Me-too validation study for in vitro skin irritation test with a reconstructed human epidermis model, KeraSkin™ for OECD test guideline 439

Abstract

We conducted a me-too validation study to confirm the reproducibility, reliability, and predictive capacity of KeraSkin™ skin irritation test (SIT) as a me-too method of OECD TG 439. With 20 reference chemicals, within-laboratory reproducibility (WLR) of KeraSkin™ SIT in the decision of irritant or non-irritant was 100%, 100%, and 95% while between-laboratory reproducibility (BLR) was 100%, which met the criteria of performance standard (PS, WLR≥90%, BLR≥80%). WLR and BLR were further confirmed with intra-class correlation (ICC, coefficients >0.950). WLR and BLR in raw data (viability) were also shown with a scatter plot and Bland-Altman plot. Comparison with existing VRMs with Bland-Altman plot, ICC and kappa statistics confirmed the compatibility of KeraSkin™ SIT with OECD TG 439. The predictive capacity of KeraSkin™ SIT was estimated with 20 reference chemicals (the sensitivity of 98.9%, the specificity of 70%, and the accuracy of 84.4%) and additional 46 chemicals (for 66 chemicals [20 + 46 chemicals, the sensitivity, specificity and accuracy: 95.2%, 82.2% and 86.4%]). The receiver operating characteristic (ROC) analysis suggested a potential improvement of the predictive capacity, especially sensitivity, when changing cut-off (50% → 60-75%). Collectively, the me-too validation study demonstrated that KeraSkin™ SIT can be a new me-too method for OECD TG 439.

Keywords: KeraSkin(TM); OECD TG 439; Predictive capacity; Reconstructed human epidermis; Reproducibility; Skin irritation.