AUTHORS: Michael S. Saag, M.D., William G. Powderly, M.D., Gretchen A. Cloud, M.S., Patrick Robinson, M.D., Michael H. Grieco, M.D., Patricia K. Sharkey, M.D., Sumner E. Thompson, M.D., Alan M. Sugar, M.D., Carmelita U. Tuazon, M.D., John F. Fisher, M.D., Newton Hyslop, M.D., Jeffrey M. Jacobson, M.D., et al.
Copyright: The NEW ENGLAND JOURNAL of MEDICINE
Intravenous amphotericin B, with or without flucytosine, is usually standard therapy for cryptococcal meningitis in patients with the acquired immunodeficiency syndrome (AIDS). Fluconazole, an oral triazole agent, represents a promising new approach to the treatment of cryptococcal disease.
In a randomized multicenter trial, we compared intravenous amphotericin B with oral fluconazole as primary therapy for AIDS-associated acute cryptococcal meningitis. Eligible patients, in all of whom the diagnosis had been confirmed by culture, were randomly assigned in a 2:1 ratio to receive either fluconazole (200 mg per day) or amphotericin B. Treatment was considered successful if the patient had had two consecutive negative cerebrospinal fluid cultures by the end of the 10-week treatment period.
Of the 194 eligible patients, 131 received fluconazole and 63 received amphotericin B (mean daily dose, 0.4 mg per kilogram of body weight in patients with successful treatment and 0.5 mg per kilogram in patients with treatment failure; P = 0.34). Treatment was successful in 25 of the 63 amphotericin B recipients (40 percent; 95 percent confidence interval, 26 percent to 53 percent) and in 44 of the 131 fluconazole recipients (34 percent; 95 percent confidence interval, 25 percent to 42 percent) (P = 0.40). There was no significant difference between the groups in overall mortality due to cryptococcosis (amphotericin vs. fluconazole, 9 of 63 [14 percent] vs. 24 of 131 [18 percent]; P = 0.48); however, mortality during the first two weeks of therapy was higher in the fluconazole group (15 percent vs. 8 percent; P = 0.25). The median length of time to the first negative cerebrospinal fluid culture was 42 days (95 percent confidence interval, 28 to 71) in the amphotericin B group and 64 days (95 percent confidence interval, 53 to 67) in the fluconazole group (P = 0.25). Multivariate analyses identified abnormal mental status (lethargy, somnolence, or obtundation) as the most important predictive factor of a high risk of death during therapy (P<0.0001).
Fluconazole is an effective alternative to amphotericin B as primary treatment of cryptococcal meningitis in patients with AIDS. Single-drug therapy with either drug is most effective in patients who are at low risk for treatment failure. The optimal therapy for patients at high risk remains to be determined. (N Engl J Med 1992; 326:83–9)
CRYPTOCOCCAL meningitis, the most common life-threatening opportunistic fungal disease in patients infected with the human immunodeficiency virus (HIV) type 1, occurs in 5 to 8 percent of patients with the acquired immunodeficiency syndrome (AIDS). With the widespread implementation of more effective antiretroviral therapy and prophylactic treatment regimens designed to prevent the development of common life-threatening opportunistic infections, such as Pneumocystis carinii pneumonia, the incidence of cryptococcal disease will probably increase in parallel with improvement in survival over the next decade. Therefore, defining optimal approaches to the treatment of cryptococcal meningitis will become increasingly important.
Among patients with cryptococcal meningitis who did not have AIDS, the administration of amphotericin B combined with flucytosine over a period of four to six weeks led to success rates of 75 to 85 percent. This regimen has been less effective among patients with AIDS, in whom success rates with amphotericin B (0.3 to 0.5 mg per kilogram of body weight per day) with or without flucytosine are only 40 to 50 percent and drug intolerance is common, especially when flucytosine is used. By contrast, Larsen et al. reported success in all six patients treated with higher doses of amphotericin B (0.7 mg per kilogram per day for the first two weeks of therapy) plus flucytosine. However, even among patients successfully treated, high relapse rates necessitate the use of long-term maintenance therapy.
Fluconazole, one of the recently developed triazole antifungal agents, represents a promising new approach to the treatment of cryptococcal meningitis in patients with AIDS. Unlike earlier antifungal azoles such as miconazole and ketoconazole, fluconazole is highly active against Cryptococcus neoformans infection induced in animals and possesses novel pharmacologic properties, including bioavailability after oral administration, excellent penetration of the cerebrospinal fluid, and renal excretion of more than 80 percent of active drug. Moreover, fluconazole has been used as therapy for AIDS-associated cryptococcosis in pilot studies, with encouraging results.
Accordingly, we initiated a multicenter, randomized clinical trial to compare the efficacy and tolerability of orally administered fluconazole with that of intravenously administered amphotericin B as primary therapy for cryptococcal meningitis in patients with AIDS. This report describes the results of that trial.
The criteria for enrollment included HIV infection documented by a positive test for the HIV antibody, age of at least 18 years, and a positive cerebrospinal fluid culture for C. neoformans. Patients could be enrolled on the basis of preliminary evidence of acute cryptococcal meningitis, including a positive cerebrospinal fluid titer for cryptococcal antigen, a positive cerebrospinal fluid India ink preparation, or evidence of C. neoformanson histopathological examination of tissues from other sites in combination with abnormal cerebrospinal fluid laboratory values. However, the cerebrospinal fluid culture must have been positive within 14 days of enrollment for the patient to be eligible. The episode of cryptococcal meningitis that we studied may have been an initial episode or represented a relapse after successful treatment. Patients could be enrolled if they had received no more than 1 mg of amphotericin B per kilogram within one week before entry.
Patients were excluded from the study if they were pregnant or comatose or were considered unlikely to survive for more than two weeks. They were also excluded if they had evidence of concomitant acute or chronic meningitis not due to C. neoformans, moderate-tosevere liver dysfunction, and a creatinine clearance rate of less than 0.333 ml per second (20 ml per minute). Concomitant therapy with anticoagulants similar to warfarin, oral hypoglycemic agents, barbiturates, phenytoin, immunostimulants, lymphocyte-replacement therapy, or investigational agents was not allowed. The study protocol was reviewed and approved by the institutional review board at each study site.
After written informed consent was obtained from either the patient or the principal care giver, two patients were randomly assigned to fluconazole treatment for each patient assigned to amphotericin B treatment. Randomization was performed at each site by means of sealed envelopes and was stratified according to treatment site.
Fluconazole was administered orally in a single, 400-mg loading dose on the first day and in a dose of 200 mg daily thereafter. It could be administered intravenously in the same doses during the first week of treatment and for brief periods, up to three days, after the first week. Since the drug is excreted renally, the dose was decreased by 50 percent if the creatinine clearance rate was less than 0.83 ml per second (50 ml per minute). The dose could be increased to 400 mg after two weeks of therapy if the cerebrospinal fluid culture was still positive and either the patient’s mental status deteriorated slightly or non-life-threatening cryptococcal disease developed at a new site.
Amphotericin B was administered intravenously in a dose of at least 0.3 mg per kilogram per day or in an equivalent dose given every other day. It was temporarily discontinued if serum creatinine levels exceeded 309 μmol per liter (3.5 mg per deciliter), and was reinstituted if they fell below 265 μmol per liter (3.0 mg per deciliter). The decision to use flucytosine (150 mg per kilogram per day in four divided doses) was left to the discretion of the investigator.
After a base-line evaluation, patients were evaluated every week for the first 4 weeks and every 2 weeks thereafter until the 10-week study period was completed. At each visit, a physical examination was performed and any adverse events were assessed and recorded. Serum electrolytes were measured, renal and hepatic function tested, and hemograms obtained to ensure the patients’ safety and to compare the relative toxicity of the therapeutic agents. Serum titers of cryptococcal antigen were determined every two weeks. Lumbar punctures were performed every two weeks; the cerebrospinal fluid was analyzed for cell counts and the levels of protein, glucose, and cryptococcal antigen and was assessed by India ink staining and fungal culture. Tests for cryptococcal antigen were performed at each study site, rather than by a central laboratory.
Treatment was considered successful if the patient had clinical improvement or complete resolution of symptoms together with two consecutive negative cultures of cerebrospinal fluid samples obtained at least one week apart. Cultures of specimens from all other sites of C. neoformans infection except the urine must also have become negative by the end of therapy. Treatment was considered to have failed if the patient had quiescent disease, defined as clinically stable, progressive improvement or resolution of symptoms but persistently positive cultures for any site or body fluid other than the urine, or only one negative cerebrospinal fluid culture by the end of the 10-week treatment period; had progressive disease, defined as progressive clinical deterioration in the presence of persistently positive cultures; had a toxic reaction, defined as a serious or intolerable adverse event that was judged to be at least possibly related to the study drug by the investigator and that led to permanent discontinuation of the study drug; or died of progressive cryptococcal disease.
STUDY DESIGN AND STATISTICAL ANALYSIS
The primary efficacy analysis in this study compared the rates of successful treatment in the two groups. We assumed that if amphotericin B was no more than 30 percent more effective, as judged by successful sterilization of cerebrospinal fluid cultures, than fluconazole in the treatment of acute cryptococcal meningitis in patients with AIDS, then fluconazole could be considered to be a useful agent in treating cryptococcal meningitis.17 We sought to enroll 120 patients who could be evaluated – 80 to be assigned to fluconazole and 40 to amphotericin B. We assumed that amphotericin B treatment would result in clinical success in 60 percent of the patients, and we used a one-sided test with an alpha level of 0.025 and a power of 80 percent to detect a 30 percent lower response rate among fluconazole-treated patients. After an earlier study of cryptococcal meningitis, we estimated that 40 percent of the patients would not be compliant or eligible. Therefore, 161 patients (107 to be assigned to fluconazole and 54 to be assigned to amphotericin B) were scheduled to be enrolled. The initial analyses were performed on an intention-to-treat basis and included all eligible patients.
The treatment groups were compared by the Wilcoxon rank-sum test, Student’s paired t-test, or chi-square analysis. The Kaplan–Meier method was used in survival analyses, and the treatment groups were compared by both the log-rank and the Gehan tests. Cox proportional-hazards linear modeling was used for multivariate assessment of the relative risk of death during therapy, after adjustment for potentially confounding factors.
From April 1988 through November 1989, 235 patients were enrolled by investigators of the National Institute of Allergy and Infectious Diseases Mycoses Study Group and the AIDS Clinical Trials Group and investigators at seven independent sites. Of these 235 patients, 37 (10 [13 percent] assigned to amphotericin B and 27 [17 percent] assigned to fluconazole) were determined to be ineligible for the study for the following reasons: 17 patients were not infected with HIV (5 assigned to amphotericin B and 12 to fluconazole), 15 had negative base-line cerebrospinal fluid cultures for C. neoformans (5 assigned to amphotericin B and 10 to fluconazole), 4 assigned to fluconazole (but none assigned to amphotericin B) had received a dose of amphotericin B before enrollment greater than the dose allowed by the enrollment criteria, and 1 assigned to fluconazole had been reentered in the study after treatment failure during the study.
Four other patients were removed from the study before they received any study drug: three patients (assigned to amphotericin B) asked to be withdrawn, and the physician of the fourth patient (assigned to fluconazole) decided not to give the study drug. The outcomes of these four patients were included as treatment failures in an initial intention-to-treat analysis; however, because their inclusion made no difference to the overall outcome (P = 0.40 vs. P = 0.43) and because the inclusion of patients who never received the study drug was deemed to make the study less clinically applicable, only the 194 patients who did receive the drug were included in further analyses. Of these 194 patients, 63 were assigned to amphotericin B treatment and 131 to fluconazole. There were no significant differences between the treatment groups at base line with respect to demographic characteristics, laboratory values, or clinical signs and symptoms of cryptococcal disease.
Treatment was successful in 25 of the 63 patients assigned to amphotericin B (40 percent; 95 percent confidence interval, 26 percent to 53 percent), as compared with 44 of the 131 assigned to fluconazole (34 percent; 95 percent confidence interval, 25 percent to 42 percent) (P = 0.40 by chi-square test). The relative risk of success of amphotericin B treatment was 1.3 (95 percent confidence interval, 0.69 to 2.44). Approximately one fourth of the patients in each treatment group (17 of 63 amphotericin B recipients [27 percent] and 34 of 131 fluconazole recipients [26 percent]) had so-called quiescent disease. When these patients were combined with patients who were successfully treated, the overall response rate was 67 percent in the amphotericin B group (42 of 63 patients) and 60 percent in the fluconazole group (78 of 131 patients) (P = 0.39 by chi-square test; relative risk of success of amphotericin B treatment, 1.5; 95 percent confidence interval, 0.79 to 2.86). Disease progression occurred in a higher proportion of the patients assigned to fluconazole (20 percent vs. 11 percent), whereas drug toxicity leading to treatment failure occurred in a greater percentage of the patients assigned to amphotericin B (8 percent vs. 2 percent); neither of these differences was statistically significant.
Mortality due to progressive cryptococcal disease was 17 percent, with 9 deaths in the amphotericin B group and 24 in the fluconazole group (14 percent vs. 18 percent; P = 0.48 by chi-square test). Early deaths, defined as those occurring within the first two weeks of therapy, were more frequent in the fluconazole group (15 percent [19 of 131] vs. 8 percent [5 of 63]; P = 0.25 by chi-square test).
At 10 weeks, the response rate, determined by Kaplan–Meier estimates of the length of time to the first negative culture, was 43 percent in the fluconazole group and 46 percent in the amphotericin B group. The 95 percent confidence interval for the difference was -0.21 to 0.15, which allowed us to exclude as a reasonable possibility the assumption that amphotericin B was at least 30 percent more effective than fluconazole (P = 0.001 by one-sided t-test). The median length of time to the first negative cerebrospinal fluid culture was 42 days in the amphotericin B group (95 percent confidence interval, 28 to 71) and 64 days in the fluconazole group (95 percent confidence interval, 53 to 67). There was no significant difference between the two treatment groups in the length of time to the conversion of the cerebrospinal fluid culture from positive to negative (P = 0.25 by log-rank test). However, among the patients who were treated successfully, the median time to a negative culture was 16 days in the amphotericin B group (range, 12 to 71 days; n = 25), as compared with 30 days in the fluconazole group (range, 4 to 67 days; n = 44) (P = 0.02).
The median total dose of amphotericin B given the patients in whom treatment was successful was 24.3 mg per kilogram (range, 11.8 to 49.9), as compared with 23.1 mg per kilogram (range, 0.2 to 43.22) given the patients in whom it failed, including those who died within the first two weeks of treatment (P = 0.91 by Wilcoxon two-sample test). The mean daily dose of amphotericin B given the patients with successful treatment was no different from that given the patients with treatment failure (0.4 vs. 0.5 mg per kilogram per day; P = 0.34), and no dose–response effect was observed. Of the 63 patients assigned to amphotericin B treatment, only 9 also received flucytosine; in only 1 of these 9 patients was flucytosine discontinued because of toxicity (increased creatinine levels).
Eighty-three of the 141 fluconazole recipients received 200 mg of the drug per day throughout their course of therapy. The dose was increased to 400 mg per day in the other 48 patients for the following reasons: clinical deterioration in 16 patients, persistence of positive cerebrospinal fluid cultures in 14, no obvious reason in 10, noncompliance of an investigator in 7, and noncompliance of the patient in 1. The outcomes of these 48 patients who received an increased dose were similar to the outcomes observed in all 131 given fluconazole: 13 patients (27 percent) were successfully treated, 12 (25 percent) had quiescent disease, 15 (31 percent) had treatment failure due to disease progression and 8(17 percent) died. The median length of time to the increase in dose was 22 days.
FACTORS PREDICTING OUTCOME
The pretreatment factors identified by univariate analysis as predictive of treatment failure were a positive India ink stain (P = 0.0005; relative risk = 4.72; 95 percent confidence interval, 1.88 to 11.82), the presence of visual abnormalities (blurred vision, photophobia, diplopia, or papilledema) (P = 0.005; relative risk = 2.57; 95 percent confidence interval, 1.35 to 4.85), an age of less than 35 years (P = 0.01; relative risk = 2.24; 95 percent confidence interval, 1.25 to 4.06), and the absence of therapy with zidovudine (P = 0.05; relative risk = 2.18; 95 percent confidence interval, 1.00 to 4.76). The pretreatment factors predictive of early death were an abnormal mental status, defined as lethargy or obtundation (P = 0.00002; relative risk = 5.99; 95 percent confidence interval, 2.44 to 14.73), a cerebrospinal fluid cryptococcal-antigen titer of more than 1:1024 (P = 0.02; relative risk = 4.31; 95 percent confidence interval, 1.17 to 15.80), and a positive blood culture for C. neoformans (P = 0.03; relative risk = 3.76; 95 percent confidence interval, 1.05 to 13.33). Other factors, such as previous cryptococcal infection, absence of headache, cranial-nerve dysfunction, hyponatremia, presence of C. neoformans at extraneural sites, and antifungal treatment, were not predictive of either early death or treatment failure. The base-line cerebrospinal fluid opening pressure was recorded in only 33 percent of the patients and therefore could not be included as a factor in the proportional-hazards analysis.
In a multivariate analysis, significant pretreatment predictors of death during therapy were an abnormal mental status (P<0.0001), a cerebrospinal fluid cryptococcal-antigen titer of more than 1:1024 (P = 0.01), and a cerebrospinal fluid white-cell count less than 0.02×109 per liter (20 cells per cubic millimeter) (P = 0.04). Using these predictive factors, we grouped the patients according to their risk of death during therapy, as being at low or high risk. Thus, a patient with an abnormal mental status, a cerebrospinal fluid cryptococcal-antigen titer above 1:1024, and a cerebrospinal fluid white-cell count below 0.02×109 per liter would have a high risk of death while receiving therapy. Among the patients at low risk, mortality was 2.6 percent among those treated with amphotericin B (1 of 39 patients) and 4.9 percent among those treated with fluconazole (4 of 81 patients) (P = 0.54 by chi-square test). In contrast, among the patients at high risk, mortality was 33 percent in amphotericin B recipients (8 of 24 patients) and 40 percent in fluconazole recipients (20 of 50 patients) (P = 0.58).
Fluconazole was tolerated much better than amphotericin B; 73 percent of fluconazole recipients reported no adverse effects, as compared with 36 percent of amphotericin B recipients (P<0.0001 by chi-square test). In addition, severe toxicity requiring discontinuation of the drug was more frequent in the amphotericin B group; treatment failure due to drug toxicity occurred in 8 percent of amphotericin B recipients, as compared with 2 percent of fluconazole recipients (P = 0.12 by Fisher’s exact test). Of the five amphotericin B recipients with treatment failure due to serious toxicity, three had superimposed bacterial infections related to the infusion catheter and two had serious renal insufficiency that developed during treatment. Of the three fluconazole recipients whose treatment was stopped because of adverse events, one had elevated levels of hepatic enzymes, one had agranulocytosis that developed during treatment, and one had a seizure considered possibly related to the drug treatment.
The primary objective of our study was to assess whether oral fluconazole was at least as effective as standard intravenous therapy with amphotericin B in the treatment of acute cryptococcal meningitis in patients with AIDS. It was assumed that owing to the advantages of oral therapy, if fluconazole was not more than 30 percent less effective than intravenous amphotericin B, then fluconazole would be an acceptable agent for treatment of primary cryptococcal meningitis. The confidence interval around the difference in success rates, along with the lack of a significant difference between the two treatment groups in the overall outcome (P = 0.40), indicates that fluconazole is an acceptable alternative to amphotericin B as primary therapy for acute AIDS-associated cryptococcal meningitis.