Changing microbial epidemiology in hematopoietic stem cell transplant recipients: increasing resistance over a 9-year period

Transcription

1 214 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd Transplant Infectious Disease, ISSN Changing microbial epidemiology in hematopoietic stem cell transplant recipients: increasing resistance over a 9-year period N. Macesic, C.O. Morrissey, A.C. Cheng, A. Spencer, A.Y. Peleg. Changing microbial epidemiology in hematopoietic stem cell transplant recipients: increasing resistance over a 9-year period. Transpl Infect Dis 214: 16: All rights reserved Abstract: Infections remain important contributors to mortality in hematopoietic stem cell transplantation (HSCT). Method. We studied the evolving epidemiology and trends in susceptibility of bacterial and Candida isolates at an Australian HSCT center. A total of 528 HSCTs in 58 patients were performed from April 21 to May 21. A total of 65 isolates were eligible for study inclusion; 318 (53%) were gram-positive, 268 (44%) were gramnegative, and 19 (3%) were Candida species. Results. The most common site for isolates was blood (38 isolates, 63%). Staphylococcus aureus was the most common gram-positive organism (n = 17, 34%), but trends to increasing coagulasenegative staphylococci (P =.2) and vancomycin-resistant Enterococcus (P <.1) were observed. Escherichia coli was the most common gram-negative isolate (n = 74, 28%). Fluoroquinolone resistance increased with widespread use of protocol fluoroquinolone prophylaxis (P =.1). Carbapenem resistance was found in 44% of Pseudomonas or Acinetobacter isolates. Bloodstream infection with a multidrug-resistant organism (odds ratio 3.61, 95% confidence interval: , P =.8) was an independent predictor of mortality at 7 days after a positive blood culture. Conclusions. Antimicrobial resistance is an increasing problem in this vulnerable patient population, and not only has an impact on choice of empiric therapy for febrile neutropenia but also on mortality. N. Macesic 1, C.O. Morrissey 1,2,3, A.C. Cheng 1,3,4, A. Spencer 2,5, A.Y. Peleg 1,3,6 1 Department of Infectious Diseases, Alfred Hospital, Melbourne, Victoria, Australia, 2 Department of Haematology, Central Clinical School, Monash University, Melbourne, Victoria, Australia, 3 Department of Infectious Diseases, Central Clinical School, Monash University, Melbourne, Victoria, Australia, 4 School of Public Health and Preventive Medicine, Monash University, Melbourne, Victoria, Australia, 5 Malignant Haematology and Stem Cell Transplantation Service, Alfred Hospital, Melbourne, Victoria, Australia, 6 Department of Microbiology, Monash University, Clayton, Victoria, Australia Key words: hematopoietic stem cell transplantation; epidemiology; infection; microbial drug resistance; bacteria; Candida Correspondence to: Anton Y. Peleg, Department of Microbiology, School of Biomedical Sciences, Monash University, Building 76, Wellington Road, Clayton, VIC 38, Australia Tel: Fax: anton.peleg@monash.edu Received 18 January 214, revised 3 June 214, accepted for publication 12 July 214 DOI: /tid Transpl Infect Dis 214: 16: The last decade has seen major changes in hematopoietic stem cell transplantation (HSCT). Despite the introduction of reduced-intensity conditioning and a number of supportive care measures, infectious complications remain a significant contributor to morbidity and mortality (1). Over time, many strategies have been introduced to prevent infections, including the introduction of fluoroquinolone prophylaxis in those with prolonged neutropenia (2, 3). However, such strategies may have untoward consequences such as the emergence of resistance to fluoroquinolones and other key antibiotic classes, which may have an impact on firstline empiric therapy for febrile neutropenia. Increasing antimicrobial resistance has been associated with increasing mortality (4), and evidence has emerged that receipt of appropriate empiric antibiotics, defined as an antibiotic that has in vitro activity against the implicated organism in the first 48 h, is important for improved patient outcomes (5). It is therefore vital that we evaluate the microbiology of patients post HSCT, so that empiric prescribing for critically unwell and severely immunocompromised patients is optimized. 887

2 The objectives of this study were to determine the microbial epidemiology and trends in susceptibility of organisms cultured from patients post HSCT over a 9- year period at a major Australian HSCT center. We also aimed to assess the independent predictors of mortality for bloodstream infection (BSI) in this patient population. Patients and methods Patients and data collection We reviewed data from the Malignant Haematology and Stem Cell Transplantation Service at the Alfred Hospital, Melbourne, Australia. The service performs approximately 6 myeloablative, reduced-intensity, allogeneic, and autologous HSCTs per year primarily for hematologic malignancies in adults and is one of 2 referral services for the Australian state of Victoria. We retrospectively evaluated all adult patients undergoing HSCT admitted to the unit using a computerized database that contained demographic, clinical, and microbiological data from April 21, 21 to May 24, 21. All clinical isolates that underwent antibiotic susceptibility testing during this time period were included, including all bacterial organisms and Candida species from blood. Definitions Duplicate isolates, defined as a positive culture result of the same organism with the same antibiogram within 14 days of the initial isolate, were excluded. Coagulasenegative staphylococci (CoNS), Corynebacteria, and Lactobacilli were considered contaminants and were excluded unless 2 separate positive cultures occurred with the same antimicrobial susceptibility profile within 48 h (6), where only the first isolate was included in the analysis. Species identification and antimicrobial susceptibility testing of clinical isolates were performed using an automated system (VITEK 2; BioMerieux, Marcy l Etoile, France), and susceptibility thresholds followed the British Society for Antimicrobial Chemotherapy guidelines (7). Significant bacteriuria was defined as 1 colony-forming units 91 6 /L. Growth from intravascular catheter tips was considered to be significant in all cases with the exception of CoNS, where >15 colonies were required for significance. Organisms were defined as being susceptible or resistant. Intermediate susceptibility was categorized as resistant. For gram-positive pathogens, multidrug resistance (MDR) classification was based on resistance to specific antibiotics, including flucloxacillin (Staphylococcus aureus), vancomycin (Enterococcus species), and at least 1 of penicillin or ceftriaxone (Streptococcus pneumoniae) (8). MDR in gram-negative bacteria was defined as being non-susceptible to at least 3 different classes of antimicrobial agents to which the organism would typically be susceptible (8, 9). Seven antibiotic classes were assessed: (i) penicillins or first/second generation cephalosporins, (ii) extended-spectrum penicillin with a b-lactamase inhibitor, (iii) extendedspectrum cephalosporins, (iv) carbapenems, (v) trimethoprim (TMP) or trimethoprim-sulfamethoxazole (TMP-SMX), (vi) aminoglycosides, and (vii) quinolones. The presence of an extended-spectrum beta-lactamase was suggested through the detection of ceftriaxone resistance and the analysis of mean inhibitory concentration patterns with software used by the Advanced Expert System of the VITEK-2 (BioMerieux), then confirmed by Jarlier double-disc synergy test. For the analysis of trends in antimicrobial susceptibility, the study period was divided into 3 equal periods (April 21 May 24, May 24 May 27, and May 27 May 21). Although the study was not designed to assess attributable mortality, crude mortality was assessed at 7 and 28 days after a positive blood culture isolate. Acute graft-versus-host disease (GVHD) was defined according to the Glucksberg grade (1) and was categorized into grade 4 acute GVHD versus other grades of acute GVHD. Neutropenia was defined as an absolute neutrophil count < /L and duration of neutropenia was divided into 14 days and >14 days. Transplantation procedures and management of infections Transplantation was performed according to institutional protocols. Patients were admitted to high-efficiency particulate air-filtered rooms. Patients known to be carriers of vancomycin-resistant Enterococcus (VRE) or carbapenem-resistant gram-negative organisms were isolated in single rooms with contact precautions. No isolation precautions were used for methicillin-resistant S. aureus (MRSA). Ciprofloxacin 5 mg twice a day was used as antibacterial prophylaxis from the start of conditioning therapy until neutrophil engraftment (i.e., absolute neutrophil count of > /L) in all HSCT recipients from August 25 onwards. Patients received antifungal prophylaxis from day 1 until at least day 1. Fluconazole or itraconazole was used initially for all 888 Transplant Infectious Disease 214: 16:

3 HSCT, with voriconazole becoming the agent of choice for allogeneic myeloablative HSCT recipients and fluconazole for autologous HSCT recipients, respectively, from January 28. Allogeneic HSCT recipients were also administered TMP-SMX 16/8 mg 3 times per week as prophylaxis against Pneumocystis jirovecii. Piperacillin-tazobactam 4.5 g q 6-hourly was used as first-line empiric therapy for febrile neutropenia throughout the study duration, with gentamicin being added in the case of septic shock. Vancomycin was also used as part of first-line empiric therapy in the case of prior MRSA colonization, suspected central venous catheter infection, or if the patient was clinically unwell. Patients with anaphylaxis to beta-lactam agents were administered vancomycin and aztreonam. Statistical analysis Statistical analyses were performed using Stata 12.1 (Stata Corp., College Station, Texas, USA). Temporal trends were assessed using the chi-squared test for trend. Comparison of categorical variables was performed using the chi-squared or Fisher s exact test as appropriate. Continuous variables were compared using the Student s t-test or the Wilcoxon rank-sum test for non-parametric distributions. A P-value of.5 was considered statistically significant. To identify independent predictors of mortality at 7 and 28 days after BSI, logistic regression was performed. Potential covariates influencing mortality that were considered included age, primary hematologic diagnosis, disease status at time of HSCT, type of transplant performed, conditioning regimen, acute GVHD, duration of neutropenia, and BSI with an MDR organism. Primary hematologic diagnosis was divided into acute leukemia or other diagnoses. Disease status at time of HSCT was categorized into active disease versus complete remission after 1 or more chemotherapy regimens. Multivariable models were built in a forwards stepwise fashion using variables from the univariate analysis (P <.1), considering biological plausibility and possible confounders. As a measure of the strength of association, the odds ratio (OR) with 95% confidence intervals (95% CI) was estimated. Results Patient characteristics From April 21 to May 21, 528 HSCTs were performed in 58 patients (Table 1). Overall, 46% of patients underwent allogeneic HSCT, with the majority of these being myeloablative (62%). Acute leukemia (n = 172, 33%), multiple myeloma (n = 152, 29%), and lymphoma (n = 15, 28%) were the most common indications for transplantation. Clinical cultures and susceptibility trends A total of 65 isolates were cultured from HSCT patients and underwent antimicrobial susceptibility testing during the study period; 586 (97%) were bacterial isolates, and 19 (3%) were Candida species. The most common site of a positive culture was from blood (n = 38, 63%), followed by sputum (n = 79, 13%), urine (n = 6, 1%), wounds and superficial sites (n = 45, 8%), intravascular catheter tips (n = 14, 2%), bronchoalveolar lavage (n = 7, 1%), swabs from unspecified sites (n = 14, 3%), stool (n = 1,.2%), and unknown (n = 2,.3%). As shown in Figure 1, we observed a decreasing trend in the number of positive cultures over the study period, despite an increasing number of transplants. No change was seen in the ratio of gram-positive to gram-negative organisms over time (Fig. 1). Among gram-positive isolates, S. aureus was the most common species overall (n = 17, 34%), followed by CoNS (n = 94, 3%) and Enterococcus species (n = 6, 19%). In the most recent study period (27 21), CoNS became the most common gram-positive isolate, and a significant increase in Enterococcus species compared to the first study period (21 24) was also observed (P =.4; Fig. 2). Importantly, this increase in Enterococcus was dominated by Enterococcus faecium and was associated with a corresponding increase in vancomycin resistance (8% increased to 64%, P <.1; Fig. 3). MRSA accounted for 54% of S. aureus isolates and rates remained constant during the study (all remained susceptible to vancomycin). Of the CoNS isolates, 94% were methicillin resistant. The epidemiology of gram-negative isolates remained constant throughout the observation period, with Escherichia coli (28%), Klebsiella species (19%), Pseudomonas species (17%), Enterobacter species (12%), Acinetobacter species (6%), and Stenotrophomonas maltophilia (6%) being the most common organisms. However, proportions of susceptible organisms changed significantly (Fig. 4). Increasing resistance was noted in compared to for the following classes: fluoroquinolones (13% up to 36%, P =.1), extended-spectrum beta lactams (ESBLs) (e.g., piperacillin-tazobactam or ticarcillin-clavulanic acid; 32% to 51%, P =.38), and carbapenems (14% to 22%, P =.223), and there were Transplant Infectious Disease 214: 16:

4 Patient and transplant characteristics Variable Total (n = 528) Table 1 Continued Variable Total (n = 528) Median age, years (range) 5 (17 71) Gender (%) Male 314 (6) Primary diagnosis (%) Acute myeloid leukemia 132 (25) Acute lymphoblastic leukemia 4 (7) Multiple myeloma 152 (29) Non-Hodgkins lymphoma 113 (21) Hodgkins lymphoma 37 (7) Chronic myeloid leukemia 15 (3) Chronic lymphocytic leukemia 9 (2) Other hematological disorder 19 (4) Other non-hematological disorder 11 (2) Cyclosporine 8 (15) Cyclosporine/Mycophenolate mofetil 67 (13) Other 17 (3) Acute GVHD (%) None 89 (38) I 48 (21) II 36 (16) III 32 (14) IV 27 (12) Chronic GVHD (%) Nil 53 (34) Limited 19 (12) Extensive 82 (53) Disease state at time of HSCT (%) Complete remission 12 (23) Overall mortality (of patients transplanted during study period) (%) 267 (51) Complete remission 2nd 73 (14) Active underlying disease 311 (59) Not applicable 16 (3) HSCT, hematopoietic stem cell transplant; TBI, total body irradiation; ATG, anti-thymocyte globulin; GVHD, graft-versus-host disease; LACE, lomustine, ARA-C, cyclophosphamide, etoposide regimen. Transplant type (%) Table 1 Allogeneic 244 (46) Myeloblative 152 (29) 3 25 Gram nega ve Reduced intensity 92 (17) Autologous 281 (53) 25 2 Gram posi ve Transplants Syngeneic 3 (1) Conditioning regimens (%) Myeloablative Cyclophosphamide/TBI 56 (37) Total number of isolates Total no. of transplants Cyclophosphamide/TBI/ATG 45 (3) Other allogeneic regimen 51 (34) 5 5 Reduced intensity Fludarabine/Melphalan/ATG 44 (48) Fludarabine/Melphalan/Alemtuzumab 23 (25) Fludarabine/TBI 22 (24) Other reduced-intensity regimen 3 (3) Autologous Melphalan 142 (51) LACE 15 (38) Other autologous regimen 34 (12) GVHD prophylaxis (%) Not applicable 284 (54) Cyclosporine/Methotrexate 8 (15) Study period Fig. 1. Gram-positive versus gram-negative isolates from any site and total number of transplants. high rates of concurrent resistance (Table 2). The proportion of organisms defined as MDR also increased (11% to 22%, P =.52). The impact of fluoroquinolone use was highlighted by analyzing resistance before and after the introduction of routine fluoroquinolone prophylaxis (August 25). Fluoroquinolone resistance pre-prophylaxis was 16% and increased to 35% following prophylaxis (P =.1). Similarly, ESBL resistance increased from 31% to 5% (P =.11). 89 Transplant Infectious Disease 214: 16:

5 % of gram-posi ve isolates Study period S. aureus CoNS Enterococcus S. viridans Other Fig. 2. Epidemiology of gram-positive isolates from any site. CoNS, coagulase-negative Staphylococcus aureas; S. viridans, Streptococcus viridans. Total number of isolates Study period Percentage of isolates E. faecium E. faecalis Vancomycin resistance Fig. 3. Epidemiology and vancomycin resistance in enterococcal isolates from any site. E., Enterococcus. Percentage of isolates resistant Study period Fluoroquinolones Extended-spectrum beta lactams Cefepime Carbapenems Aminoglycosides Mul -drug resistant Fig. 4. Resistance rates to common antimicrobials among gramnegative organisms from any site. Resistance patterns of fluoroquinolone-resistant and extendedspectrum b-lactam (ESBL)-resistant gram-negative organisms from any site Antibiotic class BSIs No. of isolates resistant (%) Odds ratio 95% CI P-value Fluoroquinolone-resistant organisms ESBLs 29/42 (69) <.1 4th generation cephalosporins 13/23 (57) <.1 Carbapenems 17/42 (41) <.1 Aminoglycosides 28/51 (54) <.1 ESBL-resistant organisms Fluoroquinolones 29/67 (43) <.1 4th generation cephalosporins 17/32 (53) <.1 Carbapenems 23/67 (34) <.1 Aminoglycosides 25/67 (37) <.1 CI, confidence interval. Table 2 Gram-positive organisms caused 225 of 38 cases (59%) of BSI (Table 3). The most frequent gram-positive organisms causing BSI were CoNS (n = 92/38, 24%), Enterococcus species (n = 45/38, 12%), S. aureus (n = 35/38, 9%), and viridans Streptococci (n = 33/ 38, 9%). Of the BSI caused by CoNS, 93% were resistant to methicillin but all were sensitive to vancomycin. Of the 35 cases of S. aureus BSI, 17 (49%) were caused by MRSA. Vancomycin-resistant organisms caused 17 of 45 cases (38%) of enterococcal BSI, and vancomycin-resistant BSI showed a significant increase from 7% (1/15) in to 67% (12/18) in (P =.1). Gram-negative organisms caused 137 of 38 cases (36%) of BSI (Table 3). Trends were seen toward increasing resistance to fluoroquinolones and ESBLs. The highest proportions of resistant isolates were observed from 27 to 21, with 28% (7/25) fluoroquinolone resistance compared with 12% (7/62) isolates in and 36% (9/25) ESBL resistance compared with 26% (16/62) isolates in Carbapenem resistance remained unchanged with 8% (2/25) in and 8% (5/62) in respectively. The number of positive blood cultures decreased from 161 in (88/161, 55% grampositive) to 118 in (68/118, 58% gram-positive), and to 11 in (69/11, 68% gram-positive). Transplant Infectious Disease 214: 16:

6 Causative organisms for bloodstream infections Organism n = 38 (%) Coagulase-negative Staphylococci 92 (24) Enterococcus species 45 (12) Escherichia coli 43 (11) Staphylococcus aureus 35 (9) Klebsiella species 35 (9) Viridans Streptococci 33 (9) Candida species 18 (5) Enterobacter species 18 (5) Pseudomonas species 12 (3) Acinetobacter species 1 (3) Stenotrophomonas maltophilia 7 (2) Other gram-positive organisms 2 (5) Other gram-negative organisms 12 (3) Table 3 Candidemia constituted 5% of BSI (n = 18), with Candida glabrata (n = 1, 53%) predominating over Candida albicans (n = 3, 16%). A significant decrease was noted in candidemia from 7 isolates in to 1 isolate in (P =.2); 61% of isolates (11/18) were sensitive to all tested antifungals, and 28% (5/18) were resistant to fluconazole, with 4 C. glabrata and 1 C. krusei isolate. Transplant type and microbiology Myeloablative allografts had similar numbers of positive clinical cultures in all 3 phases after transplantation (4% occurred day 7 to day 29 post HSCT, 28% occurred day 3 to day 99, and 32% occurred after day 1). Compared to other transplant types, reducedintensity allografts had a significantly higher number of isolates occurring >1 days post HSCT (71 of 148 isolates, 48%; P <.1), whereas autologous transplants had a significantly higher number of isolates occurring day 7 to day 29 post HSCT (135/27, 65%; P <.1). Mortality from bacteremia Overall, crude mortality at 7 days after a positive blood culture was 12% (25/212 patients), with E. faecium being the most commonly isolated organism associated with mortality (6/25, 24%; Table 4). Of the 25 patients Organisms associated with mortality at 7 days after positive blood culture Organism n = 27 (%) Enterococcus faecium 6 (22) Staphylococcus aureus 3 (11) Acinetobacter calcoaceticus-baumanii complex 2 (7) Candida glabrata 2 (7) Escherichia coli 2 (7) Klebsiella pneumoniae 2 (7) Pseudomonas aeruginosa 2 (7) Stenotrophomonas maltophilia 2 (7) Candida albicans 1 (4) Citrobacter freundii 1 (4) Diphtheroid species 1 (4) Enterobacter cloacae 1 (4) Coagulase-negative Staphylococci 1 (4) Streptococcus dysgalactiae 1 (4) Table 4 who died within 7 days after a positive blood culture, 11 were infected with organisms that would not have been covered by the standard empiric therapy for febrile neutropenia. Active disease at the time of transplant (OR 3.75, 95% CI , P =.23) and BSI with an MDR organism (OR 3.61, 95% CI , P =.8) were independent predictors of death at 7 days using multivariate analysis (Table 5). Crude mortality at 28 days after a positive blood culture was 25% (53/212), with CoNS being the most common organism associated with mortality (14/53, 26%). Multivariate analysis showed that active disease at time of transplant (OR 2.48, 95% CI , P =.2), Grade 4 acute GVHD (OR 4.31, 95% CI , P =.6), and BSI with an MDR organism (OR 3.89, 95% CI , P =.1) remained independent predictors. Discussion The emergence of MDR organisms has been identified as one of the greatest challenges in the treatment of infectious diseases (11). This growing threat is of particular concern in HSCT recipients owing to their profound immunosuppression. We conducted a retrospective cohort study examining the epidemiology and susceptibility patterns of organisms cultured from HSCT recipients, including both BSIs and cultures 892 Transplant Infectious Disease 214: 16:

7 Univariate and multivariate analysis of factors associated with mortality at 7 and 28 days after bloodstream infection (BSI) Variable Total no. of patients with BSI n = 212 Death at 7 days following BSI (%) n = 25 Univariate analysis Multivariate analysis Death at Univariate analysis Multivariate analysis 28 days following BSI (%) OR (95% CI) P-value OR (95% CI) P-value n = 53 OR (95% CI) P-value OR (95% CI) P-value Age (tertile) (12.5) 17 (26.6) (1.4).81 ( ) (22.1).78 ( ) (12.7) 1. (.61.68) (26.8) 1. ( ).98 Acute leukemia as indication for HSCT 92 1 (1.9).85 ( ) (29.4) 1.5 (.8 2.8).2 Active disease/relapse at time of HSCT (16.5) 3.81 ( ) ( ) (31.4) 2.39 ( ) ( ).2 Reduced-intensity allograft 4 8 (2) 2.28 ( ) ( ) ( ).23 ATG-containing conditioning regimen (14.5) 1.51 ( ) ( ) ( ).84 Grade 4 acute GVHD 21 5 (23.8) 2.67 ( ) ( ) ( ) < ( ).6 Duration of neutropenia >14 days (1.8).82 ( ) ( ).87 MDR organism 4 1 (25) 3.49 ( ) ( ) ( ) < ( ).1 OR, odds ratio; CI, confidence interval; HSCT, hematopoietic stem cell transplant; ATG, anti-thymocyte globulin; GVHD, graft-versus-host disease; MDR, multidrug-resistant. Table 5 Transplant Infectious Disease 214: 16:

8 from other sites. In addition to myeloablative and autologous transplants, this study included a large number of reduced-intensity allogeneic transplants, thus reflecting current transplant practice. During the 9-year study period, we observed an increasing proportion of non-susceptible organisms, in particular VRE and fluoroquinolone-resistant gram-negative organisms. BSI with an MDR organism was an independent predictor of mortality at both 7 and 28 days. Fluoroquinolone prophylaxis in neutropenic patients has been shown to decrease fever and the number of microbiologically documented infections, but has not resulted in a clear mortality benefit in randomized controlled trials (2, 3). A meta-analysis suggested that antibiotic prophylaxis significantly reduced the risk of death from all causes and the risk of infection-related death, but did not lead to increased resistance in patients that received fluoroquinolone prophylaxis (12). Thus far, the long-term impact of fluroquinolone prophylaxis on microbiological epidemiology in HSCT has been poorly documented. In our study, rates of fluoroquinolone resistance in gram-negative organisms increased from 16% to 35% following institution of fluoroquinolone prophylaxis (P =.1). This rise in fluoroquinolone resistance was accompanied by a decrease in the total number of isolates cultured (including bloodstream isolates), implying that fluoroquinolone prophylaxis may have prevented infections at the cost of selecting for infections with increasingly resistant organisms. These findings also raise the concern that over time, the prevalence of fluoroquinolone-resistant organisms may rise, leading to a return of infection numbers to pre-fluoroquinolone levels but with greater resistance. Furthermore, fluoroquinolone use may drive other resistance mechanisms such as multidrug efflux pumps and ESBLs (13). In fact, in our data, we observed an increase in resistance to ESBLs post initiation of fluoroquinolone prophylaxis (from 31% to 5% [P =.11]). As a result of the demonstrated lack of mortality benefit and other results described in this study, routine fluoroquinolone prophylaxis is now limited to myeloablative allografts only and may be reduced further once mortality rates with restricted use are assessed. Data are limited regarding trends in susceptibility to other key gram-negative antimicrobial classes in this patient population (14). Our study noted increasing resistance to ESBLs and carbapenems, and increasing numbers of gram-negative MDR organisms. The concerns regarding infection with MDR organisms in febrile neutropenia have led to discussion of a strategy of de-escalation in Europe. According to this strategy, empiric therapy for MDR organisms is used first-line, with a carbapenem and vancomycin, and subsequently de-escalated according to culture results (15). Although this approach may be warranted in certain cases, the potential exists for antibiotic overuse both in terms of the choice of antibiotic and duration of therapy, and this issue has been addressed by the recently published guideline from the European Conference on Infections in Leukemia Group, which now recommends that escalation should be the standard approach in cases that do not have a severe presentation (16). Future empiric therapy for febrile neutropenia may require a more tailored approach through risk stratification of patients to allow appropriately broad antimicrobial therapy for high-risk patients, such as those with previous colonization with resistant isolates, while limiting collateral damage in the form of antimicrobial resistance. This goal could be achieved with the use of clinical risk algorithms and novel rapid diagnostic techniques such as matrix-assisted laser desorption/ ionization time-of-flight (MALDI-TOF) mass spectrometry to allow early detection of colonization and infection with resistant organisms (17). Enterococcus is now recognized as an increasingly important pathogen in HSCT recipients (18 2) and trends toward increasing vancomycin resistance have been noted previously (21). This change in grampositive epidemiology was also a key finding in our study with significant increases in numbers of enterococcal isolates and proportions of VRE. Multiple possible contributors to this evolution include an increase in enterococcal colonization (22) and widespread antibiotic use (21). Enterococcus was also associated with poor outcomes in our study: E. faecium featured as the organism most commonly associated with mortality at 7 days after BSI and the second most commonly associated with mortality at 28 days after BSI. The role of enterococcal infection in contributing to mortality is controversial, as it may be associated with severe underlying disease rather than being a direct cause of mortality (23). Nonetheless, the difficulties in treating this organism raise concerns about nosocomial acquisition and underscore the need for antimicrobial stewardship programs and effective infection prevention efforts in this patient population. With the findings of our study, our protocol for febrile neutropenia has been updated to include the addition of teicoplanin or daptomycin in addition to standard therapy in the case of previous VRE colonization. Our multivariate analysis indicated that active disease at time of transplant and BSI with an MDR organism were associated with death at 7 days after BSI. Both of these factors were associated with death at 28 days after BSI, as was Grade 4 acute GVHD. These findings 894 Transplant Infectious Disease 214: 16:

9 logically link the importance of both host and microbial factors in outcomes of BSI in HSCT patients. Active disease at the time of transplantation and acute GVHD have adverse impacts on host immunity, thus contributing to poor outcomes after infective complications. The association between BSI with an MDR organism and mortality may result from inactive empirical regimens for febrile neutropenia and the difficulties of treating resistant organisms. Although it is also possible that unmeasured confounding may account for the apparent association between BSI and outcome, the strength of association, biological plausibility, and consistency with other studies suggest that this may be a causal link. This finding adds to previous concerns regarding the negative clinical implications of resistant organisms (21, 24, 25) and suggests a need for close monitoring of evolving bacterial epidemiology. Candidemia only constituted 5% of BSI and decreased significantly over the course of the study (P =.2). Of interest, C. glabrata emerged as the most common cause of candidemia, in keeping with higher fluconazole mean inhibitory concentration in this species and its possible selection through the use of azole antifungal prophylaxis in our patient population. Slavin et al. (26) have noted that receipt of triazoles was the strongest independent predictor for fluconazole-resistant candidemia in adult cancer patients. Our study has several limitations. It was a singlecenter study and therefore the results may not be generalizable to other institutions. The distribution of transplant types may be different compared to that of other centers, and newer transplant practices such as cord blood HSCT were under-represented in this study and would likely impact upon infectious complications. The retrospective nature of the study imposed limitations on data collection regarding use of antibiotics, clinical findings, and detailed outcomes. Future prospective studies are warranted to confirm the links between antibiotic use, resistant organisms, and patient outcomes in this highly vulnerable patient population. In conclusion, our study has shown important changes in both epidemiology and susceptibility profiles of isolates after HSCT over a 9-year period. In particular, it raises concerns regarding increasing gram-negative resistance to fluoroquinolones, ESBLs, and carbapenems, in the setting of fluoroquinolone prophylaxis. Furthermore, Enterococcus has emerged as an increasingly important pathogen with high degrees of vancomycin resistance. In multivariate analysis, active disease at the time of transplantation, and BSI with an MDR organism were independently associated with mortality at 7 and 28 days after BSI. These findings have important implications for the future efficacy of fluoroquinolone prophylaxis, and highlight the importance of local epidemiologic knowledge to help guide approaches to empirical therapy for febrile neutropenia. Further prospective multicenter studies are needed to determine the threshold of fluoroquinolone resistance where the benefit of prophylaxis is lost, and the impact of such prophylaxis on other gram-negative resistance mechanisms. An approach that stratifies risk of empiric therapy for febrile neutropenia on the basis of local epidemiology, previous colonization with resistant organisms, and severity of clinical presentation, may lead to more appropriate antibiotic use, better clinical outcomes, and a decrease in the emergence of antimicrobial resistance. Acknowledgements: Thanks: We are grateful to Kerrie Watson and Jenny Muirhead for data support. A.Y.P was supported by an Australian National Health and Medical Research Council Career Development Fellowship (APP147916). Author contributions: N.M. and C.O.M.: Concept, design, data collection, drafting and revision of the article. A.C.C.: Concept, design, statistical analysis, drafting and revision of the article. A.S.: Concept, design, and data collection. A.Y.P.: Concept, design, data collection, statistical analysis, drafting and revision of the article. Conflicts of interest: The authors declare no conflict of interest. References 1. Gooley TA, Chien JW, Pergam SA, et al. Reduced mortality after allogeneic hematopoietic-cell transplantation. N Engl J Med 21; 363 (22): Bucaneve G, Micozzi A, Menichetti F, et al. Levofloxacin to prevent bacterial infection in patients with cancer and neutropenia. N Engl J Med 25; 353 (1): Cullen M, Steven N, Billingham L, et al. Antibacterial prophylaxis after chemotherapy for solid tumors and lymphomas. N Engl J Med 25; 353 (1): Giske CG, Monnet DL, Cars O, Carmeli Y. Clinical and economic impact of common multidrug-resistant gram-negative bacilli. Antimicrob Agents Chemother 28; 52 (3): Retamar P, Portillo MM, Lopez-Prieto MD, et al. Impact of inadequate empirical therapy on the mortality of patients with bloodstream infections: a propensity score-based analysis. Antimicrob Agents Chemother 211; 56 (1): O Grady NP, Alexander M, Burns LA, et al. Guidelines for the prevention of intravascular catheter-related infections. Clin Infect Dis 211; 52 (9): e162 e193. Transplant Infectious Disease 214: 16:

10 7. Andrews JM, for the BSAC Working Party on Susceptibility Testing. BSAC standardized disc susceptibility testing method (version 8). J Antimicrob Chemother 29; 64 (3): Magiorakos AP, Srinivasan A, Carey RB, et al. Multidrugresistant, extensively drug-resistant and pandrug-resistant bacteria: an international expert proposal for interim standard definitions for acquired resistance. Clin Microbiol Infect 211; 18 (3): Pop-Vicas AE, D Agata EMC. The rising influx of multidrugresistant gram-negative bacilli into a tertiary care hospital. Clin Infect Dis 25; 4 (12): Glucksberg H, Storb R, Fefer A, et al. Clinical manifestations of graft-versus-host disease in human recipients of marrow from HL-A-matched sibling donors. Transplantation 1974; 18: Spellberg B, Bartlett JG, Gilbert DN. The future of antibiotics and resistance. N Engl J Med 213; 368 (4): Gafter-Gvili A, Fraser A, Paul M, et al. Antibiotic prophylaxis for bacterial infections in afebrile neutropenic patients following chemotherapy. Cochrane Database Syst Rev 212; 1: CD Rice LB. Mechanisms of resistance and clinical relevance of resistance to b-lactams, glycopeptides, and fluoroquinolones. Mayo Clin Proc 212; 87 (2): Irfan S, Idrees F, Mehraj V, Habib F, Adil S, Hasan R. Emergence of carbapenem resistant Gram negative and vancomycin resistant Gram positive organisms in bacteremic isolates of febrile neutropenic patients: a descriptive study. BMC Infect Dis 28; 8 (1): European Conference on Infections in Leukemia. Bacterial Resistance in Haematology ECIL 4 Study Groups & Participants. Available from: Resources/Library/ECIL/Documents/ECIL4%2211% 2Bacterial%2resistance%2in%2Haematology.pdf (accessed on 21 May, 213). 16. Averbuch D, Orasch C, Cordonnier C, et al. European guidelines for empirical antibacterial therapy for febrile neutropenic patients in the era of growing resistance: summary of the 211 4th European Conference on Infections in Leukemia. Haematologica 213; 98 (12): Diekema DJ, Pfaller MA. Rapid detection of antibiotic-resistant organism carriage for infection prevention. Clin Infect Dis 213; 56 (11): Almyroudis NG, Fuller A, Jakubowski A, et al. Pre- and postengraftment bloodstream infection rates and associated mortality in allogeneic hematopoietic stem cell transplant recipients. Transpl Infect Dis 25; 7 (1): Cappellano P, Viscoli C, Bruzzi P, Van Lint MT, Pereira CAP, Bacigalupo A. Epidemiology and risk factors for bloodstream infections after allogeneic hematopoietic stem cell transplantation. New Microbiol 27; 3 (2): Kamboj M, Chung D, Seo SK, et al. The changing epidemiology of vancomycin-resistant Enterococcus (VRE) bacteremia in allogeneic hematopoietic stem cell transplant (HSCT) recipients. Biol Blood Marrow Transplant 21; 16 (11): Mikulska M, Del Bono V, Prinapori R, et al. Risk factors for enterococcal bacteremia in allogeneic hematopoietic stem cell transplant recipients. Transpl Infect Dis 21; 12 (6): Weinstock DM, Conlon M, Iovino C, et al. Colonization, bloodstream infection, and mortality caused by vancomycinresistant Enterococcus early after allogeneic hematopoietic stem cell transplant. Biol Blood Marrow Transplant 27; 13 (5): Dubberke ER, Hollands JM, Georgantopoulos P, et al. Vancomycin-resistant enterococcal bloodstream infections on a hematopoietic stem cell transplant unit: are the sick getting sicker? Bone Marrow Transplant 26; 38 (12): Peralta G, Sanchez MB, Garrido JC, et al. Impact of antibiotic resistance and of adequate empirical antibiotic treatment in the prognosis of patients with Escherichia coli bacteraemia. J Antimicrob Chemother 27; 6: Livermore DM. Bacterial resistance: origins, epidemiology, and impact. Clin Infect Dis 23; 36 (Suppl): S11 S Slavin MA, Sorrell TC, Marriott D, et al. Candidaemia in adult cancer patients: risks for fluconazole-resistant isolates and death. J Antimicrob Chemother 21; 65 (5): Transplant Infectious Disease 214: 16: