Investigating Polychlorinated Biphenyls and Cytokines in Autism Spectrum Disorder

Background: Polychlorinated biphenyls (PCBs) are persistent organic pollutants known to adversely affect neurodevelopment. Both cytokines and PCBs have independently been hypothesized to play a role in etiologic pathways of ASD, and PCBs have been shown to alter cytokine profiles. However, no study has examined the potential for cytokines to act as an effect modifier of the association between PCBs and ASD.

Objectives: This study examined whether there is an association between prenatal exposure to PCBs and ASD in a risk-enriched pregnancy cohort. This study also assessed whether levels of selected maternal circulating cytokines altered associations between PCBs and ASD.

Methods: Data are from the Early Autism Risk Longitudinal Investigation (EARLI) study, a risk-enriched longitudinal cohort in which women who had a child with ASD were enrolled during a subsequent pregnancy and followed through the sibling’s third year of life. Based on clinical assessment at 36 months using DSM-5 criteria, 37 children were diagnosed with ASD, and 127 were non-cases and included in these analyses (n=164). Concentrations of 37 PCB congeners were measured in second trimester maternal serum samples; 29 cytokines and chemokines were measured in first available maternal plasma samples. Twelve PCB congeners (28, 74, 99, 118, 138/158, 153, 170, 180, 187, 194, 196/203, 199), a summed measure of all 12 congeners, and six cytokines (IL-1β, IL-2, IL-4, IL6, IL-8, and IFN-λ) had >60% of measures above limits of detection and were included in further analyses examining odds of ASD relative to typical development. Multivariable logistic regression analyses adjusting for maternal race and education, parity, smoking during pregnancy, and gender of child were run to assess the association between PCB congeners and ASD, parameterizing PCB concentrations in tertiles. The potential for effect modification by cytokine concentration was explored by comparing PCB levels in ASD cases and non-cases in subgroups defined by cytokine concentration above or below the median.

Results: PCB congeners 138/158 and 153 had the highest concentration in the total study population. There were no statistically significant differences in geometric means of individual prenatal PCB concentrations in ASD cases and non-cases. Adjusted associations were also not statistically significant. PCB 99 had the largest odds ratio estimate, OR=3.66, but the associated 95% C.I. was very wide (0.84-15.84) and two congeners, PCB 28 and 153, had adjusted OR point estimates below 1.0, also with wide confidence intervals. There was no suggestion of potential effect modification of PCB and ASD associations by cytokine level in our preliminary explorations.

Conclusions: The results of these analyses with prospectively collected biomarkers do not provide evidence supporting an association between prenatal PCB exposure and ASD risk. However, these results were based on only 164 subjects, and further investigation of prenatal exposure to these and other persistent organic pollutants is needed.