COPD phenotypes revealed by the integrated sputum microbiome and proteome analysis in COPDMAP cohort

Lung microbiota composition has been associated with the exacerbation phenotype in COPD. We aimed to explore the interaction of sputum microbiome composition and biomarkers of host defense, and their relation to exacerbation frequency. We studied sputum samples collected in stable disease, and performed 16 S seq in 416 follow-up samples from 224 COPD patients and SOMAScan in 44 samples from 44 patients. Unsupervised clustering was used to identify lung microbiome profiles and linear modeling for association analysis. Three major lung microbiome clusters (C) were identified: C1 characterized with high relative abundance of Haemophilus, C2 with high abundance of Veillonella and Prevotella, and C3 dominated by Moraxella. C2 showed a decreased exacerbation frequency than the other clusters, and the overall high abundance of Haemophilus and low Veillonella associated with a high risk of exacerbations. High abundance of Haemophilus was associated with increased sputum p38-alpha (MAPK14), Calcineurin and CRP, while high abundance of Veillonella and Prevotella associated with low levels of these proteins. Sputum a1-antichymotrypsin showed an inverse association with the abundance of Haemophilus. Veillonella and Prevotella colonization is a marker of decreased risk of exacerbations. Haemophilus colonization is associated with increased risk of exacerbations and is characterized by a local inflammatory response and low a1-antichymotrypsin suggesting a novel role for a serine protease inhibitor as a potential modulator of Haemophilus colonization and microbiome composition in COPD.