The in vivo Pig-a gene mutation assay serves to evaluate the genotoxic potential of chemicals. In the rat blood-based assay, the lack of CD59 on the surface of erythrocytes is quantified via fluorophore-labeled antibodies in conjunction with flow cytometric analysis to determine the frequency of Pig-a mutant phenotype cells. The assay has achieved regulatory relevance as it is suggested as an in vivo follow-up test for Ames mutagens in the recent ICH M7 [25] step 4 document. However, very little work exists regarding suitable statistical approaches for analyzing Pig-a data. In the current report, we present a statistical strategy based on a two factor model involving 'treatment' and 'time' incl. their interaction and a baseline covariate for log proportions to compare treatment and vehicle data per time point as well as in time. In doing so, multiple contrast tests allow us to discover time-related changes within and between treatment groups in addition to multiple treatment comparisons to a control group per single time point. We compare our proposed strategy with the results of classical Dunnett and Wilcoxon-Mann-Whitney tests using two data sets describing the mode of action of Chlorambucil and Glycidyl methacrylate both analyzed in a 28-day treatment schedule.
Keywords: In vivo Pig-a test; Linear model; Multiple contrasts.
Copyright © 2018 Elsevier B.V. All rights reserved.