ADVANCED COMPUTATIONAL METHODS IN DOSE MODELING
Impact/Purpose:
1) Evaluate and apply advanced computational techniques and emerging 'omics technologies to improve dose modeling and, thereby, reduce uncertainty in human exposure and risk assessment.
2) Employ QSAR-based methods, including quantum-chemical approaches, to estimate critical physiochemical and biochemical constants used in PBPK/PD models.
3) Contribute to next-generation PBPK/PD models by developing methods to predict absorption, distribution, metabolism, elimination (ADME) and toxic effects on the basis of physicochemical properties.
Description:
The overall goal of the EPA-ORD NERL research program on Computational Toxicology (CompTox) is to provide the Agency with the tools of modern chemistry, biology, and computing to improve quantitative risk assessments and reduce uncertainties in the source-to-adverse outcome continuum (SOC). NERL has found it useful to envision the risk assessment paradigm as a continuum of events leading from source-to-adverse outcome. Modeling the intervening steps along the continuum is critical to identifying the factors that control the relationship between effects and source/exposure (the risk factors). The goals of this task are to: 1) evaluate and implement state-of-the-art computational methodologies used in the prediction of ADME (absorption, distribution, metabolism, excretion) parameters for chemicals and chemical mixtures and 2) provide an unbiased estimate of differential organismal state even in the absence of traditional quantifiable health effects of exposure.
Numerous physiologically based pharmacokinetic (PBPK) models have been developed for use in studies of pharmaceutical dose and efficacy, as well as for risk assessments of environmental chemicals (Anderson et al., 1997; Slob et al., 1997; Corley et al., 1997; Corely et al., 2000; Timchalk et al., 2002; Luttringer et al., 2003; Knaak et al., 2004). PBPK models are tractable because they allow the integration of disparate forms of data (i.e., in vitro, in vivo, and in silico) to accurately predict pharmacokinetic behavior of absorbed pollutants. In addition, they provide a platform for extrapolating among species, route, dose and exposure scenarios. The future development of PBPK models requires improved methods for estimating and simulating ADME parameters and predicting toxic effects of a chemical based on its physicochemical properties.
Quantitative structure-activity relationships (QSARs) are well suited for estimating certain ADME parameters ( Poulin and Krishnan, 1995; Rao et al., 2000; Fouchecourt et al., 2001; Shin-Ichi Fujiwara et al., 2003; Votano et al., 2004). In the QSAR approach, statistical correlations are found between chemical structural indices and chemical activity or behavior in biological systems (Hansch, 1971; Hansch and Leo, 1979; Goldberg, 1983). The defined relationships are then used to assign parameter values to new chemicals, or groups of chemicals, for which activity data are not available. Property-based and structure-based QSARs can reduce bias and increase the precision of parameter estimation (i.e., equilibrium binding constant, metabolic rate constants, and enzyme inhibition constants).
The mass of data produced by emerging high-throughput 'omics (i.e. genomics, metabonomics and proteomics) technologies necessitates the expansion of traditional QSARs to include 'omics profiles as additional endpoints of evaluation (Darvas and Droman, 2002; Ansede and Thakker, 2004). Concurrent evaluation of standard toxicological endpoints will provide us with the information that we need to utilize newly emerging omics technologies in modeling efforts for predictive toxicology. Other descriptors for the QSARs may be obtained from quantum mechanical or molecular dynamics predictions, that may ultimately lead to parameter estimates from ab initio calculations. The application of the values for specific molecular interactions to human physiology requires detailed models of organ systems. Thus, the data and predictions may be incorporated into PBPK models to evaluate risk. The need for novel in silico approaches is all the more pressing given the number of chemicals that the Agency must consider under different regulations (e.g., FQPA) and the emerging interest in cumulative assessments of chemical mixtures. Clearly, testing on a chemical-by-chemical basis using standard approaches is no longer sustainable.
Record Details:
Record Type:PROJECT
Start Date:10/01/2005
Projected Completion Date:10/01/2007
OMB Category:Other
Record ID:
135585
Keywords:
GENOMICS, PROTEOMICS, METABONOMICS, QSAR, PBPK, SYSTEMS BIOLOGY,
Project Information:
Progress
:This task begins in FY05, so preliminary work is beinghas been conducted under FY04 task #3906. A recent publication detailed the use of computational approaches to predict relative rates of oxidation for haloalkanes. With the assistance of Dr. JP Jones of Washington State University, a semi-empircal quantum mechanical model for cytochrome P450 was employed to predict rates of oxidation of haloalkanes on the EPA CCL (Contaminant Candidate List; http://www.epa.gov/safewater/ccl/cclfs.html). The approach described in the manuscript will be employed to predict rates of oxidation of pyrethroid and carbamate pesticides.
Tornero-Velez R, Ross MK, Granville C, Laskey J, Jones JP, DeMarini DM, Evans MV. Metabolism and Mutagenicity of Source Water Contaminants 1,3-Dichloropropane and 2,2-Dichloropropane. Drug Metabolism and Disposition 32:123-131, (2004).
A proposal has been written for a new Gordon Research Conference, "Integrative Modeling of Physiological Processes". The conference series will begin in early 2006. The goal of this conference series will be to improve our fundamental understanding of the relationships among currently disparate exposure, dose and toxicity data in animal systems (including humans) and the degree to which those relationships can accurately be extrapolated to other systems.
Relevance
: Risk assessments that incorporate well-tested PBPK models can provide accurate quantitative risk assessments when empirical data are available (Okino et al., 2003; Knaak et al., 2004; Dary et al., 2004; Okino et al., 2004). Since source-to-dose modeling is a sequential process involving multiple components, the reliability of risk assessments, based on these models, is limited by the accuracy of each component. Computational chemistry approaches such as quantitative structure-activity relationships (QSAR) offer an attractive alternative to costly and more traditional forms of parameter estimation ( Fujita, 1995; Hansch and Leo, 1979; Goldberg, 1983; Fouchecourt et al., 2001). The integration of QSAR with a PBPK model framework (Darvas and Droman, 2002) will refine the understanding of source-to-dose relationships and support the overarching goal of EPA?s Computational Toxicology initiative, i.e., to enable health-risk assessment of chemicals without the need for extensive in vivo animal testing of each new chemical. A QSAR-based PBPK model will enhance source-to-outcome predictions via the capacity for extrapolation across species, route, dose, and exposure scenario. This task was prepared to support the Food Quality Protection Act (FQPA) through Goal 4: Healthy Communities and Ecosystems of EPA?s Strategic Plan (Directions for the future) that calls for research to reduce exposure to toxic pesticides by 2008. EPA expects to achieve this goal by developing methodologies, data, models, risk assessment guidelines and toxicity testing methods and protocols. Through this task, ORD-NERL shall assemble QSAR data bases that interact with PBPK/PD models to address chemical class specific cumulative risk assessments. These assessments pertain to chemicals that act by common mechanisms of toxicity, e.g., cholinesterase-inhibiting neurotoxic insecticides ( Mileson et al., 1999, A framework for cumulative risk assessment: An ILSI workshop report; U.S. EPA, 2000: Cumulative Risk; A case study of the estimation of risk from 24 organophosphorus [sic] pesticides, U.S. EPA, 2002: Guidance on cumulative risk assessment of pesticides having a common mechanism of toxicity: http://www.epa.gov/pesticides/cumulative/rra-op/ June 10). As QSAR and PBPK/PD methodologies evolve, it is expected that modeling approaches will become increasingly congener-independent such that cumulative risk assessments may be performed on mixtures having different modes of action, e.g., cholinesterase-inhibiting insecticides (OP and carbamate) and other neurotoxic insecticides (pyrethoids). This foreword approach is important to the continued development and registration of new reduced risk chemicals (Goal 4 sub-objective 4.1.2: License pesticides meeting safety standards). QSAR and PBPK/PD in silico approaches are expected to provide a means to rapidly test exposure assumptions in the absence of data. This is because, as pollutant sources and exposure profiles may change with time, dose models elaborated through computational approaches are flexible to meet these changes. Hence, the long-term impact of this task is to provide the foundation for QSAR and PBPK/PD models that permit efficient and adaptable assessments of source-to-outcome relationships. Moreover, model results/predictions are expected to provide focus for validation experiments. Ultimately, QSAR and PBPK/PD in silico approaches are expected to reduce and in certain cases eliminate the need for laboratory animal and human subject testing.
Clients
:NERL, NHEERL, NCEA, OPPTS, OPP, Scientific community
Project IDs:
ID Code
:20707
Project type
:OMIS