Groupe de recherche en toxicologie humaine (TOXHUM), Faculté de médecine, Université de Montréal, Case Postale 6128, Succursale centre-ville, Montréal, Québec H3C 3J7, Canada
Received January 18, 2001; accepted May 9, 2001
ABSTRACT
The objectives of the present study were: (1) to develop a risk assessment methodology for chemical mixtures that accounts for pharmacokinetic interactions among components, and (2) to apply this methodology to assess the health risk associated with occupational inhalation exposure to airborne mixtures of dichloromethane, benzene, toluene, ethylbenzene, and m-xylene. The basis of the proposed risk assessment methodology relates to the characterization of the change in tissue dose metrics (e.g., area under the concentration-time curve for parent chemical in tissues [AUCtissue], maximal concentration of parent chemical or metabolite [Cmax], quantity metabolized over a period of time) in humans, during mixed exposures using PBPK models. For systemic toxicants, an interaction-based hazard index was calculated using data on tissue dose of mixture constituents. Initially, the AUCtarget tissue (AUCtt) corresponding to guideline values (e.g., threshold limit value [TLV]) of individual chemicals were obtained. Then, the AUCtt for each chemical during mixed exposure was obtained using a mixture PBPK model that accounted for the binary and higher order interactions occurring within the mixture. An interaction-based hazard index was then calculated for each toxic effect by summing the ratio of AUCtt obtained during mixed exposure (predefined mixture) and single exposure (TLV). For the carcinogenic constituents of the mixture, an interaction-based response additivity approach was applied. This method consisted of adding the cancer risk for each constituent, calculated as the product of q*tissue dose and AUCtt. The AUCtt during mixture exposures was obtained using an interaction-based PBPK model. The approaches developed in the present study permit, for the first time, the consideration of the impact of multichemical pharmacokinetic interactions at a quantitative level in mixture risk assessments.
Key Words: mixtures; PBPK modeling; risk assessment; VOCs; pharmacokinetic interactions; hazard index.
Single chemical exposure is an exception rather than the rule in the general and occupational environments. The currently used default mixture risk assessment methodologies do not take into account the consequences of potential interactions occurring between components (U.S. EPA, 1986). The occurrence of pharmacokinetic and pharmacodynamic interactions can result in lower toxicity (antagonism) or greater toxicity (synergism) of mixtures than would be expected based on the knowledge of the potency and dose of the constituents (Calabrese, 1991
). Whereas a mechanistic risk assessment framework for single chemical exposure is fairly well developed (Andersen et al., 1987
), such a framework for characterizing health risk associated with mixture exposure is still in development.
Recent advances in physiologically based pharmacokinetic (PBPK) modeling have demonstrated the feasibility of predicting the change in tissue dose of the components of complex mixtures, due to multiple pharmacokinetic interactions occurring among the constituents (Haddad et al., 1999a, 2000b
; Tardif et al., 1997
). In this modeling framework, information on the pharmacokinetic interactions at the binary level alone are sufficient to predict the magnitude of the interactions occurring in mixtures of greater complexity.
The use of such mixture PBPK models, along with the currently used dose addition and response addition approaches should facilitate the consideration of the consequences of pharmacokinetic interactions for a scientifically sound characterization of risk associated with mixture exposures. The objectives of the present study were: (1) to develop a pharmacokinetic interaction-based risk assessment methodology for mixtures containing systemic toxicants and/or carcinogens, and (2) to apply this methodology to assess the health risk associated with occupational inhalation exposure to mixtures of five volatile organic chemicals (VOCs): dichloromethane, benzene, toluene, ethylbenzene, and m-xylene.
METHODS
Pharmacokinetic interaction-based risk assessment of mixtures of systemic toxicants.
The dose addition or the hazard index (HI) approach is currently used to characterize the risk associated with exposure to noncarcinogenic chemical mixtures (ACGIH, 1999; USEPA, 1986). In this approach, the doses of the mixture components are standardized using health-based values (e.g., acceptable daily intake [ADI], reference dose [RfD], threshold limit values [TLVs]) and are summed as follows:
![]() | (1) |
This approach has been recommended and applied appropriately for components that induce the same toxic effect by identical mechanism of action. In cases where the mixture components act by different mechanisms or affect different target organs, a separate HI calculation is performed for each end point of concern. This approach lacks 2 important notions that should be considered in mechanistic mixture risk assessment: (1) tissue dosimetry of toxic moiety, and (2) possible pharmacokinetic interactions. The denominator and numerator of Equation 1 can be transformed to reflect tissue dose measures that can in turn be obtained using PBPK models. The resulting equation is similar to that proposed by Haddad et al. (1999b) for calculating biological hazard indices for use in biological monitoring of worker exposure to contaminant mixtures at workplaces. Accordingly, the interaction-based HI for systemic toxicant mixtures, based on tissue doses, can be calculated as follows:
![]() | (2) |
Pharmacokinetic interaction-based risk assessment of mixtures of carcinogens.
ACGIH (1999) addresses neither the methodological issues related to the cancer risk assessment of chemical mixtures nor uses of quantitative approaches for the risk assessment of carcinogens. However, the current state of knowledge dictates that the risk assessment of carcinogenic chemical mixtures be conducted per response additivity approach, which involves the summation of excess risk attributed to each carcinogenic mixture constituent (U.S. EPA, 1986):
![]() | (3) |
Like the dose addition approach, the currently used response addition approach neither considers the information on target tissue dose of mixture constituents nor accounts for potential interactions occurring at the pharmacokinetic level. Andersen et al. (1987) developed an approach to incorporate tissue dosimetry into cancer risk assessment of individual chemicals using PBPK modeling. Along those lines, the information on altered tissue dose simulated by mixture PBPK models can be used to account for pharmacokinetic interactions in the calculation of CRM as follows:
![]() | (4) |
The use of q*tti in Equation 4 enables us to calculate the CRM from knowledge of the target tissue dose of mixture components (TMi), which can vary due to pharmacokinetic interactions. PBPK models for individual mixture constituents can be used for estimating q*tti, whereas the mixture PBPK models are of use in estimating TMi by accounting for the interactions occurring among mixture constituents.
Estimating target tissue exposure.
Equations 2 and 4 represent essentially the proposed manner of conducting interaction-based risk assessment of exposure to chemical mixtures. These 2 equations, corresponding to noncancer and cancer risk assessments, require that the estimate of TMi be obtained with PBPK models for mixture exposures. The estimation of the target tissue dose during individual and mixed exposures, in fact, is the crucial step of the proposed risk assessment approach. The appropriate tissue dose metric (e.g., area under the concentration-time curve [AUC] for parent chemical or metabolite, maximal concentration [Cmax] of metabolite or parent chemical in tissues, amount metabolized over a period of time, and average concentration of metabolite in target tissue) should be chosen based on the state of knowledge of the mechanism of toxicity of the mixture constituents (e.g., Andersen et al., 1987).
Tissue dose can be estimated from knowledge of external exposure or administered dose, using PBPK models. These models can adequately simulate the uptake, disposition, and tissue dose of chemicals in various conditions (i.e., species, dose, scenario, and exposure route), because they are based on the mechanisms that account for the biology and chemistry of the organism, and the characteristics of the chemical. During mixed exposures, the pharmacokinetics and tissue dose of a chemical may be modified in the presence of other chemicals. When the mechanisms of interactions are known or hypothesized, it is possible, with PBPK models, to predict the altered pharmacokinetics and tissue dose of the components of a chemical mixture. It has been done for several binary mixtures (reviewed in Krishnan and Brodeur, 1994; Simmons, 1995), and recently for more complex mixtures (Haddad et al. 1999a, 2000b
; Tardif et al. 1997
). The methodology involves linking binary interactions within a PBPK model framework (Fig. 1
) to simulate the kinetics and tissue dose of constituents of mixtures regardless of their complexity (Haddad et al., 2000b
; Haddad and Krishnan, 1998
).
|
The rat model for DBTEX mixture developed and validated by Haddad et al. (2000b) was scaled to a human model by changing the rat physiological (tissue blood flow, alveolar ventilation rate, and cardiac output) and physicochemical (partition coefficients) parameters to human values (Tables 1 and 2) (Andersen et al., 1991
). The biochemical parameters (i.e., inhibition constants, maximal velocity for metabolism scaled to the body weight2/3 and Michaelis affinity constant [Km]) were kept species-invariant, except for the Km of D, which was changed to the human value specified by Andersen et al. (1991; see Table 2
). The D submodel also contained parameters and equations essential for simulating the percent carboxyhemoglobin in blood that resulted from D exposure (Andersen et al., 1991
). The consideration of the species-invariant nature of metabolic interaction constants was based on the previous observations of a mixture PBPK modeling study (Tardif et al., 1997
) in which the rat-human extrapolation of the occurrence of interactions among T, E, and X was validated with experimental data.
|
|
RESULTS
Systemic Risk Assessment for DBTEX Mixtures
The conventional and interaction-based hazard indices for CNS effects and hypoxia for various DBTEX mixtures are presented in Tables 3 and 4. The conventional HI calculations for CNS effects were done using the exposure concentrations of D, T, E, and X, whereas such calculations for hypoxia were done using the exposure concentrations of D. Examining the data for CNS effect, it can be noticed that at high concentrations the HI values, calculated with the consideration of interactions, are greater than those obtained according to the dose-addition approach that did not account for the occurrence of interactions (Table 3
). At lower exposure concentrations of DBTEX in mixtures, the difference between the conventional and interaction-based HI is smaller.
|
|
Cancer Risk Assessment for DBTEX Mixtures
According to the methodology used in the present study, the relative change in cancer risk associated with D and B during mixture exposures is a direct consequence of the change in their tissue dose metrics. The change in risk level during mixture exposures compared to single chemical exposures, as calculated using PBPK model-simulated changes in the tissue doses of D and B, is shown in Table 5. In the case of D, the GSH conjugate is the relevant dose surrogate (Andersen et al., 1991
). In the presence of competitive inhibitors (i.e., BTEX) of P450 metabolism of D, the flux of D through the GSH conjugation pathway increases, thus contributing to a greater risk level during mixed than during single exposures. For the mixture exposure scenarios considered in the present study, the cancer risk attributed to D could increase by up to a factor of 4 compared to single chemical exposure situations (Table 5
). The cancer risk attributed to B exposure, however, would decrease during mixed exposures compared to single chemical exposures, since the rate of formation of oxidative metabolites from B is reduced during concurrent exposure to DTEX (Table 5
). The simulation results presented in Table 5
indicate that the relative cancer risk due to B in DBTEX mixtures approaches unity (i.e., close to the absolute risk level associated with a single exposure to B) as the concentration of DTEX in the mixture decreases.
|
The occupational and environmental health risk assessments of chemical mixtures do not account for the quantitative impact of possible interactions among mixture components, which may vary as a function of dose and exposure scenario in animals and humans. Depending on the relative and absolute concentrations of the chemicals present in the mixture, they may result in interactions that cause departure from additivity. Interactions may be pharmacokinetic or pharmacodynamic in nature. The pharmacokinetic interactions result in a change in tissue dose of chemicals during mixture exposures compared to single exposures, and represent the most common type of interaction observed and reported in the literature (reviewed in Krishnan and Brodeur, 1991, 1994). The relative change in tissue dose of chemicals due to pharmacokinetic interactions during mixture exposures depends on the relative concentrations of components and the mechanism(s) of interactions. PBPK models are unique tools that facilitate the consideration of interaction mechanisms at the binary level to simulate the change in tissue dose of chemicals present in complex mixtures. The present study, for the first time, demonstrates the use of PBPK models in quantifying the change in the tissue dose metrics of chemicals during mixture exposures and in improving the mechanistic basis of mixture risk assessment. The application of PBPK models in mixture risk assessment has been demonstrated in this study using DBTEX mixture, for which an interaction-based PBPK model has recently been developed and validated (Haddad et al., 2000b).
According to the proposed approach, it is possible that HIinteraction-based exceeds 1 while the conventional HI value is less than unity, or vice versa. The interaction-based HI values developed in the present study are more relevant than the conventional HI because internal concentrations of the toxic entities (and not external exposure concentrations) are used for the calculation. The computed HIinteraction-based will not always be different from the conventional HI because its magnitude depends on the relative concentrations of all mixture constituents and the quantitative nature of the interaction mechanisms as included in the PBPK models. When both the HIinteraction-based and conventional HI values exceed 1, the interpretation should be limited to a qualitative indication of health risk being associated with exposure to the given chemical mixture. The difference in numerical values obtained, once they are above 1, should not be interpreted in quantitative risk terms. This is consistent with the current practice of risk assessment for systemic toxicants, either present individually or as mixtures.
The interaction-based PBPK model facilitates the prediction of the change in tissue dose of the toxic moiety of chemicals during mixture exposures to assess the cancer risk for chemical mixtures. In this approach, the potency of the mixture constituents does not change between single and mixture exposures, but it is the tissue dose that changes according to the interaction mechanism and the exposure concentration of interacting chemicals. The proposed approach then improves upon the currently used response-addition methodology by facilitating the incorporation of data on the tissue dose of chemicals in mixtures (instead of their external concentration), and by accounting for the extent of their modulation due to interactions during mixed exposures. During coexposures to chemicals that interact at the metabolic level, the tissue dose and associated cancer risk of mixture constituents may either be decreased or increased (compared to single exposures) as exemplified in this study. The magnitude and direction of the change in tissue dose during mixed exposures depend on the mechanism of pharmacokinetic interactions (e.g., metabolic inhibition or enzymatic induction) and the identity of the putative toxic moiety (e.g., parent chemical, metabolite).
The present study applied the validated rodent PBPK model to characterize the cancer and noncancer risk associated with occupational exposure to the DBTEX mixture of varying compositions by accounting for the change in tissue dose due to metabolic interactions. The simulated changes in tissue dose and risk levels for occupational mixture exposures do not necessarily reflect those that are expected in environmental exposure situations. While comparing the occupational and environmental exposure to mixtures, the interaction mechanisms are likely to remain the same in both situations, whereas the concentrations of the inhibitors differ markedly. With decreasing blood concentrations of the inhibitors, their effect on the metabolism of other mixture components becomes smaller and smaller. Using the mixture PBPK model developed in the current study, a threshold of interactions in multichemical mixtures can be established following the simulation of the exposure level impact on the magnitude of interactions. Such studies should facilitate a better understanding of the relative importance and relevance of specific interactions and interaction mechanisms in occupational and environmental exposure situations.
Even though the mixture model used in the present study accounted for the occurrence of metabolic inhibition as the interaction mechanism (Haddad et al. 2000b), induction of metabolism may occur during repeated exposure scenarios, complicating the PBPK model calculation of the magnitude of net change in tissue dose during mixed exposures. However, experimental studies have shown the absence of induction effects on D, B, T, E, and X during repeated exposures (Haddad et al., 2000a
). Therefore, the assessment presented in this paper, based on the consideration of the inhibition mechanism, is likely to describe adequately the pharmacokinetic interactions occurring in the DBTEX mixture and ensuing changes in tissue dose of the mixture constituents. The possible impact of pharmacodynamic interactions on the mixture risk was not evaluated in the present study, but it can be performed if quantitative mechanistic data on binary level interactions are available/generated. Overall, the modeling and risk assessment frameworks outlined in this study should be amenable to the use of data on other mechanisms of interactions, toxic endpoints, and dose-response relationships, if intended and if the required data are available.
An advantage of the PBPK model-based risk assessment methodology developed in this study is that the combinations of exposure concentrations of individual chemicals that will not deviate significantly from the conventional HI (i.e., < 1) or the CRM (i.e., < 1 x 106) can be determined by iterative simulation. The proposed approach should then be useful from health protection and prevention perspectives, particularly where there is a possibility of pharmacokinetic interactions among chemicals present as mixtures in the occupational environment.
ACKNOWLEDGMENTS
This work was supported by Research Grants from the Canadian Network of Toxicology Centres (CNTC), Toxic Substances Research Initiative (TSRI) of Health Canada, and Fonds de la Recherche en Santé du Québec (FSRQ). K. K. is recipient of a Research Scholarship from FRSQ (19922004).
NOTES
1 To whom correspondence should be addressed at Département de santé environnementale et santé au travail, Université de Montréal, 2375 Côte Ste-Catherine Bureau 4105, Montréal, Québec H3T 1A8, Canada. Fax: (514) 343-2200. E-mail: kannan.krishnan{at}umontreal.ca.
REFERENCES
ACGIH (1999). TLVs® and BEIs®: Threshold Limit Values for Chemical Substances and Physical Agents; Biological Exposure Indices. American Conference of Governmental Industrial Hygienists. ACGIH Worldwide, Cincinnati.
Andersen, M. E., Clewell, H. J. I., Gargas, M. L., MacNaughton, M. G., Reitz, R. H., Nolan, R. J., and McKenna, M. J. (1991). Physiologically-based pharmacokinetic modeling with dichloromethane, its metabolite carbon monoxide and blood carboxyhemoglobin in rats and humans. Toxicol. Appl. Pharmacol. 108, 1427.[ISI][Medline]
Andersen, M. E., Clewell, H. J., III, Gargas, M. L., Smith, F. A., and Reitz, R. H. (1987). Physiologically based pharmacokinetics and the risk assessment process for methylene chloride. Toxicol. Appl. Pharmacol. 87, 185205.[ISI][Medline]
Calabrese, E. J. (1991). Multiple Chemical Interactions. Lewis Publishers, Chelsea, MI.
Cox, L. A., Jr., and Ricci, P. F. (1992). Reassessing benzene cancer risks using internal doses. Risk Anal. 12, 401410.[ISI][Medline]
Haddad, S., Charest-Tardif, G., and Krishnan, K. (2000a). Physiologically based modeling of the maximal effect of metabolic interactions on the kinetics of components of complex chemical mixtures. J. Toxicol. Environ. Health 61, 209223.[ISI]
Haddad, S., Charest-Tardif, G., Tardif, R., and Krishnan, K. (1999a). Physiological modeling of the pharmacokinetic interactions in a quaternary mixture of aromatic hydrocarbons. Toxicol. Appl. Pharmacol. 161, 249257.[ISI][Medline]
Haddad, S., Charest-Tardif, G., Tardif, R., and Krishnan, K. (2000b). Validation of a physiological modeling framework for simulating the toxicokinetics of chemicals in mixtures. Toxicol. Appl. Pharmacol. 167, 199209.[ISI][Medline]
Haddad, S., and Krishnan, K. (1998). Physiological modeling of toxicokinetic interactions: Implications for mixture risk assessment. Environ. Health Perspect. 106(Suppl. 6), 13771384.[ISI][Medline]
Haddad, S., Tardif, R., Viau, C., and Krishnan, K. (1999). A modeling approach to account for pharmacokinetic interactions in the calculation of biological hazard index for chemical mixtures. Toxicol. Lett. 108, 303308.[ISI][Medline]
Krishnan, K., Andersen, M. E., Clewell, H. J., III, and Yang, R. S. H. (1994). Physiologically based pharmacokinetic modeling of chemical mixtures. In Toxicology of Mixtures: Case Studies, Mechanisms, and Novel Approaches (R. S. H. Yang, Ed.), pp. 399437. Academic Press, San Diego.
Krishnan, K., and Brodeur, J. (1991). Toxicological consequences of combined exposure to environmental pollutants. Arch. Compl. Environ. Stud. 3, 1106.
Krishnan, K., and Brodeur, J. (1994). Toxic interactions among environmental pollutants: Corroborating laboratory observations with human experience. Environ. Health Perspect. 102, 1117.
Mumtaz, M. M., and Hertzberg, R. C., (1993). The status of data in risk assessment of chemical mixtures. In Hazard Assessment of Chemicals (J. Saxena, Ed.), Vol. 8, pp. 4779. Hemisphere, Washington, DC.
Mumtaz, M. M., Sipes, I. G., Clewell, H. J., and Yang, R. S. (1993). Risk assessment of chemical mixtures: Biologic and toxicologic issues (symposium overview). Fundam. Appl. Toxicol. 21, 258269.[ISI][Medline]
Simmons, J. E. (1995). Chemical mixtures: Challenge for toxicology and risk assessment. Toxicology 105, 111119.[ISI][Medline]
Tardif, R., Charest-Tardif, G., Brodeur, J., and Krishnan, K. (1997). Physiologically based pharmacokinetic modeling of a ternary mixture of alkyl benzenes. Toxicol. Appl. Pharmacol. 144, 120134.[ISI][Medline]
U.S. EPA (1986). Guidelines for the health-risk assessment of chemical mixtures. U.S. Environmental Protection Agency. Fed. Reg. 51, 3401434025.