Study design

An in vivo assessment of the therapeutic efficacy of AL, ASAQ, and DP was conducted according to the standard WHO protocol [3]. Study participants were recently febrile children with P. falciparum monoinfection. Participants were followed with physical exams and blood samples for 28–42 days after anti-malarial administration for development of adverse medication effects or recurrent parasitaemia. Molecular analyses were performed on samples from those experiencing treatment failure.

Study population

The study took place at anti-malarial resistance sentinel sites in Benguela, Lunda Sul, and Zaire that were retained from previous therapeutic efficacy studies [6, 7]. Benguela is a southern coastal province with stable mesoendemic malaria transmission. Zaire is a forested province on the northern coast that also has stable mesoendemic transmission. Lunda Sul province, part of Angola’s eastern savannah, features hyperendemic transmission (Fig. 1) [1].

Enrollment was divided into six arms, and efficacy of each medication was assessed in two provinces. Target enrollment of one hundred participants per arm provided enough power to estimate efficacy with 95% confidence limits of ± 5%, assuming an expected efficacy of 95% and a maximum loss to follow-up and exclusion rate of 27%. AL and DP were evaluated in Lunda Sul, AL and ASAQ in Zaire, and DP and ASAQ in Benguela (Fig. 1). Two drugs were assessed sequentially in each site. Medication administration was not randomized or blinded; the first hundred participants enrolled at each site received the first ACT to be evaluated there (AL in Zaire and Lunda Sul or DP in Benguela), and the second hundred enrollees received the second drug under evaluation (ASAQ in Benguela and Zaire or DP in Lunda Sul).

Screening for participants took place at two health facilities in each province’s capital from March to July 2017. Study entrants were children with fever (≥ 37.5 °C) or a history of fever in the past 24 h, P. falciparum monoinfection on blood smear, no signs of severe malaria on physical exam, no concomitant illness or severe malnutrition, a haemoglobin greater than 8 g/dL, no anti-malarials within the last 14 days, and caregivers willing to attend all follow-up visits with study staff. Inclusion criteria were broader in Benguela compared to Lunda Sul and Zaire, allowing for older children (< 12 years vs. < 5 years) and lower parasitaemias (between 1000 and 100,000 asexual parasites/μL vs. between 2000 and 200,000 parasites/μL), due to the lower level of malaria transmission in Benguela.

Clinical monitoring

Enrolled children received a 3-day course of one of the anti-malarials under evaluation, dosed according to guidelines from the drug manufacturers. Quality-controlled AL (Ipca Laboratories, Maharashtra, India), ASAQ (Sanofi Aventis, Paris, France), and DP (Eurartesim^®; Leadiant Biosciences, Rome, Italy) were provided by WHO. For ASAQ and DP, which have once-daily dosing, all three doses were given under direct observation of study staff. For AL, which requires twice-daily dosing, each morning dose was directly observed. Evening doses were given at home, and as in previous studies, efforts were made to ensure correct administration of each dose. Parents or guardians were given the appropriate weight-based dose, called at home to confirm delivery of the dose, and instructed to bring the empty pill packages back to study staff the following morning. All AL doses were administered with a snack, containing at least 3 g of fat, to facilitate drug absorption. Parents or guardians were given individual packages of yogurt or chocolate milk to take home and were asked to bring the empty package back to study staff the following morning.

After each medication administration, children were observed for 1 h to monitor for vomiting or other side effects, including diarrhea, nausea, or excessive sweating. Children who vomited within half an hour after medication administration were given a repeat dose. Children who vomited between half an hour and 1 h after medication administration were given a repeat dose that was half of the original dose. Those with persistent vomiting were excluded from the study.

Participants made eight or ten visits to study staff over the course of 28 or 42 days. Participants in the AL and ASAQ arms were followed for 28 days, while those in the DP arms required 42-day follow-up due to the longer half-life of piperaquine [3]. All children were monitored daily during the 3 days of drug administration and 1 day following; afterwards, follow-up visits occurred weekly. An interval history, physical exam, thick and thin blood smear, and dried blood spot collection were performed at each visit, except for day 1 of follow-up, in which blood samples were only collected if a child exhibited signs of severe malaria. Participants’ haemoglobin was also measured every 2 weeks. Finally, participants in the AL arm in Zaire underwent an additional dried blood spot collection on day 7 of follow-up to measure blood lumefantrine levels, to provide further insight into the large number of late treatment failures previously observed with AL in this province [6, 7].

Children were excluded from follow-up if they developed a non-falciparum malaria infection, showed signs of severe illness, missed a follow-up visit, or took another anti-malarial. For the preliminary analysis, the primary outcomes were adequate clinical and parasitological response, failure to adequately clear parasitaemia, or recurrent P. falciparum parasitaemia. Participants were categorized as early treatment failures if their parasitaemia increased from day 0 to day 2 of follow-up, failed to decrease by at least 75% by day 3, was associated with fever on day 3, or was associated with signs of severe malaria on days 1–3. Participants who developed any parasitaemia after day 3 were classified as late treatment failures.

Molecular analysis

Microsatellite analysis and Sanger sequencing of P. falciparum parasites were performed on selected blood samples from study participants. Day 0 and day of failure samples from all cases of late treatment failure were genotyped by microsatellite analysis for reclassification as reinfection or recrudescence. Additional day 0 samples from participants not experiencing treatment failure were randomly selected for neutral microsatellite analysis to augment the number of samples for determination of allele frequencies. day 0 and day of failure samples from all cases of late treatment failure were also assessed for pfK13, pfcrt, and pfmdr1 sequence and pfmdr1 and pfpm2 copy number.

Genomic DNA was extracted using the QIAamp blood minikit (Qiagen Inc., California, USA). Seven neutral microsatellite loci spanning six chromosomes (TA1, chromosome 6; poly α, chromosome 4; PfPK2, chromosome 12; TA109, chromosome 6; and 2490, chromosome 10; C2M24, chromosome 2; and C3M69, chromosome 3) were amplified by non-nested or semi-nested PCR and their fragment lengths were measured [21–24]. Fragments of the pfcrt [25], pfmdr1 [25], and pfK13 [26] genes were amplified using previously published primers. Direct Sanger sequencing of the nested purified PCR products was performed by using a BigDye Terminator v3.1 cycle sequencing kit on an iCycler thermal cycler (Bio-Rad, California, USA). Sequence analysis was performed by using Geneious R7 (Biomatters, Auckland, New Zealand).

Pfmdr1 copy number was assessed using a previously described protocol [27]. The P. falciparum β-tubulin gene was used as the reference gene for relative quantification of pfmdr1 gene copy number. Each run included the 3D7 parasite strain as a single-copy control, the W2Mef strain (two copies), and the Dd2 strain (three to four copies). The assay was performed using the Agilent Mx3005 real-time PCR instrument (Agilent Technologies, California, USA). The copy number was determined by using the relative quantification module in MxPro3005 software (Agilent Technologies, California, USA), using the comparative ΔΔCT method, and rounded to the nearest integer.

For pfpm2 copy number analysis, the P. falciparum β-tubulin gene was again used for reference. The reverse primers of both the pfpm2 and β-tubulin genes were modified with the PET tag and labeled with FAM (pfpm2) and HEX (β-tubulin) fluorophores. The PET-PCR assays were performed using Agilent Mx3005pro thermocyclers (Agilent Technologies, California, USA).

Lumefantrine level measurement

Additional dried blood spots were collected on day 7 of follow-up from participants in the AL arm in Zaire. A volume of 50 μL of whole blood was collected on filter paper pre-treated with 0.75 M tartaric acid and stored at room temperature. Dried blood spots were eluted with an acidified acetonitrile solution followed by solid phase extraction. The extraction procedure was followed by liquid chromatographic separation using a Synergi Hydro-RP, 4 µm, 150 × 2.0 mm analytical column (Phenomenex, California, USA) with an isocratic mobile phase containing a mixture of acetonitrile, water, and formic acid (70:29.8:0.2) at a flow-rate of 300 µL/min. An AB Sciex API 3000 mass spectrometer at unit resolution in the multiple reaction monitoring mode was used to monitor the transition of the protonated precursor ions m/z 530.0 and m/z 539.1 to the product ions m/z 512.1 and m/z 521.3 for lumefantrine and the internal standard, respectively. Electrospray ionization was used for ion production [28, 29].

Day 7 lumefantrine concentrations greater than or equal to 0.2 μg/mL were considered therapeutic [30].

Statistical analysis

Uncorrected efficacy was determined by dividing the number of treatment failures in each study arm by the total number of participants classified as either adequate clinical and parasitological response or treatment failure in that arm. Parasite clearance rates on day 2 and day 3 of follow-up were also examined to evaluate artemisinin delayed response.

Kaplan–Meier estimates of survival curves were calculated using the survival package in R version 3.3.2 (R Foundation for Statistical Computing, Vienna, Austria). Data from participants excluded for reasons other than treatment failure or incorrect enrollment were included in the analysis until the day of study departure.

To determine corrected therapeutic efficacy, microsatellite data were analysed using a previously published algorithm [31] that assigns each late treatment failure a posterior probability of recrudescence. A probability of more than 0.5 was considered a recrudescence, and less than or equal to 0.5 was designated as a reinfection.

The distribution of day 7 lumefantrine drug levels was compared between cases of adequate clinical and parasitological response and cases of recrudescence and reinfection using the Kolmogorov–Smirnov test. The proportion of participants with sub-therapeutic day 7 lumefantrine levels was compared between cases of adequate clinical and parasitological response and cases of recrudescence and reinfection using Fisher’s exact test.

Triage and enrollment

Efficacy

The majority of the 38 late treatment failures were reinfections (26, 68%) rather than recrudescences (12, 32%) by microsatellite analysis (Table 2). Late treatment failures were categorized with a high degree of confidence, with the posterior probability of recrudescence greater than 90% or less than 10% in all but three instances (Additional file 2). One late treatment failure in the Zaire AL arm had a posterior probability of recrudescence of 29%, and two late treatment failures in the Zaire ASAQ arms had probabilities of 19 and 39%. Four cases of recrudescence were identified in the Zaire AL arm, three in the Lunda Sul AL arm, and five in the Zaire ASAQ arm. No recrudescences were found among participants in the Benguela DP, Lunda Sul DP, or Benguela ASAQ arms.

After exclusion of reinfections, the corrected 28-day efficacy of AL was 96% (91–100%), in Zaire and 97% (93–100%) in Lunda Sul. The corrected 28-day efficacy of ASAQ was 100% (97–100%) in Benguela and 93% (88–99%) in Zaire. The corrected 42-day efficacy of DP was 100% (96–100%) in Benguela and 100% in Lunda Sul.

Safety

Among the 586 participants enrolled within inclusion criteria, 11 participants (2%) reported adverse medication effects: four with vomiting after DP (2% of participants in DP arms), three with nausea after ASAQ (2%), two with excessive sweating after ASAQ (1%), one with vomiting after ASAQ (1%), and one with excessive sweating after DP (1%). No participants reported adverse effects after AL administration, neither under direct observation nor at home. No participants left the study due to adverse medication effects.

Molecular markers of resistance

All 76 analysed samples had only one copy of both the pfmdr1 and pfpm2 genes. However, 75 of 76 samples contained multiple P. falciparum strains.

Lumefantrine concentrations

Median day 7 lumefantrine concentrations in the Zaire AL arm were similar among participants experiencing adequate clinical and parasitological response, reinfection, and recrudescence, ranging from 0.19 to 0.23 μg/mL (Additional files 3, 4). Two (50%) out of the four cases of recrudescence observed in the AL arm had sub-therapeutic day 7 lumefantrine levels, but this did not differ significantly (p value 1.0) from rate of sub-therapeutic day 7 lumefantrine levels in participants with adequate clinical and parasitological response (35/87, 40%). Two children had undetectable lumefantrine levels, one of whom successfully responded to treatment and one of whom had a recrudescent infection. The distribution of day 7 lumefantrine concentrations was not statistically different between cases of adequate clinical and parasitological response and recrudescence (p value 0.94), between cases of adequate clinical and parasitological response and reinfection (p value 0.34), or between cases of adequate clinical and parasitological response and recrudescence/reinfection (0.80).