Comparison between rodent carcinogenicity test results of 44 chemicals and a number of predictive systems.

A comparison is made between a number of predictive systems including bacterial mutagenicity, structure alert and chronic toxicity (Tennant et al., 1990), COMPACT, Hazardexpert, and DEREK, with the outcome of the two species rodent carcinogenicity bioassay for 44 chemicals conducted under the NTP protocol. It is shown that, following minor updating of the rodent data in the light of pathology reports, there is a generally good agreement between various predictions and the actual carcinogenicity results. In particular, a combination of two methods of prediction produces an over 80% concordance with the rodent bioassay. Possible reasons for various discrepancies are discussed in light of xenobiotic metabolism and activation mechanisms of chemical carcinogenesis.