Support Care Cancer (2003) 11:362370 DOI 10.

1007/s00520-003-0463-7

O R I G I N A L A RT I C L E

M. Gorschlter C. Hahn A. Fixson U. Mey C. Ziske E. Molitor R. Horr T. Sauerbruch G. Marklein I. G. H. Schmidt-Wolf A. Glasmacher

Piperacillin-tazobactam is more effective than ceftriaxone plus gentamicin in febrile neutropenic patients with hematological malignancies: a randomized comparison

Received: 18 November 2002 Accepted: 5 March 2003 Published online: 11 April 2003 Springer-Verlag 2003

M. Gorschlter C. Hahn A. Fixson U. Mey C. Ziske T. Sauerbruch I. G. H. Schmidt-Wolf A. Glasmacher () Department of Internal Medicine I, University of Bonn, Sigmund-Freud-Str. 25, 53105 Bonn, Germany e-mail: glasmacher@uni-bonn.de Tel.: +49-228-2875507 Fax: +49-228-2875849 E. Molitor R. Horr G. Marklein Department of Medical Microbiology and Immunology, University of Bonn, Sigmund-Freud-Str. 25, 53105 Bonn, Germany

Abstract Goals: Efficacy and costs of empirical antibacterial therapy in febrile neutropenic patients are important issues. Several strategies have been reported to be similarly effective: monotherapy with cefepime, ceftazidime or a carbapenem or duotherapy with an antipseudomonal -lactam antibiotic or ceftriaxone in combination with an aminoglycoside. Piperacillin-tazobactam monotherapy is promising, but its role in this setting still has to be defined. Patients and methods: Of 212 consecutive febrile episodes in 130 neutropenic patients with hematological malignancies randomized to receive either piperacillin-tazobactam (4.5 g every 8 h; group A) or ceftriaxone (2 g once daily plus gentamicin 5 mg/kg once daily; group B), 183 episodes (98 group A, 85 group B) were evaluable for response. Results: Defervescence within 72 h without modification of the antibiotic therapy was achieved in 56/98 epi-

sodes (57.1%) in group A and in 30/85 (35.3%) in group B (P=0.0047). If fever persisted, teicoplanin plus gentamicin (group A) or teicoplanin plus ciprofloxacin (group B) were added. All patients still febrile then received meropenem, teicoplanin and amphotericin B. With these modifications of antibiotic therapy, 89.8% of patients in group A had responded at 21 days but only 71.8% in group B (P=0.005). The mean total antibiotic drug cost in group A was only 39.4% of that in group B ( 445 versus 1129; P=0.010). Conclusion: Piperacillin-tazobactam monotherapy is significantly more effective and cost-efficient than ceftriaxone plus gentamicin as firstline therapy in febrile neutropenic patients with hematological malignancies. Keywords Neutropenia Acute leukemia Febrile infections Empiric antibiotic therapy

Introduction
The guidelines of the Infectious Diseases Society of America (IDSA) published in 2002 recommend either monotherapy with cefepime or ceftazidime or a carbapenem or duotherapy with an antipseudomonal -lactam antibiotic in combination with an aminoglycoside as empiric antibacterial therapy in febrile high-risk neutropenic patients [13]. Once-daily ceftriaxone plus aminoglycoside leads to a much lower daily drug cost. A recent metaanalysis has demonstrated that the combination of cef-

triaxone with an aminoglycoside is as effective as ceftazidime plus aminoglycoside for the empiric treatment of infections in neutropenic patients [7]. Among these studies is a large randomized trial organized by the EORTC that revealed equivalent response rates to ceftriaxone plus amikacin and ceftazidime plus amikacin in 858 febrile episodes [6]. Piperacillin-tazobactam, a combination of a broadspectrum -lactam antibiotic and a potent -lactamase inhibitor, is appropriate for many infections traditionally treated empirically by double or triple antibiotic therapy

363

[9]. Several studies have shown promising results of piperacillin-tazobactam with or without aminoglycoside addition in febrile neutropenic cancer patients [1, 3, 10, 11, 17]. Recent data suggest that the addition of aminoglycosides to piperacillin-tazobactam may not be necessary [5]. However, very limited data are available on monotherapy with piperacillin-tazobactam compared to standard regimens in febrile neutropenia [11]. In particular, piperacillin-tazobactam has not been compared in a randomized study with the widely used combination of ceftriaxone plus gentamicin. In this trial the efficacy and cost-efficiency of piperacillin-tazobactam (cost 55/day) administered three times daily as monotherapy was compared with once-daily ceftriaxone plus gentamicin (cost 31/day) in high-risk patients with hematological malignancies.

overall mortality during the study period, infection-related mortality, toxicity and drug acquisition cost of antibiotic therapy. Inclusion criteria, exclusion criteria and definitions are shown in Table 1. A urinary tract infection was defined as more than 100,000 cfu/ml in aseptically collected midstream urine in the presence of fever. Patients could be enrolled for successive separate episodes of fever and neutropenia, provided a previous episode had resolved and treatment for the previous episode had been completed at least 7 days before enrollment. All analyses were performed for the entire study population and the subgroups as specified below. Response after 72 h of antibiotic therapy and overall response rates to levels 13 were also calculated separately for the patients first entries. Antimicrobial prophylaxis Antimicrobial prophylaxis for neutropenic patients consisted of mucosal antiseptics, systemic fungal prophylaxis with itraconazole [8] and oral antibiotic prophylaxis with oral colistine 200 mg every 8 h and trimethoprim-sulfamethoxazole 160 mg/800 mg every 12 h until recovery of the neutrophil count or fever requiring intravenous antimicrobial therapy. Prophylactic aciclovir was also administered orally from July 1998 onwards at a dose of 400 mg every 12 h during neutropenia. Classification of febrile episodes At the onset of fever, at least three blood cultures, cultures from urine and other presumably infected sites were obtained. The microbiology department of our university hospital analyzed all specimens by standard culture techniques. A chest radiograph was obtained within 24 h. Primary febrile episodes were classified as (1) fever of unknown origin (FUO), (2) pneumonia (radiologically confirmed) and (3) non-pneumonic microbiologically documented infection (MDI) and/or clinically documented infection (CDI). FUO was defined as the absence of isolation of any pathogen, any clinical or any radiological sign of infection. Pneumonia was defined as the presence of a pulmonary infiltration in a conventional chest radiograph or a high-resolution computed tomography of the chest. Evaluation of response Responses were evaluated 72 h after the initiation of antibacterial therapy (early evaluation) and after the completion of treatment

Patients and methods

Study design and patient population This randomized, open-label study was conducted in the Leukemia Unit of the Medical Center of the University of Bonn between July 1996 and May 1999. The trial protocol was approved according to German law by the Ethical Committee of the University of Bonn which is independent of our institution. Its guidelines for human experimentation were followed in the conduct of the trial and informed consent was obtained from the patients. Consecutive febrile neutropenic patients were randomly assigned by computergenerated lists of the allocation sequence and sealed opaque envelopes to receive either piperacillin-tazobactam 4.5 g every 8 h (group A) or ceftriaxone 2 g once daily plus gentamicin 5 mg/kg once daily (group B) as empiric therapy. If fever persisted after 72 h, teicoplanin 400 mg once daily plus gentamicin 5 mg/kg once daily were added in group A and teicoplanin 400 mg once daily plus ciprofloxacin 200 mg every 12 h in group B (level 2). After a further 72 h all patients who were still febrile received meropenem 1 g every 8 h, teicoplanin 400 mg once daily and amphotericin B 1 mg/kg once daily for at least 96 h (level 3). The primary predetermined end-point was response after 72 h of antibiotic therapy. Secondary end-points were responses to level 2 or level 3 of antibiotic therapy, the number of febrile days, Table 1 Inclusion and exclusion criteria

Inclusion criteria Fever (axillary temperature >38.0C or rectal temperatures >38.5C) and Neutropenia (WBC count <1109/l or neutrophils <0.5109/l or WBC count >1109/l but expected to decrease <1109/l within 24 to 48 h of trial entry and Age >18 years and Acute leukemia, blast crisis of chronic leukemia or myelodysplastic syndrome, high-grade lymphoma, multiple myeloma after high-dose chemotherapy with subsequent peripheral blood progenitor cell transplantation (PBPCT) or severe aplastic anemia Exclusion criteria History of hypersensitivity to any study drug or Systemic antibiotic pretreatment within 5 days before randomization (with exception of colistine, cotrimoxazole and metronidazole) or Marked renal function impairment (creatinine >25 mg/l) or Human immunodeficiency virus infection or Allogeneic bone marrow transplantation within 3 months of the start of the trial or Estimated life expectancy of 14 days or less or Pregnancy or lactation

364

(overall evaluation). Success (primary end-point) was defined as persistent (7 days) defervescence (temperature below 38C without antipyretic medication) in the absence of clinical signs of continuing infection. Failure was defined as one of the following events: persistent fever at the time of early evaluation (see above), relapse of fever within 7 days of defervescence (overall evaluation), worsening of general condition, or necessity for the addition of other parenteral antibiotics according to susceptibility testing from positive blood cultures. Statistical methods All analyses were performed using two-tailed tests with an alpha level of P<0.05. Differences between categorical variables were tested with univariate 2-tests. Differences between continuous variables were compared using the Mann-Whitney test. The study was discontinued according to the protocol because the second interim analysis revealed a significant difference at the predetermined threshold P<0.005. Efficacy was analyzed for the modified intent-to-treat population (all randomized patients who fulfilled the inclusion criteria) in a manner analogous to the study of Kern et al. [15]: episodes which violated the predefined inclusion or exclusion criteria and did not represent the population of interest were excluded. All episodes which violated the regimen of the randomly assigned therapy were included. Prospectively defined exploratory subgroup analyses were performed for patients with FUO, pneumonia or non-pneumonic documented infection, documented gram-negative and documented gram-positive infections. Subgroup analyses were additionally performed for patients with MDI, bacteremia (subgroup of the latter subgroup) and CDI without microbiological confirmation. The literature was scanned by a computer-assisted search of Medline from 1966 to 2002. Data storage and statistical tests were computed using SPSS 10.0 software (SPSS, Chicago, Ill.). Total acquisition costs of antibacterial drugs were calculated by adding the hospital pharmacy prices as of May 1999 including 16% VAT (in Euros) of all antibacterial drugs for up to 21 days of each episode. Indirect costs of antibiotic therapy were calculated by adding the costs of material not provided by the manufacturer of the antibiotic and costs for nursing time.

Comparability of treatment groups and clinical characteristics The febrile neutropenic episodes were well balanced between the two groups with respect to duration of neutropenia, administration of high-dose cytarabine therapy, use of granulocyte colony-stimulating factor (G-CSF), use of a central venous line, frequency of bacteremia, and predefined subgroup of infection (Table 2). Moreover, the patients in both groups were well balanced with respect to age, gender, and underlying malignancy (Table 3). Clinical response Response defined as sustained defervescence within 72 h was achieved in 57.1% of episodes (56/98, 95% CI 4767%) in group A and in 35.3% of episodes (30/85, 95% CI 2546%) in group B (P=0.0047). After up to 21 days, 89.8% of episodes (88/98, 95% CI 8295%) in group A responded but only 71.8% (61/85, 95% CI 6182%) in group B (P=0.005). The administration of level 3 antibiotic therapy varied between 96 h and 144 h. Further details are shown in Fig. 1. Including only the first febrile episodes in the analysis, defervescence within 72 h was achieved in 55.4% of episodes (31/56, 95% CI 4169%) in group A and in 31.4% of episodes (16/51, 95% CI 1946%) in group B (P=0.019). After up to 21 days, 92.9% of episodes (50/56, 95% CI 7896%) in group A responded but only 74.5% (38/51, 95% CI 6086%) in group B (P=0.074). Time to defervescence was significantly shorter in group A (median 3 days, interquartile range 16 days) than in group B (median 4 days, interquartile range 27 days; P=0.027). According to the protocol, the study was discontinued prematurely because an interim analysis showed statistically significant differences in overall response rates. No patient died within 72 h of the start of antibiotic therapy. After modification of initial therapy, five deaths during the study period occurred in group A: CNS bleeding (one); pneumonia (Aspergillus spp.) (two), pneumonia (other pathogen) (two). In group B, eight deaths occurred: progressive leukemia (two), septic shock (two), pneumonia (Aspergillus spp.) (two), pneumonia (unknown pathogen) (one) and neutropenic enterocolitis (one). One death in group B, but none of the deaths in group A, was caused by Pseudomonas aeruginosa. The differences in overall death rates (P=0.258) and infection-related death rates (P=0.377) were not statistically significant. Of the 183 episodes of febrile neutropenia, 88 (48.1%) were classified as FUO, 45 (24.6%) as pneumonia and 50 (27.3%) as non-pneumonic documented infection, the latter including bacteremia (35 patients), MDI of the urinary tract (6), MDI of the gastrointestinal

Results
From July 1996 to May 1999, 212 febrile episodes in 130 patients were randomized (Fig. 1). Of these 212 episodes, 15 were excluded from intent-to-treat analysis in group A and 14 in group B: 13 patients had non-prophylactic systemic antibiotics during the 5 days before randomization (7 in group A and 6 in group B), 5 had a temperature below 38.0C (3 and 2), 5 were not neutropenic (2 and 3), 2 presented with renal failure (1 and 1) and in 3 the record was missing (1 and 2). Of the remaining 183 episodes, 98 in 56 patients were evaluable for response in group A and 85 in 51 patients in group B (Fig. 1). In 163 of 183 episodes (89%) the patients suffered from acute leukemia.

365

Fig. 1 Flow chart of patient episodes according the CONSORT statement [19]. Modification according to susceptibility testing of identified organism

tract (1), CDI of the gastrointestinal tract (2), central venous line infection (1), funiculitis (1), bronchitis (1), perianal abscess (1), axillary sweat gland abscess (1) and generalized metastatic abscesses (1). After 72 h, response rates in these subgroups to piperacillin-tazobactam were higher than to ceftriaxone plus gentamicin. However, the differences were not statistically significant (Table 4) due to the smaller group sizes and the resulting loss of statistical power. In an alternative subdivision, 60 non-FUO episodes were classified as MDI with (n=49) and without (n=11) bacteremia (see also Table 5)

and 35 as CDI (Table 2). Response rates of these subgroups are shown in Table 4. In 60 febrile episodes, the causative pathogen could be isolated. Sources of isolation are shown in Table 5. In two patients with pneumonia, Aspergillus spp. was confirmed histologically. In 58 episodes (pneumonia and non-pneumonic documented infections), bacteria were considered to be responsible for the fever. Four febrile episodes were excluded from the subgroup analysis according to gram-staining as they were polymicrobial. The response rate of 31 documented infections caused

366

Table 2 Clinical characteristics and comparability of treatment groups by febrile episodes (FUO fever of unknown origin, MDI microbiologically documented infection, CDI clinically documented infection) Parameter Number of evaluable episodes Duration of neutropenia (days) (Table 1) Median Range Duration of neutropenia after enrollment (days) Median Range G-CSF administration (%) Central venous line (%) Classification of episodes (I) FUO Pneumonia Documented infection (except pneumonia) Classification of episodes (II) FUO MDI Bacteremia (including pneumonia) Other MDI CDI Urinary tract infections Gastrointestinal tract infections Central venous line infection Funiculitis Bronchitis Perianal abscess Axillary sweat gland abscess Generalized metastatic abscess Piperacillintazobactam 98 22 463 11 160 46 97 48 (49.0%) 25 (25.5%) 25 (25.5%) 48 (49.0%) 31 (31.6%) 24 (24.5%) 7 (7.1%) 19 (19.4%) 3 1 1 1 1 1 Ceftriaxone plus gentamicin 85 20 555 11 241 52 96 40 (47.1%) 20 (23.5%) 25 (29.4%) 40 (47.1%) 29 (34.1%) 25 (29.4%) 4 (4.7%) 16 (18.8%) 3 2 1 1 88 (48.1%) 45 (24.6%) 50 (27.3%) 0.938 88 (48.1%) 60 (32.8%) 49 (26.8%) 11 (6.0%) 35 (19.1%) 6 3 1 1 1 1 1 1 0.461 0.835 Total 183 0.568 P-value

Table 3 Clinical characteristics and comparability of individual patients (AML acute myeloid leukemia, ALL acute lymphoblastic leukemia, NHL highgrade non-Hodgkins-lymphoma, CML chronic myeloid leukemia (blast crisis), MM multiple myeloma)

Parameter Number of evaluable patients Age (years) Median Range Sex Male Female Underlying malignancy AML ALL CML blast crisis NHL (high grade) Other (mainly MM)

Piperacillintazobactam 56 58 1879 36 (64%) 20 (36%) 41 (73.2%) 6 (10.7%) 1 (1.8%) 6 (10.7%) 2 (3.6%)

Ceftriaxone plus gentamicin 51

P-value

0.109 50 1882 0.122 25 (49%) 26 (51%) 0.330 30 (58.8%) 8 (15.7%) 4 (7.8%) 3 (5.8%) 6 (11.8%)

367

Table 4 Response rates after 72 h by type of infection (FUO fever of unknown origin, MDI microbiologically documented infections, CDI clinically documented infections)

Group All episodes (72 h) All episodes after level 3 Total no. of deaths Infection-related deaths Classification of episodes (I) FUO Pneumonia Non-pneumonic documented infection Classification of episodes (II) FUO MDI MDI bacteremia CDI Single gram-positive infections (all) Single gram-positive infections (all except coagulase-negative staphylococci) Single gram-negative infections

Piperacillintazobactam 56/98 (57.1%) 88/98 (89.8%) 5/98 (5.1%) 4/98 (4.1%) 35/48 (72.9%) 6/25 (24.0%) 15/25 (60.0%)

Ceftriaxone plus gentamicin 30/85 (35.3%) 61/85 (71.8%) 8/85 (9.4%) 6/85 (7.1%) 21/40 (52.5%) 1/20 (5%) 8/25 (32.0%)

P-value 0.0047 0.005 0.258 0.377 0.088 0.112 0.088

35/48 (72.9%) 13/31 (41.9%) 10/24 (41.7%) 8/19 (42.1%) 6/18 (33.3%) 4/12 (33.3%) 6/9 (66.7%)

21/40 (52.5%) 8/29 (27.6%) 6/25 (24.0%) 1/16 (6.3%) 2/13 (15.4%) 2/8 (25%) 6/14 (42.9%)

0.088 0.287 0.232 0.022 0.412 1.000 0.400

Table 5 Sources of isolated causative pathogens

Source Blood culture Urine culture Tissue specimen (lung) Tracheal aspirate Central venous line Stool Total

All 49 6 2 1 1 1 60

Gram-positive 28 1 1 1 31

Gram-negative 17 5 1 23

Double infection 4 (bacterial) 2 (fungal) 6

by gram-positive bacteria to piperacillin-tazobactam (33.3%, 6/18 episodes, 95% CI 1359%) was markedly higher than to ceftriaxone plus gentamicin (15.4%, 2/13 episodes, 95% CI 245%), but the difference was not statistically significant (P=0.412) due to the low number of cases. In 23 documented infections caused by gramnegative bacteria the difference in success rates between piperacillin-tazobactam (66.7%, 6/9 episodes, 95% CI 3093%) and ceftriaxone plus gentamicin (42.9%, 6/14 episodes, 95% CI 1871%; P=0.4) was less pronounced (Table 4). Of 53 isolates from initial blood cultures (isolates from polymicrobial cultures included), 34 (64.2%) were gram-positive. Tazobactam restored susceptibility to piperacillin in five causative isolates (9.6% of 52 evaluable episodes in both groups), four of which were gram-negative, resulting in an overall susceptibility of causative isolates in 75% (39/52). Detailed susceptibility rates of the causative bacteria to the study antibiotics are shown in Table 6. In six patients (3.3%), P. aeruginosa was the causative organism. Two of three patients treated initially with piperacillin-tazobactam (one with urinary tract in-

fection and one with bacteremia) responded within 72 h. The third patient with bacteremia responded after the addition of gentamicin and teicoplanin (level 2). In contrast, none of three patients treated with ceftriaxone plus gentamicin responded within 72 h. One 47-year-old female patient died from septic shock due to P. aeruginosa 9 days after the onset of fever and 13 days after the start of neutropenia following high-dose cytarabine therapy. The strain was resistant to ceftriaxone but susceptible to piperacillin-tazobactam and gentamicin. The focus of the infection was a microbiologically confirmed necrotizing inflammation of the left upper maxilla. The course was fatal despite a change from ceftriaxone to piperacillin-tazobactam both combined with gentamicin after 4 days and surgical treatment. Two other patients survived but also failed to respond to levels 2 and 3. In one of these patients, pneumonia suspected to be caused by Aspergillus spp. developed. In the other patient, suffering from an axillary abscess, success was achieved with surgical treatment and long-term combination therapy consisting of ceftazidime, meropenem, amikacin and teicoplanin.

368

Table 6 Response to therapy and susceptibilities by infecting organisms Pathogen Total Piperacillin-tazobactam Response (day 3) Gram-positive bacteria Coagulase-negative staphylococci -Hemolysing streptococci Streptococcus mitis Streptococcus acidominimus Gemella morbillorum Staphylococcus haemolyticus Staphylococcus aureus Aerococcus spp. Clostridium sporogenes Enterococcus faecalis Gram-negative bacteria Escherichia coli Pseudomonas aeruginosa Klebsiella pneumoniae Proteus mirabilis Moraxella nonliquefaciens Morganella morganii Klebsiella oxytoca Double bacterial Aspergillus (histologically)
a In

Ceftriaxone plus gentamicin Response (day 3) Susceptibility

Susceptibility

31 11 6 5 2 2 1 1 1 1 1 23 8 5 4 3 1 1 1 4 2

2/6 3/4 0/1 0/1 0/2 0/1 0/1 0/1 1/1 2/2 2/3 1/1 1/1 0/1 0/1 1/3 0/1

1/6 3/3a 1/1 1/1 2/2 0/1 1/1 1/1 1/1 2/2 2/2a 1/1 1/1 1/1 1/1

0/5 1/2 1/4 0/1 0/1 3/6 0/2 1/3 2/3 0/1 (including P. aeruginosa) 0/1

2/5 2/2 2/4 1/1 1/1 4/4a 1/2 3/3 3/3

single cases antimicrobial susceptibility was not evaluable

Two patients suffered from proven invasive fungal infections (one with histologically confirmed Aspergillus pneumonia in each group). No fungemia occurred. No patient fulfilled the criteria for a possible invasive fungal infection. Ten episodes fulfilled the criteria for a probable invasive fungal infection (these patients showed signs suggestive of Aspergillus pneumonia on CT imaging without microbiological confirmation). The incidence of diarrhea associated with Clostridium difficile was identical (8.2%) in the two groups. Cost analysis The daily drug cost of initial empiric therapy with piperacillin-tazobactam (group A) was 55 compared to 31 with ceftriaxone plus gentamicin (group B). However, the total cost of antibacterial drugs including modifications of therapy was 2.5-fold higher in group B ( 1129, 2671950) than in group A ( 445, 2971388). This difference was highly significant (P=0.010). The mean indirect costs of antibiotic therapy were only slightly higher in group B ( 77/episode) than in group A ( 72/episode) and had no relevant influence on total costs.

Discussion
This is the first study in which the superior efficacy of monotherapy with piperacillin-tazobactam over ceftriaxone plus aminoglycoside has been demonstrated in neutropenic febrile patients. The recent guidelines of IDSA recommend the use in this setting of either monotherapy with cefepime or ceftazidime or a carbapenem or duotherapy consisting of an antipseudomonal -lactam antibiotic plus an aminoglycoside [13]. Some experts would question whether ceftriaxone plus aminoglycoside can be considered as an adequate first-line therapy for neutropenic fever. However, a recent metaanalysis [7] and a number of randomized studies [2, 6, 16, 21] have demonstrated that the combination of ceftriaxone with an aminoglycoside is as effective as ceftazidime plus aminoglycoside for the empiric treatment of infections in neutropenic patients. When combined with aminoglycoside, the limited efficacy of ceftriaxone against Pseudomonas spp. did not significantly influence response rates in these studies. Furthermore, ceftriaxone plus tobramycin was found to be as effective as cefotaxime every 12 h plus tobramycin in febrile neutropenic patients [4]. We selected ceftriaxone plus gentamicin in our randomized trial as standard therapy because of the promising results reported in the literature and the low drug acquisition costs.

369

However, the second interim analysis revealed a statistically significantly higher efficacy of piperacillintazobactam and the study had to be discontinued after inclusion of 183 episodes. The superior efficacy of piperacillin-tazobactam in our study was a result of higher response rates in all patient subgroups (FUO, pneumonia, non-pneumonic documented infections, gram-negative infections and gram-positive infections). However, due to the reduced statistical power, differences between subgroups did not reach statistical significance. Our data indicate that the role of ceftriaxone plus aminoglycoside as empiric standard therapy in febrile neutropenic patients suffering from hematological malignancies is questionable and that piperacillin-tazobactam is clearly preferable. Furthermore, our results are in accordance with those of three other large randomized studies in which piperacillin-tazobactam plus aminoglycoside (amikacin or tobramycin) was compared with ceftazidime plus aminoglycoside and in all of which the cephalosporin-based therapy was found to be inferior [3, 17, 18]. One explanation for these results may be provided by in vitro studies, in which improved suppression of resistant strains by the combination of piperacillin and tazobactam was demonstrated in comparison with elevated concentrations of piperacillin or ceftazidime alone [12]. The suppressive effect of tazobactam on the appearance of resistance was probably mediated by the -lactamase inhibitory activity [14]. The better efficacy of piperacillin-tazobactam in our study could not be attributed to a single factor, but may be explained by the results of our exploratory subgroup analyses. The response rate of 35% to ceftriaxone plus gentamicin in our study was lower than in related studies using ceftriaxone plus amikacin in which response rates of 64% to 71% were found. It appears unlikely that the substitution of amikacin by gentamicin caused a marked loss of efficacy. The differences are more probably consequences of less-strict definitions of response in those studies [2, 16]. Furthermore, they may have been due to a more serious risk of poor outcome in our patients (89% had acute leukemia). In other studies patients suffering from solid tumors were also included [2, 6, 21], the median duration of neutropenia was shorter [2, 6] and the definition of neutropenia was less stringent [6, 21]. Empiric monotherapy with broad-spectrum -lactam antibiotics has been shown to be as efficient as combination therapy with an aminoglycoside for febrile neutropenic patients in many studies. Recently, Del Favero et al. found no significant differences between piperacillintazobactam and piperacillin-tazobactam plus amikacin in terms of response rates in a double-blind randomized study including 733 evaluable patients [5]. Moreover, aminoglycoside-related adverse events such as nephrotoxicity and ototoxicity have to be considered.

During the last decade, pressure to control the growth of drug costs has increased markedly. Combination therapy with ceftriaxone and gentamicin seemed to be attractive due to the appropriate antimicrobial activity of these drugs and lower acquisition cost compared with other broad-spectrum -lactam aminoglycoside combinations. However, in our study the total antibiotic drug costs in patients receiving ceftriaxone plus gentamicin as initial therapy were 2.5-fold higher than in our control group receiving piperacillin-tazobactam. This was a result of a reduced need for second- and third-line therapy and fewer febrile and treatment days. The types of pathogens isolated from patients in our trial were consistent with those commonly associated with infection in neutropenic patients. Over the past three decades there has been a shift toward gram-positive bacteremia in febrile neutropenic patients. In most centers today, approximately 70% of bacteremic isolates are gram-positive cocci [22]. In our study, 64.2% of isolates from primary bacteremias were gram-positive organisms, which is well in line with these observations. The incidence of P. aeruginosa was low (3.3%) in both groups, and had therefore no significant influence on response rates. Although piperacillin has a broad spectrum of antibacterial activity, the increased prevalence of -lactamase-producing bacteria over recent years has led to an increased resistance to this agent. The -lactamase inhibitor tazobactam has good activity against plasmid-mediated -lactamases, staphylococcal penicillinase and chromosomal 2e -lactamases [20]. In our study, tazobactam restored the susceptibility to piperacillin of five causative isolates (9.6% of 52 evaluable episodes), four of which were gram-negative, resulting in an overall susceptibility of causative isolates in 75% of episodes (39/52). Recent guidelines recommend that empiric glycopeptides should be considered in numerous situations if the initial therapy with broad-spectrum antibiotics is not successful [13]. However, these antibiotics lead to additional cost and side effects, and especially the induction of resistant enterococci and Staphylococcus aureus has been described. In our study, initial therapy with piperacillin-tazobactam permitted a marked decrease in empiric glycopeptide use in comparison with ceftriaxone plus gentamicin. We conclude that piperacillin-tazobactam is significantly more effective and cost-efficient than ceftriaxone plus gentamicin as first-line therapy in febrile neutropenic patients with hematological malignances and a long duration of neutropenia.
Acknowledgements This study was supported by LeukmieInitiative Bonn e.V., Hoffmann-La Roche AG, Grenzach-Wyhlen, and Wyeth Pharma GmbH, Mnster, Germany.

370

References
1. Bohme A, Shah PM, Stille W, Hoelzer D (1998) Piperacillin/tazobactam versus cefepime as initial empirical antimicrobial therapy in febrile neutropenic patients: a prospective randomized pilot study. Eur J Med Res 3:324330 2. Charnas R, Luthi AR, Ruch W (1997) Once daily ceftriaxone plus amikacin vs. three times daily ceftazidime plus amikacin for treatment of febrile neutropenic children with cancer. Writing Committee for the International Collaboration on Antimicrobial Treatment of Febrile Neutropenia in Children. Pediatr Infect Dis J 16:346353 3. Cometta A, Zinner S, de Bock R, Calandra T, Gaya H, Klastersky J, Langenaeken J, Paesmans M, Viscoli C, Glauser MP (1995) Piperacillintazobactam plus amikacin versus ceftazidime plus amikacin as empiric therapy for fever in granulocytopenic patients with cancer. The International Antimicrobial Therapy Cooperative Group of the European Organization for Research and Treatment of Cancer. Antimicrob Agents Chemother 39:445452 4. Cornely OA, Bethe U, Salzberger B, Franzen C, Hartmann P, Steinmetz T, Fatkenheuer G, Seifert H, Diehl V, Schrappe M (2001) Randomized controlled monocentric comparison of once daily ceftriaxone with tobramycin and cefotaxime three times daily with tobramycin in neutropenic fever. Ann Hematol 80:103108 5. Del Favero A, Menichetti F, Martino P, Bucaneve G, Micozzi A, Gentile G, Furno P, Russo D, DAntonio D, Ricci P, Martino B, Mandelli F (2001) A multicenter, double-blind, placebocontrolled trial comparing piperacillintazobactam with and without amikacin as empiric therapy for febrile neutropenia. Clin Infect Dis 33:12951301 6. EORTC-IATCG (1993) Efficacy and toxicity of single daily doses of amikacin and ceftriaxone versus multiple daily doses of amikacin and ceftazidime for infection in patients with cancer and granulocytopenia. The International Antimicrobial Therapy Cooperative Group of the European Organization for Research and Treatment of Cancer. Ann Intern Med 119:584593 7. Furno P, Dionisi MS, Bucaneve G, Menichetti F, Del Favero A (2000) Ceftriaxone versus beta-lactams with antipseudomonal activity for empirical, combined antibiotic therapy in febrile neutropenia: a meta-analysis. Support Care Cancer 8:293301 8. Glasmacher A, Hahn C, Molitor E, Marklein G, Sauerbruch T, SchmidtWolf IG (1999) Itraconazole through concentrations in antifungal prophylaxis with six different dosing regimens using hydroxypropyl-beta-cyclodextrin oral solution or coated-pellet capsules. Mycoses 42:591600 9. Gorbach SL (1994) Piperacillin/ tazobactam in the treatment of polymicrobial infections. Intensive Care Med 20 [Suppl 3]:S27S34 10. Hazel DL, Graham J, Dickinson JP, Newland AC, Kelsey SM (1997) Piperacillin-tazobactam as empiric monotherapy in febrile neutropenic patients with haematological malignancies. J Chemother 9:267272 11. Hess U, Bohme C, Rey K, Senn HJ (1998) Monotherapy with piperacillin/ tazobactam versus combination therapy with ceftazidime plus amikacin as an empiric therapy for fever in neutropenic cancer patients. Support Care Cancer 6:402409 12. Higashitani F, Nishida K, Hyodo A, Inoue M (1995) Effects of tazobactam on the frequency of the emergence of resistant strains from Enterobacter cloacae, Citrobacter freundii, and Proteus vulgaris (beta-lactamase derepressed mutants). J Antibiot (Tokyo) 48:10271033 13. Hughes WT, Armstrong D, Bodey GP, Bow EJ, Brown AE, Calandra T, Feld R, Pizzo PA, Rolston KV, Shenep JL, Young LS (2002) 2002 guidelines for the use of antimicrobial agents in neutropenic patients with cancer. Clin Infect Dis 34:730751 14. Kadima TA, Weiner JH (1997) Mechanism of suppression of piperacillin resistance in enterobacteria by tazobactam. Antimicrob Agents Chemother 41:21772183 15. Kern WV, Cometta A, de Bock R, Langenaeken J, Paesmans M, Gaya H (1999) Oral versus intravenous empirical antimicrobial therapy for fever in patients with granulocytopenia who are receiving cancer chemotherapy. International Antimicrobial Therapy Cooperatve Group of the European Organization for Research and Treatment of Cancer. N Engl J Med 341:312318 16. Leoni F, Ciolli S, Pascarella A, Fanci R, Caporale R, Rossi FP (1993) Ceftriaxone plus conventional or single-daily dose amikacin versus ceftazidime/ amikacin as empiric therapy in febrile neutropenic patients. Chemotherapy 39:147152 17. Marie JP, Vekhoff A, Cony-Makhoul P, Fiere D, Guy H, Herbrecht R, Milpied N, Pico JL, Plantier I (1995) Piperacillin/tazobactam combination+amikacin versus ceftazidime+ amikacin in patients with neutropenia and fever. An open multicenter study. Groupe detude des Aplasies Febriles (in French). Presse Med 24:397401 18. Marie JP, Marjanovic Z, Vekhoff A, Bouvet A, Chast F, Levy V, Baudard M, Legrand O, Rio B, Delmer A, Zittoun R (1999) Piperacillin/ tazobactam plus tobramycin versus ceftazidime plus tobramycin as empiric therapy for fever in severely neutropenic patients. Support Care Cancer 7:8994 19. Moher D, Schulz KF, Altman DG, Lepage L (2001) The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomised trials. Lancet 357:11911194 20. Perry CM, Markham A (1999) Piperacillin/tazobactam: an updated review of its use in the treatment of bacterial infections. Drugs 57:805843 21. Schmid L, Jeschko M, Wilder-Smith C, Schafroth U, Thurlimann B, Pedrazzini A, Senn H (1991) Ceftriaxone and amikacin versus ceftazidime and amikacin in febrile granulocytopenia. Chemotherapy 37:346352 22. Zinner SH (1999) Changing epidemiology of infections in patients with neutropenia and cancer: emphasis on gram-positive and resistant bacteria. Clin Infect Dis 29:490494