Evaluation of seven alternative assays on the main ingredients in cosmetics as predictors of Draize eye irritation scores.

This study evaluates seven alternative assays carried out on the main ingredients in cosmetics to determine which battery is the best set and what is the best predictor of the maximal Draize rabbit eye irritation scores (MDESs). The assays consisted of the maximal primary Draize rabbit skin irritation scores (MDSSs), a cytotoxicity test on neutral red uptake using Chinese hamster lung cells (NR-EC(50)), a cytotoxicity test on MTT using normal skin fibroblasts (MTT-EC(50)), the hen's egg test-chorioallantoic membrane test using fertile chicken eggs (HET-CAM), a haemolysis test using red blood cells from Wistar rats (HC(50)), a protein denaturation test using haemoglobin from bovine (HDR), and pH. We tested 10% solutions of 24 tested chemicals, that is, 20 surfactants, three solvents and formaldehyde, to select from the assays a best set for prediction (x) and to obtain the best predictor [f(x)] based on the prediction sum of squares criterion. The selected set consisted of NR-EC(50) and HDR, and the resultant best predictor was f(x) = 74.0 - 29.52 log(NR-EC(50)) + 0.87 HDR. This predictor achieved a high score of 89.6% of the contribution ratio. For two of the 24 test chemicals, an inconsistency occurred in the criticality of corneal damage between the observed and predicted Draize eye scores, although this is not considered to affect significantly the overall results.