The long-term animal bioassay for carcinogenicity was developed during the 1960s and early 1970s primarily as a qualitative screen for carcinogenic potential. Long-term animal bioassays are now used regularly to determine whether chemical agents are capable of inducing cancer in exposed animals. The bioassays are also commonly used as a basis for making qualitative inferences about the likelihood that an agent poses a carcinogenic hazard for humans as well (IARC, 1991).
Because of practical considerations, such as the cost of maintaining large numbers of animals for long periods, the number of animals used in long-term studies is generally limited to about 50 per dose-sex-species group tested. That limits the sensitivity of the carcinogenicity bioassay: it cannot detect a small increase in tumor incidence, such as an increase of 1% or less, even in experiments that use hundreds of animals. To minimize the number of false-negative results, the bioassay design was modified early in its development. The most important modifications were extension of the testing period to cover most of the lifetime of the experimental animals (which, for practical reasons, limited the test species to small rodents with lifetimes of 2-3 years) and the use of high doses.
A carcinogenicity bioassay generally involves animals exposed at two or more doses and a control group. A higher dose generally is more likely than a lower dose to produce cancer in the test animals and hence
to increase the likelihood that a carcinogen will be detected. However, too high a dose might cause toxic effects that shorten the life of the test animals and prevent the observation of an excess tumor incidence. Those considerations led to the practice of selecting the maximum tolerated dose (MTD) as the highest dose tested (HDT) in an animal bioassay. The MTD is roughly described as the highest dose that does not alter the animals' longevity or well-being because of noncancer effects (Sontag et al., 1976; McConnell, 1989). These terms are further defined later in the report.
The MTD is generally estimated in a preliminary study by subjecting small groups of animals to a series of doses (perhaps six) for a small fraction of a lifetime (e.g., 3-months for mice and rats). The highest dose judged to cause no overt toxicity and little or no growth suppression is the estimated maximum tolerated dose (EMTD).1
Estimation sometimes results in selection of an EMTD that is too high—that causes animals to die early in life before chemically induced cancers could occur. Because it is difficult to interpret the results of animal bioassays when animals die prematurely, the bioassay design was refined to include testing at a lower dose as well—often half the EMTD (EMTD/2). Other doses (such as EMTD/5 or EMTD/10) are also used to define dose-response relationships better. Current bioassay designs have become reasonably well standardized and usually specify lifetime testing of both sexes of two species of rodents at two or more doses, the highest of which is the MTD (IARC, 1986a,b). Criteria for interpreting results obtained in these tests and for classifying them as positive, equivocal, or negative have been developed and refined (see the technical reports series of the U.S. National Toxicology Program and many others).
Although the carcinogenicity bioassay in rodents was developed primarily for qualitative screening of agents for carcinogenicity, it often provides the only quantitative information for evaluating the relationship between dose and carcinogenic response and for estimating the carcinogenic potency of an agent. Procedures for quantitative risk assessment
Since 1970, several hundred chemical agents have been tested for carcinogenicity in bioassays of standard designs. The National Toxicology Program (NTP) alone has reported on 382 bioassays, of which 195 (51%) identified the tested chemical as carcinogenic under the conditions of the bioassay in at least one species-sex group (R. Griesemer, NIEHS, pers. comm., 1991). That proportion is not representative of chemicals in general, however, because of how the chemicals were selected for testing. Most of the substances (255 of 382, or 67%) were selected for testing primarily because of suspicion of carcinogenicity, and 169 (66%) of the 255 were positive. The remainder (127 of 382, or 33%) was selected for testing mainly on the basis of human exposures and the lack of toxicity data, and only 26 (20%) of the 127 were positive (R. Griesemer, NIEHS, pers. comm., 1991).
Limitations inherent in using the MTD approach and suggestions for improvement have been the subject of controversy since its use became standard (Shubik, 1978). In recent years, the use of data from bioassays performed with the MTD has been called into further question. Some of the criticisms of such data are based on the following points:
A large percentage of chemicals tested by the NTP have been identified as carcinogenic in at least one species-sex group. Some observers believe that the test is labeling so many substances as carcinogenic that regulatory attention and public concern have been focused on many agents that pose only trivial hazards, while attention has been diverted from other agents that pose more important carcinogenic risks. The committee was given some evidence to support that charge. However, a high proportion of materials found positive in one or more species-sex groups have not been regulated (OTA, 1987).
At high doses (including MTD and MTD/2), some agents might