Towards an integrated in vitro strategy for estrogenicity testing

In order to define an in vitro integrated testing strategy (ITS) for estrogenicity, a set of 23 reference compounds representing diverse chemical classes were tested in a series of in vitro assays including proliferation and reporter gene assays. Outcomes of these assays were combined with published results for estrogen receptor (ER) binding assays and the OECD validated BG1Luc ER transcriptional activation (TA) assay and compared with the outcomes of the in vivo uterotrophic assay to investigate which assays most accurately predict the in vivo uterotrophic effect and to identify discrepancies between the in vitro assays and the in vivo uterotrophic assay. All in vitro assays used revealed a reasonable to good correlation (R2¿=¿0.62–0.87) with the in vivo uterotrophic assay but the combination of the yeast estrogen bioassay with the U2OS ERa-CALUX assay seems most promising for an ITS for in vitro estrogenicity testing. The main outliers identified when correlating data from the different in vitro assays and the in vivo uterotrophic assay were 4-hydroxytamoxifen, testosterone and to a lesser extent apigenin, tamoxifen and kepone. Based on the modes of action possibly underlying these discrepancies it becomes evident that to further improve the ITS and ultimately replace animal testing for (anti-)estrogenic effects, the selected bioassays have to be combined with other types of in vitro assays, including for instance in vitro models for digestion, bioavailability and metabolism of the compounds under investigatio