Study design
Data were collected as part of a six-arm clinical outcomes trial following the standard WHO in vivo therapeutic efficacy protocol [18]. In each of three provinces, the efficacy of two of Angola’s three first-line ACT was measured: AL and DP in the mesoendemic-stable northern province of Zaire, AL and ASAQ in the mesoendemic-stable central coastal province of Benguela, and DP and ASAQ in the hyperendemic eastern province of Lunda Sul (Fig. 1). In each provincial capital, two public health facilities were selected as sentinel sites for patient enrolment. In each province, each anti-malarial regimen was tested consecutively; there was no randomization of participants, and the study was not designed to compare efficacies of the three ACT.
Study population and enrolment
Children with symptomatic, uncomplicated Plasmodium falciparum infection presenting to the sentinel health facilities and meeting standard inclusion criteria [18] were invited to participate in the trial. To account for differences in transmission intensity, the target age range was 6 months to 5 years (59 months) in Zaire and Lunda Sul, and 6 months to 12 years (143 months) in Benguela. Similarly, the acceptable initial parasite density was 2000–200,000 parasites/µL in Zaire and Lunda Sul and 1000–100,000 parasites/µL in Benguela. In addition to parasite density, haemoglobin was measured during enrolment using HemoCue® 201 and 301 devices (AB Leo Diagnostics, Helsinborg, Sweden), with children having haemoglobin >5 g/dL eligible for enrolment.
The target sample enrolment was 100 children per arm, providing 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%.
Study procedures
Study participants were administered one of three anti-malarials: AL (Coartem®; Novartis, Basel, Switzerland), ASAQ (Winthrop; Sanofi Aventis, Paris, France), and DP (Duo-cotecxin®; Holley-Cotec, Beijing, China). Study participants in the AL arm in Benguela received the regular formulation of Coartem, whereas those in the AL arm in Zaire received the dispersible formulation. AL was procured by The President’s Malaria Initiative and provided by the Angolan National Malaria Control Programme, and the ASAQ and DP were procured and provided by WHO. All medications passed lot-quality testing, performed in laboratories in the USA (AL), the Netherlands (DP) and South Africa (ASAQ), prior to the study. The participants were treated according to manufacturers’ weight-based dosing recommendations. All three doses of ASAQ and DP were directly observed in the study health facilities. For AL, the morning doses were directly observed at the study health facilities, while the evening doses were given by participants’ guardians. To ensure appropriate administration of evening doses, guardians were counselled at the health facility and reminded in the evening by telephone. Guardians were also instructed to bring back empty blister packs the following day. Study participants not correctly completing the medication were withdrawn from the study.
Study participants were followed up daily for the first three days after initiation of treatment, and then weekly thereafter for a total of 28 days (AL and ASAQ arms) or 42 days (DP arms). At each follow-up visit, clinical response to treatment was monitored through a standardized history and physical examination and parasitological response was assessed through preparation and examination of thick blood films (with the exception of day 1). Guardians were asked about any adverse events experienced by the study participants. Blood slides were prepared and read by two dedicated study laboratory technicians following standard procedures [18], and a sub-set of slides was read by expert microscopists from the Angolan National Malaria Control Programme during supervisory visits and following study end. Dried blood spots were collected at enrolment and at all weekly follow-up visits. Additional blood spots were also collected on day 3 in Zaire and Benguela. Follow-up was suspended for all study participants developing any danger signs and/or meeting criteria for treatment failure. Treatment failures were treated with intravenous quinine. An electronic database containing clinical and parasitological data was updated daily using double data entry.
Molecular analysis
Dried blood spots were transported to CDC laboratories in Atlanta. DNA was extracted using a QIAamp Blood DNA kit (Qiagen, Hilden, Germany) from day 0 and day-of-failure samples for all treatment failures and day-0 samples from 50 randomly selected participants with adequate clinical and parasitological response (ACPR).
For all treatment failure samples and the randomly selected treatment successes, the lengths of six microsatellites [19] were measured: TA1, Polyα, PfPK2, C2M34, C3M69, and 2490. A previously described protocol for cycling [20] was modified for this study. Primer pairs with annealing temperature in the 50–60 °C range were designed and cycling conditions were optimized according to melting temperature for each primer pair. The sizes of the amplified products were assayed by capillary electrophoresis on an Applied Biosystems 3130xl sequencer (Applied Biosystems, Foster City, CA, USA). Microsatellite fragment analysis was performed using the Geneious Pro R8 microsatellite plug-in (Biomatters, San Francisco, CA, USA).
Regions in the P. falciparum K13, P. falciparum multidrug-resistance gene 1 (pfmdr1), and P. falciparum cytochrome b (pfcytb) genes were amplified and sequenced for day 0 and day of failure for all treatment failures. The P. falciparum K13 propeller domain was amplified using a P. falciparum-specific protocol described previously [21]. Two regions of pfmdr1, covering codons 86–184 and 1034–1246, were amplified using a previously described protocol [22]. The pfcytb gene was amplified using the following primers: forward primer, 5′ CTATTAATTTAGTTAAAGCACAC 3′ and reverse primer, 5′ ACAGAATAATCTCTAGCACCA 3′. 1 µL of genomic DNA was amplified using 0.5 µM of each primer, 0.2 mM dNTP, 2 mM MgCl2 and 1 U High Fidelity Taq (New England Biolabs, Ipswich, MA, USA). For the primary reaction, the following cycling parameters were used: 2 min at 98 °C, 35 cycles of 98 °C for 10 s, 46 °C for 30 s, 68 °C for 2:30 min, and final extension for 10 min at 68 °C. Amplified PCR products were visualized on a 2% agarose gel after electrophoresis. Sequencing of PCR products was performed using an ABI 3730 sequencer (Applied Biosystems, Foster City, CA, USA).
Sanger sequences were analysed using Geneious R8 software (Biomatters, San Francisco, CA, USA) to identify specific single-nucleotide polymorphism combinations. Single-nucleotide polymorphisms were only called if both the forward and reverse strands had the mutation.
Statistical analysis
Study participants not meeting any exclusion criteria were classified as one of three possible study endpoints. Study participants showing signs of severe disease while parasitaemic in the first 3 days following treatment, having a parasite density on day 2 that exceeded day 0 parasite density, clearing less than 75% of day 0 parasites on day 3, or still febrile (≥37.5 °C axillary) and parasitaemic on day 3 were classified as early treatment failures (ETFs). Study participants that were febrile and parasitaemic after day 3 or parasitaemic regardless of body temperature after day 6 were classified as late treatment failures (LTFs). Participants not meeting these criteria with a negative slide on the last day of follow-up were classified as having an ACPR.
Kaplan–Meier estimates of the survival curve for each arm were calculated and plotted. Uncorrected efficacy was calculated as the proportion ACPR (per-protocol method), as well as the Kaplan–Meier estimate of the survival function at day 28 and/or 42. Microsatellite data for LTFs were analysed using a previously published statistical algorithm which assigns each LTF a posterior probability of recrudescence [23]. LTFs with a posterior probability of recrudescence >0.5 were classified as likely recrudescences; all others were classified as re-infections. The point estimates and 95% confidence intervals of the PCR-corrected efficacies were calculated by sampling from the posterior distribution of classifications to account for the uncertainty of the correction [23]. Statistical analysis was performed using R version 3.0.1 (R Foundation for Statistical Computing, Vienna, Austria).
Ethical considerations
Parents or guardians of all study participants provided written informed consent prior to enrolment and were reimbursed for transport costs for clinic visits. The activity was classified as non-research by human subjects research boards at CDC and the Angolan Ministry of Health and was approved by the WHO Ethical Research Committee.