Prediction of skin irritation from organic chemicals using membrane-interaction QSAR analysis.

Membrane-interaction (MI) quantitative structure activity relationship (QSAR) analysis was carried out for a training set of 22 hydroxy organic compounds for which the Draize skin irritation scores, PII, had been determined. Significant MI-QSAR models were constructed in which skin irritation potency is predicted to increase with (1) increasing effective concentration of the compound available for uptake into phospholipid-rich regions of a cellular membrane, (2) increasing binding of the compound to the phospholipid-rich regions of a cellular membrane, and (3) the chemical reactivity of the compound as reflected by the highest occupied molecular orbital (HOMO) and/or lowest unoccupied molecular orbital (LUMO) of the molecule. Overall, the MI-QSAR models constructed for skin irritation are very similar, with respect to the types of descriptors, to those found for eye irritation. In turn, the skin irritation MI-QSAR models suggest a similar molecular mechanism of action to that postulated for eye irritation from MI-QSAR analysis. Significant and predictive QSAR models cannot be constructed unless test compound-membrane interaction descriptors are computed and used to build the QSAR models.