Abstract
We are referring in this paper to traditional tests of genotoxicity that are essentially detecting properties correlated with the initiating activity of chemical and physical agents. In classical initiation-promotion experiments, promotion is changing dramatically tumor frequency, but a better correlation between initiation and tumor frequency may exist when carcinogenesis experiments in rodents utilize only an initiator agent. End points, metabolism and target organs of short term test may be not optimal in respect to the real initiation process. In conclusion, we have to expect a very high level of statistical noise in the correlation between short term tests and carcinogenicity in rodents. Correlations of this type are not unique in science. For instance, the input data for weather forecasting and the real weather are of ten related with a high level of statistical noise. But this is not considered a good reason for throwing away the quantitative component of the input information. Similarly, in this context the utilization of the quantitative component of the information improves in a moderate, but not negligible way the predictive value of the information of fered from short term tests, especially when they are used in a battery. In this review we discuss the possibility that, when studying correlations between genotoxicity tests and carcinogenicity in rodents, a quantitative approach could replace with important advantages the qualitative approach generally adopted up to the present time. The qualitative approach appeared as a more modest but realistic approach to the complexities of these correlation studies. What we suggest is that a quantitative approach mathematically not less correct than the qualitative approach is feasible. More precise and useful information for risk assessment evaluations can be obtained.