Sixteen test methods with data in common for a set of 10 substances were considered during this evaluation. With the exception of test methods developed by member companies of Cosmetics Europe (i.e. DPRA, h-CLAT, MUSST and PBMDC) that provided existing data from non-blinded testing, coded substances were tested. However, for calculation of the predictivity of most methods Selleckchem ABT888 including these four, available data on additional chemicals were considered (in most cases ⩾40 substances, Table 4), so that potential impact of coded versus non-coded testing on predictivity become marginal. With the cooperation of the test method developers, additional

information relevant to a pre-defined list of criteria that addressed a number of parameters including Selleck ZD1839 the level of standardisation, existing test data, potential for throughput, transferability and accessibility was systematically collated. The outcome of this evaluation was reviewed by each test method developer, discussed at a workshop held with the method developers, and ultimately informed the prioritisation of test methods for phase II of the evaluation process. Initially, the ten test methods DPRA,

GARD, h-CLAT, KeratinoSens™, MUSST, PPRA, SenCeeTox, SensiDerm, Sens-IS and VITOSENS were prioritised based on voting by the Cosmetics Europe member companies represented in the Skin Tolerance Task Force. At a later stage, one test method was dropped because significant optimisation would be needed, while another was stopped due to organisational issues. During phases II and III of the Cosmetics Europe framework new developments of existing or up-coming methods such as the efforts by Teunis

et al., 2013 and Teunis et al., 2014, or van der Veen et al. (2015), will be monitored and considered in case they can be expected to improve the testing strategy. The basis for the testing strategy composition will be more than 100 substances, for which both LLNA and human data are available. It is planned that test results from all eight phase II methods for all substance will be available. For each test method the data considered most useful for the testing strategy composition will Florfenicol be defined. This implies that the potential contribution of read-out parameters – instead of currently applied prediction models – to the strategy will be explored, especially for the methods that on hazard assessment. For example, for the DPRA relative cysteine and lysine depletion will be used. It has to be noted that properties of the data of the various methods differ. While the methods of first priority have a few relevant read-outs to be captured, this will be more complex for other methods, such as Sens-IS or GARD that measure an array of genes. Variability of the methods will be accounted for.