A reexamination of the low prevalence of carcinogens in an early carcinogen screen.

In 1969 Innes et al. reported that 9% of 127 industrial chemicals and pesticides (including 7 positive controls) tested in a mouse bioassay were carcinogenic by oral administration. This is much lower than the prevalence of approximately 50% that has been found recently by the cancer bioassays conducted under the aegis of the U.S. National Toxicology Program (NTP) or reported in Carcinogen Potency Data Base compiled by Gold and collaborators. Using the CASE structure-activity relational expert system and the NTP cancer bioassay results as a learning set, we found that the predicted prevalence of carcinogens among the chemicals tested by Innes et al. is 62.7%. It is concluded that the bioassay protocol used by Innes et al. is less sensitive than the subsequently adopted bioassay procedures.