A high-risk airway mycobiome is associated with frequent exacerbation and mortality in COPD

Discussion

Here, we report the largest multicentre evaluation of the airway mycobiome in COPD, including participants recruited in Asia (Singapore and Malaysia) and the UK (Dundee), and additionally evaluate longitudinally mycobiome profiles during and following exacerbations.

The COPD mycobiome is diverse and demonstrates geographic variation. Very frequent COPD exacerbators are characterised by altered mycobiomes and increased interfungal interaction; however, mycobiome composition and diversity remain unaffected by the exacerbation event itself, antibiotics and/or corticosteroid use, suggesting that perturbation of the COPD mycobiome results from factors other than treatment. Two important patient clusters were identified based on mycobiome signatures: Cluster 1 characterised by significant symptoms and Saccharomyces, and Cluster 2 characterised by exacerbations, higher mortality, and Aspergillus, Curvularia and Penicillium. The latter group demonstrates systemic immune responses to these fungi. Lower mycobiome diversity during exacerbations represents an important risk for 2-year mortality.

In line with work in bronchiectasis, we observe differing mycobiome profiles between participants recruited in different countries [2]. The detected geographic variability may be attributed to contrasting climates, humidity and air quality, all important for fungal growth and survival. Host factors, including genetics, lifestyle and dietary differences, may contribute further, but remain beyond the scope of this work. Importantly, the respiratory tract represents an organ system in continuous communication with the external environment, i.e. a rich fungal source, allowing temporary passage and potentially retention of environmental fungi under appropriate conditions. The importance of geographic location and the specific environmental influence on host mycobiome profiles in relation to COPD has also been the subject of prior work [7]. Using metagenomic sequencing of outdoor and indoor air, a measurable sensitisation response to environmental fungi relates to COPD outcomes. Taken together with the findings presented in this study, geographic location and local environment remain important determinants of the host mycobiome.

Prior studies in COPD demonstrate alterations in the bacterial microbiome, particularly in frequent exacerbators [26, 27]. This current study identifies significant changes to the airway mycobiome in very frequent COPD exacerbators (three or more exacerbations per year) who interestingly illustrate a more complex mycobiome characterised by increased antagonistic interfungal interaction. COPD exacerbations may promote initial bacterial dysbiosis and disruption to fungal communities may only occur as exacerbation frequency increases. In addition to an alteration in fungal community membership with increased exacerbation frequency, dynamic variation also occurs. The increased negative (co-exclusive) interactions observed in very frequent COPD exacerbators may result in fungal overgrowth, particularly of specific fungal taxa such as Wickerhamomyces, which in turn drives the observed clinical phenotype. These fungi are widespread in natural habitats, including soil, trees and plants; however, there has been a reported case of clinical fungaemia due to Wickerhamomyces [28]. Although very frequent COPD exacerbators generally receive significantly more treatment, including antibiotics and/or systemic steroids, compared with nonexacerbators, longitudinal mycobiome analysis reveals its relative stability despite such treatment and even during exacerbations, consistent with work performed in CF, suggesting alternate mechanisms to explain the observed fungal dysbiosis in very frequent exacerbators [29, 30]. As bacteria and viruses (rather than fungi) are commonly identified triggers of COPD exacerbations, it is not unexpected that the airway mycobiome remains stable during AECOPD in our longitudinal analysis. However, with increased exacerbation frequency and its associated immunoinflammatory change, a “threshold” in very frequent COPD exacerbators may be reached where alterations to the airway mycobiome begin to appear. Such a change, occurring over time, likely involves a complex interaction between microbial kingdoms, the surrounding environment and the underlying host immunoinflammatory response rather than an acute alteration during a single COPD exacerbation where bacterial and/or viral change may be more apparent [16]. Therefore, the airway mycobiome in COPD is likely most useful in identifying patients at greatest risk of longer-term adverse outcomes, including a risk of frequent exacerbations and mortality [31, 32]. Interestingly, inhaled corticosteroid use demonstrates no impact on mycobiome profiles. This contrasts with bacteriome studies in COPD, where microbiome change is observed over the course and treatment of exacerbations, which are influenced further by COPD therapy [16]. Differences in study outcomes between bacterial and fungal microbiomes suggest that bacteria and fungi play different roles during exacerbations and are differently impacted by COPD treatments.

While our current understanding of the mycobiome's contribution to respiratory disease pathogenesis remains indeterminate, available data does suggest that specific mycobiome profiles do exist in chronic respiratory disease states, with an increased abundance of Candida and Aspergillus in CF, an increased abundance of Psathyrella, Malassezia, Termitomyces and Grifola in asthma, and an increased abundance of Aspergillus, Cryptococcus and Clavispora in bronchiectasis [2, 20, 33–35]. In the context of COPD, prior mycobiome studies have focused purely on the HIV-infected COPD patient group where the key change to airway mycobiomes observed was an increased abundance of P. jirovecii [22]. Importantly, our work did not detect any Pneumocystis fungi in our COPD cohorts; however, we did identify specific fungal taxa associated to COPD, including Alternaria, Aspergillus, Cladosporium, Cryptococcus, Mycosphaerella, Wickerhamomyces, Trametes and Penicillium. Differences in the chosen study populations and methodology used likely accounted for at least some of these differences.

Unsupervised clustering analysis reveals two clinically important COPD patient groups: one (Cluster 1) with increased symptoms (CAT score) and Saccharomyces dominance, and another (Cluster 2) with increased exacerbations and higher mortality characterised by Aspergillus, Curvularia and Penicillium. In the latter group, systemic sIgE responses to Aspergillus, Curvularia and Penicillium were detectable. The first group, i.e. characterised by increased COPD symptoms and Saccharomyces, is particularly interesting because existing data in COPD and, more recently, bronchiectasis suggest that the presence of symptoms can indicate a risk of future exacerbations and a response to treatment [36–40]. Whether these concepts hold true regarding airway fungi in COPD remains to be established. The second group, i.e. characterised by Aspergillus, Curvularia and Penicillium, should be considered “high risk”, and warrants early identification and close clinical follow-up. Aspergillus, Curvularia and Penicillium are thermophilic fungi with survival capabilities in the human airway and pathogenic potential. Importantly, these fungi commonly associate with ABPM, and airway isolation of Aspergillus and Penicillium has been associated with sensitisation and reduced lung function in asthma [41]. The importance of fungal sensitisation in COPD has recently been described by our group and contributes to COPD exacerbations in line with earlier studies in asthma, where fungal sensitisation demonstrates key roles in disease severity and exacerbations [3, 7]. Therefore, it is plausible that increases in specific airway fungi in COPD activate host immunity, with resulting sensitisation and poorer clinical outcomes [7]. Importantly, despite striking geographical, environmental and ethnic differences, both clusters were consistently identified in participants from Asia and the UK.

Evaluating the COPD mycobiome during acute exacerbations reveals that a loss of diversity is a signal for increased 2-year mortality. This is consistent with prior bacteriome studies in COPD where survival, assessed at 1 year following exacerbation, demonstrates a similar lack of diversity [17]. Fungal genera characterising poorer survival include Penicillium, Cladosporium, Trametes and Lodderomyces. Of interest, Penicillium also characterised the “high-risk” cluster in the stable COPD state, suggesting that its specific role in COPD pathogenesis warrants further study. The fungi identified in the nonsurvivors have been associated with invasive fungal infection and/or sensitisation in humans; however, their specific roles in COPD remain to be fully expounded. The COPD mycobiome clearly holds important information with potential prognostic implications and may be used to identify “high-risk” patients with worse 2-year survival following an acute exacerbation.

Our study is the largest and, to the best of our knowledge, the first multicentre COPD study to evaluate the airway mycobiome including longitudinal sampling. Despite its clear strengths, our work is limited by a small healthy cohort for comparison, all recruited from a single site (Singapore), and therefore may not be generalisable to the wider healthy population. In addition, our AECOPD cohort was relatively small (n=66) and therefore the survival analysis is limited; however, it does represent the largest study of the airway mycobiome in COPD to date, complete with 2-year survival data. Further investigations building on this work in larger cohorts will help to explicate the clinical significance and survival-related aspects of our findings. While we report unique mycobiome signatures in COPD, we did not specifically exclude patients with bronchiectasis–COPD overlap (24% (81 out of 337)), although importantly, no differences between mycobiome profiles were detectable between patients with COPD (n=256) and patients with bronchiectasis–COPD overlap (n=81). In addition, we did not assess the COPD mycobiome's potential interaction with other microbiomes (bacterial and viral). Future studies integrating the various microbiomes accounting for the host response will be important in understanding the specific impact of the mycobiome in COPD pathogenesis. Despite optimising our internal transcribed spacer protocol as previously published, inherent limitations persist in the field, including weak fungal reference databases that preclude confident resolution to the species level [42]. In addition, targeted amplicon sequencing lacks functional annotation when compared with metagenomic sequencing approaches, and therefore our findings presented here can only describe associations and cannot imply causation or demonstrate a pathogenic role for any of the detected fungal taxa. Future work focused on mechanistic confirmation should be pursued. Unlike bacterial microbiome studies in COPD [19, 43], we did not detect associations between disease severity and the COPD mycobiome; however, it does appear that the mycobiome composition is influenced beyond a certain “threshold” of COPD exacerbation frequency, exemplified by our “very frequent exacerbator” group. The COPD mycobiome may therefore represent a “marker” for specific disease traits in COPD that requires further study. Finally, our systemic sIgE assays were performed only in a subset of participants and with selected fungi, and this an avenue for further exploration. A validation of our study findings with a more comprehensive fungal allergen panel and inclusion of other ethnic populations should be pursued in future work.

In summary, we show that the COPD mycobiome provides important information in terms of clinical outcomes and prognosis, and should be further evaluated.