Background
Chronic obstructive pulmonary disease (COPD) is characterised by inflammation and irreversible airflow obstruction. Before the advent of culture-independent DNA profiling methods, the healthy lung was deemed a sterile niche while COPD samples would frequently culture
Haemophilus influenzae, Streptococcus pneumoniae, Moraxella catarrhalis and
Pseudomonas aeruginosa. It is considered that pathogenic bacteria gain a foothold in damaged airways contributing to further lung pathology through release of noxious bacterial products and provocation of host inflammation [
1,
2].
In recent years, use of high-throughput 16S rRNA gene based sequencing has demonstrated that rich, complex bacterial communities exist in the airways of both health and COPD, with overlapping bacterial composition observed [
3,
4]. In COPD aerobic, facultative and anaerobic organisms colonise the airways [
3,
5], with
Proteobacteria and
Firmicutes being the two major phyla reported in the microbiome and
Haemophilus and
Streptococcus, the respective dominant genera [
3,
4,
6,
7]. At exacerbation, shifts in bacterial composition, characterised by a relative increase in
Proteobacteria that falls in response to antibiotics has been observed, suggesting an association with the aetiology of COPD exacerbations [
8,
9]. Furthermore, the ratio of
Gammaproteobacteria to
Firmicutes identifies the sub-group with dynamic changes in their microbiome during exacerbation, suggesting a potential use of this ratio as a biomarker for targeting antimicrobial treatment [
10].
While
Proteobacteria have been associated with COPD exacerbation events, the role of the microbiome in the stable state and important differences in composition with health are unclear [
3,
4,
6]. Furthermore, changes in the microbiome that may associate with development of COPD in smokers are unclear. Differences have been shown in the nasal and oropharyngeal microbiome between smokers and non-smokers [
11] but microbiome data from healthy smokers and non-smokers is limited making the findings inconclusive to contextualize the pathological basis of the observations in COPD.
To address this, we have used sputum collected from a substantial number of well characterised healthy volunteers to investigate the impact of smoking on healthy lower airway microbiome; to explore if there are differences in microbiome between health and COPD and its association with smoking in health.
Discussion
Our study is the largest to compare the sputum microbiome between health and COPD and clear differences between these groups were identified. Firmicutes, Bacteroidetes and Actinobacteria comprised 88% of the sputum microbiome in healthy participants, with Streptococcus, Prevotella and Veillonella as the dominant genera. Haemophilus, the dominant genus in COPD was present in health at a significantly lower proportion. Healthy smokers with ≥10 PY smoking history showed a trend towards a higher ratio of Streptococcus to Prevotella.
Existing comparative respiratory microbiome data are divided, with some studies reporting an overlapping microbial composition between health and COPD [
3,
4,
26] while others have shown the COPD microbiome to be distinct [
5,
6,
27] . These discordant outcomes likely reflect the underlying heterogeneity in COPD groups and small sample sizes of healthy individuals (< 20), undermining the strength of these studies [
3‐
6,
26,
27]. However, similar to our observation, higher levels of
Proteobacteria, especially
Haemophilus, in COPD [
3‐
5] and relatively higher proportion of
Prevotella,
Veillonella and
Actinomyces species in health have been observed [
5,
6] but differences did not reach significance. Contrary to our observations, most studies have reported similar or a higher abundance of
Firmicutes and especially
Streptococcus in COPD compared with health [
4,
6]. One reason for this might be that composition varies between samples depending upon the type of treatment received, disease severity and inflammation. Previous studies have shown association between very severe COPD and eosinophilic phenotypes with dominance of
Firmicutes, while
Proteobacteria are predominant in moderate COPD and the bacterial related phenotype [
9,
28] . The COPD cohort analysed here was mainly of moderate-to-severe severity with high neutrophil counts.
Contrary to most studies, we found a higher alpha diversity in COPD compared to health [
3,
6]. Although the COPD sample reads were reanalysed with the healthy at a normalised sequence depth, they were sequenced as part of COPDMAP study which involved a much larger sample size [
13], including different disease stages, and this may have contributed higher COPD diversity. Moreover, our COPD cohort was older than the healthy group and from moderate to severe GOLD stage. Higher diversity has been associated with both increasing disease severity and age in COPD [
4,
27,
29].
Although a strong association exists between smoking and both airway inflammation and COPD, the determinants for developing COPD in smokers are not clear. We hypothesised that smoking associated microbiome changes in health would help in understanding the role of microbes in transition from health towards COPD. Consistent with previous studies, we found no significant difference between the microbiome of smokers with ≥10PY history and the < 10 PY group [
3,
4,
6,
30]. However, similar to Morris and colleagues [
30], a trend towards lower proportions of both
Bacteroidetes and
Proteobacteria in smokers with ≥10PY history was observed, suggesting subtle effects of smoking on the airway microbiome. Other pathological factors may therefore be important in shaping the microbiome in COPD. Hypoxia and chronic systemic inflammation related factors, which are features of COPD, have been reported to be associated with the airway microbiome [
31] and may be relevant to the differences observed in our COPD cohort.
PICRUSt analysis showed relatively higher lipopolysaccharide biosynthesis products in COPD. Lipopolysaccharides are present in the outer membrane of
Proteobacteria and together with pathogen-associated molecular patterns, induce strong and damaging pro-inflammatory responses. In keeping with this, our previous study showed that sputum chemokine interleukin-8, known to play a key role in COPD inflammation, is positively correlated with
Haemophilus and
Moraxella, suggesting these bacteria trigger the excessive production of this chemokine [
9]. Moreover,
Haemophilus has been implicated in a dysbiotic role by co-inclusion of its related phylotypes and depletion of
Firmicutes,
Bacteroidetes and
Actinobacteria that are involved in pathways for production of anti-inflammatory compounds [
8,
9].
Although antibiotic treatment has been associated with suppression of
Proteobacteria in COPD [
8,
9], it is not true for all cases [
10]. With the increasing urgency for effective antibiotic stewardship, research is needed to better understand the impact of both acute and long term antimicrobial therapy on the COPD microbiome. In this respect, alternate therapeutic strategies such as
H. influenzae vaccination, or highly selective antimicrobial approaches such as phage therapy may effectively reverse some dysbiotic with prognostic benefit.
A limitation of this study is that the lung microbiome has been analysed from sputum samples which can be contaminated with the microbiome of the oropharynx. However, we emphasise that this effect will have been limited by sputum plug selection for the analysis. We did not perform longitudinal sampling to demonstrate reproducibility of the sputum microbiome over time in healthy participants. For COPD we have previously demonstrated that the sputum microbiome is comparable between time-points when sampling at their stable state [
32]. The effects on the microbiome of using sputum induction as the predominant sampling technique in the healthy control group are also not known, but it is noteworthy that the predominant bacterial constituents of our healthy microbiome are consistent with the respiratory microbiome detected by investigating BAL and bronchial samples reported in previous studies [
3,
6]. This suggests that our observations are robust and representative of the bacterial composition of the lung microbiome. A major incentive to work with sputum is its compatibility with routine clinical practice as any findings are therefore more readily translated into established care pathways. In this study we have not characterized the viral and fungal communities, and this will be important to understand their role in health and disease.
Competing interests
KH, LG, ZW, VM, MYR, RCF, NFR, AJW, AJB, MRB have nothing to declare. JRB, BEM and RTS are employees and shareholders of GSK; GCD reports grants and personal fees from Astrazeneca and grants from Micom ltd and American Thoracic Society; MDT reports grants from GSK and other from Orion; JAW reports grants from GSK, grants from Johnson and Johnson, other from Novartis, other from Boehringer Ingelheim, other from Astra Zeneca, other from GSK, grants from GSK, grants from Astra Zeneca, grants from Boehringer Ingelheim, grants from Novartis; DS reports grants and personal fees from GlaxoSmithKline, grants and personal fees from AstraZeneca, grants and personal fees from Boehringer Ingleheim, grants and personal fees from Chiesi, personal fees from Cipla, personal fees from Genentech, grants and personal fees from Glenmark, grants and personal fees from Menarini, grants and personal fees from Mundipharma, grants and personal fees from Novartis, personal fees from Peptinnovate, grants and personal fees from Pfizer, grants and personal fees from Pulmatrix, grants and personal fees from Therevance, grants and personal fees from Verona; CEB reports grants and personal fees from GSK, grants and personal fees from Novartis, grants and personal fees from Genentech, grants and personal fees from Chiesi, personal fees from Sanofi/Regeneron, grants and personal fees from 4DPharma, grants and personal fees from BI, grants and personal fees from Mologics, grants and personal fees from Gossamer, grants and personal fees from AZ/MedImmune, grants and personal fees from TEVA, outside the submitted work. CEB reports grants from MRC COPDMAP, grants from AirPROM (FP7–270194), during the conduct of the study.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.