Accounting for seasonal variation of nasal gene expression improves transcriptome analysis of asthma

Abstract

INTRODUCTION: Environmental seasonal changes, including allergen exposure, pollutants, and viral prevalence, impact gene expression across human tissues. Therefore, season of sample collection is critical to consider when analyzing disease-associated transcriptional programs. In this study we aimed to investigate the effect of season on identification of asthma-associated genes. METHODS: We analyzed RNA-seq data from nasal brushes of asthma patients (n=361) and healthy controls (n=57) included in the ATLANTIS study. Samples were compared across two time periods defined by a data-driven approach: winter-spring (Nov 15 -May 14) and summer-autumn (May 15 - Nov 14), correcting for age, sex, smoking, and asthma status. Gene Set Variation Analysis (GSVA) scores of the obtained seasonal gene signatures were compared using Mann-Whitney U test in the independent PIAMA cohort (n=325) for replication. RESULTS: We identified 1,731 genes differentially expressed between the two time periods (813 higher and 918 lower in winter-spring compared with summer-autumn; FDR < 0.05). This signature was replicated in the PIAMA cohort nasal gene expression data and was independent of asthma or allergy status. Samples collected in winter-spring showed a significantly lower predicted proportion of ciliated cells (p = 0.005) and a higher proportion of basal cells (p = 0.04) compared to those from summer-autumn. Crucially, including season as a covariate in models for asthma-associated gene expression nearly doubled the number of differentially expressed genes, from 126 to 271 (FDR < 0.05). We conclude that season of sampling strongly affects the nasal transcriptome, and we demonstrate that season-adjusted analyses significantly enhance sensitivity for detecting disease-associated changes.