Incorporating Hormesis in Risk Regulation

Frank B. Cross

The author is a Herbert D. Kelleher Centennial Professor of Business Law, University of Texas at Austin.

I suspect that many readers are unfamiliar with the concept of, evidence for, or implications of, the biological principle of hormesis.¹ This is a shame, insofar as environmental policy purports to be based on scientific evidence.² There is a substantial body of evidence behind the theory that even very hazardous substances have a hormetic pattern of effects—that is, they may be beneficial to health at very low levels of exposure.

The evidence for hormesis has received virtually no attention in environmental regulation, perhaps because the concept is seen as a "front" for industry efforts at deregulation.³ Yet one sincerely concerned with public health, and not merely concerned with opposing industry, must take hormesis seriously. This Dialogue seeks to do so. I explore the implications of hormesis in the context of environmental regulation of carcinogens, where it represents the greatest departure from prevailing regulatory theory.

After summarizing the concept of hormesis briefly, I consider how it should be incorporated into existing statutory frameworks of environmental policy and paradigms of risk assessment. While hormesis is contrary to most current risk assessment policies, it actually fits quite well within current statutory frameworks. In addition, I will demonstrate how applications of hormesis can actually lead to greater stringency of regulation, in some circumstances.

I conclude by discussing how environmental statutes should be structured in a world in which substances have hormetic effects. While hormesis generally fits within existing statutes, the regulatory systems created by some of those statutes are suboptimal in dealing with hormetic effects. The sort of industry-based technologically contingent emissions standards are not well-suited for conditions of hormesis. The best policy, I argue, would focus on ambient standards and "risk cups" of the sort established in the recently enacted Food Quality Protection Act (FQPA).⁴

The Concept of Hormesis

Hormesis isa version of the traditional aphorism that the "dose makes the poison." The concept more particularly suggests that low levels of exposure to a substance may be beneficial, even if that substance is extremely hazardous at higher exposure levels. Even a substance that can cause cancer with a single, minute exposure may nevertheless exhibit net hormetic effects, so that small concentrations of exposure will be beneficial.

Hormesis is often graphically depicted as an inverted "U-shaped" or "[beta]-shaped" curve of health effects, with a hormetic zone of beneficial exposure rising to a point where defense mechanisms are overwhelmed (the "tipping point"), followed by a zone in which additional exposures have negative consequences. In the legal or regulatory context, it is easier to understand the application of hormesis as a standard "J-shaped" curve, as depicted in Figure 1.

Figure 1

[SEE ILLUSTRATION IN ORIGINAL]

[30 ELR 10779]

In this figure, the vertical axis depicts adverse health effects. This reflects the legal and regulatory aim of minimizing adverse health effects (in contrast to maximizing health).⁵ The curve declines during the hormetic zone, as exposures actually reduce the health effects suffered by exposed populations. Beyond a point on the horizontal axis, though, adverse health effects begin increasing with additional exposure.

The theoretical reason for hormetic effects lies in the body's adaptive or stimulatory response to low levels of exposure. When the body senses such exposure, defensive mechanisms are activated. This biological response over-compensates for the threat posed by the very low levels of exposure and makes the defense mechanisms more robust than before the exposure.⁶ The theory is well known and established with respect to vaccines, which often involve introducing a small exposure to a pathogenic agent in order to activate defenses and prepare the body for the threat of a larger exposure. Indeed, this effect is true of most drugs. Aspirin, for example, is beneficial at low doses but can cause toxicity at high doses. At some exposure level, the toxicity overwhelms the body's defenses, whether or not they have been stimulated by low-level exposures. At these higher exposure levels, many substances are truly hazardous.

The nature of this effect has been referred to with the metaphor of invaders and defenders.⁷ A carcinogen is like an invader in cellular health. After noticing the invader, the cell calls its volunteer defenders into action to repel the invasion. If the invader is weak (low dose exposure), the defenders fight off the invader. And the weak invasion effectively mobilized the defenders, who might otherwise be dormant. Hence, in the metaphor, a low dose invasion enhances the vigilance of the cell's defense, thereby leaving it in a stronger position to deal with other threats.

Hormesis is most commonly associated with radiation exposure. Because of the relative availability of data or the controversial nature of radiation, considerable research has been conducted on the effects of low-level exposures to radiation. Dr. Thomas Luckey published a well-known article showing hormetic effects from radiation exposure.⁸ Much subsequent research also shows a hormetic effect. In Japan, survivors of the atomic bomb who received low dose radiation have lived longer lives than those not so exposed.⁹ Another Japanese study found that residents of villages who had high radon levels in drinking water had lower cancer rates than comparable populations elsewhere.¹⁰ Considerable additional research, much of it on plants and animals, has confirmed the existence of the effect.¹¹

Hormesis is not limited to the context of radiation exposure. For chemical toxicity, researchers have shown that

if a low—and often nontoxic—exposure to radiation or a toxic chemical is administered either to cells in culture or to whole organisms, and is then followed by massive exposure to the agent that would normally seriously injure or even kill the cells or organisms, the preexposed cells or organisms display remarkable protection from toxicity and lethality.¹²

Indeed, researchers have found that hormesis is broadly generalizable as a concept.¹³ It applies in all types of species, from mammals to plants to microbial entities. Agents of all chemical classes show hormetic effects, and the concept seems to apply to all sorts of biological effects.¹⁴ Most relevant for our purposes is evidence on cancer. Most cancer studies do not employ sample sizes or sufficient doses to test for hormetic effects, but some studies are quite revealing.

One study examined the cellular initiation of cancer by genetic disruption. Careful experimentation found that cells given low concentrations of a carcinogen showed less disruption than did cells that were untreated. Evidence of the cell's adaptive response was seen, and the low dose cells had nearly a 50% better survival rate than did the untreated control group.¹⁵ Studies in rats have shown hormetic effects at the epigenetic or promoter stage of cancer development.¹⁶ Indeed, evidence of a hormetic protective effect has been seen even in studies of dioxin, commonly considered to be highly hazardous at very low exposure levels.¹⁷ The size of the beneficial hormetic effect may be considerable.¹⁸

The evidence for hormesis is not conclusive. The nature of the hormetic effect may vary among species and even within species, by age, gender, or other factors. Much of the evidence comes from animal bioassays, which are of uncertain [30 ELR 10780] value in extrapolating to humans.¹⁹ Such animal studies, however, are the fundamental basis of current regulatory toxicology. Were such research to be disregarded, many regulations would likewise have to be thrown out. Regulation is not conditioned on scientific certainty but relies upon the best scientific evidence, interpreted conservatively so as to maximize public health protection.²⁰ The scientific data-base is sufficient to incorporate hormesis into regulation, and the quantity of data is such that hormesis could become a default assumption of risk assessment, presumed until disproven in a particular circumstance.²¹

Incorporating Hormetic Effects in Regulation

Regulation has currently ignored hormetic effects. A search of the entire Federal Register has found but one reference to the concept, and that was in a notice of grant money.²² Some authors have considered the possibility of incorporating hormetic effects in environmental regulation, but these efforts have largely been highly abbreviated ones.²³ I strive to provide a somewhat more extensive analysis of how such an incorporation could transpire legally and how laws and regulations might be adapted to better suit a world of hormetic effects.

I focus this Dialogue on the environmental regulation of carcinogens. For most noncarcinogenic hazards, the U.S. Environmental Protection Agency (EPA) assumes the existence of a safe threshold of exposure and seeks to regulate only at levels surpassing this threshold. The threshold is analogous to the concept of hormesis because it presumes that there is some safe level below which risk does not increase (though it does not recognize the possibility that exposures may reduce risk). Hence, hormesis is not especially controversial for these substances. While not formally recognized as such, hormetic effects may functionally already be incorporated in regulations for non-carcinogenic substances.

EPA has recognized something like hormesis even in carcinogen regulation, in limited circumstances. Some essential nutrients, such as selenium, have been shown to be carcinogenic. In such cases, the Agency obviously has not sought to reduce exposure to levels below that essential for life. For selenium, the Agency adopted a standard that was several times the recommended dietary allowance for the mineral. This decision was not based, however, on any consideration of the hormetic, anticarcinogenic effects of low-level exposures to selenium.²⁴

The Current Regime

The scientific merits of hormesis might have little practical legal significance, if the concept were precluded by statutory command. This clearly is not the case, however. In fact, many statutes are more consistent with a theory of hormetic effects than on the zero threshold assumptions commonly employed by regulators. In this section, I consider use of hormesis under the statutes and risk assessment policies promulgated by the agencies. While hormesis does not conform to those risk assessment policies, their features arenot statutorily commanded, so they could readily be modified to incorporate hormetic effects.

Statutory Standards

The statutory standards for environmental regulation have been criticized as reflecting wishful thinking, at least with respect to carcinogenic hazards. Many of the statutory standards implicitly assume that there is some safe threshold level of exposure, below which no risk exists. Under these statutes, regulators were to set standards so that human exposures remained below the threshold at which adverse health effects began to occur. In recognition of scientific uncertainty and the precautionary nature of environmental legislation, some statutes called for a "margin of safety" below the identified threshold of exposure.

The original Clean Air Act (CAA) provision for hazardous air pollutants, such as carcinogens, called for having a standard with an "ample margin of safety to protect public health."²⁵ The standard for the most common, criteria pollutants (which may or may not be carcinogens) called for an "adequate margin of safety."²⁶ Given the regulatory disappointment with the progress under this standard, the law was amended to provided for technology-based standards by source category, but the new law still provided that such standards should be strengthened if they did not "provide an ample margin of safety."²⁷ The Clean Water Act required that standards for toxic pollutants incorporate an "ample margin of safety."²⁸ The Safe Drinking Water Act (SDWA) directs the Agency to set goals (referred to as recommended maximum containment levels (RMCLs) or maximum containment level goals (MCLGs)) that contain an "adequate margin of safety" for public health protection.²⁹ EPA is then to set enforceable standards that come as close as is "feasible" to achieving that goal.³⁰ Other statutes do not have the margin of safety language but provide for a standard such as prevention of an "unreasonable risk of injury to health or the [30 ELR 10781] environment."³¹ The recently enacted FQPA, which may act as a template for future laws, directs the Administrator to set pesticide residue standards at a level that is "safe," which is defined as a "reasonable certainty that no harm will result" from the exposure.³² The FQPA also contains a presumptive ten-fold margin of safety beyond this level to account for the possible enhanced sensitivity of children to pesticide exposures.³³

Application of the "margin of safety" statutory standards became quite complicated in the context of carcinogens. It was impossible to prove a safe level of exposure for carcinogens, and there was some theoretical reason to believe that even the smallest level of exposure could trigger cancer, called the "one-hit hypothesis." Yet this approach admits of no possible margin for safety. Former Administrator William Ruckelshaus declared that EPA's most "impossible assignment" was the pursuit of "zero cancer risk with a margin of safety below that."³⁴ The Agency struggled with the conflict, and relatively little regulation ensued under many of these statutes.

Under the zero threshold or no safe level assumption, the notion of margins of safety become incoherent. How can there be a margin of safety if there is no safe level of exposure? Agencies have struggled with this difficulty and have had varying degrees of success. The U.S. Supreme Court intervened and facilitated regulation under these statutes in the well-known Industrial Union Department, AFL-CIO v. American Petroleum Institute (Benzene)³⁵ decision. There, the Court held that the Occupational Safety and Health Act (OSH Act) should be interpreted to provide authority for regulation of only "significant" risks to public health, not all risks however small.³⁶ The OSH Act, though, did not contain margin of safety language. The Benzene holding has not been limited to the OSH Act, but its application under other standards is less clear.³⁷ Moreover, the Court affirmed the use of conservative risk assessment presumptions which, along with linear risk models, meant that agencies could regulate very low-level exposures under assumptions of linearity.³⁸

Risk Assessment Policies

Absent hormesis, agency risk assessment for carcinogens has had a hard time conforming to the decision standards of environmental legislation. Regulators have generally presumed that there is no safe level of exposure to a carcinogen (the no threshold hypothesis), which precludes any literal application of margin of safety standards. This presumption is in keeping with a general practice of using "conservative" assumptions in risk assessment (those that accept some probability of overestimating risk in order to avoid underestimating the risk).

Risks are commonly assessed by identifying carcinogenicity at very high dose exposures of animals and then extrapolating the dose to lower level exposures according to a biomathematical model. The risk assessment process explicitly incorporates conservative assumptions in order to avoid any possible underestimation of true risk.³⁹ Conservative assumptions embedded within the test process include the use of highly susceptible animal subjects, the application of high maximum tolerated doses, assumptions that mode of administration does not matter, and failure to consider pharmacokinetic differences among species.⁴⁰ The cumulative result of the various conservative assumptions may cause the overstatement of actual risk by "factors of a thousand or even a million or more."⁴¹ Such an overestimation of risk may be justified by the "precautionary principle," which posits that we should err on the side of public safety. This principle implicitly presumes, however, that low exposures can only be hazardous and could not possibly be beneficial. Once beneficial hormetic effects are considered, the basis for the highly conservative risk assessment assumptions disappears. The possibility of beneficial hormetic effects means that risk assessment should be reoriented toward obtaining the most accurate possible measure of risk, sometimes called the "most likely estimate."

EPA's most recent guidelines for carcinogen risk assessment were proposed in 1996 and relaxed the conservatism of its practice somewhat.⁴² The guidelines expressly contemplate consideration of different dose-response models and departing from the strongly held presumption of linearity in extrapolation to low levels.⁴³ Even the revised guidelines use default presumptions that are "public health conservative,"⁴⁴ though, and the default calls for a linear dose-response curve for carcinogens.⁴⁵ This choice of a default assumption is extraordinarily powerful, as the Agency seldom departs from that base.⁴⁶

One of the conservative assumptions commonly employed by risk assessors is that there is no safe threshold for carcinogen exposure and that dose-response relationships are linear. While EPA risk assessment has increasingly become more sophisticated and recognizes several models of cancer development, virtually all the assessment methodologies share the characteristics of being linear at low doses and assuming no safe level of exposure, however small. The contrast between EPA linear risk assessment models and the hormetic risk curve is indicated by Figure 2.

[30 ELR 10782]

Figure 2

[SEE ILLUSTRATION IN ORIGINAL]

If we know the risk level at a hypothetical exposure E, extrapolation to risks from lower levels depends centrally upon the assumed dose-response curve. From Figure 2, it is clear that there is a substantial disparity between the risks projected by hormetic models and those projected by linear exposure models at low levels.

Recent years have seen a glimmer of recognition for hormetic effects. In setting SDWA standards for chloroform, EPA acknowledged that the substance exhibited a "nonlinear mode of carcinogenic action"⁴⁷ but nevertheless set an exposure goal of zero, based on the Agency's long-held policy of no save level of exposure to carcinogens.⁴⁸ The latter determination was successfully challenged by an industry group. The D.C. Circuit vacated it as contrary to the best scientific evidence.⁴⁹ The case is significant because both the court and the Agency held that the best scientific evidence demonstrated a threshold for carcinogenic effects.

While neither EPA nor the D.C. Circuit discussed hormesis in regulation, the chloroform decision implicitly raises the issue. Chloroform is a byproduct of the use of chlorine as a disinfectant for drinking water, which has clear health benefits. The decision clearly recognizes thresholds and conceivably is a first step toward the incorporation of hormesis into carcinogen regulation. The following discussion considers how the hormetic model could be incorporated into existing statutory frameworks.

Effects of Hormesis on Regulations

It is commonly assumed that consideration of hormesis would lead to a reduction in regulation, at least for carcinogens. The hormetic model clearly predicts less risk at low levels than does the linear model. Relaxing the assumption of no threshold for hazardous effects would seem to permit more emissions under any safety standard. This may sometimes be true, but in theory the concept of hormesis may possibly lead to greater stringency in regulation, under certain circumstances.

Suppose that an agency is considering a particular regulation, in light of its public health benefits and the costs associated with the regulation. Some statutory standards, such as the Toxic Substances Control Act's (TSCA's) unreasonable risk standard, explicitly require such cost considerations.⁵⁰ Although a number of other environmental statutes literally proscribe cost considerations,⁵¹ it is common for agencies to consider them nonetheless when promulgating regulations.⁵² Figure 3 shows how a shift to a hormetic curve could alter the agency's calculus.

Figure 3

[SEE ILLUSTRATION IN ORIGINAL]

Suppose that the agency is considering a reduction in exposure levels from E to E<*>. Under the linear model, this would yield a health benefit amounting to H-H[1]. That health benefit will be achievable at a certain cost C. The regulation will appear desirable if the magnitude of the risk reduction is [30 ELR 10783] considered worth the cost. Under the hormetic model, though, the regulation would yield the much greater health benefit of H-H[2]. The cost C will not be changed by use of the hormetic risk assessment model. The regulation is more likely to appear desirable under a cost-benefit standard, when benefits are measured under assumptions of the hormetic model. Even absent cost considerations, the hypothetical example could call for stricter regulation under the hormetic model—sometimes H-H[1] will not meet the significant risk requirement for rulemaking when H-H[2] would do so.

Another context in which hormesis could dictate more stringent regulation arises in the context of source-specific emission or effluent standards such as the hazardous air pollutant authority of the CAA, as amended in 1990. These standards functionally assume that all increments of exposure create equivalent risk, without regard to ambient levels of exposure. Yet the evidence for hormesis denies this assumption. If ambient levels are quite high, above the tipping point, the relative risk from each additional increment of exposure seems relatively higher, as illustrated by Figure 3. Hence, the acceptability of the risks is less and the need for regulation more apparent under these circumstances.

Applications of Hormesis in Current Structures

Our environmental protection statutes were not adopted with the concept of hormesis in mind. Nor was hormesis particularly relevant to early environmental protection measures. The beneficial, hormetic effects of exposure occur at very low levels. When ambient levels were high, hormesis may well have counseled for even stricter controls, as discussed above. As environmental standards have become increasingly stringent, however, the beneficial part of the curve may come into play. This section discusses the extent to which current statutes permit incorporation of hormetic effects in government policy.

Adding consideration of hormesis would require a Kuhnian paradigm shift for regulators. The concept raises some fundamental questions about the goals of current structures. For example, if low-level exposures are beneficial, should the regulators strive to ensure that populations receive such exposures? Should standards be set to compel a minimum level of exposure to pollutants? I suspect that the answer to this should be no. Our government has occasionally acted to ensure public exposure to desirable substances, e.g., fluoridation of water, but it has not done so under the rubric of environmental protection legislation. When EPA recognized beneficial consequences from essential nutrients such as selenium, it took no action to ensure the existence of the minimum exposures for such beneficial consequences. The laws are clearly written from the perspective of avoiding exposure to harmful substances, not increasing exposure to beneficial ones. Any action to compel beneficial low-level exposures should come only after new legislation to that effect has been adopted.

Assuming that regulators are concerned only with the portion of the J-shaped curve that is going up with additional exposure, what exposure level should be regulation's target? Should regulators strive to reduce exposure only to the level (1.0) that corresponds with zero exposure or should regulators try to reduce exposure to the bottom of the trough of the J-shaped curve, where health benefits from exposure are maximized? This is a more difficult question to answer. When a statute calls for a "margin of safety," what is meant by safe? My inclination is to set the primary target at the zero exposure level rather than the bottom of the trough, because the organic legislation seems interested only in circumstances in which a substance presents a net risk to health and because this fits better with concepts of margin of safety. Even when the risk is on the upward-sloping portion of the curve, the net effect of the exposure is beneficial so long as it remains below the 1.0 level.

If the target is set at the risk prevailing at the zero exposure level, how should an agency go about setting a margin of safety? I would argue that the agency should set the standard midway between the zero exposure level and the bottom of the trough of the curve. This truly reflects the substance and spirit of the concept of margins of safety, much better than do prevailing theories. The target for standard setting under my proposal is illustrated in Figure 4.

Figure 4

[SEE ILLUSTRATION IN ORIGINAL]

Given uncertainties in science, we can never be sure exactly where the zero exposure level and the bottom of the trough are—we can only estimate based on the best available data. Margins of safety are meant to account for such inevitable scientific uncertainties. Setting the standard halfway between zero exposure and the trough preserves some margin of safety on both ends of the continuum. Such a standard protects against the danger that we have left a residual risk greater than the zero exposure level while simultaneously protecting against the danger that we have reduced exposures into the zone of beneficial effects.

Incorporating hormetic effects into the margin of safety is not the only issue relevant to hormesis in carcinogen regulation. Whatever the exposure level target, the existence of hormetic effects counsels for some particular approaches to environmental protection, as discussed in the following section. Some types of standards are ill-suited to a world of hormetic effects.

[30 ELR 10784]

Implications of Hormesis for Environmental Policy Design

Recognizing the validity of hormetic effects has important implications for statutory and regulatory design of environmental protection. Current statutes and rules have a wide range of regulatory approaches. Some statutes, such as §§ 108-110 of the CAA, provide for ambient standards, setting an allowable ambient exposure for certain substances and making provisions for source regulation so that ambient maxima are not exceeded. Because ambient levels cannot be directly regulated, however, a large number of provisions in environmental statues, especially those regulating carcinogens, provide for source-specific regulations. These regulations provide for setting emission standards for particular sources or source-categories, often based upon the technological feasibility of emission reductions. Once hormetic effects are recognized, ambient standards make much more sense.

The source-specific emission standards operate on an often unstated assumption that less exposure is inherently preferable to more. Nor do these standards typically account for prevailing ambient exposure levels for the substance. Consequently, a given standard may have a beneficial effect or a detrimental effect, depending upon the ambient levels of the regulated substance. Setting standards without regard to whether they have a positive or negative effect does not seem very rational.

Ambient standards, by contrast, can be set to ensure health protection. Such standards could take into account the dimensions of the hormetic effects curve. Ambient standards can avoid the risk of overregulation that would push exposures into the downward-sloping portion of the hormetic risk curve. They can also avoid the risk of underregulation; when exposures exceed the ambient standards they are to be reduced, and are not to be evaded by industry-specific complaints about the feasibility of further reductions.

Of course, ambient standards are not self-executing and must be translated into emission reductions from particular sources. This translation can be done in different ways. Under the CAA ambient air quality standards, the EPA sets the ambient standard, while state and local governments are to adopt implementing emission standards. Disconnecting the ambient from the emission standards, though, has not been perfectly effective. Many localities are not in compliance with the ambient standards and are called nonattainment areas, due to the hard political choices that states must make in setting the emission standards.⁵³ Political realizability has also precluded the national government from imposing any truly severe penalties for local noncompliance.

The recently enacted FQPA employs an interesting system—called a "risk cup"—that suggests an efficient means for implementing ambient standards.⁵⁴ For pesticides of a given type that act in the same manner on human health, the agency sets a maximum allowable exposure level. When this maximum, called the cup, is filled, no further sources of exposure will be permitted. This risk cup makes little sense in a world of linear risk assessment, because it may cause a shift to a more carcinogenic, though less used, alternative pesticide.⁵⁵ In a world of hormetic effects, though, the risk cup is a very sensible regulatory approach.

EPA could set the cup at the desired exposure point that avoided a risk of net adverse effects. This would help ensure that human exposures are not harmful. In the context of pesticides, the risk cup means that the Agency does not even have to concern itself with regulating particular sources, because the free market makes the maximum almost self-executing. If a given pesticide's risk cup is full, no further uses will be registered. The potential registrant can evade this problem by "buying" existing, less profitable uses and discontinuing them. In this process, public health is preserved, while the most valuable uses of the substance are enabled.

EPA's air pollution policies contain a variety of analogous procedures. The new acid deposition regulatory program set up an emissions trading system, in which EPA authorized an overall level of sulfur dioxide emissions and allowed polluters to purchase and sell rights within that ceiling.⁵⁶ The Agency has also adopted a "bubble" policy that enables individual facilities to increase emissions from one point if they provide for a comparable reduction in emissions from another point within the facility.⁵⁷ The Agency also has an "offset" policy that enables a company to construct a new source of pollution even within a nonattainment area, if the company provides for a more than offsetting reduction in pollution from other sources.⁵⁸

There are other difficult structural regulatory issues to be grappled with, once hormetic effects are recognized. For example, there is variability in the human population, and some may be especially sensitive to exposure to a given substance. The margin of safety was intended in part to deal with this issue, to protect sensitive populations. The margin of safety concept is not helpful, though, if some populations gain health benefits from a particular exposure level, while others suffer adverse health effects at those same levels. One could take a utilitarian approach and suggest that the law maximize overall public health, considering the number in each population and the magnitude of the health effects. Under such a rule, the most sensitive populations would be left to suffer adverse health effects. But there is a legitimate counterargument that we should regulate to restrict exposures to protect the sensitive population, even at a health cost to a larger population.⁵⁹ This is the sort of trade off best left to the accountable branches of government. If Congress does not act, the agencies could conduct a notice-and-comment rulemaking and choose an approach for protection of especially sensitive populations.

[30 ELR 10785]

Conclusion

The evidence for the universal existence of hormetic effects, for carcinogens or other substances, is not proved conclusively. But this is an unrealistic standard—nothing can be scientifically proved beyond dispute and prevailing linear risk assessment models are certainly not conclusively proven. The evidence for hormetic effects at low exposure levels is if anything stronger than the evidence for harmful effects at these levels. Once hormetic effects are recognized, regulatory policy must respond. Prevailing statutes generally allow room for such a response, adjusting for hormetic effects with either more or less stringent regulation. However, improved regulation would call for some statutory changes, to adapt to a world of hormetic effects.

1. You may be familiar with the concept as an avid reader of the Environmental Law Reporter, which is virtually the only environmental journal to so much as mention hormesis. See Edward J. Calabrese & Robyn Blain, A Single Exposure to Many Carcinogens Can Cause Cancer, 28 ELR 10254 (May 1998); Edward J. Calabrese, Hormesis Revisited: New Insights Concerning the Biological Effects of Low-Dose Exposures to Toxins, 27 ELR 10526 (Oct. 1997). Dr. Calabrese and his associates at the University of Massachusetts-Amherst have accumulated and summarized an enormous amount of research on dose-response to hazardous substances and hormetic effects. An organization on the Biological Effects of Low Level Exposures (BELLE), formed from the International Society of Regulatory Toxicology and Pharmacology, has done likewise and publishes a monthly newsletter that reviews much of this information.

2. Environmental statutes generally call for standards to be based on science. See Wendy E. Wagner, The Science Charade in Toxic Risk Regulation, 95 COLUM. L. REV. 1613 (1995). Wagner suggests that the standards generally are not in fact scientifically based. Id. at 1628-49. Yet she too believes that science should be crucial to standard-setting and proposes reforms to further this end. Id. at 1701-19.

3. Hormesis has also received relative little awareness among even toxicologists, for a variety of other scientific and psychological reasons. See Edward J. Calabrese & Linda A. Baldwin, The Marginalization of Hormesis, 27 TOXICOLOGICAL PHARMACOLOGY 187 (1999) (ascribing the situation to the unfortunate association of hormesis and homeopathy, focus on high-dose effects, the lack of an evolutionary mechanism to account for hormesis, and the lack of scientific advocacy for the concept).

4. Pub. L. No. 104-170, 110 Stat. 1489 (1996).

5. Environmental regulations have not been used to compel additional exposures when appropriate to improve the health of the public. The laws are aimed at preventing bad outcomes rather than creating good outcomes.

6. This adaptive response may take the form of enhanced detoxification of the invading substance or a more general defense mechanism for threats of cellular damage. See Calabrese, Hormesis Revisited, supra note 1, at 10528 (discussing stimulatory responses that include an "enhanced metabolic capacity for detoxification of the particular toxicant" and "those that offer more general protection against cellular damage caused by a wide variety of agents"). For experimental confirmation of the overcompensation effect, see, e.g., Edward J. Calabrese, Evidence That Hormesis Represents an "Overcompensation" Response to a Disruption in Homeostasis, 42 ECOTOXICOLOGY & ENVTL. SAFETY 135 (1999).

7. See, e.g., Robert L. Sielken, Cancer Dose-Response Extrapolations, 21 ENVTL. SCI. TECH. 1037 (1987).

8. Thomas D. Luckey, Physiological Benefits From Levels of Ionizing Radiation, 43 HEALTH PHYSICS 6 (1982).

9. Sadao Hattori, State of Research and Perspective on Radiation Hormesis in Japan, 3 BELLE NEWSL. 1 (1994).

10. Id. at 2.

11. See, e.g., Edward J. Calabrese & Linda Baldwin, Radiation Hormesis: Its Historical Foundations as a Biological Hypothesis, 8 BELLE NEWSL. 2 (1999).

12. Calabrese, Hormesis Revisited, supra note 1, at 10528.

13. Edward J. Calabrese et al., Hormesis: A Highly Generalizable and Reproducible Phenomenon With Important Implications for Risk Assessment, 19 RISK ANALYSIS 261 (1999); Edward J. Calabrese & Linda A. Baldwin, Can the Concept of Hormesis Be Generalized to Carcinogenesis?, 28 REG. TOXICOLOGY & PHARMACOLOGY 230 (1998) (showing that hormesis was generalizable to carcinogen exposures in different species).

14. Id. at 264.

15. Id. at 267, summarizing the research results of H.E. Lkeczkowska & F.R. Althaus, Response of Human Keratinocytes to Extremely Low Concentrations of N-methyl-N'-nitro-N-nitrosoguanidine, 367 MUT. RES. 151 (1996).

16. Id. at 268-69.

17. Id. at 270 (discussing research showing that at low levels, dioxin-treated rats displayed "substantial decreases in tumors" at various sites, including the liver).

18. See, e.g., Edward J. Calabrese & Linda A. Baldwin, The Dose Determines the Stimulation (and Poison): Development of a Chemical Hormesis Database 16 INT'L J. TOXICOLOGY 545, 553 (1997) (noting that the magnitude of the beneficial stimulatory response "has been observed as high as severalfold, but the majority of low-dose stimulations are 30-60% greater than the controls").

19. See, e.g., Lester B. Lave et al., Information Value of the Rodent Bioassay, 336 NATURE 631 (1988) (questioning validity of extrapolating results from animal tests to humans).

20. See, e.g., Celia Campbell-Mohn & John S. Applegate, Learning From NEPA: Guidelines for Responsible Risk Legislation, 23 HARV. ENVTL. L. REV. 93, 103 (1999) (discussing conservative risk assessment assumptions). Common complaints about conservatism in risk assessment have bred a response from environmentalists who contend that assessment is not so conservative. While the degree of conservatism may be exaggerated by some, the presumptions of prevailing regulatory assessment methods are generally conservative. A thorough analysis of pesticide regulation by the National Research Council of the National Academy of Sciences concluded that risk assessment was grounded in conservativism. NATIONAL RESEARCH COUNCIL, REGULATING PESTICIDES IN FOOD: THE DELANEY PARADOX 50 (1987).

21. See Edward J. Calabrese & Linda A. Baldwin, Hormesis as a Default Parameter in RfD Derivation, 17 HUMAN & EXPERIMENTAL TOXICOLOGY 444 (1998).

22. See 53 Fed. Reg. 43305, 43306 (Oct. 26, 1988).

23. See, e.g., Jefferey A. Foran, Regulatory Implications of Hormesis, 18 HUMAN & EXPERIMENTAL TOXICOLOGY 441 (1999).

24. See Kenneth A. Poirier & Michael L. Dourson, Is the Current Risk Assessment Paradigm Used by U.S. EPA and Others Compatible With the Concept of Hormesis?, 8 BELLE NEWSL. 22 (1999).

25. 42 U.S.C. § 7412(f)(2)(A), ELR STAT. CAA § 112(f)(2)(A).

26. Id. § 7409, ELR STAT. CAA § 109.

27. Id. § 7412(f)(2)(A), ELR STAT. CAA § 112(f)(2)(A).

28. 33 U.S.C. § 1317(a)(4), ELR STAT. FWPCA § 307(a)(4).

29. 42 U.S.C. § 300g-1(b)(4), ELR STAT. SDWA § 1412(b)(4).

30. Id. § 300g-1(b)(4), (5), ELR STAT. SDWA § 1412(b)(4), (5).

31. 15 U.S.C. § 2605(a), ELR STAT. TSCA § 6(a) (language from Toxic Substances Control Act).

32. 21 U.S.C. § 346a(b)(2)(A)(i).

33. Id. § 346a(b)(2)(C). See Kenneth Weinstein et al., The Food Quality Protection Act: A New Way of Looking at Pesticides, 28 ELR 10555 (Oct. 1998).

34. William D. Ruckelshaus, Stopping the Pendulum, ENVTL. F. Nov./Dec. 1995, at 25, 29.

35. 448 U.S. 607, 10 ELR 20489 (1980) (plurality opinion).

36. Id. at 612, 10 ELR at 20490.

37. For a discussion of applying the principles of the decision in the context of other statutes, see Frank B. Cross, Beyond Benzene: Establishing Principles for a Significance Threshold on Regulatable Risks of Cancer, 35 EMORY L.J. 1 (1986).

38. See 448 U.S. at 656, 10 ELR at 20501.

39. See supra note 20 regarding conservative assumptions. EPA uses conservatism in order to create an "upper bound" on the risk threatened. See, e.g., 55 Fed. Reg. 8292 (Mar. 7, 1990) (EPA emission standards for benzene transfer operations).

40. See Philip H. Abelson, Exaggerated Risks of Chemicals, 48 J. CLINICAL EPIDEMIOLOGY 173, 174-77 (1995).

41. STEPHEN BREYER, BREAKING THE VICIOUS CIRCLE 47 (1993).

42. U.S. EPA, Proposed Guidelines for Carcinogen Risk Assessment, 61 Fed. Reg. 17960 (Apr. 23, 1996).

43. Id. at 17962.

44. Id. at 17964.

45. Id. at 17968.

46. Id. at 17963 (noting that in practice, "the agency's assessments routinely have employed defaults").

47. 63 Fed. Reg. 15674, 15686 (Mar. 28, 1998).

48. Id. at 69390, 69398 (Dec. 16, 1998).

49. Chlorine Chemistry Council v. EPA, 206 F.3d 1286, 30 ELR 20473 (D.C. Cir. Mar. 31, 2000).

50. See, e.g., Cass R. Sunstein, Is the Clean Air Act Unconstitutional?, 98 MICH. L. REV. 303, 353 (1999) (noting that TSCA and Federal Insecticide, Fungicide, and Rodenticide Act tests of unreasonable risk clearly contemplate cost-benefit balancing).

51. See, e.g., American Textile Mfrs. Inst., Inc. v. Donovan, 452 U.S. 490 (1981) (the "cotton dust" case; OSH Act does not provide for cost-benefit balancing of standards); Lead Indus. Ass'n v. EPA, 647 F.2d 1130, 10 ELR 20643 (D.C. Cir.), cert. denied, 449 U.S. 1042 (1980) (the national ambient air quality standards of the CAA do not permit cost considerations); American Trucking Ass'n v. EPA, 175 F.3d 1027, 29 ELR 21071 (D.C. Cir. 1999), cert. granted sub nom. Browner v. American Trucking Ass'n, No. 99-1257, 2000 U.S. LEXIS 3577 (U.S. May 22, 2000).

52. As political bodies, agencies are unlikely to wholly disregard the economic costs of their actions, even when their enabling statutes so dictate. See Craig N. Oren, Run Over by American Trucking Part I: Can EPA Revive Its Air Quality Standards?, 29 ELR 10653, 10662 (Nov. 1999) ("EPA inevitably must therefore consider costs in [CAA] standard-setting to help decide how stringent to make the standards. Indeed, EPA decisionmakers have admitted that they examine cost data when deciding on the levels of the standards.").

53. See OFFICE OF AIR QUALITY PLANNING AND STANDARDS, U.S. EPA, NATIONAL. AIR QUALITY AND EMISSIONS TRENDS REPORT 1-2 (1995) (noting that 127 million people still lived in some nonattainment areas).

54. For a discussion of the concept, see Frank B. Cross, The Consequences of Consensus' Dangerous Compromises of the Food Quality Protection Act, 75 WASH. U. L.Q. 1155 (1997).

55. Id.

56. 42 U.S.C. §§ 7651-7651e, ELR STAT. CAA §§ 401-406. See generally Larry B. Parker et al., Clean Air Act Allowance Trading, 21 ENVTL. L. 2021 (1991).

57. For a recent discussion of the bubble and a related "netting" policy, see Stephen M. Johnson, Economics v. Equity: Do Market-Based Environmental Reforms Exacerbate Environmental Injustice?, 56 WASH. & LEE L. REV. 111, 125-26 (1999).

58. See Final Policy Statement, 51 Fed. Reg. 43814 (Dec. 4, 1986). The CAA now expressly allows companies to use offsets. 42 U.S.C. § 7475, ELR STAT. CAA § 165.

59. This argument would be grounded in the law's fundamental concern for avoiding adverse health effects rather than promoting beneficial health effects. The argument would find further support in the fact that some laws express particular concern for the health of sensitive populations.