A statistical evaluation of the reproducibility of micronucleus, sister-chromatid exchange, thioguanine-resistance and ouabain-resistance assays in V79 cells exposed to ethyl methanesulfonate and 7,12-dimethylbenz[a]anthracene.

In contrast to the "validation" of short-term in vitro genotoxicity assays by concordance with the rodent cancer bioassay, the present report describes the multiple replication of 4 short-term tests with V79 cells (micronucleus assay, MN; sister-chromatid exchange, SCE; ouabain resistance. OUR; and thioguanine resistance, TGR) within the same assay system following exposure to each of two genotoxins, ethyl methanesulfonate (direct acting) and 7,12-dimethylbenz[a]anthracene (indirect acting). Reproducibility, proportion of genotoxins correctly identified, and proportion of non-genotoxins correctly identified by each test were each determined statistically. Decision rules were formulated to declare a positive response in each assay, and overall accuracy of each was determined. Statistical analysis of the data, obtained under standardized test conditions, showed that for these two chemicals SCE identified 100% of genotoxins and 86% of non-genotoxins, with overall accuracy of prediction of 93%; TGR identified 98% of genotoxins and 74% of non-genotoxins, with overall accuracy of 86%; MN identified 78% of genotoxins and 84% of non-genotoxins, with overall accuracy of 81%; while OUR indicated 100% of genotoxins, but only 50% of non-genotoxins, and only 76% overall accuracy. The results suggested that the best overall accuracy of classification with the V79 assay system could be achieved by measurement of SCE in combination with thioguanine resistance.