Background
Monitoring the emergence and spread of drug resistance is a key priority for malaria control and elimination efforts in the modern era [
1]. The recent emergence of artemisinin resistance in South-east Asia demonstrates that drug resistance can spread rapidly across national borders and how its prevalence can vary substantially even within individual countries [
2]. Effective containment strategies will require comprehensive, accurate, real-time estimates of drug resistance status from multiple sites that can identify the potential for treatment failure within small geographic regions.
In vivo evaluations of drug susceptibility are expensive and logistically challenging, requiring a substantial time-commitment from participants and appropriate clinical research infrastructure at each study site. Sample size constraints and the confounding effects of local transmission (risk of re-infection) during follow-up can compromise comparisons between different geographic locations. In vivo evaluations are, therefore, inappropriate for rapid and comprehensive comparisons of drug resistance between distinct sites within a single country or small geographic area.
The recent development of well-characterized molecular markers that correlate strongly with in vivo drug resistance has meant that there are now feasible alternative in vitro methods of characterizing population drug resistance. Resistance to sulfadoxine-pyrimethamine (SP) in
Plasmodium falciparum is clearly linked to mutations in the dihydrofolate reductase (
pfdhfr) and dihydropteroate synthase (
pfdhps) genes with selection thought to occur as a step-wise process leading to highly resistant triple to quintuple mutant parasites [
3]. There are orthologs of these genes in
P. vivax [
4‐
8], and
pvdhfr mutations at codons 57, 58, 61, 117, 173 correlate with SP resistance [
9,
10].
P. falciparum 4-aminoquinoline resistance correlates with mutations in
pfcrt [
11] and
pfmdr1. However, the relationship between 4-aminoquinoline resistance and mutations in the
P. vivax ortholog
(pvmdr1) remains controversial [
12‐
14] even though it seems that a codon 976 mutation is associated with AQ or CQ drug resistance in the Asia–Pacific region [
13].
Of the four
Plasmodium species present in Papua New Guinea (PNG),
P. falciparum and
P. vivax predominate [
15,
16]. The prevalence of each species differs according to the geography of the country and among age groups [
17]. This complex epidemiology makes malaria control in PNG very challenging. Early malaria control efforts in PNG included mass drug administration (MDA) in the 1950s and 1960s [
18,
19]. Employed variably throughout the country, some areas received MDA with chloroquine, some with pyrimethamine and others no drug treatment at all [
20]. Anti-malarial treatment policy for case management was chloroquine (CQ) for adults or amodiaquine (AQ) for young children until 2000 [
21]. However,
P. falciparum CQ resistance had started to emerge as early as the 1970s [
22] and in vivo resistance rates as high as 47 % were documented in Madang Province by the early 1980s [
20]. Resistance in
P. vivax emerged much later, with the first well-documented CQ treatment failure described in 1989 [
23,
24] and remained at 4.5 % in East Sepik by 2000 [
25].
In 2000, PNG treatment policy was changed to incorporate SP as first-line therapy in combination with AQ or CQ. At around that time,
P. falciparum PCR-corrected failure rates ranged from 18.2 to 28.6 % for CQ-SP and 10.3 to 21.5 % for AQ-SP in different regions [
26]. For
P. vivax, estimates of resistance to AQ/CQ-SP ranged from 0 to 7 % in the same provinces [
26] whilst uncorrected treatment failure rates of 49 % (28 days) and 87 % (42 days) were documented in Madang and Sepik in 2005–2006 [
27]. However, in contrast to
P. falciparum, the absence of validated markers to distinguish recrudescence from reinfection have meant that in vivo comparisons of
P. vivax therapeutic efficacy between sites with different malaria epidemiology are confounded by both new infections and hypnozoite relapses during follow-up, especially in areas of high transmission [
28].
Following observations of heterogeneity of in vivo drug resistance in Madang and East Sepik provinces of PNG [
26,
29], the utility of an established ligase detection reaction fluorescence microsphere assay (LDR-FMA) was investigated [
29]. A novel primer set was multiplexed to characterize mutations associated with SP and 4-aminoquinoline resistance in both
P. falciparum and
P. vivax from cross-sectional population surveys conducted in these two provinces. The aim of the study was to demonstrate the feasibility of this high-throughput technology as a practical and cost-effective means of rapidly comparing susceptibility to multiple drugs at numerous geographic sites.
Discussion
The present study shows that the highly multiplexed LDR-FMA platform performed well in characterizing multiple loci related to 4-aminoquinoline and antifolate drug resistance in both P. falciparum and P. vivax. Through its successful application in low parasite density infections (from cross-sectional population surveys of mostly asymptomatic individuals) and by generating results highly concordant with existing in vivo drug resistance data (reflecting both spatial and temporal differences in population drug resistance profiles), the LDR-FMA platform can be used for rapid, low-cost application at multiple field sites. This makes it an attractive tool for national malaria programmes that require comprehensive local level characterization of drug resistance patterns.
Existing LDR-FMA assays [
33,
34] were significantly improved by using new PCR primers designed to allow visualization of the five amplified fragments of
pfmdr1,
pfcrt, pfdhfr and
pfdhps on an agarose gel in a single multiplexed assay. Compared with previously published LDR-FMA protocols, the new assay requires much less total DNA, less technician time and has lower overall assay costs. Even though the assay was performed in samples from mostly asymptomatic individuals who a had much lower parasitaemia than previous studies utilizing samples from patients with symptomatic infections [
34], it was only slightly less sensitive for
P. vivax and equivalent for
P. falciparum [
29,
34]. Multiplexing the PCR in a nested PCR protocol did not, therefore, compromise assay performance significantly. A potential problem in the
P. falciparum component of the assay related to high background signal for the wild type
pfmdr 86 allele (that raised concerns regarding the LDR primer’s specificity in polyclonal infections) was addressed using a PCR RFLP protocol. The
P. vivax assay was also improved by designing an allele specific primer targeting a new polymorphism in the
pvdhfr codon 57.
The prevalence of the
P. falciparum 4-aminoquinoline resistance mutations reported here is very similar to previously published data including a >80 % prevalence of
pfmdr1 86Y in 2000–2005 [
37‐
39] and 86–96 % prevalence of
pfcrt 76T mutation in the early 2000s in East Sepik and Madang [
33,
37‐
40]. The
pfcrt 76T mutation is almost exclusively found as part of the double mutant haplotype
SVMN
T previously described in PNG and South America [
41,
42]. While this mutation has been reaching levels near fixation (96 %) in symptomatic infections from children from the same geographical area (Mugil village, Madang Province) over the last 9 years [
43], a similar prevalence is observed here in asymptomatic individuals of all ages.
A recent analysis by Nsanzabana et al. has suggested that strong selection of
pfdhfr 59R 108N double mutant parasites occurred in East Sepik in the 3 years following the introduction of SP in the early 2000s, with a 2.5 times increase in double mutant prevalence to 60–70 % over this time [
44]. This has progressed with levels approaching fixation in the 2006 surveys and other data from the same provinces in 2005–2007 [
34]. By contrast, the frequency of
pfdhps mutations remained very low in 2006 suggesting some retention of sulfadoxine activity. However, the emergence of significant rates of single and double mutant
pfdhps in the 2010 data (combined prevalence of 31.8 %) suggests that a loss of SP efficacy may be imminent in Madang Province, thus threatening its potential value for intermittent preventive treatment in pregnancy and infants.
Despite the very high overall frequency of many mutations in the present study, at least four mutations (
pfcrt,
pfdhfr, pvdhfr haplotypes and the
pvmdr1 976F) were significantly more prevalent in Madang compared with East Sepik, which is in accord with similar observations of differences in drug resistance mutations between the two areas [
37]. One explanation for this phenomenon is that the Madang region has more functional health-care infrastructure and, therefore, its population has better access to anti-malarial drugs than the more remote populations of the East Sepik [
44], resulting in a higher selection pressure. Distribution of drugs beyond their expiration date leading to sub optimal dosage could also have contributed to the selection of drug resistant parasites.
The significant differences in
pvmdr1 976F mutation prevalence between the two provinces (72 vs 26 % in monoclonal infections) and similar differences observed for
pvdhfr mutant genotypes accord well with the observations from in vivo studies of SP + CQ or AQ efficacy of
P. vivax infections between 2003 and 2005 that demonstrated no treatment failure in East Sepik but 29 % treatment failures in Madang [
26]. Overall, the quadruple
pvdhfr genotype was identified in 38.5 % of the
P. vivax monoclonal infections. This genotype, when found in combination with a mutation at
pvmdr1 976, has previously been associated with AQ-SP treatment failure in a study conducted in PNG children [
9]. Apart from the 647
P mutation that seems to be fixed in the PNG
P. vivax population, mutations in
pvdhps were rarely found.
Plasmodium vivax is characterized by early production of gametocytes during its life cycle (prior to the first symptoms of malaria) so that transmission can occur prior to anti-malarial drug exposure, therefore delaying the selection of drug resistant isolates. However, in PNG, the high incidence of mixed
P. falciparum/
P. vivax infections may result in an early exposure of concurrent
P. vivax infections to anti-malarial drugs at the time of treatment of symptomatic malaria caused by
P. falciparum. Additionally,
P. vivax polyclonal infections are very frequent with up to 75 % of all infected carrying several genotypes [
45]. This leads to numerous opportunities for sexual recombination between parasite clones during the mosquito phase of the parasite life cycle and therefore production of meiotic recombinant parasites with drug resistant genotypes. Despite a lower
P. vivax prevalence in East Sepik than in Madang, both parasite populations have shown similar levels of genetic diversity [
36]. Therefore, as with observations of geographic differences in
P. falciparum drug resistance mutants, the higher prevalence of
pvmdr1 and
pvdhfr mutants in the Madang
P. vivax population may reflect greater drug pressure in the Madang area.
Like most malaria-endemic countries, PNG has now adopted WHO recommended artemisinin combination therapy (ACT) as the cornerstone of its anti-malarial case-management strategy. This change to ACT has already been associated with significant reductions in malaria prevalence elsewhere but the emergence and spread of artemisinin resistance in South-East Asia [
2], and observations of variable response to partner drugs (such as lumefantrine) [
46] suggest the need for ongoing surveillance of drug susceptibility to both components. Recent identification of an association of particular single nucleotide polymorphisms in the
kelch-
13 gene with artemisinin resistant phenotypes [
47] may enable incorporation of markers of artemisinin resistance in combination with those evaluated in this study. As demonstrated here, the LDR-FMA platform is highly flexible and integration of new markers can easily be achieved.
It is also notable that lumefantrine exerts a selective pressure on
pfmdr1 N86Y in an opposing direction to that of AQ and CQ [
46,
48,
49]. Therefore, reversion to wild-type genotypes following uptake of artemether-lumefantrine may indicate diminishing parasite susceptibility to lumefantrine but may also raise the prospect of resurrecting AQ or CQ as effective treatments.
Authors’ contributions
CB, LT were responsible of laboratory work, data analysis and writing up the manuscript. SJ, EM and JI participated in the laboratory work and in editing the manuscript. CK contributed species typing data and in editing the manuscript. LR, NS, BK, LJR contributed to field work and in editing the manuscript. IM, JCR, HK, TMD, PAZ and PMS helped with data analysis and writing up the paper. All authors read and approved the manuscript.