The study was designed as a multi-centre, open-label, non-inferiority, randomized controlled trial, using a sequential-step design to evaluate the efficacy and safety of a single dose treatment regimen of intravenous AmBisome® (either 7.5, 10.0, 12.5 or 15.0 mg/kg body weight) compared to the reference multiple dose regimen currently approved in the USA: 3 mg/kg body weight on days 1 to 5, 14, and 21. The single dose tested in the first cohort was 7·5 mg/kg body weight (figure 1, the consort flowchart). Two interim analyses were planned, after enrolment of 20 and 40 patients per arm, for early detection of inefficacious single doses, based on parasite clearance at day 30 (figure 2) and/or worsening clinical conditions. If the stopping rule was met, the single dose was increased by 2·5 mg/kg and recruitment into the two arms restarted. The multiple dose treatment remained the same throughout. Non-responders to treatment were considered treatment failures and received rescue medication (multiple dose AmBisome® regimen for single-dose failures and SSG for multiple dose failures).

Patients with age of at least 4 years, confirmed HIV-negative, parasitologically-confirmed non-severe VL, were enrolled in three centres: (1) Gondar University Hospital, Amhara Regional State, Northern Ethiopia; (2) Arba Minch Hospital, Gamo Gofa Zone, Southern Nations, Nationalities and Peoples Regional State, Southern Ethiopia; and per protocol amendment (3) Ministry of Health Hospital, Kassab, Gedaref State, Eastern Sudan from May 2009 to September 2010. Exclusion criteria were signs/symptoms of severe VL (patients who were very weak, unable to walk, bleeding, jaundiced, suffering from sepsis and other concomitant infections/illnesses); anti-leishmanial or unlicensed investigational treatments within six months; underlying chronic disease such as severe cardiac, pulmonary, renal, or hepatic impairment; serum creatinine outside the normal range; liver function tests more than 3 times the normal range; platelet count less than 40,000/mm3; known alcohol abuse; pregnancy or lactation; concomitant acute drug usage for malaria and bacterial infection; pneumonia within last 7 days; known hypersensitivity to AmBisome® or amphotericin B; any other condition which may invalidate the trial.

On pre-specified treatment days, AmBisome® dosage was calculated according to body weight. Preparation included reconstitution with sterile water for injection and filtration according to the manufacturer's instructions. Administration was by slow intravenous infusion in 5% dextrose solution. During infusion, patients were closely observed with regular monitoring of vital signs (blood pressure and pulse). A test dose was administered in the first 10 minutes of infusion, and the patients carefully observed for 30 minutes. The time of treatment and dosage was recorded.

Patients were randomized to receive either treatment using a computer-generated randomisation list, stratified by site. Individual treatment allocations were placed in sealed, opaque envelopes which were opened after a patient had been entered into the trial. It was not possible to blind patients or treating physicians due to the nature of the intervention.

Primary (definitive/final cure) and secondary (initial cure) efficacy endpoints were determined by parasitological assessment, as the most objective and consistent method across treatment centres at six months follow-up and at Day 30, respectively. Day 30 was considered as the end of treatment time point for initial cure assessment. Definitive cure was defined as absence of parasites in tissue aspirates (bone marrow, lymph node or spleen) with no relapse of signs and symptoms of VL during six months follow-up. Initial cure was defined on day 30 by absence of parasites in tissue aspirates. For patients with detectable parasites at initial cure assessment, clinical and biological assessments were used, in addition to parasitological results, to ascertain the need for rescue treatment at the end of study treatment at the discretion of the treating clinician, according to the protocol. Patients with a presence of parasites at initial assessment of cure but who were clinically well were invited to return after one month to further assess their status and need for rescue treatment. Patients who did not clear parasites at end of treatment but did by six months follow-up, in the absence of a requirement for additional rescue treatment, were classified as treatment successes at six months as patients who were slow responders, in line with previous trials in the region [7]. Assessment of safety involved monitoring vital signs, documentation of patients' complaints about the treatments, and haematological and biochemical measurements evaluated on days 2 to 5, 7, 14, 21 and 30, and at 3 and 6 months. Treatment emergent adverse events (TEAE) were classified according to the Medical Dictionary for Regulatory Activities (MedDRA). Treatment emergent events were those with onset between day 1 of treatment and day 60 inclusive.

Peripheral Blood Parasitaemia: In Kassab, per protocol amendment, peripheral blood samples were analysed for parasite load in a subset of 5 consenting patients in each of the 10 mg/kg single dose and multiple dose arms. For this, a validated quantitative reverse-transcriptase polymerase-chain-reaction (qRT-PCR) method targeting Leishmania 18S ribosomal RNA was used [15]. Genetic material was extracted using a modified Boom-method [15], [16], [17]. qRT-PCR analysis using a Bio-Rad CFX-96 real-time machine (Bio-Rad, Veenendaal, the Netherlands) was performed at Koninklijk Instituut voor de Tropen (KIT).

For the primary endpoint comparison, 120 patients per arm would provide 80% power to detect non-inferiority within a margin of 10%, assuming 95% cure in the reference arm, a one-sided alpha of 0·05 and 15% loss-to-follow-up. In interim analyses, 20 patients per arm would provide 90% power to detect a difference of at least 35% in parasite clearance rate at day 30, assuming 95% cure in the reference arm and a two-sided alpha of 0.05. With 40 patients per arm, there would be 90% power to detect a difference of at least 25% under the same assumptions.

Interim analyses were based on day 30 cure in the Intention-to-Treat (ITT) population. Decision-making at each interim analysis was based on a test of difference between the parasite clearance rates in the single dose arm and multiple dose arms. If the single dose arm showed significantly poorer efficacy (P<0.05), the single-dose was increased prior to re-starting recruitment. Patients allocated to the multiple-dose arm for discontinued single-dose comparisons were not included in comparisons of higher single doses.

Key assumptions for the planned final analysis were not met due to low efficacy in the multi-dose arm and the trial was terminated prematurely (see Results). Cumulative data for each treatment regimen were used to calculate the percentage of patients cured, with exact binomial 95% confidence intervals (CI), at day 30 and 6 months follow-up in ITT and per-protocol (PP) analysis populations. Patients with missing outcome data were excluded from analyses. For safety, the number and percentage of patients per arm experiencing adverse events (AEs) were summarised, for AEs with cumulative incidence higher than 10%. For parasite clearance from peripheral blood, a linear mixed effects regression model using the natural log-transformed parasite loads was applied to estimate the time to clear 50% and 90% of parasites for each individual. Model performance and significance were assessed by analysis of variance (ANOVA).

The most commonly presenting VL symptoms were fever and weight loss, followed by loss of appetite, abdominal swelling, and cough; less commonly observed were epistaxis, diarrhoea, and skin lesions. Other characteristics of patients at entry to the trial are summarized in Table 1. Overall, 82% of patients were male and about half were children. About two thirds of patients were underweight or severely underweight. Mean haemoglobin concentrations were <8·0 g/dl, and anaemia was common, but neither baseline laboratory parameters nor vital signs suggested any major difference among dose groups. Baseline characteristics were generally comparable in the multiple and the 10 mg/kg single dose group, whereas the smaller 7·5 mg/kg dose group showed some imbalances: on average, patients in this group were older and accordingly had higher body weight and larger spleen size, but also had the highest baseline parasitaemia. Patients were younger in Kassab (10·5±5·0 years) than in Gondar (20·9±6·3) or Arba Minch (17·4±10·2 in) and more often female in Kassab (33·3%) than in Gondar (10·5%) or Arba Minch (12%). In Arba Minch, they were more often normal weight (42·0%) than in Gondar (21·1%) or Kassab (28·6%).

In the first section of Table 2, parasite clearance rates at day 30 are shown for the three interim analyses. Summary data for the parasite clearance rate at day 30 and the cure rate at 6 months are shown for all patients and for those treated at each of the 3 centres. The IC and DC rates with the standard multiple dose treatment were both 85%. IC rates with single doses of 7·5 and 10 mg/kg were 50% and 73%, respectively; and DC rates were lower, at 40% and 58%, respectively. However, there were variations in treatment response between treatment centres, with poor efficacy in Kassab and Gondar, particularly with single doses. By contrast, at Arba Minch, the multiple doses as well as the single dose of 10 mg/kg resulted in complete cure, and treatment failures were observed with the 7·5 mg/kg dose only. All non-responders were cured after receiving rescue medication.

For the first interim analysis, comparing the 7·5 mg/kg single dose to the multiple dose regimen with 20 and 18 patients per arm, respectively, in the two Ethiopian sites, the stopping rule was met (Table 2: Fisher's exact test, p = 0·015). The single dose was increased to 10 mg/kg, and recruitment restarted at both Ethiopian sites and in an additional site, Kassab, Sudan. There was no significant difference in efficacy found at the next interim analysis comparing 10 mg/kg to the multiple dose arm (p = 0·748), but when 44 patients had been recruited into the multiple dose and 40 patients into the 10 mg/kg single-dose arm, the third interim analysis indicated unexpectedly low initial cure rates in both arms; 84% in the multiple dose and 73% in the single-dose arm. The stopping rule was not met (chi-squared test, p = 0.196), but based on the observed poor efficacy overall, and following discussions with the Data Safety and Monitoring Board (DSMB) and investigators, the sponsor terminated the trial.

At that time, a total of 124 patients had been enrolled into the trial; 63 had received the multiple dose regimen, 20 patients received a single dose of 7·5 mg/kg dose and 41 patients received a single dose of 10 mg/kg (figure 1).

TEAE were common regardless of dose regimen. Severity was mostly mild or moderate and only about 2% of TEAE were rated severe, mostly with respect to laboratory measurements. There was one non-fatal SAE, a pneumonia deemed unlikely related to the drug, and one death due to snakebite. AEs for which relatedness could not be excluded are listed in Table 3. These potential adverse drug reactions were seen in all the three dose groups and occurred in both the multiple and the higher single dose groups with similar frequencies.

At baseline, semi-quantitative microscopy counts on bone marrow aspirates correlated well with parasite loads in peripheral blood when assessed by qRT-PCR (R²: 0·77, p<0·01). Mean natural log-normalized parasite loads (P) were comparable in the single and multiple dose groups (6·4 lnP/mL, 95%CI: 4·6–8·2 versus 5·1 lnP/mL, 95%CI: 3·3–6·8; p = 0·358) at baseline. Three out of the 5 patients of each group had baseline blood parasite loads >50 per mL and these had clearance rates assessed and modelled over the first 7 days. Mean parasite clearance rates were significantly different between the single and the multiple dose group (0·35 per day, 95%CI: 0·00–0·70 versus 1·14 per day, 95%CI: 0·78–1·50; p = 0·012) as early as day 3 (Figure 3), corresponding to mean parasite elimination half-lives of 1·97 days (95% CI: 0·99–278) for the single-dose group and 0·61 days (95%CI: 0·46–0·89) for the multiple dose group. Time required for 90% parasite clearance for single-dose and multiple-dose groups was estimated at 6·55 (95% CI: 3·29–923) and 2·02 (95% CI: 1·53–2·95) days, respectively. One patient in the single dose group had a low blood parasite load at baseline, which increased until day 30, but was no longer detectable after rescue treatment.

AmBisome® in single doses of up to 10 mg/kg is suboptimal for eastern African VL patients. A higher dose of 21 mg/kg administered in multiple doses was also far less effective than anticipated. In a small number of patients, no treatment failures occurred in patients from south Ethiopia treated in Arba Minch Hospital either with 10 mg/kg single dose or 21 mg/kg divided doses. The tested AmBisome® regimens would not be suitable for VL treatment across eastern Africa, and an optimal single dose that could be included in shorter and simplified treatment regimens of anti-leishmanial drugs across all VL endemic regions of eastern Africa was not identified. Efficacy of AmBisome®for the treatment of VL in eastern Africa was variable and overall lower than in India.