You are here:
Applying a tiered risk-based approach to prioritizing thousands of chemicals for further evaluation: A comparison of current high throughput computational approaches
Citation:
Nicolas, C., K. Mansouri, P. McMullen, R. Clewell, M. Yoon, M. Phillips, G. Patlewicz, J. Wambaugh, AND H. Clewell. Applying a tiered risk-based approach to prioritizing thousands of chemicals for further evaluation: A comparison of current high throughput computational approaches. SOT 2018, San Antonio,TX, March 11 - 15, 2018. https://doi.org/10.23645/epacomptox.6837668
Impact/Purpose:
Abstract for a presentation at SOT annual meeting. Applying a tiered risk-based approach to prioritizing thousands of chemicals for further evaluation: A case study in exploiting computational approaches
Description:
Current computational technologies offer novel opportunities to help in the prioritization of chemicals for further evaluation. Here we present a tiered risk-based approach based on an initial triage based on the ratio between high-throughput exposure estimates and Thresholds of Toxicological Concern (TTCs), followed by automated read-across to identify suitable analogues for identifying potential endpoints and estimate potencies. To demonstrate the applicability of TTCs for the initial triage, 300+ ToxCast chemicals were processed as follows: 1) oral equivalent doses (OEDs) were calculated based on ToxCast bioactivity measurements and available metabolism data for estimating in vivo clearance, 2) TTC values were determined using the Cramer classification system, 3) OEDs and TTCs were then compared with available ExpoCast exposure estimates to determine their respective activity:exposure ratios (AERs). This evaluation demonstrated that TTCs could serve as a reasonable basis for prioritizing compounds for further evaluation. The TTC approach was then applied to the Collaborative Estrogen Receptor Activity Prediction Project (CERAPP) database, a set of ~45,000 chemicals. An in-house read-across tool was used to identify suitable analogues for the top prioritized chemicals based on available experimental and predicted data. Potential analogue identification was guided by categorizing the CERAPP compounds into structural classes (e.g., carbamates) using SMARTS. This resulted in 17 structural classes with at least 13 chemicals for which predicted and measured bioactivity data were publicly available. We also identified an optimal combination of chemical descriptors that were predictive of bioactivity across specific endpoints (e.g. estrogen receptor, skin sensitization). We then interrogated the impact that various combinations of descriptors may have on identifying regions of chemical space that might be amenable to read-across. Our endpoint specific read-across approach employed supervised machine learning for optimal analogue selection for AER-ranked compounds. This study demonstrated the utility of exploiting computational approaches as part of a tiered risk-based approach to prioritize thousands of chemicals. This abstract does not necessarily reflect U.S. EPA policy.
URLs/Downloads:
DOI: Applying a tiered risk-based approach to prioritizing thousands of chemicals for further evaluation: A comparison of current high throughput computational approaches
NICOLAS_EXPOSUREPRIORITIZATION_SOT2018_POSTER_022618_FINAL.PDF (PDF, NA pp, 2401.32 KB, about PDF)