A Volatile and Dynamic Longitudinal Microbiome Is Associated With Less Reduction in Lung Function in Adolescents With Cystic Fibrosis

Abstract

Progressive impairment in lung function caused by chronic polymicrobial airway infection remains the major cause of death in patients with cystic fibrosis (CF). Cross-sectional studies suggest an association between lung function decline and specific lung microbiome ecotypes. However, longitudinal studies on the stability of the airway microbiome are missing for adolescents with CF constituting the age group showing the highest rate of decline in lung function. In this study, we analyzed longitudinal lung function data and sputum samples collected over a period of 3 to 5 years from 12 adolescents with CF. The sputum microbiome was analyzed using 16S rRNA gene sequencing. Our results indicate that the individual course of the lung microbiome is associated with longitudinal lung function. In our cohort, patients with a dynamic, diverse microbiome showed a slower decline of lung function measured by FEV_1% predicted, whereas a more stable and less diverse lung microbiome was related to worse outcomes. Specifically, a higher abundance of the phyla Bacteroidetes and Firmicutes was linked to a better clinical outcome, while Proteobacteria were correlated with a decline in FEV_1% predicted. Our study indicates that the stability and diversity of the lung microbiome and the abundance of Bacteroidetes and Firmicutes are associated with the lung function decline and are one of the contributing factors to the disease severity.

Keywords: Pseudomonas aeruginosa; adolescent; cystic fibrosis; longitudinal study; lung microbiome; volatility analysis.

Figure 1

Microbiome composition of each sample grouped by hierarchical clustering. (A) Dendrogram representing the hierarchical clustering of the samples. (B) Colored bars showing the affiliation of the samples to a patient (pat) and the patient FEV-group (group; stable or decliner). The most abundant genera are written above the colored bars to name the clusters. Pse, Pseudomonas; Staph, Staphylococcus; Neis, Neisseria; Achr, Achromobacter; Fu, Fusobacteria; Pr, Prevotella; E/S, Escherichia/Shigella; Bord, Bordetella; Veil, Veillonella; Por, Porphyromonas; St, Stenotrophomonas; H, Haemophilus; Gem, Gemella. (C) Relative abundance of the genera, which occurs in at least one sample with a relative abundance ≥5%. All other genera are cumulated as “Others.” The letter at the bottom indicates the cluster groups.

Figure 2

Longitudinal volatility analysis and boxplot visualization of descriptive microbiome statistics grouped by the patient groups: Alpha diversity (A), relative dominance (B), and the evenness (C). The thick lines represent how the variable changes over time in the patient group while the thin lines represent each patient. *p < 0.05, NS, not significant.

Figure 3

Longitudinal feature volatility analysis on ASV level (A, C) and boxplot visualization (B, D) grouped by patient groups. The feature volatility analysis was done on ASV levels, and the boxplots are median values for each patient for all species with the respective genus name. (A, B) Shown are two examples of important commensals (Prevotella and Veillonella), which do also reveal a high global median in the feature volatility analysis. (C, D) A Pseudomonas ASV is the eighth important feature (ASV), but the first with a very high global variance indicates a high impact on the microbiome development, and it is important for distinguishing between patient groups. *p < 0.05), NS, not significant.

Figure 4

Proteobacteria are enriched for patients in the decliner group. (A) Longitudinal feature volatility analysis at the phylum level reveals the Proteobacteria as the most important feature for the microbiome progression in this study. (B) Median relative abundance for each patient of the five most important phyla grouped by the patient groups. *p < 0.05, NS, not significant.