Résumé
Background: Lung disease is variable among patients with cystic fibrosis (CF) and depends on genetic and environmental factors. To better understand the molecular determinants of lung disease variability, we carried out an epigenome-wide association study (EWAS) in sputum samples from patients with CF.Methods: We profiled 64 sputum samples using Human Methylation EPIC BeadChips and assessed the correlation between DNA methylation levels and four clinical traits: lung function (FEV1pp), lung function variation (FEV1pp slope), presence and number of pulmonary exacerbations. Sputum samples were collected at four time points over an 18-month follow-up period. Selected CpG sites were reassessed in independent sputum samples from the same cohort by pyrosequencing.Results: In the EWAS, we identified two differentially methylated CpG sites (cg11047325/SOCS3, p = 4 × 10-6; cg18608055/SBNO2, p = 6 × 10-7) that correlated with lung function. They were validated in independent sputum samples from the same cohort using pyrosequencing. Additionally, three CpG sites (cg23107754, cg03209812 and cg09600088) split patients with declining lung function from those whose lung function either improved or remained stable (accuracy = 0.80). Of interest for CF-related diabetes, one of these CpG sites (cg09600088) maps to the BRSK2 gene, which plays a role in pancreatic beta cell function. Finally, a DNA methylation signature of 23 CpG sites predicted patients with pulmonary exacerbation (accuracy = 0.84).Conclusions: We provide the first longitudinal assessment of genome-wide DNA methylation in a cohort of patient with CF and identify CpG sites that predict clinical traits of key importance for lung disease. The associated genes play a critical role in inflammation or pancreatic endocrine activity. Overall, our results underscore the emerging role of DNA methylation as a key modulator of disease severity in CF.