Epidemiology of Burkholderia cepacia complex colonisation in cystic fibrosis patients

Chronic respiratory tract infections with intercurrent acute exacerbations are the hallmark of cystic fibrosis (CF) lung disease. Pseudomonas aeruginosa and Staphylococcus aureus are recovered most frequently from respiratory secretions of CF patients 1. Burkholderia cepacia infection is much less common, but it is notorious for being associated with cross infection and the possibility of rapid deterioration, known as the cepacia syndrome 2–4. B. cepacia colonisation frequency differs greatly between CF centres; overall, the prevalence is low, but the impact on individual patient survival may be very high 5. Furthermore, the possibility of B. cepacia colonisation induces a high level of anxiety in the CF population.

Optimal isolation of B. cepacia from CF sputum requires the use of selective culture media 6. Remarkable heterogeneity among presumed B. cepacia strains and the lack of reliable identification schemes makes correct identification of B. cepacia colonisation a complex problem 5, 6. Several distinct genomovars are collectively referred to as B. cepacia complex (Bcc) 7. The most commonly isolated Bcc species in CF patients are B. multivorans, B. cenocepacia and B. stabilis. A rapid deterioration and patient-to-patient transmission is probably more likely with B. cenocepacia infection and colonisation 8, 9. A Belgian surveillance study in 1993 revealed that 12 out of 465 Belgian CF patients were Bcc colonised. These strains belonged to three different genomovars i.e. B. multivorans, B. cenocepacia and B. stabilis) 10.

The aims of the present study were to critically re-evaluate the prevalence of Bcc colonisation in Belgian CF patients during a new national 1‐yr surveillance study to determine the genomovar status of the isolates and to evaluate the clinical status of patients colonised with Bcc, as compared to patients not colonised with Bcc. In addition, the clinical course of CF patients, colonised during the 1993 surveillance study, was evaluated.

Methods

In Belgium, since 1993 CF centres have used selective MAST medium (Mast Diagnostics, Meyerside, UK) to inoculate all sputum samples and upper airway samples from CF patients. Sputum samples are examined routinely every 3 months. A throat swab is taken in patients unable to produce sputum. All 11 Belgian health facilities caring for CF patients agreed to participate in the current study and to send all B. cepacia-like isolates to the reference laboratory between February 1999 and February 2000. The isolates were first examined by means of whole-cell protein electrophoresis, and compared to a database comprising a large number of Gram-negative, nonfermentative bacteria, isolated from CF and environmental specimens 11–15. Extensive taxonomic studies have revealed that this method allows species-level identification of a wide range of Gram-negative nonfermenting bacteria, commonly misidentified as Bcc, and confirms the identification of putative Bcc isolates at the complex level 14, 15. The method has proved to be insufficiently discriminatory to identify all species within Bcc 14, 15. Therefore, Bcc bacteria were further examined by means of recA restriction fragment length polymorphism (RFLP) analysis 16 and pulsed-field gel electrophoresis (PFGE) of genomic DNA restricted with SpeI 17. Non-Bcc bacteria were further examined by using whole-cell fatty acid analysis 7. Presence of the cable pilus subunit gene 18 and the epidemic strain marker BCESM 19 were determined by PCR using established primers.

For every patient colonised by Bcc, a control patient was identified from the same CF centre as the index patient and matched for sex, pancreatic status and genotype (a homozygous F508del patient was chosen for a F508del homozygous index case, a heterozygous F508del/other severe mutation was chosen for a heterozygous F508del/other severe mutation index case; mutations were considered severe when associated with pancreatic insufficiency and reported as such in the Toronto CF consortium database 20. Patients were also matched for respiratory colonisation status, with chronic P. aeruginosa colonisation being defined as at least three sputum cultures positive for P. aeruginosa over the course of 6 months 21. If several patients qualified, the patient with the birth date closest to the index patient was chosen. There was no difference in mean age between index cases and controls (table 1⇓).

For both index and control patients, body mass index (BMI) was calculated and expressed as sd score 22. Maximal expiratory flow manoeuvres were performed according to 1994 American Thoracic Society criteria 23. Forced vital capacity (FVC) and forced expiratory volume in one second (FEV₁) were expressed as per cent of the normal value, according to Zapletal for children and the European Community for Coal and Steel for adults 24. For the time period outside the surveillance year, lung function data were extracted from the patient's chart. For every Bcc-colonised patient, the time of first colonisation was determined from the patient's chart. Data on all reported patients are included in the Belgian CF database. These patients all sign an informed consent and agree that their clinical data may be analysed and reported in an anonymous way in clinical studies.

Median age of colonised patients was compared to median age of CF patients included in the Belgian CF registry. The age distribution for the Belgian CF population in 1999 was not normal and, therefore, for this purpose, median age is reported and Mann-Whitney U‐test was used for analysis.

Since 1995, specific infection control measures have been taken in all Belgian centres to prevent patient-to-patient spread of Bcc. These include: stressing hand wash policy for all patients and personnel; individual rooms for all hospitalised CF patients; and isolation during hospital admission, as well as during outpatient clinic visits for patients colonised with Bcc.

Results

Typing results using pulsed field gel electrophoresis

An overview of typing results for Bcc isolates is given in figure 1⇓. Typing results confirmed that the patients remained colonised by the same strain during the course of the study (data not shown). Three patients were colonised by an identical strain; two patients were followed at centre E (PFGE type III) and one was followed at centre H (PFGE type III'). The PFGE DNA profile of the latter isolate differed in four DNA fragments from the profiles of the former two isolates. Such a limited degree of genomic variability is well known to occur in genomically versatile organisms like Bcc bacteria 25, and unrelated isolates differ in larger numbers of DNA fragments, as can be seen in figure 1⇓. The centres E and H are 50 km apart and the three patients claim not to have any contact with each other. Three colonised patients had already been detected as positive during the 1993 cohort study. Typing results indicate that these patients continue to be colonised by the same B. cepacia strain. In one patient, during 2002, a temporary switch from colonisation by B. stabilis to colonisation by B. multivorans occurred. At present, the patient is again colonised with B. stabilis, as he has been for 10 yrs.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.—

Pulsed-field gel electrophoresis (PFGE) pattern of the Burkholderia cepacia complex isolates in the 12 patients as indicated by their number in table 1⇑ in 1999. Different PFGE banding patterns are numbered sequentially. Types III' and IV' differ in 4 and 3 fragments, respectively, from the types III and IV banding patterns. Patients 4, 8 and 5 were included in the 1993 study and were identified as having PFGE type IV', VI and VIII respectively. For patient 4, the 1993 isolate with IV' banding pattern is also shown which differs slightly from type IV. For patients 5 and 8, the PFGE pattern of the strain was identical in 1993 and 1999. Number of samples: patient 1, n=5; patient 2, n=6; patient 3, n=2; patient 4, n=2; patient 5, n=2; patient 6, n=3; patient 7, n=1; patient 8, n=8; patient 9, n=4; patient 10, n=2; patient 11, n=1; patient 12, n=8.

Discussion

In 1999, only 12 out of 650 Belgian CF patients were colonised by Bcc. The prevailing genomovar was B. multivorans; only two patients were colonised by B. cenocepacia. The 5‐yr evolution following onset of Bcc colonisation was poor: four of the 12 index cases died, although none of the matched controls died. In addition, lung-function evolution was significantly worse in index cases than in controls. In 11 out of 12 Belgian CF patients colonised with Bcc in 1993, the outcome could be evaluated: three patients had transient colonisation, five patients died, of whom four within 5 yrs of acquiring Bcc.

The Bcc colonisation rate in Belgian CF patients is low and remains low. In 1993, 12 patients out of 465 screened were Bcc positive. In the second national surveillance study, 12 out of 650 patients were Bcc positive. For the second study, all centres were able to cooperate and that study comprised all of the CF patients known in Belgium (the Belgian CF registry for 1999 contains data on 605 patients (C. Sevens, personal communication)). This suggests that there was no increase in Bcc colonisation in Belgian CF patients from 1993 to 2000.

In the Belgian CF community, the prevailing B. cepacia genomovar varied over time. In the 1993 cohort, six of the 12 patients were colonised with B. cenocepacia; the species most commonly associated with epidemic spread of disease and rapid deterioration 8, 9. In the 1999 cohort, the majority (9 out of 12) of the patients were colonised with B. multivorans; only two patients were colonised with B. cenocepacia. This differs from the USA and Canada, where the majority of the isolates belong to B. cenocepacia 9, 26, 27. In the study by Chen et al. 27, 59 out of 60 Bcc isolates in CF patients were B. cenocepacia, and 57 of these 59 isolates represented the same clone. In Belgium, in between the two evaluation periods, strict isolation measures for Bcc-colonised patients have been instituted. These seem to have been successful so far.

The prevailing Bcc genomovar has been reported in few countries. The Belgian data differ somewhat from that which is known in both Italy and the UK as well. A nationwide survey was not reported from these countries, but a large number of Bcc isolates from CF patients was studied. A high incidence of infection caused by a single B. cenocepacia epidemic clone, possessing the cable A gene and closely related to the USA–UK clone “ET12”, was reported from two Sicilian CF centres 28. A recent study on 117 patient isolates from 40 UK hospitals reported that B. cenocepacia is most prevalent, with most isolates belonging to the epidemic clone 29. However, both in Italy and the UK, other Bcc species were isolated from CF patients. It appears that in the absence of epidemic spread of one particular strain, the incidence of Bcc infection is low (such as in Belgium) and the distribution of genomovars is more varied.

The possibility of patient-to-patient transfer of B. cepacia in the CF community has always been worrisome. From DNA fingerprinting by AP‐PCR and PFGE results in the 1993 cohort, it was demonstrated that four Belgian patients with B. cenocepacia were colonised by the same strain. These patients all attended the same clinic 10. Two patients with B. stabilis colonisation, attending distinct clinics, shared the same strain.

For the 1999 cohort, three patients with B. multivorans shared an identical strain. In the UK, most reports stress the importance of the epidemic B. cenocepacia strain; however, B. multivorans is second in prevalence and several clusters of cross-infection have been identified 29. In addition, in Italy, patient-to-patient spread was observed with genomovars other than B. cenocepacia 28. The fact that the number of patients sharing the same strain decreases, suggests that for Belgian patients cross-infection is at present less likely than acquisition from an environmental source. Bcc is being used in agriculture for plant disease control 30 and strains from all genomovars have been isolated from the environment 31.

Patients colonised with Bcc may remain colonised during prolonged periods of time. Three patients were colonised by the same strain from 1993–1999. However, transient colonisation does seem to occur. It was documented in three patients of the total group of 21 patients described in this paper. Since in these patients several sputum samples were grown on B. cepacia selective media, it is unlikely that the pathogen was missed. In some patients, however, Bcc was not isolated for several years and then their first strain reappeared up to 10 yrs later. In some of these patients, it cannot be excluded that reacquisition from the environment occurred. Reports from UK 32 and Canada 33 also incidentally mention transient Bcc infection, but a thorough evaluation of a cohort to assess the frequency of this event has not been reported previously.

Unlike the initial reports 2, 3 suggested, Bcc acquisition is not limited to patients with P. aeruginosa colonisation nor patients with severe lung disease. In the Belgian study, several patients were not colonised with P. aeruginosa prior to Bcc colonisation. At the onset of Bcc colonisation, there was no significant difference in lung function abnormality between Bcc positive patients and controls. However, in the 5 yrs following onset of colonisation, the mortality is higher and lung function deterioration is faster.

In this Belgian study, the mortality amongst Bcc-colonised patients is high; from the 1993 cohort, four of the six patients colonised with B. cenocepacia have died. One of the remaining six patients has died; this patient was included in the 1999 cohort. Four other patients out of the 1999 cohort have also died, whereas none of the control patients have died. Only the transplanted patient died with the typical febrile bacteraemic “cepacia syndrome”. The big impact of Bcc colonisation on disease severity and mortality has been reported previously 2, 3, 32 and is confirmed by the present study. This study in a low-Bcc-incidence country clearly corroborates that the Bcc problem is not limited to B. cenocepacia.

In conclusion, in the Belgian cystic fibrosis population, Burkholderia cepacia complex colonisation rate is low and stable. Burkholderia multivorans and Burkholderia cenocepacia colonisation is most common. Although most patients are colonised for a longer time, in some patients transient colonisation seems to occur. The impact of Burkholderia cepacia complex on morbidity and mortality in a cystic fibrosis population is high and not limited to Burkholderia cenocepacia isolates. In Belgian cystic fibrosis patients, most patients carry a different strain and the cross-infection rate is low.