Prenatal exposure to single chemicals belonging to the per- and polyfluoroalkyl substances (PFAS) family is associated with biological perturbations in the mother, fetus, and placenta, plus adverse health outcomes. Despite our knowledge that humans are exposed to multiple PFAS, the potential joint effects of PFAS on the metabolome remain largely unknown. Here, we leveraged high-resolution metabolomics to identify metabolites and metabolic pathways perturbed by exposure to a PFAS mixture during pregnancy. Targeted assessment of perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorooctanesulfonic acid (PFOS), and perfluorohexanesulfonic acid (PFHxS), along with untargeted metabolomics profiling, were conducted on nonfasting serum samples collected from pregnant African Americans at 6-17 weeks gestation. We estimated the overall mixture effect and partial effects using quantile g-computation and single-chemical effects using linear regression. All models were adjusted for maternal age, education, parity, early pregnancy body mass index, substance use, and gestational weeks at sample collection. Our analytic sample included 268 participants and was socioeconomically diverse, with the majority receiving public health insurance (78%). We observed 13.3% of the detected metabolic features were associated with the PFAS mixture (n = 1705, p < 0.05), which was more than any of the single PFAS chemicals. There was a consistent association with metabolic pathways indicative of systemic inflammation and oxidative stress (e.g., glutathione, histidine, leukotriene, linoleic acid, prostaglandins, and vitamins A, C, D, and E metabolism) across all metabolome-wide association studies. Twenty-six metabolites were validated against authenticated compounds and associated with the PFAS mixture (p < 0.05). Based on quantile g-computation weights, PFNA contributed the most to the overall mixture effect for γ-aminobutyric acid (GABA), tyrosine, and uracil. In one of the first studies of its kind, we demonstrate the feasibility and utility of using methods designed for exposure mixtures in conjunction with metabolomics to assess the potential joint effects of multiple PFAS chemicals on the human metabolome. We identified more pronounced metabolic perturbations associated with the PFAS mixture than for single PFAS chemicals. Taken together, our findings illustrate the potential for integrating environmental mixture analyses and high-throughput metabolomics to elucidate the molecular mechanisms underlying human health.
Keywords: PFAS; environmental mixtures; high-resolution metabolomics; mixture analysis; quantile G-computation.