Annals of the American Thoracic Society

Rationale: Although airway microbiota composition correlates with clinical measures in non–cystic fibrosis bronchiectasis, these data are unlikely to provide useful prognostic information at the individual patient level. A system enabling microbiota data to be applied clinically would represent a substantial translational advance.

Objectives: This study aims to determine whether stratification of patients according to the predominant microbiota taxon can provide improved clinical insight compared with standard diagnostics.

Methods: The presence of bacterial respiratory pathogens was assessed in induced sputum from 107 adult patients by culture, quantitative PCR, and, in 96 samples, by ribosomal gene pyrosequencing. Prospective analysis was performed on samples from 42 of these patients. Microbiological data were correlated with concurrent clinical measures and subsequent outcomes.

Measurements and Main Results: Microbiota analysis defined three groups: Pseudomonas aeruginosa dominated (n = 26), Haemophilus influenzae dominated (n = 34), and other taxa dominated (n = 36). Patients with P. aeruginosa– and H. influenzae–dominated communities had significantly worse lung function, higher serum levels of C-reactive protein (CRP), and higher sputum levels of IL-8 and IL-1β. Predominance of P. aeruginosa, followed by Veillonella species, was the best predictor of future exacerbation frequency, with H. influenzae–dominated communities having significantly fewer episodes. Detection of P. aeruginosa was associated with poor lung function and exacerbation frequency, irrespective of analytical strategy. Quantitative PCR revealed significant correlations between H. influenzae levels and sputum IL-8, IL-1β, and serum CRP. Genus richness was negatively correlated with 24-hour sputum weight, age, serum CRP, sputum IL-1β, and IL-8.

Conclusions: Stratification of patients with non–cystic fibrosis bronchiectasis on the basis of predominant bacterial taxa is more clinically informative than either conventional culture or quantitative PCR–based analysis. Further investigation is now required to assess the mechanistic basis of these associations.

The pathophysiology of non–cystic fibrosis (CF) bronchiectasis is generally considered to be characterized by an airway inflammatory response to chronic bacterial infection (1). An association between high bacterial load (as determined by culture-based approaches) and systemic inflammation, airway inflammation, and a greater risk of exacerbations (2) appears to support this hypothesis. However, although Pseudomonas aeruginosa infection is associated with more severe disease (35), the contribution of other species to airway disease is not known. Further, in a substantial proportion of patients, no recognized respiratory pathogen is isolated through conventional diagnostic culture, despite elevated inflammatory marker levels (6).

Previously, we have used a meta-analytical approach based on 16S ribosomal RNA (rRNA) gene pyrosequencing data to define the “core” bacterial community associated with bronchoalveolar lavage and sputum samples from adult patients with non-CF bronchiectasis (7). This analysis revealed that, by a considerable margin, the species that contributed most to bacterial community similarity in these subjects was Haemophilus influenzae. In contrast, P. aeruginosa was distributed more unevenly within the patient population, but was present at high levels in certain individuals. Members of the genera Streptococcus and Veillonella were also notable as present in a majority of subjects. In keeping with other reports (8), a large number of other genera not traditionally associated with airway infection were also defined as core within this patient group.

Defining a core microbiota in this way helps to identify species that are abundant and frequently detected within a patient population, and therefore more likely to be of clinical relevance. However, by focusing on similarities between patients, such meta-analysis could give the impression that airway microbiology is conserved among patients with non-CF bronchiectasis, thus supporting a uniform or nonstratified approach to antibiotic therapy. Furthermore, while airway microbiota composition correlates with important clinical measures in non-CF bronchiectasis (7), these complex data sets are not readily converted into clinical or prognostic information that can be applied in a treatment setting on an individual patient level.

We hypothesized that the dominant bacterial taxon with the airway microbiota would provide a basis for clinically relevant stratification of patients, providing prognostic information superior to that available using current analytical approaches. To test this, we compared recorded clinical measures with assessments of sputum bacterial composition, as defined by conventional diagnostic culture, quantitative PCR (Q-PCR), and 16S rRNA gene pyrosequencing, in 107 adult patients with bronchiectasis prospectively enrolled in the Bronchiectasis and Low-dose Erythromycin Study (BLESS) (9).

The BLESS was a 12-month double-blind, randomized, placebo-controlled study of low-dose erythromycin in patients with non-CF bronchiectasis (10). Patients’ details have been published previously, and are summarized in Table 1. Details of patient recruitment, inclusion criteria, and assessment protocols are provided in the online supplement.

Table 1. Patient demographic and diagnostic microbiology data

Age (yr), mean (SD)62.5 (9.8)
Female sex, no. (%)59
Duration of bronchiectasis (yr), mean (SD, range)40.5 (21.8, 1–75)
 Idiopathic, no. (%)60 (56.0)
 Infectious, no. (%)30 (28.0)
 Pinks disease, no. (%)11 (11.3)
 Ciliary dysfunction, no. (%)3 (2.8)
FEV1, mean (SD) 
 Prebronchodilator1.82 (0.72)
 Prebronchodilator, % predicted68.0 (18.3)
 Postbronchodilator1.92 (0.74)
 Postbronchodilator, % predicted71.6 (18.7)
Ex-smokers, no. (%)20 (18.7)
 Smoking history (pack-years), mean (SD)12.2 (18.1)
 Median (IQR)0.0 (0.0)
24-h sputum weight (g), median (IQR)18.0 (13.7–24.7)
SGRQ total score, mean (SD)37.9 (14.4)
LCQ score, mean (SD)14.9 (3.2)
6-min walk test (m), median (IQR)510 (475–572.5)
C-reactive protein (mg/L), median (IQR)3.55 (1.2–8.5)
Exacerbations in prior 12 mo 
 Total exacerbations, mean (SD)4.8 (2.8)
 ≥5 Exacerbations, no. (%)40 (37.4)
Medications, no. 
 Combination inhalers, ICS, LABA47
 Inhaled SABA47
 Inhaled anticholinergics13
 Inhaled corticosteroids alone13
 Inhaled LABA alone4
 Nebulized saline2
 Inhaled mannitol1
 Nebulized colistin1
Comorbidities, no. 
 Ischemic heart disease10
 Cerebrovascular disease6
 Diabetes mellitus3
Diagnostic microbiology 
 No organism reported47
Haemophilus influenzae24
Pseudomonas aeruginosa32
Alcaligenes xylosoxidans1
Moraxella catarrhalis3
Pseudomonas stutzeri1

Definition of abbreviations: ICS = inhaled corticosteroids; IQR = interquartile range; LABA = long-acting β-agonists; LCQ = Leicester Cough Questionnaire; SABA = short-acting β-agonists; SGRQ = St. George’s Respiratory Questionnaire.

Induced sputum samples were collected at baseline from 107 patients. Clinical measures were recorded as described previously (9, 11). Exacerbation frequency and antibiotic treatment data were collected for 52 weeks after sampling, and these data were used as variables in the analysis of microbiological data from 42 patients from the placebo arm of the trial.

Sputum induction was performed as described previously (10), with details provided in the online supplement. After collection, sample aliquots were frozen at −80°C, with separate aliquots used for molecular microbiology and inflammatory marker analysis.

Nucleic acid extraction and bacterial tag–encoded FLX amplicon pyrosequencing (bTEFAP) were performed as described previously (12), with detailed protocols provided in the online supplement. Sequence data generated have the accession number SRP035600.

All statistical analyses were performed with IBM SPSS Statistics (version 21; IBM, New York, NY). Boxplots were generated with the GraphPad Prism program (version 5.00; GraphPad Software, San Diego, CA), with full details provided in the online supplement.

Given the clinical importance of P. aeruginosa in non-CF bronchiectasis, and the prevalence of H. influenzae in airway samples, we initially assessed the relationships between these species and disease measures using standard diagnostic culture, Q-PCR–based detection, and Q-PCR enumeration.

Bacterial isolation frequencies by standard diagnostic microbiological surveillance are shown in Table 1. Compared with noninfected subjects, P. aeruginosa detection by culture was associated with poor lung function (mean FEV1, 63.8 vs. 74.9%, P = 0.007, Wilcoxon rank sum test, n = 107) and with the number of reported exacerbations in the prior 12 months (P = 0.011). Patients with H. influenzae infection had fewer prior infective exacerbations compared with other subjects (P = 0.026).

Agreement between culture- and PCR-based detection of P. aeruginosa was poor (κ = 0.116, P = 0.022), with detection significantly higher by PCR (91 of 107 vs. 31 of 107 samples; P < 0.001). The associations with reduced lung function (P = 0.021, postbronchodilator FEV1) and with the number of exacerbations in the prior 12 months (P = 0.026) that were observed for culture remained, despite this group now comprising 85% of patients. The frequency of H. influenzae detection in these samples was also significantly higher by PCR than by culture (92 of 107 vs. 42 of 107 samples, P < 0.001; κ = 0.094, P = 0.023).

Levels of total bacteria, P. aeruginosa, and H. influenzae per unit volume of sputum were determined for all samples by Q-PCR (Figure 1). No significant relationships between total bacterial load and neutrophil count or other disease measures, including lung function, were observed. Enumeration of P. aeruginosa and H. influenzae by this approach was the most clinically informative of the three conventional analyses. A negative correlation was found between P. aeruginosa load and lung function (FEV1, ρ = –0.28, P = 0.004) and the number of episodes of intravenous antibiotic therapy in the prior 12 months (ρ = 0.23, P = 0.016). In addition, H. influenzae load was correlated with levels of sputum IL-8 (ρ = 0.22, P = 0.035), sputum IL-1β (ρ = 0.33, P = 0.001), and serum C-reactive protein (CRP) (ρ = 0.25, P = 0.010).

Bacterial Community Composition

Sputum samples from 96 subjects were analyzed by 16S rRNA gene pyrosequencing, with a median of 18 genera detected per sample (range, 6–53 genera). The genera most commonly detected were those associated with the oropharynx and reported previously in chronic lower infections (10) (see Table 2).

Table 2. Bacterial taxa most frequently detected by 16S ribosomal RNA gene pyrosequencing, and the frequency of detection of selected recognized respiratory pathogens

TaxonSamples (n = 96)
Veillonella90 (93.8)
Prevotella84 (86.5)
Streptococcus82 (85.4)
 (of which S. pneumoniae)28 (29.2)
Neisseria69 (71.9)
Haemophilus influenzae92 (95.8)
Pseudomonas aeruginosa70 (72.9)
Stenotrophomonas maltophilia29 (30.2)
Burkholderia cepacia complex27 (28.1)
Moraxella catarrhalis21 (21.9)
Staphylococcus aureus11 (11.5)

Values in parentheses represent percentages.

Bacterial communities were typically dominated by a small number of taxa. On average, only 5.3 (±3.9) genera per sample represented more than 1% of total bacterial signal, with the dominant genus typically representing more than 70% (mean, 71.3%; SD, 27.1). The genus with the highest mean relative abundance was Haemophilus (33.4%), followed by Pseudomonas (23.5%), Prevotella (9.5%), Veillonella (8.2%), Streptococcus (5.6%), Pasteurella (2.0%), Neisseria (1.9%), Porphyromonas (1.6%), and Moraxella (1.5%). Genus richness was negatively correlated with 24-hour sputum weight (ρ = –0.38, P < 0.001), age (ρ = –0.28, P = 0.005), serum CRP (ρ = –0.26, P = 0.010), sputum IL-1β (ρ = –0.35, P = 0.001), and sputum IL-8 (ρ = –0.51, P < 0.001).

On the basis of their predominant taxon, bacterial communities could be placed into one of three groups: P. aeruginosa dominated (n = 26), H. influenzae dominated (n = 34), or dominated by a species other than these two (n = 36). Median relative abundance of the dominant taxon, and bacterial genus richness, for each subgroup, are shown in Table 3.

Table 3. Median relative abundance and bacterial genus richness of samples grouped according to dominant taxon

  Dominant Species Relative Abundance (%)Bacterial Richness (Genus)
All samples9682.152.318.015.0
Pseudomonas aeruginosa dominated2692.414.217.012.0
Haemophilus influenzae dominated3496.29.215.510.0
Dominated by another genus3637.626.028.511.3
Veillonella spp. dominated1035.412.224.016.0
Prevotella spp. dominated936.819.828.08.0

Definition of abbreviation: IQR = interquartile range.

Bacterial richness was significantly lower in communities dominated by either P. aeruginosa or H. influenzae, compared with those dominated by another bacterial species (P < 0.001, one-way analysis of variance). Hierarchical cluster analysis using the Bray–Curtis similarity measure was used to assess the extent to which group membership, based on dominant taxa, was maintained when clustering was performed using all detectable genera (see Figure E1 in the online supplement). In the majority of cases, clustering was closely related to dominant taxon group, with this effect more pronounced in those patients in whom P. aeruginosa or H. influenzae was dominant. The median relative abundance of the dominant taxa was higher in communities dominated by P. aeruginosa or H. influenzae.

Total bacterial load did not differ significantly between P. aeruginosa– and H. influenzae–dominated samples (Figure 2). However, bacterial communities dominated by P. aeruginosa had a higher total bacterial load compared with all other samples analyzed (median, 7.5 × 109 vs. 5.5 × 109 CFU/ml; P = 0.024). Agreement between detection of P. aeruginosa and H. influenzae by culture, Q-PCR, 16S rRNA gene pyrosequencing, and identification of dominant taxa by pyrosequencing, is shown in Figure E2, with community dominance by P. aeruginosa and H. influenzae comparable with the results of culture-based detection (κ = 0.72, P < 0.001, and 0.65, P < 0.001, respectively).

The median relative abundance of the dominant taxa was higher in communities dominated by P. aeruginosa or H. influenzae than in communities dominated by other species (93.7 and 37.6%, respectively, P < 0.001, Kruskal-Wallis analysis of variance). Relative abundance of the predominant taxon correlated significantly with a number of disease measures (Table 4). Notably, relative abundance of the predominant taxon (irrespective of identity) was negatively correlated with lung function (FEV1) and sputum macrophage levels, and positively correlated with duration of bronchiectasis, serum CRP, sputum IL-8 and IL-1β, and sputum neutrophil levels.

Table 4. Significant correlations between the relative abundance of the dominant bacterial taxon and clinical measures of disease

Clinical MeasureSpearman's ρSignificance (P Value)n
Pre-BD FEV1−0.2250.02796
Post-BD FEV1−0.2230.02996
Bronchiectasis duration, yr0.2210.03196
Sputum IL-80.585<0.00186
Sputum IL-1β0.626<0.00186
Sputum neutrophils, absolute0.2400.03181
Sputum neutrophils, %0.3470.00181
Sputum macrophages, absolute−0.3400.00281
Sputum macrophages, %−0.389<0.00181
24-h sputum volume, g0.3700.00296

Definition of abbreviations: BD = bronchodilator; CRP = C-reactive protein.

Bacterial communities in 36 of the samples (37.5%) were dominated by a species other than P. aeruginosa or H. influenzae. All of the predominant taxa in these patients have been reported previously in the context of lower airway infections (Figure E3). Whereas the majority of taxa were dominant in one sample only, members of the genera Veillonella and Prevotella were dominant in 10 and 9 samples, respectively. Median bacterial load for this group was 1.4 × 109 CFU/ml (range, 8.9 × 104–1.8 × 1010) with bacterial community profiles obtained for all samples. Despite this, 27 (75%) were classified as culture negative by standard diagnostic microbiology.

We next examined the ability of patient grouping based on predominant taxon to provide clinically useful stratification. Patients whose airway bacterial community was dominated by P. aeruginosa exhibited disease features more severe than those of patients dominated by H. influenzae or other bacterial species. FEV1 was lower in these patients (mean, 56.8 ± 17.6%) than in patients whose bacterial community was dominated either by H. influenzae (69.9 ± 17.9%; P = 0.010) or another species (73.1 ± 16.8%; P = 0.001). Patients with P. aeruginosa–dominated bacterial communities also had a greater number of exacerbations in the 12 months before sampling (6.1 ± 3.0) compared with both the patient group as a whole (4.7 ± 2.6; P = 0.007) and those whose bacterial community was dominated by H. influenzae (3.5 ± 1.4; P = 0.001). Levels of CRP, sputum IL-8, and sputum IL-1β did not differ between P. aeruginosa– and H. influenzae–dominated communities; however, P. aeruginosa– and H. influenzae–dominated communities had lower FEV1 (P = 0.029) and higher levels of CRP (P = 0.004), sputum IL-8 (P < 0.001), and sputum IL-1β (P < 0.001) than communities dominated by another species.

Although P. aeruginosa was detected in 22 of the 34 patients in whom H. influenzae was dominant, no significant differences were observed in clinical measures between these patients and the H. influenzae–dominated patients in whom P. aeruginosa was not detected.

Microbiological Measures as Prognostic Markers

Data were available for 42 of the placebo subjects regarding the number of exacerbations and the total days of antibiotic therapy that occurred in the 12 months after sample collection (Tables 3 and 5). Patients whose bacterial community was dominated by H. influenzae had fewer exacerbations (1.1 ± 1.1) than those dominated by P. aeruginosa (3.7 ± 2.1; P < 0.001) or other genera (2.5 ± 2.1; P = 0.047). This pattern was reflected in the number of days of antibiotic therapy that patients received, with those whose bacterial community was dominated by H. influenzae having fewer (10.8 ± 13.8 d) than those dominated by either P. aeruginosa (52.4 ± 35.1; P < 0.001) or other genera (31.2 ± 27.3; P = 0.017). No significant relationship was identified between taxon dominance and 12-month change in FEV1.

Table 5. Median relative abundance and bacterial genus richness of samples used for prognostic analysis, grouped according to dominant taxon

  Dominant Species Relative Abundance (%)Bacterial Richness (Genus)
All samples4286.360.419.518.5
Pseudomonas aeruginosa dominated1295.26.418.58.8
Haemophilus influenzae dominated1496.215.711.05.8
Dominated by another genus1634.513.929.513.0
Veillonella spp.834.610.034.513.0
Prevotella spp.433.88.629.56.3
Burkholderia spp.1(58.1) (11.0) 
Flavobacterium spp.1(16.7) (38.0) 
Leptotrichia spp.1(34.6) (27.0) 
Pasteurella spp.1(99.6) (11.0) 

Definition of abbreviation: IQR = interquartile range.

Values in parentheses indicate single measures, rather than median values.

By comparison, analysis based on culture data revealed significant relationships only between P. aeruginosa and a greater number of days of antibiotic compared with the group as a whole (P = 0.009), and with patients in whom H. influenzae was dominant (P = 0.011). No significant relationship was identified between culture-based microbiology and exacerbation frequency.

Multiple regression analysis indicated that P. aeruginosa microbiota predominance was the single best predictor of exacerbation frequency (β= 0.501, P < 0.001) and associated antibiotic burden (β= 0.595, P < 0.001), followed by dominance by a Veillonella species (β= 0.385, P = 0.005 and β= 0.385, P = 0.003, respectively). The relative risk of a patient having five or more exacerbations in the 12 months after analysis was greater by a factor of 1.60 when P. aeruginosa was dominant, and by a factor or 1.44 when a Veillonella species was dominant. By comparison, the relative risk associated with P. aeruginosa culture positivity was 1.35.

We describe a system of microbiological stratification of patients based on predominant bacterial taxa that is more closely correlated with clinical measures of disease, inflammation, and disease-related outcomes than the detection of recognized pathogens either by culture or PCR assays. This stratification system allows microbiota composition data to be directly clinically relevant for individual patients, as well as providing potential mechanistic insights in relation to host–bacteria interaction.

An initial assessment of P. aeruginosa and H. influenzae detection through culture and PCR assays indicated they were comparable in terms of their correlation with disease measures. Both approaches indicate a significant association between P. aeruginosa detection and poor lung function, as reported previously (5). A significant association was also observed between P. aeruginosa detection and subsequent exacerbation frequency, the first time this has been shown to our knowledge. Culture performed slightly better than PCR-based detection, with a significant relationship between H. influenzae isolation and low exacerbation frequency identified by culture alone. The relatively weak relationship between PCR-based detection and disease measures may result from high assay sensitivity, resulting in most patients being classified “positive” when bacteria may be below clinically relevant levels; a model supported by the finding that Q-PCR–based enumeration of bacteria provided greater clinical insight than detection alone. Here, in addition to a significant negative correlation between P. aeruginosa load and lung function, significant positive correlations were also observed between H. influenzae load and serum levels of CRP and sputum IL-8 and IL-1β levels.

In keeping with previous studies of non-CF bronchiectasis, and in studies of chronic airway disease more widely, microbiota profiling indicated that a substantial proportion of the detected bacteria represented taxa other than recognized respiratory pathogens. Although the clinical significance of many of these species in airway disease is as yet undetermined, the presence of any substantial bacterial population in the lower airways is arguably a clinical concern.

Converting complex bacterial community data into a readily accessible measure for patient stratification, particularly at the individual patient level, is a major challenge. For the first time, our stratification system conveys complex microbiota data in a clinically relevant and clinically recognizable output that has the potential to be applied at the individual patient level (by clinicians) and also provides prognostic information that is superior to existing culture-based methods. This stratification system therefore represents a substantial translational advance toward a clinically relevant tool from bacterial community profiling.

Bacterial community diversity as a gross measure of community structure has the potential to be informative in itself, requiring relatively little in-depth analysis. A correlation between lower diversity and more severe disease has been reported in other chronic respiratory infections (1315), and was also observed here. It is not clear whether this relationship is reflective of the impact of virulent pathogens (such as P. aeruginosa) outcompeting other species, increased antibiotic exposure in more severe disease, or a direct role of reduced diversity as a primary driver of disease. Although the high antibiotic treatment burden in patients with more severe disease (16, 17) seems a likely explanation, no relationship was observed here between bacterial diversity and the number of courses of antibiotics a patient received in the year before sampling.

However, richness measures do not reflect the relative abundance of clinically important species. Here, defining samples according to their predominant taxon provided a substantial advantage. In addition to the relationship between P. aeruginosa, poor lung function, and exacerbation history revealed by culture, additional significant correlations with serum and sputum inflammatory marker levels were observed by this strategy. Furthermore, grouping of patients according to dominant taxon was also more informative in terms of predicting future clinical course. For example, whereas H. influenzae culture positivity showed no significant clinical correlations, patients whose bacterial community was dominated by H. influenzae had significantly fewer exacerbations, and received significantly fewer days of antibiotics, compared with patients dominated by either P. aeruginosa or other species. In addition, community dominance by P. aeruginosa or a Veillonella species was associated with a significantly greater risk of frequent future exacerbations.

The results of this study suggest distinct clinical and prognostic associations according to bacterial taxa might exist. Pseudomonas predominance is associated with markers of more severe disease including poorer lung function, frequent exacerbations, and greater antibiotic use. In contrast, Haemophilus predominance is associated with fewer pulmonary exacerbations (historically and prospectively). However, Haemophilus bacterial load was associated with inflammatory activation assessed both locally (in sputum) and systemically (in blood); it is tempting to speculate that the host inflammatory response to Haemophilus infection, while increased, contains infection and prevents exacerbation, in contrast to infection with Pseudomonas and other dominant taxa.

Although providing a superior basis for the stratification of non-CF bronchiectasis airway infections compared with standard microbiology, identification of dominant taxa through deep sequencing represents a much greater investment of resources. However, if airway microbiota are conserved over time within individuals, infrequent assessment could still provide important prognostic information. Further, unlike culture-based analysis, which indicated that approximately one-quarter of samples analyzed were culture negative, sequencing identified a dominant bacterial taxon in all samples. Despite bacterial loads being significantly lower than in samples from other patients, they were still substantially above levels that might be expected in lower airway samples from healthy individuals (18), suggesting strongly that they represent chronic infection.

In many cases, the predominant taxa in “culture-negative” samples were recognized to be associated with lower respiratory infection, for example, Moraxella, Stenotrophomonas, or Burkholderia species. However, the genera Prevotella and Veillonella were both the dominant taxa in about one-quarter of the samples that were not dominated by either P. aeruginosa or H. influenzae. This observation is consistent with culture-based work described elsewhere (8). Further, Veillonella predominance was predictive of a high frequency of subsequent exacerbation, a finding that highlights the contribution of bacterial species outside those traditionally considered important in a respiratory context. As Veillonella species would not be isolated under conditions used for routine diagnostic testing of respiratory samples, they are likely to be underreported. The data presented here suggest their potential to act as a marker for disease progression, or indeed to play a direct role in pathogenesis in this context.

Some genera detected in high relative abundance here, including Veillonella species, are commonly associated with the oral cavity, and the potential for sample contamination is difficult to exclude. The absence of a significant divergence in microbiota composition between matched induced sputum and bronchoalveolar lavage samples has been demonstrated previously, with a high prevalence of species including members of the genera Veillonella and Prevotella in bronchoalveolar lavage fluid from this patient group (7). Further, there is evidence that although contamination of CF sputum by oral species occurs at low relative levels, the presence of oral anaerobes in sputum in high relative abundance is likely to represent lower airway colonization (19). Despite this evidence, the isolation of these bacteria directly from areas of bronchiectasis within the lung, perhaps by directed bronchoalveolar lavage, represents an important next step in determining the extent of their contribution to disease.

Given their potential clinical importance, it is worth considering the nature of Veillonella species. Members of this genus are anaerobic, gram-negative cocci and are residents of the oral cavity, gastrointestinal tract, and vagina (20). Veillonella species are commonly recovered from abscesses, aspiration pneumonias, burns, bites, and sinuses (21). In many instances, these infections are polymicrobial (21). Perhaps partly because samples from chronic respiratory infections are rarely subjected to anaerobic diagnostic microbiology, the role played by this genus in disease progression is poorly understood. However, Veillonella species are generally susceptible to most antibiotics used for the treatment of anaerobic infections, including β-lactam antibiotics, clindamycin, and metronidazole (21).

Limitations of this study include the limited number of patients from whom samples were obtained, and the basis of analysis in a single time-point sample. Here, the variability of airway microbiology both within and between patients is an important consideration, and one that points to an extension of the analysis presented here to a larger patient population. The limitations of using 16S rRNA gene pyrosequencing to characterize the airway microbiota must also be considered. The relatively short read lengths mean that identification of bacteria to a species level is not always possible, with additional methods needed when such species-level identification is important (species-specific PCR assays were used here to confirm identification of P. aeruginosa and H. influenzae).

In conclusion, this stratification system provides highly clinically relevant output from complex bacterial community data that surpasses culture-based techniques. In subjects with non-CF bronchiectasis this system enabled future exacerbation risk to be predicted, including in the substantial subset of subjects with dominant bacterial taxa that are not ordinarily identified by culture and have not previously been considered to be pathogenic.

1 . Cole PJ. Inflammation: a two-edged sword—the model of bronchiectasis. Eur J Respir Dis Suppl 1986;147:615.
2 . Chalmers JD, Smith MP, McHugh BJ, Doherty C, Govan JR, Hill AT. Short- and long-term antibiotic treatment reduces airway and systemic inflammation in non–cystic fibrosis bronchiectasis. Am J Respir Crit Care Med 2012;186:657665.
3 . Evans SA, Turner SM, Bosch BJ, Hardy CC, Woodhead MA. Lung function in bronchiectasis: the influence of Pseudomonas aeruginosa. Eur Respir J 1996;9:16011604.
4 . King PT, Holdsworth SR, Freezer NJ, Villanueva E, Holmes PW. Microbiologic follow-up study in adult bronchiectasis. Respir Med 2007;101:16331638.
5 . Martínez-García MA, Soler-Cataluña JJ, Perpiñá-Tordera M, Román-Sánchez P, Soriano J. Factors associated with lung function decline in adult patients with stable non–cystic fibrosis bronchiectasis. Chest 2007;132:15651572.
6 . Pasteur MC, Helliwell SM, Houghton SJ, Webb SC, Foweraker JE, Coulden RA, Flower CD, Bilton D, Keogan MT. An investigation into causative factors in patients with bronchiectasis. Am J Respir Crit Care Med 2000;162:12771284.
7 . Rogers GB, van der Gast CJ, Cuthbertson L, Thomson SK, Bruce KD, Martin ML, Serisier DJ. Clinical measures of disease in adult non-CF bronchiectasis correlate with airway microbiota composition. Thorax 2013;68:731737.
8 . Tunney MM, Einarsson GG, Wei L, Drain M, Klem ER, Cardwell C, Ennis M, Boucher RC, Wolfgang MC, Elborn JS. Lung microbiota and bacterial abundance in patients with bronchiectasis when clinically stable and during exacerbation. Am J Respir Crit Care Med 2013;187:11181126.
9 . Serisier DJ, Martin ML, McGuckin MA, Lourie R, Chen AC, Brain B, Biga S, Schlebusch S, Dash P, Bowler SD. Effect of long-term, low-dose erythromycin on pulmonary exacerbations among patients with non–cystic fibrosis bronchiectasis: the BLESS randomized controlled trial. JAMA 2013;309:12601267.
10 . Rogers GB, Skelton S, Serisier DJ, van der Gast CJ, Bruce KD. Determining cystic fibrosis–affected lung microbiology: comparison of spontaneous and serially induced sputum samples by use of terminal restriction fragment length polymorphism profiling. J Clin Microbiol 2010;48:7886.
11 . Bronchiectasis (Non-CF) Guideline Group, British Thoracic Society. Guideline for non-CF bronchiectasis. Thorax 2010;65(Suppl 1):i1–i58. Available from:
12 . Rogers GB, Cuthbertson L, Hoffman LR, Wing PA, Pope C, Hooftman DA, Lilley AK, Oliver A, Carroll MP, Bruce KD, et al. Reducing bias in bacterial community analysis of lower respiratory infections. ISME J 2013;7:697706.
13 . Klepac-Ceraj V, Lemon KP, Martin TR, Allgaier M, Kembel SW, Knapp AA, Lory S, Brodie EL, Lynch SV, Bohannan BJ, et al. Relationship between cystic fibrosis respiratory tract bacterial communities and age, genotype, antibiotics and Pseudomonas aeruginosa. Environ Microbiol 2010;12:12931303.
14 . Cox MJ, Allgaier M, Taylor B, Baek MS, Huang YJ, Daly RA, Karaoz U, Andersen GL, Brown R, Fujimura KE, et al. Airway microbiota and pathogen abundance in age-stratified cystic fibrosis patients. PLoS One 2010;5:e11044.
15 . van der Gast CJ, Walker AW, Stressmann FA, Rogers GB, Scott P, Daniels TW, Carroll MP, Parkhill J, Bruce KD. Partitioning core and satellite taxa from within cystic fibrosis lung bacterial communities. ISME J 2011;5:780791.
16 . Zhao J, Schloss PD, Kalikin LM, Carmody LA, Foster BK, Petrosino JF, Cavalcoli JD, VanDevanter DR, Murray S, Li JZ, et al. Decade-long bacterial community dynamics in cystic fibrosis airways. Proc Natl Acad Sci USA 2012;109:58095814.
17 . Rogers GB, Hoffman LR, Carroll MP, Bruce KD. Interpreting infective microbiota: the importance of an ecological perspective. Trends Microbiol 2013;21:271276.
18 . Koenig SM, Truwit JD. Ventilator-associated pneumonia: diagnosis, treatment, and prevention. Clin Microbiol Rev 2006;19:637657.
19 . Rogers GB, Carroll MP, Serisier DJ, Hockey PM, Jones G, Kehagia V, Connett GJ, Bruce KD. Use of 16S rRNA gene profiling by terminal restriction fragment length polymorphism analysis to compare bacterial communities in sputum and mouthwash samples from patients with cystic fibrosis. J Clin Microbiol 2006;44:26012604.
20 . Bhatti MA, Frank MO. Veillonella parvula meningitis: case report and review of Veillonella infections. Clin Infect Dis 2000;31:839840.
21 . Brook I. Veillonella infections in children. J Clin Microbiol 1996;34:12831285.

*Joint senior authors.

Correspondence and requests for reprints should be addressed to Geraint B. Rogers, Ph.D., Mater Medical Research Institute, Translational Research Institute, 37 Kent Street, Woolloongabba, QLD 4102, Australia. E-mail:

Supported by the Mater Adult Respiratory Research Trust Fund.

Author Contributions: G.B.R., N.M.M.Z., K.D.B., L.D.B., A.C.C., D.W.R., M.A.M., D.J.S. all contributed to the conception, execution, analysis, and reporting of the study, and approved the final version. G.B.R. acts as the guarantor of the manuscript.

This article has an online supplement, which is accessible from this issue’s table of contents online at

Author disclosures are available with the text of this article at


No related items
Comments Post a Comment

New User Registration

Not Yet Registered?
Benefits of Registration Include:
 •  A Unique User Profile that will allow you to manage your current subscriptions (including online access)
 •  The ability to create favorites lists down to the article level
 •  The ability to customize email alerts to receive specific notifications about the topics you care most about and special offers
Annals of the American Thoracic Society

Click to see any corrections or updates and to confirm this is the authentic version of record