Molecular surveillance for polymorphisms associated with artemisinin-based combination therapy resistance in Plasmodium falciparum isolates collected in Mozambique, 2018

Due to the threat of emerging anti-malarial resistance, the World Health Organization recommends incorporating surveillance for molecular markers of anti-malarial resistance into routine therapeutic efficacy studies (TESs). In 2018, a TES of artemether-lumefantrine (AL) and artesunate-amodiaquine (ASAQ) was conducted in Mozambique, and the prevalence of polymorphisms in the pfk13, pfcrt, and pfmdr1 genes associated with drug resistance was investigated. Children aged 6–59 months were enrolled in four study sites. Blood was collected and dried on filter paper from participants who developed fever within 28 days of initial malaria treatment. All samples were first screened for Plasmodium falciparum using a multiplex real-time PCR assay, and polymorphisms in the pfk13, pfcrt, and pfmdr1 genes were investigated by Sanger sequencing. No pfk13 mutations, associated with artemisinin partial resistance, were observed. The only pfcrt haplotype observed was the wild type CVMNK (codons 72–76), associated with chloroquine sensitivity. Polymorphisms in pfmdr1 were only observed at codon 184, with the mutant 184F in 43/109 (39.4%) of the samples, wild type Y184 in 42/109 (38.5%), and mixed 184F/Y in 24/109 (22.0%). All samples possessed N86 and D1246 at these two codons. In 2018, no markers of artemisinin resistance were documented. Molecular surveillance should continue to monitor the prevalence of these markers to inform decisions on malaria treatment in Mozambique.

falciparum is the predominant malaria parasite species in the country [3].
One of the fundamental steps toward malaria control is the rapid diagnosis and correct treatment of symptomatic cases with an effective anti-malarial [4]. Antimalarial drug resistance continues to be a major hurdle to malaria control efforts in some settings. Since replacing chloroquine (CQ) with a combination of amodiaquine (AQ) + sulfadoxine-pyrimethamine (SP) for uncomplicated malaria treatment in 2003, the Mozambique national treatment guidelines have experienced various adjustments as parasites became resistant to treatments [5]. In 2006, artemisinin-based combination therapy (ACT) was formally introduced by adopting artesunate (AS) + SP as a first-line treatment for uncomplicated P. falciparum malaria [6,7]. Subsequently, the last change occurred in 2009, when the country introduced artemether-lumefantrine (AL) and artesunate-amodiaquine (ASAQ) as the official first-line treatments, with ASAQ as a backup in situations when AL is contraindicated [6][7][8]. While ASAQ is part of the national treatment algorithm, there has been limited procurement and use. In artemisinin-based combinations, the artemisinin component is short-acting and kills the majority of parasites within the first 2 days of treatment; the remaining parasites are cleared by the longer-acting partner drug [9], thus helping to abate the acquisition of parasite resistance to the treatment. However, resistance to artemisinin derivatives, defined as delayed parasite clearance (presence of > 10% parasitaemia on day 3 after the start of treatment), has been reported in Southeast Asia [10][11][12] and Rwanda [13]. Resistance to specific anti-malarials is associated with polymorphisms, such as a single nucleotide polymorphisms (SNPs), a combination of SNPs, or gene copy number variation in drug target genes.
To monitor the efficacy of anti-malarial treatment, the World Health Organization (WHO) recommends periodic (at least every 2 years) monitoring of the first and second-line anti-malarial treatments [14] and, in addition, molecular surveillance of resistance markers is encouraged. Artemisinin partial resistance is associated with polymorphisms in the P. falciparum kelch 13 (pfk13) gene [10] and ten SNPs in pfk13 gene are currently validated molecular markers for artemisinin partial resistance: F446I, N458Y, M476I, Y493H, R539T, I543T, P553L, R561H, P574L and C580Y [14]. One of these mutations, R561H, has been reported to be present in multiple samples from different sites in Rwanda [13,15,16], highlighting the importance of conducting molecular surveillance to identify emerging artemisinin and partner drug resistance genotypes. To date, there have been no reports of pfk13 polymorphisms associated with artemisinin partial resistance in Mozambique [17,18].
Resistance to CQ is mainly associated with SNPs in the P. falciparum chloroquine resistance transporter (pfcrt) gene, resulting in an amino acid change from lysine (K76) to threonine (76T) at position 76; however, the P. falciparum multi-drug resistance (pfmdr1) gene may also play a role in CQ resistance [19,20]. The most commonly reported pfcrt mutations are observed in codons 72, 74-76 [21]. The wild type CVMNK haplotype is associated with CQ sensitivity, while the CVIET and SVMNT haplotypes are associated with CQ resistance, with CVIET being the more common of the latter two haplotypes in Africa [22,23].
The pfmdr1 gene is implicated in lower sensitivity or tolerance to several anti-malarial drugs, including CQ, AQ, and lumefantrine [22,24]. In Africa, the most relevant polymorphisms of pfmdr1 include N86Y, Y184F and D1246Y [8,25]. Mutations at positions S1034C and N1042D of pfmdr1 are rarely reported on the continent [8,26]. The 86Y mutation has been associated with decreased CQ and AQ sensitivity, while the N86 wild type codon has been implicated in decreased sensitivity to lumefantrine. The N86, 184F, and D1246 (NFD) haplotype is associated with decreased sensitivity to AL, while the 86Y, Y184, and 1246Y (YYY) haplotype is reported to be associated with decreased sensitivity to ASAQ [25]. In Mozambique, the prevalence of pfmdr1 mutations was low in the capital city of Maputo, although the alleles N86 and 184F showed a significantly increased prevalence after the introduction of ACT [8].
Molecular surveillance for drug resistant parasites is part of a comprehensive approach along with TESs for early detection and subsequent prevention of spread of resistant parasites by permitting timely implementation of appropriate alternative treatment policy decisions. This study's aim was to analyse the prevalence of molecular markers associated with P. falciparum resistance to anti-malarial drugs in the pfk13, pfmdr1, and pfcrt genes in samples collected during a 2018 TES in four sentinel sites in Mozambique.

Study sites
This study was a sub-study of a TES that evaluated the efficacy and safety of AL and ASAQ in the treatment of uncomplicated P. falciparum malaria in children aged 6-59 months in Mozambique, based on WHO-recommended protocol [27]. Malaria transmission in the country is year-round, with seasonal peaks during and after the rainy season, which occurs between October and March. The peak of the malaria transmission extends from November into April [3]. This study was conducted between February and September 2018 in four sentinel sites: Rural Hospital of Montepuez, in Cabo Delgado Province (Northern region), Moatize Health Center, in Tete Province (Central region), District Hospital of Mopeia, in Zambézia (Central region), and District Hospital of Massinga, in Inhambane Province (Southern region) (Fig. 1). These sites are distributed across the Northern, Central, and Southern regions of Mozambique, which represent areas with high, moderate, and low prevalence of malaria, respectively. The per protocol PCR-corrected efficacy results of this study will be reported elsewhere, but were greater than 95% for all four AL arms and greater than 98% for all three ASAQ arms (no ASAQ arm in Moatize).

Sample collection
Potential participants were screened for malaria parasites using microscopy at each study site. Patients were eligible for enrolment if they had uncomplicated P. falciparum mono-infection with an asexual blood density between 2000 and 200,000/µL, were aged 6-59 months, and had a fever at presentation (axillary temperature ≥ 37.5 °C) or history of fever in the last 24 h. A dried blood spot on Whatman 3-mm filter paper was prepared using 50 µL of blood collected on the day of enrolment (day 0/ pre-treatment) and on any other day the patient had a recurrent malaria infection during the follow-up period (post-treatment).

DNA extraction
DNA was extracted at the Manhiça Health Research Center Laboratory, Mozambique, from half of the dried blood spot using a QIAamp DNA Mini kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions. The DNA was eluted in 150μL of elution buffer, aliquoted and transferred to the CDC Malaria Laboratory in Atlanta, GA, USA, for molecular analysis.

Molecular genotyping of resistance markers
Molecular analysis for drug resistance markers was performed by a laboratory technician from Mozambique with the support of staff from the CDC Malaria Laboratory in Atlanta, USA [28]. For this analysis, selected pre-treatment and all post-treatment samples were used. Samples were first screened using a multiplex real-time PCR assay (PET-PCR) for detection of Plasmodium genus and P. falciparum, as previously described [29]. Polymorphisms in the pfk13 (propeller domain 389-649), pfcrt (codons 72-76), and pfmdr1 (codons 86, 184, 1034, 1042, and 1246) genes were investigated as previously described [30,31]. Briefly, both pre-treatment and post-treatment samples were used to amplify fragments of pfk13, pfcrt, and pfmdr1 by nested PCRs. Three laboratory P. falciparum parasite lines, 3D7, 7G8, and Dd2, were included as controls. Direct Sanger sequencing of the purified nested PCR products was performed using a BigDye Terminator v3.1 cycle sequencing kit on an iCycler thermal cycler (Bio-Rad, CA, USA). The reaction mixtures were precipitated in 70% ethanol to clean up dye terminators, rehydrated in 10 μL HiDi formamide, and then sequenced on a 3130xl ABI genetic analyzer (ABI Prism, CA, USA). Sequence analysis was performed using Geneious R7 (Biomatters, Auckland, New Zealand). Raw sequence reads were cleaned using default settings and reads with high-quality scores (> 30%) were further analysed using the 3D7 pfk13, pfcrt, and pfmdr1 genes as references.

Data analyses
Data were entered into a Microsoft Office Excel 2007 sheet and then exported into R 3.6.0 (R Core Team 2019) for validation, cleaning, and analysis. A statistical significance of difference in the risk of treatment failure (reinfection or recrudescence) was determined by Fisher's exact test, at a 5% significance level. All possible haplotypes from mixed infections (both wild type and mutants) were included in construction of the pfmdr1 haplotype.

Characteristics of study subjects
From the 641 patients enrolled in the TES, 110 (17%) pretreatment samples were selected for the analysis. This included all the pre-treatment samples from subjects who returned with a recurrent infection (n = 51) and 10% randomly selected pre-treatment samples from patients who did not have a recurrent infection (n = 59); however, one sample was excluded due to poor quality DNA, leaving 109 pre-treatment samples. All 51 post-treatment samples from patients who had a recurrent malaria infection were included in the analysis (Fig. 2). In the AL and ASAQ arms, 7.1% (26/368) and 3.0% (8/273) of subjects, respectively, remained parasitaemic at day 3, although no subject met criteria for early treatment failure. Table 1 summarizes the characteristics by site of 109 study participants used for molecular analysis. A total of 79 and 30 pre-treatment samples, and 48 and 3 posttreatment samples, were in the AL and ASAQ treatment arms, respectively.

Molecular markers of drug resistance pfk13 polymorphisms
All 109 pre-treatment samples and 48/51 (94.1%) of the post-treatment samples were successfully sequenced at the pfk13 gene. No polymorphisms associated with artemisinin partial resistance were observed in the propeller domain One sample from Mopeia contained a synonymous mutation at codon 469 (TGC to TGT) and three samples from Mopeia contained a synonymous mutation at codon 548 (GGC to GGT). No other synonymous or nonsynonymous mutations were found.

pfcrt polymorphisms
All 109 pre-treatment samples and 47/51 (92.2%) of the post-treatment samples were successfully sequenced at the pfcrt gene. All samples showed the wild type CVMNK haplotype.
In the pre-treatment samples, NFD was present in 29 (74.4%) and 8 (66.7%) samples from Massinga and Moatize, respectively, while NYD was present in 15 (75.0%) and 26 (68.4%) samples from Montepuez and Mopeia, respectively. For the late treatment failure samples, the NFD was observed in all samples from Moatize and Montepuez and in 72.7% of the samples from Massinga and Mopeia (Table 3).

Discussion
Mozambique has used AL and ASAQ as the two firstline anti-malarial regimens since 2009, with AL being the most widely used and ASAQ as backup for situations in which AL could not be used or is not available. This study, provides insights into the pfk13, pfcrt, and pfmdr1 genetic profiles of P. falciparum isolates from four sentinel sites throughout the country. In this study, no pfk13 mutations associated with artemisinin partial resistance were observed. These findings are encouraging and suggest that artemisinin partial resistance has not yet emerged in the four study sites selected in Mozambique. A previous study in Mozambique revealed a very low prevalence (< 1%) of four polymorphisms in the pfk13 gene (L619L, F656I, V666V, and G690G) [17]. Another study revealed the presence of a V494I K13 polymorphism, found in two samples collected after the introduction of ACT in Mozambique [32]; However, these aforementioned mutations are either synonymous or not known to be associated with artemisinin partial resistance [17]. This current study's findings are also consistent with most reports from Africa in which no, or a very low prevalence of, pfk13 mutations have been reported [33,34]. The absence of delayed parasite clearance and pfk13 mutations known to be associated with artemisinin partial resistance is reassuring for Mozambique, at least in the short term. Nevertheless, a recently published  Hb in g/dL (mean ± SD) 8.2 ± 1.7 9.5 ± 1.1 9.6 ± 1.9 9.4 ± 1.6 9 ± 1.7 Rwandan study, using samples collected between 2012 and 2015, showed that 7.4% of the specimens carried the pfk13 R561H mutation [15], known to be associated with artemisinin partial resistance. Another recent Rwandan study also detected the presence of two validated markers of artemisinin partial resistance, R561H and P574L, and delayed parasite clearance (parasitaemia at day 3) in more than 10% of the study participants in two sites [13]. Although this finding was not linked to clinical treatment failure of AL, it highlights the importance of conducting molecular surveillance to identify emerging patterns of parasites with artemisinin and partner drug resistance genotypes.
The pfcrt data from this study showed that all samples sequenced contained the wild type pfcrt haplotype (CVMNK), suggesting the return of chloroquine sensitive alleles after its use was discontinued in 2003 in Mozambique. This finding is consistent with data from previous studies carried out in Mozambique that reported an increasing frequency of the pfcrt CVMNK wild type, from 43.9 to 66.4% between 2009 and 2010 [22]. A Mozambique study from 2015 reported a very low prevalence of mutant alleles at codons M74I, N75E, and K76T; only 2.3% samples harbored the pfcrt SNP 76T [17]. A separate report from 2015 reported only 0.9% samples with the pfcrt 76T mutant allele, 3.7% samples with a mixed infection, and 95.4% samples with the wild type allele [35]. These findings confirm the likely return of CQ-susceptible P. falciparum and are similar to findings from studies conducted in other African countries that also observed a resurgence in the proportion of wild type pfcrt alleles after the discontinuation of CQ for treatment [36][37][38]. AL has been shown to select for pfcrt wild types [22,39], and the widespread use of AL in most African countries may contribute to the re-emergence of these alleles associated with CQ sensitivity [39].
Polymorphisms in pfmdr1 were only observed at codon 184, resulting in two observed haplotypes, NFD Table 2 Prevalence of pfmdr1 184 polymorphisms in pre-treatment and post-treatment samples stratified by treatment arms AL, artemether-lumefantrine; ASAQ, artesunate-amodiaquine *Statistical significance in risk of recurrent infection (reinfection or recrudescence) was determined by Fisher's exact test; three post-treatment samples (two in the AL arm and one in the ASAQ arm) failed to amplify at one or more loci and are not included in corresponding single nucleotide polymorphism and haplotype counts a Haplotype percentages may not sum to 100% because all possible haplotypes from mixed infections (both wild type and mutants) were included in the construction of haplotypes  and NYD. This is consistent with previous findings from Mozambique. In 2015, a low prevalence of 86Y (3.1%) and a higher prevalence of 184F (46.7%) were reported [17]. In addition, a high prevalence of wild type N86 (73.2%) and D1246 (96.7%) and the presence of the mutant 184F (22.7%) were reported in a 2010-2012 study [8].
The pfmdr1 gene has been implicated in lower sensitivity or tolerance to several anti-malarial drugs, including lumefantrine, CQ, and AQ [22,24], with the 86Y mutation being associated with decreased CQ and AQ sensitivity and the N86 wild type allele implicated in decreased sensitivity to lumefantrine [25]. The NFD haplotype increased in prevalence between the pre-and post-treatment samples in this study's AL arm, but this was not significant when compared with NYD. In Mozambique, NFD haplotype prevalence increased from approximately 22-38% between 2009 and 2010 [22]. While the sites from this study are not comparable to that report, the 61.5% pre-treatment NFD prevalence indicates that this haplotype is still circulating. Similar findings over time have been reported in other African countries in which AL was used as the first-line anti-malarial treatment [40,41]. Some studies showed that the pfmdr1 gene polymorphism at codons N86Y, Y184F, and D1246Y is mainly linked to AL or ASAQ drug pressure [42,43]. Stratifying by site, NFD was identified in 100% of post-treatment samples from Moatize and Montepuez and 72.7% of samples from Massinga and Mopeia, although this was not statistically significant. Notable limitations of this study include a low sample size in some sites, due to few late recurrences, and the limited number of pre-treatment samples analysed, due to budgetary restrictions.

Conclusion
Given that no pfk13 or pfcrt molecular markers of resistance were observed, the results of this study corroborate the findings of the associated TES that showed AL and ASAQ were efficacious. The high prevalence of the pfmdr1 NFD haplotype, associated with decreased sensitivity to lumefantrine in some studies, requires further investigation to fully understand the role of this haplotype in the sensitivity of the currently used artemisinin-based combinations, AL and ASAQ. Because alleles associated with artemisinin partial resistance are emerging in the East Africa region, continued molecular surveillance for early detection of these alleles as well as relevant partner drug resistant markers remains important.