Inflammatory Molecular Endotypes of Bronchiectasis Are Related to Microbiome Structure and Diversity: A European Multicenter Cohort Study

Background: Bronchiectasis is a heterogeneous disease requiring the development of biomarkers to guide precision medicine. Inflammation and infection are key drivers of disease but few studies have integrated host and microbiome data to define endotypes. This study aimed to identify clusters of patients based on airway and serum inflammatory markers and to assess the relationship between inflammatory endotypes and microbiome characteristics in bronchiectasis. Methods: A prospective cohort of 209 patients with stable bronchiectasis was enrolled from three European centers in the United Kingdom, Italy, and Spain as part of the EMBARC-BRIDGE study(NCT03791086). A k-means cluster analysis was performed to stratify patients according to levels of 33 airway and serum inflammatory markers. DNA was extracted from the sputum, and the V3 andV4 regions of the bacterial 16S rRNA gene was amplified and sequenced. The dominant bacterial phyla and genera were evaluated in each inflammatory endotype cluster. Microbiome characteristics were also compared between the clusters using alpha and beta diversity, and random forest. Results: Of the 209 patients, 115 (55.0%) were females, and the median age was69 years (interquartile range, 62-77 years). Four clusters of patients were defined according to the inflammatory profiles: Cluster 1 (milder neutrophilic; n = 94), Cluster 2 (mixed-neutrophilic and type2 inflammation; n = 65), Cluster 3 (most severe neutrophilic; n = 22), and Cluster 4 (mixed-epithelia land type 2 inflammation; n = 28). The sputum microbiome was more dominated by Proteobacteria in Clusters 2 and 3 than in Clusters 1 and 4 at the phyla level. Furthermore, at the genera level, Streprococcus and Haemophilus dominated Clusters 2-4 , whereas Streptococcus and Veillonelladominated Cluster 1. Lower microbiome diversity, measured using the Shannon-Wiener diversity index, was associated with more severe inflammation clusters (the lowest alpha diversity in Cluster3; P = 0.00003). Regarding beta diversity, centers of the four clusters were distinct on a principal coordinates analysis plot (PERMANOVA P = 0.001). A random forest plot revealed that Cluster 4was associated with Streptococcus, whereas Cluster 1 was associated with Veillonella, Prevotella, and Moryella. Clusters were represented in all three countries studied, and were essentially indistinguishable by clinical characteristics alone. Conclusion: Bronchiectasis inflammatory endotypes are associated with distinct microbiome profiles.