Childhood exposure to traffic-related air pollution and lung function at age 10 years: evidence of nutritional modification in a birth cohort.

BACKGROUND: Exposure to traffic-related air pollution has been associated with impaired lung function in children, yet heterogeneity in susceptibility related to nutritional status remains poorly understood. METHODS: We analyzed data from a Korean birth cohort with lung function measured at age 10 years. Dietary nutrient intakes were assessed at ages 6 and 8 years using food frequency questionnaires. We applied a three-step analytical framework consisting of elastic net regression to screen nutrient-NO2 interactions, LightGBM with SHAP values to explore nonlinear patterns, and conventional linear regression models for confirmation. RESULTS: Childhood exposure to nitrogen dioxide (NO2) was associated with lower forced expiratory volume in 1 s (FEV1) (β = -0.07, 95% confidence interval [CI]: -0.14, -0.003) and a lower FEV1/forced vital capacity (FVC) ratio (β = -1.70, 95% CI: -3.31, -0.09). Elastic net screening identified 4 of 33 fatty acid-related nutrients and 2 of 8 B-vitamin-related nutrients with stable interaction signals (selection frequency ≥0.60). In LightGBM-SHAP analyses, contributions of the selected nutrients to lung function varied nonlinearly across intake ranges, showing threshold, plateau, or U-shaped patterns. Only a small subset of interactions, involving eicosadienoic acid and capric acid, remained statistically significant in conventional linear regression models (p-interaction <0.05). CONCLUSIONS: Childhood NO2 exposure was associated with lower lung function at age 10 years. Nutritional status may modify this association in heterogeneous and nonlinear ways, highlighting the value of integrative analytical approaches combining machine learning and conventional epidemiological models to characterize individual susceptibility to environmental exposures.