The use of short-term tests and limited bioassays in carcinogenicity testing

A brief discussion is given of the use of short-term in vivo and in vitro tests in carcinogenicity testing. Data are presented on the performance of nine such tests and five limited bioassays, as measured in terms of their sensitivity, specificity, accuracy, and predictive value. It is concluded that tests are available which, when used in combinations such as have been proposed by various authors, are capable of correctly identifying carcinogens and noncarcinogens with high confidence. A discussion of the statistics of batteries of tests is given. This is followed by a semiquantitative graphical representation and a discussion of how, on proceeding through a testing scheme such as that proposed by J. H. Weisburger and G. M. Williams (1981, Science 214, 401-407), the cumulative cost and the probability of correctly identifying a carcinogen or a noncarcinogen change as results become available from each of the stages of such a testing scheme.