A Randomized Comparison of Empiric Or Pre-Emptive Antibiotic Therapy After Hematopoietic Stem Cell Transplantation

Home , Antibiotic prophylaxis, Empiric therapy

ORIGINAL ARTICLE A randomized comparison of empiric or pre-emptive antibiotic therapy after hematopoietic stem cell transplantation

MA Slavin1, AP Grigg2, AP Schwarer3, J Szer2, A Spencer3, A Sainani2, KA Thursky1 and AW Roberts2

1Victorian Infectious Diseases Service and Clinical Research Centre of Excellence in Infectious Diseases, Royal Melbourne Hospital, Melbourne, Victoria, Australia; 2Department of Clinical Haematology and Bone Marrow Transplant Service, Royal Melbourne Hospital, Melbourne, Victoria, Australia and 3Bone Marrow Transplant Programme, Alfred Hospital, Melbourne, Victoria, Australia

We performed a randomized comparison of pre-emptive fever in the neutropenic patient.4 Despite empiric anti- and empiric antibiotic therapy for adult patients under- bacterial therapy, mortality ranging up to 10% is asso- going allogeneic or autologous stem cell transplantation. ciated with fever and neutropenia, particularly in those One hundred and fifty-three patients were randomized to with an expected duration of neutropenia (ANC receive cefepime either pre-emptively on the day that o0.5 Â 109/l) 410 days.5 Inadequate or delayed therapy neutropenia (ANCo1.0 Â 109 cells/l) developed irrespec- of the initial pathogen has been associated with increased tive of the presence of fever, or at onset of fever and mortality6 as has been the presence of underlying comor- neutropenia (empiric). Although there was no difference bidities.5 between the two arms in the proportion of patients Prophylaxis with oral levofloxacin decreased blood- developing fever or in the median number of days of fever, stream infections (BSIs) in one randomized placebo- the time to onset of fever was a mean of 1 day longer in controlled study, although patients were exclusively each patient on the pre-emptive arm (log rank Po0.001). undergoing autologous transplantation, and most were The number of patients with bloodstream infections was receiving therapy for solid tumours rather than leukaemia significantly reduced in those receiving pre-emptive or lymphoma.7 The use of prophylactic oral fluoroquino- therapy (16/75) compared to the empiric arm (31/76) lones remains controversial8–11 and the effects of oral (Po0.01) but this did not translate into an appreciable antibiotics used prophylactically may not equate to clinical benefit as measured by days of hospitalization, intravenous agents.12 We and others have hypothesized time to engraftment, use of additional antimicrobial that pre-emptive intravenous antibiotic therapy, that is, agents or mortality at 30 days. This study does not antibiotics commenced at onset of neutropenia or on day of support the use of pre-emptive intravenous antibiotic transplant (also referred to as prophylactic antibiotics in therapy in adult stem cell transplant recipients. some recent publications), would reduce the incidence of Bone Marrow Transplantation (2007) 40, 157–163; fever and bacterial infection during the acute phase of doi:10.1038/sj.bmt.1705686; published online 30 April 2007 transplantation.2,13,14 However, pre-emptive intravenous Keywords: empiric; pre-emptive; antibiotics; SCT; antibiotic therapy has not been well evaluated in HSCT neutropenia although two non-randomized studies have suggested that patients may benefit from this approach.13,14 The present study was designed to compare the occurrence of fever, bacterial infection, antibiotic usage and outcome of adult patients undergoing either allogeneic Introduction or autologous stem cell transplant in patients randomized to receive either empiric antibiotic therapy, started at onset Fever occurs in 490% of allogeneic hematopoietic stem of fever, or pre-emptive antibiotic therapy, started at onset cell transplant (HSCT)1,2 and 80% of autologous HSCT of significant neutropenia. recipients.3 Empiric antibacterial regimens such as b-lactam or carbapenem monotherapy or a combination of beta- lactam and an aminoglycoside are recommended to treat Patients and methods

The study was conducted at two adult bone marrow Correspondence: Dr MA Slavin, Victorian Infectious Diseases Service transplant centres in Melbourne, Australia. All patients and Clinical Research Centre of Excellence in Infectious Diseases, Royal admitted for HSCT were approached for consent to Melbourne Hospital, Grattan St, Parkville, Melbourne, Victoria 3050, participate. Patients previously randomized, receiving Australia. E-mail: [email protected] systemic antibiotics (except for Pneumocystis jirovecii or Received 30 October 2006; revised 5 March 2007; accepted 13 March post-splenectomy prophylaxis) or with documented beta- 2007; published online 30 April 2007 lactam allergy, fever or infection at the time of admission Pre-emptive versus empiric antibiotics after SCT MA Slavin et al 158 for conditioning for HSCT were excluded. Fluoroquino- engraftment. The use of prophylactic antifungal, antiviral lone prophylaxis was not used in either centre during the and growth factor therapy was according to the unit period of this study. Patients were followed prospectively protocols. Briefly, for allogeneic HSCT recipients, flucona- from time of randomization through to discharge. Deaths zole was used as antifungal prophylaxis at a dose of occurring up to 3 months after transplant were recorded. 400 mg/day from the onset of neutropenia to neutrophil Maximum temperature, additional antimicrobials pre- recovery. All patients seropositive for herpes simplex virus scribed and their doses and all positive microbiology or varicella zoster virus received acyclovir 400 mg orally results were recorded daily. Date of engraftment was twice daily or dose 250 mg intravenously three times daily recorded and time to all events was from date of transplant. from onset of neutropenia until hospital discharge. The research and ethics committees of both hospitals approved the protocol. Severity of sepsis A further analysis was performed on the occurrence and Definitions severity of sepsis syndrome in the cohort of patients treated For the purpose of starting either empiric or pre-emptive at Centre 2. Charts were reviewed by an investigator not antibiotic treatment, and to account for rapidly declining involved in the initial study (AS). At the time of fever, neutrophil counts with conditioning, neutropenia was patients were categorized as to the severity of sepsis defined as ANCo1.0 Â 109/l. Time to engraftment was according to the following scale: grade 1 uneventful, grade defined as time from transplant until the neutrophil count 2 intermediate (one or more of temperature 4391C, rigors, exceeded 0.5 Â 109/l. Fever was an oral temperature of heart rate 4120/min, respiratory rate 420/min, new 38.31C on one occasion, or two temperatures of X381C pulmonary infiltrate on chest radiograph, fluid resuscita- separated by at least 1 h, not associated with infusion of tion o500 ml/h required), grade 3 severe (one or more of blood products. Severe sepsis was defined as fever altered mental status, respiratory rate increase from base- accompanied by a reduction of systolic blood pressure to line greater than 10/min, systolic blood pressure o90 mm o100 mm Hg or a 420 mm Hg decrease from baseline or Hg or decrease 440 mm Hg from baseline, pulse oximetry pulse 4120/min. BSI was any organism detected from at o90% on room air, acidosis (pHo7.36), coagulation least one blood culture. Where more than one species was abnormalities (INR41.5, APTT460 s), rise in serum isolated on a single blood culture, this was considered as creatinine 450% above baseline, rise in bilirubin one infection but designated as polymicrobial. When the 470 mmol/l, fluid resuscitation 4500 ml/h required). same organism was isolated on a sample taken within 14 Admission to intensive care unit (ICU) was also recorded. days of the first positive culture, this was counted as one infection. Standard microbiological methods were used for Transplantation protocols culture and identification of bacteria and fungi. Standard prophylaxis for graft-versus-host disease (GVHD) included the use of cyclosporine and short-course Antimicrobial protocols methotrexate.16 Corticosteroids were not used prophylacti- Pre-emptive therapy with cefepime began on the first day cally. Treatment of moderate to severe GVHD consisted of after randomization when the ANC was o1.0 Â 109/l and optimization of cyclosporin levels and commencement of iv decreasing. The empiric therapy with cefepime began when methylprednisolone 1–2 mg/kg/day or an equivalent oral a patient with a neutrophil count o1.0 Â 109/l developed a prednisolone dose. fever. The dose of cefepime was 1 g intravenously 12 hourly, which was increased to 2 g 12 hourly for severe Statistical analysis sepsis as defined above. The addition of gentamicin, Randomization was on a 1:1 ratio to receive either pre- metronidazole and vancomycin was recommended for emptive or empiric therapy in blocks of 4 with the study severe sepsis. The use of vancomycin was also permitted arm randomly allocated within those blocks and concealed for evidence of central line infection, and metronidazole for before randomization. Stratification was for the centre at diarrhoea or clinical suspicion of neutropenic enterocolitis which the transplant was performed (centre 1 or centre 2) defined by unexpected or severe nausea, vomiting, diar- and stem cell source (allogeneic or autologous). The sample rhoea, abdominal pain or ileus. The addition of empiric size was calculated at 67 patients per arm based on numbers antifungal therapy (amphotericin B at 0.5–1 mg/kg/day) required to detect a change in the incidence of fever from 95 was considered during neutropenia at or after 5 days of to 75% in the pre-emptive antibiotic arm at the 0.05 persistent fever despite antibiotics, or upon development of significance level with 80% power. The aim was to enrol new fever after 48 h without fever. If an infection was 140 patients to account for patient withdrawals and loss to identified, treatment was modified appropriately according follow-up. to antibiotic susceptibility that was performed by disc The primary end points were occurrence of fever and diffusion according to the Clinical and Laboratory days with fever during the transplant admission. Secondary 15 Standards Institute guidelines. Details of all antimicro- end points were number of bacterial BSIs, total number of bials given after transplantation were recorded. systemic antimicrobial agents added (including antiviral and antifungal as well as antibacterial agents), duration of Antifungal and antiviral prophylaxis and treatment hospitalization for transplant and 30-day mortality. All All patients were in-patients in high efficiency particulate patients randomized were analysed (intention to treat) air-filtered rooms from the start of conditioning to unless HSCT did not occur or consent for study was

Bone Marrow Transplantation Pre-emptive versus empiric antibiotics after SCT MA Slavin et al 159 withdrawn. Time was counted from the day of transplant, recipients received peripheral blood stem cells. Patient and expressed as median number of days, with 95% demographics are shown in Table 1. confidence interval (CI). Times to event outcomes were analysed by the Kaplan–Meier method. The Poisson regression approach was also used to model Time to fever and duration of fever the data as Poisson data (skewed distribution, non-negative Despite receiving pre-emptive antibiotics, 66/72 (92%) values and increasing variance as the mean increases) and patients in the pre-emptive arm developed a fever during accounting for the varying times ‘at risk’. Negative their in-patient admission, a rate similar to that observed in binomial regression uses a log transformation that adjusts the empiric arm where 73/75 (97%) patients developed for the skewness and prevents the model from producing fever and required treatment with antibiotics (P ¼ 0.16) negative predicted values. It also models the variance as a (Table 2). Data relating to onset of fever were missing for function of the mean.17 Exposure was determined as the three additional patients in the pre-emptive and one time (in days) from the date of transplant to the date of additional patient in the empiric arm. The median time discharge or date of death, whichever came earlier. The from transplant to the onset of the first fever was 7 days number of days with fever across this time period could (95% CI 7,8) in patients receiving pre-emptive antibiotics then be determined. Days without data simply did not and 6 days (95% CI 5,6 days) in patients receiving empiric contribute to the analysis, but did not cause the patients to antibiotics, as calculated by Kaplan–Meier survival curve be censored. For the analysis of days with fever, the (Po0.001) (Figure 1). The Cox proportional hazards following prognostic factors were also included in the model for time to onset of fever showed that the study model: age, transplant type, time to neutropenia from date arm remained significant (Po0.001) with patients on of transplant, use of total body irradiation (TBI) in the the empiric arm having a hazard ratio of 2.42 (95% CI conditioning regimen and hospital. These terms were included in all models, together with treatment, irrespective of attainment of statistical significance. According to the Table 1 Demographics of 151 patients receiving either pre-emptive statistical goodness of fit statistics, an overdispersed or empiric antibiotic therapy in modified intent to treat analysis Poisson model (negative binomial model) was used for all models except the model for antibiotic exposure, which was Demographic Pre-emptive n Empiric n (%) (%) (N ¼ 75) (N ¼ 76) underdispersed. Estimates of treatment mean rates were derived from the model, adjusted for age, transplant type, Center time to neutropenia (from date of transplant) use of TBI Center 1 40 (53) 40 (53) and treating hospital. Center 2 35 (47) 36 (47) A Cox proportional hazards model was fitted to model Sex for time to onset of fever including risk factors for fever. Male 47 (63) 41 (55) Variables included in the model were hospital, transplant Female 28 (37) 35 (45) type (allogeneic or autologous), use of TBI, age and time to Transplant neutropenia. Autologous 36 (48) 36 (48) For comparison of proportions, Fisher’s exact test was Allogeneic 39 (52) 40 (52) used. Statistical calculations were performed with the use of STATA software (version 8, Stata Corporation, TX, USA). Donor type Matched 31 (41) 29 (38) Mismatched 18 (24) 11 (14)

Results Preparative regimen TBI 21 (28) 25 (33) One hundred and fifty-three patients were randomized over Nonmyeloablative 9 (12) 6 (8) ATG 10 (13) 9 (12) a 26-month period. After randomization, two patients were excluded from the analysis either owing to failure of the Diagnosis transplant to proceed (n ¼ 1) or because consent for the NHL 30 (40) 21 (28) study was withdrawn before any study drug was adminis- AML 7 (9) 15 (20) ALL 6 (8) 4 (5) tered (n ¼ 1). Of the 151 patients included in the modified CML 9 (12) 9 (12) intent to treat analysis, 75 received pre-emptive and 76 Myeloma 10 (13) 9 (12) empiric therapy. Seventy-nine patients (52%) had under- Hodgkin’s lymphoma 5 (7) 6 (8) gone allogeneic HSCT and of these 58% received TBI as Other haematological diseasea 5 (7) 7 (8) b part of the conditioning regimen and 37% received Solid tumour 3 (4) 5 (7) mismatched grafts. Ex vivo T cell depletion was used in Total 75 76 only one allograft although in vivo T cell depletion with ATG (antithymocyte globulin) was used in 19 patients as Abbreviations: AML ¼ acute myeloid leukaemia; ALL ¼ acute lympho- part of the conditioning regimen. Among allogeneic blastic leukaemia; ATG ¼ antithymocyte globulin; CML ¼ chronic myeloid transplant recipients, the stem cell source was peripheral leukaemia; NHL ¼ non-Hodgkin’s lymphoma. aOther haematological disease includes myelodysplastic syndrome, blood in 44 and bone marrow in 35, and both stem cell chronic lymphocytic leukaemia, amyloid, myelofibrosis, aplastic anaemia, sources were distributed evenly between antibiotic treat- Fanconi’s anaemia. ment arms. All except one of the autologous transplant bSolid tumours: includes breast, sarcoma, germ cell tumours.

Bone Marrow Transplantation Pre-emptive versus empiric antibiotics after SCT MA Slavin et al 160 Table 2 Outcome of patients receiving pre-emptive versus empiric antibiotic therapy

End point Pre-emptive antibiotics (N ¼ 75) Empiric antibiotics (N ¼ 76)P-value

Number of patients with fever (%) 66 (92) 74 (97) 0.16 Median days to fever after SCT (range) 7 (1–31) 6 (1–10) 0.001 Median duration in days of cefepime (range) 10.5 (0–36) 7 (0–24) 0.003 Median days in hospital (range) 29.5 (13–104) 29.5 (11–133) 0.5 Median time in days to engraftment (range) 14 (10–29) 15 (9–47) 0.5 Median no. additional antimicrobials (range) 4 (1–13) 5 (1–17) 0.99 Number of patients starting empiric antifungal therapy (%) 26 (34) 42 (55) 0.01 Number of deaths at 30 days after SCT (%) 6 (8) 4 (5.3) 0.53 Number of deaths due to bacterial sepsis (%) 3 (4) 2 (2.6) 0.63

Abbreviation: SCT ¼ stem cell transplantation.

1.00 The median duration of fever did not differ between patients receiving pre-emptive or empiric antibiotics at 6 (95% CI 5,7) and 8 (95% CI 6,10) days, respectively (range 0.75 0–35 days) (P ¼ 0.51).

0.50 Addition of other antimicrobials Of the patients undergoing allogeneic transplant, 18/39 (46%) in the pre-emptive and 22/40 (55%) in empiric arms 0.25

Survival distribution function received empiric antifungal treatment (P ¼ 0.50) (Table 2). Of the patients undergoing autologous HSCT, 8/36 (22%)

0.00 from the pre-emptive and 20/36 (55%) patients from the 0 5 10 15 20 25 30 35 empiric arms received empiric antifungal therapy Days with fever (P ¼ 0.007). There was no significant difference in the STRATA: Tmt=Empiric antibiotics Censored tmt=Empiric antibiotics median number of antimicrobial agents added after starting Tmt=Pre-emptive antibiotics Censored tmt=Pre-emptive antibiotics cefepime during transplant hospitalization in either arm, Figure 1 Kaplan–Meier survival curve showing increased median time being 4 in the pre-emptive and 5 in the empiric arm from transplant to the onset of the first fever in the pre-emptive and empiric groups (7 versus 6 days, Po0.001). (P ¼ 0.99). Antibiotic utilization (measured by expressing the daily dose of each antibiotic administered as a fraction of the accepted daily treatment dose for each day of treatment after stem cell transplantation (SCT) until 1.68–3.49) for developing fever at any given time point discharge or death) when compared between the two from transplant compared to those on the pre-emptive arm. groups was not different at a mean rate of antibiotic use Fever may be caused by noninfective inflammatory of 3.38 standard antibiotic treatments/day in the pre- processes, particularly post-engraftment in patients under- emptive and 4.02/day in the empiric treatment groups going allogeneic transplantation. As pre-emptive antibio- (P ¼ 0.64). tics were hypothesized to reduce fever related to infections during the neutropenic phase of transplantation, we performed a post hoc analysis of the incidence of any fever Bacteremias from day of infusion to day of engraftment. Nine patients Bacterial BSIs occurred in 16 patients (with 18 organisms) receiving pre-emptive antibiotics and five in the empiric in the pre-emptive arm and in 31 patients (with 38 isolates) arm developed their first fever after engraftment, and were in the empiric arm (Po0.01) (Table 3). The negative excluded from this exploratory analysis. Before engraft- binomial regression demonstrated a significant difference ment, the development of fever was significantly less between the two arms that was independent of transplant frequent in the pre-emptive arm (57/69, 83%) than in the type, treating hospital, age and presence of TBI and time empiric arm (68/71, 96%; P ¼ 0.018). to neutropenia. Both Gram-positive and Gram-negative The mean number of days with fever was not different isolates were reduced in the pre-emptive arm, but particularly for the empiric arm compared with the pre-emptive arm the latter with only 2 Gram negative isolates detected in the (when other covariates were held constant, P ¼ 0.32). pre-emptive arm compared to 12 in the empiric arm Patients attending centre 1 had fewer days with fever than (Po0.01). The spectrum of Gram-positive organisms those at centre 2 (P ¼ 0.06). For the entire cohort, the differed between the two arms, with more Staphylococcus longer the duration between transplant and onset of aureus isolates in the pre-emptive arm and more Strepto- neutropenia, the fewer the days with fever (P ¼ 0.04). The coccus spp in the empiric arm. Cefepime resistance was not estimated mean rate of fever was 0.41 days/days assessed in detected in any isolate on either arm with the exception of the empiric arm and 0.37 days/days assessed in the pre- organisms expected to demonstrate intrinsic cephalosporin emptive arm. resistance (Stenotrophomonas maltophilia, Enterococcus spp,

Bone Marrow Transplantation Pre-emptive versus empiric antibiotics after SCT MA Slavin et al 161 Table 3 Organisms responsible for bacterial bloodstream infection patients with grade 2 sepsis, eight were from the pre- in patients receiving pre-emptive and empiric antibiotic therapy emptive arm and 10 from the empiric arm. Organism causing bacteremia Pre-emptive Empiric antibiotics antibiotics (N ¼ 18) (N ¼ 38) Discussion

Gram-positive organisms 16 25 Coagulase-negative Staphylococci 69To our knowledge, this is the first prospective randomized Staphylococcus aureus 51study of intravenous pre-emptive compared to empiric Streptococcus spp 1 8 antibiotics for fever and neutropenia in adult allogeneic Enterococcus spp 2 3 and autologous HSCT recipients. Consistent with our a Other 24hypothesis, patients receiving pre-emptive therapy with Gram-negative organisms 2 13 Escherischia coli 16cefepime had a significantly longer time to onset of fever Burkholderia cepacia 02after HSCT than those receiving empiric therapy. However, Stenotrophomonas maltophilia 10this difference was only of 1 day, and there was only a Otherb 05minor reduction in the incidence of fever during the neutropenic phase of transplantation (relative risk 0.86). aOther Gram-positive organisms: Clostridium tertium n ¼ 1, Bacillus sp n ¼ 2, Lactobacillus sp n ¼ 1, Proprionobacterium sp n ¼ 2. Pre-emptive antibiotics did not impact on either the bOther Gram-negative organisms: one each of Pseudomonas aeruginosa, number of days with fever overall or the duration of Proteus mirabilis, Enterobacter cloacae, Fusobacterium nucleatum, Neisseria hospital admission. Although it might be expected that by sicca. starting antibiotics earlier, total antibiotic exposure would Bold values signifies subtotals of Gram positive and negative organisms. be greater, this was not the case. There was no significant difference in number of antimicrobial agents used or total antibiotic exposure between the two arms. Consistent with our hypothesis, the number of blood- MRSA). Additionally, candidemia was detected in one stream isolates, particularly Gram-negative organisms, was pre-emptive and two empiric arm patients. reduced in those receiving pre-emptive antibiotics. As with The estimated mean rate of infection was 0.007/day in the delay in onset of fever, however, this did not translate the pre-emptive arm and 0.015/day in the empiric arm. into a reduction in total days of antibiotics, total doses of When the results were compared between patients under- antibiotics, overall proportion of patients receiving empiric going allogeneic and autologous transplantation, a similar antifungal agents, days of hospitalization or improved reduction was seen in the proportion of patients with BSIs survival. It should be noted however that there was a in the pre-emptive compared to the empiric therapy groups. reduction in empiric antifungal use in autologous but not In autologous transplant recipients, five patients receiving allogeneic HSCT recipients receiving pre-emptive antibiotic pre-emptive therapy had BSIs compared to 10 patients therapy. The overall proportion of autologous HSCT receiving empiric therapy. In allogeneic transplant recipi- recipients receiving empiric antifungal therapy was high ents, 11 patients receiving pre-emptive therapy had BSIs at 39% but may have reflected the underlying diseases of compared to 21 patients receiving empiric therapy. patients receiving autologous transplant (heavily pretreated lymphomas, myelomas or solid tumours) and the conditioning regimens used, which largely comprised mucositic Patient outcome regimens such as busulphan and melphalan (32%) and Time to neutrophil recovery was not different between the BEAM (24%) (data not shown). This reduction in empiric pre-emptive and empiric arms at 13 and 12 days, antifungal therapy is less likely to be beneficial in respectively. The median duration of hospital stay after autologous HSCT where the risk of nephrotoxicity and transplantation was 23 days for both arms (range 10–127 its associated morbidity is not as great as in allogeneic days). Overall mortality during admission for transplant- HSCT.18,19 ation was 13/72 (18%) in the pre-emptive arm and 8/73 Two earlier studies of pre-emptive intravenous therapy in (11%) in the empiric arm. There was no difference in adults undergoing SCT undertaken in the early 1990s and 30-day mortality (Table 2). Of these deaths, three in the late 1980s suggested a reduced incidence of BSI1 and fever.2 pre-emptive and two in the empiric arm were attributed to However, these trials were either not randomized2 or bacterial sepsis in the presence of positive blood cultures included other interventions such as the use of laminar (two with S. aureus and one each of Clostridium tertium, air-flow ventilation or the use of systemic non-absorbable Pseudomonas aeruginosa and Enterococcus sp). Median antibiotics.1 Recently, two studies of pre-emptive intra- time to death after transplant was 32 days (range 5–78 venous antibiotic therapy have been undertaken in adult days). In assessing the severity of sepsis in patients at one HSCT recipients, although neither was randomized and hospital, it was found that overall, 4/71 patients required comparative groups for both studies were drawn from ICU admission. Two of these were due to sepsis (one with different time periods. In autologous HSCT recipients, half Streptococcus pneumoniae bacteremia from the empiric arm of whom had solid tumours, oral norfloxacin begun at and one with S. aureus bacteremia from the pre-emptive admission was compared with piperacillin–tazobactam arm). The remaining two patients admitted to ICU had no begun on the day of transplant as prophylaxis.14 This positive blood cultures but required airway support. All retrospective historical control study showed that 76% of other (67/71) patients had grade 1 or 2 sepsis. Of the 18 the 51 patients who received piperacillin–tazobactam and

Bone Marrow Transplantation Pre-emptive versus empiric antibiotics after SCT MA Slavin et al 162 98% of the 51 patients who received norfloxacin developed studies were smaller and neither was randomized. Our fever during neutropenia. However, time to onset of fever study was underpowered for the detection of minor was significantly longer in those receiving piperacillin– differences in duration of fever or antibiotics. tazobactam and this was associated with a shorter total Our study suggests that pre-emptive intravenous anti- duration of antibiotic administration. In a pilot study, pre- biotics are unlikely to improve outcomes in patients emptive and empiric meropenem were compared in a undergoing SCT. Although time to onset of antibiotic cohort of 38 matched-related allogeneic transplant recipi- therapy could be reduced, no overall benefit in terms of ents.13 Pre-emptive meropenem was begun on the day the days of fever, morbidity or mortality could be associated neutrophil count fell to o0.5 Â 109/l. Although a significant with the delayed onset of fever. As such, these results do reduction in the occurrence of fever was observed with pre- not support the routine use of pre-emptive antibiotic emptive therapy (76% of patients treated pre-emptively therapy in patients receiving HSCT as in-patients. compared with 94% treated with empiric therapy), there was no difference in the mean number of days of fever (6 days), microbiologically documented infections or use of antifungal agents. Acknowledgements In contrast to both these studies,13,14 our study did not demonstrate a reduction in the proportion of patients We thank the nurses, medical staff and pharmacists who cared for the patients in this study, the data managers and Dr Annie developing fever. The present study did not show a reduced Solterbeck for statistical advice. use of antibiotics in either dose or duration and this finding was common to the study of Perez Simon et al.13 using meropenem in allogeneic HSCT, but differed from that of Solano et al.14 These differences in results between studies References may be due to several factors. First, differences in types of patients and type of 1 Petersen F, Thornquist M, Buckner C, Counts G, Nelson N, transplantation (increased proportion of allogeneic versus Meyers J et al. The effects of infection prevention regimens on autologous recipients and proportion receiving mismatched early infectious complications in marrow transplant patients: a four arm randomized study. Infection 1988; 16: 199–208. transplants) may have contributed to the lack of reduction 2 Roberts AW, Grigg AP, de Boer R, Panuncialman G, Brown in the number of patients with fever or antibiotic use in the G, Street A et al. A pilot study of prophylactic intravenous present study compared to the previous studies. Our study antibodies to prevent fever and infection during allogeneic included allogeneic HSCT recipients (52% of all patients bone marrow transplantation. ANZ J Med 1993; 23: 141. studied and not all were matched-related transplants). As 3 Avril M, Hartmann O, Valteau-Couanet D, Brugieres L, indicated by our findings of increased antibiotic use among Kalifa C, Lemerle J. Antiinfective prophylaxis with ceftazidime allogeneic recipients and those receiving TBI, patients and teicoplanin in children undergoing high-dose chemother- undergoing allografts have more reasons for fever, that is, apy and bone marrow transplantation. Pediatr Hematol Oncol more conditioning toxicity, methotrexate-related mucositis, 1994; 11: 63–73. GVHD and potentially a longer duration of neutropenia. 4 Hughes WT, Armstrong D, Bodey GP, Bow EJ, Brown AE, Calandra T et al. 2002 guidelines for the use of antimicrobial Thus, pre-emptive antibiotics may not have had such an agents in neutropenic patients with cancer. Clin Infect Dis impact on fever and antibiotic use in our study. Further, 2002; 34: 730–751. our study was undertaken in a population dominated by 5 Klastersky J, Paesmans M, Rubenstein EB, Boyer M, Elting L, patients with haematological malignancy requiring post- Feld R et al. The Multinational Association for Supportive transplant management in hospital. Compared to patients Care in Cancer risk index: a multinational scoring system for with diseases more amenable to outpatient-based SCT, this identifying low-risk febrile neutropenic cancer patients. J Clin population was less likely to benefit clinically from a Oncol 2000; 18: 3038–3051. reduction in the rate of fever or BSI. 6 Elting LS, Rubenstein EB, Rolston KV, Bodey GP. Outcomes Second, the differing results may reflect lesser efficacy of of bacteremia in patients with cancer and neutropenia: the drug given in our study, that is cefepime, compared observations from two decades of epidemiological and clinical trials. Clin Infect Dis 1997; 25: 247–259. with either piperacillin–tazobactam or meropenem used in 7 Bucaneve G, Micozzi A, Menichetti F, Martino P, Dionisi MS, 13,14 the two other studies. A recent meta-analysis has Martinelli G et al. Levofloxacin to prevent bacterial infection suggested that cefepime is less effective in oncology patients in patients with cancer and neutropenia. N Engl J Med 2005; with febrile neutropenia, being associated with a higher all- 353: 977–987. cause mortality and bacterial superinfection rate than other 8 Baden LR. Prophylactic antimicrobial agents and the im- beta-lactams.20 Against this possibility, no reduction in portance of fitness. N Engl J Med 2005; 353: 1052–1054. days of antibiotics was observed in the study using 9 Carratala J, Alcaide F, Fernandez-Sevilla A, Corbella X, meropenem as pre-emptive therapy in allogeneic transplant Linares J, Gudiol F. Bacteremia due to viridans streptococci recipients.13 Whether the results of the present study would that are highly resistant to penicillin: increase among Clin Infect Dis have differed had piperacillin–tazobactam been chosen neutropenic patients with cancer. 1995; 20: 1169–1173. rather than cefepime cannot be determined, but this seems 10 Carratala J, Fernandez-Sevilla A, Tubau F, Callis M, Gudiol unlikely in view of the microorganisms responsible for F. Emergence of quinolone-resistant Escherichia coli bacter- infection in patients receiving pre-emptive cefepime. emia in neutropenic patients with cancer who have received Third, the differing results could have been due to prophylactic norfloxacin. Clin Infect Dis 1995; 20: 557–560, differences in study design and sample size. Both previous discussion 561–563.

Bone Marrow Transplantation Pre-emptive versus empiric antibiotics after SCT MA Slavin et al 163 11 Kern WV, Andriof E, Oethinger M, Kern P, Hacker J, Marre 16 Sorror ML, Leisenring W, Deeg HJ, Martin PJ, Storb R. R. Emergence of fluoroquinolone-resistant Escherichia coli at a Twenty-year follow-up of a controlled trial comparing a cancer center. Antimicrob Agents Chemother 1994; 38: 681–687. combination of methotrexate plus cyclosporine with cyclos- 12 Bodey GP, Rodriguez V, Chang HY, Narboni G. Fever and porine alone for prophylaxis of graft-versus-host disease in infection in leukemic patients: a study of 494 consecutive patients administered HLA-identical marrow grafts for leuke- patients. Cancer 1978; 41: 1610–1622. mia. Biol Blood Marrow Transplant 2005; 11: 814–815. 13 Perez-Simon JA, Garcia-Escobar I, Martinez J, Vazquez L, 17 Gardner W, Mulvey EP, Shaw EC. Regression analyses of Caballero D, Canizo C et al. Antibiotic prophylaxis with counts and rates: Poisson, overdispersed Poisson, and negative meropenem after allogeneic stem cell transplantation. Bone binomial models. Psychol Bull 1995; 118: 392–404. Marrow Transplant 2004; 33: 183–187. 18 Bates DW, Su L, Yu DT, Chertow GM, Seger DL, Gomes DR 14 Solano C, Gutierrez A, Martinez F, Gimeno C, Gomez C, et al. Mortality and costs of acute renal failure associated Munoz I et al. Prophylaxis of early bacterial infections after with amphotericin B therapy. Clin Infect Dis 2001; 32: autologous peripheral blood stem cell transplantation 686–693. (PBSCT): a matched-pair study comparing oral fluoroquino- 19 Wingard JR, Kubilis P, Lee L, Yee G, White M, Walshe L lones and intravenous piperacillin–tazobactam. Bone Marrow et al. Clinical significance of nephrotoxicity in patients treated Transplant 2005; 36: 59–65. with amphotericin B for suspected or proven aspergillosis. Clin 15 Shryock TR, Apley M, Jones RN, Lein DH, Thornsberry C, Infect Dis 1999; 29: 1402–1407. Walker RD et al. (eds). Institute, C.a.L.S. M31-A2 (Electronic 20 Paul M, Yahav D, Fraser A, Leibovici L. Empirical antibiotic Document) Performance Standards for Antimicrobial Disk and monotherapy for febrile neutropenia: systematic review and Dilution Susceptibility Tests for Bacteria 2002. Clinical and meta-analysis of randomized controlled trials. J Antimicrob Laboratory Standards Institute: Wayne, Pennsylvania, USA. Chemother 2006; 57: 176–189.

Bone Marrow Transplantation