NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Research Council (US) Committee on Scientific and Regulatory Issues Underlying Pesticide Use Patterns and Agricultural Innovation. Regulating Pesticides in Food: The Delaney Paradox. Washington (DC): National Academies Press (US); 1987.

Cover of Regulating Pesticides in Food

Regulating Pesticides in Food: The Delaney Paradox.

Show details

2The Current System: Theory and Practice

The issues this report addresses arise from the interaction between two different statutes, the EPA's administrative practices and the growth in new information about pesticide toxicity and prevalence. The questions posed by the Delaney Clause can be understood only in the context of the system under which the EPA regulates pesticides. This chapter describes that system and identifies the developments that give rise to questions about the Delaney Clause. It concludes by summarizing the major issues the current study is intended to help resolve.

Registration of Pesticides Under FIFRA

The central event in the regulation of a pesticide is registration, which is EPA approval of one or more of its uses under the Federal Insecticide, Fungicide and Rodenticide Act (FIFRA). EPA registration of a pesticide use is required before the pesticide can be lawfully sold in the United States. The use of a pesticide in a manner inconsistent with the terms and conditions of its registration is unlawful.

The registration process is linked with the tolerance-setting process. Pesticides that are to be registered for use on food crops must be granted tolerances under the Food, Drug and Cosmetic (FDC) Act. Tolerances authorize and place legal limits on the presence of pesticide residues in or on raw agricultural commodities and, in appropriate cases, processed foods. The EPA will not register the use of a pesticide on food crops unless tolerances have first been granted to cover any residues expected to remain in or on food. Registration is nevertheless the logical starting point for a discussion of pesticide regulation because registration governs the uses of pesticides that result in food-borne residues.

Section 3 of FIFRA sets forth the standards for registration. The basic requirements are that the pesticide use must be able to accomplish its intended effect without causing ''unreasonable adverse effects on the environment," which the law defines as "any unreasonable risk to man or the environment, taking into account the economic, social, and environmental costs and benefits of the use of any pesticide."1 In proceedings to cancel or suspend the registration for use of a pesticide, the law further directs the EPA to consider the impact of the proposed action "on production and prices of agricultural commodities, retail food prices, and otherwise on the agricultural economy."2

Thus, FIFRA is a "balancing" statute. Congress recognized that pesticide uses can yield both risks and benefits and directed the EPA to consider both in deciding whether to permit particular uses of a pesticide. To grant registration, the EPA must conclude that the food production benefits of a pesticide outweigh any risks.

Under FIFRA the burden rests on the manufacturer or other would-be registrant to provide the data needed to support registration. The EPA regulations spell out in detail the data required in 40 CFR Parts 158 and 162.3 Required data include substantiation of the product's usefulness and disclosure of its chemical and toxicological properties, likely distribution in the environment, and possible effects on wildlife, plants, and other elements in the environment. If the applicant's data fail to prove that the product's use poses "no unreasonable adverse effects on the environment," registration is denied. In theory, the registrant continues to bear this burden even after registration and may be called on to prove its case again if new scientific data cast doubt on the EPA's original assessment of risk or balancing of benefits.

The conclusion of a successful registration process is the approval of a label for the product. This label sets forth detailed and legally binding instructions for use of the pesticide on certain crops, including any limitations or conditions on how or when the pesticide must be applied or not applied. Label specifications are generally designed to avoid adverse effects on the environment or on adjacent or future crops, to ensure efficacy, and to minimize applicator exposure.

Tolerance Setting Under the FDC Act

Pesticides that are to be registered for use on food crops must have been granted tolerances covering expected residues of the pesticide in raw and processed foods. Two different sections of the FDC Act, enacted four years apart, apply to the setting of tolerances. One, section 408, governs tolerances for pesticide residues on raw commodities. The other, section 409, governs tolerances for pesticide residues that concentrate in processed foods.

Section 408—The Statutory Standard

Congress enacted section 408 of the FDC Act in 1954 to enhance regulatory control over pesticide residues in food. It authorizes the establishment of tolerances for pesticide residues in or on raw agricultural commodities before they leave the farm gate. These tolerances are to be set at levels deemed necessary to protect the public health, while considering the need for "an adequate, wholesome, and economical food supply." Like the FIFRA standard for registration, section 408 of the FDC Act explicitly recognizes that pesticide uses confer benefits and risks and that both should be taken into account. The inquiry authorized by section 408 may not be as broad as that under FIFRA, yet section 408 clearly allows although does not compel the EPA to consider factors other than risks to human health.4

Residues of a pesticide on a raw agricultural commodity that exceed a section 408 tolerance or for which no tolerance has been established are deemed unsafe. The commodity itself is characterized as adulterated (and thus unlawful) under the FDC Act.5

Section 409

Section 409 of the FDC Act is the source of the Food and Drug Administration's (FDA) general authority to regulate the purposeful addition of substances to food. This provision empowers the FDA to require premarket approval for a varied universe of food additives, including artificial sweeteners, preservatives, chemical processing aids, animal drug residues, and packaging materials. Precisely how section 409 affects the EPA's regulation of pesticides requires some explanation.

Section 201(s) of the FDC Act initially defines the term "food additive" broadly to include "any substance the intended use of which results or may reasonably be expected to result . . . in its becoming a component of food."6 But it then expressly excludes from the definition pesticide residues in or on raw agricultural commodities, presumably because they are already covered by section 408. By necessary implication, however, pesticide residues in processed foods remain food additives and thus subject to the premarket approval requirement of section 409.7 The FDA has primary responsibility for implementing section 409, but the EPA has been delegated responsibility for regulating pesticide residues that are food additives.

Like section 408, section 409 establishes a procedure to secure approval for the uses of food additives. However, the standard for granting approvals under section 409 differs fundamentally from the risk-benefit standard of section 408. Section 409 requires the sponsor of a food additive to prove with reasonable certainty that no harm to consumers will result when the additive is put to its intended use.8 This so-called "general safety standard" for food additives is strictly risk based. It allows consideration of an additive's potential health risks and, by negative implication, seems to preclude consideration of any economic or other benefits.

In section 409, Congress also created a special rule for food additives that have been found to induce cancer in humans or animals. Under the famous Delaney Clause—enacted as a proviso to the general safety standard—no such additive can be approved (in the case of a pesticide this means "granted a tolerance") under section 409.9

A food additive that has not been approved under section 409 or that is present in food at a level exceeding a section 409 tolerance is deemed unsafe. Unsafe food additives, as well as the foodstuffs containing them, are adulterated and subject to the same enforcement procedures and penalties applicable to raw agricultural commodities.

If Congress had stopped here, pesticide residues in raw commodities and those in processed foods would be subject to different standards, but the distinction would be clear. The former would be regulated under the balancing criteria of section 408; the latter would be regulated under the risk-only standard of section 409, reinforced by the Delaney Clause. But Congress did not stop here, and it is Congress' further effort to integrate sections 408 and 409 that contributes much of the complexity in pesticide tolerance setting. In brief, not all pesticide residues in processed foods are regulated as food additives.

When it adopted section 409 in 1958, Congress realized that many, if not most, processed foods would contain at least some of the residues of pesticides lawfully present (under section 408) on the raw agricultural commodity used in their production. To facilitate regulation of pesticide residues falling within the definition of food additive—and hence requiring approval under both section 408 and section 409—Congress in effect exempted from "food additive" regulation residues that are present in a processed food at levels no higher than sanctioned on the raw agricultural commodity. Section 402(a)(2)(C) of the FDC Act provides that where a pesticide chemical has been used in or on a raw agricultural commodity in conformity with an exemption granted or a tolerance prescribed under section 408 and such raw agricultural commodity has been subjected to processing such as canning, cooking, freezing, dehydrating, or milling, the residue of such pesticide chemical remaining in or on such processed food shall, notwithstanding the provisions of sections 406 and 409, not be deemed unsafe if such residue in or on the raw agricultural commodity has been removed to the extent possible in good manufacturing practice and the concentration of such residue in the processed food when ready to eat is not greater than the tolerance prescribed for the raw agricultural commodity . . ..10

As a general rule, this proviso allows EPA approval of a residue under section 408 to suffice, as long as any residues in food processed from the raw agricultural commodity do not exceed the level authorized under section 408 (see the boxed article "Concentrating Residues"). Such residues remain subject to the balancing standard of section 408, and they escape the Delaney Clause.

Under this statutory framework, the concentration of a pesticide's residues in processed food has profound implications. To expose them, the committee first summarizes the procedural and analytical steps the EPA follows in setting tolerances under sections 408 and 409 and describes the universe of tolerances promulgated to date. Then the committee examines more fully the significance of discovering that pesticide residues concentrate in processed food.

The Tolerance-Setting Process Under Sections 408 and 409

Overview of the Process

Most tolerance-setting proceedings are initiated when a pesticide manufacturer files a petition with the EPA requesting establishment of a tolerance. The petition must be accompanied by or make reference to scientific data and technical information that the manufacturer believes satisfy the agency's data requirements. This information also must support a judgment that the tolerance can be established in compliance with statutory standards. The formal procedures for handling completed petitions under sections 408 and 409 differ slightly, but the same basic supporting data are mandated.

After reviewing a petition for completeness, the EPA publishes a notice in the Federal Register inviting comment on the proposed tolerance. At this point the underlying safety and residue data generally are not subject to examination by members of the public.11 After analyzing all the available data and considering any comments submitted in response to the proposal, the EPA either denies the petition or establishes a final tolerance. A notice announcing the EPA's action, including a brief summary of reasons, is published in the Federal Register. The tolerance is eventually codified in the Code of Federal Regulations (CFR). Both statutes permit opponents of a tolerance to object, request a hearing, and ultimately challenge the EPA's final decision in court, but these formal procedures are almost never invoked. Indeed, relatively few tolerance petitions evoke written comment from members of the public other than those affiliated with the pesticide industry.

Concentrating Residues: What Are They and When Does Delaney Apply?

The FDC Act dispenses with the need for a food or feed additive tolerance for any pesticide residue in processed food or feed when "the concentration of such residue in the processed food . . . is not greater than the tolerance prescribed for the raw agricultural commodity . . . ." Concentrating residues requiring food or feed additive tolerances must meet the act's safety standard. Under the Delaney Clause, the pesticide presumably cannot be approved if found to induce cancer in man or animal. Thus, the EPA's interpretation of the language relating to concentration can be critical. The central issue is whether the law makes the fact or the level of concentration the determining factor. Although there has been some confusion on the matter, the EPA's current position is clear. It is the fact of concentration that necessitates section 409 tolerances and thus potentially triggers the application of the Delaney Clause.

Raw agricultural commodity tolerances are based on the results of field trials designed to achieve the highest residue levels likely under normal agricultural practice. These studies include such methods as using the highest recommended application rates under weather and climatic conditions that prolong and in some cases exacerbate residues. Because of this, the tolerance is often higher than the actual residues found at harvest on crops grown in regions where application rates are lower and where residues dissipate more rapidly. In these cases a residue theoretically could concentrate during processing yet not exceed the level allowed on the raw agricultural commodity, which is the section 408 tolerance.

On occasion, tolerance petitioners have asked the EPA to set section 408 tolerances high enough to allow concentration of residues during processing to levels below these tolerances. The EPA reports it has denied all such requests and has relied on 40 CFR § 180.4 (1986), which states: "The tolerance established ordinarily will not exceed that figure which the Administrator of the Environmental Protection Agency states, in his opinion, reasonably reflects the amounts of residue likely to result."

When seeking a section 408 tolerance for a specific crop, the petitioner must address the need for section 409 tolerances by showing whether the residues concentrate as a result of specific processes such as drying, milling, or juicing. In determining whether a section 409 tolerance is required, the EPA focuses on whether residues in any processed product exceed those found on the unprocessed crop, not whether residues concentrate above some hypothetical section 408 tolerance. If residues concentrate, an average concentration factor is determined. The section 409 tolerance is set at a level equal to the section 408 tolerance multiplied by this concentration factor.

The logic of the EPA's practice is clear. A section 408 tolerance represents a residue level that may in some cases be realized. A section 409 tolerance must reflect the possible residue levels in processed foods derived from that raw commodity. (Source: 40 CFR § 180.4 (1986).)

Specific Data Requirements

Data requirements for tolerance petitions are spelled out in EPA regulations and guidelines. Much of the required information duplicates that needed to support registration under FIFRA, and is already available. Key elements of the data package include a description of the chemistry of the pesticide itself; identity and quantity of residues expected to be present in food; analytical procedures used in obtaining the residue data, which must be complete enough to permit replication by a competent analyst; residues in animal feed derived from crop by-products or from forages and resulting residues, if any, in meat, milk, poultry, fish, and eggs; and toxicity tests on the parent compound and any major impurities, degradation products, or metabolites.

The gathering and interpretation of residue chemistry data are some of the most difficult technical challenges that the registrant and the EPA face. The objective is to estimate and fully track the principal food residues, including metabolites and degradation products, that are likely to result from the commercial use of a pesticide under varying climatic and soil conditions. This generally requires extrapolation from data on a limited number of field trials in different parts of the country where the pesticide would possibly be used. In considering the level at which to set a tolerance, the EPA will generally select the highest residue levels reported in such tests.

The toxicity data required for each active ingredient and for major impurities or metabolites typically include the results of the following studies and reflect the need for data on all risks as well as those posed by residues in food:

  • Acute oral, dermal, and inhalation studies;
  • Two-generation reproduction study;
  • Chronic feeding studies on rodents and nonrodents;
  • Oncogenicity studies on mice and rats;
  • Mutagenicity studies on gene mutation, structural chromosomal aberration, and other effects toxic to genetic material;
  • Teratogenicity studies on rats and rabbits;
  • Delayed neuropathy studies on chickens; and
  • Plant and animal metabolism studies.

The agency may grant exemptions from one or more of these requirements when the petitioner can show it is scientifically appropriate to do so.

Oncogenicity and Carcinogenicity

Throughout its work, the committee encountered disparate usages of the terms oncogen, oncogenicity, carcinogen, and carcinogenicity. In conventional scientific terminology, oncogen means a substance capable of producing benign or malignant tumors. The EPA has adopted this definition.12 The term carcinogen is generally reserved for substances capable of producing malignant tumors. The committee will follow these usages in this report.

Confusion can arise when these terms are used in a regulatory context. For example, the FDA apparently interprets the Delaney Clause as prohibiting approval of carcinogens, whereas the EPA apparently treats it as prohibiting oncogens—theoretically a broader interpretation. It is unclear to the committee how significant this difference is in actual practice. It seems likely, though, that there are more oncogenic pesticides than carcinogenic ones; chronic feeding studies will sometimes reveal oncogenicity even when a pesticide's capacity to cause malignant tumors is uncertain. Thus, the EPA's more conservative approach generally expands the universe of pesticides to which the Delaney Clause applies.

The description of a substance that is merely a suspect oncogen or carcinogen also may be confusing. A substance may be characterized as an oncogen or carcinogen even though the evidence on which the statement is based may be incomplete (for example, it consists of results from a single test in one species or sex), weak (for example, a trend was seen in a chronic bioassay but not at a statistically significant level), or otherwise flawed (for example, a statisically significant effect was observed in a study of flawed design or execution). In such cases, the EPA may evaluate the potential human risk of such substances as suspect or possible oncogens or carcinogens. The criteria the EPA uses in judging whether a compound is an oncogen for the purposes of the Delaney Clause constitute a critical regulatory variable. The implications will be considered later in this report.

Finally, the important scientific distinction between substances found to be oncogenic or carcinogenic in animals and those found to have the same effects in humans is often obscured or overlooked. Because of the limits of epidemiological data, regulatory agencies typically rely solely upon animal studies in evaluating the safety of compounds for humans. In the absence of convincing data documenting causality between pesticide exposure and cancer in humans, however, it is inaccurate to refer to such substances as human oncogens or carcinogens. Rather, such chemicals are animal oncogens or carcinogens. (It should be noted, however, that the Delaney Clause does not require proof of carcinogenicity in humans.) In its scheme for categorizing evidence of carcinogenicity, the EPA properly reserves the term "human carcinogen" for substances where available human data are sufficient to support that finding. In other cases, the EPA uses the terms "probable human carcinogen" and "possible human carcinogen," depending on the strength of available animal evidence, short-term in-vitro mutagenicity assays, and any other relevant data.13

Appreciation of these distinctions and possible differences in the approaches of the EPA and the FDA is important to complete understanding of the impact of the Delaney Clause. One set of issues lies within the gray areas of the sciences of toxicology, pathology, and risk assessment. Another set clusters around the regulatory consequences of a given scientific judgment. In recent years, the EPA and the FDA have agreed in their regulatory judgments on chemicals with clear, strong indications of carcinogenic potential and on chemicals with very weak or equivocal evidence of oncogenicity. But chemicals that fall between these extremes are vexing.

In this study, the committee follows the EPA's criteria and terminology. The term oncogen will be used in cases when the EPA would judge the animal evidence sufficient to trigger the Delaney Clause. The committee will attempt to clarify its terms throughout the report and cautions readers to remember that tables and text generally depict only potential human risk, usually under worst-case scenarios.

Tolerance Setting for Non-Oncogenic Pesticides Under Section 408

The core of the typical tolerance-setting process under section 408 is the effort by the EPA to compare the quantity of residues to which humans might be exposed through consumption of pesticide-treated food with the level it judges, based on the available toxicological data, as safe. If the EPA finds that the pesticide (or its expected impurities, metabolites, and degradation products) does not cause a statistically significant increase in the incidence of tumors in animals, it concludes the Delaney Clause is not applicable. Then, the EPA calculates a safe level of exposure, following the conventional analysis of calculating an Acceptable Daily Intake (ADI) for the substance in question.

The first step in calculating an ADI is determining from the battery of toxicity studies the "no observable effect level" (NOEL) for the most sensitive toxic response that is considered to be of potential human health concern. In any study, the NOEL is the highest dose level of pesticide (consumed in the daily diet per unit of body weight) at which no adverse effect was observed. It is the dose level nearest to but less than the lowest dose producing observable indications of toxicity. The study displaying the lowest NOEL is generally selected to establish the ADI. This is calculated by dividing the NOEL by a safety factor (typically 100) to yield the ADI, which is also expressed in milligrams of pesticide per kilogram of body weight per day. (The safety factor of 100, an accepted convention in toxicology, is derived by assuming that [1] humans are 10 times as sensitive as the most sensitive animal tested, and [2] some humans are 10 times as sensitive as the least susceptible human.) Regulatory scientists regard the ADI arrived at in this fashion as a level of dietary exposure that virtually all individuals could consume on a daily basis and even exceed on occasion without experiencing adverse effects.

The next step in evaluating whether a proposed tolerance is toxicologically supportable is calculation of the theoretical maximum residue contribution (TMRC) for each food form in which the pesticide could occur. The sum of the TMRCs for all food forms represents the cumulative TMRC for the pesticide. If the TMRC for a proposed use combined with the TMRC for all other already-approved uses is less than the ADI, the proposed new tolerance that has met other requirements is generally approved. If the TMRC exceeds the ADI, the EPA either denies the tolerance or explores with the petitioner ways to lower the TMRC from the proposed or other uses.

In calculating the TMRC, the EPA seeks to avoid underestimating food consumption and exposure to residues by assuming that (1) each pesticide is used on all harvested crop acres for which a tolerance exists or is proposed and (2) pesticide residues are present at the full tolerance level in every food consumed. Together these assumptions generally exaggerate estimates of dietary exposure to residues. Very few pesticides are used on anywhere near 100 percent of the total acreage of a crop grown in the United States, and measured residues are usually below the tolerance. However, the EPA routinely uses these conservative assumptions to account for gaps in information about actual exposure and uncertainties about health effects.

Although the EPA is empowered by law to consider the benefits of pesticide use in establishing section 408 tolerances, it rarely does so. Residues that pass the foregoing ADI/TMRC analyses are regarded as safe. (Indeed, given the 100-fold safety factor and the conservative assumptions about exposure built into the TMRC, there is thought to be a wide margin of safety.) The petition is then said to be toxicologically supportable and is generally approved without examination of benefits. If the TMRC exceeds a pesticide's ADI, the agency may examine the benefits or, as noted above, explore with registrants possible ways to reduce the TMRC by changing the timing, rate, method, or diversity of crop uses for the given pesticide. Before the pesticide use is registered, however, other important potential environmental effects of the pesticide's use and other routes of human exposure must be evaluated. These include the pesticide's effects on birds, fish, and wildlife; groundwater contamination; and hazards to applicators.

Tolerance Setting for Oncogenic Compounds Under Section 408

The EPA's analysis proceeds somewhat differently if the pesticide is a suspected oncogen. In this case, the agency does not seek to identify a NOEL or calculate an ADI. Its approach is based on quantitative risk assessment models developed specifically to provide upper-bound estimates of human cancer risks, based on animal bioassay data, assuming a lifetime of exposure to the pesticide. On the basis of such risk assessments, the EPA makes a judgment about whether a given tolerance for a specific pesticide use poses an unreasonable risk to humans.

The use of quantitative risk assessment in this context raises questions that go beyond the scope of this report.14 However, a basic understanding of the methodology and limitations of risk assessment is essential to the analysis presented in subsequent chapters.

In brief, risk assessment is a complex extrapolation process. It involves first extrapolating from the effects seen at the generally high doses used in animal studies to the much lower dosages ordinarily consumed by humans in the diet. Then, one predicts from the animal test model results that might occur in humans under actual exposure conditions. The assessment of the oncogenic risk posed by any given substance thus reflects both the potency of the substance and human exposure to it. Once potency is determined, the level of risk to food consumers from a particular pesticide use is a function of exposure to residues in food: the higher the residue levels in foods (or frequency of consumption), the higher the risk.

The risk assessment process is beset by uncertainty and by gaps in knowledge, even on such basic points as the relevance of particular animal test models to humans and the true qualitative and quantitative relationships between effects seen at high doses and those likely to occur at low doses. To compensate for such gaps in knowledge, the EPA and other agencies that use quantitative risk assessment typically adopt conservative assumptions that are designed to avoid understating the potential human risk. For example, results from the most sensitive animal species are used in extrapolating from high doses to low doses, mathematical models are selected that are thought to avoid understating potential human risk, and assumptions are made concerning potential human exposure to the substance that almost certainly overstate true human exposure. Most experts believe that these conservative assumptions together produce risk estimates that represent the likely upper bound of potential human risk. It is generally accepted that the true human risk is probably less than the reported risk estimate.

Risk estimates are typically expressed in terms of the probability that an individual member of a population will experience cancer from exposure to the substance in question over his or her lifetime. Thus, a risk estimate of 1 in 1 million or (1 × 10-6) from a specified level of exposure is a statement that at the 95 percent upper-bound confidence limit, there is no greater than a I in 1 million chance that an exposed individual, or that 1 person out of 1 million exposed individuals, will experience cancer from daily lifetime exposure to the substance in question. To keep these risk estimates in perspective, all individuals now face about a 1 in 4 chance of contracting cancer. Heavy smokers face far worse odds.

In setting tolerances for oncogenic pesticides under section 408, the EPA performs the risk assessment described above and decides whether the risk posed is acceptable—that is, whether the risk is negligible enough to justify a tolerance. To the committee's knowledge, the EPA has not formally adopted any numerical cutoff for oncogenic risks it views as negligible. Without question, however, the EPA has approved many section 408 tolerances for oncogenic pesticides. (See the case studies in Appendix C.)

The committee's review of EPA tolerance actions in recent years suggests that when the estimated upper-bound risk is less than 1 in 1 million (1 × 10-6), the agency rarely disapproves a tolerance. Tolerances likely to pose greater risk than 1 in 10,000 (1 × 10-4) are rarely granted. Decisions to approve tolerances carrying an upper-bound risk between 1 in 1 million and 1 in 10,000 are generally, but not always, made after taking steps to reduce dietary exposure and confirming that risks from other routes of exposure are also small.

In reaching decisions on dietary risks that fall between 1 × 10-4 and 1 × 10-6, the EPA enlarges its inquiry. Under section 408, the agency may consider the benefits of a pesticide's use. The agency generally evaluates benefits in relation to all risks, when data are available, in deciding whether to grant a tolerance for a pesticide for which the upper-bound oncogenic risk estimate falls between 10-4 and 10-6. On occasion the EPA's consideration of benefits has been relatively thorough and its judgment has proved central to the ultimate decision. (See the case studies in Appendix C.) In most cases, however, whether dealing with an oncogen or non-oncogen the EPA rests its tolerance decision on a judgment about the safety of the pesticide residue without regard to benefits.

Tolerance Setting Under Section 409

As noted, section 408 of the FDC Act allows the EPA to consider the benefits of pesticide use and does not forbid approval of oncogenic residues. Section 409, which applies to concentrated residues in processed food, differs in both respects. It does not expressly allow the EPA (or the FDA in its consideration of direct food additives) to consider the benefits of a pesticide's use.15 In the Delaney Clause, it forbids the approval as safe of any food additive shown to ''induce cancer in man or animal."

In the case of non-oncogens, these differences in statutory language have little practical importance. The EPA evaluates the human risk associated with a proposed section 409 tolerance for a non-oncogen using the same ADI/TMRC analysis it follows under section 408. If the TMRC is less than the ADI, the residue is judged safe, and a 409 tolerance is granted.

In regulating oncogens, however, the EPA immediately confronts the Delaney Clause. When a pesticide that requires a section 409 tolerance (because its residues concentrate in some processed foods) has also been found to be oncogenic in animals, the EPA simply declines to grant a tolerance. There is no consideration of whether the residue poses a risk to humans or whether the risk might be judged acceptable. Tolerances in the CFR as of June 1986 are shown in Table 2-1.16

TABLE 2-1. Food Tolerances in the CFR.


Food Tolerances in the CFR.

It is obvious how the fact of concentration can decisively affect the regulatory fate of a pesticide use. If residues of an oncogenic pesticide occur but do not concentrate in processed food, the EPA can estimate risk, make a safety judgment, and then balance risks with the pesticide's benefits. In these cases, a food additive tolerance is not required; the raw agricultural commodity tolerance suffices. If the oncogenic pesticide concentrates in the processed food, the EPA automatically denies a tolerance without further analysis.

Although the impacts of the Delaney Clause on petitions to establish new section 409 tolerances for oncogenic pesticides are clear-cut, its impact on already-established tolerances has been very different. New information demonstrating that a pesticide has oncogenic potential and concentrates in processed food would seem to necessitate the Delaney Clause prohibition. Yet, the EPA has not invoked this provision to revoke a single such tolerance.

As shown in Table 2-2, 31 section 409 tolerances and 2,525 section 408 tolerances exist for pesticides shown to be oncogenic in animal tests.

TABLE 2-2. Food Tolerances in the CFR for 53 Oncogens.


Food Tolerances in the CFR for 53 Oncogens.

The Data Call-In Program

A driving force that will compel the EPA to confront the implications of the Delaney Clause is FIFRA's requirement that all registered pesticides be reregistered on the basis of contemporary scientific standards and data, with priority given to pesticides used on food. To implement this long-standing mandate, the EPA in 1981 instituted the Data Call-In Program, which was designed to elicit the toxicity information needed to make reregistration decisions. The agency also requests a wider range of data in their registration standards program. The toxicity and residue chemistry data generated by these programs in the future will substantially enlarge the number of pesticides that, under current law and policy, seem to require section 409 tolerances and thus could be affected by the Delaney Clause.

By 1990, the agency should have received updated oncogenicity data for nearly all pesticides registered for use on foods. It is impossible to predict which or how many active pesticide ingredients will be found to be oncogenic once all are adequately tested. But, based on past experience (53 out of 289 pesticides used on foods are determined by the EPA to be oncogenic) and assuming continuity in the EPA's interpretation of oncogenicity studies, about 20 percent of the registered pesticides for which data are submitted may be found to be oncogenic.

The EPA determines whether a pesticide residue concentrates in processed food and thus requires a section 409 or food additive tolerance on the basis of residue chemistry data. The EPA currently has complete residue chemistry data on less than 25 percent of pesticides used on foods. The agency is working to complete this segment of its data base although residue data are coming in at a slower pace than toxicity data.

The EPA confronts some difficult issues in this area. EPA scientists acknowledge that the agency's current requirements for food-processing studies do not cover all foods in which residues are likely to concentrate. Concerned food-processing companies have asked the EPA to review the need for evaluating residue concentration in 20 additional crops. The agency must decide how far to pursue metabolites, degradation products, and impurities to determine whether there is concentration. And the EPA is debating how and when to test for concentration in certain dried foods, animal feeds, animal products, and complex mixtures in highly processed foods.

It would be an enormous task to reliably determine whether residues of all pesticides now covered by section 408 tolerances concentrate in processed foods. For those pesticides with scores of raw food tolerances, it would require more time and money to satisfy residue chemistry data requirements than to develop a complete new set of chronic toxicity data. (See Appendix E for a discussion by industrial research directors.) One outcome seems clear. Completion of the data base on pesticide residues in processed foods will substantially enlarge the number of pesticide uses for which section 409 tolerances will be required. The Delaney Clause will halt many of the required tolerances because the pesticides will be found to be oncogenic in animal tests. Residue chemistry data, required to elicit information to assess dietary exposure to pesticide residues in raw and processed foods, will reveal many concentrating residues for which no 409 tolerances are now approved.

The Delaney Clause—A Closer Examination

The foregoing discussion explains the role that the Delaney Clause plays in pesticide tolerance setting and suggests why the provision will loom larger in future EPA decisions. This impending collision between law and practice triggered the current study and independently inspired a closer examination of the history and current interpretation of this noteworthy provision. The FDA, the agency principally responsible for implementing the Delaney Clause, has examined these issues for many years.

In the early 1970s the FDA first suggested the possibility of using quantitative risk assessment to evaluate the safety of substances found oncogenic or carcinogenic in animal studies. The FDA proposed this method of evaluation when it implemented the DES Proviso, which was added to the Delaney Clause in 1962.17 n As part of the Drug Amendments in 1962, Congress provided that the Delaney Clause would not bar approval of carcinogenic drugs and feed additives administered to food-producing animals if upon examination by methods acceptable to FDA no residue of the material could be found in the edible tissues of the animals.18 Convinced that the use of any animal drug would leave some residue, the FDA in 1979 interpreted this proviso to mean no residue above a level posing no significant increased risk of cancer in humans, that is, a level judged by the FDA to be safe. The agency proposed to use quantitative risk assessment to determine the residue level corresponding to an increased lifetime risk of no more than 1 × 10-6 for an individual. It termed this the sensitivity-of-the-method (SOM) approach and continues to use it in regulating residues of carcinogenic animal drugs and feed additives in human food.

Throughout the 1970s, the FDA continued to grapple with developments in science (paralleling those now confronting the EPA) that put great pressure on the traditional understanding of the Delaney Clause. Increased and more sensitive chronic toxicity testing and advances in analytical methods identified many more natural and man-made carcinogens in human food. Many of these were residual reactants, trace constituents of direct food additives, or components of packaging materials that migrated into food in minute amounts.

One source of authority for the exercise of judgment is found in the language of the clause itself. The Delaney Clause applies only if the FDA (or the EPA) finds that a substance "induces cancer when ingested by man or animal." The legislative history makes clear that Congress intended the agencies to exercise sound scientific judgment in deciding whether a substance induces cancer. The clause seems to preclude the agency from ignoring the results of an animal ingestion study solely on the basis of the conclusion that such results are irrevelant to humans. Yet, the agency is surely allowed, perhaps even required, to evaluate whether the study was properly designed and conducted. The statute also leaves open the questions of whether cancer induction must be demonstrated by more than a single well-conducted study or how to weigh conflicting results from two or more studies. In practice, a single, properly conducted, positive test has been adequate, in the EPA's judgment, to trigger a finding of oncogenicity.

The statute is also silent on exactly what "induce cancer" means. Is it sufficient that an additive or pesticide increases the incidence of benign tumors in the test animals? Or must there be convincing—or at least some—evidence of malignancy? How should benign and malignant tumors seen in the same study be interpreted? The FDA has not chosen a rigid position, but generally it has looked for evidence of malignancy before taking action solely on the basis of the Delaney Clause. EPA scientists have been more conservative. In several cases they found that a substance meets the "induce cancer" criterion where there has been no indication of malignancy (see Appendix C).

In addition to the SOM approach, the FDA has used other interpretations of the Delaney Clause intended to limit application. Two of these— the constituents policy and the de minimis interpretation—deserve discussion here.

The constituents policy rests on an interpretation of the phrase "no additive shall be deemed to be safe if it is found to induce cancer . . . ." The FDA now interprets the term "additive" to refer to the added substance as a whole and not to each of its individual constituents. Thus, if a constituent of a food additive or color additive has been found to induce cancer, but the parent additive has not, the FDA will not invoke the Delaney Clause. Instead it will evaluate the safety of the additive under the general safety standard. The carcinogenicity of the trace constituent will be taken into account by performing a quantitative risk assessment. The parent additive will be approved as safe under the general safety standard if the assessed risk from the carcinogenic constituent is insignificant. The FDA's benchmark for judging safety in this context is a lifetime upper-bound risk estimate of 1 × 10-6.

The FDA has applied its constituents policy in evaluating several residual reactants in color additives and some migrating components of packaging materials. One reviewing court upheld this interpretation of the Delaney Clause. A significant example of the application of the constituents policy to pesticide tolerance setting is discussed later in this report.

The FDA's latest and potentially most far-reaching effort to expand its discretion under the Delaney Clause is its de minimis interpretation. The FDA first used this interpretation in evaluating certain color additives that induced cancer in animals. The levels of human exposure to the colors were extremely small. Risks posed were estimated to be extremely low— in some cases orders of magnitude below 1 × 10-6.The FDA announced in June 1985 that it interpreted the Delaney Clause as not prohibiting such extremely small risks. Six months later, the agency used the de minimis concept to approve the use of methylene chloride to decaffeinate coffee, based on the conclusion that the risks posed by permitted residues were no greater than 1 × 10-6.19

The FDA's de minimis interpretation of the Delaney Clause has been more controversial than the agency's constituents policy. The de minimis interpretation recently has been challenged in a petition for judicial review of the FDA's decision to permanently list two color additives.20

Because the de minimis interpretation departs from the FDA's traditional interpretation of the Delaney Clause, the policy's legality will remain uncertain until there is a definitive court ruling.

From a policy standpoint, the FDA's de minimis interpretation is an important development. If upheld, it would replace the zero-risk interpretation of the Delaney Clause with a no-significant-risk standard. For carcinogens, the requirements of the general safety clause and the Delaney Clause would have become congruent. Even if the FDA's statutory interpretation is overruled, its policy judgment that cancer risks of less than 1 × 10-6 may be considered safe when derived by methods using a clearly defined set of conservative assumptions could have important implications for pesticide tolerance setting.

Summary of Problems and Issues Posed By the Delaney Clause

The committee identified four different risk standards in the current law which could apply to residues of an oncogenic pesticide on the food and feed forms of a single crop.

  • For residues on raw agricultural commodities consumed as food, tolerances may reflect a weighing of risks and benefits.
  • When residues concentrate in processed food, the Delaney Clause would bar any tolerances for that crop.
  • For concentrated residues in processed animal feeds such as soybean hulls, tolerances may be denied under the Delaney Clause unless approvable under the SOM approach discussed above.
  • For most hays, fodders, and other nonprocessed livestock feeds, tolerances would be granted under section 408 regardless of whether residues concentrate, based on an assessment of the cancer risks associated with dietary exposure to residues in the animal products ultimately consumed by humans.

This diversity dramatizes the problems presented by the current framework for setting tolerances for pesticide residues on agricultural commodities.


The current system treats pesticide residues inconsistently in two ways. One is exemplified by the dichotomous risk standards in sections 408 and 409. From the outset of its deliberations, the committee has been unable to identify any sound scientific or policy reason for regulating pesticides present in or on raw commodities differently than those present on processed foods. From the standpoint of consumer protection, the source of exposure—raw commodity versus processed food—seems irrelevant.

The other inconsistency is the system's disparate treatment of old and new pesticides. Old pesticides, especially those first registered before 1972, generally were not tested adequately for oncogenicity. They were approved on the basis of limited residue chemistry data, particularly concerning metabolites. Consequently, there was very limited knowledge of the pesticides' capacity to concentrate in processed food. Many of these pesticides are widely used today even though some are suspected oncogens and usually lack section 409 tolerances that would almost surely be required if complete residue chemistry data were available. Pesticides recently registered for use on foods, on the other hand, have generally been tested rigorously. With more complete residue chemistry data, the EPA is more likely to recognize the need for section 409 tolerances which, if a pesticide proves even weakly oncogenic, cannot be granted.

This inconsistency in treatment of old and new pesticides is very important. If the standards applied to new chemicals are justified to protect the public, the same standards should be applied to older pesticides. If older pesticides are judged to not pose a public health problem, then contemporary requirements restricting new, less oncogenic pesticides may be overly protective and may impede introduction of useful new pesticides.

The Issue of Concentration in Processed Foods

Another major problem derives from the current law's emphasis on whether a pesticide residue concentrates in processed food. For an oncogenic pesticide, this fact can prove crucial. If it is found to concentrate, it will be denied a section 409 tolerance under the current system. Consequently, the pesticide will lose the underlying section 408 tolerance and FIFRA registration for that use. If a crop has no recognized processed form (see the boxed article "Subsection O"), then tolerances for an oncogen can be granted if the risks are deemed acceptable. If a crop has a processed form but residues do not concentrate, an oncogen can be granted a tolerance under the general safety clause of section 409. Such differences based on the fact of concentration in certain processed foods make no discernible sense in terms of public health protection.

Paradoxical Regulatory Results

The committee can envision situations in which the current system would compel results that, at least in the short term, actually increase the human cancer risks from pesticides. For example, suppose a registered pesticide X with known oncogenic effects and an existing substitute Y which is a weaker oncogen are under review. Both agents produce roughly equal benefits for comparable uses. X does not concentrate in any processed apple products, but Y concentrates marginally. The EPA could be forced by the Delaney Clause to deny a section 409 tolerance for Y and also would be compelled to cancel its section 408 tolerance and registration. Pesticide X would claim a larger share of the market. Human cancer risk would rise, not fall.

Subsection O

The Delaney Clause prohibits a food or feed additive tolerance for any pesticide that is found to cause cancer in humans or animals when the residues of that pesticide concentrate in a processed food or feed above the level allowed in the raw agricultural commodity. The Delaney Clause does not directly govern residues of these pesticides in raw foods. The clause can have a significant effect on raw food tolerances, however. The EPA has successfully used the Delaney Clause to deny tolerances for an oncogenic pesticide on an entire crop when residues are found to concentrate in the processed portion of that crop.

As a result, the definition of a processed food is critical to the scope and impact of the Delaney Clause. The criteria that the EPA uses to define processed foods and feeds are in a nonregulatory companion to 40 CFR 158, Subpart K, entitled "Pesticide Assessment Guidelines Subdivision O: Residue Chemistry." These criteria are guidelines, not regulations. Yet, they represent the EPA's current thinking on processed versus raw foods and feeds. The EPA is currently reviewing and revising the criteria. When the criteria emerge in final form they could have a significant effect on which crops and pesticides might be most affected by the Delaney Clause. Currently, however, most pesticides lack residue studies on a range of processed foods.

Another example involves a registration application and tolerance petition for a new pesticide with data that show weak indications of oncogenicity. The new pesticide is destined for a crop use for which there are two registered, relatively potent oncogenic pesticides. Registration of the new product is delayed by a prolonged dispute over whether a metabolite causes the potential oncogenicity and whether it concentrates in processed foods. Even though approving the new chemical may reduce dietary cancer risk because it would displace more potent, approved oncogens, the EPA probably would maintain the status quo under current policy. Examples of this scenario can also be found in actions now pending before the EPA.

Another twist of the old standards versus the new creates other paradoxes. Suppose the EPA is deciding whether to cancel an old compound that poses clear oncogenic hazards. The availability of effective registered substitute chemicals is important in estimating the material's benefits and, hence, in balancing its risks and benefits. The agency's inability under current law and policy to register a new, weakly oncogenic substitute chemical when residues concentrate in processed food exaggerates the perceived benefits of the older products. Registrations and tolerances may be denied even though EPA scientists are convinced that the new chemical would pose less risk and provide essentially equal food-production benefits.

The following chapters explore the likelihood that the current system will often produce such paradoxical, indefensible results. Alternative policy constructs are explored that might help the agency more efficiently reduce public health hazards while ensuring an adequate inventory of pesticides.


1. 7 USC § 136(a) (1978).

2. Ibid.

3. 40 CFR Parts 158 and 162 (1986).

4. 21 USC § 346(b) (1984).

5. 21 USC §§ 342(a)(2)(B) and 348(a) (1984).

6. 21 USC § 321(s) (1984).

7. H.R. 2284. 84th Cong., 2d sess. (1958).

8. 21 CFR § 170.3(i) (1986).

9. 21 USC § 348(c)(3)(A) (1984).

10. 21 USC § 342(a)(2)(C) (1984).

11. This is because previously unpublished data are the registrant's confidential proprietary information. The EPA sometimes will press a petitioner to allow public access to its safety data. This occurs when the agency regards the tolerance decision as difficult or potentially controversial, such as when significant safety questions are posed.

12. 40 CFR § 162.3(bb) (1986).

13. U.S. Environmental Protection Agency. 1986. Guidelines for Carcinogenic Risk Assessment. Federal Register 51(185): 33,992-34,003.

14. National Research Council. 1983. Risk Assessment in the Federal Government: Managing the Process. Washington, D.C.: National Academy Press.

15. There is disagreement whether section 409 allows the EPA or the FDA to consider the benefits of individual food additives in deciding whether they are safe. The FDA has consistently taken the view that it does not. In the past, the FDA weighed a pesticide's benefits in deciding whether to approve a 409 tolerance. This interpretation seems difficult to reconcile with the section's language and history. It also conflicts with the view of the FDA, the agency primarily responsible for interpreting and administering this section of the FDC Act. No court squarely holds that the EPA's view is untenable.

16. 21 CFR Parts 193 and 561 (1986); 40 CFR Part 180 (1986).

17. 21 USC § 348(c)(3)(A) (1984).

18. Ibid.

19. In August 1986 the FDA applied the de minimis interpretation to approve two color additives, D&C Orange No. 17 and D&C Red No. 19. This approval inspired the interpretation's development. U.S. Food and Drug Administration. 1986. Federal Register 51(August 7): 28,331-28,346.

20. Ibid. The colors are D&C Orange No. 17 and D&C Red No. 19. The petition for review was filed by Public Citizen in the U.S. Court of Appeals for the District of Columbia Circuit. Public Citizen v. Young, No. 86-1548 (D.C. Cir. filed Oct. 9, 1986) (a decision could be rendered later this year).

Copyright © 1987 by the National Academy of Sciences.
Bookshelf ID: NBK218047


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (3.8M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...