55, 256 –265 (2000)
Copyright © 2000 by the Society of Toxicology
TOXICOLOGICAL SCIENCES
Application of a Physiologically Based Pharmacokinetic Model to
Estimate the Bioavailability of Ethanol in Male Rats: Distinction
between Gastric and Hepatic Pathways of Metabolic Clearance
Gina M. Pastino* ,1 and Rory B. Conolly†
*Schering Plough Research Institute, P.O. Box 32, Lafayette, New Jersey 07848; and †Chemical Industry Institute of Toxicology,
Research Triangle Park, North Carolina 27709
Received September 29, 1999; accepted January 24, 2000
A portion of ingested ethanol does not reach the systemic circulation in both rats and humans as indicated by higher blood
ethanol concentrations following an intravenous administration
compared to an equivalent oral administration. The mechanism
for this decrease in the oral bioavailability is not yet completely
understood. Metabolism by gastric or hepatic alcohol dehydrogenase (ADH), or both, has been implicated. However, the extent to
which each pathway of elimination contributes to the first-pass
clearance is not known. The purpose of this study was to utilize a
physiologically based pharmacokinetic (PBPK) model for ethanol
to estimate the relative contributions of hepatic and gastric metabolic clearance to the oral bioavailability of ethanol in male rats.
In the current model, calculations of hepatic-first pass metabolic
clearance accounted for the competition for metabolism between
incoming ethanol from the GI tract and recirculating ethanol. This
differs from previous methods that quantified the effect of ethanol
entering the liver from the GI tract on the overall rate of metabolism of ethanol by the liver. These models did not specifically
describe the effect of recirculating ethanol on the first-pass metabolism of ethanol, and vice versa. The dependence of bioavailability on dose and absorption rate was also investigated. The use
of a PBPK model for ethanol in rats allows a more detailed
examination of physiological and biochemical factors affecting the
bioavailability of ethanol than has previously been possible. The
analysis indicates that both gastric and hepatic first-pass metabolism of ethanol contribute to ethanol bioavailability in male rats.
Key Words: blood ethanol concentrations; gastric and hepatic
alcohol dehydrogenase; ethanol metabolism; PBPK modeling.
Despite the fact that much is known regarding the pharmacokinetics of ethanol, central issues are still not completely
understood. Of particular interest is the extent of first-pass
metabolic clearance of orally administered ethanol because
interindividual variability in this clearance potentially contributes to susceptibility to the toxic effects of ethanol. For example, women are more susceptible to many of the adverse effects
of chronic ethanol exposure. The cumulative dose of ethanol
1
To whom correspondence should be addressed. E-mail: gina.pastino@
spcorp.com.
required to elicit cirrhosis in women is less than the dose
required to elicit the same response in men (Mezey et al.,
1988). Moreover, the time period for consumption of ethanol
resulting in chronic pancreatitis is shorter for women than men
(Mezey et al., 1988). A possible mechanism for the increase in
susceptibility among women includes greater bioavailability
due to less first-pass metabolic clearance (Frezza et al., 1990).
Women have lower levels of gastric alcohol dehydrogenase
(ADH) and metabolize less ethanol during the first pass
through the gut (Frezza et al., 1990). Thus, a greater dose
reaches the systemic circulation in women.
In addition, several factors have been shown to affect the
bioavailability of ethanol that may influence toxicity. For example, administration of ethanol during fasting increases the
bioavailability as compared to administration during the fed
state (DiPadova et al., 1987; Gentry et al., 1992). That is, blood
ethanol concentrations (BEC) are higher in the fasted state than
when ethanol is given with food. The concentration of ethanol
administered can also alter the bioavailability. Humans who
ingested 4% ethanol had higher BEC than those who ingested
10% ethanol (Sharma et al., 1993). Similar effects of concentration have been found in rats (Roine et al., 1991). The effect
of the prandial state and concentration on bioavailability is, in
part, mediated by altered absorption rates. Fasting and alcohol
dilution are associated with faster gastric emptying (i.e., increased absorption rate), which in turn affects first-pass metabolic clearance (Gentry et al., 1994).
Blood ethanol concentrations (BEC) in rats following a dose
of 1.0 g/kg (Lim et al., 1993), and in humans following a dose
of 0.15 g/kg (Julkunen et al., 1985) illustrate the first-pass
metabolism (FPM) of ethanol associated with oral ingestion.
BEC are much higher following an intravenous (iv) administration when compared to an equivalent oral administration.
Although FPM typically refers to loss due to hepatic metabolism, gastric metabolism has also been implicated in the decrease in bioavailability of ethanol. The terms gastric FPM
(FPM G) and hepatic FPM (FPM H) have been introduced by
several authors to provide the distinction (Lim et al., 1993;
Roine et al. 1991). The relative contribution of each pathway
256
257
BIOAVAILABILITY OF ETHANOL
in either rats or humans is not known with certainty and is the
main focus of this paper.
Lim et al. (1993) proposed that the FPM of orally administered ethanol is due primarily to gastric ADH. Gastric ADH
activity has been measured in rats (Lamboeuf et al., 1981;
Caballeria et al., 1987), mice (Algar et al., 1983), and humans
(Hempel and Peitruszko, 1975). Histamine 2-blockers, such as
cimetidine, inhibit gastric ADH activity in vitro (Palmer, 1987)
and also increase BEC, suggesting that metabolism by gastric
ADH is at least partly responsible for the first-pass clearance
(Caballeria et al., 1989).
Despite the evidence regarding gastric FPM, other reports
have implicated the liver as the primary site of FPM and also
suggest dependency of bioavailability on the rate of ethanol
absorption (Smith et al., 1992). Levitt and Levitt (1994) used
a 2-compartment model developed for human males to illustrate that the FPM of ethanol was a result of hepatic metabolism and that gastric ADH did not contribute significantly to
FPM. Their model also illustrated the dependency of first-pass
clearance on the absorption rate of ethanol. The results from a
pharmacokinetic model developed by Derr (1993) agree with
those of Levitt and Levitt (1994).
FPM has been quantified by comparisons of the ratio of the
blood area under the curve (AUC) following oral versus intravenous (iv) or intraperitoneal (ip) administration of ethanol
(DiPadova et al., 1987). Another approach has been to quantify
the total amount (mg) of ethanol absorbed following an oral
versus an iv route of administration, with the difference being
equal to the amount of FPM (Lim et al., 1993; Roine et al.,
1991). Lim and coworkers (1993) administered ethanol by
routes that bypassed the stomach (i.e., intraduodenal and intraportal), and found BEC equivalent to those obtained following an iv administration, thereby implicating the gastric mucosa as the primary site of FPM.
The reasons for the inconsistent results of the studies described above are not entirely clear. The use of invasive techniques may have been a factor. Animals were administered
ethanol by the intraduodenal or intraportal route (Lim et al.,
1993) or via isolated liver perfusions (Matsumoto et al., 1994).
These techniques require the use of anesthetics other than
ethanol that can potentially influence results. In addition, the
use of blood AUC may not accurately estimate bioavailability
for chemicals whose metabolism can be saturated. When the
rate of metabolism is first order, the bioavailability of a chemical is typically measured by blood AUC Oral:AUV IV. However,
when metabolism is pseudo-zero order, an increase in dose
does not result in a proportional increase in blood AUC. That
is, if the dose is doubled, the blood AUC is not necessarily
doubled. Comparisons of AUC are appropriate only when BEC
are low and the rate of metabolism is proportional to BEC.
Ethanol metabolism is saturated at pharmacologically relevant
doses (i.e., doses typically consumed by social drinkers and
alcohol abusers). Therefore, blood AUC Oral:AUC IV may not
provide an accurate estimate of bioavailability under exposure
scenarios of clinical relevance.
Physiologically based pharmacokinetic (PBPK) modeling
(Pastino et al., 1997) has recently been used to characterize the
disposition of ethanol. In contrast to classical pharmacokinetic
methods, PBPK models take into consideration anatomical and
physiological processes (tissue volumes and blood flows) as
well as biochemical (metabolic rate constants) and physiochemical (partition coefficients) properties of the specific
chemical (Clewell and Andersen, 1985; Himmelstein and Lutz,
1979). These characteristics allow for a more biologically
based approach to quantifying FPM and bioavailability of
ethanol than classical methods. A PBPK model can also be
used to characterize the dosimetry of ethanol under a variety of
conditions that may alter the bioavailability. The purpose of
this study was to utilize a physiologically based pharmacokinetic (PBPK) model for ethanol to estimate the relative contributions of hepatic and gastric metabolic clearance to the oral
bioavailability of ethanol in male rats.
MATERIALS AND METHODS
PBPK Model Theoretical Development. The blood flow limited PBPK
model for ethanol previously developed for mice (Pastino et al., 1996b) and
rats (Pastino et al., 1997) was extended to include oral administration in rats.
Compartments for the present model include liver, stomach, brain, fat, rapidly
perfused tissue, slowly perfused tissue, and blood (Fig. 1). Mass-balance
equations were written describing the rate of change in ethanol concentration
for each compartment, as described by Pastino et al. (1996b, 1997).
Mass-balance hepatic metabolism of ethanol was quantified by a MichaelisMenten expression for hepatic ADH included in the equation describing the
rate of change of ethanol in the liver (dA L/dt):
dAL/dt ⫽ QL ⫻ (CA ⫺ CVL)
⫺ [VmaxH ⫻ CVL/(KMH ⫹CVL)] ⫹ KaS ⫻ AS (1)
where Q L is the blood flow to the liver (l/h), CA is the concentration of ethanol
in the arterial blood entering the liver (mg/l), CV L is the concentration of
ethanol in the venous outflow from the liver (mg/l), Vmax H (mg/h) and K MH
(mg/l) are the Michaelis-Menten constants for metabolism by hepatic ADH,
Ka S is the first order oral absorption rate constant (hr –1), and A S is the amount
of ethanol in the stomach (mg). Under the experimental conditions of the
studies used in the development of this model, it is appropriate to describe the
absorption of ethanol as first order (Holford, 1987).
The rate of change of ethanol in the stomach is a function of the rate of
absorption and gastric metabolism, as given by:
dAS/dt ⫽ ⫺[VmaxG ⫻ CGMuc/(KMG ⫹ CGMuc)] ⫺ KaS ⫻ AS
(2)
where Vmax G and K MG are the Michaelis-Menten constants for gastric ADH
metabolism and C GMuc is the concentration of ethanol in the gastric mucosa.
The equation used to estimate C GMuc was:
CGMuc ⫽ (AS/VS) ⫻ PML
(3)
where A S is the amount of ethanol in the stomach (mg), V S is the stomach
volume (l), and P ML is the fraction of ethanol in the gastric mucosa relative to
the amount of ethanol in the lumen of the stomach.
The typical tissue:blood partition coefficient, which normally describes the
258
PASTINO AND CONOLLY
The experimental data used in this model development were BEC obtained
from rat tail blood. Previous research demonstrated that during periods of
rising and declining ethanol levels, the concentration in the tail lagged behind
the arterial, jugular or femoral vein blood (Levitt et al., 1994). Concentrations
in the tail blood are not equivalent to the pooled venous blood concentration at
early time points after bolus dosing. The equation must take into account the
rate of transfer into the tail blood. Thus, the concentration of ethanol in the tail
blood (CV Tail) was described by:
CVTail ⫽ CVPooled ⫻ [1 ⫺ e (–K
Tail
⫻ T)
]
(5)
where K Tail is a first-order constant (hr –1), CV Pooled is the pooled venous blood
concentration.
The blood flows and tissue volumes for each compartment (Table 1) were
obtained from the report prepared by the International Life Sciences Institute,
Risk Science Institute, the United States Environmental Protection Agency on
“Physiological Parameter Values for PBPK Models” (International Life Science Institute, 1994). The ethanol partition coefficients for rats were determined by Kaneko et al. (1994). The kidney:blood partition coefficient was
used for the rapidly perfused compartment, and the skeletal muscle:blood
TABLE 1
Ethanol-specific Parameters Used in the PBPK Model for
Ethanol in the Male Sprague-Dawley Rat
Parameter
FIG. 1. Schematic diagram of the PBPK model for ethanol in the male
rat.
ratio of the chemical in the tissue relative to that in the blood, is not appropriate, because the partitioning of ethanol is dependent on the volume of the
stomach and changes with time. Previous research estimated that the concentration of ethanol at the active site of gastric ADH is 4% of the concentration
of ethanol in the stomach lumen (Smith et al., 1992). However, Pastino et al.
(1996a) reported that the concentration of ethanol in the gastric mucosa
exceeded 4% and changed over time relative to the amount in the stomach
lumen. In the study by Pastino et al. (1996a), rats were administered 1.0 g/kg
(16% w/v) ethanol orally and the ratio of ethanol in the gastric mucosa to the
stomach lumen (i.e., stomach contents) was measured at several time points
following the administration. This ratio is equivalent to P ML (Equation 3). A
nonlinear regression analysis of these data presented by Pastino et al. (1996a)
provided the following equation:
PML ⫽ ⫺0.0919 ⫻ log(T) ⫹ 0.1526
(4)
where T is time (h). The upper bound on P ML was set equal to 0.8. That is, at
time zero, P ML was set to 0.8, the maximum value measured by Pastino et al.
(1996b). However, the ratio ranged from 0.8 to 0.04 (Pastino et al., 1996b).
Equation 4 was used in the model to calculate the concentration of ethanol at
the active site of gastric ADH at any given time after a bolus oral dose (C GMuc,
Equation 3).
Partition coefficients b
Liver:blood
Brain:blood
Fat:blood
Rapidly perfused:blood
Slowly perfused:blood
Blood:air
Pharmacokinetic constants
Hepatic ADH Vmax (Vmax H, mg/h)
Hepatic ADH K M (K MH, mg/L)
Gastric ADH Vmax (Vmax G, mg/h)
Gastric ADH K M (K MG, mg/L)
Oral absorption rate (K A, h –1)
Tail equilibration rate (K TAIL, h –1)
Physiological constants e
Alveolar ventilation (L/h/kg)
Cardiac output (L/h/kg)
Blood flow (% cardiac output)
Brain
Liver
Fat
Rapidly perfused
Slowly perfused
Tissue volumes (% body weight):
Brain
Liver
Fat
Rapidly Perfused
Slowly Perfused
a
Estimation a
Value
Measured
Measured
Measured
Measured
Measured
Measured
0.81
0.87
0.11
0.95
0.80
2140.00
Fitted c
Measured d
Fitted c
Measured d
Fitted c
Fitted c
110.64
23.00
631.27
18400.00
4.15
3.82
Fixed
Fixed
Fixed
14.00
14.00
2.00
18.00
7.00
56.00
17.00
Fixed
0.57
3.66
7.36
4.77
73.64
Parameters were either obtained experimentally (measured) or optimized
against empirical data (fitted).
b
Kaneko, 1994.
c
See Table 2 for specific references used in the optimization procedures.
d
Caballeria et al., 1989.
e
As outlined in International Life Sciences Institute (1994).
259
BIOAVAILABILITY OF ETHANOL
TABLE 2
Summary of Data Used to Optimize for Kinetic Parameters
Parameter
Experimental blood ethanol concentration data
References
K Tail and Vmax H
Multistep IV administration of 1.0 g/kg
IV Administration of 0.25 g/kg, control and cimetidine pretreated
Oral Administration of 1.0 g/kg
Oral Administration of 1.0 g/kg
Oral Administration of 0.25 g/kg, Control and cimetidine
pretreated
Lim et al., 1993
Caballeria et al., 1989
Roine et al., 1991
Lim et al., 1993
Caballeria et al., 1989
K A and Vmax G
partition coefficient was used for the slowly perfused compartment. The
absorption and metabolic rate constants were determined as outlined below.
The mass-balance equations were solved simultaneously using a Dell Latitude computer with an Intel Pentium processor (Dell Computers, Round Rock,
TX) and the software package ACSL Tox for Windows (Pharsight Inc.,
Mountain View, CA). ACSL Tox is designed for modeling continuous systems
that can be described by time-dependent, nonlinear, differential equations and
includes optimization capabilities. The average time to run a simulation using
this system was less than 1 min. For parameter estimation, the Nelder-Mead
algorithm was used to maximize the likelihood function defined in ACSL Tox.
Rate Constants for Metabolism by Alcohol Dehydrogenase in Male Rats.
Mammalian ADH exists in multiple molecular forms (Agarwal and Goedde,
1990; Bosron and Li, 1986; Kedishvili et al., 1995). There are several
isozymes of rat ADH, specifically ADH 1, ADH 2, and ADH 3. ADH 3 is primarily responsible for ethanol metabolism in the liver whereas ADH 1 is responsible for ethanol metabolism in the stomach (Julia et al., 1987). While each
enzyme is comprised of two active subunits and require zinc and NAD ⫹ for
metabolic activity, the kinetic properties of each isozyme differ. For example,
the isoelectric points are 5.1 and 8.25– 8.4 for ADH 1 and ADH 3, respectively
(Julia et al., 1987). In addition, the K M for ethanol oxidation differs significantly. The potential for each isozyme to contribute to the in vivo elimination
of ethanol is therefore different. In the interests of model parsimony, only a
single Michaelis-Menten pathway was described in the gastric and hepatic
compartments. Attempting to describe multiple pathways in each compartment
would not be a useful exercise given the extent of the data available for
parameter estimation. The metabolic pathways described in the current model
can thus be thought of as representing the average metabolic behavior of the
enzymes capable of metabolizing ethanol.
The hepatic and gastric ADH K Ms utilized in the PBPK model were
experimentally determined by Caballeria et al. (1989; 23 mg/L and 18,400
mg/L, respectively). The remaining rate constants were optimized against
experimental BEC and included K Tail, Vmax H, Ka S and Vmax G (Table 2). K Tail
and Vmax H were optimized against BEC following an iv administration of 250
mg/kg [Caballeria et al., 1989; numerical data was kindly provided by Joan
Caballeria (personal communication)] and 1.0 g/kg (Lim et al., 1993). The
model was coded to account for the multi-step iv infusion rate used by Lim et
al. (1993). In this study, 50% of the dose (1.0 g/kg) was administered in the
first 15 min, 25% given over the next 15 min, 12.5% given over the following
90 min, and the final 12.5% over the last 240 min. The total infusion time was
6 h.
Once K Tail and Vmax H were obtained, the PBPK model was used to optimize
for Ka S and Vmax G. The data used for the determination of Ka S and Vmax G
were BEC following an oral administration of 1.0 g/kg (Lim et al., 1993; Roine
et al., 1991) and 250 mg/kg (Caballeria et al., 1989). The Caballeria et al.
(1989) data were obtained from rats pretreated with cimetidine, an inhibitor of
gastric ADH (Palmer, 1987), and in control animals receiving ethanol only. For
simulations using the data from animals pretreated with cimetidine, Vmax G
was set to zero.
Calculation of Gastric First Pass Metabolism of Ethanol. The first-pass
clearance of orally administered ethanol results from metabolism by gastric
ADH, which occurs prior to absorption from the gastrointestinal (GI) tract into
the liver, and metabolism by the liver through the first pass prior to absorption
into the systemic circulation. The amount of gastric ethanol metabolism (AM G;
mg) was calculated using the PBPK model, through integration of the Michaelis-Menten expression, for metabolism by gastric ADH (Equation 2). Gastric
FPM, represented as a percentage of the administered dose, was then calculated by:
FPMG ⫽ (AMG /Dose) ⫻ 100
(6)
Calculation of Hepatic First-Pass Metabolism of Ethanol. In order to
calculate FPM H, a distinction between the two sources of hepatic ethanol
metabolism was made. The assumptions of the PBPK model are that the liver
is well mixed and delivery of ethanol to the liver is blood-flow limited. Ethanol
enters the liver from the GI tract and as recirculating ethanol. Regardless of
how ethanol reaches the liver, it is metabolized by a saturable system having
Michaelis-Menten kinetics characterized by Vmax and K M. The total rate of
hepatic ADH metabolism (RAM T; mg/h) is described by:
RAMT ⫽ RAMR ⫺ L ⫹ RAMGI – L
(7)
where RAM R – L is the rate of metabolism of recirculating ethanol (mg/h) and
RAM GI – L is the rate of metabolism of ethanol entering the liver from the GI
tract (mg/h). RAM GI – L is the rate of first pass hepatic metabolic clearance.
Given that the liver is assumed to be well mixed, ethanol entering the liver
by one route competes for metabolism of ethanol entering by another route.
The rate of metabolism by each pathway (i.e., RAM R – L and RAM GI – L) can
therefore be described by the equation for competitive inhibition (York, 1997).
In describing the rate of metabolism of ethanol entering the liver from the GI
tract, the substrate concentration is the concentration of ethanol in the liver
resulting from newly absorbed ethanol from the GI tract, and the concentration
of the inhibitor is the concentration of recirculating ethanol. The inhibitor
affinity constant is the hepatic ADH K MH for ethanol oxidation. The rate of
metabolism of ethanol entering the liver from the GI tract is estimated by:
RAMGI-L ⫽ VmaxH ⫻ CVGI-L /[CVGI-L ⫹ KMH ⫻ (1 ⫹ CVR-L/KMH)]
(8)
where CV GI-L is the liver venous blood concentration of ethanol that results
from ethanol received from the GI tract (mg/l), Vmax H (mg/h) and K MH (mg/l)
are the Michaelis-Menten constants for metabolism by hepatic ADH, and
CV R – L (mg/l) is the liver venous blood concentration resulting from recirculating ethanol. The total liver venous BEC is:
CVL ⫽ CVGI – L ⫹ CVR – L
(9)
Therefore, Equation 8 reduces to:
RAMGI-L ⫽ VmaxH ⫻ CVGI ⫺ L /(KMH ⫹ CVL)
(10)
Integration of Equation 10 provides the amount of ethanol metabolized by
260
PASTINO AND CONOLLY
hepatic ADH during the first pass through the liver (AM H; mg), and was used
to calculate the hepatic FPM of ethanol:
FPMH ⫽ (AMH /Dose) ⫻ 100
(11)
Calculation of the Oral Bioavailability of Ethanol. The bioavailability,
represented as a percentage of the total dose administered (BIO), was calculated as:
BIO ⫽ (Dose – AMG – AMH)/Dose ⫻ 100
(12)
The bioavailability of ethanol was represented as a percentage of the administered dose since the bioavailability is the amount of administered ethanol that
reaches the systemic circulation and is not specific to a particular organ (e.g.,
liver or gut). When estimating the contribution of an organ to the overall
metabolic clearance, FPM is calculated relative to the amount presented to the
organ. In the case of hepatic FPM, the amount presented to the liver is the
amount of dose absorbed, not the dose administered. For this reason, the
decrease in oral bioavailability is not equivalent to the sum of the gastric and
hepatic FPM.
RESULTS
The ethanol pharmacokinetic parameters used in the model
were either obtained from the literature or were optimized
against previously published experimental data (Table 1). The
K MH (23 mg/L) and K MG (18,400 mg/L) were experimentally
determined by Caballeria et al. (1989). Table 2 provides an
outline of the data utilized in optimizing for the remaining
parameters, which included the Vmax H, Vmax G, Ka S and K Tail.
Experimental tail BEC following an iv administration of 250
mg/kg (Figure 2, top panel) and 1000 mg/kg (Figure 3, top
panel) were used to optimize for K Tail and Vmax H. Experimental tail BEC following an oral administration of 250 mg/kg
(Figure 2, bottom panel) and 1000 mg/kg (Figure 3, bottom)
were used to optimize for Ka S and Vmax G with the values for
K Tail and Vmax H that were estimated from the iv data.
The model predicted BEC obtained from animals pretreated
with cimetidine, an inhibitor of gastric ADH (Fig. 2, bottom
and top panels, squares). These simulations were obtained in
the absence of gastric metabolism (i.e., the Vmax G was set to
zero) while keeping all other parameters constant. That is,
Vmax H, K MH and Ka S were not changed from the simulations
of data from animals receiving ethanol only (Fig. 2, top and
bottom panels, circles), thereby strengthening the validity of
the model.
The optimizations provided accurate simulations of the experimental data, with the possible exception of the BEC, following oral administration of 1000 mg/kg reported in Lim et
al. (1993; Fig. 3, bottom panel, circles). It is unclear why the
PBPK model overpredicted the Lim et al. (1993) data but not
the Roine et al. (1991) data. The only differences between the
experimental conditions in these studies was the body weight
of the animals. The dose, concentration, and strain of rats were
the same in both studies. The PBPK model accounted for the
body weights when simulating each respective data set, as
illustrated by the differences in the PBPK model simulations
FIG. 2. Experimental versus simulated blood ethanol concentrations in
male rats following an iv (top) and oral (bottom) administration of 250 mg/kg.
The experimental data are from Caballeria et al. (1989) and are the mean
⫾SEM from 6 male Sprague-Dawley rats. In this study, animals were administered either 50 mg/kg cimetidine or saline followed by administration of
ethanol either orally or intravenously. The circles represent the data obtained
in animals pretreated with cimetidine prior to ethanol exposure and the squares
represent the data obtained in animals receiving ethanol only. Simulations of
the data obtained from cimetidine treated animals were obtained in the absence
of gastric ADH activity (i.e., Vmax G ⫽ 0) because cimetidine inhibits gastric
ADH activity. The simulation of the iv data assumed an infusion time of 5 min.
(Fig. 3, bottom panel). Although the specific ages of the
animals were not specified in each study, the animals used in
the Roine et al. (1991) study were purchased as adults whereas
the animals used in the Lim et al. (1993) study were purchased
as weanlings. It is possible that the age difference may have
affected the rates of hepatic metabolism and therefore the BEC
(Seitz et al., 1992). The simulations in Figures 2 and 3 were
obtained using a single hepatic ADH Vmax and did not account for possible age-dependent differences in hepatic metabolism.
The experimental data used for the optimization procedures
were tail BEC obtained following both oral and iv administra-
BIOAVAILABILITY OF ETHANOL
261
FIG. 4. Simulated blood ethanol concentrations following an oral administration. The solid line represents pooled venous blood and the dashed line
represents the tail blood. Tail BEC were calculated using Equation 5.
FIG. 3. Experimental versus simulated blood ethanol concentrations following an iv (top) and oral (bottom) administration of 1.0 gm/kg. The triangles
are the experimental data from Roine et al. (1991), and the circles are the
experimental data obtained from Lim et al. (1993). The experimental data from
Roine et al. (1991) are the mean ⫾SEM from 7 rats, and the experimental data
from Lim et al. (1993) are the means from 6 rats. The iv administration used
in Lim et al. (1993) was a 4-step infusion rate over a 6-h time period. Fifty
percent of the dose was administered in the first 15 min, 25% given over the
next 15 min, 12.5% given over the following 90 min, and the final 12.5% over
the last 240 min. The solid lines are the PBPK model simulations obtained
using the body weights reported for each study: 0.315 kg for the Lim et al.
(1993) study and 0.351 kg for the Roine et al. (1991) study. In each study, male
Sprague-Dawley rats were administered 1000 mg/kg ethanol as a 16% (w/v) in
saline by intragastric intubation.
tion. Previous research found that, during periods of rising and
declining ethanol levels, tail BEC lag behind concentrations in
the arterial, jugular, and femoral vein blood in rats (Levitt et
al., 1994). This is due to the low blood perfusion to tissue
water ratio in the tail. The simulations in Figure 4 illustrate the
discrepancy between pooled venous BEC and tail BEC following a bolus oral administration. At earlier time points, tail BEC
are lower than pooled venous BEC. However, these concentrations eventually equilibrate.
The model was then applied to provide quantitative estimates of first pass metabolism and bioavailability of ethanol.
As the simulated dose increased, gastric and hepatic FPM
decreased as a fraction of dose (Fig. 5). In addition, PBPK
model predictions of gastric FPM were higher than the model
predictions of hepatic FPM except at the lowest simulated dose
(100 mg/kg). At a simulated dose of 500 mg/kg, gastric FPM
was predicted to be 26% of the administered dose, whereas
hepatic FPM was predicted to be 12% of the administered
dose. Simulation of an intermediate dose (1000 mg/kg) provided estimates of 22% and 5% for gastric and hepatic FPM,
respectively. At the highest simulated dose (3000 mg/kg) gastric and hepatic FPM were predicted to be 15% and 2% of the
administered dose, respectively.
The predicted dose- and absorption rate-dependent changes
in gastric and hepatic FPM were reflected in predictions of
bioavailability. In the simulated dose range (200 mg/kg to 1200
mg/kg), the predicted bioavailability increased as the dose and
absorption rate increased (Fig. 6). The bioavailability did not
reach 100% in the simulated dose range and was greatest at the
highest dose with the fastest absorption rate.
FIG. 5. Simulated dose-dependent effects on the first-pass metabolism of
ethanol in male rats. Gastric FPM (triangles) was calculated using Equation 6
and hepatic FPM (circles) was obtained using Equation 11.
262
PASTINO AND CONOLLY
FIG. 6. Simulated effect of dose
and absorption rates on the oral bioavailability of ethanol in male rats.
The bioavailability was calculated
using Equation 12.
DISCUSSION
Previous research illustrated that a portion of ingested ethanol does not reach the systemic circulation, due to metabolism
by gastric or hepatic ADH (Roine et al., 1991; Caballeria et al.,
1987). The significance of this first pass clearance of ethanol is
that it may play a protective role against the toxic effects of
ethanol since it provides for presystemic elimination, thereby
decreasing the amount of ethanol to which sensitive tissues are
exposed. Gastric metabolism of ethanol reduces the dose of
ethanol to which the liver is exposed. Hepatic metabolism
further decreases the dose of ethanol to the remaining tissues,
such as the brain. Consequently, first pass clearance is a mitigating factor in susceptibility to the toxicity of ethanol. To
completely understand the possible protective effect of FPM,
its site and the mechanism need to be fully elucidated. The
purpose of this study was to refine a previously developed
PBPK model for ethanol in male rats to quantify the relative
contribution of gastric and hepatic FPM to the oral bioavailability of ethanol.
In calculating FPM H, the PBPK model provided a distinction
between metabolism of incoming ethanol from the GI tract and
metabolism of recirculating ethanol. Under the assumptions of
a well-stirred liver, Michaelis-Menten elimination kinetics, and
flow-limited delivery of ethanol, the ethanol entering the liver
from the GI tract necessarily competes for metabolism with
ethanol that has entered the liver through recirculation. When
the duration of absorption from the GI tract is long with respect
to the time it takes the blood to recirculate, the calculation of
FPM H must take into account the competition with recirculating ethanol. Accordingly, at least some of the ethanol will be
metabolized during the first pass through the liver, even when
concentrations in the blood are high. Thus, the bioavailability
should never reach 100%, as was illustrated by the PBPK
model.
In addition, because hepatic and gastric metabolism were
described as saturable processes, the predicted bioavailability
should be dose-dependent. As the amount of ethanol administered increases, the capacity of the liver to metabolize it,
relative to the amount presented to the liver, decreases, and
more of the ethanol escapes metabolism during the first pass.
An increase in bioavailability is therefore expected with an
increase in the dose, based on the kinetic characteristics of
gastric and hepatic ADH.
The calculated FPM H of ethanol is higher than previously
found (Levitt and Levitt, 1994) and is probably due to the
respective definition and the method for calculating FPM. The
traditional definition of FPM is the removal of chemicals
before entrance into the systemic circulation, typically by
metabolism in the gut or liver (Rozman and Klaassen, 1996).
However, Levitt and Levitt (1994) defined hepatic FPM as:
MFPM ⫽
冕
[VMAX ⫻ CL/(KM ⫹ CL)
⫺ VMAX ⫻ CLR/(KM ⫹ CLR)]dt
(13)
where V MAX and K M are the Michaelis-Menten constants for
metabolism by hepatic ADH, C L is the total concentration of
ethanol in the liver, and C LR is the concentration of ethanol in
the liver resulting from recirculating ethanol. Thus, the first
BIOAVAILABILITY OF ETHANOL
term is the total rate of liver metabolism and the second is the
rate of metabolism if the only source of ethanol to the liver was
from recirculating ethanol with no direct supply of newly
absorbed ethanol from the gastrointestinal tract (Levitt and
Levitt, 1994).
M FPM as defined by Levitt and Levitt (1994; Equation 13)
quantifies the effect of ethanol entering the liver from the GI
tract on the overall rate of metabolism and therefore makes no
provision for the competition between the ethanol entering the
liver from the GI tract and recirculating ethanol. Accordingly,
if the recirculating concentration of ethanol is high enough and
metabolism is saturated, this equation specifies that there will
be no additional metabolism of ethanol because it is already
occurring at the maximum velocity. That is, at pharmacologically relevant doses (i.e., higher doses) the first and second
terms approach equivalency, there is very little effect on the
overall rate of metabolism of ethanol, and M FPM becomes
negligible. In the current model, Equation 13 was modified to
account for the continual absorption of ethanol from the GI
tract. When this is done, the first-pass clearance increases and
bioavailability decreases. It is the portion of ingested ethanol
undergoing first-pass clearance, due to gastric and hepatic
metabolism, that is the relevant factor in understanding the
relationship between ethanol consumption, exposure to target
tissues, and biological effects.
Levitt and Levitt (1994) have also suggested that in rats and
humans the first-pass clearance of ethanol attributed to firstpass gastric metabolism might result from differences in the
rate of ethanol absorption and its influence on hepatic metabolism. That is, based on the kinetic characteristics of hepatic
ADH, the amount of ethanol cleared by the liver is extremely
sensitive to changes in the absorption rate of ethanol. Furthermore, it is the increase in this absorption rate, as the dose
increases, that results in a proportionally lower amount of
ethanol metabolism in the liver. The PBPK model simulations
agree with the suggestion that the differences in the absorption
rate might explain, at least in part, the discrepancy between the
estimates of the oral bioavailability by Levitt and Levitt (1994)
and those of Roine et al. (1991). However, the model showed
that even at a given absorption rate other factors, such as the
dose and gastric metabolism, affect measurements of first-pass
clearance and bioavailability.
While some of the literature reports support a lack of effect
of gastric metabolism on the oral bioavailability of ethanol
(Derr, 1993; Levitt and Levitt, 1994), other reports indicate
that gastric metabolism does play a role in bioavailability (Lim
et al., 1993; Roine et al., 1991). These latter studies reported
that FPM G is approximately 25 to 30% of the administered
dose in the male rat (1000 mg/kg; Lim et al., 1993; Roine et al.,
1991). At a simulated dose of 1000 mg/kg, the PBPK model
predicted FPM G to be 22%, which is in agreement with Roine
et al. (1991) and Lim et al. (1993). The model also predicted
a dose-dependent effect on FPM G; as the simulated dose increased predictions of FPM G decreased.
263
The model predicted considerable first-pass metabolic clearance attributable to both gastric and hepatic metabolism at all
simulated doses. At a dose roughly equivalent to the consumption of 1.5 standard alcoholic beverages by an average, healthy
adult male (500 mg/kg), FPM G and FPM H were predicted to be
26 and 15% of the administered dose, respectively. Thus, these
metabolic barriers are protective against toxicity because they
reduce exposure to target organs such as the liver and brain.
Differences in capacity of gastric or metabolic clearance therefore can influence susceptibility to the toxic effects of ethanol.
Moreover, the PBPK model accurately simulated experimental BEC from animals pretreated with cimetidine, a gastric
ADH inhibitor (Palmer, 1987). These simulations were obtained in the absence of gastric ADH activity while keeping all
other parameters the same. Specifically, the absorption rate and
hepatic metabolic constants were not changed from those used
to simulate data from animals receiving ethanol only. The
validity of the model was strengthened by its ability to predict
these data and further illustrate the influence of gastric ethanol
metabolism on oral bioavailability. If gastric ADH metabolism
of ethanol was not a factor in the bioavailability, it is likely that
the model would not have provided accurate predictions of
experimental BEC from animals pretreated with cimetidine.
Use of Equation 5 to describe a difference between the
mixed venous and tail vein blood concentrations of ethanol is
consistent with experimental observations (Levitt et al., 1994).
However, Equation 5 provides an empirically, rather than a
physiologically based description. The latter would be more
complex, and would require a specific description of tail vein
blood flow and perhaps of diffusional exchange of ethanol
between arterial blood flowing towards the distal regions of the
tail and venous blood returning from the distal regions. This
being the case, the parameter K Tail cannot be assigned a rigorous physiological definition. This means that the value identified by optimization against the iv data might not be optimal
for other experimental conditions, such as oral dosing. The
estimates of Vmax G and K A and the related calculations of
FPM are thus uncertain to the extent that the value of K Tail may
be experiment-specific. Although a rigorous examination of
this issue was not conducted in the present study, a preliminary
study indicated that optimization of K Tail against oral data did
not significantly affect the estimated value of Vmax G, rather,
some covariance of K Tail and K A was seen (results not shown).
Thus, the overall conclusions of this work with respect to FPM
are not likely to be sensitive to the strategy used to identify
K Tail.
The inhibition of gastric ADH activity by drugs, such as
cimetidine, may have implications for people being treated for
gastritis. The inhibition of gastric ADH activity by the concomitant use of certain H 2-receptor antagonists (e.g., cimetidine) has been demonstrated in humans (Gupta et al., 1995). In
this study, healthy male volunteers were administered 600mg/kg ethanol postprandially, before and after cimetidine treatment. Peak BEC following exposure to cimetidine were sig-
264
PASTINO AND CONOLLY
nificantly increased. Although the mean increase in peak BEC
in this study was small (approximately 3 mM), 3 subjects
exceeded the legal limit for driving while impaired, following
pretreatment with cimetidine (Gupta et al., 1995). Thus, the use
of these drugs may result in higher BEC than would otherwise
be expected and may actually potentiate ethanol toxicity. For
example, in a recent study, an oral administration of ethanol to
male rats resulted in a less pronounced decrease in hepatic
glutathione levels, as well as a quicker recovery compared to
an ip administration (Battiston et al., 1996). However, this
difference was eliminated when the rats were pretreated with
cimetidine.
As previously discussed, difference in gastric FPM may play
a role in the observed sex-dependent differences in the adverse
health effects of ethanol consumption. Frezza et al. (1990)
found that women have lower levels of gastric ADH and
metabolize less ethanol during the first pass through the liver.
In fact, the opposite is true in rats. There are no sex-dependent
differences in hepatic metabolism of ethanol, but female rats
have higher gastric ADH activity and enzyme protein levels
(Mezey et al., 1992). Although the current model was developed for male rats, it would be expected to predict lower blood
ethanol levels and reduced bioavailability in female rats as
compared to male rats.
In summary, the PBPK model developed in male rats illustrated how the observed first-pass clearance of ethanol following an oral administration can be attributed to metabolism by
both gastric and hepatic ADH, and it demonstrated the dependence of this clearance on the dose and the absorption rate. The
PBPK model was used to provide quantitative estimates of
gastric and hepatic FPM and bioavailability, and provided
insight regarding different approaches to calculate the FPM of
ethanol. The ability of the model to accurately simulate ethanol
pharmacokinetic data from several experiments, including data
from animals pretreated with cimetidine, which inhibits gastric
ADH, is consistent with roles of both gastric and hepatic
metabolism as determinants of the oral bioavailability of ethanol.
ACKNOWLEDGMENTS
We would like to acknowledge the assistance of Drs. Lester Sultatos,
Edward J Flynn, and Charles Lieber for their discussions pertaining to this
work. The authors also thank Drs. Michael Gargas and Susan Borghoff for
their review of this paper. During her fellowship at the Alcohol Research and
Treatment Center (Mount Sinai School of Medicine, Bronx, N.Y.), G.M.P. was
supported by NIH Research Training Fellowship AA07275.
REFERENCES
Agarwal, D., and Goedde, H. (1990). Pharmacogenetics of alcohol dehydrogenase (ADH). Pharmacol. Ther. 45, 69 – 83.
Algar, E. M., Seeley, T. L., and Holmes, R. (1983). Purification and molecular
properties of mouse alcohol dehydrogenase isozymes. Eur. J. Biochem. 137,
139 –147.
Andersen, M. E., Clewell H. J., III, Gargas, M. L., Smith, F. A., and Reitz,
R. H. (1987). Physiologically based pharmacokinetic modeling and the risk
assessment process for methylene chloride. Toxicol. Appl. Pharmacol. 87,
185–205.
Battiston, L., Moretti, M., Tulissi, P., Fanni-Canelles, M., and Pozzato, G.
(1996). Alcohol hepatic toxicity rat: evidence of utility of gastric ethanol
metabolism. Int. Hepatol. Commun. 6, 61–70.
Bosron, W., and Li, T. K. (1986). Genetic polymorphism of human liver
alcohol and aldehyde dehydrogenases and their relationship to alcohol
metabolism and alcoholism. Hepatology 6, 502–510.
Caballeria, J., Baraona, E., and Lieber, C. S. (1987). The contribution of the
stomach to ethanol oxidation in the rat. Life Sci. 41, 1021–1027.
Caballeria, J., Baraona, E., Rodamilans, M., and Lieber, C. S. (1989). Effects
of cimetidine on gastric alcohol dehydrogenase activity and blood ethanol
levels. Gastroenterology 96, 388 –392.
Clewell, H. J., III, and Andersen, M. E. (1985). Risk Assessment extrapolations and physiological modeling. Toxicol. Ind. Health 1, 111–131.
Derr, R. (1993). Simulation studies on ethanol metabolism in different populations with a physiological pharmacokinetic model. J. Pharmaceut. Sci. 82,
677– 682.
DiPadova, C., Worner, T. M., Julkunen, R. J., and Lieber, C. S. (1987). Effects
of fasting and chronic alcohol consumption on the first-pass metabolism of
ethanol. Gastroenterology 92, 1169 –1173.
Frezza, M., DiPadova, C., Pozzato, G., Terpin, M., Baraona, E., and Lieber,
C. S. (1990). High blood levels in women. The role of decreased gastric
alcohol dehydrogenase activity and first-pass metabolism. N. Engl. J. Med.
322, 95–99.
Gentry, R. T., Baraona, E. and Lieber, C. S. (1994). Agonist: gastric first pass
metabolism of alcohol. J. Lab. Clin. Med. 123, 21–26.
Gentry, R. T., Sharma, R., Lim, R. T., Baraona, E., DiPadova, C., and Lieber,
C. S. (1992). Food augments first-pass metabolism (FPM) of alcohol by
promoting gastric retention. Alcohol Clin. Exp. Res. 16, 607.
Gupta, A. M., Baraona, E., and Lieber, C. S. (1995). Significant increase of
blood alcohol by cimetidine after repetitive drinking of small alcohol doses.
Alcohol Clin. Exp. Res. 19, 1083–1087.
Hempel, J. D., and Pietruszko, R. (1975). Human stomach alcohol dehydrogenase: isoenzyme composition and catalytic properties. Alcohol Clin. Exp.
Res. 3, 95–98.
Himmelstein, K. J., and Lutz, R. J. (1979). A review of the application of
physiologically based pharmacokinetic modeling. J. Pharmacokinet. Biopharm. 7, 127–145.
Holford, N. H. (1987). Clinical pharmacokinetics of ethanol. Clin. Pharmacokinet. 13, 273–292.
International Life Sciences Institute, Risk Science Institute, and the United
States Environmental Protection Agency (1994). Physiological Parameter
Values for PBPK Models.
Julia, P., Farres, J., and Pares, X. (1987). Characterization of three isoenzymes
of rat alcohol dehydrogenase. Tissue distribution and physical and enzymatic properties. Eur. J. Biochem. 162, 179 – 89.
Julkunen, R. J., Di Padova, C., and Lieber, C. S. (1985). First pass metabolism
of ethanol—A gastrointestinal barrier against the systemic toxicity of ethanol. Life Sci. 37, 567–573.
Kedishvili, N. Y., Bosron, W. F., Stone, C. L., Hurley, T. D., Peggs, C. F.,
Thomasson, H. R., Popov, K. M., Carr, L. G., Edenberg, H. J., and Li, T. K.
(1995). Expression and kinetic characterization of recombinant human stomach alcohol dehydrogenase. Active-site amino acid sequence explains substrate specificity compared with liver isozymes. J. Biol. Chem. 270, 3625–
3630.
Kaneko, T., Wang, P-Y., and Sato, A. (1994). Partition coefficients of some
acetate esters and alcohols in water, blood, olive oil, and rat tissues. Occup.
Environ. Med. 51, 68 –72.
BIOAVAILABILITY OF ETHANOL
Lamboeuf, Y., de Saint Blanquat, G., and Derache, R. (1981). Mucosal alcohol
dehydrogenase- and aldehyde dehydrogenase-mediated ethanol oxidation in
the digestive tract of the rat. Biochem. Pharmacol. 30, 542–545.
Levitt, M. D., and Levitt, D. G. (1994). The critical role of the rate of ethanol
absorption in the interpretation of studies purporting to demonstrate gastric
metabolism of ethanol. J. Pharmacol. Exp. Ther. 269, 297–304.
Lim, R. T., Jr., Gentry, R. T., Ito, D., Yokoyama, H., Baraona, E., and Lieber,
C. S. (1993). First-pass metabolism of ethanol is predominantly gastric.
Alcohol Clin. Exp. Res. 17, 1337–1344.
Matsumoto, H., Matsubayashi, K., and Fukui, Y. (1994). Estimation of ethanol
first-pass effect in the perfused rat liver. Alcohol Alcohol 29, 9 –13.
Mezey, E., Kolman, C. J., Diehl, A. M., Mitchell, M. C., and Herlong, H. F.
(1988). Alcohol and dietary intake in the development of chronic pancreatitis and liver disease in alcoholism. Am. J. Clin. Nutr. 48, 148 –151.
Mezey, E., Sharma, S., Rennie, L., and Potter, J. (1992). Sex differences in
gastric alcohol dehydrogenase activity in Sprague-Dawley rats. Gastroenterology 103, 1804 –1810.
Palmer, R. H. (1987). Cimetidine and alcohol absorption. Gastroenterology 97,
1066 –1068.
Pastino, G. M., Asgharian, B., Roberts, K., Medinsky, M. A., and Bond, J.
(1997). A comparison of physiologically based pharmacokinetic model
predictions and experimental data for inhaled ethanol in male and female
B6C3F1 mice, F344 rats and humans. Toxicol. Appl. Pharmacol. 145,
147–157.
Pastino, G. M., Gentry, R. T., and Lieber, C. S. (1996a). Hepatic vs gastric
first-pass metabolism of ethanol in a physiologically-based pharmacokinetic
model. Alcohol Clin. Exp. Res. 20 (Suppl.), 35A.
265
Pastino, G. M., Sultatos, L. G., and Flynn, E. J. (1996b). Development and
application of a physiologically based pharmacokinetic model for ethanol in
the mouse. Alcohol Alcohol 31, 365–374.
Roine, R., Gentry, R. T., Hernandez-Munoz, R., Baraona, E., and Lieber, C. S.
(1990). Aspirin increases blood alcohol concentrations in humans after
ingestion of ethanol. JAMA 264, 2406 –2408.
Roine, R., Gentry, R. T., Lim, R. T., Jr., Baraona, E., and Lieber, C. S. (1991).
Effect of concentration of ingested ethanol on blood alcohol levels. Alcohol
Clin. Exp. Res. 15, 734 –738.
Rozman, K.K. and Klaassen, C. (1996). Absorption, distribution and excretion
of toxicants. In Casarett & Doull’s Toxicology: The Basic Science of
Poisons, 5th ed. (C. D. Klaassen, Ed.), p. 97. McGraw-Hill, New York.
Seitz, H. K., Xu, Y., Simanowski, U. A., and Osswald, B. (1992). Effect of
age and gender on in vivo ethanol elimination, hepatic alcohol dehydrogenase activity, and NAD ⫹ availability in F344 rats. Res. Exp. Med. 192,
205–212.
Sharma, R., Gentry, R. T., Chayes, Z., and Lieber, C. S. (1993). Higher
concentrations of alcohol yield lower blower alcohol levels because of
slower gastric emptying time and increase first pass metabolism of alcohol.
Alcohol Clin. Exp. Res. 17, 447.
Smith, T., DeMaster, E. G., Furne, J. K., Springfield, J., and Levitt, M. D.
(1992). First-pass gastric mucosal metabolism of ethanol is negligible in the
rat. J. Clin. Invest. 89, 1801–1806.
York, J. L. (1997) Enzymes: classification, kinetics, and control. In Textbook
of Biochemistry with Clinical Correlations, 4th ed. (T. M. Delvin, Ed.), pp.
127–178. Wiley-Liss, New York.