Vancomycin-Resistant Enterococcal Bloodstream Infections on a Hematopoietic Stem Cell Transplant Unit: Are the Sickgetting Sicker?

ORIGINAL ARTICLE Vancomycin-resistant enterococcal bloodstream infections on a hematopoietic stem cell transplant unit: are the sickgetting sicker?

ER Dubberke1,2, JM Hollands3, P Georgantopoulos2, K Augustin3, JF DiPersio1,4, LM Mundy5,6 and HJ Khoury1,4

1Department of Medicine, Washington University School of Medicine, St Louis, MO, USA; 2Division of Infectious Diseases, Washington University School of Medicine, St Louis, MO, USA; 3Barnes-Jewish Hospital, St Louis, MO, USA; 4Division of Oncology, Section Leukemia and Bone Marrow Transplantation, Washington University School of Medicine, St Louis, MO, USA; 5School of Public Health, St Louis University, St Louis, MO, USA and 6The Jonathan Lax Treatment Center, Philadelphia, PA, USA

Patients with hematologic malignancies and hemato- Introduction poietic stem cell transplant (HSCT) recipients are at high riskfor bacterial bloodstream infections (BSI) owing to Patients with hematologic malignancies and hematopoietic resistant organisms. Data describing the outcomes of stem cell transplant (HSCT) recipients are at high risk for vancomycin-resistant enterococcal (VRE) BSI in this bloodstream infections (BSI) that arise from organisms that patient population are limited. We performed a retro- normally colonize the human intestine.1–4 Because of their spective cohort study of all cases of VRE BSI that occured high severity of underlying illness, prolonged hospitaliza- between February 1996 and December 2002 on the tion and frequent broad-spectrum antibiotic exposures, Leukemia/HSCT unit at Barnes-Jewish Hospital. There these patients are commonly colonized with resistant were 68 episodes of VRE BSI in 60 patients with acute organisms and therefore at high risk for BSI owing to (53%) or chronic (8%) leukemia, non-Hodgkin’s lym- resistant organisms.1,2 This is important because infections phoma (22%) or other malignant hematologic disorders owing to resistant organisms are associated with increased (17%). A total of 13, 32 and 32% were recipients of morbidity, costs and mortality.5 autologous, related and matched-unrelated transplants, Enterococci are normal inhabitants of the human colon respectively. Forty-two of allograft recipients had active and are the third most frequent cause of health-care- acute graft-versus-host disease (GVHD) and 32% chronic associated infections in the United States.6–8 Up to 73% of GVHD. Only 57% were neutropenic, 52% had refrac- enterococcal BSIs are due to vancomycin-resistant strains.9 tory/relapsed malignancy and 60% had end organ Risk factors for vancomycin-resistant enterococcal (VRE) dysfunction with a median APACHE II score of 17. colonization and infection include prolonged hospitaliza- Median survival after VRE BSI was 19 days. Pneumonia, tion, intensive care unit (ICU) stay, advanced age, receipt of anti-fungal drugs and low APACHE II score at immunocompromised state, neutropenia, high severity of the time of the VRE BSI remained signiﬁcant riskfactors underlying illness, antibiotic exposure, and indwelling for death on multivariable analysis. Our analysis suggests urinary and vascular catheters,8,10–13 characteristics com- that in patients with hematological malignancies or monly found in patients on HSCT units. Several studies HSCT, VRE may not have the behavior of a virulent have investigated the signiﬁcance of VRE infections in pathogen. VRE BSI may simply be a marker of these different patient populations and controversy exists as to patients’ already existing critical medical condition. whether VRE infections are associated with worse out- Bone Marrow Transplantation (2006) 38, 813–819. comes compared to vancomycin-sensitive enterococci.14–21 doi:10.1038/sj.bmt.1705530; published online 23 October 2006 Although some studies have indicated VRE infections are Keywords: vancomycin-resistant enterococcus; blood- associated with worse outcomes in immunocompromised stream infection; transplantation; leukemia; mortality patients,19,20 data on VRE infections in patients with hematologic malignancies and HSCT recipients are limited.21,22 Therefore we conducted a retrospective cohort study to determine the patient characteristics associated with VRE BSI and risk factors for death in these patients.

Correspondence: Dr ER Dubberke, Department of Medicine, Division of Infectious Diseases, Washington University School of Medicine, 660 S Euclid, Box 8051, St Louis, MO 63110, USA. Patients and methods E-mail: [email protected] Preliminary data was presented in part at the 2006 BMT Tandem Meetings, Honolulu, HI. Study design Received 20 April 2006; revised 10 July 2006; accepted 18 August 2006; All cases of VRE BSI occurring between February 1996 published online 23 October 2006 and December 2002 on the HSCT unit at Barnes-Jewish VRE bloodstream infections on an HSCT unit ER Dubberke et al 814 Hospital were prospectively identified by the Infection qualitative PCR or culture from the blood or a quantitative Control Department. Barnes-Jewish Hospital is a 1251-bed PCRX10 000 copies/ml. CMV pneumonia was defined tertiary care facility located in St Louis, MO and is as the presence of histopathological evidence of CMV affiliated with Washington University School of Medicine. pneumonia or a positive culture by bronchoalveolar lavage The HSCT unit has 27 positive pressure, high efficiency in a patient with radiographic evidence of pneumonia. particulate air-filtered, single-occupancy rooms and can GVHD was diagnosed and graded according to standard accommodate up to six patients requiring ICU care. All definitions.25,26 Mortality attributable to VRE BSI was patients with acute leukemia and HSCT recipients are defined as a patient who had an acute clinical deterioration admitted to this unit. Nine hundred and sixty-eight temporally associated with the VRE BSI culminating in autologous and 612 allogeneic transplants were performed death. during the study period. Although there was no active screening for enteric VRE colonization during the study period, VRE screening was carried out on all stools sent for Data analysis Clostridium difficile testing. The standard antibiotic regi- Risk factors for death were analyzed by the Student’s t-test mens for neutropenic fever during the study period or Mann–Whitney U-test for continuous variables and the 2 consisted of cefepime monotherapy (ceftazidime from w test for categorical variables. Kaplan–Meier analysis was February 1996 to January 1998) for leukemics, autologous, used to calculate median survival after VRE BSI. Back- and sibling allogeneic transplants recipients and imipenem wards, stepwise Cox proportional hazards (Pp0.05 for plus vancomycin for unrelated donor allogeneic recipients. entry, PX0.1 for exclusion) were used to determine risk Washington University School of Medicine Human Studies factors for death at discharge on multivariable analysis. Committee approval was obtained for this study. For measures of association, a two-tailed Pp0.05 was Chart review was conducted with a standardized data considered to be significant. collection tool. Patient-related variables collected at the time of the VRE BSI included: age, underlying disease, stage of underlying disease, transplant history, Eastern Results Cooperative Oncology Group (ECOG) score on admission and at time of the VRE BSI, co-morbidities, graft- Epidemiological characteristics of VRE colonization versus-host disease (GVHD), immunosuppressant drugs, and BSI APACHE II score 48 h before and at the time of VRE BSI During the 83-month-study period 334 patients screened and discharge status. All inpatient antibiotics, chemo- positive for enteric VRE colonization. Sixty-eight episodes therapeutic agents and infections were collected if of VRE BSI were observed in 60 patients (for patients present within 42 days before the VRE BSI. Antibiotic with 41 VRE BSI only data from the first VRE BSI are exposures were aggregated by type of coverage (gram described from this point forward). The incidence of VRE positive, gram negative, anaerobic and fungal) as described BSI ranged between 0.6 and 0.9 VRE BSI/1000 patient days previously.17,23 Vancomycin exposure was also reported from 1996 to 1998 (Figure 1). During this period it was separately. hospital policy for all health-care workers to wear gowns and gloves before entering the room of patients colonized with VRE. In 1998 this policy was changed to gloves alone. Study definitions This was followed by a sharp rise in the rate to 2.1 VRE Episodes of VRE BSI were prospectively identified by the BSI/1000 patient-days in 2001. As a result of this increase, Infection Control Department at Barnes-Jewish Hospital. gowns and gloves were reinstituted and the rate regressed A VRE BSI was defined as X2 sets of blood cultures back to 1.3 VRE BSI/1000 patient-days. positive for VRE drawn within 24 h of each other or one set of blood cultures positive and the presence of fever 1 (438.0 C), chills or hypotension when cultures were VRE BSI Rate from 1996 through 2002 drawn. Infections were considered secondary if the same Gown use re-instituted for organism was cultured from another site, other than stool, 2.5 VRE precautions within the previous 48 h. Bloodstream infections were considered polymicrobial if more than one organism grew 2.0 from the same set of blood cultures. VRE BSIs were considered recurrent if VRE was isolated from blood 1.5 Gown use discontinued cultures after treatment with an appropriate dose and for VRE precautions duration of an effective antibiotic(s) plus documentation of 1.0 negative blood cultures or resolution of symptoms associated with the first BSI after the initiation of antibiotics. 0.5 Infections were defined by the National Nosocomial

Infection Surveillance System (NNIS) deﬁnitions for Rate (per 1000 patient days) 24 0.0 nosocomial infections. Infections were considered fungal 1996 1997 1998 1999 2000 2001 2002 if fungi grew from an otherwise sterile site or invasive disease was seen on histopathology. Cytomegalovirus Year (CMV) viremia was deﬁned as persistently positive Figure 1 VRE BSI rates during the study period.

Bone Marrow Transplantation VRE bloodstream infections on an HSCT unit ER Dubberke et al 815 Microbiological characteristics Table 1 Demographic characteristics of patients who developed a Of the 334 patients who were known to be enterically VRE BSI on a HSCT unit (n (%) or median (range)) colonized with VRE, 42 (13%) developed a VRE BSI. Characteristic Number N ¼ 60 Therefore 70% (42/60) of those who developed VRE BSI were known to be enterically colonized with VRE when Age 46 (20–75) they developed their VRE BSI. Eight (13%) VRE BSI were Male 34 (57) White 53 (88) polymicrobial BSIs. Other organisms isolated were coagu- Hospitalized in previous 30 days 41 (68) lase-negative Staphylococcus (n ¼ 4), Pseudomonas aerugi- ECOG Score on admission 2 (1–4) nosa (n ¼ 3), Acinetobacter calcoacetius (n ¼ 1) and Stenotrophomonas maltophilia (n ¼ 1). All VRE isolates Reason for admission were Enterococcus faecium except one isolate, which was Chemotherapy/Transplant 29 (48) Infection 18 (30) Enterococcus feacalis. The median duration of VRE GVHD 8 (13) bacteremia was 1 day (range 1–9); 22 (37%) of patients Othera 5 (8) had more than 1 day of VRE bacteremia. Fifty-five (92%) patients had primary VRE BSI. The secondary VRE BSI Underlying diagnosis Acute leukemia 32 (53) were attributed to the urinary tract (n ¼ 3) and intra- Lymphoma 13 (22) abdominal infections (n ¼ 2). Chronic leukemia 5 (8) Myelodysplastic syndrome 4 (7) Otherb 6 (10) Patient characteristics The median age of VRE BSI patients was 46 years (range Underlying diagnosis status 20–75) (Table 1); 57% were male and the median ECOG New diagnsosis 5 (8) score on admission was 2 (range 1–4). The most common Remission 24 (40) Refractory/Relapse 31 (52) diagnosis was acute leukemia (53%), followed by lymphoma (22%) and chronic myelogenous leukemia (8%); 24 Transplant status (40%) patients were in remission, and 31 (53%) had None 14 (23) relapsed or refractory disease. At the time of the VRE BSI, Autologous 8 (13) Allogeneic: related donor 19 (32) 19 (32%) patients were recipients of matched unrelated Allogeneic: unrelated donor 19 (32) donor transplants, 19 (32%) sibling matched transplants, eight (13%) autologous transplants and 14 (23%) had not Cancer treatment been transplanted. Seventeen (45%; 17/38) allograft re- Previous total body irradiation 34 (57) cipients were transplanted after a failed allograft and one Chemotherapy in previous 42 days 38 (63) Neutropenic (ANCo1000 PMN/ml) 34 (57) (12%) after a failed autograft. The VRE BSI occurred a median 12 days after transplant (range À12 to 770) in Infections in previous 42 days allogeneic and 10.5 days (À5 to 77) in autologous Non-VRE bloodstream infection 28 (47) recipients. At the time of the VRE BSI, 16 (42%) of the Pneumonia 21 (35) Invasive mold infection 6 (10) allograft recipients had acute GVHD at one or more sites CMV disease 13 (22) (14 skin, seven gastrointestinal, five liver) and 12 (32%) had chronic GVHD (10 skin, six gastrointestinal, four liver and Antimicrobial exposures (days) two mucosalivary). Gram-positive agents 15 (2–42) Gram-negative agents 14 (2–42) Anaerobic agents 10 (2–42) Severity of underlying illness Anti-fungals 8 (2–42) Sixty-eight percent of patients with VRE BSI were Vancomycin 10 (1–42) previously hospitalized in the 30 days before the admission Known to be colonized with VRE 42 (70) during which the VRE BSI occurred. The median length of stay before the VRE BSI was 17 days (range 0–77). Abbreviations: ANC ¼ absolute neutrophil count; CMV ¼ cytomegalo- Infections were common in the 42 days preceding the VRE virus; ECOG ¼ Eastern Cooperative Oncology Group; GVHD ¼ graft- BSI (some patients had more than one infection). Twenty- versus-host disease; HSCT ¼ hematopoietic stem cell transplant; PMN ¼ polymorphonuclear cells; VRE BSI ¼ vancomycin-resistant enter- eight (47%) patients had a non-VRE bacterial BSI in the 42 ococcal bloodstream infection. days preceding the VRE BSI, 6 (10%) had C. difficile- aOther: diffuse alveolar hemorrhage (n ¼ 1), multisystem organ failure associated disease and 21 (35%) had pneumonia. Several (n ¼ 1), jaundice (n ¼ 1), data not available (n ¼ 2). b patients had fungal infections: 10 (17%) had an invasive Other: multiple myeloma (n ¼ 2), aplastic anemia (n ¼ 2), testicular cancer candidal infection (six candidemia, one lung, one hepatos- (n ¼ 1), ovarian cancer (n ¼ 1). plenic candidiasis, one disseminated, one urinary tract infection), six (10%) had an invasive mold infection (four positive antibiotics, 56 (93%) received anti-gram-negative lung, one lung and brain, one sinus), and one had antibiotics, 44 (73%) received anti-anaerobic antibiotics, disseminated blastomycosis. Thirteen (22%) patients had 36 (60%) received anti-fungals and 49 (82%) received CMV disease (10 viremia, one pneumonia, one dissemi- vancomycin. nated and one encephalitis). Antibiotic exposures in the 42 Mucocutaneous barrier disruption was nearly ubiqui- days preceding the VRE BSI reflected the prevalence of tous. There were 66 central venous catheters in 58 (97%) infections. Fifty-seven (95%) patients received anti-gram- patients at the time of the VRE BSI (34 tunneled catheters,

Bone Marrow Transplantation VRE bloodstream infections on an HSCT unit ER Dubberke et al 816 19 non-tunneled catheters, and 15 implanted ports) and 31 1.0 (52%) had a urinary catheter. Mucositis was reported in 42 Survival (70%) patients, of whom 11 had Xgrade 2 mucositis, with function 19 (32%) receiving enteral feeding (17 nasogastric tube, 0.8 Censored two percutaneous tubes) and 20 (33%) total parenteral nutrition (TPN). As mentioned above, 13 patients had 0.6 gastrointestinal GVHD (seven acute, six chronic) and six had C. difﬁcile-associated disease. End-organ dysfunction was common: 20 (33%) patients 0.4 had liver dysfunction (serum bilirubin 42.0 mg/dl), 19

(32%) had acute neurologic dysfunction (mental status Cumulative survival 0.2 changes or new focal deﬁcit), 13 (22%) patients were mechanically ventilated, 11 (18%) patients had acute renal 0.0 dysfunction (serum creatinine 42.0 mg/dl or doubling of baseline; two required acute dialysis) and six (10%) had 0 102030405060 gastrointestinal dysfunction (ileus or bleeding requiring Days from VRE BSI to death transfusion). The mean APACHE II score 48 h before the VRE BSI was 17.4 and the mean APACHE II score at the Figure 2 Kaplan–Meier analysis of survival to discharge. time of the VRE BSI was 18.1 (P ¼ 0.23).

Table 2 Risk factors for not surviving VRE BSI hospitalization on Treatment of VRE BSI univariate analysis (n (%) or median (range)) All enterococcal blood isolates at our institution were Variable Alive Dead P-value tested for sensitivity to vancomycin, ampicillin, gentamicin n ¼ 28 n ¼ 32 and streptomycin. All isolates except one (the E. faecalis isolate) were resistant to all antibiotics tested. The one Age (years) 49 (22–75) 44 (20–67) 0.018 Hospitalized in previous 30 days 15 (47) 26 (81) 0.043 isolate that was not pan-resistant was sensitive to ampi- ECOG score X2 on admission 14 (50) 24 (75) 0.012 cillin. Upon detection of the VRE BSIs, 33 (55%) patients underwent central venous catheter removal. Thirty (50%) Transplant status 0.004 patients were treated with linezolid (three also received Allogeneic 12 (43) 26 (81) ampicillin), 16 (27%) with chloramphenicol (two also Autologous 7 (25) 1 (3) None 9 (32) 5 (16) received ampicillin), three (5%) with quinuprisitin/dalfopristin and one (o1%) patient was treated with high-dose Admitted for GVHD 1 (4) 7 (22) 0.037 ampicillin alone. In 25 (42%) patients, antibiotics with Pneumonia 5 (18) 16 (50) 0.014 activity against VRE were started within 48 h from the time Receipt of anti-fungals 11 (40) 25 (78) 0.004 Receipt of anti-anaerobic antibiotics 17 (61) 27 (84) 0.047 the culture was obtained. Ten patients did not receive any Mechanical ventilation 2 (7) 12 (38) 0.006 antibiotics with potential activity against VRE, all of whom Acute neurological dysfunction 3 (11) 16 (53) 0.001 had a single culture positive for VRE. Three died within Receipt of TPN 4 (14) 16 (50) 0.006 48 h from when the blood culture was drawn. Four of the APACHE II score at time 16.5 (7–25) 17.5 (1–32) 0.035 seven who did not die within 48 h were treated with removal of VRE BSI Treated with chloramphenicol 2 (7) 14 (44) 0.001 of their central venous catheters alone. All four survived Treatment with linezolid 17 (63) 13 (41) 0.119 4210 days after the VRE BSI. Of the three who received Central venous catheter removed 20 (71) 13 (41) 0.021 no speciﬁc treatment, one was discharged on hospice and died 29 days after the VRE BSI from disease progression. Abbreviations: ECOG ¼ Eastern Cooperative Oncology Group; GVHD ¼ The other two survived 4240 days after the VRE BSI. graft-versus-host disease; TPN ¼ total parenteral nutrition; VRE BSI ¼ None of the seven developed recurrent VRE BSI. vancomycin-resistant enterococcal bloodstream infection.

Variables associated with death during the VRE BSI Survival after VRE BSI hospitalization on univariate analysis included age, Thirty-two (53%) patients died during the VRE BSI hospitalization in the previous 30 days, ECOG score X2 hospitalization and the median survival after VRE BSI on admission, transplant status, admission for GVHD, was 19 days (Figure 2). Only four deaths were felt to be pneumonia, receipt of non-prophylactic anti-fungals or directly attributable to the VRE BSI by chart review. All anti-anaerobic antibiotic, mechanical ventilation, acute four patients developed septic shock that coincided with the neurologic dysfunction, TPN, APACHE II score at time VRE BSI without other identiﬁable causes. Two of the of VRE BSI, VRE BSI treated with chloramphenicol and patients had myelodysplastic syndrome, one had acute non-removal of the central venous catheter after VRE BSI myelogenous leukemia and one had Hodgkin’s disease. (Table 2). There was a trend towards improved survival Three patients had an allogeneic transplant (one unrelated when linezolid was used to treat the VRE BSI. Factors not donor), all of whom had GVHD. Three of the patients were associated with death include underlying disease, CMV treated with chloramphenicol and one with linezolid. Time disease, invasive mold infection, neutropenia at time of from VRE BSI to death ranged from 2 to 20 days. VRE BSI, severe mucositis, change in APACHE II score

Bone Marrow Transplantation VRE bloodstream infections on an HSCT unit ER Dubberke et al 817 Table 3 Hazard ratios associated with not surviving hospitaliza- received linezolid whereas only 27% were treated with tion in patients with VRE BSI by multivariable Cox proportional chloramphenicol. On univariate analysis we found treat- hazards ment of the VRE BSI with chloramphenicol to be Variable HR 95% CI associated with an increased risk of death. Of note, none of our patients treated with chloramphenicol experienced Admitted for GVHD 6.6 2.4–17.7 any marrow suppression owing to the chloramphenicol Pneumonia in previous 42 days 2.3 1.1–4.9 (data not shown). Receipt of anti-fungals 4.2 1.6–11.3 22 APACHE II Score at time of VRE BSI 1.1 1.0–1.2 Avery et al. described the outcomes of 12 allogeneic transplant recipients who developed VRE BSI. Ten of 12 Abbreviations: CI ¼ confidence interval; GVHD ¼ graft-versus-host dis- patients experienced VRE BSI during the neutropenic ease; HR ¼ hazard ratio; VRE BSI ¼ vancomycin-resistant enterococcal period (median 15 days post transplant, range 8–21), and bloodstream infection. mainly this occurred in patients with refractory or relapsed disease at the time of transplantation (70%). All VRE BSI patients were co-infected with other organisms including from 48 h before VRE BSI to day of VRE BSI, 41 day of gram-negative bacilli, fungi or CMV, and five patients had VRE bacteremia, initiation of treatment for VRE within intra-abdomial complications (GVHD, cholecystitis or 48 h of the VRE BSI and recurrent VRE BSI. Variables typhlitis). Patients received linezolid (five), chlorampheni- associated with death on multivariable Cox analysis col (two) and quinuprisitin/dalfopristin (n ¼ 5). Patients included hospitalization for GVHD, pneumonia, receipt died a median of 18 days after onset of VRE BSI (range 7– of non-prophylactic anti-fungals and APACHE II score at 73), and only one death was directly attributable to VRE. the time of the VRE BSI (Table 3). Results of this study are comparable to our analysis with regards to the co-morbidities present at the time of onset of VRE BSI and the low rate of death directly attributable to VRE. However, in our study, VRE BSI was observed Discussion later after allografting with only 10/38 (26%) VRE BSI occurring within 21 days post transplant in allograft In this study, we sought to describe HSCT recipients and recipients. patients with hematological malignancies who developed There were several important findings of this study that VRE BSI and determine risk factors for death over an 83- have impacted how we manage our patients. Although month-study period. We found that patients who devel- treatment with linezolid and removal of the central venous oped VRE BSI were severely ill from complications related catheter were not associated with improved survival on to the underlying disease and its treatment, and often had multivariable analysis, because they were associated with concomitant infections and frequent exposures to anti- improved outcomes on univariate analysis they have microbial agents. Although 52% of patients with VRE BSI become the standard of care for treating VRE BSI on our did not survive their hospitalization, only four deaths were HSCT unit. VRE in the US is almost completely health- felt to be directly attributable to the VRE BSI by chart care-associated and is transmitted primarily by the hands review. Enterococcus is not an inherently virulent organ- of health-care workers. Our data demonstrated a trend ism.27 This may be why three patients survived without any towards a decrease in the incidence of VRE BSI with more specific therapy. These patients had only one culture aggressive infection control measures. This trend continued positive indicating they had a transient bacteremia. By beyond the study period with continued gown and glove the time it was known VRE was the causative organism of use when caring for patients with VRE colonization (data the BSI the patients had not received appropriate not shown). This finding stresses the importance of antibiotics for several days. As repeat blood cultures did preventing the spread of VRE, and wearing gowns and not grow VRE, the primary team did not feel it necessary to gloves when caring for patients colonized with VRE has start therapy active against VRE. This is in line with our remained the standard of care on out HSCT unit. findings that VRE was not the direct cause of death in the Our study suffers from the same limitations as other majority of our patients that died after their VRE BSI. This studies focused on patients with VRE BSI and studies observation raises an important question: did patients die to determine the attributable mortality owing to infec- from the VRE BSI or with the VRE BSI? tions,14–20,28,29 that is, lack of appropriate controls, limited Recently DiazGranados and Jernigan20 reported the heterogenous objective methods controlling for the com- impact of vancomycin resistance on clinical outcomes in plexity and the severity of underlying illness and impact of 22 neutropenic patients who developed a VRE BSI and effective therapy over time. Our study does describe the compared their outcomes to 61 neutropenic patients largest cohort of HSCT recipients and patients with who developed vancomycin-senstivie enterococcal BSIs. hematologic malignancies with VRE BSI in the literature Seventy-seven (93%) of the patients had an underlying to date. We believe that VRE BSI was not the direct cause hematologic malignancy. Mortality of patients with VRE of death in most patients who died. This is supported by BSI was 64% at 60 days after the onset of bacteremia. They our findings that factors previously associated with an found vancomycin resistance to be a risk factor for death increased risk of death owing to infections, timing of owing to prolonged bacteremia (HR 4.9, CI 1.2-20.4). All antibiotics and duration of bacteremia, were not associated patients with VRE BSI in their study were treated initially with an increased risk of death in this study. In addition, with chloramphenicol. In our study, 50% of patients the APACHE II score did not change significantly from

Bone Marrow Transplantation VRE bloodstream infections on an HSCT unit ER Dubberke et al 818 48 h before the VRE BSI to the time of the VRE BSI, stream infection in patients with cancer who are colonized indicating the VRE BSI was not associated with an acute with vancomycin-resistant enterococci. Clin Infect Dis 2002; clinical deterioration in the majority of patients. However, 35: 1139–1146. these patients were extremely ill at baseline and the 13 Safdar N, Maki DG. The commonality of risk factors for additional insult of the VRE BSI could have contributed nosocomial colonization and infection with antimicrobial- Staphylococcus aureus to death in ways too subtle for us to detect. resistant , enterococcus, gram-negative bacilli, Clostridium difficile, and Candida. Ann Intern Med In conclusion, this study suggests that in HSCT 2002; 136: 834–844. recipients and patients with hematological malignancies 14 Edmond MB, Ober JF, Dawson JD, Weinbaum DL, Wenzel VRE may not have the microbiological behavior of a RP. Vancomycin-resistant enterococcal bacteremia: natural virulent pathogen. VRE BSI may simply be a marker of history and attributable mortality. Clin Infect Dis 1996; 23: disease severity in patients that are otherwise at high risk 1234–1239. for death. 15 Shay DK, Maloney SA, Montecalvo M, Banerjee S, Wormser GP, Arduino MJ et al. Epidemiology and mortality risk of vancomycin-resistant enterococcal bloodstream infections. J Infect Dis 1995; 172: 993–1000. Acknowledgements 16 Lucas GM, Lechtzin N, Puryear DW, Yau LL, Flexner CW, Moore RD. Vancomycin-resistant and vancomycin-susceptible We thank Victoria J Fraser MD, Cherie Hill and Stacey enterococcal bacteremia: comparison of clinical features and Leimbach for their indispensable assistance on this project. This outcomes. Clin Infect Dis 1998; 26: 1127–1133. work was supported by a grant from the Centers of Disease 17 Garbutt JM, Ventrapragada M, Littenberg B, Mundy LM. Control and Prevention (1W01C000333-02). Association between resistance to vancomycin and death in cases of Enterococcus faecium bacteremia. Clin Infect Dis 2000; 30: 466–472. References 18 Lautenbach E, Bilker WB, Brennan PJ. Enterococcal bacteremia: risk factors for vancomycin resistance and 1 Fainstein V, Rodriguez V, Turck M, Hermann G, Rosenbaum predictors of mortality. Infect Control Hosp Epidemiol 1999; B, Bodey GP. Patterns of oropharyngeal and fecal flora in 20: 318–323. patients with acute leukemia. J Infect Dis 1981; 144: 10–18. 19 Linden PK, Pasculle AW, Manez R, Kramer DJ, Fung JJ, 2 Schimpff SC, Young VM, Greene WH, Vermeulen GD, Pinna AD et al. Differences in outcomes for patients with Moody MR, Wiernik PH. Origin of infection in acute bacteremia due to vancomycin-resistant Enterococcus faecium nonlymphocytic leukemia. Significance of hospital acquisition or vancomycin-susceptible E. faecium. Clin Infect Dis 1996; 22: of potential pathogens. Ann Intern Med 1972; 77: 707–714. 663–670. 3 Schimpff SC. Infections in cancer patients: differences between 20 DiazGranados CA, Jernigan JA. Impact of vancomycin developed and less developed countries? Eur J Cancer 1991; 27: resistance on mortality among patients with neutropenia and 407–408. enterococcal bloodstream infection. J Infect Dis 2005; 191: 4 Schimpff SC. Gram-negative bacteremia. Support Care Cancer 588–595. 1993; 1: 5–18. 21 Small TN, Boulad F, Papadopoulos E, Jaffe D, O’Reilly R, 5 Cosgrove SE, Sakoulas G, Perencevich EN, Schwaber MJ, Sepkowitz K et al. Incidence and outcome of vancomycin- Karchmer AW, Carmeli Y. Comparison of mortality asso- resistant enterococcal bacteremia following allogeneic hema- ciated with methicillin-resistant and methicillin-susceptible topoietic stem cell transplant (HSCT). In: Program and Staphylococcus aureus bacteremia: a meta-analysis. Clin Infect abstracts of the 42nd Interscience Conference on Antimicrobial Dis 2003; 36: 53–59. Agents and Chemotherapy (San Diego). American Society for 6 Kaye D. Enterococci. Biologic and epidemiologic character- Microbiology: Washington, DC, 2002, p 325. istics and in vitro susceptibility. Arch Intern Med 1982; 142: 22 Avery R, Kalaycio M, Pohlman B, Sobecks R, Kuczkowski E, 2006–2009. Andresen S et al. Early vancomycin-resistant enterococcus 7 Low DE, Keller N, Barth A, Jones RN. Clinical prevalence, (VRE) bacteremia after allogeneic bone marrow transplant- antimicrobial susceptibility, and geographic resistance patterns ation is associated with a rapidly deteriorating clinical course. of enterococci: results from the SENTRY Antimicrobial Bone Marrow Transplant 2005; 35: 497–499. Surveillance Program, 1997-1999. Clin Infect Dis 2001; 32 23 Puzniak LA, Mayfield J, Leet T, Kollef M, Mundy LM. (Suppl 2): S133–S145. Acquisition of vancomycin-resistant enterococci during 8 Schaberg DR, Culver DH, Gaynes RP. Major trends in the scheduled antimicrobial rotation in an intensive care unit. microbial etiology of nosocomial infection. Am J Med 1991; Clin Infect Dis 2001; 33: 151–157. 91: 72S–75S. 24 Horan TC, Emori TG. Definitions of key terms used in the 9 Treitman AN, Yarnold PR, Warren J, Noskin GA. Emerging NNIS System. Am J Infect Control 1997; 25: 112–116. incidence of Enterococcus faecium among hospital isolates 25 Glucksberg H, Storb R, Fefer A, Buckner CD, Neiman PE, (1993 to 2002). J Clin Microbiol 2005; 43: 462–463. Clift RA et al. Clinical manifestations of graft-versus-host 10 Handwerger S, Raucher B, Altarac D, Monka J, Marchione S, disease in human recipients of marrow from HL-A-matched Singh KV et al. Nosocomial outbreak due to Enterococcus sibling donors. Transplantation 1974; 18: 295–304. faecium highly resistant to vancomycin, penicillin, and 26 Rowlings PA, Przepiorka D, Klein JP, Gale RP, gentamicin. Clin Infect Dis 1993; 16: 750–755. Passweg JR, Henslee-Downey PJ et al. IBMTR Severity Index 11 Wells VD, Wong ES, Murray BE, Coudron PE, Williams DS, for grading acute graft-versus-host disease: retrospective Markowitz SM. Infections due to beta-lactamase-producing, comparison with Glucksberg grade. Br J Haematol 1997; 97: high-level gentamicin-resistant Enterococcus faecalis. Ann 855–864. Intern Med 1992; 116: 285–292. 27 Moellering RC. Enterococcus Species, Streptococcus bovis, 12 Zaas AK, Song X, Tucker P, Perl TM. Risk factors for and Leuconostoc Species. In: Mandell GL, Bennett JE, development of vancomycin-resistant enterococcal blood- Dolin R (eds). Principles and Practice of Infectious Diseases,

Bone Marrow Transplantation VRE bloodstream infections on an HSCT unit ER Dubberke et al 819 Sixth edn. Elsevier Churchill-Livingstone: Philadelphia, 2005, 29 Bhavnani SM, Drake JA, Forrest A, Deinhart JA, pp 2411–2421. Jones RN, Biedenbach DJ et al. A nationwide, multicenter, 28 Lodise TP, McKinnon PS, Tam VH, Rybak MJ. Clinical case-control study comparing risk factors, treatment, outcomes for patients with bacteremia caused by vancomycin- and outcome for vancomycin-resistant and -susceptible enter- resistant enterococcus in a level 1 trauma center. Clin Infect Dis ococcal bacteremia. Diagn Microbiol Infect Dis 2000; 36: 2002; 34: 922–929. 145–158.

Bone Marrow Transplantation