Molecular surveillance for drug resistance markers in Plasmodium vivax isolates from symptomatic and asymptomatic infections at the China–Myanmar border

Plasmodium vivax is the most geographically widespread Plasmodium species and also a cause of severe malaria [1‐3]. Countries within the Greater Mekong Sub-region (GMS) have endorsed an ambitious plan to eliminate malaria by 2030 [4]. However, the proportion of malaria cases caused by P. vivax infection in Myanmar has increased steadily since 2012, especially in border areas [5]. Several features of P. vivax, including the formation of hypnozoites, the low density of infection, and the early production of gametocytes favor continuous transmission. Plasmodium vivax infections from asymptomatic carriers as a potential silent reservoir for transmission are common in both high- and low-endemic areas of Myanmar [6, 7]. Previous reports of asymptomatic Plasmodium falciparum infections carrying genes potentially associated with drug resistance suggest a possible spread of drug-resistant parasites in Myanmar [8, 9]. However, surveys of P. vivax drug resistance are scant because most drug resistance studies have focused on P. falciparum. Thus, monitoring the emergence and spread of P. vivax drug resistance, especially among asymptomatic carriers, is critical to achieve the goal of malaria elimination in the GMS.

Chloroquine (CQ) and primaquine (PQ) combination has been the frontline therapy for treating uncomplicated P. vivax cases. Plasmodium vivax resistance to CQ was first reported by Papua New Guinea in 1989 [10]. In the GMS, there have been sporadic reports of efficacy studies suggestive of emergence of CQ resistance (CQR) [11‐13]. Two recent studies at the China–Myanmar border have demonstrated the declining efficacy of CQ against P. vivax and the potential emergence of drug resistance in this parasite [14, 15]. Although sulfadoxine-pyrimethamine (SP) was rarely used to treat P. vivax, the substantial selective pressure exerted by the drug is thought to have continued during treatment of P. vivax and P. falciparum mixed-strain infections, resulting in the emergence of high-grade antifolate resistance in P. vivax populations [16]. Molecular surveillance studies indicated that P. vivax populations in southwestern Yunnan Province of China bordering Myanmar may be highly resistant to SP [17]. Because artemisinin-based combination therapy (ACT) is also used to treat mixed-species infections [18, 19], P. vivax may have been under similar drug selective pressure as P. falciparum.

Currently, the molecular mechanisms underlying CQR remain unknown. It has been proposed that P. vivax CQR may involve similar molecular mechanisms as in P. falciparum. Multidrug resistance 1 gene (pvmdr1) and putative transporter protein CG10 gene (pvcg10 or pvcrt-o), orthologous to pfmdr1 and pfcrt genes, respectively, have been suggested as possible genetic markers for CQR [20, 21]. However, the first survey of the pvcrt-o gene in clinical isolates including treatment failure cases failed to identify an association between in vivo CQR with amino acid changes of pvcrt-o, suggesting the mechanism of CQR in P. vivax may be different from that in P. falciparum [22]. The K10 insertion in the first exon of pvcrt-o was the most common but also variable in different parasite populations [20, 23, 24], though it does not appear to correlate with CQR. Analysis of pvcrt-o mutant isoforms in yeast suggests that at least some pvcrt mutations may alter P. vivax sensitivity to CQ [25]. Whereas increased expression or copy number of pvcrt-o was correlated with in vivo CQR in South America [26, 27], such a correlation was not identified in Papua Indonesia, where the level of CQR is high [28]. Recently, using a genetic cross and linkage mapping, upregulated pvcrt expression was identified as a mechanism of CQR [29]. In P. falciparum, polymorphisms in codons 86, 184, 1034, 1042 and 1246 of the pfmdr1gene were reported to be associated with CQR, which correspond to respective positions 91, 189, 1071, 1079 and 1291 in pvmdr1 [21]. In pvmdr1, in vitro studies identified the Y976F mutation as a possible marker for CQR in P. vivax [20, 30], whereas other studies did not identify such an association [31‐34]. Similarly, whereas pfmdr1 gene amplification was associated with resistance to mefloquine (MQ) in Thailand [30], increased expression of pvmdr1 and pvcrt-o was associated with CQR in Brazil [26]. Altogether, the roles of pvcrt-o and pvmdr1 in CQR in P. vivax are still not resolved [35]. Mutations in dihydrofolate reductase (pvdhfr) and dihydropteroate synthase (pvdhps) have been associated with the altered clinical response to SP. F57L, S58R, T61M and S117N in pvdhfr are linked to pyrimethamine resistance [19, 36‐39], while S382A/C, A383G, and A553G in pvdhps are responsible for sulfadoxine resistance [40]. Mutations in the propeller region of P. falciparum kelch 13 (pfk13) gene are the main genetic marker for artemisinin resistance [41]. It is logical to determine whether artemisinin drugs have imposed similar selective pressure on the pfk13 ortholog on chromosome 12 of P. vivax (pvk12) [42‐44].

Drug resistance affects the fitness and virulence of the malaria parasites [45]. This has been demonstrated in P. falciparum using in vitro growth competition [46, 47] and inferred from the reversion of resistance-mediating mutations to wild type (WT) in parasite populations after withdrawal of the drug [48]. Since less fit parasites are presumably to produce infections with lower parasitaemia, drug resistance may also affect the clinical presentations of the disease. Some mutations in pfcrt and pfmdr1 were found to have higher prevalence in children with asymptomatic parasitaemia than those with parasitaemia and fever [49]. Similarly, in the GMS, pfmdr1 amplification was more prevalent in subclinical isolates than clinical isolates [50]. Under the same premise, mutations mediating CQR in P. vivax may have differential prevalence in asymptomatic and symptomatic infections.

To test this hypothesis and to obtain more comprehensive information of polymorphisms in candidate drug resistance genes in P. vivax in the China–Myanmar border area, P. vivax parasites from asymptomatic and acute infections were genotyped at the pvmdr1, pvcrt-o, pvdhfr, pvdhps and pvk12 genes.

In Myanmar, CQR in P. vivax was reported as early as in the 1990s [12, 13]. Drug resistance in P. falciparum and P. vivax isolates of asymptomatic malaria carriers has also been reported in high- and low-endemic regions of Myanmar [8, 9]. Recently, along the China–Myanmar border, the therapeutic responses of P. vivax malaria to CQ treatment were declining [14, 15]. Thus, this study compared the potential markers for CQR and antifolate resistance in asymptomatic and symptomatic P. vivax infections from this region.

The molecular mechanisms underlying CQR are not well understood, but mutations in pvmdr1 and expression of pvcrt-o were implicated. For pvmdr1, the Y976F mutation has been reported in P. vivax isolates from many malaria-endemic regions around the world [60‐64], and is associated with a decrease in in vitro sensitivity to CQ [20]. In this study, the Y976F mutation was relatively rare, with 5.5% detected only in samples from symptomatic infections. This prevalence was much lower than that found in Cambodia (89%) [65] and Thailand (8–25%) [62]. The T985M mutation is fixed in all parasite populations in Asia and it is not associated with CQR. Similarly, F1076L has not been found to be associated with CQR both in vivo and in vitro drug assays [66]. In this study, F1076L reached high prevalence of 85.8%, which was concordant with previous reports from other Asian areas including India, Thailand and Myanmar [23, 60, 61]. The frequency of F1076L mutation in asymptomatic infections in Laiza township was twice as much as that in Shwegyin township of Myanmar [8]. The single mutation haplotype 976Y/997K/1076L was the most prevalent at the China–Myanmar border, similar to a previous report from India [60], but differed from a report from Yunnan, China, which showed WT as the dominant [67]. These geographical variations in pvmdr1 gene may suggest different drug selection pressure imposed on P. vivax population in these Asian countries.

The role of pvcrt-o in CQR is controversial. Analysis of pvcrt-o isoforms in yeast suggest that a single amino acid substitution (S249P) slightly increased CQ transport [25], indicating a mild form of CQR. Other studies found that lysine (K) insertion at position 10 of pvcrt-o gene may be associated with CQR [20, 23]. This study observed a prevalence of 34% of K10 insertion in the pvcrt-o gene, higher than that found in India (9.4%) and Thailand (0%) [60‐62], but lower than that detected in other regions of Myanmar (48.3–72.7%) [23]. A recent study showed correlation of CQR with increased expression of pvcrt [29], which could not be evaluated with the DBS samples. Continuous monitoring of clinical efficacy of CQ and candidate molecular markers including pvcrt-o expression may be necessary to assess CQR in P. vivax populations in different parts of the GMS.

Results from this study suggest high-level resistance of the P. vivax parasites from the GMS to the antifolate drug SP. Resistance to antifolate drugs in P. falciparum and P. vivax was found to be associated with point mutations in dhps and dhfr [68]. For the pvdhfr gene, mutations at codons 50, 58, 117 and 173, corresponding to residues 51, 59, 108 and 164 in pfdhfr, confer resistance to pyrimethamine [69]. Double mutations (S58R and S117N) were associated with a high level of resistance in P. vivax, whereas quadruple mutations (F57L/I, S58R, T61M and S117T) were more likely associated with SP treatment failure [38, 70]. Here, the prevalence of double or quadruple mutations (50.7%) was much lower than that found along the Thailand border (100%) and other areas of Myanmar (71%–90%) [23, 61], but much higher than what was found in southern China (9.2%) [71]. Asymptomatic isolates in this study showed a much lower prevalence of both the double and quadruple mutations than that found in southern Myanmar. Compared with the findings reported much earlier in Myanmar and Cambodia where double mutations (S58R and S117N) accounted for 91.7% to 93.8% of the sequenced samples [36, 72], multiple mutations (more than 2) in the DHFR domain were more frequently in the present study. This may be a warning sign of the growing resistance of P. vivax to pyrimethamine over time in Southeast Asia.

Mutations at codons 382, 383, 512, 553 and 585 in pvdhps, corresponding to codons 436, 437, 540, 581 and 613 in pfdhps, may confer resistance to sulfadoxine. A recent study confirmed that A383G was associated with sulfadoxine resistance than other mutations when examined in transgenic rodent parasites expressing PvHPPK-DHPS [73]. In addition, the double mutations A383G and A553G that possibly cause a disruption in the sulfadoxine-binding site in P. vivax were similar to those in P. falciparum [40]. At the China–Myanmar border, A383G reached 75%. The prevalence of a haplotype with both the A383G and A553G mutations was 36.5%, obviously lower than at the Thai–Myanmar and–Cambodian borders (61.2%), and in other endemic areas of Myanmar (73.5%) [23, 61].

For P. falciparum, triple mutations at codons 51, 59 and 108 of pfdhfr and double mutations at codons 437 and 540 of pfdhps are associated with SP treatment failures [74]. The combination of pvdhfr mutations at codons 57, 58, 61 and 117 and pvdhps mutations at codons 383 and 553 was identified in 13 (22%) P. vivax isolates, of which 6 (46.2%) were from asymptomatic carriers. These findings suggest that highly resistant P. vivax parasites to SP were present among asymptomatic and symptomatic infections at the China–Myanmar border.

Tandem repeats are a unique feature present in the pvdhfr and pvdhps, but it is not clear whether polymorphisms in these repeat regions contribute to resistance to SP. Tandem repeat region variations were observed in both asymptomatic and symptomatic infections. Consistent with previous studies, parasite isolates based on the pvdhfr repeat region were typically separated into three types [67, 75]. Type 1 tandem repeat variant was highly prevalent along with triple, quadruple or quintuple mutations, and about half of Type 3 variant co-existed with the double mutations (58R/117N). This finding is consistent with that Type 1 and Type 3 are associated with increased resistance to SP [75‐78]. It differed from the earlier findings in Cambodia where a large majority of isolates had two GGDN repeat units with double mutations (58R/117N) [72], indicating that P. vivax with antifolate resistance evolved independently in different regions of the GMS. For pvdhps gene, five types of tandem repeat variants were identified for the first time in this study. Similar to pvdhfr, the majority of pvdhps tandem repeat types co-existed with mutations conferring SP resistance. However, further studies are essential to clarify the relationship between these polymorphisms and P. vivax sensitivity to SP.

PCA was used to explore if the parasite populations in symptomatic and asymptomatic infections could be differentiated based on the haplotypes of mutations in the five candidate resistance genes. While this method may have limitations to illustrate the genetic relatedness of different parasite isolates, the analysis nonetheless showed that the two clusters largely overlapped. While this supports the notion that asymptomatic infections are important reservoirs for sustaining continued transmission of the parasites, it also showed higher prevalence of certain haplotypes in asymptomatic parasite population. For single-gene haplotypes, the WT pvmdr1 haplotype was significantly more prevalent in symptomatic patients than asymptomatic carriers, whereas the 976Y/997K/F1076L haplotype showed the opposite. In P. falciparum, some mutations in the drug transporter genes were found to confer fitness costs [46, 47]. Although the effect of pvmdr1 mutations on parasite’s fitness is unknown, such differences in the prevalence of the WT and mutant alleles in asymptomatic and symptomatic infections, which had lower and higher parasitaemias, respectively, implies that the F1076L mutation may be associated with a fitness loss in the parasites. This mutation varies greatly in different parasite populations [21, 31, 79, 80], and its functional importance for CQR remains to be formally tested. It is also noteworthy that although this study screened more than 13,000 blood samples for asymptomatic infections, only a limited number of slide-positive samples were identified and used in the analysis, thus limiting the sample size and power of the analysis. There were also differences in age distribution between the two groups, which further complicates the comparison as host immunity is correlated with age and exposure. Therefore, although this study provided baseline information on candidate drug resistance genes in P. vivax in the China–Myanmar border region, the resistance mechanisms, except for antifolate resistance, demand future investigations.

All countries in the GMS have set an ultimate goal of eliminating malaria by 2030. One of the main challenges is the resilience of P. vivax parasites to control measures, evidenced by the increased proportions of P. vivax parasites in many areas of the GMS. Molecular analysis of five potential drug resistance markers in P. vivax from the China–Myanmar border area showed prevalence of mutations in pvmdr1, pvdhfr and pvdhps, suggesting resistance to antifolate drugs and possible CQ. The higher prevalence of certain mutant alleles in asymptomatic infections also suggests fitness cost of the mutations, underscoring their potential involvement in drug resistance.