Hormesis in Regulatory Risk Assessment - Science and Science Policy

SCIENCE POLICY

Science policy can be thought of as the bridge between specific acts of science, whether experimental studies, measurements, observational studies or modeling of processes, and making social decisions in the policy realm. For many health and environmental questions this bridging function takes place through the risk assessment process. Essentially, science policy is policy about how science will be used to inform decisions. Science policy is necessary because no single scientific study is sufficient to inform decisions and analytic choices must be made in the face of uncertainty created by conflicting studies, alternative models and competing theories.

The National Research Council (NRC) panel that first tackled federal risk assessment practice also confronted the issue of science policy as reflected in a later NRC report (NRC, 1994): “Risk assessors might be faced with several scientifically plausible approaches (e.g., choosing the most reliable dose-response model or extrapolation beyond the range of observable effects) with no definitive basis for distinguishing among them. The earlier Committee [NRC 1983 (The “Red Book”)] pointed out that selection of a particular approach under such circumstances involves what it called a science-policy choice. Science policy choices are distinct from the policy choices associated with ultimate decision-making…”

The regulatory risk assessment process, as currently practiced, often has a goal of a single number characterization of risk like a Reference Dose (RfD) or Cancer Slope Factor (CSF) meaning that science policy decisions often have to be made in deciding which data to rely upon. Key questions can include:

In addition assessors confront questions about how to choose exposure measures or dose metrics. If exposures are modeled, which model should be used? How should the dose-response relationship be extrapolated from exposures in toxicology or epidemiology studies to the levels of concern? These are not simple questions and the choice made can be significant. As a simple demonstration, using analysis from Evans et al. (1994) and assuming 10,000.0000 people exposed to a background level of 3.45 ppb of formaldehyde in air, 20 m3 air breathed per day and 70 years of exposure the choice of dose-response model leads to prediction of excess cancers of 0 for a probit model, <1 for a multistage model, and about 21,000 for a one-hit model. In the absence of compelling information about the scientific appropriateness of one model or another, the choice becomes a matter of science policy.

The way in which these science policy considerations are approached varies from agency to agency in the Executive Branch of the US Government in ways that do not depend on statute or legal authority (Rosenthal et al., 1992; Rhomberg, 1997). Instead, science policy is often reflective of the mission or culture of an agency and may be informed by factors such as concern about safety, equity, or burden of proof. As a quick example, consider liver tumors found in long-term bioassays of mice. These are the most common tumors found in rodent cancer tests (Gold et al., 1991). In general, the U.S. Food and Drug Administration (FDA) downplays the relevance of these tumors to humans and numerous pharmaceutical and over the counter drugs have been approved in spite of increased rates of liver tumors in bioassays. EPA on the other hand, virtually always takes findings of mouse liver tumors as relevant to humans and, for example, many of the chemicals often found at Superfund sites (EPA, 2010b) (often why the site is considered for Superfund status) are regulated based on mouse liver tumors (e.g., EPA 1991, 1995, 1996, 2010a). There is no statutory direction to EPA to rely on liver tumors and FDA to discount them. There is no scientific information known to FDA and not to EPA. Instead, each organization has chosen a different science policy approach to dealing with a fundamental scientific uncertainty inherent in risk assessment.

For the US EPA science policy for risk assessment is often implemented through guidance documents. These include guidance for risk assessment of carcinogens, developmental toxicants and ecological stressors ¹ . These documents describe the Agency's science policy framework for EPA risk assessors and have a great influence on the way assessments are carried out in state and local governments and the private sector.

It is clear that one of the factors guiding EPA science policy is a desire to minimize the likelihood of underestimating risk when confronted by scientific alternatives. For example, the Cancer Risk Assessment Guidelines (EPA, 2005) make clear that “conservatism” is a goal guiding science policy choices about dose-response evaluation – “It is the Agency's long-standing science policy position that use of the linear low-dose extrapolation approach provides adequate public health conservatism in the absence of chemical-specific data indicating differential early-life sensitivity or when the mode of action is not mutagenic.” The notion of “public health conservatism” clearly informs both Agency science policy and its implementation. The development of rules to deal with science choices under uncertainty is important but of potentially greater significance is how those rules are used (e.g., the determination that a chemical's mode of action is mutagenic in the example above).

SCIENCE AND HORMESIS

There is an extensive literature on hormetic responses in a wide range of biological systems and settings (for reviews see Calabrese and Baldwin, 2001; Calabrese and Blain, 2005). Indeed EPA has funded research looking for, and finding, evidence of hormetic responses ² . The Agency's 2003 edition of the Annual Toxicology and Risk Assessment Conference devoted a major session to “Hormesis in Human Health and Ecological Risk Assessment.” Agency science advisors have encouraged EPA to look more closely at hormetic effects, although the advice has been controversial even within the same science panel ³ . Clearly EPA is aware of hormesis and its potential effects on the conduct of risk assessment.

At the same time, there is dissent in the scientific community. Some focuses on the science of hormesis and some on the policy implications of including hormetic effects in risk assessment procedures. As an example of the former, Mushak (2007) has written about his concerns that hormesis is ill defined, its prevalence unknowable and its biological basis unclear. He concludes in his paper that “Definition, characterization, occurrence, and mechanistic rationale for hormesis will remain speculative, absent rigorous studies done specifically for hormesis testing. Any role for hormesis in current risk assessment and regulatory policies for toxics remains to be determined.” Others have raised both scientific and policy concerns. Thayer et al. (2005) review areas in which the theory and evidence for hormesis is, in their view, lacking. These include questions of variability in response and mechanistic underpinnings. They conclude that considering hormesis in risk assessment would detract from its “public health protective” policy position. The sum up their view thusly “The claims and projections of health benefits from exposures to environmental toxicants and carcinogens are based on untested assumptions and disregard numerous well-established scientific principles that underpin a public health-protective approach to regulating exposure to toxic substances” (emphasis added).

SCIENCE POLICY - WHERE ARE WE TODAY?

We have seen that regulatory risk assessors must be quite aware of hormesis, both the evidence for the phenomenon and the concerns, both scientific and policy oriented, that have been raised. What science policy positions have been taken on hormesis? The answer is none – except by omission. At this point there is no mention of hormesis in any EPA risk assessment guidance document including those for Cancer Risk Assessment (EPA 2005), Developmental Toxicity Risk Assessment (EPA 1991) and Neurotoxicity Risk Assessment (EPA 1998). Hormesis is not explicitly excluded from consideration but no direction is given on how to incorporate it into an assessment. Without specific science policy guidance on how to make choices about when and how to consider hormesis it is highly unlikely to be incorporated into EPA assessments.

I believe hormesis is “shut out” of current risk assessment procedures for reasons of both science and science policy. The science issues may be amenable to research and investigation but science policy views may have greater influence. Several of the science questions raised by Mushak (2007) and Thayer et al. (2005) are appropriate and can be investigated experimentally and observationally. Understanding interspecies and intraspecies variability in hormetic responses, especially quantitatively, is clearly an important need to inform any use in risk assessment. Epidemiologic evidence of hormesis would likely be quite persuasive.

The science policy issues, however, are driven more by a regulatory agency's perspective on decision making under uncertainty, its view of the best way to carry out its mission, and the history and culture that influence specific choices. This is perhaps best exemplified by the EPA's notion of “health protective” or “public health conservative” science policy choices in the carcinogen risk assessment process (EPA 2005). As long as the consideration of hormesis is seen as not health protective or, in the words of Thayer et al. (2005), having the implication that it would “allow higher exposures to toxic and carcinogenic agents” we are unlikely to see changes in the science policy underlying regulatory risk assessment.

In addition, there are no widely accepted examples in which ignoring a hormetic response has resulted in public health harm although it is clearly theoretically possible that ignoring a true hormetic effect would actually contravene the notion of public health conservatism underlying much of risk assessment science policy ⁴ .

WHAT MIGHT CHANGE THINGS?

In my view, hormesis will not be incorporated into regulatory risk assessment science policy until there is demand from decision makers. The science policy culture of regulatory risk assessment suggests that the perception that considering hormesis is not public health protective means it is unlikely to be advocated by the risk assessors.

If the use of risk assessment begins to evolve from the standard setting uses of the past to the more decision focused approaches advocated in the 2009 National Research Council Report Science and Decisions (NRC 2009), risk managers may begin to request assessors to consider hormetic effects. As envisioned by the NRC risk assessments will support weighing alternative risk management options. The risk assessment will be tailored in complexity to inform the specific choices that have to be made. Inherent in this way of thinking is the realization that there are public health risks on both sides of a choice (Graham and Wiener, 1995). In this case, decision makers will want (and need) best estimates of risk not “health protective” estimates with unknown (and often different levels of) conservatism (Gray, 1996; Ohanian et al., 1997; Gray and Hammitt, 2000). Best estimates of risk will involve the consideration of all relevant scientific information (including hormesis), careful characterization of dose-response including the potential for non-linear, threshold and nonmonotonic forms, and description of variability in exposure.

Unless this sort of change in the basic approach to risk assessment and management occurs the only thing I foresee forcing a reconsideration of hormesis in regulatory risk assessment is a clear public health case that not considering low dose non-monotonicity is a threat to public health. I don't see any examples of this on the horizon.

SUMMARY

Regulatory risk assessments are clearly a mix of science and science policy. Making consistent and reproducible decisions about which studies, data, models and approaches to use in assessments is the traditional argument for science policy guidelines. The content of these policies and guidelines is informed by agency history, culture and view of their mission. Different agencies have different science policy approaches to the same risk assessment issues.

There are still science questions regarding hormesis and the way it should inform risk assessment. These include a lack of understanding of the biological basis of hormesis, lack of clarity about its definition and questions about population variability in both hormetic and adverse effects of exposures. It is my belief that future research can, and will, begin to resolve many of these issues.

However, acceptance of hormesis in science policy choices is only partly informed by advances in science. Because science policy seeks to provide guidance in the face of scientific uncertainty, and there will always be some level of uncertainty in many aspects of risk assessment, an agency's view of its mission, its history and its culture will play a large role in whether hormesis will be considered in risk assessment. Until there is demand from risk managers for best estimates of risk, I believe hormesis will continue to be excluded from risk assessment because of the belief that its inclusion would not be “health protective.”