A Quasi-Likelihood Model for the Analysis of Tumorigenicity in Multigenerational Data
*Maria Mendoza, FDA/NCTR  James J. Chen, FDA/NCTR 

Keywords: survival, quasi-likelihood, multigenerational

In animal studies where time to tumor formation is not observed, tumor incidence data can be analyzed using the Poly-3 trend test which involves weighting animals, without tumors and censored before the end of study, by a power of the proportion of time they were on study. For multigenerational studies, we are interested in analyzing the effects of dose response on tumor incidence rates in mouse litters, of which mothers have been treated with a potential toxic compound. Analysis of tumor incidence may be complicated by the effects of intra-litter correlations. These effects can be bypassed by use of quasi-likelihood, which assumes knowledge of only the first two moments. We apply quasi-likelihood to the analysis of tumor incidence in poly-3 adjusted multigenerational data. We apply our approach to real data from a study conducted at the National Center for Toxicological Research.