Comparative Analysis of Test Methods for Predicting Dermal Sensitization Potency in Essential Oils: A Component-Based Prediction Model, GARDskin Dose–Response Assay, and Local Lymph Node Assay

The dose–response adaptation of the genomic allergen rapid detection for skin sensitizers (GARDskin DR) assay is a non-animal method for predicting sensitization potency. This proof-of-concept comparative analysis evaluates outputs from a component-based prediction model (CP), GARDskin DR, and historical local lymph node assays (LLNA), exploring the applicability of CP and GARDskin DR for evaluating the no-expected-sensitization induction level (NESIL) potencies of essential oils. Eight well-studied essential oils were selected: cedarwood, cinnamon bark, clove bud, geranium, lavender, lemongrass, spearmint, and tea tree oils. Each underwent GARDskin DR as well as GC–MS, from which CP model outputs were derived. GARDskin DR, CP, and LLNA outputs were compared graphically and using preliminary descriptive statistics. Results show overall statistical concordance in material ranking between methods (Kendall’s W = 0.857, p = 0.017). Bland–Altman log-transformed bias and limits of agreement show numerically higher NESILCP compared to corresponding NESILGARD and NESILLLNA, but sensitivity analysis shows this finding is statistically sensitive to geranium oil. Spearman’s rank correlations were significant between NESILCP vs. NESILGARD and NESILLLNA but not between NESILGARD vs. NESILLLNA despite numerical similarity. Given the small sample, no statistical findings from this study are generalizable and should be interpreted descriptively. However, the NESIL results provide preliminary evidence supporting further evaluation of GARDskin DR and CP for dermal sensitization potency testing of essential oils.