Artemisinin combination therapy (ACT) has been adopted globally as the first-line treatment for uncomplicated Plasmodium falciparum malaria and has contributed to the reduction in malaria related mortality and morbidity. However, resistance to artemisinin poses a threat to the global effort to control malaria. In 2008–2009 the first verified cases of artemisinin resistance, characterized by delayed parasite clearance, were observed in western Cambodia [1]. Prior to those cases, instances of reduced parasite susceptibility to artemisinin were reported in parts of Thailand bordering Cambodia (Chantaburi, Trat, Sakaew, Sisaket, Burirum, and Surin provinces) and Myanmar (Tak province) as early as 2003 [2]. The Thai-Cambodian border region has historically been the epicenter of multi-drug resistant (MDR) malaria [3]. As resistance to earlier anti-malarial drugs spread from this region to Africa and other parts of Asia through parasite migration, there is a serious concern that a similar scenario may occur with artemisinin resistance [4]. The ACT artesunate-mefloquine (ASMQ) has been used as first-line therapy in Thailand since 1995, beginning in areas where multi-drug resistance had evolved. Use of ASMQ was extended to the rest of the country after the World Health Organization (WHO) recommended ACT for global use in the early 2000s [5]. In Thailand, ASMQ was initially introduced as a two-day regimen and in 2008 was extended to three days (three days of artesunate and two days of mefloquine, but with the same total dose as the two-day regimen) [6].

The ASMQ regimen has remained generally effective in Thailand despite high levels of mefloquine resistance with a cure rate of greater than 90% [5]. However, in several locations in Cambodia and Thailand, treatment failure rates over 10% have also been observed [7]. To date, the strongest evidence of artemisinin resistance was initially reported in western Cambodia, and subsequently other parts of Southeast Asia [1, 7–13].

To prevent the spread of artemisinin-resistant P. falciparum, the WHO and other partners initiated an artemisinin resistance containment project for the Greater Mekong Subregion in 2009 [14]. The goal was to identify and prevent artemisinin-resistant parasites from spreading outside of documented hotspot regions along the Thai-Cambodian border by ensuring proper diagnosis and full treatment of reported malaria cases [15]. Subsequently, the WHO, along with other partners, initiated the Global Plan for Artemisinin Resistance Containment and Emergency Response to artemisinin resistance in the Greater Mekong Subregion [5].

Currently, therapeutic efficacy studies (TES) are considered the gold standard for determining antimalarial drug efficacy [16]. However, the WHO recommends that TES results be complimented using molecular marker studies [17]. Therefore, it was desirable to identify a molecular marker for artemisinin resistance. Initial studies using a genome-wide association approach found two loci on P. falciparum chromosomes 10 and 13 to be associated with artemisinin resistance [12, 18]. After a long search to pinpoint a specific gene associated with artemisinin resistance, the K13 gene (PF3D7_1343700) on chromosome 13 was identified as a potential molecular marker [19]. The study identified mutations in the propeller domain of the K13 gene that were associated with artemisinin resistance as measured by ex vivo ring stage survival assays and delayed parasite clearance times [19]. Specifically, the study identified 18 non-synonymous single nucleotide polymorphisms (SNPs) in the K13 propeller domain, of which three mutations (C580Y, R539T and Y493H) were strongly associated with increased ring stage survival and delayed parasite clearance rates. The C580Y allele accounted for about 85% of all mutant K13 alleles observed in 2011–2012 in western Cambodia [19]. Most recently, a two year multi-site project by Ashley et al. further confirmed that multiple SNPs in the propeller domain of K13 were predictive of slow parasite clearance and that these mutations were found in multiple countries in the Greater Mekong Subregion [20]. That study, along with two other recent studies by Takala-Harrison et al. and Miotto et al. [13, 21], showed that the C580Y allele was the predominant allele in Cambodia, Myanmar, and Vietnam. The authors further demonstrated that the C580Y allele may have emerged independently in Cambodia and Myanmar.

Although Thailand has historically been an epicenter of resistance to several antimalarial drugs, currently there are limited data on the artemisinin resistance-associated K13 propeller mutations in this region. Here, using historical samples collected during 2007 from ten different sites in Thailand, we set forth to answer the following questions: (1) Were artemisinin resistance-associated K13 mutant alleles present in Thailand prior to the implementation of the artemisinin resistance containment projects? (2) If so, what were the prevalence and distribution of the K13 propeller mutations in Thailand? (3) What are the evolutionary histories of the different K13 mutant and wild type alleles? (4) Are the K13 mutant alleles evolving locally or are particular mutants spreading across Thailand? and (5) Is there evidence for selection of resistant K13 alleles?

To our knowledge, this study is one of the first reports to systematically analyze the artemisinin resistance K13 propeller mutations and flanking microsatellite loci in parasites collected in numerous sites in Thailand shortly before the implementation of the artemisinin resistance containment project in 2009. The samples were collected from across ten provinces, including the containment zones and areas at highest risk for malaria. This study provides evidence that (1) artemisinin resistance alleles were present in 8 out of 10 Thai provinces sampled, including two mutant alleles (R575K and S621F) not previously reported in Thailand, (2) artemisinin resistance-associated K13 alleles had evolved along the Thai-Cambodia and Thai-Myanmar border regions at least two years prior to the implementation of the artemisinin resistance containment project, (3) the artemisinin resistance-associated C580Y mutant alleles were the most common and widespread in Thailand, (4) there are clear differences in microsatellites that differentiate the C580Y mutant alleles from eastern and western parts of Thailand, and (5) the C580Y alleles appear to have had two recent, independent origins.

Our study provides insight into the prevalence and distribution of K13 mutations in Thailand as early as 2007. Interestingly, the prevalence of mutant K13 alleles reported in Pailin, Cambodia during 2007 for the C580Y (45%) and R539T (5%) alleles [19] is consistent with our findings. Across the border from Pailin, in the provinces of Chanthaburi, Sisaket and Trat, we observed that more than 42% of the parasites screened carried the C580Y mutant allele (Fig 1). The C580Y allele has since been associated with delayed parasite clearance in this region [19, 20]. The highest prevalence of the C580Y (62%) and R539T (16%) mutations is seen in the province of Sisaket, just north of Cambodia. These results are in agreement with the recent study by Ashley et al [20], which provided compelling evidence that the C580Y and R539T mutations are associated with delayed parasite clearance in both Sisaket and Ranong. Moreover, 78% (7/9) of the parasite isolates from Sisaket showed a similar clonal genetic profile, suggesting that this may have been a recent clonal expansion event (Figs 3C and S1). Similar findings were reported by Miotto et al [22], who demonstrated that three subpopulations associated with clinical resistance to artemisinin may have recently expanded in Cambodia and elsewhere in the region.

In eastern Thailand, C580Y and R539T were the only mutations observed; however, this may be due to the limited number of samples analyzed. In contrast, the provinces bordering Myanmar had the following mutations: S621F, C580Y, R575K, P574L, E556D, and N458Y. C580Y, P574L and R575K were the most commonly found alleles along the Thai-Myanmar border region (i.e. from Kanchanaburi to Chumporn). Interestingly, the C580Y and R575K mutations have recently been reported near the Thai-Myanmar border as well [23, 24]. Other mutant alleles (N458Y, S621F, and E556D) were rare and restricted to one or two sites (Fig 1), suggesting that these K13 mutations may have arisen independently. Although previous studies have confirmed a strong association between select K13 propeller domain mutations and delayed parasite clearance [19, 20, 25], it remains to be determined whether the remaining mutant alleles will have a similar association. Furthermore, it remains to be seen whether the C580Y mutation will trend towards fixation in Thailand, as was seen between 2001 and 2012 in Pailin, Cambodia [19]. Our data show that artemisinin-resistant K13 alleles did not spread to or evolve in the southernmost Yala province or the northern Mae Hong Son province during 2007. In Yala, the parasites had identical flanking and neutral microsatellite haplotypes, which is consistent with our previously published results [26], indicating a closely related clonal population.

Population differentiation analysis further reveals that parasites with the C580Y allele group together by geography (Fig 3A). Analysis of raw microsatellite haplotype data for these parasites revealed that alleles circulating in the east and west comprise two distinct lineages marked by differences in the 8.6kb locus downstream of the K13 gene (Fig 2). These data suggest the C580Y mutations may have arisen independently along the Thai-Cambodia and Thai-Myanmar borders. The reduced pattern of heterozygosity of C580Y alleles compared to wild type alleles (Fig 3) further suggests recent independent origins along the Thai-Cambodia and Thai-Myanmar borders. This interpretation of our data is consistent with the recently published work by Miotto et al. [21]. The study provided compelling evidence for the selection of the C580Y allele in the Greater Mekong Subregion [21], which is consistent with our data, and suggested that the selection process may have been under way on both sides of Thailand at the time of this study. One possible interpretation of our findings would be that the parasites migrated across Thailand prior to the independent C580Y emergence events. Recent findings by Takala-Harrison et al. and Miotto et al. are consistent with the independent emergence of the C580Y allele, which was also observed along the Myanmar-Thai border and the lower Mekong region [13, 21]. The strong association between the genetic dissimilarity and geographic distance of the C580Y mutants further supports the hypothesis that this mutation may have emerged independently in eastern and western Thailand (Fig 5).

Given the history of population movements within this region, some of the mutant alleles in the Thai-Myanmar region (C580Y and P574L) and Thai-Cambodia region (C580Y and R539T) may share common ancestry. Interestingly, in the work by Miotto et al. [21], parasites with the most common K13 mutant alleles (C580Y, I543T, R539T, and Y493H) were found in multiple countries in the region, indicating that parasite cross-border movement may have already occurred. It remains to be determined if the P574L and R575K alleles, which have been found in Myanmar [13, 23], originated in either Thailand or Myanmar.

It has been suggested that in the absence of drug pressure, parasites with some resistant mutations are less fit than their ancestral wild type counterparts [27]. However, with continued drug pressure one might expect resistant alleles, such as the C580Y mutation, to eventually become fixed in the population as has occurred with other resistance mutations in the past. The work by Ariey et al., which identified the K13 propeller as a molecular marker of artemisinin resistance, demonstrated that over the course of 11 years (2001–2012), the C580Y allele prevalence increased from 40% to 90% in Pailin, Cambodia [19]. This would suggest that the parasites carrying the C580Y mutation may be nearing fixation in the population, and therefore, no sensitive parasites will remain to outcompete them in the absence of ACT. This is very worrisome, as ACT is one of our last working treatment options for drug resistant P. falciparum.

In summary, it is evident from our study that artemisinin-resistant K13 alleles have been evolving along both the Thai-Cambodian border and Thai-Myanmar border long before the artemisinin containment project was implemented. It is further evident that the most commonly found C580Y allele had two distinct haplotypes, suggesting different patterns of origin and migration along the Thai-Cambodia and Thai-Myanmar regions.