Continued Low Efficacy of Artemether-Lumefantrine in Angola in 2019

Biennial therapeutic efficacy monitoring is a crucial activity for ensuring the efficacy of currently used artemisinin-based combination therapy in Angola. Children with acute uncomplicated Plasmodium falciparum infection in sentinel sites in the Benguela, Zaire, and Lunda Sul Provinces were treated with artemether-lumefantrine (AL) or artesunate-amodiaquine (ASAQ) and monitored for 28 days to assess clinical and parasitological responses. Molecular correction was performed using seven microsatellite markers.

risk of its spread to sub-Saharan Africa, mirroring the emergence and spread of chloroquine and sulfadoxine-pyrimethamine resistance (2,3). In addition, there have been recent reports of the independent emergence of artemisinin resistance in Rwanda (4). To mitigate these threats, the World Health Organization (WHO) recommends periodic reassessment of the efficacy of currently used antimalarials in countries where the disease is endemic (5).
While initial pre-ACT-era trials in Angola showed 100% efficacies of AL and ASAQ (6), routine biennial therapeutic efficacy monitoring in Angola's three sentinel sites has shown signs of AL efficacy hovering around the crucial WHO 90% threshold in the Zaire Province site (7-9) ( Table 1). In contrast, the lowest observed efficacies for ASAQ and DP in any site have been 93.3% and 98.8%, respectively.
Characterization of molecular markers of resistance in samples collected during therapeutic efficacy monitoring complements and facilitates the interpretation of the clinical data. To date, as in most sub-Saharan settings, no pfk13 mutations associated with artemisinin resistance have been detected in Angola either as part of therapeutic efficacy monitoring or in separate molecular surveys (7)(8)(9)(10)(11)(12). Molecular surveys in Angola have shown a predominance of the K76T pfcrt mutation (10), conclusively linked to chloroquine resistance and also potentially associated with amodiaquine resistance but lumefantrine sensitivity (13). Consistent with this putative relationship, all ASAQ treatment failure samples in the 2017 round of Angolan monitoring had the 76T mutated pfcrt allele, whereas 88% of AL treatment failures had the K76 wild-type allele. Analysis of AL treatment failure samples from the 2013, 2015, and 2017 rounds found a predominance of the N86 pfmdr1 allele previously associated with decreased sensitivity to lumefantrine (13). In addition, in the 2015 study, all pretreatment samples regardless of treatment outcome were sequenced for pfmdr1, and a large majority (86%) had the N86 allele present (11).
Here, the results of the fourth round of therapeutic efficacy monitoring in Angola's three fixed sentinel sites are reported.

RESULTS
Across all study arms, 3,616 children were screened, and 610 (16.8%) entered the study (Table 2). Twenty (3.3%) were lost to follow-up, and a further 11 (1.8%) were excluded. Rates of loss to follow-up were Յ5% across all arms. The median age varied between 2.8 years in the Zaire arms and 7.5 years in the Benguela ASAQ arm ( Table 2). The median parasite density and hemoglobin level at enrollment varied by study arm between 19,114 and 44,467 parasites/l and 10.0 to 10.8 g/dl, respectively.
The rate of day 2 slide negativity was 68% (95% confidence interval [CI], 58 to 77%) in the AL arm in Lunda Sul but was above 80% in all other arms (Table 3). Day 3 slide negativity rates ranged between 95% and 100%. There were 1 early treatment failure and 53 cases of recurrent parasitemia (Table 4). Of the 53 cases of recurrent parasitemia, 24 had a probability of recrudescence of Ͼ0.5 (Table 4; see also Tables S1 and S2 in the supplemental material).
The uncorrected Kaplan-Meier estimates of the day 28 efficacy were 84.2% (95% CI, All 103 sequenced samples from patients with early treatment failure at day 0, and day 0 and day-of-failure samples from participants with recurrent parasitemia, were wild type for pfk13 (Table 6). Nearly all (92%; 11/12) ASAQ day-of-failure samples from cases of recurrent parasitemia had the 76T pfcrt allele. In contrast, only 16% (4/25) of AL day-of-failure samples had this allele. All (21/21) AL and most (93%; 13/14) ASAQ day-of-failure samples had the N86 pfmdr1 allele.

DISCUSSION
For the third time in four rounds of Angolan therapeutic efficacy monitoring, one of the AL arms was observed to have a corrected efficacy below the key 90% threshold set by the WHO. While previous rounds have shown suboptimal AL efficacy in Zaire, the AL efficacy in Zaire for this round was 92.2%. In contrast, the AL efficacy in Lunda Sul for this round was 87.6%, compared to 96.5% in 2017. Consistent with previous findings, the ASAQ efficacy was uniformly high (above 95%) in all three sites.
Recent reports of AL efficacy estimates below 90% from the Mikalayi site in the Democratic Republic of the Congo (DRC), less than 500 km from Lunda Sul and the closest efficacy monitoring site to Lunda Sul, provide additional context of potential warning signs for AL in the region. Moreover, the low day 28 AL efficacy in Lunda Sul in this study was observed in the backdrop of a relatively low rate of day 2 clearance, as only 68% of participants were microscopy negative on day 2, compared to 97% in    (14) is a likely explanation for the large number of reinfections in the AL arms, the large number of recrudescences in the AL Lunda Sul arm likely reflects underlying decreased susceptibility to lumefantrine in the parasite population. There was no molecular evidence of artemisinin resistance, as all failure samples had wild-type pfk13 sequences. The results of analyses of pfcrt and pfmdr1 sequences were consistent with literature showing an overrepresentation of the 76T pfcrt allele in amodiaquine treatment failures and a predominance of the N86 pfmdr1 allele in AL treatment failures. Interpretation of the latter finding should be done in the context of previous reports of the high population prevalence of the N86 allele in parasites circulating in Angola and the high prevalence of the N86 allele even in ASAQ treatment failures in this study. Molecular characterization of all day 0 samples, not just failures, will allow a better understanding of the population prevalence of these markers. Notably, fixation or near fixation of the N86 allele, together with the clinical evidence of reduced efficacy, might have implications for the future of AL use in Angola.
The unavailability of quality-controlled DP precluded an assessment of its efficacy in this round of monitoring in Angola. The recent introduction of new prequalified formulations should ensure that DP returns to being evaluated in future rounds. The inability to directly observe evening AL doses means that lumefantrine drug level measurement might be warranted, as in previous rounds (9), especially in the Lunda Sul site. Consecutive assessment of the two study drugs, as opposed to randomization, and small sample sizes do not allow direct comparison of drug efficacies. Moreover, because estimates of drug efficacies are close to 90%, substantially larger sample sizes will be required for a definitive determination of whether the true efficacy is above or below this threshold in future rounds. In the meantime, consideration of alternatives to AL in Angola may be warranted. In a larger context, our findings of consistent long-term trends of suboptimal efficacy of AL in Angola, together with recent results showing

Study design.
A prospective clinical outcome trial was conducted according to the standard 2009 WHO in vivo protocol (5). Two drugs, AL and ASAQ, were consecutively tested in each of the three sites for a total of six treatment groups. The three sentinel sites included M'Banza Congo in the hyperendemic Zaire Province in the north of the country, Saurimo in the hyperendemic Lunda Sul Province in the east, and the eponymous capital of the mesoendemic Benguela Province on the central western coast (see Fig.  S1 in the supplemental material). The study was conducted from March to July 2019, coinciding with the high-malaria-transmission period.
Study population. Children with uncomplicated P. falciparum monoinfection presenting with febrile illness to sentinel site clinics were eligible for inclusion. Age and parasite density-based eligibility criteria reflected the different transmission levels, as dictated in the WHO protocol: 6 to 59 months of age and 2,000 to 200,000 parasites/l in Zaire and Lunda Sul versus 6 to 143 months of age and 1,000 to 100,000 parasites/l in Benguela. Children with signs of severe illness, antimalarial use in the preceding 2 weeks, allergy or hypersensitivity to the drugs, or hemoglobin levels of Ͻ8 g/dl were excluded from enrollment. A target of 100 participants per arm was chosen to provide a precision of 5% assuming a 25% combined exclusion and loss-to-follow-up rate and 95% efficacy.
Study procedures. Children were treated with either AL (Cipla, Mumbai, India) or ASAQ (Sanofi-Aventis, Maphar, Morocco) for 3 days. The dosage was weight based according to the manufacturers' recommendations. All ASAQ doses were directly observed by study staff and were administered with water or juice. For AL, morning doses were observed by study staff and were administered with yogurt or milk. Parents or guardians were given the evening dose and a yogurt or milk packet to give at home. Study staff telephoned parents or guardians in the evening to remind them to administer the evening dose, and compliance was further assessed by requesting parents or guardians to bring empty blister packets to the clinic the following day.
Children were monitored daily for the first 4 days and then weekly thereafter for a total of 28 days. At each visit, study staff performed clinical exams, performed blood smears (except for day 1), and collected blood on Whatman 903 filter paper (GE Healthcare Life Sciences, Marlborough, MA, USA). Hemoglobin was assessed fortnightly using HemoCue Hb301 (Hemocue, Ängelholm, Sweden) or DiaSpect (EKF Diagnostics, Barleben, Germany) hemoglobinometers.
Laboratory analysis. Blood smear microscopy was done using thick and thin smears (5) by two microscopists. Discrepancies of more than 20% between parasite density estimates were resolved by a reading by a third microscopist. Monthly supervision visits included quality control of a randomly selected 10% of slides.  (15). Paired day 0 and day-of-failure samples from cases of recurrent parasitemia and 60 randomly selected day 0 nonfailure samples were analyzed for seven microsatellite markers (TA1, poly-␣, PfPK2, TA109, TA2490, C2M34, and C3M69) (16,17). Following photo-induced electron transfer (PET)-PCR (18) confirmation of the presence of P. falciparum, fragment lengths of the microsatellite markers were measured using an ABI3130 xl genetic analyzer (Applied Biosystems, Foster City, CA) capillary sequencer and GeneMapper software (Applied Biosystems, Foster City, CA). Day 0 samples from early treatment failures (before day 7) and day 0 and day-of-failure samples from cases of recurrent parasitemia (on or after day 7) were analyzed using a targeted deep-amplicon-sequencing workflow that included the full pfk13, pfcrt, and pfmdr1 genes (19).
Statistical analysis. Data were double entered into an Excel database (Microsoft, Redmond, WA). Baseline characteristics were summarized by treatment group. The proportions of participants with negative slides on day 2 and day 3 were calculated. Individuals finishing follow-up were classified as early treatment failures, cases of recurrent parasitemia, or cases of adequate clinical and parasitological response (ACPR) according to standard WHO definitions (5). For molecular correction, microsatellite genotyping data from day 0 and day-of-failure samples from cases of recurrent parasitemia were compared using a Bayesian algorithm, and the posterior probability of recrudescence was calculated based on the observed allele data (20,21). For tabulation, recrudescences were defined when the posterior probability of recrudescence was Ͼ0.5. The corrected efficacy and corresponding 95% confidence intervals (CIs) were calculated using the Kaplan-Meier estimate of the survival function at day 28 and the posterior sample of calculated probabilities of recrudescence (20).
Ethical review. The study was reviewed by the human subjects review boards at the Angolan Ministry of Health, and the activity was approved as a nonresearch program evaluation by the Office of the Associate Director for Science at the CDC's Center for Global Health (protocol 2014-233c). Parents or guardians of study participants provided written informed consent.
Data availability. The full clinical data set has been uploaded to the WorldWide Antimalarial Resistance Network and WHO repositories. Full microsatellite genotyping data are available in Tables S1 and S2 in the supplemental material.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.5 MB.