The contribution of infection and the respiratory microbiome in acute exacerbations of idiopathic pulmonary fibrosis

The lung microbiome: a complex and dynamic community of microbes

The epithelial surfaces of the human respiratory tract have historically been described as sterile. More recently, novel culture-independent techniques have debunked this notion by demonstrating that the respiratory tract harbours a complex and dynamic community of microbes which play a role both in health and disease [14]. High-throughput sequencing of the 16S rRNA gene is the most commonly used approach to study bacterial communities. Groups of bacteria sharing similar gene sequences are clustered together and identified by comparison to 16S rRNA reference databases. The commensal, symbiotic and pathogenic microorganisms which inhabit the lower airways are collectively defined as the “lung microbiome”. In health, the primary routes of microbial immigration to the lungs include microaspiration, inhalation of air-borne bacteria and direct dispersion along mucosal surfaces [15]. Of these, microaspiration is probably the dominant mechanism, as the composition of the oral microbiome most closely resembles the lung microbiome. It is also expected to be the most relevant to IPF given the strong association with reflux [16]. Elimination is achieved through mucociliary clearance, coughing, and innate and adaptive immune defences which selectively recognise, kill and clear the microbes [17]. The quantity and composition of the microbiome is influenced by the balance between the rate of immigration and clearance. The discovery of the lung microbiome has prompted the exploration of its role in a variety of respiratory conditions, including IPF. These molecular techniques have not only enabled researchers to characterise niche-specific microbial communities in the respiratory tract of patients and healthy individuals, but have also demonstrated associations which point towards the interaction between the microbiome and the host as key in the aetiology and progression of lung disease [15, 18–20].

The role of infection and the microbiome in AE-IPF

Historically, research investigating the role of infections in the pathogenesis of IPF and as triggers of AE-IPF has focused primarily on viruses. When considering the role of infection in AE-IPF, it is important to remember these are events that, prior to the latest revised diagnostic criteria [5], specifically required the exclusion of any infective trigger [21]. Therefore, the following findings are all in cohorts of patients where overt clinical infection has been excluded.

It has been hypothesised that chronic viral infection may act as a persistent antigenic stimulus or co-factor in susceptible hosts, triggering fibrosis, and there is growing evidence both from human and animal studies to support this [22, 23]. The human herpes virus (HHV) family has received the most attention as either an aetiologic or exacerbating agent of IPF [24, 25], but a number of other viral infections have been consistently associated with IPF and AE-IPF (table 1).

There is growing evidence, albeit associative, suggesting that subclinical or occult viral infections may play a pathogenic role in the development of AE-IPF [32]. Molecular analysis of viral particles and bronchoalveolar lavage (BAL) revealed infection in 19 out of 43 cases of AE-IPF, including 12 cases of torque teno virus (TTV) in the exacerbation cohort with no viruses detected in the stable IPF group [26]. Supporting this observation, IPF subjects who were found to be seropositive for TTV-DNA demonstrated a worse survival in an independent cohort [30]. More common respiratory viral pathogens have also been detected using molecular techniques during exacerbations, including respiratory syncytial virus and cytomegalovirus [27]. Having excluded infection clinically, a cohort of 37 IPF patients experiencing an exacerbation underwent surgical lung biopsy and over one-third exhibited evidence of viral infection based on immunohistochemistry or microarray [28]. More recently, viral sequences were profiled in nasopharyngeal swabs of 30 stable IPF and 30 AE-IPF patients which revealed higher virus-positive rates in the AE-IPF cohort (60%) compared to stable disease (43%) [29]. HHV was the most prominent virus in the AE-IPF group as well as influenza A [29]. As AE-IPF have been reported following influenza A vaccination, its presence suggests that cold-associated infection may potentially trigger these events [31]. Overall these observational studies highlight a potential role for viral infections, although the supporting mechanistic role for airborne viruses is insufficient to prove their causal role in AE-IPF.

While the role of viruses has been extensively explored, the long-held, but incorrect, theory of sterility of the lung outside of clinical infection means historically little work has evaluated the role of bacteria in AE-IPF. Using quantitative methods, the first study to investigate the presence of bacterial growth in the lower airways of 22 stable IPF patients confirmed the presence of known pathogens including Haemophilus, Pseudomonas and Streptococcus in 36% of the cases [33]. More recently, molecular culture-independent techniques have been used to characterise the microbial composition in the lower airways of IPF patients. Recent studies suggest that microbial dysbiosis may be linked to disease outcome (table 2).

The Correlating Outcomes with Biochemical Markers to Estimate Time-Progression (COMET) study was the first to truly evaluate the lung microbiome in IPF patients [34]. The authors demonstrated an increased abundance of either Streptococcus or Staphylococcus was associated with a significant reduction in progression-free survival time [34]. The enrichment of these two bacteria was observed in less than half of the cohort; therefore, it remains unlikely that these organisms alone can explain the disease pathogenesis. Consistent with the COMET study, Molyneaux et al.[36] demonstrated that the lower airways of patients with IPF were not sterile and indeed were more likely to harbour potentially pathogenic Haemophilus, Neisseria and Streptococcus species than healthy controls. The authors also identified the presence of a higher overall bacterial burden in IPF, which was associated with a reduced progression-free survival time in subjects with IPF. Having characterised the microbiome in stable IPF, the same authors then set out to examine for potential changes during acute exacerbations. Using a cohort of matched stable and exacerbation patients who underwent bronchoscopy, the authors were able to demonstrate a further increase in bacterial burden during an exacerbation, despite the presence of negative clinical cultures [38]. Subjects experiencing an AE-IPF exhibited a marked change in the respiratory microbiome with an increase in Campylobacter and Stenotrophomonas species compared to stable disease. Both of these organisms are potential respiratory pathogens but the presence of Campylobacter also leads the authors to suggest their findings support the role of reflux and aspiration with AE-IPF. One weakness of the study was that there were only two paired samples from the same individuals when stable and experiencing an exacerbation. The stark changes in these two pairs of samples is intriguing, especially as in one individual there was a marked outgrowth of streptococcal species from the baseline microbiota which then dominated the exacerbation state. This result is concordant with findings from two distinct mouse models, where infection with Streptococcus pneumoniae induced exacerbation of established lung fibrosis [35].

More recently, Weng et al. [29] have employed molecular techniques to examine sputum cultures of 170 patients experiencing an exacerbation. They detected 38 different bacterial strains with gram-negative bacteria accounting for 89% of pathogens: Klebsiella pneumonia (26%), Mycobacterium tuberculosis (21%), and Acinetobacter baumannii (10%) dominated the sputum samples [29]. Together these studies demonstrate that alterations in the respiratory microbiome occur during acute exacerbations but they do not provide any functional or mechanistic insights. In a recent study, O'Dwyer et al. [37] tried to overcome this issue in an attempt to move from correlation to causation. A germ-free mouse model of fibrosis was used to explore the role of bacteria on lung inflammation and fibrinogenesis. Following bleomycin exposure, germ-free mice were protected from mortality, even though they exhibited similar severity of pulmonary fibrosis when compared to conventional mice [37]. The authors postulate that this may reflect the clinical observation that subjects with IPF not only die from progressive fibrosis but also from distinct inflammatory causes of acute-on-chronic respiratory failure including respiratory infections and AE-IPF. While there are several limitations of animal models in IPF, collectively, this study shows how preclinical germ-free models are a useful tool to investigate potential mechanisms of host-microbiota interactions and provides the first causal evidence that the microbiome participates in the pathogenesis and mortality of fibrotic lung disease.

Conclusions and future directions

The role of the respiratory microbiome in the pathogenesis and progression of IPF is just beginning to be elucidated. Clinically, acute exacerbations are important events in the natural course of IPF which carry a significant morbidity and mortality. While there is a paucity of information regarding the aetiology, pathophysiology and clinical management of acute exacerbations, an increasing number of findings indicate that infection, both viral and bacterial might be involved. Recent advances in sequencing technologies have allowed the use of molecular microbial technologies to characterise the respiratory microbiota in these patients demonstrating a clear change in the respiratory microbiome, as well as an increased bacterial burden in BAL compared to stable disease. However, the microbiome has not been fully characterised and further research should also investigate organisms other than bacteria and viruses, including fungi. It is unclear whether a higher bacterial load and an altered microbiome during an acute exacerbation reflect an active infection, increased aspiration or occur as a result of widespread diffuse alveolar damage. As sequencing DNA from BAL provides a snapshot in time of the microbial diversity, future research should incorporate longitudinal sampling before, during and after an AE-IPF. Furthermore, there is a pressing need to move from correlation to causation [39]. In order to do so, functional studies examining host-microbiome interactions as well as better models of AE-IPF are needed, given existing models fail to fully recapitulate physiologic findings in IPF. Current consensus is that all patients experiencing an exacerbation should receive broad spectrum empirical antibiotics even if no overt infection can be identified [40]. While this is likely driven by the use of immunosuppressive agents, the molecular microbial data may support this empiric therapy. Indeed, there is already some evidence that azithromycin may be beneficial in AE-IPF [41]. Ultimately, understanding the role of the microbiome in AE-IPF will allow its manipulation and may provide an opportunity for targeted therapeutic intervention.