Statistical methods to decide whether mutagenicity test data from Drosophila assays indicate a positive, negative, or inconclusive result

Two alternative hypotheses are used to distinguish among the possibilities of a positive, inconclusive, or negative result in Drosophila mutagenicity tests. In the null hypothesis one assumes that there is no difference in the mutation frequency between control and treated series. The alternative hypothesis postulates a priori that the treatment results in an increased mutation frequency that is m times the spontaneous frequency. To test against the hypotheses, the conditional binomial test according to Kastenbaum and Bowman or the chi 2 test for proportions may be applied. These 2 methods are in principle equivalent. An alternative method which is based on determining confidence limits of observed mutation frequencies also leads to the same conclusions. The practical calculations are formulated and an application is shown with a test example demonstrating the genotoxicity of the pyrrolizidine alkaloid 7-acetylintermedine in the somatic wing mosaic test. In the Appendix, the calculus for the 3 testing methods is explained with a numerical example.