BackgroundRisk substances in cosmetics have long been associated with adverse reactions. However, as risk substances become more concealed and diversified, conventional targeted analysis methods are no longer sufficient to meet regulatory requirements. ObjectiveTo construct a rapid and effective non-targeted screening method for the identification of risk substances, and to provide a high-throughput method for toxicity assessment. MethodsMolecular networking was utilized for the non-targeted screening of risk substances in facial skincare products, and the toxicity of these risk substances was evaluated through molecular docking and quantitative structure-activity relationship (QSAR) models. ResultsThrough molecular networking, we identified seven known prohibited ingredients, six of which were confirmed using standard substances. In addition, 17 potential risk substances were discovered within molecular clusters containing prohibited ingredients, including antibiotics, antihistamines, and phthalates, etc. Notably, chloramphenicol base and N-desmethyl chlorpheniramine exhibited stronger binding affinity to keratin 5/14 than chloramphenicol and chlorpheniramine through molecular docking, respectively. Additionally, toxicity prediction results indicated that the carcinogenicity of antibiotics presented gender differences in mice and rats, and two chlorpheniramine derivatives also showed carcinogenicity in mice. Moreover, of the 24 compounds, 11 showed skin sensitization, while 14 caused skin irritation. Furthermore, half of these compounds demonstrated potential developmental toxicity, and only 4-nitrobenzenethiol was found to be mutagenic. ConclusionIn this study, we developed a visualization strategy for non-targeted screening of risk substances and a high-throughput method for initial toxicity assessment of risk substances.
Read full abstract