Airway dysbiosis: Haemophilus influenzae and Tropheryma in poorly controlled asthma

Discussion

In the present study, adults with asthma had reduced bacterial diversity with the lowest diversity demonstrated in those with neutrophilic asthma. Dominant taxa were identified within specific inflammatory phenotypes. We identified a group of adult males with high abundance of Haemophilus sequences with predominantly neutrophilic airway inflammation. We also identified a group of participants with asthma who had Tropheryma sequences which we confirmed as T. whipplei. These participants most commonly exhibited an eosinophilic asthma subtype.

Our data show reduced bacterial diversity in neutrophilic asthma. This is the first examination of bacterial diversity within asthma inflammatory subtypes. Diversity has been reported to be increased [5, 11] or similar [6] between patients with asthma and healthy controls. Similar variation in diversity has been reported in patients with chronic obstructive pulmonary disease (COPD) where bacterial diversity has been reported as increased [12] or reduced [13] compared with controls. As the majority of participants were taking ICS, we are unable to determine if the use of ICS influences bacterial diversity. We corrected the diversity analysis for the dose of ICS, and found the difference in both Simpson and Shannon index was maintained after correction for ICS dose.

Studies of asthma in both adults and children report an increased frequency of sequences from the phylum Proteobacteria [4–6, 14]. Huang et al. [14], in their study of 30 adults with severe asthma, observed that a worsening ACQ score was associated with an abundance of Proteobacteria and an association between T-helper 17-related gene expression (including CXCL8 and its receptors) with Proteobacteria taxa. We extend this to show this increase in Proteobacteria taxa is most prominent in those with neutrophilic asthma. We also observed significantly fewer OTUs from bacteria in the phyla Firmicutes and Actinobacteria in neutrophilic asthma. The anaerobic bacterial phylum Fusobacteria was also less frequent in neutrophilic asthma, although not statistically different. This phylum has been associated with corticosteroid-resistant asthma [15] which is known to be noneosinophilic in nature and suggests the absence of common oral anaerobes may be an important difference between asthma subtypes.

Currently, the impact of long-term ICS use on bacterial diversity is unknown. Early studies suggest that ICS use may influence the airway microbiome, resulting in a unique cluster of COPD patients taking ICS compared with those not taking ICS [12]. Similarly there was a unique cluster of taxa among those with corticosteroid-resistant asthma, including the genera Haemophilus and Tropheryma [15]. In this study, where the majority of participants were taking high doses of ICS, we also show a high prevalence of T. whipplei and H. influenzae in poorly controlled asthma. Little is understood about the impact of inhaled medications on oral flora. ICS may promote adverse oral effects; the most commonly reported of these is a higher rate of oral Candida species detection [16] and there is some evidence of mucosal damage in children with asthma [17].

This study is limited in its cross-sectional nature as it is not possible to distinguish between cause and effect. The strengths of this study include the well-characterised study population combining detailed clinical assessment with assessment of both airway inflammation and the airway microbiome. Induced sputum has been used frequently to study the airway microbiome [18–20]. We have attempted to reduce oral contamination by selecting the lower airway portions of the sample from the salivary material; however, oral contamination cannot be excluded. Other research suggests that while lower airway samples do contain an abundance of oral microbiota, there is a lung-specific signal where some OTUs are disproportionately found in bronchoalveolar lavage compared with oral wash, particularly Haemophilus species [21].

We identified seven patients with asthma who had sequences belonging to the genus Tropheryma (phylum Actinobacteria). The only named species of this genus, T. whipplei, causes Whipple's disease, a gastrointestinal disorder characterised by arthritis, diarrhoea and weight loss [22]. Increasing evidence supports a role for T. whipplei in airway disease with identification in interstitial lung disease [23], pneumonia [24, 25] and recently in corticosteroid-resistant asthma [6]. Pulmonary symptoms may precede gastrointestinal symptoms, including intermittent fever, dry cough and shortness of breath, which suggests that airway colonisation or infection may contribute to the symptoms and exacerbations observed in asthma. Our PCR confirmation resulted in a further five participants having evidence of T. whipplei, with a PCR detection rate of 40%. Tropheryma has also been reported in asymptomatic HIV infection (13%), specifically in the lung rather than the mouth [26], and this is further evidence that our identification of Tropheryma in samples of induced sputum may originate in the lower airway. The participants with Tropheryma tended to be taking a higher dose of ICS; however, this was not statistically different. While further research with longitudinal follow-up is needed to determine if this group of patients will develop Whipple's disease, identification of Tropheryma suggests previously unidentified microorganisms may be present in the airways, possibly contributing to asthma symptoms and airway inflammation.

H. influenzae is the most common species from the phylum Proteobacteria resulting in airway infection, and has been isolated from the airways of both adults and children with asthma [1, 27, 28]. In neutrophilic asthma, there is a high incidence of H. influenzae identified using traditional culture methods, compared with other inflammatory phenotypes [2]. In this study, we observed a group of adults with asthma with a predominance of Haemophilus OTUs who were younger, atopic, nonsmoking males who tended to have higher proportions of neutrophils. Since the sensitivity of the pyrosequencing method does not allow the exact species to be determined, we conducted real-time PCR using specific probes and found that 80% of participants with poorly controlled asthma were positive for H. influenzae. While PCR does not distinguish between the presence of dead and live organisms, it is important to recognise that even dead or degraded bacteria can elicit an immune response, e.g. bacteria endotoxin can induce significant inflammatory and immune signals. We have previously described high levels of sputum endotoxin in patients with neutrophilic asthma and, in the same group, a high incidence of positive culture for H. influenzae [2].

In conclusion, we have been able to link differences in the airway microbiome to altered host inflammatory response in asthma. There are phenotype-specific alterations to the airway microbiome in asthma. Reduced bacterial diversity combined with a high prevalence of H. influenzae was observed in neutrophilic asthma, whereas eosinophilic asthma had abundant T. whipplei. These changes may modulate local airway inflammation and immune responses. Understanding the relationships between these events will contribute to the development of more specific treatment strategies.