The skin sensitization prediction model: an algorithm for real-world prediction of skin sensitization risk and minimization of human sensitization testing.

Abstract

Skin sensitization testing to ensure the safety of skincare products for public consumption has largely relied on human repeat insult patch test (HRIPT). The desire to minimize reliance on HRIPT has prompted a search for alternative methods to assess the sensitization risk of consumer products to inform decision-making about their suitability before being brought to market. The novel Skin Sensitization Prediction Model (SSPM) is a methodology that draws upon a database consisting of more than 20 years of historical HRIPT data pertaining to 1,274 unique product formulations, comprising 1,226 common ingredients, for which HRIPT testing has been performed on 203,640 human subjects. The SSPM sets modifiable thresholds for each individual ingredient of a proposed formulation and for the formulation as a whole, applying risk calculations based on dosage density, potential for skin occlusion, potential for skin barrier impairment, and potential effects on immune-primed skin. Tabulations of a formulation's risk characteristics allow for a numerical risk calculation that is compared to the preset thresholds to determine whether the product may continue its development or should be reformulated or discontinued. This methodology points to a new model for sensitization testing for a wide array of products without recourse to in vivo testing.