Endocrine Disruptor Risk Characterization: An EPA Perspective

REGULATORY TOXICOLOGY AND PHARMACOLOGY ARTICLE NO.

26, 70–73 (1997)

RT971122

Endocrine Disruptor Risk Characterization: An EPA Perspective1,2 Penelope A. Fenner-Crisp, Ph.D. Office of Pesticide Programs (7501C), U.S. Environmental Protection Agency, 401 M Street, S.W., Washington, DC 20460 Received January 31, 1997

mental Protection Agency (Browner, 1995). It was acknowledged that this policy and its associated guidance, when implemented, would be useful in the Agency’s attempts to be responsive and open in communicating to the public the complexities and challenges of environmental decision making in the face of scientific uncertainty. Supported by the core values of transparency, clarity, consistency, and reasonableness, implementation of this policy should lead to a more open decision-making process, improved decisions, and greater public support for, and understanding of, the regulatory decisions the Agency makes. When the policy was issued, it was focused, nearly exclusively, on characterization of the risks to human health. Nonetheless, EPA believes that the core values also apply to the evaluation of ecological risks. At the same time Agency offices have been developing and carrying out their implementation plans for human health risk characterization, work has been ongoing to develop the approaches for ecological risk characterization. These activities, informed by the Framework for Ecological Risk Assessment (U.S. EPA, 1992a) and the proposed Ecological Risk Assessment guidelines (U.S. EPA, 1996b), will lead to the development of Agency guidance in the not-too-distant future. The 1995 policy includes guidance which addresses the development of chemical-specific risk characterizations, given that the traditional focus has been on the evaluation of single agents or, at best, complex mixtures of structurally similar chemicals such as PCBs or PAHs. In the future, the Agency expects to conduct more complex, cumulative risk assessments and, thus, will be developing comparable guidance for site-specific and other aggregate exposure assessments, which presumes the evaluation of mixtures of more structurally and toxicologically diverse agents. EPA, in its Exposure Assessment Guidelines (U.S. EPA, 1992b), defined risk characterization as the final step in the risk assessment process that (1) integrates the individual characterizations from the hazard identification, dose–response, and exposure assessments; (2) provides an evaluation of the overall quality of the assessment and the degree of confidence the authors have in the estimates of risk and conclusions drawn;

The characterization of risk to endocrine-disruptive agents may prove to be one of the greatest challenges that the risk assessment/regulatory community has ever faced. Why is this so? The endocrine system is actually many systems, having complex interactions and interdependencies. Normal endocrine function is often dependent on cyclical events, rather than steadystate. Timing is everything, as evidenced by significant differences in adverse outcome as a function of age and stage of development. Further, the consequences of concomitant exposures to endocrine-active substances in the diet or as therapeutic agents are poorly understood. So, how should risk characterization to these agents be approached? This presentation will include the description of current practices for addressing hormonally mediated cancer and noncancer effects and offer speculation on modifications to these approaches that might be necessary as our knowledge of this area increases.

A great deal of attention is currently being focused on the issue of endocrine disruption. Substantial amounts of money have been committed to support research on the basic biology underlying this phenomenon and to develop risk assessment methodologies and tools so that we can fully understand the significance of the consequences of exposure to endocrine disruptors. As a risk assessor and a risk manager, it is my opinion that those resources must be directed to developing data and tools which will allow us to answer the questions that form the basis of a risk characterization. To do otherwise makes it only an academic exercise in satisfying one’s curiosity and does not move our understanding of the issue forward or improve our capacity to make wise regulatory decisions. In March, 1995, the EPA Administrator issued the Policy for Risk Characterization at the U.S. Environ1 Paper presented at the ISRTP meeting ‘‘Assessing the Risks of Adverse Endocrine-Mediated Effects’’ held January 13–14, 1997, in Research Triangle Park, North Carolina. 2 Any opinions expressed are those of the author and may not represent official policy of the U.S. Environmental Protection Agency.

70

0273-2300/97

AID

RTP 1122

/

6e15$$$$61

08-27-97 16:44:34

rtpa

AP: RTP

ADVERSE ENDOCRINE-MEDIATED EFFECTS

71

(3) describes risks to individuals and populations in Toxic effects due to endocrine disruption will not alterms of extent and severity of probable harm; and (4) ways be the most critical effect or most sensitive endcommunicates the results of the risk assessment to the point. To single out for evaluation a subset of endpoints risk manager/decision maker. of concern to the exclusion of the others could lead to Critical to the integrity of any risk characterization an inaccurate assessment of the risks and, subseis a discussion of its strengths and weaknesses, in- quently, the development of strategies that would not cluding the uncertainties attendant to the assess- be consistent with previously articulated regulatory ment — in the data, in the analytical tools, and in the goals (such as negligible risk, reasonable certainty of default assumptions employed in the absence of em- no harm, no unreasonable risk, etc.). pirical evidence. The Guidance which accompanied the Risk CharacShould every risk characterization be identical— terization Policy provides a framework by which a good e.g., in endpoints evaluated, populations addressed, risk characterization can be developed. This framework number/nature of exposure pathways/scenarios exam- is essentially a series of questions which, when anined, etc.? Absolutely not. Legislative mandates differ swered, will provide a sound basis upon which to judge in their scope. For example, the implementors of the the risk assessment component of the decision-making Safe Drinking Water Act would be off base if they pro- process. ceeded to conduct ecological risk assessments. On the Remembering that a good risk characterization will other hand, they should (and do) consider human expo- capture and summarize the critical aspects of the other sure via sources/pathways other than drinking the wa- phases of the risk assessment process: hazard identifiter when developing Health Advisories or Maximum cation, dose–response assessment, and exposure asContaminant Level Goals. sessment, what questions will we want answered in Should every risk characterization prepared in the the risk characterization? [The following example adcourse of decision making under a single statute be dresses only human health, although many of the same identical—in scope and level of complexity? The an- points can be made in the evaluation of ecological efswer again is absolutely not. For instance, many differ- fects. Also, not all questions would be answered for all ent types of regulatory actions are taken in the course risk characterizations. The example represents quesof regulating pesticides under the Federal Insecticide, tions pertinent to a comprehensive risk assessment. Fungicide, and Rodenticide Act and the Federal Food, Limited assessments would necessarily address fewer Drug, and Cosmetic Act. Far less effort should be ex- points, as appropriate.] pended when evaluating a proposed amendment to an existing registration (e.g., to answer the question of THE FRAMEWORK what, if any, difference in risk would accrue to a pesticide handler or a consumer of the food when increasing Hazard Identification the maximum application rate or shortening the preharvest interval for Pesticide X on Crop Y?) than would 1. What is the key toxicity study (or studies) that be required during the reregistration process of a pesti- provide(s) the basis for the health concern(s)? What is cide when a comprehensive reevaluation of hazard, ex- its/their quality? Are the data from the field or laboraposure, and risk is conducted for all aspects of human tory study(ies)? Are the data from one or more sex/ and ecological health. strain/species? What endpoints were observed? What In any case, the ‘‘right size’’ for a risk characteriza- is basis for the critical effect/most sensitive endpoint? tion must be agreed upon jointly by the risk assessor Are there supporting studies/data? Are there valid and the risk manager before the risk assessment pro- studies with countervailing data? cess begins. Without this understanding, the character2. In addition to the health effect(s) observed in ization could fail to meet the needs of the decision mak- the key study(ies), are there additional endpoints of ers, and a lot of work on the part of the risk assessors concern? could be for naught. It should be reemphasized here that effects due to A third question comes to mind. In the contexts de- endocrine disruption should not be evaluated in isolascribed thus far, should environmental agents shown to tion from those adverse effects induced via differing have the potential for endocrine disruption be treated modes/mechanisms. differently from agents which exhibit other toxicologi3. Are there epidemiologic or clinical data? cal traits with different underlying modes/mechanisms 4. What is known about mode/mechanism of action? of action, but which are also considered to be adverse? 5. Are there any nonpositive data in animals or huAt the risk of becoming repetitious, again, the answer mans? How were they considered in the hazard identiis no. Endpoints of concern that are/may be the conse- fication? quence of endocrine disruption should be evaluated in 6. When summarizing the hazard identification, the context of all other endpoints of concern induced what is the degree of confidence in the conclusions? by the agent(s) under scrutiny in the risk assessment. What alternative conclusions are supported by the

AID

RTP 1122

/

6e15$$$$61

08-27-97 16:44:34

rtpa

AP: RTP

72

PENELOPE A. FENNER-CRISP

data? What are the significant data gaps? What major assumptions were incorporated into the assessment? Dose–Response Characterization

Exposure Characterization

1. What data were used to develop the dose–response curve? Would the results differ significantly if a different data set were employed? 2. What model, if any, was used to develop the dose– response curve? What rationale supports this choice? Is chemical-specific information available to support this approach? A. For noncancer endpoints: How was the Reference dose/concentration (RfD/RfC) calculated? What assumptions/uncertainty factors were employed? What is the level of confidence in the estimates? Traditionally, EPA has assumed by default (i.e., absent chemical-specific data to the contrary) that the dose–response curve for all noncancer effects is conventional in nature and nonlinear. At this time, based upon the available data on noncancer effects known to be induced via endocrine disruption, this assumption remains appropriate. In the future, if it were determined that, for example, receptor binding data were critical for use in the derivation of an RfD or RfC, then alternative approaches would be required to accommodate the inverted U-shaped dose–response curve often seen for this measure. Also, if, as some scientists assert, the dose–response curves for one or more critical noncancer effects induced by the agent(s) under scrutiny were shown to be linear in nature, this would have to be taken into account in the decision as to whether or not an RfD or RfC could be derived and, if so, how. B. For cancer endpoints: What dose–response model was used? What was the basis for selection of that model? Are there other models that could be used with equal plausibility and scientific validity? As articulated in the proposed revisions to the cancer risk assessment guidelines (U.S. EPA, 1996a), existence or lack of mode/mechanism of action data will play a significant role in the selection of the model for evaluating the cancer dose response. While a substantial amount of evidence is accruing to suggest that the etiology of a number of tumors may have a hormonally mediated component which is nonlinear in nature, application of a nonlinear dose–response model will remain a case-by-case decision, in the short term, based in good measure on the availability of chemical-specific data to support such an approach. The default assumption continues to employ a linear approach, albeit a modification of the approach exercised in the 1986 cancer risk assessment guidelines (U.S. EPA, 1986a). 3. What routes and levels of exposure were observed/ tested, compared to expected human exposure? Are the data from the same route(s) of exposure as expected human exposures? If not, do pharmacokinetic data ex-

AID

RTP 1122

/

6e15$$$$61

ist for route-to-route extrapolation? How far must one extrapolate from the observed dose range to the range of expected/known environmental exposures?

08-27-97 16:44:34

1. What are the most significant sources of environmental exposure? 2. Which populations were assessed? Include the general population, highly exposed groups, highly susceptible groups, and (in accordance with the ‘‘kiddie risk policy’’), specifically, infants and children. 3. What was the basis for the exposure assessment: Monitoring data? Modeling? Distributional analyses? 4. What are the key descriptors of exposure? Average individual? High end? General population? High exposure group (e.g., children)? Susceptible population(s)? Other? 5. Is there reason to be concerned about cumulative or multiple exposures? Most of the risk assessments that EPA has conducted have been for single chemicals/agents, occasionally for complex mixtures. Even when conducting site-specific assessments, the Agency has tended to develop single chemical assessments and then attempted to evaluate risk in the context of the mixture of chemicals identified at that site by employing a set of principles addressing the potential for relationships and interactions between the components of the mixture. Among the default assumptions included in these principles are (1) if two or more components of the mixture share a common endpoint of toxicity, that endpoint is assumed to have the same underlying mode/mechanism of action and thus should be assessed in combination; and (2) any combination of action is assumed to be additive (U.S. EPA, 1986b). One can see here how important the role of mode/ mechanism of action data could be in the assessment of mixtures and in identifying which, if any, components of that mixture should be assessed in combination. The default assumption of additivity currently would apply to endpoints of toxicity resulting as the consequence of endocrine disruption. This default may not be appropriate, however, for the assessment of certain endocrine disruptors, if the results reported by Arnold et al. (1996) can be corroborated. If, in fact, synergistic interactions are confirmed, this observation will have to be taken into consideration when evaluating certain mixtures. Risk Conclusions and Comparisons This section of a risk characterization will be prepared de novo, reflecting an integration of the information presented in the three previous sections. This is, assuredly, the most critical section of the risk characterization. It is here where the challenge exists to intelligently sift and sort the available information and

rtpa

AP: RTP

73

ADVERSE ENDOCRINE-MEDIATED EFFECTS

present a reasoned judgment of what is known and what is not known about the risks to human health of the agent(s) being evaluated. Risk Conclusions 1. What is the overall picture of risk? 2. What are the major conclusions and strengths of the assessment in each of the three main analyses (hazard, dose–response, and exposure)? 3. What are the major limitations and uncertainties in the three main analyses? 4. What are the science policy options in each of the three analyses? What are the alternatives that were evaluated? What are the reasons for the choice(s) made?

Other Information 1. Is there other information that would be useful to the risk manager/decision maker in this situation that has not been described above? In conclusion, risk assessment is a critical ingredient of the regulatory process. The information contained in the risk assessment is weighed along with other factors, such as those related to economics, public values, political considerations, and technology, during the decision-making process. Risk characterizations which reflect the core values of transparency, clarity, consistency, and reasonableness will only serve to better inform the difficult decisions that must be made to ensure the protection of human health and the environment. REFERENCES

Risk Context 1. What are the qualitative characteristics of the hazard(s) (e.g., voluntary vs involuntary, natural vs technological)? 2. What are the alternatives to this/these hazard(s)? How do the risks compare? 3. How does/do the risk(s) characterized here compare to other risks? In this regulatory program or other similar risks that EPA has made decisions on? Can this risk be compared with those evaluated in past Agency decisions, decisions by other authorities, or common risks with which the public would be acquainted? What are the limitations of these comparisons? 4. Are there any significant community concerns which influence public perception of this/these risk(s)? Existing Risk Information 1. Have risk assessments on this agent been conducted previously by EPA or other regulatory agencies or organizations? If so, are there significantly different conclusions that merit further discussion?

AID

RTP 1122

/

6e15$$$$61

08-27-97 16:44:34

Arnold, S. F., Klotz, D. M., Collins, B. M., Vonier, P. M., Guillette, L. J., Jr., and McLachlan, J. A. (1996). Synergistic activation of estrogen receptor with combinations of environmental chemicals. Science 272, 1489–1492. Browner, C. M. (1995). EPA Risk Characterization Program. Policy for Risk Characterization at the U.S. Environmental Protection Agency. Memorandum. March 21, 1995. U.S. EPA. (1986a). U.S. Environmental Protection Agency. Guidelines for Carcinogen Risk Assessment. Fed. Reg. 51(185), 33992– 34003. U.S. EPA. (1986b). U.S. Environmental Protection Agency. Guidelines for the Health Risk Assessment of Chemical Mixtures. Fed. Reg. 51(185), 34014–34025. U.S. EPA. (1992a). U.S. Environmental Protection Agency. Framework for Ecological Risk Assessment. EPA 630/R-92/001. February, 1992. U.S. EPA. (1992b). U.S. Environmental Protection Agency. Guidelines for Exposure Assessment; Notice. Fed. Reg. 57(104), 22888– 22938. U.S. EPA. (1996a). U.S. Environmental Protection Agency. Proposed Guidelines for Carcinogen Risk Assessment; Notice. Fed. Reg. 61(79), 17960–18011. U.S. EPA. (1996b). U.S. Environmental Protection Agency. Proposed Guidelines for Ecological Risk Assessment; Notice. Fed. Reg. 61(175), 47552–47631.

rtpa

AP: RTP