![[logo] US EPA](http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
Research Highlights
Modeling the Health Effects of Harmful Agents
Equation Relating Biological Activity to Chemical Structure
T=s(d)+C T=the biological or toxicological endpoint s=the coefficient associated with the descriptors d=the descriptors computed for each molecule of interest C=a constant QSARs and VFARsThe models to supplement the risk-assessment process are the Quantitative Structure Activity Relationships (QSARs) and Virulence Factor Activity Relationships (VFARs). QSARs are mathematical equations—or models—that describe the correlations between various features of a chemical’s molecular structure and its observed toxic effects. They can be used to calculate the likely health effects of new chemicals or of chemicals whose toxicity is not well characterized. VFARs are models for biological agents and are developed to relate the architectural and biochemical components of a microorganism to its potential to cause human disease.Developing QSAR and VFAR ProductsEPA currently has four projects related to the development of QSAR models. The first is a literature search for information on chemicals (e.g., organophosphates, aldehydes). The objective of the search is to compile quantitative estimates of dose-response metrics for inhalation and/or oral exposures—for example, Lethal Dose 50 (LD50), Lowest Observable Adverse Effect Level (LOAEL), and Maximum Tolerated Dose (MTD). The data are being compiled for all species; however, our current focus is on less-than-lifetime exposures.From this literature search, between 50 and 200 chemicals/agents rich in toxicological data will be identified for each chemical class. For each of these identified agents, researchers will compile available information on the toxicokinetics (e.g., absorption, excretion) and toxico-dynamics (i.e., how the chemicals or their metabolites interact with specific bodily processes to result in toxicity). Any published cancer classifications or data on modes or mechanisms of action will be gathered as well. The second project is the development of a statistically based QSAR for at least 200 chemicals. Experimental data will be used to design a robust predictive model that will employ QSARs to estimate the LOAELs for a number of chemicals, particularly those for which there are no experimental toxicity data. The predictive ability of this QSAR model will be tested with about 40 chemicals. The third and fourth projects will make use of the results of the literature search to develop statistically based predictive QSAR models for approximately 200 to 500 chemicals that might be ingested or inhaled. Experimental toxicity benchmark data (such as LD50 and LOAEL) for various health effects will be used to develop QSARs for estimating the acute, subacute, and subchronic toxicity of these chemicals. A separate QSAR model will be developed for each benchmark and exposure route. Where data are available, the QSARs will not only predict the toxicity benchmarks, but also recognize the varying toxic effects of isomers. The predictive ability of these QSAR models will be tested on a number of chemicals with known toxicities. The bioinformatics data needed to develop VFARs are very sparse at present. EPA will conduct a feasibility study to assess the possibility of obtaining enough data to pursue the development of VFAR models. Contact: Chandrika Moudgal |