Comparative analysis of fish embryo acute toxicity (FET) test, fish cell line acute toxicity test and acute fish toxicity test: Evaluating the applicability domain of the FET test

Although the fish embryo acute toxicity (FET) test has been recognized as a potential alternative to the acute fish toxicity (AFT) test, further studies focusing on its applicability domain are required due to existing uncertainties. Improved prediction of the mode of action (MoA) could help to better define the applicability domain of the FET test. The Verhaar scheme is one of the currently used structure-based approaches for predicting MoA; however, it has limitations to define the MoA of chemicals. The objective of this study is to explore an approach for predicting the MoA of chemicals, with the aim of evaluating the applicability domain of the FET test. To this end, 40 chemicals, including three pesticides and seven pharmaceuticals with known biological targets, and 30 industrial chemicals without known biological targets were selected and the FET test and RTgill-W1 cell line assay were performed on these chemicals. Through the comparative analysis between the generated results and those from the AFT tests, it was demonstrated that an integrated approach that combines the results from several MoA prediction models can predict the applicability domain of the FET test more broadly than single-scheme approach through improved prediction of MoA. Furthermore, we explored a new effect-based approach for predicting MoA by comparing median lethal concentration (LC50) at the early and late developmental stages in the FET test. We found that the LC50 from 24 hr post-fertilization (24-hr LC50) in the FET test was similar to the median effect concentration from the cell line assay, indicating that 24-hr LC50 from the FET test mainly reflects basal cytotoxicity. These findings suggest that the ratio of the 24-hr LC50 to the 96-hr LC50 from the FET test may be useful for roughly predicting the MoA of chemicals, because the latter can include organ-specific toxicity, whereas the former mainly reflects basal cytotoxicity.