Parametric analysis of rank transformed data for statistical assessment of genotoxicity data with examples from cultured mammalian cells

The utility of rank transformation followed by parametric analysisof the ranks has been assessed for determination of the statisticalsignificance of genotoxicity data. Both non-parametric and parametricanalytical methods have defects when used to assess the significanceof results from routine regulatory tests. Superficially, therank transformation method followed by parametric analysis ofranks appears to be an ideal solution. However, we consideredthat such a test might suffer a substantial loss of power whenused to analyse normally distributed data with very low samplereplication. To test this hypothesis we took 22 data sets fromfive ‘borderline’ positive compounds in mouse lymphomatreat-and-plate assays where treatment-related increases werebetween 1.5- and 3-fold the control and analysed these resultsby Dunnett's t-test using rank transformed data and weighted,untransformed data. In theory these mouse lymphoma data shouldshow the rank transformation system at its worst in comparisonwith parametric methodology using weighted data. Surprisingly,the rank transformation methodology showed no loss of powerand, overall, performed more consistently than the weighteddata methodology. Based on this limited number of data sets,rank transformation followed by parametric analysis of ranksseems to be an approach very suitable for genotoxicity assaysin general, particularly where distributions are non-normalor of uncertain form. It combines the general applicabilityof non-parametric methods with the power of parametric analyses.However, the methodology still requires to be further validatedin use and by computer simulation.