Abstract
A quantitative structure-activity relationship (QSAR) approach has been applied to the analysis of the relation between the structural features of chemicals and the primary eye irritation in rabbits. One hundred thirty one heterogeneous chemicals including medicals, pesticides, detergents and organic solvents were used in this study. The eye irritation ratings were made in three classes on the basis of the recovery time of corneal and conjunctival damages. Thirty six descriptors were used to describe the molecules. To correlate eye irritation ratings with the descriptors, a QSAR model was formulated by the adaptive least-squares method. A three-class discrimination was made as follows ; class I included 23 chemicals which induced the damages recovering within 24 hr, class II included 64 chemicals which induced the damages persisting for more than 24 hr but recovering within 21 days, class III included 44 chemicals which induced the damages not recovering within 21 days. The discriminant function included 18 descriptors. The accuracy in classifying the chemicals was 86.3% in the recognition and 74.0% in the leave-one-out prediction. These results suggest that QSAR analysis is valuable to predict the primary eye irritation of chemicals.
