Trial Design and Oversight
We conducted an open-label, phase 3, randomized, noninferiority, multicenter trial (Advancing Cryptococcal Meningitis Treatment for Africa [ACTA]) to compare three treatment strategies (an oral combination regimen of fluconazole plus flucytosine, 1 week of amphotericin B, and the standard 2 weeks of amphotericin B) for the induction treatment of HIV-associated cryptococcal meningitis. Flucytosine and fluconazole were also evaluated as partner drugs with amphotericin B.
Participants were recruited from nine African centers: Queen Elizabeth Central Hospital, Blantyre, Kamuzu Central Hospital, Lilongwe, and Zomba Central Hospital, Zomba, Malawi; University Teaching Hospital, Lusaka, Zambia; Muhimbili, Amana, and Mwananyamala Hospitals, Dar Es Salaam, Tanzania; and Hôpital Central, Yaoundé, and Douala General Hospital, Douala, Cameroon. The protocol was approved by the London School of Hygiene and Tropical Medicine Research Ethics Committee and by all the site national research ethics committees and regulatory bodies. Written informed consent was obtained from all the patients or, in the case of patients with altered mental status, from the next of kin (consent was obtained from these patients after recovery).
Lateral-flow cryptococcal antigen tests were donated by or purchased from IMMY. Trial drugs were purchased from Bristol-Myers Squibb (amphotericin B [Fungizone]), Meda Pharmaceuticals (flucytosine), and Cipla or Medopharm (fluconazole). In places where the Pfizer fluconazole donation program was running, donated fluconazole was used when available. The trial funders, suppliers, and drug manufacturers had no role in trial design; data collection, analysis, and interpretation; or manuscript preparation. The authors vouch for the accuracy and completeness of the data and for the adherence of the trial to the protocol, available with the full text of this article at NEJM.org.
HIV-seropositive adults (≥18 years old) with a first episode of cryptococcal meningitis who tested positive on India ink staining, cryptococcal antigen assay, or both in cerebrospinal fluid (CSF) were included. Patients were excluded if they had previously received more than one dose of amphotericin B or more than one treatment dose (1200 mg) or more than seven low doses (200 mg) of fluconazole in the 2 weeks before screening, were pregnant or lactating, were taking contraindicated concomitant drugs, or had any previous adverse reactions to the trial drugs.
An alanine aminotransferase (ALT) level that was more than 5 times the upper limit of the normal range, a polymorphonuclear leukocyte count that was less than 500 per cubic millimeter, or a platelet count that was less than 50,000 per cubic millimeter were late-exclusion criteria (i.e., a patient who met one or more of these criteria at baseline was withdrawn from the trial). In addition, if an elevated creatinine level remained above 220 μmol per liter on the day after randomization despite the patient receiving rehydration, the patient was withdrawn from the trial.
Initially, patients were excluded if they had previously been exposed to antiretroviral therapy (ART). However, because it became clear that a large number of patients were presenting with cryptococcal meningitis while taking ART or with previous exposure to ART, soon after commencement of the trial (after 4% of total enrollment), a protocol amendment allowed the inclusion of these patients. Full details of the trial design can be found in the protocol and statistical analysis plan.
Interventions and Randomization
We assessed three treatment strategies (the use of an oral regimen, a 1-week amphotericin B regimen, and a 2-week amphotericin B regimen), as well as two alternative partner drugs for amphotericin B (fluconazole or flucytosine). The oral regimen consisted of fluconazole (1200 mg per day) plus flucytosine (100 mg per kilogram of body weight per day) given orally for 2 weeks. The 1-week amphotericin B regimen consisted of amphotericin B (1 mg per kilogram per day administered intravenously) plus either fluconazole (1200 mg per day) or flucytosine (100 mg per kilogram per day) for 7 days, followed on days 8 through 14 by fluconazole (1200 mg per day). The 2-week amphotericin B regimen consisted of amphotericin B (1 mg per kilogram per day administered intravenously) plus either fluconazole (1200 mg per day) or flucytosine (100 mg per kilogram per day) for 14 days.
Patients underwent block randomization individually, stratified according to site, to one of the three treatment strategies and, for patients who were assigned to an amphotericin B regimen, to one of the two partner drugs. Overall, this strategy resulted in a 2:1:1:1:1 ratio of patients assigned to receive one of the five combinations of treatment strategy and partner drug with amphotericin B. For each site, a computer-generated randomization list with block sizes of 18, 24, and 30 was produced. The trial pharmacist and clinician were responsible for conducting the randomization by sequentially drawing sealed envelopes that contained the treatment assignment for each enrolled patient.
Patients who received amphotericin B were given 1 liter of normal saline intravenously daily in addition to usual fluid requirements and preemptive potassium and magnesium (glycerophosphate) supplementation.16 Oral medications were given through a nasogastric tube if the patient was unable to swallow. Laboratory blood tests were performed regularly during the first 2 weeks of treatment. Baseline and day 7 electrocardiographic monitoring was discontinued at the advice of the data and safety monitoring committee after 100 paired electrocardiograms showed no evidence of clinically significant prolongation of the QT interval in association with fluconazole at a dose of 1200 mg per day. Lumbar punctures were performed at baseline and on days 7 and 14 for quantitative cultures.17 In addition, patients with high CSF pressure underwent daily therapeutic lumbar punctures until the pressure was controlled.5 Patients were followed for 10 weeks after randomization. After 2 weeks, fluconazole was given at 800 mg per day until ART was started at 4 weeks (or restarted in those who had discontinued ART), at 400 mg per day until 10 weeks, and at 200 mg per day thereafter. ART was prescribed in accordance with national guidelines.
The primary end point for comparison of the two experimental treatment strategies with the standard therapy of 2 weeks of amphotericin B–based treatment was all-cause mortality at 2 weeks. Two weeks was chosen in view of the noninferiority design of the trial and the fact that mortality at 2 weeks is more likely than mortality at later time points to reflect deaths from cryptococcal meningitis.18 Secondary end points included 4-week and 10-week all-cause mortality, the rate of decrease in the log10 CSF fungal count over 14 days, and clinical and laboratory-defined grade 3 and 4 adverse events.
For the comparison between partner drugs for the amphotericin B regimens, the primary end point was all-cause mortality at 10 weeks. The secondary end points were all-cause mortality at 2 weeks and 4 weeks, rate of clearance of infection, and adverse events.
A target enrollment of 680 patients (226 per strategy) was set in order to achieve 90% power to show noninferiority with a 10-percentage-point noninferiority margin and under the assumption of 15% mortality at 2 weeks in the 2-week amphotericin B groups. For the comparison of the partner drugs with amphotericin B, with the use of a superiority design and under the assumption of a 10-week mortality of 40% with one partner treatment, the trial had 90% power to detect a 35% lower mortality with the alternative partner treatment.
The primary analysis was based on the intention-to-treat population. A generalized linear model with a binomial distribution and identity link function was used to calculate differences and upper limits of the one-sided 95% confidence interval for mortality. Post-hoc analysis of the primary end point with a one-sided 97.5% confidence interval, the upper limit of which is equivalent to the upper limit of a two-sided 95% confidence interval and the use of which is equivalent to applying a Bonferroni correction (α=0.025) for two comparisons, was also performed. Correction for multiple comparisons was not applied to the analyses of secondary outcomes. The per-protocol population excluded patients who missed more than 1 day of treatment within the first 2 weeks after randomization.
All-cause mortality at 2, 4, and 10 weeks was compared between the groups with the use of log-rank tests. Kaplan–Meier plots were also constructed, and Cox regression models with treatment as a predictor were used to derive hazard ratios and two-sided 95% confidence intervals. Analyses were also performed with adjustment for prespecified covariates: site, age, sex, Glasgow Coma Scale score, CD4+ cell count, CSF fungal count at baseline, and ART status at baseline. Sensitivity analyses of all-cause mortality were performed under the assumption that all the patients who were lost to follow-up had died.
The analysis of the log10 CSF fungal count over a period of 14 days from baseline was performed with a linear mixed-effects model. For comparison with previous studies,17-19 linear regression was also used to calculate slopes of the decrease in CSF fungal count for each patient, and the mean slopes were compared between the groups.
All analyses were performed with the use of SAS software, version 9.3 (SAS Institute). Additional details are provided in the Supplementary Appendix, available at NEJM.org.