Examining dietary modifiers of associations between air pollution and autism-related outcomes in two cohorts - PROJECT SUMMARY Prenatal air pollution exposure has been repeatedly identified as risk factor for autism spectrum disorder (ASD), with support for associations with particulate matter less than 2.5 microns in diameter (PM2.5), ozone, and nitrogen dioxide (NO2). Research from other fields suggests diet may be a key modulator of air pollution risks in pathways relevant to autism, yet only one study, examining the joint effects of folate and air pollution, has been published on ASD risk. However, a range of dietary factors beyond folate, including polyunsaturated fatty acids (PUFAs) and vitamin D, may serve to offset effects of exposures. In addition, given that nutrients do not act in isolation, studying single nutrient-pollutant interactions may provide only part of the picture. Further, time- windows for such modification have not been identified. In the proposed study, we will address these gaps and examine how prenatal diet may modify air pollutant associations with ASD-related outcomes in two prospective cohorts. The Nurses’ Health Study 3 (NHS3) is an ongoing, large prospective cohort of nurses from across the US that includes a pregnancy sub-cohort (n>7,000). The Early Autism Risk Longitudinal Investigation (EARLI) is a high-risk cohort that followed mothers who already had a child with ASD through a subsequent pregnancy until that child was age 3 (n~200). ASD-related outcomes will be captured in both studies according to Social Responsiveness Scale (SRS) scores, as well as ASD diagnosis, allowing us to consider both dimensional traits across the population and diagnostic-level risks. Both studies will have air pollution exposure assignments for PM2.5, NO2, and ozone from the same method and prenatal nutrient data from validated food frequency questionnaires. Using these data, our aims are to: 1) Examine modification of air pollutant-ASD associations by folate, vitamin D, and PUFAs; 2) Evaluate interactions between air pollutants and nutrients on ASD-related outcomes within a multi-exposure framework; and 3) Examine time windows in the air pollution-ASD relationship when dietary nutrients might be most impactful. Aim 1 analyses will use linear and logistic regression to examine associations with SRS scores and ASD diagnosis, respectively, within strata of nutrients defined by deficiency status and prior interaction in each cohort. In Aim 2, we will use Bayesian Kernel Machine Regression (BKMR) to consider nutrient-pollutant interactions within the context of broader diet, accounting for a wider set of nutrients and considering potential combined effects on ASD outcomes. In Aim 3, we will use distributed lag models to consider potential critical windows of air pollutant associations with ASD-related outcomes in which nutrient modifiers may have strongest effects. In this proposal we address a critical, yet understudied, area of research. Due to the widespread occurrence of air pollution exposure, and because diet is a readily accessible, individual- level modifiable factor, findings from this project present the potential for a large public health impact.
