Global tests for analysis of multiple tumour data from animal carcinogenicity experiments.

This paper presents test procedures for testing of the overall dose effect at every tumour type site simultaneously in a chronic animal carcinogenicity bioassay. I derive global tests of the tumour prevalence rates and the tumour onset rates using the generalized estimating equations approach. The test is based on the model of a common dose effect alternative. I used the logistic regression model for simultaneous testing of non-lethal tumours, and the Poly-3 test for testing of mixtures of lethal or non-lethal tumours of different types/sites. An application illustrating the proposed tests concerns analysis of a carcinogenicity experiment designed to study the dose response of hepatocellular and bladder tumours to 2-acetylaminofluorene (2-AAF) exposure.