Professional Documents
Culture Documents
research-article2018
PRF0010.1177/0267659118766833PerfusionVogel et al.
Original paper
Perfusion
2018, Vol. 33(1S) 57–64
Veno-arterio-venous ECMO for septic © The Author(s) 2018
Reprints and permissions:
cardiomyopathy: a single-centre experience sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0267659118766833
https://doi.org/10.1177/0267659118766833
journals.sagepub.com/home/prf
Abstract
Introduction: The role of extracorporeal support for patients with septic shock remains unclear.
Methods: We conducted a retrospective analysis of our single-centre experience with veno-arterio-venous extracorporeal
membrane oxygenation (VAV ECMO) in adult patients with severe respiratory failure and septic cardiomyopathy. Clinical
data was extracted from electronic medical records including a dedicated ECMO referral and follow-up database.
Results: Twelve patients were commenced on VAV ECMO for septic cardiomyopathy for a median of four days (IQR 3.0
to 5.3) between 01/2014 and 12/2017. Five patients (41.7%) had a cardiac arrest prior to initiation of ECMO support. At
baseline, median left ventricular ejection fraction was 16.25% (IQR 13.13 to 17.5) and median PaO2/FiO2 ratio was 9 kPa
(IQR 6.5 to 12.0) [67.50 mmHg (IQR 48.75 to 90.00)]. The survival rate to hospital discharge for VAV ECMO was 75%
in this cohort. None of the surviving patients died within the follow-up period (median six month).
Conclusion: VAV ECMO is a feasible rescue strategy for a small proportion of patients with combined respiratory and
cardiac failure secondary to septic shock with septic cardiomyopathy. We provide a detailed report of our experience with
this technique. Further research is required comparing the different extracorporeal strategies directly to conventional
resuscitation and against each other.
Keywords
Extracorporeal life support; Extracorporeal membrane oxygenation (ECMO); Veno-arterio-venous extracorporeal
membrane oxygenation (VAV ECMO); Septic cardiomyopathy; Septic shock; Severe respiratory failure; Acute heart failure
Introduction
The role of extracorporeal membrane oxygenation respiratory function. However, some patients with
(ECMO) as a supportive therapy for adult patients severe pneumonia have a secondary septic cardiomy-
with septic shock remains unclear. Veno-venous (VV) opathy and in this group, a hybrid VAV configuration
ECMO corrects hypoxaemia, hypercapnoea and pH, can be used. We report the outcomes of a cohort of
which may improve cardiovascular function through a patients established on VAV-ECMO with combined
reduction in pulmonary arterial hypertension, but respiratory and cardiac failure secondary to septic
does not specifically provide mechanical haemody- shock with septic cardiomyopathy.
namic support. Veno-arterial (VA) ECMO has been
described as a potential rescue therapy for concomi-
tant cardiac failure, however, only in small cohorts and 1Department of Critical Care, Guy’s and St Thomas’ NHS Foundation
with a high variability in outcome. 1, 2 The use of hybrid Trust, Westminster Bridge Road, London, UK
configurations with venous drainage and divided 2Department of Clinical Perfusion, Guy’s and St Thomas’ NHS
return flow to arterial and venous cannulae (VAV Foundation Trust, Westminster Bridge Road, London, UK
ECMO) has been reported in patients with a combina-
Corresponding author:
tion of various acute heart and lung pathologies. 3, 4
Dominik J. Vogel, Department of Critical Care, Guy’s and St Thomas’
Hybrid configurations are usually commenced in NHS Foundation Trust, Westminster Bridge Road, London, SE1 7EH,
response to upper and lower body differential oxygen- UK.
ation as cardiac function recovers in advance of Email: dominik.vogel@gstt.nhs.uk
58 Perfusion 33(1S)
and a Seldinger technique. Standard configuration was fibrosis, granulomata or deposition of iron, and Congo
established in eleven (11/12) patients. One patient was red staining was negative for amyloid.
cannulated with a 21 Fr femoral access cannula, a 21 Fr
femoral venous cannula and a 15 Fr left femoral arterial
Pathogens and antimicrobial agents
return cannula due to small body habitus (150cm, 45kg).
A distinct causative pathogen was identified in eleven
Vasoconstrictor and inotropic support (91.7%) patients, with gram-positive cocci accountable
for the majority of cases (75%). Two patients had a
All patients received noradrenaline; hourly rates are staphylococcal superinfection on the background of
shown in Figure 1. Furthermore, every patient received influenza. One patient had multiple bacterial growth
a continuous infusion of levosimendan at a rate of 0.10 (Staph. aureus, streptococci and gram-negative rods)
micrograms/kilogram/hour, commenced either prior to detected from a sample obtained with a non-directed
or during VAV-ECMO support. Five (41.7%) patients bronchoalveolar lavage prior to retrieval but no further
received additional support with vasopressin. positive bacterial cultures; PCR revealed herpes simplex
Adrenaline, dobutamine or milrinone were given in two virus and rhinovirus. Table 1 shows the primary source
(16.7%) patients, respectively. Figure 2 shows develop- of sepsis and the identified pathogens. All patients had
ment of serum lactate concentration relative to time of been started on empirical broad-spectrum antimicro-
VAV ECMO initiation. bial agents prior to referral and this regimen was
adjusted according to local guidelines or available results
on admission. There was no evidence that any of the
Cardiac support and monitoring
causative pathogens was not covered by the agents
All twelve patients had a baseline transthoracic echocar- administered prior to initiation of VAV-ECMO.
diogram showing severe left ventricular systolic impair- Inflammatory markers are shown in Figure 3.
ment. Median left ventricular ejection fraction (LVEF) According to our local guidelines, patients with sep-
at time of VAV cannulation was 16.25% (IQR 13.13 to tic shock are treated with a combination of antibiotics
17.5). In eight patients (66.7%) echocardiographic evi- including a regimen targeting ribosomal protein synthe-
dence of significant right ventricular impairment was sis. In the first 48 hours following ECMO initiation,
found. 6/12 (50%) patients received gentamicin, 7/12 (58.3%)
As part of the routine weaning assessment, aortic received macrolides, 3/12 (25%) received clindamycin
valve time velocity integral (AV VTI) was measured and 9/12 (75%) received linezolid. Additional support-
using continuous wave Doppler.7 The mean AV VTI ive therapies included immunomodulation/toxin scav-
prior to de-cannulation from arterial support was enging with intravenous immunoglobulin in 10/12
17.7cm (IQR 16.7 to 19) and median LVEF was 41.25 (83.3%) patients (median total dose of 2g/kg, range
(IQR 34.38 to 43.75) on ≤ 1l/min arterial flow. All nine 1-3g/kg), hydrocortisone infusion 8mg/hour in 10/12
patients who survived had an echocardiogram showing (83.3%) patients and high dose Pabrinex® (vitamin B1,
their left ventricular systolic function recovered to nor- 2, 3, 6 and C) in 7/12 (58.3%) patients.
mal (LVEF 55-60%). First normal LVEF was docu-
mented after a median of nine days (IQR 4 to 12 days). Complications
All three patients who subsequently died had an improv-
ing LV function, compared to their baseline LVEF, doc- One patient in this cohort developed an ischaemic right
umented prior to death (LVEF 40, 40 and 50%). leg after arterial cannulation. Due to extreme vasocon-
One patient underwent diagnostic coronary angiog- striction the insertion of the right femoral artery back-
raphy (excluding coronary artery disease), and insertion flow cannula had failed. The patient underwent surgical
of an intra-aortic-balloon-pump (IABP) to support left extraction of the ECMO cannula, embolectomy and fas-
ventricular venting post initiation of VAV-ECMO after ciotomy, and the ECMO circuit was simultaneously
echocardiography had shown a dilated left ventricle. reconfigured to a left femoral arterial return. The patient
IABP remained in situ for 5 days. None of the remaining subsequently died when therapy was withdrawn after
patients had other types of mechanical cardiac support, cerebral oedema led to brain herniation. Another
left-ventricular decompression, or diagnostic or thera- patient required transfusion of red cells for pronounced
peutic cardiac interventions whilst on VAV ECMO sup- bleeding from the arterial ECMO cannula site on the
port. One further patient underwent an endomyocardial background of disseminated intravascular coagulopa-
biopsy after decannulation from extracorporeal thy, at a later stage requiring surgical intervention after
support. In this case, cardiac myocytes were viable with- ongoing bleeding post de-cannulation. Two patients
out hypertrophy, disarray, significant inflammation, developed a neurologic deficit secondary to ischaemic
60 Perfusion 33(1S)
Figure 1. Showing mean intravenous noradrenaline dose administered during the first three days on VAV ECMO support. Time
‘zero’ indicates time of cannulation. Bars indicate standard error of the mean.
strokes. Both survived; one with a residual right-sided Prediction and comparison of outcome
weakness and right inferior quadrantonopia, the other
made a full neurological recovery. One patient survived SOFA, APACHE II and SAVE scores as shown in Table 1
with significant disability after development of bilateral suggest a predicted probability of survival of 50%
lower limb ischaemia and a splenic infarct in the initial (median initial SOFA score 10), 50-55% (median
period of sepsis with high vasoconstrictor requirements APACHE II 23.5) and 30% (median SAVE score -5.5). 5,
8, 9 The SAVE score was calculated using available date
and at a later stage underwent bilateral below knee
amputation. within the last six hours prior to cannulation of the arte-
rial limb, regardless whether patients were converted
from VV ECMO or had arterial support initially.
Outcome
An overview of outcome in our cohort is shown in Discussion
Table 2. Patients required VAV ECMO for a median of
96 hours (IQR 71.25 to 126). Three patients (25%) died In this retrospective, observational, case series, patients
after treatment - including VAV ECMO - was with- with severe respiratory failure and septic shock due to
drawn (after two, four and five days, respectively) due septic cardiomyopathy, were managed using hybrid
to futility or devastating prognosis: two with irresolv- VAV ECMO with survival which surpassed that pre-
able multi-organ-failure and one with cerebral oedema dicted using the currently available scoring systems.
leading to brain herniation. Nine (75%) patients sur- These results are similar to those reported by Bréchot
vived to successful de-cannulation from VAV ECMO et al who found a survival rate of 71% in a cohort of 14
and none of these died within the follow-up period patients receiving VA ECMO for profound septic car-
(median six month). diomyopathy. 1 However, other reports have indicated a
Vogel et al. 61
Figure 3. Showing white cell count (WCC) in 109/l, C-reactive protein (CRP) in mg/dl and procalcitonin (PCT) in ng/ml on admission
and on day-3, respectively. Note that on admission PCT was > 100 ng/ml in eight patients and still >100 ng/ml in five patients on day-3.
Also, note the leukopenia on admission in five patients. Left ventricular ejection fraction (LVEF) is shown as box-and-whisker plot for
baseline at VAV ECMO initiation, prior to de-cannulation of the arterial limb and (for survivors) prior to ICU discharge.
Table 2. Showing outcome of all twelve patients with times as mainly staphylococcal, or streptococcal, with two
median days (IQR in brackets). cases of superinfection of Influenza. A causative
Outcomes organism was identified in 11 cases using a mixture
of traditional microbiological tests (culture) and
Time on VAV ECMO (days) 4 (3.0-5.3) advanced molecular diagnostics (16S PCR) on
Total time on ECMO (days) 9 (7.5 to 15.5) specimens obtained by bronchoalveolar lavage.
Time until LVEF normal (days) 9 (4 to 12) Staphylococcal and streptococcal species are known
Time on ventilator (days) 15 (6.5 to 32.5) to produce severe infections, made worse by the pres-
Time on renal-replacement-therapy (days) 5 (2.5 to 22.5) ence of toxins which enhance tissue necrosis and
Length of ICU stay (days) 21 (11 to 38)
cause leucocytosis. 16, 17 The degree of host response
Length of hospital stay (days) 62 (36.5 to 67.5)
was overwhelming with two thirds of cases having a
Predicted hospital survival by SAVE score 30% (30 to 42)
risk class
procalcitonin over 100ng/mL and nearly half the
Discharged alive from ICU 9/12 (75%) cohort having profound leucopenia. These are thought
Discharged alive from hospital 9/12 (75%) to be indicators of both the degree of sepsis and the
Alive at follow-up 9/12 (75%) presence of toxin producing organisms. 18-20
There were two significant complications related to
cannulation. One in a patient in whom a backflow
to recovery of cardiac function is variable and unpredict- cannula could not be placed. This patient was re-can-
able; having the venous return in situ also allows for pre- nulated following return to the hospital by vascular
served cerebral oxygenation. The additional benefit of surgeons who placed a graft on the contralateral fem-
having the venous cannula at the beginning of the ECMO oral artery. The other was secondary to bleeding,
run, is that re-cannulation to a VV configuration does although the patient had a fulminant coagulopathy.
not have to happen in an urgent manner, nor while the Cannulation for peripheral VAV ECMO was per-
patient is anticoagulated or has an ECMO-related coagu- formed at the referring hospital by a consultant inten-
lopathy. sivist-led mobile ECMO service using ultrasound and
The patient cohort was found to have a preponder- fluoroscopic guidance, a strategy previously shown to
ance of community acquired bacterial pneumonia, be safe and effective for the provision of VV ECMO. 6,21
Vogel et al. 63
15. Choi JH, Kim SW, Kim YU, et al. Application of veno- 20. Adamik B, Smiechowicz J, Jakubczyk D, et al. Elevated
arterial-venous extracorporeal membrane oxygenation in Serum PCT in Septic Shock With Endotoxemia Is
differential hypoxia. Multidiscip Respir Med 2014; 9: 55. Associated With a Higher Mortality Rate. Medicine
16. Barrett NA, Armstrong-James D, Edgeworth J, et al.
(Baltimore) 2015; 94: e1085.
Novel H1N1 influenza and Panton-Valentine leukocidin 21. Burns J, Cooper E, Salt G, et al. Retrospective
Staphylococcus aureus necrotizing pneumonia. Br J Hosp Observational Review of Percutaneous Cannulation for
Med (Lond) 2010; 71: 350–351. Extracorporeal Membrane Oxygenation. ASAIO J 2016;
17. Kahr V, Barrett NA, Shankar-Hari M, et al. A life-threat- 62: 325–328.
ening sore throat masquerading as swine flu. Lancet 22. Lockie CJA, Gillon SA, Barrett NA, et al. Severe Respira
(London, England) 2010; 375: 524. tory Failure, Extracorporeal Membrane Oxygenation,
18. Mortensen EM, Coley CM, Singer DE, et al. Causes of and Intracranial Hemorrhage. Critical care medicine
death for patients with community-acquired pneumonia: 2017; 45: 1642–1649.
results from the Pneumonia Patient Outcomes Research 23. Sato R and Nasu M. A review of sepsis-induced cardio-
Team cohort study. Arch Intern Med 2002; 162: 1059– myopathy. J Intensive Care 2015; 3: 48.
1064. 24. Lee SI, Hwang HJ, Lee SY, et al. Veno-veno-arterial extra-
19. Karlsson S, Heikkinen M, Pettila V, et al. Predictive value corporeal membrane oxygenation for acute respiratory
of procalcitonin decrease in patients with severe sepsis: distress syndrome with septic-induced cardiomyopathy
a prospective observational study. Critical care (London, due to severe pulmonary tuberculosis. J Artif Organs
England) 2010; 14: R205. 2017; 20: 359–364.