Science Inventory

Encoding Exposure Pathways as Hierarchical Binary Descriptors for Chemical Source Identification in Non-Targeted Analysis

Citation:

Isaacs, K., J. Sobus, C. Ring, J. Wambaugh, AND K. Phillips. Encoding Exposure Pathways as Hierarchical Binary Descriptors for Chemical Source Identification in Non-Targeted Analysis. SOT, Nashville, TN, March 19 - 23, 2023. https://doi.org/10.23645/epacomptox.22272214

Impact/Purpose:

This SOT abstract describes new work to encode exposure pathways as descriptors that can be analyzed to support identification of exposure sources in non-targeted analysis.

Description:

New non-targeted analysis (NTA) technologies and supporting data analysis platforms have proven increasingly effective at identifying emerging contaminants in environmental and biological samples. Once contaminants are identified, effective mitigation strategies rely on the ability to trace the point(s) of origin, or “exposure sources”, for those contaminants. To date, no methods have been developed that would allow the rapid characterization of exposure sources for chemicals identified in NTA studies. Given this need, a fingerprinting approach was developed that encodes available exposure pathway information from existing U.S. Environmental Protection Agency (EPA) Office of Research and Development databases as hierarchical binary descriptor sets. These descriptors can be queried for enrichment using lists of chemicals identified in NTA studies to provide evidence of exposure sources. In total, 475 descriptors were developed based on existing chemical use and environmental monitoring information. These descriptors represented occurrence of chemicals in broad commercial use sectors (e.g., consumer use), industrial categories (e.g., specific utilities or manufacturing types), consumer product categories (e.g., personal care), chemical functional use categories (e.g., solvent), and environmental media. Fingerprints, representing aggregations of binary descriptors, could be generated for over 26,000 chemical structures in EPA’s DSSTox database. As a proof-of-concept for linking chemicals to exposure sources, an enrichment analysis was applied to chemicals tentatively identified in pooled blood samples from women (the selected chemicals of interest were those not also found in pooled samples from men). A Fisher’s Exact test was used to identify 44 descriptors that were enriched for chemicals in the women’s blood. These descriptors spanned a variety of consumer product and chemical function categories. An undirected chemical co-occurrence network analysis helped identify related patterns in chemical use across the enriched descriptors. Specifically, a co-occurrence threshold of more than four chemicals helped elucidate patterns of chemical use as a fragrance, deodorizer, or solvent in cosmetics (e.g., lip color, lip balm, eyeliner), hair care products (e.g., shampoo and styling products), and laundry products. Given these early findings, fingerprinting and network analysis approaches show strong potential to rapidly inform exposure sources for chemicals identified in NTA studies. Complementary approaches that define directed relationships among exposure pathway components, or define a priori related descriptor sets, are also being investigated. Understanding the origin of exposure sources for chemicals sharing common mechanisms of toxicity will potentially inform mitigation priorities and risk management strategies.