Chloroquine resistant Plasmodium vivax: in vitro characterisation and association with molecular polymorphisms.

BACKGROUND: Treatment failure of chloroquine for P. vivax infections has reached high levels in the eastern provinces of Indonesia, however, in vitro characterization of chloroquine resistance and its associated molecular profile have yet to be determined.
METHODS: Using a modified schizont maturation assay we investigated the in vitro chloroquine susceptibility profile and molecular polymorphisms of P. vivax isolates collected from Papua, Indonesia, where high levels of clinical chloroquine treatment failure have been reported, and from Thailand, where chloroquine treatment is generally effective.
RESULTS: The geometric mean chloroquine IC(50) for P. vivax isolates from Papua (n = 145) was 312 nM [95%CI: 237-411 nM] compared to 46.8 nM [95%CI: 34.7-63.1 nM] from Thailand (n = 81); p<0.001. Correlating with the known clinical efficacy of the area, a cut off for chloroquine resistance was defined as 220 nM, a level exceeded in 13.6% (11/81) of Thai isolates and 65% (94/145) of Papuan isolates; p<0.001. Several sequence polymorphisms in pvcrt-o and pvmdr1, and difference in pvmdr1 copy number were identified. A Y976F mutation in pvmdr1 was present in 96% (123/128) of Papuan isolates and 25% (17/69) of Thai isolates; p<0.001. Overall, the geometric mean chloroquine IC(50) in isolates with the Y976F mutation was 283 nM [95%CI: 211-379], compared to 44.5 nM [95%CI: 31.3-63.4] in isolates with the wild type; p< 0.001. Pvmdr1 amplification occurred in 23% (15/66) of Thai isolates compared to none (0/104) of Indonesian isolates (p<0.001), but was not associated with increased chloroquine resistance after controlling for geographical location.
CONCLUSIONS: In vitro susceptibility testing of P. vivax discriminates between populations with differing levels of clinical efficacy of chloroquine. The pvmdr1 polymorphism at Y976F may provide a useful tool to highlight areas of emerging chloroquine resistance, although further studies defining its clinical correlates are needed.

Introduction

The burden of malaria caused by Plasmodium vivax has been greatly under-appreciated both in terms of its clinical spectrum and incidence of disease [1], [2]. P. vivax is the most widely distributed cause of malaria in the world affecting 40% of the worlds population and causing between 147–436 million clinical infections each year [3]. Although associated with less mortality than P falciparum it exerts a considerable morbidity particularly in children and pregnant women. Control measures are confounded by two major factors: firstly, the presence of dormant hypnozoite stages in the liver, which result in relapse infections weeks after the cure of the initial episode, and secondly the emergence of chloroquine resistance.

In most of the world chloroquine remains the first line of treatment for patients with vivax malaria. Not only is it well-tolerated and affordable, but its long half-life provides protection from early relapses following treatment. The first cases of chloroquine resistant P. vivax were reported in 1989 from PNG [4] and northern Papua (formerly Irian Jaya), Indonesia [5], [6], [7], [8]. Chloroquine monotherapy is now virtually ineffective in Papua Indonesia [8], [9], [10] with significant clinical resistance apparent throughout the Indonesian archipelago [5], [11]. More recently sporadic cases have been reported from Myanmar [12], South America [13], [14], Viet Nam [15], and Turkey [16].

Despite these clinical reports, the global prevalence of chloroquine resistant P. vivax remains poorly defined. Clinical studies are difficult to carry out and subject to individual variations in patient immune status, reinfection and frequent relapses. In vitro susceptibility assays provide an alternative means of assessing drug susceptibility of Plasmodium spp. Although these tests have been well established for P. falciparum, their application in P. vivax has been more difficult to develop due to limitations of in vitro culture methods in this species. Recently several centres have reported methods for conducting in vitro P. vivax drug susceptibility which are generally based on the P. falciparum WHO microtest using quantification of schizont maturation [17], [18], [19].

The mechanism of P. vivax chloroquine resistance is unknown and as yet no genetic markers have been identified. In P. falciparum, polymorphisms in pfcrt and pfmdr1 have been shown to confer resistance [20], [21]. However, no associations have been found between point mutations in the orthologue genes, pvcrt-o (pvcg10) and pvmdr1 and the clinical response of vivax malaria to chloroquine [22], [23]. Heterologous systems investigating the effect of pvcrt-o expression on chloroquine response showed a 2.2-fold decrease in susceptibility to chloroquine in P. falciparum transformed with pvcrt-o, suggesting a possible role of pvcrt-o in chloroquine resistant P. vivax [24].

In this study, we investigated the in vitro chloroquine susceptibility profile and molecular polymorphisms of P. vivax isolates collected from Papua, Indonesia, where high levels of clinical chloroquine resistance have been reported [10], [25] and from Thailand where chloroquine treatment is generally effective [26], [27].

Materials and Methods

Field location and sample collection

Clinical isolates were collected between 2003 and 2006 from two sites, one in Indonesia and the other in Thailand. Timika, located in the southern region of Papua Province, Indonesia, has documented clinical chloroquine resistance with day 28 failure rates following chloroquine monotherapy exceeding 65% and 16% of patients having early treatment failure [10]. At the second site at the Shoklo Malaria Research Unit, Mae Sod, Tak Province on the western border of Thailand, P. vivax remains clinically sensitive to chloroquine [27].

Patients with symptomatic infections of pure P. vivax presenting to an outpatient facility were recruited into the study and 5 ml blood samples collected by venepuncture. After removal of host white blood cells using a CF11 column, 2 ml of packed infected Red blood cells (IRBC) were divided as follows: 1 ml was cryopreserved in glycerolyte, 200 µl spotted onto a filter paper and 800 µl was used for the in vitro drug susceptibility assay. Patients were treated with dihydroartemisinin-piperaquine (Indonesia) or chloroquine (Thailand) according to local guidelines, but were not followed routinely thereafter.

In vitro drug susceptibility assay

P. vivax susceptibility to chloroquine was measured at both sites using an identical protocol modified from the WHO microtest as described previously [17]. This method was modified further by reducing the final haematocrit of the blood media mix (BMM) from 4% to 2%; and using 200 µl of BMM per well instead of 50 µl. The BMM were added to pre-dosed drug plates containing serial concentrations of chloroquine with doubling dilutions from 2992 nM to 2.92 nM (Salt). Drug plates were quality assured using P. falciparum clones with known chloroquine susceptibility: the resistant clones K1 and W2 had median chloroquine IC50s of 120 nM and 470nM respectively compared to the chloroquine sensitive clones 3D7 and FC27 with median IC50s of 17 nM and 9 nM.

A traditional candle jar was used to mature the parasites at 37.5°C (25–36 hours) at reduced oxygen concentration. Incubation was stopped when parasites present had matured to at least 40% schizonts in the drug-free control well. A thick blood film was made from each well, stained with Giemsa and examined microscopically. The number of schizonts per 200 asexual stage parasites was determined and the result for each drug concentration normalized to the control well. The dose-response data were analyzed using nonlinear regression analysis (WinNonLin 4.1, Pharsight Corperation) to obtain the IC50 values. To assess the effect of verapamil on chloroquine susceptibility, an additional assay was conducted on 16 Indonesian isolates in which 0.9 µM verapamil was added to serial dilutions of chloroquine and the IC50 calculated and compared to that of chloroquine alone.

Malaria DNA preparation and determinantion of species and hapltotype

Genomic DNA from blood spots and cryopreserved samples was extracted using QIAamp DNA mini kit (Qiagen). Plasmodium species were confirmed using multiplex PCR as previously described [28]. The haplotypes of samples containing a single P. vivax infection were then determined using three polymorphic markers, by sequencing pvama1 [29], the number of pvmsp1 bands after PCR [29] and restriction fragment polymorphism of the pvmsp3 alpha locus [30].

SNP identification in pvmdr1 and pvcrt-o genes

In order to identify relevant polymorphisms in the pvmdr1 and pvcrt-o genes in our parasite population 25 Indonesian and 7 Thai P. vivax isolates (“core” samples), were fully sequenced for both genes using primers listed in Table 1, comparing the sequences to those of the pvmdr1 (GenBank Acc. No. AY618622) and pvcrt-o (GenBank Acc. No. AF314649) of the Sal 1, a chloroquine sensitive strain from Salvador used as a reference strain in this study. All core isolates were single species, monoclonal infections. PCR conditions were as follows: a total volume of 50 µl containing 5 µl of 10×PCR buffer, 2.5 mM MgCl2, 0.20 mM each dNTP, 1 µM each primer and 1.25 U of AmpliTaq Gold DNA polymerase (Applied Biosystems), and 1 µl of genomic DNA. PCR was performed under the following conditions: 95°C for 10 minutes and 40 cycles of at 94°C for 40 seconds, 55°C for 1 minutes and 72°C for 2 minutes. PCR products were sequenced using the BigDye terminator 3.1 (Applied Biosystems). Polymorphisms which were identified in the core samples were then examined in the complete sample set.

Table 1

Primers and sequences used to study mutations in pvmdr1 and pvcrt-o and in the pvmdr1 copy number assay.

A. pvmdr1	Pvmdr1-1F	5′-CTT TTA TGC CTC TCC CCC
	Pvmdr1-1Fb	5′-AGA TTG TTC TGT AGC CGTT
	Pvmdr1-1R	5′-GCG TAA GAT GCT AAA ATG AACC
	Pvmdr1-2F	5′-ATT TAA CCT TTC AGA AAA GCT GT
	Pvmdr1-2R	5′-CCA CCT GAC AAC TTA GAT GC
	Pvmdr1-3F	5′-CTG ATA CAA GTG AGG AAG AAC TAC
	Pvmdr1-3R	5′-ACT ATC CTG GTC AAA AAA GC
	Pvmdr1-4F	5′-CCC TCT ACA TCT TAG TCA TCG
	Pvmdr1-4R	5′-TGG TCT GGA CAA GTA TCT AAAA
	Pvmdr1-5F	5′-GGA AGT TGA TGT CCC TAA AGG
	Pvmdr1-5R	5′-CCT GGC GCG TCT ACT TAG
B. pvcrt-o	pvcg10-1F	5′-CGC TGT CGAAGA GCC
	pvcg10-1R	5′-AGT TTC CCT CTA CAC CCG
	pvcg10-2F	5′-CGC CCG GTA GAA GC
	pvcg10-2R	5′-GGT GAG GCG ACA TGG
	pvcg10-3F	5′-GCT AAG GGC ACA TTT CC
	pvcg10-3R	5′-GTA GTC CTC AAA AGA CAC ACA TC
	pvcg10-4Fa	5′-TAT GAA GCA AAT CGC AAC AA
	pvcg10-4Fb	5′-CTT GAG AGT AAG GCA GCG AA
	pvcg10-4R	5′-TCA TCC AGA GAG CAA ACT TTC TA
C. pvmdr1 976	Pvmdr976 F	5′-GGA TAG TCA TGC CCC AGG ATT G
	Pvmdr976 R	5′-CAT CAA CTT CCC GGC GTA GC
	pvmdr976 internal Internal	5′-CGG CTG TAC TGA CCG GAA CGT A
D. pvmdr1 copy number	Pvmdr F	5′-CTG ATA CAA GTG AGG AAG AAC TAC G
	pvmdrR	5′-GTC CAC CTG ACA ACT TAG ATG C
	pvaldo F	5′-GAC AGT GCC ACC ATC CTT ACC
	pvaldoR	5′-CCT TCT CAA CAT TCT CCT TCT TTC C

List of primers and their sequences used to amplify and sequence pvmdr1 (A), pvcrt-o (B) and identification of the pvmdr1 Y976F mutation (C). Primers used to amplify the fragments of the pvmdr1 and P. vivax aldolase reference gene in the pvmdr1 copy number assay (D).

Determining pvmdr1 copy number

Pvmdr1 gene copy number was estimated by a novel quantitative real time SybrGreen PCR assay using the Mx4000 Multiplex Quantitative PCR system (Stratagene). A single copy gene coding for P. vivax aldolase (GenBank Acc. No. AF247063), was used as a reference (normaliser) gene for estimating the pvmdr1 copy number. Primers rt-pvmdrF, rt-pvmdrR, rt-pvaldoF and rt-pvaldoR, listed in Table 1, were used to amplify fragments of the pvmdr1 or the aldolase genes respectively.

Two plasmids, containing cloned fragments of the P. vivax aldolase gene and either one or two copies of the pvmdr1 fragment, were constructed de-novo and used as positive controls in every experiment. PCR reactions were performed in triplicates or quadruplets and contained 1× AB gene ABsolute™ QPCR SYBR® Green Mix (Cat. N AB-1166/a), 100 nM of ROX dye (passive reference dye), 1 µl of DNA template and 75 nM of each primer in a final volume of 25 µl. Cycling conditions were: 95°C for 15 min; followed by 40 cycles of 95°C 30 sec, 60°C 1 min and 72°C 30 sec. Fluorescence data was collected at the end of the annealing and extension steps 3 times at each and averaged. Following the amplification cycles, a melting curve analysis was performed to confirm that the correct products were synthesised. The text report, containing the threshold cycle (Ct) values for every well was exported into the Excel program (Microsoft©)and analysed.

The assay was optimised to achieve equal amplification efficiencies for the pvmdr1 and aldolase gene fragments within the range of DNA concentrations from 100 ng/µl to 10 pg/µl, thus the ΔΔCt method (Applied Biosystems User Bulletin N2 (P/N 4303859B) could be used and the pfmdr1 copy number (N) was calculated as follows: N = 2ΔΔCt±SD, where ΔΔCt = (Ct-Ct )-(Ct cal-Ct cal). The Ct and Ct used above are threshold cycle values for the pvmdr1 and aldolase gene respectively, whereas Ctcal is an average difference between Ctaldo and Ctpvmdr1 obtained for the positive control containing a single copy of pvmdr1 and aldolase gene fragments. The SD is a standard deviation calculated as follows: SD = √(S2 +S2 +S2 ) where S and S are the standard deviations from the average Ct calculated for 3 or 4 replicates in the pvmdr1 and pvaldo amplifications and S is an average standard deviation of the ΔCt values for the calibrator. Assessment of copy number was repeated at least twice for all isolates and the repeatability coefficient determined as 0.30 (viz 95% of repeated estimates of pvmdr1 copy number were within 0.15 of the first).

Data and sequence analysis

Analysis was performed using SPSS vs 14 for Windows (SPSS Inc, Chicago, Illinois, USA). The Mann-Whitney U test and Wilcoxon Signed-Rank test method was used for nonparametric comparisons, and Student's t-test (paired and unpaired) or one-way analysis of variance for parametric comparisons. Proportions were examined using χ2 with Yates' correction or by Fisher's exact test.

A linear regression analysis was used to determine the relationship between the log-transformed chloroquine in vitro susceptibility and country or genotype after correcting for duration of assay and initial percentage of parasites at ring stages, both previously shown to be confounding factors for in vitro susceptibility (unpublished data). In Papua, Indonesia the recurrence rate of P. vivax by day 28 following chloroquine monotherapy in 2004 was 65% [10]. We therefore defined, a priori, the clinically appropriate cut off for the IC50 as the 35th percentile of isolates from this region.

Sequences were aligned using the Gap4 program, version 4.10 freely available from http://www.mrc-lmb.cam.ac.uk/pubseq/manual/gap4_windows_2.html).To investigate the relatedness of the sequences a Clustal C program was used to create the phylogenetic trees for pvama1, and synonymous SNPs in pvmdr1 and pvcrt-o (ANGIS, http://www.angis.org.au). Unique DNA sequences described in this paper have been deposited in the GenBank under the accession No. EF458622 to EF458625.

Results

In vitro chloroquine susceptibility of Indonesian and Thai Isolates

Between April 2003 and December 2006, 247 isolates were assayed for in vitro susceptibility of which acceptable chloroquine susceptibility data could be derived in 226 (91%). Further analysis was restricted to these isolates (145 from Indonesia and 81 from Thailand); see figure 1. In total 51% (74/145) of Indonesian isolates began the assay with more than 40% ring stages prior to culture, compared to 81% (65/81) of isolates from Thailand; p<0.001. The time to reach 40% schizonts, and thus the duration of the assay, was significantly shorter in Indonesian isolates (26 hours [Range: 22–48]) compared to 36 hours [Range: 21–48] for isolates from Thailand; p<0.001.

Figure 1

The selection of samples analysed.

The geometric mean chloroquine IC50 for P. vivax isolates from Indonesia was 312 nM [95%Confidence Intervals CI: 237–411 nM] significantly higher than that for Thai isolates (46.8 nM [95%CI: 34.7–63.1 nM]); p<0.001. After ranking the 226 P. vivax isolates in order of decreasing chloroquine susceptibility, a continuous non-linear curve of chloroquine IC50 was observed (Figure 2). The 35th percentile for chloroquine IC50 in Indonesian isolates was 220 nM. Using this as an a priori cut-off for clinically relevant chloroquine resistance, 13.6% (11/81) of Thai isolates were classified as resistant in vitro.

Figure 2

Distribution of isolate chloroquine IC50.

81 Thai (closed diamonds) and 141 Indonesian (open diamonds) P. vivax isolates ranked in order of increasing chloroquine IC50.

The chloroquine IC50 values were negatively correlated with both the duration of assay and the percentage of parasites at ring stage prior to culture (rs = −0.576 and rs = −0.496 respectively; p<0.001). However, the difference in chloroquine IC50 between countries remained after controlling for these factors independently (Table 2) and in a multivariate comparison; p<0.001.

Table 2

In vitro chloroquine sensitivity (nM) of isolates from Thailand and Indonesia.

		N	Geometric Mean IC50	95% Confidence Intervals	Range	P
ALL	Indonesia	145	312	237–411	4.6–5637	P<0.001
	Thailand	81	46.8	34.7–63.1	6.7–2231
>30 Hours Duration	Indonesia	48	113	67.9–188	4.6–3024	P<0.001
	Thailand	69	33.2	26.0–42.5	6.7–430
>40% parasites at ring stage prior to culture	Indonesia	74	208	139–312	4.6–3506	P<0.001
	Thailand	65	33.7	25.8–44.1	6.7–1264

Values given overall and after selecting cultures with greater than 30 hour duration of assay or a starting with more 40% of parasites at ring stage. Criteria for duration of assay and percentage of rings in initial culture taken from Tasanor et al 2002 [19]

In total 16 Indonesian isolates were assayed for chloroquine with and without 0.9 µM verapamil. The median IC50 was not significantly different: 258 nM [Interquartile Range: 69–950 nM] for chloroquine plus verapamil compared to 157 nM [IQR: 23–332 nM] in the chloroquine alone assay; p = 0.56.

Polymorphisms in pvmdr1, pvcrt-o and pvama1 in core samples

Using Sal1 as the reference strain, sequence analysis of 32 core isolates revealed single nucleotide polymorphisms (SNP) at 5 loci of pvmdr1, two non-synonymous mutations resulting in amino acid changes at Y976F and L1076F (Table 3) and three synonymous SNPs (at codons 493, 908 and 1396). In these core isolates the Y976F mutation was significantly more prevalent in Indonesian isolates (96%, 24/25) compared to Thai isolates (43%, 3/7); p = 0.004.

Table 3

Mutations in pvmdr1 and pvcrt-o among 32 core Plasmodium vivax isolates from Indonesian and Thailand and the reference strains SAL1.

Sample	Origin	Genotype groups				Polymorphisms in pvmdr1 and pvcrt-0
						pvmdr1			pvcrt-0 exon
		Msp1 LP	Msp3 RFLP	AMA1 Sequence	Combined Genotype	Copy number	Y976F SNP	F1076L SNP	K10 Insert	I43M SNP
SAL 1	Central America	a	a	a	a	1	Y	F	-	I
ANV20	Indonesia	b	a	b	b	1	F	L	-	I
VI21	Thailand	a	b	c	c	1	Y	L	K	I
VI20	Thailand	a	a	d	d	1	F	L	K	I
VI32	Thailand	a	a	d	d	2	Y	L	K	I
VI5	Thailand	a	a	c	e	2	Y	L	K	I
VRP21	Indonesia	b	a	e	f	1	F	L	-	I
VI1	Thailand	a	a	f	g	1	F	L	-	I
PV14	Thailand	b	c	d	h	1	F	L	K	I
VI23	Thailand	a	a	g	i	2	Y	L	K	I
VP63	Indonesia	b	a	h	j	1	F	L	-	I
FC1010	Indonesia	b	a	i	k	1	F	L	-	I
UVT27	Indonesia	b	a	i	k	1	F	L	-	I
FC1232	Indonesia	b	a	c	l	1	F	L	-	I
VP56	Indonesia	b	a	c	l	1	F	L	-	I
FC1108	Indonesia	b	a	c	l	1	F	L	-	I
VRP23	Indonesia	b	a	c	l	1	F	L	-	I
VP59	Indonesia	a	a	c	m	1	F	L	-	I
FC1158	Indonesia	a	a	c	m	1	F	L	-	I
ANV15	Indonesia	b	d	c	n	1	F	L	-	I
FC1248	Indonesia	b	d	c	n	1	F	L	-	I
VRP20	Indonesia	a	c	j	o	1	F	L	-	I
UVT22	Indonesia	a	a	i	p	1	Y	L	-	I
ANV16	Indonesia	c	a	i	q	1	F	L	-	I
FC1083	Indonesia	a	c	k	r	1	F	L	-	I
UVT44	Indonesia	a	a	b	s	1	F	L	-	I
ANV18	Indonesia	c	a	i	t	1	F	L	-	I
ANV12	Indonesia	c	d	l	u	1	F	L	-	I
UVT70	Indonesia	c	a	m	v	1	F	L	-	I
FC269	Indonesia	b	d	n	w	1	F	L	-	I
FC1290	Indonesia	a	a	o	x	1	F	L	-	I
FC10122	Indonesia	b	a	l	y	1	F	L	-	I

Isolates are grouped and ordered alphabetically according to the combined pvmsp1, pvmsp3 and ama1 genotype. SNP positions and corresponding amino acid changes relative to the SAL1 reference strain are in bold.

One insert and one non-synonymous SNP were found in pvcrt-o exons, the most prevalent of which was the insertion of the trinucleotide AAG, coding for amino acid Lysine (K) at amino acid position 10 in the first exon (see table 3). The insert was found in 86% (6/7) of Thai isolates and 0% (0/25) Indonesian isolates; p<0.001.

Analysis of three loci (pvama1, pvmsp1 and pvmsp3) revealed 26 combined haplotypes in the 32 core isolates indicating that the isolates are of a diverse genetic background. Clustal-c analysis of partial pvama1 sequences from 89 isolates (Figure 3) showed no evidence that sequence diversity was less among isolates from within each field site compared to between locations.

Figure 3

Sequence relatedness among P. vivax isolates from different locations according to clustal-c analysis of pvama1 sequence.

Stars indicate Thai isolates.

P. vivax chloroquine susceptibility and pvmdr1 and pvcrt-o polymorphisms

Polymorphisms were assessed for the pvmdr1 SNP at codon 976 in an additional 165 isolates with in vitro susceptibility data and for the pvcrt-o insertion at amino acid position 10 in an additional 55 isolates (Figure 1). After combining these with the core isolates the Y976F allele was found in 96.1% (123/128) of Indonesian isolates compared to 25% (17/69) of Thai isolates (p<0.001). Overall, the geometric mean chloroquine IC50 in isolates with the Y976F mutation was 283 nM [95%CI: 211–379], significantly higher than that in isolates with the wild type allele (geometric mean = 44.5 nM [95%CI: 31.3–63.4]; p< 0.001). In Thailand, isolates with the Y976F mutation had a mean IC50 of 65.6 nM [95%CI: 29.9–144] compared to 39.0 nM [95%CI: 27.8–54.8] in those with the wild type allele (p = 0.008, after controlling for assay duration and percentage of rings pre incubation). The trend was similar in Indonesian isolates, however the proportion of Y976F mutation almost reached fixation and thus prohibited analysis of the correlation between the polymorphism and the phenotype (see figure 4).

Figure 4

Association between P. vivax chloroquine IC50 and Y976F mutations in pvmdr1.

The solid black horizontal lines show the geometric mean IC50 (nM) of the P. vivax population samples.

The pvcrt-o AAG insertion occurred in 76% (31/41) of Thai isolates, but only 2.2% (1/46) of the Indonesian isolates (p<0.001). Overall the pvcrt-o AAG insertion was associated with a significant reduction in chloroquine IC50, (geometric mean 47.6 nM [95%CI: 29.7–76.1] vs 261 nM [95%CI: 172–396]; p<0.001). After stratifying by geographical location, the AAG insertion of pvcrt-o was not linked to the pvmdr1 Y976F mutation and was not significantly associated with reduced chloroquine IC50 .

P. vivax chloroquine susceptibility and pvmdr1 copy number

The pvmdr1 copy number was successfully quantified in 86% (170/198) of isolates tested. In total 23% (15/66) of isolates from Thailand had an increased pvmdr1 copy number (13 with 2 copies and two with 3 copies) and none (0/104) from Indonesia; p<0.001. In Thailand all (15/15) of the isolates with increased copy number were wild type at 976, compared to 67% (34/51) of those with a single copy number of pvmdr1 (p = 0.007). Although isolates with increased pvmdr1 copy number had significantly lower chloroquine IC50s (geometric mean = 39.6 [95%CI: 24.5–64.1]) compared to isolates with single copies of pvmdr1 (geometric mean = 184 [95%CI: 137–247]; p<0.001); this was not apparent after stratification by country.

Discussion

In clinical studies, P. vivax remains predominantly sensitive to chloroquine in Thailand, whereas in Papua, Indonesia high grade clinical resistance is already established [10], [25], [26]. In 2004 a chemotherapeutic study at the Papuan field site demonstrated that 65% of patients failed treatment within 28 days of chloroquine monotherapy, 16% of whom had early high grade failures. Treatment guidelines were changed accordingly to an Artemisinin combination therapy for both P. falciparum and P. vivax [31], precluding further clinical studies on the use of chloroquine monotherapy in this region. In the present study we have continued our analysis of chloroquine resistance P. vivax using an identical in vitro methodology in both Indonesia and Thailand and correlating our results with the known data on the clinical efficacy of chloroquine in these regions. The Indonesian P. vivax isolates tested had a significantly higher median chloroquine IC50 and a higher proportion above the resistance threshold compared with that of Thai isolates.

The determination of chloroquine susceptibility in P. vivax using the schizont maturation method is more complicated than the same method in P. falciparum, due to the asynchrony of the vivax parasites and possible differential responses to the drug by parasites at different development stages. Patient samples with higher percentage of trophozoites and late rings require less incubation time to reach maturation (unpublished data). The decreased susceptibility to chloroquine in these samples provides a plausible explanation for our observation of the negative correlations between IC50 and culture duration as well as with percentage of rings at the start of culture. Although differentiating between these possibilities is difficult, we attempted to control for these confounding factors by stratifying our results according to culture duration and the percentage of rings at the start of culture; the differences in IC50 between isolates from Indonesia and Thailand remained.

The in vitro cut-off defining clinically relevant chloroquine resistance has yet to be defined. Using the clinical failure rate (65%) observed in the same area, we defined this from the 35th percentile as 220 nM, almost double the 100nM cut-off value for chloroquine resistance in P. falciparum. However clinical failures may have included some relapses that occur within the 28 day follow up period, and the true rate of recrudescence maybe lower. Hence this threshold is likely to be the minimum value associated with resistance.

The chloroquine IC50 of Thai isolates were significantly lower than the Indonesian isolates (Geometric mean = 46.8 vs 312 nM), although the difference was less after controlling for the duration of assay (33.2 vs 113 nM) or initial stage of parasite prior to culture (33.7 vs 208 nM). Interestingly 13.6% (11/81) of Thai isolates had a chloroquine IC50 over 220nM. Although clinical studies in Thailand in the 1990s have repeatedly demonstrated the continued efficacy of chloroquine monotherapy for P. vivax [26], [27], our in vitro results raise the possibility that clinically relevant chloroquine resistance may now be present at low prevalence along the western border of Thailand. This is corroborated by a recent clinical study from the Thai-Myanmar border demonstrating 34% P. vivax recurrence rates within 28 days of chloroquine monotherapy [32].

The correlation between in vitro susceptibility and clinical efficacy at our two study sites validates our in vitro susceptibility test and suggests that the adapted schizont maturation method may be usefully applied to investigate the emergence of drug resistance in P. vivax in other locations. Furthermore, the ability to define parasite susceptibility free from the confounding factors of host and environment provides a useful framework from which to investigate putative molecular markers of drug resistance. We used our carefully defined sample set to test for associations between the in vitro response to chloroquine and polymorphisms of the orthologues of two genes (pvmdr1 and pvcrt-o) known to be important determinants of chloroquine resistance in P. falciparum. Although previous studies have not established a link between these genes and chloroquine resistant P. vivax , these generally used a relatively small number of clinical isolates in which the phenotypic definition was possibly confounded by patient immunity, re-infection and relapses [22], [23]. Brega et al identified the pvmdr1 Y976F and 1076 mutation in a small number of Thai and Indonesian isolates, although in vitro and clinical correlates were not presented [33].

In the present study we found two polymorphisms which were correlated with in vitro chloroquine susceptibility: the pvmdr1 Y976F mutation and an insertion in the 1st exon (amino acid position 10) of pvcrt-o. Overall both polymorphisms were associated with a significant increase in chloroquine IC50. In Papua, Indonesia, where the Y976F mutation has reached fixation and the AAG insertion was almost absent it was not possible to test the relevance of these markers. However in Thailand, the Y976F mutation was present in 25% (17/69) of isolates and associated with 1.7 fold increase in IC50 to chloroquine.

To rule out the possibility that the polymorphisms were related to geographical isolation of the samples, we performed phylogeny analyses to compare the samples from two locations on pvama1, a marker unrelated to chloroquine pressure. The results did not show clustering of samples with location. In addition, we analysed pvmdr1 sequence including all synonymous changes in pvmdr1 which are presumed not to be selected by drug pressure. Again we did not see clustering of the samples with location. These analyses suggest that the Y976F is unlikely to be geographically associated with the Papua location per se, and provide further evidence for its selection by chloroquine selective pressure.

Notably a small number of isolates with high IC50 values were observed from both sites in the absence of the 976 mutation, and vice versa, suggesting that other major molecular determinants are likely to be involved. However a role of pvmdr1 in modulating chloroquine susceptibility is supported by the almost ubiquitous selection of the Y976F allele in Papua, where high grade chloroquine resistance is known to predominate.

Gene amplification of the pfmdr1 gene has been shown to be a major determinant of multidrug resistance in P. falciparum. Furthermore on the Thai-Myanmar border widespread deployment of mefloquine has been associated with high prevalence of P. falciparum isolates with increased pfmdr1 copy number and an associated decrease in susceptibility to mefloquine, quinine, lumefantrine, halofantrine and the artemisinin derivatives in P. falciparum [34]. In this study we report that amplification of pvmdr1 copy number occurs in P. vivax in Thailand, but not Papua, where mefloquine has not been used. Our data raise the prospect of similar molecular mechanisms of multi drug resistant phenotype as found in P. falciparum, although further work is needed to confirm this.

In conclusion, using an in vitro susceptibility assay, we have been able to define a spectrum of chloroquine susceptibility in P. vivax and discriminate between populations with differing levels of clinical efficacy following chloroquine monotherapy. Although the molecular mechanism underlying chloroquine resistance P. vivax may involve multigenic loci, the pvmdr1 polymorphism at Y976F may provide a useful tool to monitor the emergence of chloroquine resistance.