Drug-Resistant Polymorphisms and Copy Numbers in Plasmodium falciparum, Mozambique, 2015

One of the fundamental steps toward malaria control is the use of antimalarial drugs. The success of antimalarial treatment can be affected by the presence of drug-resistant populations of Plasmodium falciparum. To assess resistance, we used molecular methods to examine 351 P. falciparum isolates collected from 4 sentinel sites in Mozambique for K13, pfmdr1, pfcrt, and pfdhps polymorphisms and for plasmepsin2 (pfpm2) and pfmdr1 copy numbers. We found multiple copies of pfpm2 in 1.1% of isolates. All isolates carried K13 wild-type alleles (3D7-like), except 4 novel polymorphisms (Leu619Leu, Phe656Ile, Val666Val, Gly690Gly). Prevalence of isolates with pfcrt mutant (K76T) allele was low (2.3%). Prevalence of isolates with pfdhps mutant alleles (A437G and K540E) was >80%, indicating persistence of sulfadoxine/pyrimethamine resistance; however, markers of artemisinin were absent, and markers of piperaquine resistance were low. Piperaquine resistance isolates may spread in Mozambique as dihydroartemisinin/piperaquine drug pressure increases.

D uring the past decade, malaria control strategies have substantially reduced the malaria burden worldwide; several countries are advancing toward malaria elimination (1,2). A fundamental pillar for contributing to the reduction of the malaria burden has been artemisinin-based combination therapy. Unfortunately, the effectiveness of antimalarial drugs used for malaria treatment and chemoprevention during pregnancy has been threatened by the emergence of drug-resistant parasite populations (2)(3)(4)(5).
The emergence of artemisinin resistance in Plasmodium falciparum, with reduced in vivo susceptibility to artesunate, was reported in Southeast Asia (3,6). Detectable polymorphisms in the Kelch 13 (K13) propeller domain in P. falciparum associated with artemisinin resistance have subsequently provided an additional tool for monitoring resistance to antimalarial drugs (7,8). In Cambodia, polymorphisms in the K13 propeller domain (mainly Y493H, R539T, I543T, and C580Y) were associated with in vitro prolonged parasite survival rates and in vivo delayed parasite clearance rates (8,9). Recently, plasmepsin 2 (pfpm2) copy number and pfcrt C101F polymorphism have been associated with piperaquine resistance (10)(11)(12). In addition, increased pfmdr1 copies have been associated with resistance to mefloquine (in vivo, in vitro, or both) and partially to lumefantrine (13)(14)(15)(16)(17)(18). Specific point polymorphisms (at codons 86, 184, 1034, 1042, and 1246) of the pfmdr1 gene have also been linked to resistance to antimalarial drugs (19,20). In field isolates tested in vitro as well as in laboratory lines, N86Y polymorphism was associated with chloroquine resistance (21). Further, polymorphisms in the pfcrt gene have also been shown to affect parasite susceptibility to chloroquine (22), amodiaquine (23,24), and artemether/ lumefantrine (25). Recently, a nonsynonymous polymorphism in the pfcrt gene was shown to be prevalent in the genetic background of K13 mutant artemisinin-resistant isolates (26). In addition, polymorphisms in pfdhfr and pfdhps genes, specifically the quintuple mutant, including the pfdhfr substitutions N51I, C59R, and S108N, as well as the pfdhps substitutions A437G and K540E, have been associated with a failure of sulfadoxine/pyrimethamine treatment against uncomplicated P. falciparum malaria (27). In Africa, the pfdhps K540E polymorphism has been considered a useful epidemiologic marker of the quintuple mutations (28).

Drug-Resistant Polymorphisms and Copy Numbers in Plasmodium falciparum, Mozambique, 2015
The development of drug resistance could be influenced by multiple factors such as polymorphism rate, fitness costs, overall parasite load, strength of drug selection, treatment compliance, transmission intensity, host immunity, and erythrocyte disorders (29)(30)(31). Naturally acquired immunity plays a major role in the emergence and clearance of artemisinin-resistant parasites (32). Because of increasing concern over the effectiveness of the nationally recommended antimalarial drugs, the Mozambique Ministry of Health has made several changes in antimalarial drug policy. In 2002, chloroquine monotherapy was replaced with sulfadoxine/pyrimethamine/amodiaquine as the first line of treatment against uncomplicated malaria (33); 2 years later, this combination was replaced with artesunate/sulfadoxine/pyrimethamine (33). In 2008, artemether/lumefantrine was introduced to replace artesunate/ sulfadoxine/pyrimethamine (34). Molecular markers for antimalarial drug resistance have been considered useful for confirming parasite resistance, a major factor causing treatment failure. To determine whether parasites carrying these polymorphisms or gene amplifications exist in Mozambique, we conducted molecular surveillance targeting K13, pfmdr1, pfcrt, and pfdhps polymorphisms and pfpm2 and pfmdr1 copy numbers in field isolates collected from 4 sentinel sites.

Study Sites and Population
We performed a descriptive observational study on blood samples collected before artemether/lumefantrine treatment (on day 0) in 2015 from 352 symptomatic children at 4 sentinel sites in Mozambique ( Figure 1): 1) Hospital Rural de Montepuez in Cabo Delgado Province (northern region), 2) Centro de Saúde de Dondo in Sofala Province (central region), 3) Hospital Provincial de Moatize in Tete Province (central region), and 4) Hospital Rural de Chokwe in Gaza Province (southern region). In Mozambique, transmission usually peaks during the rainy season (November-April). Transmission intensity in southern Mozambique is generally low, although areas of high transmission may still occur (35). To determine molecular markers of drug resistance, we analyzed samples collected during a clinical trial conducted in 2015 (registration no. ACTRN12616001680459); the trial aimed to assess the efficacy and safety of artemether/ lumefantrine for treatment of uncomplicated P. falciparum malaria in children <5 years of age. The National Mozambican Ethical Review Committee (Mozambique) and Hospital Clínic (Barcelona, Spain) ethics review committees approved the study, and signed written informed consent was obtained from all participants' guardian or parent.

Molecular Procedures
We extracted DNA from half of a 50-µL dried blood drop on Whatman 3-mm filter paper by using a QIAamp DNA Mini kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions. We used an ABI PRISM 7500 HT Real-Time System (Applied Biosystems, Foster City, CA, USA) to amplify purified parasite DNA templates, following a previously described method (36,37). A standard curve was prepared from an in vitro culture of 3D7 strain containing known numbers of ring-infected erythrocytes. The standard curve was run in triplicate for each test with 5 serially diluted points. Parasitemia in the clinical samples was quantified by extrapolation against the standard curve.
To assess polymorphisms in the K13, pfcrt, pfmdr1, and pfdhps genes, we amplified purified DNA templates by using a 2720 Thermal Cycler (Applied Biosystems), following protocols described for K13 PCR (38) and pfcrt PCR (35). To genotype polymorphisms in pfmdr1 and pfdhps genes, we designed new assays by using Sanger sequencing and restriction fragment length polymorphisms (online Technical Appendix, https://wwwnc.cdc.gov/EID/ article/24/1/17-0864-Techapp1.pdf). A total of 6 positive controls with known K13 alleles, provided by the Institut Pasteur in Cambodia, and 4 parasite lines (3D7, 7G8, Dd2, and V1/S) with known pfcrt and pfmdr1 alleles, available in the laboratory, were also processed, amplified, and sequenced at the same time as the studied samples (PCR characteristics in online Technical Appendix Table). To determine the detection limit of Sanger sequencing, we used artificially mixed DNA samples of P. falciparum laboratory strains containing various known proportions of wild-and mutant-type alleles of pfcrt (K76T) and pfmdr1 (Y184F and S1034C) genes. To estimate polymorphism frequency, we considered isolates with mixed alleles to be mutated.
We assessed copy numbers of pfpm2 and pfmdr1 genes as described elsewhere (11) with minor changes (online Technical Appendix) by using quantitative PCR (qPCR). We performed amplification in 20-µL reaction mixtures for pfpm2, pfmdr1, and pfβ-tubulin genes, separately. We used the pfβ-tubulin gene as an endogenous control. All samples with estimated copy numbers >1.5 were defined as containing multiple copies and repeated for confirmation. The estimated copy numbers were the average of the copy number of each clone in the isolate.

Data Analyses
We calculated the proportion of the mutant alleles and isolates with multiple copies of pfpm2 and pfmdr1 genes on the basis of the number of samples with wild-and mutant-type alleles as well as isolates with single and multiple copies of the gene from P. falciparum isolates from each study site. To compare continuous data and categorical data between sites, respectively, we performed analyses of variance and χ 2 tests. We defined statistical significance as p<0.05.
The polymorphism analyses of K13, pfmdr1, pfcrt, and pfdhps genes were successful for 98.3% to 100% isolates. Because no amplifications were noticed in negative controls (with water and human genomic DNA), PCR assays were specific to P. falciparum genomic DNA only.

Detection Limit of Mixed Samples by Sanger Sequencing
We identified ¨A¨ alleles of pfcrt (K76T) and pfmdr1 (Y184F) codons in artificially mixed samples by using Sanger sequencing when the proportion of target DNA was >10%. However, we identified ¨C¨ and ¨T¨ alleles of pfcrt (K76T) and pfmdr1 (Y184F) polymorphisms, respectively, in mixed samples when their proportion was >20% (online Technical Appendix Figure 1). For pfmdr1 (S1034C) polymorphism, the minor allele was detected when its proportion was >20% in a mixed sample (online Technical Appendix Figure 1, panel C). For positive controls, we used several parasite lines with known K13, pfmdr1, and pfcrt alleles. As expected, sequencing analysis of all positive controls revealed wild-and mutant-type alleles of K13, pfmdr1, and pfcrt polymorphisms.

Copy Numbers for pfpm2 and pfmdr1
We successfully analyzed 351 (100%) samples for copy number variation in the pfpm2 and pfmdr1 genes. PCR efficiencies were 98.4% for pfpm2, 97.2% for pfmdr1, and 99.2% for pfβ-tubulin genes. As expected, the estimated pfpm2 and pfmdr1 copy numbers for the positive controls were 3-4 copies. The estimated mean (interquartile range) copy numbers were 3.51 (3.37-3.62) for pfpm2 and 3.62 (3.51-3.79) for pfmdr1 positive controls. When we used a copy number threshold of 1.5 to define multiple copies, only 4 (1.1%) and 5 (1.4%) of the 351 isolates had multiple copies of pfpm2 and pfmdr1, respectively (Table 2; Figure 2). The range of estimated pfpm2 copy numbers was 0.59-1.79 and of pfmdr1 was 0.58-1.88. The copy number of pfpm2 and pfmdr1 genes did not significantly differ between isolates from different sites. The proportion of isolates with multiple copies of the pfpm2 gene was the highest at Chokwe (2 [2.3%] of 87). Only 1 (1.1%) of 88 samples from Dondo had multiple copies of pfpm2 and pfmdr1 genes.
The proportion of isolates with A437G and K540E polymorphisms was the highest at Moatize, and the proportion with S436F, S436C, S436A, and S436H alleles was highest at Montepuez.

Discussion
We provide evidence for the presence of multiple copies of pfpm2 in 4 (1.1%) of 351 P. falciparum isolates circulating in southern Mozambique despite the absence of piperaquine drug pressure. Thus, with adequate drug pressure, isolates resistant to piperaquine may spread in Mozambique, as occurred in Southeast Asia (10,42,43).
In selected areas of Cambodia in 2008, piperaquine was introduced as a partner drug of artemisinin (44). Soon after its introduction, as early as 2010, piperaquine resistance in western Cambodia emerged at an alarming rate (45). Subsequent reports confirmed a rapid increase in failure of dihydroartemisinin/piperaquine in other parts of Cambodia (42,46,47). The most frequent K13 mutants associated with artemisinin resistance were absent in the isolates from Mozambique. We also determined that prevalence of pfcrt (K76T) and pfmdr1 (N86Y) markers of resistance are low, supporting previous evidence for the return of parasites carrying pfcrt wild-type alleles in Mozambique (35), in contrast to persistence of pfdhps (A437G [83.8%]) and K540E [82.2%]) polymorphisms, markers of sulfadoxine/pyrimethamine resistance (34). The well-characterized polymorphism in pfmdr1 (Y184F [46.7%]) was also prevalent in Mozambique. We found very low prevalence (<1%) for 4 new polymorphisms (Leu619Leu, Phe656Ile, Val666Val, and Gly-690Gly) in the K13 gene of P. falciparum isolates from Mozambique. All polymorphisms except Phe656Ile were synonymous. Previously, V494I K13 nonsynonymous polymorphism has also been reported in Mozambique (48). In Africa, K13 nonsynonymous polymorphisms have also been reported at low frequencies in isolates from Cameroon, Central African Republic, Democratic Republic of the Congo, Gabon, The Gambia, Kenya, Madagascar, Malawi, Mali, Rwanda, Togo, Uganda, Zambia, and Equatorial Guinea (38,40,49-50; references 51,52 in online Technical Appendix). The association of nonsynonymous polymorphisms with delayed parasite clearance has only recently been identified in Africa (reference 52 in online Technical Appendix).
Resistance to both chloroquine and amodiaquine has been mainly associated with a single K76T mutant allele in the pfcrt gene (22)(23)(24). In our study, its prevalence in 8 (2.3%) of 351 samples was significantly lower than that found in previous studies in Mozambique (  reference 53 in online Technical Appendix). Our pfcrt data align with previous evidence for the return of parasites carrying pfcrt wild-type alleles in Mozambique (35) and in other countries in Africa, such as Ethiopia (reference 54 in online Technical Appendix), Malawi (reference 55 in online Technical Appendix), and Cameroon (reference 56 in online Technical Appendix). The selective disadvantage of mutant parasites in the absence of drug pressure has been proposed as the leading factor contributing to the reemergence of chloroquine-susceptible parasites (reference 57 in online Technical Appendix). Because artemether/lumefantrine has been shown to select for the wild-type pfcrt 76K allele (25), this reemergence might be accelerated because of the increased use of artemether/lumefantrine as a firstline treatment for uncomplicated malaria in Mozambique (reference 53 in online Technical Appendix).
Our study also provides evidence for the presence of few P. falciparum isolates with multiple copies of the pfmdr1 gene (5 [1.4%] of 351) circulating in southern Mozambique (34). Increased pfmdr1 copies have been associated with resistance to mefloquine and partial resistance to lumefantrine (13)(14)(15)(16)(17)(18). Our study found that prevalence of the pfmdr1 N86Y mutant allele has decreased and the Y184F mutant allele has increased over time, in contrast with findings of other studies from Mozambique (34; references 55,58 in online Technical Appendix). We identified 10 new polymorphisms (L1030L, D1061D, D1127D, S1137S, L1174L, D1179D, N1189N, T1192A, F1194S, and Y1197N) that had not been previously described for the pfmdr1 gene. Among the 15 polymorphisms identified in the pfmdr1 gene, we observed significant differences between sites for the N86Y and D1179D polymorphisms only.
Of 351 children who had received adequate treatment with artemether/lumefantrine (6 doses), 2 were still positive for parasitemia on day 3 (39). These isolates contained wild-type K13 gene polymorphisms. P. falciparum-positive patients for whom artemether/lumefantrine treatment failed had parasites that carried wild-type pfcrt and pfmdr1 polymorphisms. This observation suggests that in vivo artemether/lumefantrine resistance may be caused not only by variations in the pfcrt and pfmdr1 genes but possibly by parasite selection of variations in other genes; however, drug bioavailability issues may also have contributed.
A high proportion of the P. falciparum isolates from Mozambique contained K540E (82.2%) and A437G (83.8%) mutant alleles. These mutant alleles may still not jeopardize the effectiveness of sulfadoxine/pyrimethamine for malaria prevention in Mozambique; recent findings suggest that only >90% prevalence of a pfdhps K540E polymorphism could reduce the effectiveness of intermittent preventive therapy to clear peripheral parasites and prevent new infections during pregnancy (reference 59 in online Technical Appendix). Therefore, sulfadoxine/pyrimethamine remains effective for intermittent preventive therapy during pregnancy, despite the high frequency of quintuple mutants; thus, the World Health Organization continues to recommend the use of intermittent preventive therapy to prevent malaria during pregnancy (references 60-62 in online Technical Appendix).  However, alternative antimalarial drugs for intermittent preventive therapy during pregnancy are needed because the prevalence of the K540E polymorphism in Mozambique is close to the threshold.
In conclusion, we report that prevalence of isolates with multiple copies of pfpm2 is lower than that found by previous studies in Cambodia (34.3%) and Vietnam (54.3%) (10,43), and we report the absence of K13 polymorphisms known to be associated with artemisinin resistance. We also report the return of parasites carrying pfcrt wild-type alleles (except in Chokwe) and persistence of parasites with pfdhps mutations associated with sulfadoxine/pyrimethamine resistance in Mozambique. Sulfadoxine/pyrimethamine-resistant isolates may be maintained by the constant use of intermittent preventive therapy during pregnancy, use of drug outside of hospitals, the very common use of co-trimoxazole (as prophylaxis for HIV-infected persons), and the low fitness cost of the polymorphisms (33; references 63,64 in online Technical Appendix). In contrast, the fitness cost of the pfcrt mutant allele seems to be high, probably accounting for the return of parasites carrying pfcrt wild-type alleles in Mozambique (reference 57 in online Technical Appendix). Current regional elimination efforts, as part of the G8 Malaria Elimination Initiative, may lead to more aggressive strategies involving population-wide distribution of antimalarial drugs, such as dihydroartemisinin/piperaquine, resulting in significantly increased drug pressure. Our findings might provide baseline prevalence data that enable us to directly determine the effects that increasing malaria control efforts or elimination programs will have on resistance evolution.