Although the Pullman’s (Chapter 3) were among the first to attempt a structure-activity relation- ship of carcinogenic chemicals in relating this to predicting carcinogenicity in animals and po- tentially humans, this approach has been expanded during the last several decades. This is largely the result of various methodologies for the analysis of chemical structures and the exten- sive capability of data handling by computers and more modern technologies. Figure 13.11 de- picts a summarized diagram of many of the “reactive forms” of chemical carcinogens. However, there are numerous other subtleties in structure that may be related to some carcinogenic poten- tial. While it is beyond the scope of this text to consider this subject in detail, the student may be interested in reading several reviews of the subject using different approaches, such as those of Klopman and Rosenkranz, 1994; Enslein et al., 1994; and Zhang et al., 1996. Attempts have been made to predict carcinogenicity of nongenotoxic chemicals by structure-activity relation- ships (Lee et al., 1995) and distinguishing between genotoxic and nongenotoxic chemicals by similar methods (Cunningham et al., 1998). An extensive review of the implications of structure- activity relationships for cross species extrapolation in DNA damage and repair as related to mutagenesis and carcinogenesis has also been published (Vogel et al., 1996). In ideal circum- stances, prediction of the carcinogenic action of a chemical by its structure is very much to be desired. In the final analysis, the prediction of the carcinogenic effect of specific structures in the human would be most useful, but as yet the biological effects of carcinogenic chemicals in the human are so diverse that no such ideal for extrapolation has been found.
Several other issues are also relevant to cross-species extrapolation, including differences in metabolism of chemical agents between the species. While metabolic schemes are qualita- tively similar across species, significant quantitative differences, especially in metabolic rate, partly owing to elimination kinetics, are the rule. Exposure estimation is frequently based on the daily dose administered or on plasma concentration used as a surrogate for concentrations in the tissue. Using plasma concentrations for extrapolation across species assumes that each species responds in the same manner to any given dose of an agent. In the final analysis, it may be that the best basis for cross-species comparison is serum concentration expressed as milligrams per kilogram body weight, since this better predicts tissue concentration-response effects after chronic administration (Allen et al., 1988; Monro, 1992). Thus, it is clear that the problem of extrapolation of both short- and long-term tests to carcinogenic potential in the human is much less than perfect, suggesting an important need for reevaluation and reinterpretation of the tests currently in use. Needless to say, a program to develop better extrapolative end points should be a major priority.