Rationale Epidemic Pseudomonas aeruginosa (PA) plays an important role in cystic fibrosis (CF) lung disease. A novel strain, the ‘Prairie Epidemic Strain’ (PES), has been identified in up to 30% of patients in Prairie-based Canadian CF centres.
Objective To determine the incidence, prevalence and long-term clinical impact of PES infection.
Methods A cohort of adults with CF was followed from 1980 to 2014 where bacteria isolated from clinical encounters were prospectively collected. Strain typing was performed using pulse-field gel electrophoresis and multilocus sequence typing. Patients were divided into one of four cohorts: no PA, transient PA, chronic PA with unique strains and chronic PES. Proportional Cox hazard and linear mixed models were used to assess for CF-associated respiratory death or transplantation, and rates of %FEV1 and body mass index (BMI) decline.
Results 274 patients (51.7% male) were analysed: 44––no PA, 29––transient PA, 137––unique PA, 64––PES. A total of 92 patients (33.6%) died or underwent lung transplantation (2423.0 patient-years). PES infection was associated with greater risk of respiratory death or lung transplant compared with the no PA group (aHR, 3.94 (95% CI 1.18 to 13.1); p=0.03) and unique PA group (aHR, 1.75 (95% CI 1.05 to 2.92) p=0.03). Rate of lung function decline (%FEV1 predicted) was greatest in the PES group (1.73%/year (95% CI 1.63% to 1.82%); p<0.001). BMI improved over time but at an attenuated rate in the PES group (p=0.001).
Conclusions Infection with PES was associated with increased patient morbidity through three decades and manifested in an increased risk of respiratory death and/or lung transplantation.
- Cystic Fibrosis
- Respiratory Infection
- Bacterial Infection
- Clinical Epidemiology
Statistics from Altmetric.com
What is the key question?
As the Prairie Epidemic Strain represents a prevalent transmissible strain of P. aeruginosa in Western North America, what is its clinical impact on morbidity and mortality in persons with cystic fibrosis (CF)?
What is the bottom line?
The ‘Prairie Epidemic Strain’ of P. aeruginosa is only the second epidemic strain that is associated with increased risk of death or lung transplantation, and has increased morbidity in patients with CF.
Why read on?
This cohort study represents the largest and longest study of a transmissible strain of P. aeruginosa to date, encompassing >2400 patient-years of follow-up, and links microbiological and clinical data over this period with clinical outcomes.
Cystic fibrosis (CF) is a genetic disease characterised by chronic progressive airway infection, punctuated by acute exacerbations, ultimately leading to respiratory failure. Pseudomonas aeruginosa (PA) remains the archetypal pathogen and causes chronic airway infection in 60%–70% of adults with CF.1–5 Following initial P. aeruginosa infection, the typical clinical course involves conversion to a mucoid phenotype, leading to increasing exacerbations, treatment burden and deteriorating clinical status.6–8 While P. aeruginosa infection is associated with increased morbidity and mortality in CF regardless of lung function,9 ,10 significant heterogeneity exists in the natural history of lung disease.
P. aeruginosa is commonly found in the natural and urban environmental niches, and thus patients with CF were thought to each harbour uniquely acquired strains.11–15 However, in the last three decades, multiple genetically related strains of P. aeruginosa infecting unrelated patients with CF have been described, determined to be transmitted between patients, and are thusly termed epidemic or transmissible strains. While epidemic P. aeruginosa (ePA) strains have been extensively studied, much remains to be understood regarding acquisition and natural history of infection.16 ,17 With the use of a comprehensive biobank (established in 1978), we identified a novel strain of P. aeruginosa (multilocus sequence typing (MLST) 192) infecting patients with CF attending our clinic.18 Termed the ‘Prairie Epidemic Strain’ (PES), it had a stable prevalence rate through multiple adult cohorts over 25 years, resulted in chronic stable infections and had increased antimicrobial resistance compared with unique strains.18 PES was found in 1/3 of the cohort and had a broad distribution in Western Canada.19
ePA are notorious for multidrug resistance potential and their association with adverse clinical outcomes. Transmissible strains have been associated with an accelerated rate of decline in pulmonary function to respiratory failure,20 increased risk of death or lung transplantation,21 increased frequency of pulmonary exacerbations22 ,23 and greater antibiotic treatment burden24 with toxicities.25 Accordingly, identification of transmissible strains has impacted clinical management and infection control policies on a global scale. However, these studies have been limited to short-term follow-up owing to their prospective nature. Our group, however, is ideally positioned to study the longitudinal clinical impact of ePA. Given the high prevalence of PES, persistence through time and widespread endemicity in Western North America, we sought to assess the clinical impact of the PES strain on adults with CF.
Materials and methods
The Calgary Adult Cystic Fibrosis Clinic (CACFC) provides care for all adult patients in Southern Alberta. Upon enrolment, patients consent to the routine submission and banking of bacterial isolates at each encounter. All sputum-derived pathogens are stored at −80°C with approval from the Conjoint Health Research Ethics Board (REB E23087).
Patient and bacterial strain selection
A longitudinal historical cohort study was conducted with adult patients with CF attending CACFC between 1 January 1980 and 30 December 2014. Patients were included if they were ≥18 years, had a confirmed CF clinical diagnosis including sweat chloride values ≥60 mmol/L and mutation analysis (provided the patient survived beyond 1993)26 and ≥3 months of follow-up (clinical+microbiological information). Data were recorded until 30 December 2014, the designated end point, or until censoring (ie, death, transplant, transfer). Baseline data including sex, genotype, CF-comorbidities, spirometry and body mass index (BMI), and longitudinal outcome data including dates and cause of death, and/or transplant, spirometry, BMI, sputum culture and exacerbation data, were collected through detailed chart and database review.
Patients were categorised by infecting P. aeruginosa strains as follows: (i) no history of infection, (ii) history of transient P. aeruginosa (unique or PES strain) infection where the strain was cleared spontaneously or with eradication treatments, (iii) chronic infection with unique strains and (iv) chronic infection with PES (including those patients who experienced strain displacement of their unique strains by PES as it occurred early within 2 years following clinic entry).18 Chronic infection was defined per Leeds criteria.27 For patients who developed chronic P. aeruginosa infection during or immediately prior to the study period, >50% positivity of a minimum of three cultures in a 12-month period was required. Stability of infection using previously established methods was assessed through sputum analysis at two time points: the first encounter (FE) and the most recent encounter (RE; prior to death/transplant/transfer or last clinic visit); annual sputum samples were screened for (i) patients when discordant strains were identified in order to determine when the change occurred and (ii) randomly in a subset of patients to ensure stability of infecting strains. From each relevant time point, up to five P. aeruginosa isolates from the CACFC biobank were typed, representing distinct morphotypes identified on MacConkey agar in real time, when samples were submitted as part of standard care.18
Pulsed-field gel electrophoresis (PFGE) was used as the primary screening procedure for each isolate and compared against 40 representatives of known epidemic strains as previously described.18 Isolates matching known ePA strains or those infecting ≥3 unrelated individuals were considered clonal a priori. Patients with any isolate with discordant PFGE profiles and select random isolates (quality assurance) underwent MLST using previously established methods to ensure accuracy of strain typing.28–31 Following this, patients with discordant PES isolates after the initial encounter were considered a priori to have a ‘superinfection’ wherein they underwent strain displacement. A random sample of isolates from patients with PES and unique P. aeruginosa strains identified by PFGE underwent confirmatory testing with either whole genome sequencing (WGS),32 ,33 MLST30 or a qualitative PES-specific PCR assay.34
Symmetrical and asymmetrical variables were reported as means with SDs and medians with IQRs, respectively. Continuous variables and proportions were compared between groups using t-tests and χ2 tests, respectively. For all long-term outcomes, multivariable models were used with baseline and time-varying data by adding in variables with p<0.15 in univariate analysis, or those with a strong a priori hypothesis (ie, sex, age) in a stepwise manner.35–38 For death and transplant outcomes, Kaplan-Meier survival analysis was used to conduct unadjusted between-group comparisons. Cox proportional hazard models using follow-up time were used to determine HRs for time to respiratory death or lung transplantation. To assess for left truncation bias in the Cox model, a second analysis using calendar time was conducted. Lung function (%FEV1) and BMI decline were modelled using a mixed effects model with random slopes and intercepts. Other secondary outcomes were assessed using linear generalised estimating equations. Study outcomes were established prior to data collection and analysis. Analyses were two-sided with α-significance level of 0.05 using STATA V.13.1 software (College Station, Texas, USA).
A total of 306 patients were screened (data collection+sputum typing if available), and 274 (51.7% male) were included in the analysis (table 1). Patients were excluded if they lacked ≥3 months of clinical details, had not contributed to the biobank or had chronic infection with epidemic strains of P. aeruginosa (ePA) other than PES. Of those excluded, 20 patients had no sputum for typing, 5 patients had an untypeable strain and 7 patients had other transmissible strains (two––Type B, five––Liverpool epidemic strain (LES).21 Those with established infections with other transmissible P. aeruginosa strains had transferred to the CACFC at a median age of 24.6 years from CF centres in Eastern Canada. There were no significant baseline demographic differences with regard to age, sex, BMI and lung function of the excluded group compared with the included cohort (table 1) (p>0.05).
The mean (SD) study entry age was 23.1 (8.2) for the total cohort. A total of 1308 P. aeruginosa isolates were typed by PFGE, and of these, an additional 108 underwent MLST and 86 whole genome sequencing32 ,33 (MG Surette, Professor, McMaster University, personal communication, 2016). The transient P. aeruginosa group comprised of 27 patients with unique strains and 2 patients with the PES strain. A small number of patients with chronic P. aeruginosa had mixed infection with more than one strain of P. aeruginosa, and were categorised as unique if both strains were unique, and as LES if infected with a unique and LES strain (two patients); no patients had mixed infection with PES as a constituent. Patients with chronic PES had a median of 4 isolates typed (IQR 3–7, up to 32), and those with chronic unique PA had a median of 3 (IQR 2–4, up to 11) typed.
Assessment of prevalence rates of PES infection relative to all patients with chronic P. aeruginosa infection through three decades of patients enrolling at the CACFC demonstrated rates of 37.5% prior to 1985 (3/8 patients), 20.5% for 1986–1990 (8/39 patients), 32.2% for 1990–1995 (19/59 patients), 35.9% for 1996–2000 (14/39 patients), 23.4% for 2000–2005 (13/55 patients) and 8.51% for 2005 onwards (8/94 patients). Rates were similar with the exception of the earliest cohort compared with the cohort 2005 onwards (37.5% (95% CI 3.9% to 71.0%) vs 8.5% (95%CI 2.9% to 14.2%), p=0.01) and were concordant with our prior findings.18 The mean duration of follow-up for patients in each group was: (i) no P. aeruginosa—5.61 years; (ii) transient infection—10.86 years; (iii) unique chronic P. aeruginosa—9.22 years and (iv) PES—12.33 years. The unique chronic P. aeruginosa and PES group had a higher proportion of females, and lower baseline weight, BMI and lung function compared with the group with no P. aeruginosa infection.
A total of 10 patients (60.0% male, mean age 27.4 (SD 4.9) years) developed new PES infections in the study period. Eight PES infections occurred in patients with chronic P. aeruginosa infection through ‘superinfection’, and two occurred in patients without a history of chronic P. aeruginosa. These infections occurred evenly through the study period, occurred within 2 years of entry into the CACFC, and most recently in 2013. In those experiencing strain displacement, we did not discover the PES strain on testing of any of the early isolates by PFGE or PCR.34 Of the two patients with PES infection with the absence of antecedent chronic P. aeruginosa infection, one underwent transplantation within 3 months of the event, and the second cleared the infection after eradication therapy. The incidence rate of new PES infection was 4.2 per 1000 person-years (95% CI 1.6 to 6.9 per 1000 person-years). Mean %FEV1 decline was greater following PES infection (−2.20% (−1.83% to −2.65%) vs −1.85% (−1.31% to −2.42%) per year), but did not achieve statistical significance. Conversely, BMI had a declining trend in this group following PES infection (−0.18/year (−0.01 to −0.33), p=0.04). In patients with incident PES infections, eight died (pretransplant respiratory death) or underwent lung transplantation, resulting in an incidence rate of 1.5 events/10 person-years.
Of 274 patients, a total of 92 patients (33.6%) died or underwent lung transplantation over a 35-year period (2423.0 total patient-years). Death or transplantation occurred in six patients (13.6%) with no history of P. aeruginosa infection, four patients (13.8%) with history of transient P. aeruginosa infection, 44 with (32.1%) unique chronic P. aeruginosa infection, and 38 (59.4%) in the PES group. Of the 61 deaths, 32 (52.5%) occurred prior to transplant due to end-stage lung disease based on detailed review of chart and database records including autopsy reports by two assessors (RS and MDP). The remainder of the deaths were following lung transplant (19), or due to other causes (2—cancer, 1—suicide, 1—overdose, 7—unknown).
Chronic PES infection was associated with a significantly greater risk of CF-associated respiratory death or lung transplantation compared with patients with no P. aeruginosa infection (unadjusted HR, 4.16 (95% CI 1.28 to 13.6) p=0.02; adjusted HR, 3.94 (95% CI 1.18 to 13.1) p=0.03) or those with chronic infection with unique P. aeruginosa strains (unadjusted HR, 1.77 (95% CI 1.11 to 2.82) p=0.02; adjusted HR, 1.75 (95% CI 1.05 to 2.92) p=0.03; figure 1). Chronic infection with unique P. aeruginosa strains was not significantly associated with greater risk of respiratory death or lung transplant compared with patients with no history of P. aeruginosa infection (unadjusted HR, 2.33 (95% CI 0.72 to 7.58) p=0.16; adjusted HR, 2.25 (95% CI 0.69 to 7.37) p=0.18). Genotype (deltaF508 homozygous HR 1.89 (95% CI 1.04 to 3.39)), %FEV1 predicted (HR 0.98 (95% CI 0.98 to 0.99)) and BMI (1.09 (95% CI 0.98 to 1.21)) were included in the adjusted models. When calendar time was used in the Cox model, the adjusted HR for PES infection was similar to that in the primary analysis (HR 4.11 (95% CI 1.41 to 11.98) p=0.01). When a sensitivity analysis was performed by including the unknown deaths into the Cox model, chronic PES infection was still associated with the risk of respiratory death and transplant compared with the patients with no P. aeruginosa history, although to a lesser degree (adjusted HR, 2.95 (95% CI 1.02 to 8.54) p=0.046).
Over the course of the 35-year study period, the annual mean rate of decline of patients with chronic PES infection (1.73% (95% CI 1.63% to 1.82%)) was significantly greater than those chronically infected with unique strains (1.49% (95% CI 1.42% to 1.56%) p<0.001) and those with no P. aeruginosa infection (1.18%, (95% CI 0.96% to 1.40%) p<0.001) (figure 2). Patients infected with PES had a lower BMI than those without P. aeruginosa at study entry, and slightly lower than those infected with unique strains (table 1). BMI increased in all groups over time, but the rates of mean BMI change were significantly less in both the PES (0.10/year, 95% CI 0.07 to 0.13; p=0.001) and unique chronic P. aeruginosa groups (0.10/year, 95% CI 0.08 to 0.13; p=0.001) compared with the group without P. aeruginosa infection (0.23/year, 95% CI 0.16 to 0.30) (figure 3).
Pulmonary exacerbations were measured by home, hospital and total IV antibiotic days in the study period. Although the mean number of IV antibiotic days/year was greatest in the PES group in all categories, it did not achieve statistical significance relative to the other groups. Patients with chronic unique P. aeruginosa or PES infection had a significantly greater mean annual number of hospital admissions, admitted days in hospital and clinic assessments compared with patients without P. aeruginosa infection (table 2). However, no significant differences were noted between the chronic P. aeruginosa (unique and PES) infection patient groups with regard to these outcomes.
Although PES is a newly described transmissible P. aeruginosa strain, it has infected patients with CF for more than three decades, and is likely pervasive in Western Canada with potential extension to the proximal Northwest USA.18 We therefore sought to determine if PES was associated with adverse outcomes in adults with CF in the pretransplant period.
Transmissible P. aeruginosa strains in CF are common, and a number have been well described in Europe, Australia and North America.39 ,17 ,40–42 Perhaps, the most notorious epidemic strain, LES, has been identified in >15% of patients with CF in the UK.43 Aaron et al21 conducted a prospective study in Ontario, Canada to assess the prevalence and outcomes of transmissible P. aeruginosa strains and demonstrated that 15% of patients were infected with LES. The patients with CF infected with LES had an increased rate of death or lung transplantation compared with those infected with unique P. aeruginosa strains, but had a similar rate of pulmonary function decline. This contrasts the work of Jones et al, where an increased rate of lung function decline was observed in patients with LES relative to those with chronic unique infections, but not a disproportional rate of progression to end-stage lung disease.24 In our study cohort, we observed an association for the first time with an epidemic strain of P. aeruginosa, with both increased rates of lung function decline and progression to end-stage lung disease.
Herein, we determined P. aeruginosa strain type predominantly by PFGE owing to the reduced costs, efficiency and established local expertise. Any patient with more than one strain type was assessed for an alternate strain type confirmation using WGS and MLST. Further, we confirmed our findings with random audits using WGS, MLST and a qualitative PCR assay developed specifically to identify PES.44 No cases of misidentification were observed in our audits, and the large number of screened isolates by PFGE for each patient makes it exceedingly unlikely that patients were misclassified.
When clinical outcomes were assessed as a function of P. aeruginosa infection, we observed a significantly greater risk of pretransplant CF respiratory death or lung transplantation in the PES group. PES infection resulted in a greater rate of pulmonary function decline, and likely led to the increased risk of lung transplantation seen in this group. Interestingly, the BMI of patients increased over the study period in all patient groups, but the mean BMI was lowest in the PES group at baseline and increased at a lesser rate as compared with patients without P. aeruginosa infection. This improvement in nutritional status has been well described in more recent cohorts,45 ,46 and likely follows recognition of early work on the importance of nutrition in slowing progression of chronic lung disease.47 ,48 PES was overall associated with lower baseline status, and worsened clinical outcomes in the pretransplant period. Conversely, in a recently completed study assessing the post-transplant outcomes of PES in CF, although patients were likely to be younger at the time of transplant, no significant differences were noted in survival, or secondary outcomes including acute cellular rejection, or development of bronchiolitis obliterans syndrome.49
For the other secondary outcomes, both chronically infected P. aeruginosa patient groups had a greater number of hospital admissions, admitted days and clinic assessments annually. We observed a trend in the PES group relative to the other groups in increased need for IV antibiotics, though it did not reach significance. This is in contrast to other works with the likes of Manchester epidemic strain, Australian-2 and LES strains where increased treatment requirements were clearly identified.21 ,39–41 Additionally, treating clinicians were unaware of P. aeruginosa status owing to the retrospective nature of the analysis, and although similar to Australian CF physicians,40 this was in contrast to UK studies. Consequently, knowledge of infection status may have lowered their threshold to treat due to concerns of the potential deleterious outcomes of ePA strains.
We observed a declining prevalence of PES infection in sequential cohorts of young adults transitioning to our adult CF clinic, concomitant with changes in CF attitudes and policies regarding patient interaction and infection control. Notably, incident infection with ePA was still uncommon in historical cohorts—even prior to the introduction of policies to reduce social interaction among patients. Aaron et al observed an incidence rate for ePA infection of 7.0 per 1000 person-years, which was similar to our findings with the PES strain (4.2 per 1000 person-years). However, environmental sampling did not demonstrate any transmissible P. aeruginosa strains, which is similar to the findings in other studies, and highlights the need for increased understanding of transmission events. This contrasts data which suggest that P. aeruginosa aerosol generation through cough is common, highlighting the complexity of infection acquisition.50–52 Indeed, in a recent study by our group, we observed a significant association of transmissible PA acquisition with past social behaviours, highlighting further that periods of social contact may pose substantial risk.53 ,54
The primary strength of this study is the large sample size and extended duration of follow-up of 35 years, and approximately 2400 patient-years. This study was powered to examine relevant long-term clinical outcomes including death, transplant and lung function decline. Additionally, as we typed >1300 strains and each patient had early and late sputum isolates tested, new infection or strain replacement was readily identified. Although we had a larger number of primary outcome events of death or transplantation relative to prior works21 ,24 (92 events), this was still relatively low as was reflected by the wide CIs. As patients with chronic P. aeruginosa infection (unique and PES) had lower baseline values of lung function or nutritional status, the precise casual mechanisms in relation to P. aeruginosa acquisition could not be determined with the study period and design. Our study primarily assessed prevalent cases of P. aeruginosa, thus limiting our ability to draw causal inferences relating to PES infection. Other limitations included the retrospective study design, effect modification by time, selection bias (ie, sampling of morphotypes), information bias (ie, missing data, loss to follow-up) and confounding not accounted for in the analysis. Regardless, our study was the first to assess clinical outcomes associated with a transmissible P. aeruginosa strain over multiple decades of patient follow-up time. Moving forward, collaborative studies with other centres will enable us to improve our understanding of the natural history, and transmission characteristics of PES, and more broadly, of transmissible P. aeruginosa strains in CF.
Our study demonstrates that patients with chronic PES infection, as an example of an epidemic strain of P. aeruginosa, have a greater risk of CF-associated respiratory death or transplant, as well as an accelerated rate of %FEV1 decline with progression to end-stage lung disease. However, the incidence of PES infections in adults with CF remained low even in times of encouraged socialisation. Active surveillance of paediatric and adult patients with CF is recommended for early identification and management of ePA infection given its associated deleterious effects on the CF population.
This work was presented in abstract form at the European Cystic Fibrosis Society Conference in Brussels, Belgium in June 2015:JCF;14(S1):S74.
Contributors All authors meet the criteria for authorship as required by Thorax submission guidelines. RS and JCL were primarily responsible for data collection and analysis. BW, CDS and SP were responsible for bacterial strain typing and analysis. HRR and MDP were responsible for collection and maintenance of the CACFC biobank and clinical care records and documentation. RS was responsible for the creation of the manuscript. MDP, MGS and HRR were responsible for the project's inception and supervision. All authors contributed to the development of the final manuscript. MDP serves as guarantor of the work.
Funding Cystic Fibrosis Canada (10001919).
Competing interests HRR has been on advisory boards for Novartis and Vertex. MDP has been on advisory boards for Gilead, Novartis and Roche Pharmaceuticals. MDP, HRR and MGS have received research support from Gilead. MDP and MGS are supported by Cystic Fibrosis Canada research grants funding this work. RS is supported by a Cystic Fibrosis Canada Clinical Fellowship.
Ethics approval The study was approved by the Conjoint Health Research Ethics Board at the University of Calgary (REB#E23087 & REB#152744).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.