Recently, several studies in vitro and in animal models have suggested that aminoglycosides can be given effectively and safely once daily to achieve high peak concentrations associated with maximal bacterial killing and without an increase in toxicity [6, 7]. Small clinical studies have reported equivalent clinical results and minimal or even reduced toxicity when the same total daily dose of aminoglycosides was administered in single daily doses compared with more frequent administration [6].

The convenience and potential cost-effectiveness of once daily administration of antibiotics plus concern over the cumulative toxicity associated with the use of other nephrotoxic agents (such as amphotericin B, cisplatin, vancomycin, or acyclovir) in patients with cancer prompted this study, which is the first large-scale clinical application of this concept.

Amikacin was the chosen aminoglycoside because it has been used in our last five clinical trials without the development of resistance [8-12]. Also, we have reported our experience with single daily doses of amikacin in adults and children in two pilot studies [13, 14].

The pharmacokinetics and antibacterial activity of ceftriaxone make this drug attractive for combined single-dose therapy with amikacin. Ceftriaxone has a long half-life, and its spectrum of activity includes most gram-positive cocci, which are the predominant pathogens in patients with cancer and neutropenia [11, 15, 16], and gram-negative rods (with the exception of Pseudomonas species). However, Pseudomonas are now uncommonly isolated in patients with cancer and granulocytopenia, as shown in this and other studies [11, 16, 17].

Ceftazidime is traditionally administered thrice daily and has been used in combination with amikacin as the "standard" regimen in several of our previous studies [10, 11]. With the exception of Pseudomonas, we found that the activities of ceftriaxone and ceftazidime against 365 aerobic bacteria isolated from the blood of neutropenic patients enrolled in our previous trials were virtually superimposable. This trial compared single daily administration of amikacin and ceftriaxone with multiple daily dosing of amikacin and ceftazidime in 858 febrile episodes in granulocytopenic patients with cancer.

Materials and Methods

Participating Centers and Patient Eligibility

The International Antimicrobial Therapy Cooperative Group of the European Organization for Research and Treatment of Cancer (EORTC) consisted of 21 tertiary care or university medical centers located in Europe (n = 18), in the United States (n = 2), and in the Middle East [n = 1]. Patients with cancer were eligible for this study if they had 1) granulocytopenia [<1000 granulocytes/mm³]; 2) fever (defined as a temperature 38 °C [ 100.4 °F] on one occasion in the absence of an obvious noninfectious cause of fever, such as blood product, cytotoxic drug administration, or tumor fever); and 3) not received parenteral antibiotics for at least 4 days before randomization. Patients who were treated with any prophylactic antimicrobial agent were eligible for the study; however, antibacterial prophylaxis was discontinued with the institution of intravenous antibiotics. Patients could re-enter the protocol provided that they had not received antibiotics for at least 4 days and that the second randomization did not occur during the same episode of granulocytopenia, a standard practice in trials in these patients.

Patients were excluded if 1) they were allergic to any of the trial antibiotics; 2) they required hemodialysis or peritoneal dialysis or had a serum creatinine concentration greater than 300 µmol/L; or 3) they were pregnant or nursing.

The protocol was approved by the EORTC Protocol Review Committee (EORTC study number 46882) and by the Human Subjects or Ethics Committee of each participating institution. The trial was conducted in accordance with the Recommendations Guiding Medical Doctors in Biomedical Research Involving Human Subjects of the Declaration of Helsinki. Written or oral informed consent was obtained, depending on the requirements of the local Ethics Committee.

Randomization Procedure

Patients were randomly allocated by drawing consecutive sealed envelopes to one of the following two regimens: amikacin plus ceftazidime (control regimen) or amikacin plus ceftriaxone (experimental regimen). Randomization was stratified by institution and by underlying condition (leukemia, lymphoma, or bone marrow transplantation compared with solid tumors). Therefore, two randomization lists were established for each institution (one for patients with leukemia, lymphoma, and bone marrow transplantation and one for patients with solid tumors). We used random permuted blocks of six, so that equal numbers of patients would be randomized to each trial arm with every six patients entered.

Study Design, End Points, and Sample Size

The specific hypothesis tested in this prospective, randomized, multicenter study was whether once daily amikacin and ceftriaxone was as effective as multiple daily amikacin and ceftazidime for the empiric therapy of febrile patients with cancer and granulocytopenia. Secondary end points included the frequency of nephrotoxicity and ototoxicity.

To determine the sample size for the study we assumed that: 1) among evaluable patients the overall response rate to ceftazidime and amikacin would be 75%, as observed in previous studies by our group [10, 11]; 2) a difference in efficacy of less than 10% was necessary to conclude that both treatment groups had a similar efficacy; and 3) 85% of all randomized patients would be evaluable for response to treatment. To ensure with a probability of 90% (that is, ß = 0.10) that the upper 95% confidence interval (CI) (that is, = 0.05) for the true difference in response rate between the two treatment groups would not exceed 10%, 322 evaluable patients were needed in each study arm. Thus, assuming a 15% nonevaluability rate, 760 patient entries was required. Interim analysis was not done.

Antibiotic Treatment

The antibiotics were mixed in 50 mL of 5% dextrose in water and were sequentially infused intravenously (the ß-lactam was administered first). The infusion time was 15 minutes for each of the antibiotics, except for single-dose amikacin, which was infused in 60 minutes. The 60-minute infusion time for amikacin given every 24 hours was chosen because of limited experience with shorter infusion times when the study was designed. Ceftriaxone was administered once daily at a dose of 30 mg/kg in adults (the average adult received 2 g) and 80 mg/kg in children younger than 12 years. Ceftazidime was administered as 100 mg/kg per day in three divided doses. An average adult received 2 g of ceftazidime every 8 hours. Amikacin was administered as 20 mg/kg per day, given either once daily or in three divided doses. Maximal daily doses were 2 g for ceftriaxone, 6 g for ceftazidime, and 1.5 g for amikacin.

Twice-weekly monitoring of the serum concentrations of amikacin was recommended. The protocol suggested that peak (30 minutes after the end of infusion) and trough (preinfusion) serum amikacin concentrations of 26 to 32 µg/mL (8-hour group) and less than 10 µg/mL (for both 24-hour and 8-hour groups), respectively, should be achieved. No recommendation was made for peak serum levels of once daily amikacin, because no such data were available for patients with cancer and neutropenia when the study was designed. Monitoring of the serum levels of the ß-lactam antibiotics was not required. The minimum duration of therapy was 7 days, except in patients with noninfectious fever, in whom treatment was discontinued after 4 days.

Patient Evaluation and Classification of Febrile Episodes

All patients were evaluated before and during the trial according to previously published procedures [10, 11]. Case report forms were reviewed by the Data Manager and by the Data Review Committee as previously published [11].

Primary febrile episodes were classified as 1) microbiologically documented infections with or without bacteremia; 2) clinically documented infections; 3) unexplained fever [formerly classified as possible infections]; or 4) noninfectious fever (such as fever related to transfusion of blood products, chemotherapy, or the underlying cancer) according to definitions published previously [9, 11].

Bacteria isolated from blood were sent to the Microbiological Reference Center (Clinical Microbiology Laboratory, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland), where standardized identification methods and antibiotic susceptibility tests were done [18-20]. Evaluation of susceptibility data was done using the zone diameter interpretive standards and equivalent minimum inhibitory concentration recommended by the National Committee for Clinical Laboratory Standards: Break points for resistance were an inhibitory zone diameter of 14 mm or less for ceftazidime and amikacin and of 13 mm or less for ceftriaxone or as a minimum inhibitory concentration of 32 µg/mL or more for ceftazidime and amikacin and 64 µg/mL or more for ceftriaxone.

Evaluation of Response

A patient's trial was evaluated as a success if fever and clinical signs of infection (whenever present) resolved and if the infecting microorganism [whenever isolated] was eradicated without change of the allocated antibacterial therapy. The response had to be maintained for at least 4 days after discontinuation of therapy to qualify as a treatment success. In contrast, a patient's trial was evaluated as a failure if 1) the patient died of the primary infection; 2) a bacteremia persisted beyond the first 24 hours of therapy; 3) a breakthrough bacteremia was documented; 4) no response was seen to empiric therapy [that is, the pathogen or the fever persisted and the patient's clinical condition did not improve, requiring change of antibacterial therapy]; or 5) the primary infection relapsed within 4 days after discontinuation of therapy. The addition of any antiviral or antifungal agent without concomitant change in protocol antibacterial therapy was not considered as a treatment failure. A patient's trial was classified as not evaluable for response to protocol therapy if the patient had a viral, fungal, parasitic, or mycobacterial infection; if a major protocol violation occurred; or if a noninfectious cause of fever was documented.

Toxicity

All toxicities were evaluated as probably antibiotic related if noted in the absence of other toxic agents or predisposing conditions and as possibly antibiotic related in all other instances.

Nephrotoxicity was defined as an increase in serum creatinine of at least 45 µmol/L (baseline creatinine was measured before the onset of fever). Ototoxicity was defined as an alteration of inner-ear function, either vestibular or auditory, without a discernible physical cause. Testing of auditory function was available in 11 of 21 centers and included otoscopic examination (to exclude otitis or the presence of an earwax plug) followed by a pure-tone air conduction audiogram (pulsed sound). Audiograms were done by a technician or a physician in a specialized facility whenever possible. If a patient was unable to leave his or her room, audiometry was done in the patient's room with maximum reduction of ambient noise. Tested frequencies included 0.25, 0.5, 1, 2, 4, and 8 kHz. Audiograms were done before or within 24 hours of antibiotic initiation (baseline) and after therapy. Auditory toxicity was defined as an increase of 20 dB or more in auditory threshold at one or more frequencies in one or both ears. Patients in whom baseline audiograms were not available were excluded from the analysis, unless audiograms done during and after therapy or after therapy alone showed physiologic auditory activity (that is, within normal limits for the patient's age). In such cases, the patient trial was evaluated as showing no evidence of auditory toxicity. At the completion of the study, all audiograms were examined by one otorhinolaryngologist who was blinded with respect to the allocated antibiotic regimen.

Vestibular function testing was done at the bedside. The presence of one or more of the following symptoms or signs was considered evidence of vestibular dysfunction: nystagmus, vertigo accompanied by nausea and vomiting, gait disturbances or instability (Romberg and Unterberger tests), or past-pointing.

Hepatotoxicity was defined as an increase of transaminase, bilirubin, or alkaline phosphatase levels to 1.5 times above baseline values and normal ranges. Hypokalemia was defined as a decrease in the serum potassium level of 1.0 mmol/L or more (without concomitant supply of potassium) or of 0.5 mmol/L or more (with concomitant supply of potassium).

Further Infections and Death

Further infections were defined as those caused by a new organism not recognized as the initial infecting pathogen and occurring either during therapy or within 1 week after discontinuation of protocol antibiotics. Death was attributed to infection when it occurred as a direct consequence of either the presenting infection or a further infection.

Statistical Analysis

Statistical analyses were done using SPS PC V3.0 and BMDP [21, 22]. Inferential univariate analyses included chi-square tests or the Fisher exact test for the comparison of categoric variables with a continuity correction in the case of dichotomous variables. Mann-Whitney tests were used for the comparison of continuous variables and log-rank tests were used for the comparison of survival curves estimated by the method of Kaplan and Meier. Inferential multivariable analysis included logistic regression to estimate the probability distribution of a dichotomous variable (with a stepwise method for the selection of the explanatory variables). The McNemar test was used to compare the response rates of first and second febrile episodes in patients entered into the study more than once. All significance probabilities were calculated for two-tailed tests.

The protocol specified that patients could re-enter the study during separate episodes of granulocytopenia. For these patients, statistical analyses were conducted with the assumption that observations coming from multiple febrile episodes in the same patient were independent. To further address this point, the analyses were done with and without multiple entries.

Results

Top Results Discussion Author & Article Info References

From May 1988 to September 1990, 858 febrile episodes occurring in 677 patients were entered in the trial. A total of 164 episodes were not evaluable for response to therapy for the following reasons (amikacin plus ceftriaxone and amikacin plus ceftazidime, respectively): less than 7 days of therapy (15 compared with 20 episodes); documented nonbacterial infections (fungal infections, 9 compared with 14 episodes; viral infections, 6 compared with 3 episodes); tuberculosis (amikacin plus ceftazidime: 1 episode); noninfectious fever (15 compared with 17 episodes); unjustified discontinuation of protocol therapy (10 compared with 16 patients); continuation of antibacterial prophylaxis during trial (10 compared with 8 episodes); concomitant prescription of nonprotocol antibiotics (4 compared with 3 episodes); incorrect allocation of trial drugs (4 compared with 0 episodes); early discharge from the hospital before completion of protocol therapy (2 compared with 2 episodes); discontinuation of treatment because of toxicity (1 compared with 2 episodes); and patient's withdrawal from treatment (1 compared with 1 episode).

Of the 694 evaluable episodes, 350 were treated with amikacin plus ceftriaxone given once daily (24-hour group), and 344 were treated with amikacin plus ceftazidime given thrice daily (8-hour group). The detailed characteristics of all evaluable patient-episodes in each trial arm are described in Table 1. No statistically significant differences were noted between the two treatment groups in any of the parameters.

Documentation of Febrile Episodes

Microbiologically documented infections occurred in 205 (29.5%) episodes, of which 170 were bacteremias. These bacteremias were due to multiple organisms in 19 episodes and to single organisms in 151 episodes. Of the latter, 104 (69%) were caused by gram-positive bacteria and 47 (31%) by gram-negative bacteria. Of the 35 nonbacteremic infections, 24 were due to gram-negative bacteria and 11 to gram-positive bacteria. One hundred ninety-one infections (27.5%) were clinically documented, and 298 febrile episodes (43%) were classified as unexplained fevers.

Response Rates

The overall responses to each trial regimen were equivalent (Table 2). For all evaluable episodes, 249 of 350 (71%) patients in the 24-hour group and 256 of 344 (74%) patients in the 8-hour group were classified as treatment successes (difference, –3%;CI, –10% to 3%) (P > 0.2). No differences were observed between the two treatment groups in time to defervescence, time to failure, and reasons for failure (all P values >0.2). The response rates in each treatment group were also similar in the two randomization strata. For patients with leukemia, bone marrow transplantation, or lymphoma, success rates were 68% in the 24-hour group and 70% in the 8-hour group (difference, –2%;CI, –10% to 5%) (P > 0.2); for patients with solid tumors, success rates were 83% and 89%, respectively (difference, –6%;CI, –17% to 5%) (P > 0.2).

Responses for each category of infection are shown in Table 3. No statistically significant differences were seen between the two treatment groups in any subgroup of infection, even in patients with poor prognostic factors such as those with 100 granulocytes/mm³ or less throughout therapy (response rates, 43 of 107 patients [40%] in the 24-hour group compared with 44 of 90 patients [49%] in the 8-hour group; difference, –9%;CI, –23% to 5%; P > 0.2). The response rates in 170 bacteremic patients were 50% in the 24-hour group and 51% in the 8-hour group (difference, –1%;CI, –16% to 14%) (gram-positive bacteremia, 44% in both treatment groups; gram-negative bacteremia, 65% and 70%, respectively). In patients with single-organism bacteremia, the response rates for the most frequent isolates were 11 of 23 (48%) and 10 of 20 (50%), respectively, for viridans streptococci; 9 of 24 (38%) and 6 of 15 (40%), respectively, for coagulase-negative staphylococci; 3 of 5 (60%) and 11 of 15 (73%), respectively, for Escherichia coli; and 1 of 5 (20%) and 3 of 5 (60%), respectively, for Pseudomonas aeruginosa. None of the patients with P. aeruginosa bacteremias died of primary infection.

Susceptibilities to ceftriaxone, ceftazidime, and amikacin, respectively, were 100%, 89%, and 21% for viridans streptococci; 40%, 28%, and 79% for coagulase-negative staphylococci (40% of these strains were susceptible to oxacillin and 100% to vancomycin); 100% for all three antibiotics for E. coli; and 25%, 100%, and 90% for P. aeruginosa.

Sites of Infection

A site of infection was identified in 279 patient-episodes (40%). The distribution of infectious sites in the two treatment groups did not differ statistically and included (percentage of total; 24-hour group compared with 8-hour group) oral cavity and pharynx in 92 (33%, 46 compared with 46), lower respiratory tract in 78 (28%, 42 compared with 36), skin and soft tissue in 50 (18%, 25 compared with 25), and intravenous access site in 18 (6%, 13 compared with 5) patient-episodes. Other sites included urinary, upper respiratory, and gastrointestinal tracts in the remaining 41 patient-episodes (15%, 29 compared with 12).

Aminoglycoside Concentrations

Peak (30 minutes after the end of a 60-minute infusion) and trough (preinfusion) amikacin serum concentrations were measured at least once in 191 and 250 patients, respectively, in the 24-hour group and in 186 and 234 patients, respectively, in the 8-hour group. Median (mean ±SD) peak serum concentrations of amikacin were 45.6 µg/mL (48.3 ± 20.2) in the 24-hour group and 21 µg/mL (22.9 ± 13.8) in the 8-hour group (P < 0.001). Median (mean ±SD) trough concentrations were 0.9 µg/mL (2.1 ± 5.1) and 2.0 µg/mL (4 ± 6.1), respectively (P < 0.001).

Amikacin levels in serum were determined on multiple occasions (two to eight times) in 135 patients in each treatment group. These serial measurements did not show any statistically significant changes in the serum levels of amikacin over time in patients with stable serum creatinine concentrations.

Toxicity

Nephrotoxicity occurred in 12 of 351 patients (3%) evaluable for toxicity in the 24-hour group (11 adults and one child) and in 8 of 345 such patients (2%) in the 8-hour group (all adults) (difference, 1%; CI, –1% to 4%; P > 0.2). However, as seen in Table 4, nephrotoxicity did not develop (that is, serum creatinine concentrations were stable) until other nephrotoxic drugs (amphotericin B in 10 and 2 patients, respectively; glycopeptide antibiotics in 4 and 3 patients, respectively; and furosemide in 1 patient in each group) were used in 11 of the 12 patients (92%) in the 24-hour group and in 3 of the 8 patients (37.5%) in the 8-hour group (P = 0.02). Nephrotoxicity occurred later in the 24-hour group than in the 8-hour group (median, 10 days compared with 7 days; P = 0.048), and the median increase in serum creatinine was also lower (75 compared with 120 µmol/L, P = 0.06). Thus, in the 24-hour group, aminoglycoside-related nephrotoxicity was milder, delayed, and occurred almost exclusively after addition of other nephrotoxic agents. Renal dysfunction reversed in all evaluable patients.

Patients developing nephrotoxicity were older (median, 53 years; range, 16 to 74 years in the 24-hour group and 61.5 years; range, 41 to 76 years in the 8-hour group) and had a higher initial serum creatinine level (median, 79.5 µmol/L; range, 50 to 221 µmol/L in the 24-hour group; and 92 µmol/L; range, 60 to 240 µmol/L in the 8-hour group) than did patients without renal dysfunction (see Table 1 for comparison). In a logistic regression analysis, serum creatinine at study entry (P < 0.001) and patient age (P = 0.01) were covariates associated with nephrotoxicity. In the 8-hour treatment group, the odds ratio for nephrotoxicity by treatment adjusted for serum creatinine and patient age was 0.67 (CI, 0.27 to 1.70). Therefore, even after adjustment for the covariates associated with nephrotoxicity in the logistic regression model, neither a clinically nor a statistically significant difference in nephrotoxicity was found between the two treatment groups.

Audiometric testing was available in 11 centers and was successfully done in 144 patients (95 adults and 49 children). Compared with the entire study sample, the tested patients had fewer solid tumors as underlying diseases (14% compared with 24%, P = 0.01) and were somewhat younger (median age, 25 compared with 35 years; P = 0.09). Seventy patients were evaluated in the 24-hour group and 74 patients in the 8-hour group. These two groups of patients were similar with respect to age, sex, body weight, underlying diseases, presence of shock, type of infection, serum creatinine concentration at entry, and mortality. Auditory toxicity was documented in 6 of 70 patients (9%) in the 24-hour group (5 adults and one child) and in 5 of 74 patients (7%) in the 8-hour group (all adults) (difference, 2%; CI, –7% to 11%; P > 0.2). In all but three of these patients, the increase of 20 dB or more in the auditory threshold occurred at two or more adjacent frequencies. Hearing loss persisted in all patients who had subsequent audiograms. The two treatment groups did not differ statistically with respect to the severity of hearing loss, which was symptomatic in only one patient. One patient in the 8-hour group developed concomitant ototoxicity and nephrotoxicity. In both treatment groups, neither the peak nor the trough serum concentrations of amikacin were higher in patients with ototoxicity than in those without it. Minimal vestibular toxicity was reported in only one patient (24-hour group), who quickly recovered.

Other toxicities included hepatotoxicity (reported in 9% of the patients in the 24-hour group and in 11% of those in the 8-hour group), hypokalemia (8% and 9%, respectively), nausea and vomiting (3% and 4%, respectively), rash and urticaria (4% and 3%, respectively), and other side effects (1% in both groups).

Further Infections and Mortality

Further infections developed in 53 of 350 patients (15%) in the 24-hour group and in 40 of 344 patients (12%) in the 8-hour group (difference, 3%; CI, –2% to 9%; P > 0.2). These included 30 and 31 microbiologically documented infections, respectively, of which 22 were bloodstream infections in each of the two treatment regimens, and 23 and 9, respectively, were clinically documented infections.

The bloodstream isolates were (24-hour group compared with 8-hour group) coagulase-negative staphylococci (eight compared with five isolates), viridans streptococci (two compared with four isolates), Enterococcus species (five compared with two isolates), S. aureus (24-hour group: one isolate), Corynebacterium species (8-hour group: five isolates), other gram-positive bacteria (8-hour group: two isolates), Pseudomonas species (two isolates compared with one isolate), other gram-negative bacteria (four compared with two isolates), and Candida species (three compared with two isolates). In both treatment groups, most secondary infections were caused by pathogens resistant to the allocated antibiotics.

Further infections occurred at a median of 9 days (range, 2 to 27 days) after randomization in the 24-hour group and at 8 days (range, 2 to 16 days) in the 8-hour group (P > 0.2). In both groups, most infections developed when patients were profoundly granulocytopenic (granulocyte counts 100/mm³ in 76% and 78%, 101 to 500/mm³ in 11% and 10%, and >500/mm³ in 13% and 12%, respectively).

The overall mortality rate was 11% in both groups (difference, 0%; CI, –5% to 5%). Deaths occurred at a median of 15 days (range, 2 to 49 days) after study entry in the 24-hour group and at 17 days (range, 2 to 81 days) in the 8-hour group (P > 0.2). Causes of death included extensive cancer (10 and 15 cases, with concomitant infection in 5 and 8 cases, respectively), primary infection (7 cases in both groups), further infection (9 and 4 cases, respectively), hemorrhage (6 cases in both groups), and other causes (7 and 6 cases, respectively).

Discussion

Top Results Discussion Author & Article Info References

The results of this large clinical trial showed that a single daily dose of amikacin and ceftriaxone is as effective as and no more nephrotoxic or ototoxic than conventional therapy with multiple daily doses of amikacin and ceftazidime. The latter antibiotic combination is a standard regimen for the treatment of febrile patients with cancer and granulocytopenia, to whom antibiotics are usually administered in multiple daily doses based on the pharmacokinetics of the individual drugs used. Added to the many other drugs used to treat patients with cancer, these multiple administrations result in a large number of daily interventions through the patient's intravenous line, thus increasing the risk for complications. The demonstrated efficacy of a once daily regimen would allow more convenient and cost-effective therapy.

The rationale for high single-dose aminoglycoside therapy has been extensively reviewed recently [6, 7, 23]. Small studies of single daily dosing of aminoglycosides have been published in recent years [24-34]. In all but one of these reports [34], equivalent clinical outcomes and similar rates of oto- and nephrotoxicity were reported; however, most of these studies included a limited number of patients, thus entailing a risk for false-negative results.

Our study showed with sufficient power that both the 24-hour and 8-hour aminoglycoside-containing regimens had equivalent clinical and microbiologic efficacies (see Tables 2 and 3). The study included 205 episodes of microbiologically documented infections, of which 170 (83%) were bacteremias. In these documented bacterial infections, no difference in response was observed between single daily dose amikacin plus ceftriaxone and multiple daily dose amikacin plus ceftazidime. This number of bacteremias is several times greater than that reported in other studies of single daily dose aminoglycosides [25-28, 34, 35] and that noted in pivotal clinical trials with multiple daily doses of aminoglycosides [36-40]. Also, the frequency of microbiologically documented infections reported here is higher than that consistently reported during the last several decades in the myriad studies of patients with cancer and neutropenia. However, as in all clinical trials of empiric therapy in neutropenic patients, the specific infecting organism could not be identified in all patients. We recognize that unexplained fevers in these patients certainly respond to monotherapy with a cephalosporin alone, as we and others have shown [10, 15, 17, 41, 42].

The similarity in response rates for the 205 bacterial infections is unlikely to be attributed to the use of different cephalosporins in the two treatment groups. Indeed, the results of in-vitro susceptibility testing against the bacteria isolated in this trial did not show major differences between ceftriaxone and ceftazidime, confirming previous data obtained in 365 bacteremic isolates. Moreover, in view of the high serum concentrations achieved with both ceftriaxone and ceftazidime, any minor differences in in-vitro susceptibility are unlikely to be clinically relevant.

The use of a single daily dose of amikacin, which was infused in 60 minutes, resulted in higher peak serum concentrations compared with multiple daily administrations of the same total dose of amikacin, which was infused in 15 minutes (median, 45.6 compared with 21 µg/mL; P < 0.001). The serum levels of amikacin reported in this study confirm the results of previously published studies, both for the 8-hour group [11, 43] and for the 24-hour group [13, 14, 44]. Preliminary analysis of population pharmacokinetics using Nonlinear Mixed Effects Modeling suggests slightly greater daily amikacin exposure in patients in the 24-hour group (Strayer A. Unpublished data).

Despite higher serum levels of amikacin, the once daily regimen was not associated with an increase in toxicity (see Table 4). With more than 340 patients per treatment arm and a nephrotoxicity of 2% in the 8-hour group, this study had a power of 80% to detect at a 5% significance level an absolute increase of nephrotoxicity as small as 5% in the 24-hour group [two-tailed test]. The observed incidence of nephrotoxicity was 3% in the 24-hour group and 2% in the 8-hour group and did not differ statistically. Furthermore, nephrotoxicity occurred later in the 24-hour group and primarily only when other nephrotoxic agents had been added to the regimen. Also, the increases in serum creatinine concentration were smaller in the 24-hour group than in the 8-hour group. Similar to the present results, ter Braak and colleagues [27] reported the delayed onset of nephrotoxicity in 127 evaluable patients receiving netilmicin every 24 hours compared with those receiving the drug every 8 hours. In a recent study, a once daily dosing regimen of gentamicin was shown to be as efficacious as but less nephrotoxic than more frequent dosing [34].

Our study included the largest number of patients treated with single daily doses of aminoglycosides. Previous studies in granulocytopenic patients with cancer have reported incidences of aminoglycoside-related nephrotoxicity ranging from 0% to 22% [8, 10-12, 45-51]. Similar frequencies of nephrotoxicity have also been reported in clinical trials in non-neutropenic patients [34, 36-40]. The low incidence of nephrotoxicity in both treatment groups in our study may reflect the short duration of therapy (median, 8 days) and the inclusion of relatively young patients (median age, 28 years). A logistic regression analysis showed that age and elevated serum creatinine level at study entry were associated with the development of nephrotoxicity. This finding supports those of other studies on risk factors for nephrotoxicity in patients treated with aminoglycosides [5, 40]. Although the incidence of nephrotoxicity with aminoglycoside-containing combinations has been reported to vary with the ß-lactam antibiotic when a first-generation cephalosporin was included [47, 52], this finding has not been reported with second- or third-generation cephalosporins, and no reason exists to suspect that ceftazidime and ceftriaxone differ in their nephrotoxic potentials.

Audiometric examinations are difficult to do in clinical trials of antibiotic therapy. Such examinations are even more complicated in patients with cancer and granulocytopenia who are quite ill and often confined to their rooms. In this trial, simple audiometric testing was successfully done in 144 patients, who comprised only 21% of the study sample. Nonetheless, to our knowledge, this is the second largest number of audiometric examinations done in patients treated with aminoglycosides. Ototoxicity was 9% in the 24-hour group and 7% in the 8-hour group (difference, 2%; CI, –7% to 11%; P > 0.2). This value is equivalent to that reported in most studies of aminoglycoside-related ototoxicity [3640, 45, 46, 53, 54]. Although the universal standard for evaluating aminoglycoside-induced ototoxicity has not been established [55] and the tests used here were limited to a 0.25- to 8-kHz range, the results obtained did not differ statistically in the two patient groups. However, with the relatively small number of patients evaluated and with an ototoxicity of 7% in the 8-hour group, this study only has a power of 80% to detect at a 5% significance level an absolute difference of 18% or greater in ototoxicity between the two treatment groups (two-tailed test).

The choice of different cephalosporins in the two treatment arms allowed a realistic comparison of single compared with multiple daily administration of aminoglycoside-containing antibiotic combinations. Although this study focuses on the differences in aminoglycoside-related toxicity, it also suggests that single daily antibiotic administration is feasible and useful in granulocytopenic patients with cancer. As well as maintaining effective and safe treatment of bacterial infections in these patients, the once daily administration of antibiotics should reduce cost and patient inconvenience and also allow convenient intravenous therapy on an outpatient basis.

Presented in part (abstract number 1157) at the 31st Interscience Conference on Antimicrobial Agents and Chemotherapy, October 1991, Chicago, Illinois.

Appendix

The International Antimicrobial Therapy Cooperative Group of the European Organization for Research and Treatment of Cancer is composed of the following individuals:

Writing committee: Thierry Calandra, Stephen H. Zinner, Claudio Viscoli, Robrecht de Bock, Harold Gaya, Francoise Meunier, Jean Klastersky, and Michel P. Glauser

Data managers: Daniele Ninove and John Langenaeken, Institut Jules Bordet, Brussels, Belgium

Statistician: Marianne Paesmans, Institut Jules Bordet, Brussels, Belgium

Microbiology Reference Center: Marika Galazzo, Marlyse Giddey, and Jacques Bille, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland

Consultant: Andree Hadj-Djilani, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland

Appendix Principal Investigators and Participating Centers

L. Massimo, C. Moroni, E. Castagnola, Istituto G. Gaslini, Genova, Italy; M. Sanz, Hospital Universitario La Fe, Valencia, Spain; A. Ferster, Hopital Universitaire des Enfants, Brussels, Belgium; R. de Bock, Universitair Ziekenhuis Antwerpen, Edegem, Belgium; F. Meunier and J. Klastersky, Institut Jules Bordet, Brussels, Belgium; A. Padmos, King Faisal Specialist Hospital, Riyadh, Saudi Arabia; J. Gallagher, Geisinger Clinic-Cancer Center, Danville, Pennsylvania; A. Cometta, M. P. Glauser and T. Calandra, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland; A. Lopez, Hospital General Vall d'Hebron, Barcelona, Spain; A. Martinez-Dalmau, Hospital Xeral de Vigo, Spain; E. Pogliani, Ospedale San Gerardo, Monza, Italy; R. Hemmer, M. Dicato, F. Ries, Centre Hospitalier de Luxembourg, Luxemburg; A. Porcellini, Ospedale Civile, Pesaro, Italy; J. C. Legrand, Hopital Civil, Charleroi, Belgium; A. Porcellini, Centro Trapianti di Midollo Osseo, Cremona, Italy; J-M. Estavoyer, Centre Hospitalier Universitaire, Besancon, France; F. Follath, Kantonsspital, Basel, Switzerland; B. Seitanides, Metaxas Memorial Hospital, Pireaus, Greece; S. Zinner, M. Browne, Roger Williams General Hospital, Providence, Rhode Island; J. Nikoskelainen, Turku University Central Hospital, Turku, Finland; M. Rossi and G. Masera, Ospedale di Monza, Monza, Italy.