Abstract
Prenatal exposure to polychlorinated biphenyls (PCBs) has a detrimental effect on early cognitive development. Based on these observations, some researchers suggested that prenatal exposure to PCB may be an environmental cause of autism spectrum disorder (ASD). To investigate the potential link between prenatal exposure to PCB, we analyzed the link between the level of prenatal PCB exposure and ASD risk evaluated at 18 months of age and behavioral problems at 5 years old based on longitudinal birth cohort data collected in urban areas in Japan based on the data from 115 mother-infant pairs. Logistic regression analysis revealed a significant association between ASD risk at 18 months of age and the factor scores of the principal components (PCB PCs) obtained by compressing the exposure level to PCB congeners. There was no reliable relationship between PCB PCs and problematic behaviors at 5 years of age. Furthermore, machine learning-based analysis showed the possibility that, when the information of the pattern of infants' spontaneous bodily motion, a potential marker of ASD risk, was used as the predictors together, prenatal PCB exposure levels predict ASD risk at 18 months of age. Together, these findings support the view that prenatal exposure to PCBs is associated with the later emergence of autistic behaviors and indicate the predictability of ASD risk based on the information available at the neonatal stage.