Clinical Science (1993) 84, 681-685 (Printed in Great Britain)
681
Estimation of the hepatic extraction ratio of lndocyanine
Green in swine
David F. KISOR', Reginald F. FRYE' and Kenneth A. KUDSKl
'Department of Pharmacy and Therapeutics, School of Pharmacy, University of Pittsburgh,
Pittsburgh, Pennsylvania, U.S.A., and 2Department of Surgery, College of Medicine, University
of Tennessee, Memphis. Tennessee, U.S.A.
(Received 16 November 1992/9 February 199); accepted 19 February 1993)
1. The hepatic extraction ratio of Indocyanine Green
was measured directly by trans-hepatic catheterization in 14 outbred swine (eight well-fed, six malnourished). A specific two-compartment pharmacokinetic model was fitted to the arterial Indocyanine
Green concentration-time data and used to estimate
the hepatic extraction ratio of Indocyanine Green.
2. The specific two-compartment pharmacokinetic
model was modified to represent more accurately the
physiological process of Indocyanine Green removal.
Simulations were performed using this new model to
estimate the hepatic extraction ratio of Indocyanine
Green in the swine.
3. Similarly to previous findings, our data showed
that the original model consistently overestimated the
hepatic extraction ratio of Indocyanine Green (i.e. the
model estimate was compared with the true, directly
measured value).
4. The comparison of the modified model and the
original model clearly indicates the reason for the
overprediction of the hepatic extraction ratio of
Indocyanine Green by the latter. The simulations
using the new model indicate that the percentage of
binding of Indocyanine Green to its transport protein
(glutathione Stransferase) for removal in the bile will
affect the estimation of the hepatic extraction ratio
of Indocyanine Green. Thus, the amount of Indocyanine Green available for removal is less than that
assumed by the original model.
blood flow (QH) [6-91. This relationship has led to a
minimally invasive method for estimating Q H
[lo, 113, where the blood clearance of ICG is taken
as an estimate of 'effective' hepatic blood flow (i.e. it
is assumed that the extraction ratio of ICG approaches unity). The EH of ICG is lower in patients
with cirrhosis [12] and in some animal species
(swine [13, 141, cats [15]). In these cases, the
clearance of ICG is more dependent on overall
hepatic function (biliary excretion) rather than QH.
The direct measurement of the EH of ICG
requires the placement of an hepatic venous catheter. This is not practical in healthy, normal subjects
or most patient investigations and may result in the
death of laboratory animals (after removal of the
catheter).
An indirect method for estimating the EH of ICG
without hepatic venous catheterization was reported
by Grainger et al. [16]. This approach is based on a
specific two-compartment pharmacokinetic model
[17] describing the disposition of ICG after an
intravenous bolus dose.
The present report describes the comparison of
the direct determination of the EH-of ICG with the
indirect method. A modification of the model presented by Grainger et al. [16] is described. Simulations using the amended model are presented to
indicate the influence of transport protein (GST)
binding on the estimation of the E H of ICG.
MATERIALS A N D METHODS
INTROD UCTlO N
Indocyanine Green (ICG) is an anionic dye frequently used to estimate hepatic function and
hepatic blood flow in animals and man [l-31. The
dye is highly bound to albumin and a-lipoproteins
[4] and is eliminated unchanged in the bile via
transport by glutathione S-transferase (GST) [S].
The hepatic extraction ratio (EH) of ICG in
healthy normal man is about 0.62-0.88, indicating
that its clearance is primarily dependent on hepatic
Animals
All experiments were approved by the Ohio State
University Laboratory Animal Review Committee.
Fourteen outbred swine (nine male, five female)
weighing 19-28 kg were studied. These animals were
being studied in a protocol designed to examine the
effect of nutritional status on organic anion clearance [14]. On each study day the animal was
anaesthetized with ketamine and pentobarbital,
intubated and ventilated. Each animal had the
Key words: extraction ratio, lndocyanine Green, model.
Abbreviations: E,, hepatic extraction ratio; E,, model-estimated hepatic extraction ratio; EN, modified-model-simulated hepatic extraction ratio; GST, glutathione Stransferase;
ICG, lndocyanine Green: QH. hepatic blood flow.
Correspondence: Dr David F. Kisor, Department of Pharmacy and Therapeutics, School of Pharmacy, University of Pittsburgh, 904 Salk Hall, Pittsburgh, PA 15261. U.S.A.
D. F. Kisot et al.
682
following catheters placed: a Swan-Ganz catheter
was inserted through a right femoral cut-down into
a pulmonary artery for cardiac output determinations; another catheter was placed into the contralateral femoral artery to allow arterial sampling;
under fluoroscopy, a catheter was placed (5-7 cm)
into a major right hepatic vein. Throughout each
study fluoroscopy was used to confirm the placement of the hepatic vein catheter. Twenty millilitres
of blood were collected for preparation of standard
curves. Each animal received a rapid (10-20 s) intravenous bolus of ICG (5mg/kg). From each catheter,
blood samples were obtained at 5, 10, 15, 30, 45, 60,
75, 90, 105, 120, 135, 150, 165 and 180min after the
administration of the ICG. Blood samples were
transferred to heparin-containing Venoject tubes
and immediately centrifuged. Plasma was separated
and stored at -20°C until analysed (within 72 h).
I
k=
Fig. 1. Open two-compartment model fit to arterial plasma ICG
concentration-time data
Analytical procedure
A modification of the h.p.1.c. analytical procedure
of Rappaport and Theissen [18] was used and was
described previously [141. The intra-day and interday coefficients of variation were both 5.6%.
Fig. 2. Modified open two-compartment model (schematic) used for
simulations
The model estimate of the EH of ICG (EM)was
calculated in the following manner [16]:
Data analysis
Plasma ICG concentration versus time data for
each animal were analysed by log-linear regression
for estimation of the terminal slope and elimination
rate constant (KJ. The terminal slope was used to
extrapolate for total area under the plasma ICG
Concentration versus time curve from the last measured time point to infinity. The areas under the
arterial and hepatic vein ICG concentration versus
time curves were calculated using the linear trapezoidal rule. The EH at steady-state was calculated in
the manner described by Shand et al. [19]; the
hepatic extraction ratio (EH) is calculated as
(AUC,,,-AUC,,)/AUC,,,, where AUC,,, is the area
under the arterial plasma ICG concentration versus
time curve and AUCh, is the area under the hepatic
vein plasma ICG concentration versus time curve.
The method of residuals [20] was used to obtain
estimates of the disposition rate constants (aand jl),
and the corresponding zero-time intercepts ( A and
B) from the arterial plasma ICG concentration-time
curve. The two-compartment open model described
by Rowland et al. [I71 (Fig. 1) with elimination
from the second compartment was fitted to the
arterial data using PCNONLIN [21]. The
PCNONLIN program was written to generate estimates of the ‘micro’ rate constants k , , , kzo and k z l
as follows:
+
+
k 1 2= ( d jlB)/(A B)
(1)
where EM is the model-estimated EH of ICG, and X ,
is the amount of ICG in V, (Fig. 1). A modification
was made to the model in that X , , the amount of
ICG in the liver, can be described as the amount
available for efflux ( X , ) plus the amount available
for removal ( X r ) . Thus, X , = X , + X , . The modified
model was used in simulations in the following
manner:
where EN is the modified-model-simulated EH of
ICG, and X , , X , , K , , and kzo are described as
above. The values for X , and X , were altered from
0 to 1 in inverse proportion as to simulate changes
in the availability of ICG for emux or removal. The
modified model is presented in Fig. 2.
Statistical analysis
The degree of association between the predicted
(model estimates, EM and EN) and the true (directly
measured) EH of ICG was examined by calculating
the correlation coefficients.
RESULTS
The arterial plasma ICG concentration versus
time data fell in a biexponential fashion re-
Extraction ratio of lndocyanine Green
683
Table 1. Direct and modelderived hepatic extraction ratios of ICG
in swine. Abbreviation: NM. not mwurable.
70
Animal no.
Em
EH
(measured)
(model)
~
10
-
A
A
A
A
A
A
A
A
A
&
Time (min)
Fig. 3. Mean arterial plasma ICG concentration-time curve for all of
the animals
.-0
I
2
3
4
5
6
7
8
9
10
II
I2
13
14
0.3 I2
0.267
0.272
0.221
0. I32
0.192
0.266
0.210
NM*
0.039
NM
0.07 I
0.121
0.099
0.086
0. I27
0.022
0.254
NM
NM
0. I82
Mean
SD
0.094
0.026
0.25 I
0.306
0.279
0.307
0.314
0.087t
0.049
0.256$
0.052
t n = 10.
Sn=
I
14.
Directly measured extraction ratio (EH)
Fig. 4. Correlation between original ( 0 )and new ( 0 )model estimates of the hepatic extraction ratio of ICG and the ‘true’ directly
measured value, EH. -=unity.
presentative of a two-compartment model (Fig. 3).
There was a poor correlation between the directly
measured EH of ICG and the model-derived EM of
ICG in the ten animals where both values were
available (0.087f0.049 versus 0.256f0.052; r =0.2).
The mean value of the EH as compared with the EN
for the ten animals was 0.087 0.049 versus
0.080+0.019 when X,:Xe=0.2:0.8. The r value
improved slightly (0.31) and the points were closer
to the line of unity (Fig. 4). The directly measured
EH of ICG was less than 0.2 in all the animals. The
original model consistently overpredicted the true
extraction ratio (Table 1).
In the simulation study alterations of X, indicated
that in all situations where X, is less than SO%, the
EN is less than that estimated by the model of
Grainger et al. [I61 (Table 2). In the case where
X,=0.5 and Xe=0.5, the modified model reduces to
the model of Grainger et al. [16] (EN=E,=0.256)
C161.
DISCUSSION
The pharmacokinetics of ICG appeared linear in
all animals at the dose of 5mg/kg. Linearity at this
dose has been reported in other species [22]. We
observed linearity at doses of 0.2, 1.0 and 5.0mg/kg.
However, non-linearity was observed at a dose of
23 mg/kg (Scott et al., unpublished work). The use
of the 5mg/kg dose with the h.p.1.c. assay allowed
Table 2. Simulated extraction ratios (EN) of ICG
Animal no. km
kx
EN
X, ... 0.100
X, ... 0.900
I
2
3
4
5
6
7
8
9
10
II
I2
13
14
0.02063
0.0 I074
0.00192
0.00721
0.0 I9 I3
0.00492
0.00709
0.00479
0.0044 I
0.02095
0.04183
0.01353
0.01778
0.02035
0.04551
0.02947
0.00514
0.02544
0.12553
0.02069
0.01955
0.01800
0.01293
0.06256
0.09502
0.03502
0.04007
0.04455
0.200
0.800
0.300
0.700
0.400 0.500
0.600 0.500
0.048
0.102
0.163
0.039
0.040
0.031
0.017
0.026
0.039
0.029
0.036
0.036
0.047
0.041
0.047
0.084 0.135
0.312
0.267
0.272
0.221
0.132
0.192
0.266
0.210
0.254
0.251
0.306
0.279
0.307
0.314
0.256
0.052
0.048
Mean 0.037
SD 0.009
0.100
0.103
0.160
0.164
0.232
0.196
0.199
0.159
0.092
0.137
0.195
0.151
0.185
0.182
0.227
0.205
0.228
0.233
0.080
0.019
0.129
0.030
0.187
0.041
0.085
0.066
0.037
0.056
0.083
0.062
0.079
0.077
0.099
0.088
0.138
0.108
0.061
0.092
0.135
0.102
0.127
0.125
0.159
0.142
the characterization of the biexponential decline in
plasma ICG concentration.
The biexponential decline in plasma ICG concentration suggests that a two-compartment model
may best be fitted to the data. Of the three possible
two-compartment open models, the model with elimination from the second compartment was chosen.
This model best describes the disposition of ICG
which includes hepatic uptake (from V, to V,) and
biliary excretion (elimination from V,) (Fig. 1). This
model was used to estimate the EH of ICG as
offered by Grainger et al. [16j.
The measured EH of ICG was low in all animals
684
D. F. Kisor et al.
(range: not measurable to 0.18). In some cases the
EH was not measurable because there was no
observed difference between AUC,,, and AUC,,,.
This suggested that in these animals the extraction
ratio of ICG was very low. However, the measurement of the EH in these animals was obscured by
acceptable analytical variability. The low EH of ICG
in the swine indicates the dependence of ICG
clearance on hepatic function rather than hepatic
blood flow ( Q H ) . Hepatic blood flow was not calculated because of the very low measured EH. Small
differences (e.g. acceptable assay variability) in the
measurement of EH would result in large calculated
changes in QH.
The indirect method of estimating EH consistently
overpredicted the true value. Questions of intrahepatic shunting and model mis-specification (i.e.
two-compartment versus three-compartment) have
been cited as possible reasons for the failure of the
model [22, 231. For instance, Burns et al. [24]
found that the disposition of ICG was best described by a triexponential model. With respect to
the use of the specific two-compartment model to
estimate the EH of ICG, the following is presented:
the amount of dye in V, 'seen' for emux may not be
the same as the amount 'seen' for elimination.
Specifically, the amount of ICG ( X , ) in V2 is the
sum of the amount available for efflux ( X , ) and the
amount available for elimination ( X , ) . The relationship for the determination of EH [elimination
rate/(efflux +elimination rate)] would be written as
X , k 2 0 / ( X , k 2 ,+X,k,,), and it is clear that the estimated EH would be less than that using the relationship (eqn. 4) as described by Grainger et al.
[16]. The difference between X 2 and X , explains
the overprediction of the EH. Only in the case
where
X,=X,
can
the
expression
E=
X 2 k 2 0 / ( X 2 kI2+ X,k,,) be reduced to k 2 , / ( k 2 , + k20).
Physiologically, the active process of organic anion
uptake into the hepatocyte, binding to the cytosolic
transport protein (ligandin; GST) and biliary excretion would result in elimination. Drug not bound in
the cytosol would not be removed and would be
available for efflux 1243. While it is possible that
50% of ICG is bound to cytosolic GST, it is
unlikely that this relationship would hold under
conditions of physiological stress. Our simulations
suggest that 21-220/, of ICG is bound to GST and
available for removal, which results in an EN of
approximately 0.087 (i.e. the same value as determined directly). It has been shown that increases or
decreases in GST correlate with the disappearance
rates of bilirubin and bromosulphophthalein
[25-271. Our simulations would predict a similar
relationship for ICG.
Our data agree with earlier reports [22, 231
showing the inadequacies of the indirect method
with consistent overprediction of EH and do not
support the use of the model. Our simulations serve
only to offer, in theory, a reason for the consistent
overestimation of the EH of ICG by the original
model. As X, and X , cannot be measured in our
modification of the original two-compartment
model, further work is needed to define more clearly
the appropriate model for the determination of the
EH of ICG without hepatic venous catheterization.
ACKNOWLEDGMENTS
We are grateful for the input of Dr Ronald A.
Scott, and we acknowledge the skillful secretarial
assistance of Ms Helen Jarosz.
REFERENCES
I. Hunton DB, Bollman JL, Hoffman HN. Studies of hepatic function with
indocyanine green. Gastroenterology 1960; 3 9 317-24.
2. Branch RA. James ]A. Read AE. The clearance of antipyrine and indocyanine
green in normal subjects and in patients with chronic liver disease. Clin
Pharmacol Ther 1976; 2 0 81-9.
3. Kisor DF. Wooding-Scott RA. Effect of theophylline on estimated hepatic
blood flow. Pharmacotherapy 1990; 10 406-9.
4. Baker KJ, Bradley SE. Binding of bromosulfophthalein (BSP) and indocyanine
green (ICG) by plasma alpha-I-lipoproteins. Proc Soc Exp Biol Med 1966; 122:
957-63.
5. Paumgartner G. The handling of indocyanine green by the liver. Schweir Med
Wochenschr 1975; 105 1-30.
6. Leevy CM, Mendenhall CL, Lesko W, Howard MM. Estimation of hepatic
blood flow with indocyanine green. J Clin Invest 1962; 41: 1169-79.
7. Weigand BD, Ketterer SG, Rapapport E. The use of indocyanine green for
the evaluations of hepatic function and blood flow in man. Am J Dig Dis 19M);
5: 427-36.
8. Reemtsma K. Hattinger GC, DeGraff AC. The estimation of hepatic blood
flow using indocyanine green. Surg Gynecol Obstet 1960; 1 1 0 353.
9. Caesar J, Shaldon S, Chandussi L, Guevara L, Sherlock S. The use of
indocyanine green in the measurement of hepatic blood flow and as a test of
hepatic function. Clin Sci 1961; 21: 43-57.
10. Quatury RR, Khan MB, Nallathambi M. Davis K. Staht WM. Cimetidine and
hepatic blood flow in polytrauma patients. Crit Care Med 1985; 13: 436-7.
I I. Bauer LA, Stenwall M, Horn JR. Davis R, Opheim K, Greene L. Changes in
antipyrine and indocyanine green kinetics during nifedipine, verapamil. and
diltiazem. Clin Pharmacol Ther 1986; 40: 239-42.
12. Navasa M, Bosch J, Mastai R. et al. Noninvasive measurement of hepatic
extraction, intrinsic clearance and hepatic blood flow from two-compartment
analysis of the plasma disappearance curve of indocyanine green in patients
with cirrhosis. Hepatology 1986; 3: S29.
13. Paschen U, Muller MI. Liver blood flow measured by indocyanine green in
conscious unrestrained miniature pigs. Res Exp Med 1987; 187: 71-9.
14. Kudsk KA, Kisor DF. Waters B, Mirtallo JM, Campbell AJ, Wooding-Scott RA.
Effects of nutritional status upon organic anion clearance by the swine liver.
Surgery 1992; 111: 188-94.
15. Krarup, K. Larson ].A. The influence of dye infusion rate and hepatic
plasma flow on indocyanine green clearance. Scand J Lab Invest 1976; 3 6
1834.
16. Grainger SL. Keeling PWN, Brown IMH, Marigold JH, Thompson RPH.
Clearance and noninvasive determination of the hepatic extraction of
lndocyanine Green in baboons and man. Clin Sci 1983; 64: 207-12.
17. Rowland M. Benet L. Graham GG. Clearance concepts in pharmacoketics.
J Pharmacokinet Biopharmacol 1973; I: 123-36.
18. Rappaport PL. Theissen JJ. High-pressure liquid chromatographic analysis of
indocyanine green. J Pharmacol Sci 1982; 71: 157-62.
19. Shand DG, Kornhauser DM, Wilkinson GR. Effects of route of administration
and blood flow on hepatic drug elimination. J Pharmacol Exp Ther 1975; 195
42432.
20. Gibaldi M, Perrier D. Method of residuals. In: Pharmacokinetics. New York
Marcel Dekker Inc., 1982; 433-44.
21. Metzler CM. Weiner DL. PCNONLIN. Statistical Consultants Inc., Gainesville,
FL, 1989.
Extraction ratio of lndocyanine Green
22. Burns E. Ball CE, Christie JP. Broadhead GD, Tucker GT, Bax ND. Direct
and indirect measurement of the hepatic extraction ratio of lndocyanine
Green in the rat. Clin Sci 1989; 76: 503-8.
23. Clements D. West R. Elias E. Comparison of bolus and infusion methods for
estimating hepatic blood flow in patients with liver disease using indocyanine
green. J Hepatol 1987; 5 282-7.
24. Burns E, Triger DR, Tucker GT, Bax NDS. lndocyanine Green elimination in
patients with liver disease and in normal subjects. Clin Sci 1991; 80: 155-60.
685
25. Klaassen CD, Watkins JB. Mechanisms of bile formation, hepatic uptake, and
biliary excretion. Pharmacol Rev 1984; 36: 1-67.
26. WolkoH AW, Chung CT. Identification, purification and partial
characterization of an organic anion binding protein from rat liver cell
membrane. J Clin Invest 1980; 65 115261.
27. Stein LB, Mishkin S, Fleischner G, et al. Effect of fasting on hepatic ligandin.
Z protein, and organic anion transfer from plasma in rats. Am J Physiol 1976;
231: 1371-6.