Fragrance allergens: Classification and ranking by QSAR

Area of Science:

• Toxicology
• Computational Chemistry
• Dermatology

Background:

• Quantitative structure-activity relationship (QSAR) models are crucial for predicting chemical properties.
• Assessing skin penetration and cell-mediated immunity is vital for identifying potential allergens.
• Fragrance chemicals require rigorous safety evaluations due to common use.

Purpose of the Study:

• To validate existing QSAR models for predicting skin penetration and cell-mediated immunity.
• To evaluate the discriminating and grading power of QSAR models on known fragrance allergens.
• To test the performance of classification and rank models for allergen identification.

Main Methods:

• Validation of QSAR models using a test set of 74 fragrance chemicals (allergens and non-allergens).
• Classification test based on human experience; rank model tested with human and guinea pig data.
• Performance metrics included sensitivity, specificity, and concordance.

Main Results:

• The classification model achieved 90% sensitivity and 100% specificity on 62 compounds, with 92% concordance.
• The rank model correctly graded 88% of compounds, showing 95% sensitivity for allergens.
• Combined models demonstrated 93% overall concordance on the test set.

Conclusions:

• Validated QSAR models demonstrate significant potential for predicting skin sensitization and classifying fragrance chemicals.
• The models show high accuracy in identifying allergens, aiding in risk assessment and product safety.
• Further refinement could improve classification of indeterminate compounds and enhance grading accuracy.