Can Regulatory Agency Actions Accommodate Hormesis, Adaptation and Mechanisms of Toxicity?

Christopher M. Teaf, Ph.D. *

Center for Biomedical & Toxicological Research

2035 East Dirac Drive, Suite 226 HMB

Florida State University

Tallahassee, FL 32310

Phone: 850-644-3453

Fax: 850-574-6704


*Address all correspondence to:


The phenomenon of seemingly paradoxical beneficial effects at low doses for agents that are otherwise toxic at higher doses (hormesis) is well-established in the scientific literature, and covers processes that may be related to adaptation, repair and stimulation of some biological processes (e.g., immunity). Although it is recognized in the field of toxicology that one or more mechanisms of action are important considerations, and although most public health and environmental regulatory agencies have theoretical procedures in place to permit consideration of mechanistic information, instances of it's inclusion in standard setting are rare. This has resulted in cases where broadly used guidance values (e.g., RfDs) appear to be inconsistent with, for example, typical recommended dietary intakes. The implementation of proposals to accommodate hormesis and related phenomena, which clearly exist, within risk-based evaluations may be best served by specific efforts to provide detailed characterization of the process for an example substance. Persuasive demonstration and definition of dose-related physiological effects, coupled with reproducible dose/response information to validate earlier reports for a consensus sentinel compound, are needed to support widespread incorporation into conventional risk determinations.


The term hormesis is well-established in the scientific literature, though the early reports were primarily of an observational nature. Many of the historical reports and studies were concerning radiological exposures, though a large number of chemical examples more recently have become commonly recognized (Calabrese and Baldwin, 1998). The term "hormesis" is descriptive, and surely covers a wide range of phenomena which may be reflective of one or more of the following processes: adaptation (cellular or biochemical), damage repair, and stimulation of biochemical processes such as sequestration, immunity, and metabolic degradation. The breadth and chemical diversity of agents for which effects have been reported, and the variable degrees of response ensure that different mechanisms are at work. For entire classes of chemicals in the medical arena, the comparison of beneficial effects (therapeutic) weighed against adverse effects (toxicity) is a standard process which clearly addresses mechanisms (Gaylor et al., 1998).

The dilemma posed by accounts of hormesis, or at least the reports of apparent paradoxical responses caused by a range of doses for some agents, is that we can readily acknowledge its existence but have a more difficult time making practical use of the information. This is especially true in an environmental regulatory context, where the paradigm of increasing response with increasing dose underlies virtually all state and federal programs of contaminated site remediation, for all media. This discussion considers two principal questions, primarily in the context of environmental decision-making:

· Theoretical Considerations: Does an understanding of the mechanism of toxicity for a chemical, or potential adaptive mechanisms in an organism, affect the way in which environmental and public health regulatory agencies assess risks from exposure to toxic substances?

· Nuts and Bolts: If low doses of chemicals induce, in some cases, apparently beneficial effects, how can environmental and public health regulatory agencies incorporate that information in assessing the risks from higher doses?


To the extent that mechanistic information regarding toxic sequelae is known for a particular substance, it often may be acknowledged in supplementary information but rarely is explicitly taken into account in the assessment of risks by regulatory agencies. An exception, of course, is the determination of whether or not a chemical may be classified as a carcinogen. Even here, however, within those agents that are so classified, life is not simple. Take as an example the federal Proposed Guidelines for Carcinogen Risk Assessment (U.S. EPA, 1996) and the recent debate concerning application of those guidelines to a revision to the recommended Maximum Contaminant Level (MCL) for chloroform. The Guidelines discuss the reasons for, importance of, and methods to address the incorporation of mechanistic information into carcinogen risk assessment, as an adjunct to or replacement for conventional linear modeling approaches. A huge volume of relevant information is available on the toxicology, mechanisms of action, and occurrence of chloroform, yet the proposal to raise the MCL from its present level to a new value of 300 mg/L engendered a significant controversy. Ultimately, the agency made a decision not to incorporate the mechanistic information following extensive public comment. This is especially troublesome in that many scientists had viewed chloroform as an excellent example of just the type of case envisioned by the 1996 Guidelines, particular for putative nongenotoxic carcinogens.

One likely reason for this situation, at least in the environmental context, is that the establishment of applicable remedial goals for environmental media at the state and federal levels fundamentally relies upon toxicological limits defined by the U.S. EPA Reference Dose (RfD) for noncarcinogenic effects and the Cancer Slope Factor (CSF) or Potency Factor for cancer health effects. These ostensibly health-based values are typically derived from a combination of animal studies, an identification of appropriate departure points (e.g., LOAEL, NOAEL), and the stepwise application of Uncertainty Factors to estimate a threshold (or risk-defined value in the case of carcinogens) which represents an "acceptable" condition. These RfD's and CSFs are then used by virtually all U.S. EPA regional offices and state regulatory programs to develop default criteria or preliminary remediation goals that assume default residential or commercial exposure scenarios. As a practical matter, reliance on these benchmark values creates some interesting inconsistencies or potential conflicts with the conclusions of other entities, such as the U.S. Food and Drug Administration (FDA), and the National Academy of Sciences (NAS, 1989) which make recommendations regarding dietary intake of certain essential nutrients. Selenium, for example, which is an essential antioxidant and enzyme component, has an oral RfD value of 5E-03 mg/kg·day (5 µg/kg·day) and the upper end of the FDA recommended range is nearly 3 µg/kg·day for an adult male or 4 µg/kg·day for an adult female (based on 200 µg/day recommended intake and body weights of 70 and 50 kg, respectively). A similar oddity occurs in the case of fluoride, where thousands of U.S. municipalities intentionally fortify their water supplies with the substance as a protective agent for teeth, yet cleanup requirements may be set in the same states for groundwater with less flouride than is present in the municipal water.

At present the FDA can, and in some cases does, consider beneficial as well as adverse effects (Gaylor et al., 1998). Similarly, U.S. EPA has no formal policy regarding beneficial effects of low level exposures, but has the capability to take such information into account, given adequate available detail (Davis and Farland, 1998). A parallel approach is taken by the U.S. Consumer Product Safety Commission (CPSC), which has a mechanism in place to address specific cases, but such instances are likely to be rare in practice. More to the point, Babich (1998) points out that the CPSC guidelines were not meant to be applied "mechanically". The Agency for Toxic Substances and Disease Registry (ATSDR) acknowledges the value of such information, and has taken it into consideration in its decisions concerning agents such as the essential nutrients chromium, manganese and zinc (De Rosa et al., 1998). Personal experience of this author, and a review of occasional compilations of selected state environmental regulatory agency guidelines (e.g., Judge et al., 1997) suggests that mechanisms of action are not commonly addressed in a significant way in standard or guideline setting exercises at present.


When considering the issue of how regulatory agencies should (or could) address the concept of adaptation and beneficial low dose effects, one is inevitably faced with examples such as arsenic. This metal, in its many organic and inorganic forms, has been extensively studied. Yet in recent months the complex nature of the "bad vs good?" question has been elevated further for arsenic. On one hand, we have knowledge of the continuing crisis in Bangladesh, where millions may be adversely affected by high levels or arsenic in drinking water (Lepkowski, 1998). On the other hand, we have a contemporary report in the New England Journal of Medicine in which arsenic trioxide was effective in causing complete remission of one form of leukemia in patients receiving approximately 0.15 mg/kg·day (Soignet et al., 1998). And arsenic is one of the substances with the Gordian knot-like situation of being regulated as a Known Human Carcinogen by a variety of agencies, while also being regularly present in the diet and even exhibiting plausible evidence of essentiality (ATSDR, 1998; U.S. EPA, 1988; NAS, 1977). Given these characteristics and recent heath-related developments, coupled with its dominant role in determining cleanups at many sites, arsenic may present a case study for documentation of the often qualitative distinction between low dose and high dose effects.

There are regular discussions concerning the derivation and potential improvements to establishment of threshold level toxicological benchmarks. To the extent that it is useful for environmental agencies to develop tabular presentation of consistently derived but admittedly default remedial goals or guidelines, the use of RfDs and CSFs is a reasonable exercise. However, in cases where a well-grounded case can be made that different controlling mechanisms operate at low dose than at high dose, that information should be incorporated at least in annotated form in the guideline documents. Nearly every state which has established environmental risk-based programs has included the option to present site-specific, or even chemical-specific information, in support of a site cleanup decision. For example, Texas, Florida, Louisiana, California and many others provide default guidance, but also preserve the right of the responsible party to submit mitigating information. In common experience, this includes information such as exposure frequency and land use controls, and it is not clear that most states have risk assessment staff with appropriate experience to make chemical-specific decisions. Nevertheless, incorporation of such considerations into the RfDs and CSFs, perhaps by U.S. EPA, would lessen the burden.

Another area of important research and practical application is that of the quantitative modeling that is used to develop the CSFs and to a lesser degree the RfDs. The overwhelming majority of CSFs have been developed on the linear or linearized model format for toxicological actions at very low doses. Therefore, it probably will be necessary to address agents such as arsenic, with large underlying exposure databases, in an attempt to reconcile the dilemma described earlier in this section. Revisiting the foundation by which most arsenic environmental decisions are made (e.g., the very restrictive CSF) may assist in addressing criteria such as those in U.S. EPA Region IV, Region VI, Region IX and many individual states where default recommended residential soil cleanup criteria in the range of 0.4 to 0.8 mg/kg are associated with arsenic intakes (<0.1 mg/day) that are several orders of magnitude less than typical arsenic dietary intake (10 to 50 mg/day; ATSDR, 1998), and perhaps below potentially beneficial effects as well.


Aldous Huxley is credited with having observed that "facts do not cease to exist because they are ignored". The question is not so much whether information regarding hormesis, mechanistic data, and demonstration of beneficial effects of low dose exposures should be taken into consideration, but rather how to do that effectively within the existing regulatory structures and procedures. In principle, as noted previously and as articulated elsewhere by numerous agency staff, the incorporation of such information seems feasible under existing procedures at environmental regulatory agencies, by invoking the "case-by-case" provisions for risk assessment that are present in many state and federal rules. If we acknowledge the difficulty inherent in the implementation of requests to an agency that depart from the norm, the onus is on us, the scientific community, to provide adequate detail and reproducibility in the data to persuade regulators that decisions made on the basis of those data will not put the public at unacceptable risk.

As an adjunct to the essential searches of the historical literature that are designed to find examples of inverse dose-response curves, apparent paradoxical results and hormetic effects, the most constructive approach may be to identify the most stellar single representative from among the available examples of these phenomena, compile the evidence in a comprehensive way for that chemical, and carry it forward to establish a practical precedent that an agency or multiple agencies can implement in an existing program. Beneficial effects of chemicals at low doses exist, of that there can be no doubt. However, without an initial successful persuasive effort in hand, however simple that single case may be, any mere philosophical acceptance of the general phenomenon by the regulatory community will be without real benefit.


ATSDR. 1998. Toxicological Profile for Arsenic (update). Agency for Toxic Substances & Disease Registry, Atlanta, GA. .

Babich, M.A. 1998. Risk assessment of low-level chemical exposures from consumer products under the U.S. Consumer Product Safety Commission chronic hazard guidelines. Envir. Health Perspect. 106 (Supplement1):387-390.

Calabrese, E.J. and L.A. Baldwin. 1998. Hormesis as a biological hypothesis. Envir. Health Perspect. 106 (Supplement1):357-362.

Davis, J.M. and W.H. Farland. 1998. Biological effects of low-level exposures: A perspective from U.S. EPA scientists. Envir. Health Perspect. 106 (Supplement1):379-381.

De Rosa, C.T., et al. Public health implications of environmental exposures. Envir. Health Perspect. 106 (Supplement1):369-378.

Gaylor, D.W., P.M. Bolger and B.A. Schwetz. U.S. Food and Drug Administration perspective of the inclusion of effects of low-level exposures in safety and risk assessment. 1998. Envir. Health Perspect. 106 (Supplement1):391-394.

Judge, C., P. Kostecki and E. Calabrese. 1997. State summaries of soil cleanup standards. Soil and Groundwater Cleanup (November 1997): 10-34.

Lepkowski, W. Arsenic crisis in Bengladesh. Chem. Engin. News 76(46): 27-29.

NAS. 1977. Drinking Water and Health. National Academy of Sciences, Washington, DC.

NAS. 1989. Recommended Dietary Allowances (10th Edition). National Academy of Sciences, Washington, DC.

Soignet, S.L., et al. 1998. Complete remission after the treatment of acute promyelocytic leukemia with arsenic trioxide. New Engl. Journ. Med. 339 (19): 1341-1348.

U.S. EPA. 1988. Special report on ingested inorganic arsenic: skin cancer, nutritional essentiality. Risk Assessment Forum, EPA Report EPA/625/3-87/013, Washington, DC..

U.S. EPA. 1996. Proposed Guidelines for Carcinogen Risk Assessment. Office of Research and Development,
Washington, DC.. April, 1996.