U.S. Environmental Protection Agency, Program Contract Service Center (3803R), 401 M Street, SW, Washington, DC 20460-0001

B -- DEVELOP FRAMEWORK FOR ESTIMATING THE UPPER-BOUND RISKS OF MIXTURES OF CHEMICALS SOL RFQ-DC-98-00217 DUE 091898 POC Valoree S. Lilley, lilley.valoree@epamail.epa.gov E-MAIL: click here to contact the contract specialist, lilley.valoree@epamail.epa.gov. The Environmental Protection Agency under the Simplified Acquisition Procedures of FAR Part 13, intends to issue a sole source purchase order to Dr. R. M. Putzrath. This SIC code for this requirement is 8731. The EPA is FACNET certified. BACKGROUND: Methods currently used by EPA for evaluating risks from exposure to mixtures of chemicals do not explicitly address how uncertainty and variability will be evaluated. Some methods combine upper-bound risks or lower-bound doses of component chemicals while others do not explicitly address the issue of uncertainty and variability. In the absence of mixture- specific information, the risks may be over- or under-estimated. Even when risks for individual pesticides on foods are appropriately regulated, regulators and consumers still need to evaluate the risks of mixtures of chemicals with uncertainties and variabilities treated in comparable ways. The goals of this research are two-fold: (1) to determine mathematically and toxicologically appropriate methods forestimating upper-bound risks for mixtures of chemicals and (2) to examine the effect of choice of method for estimating the cumulative risk of mixtures on the upper-bound estimates of risk. The upper-bound estimate will depend on both the mixture model and the method used to estimate the combined uncertainty and/or variability. At question is whether these differences will be significant within a regulatory framework. The Guidance on cumulative risk planning and scoping states that cumulative risk under the Food Quality Protection Act (FQPA) may be defined using terms such as aggregate exposure by multiple pathways and routes of exposure or cumulative effects to mixtures of pesticide chemical residues with the same mechanism of toxicity. Simply adding upper-bound risks can serve as a screen, i.e., a bounding risk that if not exceeded is unlikely that a true risk exists. Performing arithmetical procedures on upper-bounds of risk or lower-bounds of dose, however, is mathematically imprecise. Estimating the upper-bound risk from exposure to mixtures of chemicals as precisely as possible is essential to achieve the goals of adequate protection while guarding against regulating overestimated risks. For pesticides in particular, if a procedure produces exaggerated risks another risk may result, e.g., not obtaining the health benefits of eating fruits and vegetables because the estimated risks seem unacceptable. More accurate methods for estimating upper-bounds of aggregated and cumulative risks exist. These include combining distributions by Monte Carlo simulations or by combining best estimates and calculating a combined upper-bound through meta-analytic techniques. In addition, providing more accurate information on the upper-bound risks, when such analysis match the needs of the decision, is in accord with the National Academy of Sciences' recommendations. Current methods for evaluating risks from exposure to mixtures of chemicals do not explicitly address the issue of differences in the uncertainty of our knowledge about the toxicity of the individual components of the mixture. Depending on the method used for estimating the cumulative risk, different methods will be necessary to consider the effect of uncertainty and variability on the estimated risk. For example for non-cancer risks, the hazard index method uses acceptable levels of exposure such as RfDs, each of which includes an estimate of the individual constituent's uncertainty and variability. The cumulative evaluation, however, is based on fractions of lower-bound doses, which is a less than optimal method for estimating the combined hazard. Similarly, the method for evaluating combined carcinogenic effects adds upper-bound risk estimates that are based on upper-bound uncertainty estimates of carcinogenic potency and the same mathematical limitations apply. In contrast, toxicity equivalence factors (TEFs) may be based on a best estimate of the relative potency of the constituents or they may be designed to be protective, i.e., upper-bound estimates of the relative potency, to account for variations across endpoints or species. Methods for estimating the cumulative uncertainty for the mixture, therefore, would depend on which of these methods were used. Both the hazard index and TEF approaches are likely to require use of the no observed adverse effect level (NOAEL), uncertainty factors (UFs), and modifying factors (MFs) that comprise the current estimation of an RfD for the individual chemicals. In addition to the two methods currently in use by EPA, other methods for estimating the risk of mixtures that have been published in the peer-reviewed literature may be considered, if their use provide additional insight in determining the uncertainty associated with the risk of a mixture of chemicals. The method for estimating the uncertainty of the cumulative risk will again depend on the model selected. The method frequently used for determining the cancer risk from a mixture of chemicals involves adding the upper-bound risks from the individual chemicals. This methodsuffers from the same limitations as combining RfDs, i.e., arithmetic procedures should not be performed on the bounds. Therefore, techniques similar to those used for non-cancer risk assessment, such as Monte Carlo simulations or meta-analysis, will likely provide more accurate estimates of the upper-bound for the risk for the mixture. Some of the constraints on upper-bound risks for mixtures of chemicals have been examined, although under the constraint of a relatively large number of assumptions. The proposed Guidance for risk assessment of carcinogens suggests that a point of departure from observed data be used as the starting point for both linear and non-linear default extrapolations. The point of departure linear method not only simplifies determining the risk for individual chemicals, but may also simplify the estimation of an upper-bound risk for mixtures of carcinogens. The second application using the point of departure is the margin of exposure (MOE) for cases where the dose-response curve is more likely to be non-linear. Methods for evaluating the upper-bound risks with the combined uncertainty and variability for mixtures of such carcinogens may be limited to simulations such as Monte Carlo methods. Uncertainty, variability, and quality of the data and methods for estimating the aggregate risks could be large. And the permutation for combining these factors are large. To succeed the initial research will be limited to mixtures of chemicals that act by the same mechanism, as per FQPA and Presidential/Congressional Commission. Variability due to factors such as age or gender may require different methods for combining risks across mixtures or routes of exposure. Alternatively, separate assessments of the risks of mixtures may be required to portray uncertainty for each of the components of variability, as recommended by the NAS. This approach may be particularly useful in communicating differences due to variability for easily identifiable subgroups as the increased information should enhance both the clarity and the transparency of the analysis and allow for better understanding by stakeholders.STATEMENT OF WORK: The contractor will develop a framework for estimating the upper-bound risks of mixtures of chemicals using the uncertainty and variability of the individual constituents of the mixture. To proceed, a series of case studies of pesticides with similar mechanisms of action, e.g., organophosphates or carbamates will be analyzed. The focus of the efforts is to produce results immediately applicable under the provisions of the FQPA. For non-cancer effects, the contractor will start by examining the science, mathematics, and assumptions under the hazard index and TEF methods. The contractor will examine at least one of the other method for estimating risks of mixtures published in the peer-reviewed literature. For carcinogenic effects, the contractor will examine the two methods proposed in the 1996 EPA cancer assessment guidlines. The case studies are to provide a series of examples which may include: several endpoints, several methods for combining the risks, or several methods for evaluating uncertainty and variability of the mixture from the uncertainty and variability of the constituents. The contractor will examine the results of the case studies along with the methods currently used for evaluating mixtures. The strengths and limitations of the various approaches will be examined to evaluate the various methods for using existing information to estimate the upper-bound risk for the mixture. No new toxicological assessments is expected to be performed. Instead the contractor will use evaluated chemicals/pesticides and prior decisions on methods and assumptions of EPA evaluated chemicals. Thus, if an RfD is disaggregated into its NOAEL, UFs, and MFs for reanalysis as part of a mixture, the values for each of these parameters will be retained. Similarly, as points of departure are developed for various carcinogens, the contractor will use these values. Alternatively, the contractor will determine a "point of departure" from existing cancer potency factors for the selected chemicals in order to develop the methods, even if the numbers may need to be changed as the new guidance is implemented. After sufficient results have been developed, the contractor will work with the EPA project officer to plan and convene a panel of qualified and interested EPA professionals to review the initial results for ideas for improvement. Final and significant interim findings will be presented at professional society meetings as well as at an open meeting at EPA headquarters. After the results and conclusions are complete, the contractor will assist the EPA project officer to plan a workshop to an appropriate professional society, such as the Society for Risk Analysis. SEE Note 22 Posted 09/01/98 (W-SN244442). (0244)

Loren Data Corp. http://www.ld.com (SYN# 0015 19980903\B-0002.SOL)