TOXICOLOGYANDAPPLIEDPHARMACOLOGY

A Physiologically Disposition

JOHN

W.

106,433-447(1990)

Based Toxicokinetic Model for the Uptake of Waterborne Organic Chemicals in Fish’

NICHOLS,*.~

MICHAEL L.G~m.xs,t~~ *United

JAMES

M.

McKIM,*

HARVEYJ.CLEWELL

MELVIN E. ANDERSEN,?.,~

111,-fAND RUSSELLJ. ERICKSON*

States Environmental Protection Agenqy, Environmental Research Laboratory-Duluth. Boulevard, Duluth, Minnesota 55804; and tArmstrong Aerospace Medical Research ki’right-Patterson Air Force Base, Dayton, Ohio 45433

Received

March

and

26, 1990; accepted August

6201 Congdon Laboratory,

15. 1990

A Physiologically Based Toxicokinetic Model for the Uptake and Disposition of Waterborne Organic Chemicalsin Fish. NICHOLS, J. W., MCKIM, J. M., ANDERSEN, M. E., GARGAS, M.L., CLEWELL, H. J., III, AND ERICKSON, R. J. (1990). Toxicol. Appl. Pharmacol. 106, 433-447. A physiologically based toxicokinetic model was developed to predict the uptake and disposition of waterborne organic chemicals in fish. The model consists of a set of mass-balance differential equations which describe the time course of chemical concentration within each of five tissue compartments: liver, kidney, fat, and richly perfused and poorly perfused tissue. Model compartmentalization and blood perfusion relationships were designed to reflect the physiology of fishes. Chemical uptake and elimination at the gills were modeled as countercurrent exchange processes, limited by the chemical capacity of blood and water flows. The model was evaluated by exposing rainbow trout (Oncorhynchus mykiss) to pentachloroethane (PCE) in water in fish respirometer-metabolism chambers. Exposure to 1500, 150, or 15 pg PCE/liter for 48 hr resulted in corresponding changes in the magnitude of blood concentrations without any change in uptake kinetics. The extraction efficiency for the chemical from water decreased throughout each exposure, declining from 65 to 20% in 48 hr. Extraction efficiency was close to 0% in fish exposed to PCE to near steady state (264 hr), suggesting that very little PCE was eliminated by metabolism or other extrabranchial routes. Parameterized for trout with physiological information from the literature and chemical partitioning estimates obtained in vitro, the model accurately predicted the accumulation of PCE in blood and tissues, and its extraction from inspired water. These results demonstrate the potential utility of this model for use in aquatic toxicology and environmental risk aSSeSSment. 0 1990 Academic PESS. 1~

With continuing, rapid growth in the inventory of xenobiotic industrial chemicals, it has become clear that individual testing of compounds to determine their environmental impacts is not possible and that a predictive capability is required. Fish and other aquatic or-

ganisms are exposed to chemicals in water and to chemicals complexed with food, particulates, and dissolved organic material. Estimating chemical dose to the animal therefore presents a difficult challenge. Current interest in predicting the kinetics of chemical uptake and disposition has arisen because of the need to understand differences in species sensitivity and to permit extrapolation between different routes of exposure. Moreover, there is an increasing need to relate toxic effects to the accumulation of chemicals in specific target tissues. Fish are known to possess a variety of

’ The U.S. Government’s right to retain a nonexclusive royalty-free license in and to the copyright covering this paper, for governmental purposes, is acknowledged. * To whom reprint requests and other correspondence should be addressed. 3 Present address: Chemical Industry Institute of Toxicology, Research Triangle Park, NC 27709. 433

0041-008X/90

$3.00

Copyright 0 1990 by Academic Press, Inc. All rights of reproduction in any form resewed.

434

NICHOLS

pathways for metabolism of xenobiotic chemicals, but without kinetic information the toxicological significance of these pathways is difficult to assess. Tissue dose estimates are also needed for carcinogenicity tests, in which fish have shown promise as an alternative to conventional rodent models (NIH, 1984; Powers, 1989). Kinetic models of chemical uptake and disposition in fish have largely been directed toward describing the accumulation of poorly soluble, lipophilic compounds (Neely et al., 1974;Neely, 1979). Thesemodels are generally composed of one homogeneous compartment into which the chemical accumulates as a function of uptake and elimination rate constants. While successfulfor many chemicals, these models cannot be usedwith compounds that display multiexponential kinetic behavior. Recently, multicompartmental analysis has been used to describe the uptake and elimination of chemicals by fish (Karara and Hayton, 1984; Stehly and Hayton, 1989). Varying in complexity, these models have proven useful as tools for understanding chemical accumulation in tissues and organs. In practice, however, the limitations of curve fitting restrict such models to two or three compartments. More importantly, the use of compartmental models for extrapolation is limited by the need to empirically fit a new set of parameters for each new compound, species,and route of exposure. One promising alternative to compartmental models is the development of physiologically based kinetic models (Himmelstein and Lutz, 1979; Gerlowski and Jain, 1983; Rowland, 1985). A physiologically based model defines an organism in terms of its anatomy, physiology, and biochemistry. This leads to a better understanding of the disposition and toxicity of a chemical and provides a basisfor extrapolation between species over a wide range of conditions. At present physiologically basedtoxicokinetic (PB-TK) models are being usedin risk assessment (Clewell and Andersen, 1985; Andersen et al., 1987) and have been successfully employed to extrapolate kinetic

ET

AL.

data between mammals (King et al., 1983; Lutz et al., 1984; Ramsey and Andersen, 1984). The application of physiologically based models to fish hasbeen limited to the pioneering work of Zaharko, Bungay, and co-workers (Zaharko et al., 1972; Bungay et al., 1976). Adapted from a mouse model, these models were designedto simulate the kinetics of compounds injected intravenously; neither model included a gill description, and elimination was considered to take place only via the urine and feces. In this paper we present a PB-TK model for the uptake and disposition of waterborne organic chemicals in fish. Our objective was to construct a model that would provide estimates of kinetic behavior without the prior need to expose fish. The model incorporates a flow-limited description.of chemical flux at fish gills (Erickson and McKim, 1990) and was GILLS

CART

QM

POORLY PERFUSED TISSUE GROUP

x 0.4

CVM

-

KIDNEY

Q~ ‘ART

d CART

‘VK

q--t--tMETABOLISM,

BILIARY

EXCRETION

ANDiOR FIRST ORDER ELIMINATION

FIG. 1. Schematic representation of a physiologically based toxicokinetic model for the uptake and disposition of waterborne organic chemicals in fish. Symbols used in the model schematic are presented in Table 1.

A PB-TK

MODEL

evaluated by exposing rainbow trout (Oncorhynchus m&.s$ to pentachloroethane (PCE) in fish respirometer-metabolism chambers. Parameterized for trout, the model accurately predicted the uptake of PCE and its accumulation in various tissues and organs. METHODS

FOR

TABLE ABBREVIATIONS PHYSIOLOGICALLY

F,9FT _

AND

1

SYMBOLS

BASED

USED

TO DESCRIBE

TOXICOKINETIC

MODEL

A FOR

.Y__

bw

Q” m QW

Model Development and Pa~ameterizat~on Physiologicaljsh model. A PB-TK model for fish was developed from an inhalation model for rats (Ramsey and Andersen, 1984) and incorporates changes reflecting the physiology of fishes (Fig. 1). Symbols and abbreviations are given in Table 1. Differential equations which describe the quantitative behavior of chemicals in the model are presented in the Appendix. Features unique to the fish model include an emphasis on portal blood flows, both to the liver and to the kidney, and a countercurrent gill description. Chemical flux at the tissues was assumed to be flow limited; that is, chemical equilibrium was considered to exist between the tissues in each compartment and blood exiting the compartment. Additional features of the model include exponential replacement functions to correct for dilution in the A and B sides of the exposure apparatus (Fig. 2). The differential equations which make up the model were solved simultaneously by numerical integration using a commercial software package (Advanced Continuous Simulation Language, Mitchell and Gauthier Associates, Concord, MA). Fish gill description. Chemical flux at fish gills was modeled as a countercurrent process regulated by gill ventilation, blood perfusion. and the relative affinity of the chemical for blood and water (Erickson and McKim, 1990). This description considers only those limitations imposed by the capacity of blood and water flows to deliver and remove the chemical, and assumes equilibrium between blood and water at perfused gill lamellae. Derivation of the fish gill description is presented in the Appendix. Model parameterization. Parameterization of the fish model requires: (1) estimates of cardiovascular and respiratory function, (2) compartment volumes and blood perfusion rates, (3) metabolic and excretory rates, and (4) equilibrium partition coefficients for blood and tissues. Physiological inputs for rainbow trout were obtained from the literature and are presented in Table 2. Cardiac output (Q,) at 12°C was estimated from the linear relationship between cardiac output and temperature established by Barron et al. (1987). Ventilation volume (Q”) and oxygen consumption rate (VOJ were patterned after values obtained by McKim, Bradbury and co-workers (McKim and Heath, 1983; McKim et al., 1987a,b; Bradbury et al., 1989). Effective respiratory volume (Q,,,), defined as the volume of inspired water that comes to equilibrium with

435

FISH

Subscripts

Body weight (kg) Ventilation volume (liters water/hr) Oxygen consumption rate (mg O,/hr) Effective respiratory volume (liters water/hr) Concentration in inspired water &g/liter) Concentration in expired water (pg/liter) Chemical extraction efficiency (“i reduction Of GmpP) Blood:water partition coefficient (pg/liter in blood/pg/liter in water) Cardiac output (liters blood/hr) Concentration in arterial blood (fig/liter) Concentration in mixed venous blood Wliter) First order elimination rate constant (/hr) Michaelis-Menten constant for enzymatic reaction (pg/liter blood) Maximum enzymatic reaction rate (pg/hr)

(i) for compartments 1 f m r k

Liver compartment Fat compartment Poorly perfused compartment Richly perfused compartment Kidney compartment

Arterial blood flow to the compartment (liters/hr) Compartment volume (liters) Concentration in the compartment Wlited Amount of chemical in the compartment hi9 Amount of chemical metabolized within the compartment (pg) Concentration in blood exiting the compartment (pg/liter) Tissue:blood partition coefficient (pg/liter tissue/pg/liter in blood) For compartments blood

receiving

mixed,

arterial,

and portal

Summed blood flow to the compartment (liters/hr) Concentration in mixed blood (&g/liter) Portal blood flow to the compartment (liters/hr) Concentration in portal blood (pg/liter)

in

436

NICHOLS

FIG. 2. Respirometer-metabolism chamber used to collect toxicokinetic information from rainbow trout (after McKim and Goeden, 1982). A latex oral membrane sewed to the fish’s mouth separates the A and B chambers. Water inspired by fish is exhausted to the B chamber, allowing direct calculation of chemical extraction efficiency. A second membrane between the B and the C chambers limits chemical exposure to the head region. Standpipes in each chamber ehminate the development ofdifferent hydrostatic head pressures.

ET

AL.

flow exiting the richly perfused compartment. Portal blood flow to the kidney compartment was set equal to 0.6 times blood flow to the poorly perfused compartment, consistent with the observation that the trunk musculature in salmonids drains to the kidney via the caudal vein (Smith and Bell, 1975). Arterial and portal blood supplies to the liver and kidney were assumed to mix before entering each compartment. Special consideration was given to the fat tissue compartment because its volume and blood perfusion both are poorly known and have a large impact on model outputs. The volume of the fat compartment was estimated from total lipid content data for tissues from adult rainbow trout (Table 3). Close inspection of these data suggested that most fat is stored in association with “viscera” or in deposits along the back, just under the skin, and between muscle layers (“carcass”). Lean tissue contained 2.5% lipid by weight, while samples of fat from the abdominal cavity contained 8 1.6% lipid. The fat tissue content of the viscera and carcass were estimated from the relationship (tissue weight)(lipid

= (fat tissue content)(8 where

blood in perfused gill lamellae, was calculated as the ratio of oxygen consumption rate to the drop in oxygen concentration in water flowing through secondary lamellar channels (Erickson and McKim, 1990). Arterial blood flows to internal organs were modeled after those estimated by Barron et al. (1987) using radiolabeled microspheres, Blood flow to the richly perfused compartment4 (23.0%), expressed as a percentage of Qc, was calculated as the sum of arterial flows to the stomach, pyloric ceca. intestine, spleen, and gonads. Arterial blood flow to the poorly perfused compartment (60.0%) was calculated as the mean of values determined for white muscle by Randall and Daxboeck (1982) Neumann et al. (1983), and Barron et at’. (1987) plus 10% to account for blood flow to the fins, skin, and bones. Blood flow estimates obtained using radiolabeled microspheres generally result in underestimation of total blood flow to organs (liver and kidney) receiving portal blood because of the series arrangement of capillary beds. Total hepatic blood flow was therefore assumed to consist of an arterial supply equal to the value (2.9%) reported by Barron et al. (1987), and a portal supply equal to blood

4 The term “richly perfused” refers to the high level of volume-normalized blood flow to this compartment. Dividing blood flow as a percentage of cardiac output by compartment volume as a percentage of total volume results in a value of 0.23/0.063, or approximately 3.6. Calculated in the same way, the ratio of blood perfusion to compartment volume in the “poorly perfused” compartment is approximately 0.7.

content)

tissue weight

1.6) + (tissue weight - fat tissue content)(2.50),

and fat tissue content

TABLE

are expressed

2

PHYSIOLOGICAL PARAMETERSUSEDINAPHYSIOLOGICALLY BASED TOXICOKINETIC MODEL FOR ADULT RAINBOW TROUT"~~ Body weight (kg) Ventilation volume (liters water/hr) Effective respiratory volume (liters water/hr) Oxygen consumption rate (mg O,/hr) Cardiac output (liters blood/hr) Arterial blood flow to tissues (liters blood/hr)’

bw Q”

1.0 10.60

QW

7.20 63.0 2.07

V02

QC t :r

Qk Tissue group

volumes

(liters)

’ Abbreviations and symbols b Values given are for a 1 kg of physiological parameters to the text. ’ Calculated by multiplying of Qc times 2.07 liters blood/hr.

V Vf VtIl K vk

0.060 0.176 1.242 0.476 0.116 0.013 0.098 0.818 0.063 0.008

are as defined in Table 1. trout at 12°C. The scaling body weight is described in blood

flow as a percentage

A PB-TK TABLE TOTAL

MODEL

3

LIPID CONTENT OF TISSUES FROM ADULT RAINBOW TROUT” Tissue weight as a % of body weight b

Tissue Lean tissues Muscle Spleen Liver Gill Fatty tissues Kidney Brain Carcass Viscerad Adipose fat

Total lipid as a % of tissue wet weight

67.0’ 0.16 1.27 2.47

2.31 2.13 2.96 2.53 mean = 2.50

0.76 0.49 80.79 6.92

4.77 7.71 9.60 31.36 81.60

a Rainbow trout, weighing 750-1000 g, were dissected and the tissues kept on ice until analyzed. Lipids were extracted with methylene chloride and quantified microgravimettically. Values are means; N = 10 fish. ’ Determined from tissues used in the analysis of total lipid content, unless otherwise noted. ’ From Stevens (1968). d Includes stomach, intestines, pyloric caeca, spleen. gonads, and viscera-associated fat.

as percentages of total body weight, and lipid content is expressed as a percentage of tissue weight. Solving for the fat content of viscera, (6.92)(3

1.36) = (fat tissue content)(8 + (6.92

fat tissue content

of viscera

- fat tissue content)(2.50) = 2.52% of body weight,

= (fat tissue content)(E + (80.79

fat tissue content

I .6)

- fat tissue content)(2.50)

of carcass = 7.25% of body

FISH

437

Liver and kidney volumes as a percentage of total were measured directly from organs excised from large (7501000 g) rainbow trout (N = 9; mean + SD for liver = 1.26 + 0.50% and for kidney = 0.76 i 0.32%). The volume of the richly perfused compartment was estimated by summing values reported by Barron et al. (1987) for the stomach, intestines, pyloric caeca, gall bladder, spleen, and gonads. The volume of the poorly perfused compartment was calculated as the difference between total volume and the summed volumes of the other compartments. Scaling. Cardiac output was scaled directly to body weight, as suggested by Barron et al. (1987) for large (loo1000 g) rainbow trout. Ventilation volume was scaled to body weight using the relationship, Q” = a:. Wp, where CY = 10.6, p = 0.75, and W is the weight of the fish in kg (Adolph, 1949). Compartment volumes were scaled directly to body weight. Extrabranchial elimination, including metabolism. The model incorporates both first-order and nonlinear (Michaelis-Menten) elimination from the liver compartment (see Appendix). For most compounds, however, metabolic and excretory pathways in fish are poorly known. In the present study, rate and capacity parameters were initially set at zero and the exposure data examined for evidence to the contrary. Chemical partitioning. Blood:water and tissue:blood equilibrium partition coefficients were estimated using an in vitro vial equilibration method (Gargas et al., 1989), modified for use with fish (Hoffman and Bertelsen, 1990). Partition coefficients for PCE are shown in Table 4, along with in vivo partitioning information from fish exposed to near steady state (264 hr). Chemical partitioning to the richly perfused compartment was assumed to be identical to that for the liver compartment.

Model Validation

1.6)

and carcass, (80.79)(9.60)

FOR

weight,

and summing resulted in a value of 9.8% of body weight. Assuming al1 tissues to have the same specific gravity, this value was then used to estimate fat volume as a percentage of total body volume. Blood perfusion to the fat compartment (8.5%) was calculated as the difference between Q, and blood flow already accounted for by the other compartments. This value is similar to that used for the poorly perfused compartment, when normalized for tissue volume.

Animals. Rainbow trout, weighing 600 to 1000 g, were held in a 4-f&diameter circular fiberglass tank and maintained as previously described (McKim et al., 1985). Fish that required surgical preparation were fasted for 24 hr before operating. All trout were acclimated to the experimental temperature (12°C) for several weeks before use. Both free-swimming and chambered exposures were conducted using unfiltered Lake Superior water. Chambered exposures. The accuracy of model simulations was evaluated by exposing rainbow trout to PCE in water in fish respirometer-metabolism chambers (Fig. 2), permitting direct in viva measurement of chemical absorption, distribution, and elimination (McKim and Goeden, 1982; McKim et al., 1985). Fish were cannulated from the dorsal aorta to permit serial sampling of blood. A latex membrane sewed to the fish’s mouth separated inspired and expired water Bows, allowing calculation of chemical extraction efficiency. A second membrane limited exposure to the head region.

438

NICHOLS

ET

TABLE

AL. 4

PARTITIONINGOFPENTACHLOROETHANETOBLOODANDTISSUESFROM

RAINBOWTROUT In vitro

In vivo

partitioninga Blood:water partition coefficient Tissue:blood partition coefficients Fat Kidney Liver Richly perfused Poorly perfused (muscle)

partitioning

Pbw

25.82

Pf S pi p, PII?

85.80 + 28.80 3.15 i 1.17 2.83 f 0.73 c 3.22 i

’ Equilibrium partition coefficients determined by Hoffman and Bertelsen b Determined by analysis of tissues from fish exposed to 150 +g PCE/liter pg PCE/liter) for 264 hr. Values are means f SD, N = 5 fish. ‘Set equal to partition coefficients for liver.

Fish were placed in the exposure chambers following surgery and allowed 24 hr to recover before dosing. Saturated stock solutions of PCE were generated using a liquid-liquid saturator (M. Kahl, AScI Corp., ERL-Duluth, Duluth, MN, personal communication) and pumped into a mixing cell to obtain desired PCE concentrations. Each chamber received approximately 500 ml exposure water/ min. Nominal exposure levels (15.0, 150.0, and 1500.0 pg/liter) were chosen to provide an opportunity to observe saturation of metabolic pathways, while remaining low enough to have no measurable effect on the physiology of test animals. The highest PCE concentration tested was approximately 20% of the 96 hr LC50 value for fathead minnows (Ahmad et al.. 1984). Measured concentrations are reported as the means + SD. Blood (50 or 100 pliter) and water samples were collected concurrently. Cannulae were filled between sampling periods with heparinized (250 lU/ml) Cortland physiological saline. The total of all blood samples was limited to 10% of the estimated blood volume. QY and I/O* were monitored automatically every 15 min throughout each experiment. Free-swimming exposures. A group of four trout were exposed to 150 pg PCE/liter for 168 hr under free-swimming conditions in an effort to load fish with chemical to near steady state. Fish were transferred directly from holding tanks to a 3-f&diameter circular fiberglass tank receiving 2.5-liter exposure water/min. Water and tissue analysis. Chemical analysis was limited to determination of parent chemical concentrations in water, blood, and tissues. Water and blood were extracted by pipetting samples directly into hexane. At the end of each exposure, fish were killed by anesthetic overdose and dissected to obtain the liver and kidney, and samples of fat and muscle. Tissues were extracted by homogenizing in ice-cold sodium sulfate and hexane. Hexane extracts

+

7.05

2.14

20.66

k

b 4.45

126.34 f 31.60 5.87 + 1.93 2.86 f 0.69 2.91 i 0.27

( 1990). Values are means Z! SD; N = 6 fish. (measured concentration = 187.1 t 10.0

were analyzed on a Hewlett Packard 5880A gas chromatograph equipped with a 63Ni electron capture detector and a 1.83 m X 2 mm i.d. glass column packed with 3% OV-101 SO/l00 mesh Gas Chrom Q. Inlet, oven, and detector temperatures were 250,65. and 3OO”C, respectively. The carrier gas was 95% argon/5% methane, flowing at 30 ml/min. Hexane:sample volume ratios were adjusted for each sample type and exposure level to provide extracts that could be injected directly onto the gas chromatograph. The limit of detection using this method was 0.5 ,ug PCE/ kg. The mean coefficients of variation of duplicate water, blood, muscle. fat, liver, and kidney samples were 0.02, 0.05, 0.30, 0. LO, 0.12, and 0.34. respectively. Sample spike recoveries ranged from 90 to 110% and were not used to correct the data.

RESULTS

PCE Exposure Studies Studies with rainbow trout exposed to PCE provided three types of kinetic information: (1) the time course of chemical concentration in arterial blood, (2) the time course of chemical extraction efficiency from inspired water, and (3) chemical concentrations in blood and tissuesat termination. These data were subsequently used to evaluate the accuracy of model simulations. Blood time course data from fish exposed to 150 pugPCE/liter are plotted in Fig. 3 using a numerical scale to illustrate the kinetics of accumulation. Ex-

A PB-TK

MODEL

FOR

439

FISH

HOURS

FIG. 3. Time course (48 hr) for pentachloroethane concentration in arterial blood. In this and the following figures, model simulations were prepared by setting the input concentration to that used in the corresponding validation study. The model simulation is shown as a solid line, the observed values as individual points. The data in this figure were obtained from fish exposed to 150 +g PCE/liter (measured water concentration = 140.3 + 1.52 pg PCE/ liter). Each point represents the mean t SD; N = 4 fish.

posure to 1500, 150, or 15 pg PCE/liter resulted in corresponding changes in the magnitude of blood concentrations without any apparent change in uptake kinetics (Fig. 4). Chemical extraction efficiency, calculated as (Gnsp - Gxp)/Gnsp 2 declined during the course of each experiment but was always greater than 0% at 48 hr, usually ranging from 15 to 25% (Fig. 5). Chemical residues in blood and tissues, expressed as blood:water and tissue:blood concentration ratios, are presented in Table 5.

Evaluation

FIG. 4. Dependence of pentachloroethane accumulation in arterial blood on inspired chemical concentration. Model simulations are shown as solid lines, the observed values as individual points. The data were obtained from fish exposed to 1500, 150, or 15 fig PCE/liter (measured water concentrations were 1648.0 f 79.0, 140.3 + 15.2, and 17.3 k 3.0 pg PCE/liter, respectively). Each point represents the mean: N = 4 fish.

that fish had not reached steady state or that a substantial amount of chemical was being cleared systemically. To determine whether fish had reached steady state, trout were loaded with PCE for 1 week under free-swimming conditions, then transferred to respirometermetabolism chambers where they were dosed for an additional 96 hr (264 hr total). Extrac-

1w 1

of ModeI Performance

The fish PB-TK model was initially used to simulate chemical extraction efficiency. With metabolic rate and capacity parameters set at zero, the model accurately predicted the extraction of PCE from inspired water (Fig. 5). Extraction efficiency data reflect both chemical uptake by fish and, at steady state, the extent to which the chemical is eliminated by metabolism or other systemic (extrabranchial) routes. The appearance of relatively high extraction efficiencies at 48 hr suggested either

FIG. 5. Time course (48 hr) for extraction of pentachloroethane from inspired water. Chemical extraction efficiency was calculated as (C,,,, - CeXp)/Cinsr,. The model simulation is shown as a solid line, the observed values as individual points. The data were obtained from fish exposed to 150 Fg PCE/liter (measured water concentration = 140.3 f 15.2 pg PCE/liter). Each point represents the mean f SD; N = 4 fish.

440

NICHOLS

ET AL.

tion efficiency at 264 hr was close to 0% (Fig. 6), indicating that fish were at or near steady state and that systemic elimination of PCE was small. The extended time course for extraction of PCE from water was accurately simulated by the model. Chemical concentrations in arterial blood were generally close to model predictions (Figs. 3 and 4); however, the model tended to underestimate levels observed during the first 4 hr of exposure, while overestimating data from later sampling periods. Overestimation of concentrations observed at later sampling times may have been due to the effects of blood sampling. Blood hematocrit values in chambered trout decreased approximately 20% during the course of each experiment. A decrease in hematocrit, particularly if accompanied by a decrease in the concentration of plasma lipids, could reduce the chemical binding capacity of blood and cause a progressive decline in the rate of chemical uptake at the gills. Underprediction of chemical concentrations at early time points may have been due to overestimation of cardiac output, or to deviation from the assumption of flow-limited tissue distribution, such that venous blood returning to the gills contained more chemical than predicted.

0

FIG. 6. Time course (264 hr) for extraction of pentachloroethane from inspired water. The model simulation is shown as a solid line, the observed values as individual points. Values from 0 to 48 hr are from Fig. 5, while later points are from fish loaded with PCE for 1 week under free-swimming conditions before being placed in respirometer-metabolism chambers. Each point represents the mean i SD: N = 4 fish.

Chemical residues in blood and tissues were converted to blood:water and tissue:blood concentration ratios to adjust for differences in exposure concentration and to provide a basis for comparison with in vitro partitioning estimates (Table 5). Blood:water and tissue: blood concentration ratios were similar in chambered fish exposed to 15 or 1500 pg PCE/ liter, suggesting that PCE accumulated in

TABLE PENTACHLOROETHANE

RESIDUESINBLOODANDTISSUESFROMRAINBOWTROUT,EXPRESSEDASBLOOD:WATER AND TISSUE:BLOOD CONCENTRATION RATIOS Free-swimming 264 hr

Blood:water Tissue:blood Fat Kidney Liver Muscle

5

20.66 ?I 4.45 126.34 5.87 2.86 2.91

f 31.60 f 1.93 -t 0.69 ? 0.27

exposuresa

48-hr

59.51 4.04 2.11 1.74

exposures*

15.0 fig/liter

48 hr 17.96 i

chambered

3.75

10.13 f

i 11.74 -t 1.82 + 0.61 + 0.45

a Values are means f SD; N = 4 (48 hr) or 5 (264 hr) fish. Measured and 187.1 + 10.0 (264 hr) Fg PCE/liter. b Values are means F SD; N = 4 fish. Measured water concentrations liter, respectively.

119.98 6.76 4.48 4.52 water were

f + i +-

1500.0 fig/liter

1.53

9.56 YE 2.03

24.16 4.20 0.70 3.92

concentrations

91.74 4.58 4.58 2.53 were

+- 20.91 f 1.60 f 1.22 f 1.70

190.8 f 24.4 (48 hr)

17.3 + 3.0 and 1648.0

+- 79.0 pg PCE/

A PB-TK

MODEL

blood and tissues in proportion to the exposure concentration. A comparison of free-swimming and chambered animals exposed for 48 hr suggests that PCE concentrations in blood were reduced in chambered fish for the reasons discussed above. This in turn accounts for observed differences in tissue:blood concentration ratios (free-swimming values being lower) because blood is the denominator in this ratio. Generally, however, the pattern of accumulation in free-swimming animals was similar to that of chambered animals, the rank order of concentration in tissue being fat $ kidney > liver = muscle. Blood:water and tissue:blood concentration ratios increased in free-swimming fish between 48 and 264 hr, the fat:blood ratio more than for other tissues. The model predicts that PCE concentrations in blood increase rapidly during the first 48 hr of exposure, followed by slow accumulation (Fig. 7). In contrast, the fat compartment is both poorly perfused and has a high capacity for PCE. These factors combine to retard the kinetics of uptake into this compartment such that the proportional increase in PCE concentration between 48 and 264 hr is greater in fat than it is in blood or other tissues. Although insufficient to assess the kinetics of chemical accumulation in fat and blood, residue levels at the termination of free-swimming exposures were in good agreement with those predicted by the model.

DISCUSSION The objective of this research was to develop a PB-TK model for fish that accurately predicts the uptake and disposition of waterborne organic chemicals, while remaining simple enough to be applied to problems in aquatic toxicology and environmental risk assessment. PB-TK models for fish have been published previously (Zaharko et al., 1972; Bungay et al., 1976). The major contribution of the present work was to combine an established PBTK model structure with a physiologically based description of fish gills, thereby provid-

FOR

441

FISH

HOURS

FIG. 7. Accumulation of pentachloroethane in fat and arterial blood. Model simulations are shown as solid lines, the observed values as individual points. Chemical residues were measured in fat (solid circles) and blood (open circles) from fish exposed to 150 pg PCE/liter for 48 or 264 hr (measured water concentrations were 190.8 + 24.4 and 187.1 + 10.0 pg PCE/liter, respectively). Each pomt represents the mean f SD; N = 4 fish.

ing a means of modeling an environmentally relevant route of exposure (inhalation). Model performance was evaluated by exposing rainbow trout to PCE in water in fish respirometer-metabolism chambers. Parameterized with physiological information from the literature, the model accurately predicted the uptake and disposition of PCE in trout. The selection of PCE for use in developing a PB-TK model for fish was based upon four considerations. First, the PB-TK model developed by Ramsey and Andersen ( 1984) and adapted in this paper for use with fish, has been successfully employed to predict the toxicokinetics of PCE in rats (Gargas and Andersen, 1989). Second, it was anticipated that PCE would exhibit rapid uptake and accumulation kinetics, making it possible to obtain data from fish at or near steady state. Third, because PCE is volatile, it was possible to obtain necessary chemical partitioning estimates using an established in vitro procedure (Gargas et al., 1989). Finally, the physical and chemical properties of PCE, in particular its low molecular volume, moderate lipophilicity, and absence of charged groups, made it likely that chemical partitioning between blood and tissues was flow-limited. Compounds that exhibit significant diffusion limitation may be mod-

442

NICHOLS

eled within a physiological framework by incorporating estimates for diffusion parameters. This, however, adds to the complexity of the analysis. Physiological inputs to PB-TK models are available in the literature for many higher vertebrates, but information for most aquatic vertebrates is scarce. Rainbow trout are a notable exception. The physiology of trout has been investigated in detail and information is available for ventilation volume, oxygen consumption rate, cardiac output, organ blood flows as a percentage of cardiac output, and major organ and tissue volumes (see Methods: Model Development and Parameterization) as well as bile flow rates and biliary excretion rates (Schmidt and Weber, 1973). In addition, the physiology of trout has been examined with respect to the effects of various environmental parameters, including temperature (Barron et al., 1987) and dissolved oxygen content (Randall, 1982). The development of a successful PB-TK model must be viewed as an iterative process. As each compound is tested, a sensitivity analysis is undertaken to determine the relative contribution of individual parameters to model performance. This information can then be combined with the modeler’s confidence in the parameter estimates to determine whether and how the model should be amended, and which of the parameters require further investigation. Of the parameters examined to date, those which had large impacts on model outputs included ventilation volume, blood perfusion to the fat compartment, and chemical partitioning between water and blood and between blood and tissues. Ventilation volume is the best known of these input parameters because it can be measured directly. The gill description developed by Erickson and McKim (1990) predicts that chemical uptake will vary directly with ventilation volume when the capacity of blood to take up a compound at the gills exceeds the amount of chemical presented in inspired water (see Appendix). Empirical data from rainbow trout suggest that this condition is usually

ET

AL.

satisfied when the log of the octanol:water partition coefficient for the compound (log K,,) exceeds 3.0 (McKim et al., 1985). The high initial extraction efficiencies observed for PCE, which has a log K,, of about 3.6, confirmed that uptake of this compound was ventilation-limited. In contrast to ventilation volume, blood perfusion to the fat compartment is poorly known. Fat is generally thought to be poorly perfused; however, most of the fat in fish is distributed throughout the muscle and under the skin and is not accessible to analysis using radiolabeled microspheres. The kinetics of PCE uptake and accumulation in trout were accurately simulated when volume-normalized perfusion of the fat compartment was comparable to that of the poorly perfused (muscle) compartment. It is possible, however, that a compound initially taken up by the muscle distributes to the fat by local diffusion. The need for accurate chemical partitioning information cannot be overstated; partition coefficients drive the model, determining chemical uptake rate at the gills, steady-state concentrations in blood and tissues, and the overall kinetics of accumulation. Blood:water and tissue:blood partitioning estimates were obtained using an in vitro vial equilibration method. This method is limited, however, to use with volatile compounds. The use of PBTK models with nonvolatile compounds will require development and validation of other methods for estimating these partition coefficients. Of the parameters required to run a PB-TK model, those relating to chemical elimination are among the most difficult to obtain. By loading trout with PCE before exposing them in the chambers, it was possible to demonstrate that systemic elimination was small. We recognize that this will not always be the case. The metabolizing, secretory, and active transport systems of fish are well developed and have been shown to play a major role in determining chemical kinetics. Fish respirometer-metabolism chambers lend themselves to the task of understanding systemic elimination

A PB-TK

MODEL

by providing a means for conducting massbalance experiments. These experiments are usually performed using radiolabeled compounds, permitting quantitation of parent and “other” compounds. In this manner it has been possible to demonstrate that elimination has occurred and identify the organ(s) to which it is localized (McKim and Health, 1983; McKim et al., 1986; Bradbury et al., 1986). The incorporation of xenobiotic metabolism into PB-TK models presents a special challenge and has been the focus of a number of studies with mammals (Sat0 and Nakajima, 1979; Hilderbrand et al., 198 1; Lutz et al., 1984; Andersen et al., 1984; Gargas et al., 1986; Reitz et al., 1988). Anticipating that fish models will eventually be used to describe the kinetics of compounds that undergo significant metabolism, consideration should be given to the choice of trial compounds, initially favoring those for which metabolic and other elimination pathways are well characterized and easily monitored. Dose estimates in fish are difficult to obtain because chemical uptake at the gills varies as a complex function of physical, chemical, and physiological factors (McKim et al., 1985; Gobas and MacKay, 1987; Erickson and McKim, 1990). To date, a large amount of acute toxic effects data have been collected for a variety of fish species. Lacking a means of estimating dose to the animal, these data are usually correlated with the exposure concentration, which provides no explanation for observed differences in species sensitivity. A PBTK model for these fish species could contribute greatly to our understanding of the effects database by allowing comparisons to be made on a true dose basis. PB-TK models for fish have the potential to be of use in risk assessment, comparative toxicology, carcinogenicity testing, and the study of metabolic transformation. The most important contribution of a PB-TK model is that it provides estimates of the time course of chemical concentration in specific tissues of interest. Thus, estimates of chemical concentration in muscle and associated fat could

FOR

443

FISH

be used in residue-based risk assessment for human consumption of fish. Alternatively, concentration estimates in the kidney or liver could be correlated with the appearance of metabolic products or with an increased incidence of neoplasia. As additional physiological and partitioning information becomes available it should be possible to extrapolate toxicokinetic information between fish species and from fish to higher vertebrates. Future work must therefore include studies designed to validate the model for use with other fish species, classes of chemicals, and routes of exposure. APPENDIX

Physiological model. The PB-TK model for fish (Fig. 1) consists of a set of simultaneous differential equations which describe the rate of change of the amount of chemical within each compartment. For any given chemical, one need only change chemical-specific partitioning parameters and the input concentration to obtain computer simulations of chemical uptake and disposition. Using symbols and abbreviations presented in Table 1, general equations describing the kinetics of a chemical within a compartment (i) receiving arterial blood may be written as dAJdt = Qi(Ca* - Cvi)

(1)

Cl = AJ Vi

(2)

Cvi = CJPi.

(3)

From mass-balance considerations, the rate of change of the amount of chemical in a compartment is equal to the blood flow rate times the arterial-to-venous difference in chemical concentration (Eq. (1)). Chemical flux was considered to be flow-limited and the chemical assumed to distribute homogeneously within each compartment. Integrating Eq. (1) yields an equation having the general form t

Aj =

s0

dAi.

(4)

444

NICHOLS

This system of equations must be expanded for compartments receiving both arterial and portal blood because a fraction of the compound originally contained in portal blood has been removed during prior perfusion of another (upstream) compartment:

dAi/dt = Qmi(Cmi - Cvi)

(5)

Qmi= Qi + Qpi (6) Crni= l(QdQmXG>l+ L(QdQmJ(CpJl(7) In practice, the variable Q,i is defined as equal to all or a fraction of arterial flow (Qi) to the upstream compartment. Similarly, Cpi is set equal to the chemical concentration in venous blood exiting the upstream compartment. Chemical concentration in mixed blood is calculated by summing the flow-weighted contributions of arterial and portal blood

ET

of equation, where V,,, (pg/hr) is the maximum rate of reaction and K, (pug/liter blood) equals the chemical concentration in venous blood (Cvi) leaving the compartment at onehalf Vmax:

dA,Jdt

dAJdt = Qi * (Cati - Cvi) - dAJdt.

(8)

Referenced to chemical concentration in venous blood exiting the eliminating compartment, first-order elimination may be described by the equation

dAJdt

= Kfc * C,i * Vi.

(9)

Saturable pathways, which may include metabolism or active secretion to urine or bile, can be described by a Michaelis-Menten-type

= V,,,*

CviItKm

+

Gil-

(10)

Fish gill description. A general expression for chemical flux at fish gills (I;,, pg/hr) was derived by Erickson and M&m (I 990) and relates flux to an exchange coefficient (k,) and the difference in chemical activities in venous blood and inspired water: Fg

=

kx * (Cinsp

-

cven/pbw>-

(11)

Chemical flux can also be related to water and blood flow rates by simple mass-balance considerations: Fg =

(Eq. (7)). Extrabranchial

elimination. Chemical elimination, by metabolism or secretion to the external environment, probably occurs to a limited extent in all tissues, but is usually modeled in physiologically based models as a process localized to one compartment. The choice of this compartment and of the kinetics of elimination (zero order, first order, or nonlinear) varies with species, the chemical in question, and a host of other variables. In general, however, elimination may be modeled by incorporating an expression for the rate of metabolism (aTA,,/dt, pg/hr) into the massbalance equation for the eliminating compartment (Ramsey and Andersen, 1984; Gargas et al., 1986):

AL.

=

Qw* (Gsp -Cexp) Qc- CC,, - GA

(12)

Equating Eqs. ( 11) and (12), an expression is obtained for the exchange coefficient (k,) in terms of blood and water flows and chemical concentrations: k = x

Qw.(Gasp Cinsp

-

Qc-(Cm

=

-

Gnsp

-

- Cap)

cven/pbw GJ

cven/pbw

(13) ’

In the absence of diffusion limitations, chemical exchange at fish gills is assumed to proceed to equilibrium. For countercurrent exchange this means that arterial blood equilibrates with inspired water (blood flow limited exchange, Eq. (14)), or that venous blood equilibrates with expired water (ventilation limited exchange, Eq. (15)): cart =

Cinsp’

Pbw

(14)

or cex,

=

~v,,,lpb,~

(15)

When the equilibration of inspired water with arterial blood limits chemical exchange, k, is

A PB-TK

solved for by substituting (13):

MODEL

Eq. (14) into Eq.

k, = 0,. Pbw,

(17)

Q,.

Substituting Eqs. (16) and (17) into Eq. (1 l), a general expression is derived for chemical flux at fish gills, in which k, equals the flow term that is limiting: Fg

=

MINQ,,

Qc-PbvJ

’ fCinsp

- cv,,/pbw).

t18)

Dilrltion correction for the exposure chambers. Becausewater entering the A side of the exposure chambersor exhaustedby ventilation to the B sideis diluted by water already present in each compartment, an instantaneous measurement of chamber concentration may not accurately represent chemical concentrations in incoming or exhausted water. This in turn results in an error in calculation of chemical extraction efficiency, the magnitude of which isgreatestwhen the chemical is first introduced or when the flow of the chemical is terminated. To correct for the effect of dilution, an exponential replacement function must be introduced, as shown for inspired water, Gnsp

= Ci, * (1 - e@‘),

(19)

or for chemical concentration in the B side of the exposure chamber (cb), cb = cexp- ( 1 - esk’).

(20)

In each case,the exponent k equalswater flow rate to the chamber (liters/hr) divided by the chamber volume (liters).

ACKNOWLEDGMENTS This work was sponsored by the Air Force Office of Scientific Research, Air Force Systems Command, U.S. Air Force, under Grant AFOSR-ISSA-89-0060. The authors thank Gregory Lien, Alex Hoffman, and Patrick

445

FISH

Tierney for their technical assistance, and Frank Dr. Philip Cook for reviewing the manuscript.

(16)

Similarly, when the equilibration of venous blood with expired water limits chemical exchange, Eq. (13) reduces to kx =

FOR

Stay and

REFERENCES

ADOLPH, E. F. (1949). Quantitative relations in the physiological constitutions of mammals. Science 109, 579585. AHMAD, N., BENOIT, D., BROOKE, L., CALL, D., CARLSON, A.: DEFOE, D., HUOT, J., MORIARITY, A., RICHTER, J., SHUBAT, P., VEITH, G., AND WALBRIIXE, C. (1984). Aquatic toxicity tests to characterize the hazard of volatile organic chemicals in water: A toxicity data summary. EPA Project Summary, EPA-6OO/S3-84-009. ANDERSEN, M. E., GARGAS, M. L., AND RAMSEY, J. C. (1984). Inhalation pharmacokinetics: Evaluating systemic extraction, total in viva metabolism, and the time course of enzyme induction for inhaled styrene in rats based on arterial blood: Inhaled air concentration ratios.

Toxicoi. Appt. Pharmacol. 73, 176- 187. ANDERSEN, M. E., CLEWELL, H. J., GARGAS, M. L., SMITH, F. A., AND REITZ, R. H. (1987). Physiologically based pharmacokinetics and the risk assessment process for methylene chloride. Toxicol. A&. Pharmacol. 87, 185-205. BARRON, M. G., TARR, B. D., AND HAYTON, W. L. (1987). Temperature-dependence of cardiac output and regional blood flow in rainbow trout, Salmo gairdneri Richardson. J. Fish Biof. 31, 735-744. BRADBURY, S. P., COATS, J. R., AND MCKIM, J. M. (1986). Toxicokinetics of fenvalerate in rainbow trout (Salmo

gairdneri). Environ. To.xicol. Chem. 5, 567-576. BRADBURY, S. P., HENRY, T. R., NIEMI. G. J., CARLSON, R. W., AND SNARSKI, V. M. (1989). Use of respiratorycardiovascular responses of rainbow trout (Salmo gairdneri) in identifying acute toxicity syndromes in fish: Part 3. Polar narcotics. Environ. Toxicol. Chem. 8,241261. BUNGAY,~. M., DEDRICK, R. L., AND GUARINO, A. M. (1976). Pharmacokinetic modeling of the dogfish shark (Squalus acanthias): Distribution and urinary and biliary excretion of phenol red and its glucuronide. J. Phar-

macokinet. Biopharm. 4(5), 377-388. CLEWELL, H. J., AND ANDERSEN, M. E. (1985). Risk assessment extrapolations and physiological modeling. To.xicol. Ind. Health l(4). 11 I- 13 1. ERICKSON, R. J.. AND MC&M, J. M. (1990). A simple flow-limited model for exchange of organic chemicals at fish gills. Environ. Toxicol. Chem. 9, 159-165. GARGAS, M. L., ANDERSEN, M. E., AND CLEWEU, H. J. (1986). A physiologically based simulation approach for determining metabolic constants from gas uptake data.

Toxicol. Appl. Pharmacol. 86, 34 l-352. GARGAS, (1986).

M. L., CLEWELL, H. J., AND ANDERSEN. Metabolism of inhaled dihalomethanes

M. E.

in viva:

446

NICHOLS

Differentiation of kinetic constants for two independent pathways. Toxicol. Appl. Pharmacol. 82, 21 l-223. GARGAS, M. L., AND ANDERSEN, M. E. (1989). Determining kinetic constants of chlorinated ethane metabolism in the rat from rates of exhalation. Toxicol. Appl. Pharmacol. 99,344-353. GARGAS, M. L., BURGESS,R. J., VOISARD, D. E., CASON, G. H., AND ANDERSEN, M. E. (1989). Partition coefficients of low-molecular-weight volatile chemicals in various liquids and tissues. Toxicol. Appl. Pharmacol. 98,87-99. GERLOWSKI,L. F., AND JAIN, R. K. (1983). Physiologically based pharmacokinetic modeling: Principles and applications. J. Pharm. Sci. 72, 1103-l 126. GOBAS, F. A. P. C., AND MACKAY, D. (1987). Dynamics of hydrophobic organic chemical bioconcentration in fish. Environ. Toxicol. Chern. 6, 495-504. HILDERBRAND, R. L., ANDERSEN, M. E., AND JENKINS, L. J., JR. (198 1). Prediction of in vivo kinetic constants for metabolism of inhaled vapors from kinetic constants measured in vitro. Fundam. Appl. Toxicol. 1,403-409. HIMMELSTEIN, K. J., AND LUTZ, R. J. (1979). A review of the applications of physiologically based pharmacokinetic modeling. J. Pharmacokinet. Biopharm. 7, 127145. HOFFMAN, A. D., BERTELSEN, S. L., AND GARGAS, M. L. (1990). A gas equilibration method jbr determination of in vitro partition coeficients in Jish. Report, USEPA, Environment Research Laboratory-Duluth, Duluth, MN. KARARA, A. H., AND HAYTON, W. L. (1984). Pharmacokinetic model for the uptake and disposition of di-2ethylhexyl phthalate in sheepshead minnow (Cyprinodon variegatus). Aqzrat. Toxicol. 5, 181-195. KING, F. G., DEDRICK, R. L.. COLLINS. J. M., MATTHEWS, H. B., AND BIRNBAUM, L. S. (1983). Physiological model for the pharmacokinetics of 2,3,7,8-tetrachlorodibenzofuran in several species. Toxicol. Appl. Pharmacol. 67,390-400. LUTZ, R. J., DEDRICK, R. L., MATTHEWS, H. B., ELING, T. E.. AND ANDERSON, M. W. (1977). A preliminary pharmacokinetic model for severalchlorinated biphenyls in the rat. Drug Metab. Dispos. 5(4), 386-396. LUTZ, R. J., DEDRICK, R. L., TUEY, D., SIPES,I. G.: ANDERSON,M. W., AND MATTHEWS, H. B. (1984). Comparison of the pharmacokinetics of several polychlorinated biphenyls in mouse, rat, dog, and monkey by means of a physiological pharmacokinetic model. Drug Metab. Dispos. 12(5), 527-535. MCKIM, J. M., AND GOEDEN, H. M. (1982). A direct measure of the uptake efficiencyof a xenobiotic chemical across the gills of brook trout (Salvelinus fontinalis) under normoxic and hypoxic conditions. Camp. Biochem. Physiol. C 12, 65-14. MCKIM, J. M.. AND HEATH, E. M. (1983). Dose determinations for waterborne 2:5,2’,5’-[14C]tetrachlo-

ET AL. robiphenyl and related pharmacokinetics in two species of trout (Salmo gairdneri and Salvelinus forztinalis): A mass-balance approach. Toxicol. Appl. Pharmacol. 68, 177-187. MCKIM, J. M., SCHMIEDER, P., AND VEITH, G. (1985). Absorption dynamics of organic chemical transport across trout gills as related to octanol-water partition coefficient. Toxicol. Appl. Pharmacol. 77, l-10. MCKIM, J. M., SCHMIEDER, P. K., AND ERICKSON, R. J. (1986). Toxicokinetic modeling of [‘%I]pentachlorophenol in the rainbow trout (Salmo gairdneri). Aquat. To.xicol. 9, 59-80. MCKIM, J. M., SCHMIEDER,P. K., NIEMI, G. J., CARLSON, R. W., AND HENRY, T. R. (1987a). Use of respiratorycardiovascular responses of rainbow trout (Salmo gairdneri) in identifying acute toxicity syndromes in fish: Part 2. Malathion, carbaryl, acrolein, and benzaldehyde. Environ. Toxicol. Chem. 6, 3 13-328. MCKIM, J. M., SCHMIEDER, P. K., CARLSON, R. W., HUNT, E. P., AND NIEMI. G. J. (1987b). Use of respiratory-cardiovascular responses of rainbow trout (Salmo gairdneri) in identifying acute toxicity syndromes in fish: Part 1. Pentachlorophenol, 2,4-dinitrophenol, tricaine methanesulfonate, and 1-octanol. Environ. Toxicol. Chem. 6, 295-312. NEELY, W. B., BRANSON,D. R., AND BLAU, G. E. (1974). Partition coefficient to measure bioconcentration potential of organic chemicals in fish. Environ. Sci. Tech~~01.8(13), 1113-l 115. NEELY, W. B. (1979). Estimating rate constants for the uptake and clearance of chemicals by fish. Environ. Sci. Technol. 13(12), 1506-1510. NEUMANN, P. HOLETON, G. F., AND HEISLER, N. (1983). Cardiac output and regional blood flow in gills and muscles after exhaustive exercise in rainbow trout (Salmo gairdneri). J. Exp. Biol. 105, 1-14. NIH (1984). Use of small fish species in carcinogenicity testing. NCI Monograph 65, NIH 84-2653. POWERS, D. A. (1989). Fish as model systems. Science 246,352-357. RAMSEY, J. C., AND ANDERSEN, M. E. (1984). A physiologically based description of the inhalation pharmacokinetics of styrene in rats and humans. Toxicol. Appl. Pharmacol. 73, 159-175. RANDALL, D. J. (1982). The control of respiration and circulation in fish during exercise and hypoxia. J. Exp. Biol. 100,275-288. RANDALL, D. J.. AND DAXBOECK, C. (1982). Cardiovascular changes in the rainbow trout (Salmo gairdneri Richardson) during exercise. Canad. J. Zool. 60, 11351140. REITZ, R. H., MENDRALA, A. L., PARK, C. N., ANLIERSEN, M. E., AND GUENGERICH, F. P. ( 1988). Incorporation of in vitro enzyme data into the physiologically-based pharmacokinetic (PB-PK) model for methylene chlo-

A PB-TK

MODEL

ride: Implications for risk assessment. Toxicol. Lett. 43, 97-l 16. ROWLAND, M. (1985). Physiologic pharmacokinetic models and interanimal species scaling. Pharmacol.

Ther. 29, 49-68. SATO, A., AND NAKAJIMA, T. (1979). A vial-equilibration method to evaluate the drug-metabolizing enzyme activity for volatile hydrocarbons. Toxicol. Appl. Pharrnacol. 47, 4 l-46. SCHMIDT, D. C., AND WEBER, L. J. (1973). Metabolism and biliaty excretion of sulfobromophthalein by rainbow trout (Salmo gairdneri). J. Fish. Res. Board Can. 30, 1301-1308. SMITH, L. S., AND BELL, G. R. (1975). A Practical Guide to the Anatomy and Physiology of Pacific Salmon. Miscellaneous Special Publication 27, Department of the

FOR

447

FISH

Environment. Canada.

Fisheries

and Marine

Service,

Ottawa,

STEHLY, G. R., AND HAYTON, W. L. (1989). Disposition of pentachlorophenol in rainbow trout (Sulmo guirdnerr]: Effect of inhibition of metabolism. Aquat. Toxicol. 14,l 131-14s. STEVENS, E. D. (1968). The effect of exercise on the distribution of blood to various organs in rainbow trout. Camp. Biochem. Physiol. 25, 615625. ZAHARKO, D. S., DEDRICK. R. L., AND OLIVERIO, V. T. (1972). Prediction of the distribution of methotrexate in the sting rays Dasyatidae sabina and sayi by use of a model developed in mice. Comp. Biochem. Physiol. A 42, 183-194.