Prenatal exposure to polycyclic aromatic hydrocarbons and cognition in early childhood.

Background: Epidemiological evidence for gestational polycyclic aromatic hydrocarbon (PAH) exposure and adverse child cognitive outcomes is mixed; little is known about critical windows of exposure.

Objective: We investigated associations between prenatal PAH exposure and child cognition in a large, multi-site study.

Methods: We included mother-child dyads from two pooled prospective pregnancy cohorts (CANDLE and TIDES, N = 1,223) in the ECHO-PATHWAYS Consortium. Seven urinary mono-hydroxylated PAH metabolites were measured in mid-pregnancy in both cohorts as well as early and late pregnancy in TIDES. Child intelligence quotient (IQ) was assessed between ages 4-6. Associations between individual PAH metabolites and IQ were estimated with multivariable linear regression. Interaction terms were used to examine effect modification by child sex and maternal obesity. We explored associations of PAH metabolite mixtures with IQ using weighted quantile sum regression. In TIDES, we averaged PAH metabolites over three periods of pregnancy and by pregnancy period to investigate associations between PAH metabolites and IQ.

Results: In the combined sample, PAH metabolites were not associated with IQ after full adjustment, nor did we observe associations with PAH mixtures. Tests of effect modification were null except for the association between 2-hydroxynaphthalene and IQ, which was negative in males (β = -0.67 [95%CI:-1.47,0.13]) and positive in females (β = 0.31 [95%CI:-0.52,1.13])(p = 0.04). In analyses across pregnancy (TIDES-only), inverse associations with IQ were observed for 2-hydroxyphenanthrene averaged across pregnancy (β = -1.28 [95%CI:-2.53,-0.03]) and in early pregnancy (β = -1.14 [95%CI:-2.00,-0.28]).

Significance: In this multi-cohort analysis, we observed limited evidence of adverse associations of early pregnancy PAHs with child IQ. Analyses in the pooled cohorts were null. However, results also indicated that utilizing more than one exposure measures across pregnancy could improve the ability to detect associations by identifying sensitive windows and improving the reliability of exposure measurement. More research with multiple timepoints of PAH assessment is warranted.