|Home | About | Journals | Submit | Contact Us | Français|
The aim of this study was to provide the first comprehensive spatiotemporal picture of Plasmodium falciparum resistance in various geographic areas in Madagascar. Additional data about the antimalarial resistance in the neighboring islands of the Comoros archipelago were also collected. We assessed the prevalence of pfcrt, pfmdr-1, pfdhfr, and pfdhps mutations and the pfmdr-1 gene copy number in 1,596 P. falciparum isolates collected in 26 health centers (20 in Madagascar and 6 in the Comoros Islands) from 2006 to 2008. The in vitro responses to a panel of drugs by 373 of the parasite isolates were determined. The results showed (i) unusual profiles of chloroquine susceptibility in Madagascar, (ii) a rapid rise in the frequency of parasites with both the pfdhfr and the pfdhps mutations, (iii) the alarming emergence of the single pfdhfr 164L genotype, and (iv) the progressive loss of the most susceptible isolates to artemisinin derivatives. In the context of the implementation of the new national policy for the fight against malaria, continued surveillance for the detection of P. falciparum resistance in the future is required.
In recent decades, the emergence and subsequent spread of Plasmodium falciparum chloroquine (CQ)- and sulfadoxine-pyrimethamine (SP)-resistant parasites across areas where malaria is endemic have been a challenge to malaria control programs (41, 44). Substantial advances toward gaining an understanding of the genetic basis of antimalarial drug resistance have been made (14). Molecular evolutionary studies have concluded that the CQ-resistant P. falciparum chloroquine resistance transporter (pfcrt) and high-level pyrimethamine-resistant dihydrofolate (pfdhfr) alleles have emerged in a limited number of independent foci, from which they have rapidly spread in the local vicinity and have then invaded areas continent-wide and transferred between continents (1, 36). These lessons of the past have, first, stimulated changes in antimalarial treatment policies by introducing combinations of drugs that act on different targets and, second, resulted in the implementation of effective monitoring systems to detect as early as possible the emergence of resistant parasites on the basis of the assessment of the therapeutic efficacies of antimalarials (25, 46), determination of the decreased sensitivity of the parasites to drugs in vitro (4), and the detection of an increasing prevalence of molecular markers related to drug resistance (24).
According to data published from 2002 to 2006, the epidemiological features of P. falciparum CQ and SP resistance differ considerably between Madagascar and the Comoros Islands, two countries located close to each other in the southwestern Indian Ocean (43). In vitro CQ resistance was moderate in Madagascar (29, 33, 45), although the level of therapeutic efficacy was declining. During that time, the rate of CQ resistance was high in the Comoros Islands (22, 23, 30). Likewise, pyrimethamine resistance was absent in Madagascar (28, 32) but was present at high levels in the Comoros Islands (23). The most recent in vivo data obtained on the basis of the WHO 28-day follow-up protocol, conducted in 2006 and 2007 at multiple sites, have confirmed that resistance to all antimalarials except CQ in Madagascar remains rare. Indeed, the prevalence of the clinical failure of treatment with amodiaquine, SP and the combination artesunate and amodiaquine was <5%, while the rate of failure of treatment with CQ was 44% (19). However, the recent demonstration of the introduction of multidrug-resistant P. falciparum parasites into Madagascar from the Comoros Islands (18) and the emergence of the uncommon dihydrofolate reductase I164L genotype in P. falciparum parasites (17) suggest that the situation is currently changing in Madagascar.
In this context and in order to help with the rationalization of the malaria elimination policy recently launched by the Malagasy government (withdrawal of CQ in favor of the combination of artesunate plus amodiaquine as first-line treatment and SP usage for intermittent preventive treatment for pregnant women), a large-scale survey was designed and carried out between 2006 and 2008. The aim was to provide a comprehensive spatiotemporal picture of P. falciparum resistance in several geographic areas of Madagascar. We report here the prevalence of P. falciparum parasites harboring mutations correlated with resistance to some quinolines, namely, pfcrt and P. falciparum multidrug resistance gene 1 (pfmdr-1), or SP resistance (pfdhfr, pfdhps) or presenting an increased pfmdr-1 gene copy number, along with the in vitro responses of the parasites to a panel of drugs, including CQ, mefloquine (MF), amodiaquine, quinine (QU), and artemisinin derivatives. In addition, information related to the risk factors that contribute to the spread of antimalarial drug resistance, such as antimalarial resistance in the neighboring islands of the Comoros archipelago, drug pressure, and population movement in Madagascar, was collected (8, 42).
P. falciparum isolates were obtained from febrile patients seeking treatment for malaria at public health centers. Comorian isolates were collected from blood from finger pricks placed on filter paper at six different sites from May to June 2006, and those from Madagascar were collected from blood obtained by venipuncture at sites involved in the antimalarial drug resistance surveillance network from 2006 to 2008 (Fig. (Fig.1).1). Blood samples collected in EDTA tubes were transported to Antananarivo, Madagascar, at +4°C within 24 to 48 h of collection. Giemsa-stained blood smears were examined to check for monoinfection with P. falciparum, and the parasite density was determined. The samples were processed for in vitro assays, and aliquots were stored at −20°C before genomic DNA extraction.
In vitro assays were performed by using the classical isotopic 48-h test, as described previously (34). The final drug concentrations ranged from 12.5 to 1,600 nM for CQ, 2.5 to 320 nM for monodesethylamodiaquine (MDA), 25 to 3,200 nM for QU, 0.25 to 40 nM for dihydroartemisinin (DHA), and 3.1 to 400 nM for MF. For each drug tested, three control wells were drug free, and each concentration (nine for CQ and eight for each of the other drugs) was studied in duplicate or triplicate. Tests were considered interpretable if the level of [3H]hypoxanthine incorporation was >1,000 cpm in the drug-free wells. The validity and the reproducibility of the assays were controlled by using P. falciparum reference lines (3D7 Africa and FCM29 Cameroon). The results were expressed as the 50% inhibitory concentration (IC50), defined as the concentration at which 50% of the level of incorporation of [3H]hypoxanthine was inhibited compared to the level of incorporation in the control wells. The IC50 was calculated by using a probit/logit regression of the percentage of growth inhibition for each drug.
Parasite DNA was extracted from blood spots with Instagene matrix (Bio-Rad, Marnes la Coquette, France), according to the manufacturer's instructions, or directly from 100 μl infected blood by the phenol-chloroform method (27). The parasite species was confirmed by real-time PCR, as described by de Monbrison et al. (5). The primers and PCR conditions used are detailed in Table Table1.1. Sequencing reactions were carried out with a ABI Prism BigDye Terminator cycle sequencing ready reaction kit and were run on a model 3730 xl genetic analyzer (Applied Biosystems, Courtaboeuf, France). Electrophoregrams were visualized and analyzed with CEQ2000 genetic analysis system software (Beckman Coulter, Villepinte, France). The amino acid sequences were compared with the wild-type amino acid sequences (GenBank accession numbers, AF030694 for pfcrt, XM_001351751 for pfmdr-1, XM_001351443 for pfdhfr, and Z30654 for pfdhps). The presence of single nucleotide polymorphisms (SNPs) was confirmed by reading both the forward and the reverse strands. Parasites with mixed alleles (in which both wild-type and mutant alleles were present) were considered mutants for estimation of the prevalence of the SNPs. Haplotypes for drug resistance markers were reconstructed from the full sequence presenting an unambiguous single allele signal at all positions.
The P. falciparum pfmdr-1 copy number was measured by TaqMan real-time PCR (Rotor-Gene 6000; Corbett Research, Sydney, Australia) relative to the single copy of the ß-tubulin (used as a housekeeping gene), as described previously (26). All samples were run in duplicate in 25-μl reaction mixtures containing JumpStart Taq ReadyMix (Sigma, Taufkirchen, Germany), 300 nM each forward and reverse primer, 100 nM each probe, and 4 μl template DNA. Amplifications were performed under the following conditions: 94°C for 2 min, followed by 40 cycles of 94°C for 15 s, 58°C for 60 s, and 72°C for 60 s. For each run, the pfmdr-1 copy number was measured relative to the numbers in two standard calibrator parasite clonal lines, the 3D7 Africa line (which has one copy of pfmdr-1) and line Dd2 (which has three copies of pfmdr-1), by the ΔΔCT method (where CT is the threshold cycle) (15). Reference DNA clone W2 (which has three copies of pfmdr-1) was used as the quality control in each run.
The multiplicity of infection, defined as the highest number of alleles detected at either of the two loci, was estimated by using an allelic family-specific nested PCR (MAD20, K1, and RO33 for pfmsp-1 and 3D7 Africa and FC27 for pfmsp-2), as described previously (40). All PCR amplifications contained a positive control (genomic DNA from strains W2, HB3, and 3D7 Africa) and a negative control (no target DNA). Multilocus genotype analysis for drug resistance markers was performed with monoclonal isolates, i.e., isolates in which a single pfmsp-1 and/or pfmsp-2 allelic form was detected.
Between April and May 2006, cross-sectional surveys were carried out in Madagascar. Thirty households from each of the seven sites were randomly selected. Three randomly selected individuals in each household were interviewed about their recent travel and antimalarial drug consumption, as described by Gardella et al. (8). Data collected from the individuals and the household were secondarily aggregated to characterize each site.
EpiInfo software (version 6.0; CDC, Atlanta, GA) and MedCalc software (version 126.96.36.199; Mariakerke, Belgium) were used for data analysis. Categorical variables were compared by using the χ2 or Fisher's exact test, and continuous variables were compared by using the independent-sample t test. Data from the in vitro assays were analyzed after logarithmic transformation and were expressed as geometric means with 95% confidence intervals. A P value of <0.05 was considered statistically significant.
The study protocol was reviewed and approved by the Ethics Committee of the Ministry of Health of Madagascar (approval number 007/SANPF/2007; registration number ISRCTN36517335). Informed written consent was provided by all patients or their parents/guardians before inclusion in the study.
The exact sequence of each new mutant allele has been submitted to the GenBank database (accession numbers FJ941830 to FJ941841).
A 600-bp fragment (codon 6 to 206 of the pfdhfr gene) was sequenced from 1,195 clinical isolates (269 Comorian and 926 Malagasy isolates collected from 2006 to 2008). Seven novel mutations were detected in 30 samples (2.5%): two synonymous mutations (codon G26G, n = 2; codon L73L, n = 1) and five nonsynonymous mutations, either specific for Madagascar (G26R, n = 2; K27E, n = 1; K72I, n = 13; D91H, n = 1) or shared by both countries (N88S, 6 Comorian and 4 Malagasy isolates). The four major SNPs related to pyrimethamine resistance (N51I, C59R, S108N, I164L) were observed, with there being eight different alleles overall (Table (Table2).2). In 2006, four alleles (NCNI, ICNI, NRNI, and IRNI, where the boldface and underscore indicate amino acids conferring resistance; Table Table2)2) were more frequent in the Comoros Islands than in Madagascar (P < 0.05). The allele with quadruple mutations (IRNL), which conferred a high level of resistance to antifolates, was found at a low frequency (3/269 [1.1%]) in the Comoros Islands, whereas the single 164L mutation (NCSL) was observed in three isolates in Madagascar (3/292, 1.0%). The spatial-temporal distribution of the two most prevalent alleles (IRNI and NCSL) in 2006 to 2008 is given in Fig. Fig.2a2a.
The sequence of a 690-bp fragment (codons 419 to 649 of the pfdhps gene) was generated for 939 clinical isolates (269 Comorian and 670 Malagasy isolates collected in 2006 and 2007). Eight SNPs related to sulfadoxine resistance (S436A, S436C, S436F, A437G, K540E, A581G, A613S, A613T) were detected and were displayed by 10 alleles (Table (Table2).2). The 540E mutation was found only in the Comoros Islands, whereas the 581G, 613S, and 613T mutations were observed only in Madagascar. In 2006, the CAKAA allele was more frequent in the Comoros Islands, whereas the SGKAA allele was more frequent in Madagascar (P < 0.05). Although it was predominant mainly in the northwest in 2006 (11/54, 20.7%), the frequency of the SGKAA allele increased in 2007 in the central west (34/95 [35.8%] to 90/136 [66.2%]; P < 10−4) and the southwest (40/104 [38.5%] to 34/58 [58.6%]; P = 0.02), reaching a prevalence >50% in the western part of Madagascar.
Three fragments of 450 bp (from codon 37 in exon 2 to codon 125 in exon 3), 447 bp (from codon 121 in exon 3 to codon 211 in exon 4), and 850 bp (from codon 199 in exon 3 to codon 316 in exon 7) located in the pfcrt gene were sequenced from 959 clinical isolates (266 Comorian and 693 Malagasy isolates collected in 2006 and 2007). Four SNPs related to CQ resistance (N74I, M75E, K76T, A220S) and four alleles were observed. Among the isolates from Madagascar, the mutant-type IETA allele was very rare compared to its incidence among the isolates from the Comoros Islands (P < 0.05) (Table (Table2).2). Interestingly, the 76T mutation, which is key for CQ resistance, was observed in an allele with triple mutations (IETA) and an allele with quadruple mutations (IETS), while the 220S mutation, which acts in synergy with 76T for the establishment of CQ resistance, was observed as an allele with a single mutation (NMKS) as well as an allele with quadruple mutation (IETS).
Two fragments of 549 bp (from codons 23 to 206) and 938 bp (from codons 966 to 1278) were sequenced from 788 clinical isolates collected in 2006 and 2007 (271 from the Comoros Islands and 517 from Madagascar). Five novel mutations were detected in 10 Malagasy samples (1.9%): four synonymous mutations (codon G102G, n = 1; codon 182G, n = 1, codon T1069T, n = 1; codon E1095E, n = 6) and the S112L nonsynonymous mutation (n = 1). Among the previously described SNPs, only N86Y, Y184F, and D1246Y were found in samples from both countries, with there being eight alleles overall (Table (Table2).2). In 2006, the YYY and YFY alleles were more common in the Comoros Islands (P<0.05), while the NYD, YFD, and NFY alleles were more common in Madagascar (P < 0.05). The spatial distribution of the major alleles observed in Madagascar in 2006 to 2007 is presented in Fig. Fig.2b2b.
The pfmdr-1 copy number was determined for 233 Comorian and 350 Malagasy isolates. The mean copy numbers were 0.96 and 0.92 in the Comoros Islands and Madagascar, respectively. When the value was rounded to the nearest integer, no pfmdr-1 amplification was observed.
Multiplicity of infection, analyzed for a subset of randomly chosen isolates (472 from Madagascar and 86 from the Comoros Islands), showed that the proportion of monoclonal infections was not significantly different between the two countries (172/472 [36.4%] in Madagascar and 36/86 [41.8%] in the Comoros Islands; P = 0.40). The results of multilocus genotype analysis of 75 isolates that had a single allelic form and that were considered monoclonal are presented in Fig. Fig.33.
Among the 420 Malagasy P. falciparum isolates tested, 372 (88.6%), 373 (88.9%), 373 (88.9%), 370 (88.1%), and 281 (66.9%) were successfully assayed for their susceptibilities to CQ, MDA, QU, DHA, and MF, respectively. The in vitro activities of the five antimalarial drugs are summarized in Table Table3.3. The prevalence of isolates classified as having in vitro resistance to CQ, MDA, or MF was low; and there was no indication of temporal fluctuations. In contrast, a significantly decreased in vitro susceptibility to DHA was observed between 2006 and 2008 (0.76 nM, 1.02 nM, 1.83 nM in 2006, 2007, and 2008, respectively; P < 10−3), as shown in Fig. Fig.4.4. Interestingly, this was not associated with increased IC50s in some isolates but, rather, was associated with a drastic reduction in the number of isolates presenting a very low IC50 for DHA.
Among the 713 individuals interviewed (mean age, 23.4 years; age range, 1 to 90 years), 18% (range, 8% to 30%) declared that they had been prescribed an antimalarial treatment at the health center or hospital in the previous 30 days. Surprisingly, QU was the antimalarial most commonly prescribed by practitioners (48%; range, 28% to 77%), followed by CQ (34%; range, 8% to 53%) and SP (16%; range, 11% to 24%). The combination of artesunate plus amodiaquine was not prescribed, despite the official recommendations. Among the 37% (range, 19% to 51%) of individuals who had taken an antimalarial drug at home, 80% (range, 65% to 88%) declared that they had used CQ, 15% (range, 2% to 35%) declared that they had used SP, and 4% (range, 0% to 10%) declared that they had used QU.
A population mobility survey showed that one-third of the individuals (range, 13% to 48%) lived at another site 1 year before the interview. The proportion of individuals who traveled outside the site in the previous 30 days was estimated to be 6.6% (range, 2% to 15%).
The present study provides the first comprehensive picture of antimalarial drug resistance in Madagascar at a time that coincides with the implementation of the new national policy on antimalarial use (20) and completes previous results on in vivo drug efficacy (19). The work conducted within the antimalarial drug resistance surveillance network in Madagascar involved in vitro assays and sequencing of key loci implicated in drug resistance. A large number of P. falciparum isolates collected at health centers from multiple sites between 2006 and 2008 were studied. Hitherto, Madagascar was relatively unaffected by multiresistant P. falciparum parasites, especially parasites with the pfdhfr and pfcrt mutations (28, 29), and drug resistance in Madagascar displayed unique and uncommon features and markedly differed from the drug resistance observed in the Comoros Islands.
Comparative analysis of the prevalence of the pfdhfr SNPs showed the rapid emergence and spread of the IRNI allele in Madagascar, where parasites with pfdhfr mutations had not previously been detected (28). These findings are consistent with those presented in our previous report, in which we indicated that resistant P. falciparum parasites were introduced into Madagascar by travelers coming from the Comoros Islands (18). Resistant parasites were highly prevalent in the north/central western part of Madagascar in 2006 (>50%), and we observed their rapid spread along the north-to-south axis, reaching first the southwest (10-fold increase in 3 years) and then the eastern part of the country (1.5-fold increase in the central east and 3-fold increase in the southeast). These results are also concordant with those of in vivo efficacy studies, which showed that SP treatment failures were more frequent in the west (12.5%) and in the south (6.0%) (19). We suppose that the dissemination of the pfdhfr allele with triple mutations has been accelerated since 2006 by the massive use of SP, because of its better fitness compared to that of the wild type (38) and/or because of human population movements. A rapid rise in the prevalence of parasites with the single 164L mutation was observed in the south-to-north direction and is a worrying development. This single pfdhfr mutation was previously detected in the south of Madagascar in 2006 (17). Although it is rare in Africa (12, 16), it is obvious that this single pfdhfr 164L allele, which has not been described before, represents a serious threat that may lead to the emergence of a mutant with quadruple mutations, which will render SP ineffective. The mutant with quadruple mutations could arise locally by recombination between mutant parasites with triple mutations and parasites with the single 164L mutation (especially in areas with high rates of transmission, such as the southeast, where the prevalence of the both genotypes is high), or by stepwise selection of the 51I, 59R, and 108N mutations from the mutant parasite with the single 164L mutation, as suggested by our recent detection of a mutant parasite with a double mutation, NCNL (unpublished data). This adds to the threat of the possible spread of the mutant with quadruple mutations from the Comoros Islands.
Of the pfdhps point mutations related to sulfadoxine resistance, only A437G was prevalent in both countries, confirming previous findings of studies performed in the Comoros Islands (23). This SNP, which had never previously been described in Madagascar, showed the same prevalence and spatiotemporal evolution as the pfdhfr triple mutation. The rapid, concomitant rise in the frequencies of mutant parasites with both the pfdhfr and pfdhps mutations is challenging the current recommendation to use SP for intermittent preventive treatment of pregnant women (13).
The situation with regard to in vitro CQ resistance and CQ resistance-associated mutations in Madagascar is unique. The scarcity of parasites with the pfcrt mutation (3/693 isolates; Table Table2)2) and isolates resistant to CQ in vitro (3.2%; Table Table3)3) is striking in view of the high rate of CQ treatment failures (44%) observed in our survey of the efficacy of antimalarial drug treatment (19). Thus, CQ resistance in Madagascar is different from that in other African countries where malaria is endemic and the Comoros Islands. CQ treatment failure was first described in 1975 (9) and reached Madagascar early, at the same time that it reached East Africa (3). Since then and despite the massive use of CQ, the rate of CQ resistance has remained stable over the past 30 years, with the incidence of treatment failures ranging from 30% to 50% (31). This unique epidemiological situation, despite the regular introduction of pfcrt mutant-type alleles from the Comoros Islands (18), raises several questions. (i) Why is the spread of parasites with the pfcrt mutation rare in Madagascar, despite the huge pressure from the use of CQ? (ii) Is monitoring of isolates for the presence of the pfcrt mutant allele the best way to assess the incidence of CQ resistance in Madagascar, as suggested by Ariey et al. (2)? (iii) Are mechanisms unrelated to the pfcrt mutations responsible for the failures of CQ therapy?
To evaluate the contribution of pfmdr-1 to the unusual CQ susceptibility profiles in Madagascar, we analyzed pfmdr-1 SNPs and gene copy number, as Pgh-1 (encoded by mdr-1) is known to modulate P. falciparum drug sensitivity (37). This showed that SNPs at codons 86, 184, and 1246 in pfmdr-1 were highly prevalent, likely reflecting the widespread use of quinolines, such as CQ and QU. The elevated frequency of pfmdr-1 mutant alleles (YYD, NFD, YFD, and YFY) associated with CQ resistance in some studies (6, 7, 35), in a context of a low frequency of pfcrt mutant genotypes, may account for the high rate of late clinical failure of CQ treatment in the absence of in vitro CQ resistance. Further field studies are needed to clarify this issue and obtain a better understanding of the usefulness of the pfmdr-1 marker for predicting in vivo CQ resistance in Madagascar. pfmdr-1gene amplification has been associated with MF resistance in vivo (26) and in vitro (39). Interestingly, there was no evidence for any increase in pfmdr-1 copy number in Madagascar or in the Comoros Islands in the context of a lack of MF usage. In line with this, all the IC50s for MF were in the range classified as susceptible. Our observation that QU was the most frequently prescribed drug in the public sector in 2006 is informative and may point to the specific selective pressures exerted on the parasite population in Madagascar. Additional factors contributing to the uncommon drug resistance situation in Madagascar may be its geographical isolation, the broad range of malaria epidemiological strata in which the four Plasmodium species are present, and the admixture of inhabitants with multiple ethnic origins.
Finally, concern is raised by our observations of a significant decrease in DHA susceptibility between 2006 and 2008, since the implementation of the combination of artesunate plus amodiaquine in Madagascar. In the present study, we witnessed a progressive loss of the most susceptible isolates during a very short period of time. Close monitoring of in vitro susceptibility to artesunate and its companion drugs should be actively pursued, along with an exploration of the molecular markers including pfcrt and pfmdr-1 associated with some failures of treatment with artemisinin combined therapies in East Africa (10, 11, 21).
In conclusion, the work reported here provides evidence that the antimalarial drug resistance patterns in Madagascar are rapidly changing because of gene flow from the Comoros Islands and the local emergence and spread of polymorphisms. These changes reflect the efficiency of antimalarial drugs at eliminating susceptible parasites, but with the dark side of the coin being that the parasite population gets progressively enriched in less susceptible individuals. In this context, the establishment of regional and global monitoring networks such as the Worldwide Antimalarial Resistance Network (www.wwarn.org) is crucial to scaling up surveillance, rapidly sharing information on drug resistance, and helping with minimizing delays in implementing changes to antimalarial drug policies.
We thank the patients and health care workers involved in the national network for the surveillance of malaria resistance in Madagascar (Réseau d'Etude de la Résistance), from which the samples used in the present study were obtained, and the staff of the Ministries of Health of Madagascar and of the Comoros Islands for their collaboration. We are grateful to Marco Faniriko and Perlinot Herindrainy, who performed the interviews.
The following reagents were obtained through the Malaria Research and Reference Reagent Resource Center, Division of Microbiology and Infectious Diseases, NIAID, NIH: 3D7 Africa (MRA-102G), Dd2 (MRA-150G), HB3 (MRA-155G), and W2 (MRA-157G).
This study was supported by grants from Natixis/Impact Malaria through the Observatoire de la Résistance aux Antipaludiques Project and the Genomics Platform, Pasteur Génopôle, Pasteur Institute, France. Sample collection was funded by the Global Fund Project for Madagascar, Round 3 (Community Action to Roll Back Malaria grant no. MDG-304-G05-M). Valérie Andriantsoanirina is a graduate Ph.D. student funded by the Institut Pasteur de Madagascar (Bourse Girard) and the Direction des Affaires Internationales (Institut Pasteur).
Published ahead of print on 24 August 2009.