Structure-activity studies of chemical carcinogens: use of an electrophilic reactivity parameter in a new QSAR model.

Electrophilic reactivity data for 142 compounds were obtained from the literature, and were used to establish the contribution of different functional groups and molecular determinants to this property. An equation containing approximately 20 molecular determinants was derived (r = 0.89); this provided a system for estimating the electrophilic reactivity for other compounds on the basis of their molecular structure. The contribution of the estimated electrophilicity to the structure-activity relationship (SAR) studies of carcinogens was tested. In a previous work, we studied a set of 137 carcinogens and non-carcinogens belonging to different chemical classes, and established an SAR on the basis of four physical-chemical descriptors. The mathematical model consisted of a pattern recognition method specifically designed by us for the non linear situations typical of large non congeneric sets of chemicals. The addition of the estimated electrophilicity parameter increased the overall performance of the system (from the previous 85-90% retrospective correct classification). In particular, the estimated electrophilicity remarkably contributed to the identification of carcinogens (their correct classification increasing from the previous 86% to 97%). The derived SAR was tested by applying it to 11 human carcinogens not included in the training set. Their carcinogenicity was correctly predicted for 10 chemicals out of 11.