Joint poster with L’Oreal: Point of Departure for risk assessment
Improved Confidence of Quantitative Sensitizing Potency Assessment for Point of Departure Using GARD®skin Dose-Response
Presented at SOT and Eurotox 2024
Conclusion
In conclusion, the readouts from GARDskin Dose-Response derive a quantitative continuous potency estimate of skin sensitizers that may be used directly as a PoD for a seamless integration into downstream NGRA.
Abstract
Identification of skin sensitization hazard and potency characterization are central aspects of risk assessment of chemicals. Current legislation advocates a transition from hazard assessment using in vivo methods to UN GHS potency subclassification and quantitative risk assessment by use of New Approach Methodologies (NAM:s) as well as Defined Approaches (DA). However, the ability of NAM assays to quantitatively estimate sensitizing potency and thereby establish a point of departure (POD) for next-generation risk assessment (NGRA) strategies is currently lacking.
To this end, the GARDskin Dose-Response (DR) method, adapted from the OECD TG 442E method GARDskin, was recently introduced. The GARDskin DR method evaluates test chemicals in a titrated range of concentrations, in order to investigate the dose-response relationship between the output from the GARDskin prediction algorithm (Decision Values; DV:s) and test chemical concentration. The combined information can be used to derive a quantitative estimation of sensitizing potency, defined as the cDV0-value, i.e, the least required dose required to elicit a positive response by the prediction model.
The current work focuses on optimizing the ability of GARDskin DR to derive a quantitative POD based on conversion to a composite Potency Value (PV; µg/cm2), taking into account both human and in vivo reference data sources. A total of 25 chemicals were used to construct predictive regression models fitted to reference PV:s. Results show that the updated models fitted to reference PV:s produced more accurate potency predictions compared with models fitted with, and aiming to predict, only LLNA EC3 and NOEL, respectively. Mean fold-change errors ranged between 2.8 and 3.2, with predicted POD:s being within or close to the range of the variation of the historical in vivo data. In addition, uncertainty in predictions was reduced, as estimated by a minimum 2-fold reduction of 95%-confidence intervals, when comparing models fitted to reference PV:s with models fitter with only LLNA EC3 and human NOEL, respectively.
In conclusion, these improvements constitute a major step forward for the ability of NAM:s to assess quantitative sensitizing potency. It demonstrates how GARDskin Dose-Response can accurately estimate a POD and be incorporated into downstream strategies for quantitative risk assessment (QRA), to ultimately contribute to the assessment of safe use levels of chemicals.
Keywords: NAM, GARDskin Dose-Response, Sensitizing potency, Quantitative risk assessment, Point of departure
