Background
Resistance of
Plasmodium falciparum to anti-malarial drugs in Papua New Guinea (PNG) began with chloroquine (CQ) in the 1970s [
1] and has since extended to amodiaquine [
2] and sulphadoxine-pyrimethamine (SP) [
3]. Because of this trend, together with efficacy data from a large-scale, multi-arm, treatment trial conducted in coastal PNG from 2005 to 2007 [
4] and World Health Organization management guidelines at the time [
5], artemisinin combination therapy (ACT) was introduced nationally as recommended therapy for uncomplicated malaria in 2010 [
6]. Artemether (AM) plus lumefantrine (LM) is currently first-line and dihydroartemisinin (DHA) plus piperaquine (PQ) second-line treatment in PNG, but artemisinin plus naphthoquine (NQ) is also available in the private sector [
7]. Resistance to artemisinin derivatives has, however, emerged in recent years in Southeast Asia [
8], and is a concern for countries such as PNG in which ACT is now widely used.
Regular testing using economical, robust and sensitive
in vitro anti-malarial drug susceptibility assays is an integral part of the surveillance for parasite resistance [
9]. Of the different methods currently available, those based on fluorescence measurements of parasite growth using inexpensive intercalating DNA stains such as Sybr Green and Pico Green have proved efficient and inexpensive without loss of sensitivity [
10,
11]. Additional insight into mechanisms of resistance is provided by detection of single nucleotide polymorphisms in parasite genes determining drug response [
12], including mutations in the
P. falciparum CQ transporter (
pfcrt), multidrug resistance 1 (
pfmdr1), dihydrofolate reductase (
pfdhfr), and dihydropteroate synthetase (
pfdhps) genes.
The most recent parasite drug resistance data from PNG were collected as part of the comparative intervention trial conducted in coastal Madang and East Sepik Provinces between 2005 and 2007 [
13,
14]. Most of the isolates tested proved resistant to CQ
in vitro but not to other ACT partner drugs or to the artemisinin derivatives themselves [
13]. Consistent with this finding and previous heavy 4-aminoquinoline/SP use, there was near-fixation of
pfcrt K76T,
pfdhfr C59R and S108N, and
pfmdr1 N86Y alleles, while multiple mutations were frequent [
14].
To determine whether there has been any recent change in P. falciparum drug resistance in the north coastal PNG area, the in vitro susceptibility of local P. falciparum isolates collected between 2011 and 2013 to artemisinin derivatives and ACT partner drugs were re-assessed, and the prevalence of drug resistance markers in the same parasite strains re-examined.
Methods
Study sites, patients and ethical approval
Venous blood samples were obtained from 52 children aged six months to five years with an uncomplicated
P. falciparum mono-infection at a parasitaemia >0.5% who were recruited at Mugil (n = 43) and Alexishafen (n = 9) health centres in Madang Province to a randomized, comparative, efficacy trial of the ACT AM-LM and artemisinin-NQ (Australian New Zealand Clinical Trials Registry ACTRN12610000913077) [
15]. The study received ethical approval from the Medical Research Advisory Committee of the PNG Department of Health (MRAC #10.39). In all cases, informed consent was obtained from the parents or legal guardians before recruitment and blood sampling.
Drug susceptibility assays
A Sybr Green fluorescence assay was used to assess drug susceptibility. All assays were carried out at the PNG Institute of Medical Research in Madang. The methodology used, a modified version of that first described by Smilkstein
et al. [
11], has been previously validated against tritium hypoxanthine incorporation,
Pf lactate dehydrogenase (
PfLDH), light microscopic schizont maturation, and flow cytometry-based drug susceptibility assays using the laboratory-adapted parasite strains 3D7, E8B and W2 [
16]. For the present series of experiments, the 3D7 strain was used as reference with a mean CQ
in vitro concentration required for 50% parasite growth inhibition (IC
50) value of 14.3 nM. This compares with IC
50 values of 18–20 nM for tritiated hypoxanthine isotopic assay and 23–33 nM for
PfLDH assay using this strain in our laboratories at Fremantle Hospital in Australia (unpublished observations).
The anti-malarial compounds used in this assay were purchased from Sigma-Aldrich, St Louis, MI, USA (CQ diphosphate), Santa Cruz Biotechnologies, Santa Cruz, CA, USA (pyronaridine (PY) tetraphosphate), Hubei Onward Bio Development Co Ltd, Enshi City, Hubei, China (DHA, artesunate (AS), AM, LM) or kindly donated by Mangalam Pty Ltd, Bangalore, India (PQ phosphate and NQ phosphate). Solutions of 10 mM concentration were prepared for each drug in an appropriate solvent (CQ, PQ and PY in deionized water; AM in methanol; NQ in 50% v/v ethanol; LM in 1:1:1 v/v linoleic acid/Tween 80/ethanol; DHA in 70% v/v ethanol; AS in ethanol). These solutions were further diluted to a stock 1 mM concentration in deionized water. After sterile filtration, stock solutions were aliquoted into airtight microcentrifuge tubes and stored at −20°C. A fresh aliquot was used for each assay.
Red blood cells from slide-positive children were washed three times in standard RPMI 1640-based malaria cell culture medium [
17] and, if necessary, diluted to 0.5-1.0% parasitaemia with red blood cells from a malaria-naïve donor of blood type O Rhesus negative. The culture medium consisted of RPMI 1640 HEPES (Sigma Aldrich, St Louis, MO) supplemented with 92.6 mg/L L-glutamine (Sigma Aldrich, St Louis, MO), 500 μg/L gentamicin (Sigma Aldrich, St Louis, MO), 50 mg/L hypoxanthine (Sigma Aldrich, St Louis, MO) and 0.5% w/v Albumax II lipid rich BSA (Life Technologies, Mulgrave, Victoria, Australia) [
16]. Drug dilutions were set up in 96-well plates in triplicate, with eight dilutions for each drug. The haematocrit was set at 1% and the liquid volume per well was 200 μL. The assay plates were incubated for 48 hr in a candle jar using the method of Trager and Jensen [
18], after which 50 μL of a red cell lysis buffer/Sybr green (Invitrogen, Carlsbad, CA, USA) mixture were added to each well. The plate was incubated for 15 min in the dark. Fluorescence was read on a microplate reader (Fluostar Optima, BMG Labtec, Offenburg, Germany) equipped with a 484 nm excitation filter and a 520 nm absorbance filter.
Molecular analysis
Parasite isolates where tested for genetic markers associated with drug resistance using a multiplex polymerase chain reaction ligase detection reaction fluorescent microsphere assay (PCR-LDR-FMA) assay as previously described [
14,
19]. In brief, PCR-LDR-FMA was performed using established primer sequences to detect single nucleotide polymorphisms in the known resistance loci of
pfdhfr (codons 51, 58, 108, 164)
, pfdhps (codons 436–437, 540, 581, 613)
, pfcrt (codons 72–76) and
pfmdr1 (codons 86, 184, 1034, 1042, 1246)
. Fluorescent products were detected using a Bio-Plex analyzer (Bio-Rad, Hercules, CA, USA). Data analysis was conducted as described previously [
20,
21].
Data analysis
Concentrations of anti-malarial drugs for each isolate and anti-malarial were log-transformed and the fluorescence values were normalized such that the smallest value in each dataset represented 0 and the largest value (drug-free control) unity. The dose–response curve Y = 100/(1 + 10
k (logIC50-logX)) was then fitted to each dataset, where Y corresponds to the percentage of growth at drug concentration X, and k is the Hill slope. For calculations of means and 95% confidence intervals (CI) as well as for analysis of associations between pairs of different anti-malarial drugs, the log10 IC50 values were used as these were normally distributed by Kolmogorov-Smirnov test. Associations between IC50 values were determined using Pearson’s correlation coefficient, and significant pairwise correlations (P < 0.05) were considered moderate for 0.3 ≤ r ≤0.50 or strong for r >0.5.
Factor analysis was conducted after testing the cross-correlation matrix for sphericity using Bartlett’s Test and using the Kaiser-Maier-Olkin statistic to determine the appropriateness of the data for this analysis. The distribution of the Eigenvalues of the cross-correlation matrix indicated that factoring into two components was the most appropriate approach. Since it was hypothesized that the underlying factors relate to the mechanisms of drug action and are thus related to each other, a non-orthogonal (direct-oblimin) rotation was applied to the solution.
Discussion
The present data demonstrate that there have been changes in the drug resistance characteristics of parasite isolates collected between 2005–2007 and 2011–2013 from areas of north coastal PNG with intense malaria transmission. Although a different methodology was used to assess
in vitro sensitivity in the present study, and notwithstanding limitations in assigning thresholds for
in vitro drug sensitivity [
29], more strains appeared CQ-sensitive than in 2005–2007 [
13,
14] despite the majority retaining the mutant
pfcrt K76T allele over time. There were also apparent temporal reductions in the IC
50s of LM and NQ, while those for PQ and DHA increased albeit still within relatively low nM ranges. The proportion of parasites carrying the wild-type
pfdhps gene had fallen over time and more mutations had appeared in
pfdhps in the 2011–2013 isolates, consistent with continued use of SP in the study area. Factor analysis suggested that the
in vitro susceptibilities of PNG
P. falciparum strains to LM and PY may be unrelated to those of other long half-life ACT partner drugs, with the IC
50 of PY clustering with those of the artemisinin derivatives. Interpretation of the present and previous data needs to take into account several factors. These comprise i) temporal changes in anti-malarial drug use in the study areas, ii) potential effects of the introduction of non-pharmacological strategies to reduce malaria transmission, and iii) differences in assay methodology between 2005–2007 and 2011–2013.
Recommendations regarding replacement of regimens based on CQ and SP by ACT for treatment of uncomplicated malaria in PNG children were implemented in 2010 [
6], but translation of this policy into practice has been slow. In addition, the use of CQ and SP as first-line intermittent preventive therapy (IPT) in pregnancy has continued, and an IPT trial in infants involving SP was conducted in the Mugil area (from where most of the present isolates were collected) between 2006 and 2010 [
30]. Therefore, CQ-SP drug pressure had been reduced, but not eliminated, over a period of two to three years leading up to isolate collection in the study area.
The dynamics governing repopulation by CQ-sensitive strains in areas in which CQ treatment pressure has been removed completely are not well understood, but the time-scale is probably approaching a decade [
24,
31,
32]. The fact that the present isolates were collected after a short period of incomplete removal of CQ-SP pressure is reflected in the present molecular analyses which showed no reduction or an increase in parasites carrying genetic markers that correlate with CQ-SP resistance. However, CQ resistance mutations are frequently found in isolates that show
in vitro susceptibility [
24,
25,
33], and there is also evidence that CQ IC
50 values can fall relatively quickly (within a few years) after reduction in drug pressure [
34,
35].
The introduction of long-lasting insecticide-impregnated bed nets (LLINs), such as was started on a large scale in PNG in 2004 [
36], could theoretically also attenuate drug pressure by reducing malaria transmission. The evidence for this effect on molecular resistance markers in studies from sub-Saharan Africa is conflicting [
37,
38]. However, these studies were relatively short-term compared with the time needed for re-establishment of full sensitivity after drug withdrawal [
24,
31,
32] and no
in vitro susceptibility data were presented. It remains possible that increasing LLIN use in coastal PNG between 2005–2007 and 2011–2013, together with the partial replacement of CQ-SP by ACT, both contributed to the lower IC
50 for CQ in the present study.
There is no accepted standardized protocol for determining
in vitro anti-malarial drug susceptibilities and the results may differ according to the methodology employed. There is evidence that the
PfLDH assay generates higher IC
50 values than other methodologies including the Sybr Green assay used in the present study [
16,
39,
40]. However, the reported differences are typically modest (typically 10–30 nM across a range of IC
50 values, as seen with our own data for the 3D7 strain) compared with the substantial reduction in CQ IC
50 observed between 2005–2007 and 2011–2013. The increase in PQ IC
50 over time in north coastal PNG might appear paradoxical given than CQ and PQ susceptibility have both been considered to reflect
pfcrt mutations [
41]. Nevertheless, not all studies show this relationship [
42], while the PQ IC
50 values in both time periods were both well below the conventional 100 nM cut-point for resistance in all but one isolate in the present study. The small temporal increase in DHA IC
50 may also be of no clinical significance given that the IC
50 value in all isolates was in the very low nM range. Standard drug susceptibility assays do not detect early stage artemisinin resistance defined by a slow parasite clearance time for which there is now a molecular marker [
43], but the isolates were obtained between 2011 and 2013 from patients in a clinical trial [
15] in which there was no evidence of longer parasite clearance times after AM-LM than in the equivalent trial conducted from 2005 to 2007 [
4].
Several studies have shown moderate to strong correlations between
in vitro parasite responses to CQ and PY [
44,
45] while in others, including the present study, there has been no such association [
46,
47]. The future of PY-containing ACT is uncertain because of hepatotoxicity [
48]. However, it does not appear to exhibit cross-resistance with CQ in the present parasite isolates, which would be an advantage if PY-based ACT became available for repeated use in PNG. The present data confirm the moderate associations between CQ, PQ and NQ which were also observed previously in north coastal PNG [
13]. Although the IC
50s for the latter two compounds are relatively low, their significant association with CQ susceptibility may have implications for the longevity of ACT formulations incorporating them. The lack of clustering of LM susceptibility with other longer half-life anti-malarial drugs suggests that it may have an independent mechanism of action. As has been done by other groups [
49-
51], we included the three artemisinin drugs in common clinical use, even though DHA is the active metabolite of both AS and AM
in vivo, since there is evidence that their activity against
P. falciparum in vitro is not uniform [
52]. Consistent with this latter observation, the association between AM and DHA was only moderate while those between AS and both AM and DHA were the strongest observed.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
All authors contributed to the design of this study, collection of data, analysis of data and/or the interpretation of the results, and to the writing of the manuscript. All authors edited and approved the final version of the manuscript.