Lung Microbiota and Metabolites Collectively Associate with Clinical Outcomes in Milder Stage Chronic Obstructive Pulmonary Disease

Abstract

Rationale: Chronic obstructive pulmonary disease (COPD) is variable in its development. Lung microbiota and metabolites collectively may impact COPD pathophysiology, but relationships to clinical outcomes in milder disease are unclear. Objectives: Identify components of the lung microbiome and metabolome collectively associated with clinical markers in milder stage COPD. Methods: We analyzed paired microbiome and metabolomic data previously characterized from bronchoalveolar lavage fluid in 137 participants in the SPIROMICS (Subpopulations and Intermediate Outcome Measures in COPD Study), or (GOLD [Global Initiative for Chronic Obstructive Lung Disease Stage 0-2). Datasets used included 1) bacterial 16S rRNA gene sequencing; 2) untargeted metabolomics of the hydrophobic fraction, largely comprising lipids; and 3) targeted metabolomics for a panel of hydrophilic compounds previously implicated in mucoinflammation. We applied an integrative approach to select features and model 14 individual clinical variables representative of known associations with COPD trajectory (lung function, symptoms, and exacerbations). Measurements and Main Results: The majority of clinical measures associated with the lung microbiome and metabolome collectively in overall models (classification accuracies, >50%, P < 0.05 vs. chance). Lower lung function, COPD diagnosis, and greater symptoms associated positively with Streptococcus, Neisseria, and Veillonella, together with compounds from several classes (glycosphingolipids, glycerophospholipids, polyamines and xanthine, an adenosine metabolite). In contrast, several Prevotella members, together with adenosine, 5'-methylthioadenosine, sialic acid, tyrosine, and glutathione, associated with better lung function, absence of COPD, or less symptoms. Significant correlations were observed between specific metabolites and bacteria (P_adj < 0.05). Conclusions: Components of the lung microbiome and metabolome in combination relate to outcome measures in milder COPD, highlighting their potential collaborative roles in disease pathogenesis.

Keywords: bronchoscopy; chronic obstructive pulmonary disease; lung function; metabolomics.

Figure 1.

Classification accuracies and the lung bacteria and untargeted metabolomics features (hydrophobic fraction) most strongly associated with each clinical outcome, as modeled in Framework 1. Results from DIABLO feature selection followed by elastic net models adjusted for age, sex, race, current smoking, inhaled corticosteroid use, and recent antibiotic use. (A) Mean out-of-sample classification accuracies. Red dashed line = 50% accuracy (random chance). Asterisks indicate mean model performance > random chance (one-sided t test). (B) Most predictive microbial and untargeted metabolomic features from adjusted elastic net models for outcomes whose classification accuracy exceeded random chance. Bacterial OTUs are displayed alphabetically, and metabolites are displayed by class membership with superclasses of interest indicated. See Table E5 for full IDs and class information. Metabolite names with >50 characters were relabeled as “Class-Name-#”. Superclasses are intended to highlight metabolite groups of interest; in particular, lipids. “Other” refers to compounds for which the superclass was unknown or the metabolite name, as displayed, provides indication of metabolite class. BDR = bronchodilator response; CAT = COPD Assessment Test; COPD = chronic obstructive pulmonary disease; FEF_25–75 = maximum midexpiratory flow; HCU+AB/S = exacerbation requiring healthcare utilization and antibiotics/steroid treatment; ns = not significant; OTU = operational taxonomic unit; SGRQ = St. George’s Respiratory Questionnaire.

Figure 2.

Correlation heatmap between lung microbiota and untargeted metabolomics features that were most strongly associated with the clinical measures. Only OTUs and metabolites having at least one significant correlation based on P_adj < 0.01 (Benjamini–Hochberg corrected) are shown. *0.01 ⩽  P_adj < 0.05, **0.001 ⩽ P_adj < 0.01, and ***P_adj < 0.001. OTU = operational taxonomic unit.

Figure 3.

Classification accuracies and the lung bacterial and metabolites from targeted metabolomics most strongly associated with each clinical outcome, as modeled in Framework 2. (A) Mean out-of-sample classification accuracies for each outcome. Asterisks indicate mean model performance > random chance (one-sided t test). (B) The most predictive features from adjusted elastic net models for outcomes whose classification accuracy exceeded random chance. (C) Correlation heatmap between lung microbiota and metabolites that were most strongly associated with the clinical measures. Only OTUs and metabolites having at least one significant correlation based on P_adj < 0.05 (Benjamini–Hochberg correction) are shown. *0.01 ⩽ P_adj < 0.05, **0.001 ⩽ P_adj < 0.01, and ***P_adj < 0.001. For definition of abbreviations, see Figure 1.

Figure 3.

Classification accuracies and the lung bacterial and metabolites from targeted metabolomics most strongly associated with each clinical outcome, as modeled in Framework 2. (A) Mean out-of-sample classification accuracies for each outcome. Asterisks indicate mean model performance > random chance (one-sided t test). (B) The most predictive features from adjusted elastic net models for outcomes whose classification accuracy exceeded random chance. (C) Correlation heatmap between lung microbiota and metabolites that were most strongly associated with the clinical measures. Only OTUs and metabolites having at least one significant correlation based on P_adj < 0.05 (Benjamini–Hochberg correction) are shown. *0.01 ⩽ P_adj < 0.05, **0.001 ⩽ P_adj < 0.01, and ***P_adj < 0.001. For definition of abbreviations, see Figure 1.

Figure 4.

Bacterial and metabolite features associated with BAL neutrophil percentages and correlations between these features per Framework 1 (A and B) or Framework 2 (C and D). *0.01 ⩽ P_adj < 0.05, **0.001 ⩽ P_adj < 0.01, and ***P_adj < 0.001.