Association between a Mixture of Per- and Polyfluoroalkyl Substances (PFAS) and Inflammatory Biomarkers in the Atlanta African American Maternal-Child Cohort

Abstract

Per- and polyfluoroalkyl substances (PFAS) have been identified as environmental contributors to adverse birth outcomes. One potential mechanistic pathway could be through PFAS-related inflammation and cytokine production. Here, we examined associations between a PFAS mixture and inflammatory biomarkers during early and late pregnancy from participants enrolled in the Atlanta African American Maternal-Child Cohort (N = 425). Serum concentrations of multiple PFAS were detected in >90% samples at 8-14 weeks gestation. Serum concentrations of interferon-γ (IFN-γ), interleukin 6 (IL-6), interleukin 10 (IL-10), tumor necrosis factor-α (TNF-α), and C-reactive protein (CRP) were measured at up to two time points (8-14 weeks and 24-30 weeks gestation). The effect of the PFAS mixture on each inflammatory biomarker was examined using quantile g-computation, Bayesian kernel machine regression (BKMR), Bayesian Weighted Sums (BWS), and weighted quantile sum (WQS) regression. Across all models, the PFAS mixture was associated with increased IFN-γ, IL-10, and TNF-α at both time points, with the strongest effects being observed at 24-30 weeks. Using quantile g-computation, increasing concentrations of a PFAS mixture were associated with a 29% (95% confidence interval = 18.0%, 40.7%) increase in TNF-α at 24-30 weeks. Similarly, using BWS, the PFAS mixture was associated with increased TNF-α at 24-30 weeks (summed effect = 0.29, 95% highest posterior density = 0.17, 0.41). The PFAS mixture was also positively associated with TNF-α at 24-30 weeks using BKMR [75th vs 50th percentile: 17.1% (95% credible interval = 7.7%, 27.4%)]. Meanwhile, PFOS was consistently the main drivers of overall mixture effect across four methods. Our findings indicated an increase in prenatal PFAS exposure is associated with an increase in multiple pro-inflammatory cytokines, potentially contributing to adverse pregnancy outcomes.

Keywords: Cytokines; Inflammatory biomarkers; Per- and polyfluoroalkyl substances (PFAS).

Figure 1.

Overall effect (95% confidence intervals) of the serum PFAS mixture (ng/mL) on serum inflammatory cytokines (mg/dL) at visit 1 (N=412) and visit 2 (N=330), estimated using quantile g-computation. Note: The mean (standard deviation) for gestational age at sample collection in weeks at visit 1 and visit 2 is 11.31 (2.16) and 26.55 (2.46), respectively. Models were adjusted for gestational age at sample collection, maternal age, education, prenatal BMI and parity. Estimates are interpreted as the percent change in each cytokine in association with a simultaneous one quartile increase in the PFAS mixture. Abbreviations: IFN-γ, Interferon gamma; IL-10, Interleukin 10; IL-6, Interleukin 6; TNF-α, Tumor necrosis factor; CRP, C-reactive protein; PFAS, perfluoroalkyl substances.

Figure 2.

Overall effect (95% credible interval) of the serum PFAS mixture (ng/mL) on serum inflammatory cytokines (mg/dL) at visit 1 (N=412) and visit 2 (N=330), estimated using BKMR. Note: The mean (standard deviation) for gestational age at sample collection in weeks at visit 1 and visit 2 is 11.31 (2.16) and 26.55 (2.46), respectively. Models were adjusted for gestational age at sample collection, maternal age, education, prenatal BMI and parity. The overall effect is interpreted as the expected percent change in each cytokine when setting all PFAS at the 25^th and 75^th percentile versus the 50^th percentile. Abbreviations: IFN-γ, Interferon gamma; IL-10, Interleukin 10; IL-6, Interleukin 6; TNF-α, Tumor necrosis factor; CRP, C-reactive protein; PFAS, perfluoroalkyl substances.