Epidemiology, treatment and outcomes of bloodstream infection due to vancomycin-resistant enterococci in cancer patients in a vanB endemic setting

Background Vancomycin-resistant enterococcus (VRE) is an important cause of infection in immunocompromised populations. Few studies have described the characteristics of vanB VRE infection. We sought to describe the epidemiology, treatment and outcomes of VRE bloodstream infections (BSI) in a vanB predominant setting in malignant hematology and oncology patients. Methods A retrospective review was performed at two large Australian centres and spanning a 6-year period (2008–2014). Evaluable outcomes were intensive care admission (ICU) within 48 h of BSI, all-cause mortality (7 and 30 days) and length of admission. Results Overall, 106 BSI episodes were observed in 96 patients, predominantly Enterococcus faecium vanB (105/106, 99%). Antibiotics were administered for a median of 17 days prior to BSI, and 76/96 (79%) were neutropenic at BSI onset. Of patients screened before BSI onset, 49/72 (68%) were found to be colonised. Treatment included teicoplanin (59), linezolid (6), daptomycin (2) and sequential/multiple agents (21). Mortality at 30-days was 31%. On multivariable analysis, teicoplanin was not associated with mortality at 30 days. Conclusions VRE BSI in a vanB endemic setting occurred in the context of substantive prior antibiotic use and was associated with high 30-day mortality. Targeted screening identified 68% to be colonised prior to BSI. Teicoplanin therapy was not associated with poorer outcomes and warrants further study for vanB VRE BSI in cancer populations.

Studies evaluating risks for BSI and clinical outcomes have focussed on vanA VRE as the predominant genotype or have been performed in settings where vanA strains are dominant [2,10]. Although there have been recent reports of increased vanA in Australia, vanB has historically been the predominant genotype [9]. Few studies describe risk factors and outcomes of vanB VRE BSI particularly in the high-risk hematology and oncology population. Two previous case-control studies have described risk factors for vanB VRE BSI in unselected patients [2,14]. They described the use of central venous access devices (CVAD), neutropenia, allogeneic hematopoietic transplant, urinary catheterisation and duration of metronidazole therapy as risk factors. A study focusing on vanB VRE BSI in patients with hematological malignancy found acute myeloid leukemia and vancomycin therapy as risk factors [15]. A single case control study investigating factors influencing mortality and length of stay found prior intensive care unit stay and burden of comorbidities to be associated with mortality and linezolid therapy with lower mortality [1]. However, this study was not restricted to patients with malignancy and the median duration of neutropenia was only 1 day.
The objective of this study was to describe the epidemiology, treatment and outcomes of VRE BSI in patients with solid tumours or hematological malignancies in a vanB endemic setting at two large Australian healthcare facilities. We also sought to describe the prevalence of risk factors previously associated with vanA BSI in this population in order to inform targeted interventions for improved patient outcomes.

Setting
Retrospective chart review was performed at two adult tertiary hospitals in Melbourne, Australia: The Royal Melbourne Hospital (RMH) and the Peter MacCallum Cancer Centre (PMCC). The study period was defined as 1 January 2008 to 31 December 2013 (RMH) and 1 January 2008 to 30 June 2014 (PMCC). Both centres are major tertiary referral teaching hospitals with oncology and haematology units, including autologous (PMCC) and allogeneic (RMH) bone marrow transplant services. During this period, active surveillance for VRE was performed at both centres by collection of weekly perianal swabs from inpatients with no known history of VRE colonisation or infection. Patients who were VRE colonized on surveillance swabs or with a previous history of VRE in a clinical isolate were placed in contact precautions requiring the use gowns and gloves on entry to the patient room.

Study population
Cases of vancomycin-resistant E. faecium or E. faecalis BSI during the defined study period were identified from laboratory extracts. Patients with underlying hematological or oncological diseases were eligible for inclusion. All patients with at least one positive blood culture isolating VRE were classified as having a BSI. Blood cultures were obtained with aseptic technique via peripheral venepuncture or through a CVAD.

Laboratory confirmation
Chromogenic medium (ChromID VRE agar plate, Biomerieux) was used for detection and differentiation of VRE from screening swabs. Identification of Enterococci in bloodstream isolates was based on VITEK 2 (Biomerieux). Genetic testing for vanA and vanB were not routine during the study period. Clinical isolates were deemed to be vanB if phenotypically susceptible to teicoplanin and resistant to vancomycin by VITEK 2 (Biomerieux) using Clinical and Laboratory Standards Institute (CLSI) guidelines (CLSI 2015, version M100-S25). Isolates resistant to teicoplanin were confirmed to be vanA by PCR (Xpert vanA/vanB, Cepheid, Sunnyvale, USA).

Data collection
Data including patient demographics, underlying disease, comorbidities, presence of CVAD or urinary catheters, WHO mucositis grade, treatment and outcomes were captured by retrospective chart review and using a standardised data collection tool. Consistent with previous studies, the Chronic Disease Score specific to VRE (CDS-VRE) was used as a quantitative measure of comorbidities for VRE infection [2,16]. The number of days of antibiotic use in the 30 days preceding the first positive blood culture was obtained by medication chart review. The antibiotic used, sequence and duration of antibiotics were recorded. Evaluable outcomes included intensive care unit (ICU) admission within 48 h of first positive blood culture for VRE, all-cause mortality at 7 and 30 days, and length of hospital admission after BSI.

Definitions
Days of neutropenia in the 30 days prior to first positive blood culture were defined as the number of days where the absolute neutrophil count was ≤0.5 × 10^9/L. Hypoalbuminemia was defined as a serum albumin of < 35 g/ L. VRE colonization was defined as a positive rectal swab for VRE prior to first positive blood culture or previous VRE infection. Polymicrobial infection was defined as isolation of one or more additional organisms in a VREpositive blood culture.

Ethics review
The study was reviewed and approved by Human Research Ethics Committees at the Royal Melbourne Hospital (reference number: QA2015054) and Peter MacCallum Cancer Centre (reference number: 16/35R).

Statistical analysis
Data analysis was performed using IBM SPSS Statistics version 26.0 (IBM Corp., Armonk, NY, USA). Categorical variables were expressed as frequencies. Continuous variables were reported as mean or median for normally or non-normally distributed data, respectively. Where a patient had more than one episode of VRE BSI, only the first episode was included in analysis. Univariable and multivariable logistic regression analyses were performed to evaluate the association between outcomes and potential risk factors, with covariates having p ≤ 0.2 included in multivariable models. To enable analysis of treatment outcomes, patients who only received teicoplanin, linezolid or daptomycin throughout the treatment course were grouped separately to patients receiving multiple or sequential antibiotics. Patients who were not treated due to end-of-life care were excluded from the logistic regression analyses.

Patient characteristics
A total of 106 bloodstream infections in 96 patients were identified. Ten patients had more than one episode of BSI. Most patients had an underlying hematological malignancy (83/96, 86%), with the most common hematological malignancy being acute myeloid leukemia (40/83, 48%). There was no significant change in the number of cases per year through the study period. Patient characteristics are summarized in Table 1.
All isolates were Enterococcus faecium ( Table 2). All cases demonstrated vanB phenotype except for one case of vanA VRE BSI during the study period. Twenty-three (24%) cases were polymicrobial BSI.
Seventy two out of 96 patients were screened for VRE prior to BSI. Of these, forty-nine (68%) screened positive.

Treatment and outcomes
Teicoplanin was used at any point during the treatment course in 80 patients (80/95, 84%) although 21 patients had sequential therapy with teicoplanin, daptomycin or linezolid. Teicoplanin was used alone throughout the treatment course in 59 patients, linezolid alone in 6 patients and daptomycin alone in 2 patients. Data regarding targeted antibiotic treatment were unavailable for 1 patient. Median duration of treatment in patients who received teicoplanin as a part of their treatment was 14 days (range 1-60). Teicoplanin was given as a 400-1200 mg load 12-hourly (6-12 mg/kg) for three doses then as 400-1200 mg daily maintenance with adjustment for renal function. No loading doses were given in 3 cases.
Seven patients were not treated (7/95, 7%). In four instances, patients were not treated as infection episodes were deemed to be line-related with prompt clinical improvement observed after line removal or were deemed to be contaminants by the treating physician. In three instances, patients were receiving end-of-life care and targeted therapy was therefore not commenced.

Risks for ICU admission and mortality
With respect to risk factors for ICU admission within 48 h of blood culture collection (Table 3), multivariable analysis demonstrated mucositis grade was independently associated with ICU admission (OR 1.81, 95% CI 1.04-3.14) and targeted treatment with teicoplanin was associated with lower odds of ICU admission (OR 0.14, 95% CI 0.03-0.66). Patients who only received linezolid or daptomycin therapy were not inputted into the regression model due to low numbers. With respect to risk factors for 30-day mortality (Table 4), multivariable analysis revealed ICU admission within 48 h of BSI to be associated with higher odds of 30-day mortality (OR 4.16, 95% CI 1.08-16.00). Teicoplanin monotherapy throughout the treatment course was not associated with 30-day mortality (OR 0.57, 95% CI 0.20-1.64).

Discussion
Our study describes the characteristics and outcomes of predominantly vanB-phenotype VRE BSI in a tertiary hematology and oncology setting. The most common underlying condition in this group of patients was acute myeloid leukemia (AML). We found substantive antibiotic use prior to onset of BSI and high 30-day mortality after VRE BSI similar to that described for vanA VRE [18][19][20]. We observed targeted treatment with teicoplanin to not be associated with 30-day mortality. Colonization with VRE was identified by screening prior to the onset of BSI in a large proportion (68%) of patients, supporting early commencement of VRE-active empiric therapy as a component of appropriate sepsis management in patients with known colonization.
Previous studies including predominantly vanA isolates have identified prior antibiotic exposure, hypoalbuminemia, neutropenia and the presence of indwelling venous catheters to be risk factors for development of BSI [14,15,18,[21][22][23]. Our study found a median of 11 days of hypoalbuminemia and 11 days of neutropenia prior to onset of BSI. Prior studies have demonstrated a median of 13 days of hypoalbuminemia and between 1 and 32 days of neutropenia with longer durations of neutropenia in allogeneic hematopoietic transplant patients [2,14,18]. Most patients in our study (81%) had a CVAD at the time of BSI. AML and exposure to vancomycin have been associated with vanB VRE bacteraemia previously [15]. Consistent with these findings, we found AML as the most frequent underlying condition. In addition, we found a median of 17 antibiotic days exposure prior to BSI, generally related to administration of broad-spectrum agents (meropenem, piperacillintazobactam or vancomycin). While the cumulative duration of antibiotic exposure was not reported, the median duration of exposure to vancomycin, piperacillintazobactam and meropenem is similar to that reported previously [1,21].
We found death at 30 days was 31% and comparable to previous Australian studies of vanB VRE BSI between 21 and 36% [1,15]. Mortality after VRE BSI among patients with hematological malignancies was also similar to previous reports [18][19][20]. Few previous studies have investigated treatment outcomes associated with teicoplanin for vanB VRE BSI. A case-control study in Australia suggested that linezolid treatment for vanB VRE was associated with lower in-hospital mortality than teicoplanin [1]. We found that treatment with teicoplanin monotherapy was associated with lower rates of ICU admission within 48 h of VRE BSI. However, this may have been influenced by the selective use of linezolid or daptomycin in more unwell patients, rather than an effect of teicoplanin itself. Comparison is also limited by the smaller number of patients receiving linezolid We identified mucositis to be associated with need for ICU admission within 48 h of VRE BSI. This may reflect a greater disturbance of innate immunological barriers predisposing to infection. There was a concurrent high rate of polymicrobial BSI in our cohort consisting of almost a quarter of all episodes. However, mucositis and polymicrobial BSI were not associated with higher 30day mortality.
Others have proposed that VRE BSI may contribute to mortality or may just be a marker for severe underlying disease with conflicting reports on the effectiveness of early empiric therapy [1,4,[24][25][26][27]. It has been reported that in patients with hematological malignancy, rates of severe sepsis within 2 days of VRE BSI can be as high as 36% [6]. Sixteen percent of our patients required ICU admission within 48 h of onset of BSI and the need for ICU admission was associated with 30-day mortality on multivariable analysis, although the confidence interval was wide (OR 4.16, 95% CI 1.08-16.00). Septic shock at  onset of BSI has previously been associated with mortality at 28-days with a hazard ratio of 1.91 [26]. The median length of hospitalization after VRE BSI was 18 days. This was the comparable to previous studies for vanB VRE BSI and for malignant hematology patients in a vanA endemic setting [1,20].
Rectal colonization with VRE has been reported as a risk factor for subsequent BSI in recipients of allogeneic hematopoietic stem cell transplantation [4,28,29]. Notably, 68% of our cohort who were screened for VRE prior to BSI were positive, highlighting the potential benefits of screening to guide empiric therapy. We suggest that while screening for colonization by collection of rectal swabs will not identify all patients who develop VRE BSI, when combined with other clinical risk factors such as mucositis and prior antibiotic exposure, it can help identify patients at high risk. Risk scores, such as that developed by Webb et al. may aid risk stratification [28], which may then inform treatment decisions (e.g. administration of VRE-active antimicrobial agents) in high-risk, and unwell patients. Another approach could be the early addition of VRE-active antibiotics in patients known to be VRE colonized who develop positive blood cultures with gram-positive cocci resembling streptococci/enterococci. Protocols for neutropenic sepsis at our institutions have changed over recent years to reflect the latter approach.
A limitation of this study is the retrospective design, and inability to confirm causal association between risk factors and outcomes. Genotyping for vanA/vanB genes were not performed as routine during the study period. VRE isolates with teicoplanin-susceptible phenotype but harbouring the vanA gene have been described. However, Australia-wide national surveillance had described low rates of vanA during the study period and undetected vanA isolates were unlikely to have contributed significantly to the results [30].

Conclusion
In summary, we report clinical characteristics and outcomes of a large cohort of patients with vanB VRE bacteremia within a hematology and oncology setting. VRE BSI in a vanB predominant environment occurred in the context of significant antibiotic use and was associated with high 30-day mortality similar to that described for vanA BSI. We did not find that teicoplanin was associated with poorer outcomes and further comparative study of teicoplanin as an option for treatment in vanB VRE BSI is warranted. Screening for VRE carriage with rectal swabs identified more than two thirds of patients with subsequent VRE BSI, highlighting the benefit of surveillance for informing decisions regarding early effective treatment in the setting of sepsis within this vulnerable population. Authors' contributions OX performed literature review, data collection, data analysis and manuscript preparation. LW conceived project design and carried out data collection, data analysis and manuscript preparation. MS and AB contributed to project design, supervision and manuscript preparation. BT and AD assisted with data collection and manuscript review. All authors read and approved the final manuscript.

Funding
No funding was received to perform this study.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
The study was reviewed and approved by Human Research Ethics Committees at the Royal Melbourne Hospital (reference number: QA2015054) and Peter MacCallum Cancer Centre (reference number: 16/ 35R). All data was anonymised before use in the study.

Consent for publication
Not applicable.