Associations of early life per- and polyfluoroalkyl substances (PFAS) exposure with body mass index and risk of overweight or obesity at age 2-18 years: Mixture analysis in the prospective Boston Birth Cohort

Abstract

Background: Per- and polyfluoroalkyl substances (PFAS) are a class of widespread persistent chemicals, which may have obesogenic effects during the fetal period. This study investigated the long-term association of maternal plasma PFAS concentrations at delivery and their mixture with child body mass index (BMI) and the risk of Overweight or Obesity (OWO) at the age of 2-18 years.

Methods: The study included 1189 mother-child dyads from the prospective Boston Birth Cohort. Eight PFAS were measured in maternal plasma samples collected 24-72 h after delivery. Outcomes were BMI Z-score and OWO status of children at 2-18 years. The exposure-outcome associations were evaluated with linear and modified Poisson mixed-effects regression for individual PFAS and Bayesian kernel machine regression and quantile-based g-computation models for PFAS mixture. We explored the effect modification by maternal pre-pregnancy OWO, child age, sex, and race.

Results: Maternal plasma samples had PFAS detection frequencies from 87 % to 100 % and geometric means ranging from 0.11 to 3.67 ng/mL. PFHpS and PFHxS were associated with higher child BMI Z-score. Such associations were stronger in children aged 6-12 years and 13-18 years than in 2-5 years. Stratified by maternal pre-pregnancy OWO, significant associations of the PFAS mixture with child BMI Z-score were only found in children of non-OWO mothers. In children aged 13-18 years, children with both high maternal plasma PFDeA, PFNA, and PFOA concentrations and maternal OWO had the highest risks of OWO compared to children with either only. Such synergistic effects were not found in younger children.

Conclusions: Early life exposure to individual PFAS and their mixture were associated with a higher risk of childhood OWO, with stronger associations observed in older child age groups and in children of non-OWO mothers. Synergistic effects of PFAS exposures and maternal pre-pregnancy OWO became evident in adolescence.

Keywords: BMI; Childhood overweight or obesity; DOHaD theory; Maternal overweight or obesity; PFAS; Prenatal exposure.

Fig. 1.

Association of maternal plasma PFAS with repeated child BMI Z-score at the age of 2–18 years examined by linear and modified Poisson mixed-effects models among 1,189 mother–child dyads in the Boston Birth Cohort. Note: Abbreviations: PFAS – per- and polyfluoroalkyl substances; BMI – body mass index; CI – confidence interval; Me-PFOSA-AcOH – 2-(N-Methyl-perfluorooctane sulfonamido) acetic acid; PFDeA – perfluorodecanoic acid; PFHpS – perfluoroheptanesulfonic acid; PFHxS – perfluorohexanesulfonic acid; PFNA – perfluorononanoic acid; PFOA – perfluorooctanoic acid; PFOS – perfluorooctanesulfonic acid; PFUnA – perfluoroundecanoic acid. Models were adjusted for maternal age, education, race, maternal pre-pregnancy BMI, diabetes, child’s sex, birthweight-for-gestational age (ie. AGA, SGA, LGA), year of birth, delivery mode, and parity. The estimates and 95 % CI of associations of PFAS with child BMI Z-score and child OWO are reported in Table S2.

Fig. 2.

Association of maternal plasma PFAS mixture with child BMI Z-score and child OWO status at the age of 2–18 years examined by Bayesian kernel machine regression and quantile-base g-computation models among 1,189 mother–child dyads in the Boston Birth Cohort. Note: The points in BKMR figures mean the estimates of outcome as all the PFAS exposure are fixed at the same quantiles (from 10th to 90th) compared to when all PFAS are held at their medians, and the vertical lines mean the 95 % confidence interval of the estimate. The figures of overall models suggest increased child BMI Z-score and risk of child OWO as cumulative levels across all PFAS exposures increase. Abbreviations: PFAS – per- and polyfluoroalkyl substances; BMI – body mass index; OWO – overweight or obesity. Models were adjusted for maternal age, education, race, maternal pre-pregnancy BMI (only in overall models), diabetes, child’s sex, birthweight-for-gestational age (ie. AGA, SGA, LGA), year of birth, delivery mode, and parity. The estimates, 95 % CI, and weights of each exposure in quantile-based g-computation models are reported in Table S4.

Fig. 3.

Joint association stratified by child age group of maternal plasma PFAS and maternal OWO status with child BMI Z-score and child OWO status among 1,189 mother–child dyads in the Boston Birth Cohort. Note: Abbreviations: PFAS – per- and polyfluoroalkyl substances; BMI – body mass index; CI – confidence interval; Me-PFOSA-AcOH – 2-(N-Methyl-perfluorooctane sulfonamido) acetic acid; PFDeA – perfluorodecanoic acid; PFHpS – perfluoroheptanesulfonic acid; PFHxS – perfluorohexanesulfonic acid; PFNA – perfluorononanoic acid; PFOA – perfluorooctanoic acid; PFOS – perfluorooctanesulfonic acid; PFUnA – perfluoroundecanoic acid. Models were adjusted for maternal age, education, race, diabetes, child’s sex, birthweight-for-gestational age (ie. AGA, SGA, LGA), year of birth, delivery mode, and parity. The reference group is children of non-OWO mother and low PFAS exposures. Low PFAS exposures were defined as maternal plasma PFAS concentrations in Quartile 1–3, and high PFAS exposures were defined as maternal plasma PFAS concentrations in Quartile 4. We classified low/high as Quartile 1–3/Quartile 4 because we observed similar estimates of associations in Quartile 1–3 in Table S6. The estimates and 95 % CI of associations presented in this figure are reported in Table S8.