Study design: A randomized, open-label phase IV clinical trial was conducted from September 2018 to February 2019 in three health centers in Uganda: Aduku Health Centre IV, Kwania district, in northern Uganda, Arua Regional Referral Hospital; Arua district, in northwestern Uganda; and Masafu District Hospital, Busia district, in Eastern Uganda (figure 1). All sites experience perennial malaria transmission with high transmission intensity. A 2018–2019 malaria indicator survey showed parasite prevalence of 13% in Kwania district, 22% in Arua district and 21% in Busia district (23).
Study population and enrollment: Inclusion criteria were children aged 6 months to 10 years with symptomatic, uncomplicated P. falciparum infection, with a parasite density of 2,000 – 200,000 parasites/µl measured by microscopy; axillary temperature ≥ 37.5°C or history of fever in the prior 24 hours; weight ≥ 5 kg; absence of severe malnutrition, defined as a z-score <-3, symmetric edema, or mid-upper arm circumference <110 mm; no history of serious adverse effects or hypersensitivity reactions to study medications; no recent use of medications which could interact with the study medications; no evidence of severe malaria or danger signs; no evidence of concomitant febrile illness or other known underlying chronic or severe diseases; declared consent from a parent or guardian and agreement for follow-up for 42 days; absence of repeated vomiting with the first dose of study medication; and hemoglobin > 8.0 g/dl. Microscopic blood examination was performed by trained microscopists using thick and thin Giemsa-stained blood smears on the same slide to determine parasite density and species, respectively. Laboratory results were not available until after enrollment, hence patients could be excluded following enrollment based on test results.
Randomisation: A computer generated randomization list was created for each of the study sites by an individual not involved in the study. Sequentially numbered, sealed, opaque envelopes containing the treatment assignment were prepared and secured in a locked cabinet accessible to the study nurse. The study nurse, who was unblinded, dispensed the assigned medications but was not involved in the evaluation of participants. Patients and other study staff were informed of treatment assignments after randomisation.
Interventions: Study participants were administered one of two ACTs: AL (Coartem® [Novartis] Basel, Switzerland; 20 mg artemether/120 mg lumefantrine) or DP (D-Artepp® [Forsun] Shanghai, China; 40 mg dihydroartemisinin/320 mg piperaquine) supplied by WHO. For both treatment arms, no food was given to the study participants, but the parents were advised to feed their children after administration of the study medication. All the doses of the study medication were administered as directly observed treatment by the study nurse and the study participants were monitored for 30 minutes after administration of study drugs. AL was administered twice a day for three days (six doses total), and DP was administered once a day for three days (three doses total). The dosage was determined using the weight-based dosage as indicated in the WHO malaria treatment guidelines (3). If a child vomited within 30 minutes of administration, the medication was readministered. Children with fever were given paracetamol and those with a hemoglobin level <10 g/ dL were treated with ferrous sulfate and anthelminthics as per Integrated Management of Childhood Illness Guidelines (24).
Study participants were followed daily for the first three days after initiation of treatment and then weekly thereafter for a total of 28 days (AL) or 42 days (DP). Study participants were also encouraged to visit the study clinic on any other day they felt unwell. At each follow-up visit, clinical response to treatment was monitored through a standardized history and physical examination, and parasitological response was assessed through examination of thick and thin blood films. Hemoglobin level was measured on days 0, 21, 28, 35, and 42 using a portable spectrophotometer (Hemocue). A dried blood spot (DBS) was collected on the day of enrollment and follow-up days for the parasite molecular studies.
Treatment failures as defined by WHO (13) were treated with quinine tablets (10 mg/kg) every 8 hours for 5 days. Patients with evidence of severe malaria or danger signs (including hemoglobin <5 g/dL, convulsions, lethargy, inability to drink or breastfeed, repeated vomiting, and inability to sit/stand because of weakness) were referred for treatment with parenteral artesunate. Participants who took antimalarial medications outside the study, experienced adverse events requiring a change in treatment, withdrew informed consent, or were lost to follow-up were excluded at the time of these events.
Malaria parasitemia was diagnosed using thick smears stained for 10 minutes with 10% Giemsa. Follow-up thick and thin smears were stained with 2% Giemsa for 30 minutes and used to determine the actual parasite density, species, and presence of gametocytes. Parasitaemia was measured by counting the number of asexual parasites against 200 leucocytes in thick blood films and thin films were used for detection of the different parasite species. Parasite density per µl of blood was calculated by multiplying the total number of parasite counted by 40, assuming that 1 µl of blood had a mean count of 8000 leucocytes (25, 26). When more than 500 parasites were identified before counting 200 leucocytes, counting was stopped and parasitemia was calculated using the actual number of leucocytes counted. A blood smear was declared negative when examination of 100 high power fields did not reveal the presence of malaria parasites. For quality control, each smear was read by two microscopists, with disagreements, defined as differences between the two microscopists in species diagnosis, in parasite density of > 50%, or in the presence of parasites, settled by a third microscopist. The final parasitemia was calculated as the average between the two readings for readings without disagreement. For readings with a disagreement, the final parasitemia was calculated as the average of the third microscopist’s and the closet of the first two readings; the third microscopist’s reading was taken as the final species.
Molecular markers of antimalarial drug resistance and microsatellite markers were analysed at the Centers for Disease Control and Prevention (CDC) Malaria Laboratory in Atlanta, USA(27, 28). Parasite genomic DNA was extracted from DBS collected on day 0 and day of recurrent infection using the QIAamp blood mini-kits (Qiagen GmbH, Hilden, Germany) according to the manufacturer’s instructions (29). The PET-PCR assay (30) was used to analyse the quality of the extracted parasite genomic DNA and to confirm the presence of P. falciparum.
Paired samples with recurrent infections (day 0 and day of recurrent infection) were analysed to distinguish recrudescence from reinfection using seven neutral microsatellite markers (TA1, Poly-α, PfPK2, TA109, TA2490, C2M34 and C3M69) over six chromosomes (31, 32). Fragment size was measured by capillary electrophoresis on ABI 3033 (Applied Biosystems) and scored using GeneMarker® V2.6.3 (SoftGenetic, LLC, PA, USA). A Bayesian probabilistic algorithm was used to distinguish recrudescent from new infections, accounting for classification uncertainty with multi-parasite genetic diversity (28, 33).
Sanger sequencing was used to investigate Pfk13 propeller domain (codon positions: 389–649), as previously described (34), in all the day 0 and day of recurrent infection samples in both study arms. The Pfmdr1 (codon positions: 86, 184, 1034, 1042, and 1246) mutations were investigated in paired day 0 and day of recurrent infection samples in the AL arm using a previously described method (35). SNPs were identified using the Geneious software package (Biomatters, Inc., San Francisco, CA). The 3D7 Pfk13 and Pfmdr1 were used as reference sequences. Heterozygous SNPs (double peaks representing mixed infections) were identified using the heterozygous caller plug-in tool in Geneious with a minor allele threshold of at least 30%. For samples with mixed infections and SNP variations at multiple sites, each possible haplotype constructed from the observed SNPs was reported for Pfmdr1. Detection of P. falciparum plasmepsin 2 copy number was performed in paired day 0 and recurrent infection samples in the DP arm using an Agilent Mx3005 real-time PCR machine (Agilent Technologies, California, USA), according to previously described protocols(36)
Outcomes: The primary outcome was parasitemia, assessed by microscopy within 28 (AL and DP) or 42 days (DP) of treatment, either unadjusted or adjusted to distinguish recrudescence from new infection. Recrudescence is defined as the recurrence of asexual parasitemia of the same genotype(s) that caused the original illness, due to incomplete clearance of asexual parasites after antimalarial treatment while new infection is an infection that follows a primary infection which is often (but not always) different from that which caused the initial infection(13, 14).
Outcomes were classified according to WHO guidelines(13) and included early treatment failure (ETF; danger signs, complicated malaria, or failure to adequately respond to therapy on days 0–3), late clinical failure (LCF; danger signs, complicated malaria, or fever and parasitemia on days 4–28/42), late parasitological failure (LPF; asymptomatic parasitemia on days 7–28/42), and adequate clinical and parasitological response (absence of parasitemia through follow-up). Secondary outcomes included prevalence of fever and parasitemia on days 1–3, the prevalence of gametocytes during follow-up, risk of adverse events, and presence of genetic polymorphisms associated with antimalarial drug resistance. Adverse events were evaluated at each study visit, graded according to WHO and U.S. National Institute of Allergy and Infectious Diseases scales, and defined based on International Conference on Harmonization guidelines as untoward medical occurrences. Serious adverse events included death, a life-threatening experience, hospitalization, incapacity, or events that required medical or surgical intervention to prevent serious outcomes(37, 38).
Sample size: Both AL and DP were studied at each site, with a target of 100 study participants in each treatment arm. The sample size calculation was based on estimated risks of treatment failure (outcomes classified as ETF, LCF, or LPF) of the treatment regimens to be studied (13). Based on prior data, the risk of treatment failure for both AL and DP were estimated at 5% with a 5% margin of error and 95% confidence interval. The calculated sample size was adjusted by 35% to account for recurrent infection rates, thus the total target sample size per site per treatment arm was 100.
Data management and statistical methods: Data were double entered into Microsoft Access and analyzed using Stata, version 14.2 (Stata). Per-protocol (proportional) and cumulative efficacy were calculated by arm and site at 28- or 42-days follow-up. Unadjusted outcomes were censored for loss to follow up or exclusion. Outcomes adjusted by genotyping were censored for loss to follow-up or exclusion, new infections, or failure of genotyping. The posterior probabilities of recrudescence generated using the Bayesian algorithm to distinguish recrudescence from reinfection were used to generate the per protocol (proportional) efficacies, and posterior sampling was used to generate the Kaplan-Meier estimates and 95% confidence intervals. Incident adverse events were reported by arm.