Citation:

Nelms, M., C. Mellor, M. Angrish, B. Chorley, S. Enoch, J. Madden, J. Simmons, M. Cronin, AND S. Edwards. Using Adverse Outcome Pathways to Build Chemical Groups: A Case Study for Hepatic Steatosis. SOT Annual Meeting, Baltimore, MD, March 12 - 16, 2017.

Impact/Purpose:

Adverse Outcome Pathways (AOPs) can inform risk assessment involving chemical mixtures by providing a structured description of the mechanisms underlying the chemical toxicity. A first step in this process is defining chemical groups based on specific molecular initiating events and shared structural features. In the short term, this information can provide additional confidence in decisions regarding whether a dose addition or independent action assumption should be made. In the longer term, the AOP can be used to identify intermediate key events as surrogates for the adverse outcome in cases where independent action is required. The AOP can also be used to identify key events that could be used to evaluate dose addition assumptions where the confidence in that choice is not sufficient for the decision context.

Description:

The Adverse Outcome Pathway (AOP) framework systematically documents the mechanisms underlying effects of chemicals. Ideally, the AOP traces the mechanism to the initial interaction of chemicals with the biological system. Thus, AOPs should help inform chemical grouping by identifying chemicals that interact with a common biological molecule. Establishing such groups will aid both toxicological evaluation and risk assessment of chemical mixtures. Here, we developed mechanism-based chemical groups by combining high-throughput toxicity data (HTT) and structural alerts based on an AOP network for hepatic steatosis (Angrish et al 2016). We extracted HTT data from ToxCast for nine nuclear receptors (NRs) associated with steatosis: AHR, ER, FXR, GR, LXR, PPAR, PXR, RAR, RXR. We identified chemicals that were active in at least one out of the suite of assays for each individual NR (assay suites ranged from 2 to 20). The number of chemicals active in at least one assay varied across the nine NRs with LXR and RAR having the fewest (76 actives) and ER the most (1439 actives). We profiled each list of chemicals against previously developed structural alerts relating to NR initiation of steatosis (Mellor et al 2016). Increasing the number of positive assays required to consider a chemical active revealed a positive association between the percentage of chemicals that triggered an alert for the specific NR (e.g. chemicals active in ER assays triggered an ER alert) for five of the nine NRs (ER, FXR, GR, PPAR, PXR). Assuming this is due to a reduction in false positives, an iterative process of assay filtering followed by structural evaluation identified the optimal balance between group size and false positives for each NR. Existing alerts for three of the nine NRs (LXR, RAR, and RXR) only identified between 0-1% of the chemicals that were active in at least one NR-specific assay highlighting the need to expand the chemical space of alerts for these NRs. This study shows the value of combining information garnered from HTT assays with chemical structure information held in structure-activity relationships such as structural alerts to identify chemical groupings. [This abstract does not reflect the views or policies of the EPA.]

Record Details: