Page 142 - The Toxicology of Fishes

122 The Toxicology of Fishes

6
using isolated hepatocytes when expressed in comparable units (per 10 cells). Similarly, Worboys et al.
(1996a,b) found that CL invitro,int values for several compounds (expressed as V max /K on a whole-liver
m
basis) calculated using data from liver slices were lower than those obtained using isolated hepatocytes.
Moreover, the extent of this difference increased with increasing metabolic activity. These studies suggest
that slice metabolism rates are influenced by the rate of chemical diffusion from the culture medium
into the center of the slice.
Working with rainbow trout, Cravedi et al. (1999) found that isolated hepatocytes were well suited
for in vitro studies of biotransformation but in some cases failed to produce metabolites found in vivo
in urine and bile. Research with mammals suggests that it is common for chemicals to be transformed
by more than one metabolic reaction. Each of these reactions may exhibit saturable kinetics, charac-
terized by a different set rate and affinity constants. Under these circumstances, it is possible for one
pathway to predominate at one substrate concentration while another predominates at higher or lower
concentrations.
In several studies with mammals, in vitro metabolic parameters have been evaluated by comparison
with in vivo parameters determined using PBTK models. This approach is based on the assumption that
a PBTK model is “correct” with respect to all nonmetabolic aspects of parent chemical disposition. The
model can then be used to fit in vivo metabolism parameters by simulating the disappearance of parent
compound or, less frequently, the appearance of metabolites. Depuration data are generally preferred
for this purpose because of the absence of complications associated with chemical uptake and the rapid
phase of internal distribution. Not surprisingly, the agreement between in vitro metabolism data and
fitted in vivo parameters has been variable (Fiserova-Bergerova, 1995). In addition to limitations of the
in vitro systems themselves, substantial metabolism may occur in tissues other than those to which the
metabolism was ascribed in the model (and for which in vitro data are available). For this reason, fitted
in vivo parameters are best thought of as apparent values representing the summed activities of all
relevant enzyme systems and tissues. Law et al. (1991) used K and V max data from a trout hepatocyte
m
system to develop a PBTK model for pyrene. Total pyrene clearance greatly exceeded that predicted by
in vitro metabolism, requiring the incorporation of a fitted clearance constant.
Metabolic parameters determined from in vivo studies can also be incorporated into PBTK models.
Several authors have attempted to characterize in vivo metabolism by measuring products retained by
fish or eliminated in bile, urine, feces, and exposure water (Bradbury et al., 1986, 1993; Cravedi et al.,
1999; McKim et al., 1986; Stehly and Hayton, 1989a). In practice, however, these measurements are
very difficult to make due to low metabolite concentrations and incomplete extraction of samples. One
approach to dealing with these problems is to employ a technique called microdialysis to measure
metabolite concentrations in blood and tissues. McKim et al. (1993) implanted a microdialysis probe
into the dorsal aorta of rainbow trout and measured phenol and its major phase II metabolites (phenyl-
glucuronide and phenylsulfate) in the blood of fish exposed to phenol in water. More recently, Solem
et al. (2003) used a microdialysis probe to deliver a parent compound (phenol) to the liver in rainbow
trout and measure the production of phase I metabolic products (hydroquinone and catechol). An
important advantage of microdialysis sampling method is that dialysate samples are free of protein and
can be analyzed without extraction. These experiments provide qualitative information about the identity
and relative concentrations of metabolic products and can be performed without the need to expose
whole fish. Improved in vivo calibration procedures and a knowledge of chemical concentration gradients
around the microdialysis probe are required, however, before this information can be used to estimate
metabolic rate constants.

Utility of Physiologically Based Fish Models
In the true sense of the word, a PBTK model cannot be said to be valid, only that it does or does not
reproduce observed kinetics. Confidence in the specification of physiological parameters for a single
species is gained if the same set of inputs provides acceptable simulations for several compounds
exhibiting diverse partitioning behavior. Similarly, confidence in model structure derives from the ability
to simulate the kinetic behavior of the one or more chemicals in several different species by making
appropriate changes in physiological inputs.