Chronic obstructive pulmonary disease (COPD) is one of the most prevalent diseases worldwide. Episodes of acute exacerbations of COPD (AECOPD) are associated with disease severity and progression. Although substantial progress has been made in understanding the dynamics of AECOPD, little is known about the sputum microbiome of AECOPD in the Chinese population.

In this study, we characterized the sputum microbiomes from sputum specimens collected from healthy controls (n = 10), stable (n = 4), AECOPD (n = 36), and recovery (n = 18) stages by sequencing the V3-V4 region of the 16S rRNA gene with a HiSeq sequencer.

Streptococcus was the most dominant genus among all the different types of sputum. A random forest model was developed to identify bacterial taxa that differentiate AECOPD samples from others. Most of the top predictors, except Pseudomonas, were less abundant in AECOPD samples. We also developed random forest models to differentiate subtypes of AECOPD based on blood eosinophil counts, the frequency of AECOPD, and sputum eosinophils. Bacterial taxa associated with Pasteurellaceae, Fusobacterium, Solobacterium, Haemophilus, Atopobium, Corynebacterium and Streptococcus, were enriched in the sputum microbiomes of eosinophilic AECOPD. Random forest models also demonstrate that a total of 2 bacterial OTUs were needed to differentiate frequent from non-frequent AECOPDs, and 23 OTUs were enough to accurately predict sputum-eosinophilic (sputum eosinophilic concentration ≥ 3%) AECOPD.

This study expanded our understanding of the sputum microbiome associated with different subtypes and clinical status of patients with AECOPD in a Chinese cohort, which provides insights into novel and more targeted management of the different subtypes of AECOPD.