Chapter 1

Introduction and Summary

CONTENTS

PageIntroduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Summary .. .. . $ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

Agency Policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4Required Animal Testing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Guidelines for Testing Protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Risk Assessment Policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6Carcinogen Assessment Policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7Federal Assessment and Regulation of Carcinogens . . . . . . . . . . . . . . . . . . . . . . . . 9Type of Evidence: Human or Animal Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15The National Toxicology Program (NTP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . l6Regulatory Responses to NCI/NTP Test Results and the Annual Report . . . . . 18Future Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

FiguresFigure No. Pagel-1. Elements of Risk Assessment and Risk Management. . . . . . . . . . . . . . . . . . . . . 8l-2. Agency Actions on Annual Report and Positive NCI/NTP Chemicals . . . . . 20

Chapter 1

Introduction and Summary

INTRODUCTION

Over the years, laws have been enacted to pro-tect the health of Americans, with particular em-phasis on protection against cancer. By and large,these laws provide for reducing or eliminating ex-posures to external chemical carcinogens withwhich people come into contact—in the food sup-ply; in drinking water; in pharmaceutical drugsand other consumer products; in work environ-ments; in ambient air, water, and soil. Most casesof cancer, however, are not caused by these typesof carcinogenic exposures.

Instead, according to the best interpretation ofthe evidence currently available, most result from“lifestyle” factors, of which the details are onlyslowly becoming clear. One—tobacco smoking—stands out the clearest of all, and alone is the causeof more than one-third of all deaths from cancereach year in the United States. The more poorlydefined lifestyle factors include such items as over-all dietary balance and aspects of sexual behavior;others, slightly better defined, include exposureto sunlight (see OTA 1981 for a fuller discussionof causes of cancer). In addition to lifestyle fac-tors, viruses are potentially great, but currentlyunquantifiable, contributors to the overall can-cer burden. Nevertheless, those carcinogenicchemicals that can be identified specifically andcan be controlled are important for those very rea-sons: they are avoidable. And often, unlike ciga-rette smoking, exposure to them is involuntary.Furthermore, the potential for introducing new,potent, carcinogens is very real.

For the laws addressing chemical carcinogensto be effective, there must be means of identify-ing substances that have caused, or would cause,human beings to get cancer. Once the substanceshave been identified, regulatory decisions can bemade about whether and how to control expo-sures. Both the process for finding out which sub-stances already in the human environment arecausing cancer in the population (through epi-demiologic studies) and the process for predict-ing carcinogenicity in humans before people areexposed (by testing in the laboratory and in ex-

perimental animals) are imperfect, and interpre-tation of the results of such studies is contentious.While efforts to develop improved methods foridentifying carcinogens continue, current and pastregulatory decisions have, of necessity, embodiedmany untested and some untestable assumptions.

This OTA background paper responds to a re-quest from the House Committee on GovernmentOperations and its Subcommittee on Intergov-ernmental Relations and Human Resources to ex-amine Federal activity in testing chemicals forcarcinogenicity and the use of test results by reg-ulatory agencies.

In this background paper, OTA addresses thefollowing specific questions:

●

●

●

What policies for regulating carcinogens haveFederal agencies adopted? What guidance dothese policies provide about identifying,assessing, and regulating chemical carcino-gens? What kind of evidence, human or ani-mal, do the agencies require to identify achemical qualitatively as carcinogenic? Howdo the agencies intend to conduct quantita-tive risk assessments?What chemicals have actually been regu-lated? What evidence provided the basis forthese regulations? How long does the regu-latory process take?How is Federal carcinogenicity testing orga-nized? How are chemicals chosen for suchtesting? After the chemicals are tested, arethe chemicals that test positive regulated?Have agencies regulated the chemicals listedin the Federal Government’s Annual Reporton Carcinogens?

Chapter 2 of this background paper comparesthe formal Federal policies for identifying andassessing the risks from carcinogenic chemicals.Chapter 3 lists the carcinogenic chemicals thathave been regulated by each Federal regulatoryagency. Federal agencies with the greatest rolesin regulating chemical carcinogens are the Foodand Drug Administration (FDA) for foods, cos-

3

4

metics, and human and animal drugs; the Occupa-tional Safety and Health Administration (OSHA)for worker exposure in most industries; the MineSafety and Health Administration (MSHA) forworker exposure in mines; the Consumer Prod-uct Safety Commission (CPSC) for consumerproducts; and the Environmental ProtectionAgency (EPA). EPA is charged with regulating airpollutants under the Clean Air Act (CAA); waterpollutants under Clean Water Act (CWA); drink-ing water contaminants under the Safe DrinkingWater Act (SDWA); pesticides under the FederalInsecticide, Fungicide, and Rodenticide Act(FIFRA); toxic chemicals under the Toxic Sub-stances Control Act (TCSA); and hazardouswastes under the Resource Conservation and Re-covery Act (RCRA) and the Comprehensive Envi-ronmental Response, Compensation, and Liabil-ity Act (CERCLA).

Chapter 4 describes the National ToxicologyProgram (NTP), the home of the Federal testingprogram, and its carcinogenicity testing. Chap-ter 5 examines the regulatory responses to posi-tive results from Federal carcinogenicity bioassaysand to the chemicals listed in the Annual Report

SUMMARY

Agency Policies

Over the last decade, several Federal agencieshave issued guidelines and policies detailing howthey intend to identify, evaluate, and regulate car-cinogens. These guidelines encompass the designof animal carcinogenicity bioassays, the interpre-tation of data from human and animal studies,and the assumptions that should or will be madewhen assessing human risk from such studies.

The assumptions in these documents representscientific views and policy judgments about car-cinogen assessment. Some assumptions are madebecause, though appropriate data might be ob-tained with current techniques, the data are sim-ply not available in a particular case. Other moregeneral assumptions take the place of experimentalevidence that may be developed with further re-search. Finally, some assumptions are employedbecause of ethical considerations and the inher-

on Carcinogens. Appendix A describes the Fed-eral statutes that have been most important in reg-ulating carcinogenic chemicals.

The scope of this background paper is limitedto “chemicals” that have been tested, listed, or reg-ulated by the Federal Government for carcinoge-nicity. The term “chemical” is used broadly hereto encompass substances, mixtures, groups of sub-stances, and exposures. This background paperdoes not examine the regulation of radiationsources licensed by the Nuclear Regulatory Com-mission, electronic radiation (including, for ex-ample, x-ray machines, which are regulated byFDA), ultraviolet radiation, alcohol, and tobacco.Depending on statutory mandate, Federal regu-latory decisions can be based on such factors ascontrol technologies and costs, in addition torisks. Agency procedures for developing informa-tion on these factors will not be discussed in thisbackground paper. Moreover, while very impor-tant, other related efforts not covered here arethose of industry and the private sector to testchemicals for carcinogenicity and implementvoluntary controls to reduce exposures to car-cinogens.

ent limits of experimental methods. The use ofassumptions, the frequent absence of data, thepotential economic implications of governmentregulation, and underlying political disputes aboutthe desirability of regulation, combine to makethe assessment of carcinogenicity and the devel-opment of corresponding regulations subjects ofintense debates.

It is now common to distinguish between riskassessment and risk management: risk assessmentcharacterizes the adverse health effects of humanexposures to environmental hazards; risk man-agement is the choosing of regulatory options.Both risk assessment and risk management incor-porate policy choices and reflect the values of therisk assessors and managers. Some agencies haveattempted to establish separate staffs for the twotasks, but this separation does not eliminate theneed to make policy choices about the assump-tions used in risk assessments.

5

The values and policy preferences of decision-makers, risk assessors, and representatives of in-dustry, labor unions, environmental organiza-tions, and public interest groups often differ.Scientists disagree about the nature of scientificevidence. These differences explain some of thepast controversies over the regulation of specificcarcinogenic chemicals and the development ofagency policies.

In 1983, a committee of the National ResearchCouncil recommended the development of uni-form guidelines for conducting risk assessments.The committee described several advantages anddisadvantages of such guidelines. They have theadvantages of promoting quality control, con-sistency, predictability, public understanding,administrative efficiency, and improvements inmethods. In addition, guidelines serve an impor-tant role within agencies in training new staff inagency practices. The potential disadvantages ofsuch guidelines include oversimplification, inap-propriate mixing of scientific knowledge with riskassessment policy, misallocation of agency re-sources to the task of developing guidelines, andinsensitivity to scientific developments. Somehave hoped that policies for assessing and regu-lating carcinogens would speed regulatory activ-ity. Others have tried to use such policies tochange the direction of risk assessment and regu-lation.

While much effort has been devoted to devel-oping guidelines and policies for carcinogenassessment and regulation, it is not clear howmuch effect they have actually had. They do pro-vide points of reference for discussions on par-ticular regulatory issues. Nevertheless, while thereare important disagreements among regulatoryagencies, industries, and other groups on generalissues, many disagreements concern interpreta-tions of evidence in particular cases. Everyonemay agree, for example, that animal data can beused to identify potential human carcinogens, yetthey may disagree about the applicability of re-sults from particular animal experiments in assess-ing particular chemicals, especially commerciallyimportant ones. Adoption of general guidelinescannot resolve these specific disputes.

Agency policies and guidelines have varied con-siderably in their flexibility, formality, and com-prehensiveness. They have also evolved, gener-ally becoming more complex and detailed.

This background paper considers two distinct,but related, types of guidelines: agency require-ments for animal carcinogenicity studies; andagency policies on identifying, assessing, and reg-ulating carcinogens.

Required Animal Testing

FDA and EPA have required industry to con-duct carcinogenicity testing of food and color ad-ditives, animal drugs, animals, human drugs, pes-ticides, and toxic substances.

FDA requires carcinogenicity testing for a pro-posed food additive only if it falls into certainchemical categories and its expected concentra-tion in food exceeds specified levels. For an ani-mal drug, testing may be required depending onthe expected extent of its use in animals, the levelsof drug residues, and the potential toxicity of thedrug as determined from chemical structure,short-term tests, and other data. FDA requirescarcinogenicity testing for new human drugs thatare expected to have chronic or widespread use,although this requirement has not been appliedto drugs marketed prior to 1968. For some of theseolder drugs, which are used widely today, FDAhas requested studies from NTP rather than fromdrug manufacturers.

EPA may require animal carcinogenicity studiesof pesticides when they generate some toxicologicconcern, when they will be used on food, or whentheir use will result in significant human exposure.Considerable delays have occurred in requiringtest data on pesticides marketed prior to 1972. Un-der TSCA, EPA may require testing for newchemicals or for existing chemicals.

Guidelines for Testing Protocols

OTA compared the bioassay study designs forsuspected carcinogens that are specified by sev-eral Federal agencies. FDA and EPA have issuedguidelines for the design of toxicologic studies, in-

6

eluding those of carcinogenicity. FDA has reliedon nonregulatory guidelines, such as its “RedBook,” for studies required of new food and coloradditives. For human drugs, a joint workshopsponsored by FDA and the Pharmaceutical Man-ufacturers’ Association (PMA) discussed the de-sign of studies. Although FDA decided not to is-sue guidelines under its own name, the guidelineswere published by PMA. EPA issued as regula-tions separate testing guidelines for pesticides andfor toxic substances. The National Cancer Insti-tute (NCI) and NTP test guidelines were also con-sidered in this OTA comparison.

Federal agency guidelines are generally consist-ent about major features of the study design. Theyspecify testing in two animal species, which inpractice are usually rats and mice. The two old-est guidelines (those of NCI and PMA) require atleast two dose groups in addition to a controlgroup. All other guidelines suggest the use of threedose groups and a control group. The guidelinesagree that to maximize the sensitivity of a studyin detecting carcinogenic effects, the highest dosein the study must be set as high as possible with-out shortening the animals’ lives because of non-carcinogenic toxic effects.

Risk Assessment Policies

OTA also compared Federal agencies’ policieson identifying and assessing carcinogens. Thesepolicies were issued under a variety of circum-stances and are organized in different ways. Insome cases, the policies are relatively informalstatements of current scientific understandingabout how carcinogens might be identified. Inother cases, they constitute formally adopted reg-ulations, specifying how an agency will identifycarcinogens and limiting the kinds of argumentsand evidence to be considered in specific regula-tory proceedings. In between these two extremes,some documents outline an agency’s standard pro-cedures and discuss problematic areas of interpre-tation, including the inference assumptions thatthe agency will use.

Several agency policies have taken a regulatoryform, for example, OSHA’s 1980 policy. OSHAintended to collect evidence and testimony on“generic” issues in carcinogen identification and

regulation, make decisions on these issues, andthen rely on these decisions and presumptions infuture proceedings. The policy might be termeda “presumption-rebuttal” approach, providingstrong presumptions and limited room for rebut-tal. The framers of this policy hoped it would limitdebate in subsequent regulatory proceedings andthereby speed carcinogen regulation. That hopehas not been realized. Two carcinogens with oc-cupational exposures, ethylene oxide and as-bestos, have been regulated since the publicationof OSHA’s policy.

CPSC attempted to adopt carcinogen assess-ment guidelines in 1978. CPSC was sued, and theguidelines were struck down by a reviewing court.Subsequent to this decision, CPSC formally with-drew its policy.

FDA has been working on a regulatory defini-tion of allowable animal drug residues in humanfood since 1973. This definition specifies how sen-sitive an analytic technique must be, hence thedefinition is called “sensitivity of method” (SOM).It was first proposed in 1973, made final in 1977,challenged in court and sent back to FDA, repro-posed in 1979, then proposed for a third time in1985. The final rule has still not been issued.

Other agency policies provide guidelines forconducting risk assessments. EPA’s 1976 “interim”guidelines and its 1986 carcinogen risk assessmentguidelines are examples of this approach, whichdiscusses scientific issues, sets forth flexible as-sumptions, and specifies an analysis based on theweight of the evidence.

A 1979 Interagency Regulatory Liaison Group(IRLG) policy and a 1985 guideline issued by theWhite House Office of Science and TechnologyPolicy (OSTP) both discussed current knowledgeof carcinogenesis and related risk assessment tech-niques. These documents are important becausethey represent the results of extensive discussionsamong scientists from many agencies. One goalof these discussions was to develop a consensusamong the agencies on these issues.

Not all agency programs have adopted policieson risk assessment. For example, FDA does nothave a formal risk assessment policy on food andcolor additives. However, FDA’s Center for Food

7

Safety and Applied Nutrition has established aformal committee for considering evidence on thecarcinogenicity of food and color additives. FDAhas published a policy on regulating additives withcarcinogenic impurities and has developed a pol-icy incorporating a de minimis approach to reg-ulating the safety of food and color additives. Forevaluating the safety of human drugs, FDA re-quires different kinds of tests depending on theexpected duration of human use of the drug, butit has never issued guidelines for evaluating orassessing animal carcinogenicity test results. Itdoes provide guidance, however, on preparingdata for statistical analysis.

Carcinogen Assessment Policies

A National Research Council committee dividedrisk assessment into four distinct parts: hazardidentification, dose-response assessment, exposureassessment, and risk characterization (137). Haz-ard identification is the qualitative identificationof a substance as a human or animal carcinogen.In dose-response assessment, the relationship be-tween the level of exposure or the dose and theincidence of disease is described. The two mostimportant aspects of the second step are extrap-olating from information on incidence at highdoses to predict incidence at lower doses and, inthe case of risk assessments based on animal data,converting animal doses into equivalent humandoses. Exposure assessment estimates the fre-quency, duration, and intensity of human ex-posures to the agent in question. Finally, riskcharacterization relies on information from bothdose-response and exposure assessments to esti-mate the expected risk, as well as to explain thenature of the risk and any uncertainties in assess-ing it.

Figure 1-1 illustrates these steps, which even-tually lead to information useful for risk manage-ment decisions. Each step involves some uncer-tainty, owing either to inadequate data on theparticular agent or to uncertainty about its mech-anisms of toxicity.

Hazard Identification

In many situations of regulatory interest, thereare few toxicity data of any sort. When data are

available, the agencies value epidemiologic studiesas the most conclusive evidence for human car-cinogenicity, presume that substances found tobe carcinogenic in animals in long-term bioassayspresent carcinogenic hazards to humans, and useshort-term test results as supportive information.Analyses of structure-activity relationships (anal-yses based on the structural similarity of a sub-stance to other known carcinogens) are usedmostly when there are no other data (e.g., to iden-tify new chemicals that should have additionaltesting prior to large-scale manufacture).

All Federal policies accept the use of animaldata in predicting human effects. While it is notknown with certainty that all animal carcinogensare also human carcinogens, most well-studied hu-man carcinogens show some evidence of carcino-genicity in animals.

While agencies accept animal data, determin-ing exactly what evidence demonstrates that asubstance is an animal carcinogen is more com-plex. Generally, the agencies accept data derivedfrom use of the maximum tolerated dose, and thenuse the increased incidence of malignant or be-nign tumors to demonstrate carcinogenicity. Pol-icies usually state that positive results in animalsoutweigh negative epidemiologic results, and thatpositive results in one species outweigh negativeresults in another.

Dose-Response Assessment

Prior to 1970, there was considerable doubtabout the utility of quantitative assessments. Dur-ing the 1970s and 1980s, the agencies began usingthese assessments for carcinogens. In 1973, FDAspecified the use of quantitative risk assessmentin the proposed SOM for evaluating animal drugs.In 1978 and 1979, FDA conducted risk assessmentsfor the environmental contaminants aflatoxinsand polychlorinated biphenyls (PCBS). FDA firstused risk assessment to determine the risk of car-cinogenic impurities of color additives in 1982 andof food and color additives themselves in 1985 and1986, The first EPA risk assessment, in 1975, con-cerned vinyl chloride. In 1976, EPA establishedits Carcinogen Assessment Group and publishedits “interim” guidelines on risk assessment. CPSC’sfirst use of risk assessment came with its evalua-

8

Figure 1-1 .—Elements of Risk Assessment and Risk Management

I

I

I

SOURCE: National Academy of Sciences/National Research Council, Risk Assessment in the Federal Government: Managing the Process (Washington, DC: NationalAcademy Press, 1983).

tion of tris(2,3-dibromopropyl)phosphate (Tris)in 1977. While OSHA had first prepared a quanti-tative risk assessment in 1976 for worker exposureto coke oven emissions, it resisted calls for in-creased use of these assessments until the SupremeCourt’s 1980 decision on the benzene standard.Today, although there are still many uncertain-ties associated with quantitative risk assessment,all of these agencies use it.

The agencies all assume that human risk esti-mates can be derived from animal data, that car-cinogenic chemicals do not have no-effects thresh-olds, and that risk estimates should be based onresults from the most sensitive animal species. Allthe agencies use mathematical models that assumelow-dose linearity for extrapolating from the dosestested in the animal experiment to the doses ofregulatory interest, although they differ on themathematical technique to use, whether the fo-cus should be on the “upper confidence limit” orthe “maximum likelihood estimate, ” and the

method of converting animal doses into humandoses. The general approach is to develop riskestimates with assumptions designed to err on theside of safety. The agency policies do not distin-guish among chemicals thought to have differentmechanisms of action (e.g., between “initiators”and “promoters”). The agencies are only begin-ning to explore the use of pharmacokinetic mod-eling techniques, and thus have not discussed thesein detail in their policies.

Exposure Assessment

Agency policies give much less detailed guid-ance on how human exposures to specific chemi-cals should be estimated. While EPA has issuedexposure guidelines, the predominant approachin those guidelines and in actual agency practiceis to make evaluations case by case. The lack ofdetailed guidelines does not diminish the great im-portance of considering exposure in estimating hu-man risk.

9

Risk Characterization

Several policies discuss risk characterization,mentioning alternative ways to describe estimatedrisk and various sources of uncertainty. Some pol-icies also specify a method of classifying carcino-gens, for example, by the weight of evidence forcarcinogenicity. Considering the weight of evi-dence, that is, using all available information ona chemical’s effects, has received more attentionin recent policies.

Federal Assessment and Regulationof Carcinogens

Federal statutes authorize agencies to set ex-posure standards, residue limits, tolerances, andemissions standards for carcinogenic chemicalsfound in air, water, food, and the workplace.Some statutes authorize or require the outrightbanning of carcinogenic substances or productscontaining them; in other cases, agencies may setrules for a product’s use.

Under this authority, a number of carcinogenshave been regulated, although the agencies havenot acted on all of the exposures known to presentcarcinogenic risk. While some time is required toprepare the analyses necessary for regulatory ac-tion and to respond to public comment, there havealso been lengthy delays between knowing theoutcome of human epidemiologic studies or ani-mal bioassays and publishing proposed regula-tions, and delays between the publication of pro-posed and final rules. Regulations on carcinogenshave frequently been challenged in court by in-dustry, labor unions, environmental organiza-tions, or other groups. In some cases the courtshave ruled that the agencies exceeded their author-ity, although in other cases the courts have com-pelled the agencies to act.

Many chemical exposure limits set by the gov-ernment or recommended by private individualsand organizations were established primarily toprotect people from noncarcinogenic toxicities—effects that manifest themselves at the time of ex-posure or shortly thereafter. But cancer is an in-sidious disease. People can be exposed to carcino-gens at levels that do not cause any immediatelyapparent adverse effects. These exposures, how-

ever, can crucially injure individual cells, lead-ing to cancer many years later. Thus, regulatory

standards to protect the public from carcinogenexposures will need to be set at levels much lowerthan those designed to protect against acute tox-icities.

In general, a standard that reduces exposuresbased on concern for one health effect will do sofor all health effects associated with that chemi-cal. But a standard based on noncarcinogenic tox-icities may not reduce exposures sufficiently toprotect against cancer. Significantly, many Fed-eral standards regulating carcinogenic chemicalswere set originally to protect against noncarcino-genic toxicities and have not been updated to takeaccount of carcinogenic effects.

OSHA

Congress passed the Occupational Safety andHealth Act in 1970. In 1971, OSHA adopted alarge number of startup standards, setting ex-posure limits on about 400 specific chemicals.These exposure standards consisted largely of the1968 recommendations of the American Con-ference of Governmental Industrial Hygienists(ACGIH) and had been developed primarily toprotect workers from noncarcinogenic toxicities.While the ACGIH recommendations are updatedannually, OSHA standards are not.

From 1972 to 1986, OSHA issued health stand-ards covering 22 carcinogens, many of which hadbeen regulated by the 1971 standards. Most ofthese carcinogen standards have aroused con-troversy. Of 9 final actions on carcinogens regu-lated individually (including 2 on asbestos), 7 re-sulted in court challenges. In OSHA’s regulationof a group of 14 carcinogens, the final standardsfor 2 chemicals were challenged. Permanent stand-ards for 2 chemicals were struck down as a resultof such challenges.

National Institute for OccupationalSafety and Health (NIOSH)

One role of NIOSH is to identify substancesthat pose potential health problems and recom-mend exposure levels to OSHA. However, OSHAhas not responded to many NIOSH recommen-

10

dations. Since 1971, NIOSH recommendationshave addressed 71 different chemicals or processesthat they determined to be carcinogenic. OSHAhas issued health standards for 21 of the 71 chem-icals or processes. Two of these OSHA standardswere struck down by the courts. Thus, 19 of the71 NIOSH recommendations on carcinogens haveactually been addressed by OSHA regulations. Ofthe 50 chemicals or processes that are not the sub-jects of a final OSHA standard based on car-cinogenicity, many are still regulated under the1971 startup standards. OSHA has proposed reg-ulations for four, but is actively working on a fi-nal standard for only one. No OSHA proposalshave been issued for the remaining 46 chemicalsor processes.

OSHA has criticized the quality of early NIOSHcriteria documents, yet OSHA’s failure to respondwith standards highlights OSHA’s regulatory dif-ficulties. Increasingly, OSHA’s regulatory agendais being set by outside groups, in the form of pe-titions, court orders, congressional directives, andEPA referrals, including those on seven chemicalcarcinogens that EPA formally or informally re-ferred under TSCA. OSHA has proposed a stand-ard for one of these referred substances.

MSHA

MSHA regulation covers coal mines and metaland nonmetal mines. Regulation of toxic ex-posures in mines consists largely of reference tothe 1972 and 1973 recommendations of ACGIH,depending on the type of mine. The ACGIH rec-ommendations are updated annually, whileMSHA has changed few of its standards.

In the late 1970s, MSHA regulated asbestos ex-posures for surface mines (using the exposure limitOSHA issued in 1972) and the chemicals OSHAincluded in its “14-carcinogens standard. ” MSHAhas also proposed revised standards for under-ground exposure to radon daughters.

OSHA set a stricter standard for asbestos in1986, but MSHA has not followed suit. More-over, MSHA’s current asbestos standard does notapply to exposures in underground coal mines.The increased use of diesel engines in undergroundcoal mines has exposed workers to fumes. WhileMSHA has standards for such exposures in metal

and nonmetal mines, these standards were notbased on carcinogenicity. MSHA is developinga proposed standard for diesel exposures in coalmines.

FDA Actions on Food and Color Additives

Since congressional enactment in 1958 of theDelaney clause, which prohibits the use of foodadditives determined to cause cancer, FDA hasidentified over 60 relevant carcinogenic chemicals.They include direct food additives, indirect foodadditives (chemicals that might migrate frompackaging material or manufacturing processesinto foods or beverages), color additives, cosmeticingredients, contaminants or potential contami-nants of food or color additives, and environ-mental or unavoidable contaminants of food.

The regulation of food additives received muchpublic attention when FDA banned cyclamatesand proposed to ban saccharin. FDA has actu-ally banned seven direct food additives. Its pro-posed ban of saccharin was barred by congres-sional action.

The review of provisionally approved color ad-ditives, begun in 1962 under the Color AdditiveAmendments of 1960, has been lengthy. It hastaken until now to obtain required toxicity dataand make regulatory decisions about many of thesubstances on the list. FDA has banned a totalof 10 color additives, while a number of othercolor additives were withdrawn from the marketby their sponsors who sometimes chose not toconduct the FDA-required testing.

In the last few years FDA policy on regulatingfood and color additives has also changed. Priorto 1982, FDA banned several color additives be-cause they were shown to be carcinogenic or con-taminated with a carcinogen. Since 1982, FDA haspermanently listed several color additives eventhough they contain known carcinogens. The newpolicy states that, if a color additive itself doesnot cause cancer in humans or animals, but a con-taminant of the additive does, FDA will regulatethis color additive based on the general safety pro-visions of the act. Under this policy, the carcino-genic impurities are not considered to trigger therequirements of the Delaney clause. FDA will esti-mate potential risk using quantitative risk assess-

11

ment techniques and if the risk of the impuritiesis estimated to be low, FDA will permit the useof the color additive.

In 1985 and 1986, FDA took action to allowuse of food and color additives that were them-selves carcinogenic, basing its action on quantita-tive risk assessment. In 1985, FDA proposed toallow the continued use of methylene chloride fordecaffeinating coffee by limiting the allowable res-idue, rather than to ban the chemical’s use en-tirely. Several color additives were identified byFDA as carcinogenic in 1982 and 1983 based onthe results of animal bioassays. After performingrisk assessment calculations, FDA announced in1986 that it was permanently listing these addi-tives because their estimated carcinogenic riskswere low. FDA believes such actions are legallypermissible under the interpretation that the Food,Drug, and Cosmetic Act allows FDA to ignorede minimis risks, despite the seemingly absolutelanguage of the Delaney clause. In February 1987,FDA argued further that because the estimatedrisk in humans was low, the color additives inquestion would not be considered, for purposesof the Delaney clause, to be animal carcinogenseither.

Indirect food additives are generally packag-ing material—various plastics and adhesives usedto hold foods and liquids—and materials that con-taminate foods in the manufacturing process. FDAhas banned two indirect food additives. Other in-direct additives containing carcinogenic impuri-ties have been regulated by prescribing conditionsfor “safe use. ”

In the mid-1970s, FDA prohibited the use ofbottles made from polymers of acrylonitrile andvinyl chloride, because these chemicals mightleach into liquids. FDA’s position was rejected bythe courts. In the 1980s, FDA issued a rule to al-low acrylonitrile copolymer bottles and proposedto allow polyvinyl chloride bottles, arguing thatnew manufacturing technology can ensure mini-mal leaching from these bottles.

FDA can set regulatory tolerances or actionlevels for environmental or unavoidable contami-nants. It has set tolerances for PCB contamina-tion of fish and action levels for aflatoxins,

dimethylnitrosamines (in malt beverages), and N-nitrosamines (in baby bottle nipples).

FDA Actions on Animal Drugs

FDA has identified 14 chemicals associated withanimal drugs that might leave carcinogenicresidues in animal tissues. Such residues had beensubject to the Delaney clause, but in 1962 Con-gress amended the Food, Drug, and Cosmetics Actto permit the use of carcinogenic drugs in animals,providing carcinogenic residues cannot be de-tected in meat or milk using FDA-approved meth-ods. FDA has banned diethylstilbestrol (DES)from use in animals and has required residuestudies on six other substances. FDA has proposedto withdraw approval for seven. One animal drugwas withdrawn by the sponsor and there is noreported action for several others. As mentionedabove, FDA has been working for 14 years onregulatory guidelines specifying the SOM for de-termining the presence of harmful animal drugresidues.

FDA Actions on Human Drugs

In regulating carcinogens in human drugs, FDAhas issued rules on six substances or groups of sub-stances. Two were removed from the market, onewas voluntarily recalled, and cautionary labelingwas required on three. When a drug is determinedto be carcinogenic, the drug’s labeling for physi-cians is usually updated informally. Many, butnot all, carcinogenic drugs on the market are, infact, anticancer drugs. Treatment in these casesinvolves balancing the risk of future cancer againstthe benefit of treating a diagnosed cancer today.

CPSC

Since its creation in 1970, CPSC has evaluatedand attempted to regulate or begun to regulateeight chemicals (or groups of chemicals) for car-cinogenicity. CPSC regulations have often beenoverruled by the courts. although in the case ofTris-treated children’s pajamas, CPSC developedan alternative strategy to remove the productfrom the market. In 1981, CPSC issued a rule reg-ulating hazardous urea-formaldehyde foam insu-lation (UFFI), a rule that was also struck downby the courts.

12

In other cases, use of chemicals in consumerproducts stopped, even though regulation was notfinal or had been overturned in court. In somecases, CPSC has been able to negotiate voluntaryactions by manufacturers, such as the 1979 volun-tary recall of hairdryers containing asbestosshields.

EPA Actions Under the Clean Air Act

Since the 1970 enactment of the Clean Air Act,EPA has, often under legal pressure, listed sevencarcinogens and issued hazardous air pollutantemission standards on six, although one of theseactions was based on noncarcinogenic toxicity.

Although the Clean Air Act provides EPA oneyear to issue regulations after a substance is listed,this deadline was met only in the case of vinylchloride. EPA has taken an average of almost 4%years from the date of listing to final action forthe six carcinogens on which it has issued finalrules. During the time between the listing and reg-ulation of benzene, one major industrial sourceof benzene had changed its process and eliminatedrelease of the chemical.

EPA has created a new type of action in addi-tion to listing: an “intent to list” decision. Accord-ing to EPA, the intent to list a substance as a haz-ardous pollutant does not legally bind the agencyas does a listing decision. EPA has indicated theintent to list for 10 substances, but none as yethas been listed and therefore none regulated.

EPA Actions Under the Clean Water Act

Important amendments to the Clean Water Actwere enacted in 1972, 1977, 1981, and 1987. From1972 to 1975, EPA issued toxic effluent standardsfor six categories of pollutants, under court or-der. In a consent decree, EPA agreed to regulatetoxic pollutants by industry and by specifying thetechnology to be used. EPA agreed to issue ef-fluent limitations for 65 categories of toxic sub-stances, including 29 judged to be carcinogenicaccording to the water quality criteria documentsthat were also developed under this decree.

EPA has focused on 126 chemicals within these65 classes of pollutants, but not all of these chem-icals are regulated for every industry. In addition,

EPA has not established effluent limitations fortoxic pollutants from the organic chemicals indus-try, and current regulation of the pesticides in-dustry does not limit the discharges of most toxicpollutants in that industry. EPA had issued newregulations for the pesticides industry, but theywere challenged in court and are now being recon-sidered by EPA. Again, this regulatory activityhas taken considerable time (from the 1976 con-sent decree until today), has involved the courtson a number of occasions, and is not yet finished.Further, while the list of 126 chemicals was cho-sen based on known toxicity and probable pres-ence in water, and represented the best efforts ofthe participants at the time, more recent data re-veal that many of the chemicals most commonlyfound in industrial discharges are not on this list.

EPA has also prepared nonbinding water qual-ity criteria documents for States to use in devel-oping water quality standards and requirementsfor specific discharge permits. However, only 7of the 29 water quality criteria set for carcinogenshave been adopted by one or more States. Foronly one of these substances (arsenic) have morethan one-fourth of the States issued a water qual-ity standard, although in some States that havenot taken legislative action, individual dischargepermits impose limitations based on the waterquality criteria.

EPA Actions Under the SafeDrinking Water Act

In 1975, EPA issued the “interim” drinkingwater standards still used today for several inor-ganic and organic chemicals and for microbialcontaminants. These standards were based on the1962 recommendations of the U.S. Public HealthService for noncarcinogenic toxicities. EPA alsoissued regulations for radionuclides in 1976 andfor total trihalomethanes in 1979, two groups ofsubstances presenting carcinogenic hazards.

Following the congressionally mandated reportson drinking water by the National Academy ofSciences (the first of six volumes was published in1977), EPA was required to publish proposed rec-ommended maximum contaminant levels (RMCLs)and then to issue maximum contaminant levels(MCLs) for particular chemicals found in drink-

13.

ing water. The MCLs are to be set as close to theRMCLs as is feasible. After considering a 1978proposed regulation to set generic standards fortreating surface water supplies, EPA decided tocontinue focusing on individual substances.

In 1982 and 1983, EPA published two AdvancedNotices of Proposed Rule-making (ANPRMs) list-ing 83 chemicals of concern. In 1983 and 1985,it proposed RMCLs for inorganic substances,volatile organic compounds, and synthetic organiccompounds. EPA issued final RMCLs for eightvolatile organic compounds in November 1985.It has not yet issued final RMCLs for the inor-ganic substances and the synthetic organic com-pounds, and has not proposed RMCLs for radio-nuclides. To date, EPA has issued final MCLs fornine chemicals, five of which are judged to havesufficient evidence for carcinogenicity, and oneto have limited evidence.

Congress was concerned that drinking waterstandards were not being set quickly enough, soin the 1986 reauthorization of the act, it set dead-lines for EPA to regulate the 83 chemicals that hadbeen identified as candidates for regulation in 1982and 1983. These 83 substances include 51 in theprocess of being regulated. In addition, 52 healthadvisories have been issued by EPA. Many ofthese provide information on potential carcino-gens in drinking water.

EPA Actions Under FIFRA

To prevent unreasonable adverse effects onhealth and the environment, FIFRA authorizesEPA to screen pesticides before they enter the mar-ket and to regulate through reregistration the pes-ticides that were already on the market in 1972.In both cases, EPA may require manufacturers toconduct toxicity tests, including long-term bio-assays for carcinogenicity.

FIFRA was substantially rewritten in 1972. Atthat time there were about 50,000 pesticide prod-ucts and 600 active ingredients previously regis-tered by the Federal Government that neededreregistration under the new law. The reregistra-tion process has taken longer than originally an-ticipated. It was to have been completed by 1976,but in 1975 Congress extended the deadline to1977, and in 1978 Congress dropped the deadline

completely because of the large number of sub-stances not yet reregistered. This task will OCCUPYEPA for many years.

For a number of active ingredients subject toreregistration, EPA has lacked sufficient informa-tion to judge their carcinogenic effects. EPA is tak-ing steps to obtain this information. Still, as ofMarch 31, 1986, it had identified at least 81 car-cinogenic active pesticide ingredients. Of these,18 have been canceled or restricted, Daminozide(Alar) is still undergoing review, and 15 have beenvoluntarily canceled. However, cancellationsoften cover only some uses. Other uses of the pes-ticide continue, although EPA may set additionalrequirements, for example, requiring workers towear protective clothing. Special Reviews (SRs)for the substances EPA canceled or restricted re-quired from 13 to 88 months, taking an averageof about 44 months.

Another 18 chemicals have also been subjectsof SRs. The SRs have been completed for 10 car-cinogens, and these chemicals have not been can-celed based on EPA judgments weighing risks andbenefits. For the remaining 8 chemicals, SRs arenot yet complete. Finally, EPA has identified 29carcinogens, but has not started SR or cancella-tion proceedings for any of these.

Thus, EPA has identified 47 carcinogenic ac-tive pesticide ingredients that have not been can-celed. For 13 of these EPA has determined thatlow exposure, low risk, or the weight of evidencefor carcinogenicity suggest no action need betaken.

In addition to considering active ingredients,EPA has indicated that about 55 inert ingredientsare of “high concern, ” with 28 of these showingcarcinogenic effects. In 1987, EPA announced forthe first time that it was taking steps to addresssome of the hazards of these ingredients.

EPA Actions Under TSCA

EPA actions under TSCA cover both new andexisting chemicals. For new chemicals, the prin-cipal focus is the premanufacture review process.If after review of the manufacturer’s premanufac-ture notice (PMN), EPA decides that there is causefor concern, it can request or require that addi-

14

tional toxicity testing be done, that certain con-trols be used when working with the chemical,and that the manufacturer notify EPA before be-ginning a significant new use of the chemical.

From mid-1979, when the PMN program be-gan, until September 1986, EPA received 7,356valid PMNs. Of these, 80 percent or 5,671 re-quired no further action, according to EPA. Ofthe remaining chemicals, 523 were subject to somekind of action; an unknown number of theseraised concerns about carcinogenicity. About halfthe time, EPA attention led to the manufacturer’sinformally and voluntarily agreeing to testing,control actions, or withdrawal of the PMN. Forthe remaining cases, EPA took more formal ac-tion, although often with the manufacturer’sconsent.

The lack of information in the PMNs is a po-tential problem. In 1983, OTA found that abouthalf the submitted PMNs reported no toxicity in-formation and “only 17 percent of PMNs haveany test information about the likelihood of thesubstance’s causing cancer, birth defects or mu-tations. ” Because many PMNs do not provide anytoxicity test data, EPA uses information on chem-ical structure-activity relationships to attempt topredict the hazards that a substance may present.

For existing chemicals, EPA can require toxic-ity and environmental effects testing, designatethe chemical for accelerated review, or requiremanufacturers to report on production and uses,provide EPA with any studies they have con-ducted, or report significant new uses. EPA canalso issue regulations restricting or banning theproduction of a chemical or limiting its uses.

TSCA established an Interagency Testing Com-mittee (ITC) to make recommendations on neededtesting for toxicity and environmental effects. Inthe early years of the program, EPA’s responsesto the ITC recommendations provoked concern,both because of EPA delays in deciding whetherto test and because of the particular administra-tive arrangements chosen for obtaining test data.In addition to the ITC recommendations, EPAcould select other chemicals for testing. So far,this has not occurred often, although this may bechanging.

A rule issued under section 8(a) of TSCA re-quires manufacturers to provide informationabout the production and uses of a chemical,while a rule adopted under section 8(d) requiresthat manufacturers submit to EPA unpublishedhealth and safety studies. EPA has issued 8(a) and8(d) rules for all the substances recommended byITC, but until recently for few additional chemi-cals. EPA has recently received data from manu-facturers as part of its effort to update its inven-tory on all chemicals in commerce.

Sufficient toxicity information is available onsome existing chemicals to show they are carcino-genic. For these chemicals, the issues are deter-mining whether the risks of cancer are “un-reasonable” and what actions may be needed toreduce or eliminate such risks. EPA’s Office ofToxic Substances, which is in charge of the TSCAprogram, has identified 38 chemicals or chemi-cal classes as carcinogenic and has prepared riskassessments for 21 of these.

But beyond the development of risk assessmentsand the gathering of other information, regulatoryactions on existing chemicals have been limited.Four chemicals have been designated for an ac-celerated review under section 4(f) (4,4’ -methyl-enedianiline, 1,3-butadiene, formaldehyde, meth-ylene chloride). Consideration of the regulationof occupational exposures to these chemicals hasbeen referred formally or informally to OSHAsince TSCA provides for referrals if EPA believesanother agency may be able to address a hazard.Under TSCA authority, EPA began proposingSignificant New Use Rules (SNURs) for existingchemicals considered to be carcinogenic. How-ever, actions on carcinogens began in 1984, nearly7 years after TSCA’s enactment. For carcinogenicchemicals, EPA has now proposed six SNURs oneight existing chemicals and has issued four.

Section 6 of TSCA provides wide-ranging au-thority to limit production and uses of chemicals,including the authority to ban a substance. EPAhas proposed section 6 action on PCBs, asbestos,chlorofluorocarbons, and metalworking fluids.PCBS were banned by Congress in TSCA itself;EPA regulations cover implementing that ban andarranging for disposal of PCBs. EPA has alsobanned propellant uses of chlorofluorocarbons,

15

but it has not yet taken action on the most im-portant uses of this group of chemicals, which areused in refrigeration and air-conditioning. Finally,EPA has issued rules on identification of asbestosin schools and proposed rules to require removalin certain cases. EPA has also regulated asbestosexposures for certain workers not covered by theOSHA asbestos standard, although it has nottaken final action on a major proposal to limitand eventually ban asbestos use. The proposal onmetalworking fluids is also not yet final.

EPA Actions Under RCRA

RCRA regulates the generators, transporters,storers, and disposers of hazardous wastes. EPA’slists of hazardous wastes cover 361 commercialchemicals and 85 industrial waste processes. Whenpossible, EPA has emphasized waste streams fromcommercial processes rather than specific hazard-ous substances, to relieve waste generators of test-ing burdens and uncertainties in “relating a wastecontaining many substances to a list of specificsubstances. ” EPA has also issued a list of toxicchemicals as Appendix VIII of its RCRA stand-ards. Wastes containing chemicals on this list maybe deemed hazardous wastes.

EPA has made limited changes in its list ofRCRA hazardous wastes. For example, since 1980EPA has added five wastes to the RCRA list. Inthe 1984 RCRA amendments Congress employed“hammers’’ -congressionally enacted prohibitionsagainst disposal of certain groups of chemicals un-less EPA has acted to specify treatment techniquesfor those wastes. In addition, Congress mandatedthat EPA review, over a 3-year period, the entireRCRA list of hazardous wastes.

EPA Actions Under CERCLA

Commonly known as Superfund, CERCLA wasenacted in 1980. CERCLA requires EPA to iden-tify reportable quantities for hazardous substancesand set requirements for notification of environ-mental releases.

Congress specifically included in the definitionof hazardous substances those chemicals alreadyregulated under several environmental statutes.In addition, Congress set reportable quantities forthese substances at 1 pound (except for reporta-

ble quantities specified under the Clean WaterAct) until EPA could set more appropriate report-able quantities. In May 1983, EPA published itsinitial list of hazardous substances. Since 1983,19 substances have been added to the CERCLAlist yielding a total of 717 substances. Most of theregulatory activity on the CERCLA list has beenin modifying the reportable quantities. In 1987,EPA proposed modified reportable quantities forCERCLA carcinogens. Of the CERCLA hazard-ous substances, 191 have been identified by EPAas “potential carcinogens” or as substances “hav-ing carcinogenic potential. ”

EPA’s Carcinogen Assessment Group (CAG)

As mentioned above, CAG was established in1976 to centralize the conduct of carcinogen riskassessments at EPA. Major CAG assessments arethorough reviews of the carcinogenic risks of par-ticular chemicals, including both qualitativeevaluation of the weight of evidence for carcinoge-nicity and quantitative dose-response estimates.To date, CAG has prepared full assessments on57 chemicals.

Office of Management and Budget (OMB)

Although not a regulatory agency, OMB hasbecome an important actor in developing Federalregulations through their review of proposed reg-ulations under Executive order 12291 and thePaperwork Reduction Act. This review has ledto delays in proposing and issuing standards oncarcinogens. OMB has also publicly questionedsome of the regulatory agencies’ assumptions inconducting risk assessments. The methods OMBused in commenting on a proposed OSHA for-maldehyde standard ran counter to some of theassumptions typically used by the regulatoryagencies and incorporated in agency policies onidentifying and assessing carcinogens.

Type of Evidence: Human orAnimal Data

Agencies use the hazard data available at thetime of their action, most generally, data fromhuman or animal studies. OTA has attempted tocharacterize the evidence that agencies have usedin regulating carcinogens.

16

FDA has relied mostly on animal evidence inevaluating food additives, color additives, humandrugs, and animal drugs.

CPSC has used both human and animal evi-dence, although in its action on Tris and at-tempted regulation of formaldehyde, it reliedupon animal evidence only.

Of the 57 chemicals covered by CAG healthassessments, 40 have been assessed based on“sufficient” animal evidence. Nine more were sup-ported by sufficient human evidence and all butone of these were also supported by sufficient ani-mal evidence. EPA judged the remaining 8 chem-icals to have inadequate human evidence andlimited animal evidence.

Most cancellations and restrictions of pesticideshave been based on the results of carcinogenicitytests in at least two animal species. Nearly allTSCA hazard identifications and risk assessmentsare based on animal data.

There is some evidence of the carcinogenicityof the 35 chemicals proposed for regulation un-der the Safe Drinking Water Act, but EPA believesthat the evidence for the carcinogenicity of 8 ofthese in drinking water has not been establishedand thus is basing RMCLs for these chemicals onnoncarcinogenic effects. EPA’s classification of theother 27 drinking water contaminants as carcino-gens relied mostly on animal evidence.

The original RCRA list of hazardous wastes andCERCLA list of hazardous waste reportable quan-tities were developed largely without specific con-cern for carcinogenicity, although the original reg-ulations on which these lists were based may havehad this concern. Recently proposed adjustmentsin the CERCLA list of reportable quantities clas-sify 191 chemicals as potential carcinogens: 14based on sufficient human evidence, 110 on suffi-cient animal evidence, and 20 on limited animalevidence. Most (40) of the remaining chemicalswere classified based on a parent element (e.g.,inorganic compounds of arsenic were classifiedbased on the carcinogenicity of arsenic), althoughfor 7 chemicals EPA had no evidence of carcinoge-nicity.

OSHA and EPA’s Clean Air Program havebased regulation on human data most of the time,though there are indications this may be changing.

Of OSHA’s eight regulations on individual car-cinogens, seven were based on at least some evi-dence of human carcinogenicity. The other car-cinogen, 1,2-dibromo-3-chloropropane (DBCP),was regulated primarily because it caused infer-tility in men. The evidence of its carcinogenicityconsists of animal data. Regulation of three car-cinogens under the “14-carcinogen standard” wasbased on human evidence, that of nine on animalevidence. The remaining two substances were reg-ulated because of their chemical relationship toother carcinogens. Most OSHA regulations of car-cinogens based only on animal evidence occurredwith the regulation of the 14 carcinogens in 1974.Standards since then have been based mostly onhuman data, although for OSHA’s 1984 regula-tion of ethylene oxide the primary evidence forits carcinogenicity is animal data. The primaryevidence for several chemicals now being consid-ered for regulation, including formaldehyde andmethylene chloride, is animal evidence.

For the five substances regulated primarily ascarcinogens under the Clean Air Act, EPA has re-lied on human evidence of carcinogenicity. EPA’sintent-to-list decisions for eight of ten substanceshave relied on animal bioassays for evidence ofcarcinogenicity; the other two substances showboth animal and human evidence of carcinoge-nicity.

The National Toxicology Program(NTP)

Since 1961, the Federal Government has beendeveloping a testing program for determining thecarcinogenicity of chemicals, first at NCI, andsince 1978, at NTP. The program encompasseslong-term animal studies and other tests to deter-mine carcinogenic activity. NTP is probably thelargest such testing program in the world, and isthus important in advancing knowledge of car-cinogenic chemicals.

17

Early testing at NCI focused primarily on un-derstanding the etiology and biological mecha-nisms of cancer. In the late 1960s, the FederalGovernment expanded carcinogenicity testing.Today, NTP bioassays and other tests provide im-portant information for developing risk assess-ments and issuing regulations.

NTP was created to coordinate the toxicity test-ing of the then Department of Health, Education,and Welfare and to provide a mechanism for reg-ulatory agencies (and others) to request bioassayson chemicals of regulatory interest. The NTP bud-get consists of contributions from several differ-ent agencies in the Department of Health andHuman Services (FDA/National Center for Tox-icological Research (NCTR), CDC/NIOSH, andthe National Institute of Environmental HealthSciences (NIEHS)), although the lion’s share offunds derive from NIEHS. The Director of NIEHSis also the Director of NTP. Activities of the con-tributing agencies are coordinated by the NTPSteering Committee, which consists of the headsof these agencies and the NTP Director. Formalauthority to approve and monitor the general planof NTP activities is vested in an Executive Com-mittee that consists of the heads of the four ma-jor health and environmental regulatory agencies(CPSC, EPA, FDA, and OSHA), the heads of fourresearch agencies (National Institutes of Health(NIH), NCI, NIEHS, and NIOSH) and the Assis-tant Secretary for Health of the Department ofHealth and Human Services (DHHS). This struc-ture allows both the regulatory agencies and re-search agencies a voice in planning and operat-ing NTP.

The nomination of chemicals for NTP testingis invited from any source, including the regula-tory and research agencies. NTP’s established pro-cedures to evaluate nominations include reviewby the interagency Chemical Evaluation Commit-tee, solicitation of public comments, review byNTP’s Board of Scientific Counselors, and finaldecision by the NTP Executive Committee.

After selection, a protocol is prepared and test-ing begins. Testing consists of various preliminarystudies, a long-term dosing regimen (which by it-

self takes 2 years), sacrifice, and pathologic ex-amination, including microscope studies of tissuesand tumor diagnoses. NTP has established pro-cedures for ensuring the quality of these diagno-ses, which are crucial to determining the final bi-oassay results. The resulting data are analyzed andthe draft technical report is submitted to a peerreview committee. Peer reviewers have the train-ing and experience appropriate to judge the qual-ity of the bioassay and to interpret bioassay re-sults. NTP has chosen to include on its peer reviewcommittees people of different perspectives, in-cluding academics and representatives of indus-try, environmental organizations, and laborunions.

The number of chemicals tested depends pri-marily on the resources available. The NTP bud-get increased approximately 40 percent between1979 and 1981. From fiscal year 1981 to 1987 thetotal NTP budget rose from $70.5 to $77.9 mil-lion, which, after adjustment for inflation, rep-resents a small decline. Budget reductions neces-sitated by the Gramm-Rudman-Hollings Act haveaffected NTP. Recently, NCTR discontinued long-term NTP animal tests on one antihistamine andcontinued two other tests only because NIEHSagreed to pay 75 percent of the costs to completethe 2-year exposure phase. NTP has now agreedto fund completion of these two studies. Givencurrent resources, more chemicals are nominatedthan can be tested.

The entire process—nomination, selection, pre-liminary testing, chronic testing, necropsy, dataanalysis, review, and publication—is a long one.OTA examined the process for a group of chem-icals reviewed by NTP’s Chemical EvaluationCommittee in fiscal year 1981 and 1982. None ofthese chemicals has passed through the entire test-ing process. Of the 30 chemicals approved for test-ing in those 2 years, 4 have reached the stage ofchronic testing.

The time from nomination to selection is morethan 2 years for most chemicals. Some shorten-ing of this period should be possible. But muchof the remaining time required (between selectionand beginning chronic exposures) is difficult to

18

shorten because it is used to develop informationimportant for the design, conduct, and interpre-tation of the bioassay.

The nomination process raises at least two is-sues. First, nominations and selections are impor-tant because they may set the regulatory agendafor the following decade. Today, several agenciesare working on regulations for such chemicals asmethylene chloride, 1,3-butadiene, 4,4’-methylenedianiline, and benzene, which NTP tests showedto be carcinogenic. These test results and the re-sulting regulatory action proceed in part fromselection decisions of a number of years ago.

Second, NTP’s recent decisions on testing thebenzodiazepines (which include Valium andLibrium) raise the issue of who should pay for car-cinogenicity testing—government manufacturers,drug sponsors, pesticide registrants, or others.There are advantages to testing through commonprotocols and in the Federal Government’s pro-gram. There is also reason to argue that the man-ufacturers and sponsors of chemicals have aresponsibility to pay for the toxicity tests of theirproducts.

Regulatory Responses to NCI/NTPTest Results and the Annual Report

NCI/NTP Bioassay Results

As of June 1987, the NCI/NTP bioassay pro-gram has completed testing of 308 chemicals ina total of 327 studies. Chemicals are typicallytested in both sexes of rats and mice, for a totalof four “experiments. ” At the end of the study,the results of each experiment are classified asclear evidence, some evidence, equivocal evi-dence, or no evidence for carcinogenicity, or asan inadequate test.

OTA has analyzed the regulatory uses of theNCI and NTP test results subject to peer reviewand audit approval by September 1986. These re-sults represent 284 chemicals studied in 295 tests.For the analysis, “clear evidence” and “some evi-dence” for carcinogenicity were grouped as “posi-tive” results. The chemicals tested were groupedbased on the number of the four experiments foreach that showed positive results. Of the 284chemicals, 36 yielded four positive results, 25

three positives, 51 two positives, and 32 one posi-tive result, for a total of 144 chemicals testing posi-tive in at least one experiment.

OTA did not incorporate any additional dataon the affected animal tumor sites, on whetherboth high and low doses (or all three doses in athree-dose experiment) produced a response, oron chemicals’ estimated potencies. The groupingof substances for this analysis is also based onlyon the results of NCI/NTP testing. OTA has notused the bioassay results of others or the resultsof human epidemiologic studies.

Annual Report on Carcinogens

In 1978, Congress mandated that the DHHSpublish an annual report listing all known car-cinogenic substances and substances reasonably

thought to be carcinogenic to which a significantnumber of people in the United States are exposed.Furthermore, the report is to describe regulatory

actions on these substances, and estimate howmuch those actions have reduced risk. The legis-lation’s first sponsors thought this discussionwould help focus on chemical exposures that stillpresent risks, and thus on areas for regulatory

activity.

The substances discussed in the report are cho-sen by an interagency committee, including rep-resentatives of CPSC, EPA, FDA, NCI, NIEHS,NIOSH, the National Library of Medicine, andOSHA. The committee bases its decisions on theprevious Annual Report, lists of chemicals judgedto be supported by sufficient evidence for car-cinogenicity by the International Agency for Re-search on Cancer (IARC), and animal testing re-sults from NTP and other peer-reviewed studies.They publish the list of possible additions for com-ments and then make their final selections, Thelatest Annual Report, the fourth, lists a total of148 substances, groups of substances, and expo-sures. For this analysis, OTA eliminated double-counted chemicals in this list for a total of 145chemicals.

OTA Analysis

OTA examined regulatory responses to threegroups of chemicals: all NCI/NTP-tested chemi-cals with at least one positive experiment, the

19

NCI/NTP chemicals with three or four positiveexperiments, and the chemicals listed in the fourthAnnual Report on Carcinogens. While OTA ana-lyzed the three separately, in fact there is someoverlap of the three lists. All the chemicals test-ing positive in three or four experiments of coursealso tested positive in at least one experiment. Inaddition, many of the chemicals with three or fourpositive results have been listed in the AnnualReport.

OTA focused on the chemicals of potential reg-ulatory interest for each agency or program: thechemicals found in specific environmental media,such as air or drinking water, occupational set-tings, consumer products, pesticides, food, anddrugs. Information on exposures is, unfortu-nately, often simply unavailable. Quantitative in-formation is particularly difficult to obtain. SoOTA relied on information on estimated produc-tion levels, estimated number of workers exposed,and qualitative data on the presence of particu-lar chemicals in given situations. Even using thisinformation on regulatory jurisdictions, OTAfound apparent gaps in regulatory coverage. Fig-ure 1-2 summarizes OTA’s analysis of agency ac-tions and nonactions on chemicals in their juris-dictions.

The impact of these regulatory gaps on humanhealth depends on factors not analyzed by OTA,including the extent and magnitude of exposures,the potency of the chemicals, and other poten-tially synergistic or antagonistic exposures andrisk factors. Many agency analyses conducted todevelop information prior to regulation on infor-mation hazards, risks, control technologies, costs,and other factors—have not been included in theactions discussed here.

Regulation of Chemicals Tested by NCI/NTP

While a number of regulatory actions appearto have been based directly on positive NCI/NTPtest results, there also appear to be substantialgaps in regulatory activity. In the NCI/NTP bi-oassay program, 144 chemicals tested positive inat least one experiment. Considering each agencyand program individually reveals that no agencyhas regulated more than a third of the chemicals

with positive test results. More typically, an in-dividual agency will have acted out of concernfor carcinogenicity on 5 to 30 of the 144 chemicals.

FDA has taken action on 17 of the 48 positiveNCI/NTP chemicals associated with food addi-tives, color additives, or cosmetics. The balancehave been evaluated, but have not been subjectto further action. FDA has acted on 4 of the 5positive NCI/NTP chemicals associated with ani-mal drugs, and 6 of the 12 positive NCI/NTPchemicals that are human drugs bear labeling thatwarns of carcinogenicity. OSHA has set exposurestandards for 29 of the 53 positive NCI/NTPchemicals that are of interest in the workplace,although 27 of these 29 are regulated by stand-ards based on concern for noncarcinogenic tox-icity, which were adopted by OSHA in 1971.NIOSH has provided OSHA with recommenda-tions on 31 of the 62 positive NCI/NTP chemi-cals in its OTA-defined jurisdiction. Regulatoryaction or voluntary exposure reductions haveoccurred for 8 of the 14 positive NCI/NTP chem-icals in CPSC’s jurisdiction. EPA has listed un-der the Clean Air Act 2 of 12 positive NCI/NTPchemicals within the act’s jurisdiction. Water qual-ity criteria have been prepared for 14 of the 27positive NCI/NTP chemicals in the jurisdictionof the Clean Water Act. Of the 14 positiveNCI/NTP chemicals in the jurisdiction of the SafeDrinking Water Act, 12 have been addressed bysome regulatory attention, although for many ofthese, the regulatory process is not yet finished.EPA has developed information on 53 of the 144positive NCI/NTP chemicals in the TSCA’s juris-diction. For 5 of the 144 chemicals, EPA has is-sued SNURs, begun accelerated reviews, or takenaction under section 6 of the act. Under FIFRA,there have been EPA-ordered or voluntary can-cellations for 13 of the 22 positive NCI/NTPchemicals used as active pesticide ingredients. Ofthe 144 positive NCI/NTP chemicals, 41 havebeen included in RCRA’s list of hazardous wastesor its Appendix VIII list, while 47 of the 144positive NCI/NTP chemicals are listed underCERCLA. CAG has prepared health assessmentsfor 22 of the 144 positive NCI/NTP chemicals.No actions have occurred for 43 of the 144 posi-tive NCI/NTP chemicals.

20

21

Limiting attention to those chemicals with threeor four positive experiments reveals that agenciesand programs have each acted on 1 to 22 of the61 NCI/NTP chemicals with these results. Chem-icals with three or four positive experiments willgenerate greater concern because in these casesthere are positive results from both rats and mice.FDA has taken some regulatory action on 7 ofthe 19 chemicals with three or four positive ex-periments associated with food or color additivesor cosmetics. The one animal drug with three orfour positive results has been revoked whiles ofthe 6 chemicals with three or four positive exper-iments have been removed from human drugs orhave been labeled for carcinogenicity. OSHA hasregulated 16 of the 30 chemicals with three or fourpositive experiments that are in its jurisdiction.One of these standards is based on carcinogenic-ity. NIOSH has made recommendations on 13 ofthe 39 chemicals in its jurisdiction with three orfour positive results. In CPSC’s jurisdiction, 4 of7 chemicals have been subject to regulatory orvoluntary action. Under the Clean Air Act, EPAhas listed one of eight chemicals with three or fourpositive results. Water quality criteria have beenissued for 7 of 10 chemicals in the Clean WaterAct jurisdiction, and some regulatory action hasoccurred for 6 of the 7 chemicals under the juris-diction of the Safe Drinking Water Act. Informa-tion has been developed under TSCA for 22 ofthe 61 chemicals with three or four positive ex-periments and SNURs, accelerated reviews, andsection 6 actions have addressed 2 of the 61. EPA-ordered and voluntary cancellations have oc-curred for 5 of the 11 active pesticide ingredientswith three or four positive experiments. RCRAlists include 22 of the 61 chemicals with three orfour positive experiments, and the CERCLA listcovers 22 of the 61. CAG assessments address 9of the 61. No actions have addressed 23 of the61 chemicals with three or four positive exper-iments.

Regulation of Chemicals Listed inthe Annual Report on Carcinogens

All the Annual Report chemicals have been ad-dressed by at least one agency, although a largenumber of these chemicals have not been actedon by all the agencies and programs that might

have an interest in them. Except for chemicals onthe lists adopted under RCRA and CERCLA, noagency has regulated as many as half the chemi-cals included in the Annual Report. Generally,agencies have acted on 5 to 60 of these 145 An-nual Report chemicals.

FDA has acted on 46 of the 52 Annual Reportchemicals in its jurisdiction for food and color ad-ditives and cosmetics, and on 2 of the 6 AnnualReport chemicals used as animal drugs. Of the 31Annual Report chemicals with human drug uses,26 have been removed from the market or havecarcinogenicity warning labels. OSHA has ex-posure standards for 52 of 110 Annual Reportchemicals in its jurisdiction; 17 of these standardsare based on carcinogenicity. All Annual Reportchemicals are covered by OSHA’s hazard com-munication standard. NIOSH has made recom-mendations on 59 of the 112 Annual Report chem-icals in its jurisdiction. Voluntary and regulatory

actions have been taken on 18 of the 23 AnnualReport chemicals in CPSC’s jurisdiction. EPA list-ings under the Clean Air Act address 6 of 15 An-nual Report chemicals in the act’s jurisdiction. For48 of 65 Annual Report chemicals in the jurisdic-tion of the Clean Water Act, water quality cri-teria have been prepared. Interim standards un-der the Safe Drinking Water Act, and the currentRMCL/MCL process address 21 of 32 Annual Re-port chemicals within the act’s jurisdiction. EPAhas developed information on 28 of the 145 An-nual Report chemicals in the TSCA jurisdictionand issued SNURs, started accelerated reviews,or section 6 actions on 6 of the 145. EPA-orderedand voluntary cancellations have affected 12 ofthe 24 Annual Report chemicals used as active in-gredients in pesticides. The RCRA lists address97, and the CERCLA lists 95 of the 145 AnnualReport chemicals. CAG assessments cover 78 ofthe 145.

Comments on the OTA Analysis

In comments on a draft of this background pa-per, officials of Federal regulatory agencies em-phasized their belief that they have acted appro-priately in regulating the chemicals tested byNCI/NTP and the chemicals in the Annual Re-port. They pointed out that statutes require they

22

assess the risks and benefits of using chemicals,and the technical feasibility and costs of regula-tory action. Because of these considerations, aswell as their judgments about the weight of evi-dence for carcinogenicity, in some cases they havedecided not to regulate substances. In other cases,the chemicals are being considered as subjects ofregulatory action.

Future Improvements

Today the hope for a more complete under-standing of cancer causation rests on research intobiochemical markers, pharmacokinetics, and mo-lecular mechanisms. Nevertheless, science cannotnow answer all the questions that are raised inthis field. Even in the face of such uncertainty,however, it is important to take action to pro-tect public health.

Ever since the development of carcinogenicitybioassays, there has been skepticism about thereliability of animal results for estimating humanrisk. The Federal agencies have usually assumedthe usefulness of animal test results. However, reg-ulated industries have often disputed these resultsin particular cases and express concern that soci-ety not impose unnecessary regulations. These dis-putes are not likely to go away.

To force regulatory action, Congress has legis-lated a variety of statutory mechanisms. The mostcommon of these have been statutory deadlines,which have sometimes led to regulatory action,but are also frequently missed by the agencies.In the 1984 RCRA amendments, Congress in-

cluded “hammers”-statutory provisions that gointo effect if EPA misses particular deadlines. Con-gress has also mandated requirements, such asTSCA’S ban of PCBs, and agency adoption orconsideration of designated lists of chemicals. Inone case (that of saccharin regulation), Congressprohibited an agency from acting. A final con-gressional mechanism is requiring agencies to con-sider or respond to recommendations of anotheragency or organization. For example, OSHA mustconsider the recommendations of NIOSH, EPAmust respond to nominations of chemicals by theITC, and, in the original Safe Drinking Water Act,EPA was to respond to National Academy of Sci-ences recommendations.

In light of the regulatory gaps revealed byOTA’s analysis of agency responses to positiveNCI/NTP bioassay results and the list of chemi-cals in the Annual Report on Carcinogens, Con-gress may wish to consider a statutory require-ment mandating that agencies regulate thesechemicals or at least publicly respond to thesesources of information, even if, for various rea-sons, they choose not to regulate. On the otherhand, such a requirement might make develop-ing the Annual Report or selecting chemicals forNTP testing more difficult. In addition, regula-tory action may not always be necessary and, iftaken, may impose costs on regulated industries.Finally, in light of the importance exposures playin determining the need for regulation, it mightbe appropriate to develop additional informationon the extent of human exposures to thesechemicals.