Computational analysis of the ToxCast estrogen receptor agonist assays to predict vitellogenin induction by chemicals in male fish

Endocrine disruptors, especially estrogen receptor (ER) agonists, have received considerable research attention. While there are several mechanistic endpoints for ER agonism in the Endocrine Disruptor Screening Program, there have been growing efforts to develop high-throughput screening assays and computational models to reduce testing cost, time, and animal use. For example, there are 16 ER agonist assays and an integrated computational model in ToxCast. In the present study, we examined the relationship between ToxCast ER agonist assays and model activity to male vitellogenin induction in the Fish-Short Term Reproduction Assay. It was found 15/16 of the assays significantly predicted potency ranks for 10 common ER agonists, and 7/16 of the assays had a significant linear correlation. The integrated model also provided comparable performance to most assays. Thus, the ToxCast ER agonist assays and model may be useful to identify endocrine disruptors and predict reproductive outcomes in fish.