and SEP
stimulation in an alternated fashion should be available. For lower extremity muscle MEPs, the 100 mA
maximum output of most certified devices in Europe
and North America is not always sufficient. However,
aneurysm surgery monitoring can be satisfactorily performed with arm MEPs. Medico-legal implications
should be taken into consideration when using noncertified devices (Nuwer, 2002).
59.8. Limitations
EPs can obviously monitor only certain areas and
pathways, as discussed in more detail earlier, though
the combination of different recording enhances the
sensitivity of monitoring. Technical and anesthetical
reasons for EP changes must be carefully ruled out
to minimize false alarms and inadequate changes of
the surgical strategy.
It is a common misunderstanding that the surgeon
must decide on monitoring results alone how to
VASCULAR SURGERY
thenar
MEPs
hypothenar
811
SEPs
median n.
Dissecting
M1
Dissecting 2
LSAs off
aneurysm
Closing
Clip
Warning
Inspection:
Kinking of
3rd LSA:
cushioning
MD: OK
500 V
20 ms
200 V
20 ms
2 V
20 ms
Incidental M1 (LSA) aneurysm

Postoperative transient paresis, fine movement impairment
Fig. 6. Motor evoked potential (MEP) deterioration triggers intervention to prevent permanent new deficit. MEPs deteriorated after dissection of the lateral leticulostriate group of perforating arteries for clipping of a proximal M1 aneurysm much
like in the case of Fig. 5. Somatosensory evoked potentials (SEPs) remained stable. A second look triggered by this monitoring result revealed kinking of one perforating branch which was eliminated. There was transient light new paresis postoperatively, and a more protracted fine movement disturbance. Serial CCTs revealed some early subcortical hypodensity
extending into the internal capsule, but only a small definite infarction within the basal ganglia on a late control scan.
proceed in a given situation. The surgeon must be

aware of the specific conditions in a given case, from
the approach and the previous course of the operation
to the local anatomy of the aneurysm, in order to adequately appreciate the monitoring results.
The limited repertoire of surgical measures that
can be triggered by an adverse monitoring result constitutes another limitation, though not inherent to the
method, in particular with neurovascular procedures.
It is, for example, not always possible to reverse temporary clipping immediately, and trapping of a ruptured aneurysm may be the only option at hand.
Furthermore, the cause for EP deterioration may not
always become clear, for example, kinking or spasm
of small perforators may go undetected despite all
inspection. It requires an experienced surgeon to
decide on the most likely complication and the most

adequate reaction in such a situation.
The feasibility of EPs is never 100%. Successful
recordings can be achieved in about 9099% of cases
(Neuloh and Schramm, 2004a; Szelenyi et al., 2005).
However, the movement artifact associated with transcranial stimulation may be incompatible with the
microvascular procedure, and MEP recordings at a
reasonable signal-to-noise ratio may be impossible in
those cases (Neuloh and Schramm, 2004b). Steadystate partial muscle relaxation (Kalkman et al., 1992)
is difficult realize, and direct cortical stimulation via
subdural electrodes is not possible if bilateral recordings are required.
Finally, the lack of evidence from controlled studies at the present point of time still leaves it to the
812
individual surgeon how to weight the monitoring data

against other intraoperative observations.
59.9. Outlook to future developments
Whereas SEPs and, to some extent, BAEPs are
fully established monitoring methods during aneurysm surgery, the optimum MEP stimulation technique (transcranial vs. direct cortical) still needs to
be determined. There are ongoing attempts to modify
the VEP stimulation paradigm in order to obtain
responses reliable enough for monitoring of the distal
PCA territory. Processed EEG and adjunctive techniques such as direct measurement of brain perfusion
and oxygenation may help to overcome some of the
limitations of EP monitoring. However, all of these
methods will be fully appreciated only if their benefit
is ascertained by future controlled studies.
References
Bailes, JE, Tantuwaya, LS, Fukushima, T, Schurman, GW
and Davis, D (1997) Intraoperative microvascular Doppler sonography in aneurysm surgery. Neurosurgery,
40(5): 965970; discussion 970972.
Branston, NM, Ladds, A, Symon, L and Wang, AD (1984)
Comparison of the effects of ischaemia on early components of the somatosensory evoked potential in
brainstem, thalamus, and cerebral cortex. J. Cereb.
Blood Flow Metab., 4(1): 6881.
Buchthal, A, Belopavlovic, M and Mooij, JJ (1988) Evoked
potential monitoring and temporary clipping in cerebral
aneurysm surgery. Acta Neurochir., 93(12): 2836.
Carter, LP, Raudzens, PA, Gaines, C and Crowell, RM
(1984) Somatosensory evoked potentials and cortical
blood flow during craniotomy for vascular disease. Neurosurgery, 15(1): 2228.
Cedzich, C, Schramm, J and Fahlbusch, R (1987) Are
flash-evoked visual potentials useful for intraoperative
monitoring of visual pathway function? Neurosurgery,
21(5): 709715.
63: 201214.
Deletis, V (2002) Intraoperative neurophysiology and methodologies used to monitor the functional integriy of the
motor system. In: V Deletis and J Shils (Eds.), Neurophysiology in Neurosurgery. Academic Press, London,
pp. 2551.
Drake, ME, Jr., Pakalnis, A, Padamadan, H and Hietter, SA
(1990) Auditory evoked potentials in vertebrobasilar
transient ischemic attacks. Clin. Electroencephalogr.,
21(2): 96100.

Firsching, R, Synowitz, HJ and Hanebeck, J (2000) Practicability of intraoperative microvascular Doppler sonography in aneurysm surgery. Minim. Invasive Neurosurg.,
43(3): 144148.
Gilsbach, JM and Harders, AG (1989) Microvascular and
transcranial Doppler sonographic evaluation of cerebral
aneurysm flow pattern. Neurol. Res., 11(1): 4148.
Hacke, W (1985) Neuromonitoring during interventional neuroradiology. Cent. Nerv. Syst. Trauma, 2(2): 123136.
Hacke, W, Zeumer, H and Berg-Dammer, E (1983) Monitoring of hemispheric or brainstem functions with neurophysiologic methods during interventional neuroradiology. AJNR Am. J. Neuroradiol., 4(3): 382384.
Holland, NR (1998) Subcortical strokes from intracranial
aneurysm surgery: implications for intraoperative neuromonitoring. J. Clin. Neurophysiol., 15(5): 439446.
Horiuchi, K, Suzuki, K, Sasaki, T, Matsumoto, M, Sakuma, J,
Konno, Y, Oinuma, M, Itakura, T and Kodama, N (2005)
Intraoperative monitoring of blood flow insufficiency during surgery of middle cerebral artery aneurysms. J. Neurosurg., 103(2): 275283.
John, ER, Chabot, RJ, Prichep, LS, Ransohoff, J, Epstein, F
and Berenstein, A (1989) Real-time intraoperative monitoring during neurosurgical and neuroradiological procedures. J. Clin. Neurophysiol., 6(2): 125158.
Kalkman, CJ, Drummond, JC, Kennelly, NA, Patel, PM and
Partridge, BL (1992) Intraoperative monitoring of tibialis
anterior muscle motor evoked responses to transcranial
electrical stimulation during partial neuromuscular
blockade. Anesth. Analg., 75(4): 584589.
Kawahara, N, Sasaki, M, Mii, K, Tsuzuki, M and Takakura,
K (1989) Sequential changes of auditory brainstem
responses in relation to intracranial and cerebral perfusion pressure and initiation of secondary brainstem
damage. Acta Neurochir. (Wien), 100(34): 142149.
Kombos, T, Suess, O, Ciklatekerlio, O and Brock, M
(2001) Monitoring of intraoperative motor evoked
potentials to increase the safety of surgery in and
around the motor cortex. J. Neurosurg., 95(4): 608614.
Kothbauer, K, Deletis, V and Epstein, FJ (1997) Intraoperative spinal cord monitoring for intramedullary surgery: an essential adjunct. Pediatr. Neurosurg., 26(5):
247254.
Kuroda, S, Yonekawa, Y, Kawano, T, Yamashita, K,
Handa, H, Gotoh, Y, Tanaka, K, Kaku, Y and Taki, W
(1991) SEP monitoring during balloon occlusion test
or operation for vertebro-basilar aneurysms. No Shinkei
Geka, 19(4): 343348.
Lasjaunias, P, Wuppalapati, S, Alvarez, H, Rodesch, G and
Ozanne, A (2005) Intracranial aneurysms in children
aged under 15 years: review of 59 consecutive children
with 75 aneurysms. Childs Nerv. Syst., 21(6): 437450.
Little, JR, Lesser, RP, Luders, H and Furlan, AJ (1983)
Brainstem auditory evoked potentials in posterior circulation surgery. Neurosurgery, 12(5): 496502.
VASCULAR SURGERY
Liu, AY, Lopez, JR, Do, HM, Steinberg, GK, Cockroft, K
and Marks, MP (2003) Neurophysiological monitoring
in the endovascular therapy of aneurysms. AJNR Am. J.
Neuroradiol., 24(8): 15201527.
Lopez, JR, Chang, SD and Steinberg, GK (1999) The use
of electrophysiological monitoring in the intraoperative
management of intracranial aneurysms. J. Neurol. Neurosurg. Psychiatry, 66(2): 189196.
modalities. Can. J. Anaesth., 41(2): 9297.
Mizoi, K and Yoshimoto, T (1991) Intraoperative monitoring of the somatosensory evoked potentials and cerebral
blood flow during aneurysm surgery safety evaluation
for temporary vascular occlusion. Neurol. Med. Chir.
(Tokyo), 31(6): 318325.
Molyneux, A, Kerr, R, Stratton, I, Sandercock, P, Clarke,
M, Shrimpton, J and Holman, R (2002) International
subarachnoid aneurysm trial (ISAT) of neurosurgical
clipping versus endovascular coiling in 2143 patients
with ruptured intracranial aneurysms: a randomised
trial. Lancet, 360(9342): 12671274.
Molyneux, AJ, Kerr, RS, Yu, LM, Clarke, M, Sneade, M,
Yarnold, JA and Sandercock, P (2005) International
subarachnoid aneurysm trial (ISAT) of neurosurgical
clipping versus endovascular coiling in 2143 patients
with ruptured intracranial aneurysms: a randomised
comparison of effects on survival, dependency, seizures,
rebleeding, subgroups, and aneurysm occlusion. Lancet,
366(9488): 809817.
Momma, F, Wang, AD and Symon, L (1987) Effects of
temporary arterial occlusion on somatosensory evoked
responses in aneurysm surgery. Surg. Neurol., 27(4):
343352.
Mooij, JJ, Buchthal, A and Belopavlovic, M (1987) Somatosensory evoked potential monitoring of temporary middle
cerebral artery occlusion during aneurysm operation.
Neu, M, Strauss, C, Romstock, J, Bischoff, B and Fahlbusch,
R (1999) The prognostic value of intraoperative BAEP
patterns in acoustic neurinoma surgery. Clin. Neurophysiol., 110(11): 19351941.
Neuloh, G and Schramm, J (2002) Intraoperative neurophysiological mapping and monitoring for supratentorial procedures. In: V Deletis and JL Shils (Eds.),
Neurophysiology in Neurosurgery. Academic Press/
Elsevier Science, London, 339401.
Neuloh, G and Schramm, J (2004a) Monitoring of motor
evoked potentials compared with somatosensory evoked
potentials and microvascular Doppler ultrasonography
813
in cerebral aneurysm surgery. J. Neurosurg., 100(3):
389399.
Neuloh, G and Schramm, J (2004b) Motor evoked potential
monitoring for the surgery of brain tumours and vascular
malformations. Adv. Tech. Stand. Neurosurg., 29: 171228.
Neuloh, G and Schramm, J (2005) What the surgeon wins,
and what the surgeon loses from intraoperative
neurophysiologic monitoring? Acta Neurochir. (Wien),
147(8): 811813.
(2004) Motor evoked potential monitoring with supratentorial surgery. Neurosurgery, 54(5): 10611070; discussion 10701072.
Niimi, Y, Sala, F, Deletis, V, Setton, A, De Camargo, AB
and Berenstein, A (2004) Neurophysiologic monitoring
and pharmacologic provocative testing for embolization
of spinal cord arteriovenous malformations. AJNR Am.
J. Neuroradiol., 25(7): 11311138.
Nuwer, MR (1993) Intraoperative electroencephalography.
Nuwer, MR (2002) Regulatory and medicallegal aspects
of intraoperative monitoring. J. Clin. Neurophysiol.,
19(5): 387395.
Polo, G, Fischer, C, Sindou, MP and Marneffe, V (2004)
Brainstem auditory evoked potential monitoring during
microvascular decompression for hemifacial spasm:
intraoperative brainstem auditory evoked potential
changes and warning values to prevent hearing loss
prospective study in a consecutive series of 84 patients.
Quinones-Hinojosa, A, Alam, M, Lyon, R, Yingling, CD
and Lawton, MT (2004) Transcranial motor evoked
potentials during basilar artery aneurysm surgery: technique application for 30 consecutive patients. Neurosurgery, 54(4): 916924; discussion 924.
Rohde, V, Will, BE, Hahn, G, Bien, S and Zentner, J
(1999) Motor evoked potentials during embolization of
arteriovenous malformations for the detection of ischemic complications. Zentralbl. Neurochir., 60(2): 7480.
Ross, IB and Dhillon, GS (2005) Complications of endovascular treatment of cerebral aneurysms. Surg. Neurol.,
64(1): 1218; discussion 1819.
Neuroprotective role of neurophysiological monitoring
Ann. N. Y. Acad. Sci., 939: 126136.
why, when, how? Childs Nerv. Syst., 18(67): 264287.
814
Schramm, J, Koht, A, Schmidt, G, Pechstein, U, Taniguchi, M
and Fahlbusch, R (1990) Surgical and electrophysiological
observations during clipping of 134 aneurysms with
evoked potential monitoring. Neurosurgery, 26(1): 6170.
Schramm, J, Zentner, J and Pechstein, U (1994) Intraoperative SEP monitoring in aneurysm surgery. Neurol. Res.,
16(1): 2022.
Stendel, R, Pietila, T, Al Hassan, AA, Schilling, A and
Brock, M (2000) Intraoperative microvascular Doppler
ultrasonography in cerebral aneurysm surgery.
J. Neurol. Neurosurg. Psychiatry, 68(1): 2935.
Suzuki, K, Kodama, N, Sasaki, T, Matsumoto, M, Konno, Y,
Sakuma, J, Oinuma, M and Murakawa, M (2003) Intraoperative monitoring of blood flow insufficiency in the
anterior choroidal artery during aneurysm surgery.
J. Neurosurg., 98(3): 507514.
Symon, L (1980) The relationship between CBF, evoked
potentials and the clinical features in cerebral ischaemia. Acta Neurol. Scand. Suppl., 78: 175190.
Symon, L, Momma, F and Murota, T (1988) Assessment of
reversible cerebral ischaemia in man: intraoperative monitoring of the somatosensory evoked response. Acta Neurochir. Suppl. (Wien), 42: 37.
Szelenyi, A, Jung, CS, Schon, H and Seifert, V (2002)
Brain tissue oxygenation monitoring supplementary to

somatosensory evoked potential monitoring for aneurysm surgery. Initial clinical experience. Neurol. Res.,
24(6): 555562.
Szelenyi, A, Bueno de Camargo, A, Flamm, E and Deletis,
V (2003) Neurophysiological criteria for intraoperative
prediction of pure motor hemiplegia during aneurysm
surgery. Case report. J. Neurosurg., 99(3): 575578.
Szelenyi, A, Kothbauer, K, Bueno de Camargo, A, Langer, D,
Flamm, ES and Deletis, V (2005) Motor evoked potential
monitoring during cerebral aneurysm surgery: technical
aspects and comparison of transcranial and direct cortical
stimulation. Neurosurgery, 57(4 Suppl): 331338; discussion 331338.
Taniguchi, M, Cedzich, C and Schramm, J (1993) Modification of cortical stimulation for motor evoked potentials under general anesthesia: technical description.
Wiedemayer, H, Fauser, B, Armbruster, W, Gasser, T and
Stolke, D (2003) Visual evoked potentials for intraoperative neurophysiologic monitoring using total intravenous
Winn, HR (2004) Youmans Neurological Surgery:
A Comprehensive Reference Guide to the Diagnosis
and Management of Neurosurgical Problems. W.B.
Zhou, HH and Kelly, PJ (2001) Transcranial electrical motor
evoked potential monitoring for brain tumor resection.

M.R. Nuwer (Ed.)
815
CHAPTER 60
Spinal cord monitoring during descending

aortic procedures
David B. MacDonalda,* and Charles C.J. Dongb
a
Department of Neurosciences, King Faisal Specialist Hospital and Research Center, Riyadh 11211, Saudi Arabia
b
Department of Surgery, Vancouver General Hospital, Vancouver, BC V5Z 1M9, Canada
60.1. Introduction
Procedures on the descending aorta risk devastating
spinal cord infarction due to temporary ischemia or
permanent devascularization. Evoked potential monitoring might reduce this risk by guiding measures to
restore perfusion. However, the abrupt onset of ischemia, its rapid progression to infarction, and the frequent occurrence of confounding factors that also
affect evoked potentials all present significant challenges. Practitioners should understand aortic pathology, spinal cord blood supply, aortic procedures and
adjuncts, cord ischemia, infarction and their neurophysiological effects, confounding factors, possible
interventions, and the impact on outcome.
60.2. Aortic pathology
60.2.1. Descending aneurysms
Aneurysms of the descending aorta consist of a saccular
or fusiform progressive dilatation usually due to atherosclerosis that destroys aortic wall elastin and collagen.
Other causes include connective tissue or inflammatory
disorders, chronic dissection, and trauma (Isselbacher,
2005). Thoracic aneurysms are localized above the diaphragm, while thoracoabdominal aneurysms extend
below it and are classified, according to Crawford et al.
(1986; Fig. 1). Type I involves most of the descending
thoracic and upper abdominal aorta. Type II extends farther to most of the abdominal aorta. Type III involves the
distal thoracic aorta and varying segments of the
*
Correspondence to: David B. MacDonald, M.D., FRCP

(C), Section of Neurophysiology, Department of Neurosciences, King Faisal Specialist Hospital and Research
Center, MBC 76, PO Box 3354, Riyadh 11211, Saudi Arabia.
Tel.: 966-1-464-7272, ext. 32772; fax: 966-1-442-4763.
E-mail: dbmacdon@yahoo.com (D.B. MacDonald).
abdominal aorta. Type IV involves most of the abdominal aorta. Aneurysms may be found incidentally on a
chest X-ray or computed tomography (CT) scan performed for evaluation of other conditions or during routine examination, or present with chest or back pain,
rupture or other complications. The risk of fatal rupture
increases after a 56 cm diameter, which mainly
determines the threshold for life-saving intervention.
60.2.2. Coarctation
Aortic coarctation consists of a congenital narrowing,
causing a pressure gradient, usually near the ductus
arteriosus. There may also be cardiac anomalies
(Serfontein and Kron, 2002; Rao, 2005). The blockage can increase blood pressure in the arms and head,
reduce pressure in the legs, and seriously strain the
heart. The upper body hypertension and delayed, or
absent, femoral pulses can reveal the diagnosis in
infants, children, or adults. The hypertension may
also present with heart failure, aortic dissection,
endocarditis, or stroke. These complications reduce
life expectancy, and repair improves longevity.
60.3. Spinal cord arterial anatomy and
blood flow
60.3.1. Normal arterial anatomy
There are several excellent reviews of spinal cord
arterial supply (Mawad et al., 1990; Cheshire et al.,
1996; Connolly, 1998; Sloan, 2004). To summarize,
the anterior spinal artery runs down the anterior sulcus
and the two posterior spinal arteries run down the left
and right dorsal surface. A pial plexus of anastomoses
between these longitudinal vessels supplies the outer
rim of the cord. Penetrating branches of the posterior
spinal arteries supply the dorsal columns and dorsal
816
D.B. MACDONALD AND C.C.J. DONG
II
III
IV
Fig. 1. The Crawford classification of thoracoabdominal aneurysms, from Jacobs and Mess (2003) with permission from
Elsevier.
horns. The anterior spinal artery supplies the ventral

two-thirds of the cord through alternating left, and right,
sulcocommissural branches that perfuse the anterior
and intermediate gray, ventral portions of the dorsal
horns, and inner white matter including the corticospinal tracts. Their density is greatest in the lumbosacral
and cervical enlargements, corresponding to gray
matter volume.
Several radicular arteries feed the longitudinal
arteries along their course. The subclavian arteries provide cervical radicular supply through branches of
their thyrocervical and costocervical trunks, as well
as the vertebral arteries that also form the superior origin of the longitudinal arteries at the foramen magnum.
Thoracolumbar radicular supply comes from the
paired intercostal and lumbar segmental arteries of
the aorta. Branches of the internal iliac arteries in the
pelvis complete the inferior radicular supply; they also
supply the lumbosacral roots and plexi.
Most of the 31 pairs of radicular arteries patent in

the human embryo involute to leave a few variable
feeding vessels, making it uncertain which segmental
arteries are critical for individuals. Only one or two
small radicular arteries arising from intercostal
arteries at about T7 feed the mid-thoracic cord that
is therefore considered an anatomical watershed
zone. The large radicular artery of Adamkiewicz is
the dominant supply of the thoracolumbar cord and
usually arises between T8 and L1 on the left, but
may originate from T5 to L4 on either side. A higher
origin may increase the importance of inferior radicular supply to the lumbosacral cord.
60.3.2. Pathologic alterations
Aortic pathology compounds the uncertainty about
critical segmental vessels. Atherosclerosis occludes
many intercostal and lumbar arteries so that Jacobs
VASCULAR SURGERY
817
et al. (2002a) found a mean of only five patent

arteries between T5 and L5. They suggested that a
compensatory collateral network develops with
increasing reliance on lumbar and iliac supply and
supported the idea of Griepp et al. (1996) that a dominant artery of Adamkiewicz may no longer even
exist. Thus, one or more other segmental arteries
may assume critical importance, or there may be no
critical supply arising from the aneurysm.
With coarctation, a variably extensive collateral system involving numerous thoracic arteries develops to
perfuse the distal aorta and intercostal flow may even
be reversed (Serfontein and Kron, 2002; Rao, 2005).
above 60 mm Hg to perfuse distal radicular arteries.

CSF drainage increases cord perfusion pressure and
moderate systemic hypothermia (30 C), epidural
spinal cord cooling, or even deep hypothermia
increases ischemic tolerance. Pharmacologic agents
targeting ischemia biochemistry are experimental.
Preoperative identification of the artery of Adamkiewicz for reimplantation is controversial. Alternatively, peri-diaphragmatic segmental arteries, or none,
may be reattached; most patients tolerate the sacrifice
of all patent segmental arteries within the aneurysm,
but others do not, particularly when 10 or more are sacrificed (Griepp et al., 1996).
60.3.3. Spinal cord blood flow
60.4.3. Open coarctation repair
Marcus et al. (1977) demonstrated that blood flow to

spinal cord gray matter is about four times greater
than white matter. In addition, the lumbar and cervical enlargements have about 40% higher flow than
the mid-thoracic cord, where gray matter volume is
less. Furthermore, autoregulation maintains constant
blood flow within a range of blood pressure, while
hypoxia and hypercapnia increase, and hypocapnia
decreases flow. Finally, nerve stimulation selectively
increases gray matter blood flow in the activated cord
segment. Thus, spinal cord blood flow is distributed
and regulated according to metabolic need.
Coarctation surgery involves aortic cross-clamping

above and below the stenotic segment to repair it
using a variety of techniques (Connolly, 1998;
Massey and Shore, 2004). The collateral arterial system
generally protects against cord ischemia, but adjuncts
are used when distal flow is judged inadequate.
60.4. Aortic procedures and adjuncts
60.4.4. Endovascular procedures

Endovascular stent grafting of aneurysms and balloon
angioplasty with or without stenting for coarctation is
minimally invasive alternatives for selected patients
(Cheung et al., 2005; Isselbacher, 2005; Rao, 2005).
There is still a risk of cord infarction due to temporary
or permanent interference with radicular blood flow.
60.4.1. Open aneurysm repair

In a simple clamp, cut and sew aneurysm repair,
after clamping the proximal aorta, the aneurysm is
incised longitudinally and replaced by a synthetic
tubular graft to which visceral and often some segmental arteries are reattached before clamp release (Crawford et al., 1986). Cord infarction may occur because
distal radicular arteries are temporarily unperfused or
sacrificed and because clamp-induced proximal
hypertension raises cerebrospinal fluid (CSF) pressure, further compromising cord perfusion pressure.
60.4.2. Adjuncts
Several adjuncts aim to reduce the chance of infarction
(Connolly, 1998; Webb and Williams, 1999; Wan
et al., 2001). Sequential segmental aortic clamping
with pump-controlled atriofemoral bypass avoids
proximal hypertension and keeps distal aortic pressure
60.5. Spinal cord ischemia and infarction

Because of their higher metabolic rate and blood flow
needs, spinal neurons are more rapidly disabled by
ischemia and more susceptible to infarction than white
matter (Mathe et al., 1998). Anterior horn and intermediate gray neurons seem more vulnerable than dorsal
horn neurons do (Gonzalez-Fajardo et al., 1998; Mathe
et al., 1998; Kakinohana et al., 2005). Thus, infarction
is often a central cord lesion of the anterior gray so that
magnetic resonance imaging (MRI) scans typically
show owls eye abnormality limited to the anterior
horns in axial images (Mawad et al., 1990; Masson
et al., 2004). More severe infarctions and MRI
abnormalities extend to the dorsal horns, then inner
white matter and eventually the transverse extent of
the cord (Mawad et al., 1990).
Alpha motor neuron infarction causes segmental
flaccid weakness and subsequent denervation, which
818
may be the only deficit; sphincter paralysis occurs

with sacral gray involvement (Cheshire et al., 1996;
Mathe et al., 1998; Servais et al., 2001; Gallitto
et al., 2005). When some alpha motor neurons survive, destruction of inhibitory interneurons may
cause spasticity that does not necessarily indicate
tract damage (Mathe et al., 1998; Kakinohana et al.,
2005). Destruction of second-order sensory neurons
causes segmental pain and temperature loss (Mathe
et al., 1998). When white matter damage also occurs,
there are long tract signs below the infarction, including dorsal column deficits with extensive lesions.
Rarely, the lumbosacral roots or plexi may be
infarcted with internal iliac occlusion (Connolly,
1998).
Infarctions during aortic surgery can be likened to
cardiovascular collapse, after which lumbosacral gray
matter susceptibility is greatest, followed by cervical,
and finally thoracic involvement (Duggal and Lach,
2002). Isolated mid-thoracic infarction is rare, indicating that metabolic rate is more important than arterial
density. Cervical perfusion is preserved during aortic
surgery so that most infarctions involve the thoracolumbar and sacral cord. The multi-segmental lesions
commonly produce pencil like longitudinal abnormality in sagittal MRI scans (Mawad et al., 1990;
Masson et al., 2004).
Infarction likelihood increases with ischemia
degree, duration, and concurrent metabolic rate. Thus,
paraplegia risk rises with aortic clamp time, particularly after 30 min. However, infarction may begin
within 10 min of marked ischemia (Duggal and Lach,
2002; Kakinohana et al., 2005). Hypothermia prolongs
tolerance by reducing metabolic rate. Cellular biochemical reactions following reperfusion may cause
further neuronal damage, known as reperfusion injury
(Wan et al., 2001).
60.6. Neurophysiological effects of spinal cord
ischemia and infarction
Moderate ischemia causes functional failure of viable
neurons, while severe ischemia also initiates the
process of infarction. Evoked potential loss cannot
discriminate between the two. With reperfusion, neuronal recovery follows moderate ischemia, whereas
infarction follows severe ischemia depending on
duration, degree, and temperature (Sloan, 2004).
Evoked potentials from undamaged structures
recover, while those from damaged structures show
no or incomplete recovery.
60.6.1. Spinal cord white matter potentials

60.6.1.1. Somatosensory evoked potentials
Averaged scalp somatosensory evoked potentials
(SEPs) are conducted to the brain primarily through
the dorsal columns. Consequently, they are usually preserved with established spinal cord infarction, though
abnormalities may be seen with extensive lesions
involving dorsal white matter (Cheshire et al., 1996;
Servais et al., 2001; Kalita et al., 2003; Gallitto et al.,
2005).
Scalp SEPs have been studied in animal models of
acute cord ischemia (Grubbs et al., 1988; Hamaya,
1993) and in human aortic surgery (Cunningham
et al., 1982, 1987; Crawford et al., 1988; Schepens
et al., 1994; Galla et al., 1999; Guerit et al., 1999; Meylaerts et al., 1999a; Dong et al., 2002; MacDonald and
Janusz, 2002; Weigang et al., 2005b). It has been found
that though ischemia may cause attenuation, there is a
delay of 334 min, and gray matter infarction can
occur without attenuation. With reperfusion, persistent
SEP loss predicts infarction, but it is rare. More often,
reappearance indicates resumption of dorsal column
conduction but does not exclude gray matter infarction, particularly after 30 min of attenuation.
Averaged spinal epidural SEPs are ascending
dorsal column volleys recorded with invasive electrodes and have similar fundamental properties in
animal models of acute cord ischemia (Elmore et al.,
1991; Kai et al., 1994; Gonzalez-Fajardo et al., 1998).
They have not been studied in human aortic surgery.
60.6.1.2. Evoked spinal cord potentials
Averaged evoked spinal cord potentials (ESCPs) are
mixed sensory and motor tract action potential volleys
elicited by stimulating one end of the cord and recording at the other using invasive electrodes. They have
been studied in animal models of cord ischemia
(Machida et al., 1990; Hamaya, 1993; Kai et al.,
1994; Gonzalez-Fajardo et al., 1998) and in human
aortic surgery (Ihaya et al., 1990; Okamoto et al.,
1992; Stuhmeier et al., 1993; Grabitz et al., 1996;
Dudra et al., 1997; Matsui et al., 1997; Sueda et al.,
2000). It has been found that though ischemia may
cause attenuation, there is a delay of up to 15 or more
minutes, and gray matter infarction can occur without attenuation. With reperfusion, persistent ESCP
depression predicts infarction, but it is rare. More
often, reappearance indicates resumption of tract conduction, but does not exclude gray matter infarction
especially after attenuation exceeding 40 min.
VASCULAR SURGERY
60.6.1.3. Spinally elicited peripheral nerve potentials

Averaged peripheral nerve potentials following invasive rostral cord stimulation predominantly reflect
antidromic conduction through the dorsal columns
(Minahan et al., 2001). Consequently, their behavior
should resemble scalp and epidural SEPs. Kai et al.
(1994) found a 12.5% failure to predict a positive
wake-up test during acute cord ischemia in dogs,
but these potentials have not been evaluated in
human aortic surgery.
60.6.1.4. Epidural motor evoked potentials
Averaged spinal epidural motor evoked potentials
(MEPs) are corticospinal tract action potential
volleys usually evoked by single-pulse transcranial
electric stimulation (TES) (see Deletis and Salas
Chapter 16, this volume). Their size is proportional
to the number of conducting axons at the recorded
level and thus becomes too small to reliably record
at the threatened lumbosacral cord where the corticospinal tracts end.
These potentials have been studied in animal
models of cord ischemia (Kraus et al., 1990; Elmore
et al., 1991; Reuter et al., 1992; Shokoku et al.,
1993; De Haan et al., 1996), but in only a few human
aortic surgeries (Sueda et al., 2000; MacDonald and
Janusz, 2002). It has been found that though ischemia may cause attenuation, there is a delay of 3
22 min and infarction can occur without attenuation
when ischemia occurs below the recorded level or
spares the corticospinal tracts. With reperfusion, persistent epidural MEP loss predicts infarction but
reappearance indicating resumption of corticospinal
tract conduction does not exclude gray matter
infarction.
60.6.2. Spinal cord gray matter potentials
60.6.2.1. Tibial lumbar somatosensory evoked
potentials
Because the averaged lumbar potential (LP) at T12
following tibial nerve stimulation is generated by
lumbosacral dorsal horn neurons, it might rapidly
attenuate with ischemia. Indeed, Guerit et al.
(1997) described earlier LP than scalp SEP deterioration during two coarctation repairs and noted that
the LP was easily recorded in young children. However, significant LP alterations had delays of 5 and
15 min following aortic cross-clamping in these
two cases. In addition, 13 observations of simultaneous loss (mean delay 7.6 min) and 12 observations
819
of earlier scalp SEP attenuation (mean delay

7.4 min) were made during aneurysm surgeries
(Guerit et al., 1996, 1999). Interestingly, Mizrahi
and Crawford (1984), Fava et al. (1988) and
Faberowski et al. (1999) defined spinal cord ischemia
by scalp SEP loss with LP preservation. MacDonald
and Janusz (2002) included the LP in 11 adult aneurysm surgeries and found transient loss in one, but
abandoned its use due to low signal-to-noise ratio.
60.6.2.2. Segmental motor reflexes
Unaveraged segmental reflexes mediated through
alpha motor neurons, such as the H-reflexes have
been evaluated in a cat model of spinal cord ischemia
(Kolenda et al., 1997). Only 1 min of ischemia
caused disappearance and the likelihood of incomplete recovery increased linearly with ischemia duration above 3 min. These potentials have not been
applied to human aortic surgery.
60.6.2.3. Transcranial electric peripheral nerve
potentials
Averaged peripheral nerve potentials following TES
are mediated through alpha motor neurons activated
by corticospinal volleys. Hamaya (1993) found that
these potentials disappeared and predicted anterior
horn cell infarction at a distal aortic pressure of
45 mm Hg in dogs; SEPs and ESCPs were unaffected. Kraus et al. (1990) found attenuation within
4 min during aortic occlusion in dogs. Reuter et al.
(1992) found loss within 1 min that predicted gray
matter infarction in dogs. These potentials are
untested in human aortic surgery.
60.6.2.4. Spinally elicited muscle responses
Unaveraged muscle responses to invasive spinal cord
stimulation are mediated through alpha motor neurons
theoretically activated by both motor tract and antidromic Ia afferent volleys. These potentials have been
studied in animal models of cord ischemia (Machida
et al., 1990; Svensson et al. 1991; Gonzalez-Fajardo
et al., 1998), but in only a few human aortic surgeries
(Svensson et al., 1991). It has been found that loss
occurs within 510 min of acute ischemia and earlier
than epidural SEPs or ESCPs. High-thoracic stimulation appears more sensitive than lumbar stimulation
does, possibly due to lumbosacral root activation.
Persistent loss predicts anterior horn infarction and
reappearance within 30 min may predict recovery
from ischemia.
820
60.6.2.5. Transcranial magnetic muscle motor

evoked potentials
Unaveraged transcranial magnetic stimulation (TMS)
muscle MEPs are mediated through alpha motor neurons activated by corticospinal volleys. They are
absent, small, or delayed in established spinal cord
infarction (Berlit et al., 1992; Kalita et al., 2003).
Qayumi et al. (1997) studied their disappearance
during spinal cord ischemia in pigs under alphachloralose anesthesia; amplitude was considered too
variable to evaluate. Aortic clamping abolished
MEPs at a mean of 2 min. Occlusion for 25 min
caused no deficit and MEPs reappeared at a mean
of 29 min. Occlusion for 30 min caused paraplegia
and MEP reappearance was delayed to a mean of
40 min. Without aortic clamping, ligation of all intercostal arteries caused no MEP loss or deficit, whereas
ligation of all intercostal and lumbar arteries caused
persistent MEP loss and paraplegia. There are no
studies in human aortic surgery with this technique.
L TA
L 1stDI
2.5 mV
60.6.2.6. Transcranial electric muscle motor evoked

potentials
Pulse train TES muscle MEPs are also mediated
through alpha motor neurons activated by corticospinal
volleys, but are more practical and consistently obtained
under appropriate anesthesia (see Mendiratta and
Emersons Chapter 18, this volume). Consequently,
they have received extensive study in animal models
of acute cord ischemia (De Haan et al., 1996; Meylaerts
et al., 2000; Lips et al., 2002a,b; Murakami et al., 2004)
and in human aortic surgery (De Haan et al., 1997;
Meylaerts et al., 1999a; Van Dongen et al., 2001; Dong
et al., 2002; Jacobs et al., 2002b; MacDonald and
Janusz, 2002; Jacobs and Mess, 2003; Weigang et al.,
2005b).
It has been found that although spontaneous variability may interfere with measurement, leg muscle
MEPs disappear or markedly attenuate with acute
ischemia in about 2 min (Fig. 2). Occasional longer
delays may reflect corticospinal tract failure in an
R TA
1 mV
R 1stDI
0.5 mV
5 mV
10:33
10:36
10:38
10:40
10:42
10:46
10:47
10:49
10:51
100 ms
Fig. 2. Transcranial electric muscle motor evoked potential (MEP) response to acute cord ischemia during thoracoabdominal aneurysm surgery. TA, tibialis anterior; 1stDI, first dorsal interosseous. The aorta was clamped at 10:33. Leg MEPs
disappeared or markedly attenuated within 3 min. Unaffected hand MEPs excluded technical or systemic causes and unaffected tibial peripheral and cortical SEPs (not shown) excluded leg or cerebral ischemia. MEP restoration followed clamp
release at 10:38. Modified from MacDonald and Janusz (2002), with permission from Lippincott, Williams and Wilkins.
VASCULAR SURGERY
ischemic segment above an unaffected lumbosacral

cord. Decrements frequently occur without scalp
SEP attenuation, or lead by 234 min. Stable MEPs
predict the absence of infarction, and persistent loss
predicts paraplegia. Restoration correlates with motor
integrity, but reappearance after a 40-min absence or
incomplete amplitude restoration may indicate a risk
of partial motor deficit. In addition, Kakinohana et al.
(2005) found full amplitude recovery and spared
alpha motor neurons but spastic paraparesis due to
interneuron infarction in rats. They also described a
patient with 1-mV postoperative TES leg muscle
MEPs despite spastic paraparesis following unmonitored aortic surgery.
60.6.3. Evoked potential summary
Whether sensory, motor or mixed, potentials mediated
through spinal cord white matter conduction respond
inconsistently and slowly to acute cord ischemia.
Their recovery indicates resumed conduction in
undamaged tracts, but does not exclude gray matter
infarction. Potentials mediated through alpha motor
neurons consistently show rapid loss or marked attenuation and noninvasive pulse train TES muscle MEPs
are currently the method of choice. Reappearance of
these potentials indicates surviving alpha motor neurons, but may not exclude damage to a subpopulation
or to interneurons, depending on attenuation duration
and recovery amplitude, and possible spontaneous
variability.
60.7. Confounding factors
Other factors that attenuate evoked potentials are frequent during aortic surgeries. These likely account
for reported false-positive SEPs (Crawford et al.,
1988; Meylaerts et al., 1999a) and could cause
MEP monitoring errors.
60.7.1. Anesthesia
Effective anesthetics for monitoring TES muscle
MEPs during aortic surgery include propofol/opioids
(Dong et al., 2002; MacDonald and Janusz, 2002),
ketamine/sufentanil (Jacobs et al., 2002b), diazepam/propofol/fentanyl/nitrous oxide (Van Dongen
et al., 2001), and benzodiazepine/fentanyl (Weigang
et al., 2005b). Because increments or boluses may
attenuate muscle MEPs and cortical SEPs, stable
anesthesia is desirable, but not always possible.
821
60.7.2. Neuromuscular blockade

Tightly controlled partial neuromuscular blockade is
compatible with muscle MEP monitoring (Van
Dongen et al., 2001; Jacobs et al., 2002b). Nevertheless, blockade potentiation may arise from the
administration of magnesium and other substances
(De Haan and Kalkman, 2001). Consequently, relaxants can also be omitted after intubation without adverse
effects on monitoring or surgery (Dong et al., 2002;
MacDonald and Janusz, 2002; Weigang et al., 2005b).
60.7.3. Other systemic factors
Hemodynamic values often fluctuate widely. Moderate systemic or regional cord hypothermia primarily
increases evoked potential latency and is compatible
with effective monitoring (Meylaerts et al., 1999b).
Deep hypothermia disables synaptic potential monitoring until rewarming. Scalp edema due to fluid
and electrolyte imbalance may reduce TES efficiency
and cortical SEP amplitude (MacDonald and Janusz,
2002).
60.7.4. Potential fade
Even with stable anesthesia and no scalp edema, there
is an unexplained tendency for muscle MEPs and cortical SEPs to gradually fade. Stimulus increments may
be needed to maintain MEPs and cortical SEP amplitudes may fall below arbitrary critical levels based on
earlier traces (Meylaerts et al., 1999a; MacDonald
and Janusz, 2002; Weigang et al., 2005a).
60.7.5. Cerebral ischemia
Generalized or focal cerebral ischemia due to hypotension or left subclavian artery occlusion with proximal
aortic clamping is common (Guerit et al., 1996; MacDonald and Janusz, 2002). Cortical SEPs and muscle
MEPs attenuate or disappear; subcortical SEPs also
attenuate if the brainstem is affected.
60.7.6. Limb ischemia
Frequent limb ischemia causes progressive SEP and
muscle MEP loss in the affected limb(s) over 20
30 min (Guerit et al., 1999; De Haan and Kalkman,
2001; MacDonald and Janusz, 2002). Consequently,
atriofemoral bypass is essential for monitoring leg
potentials. However, the occlusive femoral bypass
822
cannula renders one leg ischemic. Cannulating a femoral side arm graft, and other means of perfusing the
leg prevent this, but ischemia may still occur during
femoral repair at closure. Bypass malfunction, or discontinuation to complete the distal aortic anastomosis,
may cause bilateral leg ischemia. Finally, left subclavian artery occlusion produces left arm ischemia.
60.7.7. Technical obstacles
Dissimilar metal artifact from aortic clamps contacting
the thoracic retractor may disable monitoring and
interposing cloth between metal surfaces prevents
this (MacDonald and Janusz, 2002). Anticoagulation
potentiates the risk of epidural hemorrhage from invasive spinal electrodes. Other standard operating room
obstacles are compounded by the amount of equipment and personnel, but are surmountable.
60.7.8. Differentiating confounding factors
Monitoring the upper limbs controls for technical,
anesthetic, neuromuscular, and other systemic factors
including potential fade (Shahin et al., 1996; De Haan
and Kalkman, 2001; MacDonald and Janusz, 2002).
Approximately parallel four-limb changes point to
systemic causes, whereas focal decrement indicates
localized compromise (MacDonald and Janusz,
2002). Four-limb cortical SEPs identify cerebral ischemia, and peripheral SEPs identify limb ischemia
(Guerit et al., 1999; MacDonald and Janusz, 2002).
Subcortical and lumbar SEPs may provide additional
information (Guerit et al., 1999), but they substantially
delay feedback due to low signal-to-noise ratio (MacDonald and Janusz, 2002; MacDonald et al., 2005).
60.8. Impact on outcome
60.8.1. Open aneurysm repair
The risk of spinal cord infarction in unmonitored thoracic aneurysm surgery is 320% (Connolly, 1998).
Crawford et al. (1986) found the risk for unmonitored
types I, II, III, and IV simple thoracoabdominal aneurysm repair to be 10%, 28%, 3%, and 2%, respectively. Other reported rates vary with proportions of
aneurysm type and adjunct use, but the risk is consistently highest for type I and particularly type II
aneurysms. Cord infarction rates remain at 516%
with adjuncts (Webb and Williams, 1999; Jacobs
et al., 2002b). The lack of randomized trials makes

it difficult to formally determine if monitoring
reduces this risk. However, there is enough data to
estimate the impact of SEP, ESCP, and TES muscle
MEP monitoring.
60.8.1.1. SEP monitoring
While it is likely that SEP monitoring has benefited
selected patients, the cumulative infarction rate
among over 700 reported surgeries is 10.0%
(Table 1). Series with lower rates have had small proportions of high-risk aneurysms. Numerous falsepositive results have been reported.
It must be pointed out that many studies have used
lower-extremity scalp SEPs alone, which limits specificity. Guerit et al. (1999) proposed that multilevel
popliteal fossa, LP, subcortical, and cortical tibial
SEP recordings enhance sensitivity and specificity.
In 63 mixed-etiology aortic surgeries, they observed
only three confirmed cord infarctions (4.8%), two
of which were delayed postoperative infarctions.
However, three other patients had lower motor neuron
deficits attributed to neuropathy because their spinal
cord MRIs were normal. Since MRIs can be negative
(Mawad et al., 1990; Cheshire et al., 1996; Masson
et al., 2004), these undetected motor deficits could
have been due to alpha motor neuron infarction.
Including them would double the deficit rate and aortic
pathology extended to the abdomen in only 25% of
patients. Thus, multilevel SEP methodology has not
been shown to substantially impact overall outcome.
60.8.1.2. Evoked spinal cord potential monitoring
Similarly, ESCP monitoring has not reduced overall
risk (Table 1). The cumulative paraplegia rate of
14.0% in nearly 300 reported aneurysm surgeries
makes the possibility of invasive electrode complications difficult to justify.
60.8.1.3. Transcranial muscle MEP monitoring
Turning to TES muscle MEP series, Weigang et al.
(2005b) reported their initial experience in 19 aneurysm surgeries (42% type II). Five patients had stable
leg MEPs and no deficit. Although two had persistent
MEP loss and paraplegia (10.5%), nine had MEP
reappearance following intervention and no deficit.
They noted that the rapidity of MEP loss provided a
better guide to intervention than tibial SEPs did.
Dong et al. (2002) and MacDonald and Janusz
(2002) reported on 25 thoracic and 31 thoracoabdominal aneurysm surgeries. No patient with consistently
VASCULAR SURGERY
823
Table 1
Reported spinal cord infarction rates in monitored aortic aneurysm surgery
Modality
First author
Year
SCI
SEPsa
Mizrahi
Cunningham
Crawford
De Mol
Schepens
Griepp
Shahin
Galla
Guerit
Wada
Cumulative
1984
1987
1988
1990
1994
1996
1996
1999
1999
2001
13
33
198
14
43
95
40
149
63
82
730
1
5
35
3
7
2
2
12
3
3
73
7.7
15.2
17.7
21.4
16.3
2.1
5.0
8.1
4.8
3.7
10.0
ESCPs
Okamoto
Grabitz
Dudra
Matsui
Cumulative
1992
1996
1997
1997
21
167
15
83
286
1
23
5
11
40
4.8
13.8
33.3
13.3
14.0
TES MEPsb
Van Dongen
Dong
Jacobs
Weigang
Cumulative
2001
2002
2003
2005
118
56
260
19
453
5
3
6
2
16
4.2
5.4
2.3
10.5
3.5
SEP monitoring series with low infarction rates had small proportions of high-risk aneurysms.
All MEP monitoring series had large proportions of high-risk aneurysms.
SEPs, somatosensory evoked potentials; ESCPs, evoked spinal cord potentials; TES MEPs, transcranial electric muscle motor evoked
potentials; SCI, spinal cord infarction.
b
present leg MEPs had paraplegia. Of 16 patients with

intraoperative MEP loss, only four had delayed tibial
SEP reduction. Reappearance followed intervention
in 13, none of whom had paraplegia. Three with persistent MEP loss, but SEP recovery, had spinal cord
infarction (5.4%). They concluded that leg MEPs
best-guided intervention and that four-limb recordings including peripheral/cortical SEPs provided
rapid differential information enhancing interpretation. Subcortical and lumbar SEPs in a subgroup of
15 patients were too slowly resolved to be effective.
Van Dongen et al. (2001) reported on 118 aneurysm surgeries (67% type II). Forty-two patients
had >50% leg MEP attenuation with aortic clamping
accompanied by tibial SEP reduction in only five.
Restoration without deficit typically followed intervention. Of 18 patients with closing MEP amplitudes
below 50% of baseline, four had cord infarction.
One patient with stable potentials had delayed postoperative infarction. They believed that the 4.2%
infarction rate would have been substantially higher

without MEP monitoring.
Jacobs et al. (2002b) and Jacobs and Mess (2003)
reported a remarkable series of 80 type I, 111 type II,
41 type III, and 28 type IV thoracoabdominal aneurysm surgeries. About 40% of the 260 patients had
>75% MEP reduction or disappearance that usually
occurred within 13 min of cross-clamping a critical
aortic segment. Almost all had restoration following
intervention. Two patients with about 20% of baseline amplitude leg MEPs unilaterally at closure had
temporary leg monoparesis; only one had persistent
bilateral leg MEP loss and immediate paraplegia.
Three patients had postoperative day 3, 6, or 9 paraplegia. The spinal cord infarction rate was only 2.3%
(1.2% intraoperative).
MEP-guided interventions in this series included
raising distal aortic pressure even beyond 60 mm Hg,
selective segmental artery reattachment, increased
CSF drainage, clamp repositioning, and selective
824
hypothermia. Jacobs et al. (2002b) also performed

immediate aortic endarterectomy when MEP loss
accompanied clamping of an aortic segment with no
visibly patent segmental arteries. When back bleeding vessels were then identified, MEP restoration followed their reattachment, indicating that arteries with
occluded ostia might still participate in collateral
cord perfusion.
The cumulative 3.5% rate of spinal cord infarction
in over 450 reported surgeries with a large proportion
of high-risk aneurysms strongly suggests that leg
muscle MEP monitoring can favorably impact outcome (Table 1). Concurrent SEPs provide differential
information and congruent ischemic decrements
imply greater severity.
It has also been found that scalp SEPs are difficult

or impossible to monitor in young children and
infants. However, TES muscle MEPs are readily monitorable in children as young as 11 months (unpublished personal observations) and might therefore be
especially valuable for young coarctation patients.
60.8.3. Endovascular procedures
60.8.1.4. Delayed paraplegia

Delayed paraplegia can follow hypotension with marginal cord perfusion or thrombosis of reattached segmental arteries. Raising blood pressure and draining
CSF may reverse the process (Cheung et al., 2002).
Intraoperative signs of ischemia should increase vigilance, and selective postoperative monitoring until
the patient is alert enough for clinical assessment
may be worthwhile (Guerit and Dion, 2002), but has
not been shown to improve overall outcome. Postoperative SEPs are feasible, but TES may be too painful;
TMS, MEPs, and H-reflexes have not been explored.
Since infarction can occur after several days, there
will remain an irreducible risk of delayed paraplegia.
Spinal cord infarction complicates 412% of endovascular aortic aneurysm treatments (Cheung et al.,
2005), but there is little experience with monitoring.
Bafort et al. (2002) reported the use of unchanged tibial
SEPs during temporary occlusion to predict safe stent
deployment in one patient. Cheung et al. (2005) monitored SEPs in 15 patients. One patient had persistent tibial scalp SEP loss with paraplegia and another had
reappearance following intervention and no deficit.
Koizumi et al. (2004) reported ESCP-guided retrievable
stent-grafting of 87 thoracic aneurysms; three patients
(3.4%) had postoperative paraparesis. Weigang et al.
(2005b) monitored TES muscle MEPs and scalp SEPs
during stent-grafting in nine patients. One had MEP loss
restored following intervention and no deficit; SEPs
were unaltered. The others had no alterations or deficits.
These preliminary results suggest that monitoring may
be valuable, but the overall impact on outcome is
unknown. Although spinal cord infarction can complicate coarctation balloon angioplasty (Ussia et al.,
2001), there are no reports of monitoring.
60.8.2. Open coarctation repair
60.9. Conclusion
Cord infarction complicates only 0.31.5% of unmonitored coarctation repairs (Faberowski et al., 1999; Serfontein and Kron, 2002; Massey and Shore, 2004).
Five series totaling 169 infants, children, and adults
have evaluated SEP monitoring (Kaplan et al., 1986;
Pollock et al., 1986; Dasmahapatra et al., 1987; Guerit
et al., 1997; Faberowski et al., 1999). It has been found
that tibial SEP decrements correlate with low-distal aortic pressure and occur in 2647% of patients during aortic clamping. All reported decrements resolved after
clamp release, or following intervention (repositioning
clamps, raising blood pressure, or even abandoning surgery). One cord infarction occurred despite SEP recovery after 14 min. Thus, the cumulative 0.6% infarction
rate matches the expected risk. However, the low incidence requires many more surgeries to evaluate impact,
and the possible value of muscle MEP monitoring
remains untested.
Monitoring descending aortic procedures requires

rapid differential feedback. Spinal cord ischemia
and infarction begins in, and may be limited to, the
anterior horns while tracts are resistant. Therefore,
potentials mediated through spinal cord white matter
conduction respond inconsistently and slowly to
ischemia and have not been shown to impact overall
outcome. In contrast, potentials mediated through
alpha motor neurons show rapid attenuation that is
reversible with timely reperfusion. Consequently,
there is now strong evidence that TES leg muscle
MEP monitoring positively impacts descending aortic aneurysm surgery outcome when used to guide
aggressive intervention. Concurrent upper limb and
SEP monitoring provides useful differential information. The impact of monitoring on coarctation
surgery, endovascular procedures, and delayed paraplegia remains to be determined.
VASCULAR SURGERY
References
Bafort, C, Astarci, P, Goffette, P, El Khoury, G,
Guerit, JM, De Tourtchaninoff, M and Verhelst, R
(2002) Predicting spinal cord ischemia before endovascular thoracoabdominal aneurysm repair: monitoring
somatosensory evoked potentials. J. Endovasc. Ther., 9:
289294.
Berlit, P, Klotzsch, G, Rother, J, Assmus, HP, Daffertshofer, M and Schwartz, A (1992) Spinal cord infarction:
MRI and MEP findings in three cases. J. Spinal Disord.,
5: 212216.
Cheshire, WP, Santos, CC, Massey, EW and Howard, JF, Jr.
(1996) Spinal cord infarction: etiology and outcome.
Neurology, 47: 321330.
Cheung, AT, Weiss, SJ, McGarvey, ML, Stecker, MM,
Hogan, MS, Escherich, A and Bavaria, JE (2002) Interventions for reversing delayed-onset postoperative paraplegia
after thoracic aortic reconstruction. Ann. Thorac. Surg., 74:
413419.
Cheung, AT, Pochettino, A, McGarvey, ML, Appoo, JJ,
Fairman, RM, Carpenter, JP, Moser, WG, Woo, EY and
Bavaria, JE (2005) Strategies to manage paraplegia risk
after endovascular stent repair of descending thoracic
aortic aneurysms. Ann. Thorac. Surg., 80: 12801288.
Connolly, JE (1998) Hume Memorial lecture. Prevention of
spinal cord complications in aortic surgery. Am. J. Surg.,
176: 92101.
Crawford, ES, Crawford, JL, Safi, HJ, Coselli, JS, Hess,
KR, Brooks, B, Norton, HJ and Glaeser, DH (1986)
Thoracoabdominal aortic aneurysms: preoperative and
intraoperative factors determining immediate and longterm results of operations in 605 patients. J. Vasc. Surg.,
3: 389404.
Crawford, ES, Mizrahi, EM, Hess, KR, Coselli, JS, Safi, HJ
and Patel, VM (1988) The impact of distal aortic perfusion and somatosensory evoked potential monitoring on
prevention of paraplegia after aortic aneurysm operation.
Cunningham, JN, Jr., Laschinger, JC, Merkin, HA, Nathan,
IM, Colvin, S, Ransohoff, J and Spencer, FC (1982) Measurement of spinal cord ischemia during operations upon
the thoracic aorta: initial clinical experience. Ann. Surg.,
196: 285296.
Cunningham, JN, Jr., Laschinger, JC and Spencer, FC (1987)
Monitoring of somatosensory evoked potentials during
surgical procedures on the thoracoabdominal aorta. IV.
Clinical observations and results. J. Thorac. Cardiovasc.
Surg., 94: 275285.
Dasmahapatra, HK, Coles, JG, Taylor, MJ, Cass, D, Couper,
G, Adler, S, Burrows, F, Sherret, H, Trusler, GA and Williams, WG (1987) Identification of risk factors for spinal
cord ischemia by the use of monitoring of somatosensory
evoked potentials during coarctation repair. Circulation,
76(3 Pt 2): III14III18.
825
De Haan, P and Kalkman, CJ (2001) Spinal cord monitoring: somatosensory- and motor-evoked potentials.
Anesthesiol. Clin. North America, 19: 923945.
De Haan, P, Kalkman, CJ, Ubags, LH, Jacobs, MJ and
Drummond, JC (1996) A comparison of the sensitivity
of epidural and myogenic transcranial motor-evoked
responses in the detection of acute spinal cord ischemia
in the rabbit. Anesth. Analg., 83: 10221027.
De Haan, P, Kalkman, CJ, De Mol, BA, Ubags, LH, Veldman, DJ and Jacobs, MJ (1997) Efficacy of transcranial
motor-evoked myogenic potentials to detect spinal cord
ischemia during operations for thoracoabdominal aneurysms. J. Thorac. Cardiovasc. Surg., 113: 87100.
Dong, CC, MacDonald, DB and Janusz, MT (2002) Intraoperative spinal cord monitoring during descending thoracic and thoracoabdominal aneurysm surgery. Ann.
Dudra, J, Shiiya, N, Matsui, Y, Sakuma, M, Ishii, K,
Asada, M and Yasuda, K (1997) Operative results of
thoracoabdominal repair for chronic type B aortic dissection. J. Cardiovasc. Surg. (Torino), 38: 147151.
Duggal, N and Lach, B (2002) Selective vulnerability of
the lumbosacral spinal cord after cardiac arrest and
hypotension. Stroke, 33: 116121.
Elmore, JR, Gloviczki, P, Harper, CM, Pairolero, PC,
Murray, MJ, Bourchier, RG, Bower, TC and Daube, JR
(1991) Failure of motor evoked potentials to predict neurologic outcome in experimental thoracic aortic occlusion. J. Vasc. Surg., 14: 131139.
Faberowski, LW, Black, S, Trankina, MF, Pollard, RJ, Clark,
RK and Mahla, ME (1999) Somatosensory-evoked
potentials during aortic coarctation repair. J. Cardiothorac. Vasc. Anesth., 13: 538543.
Fava, E, Bortolani, EM, Ducati, A and Ruberti, U (1988)
Evaluation of spinal cord function by means of lower
limb somatosensory evoked potentials in reparative
aortic surgery. J. Cardiovasc. Surg. (Torino), 29:
421427.
Galla, JD, Ergin, MA, Lansman, SL, McCullough, JN,
Nguyen, KH, Spielvogel, D, Klein, JJ and Griepp, RB
(1999) Use of somatosensory evoked potentials for thoracic and thoracoabdominal aortic resections. Ann.
Thorac. Surg., 67: 19471952.
Gallitto, G, La Spina, P, Granata, F, Quartarone, A, Savica,
R and Musolino, R (2005) Spinal gray matter infarction
after aortic surgery: a case of persistent pure flaccid
paraplegia. Cerebrovasc. Dis., 19: 345347.
Gonzalez-Fajardo, JA, Toledano, M, Alvarez, T and
Vaquero, C (1998) Monitoring of evoked potentials during spinal cord ischaemia: experimental evaluation in a
rabbit model. Eur. J. Vasc. Endovasc. Surg., 16:
320328.
Grabitz, K, Sandmann, W, Stuhmeier, K, Mainzer, B, Godehardt, E, Ohle, B and Hartwich, U (1996) The risk of
ischemic spinal cord injury in patients undergoing graft
826
replacement for thoracoabdominal aortic aneurysms.
J. Vasc. Surg., 23: 230240.
Griepp, RB, Ergin, MA, Galla, JD, Lansman, S, Khan, N,
Quintana, C, McCollough, J and Bodian, C (1996)
Looking for the artery of Adamkiewicz: a quest to minimize paraplegia after operations for aneurysms of the
descending thoracic and thoracoabdominal aorta.
Grubbs, PE Jr., Marini, C, Toporoff, B, Nathan, I, Basu, S,
Acinapura, AJ and Cunningham, JN, Jr. (1988) Somatosensory evoked potentials and spinal cord perfusion
pressure are significant predictors of postoperative neurologic dysfunction. Surgery, 104: 216223.
Guerit, JM and Dion, RA (2002) State-of-the-art of neuromonitoring for prevention of immediate and delayed
paraplegia in thoracic and thoracoabdominal aorta surgery. Ann. Thorac. Surg, 74: S1867S1869.
Guerit, JM, Verhelst, R, Rubay, J, Khoury, G, Matta, A and
Dion, R (1996) Multilevel somatosensory evoked potentials (SEPs) for spinal cord monitoring in descending
thoracic and thoraco-abdominal aortic surgery. Eur. J.
Cardiothorac. Surg., 10: 93103.
Guerit, JM, Witdoeckt, C, Rubay, J, Matta, A and Dion, R
(1997) The usefulness of the spinal and subcortical
components of the posterior tibial nerve SEPs for spinal
cord monitoring during aortic coarctation repair. Electroencephalogr. Clin. Neurophysiol., 104: 115121.
Guerit, JM, Witdoeckt, C, Verhelst, R, Matta, AJ, Jacquet,
LM and Dion, RA (1999) Sensitivity, specificity, and
surgical impact of somatosensory evoked potentials in
descending aorta surgery. Ann. Thorac. Surg., 67:
19431946.
Hamaya, H (1993) Study on prevention of paraplegia during occlusion blockade of the thoracic aorta
examination of the evoked potential in the motor nerve
induced by stimulation of the motor area of cerebral
cortex. Nippon Kyobu Geka Gakkai Zasshi, 41:
13471356.
Ihaya, A, Morioka, K, Noguchi, H, Kimura, T, Nishii, H,
Hiramatsu, Y, Chiba, Y and Muraoka, R (1990) A case
report of descending thoracic aortic aneurysm associated
with anterior spinal artery syndrome despite no marked
ESP changes. Kyobu Geka, 43: 843846.
Isselbacher, EM (2005) Thoracic and abdominal aortic
aneurysms. Circulation, 111: 816828.
Jacobs, MJ and Mess, WH (2003) The role of evoked
potential monitoring in operative management of type
I and type II thoracoabdominal aortic aneurysms.
Semin. Thorac. Cardiovasc. Surg., 15: 353364.
Jacobs, MJ, De Mol, BA, Elenbaas, T, Mess, WH, Kalkman,
CJ, Schurink, GW and Mochtar, B (2002a) Spinal cord
blood supply in patients with thoracoabdominal aortic
aneurysms. J. Vasc. Surg., 35: 3037.
Jacobs, MJ, Elenbaas, TW, Schurink, GW, Mess, WH and
Mochtar, B (2002b) Assessment of spinal cord integrity

during thoracoabdominal aortic aneurysm repair. Ann.
Thorac. Surg, 74: S1864S1866.
Kai, Y, Owen, JH, Allen, BT, Dobras, M and Davis, C (1994)
Relationship between evoked potentials and clinical status in spinal cord ischemia. Spine, 19: 11621167.
Kakinohana, M, Kawabata, T, Miyata, Y and Sugahara, K
(2005) Myogenic transcranial motor evoked potentials
monitoring cannot always predict neurologic outcome
after spinal cord ischemia in rats. J. Thorac. Cardiovasc. Surg., 129: 4652.
Kalita, J, Srivastava, AK, Mittal, P, Sharma, VP and Misra,
UK (2003) Evoked potential changes in ischaemic myelopathy. Electromyogr. Clin. Neurophysiol., 43:
211215.
Kaplan, BJ, Friedman, WA, Alexander, JA and Hampson,
SR (1986) Somatosensory evoked potential monitoring
of spinal cord ischemia during aortic operations. Neurosurgery, 19(1): 8290.
Koizumi, N, Obitsu, Y, Koide, K, Sato, K, Saiki, N,
Watanabe, Y, Ichihashi, H, Yokoi, Y, Shimazaki, T,
Kawaguchi, S and Ishimaru, S (2004) Evaluation of
spinal cord ischemia in endovascular stent graft repair
and surgical operation of descending thoracic or thoracoabdominal aortic aneurysms. Kyobu Geka, 57:
262267.
Kolenda, H, Steffens, H, Gefeller, O, Hagenah, J and
Schomburg, ED (1997) Critical levels of spinal cord
blood flow and duration of ischemia for the acute recovery of segmental spinal cord responses in cats. J. Spinal.
Disord., 10: 288295.
Kraus, KH, Pope, ER, OBrien, D and Hay, BL (1990) The
effects of aortic occlusion on transcranially induced
evoked potentials in the dog. Vet. Surg., 19: 341347.
Lips, J, De Haan, P, Bouma, GJ, Jacobs, MJ and Kalkman,
CJ (2002a) Delayed detection of motor pathway dysfunction after selective reduction of thoracic spinal cord
blood flow in pigs. J. Thorac. Cardiovasc. Surg., 123:
531538.
Lips, J, De Haan, P, De Jager, SW, Vanicky, I, Jacobs, MJ
and Kalkman, CJ (2002b) The role of transcranial motor
evoked potentials in predicting neurologic and histopathologic outcome after experimental spinal cord
ischemia. Anesthesiology, 97: 183191.
MacDonald, DB and Janusz, M (2002) An approach to intraoperative monitoring of thoracoabdominal aneurysm surgery. J. Clin. Neurophysiol., 19: 4354.
MacDonald, DB, Al Zayed, Z and Stigsby, B (2005) Tibial
somatosensory evoked potential intraoperative monitoring: recommendations based on signal to noise ratio
analysis of popliteal fossa, optimized P37, standard
P37 and P31 potentials. Clin. Neurophysiol., 116:
18581869.
Machida, M, Yamada, T, Ross, M, Kimura, J and Hitchon,
P (1990) Effect of spinal cord ischemia on compound
muscle action potentials and spinal evoked potentials
VASCULAR SURGERY
following spinal cord stimulation in the dog. J. Spinal.
Disord., 3: 345352.
Marcus, ML, Heistad, DD, Ehrhardt, JC and Abboud, FM
(1977) Regulation of total and regional spinal cord
blood flow. Circ. Res., 41: 128134.
Massey, R and Shore, DF (2004) Surgery for complex
coarctation of the aorta. Int. J. Cardiol., 97(Suppl. 1):
6773.
Masson, C, Pruvo, JP, Meder, JF, Cordonnier, C, Touze, E,
De La Sayette, V, Giroud, M, Mas, JL and Leys, D
(2004) Spinal cord infarction: clinical and magnetic resonance imaging findings and short term outcome.
Mathe, JF, Richard, I, Roger, JC, Potagas, C, El Masry, WS
and Perrouin-Verbe, B (1998) Ischaemic myelopathy
following aortic surgery or traumatic laceration of the
aorta. Spinal Cord, 36: 110116.
Matsui, Y, Shiiya, N, Ishii, K, Murashita, T, Sasaki, S,
Sakuma, M and Yasuda, K (1997) The reliability of
evoked spinal cord potentials elicited by direct stimulation of the cord as a monitor of spinal cord ischemia
during temporary occlusion of the thoracic aorta. Panminerva Med., 39: 7884.
Mawad, ME, Rivera, V, Crawford, S, Ramirez, A and Breitbach, W (1990) Spinal cord ischemia after resection of
thoracoabdominal aortic aneurysms: MR findings in 24
patients. AJNR Am. J. Neuroradiol., 11: 987991.
Meylaerts, SA, Jacobs, MJ, Van Iterson, V, De Haan, P and
Kalkman, CJ (1999a) Comparison of transcranial motor
evoked potentials and somatosensory evoked potentials
Surg., 230: 742749.
Meylaerts, SA, De Haan, P, Kalkman, CJ, Lips, J, De Mol,
BA and Jacobs, MJ (1999b) The influence of regional
spinal cord hypothermia on transcranial myogenic
motor-evoked potential monitoring and the efficacy of
spinal cord ischemia detection. J. Thorac. Cardiovasc.
Surg., 118: 10381045.
Meylaerts, SA, De Haan, P, Kalkman, CJ, Jaspers, J,
Vanicky, I and Jacobs, MJ (2000) Prevention of paraplegia in pigs by selective segmental artery perfusion
during aortic cross-clamping. J. Vasc. Surg., 32:
160170.
Mizrahi, EM and Crawford, ES (1984) Somatosensory
evoked potentials during reversible spinal cord ischemia
in man. Electroencephalogr. Clin. Neurophysiol., 58:
120126.
Murakami, H, Tsukube, T, Kawanishi, Y and Okita, Y
(2004) Transcranial myogenic motor-evoked potentials
after transient spinal cord ischemia predicts neurologic
outcome in rabbits. J. Vasc. Surg., 39: 207213.
827
Okamoto, Y, Murakami, M, Nakagawa, T, Murata, A and
Moriya, H (1992) Intraoperative spinal cord monitoring
during surgery for aortic aneurysm: application of spinal cord evoked potential. Electroencephalogr. Clin.
Pollock, JC, Jamieson, MP and McWilliam, R (1986)
Somatosensory evoked potentials in the detection of
spinal cord ischemia in aortic coarctation repair. Ann.
Thorac. Surg., 41: 251254.
Qayumi, KA, Janusz, MT, Jamieson, EW, Chow, VD and
Dry, GM (1997) Transcranial magnetic stimulation:
use of motor evoked potentials in the evaluation of surgically induced spinal cord ischemia. J. Spinal Cord
Med., 20: 395401.
Rao, PS (2005) Coarctation of the aorta. Curr. Cardiol.
Rep., 7: 425434.
Reuter, DG, Tacker, WA, Jr., Badylak, SF, Voorhees, WD
3rd and Konrad, PE (1992) Correlation of motor-evoked
potential response to ischemic spinal cord damage.
Schepens, MA, Boezeman, EH, Hamerlijnck, RP, Ter
Beek, H and Vermeulen, FE (1994) Somatosensory
evoked potentials during exclusion and reperfusion of
critical aortic segments in thoracoabdominal aortic
aneurysm surgery. J. Card. Surg., 9: 692702.
Serfontein, SJ and Kron, IL (2002) Complications of coarctation repair. Semin. Thorac. Cardiovasc. Surg. Pediatr.
Card. Surg. Annu., 5: 206211.
Servais, LJ, Rivelli, SK, Dachy, BA, Christophe, CD and
Dan, B (2001) Anterior spinal artery syndrome
after aortic surgery in a child. Pediatr. Neurol., 24:
310312.
Shahin, GM, Hamerlijnck, RP, Schepens, MA, Ter Beek, HT,
Vermeulen, FE and Boezeman, EH (1996) Upper and
lower extremity somatosensory evoked potential recording during surgery for aneurysms of the descending
thoracic aorta. Eur. J. Cardiothorac. Surg., 10:
299304.
Shokoku, S, Uchida, H and Teramoto, S (1993) An experimental study on spinal cord ischemia during crossclamping of the thoracic aorta: the monitoring of spinal
cord ischemia with motor evoked potential by transcranial stimulation of the cerebral cortex in dogs. Surg.
Today, 23: 10681073.
Sloan, TB (2004) Electrophysiologic monitoring during
surgery to repair the thoracoabdominal aorta. Semin.
Cardiothorac. Vasc. Anesth., 8: 113125.
Stuhmeier, KD, Grabitz, K, Mainzer, B, Sandmann, W and
Tarnow, J (1993) Use of the electrospinogram for predicting harmful spinal cord ischemia during repair of
thoracic or thoracoabdominal aortic aneurysms. Anesthesiology, 79: 11701176.
Sueda, T, Okada, K, Watari, M, Orihashi, K, Shikata, H and
Matsuura, Y (2000) Evaluation of motor- and sensoryevoked potentials for spinal cord monitoring during
828
thoracoabdominal aortic aneurysm surgery. Jpn. J. Thorac.
Svensson, LG, Patel, V, Robinson, MF, Ueda, T, Roehm,
JO, Jr. and Crawford, ES (1991) Influence of preservation or perfusion of intraoperatively identified spinal
cord blood supply on spinal motor evoked potentials
and paraplegia after aortic surgery. J. Vasc. Surg., 13:
355365.
Ussia, GP, Marasini, M and Pongiglione, G (2001) Paraplegia following percutaneous balloon angioplasty of aortic
coarctation: a case report. Catheter. Cardiovasc. Interv.,
54: 510513.
Van Dongen, EP, Schepens, MA, Morshuis, WJ, Ter Beek,
HT, Aarts, LP, De Boer, A and Boezeman, EH (2001)
Thoracic and thoracoabdominal aortic aneurysm repair:
use of evoked potential monitoring in 118 patients.
J. Vasc. Surg., 34: 10351040.

Wan, IY, Angelini, GD, Bryan, AJ, Ryder, I and Underwood, MJ (2001) Prevention of spinal cord ischaemia
during descending thoracic and thoracoabdominal
aortic surgery. Eur. J. Cardiothorac. Surg., 19:
203213.
Webb, TH and Williams, GM (1999) Thoracoabdominal
aneurysm repair. Cardiovasc. Surg., 7: 573585.
Weigang, E, Hartert, M, Sircar, R,V, Samson, P, Pitzer, K,
Genstorfer, J, Zentner, J and Beyersdorf, F (2005a) Setup
of neurophysiological monitoring with tcMEP/SSEP during thoracoabdominal aneurysm repair. Thorac. Cardiovasc. Surg., 53: 2832.
Weigang, E, Hartert, M, Von Samson, P, Sircar, R, Pitzer,
K, Genstorfer, J, Zentner, J and Beyersdorf, F (2005b)
Thoracoabdominal aortic aneurysm repair: interplay of
spinal cord protecting modalities. Eur. J. Vasc. Endovasc. Surg., 30: 624631.

M.R. Nuwer (Ed.)
829
CHAPTER 61
Intraoperative monitoring during cardiac surgery

Jean-Michel Guerit*
Clinique Edith Cavell, Unite dExplorations Electrophysiologiques du Syste`me Nerveux, CHIREC, B-1050 Brussels, Belgium
61.1. Neurological and neuropsychological

outcome of cardiac surgery
Although the introduction of cardiopulmonary bypass
(CPB) caused a significant decrease in the morbidity
and mortality of cardiac surgery, neurological complications remain a significant concern. It is difficult to
estimate their actual rate as reported outcomes depend
on several methodological factors (prospective vs. retrospective studies, evaluation by a neurologist vs. nonneurologists, rough neurological examination vs. fine
assessment of cognitive function, short-term vs. longterm outcome studies) (Selnes et al., 2006). Most studies were performed after coronary artery bypass graft
(CABG) surgery, which is by far the most frequently
performed heart surgery. Other results were gained
from open-heart surgery, valvular replacement, or
repair of congenital malformations. Noteworthy, most
of these procedures are performed under CPB.
Stroke is the most recognized complication with a
frequency of 25% in large series limited to a rough
neurological examination. Risk factors for stroke are
largely comparable to those involved in noncardiac
procedures: proximal aortic atherosclerosis, arterial
hypertension, previous neurological disorders, excessive consumption of alcohol, and older age (Ricotta
et al., 1995; Roach et al., 1996). In the study by Ricotta
et al. (1995) stroke was due to aortic emboli in 32% of
cases, emboli of cardiac origin in 12%, systemic hypotension in 12%, concomitant vascular disease in 11%,
and to undetermined cause in 33%.
Neurocognitive dysfunction is the second wellrecognized complication. Its rate actually depends on
the time of examination and the thoroughness of the
neuropsychological examination. Thus, prospective
*
Correspondence to: Jean-Michel Guerit, M.D., Ph.D., Stress

and Craniofacial Pain Clinic, CHIREC, Avenue Louise 390,
Tel.: 32-2-6480081; fax: 32-2-626-9909.
E-mail: guerit.nccn@skynet.be (J.-M. Guerit).
studies using extensive neuropsychological test batteries found that 3575% of patients had cognitive
impairment within 710 days, and 1030% had persisting deficit at 36 months (Sotaniemi et al., 1986). More
recent studies demonstrated that cognitive decline after
CABG was transient and reversible and that most, but
not all, patients return to their baseline performance
between 3 and 12 week after surgery (Selnes et al.,
2006). Risk factors for early encephalopathy or cognitive impairment are the same as for strokes but also
include a history of peripheral vascular disease or prior
CABG, postoperative atrial fibrillation, and cerebral
atrophy (Sila, 1998). Long-term cognitive changes were
also described, but it is actually difficult to differentiate
decline due to the CABG procedure itself, age-related
changes, or the natural history of the underlying cerebrovascular disease (Selnes et al., 2006).
Failure to awaken and intracranial bleeding are
other rare but particularly severe complications occurring in less than 0.2% and 0.03% of cases, respectively
(Sila, 1998).
61.2. Pathophysiology of neurological
complications
Neurological complications occurring during cardiac
surgery are vascular in origin. Two types of mechanisms may be involved: embolism or hypoperfusion.
61.2.1. Embolism
The major causes of embolism are the release of an aortic arch atheromatous plaque during aorta cannulation,
macro- or microemboli from the CPB device, or valve
debris or air and particles during mechanical deflating
of the heart (Sila, 1998). As already mentioned, macroembolism of identified aortic of cardiac origin accounts
for near 50% of strokes (Ricotta et al., 1995).
Several studies using transcranial Doppler have
demonstrated that showers of microemboli are
J.-M. GUERIT
830
common during cardiac surgery, particularly during

aortic cannulation and clamping/declamping (Mullges
et al., 2001). Interestingly, Brown et al. (1999)
described neuropathological lesions suggestive of
microembolism in the brains of patients who died
shortly after surgery, but not in those who died 1 week
or longer thereafter. This progressive disappearance of
lesions of microembolic origin should be paralleled
with the early improvement of cognitive functions.
61.2.2. Hypoperfusion
Cerebral blood flow (CBF) autoregulation comprises a
set of mechanisms that maintain a sufficient brain perfusion when blood pressure (BP) decreases. It typically operates between mean BPs of the order of 60
and 150 mm Hg (Paulson et al., 1990) but could
remain efficient down to a mean arterial BP of
30 mm Hg (Sila, 1998). By contrast, these can be
reduced owing to a preexisting vasculopathy or merely
less efficient in older age. Self-regulation mechanisms
also seem less efficient in the presence of a nonpulsating blood flow, like that obtained under CPB (Andersen et al., 1985). This makes that CBF autoregulation
may be overtaken in the current BP regimens of cardiac surgery. Furthermore, focal narrowing(s) of one
or both carotid arteries or intracranial vessels can
increase the sensitivity of the corresponding brain
region to an otherwise harmless CBF decrease.
Even in the absence of any focal artery narrowing,
systemic hypoperfusion can be associated with focal
brain lesions due to the higher sensitivity of these
regions at the boundary between the territories of the
three main cerebral arteries (last-meadows phenomenon). These regions are the parietooccipital junction at the boundary between the territories of the
sylvian and posterior cerebral arteries, and the paramedial frontal cortex at the boundary between the territories of the sylvian and anterior cerebral arteries.
61.2.3. Strategies to improve prognosis
Based on these mechanisms, several methods were
proposed to improve prognosis. The probability of
embolism can be decreased through improvement of
filtration techniques in CPB devices or intraoperative
echocardiography to identify plaques and visualize a
safe site for aortic manipulation and cannulation.
Brain hypoperfusion can be avoided by maintaining BP in a range that remains compatible with an
efficient CBF autoregulation. Whenever a sufficient
BP cannot be maintained (owing, for instance, to

too high a risk of bleeding or myocardial overload
when no CPB is used), hypothermia can increase
the brain resistance to hemodynamic disturbances
through a decrease of its metabolic needs. However,
hypothermia has also been reproached to worsen
myocardial performance and to decrease further the
range of autoregulation so that it should be preferred
only for the patients with higher risk of neurological
complications (Martin et al., 1994). This fact that the
efficiency range of autoregulation depends on such
miscellaneous factors implies that an optimal BP
should ideally be determined individually on the
basis of an online determination of brain function.
61.2.4. The ischemic penumbra
A progressive decrease in brain perfusion is first associated with functional, but still reversible, alterations
in neuronal transmission and, thereafter, with neuronal
destruction. This first step, in which neurons are no
more functional but still alive and salvageable by
reperfusion, is known as the ischemic penumbra zone
(Astrup et al., 1981; Fig. 1). The time spent in ischemic
penumbra that is, the time of ischemia the brain can
cope before irreversible lesions occur is inversely
related to the importance of the CBF decrease, and
for a given degree of decrease, to the initial brain
metabolism: the higher the metabolism, the shorter
Cerebral blood flow
(ml/100 g/min)
Normal brain function
20
Ischemic penumbra zone
Neuronal destruction
Time
Fig. 1. The ischemic penumbra zone. A cerebral blood

flow (CBF) decrease down to less than 20 ml/100 g/min
initially causes a reversible decrease in neuronal function,
followed by irreversible neuronal destruction. The time the
brain can spend in this so-called ischemic penumbra
zone is inversely related to the degree of CBF decrease.
VASCULAR SURGERY
the ischemic penumbra duration. Metabolism, on its

turn, depends on both intrinsic brain properties (it is
higher in the gray vs. white matter) and extracerebral
factors (particularly, brain temperature).
Most importantly, as it is associated with neuronal
functional alterations, the entry of the brain in the
ischemic penumbra zone is associated with alterations
of brain electrogenesis, which can be detected by
electroencephalography (EEG) or evoked potentials
(EPs). In other words, EEG or EPs may provide early
warning signals of impending ischemia at a reversible
stage, all the more as they largely depend on the integrity of the cerebral cortex, which, for metabolic reasons, has the highest sensitivity. This is the rational
for intraoperative monitoring (IOM). Of course, the
actual usefulness of such detection will depend on
the strategy which can be developed to solve it.
61.3. How IOM helps in these procedures,
how it changes the procedure?
If one considers pathophysiology, IOM can be useful
in three respects: as a potential indicator of the time
of occurrence and functional consequences of embolism, as a tool to optimize BP, and as a follow-up of
deep hypothermia.
At first glance, IOM is of little help to prevent
neurological complications of embolism, at least for
two reasons: embolism detection is best achieved
with ultrasonography and does not belong to the primary scope of EEG and EPs, and, when embolism
has been detected, it is, in principle, too late. Both
arguments need to be qualified. First, ultrasonography just detects the passage of emboli through
carotid arteries or, for intracranial Doppler, the
arrival of macroemboli in the intracranial circulation,
but it does not assess the functional consequences of
embolism in terms of neuronal dysfunction. Hence,
this may sometimes be achieved with EEG and
EPs, provided that their generators should belong to
the involved brain area (which is usually the case
for sylvian embolism). The fact that an embolism
that was detected by ultrasonography actually causes
more or less localizable brain dysfunction might be
used as an argument for thrombolysis or neuroprotective treatment and as a tool to follow-up neuronal
function in response to that treatment. In this respect,
intraoperative-monitored embolism could constitute a
dream model for the evaluation of new neuroprotective agents as it allows picking up events at very
early stage and, at the same time, quantitatively
831
following up the improvement of neuronal function,

if any. Second, ultrasonography is not always available. Third, identifying the exact cause of an intraoperative accident or localizing it in time may help
identify some critical or hazardous surgical steps,
thereby improving strategy in subsequent procedures.
By contrast, IOM can be of a major help for these
focal or bilateral changes that are due to hemodynamic disturbances, which, in principle, should be
reversed by increasing BP, even if it remains to be
proven that maintaining the brain in the ischemic
penumbra zone for a limited amount of time is really
harmful. This requires controlled studies, which are
not available at the present time. On the contrary,
and provided that the cerebral zone at risk should
be accordingly targeted, the absence of IOM changes
is an excellent argument to reassure the surgeon by
telling him that there is no impending ischemia,
which allows him either to maintain perfusion pressure at the same low level or to go on without undue
haste with his procedure.
The last application of IOM in cardiac surgery
concerns operations performed under deep hypothermia. Ideally, IOM should meet two objectives. First,
and most evident, it should provide an endpoint for
a sufficient level of brain protection that is based
on the disappearance of some activities (see below).
Second, IOM should remain compatible with the
detection of brain dysfunction. Indeed, the brain protection that hypothermia provides only spans the
intraoperative period. However, any surgical maneuver carrying some risk for a given brain area may
remain undetected if this area is made silent by hypothermia, in which case long-term sequels are to be
feared. Consequently, the choice of an optimal IOM
tool in hypothermia should be a trade-off between
both sensitive and resistant activities.
61.4. Which methods are applied to this
procedure?
Thus far, the EEG has been more popular than EPs
and, consequently, has been the object of many more
validating studies. This is partly due to historical reasons as the EEG was first used in the operating theater in the mid-1950s. It is also likely that many
non-neurophysiologists wrongly consider it to be a
more straightforward tool to evaluate brain function
than EPs. Theoretical reasons were also put forward
to advocate the use of EEG: it would provide online
assessment of brain function contrary to EPs that
J.-M. GUERIT
832
are only obtained after the averaging time, and it

would provide more widespread assessment than
EPs. As we shall see later on, all these considerations
need to be qualified.
61.4.1. EEG and SEP changes that may be
observed in cardiac surgery
N20
P27
C93-A2
Grand average
N24
P24
1.25 V
10 ms
Grand average
The EEG changes that are suggestive of impaired

brain perfusion and impending brain ischemia were
summarized in the chapter on carotid surgery (Burke
et al., 1999). Although these changes can be observed
visually, several types of computer-based equipment
are available to compress the EEG (compressed spectral array, CSA) or represent it through trend curves
of derived EEG variables: slow-to-fast frequencies
ratios, asymmetry indexes, burst-suppression ratio
(see Chapter 6 by Van Huffelen, this volume). Not
only the nature of changes but also their dynamics
need to be considered (Schwartz et al., 1973). Slow
changes are the most frequent (68% in the series of
Schwartz et al., 1973), are usually related to physiological factors (changes in anesthesia, hypothermia),
and are not associated with any long-term neurological complications. Quick changes (i.e., changes
occurring within less than 12 min) are usually indicative of developing ischemia; the associated outcome
depends on the number of observed episodes and
their degree of reversibility (Schwartz et al., 1973;
Nevin et al., 1989). In principle, all these changes
can be focal or generalized, according to the pathophysiology of the process at the origin of ischemia,
and the EEG capability of detecting focal changes
depends directly on the number of channels. Thus,
the International Federation of Clinical Neurophysiology (IFCN) recommendations suggest that,
although the whole 1020 array of 21 electrodes will
be applied in some cases, fewer often suffice, for
instance symmetrical derivations such as F4-P4 and
F3-P3 (Burke et al., 1999). Noteworthy, such simplified montage may be insufficient to pickup ischemia
restricted to boundary zones.
Somatosensory evoked potentials (SEPs) are the
most widely used EP tool. Indeed, visual EPs are
too variable to be usable in anesthetized patients
and brainstem auditory EPs only test the pontine
auditory pathways, which are usually spared in cardiac surgery. Most recordings were performed with
median nerve SEPs. We strongly suggest recording
these with an ear reference, which allows, first,
making the difference between parietal and frontal
P45
Fpz-A2
N30
P14
Fig. 2. Normal median nerve SEPs (right median nerve

stimulation). The use of an ipsilateral ear reference allows
recording the lemniscal P14. Parietal components (top
recordings) include N20, P24, P27, and P45. Frontal components include P20 and N24 (which are likely to be the
positive and negative counterparts of N20 and P24, respectively), and N30. Analysis time: 100 ms (10-ms prestimulus
baseline).
SEP components, and, second, recording the P14

(Fig. 2). As this component comes from the lower
brainstem, that is, a region that escapes most neurological complications of cardiac surgery, it is particularly resistant and can therefore be used as the
criterion that severe cortical abnormalities are not
the consequence of an upstream problem (technical
failure of stimulators, peripheral nerve dysfunction,
and/or spinal cord dysfunction).
Interestingly, there is good evidence that the frontal N30 is generated close to the supplementary
motor area (Desmedt et al., 1987), which lies at the
boundary between the anterior cerebral and sylvian
artery territories, and that the parietal P45 is generated close to the parietooccipital junction, which lies
at the boundary between the sylvian artery and posterior cerebral artery territories (Fava et al., 1992).
Thus, both N30 and P45 may be expected to be particularly sensitive to boundary zone ischemia occurring after systemic hypoperfusion. Noteworthy, the
presence or absence of the frontal N30 and the parietal P45 is the criterion allowing differentiating mild
from moderate SEP alterations in our classification
or mild, moderate, and severe SEP abnormalities in
carotid endarterectomy (Guerit et al., 1997; Witdoeckt
et al., 1997; see Fig. 1 of chapter on carotid endarterectomy, Chapter 57, in this volume). Mild SEP abnormalities consist of the disappearance of the frontal
VASCULAR SURGERY
N30 and/or parietal P45, moderate abnormalities to the

additional disappearance of P27, and severe abnormalities of the disappearance of P24 and/or N20. Our
recent experience indicates that this classification,
which was initially proposed in carotid endarterectomy, is also usable in cardiac surgery (see Table 2 of
chapter on carotid endarterectomy, Chapter 57, in this
volume).
Noteworthy, the parietal N20 component is particularly resistant to hypoperfusion so that its disappearance (with a preserved P14) may be indicative of
either a particularly severe anoxic insult in systemic
hypoperfusion (symmetrical changes) or damage,
usually of embolic origin, in the subcortical somatosensory pathways. Conversely, using the disappearance or a more than 50% amplitude decrease of
N20 is unlikely to be sensitive enough to detect
mild-to-moderate systemic hypoperfusion.
SEPs can also be obtained through stimulation of
the posterior tibial nerve (PTN). Theoretically, they
can usefully complement median nerve SEPs as their
cortical components are generated in the paramedian
parietal and frontal areas, which depend on the anterior cerebral artery, contrary to median nerve SEPs
whose cortical components mainly come from the
sylvian artery territory. They are rarely used in cardiac surgery, except in some complex procedures in
which both the brain and the spinal cord have to be
simultaneously monitored.
61.4.2. EEG or SEPs: what technique should be
preferred, if any?
Although it is true that the EEG provides more continuous assessment, the actual usefulness of continuous assessment of brain function in impending
ischemia may be questioned. Actually, this issue
can be likened as that of the optimal sampling frequency in signal analysis, which actually depends
on the signal frequency. Here, the key issues are,
on the one hand, the dynamics of brain ischemia,
which always take some time to become evident,
even on the EEG, and, on the other hand, the time
taken for interpretation. Thus, Clute and Levy
(1990) monitored EEG changes during controlled
periods of cardiac arrest and found that the mean
time of onset of EEG changes was 10.2 s (range
3.321.1 s) after induction of ventricular fibrillation.
Moreover, although these changes usually consisted
of slowing and attenuation (in 82% of cases), other
changes were less typical of brain ischemia (loss of
833
delta wave activity in 7% of cases and an increase

in amplitude of theta activity in 11% of cases, which
may still increase the time required for EEG interpretation). Finally, as they just reflect the entry of the
brain in the ischemic penumbra zone, the appearance
of EEG and/or EP changes does not mean that irreversible lesions did occur (for a recent review, see
Koenig et al., 2006). Given that in operating room
(OR) conditions (curarized patient, relatively high
stimulus intensity) current acquisition systems allow
getting interpretable SEPs in less than 20 s (60100
sweeps at 35 stimuli/s), the fact that EEG provides
continuous assessment cannot be considered as a real
advantage with respect to SEPs, exactly like it is not
useful to increase sampling frequency far above the
Nyquist frequency.
Whether EEG actually provides more widespread
assessment than EPs should also be qualified. As
already mentioned, this would perhaps be true with
multichannel recording but this is not in that way that
the EEG is usually recorded in the OR (Burke et al.,
1999). The actual issue is to adapt the technique to
target these neural structures, which are the most at
risk, that is, the parietooccipital and paramedial frontal cortices in systemic hypoperfusion and the territory of the sylvian artery, which is statistically the
most exposed in macroembolism. Hence, we have
seen that SEPs were able to target these regions at
least as well as the EEG. Moreover, SEPs are also
sensitive to lower brainstem ischemia, which cannot
be detected with the EEG. Therefore, we estimate
that this argument of a more widespread assessment
is actually a false one. Of course, both EEG and
EPs can miss ischemic damage that would occur in
more restricted brain areas.
In summary, there is no definite argument that
would favor EEG versus SEPs as a warning tool of
developing ischemia. The EEG could a bit more
sensitive. On the other hand, the EEG is also
extremely sensitive to environmental noise, contrary
to SEPs that are relatively protected by the averaging process and, therefore, according to our experience, more specific. Theoretically, the EEG is a
better tool whenever IOM has to be used to reassure
the surgeon that nothing happens, while SEPs are
better and safer to detect abnormalities that are
important enough to reorientate surgical strategy.
This favors the combined used of both techniques.
If one of both must be chosen, we suggest guiding
the choice as a function of each laboratories own
expertise.
J.-M. GUERIT
834
61.4.3. EEG and SEPs in deep hypothermia
N20
35.0 C
The basic principle of hypothermia or barbiturate coma

is to increase the brain resistance to a drop in CBF
through a decrease in its metabolism. This may prevent
irreversible brain lesions through a rightward shift of the
limit of the ischemic penumbra zone (Fig. 1). As mentioned above, hypothermia allows brain protection only
throughout the hypothermic period and may hide potentially dangerous maneuvers.
Hypothermia causes well-known and rather stereotyped EEG and SEP changes (Figs. 3 and 4).
EEG changes first consist in a slowing of background
rhythms (at 2930 C) followed by a burst-suppression pattern at 2022 C, and electrocerebral
silence (ECS) at 1518 C (Prior, 1973). ECS is considered the endpoint for hypothermic circulatory
arrest. SEP changes consist of a latency increase of
all components, followed by the gradual disappearance of N30 (at a mean temperature of 30 C), P27
(27 C), N20 (21 C), and P14 (17 C) (Guerit
et al., 1990; Fig. 3). We routinely use SEPs in surgical repairs of the aortic arch under deep hypothermic
circulatory arrest in which case P14 disappearance is
used as the endpoint (Guerit et al., 1994).
P14
P24 P27
26.2 C
N20
P13 P14
19.1 C
P24
P13 P14
15.0 C
P13
P14
Fig. 3. Median nerve SEPs in hypothermia. Superimposition of ipsilateral and contralateral parietal recordings, the
shaded area corresponds to the cortical components. At
26.2 C, P14 has become bilobed, consisting of both
P13 and P14 components; P27 has disappeared. There is
a marked increase in N20-P24 latency. At 19.1 C, ipsilateral and contralateral recordings do not differ anymore,
which means that all cortical components have disappeared. At 15 C, the P14 component starts to disappear;
only P13 persists.
Again, there is no definite argument that would favor

EEG or SEPs. Of course, neither the EEG nor the SEPs
can be used to detect brain ischemia when ECS or P14
and/or N20 disappearance are used as endpoints. By
1.2
Amplitude P14 ( mV)
6
Prediction interval (95%)
Confidence interval (95%)
0.8
0.6
0.4
0.2
10
r2 = 0.8388
0.7
15
20
25
30
35
Temprature (degrees)
40
Amplitude N20 ( mV)

%N20 lat = 9.05
5.0
Amplitude N30 (mV)
0.0
1
0.9
0.8
1.0
4.0
4.0
3.0
2.0
1.0
0.0
3.0
2.0
1.0
0.0
40
35
30
25
20
15
10
Fig. 4. Quantitative relationship between body temperature and N20 latency (left side) and N30, N20, and P14 amplitude.
There is an exponential evolution of N20 latency. Note the successive disappearance of N30, N20, and P14 (from Ghariani
et al. (2000) with permission. # 2000 International Federation of Clinical Neurophysiology).
VASCULAR SURGERY
835
contrast, intermediate levels of hypothermia (2225 C)

may remain compatible with the SEP detection of ischemia, while the EEG is no more usable (Fig. 5).
61.5. Clinical considerations when potentials change
Table 1 summarizes the main relevant causes of EEG
and/or EP changes during cardiac surgery. Overall
changes can be subdivided into five categories: physiological modifications (level of anesthesia, body
temperature), changes due to an increase or a decrease
in cerebral perfusion, normal individual variability,
peripheral nerve pathology, or technical failure. A
two-step procedure is suggested to identify these
factors: (1) put forward a hypothesis based on recent
modifications of the level of anesthesia, patients
F3-A1
C3-A1
F4-A2
C4-A2
Fpz-A2
C6sp
Cerv.g
100 ms
5 V
N20
C4
100 ms
5 V
N20
N20
Fpz
100 ms
5 V
Cerv.dr
100 ms
10 V
C3
100 ms
5 V
Fpz
100 ms
10 V
A 35 C
B 31 C
C 27 C
E 24 C left CCC
F 24 C BP restoration
F3-A1
C3-A1
F4-A2
C4-A2
Fpz-A3
C6spCerv.g
100 ms
5 V
C4
100 ms
5 V
Fpz
100 ms
5 V
Cerv.dr.
100 ms
10 V
C3
100 ms
5 V
Fpz
100 ms
10 V
D 24 C
Fig. 5. Combined recording of EEG and median nerve SEPs in a complex cardiac procedure under deep hypothermia.
AC: from 35 to 27 C; DF: 24 C (magnification 2). Note the progressive slowing down of the EEG from 35 to
27 C, followed by its progressive flattening. At 24 C, there are some rare bursts of EEG activity (F) but the EEG is no
more analyzable. SEP evolution is in keeping with that described in Fig. 3 with the progressive disappearance of P45
and P27 and a marked increase in P14 and N20 latency. However, N20 still persists even at 24 C. This allowed detection
of left-hemisphere ischemia alters left carotid cross-clamping [see arrow in E, comparison with immediately preceding
waveform] and its recovery after BP restoration (F).
J.-M. GUERIT
836
Table 1
The main causes of EEG or SEP abnormalities in
cardiac surgery
D
Unilateral
Bilateral
Mild
# Perfusion
pressure
Moderate
# Perfusion
pressure
# Perfusion
pressure
Embolism of
sylvian artery
" Level of anesthesia

# Body temperature
# Perfusion pressure
# Perfusion pressure
Severe
Embolism of basilar
artery (SEPs)
Circulatory arrest
parameters, and/or intraoperative events, and (2) check

whether the observed changes are compatible with the
hypothesized mechanism or not.
The first step illustrates the prime importance of
getting continuous control on patients parameters,
either through a remote display or, ideally, direct visualization of parameters on the EEG screen. Control of
intraoperative events can be achieved through oral
communication with the surgeon or, better, through
video control of the operative field.
The following criteria help achieve the second step:
The preservation of P14 constitutes an almost
absolute warranty that no technical failure

occurred and that SEP changes are not due to
peripheral nerve dysfunction. This is due to the
remarkable P14 robustness (even after total circulatory arrest, P14 persists for about 1 min after
N20 disappearance).
These changes which are due to physiological modifications (level of anesthesia, body temperature)
are always bilateral and symmetrical. Conversely,
those changes which are due to changes in perfusion pressure may be unilateral or bilateral, focal or
diffuse.
While these EEG changes that are due to hypothermia are very nonspecific and sometimes hardly
distinguished from those that are due to other factors, body temperature gives rise to very specific
SEP changes initially consisting of an increase in
latency of all peaks, including P14 without any
major morphological change in the overall aspect
of waveforms. The increase in P14 latency may
help to differentiate hypothermia from an increase in
the level of some anesthetics (propofol, halogenated
agents) that can also cause an increase in the P14

N20 interpeak latency. The preservation of waveform morphology helps to differentiate these
changes that are due to hypothermia from those
which are due to ischemia.
Rapidly occurring moderate-to-severe abnormalities are usually not the consequence of physiological changes and their cause should be identified as
quickly as possible (if necessary by direct interpellation of the anesthesiologist or the surgeon).
Conversely, mild changes usually do not constitute
an emergency situation. In the authors experience,
these justify just having a look on patients parameters and to the overall mood of the operative room.
If no obvious cause is identified, one or two control
recordings are performed and an explicit intervention is only performed in case of worsening of the
waveforms. If waveforms come back to normal, this
is interpreted as reflecting individual variability;
very often this could already be forecasted by baseline inspection.
61.6. Which surgeons expectancies can
or cannot be met?
As a very general principle, the increased safety that
IOM provides to the surgeon can sometimes lead him
to perform acts that he would never have dared in the
absence of monitoring. That is to say that the surgeon
should be aware not only of the exact capabilities,
but also of the limitations of IOM. Actually, it is
the neurophysiologists job not only to provide him
with that information but also to check that he correctly understands and accepts it. False safety is
worse than no safety at all and I personally would
prefer to refuse to work with a surgeon in whom I
am not fully confident.
Even if it looks evident at first glance, the surgeon
should be clearly reminded that neurological complications can be of both embolic and hemodynamic
origin and that, evidently, IOM has no power to prevent embolism . . . Similarly, he should be clearly
informed about what brain regions are or are not
within the scope of IOM so that a small embolism
involving an unexplored region may remain undetected. Next, he must be made aware that the process
of ischemia may take some time so that it is usually
illusory to detect mild-to-moderate changes within
the second following a suspicious maneuver. Finally,
he must be aware that IOM just follows up the
VASCULAR SURGERY
current brain status but may be unable to forecast late

complications. At this respect, he should be made
aware that beyond its protective role, hypothermia
may prevent from detecting complications that will
occur later on.
When should one warn the surgeon? This issue is
largely related to that of abnormality criteria. Early
warning (or less severe abnormality criteria) provides
the surgeon with the opportunity immediately to
associate neurophysiologic alterations with intraoperative events; the other side of the coin is that it
increases the number of false positives, places the
surgeon under stress to no purpose, and unduly prolongs the operation. Late warning (or more severe
abnormality criteria) potentially increases the risk of
neurological damage and complicates the display of
associations between neurophysiologic abnormalities
and intraoperative events. I have adopted an intermediate attitude: I always warn the surgeon as soon as worrying neurophysiologic abnormalities are observed,
but he knows that this warning should be confirmed
before an urgent reaction is needed; this allows him
immediately reacting to any obvious surgical problem without undue stress in the absence of such
evidence. Similarly, I prefer to use continuous EP or
EEG worsening rather than isolated abnormalities as
the criterion that something wrong is happening.
The ability to select relevant information and reject
false positives is above all a matter of neurophysiologists expertise.
61.7. What new types of monitoring would
be desirable in the future?
Theoretically, some of the limitations that were mentioned in the previous paragraph can be bypassed by
specific methods. From a strictly neurophysiologic
standpoint, increasing the number of EEG channels,
multiplying the sensory modalities used for EP
recording, combining motor EPs, increasing the number of quantitative indices which may be derived
from frequency analysis of the EEG can allow examining more and more widespread brain areas. On the
other hand, neurophysiologic monitoring evaluates
just one aspect, that is, brain function and it has a
poor etiologic specificity. That is to say that neurophysiologic IOM should be considerably enriched
through its integration into the overall context: anesthetic and surgical context, ultrasonography, direct
measurements of brain metabolism. On its turn, combining IOM to these other measurements will help
837
to increase their own specificity. Moreover, to base

decisions on a cluster of overlapping arguments
increases system redundancy and, therefore, safety.
This justifies the concept of multimodality monitoring, which, in order to prevent late complications,
should also be pursued in the postoperative period
in the ICU (Edmonds, 2002; Murkin, 2004). The
reverse side of the coin is that, at the same time,
multiplying parameters increases the complexity of
decisions and, therefore, the risk of errors. There
is a need to develop new pattern recognition algorithms, possibly through neuronal networks or expert
systems.
On the other hand, there are more than 800,000
myocardial revascularization procedures which are
performed annually throughout the world, and a high
percentage of these would justify monitoring. This is
largely beyond the neurophysiologic offer so that
IOM has currently to be limited to the procedures
that are the most at risk of neurological complications, particularly in case of associated carotid stenosis in symptomatic or asymptomatic patients. This
justifies the need to develop methods of remote
IOM, probably through telemedicine, whose advantage would not be limited to the intraoperative period
but would also include the postoperative stay in the
ICU.
References
Andersen, K, Waaben, J, Husum, B, Voldbly, B, Bodker,
A, Hansen, AJ and Gjedde, A (1985) Non-pulsating cardiopulmonary bypass disrupts flow-metabolism couple
in the brain. J. Thorac. Cardiovasc. Surg., 90: 570579.
cerebral ischemia the ischemic penumbra. Stroke,
12: 723725.
Brown, WR, Moody, DM and Challa, CR (1999) Cerebral fat
embolism from cardiopulmonary bypass. J. Neuropathol.
Exp. Neurol., 58: 109119.
Yingling, CD and Jones, SJ (1999) Intraoperative monitoring. In: G Deuschl and A Eisen (Eds.), Recommendations for the Practice of Clinical Neurophysiology:
Guidelines of the International Federation of Clinical
Neurophysiology (EEG Suppl. 52). International Federation of Clinical Neurophysiology, Elsevier Science,
Clute, HL and Levy, WJ (1990) Electroencephalographic
changes during brief cardiac arrest in humans. Anesthesiology, 73: 821825.
Desmedt, JE, Nguyen, TH and Bourguet, M (1987) Bitmapped color imaging of human evoked potentials with
838
reference to the N20, P22, P27 and N30 somatosensory
response. Electroencephalogr. Clin. Neurophysiol.,
68: 119.
monitoring for cardiac surgery. Heart Surg. Forum, 5:
225228.
Fava, E, Bortolani, E, Ducati, A and Schieppati, M (1992)
Role of SEP in identifying patients requiring temporary
shunt during carotid endarterectomy. Electroencephalogr. Clin. Neurophysiol., 84: 426432.
Ghariani, S, Matta, A, Dion, R and Guerit, JM (2000) Intraand postoperative factors determining neurological
complications after surgery under deep hypothermic circulatory arrest: a retrospective somatosensory evoked
potential study. Clin. Neurophysiol., 111: 10821094.
Guerit, JM, Sove`ges, L, Baele, P and Dion, R (1990)
Evoked potentials in profound hypothermia for ascending aorta repair. Electroencephalogr. Clin. Neurophysiol., 77: 163173.
Guerit, JM, Verhelst, R, Rubay, J, El Khoury, G, Noirhomme, P, Baele, P and Dion, R (1994) The use of
somatosensory evoked potentials to determine the optimal degree of hypothermia during circulatory arrest.
J. Card. Surg., 9: 596603.
Guerit, JM, Witdoeckt, C, De Tourtchaninoff, M, Ghariani,
S, Matta, A, Dion, R and Verhelst, R (1997) Somatosensory evoked potential monitoring in carotid surgery.
I. Relationships between quantitative SEP alterations
and intraoperative events. Electroencephalogr. Clin.
Koenig, MA, Kaplan, PW and Thakor, NV (2006) Clinical
neurophysiologic monitoring and brain injury from cardiac arrest. Neurol. Clin., 24: 89106.
Martin, TD, Craver, JM, Gott, JP, Weintraub, WS, Ramsay,
J, Mora, CT and Guyton, RA (1994) Prospective, randomized trial of retrograde warm blood cardioplegia:
myocardial benefit and neurologic threat. Ann. Thorac.
Surg., 57: 298302.
Mullges, W, Franke, D, Reents, W and Babin-Ebell, J
(2001) Brain microemboli counts during extracorporeal
circulation depend on aortic-cannula position. Ultrasound Med Biol., 27: 933936.
J.-M. GUERIT
Murkin, JM (2004) Perioperative multimodality monitoring: an overview. Semin. Cardiothorac. Vasc. Anesth.,
8: 167171.
Nevin, M, Colchester, ACF, Adams, S and Pepper, JR
(1989) Prediction of neurological damage after cardiopulmonary bypass surgery. Anaesthesia, 44: 725729.
Paulson, OB, Strandgaard, S and Edvinsson, L (1990)
Cerebral autoregulation. Cerebrovasc. Brain Metab.
Rev., 2: 161192.
Prior, PF (1973) The EEG in Acute Cerebral Anoxia.
Excerpta Medica, Amsterdam.
Ricotta, JJ, Faggioli, GL, Castilone, A and Hassett, JM
(1995) Risk factors for stroke after cardiac surgery:
Buffalo Cardiac-Cerebral Study Group. J. Vasc. Surg.,
21: 359364.
Roach, GW, Kanchuger, M, Mangano, CM, Newman, M,
Nusmeier, M, Wolman, M, Aggarwal, A, Marschall, K,
Graham, SH, Ley, C, Ozanne, G and Mangano, DT
(1996) Adverse cerebral outcome after coronary bypass
surgery. Multicenter study of perioperative Ischemia
Research Group and the Ischemia Research and Education Foundation Investigators. N. Engl. J. Med., 335:
18571863.
Schwartz, MS, Colvin, MP, Prior, PF, Strunin, L, Simpson,
BR, Weaver, EJM and Scott, DF (1973) The cerebral
function monitor: its value in predicting the neurological
outcome in patients undergoing cardio-pulmonary bypass. Anaesthesia, 28: 611618.
Selnes, OA, McKhann, GM, Brorowicz, LM and Grega, MA
(2006) Cognitive and Neurobehavioral dysfunction after
cardiac bypass procedures. Neurol. Clin., 24: 133145.
Sila, CA (1998) Neurologic complications of vascular surgery. Neurol. Clin., 16: 920.
Sotaniemi, KA, Mononen, H and Hokkanen, ET (1986)
Neuropsychological outcome after open-heart surgery:
a five-year neuropsychological follow-up study. Stroke,
17: 410416.
Witdoeckt, C, Ghariani, S and Guerit, JM (1997) Somatosensory evoked potential monitoring in carotid surgery.
II. Comparison between qualitative and quantitative
scoring systems. Electroencephalogr. Clin. Neurophysiol., 104: 328332.
SECTION IV
Intensive Care Monitoring

M.R. Nuwer (Ed.)
840
CHAPTER 62
Neonatal ICU monitoring

Lena Hellstrom-Westasa,*, Linda S. De Vriesb and Ingmar Rosenc
a
Neonatal Intensive Care Unit, Department of Pediatrics, University Hospital, SE-22185 Lund, Sweden
b
Department of Neonatology, Wilhelmina Childrens Hospital, 3508 AB Utrecht, The Netherlands

c
Department of Clinical Neurophysiology, University Hospital, SE-22185 Lund, Sweden
62.1. Introduction
The overall aim of the care in the sick newborn baby is
survival with an intact neurological and developmental
outcome. Paradoxically, on line continuous functional
monitoring in the neonate has for a long time been
focused on cardiorespiratory variables rather than cerebral function. The electrophysiological brain activity as
reflected by electroencephalography (EEG) is well
established as a tool for providing information about
the current metabolic state of the brain and the occurrence of epileptic seizure episodes. In neonatal care,
EEG has been extensively used for estimation of the
degree of cerebral maturation in preterm infants (Connell
et al., 1987; Lamblin et al., 1999) and for detection of
abnormal patterns indicating focal and global cerebral
lesions (Holmes and Lombroso, 1993). In the neonatal
setting as well as in intensive care in general, EEG has
been recorded intermittently, at best serially rather than
continuously. Performing a full multichannel EEG in a
newborn requires specialized technical skill in securing
correct electrode positioning and impedances, identification of extracerebral biologic and nonbiologic artifact
sources. Among clinical neurophysiologists interpretation of neonatal EEG is considered as a demanding
task considering the different features related to different
gestational age, activity state, and medication.
The main disadvantage with intermittent conventional EEG during neonatal care is the difficulty in discriminating emerging trends of development of the
electrocerebral activity over hours and days. If at all
*
Correspondence to: Lena Hellstrom-Westas, M.D., Ph.D.,

Neonatal Intensive Care Unit, Department of Pediatrics,
University Hospital, SE-22185 Lund, Sweden.
Tel.: +46-46-171000 (switchboard); +46-46-178068 (NICU);
fax: +46-46-178430.
E-mail: lena.hellstrom-westas@med.lu.se (L. HellstromWestas).
possible, it takes special skills not usually available at

the intensive care unit (ICU) to identify such long-term
changes of EEG patterns. In order to have an impact on
the intensive care, monitoring of the electrocortical
activity would have to be continuous and simple in
terms of recording equipment and number of recording
electrodes. Furthermore, EEG features which are
immediately relevant for clinical decisions should be
continuously available at the bedside and made
possible to interpret by the attending physician day
and night. Such techniques have been applied for a
long time using fast fourier transformation (FFT) analysis of the EEG spectral content and presented
as compressed spectral arrays (CSA) or spectral edge
frequency (SE) and used in intensive care and
anesthesia (see relevant chapters in this book) as well
as in neonatal monitoring (Inder et al., 2003).
62.2. Amplitude-integrated EEG
Amplitude-integrated EEG (aEEG) is a technique for
simplified EEG monitoring which has found an
increasing clinical application in neonatal intensive
care units (NICUs). The method is based on a timecompressed (usually 6 cm/h) semilogarithmic (linear
010 mV, logarithmic 10100 mV) display of the
peak-to-peak amplitude values of EEG passed through
an asymmetrical bandpass filter that strongly enhances
higher over lower frequencies with suppression of
activity below 2 Hz and above 15 Hz in order to minimize artifacts from sources as sweating, movements,
muscle activity, and interference (Fig. 1). The bandwidth reflects variations in minimum and maximum
EEG amplitude. The semilogarithmic display enhances
identification of changes in the low-voltage range and
avoids overloading of the display at high amplitudes.
The cerebral function monitor (CFM) was designed
in the 1960s by Maynard and originally applied in adult
intensive care by Prior (Maynard et al., 1969). The
INTENSIVE CARE MONITORING
841
CFM 6000 Filter

db 0
12 db/
decade
Amplification/attenuation (dB)
60 db/
decade
5
120 db/
decade
20
0
5
10
15
20
25
30
dB
10
40
15
20
25
30
60
35
1
5
10
50 100
Frequency in Hz (log scale)
40
EEG frequency, Hz
Semilogarithmic scaling
60.00
50.00
mm
40.00
30.00
20.00
10.00
60
40
20
0
Amplitude, mV
20
40
60
Fig. 1. aEEG filter settings in the Olympic CFM 6000, and Viasys Nicolet One (upper right), filter settings are similar
in other aEEG monitors. All monitors also display the aEEG in a semilogarithmic fashion, as shown by the lower gray
schematic drawing (linear display of amplitudes between 0 and 10 mV, and logarithmic from 10 to 100 mV).
method was first applied to newborn babies in the late

1970s and early 1980s (Viniker et al., 1984). The original CFM concept has been developed and several new
machines, based on digital technique, are available. In
addition to the aEEG trend, these equipments display
the original EEG signal in present and passed time. This
is a main advantage compared with the original CFM
since artifacts and EEG seizure patterns are easier to
identify. Published reports on neonatal aEEG include
both clinical and experimental studies (Bunt et al.,
1996; De Vries and Hellstrom-Westas, 2005). The
finding that the aEEG is suitable for very early prediction of outcome after perinatal asphyxia is presently
being used as an inclusion criterion in studies of therapeutic hypothermia (Gluckman et al., 2005) and has
resulted in spreading use of the technique. The clinical
experiences with neonatal aEEG monitoring have been
exemplified in an atlas (Hellstrom-Westas et al., 2003).
For new users of aEEG, the value of the method

is usually apparent when clinical aEEG reveals
abnormal brain activity patterns that would otherwise pass unrecognized such as subclinical seizure
activity or transient deterioration due to unexpected
complications such as hypoglycemia or pneumothorax (Hellstrom-Westas et al., 2003). It is, however,
important to be aware of limitations and risks of
overinterpretation when using aEEG (Toet et al.,
2002, Rennie et al., 2004). The possibility of
extracting detailed information is lost in exchange
for markedly improved possibilities of following
long-term trends and changes in the overall cerebroelectrical activity. Therefore, continuous aEEG
does not replace but is complementary to standard
EEG. We recommend that at least one standard
EEG, preferably including a period of quiet sleep,
should be recorded in infants monitored with aEEG.
842
62.2.1. Electrodes
Both thin subdermal needle electrodes and disk and
hydrogel electrodes have been used. Although application of needle electrodes may be considered more
invasive, it involves less handling, and the impedance usually remains below 5 kO for several days.
Silver-silver chloride disk electrodes are used for
standard EEG but need training to give good-quality
recording. Some adhesive electrodes seem to work
after skin preparation, but they should not be placed
on the forehead or over the temples since forehead
and temporal muscle activity and eye movements
interfere with the recording.
62.2.2. Correlation between aEEG and EEG
In the neonatal setting aEEG is usually derived from
one pair of biparietal (P3-P4) electrodes or two channels, one from each hemisphere (C3-P3 and C4-P4,
or F3-P3 and F4-P4). Although previously singlechannel recordings dominated, this is now mainly
recommended for very preterm infants. The aEEG
has been compared with EEG both regarding epileptic seizure detection and background features. In general, there is a good correlation between main
findings in the aEEG/CFM and EEG.
62.2.2.1. Seizure patterns
In the aEEG, onset of seizure activity is usually seen
as an abrupt increase of maximal and minimal aEEG
amplitude, sometimes only minimal amplitude, often
followed by a transient postictal amplitude depression (Fig. 2A). Repeated seizures are identified as
repeated peaks in the aEEG trace, sometimes
described as a saw-tooth pattern (Fig. 2B). The
aEEG is complementary to the EEG when diagnosing
epileptic seizure activity. It is not unusual that the
aEEG, which may run for hours and days, detects
epileptic seizure activity which is not seen when the
standard EEG is recorded (Fig. 3). The appearance
of a repeated saw-tooth pattern in the aEEG trace
can sometimes also facilitate identification of an
unclear rhythmic EEG pattern as an ictal EEG phenomenon. Although all seizures with duration more
than 30 s, recorded simultaneously in five-channel
tape recorded EEG, could be identified in the
single-channel biparietal aEEG/CFM recording,
some focal, low-amplitude and brief seizures may
be missed (Bjerre et al., 1983; Hellstrom-Westas,
1992; Toet et al., 2002; Rennie et al., 2004). With a
M-WESTAS ET AL.
L. HELLSTRO
two-channel aEEG recording, focal seizures often

appear more clearly in one unilateral channel than
in the contralateral or biparietal recording channel
(Toet et al., 2002; Fig. 4). In our experience, patients
displaying focal seizures which are difficult to discriminate in the aEEG trace during the course of continuous monitoring usually also develop more
widespread seizure patterns which are easily detectable. The background EEG activity (see below) has
an effect on how easily seizure patterns are discriminated in the aEEG trace. With a highly abnormal flat
or discontinuous background, the onset of seizures is
often more easily discernible than with a normal continuous EEG background (Fig. 5). Continuous spiking
will not cause any abrupt changes of the aEEG trace
and would be missed in an aEEG record without access
to the original EEG trace (Hellstrom-Westas, 1992).
The combined display of aEEG and original EEG
traces in the new machines is a valuable asset in discriminating seizure patterns from episodic artifacts
such as those produced by caregiving (Fig. 6).
62.2.2.2. Background
The characterization of the background EEG activity
may sometimes differ slightly between the aEEG and
EEG. A discontinuous aEEG with low interburst amplitude at a level above zero may be classified as a
suppression-burst pattern in the EEG (Toet et al.,
2002). This may be due to the high sensitivity of the
aEEG in the low-amplitude range. However, close
inspection of the original EEG signal is necessary in
order to exclude interference from electrocardiogram,
electronic equipment, or high-frequency ventilation
(Fig. 7).
62.2.3. Normal aEEG
The normal aEEG pattern changes with gestational
age (Verma et al., 1984; Viniker et al., 1984; Thornberg and Thiringer, 1990; Burdjalov et al., 2003;
Olischar et al., 2004; Sisman et al., 2005). As predicted from EEG, the aEEG in the very preterm
infant is mainly discontinuous and becomes gradually more continuous with increasing gestational
age (Connell et al., 1987). The normal discontinuous
EEG in very preterm infants, trace discontinue, with
low-amplitude EEG activity between bursts, should
be distinguished from the abnormal pattern burst suppression with inactive (isoelectric) interburst intervals (Lombroso, 1985; Fig. 8). This difference is
often possible to distinguish in the aEEG trace: the
843
Fig. 2. A: A single seizure is often seen as an abrupt increase in aEEG amplitude followed by brief postictal amplitude
depression. This term infant had unilateral seizures due to a left-sided cerebral infarction but the seizures were also detectable on this 2-hour duration biparietal recording, the duration of the seizure was almost 20 min. Below the aEEG is 25 s of
EEG, corresponding to the peak of the seizure (gray vertical line through the aEEG trace). B: Repeated seizures and status
epilepticus is often seen as a saw-tooth pattern as in this term infant with a 4-hour duration aEEG. The 25 s of EEG
below represents the aEEG traces at the vertical gray line. Another EEG trend, a spectrogram, is displayed between the
aEEG and the EEG.
F3-T3
T3-P3
F3-C3
C3-P3
F4-T4
T4-P4
F4-C4
C4-P4
T3-C3
C3-Cz
Cz-C4
C4-T4
Pg1-Pg2
ECG1
Fig. 3. This term infant developed a middle cerebral artery infarction on the left side. The 6-hour duration aEEG trends are
derived from the left (F3-P3) and right (F4-P4) hemispheres, respectively. The upper aEEG tracing from the left side shows
a slightly discontinuous background with sleepwake cycling (SWC) and repeated brief seizures. Some seizure activity is
also discernible on the lower aEEG trace from the right side. Thirty-five seconds of EEG representing the vertical gray line
is shown below the aEEG tracings. A standard EEG (below) performed on the same day shows rhythmic activity over the
left central region but not seizure activity.
100
V
845
1 2Left
aEEG
6 7
1 Right
2
aEEG
6 7
10
5
0
100
V
10
5
0
300
V
12:16
Wed 23 Feb
1 2
aEEG Cross-cerebral
13:16
14:16
14:55
6 7
100
50
25
10
5
3
0
11:16
12:16
13:16
14:16
14:47
Fig. 4. Three simultaneous aEEG derivations (left, right, cross cerebral) from a moderately preterm infant with necrotizing
enterocolitis and hypoglycemia. The aEEG shows a status epilepticus (saw-tooth pattern) during the first half of the 3.5-h
recording. The majority of the seizures are detectable in all three channels.
50 EEG Waveform Left

V
25
0
25
50
50 EEG Waveform Right
V
25
0
25
50
5
100 aEEG Left

V
+1
+2
10
11
12
13
14
10
11
12
13
14
+3
+4
+5
seconds
10
5
0
100 aEEG Right
V
10
5
0
18:12
19:12
20:12
21:12
21:31
Fig. 5. Seizures are often easier to detect in the aEEG when they appear on a severely depressed background. The twochannel example from a term asphyxiated infant with classical watershed infarcts on MRI and normal outcome at
18 months shows four brief seizures. The EEG from the first seizure is displayed above the aEEG trends.
Fig. 6. It is sometimes difficult in the aEEG to distinguish changes due to care procedures from epileptic seizure activity.
This example shows a 6-h recording where two similar changes appear in the aEEG. The first (A) was created by a care
procedure, and the second (B) by an epileptic seizure, as shown by the corresponding EEG tracings below.
Fig. 7. This recording was obtained from an extremely preterm infant, born at 24 gestational weeks. The ventilator was
changed to high-frequency ventilation (HFV) just before the infant was given surfactant (vertical gray line). The abrupt
increase (2 mV) in the minimum level is caused by the HFV. The monorhythmic 16 Hz HFV artifact is displayed as a peak
in the inserted Frequency Graticule. The surfactant administration also results in a clear, but rather discrete decrease in burst
density. There is no clear explanation for this phenomenon, which has been described previously.
Fig. 8. aEEG background patterns with corresponding EEGs below. A: Continuous background (C) with SWC in healthy fullterm infant (two channels, EEG upper panel, aEEG lower panel). B: Continuous and discontinuous aEEG background with
immature SWC in an infant with Dandy Walker malformation at 35 weeks gestation (one channel, aEEG upper panel, EEG
lower panel). C: Discontinuous (DC) background which gradually becomes more continuous, as seen by the rise in the
minimum amplitude, in a full-term infant after cardiac surgery. D: DC background in normal very preterm infant, maximum
amplitude is often higher and the variability in the minimum amplitude is larger than in full-term infants with DC pattern.
E: Burst-suppression with 100 bursts/h (BS) in moderately sedated preterm infant. F: BS with <100 bursts/h (BS-) in
severely asphyxiated full-term infant. G: Low voltage (LV) in a severely asphyxiated infant. H: Flat (FT) aEEG and EEG
in severely asphyctic full-term infant. The raised aEEG baseline, between 3 and 5 mV, is caused by interference from electrocardiogram, which can be seen in the EEG trace. The burst-like changes on the flat aEEG background are caused by movement
artifacts. Published with permission from Hellstrom-Westas et al. (2006). # 2006 by the AAP.
848
burst suppression pattern has a straight low margin at

01 (2) mV, while the trace discontinu pattern has a
more variable lower margin around 06 (7) mV.
Cyclic variations in the aEEG background (sleep
wake cycling, SWC) presumably reflecting cyclic
alterations between periods of quiet sleep and periods
of active sleep/wakefulness can be seen in well infants
from around 2627 weeks gestation (HellstromWestas et al., 1991; Kuhle et al., 2001). A recently
published study evaluating SWC in relation to standard EEG in stable preterm infants between 25 and
30 weeks supports the aEEG findings (Scher et al.,
2005). SWC develops with increasing maturation,
and from 3132 weeks gestation quiet sleep periods
are clearly discernible in the aEEG trace as distinct
periods with increased bandwidth. At term, these
periods represent trace alternant EEG patterns
(Fig. 9).
62.2.3.1. aEEG in the full-term infant
In full-term infants, aEEG is an excellent method for
evaluating cerebral function and cerebral recovery
after hypoxic-ischemic insults, for example perinatal
asphyxia and apparent life-threatening events
(ALTE; Bjerre et al., 1983). Besides contributing to
the clinical evaluation, the immediately available
aEEG pattern allows early information to the parents.
Infants needing intensive care are at risk of developing cerebral complications due to circulatory instability (e.g., sepsis), hypoxia (e.g., persistent pulmonary
hypertension, meconium aspiration, cardiac malformations, or diaphragmatic hernia), hypoglycemia, or
seizures (Hellstrom-Westas et al., 1985, 2003). Cerebral symptoms of cerebral dysfunction are often difficult to detect due to the medical condition itself or
because sedatives or analgesics have been administered. In such infants electrocortical activity usually
remains unaffected or only moderately depressed,
unless high doses of antiepileptic drugs or sedatives
have been given. A continuous aEEG with amplitudes between 10 and 25 mV, with some cyclic
variation or a fully developed SWC, is usually a
reassuring sign of noncompromised brain function
in a full-term infant. In the very preterm infant, a
discontinuous aEEG background pattern with dense
burst activity, and some smooth cyclical variability
of the background amplitude, would indicate the
same.
The aEEG has been shown to be very sensitive
for early prediction of outcome in asphyxiated fullterm newborns, as shown by several investigators
M-WESTAS ET AL.
L. HELLSTRO
(Thornberg and Ekstrom-Jodal 1994; Eken et al.,

1995; Hellstrom-Westas et al., 1995a; Al Naqeeb
et al., 1999; Toet et al., 1999; Shalak et al., 2003;
Ter Horst et al., 2004a; Osredkar et al., 2005; Van
Rooij et al., 2005). A continuous or slightly discontinuous aEEG pattern during the first 6 h is associated with a high chance of cerebral recovery and
normal outcome. Presence of SWC appearing in the
aEEG within the first 36 h in infants with moderate
hypoxic-ischemic encephalopathy (HIE) was associated with favorable outcome as compared to infants
with later appearing SWC (Osredkar et al., 2005).
Presence of seizures in aEEG traces of asphyxiated
infants has not been as clearly associated with outcome as background activity. This may be due to
the effect of the global cerebral ischemic insult,
with actually fewer seizures in the most severely
asphyxiated infants with poor outcome (HIE grade 3)
than in the more moderately asphyxiated infants
(HIE grade 2). However, it is our impression that
recurrent seizure activity, or status epilepticus, is associated with a worse outcome, but this has not been
studied from a perspective including quantification of
seizures. An abnormal aEEG before and during
ECMO (extracorporeal membrane oxygenation) treatment in newborn infants with severe respiratory failure also seems to have a high sensitivity (1.0) and
specificity (0.75) for death or intracranial neuropathology (Pappas et al., 2006). In infants needing ECMO it
is not uncommon that brain function may be affected
by prior hypoxia-ischemia and seizures.
62.2.3.2. aEEG in the preterm infant
Although the aEEG has not been evaluated in preterm infants as extensively as in full-term infants,
several studies have shown the utility of the technique also in preterm infants. Two prospective evaluations of aEEG in very preterm infants during the
first days have been published. Both showed correlation between degree of the intraventricular hemorrhage (IVH) (or periventricular hemorrhage) and the
depression of the aEEG background, quantified as
the amount of continuous activity above 3 mV. A
majority of infants developing IVH had epileptic seizure patterns in the aEEG tracing, often entirely subclinical (Greisen et al., 1987; Hellstrom-Westas
et al., 1991). The number of bursts per hour was
found to be associated with outcome in preterm
infants with large (grade 34) IVH already during
the first 2448 h of life (Hellstrom-Westas et al.,
2001). A relatively good outcome could be predicted
849
Fig. 9. The upper recording is from a 3-day-old stable infant born at 27 weeks gestation. An intermediate trace shows the
variability in interburst intervals during the different sleepwake states. The tracing below shows well-developed SWC in a
recovering term asphyxiated infant. The EEG samples represent quiet sleep periods at the vertical gray lines in the aEEG
tracings.
when the maximum number of bursts per hour

exceeded 135. Presence of SWC during the first
week of life was associated with better outcome,
but presence of epileptic seizure activity was not
associated with worse outcome in these infants.
62.2.3.3. Effect of medications on aEEG

Administration of morphine, phenobarbital, lidocaine, and midazolam may depress aEEG activity
(Hellstrom-Westas et al., 1988; Bell et al., 1993;
Ter Horst et al., 2004b; Van Leuven et al., 2004).
850
Morphine and sufentanil are also known to depress

EEG activity (Young and Da Silva, 2000; Nguyen
The Tich et al., 2003). Lidocaine, given as infusion
for treatment of recurrent seizures, often results in a
gradual depression of the aEEG developing into discontinuous or burst-suppression (BS) pattern. This
becomes more evident when the infusion is stopped
and the aEEG background recovers to a more continuous pattern. In our experience, a loading dose of
phenobarbital (1020 mg/kg) may result in moderate
depression of background activity. In full-term
infants there is usually no major change in background activity. If the phenobarbital loading dose
results in severe depression this is often a sign of
more severely compromised cerebral function. This
is similar to effects found with midazolam in neonates with HIE (Van Leuven et al., 2004). Diazepam
often results in a profound depression of aEEG activity in preterm infants, while the effect in full-term
infants often is less marked. In preterm infants, surfactant administration may result in a transient aEEG
depression for about 10 min (Hellstrom-Westas et al.,
1992). The cause of the aEEG depression is not
known. It is associated with a rise in cerebral blood
volume (as measured with near-infrared spectroscopy), but there is no relation to transient hypotension or changes in blood gases (Skov et al., 1992).
62.2.4. Classification and description of a
neonatal aEEG recording
aEEG tracings have been described and classified in
several different ways, depending on whether normal
or abnormal features have been addressed or whether
full-term or preterm infants have been in focus.
62.2.4.1. aEEG in normal infants
A number of publications, two of them already from
1984, have described normal aEEG patterns in fullterm and preterm infants (Viniker et al., 1984; Verma
et al., 1984; Thornberg and Thiringer, 1990; Burdjalov et al., 2003; Olischar et al., 2004; Sisman et al.,
2005). Among the earlier studies, both Verma et al.
(1984) and Viniker et al. (1984) found that the aEEG
feature which most clearly was related to maturation
in healthy infants was the lower edge (amplitude) of
the quiet sleep trace. In 1990, Thornberg and Thiringer (1990) presented a study on normal aEEG development in preterm and full-term infants who had an
uneventful neonatal period and were neurologically
normal at follow-up, and normative data for
M-WESTAS ET AL.
L. HELLSTRO
minimum and maximum amplitudes during wakefulness and sleep were presented. Burdjalov et al.
(2003) studied 30 infants with gestational ages
2439 weeks, serially on 146 occasions, twice during
the first 3 days of life and then weekly or biweekly.
A scoring system was developed, evaluating continuity, cyclic (SWC) changes, amplitude of lower border, and bandwidth. The range of summarized score
points was 013. The total score correlated with gestational and postconceptional ages, the highest total
scores was attained at 3536 weeks postconceptional
age gestation. Abnormal patterns, for example, BS or
seizures were not included in the scoring system.
Olischar et al. (2004) recorded very preterm infants,
born at 2329 gestational weeks and without cerebral
ultrasound abnormalities, and defined quantitatively
three different patterns: discontinuous low voltage,
continuous, and discontinuous high voltage. Further
normative data have recently been collected by Sisman et al. (2005) from preterm infants without neurological abnormalities, born 2532 gestational weeks
recorded biweekly from 2448 h to 35 postmenstrual
weeks. Amplitude data largely substantiated those of
Thornberg and Thiringer, and clear SWC was present
from 29 gestational weeks.
62.2.4.2. aEEG in infants with compromised
brain function
Several studies have addressed and characterized
abnormal aEEG records. Bjerre et al. (1983) in one of
the earlier studies described background patterns
as continuous or interrupted (discontinuous). The
recorded infants included asphyxiated preterm infants,
full-term asphyxiated infants, and infants up to
5 months who had suffered ALTE. Cerebral recovery
was associated with an initial continuous tracing or a
change in background pattern from interrupted to
continuous within 12 days of the hypoxic-ischemic
insult. Hellstrom-Westas et al. (1995a) classified
aEEG from asphyxiated full-term infants as continuous normal voltage (CNV), BS, continuous extremely
low voltage (CLV), and flat (FT). Toet et al. (1999)
used a similar classification with the addition of discontinuous normal voltage (DNV) to the previous four
patterns. Both classifications show very high correlation with outcome. Infants with CNV or DNV during
the first 6 h of life are likely to survive without
sequelae, while infants with BS, CLV, or FT have a
high risk for death or severe handicap. Al Naqeeb
et al. (1999) used a classification including three categories for normal and abnormal aEEGs in full-term
851
infants, based on median upper and lower margin

amplitude of the widest band of the aEEG trace. The
aEEG background activity was classified as normal
amplitude when the upper margin was >10 mV and
the lower margin was >5 mV; moderately abnormal
when the upper margin was >10 mV and the lower
margin was <5 mV; and suppressed when the upper
margin was <10 mV and lower margin was <5 mV.
Seizure activity was defined, but not SWC. This
classification was recently used in a randomized multicenter study on postasphyctic head cooling. The
intervention improved outcome in infants with a
moderately abnormal aEEG pattern before 5.5 h
(Gluckman et al., 2005).
Recovery over time is a general feature of electrocortical background abnormalities after an insult. A
few full-term postasphyctic infants with CLV/FT or
BS patterns at 6 h attained a CNV pattern within
24 h and half of them survived without sequelae
(Van Rooij et al., 2005).
Table 1
62.2.4.3. Proposal for a new classification

Several models for classification of aEEG patterns
have been suggested. It is obvious that some classifications are relevant only for a certain group of NICU
patients, for example, asphyxiated full-term or normal
preterm infants. We therefore propose a classification
of aEEG background patterns, based on EEG terminology, which could be used in all newborn infants
(Table 1, Fig. 8; Hellstrom-Westas et al., 2006). In this
proposal we have used and modified some of the classifications described above. The classification does
not include evaluation of background patterns and
amplitudes in relation to normative data for different
gestational ages since this has been accounted for
already (Hellstrom-Westas et al., 2006). It should be
noted that many EEG terms (e.g., focal, multifocal,
sharp waves, delta pattern) are not relevant for aEEG
since a trend monitor does not provide this type of
information. Similar to basic EEG interpretation we
suggest that pattern recognition forms the basis also
for interpretation of the aEEG trace.
The amplitude of the electrocortical activity is
extremely important, but this measure must be handled with caution since voltage may be affected by for
example, interelectrode distance and scalp edema.
Furthermore, extracerebral signals like electrocardiogram and high-frequency oscillation ventilation may
add to the trace and interfere with the interpretation
of a discontinuous signal as DNV versus BS or LV
versus FT. Close inspection of the original EEG trace
accompanying the aEEG trace in modern equipments
Sleepwake cycling
Sleepwake cycling (SWC) in the aEEG is characterized
by smooth sinusoidal variations, mainly in the
minimum amplitude. The broader bandwidth
represents discontinuous background activity during
quiet sleep (trace alternant EEG in term infants),
and the more narrow bandwidth corresponds to
more continuous activity during wakefulness and
active sleep
No SWC: No cyclic variation of the aEEG
background
Imminent/immature SWC: Some, but not fully developed, cyclic variation of the lower amplitude, but not
developed as compared to normative gestational age
representative data
Developed SWC: Clearly identifiable sinusoidal variations between discontinuous and more continuous
background activity with cycle duration 20 min
Suggested classification of aEEG patterns in preterm

and term infants
Background pattern
Describes the dominating type of electrocortical activity in
the aEEG trace.
Continuous (C): Continuous activity with lower (minimum) amplitude around 7 to 10 (5) mV and maximum
amplitude 1025 (50) mV
Discontinuous (DC): Discontinuous background with
minimum amplitude variable, but below 5 mV, and maximum amplitude above 10 mV
Continuous low voltage (CLV): Continuous background
pattern of very low voltage (around or below 5 mV)
Burst-suppression (BS): Discontinuous background with
minimum amplitude without variability at 01 (2) mV,
and bursts with amplitude >25 mV
BS denotes burst density 100 bursts/h, and BS
means burst density <100 bursts/h
Inactive, flat (FT): Mainly inactive (isoelectric tracing)
background below 5 mV.
Seizures
Epileptic seizure activity in the aEEG is usually seen
as an abrupt rise in the minimum amplitude, usually
also a simultaneous rise in the maximum amplitude.
The raw-EEG should show simultaneous seizure
activity with a gradual buildup and then decline,
in frequency and amplitude, of repetitive spikes
or sharp-wave or activity with duration at
least 510 s
Single a solitary seizure
Repetitive single seizures appearing more frequently than at 30-min intervals
Status epilepticus continuously ongoing seizure activity >30 min
852
is highly recommended. Nevertheless, normative

values for minimum and maximum amplitudes of the
aEEG at different gestational ages have been published (Verma et al., 1984; Viniker et al., 1984; Thornberg and Thiringer, 1990; Sisman et al., 2005) and are
very helpful for assisting in evaluating aEEG recordings in relation to normal traces for a certain gestational age. Especially the minimum amplitude at
quiet sleep periods is valuable to assess since it
increases with gestational age up to term. Furthermore,
short-time variability of minimum amplitude is a sign
separating a discontinuous EEG from a BS pattern.
62.3. Discussion
aEEG is a method for continuous long-term monitoring of brain activity which has proved to be very successful in newborn infants of all gestational ages, and
which will probably gain more widespread use in
NICUs. The simplicity of the method makes it possible to apply and interpret around the clock by the
neonatal staff, and the interrater reliability is usually
excellent (Al Naqeeb et al., 1999; Toet et al., 2002;
Thorngren-Jerneck et al., 2003; Ter Horst et al.,
2004a). Due to the reduced number of electrodes,
single-channel biparietal aEEG is suitable also for
monitoring the most preterm infants.
A common classification of patterns would be beneficial and increases the understanding of the method.
A number of studies have shown that the previously
used classifications are relevant for identifying
abnormalities that could lead to early intervention.
They have been designed for special purposes and
are not generally applicable to all clinical situations.
The three main features which can be extracted from
neonatal aEEGs, background activity, SWC, and electrographic seizure patterns are reflecting different
physiological and pathophysiological mechanisms
with different clinical implications in different groups
of infants, and should be described separately. We are
therefore not suggesting an overall rating score for
clinical aEEG.
The aEEG identifies subclinical seizure activity
that would otherwise pass without detection, but we
do not have a consensus how to treat clinically silent
seizures. However, for the neonatologist caring for a
sick infant, knowledge about such seizure activity is
usually of great clinical value since it may direct further treatment and investigation. In this context, however, all users of aEEG must be aware that seizure
M-WESTAS ET AL.
L. HELLSTRO
identification with a single- or two-channel monitor

is good, but that some seizures may pass unrecognized. Especially, brief, single, and very focal seizures as well as continuous spiking may be missed
by inspection of the aEEG trace only. In our experience, the additional display of the original EEG signal provided in modern equipments improves the
sensitivity and makes it possible to discriminate single seizures from periods of artifacts due to, for
example, patient care procedures. Close collaboration
with neurologists and neurophysiologists is recommended and such collaboration is made easier with
the recording equipment connected to a hospital network. Standard EEGs should be recorded frequently
in infants undergoing aEEG, especially when this
shows an abnormal pattern. More advanced systems
for neurophysiological monitoring in the NICU are
available, for example video-EEG. However, to our
knowledge, video-EEG has not been used for standard clinical monitoring in a large population of
high-risk infants in the NICU, probably because it
is difficult to combine with other ICU devices and
requires specialized interpretation resources. However, in infants with severe and recurrent seizures, a
period with video-EEG is probably the best method
for close-up monitoring.
An important issue is how many channels of EEG
should be integrated into aEEG traces. For standard
monitoring one or two channels are probably sufficient. When abnormalities are identified and localized in standard EEG the number and positioning
of aEEG recordings may be modified. With a bilateral two-channel aEEG monitoring one may be able
to identify gross asymmetries in background and seizure patterns (Toet et al., 2002). These types of findings will have to be substantiated by multichannel
standard EEG to make certain that the asymmetry is
not due to an artifact. A multichannel system as standard monitoring is not usually feasible since it may
hinder the care and disturb the patient, which is especially relevant for the fragile very preterm infant.
The CFM algorithm for aEEG was developed by
Maynard and Prior for use in adult intensive care
but has been found to be useful also in neonatal
EEG monitoring, although it would probably have
been different if it had been developed de novo for
neonatal applications. Probably, the ideal EEG trend
monitor for neonatal use has still to be developed.
The addition of an original EEG trace is of value.
Calculation of burst density from aEEG traces of discontinuous EEG is tedious and should be replaced by
a direct additional trend displaying bursts per hour

or interburst interval. Furthermore, spectral content
of the EEG, expressed as, for example, SE has been
found to be of relevance in experimental (Fraser
et al., 2005) and clinical studies (Inder et al.,
2003) of white matter injury in premature subjects.
Recent studies show that very low frequency (DC)
components are predominant in the immature EEG
(Vanhatalo et al., 2002) and may become relevant
to include into a continuous monitoring paradigm
in the future. Automatic seizure detection is an issue
that is currently explored. Probably, EEG trend
analysis will be combined in integrated systems
with other modalities such as near-infrared spectroscopy for continuous evaluation of brain function in
sick neonates.
Studies in adult intensive care indicate that use of
EEG monitoring decreases costs (Vespa et al., 1999),
but no such studies are available in newborn infants.
Furthermore, besides the possibility of early identification of risk for brain injury and the possibility of
intervention with hypothermia, there are no studies
indicating that aEEG actually improves outcome.
The incidence of later epilepsy has been reported to
be lower in centers treating clinical as well as subclinical seizures than in cohorts where only clinical
seizures were treated (Clancy and Legido, 1991;
Hellstrom-Westas et al., 1995b; Toet et al., 2005).
In conclusion, aEEG has become a standard method
for simple brain monitoring in neonatal intensive care.
The aEEG is derived from the original EEG signal in a
well-defined way, presently copying the algorithm
originally developed by Maynard and Prior. Several
instruments are available providing the aEEG signal
with slightly different displays. After some training, a
user can classify some hours of aEEG recording
according to background activity, presence of seizures
and SWC, and take clinical decisions based on a large
body of evidence from studies in newborn babies. This
makes around-the-clock brain monitoring possible in
every NICU. New digital instruments provide access
to the original EEG signal and alternate ways of trend
analysis. Further research will show if these options
are clinically useful.
References
Al Naqeeb, N, Edwards, AD, Cowan, F and Azzopardi, D
(1999) Assessment of neonatal encephalopathy by amplitude integrated electroencephalography. Pediatrics, 103:
12631271.
853
Bell, AH, Greisen, G and Pryds, O (1993) Comparison of
the effects of phenobarbitone and morphine administration on EEG activity in preterm babies. Acta Paediatr.,
82: 3539.
Bjerre, I, Hellstrom-Westas, L, Rosen, I and Svenningsen,
NW (1983) Monitoring of cerebral function after severe
birth asphyxia in infancy. Arch. Dis. Child., 58: 9971002.
Bunt, JE, Gavilanes, AW, Reulen, JP, Blanco, CE and Vles,
JS (1996) The influence of acute hypoxemia and hypovolemic hypotension on neuronal brain activity
measured by the cerebral function monitor in new-born
piglets. Neuropediatrics, 27: 260264.
Burdjalov, VF, Baumgart, S and Spitzer, AR (2003) Cerebral function monitoring: a new scoring system for the
evaluation of brain maturation in neonates. Pediatrics,
112: 855861.
Clancy, RR and Legido, A (1991) Postnatal epilepsy after
EEG-confirmed neonatal seizures. Epilepsia, 32: 6976.
Connell, JA, Oozeer, R and Dubowitz, V (1987) Continuous 4-channel EEG monitoring: a guide to interpretation, with normal values, in preterm infants.
Neuropediatrics, 18: 138145.
De Vries, LS and Hellstrom-Westas, L (2005) Role of cerebral function monitoring in the newborn. Arch. Dis.
Child. Fetal Neonatal Ed., 90: F201F207.
Eken, P, Toet, MC, Groenendaal, F and De Vries, LS
(1995) Predictive value of early neuromaging, pulsed
Doppler and neurophysiology in full term infants with
hypoxic-ischaemic encephalopathy. Arch. Dis. Child.,
73: F75F80.
Fraser, M, Bennet, L, Gunning, M, Williams, C, Gluckman,
PD, George, S and Gunn, AJ (2005) Cortical electroencephalogram suppression is associated with post-ischemic
cortical injury in 0.65 gestation fetal sheep. Brain Res.
Dev. Brain Res., 154: 4555.
Gluckman, PD, Wyatt, JS, Azzopardi, D, Ballard, R,
Edwards, AD, Ferriero, DM, Polin, RA, Robertson, CM,
Thoresen, M, Whitelaw, A and Gunn, AJ (2005) Selective head cooling with mild systemic hypothermia after
neonatal encephalopathy: multicentre randomised trial.
Lancet, 365: 663670.
Greisen, G, Hellstrom-Westas, L, Lou, H, Rosen, I and
Svenningsen, NW (1987) EEG depression and germinal
layer haemorrhage in the newborn. Acta Paediatr.
Scand., 76: 519525.
Hellstrom-Westas, L, Rosen, I and Svenningsen, NW
(1985) Silent seizures in sick infants in early life. Acta
Paediatr. Scand., 74: 741748.
Hellstrom-Westas, L, Westgren, U, Rosen, I and Svenningsen, NW (1988) Lidocaine treatment of severe seizures
in newborn infants. I. Clinical effects and cerebral electrical activity monitoring. Acta Paediatr. Scand., 77:
7984.
Hellstrom-Westas, L, Rosen, I and Svenningsen, NW (1991)
Cerebral function monitoring during the first week of life
854
in extremely small low birthweight (ESLBW) infants.
Neuropediatrics, 22: 2732.
Hellstrom-Westas, L, Bell, AH, Skov, L, Greisen, G and
Svenningsen, NW (1992) Cerebroelectrical depression
following surfactant treatment in preterm neonates.
Pediatrics, 89: 643647.
Hellstrom-Westas, L, Rosen, I and Svenningsen, NW
(1995a) Predictive value of early continuous amplitude
integrated EEG recordings on outcome after severe
birth asphyxia in full term infants. Arch. Dis. Child.,
72: F34F38.
Hellstrom-Westas, L, Blennow, G, Lindroth, M, Rosen, I
and Svenningsen, NW (1995b) Low risk of seizure
recurrence after early withdrawal of antiepileptic treatment in the neonatal period. Arch. Dis. Child., 72:
F97F101.
Hellstrom-Westas, L, Klette, H, Thorngren-Jerneck, K and
Rosen, I (2001) Early prediction of outcome with aEEG
in preterm infants with large intraventricular hemorrhages. Neuropediatrics, 32: 319324.
Hellstrom-Westas, L, De Vries, LS and Rosen, I (2003) An
Atlas of Amplitude-Integrated EEGs in the Newborn.
Parthenon Publishing, London, pp. 1150.
Hellstrom-Westas, L, Rosen, I, De Vries, LS and Greisen,
G (2006) Amplitude integrated EEG: classification and
interpretation in preterm and term infants. NeoReviews,
7: e76e87.
Holmes, GL and Lombroso, CT (1993) Prognostic value of
background patterns in the neonatal EEG. J. Clin. Neurophysiol., 10: 323352.
Inder, TE, Buckland, L, Williams, CE, Spencer, C,
Gunning, MI, Darlow, BA, Volpe, JJ and Gluckman,
PD (2003) Lowered electroencephalographic spectral
edge frequency predicts the presence of cerebral white
matter injury in premature infants. Pediatrics, 111:
2733.
Kuhle, S, Klebermass, K, Olischar, M, Hulek, M, Prusa, AR,
Kohlhauser, C, Birnbacher, R and Weninger, M (2001)
Sleep-wake cycles in preterm infants below 30 weeks
of gestational age: preliminary results of a prospective
amplitude-integrated EEG study. Wien. Klin. Wochenschr.,
113: 219223.
Lamblin, MD, Andre, M, Challamel, MJ, Curzi-Dascalova,
L, dAllest, AM, De Giovanni, E, Moussalli-Salefranque,
F, Navelet, Y, Plouin, P, Radvanyi-Bouvet, MF, SamsonDollfus, D and Vecchierini-Blineau, MF (1999) Electroencephalography of the premature and term newborn.
Maturational aspects and glossary. Neurophysiol. Clin.,
29: 123219.
Lombroso, CT (1985) Neonatal polygraphy in full-term and
premature infants: a review of normal and abnormal findings. J. Clin. Neurophysiol., 2: 105155.
Maynard, D, Prior, PF and Scott, DF (1969) Device for
continuous monitoring of cerebral activity in resuscitated patients. Br. Med. J., 4: 545546.
M-WESTAS ET AL.
L. HELLSTRO
Nguyen The Tich, S, Vecchierini, MF, Debillon, T and
Pereon, Y (2003) Effects of sufentanil on electroencephalogram in very and extremely preterm neonates. Pediatrics, 111: 123128.
Olischar, M, Klebermass, K, Kuhle, S, Hulek, M, Kohlhauser,
C, Rucklinger, E, Pollak, A and Weninger, M (2004) Reference values for amplitude-integrated electroencephalographic activity in preterm infants younger than 30
weeks gestational age. Pediatrics, 113: e61e66.
Osredkar, D, Toet, MC, Van Rooij, LGM, Van Huffelen,
AC, Groenendaal, F and De Vries, LS (2005) Sleep-wake
cycling on amplitude-integrated EEG in full-term
newborns with hypoxic-ischemic encephalopathy. Pediatrics, 115: 327332.
Pappas, A, Shankaran, S, Stockmann, PT and Bara, R (2006)
Changes in amplitude-integrated electroencephalography
in neonates treated with extracorporeal membrane oxygenation: a pilot study. J. Pediatr., 148: 125127.
Rennie, JM, Chorley, G, Boylan, GB, Pressler, R, Nguyen, Y
and Hooper, R (2004) Non-expert use of the cerebral function monitor for neonatal seizure detection. Arch. Dis.
Scher, MS, Johnson, MW and Holditch-Davis, D (2005)
Cyclicity of neonatal sleep behaviors at 25 to 30 weeks
postconceptional age. Pediatr. Res., 57: 879882.
Shalak, LF, Laptook, AR, Velaphi, SC and Perlman, JM
(2003) Amplitude-integrated electroencephalography
coupled with an early neurologic examination enhances
prediction of term infants at risk for persistent encephalopathy. Pediatrics, 111: 351357.
Sisman, J, Campbell, DE and Brion, LP (2005) Amplitudeintegrated EEG in preterm infants: maturation of background pattern and amplitude voltage with postmenstrual
age and gestational age. J. Perinatol., 25: 391396.
Skov, L, Hellstrom-Westas, L, Jacobsen, T, Greisen, G and
Svenningsen, NW (1992) Acute changes in cerebral
oxygenation and cerebral blood volume in preterm
infants during surfactant treatment. Neuropediatrics,
23: 126130.
Ter Horst, HJ, Sommer, C, Bergman, KA, Fock, JM, Van
Weerden, TW and Bos, AF (2004a) Prognostic significance of amplitude-integrated EEG during the first 72
hours after birth in severely asphyxiated neonates.
Pediatr. Res., 55: 10261033.
Ter Horst, HJ, Brouwer, OF and Bos, AF (2004b) Burst
suppression on amplitude-integrated electroencephalogram may be induced by midazolam: a report on three
cases. Acta Paediatr., 93: 559563.
Thornberg, E and Ekstrom-Jodal, B (1994) Cerebral function monitoring: a method of predicting outcome in
term neonates after severe perinatal asphyxia. Acta Paediatr., 83: 596601.
Thornberg, E and Thiringer, K (1990) Normal patterns of
cerebral function monitor traces in term and preterm neonates. Acta Paediatr. Scand., 79: 2025.

Thorngren-Jerneck, K, Hellstrom-Westas, L, Ryding, E and
Rosen, I (2003) Cerebral glucose metabolism and early
EEG/aEEG in term newborn infants with hypoxicischemic encephalopathy. Pediatr. Res., 54: 854860.
Toet, MC, Hellstrom-Westas, L, Groenendaal, F, Eken, P
and De Vries, LS (1999) Amplitude integrated EEG 3
and 6 hours after birth in full term neonates with
hypoxic-ischaemic encephalopathy. Arch. Dis. Child.
Fetal Neonatal Ed., 81: F19F23.
Toet, MC, Van Der Meij, W, De Vries, LS and Van Huffelen, AC (2002) Comparison between simultaneously
recorded amplitude integrated EEG (Cerebral Function
Monitor) and standard EEG in neonates. Pediatrics,
109: 772779.
Toet, MC, Groenendaal, F, Osredkar, D, Van Huffelen, AC
and De Vries, LS (2005) Postneonatal epilepsy following amplitude-integrated EEG-detected neonatal seizures. Pediatr. Neurol., 32: 241247.
Vanhatalo, S, Tallgren, P, Andersson, S, Sainio, K, Voipio, J
and Kaila, K (2002) DC-EEG discloses prominent, very
slow wave activity patterns during sleep in preterm
infants. Clin. Neurophysiol., 113: 18221825.
855
Van Leuven, K, Groenendaal, F, Toet, MC, Schobben,
AF AM, Bos, SAJ, De Vries, LS and Rademaker,
CMA (2004) Midazolam and amplitude integrated
EEG in asphyxiated full-term neonates. Acta Paediatr.,
93: 12211227.
Van Rooij, LG, Toet, MC, Osredkar, D, Van Huffelen, AC,
Groenendaal, F and De Vries, LS (2005) Recovery of
amplitude integrated electroencephalographic background patterns within 24 hours of perinatal asphyxia.
Arch. Dis. Child. Fetal Neonatal Ed., 90: F245F251.
Verma, UL, Archbald, F, Tejani, N and Handwerker, SM
(1984) Cerebral function monitor in the neonate. I. Normal patterns. Dev. Med. Child Neurol., 26: 154161.
Vespa, PM, Nenov, V and Nuwer, MR (1999) Continuous
EEG monitoring in the intensive care unit: early findings
and clinical efficacy. J. Clin. Neurophysiol., 16: 113.
Viniker, DA, Maynard, DE and Scott, DF (1984) Cerebral
function monitor studies in neonates. Clin. Electroenceph., 15: 185192.
Young, GB and Da Silva, OP (2000) Effects of morphine on
the electroencephalograms of neonates: a prospective,
observational study. Clin. Neurophysiol., 111: 19551960.

M.R. Nuwer (Ed.)
856
CHAPTER 63
ICU EEG monitoring for acute seizures and

status epilepticus
Paul M. Vespa*
Department of Neurosurgery and Neurology, David Geffen School of Medicine, University of California,
Los Angeles, CA 90095, USA
63.1. Introduction
Status epilepticus (or more broadly, repetitive seizures)
has been a well-known phenomenon in clinical neurology since ancient times, with the term resulting from
French slang used at the Salpetrie`re and Bicetre hospitals in the 1800s (Kinnier Wilson and Reynolds,
1990). Yet today, the diagnosis and management of status epilepticus is more controversial than ever (Walker
et al., 2005). Status epilepticus is a common neurological critical illness that affects 120,000200,000 patients
per year in the United States with an overall acute
morality rate of 22%, and even higher among the
elderly. The mortality and outcome varies according
to the different types of status epilepticus, the age of
the patient, and the etiology of the status epilepticus.
Patients in status epilepticus present with a variety of
presenting illnesses, including acute brain injury, central nervous system infection, acute intoxication, acute
withdrawal from intoxication, acute renal and electrolyte abnormalities, and endocrine emergencies. This
broad etiological basis suggests that status epilepticus
can be triggered by a variety of insults or is a fundamental response by the brain to acute injury to the brain that
results from a metabolic or neurochemical alteration in
the brain. The etiological basis does play a role in eventual prognosis and in the therapeutic potential of acute
intervention, but in most cases the treatment of status
epilepticus can result in eventual clinical recovery of
the patient. Thus, understanding the treatment modalities that will result in identification and successful
Correspondence to: Paul M. Vespa, MD, Department of

Neurosurgery and Neurology, David Geffen School of
Medicine, CHS 18-218, UCLA Medical Center, Los
Angeles, CA 90095, USA.
Tel.: 1-310-206-6969; fax: 1-310-794-2147.
E-mail: PVespa@mednet.ucla.edu (P.M. Vespa).
treatment of status epilepticus is vitally important to

treating critically ill neurologic patients.
63.2. Definitions of status epilepticus
The traditional definition of status epilepticus has been
the state of continuous seizures for 30 min or recurrent
seizures without the return to normal level of consciousness. More recently, a revision of this definition
has been proposed by Lowenstein et al. (1999). The
new definition is ongoing seizures for more than
5 min. This revision is based on several factors. The
most convincing factor is that the expected time course
of the most common forms of seizures in chronic
epileptics undergoing closed-circuit television and
electroencephalography (CCTV-EEG) monitoring is
12 min. Using the CCTV-EEG method, the large
majority of seizures are quite brief with resolution of
mental status including immediate recall being reestablished within minutes of the end of the seizure. In
addition, persistent unresponsiveness is associated
with persistent nonconvulsive epileptiform activity
under these conditions. The definition also includes
EEG evidence of seizure activity after a clinical
seizure in which a return to consciousness has not
occurred.
There are many types of status epilepticus, including
generalized tonic clonic status epilepticus, complex partial status epilepticus, absence status epilepticus, and
nonconvlusive status epilepticus. Some of these constitute a spectrum of chronic illness and/or part of the clinical interictal state (i.e., absence status epilepticus),
while most of these are part of an acute illness and hence
represent a distinct state of electrical instability and
clinical condition. In these cases of acute illness, nonconvulsive seizures or complex partial status epilepticus is the most frequent form of status epilepticus, and
the diagnosis is highly dependent on continuous EEG
(cEEG) monitoring.
63.3. Methods of cEEG monitoring

The methods used in our intensive care unit (ICU) have
been previously published (Vespa et al., 1999b, 2002,
2003b) after being clinically established as a brain monitor in 1994. The method has been to monitor the brain in
the neurocritical care unit just like the heart is monitored
in the cardiac care unit. Briefly, a14 channel EEG (F4Cz, T4-Cz, P4-Cz, O2-Cz, F3-Cz, T3-Cz, P3-Cz,
O1-Cz, F4-T4, T4-P4, P4-O2, F3-T3, T3-P3, and P3O1) was continuously monitored at the patients bedside
beginning at the earliest opportunity after admission to
the ICU. This results in at least 1014 channels of EEG
being monitored and displayed and insures comprehensive geographic coverage of all brain regions. However,
the specificity of exact localization within these large
regions is, by necessity, somewhat restricted. The cEEG
was continuously displayed at the bedside, 24 h/day, for
moment-to-moment online observation by physicians
and nurses. A physician trained in interpretation of
EEG reviewed the ongoing EEG activity at the bedside
at least three times each day and additionally when cued
by the bedside nurse of suspicious EEG activity. Seizures were detected in one of three ways: Online identification of seizures by the neuro-ICU nurse or
neurointensivist, by the total power trend seizure detection method, or by detection during regularly scheduled
EEG segment review. The date and time of the seizure
and the clinical behavior noted by the bedside neuroICU nurse or neurointensivist were recorded. Others
have used video-EEG monitoring in the ICU for patients
with seizures and status epilepticus (Pandian et al.,
2004). In some cases, the addition of video information
is critical to the diagnosis and treatment of these patients.
857
Figures 13 show the typical setup and appearance of

EEG in the ICU. The cEEG is networked through our
hospital information technology system and is stored
for review on a large computer server. This server as well
as each EEG machine in the ICU is electronically accessible through the Internet. In this way, the EEG can be
assessed remotely by an expert physician reader and
important interpretation and treatment decisions can be
rendered in real time. An example of this server can be
accessed by contacting the author for access information.
Others have made use of limited number of electrode
arrays and only quantitative EEG (qEEG) and automated
seizure detection software. Some have suggested that
single-channel systems, such as the BIS monitor (Aspect
Medical, Newton, MA) or the SNAP monitor (Everest
Biomedical, Chesterfield, MO), can be used to monitor
patients with status epilepticus. However, most agree that
these limited quantitative methods are not appropriate for
monitoring seizure activity. Gotman and colleagues have
recently reported progress in creating an automated system using fuzzy logic and neural networks that can reliably imitate an experienced electroencephalographer,
which may decrease the time spent differentiating real
from artifact-induced abnormalities in qEEG trends
(Si et al., 1998). Several characteristic alterations in
qEEG trends have been repeatedly recognized as
markers of seizure activity. The finding of increases in
amplitude of total power and/or in the alpha/delta ratio
occurs reliably with high-amplitude seizure or repetitive
epileptiform spike discharges. However, similar indiscriminant changes in total power and/or alpha/delta ratio
occur with muscle activity or electrode artifacts. Reduction in the amplitude of total power occurs with burst
suppression and/or marked attenuation of the EEG.
Fig. 1. ICU continuous EEG (cEEG) set up and enlarged image of EEG and qEEG.
Fig. 2. ICU seizure.
Fig. 3. Example of raw EEG with compressed spectral array.
63.4. Findings of seizures and status epilepticus

in cEEG monitoring
Jordan (1995) reported an initial study of cEEG monitoring in a neuroscience ICU population of 124 patients.
The admission diagnosis of these patients included
stroke, intracerebral hemorrhage, seizures, metabolic
coma, brain tumors, and brain trauma. Overall seizures occurred in 35% of patients during the ICU
course, with over three-fourth demonstrating nonconvulsive seizures or nonconvulsive status epilepticus.
Seizures were not readily clinically diagnosed by
obvious motor manifestations and the diagnosis
required EEG evaluation. The duration of monitoring
was determined by the incidence of epileptiform
activity. The impact of cEEG on clinical decision
was determined using the available EEG information
in making the following decisions: (1) initiating or
modifying anticonvulsants, (2) obtaining a CT or MRI
scan, or (3) adjusting cerebral perfusion or mean arterial blood pressure. Overall, the cEEG was decisive in
these decisions in 51% of patients and made a significant contribution in the remaining 31% of patients.
Thus cEEG monitoring detected abnormal subclinical
pathophysiology that could be treated and contributed
directly to patient care in a large number (82%)
of patients. Similar reports of a high frequency of nonconvulsive seizures in neurocritical care patients had
been made with the incidence ranging from 11% (Litt
et al., 1994) to 55% (GrandMaison et al., 1991).
The incidence of nonconvulsive seizures after traumatic brain injury has been recently established. In a
seminal study, Vespa et al. (1999a) used cEEG monitoring in patients with moderate and severe traumatic brain
injury (Glasgow Coma Score 312). Monitoring was
performed for 710 days after injury from the time of
admission in the ICU. The EEG was displayed at the
bedside and the nurses instructed to identify electrographic generalized seizure activity. Electrographic seizures without clinical signs of convulsions were the
most common form of seizure (Fig. 2). Sudden
increases in total power have been seen in patients with
electrographic seizures and thus were used as a nonspecific marker of important EEG segments that required
review (Fig. 3). Using these methods, 22% of 91
patients were found to have seizures, with a majority
(57%) demonstrating nonconvulsive seizure activity.
Most seizures occurred within the initial week after
the brain injury and the incidence of seizures was neither affected by therapeutic phenytoin levels or by
the withdrawal of ethanol, nor by the type of brain
859
injury. It should be noted that only moderate-tosevere brain injury patients were monitored and that
the incidence of nonconsulsive seizures in patients
with mild brain injury is not well documented. The
absence of overt clinical signs of seizure activity was
similar to the observations of ongoing nonconvulsive
status epilepticus in other neurologic populations
reported by Young et al. (1996), GrandMaison et al.
(1991), and Scholtes et al. (1996). Patients who exhibited posttraumatic status epilepticus uniformly died
whereas patients with isolated seizures had no apparent difference in mortality rate. In a follow-up assessment of 315 patients with moderate-to-severe brain
injury, seizures were present in 27% of patients and
were a factor in increasing mortality.
The incidence of seizures after brain ischemia has
relied upon studies of clinical signs of seizure activity.
The incidence of clinically defined seizures after bland
ischemia stroke varies from 5% to 17% (Shinton et al.,
1988). The hospital-based studies demonstrated the
highest rate of poststroke seizures (Olsen et al.,
1987; Lancman et al., 1993). Indeed the incidence of
seizures after stroke increases with large territory
ischemic injury and with cardioembolic stroke. The
use of cEEG after ischemic stroke suggests that the
incidence of nonconvulsive seizures is much higher
than the previous studies, which are based on clinical
signs of seizures, would suggest. In the work of
Jordan (1995) using cEEG in 57 consecutive patients
admitted to the ICU with cerebral ischemia, 26% of
the patients had EEG-defined nonconvulsive seizures
during the period of monitoring. In a recent study
using cEEG, 6% of ischemic stroke patients demonstrated nonconvulsive seizure activity (Vespa et al.,
2003b). This occurred in the setting of malignant brain
edema but not in conjunction with hemicraniectomy
or specifically with arterial recanalization.
The incidence of seizures after intracerebral hemorrhage has been noted to be higher than after ischemic stroke. Immediate and early seizures occur in the
range of 2.818.7% with the frequency of status epilepticus between 1.1% and 2%. In a recent study by
Vespa et al. (2003b) the incidence of seizures after
intracerebral hemorrhage was 28% compared with
6% incidence in ischemic stroke. In this study, cEEG
was used to identify nonconvulsive seizures and most
likely is the primary reason why the incidence rate is
higher than in most previously cited studies. In a recent
report from the Columbia group, Claassen et al. (2004)
used similar cEEG methods and detected an 18% incidence rate of seizures in a mixed population of
860
P.M. VESPA
Table 1
The incidence of seizures using cEEG monitoring in a neurologic critical care illness population
ICU cEEG studies
Percentage with seizures
Principal diagnosis
Jordan, 1995
Vespa et al., 1999a
Vespa et al., 1999b
Vespa et al., 2003b
Claasen et al., 2004
Pandian et al., 2004
Young and Doig, 2005
124
91
300
65
204
105
55
35
22
21
28
17
42
20
Neurologic critical illness

Brain trauma
Subarachnoid hemorrhage/brain trauma
Primary intracerebral hemorrhage
Subarachnoid hemorrhage
hemorrhage and traumatic brain injury is double than

that of premonitoring natural history studies. Thus
cEEG detects nonconvulsive seizures and therefore
permits treatment.
intracerebral hemorrhage, subarachnoid hemorrhage,

and ischemic stroke. Again, Claassen et al. found that
most seizures occurred within the first week after
injury. The seizure rate in brain hemorrhage patients
may be higher due to the relative toxicity of intraparenchymal blood. Blood products, especially thrombin, are
known to be proepileptogenic and are often used for
experimental models of epilepsy (Engstrom et al.,
2001). The presence of blood may elicit seizures in
the injured brain since the rate of seizures is higher in
patients with nontraumatic and traumatic intraparenchymal bleeding compared with patients with bland
infarctions. It is clear that the use of cEEG monitoring
has resulted in an increased number of seizures being
detected, as reflected by the data in Table 1. Table 1
shows the incidence of seizures using cEEG monitoring in a neurologic critical care illness population.
The incidence rate of seizures after intracranial
63.5. ICU seizures add damage to the brain

Seizures occurring after brain injury or brain hemorrhage elicit a pathophysiological response at a time
when the brain is most vulnerable. These seizures lead
to additional secondary injury through a variety of
mechanisms. Posttraumatic seizures give rise to increased levels of extracellular glutamate (Vespa et al.,
1998). This increase in glutamate concentration exceeds
the concentration known to induce cell death and
swelling, and is persistent across many days after the
injury. The sustained excitotoxic injury associated with
seizures gives rise to lipid peroxidation and membrane
Table 2
Use of cEEG monitoring in treatment of status epilepticus
Step/drug
Clinical/drug
cEEG target
Potential complications
1. Diagnosis
Convulsions
Unresponsive/coma
Lorazepam/DPH
Epileptiform
Cellular injury, hemodynamic

instability, death
Hypotension, respiratory
Compromise
Hypotension
Hypotension
Hepatic injury
Systemic infections
2. Induction
6. Lightening
DPH
Continuous infusion
Midazolam
Propofol
Pentobarbital
Same step as in step 4 for
4872 h
Taper infusion
Suppress seizures
Beta activity
Beta/theta
Seizures/PLEDs
(0.52 mg/kg/h)
(50150 mg/kg/min)
(215 mg/kg/h)
Burst suppression
1020 s intervals
Beta/theta
7. Secondary prevention
High-dose AEDs
Beta/theta
3. Monitoring
4. Monitoring
5. Suppression
Hypotension, respiratory
Increased blood pressure
Rebound seizures on cEEG
Drugdrug interactions
Hepatic dysfunction
Abbreviations: DPH, phenytoin; PLEDs, pseudoperiodic lateralized epileptiform discharges; AEDs, antiepileptic drugs; cEEG, continuous
electroencephalography monitoring.
861
disruption (Vespa et al., 2002) with the resultant

increases in extracellular lipid breakdown product
(e.g., glycerol) levels. In addition, in the acute setting,
seizures increase glucose metabolism and lead to an
increase in the intracranial pressure. These mechanisms
most likely were responsible for the observation that seizures after brain hemorrhage lead to increased brain
swelling and midline shift (Vespa et al., 2003b). The latter study showed that compared with brain hemorrhage
patients without seizures, patients who had posthemorrhagic seizures had a dramatic increase in midline shift
and increased mortality. Indeed, seizures alone were
an independent factor in patient mortality, after
controlling for traditional prognostic factors (Table 2).
One can conclude that seizures led directly to worsened
brain edema and poor outcome and thus are not benign.
The role of seizures in the natural history of brain injuries
is not well established, but studies so far indicate that
seizures enhance and worsen the natural history of
brain injury. It is safe to conclude that seizures occur commonly after hemorrhagic brain injury, and in this context
are not benign, and must be detected and subsequently
treated.
63.6. Goal-directed treatment of seizures based
on cEEG
Patients with seizures after brain hemorrhage or cerebral injury are prone to increased brain edema, shift
and elevated intracranial pressure (Vespa et al.,
2003a). These events occur during the initial week after
injury, which overlaps with the peak incidence of posttraumatic seizures. Several principles of goal-directed
seizures can be derived from the current literature:
(1) cEEG should be started immediately upon presentation of the brain injury or hemorrhage. (2) Brain
hemorrhage patients (traumatic or nontraumatic) have
a higher risk than ischemic stroke patients and should
get priority for monitoring if resources are limited.
(3) Monitoring should continue for at least 5 days given
the temporal course of seizures. (4) The lack of clinical
seizure activity is not an indication to avoid cEEG since
the preponderance of seizures is nonconvulsive.
Treatment to avoid or limit the number of seizures
includes the use of long-acting anticonvulsants as well
as the use of continuous infusions of anticonvulsants
such as midazolam, propofol, and pentobarbital. cEEG
is a useful monitor for titration of these medications.
Anticonvulsants are typically quite effective at stopping the seizure activity when used as continuous
infusions.
Fig. 4. Example case of using continuous EEG (cEEG) for

the treatment of status epilepticus.
A stepwise approach to the treatment of status epilepticus is outlined in Table 2. This is a stepwise process in which the diagnosis and treatment of status
epilepticus progresses sequentially from the lowest
to highest steps, according to the target EEG. cEEG
is started as soon as possible upon diagnosis of seizures or suspected status epilepticus. For the nonmoving patient, cEEG will be needed to diagnose
862
nonconvulsive status epilepticus or complex partial

status epilepticus. Once the cEEG is started, it is then
the primary treatment target, and the indicated treatment
endpoint is outlined in steps 27 of Table 2. cEEG
is essential at all steps along the process, and intermittent EEG monitoring alone is not sufficient in
order to direct and modify the intensive care that
is required to suppress seizures. It should be noted
that steps 5 and 6 are frequently repeated in cases
of refractory status epilepticus, with the total duration of suppression phase lasting weeks to months
in some cases. Moreover, it is unclear whether complete suppression rather than burst suppression is
required or desired during step 5. The literature suggests that deeper suppression may be useful in
selected cases.
The corresponding changes in cEEG are shown in
Fig. 4. The sequence of changes from seizure to suppression to recovery/secondary prevention of new
seizures is shown in this figure.
63.7. Summary
cEEG monitoring is a powerful tool in the critical
care of patients with status epilepticus. It provides
the real-time assessment of intensive treatments to
stop seizures and management brain function during
a life-threatening illness. Much in the same way that
electrocardiographic monitoring is a required element of caring for the myocardial ischemia patient,
cEEG is a required element of care of the status
epilepticus patient.
References
Claassen, J, Mayer, SA, Kowalski, RG, Emerson, RG and
Hirsch, LJ (2004) Detection of electrographic seizures
with continuous EEG monitoring in critically ill
patients. Neurology, 62(10): 17431748.
Engstrom, ER, Hillered, L, Flink, R, Kihlstrom, L, Lindquist,
C, Nie, JX, Olsson, Y and Silander, HC (2001) Extracellular amino acid levels measured with intracerebral
microdialysis in the model of posttraumatic epilepsy
induced by intracortical iron injection. Epilepsy Res.,
43(2): 135144.
GrandMaison, F, Reiher, J and Laduke, CP (1991) Retrospective inventory of EEG abnormalities in partial status
elilepticus. Electroencephalogr. Clin. Neurophysiol., 79:
264270.
Jordan, KG (1995) Neurophysiologic monitoring in the
neuroscience intensive care unit. Neurol. Clin., 13:
579626.
P.M. VESPA
Kinnier Wilson, JV and Reynolds, EH (1990) Translation
and analysis of a cuneiform text forming part of a Babylonian treatise on epilepsy. Med Hist., 34: 185198.
Lancman, ME, Crolinstock, A, Morseini, J and Ganillo, R
(1993) Risk factors for developing seizures after a
stroke. Epilepsia, 34: 141143.
Litt, B, Dizon, L and Ryan, D (1994) Fatal nonconvulsive
status epilepticus in the elderly. Epilepsia, 35: 1017.
Lowenstein, DH, Bleck, T and MacDonald, RL (1999) Its
time to revise the definition of status epilepticus.
Epilepsia, 40(1): 120122.
Olsen, TS, Hogenhaven, H and Thage, O (1987) Epilepsy
after stroke. Neurology, 37: 12091211.
Pandian, JD, Cascino, GD, So, EL, Manno, E and Fulgham,
JR (2004) Digital video-electroencephalographic monitoring in the neurological-neurosurgical intensive care
unit: clinical features and outcome. Arch. Neurol.,
61(7): 10901094.
Scholtes, FB, Renier, WO and Meinardi, H (1996) Nonconvulsive status epilepticus: causes, treatment, and
outcome in 65 patients. J. Neurol. Neurosurg. Psychiatry, 61: 9395.
Shinton, RA, Gill, JS and Melnick, AK (1988) The frequency, characteristics, and prognosis of epileptic seizures at the onset of stroke. J. Neurol. Neurosurg.
Si, Y, Gotman, J, Pauspathy, A, Flanagan, D, Rosenblatt, B
and Gottesman, R (1998) An expert system for EEG monitoring in the pediatric intensive care unit. Electroencephalogr. Clin. Neurophysiol., 106: 488500.
Vespa, PM, Ronne-Engstrom, E, Smith, M, Hovda, DA, Shalmon, E, Martin, NA and Becker, DP (1998) Increase in
extracellular glutamate caused by reduced cerebral perfusion pressure and seizures after human traumatic brain
injury: a microdialysis study. J. Neurosurg., 89: 971982.
Vespa, PM, Nuwer, MR, Nenov, V, Ronne-Engstrom, E,
Hovda, DA, Martin, NA and Becker, DP (1999a)
Increased incidence and impact of nonconvulsive and
convulsive seizures after traumatic brain injury as
detected by continuous EEG in the intensive care unit.
J. Neurosurg., 91: 750760.
Vespa, PM, Nenov, V and Nuwer, MR (1999b) Continuous
EEG monitoring in the intensive care unit: early
findings and clinical efficacy. J. Clin. Neurophysiol.,
16(1): 113.
Vespa, P, Martin, NA, Nenov, V, Glenn, T, Bergsneider, M,
Kelly, D, Becker, DP and Hovda, DA (2002) Delayed
increase in extracellular glycerol with post-traumatic electrographic epileptic activity: support for the theory that seizures induce secondary injury. Acta Neurochir. Suppl., 81:
355357.
Vespa, P, McArthur, D, Glenn, T, OPhelan, K, Etchepare,
M, Kelly, D, Bergsneider, M, Martin, NA and Hovda,
DA (2003a) Persistently reduced levels of extracellular
glucose early after traumatic brain injury correlate with

poor outcome at six months: A microdialysis study.
J. Cereb. Blood Flow Metab., 23: 865877.
Vespa, P, OPhelan, K, Mirabelli, J, Shah, M, Starkman, S,
Kidwell, C, Saver, J, Nuwer, M, Frazee, J, Mcarthur, D
and Martin, NA (2003b) Acute seizures after intracerebral hemorrhage: a factor in progressive midline shift
and outcome. Neurology, 60: 14411446.
Walker, M, Cross, H, Smith, S, Young, C, Aicardi, J, Appleton, R, Aylett, S, Besag, F, Cock, H, Delorenzo, R, Drislane, F, Duncan, J, Ferrie, C, Fujikawa, D, Gray, W,
Kaplan, P, Koutroumanidis, M, ORegan, M, Plouin, P,
Sander, J, Scott, R, Shorvon, S, Treiman, D, Wasterlain,
863
C and Wieshmann, U (2005) Nonconvulsive status epilepticus: epilepsy research foundation workshop reports.
Epileptic Disord., 7(3): 253296.
Young, GB and Doig, GS (2005) Continuous EEG monitoring in comatose intensive care patients: epileptiform
activity in etiologically distinct groups. Neurocrit. Care,
2(1): 510.
Young, GB, Jordan, KG and Doig, GS (1996) An assessment of nonconvulsive seizures in the intensive care
unit using continuous EEG monitoring: an investigation
of variables associated with mortality. Neurology,
47(1): 8389.

M.R. Nuwer (Ed.)
864
CHAPTER 64
ICU EEG monitoring for vasospasm and other focal

cortical disorders
Jan Claassen and Lawrence J. Hirsch*
Division of Critical Care Neurology, Neurological Institute, Columbia University, New York, NY 10032, USA
64.1. Introduction
Recent technological developments have made digital,
continuous EEG monitoring (cEEG) widely available.
Primarily used to diagnose nonconvulsive seizures
and to monitor treatment of status epilepticus (see
Chapter 63 in this volume), this technique is increasingly used to detect focal brain abnormalities such as
delayed cerebral ischemia (DCI) from vasospasm,
recurrent or progressive ischemia in the stroke patient,
and any other acute brain event. A major hurdle to overcome is the labor intensiveness of interpreting cEEG.
Digital EEG data can be transformed into power spectra by fast Fourier transformation (FFT), creating a
large number of possible quantitative EEG (qEEG)
parameters. These can be displayed as numbers or
graphically as compressed spectral arrays (CSAs),
histograms, or staggered arrays (Bricolo et al., 1987;
Scheuer and Wilson, 2004; Hansen and Claassen,
2005). Subtle changes in these qEEG parameters may
alert the clinician to suspect a clinical change. With
the advent of powerful microprocessors, data processing of this type can now be performed in real time at
the patients bedside. qEEG analysis is readily available since most manufacturers have integrated it to
some extent into their software packages. This technique may be particularly useful in unmasking subtle
focal changes in the EEG that may go unnoticed in
the raw EEG. Although little human data exist at this
point, some studies suggest that intracranial recording
of cortical spreading depression (CSD) and periinfarct
depolarizations may also be a powerful tool in
Correspondence to: Lawrence J. Hirsch, MD, Department

of Neurology, Comprehensive Epilepsy Center, Neurological Institute, Columbia University, Box NI-135, 710 W.
168th Street, New York, NY 10032, USA.
Tel.: 1-212-305-0458; fax: 1-212-305-1450.
E-mail: ljh3@columbia.edu (L.J. Hirsch).
monitoring the injured brain and preventing secondary

injury (Strong et al., 2002; Fabricius et al., 2006).
Ideally these techniques are integrated into multimodal
brain monitoring of the severely brain-injured patient,
also including blood pressure, oxygen saturation, temperature, cerebral perfusion pressure (CPP), intracranial
pressure (ICP), microdialysis, transcranial Doppler
(TCD) sonography, and brain tissue oxygen tension
(Wartenberg and Mayer, 2005).
Daly and Markand (1990) extensively reviewed the
different EEG patterns that focal brain lesions may
cause. These include arrhythmic delta activity, intermittent rhythmic slow activity, or epileptiform discharges; all of these may be seen regionally (i.e., just
a few electrodes) or involving one or both hemispheres
(Daly and Markand, 1990).
64.2. Clinical applications
Currently the major clinical application of EEG monitoring in the intensive care unit (ICU) setting is the
detection of nonconvulsive seizures and monitoring
the treatment of status epilepticus. For many years
monitoring cerebral blood flow (CBF) with EEG was
exclusively done in the operating room, specifically
during carotid endarterectomy (Sharbrough et al.,
1973; Zampella et al., 1991), which is discussed in
detail in Chapter 57 in this volume. Increasingly, the
detection of ischemia in different clinical scenarios
has become of interest in the ICU. One particularly
promising application involves the detection of DCI
due to vasospasm after subarachnoid hemorrhage
(SAH) which will be discussed under Section 64.2.1.2.
Other less well-studied focal brain abnormalities
include recurrent stroke after acute ischemic stroke
(AIS) or transient ischemic attack (TIA) (Section
64.2.1.1), the detection of intracranial hemorrhage
(ICH) expansion (Section 64.2.2), and the documentation of ICP changes (Section 64.2.3).
64.2.1. Cerebral ischemia

Cortical layers 3 and 5, which are particularly sensitive
to oxygen deficits, contribute most to the generation of
electrical dipoles detected by the EEG (Jordan, 1999,
2004; Ebersole, 2003). These areas are selectively sensitive to hypoxia and ischemia (Courville, 1958). EEG
changes due to energy and ion pump failure occur
when CBF falls below 2530 ml/100 g/min, at a time
when therapeutic interventions might prevent permanent brain damage as infarction does not occur until
CBF falls below 18 ml/100 g/min (Astrup et al.,
1981; Sundt et al., 1981; Baron, 2001). Brain damage
may be reversible down to a CBF of 12 ml/100 g/min
(Table 1; Jordan, 2004). Mild hypoxia (CBF 25
35 ml/100 g/min) may result in subtle decreases in
the amplitude of fast activity (>13 Hz) (Nagata et al.,
1989; Jordan, 2004), whereas cerebral infarction or
increased ischemia usually results in polymorphic
delta and more pronounced attenuation of fast frequencies, including sleep spindles (Cohn et al., 1948;
Nuwer et al., 1987; Niedermeyer, 2005). These EEG
findings reflect disrupted metabolism (cerebral metabolic rate of oxygen) as seen with positron emission
tomography (Nagata et al., 1989) and abnormal CBF
as seen with Xenon-CT studies (Ingvar et al., 1976;
Tolonen and Sulg, 1981). Several studies of patients
with ischemic strokes have demonstrated a relationship between focal ischemia and qEEG monitoring
parameters such as topographic power maps (Nuwer
et al., 1987; Murri et al., 1998; Luu et al., 2001;
Fernandez-Bouzas et al., 2002). Additionally, EEG is
very sensitive for recovery and may demonstrate
recovery of brain function from reperfusion earlier
than the clinical exam (Jordan, 1999). These observations make the EEG a potentially very useful tool to
continuously monitor brain perfusion.
865
64.2.1.1. Acute ischemic stroke

In AIS the diagnosis is usually made under great time
constraints in the emergency room based on a combination of the clinical history, the neurological examination, and the head CT. AISs are best visualized
on diffusion- and perfusion-weighted MRI. Prior to
the advent of readily available MRIs there was a
role for EEG in supplementing the physical exam
and head CT to diagnose AIS (Jordan, 2004). Among
91 patients with AIS, 36% initially had normal CT
scans, and 48% (N 16) of these showed acute lateralized EEG abnormalities (MacDonell et al., 1988).
All 16 showed cortical infarctions on follow-up
CT scans corresponding to the EEG findings. EEG
was able to detect cortical infarcts with a much higher
sensitivity (76%) than subcortical infarcts (30%;
MacDonell et al., 1988). Among 17 of 20 AIS patients,
Nuwer et al. (1987) described a close topographic
correlation between infarct location and focal increases
in delta and decreases in alpha activity as measured by
frequency analysis and topographic mapping. Murri
et al. (1998) found that using quantitative analysis
of EEG (topographical maps based on absolute and
relative power in the different frequency spectra)
helped to localize focal hypoperfusion in AIS
patients. However, Luu et al. (2001) pointed out that
at least 64 channel recordings are necessary to
achieve an adequate EEG-based localization of acute
ischemia recording densities when compared to clinical and MRI standards. Finnigan et al. (2004)
demonstrated a close relationship between the
perfusion-weighted imaging (PWI) lesion volume
measured as the mean transit time on MRI and the
acute delta change index, a qEEG measure reflecting
the rate of change of average delta power in the scalp
EEG, of 11 patients admitted with AIS (see Figs. 1
and 2). Remarkably, the authors recorded EEGs in
Table 1
Relationship between ischemia, EEG change, and neuronal injury
CBF level (ml/100 g/min)
EEG change
Degree of neuronal injury
3570
2535
1825
1218
<810
Normal
Loss of fast (beta) frequencies
Slowing of background to 57 Hz
Slowing to 14 Hz delta
Suppression of all frequencies
No injury
Reversible
Reversible
Reversible
Neuronal death
Adapted from Jordan (2004), with permission.
866
J. CLAASSEN AND L.J. HIRSCH
Fig. 1. Correlation of early change in EEG focal delta power with outcome and imaging in acute ischemic change: example
with early decreasing delta power and excellent early recovery. (A and B) Axial and left lateral EEG scalp delta power
maps acquired 6.5 h after onset of symptoms. (C) Initial DWI (6 h); (D) initial MTT map; (E and F) axial and lateral delta
power maps at 13 h; (G) 15-h DWI scan; and (H) 30-day T2 MRI [adapted from Finnigan et al. (2004) with permission].
Fig. 2. Correlation of early change in EEG focal delta power with outcome and imaging in acute ischemic change: example
with early increasing delta power and poor outcome (died). (A and B) Axial and lateral delta power maps acquired 9 h after
onset of symptoms; (C) initial DWI (6 h); (D) initial MTT map; (E and F) axial and right lateral delta power maps at 17 h;
(G) 15-h DWI scan; and (H) 15-h T2 MRI [adapted from Finnigan et al. (2004) with permission].
all 11 patients within 7 and 16 h of symptom onset

(Finnigan et al., 2004). Despite these correlations
between early EEG findings and later clinical and
radiographic outcome, EEG currently has a minimal
role in diagnosing or localizing AIS given the availability of MRI except perhaps in cases of laminar
necrosis [in which MRI, including diffusionweighted imaging (DWI), may be negative or show
only subtle findings for days (McKinney et al.,
2004), see the case shown in Fig. 3], ischemia without infarction, or borderline blood flow.
More importantly, cEEG monitoring may be used
as a continuous monitor of brain perfusion to detect
progressive or recurrent ischemia (Niedermeyer,
2005). AIS patients with recanalization may have
reocclusion in up to 34% of cases (Alexandrov and
Grotta, 2002). Both patients with stroke and those
with TIA have an increased risk of early recurrent
ischemic stroke (Rothwell et al., 2005), which may
be picked up with cEEG monitoring. Despite theoretical advantages and some clinical evidence, today
only few institutions use cEEG monitoring for AIS
patients (Jordan, 2004). Jordan (1995) found that in
AIS patients treated with hypervolemic hypertensive
therapy (HHT), focal slowing in the raw EEG normalized once regional CBF (measured with Xenon-CTCBF imaging) was elevated above the ischemic
threshold. Wood et al. (1984) correlated qEEG (percentage of theta and delta activity over total activity)
with baseline and posttreatment CBF measured with
133-Xenon SPECT in patients with AIS in the medial
cerebral artery (MCA) territory during isovolemic
hemodilution. Improved background activity was seen
in response to increases in regional CBF and clinical
improvement. In patients with atherosclerotic AIS,
Suzuki et al. (1990) reported good correlation between
topographic maps based on qEEG parameters (percentage of time and amplitude obtained by the
waveform recognition method) and controlled hypertension and hypotension. Six of 11 patients (54%)
showed qEEG improvement during hypertension,
and 12 of 18 patients (66%) showed qEEG worsening
with hypotension (Suzuki et al., 1990). The acute delta
change index was found to reflect well the effects of IV
tPA in a patient with AIS (Finnigan et al., 2004). Van
Putten and Tavy (2004) found a close relationship
between the clinical condition of 21 AIS patients
using the NIH stroke scale (NIHSS) score and the
qEEG using the brain symmetry index. These authors
suggest that changes in the continuously assessed
brain symmetry index may alert clinicians to
867
reexamine the patient and detect a clinical change as

early as possible.
In addition to hypoperfusion, AIS patients are at
risk of sustaining a number of complications such as
seizures (1.815%; Jordan, 2004), cerebral edema
with increased ICP, ICHs, and side effects from treatment. Up to 42% of AIS patients deteriorate within
the first 24 h (Britton and Roden, 1985). IV tPA therapy has a 710% complication rate of intracerebral
hemorrhage (Quality Standards Subcommittee of the
American Academy of Neurology, 1996). These
patients are also at an increased risk of hypotension,
fever, hyperglycemia, and other systemic perturbations (Castillo et al., 2004). In animals, seizures may
increase infarct size (Camilo and Goldstein, 2004;
Williams et al., 2004), and epidemiological studies
have demonstrated a threefold increase in mortality
when status epilepticus complicated AIS, independent
of infarct size or initial severity (Waterhouse et al.,
1998). Although little data exist in AIS patients,
hemorrhages and ICP elevation may also be detected
by cEEG. Schneider and Jordan (2005) have suggested
that regional attenuation without delta (RAWOD) is a
distinctive early EEG pattern reflecting massive hemispheric infarction and should be considered to help
identify patients that may benefit or be excluded from
IV tPA (Schneider and Jordan, 2005).
Some preliminary human data suggest that CSDlike events documented on intracranial electrocorticographic recordings occur in patients with acute,
focal brain damage (Strong et al., 2002; Fabricius
et al., 2006; see Fig. 4). In a small study of 12
acutely brain-injured patients [7 with spontaneous
ICH, 5 with traumatic brain injury (TBI)], electrocorticographic activity was recorded from the
immediate vicinity of the injured cortex (Fabricius
et al., 2006) and CSD was detected in 6 of these
patients (4 of 5 TBI, 2 of 7 ICH patients). In two
of these patients they also observed periinfarct
depolarizations in electrically silent cortical tissues
at the margins of injury and hypothesized that these
may contribute further to tissue damage in acute
brain lesions. Animal data suggest that in models
of transient focal ischemia and reperfusion, similar
peri-infarct depolarizations contribute to secondary
injury such as delayed edema, intracranial hypertension, and hypoperfusion (Hartings et al., 2005).
If future studies confirm these findings, monitoring of CSD and periinfarct depolarizations with
electrocorticography may allow real-time detection
of potentially modifiable physiological processes
868
involved in early secondary injury in acutely braininjured patients.

Prognosis. A number of studies investigated the
ability of acute or subacute EEG data to predict
long-term outcome after AIS (Kayser-Gatchalian and
Neundorfer, 1980; Ahmed, 1988; Cillessen et al.,
1994; Cuspineda et al., 2003; Finnigan et al., 2004).
Poor outcome was found to correlate with contralateral
(Ahmed, 1988) and generalized background slowing
(Kayser-Gatchalian and Neundorfer, 1980). Among
55 patients with supratentorial ischemia, early EEG
was able to differentiate between poor and good functional outcome more accurately than the admission
functional impairment (measured by Rankin score;
Cillessen et al., 1994). EEG findings that were found
to relate to poor outcome included continuous polymorphic delta with depression of alpha or beta activity
and the degree of background suppression, whereas
good outcome was seen with the absence of slow
activity with minimal decrease in background frequencies, and intermittent theta-delta activity with
slight asymmetry of background activity. Both raw
EEG and qEEG parameters were similarly related

to outcome measures in this study (Cillessen et al.,
1994).
More recent studies have used qEEG parameters to
predict outcome after AIS. Among 28 patients with
acute MCA stroke, qEEG parameters (Z-values of
absolute power from the four classic frequency bands
calculated from serially obtained EEGs within 72 h
of stroke onset) and a clinical examination scale
(Canadian Neurological Scale) were used to predict
functional disability (Cuspineda et al., 2003). Remarkably, these qEEG parameters were more often able to
correctly predict residual functional disability than
the clinical scale. Similarly, among 11 AIS patients,
a change in the power of slow frequencies (acute delta
change index) based on EEGs obtained within 16 h of
stroke onset was highly related to 30-day NIHSS score
(Finnigan et al., 2004). This qEEG parameter correlated as closely with clinical outcome as the lesion
size on perfusion-weighted MRI scans, and (surprisingly) correlated more closely with outcome than
DWI.
Fig. 3. Seventy-year-old man with persistent left MCA syndrome after left carotid endarterectomy, negative MRI with DWI
2 days in a row, but markedly asymmetric EEG, most likely due to laminar necrosis from intraoperative or perioperative
ischemia. Postoperative angiogram, MR perfusion, and transcranial Doppler studies showed no occlusion, major stenosis,
or delayed perfusion. A: MRI with DWI obtained about 28 h postoperatively, with no major abnormalities. B: MRI with
FLAIR, also at 28 h postoperatively and with no major abnormalities.
(Continued)
869
Fig. 3. Contd. C: EEG from the same day (24 h postoperatively) showing a clear asymmetry of faster activity consistent with
diffuse cortical dysfunction such as from ischemia/infarct. This abnormality was persistent throughout his recording beginning at
the time of EEG hookup less than 24 h postoperatively. No other imaging or vascular study showed an abnormality to explain his
clinical syndrome in the first several days. D: MRI with DWI 9 days postoperatively, now showing widespread abnormalities.
E: MRI with FLAIR, also 9 days postoperatively and showing widespread increased signal in the left hemisphere cortex.
870
Fig. 4. Cortical spreading depression (CSD) and periinfarct depolarizations (PID) in acutely injured brain. Three-hour recording of ECoG from Channels A to D. The three sets of traces represent the same period: upper four traces show the unfiltered
signal (full scale 3 mV), middle four traces show the integrated signal (full scale 100 mV s), and lower four traces show the
power of the 0.570 Hz band of the signal (full scale 0.05 mV2). Baseline ECoG activity showed burst-suppression pattern: 2 s
bursts and 1030 s suppressions, amplitude 3001000 mV. Initially, a CSD spreading from Channel A to D depressed this
ECoG activity for 3040 min. The CSD was accompanied by slow potential changes (SPCs) and spread from Channels B
to D at a velocity of 23 mm/min (thin arrows). After 1 h another CSD accompanied by SPCs spread from D to A (thick
arrows). This time, the ECoG activity did not recover. After 29 min, SPCs spread from channel D to A with exactly the same
time sequence and shape as detected during the last CSD. The ECoG remained depressed indicating compromised metabolism.
The event was therefore classified as a PID. Two stereotyped PIDs followed after intervals of 32 and 39 min. Just before the
last of these PIDs slight recovery of ECoG in Channel D was noticed (lower right). During 5 h a total of 21 stereotyped PIDs
were recorded. These PIDs were either similar to those in the figure or spread in the direction A to D in a second stereotyped
pattern [adapted from Fabricius et al. (2006) with permission of the Oxford University Press].
64.2.1.2. DCI from vasospasm after SAH

Subarachnoid blood will result in diffuse background
slowing in the theta and delta range with the amplitude
and frequency relating to the degree of impaired consciousness (Daly and Markand, 1990), and will also
frequently lead to a disruption of the posterior dominant (alpha) rhythm (Niedermeyer, 2005). Focal or
lateralized slowing, usually ipsilateral to the location
of the ruptured aneurysm, appears with parenchymal
extension of the hemorrhage (Roseman et al., 1951).
In the pre-CT scan era, research focused on using focal
EEG findings to help in localizing the aneurysm
location in these patients (Roseman et al., 1951). With

the advent of readily available CT scanning and angiography, this objective has become obsolete.
Among 116 consecutive SAH patients that underwent cEEG monitoring at our center, 16% had some
kind of periodic epileptiform discharge while undergoing cEEG monitoring (lateralized in 11%; generalized in 5%; Claassen et al., 2006). Nonconvulsive
seizures were seen in 6% and nonconvulsive status epilepticus in 4% of cases. The EEG was nonreactive
to external stimuli in 14%, showed no state changes
in 14%, and stage II sleep transients were absent in
85%. Other EEG findings included frontal rhythmic

delta activity (FRDA; 8%) and stimulus-induced
rhythmic, periodic, or ictal discharges (SIRPIDs; 8%;
Hirsch et al., 2004; Claassen et al., 2006).
DCI is defined as clinical deterioration and/or
infarction due to vasospasm after SAH and is one of
the major in-hospital complications of this disease.
Angiographic vasospasm is detected in 5070% of
patients with SAH (Weir et al., 1978). DCI occurs
in 1946% of SAH patients (Hijdra et al., 1988;
Murayama et al., 1997; Charpentier et al., 1999; Hop
et al., 1999; Qureshi et al., 2000; Claassen et al.,
2001) and may be seen in up to 54% of poor grade
SAH patients (HuntHess grade 4 or 5, i.e., stupor or
coma) (Kirmani et al., 2002). Clinically silent infarction accounts for approximately one-fourth of patients
with DCI (Claassen et al., 2001; Shimoda et al., 2001)
and is associated with poor clinical grade and poor outcome (Claassen et al., 2001; Kirmani et al., 2002).
HHT and angioplasty are used to treat vasospam and
may potentially prevent infarction if started early
enough. Therefore, timely diagnosis is of crucial
importance in these patients.
The gold standard to diagnose vasospasm in a patient
with DCI is cerebral angiography which cannot easily
be performed frequently or continuously, cannot be performed onsite in the ICU, has a low but significant risk
associated with it, and is rather expensive. Currently
serial clinical exams, which are particularly limited in
patients with altered mental status, and TCD examinations, which are often inaccurate (Grosset et al., 1993),
are used to alert physicians that a patient may have
DCI and angiography may be warranted. These techniques are only intermittent whereas cEEG can be utilized
continuously. Suspected ischemia in the cEEG would
ideally lead to rapid examination and confirmatory testing such as with cerebral angiography. Similarly, the
EEG could be used to help guide therapy, such as by
monitoring the functional effect of blood pressure
manipulations or changes in position (head elevation,
Trendelenburg, etc.).
EEG patterns, such as broad, repetitive slow waves,
termed axial bursts, are highly correlated with clinical or angiographic evidence of vasospasm (up to 97%
of the time) (Rivierez et al., 1991). In an effort to make
this technique both more sensitive for detecting vasospasm and more practical, quantitative analysis of
cEEG has been used to detect DCI due to vasospasm
in SAH patients (Labar et al., 1991; Vespa et al.,
1997; Claassen et al., 2004). With the advent of powerful microprocessors, data processing of this type can
871
now be performed in real time at the patients bedside.

The qEEG parameter that best correlates with clinically significant ischemia is controversial but most
authors agree that a ratio of fast over slow activity
(e.g., alpha over delta activity) or variability of
such activity (e.g., relative alpha variability) is the
most effective approach (Vespa et al., 1997; Claassen
et al., 2004; see Fig. 5). A number of qEEG parameters
have been shown to correlate with DCI or angiographic vasospasm: trend analysis of total power
(130 Hz) (Labar et al., 1991), variability of relative
alpha (visual scoring of tracings of 614 Hz/120 Hz)
(Vespa et al., 1997), and poststimulation alpha/delta
ratio (PSADR, 813 Hz/14 Hz) (Claassen et al., 2004).
These qEEG parameters may detect changes days prior
to clinical changes (Labar et al., 1991; Vespa et al.,
1997). Among primarily good grade SAH patients,
changes in two-channel qEEG preceded clinical symptoms suggesting DCI in 4 of 11 cases (Labar et al,
1991). Vespa et al. (1997) found that a decrease in the
visual scoring of the alpha variability preceded TCD
or angiographic detection of vasospasm by at least
2 days in 10 of 14 SAH patients. In this study, qEEG
was 100% sensitive for angiographically defined vasospasm (19 of 19) (Vespa et al., 1997). Importantly, all
of these studies found that focal ischemia sometimes
resulted in global or bilateral changes in the EEG
(Vespa et al., 1997; Claassen et al., 2004).
cEEG monitoring provided independent prognostic
information in poor grade SAH patients after
controlling for age, clinical exam (HuntHess grade),
and the presence of intraventricular hemorrhage
(IVH) on admission CT scan (Claassen et al., 2006).
Poor outcome (modified Rankin score 46, i.e., dead
or severely disabled) was associated with the absence
of sleep architecture (OR 4.3, 95%-CI 1.117.2) and
the presence of periodic lateralized epileptiform
discharges (PLEDs; OR 18.8, 95%-CI 1.6214.6).
In addition, all patients with absent EEG reactivity
(N 8), or with generalized periodic epileptiform
discharges (N 12) or bilateral-independent PLEDs
(N 5), and 92% (11 of 12) of patients with nonconvulsive status epilepticus had poor outcome (Claassen
et al., 2006). In an earlier study from the same center
by Dennis et al. (2002), all eight SAH patients that
were found to have NCSE on cEEG monitoring died.
64.2.2. Intracerebral hemorrhage
Depending on the ICH location different patterns
have been observed in the raw EEG. In deep capsular
872
1.0
Left
anterior
0.5
Right
0.5
anterior
1.0
Left
posterior
0.5
1.0
Right
posterior
0.5
14
GCS
10
7
Day 4
Day 6
Day 7
Day 8
Fig. 5. Detecting delayed cerebral ischemia (DCI) from vasospasm after subarachnoid hemorrhage (SAH). Alpha/delta
ratio (ADR) calculated every 15 min and glasgow coma score (GCS) shown for days 68 of continuous EEG (cEEG) monitoring. Fifty-seven-year-old woman admitted for acute SAH (admission HuntHess grade 4) from a right posterior communicating aneurysm. Admission angiography did not show vasospasm. The aneurysm was clipped on SAH day 2. No infarcts
were seen on postoperative CT. Postoperatively she had a GCS of 14. cEEG monitoring was performed from SAH days
38. The ADR progressively decreased after day 6, particularly in the right anterior region (thick vertical gray arrow), to
settle into a steady trough level later that night, reflecting loss of fast frequencies, and increased slowing over the right
hemisphere in the raw cEEG. On SAH day 6, flow velocities in the right MCA were marginally elevated (144 cm/s), but
the patient remained clinically stable with HHT. On day 7, the GCS dropped from 14 to 12 and a CT scan showed a right
internal capsule and hypothalamic infarction. Angiography demonstrated severe distal right MCA and left vertebral artery
spasm; however, due to the marked tortuosity of the parent vessels and the location of vasospasm, a decision was made not
to perform angioplasty, but to infuse verapamil and papaverine. This resulted in a marked but transient increase of the right
anterior and posterior alpha/delta ratios (shaded area). Later that day the patient further deteriorated clinically to a GCS of
7, with a new onset left hemiparesis, and died on SAH day 9 from widespread infarction due to vasospasm [adapted from
Claassen et al. (2004) with permission from Elsevier].
hemorrhages, delta activity is seen over the affected

hemisphere, at times occurring in rhythmic runs of
moderate amplitude (Hirose et al., 1981; Niedermeyer,
2005). Thalamic hemorrhages may lead to ipsilateral
delta activity, a reduction or enhancement of the alpha
rhythm depending on the precise location within the
thalamus, and lack of sleep spindles (Jasper and Van
Buren, 1953; Niedermeyer, 2005). Bleeds located in
the mesencephalon may cause diffuse theta activity
(Niedermeyer, 2005), those in the lower brainstem
diffuse attenuation or an unreactive but preserved
posterior dominant rhythm (Loeb et al., 1959;
Niedermeyer, 2005). Though theoretically feasible,
the ability of EEG to detect rebleeding after SAH is
increasingly difficult to study since it occurs in only
approximately 6.9% in more recent series (Naidech
et al., 2005).
Electrographic seizures are documented in 18% to
28% on cEEG monitoring of patients with nontraumatic ICH (Vespa et al., 2003; Claassen et al.,
2007). During cEEG the vast majority of seizures
are only noticed on cEEG and not clinically detectable (Claassen et al., 2007). Electrographic seizures
were associated with increasing midline shift (p < 0.03;
Vespa et al., 2003) and twice as common (31% vs.
14%; OR 9.5, 95%-CI 1.753.8) in patients with
expanding hemorrhages (growing by more than
30% on the 24 hour follow-up CT; Claassen et al.,
2007). This may indicate that a more prolonged and
active bleeding process may serve as at trigger of seizure activity after ICH, or that seizures lead to additional bleedings perhaps via increased blood flow
known to accompany focal seizures. There was a trend
worse outcome (p < 0.06, Vespa et al., 2003; and
45% vs. 20%, univariate only p = 0.03, Claassen)
and a significant worsening in NIH stroke scale
scores in patients with electrographic seizures (Vespa
et al., 2003). Periodic epileptiform discharges were
more frequently seen in hemorrhages closer to the
cortex and were independently associated with worse
outcome (Claassen et al., 2007).
64.2.3. Intracranial pressure
Although not strictly a focal disorder, we include a
discussion of the EEG effects of ICP here as it is
directly relevant to the care of the same patients
critically ill patients with acute brain injuries and
it is an example of the many types of acute unanticipated cerebral events that can be readily detected in
real time with cEEG.
873
Relatively few studies have explored the relationship

between EEG patterns and ICP. EEG changes were not
seen in monkeys during experimental gradual ICP rises
at levels approaching mean systemic arterial pressure
(Langfitt et al., 1966). In humans, no specific EEG pattern has been reliably correlated with mean ICP values
(Munari and Calbucci, 1981; Chiappa and Ropper,
1984). During phases without pressure waves and stable
elevated ICP, the EEG contains regular high-voltage
slow waves, while an alternating EEG is seen in those
patients with Lundberg B waves (which typically occur
at 0.52 waves/min). In six brain-injured patients, EEG
activity (measured as oscillations of spectral edge frequency at a mean frequency of 26 9 MHz) and CBF
fluctuations (measured by TCD as mean flow velocity
in the MCA) were closely correlated with intracranial
Lundberg B waves as a measure of ICP fluctuations
(Lescot et al., 2005). It appears likely that EEG activity
is not reliably influenced by increased ICP until very
high levels are reached and CPP and CBF are compromised (Claassen and Hansen, 2005). In fact, in patients
with benign intracranial hypertension the EEG and neurological function are generally normal (Sidell and
Daly, 1961; Daly and Markand, 1990).
As mentioned above, cerebral edema may be associated with midline shift and by inference related
to electrographic seizures. In a small series of
10 patients with space-occupying lesions Fernandez-Bouzas et al. (1997) found current source densities of the theta frequency (based on Z-values
calculated between patient data and normals) to be
related to the edema volume. Current source density
maps represented the site of the lesion and of the
edema better than other qEEG parameters (e.g., voltage estimates; Fernandez-Bouzas et al., 1997).
FIRDA, although being encountered in a number of
conditions and nonspecific, is often seen with dilation of the third ventricle as in aqueductal stenosis
and in ICP dysregulation. In patients with SAH, the
appearance of FIRDA often parallels the development of hydrocephalus, suggesting the need for additional CSF drainage (Riemer et al., 1998). We have
encountered at least one case where cEEG detected
a clinically important rise in ICP (due to lumbar
drain failure) prior to clinical recognition (see Fig. 6).
64.2.4. Others
Focal EEG findings are sometimes helpful by suggesting specific pathological processes, such as PLEDs in
herpes encephalitis (Daly and Markand, 1990).
874
Fig. 6. Elevated intracranial pressure (ICP) detected by continuous EEG (cEEG). Fifty-six-year-old stuporous woman with
severe hepatic cirrhosis and cryptococcal meningitis. Lumbar drain was placed to help manage raised ICP. A: Baseline
abnormal EEG at 4:09 p.m. B: EEG at 4:18 p.m. showing new diffuse attenuation.
(Continued)
875
Fig. 6. Contd. C: EEG at 4:27 p.m. showing progressive diffuse attenuation that was unreactive to stimulation. This was
noted by the reading EEG fellow and the clinical team was notified. Subsequent urgent exam triggered by the EEG findings
showed worsened alertness and rising blood pressure. After adjusting her lumbar drain (no CSF had been obtained for a few
hours), BP quickly normalized, and EEG and exam improved gradually. D: EEG at 10:13 p.m. showing some return of prior
EEG pattern, but in a discontinuous pattern. Now reactive to stimulation.
(Continued)
876
Fig. 6. Contd. E: EEG at 01:35 a.m. showing return to continuous activity, starting to approach the baseline EEG pattern.
64.3. Limitations of EEG for the detection of

focal CNS abnormalities
Some of the limitations of cEEG monitoring in
recording generalized abnormalities are less significant when focal abnormalities are to be recorded.
Generalized changes are often difficult to differentiate from state changes and medication effects. Since
these usually affect the entire brain to an equal
extent, focal changes can often still be visualized.
A major limitation in the setting of ischemia and
stroke is the inability of EEG to detect most small
deep (lacunar) infarcts, and the possibility of missing
small but clinically significant cortical events.
Generally a major drawback of this technique is the
labor intensiveness of getting patients connected to
cEEG, maintaining good recordings, and interpreting
large amounts of data. As discussed above, qEEG
parameters have the potential to significantly cut down
on reading time of cEEG files. Importantly, qEEG
should never be interpreted in isolation but should
always be seen in the context of the underlying raw
EEG. Substantial changes of a qEEG parameter may be
caused by underlying artifact alone or by noncerebral
factors such as scalp swelling. Similar to raw EEG
reading, interpretation of qEEG parameters should not

be attempted without proper training in electroencephalography. Unfortunately it is unrealistic to expect 24-h
coverage by an electroencephalographer in all neurological ICUs. Possibly, well-trained ICU staff or specially trained technicians akin to cardiac telemetry
technicians could screen the qEEG and raw EEG, with
definitive interpretation available remotely via a webbased link to the electroencephalographer. Remote
access to real-time EEG is quite feasible (Scheuer and
Wilson, 2004), and we have been using this routinely
for the past few years. qEEG and verified underlying
EEG changes would then prompt further testing or therapeutic interventions.
64.4. Impact of cEEG monitoring
In the future, mulimodality monitoring should incorporate cEEG monitoring and continuously generate a
number of prespecified qEEG parameters specifically
tailored to the individual patient (e.g., alpha/delta ratio
to detect ischemia). These brain monitoring parameters would generate an alarm once certain thresholds would be passed and would result in a patient
evaluation by the nurse or physician. If clinical
suspicion corroborates the monitoring alarm (i.e., it is

not a false positive), a more definitive test would
be ordered (e.g., MRI or cerebral angiogram). cEEG
may also help by assessing response to treatment and
prognosticating.
Ultimately, routine use of cEEG will lead to
immediate real-time detection of most major cerebral
events that occur in high-risk inpatients. As acute
treatments for ischemia and hemorrhage continue to
evolve (they were nonexistent a decade ago), realtime cEEG monitoring has the potential to lead to
substantial improvement in neurological outcome
and significant public health benefit. As we continue
to advance in neurocritical care, we must not forget
the older technique of EEG as it remains the only
convenient means of continuous real-time functional
monitoring of brain function. The onus is on the neurology, clinical neurophysiology, and neurocritical
care communities to collaborate and maximize the
use of these techniques.
References
Ahmed, I (1988) Predictive value of the electroencephalogram in acute hemispheric lesions. Clin. Electroencephalogr., 19: 205209.
Alexandrov, AV and Grotta, JC (2002) Arterial reocclusion
in stroke patients treated with intravenous tissue plasminogen activator. Neurology, 59: 862867.
cerebral ischemia the ischemic penumbra. Stroke, 12:
723725.
Baron, JC (2001) Perfusion thresholds in human cerebral
ischemia: historical perspective and therapeutic implications. Cerebrovasc. Dis., 11: 28.
Bricolo, A, Turazzi, S, Faccioli, F, Odonnizzi, F, Scirratta,
G and Erculiani, P (1987) Clinical application of compressed spectral array in long-term EEG monitoring of
comatose patients. Electroencephalogr. Clin. Neurophysiol., 45: 211225.
Britton, M and Roden, A (1985) Progression of stroke after
arrival at hospital. Stroke, 16: 629632.
Camilo, O and Goldstein, LB (2004) Seizures and epilepsy
after ischemic stroke. Stroke, 35: 17691775.
Castillo, J, Leira, R, Garcia, MM, Serena, J, Blanco, M and
Davalos, A (2004) Blood pressure decrease during the
acute phase of ischemic stroke is associated with brain
injury and poor stroke outcome. Stroke, 35: 520526.
Charpentier, C, Audibert, G, Guillemin, F, Civit, T, Ducrocq,
X, Bracard, S, Hepner, H, Picard, L and Laxenaire, MC
(1999) Multivariate analysis of predictors of cerebral
vasospasm occurrence after aneurysmal subarachnoid
hemorrhage. Stroke, 30: 14021408.
877
Chiappa, KH and Ropper, AH (1984) Long-term electrophysiologic monitoring of patients in the neurology
intensive care unit. Semin. Neurol., 4: 469479.
Cillessen, JP, Van Huffelen, AC, Kappelle, LJ, Algra, A
and Van Gijn, J (1994) Electroencephalography
improves the prediction of functional outcome in the
acute stage of cerebral ischemia. Stroke, 25: 19681972.
Claassen, J, Bernardini, GL, Kreiter, K, Bates, J, Du, YE,
Copeland, D, Connolly, ES and Mayer, SA (2001)
Effect of cisternal and ventricular blood on risk of
delayed cerebral ischemia after subarachnoid hemorrhage: the Fisher Scale revisited. Stroke, 32:
20122020.
Claassen, J, Hirsch, LJ, Frontera, JA, Fernandez, A,
Schmidt, M, Kapinos, G, Wittman, J, Connolly, ES,
Emerson, RG and Mayer, SA (2006) Prognostic significance of continuous EEG monitoring in patients with
poor-grade subarachnoid hemorrhage. Neurocrit. Care,
4: 103112.
Claassen, J, Hirsch, LJ, Kreiter, KT, Du, EY, Connolly, ES,
Emerson, RG and Mayer, SA (2004) Quantitative continuous EEG for detecting delayed cerebral ischemia
in patients with poor-grade subarachnoid hemorrhage.
Claassen, J, Jette, N, Chum, F, Green, R, Schmidt, M, Choi,
H, Jirsch, J, Frontera, JA, Connolly, ES, Emerson, RG,
Mayer, SA and Hirsch, LJ (2007) Electrographic seizures and periodic discharges after intracerebral hemorrhage. Neurology, 69: 13561365.
Cohn, HR, Raines, RG, Mulder, DW and Neumann, MH
(1948) Cerebral vascular lesions: electroencephalographic and neuropathologic correlations. Arch.
Neurol., 60: 163181.
Courville, CB (1958) Etiology and pathogenesis of laminar
cortical necrosis; its significance in evaluation of
uniform cortical atrophies of early life. Arch. Neurol.
Cuspineda, E, Machado, C, Aubert, E, Galan, L, Llopis, F
and Avila, Y (2003) Predicting outcome in acute stroke:
a comparison between QEEG and the Canadian Neurological Scale. Clin. Electroencephalogr., 34: 14.
Daly, DD and Markand, ON (1990) Focal brain lesions. In:
DD Daly and TA Pedley (Eds.), Current Practice of
Clinical Electroencephalography. Raven Press, Ltd.,
New York, 2nd ed., pp. 335370.
Dennis, LJ, Claassen, J, Hirsch, LJ, Emerson, RG, Connolly, ES and Mayer, SA (2002) Nonconvulsive status
epilepticus after subarachnoid hemorrhage. Neurosurgery, 51: 11361143.
Ebersole, J (2003) Cortical generators and EEG voltage
fields. In: JS Ebersole and TM Pedley (Eds.), Current
Practice of Clinical Electroencephalography. Lippincott,
Williams & Wilkins, New York, 3rd ed., pp. 1231.
Fabricius, M, Fuhr, S, Bhatia, R, Boutelle, M, Hashemi, P,
Strong, AJ and Lauritzen, M (2006) Cortical spreading
878
depression and peri-infarct depolarization in acutely
injured human cerebral cortex. Brain, 129: 778790.
Fernandez-Bouzas, A, Harmony, T, Marosi, E, Fernandez,
T, Silva, J, Rodriguez, M, Bernal, J, Reyes, A and
Casian, G (1997) Evolution of cerebral edema and its
relationship with power in the theta band. Electroencephalogr. Clin Neurophysiol., 102(4): 279285.
Fernandez-Bouzas, A, Harmony, T, Fernandez, T, Aubert, E,
Ricardo-Garcell, J, Valdes, P, Bosch, J, Casian, G and
Sanchez-Conde, R (2002) Sources of abnormal EEG
activity in spontaneous intracerebral haemorrhage. Clin.
Electroencephalogr., 33: 7076.
Finnigan, SP, Rose, SE, Walsh, M, Griffin, M, Janke, AL,
McMahon, KL, Gillies, R, Strudwick, MW, Pettigrew,
CM, Semple, J, Brown, J, Brown, P and Chalk, JB
(2004) Correlation of quantitative EEG in acute ischemic stroke with 30-day NIHSS score: comparison with
diffusion and perfusion MRI. Stroke, 35: 899903.
Grosset, DG, Straiton, J, McDonald, I, Cockburn, M and Bullock, R (1993) Use of transcranial Doppler sonography to
predict development of a delayed ischemic deficit after
subarachnoid hemorrhage. J. Neurosurg., 78: 183187.
Hansen, HC and Claassen, J (2005) EEG and evoked potentials in neuroanesthesia, intraoperative neurological monitoring, and neurointensive care. In: E Niedermeyer and
F Lopes Da Silva (Eds.), Electroencephalography: Basic
Principles, Clinical Applications, and Related Fields.
Lippincott, Williams & Wilkins, Baltimore, 5th ed.,
pp. 11371164.
Hartings, JA, Tortella, FC and Rolli, ML (2005) AC electrocorticographic correlates of peri-infarct depolarizations
during transient focal ischemia and reperfusion. J. Cereb.
Blood Flow Metab., 26: 696707.
Hijdra, A, Van Gijn, J, Nagelkerke, NJ, Vermeulen, M and
Van Crevel, H (1988) Prediction of delayed cerebral
ischemia, rebleeding, and outcome after aneurysmal subarachnoid hemorrhage. Stroke, 19: 12501256.
Hirose, G, Saeki, M, Kosoegawa, H, Takado, M, Yamamoto, T and Tada, A (1981) Delta waves in the EEGs
of patients with intracerebral hemorrhage. Arch.
Neurol., 38: 170175.
Hirsch, LJ, Claassen, J, Mayer, SA and Emerson, RG
(2004) Stimulus-induced rhythmic, periodic, or ictal
discharges (SIRPIDs): a common EEG phenomenon in
the critically ill. Epilepsia, 45: 109123.
Hop, JW, Rinkel, GJ, Algra, A and Van Gijn, J (1999) Initial loss of consciousness and risk of delayed cerebral
ischemia after aneurysmal subarachnoid hemorrhage.
Stroke, 30: 22682271.
Ingvar, DH, Sjolund, B and Ardo, A (1976) Correlation
between dominant EEG frequency, cerebral oxygen
uptake and blood flow. Electroencephalogr. Clin.
Jasper, HH and Van Buren, J (1953) Interrelationship
between cortex and subcortical structures: clinical

electroencephalographic studies. Electroencephalogr.
Jordan, KG (1995) Neurophysiologic monitoring in the
neuroscience intensive care unit. Neurol. Clin., 13:
579626.
Jordan, KG (1999) Continuous EEG monitoring in the neuroscience intensive care unit and emergency department.
Jordan, KG (2004) Emergency EEG and continuous EEG
monitoring in acute ischemic stroke. J. Clin. Neurophysiol., 21: 341352.
Kayser-Gatchalian, MC and Neundorfer, B (1980) The
prognostic value of EEG in ischaemic cerebral
insults. Electroencephalogr. Clin. Neurophysiol., 49:
608617.
Kirmani, JF, Qureshi, AI, Hanel, RA, Siddiqui, AM,
Safdar, A and Yahia, AM (2002) Silent cerebral infarction in poor-grade patients with subarachnoid hemorrhage. Neurology, 58(7): 159.
Labar, DR, Fisch, BJ, Pedley, TA, Fink, ME and Solomon,
RA (1991) Quantitative EEG monitoring for patients
with subarachnoid hemorrhage. Electroencephalogr.
Langfitt, TW, Tannanbaum, HM, Kassell, NF and Zaren, H
(1966) Acute intracranial hypertension, cerebral blood
flow, and the EEG. Electroencephalogr. Clin. Neurophysiol., 20: 139148.
Lescot, T, Naccache, L, Bonnet, MP, Abdennour, L, Coriat,
P and Puybasset, L (2005) The relationship of intracranial pressure Lundberg waves to electroencephalograph
fluctuations in patients with severe head trauma. Acta
Neurochir. (Wien.), 147: 125129.
Loeb, C, Rosadini, G and Poggio, GF (1959) Electroencephalograms during coma; normal and borderline
records in 5 patients. Neurology, 9: 610618.
Luu, P, Tucker, DM, Englander, R, Lockfeld, A, Lutsep, H
and Oken, B (2001) Localizing acute stroke-related
EEG changes: assessing the effects of spatial undersampling. J. Clin. Neurophysiol., 18: 302317.
MacDonell, RA, Donnan, GA, Bladin, PF, Berkovic, SF
and Wriedt, CH (1988) The electroencephalogram and
acute ischemic stroke. Distinguishing cortical from
lacunar infarction. Arch. Neurol., 45: 520524.
McKinney, AM, Teksam, M, Felice, R, Casey, SO, Cranford, R, Truwit, CL and Kieffer, S (2004) Diffusionweighted imaging in the setting of diffuse cortical laminar
necrosis and hypoxic-ischemic encephalopathy. AJNR
Am. J. Neuroradiol., 25: 16591665.
Munari, C and Calbucci, F (1981) Correlations between intracranial pressure and EEG during coma and sleep. Electroencephalogr. Clin. Neurophysiol., 51: 170176.
Murayama, Y, Malisch, T, Guglielmi, G, Mawad, ME,
Vinuela, F, Duckwiler, GR, Gobin, YP, Klucznick,
RP, Martin, NA and Frazee, J (1997) Incidence of cerebral vasospasm after endovascular treatment of acutely

ruptured aneurysms: report on 69 cases. J. Neurosurg.,
87: 830835.
Murri, L, Gori, S, Massetani, R, Bonanni, E, Marcella, F
and Milani, S (1998) Evaluation of acute ischemic
stroke using quantitative EEG: a comparison with conventional EEG and CT scan. Neurophysiol. Clin., 28:
249257.
Nagata, K, Tagawa, K, Hiroi, S, Shishido, F and Uemura,
K (1989) Electroencephalographic correlates of blood
flow and oxygen metabolism provided by positron
emission tomography in patients with cerebral infarction. Electroencephalogr. Clin. Neurophysiol., 72:
1630.
Naidech, AM, Janjua, N, Kreiter, KT, Ostapkovich, ND,
Fitzsimmons, BF, Parra, A, Commichau, C, Connolly,
ES and Mayer, SA (2005) Predictors and impact of aneurysm rebleeding after subarachnoid hemorrhage. Arch.
Neurol., 62: 410416.
Niedermeyer, E (2005) Cerebrovascular disorders and EEG.
In: E Niedermeyer and F Lopes da Silva (Eds.), Electroencephalography. Urban & Schwarzenberg Inc., Baltimore, 5th ed., pp. 339362.
Nuwer, MR, Jordan, SE and Ahn, SS (1987) Evaluation of
stroke using EEG frequency analysis and topographic
mapping. Neurology, 37: 11531159.
Quality Standards Subcommittee of the American Academy of Neurology (1996) Practice Advisory: thrombolytic therapy for acute ischemic strokesummary
statement. Neurology, 47: 835839.
Qureshi, AI, Sung, GY, Razumovsky, AY, Lane, K, Straw,
RN and Ulatowski, JA (2000) Early identification of
patients at risk for symptomatic vasospasm after aneurysmal subarachnoid hemorrhage. Crit. Care Med., 28:
984990.
Riemer, G, Hansen, HC, Theis, O and Kunze, K (1998) Der
Wert des EEG fur die Hydrozephalusdiagnostik bei
Subarachnoidalblutungen. Klin. Neurophysiol., 28: 19.
Rivierez, M, Landau Ferey, J, Grob, R, Grosskopf, D and
Philippon, J (1991) Value of electroencephalogram in
prediction and diagnosis of vasospasm after intracranial
aneurysm rupture. Acta Neurochir., 110: 1723.
Roseman, E, Bloom, BM and Schmidt, RP (1951) The
electroencephalogram in intracranial aneurysms. Neurology, 1: 2538.
Rothwell, PM, Giles, MF, Flossmann, E, Lovelock, CE,
Redgrave, JN, Warlow, CP and Mehta, Z (2005) A simple score (ABCD) to identify individuals at high early
risk of stroke after transient ischaemic attack. Lancet,
366: 2936.
Scheuer, ML and Wilson, SB (2004) Data analysis for continuous EEG monitoring in the ICU: seeing the forest
and the trees. J. Clin. Neurophysiol., 21: 353378.
Schneider, AL and Jordan, KG (2005) Regional Attenuation
WithOut Delta (RAWOD): a distinctive EEG pattern that
can aid in the diagnosis and management of severe acute
879
ischemic stroke. Am. J. Electroneurodiagn. Technol., 45:
102117.
Sharbrough, FW, Messick, JM, Jr. and Sundt, TM, Jr. (1973)
Correlation of continuous electroencephalograms with
cerebral blood flow measurements during carotid endarterectomy. Stroke, 4: 674683.
Shimoda, M, Takeuchi, M, Tominaga, J, Oda, S, Kumasaka, A and Tsugane, R (2001) Asymptomatic versus
symptomatic infarcts from vasospasm in patients with
subarachnoid hemorrhage: serial magnetic resonance
imaging. Neurosurgery, 49: 13411348.
Sidell, AD and Daly, DD (1961) The electroencephalogram
in cases of benign intracranial hypertension. Neurology,
11: 413417.
Strong, AJ, Fabricius, M, Boutelle, MG, Hibbins, SJ, Hopwood, SE, Jones, R, Parkin, MC and Lauritzen, M
(2002) Spreading and synchronous depressions of cortical activity in acutely injured human brain. Stroke, 33:
27382743.
Sundt, TM, Jr., Sharbrough, FW, Piepgras, DG, Kearns,
TP, Messick, JM, Jr. and OFallon, WM (1981) Correlation of cerebral blood flow and electroencephalographic
changes during carotid endarterectomy: with results of
surgery and hemodynamics of cerebral ischemia. Mayo
Clin. Proc., 56: 533543.
Suzuki, A, Yoshioka, K and Yasui, N (1990) Clinical application of EEG topography in cerebral ischemia: detection of functional reversibility and hemodynamics.
Brain Topogr., 3: 167174.
Tolonen, U and Sulg, IA (1981) Comparison of quantitative EEG parameters from four different analysis techniques in evaluation of relationships between EEG and
CBF in brain infarction. Electroencephalogr. Clin. Neurophysiol., 51: 177185.
Van Putten, MJ and Tavy, DL (2004) Continuous quantitative EEG monitoring in hemispheric stroke patients
using the brain symmetry index. Stroke, 35:
24892492.
Vespa, PM, Nuwer, MR, Juhasz, C, Alexander, M, Nenov,
V, Martin, N and Becker, DP (1997) Early detection of
vasospasm after acute subarachnoid hemorrhage using
continuous EEG ICU monitoring. Electroencephalogr.
Vespa, PM, OPhelan, K, Shah, M, Mirabelli, J, Starkman,
S, Kidwell, C, Saver, J, Nuwer, MR, Frazee, JG,
McArthur, DA and Martin, NA (2003) Acute seizures
after intracerebral hemorrhage: a factor in progressive
midline shift and outcome. Neurology, 60: 14411446.
Wartenberg, KE and Mayer, SA (2005) Multimodal brain
monitoring in the neurological intensive care unit:
where does continuous EEG fit in? J. Clin. Neurophysiol., 22: 124127.
Waterhouse, EJ, Vaughan, JK, Barnes, TY, Boggs, JG,
Towne, AR, Kopec-Garnett, L and DeLorenzo, RJ
(1998) Synergistic effect of status epilepticus and
880
ischemic brain injury on mortality. Epilepsy Res., 29:
175183.
Weir, B, Grace, M, Hansen, J and Rothberg, C (1978) Time
course of vasospasm in man. J. Neurosurg., 48: 173178.
Williams, AJ, Tortella, FC, Lu, XM, Moreton, JE and Hartings, JA (2004) Antiepileptic drug treatment of nonconvulsive seizures induced by experimental focal brain
ischemia. J. Pharmacol. Exp. Ther., 311: 220227.

Wood, JH, Polyzoidis, KS, Epstein, CM, Gibby, GL and
Tindall, GT (1984) Quantitative EEG alterations after
isovolemic-hemodilutional augmentation of cerebral perfusion in stroke patients. Neurology, 34: 764768.
Zampella, E, Morawetz, RB, McDowell, HA, Zeiger, HE,
Varner, PD, McKay, RD and Halsey, JH, Jr. (1991) The
importance of cerebral ischemia during carotid endarterectomy. Neurosurgery, 29: 727730.
SECTION V
Other Issues

M.R. Nuwer (Ed.)
882
CHAPTER 65
Safety issues during surgical monitoring

David B. MacDonalda,* and Vedran Deletisb
a
Department of Neurosciences, King Faisal Specialist Hospital and Research Center, 11211 Riyadh, Saudi Arabia
Center for Endovascular Surgery and Intraoperative Neurophysiology, Institute for Neurology and Neurosurgery, Roosevelt Hospital,
New York, NY 10019, USA
65.1. Introduction
Intraoperative monitoring (IOM) normally improves
patient safety, but like any medical procedure can inadvertently lead to harm. This chapter addresses the
nature and avoidance of hazards to patients and hospital
staff in terms of electrical safety, procedure-specific
safety, infection control, and essential performance.
65.2. Electrical safety
Electrical safety involves the proper manufacture, use,
and maintenance of medical electrical devices to minimize the chance of electric shock, fire, burn, or hazardous
output. In this regard, IOM devices must comply with
International Electrotechnical Commission standards
(IEC, 1998, 2005) that contain detailed information
about hazards and means of protection. The US National
Fire Protection Agency standards for health care facilities (National Fire Protection Association [NFPA],
2005) contain further electrical safety information.
65.2.1. Electric shock
A shock occurs when electricity passing through the
body between at least two contact points stimulates
excitable tissue. The effects depend on current strength
and type (alternating or direct), duration, frequency,
entry and exit site, and contact surfaces size and
impedance. There may be perception, pain, muscle
contractions, convulsions, tissue necrosis, or sustained
ventricular fibrillation (VF) which is the usual cause of
*
Correspondence to: David B. MacDonald, MD, FRCP(C),

ABCN, Section of Clinical Neurophysiology, Department of
Neurosciences, King Faisal Specialist Hospital and Research
Center, MBC 76, PO Box 3354, 11211 Riyadh, Saudi Arabia.
Tel: 966-1-464-7272, ext. 32772; fax: 966-1-442-4763.
E-mail: dbmacdon@yahoo.com (D.B. MacDonald).
death from electrocution and, therefore, is a major

concern (IEC, 2005; NFPA, 2005).
65.2.1.1. Risk of cardiac failure
The risk of VF varies with current frequency and
strength. Risk is greatest in the 10200 Hz range,
much less for direct current, substantially less at
1,000 Hz and decreases rapidly toward higher frequencies (IEC, 2005). Thus, 50- to 60-Hz mains frequency
is particularly hazardous. Estimates for the probability
of VF with 50- to 60-Hz current through an intracardiac electrode are listed in Table 1. A small risk with
zero current is due to mechanical stimulation from the
intracardiac electrode. There is no known minimum
current with zero probability. Swerdlow et al. (1999)
recently suggested a modest upward revision of these
probabilities because they found that pump failure as
a result of continuous capture simulating ventricular
tachycardia occurs at lower thresholds than VF.
65.2.1.2. Leakage currents
Of course, IOM devices are not designed for and must
never have direct or indirect connection to intracardiac
electrodes. However, like all medical electrical
devices they produce leakage currents (nonfunctional
currents from conductive, capacitive, or inductive circuit elements) that flow through the patient or operator
via other contacts.
The current density generated at the heart by current entering the chest is about 50 mA/mm2 per
ampere (IEC, 2005). Thus, a 1,000-mA transthoracic
50- to 60-Hz current that would produce painful muscle contractions should cause 50 mA/mm2 at the heart
and an 0.01 probability of cardiac failure. A 500-m
A transthoracic current that might not be perceived
should cause only 0.025 mA/mm2 at the heart and a
negligible chance of cardiac failure; the same current
directly to the heart could be fatal. Thus, cardiac failure due to excessive leakage current entering the
OTHER ISSUES
883
Table 1
Estimates for the probability of ventricular fibrillation
with 50- to 60-Hz current through an intracardiac
electrode
Current (mA)
Probability
500
300
200
100
50
10
0
1
0.95
0.50
0.05
0.01
0.002
0.001
Source: From IEC (2005) with permission from Springer, Wien

New York.
chest (e.g., between the two arms) or as a result of

otherwise innocuous leakage current somehow
directly reaching the heart is a serious hazard that
must be prevented.
There are three types of leakage current relevant to
IOM safety: (1) touch current, (2) patient leakage current, and (3) patient auxiliary current (IEC, 2005).
Touch current, also known as chassis leakage current
(NFPA, 2005), flows from any accessible part of the
device enclosure to the earth (ground) or to another
part of the enclosure via an external path. Touch current usually occurs when the patient or operator contacts the enclosure while having a connection with
the earth. The patient can make indirect contact, for
example, through the operator or unseated leads. Note
that in order to reduce the chance of completing a
touch current circuit dead metal operating room
items must not be earthed (NFPA, 2005).
Patient leakage current flows from a patient via the
patients connection to the earth, or originates from
an external voltage source on the patient and flows
from the patient via the patients connection to the
earth. Patient auxiliary current flows between any
patient connection and all other patient connections.
It may serve a nonphysiologic function (e.g., impedance monitoring) or be incidental (e.g., amplifier bias
current).
Leakage currents are especially relevant to IOM for
several reasons. First, the necessary patient connections commonly create transthoracic current paths.
Second, the typically numerous connections increase
total patient leakage current, the vector sum of individual connections (IEC, 2005). Third, routine patient
connections to other electrical devices introduce
external voltage sources and earth pathways. Finally,

central intravenous catheters commonly inserted
through the internal jugular or subclavian veins are
conductive pathways to the heart. These must not have
direct or indirect electrical contact with any part of the
IOM device (IEC, 2005). They should not be touched
by the operator who might be contacting the IOM
device and should not be close to patient connections;
Erbs point is one nearby site. The worst cases of direct
intracardiac leakage current or of mains voltage at a
patient connection must not be permitted.
65.2.1.3. Means of protection
The basic design requirement for medical electrical
device shock prevention is the use of two means of
protection, consisting of protective earthing, insulation, or protective impedance (IEC, 2005). Each element used as a means of protection must comply with
detailed technical specifications (IEC, 2005).
65.2.1.4. Single fault safe
A corollary requirement is that the device must be single fault safe. This means that loss of one method of
protection does not create unacceptable risk (IEC,
2005). The single fault condition can arise from spontaneous component failure, the stress of normal hospital use, or foreseeable misuse, for example, a broken
protective earthing conductor in a power cord. Some
single faults are safe because they engage an overcurrent protective device or obvious warning signal. The
safety of other single faults depends on sufficiently frequent periodic inspection and repair to make the likelihood of a second fault before the next inspection
negligible.
65.2.1.5. Leakage current limits
The fundamental principle is that no hazardous
voltage or current between any accessible part and
the earth or another accessible part will exist (IEC,
2005). This must be confirmed by leakage current testing in normal and simulated single fault conditions.
Custom-built or custom-modified devices should be
in compliance. Allowable limits are based on the type
of patient connections. IOM devices are required to
have patient connections sufficiently isolated from
earthed parts and other accessible parts for safe external or internal patient connections, excluding cardiac
connection (IEC, 1998, 2005). Cardiac connections
require greater leakage current limitation.
Table 2 itemizes limits relevant to IOM devices
(IEC, 2005). Higher values (not to exceed 10,000 mA)
884
D.B. MACDONALD AND V. DELETIS
Table 2
International electrotechnical commission leakage current limits relevant to IOM electrical devices
Touch current
Auxiliary patient current
Patient leakage currentb
Total patient leakage currentb
Alternating current 1 kHza
Direct current
NC (A)
SFC (A)
NC (A)
SFC (A)
100
100
100
500
500
500
500
1,000
10
10
50
50
50
100
Source: From IEC (2005) with permission from Springer, Wien New York.
At frequencies above 1 kHz, no leakage current shall exceed 10,000 mA.
b
In the special condition of an external voltage (e.g., mains voltage) on a patient connection, current shall not exceed 5,000 mA.
NC, normal condition; SFC, single fault condition.
are allowed for frequencies above 1,000 Hz because

these have a lower risk for cardiac failure. In addition,
patient leakage current must not exceed 5,000 mA in
the special condition of external voltage on a patient
connection, including the event of mains voltage connection. This could theoretically cause 0.25 mA/mm2
at the heart and still has a low probability of cardiac
failure, except in the circumstance of a direct path to
the heart. Note that jurisdictional deviations from IEC
limits exist and that limits are periodically revised.
65.2.1.6. Inspection and maintenance
Testing must be done at the time the device is manufactured, at installation, on signs of damage, after
repair, and at least biannually or more frequently if
indicated by the manufacturer or hard use (IEC,
2005; NFPA, 2005). Note that IOM systems consisting of multiple interconnected devices supplied by a
single power cord are tested as an assembly. Failure
to periodically test equipment increases the likelihood of a seriously hazardous double fault condition
arising (IEC, 2005).
65.2.1.7. Power cords and patient lead connectors
Most IOM devices utilize protective earthing as one
means of protection, but potentially dangerous power
outlet or cord defects and damage are common
enough (Medical Device Safety Reports [MDSR],
1979). Consequently, power outlets should be well
constructed and not contain tin-soldered wiring that
may loosen over time. Power cords must be heavy
duty, exceed the instruments requirements, have
strain relief at each end, and not have strain-prone
right angle plugs (MDSR, 1979; NFPA, 2005).
Inspection of their integrity is to be included with
device maintenance. Extension cords are discouraged
but permissible when they meet power cord requirements (NFPA, 2005).
Fatal electrocution and serious injuries have
occurred with accidental insertion of electrocardiographic (ECG) patient leads with unprotected pin
connectors into power receptacles (MDSR, 1993a).
Therefore, connectors must now be touch proof,
designed not to fit into power or other inappropriate
active receptacles, or electrically contact any other
conductive surface when not seated in their intended
receptacles (IEC, 2005; NFPA, 2005).
65.2.2. Fire hazards
IOM personnel should be aware that about 100
operating room fires occur annually in the United
States and cause serious injuries and death (Daane
and Toth, 2005). IOM devices must comply with
standards that make the likelihood of electrical fire
or excessively hot surfaces remote (IEC, 2005;
NFPA, 2005). Any evidence of overly hot parts
(>40 C) or electrical burning should be immediately
inspected.
Because of the enriched oxygen and nitrous oxide
atmosphere in the intubated oropharynx, many fires
involve surgery of the head and neck region. Most
fires are due to electrosurgical ignition of flammable
materials or liquids. Alcohol-based skin preparation
solutions are discouraged or even banned in some
jurisdictions (OR Manager, 2005). If used, flammable liquids must not pool under surgical drapes and
must fully dry before draping. This applies to alcohol
swabs, acetone, and collodion that contains ether and
alcohol. These must not be in open use during electrosurgery. IOM personnel should be trained in basic
fire prevention and control.
OTHER ISSUES
65.2.3. Electrical burns

Electrical burns at IOM electrodes are rare, but might be
underreported. When they occur they can be painful, disfiguring, or even require plastic surgery (Russell and
Gaetz, 2004). Other mechanisms should be considered
before attributing skin lesions to electrical burn (MDSR,
1993b). Pressure necrosis can occur underneath rigid bar
stimulating electrodes that should not be used for IOM
(Food and Drug Administration [FDA], 1992). Minor
skin preparation abrasions are common but normally
heal spontaneously. Unusual causes of electrode skin
injuries include traumatic scalp alopecia, calcinosis
cutis, or idiosyncratic contact dermatitis (Morris et al.,
1992; Puig et al., 1998; Avenel-Audran et al., 2003).
Thermal injury from heat lamps or surfaces over
40 C, chemical burns from pooled skin preparation
solutions, and mechanical injuries due to pressure or
shearing forces may occur near electrode sites.
There are three mechanisms for electrical burns:
(1) stray high-intensity radiofrequency (RF) current,
(2) sustained low-intensity direct current, and rarely
(3) patient connection to mains power (MDSR,
1993b; Russell and Gaetz, 2004). Burns can occur
at IOM, ECG, and other monitoring electrodes, or
at any other conductive contact, such as a metallic
part of the operating room table, but most frequently
occur under electrosurgical unit (ESU) dispersive
electrodes (MDSR, 1993c).
65.2.3.1. Electrosurgical RF burns
Electrosurgery is not electrocautery, which applies an
electrically heated tip. Instead, the unheated ESU
blade passes 0.33-MHz current to cut or coagulate
adjacent tissue. Nonconductively coupled currents
(capacitive, inductive, and radiative) through extrinsic conductors are more prominent with RF than with
lower frequency circuits (NFPA, 2005). These are
less predictable than conductive circuits and contribute to the large electrical artifacts that disable monitoring during ESU activation.
Bipolar ESU current is of relatively low intensity
and flows through tissue between the two tips of forceps used to focus heating. This mode does not cause
distant burn injuries because no significant current
passes through the patient, and nonconductive coupling is too weak to generate burns at other patient
contacts (NFPA, 2005).
Stronger monopolar ESU current flows from the
blade into and through the patient. Current exiting the
patient is distributed between all available patient
885
connections according to their relative impedance and

distance from the blade (NFPA, 2005). That portion
flowing to IOM electrodes contributes to larger artifacts than bipolar ESU and often temporarily saturates
amplifiers. However, most current flows safely back
to the ESU through its large dispersive electrode fixed
to the patients skin near the operative site because this
provides a low-impedance short-distance pathway.
High current density at the small blade generates
the intended heat in adjacent tissue, while heating is
negligible under the large dispersive electrode having
low current density. However, small area or high
impedance as a result of poor application or partial
detachment may lead to burns at the dispersive electrode, or even at other patient connections where
more current must then exit (MDSR, 1993c; NFPA,
2005). Repeated surgical requests to increase power
are a warning sign of an ESU fault and risk of
burn. The worst case is dispersive electrode failure
due to detachment, disconnection, or cable damage.
Although some ESUs self-disable in this condition,
a cable short circuit can cause this feature to fail.
Even with an intact dispersive electrode, burns can
develop at other patient connections located near the
operative site or between the ESU blade and dispersive electrode (Parikh et al., 2003). Furthermore, stray
RF currents from nonconductive coupling or a defective ESU can become strong enough to cause burns
(Fig. 1; Isgum and Deletis, 2003; NFPA, 2005). This
may be more likely when the ESU is activated without
the blade contacting the patient. Leads crossing or
running parallel to ESU cables and looped or coiled
leads promote RF coupling (NFPA, 2005).
The likelihood of RF burns also varies with IOM
electrode impedance and area (Russell and Gaetz,
2004; NFPA, 2005). Current passing through higher
impedance generates more heat and small electrodes
increase current density. The subdermal needles
commonly used for IOM have small surface area.
Consequently, most IOM electrode burns involve
needles that have even been vaporized by the intensity of generated heat.
65.2.3.2. Protection against electrosurgical
RF burns
Several simple measures reduce the likelihood of an
ESU RF burn (MDSR, 1993c; Russell and Gaetz,
2004; NFPA, 2005). The dispersive electrode should
be large and uniformly attached to clean, hairless dry
skin over soft tissue close to the surgical site and avoiding bony prominences. It should be applied after patient
886
Fig. 1. Examples of electrosurgical unit (ESU) radiofrequency (RF) burn accidents. (Left) Burns at tibial nerve stimulation
electrode sites. (Center) Scarring and alopecia at a transcranial electric stimulation (TES) electrode site. (Right) Burn at the
site of a surgical head frame pin. Each was due to excessive RF leakage current from a defective ESU.
positioning and its integrity rechecked after repositioning or requests for more power. The ESU should have a
dispersive electrode monitoring system and should be
inspected regularly and after any burn incident. It
should not be activated without blade contact. Monitoring devices should be located away from the ESU;
leads should hang freely in the air without coils or loops
and should not cross or run parallel to ESU cables. IOM
electrodes should have low impedance and preferably
large area. They should not be near the ESU blade or
between the blade and dispersive electrode. Needle
electrodes require particular caution.
65.2.3.3. Magnetic resonance imaging RF burns
RF coupling during magnetic resonance imaging
(MRI) can cause burns at ECG electrodes, pulse
oximetry sensors, and underneath monitoring leads
(Shellock and Crues, 2004). The development of
intraoperative MRI for spinal and neurosurgical procedures makes this and the possibility of ferromagnetic projectiles IOM safety issues.
Safety guidelines for MRI suites (Shellock and
Crues, 2004) advise care in the use and arrangement
of patient monitoring electrodes, leads, and devices.
Nonferromagnetic items are required (e.g., silver, gold,
platinum, iridium, carbon, plastic). Unnecessary electrically conductive items should be removed. Leads
should not form loops or coils. Thermal or electrical
insulation should be placed under leads. Monitoring
devices should be specifically designed and tested for
MRI compatibility.
Extraoperative electroencephalography (EEG) and
evoked potential recordings in MRI suites have been
accomplished safely using a variety of novel systems
(Schomer et al., 2000; Christmann et al., 2002). These

have included silver, gold, or special electrode materials
such as plastic or carbon, short twisted-pair leads, protective impedance, specially designed MRI compatible
amplifiers, fiber-optic cabling, and remote, specialized,
or shielded recording devices and stimulators.
Intraoperative MRI scanner designs enhance safety
and patient access (Schmitz et al., 2003). Some scanners are open and have low 0.2 T field intensity.
Stronger closed 1.5 T units have a table that swivels
out from the core when not scanning to bring the
patient outside the 5 G line, where it might be possible
to use some standard monitoring devices. For example,
electrocorticography (ECoG) and cortical somatosensory evoked potential (SEP) mapping have been safely
accomplished. However, MRI compatible anesthesia
and surgical devices are used. We are not aware of
any specially designed MRI compatible IOM systems
or guidelines for IOM safety during intraoperative
MRI. It would seem prudent to adopt general MRI
safety measures and consult with experienced centers.
65.2.3.4. Direct current burns
Sustained direct current of low intensity can cause
electrolytic formation of sodium hydroxide and
hydrogen at the cathode and chlorine and oxygen at
the anode (Russell and Gaetz, 2004). Over time, the
skin breaks down, the process accelerates, and fullthickness burns develop, particularly at the cathode.
Consequently, allowable leakage currents are substantially lower than for alternating current (Table 2).
Component failures or accidents such as saline spilling into a headbox can establish abnormal low-level
DC currents through patient electrodes (FDA, 1997).
OTHER ISSUES
Evidence of equipment malfunction should raise concern and equipment should be protected from the
ingress of fluids, including not mounting it on IV poles
(MDSR, 1993d).
65.2.4. Hazardous output
Because repetitive stimulation is a routine part of IOM,
injury to targeted neural structures or intervening tissues are safety concerns. Theoretically, excessive
acoustic stimuli might cause auditory system injury,
but we found no reports of such an event. It appears that
the IEC 125-dB hearing threshold level (HTL) limit
makes this exceedingly unlikely (IEC, 1998). Similarly,
intraoperative photic stimulation has not been reported
to cause visual system damage and is rarely used.
Excessive electrical or magnetic stimulation could
cause injury from tissue heating, excitotoxicity, or electrochemical injuries at the electrodetissue interface
(Agnew and McCreery, 1987). Other potential hazardous output issues are discussed in the procedurespecific sections of this chapter. Since magnetic stimulation is now rarely applied intraoperatively, the main
concern is electrical stimulation.
65.2.4.1. Parameters of electrical stimulation
Electrical stimulation between a positive anode and negative cathode is referred to as bipolar when both electrodes are near the neural target and monopolar when one
electrode is distant. Stimuli used for IOM are normally
rectangular pulses with a selected duration (D) of
0.050.5 ms and intensity (I) in milliamperes. A single
pulse is either monophasic or biphasic if there is second
phase of opposite polarity following the first. Charge (Q)
in microcoulombs (mCo) is the product of I D and is
the most relevant parameter for neural stimulation.
Charge density (QD) in mCo/cm2 is defined by Q per
electrode area. It falls off rapidly according to the square
of the distance and the impedance of tissues between the
electrode and target. Pulse trains consist of a continuous
series of pulses having a selected interpulse interval,
duration, and train repetition rate. Total charge and total
charge density are defined as Q and QD times the total
pulse number (twice the pulse number for symmetric
biphasic pulse trains). Voltage (V) I resistance (R)
in ohms. Energy in Joules (J) is defined by V Q (which
equals I2 R D) and produces heat.
65.2.4.2. Protection against electrical thermal injury
The IEC stipulates that electrical stimulators should
not exceed 50 mJ per pulse with a load resistance
887
of 1,000 O (IEC, 1998). The rationale is that all

known clinical applications can be achieved without
exceeding this limit, which is validated by the lack
of reports of confirmed stimulus-induced skin or
neural thermal injury.
In practice, pulse energy is normally much lower.
Considering IOM SEP stimulators that are limited to
100-mA constant current output, a typical 25-mA
pulse of 0.2-ms duration through 1,000-O impedance
will produce only 0.125 mJ. This would rise fourfold
through 4,000-O impedance. A 100-mA pulse of
1-ms duration through 4,000-O impedance could
reach 40 mJ, but is unlikely to be used. Caution is
advisable when using subdermal needles that
increase current density. It is also prudent to ensure
low electrode impedance and avoid needlessly high
intensity. For example, peripheral SEPs can determine supramaximal intensity beyond which no further increase of scalp SEPs could occur (Mac
Donald et al., 2005).
Considering direct nerve stimulation, thresholds
with 0.05-ms pulses are usually below 1 mA and
supramaximal stimuli are a few milliamperes, so that
pulse energy is very low. Low-range nerve stimulators circumvent the possibility of inadvertently
setting needlessly high intensity, but even if this
was to occur, energy should still be insufficient to
cause thermal nerve injury (Agnew and McCreery,
1987). Similar considerations apply to direct spinal
cord or cortical stimulation.
Considering transcranial electric stimulation
(TES), high scalp charge is needed to reach the brain.
Either long pulse (0.5 ms) stimulators of up to
240-mA output or short pulse (0.05 ms) stimulators
of up to 1,500-mA output are used to achieve similar
charge (MacDonald, 2002). Average TES impedance of EEG cups, EEG needles, and spiral needles
is about 1,100, 800, and 500 O, respectively. Larger
electrodes can reduce this to below 400 O, but are
not commonly used (Journee et al., 2004). Maximum long-pulse stimulator output through 1,100-O
impedance could produce about 32 mJ at the scalp,
insufficient to cause scalp burn. However, maximum
short-pulse stimulator output through 5001,100-O
impedance could produce between 56 and 75 mJ,
exceeding the IEC limit and introducing the possibility
of scalp thermal injury (MacDonald, 2006). A review
encompassing over 15,000 TES-monitored cases identified two vertex scalp burns that might have been
due to short-pulse TES (MacDonald, 2002). Therefore, caution is advisable when operating short-pulse
888
stimulators near maximum output and spiral or larger

electrodes should be preferred for these stimulators.
Low EEG cup electrode impedance should be
ensured when used for TES. Energy dissipates rapidly through the skull and there has been no evidence
for cerebral thermal injury. Stimulators designed
for TES should not be used for direct cortical
stimulation.
65.2.4.3. Protection against neural excitotoxicity
Animal experiments designed to evaluate the safety of
chronic neural stimulation have shown that prolonged,
continuous, high-frequency electrical stimulation over
many hours or days can damage peripheral nerves and
cortical neurons through excitotoxicity (Agnew and
McCreery, 1990; McCreery et al., 1990; McCreery
et al., 1995). Peripheral nerve axonal injuries occur
with 50-Hz stimulation and charge injury threshold
decreases with higher frequency. However, minimizing frequency (<20 Hz) or total stimulation time and
using intermittent stimulation reduces or abolishes
excitotoxicity even at high intensity. Therefore, the
parameters typically used for intraoperative percutaneous or direct nerve stimulation (less than 10 Hz,
intermittent) are exceedingly unlikely to be excitotoxic. Indeed, we have not found any reports suggesting
excitotoxic nerve injury during IOM. Nevertheless,
prolonged high-frequency (>20 Hz) stimulation should
be avoided. Similar considerations might apply to spinal
cord stimulation.
With direct cortical stimulation, charge and charge
density have been identified as excitotoxic cofactors
so that higher charge can be tolerated with lower
charge density and vice versa (McCreery et al., 1990;
MacDonald, 2002). Thus, 0.4 mCo at 40 mCo/cm2
and 6 mCo at 12 mCo/cm2 have each been identified
as injury thresholds during continuous 50-Hz biphasic
cat cortex stimulation lasting many hours. Total
charge and total charge density appear to be augmentative injury factors in these experiments that have
not explored the effect of brief intermittent train
stimulation.
In humans, Gordon et al. (1990) did not find any
histologic damage after 50-Hz intermittent pulse
trains of up to 4.5 mCo and 57 mCo/cm2 lasting several seconds. Traditional 5060-Hz cortical stimulation with small-diameter hand-held probes can
produce charge density exceeding 300 mCo/cm2 and
is considered safe (American Academy of Neurology, 1990; Gordon et al., 1990). Thus, there are no
reports of intraoperative cortical excitotoxicity and
no injury threshold for brief intermittent pulse trains

has been defined. Nevertheless, it would seem prudent to favor larger electrodes, to minimize charge
density, and to follow published protocols that have
not been found injurious. Parameters below experimental injury thresholds make injury exceedingly
unlikely and staying below the maximum values
evaluated by Gordon et al. would likely be safe. Similar considerations apply to the direct cortical pulse
train motor evoked potential (MEP) technique.
An analysis of currently available TES stimulators
found that charge and charge density in the brain are
not likely to exceed experimental injury thresholds
even at maximum output and are, therefore, exceedingly unlikely to cause cerebral excitotoxicity (MacDonald, 2002). This is particularly true considering
the very brief intermittent stimuli. Nevertheless, it
would seem prudent to avoid charge above that
needed to evoke the desired response and to follow
published protocols that have not been found injurious. Note, MEP threshold charge is about 35% lower
with short pulses, while long pulses have the advantage of speeding D-wave recovery between closely
timed pulses (Deletis et al., 2001; Bartley et al.,
2002; Hausmann et al., 2002). There is currently no
evidence for excitotoxicity with either technique
(MacDonald, 2002).
65.2.4.4. Protection against electrochemical
neuronal injury
Electrochemical properties of the tissueelectrode
interface have been comprehensively reviewed by
Merrill et al. (2005). To summarize, a Helmholtz double layer forms on a metal electrode contacting tissue.
It consists of electrons in the metal and ions in the
extracellular fluid with a layer of water between, forming a capacitor with an equilibrium (no current)
charge. When a pulse is given, the double layer charge
is altered as electrons are added or removed on the
metal side and the extent to which it is altered determines how current will flow into the tissue during the
pulse. Near equilibrium, capacitive current flows
through ions being attracted or repelled. Far from equilibrium, faradic current flows through toxic electrochemical reactions that transfer electrons across the
interface.
Pulse trains progressively shift the double layer
charge away from equilibrium until reaching a steady
state. If this entails faradic pulse current, then toxic
products can accumulate. A quantity of these can be
tolerated or buffered, so the likelihood of injury
OTHER ISSUES
depends on train duration. Monophasic trains drive

the double layer charge and, more importantly, any
consequent reactions in the same direction and cause
electrochemical neuronal injury in prolonged animal
stimulation experiments.
In the 1950s, Lilly devised biphasic pulses (the
Lilly waveform) to drive reactions in alternate
directions (Girvin, 1978). This prevents electrochemical injury in animal experiments, but its relevance
to brief intermittent pulse trains is questionable.
For example, early investigations on humans including one by Penfield used intermittent monophasic
pulse trains lasting seconds without clinical signs of
injury (Girvin, 1978). Biphasic trains also tend to
be less effective for neural activation. Nevertheless,
biphasic pulse trains are recommended for traditional
5060 Hz cortical mapping. Whether the same should
apply to direct high-frequency spinal cord stimulation
or not is unknown.
The very brief direct cortical pulse train technique
to elicit MEPs uses intermittent monophasic trains of
only three to five pulses lasting a fraction of a second. No clinical signs suggesting electrochemical
injury have been reported. It, therefore, currently
appears safe for this technique to employ monophasic
stimuli, pending further information to the contrary.
Note that electrochemical injury occurs only at the
neuralelectrode interface and is of no concern when
electrodes do not directly contact neural tissue.
889
to guide deep brain procedures. ECoG adds little risk

to that of craniotomy, but the value of inserting acute
depth electrodes should be weighed carefully. Microand macroelectrode insertions do have an incidence
of intracerebral hemorrhage, but this is mainly the
surgeons decision (Gorgulho et al., 2005).
Spinal epidural or subdural electrodes have been
used for a variety of purposes. They may be placed
by the surgeon after opening, inserted percutaneously
into the epidural space by an anesthesiologist, or
introduced percutaneously in the lumbar subdural
space and threaded upward.
To our knowledge, no major intraspinal adverse
events have been reported with any of these techniques. However, improper insertion may coil the epidural lead and could theoretically cause a knot
requiring surgical removal (Fig. 2). Nerve root or spinal cord trauma and infection are potential concerns
(Giebler et al., 1997; MacDonald, 2002). In particular,
65.3. Procedure-specific safety issues

Special safety concerns arise with certain IOM procedures, such as invasive techniques, direct cortical
stimulation, brainstem stimulation, spinal cord stimulation, electromyography (EMG), and SEPs. In particular, the safety of TES MEP monitoring has been
controversial.
65.3.1. Invasive techniques
Invasive techniques insert or place stimulating and/or
recording electrodes near, on, or within the brain or
spinal cord. Some advantages to these methods
include anesthesia resistance, high signal-to-noise
ratio, and access to deep structures. However, they
unavoidably carry an additional risk of hemorrhagic,
traumatic, or infectious complications.
Some invasive techniques are necessary for the
indicated purpose, such as ECoG during epilepsy
operations and micro- or macroelectrode recordings
Fig. 2. Improper percutaneous epidural recording electrode

insertion.
890
devastating intraspinal hemorrhage is a known complication of lumbar or epidural puncture. For example,
Rodi et al. (2003) reported a patient who required
emergency laminectomy to evacuate a hematoma after
epidural anesthesia for orthopedic surgery. This
patient had ankylosing spondylitis, which is a known
risk factor for epidural hemorrhage, as are anticoagulation, bleeding disorders, and difficult or repeated
puncture. A review of epidural anesthesia identified a
risk of only 1/150,000 procedures (Vander Meulen
et al., 1994). Still, in a review of over 600 spinal hematomas, 10% were due to lumbar or epidural puncture
for anesthesia, some with permanent neurologic injury
(Kreppel et al., 2003).
As a rule, invasive techniques should be used only
when necessary otherwise noninvasive methods
should be employed. For example, epidural D-wave
MEP recordings are justified during cerebral or intramedullary spinal cord tumor surgery because they
provide important corticospinal tract information that
predicts long-term motor outcome while noninvasive
muscle MEPs alone do not (Deletis, 2002; Yamamoto et al., 2004; Fujiki et al., 2006). They are no
longer justified during aortic surgery because TES
muscle MEPs are sufficient and superior for cord
ischemia detection (Chapter 60 by MacDonald and
Dong, this volume). They may no longer be justified
during spinal deformity operations because noninvasive MEP methods are now sufficient and because
of the recently discovered high incidence of false
D-wave results in these surgeries (Ulkatan et al.,
2006). Invasive spinal cord stimulation for myogenic
MEPs has become questionable since noninvasive
TES is both effective and more selective. Similarly,
the justification for epidural SEP monitoring is questionable now that favorable anesthesia and other
methodology for satisfactory noninvasive SEP monitoring are available (MacDonald et al., 2005). As a
final example, metal electrodes screwed into the skull
for pulse train MEPs (Watanabe et al., 2004) seem
unnecessarily invasive since scalp TES is sufficiently
effective (Szelenyi and Deletis, 2004).
65.3.2. Direct cortical stimulation
Even with otherwise noninjurious stimuli, seizure
induction remains a hazardous output issue during
direct cortical stimulation. Most provoked seizures
are focal, self-limited, or readily aborted with a variety of techniques, but a generalized convulsion could
cause serious injury or sequelae.
Single cortical pulses have not induced seizures

(MacDonald, 2002). However, traditional 5060-Hz
pulse trains lasting 15 s frequently provoke afterdischarges (seizure patterns) that build to clinical seizures in 520% of operations employing this
technique (MacDonald, 2002; Sala et al., 2002; Szelenyi et al., 2005). In addition, the method is often
difficult or unsuccessful in young children and generally requires an awake craniotomy that may be arduous or impossible for the patient.
Interestingly, a brief 5060-Hz pulse train for as
short as 0.3 s can terminate an afterdischarge (Lesser
et al., 1999). Furthermore, very brief (<0.03 s) highfrequency pulse trains for direct cortical MEPs infrequently provoke seizures (MacDonald, 2002; Sala
et al., 2002). This technique vastly reduces total
charge, uses general anesthesia, and usually succeeds
in young children (Sala et al., 2002). Thus, combined
cortical pulse train MEP and SEP mapping provides
sensorimotor localization with less hazardous output
and wider efficacy. The more hazardous traditional
technique remains necessary for language mapping
and requires concurrent ECoG for afterdischarge
detection to reduce the chance of false localization or
seizure induction.
65.3.3. Electrical stimulation of the brainstem
Direct electrical stimulation of the brainstem with a
hand-held probe to map motor nuclei of the lower
cranial nerves may cause cardiac arrhythmia and
blood pressure alterations that subside when stimulation is ceased (Morota et al., 1995). No serious complications or morbidity has been reported, but the
practitioner must be aware of this effect.
65.3.4. Spinal cord stimulation
Invasive spinal cord stimulation can produce strong
paraspinal muscle contractions that may interfere
with surgery. More importantly, this can cause potentially serious lactic acidosis even with moderate neuromuscular blockade when high-frequency (16 Hz)
thoracic stimulation is used to evoke caudal spinal
cord potentials (Tohdoh et al., 2001). Thus, excessive
muscle contraction is another example of hazardous
output that can occur with otherwise noninjurious
stimulus parameters. If used, this type of spinal cord
monitoring technique requires full neuromuscular
blockade. If spinal cord stimulation is used to evoke
myogenic MEPs, full relaxation is contraindicated
OTHER ISSUES
and therefore, care needs to be taken to avoid excessively strong or rapid paraspinal muscle contractions.
65.3.5. Electromyography
891
Table 3
Identified adverse events in over 15,000 TES MEP monitoring cases
Adverse event
Published
Unpublished
Total
While EMG monitoring is normally safely performed

with surface or subdermal needle electrodes, there
are some unique safety concerns (Al-Shekhlee et al.,
2003). Deep needle insertions risk intramuscular
hematoma or possibly nerve injury and should generally be avoided. In particular, deep needle insertions
into intercostal or diaphragm muscles risk potentially
serious pneumothorax and should not be done.
Tongue or lip
laceration
Mandibular fracture
Seizure
Cardiac arrhythmia
Scalp burn
Intraoperative
awareness
26
29
1
0
0
0
0
0
5
5
2
1
1
5
5
2
1
Based on literature review, unpublished clinical experience of several investigators, and information from Digitimer, Ltd. Reproduced from MacDonald (2002), with permission.
Safe noninvasive SEP techniques have been established through over 30 years of intraoperative experience. However, incomplete neuromuscular blockade
for muscle MEP monitoring now permits strong muscle twitches during peripheral nerve stimulation for
SEPs. This regularly interferes with pulse oximetry
sensors placed on digits. More importantly, otherwise
innocuous stimulation might now become a form of
hazardous output if muscle contractions are excessive. For example, tibial compartment syndrome
requiring surgical decompression has occurred following 10-Hz tibial nerve stimulation at the popliteal
fossa to evoke epidural SEPs when neuromuscular
blockade was omitted for concurrent TES muscle
MEPs (Weston, 2002). The SEP technique had been
developed by Jones et al. (1983) in an era of routine
neuromuscular blockade. Thus, rapid proximal nerve
stimulation should be avoided in the absence of
muscle relaxation.
65.3.7. Transcranial electric stimulation
Hazardous output concerns regarding TES that have
not already been discussed include the possibility
of seizures, kindling, injury as a result of patient
movement, bite injuries, and other complications
(MacDonald, 2002). In practice, there have been
remarkably few identified adverse events (Table 3).
Note that there is currently no evidence for a safety
difference between suprathreshold- and thresholdlevel TES.
65.3.7.1. Seizures
Seizures have rarely occurred during surgeries monitored with pulse train TES and no associated
morbidity has been reported (MacDonald, 2002).

Their rarity makes it uncertain what proportion was
due to stimulation as opposed to anesthesia, which
can also rarely induce seizures. Seizure rarity is
likely due to the very brief (<0.03 s) trains, anesthesia, and the general lack of predisposing patient factors (MacDonald, 2002). No seizures have been
reported with single-pulse TES.
65.3.7.2. Kindling
Kindling is an experimental method to induce epilepsy in animals and has never been demonstrated
in humans. The neural targets and stimulus parameters of TES MEP techniques differ markedly from
kindling models and there are no observations of
postoperative epilepsy attributable to TES (MacDonald, 2002).
65.3.7.3. Movement-related injury
The patient twitch induced by pulse train TES could
cause injury if a vital structure is jolted into or torn
away from a surgical tool, but there are currently no
reported incidents (MacDonald, 2002). This concern
arises mainly during neurosurgery and neck surgery.
Technical adjustments may help. For example, omitting leg MEPs that require strong stimuli is helpful during posterior fossa surgery. When this is done,
alternating C3-Cz and C4-Cz TES may reduce the
overall magnitude of patient movement by producing
predominantly unilateral twitches. When leg MEPs
are considered essential, C1/2 or Cz-(Cz 6 cm) stimulus montages may produce less movement than C3/4
(Deletis, 2002; Neuloh and Schramm, 2002), but are
892
65.3.7.6. TES contraindications

Relative TES contraindications include epilepsy, cortical lesions, convexity skull defects, raised intracranial
pressure, cardiac disease, proconvulsant medications
or anesthetics, intracranial electrodes, vascular clips or
shunts, and cardiac pacemakers or other implanted biomedical devices (MacDonald, 2002). Each must be
weighed against the risk of omitting MEPs; such
patients have been monitored uneventfully. An intraoperative seizure or cardiac arrhythmia is an indication
to consider aborting TES.
The benefits of MEP monitoring clearly outweigh
the risks, and the techniques are thus sufficiently safe
for clinical use in expert hands (MacDonald, 2002).
This is affirmed by clearance of TES stimulators in
many countries including the United States.
sometimes insufficient. Video monitoring and careful

stimulus timing are other preventive strategies. Partial
neuromuscular blockade may dampen but not eliminate
movement and complicates interpretation.
65.3.7.4. Bite injuries
Bite injuries due to jaw muscle contractions during
pulse train TES are the most common but still infrequent complication (MacDonald, 2002). Most tongue
or lip lacerations heal spontaneously, but a few have
needed surgical repair and one mandibular fracture has
resulted from threshold-level TES (Calancie et al.,
1998). Even one instance of potentially life-threatening
endotracheal tube rupture requiring emergency reintubation has occurred (MacDonald, 2006). Protective soft
bite-blocks are recommended.
65.4. Infection control

Nosocomial infection and transmissible diseases
including hepatitis, HIV, and JakobCreutzfeldt disease are safety concerns for IOM patients and personnel. The principles of standard precautions and
additional transmission-based precautions must be
respected (American Electroencephalographic Society, 1994; International Organization of Societies
for Electrophysiological Technology, 1999; World
Health Organization, 2004). Standard precautions
assume that every person is potentially infectious and
consist of hand washing, personal protective equipment, proper handling of patient care equipment,
prevention of needle-stick injuries, environmental
cleaning, and waste management.
65.4.1. Hand washing
QRS
Hand washing with soap, antimicrobial agent, or

waterless antiseptic is fundamental. It must be done
between contact with different patients or different
procedures on the same patient, after contact with
QRS
QRS
QRS
QRS
QRS
65.3.7.5. Other adverse effects

Other potential adverse effects do not appear to be
prominent (MacDonald, 2002). There have been no
observations of neuropsychologic sequelae, headache, or disturbances of hormonal or hematological
homeostasis.
Transient cardiac arrhythmia or blood pressure
alteration may rarely be observed during TES, but
the relationship is unclear. Deep current penetration
to the hypothalamus or brainstem is a possible explanation and another reason to avoid needlessly high
intensity. A parasitic TES current path from scalp
SEP electrodes through the headbox into leg electrodes and then traveling through the heart on the
way back to the head is another possible mechanism
(Journee, 2003). Using separate MEP and SEP monitoring devices would eliminate this possibility but
might be impractical. With a single device, it may be
advisable to use separate headboxes for leg and scalp
recording leads or disconnect scalp SEP electrodes
during TES (Journee, 2003). Cardiac arrhythmia must
be distinguished from TES artifacts that may appear in
the ECG (Fig. 3).
Fig. 3. Single-pulse transcranial electric stimulation (TES) artifact (*) simulating cardiac arrhythmia. Note the absence of a synchronous pulse wave following the TES artifacts.
OTHER ISSUES
blood or body fluids, after removing gloves, and on

entering and exiting the operating room.
893
equipment must be thoroughly cleaned and disinfected. Cleaning materials must be placed in appropriate disposal bags.
65.4.2. Personal protective equipment

65.4.4. Needle-stick injuries
Personal protective equipment provides a physical
two-way barrier against microorganism transfer
between the patient and the wearer. Their use is guided
by the circumstances and likelihood of exposure.
Operating room routines such as caps, masks, boots,
clothing, and respecting sterile items must be followed. Gloves must be worn for any possible contact
with blood or body fluids. They are mandatory for needle insertion and skin preparation that might produce
exudates or bleeding. Gloves must be removed after
use, placed in appropriate disposal bags, and should
not contact surfaces or the IOM instrument.
65.4.3. Patient care equipment
Apart from EMG or near-nerve recordings, needle
electrodes are no longer used in neurophysiology
laboratories because they are a source of blood exposure, needle-stick injury, and patient infection if
reused. Nevertheless, they are practical, effective, and
commonly used for IOM. Only sterile single-use needles should be used. The skin should first be cleansed
with an appropriate antimicrobial solution that should
fully dry. Used electrodes are placed in a sharps box
for destruction. Similarly, invasive electrodes should
be single-use and destroyed after they are used.
For surface electrodes, the necessary skin preparation with light sandpaper or mildly abrasive solutions
should not be so aggressive as to cause bleeding or
exudation. Adhesive disk electrodes are placed in
appropriate disposal bags after use. Scarification with
a blunted needle through collodion-fixed EEG cup
electrodes after application is discouraged, but occasionally necessary to further reduce impedance. If
used, the blunted needle must be sterile, single-use,
and disposed of in a sharps box. After use, reusable
EEG cup electrodes must be brushed under hot detergent using gloved hands or in an ultrasound bath to
remove surface matter. They are then immersed in a
suitable disinfectant (e.g., 1% hypochlorite solution
for 10 min) followed by rinsing, or autoclaved at
121 C for 15 min or 134 C for 3 min.
The entire IOM instrument including cables, simulators, headboxes, and other parts should be wiped
with detergent or an appropriate disinfectant (e.g.,
0.1% hypochlorite) after use. Body fluid spills on
Disposable needles should be held by the stem, not

recapped, and placed in a sharps box. Needle-stick
injuries must be immediately reported to the hospital
occupational health department. Bleeding from the
injury site should be encouraged and hands should
be washed with soap and water.
65.4.5. Additional precautions and special
considerations
Patients with active infectious diseases requiring airborne precautions (e.g., active tuberculosis), droplet
precautions (e.g., pneumonia or meningitis), or contact
precautions (e.g., multiresistant Staphylococcus aureus)
are unlikely to undergo surgery needing IOM. Established guidelines must be followed in the event that
they do (World Health Organization, 2004). Hepatitis
A and B staff immunization is recommended. Hospital
postexposure policies for HIV, hepatitis, and other
infectious diseases must be adhered to. Following use
on a patient with suspected JakobCreutzfeldt disease,
EEG cup electrodes must be sterilized (e.g., 2.5% hypochlorite for 1 h or autoclaving at 134 C for at least
18 min) and other electrodes must be destroyed.
65.5. Essential performance
Modern concepts of basic safety include essential
performance, defined as performance necessary to
achieve freedom from unacceptable risk (IEC, 2005).
Essential performance relevant to IOM requires correct output of diagnostic information that is likely to
be relied upon to determine treatment, where incorrect
information could lead to an inappropriate treatment
that would present an unacceptable risk to the patient
(IEC, 2005). Clearly, this applies to IOM equipment
and to IOM practice.
65.5.1. Essential performance of IOM equipment
Loss of essential performance of IOM equipment
occurs when a feature or function is degraded to a point
that the device is no longer suitable for its intended use
and this leads to an unacceptable risk. For example,
failure to monitor leaves the patient at risk of injury
894
that monitoring was intended to help avoid. Data corruption or loss and inaccurate stimulus output or signal
display could lead to inappropriate surgical decisions
that harm the patient. The device manufacturer must
perform a risk management process complying with
standards of the International Organization for Standardization (www.iso.ch). This must ensure that the
probability of loss of essential performance in normal
use and foreseeable misuse is low enough to make
the residual risk acceptable. Note that immunity from
electromagnetic interference is an element of IOM
device essential performance (IEC, 2004).
Rough handling, misuse, and foreseeable accidents
that could cause loss of IOM device performance
should be avoided. One simple example is not mounting device components underneath fluids that could
spill and disable the equipment.
Standard laboratory evoked potential methods do

not promote rapid monitoring. Instead, acquisition
can be accelerated through optimal anesthetic, stimulus, filtering, and derivation selections emphasizing
higher signal-to-noise ratio potentials (MacDonald
et al., 2004, 2005). Low signal-to-noise ratio potentials
(e.g., posterior tibial P31) may require hundreds or
thousands of sweeps to achieve sufficient reproducibility for valid interpretation. Simply averaging fewer
sweeps of such potentials leaves excessive residual
noise distortion which may lead to misinterpretation.
Every reasonable effort should be made to optimize feedback rapidity without compromising reproducibility. There is no general agreement on how
rapid feedback should be, but a 2-min or better feedback interval for all monitored modalities together is
one suggested, reasonable goal when monitoring
four-limb SEPs and MEPs (MacDonald et al., 2003).
65.5.2. Essential performance in IOM practice

Given accurate usable instrumentation, the practitioner must apply the device appropriately to provide
timely, correct information and avoid slow or misleading information which would put the patient at
risk. This more nebulous and controversial aspect of
essential performance is incompletely addressed by
existing standards or guidelines. However, it is probably the most important IOM safety issue of all.
65.5.2.1. Protection against slow surgical feedback
Harm could easily stem from slow, even though correct surgical feedback. This issue relates mainly to
averaged evoked potentials. Some neurologic injury
mechanisms, such as laceration, aspiration, or coagulation are instantaneous and irreversible no matter
how quickly detected. Others, such as ischemia, compression, or traction are reversible with timely intervention. However, left uncorrected for long enough,
permanent damage ensues. Furthermore, to intervene
appropriately the surgeon should have an idea about
what may be causing the impending injury, and this
is more likely to be evident with rapid feedback allowing correlation to surgical maneuvers.
How often slow feedback leads to harm is unknown.
However, delayed spinal cord ischemia detection is one
reason why SEPs do not substantially reduce paraplegia
rates during aortic surgery (Chapter 60 by MacDonald
and Dong, this volume). Therefore, the likelihood of
successful intervention relates to quick detection. Excessively slow feedback as a result of choice of methodology could become a failure of essential performance.
65.5.2.2. Protection against misleading information

Harm can also result from misleading information or
interpretation causing inappropriate surgical action or
inaction. Possible examples include incorrect localization of vital structures or abnormalities, incorrect
warning of neurologic compromise (false positive,
false alarm), and incorrect lack of warning (false
negative, false reassurance). Possible examples of consequent inappropriate action include excision of eloquent cortex, unnecessary decision to abandon spinal
deformity correction, and unnecessary decision to stop
tumor resection. Examples of inappropriate inaction
could include not resecting an epileptic focus, not
restoring spinal cord perfusion during aortic surgery,
and not modifying spinal deformity correction. It is
impossible to know how often such events take place,
but false results have been reported in a variety of settings (e.g., Ulkatan et al., 2006). This may be particularly likely when predictions are attempted for an
unmonitored system, nonselective or incomplete
methods are applied, systemic and technical control
is lacking, or expert interpretation is unavailable.
Thus, misleading information as a result of inadequate methodology or lack of expertise can become a
failure of essential performance. Expert care emphasizing modality-selective methods and technical/systemic
control can minimize the likelihood of misleading
information. Some examples include: SEPs, spinal
cord evoked potentials, and spinally elicited peripheral
nerve potentials (previously misnamed neurogenic
MEPs) may miss motor compromise, but selective
TES MEP methods make this unlikely (Ihaya et al.,
OTHER ISSUES
1990; Deletis, 2001; MacDonald and Janusz, 2002;

Minahan et al., 2002; Yamamoto et al., 2004). Loss of
TES muscle MEPs, for fear of permanent paralysis,
could prompt incomplete spinal cord tumor removal
and risk of recurrence, whereas including D-wave
monitoring may allow complete removal with knowledge of probable long-term recovery (Deletis, 2002).
Systemic alterations or fade affecting leg potentials
can produce apparently false results, but the addition
of upper limb recording helps prevent this by providing
systemic control (Meylaerts et al., 1999; MacDonald
and Janusz, 2002; MacDonald et al., 2003). As a final
example, peripheral SEPs identify stimulus failure or
limb ischemia that can cause otherwise misleading
decrements (MacDonald and Janusz, 2002).
The monitorist should choose selective methods
appropriate for the surgery and employ systemic/technical control whenever possible. It should be clearly
understood what structure(s) are being monitored and
no assumptions about unmonitored structures should
be made. Inherent limitations of the methods applied
should be understood. An expert IOM interpreter
should at least design and supervise the recordings
and be present for irrevocable surgical decisions. Preferably, the same expertise would always be present,
but in any case should be readily available.
65.6. Conclusion
IOM intends to improve patient safety, but like all
medical endeavors entails some hazards for patients
and practitioners. Adhering to the principles of electrical safety, procedure-specific safety including the
reservation of invasive methods for justified indications, and infection control should provide protection. Modern concepts of basic safety extend to
essential performance that applies to IOM devices
and practice. Every reasonable effort should be made
to reduce the likelihood of a failure in the essential
performance of IOM. Expert care and interpretation
emphasizing rapid selective methods with technical
and systemic control can help accomplish this.
References
safety with chronically implanted nerve electrodes. Epilepsia, 31(Suppl. 2): S27S32.
895
Al-Shekhlee, A, Shapiro, BE and Preston, DC (2003) Iatrogenic complications and risks of nerve conduction studies and needle electromyography. Muscle Nerve, 27:
517526.
American Academy of Neurology (1990) Assessment: intraoperative neurophysiology. Report of the Therapeutics and
Technology Assessment Subcommittee of the American
Academy of Neurology. Neurology, 40: 16441646.
American Electroencephalographic Society (1994) Report
of the committee on infectious disease. J. Clin. Neurophysiol., 11: 128132.
Avenel-Audran, M, Goossens, A, Zimerson, E and Bruze,
M (2003) Contact dermatitis from electrocardiographmonitoring electrodes: role of p-tert-butylphenol-formaldehyde resin. Contact Dermatitis, 48: 108111.
Bartley, K, Woodforth, IJ, Stephen, JP and Burke, D (2002)
Corticospinal volleys and compound muscle action
potentials produced by repetitive transcranial stimulation during spinal surgery. Clin. Neurophysiol., 113(1):
7890.
Calancie, B, Harris, W, Broton, JG, Alexeeva, N and Green,
BA (1998) Threshold-level multipulse transcranial
electrical stimulation of motor cortex for intraoperative
monitoring of spinal motor tracts: description of method
and comparison to somatosensory evoked potential monitoring. J. Neurosurg., 88: 457470.
Christmann, C, Ruf, M, Braus, DF and Flor, H (2002) Simultaneous electroencephalography and functional magnetic
resonance imaging of primary and secondary somatosensory cortex in humans after electrical stimulation. Neurosci. Lett., 333: 6973.
Daane, SP and Toth, BA (2005) Fire in the operating room:
principles and prevention. Plast. Reconstr. Surg., 115:
73e75e.
Deletis, V (2001) The motor inaccuracy in neurogenic
motor evoked potentials (Editorial). Clin. Neurophysiol., 112: 13651366.
of the motor system. In: V Deletis and JL Shils (Eds.),
Neurophysiology in Neurosurgery. Academic Press,
San Diego, CA, pp. 2551.
Deletis, V, Isgum, V and Amassian, VE (2001) Neurophysiological mechanisms underlying motor evoked potentials in anesthetized humans. Part 1. Recovery time of
corticospinal tract direct waves elicited by pairs of
transcranial electrical stimuli. Clin. Neurophysiol.,
112(3): 438444.
Food Drug Administration [FDA] (1992) Medical Device
Reporting (MDR) Database. Adverse event report
[FDA web site]. Available at: http://www.accessdata.
fda.gov/scripts/cdrh/cfdocs/cfMDR/Detail.CFM?ID
463350. Accessed 5 Jan 2006.
Food Drug Administration [FDA] (1997) Manufacturer and
User Facility Device Experience (MAUDE) Database.
896
Adverse event report [FDA web site]. Available at: http://
www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfMAUDE/
Detail.CFM?MDRFOI__ID 82088. Accessed 5 Jan
2006.
Fujiki, M, Furukawa, Y, Kamida, T, Anan, M, Inoue, R,
Abe, T and Kobayashi, H (2006) Intraoperative corticomuscular motor evoked potentials for evaluation of
motor function: a comparison with corticospinal D and
I waves. J. Neurosurg., 104: 8592.
Giebler, RM, Scherer, RU and Peters, J (1997) Incidence of
neurologic complications related to thoracic epidural
catheterization. Anesthesiology, 86: 5563.
Girvin, JP (1978) A review of basic aspects concerning
chronic cerebral stimulation. In: IS Cooper (Ed.), Cerebellar Stimulation in Man. Raven Press, New York, pp. 112.
cortical electrical stimulation in the human: histopathologic confirmation. Electroencephalogr. Clin. Neurophysiol., 75: 371377.
Gorgulho, A, De Salles, AA, Frighetto, L and Behnke, E
(2005) Incidence of hemorrhage associated with electrophysiological studies performed using macroelectrodes
and microelectrodes in functional neurosurgery. J. Neurosurg., 102: 888896.
Hausmann, ON, Minb, K, Boosb, N, Ruetsch, YA, Ernia, T
and Curta, A (2002) Transcranial electrical stimulation:
significance of fast versus slow charge delivery for
intra-operative monitoring. Clin. Neurophysiol., 113:
15321535.
Ihaya, A, Morioka, K, Noguchi, H, Kimura, T, Nishii, H,
Hiramatsu, Y, Chiba, Y and Muraoka, R (1990) A case
report of descending thoracic aortic aneurysm associated
with anterior spinal artery syndrome despite no marked
ESP changes. Kyobu Geka, 43: 843846.
International Electrotechnical Commission [IEC] (1998)
60601-2-40 (Ed. 1): medical electrical equipment Part
2-40: Particular requirements for the safety of electromyographs and evoked response equipment [IEC web
site]. Available at: www.iec.ch. Accessed 12 Dec 2005.
60601-1-2 (Ed. 2.1): medical electrical equipment Part
1-2: General requirements for safety Collateral standard: Electromagnetic compatibility Requirements
and tests [IEC web site]. Available at: www.iec.ch.
Accessed 12 Dec 2005.
60601-1 (Ed. 3): Medical electrical equipment Part
1: General requirements for basic safety and essential
performance [IEC web site]. Available at: www.iec.ch.
Accessed 12 Dec 2005.
International Organization of Societies for Electrophysiological Technology (1999) Guidelines for infection
control in the clinical neurophysiology department
[OSET web site]. Available at: http://www.oset.org/
publications.htm. Accessed 10 Feb 2006.

Isgum, V and Deletis, V (2003) Dangerous high frequency
current leakage during neurophysiological intraoperative monitoring. In: Proceedings of the Symposium on
Intraoperative Neurophysiology, Ljubjana, 1718 Oct
2003 [Ljubjana Institute of Clinical Neurophysiology
web site], pp. 6972. Available at: www.kclj.si/ikn/
Dejavnosti/FAGA/2003/INVI2003.HTM.
Accessed
20 Nov 2003.
Jones, SJ, Edgar, MA, Ransford, AO and Thomas, NP (1983)
A system for the electrophysiological monitoring of the
spinal cord during operations for scoliosis. J. Bone Joint
Surg. Br., 65(2): 134139.
Journee, HL (2003) Electrical safety in intraoperative monitoring. In: Proceedings of the Symposium on Intraoperative Neurophysiology, Ljubjana, 1718 Oct 2003
[Ljubjana Institute of Clinical Neurophysiology web site],
pp. 6568. Available at: www. kclj. si/ikn/Dejavnosti/
FAGA/2003/INVI2003.HTM. Accessed 20 Nov 2003.
Journee, HL, Polak, HE and De Kleuver, M (2004) Influence
of electrode impedance on threshold voltage for transcranial electrical stimulation in motor evoked potential monitoring. Med. Biol. Eng. Comput., 42(4): 557561.
Kreppel, D, Antoniadis, G and Seeling, W (2003) Spinal
hematoma: a literature survey with meta-analysis of 613
patients. Neurosurg. Rev., 26: 149.
Lesser, RP, Kim, SH, Beyderman, L, Miglioretti, DL,
Webber, WR, Bare, M, Cysyk, B, Krauss, G and
Gordon, B (1999) Brief bursts of pulse stimulation terminate afterdischarges caused by cortical stimulation.
Neurology, 53: 20732081.
MacDonald, DB (2002) Safety of intraoperative transcranial electrical stimulation motor evoked potential
monitoring. J. Clin. Neurophysiol., 19: 416429.
MacDonald, DB (2006) Intraoperative motor evoked potential monitoring: overview and update. J. Clin. Monit.,
20(5): 347377. EPub ahead of print July 11, 2006.
MacDonald, DB and Janusz, M (2002) An approach to intraoperative monitoring of thoracoabdominal aneurysm surgery. J. Clin. Neurophysiol., 19: 4354.
MacDonald, DB, AlZayed, Z, Khoudeir, I and Stigsby, B
(2003) Monitoring scoliosis surgery with combined
multiple pulse transcranial electric motor and cortical
MacDonald, DB, Stigsby, B and Al Zayed, Z (2004)
A comparison between derivation optimization and
Cz0 -FPz for posterior tibial P37 somatosensory evoked
potential intraoperative monitoring. Clin. Neurophysiol.,
115: 19251930.
MacDonald, DB, Al Zayed, Z and Stigsby, B (2005) Tibial
somatosensory evoked potential intraoperative monitoring: recommendations based on signal to noise ratio analysis of popliteal fossa, optimized P37, standard P37 and
P31 potentials. Clin. Neurophysiol., 116: 18581869.
McCreery, DB, Agnew, WF, Yuen, TG and Bullara, L (1990)
Charge density and charge per phase as cofactors in neural
OTHER ISSUES
injury induced by electrical stimulation. IEEE Trans.
Biomed. Eng., 37: 9961001.
McCreery, DB, Agnew, WF, Yuen, TG and Bullara, LA
(1995) Relationship between stimulus amplitude, stimulus frequency and neural damage during electrical stimulation of sciatic nerve of cat. Med. Biol. Eng. Comput.,
33(3 Spec. No.): 426429.
Medical Device Safety Reports [MDSR] (1979) Electrical
Plugs: A Compendium of Problems [MDSR web site].
Available at: www.mdsr.ecri.org. Accessed 10 Jan 2006.
Medical Device Safety Reports [MDSR] (1993a) Risk of
Electric Shock from Patient Monitoring Cables and Electrode Lead Wires [MDSR web site]. Available at: www.
mdsr.ecri.org. Accessed 10 Jan 2006.
Medical Device Safety Reports [MDSR] (1993b) Investigating Device-Related Burns [MDSR web site].
Available at: www.mdsr.ecri.org. Accessed 10 Jan
2006.
Medical Device Safety Reports [MDSR] (1993c) ESU
Burns from Poor Dispersive Electrode Site Preparation
[MDSR web site]. Available at: www.mdsr.ecri.org.
Accessed 10 Jan 2006.
Medical Device Safety Reports [MDSR] (1993d) Electronic
Equipment Mounted on IV Poles [MDSR web site].
Available at: www.mdsr.ecri.org. Accessed 10 Jan 2006.
Merrill, DR, Bikson, M and Jefferys, JG (2005) Electrical
stimulation of excitable tissue: design of efficacious
and safe protocols. J. Neurosci. Meth., 141: 171198.
Meylaerts, SA, Jacobs, MJ, Van Iterson, V, De Haan, P and
Kalkman, CJ (1999) Comparison of transcranial motor
evoked potentials and somatosensory evoked potentials
Surg., 230: 742749.
Morota, N, Deletis, V, Epstein, FJ, Kofler, M, Abbott, R,
Lee, M and Ruskin, K (1995) Brainstem mapping
neurophysiological localization of motor nuclei on the
floor of the fourth ventricle. Neurosurgery, 37:
922930.
Morris, HH, Klem, G and Gilmore-Pollak, W (1992) Hair
loss after prolonged EEG/video monitoring. Neurology,
42: 14011402.
National Fire Protection Association [NFPA] (2005)
NFPA 99: Standards for Health Care Facilities [NFPA
web site]. Available at: www.nfpa.org. Accessed
2 Dec 2005.
Neuloh, G and Schramm, J (2002) Intraoperative neurophysiology of supratentorial procedures. In: V Deletis and JL
Shils (Eds.), Neurophysiology in Neurosurgery. Academic
Press, San Diego, CA, pp. 339401.
OR Manager (2005) Fire marshal bans alcohol-based preps.
OR Manager, 21: 13.
897
Parikh, SN, Mehlman, CT and Keith, RW (2003) A thirddegree burn caused by a neurogenic motor-evoked
potential monitoring electrode during spinal surgery: a
case report. Spine, 28: E21E24.
Puig, L, Rocamora, V, Roman, J, Saavedra, M and Alomar,
A (1998) Calcinosis cutis following calcium chloride
electrode paste application for auditory-brainstem evoked
potentials recording. Pediatr. Dermatol., 1: 2730.
Rodi, Z, Straus, I, Denic, K, Deletis, V and Vodusek, DB
(2003) Transient paraplegia revealed by intraoperative
neurophysiological monitoring: was it caused by the epidural anesthetic or an epidural hematoma? Anesth.
Analg., 96: 17851788.
Russell, MJ and Gaetz, M (2004) Intraoperative electrode
burns. J. Clin. Monit. Comput., 18: 2532.
neurophysiological monitoring in pediatric neurosurgery: why, when, how? Childs Nerv. Syst., 18: 264287.
Schmitz, B, Nimsky, C, Wendel, G, Wienerl, J, Ganslandt,
O, Jacobi, K, Fahlbusch, R and Schuttler, J (2003)
Anesthesia during high-field intraoperative magnetic
resonance imaging experience with 80 consecutive
cases. J. Neurosurg. Anesthesiol., 15: 255262.
Schomer, DL, Bonmassar, G, Lazeyras, F, Seeck, M, Blum,
A, Anami, K, Schwartz, D, Belliveau, JW and Ives, J
(2000) EEG-linked functional magnetic resonance imaging in epilepsy and cognitive neurophysiology. J. Clin.
Shellock, FG and Crues, JV (2004) MR procedures: biologic
effects, safety, and patient care. Radiology, 232: 635652.
Swerdlow, CD, Olson, WH, OConnor, ME, Gallik, DM,
Malkin, RA and Laks, M (1999) Cardiovascular collapse caused by electrocardiographically silent 60-Hz
intracardiac leakage current: implications for electrical
safety. Circulation, 99: 25592564.
Szelenyi, A and Deletis, V (2004) Motor evoked potentials.
J. Neurosurg., 101: 563564; author reply 564.
Szelenyi, A, Kothbauer, K, De Camargo, AB, Langer, D,
Flamm, ES and Deletis, V (2005) Motor evoked potential
monitoring during cerebral aneurysm surgery: technical
aspects and comparison of transcranial and direct cortical
stimulation. Neurosurgery, 57(4 Suppl.): 331338.
Tohdoh, Y, Sumita, S, Kawamata, T, Omote, K, Kawana, S
and Namiki, A (2001) Acute respiratory and metabolic
acidosis induced by excessive muscle contraction during spinal evoked stimulation. Br. J. Anaesth., 86:
589593.
Ulkatan, S, Neuwirth, M, Bitan, F, Minardi, C, Kokoszka, A
and Deletis, V (2006) Monitoring of scoliosis surgery
with epidurally recorded motor evoked potentials (D
Wave) revealed false results. Clin. Neurophysiol.,
117(9): 20932101.
Van der Meulen, EP, Van Aken, H and Vermylen, J (1994)
Anticoagulants and spinal-epidural anesthesia. Anesth.
Analg., 79: 11651177.
898
Watanabe, K, Watanabe, T, Takahashi, A, Saito, N, Hirato,
M and Sasaki, T (2004) Transcranial electrical stimulation through screw electrodes for intraoperative monitoring of motor evoked potentials. Technical note.
J. Neurosurg., 100: 155160.
Weston, P (2002) Anterior Tibial Compartment Syndrome
Post Scoliosis Surgery [Neuromonitoring List web site].
neuromon-l log0204. Available through: www.neuromon.
com. Accessed 11 Nov 2002.

World Health Organization (2004) Practical guidelines
for infection control in health care facilities [WHO web
site]. Available at: http://www.wpro.who.int/publications/
PUB_9290222387.htm. Accessed 2 Dec 2005.
and Fukaya, C (2004) Intraoperative monitoring of the
corticospinal motor evoked potential (D wave): clinical
index for postoperative motor function and functional
recovery. Neurol. Med. Chir. (Tokyo), 44: 169180.

M.R. Nuwer (Ed.)
899
CHAPTER 66
Monitoring the depth of anesthesia

George A. Mashour1 and Carl Rosow*
Department of Anesthesia and Critical Care, Massachusetts General Hospital, and Harvard Medical School, Boston, MA 02114, USA
66.1. Introduction
During the last decade, several companies have introduced monitors into clinical practice that utilize processed electroencephalography (EEG) or auditory
evoked potentials to assess the state of consciousness
during general anesthesia. The first of these monitors,
the bispectral index (BIS, Aspect Medical Systems,
Newton, MA), has been discussed in over 2,000 publications and is now installed in about 65% of domestic
operating rooms. The appropriate use of consciousness monitors and the validity of their measurements
have been subjects of intense interest and debate in the
anesthesia community. The most contentious issue concerns their role in the prevention of unintended intraoperative awareness. This chapter will review relevant
information on BIS and briefly describe the other commercially available devices. Our emphasis will be less
on the technical aspects of measurement than the clinical need for monitoring and the ways the information
can influence clinical outcome.
General anesthesia is a clinical state that can be
achieved in numerous ways, but the endpoints are virtually always the same. An adequately anesthetized
patient is unconsciousness, his/her responses to pain
are diminished, and he/she is immobile during surgery.
A complete general anesthetic agent like diethyl ether
or chloroform can produce all of these effects to varying
degrees, and the concept of anesthetic depth is rooted
in the earliest studies of these old drugs. The stages or
planes of ether anesthesia (Snow, 1847; Guedel,
1937) were defined by a combination of therapeutic
Correspondence to: Carl Rosow, M.D., Ph.D., Department

of Anesthesia and Critical Care, Massachusetts General
Hospital, 55 Fruit Street, Boston, MA 02114, USA.
E-mail: crosow@partners.org (C. Rosow).
1
Present Address: Division of Neuroanesthesiology,
University of Michigan Medical School, Ann Arbor,
MI 48109, USA.
effects (hypnosis, amnesia, analgesia, and immobility),

adverse effects (hypotension, excitation, and emesis),
and various other physiologic effects (pupillary function
and respiratory pattern). This holistic notion of anesthetic depth was qualitative and unique to ether, making
it difficult to compare one anesthetic drug to another.
Hypnosis, analgesia, and muscle relaxation the
so-called triad of anesthesia (Gray and Rees, 1952)
are now understood to be largely independent processes, and modern anesthesia is usually accomplished
by combining drugs that produce each of these effects
selectively. It follows that anesthetic depth is an
ambiguous concept (Jantti, 2005), and the overall state
of anesthesia is unlikely to be described adequately
by a single parameter. Even if there were such a parameter, its application to patient care would not necessarily be straightforward. For example, anesthesia
dosing will often be determined by how much medication a patient tolerates as well as how much he needs.
Most anesthetic drugs have very narrow therapeutic
indices, so administration of an excessive dose may
depress cardiovascular and respiratory function or lead
to delayed awakening. On the other hand, a patient
whose cardiovascular condition precludes adequate
doses of anesthetics is at increased risk for intraoperative awareness and recall.
The task for the anesthesiologist is made more challenging by two additional factors: there is enormous
variability in anesthetic requirement between patients,
and the clinical endpoints used to determine anesthetic
depth are completely nonspecific. The second point
deserves some amplification. An anesthetized patient
is limited to two broad types of response, irrespective
of the cause. These responses are autonomic (hypertension, tachycardia, sweating, tearing) or somatic (movement, coughing, respiratory effort). A rapid heart rate
could be a response to pain, but it might also indicate
hypovolemia or the administration of atropine. A slow
heart rate could be a response to morphine, a betablocker, or a vagal reflex. Lack of movement could be
an analgesic response to isoflurane or paralysis due to
900
pancuronium. The anesthesiologist must make these

distinctions entirely on the basis of clinical context.
For example, the occurrence of hypertension or tachycardia immediately after incision or some other painful
stimulus would usually be an indication to administer
analgesic agents. Most important for our subsequent
discussions, none of these autonomic or somatic endpoints is a sensitive or specific indicator that consciousness is returning.
What about movement? If a patient moves during
surgery, does it indicate awakening? In the 1960s, the
concept of minimum alveolar concentration (MAC)
was introduced as a measure of anesthetic potency
for volatile anesthetics like halothane (Eger et al.,
1965). MAC is defined as the lung alveolar concentration of inhaled anesthetic that prevents movement
in response to a noxious stimulus in 50% of patients.
A series of experiments performed in the 1990s
proved that MAC of an anesthetic reflects depression
of the spinal cord rather than the brain (Antognini
and Schwartz, 1993; Rampil, 1994). Interestingly, a
subjects response to voice command (unquestionably a brain effect) can be suppressed by about onethird the amount of inhaled or intravenous anesthetic
needed to prevent movement. We can reasonably
conclude that movement is unrelated to the neuroanatomic substrate of consciousness.
66.2. The EEG and anesthesia
Clinicians have had no specific way to measure
unconsciousness, the one component of the triad that
defines general anesthesia most clearly. This is also
the component most logically addressed by EEG techniques. Gibbs et al. (1937) reported that EEG measurements were sensitive to the effects of general
anesthetics. Since that time, these effects have been
studied extensively (discussed by Sloan and Jantti in
Chapter 5 of this volume). Although valuable for monitoring neurologic function in specialized or high-risk
surgeries, the routine use of unprocessed EEG data
to assess anesthetic depth in the operating room is
impractical for a number of reasons. First, there
is no unique electrical signature of an unprocessed
EEG that is invariant among all general anesthetics.
Second, the standard EEG device and electrode array
are too cumbersome for most routine operations. Third,
an expert interpreter is usually needed for analysis.
Most anesthetics produce a concentration-related
shift in the EEG power to lower frequencies, and
there have been numerous attempts to utilize simple
G.A. MASHOUR AND C. ROSOW
derivatives of the power spectrum to estimate anesthetic effect. Fourier transformation yields a single
number like median or spectral edge frequency, and
these numbers have proven to be quite useful for
measuring the effects of single drugs under controlled experimental conditions (Scott and Stanski,
1987). Unfortunately, power spectral measures have
not performed well to measure depth because
changes in frequency distribution do not bear a simple, monotonic relation to anesthetic dose. With
increasing doses of propofol, thiopental, benzodiazepines, and inhaled agents the EEG beta power usually increases and then decreases when theta and
after that, delta rhythms predominate. This means
that the median frequency (or spectral edge frequency) will also increase then decrease as the anesthetic dose is raised, so the same number can apply
to two completely different clinical states. Another
major problem for the use of power spectral analysis
is the occurrence of suppression. At high concentrations, most of the anesthetic drugs will produce an
isoelectric EEG, so any calculations based on power
will become unstable (Rampil, 1998).
66.3. Modern depth of anesthesia monitors
There have been numerous attempts to make depth of
anesthesia monitors utilizing endpoints as diverse as
heart rate variability, pupillometry, and facial electromyography. Real progress on EEG-based monitors did
not occur until high-speed microprocessors made it
possible to perform complex waveform analysis in
real time. There are now a number of EEG-based
depth of anesthesia monitors in various stages of
development. We include a brief description of five
that are commercially available (three in the United
States). Most monitors utilize spontaneous cortical
EEG, and one is based upon cortical evoked potentials. At least two monitors (BIS and spectral entropy)
utilize information on EEG synchronization, that is,
phase locking and harmonic interaction among the
various frequency components of the EEG. The relevance of synchronization or bicoherence to anesthetic
depth is also discussed by Sloan and Jantti in Chapter
5 of this volume.
BIS development has been well described and
serves as a model for the development of other

monitors. The BIS algorithm was derived empirically by analyzing digitized EEG recordings from
2,000 patients who had received a variety of
OTHER ISSUES
commonly used general anesthetics, sedativehypnotics, and opioids (Rosow and Manberg,
2001). Segments of EEG were compared to a clinical measure of hypnosis based on a standard categorical scale. Candidate spectral and bispectral
features of these EEG segments were computed
and tested for their ability to distinguish among the
clinically described hypnotic states. The best of
these features were then combined using multivariate statistical modeling to form a composite index.
The final BIS (a dimensionless number from 0 to
100) is the linear transformation of a scale that
combines four EEG features: a power spectral measure (relative beta power), two methods of suppression detection, and a bispectral measurement that
reflects EEG synchrony. Each of the four parameters is used to characterize a different portion of the
anesthetic continuum. The current BIS algorithm
has gone through a number of software revisions
that have each been revalidated. The clinical monitor uses a single referential forehead lead (BisSensor) in a position that does not correspond to the
standard 1020 system. The adhesive electrode strip
uses a printed circuit and chip technology to perform an automatic impedance check and then compute BIS in real time. In the most common monitor
configuration, a value for BIS is computed using a
15-s smoothing window that is updated every 2 s.
Thus, the number lags about 57 s behind the
patient.
Patient State Index/SEDLine (PSI; Hospira, Lake
Forest, IL) was also derived empirically from an
existing EEG database. The published studies have
used EEG recorded from four locations: FPI, FPz,
Cz, and Pz, referenced to linked ears. PSI (0100) is
a linear transformation of a scale that combines
suppression detection and six different power spectral features describing not only the distribution of
power versus frequency, but also the changes from
front to back (Dressler et al., 2004). The latest iteration of PSI uses a four-lead forehead montage that
(by definition) discards the information on anteroposterior distribution. The new array has been revalidated with the old database, although there is
little published information on this latest version.
Entropy monitors (SE, RE; Datex-Ohmeda Divison,
Instrumentarium Corp, Helsinki, Finland) are based
on the concept of information entropy posited by
Shannon (1948) and measure the increase in synchronization (decrease in randomness) that occurs
901
as anesthesia deepens. The commercial entropy

monitor displays two measurements: (1) state
entropy (SE), which uses information from 0.8 to
32 Hz reflecting the EEG dominant spectrum, and
(2) response entropy (RE), which increases the
bandwidth from 0.8 to 47 Hz and therefore includes
both EEG and electromyographic spectra (Vakkuri
et al., 2004). The forehead montage and electrode
are similar to BIS.
The Narcotrend (MonitorTechnik, Bad Bramstedt,
Germany) also uses a frontal EEG montage and
pattern recognition to categorize 14 stages and substages of anesthesia (Kreuer et al., 2004a). It is
based on a similar developmental process as the
BIS, although with a distinct algorithm.
Midlatency auditory evoked potential (MLAEP;
AAI, A-Line AEP Monitor, Danmeter A/S, Odense,
Denmark) uses a standard neurophysiologic measurement collected in the same way as brainstem
auditory evoked potentials (BAEP): A series of
auditory stimuli is administered, and the EEG is
recorded over the sensory cortex and signal averaged (Scheller et al., 2005; Schneider et al., 2005).
MLAEP occur 15100 ms after the stimulus and
reflect the earliest cortical processing of auditory
information. Unlike BAEP, MLAEP are relatively
sensitive to the effects of anesthetics. There is, thus
far, no general agreement on the best way to analyze
these waveforms. Different investigators use different waves (commonly Na and Nb) and may select
amplitude, latency, or a proprietary index based
on a derivative of these. Signal averaging of 256
1024 stimuli has been used in research studies,
although the commercially available AEP monitor
can use autoregressive modeling techniques to
extract a number in less than 2 s.
66.4. Validation studies
It is important to emphasize that BIS and the other
monitors are all intended to measure unconsciousness, not analgesia. The first validation studies of
the BIS monitor in the early 1990s were based on
the misconception (discussed previously) that movement in response to pain could be predicted with an
EEG monitor. Actually, the first studies concluded
that BIS could predict both movement and hemodynamic responses to noxious stimuli (Kearse et al.,
1994a,b), but the performance became less reliable
when large doses of analgesic drugs like opioids
902
were administered (Sebel et al., 1997). This is because

opioids produce analgesic effects that are not reflected
in the EEG. After these studies were analyzed, the
analgesic endpoints of movement and hemodynamic
response were abandoned, and the BIS was completely
redeveloped to be a predictor of response to voice. All
BIS versions labeled 2.x or higher use this sedativehypnotic endpoint.
There were several key studies validating BIS in
volunteers. A multicenter trial (Glass et al., 1997)
investigated the effects of stepwise increasing and
decreasing concentrations of anesthetics on BIS. The
sedative-hypnotic drugs studied were isoflurane, propofol, and midazolam. At each step, drug concentration was measured, BIS was recorded, and the
clinical response of the subject was assessed with a
standard categorical scale. The results indicated
(Fig. 1) that the probability of response to voice
decreased as BIS decreased, and this relationship was
the same for all three anesthetics. This last finding is
key, since a monitor that worked only for a specific anesthetic would have limited utility. A volunteer
study by Kearse et al. (1998) extended these findings
to show that BIS correlated with the ability of a subject
to process two-part verbal commands. Flaishon et al.
(1997) tested the hypothesis that BIS could detect
when a patient had regained consciousness after induction of general anesthesia. They administered a single
intravenous bolus dose of propofol or thiopental plus
a neuromuscular blocker and monitored BIS continuously. A cuff was kept inflated on one arm to maintain
the ability to show a motor response, and patients were
asked every 30 s to squeeze the investigators hand.
The duration of unconsciousness was highly variable,
Probability of response
1.00
Isoflurane
Midazolam
0.50
Propofol
0.00
20
40
60
80
100
BIS
Fig. 1. Probability of response to voice versus BIS in volunteer subjects. The probability is very low at values below
60, irrespective of the anesthetic. [Adapted with permission
from Glass et al. (1997).]
but a BIS value above 60 was highly correlated with

the return of response to verbal command. Since these
early trials, BIS has been validated with most of the
common anesthetic agents, either alone or in combination. It has also been validated in children (Denman
et al., 2000; Whyte and Booker, 2004), the elderly
(Katoh et al., 2000), and during pregnancy (Chin and
Yeo, 2004).
There is much less published information on the
other four monitors, but each has been validated in
an appropriately designed trial that deals with relevant endpoints for anesthesiologists (often a wellcharacterized measure like the Observers Assessment
of Sedation Scale). In most studies their sensitivity
and specificity for detecting the conscious state are
better than 8090%, a performance that greatly exceeds
measured or predicted drug concentrations. There are
obviously some caveats to these generalizations:
Each monitor produces a graded response, although
each may differ in its ability to distinguish a light

and deep hypnotic state. The data suggest that BIS
and PSI decrease in a gradual fashion with increasing anesthetic dose. MLAEP is more likely to be
lost entirely at a relatively light level of anesthesia.
Prediction of state only occurs in the controlled
setting of an experiment, and these monitors do not
actually predict responses during surgery (neither
do blood pressure nor heart rate). Each monitor
analyzes a brief epoch of EEG data and synthesizes
a snapshot of the patients state during the preceding 1530 s. The patients state of consciousness
30 s later will depend entirely on the surgical
context what stimuli are anticipated, and what
drugs have been given.
The monitors appear to work with common anesthetics, either alone or in combination. However, it
is clear that any single algorithm may not be able to
handle certain drugs with unique effects on the
cortical EEG (e.g., most can be disrupted by ketamine). Each may have specific sensitivity to sources
of electrical artifact in the clinical environment.
Each algorithm has been derived empirically
usually with data from a relatively healthy patient
population. One cannot simply assume that it is valid
to apply the same algorithms to other patient populations (e.g., those with brain injury or metabolic
derangement) without appropriate studies. This is
a particularly important consideration if the monitor
is to be used for assessing sedation/hypnosis in
intensive care unit populations.
OTHER ISSUES
66.5. Comparison of the BIS with other

EEG-based monitors
BIS is generally the standard for comparison for consciousness monitoring, although there have been surprisingly few direct comparison studies, and most of
them are underpowered. Chen et al. (2002) demonstrated that the PSI correlated well with the BIS during
induction, maintenance, and emergence of anesthesia.
The authors also concluded that there was greater
interpatient variability using the PSI compared to the
BIS. The PSI and BIS were found to be comparable
in a small study of intravenous and inhalational anesthesia, with the PSI reported to have less electrocautery artifact (White et al., 2004).
The Narcotrend monitor has been shown to correlate with predicted propofol concentrations as well as
the BIS (Kreuer et al., 2004b). The Narcotrend index
also correlated well with the BIS during propofol and
remifentanil anesthesia in a small number of patients
(n 26) (Schmidt et al., 2004a). The BIS and Narcotrend indexes were shown to be comparable in detecting isoflurane anesthesia (Kreuer et al., 2004a).
SE and RE showed a close correlation with BIS in
assessing sevoflurane anesthesia (Ellerkmann et al.,
2004), as well as propofol-remifentanil anesthesia
(Schmidt et al., 2004b). In a small study evaluating loss
of response to verbal command and loss of consciousness during propofol anesthesia, it was suggested that
SE might be more useful than BIS (Iannuzzi et al.,
2005). A recent study also demonstrated comparable
performance of BIS and entropy modules during both
intravenous and inhalational anesthesia, with less electrocautery interference by the entropy monitor (White
et al., 2006).
66.6. Are consciousness monitors clinically
useful?
66.6.1. Improved recovery
Interestingly, the Food and Drug Administration
(FDA) demanded information on clinical utility
prior to approving the BIS and patient state index
(PSI) monitors. Assuming one has an accurate measure of hypnosis, it is easy to postulate that one might
improve clinical care by either decreasing the utilization of hypnotic drugs, shortening recovery time, or
somehow improving the quality of recovery. The initial utility trial for the BIS monitor was published in
1997, and it is a model for these investigations. In this
903
multicenter double-blind trial, 240 patients were given

propofol, alfentanil, nitrous oxide anesthesia, and randomized to receive BIS-titrated (BIS) or standard
practice (SP) dosing of propofol. The guidelines for
BIS titration specified keeping BIS between 45 and
60 during the procedure and letting it increase to 60
75 beginning 15 min before the anticipated end of the
procedure. Those in the BIS-titrated group had a faster
emergence, a shorter stay in recovery, and received
23% less propofol.
The PSI, Narcotrend, and MLAEP monitors have
been studied in very similar trials (Drover et al.,
2002; Kreuer et al., 2003; Recart et al., 2003) and
demonstrated similar benefits in faster recovery and
lower drug usage. There are now over 20 utility studies
published as full articles, most of them dealing with
BIS. Johansen et al. (2000) conducted a large openlabel study when BIS was introduced into an academic
medical center. They noted recovery improvements
when BIS monitoring was utilized, but only when
intraoperative BIS values were within a target range
of 5065. At values below this, the monitor conferred
no advantage. The obvious conclusion is that the
benefit comes from alteration in practice (i.e., titrating
to a more appropriate hypnotic endpoint), not simply
from applying the monitor. The unresolved issue from
these trials remains whether an institution will achieve
meaningful cost savings by using this technology and
whether the savings will partially or totally outweigh
the direct costs of the monitors.
66.6.2. Reducing the risk of intraoperative
awareness
Consciousness monitors may reduce the incidence of
awareness with recall. The arguments on this issue
have been heated, as a number of anesthesiologists
feel strongly that the risk of awareness has been
exaggerated by the media and by those who wish to
sell monitors. A few facts now seem absolutely clear:
the incidence of intraoperative awareness with
explicit recall is low, but significant, and it depends
upon the particular population being studied. The
huge prospective studies of Sandin et al. (2000) in
Sweden and Sebel et al. (2004) in the United States
suggest that 0.15% is a good estimate for the general
population (i.e., 15 cases for every 10,000 general
anesthetics). The figures are higher for certain highrisk populations like emergency obstetrics, cardiac
surgery, and trauma.
904
An unknown proportion of patients experiencing

awareness may subsequently develop serious psychological sequelae, including posttraumatic stress disorder (Osterman and Van der Kolk, 1998). In 2004, the
Joint Commission on Accreditation of Hospitals
Organization issued a Sentinel Event Alert stating
that intraoperative awareness is a significant clinical
problem and requiring hospitals to have protocols
for preventing, detecting, and treating it.
A reliable tool to detect intraoperative awareness
would thus be an important clinical advance. Although
there was initially controversy over whether consciousness monitors could reduce the risk of awareness
(OConnor et al., 2001; Kalkman and Drummond,
2002), recent studies have demonstrated a significant beneficial effect. At this writing, data on the prevention of awareness are only available for the BIS
monitor.
Ekman et al. (2004) followed up on the previously
mentioned study of Sandin establishing the incidence
of awareness. BIS monitors were installed in all
operating rooms of the same Swedish hospitals,
so the investigators prospectively studied the next
4,945 patients who were to receive general endotracheal anesthesia with neuromuscular blockade. These
patients were monitored with the BIS, with values
recommended between 40 and 60. The incidence of
awareness and explicit recall in this population was
0.04% compared with 0.18% in the historical control
group of 7,826 patients who did not have cerebral
monitoring. The incidence of 0.04% in this cohort
represented two patients, both of whom had BIS
values >60 for 4 min or more.
Myles et al. (2004) completed a prospective, randomized, double-blinded study of BIS-guided anesthetic management of 2,463 patients at high risk for
awareness (baseline incidence 1%). Postoperative
interviews identified two patients who experienced
confirmed episodes of awareness with explicit recall
in the BIS-guided group and 11 in the control group
(0.16% vs. 0.9%, p 0.022). One BIS-monitored
patient with awareness had an episode during which
the BIS had reached 80, while the other had a
period in which the BIS reached 5559. The cases
of awareness in both studies indicate that, like other
intraoperative monitors, the BIS needs to be actively
assessed and acted upon. The investigators in the
Myles study also found that the use of the BIS has
a statistically significant effect on the anesthetic
recovery time in high-risk patients. On the basis of
these studies, the FDA extended the approved
indications for the BIS monitor . . . to reduce the risk

of awareness in adult patients undergoing general
anesthesia. Further studies of cerebral monitors
and awareness are clearly warranted.
66.6.3. Reducing delayed morbidity and/or
mortality
While subtherapeutic doses of anesthesia are associated with awareness, some very provocative data
suggest that supratherapeutic doses may be associated with delayed postoperative morbidity and mortality. In a prospective study of 1,064 patients, Monk
et al. (2005) demonstrated that cumulative deep
hypnotic time was a significant variable that was
independently associated with 1-year mortality after
noncardiac surgery (Table 1). For every hour of time
the BIS was <45, the risk of death at 1 year postoperatively increased by 24.4%, for a relative risk of
1.244. This finding is both surprising and unexplained: the study was not originally designed to look
at mortality, and some of the deaths were due to cancer or other causes that do not have an obvious relationship to anesthetic depth. The findings were
confirmed in a retrospective study of 5,057 patients,
who had a 20% increased risk of 1-year postoperative
mortality for every hour the BIS was <45 (Lennmarken et al., 2003). The Monk study clearly needs
to be repeated, but it also emphasizes the need for
accurate assessment of anesthetic depth. For anesthetic management based on the BIS, it also suggests
the range of 4560 may be optimal for surgical
anesthesia.
Table 1
Multivariate predictors of postoperative mortality
at 1 year
Predictor
Relative risk
p value
Charlson Comorbidity
Score (3 vs. 02)
Cumulative time BIS < 45
(per hour)
Systolic blood pressure <
80 (per minute)
16.116
<0.0001
1.244
0.0121
1.036
0.0125
The Charlson Score is a validated index for quantifying co-existing

medical illness. [Adapted with permission from Monk et al.
(2005).]
OTHER ISSUES
905
66.7. Limitations of the BIS
66.8. Conclusion
There are a number of limiting and confounding

factors associated with BIS monitoring. One such limitation is that BIS values remain unchanged or even
elevated during nitrous oxide anesthesia (Yamamura
et al., 1981; Rampil et al., 1998). Thus, BIS monitoring
may be less effective in situations where only nitrous
oxide (i.e., no propofol or potent volatile agent) is used
in conjunction with an intravenous opioid such as
remifentanil (Coste et al., 2000). This light anesthetic
technique is sometimes selected for cases where
intraoperative awakening is planned or when minimal
depression of evoked potentials is desired. The intravenous anesthetic ketamine may also elevate BIS
levels (Sakai et al., 1999; Vereecke et al., 2003) perhaps reflecting its ability to increase the beta power
of the EEG spectrum (Hering et al., 1994). It is of interest that both ketamine and nitrous oxide are thought to
act primarily through N-methyl-D-aspartate glutamate
receptors (Sato et al., 2005). The anesthetic effect of
xenon, which also acts on this receptor system, is similarly not well tracked by BIS (Goto et al., 2000). It is
of further interest that while spectral entropy values
are increased by ketamine in a similar manner to the
BIS (Hans et al., 2005a), addition of nitrous oxide
has been shown to decreases SE and RE values (Hans
et al., 2005b).
The BIS has been shown to have lower values
possibly appropriately in individuals with neurologic disorders such as Alzheimers (Renna et al.,
2003), severe hypoglycemia (Vivien et al., 2002; Wu
et al., 2002) and cerebral ischemia (Merat et al.,
2001). Furthermore, one subject who had a genetically
determined low-voltage EEG recording was reported
to have a very low BIS value when awake (Schnider
et al., 1998). This problem has not been reported
again despite the large numbers of patients known
to have low-voltage EEGs. BIS values are elevated
by sources of high-frequency artifact, particularly
frontalis muscle EMG. It is not surprising that muscle relaxants can lower BIS when such artifact is
present (Liu et al., 2005). Finally, BIS values may be
affected by interference from other intraoperative
devices such as electrocautery (Dahaba, 2005), pacemakers (Gallagher, 1999), endoscopic shavers
(Hemmerling and Migneault, 2002), and warming
devices (Hemmerling and Fortier, 2002). See Dahaba
(2005) for a review of conditions that may result
in BIS values not accurately correlated with hypnotic
state.
Current clinical practice in anesthesiology is focused

on balanced anesthetic technique in which a number
of drugs with distinct effects are delivered to achieve
desired goals. In contrast to the past use of exclusively
inhalational anesthesia, it is now routine to administer
neuromuscular blockers for the suppression of movement, opiates for pain control, benzodiazepines for
amnesia, and beta-blockers for the attenuation of cardiovascular responses. The use of these adjuncts to
general anesthetics masks the behavioral endpoints
that were once measured or observed in order to assess
anesthetic depth. It is of interest that the BIS monitor
has been used to assess the original stages of ether
anesthesia described in the nineteenth century, with
Stage 3 (surgical anesthesia) occurring at a BIS value
of 40 (Bhargava et al., 2004).
EEG-based technologies play an increasingly
important role in the maintenance of general anesthesia within the therapeutic window. The BIS monitor
has been clinically validated as a measure of hypnosis
in the adult and pediatric population and has demonstrated efficacy in the prevention of intraoperative
awareness. Furthermore, the suggestion that BIS
values <45 may be associated with postoperative mortality at 1 year indicates the need for further study of
the physiologic sequelae of deep anesthesia. Perhaps
the refinement of closed-loop systems in which
anesthetic delivery is controlled based on feedback
from BIS values will aid in maintaining appropriate
depth of anesthesia (Absalom and Kenny, 2003;
Locher et al., 2004).
The inability of the BIS to perform accurately
with ketamine, xenon, and nitrous oxide indicates
that further study and development of cerebral monitors is required. In many ways, it is surprising that
a single EEG algorithm has worked so well for so
many different inhaled and injected anesthetic
drugs perhaps this will teach us something about
their common neurophysiologic effects. It is clear
that the neurophysiology of anesthetic depth remains
linked to the persistent questions regarding the neural
correlates of consciousness and cognition.
References
Absalom, AR and Kenny, GNC (2003) Closed-loop control
of propofol anaesthesia using bispectral index: performance assessment in patients receiving computercontrolled propofol and manually controlled remifentanil
906
infusions for minor surgery. Br. J. Anaesth., 90(6):
737741.
Antognini, JF and Schwartz, K (1993) Exaggerated anesthetic requirements in the preferentially anesthetized
brain. Anesthesiology, 79: 12441249.
Bhargava, AK, Setlur, R and Sreevastava, D (2004) Correlation of bispectral index and Guedels stages of ether
anesthesia. Anesth. Analg., 98(1): 132134.
Chen, X, Tang, J, White, PF, Wender, RH, Ma, H,
Sloninsky, A and Kariger, R (2002) A comparison of
patient state index and bispectral index values during
the perioperative period. Anesth. Analg., 95(6):
16691674.
Chin, KJ and Yeo, SW (2004) Bispectral index values at
sevoflurane concentrations of 1% and 1.5% in lower
segment cesarean delivery. Anesth. Analg., 98(4):
11401144.
Coste, C, Guignard, B, Menigaux, C and Chauvin, M
(2000) Nitrous oxide prevents movement during orotracheal intubation without affecting BIS value. Anesth.
Analg., 91(1): 130135.
Dahaba, AA (2005) Different conditions that could result in
the bispectral index indicating an incorrect hypnotic
state. Anesth. Analg., 101: 765773.
Denman, WT, Swanson, EL, Rosow, D, Ezbicki, K,
Connors, P and Rosow, CE (2000) Pediatric evaluation
of the bispectral index (BIS) monitor and correlation
of BIS with end-tidal sevoflurane concentration in
infants and children. Anesth. Analg., 90(4): 872877.
Dressler, O, Schneider, G, Stockmanns, G and Kochs, EF
(2004) Awareness and the EEG power spectrum: analysis
of frequencies. Br. J. Anaesth., 93: 806809.
Drover, D, Lemmens, HJ, Pierce, ET, Plourde, G, Loyd, G,
Ornstein, E, Prichep, LS, Chabot, RJ and Gugino, L
(2002) Patient state index (PSI): titration of delivery
and recovery from propofol, alfentanil, and nitrous
oxide anesthesia. Anesthesiology, 97: 8289.
Eger, EI, II, Saidman, LJ and Brandstater, B (1965) Minimum alveolar anesthetic concentration: a standard of
anesthetic potency. Anesthesiology, 26: 756763.
Ekman, A, Lindholm, ML, Lennmarken, C and Sandin, R
(2004) Reduction in the incidence of awareness using
BIS monitoring. Acta Anaesthesiol. Scand., 48: 2026.
Ellerkmann, RK, Liermann, VM, Alves, TM, Wenningmann, I, Kreuer, S, Wilhelm, W, Roepcke, H, Hoeft, A
and Bruhn, J (2004) Spectral entropy and bispectral index
as measures of the electroencephalographic effects of
sevoflurane. Anesthesiology, 101: 12751282.
Flaishon, R, Windsor, A, Sigl, J and Sebel, PS (1997)
Recovery of consciousness after thiopental or propofol.
Bispectral index and isolated forearm technique. Anesthesiology, 86(3): 613619.
Gallagher, JD (1999) Pacer-induced artifact in the bispectral index during cardiac surgery. Anesthesiology, 90(2):
636.

Gibbs, FA, Gibbs, LE and Lennox, WG (1937) Effect on
the electroencephalogram of certain drugs which influence nervous activity. Arch. Intern. Med., 60: 154166.
Glass, PS, Bloom, M, Kearse, L, Rosow, C, Sebel, PS and
Manberg, P (1997) Bispectral analysis measures sedation
and memory effects of propofol, midazolam, isoflurane,
and alfentanil in healthy volunteers. Anesthesiology,
86(4): 836847.
Goto, T, Nakata, Y, Saito, H, Ishiguro, Y, Niimi, Y, Suwa, K
and Morita, S (2000) Bispectral analysis of the electroencephalogram does not predict responsiveness to verbal
command in patients emerging from xenon anaesthesia.
Br. J. Anaesth., 85(3): 359363.
Gray, T and Rees, G (1952) The role of apnoea in anaesthesia for major surgery. BMJ, 2: 891892.
Guedel, A (1937) Inhalation Anaesthesia: A Fundamental
Guide. Macmillan, New York.
Hans, P, Dewandre, P-Y, Brichant, JF and Bonhomme, V
(2005a) Comparative effects of ketamine on Bispectral
Index and spectral entropy of the electroencephalogram
under sevoflurane anaesthesia. Br. J. Anaesth., 94(3):
336340.
Hans, P, Dewandre, P-Y, Brichant, JF and Bonhomme, V
(2005b) Effects of nitrous oxide on spectral entropy of
the EEG during surgery under balanced anaesthesia with
sufentanil and sevoflurane. Acta Anaesthesiol. Belg., 56
(1): 3743.
Hemmerling, TM and Fortier, JD (2002) Falsely increased
bispectral index values in a series of patients undergoing cardiac surgery using forced-air-warming therapy
of the head. Anesth. Analg., 95(2): 322323.
Hemmerling, TM and Migneault, B (2002) Falsely increased
bispectral index during endoscopic shoulder surgery attributed to interferences with the endoscopic shaver device.
Anesth. Analg., 95(6): 16781679.
Hering, W, Geisslinger, G, Kamp, HD, Dinkel, M, Tschaikowsky, K, Rugheimer, E and Brune, K (1994) Changes
in the EEG power spectrum after midazolam anaesthesia combined with racemic or S- () ketamine. Acta
Anaesthesiol. Scand., 38(7): 719723.
Iannuzzi, M, Iannuzzi, E, Rossi, F, Berrino, L and Chiefari, M
(2005) Relationship between bispectral index, electroencephalographic state entropy and effect-site EC50 for propofol at different clinical endpoints. Br. J. Anaesth., 94(5):
613616.
Jantti, V (2005) From crystal ball towards cognitive anaesthesiology. Acta Anaesthesiol. Scand., 49: 273276.
Johansen, JW, Sebel, PS and Sigl, JC (2000) Clinical impact
of hypnotic titration guidelines based on EEG bispectral
index (BIS) monitoring during routine anesthetic care.
J. Clin. Anesth., 12: 433443.
Kalkman, CJ and Drummond, JC (2002) Monitors of depth of
anesthesia, quo vadis? Anesthesiology, 96(4): 784787.
Katoh, T, Bito, H and Sato, S (2000) Influence of age on
hypnotic requirement, bispectral index, and 95%
OTHER ISSUES
spectral edge frequency associated with sedation
induced by sevoflurane. Anesthesiology, 92(1): 5561.
Kearse, LA, Jr., Manberg, P, Chamoun, N, DeBros, F and
Zaslavsky, A (1994a) Bispectral analysis of the electroencephalogram correlates with patient movement to skin
incision during propofol/nitrous oxide anesthesia. Anesthesiology, 81(6): 13651370.
Kearse, LA, Jr., Manberg, P, DeBros, F, Chamoun, N and
Sinai, V (1994b) Bispectral analysis of the electroencephalogram during induction of anesthesia may predict
hemodynamic responses to laryngoscopy and intubation. Electroencephalogr. Clin. Neurophysiol., 90(3):
194200.
Kearse, LA, Jr., Rosow, C, Zaslavsky, A, Connors, P,
Dershwitz, M and Denman, W (1998) Bispectral analysis
of the electroencephalogram predicts conscious processing
of information during propofol sedation and hypnosis.
Kreuer, S, Biedler, A, Larsen, R, Altmann, S and Wilhelm,
W (2003) Narcotrend monitoring allows faster emergence and a reduction of drug consumption in
propofol-remifentanil anesthesia. Anesthesiology, 99
(1): 3441.
Kreuer, S, Bruhn, J, Larsen, R, Grundmann, U, Shafer, S
and Wilhelm, W (2004a) Application of bispectral
index and Narcotrend index to the measurement of the
electroencephalographic effects of isoflurane with and
without burst suppression. Anesthesiology, 101(4):
847854.
Kreuer, S, Wilhelm, W, Grundmann, U, Larsen, R and
Bruhn, J (2004b) Narcotrend index versus bispectral
index as electroencephalogram measures of anesthetic
drug effect during propofol anesthesia. Anesth. Analg.,
98(3): 692697.
Lennmarken, C, Lindholm, M-L, Greenwald, SD and
Sandin, R (2003) Confirmation that low intra-operative
BIS levels predict increased risk of post-operative mortality (Abstract). Anesthesiology, 99: A303.
Liu, N, Chazot, T, Huybrechts, I, Law-Koune, JD, Barvais, L
and Fischler, M (2005) The influence of a muscle relaxant
bolus on bispectral and datex-ohmeda entropy values during propofol-remifentanil induced loss of consciousness.
Anesth. Analg., 101: 17131718.
Locher, S, Stadler, KS, Boehlen, T, Bouillon, T,
Leibundgut, D, Schumacher, PM, Wymann, R and
Zbinden, AM (2004) A new closed-loop control system
for isoflurane using bispectral index outperforms manual control. Anesthesiology, 101(3): 591602.
Merat, S, Levecque, JP, Le Gulluche, Y, Diraison, Y,
Brinquin, L and Hoffmann, JJ (2001) BIS monitoring
may allow the detection of severe cerebral ischemia.
Can. J. Anaesth., 48(11): 10661069.
Monk, TG, Saini, V, Weldon, BC and Sigl, JC (2005)
Anesthetic management and one-year mortality after
noncardiac surgery. Anesth. Analg., 100(1): 410.
907
Myles, PS, Leslie, K, McNeil, J, Forbes, MT and Chan, A
(2004) Bispectral index monitoring to prevent awareness during anaesthesia: the B-aware randomised controlled trial. Lancet, 363: 17571763.
OConnor, MF, Daves, SM, Tung, A, Cook, RI, Thisted, R
and Apfelbaum, J (2001) BIS monitoring to prevent
awareness during general anesthesia. Anesthesiology,
94(3): 520522.
Osterman, JE and Van der Kolk, BA (1998) Awareness
during anesthesia and posttraumatic stress disorder.
Gen. Hosp. Psychiatry, 20: 274281.
Rampil, IJ (1994) Anesthetic potency is not altered after
hypothermic spinal cord transection in rats. Anesthesiology, 80: 606610.
Rampil, IJ (1998) A primer for EEG signal processing in
anesthesia. Anesthesiology, 89: 9801002.
Rampil, IJ, Kim, JS, Lenhardt, R, Negishi, C and Sessler, DI
(1998) Bispectral EEG index during nitrous oxide administration. Anesthesiology, 89(3): 671677.
Recart, A, White, PF, Wang, A, Gasanova, I, Byerly, S and
Jones, SB (2003) Effect of auditory evoked potential
index monitoring on anesthetic drug requirements and
recovery profile after laparoscopic surgery: a clinical
utility study. Anesthesiology, 99(4): 813818.
Renna, M, Handy, J and Shah, A (2003) Low baseline
bispectral index of the electroencephalogram in patients
with dementia. Anesth. Analg., 96(5): 13801385.
Rosow, C and Manberg, PJ (2001) Bispectral index monitoring. Anesthesiol. Clin. North Am., 19: 947966, xi.
Sakai, T, Singh, H, Mi, WD, Kudo, T and Matsuki, A
(1999) The effect of ketamine on clinical endpoints of
hypnosis and EEG variables during propofol infusion.
Acta Anaesthesiol. Scand., 43(2): 212216.
Sandin, RH, Enlund, G, Samuelsson, P and Lennmarken, C
(2000) Awareness during anaesthesia: a prospective
case study. Lancet, 355: 707711.
Sato, Y, Kobayashi, E, Murayama, T, Mishina, M and
Seo, N (2005) Effect of N-methyl-D-aspartate receptor
epsilon1 subunit gene disruption of the action of general anesthetic drugs in mice. anesthesiology, 102(3):
557561.
Scheller, B, Schneider, G, Daunderer, M, Kochs, EF and
Zwissler, B (2005) High-frequency components of auditory evoked potentials are detected in responsive but not
in unconscious patients. Anesthesiology, 103: 944950.
Schmidt, GN, Bischoff, P, Standl, T, Hellstern, A,
Lankenau, G, Hilbert, M and Schulte am Esch, J
(2004a) Comparative evaluation of Narcotrend, Bispectral Index, and classical electroencephalographic variables during induction, maintenance, and emergence of
a propofol/remifentanil anesthesia. Anesth. Analg., 98(5):
13461353.
Schmidt, GN, Bischoff, P, Standl, T, Hellstern, A, Teuber,
O and Schulte am Esch, J (2004b) Comparative evaluation of the Datex-Ohmeda S/5 Entropy Module and the
908
Bispectral Index monitor during propofol-remifentanil
Schneider, G, Hollweck, R, Ningler, M, Stockmanns, G and
Kochs, EF (2005) Detection of consciousness by electroencephalogram and auditory evoked potentials. Anesthesiology, 103: 934943.
Schnider, TW, Luginbuhl, M, Petersen-Felix, S and
Mathis, J (1998) Unreasonably low bispectral index
values in a volunteer with genetically determined
low-voltage electroencephalographic signal. Anesthesiology, 89(6): 16071608.
Scott, JC and Stanski, DR (1987) Decreased fentanyl and
alfentanil dose requirements with age. A simultaneous
pharmacokinetic and pharmacodynamic evaluation.
J. Pharmacol. Exp. Ther., 240(1): 159166.
Sebel, PS, Lang, E, Rampil, IJ, White, PF, Cork, R, Jopling,
M, Smith, NT, Glass, PS and Manberg, P (1997) A multicenter study of bispectral electroencephalogram analysis for monitoring anesthetic effect. Anesth. Analg.,
84(4): 891899.
Sebel, PS, Bowdle, TA, Ghoneim, MM, Rampil, IJ, Padilla,
RE, Gan, TJ and Domino, KB (2004) The incidence of
awareness during anesthesia: a multicenter United States
study. Anesth. Analg., 99: 833839.
Shannon, CE (1948) A mathematical theory of communication. Bell Syst. Tech. J., 27: 379423; 623656.
Snow, J (1847) On the Inhalation of the Vapour of Ether in
Surgical Operations. John Churchill, London.
Vakkuri, A, Yli-Hankala, A, Talja, P, Mustola, S, TolvanenLaakso, H, Sampson, T and Viertio-Oja, H (2004)
Time-frequency balanced spectral entropy as a measure

of anesthetic drug effect in central nervous system during
sevoflurane, propofol, and thiopental anesthesia. Acta
Anaesthesiol. Scand., 48: 145153.
Vereecke, HE, Struys, MM and Mortier, EP (2003) A comparison of bispectral index and ARX-derived auditory
evoked potential index in measuring the clinical interaction between ketamine and propofol anaesthesia. Anaesthesia, 58(10): 957961.
Vivien, B, Langeron, O and Riou, B (2002) Increase in
bispectral index (BIS) while correcting a severe hypoglycemia. Anesth. Analg., 95(6): 18241825.
White, PF, Tang, J, Ma, H, Wender, RH, Sloninsky, A and
Kariger, R (2004) Is the patient state analyzer with the
PSArray2 a cost-effective alternative to the bispectral
index monitor during the perioperative period? Anesth.
Analg., 99(5): 14291435.
White, PF, Tang, J, Romero, GF, Wender, RH, Naruse, R,
Sloninsky, A and Kariger, R (2006) A comparison of
state and response entropy versus bispectral index
values during the perioperative period. Anesth. Analg.,
102(1): 160167.
Whyte, SD and Booker, PD (2004) Bispectral index during
isoflurane anesthesia in pediatric patients. Anesth.
Analg., 98(6): 16441649.
Wu, CC, Lin, CS and Mok, MS (2002) Bispectral index
monitoring during hypoglycemic coma. J. Clin. Anesth.,
14(4): 305306.
Yamamura, T, Fukuda, M, Takeya, H, Goto, Y and
Furukawa, K (1981) Fast oscillatory EEG activity
induced by analgesic concentrations of nitrous oxide
in man. Anesth. Analg., 60: 283288.

M.R. Nuwer (Ed.)
909
CHAPTER 67
Monitoring of cerebral perfusion with transcranial

Doppler ultrasound
Harvey L. Edmonds*
Surgical Monitoring Associates, Inc., Louisville, KY 40207, USA
Embolization, hypoperfusion, and hyperperfusion are

all important sources of brain injury associated with
surgery and endovascular procedures. Transcranial
Doppler (TCD) ultrasound is currently the only direct
continuous measure of these pathophysiologic processes. The following description of TCD monitoring
examines both the positive attributes and limitations
of this still-evolving technology.
67.1. Principles of TCD measurement
Recent advances in ultrasound technology now permit
diagnostic color-flow imaging of large intracranial vessels. However, practical considerations generally preclude the use of transcranial imaging for monitoring.
As a result, continuous TCD monitoring primarily utilizes flow-velocity measurement rather than imaging.
Measurement is based on electrical activation of a piezoelectric crystal to initiate high-frequency vibrations.
More than two decades ago, Aaslid et al. (1982) discovered that 12 MHz ultrasound waves produced from
such crystals were capable of penetrating adult human
skull in the thinnest temporal regions superior to the
zygoma and anterior to the tragus of the ear. Probes containing both a transmitter (i.e., crystal) and receiver (i.e.,
microphone) enabled production and recording of the
ultrasonic echoes. Since erythrocytes effectively reflect
ultrasound waves, the direction and velocity of red cell
movement through the largest intracranial arteries and
veins could be determined using the Doppler principle
(i.e., the frequency shift of an echo returning from a
moving object is proportional to its velocity). By convention, flow directed toward the probe was registered
*
Correspondence to: Harvey L. Edmonds, Jr., Ph.D., Cardiovascular Services, Surgical Monitoring Associates,
Inc., 830 Huntington Road, Louisville, KY 40207, USA.
Tel. 1-502-262-3976; fax: 1-502-290-1751.
E-mail: lharvo@louisville.edu (H.L. Edmonds).
as positive velocity, while flow directed away from

the probe had negative velocity.
Ultrasonic insonation of large vessel laminar blood
flow creates a series of echoes with differing frequencies. As with quantitative electroencephalography
(EEG), Fourier spectral analysis has been used to
visualize the pattern of the TCD-shift frequency distribution. Analogous to the EEG spectral edge frequency,
the upper edge of the Doppler spectrum was termed the
flow-velocity envelope. It represents an echo from
the highest velocity erythrocytes in the center of the
vessel lumen. Display of individual spectra in rapid
succession, in a process resembling the EEG density
spectral array, permits characterization of time-related
changes in the TCD flow-velocity spectral pattern.
During one cardiac cycle, changes in the time domain
spectral display produce a waveform that resembles
the familiar arterial pressure tracing (Fig. 1). Thus,
peak systolic, end-diastolic, and mean flow velocity
derived from the frequency spectral waveform relate
to the cyclic alterations in both local cerebral blood
flow and flow velocity.
Insonation depth, representing the distance from
the skin to the intra- or extracranial vascular segment
under investigation, is controlled by selection of the
time interval (i.e., gating) during which pulsed echoes
are recorded. The use of multiple gates permits simultaneous interrogation of several vascular segments
within the same or neighboring vessels. In addition to
depth, the intensity (i.e., power) of the ultrasonic transmission, amplifier gain, and audio characteristics of
the echo recording as well as length of the interrogated
vascular segment (i.e., sample volume in millimeters)
are user-controlled. Only a few of the currently available sonographs enable selection of the very small
sample volumes appropriate for neonates and infants.
Those clinicians planning pediatric applications for
TCD monitoring should carefully consider this hardware limitation prior to purchase.
H.L. EDMONDS
10
+3
+120
+2
+80
+1
+40
30
40
50
60
70
80
+4
30
40
50
60
70
80
+3
+120
+2
+80
+1
+40
kHz
SAMPLE
1
RMCA
DEPTH
mm
VESSEL
46
POWER
30
9
VOLUME
44
2
kHz
SAMPLE
cm/s
47
POWER
30
40
50
60
70
80
mm
DEPTH
30
40
50
60
70
80
+4
cm/s
LMCA
mm
VESSEL
mm
910
Fig. 1. The transcranial Doppler (TCD) time domain spectral displays of Doppler shift frequencies (left vertical axis) represent erythrocyte velocity (right vertical axis) within the left (second panel) and right (fourth panel) middle cerebral artery.
The velocity measurement is made within a small segment of the artery. The distance from the ultrasound probe face to the
central point of this vascular segment is termed insonation depth and typically displayed adjacent to the TCD spectra. The
alternative M-mode display represents echoes simultaneously recorded from a wide vascular segment. The M-mode permits
detection of high-intensity microembolic signals (MES), shown here as dark transients, throughout the entire vascular segment, whereas the spectral display examines only a single small region.
M-mode ultrasound display extends the multigating

process through the use of many overlapping gates
(Fig. 1). As a result, the intensity of color-coded echoes
is used to visualize the direction and magnitude of flowvelocity signals over a wide-depth range (e.g., 50 mm).
In addition to characterizing cerebral hemodynamics, TCD allows detection of cerebral emboli. Intravascular foreign material, gas or particulate, is echogenic
because of acoustic impedance exceeding that of erythrocytes. These high-intensity transients thus represent microembolic signals (MES; Fig. 1). Multigating
(Fig. 2) and M-mode (Fig. 1) facilitate discrimination
between these pathologic ultrasonic signals and highintensity artifacts resulting from radiofrequency interference or acoustic coupling. Discrimination is based
on the principle that emboli propagate with blood
motion, while artifacts do not. For example, Smith
et al. (1996) used a 10 mm gate separation to distinguish emboli from artifact. Time delay in detection
of the high-intensity transients was 11.04 ms for
emboli and 0.08 ms for artifacts. M-mode further
enhances characterization of emboli since it permits
visualization of their movement through the vascular

tree.
TCD is most frequently used as a diagnostic aid.
With this spatial domain mode, insonation through
the orbits, foramen magnum and temporal ultrasonic
windows, and submandibular site allows one-time
flow-velocity spectra and/or M-mode images to be
obtained from key large intra- and extracranial arteries
and veins of both the anterior and posterior circulation.
Regional cerebral hemodynamic abnormalities are
inferred by comparison of the patients spectral waveform characteristics with vessel-specific age-corrected
normative values.
In contrast, the TCD time domain surgical monitoring mode involves continuous insonation of one key
intracranial artery from each hemisphere. Cerebral
hemodynamic abnormality is inferred by assessing the
magnitude of change in flow-velocity spectral characteristics from a preincision individualized reference.
If the insonation angle, vessel diameter, and blood
viscosity remain nearly constant, change in mean
flow velocity reflects similar change in volume flow
OTHER ISSUES
911
2.8
100
50
50
50
50
100
100
50
50
50
50
100
Vm
dB Cm/S
33dB
13:03:09
ms
50
Fig. 2. The left panels depict multigated spectral displays of flow-velocity spectra obtained from the middle cerebral artery
at insonation depths of 49 and 54 mm during nonpulsatile cardiopulmonary bypass. The white transients represent microembolic signals (MES). The left panels visualize the acoustic signature of these emboliform high-intensity transients.
The 2.8 ms difference in onset delay confirms that the signals are of embolic origin and not acoustic artifact.
(Clark et al., 1996). Constancy of insonation angle may

be achieved by mechanical fixation of the ultrasound
probe to the cranium at the site of an acoustic window.
Stability of middle cerebral artery diameter has been
directly observed during craniotomy, even with substantial alterations in blood pressure and arterial CO2 tension
(Giller et al., 1993). Since this artery typically carries
40% of the hemispheric blood flow, it is the favored site
for surgical TCD monitoring. A relatively constant
blood viscosity may be anticipated except during major
blood loss, hemodilution, transfusion, or hypothermia.
However, even in these situations, TCD still may provide valuable information as discussed below.
Clinically noteworthy velocity change has been
established by the relationships between blood flow
velocity and apparent cerebral dysfunction or clinical
outcome. The first approach is exemplified by the
TCD spectral changes observed in both conscious
and anesthetized patients. During tilt-table testing,
syncope is preceded by either a >60% decrease in
mean flow velocity or an end-diastolic flow velocity
decline to zero (Edmonds et al., 1998). Diastolic flow
cessation occurs as the postcapillary transmural pressure gradient becomes insufficient to prevent venular
collapse (i.e., critical closing pressure). Although these
changes do not necessarily represent ischemic injury
per se, they identify hypoperfusion of a magnitude sufficient to disrupt normal cerebral function. With
carotid endarterectomy under general anesthesia,
clamp-related decreases in mean velocity greater than
60% result in blood flow less than 20 ml/100 g/min
and pathologic EEG suppression (Jrgensen, 1995).
Outcome-oriented relationships have identified
the risk of brain injury associated with a specified
decrease or increase in flow velocity. Thus, during

carotid endarterectomy Halsey et al. (1989) found that
a flow-velocity decline to <20% of preincision baseline was related to significantly increased stroke risk.
67.2. Details of TCD monitoring
Each of the following issues is important in the
acquisition of high-quality information on cerebral
hemodynamics that is of maximum benefit to both
patient and surgical team.
67.2.1. Selection of monitoring site(s) and method
of probe fixation
The choice of vessel(s) to be monitored is dictated by
patient anatomy and characteristics of the surgical procedure. With supine patient positioning, the middle
cerebral artery is favored because of its large flow volume and relative ease of identification. After bilateral
probe placement on temporal indentations (Fig. 3),
positive pulsatile flow-velocity spectra obtained at
depths of 55 mm or less in the adult cranium arise from
the middle cerebral artery or one of its branches. Three
temporal windows are often present: the anterior,
middle, and posterior.
Unfortunately, useful spectra may be unobtainable
because of intracranial vascular disease and/or temporal hyperostosis. The latter occurs most frequently
in geriatric females, particularly those of African
American or Oriental ancestry. If transtemporal insonation is ineffective and insonation at a site proximal
to the circle of Willis is sufficient, the extracranial
portion of the internal carotid may be used via
912
H.L. EDMONDS
Fig. 3. The left two panels illustrate temporal window probe attachment systems. The system shown in the lower panel relies
on a restrictive headband, while the one in the upper panel does not. The right panels show that both restrictive and nonrestrictive systems may be used to fix the probe at a submandibular site in patients lacking a useful temporal ultrasonic window.
submandibular probe fixation (Topcuoglu et al.,

2003). With our aging and progressively atherosclerotic patient population, there is a growing necessity
for this unconventional site selection. Since not all
TCD manufacturers offer the capability of submandibular fixation, this feature should be carefully considered prior to new hardware purchases.
Neonates and infants offer a special challenge
because their soft cranium precludes the use of
restrictive headbands to secure probe holders. Some
manufacturers do offer small holders that attach
directly to skin with self-adhesive disks. Insonation
of the vertebrobasilar region of the posterior circulation in the prone patient represents another challenge.
There are currently no commercially available probe
fixation systems specifically designed for this
approach. However, some of the holders that rely
on stabilization by either soft elastic headbands or
self-adhesive disks may be used in this circumstance.
Final insonation site selection and proper probe
orientation are most challenging if not attempted
until after anesthetic induction. Optimal efficiency
and effectiveness are achieved by establishing them
prior to the patients entry into the operating room.

The newest generation of small battery-operated
TCD monitors makes preoperative bedside assessment practical.
67.2.2. Vasomotor reserve and cerebral
autoregulation
An appreciation of vasomotor reserve is important in
the management of patients under general anesthesia
(Fig. 4). With normal reactivity of cerebral arterioles
to CO2 of 4%/mm, hyperventilation-induced hypocapnia can readily produce velocity decreases of
>30% and EEG evidence of cerebral ischemia
(Halpern et al., 2003). As a consequence of cerebrovascular disease, intracranial arteries may not necessarily be equally reactive to CO2. Measurement of
individualized responses of the left and right middle
cerebral artery flow-velocity signals to CO2 change
are necessary to define hemispheric vasomotor
reserve (Brown et al., 1986; Goto et al., 2001; Yoon
et al., 2001; Fig. 5). Although CO2 reactivity is a process distinct from pressure autoregulation, the two
OTHER ISSUES
913
120
CM/S
Middle cerebral
3.00
kHz
40
CM/S
Velocity 60 cm/s
Temporal artery
3.00
kHz
Velocity 22 cm/s
CO2 38
66% increase
14% increase
CO2 48
120
CM/S
Velocity 100 cm/s
3.00
kHz
3.00
kHz
40
CM/S
Velocity 25 cm/s
Fig. 4. This figure illustrates the increased sensitivity of intracranial arterioles to changes in arterial CO2 tension. Note that
a 10 mm Hg increase in end-tidal CO2 produced a 66% increase in peak middle cerebral artery velocity compared with only
a 14% in the ipsilateral temporal artery.
Left MCA
at which cerebral blood flow becomes dependent on

cerebral perfusion pressure). Buijs et al. (1992) were
among the first to use TCD to document the preservation of autoregulation during pediatric surgery.
Importantly, they noted disruption of autoregulation
during cardiopulmonary bypass, particularly during
deep hypothermic circulatory arrest (DHCA). With
TCD monitoring, Neri et al. (2004) found that the
negative impact of DHCA on autoregulation could
also be established in adult patients. They further
showed that selective cerebral perfusion through
carotid arteries during systemic circulatory arrest preserved autoregulation.
+80
+40
cm/s
are interrelated. Near-normal reactivity is typically a

precondition for the pressure independency of cerebral blood flow (Van Lieshout et al., 2003). Vasomotor reserve is influenced by arterial oxygen tension so
that CO2 reactivity is significantly increased under
the hyperoxic conditions typical to general anesthesia
(Johnston et al., 2003).
Knowledge of the individualized cerebral autoregulatory response provides valuable information for
the optimization of blood pressure management.
Examination of the relationship between mean arterial pressure and blood flow velocity establishes the
autoregulatory range and its lower limit (i.e., the point
0
+120
+80
+40
cm/s
Right MCA
Fig. 5. CO2 can be quickly assessed in the anesthetized patient by examining the magnitude of cyclic fluctuation in peak
cerebral blood flow velocity associated with mechanical ventilation. Note the diminished fluctuation in the left middle cerebral artery (MCA) spectral trace, suggesting reduced vasomotor reserve and dysautoregulation. In such patients, careful
control of systemic pressure is required to avoid regional cerebral hypoperfusion.
914
Preoperative assessment of cerebral autoregulation is of little value for surgical monitoring because
of the confounding influence of general anesthesia.
Nitrous oxide (Iacopino et al., 2003) and all of the
volatile anesthetic agents increase cerebral blood
flow and velocity through dilation of arteriolar resistance vessels (Matta et al., 1999). As a result, these
anesthetics may disrupt autoregulation, producing
pressure-passive cerebral perfusion (Bedforth et al.,
2000). The extent of the disruption appears to be
agent specific and unpredictable. Thus, autoregulation should be objectively established by examination
of the individualized arterial pressurecerebral blood
flowvelocity relationship after the establishment of
surgical anesthesia and final patient positioning.
Hypotension associated with spinal anesthesia
may also result in pressure-passive cerebral perfusion
in susceptible patients. For example, during spinal
anesthesia for hernia repair in a group of former preterm infants, blood flow velocity was linearly related
to mean arterial pressure (Bonnett et al., 2004).
67.3. Surgical, procedural, and anesthetic
applications of TCD monitoring
67.3.1. Cerebral hypoperfusion
67.3.1.1. Inflow obstruction
67.3.1.1.1. Inadequate collateral flow. Carotid
occlusion during endarterectomy represents one of
the most common causes of iatrogenic blood flow
obstruction to the brain. Low power orbital insonation
enables detection of carotid siphon vascular abnormalities not dectectable by submandibular duplex Doppler
scan. Without knowledge of the functional status of the
intracranial carotid and circle of Willis, the complete
hemodynamic significance of a carotid bifurcation
lesion cannot be appreciated (Doblar, 1996).
Spencer et al. (1992) examined the relationship
between clamp-related decreases in flow velocity and
carotid artery stump pressure. An exponential function
was described with zero velocity occurring at a stump
pressure of 15 mm Hg. The authors concluded that
TCD provided an excellent indicator as to the necessity of shunting. Similarly, Kalra et al. (1994) found
that patients with stump pressures below 30 mm Hg
had significantly lower flow velocities than those with
higher pressures.
Permanent ligation of a carotid artery often occurs
during venoarterial extracorporeal membrane oxygenation (Crombleholme et al., 1990) and neurosurgical
H.L. EDMONDS
treatment of intracranial aneurysms and tumors. In

both circumstances, carotid sacrifice in the absence
of effective collateral perfusion may lead to cerebral
infarction. Preprocedural TCD with manual (Giller
et al., 1994) or intravascular balloon (Schneweis
et al., 1997) carotid occlusion has been used to identify
hemispheric dependence on ipsilateral carotid flow.
The appearance of neurologic sequelae correlated
with the magnitude of the relative velocity decreases
from the individualized preocclusion reference, but
not with the absolute velocity. With both occlusion
techniques, observed neurologic signs of transient
focal deficit occurred consistently with velocity
decreases greater than 65%.
An accurate appreciation of collateral flow is also
essential with attempted selective cerebral perfusion
during DHCA for aortic arch reconstruction (Neri
et al., 2004). Doblar (2004) summarized the evidence
that functionally inadequate collateral flow from a single carotid or axillary arterial cannula may be expected
in more than one-quarter of adult vascular surgery
patients. Bilateral TCD monitoring provides the only
means to directly document the initiation and maintenance of bihemispheric selective antegrade cerebral
perfusion through large arteries.
Temporary ligation of an intracranial artery is usually necessary for surgical treatment of a giant aneurysm. Patient outcome is dependent, in part, on the
functional integrity of collateral vessels. Preoperative
TCD monitoring during manual carotid compression
defines collateral perfusion, while intraoperative
microvascular Doppler documents collateral flow during temporary surgical (Boecher-Schwarz et al., 1995)
or endovascular (Yamasaki et al., 2000) compression
or occlusion of intracranial vessels.
67.3.1.1.2. Balloon occlusion of intracranial
artery. TCD has been used to monitor the effectiveness of balloon angioplasty for treatment of symptomatic vasospasm following subarachnoid hemorrhage
(Newell et al., 1989). An absence of velocity increase
after dilation consistently and correctly predicted poor
outcome.
67.3.1.1.3. Perfusion catheter or balloon occlusion
of aortic arch vessels. Cardiac surgery provides ample
opportunity for TCD monitoring to detect and correct
inflow obstruction. Minimally invasive port-access
coronary artery bypass utilizes a balloon-tipped catheter to occlude the aorta and deliver cardioplegia solution. Unexpected balloon migration with resulting
OTHER ISSUES
915
innominate occlusion has been promptly detected and

successfully corrected with the aid of bilateral TCD
monitoring of the middle cerebral artery (Grocott
et al., 1998). TCD is particularly useful for the detection of balloon migration because the right mainstem
bronchus and esophagus, may obscure transesophageal echocardiographic (TEE) imaging of the aortic
arch. Bilateral radial artery pressure monitoring alone
may be insufficient to detect innominate occlusion.
Pressure recordings may be of poor quality, asymmetric, or confounded by the initiation of cardiopulmonary bypass.
67.3.1.1.4. Malperfusion syndrome. With an acute
dissection involving the ascending aorta or arch, initiation of cardiopulmonary bypass via a femoral
artery perfusion cannula may result in a sudden flow
obstruction of one or more cranial vessels. Flow
through the false lumen may force a portion of torn
intima across the opening of arch vessels. Since the
obstruction may not involve upper limb arterial pressure monitoring sites, early detection and correction
by TCD is essential to avoid catastrophic brain injury
(Fig. 6).
67.3.1.1.5. Retrograde cerebral perfusion. An
unusual type of inflow obstruction may occur with
attempted retrograde cerebral perfusion (RCP) and
deep hypothermic systemic circulatory arrest during
aneurysm repair involving the aortic arch. Interest
in this perfusion technique continues due to demonstrations of improved outcome with RCP compared
with that expected from total circulatory arrest
(Appoo et al., 2006). However, RCP remains controversial partly because of technical uncertainties
80
CM/S
Left MCA
MEAN
02:05
22:05
100
CM/S
involved in the initiation and maintenance of retrograde flow through the cerebral venous system. The
uncertainties include the presence of a functional
valve in the internal jugular vein (Okamoto et al.,
1993), flow restriction from vascular torsion, flow
redirection into the azygous drainage, and perfusion
pressure insufficient to reexpand collapsed cerebral
veins (Doblar, 2004). Using both TCD and transcranial near-infrared spectroscopy, our group observed
that perfusion pressure and flow rate well in excess
of the oft-recommended 25 mm Hg and 300 ml/min
upper limits (Appoo et al., 2006) may be needed
initially to expand venous structures and start retrograde flow (Ganzel et al., 1997; Fig. 7). These observations were later confirmed by Estrera et al. (2003)
using M-mode TCD to guide RCP perfusion management. RCP with a middle cerebral artery mean velocity of 15 cm/s was produced in 39/40 operations with
an average perfusion pressure of 32 10 mm Hg.
67.3.1.2. Outflow obstruction
Selective decrease in diastolic flow velocity represents
a functional increase in cerebral vascular resistance.
Declining diastolic velocity may be due to decreased
input energy (i.e., lower stroke volume or myocardial
contractility; Fig. 8) or resistance increase (i.e., hypocapnic arteriolar constriction or venular mechanical
compression). Rodriguez et al. (1997) used a loss of
diastolic velocity to detect malposition of a superior
vena caval perfusion cannula prior to the onset of
cardiopulmonary bypass.
Cerebral swelling initially compromises perfusion
by mechanical obstruction of compliant intracranial
venous structures. This circumstance may be readily
detected by TCD as diastolic flow reversal (Fig. 9).
0
2.50
kHz
Right MCA
MEAN
60
CM/S
22:05
02:05
80
CM/S
0
2.00
kHz
Aortic X-clamp
0
011117GR
Fig. 6. Application of the aortic cross-clamp resulted in an immediate loss of the transcranial Doppler (TCD) flow-velocity
signal in both the left and right middle cerebral arteries (MCA) during this surgical correction of an acute aortic arch dissection. The surgeon immediately discontinued cardiopulmonary bypass and selected an alternative cannulation site. TCD
confirmed the maintenance of bilateral cerebral perfusion upon return to bypass.
916
H.L. EDMONDS
30 Antegrade
20
Retrograde
10
0
10
Fig. 7. The left portion of this right middle cerebral artery

flow-velocity spectrum shows flow cessation during the
brief period of total circulatory arrest needed for initiation
of retrograde cerebral perfusion. Successful initiation and
maintenance of retrograde cerebral flow is documented in
the right portion of the figure as a negative velocity spectrum (i.e., reversal in flow direction).
Guerra et al. (1999) described the benefits of decompressive craniectomy and the role of TCD in documenting improved cerebral hemodynamics.
67.3.2. Cerebral hyperperfusion
Powers and Smith (1990) and later Spencer (1997)
identified postocclusion hyperperfusion as an
important cause of cerebral infarction following
carotid endarterectomy. Muller et al. (1999) confirmed these observations by finding a 14% incidence of pathologically elevated middle cerebral
artery blood flow velocity ipsilateral to the carotid
endarterectomy site. Twenty percent of the highvelocity cases experienced a clinical hyperperfusion
syndrome.
140
CM/S
Stroke volume decrease
Selective antegrade cerebral perfusion during deep

hypothermic systemic circulatory arrest has been
used in both adult and pediatric patients during
reconstruction of the aortic arch. Arterial flow from
the cardiopulmonary bypass machine is directed to
the brain through one or two extracranial arteries.
Particularly in neonates and infants, there is risk of
iatrogenic cerebral hemorrhage from hyperperfusion.
Under these circumstances, TCD permits immediate
detection and correction of excessive pump flow
(Ganzel et al., 1997; Fraser and Andropoulos, 2004).
67.3.3. Flowmetabolism uncoupling
Normally, cerebral blood flow and metabolism are
tightly coupled. As a result, TCD may aid in the detection of hypermetabolic states such as that associated
with seizures. Boylan et al. (1999) described transient
velocity increases associated with EEG evidence of
seizures, independent of changes in arterial pressure.
Underlying pathology may also lead to an uncoupling of blood flow and metabolism. In patients with
fulminant hepatic failure, both uncoupling and dysautoregulation occur (Ardizzone et al., 2004). During
orthotopic liver transplantation, recoupling occurs in
parallel with a return in liver function. By the end of
surgery, autoregulation may be restored.
Uncoupling of cerebral blood flow and metabolism may also be iatrogenic. During cardiopulmonary bypass with planned DHCA for aortic arch
000214MHZ
0
140
CM/S
Increase
000214MHZ
Fig. 8. Rapid decline in stroke volume is manifested primarily as a decrease in end-diastolic flow velocity (upper panel).
Improvement in stroke volume reverses this transcranial Doppler (TCD) pattern.
OTHER ISSUES
917
120
CM/S
011101Js
Fig. 9. This right middle cerebral flow-velocity spectrum

shows reversal of flow direction during diastole, an established indication of antegrade flow cessation in an obtunded
patient after clipping of a cerebral aneurysm. Following
removal of the bone flap, the absence of diastolic flow reversal served as a harbinger of clinical improvement.
reconstruction, rapid, deep cooling may induce cerebral vasoparesis that becomes evident during
rewarming (Fig. 10). As a result, cerebral blood flow
velocity remains low and unchanged, despite the
temperature-induced exponential increase in brain
metabolic demand (Greeley et al., 1989). This mismatch between oxygen demand and supply represents ischemia. Erlich et al. (2001) demonstrated
that following arrest, an initial brief period of cold
reperfusion prior to rewarming facilitated recoupling
of flow and metabolism. Examination of the relationship between flow velocity and cranial temperature
during rewarming offers a simple method to document the success of this strategy.
67.3.4. Embolization
Paradoxical cerebral embolism may occur in the presence of a right-to-left intracardiac shunt. The incidence
PEAK
11:55
67.3.4.1.2. Atheroma. Aortic manipulation and

instrumentation may disrupt atherosclerotic plaque
and lead to cerebral infarction. Numerous studies have
100
CM/S
Uncoupled
CA
15:55
0
XXXX
100
CM/S
3.00
kHz
0
0
67.3.4.1. Particulate
67.3.4.1.1. Lipids. Fat embolism may result from
long bone fracture or arthroplasty. Riding et al.
(2004) used contrast TCD to identify right-to-left cardiac shunt in a series of arthroplasty patients. Intraoperative MES were recorded in 44% of the patients,
with the embolic count proportional to the TCD estimate of shunt size. Large aggregate MES counts preceded postoperative confusion and/or pancreatitis.
Lipid microemboli may also be infused into the
brain during cardiopulmonary bypass. Their primary
source appears to be the fat-containing cardiotomy
suction material. Deformable lipid spheres may pass
from the venous reservoir through the porous membrane oxygenator and arterial filter into the cerebral
circulation (Brooker et al., 1998).
100
CM/S PEAK
Coupled
SCP
140
CM/S
of patent foramen ovale was recently determined to be

15% in healthy control subjects, but a higher 27% in
patients with obstructive sleep apnea (Beelke et al.,
2003). Droste et al. (1999) compared TCD and TEE in
shunt identification. Using microbubble-containing
contrast media and a Valsalva maneuver, TCD detected
MES indicative of a shunt with a sensitivity of 100%
and specificity of 65% compared with TEE. Recently,
contrast TCD and intracardiac echocardiography were
used to guide endovascular closure of a patent foramen
ovale in a patient too ill to tolerate TEE and general
anesthesia (Zanchetta et al., 2003).
XXXX
2.50
kHz
Fig. 10. The left upper panel is a right middle cerebral artery peak flow-velocity trend obtained during the cardiopulmonary
bypass portion of an aortic arch reconstruction involving a period of deep hypothermic circulatory arrest with selective cerebral
perfusion (SCP) via the right axillary artery. Note the maintenance of low flow velocity during SCP and the marked increase
during rewarming (i.e., flow-metabolism coupling). In contrast, the right upper panel shows loss of signal in another patient
during total circulatory arrest (CA) and an absence of large velocity increase during rewarming (i.e., uncoupling).
918
used TCD to evaluate technical advances designed to

reduce aortic manipulation (Calafiore et al., 2001) or
trap atheromatous emboli (Eifert et al., 2003). TCD
monitoring with conventional cardiopulmonary bypass
technology was used to identify the optimal site for
aortic cannulation (Borger et al., 1999).
67.3.4.1.3. Thrombus. Spencer et al. (1990) used
TCD to detect both gaseous and particulate MES during and after carotid endarterectomy. Intraluminal
platelet thrombus was the primary source of particulate MES. In the early postoperative period, the persistent presence of many MES was associated with
the development of new cerebral infarction. Levi
et al. (1998) confirmed and extended these findings.
They found that an MES rate of >1/min during the
first postoperative hour was highly predictive for
the development of ipsilateral focal cerebral ischemia. These results were confirmed by Naylor et al.
(2000).
Thromboemboli are a potential etiology for neurologic deterioration following cerebral aneurysm
surgery. Giller et al. (1998) determined a 4.4% incidence of MES in postsurgical aneurysm patients.
CT evidence of cerebral ischemia occurred in 82%
of patients with MES compared with a 24% incidence in MES-free patients.
H.L. EDMONDS
67.3.4.2.2. Massive gas embolization. Our group

reported on the use of TCD in the detection and successful management of massive air embolism during
pediatric cardiac surgery (Yeh et al., 2003). The
embolism was manifested by the continuous highintensity signal with the acoustic signature of gas
bubbles (Fig. 11). After rapid deep cooling, a brief
period of RCP forced cerebral gas bubbles into the
aortic root for aspiration. This experience was consistent with previous recommendations for emergent
RCP in the treatment of this catastrophic event (Mills
and Ochsner, 1980; Brown et al., 1987).
67.4. Limitations of TCD monitoring
67.4.1. Effective monitoring is user and patient
dependent
Because temporal bone may distort ultrasonic beam
shape in variable and unpredictable ways (Deverson
et al., 2000), success in obtaining a useful TCD
depends on patient cranial anatomy as well as sonographer experience and continued practice (Shen et al.,
1999). The extreme variation in interuser TCD success
cm/s
100
80
67.3.4.2. Gaseous
67.3.4.2.1. Microemboli. Numerous studies spanning nearly 15 years [Padayachee et al. (1987,
1988) to Rodriguez et al. (2005)] have described
the presence of MES associated with the use of cardiopulmonary bypass. The aggregate number of
MES was influenced by the choice of aortic cannula
design and cannulation site, oxygenator, venous reservoir, arterial filter, cardiac vent system, and perfusion technique.
Studies have also reported the continuous appearance of MES in patients with prosthetic heart valves.
The rate of MES generation is influenced by valve
design, but the clinical significance of these devicedependent differences is not well understood (Georgiadis et al., 1996). Kaps et al. (1997) used inspiration of 100% oxygen to conclude that the MES
appeared to be predominantly of gaseous composition. High oxygen tension reduced cavitation effects
and lowered the MES rate to 2% of that observed
during room air inspiration. In vitro studies subsequently confirmed that the MES were, indeed, cavitation bubbles (Potthast et al., 2000).
60
40
20
0
20
Trend
180
cm/s
0
13:00
13:30
14:00
Fig. 11. The upper panel indicates the appearance of continuous high-intensity signals suggestive of massive air
embolism. Prior to initiation of heroic treatment measures,
the cerebral consequences of this ultrasonic storm were
confirmed by multilead EEG and bilateral cerebral oximetry. The spike in the right middle cerebral artery flowvelocity trend shown in the lower panel is artifactual, but
does identify the precise time of embolization.
OTHER ISSUES
in obtaining transtemporal signals is exemplified by

two contemporaneous carotid endarterectomy studies
from neighboring institutions. Jansen et al. (1993) succeeded in recording TCD signals in 130 consecutive
surgeries, whereas Dinkel et al. (1994) were unable
to produce useful ultrasonic waveforms in 40% of
176 operations. A recent study duplicated the latter
experience, failing to obtain TCD signals in 37% of
91 operations (Rowed et al., 2004).
67.4.2. Absolute cerebral blood flow velocity
is unrelated to flow
Velocity and flow are differentially influenced by
physiologic variables such as blood viscosity. Not surprisingly, Sohn et al. (1997) and Metry et al. (2002)
found significant inverse correlation between velocity
and hematocrit, while serum fibrinogen only weakly
influenced velocity.
Giller et al. (1995) reported on four cases of
abnormally elevated velocity following arterial dilation for treatment of cerebral vasospasm. Angiography revealed the high velocities to be associated
with excessive dilatation (i.e., hyperemia) rather than
persistent vasospasm.
Of greatest concern is the inability of current
sonographs to distinguish accidental probe movement
from flow cessation. In either case, the spectral or
M-mode display vanishes, but the cause is uncertain.
67.4.3. Automated detection of MES is unreliable
In 1997, an international expert panel concluded that
no current TCD system of automatic embolus detection had the required sensitivity and specificity for
clinical use (Ringelstein et al., 1998). Near the end of
the subsequent decade, the expert consensus had not
changed, as they opined that the sensitivity and positive predictive value of automated algorithms to detect
MES using either 1 or 2 MHz frequencies are unacceptably low for clinical practice (Droste et al.,
2005). Similarly, the newly developed dual-frequency
TCD appeared to better discriminate between gaseous
and particulate emboli than single-frequency devices,
but it is not accurate enough for use in clinical or
research studies (Markus and Punter, 2005). Thus,
current TCD devices cannot determine either the size
or composition of emboli and, at best, provide a semiquantitative estimate of aggregate embolic count.
Curiously, current sonographs have automated
detection of individual emboliform high-intensity
919
transient signals but not of continuous activity

signifying massive gas embolism. This potentially
lethal event is detectable only by continuous vigilance of the TCD graphic display or the audio signal.
The clinical significance of detected MES is controversial. Although some studies have found correlation between the number of MES and signs of
neurologic injury or cognitive decline, others have
not. A multicenter trial of 6 different prosthetic heart
valves revealed no difference in MES between
patient with and without neurologic complications
(Georgiadis et al., 1996). The authors speculated that
most of the detected MES were tiny and short-lived
cavitation bubbles of limited hemodynamic consequence. A similar lack of difference was observed
in a comparison of coronary artery bypass surgery
with and without (i.e., off-pump) cardiopulmonary
bypass. Although the average aggregate embolic
count in off-pump operations was significantly lower,
there were no differences in the incidence of postoperative neurologic injury or neuropsychometric performance (Lund et al., 2003). Ferrari et al. (2004)
performed TCD monitoring during transcatheter closure of patent foramen ovale and atrial septal defects.
MES were detected in 96% of the procedures with an
aggregate maximum count of 25. Despite the high
incidence of cerebral embolization, no clinically
apparent neurological or neuropsychological deficits
were observed.
67.5. Impact of TCD monitoring on surgical
outcome
In contrast to surgical monitoring with quantitative
EEG (Myles et al., 2004) or cerebral oximetry
(Murkin et al., 2004; Casati et al., 2005), TCD lacks
adequately powered randomized prospective evidence
of improved clinical outcome or economic benefit.
Nevertheless, as with TEE, retrospective or uncontrolled prospective studies have documented the association of ultrasonic monitoring with improved
clinical benefit. Spencer (1997) used a retrospective
series of 500 carotid endarterectomy surgeries to
examine the clinical impact of intraoperative TCD
monitoring. As the surgeons learned to interpret the
information provided by TCD, the incidence of neurologic complication significantly declined from 7% in
the first 100 operations to 2% in the remaining 400.
Post hoc TCD analysis identified the causes of neurologic injury as embolic (54% of injuries), hyperperfusion (29%), and hypoperfusion (17%).
920
Lennard et al. (1997) used a prospective nonrandomized design to define perioperative TCD monitoring impact on carotid endarterectomy outcome.
Postoperatively, TCD monitoring continued for 4 h.
The appearance of MES at rates exceeding 2/min in
the first postoperative hour initiated antiplatelet
therapy. In this 100-patient study, the combined
intraoperative and postoperative TCD monitoring
resulted in a 0% incidence of perioperative morbidity
and mortality.
Our group has obtained evidence of TCD clinical
and economic benefit through studies utilizing this
technology as part of a multimodality approach (i.e.,
multichannel EEG, TCD, and cerebral oximetry) to
surgical neuromonitoring. Austin et al. (1997) retrospectively examined outcome in pediatric cardiac surgical patients before and after introduction of
neuromonitoring. Interventions were made in 70% of
the monitored patients, half of which were initiated
by TCD. Monitored patients had significant reductions
in neurologic complications and hospital costs.
Edmonds (2002) retrospectively examined outcome
in two groups of coronary artery bypass operations
performed by a single surgeon. Compared with the
unmonitored cohort, the monitored cohort demonstrated significant reductions in the incidence of neurologic injury, length of hospital stay, and total
hospital charges. Neuromonitoring-initiated interventions altering patient management occurred in 59%
of the operations. Interventions attributed primarily
to TCD occurred in 13%.
67.6. Summary
TCD monitoring provides a continuous measure of
change in cerebral hemodynamics that is not available
by any other technology. The information is clinically
valuable and, in some cases, potentially lifesaving.
Despite these attributes and a two-decade experience,
this promising ultrasound technology remains substantially underutilized. As with TEE, those individuals
willing to acquire the requisite TCD skills and adapt
to its technical limitations can expect to reap the
rewards of improved patient care.
Acknowledgments
The author is grateful for the technical assistance of
Ermina Mujadzic, MD, Aida Sehic, MD, Mary
Thomas, M.A., and Henry Ton, MD, and the editorial
assistance of Jeanne F. Edmonds. Neither the author
H.L. EDMONDS
nor any family member has a personal financial relationship with any ultrasound manufacturer; however,
each of the following companies has provided research
support: DWL Instruments, Medasonics Neuroguard,
Viasys Neurocare, Spencer Technologies, and Terumo.
Studies conducted by our group cited above were
supported, in part, by the WHAS Crusade for Children,
Kosair Childrens Hospital, and the other hospitals of
Norton Healthcare and Jewish Hospital Heart and Lung
Institute, all of Louisville, KY, USA.
References
Aaslid, R, Markwalder, TM and Nornes, H (1982) Noninvasive transcranial Doppler ultrasound recording of
flow-velocity in basal cerebral arteries. J. Neurosurg.,
57: 769774.
Appoo, JJ, Augoustides, JG, Pochettino, A, Savino, JS,
McGarvey, ML, Cowie, DC, Gambone, AJ, Harris, H,
Cheung, AT and Bavaria, JE (2006) Perioperative outcome in adults undergoing elective deep hypothermic
circulatory arrest with retrograde cerebral perfusion in
proximal aortic arch repair: evaluation of protocolbased care. J. Cardiothorac. Vasc. Anesth., 20: 37.
Ardizzone, G, Arrigo, A, Panaro, F, Ornis, S, Colombi, R,
Distefano, S, Jarzembowski, TM and Cerruti, E (2004)
Cerebral hemodynamic and metabolic changes in
patients with fulminant hepatic failure during liver
transplantation. Transplant. Proc., 36: 30603064.
Austin, EH, III, Edmonds, HL, Jr., Auden, SM, Seremet, V,
Niznik, G, Sehic, A, Sowell, MK, Cheppo, CD and
Coprlett, KM (1997) Benefit of neurophysiologic monitoring for pediatric cardiac surgery. J. Thorac. Cardiovasc. Surg., 114: 707715.
Bedforth, NM, Hardman, JG and Nathanson, MH (2000)
Cerebral hemodynamic response to the introduction of
desflurane: a comparison with sevoflurane. Anesth.
Analg., 91: 152155.
Beelke, M, Angeli, S, Del Sette, M, Candolfo, C, Cabano,
ME, Canavaro, P, Nobili, L and Ferrillo, F (2003) Prevalence of patent foramen ovale in subjects with obstructive sleep apnea: a transcranial Doppler ultrasound
study. Sleep Med., 4: 219223.
Boecher-Schwarz, HG, Ungersboeck, K, Ulrich, P,
Mueller-Forell, W, Smolders, D and Perneczky, A
(1995) Pre- and intraoperative methods of controlling
cerebral circulation in giant aneurysm surgery. Neurosurg. Rev., 18: 8593.
Bonnett, MP, Larousse, E, Asehnoune, K and Benhamou,
D (2004) Spinal anesthesia with bupivacaine decreases
cerebral blood flow in former preterm infants. Anesth.
Analg., 98: 12801283.
Borger, MA, Taylor, RL, Weisel, RD, Kulkarni, G, Benaroia, M, Rao, V, Cohen, G, Fedorko, L and Feindel, CM
OTHER ISSUES
(1999) Decreased cerebral emboli during distal aortic
arch cannulation: a randomized clinical trial. J. Thorac.
Boylan, GB, Panerai, RB, Rennie, JM, Evans, DH, RabeHesketh, S and Binnie, CD (1999) Cerebral blood
flow-velocity during neonatal seizures. Arch. Dis.
Brooker, RF, Brown, WR, Moody, DM, Hammon, JW, Jr.,
Reboussin, DM, Deal, DD, Ghazi-Birry, HS and Stump,
DA (1998) Cardiotomy suction: a major source of brain
lipid emboli during cardiopulmonary bypass. Ann.
Thorac. Surg., 65: 16511655.
Brown, JW, Dierdorf, SF, Moorthy, SS and Halpin, M
(1987) Venoarterial cerebral perfusion for treatment of
massive arterial air embolism. Anesth. Analg., 66:
673674.
Brown, MM, Wade, JPH, Bishop, CCR and Russell, RW
(1986) Reactivity of the cerebral circulation in patients
with carotid occlusion. J. Neurol. Neurosurg. Psychiatry, 49: 899904.
Buijs, J, Van Bel, F, Nandorff, A, Hardjowijono, R,
Stijnen, T and Ottenkamp, J (1992) Cerebral blood flow
pattern and autoregulation during open-heart surgery in
infants and young children: a transcranial Doppler ultrasound study. Crit. Care Med., 20: 771777.
Calafiore, AM, Bar-El, Y, Vitolla, G, Gianmarco, G, Teodori,
G, Iaco, AL, DAlessandro, S and Di Mauro, M (2001)
Early clinical experience with a new sutureles anastomotic device for proximal anastomoses of the saphenous
vein to the aorta. J. Thorac. Cardiovasc. Surg., 121:
854858.
Casati, A, Fanelli, G, Pietropaoli, P, Proietti, R, Tufano, R,
Danelli, G, Fierro, G, De Cosmo, G and Servillo, G
(2005) Continuous monitoring of cerebral oxygen saturation in elderly patients undergoing major abdominal
surgery minimizes brain exposure to potential hypoxia.
Anesth. Analg., 101: 740747.
Clark, JM, Skolnick, BE, Gelfand, R, Farber, RE,
Stierheim, M, Stevens, WC, Beck, G, Jr. and
Lambertsen, CJ (1996) Relationship of 133Xe cerebral
blood flow to middle cerebral arterial flow-velocity in
men at rest. J. Cereb. Blood Flow Metab., 16:
12551262.
Crombleholme, TM, Adzick, NS, DeLorimier, AA,
Longaker, MT, Harrison, MR and Charlton, VE (1990)
Carotid artery reconstruction following extracorporeal
membrane oxygenation. Am. J. Dis. Child., 144:
872874.
Deverson, S, Evans, DH and Bouch, DC (2000) The effects
of temporal bone on transcranial Doppler ultrasound
beam shape. Ultrasound Med. Biol., 26: 239244.
Dinkel, M, Langer, H, Loerler, H, Rugheimer, E and
Schweiger, H (1994) Neuromonitoring in carotid surgery: possibilities and limits of transcranial Doppler
ultrasound. Vasa, 23: 337344[German].
921
Doblar, DD (1996) Cerebrovascular assessment of the
high-risk patient: the role of transcranial Doppler ultrasound. J. Cardiothorac. Vasc. Anesth., 10: 314.
Doblar, DD (2004) Intraoperative transcranial ultrasonic
monitoring for cardiac and vascular surgery. Sem. Cardiothorac. Vasc. Anesth., 8: 127145.
Droste, DW, Kriete, JU, Stypmann, J, Castrucci, M,
Wiochte, T, Tietje, R, Weltermann, B, Young, P and
Ringelstein, EB (1999) Contrast transcranial Doppler
ultrasound in the detection of right-to-left shunts: comparison of different procedures and different contrast
agents. Stroke, 30: 18271832.
Droste, DW, Lerner, T, Dittrich, R, Ritter, M and
Ringelstein, EB (2005) Comparison of a 1-MHz and a
2-MHz probe for microembolus detection using transcranial Doppler ultrasound. Neurol. Res., 27: 471476.
monitoring for cardiac surgery. Heart Surg. Forum, 5:
225228.
Edmonds, HL, Jr., Singer, I, Sehic, A and Strickland, TJ
(1998) Multimodality neuromonitoring for neurocardiology. J. Interven. Cardiol., 11(3): 197204.
Eifert, S, Reichenspurner, H, Pfefferkorn, T, Baur, B, Von
Schlippenback, C, Mayter, TE, Hamann, G and
Reichart, B (2003) Neurological and neuropyschological examination and outcome after use of an intra-aortic
filter device during cardiac surgery. Perfusion, 18:
5560.
Erlich, MP, McCullough, J, Wolfe, D, Zhang, N, Shiang,
H, Weisz, D, Bodian, C and Griepp, RB (2001) Cerebral effects of cold reperfusion after hypothermic circulatory arrest. J. Thorac. Cardiovasc. Surg., 121:
923931.
Estrera, AL, Garami, Z, Miller, CC, III, Sheinbaum, R,
Huynh, TT, Porat, EE, Winnerkvist, A and Safi, HJ
(2003) Determination of cerebral blood flow dynamics
during retrograde cerebral perfusion using power
M-mode transcranial Doppler. Ann. Thorac. Surg., 76:
704709.
Ferrari, J, Baumgartner, H, Tentschert, S, Dorda, V, Lang,
W, Willfort-Ehringer, A, Probst, P and Lalouschek, W
(2004) Cerebral microembolism during transcatheter
closure of patient foramen ovale. J. Neurol., 251:
825829.
Fraser, CD, Jr. and Andropoulos, DB (2004) Neurologic
monitoring for special cardiopulmonary bypass techniques. Semin. Thorac. Cardiovasc. Surg. Pediatr.
Card. Surg. Annu., 7: 125132.
Ganzel, BL, Edmonds, HL, Jr., Pank, JR and Goldsmith, LJ
(1997) Neurophysiological monitoring to assure delivery of retrograde cerebral perfusion. J. Thorac. Cardiovasc. Surg., 113: 748757.
Georgiadis, D, Kaps, M, Berg, J, Mackay, TG, Dapper, F,
Faicheney, A, Wheatley, DJ and Lees, KR (1996)
Transcranial Doppler detection of microemboli in
922
prosthetic heart valve patients: dependency upon valve
type. Eur. J. Cardiothorac. Surg., 10: 253257.
Giller, CA, Bowman, G, Dyer, H, Mootz, L and Krippner, W
(1993) Cerebral arterial diameters during changes in
blood pressure and carbon dioxide during craniotomy.
Giller, CA, Mathews, D, Walker, B, Purdy, P and
Roseland, AM (1994) Prediction of tolerance to carotid
artery occlusion using transcranial Doppler ultrasound.
Giller, CA, Purdy, P, Giller, A, Batjer, HH and Kopitnik, T
(1995) Elevated transcranial Doppler ultrasound velocities following therapeutic arterial dilation. Stroke, 26:
123127.
Giller, CA, Giller, AM and Landreneau, F (1998) Detection
of emboli after surgery for intracerebral aneurysms.
Goto, T, Baba, T, Honma, K, Shibata, Y, Arai, Y, Uozumi,
H and Okuda, T (2001) Magnetic resonance imaging
findings and postoperative neurologic dysfunction in
elderly patients undergoing coronary artery bypass
grafting. Ann. Thorac. Surg., 72: 137142.
Greeley, WJ, Ungerleider, RM, Kern, FH, Brusino, FG, Smith,
LR and Reves, JG (1989) Effects of cardiopulmonary
bypass on cerebral blood flow in neonates, infants and children. Circulation, 80(Suppl. I): 209215.
Grocott, HP, Smith, MS, Glower, DD and Clements, FM
(1998) Endovascular aortic balloon clamp malposition
during minimally invasive cardiac surgery: detection
by transcranial Doppler monitoring. Anesthesiology,
88: 13961399.
Guerra, WK, Gaab, MR, Dietz, H, Mueller, JU, Piek, J and
Fritsch, MJ (1999) Surgical decompression for traumatic brain swelling: indications and results. J. Neurosurg., 90: 187196.
Halpern, P, Neufeld, MY, Sade, K, Silbiger, A, Szold, O,
Bornstein, NM and Sorkine, P (2003) Middle cerebral
artery flow-velocity decreases and electroencephalogram
(EEG) changes occur as acute hypercapnia reverses.
Intensive Care Med., 29: 16501655.
Halsey, JH, McDowell, HA, Gelmon, S and Morawetz, RB
(1989) Blood velocity in the middle cerebral artery and
regional cerebral blood flow during carotid endarterectomy. Stroke, 20: 5358.
Iacopino, DG, Conti, A, Battaglia, C, Siliotti, C, Lucanto, T,
Santamaria, LB and Tomasello, F (2003) Transcranial
Doppler ultrasound study of the effects of nitrous
oxide on cerebral autoregulation during neurosurgical
anesthesia: a randomized controlled trial. J. Neurosurg.,
99: 5864.
Jansen, C, Vriens, EM, Eikelboom, BC, Vermeulen, FE, Van
Gijn, J and Ackerstaff, RG (1993) Carotid endarterectomy
with transcranial Doppler and electroencephalographic
monitoring. A prospective study in 130 operations. Stroke,
24: 665669.
H.L. EDMONDS
Johnston, AJ, Steiner, LA, Balestreri, M, Gupta, AK and
Menon, DK (2003) Hyperoxia and the cerebral hemodynamic responses to moderate hyperventilation. Acta
Anaesthesiol. Scand., 47: 391396.
Jrgensen, LG (1995) Transcranial Doppler ultrasound for
cerebral perfusion. Acta Physiol. Scand., 625(Suppl.):
144.
Kalra, M, Al-Khaffaf, H, Farrell, A, Wallbank, WA and
Charlesworth, D (1994) Comparison of measurement
of stump pressure and transcranial measurement of
flow-velocity in the middle cerebral artery in carotid
surgery. Ann. Vasc. Surg., 8: 225231.
Kaps, M, Hansen, J, Weiher, M, Tiffert, K, Kayser, I and
Droste, DW (1997) Clinically silent microemboli in
patients with artificial prosthetic aortic valves are predominantly gaseous and not solid. Stroke, 28: 322325.
Lennard, N, Smith, J, Dumville, J, Abbott, R, Evans, DH,
London, NJ, Bell, PR and Naylor, AR (1997) Prevention
of postoperative thrombotic stroke after carotid endarterectomy: the role of transcranial Doppler ultrasound. J.
Vasc. Surg., 26: 579584.
Levi, CR, OMalley, HM, Fell, G, Roberts, AK, Hoare, MC,
Royle, JP, Chan, A, Beiles, BC, Chambers, BR, Bladin,
CF and Donnan, GA (1998) Transcranial Doppler
detected cerebral microembolism following carotid endarterectomy. High microembolic signal loads predict postoperative cerebral ischaemia. Brain, 120(Pt. 4): 621629.
Lund, C, Hol, PK, Lundblad, R, Fosse, E, Sundet, K,
Tennoe, B, Brucher, R and Russell, D (2003) Comparison of cerebral embolization during off-pump and onpump coronary artery bypass surgery. Ann. Thorac.
Surg., 76: 765770.
Markus, HS and Punter, M (2005) Can transcranial Doppler
discriminate between solid and gaseous microemboli?
Assessment of a dual-frequency transducer system.
Stroke, 36: 17311734.
Matta, BF, Heath, KJ, Tipping, K and Summors, AC (1999)
Direct cerebral vasodilatory effects of sevoflurane and isoflurane. Anesthesiology, 91: 677680.
Metry, G, Spittle, M, Rahmati, S, Giller, C, Giller, A,
Kaufman, A, Schneditz, D, Manno, E, Brener, Z,
Boniece, I, Ronco, F, Ronco, C and Levin, NW (2002)
Online monitoring of cerebral hemodynamics during
hemodialysis. Am. J. Kidney Dis., 40: 9961004.
Mills, NL and Ochsner, JL (1980) Massive air embolism
during cardiopulmonary bypass. J. Thorac. Cardiovasc.
Surg., 80: 708717.
Muller, M, Behnke, S and Walter, P (1999) Cerebrovascular resistance and blood flow-velocity changes after
carotid endarterectomy. Vasa, 28: 279282.
Murkin, JM, Iglesias, I, Bainbridge, D, Adams, S, Schaefer,
B, Irwin, B and Fox, S (2004) Monitoring cerebral oxygen saturation significantly decreases major organ morbidity in CABG patients: a randomized, blinded study.
Heart Surg. Forum, 7: 515 [abstract].
OTHER ISSUES
Myles, PS, Leslie, K, McNeil, J, Forbes, A and Chan, MT
(2004) Bispectral index monitoring to prevent awareness during anaesthesia: the B-Aware randomized controlled trial. Lancet, 363: 17571763.
Naylor, AR, Hayes, PD, Allroggen, H, Lennard, N, Gaunt, ME
and Thompson, MM (2000) Reducing the risk of carotid
surgery: a 7-year audit of the role of monitoring and quality
control assessment. J. Vasc. Surg., 32: 750759.
Neri, E, Sassi, C, Barabersi, L, Massetti, M, Pula, G, Buklas,
D, Tassi, R and Giomarelli, P (2004) Cerebral autoregulation after hypothermic circulatory arrest in operations on
the aortic arch. Ann. Thorac. Surg., 77: 7279.
Newell, DW, Eskridge, JM, Mayberg, MR, Grady, MS and
Winn, HR (1989) Angioplasty for the treatment of symptomatic vasospasm following subarachnoid hemorrhage.
J. Neurosurg., 71: 654660.
Okamoto, H, Sato, K, Matsuura, A, Ogawa, Y, Asakura, T,
Hoshino, M, Seki, A, Abe, T and Yasuura, K (1993)
Selective jugular cannulation for safer retrograde cerebral perfusion. Ann. Thorac. Surg., 55: 538540.
Padayachee, TS, Parsons, S, Theobold, R, Linley, J,
Gosling, RG and Deverall, PB (1987) The detection of
microemboli in the middle cerebral artery during cardiopulmonary bypass: a transcranial Doppler ultrasound
investigation using membrane and bubble oxygenators.
Ann. Thorac. Surg., 44: 298302.
Padayachee, TS, Parsons, S, Theobold, R, Gosling, RC and
Deverall, RB (1988) The effect of arterial filtration on
reduction of gaseous microemboli in the middle cerebral artery during cardiopulmonary bypass. Ann.
Thorac. Surg., 45: 647649.
Potthast, K, Erdonmez, G, Schnelke, C, Sellin, L, Sliwka, U,
Schondube, F, Eichler, M and Ruel, H (2000) Origin and
appearance of HITS induced by prosthetic heart valves:
an in vitro study. Int. J. Artif. Organs, 23: 441445.
Powers, AD and Smith, RR (1990) Hyperperfusion syndrome after carotid endarterectomy: a transcranial
Doppler evaluation. Neurosurgery, 26: 5660.
Riding, G, Daly, K, Hutchinson, S, Rao, S, Lovell, M and
McCollum, C (2004) Paradoxical cerebral embolization.
An explanation for fat embolism syndrome. J. Bone
Joint Surg. Br., 86: 9598.
Ringelstein, EB, Droste, DW, Babikian, VL, Evans, DH,
Grosset, DG, Kaps, M, Markus, HS, Russell, D and
Siebler, M (1998) Consensus on microembolus detection by TCD. International Consensus Group on Microembolus Detection. Stroke, 29: 725729.
Rodriguez, RA, Cornel, G, Semelhago, L, Splinter, WM
and Weerasena, NA (1997) Cerebral effects in superior
vena caval cannula obstruction: the role of brain monitoring. Ann. Thorac. Surg., 64: 18201824.
Rodriguez, RA, Williams, KA, Babaev, A, Rubens, F and
Nathan, HJ (2005) Effect of perfusionists technique on
cerebral embolization during cardiopulmonary bypass.
Perfusion, 20: 310.
923
Rowed, DW, Houlden, DA, Burkholder, LM and Taylor,
AB (2004) Comparison of monitoring techniques for
intraoperative cerebral ischemia. Can. J. Neurol. Sci.,
31: 347356.
Schneweis, S, Urbach, H, Solymosi, L and Ries, F (1997)
Preoperative risk assessment for carotid occlusion by
transcranial Doppler ultrasound. J. Neurol. Neurosurg.
Psychiatr., 62: 485489.
Shen, Q, Stuart, J, Venkatesh, B, Wallace, J and Lipman, J
(1999) Inter-observer variability of the transcranial
Doppler ultrasound technique: impact of lack of practice on the accuracy of measurement. J. Clin. Monit.
Comput., 15: 179184.
Smith, JL, Evans, DH, Fan, L, Bell, PR and Naylor, AR
(1996) Differentiation between emboli and artefacts
using dual-gated transcranial Doppler ultrasound. Ultrasound Med. Biol., 22: 10311036.
Sohn, YH, Kim, GW and Kim, JS (1997) Do hematocrit
and serum fibrinogen influence transcranial Doppler
measurements? J. Korean Med. Sci., 12: 405408.
Spencer, MP (1997) Transcranial Doppler monitoring and
causes of stroke from carotid endarterectomy. Stroke,
28: 685691.
Spencer, MP, Thomas, GI, Nicholls, SC and Sauvage, LR
(1990) Detection of middle cerebral artery emboli during
carotid endarterectomy using transcranial Doppler ultrasonography. Stroke, 21: 415423.
Spencer, MP, Thomas, GI and Moehring, MA (1992) Relation between middle cerebral artery blood flow-velocity
and stump pressure during carotid endarterectomy.
Stroke, 23: 14391445.
Topcuoglu, MA, Palacios, IF and Buonanno, FS (2003)
Contrast M-mode power Doppler ultrasound in the
detection of right-to-left shunts: utility of submandibular internal carotid artery recording. J. Neuroimaging,
13: 315323.
Van Lieshout, JJ, Wieling, W, Karemaker, JM and Secher,
NH (2003) Syncope, cerebral perfusion and oxygenation. J. Appl. Physiol., 94: 833848.
Yamasaki, T, Moritake, K, Akiyama, Y and Oi, S (2000)
Microvascular Doppler ultrasound-assisted neuroendoscopic surgery. Br. J. Neurosurg., 14: 464467.
Yeh, TJ, Jr., Austin, EH, III, Sehic, A and Edmonds, HL,
Jr. (2003) Role of neuromonitoring in the detection
and correction of cerebral air embolism. J. Thorac. Cardiovasc. Surg., 126(2): 589591.
Yoon, BW, Bae, HJ, Kang, DW, Lee, SH, Hong, KS,
Kim, KB, Park, BJ and Roh, JK (2001) Intracranial
cerebral artery disease as a risk factor for central nervous system complications of coronary artery bypass
graft surgery. Stroke, 32: 9499.
Zanchetta, M, Rigatelli, G and Onorato, E (2003) Intracardiac echocardiography and transcranial Doppler ultrasound to guide closure of patent foramen ovale.
J. Invasive Cardiol., 15: 9396.

M.R. Nuwer (Ed.)
924
CHAPTER 68
Cerebral oximetry as a tool in the operating room

and intensive care unit
Jeffrey R. Balzera,b,*, Donald Crammonda, Miguel Habeycha,1
and Robert J. Sclabassia,b,c,d
a

b

c
Department of Psychiatry, University of Pittsburgh, Pittsburgh, PA 15213, USA
Electrical, Mechanical and Biomedical Engineering, University of Pittsburgh, Pittsburgh, PA 15213, USA
68.1. Introduction
Protection of the neural axis during a variety of surgical
procedures has traditionally been achieved via the
application of electrophysiological measures including
electroencephalography (EEG), somatosensory evoked
potentials (SEPs), brainstem auditory evoked potentials
(BAEPs), and motor evoked potentials (MEPs). These
electrophysiological techniques are employed as a
means by which to detect ischemic insult that can result
in iatrogenic injury and prevent subsequent poor neurologic outcomes in these patients.
As an adjunct to these electrophysiological techniques, a noninvasive cerebral oxygenation measurement
technique has been developed (Jobsis, 1977; McCormick et al., 1991) which has proven to be effective in
detecting early cerebral ischemia while there is still time
for the prevention of permanent neurological injury.
Transcranial cerebral oximetry, based on principles of
near-infrared spectroscopy (NIRS), has been successfully applied in the operating room and intensive care
unit (ICU). To date, cerebral oximetry has been widely
employed and has provided valuable information
concerning brain oxygenation in patients with cerebral
ischemia and impending neurological damage. Cerebral
oximetry has been used as an adjunctive measure during
*
Correspondence to: Jeffrey R. Balzer, Ph.D., Department

of Neurological Surgery, University of Pittsburgh Medical
Center, Suite B-400, 200 Lothrop Street, Pittsburgh, PA
15213, USA.
Tel.: 1-412-648-2570; fax: 1-412-383-8999.
E-mail: balzerjr@upmc.edu (J.R. Balzer).
1
Present Address: Department of Neurological Surgery,
adult and pediatric cardiothoracic procedures, carotid

endarterectomy (CEA), neuroendovascular procedures,
and in the ICU in critically ill patients in danger of intracranial hypertension and vasospasm.
This chapter will serve to discuss the technique of
NIRS and cerebral oximetry as well as document its
usefulness during a variety of surgical procedures and
in the ICU. It also will discuss throughout any interpretational and technical limitations of the technique.
68.2. NIRS technique
NIRS has long been known to be able to pass through
tissue with ease and was used in the early 1900s for
various tissue measurements. It was not until Jobsis,
in 1977, who studied the oxygenation of the brain
using light of various wavelengths from ultraviolet to
infrared, that qualitative changes in cerebral oxygenation were followed and recorded using this device.
Cerebral oximetry involves transcranial recording
and depends on two basic essentials. First, nearinfrared light (6501,100 nm) passes through human
tissue, including bone up to depths of 1218 mm and
second, the light is partially absorbed by chromophores in the brain, namely oxygenated (HbO2) and
deoxygenated (Hb) hemoglobin (McCormick et al.,
1991). The difference in the absorption spectra
between HbO2 and Hb yields the oxygenated to total
hemoglobin ratio or the oxygen saturation.
The difficulty in reflectance spectrophotometry as
it is used in NIRS tissue oximetry is that the distance
between where the light enters the tissue and the point
of detection is unknown and variable, depending on
the degree of light scattering and on the amount of
absorbing material in the tissue. Near-infrared photons
OTHER ISSUES
Fig. 1. The Somanetics 4-channel INVOS system is

shown. New generation device can be used for cerebral
oximetry, somatic oximetry, or both simultaneously. Note
that bilateral rSO2 values (L and R) are displayed as single
values representing cerebral hemispheric oxygenation
(Courtesy Somanetics, Inc., Troy, MI).
are introduced into the skin over the forehead. Once

scattered inside the scalp, bone, and brain, some fraction of the injected photons survive to return and exit
the skin, termed reflectance. By measuring the quantity of photons returning to the sensor as a function of
wavelength, the spectral absorption of the underlying
tissue can be inferred and conclusions about average
oxygenation can be made. The primary chromophores
of interest are few and include variable degrees of
hemoglobin oxygenation and the redox state of cytochrome oxidase. It is important to emphasize, however, that the absorbance signal of cytochrome
oxidase is one-tenth that of the signal of HbO2 and is,
therefore, much more difficult to detected. Oximetry
devices currently in use measure oxygen saturation of
all the blood in the vascular bed being assayed, which
is typically 7080% venous and capillary with the
remainder being arterial (Watzman et al., 2000).
Oximetry values are typically presented as a saturation
index (rSO2) (Fig. 1) for each hemisphere that is being
assayed. This venous-weighted average reflects the
balance between cerebral oxygen supply and demand
(Nemoto et al., 1998).
68.3. Application of NIRS sensors on the head
for cerebral oximetry
Disposable stick-on sensors hold multiple optodes
used for cerebral oxygenation resolution. Light-tight
seals around the sensor should be established to aid
925
in the suppression of artifact from ambient light

sources. The optodes and sensor should be isolated
from electrodes being used for electrophysiologic
recordings to prevent artifactual contamination or
degradation of signal resolution and great care should
be taken to avoid excessive or prolonged compression of the underlying skin especially in the intensive
care setting.
A near-infrared light-emitting diode transmits the
signal which is subsequently detected by a pair of
sensors positioned 3 and 4 cm from the light-emitting
source (Fig. 2). Measurements of tissue oxygenation
using currents technologies are considered regional
in nature. Therefore, placement of sensors on the
head or body is very important (Kishi et al., 2003).
In cerebral oximetry, the depiction of sensors shows
that they are placed such that the inferior edge of
the sensor is just above the eyebrows over the frontal
lobes bilaterally (Fig. 3). However, in this placement,
it is very likely that either the detectors or sensory
may be placed over the frontal sinus. The frontal
sinus can vary considerably in size and overall area.
It can range from a very small sinus situated exclusively between the orbits to a larger sinus extending
up to the middle of the forehead and laterally down
to the eyes or, in some cases, cover the entire forehead. Although such large sinuses are rare, it is best
to appreciate how variable sinus size can be if faced
with odd readings after placement of sensors (Sehic
and Thomas, 2000). A more lateral placement of
Light Source
Shallow
Detector
Deep
Detector
Somanetics Corporation
Fig. 2. Oximetry sensors are composed of two nearinfrared detectors embedded into the adhesive patch which
is affixed to any area of scalp devoid of hair. Sensor 1 is
the shallow sensor responsible for the measurement of
mostly extracerebral signals (skull, scalp) and sensor 2 samples deeper cortical tissues. The difference between these
two values accounts for the actual rSO2 values (Courtesy
Somanetics, Inc., Troy, MI).
926
J.R. BALZER ET AL.
detector and its signal is subtracted from that of the

far detector to suppress the effects of extracranial tissue (Samra et al., 1999). This effectively suppresses
the effects of extracerebral blood but does not remove
it completely. While the potential exists to reduce the
extracerebral contribution to zero, there are minor differences in the degree of bone and skin contamination
seen between individuals likely to result in variations
of approximately 0%. While the extracerebral contribution is not zero, approximately 85% of the signal,
on average, is from intracranial cerebral tissue.
68. 5. Rationale for application of
intraoperative NIRS
Fig. 3. Depiction of sensors placed above the upper orbital

rim on the scalp and their relationship to frontal lobe and
cerebral vasculature (Courtesy Somanetics, Inc., Troy, MI).
the oximetry sensors also will allow for the monitoring

to extend to part of the middle cerebral artery (MCA)
territory which is important for monitoring for cerebral
ischemia. If the patient has a high or receding hairline
the sensors can be placed at an angle higher up on the
frontal pole but with care to be lateral so as to cover
the MCA territory. If the patient is bald, sensors can
be placed parallel to the sagittal sinus in line with the
sensor aligned on a parasagittal line on the outer edges
of the eyes to cover the MCA territory or along a line
centered on the eyes for coverage of the MCA/ACA
watershed zone.
Blood between the sensor and the brain can also be
a source of erroneous readings. Therefore, ensure that
sensors are not placed over hematomas or other
sources of pooled blood such as subarachnoid blood
over the convexities. Placement over previously
infracted brain tissue, arterial-venous malformations,
or other anomalies is also not recommended.
68.4. Suppression of extracerebral signal
Current cerebral oximeters can suppress contamination of the signal by the skin and bone overlying the
brain. The subtraction of the NIRS signal representing
skin and bone is accomplished by spatially resolving
the signal with a near (3 cm) detector (photodiode)
and a far (4 cm) detector (Cui et al., 1991). The near
detector measures light that has not penetrated as
deeply into the brain as the light measured by the far
Information concerning changes in the balance of

cerebral oxygen supply which is not provided by
other techniques such as evoked potentials or blood
flow data is provided by NIRS. Evoked potential variations can indicate alterations in function in specific
regions or pathways in the brain but does not necessarily provide specific information concerning the
underlying cause for the change such as ischemia.
Likewise, transcranial Doppler can provide information concerning changes in blood flow velocity and
inferences concerning blood flow but does not yield
information about the status of the target areas which
might be at risk. For these reasons, the adjunctive use
of NIRS with these other modalities can aid the neuromonitoring team in determining the development
of an oxygen imbalance, the nature of the imbalance,
the corrective action which may be appropriate, and
the result of said corrective intervention.
An important advance with regards to utilizing
and interpreting NIRS changes in the operating room
was the establishment of normative rSO2 values in
normal healthy adults. Kim et al. (2000) showed a
normal rSO2 value of 71 6% in healthy subjects,
while a lower value with greater variability has been
described for adult patients with preexisting cardiovascular disease (Edmonds et al., 2004). Other abnormal values also have been defined. For example,
intraoperative rSO2 values <50% (Cho et al., 1998)
or <40% (Yao et al., 2004) in adults had been shown
to be associated with cerebral ischemia and iatrogenic cerebral injury evidenced by postoperative
decline in neurocognitive function. Additionally,
low rSO2 values (<45%) lasting for protracted periods of time (>180 min) have been associated with
brain injury in infants as evidenced by magnetic resonance imaging techniques.
OTHER ISSUES
It should be pointed out that these normative

values represent a hemoglobin saturation measure
which is highly regional and that a single measurement versus interrogation of values and changes over
time may not reflect an accurate depiction of cerebral
health? This notion is supported by the fact that
normal values have been obtained from cadavers
(Schwarz et al., 1996), patients with brain death
(Germon et al., 1994), and from patients with cerebral infarction (Nemoto et al., 2000).
68.6. Cardiovascular surgery
68.6.1. Adult
Routine cardiothoracic surgical procedures have the
potential to result in serious neurologic complications
and long-term cognitive deficits translating to significant cost to the healthcare system (Roach et al., 1996;
Newman et al., 2001). As is the case for other surgical
procedures and pathologies that will be discussed in this
chapter, a number of neuromonitoring modalities have
been used during adult cardiothoracic surgery in an
attempt to reduce the frequency of neurologic iatrogenic
injury. In order to identify the appropriate monitoring
modalities to be used, it is important to attempt to
identify the major causes of brain injury during these
procedures. It has been reported that brain injury related
to cardiothoracic procedures results from three major
sources, hypo- or hyperperfusion, embolic events, and/
or systemic inflammatory responses (Taylor, 1998).
Two of these problems, perfusion abnormalities
and embolic events, have been successfully detected
and corrected to some degree using a multimodality
monitoring approach using simultaneous EEG, transcranial Doppler, and cerebral oximetry. I will focus
on the application of cerebral oximetry during various
cardiothoracic procedures, its ability to accurately
detect changes in cerebral oxygenation, and its success
in reducing iatrogenic injury and ultimately improving
outcomes in this patient population.
Cerebral oximetry has been used, in addition to
other neuromonitoring modalities, quite extensively
during adult cardiovascular procedures (Edmonds,
2001; Andropoulos et al., 2004; Novitzky, 2005). Several authors have described the benefit of using NIRS
during these procedures in reducing potential iatrogenic cerebral injury that might occur secondary to circulatory arrest and deep hypothermia (Ogino et al.,
1998; Prabhune et al., 2002; Edmonds et al., 2004;
927
rSO2
Induction
80
Left Cortex
Right Cortex
70
60
CO2 29
50
040817mc
40
12:15
12:45
23
13:15
13:45
14:15
Fig. 4. Regional cerebral oxygen saturation (rSO2) changes

from the initiation of monitoring in the awake patient until
the onset of cardiopulmonary bypass at 14:00 h. The transient decline to the intervention threshold at 13:18 h was
precipitated by progressive hypocapnia. The desaturation
was promptly corrected by increasing arterial CO2 partial
pressure toward normocapnia. The transient desaturation
at the onset of cardiopulmonary bypass reflected the passage of pump priming solution through the cerebral vasculature. [Reprinted from Edmonds (2005) with permission
from Blackwell Publishing.]
Goldman et al., 2004; Murkin, 2004; Orihashi et al.,

2004; Yao et al., 2004; Edmonds, 2005; Dent
et al., 2006; Fig. 4), while others have suggested
that NIRS is not predictive of postoperative cognitive
performance after undergoing coronary artery bypass
grafting (Reents et al., 2002) and that NIRS is
less sensitive than jugular bulb oximetry in assaying
oxygenation insufficiencies during these procedures
(Sapire et al., 1997; Shaaban Ali et al., 2001).
In support of the use of NIRS during these procedures, Ogino et al. studied 12 consecutive patients with
retrograde cerebral perfusion in conjunction with profound hypothermic circulatory arrest undergoing
aortic arch surgery. The authors found a gradual
decrease in rSO2 in all patients during hypothermic circulator arrest even when it was combined with retrograde perfusion. On average, rSO2 values decreased
to 46 8.7%. One patient developed a permanent neurological injury and was the patient with the greatest
decrease in rSO2. A linear correlation was observed
between both the retrograde cerebral perfusion flow
rate and the ratio of post- to prehypothermic arrest
rSO2 and the rate of decrease in rSO2. A safe duration
of hypothermic circulatory arrest under different conditions of retrograde perfusion using rSO2 as a test
for adequate brain protection was developed.
In an attempt to examine the relationship between sustained decreases in rSO2 recorded from the frontal lobes
to the occurrence of neurological injury, Orihashi et al.
928
studied 59 consecutive patients undergoing cardiac surgery with selective cerebral perfusion. The authors found
that the durations for which rSO2 decreased were significantly longer for the patients with neurological events
than for those patients who did not succumb to these
worsened outcomes.
In a study with a larger patient population, Yao
et al. studied 101 patients undergoing elective cardiac
surgery with cardiopulmonary bypass to evaluate the
relationship between changes in rSO2 and neuropsychological function after surgery. These authors
found that patients with rSO2 values less than 35%
had significantly higher incidences of postoperative
neuropsychological dysfunction than those patients
with rSO2 values which persistently stayed above
35%. Further, a correlation between the absolute
amount of time rSO2 values were below a certain
value (<40%) and poor neuropsychological outcomes was demonstrated.
In an even larger study, Goldman et al. used cerebral oximetry to monitor rSO2 in all cardiac surgery
patients for 18 months totaling 1,034 patients. rSO2
was optimized during the surgical procedures by
modifying oxygen delivery and consumption variables in an attempt to maintain baseline rSO2 values.
The author went on to compare the incidence of
stroke in the oximetry group to a group of patients
(n 1,245) who underwent cardiac surgery without
the use of oximetry. Results showed that despite the
fact that the oximetry group had sicker patients overall, this group had significantly fewer permanent
strokes than the nonoximetry group. Additionally,
the oximetry group had significantly fewer patients
requiring prolonged ventilation as well as a significantly shorter hospital stay compared with the
nonmonitored group.
Many of the significant neurological events that
occur during cardiac surgery can be detected using multimodality intraoperative neuromonitoring, including
cerebral oximetry. Subsequent to their detection, prevention of these events can occur via perfusion adjustments, changes in oxygenation, and administration of
anesthetic regimens. NIRS, as a component to this multimodality approach, has proven effective in not only
reducing neurological deficits after adult cardiac surgery but also has resulted in reductions in hospital stay
lengths (Yao et al., 2001; Iglesias et al., 2003) and overall cost saving to the hospital system (Edmonds et al.,
1999; Goldman et al., 2006). While these data are very
promising, a more rigorous prospective analysis, void
J.R. BALZER ET AL.
of the biases observed in the above studies (i.e., selection and control population), still needs to be completed
in order to more definitively prove the clinical utility of
NIRS during cardiac surgical procedures.
68.6.2. Pediatric
As with adult cardiac surgery, iatrogenic neurologic
injury occurs with some frequency in children undergoing surgical procedures for repair of congenital
heart defects. Mechanisms of neurologic injury are
strikingly different in the pediatric population compared to adults. The causes of neurological complications in children are multifactorial and may include
low cardiac output, hypothermic circulatory arrest,
hypoxemia, and complications associated with
bypass. It also has been suggested that children with
congenital heart defects have other developmental
pathologies, namely, low baseline cerebral blood
flow. While the sources for injury may be different
in this population, as with the adult cardiac population, multimodality noninvasive neuromonitoring
techniques can prove to be extremely valuable in
detecting and preventing ischemic injury.
Several important interpretive issues have been
encountered with the use of NIRS in the pediatric
cardiac population. One problematic issue has been
the ability to define a normative range for rSO2
values in pediatric patients. Large interpatient variability has been observed in the normal neonatal population (Menke et al., 2003), in the critically ill infant
patient population (Weiss et al., 2005), and in those
with congenital heart defects (Kurth et al., 2001; Fenton et al., 2005). Despite this issue, as with adult cardiac surgery, numerous studies support the use of
NIRS during a variety of pediatric cardiac procedures
(Kurth et al., 1995; Austin et al., 1997; Daubeney
et al., 1998; Pigula et al., 2000, 2001; Morimoto
et al., 2003; Andropoulos et al., 2003a,b, 2004a,b;
Hoffman et al., 2004; Ing et al., 2004; Lozano and
Mossad, 2004; Fenton et al., 2005; Gottlieb et al.,
2006; Scholl et al., 2006; Fig. 5).
In two small studies, Kurth et al. and Daubeney
et al. reported similar results. These authors found
that rSO2 values displayed significant changes during
various times during the cardiac procedures. For
example, significant changes from baseline were
observed during caval cannulation, with the institution of bypass, during circulatory arrest and during
OTHER ISSUES
929
100
90
80
rSO2i
70
60
50
40
30
20
Time A
Left
Right
10
0
Time (min)
Fig. 5. Oximetric changes observed during pediatric cardiac procedure. Note precipitous and abrupt decrease in
cerebral saturation (Time A) which was coincident with
cardiopulmonary bypass. Repositioning of the aortic cannula resulting in normalization of rSO2 values. [Reprinted
from Gottlieb et al. (2006) with permission from Blackwell
Publishing.]
reperfusion. In some cases, a correlation was

observed between the change in rSO2 and clinical
outcome, moreover, both studies demonstrated the
value of NIRS for monitoring oxygen balance especially during critical potions of these cases where
the probability of cerebral ischemia is highest.
Fenton et al. recorded rSO2 continuously in 143
infants undergoing repair of congenital heart defects
performed utilizing cardiopulmonary bypass. The
authors correlated pre- and postoperative rSO2 with
physiology and outcome. Mean preoperative rSO2
was 64% overall and was found to be significantly
lower in patients with left to right shunt physiology
but not in cyanotic patients without the shunts. Postoperative rSO2 values were found to be lower in both the
shunt and cyanotic infants. Perioperative death in this
study was associated with baseline preinduction rSO2
values of less than 50%. The authors argue that correlative rSO2 data obtained in this study and preoperative
optimization of cerebral oxygenation may ultimately
improve outcome in this patient population.
In a larger retrospective cohort study performed by
Austin et al., multimodality monitoring, including
NIRS, was performed in 250 patients. Cerebral ischemia was defined using oximetry as a decrease in
rSO2 of more than 20% of the baseline established just
before aortic cannulation which persisted for more
than 3 min. The study examined the potential benefit
of interventions based on the monitoring in decreasing
postoperative neurological deficits, length of hospital
stay, and overall cost savings. The authors reported

changes in brain perfusion or metabolism in 70% of
the patients monitored. In 74% of those patients, interventions which altered patient management were
made based on neurophysiologic changes. Neurological deficits were observed to occur in 7% of
patients who did not show significant neurophysiologic changes, 6% of patients with intervention, and
26% of patients without intervention. Survivors
length of stay averaged 6 days in the no change and
intervention groups and 9 days for the no intervention
group. Additionally, the authors rationalized hospital
expenditure for multimodality neurophysiological
monitoring based on the savings in complication costs
prevented by implementing intraoperative monitoring
techniques, including NIRS, during these cardiac procedures. Taken together these data, despite the potentials for statistical bias in this study, suggest that
surgical interventions based on information garnered
from intraoperative neurophysiological monitoring
decrease postoperative neurologic complications and
the length of hospital stay in these patients.
Despite these studies supporting the use of NIRS
during pediatric cardiac procedures, other authors
have shown either no benefit to the addition of NIRS
in pediatric cardiac procedures (Kussman et al.,
2005) or were not able to correlate rSO2 with venous
jugular bulb saturation (Nagdyman et al., 2005) or
found wide limits of agreement between the two
(Tortoriello et al., 2005) in the pediatric cardiac population and hence are cautious in proposing NIRS
technique application in these cases.
In general, the application of NIRS techniques during pediatric cardiac procedures appears to be a
promising tool in its ability to detect changes in the
balance between cerebral oxygen delivery and the
brains metabolic demand. This is of particular importance given that the pediatric cardiac team, once aware
of any cerebral oxygenation imbalance, has a wide
range of variables under their control which can influence or reset the imbalance thus preventing potential
cerebral ischemia. While valuable, NIRS should only
be used in conjunction with other neuromonitoring
modalities (i.e., EEG and TCD) so that a complete
and thorough intraoperative assessment of cerebral
integrity can be made.
68.7. Carotid endarterectomy
Occurrence of cerebral ischemia during CEA performed under general anesthesia is commonly
930
detected by means of SEP and EEG monitoring.

In fact, EEG monitoring has long been considered
the gold standard of ischemia measurement during
these procedures (Blume and Sharbrough, 1993).
While these measures have proven to be valuable in
reducing intraoperative ischemia and improving clinical outcomes, there remain some limitations with
regard to their use and interpretation during CEA. To
address these potential shortcomings, a multimodality
approach to intraoperative neurophysiological monitoring during CEA has been adopted (Isley et al.,
1998). One tool in this armamentarium involves the
clinical application of NIRS as an adjunctive measure
of cerebral ischemia during CEA. Cerebral oximetry
enables for direct, continuous and noninvasive assessment of cerebral tissue oxygenation in addition to
electrophysiological and blood flow velocity measurements typically employed.
Cerebral oximetry does not measure blood flow
directly but rather is a reflection of the balance
between oxygen supply and demand, that is, oxygen
availability. For this reason, cerebral oximetry can be
used as a valuable adjunct during carotid crossclamping to indicate variations in cerebral perfusion
and ultimately to predict cerebral ischemia. It is of
the utmost importance to place the sensor over the
MCA territory whether it be on the frontal pole or over
the parietal cortex in the case of a bald or balding
patient. For intraoperative monitoring, the oximeter
is optimally configured to store at a high rate of every
10 s and a trend rate of 2/min. Baseline cerebral oximetry measurements should be made in the awake
patient prior to the induction of anesthesia. The systolic blood pressure also should be noted during this
time and correlated to the oximetry readings. This will
provide a baseline for the assessment of the patients
cerebral oxygenation corresponding to an arterial pressure in the awake state for intraoperative comparison
during general anesthesia.
Numerous reports in the literature exist with regard
to using cerebral oximetry during CEA in both
the anesthetized and awake patient (Duncan et al.,
1995; Cho et al., 1998; De Letter et al., 1998; Samra
et al., 2000; Takeda et al., 2000; Cuadra et al., 2003;
Kragsterman et al., 2004; Mille et al., 2004; Rigamonti
et al., 2005). Cerebral oximetry, with varying degrees
of sensitivity, has been shown to positively correlate
with changes in EEG, TCD, stump pressures,
and median nerve SEPs, all of which have been associated with cerebral ischemia detection during
endarterectomy.
J.R. BALZER ET AL.
In a report by Cho et al. (1998), a critical threshold

for changes in regional cerebral oxygen saturation
(rSO2) based on a correlation with SEP change during
cross-clamping was reported. These authors demonstrated that an absolute change in 10 points from baseline showed a positive correlation with a decrease in
the P30 peak of the median nerve SEP. In this study,
a shunt was placed if P30 amplitudes decreased by
50% or greater. However, there was not good correlation between the magnitude of the reduction in amplitude of the P30 and the magnitude of the decrease in
rSO2 other than the fact that a decrease of 10 points
in rSO2 correlated with any change in SEP.
Other studies comparing oximetry with SEP, EEG,
stump pressures, and TCD velocity changes greater
than 60% have correlated both absolute and percent
changes in rSO2 with these other measures with varying degrees of sensitivity and specificity but clearly
represent the technique as having adjunctive value
(Plyuscheva et al., 1995; Williams et al., 1995; Kuroda
et al., 1996; Duffy et al., 1997; De Letter et al., 1998;
Takeda et al., 2000; Kragsterman et al., 2004).
Further support of the use of oximetry during CEA
was demonstrated by Cuadra et al. (2003) who demonstrated that generally anesthetized patients demonstrated a statistically significant change in rSO2 as a
result of clamping and shunting the internal carotid
artery. In 40 endarterectomies, they showed that
clamping of the internal carotid artery resulted in an
average ipsilateral rSO2 drop of 12.3%. Shunt insertion increased rSO2 values by an average of 10.3%.
Contralateral rSO2 values for the same periods of
clamping and shunting did not change significantly.
Mille et al. (2004) likewise showed that a relative
decrease in rSO2 of <20% from preclamp values was
highly predictive of adequate perfusion and did not
necessitate shunt placement. In line with these
results, similar studies in patients undergoing CEA
under regional anesthesia showed that cerebral rSO2
decreased significantly on carotid cross-clamping
and increased significantly with shunting or unclamping of the same artery (Carlin et al., 1998; Samra et al.,
2000; Fig. 6).
Other studies, attempting to make the same type of
SEP/oximetry correlation during CEA, showed no correlation between oximetry values and complete loss of
SEP signals (Beese et al., 1998). In this study, 317 generally anesthetized patients undergoing CEA had
simultaneous SEP and rSO2 recordings performed.
They reported that a statistically significant decrease
in rSO2 could be found in patients both with and
OTHER ISSUES
931
% Cerebrovascular Oxygen Saturation
90
85
80
75
70
Cross-Clamp
65
60
55
Hypoventilation
50
45
Ipsilateral
Contralateral
8:08:27
8:12:43
8:16:49
8:20:55
8:25:00
8:29:04
8:34:25
8:38:35
8:42:50
8:46:55
8:51:00
8:55:07
8:59:11
9:03:17
9:07:22
9:11:25
9:15:30
9:19:35
9:23:42
9:27:48
9:31:55
9:35:59
9:40:06
9:44:11
9:48:24
9:52:32
9:56:38
10:00:46
40
Clock Time
Fig. 6. Oximetric changes during carotid endarterectomy. Note that early in procedure a bilateral decrease was observed
with hypoventilation and later, during cross-clamp, an abrupt unilateral change was observed corresponding to the operative
hemisphere [reprinted from Samra et al. (1996) with permission from Lippincott, Williams & Wilkins].
without significant changes in cortical SEP recordings

and concluded that while SEP recordings appear to
continue to be a sensitive measure of cerebral ischemia, they were unable to identify critical values of
cerebral ischemia using rSO2 measurements. Similar
to the results found in the previous report, other
authors have reported low specificity of rSO2 monitoring compared with TCD (Grubhofer et al., 2000) and
EEG (Vets et al., 2004; Rigamonti et al., 2005). There
exists one report in the literature, albeit a case report,
of a postoperative neurological deficit occurrence
despite normal cerebral oximetry during CEA (Laffey
et al., 2000). In this report, the authors describe an
instance of unchanged rSO2 values in the presence of
a major ipsilateral postoperative embolic stroke. This
lack of sensitivity occurred despite what appeared to
be typical and correct sensor placement and baseline
rSO2 values that read 68.
Using these studies as a guide, the threshold for
shunting using the oximeter alone would be a 10
15 point absolute (or 20% relative) drop in oximetry
from baseline. However, in deciding whether to place
a shunt, it is best to use the oximetry data in concert
with EEG, SEP, and TCD data in order to make the
most complete decision concerning cerebral ischemia.
In addition to the intraoperative ischemic injury
that may occur during cross-clamping of the carotid
artery, postoperative hyperperfusion and reperfusion
syndromes have been reported (Karapanayiotides

et al., 2005) which can result in significant neurological deficits. Oximetry may allow for an assay of hyperperfusion and lack of autoregulation after unclamping
in these procedures (Ogasawara et al., 2003). Overall,
while its utilization has been promising during endarterectomy procedures, a lack of a standardization with
regard to baselines and percent change has resulted in
the use of cerebral oximetry as an adjunctive measure
during there procedures and should be used in conjunction with SEP, EEG, and/or TCD measurements during a multimodality approach to the neuromonitoring.
68.8. ICU Monitoring
Delayed cerebral ischemia after subarachnoid hemorrhage (SAH), secondary to vasospasm, and after
severe closed head injury (CHI), secondary to significant rises in intracranial pressure, accounts for a significant percentage of morbidity and mortality in these
patient populations (Kassell et al., 1985; Marmarou
et al., 1991; Weir, 1995). While noninvasive (i.e.,
TCD) and invasive (i.e., angiography, jugular venous
saturation, extraventricular drain placement) techniques for measuring and detecting these pathologies
exist, a noninvasive, real-time bedside indicator of
these conditions is extremely desirable (Armonda
et al., 1999; Bader, 2006).
932
In an attempt to utilize such a noninvasive, realtime device, several authors have reported the successful use of cerebral oximetry for the detection of
cerebral oxygen desaturation in these patient populations (Kirkpatrick et al., 1995; Berre et al., 1996;
Holzschuh et al., 1997; Ekelund et al., 1998; Kerr
et al., 1999, 2000; McGrade et al., 1999; Armonda
et al., 2001; Dunham et al., 2002; Ahmad et al.,
2004). Other authors have reported skepticism with
regard to the use of cerebral oximetry in these patient
populations particularly as it relates to its lack of correlation to jugular oxygen saturation (Lewis et al.,
1996; Minassian et al., 1999).
In a study by Ekelund et al., 14 neurosurgical ICU
patients with aneurysmal SAH underwent daily
simultaneous TCD and cerebral oximetry studies. In
an interesting exception, the oximetry measurements
were conducted transtemporally, over the same site
or window where the TCD measurements were
made. Significant correlations were found between
all absolute values that were measured. Specifically,
all patients with TCD mean flow velocities of
>120 cm/s also had rSO2 values <60%. Conversely,
patients with normal rSO2 (deemed to be 63% by
the authors) all had normal or only moderately
increased TCD velocities. The authors conclude,
albeit a small study, that the results support the use
of cerebral oximetry for detecting cerebral ischemia
after aneurysmal SAH.
Kirkpatrick et al. studied the use of cerebral oximetry in the CHI population as an adjunctive tool to
other more invasive measures such as ICP monitoring, jugular venous oximetry, and cerebral perfusion
pressure. Over a 12-month period, this group studied
14 patients who had suffered moderate to severe
closed head injuries using a multimodality approach.
One interesting finding of this study, which is consistent with our own experience at the University of
Pittsburgh with utilizing oximetry and these other
modalities in the ICU, was the level of signal reliability. The authors recorded a total of 886 h of data but
found that less than 50% of that data was suitable for
final analysis. The most common cause for data
exclusion with regard to cerebral oximetry was a failure to record at signal secondary to ambient light
reaching the sensors. This is entirely consistent with
our experience with continuous cerebral oximetry
monitoring in the neurosurgical ICU. Despite these
technical issues, a correlation between rSO2 changes
and variations in ICP, cerebral perfusion pressure,
and cerebral blood flow was observed (Fig. 7). The
J.R. BALZER ET AL.
60
ICP
0
70
CPP
0
5
HBO2
5
5
HB
5
80
SJO2
0
150
LDF
0
80
FV
0
15 mins
Fig. 7. Spontaneous ICP spike (arrow) recorded from

22-year-old closed head injury patient. Complimentary
changes can be observed in oximetric measures which have
a very high correlation with rSO2 and Doppler measures
[reprinted from Kirkpatrick et al. (1995) with permission
from the American Association of Neurological Surgeons].
authors conclude that NIRS shows promise as a noninvasive technique which can warn the ICU team of
potentially cerebral anoxic events with a high degree
of specificity.
Other applications for cerebral oximetry utilization in the ICU have been reported in the literature
including its use in the stroke patients (Nemoto
et al., 2000) as well as several reports relating its usefulness in seizure detection and identification
(Steinhoff et al., 1996; Adelson et al., 1999; Sokol
et al., 2000; Fig. 8).
OTHER ISSUES
933
75
70 rSO2
65
60
55
50
45
MC11-9/10-96
40
30
HRMNS
10:33:12
11:27:05
12:21:10
15:52:47
16:46:39
17:40:43
18:34:46
19:28:40
20:22:42
21:16:36
22:10:36
23:04:33
23:58:26
0:50:13
1:41:27
2:32:36
3:23:48
4:15:00
5:06:11
5:57:27
6:48:39
7:39:51
8:31:07
9:22:16
10:13:32
11:04:45
11:55:57
15:49:36
35
Fig. 8. Continuous oximetry measurements from a 4-month-old male with a GCS of 6 after a severe closed head injury. Spiking (high-frequency changes in rSO2) noted in oximetry tracing was coincident with seizure activity which was verified with
simultaneous continuous EEG recordings. [Reprinted from Adelson et al. (1999) with permission from Blackwell Publishing.]
68.9. Conclusions
The measurement of rSO2 using cerebral oximetric
techniques provides for an exclusive noninvasive perspective on cerebral hemodynamics and cerebral
oximetry is a legitimate monitoring modality for the
measurement of changes and trends in cerebral cortical
oxygen saturation. While met with some skepticism,
cerebral oximetry has been shown to have benefit in
preventing and predicting cerebral injury than can
occur during a variety of surgical procedures including
adult and pediatric cardiac procedures and CEA.
This being said, cerebral oximetry does have limitations as an intraoperative and ICU monitoring
modality that need to be recognized for the successful implementation of this unique measure. First,
because actual noninvasive measures of brain oxygen
saturation do not exist, there is no direct validation of
the accuracy of rSO2 via NIRS techniques. Second,
baseline rSO2 appears to be influenced by the placement of the sensors on the patients scalp. Placement
is currently limited to glabrous skin. Third, due to
the constraints of sensor placement, rSO2 values represent a focal measurement of oxygen saturation in
the prefrontal cortex. Regional hypoperfusion may
occur in other areas of the cerebral cortices as well
as in infratentorial regions which may not be detected
using the current technology.
It also should be noted that the information garnered via cerebral oximetry is not only different from
but also may not always coincide directly with other

physiological measures such as EEG and SEPs. For this
reason, and because of the above-stated limitations, it is
advisable that cerebral oximetry be used as an adjunctive monitoring modality and should not be used as a
sole measure of cerebral integrity during intraoperative
surgical procedures and ICU monitoring.
References
Adelson, PD, Nemoto, E, Scheuer, M, Painter, M, Morgan,
J and Yonas, H (1999) Noninvasive continuous monitoring of cerebral oxygenation periictally using nearinfrared spectroscopy. Epilepsia, 40: 14841489.
Ahmad, S, Grindlinger, GA and Desjardins, S (2004) Noninvasive cerebral oximetry in patients with traumatic brain
injury (TBI). Crit. Care Med., 32(Suppl. A104): 375.
Andropoulos, DB, Stayer, SA, McKenzie, ED and Fraser,
CD, Jr. (2003a) Novel cerebral physiologic monitoring
to guide low-flow cerebral perfusion during neonatal
aortic arch reconstruction. J. Thorac. Cardiovasc. Surg.,
125: 491499.
Andropoulos, DB, Stayer, SA, McKenzie, ED and Fraser,
CD, Jr. (2003b) Regional low-flow perfusion provides
comparable blood flow and oxygenation to both cerebral hemispheres during neonatal aortic arch reconstruction. J. Thorac. Cardiovasc. Surg., 126: 1712
1717.
Andropoulos, DB, Diaz, LK, Fraser, CD, McKenzie, ED and
Stayer, SA (2004a) Is bilateral monitoring of cerebral
oxygen saturation necessary during neonatal aortic arch
reconstruction? Anesth. Analg., 98: 12671272.
934
Andropoulos, DB, Stayer, SA, Diaz, LK and Ramamoorthy, C (2004b) Neurological monitoring for congenital heart surgery. Anesth. Analg., 99: 13651375.
Armonda, RA, McGee, B, Veznadaraglu, E and Rosenwasser, RH (1999) Near-infrared spectroscopy (NIRS) measurements of cerebral oximetry in the neurovascular
ICU. Crit. Care Med., 27: 173.
Armonda, RA, Benitiz, R, Forget, T, Thomas, JE and
Rosenwasser, RH (2001) Combined cerebral oximetry
and transcranial Doppler for vasospasm detection. Presented at the AANS/CNS Section on Cerebrovascular
Surgery, February 912, Waikoloa, Hawaii.
Austin, EH, Edmonds, HL, Auden, SM, Seremet, V, Niznic, G, Sehic, A, Sowell, MK, Cheppo, CD and Corlett,
KM (1997) Benefit of neurophysiologic monitoring for
pediatric cardiac surgery. J. Thorac. Cardiovasc. Surg.,
114: 707717.
Bader, MK (2006) Recognizing and treating ischemic
insults in the brain: the role of brain tissue oxygen monitoring. Crit. Care Nurs. Clin. North Am., 18: 243256.
Beese, U, Langer, H, Lang, W and Dinkel, M (1998) Comparison of near-infrared spectroscopy and somatosensory evoked potentials for the detection of cerebral
ischemia during carotid endarterectomy. Stroke, 29:
20322037.
Berre, J, Melot, C, Moraine, J and Kahn, RJ (1996) Evaluation of near-infrared spectroscopy during changes in
PaCO2 in comatose patients with brain damage. Crit.
Care Med., 24: A68.
Blume, WT and Sharbrough, FW (1993) EEG monitoring
during carotid endarterectomy and open heart surgery.
In: E Niedermeyer and F Lopes da Silva (Eds.),
Electroencephalography: Basic Principles, Clinical
Applications, and Related Fields. Williams and Wilkins, Baltimore, 3rd ed., pp. 747756.
Carlin, RE, McGraw, DJ, Calimlim, JR and Mascia, MF
(1998) The use of near-infrared cerebral oximetry in
awake carotid endarterectomy. J. Clin. Anesth., 10:
109113.
Cho, H, Nemoto, E, Yonas, H, Balzer, J and Sclabassi, RJ
(1998) Cerebral monitoring by means of oximetry and
somatosensory evoked potentials during carotid endarterectomy. J. Neurosurg., 89: 533538.
Cuadra, SA, Zwerling, JS, Feuerman, M, Gasparis, AP and
Hines, GL (2003) Cerebral oximetry monitoring during
carotid endarterectomy: effect of carotid clamping and
shunting. Vasc. Endovasc. Surg., 37: 407413.
Cui, W, Kumar, C and Chance, B (1991) Experimental
study of migration depth for the photons measured at
sample surface. SPIE, 143: 180191.
De Letter, JA, Sie, TH, Moll, FL, Algra, A, Eikelboom, BC
and Ackerstaff, GA (1998) Transcranial cerebral oximetry during carotid endarterectomy: agreement between
frontal and lateral probe measurements as compared with
J.R. BALZER ET AL.

an electroencephalogram. Cardiovasc. Surg., 6:
373377.
Daubeney, PEF, Smith, DC, Pilkington, SN, Lamb, RK,
Monro, JL, Tsang, VT, Livesey, SA and Webber, SA
(1998) Cerebral oxygenation during paediatric cardiac surgery: identification of vulnerable periods using near infrared spectroscopy. Eur. J. Cardiothorac. Surg., 13:
370377.
Dent, CL, Spaeth, JP, Jones, BV, Schwartz, SM, Glauser,
TA, Hallinan, B, Pearl, JM, Khoury, PR and Kurth, CD
(2006) Brain magnetic resonance imaging abnormalities
after the Norwood procedure using region cerebral perfusion. J. Thorac. Cardiovasc. Surg., 131: 190197.
Duffy, CM, Manninen, PH, Chan, A and Kearns, CF (1997)
Comparison of cerebral oximeter and evoked potential
monitoring in carotid endarterectomy. Can. J. Anaesth.,
44: 10771081.
Duncan, LA, Ruckley, CV and Wildsmith, JAW (1995)
Cerebral oximetry: a useful monitor during carotid artery
surgery. Anaesthesia, 50: 10411045.
Dunham, CM, Sosnowski, C, Porter, JM, Siegal, J and
Kohli, C (2002) Correlation of noninvasive cerebral
oximetry with cerebral perfusion in the severe head
injured patient: a pilot study. J. Trauma, 52: 4046.
Edmonds, HL (2001) Advances in neuromonitoring for cardiothoracic and vascular surgery. J. Cardiothorac. Vasc.
Anesth., 15: 241250.
Edmonds, HL (2005) Protective effect of neuromonitoring
during cardiac surgery. Ann. NY Acad. Sci., 1053:
1219.
Edmonds, HL, Thomas, MH, Yu, Q and Pollok, SB (1999)
Cost of coronary artery bypass graft (CABG) surgery
reduced my multimodality monitoring. Anesth. Analg.,
88: SCA26.
Edmonds, HL, Ganzel, BL and Austin, EH (2004) Cerebral
oximetry for cardiac and vascular surgery. Semin. Cardiothorac.Vasc. Anesth., 8: 147166.
Ekelund, A, Kongstad, P, Saveland, H, Romner, B, Reinstrup, P, Kristiansson, K-A and Brandt, L (1998) Transcranial cerebral oximetry related to transcranial
Doppler after aneurysmal subarachnoid hemorrhage.
Acta Neurochir. (Wien.), 140: 10291036.
Fenton, KN, Freeman, K, Glogowski, K, Fogg, S and Duncan, KF (2005) The significance of baseline cerebral
oxygen saturation in children undergoing congenital
heart surgery. Am. J. Surg., 190: 260263.
Germon, TJ, Manara, AR, Kane, NM and Nielson, RJ
(1994) Possible limitations of cerebral oximetric monitoring in adults. Anaesthesia, 49: 639640.
Goldman, S, Sutter, F, Ferdinand, F and Trace, C (2004)
Optimizing intraoperative cerebral oxygen delivery using
noninvasive cerebral oximetry decreases the incidence of
stroke for cardiac surgical patients. Heart Surg. Forum, 7
(5): E376E381.
OTHER ISSUES
Goldman, SM, Sutter, EP, Wertan, MA, Ferdinand, FD, Trace,
CL and Samuels, LE (2006) Outcome improvement and
cost reduction in an increasingly morbid cardiac surgery
population. Semin. Cardiothorac. Vasc. Anesth., 10:
171175.
Gottlieb, EA, Fraser, CD, Andropoulos, DB and Diaz, LK
(2006) Bilateral monitoring of cerebral oxygen saturation results in recognition of aortic cannula malposition
during pediatric congenital heart surgery. Paediatr.
Anaesth., 16: 787789.
Grubhofer, G, Plochl, W, Skolka, M, Czerny, M, Ehrlich,
M and Lassnigg, A (2000) Comparing Doppler ultrasonography and cerebral oximetry as indicators for shunting in carotid endarterectomy. Anesth. Analg., 91:
13391344.
Hoffman, GM, Stuth, EA, Jaquiss, RD, Vanderwal, PL,
Staudt, SR, Troshynski, TJ, Ghanayem, NS and Tweddell,
JS (2004) Changes in cerebral and somatic oxygenation
during stage 1 palliation of hypoplastic left heart syndrome
using continuous regional cerebral perfusion. J. Thorac.
Holzschuh, M, Woertgen, C, Metz, C and Brawanski, A
(1997) Dynamic changes of cerebral oxygenation
measured by brain tissue oxygen pressure and near infrared
spectroscopy. Neurol. Res., 19: 246248.
Iglesias, I, Murkin, JM, Bainbridge, D and Adams, S
(2003) Monitoring cerebral oxygenation saturation significantly decreases postoperative length of stay: a prospective randomized blinded study. Heart Surg. Forum,
6: 204.
Ing, RJ, Lawson, DS, Jaggers, J, Schulman, S, Shearer, IR
and Kern, FH (2004) Detection of unintentional partial
superior vena cava occlusion during a bidirectional
cavopulmonary anastomosis. J. Cardiothorac. Vasc.
Anesth., 18: 472474.
Isley, MR, Cohen, MJ, Wadsworth, JS, Martin, SP and
OCallaghan, MA (1998) Mulitmodality neuromonitoring for carotid endarterectomy surgery: determination of critical cerebral thresholds. Am. J.
Electroneurodiagn. Technol., 38: 65122.
Jobsis, FF (1977) Non-invasive infrared monitoring of cerebral and myocardial oxygen sufficiency and circulatory
parameters. Science, 198: 12641267.
Karapanayiotides, T, Meuli, R, Devuyst, G, PiechowskiJozwiak, B, Dewarrat, A, Ruchat, P, Von Segesser, L
and Bogousslavsky, J (2005) Postcarotid endarterectomy hyperperfusion or reperfusion syndrome. Stroke,
36: 2126.
Kassell, NF, Sasaki, T, Colohan, ART and Nazar, G (1985)
Cerebral vasospasm following aneurysmal subarachnoid
hemorrhage. Stroke, 16: 562572.
Kerr, MA, Nemoto, EM, Yonas, H and Kassam, AB (1999)
Cerebral oximetry by near-infrared spectroscopy
(NIRS) as an early indicator of delayed cerebral
935
ischemia (DCI) following subarachnoid hemorrhage
(SAH). Crit. Care Med., 27: 132.
Kerr, ME, Mario, D, Sereika, SM, Weber, BB, Orndoff,
PA, Henker, R and Wilberger, J (2000) The effect of
cerebrospinal fluid drainage on cerebral perfusion in
traumatic brain injured adults. J. Neurosurg. Anesthesiol., 12: 324333.
Kim, MB, Ward, DS, Cartwright, CR, Kolano, J, Chlebowski, S and Henson, LC (2000) Estimation of jugular
venous O2 saturation from cerebral oximetry or arterial
O2 saturation during isocapnic hypoxia. J. Clin. Monit.
Comput., 16: 191199.
Kirkpatrick, PJ, Smielewski, P, Czosnyka, M, Menon, DK
and Pickard, JD (1995) Near-infrared spectroscopy use
in patients with head injury. J. Neurosurg., 83: 963970.
Kishi, K, Kawaguchi, M, Yoshitani, K, Nagahata, T and
Furuya, H (2003) Influence of patient variables and sensor location on regional cerebral oxygenation saturation
by INVOS 4100 near-infrared spectrophotometers.
Kragsterman, B, Parsson, H and Bergqvist, D (2004) Local
haemodynamic changes during carotid endarterectomy
the influence on cerebral oxygenation. Eur. J. Vasc.
Endovasc. Surg., 27: 398402.
Kuroda, S, Houkin, K, Abe, H, Hoshi, Y and Tamura, M
(1996) Near-infrared monitoring of cerebral oxygenation
state during carotid endarterectomy. Surg. Neurol., 45:
450458.
Kurth, CD, Steven, JM and Nicolson, SC (1995) Cerebral
oxygenation during pediatric cardiac surgery using deep
hypothermic circulatory arrest. Anesthesiology, 82:
7482.
Kurth, CD, Steven, JL, Montenegro, LM, Watzman, HM,
Gaynor, JW, Spray, TL and Nicolson, SC (2001) Cerebral oxygen saturation before congenital hear surgery.
Ann. Thorac. Surg., 72: 187192.
Kussman, BD, Wypij, D, DiNardo, JA, Newburger, J, Jonas,
RA, Bartlett, J, McGrath, E and Laussen, PC (2005) An
evaluation of bilateral monitoring of cerebral oxygenation
saturation during pediatric cardiac surgery. Anesth. Analg.,
101: 12941300.
Laffey, JG, Mehigan, D and Boylan, JF (2000)
Postoperative neurological deficit despite normal cerebral oximetry during carotid endarterectomy. J. Clin.
Anesth., 12: 573574.
Lewis, SB, Myburgh, JA, Thorton, EL and Reilly, PL
(1996) Cerebral oxygenation monitoring by nearinfrared spectroscopy is not clinically useful in patients
with severe closed head-injury: a comparison with jugular venous bulb oximetry. Crit. Care Med., 24:
13341338.
Lozano, S and Mossad, E (2004) Cerebral function monitors
during pediatric cardiac surgery: can they make a difference? J. Cardiothorac. Vasc. Anesth., 18(5): 645656.
936
Marmarou, A, Anderson, RL and Ward, JD (1991) Impact
of ICP instability and hypotension on outcome in
patients with severe head trauma. J. Neurosurg., 75:
559566.
McCormick, PW, Stewart, M, Goetting, MG, Dujovny, M,
Lewis, G and Ausman, JI (1991) Noninvasive cerebral
optical spectroscopy for monitoring cerebral oxygen
delivery and hemodynamics. Crit. Care Med., 19:
8997.
McGrade, HC, El-Feky, W, Andrefsky, J and Frank, JL
(1999) Cerebral venous oxygenation monitoring in
patients with tissue shifts and raised ICP. Neurology,
52: A60.
Menke, J, Voss, U, Mller, G and Jorch, G (2003) Reproducibility of cerebral near-infrared spectroscopy in neonates. Biol. Neonate, 83: 611.
Mille, T, Tachimiri, ME, Klersy, C, Ticozzelli, G, Bellinzona, G, Blangetti, I, Pirrelli, S, Lovotti, M and Odero,
A (2004) Near infrared spectroscopy monitoring during
carotid endarterectomy: which threshold value is critical? Eur. J. Vasc. Endovasc. Surg., 27: 646650.
Minassian, AT, Poirier, N, Pierrot, M, Menei, P, Granry,
JC, Ursino, M and Beydon, L (1999) Correlation
between cerebral oxygen saturation measured by nearinfrared spectroscopy and jugular oxygen saturation in
patients with severe close head injury. Anesthesiology,
91: 985990.
Morimoto, Y, Niida, Y, Hua, Y, Kemmotsu, O, Murashita,
T and Yasuda, K (2003) Changes in cerebral oxygenation in children undergoing surgical repair of ventricular
septal defects. Anaesthesia, 58: 7783.
Murkin, JM (2004) Perioperative detection of brain oxygenation and clinical outcomes in cardiac surgery. Semin. Cardiothorac. Vasc. Anesth., 8: 1314.
Nagdyman, N, Fleck, T, Schubert, S, Ewert, P, Peters, B,
Lange, PE and Abdul-Khaliq, H (2005) Comparison
between cerebral tissue oxygenation index measured by
near-infrared spectroscopy and venous jugular bulb saturation in children. Intensive Care Med., 31: 846850.
Nemoto, EM, Yonas, H and Kassam, A (1998) Cerebral
oximetry: cerebral oxygen supply and demand. J. Neurosurg. Anesthesiol., 10(4): 266.
Nemoto, EM, Yonas, H and Kassam, A (2000) Clinical experience with cerebral oximetry in stroke and cardiac arrest.
Crit. Care Med., 28: 10521054.
Newman, MF, Kirchner, J, Phillips-Bute, B, Gaver, V, Grocott, H, Jones, RH, Mark, DB, Reves, JG and Blumenthal,
JA (2001) Longitudinal assessments of neurocognitive
function after coronary-artery bypass surgery. N. Engl.
J. Med., 344: 395402.
Novitzky, D (2005) Reducing the risk of myocardial revascularization: relevance of multimodal brain monitoring.
Semin. Cardiothorac. Vasc. Anesth., 9: 131137.
Ogasawara, K, Konno, H, Yukawa, H, Endo, H, Inoue, T and
Ogawa, A (2003) Transcranial regional cerebral oxygen
J.R. BALZER ET AL.

saturation monitoring during carotid endarterectomy as
a predictor of postoperative hyperperfusion. Neurosurgery, 53: 309314.
Ogino, H, Ueda, Y, Sugita, T, Morioka, K, Sakakibara, Y,
Matsubayashi, K and Nomoto, T (1998) Monitoring of
regional cerebral oxygenation by near-infrared spectroscopy during continuous retrograde cerebral perfusion
for aortic arch surgery. Eur. J. Cardiothorac. Surg.,
14: 415418.
Orihashi, K, Sueda, T, Okada, K and Imai, K (2004) Nearinfrared spectroscopy for monitoring cerebral ischemia
during selective cerebral perfusion. Eur. J. Cardiothorac. Surg., 26: 907911.
Pigula, FA, Nemoto, EM, Griffith, BP and Siewers, RD
(2000) Regional low-flow perfusion provides cerebral
circulatory support during neonatal aortic arch reconstruction. J. Thorac. Cardiovasc. Surg., 119: 331339.
Pigula, FA, Gandhi, SK, Siewers, RD, Davis, PJ, Webber,
SA and Nemoto, EM (2001) Regional low-flow perfusion provides somatic circulatory support during neonatal aortic arch surgery. Ann. Thorac. Surg., 72:
401407.
Plyuscheva, N, Doblar, DD, McDowel, H and Jordan, W
(1995) Regional oxygen saturation (rSO2), TCD velocity
(FV), and EEG after crossclamping during carotid endarterectomy (CEA). J. Neurosurg. Anesthesiol., 7: 303.
Prabhune, A, Sehic, A, Spence, PA, Church, T and
Edmonds, HL (2002) Cerebral oximetry provides early
warning of oxygen delivery failure during cardiopulmonary bypass. J. Cardiothorac. Vasc. Anesth., 16:
204206.
Reents, W, Muellges, W, Franke, D, Babin-Ebell, J and
Elert, O (2002) Cerebral oxygen saturation assessed by
near-infrared spectroscopy during coronary artery
bypass grafting and early postoperative cognitive function. Ann. Thorac. Surg., 74: 109114.
Rigamonti, A, Scandroglio, M, Minicucci, F, Magrin, S,
Carozzo, A and Casati, A (2005) A clinical evaluation of
near-infrared cerebral oximetry in the awake patient to
monitor cerebral perfusion during carotid endarterectomy.
J. Clin. Anesth., 17: 426430.
Roach, GW, Kanchuger, M, Mangano, CM, Newman, M,
Nussmeier, N, Wolman, R, Aggarwal, A, Marschall,
K, Graham, SH and Ley, C (1996) Adverse cerebral
outcomes after coronary bypass surgery. Multicenter
study of perioperative ischemia research group and the
ischemia research and education foundation investigators. N. Engl. J. Med., 335: 18571863.
Samra, SK, Dorje, P, Zelenock, GB and Stanley, JC (1996)
Cerebral oximetry in patients undergoing carotid endarterectomy under regional anesthesia. Stroke, 27: 4955.
Samra, SK, Stanley, JC, Zelenock, GB and Dorje, P (1999)
An assessment of contributions made by extracranial
tissues during cerebral oximetry. J. Neurosurg. Anesthesiol., 11: 15.
OTHER ISSUES
Samra, SK, Dy, EA, Welch, K, Dorje, P, Zelenock, GB and
Sanley, JC (2000) Evaluation of a cerebral oximeter as a
monitor of cerebral ischemia during carotid endarterectomy. Anesthesiology, 93: 964970.
Sapire, KJ, Gopinath, SP, Farhat, G, Thakar, DR, Gabrielli,
A, Jones, JW, Robertson, CS and Chance, B (1997)
Cerebral oxygenation during warming after cardiopulmonary bypass. Crit. Care Med., 25: 16551662.
Scholl, FG, Webb, D, Christian, K and Drinkwater, DC (2006)
Rapid diagnosis of cannula migration by cerebral oximetry
in neonatal arch repair. Ann. Thorac. Surg., 82: 325327.
Schwarz, G, Litcher, G, Kleinert, R and Jobstmann, R
(1996) Cerebral oximetry in dead subjects.
Sehic, A and Thomas, MH (2000) Cerebral oximetry during carotid endarterectomy: signal failure resulting from
large frontal sinus defect. J. Cardiothorac. Vasc.
Anesth., 14: 444446.
Shaaban Ali, M, Harmer, M, Vaughan, RS, Dunne, JA and
Latto, IP (2001) Spatially resolved spectroscopy
(NIRO-300) does not agree with jugular bulb oxygen saturation in patient undergoing warm bypass surgery. Can.
J. Anaesth., 48: 497501.
Sokol, DK, Markand, ON, Daly, ED, Luerssen, TG and
Malkoff, MD (2000) Near infrared spectroscopy (NIRS)
distinguishes seizure types. Seizure, 9: 323327.
Steinhoff, BJ, Herrendorf, G and Kurth, CD (1996) Ictal
near-infrared spectroscopy in temporal lobe epilepsy: a
pilot study. Seizure, 5: 91101.
Takeda, N, Fujita, K, Katayama, S and Tamaki, N (2000)
Cerebral oximetry for the detection of cerebral ischemia
during temporary carotid artery occlusion. Neurol. Med.
Chir. (Tokyo), 40: 557562.
Taylor, KM (1998) Brain damage during cardiopulmonary
bypass. Ann. Thorac. Surg., 65: S20S26.
937
Tortoriello, TA, Stayer, SA, Mott, AR, McKenzie, ED,
Fraser, CD, Andropoulos, DB and Chang, AC (2005) A
noninvasive estimation of mixed venous oxygen saturation using near-infrared spectroscopy by cerebral oximetry in pediatric cardiac surgery patients. Paediatr.
Anaesth., 15: 495503.
Vets, P, Ten Broecke, P, Adriaensen, H, Van Schil, P and
De Hert, S (2004) Cerebral oximetry in patients undergoing carotid endarterectomy: preliminary results. Acta
Anaesthesiol. Belg., 55: 215220.
Watzman, HM, Kurth, CD, Montenegro, LM, Rome, J,
Steven, JM and Nicolson, SC (2000) Arterial and
venous contributions to near infrared cerebral oximetry.
Weir, B (1995) The pathophysiology of cerebral vasospasm. Br. J. Neurosurg., 9: 375390.
Weiss, M, Dullenkopf, A, Kolarova, A, Schulz, G, Frey, B
and Baenziger, O (2005) Near-infrared spectroscopic
cerebral oxygenation reading in neonates and infants is
associated with central venous oxygen saturation. Paediatr. Anaesth., 15: 102109.
Williams, IM, Mead, GE, Picton, AJ, Farrell, A, Mortimer,
AJ and McCollum, CN (1995) The influence of contralateral stenosis and occlusion on cerebral oxygen saturation
during carotid artery surgery. Eur. J. Vasc. Endovasc.
Surg., 10: 198206.
Yao, FSF, Levin, SK, Wu, D, Illner, P, Yu, J, Huang, SW and
Tseng, CC (2001) Maintaining cerebral oxygen saturation during cardiac surgery shortened ICU and hospital
stays. Anesth. Analg., 92: SCA86.
Yao, FSF, Tseng, CCA, Ho, CYA, Levin, SK and Illner, P
(2004) Cerebral oxygen desaturation is associated with
early postoperative neuropsychological dysfunction in
patients undergoing cardiac surgery. J. Cardiothorac.
Vasc. Anesth., 18: 552558.

M.R. Nuwer (Ed.)
939
Subject Index
Acoustic nerves, see Auditory pathway

Action potential, see Nerve action potential
Active transport
cell membrane, 1011
Analgesic agents, see also Anesthetic agents/effects
electroencephalography, 90
Anatomy
lumbosacral plexus surgery, 731732
major neural structures, 5458
meningeal coverings, 4447, 5253, 55
micturition/sexual function, 739746
motor system, 5868
motor system excitation, 288292
sacral roots/nerves, 423424
sensory system, 6876
skull bones, 45
spinal endovascular disorders, 651653
subarachnoid space, 5354
tethered cord syndrome, 698690
vascular system, 4854
ventricular system, 4748, 52
Anesthetic agents/effects
block, 41
brainstem mapping, 354
burst suppression, 7982
depth
awareness risk, 903904
bispectral index, 899905
clinical usefulness, 903
limitations, 905
modern monitors, 900901
morbidity/mortality reduction, 904
validation studies, 901902
direct cortical stimulation, 155
electrocorticography, 147
epileptiform activity, 87
frequency bands, 8587
GABAA receptors, 8788
muscle relaxants, 91
NMDA inhibitors, 8990
opioids, 90
other sedatives, 9091
patterns and topography, 8284
periodic patterns, 87
regional anesthesia, 91
evoked potentials, 94116
inhalational agents, 97103
intravenous analgesic agents
barbiturates, 106
benzodiazepine, 107
dexmeditomidine, 105106
droperidol, 110111
etomidate, 107108
ketamine, 105
opioids, 104105
propofol, 108110
sedative/hypnotic agents, 106, 115
local/regional anesthesia, 111
nitrous oxide, 103104
physiology, 112115
facial nerve monitoring, 374
lumbar stenosis, 680
magnetic cortical stimulation, 293295
membranes and ion channels, 7778
middle ear/mastoid surgery, 560
motor evoked potential, 223225
neurogenic mixed evoked potential, 274
oculomotor/lower cranial nerve monitoring, 386
pain surgery, 703
selected dorsal rhizotomy, 445
sleep and arousal, 7879
somatosensory evoked potential, 187, 195196
spinal intramedullary tumors, 635
spinal pedicle screw instrumentation, 418419
toxic effects, 7980
transcranial cortical stimulation, 434
Aneurysms
aortic procedures, 815, 817, 822
intracranial, 801812
Anogenital system, see Sacral roots/nerves,
and Neuro-urologic monitoring
940
Anorectal system, see Sacral roots/nerves,

and Neuro-urologic monitoring
Anterior cervical spine surgery, see also
Corticospinal tract monitoring
recurrent laryngeal nerve, 602
Aortic procedures
adjuncts, 817
confounding factors
anesthesia, 821
brain ischemia, 821
differentiating, 822
hemodynamics, 821
hypothermia, 821
limb ischemia, 821822
neuromuscular blockade, 821
potential fade, 821
technical obstacles, 822
endovascular procedures, 817
impact/outcome
open aneurysm repair, 822824
open coarctation repair, 824
open aneurysm repair, 817
open coarctation repair, 817
pathology
coarctation, 815
descending aneurysms, 815
somatosensory evoked potential, 818824
spinal cord arteries
anatomy, 815816
blood flow, 817
pathology, 816817
spinal cord ischemia/infarction, 817821
gray matter potentials, 819821
white matter potentials, 818819
transcranial electrical stimulation, 819824
Astrocyte
resting potential, 16
Auditory evoked potential, see Brainstem auditory
evoked potential
Auditory pathway
auditory brainstem response, 566588
cerebellopontine angle lesions, 568
epidermoid tumors, 573
functional microsurgery, 582588
surgery examples, 574582
cochlear vs. vestibular nerves, 572
meningioma surgery, 572573
schwannoma surgery, 570572
vascular compression syndrome, 573574
SUBJECT INDEX
Axolemnal disorders
peripheral and cranial nerve, 4041
Axoplasmic disorders
Basal ganglia
motor system, 67
Bispectral index
anesthetic depth, 899905
Brachial plexus surgery
entrapments, 725
nerve root assessment, 459462
neural tumors/lesions, 726727
biopsy, 726
traumatic lesions, 720725
nerve transfer, 723725
neuromas in continuity, 722
pre- vs. postganglionic, 721722
Brain
continuous electroencephalography
cerebral ischemia, 865871
clinical applications, 864
impact, 876877
intracerebral hemorrhage, 871873
intracranial pressure, 873876
limitations, 876
intensive care unit
major neural structures, 6061
mapping
motor areas of cortex, 66
perfusion
transcranial Doppler ultrasound,
909920
tumor, 491502
vascular system, 5557
Brainblood flow
electroencephalogram, 112
Brain perfusion
transcranial Doppler ultrasound,
909920
Brainstem
anatomy, 48
auditory evoked potential, 334347
corticospinal tract mapping, 323326
lesions, 522531
mapping, 350362
safety issues, 890
SUBJECT INDEX
Brainstem auditory evoked potential, see also

Evoked potential
analysis/interpretation, 340341
anesthetic depth monitor, 901
auditory pathway, 566588
brainstem lesions, 527530
carotid balloon test occlusion, 798799
cochlear microphonic, 335, 340341
components/sources, 335338
cranial base surgery, 537
intracranial aneurysms, 802
intraoperative changes
auditory system dysfunction, 342346
brainstem, 346
cochlea/labyrinth, 344345
eighth nerve, 345346
physiological effects, 342
technical problems, 341342
middle ear/mastoid surgery, 560, 563
recording, 339340
stimulation, 338339
surgical procedures, 340
Brainstem lesions
brainstem auditory evoked potential, 527530
mapping
anesthesia, 524
clinical value, 525527
literature review, 525527
recording technique, 524
safety aspects, 524525
stimulation
site, 525
technique, 522524
motor tract
clinical applications, 529531
motor evoked potential recording, 528529
motor evoked potential stimulation, 527528
Brainstem mapping
anesthetic effects, 354
clinical cases, 360362
electromyography, 351353
fourth ventricle floor
anatomy, 350351
function assessment, 351
interpretation of data, 357360
protocol, 354355
stimulation
current vs. voltage, 355
monopolar vs. bipolar, 355
parameters/safety, 353354
941
structures, 354
techniques, 351
technical considerations, 355357
Brain surgery
cranial base, 534543
epilepsy, 484488
movement disorders, 508517
tumors, 491503
Brain tumor
Brocas area, 494
calculation
angular gyrus, 494
corpus callosum, 498
frontal eye fields, 494496
improvements, 501502
language
dominant insula, 494
premotor cortex, 494
magnetic resonance imaging, 495, 500
methodology, 492493
pathophysiology, 493501
plasticity
intraoperative, 499
postoperative, 499
preoperative, 498499
surgical use, 499501
spatial awareness
right supramarginal gyrus, 496
posterior temporal areas, 496
subcortical pathways
language, 497498
motor, 496
somatosensory, 496
spatial awareness, 498
visual, 496
supplementary motor area, 493
Wernickes area, 494
Bulbocavernosus reflex response
sacral roots/nerve monitoring, 428430
Burst suppression
Carbon dioxide tension
transcranial Doppler ultrasound, 912913
Cardiac disease
Cardiac surgery, see Cardiovascular surgery
942
Cardiovascular surgery
advantages of monitoring, 831
changes, 832833
clinical aspects, 835836
deep hypothermia, 834835
preferred technique, 833
expectations, 836837
future, 837
nerve injury, 602
neurological complications
embolism, 829830
hypoperfusion, 830
ischemic penumbra, 830831
prognosis improvement, 830
outcome, 829
changes, 832833
clinical aspects, 835836
deep hypothermia, 834835
preferred technique, 833
transcranial cerebral oximetry
adult, 927928
pediatrics, 928929
Carotid balloon test occlusion
angiography, 793
cerebral blood flow, 791
protocol, 793794
Carotid endarterectomy
algorithms, 783
blood pressure adjustment, 780
cerebral blood flow, 785
contraindications, 781782
electroencephalography alterations, 777788
expectancies, 786788
future, 788
intraoperative parameters, 782784
nerve injury, 601602
outcome, 776777
pre-operative parameters, 782784
sensitivity/specificity, 785786
shunt, 779780
somatosensory evoked potential alterations,
777788
transcranial cerebral oximetry, 929931
SUBJECT INDEX
Cauda equine lesions, see Tethered cord syndrome

Cell membrane
transmembrane ion gradients, 10
equilibrium potential, 11
membrane potential, 12
neuronal excitability, 14
active transport, 1011
ion channels, 1214
Cerebellopontine angle, see Auditory pathway
Cerebellum
motor system, 6768
Cerebral, see Brain
Cervical myelopathy
anterior spinal artery, 672673
literature review, 673676
methodology
intervention criteria, 673
motor evoked potential, 672673, 675
spondylosis, 672
Chest surgery, see Neck and chest surgery
Cochlear microphonic
brainstem auditory evoked potential,
335, 340341
Compartment syndrome
epidural somatosensory evoked potential, 210
Compound muscle action potential
oculomotor/lower cranial nerve monitoring,
384394
peripheral nerve stimulation, 364369
transcranial electrical stimulation
anesthesia, 263270
safety, 263
technique, 261263
transcranial magnetic stimulation, 296299,
304, 310
Computer analysis/network, see Technological
advances
Conductance
action potential, 21
Control pathways
motor system, 6668
Coronary artery bypass graft
cardiac surgery, 829
Cortical stimulation, see Direct cortical
stimulation
SUBJECT INDEX
Corticospinal tract mapping

anatomy
brainstem, 320
cortical, 319
internal capsule, 319320
spinal cord, 320
brainstem stimulation, 324326
cortical level
brief pulse trains, 321322
prolonged trains, 322
somatosensory evoked potential polarity
reversal, 321
time-locked electromyography, 321322
dorsal column midline, 326327
internal capsule
subcortical stimulation, 322323
within spinal cord, 327328
Corticospinal tract monitoring
anatomy, 4951, 55
aortic procedures, 815824
cervical myelopathy, 671676
compound muscle action potential,
260270
D-wave collision technique
fast neuron number, 244246
mapping, 244
endovascular disorders, 651669
extramedullary tumors, 618629
fractures, 618629
fusion surgery, 678687
intramedullary tumors, 632647
methodology
cortical stimulation, 252253
D-waves, 237241
earthing, 254
electrical stimulation, 235237
epidural electrodes, 254
epidural recording, 254
filtering, 254
I-waves, 237241
tibial nerve stimulation, 253254
motor evoked potential with somatosensory
evoked potential, 252259
advantages/disadvantages, 257258
protocol, 258
motor evoked potential limb muscles
D- and I-waves
build up phenomenon, 243
facilitation, 241
943
number, 243
recovery time, 242243
motor system, 6266
pain surgery, spinal cord stimulation, 703
safety issues, 890891
scoliosis, 608615
spinal pedicle screw, 404419, 685
transient paraplegia, 248249
volleys
abnormality criteria, 257
real-time monitoring, 257
reproducibility, 257
size/morphology, 254255
stimulus intensity, 255256
Cranial base surgery
cardiovascular monitoring, 537538
clinical outcome, 534
craniocervical junction, 542543
electromyography
evoked activity, 535536
spontaneous activity, 544
visual vs. aural, 536
evoked potentials
brainstem, 537
motor, 537
olfactory, 536
somatosensory, 537
visual, 536537
foramen magnum, 542543
new technology, 543
surgical approach
anterior fossa, 542
middle fossa, 541
posterior fossa, 539541
Cranial nerves, see also Peripheral/Cranial nerve
anatomy, 48
Data acquisition
neurophysiological monitoring, 466479
Deep brain stimulation
clinical exam, 164
equipment
amplifiers/filters, 165166
anatomical atlases, 166
discriminators, 166
microdrive, 165
944
Deep brain stimulation (Continued)

recording electrodes, 165
stereotactic frame, 164165
stimulating electrodes, 165
pain surgery, 704, 715
Parkinsons disease, 509517
patient selection, 163164, 510511
procedures
anesthesia, 166167
physiologic map, 169
recordings, 167168
stimulation, 168169
troubleshooting, 169
target/procedure selection, 164
Diffusion tensor imaging
Direct cortical stimulation, see also Transcranial
electrical stimulation, and Brain surgery
anesthetic protocol, 155
brain mapping
language, 158159
motor system, 157
sensory system, 157158
subcortical stimulation, 159
brain tumor surgery, 491502
concurrent electrocorticography, 151152
localizing value, 154155
pediatrics, 159160
safety
charge density, 155156
electrodes, 156
kindling, 157
seizure, 156157
Dorsal root entry zone
pain surgery, 700715
procedure
nucleus caudalis, 711713
spinal cord, 708711
D-wave
corticospinal tract monitoring, 235250, 252259
Electrocochleography
Electrocorticography
activity, 147
SUBJECT INDEX
applications/limitations, 142145
co-registration imaging, 145146
digital acquisition system, 146
electrodes, 145
reference/ground, 146
stereotactic placement, 145146
epilepsy surgery, 485488
magnetic resonance imaging, 145146
montage, 146
physiologic range, 141142
positron emission tomography, 145146
safety, 145
single photon emission computed tomography,
145146
troubleshooting
channels not recording, 147
complications, 147148
electrode artifact, 147
Electroencephalography
anesthetic effects, 7791, 128129
brainblood flow, 112
depth, 899905
epileptiform activity, 87
frequency bands, 8587
GABAA receptors, 8789
NMDA inhibitors, 8990
opioids, 90
other sedatives, 9091
patterns and topography, 8284
periodic patterns, 87
regional anesthesia, 91
bispectral index, 899905
carotid endarterectomy, 777788
display, 138139
electrodes, 129
general equipment, 129
hypothermia, 132134
intensive care unit
epilepsy, 856862
focal cortical disorders, 864877
neonatal, 840853
vasospasm, 864877
ischemia, 130132
montages, 129130
processed monitoring, 134138
safety issues, 887
SUBJECT INDEX
Electromyography
brachial plexus surgery
neural tumors/lesions, 726
corticospinal tract mapping, 321322, 327
free running
application, 402403
electrode placement, 401
pitfalls, 402
potential recognition, 396397
recording electrodes, 397401
recording parameters, 401402
spinal procedures, 405406, 412414
oculomotor/lower cranial nerve monitoring, 384394
pelvic surgery, 754761
safety issues, 891
spinal tumor/fracture, 618620, 625629
total hip arthroplasty, 748750
Energy failure
Epidural somatosensory evoked potential, see also
Evoked potential
advantages, 210
applicability, 202203
corticospinal tract monitoring, 252259
electrodes
recording, 203
stimulating, 204
generators, 206210
intervention criteria, 205206
recording
bipolar vs. monopolar, 204
parameters, 203204
stimulation
parameters, 205
sites, 204205
Epilepsy
acute seizures
treatment, 861862
945
electroencephalography
intensive care unit, 856862
intensive care unit
neonatal electroencephalography, 842843
status epilepticus
definitions, 856
surgery
neurophysiology, 484488
Equilibrium potential
cell membrane, 11
Esophagus surgery
nerve injury, 604
Evoked potential
inhalational agents, 97103
intravenous analgesic agents, 104106
local/regional anesthesia, 111
nitrous oxide, 103104
physiology, 112115
neurogenic mixed evoked potential, 273280
pain surgery
dorsal root entry zone procedure, 708713
somatosensory evoked potential
dermatomal stimulation, 190198
scalp/cervical recording, 180189
visual evoked potential, 172176
External anal sphincter
Facial nerve
electrodes
recording, 372
stimulating, 372373
electromyography
anesthetics, 374
limitations/pitfalls, 378380
mechanical activity, 376377
other modalities, 380382
946
Facial nerve (Continued)

principles, 375
spontaneous activity, 376
hemifacial spasm, 547554
iatrogenic injury, 371
instrumentation, 371372
microvascular decompression
abnormal muscle response, 549516
advantages, 525
facial/auditory nerve monitoring, 553554
risk of injury, 553554
surgeon/physiologist, 552553
parotid surgery, 556558
stimulation parameters, 373
Final common pathway
motor system, 6163
Fusion surgery
F-waves
GABA receptors
anesthetic effects, 78, 8789, 96
Genitourinary tract, see Neuro-urologic
monitoring, and Sacral roots/nerves
Glial cells
Globus pallidus interna
movement disorders, 509512, 516
Glossopharyngeal nerve
nerve injury, 604
Heart, see Cardiac
Hemifacial spasm
clinical signs, 547
electrophysiology, 547
facial nerve decompression, 547554
pathophysiology
anatomical location, 548549
cause, 549
treatment, 547548
Hip arthroplasty, see Total hip arthroplasty
Histology
myelinated motor nerve, 35
peripheral nerve, 35
History
intraoperative monitoring, 24
transcranial magnetic stimulation,
282283
SUBJECT INDEX
Hoffman reflex
H-reflex
Hypotension
anesthetic effects
Hypothermia
anesthetic effects
heart surgery, 834
Infection control
Intensive care unit
epilepsy, 856862
focal cortical disorders, 864877
newborn, 840853
vasospasm, 864877
Intracranial aneurysms
anterior circulation, 804805
brain retraction, 808
future, 812
limitations, 810812
mechanical vasospasm, 808
outcome, 808809
posterior circulation, 805809
staffing/equipment, 810
vessel occlusion
inadvertent, 807
permanent, 808
temporary, 807
Intraoperative monitoring
technologist staffing, 4
supervision, 4
history, 24
physician training, 45
Ion channels
blockade
pathophysiology, 33
calcium channel, 14
cell membrane, 1214
SUBJECT INDEX
Ion concentration neuron

physiology, 8
Ion gradient
transmembrane, 8, 10
Ischemia
anesthetic effects
brain, 821, 865871
limb, 821
spinal cord, 817821
I-wave
combination with motor evoked potential,
241243, 246248
corticospinal tract monitoring, 235250,
252259
Kyphoscoliosis
Language
Limb muscles
corticospinal tract monitoring, 235250,
252259
Local potentials
action potential comparison, 2122
characteristics, 1820
ion channel, 1214
ionic basis, 18
Lumbar stenosis
anesthesia, 680
cases, 680685
fusion surgery, 678687
outcome, 680
techniques, 678
Lumbosacral plexus surgery
anatomy, 731732
F-waves, 736737
Hoffman reflex, 736
motor nerve conduction, 735736
recommendations, 737
947
Magnetic resonance imaging

brachial plexus surgery, 722, 726
brain surgery, 493, 501502
pain surgery
motor cortex stimulation, 715
radiofrequency burns, 886
spinal extramedullary tumor/fracture, 619
spinal tumors, 633634
Magnetic stimulation, see Transcranial magnetic
stimulation
Mastoid, see Middle ear/Mastoid surgery
Membrane potential
physiology, 89, 12
theoretical model, 17
Meningeal coverings
anatomy, 4447, 5253, 55
Meningioma surgery
Microvascular decompression
Micturition/Sexual function
bulbocavernosus reflex, 744745
filum terminale, 744
functional anatomy, 739
intracavernosus pressure, 745746
intrarectal pressure, 745746
intravesical pressure, 745746
motor roots, 744
pudendal nerve, 743744
recording technique, 740743
sphincter motor response, 744745
stimulation technique, 739740
Middle ear/Mastoid surgery
anesthesia, 560
auditory brainstem response, 560, 563
electrocochleography, 560563
facial nerve
complications, 560
cost of monitoring, 559
indications, 560
methods, 561563
practice patterns, 559560
surgical draping, 561
948
Motor evoked potential

anesthetic effects
axonal conduction, 223
neural synaptic transmission, 224225
neuromuscular junction, 223224
brachial plexus surgery, 722
cervical myelopathy, 672673, 675
compound muscle action potential, 260270
corticospinal tract mapping, 321322, 324
combined with somatosensory evoked potential,
252259
D- and I-wave spinal cord, 241243, 246248,
252259
external anal sphincter, 434438
limb muscles, 235250
nerve root assessment
brachial plexus reconstruction, 459462
interpretation, 459
techniques, 455456
neurophysiological circuitry, 218219
pelvic surgery, 758
physiology
children, 229
epidural recording, 225226
facilitation by conditioning, 227229
H-reflex, 226
risks, 229230
safety issues, 888, 890891
scoliosis, 612615
spinal intramedullary tumors, 635647
stimulation
axons, 219
cortical, 222
electrical aspects, 222223
electrodes, 220223
transcranial
electrical, 219230
magnetic, 219221
transcranial electrical stimulation
muscle recording, 260270
transcranial magnetic stimulation, 297305, 311313
SUBJECT INDEX
trigeminal rhizotomy, 704

urethral sphincter, 434438
Motor system
anesthetic effects
brainstem lesions
motor tract, 527531
control pathways, 6668
corticospinal tract, 6266
deep brain stimulation, 163170
final common pathway, 6163
neurogenic mixed evoked potential, 275,
278279
pain surgery, cortex stimulation, 704, 713715
somatosensory evoked potential cortex mapping,
211214
transcranial magnetic stimulation
anatomy/physiology, 289292
Movement disorders
adverse events, 516517
general principles, 509511
history, 508509
patient selection, 511
results, 514516
surgical technique, 511514
Muscle relaxants
Myelin disorders
N20 waves
Near-infrared spectroscopy
Neck and chest surgery
glossopharyngeal nerve, 604
heart surgery, 603
neck surgery, 601603
thoracic surgery, 603604
thyroid surgery, 590601
vagus nerve, 601602
Neonatal intensive care unit
electroencephalography amplitude integrated
background, 842
cerebral function motor, 840841
SUBJECT INDEX
classification
abnormal brain function, 850851
new model, 851852
normal infants, 850
electrodes, 842
filter settings, 841
full-term, 848
medication effect, 849850
preterm, 848849
seizure patterns, 842843
Nerve action potential
brachial plexus surgery, 722, 725726
conductance, 21
cutaneous nerve, 39
dorsal root
excitability, 2223
ionic basis, 2022
local potential comparison, 2122
nerve trunk, 38
patterns of activity, 2526
peripheral nerve stimulation, 364369
physiology, 810
propagation, 2325
sensory nerve
strengthduration curve, 39
threshold, 20
Nerve fiber types, 40
Nerve root assessment
Nerve transfer
brachial plexus surgery, 723725
Neurogenic mixed evoked potential
anesthetic parameters, 274
motor tract assessment
collision techniques, 278
myogenic recordings, 278279
recording methods, 274
research
animal performance, 275276
clinical performance, 277278
tract localization, 276277
signal interpretation, 274
stimulation methods
epidural spinal, 274
percutaneous cervical lamina, 273274
within surgical field, 274
949

Neuromodulation
classic neurotransmission, 30
neurotransmitters, 2829
Neuronal disorders, see also Pathophysiology
axolemmal disorders, 4041
axoplasmic disorders, 4142
energy failure, 3033
myelin disorders, 4243
transient neuronal disorders, 31
Neuronal excitability
cell membrane, 14
Neurophysiology
brain tumor surgery, 491502
technological advances in monitoring, 464480
Neurotransmission, see Synaptic transmission,
and Neurotransmitters
Neurotransmitters
biosynthesis/storage/reuptake, 2627
postsynaptic effects
classic neurotransmission, 28
neuromodulation, 2829
Neuro-urologic monitoring, see also
Sacral roots/nerves
Newborn, see Neonatal
NMDA receptors
Oculomotor/lower cranial nerve monitoring
anesthesia, 386
assessor nerve, 392
function, 390
glossopharyngeal nerve, 390
hypoglossal nerve, 393
technique, 386390
vagus nerve, 390392
Operating room
950
Opioids
Oscilloscope
Oximetry, see Transcranial cerebral oximetry
Pain surgery
anatomy, 700701
classification
ablative, 702703
augmentative, 703704
dorsal root entry zone, 700715
Paraplegia
transient
Parkinsons disease, see also
Movement disorders
deep brain stimulation, 509517
Parotid surgery
facial nerve
indications, 558
practice patterns, 557
Pathophysiology
brain tumors, 493501
energy failure, 3033
ion channel blockade, 33
myelin disorders, 4243
synaptic transmission, 33
transient neuronal disorders, 31
Pediatrics
cardiovascular surgery
Pelvic surgery, see also Neuro-urologic
monitoring, and Sacral roots/nerves
acetabular fracture
clinical outcomes, 760761
monitoring techniques, 759
pelvic ring
clinical outcomes, 756758
limitations, 758
monitoring techniques, 754756
SUBJECT INDEX
Peripheral nerve stimulation

corticospinal tract monitoring
motor evoked potential combined with
lesions, 367369
nerve conduction techniques, 364367
pain surgery, 703
technical problems, 367
Peripheral/cranial nerve
axolemnal disorders, 4041
axonal transport, 3637
axoplasmic disorders, 4142
blood supply, 60
entrapment, 767
fascicular biopsy, 769
fascicular nerve transfer, 769770
histology, 3436
nerve fiber types, 40
nerve transaction, 765767
neuromas in continuity, 764765
physiology, 3740
procedures around nerve, 770772
stimulation, 252259, 364369, 703
tumors, 767769
Physician training
Physiology
action potential, 2026
cell membrane, 1014
clinical correlations, 3033
ion concentration neuron, 8
local potentials, 1620
membrane potential, 89, 12
overview, 710
synaptic transmission, 2630
Positron emission tomography
Pudendal nerve
Recurrent laryngeal nerve
basic principles, 594595
clinical usefulness, 595
glottis pressure, 593
laryngeal palpation, 591
muscle relaxation, 599600
SUBJECT INDEX
neck and chest surgery, 590604

needle electrodes
vocal cords, 592593
nerve identification, 596597
pitfalls, 595596
predictive value, 598599
risk factors, 597
surface electromyography, 592593
thyroid surgery, 590600
vocal cord
movement, 594
needle electrodes, 591592
postoperative function, 596
Resting potential
extracellular calcium, 16
glial cells, 16
oscilloscopic recording, 15
sodium pump, 16
steady state, 16
Rhizotomy, see also Selected dorsal rhizotomy
pain surgery, 702
trigeminal, 704707
Sacral roots/nerves, see also Neuro-urologic
monitoring
anatomy, 423424
mapping
nervous structures, 428430
neural tissue, 425427
pudendal afferents, 427428
neurophysiological tests, 424425
neurophysiology, 425430
Safety issues
brainstem, 524525, 554555
electric shock
heart failure, 882
inspection/maintenance, 884
leading current limits, 883884
leakage currents, 882883
patient lead connectors, 884
power cords, 884
protection, 883
single fault safe, 883
electrical burns
direct current, 886887
951

protection, 885886
radiofrequency, 885886
electrical stimulation parameters, 887
essential performance
monitoring equipment, 893894
monitoring practice
misleading information, 894895
slow feedback, 894
fire hazards, 884
infection control
extra precautions, 893
hand washing, 892893
needle-stick injuries, 893
protective equipment
patient care, 893
personnel, 893
protection against injury
electrical thermal, 887888
neural excitotoxicity, 888
neuronal electrochemical, 888889
specific procedures
brainstem electrical stimulation, 890
invasive techniques, 889890
spinal cord stimulation, 890891
Schwannoma surgery
Scoliosis
limitations/problems, 615
multicenter survey, 614615
outcome, 613614
surgical complications, 612
techniques, 609612
Sedatives
Seizure, see Epilepsy
Selected dorsal rhizotomy
neurophysiological techniques, 439440
Peacocks criteria, 440441
PhillipsParks grading scheme,
441442
preservation of bowel/bladder function
anal sphincter recording, 443444
criteria reliability, 444445
pudendal afferent mapping, 443
surgical approaches, 440
952
Sensorimotor cortex mapping

Sensory system
anatomy
pain and temperature, 6975
proprioception, 6974
touch, 6970
vibration, 69
hemispheric distribution, 68
peripheral nerves
cutaneous distribution, 7071
somatosensory evoked potential cortex mapping,
211214
Sexual function, see Micturition/Sexual function
Single photon emission computed tomography
Skull bones, 45
Sleep and arousal
anesthetic effects, 7879, 96
Sodium pump
Somatosensory evoked potential, see also Evoked
potential
brainstem lesions, 527, 531
carotid endarterectomy, 777788, 929931
cervical myelopathy, 671676
corticospinal tract monitoring
combined with motor evoked potential, 252259
dermatomal stimulation
data acquisition, 193
electrodes, 193
operating room, 193198
sites, 192193
stimulator, 193
technique, 192
epidural somatosensory evoked potential,
202210
nerve root assessment
interpretation, 459
techniques, 455456
SUBJECT INDEX
pain surgery
safety issues, 887, 890891, 894895
scalp/cervical recording
alarm criteria, 185188
false results, 188
generators, 184186
peaks
lower extremity, 184185
upper extremity, 185186
recording
filters, 183184
lower extremity channels, 182183
upper extremity channels, 183
stimulation, 181
rate/intensity, 181182
sites, 181
techniques, 180
validity, 189
scoliosis, 612615
sensorimotor cortex mapping
method, 211212
phase reversal, 211214
spinal intramedullary tumor, 636640
tethered cord syndrome, 693694, 696
transcranial magnetic stimulation, 298305
Somatosensory pathways
anatomy, 7076
Spinal cord monitoring, see Corticospinal tract
monitoring
Spinal endovascular disorders
anatomy, 651653
angiography, 653654
clinical application
tumors, 665669
vascular malformations, 659665
pharmacological provocative testing, 654659
Spinal extramedullary tumors
case studies, 628629
electromyography, 618620, 625629
en bloc resections, 621622
SUBJECT INDEX
intradural, 620621
metastasis
palliative surgery, 621
Spinal fractures
case studies, 628629
decompression, 619
electromyography, 618620, 625629
instrumentation, 620
positioning, 619
reduction, 619620
Spinal intramedullary tumors
perspectives, 646647
sensitivity, 643645
treatment
anesthesiology, 635
clinical presentation, 633
diagnostic studies, 633634
outcome, 635
surgery, 634635
value, 645646
Spinal pedicle screw
anesthetics, 418419
cervical spine
free running electromyography, 405406
muscle selection, 407408
recording/stimulation parameters, 408
stimulus evoked electromyography, 406407
decompression, 404405
lumbosacral spine
electromyography discrepancies, 417418
free running electromyography, 412414
recording/stimulation parameters, 418
stimulus evoked electromyography, 414417
thoracic spine
953
Spinal tract monitoring, see Corticospinal tract
monitoring
Spine surgery, see Corticospinal tract monitoring
Stroke, see Brain
Subarachnoid space
anatomy, 5254
Subthalamic nucleus
deep brain stimulation, 167173, 510
Supervision
Sympathectomy
pain surgery, 702
Synaptic transmission
electrical synapses, 2930
neuromodulation, 30
neurotransmitters, 2629
pathophysiology, 34
Technological advances
data acquisition
analog filtering, 468
analog to digital conversion, 469470
common mode rejection ratio, 469
digital filtering, 470
multimodality signal averaging, 470
remote monitoring system
network structure, 471
NeuroNet communication protocol, 473479
sensitivity, 469
signal amplification, 468469
stimulus timing, 470
user interfaces
data display, 470471
multimodality
neurophysiological variables, 465
Tethered cord syndrome
anatomy/pathology, 689690
future direction, 697
monitoring paradigms
motor pathway, 691692
proposed, 697
sensory pathway, 693694
sphincter function, 694
somatosensory evoked potential, 693694, 696
954
Tethered cord syndrome (Continued)

surgical considerations, 690691
symptoms, 690
utility, 696697
Thermocoagulation
Thoracic surgery
nerve injury, 602603
Thyroid surgery
Tonsillectomy
glossopharyngeal nerve injury, 604
Total hip arthroplasty
studies, 748750
Transcranial cerebral oximetry
cardiovascular surgery
adult, 927928
pediatric, 928929
carotid endarterectomy, 929931
intensive care unit, 931933
near-infrared spectroscopy
head sensors, 925926
rationale, 925926
suppression extracerebral signal, 925
technique, 924925
Transcranial cortical stimulation, see Transcranial
electrical stimulation
Transcranial Doppler ultrasound
cerebral autoregulation, 912914
cerebral hyperperfusion, 916
cerebral hypoperfusion
aortic arch vessels, 914915
balloon occlusion, 914915
inadequate collateral flow, 914
intracranial artery, 914
malperfusion syndrome, 915
outflow obstruction, 915916
retrograde cerebral perfusion, 915
embolization
particulate
atheroma, 917918
gaseous, 918
lipids, 917
thrombus, 918
flowmetabolism uncoupling, 916971
impact on surgical outcome, 919920
limitations, 918919
principles, 909911
probe fixation, 911912
site selection, 911912
vasomotor reserve, 912914
SUBJECT INDEX
Transcranial electrical stimulation, see also Direct

cortical stimulation
anesthetics, 434
children, 229
corticospinal tract monitoring, 235250, 252259
direct cortical stimulation, 152154, 434438
combined with somatosensory evoked potential,
252259
external anal sphincter, 434438
muscle recording, 260270
urethral sphincter, 434438
radiofrequency burns, 886
bite injuries, 892
kindling, 891
movement-related injury, 891892
other adverse effects, 892
seizures, 891
technique, 434435
transcranial magnetic stimulation, 306308,
313314
Transcranial magnetic stimulation
anatomy/physiology
motor system excitation, 288292
anesthetic effects
descending motor systems, 292294
compound muscle action potential, 296299, 310
historical perspective, 282283
single pulse, 294303
coil shape, 294295
techniques
primer, 283288
train stimuli, 302306
transcranial electrical stimulation, 306308,
313314
Trigeminal rhizotomy
pain surgery, 704707
Urethral sphincter
Urogenital system, see Sacral roots/nerves
SUBJECT INDEX
Vagus nerve
neck surgery, 601602
Vascular compression syndrome
Vascular malformations
Vascular system
anatomy, 4452
brain, 5557
peripheral nerve, 60
spinal artery, 5859
Vascular tumors
Vasospasm
intensive care unit
955
Ventricular system
anatomy, 4749, 52
Vestibular nerves, see Auditory pathway
Visual evoked potential, see also Evoked potential
cases, 174175
efficacy, 175176
methods, 172173
postoperative deficits, 175
surgery monitoring, 174

Iom Marc Nuwer PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Iom Marc Nuwer PDF

Uploaded by

Copyright:

Available Formats

HANDBOOK OF CLINICAL NEUROPHYSIOLOGY

This page intentionally left blank

Handbook of Clinical Neurophysiology

Amsterdam  Boston  Heidelberg  London  New York

Radarweg 29, 1043 NX, Amsterdam, The Netherlands

# 2008, Elsevier B.V. All rights reserved.

Printed in the Netherlands

This page intentionally left blank

This page intentionally left blank

Department of Anesthesiology and Perioperative Medicine, Brigham and

D ABNM, University of Alexandria, 507 El Horreya Ave., School of

Department of Orthopedic Surgery, Mayo Clinic, 200 First Street SW,

Departments of Neurological Surgery and Neuroscience, University of

Department of Surgery/Neurosurgery, University of California, Los Angeles,

Center for Endovascular Surgery and Intraoperative Neurophysiology,

Mayo Clinic, Brachial Plexus Clinic, Rochester, MN 55905, USA.

Office of Research and Development, College of Health Sciences, Level 2,

Department of Surgery/Head and Neck Surgery, University of California,

Department of Surgery/Neurosurgery, University of California, Los Angeles,

Department of Neurology, Division of Epilepsy and EEG, Mayo Clinic, 200

Division of Critical Care Neurology and Comprehensive Epilepsy Center,

Department of Neurosurgery, Feinberg School of Medicine, Northwestern

Department of Neurological Surgery, University of Pittsburgh Medical

Department of Neurology, W8A, Mayo Clinic College of Medicine, 200

Department of Neurology, E8B, Mayo Clinic College of Medicine, 200 First

Intraoperative Neurophysiology, Institute for Neurology and Neurosurgery,

Department of Neonatology, Wilhelmina Childrens Hospital, UMC, PO Box

Department of Surgery, Vancouver General Hospital, EEG Lab., CP Ground

Department of Neurosurgery, Hopital Gui de Chauliac, CHU de Montpellier,

Cardiovascular Services, Surgical Monitoring Associates, Inc., 3712

Department of Clinical Neurophysiology, Cedars Sinai Hospital, 8700

Department of Clinical Neurology, Columbia University College of

Department of Neurology, Geffen School of Medicine, UCLA, 710

Department of Neurology, University of Kentucky Chandler Medical Center,

Division of Neurosurgery, Toronto Western Hospital, 399 Bathurst Street,

Department of Neurology, Keck School of Medicine, University of Southern

Department of Anesthesiology and Perioperative Medicine, Brigham and

Department of Neurological Surgery, University of Pittsburgh Medical

Department of Clinical Neurodiagnostic, Sharp Memorial Hospital, 7901

Neonatal Intensive Care Unit, Department of Pediatrics, University Hospital,

Department of Anesthesiology, Hotel Dieu, 3840 St. Urbain, Montreal, PQ

Comprehensive Epilepsy Center, Department of Neurology, Columbia

Department of Otolaryngology, New York University School of Medicine,

Department of Orthopedic Surgery, Mayo Clinic, 200 First Street SW,

Department of Clinical Neurophysiology, Tampere University Hospital,

Department of Clinical Neurophysiology, The National Hospital for

Department of Neurosurgery, University Medical Center Groningen, P.O.

Division of Neurosurgery and Spinal Program, Krembil Neuroscience

Division of Neurosurgery, Kantonsspital Luzern, CH-6000 Lucerne 16,

Department of Otolaryngology, New York University School of Medicine,

Department of Neurology, Montefiore Medical Center, 111 East 210th Street,

Department of Neurology and Neurological Sciences, Intraoperative

Section of Clinical Neurophysiology, Department of Neurosciences, King

Department of Neurology, Georgetown University, 3800 Reservoir Road

Department of Anesthesia and Critical Care, Massachusetts General Hospital

Department of Clinical Neurology, Columbia University College of

Department of Neurology, Georgetown University, 3800 Reservoir Road

Department of Orthopedic Surgery, Saint Louis University School of

School of Behavioral and Brain Sciences, University of Texas at Dallas, GR

Neurochirurgische Universitatsklinik, D-53105 Bonn, Germany.

Center for Endovascular Surgery, Institute for Neurology and Neurosurgery,

Department of Neurology and Clinical Neurophysiology, UCLA School of

Department of Clinical Neurophysiology, UCLA School of Medicine, Reed

Amsterdam Boston Heidelberg London New York