Glucose measurement in patients with diabetes

mellitus with dermal interstitial fluid

JOHN P. BANTLE and WILLIAM THOMAS

MINNEAPOLIS,MINNESOTA

Although measurement of capillary blood glucose remains the standard method of

self-monitoring for persons with diabetes mellitus, a less-invasive method of moni-
toring would be desirable. Measurement of dermal interstitial fluid glucose might
meet this need. To test this possibility, plasma glucose, capillary blood glucose
(current standard), and dermal interstitial fluid glucose were measured in 17 sub-
jects with type I diabetes during a 5-hour pre- and postprandial period when
plasma glucose was changing rapidly. The objective was to assess the ability of
dermal interstitial fluid glucose to accurately predict plasma glucose over a wide
range of potential glucose concentrations. Dermal interstitial fluid glucose was
highly correlated with plasma glucose (r --- 0.95, p < 0.0001). The mean absolute
and percent differences between dermal interstitial fluid glucose and plasma glu-
cose were 1.2 mmol/L (21 mg/dl) and 10.6%, respectively. The kinetics of dermal
interstitial fluid glucose and plasma glucose were similar. There was no significant
difference between dermal interstitial fluid glucose and plasma glucose in mean
glucose excursion, peak glucose concentration, or time to peak glucose concen-
tration. The correlation between dermal interstitial fluid glucose and plasma glucose
was as strong as the correlation between capillary blood glucose and plasma
glucose. In conclusion, dermal interstitial fluid glucose can be used to estimate
plasma glucose, and has the potential to be used for monitoring patients with
diabetes mellitus (J Lab Clin Med 1997;130:436-4I)

Abbreviations: G C R C = General Clinical Research Center, University of Minnesota

he results of the Diabetes Control and Com- persons with diabetes attempt to achieve prepran-

T plications Trial 1 have led to the recommenda-

tion that most persons with diabetes mellitus
begin treatment programs designed to achieve nor-
mal or near-normal glycemia. 2-4 The American Di-
abetes Association currently recommends that all
From the Department of Medicine, General Clinical Research
Center, and Division of Biostatistics, School of Public Health,
University of Minnesota.
Supported by a grant from Integ Inc. and by General Clinical
Research Center grant MO1-RR-00400 from the National Insti-
tutes of Health.
Submitted for publication Feb. 18, 1997; revision submitted May
28, 1997; accepted June 2, 1997.
Reprint requests: John P. Bantle, MD, Box 504 UMHC, Univer-
sity of Minnesota, Minneapolis, MN 55455.
Copyright © 1997 by Mosby-Year Book, Inc.
0022-2143/97 $5.00 + 0 5/1/83758
dial blood glucose values of 4.4 to 6.7 mmol/L (80 to
120 mg/dl), bedtime blood glucose concentrations of
5.6 to 7.8 mmol/L (100 to 140 mg/dl), and hemoglo-
bin AIC concentrations of less than 7.0%. 3 An es-
sential feature of treatment programs designed to
achieve these goals is a means of monitoring blood
glucose concentration. At present, glucose monitor-
ing requires that a capillary blood sample be ob-
tained by finger prick, placed on a reagent-impreg-
nated strip, and read either visually or with a blood
glucose meter. Intensive treatment of type I diabetes
mellitus often requires self-monitoring of blood
glucose four times daily. 4 The finger pricks neces-
sary to obtain blood samples are mildly painful and
often inconvenient. Moreover, many persons who
frequently monitor their own blood glucose levels
complain that their fingertips become tender

436
J Lab Clin Med
Volume 130, Number 4
and painful because of the frequent punctures. For
all of these reasons, a less-invasive method of mon-
itoring blood glucose would be desirable.
An alternative m e d i u m to blood for monitoring
glucose could be a body tissue or fluid. Subcutane-
ous tissue or fluid might be used in this regard, but
there has been controversy about the accuracy with
which subcutaneous glucose concentrations reflect
blood glucose concentrations. 5 Recently T a m a d a et
al. 6 reported that a glucose flux obtained transder-
mally through the use of a low-level electric current
showed a quantitative relationship to serum glucose.
Thus, dermal interstitial fluid might be a good me-
dium for glucose monitoring.
With this in mind, we used a prototype sampling
device developed by Integ Inc., St. Paul, Minn., to
collect minute quantities of dermal interstitial fluid
to determine the ability of the dermal interstitial
fluid glucose concentration to accurately predict
plasma glucose concentration over a wide range of
potential glucose concentrations. Of particular con-
cern was the predictive capability of the dermal
interstitial fluid glucose concentration at times when
plasma glucose concentration was rapidly rising or
falling. This was, therefore, a feasibility study under-
taken to determine whether dermal interstitial fluid
might be an acceptable alternative to blood for glu-
cose monitoring.
METHODS
Subjects. Twenty subjects with type I diabetes partici-
pated in this study, but data for 3 subjects were excluded
from the analysis for reasons described later. Therefore,
data from 17 subjects were included in this report. These
17 subjects were 8 women and 9 men whose mean age was
41 years (range: 21 to 70 years) and whose mean duration
of diabetes was 20 years (range: 6 months to 48 years).
Nine subjects were known to have retinopathy, 7 to have
neuropathy, and 2 to have nephropathy. One subject had
had a kidney transplant. All subjects were receiving sub-
cutaneous insulin therapy. Seven subjects received insulin
from insulin infusion pumps, 2 subjects took four insulin
injections daily, 4 subjects took three insulin injections
daily, 2 subjects took two insulin injections daily, and 2
subjects took one insulin injection daily.
Study protocol. Each subject was studied on one occa-
sion. Subjects were asked to report to the GCRC before
breakfast on study mornings. Each subject had a venous
catheter placed in a forearm vein from which timed blood
samples could be obtained. Baseline samples of venous
blood and dermal interstitial fluid were obtained 30 min-
utes and 15 minutes before breakfast, respectively. After
the second baseline samples were obtained, each subject
took his or her usual pre-breakfast dose of subcutaneous
insulin. Subjects were allowed to adjust their insulin doses
according to the blood glucose values found in their 15-
Bantle and Thomas 437
minute samples. Fifteen minutes later (at t = 0), each
subject was provided a breakfast with a calorie level
judged to be appropriate for that individual. Eighteen
additional venous blood and dermal interstitial fluid sam-
ples were obtained at 15-minute intervals from 15 to 270
minutes after breakfast. Finger-prick capillary blood sam-
ples were obtained 30 minutes before breakfast and at
every second sampling time thereafter (every 30 minutes).
GCRC nurses collected all samples. One subject, in whom
symptomatic hypoglycemia developed during the study,
was treated with orally administered carbohydrate. After
the last samples were obtained, the intravenous catheters
were removed and the studies ended. Subjects were pro-
vided lunch before they left the GCRC.
Dermal interstitial fluid was collected with a sterile,
disposable, 29-gauge hypodermic cannula attached to a
depth-limiting spacer. A 0.5 mm 3 capillary tube (Drum-
mond Scientific Company, Broomall, Pa.) was attached to
the blunt end of each cannula with heat shrink tubing.
This assembly was loaded into a prototype sampling de-
vice with a window for observing the capillary tube. The
sampling device was placed on the skin of a subject's
forearm and steady pressure was applied. The spacer
limited the penetration of the cannula to a maximum
depth of 1.4 mm. As the cannula penetrated the dermis,
the pressure created allowed interstitial fluid to flow into
the capillary tube. The study nurse visually monitored the
collection of interstitial fluid through the window in the
sampling device. It usually took between 5 and 15 seconds
to collect a sample. After the sample was collected, the
cannula assembly was removed from the sampling de,rice
and again visually inspected by the study nurse. A sample
was deemed acceptable if visual inspection showed that
the capillary tube was filled and contained no air bubbles
or visible blood. A new cannula assembly was used for
collection of each sample.
The study nurse either could not obtain or deemed
unacceptable a total of 14 interstitial fluid samples. All
acceptable samples were numbered and immediately
stored at -70 ° C.
The protocol for the study was approved by the Uni-
versity of Minnesota Committee on the Use of Human
Subjects in Research. Written informed consent was .ob-
tained from all subjects.
Laboratory methods. All dermal interstitial fluid sam-
pies were transported from the GCRC to the Integ Lab-
oratory on dry ice and were stored at -70 ° C until analysis.
The samples were identified to Integ personnel only by
number (i.e., all Integ personnel were masked with regard
to the plasma and capillary glucose values of the intersti-
tial fluid samples). Before glucose analysis, each capillary
tube was microscopically reexamined to confirm that the
samples met the acceptance criteria. Unacceptable sam-
ples were excluded from glucose analysis. On microscopic
analysis, the dermal interstitial fluid samples from three
consecutive subjects were found to contain a contaminant
that was of indeterminate origin but was thought to orig-
inate in the capillary tubes. Because the nature of the
 J Lab Clin Med
438 Bantle and Thomas October 1997

%o
O

o ooo
oj o

E 0 o o~ o 0
E o0O o ~
v OO

8 o °°'°,~ o 8

o o°~e°~o °Oo o
72
-.,i
o
o o--~ Oo o%
Ii O0 ~ 0
°

e-

t~

m
tO --

O --
Y0
I I
5
I
10
I
15
I
20
I
25

Plasma Glucose (mmol/L)

Fig. 1. Plasma glucose versus dermal interstitial fluid glucose concentrations for all observations of all
subjects (n = 223); the line shown is the line of identity (x = y).

T a b l e I. Comparisons b e t w e e n plasma glucose T a b l e II. Comparisons b e t w e e n plasma glucose

a n d dermal interstitial fluid glucose a n d b e t w e e n a n d dermal interstitial fluid glucose in 16 subjects*
plasma glucose a n d capillary b l o o d glucose Dermal
Plasma glucose Plasma glucose interstitial p
versus dermal versus capillary Plasma fluid Value
glucose glucose
Glucose excursion in 8,9 + 2.3 9.0 +- 3.1 0.86
Number of samplescompared 223 154 mmol/L (mg/dl) (161 +42) (163-+55)
Slope of regression line 1.010 0,865 Peak glucose concentration 16.2-+4.2 16.8_+4.2 0.07
Correlation coefficient 0.95 O.87 in mmol/L (mg/dl) (292 _+ 76) (303 -+ 76)
Mean absolute difference 1.2 mmol/L 1.4 mmol/L Time to peak glucose 132 _+ 44 143 + 51 0.15
(21 mg/dl) (26 mg/dl) concentration (min.)
Mean percent difference 10.6% 12,1%
*Allvaluesare mean_+SD.
Percent within 10% of 56% 55%
plasma glucose value
Percent within 20% of 88% 89% Biochemistry Laboratory at the University of Minnesota
plasma glucose value
Hospital, with either a 2300 Stat Plus (Yellow Springs
Instruments, Yellow Springs, Ohio) or an Ektachem 700
XR Analyzer (Eastman-Kodak, Rochester, N.Y.). Inter-
contaminant could not be conclusively determined, a de- assay coefficients of variation for Chemtrak Level 1 (3.5
cision was made to exclude all samples from these three mmogL [63 mg/dl]) control solution (Medical Analysis
subjects. This decision was made before glucose analysis. Systems Inc., Camarillo, Calif.) were 2.8% and 1.5%,
Plasma glucose concentrations were dtermined in the respectively, and for Chemtrak Level 3 (15.3 mmol/L
J Lab Clin Med
Volume 130, Number 4 Bantle and Thomas 439

C~l

O
O

o
0 _
..1 o

0
E rO O

E
v

0 o o
0 °o
o o
"(3
0
oo :°°
m 08 / ~ O o
o~O~ ~ 60 o
i 'o°~/~o o
o - o

o
o o
o o

LO o
o
o

0
Y
I I I 1 I I
0 5 10 15 20 25

Plasma Glucose (mmol/L)

Fig. 2. Plasma glucose versus capilla~r blood glucose concentrations for all observations of all subjects
(n = 154); the line shown is the line of identity (x = y).

[275 mg/dl]) control solution were 1.5% and 1.4%, respec-
tively. Capillary blood glucose concentrations were mea-
sured with the Yellow Springs Instruments 2300 Stat Plus.
Dermal interstitial fluid glucose concentrations were
measured in the Integ Inc. laboratory with a manual
hexokinase method and an unblanked hexokinase re-
agent and standards (Sigma Diagnostics, St Louis,
Mo.) according to the manufacturer's instructions. A
proportional adjustment was made in the measurement
method to compensate for the 0.5 ~1 volume of inter-
stitial fluid samples, which was smaller than the 10 txl
volume of serum or plasma samples typically analyzed
with this method. The dermal interstitial fluid samples
were expressed into microcentrifuge tubes containing
hexokinase reagent. The capillary tubes were rinsed
three times with the hexokinase reagent. Each sample
was read with a spectrophotometer (Genesys 5; Spec-
tronic, Milton Roy, Rochester, N.Y.) at 340 nm. For the
manual hexokinase assay, the interassay coefficient of
variation for Chemtrak Level 1 control solution was
8.0% and for Chemtrak Level 3 control solution was
5.3%.
The Chemtrak Level 1 and Level 3 control solutions
were used as intermethod controls. The dermal interstitial
fluid glucose values were corrected linearly so that the
values for the Chemtrak Level 1 and the Chemtrak Level
3 control solutions obtained at the Biochemistry Labora-
tory of the University of Minnesota agreed with those
obtained at the Integ laboratory.
Statistical analysis. Analysis of the relationship between
dermal interstitial fluid and plasma glucose values was
done so as to take into account the dependence among
repeated measures of glucose for each subject. This was
done by multiple regression of the dermal interstitial fluid
glucose values (dependent variable) on the plasma glu-
cose values (independent variable) as parallel lines with a
common slope and separate intercepts for each subject.
There was no evidence of different slopes among subjects.
A similar analysis was done for the capillary and plasma
glucose values. The slopes of these regression lines are
given in Table I and are adjusted for the repeated mea-
sures. For comparison with the literature, we also pro-
vided correlation coefficients, mean absolute differences,
and mean percent differences, although all of these ig-
 J Lab Clin Med
440 Bantle and Thomas October 1997

Subject 5 Subject 10

tO

...1
0
tO
E
V
E Q

t- 0
i i i 4 t u n i
O 0 50 100 150 200 250
0 50 100 150 200 250

c--
Subject 12 Subject 13
0
c-
O
0

/:...
tO tO
o

0 0
C

LO tO

0 0
i i p i i

0 50 100 150 200 250 0 50 100 150 200 250

Minutes from Breakfast (Time = O)

Fig. 3. Glucose determinations over 300 minutes for the four subjects with the greatest excursions in
glucose values. Plasma glucose values are represented by solid lines, dermal interstitial fluid glucose values
by solid circles, and capillary blood glucose values by open circles.

nored the dependence arising from the repeated measure-
ments. For Table II, glucose excursions were calculated as
the range from minimum to maximum values. Time to
peak glucose concentration was estimated for each subject
as the length of time needed for the glucose concentration
to reach its largest value. One subject withdrew from the
study early, and the data for this subject were excluded
from Table II, in which the data were based on the
remaining 16 subjects. Paired t tests were used to compare
end points and as the basis for calculations to determine
the power of the study to find differences in glucose
excursion, peak glucose concentration, and time to peak
glucose concentration.
RESULTS
The observed values of dermal interstitial fluid
glucose and plasma glucose concentrations for all
subjects are compared graphically in Fig. 1 and nu-
merically in Table I. Dermal interstitial fluid glucose
values were highly correlated (r = 0.95,p < 0.0001)
with plasma glucose values across the range of ob-
served plasma glucose values (2.1 to 23.4 mmol/L
[37 to 422 mg/dl]). Comparisons of capillary blood
glucose with plasma glucose values are provided in
Fig. 2 and Table I. Capillary blood glucose values
were not better correlated with plasma glucose val-
ues than were dermal interstitial fluid glucose values,
and both the mean absolute difference and mean per-
cent difference between capillary blood glucose and
plasma glucose values were somewhat greater than the
corresponding differences between dermal interstitial
fluid glucose and plasma glucose values.
Plasma glucose, dermal interstitial fluid glucose,
and capillary blood glucose values from the four
individual subjects with the largest excursions in
plasma glucose are presented in Fig. 3. In all four
subjects, the three glucose curves were similar. Data
on glucose excursion, peak glucose concentration,
and time to peak glucose concentration for 16 sub-
jects are provided in Table II. There were no signif-
icant differences between plasma and dermal inter-
stitial fluid for any of these parameters, although the
difference for peak glucose concentration ap-
J Lab Clin Med
Volume 130, Number 4
proached significance (p = 0.07). To put these re-
sults in perspective, the sizes of the differences the
study had power to find were calculated. With 16
subjects, there was an 80% chance (power: r = 0.8)
of finding a significant difference between dermal
interstitial fluid glucose and plasma glucose values if
the true differences were larger than 1.3 mmol/L (24
mg/dl) in glucose excursion, 0.9 mmol/L (16 mg/dl)
in peak glucose concentration, or 20 minutes: in time
to peak glucose concentration.
DISCUSSION
Our data demonstrated that dermal interstitial
fluid glucose concentration was highly correlated
with plasma glucose concentration in a population
of subjects with type I diabetes, many of whom had
rapidly rising and then falling postprandial plasma
glucose concentrations. Inspection of data t0r indi-
vidual subjects, as shown in Fig. 3, suggested that the
dermal interstitial fluid glucose concentration rose
and fell in nearly synchronous fashion with the
plasma glucose concentration. The correlation be-
tween dermal interstitial fluid glucose and plasma
glucose values matched or exceeded many of the
correlations between capillary blood glucose and
reference methods establishing capillary blood glu-
cose testing in the literature. 7-a4 In our study, dermal
interstitial fluid glucose values also matched those
for capillary blood glucose in all comparisons with
plasma glucose concentrations.
Although these results suggest that analyzing der-
mal interstitial fluid holds promise as a less-invasive
method for measuring glucose concentrations, a num-
ber of questions remain. Further research should ad-
dress the correlation between dermal interstitial fluid
glucose and plasma glucose values at glucose concen-
trations below 5 mmol/L. Although our study was not
designed for this purpose, accurate measurement of
glucose concentrations in the low range is of critical
significance to persons with diabetes mellitus.
In addition, studies with a fully integrated glucose
monitoring device for home use will be necessary to
determine the accuracy and reproducibility with
which it can measure dermal interstitial fluid glucose
concentrations. It will also be necessary to determine
whether the analytic precision of the procedure will be
adequate when it is performed by persons with diabe-
tes. Moreover, it remains to be determined whether
repeated sampling of dermal interstitial fluid udll have
greater acceptance within the diabetic commtmity than
does capillary blood sampling.
Bantle and Thomas 441
Our preliminary data suggest that measurement
of dermal interstitial fluid glucose concentrations
can be used to estimate plasma glucose concentra-
tions. If the performance issues raised in this study
can be resolved, and the collection techniques can
be performed satisfactorily by patients, this less-
invasive method of glucose monitoring could pro-
vide a new option for patients with diabetes mellitus.
Conflict-of-interest note: William Thomas is a stockholder in
Integ Inc.
REFERENCES
1. The Diabetes Control and Complications Trial Research
Group. The effect of intensive treatment of diabetes on the
development and progression of long-term complications in
insulin-dependent diabetes mellitus. N Engl J Med 1993;329:
977-86.
2. Lasker RD. The Diabetes Control and Complications Trial:
Implications for policy and practice. N Engl J Med 1993;329:
1035-6.
3. Eastman RC, Siebert CW, Harris M, Gorden P. Implications
of the Diabetes Control and Complications Trial. J Clin
Endocrinol Metab 1993;77:1105-7.
4. American Diabetes Association. Standards of medical care
for patients with diabetes mellitus. Diabetes Care 1994-;17:
616-23.
5. Schmidt F J, Slutter W J, Sehoonen AJM, Glucose concentra-
tion in subcutaneous extracellular space. Diabetes Care 1993;
16:695-700.
6. Tamada JA, Bohannon NJV, Potts RO. Measurement of
glucose in diabetic subjects using noninvasive transdeimal
extraction. Nature Med 1995;1:1198-1201.
7. Brunton WAT, Steel JM, Percy-Robb IW. An assessment of
a reflectance meter system for measurement of plasma or
blood glucose in the clinic or side ward. Clin Chim Acta
1977;75:359-64.
8. Wafford S, Gale EAM, Allison SP, Tattersall RB. Self-mon-
itoring of blood glucose. Lancet 1978;1:732-5.
9. Peterson CM, Forhan SE, Jones RL. Self-management: an
approach to patients with insulin-dependent diabetes melli-
tus. Diabetes Care 1980;3:82-7.
10. Sonksen PH, Judd S, Lowy C. Home monitoring of blood
glucose: new approach to management of insulin-dependent
diabetic patients in Great Britian. Diabetes Care 1980;3:100-7.
11. Reeves ML, Forhan SE, Skyler JS, Peterson CM. Compari-
son of methods for blood glucose monitoring. Diabetes Care
1981;4:404-6.
12. Clements RS, Keane NA, Kirk KA, Boshell BR. Comparison
of various methods for rapid glucose estimation. Diabetes
Care 1981; 4:392-395.
13. Shapiro B, Savage PJ, Lomatch D, et al. A comparison of
accuracy and estimated cost of methods for home blood
glucose monitoring. Diabetes Care 1981;4:396-403.
14. Nelson, JD, Woelk MA, Sheps S. Self glucose monitoring: a
comparison of the Glucometer, Glucoscan and Hypocount B.
Diabetes Care 1983;6:262-267.