Carbapenems-resistant Pseudomonas aeruginosa bloodstream infection in Hematopoietic stem cell transplant patients: in vitro synergy and clinical outcome

Summary Bloodstream infection (BSI) caused by Pseudomonas aeruginosa has high mortality in hematopoietic stem cell transplant (HSCT) recipients. Objectives: To evaluate clinical, in vitro synergy and molecular features of 30 BSI caused by carbapenems-resistant P. aeruginosa (CRPA) in HSCT patients. Methods: Demographic and clinical data including treatment were collected and a database was built using EPIINFO, bivariate and multivariate analysis were run to assess outcome using SPSS. In vitro synergy using time-kill assays, pulsed-field electrophoresis (PFGE) and PCR for carbapenemases and virulence genes were performed for all isolates. Whole genome sequence (WGS) of main clones was done by Nextera XT, using Illumina MiSeq technology. Results: Mortality was 71%; most patients who died were allogeneic HSCT (78%) (p=0.02). Clinical isolates showed a high resistance level to meropenem: 50% had a MIC of 512 µg/mL; two thirds were susceptible to amikacin (MIC 2-512 µg/mL) and 100% to colistin. Many (17/30) isolates achieved synergistic effect with meropenem plus colistin but not with amikacin. None antagonism was observed. The presence of synergy by time-kill between colistin and meropenem showed a tendency towards a better outcome (HR 0.68 95%CI 0.02-1.02; p=0.06). The most frequent carbapenemase gene identified was blaSPM, and six co-harboured both blaKPC and blaSPM. Isolates presented genes related with virulence factors such as toxA, exoS and more patients with BSI caused by P. aeruginosa harbouring gene lasB evolved to death. The WGS showed that all five clones harboured SPM-1, Tn4371 and belonged to ST277; however, the resistance and virulence genes differ among the clones. Conclusion: CRPA in patients. clones Tn4371


Background
Healthcare-associated infections (HAIs) are currently among the major challenges to the quality of patient care in hematopoietic stem cell transplantation (HSCT). Transplanted patients acquire HAIs due to impaired immunity, hospitalization, and use of invasive devices [1].
Bloodstream infection (BSI) caused by Pseudomonas aeruginosa has especially high morbidity and mortality in this population of patients and its empirical treatment is obligatory during febrile neutropenia [1][2][3]. These bacteria can harbour multiple antibiotic resistance mechanisms, leading to limited therapeutic options [4]. Recently, during an outbreak in our hospital, we conduct a case-control study and observed that prior use of carbapenem was the only independent risk factor for carbapenems-resistant P. aeruginosa (CRPA) BSI (P=0.043) [5].
To date, few studies have evaluated antimicrobial combinations and in vitro synergy against CRPA [6][7] In addition to resistance, the expression of several virulence genes by P. aeruginosa strains such as elastase and endotoxins, involved on tissue degradation and biofilm, has been associated with more severe infections and higher mortality [9-11].
We described in vitro synergy assays results in carbapenems-resistant Pseudomonas aeruginosa (CRPA) isolates, including clinical and microbiological data.

Methods
Thirty patients with BSI due to CRPA were included in the study conducted at the bone marrow unit of Hospital das Clínicas, a 1,900-bed, tertiary-care public teaching hospital, from December 2011 through December 2014.

Bloodstream infection diagnostic was based on Centers for Diseases Control and
Prevention (CDC) criteria [12]. Data on epidemiological and clinical characteristics of infection and outcome of HSCT recipients were obtained from electronic medical records and Pitt bacteraemia score was calculated [13]. All transplanted recipients had a central venous catheter and received levofloxacin as antibiotic prophylaxis during neutropenia.
Initial antibiotic therapy was considered appropriate when colistin (COL) was administered within 24 hours after obtain blood culture. All patients received 2g of meropenem (MERO) tid, prolonged infusions (3 hours) adjusted for kidney function when indicated. Amikacin (AMK) was added to the combination therapy with colistin and meropenem whenever CRPA was susceptible. Pseudomonas aeruginosa ATCC27853 was used as control.

In vitro synergy
The synergistic effect was investigated by two methods checkerboard and Time-kill assays.

Checkerboard assay
The CRPA isolates were exposed to combination of two drugs, and checkerboard microdilution testing was performed in duplicate and evaluated after 20-24 hours of incubation at 35°C. Growth and sterility controls were tested in all plates. Colistin, meropenem and amikacin were combined at the respective minimum inhibitory concentrations determined by microdilution. The antimicrobial agents were diluted from the stock solution, and left at concentrations 4 times higher than the final concentration in plate, and then a serial dilution was performed. The results were interpreted using 2well method, with synergy defined as the absence of growth in wells containing 0.25 x MIC of both drugs and 2 x MIC of both drugs [15].

Time-kill assay
Time-kill assay was performed in duplicate with drugs alone and combined at 1x MIC and 0.5x MIC as previously reported [16]. Flasks containing Mueller Hinton broth and the drug were inoculated with testing organisms at a density of ~10 6 cfu/mL, a final volume of 10 mL, and incubated in a shaker at 35°C in ambient air. Aliquots were removed at time 0 and 2, 4, 6, and 24 hours' post-inoculation and serially diluted in 0.85% sodium chloride solution. Diluted samples of 0.01 mL were plated in duplicate on Müeller Hinton agar, and the colonies were counted (log 10 cfu/mL) after 20 hours of incubation at 37°C. Synergism was interpreted as a ≥ 2 log 10 decrease in colony count with the antimicrobial combination compared to the most active single agent; the combination was considered antagonistic for a ≥ 2 log 10 increase in cfu/mL, and indifferent for a < 2 log 10 increase or decrease in count with the combination compared with the most active drug alone [16][17].

Resistance and virulence genes
Polymerase chain reaction (PCR) for the carbapenemase genes, bla SPM , bla VIM , bla NDM, and bla KPC and for virulence genes, lasB, exoS, phZN, toxA and ecfX which is intrinsic in P.
aeruginosa were performed as previously described. 18,19 The molecular profile was assessed by pulsed field gel electrophoresis (PFGE), using S-peI restriction enzyme (Fermentas, USA) in chromosomal DNA Ultrapure Agarose (Invitrogen, Life Technologies).

Results
Among 40 patients affected during entire study period, 30 were included since their bacterial isolates were available and feasible. Posteriorly two were excluded from clinical analysis because their HSCTs were postponed. They were mostly female (18/28); the median age was 42 years old. Acute leukaemia was the most frequent underlying disease (12/28), followed by lymphomas (6/28). A small proportion of patients have disease in full remission (6/28). Nineteen were allogeneic HSCT. Seventeen of twenty-five were not colonized by CRPA prior to infection. The mean Pitt score for bacteraemia severity at the time of infection was 1.10; four patients (14%) had score> 4. The median length of stay before infection was 19 days. Twenty-six patients (93%) had neutropenia at the time of a positive blood culture; 14-Day mortality was 71% (20/28), 95% of them in the first seven days onset of CRPA-BSI (Table 1).
In vitro susceptibility tests of thirty clinical isolates showed a high proportion of resistance to meropenem: 50% (15/30) had a MIC of 512 ug/mL, 47% (14/30) a MIC of 256 ug/mL and one isolate had a MIC of 16 ug/mL. Sixty-six per cent were resistant to amikacin (MIC 2-512 ug/mL), and all of them were susceptible to colistin; moreover, the last four isolates presented higher colistin MICs (4.0 ug/mL). The most frequent carbapenemase identified was SPM, and six co-harboured both bla KPC and bla SPM genes. None carried bla VIM or bla NDM ( Table 2).
The combination of COL plus MERO assessed by time-kill achieved synergy in 57% of isolates (17/30) of COL plus AMK in two of 30 isolates (6.6%), and of MERO plus AMK in 33% (10/30), only in isolates already susceptible to AMK. Of the antimicrobial combinations performed by checkerboard using two-well method interpretation the highest number of synergy was found between MERO plus AMK (9/30). There was no synergistic effect between COL and AMK. No antagonism was observed.
Regarding synergy and treatment outcome, twenty-five patients who received polimyxinbased therapy were analysed separately (Table 3). Most patients who died were allogeneic HSCT recipients (78%) (p=0.02). The presence of synergy by time-kill between colistin and meropenem showed a tendency towards to better outcome (HR 0.68 95%CI 0.02-1.02; p=0.06), which was not observe with the combination between meropenem and amikacin.
All five virulence genes searched were isolated in 21 isolates (70%) by PCR technique.
The analysis revealed the presence of several genes related to resistance to aminoglycosides, beta-lactam, fluoroquinolone, phenicol and sulphonamide. Regarding virulence genes, all the strains harboured genes involved on quorum sensing, biofilm formation, adhesion and invasion process and cytotoxicity.

Discussion
Our findings demonstrated a high mortality of CRPA-BSI in HSCT patients, and that death was more frequent among allogeneic patients, similar to those found in the literature [22].
In our cohort, however, Pitt bacteraemia score did not discriminate severe cases as observed in intensive care units [10]. Interestingly, prior gut colonization was also not common in this population, maybe due to low sensitivity of traditional culture methods to assess colonization or as an indication of cross-transmission as important source for HAI infection in our hopsital 5 . Although, five clones were found, we observed a predominant clone, showing that cross -infection occurred during the study period. In contrast with previous studies [6], the combination with amikacin was not synergistic in our casuistic, and even in a small population of patients, this finding highlights possible harmful effects of this combination, such as nephrotoxicity, with no apparent benefit.
Combination therapy is tempting since it increases the likelihood of success in initial therapy, which is crucial in a population as vulnerable as neutropenic. Hu et al. (2013) conducted a meta-analysis of combination therapy versus monotherapy for P. aeruginosa bacteraemia included 10 clinical studies in a meta-analysis but did not find a mortality difference. Up to now, the best experimental results occurred with doripenem as part of in vitro combination [8], but the clinical experience with this drug in the HSCT population is scarce. Doripenem is not available in Brazil, thus, we could not evaluate it as an option to treat our patients [25][26].
The isolates harboured important virulence attributes such as exoS, toxA. Recently, a study showed an association between exoU gene expression and mortality among 590 patients with P. aeruginosa bacteremia. 8 The exoU gene was not identified in our clones probably because they are multidrug resistant. We found that patients with BSI caused by isolated harbouring gene lasB evolved more often to death. This virulence factor is associated with bacterial enzyme production of elastase that degrades immunoglobulin and complement factor [27].
The most frequent carbapenemase identified was bla SPM; however, six co-harboured bla KPC and bla SPM . The bla SPM was located in the Tn4371, alerting for the potential of dissemination of this lineage in the world [28][29][30][31][32][33][34] Although, all clones belonged to ST277, which is common in our country and has been described in several outbreaks including patients who travelled to Brazil [29][30][31][32][33][34]. The WGS brought important information regarding resistance and virulence genes that differs among the clones. For instance, 16S rRNA methyltransferase gene rmtD1, which confers high-level resistance to all aminoglycosides and has been associated to ST277 was present only in clones A, D and E. Clone C did not harbour fosA; it carried exoS and toxA of Type Three Secretion System (TTSS) and lacked the exoY gene. The analyses also confirm this large number of virulence factors such as exoS and exoY, as well as quorum sensing, biofilm and genes of phenazine operons that are responsible for increasing intracellular oxidative stress [8,25].
There are few studies of CRPA-BSI especially in such a homogeneous patient population.
However, the small number of patients is an important limitation of our study.

Conclusions
The CRPA-BSI had high mortality in HSCT patients and even higher among allogeneic patients. Pitt bacteraemia score did not discriminate severe cases as observed in intensive care units. Treatment using combination of meropenem with colistin, showed a tendency towards lower mortality and displayed high in vitro synergy. Further investigation using antibiotic combination treating a large population is essential to confirm our findings.

Availability of data and materials
The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.   Dendrogram showing the relatedness of PFGE patterns among thirty isolates of Pseudomonas aeruginosa in BSI construct using 0.5 otimization and 1.5 tolerance.