Cohort composition and sequencing data

A total of 35 subjects were enrolled, as outlined in Table 1. The NCFB cohort included 10 individuals, eight of whom had available matched sputum and nasopharyngeal samples. The CF cohort comprised 13 members, nine of whom had paired samples of sputum and nasopharyngeal swabs. None of the included CF patients were undergoing highly effective CFTR modulator therapy at the time of recruitment. In the healthy cohort, nine subjects had paired sputum and swab samples. Overall, of the 61 processed samples, 52 were paired, originating from 26 patients, while nine samples were unpaired. All samples were utilized for cohort comparisons regarding the upper or lower respiratory tract microbiome. However, for correlation assessments between the upper and lower respiratory tracts, only the 52 paired samples were used. Statistical analysis did not reveal any significant age differences among the cohorts (p value = 0.12). In total, 19 subjects were using antimicrobial agents continuously, with seven of them belonging to the NCFB cohort and 12 to the CF cohort (Table 1).

Table 1 Composition of the cohorts

Clinical data extracted from the patients’ records are provided in Table 2. Most patients in both disease cohorts had chronic Pseudomonas aeruginosa colonization. The antibiotic regimen included oral azithromycin and inhaled colistimethate, gentamicin, and tobramycin, representing a macrolide, a polymyxin, and two aminoglycosides, respectively, in addition to the oral antifungal agent itraconazole. Three individuals (be_05, cf_07, and cf_10) exhibited antimicrobial resistance profiles, as assessed using classical microbiological techniques (Table 2).

Table 2 Clinically relevant data of the NCFB (coded as be) and CF (coded as cf) cohorts

Overall, total DNA from 31 sputum and 30 nasopharyngeal samples was subjected to high-throughput sequencing. Following trimming, approximately 0.73% of the reads were eliminated. Human reads constituted nearly 99% of the total reads. Despite the high number of human reads, the sequencing coverage was sufficient to reach the plateau on the rarefaction curves, demonstrating adequate capture of the microbial community and resistance marker diversity in the samples (Figure S1). The data related to each sample can be found in supplementary Table_S1.

Cystic fibrosis and non-cystic fibrosis bronchiectasis influence microbiota diversity in sputum

To understand the impact of CF and NCFB on microbiota, we conducted taxonomic profiling of the samples. The major microbial component of the lungs and nasopharynx was bacteria, representing 48% of cleaned reads. Viruses accounted for approximately 0.35% of cleaned reads, fungi for 0.05%, and archaea for 0.0001%. Detailed information is presented in supplementary Table_S2, Table_S3, and Table_S4.

To assess the influence of CF and NCFB on microbial composition, a diversity evaluation was performed. A comparison of Shannon’s index among the three cohorts revealed a diminished alpha diversity in both pathological cohorts compared to the healthy one in sputum samples, which represent the lower airways (Fig. 1A). However, this difference was not observed in nasopharyngeal samples, which represent the upper airways (Fig. 1B). The same pattern was observed when analyzing other alpha diversity indices such as Simpson and evenness (Figure S2).

Fig. 1

Diversity analysis in NCFB and CF patients and healthy subjects. A Alpha diversity in sputum microbiota. Sputum barplot illustrating Shannon diversity as a measure of alpha diversity. Healthy subjects display higher alpha diversity compared to individuals with CF and NCFB. B Alpha diversity in nasopharyngeal swab microbiota. Nasopharyngeal swab barplot illustrating Shannon diversity as a measure of alpha-diversity. No significant differences were observed among the three cohorts. C Interpersonal variation in sputum microbiota. Sputum barplot representing intra-cohort Bray–Curtis dissimilarity. The points denote pairwise Bray–Curtis distance combinations, revealing increased interpersonal variation within the CF cohort. D Interpersonal variation in nasopharyngeal swab microbiota. Nasopharyngeal swab barplot representing intra-cohort Bray–Curtis dissimilarity. The points denote pairwise Bray–Curtis distance combinations, revealing increased interpersonal variation within the CF cohort. E Sputum microbiota cluster analysis. PCoA of Bray–Curtis distances in sputum microbiota, showing the distinct separation of healthy subjects from disease cohorts. While the CF cohort exhibits less clustering than the NCFB group, there is a noticeable overlap between the two disease cohorts. F Nasopharyngeal swab microbiota cluster analysis. PCoA of Bray–Curtis distances in nasopharyngeal swab microbiota, showing clustered distribution among all three cohorts. Ellipses denote 95% confidence. Statistical significance was assessed using the Kruskal–Wallis test followed by Dunn’s post hoc test for all performed analyses as none assumed a normal distribution (* P ≤ 0.05, ** P ≤ 0.01, and **** P ≤ 0.0001)

To investigate the heterogeneity within cohorts, we compared subjects within the same cohort using the Bray–Curtis dissimilarity index. The CF cohort exhibited a significantly higher mean Bray–Curtis index compared to both NCFB and healthy cohorts (Fig. 1C and D), indicating a more substantial dissimilarity among individuals affected by CF. This pattern was observed in both sputum and nasopharyngeal swab samples.

PCoA, coupled with PERMANOVA comparisons (beta diversity), was applied to evaluate potential differences in microbiome community composition among CF, NCFB, and healthy subjects. Sputum samples exhibited noticeable stratification between the healthy and pathological cohorts, as shown in Fig. 1E. Healthy and NCFB subjects displayed a clustered distribution. In contrast, the CF cohort exhibited a lack of homogeneity, indicated by higher dispersion in the PCoA. Statistical differences were assessed using PERMANOVA, revealing that both CF and NCFB groups significantly differed from the healthy cohort (adjusted p values of 0.0003 and 0.0087, respectively). There was no significant difference between the CF and NCFB cohorts (adjusted p value = 0.1167). In contrast, the PCoA of nasopharyngeal samples indicated dispersed distribution regardless of the cohort, as shown in Fig. 1F. Individuals with CF also displayed a heterogeneous distribution, indicating a higher degree of dispersion. Subsequent PERMANOVA analysis confirmed that the nasopharyngeal composition of all three cohorts was not substantially different.

Microbiota composition and diversity in sputum samples: contrasting impact of cystic fibrosis and non-cystic fibrosis bronchiectasis

To elucidate disparities in diversity across the three cohorts, a comprehensive compositional analysis was conducted. This analysis involved the computation of the relative abundance of each sample and cohort, followed by the generation of phylum-level plots for both sputum and nasopharyngeal specimens. The lower respiratory tract samples exhibited more pronounced distinctions (Fig. 2A), while the nasopharyngeal samples demonstrated a higher degree of compositional consistency across all three cohorts, as illustrated in Fig. 2B.

Fig. 2

Phylum-level composition and core microbiota. A Sputum microbiota: Phylum-level composition of sputum microbiota in NCFB and healthy cohorts. Thinner bars represent individual subjects, while thicker bars represent the cohort average. B Nasopharyngeal swab microbiota: Phylum-level composition of nasopharyngeal swab microbiota as shown in (A). C and D Core microbiota at the genus level in the sputum (C) and nasopharyngeal swab (D) samples shown as Euler diagrams depicting unique and shared components

To achieve a more precise compositional evaluation of the microbiome among the cohorts, the core components of the microbiota at the genus level were determined, considering a prevalence of 99.9% and an abundance of at least 0.1% (Fig. 2C and D). The phylum-level distinctions were assessed using normalized read counts for each phylum as the basis for comparison (Fig. 3). We selected prevalent phyla for our comparative analysis, including Actinobacteria, Proteobacteria, Fusobacteria, Firmicutes, and Bacteroidetes, as well as the Firmicutes to Bacteroidetes (F/B) ratio.

Fig. 3

Phylum composition comparison. A-F Sputum analysis: Comparison of phylum-level composition between NCFB, CF, and healthy cohorts. G-L Nasopharyngeal swab analysis: Comparison of phylum-level composition in nasopharyngeal swabs for the above cohorts. Statistical significance: The Kruskal–Wallis test was applied followed by Dunn’s post hoc test, except for Bacteroidetes analysis, which used ordinary one-way ANOVA followed by Tukey’s multiple comparison test (* P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, and **** P ≤ 0.0001)

Sputum samples exhibited substantial dissimilarities in both the core composition of the microbiota (Fig. 2C) and phylum-level comparisons (Fig. 3A-F). In contrast, the core microbiota components of nasopharyngeal samples in both diseases were included in the healthy core microbiota (Fig. 2D), and phylum-level differences were observed in only a limited subset (Fig. 3G-L).

In sputum samples, the healthy cohort exhibited a core composition comprising the Actinomyces, Campylobacter, Fusobacterium, Leptotrichia, Neisseria, Prevotella, Rothia, Streptococcus, and Veillonella genera, associated with higher counts of Actinobacteria (Fig. 3A), Fusobacteria (Fig. 3C), Firmicutes (Fig. 3D), and Bacteroidetes (Fig. 3E). The CF cohort exhibited only four genera in its core microbiota: Actinomyces, also present in healthy subjects, and Burkholderia, Cutibacterium, and Rhizobium, which were exclusive to the disease cohort. The NCFB cohort displayed the Actinomyces, Burkholderia, Campylobacter, Cutibacterium, Rhizobium, Streptococcus, Capnocytophaga, Lautropia, and Treponema genera, with the last three being exclusive to the NCFB core. The F/B ratio was significantly higher in the CF cohort compared to both NCFB and healthy cohorts. The healthy and NCFB cohorts exhibited a near 1:1 F/B ratio, indicating a proportional composition (Fig. 3F). This distinction in the F/B ratio was the sole dissimilarity observed between the CF and NCFB cohorts.

The comparison of nasopharyngeal phyla and core microbiota revealed fewer disparities among the cohorts. Fusobacteria and Bacteroidetes were both enriched in the healthy cohort in comparison to the CF cohort (Fig. 3I and K). Additionally, the NCFB cohort did not show any major discernible differences from the CF and healthy cohorts.

The correlation of phyla with patients’ lung function was assessed through linear regression of forced expiratory volume in 1 second (FEV1)—an established measure of lung function—with phylum counts. None of the models demonstrated a correlation with lung function (Figure S3 and S4).

Species differential abundance across cohorts

Differential species abundance analysis was conducted through pairwise comparisons among the three cohorts. To enhance the precision of the analysis, only species exhibiting a log2FoldChange (log2FC) above 2 were considered. The volcano plots illustrating each comparison can be found in Figure S5, and the significant differential abundance data of each sample can be accessed in supplementary Table_S5, Table_S6, and Table_S7.

When the sputum samples of the healthy cohort were juxtaposed with those of the CF cohort (Fig. 4A), the species enriched in the healthy cohort predominantly comprised commonly found components of the oral microbiome, such as members of genus Prevotella. Enrichment of Ackermannviridae sp. phage was also observed in the healthy cohort. The CF cohort exhibited an array of enriched species, including known pathogens from the genera Pseudomonas, Staphylococcus, and Xanthomonas (associated with Stenotrophomonas), and numerous entries from the order Burkholderiales.

Fig. 4

Pairwise differential abundance analysis in sputum samples. A Healthy vs. CF: Differential abundance analysis comparing healthy subjects to individuals with CF. B Healthy vs. NCFB: Differential abundance analysis comparing healthy subjects to individuals with NCFB. C CF vs. NCFB: Differential abundance analysis comparing the CF and NCFB cohorts. DESeq2 was employed for the analysis, with significance defined by an adjusted p value below 0.01. The figures represent the top 15 differentially abundant species, with a log2FC of 2 for all the comparisons

Comparing the healthy and NCFB cohorts revealed a lower number of microorganisms exhibiting differential abundance relative to the comparison of the CF and healthy cohorts (Fig. 4B). The enriched species identified in healthy subjects included those from the Alloprevotella, Moraxella, Treponema, Dialister, Leptotrichia, Mogibacterium, and Olsenella genera. Conversely, NCFB subjects exhibited the enrichment of only two microorganisms: an unclassified Pseudomonas and P. aeruginosa.

When the CF and NCFB cohorts were compared (Fig. 4C), the CF cohort exhibited enrichment of bacteria belonging to the order Burkholderiales, indicating the primary distinction between them. Additionally, Staphylococcus aureus was significantly enriched in the CF group. In contrast, NCFB subjects displayed a differential abundance of Haemophilus influenzae and Prevotella shahii compared to those with CF. In the nasopharyngeal differential abundance comparison conducted among the three cohorts, only Corynebacterium propinquum displayed significant enrichment in the NCFB group in relation to the healthy cohort.

Differential functional analysis based on metagenomics data

To explore how specific microbial functions may contribute to the disease’s manifestations and outcomes, we conducted a differential functional analysis based on the microbiome profiles. Results from the Kyoto Encyclopedia of Genes and Genomes (KEGG) Ontology (KO), Clusters of Orthologous Groups (COG), and Pfam annotations were compared among cohorts. There were differences in the identified function abundance, specifically between the CF and healthy cohorts, within the KO and Pfam annotations. Three accessions appeared enriched in the healthy cohort (choline-binding repeat, leucine-rich repeat, and SusC), while a series of processes related to basic metabolism and virulence were enriched in the CF cohort (Fig. 5). A more detailed display of the functional analysis data can be found in supplementary Table_S8 and Table_S9.

Fig. 5

Pairwise functional differential abundance analysis comparing healthy and CF cohorts. A and B. Differential abundance analysis comparing functional enriched processes in the microbiome in Pfam (A) and KO (B). Green bars represent the CF cohort and blue bars represent the healthy cohort. DESeq2 was employed for the analysis, with significance defined by an adjusted p value below 0.05. The figures represent the top 15 significantly differentially abundant processes

Antimicrobial resistance genes differ across cohorts in the lower respiratory tract

Considering the impact of antimicrobial resistance on treatment failure and the spread of resistant strains, our investigation included resistome profile determination, encompassing all cohorts herein analyzed. The composition of identified antimicrobial resistance genes is presented in supplementary Table_S10 and Table_S11. The lower respiratory tract samples displayed distinct compositional variations among the three cohorts under study, a pattern that was clearly discernible through the PCoA visualization (Fig. 6A). This observed disparity was confirmed through PERMANOVA analysis. The adjusted p values for the comparisons were 0.0003 for the healthy vs. CF cohort, 0.0069 for the healthy vs. NCFB cohort, and 0.0402 for the CF vs. NCFB cohort. Among the nasopharyngeal samples, no significant differences were observed (Fig. 6B).

Fig. 6

Resistome analysis. A Sputum resistome cluster analysis: PCoA of Bray–Curtis distances in the sputum resistome, showing distinct separation of NCFB, CF, and healthy cohorts. B Nasopharyngeal swab resistome cluster analysis: PCoA of Bray–Curtis distances in the nasopharyngeal swab resistome, revealing no significant differences between cohorts. C Sputum antimicrobial resistance gene incidence: The CF cohort exhibits a higher incidence of resistance genes compared to the NCFB and healthy cohorts. D Nasopharyngeal swab antimicrobial resistance gene incidence: The CF cohort has a higher incidence of resistance genes than the healthy cohort. E Sputum antimicrobial resistance mechanism composition: NCFB and CF cohorts display increased antibiotic efflux, while the healthy cohort exhibits a homogeneous composition. F Nasopharyngeal swab antimicrobial resistance mechanism composition: NCFB, CF, and healthy cohorts show a similar distribution of resistance mechanisms. Ellipses denote 95% confidence. Statistical significance was assessed by the Kruskal–Wallis test followed by Dunn’s post hoc test (* P ≤ 0.05 and *** P ≤ 0.001)

Considering the lower respiratory tract, a significantly higher frequency of resistance genes within the CF cohort than in both the NCFB and healthy cohorts was found (Fig. 6C). This difference was also identified in nasopharyngeal samples, in which the CF cohort exhibited a significantly higher frequency of resistance genes compared to the healthy cohort. However, when compared to the NCFB cohort, no statistical significance was observed (Fig. 6D).

Distinct patterns of resistance mechanisms were discernible among the three cohorts, as depicted in Fig. 6E and F. In the lower respiratory tract, enrichment of antibiotic efflux mechanisms was evident in both disease cohorts, characterized by an uneven distribution of other mechanisms. In contrast, healthy subjects exhibited a more uniform composition of resistance mechanisms within this context (Fig. 6E). The dynamics of resistance mechanisms in nasopharyngeal samples differed from those observed in sputum samples, with a similar distribution among the three cohorts (Fig. 6F).

The vast majority of identified resistance genes in the disease cohorts were found in P. aeruginosa, as indicated in Tables S10 and S11. The variations observed in resistome analysis can be solely attributed to pathogen composition. In the case of P. aeruginosa, PERMANOVA analysis did not reveal a significant difference between disease cohorts. The top 10 most abundant resistance genes identified per cohort are displayed in Table 3. In the CF cohort, multidrug resistance was observed for Burkholderia cepacia and Burkholderia pseudomallei, while in the NCFB cohort, the resistance genes were associated with Haemophilus influenzae. It is noteworthy that Burkholderia resistance genes were exclusively found in the CF cohort, while Haemophilus resistance genes were solely identified in the NCFB cohort. The healthy cohort exhibited a more concise profile, where mel, tet(Q), and CfxA2 led to resistance to only one drug class. There was a clear alignment between resistome data and microbiological resistance determined in patients (Table 2). The data compiled from the resistome analysis related to microbial resistance in patients be_05, cf_07, and cf_10 are displayed in Table 4.

Table 3 Most abundant resistance genes in the CF, NCFB and healthy cohortsTable 4 Antimicrobial resistance genes identified.