Comparative evaluation of five in vitro tests for assessing the eye irritation potential of hair-care products

This study compared five methods, the isolated rabbit eye (IRE), bovine cor neal opacity and permeability (BCOP), EpiOcular(TM), fluorescein leakage (F L) and neutral red release (NRR) assays, for predicting the eye irritation potential of hair-care formulations. Ten shampoo and seven conditioner form ulations of known ocular irritation potential were tested. Each group inclu ded a market-acceptable formulation as a comparative benchmark. Predictions of ocular irritation were made by using classification models (IRE, BCOP a nd EpiOcular assays) or by direct comparison with benchmarks (IRE, EpiOcula r, FL and NRR assays). The BCOP assay was less sensitive than the IRE test in discriminating between formulations of different irritation potentials, and did not perform as well as the other assays in identifying mild formula tions. All of the assays appeared to be better at discriminating correctly between the shampoos than between the conditioners, The EpiOcular assay sho wed the closest concordance between the in vivo results and the in vitro da ta from cell-based assays (particularly for shampoos). The FL assay also sh owed a high concordance (particularly for conditioners). There was a tenden cy for these in vitro assays to over-predict eye irritation potential, but there was no under-prediction and they were particularly successful at iden tifying mild formulations. The NRR assay was less predictive with both sham poos and conditioners. The results from this comparative evaluation fully s upport the continued use of the IRE test as a suitable alternative to in vi vo eye irritation testing in rabbits, although it also over-predicted the i rritancies of several of the formulations. The value of using concurrent be nchmarks (reference standards), appropriate to the materials being tested, in interpreting the data obtained from in vitro tests, was also demonstrate d. Overall, the results indicate that further comparisons of the IRE, EpiOc ular and FL assays are warranted using much larger numbers of test material s.