The influence of chemical structure on the extent and sites of carcinogenesis for 522 rodent carcinogens and 55 different human carcinogen exposures.

Gold and her colleagues have tabulated the results of rodent bioassays on 522 chemicals and have analysed the data. The present study complements those analyses by providing a perspective from the viewpoint of the chemical structure of the carcinogens. The chemical structure of each of the carcinogens is displayed and the Gold database is represented with the test agents as the primary variable. The carcinogens are gathered into six chemical classes and each chemical is assessed for structural alerts to DNA reactivity. The database is then analysed using an integration of the following parameters: bioassay in rat, mouse or both; structural alert status; chemical class; sites and multiplicity of carcinogenesis, and trans-species carcinogenicity. A series of Figures is presented that enables rapid acquaintance with what represents the core database of rodent carcinogenicity. The several analyses presented combine in endorsing the reality of two broad classes of rodent carcinogen--presumed DNA-reactive and others (putative genotoxic and non-genotoxic carcinogens, but semantics have been largely avoided). Vainio and his colleagues have tabulated 55 situations in which humans have succumbed to chemically induced cancer, and have listed the tissues affected. This database of human carcinogens has been analysed in the present study as done for the rodent carcinogen database, and comparisons made between the two. The predominance of putative genotoxic carcinogens in the human database was confirmed, as was the reality of putative non-genotoxic carcinogenicity in humans. It is concluded that putative genotoxic rodent carcinogenesis can be correlated both with chemical structure and the extent and nature of the induced effect, and that it is of clear relevance to humans. In contrast, it is concluded that putative non-genotoxic rodent carcinogenesis is more closely related to the test species than to the test chemical, and that it is essentially unpredictable in the absence of mechanistic models. In the absence of such models nongenotoxic carcinogenic effects should be extrapolated to humans with caution. Progress in the accurate prediction and extrapolation of rodent carcinogenicity will be helped by a common, if only temporary, enabling acceptance that not all carcinogens are intrinsically genotoxic.