Tests for effects on tumor frequency and latency in multiple dosing photococarcinogenicity experiments

A multiple induction test procedure that extends a single induction test procedure proposed by Kodell and Chen (Biometrical J. 43(4) (2001) 447) based on the work of Kokoska et al. (Anticancer Res. 13 (1993) 1357) is introduced. The new procedure can detect overall differences between two groups as well as isolate differences in the distribution of the number of induced tumors and the distribution of their times to observation. This “frequency-latency” procedure is illustrated with an analysis of data from a multiple dosing experiment, using a likelihood ratio method of testing. The results of the frequency-latency test are compared to those of the logrank test, the negative binomial test, and the test proposed by Dunson et al. (Toxicol. Sci. 55 (2000) 293). A Monte Carlo simulation study is performed to evaluate the accuracy of the parameter estimates of the frequency-latency procedure as well as to study the Type I error rates of the frequency-latency test, the logrank test, and the negative binomial test.