PREDICTION OF CARCINOGENICITY FROM 2 VERSUS 4 SEX-SPECIES GROUPS IN THE CARCINOGENIC POTENCY DATABASE

Prediction of a positive result in rodent carcinogenesis bioassays usi ng two instead of four sex-species groups is examined for the subset o f chemicals in the Carcinogenic Potency Database that have been tested in four sex-species groups and are positive in at least one (n = 212) . Under the conditions of these bioassays, a very high proportion of r odent carcinogens that are identified as positive by tests in four gro ups is also identified by results from one sex of each species (86-92% ). Additionally, chemicals that are classified as ''two-species carcin ogens'' or ''multiple-site carcinogens'' on the basis of results from four sex-species groups are also identified as two-species or multiple -site carcinogens on the basis of two sex-species groups. Carcinogenic potency (TD50) values for the most potent target site are similar whe n based on results from two compared to four sex-species groups. Eight y-five percent of the potency values are within a factor of 2 of those obtained from tests in 4 sex-species groups, 94% are within a factor of 4, and 98% are within a factor of 10. This result is expected becau se carcinogenic potency values are constrained to a narrow range about the maximum dose tested in a bioassay, and the maximum doses administ ered to rats and mice are highly correlated and similar in dose level. Information that can be known in advance of a 2-yr bioassay (mutageni city, class, route, and maximum dose to test) does not identify groups of rodent carcinogens for which four sex-species groups are required to identify carcinogenicity. The range of accurate prediction of carci nogenicity using only male rats and female mice is 93% among mutagens and 88% among nonmutagens; for various routes of administration, 88-10 0%; for various chemical classes, 75-100%; and for various levels of t he maximum dose tested, 81-100%. Results are similar for the pair male rats and male mice. Using a strength of evidence approach, weaker car cinogens are somewhat less likely than stronger carcinogens to be iden tified by two sex-species groups. Strength of evidence is measured usi ng the proportion of experiments on a chemical that are positive, the extent to which tumors occur in animals that die before terminal sacri fice, and whether the chemical induces tumors at more than one site an d in more than one species.