Office of Research and Development Publications

Application of computational and high-throughput in vitro screening for prioritization (ECHA workshop)

Citation:

Judson, R. Application of computational and high-throughput in vitro screening for prioritization (ECHA workshop). ECHA Workshop and Meeting, Helsinki, FINLAND, April 20, 2016. https://doi.org/10.23645/epacomptox.5071450

Impact/Purpose:

Presentation at ECHA Workshop and Meeting in Helsinki, Finland on combined use of in vitro and QSAR models for screening and prioritization for estrogenic endocrine disruption.

Description:

Abstract: There are tens of thousands of man-made chemicals to which humans are exposed, but only a fraction of these have the extensive in vivo toxicity data used in most traditional risk assessments. This lack of data, coupled with concerns about testing costs and animal use, are driving the development of new methods for assessing the risk of toxicity. These methods include the use of in vitro high-throughput screening assays and computational models. This talk will review a variety of high-throughput, non-animal methods being used at the U.S. EPA to screen chemicals for their potential to be endocrine disruptors as part of the Endocrine Disruptor Screening Program (EDSP). These methods all start with the use of in vitro assays, e.g. for activity against the estrogen and androgen receptors (ER and AR) and targets in the steroidogenesis and thyroid signaling pathways. Because all individual assays are subject to a variety of noise processes and technology-specific assay artefacts, we have developed methods to create consensus predictions from multiple assays against the same target. The goal of these models is to both robustly predict in vivo activity, and also to provide quantitative estimates of uncertainty. This talk will describe these models, and how they are validated against both in vitro and in vivo reference chemicals. The U.S. EPA has deemed the in vitro ER model results to be of high enough accuracy to be used as a substitute for the current EDSP Tier 1 in vivo uterotrophic assay. Issues with developing these reference chemical sets will also be discussed – this is a critical component in the validation of any new method. Because the data on which reference chemical activity is based is itself variable, consistent and objective approaches to develop reference chemical sets need to be developed. Given the validated models for endocrine-related endpoints, we have also worked with a broad set of collaborators to develop multiple QSAR models of ER activity using the consensus in vitro model predictions as training data. We find that, just as for in vitro assays, the performance of a consensus of QSAR models is more accurate than that of the individual models. By using a combination of these in vitro and QSAR models, we have been able to make first-order predictions of the estrogenicity potential for the large majority of the EDSP universe (beyond even what is practical for testing using the in vitro methods), and are currently developing similar methods for the other endocrine-related pathways. Disclaimer: This abstract does not necessarily reflect U.S. EPA policy.