Modeling the Ames Salmonella/microsome assay

Of the many short-term tests for mutagenicity that have been proposed in recent years, the Ames Salmonella/microsome assay is the single most widely used and most thoroughly validated in vitro test system. This assay uses cells cultured in a soft agar containing a trace amount of histidine to allow growth of auxotrophic bacteria, and is designed to detect reverse mutations from auxotrophic cells to histidine-independent prototrophic cells. In this paper, statistical models that have been proposed for the analysis of Ames test data are reviewed, including those of a mechanistic and empirical nature. An extension to the class of biologically based models derived by Margolin, Kaplan, and Zeiger (1981, Proceedings of the National Academy of Sciences 78, 3779-3783) is proposed by allowing for diffusion of histidine within the plate agar. Quasi-likelihood methods for estimating the model parameters are presented, and applied to 1,120 data sets from a recent collaborative trial sponsored by the International Programme on Chemical Safety.