A prevalidation study on in vitro tests for acute skin irritation. results and evaluation by the Management Team

A prevalidation study on in vitro tests for acute skin irritation was conducted during 1999 and 2000. The overall objective of validation in this area, of which this prevalidation study is an initial stage, is to identify tests capable of discriminating irritants (I) from non-irritants (NI), as defined according to European Union (EU) risk phrases ("R38"; no classification) and the harmonised OECD criteria ("Irritant"; no label). This prevalidation study specifically addressed aspects of: protocol refinement (phase I), protocol transfer (phase II), and protocol performance (phase III), in accordance with the prevalidation scheme defined by the European Centre for the Validation of Alternative Methods (ECVAM). The tests evaluated were: EpiDerm (phases I, II and III), EPISKIN (phases I, II and III), PREDISKIN (phases I and II, and additional protocol refinement), the non-perfused pig ear method (phases I and II, and additional protocol refinement), and the mouse skin integrity function test (SIFT; phases I and II). Modified, standardised test protocols and well-defined prediction models were available for each of the tests at the end of phase I. The results of phase I (intralaboratory reproducibility) were sufficiently promising for all of the tests to progress to phase II. Protocol transfer between the Lead Laboratory and Laboratory 2 was undertaken for all five tests during phase II, and additional refinements were made to the test protocols. For EpiDerm, EPISKIN and the SIFT, the intralaboratory and interlaboratory reproducibilities were acceptable; however, better standardisation of certain aspects of the test protocols was needed prior to commencing phase III. Neither PREDISKIN nor the pig ear test performed sufficiently well in phase II to progress to phase III. The PREDISKIN protocol was overly sensitive, resulting in the prediction of all the NI chemicals as I. The variability in the pig ear test results was too great, indicating that the test would show limited predictive ability. In additional studies (a repeat of phase I), further modification of the PREDISKIN protocol and a change in the prediction model considerably improved the ability of the test to distinguish I from NI chemicals. However, attempts to improve the intralaboratory reproducibility of the pig ear test were unsuccessful. In phase III an initial assessment of the reproducibility and predictive ability, in three independent laboratories per test, was undertaken for the EpiDerm and EPISKIN tests (the SIFT was a late inclusion in the prevalidation study, and is being evaluated in a separate phase III study). A set of 20 coded chemicals (10 I, 10 NI) were tested with the final, refined, test protocols. The intralaboratory reproducibility was acceptable for both EpiDerm and EPISKIN. The interlaboratory reproducibility was considered to be acceptable for EPISKIN; however, for EpiDerm, analysis of variance (ANOVA) indicated that there was a statistically significant laboratory effect on the overall variability, suggesting that the interlaboratory transferability of the test needs to be improved. The EpiDerm test had an overall accuracy of 58%, with an over-prediction rate of 37% and an under-prediction rate of 47%. The EPISKIN test had an overall accuracy of 58%, showing an under-prediction rate of 23% and an over-prediction rate of 60%. It is concluded that, as yet, none of the tests evaluated in this prevalidation study are ready for inclusion in a formal validation study on in vitro tests for acute skin irritation. Overall protocol performance of the SIFT is currently being evaluated in a phase III study. Further studies are also in progress to improve the test protocols and prediction models for EpiDerm and EPISKIN.