Co-colonisation with Aspergillus fumigatus and Pseudomonas aeruginosa is associated with poorer health in cystic fibrosis patients: an Irish registry analysis

The male cohort made up the bigger subset of the CF population at 58.6%, reflecting higher mortality rates in females in adolescence and early adulthood [22]. The age distribution in our dataset was 4–69 years and the median age was 18.1 years. The demographic and clinical data of the cohorts are depicted in Table 1. As expected the FEV₁ percent predicted was lower in older patients reflecting a gradual decline with age (Fig. 1).

The prevalence of AF and PA co-colonisation was 3.1% in the Irish CF population. The AF + PA group included AFp + PA (n = 4), AF + PA (n = 5), AF + PAp (n = 7) and AFp + Pap (n = 8) subgroups. No significant differences in FEV₁, number of hospitalisations, number of respiratory exacerbations and antimicrobial prescribing were noted between these subgroups (Additional file 2: Table S1). AF colonisation was most prevalent in pre-adolescents and adolescents (Fig. 1), which contradicts current opinion that AF is more commonly an organism encountered in adulthood. To our knowledge this has not been previously reported. The prevalence of AF colonisation in the Irish CF population was 11%, of which 5% had persistent and 6% intermittent colonisation (Table 1). Previously AF prevalence rates in CF have been reported to be up to 58% using sputum culture [8]. PA becomes more prevalent with age and FEV₁ decline, corroborating previous findings (Fig. 1) [23]. The prevalence of PA in the Irish CF population was 31%, of which 19% had persistent and 12% had intermittent colonisation (Table 1). The prevalence of PA colonisation in Ireland is lower than the 48.7% prevalence reported in the US [17]. Employing similar criteria for the classification of persistent and intermittent PA colonisation as our study, Lee et al., (2003) [21] determined the prevalence of persistent PA to be 18% consistent with our data and intermittent colonisation to be 34%, which is higher than our findings. Proesmans et al., (2006) [24] determined the prevalence of persistent and intermittent PA colonisation to be 31 and 18% respectively which are higher than our prevalence rates. Many of these studies differ in the criteria employed for categorising colonisation status making comparisons between them difficult. We employed stringent criteria to classify colonisation status in line with the Leed’s criteria [21] as outlined in our methods. The Leed’s criteria have been independently evaluated [24] and are the classification criteria recommended in the European CF Society patient registry guidelines [25]. We did not consider positivity of AF and PA in one sputum sample in a year as intermittent, as other studies have done [6‐8, 26], to rule out inclusion of chance events, false positives or sputum contamination. For these reasons our prevalence for AF and PA colonisation may be lower than reported elsewhere [6‐9, 15, 21, 24, 26]. Consensus criteria for the definition of intermittent and persistent AF colonisation, comparable to the Leeds criteria for PA, need to be formulated. We acknowledge that a limitation of this study is its cross-sectional design and multi-year analysis of registry data would provide further detail on transition from intermittent to persistent colonisation status and the impact on disease outcome measures.

Patients co-colonised with both AF and PA had an FEV₁ 13.8% lower than non-colonised patients (p = 0.016) (Fig. 2) and co-colonised patients had comparable reductions in lung function to patients persistently colonised with PA. Consistent with our data, Amin et al., (2010) [9] reported that persistent infection with both AF and PA was associated with lower lung function. The transition of PA colonisation from intermittent to persistent is a disease milestone that cannot be reversed and is associated with worse prognosis [18, 27]. We have shown that patients co-colonised with PA and AF had similar reductions in lung function as patients with persistent PA. This is the first time this has been reported. In a recent longitudinal study the sputum microbiology and clinical outcomes of 770 adolescents with CF were recorded and AF was the only species that was associated with an increased risk for infection with PA [28]. This highlights the importance of monitoring the co-colonisation status of CF patients. Perhaps AF colonisation establishes a milieu that enables PA to become persistent and attention to AF during this phase may prevent progression to persistent PA. Further data is required to explore this hypothesis.

Patients with persistent PA colonisation had an FEV₁ 11.9% lower than patients who were intermittently colonised with PA (unadjusted p = 0.0074), which remained significant following adjustment for confounding factors (p < 0.001) (Fig. 2). Previously intermittent PA colonisation has been associated with lower FEV₁ [21] (when adjusted for age and gender only). Following data adjustment for confounding factors (age, gender and CFTR mutation) the FEV₁ of patients intermittently colonised with PA were not significantly different from the FEV₁ of non-colonised patients. Persistent PA colonisation negatively impacted on lung function in agreement with others [21, 23]. Early onset PA colonisation can be eradicated in over 90% of colonised patients by antibiotic therapy [29], therefore efforts should be targeted towards early detection of PA colonisation and preventing establishment of persistent colonisation. Interestingly 2 studies have now shown that polymerase chain reaction methods can detect PA 4.5 months [30] to 8 months [31] in advance of current culture-based techniques used commonly in diagnostic laboratories. This provides evidence to support the use of molecular-based methods for superior detection of PA.

Patients with intermittent and persistent AF colonisation had 4 and 11% lower FEV₁, respectively, than non-colonised patients however these were not statistically significant. A previous study showed patients with AF sensitisation or persistent carriage had 16.5% lower FEV₁ than the control group [32]. Amin et al., (2010) [9] showed that patients persistently colonised with AF had a 3.6% decrease in FEV₁ compared to patients who were clear of AF, however when the data was adjusted for baseline pulmonary function, this decrease was not significant. In agreement with our data, other studies suggest no correlation between AF colonisation and a decline in pulmonary function [33, 34].

For the first time we have shown that being culture positive for both AF and PA is associated with similar levels of hospital admissions, exacerbations and antimicrobial prescribing as being persistently colonised with PA (Table 2). Patients co-colonised with both AF and PA had a 165% increase in hospital admissions per person (p = 0.023), a 112% increase in the number of respiratory exacerbations per person (p = 0.042) and 48% increase in antimicrobial prescribing (p = 0.009) when compared to patients who were clear of both pathogens (Table 2). Co-colonised patients had a 64% increase in respiratory exacerbations compared to patients intermittently colonised with PA (p = 0.015) (Table 2).

Patients persistently colonised with PA had an 89% increase in the number of hospital admissions per person within the year (p = 0.004), a 91% increase in respiratory exacerbations (p < 0.001) and a 55% increase in antimicrobial prescribing (p < 0.001) compared to non-colonised patients (Table 2). In comparison to intermittently colonised patients, those persistently colonised with PA had a 60% increase in the number of exacerbations (p = 0.001) and 23% increase in antibiotic prescribing (p = 0.014) (Table 2). There was no increase in the number of hospitalisations, respiratory exacerbations or antimicrobial prescribing in patients who were intermittently colonised with PA compared to non-colonised patients.

Patients intermittently or persistently colonised with AF showed no increase in the number of hospitalisations or respiratory exacerbations per person compared to non-colonised patients. However intermittently and persistently colonised patients had a 63% and 90% increase in antimicrobial prescribing compared to non-colonised patients (p = <0.001 and p = 0.001 respectively) (Table 2). The prevalence of Aspergillus sp was previously shown to be higher in CF patients receiving prophylactic antibiotic therapy [33] and we also observed an association between AF colonisation and antimicrobial prescribing. We speculate that this finding may reflect antimicrobial suppression of bacteria facilitating fungal growth. It is also possible that AF colonisation establishes a milieu suitable for bacterial colonisation, which would impact on antimicrobial prescribing. No differences were observed between the cohorts regarding inhaled steroid use, steroids taken daily and steroids taken alternative days (Additional file 3: Figure S1).

A total of 14.7% of patients with CF had diabetes requiring insulin in 2013. Patients persistently colonised with PA had a higher prevalence of CFRD diagnosis than non-colonised patients (p = 0.012) (Additional file 4: Table S2). A direct link between insulin deficiency and clinical decline has been shown and patients with CFRD were more likely to be persistently colonised with Pseudomonas sp. [35].

The overall prevalence of ABPA was 5.9%. Patients co-colonised with AF and PA had the highest prevalence of ABPA (17.4%) but this was not significantly higher than the other groups (Additional file 4: Table S2). Interestingly, patients intermittently or persistently colonised with AF showed no increased prevalence of ABPA. It is not currently understood why some AF colonised patients are vulnerable to developing ABPA and others are not. A number of single nucleotide polymorphisms (SNPs) have been linked to susceptibility to ABPA. SNPs in the IL-4 receptor alpha chain (IL-4Rα) gene [36], the mannose-binding lectin (MBL) 2 gene [37, 38], the toll-like receptor 9 (TLR9) gene [39] and the pulmonary surfactant protein (SPA-2) gene [40] have been identified as genetic risk factors for the development of ABPA. In particular, SNPs in the promoter region of the IL-10 gene were found to be associated with Aspergillus colonisation and ABPA in patients with CF. Further studies in the CF setting could yield interesting insights into the susceptibility of CF patients to Aspergillus colonisation and development of allergic disease. Culture of AF from respiratory secretions of CF is only a secondary criterion for the diagnosis of ABPA, due to the large number of asymptomatically AF colonised CF patients. A retrospective study on 236 patients with CF found that 25% of patients had positive AF culture results but of these only 15 patients (6.5% of the total population) were ABPA positive [41]. In parallel with our findings, positive culture results for Aspergillus did not seem to represent a specific risk factor for ABPA [40, 42].

The CF airways are colonised with a diverse community of microorganisms from an early age [2‐4]. Interactions between the varieties of dominant and low abundance microorganisms in these communities are likely to impact patient health. The most commonly isolated bacteria and fungi from CF samples are PA and AF, respectively and we have shown that co-colonisation with these two microorganisms is associated with poor clinical outcomes. Previous studies have shown that PA and AF are capable of interacting and competing, which may have implications on CF airway disease. Pyocyanin and 1-hydroxy-phenazine produced by PA both have anti-AF activity [43] and PA has been shown to inhibit AF biofilm formation by direct cell contact and by excreted molecules [44, 45]. More recently the Pf4 bacteriophage produced by PA was shown to reduce both biofilm formation and pre-formed biofilms of AF and this effect was linked to inhibition of AF metabolism via Pf4 iron sequestration [46]. AF has been shown to enhance the production of PA elastase when in co-culture and these co-culture supernatants were more damaging to lung epithelial cell lines than PA supernatants alone [47]. We hypothesize that these PA-AF interactions may play a role in the damaging pathology associated with CF airway disease and may explain why patients who are co-colonised with PA and AF have a poorer prognosis than those who are clear of both pathogens. Further studies on the interaction of these two microorganisms are warranted to expand our understanding of how co-colonisation may impact the CF airways.