Interlaboratory validation of the in vitro eye irritation tests for cosmetic ingredients. (1) Overview of the validation study and Draize scores for the evaluation of the tests.

A three-step interlaboratory validation of alternative methods to the Draize eye irritation test (Draize test) was conducted by the co-operation of 27 organizations including national research institutes, universities, cosmetic industries, kit suppliers and others. Twelve alternative methods were evaluated using 38 cosmetic ingredients and isotonic sodium chloride solution. Draize tests were conducted according to the OECD guidelines using the same lot of test substances as was evaluated in the alternative tests. Results were as follows. (1) Variation in Draize scores was large near the critical range (maximal average Draize total scores (MAS)=15-50) for the evaluation of cosmetic ingredients. (2) Interlaboratory variation was relatively small for the alternative tests. The mean coefficients of variation (CV%) were less than 50 for all assays except for the hen's egg-chorioallantoic membrane test (HET-CAM), chorioallantoic membrane-trypan blue staining test (CAM-TB) and haemoglobin denaturation test (HD). The CV% of these three methods came into the same range as the other tests when non-irritants were excluded from the data analysis. (3) Results for acids (pH of 10% solution <2.5), alkalis (pH of 10% solution >11.5) and alcohols (lower mono-ol) in cytotoxicity tests clearly deviated from the other samples in the comparison of cytotoxicity with Draize results. (4) Pearson's correlation coefficients (r) between results from cytotoxicity tests using serum and MAS were -0.86 to -0.92 for samples excluding acids, alkalis and alcohols. (5) When the samples were divided into liquids and powders, r of CAM-TB increased from 0.71 for all samples to 0.80 and 0.92, respectively. (6) Spearman's rank correlation coefficients between the results of alternative methods and MAS were relatively high (r>0.8) in the case of HET-CAM and CAM-TB. Those for cytotoxicity tests were high if the data for acids, alkalis and alcohols were excluded (SIRC-CVS: r=0.945, SIRC-NRU: r=0.931, HeLa-MTT: r=0.926, CHL-CVS: r=0.880). Exclusion of data for powdered samples also increased the coefficient of HET-CAM and CAM-TB to 0.831 and 0.863, respectively. These results suggest that no single method can constitute an evaluation system applicable to all types of test substances by itself. However, several methods will be useful for the prediction of eye irritation potential of cosmetic ingredients if they are used with clear understanding of the characteristics of those methods.