Prenatal exposure to perfluoroalkyl substances predicts multimodal brain structural and functional outcomes in children aged 5 years: a birth cohort study

Abstract

<p><strong>Background:</strong> Perfluoroalkyl and polyfluoroalkyl substances (PFAS) are ubiquitous persistent organic pollutants associated with adverse health outcomes in humans. Although they are associated with autism spectrum disorder and behavioural outcomes, whether PFAS affect brain development is unclear. We aimed to characterise the relationship between maternal PFAS and brain structure and function in typically developing children.</p><p><strong>Methods:</strong> This study was set within the FinnBrain Birth Cohort Study, a prospective observational study that enrolled mothers from three clinics in Turku, Finland, during their first trimester of pregnancy. Maternal serum samples at gestational week 24 were analysed for PFAS by mass spectrometry and, at age 5 years, children were assessed through structural, diffusion-weighted, and functional MRI. Whole-brain voxel-level and vertex-level maps of grey matter volume, white matter fractional anisotropy and mean diffusivity, and cortical thickness and surface area were combined to compute ten independent components. Data were analysed by correlation network, elastic net regression, and multivariate linear regression with multiple testing correction.</p><p><strong>Findings:</strong> Pregnant mothers were enrolled into the birth cohort study between Dec 1, 2011, and April 30, 2015, and study visits at age 5 years took place between Oct 1, 2017, and March 31, 2020. This analysis involved 51 mother-child dyads for whom maternal PFAS concentrations and structural MRI data for the child were available. PFAS concentrations in maternal serum samples were mostly 0-1 ng/mL. Maternal perfluorononanoic acid (PFNA; R2=0·13, β=0·39 [95% CI 0·09-0·69], padj=0·024) and linear perfluorooctanoic acid (PFOA; 0·13, 0·36 [0·09-0·63], padj=0·022) linearly predicted a multimodal component dominated by corpus callosal integrity, whereas branched perfluorohexanesulphonic acid (PFHxS; R2=0·12, β=0·31, padj=0·036) and branched PFOA (R2=0·14, β=0·36, padj=0·016) predicted a component comprising mainly occipital cortex volume and surface area. Branched perfluorooctanesulphonic acid predicted hypothalamic microstructure (R2=0·10, β=0·29, p=0·026). PFNA, linear PFOA, and branched PFOA are associated with increased functional connectivity in the right precentral gyrus, whereas branched PFHxS predicts decreased connectivity in the intracalcerine cortices. Associations were not influenced by sex assigned at birth, but were related to PFAS chemical structure.</p><p><strong>Interpretation:</strong> We show an association between prenatal PFAS exposure and brain developmental outcomes in children. These findings are pertinent given the ubiquitous circulation of PFAS in humans and the extreme environmental persistence of these substances.</p>