Clinical study

The National Malaria Control Programme of Cambodia (CNM) conducted three therapeutic efficacy studies at Sampovloun Referral Hospital, Battambang Province (western Cambodia): 1. artemether-lumefantrine between June and August 2002 (AL2002, N = 55), 2. artemether-lumefantrine between October 2003 and March 2004 (AL2003, N = 80) [8] and 3. artesunate-mefloquine between October 2003 and March 2004 (AM2003, N = 55) [9]. One therapeutic efficacy study of AM was conducted at Veal Veng Referral Hospital, Pursat Province (western Cambodia) between July and August 2004 (AM2004, N = 85) [9]. The studies followed the WHO standard protocol for the assessment of the efficacy of anti-malarial drugs [10]. Briefly, all patients over six years of age and weight > 16 kg presenting with fever, defined as axillary temperature ≥ 37.5°C or history of fever during 24 h prior to consultation, and a positive smear with Plasmodium falciparum mono-infection, with parasite density of 1,000–150,000 parasites/μl, were included in the studies. Informed consent was given by the patient or by a parent/guardian for children. Enrolled patients had to stay in hospital until blood smear became negative and had to commit to completing the 28-day follow-up. Exclusion criteria are: one or more of the general danger signs or any sign of severe and complicated malaria, pregnancy, febrile diseases other than malaria and severe malnutrition, and known hypersensitivity or contraindication to the study drugs. A total dose of 12 mg/kg artesunate and 25 mg/kg mefloquine were given over 3 days (AM group). The daily 4 mg/kg of artesunate were divided into two equal doses, one in the morning and one in the evening on day 0, and 4 mg/kg single dose on day 1 and 2 with a maximum adult dose of 600 mg in total. The 25 mg/kg of mefloquine were also divided into two equal doses in the morning and in the evening on day 0. The maximum dose of mefloquine was limited to 1,500 mg for adults as the side effects are common beyond this dose [9]). In the artemether-lumefantrine study group (AL group), subjects were treated with 20 mg/kg artemether and 120 mg/kg lumefantrine the first day at 0 and 8 hours and twice daily (morning and late afternoon) on the following two days (Coartem^®, Novartis, Switzerland), i.e. 8–12 h apart (the total dose of 24 tablets for an adult as recommended by the manufacturer). In the AL2003 study, each dose of artemether-lumefantrine was provided with 250 ml milk and 5 pieces of coconut biscuit [8]. Patients were checked by malaria smear for the presence of parasites on study days 1, 2, 3, 7, 14, 21 and 28. If fever occurred at any time between the scheduled study days, a malaria smear was also done. People who did not present on their own were actively sought for. The therapeutic response was classified as early treatment failure (ETF), late clinical failure (LCF), late parasitological failure (LPF) and adequate clinical and parasitological response (ACPR) according to WHO protocols.

Site and blood sampling

Blood samples (5 ml) from patients were collected before treatment in EDTA Vacutainers, and were transported to the Institut Pasteur in Cambodia at 4°C within 48 hours of collection for in vitro assays. An aliquot of each isolate was frozen at -80°C for molecular analysis. If parasitaemia recurred, an additional blood sample was collected onto 3 M Whatman filter paper and transported to the laboratory at room temperature, then kept at -20°C until DNA extraction.

Preparation of drugs and plates

Quinine hydrochloride was obtained from Sigma (Germany). Mefloquine, chloroquine diphosphate, artemether and artesunate were obtained from WHO/TDR Drug Discovery Research (Geneva). Lumefantrine was obtained from Novartis Pharma (Vietnam). Stock solutions of chloroquine diphosphate were prepared in water (Biosedra, France) and stock solutions of quinine hydrochloride, mefloquine, lumefantrine, artemether and artesunate were prepared in ethanol. Further two-fold serial dilutions were prepared in distilled water (Biosedra, France). The final concentrations ranged from 0.05 to 51.2 nM for artesunate, 1 to 1024 nM for mefloquine, 5 to 5120 nM for chloroquine, 6.2 to 6400 nM for quinine, 0.1953 to 200 nM for lumefantrine and 0.097 to 100 nM for artemether. Each concentration was used to coat two wells of a Falcon 96-well, flat-bottom plate (ATGC, France). Forty microliters of drug solutions prepared as described above was added to each well and plates were then dried at room temperature in a sterile cabinet. The pre-dosed plates were finally kept frozen at -20°C until use. Their suitability for in vitro testing was regularly monitored using reference strains maintained in continuous culture (at Institut Pasteur in Cambodia) and presenting known responses to the various drugs tested.

In vitro assay

The in vitro drug sensitivity of the Cambodian isolates was assessed by use of a classical isotopic 48 h test [11]. Briefly, fresh blood samples were washed three times with RPMI 1640 medium (GibcoTM, Invitrogen Corporation, France) by centrifugation (800 g, 10 min). The parasites were then tested directly without culture adaptation. The infected erythrocytes (1.5% haematocrit, 0.1–1% parasitaemia) were suspended in complete RPMI medium supplemented with 10% AB+ human serum that was heat inactivated for 30 min at 56°C (Biomedia, France) and buffered with 25 mM/l Hepes and 25 mM/l NaHCO3. The mixture was distributed (200 μl per well) into the 96-well test plates that had been pre-coated with anti-malarial agents. Each plate included two drug-free control wells and one control well without parasites. The culture plates were incubated for 48 h at 37°C in a 5%CO2 atmosphere. [³H]-hypoxanthine (0.5 μCi/well; Amersham Biosciences, France) was used to assess parasite growth. Each isolate was tested once in duplicate in the microplates with serial dilutions of drugs. Drug response was quantified by monitoring [³H]hypoxanthine uptake in a Wallac MicroBeta Trilux counter (Perkin-Elmer, France). At the end of the incubation period, the plates were frozen at -20°C and thawed to lyse the cells. After collection on glassfiber filter paper using a cell harvester, the amount of [³H]-hypoxanthine incorporated into the parasites nucleoprotein was determined. The results of in vitro assay are expressed as the 50% inhibitory concentration (IC₅₀), defined as the concentration at which 50% of the incorporation of [³H]-hypoxanthine was inhibited, as compared with in the drug-free control wells. Parasite growth was measured by using a log probit approximation to determine the IC₅₀values.

DNA extraction

Parasite DNA was extracted from frozen blood aliquots (200 μl) using High Pure PCR template preparation kit (Roche^®) following the manufacturers protocol. Blood stored on filter paper was extracted using QIAamp DNA Mini Kit (Qiagen) [12].

Genetic polymorphism

Paired samples taken at enrollment and at recurrence of parasitaemia were used to distinguish between recrudescence and re-infection. The number of variants in three polymorphic genes (msp1, msp2 and glurp) was determined as described previously [13, 14]. If the sample taken at recurrence contained either a subset of, or the same variants as the enrollment specimen, the infection was classified as recrudescence. If not, the infection was classified as a re-infection. The amplification of a single fragment at these three loci indicated that the parasite population was mono-infected (single genotype). Detection of two or more PCR bands, at one or more loci, indicated that the isolate contained multiple infections.

PCR amplification and direct sequencing were used to determine the single nucleotide polymorphism (SNP) at four pfmdr1 sites (positions 86, 184, 1042, and 1246) at enrollment date. The primer sequences used for the analysis of the N86Y and Y184F SNPs were: first amplification, 5'-TTA AAT GTT TAC CTG CAC AAC ATA GAA AAT T-3' (forward) and 5'-CTC CAC AAT AAC TTG CAA CAG TTC TTA-3' (reverse); nested amplification, 5'-TGT ATG TGC TGT ATT ATC AG GA-3' (forward) and 5'-CTC TTC TAT AAT GGA CAT GGT A-3' (reverse). The primer sequences used for the analysis of the N1042D and D1246Y SNPs were: first amplification, 5'-AAT TTG ATA GAA AAA GCT ATT GAT TAT AA-3' (forward) and 5'-TAT TTG GTA ATG ATT CGA TAA ATT CAT C-3' (reverse); nested amplification, 5'-GAA TTA TTG TAA ATG CAG CTT TA-3' (forward) and 5'-GCA GCA AAC TTA CTA ACA CG-3' (reverse). PCR products were sent to Genopole, Institut Pasteur Paris (IPP) for sequencing. Sequencing reactions were performed from both ends using ABI Prism BigDye Terminator Cycle Sequencing-Ready Reaction kits, and were run on a 3700 Genetic Analyzer (Applied Biosystems) as described previously [15]. Sequencing results were assembled, individually visually examined, and aligned using ABI Prism^® SeqScape^® software version 2.5 (Applied Biosystems).

Real time PCR

Real time PCR was performed with an ABI PRISM 7000 sequence detection system (Applied Biosystems, Foster City, CA). Pfmdr1 copy number was assessed for all enrollment samples using a previous assay [5]. This assay consisted of combining both the primers and a FAM-TAMRA probe specific to a conserved region of pfmdr1, and the primers and a VIC-TAMRA probe specific for β-tubulin, all in one reaction. The primers were synthesized by Sigma-Proligo (Proligo Singapore Pty Ltd) and the probes were synthesized by Applied Biosystems UK. All samples were run in duplicate in 25 μl reactions. The reagents used for each unknown sample or standard were 1X Abgene QPCR ROX Master mix (2X, Abgene), 300 nM of each forward and reverse pfmdr1 primer, 150 nM of pfmdr1 probe, 100 nM of each forward and reverse β-tubulin primer and β-tubulin probe, 2 μl of template DNA and sterile water (Biosedra, Fresenius Kabi, France). DNA from strains 3D7 and Dd2 were included on each plate. The 3D7 DNA and Dd2 DNA were acquired from MR4 [16]. The reaction mixtures were prepared at 4°C in a 96-well optical reaction plate (Applied Biosystems) covered with optical adhesive covers (Applied Biosystems). The thermal cycling conditions were 50°C for 2 min, 95°C for 15 min and then 50 cycles of 95°C for 15 s and 60°C for 1 min. The threshold was placed to optimize the threshold cycle (C _T ) value for the first standardization reaction, and all subsequent C _T values were obtained by using the same threshold value. Pfmdr1 copy number was calculated according to the following formula: copy number = (E _βtubulin )^{CT(βtubulin)}/(E_{pfmdr 1})^{CT(pfmdr 1)}. The Efficiency (E) of the β-tubulin was assumed to be 2. Each plate included 1 well of 3D7 DNA (which has a copy number of 1) and 1 well Dd2 (which has 4 copies). The efficiency of the pfmdr1 was back calculated using CT's for 3D7 and Dd2 in the previous formula [17, 18].

Statistical analysis

All statistical analyses were performed using STATA (StataSE 8, Stata Corporation, College Station, Texas). The in vitro activity of anti-malarials is expressed as the geometric means of the IC₅₀s for all isolates. Drug concentrations were transformed into logarithms. Linear regression analysis was used to assess the relationship between the prevalence of parasite and pfmdr1 copy number. The association between pfmdr1 copy number and treatment failure or IC₅₀ of each anti-malarial drug was tested by Wilcoxon rank-sum (Mann-Whitney) test. The relationship between molecular changes in pfmdr1 and treatment failure was estimated using survival analysis. Patients who had been lost to follow-up or who presented again with re-infection were included but censored at their last visit or when they were diagnosed with a re-infection. All variables were assessed for the proportional hazard assumption. The main outcome was time to recrudescence. Previously identified risks for treatment failure with mefloquine and artesunate, and with lumefantrine and artemether, were included in Cox's regression analysis when assessing the association of pfmdr1 polymorphisms and treatment response. For all statistical tests, the significance level was set at p = 0.05.

Ethical approval

This study was approved by the National Ethics Committee for Health Research, Ministry of Health, Cambodia, and the Technical Review Group of WHO/WPRO.

Genetic diversity

At enrollment, 187 of 227 (82%) patient blood samples contained only one parasite population, with a single genotype at the msp1, msp2 and glurp loci, whereas 40 (18%) had multiple infections containing two (38/227 or 17%) or three (2/227 or 1%) distinct genotypes. The number of parasite populations in three isolates remained undetermined due to the poor quality of typing with three markers. The mean number of clones per isolate was 1.18 and standard deviation was 0.41.

Pfmdr1 polymorphism

One hundred and sixty two patient blood samples were randomly selected for pfmdr1 sequencing at position of pfmdr1-86, 184, 1042 and 1246. One hundred and fifty six patient blood samples (96.3%) gave interpretable results (Figure 1). Out of these, 18 were unsuccessfully sequenced at pfmdr1-86, 17 were unsuccessful at pfmdr1-184, 1 was unsuccessful at pfmdr1-1042 and 1 was unsuccessful at pfmdr1-1246. No Pfmdr1-86-Tyr and pfmdr1-1246-Tyr mutants were observed. The results showed 125 (91%) pfmdr1-184-Phe mutant isolates and 11 (8%) pfmdr1-1042-Asp mutant isolates. No Pfmdr1-86-Tyr and pfmdr1-1246-Tyr mutants were observed.

Pfmdr1 copy number

The pfmdr1 copy number was determined for 115 isolates in the AM group and for 109 isolates in the AL group, all collected at the day of enrollment (Figure 1). An overall mean copy number of 1.55 with a range from 0.56 to 4.91. Control DNA gave reproducible results as Dd2 had a mean copy number of 4.05 with a standard deviation of 0.47, and 3D7 control group had a mean copy number of 1.16 with a standard deviation of 0.14.

When rounded to the nearest integer, 182 (81%) isolates had a pfmdr1 copy number of one, 28 (12%) had two copies, 8 (4%) had three copies, and 6 (3%) had four copies. Overall, copy number was not related to baseline parasitaemia, age or gender (results not shown). No pfmdr1 polymorphisms were found to be associated with increased copy number. Genetic polymorphisms at the msp1, msp2 and glurp loci showed no indication of clonal population structure in high copy number isolates.

In vitro drug sensitivity

In vitro sensitivities to artesunate, mefloquine, quinine and chloroquine were determined for 226 isolates. Microscopic examination showed that the parasite loads were between 0.02 and 14.4%, with a mean parasitaemia of 1.27%, CI 95% (1.02 – 1.52). Consistent with previous findings indicating that parasitaemia is critical for adequate in vitro testing, interpretable data for at least one of the drugs tested were obtained only for about half of the isolates (43.8%) (Figure 1). Other samples with low parasite counts (< 0.2%, n = 60) resulted in poor quality data and were not included in the study. In vitro sensitivity for artemether and lumefantrine were only tested on the AL group with 40 and 50 interpretable results, respectively (Figure 1). The geometric mean IC₅₀ values for mefloquine, artesunate, quinine, chloroquine, lumefantrine and for artemether were 20.2 nM/l (n = 92, CI95% 16.1 – 25.3 nM, range 2.3 – 132.9), 1.3 nM/l (n = 99, CI95% 1.1 – 1.5 nM, range 0.3 – 11.1), 131.6 nM/l (n = 90, CI95% 115 – 150.7 nM, range 19.94 – 615.9), 196.1 nM/l (n = 90, CI95% 176.1 – 218.3 nM, range 54.1 – 663.3), 19.27 nM/l (n = 50, CI95% 15.8 – 23.4 nM, range 3.6 – 79.2) and 1.00 nM/l (n = 40, CI95% 0.6 – 1.6 nM, range 0.2 – 42.2), respectively.

In vivo study

Pfmdr1 copy number was lower in the ACPR group compared to the LTF group (n = 112, p < 0.001) for AM but not for AL treatment (n = 96, p = 0.257) (Figure 2). The proportional hazards assumption was violated when pfmdr1 copy number was classified as a single copy versus multiple copies. Therefore, to allow statistical analysis of copy number data, two groups were build: low (< 3 copies) and high (≥ 3 copies). A pfmdr1 copy number that is ≥ 2.5 is rounded up to 3. This way of coding copy number met the proportional hazards assumption.

In the AM group (N = 115), crude hazard ratio for increased copy number (≥ 3 copies compared to < 3 copies) was 7.80 [95%CI: 2.09–29.10], p = 0.002. After controlling for baseline parasitaemia and gender in a Cox's regression model, the hazard ratio was 12.28 [95%CI: 3.19–47.30], p < 0.001. There was no association between increased copy number and treatment failure in the AL group [HR = 1.03 [95%CI: 0.24–4.46], p = 0.962] (Figure 3).