Analysis of animal carcinogenicity experiments with multiple tumor types.

Because of the occult nature of tumors, the time-to-tumor is usually unobservable in animal studies of chemical carcinogenicity. In this paper, we propose new test approaches for simultaneous comparison of the marginal tumor incidence rates with several tumor types. We first set forth a model for tumor development and discuss the identifiability of parameters in this model. We then combine the results of tests for each marginal tumor incidence rate to develop simultaneous tests of all marginal tumor incidences. We apply the proposed methods to data from the ED01 study. This study, which used about 24,000 mice, was designed to determine the carcinogenic effect of the compound 2-acetylaminofluorene (2-AAF) at low dosages. We also present some simulation results to illustrate the application of simultaneous test methods to small experiments using 100 mice per dose group.