Description

Histoplasmosis is a serious endemic mycosis that may disseminate in immunocompromised patients. The disease in endemic in the American continent, particularly Brazil. Patients with advanced HIV infection are susceptible to disseminated histoplasmosis, an AIDS-defining illness. According to international guidelines, induction therapy for disseminated histoplasmosis involves the use of liposomal amphotericin B for two weeks, but access to this medication is limited in several regions of the globe. A phase II trial showed promising results with the use of a single high dose of liposomal amphotericin B in this context. Here we propose a phase III study aimed to evaluate non-inferiority of induction therapy with liposomal amphotericin B for disseminated histoplasmosis in AIDS, comparing 10 mg/kg (interventional arm) versus 3 mg/kg for two weeks (standard therapy) regarding two-week mortality and superiority in a Desirability of Outcome Ranking (DOOR). Induction therapy will be followed by oral itraconazole for one year for all patients. A Data Safety Monitoring Board (DSMB) will be established with the aim of defining whether the study needs to be stopped early for efficacy or harm to the study participants. The group will meet every 12 months to review the study data.

A steering committee made up of external members will advise and evaluate the study. Meetings will be held every 3 months. In addition, a medical committee made up of members of the study will be responsible for monitoring the progress of the study in order to maintain quality in all its aspects, with weekly meetings.

For data analysis, continuous variables will be described using mean, standard deviation, median, interquartile range, minimum and maximum. Categorical variables will be described using absolute and relative frequencies. The Kaplan-Meier method will be used to describe overall survival. To assess the primary outcome, the proportions in each arm and the respective 90% confidence intervals will be evaluated. Continuous variables will be compared using two-sample t-tests, paired-sample t-tests, Mann-Whitney test, Wilcoxon signed rank test, one-way ANOVA or Kruskal-Wallis test, as appropriate and if necessary. Categorical variables will be compared with Fisher's exact test or chi-squared test, as appropriate. Ordinal DOOR analysis will be done with logistic regression to determine odds ratios.

To control the type I error rate for testing of the primary and major secondary endpoint, a hierarchical strategy will be used. Superiority assessments after successful testing of non-inferiority hypotheses will be performed. There is no multiplicity argument affecting this interpretation, as this approach corresponds to a simple closed testing procedure. The sample size calculation will consider the overall 2-week mortality in the L-AmB control observed in the phase II study (i.e. ~8%). The planned calculation is 279 patients (127 patients per study arm). The sample size is based on a power of 90% to detect a non-inferiority margin of 10% with a two-tailed p-alpha of 5% (i.e. one-sided confidence interval margin of 90%). An expectation of 10% of patients lost to follow-up is added, bringing the sample size to 279 patients (approximately 140 per arm). If the mortality observed in the study is higher than expected, a larger sample size will be necessary. The data will be analyzed using SPSS 27.0 software. If non-inferiority is achieved, the study will be tested for superiority using the DOOR scare. An a priori adaptive sample size is proposed to maintain statistical power if the assumption about two-week mortality is incorrect. A hierarchical testing strategy is proposed to test for superiority of key secondary endpoints of amphotericin-related laboratory toxicity and a DOOR scale. A sample size of 150 participants per arm in a parallel two-group design will be used to test whether distribution of DOOR scores differs between groups (H0: μ1 - μ2 = 0 versus H1: μ1 - μ2 ≠ 0). The comparison will be made using a two-sided, two-sample Mann-Whitney U test, with a Type I error rate α of 0.05. The common standard deviation for both groups is assumed to be 1.5, and the underlying data distribution is assumed to be normal. To detect a difference in means of 0.5 with 80% power, the number of needed subjects will be 300.

Financial support for this study was provided by the following institutions:

Gilead - donation of medication and financial support (USD 393,600); Financiadora de Estudos e Projetos (FINEP/MCTI - Brazil) (USD 355,883.10); and IMMY: donation of diagnostic devices (50 boxes - HGM201, 51 boxes - CR2025);