Screening of patients and sample collection

Dried blood spot samples were collected from patients with microscopically confirmed P. falciparum infection during a one-year period from October 2013 to September 2014 in five malaria-endemic areas. The studied sites were Ann in Rakhine State, Homemalin in Sagaing State, Myit Kyi Nar in Kachin State, Kyauk Me in Northern Shan State, and Phruso in Kayah State. Accurate and recent P. falciparum transmission data are not available from these sites, although up to 2010 the three administrative regions spanning the western Myanmar border (Rakhine and Chin States and Sagaing Region) appeared to have the highest transmission levels in the country (generally PfPR2-10 >5 to <40 %, or ‘intermediate levels’) with lower prevalence in central and eastern areas [17].

Febrile patients in the respective study sites were recruited by passive case sampling at local clinics run by the Vector Borne Disease Control Programme and by active case sampling at nearby villages. Positive rapid diagnostic test (RDT) results for falciparum malaria were confirmed by thick and thin blood film examination. Patients with parasite counts of 250 parasites/µl or more in the thick blood film were eligible for the study. After obtaining written informed consent, dried blood spot samples from participants were collected on Whatman 3MM filter paper (Maidstone, UK). The short-term efficacy of artemether-lumefantrine therapy via assessment of day-3 parasitaemia (after completion of directly observed treatment) in a sub-set of patients studied in Homemalin and Kyauk Me were also done. The study was approved by the Ethical Committees of Faculty of Tropical Medicine, Mahidol University and Department of Medical Research (lower Myanmar).

Detection of k13 mutation

K13 sequences corresponding to amino acids 210–726 were obtained; results from an initial set of 74 samples were published as a part of a previous study [18] (Additional file 1). Nested PCR methods were as described previously [14]. Briefly, each PCR reaction contained 10 mM Tris-HCL buffer, 2 mM or 3 mM MgCl₂, 250 µM 4-deoxynucleotide triphosphate (dNTPs), 250 nM oligonucleotide primers, 0.5 units (2 units for the second nested PCR) Platinum Taq DNA polymerase (Invitrogen, USA) and 2 µL of genomic DNA template (or 5 µL of PCR product from the first nested PCR) within a final reaction volume of 100 µL. A total of 25 PCR cycles for nested 1 and 35 cycles for nested 2 were undertaken using a MyCycler ™ thermal cycler (Bio-Rad Laboratories, USA). The temperature cycles consisted of pre-denaturation at 94 °C for 5 min followed by 94 °C for 1 min, 58 °C for 2 min (1 min for the second nested PCR), 72 °C for 2 min (1 min for nested 2) and final extension at 72 °C for 10 min. DNA template from the AQ strain (without k13 mutation) was used as a positive control and a DNA free template was also included as negative control in each PCR reaction. Sequencing of k13 was obtained at Macrogen, Republic of Korea. Multiple nucleotide sequence alignments were analysed by BioEdit version 7.1.3.0, using 3D7 k13 sequence as a reference to detect point mutations in the gene. PCR and sequencing reactions were repeated when amplification did not occur or when the relevant portion of the sequencing chromatogram was unclear at any point.

Analysis of pfmdr1 copy number variation

Copy number variation in the pfmdr1 gene was assessed using a 5-plex Rotor Gene Q (Qiagen, Valencia, USA). The real-time PCR method was described previously [16, 19]. Briefly, each 10 µL PCR reaction contained 5 µL of 2x QuantiTect Multiplex PCR NoROX Master Mix (containing HotStarTaq DNA Polymerase and QuantiTect Multiplex PCR Buffer), 0.4 µM of each forward and reverse primer for both the pfmdr1 gene and the Pf β-tubulin gene, 0.2 µM of each probe, and 2.5 µL of template DNA. In each run of real-time PCR, 3D7 and H26 clones were used as calibration controls with single and multiple copies of pfmdr1 gene, respectively, and negative control containing no template DNA was also included. Test samples were assayed in triplicate. The temperature profiles for real-time PCR were pre-denaturation at 95 °C for 15 min followed by 45 cycles of 94 °C for 60 s and 60 °C for 60 s. Fluorescence data were expressed as normalized reporter signal. The detection threshold was set above the baseline value for fluorescence of the first 15 cycles. The threshold cycle (Ct) is when the increase in reporter signal first rises above this threshold. Results were analysed by a comparative ∆∆Ct method where ∆∆Ct = (Ct of Pfmdr1 gene−Ct of P fβ-tubulin) of sample- (Ct of Pfmdr1 gene−Ct of Pf β-tubulin) of Pf 3D7 calibrator. The pfmdr1 copy number was calculated as 2^∆∆Ct. Samples with copy number between 1.45 and 1.6, and samples with a spread of ∆∆Ct among the three triplicates of more than 1 were repeated and the second result used.

Statistical analysis

Qualitative variables were presented as proportions with confidence intervals calculated by Wilson’s method and comparison of proportions undertaken by the Chi squared test. Quantitative variables were expressed as median (IQR).

K13 sequence

The overall prevalence of k13 propeller mutations, and the most common mutation in each site differed among sites (Table 1; Fig. 1). Propeller mutation was less common in samples from sites in western border regions (Ann and Homemalin, 29 of 155 isolates) compared to samples from the other three sites (40 of 51 isolates; p < 0.0001). F446I was the most common mutation in Myit Kyi Nar and Homemalin while P574L was most common in Kyauk Me.

Five different non-synonymous mutations in the non-propeller (‘stem’) region of k13 (corresponding to amino acids 210–440) were found in Homemalin or Myit Kyi Nar; three of these were previously unreported. Two infections with previously unreported synonymous mutations (one with E424E in Ann and one with D343D in Homemalin) were also detected.

pfmdr1 copy number variation

Day-3 parasitaemia after artemether-lumefantrine therapy

Blood films from 36 patients (26 in Homemalin and ten in Kyauk Me) taken on day 3 following completion of a directly observed regimen of artemether-lumefantrine therapy were obtained. Only one patient from Homemalin had a positive blood film at this time (parasite count: 50 parasites/µl). Molecular testing showed wild-type k13 and single copy pfmdr1. Among the other 35 samples with negative day-3 parasitaemia, 19 were wild-type k13, nine had propeller mutations, two had non-propeller mutations and five could not be sequenced.

K13 assessment

This study adds to the knowledge of anti-malarial resistance in Myanmar. Approximately one-third of isolates had k13 propeller mutations, with around three-quarters of these being mutations known to be associated with artemisinin resistance. Of the five locations surveyed, only Ann in Rhakhine State had a k13 propeller mutation proportion of less than 10 %. This finding is consistent with previous clinical and molecular data indicating that artemisinin resistance was not present in western Myanmar [9, 21], or in neighbouring Bangladesh [22]; the data also fit with lower levels of transmission (broadly found in the sites outside the western border regions) being associated with the most resistant parasites [23]. The high rate of mutation in Myit Kyi Nar (Kachin State) is consistent with data from neighbouring provinces of China where the F446I mutation is also the predominant type of k13 mutation and associated with slow parasite clearance [12, 24] and reduced in vitro susceptibility [24]. F446I was also the most common mutation in Homemalin, near the northwest border with India, where it was found in 13 of 91 isolates (14.3 %). The P574L mutation was also present in three sites, consistent with its identification in seven provinces [18, 25]; it has been shown to be associated with a relatively long parasite clearance half-life [25].

Most k13 mutations previously associated with slow parasite clearance (in Cambodia, Thailand and Vietnam) were not found in this study. This may reflect different levels of drug pressure and different population histories among study sites, although it is not possible to relate the mutations observed in different locations in Myanmar with particular drug usage histories, since these are largely unknown. The study also documents the presence at low frequency of a number of mutations in the ‘stem’ region of the k13 protein outside the propeller region. Although clearly less common than propeller mutations, mutations in this highly conserved region may have functional consequences [26]. Further studies of k13 mutation in conjunction with parasite clearance assessment or in vitro studies [27] should help to determine the relationship between this expanding list of observed mutations and phenotypic artemisinin resistance.

pfmdr1 copy number variation

In Southeast Asia and in laboratory studies, amplification of the pfmdr1 gene has been shown to be associated with decreased in vitro sensitivity to both mefloquine and lumefantrine [16, 28] and it is associated with mefloquine-artesunate failure in Cambodia [29, 30], presumably in concert with propeller mutations in k13 (present in high proportions in Cambodia for more than a decade). Recrudescences after the six-dose artemether-lumefantrine regimen have not been firmly linked with pfmdr1 amplification [30], although relatively few artemether-lumefantrine efficacy data are available from areas with artemisinin resistance.

Prior to this study, few surveys had been undertaken examining copy number variation in the pfmdr1 gene in Myanmar [31], although considerable data have been obtained from areas along the Thai-Myanmar [5, 32, 33] and China-Myanmar border areas [34, 35]. These studies have generally shown a prevalence of multiple copies of pfmdr1 gene of less than 10 %. In the current study the overall median copy number was 1.03 (IQR = 0.86–1.2) and the proportion of samples with multiple copy numbers (more than 1.5) was less than 10 % in three out of five studied sites (Ann, Homemalin and Myit Kyi Nar).

This generally low rate of isolates with multiple pfmdr1 copies across the country reflects the fact that although ACT, artesunate-mefloquine and artemether-lumefantrine, have been policy for treatment of P. falciparum malaria since 2005, it has been challenging to deploy them particularly in remote regions. Parasites carrying multiple copies of pfmdr1 also have lower fitness in the absence of drug pressure [36]. The higher rates of multiple copies found in eastern areas (Kyauk Me and Phruso) may reflect the relatively small sample numbers (13 and six from Kyauk Me and Phruso, respectively). Nevertheless, the findings raise the suspicion of possible mefloqine resistance in these sites, perhaps related to population migration in and out of areas where relevant ACT are used. Molecular studies with larger sample size combined with clinical efficacy studies are needed to determine more accurately the drug resistance profile in these areas.

Day-3 parasitaemia after artemether-lumefantrine therapy

Among 36 patients assessed in Homemalin and Kyauk Me for day-3 parasitaemia after completion of artemether-lumefantrine therapy, all patients cleared parasite on day 3 except one from Homemalin. This is quite surprising given the overall proportion of isolates with k13 propeller mutations in these regions (28.6 % in Homemalin and 100 % in Kyauk Me), and the day-3 positivity rate appears lower than the previous clinical study done in Shwe Kyin in Lower Myanmar (9 %) [14]. Different starting parasitaemias might explain this discrepancy (minimum count was only 250 parasite/µl in the current study and median parasitaemia was consequently substantially lower). Another possibility is that the quantitative aspects of the relationship between k13 propeller mutation and day-3 positivity may differ according to the specific k13 mutation. Independent studies at the Myanmar-China border indicate that the proportion of day 3-positive patients after artemisinin derivatives is around 5–20 % but the proportion of isolates with k13 propeller mutations (mostly F446I) was around 40–60 % [12, 24]. This differs from the situation in the lower Mekong region where the proportion of day 3-positive patients broadly matches that of k13 propeller mutants [14]. Hence, certain mutations might have relatively mild phenotypes compared to better-characterized k13 mutants, such as C580Y. A number of host variables (including immunity, pharmacology and genetics) might also be contributing factors. Further, the relatively small number of patients assessed for day-3 parasitaemia limits the precision of the estimate of parasite clearance time for artemether-lumefantrine therapy.