When developing new cosmetics, it is extremely important to consider the safety of consumers. Absence of potential irritancy is generally assessed using an OECD TG439 compliant Reconstructed Human Epidermis (RHE) systems and MTT assays, resulting in an irritant/not irritant classification. To gain insight into the irritancy of molecules/finished cosmetic products and to predict the outcome of irritation tests performed on subjects whatever their nature, we developed a test that uses skin explants and histological analysis. Results showed that this irritation test is sensitive enough to accurately and repeatably detect known irritants. If the diverse origin of the skin explants used led to variability in the histological alterations scored, the overall grading of irritancy is highly reproducible. Finally, when testing 120 non-alcoholic cosmetics of various galenic forms, comparison of data between the ex vivo irritation tests and of a 24-/48-h human patch test revealed a single false negative, very close to the limit, and a 10% false positive rate. It was not possible to calculate the sensitivity of the ex vivo irritation test; however, its specificity was 89.9% and its accuracy was 89.1%. Similar results, with a slightly higher false positive rate, were found when testing 49 alcoholic cosmetics. These values exceed the minimum requirements of OECD TG439.

在开发新化妆品时，考虑消费者的安全是极其重要的。通常使用符合 OECD TG439 标准的重建人表皮 (RHE) 系统和 MTT 分析来评估是否存在潜在刺激性，从而得出刺激性/非刺激性分类。为了深入了解分子/成品化妆品的刺激性并预测对受试者进行的刺激性测试的结果，无论其性质如何，我们开发了一种使用皮肤外植体和组织学分析的测试。结果表明，这种刺激性测试足够灵敏，可以准确且可重复地检测已知刺激物。如果使用的皮肤外植体的不同来源导致评分的组织学改变的可变性，则刺激性的整体分级是高度可重复的。最后，离体刺激试验和 24/48 小时人体斑贴试验显示单个假阴性非常接近极限，假阳性率为 10%。无法计算离体刺激试验的敏感性；然而，其特异性为 89.9%，准确率为 89.1%。在测试 49 种酒精化妆品时发现了类似的结果，但假阳性率略高。这些值超过了 OECD TG439 的最低要求。