Monotherapy with Intravenous Followed by Oral High-Dose Ciprofloxacin versus Combination Therapy with Ceftazidime plus Amikacin as Initial Empiric Therapy for Granulocytopenic Patients with Fever

ABSTRACT The aim of the present study was to obtain clinical experience with the use of high-dose ciprofloxacin as monotherapy for the treatment of febrile neutropenia episodes (granulocyte count, <500/mm3) compared to a standard regimen and to clarify whether ciprofloxacin administration may be switched to the oral route. In a prospective randomized study ciprofloxacin was given at 400 mg three times a day (t.i.d.) for at least 72 h followed by oral administration at 750 mg twice a day (b.i.d). That regimen was compared with ceftazidime given intravenously at 2 g t.i.d. plus amikacin given intravenously at 500 mg b.i.d. The frequency of successful clinical response without modification at the end of therapy was almost identical for ciprofloxacin (50% [62 of 124 patients]) compared with that for ceftazidime plus amikacin (50.8% [62 of 122 patients]) in an intent-to-treat analysis; the frequencies were 48.3% (57 of 118 patients) versus 49.6% (56 of 113 patients), respectively, in a per-protocol analysis (P values for one-sided equivalence, 0.0485 and 0.0516, respectively; δ = 10%), with no significant differences among patients with bacteremia and other microbiologically or clinically documented infections and fever of unknown origin. For 82 (66.1%) patients, it was possible to switch from parenteral ciprofloxacin to the oral ciprofloxacin, and the response was successful for 61 (74.4%) patients. The efficacies of the regimens against streptococcal bacteremias were 16.6% (one of six patients) for the ciprofloxacin group and 33.3% (one of three patients) for the combination group (it was not statistically significant), with one breakthrough streptococcal bacteremia observed among the ciprofloxacin-treated patients. Adverse events were mostly self-limited and were observed in 27 (20.6%) ciprofloxacin-treated patients and 26 (19.7%) patients who were receiving the combination. This study demonstrates that high-dose ciprofloxacin given intravenously for at least 3 days and then by the oral route is therapeutically equivalent to the routine regimen of intraveneous ceftazidime plus amikacin even in febrile patients with severe neutropenia (polymorphonuclear leukocyte count, <100 mm3). However, it is very important that before an empirical therapy is chosen each hospital determine bacteriologic predominance and perform resistance surveillance.

Cancer patients who become deeply neutropenic as a result of intensive myelosuppressive chemotherapy are at high risk of developing life-threatening infections, and unless they are treated at the first sign of infection, the rate of mortality is high (2,16). Combinations of antibiotics, for example, an antipseudomonal beta-lactam plus an aminoglycoside, have been preferred as they may provide broad coverage, have high levels of bactericidal activity, and have potential synergistic effects, and there is the possibility that they protect against the development of resistance (7,18,19,28). Predominantly, the combination of ceftazidime plus amikacin has been established as a standard regimen (10,17,23). However, aminoglycosides' nephro-and ototoxic potentials are well documented. To avoid the latter effect, the efficacy of antibiotic monotherapy such as therapy with ceftazidime, cefepime, and the carbapenems has been studied and demonstrated in several studies (4,(24)(25)(26)(27). As ciprofloxacin is a potent agent against gram-negative bac-teria including Pseudomonas aeruginosa, it was expected that it would become a potential candidate for single-agent therapy in febrile neutropenic patients. While the efficacy of ciprofloxacin in combination with various antibiotics of different classes was demonstrated in early clinical trials (3,13,14), only a few studies with ciprofloxacin as monotherapy have been reported (1,11,20). However the poor in vitro activity of ciprofloxacin against gram-positive cocci (12,29) has also been indicated in vivo by Meunier et al. (20), who reported a poor response against infections caused by gram-positive bacteria, especially streptococci, in febrile neutropenic patients. That event caused premature discontinuation of the trial in which intravenous (i.v.) ciprofloxacin at a low dose of 200 to 300 mg twice a day (b.i.d.) was less effective than piperacillin plus amikacin (20). The latter observation becomes more important when one considers the changing pattern of the prevalence of pathogens in favor of gram-positive microorganisms over the past 10 years (10,22,30). With the release of an i.v. dose of 400 mg given three times a day (t.i.d.), which is equivalent regarding bioavailability and serum concentrations in serum to 750 mg given orally (p.o.) b.i.d., renewed interest in ciprofloxacin monotherapy in febrile neutropenic patients has arisen. This study was designed to compare administration of ciprofloxacin, given initially at the higher i.v. dose (400 mg t.i.d.), followed by administration by the p.o. route (a 750-mg tablet given b.i.d.), with the standard combination regimen of ceftazidime plus amikacin as empiric treatment in patients with febrile neutropenia.

MATERIALS AND METHODS
Criteria for eligibility. The protocol design was based on the guidelines published by the Immunocompromised Host Society Consensus Panel (8) Neutropenic patients (age, Ͼ18 years) with underlying hematological malignancies (acute leukemia, non-Hodgkin's lymphoma) or aplastic anemia who became febrile were eligible to enter the study. Fever was indicated by a single reading of an oral temperature of Ͼ38.5°C or two different readings of 38°C within 12 h. Neutropenia was defined as an absolute granulocyte count of less than 500 polymorphonuclear leukocytes per mm 3 or a count of 1,000/mm 3 that was expected to fall rapidly within 24 to 48 h after chemotherapy to less than 500/mm 3 . Persistent neutropenia was indicated by a granulocyte count that was below 500/mm 3 and that remained low 5 to 7 more days after the initiation of antimicrobial therapy.
The trial was conducted in accordance with the Declaration of Helsinki and all applicable national requirements. The protocol was approved by the local ethical committee of each of the participating centers. All patients admitted into the study were informed of the investigative nature of the study and gave their written informed consent.
Patients were excluded from the trial if they (i) had a known hypersensitivity to one of the study drugs, (ii) had received any systemic antibiotic therapy within the previous 72 h (oral antibiotic prophylaxis, with the exception of quinolone prophylaxis, was allowed until the start of empiric therapy), (iii) were suspected of suffering from any infection in which anaerobes were implicated, (iv) were moribund with a high probability of dying within 48 h after enrollment, (v) were suffering from renal insufficiency (serum creatinine level, Ͼ3 mg/dl), (vi) had liver dysfunction (i.e., elevated bilirubin or serum glutamic oxalacetic transaminase-serum glutamic pyruvic transaminase levels [more than three times the upper limit of normal]), or (vii) were pregnant or nursing women. Patients were subsequently assigned to one of the two treatment groups in accordance with a randomization code generated at Bayer AG for each center.
Clinical and laboratory monitoring. Prior to the start of treatment, a complete history was obtained and physical examination, routine chest X rays, and extensive routine laboratory tests including a urine culture and two sets of blood cultures (from different venipunctures) were performed. Other cultures were taken as clinically indicated. Follow-up studies including hematological analysis, blood chemistry, coagulation tests, and urinalysis were performed as required by the study protocol. In case of persisting fever, blood cultures were repeated daily. The susceptibilities of all isolated infecting microorganisms were tested by the standard antibiotic disk technique by the modified Kirby-Bauer procedure by using the zone diameter interpretive standards and equivalent MICs recommended by the National Committee for Clinical Laboratory Standards (21).
Classification of febrile episodes. Febrile episodes were classified as microbiologically documented infections (subclassified into those with or without bacteremia), clinically documented infections, and fever of unknown origin.
Evaluation of response to therapy. Therapeutic response was evaluated 72 to 120 h after the onset of empiric therapy (early evaluation) and at the end of therapy (late evaluation). Treatment was considered successful if fever (temperature, Յ37.5°C) and clinical signs (whenever present) of infection were resolved without relapse for at least 7 days after the discontinuation of therapy and the infecting microorganism was eradicated (for microbiologically documented infections) without any change to the assigned antibiotic regimen or the addition of antibiotics to the assigned regimen. Treatment was classified as a failure if the (i) the patient died from either the presenting infection or another one; (ii) the implicated pathogen or fever, or both, persisted and the patient's clinical condition did not improve, requiring any modification to the initial antimicrobial regimen; (iii) a clinical or microbiological relapse occurred within 7 days after the discontinuation of therapy; or (iv) a superinfection was observed. Patients infected with microorganisms resistant to the study drug were not excluded from clinical evaluation. This was decided because of the indication "empirical therapy." Treatment was considered nonevaluable if the patient had a proven viral or fungal infection or if a major protocol violation occurred. Bacteriological responses were defined as follows: eradication, causative organism absent at end of therapy; relapse, causative organism absent at the end of therapy but reappearance of the same pathogen (as indicated by species and susceptibility testing) at a follow-up 7 days posttherapy; superinfection, appearance of a new infection at any site that was caused by another organism and that occurred either during therapy or within the week after the discontinuation of antimicrobials; persistence, causative organism present at the end of therapy.
Antimicrobial drug regimens. Patients initially received either ciprofloxacin at 400 mg i.v. t.i.d. over a period of 1 h or ceftazidime at (2 g i.v. over 10 min every 8 h) plus amikacin (15 mg/kg of body weight/day i.v. over 30 min divided into two doses). If after 72 h patients who were receiving ciprofloxacin responded successfully and patients were able to tolerate oral medication, parenteral cipro-floxacin therapy was switched to the p.o. route at a dosage of 750 mg b.i.d. Patients in the comparator group continued the i.v. regimen.
Blood ciprofloxacin levels. In a limited group of five patients ciprofloxacin levels in serum were measured after administration of the third dose of p.o. therapy at 0.5, 1, 2, 3, 4, 8, and 12 h postdosing. Excluded from the kinetic study were patients with renal insufficiency, vomiting, diarrhea, mucositis, and gastrointestinal bleeding. Ciprofloxacin levels were measured by the high-pressure liquid chromatography method.
Duration of therapy. The duration of therapy ranged from 7 to 14 days. Study patients with a successful response to therapy received the protocol medication for a minimum of 7 days, the last 5 days of which had to be without fever, unless a clinical deterioration, adverse reaction, or death occurred. For patients with an early recovery from neutropenia, however (granulocyte count, Ͼ1,000/mm 3 ), and a complete resolution of signs and symptoms of infection, even after 3 days of defervescence and a minimum duration of 5 days of therapy, antibiotic treatment could be stopped. If the patient showed neither improvement of clinical signs or symptoms nor a decrease in fever, therapy was discontinued after 72 h and other antimicrobials were instituted. Usually, imipenem was empirically prescribed, with the addition of vancomycin in case of signs of inflammation at the entry of central i.v. catheters or septic shock. Those patients who did not become afebrile after 72 h but who had clinical improvement and whose fevers responded continued therapy for 2 more days. However, medications were withdrawn if they were still febrile after day 5.
Toxicity. Nephrotoxicity was determined as an increase in the serum creatinine concentration Ն25% above the baseline concentration, provided that an increase of a magnitude of Ն0.5 mg/dl occurred. Ototoxicity was defined as a decline in inner ear function, either auditory or vestibular, without discernible physical causes. To document auditory toxicity, audiograms were performed at the discretion of the investigators. Hepatotoxicity was defined as a rise in aminotransferase (asparagine aminotransferase, alanine aminotransferase), bilirubin, and/or alkaline phosphatase levels three times or more above the baseline levels in the absence of other apparent causes (e.g., chemotherapy). Diarrhea was considered as two or more uniformed stools or one or more watery stools per day. The relationship of adverse events to the study drugs was defined as follows: probable, follows a reasonable temporal sequence from drug administration, or possible, follows a reasonable temporal sequence from drug administration but could have been produced by the patient's clinical state or other modes of therapy administered to the patient.
Sample size estimation and statistical plan. The primary objective of this study was to compare the clinical success rates of both study regimens. On the basis of previously published data (8), a success rate of 75% for treatment with ceftazidime plus amikacin and a success rate of 80% for treatment with ciprofloxacin were assumed. By using 10% as a clinically relevant difference in success rates (␣ ϭ 0.05, ␤ ϭ 80%), the sample size estimation resulted in the need for 108 valid patients in each treatment group to prove the hypothesis that high-dose ciprofloxacin therapy is not less effective than therapy with ceftazidime plus amikacin. The expected success rates assumed here were based on previous studies not performed according to the guidelines applied in this study.
The primary efficacy variable was clinical outcome (success and initial response without modification versus failure), and it was analyzed by a one-sided modification of the Mantel-Haenszel statistic method. A supportive analysis of the clinical outcome that consisted of a one-sided 95% confidence interval as well as a two-sided 90% confidence interval was performed. A secondary efficacy variable, bacteriological response (eradication versus persistence and superinfection), was analyzed exploratively in the same way that the primary efficacy variable was.

RESULTS
The study was started in May 1992 and was completed in May 1995. A total of 263 febrile neutropenic patients were randomized: 131 in the ciprofloxacin group and 132 in the ceftazidime plus amikacin group. In 79.1% of the patients acute leukemia was the underlying disease, another 17.9% of the patients suffered from high-grade malignant non-Hodgkin's lymphoma, and the remaining 3% of the patients suffered from aplastic anemia and other malignances ( Table  1). Demographic data for patients in both treatment groups were generally comparable. No difference in the rate of acute myelogenous leukemia or non-Hodgkin's lymphoma was found (P ϭ 0.177 and P ϭ 0.382, respectively). Infection was hospital acquired in 59% of the patients in the ciprofloxacin group and 60% of the patients in the ceftazidime plus amikacin group. Two hundred forty-six patients were valid for intent-to-treat (ITT) analysis (124 in the ciprofloxacin group and 122 in the ceftazidime plus amikacin group), while 231 of them were valid for per-protocol (PPR) analysis (118 in the ciprofloxacin group and 113 in the ceftazidime plus amikacin group). Among the patients not eligible for ITT analysis, 7 were randomized into the ciprofloxacin regimen and 10 were randomized into the ceftazidime plus amikacin regimen. The reasons for noneligibility for the patients in the two groups were temperature of Ͻ38°C (two and two patients, respectively), wrong diagnosis (one and four patients, respectively), and missing clinical evaluation (four and four patients, respectively). The reasons for the exclusion of 15 patients (valid for ITT analysis) from the PPR analysis for the patients in the two groups were as follows: antibacterial prophylaxis was not withdrawn (four and two patients, respectively), a granulocyte count of Ͻ500 cells/mm 3 was never fulfilled (one and five patients, respectively), concomitant i.v. antibiotics like vancomycin or teicoplanin were used (one and one patients, respectively), and ciprofloxacin treatment was underdosed (one patient). Four patients given ciprofloxacin and seven patients given ceftazidime plus amikacin had granulocyte counts between 500 and 1,000 cells/mm 3 at enrollment. A rapid fall below 500 cells/mm 3 (in 48 h) was seen in three patients in the ciprofloxacin group and two patients in the ceftazidime plus amikacin group, with the decrease remaining in all patients in the subsequent days.
Response to therapy. The response to therapy was first determined at 72 h after the onset of empiric therapy. Of the evaluable febrile patients at this time, 66 of 124 (53.2%) in the ciprofloxacin group and 70 of 122 (57.4%) in the ceftazidime plus amikacin group had successful clinical responses. Patients with bacteremia had a better initial response (16 of 31 [51.6%]) when they were treated with ciprofloxacin than when they were treated with the comparator drugs (11/29 [37.9%]). However, this difference was not statistically significant. An additional 17 patients (13.7%) in the ciprofloxacin group and 13 patients (10.7%) in the ceftazidime plus amikacin group who did not become afebrile after 72 h but who had either clinical improvement or who were at least clinically stable and for whom no change in empiric therapy was required became afebrile after another 48 h. Therefore, in total, at day 5 of therapy 66.9 and 68.1% of the patients in the two groups, respectively, were considered to have been successfully treated ( Table 2). It  should be pointed out that because therapy was initiated on an empirical basis, patients infected or colonized with organisms resistant to the study drug were not excluded from the clinical evaluation. Therefore, among the patients in the ciprofloxacin group, 13 patients with bacteremia caused by ciprofloxacinresistant microorganisms were clinical failures; however, 2 patients with urinary tract infections caused by organisms resistant to ciprofloxacin were clinical successes. Twelve microorganisms resistant to ceftazidime were isolated from patients in the group treated with ceftazidime plus amikacin, and 11 of the patients were clinical failures; 1 patient with bacteremia caused by Escherichia coli resistant to ceftazidime but susceptible to amikacin was a clinical success.
In the ITT analysis a successful clinical outcome was documented for 50.0% of the patients receiving the ciprofloxacin treatment, whereas a successful clinical outcome was documented for 50.8% of the patients receiving the ceftazidime plus amikacin treatment. The clinical response rate in the PPR analysis was 48.3% (57 of 118) for patients receiving ciprofloxacin and 49.6% (56 of 119) for patients receiving ceftazidime plus amikacin. Success rates for both treatment groups were further analyzed by documented type of infection (Table  3). Of the 75 patients (33.7%) with microbiologically documented infections, 60 (26.5%) presented with bacteremia which was caused in 55 patients by a single organism and in 5 patients by multiple organisms (Table 3). Among the patients with bacteremia caused by a single organism, 27 (49.1%) patients were infected with gram-positive organisms and 28 (50.9%) were infected with gram-negative organisms. Coagu-lase-negative staphylococci (n ϭ 15 patients) predominated, followed by E. coli (n ϭ 12), P. aeruginosa (n ϭ 10) and viridans group streptococci (n ϭ 5) (Table 3). For all documented cases of bacteremia, the success rates for the ciprofloxacin and ceftazidime plus amikacin regimens were similar (45.1 versus 44.8%). Ciprofloxacin seemed to be the more effective treatment for bacteremia caused by gram-positive organisms (35.7 versus 23% in the two treatment groups), whereas ceftazidime plus amikacin therapy seemed to be the more effective treatment for bacteremia caused by gram-negative organisms (66.7 versus 56.2%). However those differences were not statistically significant (P ϭ 0.22 and P ϭ 0.09, respectively). Seventy-three patients had nonbacteremic microbiologically documented infections (n ϭ 15) or a clinically documented infection (n ϭ 58). The rates of response to therapy by infection site were similar for both treatment regimens (Table 4).
One hundred thirteen febrile patients were classified as suffering from fever of unknown origin. The success rates observed for the ciprofloxacin and the combination therapies were comparable: 32 of 56 (57.1%) and 32 of 57 (56.1%), respectively. For the subgroups of patients with persistent neutropenia, 25 of 60 patients (41.6%) treated with ciprofloxacin improved, whereas 37 of 68 patients (54.4%) treated with ceftazidime plus amikacin improved (P was not significant).
A total of 75 causative organisms were isolated from 71 patients (37 in the ciprofloxacin group and 34 in the ceftazidime plus amikacin group); 60 of the organisms were isolated from blood. Of these 60 organisms, 35 (46.7%) were gram- Adverse events. All 263 patients included in the study were evaluable for safety, including 131 in the ciprofloxacin group and 132 in the ceftazidime plus amikacin group. The proportions of patients who experienced adverse events were similar in both treatment groups: 27 (20.6%) receiving ciprofloxacin and 26 (19.7%) receiving ceftazidime plus amikacin. Regarding single adverse events 8 were reported as probable and 13 were reported as possible for the ciprofloxacin group, whereas 6 were reported as probable and 20 were reported as possible for the comparator treatment group. The types of adverse reactions are described in detail in Table 6. Because of adverse events, p.o. ciprofloxacin was discontinued prematurely in one patient with pseudomembranous colitis and ceftazidime plus amikacin was discontinued in five patients.
Blood ciprofloxacin levels. Mean Ϯ standard deviation serum ciprofloxacin levels at 0.5, 1, 2, 4, 6, 8, and 12 h after the administration of ciprofloxacin at 750 mg p.o. were found to be

Mortality.
A total of 12 deaths (5%) were reported during the study: 7 in the ciprofloxacin group and 5 in the ceftazidime plus amikacin group. All patients who died had persisting profound neutropenia. Two of them responded to ciprofloxacin and became afebrile but died of cerebral hemorrhage. One patient with pneumonia that was not microbiologically documented died on day 3 with septic shock. Another two patients with pneumonia that did not respond to ciprofloxacin died 2 and 3 days, respectively, after their treatments were shifted to other antibiotic regimens. The first patient developed septic shock due to bacteremia caused by P. aeruginosa resistant to all protocol agents, and in the second patient pneumonia was complicated by adult respiratory distress syndrome. Finally, antibiotic treatments for two more patients who did not respond to ciprofloxacin were changed, but the patients died on days 13 and 19 after the beginning of treatment, respectively, with the patients never becoming afebrile. The documented causes of death were pulmonary infection in the one patient and renal and hepatic failure in the other one. Among the patients who were given ceftazidime plus amikacin, one died on day 2 because of cerebral hemorrhage. Four other nonresponders to therapy were given alternative empiric antibiotic regimens. The causes of death were severe pneumonia (not microbiologically documented) with respiratory failure (n ϭ 2), septic shock without microbiologically documented infection (n ϭ 1), and septic shock due to bacteremia caused by methicillin-resistant S. aureus (MRSA) (n ϭ 1) resistant to all protocol agents. In patients who died because of sepsis, alternative therapeutic regimens included imipenem (n ϭ 2) or vancomycin plus ceftazidime (n ϭ 1) or the combination of imipenem plus vancomycin (n ϭ 5).

DISCUSSION
Ciprofloxacin is very active against members of the family Enterobacteriaceae and P. aeruginosa, which are frequently implicated in infections in neutropenic hosts. However, its in vitro activity against MRSA and streptococci is minimal (9,12,29). Because of the changing pattern of infection in neutropenic patients over the past 10 years in favor of gram-positive cocci (10,22,30) and to fill the gaps in the antimicrobial activity of ciprofloxacin, in preliminary studies ciprofloxacin was given to febrile neutropenic patients at the conventional dose of 200 mg b.i.d., mostly combined with antibiotics active against grampositive cocci and rarely as monotherapy. Kelsey et al. (14) compared ciprofloxacin combined with benzylpenicillin to piperacillin plus netilmicin and reported total response rates of 66 and 65%, respectively, while a favorable response rate of 78% for patients receiving ciprofloxacin plus teicoplanin versus a favorable response rate of 49% for patients given piperacillin plus gentamicin was observed by the same investigators (13). Testing ciprofloxacin as monotherapy at a dose of 400 mg b.i.d., Johnson et al. (11) reported similar success rates for patients treated with ciprofloxacin (44%) and azlocillin plus netilmicin (48%). Among patients receiving ciprofloxacin at 200 mg i.v. daily, Bayston et al. (1) described successful outcomes for 71% of patients with treatment modification and successful outcomes for only 14% of patients without treatment modification; these rates were 64 and 28%, respectively, among patients receiving ceftazidime at 2 g t.i.d.
From the reported overall results it is evident that within the limited power of this study, monotherapy with high-dose ciprofloxacin i.v. and then ciprofloxacin p.o is as effective as the standard combination of ceftazidime plus amikacin (50 versus 50.8%). The findings of lower efficacies than those predicted in the initial power calculation (about 75%) should probably be attributed to the fact that success rates were taken from previous studies not based on Immunocompromised Host Society (8) guidelines applied in this study. According to the Immunocompromised Host Society, success was defined as a lasting return of body temperature from a fever to a normal level (Ͻ37.5°C) and a resolution of all signs and symptoms of infection without the addition of any other antimicrobial agent; any modification was considered a failure. It was not anticipated during the planning for the study that compliance with Immunocompromised Host Society guidelines would lead to lower success rates. It may be confusing because the wide ranges of results reported from various trials that indicate treatment success are different. This can be explained by the variety of definitions used to evaluate the therapeutic efficacy of empiric therapy, therefore demonstrating the necessity of clear and similar terms to render results from different studies comparable (5,17).
In the evaluation of the causative organisms, it was apparent that both regimens are more effective against bacteremias caused by gram-negative organisms than those caused by grampositive organisms (Tables 3 and 4), at least among patients from the two Greek tertiary-care hospitals which participated in this study. In the latter settings the following rates of resistance to ciprofloxacin and ceftazidime among the major gramnegative organisms isolated from 1992 to 1995 were observed: E. coli, 6 versus 3%; P. aeruginosa, 20 versus 29%; and methicillin-susceptible S. aureus and MRSA strains resistant to ciprofloxacin, 5 versus 95%. The numerical superiority of the combination therapy against bacteremia caused by gram-negative organisms (66.7 versus 56.2%) may be explained by a synergistic effect of the combination, but it should be pointed out that ciprofloxacin was at least as effective as ceftazidime plus amikacin against P. aeruginosa infections (71.4 versus 66.7%) ( Table 3). On the other hand, the fact that in this study treatment with ciprofloxacin resulted in a higher success rate against bacteremia caused by gram-positive cocci (35.7 versus 23.1%) may be attributed to its better efficacy against coagulase-negative staphylococci (57.1 versus 12.5%). However, the differences were not statistically significant.
In a clinical study conducted by the European Organization for the Research and Treatment of Cancer (EORTC), an over-  Table 3). The test for the null hypothesis that treatment with ciprofloxacin has a success rate more than 10% lower than that of treatment with ceftazidime plus amikacin could be rejected at the 5% level (P ϭ 0.485), but the lower limit of the 95% one-sided confidence interval was from Ϫ12.12 to infinity (due to different underlying test procedures). The 95% two-sided confidence interval goes from Ϫ14.13 to 12.49. Current U.S. Food and Drug Administration regulations require an equivalence delta of 20% for demonstration of noninferiority for success rates in the better of the two arms of less than 80%. Therefore, according to U.S. Food and Drug Administration criteria noninferiority was demonstrated in this study.
Recently, an effort has been made to replace parenteral antibiotics with p.o. therapy, at least for low-risk patients with febrile cancer and neutropenia (6). To our knowledge, this is the second study with febrile neutropenic patients reported in the literature in which ciprofloxacin was switched from the i.v. to the p.o. route with a successful outcome. In two studies (6,15), it was shown that in febrile low-risk hospitalized patients with cancer who had neutropenia that was expected to resolve within 10 days, oral empirical therapy with ciprofloxacin at 750 mg b.i.d. plus amoxicillin-clavulanate at 625 mg t.i.d. was as effective as i.v. therapy. In the present study, in contrast to the results of previously reported trials, more than 75% of the patients included in the study were suffering from acute leukemia, were mostly deeply neutropenic (granulocyte counts, Ͻ100 mm 3 ), and were expected to have protracted neutropenia. The fact that 82 of the 124 patients in the ciprofloxacin group could be switched from i.v. to p.o. therapy and had a response rate of 74.4% indicates that the rate of relapse after an initial response does not increase under p.o. therapy compared to that with continued standard i.v. therapy. On the other hand, serum drug levels after p.o. intake among patients who did not have mucositis were found to be in the expected range, but with rather delayed and prolonged peaks.
The death rates were similar (5%) in the two treatment groups either during treatment or at the follow-up period, which is a rather low percentage for similar high-risk populations of patients. The tolerability of both study regimens was good. The reported adverse events in three patients who were receiving p.o. ciprofloxacin necessitated discontinuation of drug in only one patient, who had pseudomembranous colitis (Table 6).
In this study the number of febrile patients who were successfully treated without modification with high-dose ciprofloxacin alone was comparable to the number of patients who were successfully treated with the standard antibiotic combination regimen. However, it is very important for the appro-priate selection of empirical treatment in the febrile neutropenic host that each hospital determine bacteriologic predominance and perform resistance surveillance. A monotherapy approach with ciprofloxacin may not be recommended in centers in which infections caused by gram-positive cocci are clearly dominant or in centers with significant numbers of infections caused by ciprofloxacin-resistant gram-negative organisms. There is no doubt that the standard combination of ceftazidime plus amikacin should also not be used in centers with significant numbers of infections caused by ceftazidime-resistant gram-negative organisms.