Besides other aspects it could help to distinguish compound-specific wash-in effects from barrier-disruption related effects. In contrast to the recommendation of the OECD-Guideline we decided against 3H-sucrose as ISTD because of poor information about applicability and the set limit value of 5% absorption (Walters et al., 1997). Moreover, the very high hydrophilic compound GSI-IX sucrose is not representative for routinely tested lipophilic test compounds. In accordance with the above-mentioned ‘applicability domain’ for integrity tests, the ISTD should be selected on the basis of the physico-chemical properties of the test compound, to indicate representatively the barrier function

in relation to the respective pathway through the skin. Another suggested reference compound for ISTD is phenol red. Yet a 100 times higher concentration of phenol red is needed to achieve the same analytical sensitivity as the 3H-labeled reference compounds and high concentrations increase the risk to influence the test results (Dugard and Scott, 1986). To get a first impression of the performance of different ISTDs, 3H-caffeine and 3H-mannitol were tested in parallel to 3H-testosterone in human skin experiments. The combination ABT263 3H-testosterone and 14C-MCPA resulted in moderate and weak correlations (R2 0.52 and 0.16 for AD and

maxKp comparison, respectively). This is probably due to the divergent physico-chemical properties (logP 3.32 and −0.71 (at pH 7) and MW 288.4 and 200.6 g mol−1 for testosterone and MCPA, respectively), but also due to the narrow absorption range which was covered. In fact, once the absorption range was expanded, as done in the special investigation with damaged and undamaged rat skin, the correlation was improved (R2 0.859 and 0.911 against AD and maxKp, respectively). Weak correlations were obtained with 3H-mannitol as ISTD with 14C-testosterone (R2 0.34 and 0.14 for

AD and maxKp comparison, respectively) and 14C-caffeine (R2 0.20 and 0.40 for AD and maxKp comparison, respectively). Also in this case, the distance of the logP values for the very polar ISTD 3H-mannitol and the rather lipophilic test compounds was probably too large. For the combination 14C-testosterone and 3H-caffeine, having closer logP values, the best correlations over with human skin were obtained (R2 0.62 and 0.81 for AD and maxKp comparison, respectively). However, the reverse case (3H-testosterone and 14C-caffeine) resulted in weaker correlations (0.59 for maxKp comparison) and even no correlation (R2 0.04 for AD comparison) – probably due to a lower number of replicates (n = 5) and one obvious outlier. Summing up, an ISTD with close physico-chemical properties to the test compound is preferable; however, the results imply that also ISTDs with a certain distance to the test compound are applicable. Finally, the suitability of the current ISTD approach was proven by the independence of 14C-analytics by LSC in the presence of 3H (Fig.