Development of prediction models for three in vitro embryotoxicity tests in an ECVAM validation study

Since 1997 the National Center for Documentation and Evaluation of Alternative Methods to Animal Experiments, ZEBET, in Berlin, has been coordinating a validation study aimed at prevalidation and validation of three in vitro embryotoxicity tests, funded by the European Center for the Validation of Alternative Methods (ECVAM) at the Joint Research Center (JRC, Ispra, Italy). The tests use the cultivation of postimplantation rat whole embryos (WEC test), cultures of primary limb bud cells of rat embryos (micromass or, MM, test), and cultures of a pluripotent mouse embryonic stem cell line (embryonic stem cell test or EST). Each of the tests was performed in four laboratories under blind conditions. In the preliminary phase of the validation study 6 out of 20 test chemicals comprising different embryotoxic potential (non, weakly, and strongly embryotoxic) were tested. The results were used to define biostatistically based prediction models (PMs) to identify the embryotoxic potential of test chemicals for the WEC test and the MM test. The PMs developed with the results of the preliminary phase of the validation study (training set) will be evaluated with the results of the remaining 14 test chemicals (definitive phase) by the end of the study. In addition, the existing, improved PM (iPM) for the EST, which had been defined previously, was evaluated using the results of the preliminary phase of this study. Applying the iPM of the EST to the results of this study, in 79% of the experiments, chemicals were classified correctly according to the embryotoxic potential defined by in vivo testing. For the MM and the WEC test, the PMs developed during the preliminary phase of this validation study provided 81% (MM test) and 72% (WEC test) correct classifications. Because the PM of the WEC test took into account only parameters of growth and development, but not cytotoxicity data, a second PM (PM2) was developed for the WEC test by incorporating cytotoxicity data of the differentiated mouse fibroblast cell line 3T3, which was derived from the EST. This approach, which has previously never been used, resulted in an increase to 84% correct classifications in the WEC test.