Today the gold standard for determining potential carcinogenic activity of a chemical is through the use of the chronic 2-year bioassay for carcinogenicity in rodents. This assay involves test groups of 50 rats and mice of both sexes and at two or three dose levels of the test agent. The
Table 13.4 Animal Models of Neoplastic Development
animals should be susceptible but not hypersensitive to the tested effect. In general, two strains are typically used by regulatory agencies in the United States, the B6C3F1 mouse and the F344 rat. The format for the bioassay is seen in Figure 13.7. Quite simply, animals at about 8 weeks of age are placed on the test agent at the various doses for another 96 weeks of their lifespan. The test agent may be administered by dietary feeding, by gavage on a regular basis, or by inhalation in rather complex chambers. A variety of pretest analyses are carried out, such as those for acute toxicity, route of administration, and determination of the maximum tolerated dose (MTD). The use of the MTD has been challenged by many, arguing that the toxic effects of high doses of an agent can cause a replicative response in normal cells that could lead to an increase in neoplasia quite secondary to the effects of the agent itself (Cutler et al., 1997; Haseman and Lockhart,1994). This is supported by the finding of a very high percentage, nearly half in some instances, of agents exhibiting no potential for mutagenicity but inducing neoplasia at the MTD (Gold et al., 1993). Furthermore, these two strains of rodents have a significant spontaneous tumor inci- dence, as can be noted in Table 13.5.
Because so many research dollars go into carcinogenicity testing and the data resulting from such studies are expected to be useful not only in hazard identification but also in risk estimation, an acceptable scientific protocol with quality assurance must be followed to produce scientifically and statistically valid data. A variety of factors relevant to the acceptable outcome of a carcinogenicity study are considered, including animal husbandry; the identity and purity of the test compound and identification of any contaminants; the homogeneity, stability, and physi- cal properties of the test compound under various storage conditions; and the solubility, stability, and availability of the test compound in the solvent. In addition, the formulation should be either that which is to be administered to humans or that which permits bioavailability in the test or- ganism. The environment of the rodent is also important, and care should be taken to control for sources of variability in the animals, their diet, and their housing. While the usual comparison in animal studies is the concurrent control, for a number of situations historical controls may be more appropriate (Haseman et al., 1997).
The underlying basis for risk extrapolation from animals to human is that the animal is a good model for human cancer development. In fact, 2-year bioassay models have been used to detect the compounds listed by the International Agency for Research on Cancer (IARC) (Vainio et al., 1991) as known human carcinogens. Also, most known human chemical carcinogens have a carcinogenic potential in animals that supports the results of epidemiological studies (Vainio et al., 1985). Exceptions include ethanol and arsenic. In addition, it has now become evident that some neoplastic responses to chemicals in animals are unique to the rodent and species as well as the sex involved. These include such responses as thyroid neoplasia (McClain, 1989), the in- duction of α2u-globulin (Swenberg et al., 1985) resulting in renal neoplasms in male rats, and peroxisome proliferation (Ashby et al., 1994) associated with the induction of hepatic neoplasia in rats. In addition, a significant problem that has arisen in the continued use of the chronic bio- assay is the requirement for ad libitum feeding. This results in animals, especially in rats, of extreme weight by the end of the 2 years; many will have died spontaneously prior to the end of
Table 13.5 Spontaneous Tumor Incidence (Combined Benign and Malignant) in Selected Sites of the Two Species, B6C3F1 Mice and F344 Rats, Used in the NCI/NTP Bioassay
the test. Such complications are now being remedied by the use of dietary restriction in the chronic bioassay for the 2-year period. As shown in Chapter 8, this phenomenon reduces sponta- neous cancer incidence and extends lifespan in rodents, and its usefulness in the refinement of the 2-year chronic bioassay is now becoming more appreciated (Keenan et al., 1996; Allaben et al., 1996).
The statistical analysis of results obtained in chronic bioassays has also been difficult when the analysis results in relatively few neoplasms in test animals. As can be seen from Table 13.6, a relatively high percentage of animals must bear tumors before a statistically significant result
Table 13.6 Percentage of Animals with Tumors (Rx) Administered a Test Agent Required to Obtain Statistical Significance When Compared with Control Animals with Tumors (Co)
can be obtained in the face of significant development of spontaneous lesions. Since the latter phenomenon is clearly a problem in these animals (Table 13.5), borderline results become a very difficult problem for regulatory agencies in determining whether or not a compound actually is carcinogenic in the assay or not. An exception to this is when a very unusual histogenetic type of neoplasm not seen spontaneously is found in the test animals at a significant, even very low level (Chu et al., 1981; Basu et al., 1996). The enumeration of all neoplasms versus those in specific tissues also can raise difficulties in interpretation of the bioassay. Despite these criticisms and problems, the chronic 2-year bioassay continues to be the major basis for regulatory action in this country and in many countries throughout the world.