Multivariate statistical methods in toxicology. III. Specifying joint toxic interaction using multiple regression analysis

Abstract

Multiple regression is widely employed to study the contribution of components to the toxicologic effect of a mixture. Here, use is made of the fact that data obtained from standard curves of substances and from their mixtures are separable in regression analysis. Thus, under an assumption of additivity of responses, regression coefficients obtained for components in mixtures alone should be the same as for the individual substances. A t-test is developed such that nonsignificant t values support additivity, negative significant values support antagonism, and positive significant values support synergism. The results are applied to data on the mutagenicity of binary mixtures of azaserine, 4-nitroquinoline N-oxide, and 9-aminoacridine in TA 100 in the Ames assay.