Regulatory Forum Opinion Piece

Abstract

Long term bioassays in animals cannot reliably forecast unknown potential but distant human risks, and especially cancer risks. The genetic, anatomic, physiologic, behavioral and environmental adaptations of rats and mice - the officially prescribed animals - are not relevant to humans. Even bioassay results for the two prescribed species are not mutually predictive. The dearth of human relevance is augmented by arbitrary and incongruous default assumptions, also officially prescribed for the conduct and interpretation of bioassays in rats and mice. Moreover, and contrary to publicized perceptions, bioassay results are freely evaded in the markup of regulations, with the imposition of arbitrary safety factors and the guided opinions of ad hoc appointed advisory committees. Regardless of bioassay results, actual regulations of unknowable distant risks end up allowing those minimum exposures that are still compatible with uses deemed necessary or useful for the common welfare. Thus it would seem sensible to do away with very costly long-term bioassays irrelevant to humans and whose results are anyway bypassed, and to focus regulations on short-term effects relevant to humans, and on transparent cost and benefit considerations toward minimizing useful exposures.

Two-year or lifetime animal bioassays are presumed to forecast potential long-term human risks that cannot be tested in people, and especially human cancer risks. As tools of regulation, such tests are not academic exercises but legal justifications for enforcements that can deter useful applications, severely restrict business, impose drastic fines, and even detain transgressors. By expectation, those enforcements ought to rely on objective and testable evidence, at least in free societies served by fair governments. Do lifetime animal bioassays provide objective and testable evidence of human risk?

The answer to this question is not reassuring. Rats and mice that by law must populate long-term bioassays are clearly not human, for their genetic, physiologic, lifestyle, longevity, behavioral, environmental, and dietary adaptations are very much different. In fact, staying with cancer risk as the most common long-term endpoint, bioassay results in rats do not predict results in mice any better than tossing a coin, and similar discrepancies are also observed within strains of rats and mice and with other species.

Even without counting the uncertainties of gross pathology and histological interpretations, it is manifest that animal bioassays are not objective models of long-term human risks. The International Agency for Research on Cancer (IARC 1980) recognized long ago that such bioassays do not provide scientifically justified proxies of human risks  and in a 1983 seminal report the National Academy of Sciences warned that the uncertainties are insurmountable (National Research Council, National Academy of Sciences 1983). In 1981, supported by an Assistant Secretary of Health and the directors of the National Cancer Institute and of the Occupational Safety and Health Administration, the late Dr. David Rall, the then director of the National Institute of Environmental Health Sciences and of the National Toxicology Program, testified to an astonished Congressman Albert Gore that bioassays did not provide scientific evidence of human cancer risk. He went on to plead that Congress and Americans ought to have faith [sic] in the judgment of experts who examine animal entrails (Rall 1981).

Since today bioassays remain just as bereft of testable evidence of human risk, how is it possible they still provide mainstay regulatory justifications? The mandate for this practice is commonly attributed to the 1958 Delaney Amendment to the Federal Food, Drugs, and Cosmetics Act (FFDCA), which prohibits additives that cause cancer in animals. In fact, the intent of this amendment was rapidly extended, by implication if not by letter, to regulations beyond the FFDCA, and has influenced the National Toxicology Program: the prime supplier of regulatory incentives outside the Food and Drug Administration. While the attribution to Delaney may be correct, it would have been sensible to ask Congress for clarification on how animal tests should be done, given the obvious and serious handicaps of bioassays. Instead, regulators took advantage of their delegated authority and opted to decree on their own how animal tests should be performed and interpreted. The latest summary of required rules can be found in a 2005 document issued by the U.S. Environmental Protection Agency.

Among other requirements, those rules spell out a series of mandatory default assumptions, which uphold that rats and mice must be taken as long-term proxies for humans, that tests must be conducted at overtly toxic maximum tolerated doses, that challenging animals at such doses during their lifetimes is equivalent to smaller doses that humans may experience, that metabolism of animals at maximum tolerated doses is equivalent to the metabolism of humans at very low doses, that benign lesions are to be considered on par with malignant ones, that the route of exposure is irrelevant, that extrapolations from high doses in animals to low doses in humans must be linear, and more kindred assumptions extending to the odd assertion “[a]nimal studies are conducted at high doses in order to provide statistical power. . . .”

Obviously, not a defensible prescription for a scientific approach, this arrangement appears intent to maximize the probability of cancer signals from bioassays, and thus to provide regulators with more tools for action. Are those signals relevant to human risks? Indeed, executed as ordered, bioassays readily indict food and wear items in common use for centuries, which no one would seriously ban or restrict. A sensible haven for items generally recognized as safe (GRAS) was quickly set up by the Food and Drug Administration. Still more ways were found by all agencies to bypass inconvenient bioassays—positive or negative—that could be in the way of what advocacies, the media, the agencies, Congress, and people may expect.

Two complementary devices have been adopted. One is the liberal use of precautionary but subjective safety factors introduced at various steps during the regulatory markup; the other is the appointment of advisory committees set up to mitigate the apparent raw arbitrariness of rulemaking. Noticeably, however, those committees could not rise above subjectivity, confronted as they are with the dialectical uncertainties of “weight of evidence” arguments and not with defensible scientific data relevant to humans (Science and Decisions 2009). To retain control and rein in heresy, regulators have invoked conflict of interest obstacles, which they use to select and exclude potentially dissenting committee members, and thus avoid “… so much constant debate that the rulemaking apparatus simply seizes up entirely” (Science for Policy Project 2009). Strange arrangements for sure, where infallible and conflict of interest free prosecutors are also the judges, write the rules, and appoint jury members. Be as it may, the predictable end result is that regulations—by nature restrictive—end up banning items, or allowing the very least exposures compatible with uses that are or seem necessary for the common welfare.

All in all, and even forgoing obvious ethical considerations, the argument for long-term animal bioassays looks rationally and functionally lame. Although paraded as scientific centerpieces of regulation, bioassays only provide an illusion of scientific support. They cannot offer quantitative or qualitative scientific evidence relevant to human risk, and mercifully their results are freely thwarted by the application of extensive safety factors, and the deferential opinions of handpicked advisory committees.

Supposable long-term risks that cannot be tested in humans will remain inevitably hypothetical, unless later discovered by epidemiologic evidence. Long-term bioassays are not helpful, for their outcomes will diverge in different species under the influence of genetics, physiology, diet, lifestyle, life span, environment, diseases, and other determinants. Adding uncertainty, in the case of cancer and other long-term disorders, random and time-dependent cascades of causal events are most likely different for each pathology, both in kind and in timing. For these and many more reasons, in silico or in vitro genomic, metabonomic, and other “omic” approaches are even less likely to provide integrated causal evaluations of long-term pathologies that could be generalized. By contrast, short-term effects would still be useful after appropriate allometric considerations, offering credible real-time responses from relatively stable somatic systems conserved in many species, including humans. Acute toxicities, adverse genetic outcomes, immunosuppressive signals, tissue proliferation, injury, inflammation, and other warnings: all would continue to assist in the regulation of agents whose long-term risks remain unknown.

For those risks, the prudent approach is to heed the advice of Paracelsus and to regulate for the least exposures compatible with uses considered socially desirable. Permitted exposures would depend on the level of urgency or essentiality of a proposed use, an agent’s index of effectiveness, its short-term toxicities, toxic kinetics, somatic and environmental persistence, what populations might be exposed, and kindred considerations. Public health would be further advanced by a program of post-marketing epidemiologic surveillance similar but more pervasive than what has been long done for approved drugs and other therapeutic agents.

Could we miss long-term hazards and risks? Of course, but this is what we do and must do every day, as we manage to survive in a sea of inescapable hazards. Despite illusory protestations of scientific rigor, this is what current regulations are actually doing, and what they would still be doing in the absence of hapless and extravagantly expensive chronic bioassays. Salutary in other ways, the demise of these tests could just be triggered by the current global funding woes: it would redirect massive resources worldwide, return thousands of scientists to productive research, spare countless animals, and restore a much needed credibility to regulation.