Preventing Fungal Infection in Neutropenic Patients with Acute Leukemia: Fluconazole Compared with Oral Amphotericin B

Methods

Eligible patients included consecutive adults who had acute leukemia, were hospitalized at participating centers, and were receiving cytotoxic therapy likely to induce neutropenia (neutrophil count < 1000/mm³) within 7 days. Patients received remission-induction or reinduction therapy according to the GIMEMA protocols [9, 10]. We excluded from the study before randomization patients younger than 14 years, patients with a history of hypersensitivity to triazoles, patients treated with antifungal therapy in the previous 15 days, patients with evidence of a preexisting systemic fungal infection, and patients who had nasal colonization with Aspergillus spp.

Study Protocol

After informed consent was obtained, the patients were randomly assigned to receive either fluconazole, 150 mg as a once-daily capsule, or amphotericin B suspension, 500 mg every 6 hours. Patients were randomly assigned to treatments using random permuted blocks of 10 containing different and balanced sequences of the two regimens. Antifungal prophylaxis was started 1 to 3 days before the administration of cytotoxic chemotherapy and continued until the neutrophil count returned to 1000/µL or a systemic fungal infection was proved or suspected. All patients received oral ciprofloxacin, 500 mg twice daily, as antibacterial prophylaxis [11]; antiviral prophylaxis and central venous catheters were used according to autonomous decisions made at each participating center. The patients were treated under conventional ward conditions or in single rooms, depending on the center. Prophylactic granulocyte transfusions and colony-stimulating factors were not used.

All patients were examined daily for clinical signs of fungal infection. When axillary temperature increased to more than 38 °C or infection was suspected, samples for microbiological cultures, including at least three separate blood specimens, were obtained, prophylactic therapy with ciprofloxacin was discontinued, and treatment with amikacin, ceftazidime, and a glycopeptide antibiotic (teicoplanin or vancomycin) was started; if fever persisted despite 4 to 6 days of systemic antibiotics, empiric intravenous amphotericin B was administered. Documented systemic fungal infections were treated with systemic antifungal agents (mainly intravenous amphotericin B), and superficial fungal infections were treated with topical antifungal agents.

To compare the efficacy and tolerability of the two prophylactic regimens, the following variables were measured: documented systemic fungal infection; suspected systemic fungal infection; superficial fungal infection; the interval to the development of documented systemic fungal infection or to the use of empiric antifungal therapy; compliance; treatment interruption caused by side effects; and mortality.

Definition of Fungal Infection

Superficial fungal infection was defined as clinically apparent infection of the oropharynx or skin, along with positive cultures; a suspected case of systemic fungal infection was defined as any episode of fever that persisted despite 4 to 6 days of empiric antibiotic therapy, for which empiric intravenous amphotericin B therapy was administered; definite systemic fungal infection was defined as one in which there was both clinical evidence of blood or tissue infection and a culture or biopsy specimen from the involved site showing a pathogenic fungal organism [7].

Compliance

Compliance was monitored by the nurse who counted capsules of fluconazole and measured the volume of amphotericin B oral suspension each day and recorded these data on the clinical report form. Compliance was defined as excellent if the patient took all the drug doses, as good if the patient missed fewer than three consecutive doses or took more than 80% of the total number of doses, and as poor if the patient missed more than three consecutive doses or took less than 80% of the total number of doses.

Statistical Analysis

Statistical analysis was done at the GIMEMA Infection Program Data Center with the SAS package (SAS Institute, Inc., Cary, North Carolina). Results are reported for all patients enrolled in the study (intention-to-treat analysis). Except for three patients randomly assigned to fluconazole and two patients assigned to amphotericin B who did not receive the study drugs and six additional patients in the fluconazole group and five in the amphotericin B group who had a duration of neutropenia of less than 7 days, all other patients were evaluable for the clinical efficacy analysis.

The chi-square test with a correction for continuity, or the Fisher exact test when appropriate, was used to compare differences in proportions between the two groups. The log-rank test was used to compare the Kaplan-Meier survival curves. The Student unpaired t-test was used to compare the means. Confidence intervals (CIs) of 95% are given where appropriate.

Results

A total of 820 patients with acute leukemia and neutropenic episodes from 30 centers were studied; 420 were randomly assigned to receive fluconazole, and 400 were randomly assigned to receive oral amphotericin B. The two groups of patients were similar in sex, age, underlying diseases, type of chemotherapy, protective environment, use of central venous catheters, and duration and severity of neutropenia. Patients receiving first-induction chemotherapy were equally distributed in the two treatment groups (Table 1).

Systemic Fungal Infection

Proven systemic fungal infection occurred in 11 (2.6%) fluconazole recipients and in 10 (2.5%) amphotericin B recipients (P > 0.2). The distribution of fungal isolates was similar in both groups (Table 2): Candida spp. caused 55% of systemic infections in fluconazole recipients and 70% in amphotericin B recipients; no difference was found in the isolation of different Candida spp., including C. krusei, between the two groups. Rates of infections caused by Aspergillus spp. were 45% in fluconazole recipients and 30% in amphotericin B recipients, a difference of 15 percentage points (95% CI for difference, −25% to 56%, P > 0.2), and the Aspergillus isolates were equally distributed. Fungemia caused by Candida spp. was documented in five patients receiving fluconazole and in three treated with amphotericin B. The characteristics of the patients with proven cases of systemic fungal infection and their clinical outcomes are summarized in Table 3.

Overall, the sites of infection between the two treatment groups were similar (P > 0.2). Simple fungemia caused by Candida isolates was documented in three patients in each group (two cases of C. krusei and one of C. parapsilosis in fluconazole recipients; one case each of C. albicans, C. krusei, and C. parapsilosis in amphotericin B recipients), and tissue infection was documented in three fluconazole recipients (C. tropicalis, C. albicans, and C. parapsilosis), and two amphotericin recipients (Candida spp., C. albicans). In patients receiving amphotericin B, esophagitis caused by Candida spp. and urinary tract infection caused by C. tropicalis were also documented.

Tissue infection caused by Aspergillus spp. occurred in five fluconazole recipients (four cases of pneumonia and one disseminated infection) and in three amphotericin B-treated patients (two cases of pneumonia and one case of disseminated infection).

Deaths from fungal infection were similar. Candida krusei fungemia and C. albicans and C. parapsilosis tissue infections caused death in three fluconazole recipients; C. albicans fungemia and Candida spp. tissue infection caused death in two amphotericin B recipients. Aspergillus pneumonia caused two deaths in the fluconazole group and one death in the amphotericin B group.

The interval to the documented systemic fungal infection was 21 days in fluconazole recipients and 15 days in amphotericin B recipients, a nonstatistically significant difference (95% CI for difference, −3 to 15 days; P = 0.15).

Superficial Fungal Infection

Superficial infections were reported in 7 of the 420 patients receiving fluconazole (1.7%) and in 11 of 400 of those receiving amphotericin B (2.7%), a difference of 1 percentage point (CI for difference, −0.9% to 3%; P > 0.2). The site of superficial infection was almost exclusively the oral cavity (17 of 18), and fungal isolates were usually C. albicans (13 of 18) in both groups (Table 4).

Suspected Systemic Fungal Infection

A lower rate of suspected systemic fungal infections requiring empiric intravenous antifungal therapy occurred in fluconazole recipients (67 of 420, 16%) compared with amphotericin B recipients (85 of 400, 21%), but this difference does not reach a statistically significant level (a difference of 5 percentage points, CI for difference, −0.02% to 10%; P = 0.07). The mean interval to the empiric antifungal therapy was 20 days (3 to 65 days) in fluconazole recipients and 18 days (1 to 41 days) in amphotericin B recipients, a difference of 2 days (CI for difference, −0.35 to 4.5 days; P = 0.10), and the duration of antifungal therapy was also similar: 13 days (3 to 37 days) and 12 days (1 to 49 days), respectively, a difference of 1 day (CI for difference, −0.7 to 3.91 days; P = 0.16).

Patients Receiving First-Induction Chemotherapy

A separate analysis of the subgroup of 417 patients with neutropenia receiving first-induction chemotherapy showed they did not differ from the other patients for demographic characteristics, use of protective environment, use of central venous catheters, and mean duration of neutropenia. Further, no difference was found in this subgroup of patients between the two treatment arms for any of the following variables: systemic infections, suspected fungal infections requiring the empiric use of intravenous amphotericin B, superficial infections, or deaths caused by fungal infection.

Mortality

The comparison of the survival curves showed no difference between the two treatment groups (P > 0.2). The overall mortality was similar: Forty-four (10%) of 420 fluconazole recipients died and 40 (10%) of 400 amphotericin B recipients died. Deaths were attributed to fungal infection in 5 fluconazole recipients and in 3 amphotericin B recipients. Nonfungal infective causes of death (22 in fluconazole recipients and 24 in amphotericin B recipients) and noninfective causes of death (17 in fluconazole recipients and 13 in amphotericin B recipients) were equally distributed in both groups.

Compliance and Adverse Reactions

Compliance differed between the two treatment groups (P < 0.01). Compliance was classified as excellent for 90% of fluconazole recipients compared with 72% of amphotericin B recipients, a difference of 18 percentage points (CI for difference, 13% to 23%); as good for 9% of fluconazole recipients and 22% of amphotericin B recipients, a difference of 13 percentage points (CI for difference, 8% to 18%); and as poor for 1% of fluconazole recipients and 6% of amphotericin B recipients, a difference of 5 percentage points (CI for difference, 2% to 7%).

Side effects were prevalent in amphotericin B recipients: Of 400 patients, 28 (7%) had gastrointestinal disturbances (nausea, vomiting) that required interruption of drug administration in 13 patients (3%). Fluconazole recipients had gastrointestinal disturbances (3 of 420, 0.7%), reversible increases in aminotransferase levels (2 of 420, 0.5%), and skin rash (1 of 420, 0.2%). No treatment with fluconazole had to be interrupted because of side effects. Overall, the difference in side effects was statistically significant favoring fluconazole (1.4% compared with 7%; a difference of 5.6 percentage points; CI for difference, 2% to 8%).

Discussion

Our study showed that antifungal prophylaxis with fluconazole in patients with acute leukemia was at least as effective as oral amphotericin B in preventing systemic and superficial fungal infections and the empiric use of intravenous amphotericin B. Similar results were obtained in two small studies done in 50 and 59 neutropenic patients with hematologic malignancies receiving oral fluconazole, 50 or 200 mg/d, and amphotericin B, 1200 or 1600 mg/d, respectively [12, 13]. Unexpectedly, the use of an antifungal agent with systemic activity seems not to offer any advantage over a nonabsorbable agent in reducing documented systemic fungal infection: This may have been because these infections have a gastrointestinal site of origin [14], because of poor bioavailability of fluconazole in neutropenic patients with chemotherapy-induced gastrointestinal mucositis [15], or because of the emergence of fungal strains resistant to fluconazole [16].

Using fluconazole in a once-daily dose of 150 mg in a larger sample, we observed the same rate of systemic fungal infection that occurred in the patients with acute leukemia (4%) described by Winston and colleagues [8] and in patients having bone marrow transplantation (3%) described by Goodman and colleagues [7]. In these studies, fluconazole was used at a daily dose of 400 mg. A recent European multicenter study using fluconazole in a daily dose of 50 mg in patients with hematologic malignancies (including patients with acute leukemia and bone marrow transplant recipients) showed a 3% rate of documented systemic fungal infection [17]. Our results raise some concern about the need for high doses of fluconazole (400 mg/d) for antifungal prophylaxis. The use of a lower dose of fluconazole (a single daily capsule of 150 mg) compared with 400 mg/d (four 100-mg capsules) seems to be equally effective and can save an average of $250 for the cost of the antifungal agent for each neutropenic episode.

The empiric use of intravenous amphotericin B in our study (16% of fluconazole recipients and 21% of oral amphotericin B recipients) compared favorably with the rate reported by Winston and colleagues in a similar group of patients (74% of placebo recipients and 64% of fluconazole recipients) [8]. A possible explanation for the lower use of amphotericin B in our patients is that precise guidelines were given about both empiric antibiotic therapy and time of starting intravenous amphotericin B, removing important variables that may influence this particular outcome measure.

Fluconazole and amphotericin B had similar efficacy in preventing superficial fungal infections; the 1.7% rate of superficial fungal infection observed in fluconazole recipients in our study is similar to or compares favorably with the rates seen in other studies in patients treated with fluconazole: 2% in patients with various hematologic malignancies [17], 6% in patients with acute leukemia [8], and 8.4% in marrow recipients [7].

The high rate of systemic (8%) and superficial fungal infection (15%) documented in placebo recipients with acute leukemia in other studies [8] compared unfavorably with our results in patients receiving amphotericin B suspension and suggests a caveat for the use of placebo in randomized, controlled studies of antifungal prophylaxis.

One source of concern about the use of any prophylactic antimicrobial agent is the potential for the selection or development of resistant organisms during therapy. The extensive prophylactic use of polyenes as antifungal agents in some units has been shown to potentiate the emergence of resistant strains of yeasts [18, 19] and, because amphotericin B remains the drug of choice for treatment of invasive fungal infections, this may represent a major disadvantage to the prophylactic use of polyenes. The emergence of C. krusei as the chief systemic pathogen in patients with marrow transplants has been reported by some centers and has been ascribed to the prophylactic use of fluconazole [16]. The lack of in vitro activity of fluconazole against C. krusei may contribute to this increased prevalence, although the prophylactic use of fluoroquinolones has also been proposed as an independent risk factor [16]. Although we have not done a systematic survey of fungal colonization during prophylaxis and our patients uniformly received prophylactic fluoroquinolones, we observed no excess of infection with C. krusei or any other Candida spp. among fluconazole recipients with respect to amphotericin B recipients. No large increase in C. krusei infections with fluconazole in the two randomized, placebo-controlled trials was observed [7, 8], and some centers reported no increase [20, 21]. In our study, Aspergillus spp. caused more than one third of all documented systemic fungal infections. Fluconazole lacks activity against Aspergillus spp.; the use of a relatively high dose (400 mg/d) in the hope of offering some protection against this resistant fungal organism gave inconclusive results [7, 8]. Although systemic amphotericin B is active against Aspergillus spp., oral nonabsorbable suspension seems to be unable to reduce Aspergillus infections; this is not surprising because nasal colonization seems to be the risk factor for disseminated aspergillosis.

The use of a systemic antifungal agent did not influence the rate of overall and fungus-related deaths, which was similar in the two groups of patients. This finding was not unexpected; mortality in neutropenic patients is influenced by several factors that are unrelated to the type of prophylaxis used, such as the response to antibiotic and antifungal therapy, the severity of underlying disease, and complications other than infection.

Compliance was better for fluconazole than for oral amphotericin B; this seems to be related to the lower rate of gastrointestinal side effects that occurred in fluconazole recipients, supporting the advantage of a once-daily capsule over an unpalatable suspension taken four times a day. However, no relation was found between compliance and occurrence of documented systemic fungal infection.

Although we were concerned about the potential liver toxicity of fluconazole, we found only two cases of reversible elevation in the alanine aminotransferase levels in fluconazole recipients.

Fluconazole was at least as effective as oral amphotericin B in preventing systemic and superficial fungal infection and the empiric use of amphotericin B in neutropenic patients with acute leukemia; oral amphotericin B is less expensive but fluconazole is better tolerated.

Appendix

The following are additional participants in the study: M. Montillo, MD, Istituto di Clinica Medica, Universita di Ancona; G. Barbabietola, MD, Istituto di Ematologia, Universita di Perugia; A. Dinota, MD, Divisione di Ematologia, Ospedale di Potenza; L. Pagano, MD, Istituto di Semeiotica Medica, Universita Cattolica, Roma; P. Jacopino, MD, Divisione di Ematologia, Ospedale di Reggio Calabria; N. Isabella, MD, Divisione di Ematologia, Ospedale Molinette di Torino; R. Fontana, MD, Sezione di Ematologia, Universita di Napoli; G. Landonio, MD, Divisione Falck, Ospedale Niguarda, Milano; G. Broccia, MD, Ospedale Oncologico Businco, Cagliari; L. Cudillo, MD, Divisione di Ematologia, Ospedale S. Eugenio, Roma; S. Di Fazio, MD, Cattedra di Ematologia, Universita di Catania; P. Galieni, MD, Divisione di Ematologia, Ospedale Sclavo, Siena; L. Pacilli, MD, Divisione di Ematologia, Ospedale S. Camillo, Roma; M. Monaco, MD, Divisione di Ematologia, Ospedale di Foggia; A.M. Carella, MD, Divisione di Ematologia, Ospedale S. Martino, Genova; A. Chierichini, MD, Divisione di Ematologia, Ospedale di Latina; G. Quintini, MD, Clinica Medica, Universita di Palermo; A. Gabbas, MD, Divisione di Ematologia, Ospedale di Nuoro; P. Citarella, MD, Cattedra di Ematologia, Universita di Palermo; A. Dallasta, MD, Divisione di Ematologia, Ospedale di Reggio Emilia; F. Ronconi, MD, Divisione di Ematologia, Ospedale di Avellino; F. Dore, MD, Istituto di Ematologia, Universita di Sassari.