An alternative linear trend analysis for assessing the results of the mouse lymphoma assay

Abstract

As recommended by the mouse lymphoma assay (MLA) Workgroup of the International Workshop on Genotoxicity Testing (Aberdeen, 2003), a trend test is critical if an induced mutant frequency (MF) of at least 126 × 10(-6) (global evaluation factor, GEF) is achieved at one or more test concentrations. Only those responses that both achieve the GEF and a significant trend are biologically relevant. While no specific trend test was recommended by the Workshop, a trend test was recommended by the UK Environmental Mutagen Society (1989). The test uses MF (untransformed) averaged over replicate cultures following a consistency test (against a historical heterogeneity factor) in a weighted linear regression with chi-square (χ(2)) test for slope and returns significant results in virtually all cases that are positive for the GEF, including those with no apparent dose-response. We have explored an alternative method where the natural logarithm of MF and its variance are estimated for each replicate culture separately and used in a weighted ordinary linear regression with t-test for slope. Using test cases positive for the GEF, the P-value from this model is shown to be sensitive to changes in the number of replicates, the shape and magnitude of mutant induction, in contrast to the χ(2) model. Cases with no apparent dose-response and thereby questionable biological significance are tested negative by our method but positive by the χ(2) model. Our method is thus straight-forward and provides a meaningful complement to the GEF in assessing the biological significance of the MLA results.