Associations of prenatal per- and polyfluoroalkyl substances with whole blood folate levels in pregnant women in the Health Outcomes and Measures of the Environment (HOME) Study

Abstract

Background: Folate plays a critical role during pregnancy, preventing neural tube defects and possibly adverse neurodevelopment. Per- and polyfluoroalkyl substances (PFAS) are synthetic chemicals that may decrease folate levels. Although some studies have found associations between PFAS and folate, we are unaware of studies conducted in pregnant women. To address this knowledge gap, we evaluated associations between PFAS and whole blood folate (WBF) in pregnant women.

Methods: We used data from 288 pregnant women in the Health Outcomes and Measures of the Environment (HOME) Study, a pregnancy and birth cohort in the Cincinnati Ohio area. We measured eight serum PFAS and WBF concentrations at 16 weeks' gestation. We used linear regression to estimate the effect of each PFAS on WBF, and quantile-based g-computation and Bayesian kernel machine regression (BKMR) to investigate the joint effect of PFAS on WBF, adjusting for parity, prenatal vitamin intake, maternal race/ethnicity, household income, maternal age, and second trimester smoking status in all models. In addition, we investigated interactions between PFAS using BKMR.

Results: We did not observe inverse associations of individual PFAS or their mixture with WBF, nor interactions between PFAS in the BKMR model in pregnant women.

Conclusion: Future studies could consider WBF measures in late pregnancy to evaluate other periods of susceptibility. Furthermore, as people are exposed to multiple PFAS, future studies should continue to consider joint PFAS exposure.

Keywords: Folate; Mixture analysis; Per- and polyfluoroalkyl substances; Pregnancy.

Figure 1.

Serum PFAS concentrations (ng/mL) in pregnant women at 16 weeks’ gestation in the HOME Study (2003–2006), N = 288. HOME indicates Health Outcomes and Measures of the Environment Study; IQR, interquartile range; MeFOSAA, 2-(N-Methyl-perfluorooctane sulfonamido) acetic acid; PFAS, per- and polyfluoroalkyl substances; PFHxS, perfluorohexane sulfonic acid; PFNA, perfluorononanoic acid; PFOA, perfluorooctanoic acid; PFOS, perfluorooctane sulfonic acid.

Figure 2.

Adjusted difference in WBF (nmol/L) with higher serum PFAS concentrations among pregnant women (16 weeks’ gestation) in the HOME Study (2003–2006), N = 288. PFAS concentrations are log2-transformed and concentrations below the LOD are imputed with LOD/√2. All models are linear regression models, adjusted for parity, prenatal vitamin intake, maternal race, annual household income, maternal age, and smoking status in this analysis. EtFOSAA indicates 2-(N-ethyl-perfluorooctane sulfonamido) acetic acid; HOME, Health Outcomes and Measures of the Environment Study; MeFOSAA, 2-(N-methyl-perfluorooctane sulfonamido) acetic acid; PFAS, per- and polyfluoroalkyl substances; PFDA, perfluorodecanoic acid; PFHxS, perfluorohexane sulfonic acid; PFNA, perfluorononanoic acid; PFOA, perfluorooctanoic acid; PFOS, perfluorooctane sulfonic acid; WBF, whole blood folate.

Figure 3.

Adjusted differences in WBF levels (nmol/L) by PFAS mixture concentration quantile (25th–75th percentile) among pregnant women (16 weeks’ gestation) in the HOME Study (2003–2006) using BKMR, N = 288. The BKMR model evaluated a mixture of 5 PFAS (PFOS, PFOA, PFNA, PFHxS, and MeFOSAA) on whole blood folate levels, adjusted for parity, prenatal vitamin intake, maternal race/ethnicity, household income, maternal age, and log 2 maternal cotinine. The model used a Gaussian kernel, Markov Chain Monte Carlo algorithm, and ran 50,000 iterations. BKMR indicates Bayesian kernel machine regression; HOME, Health Outcomes and Measures of the Environment Study; PFAS, per- and polyfluoroalkyl substances; WBF, whole blood folate.

Figure 4.

Adjusted associations of each PFAS serum concentrations with WBF levels (nmol/L), with other PFAS concentrations fixed at the 10th, 50th, and 90th percentiles, in pregnant women (16 weeks’ gestation) in the HOME Study (2003–2006) using BKMR, N = 288. The BKMR model evaluated a mixture of 5 PFAS (PFOS, PFOA, PFNA, PFHxS, and MeFOSAA) on whole blood folate levels, adjusted for parity, prenatal vitamin intake, maternal race/ethnicity, household income, maternal age, and log 2 maternal cotinine. The model used a Gaussian kernel, Markov Chain Monte Carlo algorithm, and ran 50,000 iterations. BKMR indicates Bayesian kernel machine regression; HOME, Health Outcomes and Measures of the Environment Study; MeFOSAA, 2-(N-methyl-perfluorooctane sulfonamido) acetic acid; PFAS, per- and polyfluoroalkyl substances; PFHxS, perfluorohexane sulfonic acid; PFNA, perfluorononanoic acid; PFOA, perfluorooctanoic acid; PFOS, perfluorooctane sulfonic acid; WBF, whole blood folate.