Survival from invasive fungal infections has generally been poor in neutropenic patients with acute leukemia. In many cases, this poor survival can be linked to delays in diagnosis. Obstacles to the rapid diagnosis of invasive fungal infection include difficulty in isolation of fungi in cultures, inability to perform biopsies or other invasive diagnostic procedures in patients with disorders of coagulation, and the nonavailability of reliable serologic tests [8]. Because of these difficulties, amphotericin B is frequently administered empirically to patients with neutropenia and persistent fever refractory to antibacterial therapy [3, 9, 10]. This approach, however, is limited by the toxicity of amphotericin B, especially in patients receiving other nephrotoxic drugs.

The problems associated with effective therapy of serious fungal infections in neutropenic patients with acute leukemia have been the stimulus for using antifungal drugs for prophylaxis. Unfortunately, prophylaxis with oral agents such as nystatin, clotrimazole, and ketoconazole has produced inconsistent results either due to lack of efficacy or poor patient compliance [11]. Similarly, prophylaxis with parenteral drugs like intravenous miconazole or amphotericin B has been used only on a limited basis because of concerns about toxicity and overall effectiveness [12, 13]. Thus, there is currently no uniformly accepted or proven approach for prevention of fungal infections in neutropenic patients with acute leukemia.

Fluconazole is a new triazole antifungal agent with activity against many common fungal pathogens causing infection in patients with acute leukemia [14]. Fluconazole has a favorable pharmacokinetic profile that includes a long serum half-life, making once-daily administration possible, more consistent absorption from the gastrointestinal tract than that of ketoconazole, excellent penetration into the cerebrospinal fluid, and elimination predominantly by renal mechanisms. Significant side effects related to fluconazole have been uncommon and occur less frequently than those associated with amphotericin B. Fluconazole is currently approved for treatment of Candida and cryptococcal infections [15, 16]. Studies in neutropenic animal models also suggest that fluconazole may be effective for prevention of Candida infection and for treatment of invasive aspergillosis when given at high doses [17, 18]. Similar studies in neutropenic bone marrow transplants found that prophylactic fluconazole prevents both systemic and superficial fungal infections [19]. For these reasons, we did a double-blind, placebo-controlled trial of prophylactic fluconazole in neutropenic patients undergoing chemotherapy for acute leukemia.

Methods

Top Methods Results Discussion Author & Article Info References

Patients

Patients were eligible for the study if they satisfied the following criteria: 1) undergoing chemotherapy for acute leukemia or the blast crisis of chronic myelogenous leukemia; 2) 13 years of age or older; 3) anticipated neutropenia of less than 500 neutrophils per mm³ for 7 or more days; 4) no clinical evidence of fungal infection at time of study entry; 5) no systemic antifungal therapy within the 2 weeks before randomization; and 6) no allergy to the imidazoles or azoles. Patients with moderate or severe liver disease (aspartate aminotransferase [AST], alanine aminotransferase [ALT], or alkaline phosphatase greater than five times the upper limit of normal or a total bilirubin greater than 43 µmol/L) and patients with renal insufficiency (creatinine clearance of less than 0.83 mL/s) were excluded from the study. Our trial involved 18 oncology centers. Informed consent approved by the institutional review board at each center was obtained from each patient.

Study Drugs

Eligible patients were randomly assigned to receive either prophylactic fluconazole or placebo in a double-blind fashion. The fluconazole or placebo was begun at the time of initiation of chemotherapy and administered in identically appearing capsules. A capsule contained 100 mg of fluconazole, and each patient received four capsules (400 mg) as a single daily dose. For patients unable to tolerate oral capsules, the fluconazole or placebo was administered intravenously. The intravenous dose of fluconazole was 200 mg every 12 hours and was infused over 1 hour. The daily dose of fluconazole was reduced in patients who developed renal failure [20].

Prophylaxis with the study drug was continued through the course of chemotherapy and neutropenia and until 7 days after the neutrophil count reached 1000 cells per mm³ or greater. The maximum duration of prophylaxis was 10 weeks. Prophylaxis was discontinued if one of the following events occurred: 1) development of a documented invasive fungal infection; 2) initiation of empiric systemic antifungal therapy [amphotericin B] for clinically suspected invasive fungal infection; 3) adverse side effects related to the study drug; or 4) patient's inability to continue in the study due to noncompliance or death. Patients who developed a documented superficial fungal infection could be treated with topical clotrimazole while continuing to receive the study drug.

Laboratory Procedures

Complete blood counts, prothrombin times, blood urea nitrogen levels, serum creatinine and electrolyte determinations, urinalyses, and liver function studies (AST, ALT, alkaline phosphatase, and total bilirubin) were obtained at the time of study entry, once or twice weekly during the study, and at the end of prophylaxis to assess patients for drug-related side effects. Patients were also examined at least twice weekly for clinical symptoms and signs of adverse effects related to the study drugs. A serum pregnancy test was done on all women of child-bearing potential before the study began.

Surveillance cultures of the nasopharynx, oropharynx, axillae, urine, perirectal area, and stool were done at the time of study entry, once weekly during the study, and at the end of prophylaxis to determine the presence of fungal colonization. Cultures of blood and other suspected sites of fungal infection were obtained during the study whenever a patient's clinical condition suggested the possibility of fungal infection. Amphotericin B therapy was initiated in accordance with previously established guidelines when a reasonable clinical suspicion of systemic fungal infection existed [3, 9, 10].

Definition of Fungal Colonization and Infection

Fungal colonization was defined as the presence of a fungus in one or more surveillance cultures in the absence of any clinical symptoms or signs of infection. Superficial fungal infections were diagnosed by the isolation of a fungus from the skin, oropharynx, or gastrointestinal tract in association with signs of inflammation, ulcerations, plaques, or exudates not explainable by other pathogens. Invasive fungal infections were diagnosed by the presence of fungus in the blood, pulmonary tissue or secretions, sinuses, soft tissues, or other organ structures in association with symptoms and signs of infection not explainable by other pathogens.

Data Collection and Statistical Analysis

Data required by the study protocol were collected and recorded in case report forms by the investigators at each oncology center. Barton and Polansky Associates independently reviewed all case report forms for accuracy and compliance with the protocol by comparing the case report forms with patients' medical and pharmacy records. The case report forms were then submitted to clinical research personnel at Pfizer Central Research for review and entry of data into computer programs. All reviews, classifications of infections, and data entry were done blindly before the statistical analyses were performed. Proven fungal infections were required to meet the definitions of infection established by the protocol and approved by the Federal Drug Administration before the study. There was no interim analysis.

All statistical tests were performed as two-tailed tests. The Fisher exact test was used to compare differences in proportions, whereas the equality of two distributions was compared by the Wilcoxon rank-sum test. Univariate comparisons of times to specific events were performed by using Kaplan-Meier estimates of survival distributions and the Gehan generalized Wilcoxon test [21]. The SAS procedure LIFETEST was used for these comparisons [22]. The Cochran-Mantel-Haenszel chi-square test was used to check the equality of mean scores of ordinal response variables adjusted for center effect. Except for one placebo patient who did not receive the study drug and one fluconazole patient with invasive fungal infection at baseline, all patients were included in the efficacy analysis (intent-to-treat analysis). Center by treatment interactions were tested by using the Breslow-Day test of homogeneity of odds ratio [23].

Results

Top Methods Results Discussion Author & Article Info References

Patient Characteristics

Two hundred fifty-seven patients were enrolled into the study. One patient randomized to the placebo group did not receive the study drug and was excluded from all analyses. The characteristics of the other 256 patients are summarized in Table 1. One hundred thirty-two patients received placebo, and 124 patients were given fluconazole. The two groups of patients were similar in terms of age, sex, underlying disease, and baseline fungal colonization. Because daily neutrophil counts were not done in all patients, the median duration of neutropenia in each study group could not be determined. The nadir of neutropenia, however, was similar in the placebo and fluconazole patients (Table 1). Only three patients in each group did not have neutrophil counts less than 500 per mm³ at some time during the study. The median number of on-study days was less in the placebo group (14 days) than in the fluconazole group (19 days) (P = 0.01) due to the more frequent and earlier removal of placebo patients from the study who required treatment with amphotericin B for suspected or documented invasive fungal infection. Thirty-nine (30%) of the placebo patients and 27 (22%) of the fluconazole patients were treated with clotrimazole. Prophylactic colony stimulating factors were not used.

Study Drug Administration

Except for one fluconazole patient who received only intravenous drug, all the placebo and fluconazole patients received oral study drug at some time during the study. The median duration of oral dosing was 16.0 days (range, 1 to 72 days) in the fluconazole group and 14.0 days (range, 1 to 52 days) in the placebo group (P = 0.15). Twenty-five of the 124 (20%) fluconazole patients and 14 of the 132 (11%) placebo patients also received study drug intravenously at some time during the study (P = 0.04). The median duration of intravenous dosing was 8.0 days (range, 1 to 35 days) in the fluconazole group and 3.5 days (range, 1 to 15 days) in the placebo group (P = 0.06).

Effect on Fungal Infection

The types of proven fungal infection and the causative organisms are summarized in Tables 2 and 3. Proven fungal infection occurred in 27 of 132 (21%) placebo patients but in only eleven of 123 (9%) fluconazole patients (9%) (95% CI, 3% to 20%; P = 0.02). Both superficial and invasive fungal infections were less frequent in the fluconazole patients Table 2, although relatively few invasive fungal infections occurred in either study group. Oropharyngeal candidiasis was the most common type of superficial fungal infection in both study groups (16 cases in the placebo group, 6 cases in the fluconazole group; CI, 1% to 14%; P = 0.05). Cutaneous fungal infection occurred in four placebo patients but in none of the fluconazole patients (P = 0.1). One fluconazole patient had a rectal infection caused by an unspecified yeast organism, and another fluconazole patient developed a Torulopsis glabrata infection of the oropharynx. Disseminated candidiasis was the most common invasive fungal infection (six cases in the placebo group, one case in the fluconazole group; CI, 0% to 8%; P = 0.1). Aspergillus infection of the lungs or sinuses occurred in two placebo patients and three fluconazole patients. Other invasive fungal infections in the placebo patients were Trichosporon beigelii fungemia, Fusarium species pneumonia, and a necrotizing fasciitis caused by a Rhizopus species. One fluconazole patient had a fungal pneumonia caused by an unspecified fungus that was diagnosed by histologic examination of autopsy tissue. The study center had no effect on the incidence of proven fungal infection.

Candida organisms were the most common pathogens in both study groups (see Table 3). Superficial or invasive fungal infection caused by Candida occurred in 22 of the 132 [17%] placebo patients but in only 7 of 123 (6%) fluconazole patients [CI, 3% to 19%; P = 0.005]. Two placebo patients and one fluconazole patient had a fungemia caused by Candida krusei. The time to onset of proven fungal infection was shorter in the placebo group compared with the fluconazole group (Figure 1); P = 0.002).

View larger version (19K): [in this window] [in a new window]   Figure 1. Time to development of fungal infection. Kaplan-Meier product-limit estimates of survival distributions of time to development of proven superficial or invasive fungal infection in the fluconazole (F) (———) and placebo (P) (——) patients. The difference in time of onset of proven fungal infection between fluconazole and placebo patients has a P value of 0.002.

Empiric antifungal therapy with amphotericin B was used frequently in both study groups and was the most common reason for discontinuation of the study drug and removal of patients from the study. Ninety-eight of 132 (74%) placebo patients and 79 of 123 (64%) fluconazole patients received empiric amphotericin B (CI, –1% to 21%; P = 0.1). The time to initiation of amphotericin B, however, was delayed by prophylactic fluconazole (Figure 2). The P value for the difference in time distributions of starting amphotericin B was 0.03 with corresponding median times of 14 days in the placebo group and 20 days in the fluconazole group. Because the frequent and early use of empiric amphotericin B may have obscured the chances of showing an effect of fluconazole on prevention of documented invasive fungal infections by decreasing the risk for infection, the subgroup of patients who did not receive empiric amphotericin B was examined for the probability of developing an invasive fungal infection by using Kaplan-Meier product limit estimates. As shown in Figure 3, the probability of an invasive fungal infection was lower in fluconazole patients compared with placebo patients among patients not receiving empiric amphotericin B for suspected fungal infection (P = 0.03).

View larger version (21K): [in this window] [in a new window]   Figure 2. Time to initiation of amphotericin B therapy. Kaplan-Meier product-limit estimates of survival distributions of time to initiation of empiric amphotericin B therapy in the fluconazole (F) (———) and placebo (P) (——) patients. The difference in time of initiation of empiric amphotericin B between fluconazole and placebo patients has a P value of 0.03.

View larger version (18K): [in this window] [in a new window]   Figure 3. Time to development of invasive fungal infection. Kaplan-Meier product-limit estimates of survival distributions of time to development of proven invasive fungal infection in fluconazole (F) (———) and placebo (P) (——) patients not receiving empiric amphotericin B for suspected fungal infection. The difference in time of onset of proven invasive fungal infection between fluconazole and placebo patients has a P value of 0.03.

Effect on Fungal Colonization

The effects of prophylactic fluconazole on fungal colonization are shown in Table 4. Fungal colonization increased in placebo patients but decreased in fluconazole patients. Eighty-three of 122 (68%) placebo patients and 34 of 119 (29%) fluconazole patients were colonized with fungus at the end of prophylaxis (CI, 28% to 51%; P < 0.001). Fluconazole was especially effective in eliminating colonization with Candida albicans (53 of 122 [43%] placebo patients colonized compared with 10 of 119 [8%] fluconazole patients; CI, 25% to 45%; P = < 0.001). Fluconazole also decreased colonization by other Candida species except for Candida krusei. There was no increase in colonization with Candida krusei, Torulopsis glabrata, or Aspergillus species associated with prophylactic fluconazole.

Side Effects

Adverse clinical events and laboratory abnormalities considered by the blinded investigators to be related to the study drug were similar in the placebo and fluconazole patients. Nineteen of 132 (14%) placebo patients and 18 of 124 (15%) fluconazole patients experienced clinical side effects classified as drug-related (CI, –9% to 9%; P > 0.2). Similarly, 25 (19%) placebo patients and 22 (18%) fluconazole patients had abnormal laboratory values related to the study drug (CI, –8% to 11%; P > 0.2). Four fluconazole patients and five placebo patients were removed from the study due to drug-related side effects. The most common clinical adverse effects in placebo and fluconazole patients, respectively, were skin rash (13 patients compared with 7 patients), nausea and vomiting (five patients compared with nine patients), headaches and dizziness (one patient compared with four patients), pruritus (three patients compared with one patient), and diarrhea (three patients compared with one patient). The most common abnormal laboratory values in placebo and fluconazole patients were, respectively, elevations of total bilirubin (10 patients compared with nine patients), AST (four patients compared with one patient), ALT (seven patients compared with six patients), alkaline phosphatase (six patients compared with five patients), and creatinine or blood urea nitrogen (two patients compared with two patients).

Except for abnormal elevations of ALT, clinical side effects and abnormalities in laboratory tests considered unrelated to the study drug also occurred with similar frequency among placebo and fluconazole patients. Twenty-five of 113 (22%) fluconazole patients but only 14 of 120 (12%) placebo patients had abnormal elevations of ALT considered unrelated or related to the study drug (CI, –20% to –1%;P = 0.04). Treatment groups did not differ in the pattern of neutrophil decline or recovery, the severity of thrombocytopenia, or the incidence of eosinophilia.

Survival

Only five patients died while taking the study drug (one fluconazole patient; four placebo patients). The fluconazole patient died from gram-negative septicemia after 14 days of prophylaxis. One placebo patient also died from septicemia, whereas three other placebo patients died from intracranial hemorrhage.

Patients were followed for up to 90 days after discontinuation of prophylaxis. Twenty-six of 124 (21%) fluconazole patients and 24 of 132 (18%) placebo patients died either during prophylaxis or within 90 days after the end of prophylaxis (CI, –13% to 7%; P > 0.2). Although the overall mortality was similar in the fluconazole and placebo groups, patients with a proven invasive fungal infection had an increased risk for death. Seven of 15 (47%) fluconazole or placebo patients with proven invasive fungal infection died compared with 43 of 240 (18%) patients without invasive fungal infection (CI, 3% to 55%; P = 0.01). No deaths were attributed to the study drug.

Discussion

Top Methods Results Discussion Author & Article Info References

The high rates of morbidity and mortality of invasive fungal infections in neutropenic patients undergoing chemotherapy for acute leukemia have made effective antifungal prophylaxis a reasonable and worthy goal. Nevertheless, despite many previous studies in neutropenic patients, no general consensus has been reached on what regimen, if any, should be used for prophylaxis [11, 24]. The oral polyenes (nystatin or amphotericin B) and older imidazoles (miconazole, ketoconazole, and clotrimazole) have been studied most often. Diverse results emerged from these studies, and no consistent benefit could be demonstrated for any particular drug. Many of the trials involved small numbers of patients, were not adequately controlled, and used inadequate doses of the study drug. Problems associated with the use of the polyenes and the older imidazoles for antifungal prophylaxis include poor patient compliance because of gastrointestinal toxicity (oral nystatin and oral amphotericin), poor absorption from the gastrointestinal tract and lack of effect on invasive fungal infections (topical clotrimazole), inconsistent serum concentrations in patients being treated with antacids or cimetidine (oral ketoconazole), and a high incidence of adverse effects with parenteral administration (intravenous miconazole) [11, 24, 25].

In our trial involving 257 patients, fluconazole decreased both fungal colonization and infection in neutropenic patients with acute leukemia (see Tables 2, 3, and 4). These effects were most pronounced for colonization and superficial infections caused by Candida organisms except for Candida krusei, which caused infrequent colonization and infection in both study groups. The time to onset of proven fungal infection and the use of empiric amphotericin B were also delayed in patients receiving fluconazole (see Figures 1 and 2). Similar results were recently found in a large trial of prophylactic fluconazole (400 mg orally once daily or 200 mg intravenously every 12 hours) in patients neutropenic after bone marrow transplantation [19]. In that study, both invasive fungal infections (28 in 177 [16%] placebo patients compared with 5 in 179 [3%] fluconazole patients; P < 0.001) and superficial fungal infections (59 in 177 [33%] placebo patients compared with 15 in 179 [8%] fluconazole patients P < 0.001) were reduced. Perhaps due to a smaller number of patients, different risk factors, and the low overall incidence of invasive fungal infection, this study in patients with acute leukemia showed a reduction primarily in superficial but not invasive fungal infections. Oral fluconazole (50 mg per day) has also been found to be an effective agent for prophylaxis of superficial oropharyngeal candidiasis in patients with cancer but without neutropenia [26].

In this study, as well as in the bone marrow transplant trial [19], the number of Aspergillus infections in either the placebo or fluconazole patients was small, despite the large number of patients evaluated at different oncology centers throughout the United States (see Table 3). The reasons for this low incidence of Aspergillus infection in a highly susceptible population are not entirely clear. Aspergillosis is mainly an airborne disease, and patients may develop clinical disease as the result of either exogenous contamination, particularly during epidemics, or from previous endogenous colonization of the respiratory tract [5, 7, 27, 28]. Isolation of patients in units equipped with high-efficiency particulate air (HEPA) filters or in laminar air-flow rooms has been advocated to prevent Aspergillus infection. Laminar air-flow rooms were not used for patients participating in this study, and some but not all the study centers hospitalized patients in rooms with HEPA filters. On the other hand, the early and frequent use of empiric amphotericin B may have affected the incidence of Aspergillus infection. Prolonged neutropenia is a significant risk factor for aspergillosis, and few patients develop infection early in the course of neutropenia compared with patients with neutropenia lasting 3 weeks or longer [29]. Thus, the removal of more than 60% of the patients from the study by day 20 of the study period due to initiation of empiric amphotericin B for suspected fungal infection (see Figure 2) may have lessened the risk for Aspergillus infection. Some cases of Aspergillus infection may have also been overlooked in patients not able to undergo the necessary diagnostic procedures required by the study protocol to confirm the diagnosis. Nevertheless, the 2.3% incidence of Aspergillus infection in this study [6 infections in 256 patients] is actually similar to the 2.6% incidence of Aspergillus infection found in postmortem examinations at one large cancer hospital [30].

The frequent use of empiric amphotericin B for suspected fungal infection in this study may have been related to several factors and could have decreased the likelihood of demonstrating a substantial effect of fluconazole on prevention of invasive fungal infections. Because early diagnosis of invasive fungal infection in neutropenic patients is often difficult and unreliable, empiric amphotericin B is now considered standard treatment for neutropenic patients who are persistently or recurrently febrile despite appropriate antibacterial agents [3, 9, 10, 31]. Several trials have shown that empiric amphotericin B therapy reduces the incidence and mortality of invasive fungal infections [9, 10, 31]. Because the physicians managing the patients in this trial were blinded to the patient's prophylactic regimen (placebo or fluconazole), there may have been a greater tendency to initiate amphotericin B due to concern that patients were receiving a placebo. Similarly, the finding of fungi in the surveillance cultures required by the study may have influenced physicians' judgment about the use of amphotericin B in a persistently febrile patient. Despite these competing variables, we still found that empiric amphotericin B was used earlier in the placebo patients compared with the fluconazole patients (see Figure 2) and that the probability of invasive fungal infection was lower in fluconazole patients compared with placebo patients among a subgroup of patients not given amphotericin B for suspected fungal infection (see Figure 3). Until a reliable and sensitive test for rapid diagnosis of invasive fungal infection becomes available for routine clinical use, any trial of antifungal prophylaxis in the neutropenic patient will need to accommodate the common practice of using empiric amphotericin B.

One of the concerns about the use of any antimicrobial agent for prophylaxis is the emergence of organisms resistant to the agent used for prophylaxis. Despite questions about the standardization and reliability of susceptibility testing of fungi to antifungal drugs [32], certain fungi are inhibited in vitro only by relatively high concentrations of fluconazole. These fungi include Candida krusei and Aspergillus species [15]. We did not observe any increase in the incidence of colonization or infection by Candida krusei or Aspergillus species in patients receiving prophylactic fluconazole (see Tables 3 and 4). Similar results were obtained in the prospective, randomized trial of prophylactic fluconazole in 357 patients with bone marrow transplants [19]. Nonetheless, one fluconazole patient did develop Candida krusei fungemia, and three fluconazole patients had invasive Aspergillus infection. Candida krusei colonization and infection in patients given prophylactic fluconazole have also been noted in one retrospective, nonrandomized study at a single oncology center [33] as well as in several case reports [34, 35]. Thus, although an increase in the incidence of fluconazole-resistant fungal infections in patients treated prophylactically with fluconazole is a concern, the magnitude of this potential problem remains to be determined when the drug is used more extensively. Because Candida krusei and Aspergillus infections were increasing in immunocompromised patients even before fluconazole became available [5-7, 36], it is possible that elimination of other fungal infections by prophylactic fluconazole may only leave these fungi as persistent pathogens without necessarily increasing their incidence.

Prophylactic fluconazole at a dose of 400 mg per day was well tolerated. Most patients took all the drug orally, and only 20% of the fluconazole patients (25 of 124) and 11% of the placebo patients (14 of 132) required intravenous study drug (P = 0.04). Nausea, headaches, skin rash, abdominal pain, vomiting, and diarrhea have been the most frequent adverse effects associated with fluconazole [14, 15]. In this study, the incidence of these side effects was similar in the fluconazole and placebo groups. Laboratory abnormalities related to study drug, including liver and renal function tests, were also similar in the fluconazole and placebo patients. On the other hand, elevations in ALT considered unrelated or related to the study drug occurred more often in fluconazole patients (25 of 113 [22%]) than in placebo patients (14 of 120 [12%]) (CI, –20% to –1%;P = 0.04). Evaluation of liver function tests in patients with acute leukemia is difficult due to the competing effects of chemotherapy, other drugs, blood transfusions, and the leukemia itself on hepatic function. Nevertheless, hepatotoxicity due to azoles is a concern [37], and liver function tests should be closely monitored in patients receiving prophylactic fluconazole.

The usual dose of fluconazole for treatment of fungal infections is 100 to 400 mg per day. We chose the higher dose of 400 mg per day for prophylaxis because it was anticipated that higher doses of fluconazole may have some effect on Aspergillus as well as Candida infections [18]. The low incidence of Aspergillus infections in this study precluded any conclusions about the effectiveness of prophylactic fluconazole for prevention of Aspergillus infections. Whether a lower dose of fluconazole (100 to 200 mg per day) would be as effective as the higher dose (400 mg per day) is unknown and requires further study in controlled clinical trials.

Despite a reduction in proven fungal infections in the fluconazole patients, the overall mortality in patients followed for up to 90 days after discontinuation of prophylaxis was similar in the fluconazole and placebo patients (26 of 124 [21%] fluconazole patients compared with 24 of 132 [18%] placebo patients; CI, –13% to 7%; P > 0.2). This result is not surprising because overall mortality in patients with acute leukemia is greatly determined by prognostic factors associated with the patient's underlying disease and its responsiveness to chemotherapy in addition to the occurrence of infection [38, 39]. Nonetheless, patients without proven invasive fungal infection in this study had a lower risk for death (43 of 240 [18%] patients without infection died compared with 7 of 15 [47%] patients with infection; CI, 3% to 55%; P = 0.01). In the trial of prophylactic fluconazole in patients neutropenic after bone marrow transplantation, mortality caused by invasive fungal infection was reduced in patients receiving prophylactic fluconazole compared with patients given a placebo (1 of 179 [0.5%] patients compared with 10 of 177 [6%] patients; P < 0.001) [19]. These lower mortality rates in patients without invasive fungal infection support the concept of prophylaxis over treatment of established infection for reduction of deaths from fungal infection in neutropenic patients.

Prophylactic fluconazole decreased fungal colonization and infection in neutropenic patients undergoing chemotherapy for acute leukemia. These effects were mostly evident for colonization and superficial infections caused by Candida organisms other than Candida krusei. Fluconazole could not be clearly shown to be effective for preventing invasive fungal infections, reducing the use of amphotericin B, or decreasing mortality. The fluconazole was well tolerated and not associated with any significant adverse effects.

Participating Centers and Investigators

UCLA Center for the Health Sciences, Los Angeles, California. Principal investigator: Drew J. Winston, MD, Coinvestigators: Winston G. Ho, MD, and Richard E. Champlin, MD, Study Coordinator: Nancy Kaufman, RN. Harper Hospital, Wayne State University, Detroit, Michigan. Principal investigator: Pranatharthi H. Chandrasekar, MD, Study Coordinator: Cynthia M. Gatny, RN. University Hospitals of Cleveland, Case Western Reserve University, Cleveland, Ohio. Principal investigator: Hillard M. Lazarus, MD, Coinvestigator: Richard J. Creger, PharmD, Study Coordinator: Diana Wilson. University of Minnesota Hospital and Clinics, Minneapolis, Minnesota. Principal investigator: Jesse L. Goodman, MD, Coinvestigators: Daniel Weisdorf, MD, David Hermann, MD, Study Coordinators: Barbara Kringstad, RN, Jeanne Harkness, RN, Study Monitor: Ann Erdtman, Pharmacy: Deborah Curtis, Darlette Luke, PharmD. Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania. Principal investigator: George Talbot, MD, Jeffrey L. Silber, MD, Coinvestigators: Peter Cassileth, MD, Edward Stadtmauer, MD, Jean Scholtz, PharmD, Jeffrey Ball, PharmD, Amy Graziani, PharmD. Westchester County Medical Center, New York Medical College, Valhalla, New York. Principal investigator: Harold Horowitz, MD, Coinvestigators: Eric Feldman, MD, Zalmen Arlin, MD, Tauseef Ahmed, MD, Gary P. Wormser, MD, Study Coordinator: Gilda Forseter, RN. Western Pennsylvania Hospital, Pittsburgh, Pennsylvania. Principal investigator: Richard K. Shadduck, MD, Coinvestigator: Craig S. Rosenfeld, MD. Johns Hopkins Medical Institutions, Baltimore, Maryland. Principal investigator: Judith E. Karp, MD. Vanderbilt University Medical Center, Nashville, Tennessee. Principal investigator: Richard S. Stein, MD and Steven N. Wolf, MD. University of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma. Principal investigator: Ronald A. Greenfield, MD. St. Joseph's Cancer Institute, Florida. Principal investigator: Cameron Tebbi, MD. University of Texas Health Sciences Center, San Antonio, Texas. Principal investigator: David H. Boldt, MD. Thomas Jefferson University, Jefferson Medical College, Philadelphia, Pennsylvania. Principal investigator: Mark K. Sachs, MD. University of Texas Medical Branch, Galveston, Texas. Principal investigator: Walter H. Harvey, DO. University of Michigan Medical Center, Ann Arbor, Michigan. Principal investigator: Samuel Silver, MD. Veterans Affairs Medical Center, Gainesville, Florida. Principal investigator: Bradley Bender, MD. Temple University Hospital, Philadelphia, Pennsylvania. Principal investigator: Kenneth F. Mangan, MD. Barnes Hospital, Washington University School of Medicine, St. Louis, Missouri. Principal investigator: William Powderly, MD. Pfizer Central Research, Groton, Connecticut. Principal investigator: Donald N. Buell, MD, Statistician: Muhammed Z. Islam, PhD.