Application of the carcinogenicity prediction and battery selection method to recent National Toxicology Program short-term test data.

Identification of potentially cancer-causing chemicals is a priority in our society. Short-term assays for mutation or chromosomal damage, which are rapid, inexpensive, and reproducible, have found widespread use; however, concern has arisen recently because such assays do not coincide completely with the standard rodent bioassay for carcinogenesis. Lack of perfect correlation is not surprising, given the complex, multicausal nature of the carcinogenic process. We have developed methodologies for interpreting short-term tests to predict carcinogenicity, which allow consideration of the influence of the proportion of carcinogens expected in the tested chemicals, the complexities of the rodent carcinogenesis bioassay, and factors affecting the worth of information. These methodologies are applied to a set of data on genotoxicity and carcinogenicity of 73 chemicals (NTP-73) recently published by the National Toxicology Program; they illustrate that with this approach, batteries of short-term tests can indeed be predictive of rodent carcinogenicity or noncarcinogenicity and that batteries are more predictive than the Salmonella assay alone. The analysis is validated using an additional group of chemicals with results in the same short-term tests as NTP-73.