Implications of chemical mixtures in public health practice.

The Agency for Toxic Substances and Disease Registry (ATSDR) is a federal public health agency that investigates and strives to prevent human health problems produced by exposure to toxic chemicals and their mixtures in the environment. Most human exposures involving toxic chemicals or mixtures are thought to originate from environmental and occupational sources; however, concurrent exposures are also likely from other sources, such as prescription and nonprescription drugs, indoor air pollutants, alcohol, and tobacco smoke. Thus, in evaluating the potential hazard following exposure to environmental mixtures, ATSDR not only considers the inherent joint toxicity of the mixture but also the influence of environmental, demographic, occupational, and lifestyle factors. To foster these goals, ATSDR has pursued a Mixtures Research and Assessment Program that consists of three component efforts: trend analysis, joint toxicity assessment, and experimental testing. Through trend analysis, ATSDR sets priorities for environmental mixtures of concern for which joint toxicity assessments are conducted as needed. If data are not available to conduct appropriate assessments, a research agenda is pursued through established extramural mechanisms. Ultimately, the data generated are used to support ATSDR's work at sites involving exposure to chemical mixtures. This pragmatic approach allows testable hypotheses or research needs to be identified and resolved and enhances our understanding of the mechanisms of joint toxicity. Several collaborative and cooperative efforts with national and international organizations such as the Toxicology and Nutrition Office, the Netherlands, and the Department of Energy are being pursued as part of these activities. ATSDR also develops guidance manuals to consistently and accurately apply current methodologies for the joint toxicity assessment of chemicals. Further, expert panels often are assembled to resolve outstanding scientific issues or obtain expert advice on pertinent issues. Recently, the need for studies on chemical mixtures has been proposed as one of the six priority areas the agency identified in its agenda for public health environmental research. This has been reinforced through the agency's close work with communities whose leaders have spoken passionately about their concern for information on exposures to chemical mixtures. The five other priority research areas the agency identified are exposure, susceptible populations, communities and tribal involvement, evaluation/surveillance of health effects, and health promotion/prevention.

PBPK modeling of the metabolic interactions of carbon tetrachloride and tetrachloroethylene in B6C3F1 mice.

Potential exists for widespread human exposure to low levels of carbon tetrachloride (CT) and tetrachloroethylene (TET). These halocarbons are metabolized by the cytochrome P450 system. CT is known to inhibit its own metabolism (suicide inhibition) and to cause liver injury by generation of metabolically derived free radicals. The objective of this research was to use develop a physiologically based pharmacokinetic (PBPK) model to forcast the metabolic interactions between orally administered CT and TET in male B6C3F1 mice. Trichloroacetic acid (TCA), a stable metabolite of TET, was used as a biomarker to assess inhibition of the cytochrome P450 system by CT. Metabolic constants utilized for CT were 1.0mg/kg/h for Vmaxc_CT and 0.3 for Km_CT (mg/l). Values for TET (based in TCA production), were 6.0mg/kg/h for Vmaxc_TET was 3.0mg/l for Km_TET. The rate of loss of metabolic capacity for CT (suicide inhibition) was describe as: Vmaxloss ( mg / h )=- Kd ( RAM × RAM ) , where Kd (h/kg) is a second-order rate constant, and RAM (mg/h) is the Michaelis-Menten description of the rate of metabolism of CT. For model simplicity, CT was assumed to damage the primary enzymes responsible for metabolism of CT (CYP2E1) and TET (CYP2B2) in an equal fashion. Thus, the calculated fractional loss of TET metabolic capacity was assumed to be equivalent to the calculated loss in metabolic capacity of CT. Use of a Kd value of 400h/kg successfully described serum TCA levels in mice dosed orally with 5-100mg/kg of CT. We report, for the first time, suicide inhibition at a very low dose of CT (1mg/kg). The PBPK model under-predicted the degree of metabolic inhibition in mice administered 1mg/kg of CT. This PBPK model is one of only a few physiological models available to predict the metabolic interactions of chemical mixtures involving suicide inhibition. The success of this PBPK model demonstrates that PBPK models are useful tools for examining the nature of metabolic interactions of chemical mixtures, including suicide inhibition. Further research is required to compare the inhibitory effects of inhaled CT vapors with CT administered by oral bolus dosing and determine the interaction threshold for CT-induced metabolic inhibition.

Prediction of tissue-air partition coefficients: a comparison of structure-based and property-based methods.

Three linear regression methods were used to develop models for the prediction of rat tissue-air partition coefficient (P). In general, ridge regression (RR) was found to be superior to principal component regression (PCR) and partial least squares regression (PLS). A set of 46 diverse low molecular-weight volatile chemicals was used to model fat-air, liver-air and muscle-air partition coefficients for male Fischer 344 rats. Comparisons were made between models developed using descriptors based solely on molecular structure and those developed using experimental properties, including saline-air and olive oil-air partition coefficients, as independent variables, indicating that the structure-property correlations are comparable to the property-property correlations. Multiple structure-based models were developed utilizing various classes of structural descriptors based on level of complexity, i.e. topostructural (TS), topochemical (TC), 3-dimensional (3D) and calculated octanol-water partition coefficient. In most cases, the structure-based models developed using only the TC descriptors were found to be superior to those developed using other structural descriptor classes. Haloalkane subgroups were modeled separately for comparative purposes, and although models based on the congeneric compounds were superior, the models developed on the complete sets of diverse compounds were acceptable. Comparisons were also made with respect to the types of descriptors important for partitioning across the various media.

Gene induction studies and toxicity of chemical mixtures.

As part of its mixtures program, the Agency for Toxic Substances and Disease Registry (ATSDR) supports in vitro and limited in vivo toxicity testing to further our understanding of the toxicity and health effects of chemical mixtures. There are increasing concerns that environmental chemicals adversely affect the health of humans and wildlife. These concerns have been augmented by the realization that exposure to chemicals often occurs to mixtures of these chemicals that may exhibit complex synergistic or antagonistic interactions. To address such concerns, we have conducted two studies with techniques that are being used increasingly in experimental toxicology. In the first study, six organochlorine pesticides (4,4 -DDT, 4,4 -DDD, 4,4 -DDE, aldrin, dieldrin, or endrin) were selected from the ATSDR Comprehensive Environmental Response, Compensation and Liability Act of 1980 (or Superfund) priority list and tested for their ability to modulate transcriptional activation of an estrogen-responsive reporter gene in transfected HeLa cells. In these assays, HeLa cells cotransfected with an expression vector encoding estrogen receptor and an estrogen-responsive chloramphenicol acetyltransferase (CAT) reporter plasmid were dosed with and without selected environmental chemicals either individually or in defined combinations. Estradiol consistently elicited 10- to 23-fold dose-dependent inductions in this assay. By contrast, all six of the organochlorine pesticides showed no detectable dose-related response when tested either individually or in binary combinations. Thus, these chemicals as binary mixtures do not exhibit any additional estrogenicity at the levels tested in these assays. In the second study, arsenic [As(V)], cadmium [Cd(II)], chromium [Cr(III, VI)], and lead [Pb(II)] were tested in a commercially developed assay system, CAT-Tox (L), to identify metal-responsive promoters and to determine whether the pattern of gene expression changed with a mixture of these metals. This assay employs a battery of recombinant HepG2 cell lines to test the transcriptional activation capacity of xenobiotics in any of 13 different signal-transduction pathways. Singly, As(V), Cd(II), Cr(III, VI), and Pb(II) produced complex induction profiles in these assays. However, no evidence of synergistic activity was detected with a mixture of Cd(II), Cr(III), and Pb(II). These results have shown metal activation of gene expression through several previously unreported signal-transduction pathways and thus suggest new directions for future studies into their biochemical mechanisms of toxicity. In conclusion, the (italic)in vitro(/italic) methods used in these studies provide insights into complex interactions that occur in cellular systems and could be used to identify biomarkers of exposure to other environmental chemical mixtures.

Effects of glutathione transferase theta polymorphism on the risk estimates of dichloromethane to humans.

The carcinogenic potential of dichloromethane (DCM) has been linked to its metabolism to formaldehyde by glutathione-S-transferase theta 1 (GSTT1). GSTT1 is polymorphic in humans. The frequency of the GSTT1 homozygous null genotype ranges from 10 to 60% in different ethnic and racial populations around the world. We investigated how varying GSTT1 genotype frequencies would impact cancer risk estimates for DCM by the application of Monte Carlo simulation methods in combination with physiologically based pharmacokinetic (PBPK) models. The PBPK model was used to estimate the DNA-protein cross links (DPX) caused by metabolism of DCM based on an earlier model. Cancer potency of DCM was obtained by the application of the estimated DPX amounts to the results of a carcinogenicity study by National Toxicology Program in B6C3F(1) mice. Human risks were estimated based on the carcinogenic potency of DCM to mice and PBPK-predicted amounts of DPX formed in humans. The Monte Carlo simulations were used to provide distributions of risk estimates for a sample of 1000 PBPK runs, each run representing a collection of biochemical and physiological parameters for a single person (with and without polymorphism included in the model). Our results show that average and median risk estimates were 23-30% higher when GSTT1 polymorphism was not included at inhalation DCM doses of 1000, 100, 10, and 1 ppm. This increase in risk was significantly reduced when it was based on the 95th percentile measure for all the doses. The specific effect of this polymorphism on population risk was further investigated by varying the probability that an individual may have a nonfunctional form of the enzyme at a constant dose level of 10 ppm of DCM. Higher values of this probability resulted in a corresponding decrease in risk. Again, this drop in population risk was not as significant when the 95th percentile measure was used.

Replication potential of cells via the protein kinase C-MAPK pathway: application of a mathematical model.

A mechanistically based mathematical model is used to investigate some of the important factors in priming hepatocytes to enter the G1 phase of the cell cycle. The model considers all of the relevant biochemical mechanisms from signal-receptor binding to the elevation of AP-1 (activation protein transcription factor) levels. Focus is centered on the chain of biochemical events governing the sequential activation of protein kinase C (PKC), mitogen-activated protein kinase (MAPK) and AP-1. Factors such as amplitude and duration of growth factors signals, the kinetics of guanosine diphosphate (GDP) to guanosine triphosphate (GTP) conversion, and the negative feedback control mechanisms governing initial steps in cellular replication were theoretically examined. The results of our theoretical assessments support the finding that specific mutations along the PKC-AP1 pathways can have a critical effect on the rate at which cells enter the division cycle.

Physiologically based pharmacokinetics model of primidone and its metabolites phenobarbital and phenylethylmalonamide in humans, rats, and mice.

Physiologically based pharmacokinetic modeling of the parent chemical primidone and its two metabolites phenobarbital and phenylethylmalonamide (PEMA) was applied to investigate the differences of primidone metabolism among humans, rats, and mice. The model simulated previously published pharmacokinetic data of the parent chemical and its metabolites in plasma and brain tissues from separate studies of the three species. Metabolism of primidone and its metabolites varied widely among a sample of three human subjects from two separate studies. Estimated primidone metabolism, as expressed by the maximal velocity Vmax, ranged from 0 to 0.24 mg. min-1.kg-1 for the production of phenobarbital and from 0.003 to 0. 02 mg.min-1.kg-1 for the production of PEMA among three human subjects. Further model simulations indicated that rats were more efficient at producing and clearing phenobarbital and PEMA than mice. However, the overall metabolism profile of primidone and its metabolites in mice indicated that mice were at higher risk of toxicity owing to higher residence of phenobarbital in their tissues and owing to the carcinogenic potential of phenobarbital as illustrated in long-term bioassays. This result was in agreement with a recently finished National Toxicology Program (NTP) carcinogenicity study of primidone in rats and mice.

Enhanced regional expression of glutathione S-transferase P1-1 with colocalized AP-1 and CYP 1A2 induction in chlorobenzene-induced porphyria.

A distinct, nonfocal expression pattern was observed for glutathione S-transferase P1-1 (rGSTP1-1) in rats exposed to either hexachloro-(HCB) or pentachlorobenzene (PeCB). The nonfocal expression was localized to the centrilobular region with the most intense staining nearest the central vein. A Western blot analysis revealed a 5- and 15-fold induction of rGSTP1-1 with PeCB and HCB treatment on an equal molar basis, respectively. Evaluation of porphyrin fluorescence also revealed a centrilobular accumulation with average porphyrin measurements of 0.319, 0.590, and 0.206 micrograms/g tissue for PeCB, HCB, and corn-oil controls, respectively. Due to the role of Activator Protein-1 (AP-1) in rGSTP1-1 expression and CYP 1A2 in the pathogenesis of porphyria cutanea tarda, immunohistochemical localization of c-jun, c-fos, and CYP 1A2 was also performed. Increased expression and colocalization within the liver lobule was observed for c-jun, c-fos, CYP 1A2, rGSTP1-1, and areas of porphyrin accumulation. These observations are consistent with studies that have associated the induction of GST-P with jun- and fos-related gene products.

Integrated approaches for the analysis of toxicologic interactions of chemical mixtures.

Although an overwhelmingly large portion of the resources in toxicologic research is devoted to single chemical studies, the toxicology of chemical mixtures, not single chemicals, is the real issue regarding health effects of environmental and/or occupational exposure to chemicals. The relative lack of activities in the area of toxicology of chemical mixtures does not suggest ignorance of the importance of the issue by the toxicology community. Instead, it is a reflection of the difficulty, complexity, and controversy surrounding this area of research. Until recently, much of the literature on the toxicology of chemical mixtures has been either very focused on certain specific interaction studies or slanted toward broad-based, relatively vague theoretical deliberation. The typical interaction study involved binary mixtures at relatively high dose levels with acute toxicities as endpoints. Although the theoretical papers have been valuable contributions, little is available on actual, practical experimental approaches toward a systematic solution of this immensely complex area of research. We present here a broad discussion on the important issues of the toxicology of chemical mixtures. First, we provide some background information with respect to the problem and significance of toxicology of chemical mixtures in relation to some of the real life issues. Second, we review and compare the existing experimental approaches relevant to toxicologic interactions of chemical mixtures. Third, we propose three integrated approaches that involve the combination of physiologically based pharmacokinetic/pharmacodynamic (PBPK/PD) modeling with: (1) Monte Carlo simulation, (2) median effect principle (MEP), and (3) response surface methodology (RSM). These modeling approaches, coupled with very focused mechanistically based toxicology studies, could be the basis for solving the problems of toxicology and risk assessment of chemical mixtures.

Physiologically based pharmacodynamic modeling of an interaction threshold between trichloroethylene and 1,1-dichloroethylene in Fischer 344 rats.

Physiologically based pharmacokinetic modeling (PBPK) and gas uptake experiments have been used by researchers to demonstrate the competitive inhibition mechanism between trichloroethylene (TCE) and 1,1-dichloroethylene (DCE). Expanding on their work, we showed that this pharmacokinetic interaction was absent at levels of 100 ppm or less of either chemical in gas uptake systems. In this study, we further illustrate the presence of such an interaction threshold at the pharmacodynamic level by examining the interaction effect of either chemical on the other's ability to bind and deplete hepatic glutathione (GSH) in Fischer 344 rats. However, at this end point, the pharmacodynamic interaction is complicated by the ability of the liver to resynthesize GSH in response to its depletion. To quantitatively resolve the interaction effects on GSH content from the resynthesis effects, physiologically based pharmacodynamic (PBPD) modeling is applied. Initially, the PBPD model description of hepatic GSH kinetics was calibrated against previously published data and by gas uptake experiments conducted in our laboratory. Then, the model was used to determine the duration of the gas uptake exposure experiments by identifying the critical time point at which hepatic GSH is at a minimum in response to both chemicals. Subsequently, gas uptake experiments were designed following the PBPK/PD model predictions. In these model-directed experiments, DCE was the only chemical capable of significantly depleting hepatic GSH. The application of TCE to the rats at concentrations higher than 100 ppm obstructed the ability of DCE to deplete hepatic GSH. Since the metabolites of DCE bind to hepatic GSH, this obstruction signaled the presence of metabolic inhibition by TCE. However, TCE, at concentrations less than 100 ppm, was not effective in inhibiting DCE from significantly depleting hepatic GSH. The same observations were made when the ability of DCE to cause hepatic injury, as measured by aspartate aminotransferase serum activity, was assessed. Both conclusions validated the previous findings of the presence of the interaction threshold at the pharmacokinetic level.

Exploration of an interaction threshold for the joint toxicity of trichloroethylene and 1,1-dichloroethylene: utilization of a PBPK model.

Physiologically based pharmacokinetic (PBPK) modeling and gas uptake experiments were utilized to verify the competitive inhibition mechanism of interaction between trichloroethylene (TCE) and 1,1-dichloroethylene (DCE) and to investigate the presence of an interaction threshold between the two chemicals. Initially, gas uptake experiments were conducted on Fischer 344 rats where the initial concentrations of both DCE and TCE were 2000:0, 0:2000, 2000:2000, 1000:0, 1000:1000, and 500:500 ppm, respectively. When the different modes of inhibition interactions (competitive, uncompetitive and noncompetitive) were employed in the PBPK model, the model description of the competitive inhibition provided the best description of the declining concentrations in the gas uptake chamber. Furthermore, to predict the range at which the interaction threshold would be found, the PBPK model included a mathematical description of the percentage of enzyme sites occupied by either chemical in the presence or the absence of the other. By comparing the percentage of occupied sites by one chemical, in the presence of the other, to those sites occupied in the absence of the latter, the PBPK model predicted a range of concentrations (100 ppm or less) of either chemical where the competitive inhibition interaction would not be observed. Consequently, gas uptake experiments were designed where the initial concentration was selected at 2000 ppm for one chemical while the other chemical was set at 100 in one experiment and 50 ppm in another. Under these conditions, the best stimulation to the concentration depletion curves in the gas uptake system of the chemical in the higher concentration was obtained when the PBPK model was run under the assumption of no-interaction. This substantiated the model predictions of the presence of observable interaction only at concentrations higher than 100 ppm.

Physiologically based pharmacokinetic/pharmacodynamic modeling of the toxicologic interaction between carbon tetrachloride and Kepone.

Carbon tetrachloride (CCl4) lethality in Sprague-Dawley rats is greatly amplified by pretreatment of Kepone (decachlorooctahydro-1,3,2-metheno-2H-cyclobuta[cd] pentalen-2-one). The increase in lethality was attributed to the obstruction of liver regenerative processes. These processes are essential for restoring the liver to its full functional capacity following injury by CCl4. Based on the available mechanistic information on Kepone/CCl4 interaction, a physiologically based pharmacokinetic/pharmacodynamic (PBPK/PD) model was constructed where the following effects of Kepone on CCl4 toxicity are incorporated: (1) inhibition of mitosis; (2) reduction of repair mechanism of hepatocellular injury; (3) suppression of phagocytosis. The PBPK/PD model provided computer simulation consistent with previously published time-course results of hepatotoxicity (i.e., pyknotic, injured and mitotic cells) of CCl4 with or without Kepone. As a further verification of this model, the computer simulations were also consistent with exhalation kinetic data for rats injected with different intraperitoneal (i.p.) doses of CCl4 in our laboratory. Subsequently, the PBPK/PD model, coupled with Monte Carlo simulation, was used to predict lethalities of rats treated with CCl4 alone and CCl4 in combination with Kepone. The experimental lethality studies performed in our laboratories were as follows: Sprague-Dawley rats were given either control diet or diet containing 10 ppm Kepone for 15 days. On day 16, rats in the Kepone treated group were given i.p. doses of 0, 10, 50, and 100 microliters/kg CCl4 (n = 9) while control rats were exposed to 0, 100, 1000, 3000, and 6000 microliters/kg CCl4 (n = 9). Lethality was observed at the 1000 (1/9), 3000 (4/9), and 6000 (8/9) microliters/kg doses for the control group and at the 50 (4/9) and 100 (8/9) microliters/kg for the treated group. Based on Monte Carlo simulation, which was used to run electronically 1000 lethality experiments for each dosing situation, the LD50 estimates for CCl4 toxicity with and without Kepone pretreatment were 47 and 2890 microliters/kg, respectively. Monte Carlo simulation coupled with the PBPK/PD model produced lethality rates which were not significantly different from the observed mortality, with the exception of CCl4 at very high doses (e.g., 6000 microliters/kg, p = 0.014). Deviation at very high doses of the predicted mortality from the observed may be attributed to extrahepatic systemic toxicities of CCl4, or solvent effects on tissues at high concentrations, which were not presently included in the model. Our modeling and experimental results verified the earlier findings of Mehendale (1990) for the 67-fold amplification of CCl4 lethality in the presence of Kepone. However, much of this amplification of CCl4 lethality with Kepone pretreatment was probably due to pharmacokinetic factors, because when target tissue dose (i.e., model estimated amount of CCl4 metabolites) was used to evaluate lethality, this amplification was reduced to 4-fold.

Physiologically based pharmacokinetic/pharmacodynamic modeling of chemical mixtures and possible applications in risk assessment.

Human exposure to chemicals, be it environmental or occupational, is rarely, if ever, limited to a single chemical. Therefore, it is essential that we consider multiple chemical effects and interactions in our risk assessment process. However, with the almost infinitely large number of chemical mixtures in the environment, systematic studies of the toxicology of these chemical mixtures with conventional methodologies and approaches are impossible because of the immense resources and unrealistically long durations required. Thus, the development of predictive and alternative toxicology method is imperative. In order to have a reasonable chance to deal with the complex issue of toxicology of chemical mixtures, we believe that the following concepts must be considered: (1) the exploitation of recent advances in computational technology; (2) the utilization of mathematical/statistical modeling; (3) coupling computer modeling with very focused, mechanistically based, and short-term toxicology studies. Our approach is, therefore, the utilization of physiologically based pharmacokinetic/pharmacodynamic (PB-PK/PD) modeling, coupled with very focused, model-directed toxicology experiments as well as other statistical/mathematical modeling such as isobolographic analysis and response surface methodology. Tissue dosimetry at the pharmacokinetic and pharmacodynamic levels is achievable with simple and complex, but chemically defined, mixtures. Our long-term goal is to formulate innovative risk assessment methodologies for chemical mixtures. In this presentation, one of our specific research projects is described: PB-PK/PD modeling of toxicologic interactions between Kepone and carbon tetrachloride (CCl4) and the coupling of Monte Carlo simulation for the prediction of acute toxicity.

The use of physiologically-based pharmacokinetic/pharmacodynamic dosimetry models for chemical mixtures.

Human exposure to chemicals is rarely, if ever, limited to a single chemical. Therefore, it is essential that we consider multiple chemical effects and interactions in our risk assessment process. However, with the almost infinitely large number of chemical mixtures in the environment, systematic studies of the toxicology of these chemical mixtures with conventional methodologies and approaches are impossible because of the immense resources and unrealistically long durations required. Thus, the development of 'Predictive and Alternative Toxicology' is imperative. At Colorado State University (CSU), our research effort is entirely devoted to this challenge. In order to have a reasonable chance to deal with the complex issue of toxicology of chemical mixtures, we believe that the following concepts must be considered: (1) the utilization of computer, (2) the exploitation of mathematical/statistical methodologies; (3) developing very focused, mechanistically based, and short-term toxicology studies; (4) coupling computer/mathematical modeling with mechanistically-based toxicology. Our strategy therefore the utilization of physiologically-based pharmacokinetic/pharmacodynamic (PBPK/PD) modeling, coupled with very focused, model-directed toxicology experiments as well as other statistical/mathematical methodologies such as Monte Carlo simulation, isobolographic analysis, and response surface methodology. We believe that 'Predictive and Alternative Toxicology' in terms of tissue dosimetry at the pharmacokinetic and pharmacodynamic levels is achievable with simple and complex but chemically defined mixtures. In this presentation, we describe two ongoing research projects as an illustration of our 'Bottom-Up' and 'Top-Down' approaches for handling the chemical mixtures: (1) PBPK/PD modeling of toxicologic interactions between Kepone and carbon tetrachloride (CCl4) and the coupling of Monte Carlo simulation for the prediction of acute toxicity; (2) the conceptual development of PBPK/PD modeling for a more complex chemical mixture of seven groundwater contaminants from hazardous waste sites and the consideration of subfractionation of this chemical mixture.

The application of physiologically based pharmacokinetic/pharmacodynamic (PBPK/PD) modeling for exploring risk assessment approaches of chemical mixtures.

When dealing with health impacts of environmental or occupational exposure such as groundwater contamination from or remediation effort associated with hazardous waste sites, we are obviously not facing individual, single chemicals. Thus, we are immediately confronted with the following questions: (1) Is single chemical risk assessment approach applicable to the multiple chemicals in hazardous waste sites? (2) How do we handle risk assessment of chemical mixtures? Although there were pioneering and commendable efforts from the USEPA to formulate guidelines for risk assessment of chemical mixtures, these guidelines were principally based on additivity concept. As new scientific advances are made, improvement and refinement of risk assessment methodology will be anticipated. At Colorado State University (CSU), our research effort is devoted to the challenges and potential applications of physiologically based pharmacokinetic/pharmacodynamic (PBPK/PD) modeling in the risk assessment of chemical mixtures. With the ultimate goal of Predictive Toxicology, 3 specific research projects are described: (1) PBPK/PD modeling of toxicologic interactions between trichloroethylene (TCE) and 1,1-dichloroethylene (1,1-DCE) and the investigation and defining of an 'Interaction Threshold'; (2) PBPK/PD modeling of toxicologic interactions between Kepone and carbon tetrachloride (CCl4) and the coupling of Monte Carlo simulation for the prediction of acute toxicity; (3) PBPK modeling of the inhibition of pharmacokinetics and enzyme kinetics of TCE caused by low-level, repeated dosing of a chemical mixture of 7 groundwater contaminants. Since this paper is meant to be a commentary and the emphasis is on approaches for dealing with chemical mixtures, detailed presentation of data is avoided. These examples illustrate partially our ongoing research activities and the related ideas with respect to possible novel risk assessment applications to chemical mixtures.