Therefore methods that were considered to show potential for prediction of skin sensitisation potency and that use gene regulation or proteomics as biomarkers (GARD, SensiDerm™, Sens-IS, SenCeeTox and VITOSENS) were also selected. In addition, the PPRA as a potential improvement of the DPRA was prioritised. Cosmetic Europe’s Skin Tolerance Task Force is developing a data integration approach for the skin sensitisation safety assessment of cosmetic ingredients. This requires a non-animal testing strategy, which Epigenetics inhibitor delivers skin sensitisation
potency predictions. It is of utmost importance that the strategy is developed in a way that ensures all stakeholders will have a high level of confidence in the produced results. Confidence will be built by (a) incorporating current mechanistic understanding – guided by the OECD AOP, (b) the amount and quality of data used in strategy construction, (c) transparent and objective strategy
composition (Jaworska and Hoffmann, 2010) and (d) satisfactory predictive performance. It will need to offer flexibility to adjust to specific purposes, e.g. for cases requiring only hazard identification not potency estimation, and demands (including applicability domain issues). Therefore, we envisage that the term ‘strategy’ is used here to collectively describe an array of testing and data integration approaches. It is planned that a default or standard strategy for potency prediction will be developed, that is intended for cases without any relevant a priori information on the substance to be tested. In other cases where a priori information exists, or the purpose is not potency estimation, modifications to this check details default and/or specifically tailored strategies will be available. A-priori information may include (i) physico-chemical properties, including molecular weight, the octanol–water partition coefficient
and physical form at room temperature, As examples, approaches to confirm the expectation of a substance being a non-sensitiser or approaches specially suited for lipophilic substances (which may be difficult to test Tolmetin in an aqueous, cell line based assay) are likely to be required. The testing strategy is expected to provide an ordinal resolution of the potency spectrum preferably distinguishing five categories (non, weak, moderate, strong, extreme). The references for assessing the strategy’s performance will be human (six categories) as proposed by Basketter et al. (2014) and LLNA (EC3, categorised in 5 classes) data. EC3 values will be harvested from existing publications and qualify for inclusion only if certain criteria, including the specification of the vehicle, test concentrations and stimulation indices, are fulfilled. Variability associated with replicate EC3 values will be taken into account. To reach this aim of developing a non-animal testing strategy for potency prediction, three phases frame our efforts (Fig. 2).