Determining a point of departure for skin sensitization potency and quantitative risk assessment of fragrance ingredients using the GARDskin dose-response assay.

Summary

Using 100 fragrance ingredients spanning multiple protein-reactivity mechanisms, GARDskin DR provided quantitative potency predictions for skin sensitization with 81% approximate accuracy and mean 3.15-fold error versus NESIL. The assay supports next-generation risk assessment and may reduce reliance on animal tests in cosmetic safety evaluations.

Key Findings

• Tested 100 fragrance ingredients with diverse reactivity alerts (Schiff base, Michael addition, SN2, acylation) using GARDskin DR.
• Exact potency category accuracy was 37%, with 81% approximate accuracy (exact or ±1 category).
• Merging weak/very weak classes improved total accuracy to 53% and approximate accuracy to 98%.
• Mean prediction error was 3.15-fold versus NESIL and 3.36-fold versus LLNA EC3.

Clinical Implications

While not a clinical tool, the assay can guide product formulation to minimize sensitization risk, support labeling decisions, and reduce adverse reactions in consumers.

Limitations

• Exact potency category accuracy is modest (37%) and requires further refinement.
• In vitro assay; external validation against human patch test data and real-world formulations is needed.

Future Directions

Expand external validation across independent datasets, integrate with complementary NAMs for tiered potency estimation, and link predictions to human patch test databases for calibration.