INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Chronic infection of the lower respiratory tract is responsible for much of the morbidity and mortality of cystic fibrosis (CF). Mutations altering function of the cystic fibrosis transmembrane regulator (CFTR) impair chloride transport and lead to a milieu favoring colonization by bacteria, particularly Staphylococcus aureus and Pseudomonas aeruginosa. More recently, another microbe, Burkholderia cepacia, has been recognized as a frequent colonizer in patients with CF. Conflicting results on the virulence of B. cepacia can be found in the literature. The majority of the early literature and clinical experience suggested an accelerated decline in clinical status and increased mortality in CF patients colonized with B. cepacia compared with P. aeruginosa (1). A more varied response to B. cepacia colonization with no difference in outcome has been reported in more recent studies (5), whereas an accelerated rate of decline in respiratory function in the B. cepacia- colonized CF patient is still reported by others (8).

Suboptimal matching on clinical parameters may be the limiting factor responsible for the worse outcome after B. cepacia acquisition that has been previously reported. Changes in the clinical management of CF patients colonized with B. cepacia in the last 10 yr may also contribute to the different outcomes after colonization with this organism. In clinics with the highest prevalence of B. cepacia colonization, new acquisition rates have slowed considerably and in clinics with low prevalence significant increases have generally not been observed (7, 9, 10). The purpose of this study was to determine whether a contemporary cohort of CF patients colonized with B. cepacia have a worse outcome than CF patients with similar disease severity not harboring this organism. Typing of the patients' B. cepacia strains (11) was also performed to determine if they were colonized with a predominant type or multiple different strains.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Population

The sample was selected from 125 patients who attend the Adult Cystic Fibrosis Clinic at St. Paul's Hospital and who had attended as children the Pediatric Cystic Fibrosis Clinic at B.C. Children's Hospital in Vancouver, British Columbia. CF was diagnosed with standard sweat chloride testing. Our adult sample consisted of 36 patients consistently colonized with the pathogen B. cepacia. Time of acquisition of B. cepacia for CF patients was defined as the time point after which B. cepacia was consistently being isolated. All dates for B. cepacia acquisition were established within a 6-mo interval. CF patients displaying sporadic growth of B. cepacia (i.e., pathogen isolated once per year in a 2 to 3 yr interval out of a minimum 4 samples per year) were not included in the study as part of either the case or control groups. CF patients with negative cultures for B. cepacia were selected for controls.

Patients attending clinic who had been colonized with B. cepacia for a minimum of 6 mo (on at least two occasions) confirmed by a reference laboratory were identified for the study. Inclusion criteria for cases (i.e., colonized with B. cepacia) and controls (i.e., not colonized with B. cepacia) were as follows: (1) Patients who had available bacteriological and lung function data during clinically stable periods for at least 2 yr preacquisition of B. cepacia or prematch date with a minimum of 2 clinic assessments per year. (2) Patients had 1.5 to 2 yr postacquisition data. (3) B. cepacia cases with less than 2 yr postacquisition data due to death were included in the study. (4) Patients who underwent transplantation had no further data entered after transplantation and outcome was identified as transplantation.

Matching Criteria

Controls were selected from the same center as the B. cepacia patient cohort, but were not colonized at any time with B. cepacia. Controls were matched on the following parameters to B. cepacia colonized cases at time of acquisition of the pathogen: gender, age (± 1 yr), severity of airflow obstruction (percent predicted forced expiratory volume in one second [% pred FEV₁] ± 10%), height (± 5 cm), weight (± 8 kg), and pancreatic sufficiency status.

Outcome Parameters

Data were compiled from the patients' medical charts. Spirometry (FEV₁ and FVC) and body weight measurements performed during stable outpatient clinic visits were collected. The primary endpoint for this study was whether or not the patient was alive 2 yrs after acquisition of B. cepacia or match date in the noncepacia controls. Long-term survival was also examined. Cause of death was established from autopsy results if obtained or otherwise from the death certificate. The number of hospitalizations for pulmonary exacerbations was compiled for 2-yr pre- and 2-yr postacquisition and long-term follow-up intervals relating to the time of acquisition or matching. Specifically, the number of days of intravenous antibiotic treatment for pulmonary exacerbations was examined. Hospitalizations for causes entirely unrelated to pulmonary exacerbations were excluded. For each hospital admission, spirometry and body weight measurements performed close to the time of discharge were used for that encounter. The number of routine and ill clinic visits was compiled for 2-yr pre- and 2-yr postacquisition and long-term follow-up intervals relating to the time of acquisition or matching. A clinic visit to the CF clinic where the patient attends as part of regular follow-up and where the patient's overall status was described as stable was classified as a routine clinic visit. An ill clinic visit was defined as a visit initiated by the patient (or as part of their regular follow-up) where the patient had an increase above baseline in cough, sputum production, or dyspnea and decrements in lung function and nutritional status. Spirometry was carried out according to American Thoracic Society criteria (12). The best value of a minimum of 3 adequate postbronchodilator measurements was taken with the FEV₁ and the FVC of the best 2 of these efforts not varying by more than 5% or 100 ml, whichever was greater.

Identification and Typing of B. cepacia

Organisms suspected of being B. cepacia were sent to the Canadian Pseudomonas Repository Laboratory where B. cepacia confirmation was performed as described by Henry and coworkers (13). Isolates were checked for purity, then screened for growth on FC agar (14) and BCSA agar (13). Organisms that were FC-negative and BSCA-positive were set up to the API Rapid NFT strip (Biomerieux Vitek Inc., Hazelwood, MO), supplemented by glucose, maltose, lactose, mannitol, xylose, and sucrose oxidation/fermentation sugars and lysine decarboxylase. Organisms that matched the identification criteria of Henry and coworkers (13) were confirmed as B. cepacia. Typing of B. cepacia was performed by randomly amplified polymorphic DNA fingerprinting (RAPD) as described in Mahenthiralingam and coworkers (11). Sequential isolates of B. cepacia recovered from individual patients were typed whenever possible over a minimum period of 1 yr to ensure consistency of colonization with a single strain type.

Statistical Analysis

The McNemar test (15) was used to compare the proportion of cases and controls still alive for two time periods: 1) 2 yr after the date of acquisition of B. cepacia or match date and 2) after long-term follow-up. For each subject, we calculated the annual rates of change of % pred FEV₁, percent predicted forced vital capacity (% pred FVC), and weight using ordinary least-squares regression. Rates of change were calculated separately for three different time periods: the 2-yr preacquisition period, the 2-yr postacquisition period, and the long-term follow-up period. The Wilcoxon signed rank test (15) was used to compare the rates of change in % pred FEV₁, % pred FVC, and weight for the 2-yr preacquisition period to the 2-yr postacquisition period in B. cepacia cases and their controls. To determine if these parameters undergo more rapid change in the period following acquisition of B. cepacia, we calculated the average difference in rates of change of the three parameters between B. cepacia cases and controls and tested for statistical significance using the Wilcoxon signed rank test (15). These comparisons were done for both the 2-yr postacquisition period and the long-term follow-up period.

Differences in the average number of days of intravenous antibiotic treatment for pulmonary exacerbations per year were compared between cases and controls in a similar fashion. Outpatient clinic visits were similarly analyzed and classified as routine, or ill. Descriptive statistics were calculated for match parameters and time to death or transplant. The level of significance was set at p  0.05. Post hoc power analysis of lung function (i.e., % pred FEV₁ and % pred FVC) and weight for the 2-yr postacquisition follow-up was performed.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our case sample consisted of 36 patients colonized with B. cepacia (25% of our clinic population). Three cases were dropped owing to insufficient pre- and postacquisition data; eight cases had recently been colonized with the organism and had less than 1-yr postacquisition data. We were unable to match four cases on the following factors: age (n = 1), age and % pred FEV₁ (n = 2), and height and weight (n = 1). Of the 21 cases that were successfully matched, 10 were female. A comparison of B. cepacia cases and their controls on match variables is provided in Table 1. Also presented in Table 1 are the mean group differences within matched pairs on match variables.

There was no difference in survival in the 2-yr postacquisition period: three B. cepacia cases and no matched controls died in this period (p = 0.25). However, a significant survival advantage for the controls was detected in the long-term follow-up (p = 0.04). In eight pairs the B. cepacia patient died but the control did not and in one pair the case and control both died. The range of time to death after time of acquisition of B. cepacia was 0.77 to 14.6 yr. The range of time to death following match date for the controls was 3.1 to 11.8 yr. Table 2 presents the characteristics of B. cepacia cases and controls who died during the study interval. Specifically, Table 2 presents the gender, age at the time of acquisition of B. cepacia, B. cepacia DNA RAPD group strain, their % pred FEV₁ at time of colonization with B. cepacia, time from acquisition of B. cepacia to death (or alternatively if only the control is deceased the status of the B. cepacia case), the status of the control pair, and time to death of the control pair.

The difference between annual rates of change before and after acquisition or match date for lung function and weight was calculated and compared within matched pairs. There were no significant differences between the B. cepacia cases and controls for these parameters. Similar annual rates of change for both the 2-yr postacquisition and long-term postacquisition intervals were found for B. cepacia case and control pairs. Statistical results are tabulated in Table 2 for % pred FEV₁, % pred FVC, and body weight for 2 yr pre- to 2-yr postacquisition, the 2-yr postacquisition, and long-term follow-up intervals.

Differences in the average number of days of intravenous antibiotic treatment for pulmonary exacerbations per year were compared between cases and controls in a similar fashion. The number of days of intravenous antibiotic treatment for pulmonary exacerbations was tabulated for B. cepacia cases and controls (Table 3). There were no significant differences in the 2-yr preacquisition interval for the average number of days of intravenous therapy per year of follow-up. B. cepacia cases were prescribed intravenous antibiotic therapy for a significantly higher number of days per year follow-up for both the 2-yr postacquisition interval and the long-term follow-up than the controls.

Mean number of routine and ill outpatient visits to the CF clinic, per year of follow-up, were compared between B. cepacia cases and controls in a similar fashion. We found no significant differences between the B. cepacia cases and their controls for the 2 yr preacquisition (0.61 routine visits/yr, p = 0.22 and 0.03 ill visits/yr, p = 0.91), 2 yr postacquisition (0.23 routine visits/yr, p = 0.57 and 0.17 ill visits/yr, p = 0.55) and long-term follow-up (0.20 routine visits/yr, p = 0.59 and 0.09 ill visits/yr, p = 0.66) intervals.

Figure 1 displays the B. cepacia DNA RAPD group strain distribution at our center. We were unable to genotype one B. cepacia case as there were no samples stored after the patients' death to analyze. We have tabulated both the B. cepacia cases we were able to match to B. cepacia-negative controls and also cases not matched (does not include newly diagnosed/ colonized cases). The most prevalent B. cepacia DNA RAPD group whose patients at our center have become colonized is DNA RAPD group 4. There are six patients colonized with B. cepacia DNA RAPD group 2; the majority of cases (all but one) have relocated from Eastern Canada and were already colonized with the organism when they came to our center. A number of patients genotyped were colonized with singular strains; these are noted as "unique" in the figure and represent strains that have only been isolated from these patients.

View larger version (14K): [in this window] [in a new window]   Figure 1.   B. cepacia DNA RAPD strain type distribution. Tabulated are both the cases that were matched to non-B. cepacia controls (solid black bars) and also cases not matched (does not include newly diagnosed/colonized cases, i.e., n = 8).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

An excess in mortality has been reported in most studies after colonization or infection of CF patients with B. cepacia. While true in our study, this augmented mortality did not manifest in the first 2 yr postacquisition. Indeed, there were several other features of disease activity that failed to discriminate between B. cepacia and matched controls up to 2 yr postacquisition. We found no acceleration in the decline of spirometric measurements over this interval, providing no substantiation for worsening airflow obstruction after initial infection with B. cepacia. The number of outpatient visits to the CF Clinic, whether they were scheduled or premature visits owing to illness, were comparable. Weight loss, which we have found to be a relatively sensitive parameter for progression of respiratory disease in CF, also did not differ between B. cepacia cases and controls. After acquisition of this pathogen, however, the B. cepacia patients exhibited a higher frequency of intravenous antibiotic therapy in both the 2-yr and long-term follow-up postacquisition intervals. Furthermore, long-term mortality was increased in B. cepacia in comparison to CF patients not harboring the organism.

A more aggressive early course than we observed has been reported in most previous studies examining the consequences of B. cepacia acquisition (1, 8, 16). The deleterious effect on survival was most evident in early studies with the emergence of B. cepacia as a true pathogen in CF. Isles and coworkers (1), reviewing their experience from 1970 to 1981 reported seven deaths in 18 patients with B. cepacia, as compared with 4 of 67 patients whose sputum cultures were negative for this organism. This was an uncontrolled study with at least a proportion of the excess mortality that could be ascribed to the older age and reduced lung function of patients at the time of colonization with B. cepacia. In a case control study of pediatric patients matched for overall severity of disease, Tablan and colleagues (2) found an appreciably higher mortality in B. cepacia patients than control subjects within the first year of acquisition of this organism. An alarming 49% of patients harboring B. cepacia died, and in several patients this organism was first identified at autopsy. Muhdi and colleagues (8) also reported a 30% higher mortality in the B. cepacia group compared with the CF control group within 2 yr postacquisition. In addition to higher mortality several groups have reported that the identification of B. cepacia in respiratory secretions harbors an accelerated decline in lung function (5, 16). Our rate of decline in % pred FEV₁ based on the long-term follow-up results for both B. cepacia cases and controls (based on age and % pred FEV₁ at the time of acquisition of B. cepacia or match date) was lower than predicted for non-B. cepacia CF patients by Corey and associates (17). Although our mean % pred FEV₁ values were similar to those reported by Corey and associates (17) for pancreatic insufficient CF patients (mean age of 20 yr), the predicted mean rate of decline for the Toronto group was higher (i.e., 2.6% per year versus 1.6% and 2% per year for B. cepacia cases and controls, respectively in our study).

Although acquisition of B. cepacia adversely affects survival there is a very heterogeneous clinical course after identification of this organism. We found an increase in long-term mortality in B. cepacia cases, but did not observe any difference in 2-yr survival in the B. cepacia case-control pairs. We acknowledge that the absence of a difference in 2-yr mortality may be attributed to inadequate power related to our sample size. Nevertheless, the comparable changes in lung function parameters over this interval suggest an initial course in our B. cepacia cohort that is less aggressive than previously reported. There are several other potential reasons for this finding. B. cepacia patients may have benefitted from the more frequent intravenous antibiotic administration and hospitalization that we observed for this group. Comparable survival rates may therefore have been the result of more aggressive medical intervention. In a subset of patients, a rapid deterioration culminating in death may occur. We did not observe any patients in our study who manifested this "Cepacia syndrome." A lack of representation of this type of course may therefore have conferred a survival advantage in comparison to other studies where a number of these individuals were identified.

There are a number of potential host factors that may explain a variable clinical course after infection with B. cepacia. It has been established that acquisition of the B. cepacia often occurs in older patients with more advanced pulmonary disease (2, 16). In our study, we attempted to closely match for disease variables between cases and controls to minimize bias related to host factors. In other studies, controls were not included (16) and in those where a control group was included there was less rigidity in controlling for clinical parameters (1, 5, 8). An augmented decline in respiratory function and mortality therefore may have been accounted for in part because B. cepacia patients had more severe disease. Tablan and colleagues (2) had closely matched for parameters characterizing disease severity and also reported early excess mortality. Therefore, host factors can account for only a portion of the discrepancy in clinical course.

Bacterial factors may account for some of the variability in the disease course of CF patients. The phenotype of B. cepacia isolates recovered from patients with CF may vary considerably even within a single strain type (18). The increased virulence among certain strain types (19) may be related to the cable pilus gene (20), the B. cepacia epidemic strain marker (BCESM) (11), and antibiotic resistance (21). The strain often referred to as the hypertransmissible strain (22) is B. cepacia RAPD group 2 which is a cable pilus gene encoding strain. This strain is epidemic among CF patients in the United Kingdom and Eastern Canada (11, 22, 23). B. cepacia RAPD group 2 colonized a total of six patients at our center of whom five had relocated from Eastern Canada. Mahenthiralingam and coworkers have identified this strain to be the main strain type infecting CF patients in Eastern Canada (23). Consequently, the increased mortality and decline in lung function that has been reported in B. cepacia patients from Eastern Canada may be related specifically to colonization with this strain type. Colonization with B. cepacia was not associated with a significant decline in lung function in the Glasgow epidemic described by Whiteford and colleagues (9). A recent study by Vandamme and colleagues (24) identified the strain infecting this cohort of patients to lack the cable pilus gene and also the BCESM and has been named Burkholderia multivorans. Patients from our center were predominately infected with B. cepacia RAPD types 1, 4, and 6, which are all strains capable of patient-to-patient spread (11). Five of our patients were colonized with strain types that were unique to each patient. These strains lack the B. cepacia epidemic strain marker and there has been no evidence of spread of these strains to other CF patients (23). Preliminary information (unpublished data) suggests that these strains may be less pathogenic and might further contribute to a less aggressive course in our patient group. Further study will be required to determine the exact clinical risk and bacterial factors associated with the virulence of given B. cepacia strains.

We observed an interesting dichotomy between the lack of change in long-term lung function parameters and an augmented mortality in the B. cepacia group. This is surprising, as the excessive mortality is generally ascribed to increasing lung damage associated with this organism. Similar to our study, no difference in the rate of change in FEV₁ after acquisition of B. cepacia was found by Isles and coworkers (1), despite an increase in mortality rates in these patients. The close matching of B. cepacia cases to controls gave us the opportunity to examine the effect of B. cepacia acquisition on the course of CF while controlling for severity of disease at the time of acquisition. To ensure that we did have a sufficient number of patients to detect differences between B. cepacia cases and controls for changes in lung function we ran a posthoc power analysis. This analysis revealed that we had adequate power to detect a 5% change per year in % pred FEV₁ (0.80), % pred FVC (0.56), and weight (0.90) for the postacquisition period.

It is also plausible that spirometric measures may poorly reflect pathological features of the disease. The discrepancies between measures of airflow obstruction and histological abnormalities have been observed in other lung diseases. Although FEV₁ and FVC do offer overall prognostic information, changes in these parameters, particularly in those with advanced disease, have limited individual predictability of mortality. Evidence has accrued to implicate an enhanced immunologic host response for a proportion of the lung pathology in CF. Upregulated immunologic mechanisms may be operative with B. cepacia infection. Increased mortality may be related to the local and systemic effects of cytokines and inflammatory cell products (25). An absence of change in lung function parameters predominantly reflecting larger airway function therefore would not be surprising. Finally, in our study and in the majority of others, lung function parameters were selected when the patients were clinically stable. This decision may have masked greater decreases in lung function during exacerbations in B. cepacia patients.

Ideally the clinical significance of B. cepacia needs to be evaluated prospectively. Considering the small number of CF patients acquiring this pathogen, a long-term multicenter study would be necessary. As our study was retrospective, there are limitations to the generalizability of our results. Reliance on clinic visits placed limitations on and precluded standardization of timing for data acquisition. A universal definition of a pulmonary exacerbation does not currently exist. We cannot refute, therefore, that more frequent antibiotic utilization observed in our study was an attempt by physicians and/or patients to counter the belief of a more aggressive course with acquisition of B. cepacia.

In summary, our experience supports an increased long-term mortality in CF patients with persistent isolation of B. cepacia from the respiratory tract. This excess mortality was not associated with greater decrement in lung function testing when matching for the severity of disease at the time of acquisition, but was associated with increased frequency of pulmonary infections and more severe disease at the time of onset of B. cepacia colonization. Particularly in more advanced disease, changes in spirometric measures are limited in individual prognostication of mortality. Considerable intersubject variability in disease course is evident with B. cepacia infection. In many of our cases the augmentation of disease activity was modest, with no increase in 2-yr mortality observed compared with matched controls. This contrasts with other studies, which reported significant increases in 1- to 2-yr mortality. It is conceivable that differences in clinical outcome (i.e., patient-to-patient spread and mortality) across studies may be associated with specific strains of B. cepacia which may exhibit a more virulent course in the colonized CF patient. Further work is necessary to better define host and bacterial factors that can explain this heterogeneity in response.