Background
Chronic obstructive pulmonary disease (COPD) is a heterogeneous disease characterised by airflow obstruction and persistent airway inflammation [
1]. COPD patients show increased susceptibility to bacterial infection, through mechanisms such as decreased bacterial phagocytosis [
2]. Chronic bacterial airway colonisation may occur with potentially pathogenic microorganisms (PPMs) including
Haemophilus influenzae (H. influenzae), Moraxella catarrhalis (M. catarrhalis) and
Streptococcus pneumoniae (S. pneumoniae) [
3‐
5].
The presence of bacterial infection in COPD patients during the stable state (i.e. not during exacerbations) is associated with increased airway neutrophil numbers [
6,
7]
. Furthermore
, it appears that different bacterial phyla are associated with different profiles of airway inflammation in COPD patients; raised blood and sputum eosinophils are associated with increased presence of the Bacteroidetes phylum [
8], while low sputum eosinophil levels have been associated with increased
H. influenzae presence [
9]. Additionally, a study of microbiome, transcriptome and proteome profiling showed that
H. influenzae presence in the stable state was associated with a unique profile of inflammation, including increased sputum neutrophil counts [
10,
11].
The analyses of randomised clinical trials have demonstrated that greater inhaled corticosteroid (ICS) benefits are observed at higher blood eosinophil counts [
12]. Consequently, the Global initiative for the management of chronic Obstructive Lung Disease (GOLD) report recommends the use of blood eosinophil counts to guide ICS treatment in COPD patients with a history of exacerbations [
1]. Higher blood eosinophil counts in COPD patients are associated with increased eosinophil numbers in bronchial tissue, broncho-alveolar lavage and sputum, increased levels of T2 cytokines and greater basement membrane thickening [
13]. The mechanistic reasons for increased ICS effects at higher blood eosinophil counts may be related to an inflammatory profile associated with increased levels of T2 cytokines, but an association with a different microbiome profile may also be important [
10].
Using the COPDMAP cohort, we have further studied the relationship between sputum cell counts and bacterial species in the stable state. We evaluated the stability of the relationship between the microbiome and airway inflammation using repeated samples at 6 months, primarily focusing on bacterial load measured by quantitative polymerase chain reaction (qPCR) and sputum cell counts. Associations between these parameters and both 16S rRNA-gene based microbiome analysis and blood leucocyte counts are also reported. A focus of the analysis of repeated samples was to investigate the persistence of bacterial colonisation over 6 months, and its associations with sputum cell counts and clinical characteristics.
Discussion
This analysis of the COPDMAP cohort demonstrated a subgroup of COPD patients with H. influenzae colonisation in the stable state, associated with increased neutrophil and decreased eosinophil numbers in sputum. Although colonisation status and microbiome changed over time in some patients, we observed that approximately 40% of patients within the H. influenzae group at baseline had persistent H. influenzae colonisation at 6 months, with a similar profile of airway inflammation. H. influenzae colonisation was also associated with increased blood neutrophils numbers. In contrast, there was no association between neutrophilic inflammation and S. pneumoniae colonisation. These results reveal a distinct COPD subgroup with H. influenzae colonisation that persists over time and is associated with increased neutrophilic inflammation in both the lungs and systemic circulation.
The 16S rRNA microbiome results supported the qPCR analysis, showing increased Haemophilus abundance in the H. influenzae group defined by qPCR at baseline, with no other genus enriched in this group. In contrast, a variety of other bacterial genus (not Haemophilus) were enriched in the S. pneumoniae group. Additionally, most of the > 1PPM group (92%) were colonised with H. influenzae, and there was a trend towards increased sputum neutrophil counts in this group. Overall, these qPCR and 16S rRNA results indicate that Haemophilus colonisation may occur with or without the presence of other colonising bacteria, but that the presence of H. influenzae is the key component causing the association with neutrophil counts.
H. influenzae and > 1PPM groups displayed the greatest stability over 6 months compared to
S. pneumoniae and
M. catarrhalis. It was not possible to determine if the changes in inflammatory cell counts were solely attributed to the presence of
H. influenzae or rather the persistence of airway colonisation in these patients, although there is evidence to support the former [
10]. Significant associations were demonstrated using both qPCR and 16s rRNA gene sequencing for
Haemophilus measurements at baseline and 6 months. Sputum neutrophils showed a similar pattern. Changes in airway neutrophils on an individual basis over time showed remarkable concordance with that of
H. influenzae colonisation in the subgroup (n = 69) with sputum cell counts and microbiology data at both baseline and 6 months. The presence of
H. influenzae appeared to influence the sputum neutrophil percentage, with persistently higher neutrophils being observed in those with
H. influenzae present at both visits, and concordant changes in neutrophils observed when
H. influenzae presence changed (Fig.
6). We also observed that persistent
H. influenzae colonisation was associated with a lower FEV
1 predicted, and a numerically higher exacerbation rate (although this difference was not statistically significant, possibly due to the small sample size, n = 9). The clinical phenotype relating to persistent colonisation with
H. influenzae is likely to be of important clinical relevance and warrants further investigation in larger longitudinal cohort analysis.
Previous studies have shown associations between total airway bacterial load and neutrophilic inflammation using sputum [
6,
19] and broncho-alveolar lavage [
20] samples.
H. influenzae presence in stable state sputum samples has been associated with increased sputum neutrophil counts [
21,
22] and other pro-inflammatory biomarkers [
19]. The present study confirms these observations regarding
H. influenzae, and extends our knowledge by showing that
H. influenzae persists in the stable state in a COPD subgroup, and is associated with persistent sputum neutrophilic inflammation and a skewing away from eosinophilic inflammation. Furthermore, we show an association between
H. influenzae and systemic neutrophilic inflammation. Increased systemic inflammation is associated with co-morbidities including cachexia and cardiovascular disease, highlighting the potential clinical importance of our findings [
23,
24].
Some COPD studies have shown that greater neutrophilic airway inflammation in the stable state is associated with poor clinical outcomes [
25], including FEV
1 decline [
26], while other studies have reported little convincing evidence of any relationship [
27]. Perhaps the variation between studies is due to differences between populations in the extent of bacterial colonisation with
H. influenzae which is known to increase exacerbation susceptibility, particularly when exposure to human rhinovirus occurs [
28].
Clinical trials have shown that lower sputum eosinophil numbers are associated with a reduced response to corticosteroid treatment [
29]. The clinical use of blood eosinophils to direct ICS treatment is based on the association between blood and lung eosinophil numbers [
12,
13]. Our findings implicate
H. influenzae presence as a determinant of ICS response, as this bacterium skews the airway inflammation profile away from eosinophilic inflammation which is more ICS responsive. Furthermore, it has recently been reported that lower blood eosinophil counts are associated with increased chronic bacterial infection [
30].
H. influenzae causes secretion of CXCL8 (a neutrophil chemokine) from alveolar macrophages in a corticosteroid insensitive manner in response to exposure [
31], potentially explaining the observed skewing of neutrophil relative to eosinophil numbers.
H. influenzae can persist in the lung through various evasive pathogenic features, for example by formation of biofilms and resistance to neutrophil extracellular traps [
32,
33]. The strains of
H. influenzae present in the lungs of COPD patients may change over time [
34]; our analysis does not define whether
H. influenzae persistence at both baseline and 6 months are the same or different strains. Regardless, increases in neutrophilic airway inflammation associated with
H. influenzae colonisation at baseline and 6 month visits were similar.
It has been reported that increased blood neutrophil numbers are associated with increased pneumonia risk in COPD patients [
35]. These reports are compatible with the links between infection, microbiome dysbiosis and systemic inflammation reported here. Although we report only a relative increase in systemic neutrophil levels, the neutrophil–lymphocyte ratio (NLR) has received growing interest as a marker of systemic inflammation [
36], and as a marker of severity for sepsis, pneumonia and cardiovascular diseases [
36,
37]. We found the NLR to be significantly higher in those colonised with
H. influenzae, further supporting a role for the species in driving systemic inflammation and pneumonia risk.
M. catarrhalis has been previously described as an exacerbation-related opportunistic pathogen [
5,
10]. We report a lower stable state prevalence of
M. catarrhalis (8.1%) than previous studies reporting 16% and 11% prevalence [
38,
39]. These previous studies reported greater exacerbation frequencies prior to sampling (mean/year 3.6 and 3.5 [
38,
39] compared with 1.89 in our analysis), possibly explaining the differences.
A limitation of this study is the reduced sample size during repeat sampling due to patient drop outs, exacerbations causing delayed stable visits and failure to produce a sputum sample at every visit. A large proportion of sputum samples analysed were spontaneously produced (83% and 88% at baseline and 6 months respectively). Previous studies have reported no differences in sputum cell percentages between induced and spontaneous samples [
6,
17,
40], with our results agreeing with these findings. We observed a higher total cell count, and total neutrophil and macrophage counts in induced samples. Some caution should be applied not to over-interpret these findings, as the number of induced samples was much smaller, and no difference in total cell counts has been previously reported when paired samples from the same subjects were analysed [
40]. Furthermore, this finding for total cell counts was not associated with any differences in bacteriology data. Nevertheless, our findings suggest the possibility that sputum induction results in more cellular samples compared to spontaneously obtained sputum, and this may have been a source of variation in this study [
41]. Also, a large proportion of patients reported current use of ICS. It has been reported that higher ICS doses are significantly associated with greater total airway bacterial load [
4,
7], but we did not see a difference in ICS use between groups, perhaps due to the high overall ICS use. Some patients were current smokers (34%); a previous analysis of the COPDMAP data reports no difference in microbiome between smoking and non-smoking individuals [
42]. We were unable to determine if any changes in colonisation status were affected by antibiotics or prior vaccination. Finally, our analyses did not extend to investigation of the molecular mechanisms pertaining to the relationship between
H. influenzae colonisation and neutrophilic inflammation.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.