Roberts, D. W.; Aptula, A. O.; Patlewicz, G. Mechanistic Applicability Domains for Non-Animal Based Prediction of Toxicological Endpoints. QSAR Analysis of the Schiff Base Applicability Domain for Skin Sensitization. Chem. Res. Toxicol. 2006, 19, 9, 1228–1233.

Abstract

Several recent (1999 onward) publications on skin sensitization to aldehydes and ketones, which can sensitize by covalent binding to skin protein via Schiff base formation, present QSARs based on the Taft sigma* parameter to model reactivity and log P to model hydrophobicity. Here, all of the data are reanalyzed together in a stepwise self-consistent way using the parameters log P (octanol/water) and sum_sigma*, the latter being the sum of Taft sigma* values for the two groups R and R' in RCOR'. A QSAR is derived: pEC3 = 1.12(0.07) sum_sigma* + 0.42(0.04) log P - 0.62(0.13); n = 16 R2 = 0.952 R2adj = 0.945 s = 0.12 F = 129.6, based on mouse local lymph node assay (LLNA) data for 11 aliphatic aldehydes, 1 alpha-ketoester and 4 alpha,beta-diketones. In developing this QSAR, an initial regression equation for a training set of 10 aldehydes was found to predict a test set consisting of the other 6 compounds. The QSAR is found to be well predictive for LLNA data on a series of alpha,gamma-diketones and also correctly predicts the nonsensitizing properties of simple dialkylketones. It is shown to meet all of the criteria of the OECD principles for applicability within regulatory practice. In view of the structural diversity within the sets of compounds considered here, the present findings confirm the view that within the mechanistic applicability domain the differences in sensitization potential are dependent solely on differences in chemical reactivity and partitioning.