Comparison of human skin irritation patch test data with in vitro skin irritation assays and animal data

Background: Efforts to replace the rabbit skin irritation test have been underway for many years, encouraged by the EU Cosmetics Directive and REACH. Recently various in vitro tests have been developed, evaluated and validated. Objective: A key difficulty in confirming the validity of in vitro methods is that animal data are scarce and of limited utility for prediction of human effects, which adversely impacts their acceptance. This study examines whether in vivo or in vitro data most accurately predicted human effects. Methods: Using the 4‐hr human patch test (HPT) we examined a number of chemicals whose EU classification of skin irritancy is known to be borderline, or where in vitro methods provided conflicting results. Results: Of the 16 chemicals classified as irritants in the rabbit, only five substances were found to be significantly irritating to human skin. Concordance of the rabbit test with the 4‐hr HPT was only 56%, whereas concordance of human epidermis models with human data was 76% (EpiDerm) and 70% (EPISKIN). Conclusions: The results confirm observations that rabbits overpredict skin effects in humans. Therefore, when validating in vitro methods, all available information, including human data, should be taken into account before making conclusions about their predictive capacity.