Predictive values of traditional animal bioassay studies for human perinatal carcinogenesis risk determination.

The many physiological, biochemical, and structure differences between rodents and humans, especially with regard to gestation and fetal development, invite questions as to the utility of rodent models for the prediction of risk of perinatal carcinogenesis in humans and for extrapolation of mechanistic studies. Here, the relevance of basic generalities, derived from rodent perinatal studies, to human contexts is considered. The cross-species usefulness of these generalities was upheld by the example of carcinogen activation and detoxification as determining factors. These have been established in rodent studies and recently indicted in humans by investigations of genetic polymorphisms in cytochromes P450, N-acetyltransferase, myeloperoxidase, quinone reductase, and glutathione S-transferase. Also, published data have been analyzed comparatively for diethylstilbestrol and irradiation, the two known human transplacental carcinogenic agents. At similar doses to those experienced by humans, both diethylstilbestrol and X- and gamma-irradiation in rodents and dogs yielded increased tumors at rates similar to those for humans. In rodents, there was a clearly negative relationship between total diethylstilbestrol dose and tumors per dose unit, and a similar pattern was suggested for radiation. Diethylstilbestrol had transgenerational effects that did not diminish over three generations. Overall, this analysis of the published literature indicates that there are basic qualitative and quantitative similarities in the responsiveness of human and rodent fetuses to carcinogens, and that dose effects may be complex and in need of further investigation.