An Exposome-Wide Interaction Study to Identify Hidden Environmental Factors Affecting Susceptible Populations: Application to Chemical-Social Vulnerability Interactions on Cognitive Function

Background: Exposome-wide association studies (ExWAS) systematically evaluate numerous environmental exposures, but typically focus on the mean effect, determining whether the association is statistically significant for the average individual. This approach may overlook environmental chemicals that specifically affect susceptible subpopulations. We conducted an exposome-wide interaction study (ExWIS) to identify environmental chemicals associated with cognitive function in the context of social vulnerability. Methods: We examined associations between 147 biomarker-based chemicals and cognitive function among 4,982 adults aged 60 and older in the National Health and Nutrition Examination Survey (NHANES) 1999-2000 and 2011-2014. Social vulnerability index (SVI) was calculated as a composite measure encompassing race/ethnicity, income, education, health insurance, housing type, food security, and employment. Initially, we ran survey-weighted linear regressions between each chemical and cognition, adjusting for SVI, age, sex, smoking status, serum cotinine, fish consumption, NHANES cycle, and urinary creatinine (marginal association models). Subsequently, we repeated these models with a linear interaction term between each chemical and SVI (interaction models). Results: In marginal association models, 16 chemicals including smoking-related compounds, polyaromatic hydrocarbons, and metals were associated with cognitive function (unadjusted p<0.05). In interaction models, four additional chemicals—urinary mercury, diethylthiophosphate, perfluorononanoate, and perfluoroundecanoate—not identified in the marginal association models showed significant interactions with SVI. Discussion: This study demonstrates the utility of the ExWIS framework to identify environmental factors that affect susceptible populations but may be missed by the mean effect-based ExWAS approach. Considering interaction effects is essential for accurate environmental risk assessment and identification of neurotoxicants in at-risk populations.