Performance Statistics of Defined Approaches for Eye Hazard Identification of Non-Surfactant Liquids to Distinguish Between the Three United Nations Globally Harmonized System Categories

Abstract

Introduction: Cosmetics Europe develops two defined approaches (DAs) for eye hazard identification, that is, addressing serious eye damage (Cat. 1), eye irritation (Cat. 2), and the absence thereof (No Cat.), for nonsurfactant liquid test substances. DAL-1 combines four physicochemical properties, the reconstructed human cornea-like epithelium (EpiOcular™ eye irritation test [EIT]: VRM1 and SkinEthic™ human corneal epithelium EIT: VRM2) test method (OECD test guideline [TG] 492), and bovine corneal opacity and permeability (BCOP) test method (OECD TG 437). DAL-2 combines the short time exposure test method (OECD TG 491) and the BCOP test method (OECD TG 437). In both DAs, the BCOP with the laser light-based opacitometer is used, however, only opacity is used to identify liquids that cause serious eye damage. Materials and Methods: The performance of the DAs to distinguish between the three United Nations Globally Harmonized System of Classification (UN GHS) categories was compared against the minimum performance values proposed by Cosmetics Europe to the Organisation for Economic Co-operation and Development expert Group on eye/skin irritation/corrosion and phototoxicity and agreed upon by the experts. Furthermore, the class-specific performance metrics are also provided for each DA. Results: The balanced accuracy of DAL-1 with validated reference method—EpiOcular EIT (VRM1), DAL-1 with validated reference method—SkinEthic HCE EIT (VRM2), and DAL-2 was 69.2%, 75.2%, and 74.3%, respectively. DAL-1 with VRM1 identified 76.5% of Cat. 1 (N = 17), 59.1% of Cat. 2 (N = 22), and 72.1% of No Cat. (N = 55) correctly. DAL-1 with VRM2 identified 76.5% of Cat. 1 (N = 17), 68.7% of Cat. 2 (N = 23), and 80.4% of No Cat. (N = 46) correctly. DAL-2 identified 81.2% of Cat. 1 (N = 17), 56.3% of Cat. 2 (N = 24), and 85.3% of No Cat. (N = 123) correctly. Conclusion: The percentage of correct predictions was greater than the proposed minimum performance values of 75% for Cat. 1, 50% for Cat. 2, and 70% for No Cat. Therefore, DAL-1 and DAL-2 have shown to successfully distinguish between the three UN GHS categories for eye hazard identification.