Impact of empiric treatment for vancomycin-resistant Enterococcus (VRE) in colonized patients early after allogeneic hematopoietic stem cell transplantation

Abstract

In recent years, vancomycin-resistant Enterococcus (VRE) colonization is increasingly being encountered among transplant populations, and VRE has become one of the leading causes of bacteremia early after allogeneic hematopoietic stem cell transplantation (allo-HSCT). There are sparse data regarding the effect of empiric VRE therapy for febrile, neutropenic allo-HSCT recipients colonized with VRE. All allo-HSCT patients aged 18 years and older who developed VRE bacteremia (VREB) between 2005–2014 were identified and categorized as to whether they received empiric versus (vs) directed VRE therapy. There were 434 (33%) VRE-colonized and 872 (67%) non-VRE-colonized patients during the study period, and 172 of 434 (40%) VRE-colonized patients received empiric therapy. There was no significant difference in incidence of VREB among colonized patients who did or did not receive empiric therapy (28/172, 16% vs 55/262, 21%; P=0.22). There was a total of 95 patients with VREB, of which the majority (83/95, 87%) was known to be VRE-colonized. Of the 95 VREB episodes, 29 (31%) received empiric VRE therapy, whereas 66 (69%) received directed therapy. No significant differences in clinical outcome, including median duration of bacteremia (2 vs 2 days, P=0.39), recurrent VREB (3/29, 10% vs 5/66, 8%, P=0.65), 30-day all-cause mortality (1/29, 3% vs 4/66, 6%, P=0.62), or VRE-attributable mortality (1/29, 3% vs 1/66, 2%, P=0.55) were observed between the empiric- and directed-therapy groups. Kaplan-Meier curve showed no significant difference in survival at 30 days in allo-HSCT recipients with VREB who received empiric vs directed therapy (97% vs 94%, P=0.62). We recommend against empiric use of VRE-active agents for fever and neutropenia in VRE-colonized patients undergoing allo-HSCT.

INTRODUCTION

Vancomycin-resistant Enterococcus (VRE) is now a common problem among allogeneic hematopoietic stem cell transplant (allo-HSCT) recipients and is the leading cause of nosocomial bloodstream infections (BSI) at some centers, especially in the early post-transplant period [1, 2]. Gastrointestinal colonization with VRE is a strong predictor of BSI in this population [3–5]. The progression to VRE bacteremia (VREB) early post-transplant has been reported to occur in 11–34% of VRE-colonized patients [4–7]. Although VRE is often considered to be an organism of limited virulence, some data suggest that VREB can be associated with severe presentations (e.g., septic shock) in allo-HSCT recipients [1].

While the American Society for Blood and Marrow Transplantation (ASBMT), Infectious Diseases Society of America (IDSA), and the Society for Healthcare Epidemiology of America (SHEA) do not recommend routine screening for VRE, it may be reasonable to consider based on local epidemiology and/or evidence for ongoing transmission in a HSCT unit [8]. A natural question that follows is how to optimally manage VRE-colonized patients during the transplant course. The 2010 IDSA guideline states that modification to initial empiric therapy for fever and neutropenia may be considered for patients at risk for infection with antibiotic-resistant organisms, including VRE, particularly if the patient’s condition is unstable or if the patient has positive blood culture results suspicious for resistant bacteria [9]. Whether such practice would improve clinical outcomes is not clear, as there are no prospective studies to date. In a recent retrospective report, investigators from Brazil found that empiric use of linezolid in VRE-colonized hematology and allo-HSCT patients did not change in-hospital mortality [10].

At our center, linezolid was used for empiric therapy in combination with piperacillin/tazobactam (or if penicillin-allergic, cefepime) with fever after allo-HSCT in VRE-colonized neutropenic patients from June 2006 to January 2011. We conducted a review of all adult allo-HSCT recipients who had pre-engraftment VREB between January 1, 2005 and December 31, 2014 to compare all-cause mortality at 30 days in patients who received empiric versus (vs) directed VRE therapy.

MATERIALS AND METHODS

This study was reviewed and approved by the Institutional Review Board at Memorial Sloan Kettering Cancer Center (MSKCC), a 473-bed tertiary care cancer center in New York City, NY. All adults 18 years and older who underwent allo-HSCT and developed VREB between January 1, 2005 and December 31, 2014 were identified and categorized as to whether they received empiric vs directed VRE therapy. Data were collected from institutional transplant, pharmacy, and infection control databases, as well as microbiology records. Abstracted data included demographics, underlying cancer diagnosis, co-morbidities (e.g., type 2 diabetes, chronic renal insufficiency), conditioning regimen, transplant and donor types, VRE screening results, and reason for intensive care unit (ICU) transfer. The reason for ICU transfer was evaluated to determine if the critical illness was due in part to VREB. Clinical variables within 24 hours of blood culture collection, including fever (≥38.1°C), hypotension (<90/50 mm Hg), alteration in mental status, presence of moderate to severe mucositis [11], and signs of central venous catheter (CVC) infection (e.g., focal redness at catheter site), were also recorded. The primary outcome was all-cause mortality at 30 days after the incident VRE blood culture between the two groups. Secondary outcomes included time-to-clearance of VREB, recurrence of VREB, and VRE-attributable mortality.

VRE surveillance:

During the study period, patients admitted for allo-HSCT underwent rectal swab screening for VRE colonization initially by culture on BBL Campy CVA agar (January 2005-June 2013) and then by a commercial polymerase chain reaction (PCR) system that targets the Van A gene since July 2013 [12]. Screening was obtained on admission to the transplant unit and then weekly until discharge throughout the study period. All VRE-colonized and/or infected persons were placed on contact barrier precautions.

Laboratory methods:

For blood culture specimens, if gram positive cocci in pairs and chains were detected on the Gram stain, further identification with catalase, pyrrolidonyl arylamidase (PYR) test, vancomycin disc susceptibility, and a MicroScan® Positive ID 2 Panel was performed until September 2014 when matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry (Vitek MS; bioMérieux, St. Louis, MO) was implemented. Sensitivities were performed by using MicroScan® POS MIC 34 panel (Beckman Coulter, Brea, CA).

Antibiotic management:

Since 2006, we implemented intravenous (IV) vancomycin prophylaxis, alone or with a fluoroquinolone (FQ), during the pre-engraftment period to prevent viridans streptococcal bacteremia [2, 13]. Antibiotic prophylaxis was started 2 days prior to stem cell infusion (day-2). FQ prophylaxis was continued until neutrophil engraftment or onset of fever and neutropenia, whichever occurred first. Vancomycin prophylaxis was continued until day+7, after which continuation of vancomycin was at the discretion of the treating physician as needed for management of fever and neutropenia or treatment of a documented infection.

Between June 2006 and January 2011, VRE-colonized, febrile, neutropenic patients were switched from IV vancomycin plus/minus FQ prophylaxis to empiric IV linezolid 600 mg every 12 hours in combination with IV piperacillin/tazobactam 4.5 gm every 6 hours (or if penicillin-allergic, IV cefepime 2 gm every 8 hours). Empiric linezolid was continued until blood cultures were negative for 48 hours, at which point the linezolid was stopped or switched back to IV vancomycin. Sporadic use of linezolid as empiric VRE therapy was observed before June 2006 or after January 2011 at the discretion of the primary team; these cases were included in the analysis. Linezolid was the recommended VRE-active agent to be used for empiric intent based on local susceptibilities; daptomycin was rarely used [3].

Definitions:

Standard consensus definitions for fever, neutropenia, and sepsis were used [9, 14]. Neutrophil engraftment was defined as 3 consecutive days with an absolute neutrophil count (ANC) >500 cells/mm³. Empiric treatment for VRE referred to initiation of linezolid (or rarely, IV daptomycin 6 mg/kg daily) with fever in a VRE-colonized, neutropenic patient in the absence of known positive blood culture(s) for VRE. Directed VRE therapy referred to initiation of a VRE-active agent when gram positive cocci in pairs and chains was visualized on Gram stain from one or more positive blood cultures in a VRE-colonized, febrile, neutropenic patient; antibiotic therapy was modified based on final identification and sensitivities of the organism. Only VREB occurring prior to neutrophil engraftment was considered for this study. VREB was attributed to a gastrointestinal source following the National Health Safety Network (NSHN)’s mucosal barrier injury criteria [15]. Persistent VREB occurred if the patient had positive blood cultures >3 calendar days. Mortality was attributed to VREB if the patient did not clear VRE at time of death and/or the patient never recovered from sepsis.

Statistical analysis:

Characteristics of patients with VREB who received empiric vs directed VRE therapy were compared using a Chi-square or Fisher’s exact test for categorical variables and Mann-Whitney U test for continuous variables. Stepwise multiple logistic regression analysis was used for odds ratio in multivariate analyses. All variables from univariate analysis were entered into the model if P < 0.3 and stayed in the final model if P <0.1. Overall survival probabilities were estimated using the Kaplan-Meier method. Log-rank test was used in time-to-event analyses. A P value < 0.05 was considered statistically significant. All statistical analyses were performed using SAS 9.4 (SAS, Cary, NC).

RESULTS

VRE colonization and bacteremia:

During the study period (January 1, 2005 to December 31, 2014), 1,306 adult allo-HSCT were performed at MSKCC; 71 patients underwent 2 transplants. There were 434 (33%) VRE-colonized and 872 (67%) non-VRE-colonized patients. Eighty-three of 434 (19%) VRE-colonized patients developed VREB, whereas 12/872 (1%) non-VRE-colonized patients developed VREB. All 12 patients had undergone VRE screening by culture.

Approximately 40% (172/434) of VRE-colonized patients received empiric VRE therapy (Figure 1). There was no significant difference in incidence of VREB among colonized patients who did or did not receive empiric therapy (28/172, 16% vs 55/262, 21%; P=0.22). Of 172 recipients of empiric VRE therapy, 139 (81%) received it between June 2006 and January 2011 when this practice was standard of care. Linezolid (164/173, 95%) was the primary agent for empiric VRE therapy.

Figure 1.

Patient flowchart

*Twelve of 872 non-VRE-colonized patients had VREB of whom 1 received empiric therapy and 11 received directed therapy.

Abbreviations: VRE, vancomycin-resistant enterococcus; VREB, vancomycin-resistant enterococcal bacteremia

In all, there were 95 allo-HSCT recipients with pre-engraftment VREB, of whom 83 of 95 (87%) were tested and were known to be VRE-colonized. Of the 95 VREB episodes, 29 (31%) received empiric VRE therapy, whereas 66 (69%) received directed VRE therapy. In the empiric VRE treatment group, most (25/29, 86%) developed VREB on the same day as the initiation of linezolid.

Demographic and clinical characteristics of patients with VREB:

The median age of patients with VREB was 59 years (range, 20–70 years) (Table 1). Fifty-nine patients (62%) were men. The majority (83/95, 87%) of patients with VREB had acute leukemia or myelodysplastic syndrome. There were no significant differences in age, sex, underlying malignancy, conditioning regimen, transplant type, prior VREB, gastrointestinal or CVC source, hypotension, altered mental status, or ICU transfer between the two groups (Table 1). The empiric VRE therapy group was more likely to have matched related donors, diabetes, chronic renal insufficiency, and fever (Table1).

Table 1.

Comparison of demographics and other clinical characteristics in patients with VREB who received empiric versus directed VRE therapy

	Empiric therapy N=29 (%)	Directed therapy N=66 (%)	P value
Median age (range)	58 (26–69)	59 (20–70)
Male sex	22 (76)	37 (56)	0.11
Underlying malignancy
Acute leukemia	24 (83)	49 (74)	0.46
Chronic leukemia	0 (0)	3 (5)
Lymphoma	2 (7)	2 (3)
Other*	3 (10)	12 (18)
Conditioning
Myeloablative	15 (52)	38 (57)	0.76
Reduced intensity	14 (48)	27 (41)
Non-myeloablative	0 (0)	1 (2)
Transplant type
Conventional	3 (10)	12 (18)	0.68
Cord	3 (10)	8 (12)
T cell depletion	23 (80)	46 (70)
Donor type
Matched related	15 (52)	17 (25)	0.01
Matched unrelated	11 (38)	23 (35)
Mismatched related	0 (0)	1 (2)
Mismatched unrelated	3 (10)	25 (38)
Diabetes	11 (38)	12 (18)	0.04
Chronic renal insufficiency	5 (17)	3 (5)	0.05
Prior VRE bacteremia	4 (14)	4 (6)	0.24
Fever ≥38.1°C	28 (97)	39 (59)	<0.001
Hypotension (<90/50)	0 (0)	6 (9)	0.17
Altered mental status	2 (7)	8 (12)	0.72
ICU transfer due to VREB	1 (3)	11 (17)	0.10
Gastrointestinal source	22 (76)	51 (77)	0.88
Central venous catheter source	7 (24)	15 (23)	0.88

^*Myelodysplastic syndrome (n=10), myeloproliferative disorder (n=1), multiple myeloma (n=3), T cell leukemia/lymphoma (n=1)

Abbreviations: ICU, intensive care unit; VRE, vancomycin-resistant enterococcus; VREB, vancomycin-resistant enterococcal bacteremia

Outcomes:

There were no significant differences in 30-day all-cause mortality, median duration of VREB, recurrent VREB, and attributable mortality due to VREB between the empiric and directed VRE treatment groups (Table 2). We then analyzed factors to examine their association with mortality via logistic regression. Prior VREB, hypotension, and persistent VREB were associated with 30-day mortality, whereas age, sex, conditioning regimen, transplant and donor type, diabetes, fever, altered mental status, and empiric VRE treatment were not associated with 30-day all-cause mortality on univariate analysis (Table 3). None of the factors remained significant on multivariate analysis (Table 3). Kaplan-Meier curve showed no significant difference in survival at 30 days in allo-HSCT recipients with VREB who received empiric vs directed VRE therapy (Figure 2).

Table 2.

Comparison of outcomes in patients with VREB who received empiric versus directed VRE therapy

	Empiric therapy N=29 (%)	Directed therapy N=66 (%)	P value
Median VREB duration (range)	2 (1–9)	2 (1–19)	0.39
Recurrent VREB	3 (10)	5 (8)	0.65
Attributable mortality	1 (3)	1 (2)	0.55
30-day all-cause mortality	1 (3)	4 (6)	0.62
60-day all-cause mortality	4 (14)	8 (12)	0.84

Abbreviations: VRE, vancomycin-resistant enterococcus; VREB, vancomycin-resistant enterococcal bacteremia

Table 3.

Univariate and multivariate regression analysis for predictors of 30-day all-cause mortality in allogeneic hematopoietic stem cell transplant recipients with VREB

	Predictors of 30-Day All-Cause Mortality
Variable	Univariate Odds Ratio (95% CI)	P	Multivariate Odd Ratio (95% CI)	P
Age	1.06 (0.95–1.17)	0.31
Sex (Male)	0.39 (0.06–2.43)	0.31
Conditioning
Myeloablative	Ref
Reduced intensity	2.01 (0.32–12.65)	0.97
Non-myeloablative	NA
Transplant type
Conventional	Ref
Cord	1.45 (0.17–12.23)	0.12
T cell depletion	0.10 (0.01–1.13)	0.03
Donor type
Matched related	Ref
Matched unrelated	NA
Mismatched related	NA
Mismatched unrelated	5.17 (0.54–49.27)	0.94
Diabetes	2.19 (0.34–13.99)	0.41
Prior VRE bacteremia	9.33 (1.30–67.03)	0.03	9.38 (0.8–110.54)	0.08
Fever ≥38.1°C	0.61 (0.01–3.86)	0.60
Hypotension (<90/50)	14.33 (1.84–111.44)	0.01	11.34 (0.9–142.76)	0.06
Altered mental status	2.25 (0.23–22.38)	0.49
Empiric VRE therapy	0.55 (0.06–5.18)	0.60
Persistent VREB >48 hours	11.65 (1.24–109.66)	0.03	9.44 (0.88–101.33)	0.08

Abbreviations: CI, confidence interval, NA, not applicable; VRE, vancomycin-resistant enterococcus; VREB, vancomycin-resistant enterococcal bacteremia

Figure 2.

Kaplan-Meier curve of survival at 30 days comparing allogeneic hematopoietic stem cell transplant recipients with VREB who received empiric (n=28) versus directed (n=62) VRE therapy

Abbreviations: VRE, vancomycin-resistant enterococcus; VREB, vancomycin-resistant enterococcal bacteremia

DISCUSSION

Our findings that were derived from a large retrospective cohort of over 1,300 allo-HSCT recipients with high VRE colonization prevalence (33%) at the time of transplant failed to demonstrate any clinical benefit of empiric therapy among colonized patients.

Gram-positive bacteria remain a dominant cause of pre-engraftment bloodstream infections in allo-HSCT [16, 17], and multidrug-resistant pathogens, including VRE, are frequently isolated [1, 2]. At our institution, VRE is the leading cause of pre-engraftment BSI in allo-HSCT recipients [2], and VRE invasion of the bloodstream has been shown to be preceded by its predominance in the gastrointestinal tract [18, 19].

Although gram-positive organisms are now among the most common bacteria causing infection early after transplant and chemotherapy, empiric use of vancomycin has shown no clinical benefit or difference in all-cause mortality with the conclusion being that vancomycin can be safely deferred until the documentation of an infection that would warrant its use [20, 21]. For VRE infections, the empiric use of gram-positive agents such as linezolid, daptomycin, quinupristin-dalfopristin, and tigecycline among patients with known VRE colonization undergoing induction chemotherapy for acute leukemia or allo-HSCT has not been prospectively evaluated in clinical studies. It has been reasoned that results of VRE surveillance can potentially guide antimicrobial therapy for fever and neutropenia, but the benefit of empirically initiating VRE-active therapy while awaiting blood culture results must be carefully weighed against the potential harms of this approach [1, 22]. The most recent iteration of the IDSA practice guideline for use of antimicrobial agents in high-risk febrile patients with neutropenia recommends considering early use of VRE-active agents in areas of high endemicity [9].

In our current study, the majority (137/172, 81%) of VRE-colonized patients received empiric therapy as per institutional guideline when it was in effect between June 2006 and January 2011. Our study findings showed no clinical benefit of using linezolid as empiric treatment in VRE-colonized, febrile, neutropenic patients early after HSCT. Specifically, no significant differences in median VREB duration, recurrent VREB, VRE-attributable mortality, or 30-day all-cause mortality were found when comparing patients with VREB who received empiric therapy to those who did not. No difference in 60-day all-cause mortality was seen either. When we looked at factors affecting 30-day all-cause mortality, empiric VRE therapy was not significant on either univariate or multivariate analysis. In another study, Lisboa and colleagues considered linezolid on a case-by-case basis for VRE-colonized patients upon persistence of fever after 48 hours of broad-spectrum gram-negative coverage and a glycopeptide antibiotic [10]. They also found no impact of this practice on mortality, which instead appeared to be associated with other factors such as persistence of neutropenia.

The reason that empiric therapy may not have been effective in improving survival is not clear. Poor outcomes seen in allo-HSCT recipients with VREB have been shown in several studies to not have been directly attributable to VRE [23–25]. VREB in and of itself may simply be an indicator of gut inflammation, such as toxicity from the conditioning regimen leading to mucositis or from graft-versus-host-disease (GVHD), both of which are known to be associated with morbidity and mortality in allo-HSCT recipients [10, 25, 26]. Thus, delay in VRE-active therapy may not have detrimental consequences in contrast to BSIs due to gram-negative bacteria or Candida [27, 28]. In our study, empiric therapy was administered a median of 2 hours from the incident blood culture in comparison to a median of 24 hours in patients receiving directed therapy. This delay did not appear to convey clinically measurable differences in examined outcomes. The increasing adoption of rapid diagnostics (e.g., peptide nucleic acid fluorescence in situ hybridization (PNA FISH), matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry, multiplexed nucleic acid amplification tests (NAAT)) by clinical microbiology laboratories may further make the empiric strategy superfluous, as these technologies combined with antimicrobial stewardship intervention have been shown to improve time-to-appropriate therapy [29–31].

Our study has several limitations. The study design is retrospective, so undetected biases may have been present that determined who did and who did not receive early VRE-active treatment. Because linezolid was by far the most frequently used VRE-active agent, our finding may be restricted only to linezolid use and not to other VRE-active agents. However, we and others have reported increasing enterococcal resistance to daptomycin, which makes routine use of this agent problematic [3, 32, 33]. Consideration of quinupristin/dalfopristin or tigecycline for empiric therapy may be limited by factors such as side effect profile (e.g., quinupristin/dalfopristin and myalgias/arthralgias, tigecycline and nausea/vomiting), tissue versus serum distribution (e.g., tigecycline), and safety (e.g., tigecycline) [34, 35]. Given the low 30-day mortality rates in both arms, our sample size may also not have been sufficient to detect differences.

Because our findings showed no benefit of empiric VRE therapy on clinical outcomes and mortality, we stopped this practice at our center in early 2011. Instead administration of VRE-active agents early after transplant is contingent upon the recovery of gram positive cocci in pairs and chains from blood cultures in allo-HSCT patients known to be colonized with VRE. Similarly, the most recent guideline for the treatment of fever and neutropenia by the German Society of Hematology and Medical Oncology (DGHO) does not recommend the addition of linezolid to empiric first-line treatment in VRE-colonized patients [36].

In conclusion, empiric therapy did not impact 30-day all-cause mortality, median duration of VREB, recurrent VREB, or VRE-attributable mortality in allo-HSCT recipients with pre-engraftment VREB. Based on our findings, we do not recommend empiric VRE therapy in colonized patients undergoing allo-HSCT with fever and neutropenia.

Highlights.

• VRE colonization occurred in 33% of allogeneic HSCT patients.

• VRE bloodstream infection (BSI) occurred in 19% of VRE-colonized patients.

• Empiric linezolid for fever and neutropenia had no impact on 30-day mortality.

• Empiric linezolid had no effect on duration or recurrence of VRE BSI.

• We do not support empiric therapy in VRE-colonized allogeneic HSCT patients.