Perinatal Air Pollution Exposure and Autism, with New Results in the Nurses' Health Study-2

Background:   Over the past several years, research into environmental toxicants as possible risk factors for autism has made steady progress.  We review this progress with respect to metals and other air pollutants, beginning with early suggestive findings from methodologically weak studies that utilized exposure measures made after diagnosis or that recruited potentially biased samples of cases and controls. We trace the improvements in exposure assessment and study design, such as the use of county level mercury releases in an ecologic study in Texas to estimate exposure independently from autism, to the use of air pollutant exposure models or distance to road as a proxy for traffic-related pollutants in individual-level studies. This emerging evidence points to a possible association between perinatal exposure to hazardous air pollutants and autism.

Objectives:   To examine the association between perinatal exposure to air pollution and risk of autism spectrum disorder in a large national cohort, building upon the advances made in previous work.

Methods:   We used logistic regression to estimate odds ratios and 95% confidence intervals (CI) for the association between U.S. Environmental Protection Agency modeled levels of hazardous air pollutants at the time and place of birth and risk of autism in the children of mothers in the Nurses’ Health Study II, which includes women from across the entire United States (cases=325; controls=22,120).  Our analyses focused on pollutants associated with autism in prior research.  We adjusted for possible ascertainment bias by both family-level socioeconomic status (e.g., income, partner’s education, maternal grandparents’ education) and census-tract-level socioeconomic measures (e.g., tract median income, percent college educated).  We also examined possible differences in the relationship between autism and pollutant exposure by child’s sex.  We ran multi-pollutant models and explored the novel statistical approach of random forests to try to identify particular toxicants responsible for associations among many correlated exposures.

Results:   Perinatal exposure to higher levels of diesel, lead, manganese, mercury, methylene chloride, and a combined measure of metals were linearly related to increased risk of autism (p<.05 for each).  These associations persisted only among boys in analyses stratified by sex.  Odds ratios for autism in boys in the highest versus lowest quintile of exposure to these pollutants were 1.72 (95% CI: 1.15-2.56) for lead, 1.52 (95% CI: 1.03-2.23) for manganese,  1.91 (1.11-3.28) for mercury, 2.23 for diesel (95% CI: 1.03-4.81), and 1.82 (95% CI: 1.19-2.79) for methylene chloride.  Multi-pollutant models suggested mercury and methylene chloride to be the most robustly associated with autism. Random forests suggested exposure to arsenic, mercury, chromium, trichloroethylene, and manganese as most important predictors of autism status.

Conclusions:   Gestational exposure to air pollution may increase risk for autism.  Exposure to metals and other air pollutants have known effects on the immune system, can dysregulate calcium processes in the brain, and can be directly neurotoxic.  Such mechanisms could underlie the association between these pollutants and autism.  Methodological challenges persist, though, in determining which toxicant(s) among many correlated ones are causal factors.