Microbial burden and viral exacerbations in a longitudinal multicenter COPD cohort

Abstract

Background: Chronic obstructive pulmonary disease (COPD) is a heterogeneous disease characterized by frequent exacerbation phenotypes independent of disease stage. Increasing evidence shows that the microbiota plays a role in disease progression and severity, but long-term and international multicenter assessment of the variations in viral and bacterial communities as drivers of exacerbations are lacking.

Methods: Two-hundred severe COPD patients from Europe and North America were followed longitudinally for 3 years. We performed nucleic acid detection for 20 respiratory viruses and 16S ribosomal RNA gene sequencing to evaluate the bacterial microbiota in 1179 sputum samples collected at stable, acute exacerbation and follow-up visits.

Results: Similar viral and bacterial taxa were found in patients from the USA compared to Bulgaria and Czech Republic but their microbiome diversity was significantly different (P < 0.001) and did not impact exacerbation rates. Virus infection was strongly associated with exacerbation events (P < 5E-20). Human rhinovirus (13.1%), coronavirus (5.1%) and influenza virus (3.6%) constitute the top viral pathogens in triggering exacerbation. Moraxella and Haemophilus were 5-fold and 1.6-fold more likely to be the dominating microbiota during an exacerbation event. Presence of Proteobacteria such as Pseudomonas or Staphylococcus amongst others, were associated with exacerbation events (OR > 0.17; P < 0.02) but more strongly associated with exacerbation frequency (OR > 0.39; P < 4E-10), as confirmed by longitudinal variations and biotyping of the bacterial microbiota, and suggesting a role of the microbiota in sensitizing the lung.

Conclusions: This study highlights bacterial taxa in lung sensitization and viral triggers in COPD exacerbations. It provides a global overview of the diverse targets for drug development and explores new microbiome analysis methods to guide future patient management applications.

Conflict of interest statement

All authors are employees and share holders of AstraZeneca.

Figures

Fig. 1

Sampling timeline and composition. (a) Timeline from Oct 2011 to May 2014, COPD patients were enrolled and sampled during scheduled wellness visits (blue arrows), and any unscheduled visits (red and purple arrows) within 3 days of an acute exacerbation or exacerbation follow-up visit. Samples were considered stable if collected 31 days post-hospitalization or ARI. Dotted arrows correspond to samples collected at scheduled wellness visits that incidentally corresponded to acute exacerbation events (1.2%) and exacerbation follow-up visits (65%). Piecharts represent the proportion of patients from Europe and USA (b), the proportion of samples collected at each disease state (c) and the proportion of samples associated with antibiotics (Abx) and inhaled corticosteroids (CS) taken in the past 7 days (d)

Fig. 2

Microbiota composition and diversity is associated with geography, not with acute exacerbation or viral infections. (a) Taxonomic barplot of major bacterial phyla and genera in samples grouped by geography and disease state, and their respective Shannon diversity index represented as boxplots with interquantile range whiskers. (b) Percentage of samples with microbiota predominant with nine most common COPD bacterial taxa, and (c) percentage of samples positive for 7 most common respiratory viruses are plotted at stable (green), acute exacerbations (red) and follow-up visits (purple), and their respective Shannon diversity index. * P < 0.01, **P < 0.001, *** P < 0.0001, lower and higher statistical significant diversity compared to the average are noted in red and blue, respectively. HMPV, Human Metapneumovirus

Fig. 3

Risk factors of COPD exacerbations. Adjusted odds ratio of (a) viral infections (b) bacterial abundance (top/bottom quartile) and (c) demographics and clinical history features to be associated with acute exacerbation events or patients with frequent exacerbations (≥2 events/ year). Significance are presented in red (positive association) and green (negative association). Orange dots represent non-significant odds ratio. Horizontal bars represent the 95% confidence interval. Genera names are in bold and species italicized

Fig. 4

Longitudinal variations of the COPD sputum microbiota at stable state is associated with higher disease burden. a Longitudinal Taxa barplot at Genus level of all 42 patients representing the top (variable) and bottom (consistent) quartile of median weighted UniFrac distances at stable states. b Proportions of patients with low or high longitudinal microbiota variability associated with geography, and boxplots with interquantile range whiskers for the frequency of exacerbations, and of viral infections

Fig. 5

Timeline of consecutive viral detection. Longitudinal viral detection in 14 patients as indicated by a cross and lines are colored according to virus species, strain or subtype. Exacerbation events are indicated by a filled circle, and frequent exacerbator phenotypes are indicated by plus or minus signs

Fig. 6

Biotyping clusters of COPD sputum microbiota at stable, acute exacerbation and follow-up visits. For each state, optimal cluster number analysis using the Calinski-Harabasz index, scatter diagram of samples clustered by between-class principal component analysis, contribution of major bacterial genera in means per class and proportion of samples belonging to each cluster plotted by region, frequency of exacerbation, frequency of of viral exacerbations, antibiotics use, corticosteroids use and CODP duration are presented at stable (a), acute exacerbated (b) and exacerbated follow-up visits (c)