Abstract
"One-chemical-at-a-time" approaches are typically used to investigate adverse health outcomes associated with exposure to environmental chemicals during pregnancy. However, nontargeted high-resolution mass spectrometry (HRMS) approaches can instead measure complex, real-world mixtures of xenobiotics and metabolic products and may better explain mechanisms of toxicity and identify biomarkers of exposure than single-chemical approaches during pregnancy, a particularly sensitive exposure window. Here, our objective was to use a nontargeted, HRMS approach that monitors endogenous and xenobiotic compounds to identify associations of pollutant mixtures with metabolic pathways in serum of 100 pregnant women from the MARCH cohort (Michigan Archive for Research on Child Health). Xenobiotic mixtures were identified based on in-house libraries as well as a discovery-based approach, FluoroMatch 3.0, to identify nonlegacy/emerging fluorinated chemicals in these same samples. Metabolic pathways of interest were determined using MetaboAnalyst software and effect estimates on metabolic profiles were estimated for both individual contaminants and mixtures of all identified chemicals. 415 endogenous metabolites and 21 individual chemical pollutants were detected with high identification confidence and included per- and poly-fluoroalkyl substances (PFAS), phthalates, bisphenols, organophosphate esters (OPEs), and parabens. An additional 105 tentative fluorinated compounds were identified via FluoroMatch. As classes of toxicants, bisphenol and PFAS mixtures showed the greatest number of associations with metabolic pathways related to arginine, proline, glycine, serine, and threonine metabolism. When the total mixture of all 21 contaminants was modeled after covariate-adjustment, biosynthesis of unsaturated fatty acids remained (FDR < 0.05). Taken together, these findings illustrate that serum concentrations of environmental contaminants may be associated with some metabolic pathways in pregnancy and that mixture modeling identified fatty acid metabolism as a possible pathway of interest for future validation. These findings further demonstrate the potential of robust nontargeted exposome-wide approaches as tools to study and identify the mechanisms of toxicity underlying human disease.