TOXICOLOGICAL SCIENCES 77, 219 –229 (2004)
DOI: 10.1093/toxsci/kfh032
Advance Access publication December 2, 2003
A Physiologically Based Toxicokinetic Model for Dietary Uptake of
Hydrophobic Organic Compounds by Fish
II. Simulation of Chronic Exposure Scenarios
John W. Nichols,* ,1 Patrick N. Fitzsimmons,* and Frank W. Whiteman*
*Mid-Continent Ecology Division, National Health and Environmental Effects Research Laboratory, Office of Research and Development,
U.S. Environmental Protection Agency, Duluth, Minnesota, 55804
Received on June 4, 2003; accepted on October 23, 2003
A physiologically based toxicokinetic (PBTK) model for dietary
uptake of hydrophobic organic compounds by fish was used to simulate dosing scenarios commonly encountered in experimental and
field studies. Simulations were initially generated for the model compound [UL- 14C] 2,2ⴕ,5,5ⴕ-tetrachlorobiphenyl ([ 14C] PCB 52). Steadystate exposures were simulated by calculating chemical concentrations in tissues of the predator corresponding to an equilibrium
distribution between the fish and water (termed the bioconcentration
or BCF residue data set). This residue data set was then varied in a
proportional manner until whole-body chemical concentrations exhibited no net change for each set of exposure conditions. For [ 14C]
PCB 52, the proportional increase in BCF residues (termed the
biomagnification factor or BMF) required to achieve steady state in a
food-only exposure was 2.24, while in a combined food and water
exposure the BMF was 3.11. Additional simulations for the food and
water exposure scenario were obtained for a set of hypothetical
organic compounds with increasing log K OW values. Using gut permeability coefficients determined for [ 14C] PCB 52, predicted BMFs
increased with chemical log K OW, achieving levels much higher than
those reported in field sampling efforts. BMFs comparable to measured values were obtained by reducing permeability coefficients
within each gut segment in a log K OW– dependent manner. This
predicted decrease in chemical permeability is consistent with earlier
work, suggesting that dietary absorption of hydrophobic compounds
by fish is controlled in part by factors that vary with chemical log
K OW. Relatively low rates of metabolism or growth were shown to
have a substantial impact on steady-state biomagnification of chemical residues.
Key Words: physiologically based model; fish; dietary uptake.
Recent studies of dietary uptake by fish have focused on
factors that promote the uptake of hydrophobic compounds,
including the absorption of dietary lipid and reductions in meal
1
To whom correspondence should be addressed at U.S. Environmental
Protection Agency, Mid-Continent Ecology Division, 6201 Congdon Boulevard, Duluth, MN 55804. Fax: (218) 529-5003. E-mail: [email protected].
Toxicological Sciences vol. 77 no. 2 © Society of Toxicology 2004; all rights
reserved.
volume (Gobas et al., 1993a,b, 1999). The “digestion hypothesis” states that these processes tend to increase chemical
activity in the gut contents above that of the meal, potentially
resulting in biomagnification of chemical residues (i.e., lipidnormalized concentrations in fish greater than those of their
food). Support for this hypothesis has been obtained in studies
with several fish species. In feeding studies with guppies
(Poecilia reticulata) and goldfish (Carassius auratus), the feces:food fugacity ratio increased with chemical log K OW, attaining a maximum value of 4.6 (in guppies) for the pesticide
mirex (log K OW of 7.2; Gobas et al., 1993b). The intestinal
contents:food fugacity ratio in a natural population of white
bass (Morone chrysops) also increased with chemical log K OW,
attaining a value of about 2.2 for 2,2⬘,3,4,4⬘,5,5⬘-heptachlorobiphenyl (PCB 180, log K OW of 7.36; Russell et al., 1995). In
rainbow trout (Oncorhynchus mykiss) and rock bass (Ambloplites rupestris) exposed to 2,2⬘,4,4⬘,6,6⬘-hexachlorobiphenyl
(PCB 155, log K OW of 7.2), the chemical fugacity in chyme
following uptake of dietary lipid exceeded that of food by a
factor of 7 to 8 under both laboratory and field conditions
(Gobas et al., 1999).
Adopting the digestion hypothesis, the maximum extent to
which a compound can biomagnify in fish depends on the
feeding rate (F D), the fecal production rate (F F), and the gut
contents:organism chemical partitioning coefficient (K GB), according to the following relationship given by Gobas et al.
(1993a): F D/(F F K GB). However, the processes that actually
control chemical flux across the gastrointestinal epithelium,
and ultimately the extent to which chemicals accumulate in
fish, remain incompletely understood. Additional questions
pertain to the manner in which dietary uptake and biomagnification vary with attributes of the ingested compound (e.g.,
hydrophobicity and susceptibility to metabolic biotransformation) and the exposed animal (e.g., gastrointestinal anatomy
and physiology).
In a companion report, we described a physiologically based
toxicokinetic (PBTK) model for dietary uptake of hydrophobic
219
220
NICHOLS, FITZSIMMONS, AND WHITEMAN
organic chemicals by fish (Nichols et al., 2004). The gut
portion of this model consists of four compartments corresponding to the stomach, pyloric ceca, upper intestine, and
lower intestine. This structure provides for temporal and spatial
resolution of factors that control chemical uptake and elimination, and permits a mechanistic exploration of these factors by
allowing the modeler to adjust parameter values and modeling
assumptions for individual gut segments. The model was initially calibrated using data from rainbow trout that were fed a
single meal of fathead minnows (60 days old) contaminated
with [UL- 14C] 2,2⬘,5,5⬘-tetrachlorobiphenyl ([ 14C] PCB 52;
Nichols et al., 2001). Dietary exposures of environmental
concern are more likely, however, to occur over long periods of
time and involve many individual feedings. Accordingly, the
goals of the present study were as follows: (1) to simulate
different chronic exposure scenarios for [ 14C] PCB 52 and
compare modeled results to published data for PCB 52 (unlabeled); (2) to simulate long-term dietary exposures to a set of
hypothetical organic compounds and compare the results with
trends that have been reported with respect to chemical log
K OW; and (3) to use the model to investigate potential impacts
of biliary elimination, metabolic biotransformation, and
growth on chemical biomagnification in fish.
MATERIALS AND METHODS
PBTK model. A PBTK model for dietary uptake of hydrophobic compounds by fish is described in a companion report (Nichols et al., 2003).
Mass-balance differential equations were solved by numerical integration
using a commercial software package (ACSL; Aegis Technologies, Huntsville,
AL) to obtain a complete solution set for each time point.
[ 14C] PCB 52 dosing scenarios. The goal of this study was to use the
dietary uptake model to simulate dosing scenarios commonly encountered in
experimental studies and field exposures. Using [ 14C] PCB 52 as a model
compound, these scenarios were as follows: (1) chronic exposure to contaminated water; (2) chronic exposure to contaminated food; and (3) chronic
exposure to contaminated food and water.
Although in theory it would be possible to simulate a large number of
individual feedings over time periods lasting months to years, this approach
was determined to be impractical and unnecessary. Instead, the model was
used to solve for [ 14C] PCB 52 concentrations in trout tissues that would be
expected under steady-state conditions, given a defined exposure scenario. The
first step was to calculate the [ 14C] PCB 52 concentration in water expected
from an equilibrium distribution between the water and contaminated fathead
minnows from the initial feeding studies (Nichols et al., 2001). A fat:water
partitioning coefficient (P F:W) for [ 14C] PCB 52 was calculated as the product
of fat:blood (P F:B) and blood:water (P B:W) partitioning values given by Nichols
et al. (2004) for rainbow trout: (4500)(140.2) ⫽ 630,960. This value was
multiplied by the reported lipid content (3.3%) of fathead minnows to give a
whole-body bioconcentration factor of 20,820 (BCF; wet weight basis, defined
as the [ 14C] PCB 52 concentration in fish [ng/g]/[ 14C] PCB 52 concentration in
water [ng/ml]). The average measured [ 14C] PCB 52 concentration in fathead
minnows (1663 ng/g) was then divided by the BCF to give an estimated water
concentration of 0.08 ␮/L. This approach assumes that fat:water partitioning in
fathead minnows is similar to that in trout and ignores the contribution of [ 14C]
PCB 52 associated within nonlipid portions of the fish. Errors introduced by
these assumptions are probably small, however, given the likely similarity of
fat:water partitioning in both species (even as P F:B and P B:W vary) and the
overwhelming importance of lipid as a repository for hydrophobic chemicals in
fish.
The next step was to calculate the concentration of [ 14C] PCB 52 in each
trout tissue, assuming chemical equilibrium with an exposure water concentration of 0.08 ␮/L. This was accomplished by calculating tissue:water partitioning values (P T:W) for all tissues as products of tissue:blood (P T:B) and P B:W
values given in Table 2 of Nichols et al. (2004) and multiplying by 0.08. The
[ 14C] PCB 52 concentration in each tissue was then multiplied by the estimated
tissue volume to give the total mass of chemical. The results of these calculations are provided in Table 1. A whole-fish BCF can also be calculated by
summing the volume-weighted contributions of each tissue. Using data given
in Table 1, the whole-fish BCF for subadult trout used by Nichols et al. (2001)
is about 27,600, which is intermediate to values reported previously for
guppies (18,200) and goldfish (49,000; Connell and Hawker, 1988) and lower
than the BCF (200,000) estimated for large adult trout (Oliver and Niimi,
1985). Taken together, these calculations provide a set of individual tissue and
whole-body chemical concentrations for the predator (trout) and prey (fathead
minnows) that are in equilibrium with an aqueous [ 14C] PCB 52 concentration
of 0.08 ␮g/L.
Chronic exposures to [ 14C] PCB 52 were simulated by adjusting the mass of
14
[ C] PCB 52 in tissues to result in no net change in the predicted whole-body
chemical concentrations. These adjustments were made by configuring the
model so that [ 14C] PCB 52 masses in Table 1 could be multiplied by a fixed
value, referred to as the biomagnification factor (BMF) because of its equivalence to BMFs reported in laboratory and field studies with fish (i.e., a unitless
factor by which the lipid-normalized chemical concentration in a predator
exceeds that of its food). The value of the BMF for a given exposure scenario
was then determined by successive approximation. Consecutive feedings were
simulated by setting the [ 14C] PCB 52 concentration in each meal equal to that
of the first. Fish were assumed to eat one meal of contaminated prey every 48 h
equal to 4% of their body weight, and changes in pyloric ceca:blood and upper
intestine:blood chemical partitioning associated with uptake and processing of
dietary lipid were repeated as necessary to simulate multiple feeding events.
Previous studies have shown that trout have an extremely limited capacity to
metabolically transform [ 14C] PCB 52 (Nichols et al., 2001). Metabolic clearance was, therefore, assumed to be negligible during these initial simulations.
TABLE 1
Predicted Equilibrium Residues of [ 14C] PCB 52 in Tissues and
Intestinal Contents of Subadult Rainbow Trout Exposed to 0.08
␮g/l [ 14C] PCB 52 in Water
Tissues and intestinal
contents
Volume
(g) a
[ 14C] PCB 52
concentration
(␮g/kg tissue)
[ 14C] PCB 52
mass (␮g)
Fat
Kidney
Liver
Richly perfused
Poorly perfused
Stomach
Pyloric ceca
Upper intestine
Lower intestine
Chyme (pyloric ceca) b
Chyme (upper intestine) b
Feces b
4.15
0.83
1.35
1.56
90.93
1.56
2.18
0.73
0.52
1.25
0.50
0.10
50,472
493
749
749
194
306
864
864
518
173
173
259
209.46
0.41
1.01
1.17
17.64
0.48
1.88
0.63
0.27
0.22
0.09
0.05
a
Calculated based on an average whole-body weight of 103.8 g.
Volumes correspond to the empty state of each gut segment; see Nichols et
al. (2004).
b
221
DIETARY MODEL FOR FISH—CHRONIC EXPOSURES
The effects of chemical elimination across the skin or in urine were also
ignored.
Log K OW␥ dependence of BMFs, diffusion rate constants, and net assimilation efficiency. Additional simulations were run for a set of hypothetical
compounds with log K OW values of 6.5, 7, 7.5, and 8. These simulations were
limited to the combined food and water exposure scenario. P B:W and P F:W
partitioning coefficients were estimated using empirical relationships given by
Fitzsimmons et al. (2001) and Bertelsen et al. (1998), respectively (Table 2).
P F:B was then calculated from the quotient of these two values. Empirical
equations relating kidney:water and muscle:water partitioning to chemical log
K OW have also been reported for fish (Bertelsen et al., 1998). However, the
slope terms in these fitted equations are close to that for blood:water partitioning. The result is that estimated tissue:blood partitioning coefficients (P T:B)
change very little with log K OW. Therefore, P T:B values for all tissues except fat
were set equal to those determined for [ 14C] PCB 52 (Nichols et al., 2004).
Bile:liver partitioning values were also set equal to that for [ 14C] PCB 52.
For simplicity, it was assumed that all chemicals were freely dissolved in
water and available for uptake across the gills. The binding of hydrophobic
compounds to dissolved organic matter in water can substantially reduce
branchial uptake of high log K OW compounds by fish (Black and McCarthy,
1988), but this binding should have a similar effect on chemical accumulation
by both the predator and its prey resulting in proportional impacts on steadystate chemical concentrations.
The first set of simulations was generated by adjusting chemical residues in
fish tissues (i.e., BMFs) to result in no net change in whole-body chemical
concentration. Additional simulations were then obtained by fixing BMFs for
the three most hydrophobic compounds and adjusting permeability coefficients
within the GI tract to result in no net change of chemical residues. Guidance
on the specification of BMFs was obtained from the literature. Specifically,
laboratory and field data suggest that, within the log K OW range of 6 to 8, BMFs
for fish seldom exceed 10 for persistent, poorly metabolized compounds and
more commonly range from 3 to about 5 (Gobas et al., 1993b, 1999; Russell
et al., 1995). Additional data suggest that steady-state bioaccumulation factors
(BAFs) decrease with chemical log K OW at very high (⬎7) log K OW values
(Burkhard, 1998; Thomann, 1989). The BAF is defined in this case as the
lipid-normalized chemical concentration in fish divided by the free chemical
concentration in water. Because BAFs and BMFs are manifestations of the
same uptake and elimination processes, this observed decline in BAFs is likely
to have been associated with a log K OW-dependent decline in BMFs (had the
chemical concentration in prey been measured).
Initially, BMFs for log K OW 7, 7.5, and 8 compounds were set equal to 5.
BMFs for the same three compounds were then equal to 5, 3, and 1, respectively. In each case, permeability coefficients for the pyloric ceca, upper
TABLE 2
Predicted Equilibrium Blood:Water, Fat:Water, and Fat:Blood
Partitioning Coefficients for a Set of Hypothetical Organic Compounds in Rainbow Trout
intestine, and lower intestine were varied in a proportional manner, maintaining the relationship established previously by fitted values for [ 14C] PCB 52
(approximately 1.0:0.2:0.08). Finally, models with fitted permeability coefficients were used to simulate three consecutive feedings of contaminated prey
to a previously unexposed fish to evaluate the impact of these adjustments on
net assimilation efficiency, defined as the percentage of chemical consumed by
the animal that is retained within its tissues.
Biliary elimination, metabolism, and growth. Biliary elimination is calculated in the model as the product of bile flow rate and a liver:bile concentration ratio, which represents the net result of chemical partitioning and active
secretion of parent compounds into bile (Nichols et al., 2004). The concentration ratio was set equal to that (0.67) determined in earlier feeding studies
with [ 14C] PCB 52 (Nichols et al., 2001). The importance of biliary elimination
was then evaluated by setting the bile flow rate equal to zero and examining the
effect of this change on BMFs for [ 14C] PCB 52 and a set of hypothetical high
log K OW compounds.
The gut model provides for the possibility of metabolic biotransformation in
both the liver and gut tissues. Potential effects of metabolism were examined
by adjusting first-order rate constants in the liver, pyloric ceca, and upper
intestine. Following Nichols et al. (1990), these rate constants were referenced
to the chemical concentration in venous blood exiting each tissue. Initially, the
metabolism rate constant in liver was set equal to 0.02/h. The model was then
used to calculate BMFs for [ 14C] PCB 52 and a set of hypothetical high log
K OW compounds. When the rate of liver metabolism is low, hepatic clearance
is not limited by the rate of chemical delivery to the liver in blood or
redistribution among other tissues. Under these circumstances, the impact of
metabolism on whole-animal kinetics can be approximated by collapsing the
PBTK model into a single, well-stirred compartment. Using tissue volumes
and partitioning coefficients given by Nichols et al. (2004), it can be shown
that a liver metabolism rate constant of 0.02/h equates to a whole-body
elimination rate constant of about 0.0004/h or approximately 0.001/day. This
value corresponds to an elimination half-life (T 21) of about 700 days, assuming
that metabolism is the only route of elimination. In a second set of simulations,
elimination rate constants for the pyloric ceca and upper intestine were also set
equal to 0.02/h. Combined with hepatic metabolism, this results in a wholebody elimination rate constant of about 0.003/day and T 21 of approximately
220 days. In a third modeling exercise, the metabolism rate constant for liver
was set equal to 0.2/h, resulting in a whole-body elimination rate constant of
0.01/day and T 21 of about 70 days.
The effect of growth on model performance was evaluated by calculating
fish body weight using a zero-order growth rate term. The starting weight was
set equal to that (103.8 g) used in previous simulations. The growth rate (in
g/h) was then adjusted to achieve an annual size increase of 10 or 50 g.
Parameters within the model that scale to body weight, including tissue
volumes, gut surface areas, physiological parameters (e.g., cardiac output), and
metabolism rates, were recalculated at each time step. Meal size and the
amount of chemical consumed by fish were also adjusted to maintain the same
weight-normalized dose at each feeding interval.
RESULTS
Chemical log K OW
Partitioning
coefficient
6.5
7.0
7.5
8.0
Blood:water
Fat:water
Fat:blood
7330
1,445,440
197
16,980
4,073,800
240
39,360
11,481,540
292
91,200
32,359,370
355
Note. Blood:water partitioning was calculated from the empirical relationship given by Fitzsimmons et al. (2001) for neutral organic compounds.
Fat:water partitioning was calculated from the empirical relationship given by
Bertelsen et al. (1998) for all fish species, ignoring the contribution of tissue
water. Fat:blood partitioning was calculated as the quotient of fat:water and
blood:water partitioning values.
Figures 1 and 2 show the simulated time-course for [ 14C]
PCB 52 in rainbow trout exposed under two hypothetical
scenarios. The results of a water-only exposure are shown in
Figures 1A and 1B. When chemical residues in tissues were set
equal to those corresponding to the BCF residue set, the model
predicted that [ 14C] PCB 52 would be eliminated by fish. A
steady-state condition was established by multiplying the BCF
residue set by a BMF of 0.86. Because this exposure scenario
did not include [ 14C] PCB 52 in the diet, this BMF cannot be
interpreted in terms of a relationship between the fish and its
food. A value less than 1.0 indicated, however, that elimination
222
NICHOLS, FITZSIMMONS, AND WHITEMAN
using gut permeability coefficients for [ 14C] PCB 52. Steadystate BMFs obtained from this effort are shown in Figure 3,
along with the value determined for [ 14C] PCB 52. BMFs
estimated in this manner increased nonlinearly with chemical
log K OW, attaining a value of 27.5 for the log K OW 8 compound.
Steady-state BMFs for all compounds were reduced when
fish growth or metabolic biotransformation was incorporated
into the model (Fig. 3). The effect of a 10 g annual size
increase was similar to that achieved using a liver metabolism
rate constant of 0.02/h (0.001/day on a whole-body basis). The
same level of metabolism, when ascribed to both the liver and
gut tissues, reduced steady-state BMFs for all compounds to
about the same extent as a 50 g annual size increase. The effect
of metabolism or growth on steady-state BMFs increased with
chemical log K OW when expressed as a percentage reduction in
values obtained without these influences. In every case, however, BMFs increased with chemical log K OW.
Subsequently, the steady-state BMF was fixed at a value of
5 and the model was used to fit a set of gut permeability
coefficients for compounds with log K OW values of 7, 7.5, and
8. Permeability coefficients resulting in steady-state conditions
FIG. 1. Simulated steady state kinetics for a waterborne exposure to [ 14C]
PCB 52. (A) Concentration time course in arterial blood (dotted/dashed line),
contents of the lower intestine (dashed line), and calculated on a whole-body
basis (solid line). (B) Chemical flux within the pyloric ceca (dashed line),
lower intestine (dotted/dashed line), and at the gills (solid line).
of [ 14C] PCB 52 to feces reduced chemical concentrations in all
tissues below the levels predicted from an equilibrium distribution between the fish and water. In simulations encompassing more than one feeding event, this elimination also caused
a small but regular fluctuation in whole-body chemical concentration.
The results of a simulated chronic exposure to [ 14C] PCB 52
in food only are given in Figures 2A and 2B. A steady-state
condition was established by multiplying the BCF residue set
by a BMF of 2.24. The predicted [ 14C] PCB 52 concentration
time course in contents of the lower intestine is shown to
emphasize the role of fecal elimination. [ 14C] PCB 52 concentrations expressed on a whole-body basis reflect changes in
chemical distribution among tissues as well as the balance
between uptake and elimination. Similar patterns were generated for a chronic food and water exposure. Under these
conditions, however, a BMF of 3.11 was required to achieve a
steady state. [ 14C] PCB 52 flux was positive (i.e., inwardly
directed) at all times in each gut segment under both the
food-only and combined food and water exposure scenarios.
Additional simulations were run for a set of hypothetical
compounds with log K OW values of 6.5, 7, 7.5, and 8, again
FIG. 2. Simulated steady-state kinetics for a food-only exposure to the
[ 14C] PCB 52. (A) Concentration time course in arterial blood (dotted/dashed
line), contents of the lower intestine (dashed line), and calculated on a wholebody basis (solid line). (B) Chemical flux within the pyloric ceca (dashed line),
lower intestine (dotted/dashed line), and at the gills (solid line). Simulations
were obtained by setting TCB residues in tissues equal to 2.24 times those of
a fish in thermodynamic equilibrium with water.
DIETARY MODEL FOR FISH—CHRONIC EXPOSURES
FIG. 3. Steady-state biomagnification factors determined for [ 14C] PCB 52
and a set of hypothetical organic compounds with increasing log K OW values
in simulated food and water exposures. Filled circles: growth and metabolic
rate constants set equal to zero; open circles: growth rate of 10 g/year; filled
triangles: metabolism in the liver (equivalent to 0.001/day on a whole-body
basis); open triangles: growth rate of 50 g/year; filled squares: metabolism in
the liver and tissues of the pyloric ceca and upper intestine (0.003/day); open
squares: metabolism in the liver (0.01/day).
under this constraint declined with chemical log K OW, but the
relationship was nonlinear (Fig. 4; permeability coefficients are
expressed as a percentage of values determined for [ 14C] PCB
52). A further decease in fitted permeability coefficients was
observed when steady-state BMFs were set equal to 5, 3, and
1 for the log K OW 7, 7.5, and 8 compounds, respectively.
The effect of biliary elimination on biomagnification was
investigated by setting the bile flow rate equal to zero. Steadystate BMFs for all compounds were essentially identical to
those obtained using the model with biliary elimination when
gut permeability coefficients were set equal to those for [ 14C]
PCB 52. However, lower fitted permeability coefficients were
obtained for the log K OW 7, 7.5, and 8 compounds when the
maximum BMF was set equal to 5 (Fig. 4).
The incorporation of metabolism or growth into the model
resulted in higher fitted permeability coefficients for high log
K OW compounds. This increase in chemical permeability was
required to offset the effect of metabolism or growth on a fixed
level of bioaccumulation. The overall effect of these changes
was to shift the relationship between predicted chemical permeability and chemical log K OW to the right (Fig. 5).
Optimized models for each compound were then used to
estimate net assimilation efficiency in food-only exposures of
previously unexposed fish. Assimilation efficiencies deter-
223
mined in this manner are given in Figure 6, along with measured values for a variety of halogenated hydrocarbons and fish
species compiled by Muir and Yarechewski (1988). Simulated
assimilation efficiencies declined in a nonlinear manner with
chemical log K OW, but the shape of this curve differed from that
describing the relationship between gut permeability and
chemical log K OW. Decreasing the steady-state BMF to 3 (log
K OW 7.5 compound) or 1 (log K OW 8 compound) substantially
reduced the predicted assimilation efficiency for each chemical. A plot of net assimilation efficiency against fitted permeability coefficients from all model simulations yielded a curvilinear relationship (Fig. 7). The shape of this curve suggests
that dietary assimilation efficiency does not change in proportion to chemical permeability across the gut. Instead, large
changes in gut permeability have a relatively smaller impact on
dietary uptake efficiency.
Assimilation efficiency predictions were unaffected by fish
growth (data not shown). This result was anticipated because
assimilation efficiency is calculated on the basis of chemical
mass, which (unlike chemical concentration) does not change
with growth. Metabolic clearance was expected to reduce net
assimilation efficiency by decreasing the amount of chemical
retained over time. The magnitude of this effect was slight,
however, because low chemical concentrations at early time
points resulted in very low metabolic elimination rates. Assim-
FIG. 4. Fitted gut permeability coefficients from simulated food and water
exposures. Permeability coefficients are expressed as percentages of values
determined by Nichols et al. (2004) for [ 14C] PCB 52. Filled circles: assumed
maximum BMF of 5 with chemical elimination in bile; open circles: assumed
BMFs of 3 and 1 for compounds with log K OW values of 7.5 and 8.0,
respectively, with chemical elimination in bile; filled triangles: assumed maximum BMF of 5 without chemical elimination in bile.
224
NICHOLS, FITZSIMMONS, AND WHITEMAN
FIG. 5. The effect of growth and metabolism on fitted gut permeability
coefficients from simulated food and water exposures. Permeability coefficients are expressed as percentages of values determined by Nichols et al.
(2004) for [ 14C] PCB 52. All simulations were generated assuming a maximum
BMF of 5. Filled circles: growth and metabolic rate constants set equal to zero;
open circles: growth rate of 10 g/year; filled triangles: metabolism in the liver
(equivalent to 0.001/day on a whole-body basis); open triangles: growth rate of
50 g/year; filled squares: metabolism in the liver and tissues of the pyloric ceca
and upper intestine (0.03/day).
that this fecal elimination is sufficient to reduce chemical
concentrations in tissues below those expected from an equilibrium between the fish and water.
The expected outcome in chronic waterborne exposures to
compounds that are more hydrophobic than [ 14C] PCB 52 is
less certain. Limited data suggest that the maximum branchial
uptake rate declines with increasing chemical log K OW at log
K OW values greater than about 6, when expressed on the basis
of total aqueous chemical concentration (McKim et al., 1985).
The most likely explanation for this trend is a log K OWdependent decrease in bioavailability due to chemical complexation with dissolved organic matter (Erickson and McKim,
1990). Under these circumstances alone, chemical elimination
in feces might be expected to have a large impact on chemical
accumulation, but a decline in gut permeability coefficients
with increasing log K OW (see below) would tend to oppose this
effect.
It is unlikely, however, that environmental exposures to high
log K OW compounds would be limited to contact with the
contaminant in water. Instead, the diet is thought to be the
principal route of exposure for compounds with log K OW values
greater than about 5 (Bruggeman et al., 1984). As indicated
previously, the digestion of contaminated food may result in
chemical residues in fish higher than those expected from
equilibrium partitioning with water. Under these circumstances, the gills become a route of chemical elimination and
not uptake. Recently, Fitzsimmons et al. (2001) measured the
branchial elimination of four PCB congeners with log K OW
values ranging from 6.1 to 8.2, and used this information to
calculate a set of apparent P B:W values based on total chemical
ilation efficiencies for all compounds increased when bile flow
was set equal to zero, but the extent of these changes was also
slight.
DISCUSSION
In a companion report, we described a PBTK model for
dietary uptake of hydrophobic chemicals by fish (Nichols et al.,
2004). In the present study, we used this model to simulate a
variety of chronic exposure scenarios as a means of evaluating
factors that control the uptake and accumulation of these compounds by fish. Initially, the model was used to simulate a
chronic waterborne exposure to [ 14C] PCB 52. Because of its
hydrophobicity, the maximum rate at which [ 14C] PCB 52 can
be taken up from water is determined by the capacity of
inspired water to deliver freely dissolved chemical to the gills
(Erickson and McKim, 1990). [ 14C] PCB 52 absorbed across
the gills distributes within the animal, and, over time, the
concentration in blood increases, decreasing the rate of
branchial uptake by reducing the activity gradient for diffusion.
A portion of the absorbed chemical partitions into contents of
the GI tract and is eliminated in feces. Model simulations
resulting in a steady-state BMF less than unity (0.86) suggest
FIG. 6. Dietary assimilation efficiencies in food-only exposures of previously unexposed fish. The value for [ 14C] PCB 52 (log K OW ⫽ 6.1) is that
reported by Nichols et al. (2001) from feeding studies with rainbow trout.
Simulated values for a set of hypothetical organic compounds with log K OW
values of 6.5, 7, 7.5, and 8 were obtained using models with fitted gut
permeability coefficients and chemical elimination in bile. Filled circles:
assumed maximum BMF of 5; filled squares: assumed BMFs of 3 and 1 for
compounds with log K OW values of 7.5 and 8.0, respectively. Measured values
compiled by Muir and Yarechewski (1988) are shown as open triangles.
DIETARY MODEL FOR FISH—CHRONIC EXPOSURES
FIG. 7. Dependence of dietary assimilation efficiency on fitted gut permeability coefficients. Permeability coefficients, expressed as percentages of
values determined for [ 14C] PCB 52, correspond to those that resulted in the
assimilation efficiencies shown in Figure 6. Filled circles: assumed maximum
BMF of 5 for compounds with log K OW values of 6.5, 7, 7.5, and 8; open
circles: assumed BMFs of 3 and 1 for compounds with log K OW values of 7.5
and 8, respectively.
concentrations in blood and water. A plot of log P B:W against
chemical log K OW yielded a linear relationship similar to that
obtained from an earlier study of less hydrophobic compounds
(Bertelsen et al., 1998). The observed linearity of this relationship was interpreted as evidence for an equilibrium at the gills
between the freely dissolved chemical concentration in water
and the total chemical concentration in blood. For the purposes
of the present study, this relationship is important because it
provides a basis for predicting branchial elimination when
BMFs are greater than 1.
In simulated steady-state dietary exposures to [ 14C] PCB 52,
chemical residues in tissues exceeded those associated with a
steady-state waterborne exposure. The extent of biomagnification predicted by the model differed somewhat depending on
whether the chemical was assumed to be present in food only
(BMF ⫽ 2.24) or both food and water (BMF ⫽ 3.11). This
distinction is important because laboratory studies with hydrophobic compounds are often conducted by spiking chemicals
into a prepared diet, while environmental exposures are more
likely to include both food and water routes. In feeding studies
with guppies and goldfish, the potential for biomagnification of
PCB 52 (unlabeled) was evaluated by measuring the feces:food
fugacity ratio after 14 days of exposure (Gobas et al., 1993b).
The reported ratio for goldfish (1.2) is somewhat lower than the
modeled BMF for [ 14C] PCB 52 in trout (food-only exposure),
while the ratio for guppies (2.3) is essentially identical. Direct
225
evidence for biomagnification of PCB 52 in a natural setting
was given by Russell et al. (1995). Lipid-normalized PCB 52
concentrations in Lake Erie white bass were 2.7 times higher
than those measured in their principal prey, the emerald shiner
(Notropis atherinoides). This field-derived BMF value compares favorably to the modeled BMF for trout (food and water
exposure).
The effect of chemical hydrophobicity on dietary uptake was
examined by simulating chronic exposures to a set of hypothetical organic compounds. When gut permeability coefficients were set equal to those used to simulate [ 14C] PCB 52
kinetics, fitted BMFs for the two most hydrophobic compounds
(log K OW 7.5 and 8) exceeded values previously reported for
chemicals of comparable hydrophobicity. BMFs for all compounds were subsequently limited to a maximum value of 5,
and the model was used to fit a set of gut permeability coefficients resulting in steady-state chemical concentrations. Expressed as a percentage of values used to model [ 14C] PCB 52,
these fitted permeability coefficients decreased in a nonlinear
manner with chemical log K OW, with the greatest decline occurring in the log K OW range of 6.5 to 7 (Fig. 4). A further
decrease in fitted permeability coefficients was observed when
BMFs for the two most hydrophobic compounds were assumed
to decline with chemical log K OW.
Eliminating biliary clearance from the model had little impact on chemical accumulation when gut permeability coefficients were set equal to fitted values for [ 14C] PCB 52. When
the maximum steady-state BMF was set equal to 5, however,
this loss of biliary clearance resulted in a log K OW-dependent
decrease in fitted permeability coefficients for high log K OW
compounds. The fact that this occurred only when gut permeability was very low to begin with (due to the imposed limitation on BMF values) shows that biliary clearance impacts
whole-animal kinetics only when conditions limit the extent to
which chemicals secreted into bile can be reabsorbed from the
GI tract. In this context, hepatic metabolism contributes to
biliary elimination of hydrophobic compounds by transforming
them into polar products that are retained within the GI tract
and eliminated in feces.
Detailed kinetic data, including chemical concentrations in
gut contents and tissues, have not been collected for compounds other than [ 14C] PCB 52. Lacking this information, it is
not possible to independently evaluate the magnitude of fitted
permeability coefficients for hypothetical high log K OW compounds. An indirect test of the model can be accomplished,
however, by comparing published dietary assimilation efficiencies to values predicted by the model using fitted gut permeability coefficients. A plot of simulated assimilation efficiencies is shown in Figure 6, along with measured values
compiled by Muir and Yarechewski (1988) for a variety of
halogenated hydrocarbons and fish species. Additional data
sets have been published by Gobas et al. (1988) and Opperhuizen and Sijm (1990). However, a review of sources suggests
that there is considerable overlap between these summaries and
226
NICHOLS, FITZSIMMONS, AND WHITEMAN
that given by Muir and Yarechewski (1988). An examination
of Figure 6 shows that measured assimilation efficiencies generally range from 40 to 80% for compounds with log K OW
values less than 6. In the log K OW range from 6.5 to 8,
assimilation efficiencies tend to decline with chemical log K OW.
Assimilation efficiencies less than 10% have been reported for
several compounds with log K OW values between 8 and 10.
Modeled assimilation efficiencies from the present study
described a declining trend with chemical log K OW when the
maximum BMF was set equal to 5 (Fig. 6). An additional
decrease in dietary assimilation efficiency was predicted when
BMFs for the log K OW 7.5 and 8 compounds were set equal to
3 and 1, respectively. Generally, however, modeled assimilation efficiencies were higher than estimates reported by other
investigators. Most of the feeding studies conducted to date
have employed small fish species such as guppies and fathead
minnows or juveniles of larger species. Very few studies have
been performed using larger animals, and fewer still have
employed chemicals that were naturally incorporated into live
prey items. Previously, Nichols et al. (2001) noted that the
measured assimilation of [ 14C] PCB 52 by subadult rainbow
trout was higher than any value previously reported for this
compound. This finding suggests that the conditions under
which the study was performed (species, life stage, feeding
rate, the use of naturally contaminated prey items, and so on),
and upon which the current PBTK model was based, favor
highly efficient chemical uptake.
Throughout this study, modeling efforts were aided by the
adoption of several simplifying assumptions. Two of these
assumptions, zero growth and no metabolic biotransformation,
merit special attention. As indicated previously, BAFs for
some hydrophobic compounds have been shown to decline at
very high log K OW values (Burkhard, 1998; Thomann, 1989).
Because they arise from the same kinetic processes, this decrease in BAFs is probably associated with a log K OW-dependent decrease in BMFs. One question is whether growth dilution of chemical residues could bring about a decline in BMFs
(and by extension, BAFs). The gut model was used to evaluate
this question by incorporating a constant rate of growth into
steady-state food and water simulations.
Predicted BMFs for all compounds were substantially reduced by the incorporation of low (10 g/year) to moderate (50
g/year) rates of growth when compared with BMFs obtained
without growth (Fig. 3). When expressed as percentages of
body weight, these growth rates correspond to annual size
increases of about 10 and 50%, respectively. The intent of this
exercise was to model growth rates expected for adult fish
because biomagnification is generally assessed in adult animals. Growth rates in juvenile fish may be much higher than
those modeled here. However, feeding rates, dietary preferences, and food conversion efficiency also change as fish grow
and develop. A model-based evaluation of these factors is
beyond the scope of the present study (and would probably
require a bioenergetics-based component to the model). For
this effort, it is important to note that, for each growth rate and
set of assumed gut permeability coefficients, modeled BMFs
increased with chemical log K OW. The conclusion drawn from
this analysis is that growth dilution may contribute to a decline
in steady-state chemical concentration for any single compound but is unlikely to be responsible for a log K OW-dependent decline in BMFs.
Metabolic biotransformation also has the potential to reduce
chemical biomagnification in fish. To investigate this possibility, models developed to describe steady-state food and water
exposures were amended to include relatively low rates of
metabolism in the liver (0.001 to 0.01/day, on a whole-body
basis) or both the liver and tissues of the upper GI tract (Fig. 3).
For comparison, Van der Linde et al. (2001) used a modelbased analysis of measured elimination rates to estimate
whole-body metabolism rate constants in fish. Rate constants
ranging from 0.01 to 0.1/day were estimated for several chlorinated dioxins and furans as well as some PAHs. Metabolism
rates for a set of halogenated benzenes and biphenyls could not
be estimated but were less than 0.01/day and probably close to
zero.
Based on this evaluation, it is clear that metabolism can have
a substantial effect on the accumulation of high log K OW
compounds. PBTK models are well suited to incorporation of
metabolism data from both in vitro and in vivo studies as such
data are collected. An important advantage of this approach,
when compared to whole-animal bioaccumulation models, is
that metabolism may be localized to the tissues and organs
where it occurs. In studies with in situ gut preparations, the GI
tract was shown to play an important role in metabolizing
PAHs by fish, altering their form and limiting systemic bioavailability (Kleinow et al., 1998; Van Veld et al., 1988).
Operating in series with first-pass metabolism in the liver, this
activity can substantially reduce bioaccumulation of contaminant residues (James and Kleinow, 1994; Kleinow and James,
2001; Van Veld, 1990). Alternatively, PBTK models could be
used to translate metabolism rates determined on a wholeanimal basis into tissue-specific metabolism rates, provided
that the principal metabolizing tissues were known.
A third assumption employed in this effort is that chemical
residues in food can be predicted from an equilibrium chemical
distribution between the food item and water. Given this assumption, the BMF is equal to the ratio of a lipid-normalized
BAF (for the predator) and a lipid-normalized BCF (for its
food), both of which are referenced to the free chemical concentration in water. It is possible, however, that the same
considerations that result in biomagnification of chemicals
within the predator operate to some extent also in their prey
(due to consumption of contaminated food items). Under these
circumstances, chemical concentrations in prey could exceed
those predicted from an equilibrium chemical distribution with
water. Responding to this increased input, chemical concentrations within the predator would tend to rise, increasing the
gradient for chemical elimination across the gills. For a high
DIETARY MODEL FOR FISH—CHRONIC EXPOSURES
log K OW compound, however, this increase in branchial elimination would be expected to have only a slight effect on the
steady-state BMF (defined in this case as a ratio of lipidnormalized BAFs), given the dominant role of dietary uptake
in controlling chemical biomagnification.
The processes that control dietary uptake of hydrophobic
organic compounds have been studied in several vertebrate
systems. For example, Drouillard and Norstrom (2000) exposed ring doves to a mixture of PCB congeners spiked into a
pelleted diet. Although there was a slight trend toward decreasing assimilation efficiency with chemical log K OW, all of the
compounds were taken up with high efficiency. Moreover, the
kinetics of chemical appearance in blood did not vary with log
K OW, as would be expected if diffusion rates varied with
relative hydrophobicity. Based on these observations, it was
concluded that chemical uptake from the small intestine is not
rate-limited by simple diffusion of contaminant molecules but
is controlled by the collisional contact of lipid micelles with
intestinal cell membranes.
The dietary absorption of several hydrophobic compounds
by humans was well correlated with blood lipid levels but
could not be explained on the basis of a diffusion gradient
between blood and contents of the GI tract (Schlummer et al.,
1998). To account for this discrepancy, the authors hypothesized that lipid uptake by intestinal tissues creates a transient
inward gradient for diffusion (this effect was termed the “fat
flush” hypothesis) and proposed a two-step model for chemical
uptake in the GI tract, with absorption (small intestine) and
elimination (large intestine) occurring as distinct processes,
both of which are controlled by diffusion gradients. Support for
this hypothesis was obtained in studies demonstrating the following: (1) that an increase in the chemical capacity of the gut
contents leads to increased fecal elimination of ingested compounds (Geusau et al., 1999; Moser and McLachlan, 2001);
and (2) that net absorption efficiency depends on the chemical
concentration in the diet relative to that of the exposed individual, becoming maximal at high dietary intake levels (Moser
and McLachlan, 2001).
The role of lipid micelles in dietary uptake of hydrophobic
compounds by fish was studied by Doi et al. (2000). Using an
in situ channel catfish intestinal preparation, the bioavailability
of 3,3⬘,4,4⬘-tetrachlorobiphenyl (PCB 77) was shown to be
related to the composition of mixed micelles. This finding
suggests that dietary uptake may depend, to some extent, on
both the chemical capacity of lipid micelles as well as interactions of component molecules with the brush boarder membrane. However, Doi et al. (2000) also observed that chemical
pretreatment of catfish decreased the uptake rate for PCB 77,
suggesting that simple diffusion plays some role in controlling
dietary uptake of hydrophobic compounds. Previously, Vetter
et al. (1985) used autoradiographic methods to show that
[ 14C]-benzo(a)pyrene remained associated with lipid throughout lipolysis, lipid absorption, and the formation of intracellular fat droplets in the gut epithelium. This technique under-
227
scores the close association between hydrophobic compounds
and lipids throughout digestion but provides little information
on processes that actually limit the rate of absorption at the
gastrointestinal epithelium.
Contradictory results have been obtained by investigators
who varied the amount of lipid added to prepared fish diets. In
work cited by Van Veld (1990), the addition of triglyceride
increased the intestinal absorption of DDT, benzo[a]pyrene,
and a model PCB in killifish. Maximum absorption occurred
when fish were fed a triglyceride-enriched diet designed to
mimic the characteristics of a natural diet. However, high-fat
diets are also known to slow the digestion of lipids by fish.
Under these circumstances, undigested lipid may compete with
lipid micelles for hydrophobic organic contaminants, reducing
dietary assimilation efficiency (Van Veld, 1990). The impact of
dietary lipid content on chemical uptake efficiency may also
depend partly on chemical hydrophobicity. Working with goldfish, Gobas et al. (1993a) found that dietary uptake of some
very hydrophobic compounds (log K OW ⬎ 6.3) declined with
an increase in dietary lipid content, while uptake of some
moderate-to-high log K OW compounds (4.5 ⬍ log K OW ⬍ 6.3)
did not vary among treatment groups.
The model used in the present study provides a mechanistic basis for interpreting BMFs and assimilation efficiencies determined in both experimental and field studies with
hydrophobic compounds. The model structure, and in particular the use of time-dependent changes in chemical partitioning to gut contents and tissues, simulates the “fat flush”
effect proposed by Schlummer et al. (1998) and the increase
in chemical activity of gut contents hypothesized by Gobas
et al. (1988). In a companion report, the model was used to
describe the results of a one-time feeding study with [ 14 C]
PCB 52 (Nichols et al., 2004). Based on this analysis, it was
suggested that a resistance to chemical flux prevented the
establishment of a chemical equilibrium between gut contents and tissues. Furthermore, it was suggested that diffusion across the gastrointestinal epithelium was unlikely to
be the sole determinant of [ 14 C] PCB 52 uptake rate. In the
present study, we used the same model to simulate chronic
exposures to [ 14 C] PCB 52 and a set of hypothetical high log
K OW compounds. The results of this study suggest that the
resistance to dietary uptake increases substantially with
chemical log K OW and that this increase is responsible for
the log K OW -dependent decline in assimilation efficiency
observed in several feeding studies with fish. The factors
that limit the rate of dietary uptake in fish remain unknown.
The present work suggests, however, that one or more of
these factors varies with chemical log K OW . This conclusion
is consistent with an earlier model-based evaluation of dietary assimilation efficiencies in fish (Gobas et al., 1988).
A comparison of data from fish, birds, and mammals
suggests that factors that control dietary uptake of hydrophobic organic compounds differ among taxa. The GI tract
of fish, in relation to animal size, is much shorter than that
228
NICHOLS, FITZSIMMONS, AND WHITEMAN
of most higher vertebrates. Under these conditions, it may
not be possible to achieve the spatial separation of diffusive
uptake and elimination processes suggested by studies with
humans (Schlummer et al., 1998). Experimental support for
this conclusion was provided by Gobas et al. (1999) in
studies with rainbow trout fed a diet contaminated with PCB
155. The fugacity of PCB 155 within the intestinal tract was
double that of the food but did not differ significantly among
three intestinal segments. Micelle-mediated transport may
play a role in controlling the rate of chemical delivery to
absorbing gut surfaces in fish (Doi et al., 2000) but does not
appear to be the dominant rate-controlling process for uptake indicated in studies with ring doves (Drouillard and
Norstrom, 2000).
ACKNOWLEDGMENTS
The authors thank Dr. Russell Erickson and Dr. Frank Gobas for their
insightful reviews of this manuscript. The information in this document has
been funded in part by the U.S. Environmental Protection Agency. It has
been subjected to review by the National Health and Environmental Effects
Research Laboratory and approved for publication. Approval does not
signify that the contents reflect the views of the agency, nor does mention
of trade names or commercial products constitute endorsement or recommendation for use.
REFERENCES
Bertelsen, S. L., Hoffman, A. D., Gallinat, C. A., Elonen, C. M., and Nichols,
J. W. (1998). Evaluation of log K OW and tissue lipid content as predictors of
chemical partitioning to fish tissues. Environ. Toxicol. Chem. 17, 1447–
1455.
Fitzsimmons, P. N., Fernandez, J. D., Hoffman, A. D., Butterworth, B. C., and
Nichols, J. W. (2001). Branchial elimination of superhydrophobic organic
compounds by rainbow trout (Oncorhynchus mykiss). Aquat. Toxicol. 55,
23–34.
Geusau, A., Tschachler, E., Meixner, M., Sandermann, S., Päpke, O., Wolf, C.,
Valic, E., Stingl, G., and McLachlan, M. S. (1999). Olestra increases fecal
excretion of 2,3,7,8-tetrachlorodibenzo-p-dioxin. Lancet 354, 1266 –1267.
Gobas, F. A. P. C., McCorquodale, J. R., and Haffner, G. D. (1993a). Intestinal
absorption and biomagnification of organochlorines. Environ. Toxicol.
Chem. 12, 567–576.
Gobas, F. A. P. C, Muir, D. C. G., and Mackay, D. (1988). Dynamics of dietary
bioaccumulation and fecal elimination of hydrophobic organic chemicals in
fish. Chemosphere 17, 943–962.
Gobas, F. A. P. C., Wilcockson, J. B., Russell, R. W., and Haffner, G. D.
(1999). Mechanism of biomagnification in fish under laboratory and field
conditions. Environ. Sci. Technol. 33, 133–141.
Gobas, F. A. P. C., Zhang, X., and Wells, R. (1993b). Gastrointestinal magnification: The mechanism of biomagnification and food chain accumulation
of organic chemicals. Environ. Sci. Technol. 27, 2855–2863.
James, M. O., and Kleinow, K. M. (1994). Trophic transfer of chemicals in the
aquatic environment. In Aquatic Toxicology. Molecular, Biochemical, and
Cellular Perspectives (D. C. Malins and G. K. Ostrander, Eds.), pp. 1–35.
Lewis, Boca Raton, FL.
Kleinow, K. M., and James, M. O. (2001). Response of the teleost gastrointestinal system to xenobiotics. In Target Organ Toxicity in Marine and
Freshwater Teleosts (D. Schlenk and W. H. Benson, Eds.), Vol. 1, pp.
269 –362. Taylor and Francis, London.
Kleinow, K. M., James, M. O., Tong, Z., and Venugopalan, C. S. (1998).
Bioavailability and biotransformation of benzo(a)pyrene in an isolated perfused in situ catfish intestinal preparation. Environ. Health Perspec. 106,
ISS 3, 155–166.
McKim, J., 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, 1–10.
Moser, G. A., and McLachlan, M. S. (2001). The influence of dietary concentration on the absorption and excretion of persistent lipophilic organic
pollutants in the human intestinal tract. Chemosphere 45, 201–211.
Black, M. C., and McCarthy, J. F. (1988). Dissolved organic macromolecules
reduce the uptake of hydrophobic organic contaminants by the gills of
rainbow trout (Salmo gairdneri). Environ. Toxicol. Chem. 7, 593– 600.
Muir, D. C. G., and Yarechewski, A. L. (1988). Dietary accumulation of four
chlorinated dioxin congeners by rainbow trout and fathead minnows. Environ. Toxicol. Chem. 7, 227–236.
Bruggeman, W. A., Opperhuizen, A., Wijbenga, A., and Hutzinger, O. (1984).
Bioaccumulation of super-lipophilic chemicals in fish. Toxicol. Environ.
Chem. 7, 173–189.
Nichols, J. W., Fitzsimmons, P. N., Whiteman, F. W., Dawson, T. D.,
Babeu, L., and Juenemann, J. (2004). A physiologically based toxicokinetic model for dietary uptake of hydrophobic organic compounds by
fish. I. Feeding studies with 2,2⬘,5,5⬘-tetrachlorobiphenyl. Toxicol. Sci.
77, 206 –218.
Burkhard, L. P. (1998). Comparison of two models for predicting bioaccumulation of hydrophobic organic chemicals in a Great Lakes food web. Environ. Toxicol. Chem. 17, 383–393.
Connell, D. W., and Hawker, D. W. (1988). Use of polynomial expressions to
describe the bioconcentration of hydrophobic chemicals by fish. Ecotoxicol.
Environ. Saf. 16, 242–257.
Doi, A. M., Holmes, L. E., Li, C.-L. J., Venugopal, C. S., James, M. O., and
Kleinow, K. M. (2000). Effect of micelle fatty acid composition and
3,4,4⬘,4⬘-tetrachlorobiphenyl (TCB) exposure on intestinal [ 14C]-TCB bioavailability and biotransformation in channel catfish in situ preparations.
Toxicol. Sci. 55, 85–96.
Drouillard, K. G., and Norstrom, R. J. (2000). Dietary absorption efficiencies
and toxicokinetics of polychlorinated biphenyls in blood plasma of ring
doves following initial exposure to an arochlor contaminated diet. Environ.
Toxicol. Chem. 19, 2707–2714.
Erickson, R. J., and McKim, J. M. (1990). A model for exchange of organic
chemicals at fish gills: flow and diffusion limitations. Aquat. Toxicol. 18,
175–198.
Nichols, J. W., Fitzsimmons, P. N., Whiteman, F. W., Kuehl, D. W.,
Butterworth, B. C., and Jenson, C. T. (2001). Dietary uptake kinetics of
2,2⬘,5,5⬘-tetrachlorobiphenyl in rainbow trout. Drug Metab. Disposit. 29,
1013–1022.
Nichols, J. W., McKim, J. M., Andersen, M. E., Gargas, M. L., Clewell, H. J.,
III, and Erickson, R. J. (1990). A physiologically based toxicokinetic model
for the uptake and disposition of waterborne organic chemicals in fish.
Toxicol. Appl. Pharm., 106, 433– 447.
Oliver, B. G., and Niimi, A. J. (1985). Bioconcentration factors of some
halogenated organics for rainbow trout: Limitations in their use for prediction of environmental residues. Environ. Sci. Technol. 19, 842– 849.
Opperhuizen, A., and Sijm, D. T. H. M. (1990). Bioaccumulation and biotransformation of polychlorinated dibenzo-p-dioxins and dibenzofurans in
fish. Environ. Toxicol. Chem. 9, 175–186.
Russell, R. W., Lazar, R., and Haffner, G. D. (1995). Biomagnification of organochlorines in Lake Erie white bass. Environ. Toxicol. Chem. 14, 719 –724.
DIETARY MODEL FOR FISH—CHRONIC EXPOSURES
Schlummer, M., Moser, G. A., and McLachlan, M. S. (1998). Digestive
tract absorption of PCDD/Fs, PCBs, and HCB in humans: Mass balances
and mechanistic considerations. Toxicol. Appl. Pharmacol. 152, 128 –
137.
Thomann, R. V. (1989). Bioaccumulation model of organic chemical distribution in aquatic food chains. Environ. Sci. Technol. 23, 699 –707.
Van der Linde, A., Hendriks, A. J., and Sijm, D. T. H. M. (2001). Estimating
biotransformation rate constants of organic chemicals from modeled and
measured elimination rates. Chemosphere 44, 423– 435.
229
Van Veld, P. A. (1990). Absorption and metabolism of dietary xenobiotics by
the intestine of fish. Rev. Aquat. Sci. 2, 185–203.
Van Veld, P. A, Patton, J. S., and Lee, R. F. (1988). Effect of preexposure to
dietary benzo[a]pyrene (BP) on the first-pass metabolism of BP by the
intestine of toadfish (Opsanus tau): In vivo studies using portal vein–
catheterized fish. Toxicol. Appl. Pharmacol. 92, 255–265.
Vetter, R. D., Carey, M. C., and Patton, J. S. (1985). Coassimilation of dietary
fat and benzo(a)pyrene in the small intestine: An absorption model using the
killifish. J. Lipid Res. 26, 428 – 434.