In utero PFAS exposure measured in the umbilical cord blood and child neurodevelopmental outcomes at 24 months: a prospective birth cohort study

Background Prenatal exposure to per- and polyfluoroalkyl substances (PFAS) may be associated with child neurodevelopment, but evidence based on fetal-proximal exposure biomarkers, mixture effects, and domain-specific outcomes remains limited. Methods In the Shanghai Birth Cohort ( n = 1,668 mother-child pairs), we quantified 32 PFAS compounds in umbilical cord blood and assessed neurodevelopment at 24 months using the Bayley Scales of Infant and Toddler Development, Third Edition (BSID-III). Associations were evaluated using adjusted single-pollutant regression models as the primary analyses. Developmental risk defined as BSID-III composite score < 90 was evaluated as a complementary outcome, and weighted quantile sum (WQS) regression was used to evaluate PFAS mixture associations. Results Nineteen PFAS compounds had detection rates greater than 65% and were included in association analyses. In adjusted single-pollutant models, several long-chain PFAS were inversely associated with cognitive and language scores. PFNA was associated with lower cognitive (β = −5.39, 95% CI: −7.22, − 3.57) and language scores (β = −3.62, 95% CI: −4.94, − 2.31), as well as higher odds of cognitive developmental risk defined as BSID-III score < 90 (OR = 1.95, 95% CI: 1.40, 2.71). PFUdA, PFTrDA, and PFDoA showed similar inverse associations. The emerging substitute 8:2 Cl-PFESA was also inversely associated with cognitive (β = −1.86, 95% CI: −2.52, − 1.19) and language scores (β = −1.21, 95% CI: −1.68, − 0.73). PFHxA was positively associated with cognitive (β = 1.53), language (β = 0.83), and motor scores (β = 1.15), and PFOA was positively associated with cognitive scores (β = 3.37). In negative-direction WQS regression, the PFAS mixture index was inversely associated with cognitive (β = −4.66, 95% CI: −6.51, − 2.80) and language scores (β = −3.14, 95% CI: −4.65, − 1.63), whereas associations with motor, social-emotional, and adaptive behavior scores were less consistent. No consistent associations were observed for social-emotional or adaptive behavior domains after FDR correction. Conclusion Higher cord blood concentrations of selected long-chain PFAS and 8:2 Cl-PFESA were associated with lower cognitive and language performance at 24 months. WQS regression also suggested inverse mixture associations with cognitive and language scores, but these mixture findings should be interpreted cautiously because WQS is directionally constrained and single-pollutant results were not uniformly in the same direction. These findings support continued evaluation of legacy and emerging PFAS in relation to early neurodevelopment.