Reactive substructures in the prediction of aquatic toxicity data

Abstract

Baseline QSAR models relating acute toxicity to log P ow are analyzed for predicting fish LC 50 data of various compounds. Mostly ( > 75%), the high dependency of aquatic toxicity on lipophilicity resulted in agreement of calculated and experimental data. Deviations occur with respect to compounds revealing excess toxicity. From fathead minnow LC 50 data for 500 organic chemicals, the toxicity of 122 compounds was found to be severely underestimated on the basis of log P ow alone. Among these ‘outliers’, 14 reactive substructures were identified, which accounted for > 73% of the compounds with excess toxicity. The application of a baseline QSAR supplemented by indicators for the reactive substructures also allowed the prediction of toxicity data based on other endpoints, indicating similar modes of interaction in various aquatic tests.