High Blood Eosinophil Count at Stable State is Not Associated with Airway Microbiota Distinct Profile in COPD

Introduction

Chronic obstructive pulmonary disease (COPD) is characterized by chronic respiratory symptoms including dyspnea, cough, sputum production, and exacerbations, due to abnormalities of the airways and/or alveoli that cause persistent, often progressive airflow limitation.¹ COPD is a heterogeneous disease regarding clinical features, with high morbidity and mortality. The pathogenesis of COPD is still not completely elucidated, including structural and inflammatory changes.¹ Blood eosinophil count (BEC) has been recently identified as a biomarker of interest in stable COPD. Greater BEC levels have been associated with greater FEV₁ decline in the absence of inhaled corticosteroid (ICS) treatment, and with the risk of exacerbation.^1,2 It is currently recommended to use BEC, not as a standalone biomarker but together with exacerbation history, to identify patients with the greatest likelihood of ICS treatment benefit.^1,3

Airway microbiota at stable state might participate in the heterogeneity of COPD clinical presentation. Previous studies performed in a stable state identified associations between the sputum microbiome and COPD phenotypes.^4–6 Associations with COPD inflammatory phenotypes including eosinophils have been recently investigated, suggesting that different profiles of bacteria phyla might be associated with different profiles of inflammation: low sputum eosinophil count in the presence of potential pathogenic microorganism (PPM),^7,8 more diverse respiratory microbiome in subjects with BEC ≥2%.⁹ However, the relationships between BEC and airway microbiota and its clinical implications remain to be determined.

The most frequently used techniques for airway microbiota analyses (PCR amplification, sequencing of the bacterial 16S ribosomal RNA gene) do not involve bacterial culture. We previously described an extended conventional culture-based approach, allowing the identification of bacteria restricted to viable strains.^6,10 In this study, we applied this culture technique and compared viable airway microbiota in COPD patients at a stable state with high and low BEC.

Methods

Patients

Patients with mild to severe COPD were included prospectively in the RINNOPARI cohort (Recherche et INNOvation en PAthologie Respiratoire Inflammatoire; University Hospital of Reims, France; NCT02924818) as described previously⁶ The regional ethics committee approved the study (Comité de Protection des Personnes—Dijon EST I, no. 2016-A00242-49). All patients provided written consent. Exclusion criteria were other pulmonary diseases, including asthma, bronchiectasis, cystic fibrosis, bronchopulmonary allergic aspergillosis, (CF), or pulmonary fibrosis. Stable state was defined as 4 weeks or more from the last exacerbation (defined as an acute worsening of respiratory symptoms resulting in additional therapy).^6,11 COPD was defined by postbronchodilator FEV₁/FVC <70%. The severity of COPD was determined by spirometric classification (GOLD 1: FEV₁ ≥ 80%; GOLD 2: 50% ≤ FEV₁ < 80%; GOLD 3: 30% ≤ FEV₁ < 50%; GOLD 4: FEV₁ < 30%), and ABE classification depending on exacerbations, dyspnea and CAT score.¹ Emphysema was visually assessed and quantified from CT scan images as previously described.^12,13 BEC was performed at inclusion, defining 2 groups of patients with low BEC <300/mm³ and high BEC ≥300/mm³, using the cut-off proposed by GOLD 2023.¹

Sputum Analysis

Sputa were collected, and an extended microbiological culture was performed, as previously described.⁶ After liquefaction by N-acetylcysteine, serial dilutions (from 1/1000 to 1/100,000) of the sputum were performed and cultured at 37°C for 48 h (aerobic cultures), or 5% CO₂ for 5 days (anaerobic cultures) on several agar media including Columbia blood, chocolate, Schaedler, and Pseudomonas-selective cetrimide (Thermo Fisher Scientific, USA) We next quantified all morphologically distinct colonies as colony-forming unit (CFU) per milliliter. Colonies were then identified using MALDI-TOF mass spectrometry (MALDI Biotyper^®, Bruker Daltonics, Germany). We estimated the viable airway microbiota α-diversity using the Shannon index.

Statistical Analysis

The descriptive data are expressed as numbers (percentages), median [25^th-75^th quartiles], or mean values ± standard deviation, when appropriate. Qualitative variables were compared by the chi-square test or Fisher exact test. Quantitative variables were compared using the t-test or Mann–Whitney test. A p-value <0.05 was considered significant. The dissimilarities in bacterial communities between low BEC <300/mm3 and high BEC ≥300/mm3 groups were visualised in a low-dimensional Euclidean space, using unsupervised principal component analysis (PCA), that was plotted along the first two principal components (the two explaining most of the variance).

Results

Patients

Fifty-nine subjects were included in the study, 40 in the low BEC group (BEC<300/mm³, 67.8%) and 19 in the high BEC group (BEC ≥ 300/mm³, 32.2%). Subjects’ characteristics are detailed in Table 1 and Supplementary Table 1. Briefly, they were predominantly men (57.6%), mean age 61 ± 9 years. COPD was severe to very severe in 57.6%, 66.1% had one or more exacerbations in the last year, with a median of 2 exacerbations. Two patients received long-term oral corticosteroids at inclusion, including one in low BEC group, and one in high BEC group, with no significant differences between groups (Table 1 and Table S1). When compared with the low BEC group, the high BEC group was characterised by a higher number of exacerbations per patient (median 2 [0–4] vs 1 [0–3], p<0.001), more pronounced symptoms of cough and sputum (p<0.05, CASA-Q, Table 1) and a trend for lower smoking exposure (p=0.06) and more frequent antibiotic course (p=0.06). We did not find any differences in terms of inhaled treatment, lung function, or emphysema score.

Table 1 Patients’ Characteristics

Microbiology

The extended culture method was applied to 59 sputa (1 sample per patient) to determine the viable airway microbiota. We identified 386 bacteria, from 71 different species, representing a mean of 6.6 bacteria per sample (Table S2). We compared the viable microbiota of the low and high BEC groups by PCA analysis and we found no difference in the global composition of the microbiota of the 2 groups (Figure 1A). The number of bacteria per sample and the α-diversity of the microbiota did not differ between low and high BEC groups (Figure 1B and C). The repartition of the bacterial phyla was similar in the 2 groups, with a predominance of Firmicutes (Figure 1D). We observed the same repartition of different genera: Streptococcus, Rothia, Veillonella, Neisseria, and Actinomyces were predominant and represented more than 65% of the bacteria identified (Figure 1E). The prevalence of the different species in the 2 groups was analysed and no difference was found between the low and high BEC groups (Figure 1F). The most common bacteria in both groups were Streptococcus oralis/mitis/pneumoniae, identified in more than 90% of samples, followed by Veillonella parvula/dispar/atypica found in more than 50% of samples. Bacterial quantifications of the viable microbiota ranged from 10² CFU/mL to 10⁹ CFU/mL, with a median of 10⁶ CFU/mL and no difference between the 2 groups (Table S2).

Figure 1 Bacterial analysis of airway microbiota in low and high BEC groups. (A) Principal component analysis (PCA) of the airway microbiota. The axes are the first eigenvalues, the ones explaining the most variance of the dataset. Individual patients are represented in blue triangles (high BEC group, n=19) and green dots (low BEC group, n=40). The largest blue triangle and red dot at the center of the ellipses represent the 95% confidence interval. (B) Number of species per sample. (C) Alpha diversity of viable microbiota (Shannon index). (D) Phyla distribution. (E) Genus distribution. (F) Bacteria prevalence. Note: Bacteria with less than 5% frequency are not listed.

In this cohort of COPD patients at stable state, some PPM were detected: Staphylococcus aureus (n = 8, 13.6%), Haemophilus influenzae (n = 6, 10.2%), Moraxella catarrhalis (n = 5, 8.5%) and Pseudomonas aeruginosa (n = 3, 5.1%), with the same prevalence in the 2 groups (n=16; 40.0% in the low BEC group vs n=7, 36.8% in the high BEC group) (Table S2).

Discussion

In this cross-sectional analysis, focusing on COPD patients at stable state and using an original extended conventional culture-based approach that allows the identification of bacteria restricted to viable strains, we identified that high BEC is not associated with a distinct profile of viable airway microbiota, despite a BEC-associated clinical phenotype.

In our study, the high BEC group was characterised by a clinical phenotype including more frequent exacerbations in the last year, a trend for a more frequent antibiotic course, and more pronounced cough and sputum symptoms (CASA-Q). Elevated BEC at a stable state has been previously identified as a biomarker for exacerbation risk in COPD, with an incidence rate ratio of exacerbation of 1.32 for BEC ≥300/mm³ in the COPDgene study.² An association between sputum inflammatory cells and symptoms at stable state has been shown, with neutrophilic inflammation being associated with cough, and eosinophilic inflammation with dyspnea.^14,15 A recent post-hoc analysis suggested that COPD patients with both high eosinophil levels in sputum (≥3%) and chronic bronchitis might present a distinct profile of gene expression, characterised by the overexpression of T2- and phosphodiesterase-4-inhibitors-related genes.¹⁶ The high BEC group in our study was further characterised by a trend for lower smoking exposure, but a similarly altered lung function. This might indirectly reflect the previously observed accelerated lung function decline in COPD subjects with high eosinophil counts.¹⁷

We did not identify a distinct profile of viable airway microbiota in the high BEC group when compared with the low BEC group. Firmicutes and Streptococcus were the predominant bacteria phylum and species respectively, in line with previous analyses performed in COPD at a stable state.^5,18 The previously reported decrease in Proteobacteria abundance and increase in Firmicutes phyla in subjects with high BEC,¹⁹ or the more diverse microbiome in subjects with BEC ≥2%⁹ were not found in our study. It must be pointed out that these previous studies used gene sequencing techniques (16S rRNA), which are not able to discriminate between viable and non-viable strains. Previous studies described the presence of PPM in subjects with low sputum eosinophil count.^7,8 However, these studies used qPCR detection restricted to 3 species (H. influenzae, M. catarrhalis, and S. pneumoniae), therefore also including both viable and non-viable strains. Our study did not confirm those results.

Our study suffers from several limitations. Although monocentric, the patient’s recruitment was prospective and all benefited from an in-depth phenotypic characterisation. Exacerbation frequency was estimated retrospectively in the last year before inclusion. We performed only a one-point BEC assessment at inclusion, while BEC is known to vary over time in COPD.²⁰ However, the clinical phenotype of the high BEC group including more frequent exacerbation matches with previous studies.² Inhaled treatment was heterogeneous with 32% of the patients using ICS. ICS treatment can alter the microbiome in the small airways of patients with COPD and might have an impact on our results.²¹ However, inhaled treatment strategies including ICS in our study did not significantly differ between high and low BEC groups. Finally, our extended conventional culture-based approach may have lower sensitivity than a metagenomic approach for a more in-depth characterisation of the lung microbiota.

Conclusion

In our study using an original extended conventional culture-based approach, high BEC in COPD at a stable state was associated with a clinical phenotype including frequent exacerbation and more cough and sputum symptoms, but no distinct profile of viable airway microbiota compared with low BEC.