Plasmodium Falciparum Phenotypic and Genotypic Resistance Prole During the Emergence of Piperaquine Resistance From Royal Thai Army and Civilian patients in Northeastern Thailand

Malaria remains a public health problem in Thailand, especially along its borders where highly mobile populations can contribute to persistent transmission. This study aimed to determine resistant genotypes and phenotypes of 112 Plasmodium falciparum isolates from patients along the Thai-Cambodia border during 2013-2015. The majority of parasites harbored pfmdr1-Y184F mutation. A single pfmdr1 copy number, had CVIET haplotype of amino acids 72-76 of pfcrt and no pfcytb mutations. All isolates had a pfk13 point mutation (R539T, R539I, and C580Y), and increased ring-stage survival assay (except R539I). Multiple copies of pfpm2 and pfcrt-F145I were rst detected in 2014 (12.8%) and increased to 30.4% in 2015. Parasites containing either multiple pfpm2 copies with and without pfcrt-F145I or a single pfpm2 copy with pfcrt-F145I exhibited the elevated IC 90 values of piperaquine. Antimalarials were detected in 21 samples (18.8%) by mass spectrometer, and 12 samples (10.7%) by bioassay. Collectively, the emergence of these resistance patterns mirrored the reports of dihydroartemisinin-piperaquine treatment failures in the adjacent province of Cambodia, Oddar Meanchey, suggesting a migration of parasites across the border. As malaria elimination efforts ramp up in Southeast Asia, host nations militaries and other groups in border regions must be a focus for interventions prior to treatment. Two microscopists examined Giemsa-stained peripheral blood smears for each volunteer to determine malaria species infection and parasite densities for blood stages. Venous blood samples were collected in EDTA tubes for DNA extraction and molecular characterization and in sodium-heparin tubes for ex vivo bioassay and in vitro drug sensitivity assay. Plasma was separated from blood, frozen at -80 °C, and infected packed red blood cells were cryopreserved. All blood and processed blood samples were stored at -80C and transported in dry ice to AFRIMS for molecular characterization, ex vivo bioassay, and in vitro culture adaptation and drug sensitivity testing. selected culture-adapted clinical isolates following published methods with slight modications 14,47 . Parasite cultures were synchronized using 5% D-sorbitol and 75% Percoll to obtain 0 to 3-h post-invasion rings which were adjusted to 0.5-1% starting parasitemia with a 2% hematocrit in culture media (0.5% Albumax RPMI 1640 with 2.5% AB serum), and cultured in a 48-well microplate with 700 nM DHA and 0.1% DMSO in separate wells. The culture plate was then incubated for 6 hours at 37 °C in modular incubator chambers and gassed with 5% CO 2 , 5% O 2 and 90% N 2 . Cells were then washed once, resuspended in a drug-free medium, and cultured further for 66 hours. Susceptibility to DHA was assessed microscopically on thin lms by estimating the percentage of viable parasites, relative to control (% survival rate). For the controls, the RSA 0-3h was also performed on P. falciparum reference clones W2, IPC-4884 and IPC-5202 (BEI Resources, NIAID, NIH, USA). A survival rate > 1% was deemed resistant for RSA. subgroups. It can be clearly seen that parasite isolates containing multiple pfpm2 copies alone (group III), the pfcrt-F145I alone (group V), or the combination of both markers exhibited noticeably high IC 90 values, indicative of PPQ resistance phenotype. The current study showed that the pfk13-C580Y mutation was predominant, rising from 63% in 2013 100% in 2015. These ndings are in good agreement with those reported by others 42,61 in that several pfk13 mutations developed in Sisaket (62% 580Y) 42 but by the C580Y mutation became the sole predominant mutation found 61 . This pattern was not followed by other regions of Thailand. In southern Thailand (Yala Province) very few isolates with pfk13 mutations have been detected 61,65 . The pfk13-R539T has been previously identied in Thailand, Cambodia, Laos, and Vietnam 66-69 ; however, the pfk13-R539I found in this study has not yet been reported in Thailand. The pfk13-R539I mutation was previously observed at 0.9% in Ghanaian isolates 70 . As assessed by the RSA 0-3h , the R539I mutation R539I mutation provided the lower stabilization score, implying an effect on the protein structure and function 70 . All the collected isolates had a wild-type cytochrome b gene (pfcytb) at both codons 258 and 268 (those associated with atovaquone/proguanil treatment failure 36,38,71 ) which is consistent with other isolates collected along Thai-Myanmar and Thai-Cambodia borders during 1998-2005 72 , and while in the current study, decreased susceptibility to QN, CQ, and MQ was observed, causality cannot be established. Different results were observed from the P. falciparum isolates from the southern part of Thailand, with the pfmdr1 86Y allele signicantly more common 76


Introduction
An estimate of 229 million malaria cases occurred worldwide in 2019 1 and artemisinin-based combination therapies (ACT) have been the rst-line drug treatment for uncomplicated Plasmodiumfalciparum infection for most malaria endemic areas; however, resistance to both artemisinin (ART) and its partner drugs has increased over the years at a pace requiring intensive surveillance and monitoring.
The rst case reports of ART resistance emerged in Thailand in the late 1990s 2  is to achieve malaria elimination. Much of the research conducted at AFRIMS has mil-to-mil engagement with Thai and Cambodian armed forces, which allows for clinical and surveillance studies to be conducted in these high risk populations, resulting in interventions tailored to the military and shared with the MoPH Following the case reports of ACT failure, the Artemisinin Resistance: Con rmation, Characterization and Planning for Containment Project (ARC) in 2006 showed a delayed clearance resistance phenotype in Cambodia along the Thai border which continued to spread across mainland Southeast Asia [4][5][6][7][8][9] . But it was not until 2014 that an ART resistance molecular marker was identi ed with mutations in the propeller domain of P. falciparum Kelch-13 gene (pfk13) (PF3D7_1343700) [10][11][12][13] , which can be con rmed by the ring-stage survival assay (RSA) 10,14 and/or delayed parasite clearance times in clinical trials 4,5 . Multiple copies of P. falciparum multi drug resistance 1 (pfmdr1) (PF3DF_0523000) is a well-established marker for MQ resistance 15,16 , while the pfmdr1 single-nucleotide polymorphisms (SNPs) have been associated with modulation of parasite tolerance or susceptibility to several antimalarial drugs including quinine (QN), amodiaquine (AQ), chloroquine (CQ), MQ and lumefantrine (LUM) 17 . The main pfmdr1 SNPs associated with drug resistance include N86Y, Y184F, S1034C, and N1024D 18-23 . The wild-type N86 in pfmdr1 has been associated with increased tolerance to the artemether (ATM) and LUM 18, 23,24 , while parasites harboring the N86Y mutation exhibit an increased susceptibility to MQ 25 . An ACT alternative to AS-MQ is dihydroartemisinin-piperaquine (DHA-PPQ) and the rst PPQ resistance markers identi ed were multiple copies of P. falciparum plasmepsin 2 (pfpm2) (PF3D7_1408000) 26-29 , followed by mutations in P. falciparum chloroquine resistance transporter (pfcrt) (PF3D7_0709000) downstream of the 4-aminoquinoline resistance locus (positions 72-76 with K76T) [30][31][32][33] , and the E415G mutation in P. falciparum exonuclease (pfexo) (PF3D7_1362500) 26, 27,34 . PPQ resistance can be characterized by the elevated IC 90 values 35 and PPQ survival assay (PSA) with a survival rate of more than 10%, de ning a PPQ resistant phenotype 30 . Atovaquoneproguanil (ATQ-PG) is not an ACT but ATQ resistance subsequently emerged which was linked to speci c mutations in P. falciparum mitochondrial cytochrome B (pfcytb) (AY282930.1) in particular the mutation at positions 258 and 268 [36][37][38] .
Twenty years after the rst ACT failures, there are now descriptions of the spread of a single multidrug resistant malaria parasite lineage (PfPailin) across the eastern GMS 39,40 . Efforts in mapping parasite population structure and gene ow can assist in understanding and predicting the spreading pattern of resistance. Military populations at border areas who are deployed to endemic areas should be integrated in surveillance and monitoring efforts as they are on the front-lines of malaria transmission. Here we report resistance characteristics of P. falciparum clinical isolates collected from Thai soldiers and civilian patients between 2013 and 2015 and show the trends in drug resistance and spread. At that time, the national treatment guidelines in Thailand recommended AS-MQ for P. falciparum malaria. During this same period, we reported dramatic loss of e cacy of DHA-PPQ (54% treatment failure) in a study that was conducted by AFRIMS in Anlong Veng, 12 km from the Thai border of Sisaket Province 41 . As the neighboring countries strive to achieve malaria elimination, there is ongoing concern about the risk of malaria transmission across borders, where treatment guidelines are different.
While there is a pressing need for new classes of antimalarials, intensive malaria surveillance to track drug resistance is required in high risk military populations to achieve malaria control so that these border foci of infections no longer pose a threat to elimination.
Here, the investigation into in vitro drug susceptibility and molecular drug resistance marker pro ling of P. falciparum isolates in patients presenting to Thai military health facilities is reported.

Materials And Methods
Study setting, protocol and subjects This minimal risk surveillance study was open between July 2013-September 2015, enrolling Thai adults (aged 18 years and over) presenting with uncomplicated P. falciparum or P. falciparum/P. vivax mixed-infections at military health facilities. The enrollment population included soldiers, border police, or their family members and other villagers located near the Royal Thai Army (RTA) health clinics in Sisaket and Surin Provinces, located in northeastern Thailand on the Thai-Cambodia border. Inclusion criteria were asexual parasitemia per rapid diagnostic test (RDT) or blood lm, and no malaria infection or antimalarials taken within the past seven days. The protocol was approved by the Walter Reed Army Institute of Research Institutional Board, Institute for Development of Human Research Protection, and RTA Institutional Review Board. All research was performed in accordance with relevant guidelines and regulations. All study subjects provided written informed consent prior to participation. Goal enrollment was 50 malaria cases per year, balancing the number of isolates needed to characterize resistant genotypes/phenotypes with the local population size and expected incidence of cases that could be enrolled.

Sample collection and preparation
Patients diagnosed with P. falciparum infection at RTA clinics were subjected to peripheral venipuncture prior to treatment. Two microscopists examined Giemsa-stained peripheral blood smears for each volunteer to determine malaria species infection and parasite densities for blood stages. Venous blood samples were collected in EDTA tubes for DNA extraction and molecular characterization and in sodium-heparin tubes for ex vivo bioassay and in vitro drug sensitivity assay. Plasma was separated from blood, frozen at -80 °C, and infected packed red blood cells were cryopreserved. All blood and processed blood samples were stored at -80C and transported in dry ice to AFRIMS for molecular characterization, ex vivo bioassay, and in vitro culture adaptation and drug sensitivity testing.

Copy number variation assay
To determine copy numbers of pfmdr1 and pfpm2 genes, real-time quantitative PCR (qPCR) was performed on genomic DNA as previously described 15,28,43 with some modi cations. For pfmdr1, ampli cation reactions were performed according to the TaqMan Real-time PCR methods using ABI 7500 Real-time PCR system (Applied Biosystems) with 200 nM of each forward and reverse primer (Table S2) and 2 ng of DNA template while Rotor-Gene Q (QIAGEN, Valencia, CA) using Type-it ® HRM™ kit was employed for pfpm 28 . The primers and probes (Table S2) used were as previously described to amplify the following loci: pfmdr1 (PF3D7_0523000) and pfpm2 (PF3D7_1408000), respectively 28 . For the housekeeping gene, β-tubulin (PF3D7_1008700), â-tubulin forward and reverse primers were designed and used as a reference control for all experiments with the same validated PCR conditions as target primers. P. falciparum 3D7 and Dd2 were used as references for single and multiple copy numbers of pfmdr1, respectively. All samples including the references clones were performed in duplicate. The average copy number values for each genes were calculated using 2 −ΔΔCt method. Parasites with copy number greater than 1.5 copies for pfmdr1 15 and 1.6 copies for pfpm2 28 were interpreted to contain multiple copies of each gene.

In vitro Culture adaptation and maintenance of parasites
Of 112 collected samples, 86 samples were subjected for in vitro culture adaptation. The exclusion criteria of the in vitro culture adaptation were P. falciparum/P. vivax mixed infections, % parasitemia < 0.05, and ex vivo bioassay activity > 250 nM (DHA equivalent). Culture adaptation of the parasites was performed using the modi cation method of Trager and Jensen 44 . Parasites were maintained in fresh human erythrocytes (O + ) in RPMI-1640 medium (Sigma), containing 15% AB + human serum (heat inactivated and pooled) and supplemented with 25 mM HEPES, 25 mM sodium bicarbonate, and 0.1 mg/mL gentamicin. Human blood products (erythrocytes and serum) were obtained from the Thai Red Cross. Culture asks were gassed with 5% CO 2 , 5% O 2 , 90% N 2 and incubated at 37 °C.
In vitro 72 hours drug susceptibility by Histidine rich protein 2 (HRP2) Drug susceptibility test using HRP2-ELISA to measure 50% or 90% inhibitory concentration (IC 50 and IC 90 ) was performed following previously published methods [45][46][47] . Dried drug-coated plates containing antimalarial drugs as described in Chaorattanakawee et al, 35,46 were used and in vitro drug susceptibility testing was carried out for control reference clones (W2, D6, C2B) (Malaria Research & Reference Reagent Resource, Manassas, Vermont, USA) as described previously 35 . IC 50 s and IC 90 s were estimated by nonlinear regression analysis using GraphPad Prism version 6.0.

Ring-stage survival assay (RSA)
In vitro RSA 0-3 h was performed on 0-3 h post-invasion rings obtained from selected culture-adapted clinical isolates following published methods with slight modi cations 14,47 . Parasite cultures were synchronized using 5% D-sorbitol and 75% Percoll to obtain 0 to 3-h post-invasion rings which were adjusted to 0.5-1% starting parasitemia with a 2% hematocrit in culture media (0.5% Albumax RPMI 1640 with 2.5% AB serum), and cultured in a 48-well microplate with 700 nM DHA and 0.1% DMSO in separate wells. The culture plate was then incubated for 6 hours at 37 °C in modular incubator chambers and gassed with 5% CO 2 , 5% O 2 and 90% N 2 . Cells were then washed once, resuspended in a drug-free medium, and cultured further for 66 hours. Susceptibility to DHA was assessed microscopically on thin lms by estimating the percentage of viable parasites, relative to control (% survival rate). For the controls, the RSA 0-3h was also performed on P. falciparum reference clones W2, IPC-4884 and IPC-5202 (BEI Resources, NIAID, NIH, USA). A survival rate > 1% was deemed resistant for RSA.
Ex vivo bioassay P. falciparum-based bioassay was carried out to measure the antimalarial activity of patient plasma identifying if study volunteers were likely to have recently taken antimalarial drugs. Plasma was prepared from blood collected on the screening day for evaluation according to previously published methods 48, 49 . In brief, the complement-inactivated samples were serially diluted and applied in one column to 96-well microplate at 50 µL/well. Two columns with serial dilutions of spiked plasma were added to each plate as controls.
In addition to the plates with unknown samples, one plate was dosed with six serial dilutions in duplicate of DHA (from 100 to 2.5 ng/mL). A suspension of 200 µL of malaria parasite-infected red blood cells (W2 clone; 0.05% parasitemia with > 80% rings at a 1.7% hematocrit) was added to all wells. The microplates were placed into a chamber, ushed with a mixture of gas consisting of 5% CO 2 , 5% O 2 and 90% N 2 , and plated into an incubator at 37°C for 72 h. HRP2-ELISA was performed as described above.

Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS) Analysis
To detect baseline antimalarials prior to treatment in the study population, plasma samples were extracted by using protein precipitation 2-fold volume of acetonitrile containing internal standard, then 1-minute vortex mixing, 10 minute-centrifuged, ltered supernatant with 0.22 µm PTFE lter and then transferred to HPLC vial. The LC separation was performed on ACQUITY UPLC (Waters) coupled with tandem mass spectrometer (Xevo TQ-S, Waters) and eluted on Waters Acquity UPLC® BEH C18, 2.1 x 50 mm, 1.7 µm column at a ow rate of 0.5 mL/minute in 8 minutes run time. Mobile phase consisted of (A) 5 mM ammonium acetate pH 4.5 in water and (B) 5 mM ammonium acetate pH 4.5 in acetonitrile:methanol (50:50 v/v). The gradient started with 10% B, raised to 98% B in 6 minutes and held at this composition for 0.5 minute, decreased to 10% B and re-equilibrated for 1 minute. Column temperature was set at 40 °C. Selective mass to charge (m/z) transition for each compound was monitored as follow: AS (407.4 > 261. 23 15).

Statistical analysis
Statistical analysis was performed using GraphPad Prism version 6.0 (GraphPad Software, Inc., San Diego, CA, USA). The difference of data between groups was assessed by nonparametric Kruskal-Wallis, Mann-Whitney or Dunn's multiple comparison tests, as appropriate. Statistical signi cance was de ned as a P value< 0.05.

Results
Study population, demographic, and parasitological parameters In total, 117 individuals were enrolled but 5 individuals were excluded from analysis due to lack of P. falciparum on a blood smear in 4 patients and P. vivax monoinfection in 1 patient. Therefore, 112 individuals with uncomplicated P. falciparum were included in the analysis ( Table 1). The median age of the participants was 23 years (IQR: 22-30.75) and most were male (98.2%), in the military (83%), and from Sisaket Province (96.4%). Among rapid diagnostic test and/or microscopy positive malarial isolates, 108 samples (96.4%) were con rmed by PCR as P. falciparum infections and 4 samples (3.6%) as mixed P. falciparum and P. vivax infections. The geometric mean of parasitemia of the participants was 12,097 parasites/µL. Haplotype and Copy number variation (CNV) of P. falciparum isolates The parasite isolates were categorized into nine groups according to their genotypes of pfmdr1, pfk13, and pfcrt in combination with their CNV of pfmdr1 and pfpm2 (Table S3). Overall, the most prevalent parasites were those in group III, containing pfk13-580Y alleles, multiple pfpm2 copies, pfcrt-F145 alleles and pfmdr1 184F alleles with a single copy number. This was followed by parasites in group II which was the same as Group 3 except with pfpm2 single copy number.
The number of parasites in group II decreased from 55.6% to 13%, while those in group III increased from 7.4% to 56.5%. Parasites in group VI, containing the pfcrt-145I and multiple pfpm2 copies, were also found to be increasing over the study period. It was noted that parasites in group VII to IX, harboring the pfk13-539I/T alleles with no novel mutation on pfcrt, and pfmdr1-184F only held a single pfpm2 copy.

Prevalence of molecular markers for ART-, MQ-and PPQ-Resistance
With limited number of ACT options, we assessed if any parasites had all mutations for ART-, MQ-, and PPQ-resistance (pfk13 SNPs, pfmdr1 CNV, pfpm2 CNV, and pfcrt-F145I mutation). No tested parasites in this study carried all the aforementioned markers, although this is largely driven by having only 5 isolates with multiple copies of pfmdr1. If that marker is excluded, only 11.6% of the isolates (13/112) harbored pfk13-C580Y, multiple pfpm2 copies, and pfcrt-F145I (3 isolates in 2014 and 10 isolates in 2015). All of the 16 parasites with pfk13-R539T carried a single pfpm2 copy number with no pfcrt-F145I mutation but four of them had multiple pfmdr1 copies.  Fig. 1 shows the prevalence of pfpm2 copy number variation, pfmdr1 copy number variation, pfcrt-F145I, pfk13-C580Y and pfk13-R539T mutations over time. The prevalence of parasites with multiple pfpm2 copies associated with PPQ resistance increased from 2013 to 2015, similar to the prevalence of parasites harboring pfcrt-F145I and pfk13-C580Y mutations. In contrast, the prevalence of parasites with multiple pfmdr1 copies associated with MQ resistance and pfk13-R539T mutation decreased after 2013 through 2015. There was a modest increase in DHA and AS susceptibility, which did not change dramatically over the 3-year period. All parasite isolates had the IC 50 -ATQ lower than that of the ATQ-resistance C2B reference (IC 50 -ATQ = 13,240 nM), suggesting no ATQ-resistant phenotype. No signi cant changes in drug susceptibility of DHA, LUM and ATQ were observed during the period of this study.

In vitro Ring stage Survival Assay (RSA) and pfk13 mutations
To get a better understanding of ART-resistance, 40 isolates from 2013 to 2015 were tested in in vitro RSA 0-3h to measure % survival rate against DHA, and the association with pfk13 mutations assessed ( Fig. 2A). One isolate was excluded due to the growth rate between 0 and 72 hours being less than 1.5, leaving 39 isolates evaluable by RSA 0-3h  In the light of an association between PPQ resistance and haplotypes,

Discussion
Even though the Greater Mekong Subregion (GMS) has long been the epicentre of antimalarial drug resistance, these countries are aiming to achieve malaria elimination by 2030 50 . At the time this study was enrolling volunteers, the only validated molecular markers for ACTs were pfmdr1 but by time of writing this manuscript we developed assays to delineate markers and in vitro assays to ART, PPQ, and ATQ resistance. As the number of malaria cases in the GMS continues to decline following aggressive malaria control efforts, conducting P. falciparum therapeutic e cacy studies has become more di cult. Therefore, current efforts have to rely more on the surveillance of molecular markers in concert with in vitro data. This type of intensive surveillance of molecular and in vitro data with a minimum of 6 week patient follow up is effective and easily adapted to mobile populations such as military were the majority of volunteers enrolled. Similar approaches can be applied to migrants and occupations where follow up post treatment is di cult, including workers in forestry, agriculture or animal husbandry and refugee populations. Despite the limitations of the study, where clinical treatment outcomes were not readily available, we were able to characterize the changes in drug resistance overtime in this di cult to reach population. Since most cases of malaria in Thailand occur in border regions, areas where a military presence is required, Thai soldiers stationed at these locations are at increased risk of developing multidrug resistant malaria. There has been limited information published on drug resistance in the military cohorts and this provides new insights.
Chloroquine resistance was rst reported in the GMS since the 1950s 51 , and several groups [52][53][54][55][56] have characterized CQ resistant parasite lineages as one of four mutant pfcrt genotypes at positions 72-76 (CVIET, SVMNT, CVMNT, and CVMET; mutation underlined), with the SVMNT genotype being detected in Brazil/Peru and Melanesia lineages 57,58 . Mutations in pfcrt are associated with CQ, AQ and LUM resistance; especially, the pfcrt 72-76 CVIET and SVMNT haplotype. The 76T point mutation is the key marker, while SVMNT haplotype is required for AQ resistance 53,55 . The present results show that all collected parasite isolates harbored the pfcrt 72-76 CVIET haplotype similar to previous reports [59][60][61] . Even though CQ sensitivity of the parasite isolates was still slightly higher than that of the CQ-resistance W2 reference clone, when the geometric mean IC 50 s of the CQ-sensitive isolates of 30.1 nM 62 was applied, all the parasites collected under this study would be deemed CQ-resistant. Decreased CQ sensitivity was observed from 2013 to 2015 that can likely be attributed to continued CQ use in Thailand for treatment of P. vivax.
Parasite isolates carrying the artemisinin resistant gene have been reported in Thailand 61,63 . Some pfk13 mutations (F446I, N458Y, M476I, Y493H, R539T, I543T, P553L, R561H and C580Y) were validated for conferring artemisinin resistance 64 , while several other mutations may be deemed candidates. The current study showed that the pfk13-C580Y mutation was predominant, rising from 63% in 2013 to 100% in 2015. These ndings are in good agreement with those reported by others 42,61 in that several pfk13 mutations developed in Sisaket (62% 580Y) 42 but by 2015 the C580Y mutation became the sole predominant mutation found 61 . This pattern was not followed by other regions of Thailand. In southern Thailand (Yala Province) very few isolates with pfk13 mutations have been detected 61,65 . The pfk13-R539T has been previously identi ed in Thailand, Cambodia, Laos, and Vietnam 66-69 ; however, the pfk13-R539I found in this study has not yet been reported in Thailand. The pfk13-R539I mutation was previously observed at 0.9% in Ghanaian isolates 70 . As assessed by the RSA 0-3h , the R539I mutation appears not to be associated with ART resistance. The R539I mutation provided the lower stabilization score, implying an effect on the protein structure and function 70  Additionally, no noticeable increase in atovaquone IC 50 of parasites from this study was observed in comparison to the previous report, suggesting that atovaquone/proguanil can still be utilized for treatment and prophylaxis of multidrug-resistant P. falciparum malaria in Thailand. Of all the pfmdr1 SNPs analyzed, allelic variation was only observed in pfmdr1 position 184. This is in good agreement with the previous report by Thita et al. 73 , in which 89% of P. falciparum isolates from Thai-Cambodia border from Chanthaburi and Trat province from 1988 to 2016 had the pfmdr1 184F allele. Previous studies in Uganda and Bioko Island suggest that this allele may play a certain role in mediating resistance to some antimalarials 74,75 , and while in the current study, decreased susceptibility to QN, CQ, and MQ was observed, causality cannot be established. Different results were observed from the P. falciparum isolates from the southern part of Thailand, with the pfmdr1 86Y allele signi cantly more common 76 .
In addition to SNPs in established malaria resistance markers, pfpm2 and pfmdr1 copy numbers were quanti ed as associated with PPQ and MQ resistance, respectively. The increased trend in multiple pfpm2 copy numbers were clearly seen with a decline in multiple pfmdr1 copy numbers. Novel mutations of the pfcrt gene downstream of the 4-aminoquinoline resistance locus (positions 72 to 76), including T93S, H97Y, F145I, I218F, M343L, or G353V, were recently shown to be associated with PPQ resistance 30,32,33  and pfpm3 in the 3D7 genetic background did not alter the sensitivity of P. falciparum to PPQ, suggesting that the increase in pfpm2 copy number alone is not the sole modulator of PPQ resistance 47,77 , the initial selection of pfpm2 and pfmdr1 copy number variations developed a genetic background important for novel pfcrt mutations to emerge 78 . The increased pfpm2 copy numbers was also accompanied with the escalated numbers of pfk13-C580Y and pfcrt-F145I mutations.
Overall, from 2013-2015, P. falciparum isolates collected from the military population demonstrated progression from 62% to 100% 580Y pfk13 mutations, all with increased survival rates by RSA. Regarding ACT partner drugs, there was a decrease in multiple pfmdr1 copy numbers from 11% to 0% and a corresponding increase of multiple pfpm2 copy numbers or multiple pfpm2 copy numbers with pfcrt-F145I mutation from 7.4% to 78.3%. The rst-line antimalarial used during this time was AS-MQ, so the drive toward increasing PPQ resistance likely came from cross-border mixing of parasite populations from Cambodia. In 2013, an AFRIMS study in Anlong Veng, Cambodia, 12 km from the Cambodia-Thai border at Sisaket province, showed a similar rate of 65% C580Y pfk13 mutant 41 .
When the 54% failure rate of DHA-PPQ was observed in Anlong Veng in 2013 41 , the molecular markers for neither ART nor PPQ were available so it is only retrospectively that the association of molecular markers and in vitro data can be seen. Similarly in Sisaket, the 87% failure rate of DHA-PPQ from the TRACII study (2015-2018) 63 may have been predicted from these surveillance data.
The bene ts of having information about past medical history of malaria can help interpret ex vivo/in vitro drug testing results obtained from study samples. It also helps national programs better understand how often anti-malarial drugs are being used within

Conclusion
Altogether, data from the molecular, in vitro drug susceptibility and survival assay indicate the increased trend of pfk13-C580Y mutation, multiple pfpm2 copy numbers and pfcrt-F145I which is indicative of ART and PPQ resistance of Thai P. falciparum isolates collected near the Cambodia border, at the same time when DHA-PPQ treatment failures in Cambodia were on the rise. The sharp rise of IC 90 values was found to be linked to the presence of multiple pfpm2 copies and pfcrt-F145I mutation, whereas the pfk13-R539I mutation was not associated with ART resistance according to RSA. The observed drop in PPQ, MQ, and CYC sensitivity could be an alarming issue for areas where treatment options remain limited. In vitro characterization. (A) RSA and pfk13 mutations. RSA0-3h survival rate for standard laboratory-adapted clones (W2 for ARTsensitive control, IPC-4884 and IPC-5202 for ART-resistance control) and culture-adapted clinical isolates. The dashed line represents the 1% survival rate cut-off that differentiates ART-resistance (≥ 1% red-dashed line) from ART-sensitive (< 1%) parasites in RSAs.
Copy number variations of pfpm2 are shown on the x-axis, and PPQ IC90 is shown on the y-axis (log10 scale). Red circles represent parasites harboring pfcrt-F145I mutation. Bars represent median and interquartile range.