Infection With Transmissible Strains of Pseudomonas aeruginosa and Clinical Outcomes in Adults With Cystic Fibrosis FREE

Pseudomonas aeruginosa is a gram-negative bacterium that causes chronic endobronchial infections in 60% to 70% of adult patients with cystic fibrosis (CF).¹ Infection with P aeruginosa is associated with increased morbidity and mortality for patients with CF, irrespective of lung function.^{2 - 3} However, there is heterogeneity in the type and timing of outcome among those who are infected with P aeruginosa ; some patients experience a rapid decline in pulmonary function after infection and others harbor the organism for extended periods without any obvious adverse effects.⁴ The marked difference in prognosis among patients with P aeruginosa has not been adequately explained, but it may be due in part to differences among infecting strains.⁵

P aeruginosa transmissible strains are genetically identical strains that infect unrelated patients with CF. Transmission of P aeruginosa between siblings with CF has been well documented; however, with the exception of siblings, most patients are thought to be infected with genotypically unique strains of P aeruginosa acquired from the environment.⁶ However, reports first emerged in 1996 of a transmissible strain of P aeruginosa discovered among patients with CF in Liverpool, England.⁷ This transmissible strain was later referred to as the Liverpool epidemic strain.⁸ Further studies showed evidence of cross-infection with the Liverpool epidemic strain between patients with CF located in 15 centers in the United Kingdom.⁹ Recent reports of transmissible epidemic strains of P aeruginosa have also originated from Australian CF clinics in Melbourne and Brisbane^{10 - 11} and from a CF clinic in Manchester, England.¹²

Transmissible strains of P aeruginosa have not been described in North American patients with CF. One study from British Columbia, Canada, found that with the exception of sibling pairs with CF, patients in British Columbia were infected with unique rather than common strains of P aeruginosa.¹³ The objective of our study was to perform a multi-year prospective study of all adult patients with CF in the province of Ontario (total population: 13 million) to determine whether patients with CF were infected with transmissible strains of P aeruginosa, and if so, to determine the prevalence of infection and the incidence rates of new infection with these strains. Our second objective was to determine if infection with these strains of P aeruginosa was associated with clinically important adverse outcomes.

From September 2005 to September 2008, all adult patients with confirmed CF who attended 1 of the 7 Ontario adult CF clinics or smaller outreach clinics were approached for the 3-year prospective observational cohort. These 7 clinics and their outreach programs provide secondary and tertiary care to more than 98% of all adult patients with CF in Ontario. Patients were included in the study if they were aged 18 years or older, able to spontaneously produce sputum, and if they had a confirmed diagnosis of CF made via genetic analysis and/or sweat testing. Patients were followed up prospectively for 3 years or until December 31, 2009, when the study ended. Study follow-up was coordinated during regularly scheduled appointments at the patients' CF clinic. The research ethics boards of the participating centers approved the study, and all participants provided written informed consent.

Patients provided sputum samples on entry to the study, at 3 months, and yearly thereafter for 3 years. Sputum samples were transported on ice to the central laboratory in Ottawa, Ontario, Canada. Sputum was plated onto selective and nonselective media to detect P aeruginosa and other bacterial pathogens. If P aeruginosa was present in the sputum, 2 distinct P aeruginosa colony morphotypes from each sputum sample were selected for molecular typing, and 5 P aeruginosa isolates derived from each sputum sample were stored frozen in a mixture of brain heart infusion broth and glycerol (20%) at −70°C. A subsequent sputum culture, taken from the same patient 3 months later was similarly processed. In this fashion, 4 P aeruginosa isolates were typed from each patient over the initial 3-month period of the study to determine prevalence of infection with transmissible strains in the inception cohort. Similar procedures were followed for sputum samples obtained at 1, 2, and 3 years postenrollment.

Clinical data, including spirometry, body mass index (BMI; calculated as weight in kilograms divided by height in meters squared), and exacerbation history were collected from each patient at baseline and then annually for 3 years. Pulmonary exacerbations were defined as acute or subacute worsening of a patient's respiratory symptoms that were severe enough to warrant oral or intravenous treatment with antibiotics, which was provided at the discretion of the patient's physician. Patients who underwent lung transplantation before the 3-year end-of-study date did not contribute spirometry, BMI, or exacerbation data after their lung transplant date. Spirometry was performed according to the American Thoracic Society's standards and predicted values from Crapo et al¹⁴ were used.

Molecular genotyping of each P aeruginosa isolate was performed using pulsed-field gel electrophoresis (PFGE). Genomic DNA was prepared using a modification of a described method.¹⁵ The restriction fragment profiles were compared visually and also using BioNumerics computer software (Applied Maths, Kortrijk, Belgium), and were interpreted based on guidelines recommended by Tenover et al.¹⁶ Isolates with identical restriction fragment profiles were considered to represent a single strain. Isolates with restriction profiles that differed by 1 to 3 fragments were considered to be closely related strains evolving from a single clone. Isolates with profiles differing by 4 or more fragments were considered different strains and therefore assumed to be unrelated.

All patients who had sputum cultures positive for at least 1 common transmissible P aeruginosa strain isolated from 1 of the 2 initial sputum samples taken 3 months apart were considered to be infected with a transmissible Ontario strain in the inception cohort. For patients who could be potential new incident cases of infection with a transmissible strain, we retrieved and genotyped all 10 frozen P aeruginosa isolates from the sputums collected at baseline and at 3 months to ensure that we had not missed infection with a transmissible strain at the initial assessment. By doing this, we minimized the risk of falsely labeling a case as an incident infection when the infection may have been present at the initial assessment.

Multilocus sequence typing (MLST) was used to assess the genetic similarity of strains and was performed on P aeruginosa isolates from 35 patients. Of these patients, 10 patients were infected with strain A, 10 were infected with strain B, and 15 were infected with unique strains of P aeruginosa as determined by PFGE. Descriptions of the polymerase chain reaction primers used to amplify and sequence the 7 housekeeping genes (acsA, aroE, guaA, mutL, nuoD, ppsA, and trpE) for P aeruginosa have been published.¹⁷ The amplification product was visualized on a 1.5% agarose gel and purified from the gel using the QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany). The nucleotide sequences were determined using internal-nested primers and using standard sequencing techniques.¹⁷ Sequences were compared with those available at the P aeruginosa MLST online database (http://pubmlst.org/paeruginosa/) to determine allele numbers and sequence types.

Eight P aeruginosa isolates were obtained from Australia (2 each representing the dominant Melbourne, Brisbane, and Tasmanian transmissible strains and 2 from the second most common transmissible strain in Brisbane). Twenty isolates were obtained from the United Kingdom representing the Liverpool, Manchester, Midlands 1, and Clone C strains from Birmingham. These isolates were genotyped in our laboratory using PFGE and MLST and the results were compared with the Ontario transmissible strains.

Incident cases were patients not infected with an epidemic P aeruginosa strain based on sputum analysis performed at both study entry and 3 months, but who were discovered to be infected with such a strain on subsequent yearly sputum tests. Bacterial sampling of the incident environment (eg, patients' home, hospital, or CF clinic) was performed to rule out environmental or nosocomial acquisition of transmissible strains. Exposure to other patients with CF was ascertained via a patient questionnaire and deidentified data were linked to the study database to determine if incident cases had reported exposure to other patients known to be infected with epidemic strains.

Patients were grouped according to the following infection status categories: (1) did not grow P aeruginosa in their sputum; (2) infected with unique strains of P aeruginosa ; and (3) infected with either transmissible P aeruginosa strain A or B. Statistical comparisons between groups were limited to the strain A and B groups compared with the group infected with unique strains of P aeruginosa because these were the most clinically relevant comparisons. Continuous variables between the groups were compared using t tests and proportions were compared using χ² tests as appropriate. Random-effects mixed-linear models were used to compare the rates of decline in forced expiratory volume in the first second of expiration (FEV₁) and BMI over the 3-year study period in the patients infected with each of the 2 transmissible strains of P aeruginosa vs the corresponding rates in those patients infected with unique strains of P aeruginosa. Group × time interactions were analyzed using SAS PROC MIXED (SAS Institute Inc, Cary, North Carolina). Potential confounding effects of age, sex, BMI, baseline FEV₁, CF comorbidities (pancreatic insufficiency, diabetes, and chronic liver disease), infection with Burkholderia cepacia complex, and chronic treatment with azithromycin, dornase alfa, inhaled tobramycin, and inhaled colistin were assessed in the mixed-linear models. These confounding variables were decided a priori. Kaplan-Meier survival methods were used for between-group comparisons of unadjusted time to death or lung transplant. Cox proportional hazards models adjusted for the patient covariates were used to compare time to death or lung transplant. The clinical outcomes assessed were all decided a priori before data exploration. All statistical testing was 2-sided and was performed at a significance level of .05 using SAS software version 9.0 (SAS Institute Inc).

Of 580 patients approached to enter the study, 446 patients were enrolled (Figure 1). The mean (SD) age of the 446 enrolled patients was 29.3 (9.6) years and 255 were male (57%). The mean duration of study follow-up for patients infected with strain A of P aeruginosa was 853 days, strain B was 890 days, and unique strains was 868 days. Full 3-year follow-up was not available for 61 patients infected with P aeruginosa (14 infected with strain A, 8 infected with strain B, 1 infected with both strains A and B, and 38 infected with unique strains), who enrolled in the study after January 2007. These patients were followed up until December 2009. Vital status (death or lung transplant) was assessed for all enrolled patients until December 31, 2009.

Strains A and B of P aeruginosa were found in patients receiving care in 6 of 7 Ontario CF clinics. One small CF clinic located in Kingston, Ontario, Canada, which contributed only 8 patients to the study, did not contribute any patients infected with either transmissible strain. However, other patients living in this region and who were attending CF clinics in Toronto or Ottawa were found to be infected with strain A.

The MLST was used to determine if the 2 transmissible strains isolated in Ontario were similar to the transmissible strains identified in the United Kingdom or in Australia. The MLST typing of the Ontario transmissible strain A isolates identified them as either ST type 146 or ST type 683. Both ST 146 and ST 683 have been identified as belonging to the Liverpool epidemic strain of P aeruginosa.¹⁸ To verify that patients in Ontario were infected with the Liverpool epidemic strain of P aeruginosa, we obtained 8 known Liverpool epidemic strain isolates from CF centers in the United Kingdom. All 8 Liverpool strain isolates had identical PFGE band patterns to strain A isolates identified from Ontario. Five of the 8 Liverpool strain isolates were ST type 146.

The MLST typing of Ontario strain B isolates identified half of them as ST type 439. This ST type has not been previously described as a transmissible strain of P aeruginosa and it has never been described to infect patients with CF. The remaining 50% of Ontario strain B isolates had previously undescribed sequence types. Fifteen randomly chosen isolates from Ontario patients infected with unique strains of P aeruginosa as determined by PFGE also were typed using MLST. All 15 isolates showed unique MLST profiles and none were identified as being associated with transmissible Pseudomonas strains.

We also obtained isolates from the Midlands, Manchester, and Clone C epidemic strains from the United Kingdom. None of these other UK strains had PFGE banding patterns or MLST sequence types that matched Ontario P aeruginosa strains A or B. Similarly, we obtained isolates from the Melbourne, Brisbane, and Tasmania epidemic strains from Australia. Again none of these Australian strains had PFGE banding patterns or MLST sequence types that matched Ontario P aeruginosa strains A or B.

Thirteen patients developed possible new infections with P aeruginosa strain A or B that we discovered during annual sputum surveillance over the 3-year follow-up period. For 6 of these 13 patients, the transmissible strain was discovered to have been present in at least 1 of the 10 frozen sputum isolates recovered during the initial 3-month study period. Thus, a total of 7 patients developed new infections with P aeruginosa strains A or B over the 3-year follow-up period (Figure 1). The incidence rate of new infections with transmissible P aeruginosa strains was 7.0 per 1000 person-years (95% confidence interval [CI], 1.8-12.2 per 1000 person-years).

Environmental sampling of the incident patients' homes, including sink and shower taps, and sampling from the incident patients' CF clinics, including pulmonary function equipment and examination room sink taps, revealed positive culture evidence of P aeruginosa in 65 of 136 samples (48%), but no transmissible strains of P aeruginosa were recovered. All of the incident cases denied having relatives or friends with CF. Similarly, all denied social contact with any other patients with CF. Three patients had been hospitalized for CF in the 12-month period prior to incident infection, but all denied close contact with other patients with CF during their hospitalization. One patient had attended a CF fundraising event and may have had contact with other patients with CF.

Table 1 depicts the clinical characteristics of patients infected with P aeruginosa strain A, strain B, and unique strains and those not infected at study entry. Those patients not infected with P aeruginosa at study entry were more likely to be infected with other bacteria such as Staphylococcus aureus and B cepacia complex.

Patients infected with strain A were slightly younger and had a lower mean BMI at study entry compared with those infected with unique strains of P aeruginosa. Baseline lung function (measured as percentage predicted of FEV₁ and forced vital capacity) was not significantly different in those with strain A or B compared with those infected with unique strains.

Over the course of the 3-year observation period, the mean decline in FEV₁ percent predicted was 6.1% (95% CI, −0.3% to 12.5%) in patients with P aeruginosa strain A, 8.4% (95% CI, 3.5% to 13.4%) in patients with strain B, and 5.5% (95% CI, 3.1% to 7.9%) in those infected with unique strains. The annual median rate of decline in FEV₁ percent predicted was not significantly different for patients with P aeruginosa strain A vs those infected with unique strains (unadjusted difference: 0.64% [95% CI, −1.52% to 2.80%] P = .56; adjusted difference: 0.17% [95% CI, −1.88% to 2.22%] P = .87) (Figure 2). Similarly, the annual median rate of decline in FEV₁ percent predicted was not significantly different for patients with P aeruginosa strain B vs those infected with unique strains (unadjusted difference: 1.66% [95% CI, −1.00% to 4.30%] P = .22; adjusted difference: 2.19% [95% CI, −0.35% to 4.74%] P = .09).

Patients infected with P aeruginosa strain A had a slightly lower mean BMI compared with those infected with unique strains at study entry and this difference was maintained throughout the course of the study (Figure 2). However, mean and median BMI did not decrease in any of the 3 groups over the 3-year study. Rates of decline in BMI were not significantly different for patients with P aeruginosa strain A vs those infected with unique strains (unadjusted difference: 0.13 per year [95% CI, −0.24 to 0.50 per year] P = .49; adjusted difference: 0.10 per year, [95% CI, −0.25 to 0.44 per year] P = .58), or for those with P aeruginosa strain B vs those infected with unique strains (unadjusted difference: −0.04 per year [95% CI, −0.50 to 0.43 per year] P = .87; adjusted difference: 0.06 per year [95% CI, −0.37 to 0.48 per year] P = .80).

Patients infected with P aeruginosa strain B actually had slightly fewer pulmonary exacerbations (1.35 exacerbations/year; 95% CI, 0.89-1.81 exacerbations/year) compared with those infected with unique strains (1.93 exacerbations/year; 95% CI, 1.72-2.14 exacerbations/year) (P = .04; Table 2). Otherwise, there were no significant differences in exacerbation or hospital admission rates in the P aeruginosa strain A or B groups compared with the group infected with unique strains.

Of 446 patients, 50 died or received a lung transplant (11.2%) over the course of 3 years. Death or lung transplant occurred in 13 patients infected with P aeruginosa strain A (18.6%), 4 patients infected with strain B (11.4%), 19 patients infected with unique strains (8.7%), and 14 patients not infected with P aeruginosa (11.1%). There were no deaths or lung transplants among the 7 incident cases or among the 3 patients initially infected with both P aeruginosa A and B strains.

Infection with P aeruginosa strain A was associated with a greater 3-year risk of death or lung transplantation compared with patients infected with unique strains (unadjusted hazard ratio [HR], 2.27 [95% CI, 1.12-4.60] P = .02; adjusted HR, 3.26 [95% CI, 1.41-7.54]; P = .01). Infection with P aeruginosa strain B was not significantly associated with a greater 3-year risk of death or lung transplantation compared with patients infected with unique strains (unadjusted HR, 1.37 [95% CI, 0.47-4.02] P = .57; adjusted HR, 1.10 [95% CI, 0.34-3.63] P = .87; Figure 3).

The results of our study indicate that a sizable minority of adult Canadian patients with CF living in the province of Ontario are infected with 1 of 2 common strains of P aeruginosa. The most prevalent transmissible strain found was the Liverpool epidemic strain, which was found to infect more than 15% of Ontario patients. This same strain is known to infect approximately 11% of patients with CF who receive their care in 1 of 15 CF clinics in England and Wales.⁹ Our study is the first report to suggest that common strains of P aeruginosa are shared among patients located on different continents. Our data suggest that cross-infection with P aeruginosa has occurred widely both within Ontario and between CF centers in the United Kingdom and Canada.

It is currently unknown if infection with the Liverpool epidemic strain or with other transmissible strains of P aeruginosa is prevalent among US patients with CF. Epidemiological studies on infection with transmissible strains of P aeruginosa in the CF population in the United States have not been published.

Presumably cross-infection with transmissible strains of P aeruginosa may be resulting from close patient-to-patient contact among infected and noninfected patients. Studies of patients with CF have shown that viable P aeruginosa is easily isolated from the cough aerosols of patients with CF, and that 70% of these aerosols are of respirable size (<3.3 μm in diameter).¹⁹ This suggests that coughing may serve as a potential mechanism for airborne transmission of P aeruginosa from patient to patient. Aerosol dissemination of epidemic strains of P aeruginosa has also been documented from room air samples in which patients perform spirometry and airway clearance, implying that close physical contact between patients may not be required to spread infection with epidemic strains within CF clinics.²⁰ Recent reports have documented spread of the Liverpool epidemic strain of P aeruginosa from a CF patient to both of her parents who did not have CF,²¹ suggesting that persons without CF can also serve as temporary reservoirs of infection. Finally, spread of the Liverpool epidemic strain from a CF patient to a pet cat has been recently described,²² suggesting that family pets can also serve as reservoirs for transmissible strains within the community.

There is precedence for cross-infection of bacterial organisms between Canadian and British patients with CF. Epidemiological studies of Burkholderia cenocepacia conducted in the early 1990s revealed that the same ET12 strain of B cepacia infected patients with CF in Edinburgh, Scotland, Manchester, England, and Toronto, Ontario, Canada.^{23 - 24} The suspected index case was a patient from Edinburgh who acquired the infection in the late 1980s. This patient traveled to Canada in the summer of 1990 to attend CF summer camp along with 12 other children from the United Kingdom. Eleven of the children from the United Kingdom, and subsequently many Canadian children at the same CF camp, developed infection with the ET12 B cepacia clonal strain.²³ Later studies from London, Ontario, Canada confirmed that attendance of patients at CF summer camps was strongly correlated with B cepacia infection among Canadian pediatric patients with CF.²⁵

It is impossible to know whether infection with the Liverpool epidemic strain of P aeruginosa originated in Canada or in the United Kingdom and it is difficult to identify in hindsight how the infection may have spread from one continent to another. The Liverpool strain was first identified in Liverpool in 1996 because of its unusual phenotypic properties; isolates displayed a relatively high level of resistance to conventional antipseudomonal antibiotics.⁷ By that time, Canadian CF summer camps had been disbanded because of infection-control concerns. However, molecular typing of P aeruginosa was not widely available until the late 1990s, and it is theoretically possible that cross-infection between Canadian and UK patients may have occurred in the early 1990s at CF summer camps or elsewhere. Our laboratory first picked up the presence of a possible transmissible P aeruginosa strain infecting Canadian patients in 2004,²⁶ and this served as the stimulus for the present study. With the advent of MLST, it became possible to definitively identify this dominant Canadian transmissible strain as the Liverpool epidemic strain.

Although knowledge of infection with transmissible strains of P aeruginosa among patients with CF is important because of its implications for infection control, a sense of urgency about this issue would only be justified if infections with transmissible strains were shown to be associated with adverse clinical outcomes or prognosis. Our study is the first large, prospective cohort study to examine this issue. A retrospective case-control study of 12 patients infected with the Liverpool epidemic strain suggested that patients infected with the Liverpool strain had a greater annual loss of lung function compared with control patients with CF.²⁷ However, an 8-year single-center study from a Manchester, England clinic compared 28 patients with CF with transmissible strains of P aeruginosa (21 infected with the Manchester A strain and the other 7 infected with the Liverpool strain of P aeruginosa) with 52 patients infected with unique strains of P aeruginosa.²⁸ This study did not show differences in survival or annual changes in lung function or BMI, although patients infected with transmissible strains did receive more intravenous antibiotics compared with those infected with sporadic strains.

Results of our study shed more light on the clinical implications of infection with common transmissible strains of P aeruginosa. Those patients infected with Ontario strain A (also known as the Liverpool epidemic strain) were twice as likely to go on to death or lung transplant over the 3-year observation period compared with those infected with unique strains of P aeruginosa (19% vs 9%). Somewhat surprisingly, our study did not show that patients infected with either transmissible strain experience an accelerated decline in lung function over 3 years compared with those infected with unique strains. How do we reconcile the fact that death or lung transplant is more common in patients infected with the Liverpool strain, but lung function appears to decline at the same rate as those infected with unique strains? We think the likeliest explanation is the healthy survivor effect. Patients who died or had lung transplants did not contribute annual lung function data following these events. Because a greater proportion of patients infected with the Liverpool Epidemic strain experienced these events, censoring of the sickest patients from the Liverpool strain cohort may have resulted in apparently lower rates of decline in lung function among the healthy survivors. Similarly, because most cases of transmissible strains were prevalent cases, it is possible that the sicker patients with transmissible strains died prior to 2005, and this other form of survivor bias could explain why lung function did not decline at an accelerated rate in those with transmissible strains.

There are potential limitations of our study. The annual decline in FEV₁ percent predicted tended to be steeper in patients infected with P aeruginosa strain B compared with those infected with unique strains; however, the relatively small numbers of patients infected with strain B resulted in limited power to show a statistical difference for this comparison. Also, although most patients had at least 10 isolates of P aeruginosa genotyped over the course of the study, it is possible that some prevalent or incident cases of strain A or B could have been missed. Unmeasured confounders may also have influenced outcomes.

In summary, our study has shown that cross-infection with transmissible common strains of P aeruginosa has occurred widely both within Ontario, Canada and between CF centers in the United Kingdom and Canada. Infection with the Liverpool epidemic strain was associated with a greater risk of death or lung transplant. Differences in prognosis among patients with CF infected with P aeruginosa may be due in part to differences in specific strain types infecting individual patients.