Associations Between Long-Term Exposure To Air Pollution And Perturbations In The Transcriptome

Abstract

Epidemiologic studies have consistently shown associations between long-term exposure to ambient air pollution and chronic health effects such as cardiovascular disease, non-malignant respiratory disease, and lung cancer. We hypothesized that early indications of biological perturbations due to ambient air pollution could be detected in genome-wide gene expression profiles from peripheral blood mononuclear cells (PBMC) in healthy individuals. We assessed genome-wide gene expression in PBMC samples from study participants of two prospective cohorts from Italy (n=95) and Sweden (n=455). We estimated long-term air pollution (NOx) exposure levels for our study participants using exposure models developed within the European Study of Cohorts for Air Pollution Effects (ESCAPE). We used three complementary statistical approaches: univariate analysis (coupled with a 5% false discovery rate (FDR) control to account for multiple testing), elastic net regression, and Graphical Unit Evolutionary Stochastic Search (GUESS; a recent Bayesian variable selection approach), to identify for which genes expression level was altered by long-term exposure to NOx. There was little overlap between the distribution of NOx exposure levels in the Italian and the Swedish cohort (median; 5th-95th percentile: 94.2; 7.8-124.6 and 6.7; 4.8-19.5 μg/m3, respectively). We therefore conducted our primary analyses in the two cohorts separately. Using the univariate approach we identified one (unannotated) transcript in the Italian cohort and no transcripts in the Swedish cohort that were significantly associated at an FDR of 5%. While elastic net regression provided consistent results, GUESS provided less indication for an association with this probe in the Italian data. In summary, our study does not provide strong indications of transcriptomic changes related to air pollution within the exposure range studied.