Will the use of human subjects make pesticides safer or simply line the pockets of companies in the industry?

Charles W. Schmidt

Scientists trying to understand how poisons affect humans have always been hampered by a great limitation: Human subjects are usually off-limits in the laboratory. To fill this void, scientists rely on experimental animals, which differ physiologically from humans in many ways, and epidemiology, which is limited by enormous uncertainties over exposure and dose–response. The field of risk assessment, which is used to set regulatory standards for chemicals, is based almost entirely on predicting human effects by indirect means. To protect people from this uncertainty, regulatory standards for chemicals are often set at levels thousands of times lower than those actually considered toxic.

When the standard-setting process is viewed in this context, it isn’t surprising that some product development managers have found the temptation to use human subjects in toxicology experiments almost irresistible. However, testing humans in the laboratory represents a Pandora’s box of ethical and scientific concerns that society has grappled with for centuries.

The U.S. Environmental Protection Agency currently is engaged in a debate about using human subjects for testing, which will have important scientific, ethical, and economic consequences. The origins of the debate lie with a 1996 law, the Food Quality Protection Act (FQPA) (1), under which existing standards for more than 9000 pesticides are being evaluated for their ability to protect children. Under the law, pesticides that are found to pose a health risk to children could be subject to additional regulations that would limit their use.

Responding to the threat posed by the FQPA, several pesticide companies have turned increasingly to human experiments in an attempt to reduce uncertainties regarding the health effects of their products. So far, most of these studies have investigated oral exposures to neurotoxic pesticides in paid volunteers. The emphasis on neurotoxins derives from the fact that the FQPA directs EPA to evaluate pesticides on a “worst-first” basis. The agency has given the “worst” designation to organophosphate and carbamate insecticides, both widely used classes of pesticides that target a neurotransmitter called acetylcholinesterase (AChE) in insects and humans.

However, many public-sector stakeholders and nongovernmental organizations are angered by the prospect of human data being used to weaken pesticide regulations intended to protect children. On November 19, 1999, 45 organizations that included the Natural Resources Defense Council, the Children’s Environmental Health Network, and Physicians for Social Responsibility, sent letters to EPA Administrator Carol Browner, expressing opposition to such tests on both scientific and ethical grounds.

Struggling toward a policy
Having recognized the need for a comprehensive policy on human data submitted for regulatory purposes, EPA convened a joint panel of the Scientific Advisory Board and Scientific Advisory Panel (SAB–SAP) in late 1998 and asked it to prepare a set of recommendations. Specifically, the panel was asked to provide guidance on several points, including

• the proper role of human studies as a supplement for animal data in regulatory decision making,
• the risks and benefits of human testing to subjects and society, and
• appropriate ethical standards for human experiments.

EPA also asked the panel to provide guidance on how existing protocols for human testing, for example, the Common Rule (2) and the Declaration of Helsinki by the World Medical Association (3), would apply for studies submitted in support of pesticide registration.

The SAB–SAP panel includes leaders in the fields of toxicology, risk assessment, and ethics; it met in December 1998 and November 1999. But almost a year and a half after the panel was convened, it remains internally divided (some say “hopelessly deadlocked”), with numerous anxious stakeholders awaiting its conclusions. Panel members concede that the final report, yet to be issued, will not provide EPA with a consensus view. Meanwhile, several dozen unsolicited human studies await EPA review, most of them submitted on behalf of pesticide manufacturers since the FQPA was passed in 1996. These studies, and the aims of their sponsors to submit them for regulatory purposes, lie in limbo until EPA’s policy is established.

The English patients
In 1997, Medeval Laboratories in Manchester, England, on behalf of the California-based Amvac Chemical Co., gave adult men oral doses of the neurotoxic insecticide dichlorvos dissolved in corn oil. In September 1998, the effects of oral doses of azinphos-methyl, a neurotoxic insecticide recently banned by the EPA because of childhood health concerns, were evaluated in human studies performed by Inveresk Clinical Research Ltd., an international contract laboratory in Tranent, Scotland. Both studies focused on the pesticide’s low-dose effects on AChE, a practice that many scientists insist is safe if done properly. (These studies have not been published.)

The question of human testing surfaced as an issue for EPA in July 1998, when the Washington-based Environmental Working Group (EWG) published a report called The English Patients (4). This report revealed that the agency was accepting human studies submitted by pesticide companies, which, EWG claimed, had a “huge financial incentive to test people instead of other animals.” The report’s title reflects that most human studies on pesticides have been done in England and Scotland, for reasons neither U.S. nor British environmental officials could explain. “We felt that this was a perverse attack on a law that was meant to strengthen pesticide regulations,” says Kenneth Cook, president of EWG.

EPA officials were caught off guard by the publicity generated by the report and admitted that the agency had accepted human studies in the past. In fact, EPA’s 1998 document Guidelines for Neurotoxicity Risk Assessment (5) supports obtaining and using human data, although officials say submission of this data in the context of regulatory and licensing programs was never required or encouraged.

“What was surprising at all levels,” says Marcia Mulkey, director of Pesticide Programs at EPA, “was to learn that there had been a decision on the part of several pesticide companies to embark on a significant increase in the number and scale of these studies.”

Why would pesticide companies want to test their products on humans instead of animals? Because human data can, in some cases, permit the removal of safety factors that reduce permissible exposure levels based on animal data.

EPA and NOAELs
EPA sets pesticide tolerances (allowable residues on food) using risk assessment: Groups of laboratory animals are given consecutively lower doses of pesticide until a dose with no target effects is identified. This dose is known as the no observed adverse effects level, or NOAEL.

To reconcile the variation in animal and human response, the NOAEL is divided by a series of numerical safety (or uncertainty) factors to arrive at a health-based value called the reference dose, or RfD. The RfD is defined as the dose of a chemical that the most sensitive humans (e.g., children) can consume safely, every day, for a lifetime of 70 to 75 years. Tolerances are residue levels that won’t exceed the RfD, according to some predetermined consumption rate.

The goal of the pesticide companies is to show that uncertainty factors applied to the NOAEL are either too high or unnecessary. Their target is the “interspecies uncertainty factor”, which is used by risk assessors to account for differences in response from animals to humans. The FQPA plays a pivotal role because it introduces an additional uncertainty factor to protect children that reduces tolerances by a factor of 3 to 10—enough to push some pesticides out of the market altogether. But if a pesticide company can produce a defensible NOAEL based on a human study, the 10-fold interspecies uncertainty factor might be reduced, or even eliminated, and the tolerance for the chemical would be raised commensurately. Of course, the key to the strategy lies in producing data that show human subjects are less sensitive than animals. Pesticide companies seem to find this an acceptable gamble.

At the November 30, 1999, meeting of the SAB–SAP panel, Mulkey said that most human studies submitted to the agency thus far showed humans were “rarely 10 times more sensitive than animals,” and often “lead to a regulatory choice to tolerate more exposure.”

A question of sample sizes
Details of the panel’s final report, which currently has no release date, were not available at press time. However, a review of a preliminary draft, which observers say will closely mirror the final product, reveals a difficult attempt to reach consensus from diverging views. Panel members agree that human testing is acceptable if the subjects are “rigorously protected” and appropriate statistical design and power are used.

The sharpest divergence concerns the extent to which neurotoxic pesticides should be tested on human subjects, which is ironically the area in which the panel’s guidance is most urgently needed. The differences are particularly strong over the AChE inhibitors of particular relevance to the FQPA. Perhaps most importantly, panel members agree unanimously that human tests should never be performed merely for the purpose of deriving a NOAEL. But—opening a window of opportunity for pesticide companies—the panel implied that incidental human NOAELs might be suitable for regulatory purposes if they are included in tests with “a larger strategic plan.”

Most panel members refused to comment on this issue. However, University of Pittsburgh professor of child psychiatry and pediatrics Herbert Needleman, a panel member best known for his pioneering work on childhood effects of lead, and who strongly opposes human testing with pesticides, bases much of his position against human NOAELs on the limits of experimental sample sizes. According to Needleman, sample sizes in human studies are so small, generally fewer than 10 subjects per dose group, that researchers run the risk of missing an effect in the laboratory at a dose level that could produce an effect in the general population.

To illustrate his point, Needleman performed a power analysis (a technique used to establish minimum sample sizes for desired statistical confidence) that demonstrated that 2500 human subjects per dose group would be needed to detect a 1% increase in the background rate for mental deficits, such as cognitive impairment, learning disabilities, and behavioral problems, in children. This analysis was presented to subcommittee members at the November 30 meeting. “A 1% increase in the background rate would mean that 160,000 children under the age of five [in the United States] would be experiencing that effect,” he said. “The power analysis shows convincingly that the studies done to date have virtually no chance to find an effect, and therefore are exposing ‘volunteers’ to poisons for no scientific reason.”

Edward Gray, vice president of Jellinek, Schwartz & Connolly, a consulting company in Arlington, VA, that represents the pesticide industry, sharply disagrees with Needleman's analysis. Gray counters that EPA bases its regulatory NOAEL for organophosphates and carbamates on AChE inhibition, not mental deficits. EPA maintains that inhibition of this enzyme is a conservative endpoint for the NOAEL, because it’s a benign response that occurs at dose levels below those thought to be toxic. For Gray, a clinching rebuttal is that a statistically significant (and nonharmful) 15% drop in AChE levels from pesticide exposure in humans could be detected with typical sample sizes. Gray’s evidence for this was obtained from a report produced by Robert. L Sielken and Larry Holden on behalf of the American Crop Protection Association, a Washington, DC, lobbying group (6).

The panel also disagrees over whether human exposure to pesticides at levels that target AChE as the endpoint might actually be harmful. Some panel members suggest that human subjects can tolerate minor chemically induced fluctuations with no ill effect, and others suggest that latent or otherwise unexpected effects could occur at some time in the future. This disagreement forms the basis of the panel’s major divergence. A minority of members believe that human testing with neurotoxic pesticides should never be performed, but the majority lean toward accepting human testing with neurotoxins under certain limited conditions.

Ethical concerns
For ethicists, just the possibility of risk—regardless of toxicity—is a relevant concern. Panel member Jeffrey Kahn, the director of the Center for Bioethics at the University of Minnesota, Minneapolis, questions whether testing the effects of pesticides in human subjects is appropriate because the burdens of the research are borne by the subjects, whereas the benefits are enjoyed by others (e.g., pesticide companies). Kahn says that this splitting of risks and benefits isn’t necessarily unethical, but he cautions that it raises the potential for exploitation and argues for lowering the acceptable level of risk.

As an example of an acceptable risk–benefit split, he points to Phase I clinical trials using pharmaceuticals, in which low-level risks are borne by healthy volunteers, and the benefits go to society at large. Volunteers are allowed to take on higher risks from exposure when there is a potential for them to receive a direct medical benefit. For example, a patient with cancer might participate in a study of chemotherapy agents that are likely to be toxic. “Human subjects [in pharmaceutical trials] can view their participation in terms of their own direct medical benefit and their desire to be altruistic,” he says.

Some industry representatives draw parallels between pesticides and pharmaceuticals and suggest that human volunteers in pesticide studies do benefit from their participation, because the agents they are exposed to protect food supplies and kill insects that transmit disease. “There are a lot of social benefits to pesticides,” says Gray. “To say that all drug testing is OK, and no pesticide testing is OK, is illogical.”

A number of panel members respond skeptically to this argument and suggest that the motivation behind industry-funded human studies can be “couched in highly palatable language”. When it comes to the bottom line, they argue, most pesticide companies put their energy into these studies to protect a market for their products, not because of an altruistic desire to improve public health.

And so the true nature of the ethical debate over human testing might be reduced to a basic question: Is it ethically appropriate to perform human tests for financial purposes?

During the November meeting, panel member Samuel Gorovitz, professor of philosophy at Syracuse University, responded to this specific question. Whether industry happens to benefit financially from the experiment isn’t even EPA’s concern, he said. What matters is that the research has a legitimate goal of obtaining information that can “enhance or secure the health of the public”.

Confirming the scientific and ethical legitimacy of human studies is a delicate proposition, and not one in which EPA has much experience. According to observers, the panel will recommend that EPA rely on independent Institutional Review Boards (IRBs) to provide prospective analyses of human studies before they are actually approved. Furthermore, the panel will recommend that EPA aggressively scrutinize the structure, function, and activities of these IRBs. Protocols governing IRBs are contained in the Common Rule (2), which also contains provisions for key elements such as informed consent and appropriate selection of subjects.

The panel does appear to endorse a testing strategy that starts with animals for information about dose, mechanisms, and endpoints, and then proceeds to humans (nonpregnant adults only) for more subtle information on pharmacokinetics and biomarker response (with the exception of neurotoxic pesticides as discussed earlier). This strategy could be used to identify valuable areas for future research and poses fewer ethical questions than studies that simply establish NOAELs—which lack scientific value and whose purpose is simply to argue for higher permissible exposure.

Ultimately, these recommendations are only the first in a series that EPA will consider as it designs its final policy on human testing. Once the panel issues its report, EPA will submit it for public comment and begin a likely difficult period of deliberation. Says Mulkey, “I think we’ll be able to develop a policy position. It depends on the nature and clarity of the panel’s advice. Our idea is to move quickly on this.”

References

1. Food Quality Protection Act. Public Law 104-170 (H.R. 1627), 1996.
2. Protection of Human Subjects [Common Rule]. Code of Federal Regulations, Title 45, Part 46.
3. The World Medical Association Declaration of Helsinki: Recommendations guiding medical doctors in biomedical research involving human subjects. Adopted by the 18th World Medical Assembly, Helsinki, Finland, 1964 and as revised by the World Medical Assembly, Tokyo in 1975, in Venice, Italy, in 1983, and in Hong Kong in 1989; www.etikkom.no/NEM/REK/declaration96.htm (accessed March 23, 2000).
4. Environmental Working Group. The English Patients, Human Experiments and Pesticide Policy; Washington, DC, July 27, 1998.
5. Environmental Protection Agency. Final Guidelines for Neurotoxicity Risk Assessment; May 14, 1998; Fed. Regist. 1998, 63 (93), 262926–262954.
6. Sielken, R. L., Jr.; Holden, L. R. The substantial power of human study data to contribute to characterization of NOELs for cholinesterase inhibition in humans: a statistical analysis of recent studies. JSC Sielken, Bryan, TX; unpublished report.

Charles W. Schmidt is a freelance writer on science-related topics (171 Danforth St., Portland, ME 04102; 207-772-9672; cschmidt@gwi.net).

Return to Top