Management of initial colonisations with Burkholderia species in France, a retrospective analysis in five Cystic Fibrosis Centres

Whereas Burkholderia infections are recognized to impair prognosis in cystic fibrosis (CF) patients, there is no recommendation to date for early eradication therapy. The aim of our study was to analyse the current management of initial colonisations with Burkholderia cepacia complex (BCC) or B. gladioli in French CF Centres and its impact on bacterial clearance and clinical outcome. We performed a retrospective review of the primary colonisations (PC) observed between 2010 and 2018 in five CF Centres. Treatment regimens, microbiological and clinical data were collected. Seventeen patients (14 with BCC, and 3 with B. gladioli) were included. Eradication therapy, using heterogeneous combinations of intravenous, oral or nebulised antibiotics, was attempted in 11 patients. Six out of the 11 treated patients, and 4 out of the 6 untreated patients cleared the bacterium. Though not statistically significant, higher FEV1 and FVC at PC and consistency of treatment with in vitro antibiotic susceptibility tended to be associated with eradication. Thus, the management of PC was shown to be heterogeneous, and the statistic power of our study insufficient. Large prospective studies are needed to define who to treat, when, and how. Pending these studies, we propose, due to possible spontaneous clearance, to check the presence of Burkholderia one month after PC before starting antibiotics, at least in the milder cases, and to evaluate a combination of IV beta-lactam + oral or IV fluoroquinolone + inhaled aminoglycoside. bacterial clearance and pulmonary outcomes. We present a 2010-2018 retrospective study in five CF Centres, including 17 patients. and an inhaled aminoglycoside (tobramycin). A follow-up sputum culture can be discussed prior to introducing treatment in milder cases. Large-scale studies are required to standardize Burkholderia PC treatment in cystic fibrosis patients.


Introduction
The species-level identification of the primary and potential subsequent isolates was determined by the OBC, using ARDRA (Amplified Ribosomal DNA-Restriction Analysis) [17,21], or RecA gene sequencing [22], and was completed by genotypic analysis of BCC isolates if relevant.
Once data collection had been completed, the susceptibility of the last strain recovered before the onset of treatment was tested by the OBC, for each antibiotic administered, following the guidelines of the antibiogram committee of the French Society for Microbiology, and using PK-PD breakpoints, or species related breakpoints when available [23]. In the absence of specific values for inhaled aztreonam-lysine, the breakpoints for parenteral administration were used, whereas inhaled tobramycin was considered active given the high pulmonary concentrations delivered [24,25]. The consistency of eradication treatments with in vitro antibiotic susceptibilities or recognized activity (inhaled tobramycin) was then analysed for each patient treated. The treatment was considered to match the antibiogram when the bacterial strain was fully susceptible to at least two antibiotics, or to at least one antibiotic if combined with inhaled tobramycin.

Microbiological outcome
The microbiological follow-up allowed the classification of patients into two groups: "eradication", when the bacterium was not recovered over a follow-up period of at least one year including ≥ 3 sputum cultures, and "persistence" in case of chronic colonisation. In the "persistence" group, the in vitro antibiotic susceptibility of the post-treatment isolate was compared to that of the pre-treatment isolate.

Statistical analysis
After a verification phase in order to highlight missing, incorrect or inconsistent data, the database was locked, and information analysed. Characteristics were compared via the Chi2 test or Fisher's exact test using theoretical aspects for qualitative variables. Comparisons of quantitative variables were performed using the Student test, or the Mann-Whitney non-parametric test if the normality or homoscedasticity of the distributions had not been checked. Statistical tests were conducted with STATA v13.1, and were carried out using a bilateral approach and a type one error of 5%. A value of p<0.05 was considered significant for all analyses.

General characteristics
The general characteristics of the cohort are described in Table 1

Eradication treatment modalities
PC treatments and outcomes are detailed per patient in Table 2 Treatment modalities were heterogeneous, including intravenous (IV), oral and nebulised antibiotics.
All antibiotic regimens were combinations of two compounds or more, and comprised an IV betalactam in 8 cases (72.7%), combined with an IV (4) or inhaled (2) aminoglycoside and/or IV ciprofloxacin (3). Oral treatments (4 patients) included levofloxacin or ciprofloxacin and/or cotrimoxazole. Three patients received inhaled aztreonam-lysine. The duration of all IV courses was 14 days, whereas that of oral treatments was 21 to 28 days. Lastly, the treatment matched antibiogram data in 54.5% of patients (6/11).

Outcomes
The median duration of follow-up was 5.9 years [range 1.0-8.2]. There was no death, Burkholderia-induced septicaemia, nor cepacia syndrome to report. One patient underwent lung transplantation 2 years after PC. The Burkholderia strain was cleared in 10 patients, whereas infection was persistent in 7 patients. A comparison of patient characteristics in the two microbiological outcome groups is presented in Table 3. There was no significant difference in terms of gender, age, previous antibiotherapy, BMI, and bacterial species. FEV1 values at PC and two years after PC were lower in the "persistence" group than in the "eradication" group, but the difference did not reach statistical significance (p=0.12 and p=0.08, respectively). No correlation was found between implementation of eradication therapy, its time of onset and microbiological outcome. Four of the 6 untreated patients spontaneously cleared the bacterium, which was isolated only once in 2 of these patients, as well as in 3/6 treated patients. Eradication was obtained in 5/6 patients receiving an antibiotic regimen which was in accordance with antibiogram data, and in only 1/5 patients otherwise (p=0.08). It should be noted that beta-lactam/IV aminoglycosides combinations (+/-another molecule) were associated with treatment failures in 4/4 patients, even when the strain was susceptible to the beta-lactam used, whereas the two eradication regimens including inhaled tobramycin were successful. Five patients received combinations including a fluoroquinolone (ciprofloxacin or levofloxacin), which was active in vitro in 3 cases belonging to the "eradication" group (with 1/3 single isolation), and inactive in 2 cases (1 in the "eradication" group with a single isolation, and 1 in the "persistence" group). Lastly, inhaled aztreonam-lysine was associated with eradication in 3/3 patients, with 2/3 single isolations.
Concerning the impact of treatment on antibiotic susceptibility, the analysis of post-treatment isolates in the 5 treated patients who developed chronic infection showed that 2/5 isolates had become resistant to one of the antibiotics administered (1 to ciprofloxacin and 1 to cotrimoxazole), whereas beta-lactam susceptibilities were not significantly modified.

Discussion
Though limited to 5 of the 40 French CF Centres following Burkholderia-positive patients, our retrospective study demonstrated a great diversity in the management of initial colonisations with BCC or B. gladioli.
First of all, eradication therapy was not systematically implemented, as related by Horsley et al. in their survey of adult CF Centres in UK [26]. Indeed, while in the epidemic era, most infections with BCC persisted over years, nowadays infection is transient in about 50% of French patients [6]. As regards B. gladioli, Kennedy et al. showed a transient colonisation rate of 60% in a series of 30 North American patients [16], similar to that observed in France [17]. In the present study, spontaneous clearance was observed in 4 out of 6 untreated patients, and in the 3 treated patients with only one positive sputum culture, we cannot be sure that the bacterium would not have been cleared without antibiotic therapy, as pointed out by Horsley et al. in their study [26]. These observations could lead to discuss a short follow-up of sputum cultures before introducing eradication therapy in some patients whose characteristics have yet to be defined in greater detail. Some risk factors for progressing towards chronic colonisation have been proposed. In a study including 64 patients in gladioli are considered to be the most problematic species, and constitute a relative contraindication to lung transplantation. However, we also observed post-transplant bloodstream infections due to B. multivorans and B. vietnamiensis (OBC data), which does not allow to rule out any risk in case of colonisation with other species. In our study, eradication was attempted in 4 out of 4 patients with B.
cenocepacia, but in only 1 out of 3 patients with B. gladioli, and in 6 out of 10 patients with other species. To-date, there is no evidence clearly establishing the benefit of early eradication therapy [18] and our study failed to demonstrate the superiority of attempted eradication over therapeutic abstention. Nevertheless, as discussed above, it can be assumed that eradication treatment is relevant in patients whose infection is most likely to become chronic, namely those with the most severe forms of the disease at detection, and/or in patients colonised by given species.
Second, antibiotic regimens were heterogeneous, including IV, oral and nebulised antibiotics. Such heterogeneity was also reported in the UK survey by Horsley et al. [26], and within published studies. chloramphenicol (n=2). Antibiotics were also administered via a nebuliser (tobramycin or meropenem) on 13 occasions, generally for 12 weeks [26]. Lastly, Garcia et al. proposed a standardized aggressive protocol comprising a 21 day induction stage (IV + inhaled tobramycin, cetazidime, trimethoprim-sulfamethoxazole) and a 2 months consolidation stage (trimethoprimsulfamethoxazole, inhaled tobramycin), combined with long-term azithromycin as anti-inflammatory [33]. However, due to the low numbers of patients and the diversity of patient populations and treatment protocols, the best antibiotic strategy is not defined to date [18]. Burkholderia species are innately resistant to many antibiotics including some of those used to treat Pseudomonas aeruginosa such as aminoglycosides and colistine, and additional resistance may appear during antibiotic therapy, as observed in 2 of our 11 treated patients. Hence, the treatment of colonisations and infections with Burkholderia is particularly difficult and based on a small number of molecules mostly used in combination to limit resistance selection. These currently comprise meropenem, ceftazidime, temocillin, cotrimoxazole, ciprofloxacin or cyclines [34]. BCC is usually considered to be resistant to aztreonam, but some strains are susceptible, and Iglesias et al. reported successful eradication of BCC in two patients with bronchiectasis following inhaled aztreonam lysine courses [35]. The latest ceftazidime-avibactam combination therapy seems promising but insufficient clinical experience has been acquired to date [36]. In spite of the natural resistance to aminoglycosides, the use of inhaled tobramycin which allows the delivery of high doses are considered to be of interest in BCC infections [24], [25]. B. gladioli is generally slightly more susceptible than BCC, especially to piperacillintazobactam and aminoglycosides [17]. The European Committee on Antimicrobial Susceptibility Testing (EUCAST) does not currently advocate Burkholderia susceptibility testing as a guideline for antibiotic therapy given the lack of any obvious correlation between in vitro susceptibility and the patient's clinical course [19]. Conversely, both the Clinical and Laboratory Standards Institute (CLSI) and the antibiogram committee of the French Society for Microbiology provide specific guidelines for a limited number of molecules. In our study, consistency between treatment and antibiotic susceptibility appears to be associated with improved microbiological efficacy, since eradication was achieved in 5/6 cases (83.3%) of antibiogram consistency versus 1/5 cases (20%) of inconsistency.
Although subsequent confirmation by prospective studies involving larger patient cohorts is required, our results suggest that the antibiogram may be of interest in order to optimise the choice of antibiotics on a case-by-case basis. Based on our results and literature, a triple therapy comprising an IV beta-lactam (ceftazidime or meropenem), an oral or IV fluoroquinolone (ciprofloxacin or levofloxacin) plus an inhaled aminoglycoside (tobramycin) appears to be the most suitable option for PC eradication therapy after checking for consistency between this protocol and the resistance profile of the isolated strain. In case of B. gladioli colonisation, piperacillin-tazobactam or meropenem could be preferred to ceftazidime owing to their higher in vitro activity [17,36]. In children, trimethoprimsulfamethoxazole represents an alternative to quinolones. Based on our study and previous observations, IV/ PO treatments could be administered during 2 to 3 weeks whereas longer durations (8 to 12 weeks) could be proposed for inhaled treatments [26,33]. Due to the heterogeneity of the population of CF patients, clinical trials assessing different treatment protocols will probably be required to determine management algorithms taking into account all clinical and microbiological situations (extent of pulmonary involvement, Burkholderia species, co-pathogens). Lastly, given the specific pharmacodynamic profile of cystic fibrosis subjects, antibiotic assays should be encouraged for the purpose of dose optimisation. Given the low incidence of the Burkholderia infection in CF patients, only large-scale multicentre studies can shed light on this issue.

Conclusion
The present study has several limitations, due to the low number of patients, the diversity of species involved as well as of therapeutic approaches, leading to insufficient statistical power. Moreover, it was retrospective. Nevertheless, it highlighted the heterogeneity of therapeutic management of initial infections with Burkholderia species in French CF Centres. Though not statistically significant, treatment efficacy tended to be related to a better lung function at PC, and to consistency with in vitro antibiotic activity. Based on our observations, we suggest the implementation and evaluation of a triple eradication therapy comprising an IV beta-lactam: meropenem, or ceftazidime (BCC), or piperacillin-tazobactam (B. gladioli), an oral or IV fluoroquinolone (ciprofloxacin) and an inhaled aminoglycoside (tobramycin). A follow-up sputum culture can be discussed prior to introducing treatment in milder cases. Large-scale studies are required to standardize Burkholderia PC treatment in cystic fibrosis patients.