A composite statistical procedure for evaluating genotoxicity using cell transformation assay data

This study proposes a statistical procedure for evaluating data from the BALB/c 3T3 cell transformation assay (CTA). Three steps are applied: (i) a Dunnett type test for identifying clear negativity, (ii) a Margolin procedure for detecting and eliminating downturn data, and (iii) a maximum contrast method for evaluating the dose‐dependent increase of the response. Using Monte Carlo simulation the proposed procedure was compared to three other methods, i.e. a linear regression test, the Simpson–Margolin procedure with linear regression test (trend test), and the Omori procedure. When false positive rates were commonly controlled at a similar level for all methods, the procedure's effectiveness was greatest for dose–response patterns exhibiting non‐linear and down turn parts, being in agreement with a collaborative CTA validation study performed in Japan. Evaluation of CTA data by the proposed procedure satisfied participating researchers. Copyright © 2003 John Wiley & Sons, Ltd.