Computational identification of preservatives with potential neuronal cytotoxicity

Area of Science:

• Cosmetic Science
• Toxicology
• Computational Chemistry

Background:

• Cosmetic preservatives prevent contamination but some may pose neurotoxic risks.
• Assessing chemical neurotoxicity often relies on cytotoxicity assays, which can be resource-intensive.
• Computational methods offer an efficient alternative for prioritizing chemicals for neurotoxicity screening.

Purpose of the Study:

• To develop and validate a Quantitative Structure-Activity Relationships (QSAR) model for identifying potential neurotoxicants among cosmetic preservatives.
• To prioritize preservatives for further experimental neurotoxicity evaluation.

Main Methods:

• A robust QSAR model was constructed using 681 chemicals with oversampling and Random Forest algorithms.
• The model's predictive performance was assessed on an independent test set of 452 chemicals.
• In vitro experiments were conducted to validate the neurotoxicity of selected identified preservatives.

Main Results:

• The QSAR model demonstrated high accuracy (87.7%) within its applicability domain.
• Application of the model to 157 cosmetic preservatives identified 15 chemicals as potentially neurotoxic.
• Three of the identified preservatives were further validated through in vitro experiments, confirming potential neuronal cytotoxicity.

Conclusions:

• The developed QSAR model is a valuable tool for predicting the neurotoxicity of cosmetic preservatives.
• Further experimental investigation is warranted for the 15 identified preservatives to fully assess their neurotoxic potential.
• This approach aids in the safer selection and use of cosmetic ingredients.