ALTEX 2004 ;21 (3):107-14
Centre for Documentation and Evaluation of Alternative Methods to Animal Experiments (ZEBET) at the BfR (Federal Institute for Risk Assessment), D-Berlin.
Optimisation of the EpiDerm test protocol for the upcoming ECVAM validation study on in vitro skin irritation tests
An ECVAM-funded prevalidation study (PV) was conducted during 1999 and 2000 to identify in vitro tests capable of reliably distinguishing between skin irritants (I) and non-irritants (NI) according to European Union risk phrases ("R38" or no classification). The tests evaluated were EpiDermTM, EPISKINTM, PREDISKINTM, the non-perfused pig ear method, and the mouse skin integrity function test (SIFT). Whereas reproducibility of the two human skin model tests and SIFT was acceptable, none of the methods was deemed ready to enter a formal validation study due to their low predictivity. The ECVAM Skin Irritation Task Force therefore suggested improvements of protocols and prediction models for these tests. Furthermore, it was agreed that experience gained with the two human-skin models be shared, and a common protocol should be developed for EpiDerm and EPISKIN (Zuang et al., 2002). When we applied an improved EPISKIN protocol (Portes et al., 2002) to the EpiDerm model, an acceptable specificity (80%) was achieved, whereas the sensitivity (60%) was far too low. In 2003, the EPISKIN protocol was further refined by extension of the post-incubation period following chemical exposure. In the current study, we evaluated this EPISKIN refinement by applying it to EpiDerm. In addition, we developed technical improvements for the application of the test chemicals and rinsing procedure, which reduced the variability of results and increased the percentage of correct predictions. A set of twenty non-coded reference substances from the ECVAM prevalidation study phase III (Fentem et al., 2001) was tested with the final protocol in three independent runs. Both high sensitivity (80%) and high specificity (78%) were achieved, and the statistical probability of correct classifications was high, so that the test is now regarded ready for formal validation.