Molecular markers of resistance to amodiaquine plus sulfadoxine–pyrimethamine in an area with seasonal malaria chemoprevention in south central Niger

Although there has been great progress in malaria control, fewer than half of endemic countries account for 85% of all reported cases and most of these are in Africa [1]. In Niger, according to official estimates, the total number of confirmed and presumptive malaria cases was 846,509 in 2012 which roughly corresponds to a malaria incidence rate of 263 per thousand and malaria-related mortality was 0.12% [2]. While the entire population is exposed to the risk of malaria, the most vulnerable groups are pregnant women and children under the age of 5 years [3, 4]. Niger is within the Sahel, where malaria is markedly seasonal and most disease occurs during the rainy season from June to September. The epidemiological situation in Niger is thus suitable for the implementation of seasonal malaria chemoprevention (SMC) in children. This treatment involves the monthly administration of sulfadoxine–pyrimethamine (SP) and amodiaquine (AQ) during the season of high transmission when malaria endemicity and mortality are the highest [4]. Several studies have shown that SMC provides protection against malaria, resulting in a decrease in the number of clinical episodes and mortality, and in the incidence of severe malaria, but logistical challenges are formidable in some parts of eligible regions [5–8].

Malaria episodes during pregnancy compromise the health of the mother and the development of her fetus, particularly in primiparous women [9]. The World Health Organization (WHO) recommends that in high transmission areas, SP be given at least a month apart at the 4 antenatal visits after the first trimester for preventive treatment of malaria (IPTp) [10]. In many regions, clinical responses to SP are seriously compromised, and SP is no longer recommended for treatment of malaria episodes. However, it has been difficult to determine whether the efficacy of SP for IPTp is similarly compromised.

These uncertainties mean that decisions about whether to introduce SMC or IPTp as additional malaria preventive strategies are partly dependent on the local level of resistance to the antimalarial drugs being used [9–11]. Epidemiological monitoring of susceptibility to SP can be estimated by measuring the prevalence in the parasite population of key mutations that serve as surrogate molecular markers of the clinical response to anti-malarials. For SP, the prevalence of parasites that carry the “quintuple mutant” genotype, both the triple mutant haplotype, pfdhfr allele N51I/C59R/S108 N and the double mutant haplotype, A437G and K540E mutations of the pfdhps gene are strongly associated with clinical failure of SP treatment [10, 11]. Similarly, parasites that carry a N86Y/F184Y/D1246Y haplotype of the pfmdr1 gene are associated with reduced susceptibility to amodiaquine [12–15]. Whether SP for IPTp or SP + AQ for SMC is still protective in locations where these genotypes are common is under active investigation. Therefore, surveillance of these and other additional markers of SP and AQ response is a crucial element of future clinical decisions on implementation of SMC and IPTp in Niger.

Information on resistance to anti-malarials is scarce in Niger, but an evaluation of these markers of SP and AQ resistance can provide a baseline with which to assess any changes in these molecular markers associated with the implementation of SMC. To that end, this study reports the prevalence of mutations of the pfdhfr, pfdhps and pfmdr1 genes associated with resistance to SP and AQ in Plasmodium falciparum blood samples collected from children living in a semi-urban area of Niger before and during the implementation of SMC.

A total of 201 blood samples were collected and processed for DNA amplification. The mean age and the mean body weight (and standard deviation) of the patients included in the study were 22.8 ± 11.7 months [range: 6–55] and 9.5 ± 5.7 kg [range: 5.3–15.5], respectively, with a sex ratio of 1.04 in favour of girls. Patients included in the study had a P. falciparum monoinfection with a median parasite density 14,076 parasites/mm [range 80–202,000]). Of the 201 samples amplified by PCR with pfdhfr, pfdhps and pfmdr1 primers, 120 (59.7%), 103 (51.2%), and 95 (47.2%), respectively, produced fragments in sufficient yield for sequencing. The electropherograms showed that 13.8, 11.1 and 10.8% of the amplified pfdhfr, pfdhps, and pfmdr1 sequences, respectively, were heterogeneous consisting of a mixture of wild type and mutated residues at one or more individual polymorphic positions. These polyclonal clinical samples were not included for the calculation of the allele prevalence and reconstruction of the various haplotypes. In total, 136 amplicons of pfdhps and 120 amplicons of pfdhfr were sequenced. Combined genotypes were established using the samples that were successfully analysed at one (pfdhfr, pfdhps or pfmdr1), two (pfdhfr and pfdhps), or three (pfdhfr, pfdhps and pfmdr1) loci.

Parasites carrying 51I/59R/108N triple-mutant allele of pfdhfr and the 437G/540E allele of pfdhps are major contributors to SP treatment failure [12, 20–22]. In most West African countries, the pfdhps allele with the single mutant, 437G are common, but parasites with the triple mutant pfdhfr and the key 437G/540E substitutions are still rare (see WWARN SP surveyor).

SP intermittent preventive treatment of women in the last two trimesters of pregnancy (IPTp) appears to still provide some benefit, even in the presence of high levels of this common quintuple genotype [9]. However, the WHO recommends surveillance for this genotype, and replacement of SP for IPTp when these quintuple mutant parasites reach prevalence greater than 50%. More recently, in East and Central Africa, parasites with a triple mutant pfdhps allele 437G/540E/581G have been observed to abrogate the benefits of IPTp with SP in some areas of East and Central Africa [23, 24].

The IRN triple mutant allele of pfdhfr was present in 57% of the clinical samples. The triple mutant pfdhfr is common in West Africa as reported for neighbouring Benin and Burkina Faso and Gabon in Central Africa [25–27]. In most of West Africa and likely reflects spreading of parasites that carry the imported pfdhfr Southeast Asian allele from East African sources [28–30]. The clinical ramifications of the pfdhfr triple mutant allele may be largely counterbalanced by the absence of pfdhfr I164L.

The parasite genotypes of pfdhps in Gabi were more complex. The IRN/GAA haplotype, the most frequent in Gabi, is a marker of a lineage phylogenetically different from those previously identified in Asia and India [31]. Twenty-two % (23/108) of the clinical samples carried a mutation both at codon 437G and 613S; moreover, nine of these clinical samples carry a triple mutant allele of pfdhps that also include 581G (GKGS). Only one parasite (1/36) observed in Gabi, carried the triple pfdhfr with a unique haplotype of pfdhps, 437G/581G. It is not clear whether this strong effect of parasites carrying the 581G allele in the absence of 540E would be equally deleterious.

The mutations in position 613 of the pfdhps gene, and the replacement of an isoleucine residue at position 431 with a valine have seldom been observed outside Niger. The I431V mutation was reported the first time in 2007 in Nigeria for a limited number of clinical samples but, in contrast to what was found in Nigeria, the presence of the I431V mutation was not associated in Niger with the A581G and A613S mutations. Instead, it was associated with the S436A and A437G mutations [26, 32].

These reported prevalences of parasites with genotypes known to compromise SP efficacy in Gabi are, in any case, low and unlikely to generate a significant effect on the clinical outcome following preventive treatment with SP. The distribution and prevalence of these mutations should nevertheless carefully be explored to determine their significance in the response of parasites to SP. Most important, parasites carrying the I164L mutation of pfdhfr and the K540E mutation of pfdhps in addition to the “usual” quintuple are even more strongly associated clinical treatment failure and no parasites with this genotype were detected in Gabi before the deployment of SMC with SP + AQ.

Molecular characterization of the pfcrt and pfmdr1 markers facilitates prediction of the responses of parasites to chloroquine (CQ) and AQ, but also to mefloquine and lumefantrine. It has been shown that parasites combining both the pfcrt 76T and pfmdr1 86Y mutations and those carrying the pfmdr1 86Y/184Y/ 1246Y triple-mutant haplotype are resistant to 4-aminoquinolines and associated with clinical failure with AQ treatment [22, 23]. Selection of the pfmdr1 N86/F184/D1246 haplotype has been reported during treatment with artemether–lumefantrine and in infections with increased sensitivity to mefloquine and decreased susceptibility to lumefantrine [23]. Parasites that carry 2 or more copies of the wild type allele of pfmdr1 are also highly resistant to both lumefantrine and mefloquine [33]. In Niger, the F184Y mutation was found to be the most frequent, with 37% of sequences displaying this mutation. The 86Y, 1042D and 1246Y mutations were less common, with prevalences of around 0.1. In the study region, almost all (88%) the parasites carried the NFD or NYD haplotype, reflecting the selection exerted by AL, the first line treatment in Niger. Parasites carrying the YYY form were not detected. The pfcrt sequences were not determined in this study, but other analyses performed in southern Niger during the same period showed that the wild-type CVMNK allele (amino acids 72–76) of the pfcrt gene was present in 85% of clinical samples [34]. This is consistent with previous in vitro results, which indicated that the clinical samples from Niger respond adequately to other 4-aminoquinolines like AQ [30]. The pfmdr1 sequences characterized predict acceptable susceptibility of infections to AQ. Only 11% of the clinical samples carried the pfmdr1 86Y mutation and the 86Y/184Y/1246Y triple mutant allele was absent. It is also important to recognize that blood-stage resistance mechanisms may not prevent anti-folates from working to some degree in the liver-stages.

In contrast to other countries, SP treatment has never been recommended as a first-line anti-malarial treatment in Niger. The artemether–lumefantrine combination was introduced in 2005, to replace CQ, whereas the use of AQ has remained marginal. The artesunate–amodiaquine combination was introduced later on in 2008 to expand the options for artemisinin-based combination therapy (ACT) available in Niger. The resistance profiles currently observed thus reflect the treatment policies implemented by the health authorities in Niger over the last 10 years. In contrast to this history, the K540E mutation of pfdhps is found principally in the regions in which SP was used as a first-line treatment to replace CQ before the introduction of ACT. In West African countries in which CQ was replaced directly with ACTs, the K540E mutation emerged later. Thus, the strength and order of the drug pressures exerted on the parasite are likely to constrain and structure the parasite populations in a persistent manner [10, 12].

It is important to highlight several key limitations of our study. First, the sample size is both limited and restricted to children presenting at a health centre with fever, so the results presented here may not reflect the prevalence of these genotypes in the general parasite population. However, as children under 5 account for 20% of the total population of Niger, the expected benefit of SMC could be considerable. Although a high prevalence of the triple pfdhfr haplotype was observed, the codon 540 of the DHPS gene remained wild type, and SP was likely efficacious at least in 2012. Based on the continued efficacy of SP for prophylaxis in pregnant women, it is possible that even in the presence of the “usual” quintuple (IRN/GEA), SP might still retain efficacy for SMC.

Molecular surveys of the prevalence of parasite genotypes can provide useful information to guide decisions on local drug use, particularly for malaria prevention. It would be even more useful if the marker prevalence could be assumed to apply to other locations, or at least sites very close to the site sampled. However, a related study of both SP and AQ molecular markers was completed in 2013 only 15 km away from Gabi, and the prevalence of the pfdhps haplotype, 437G/540E was about 20% at the beginning of SMC implementation, and rose to about 50% by the end of the intervention [34]. Although these samples were collected 1 year apart, this molecular heterogeneity in clinical samples from geographically close sites is not uncommon [28, 35].

This spatial heterogeneity is likely to be common, and presents a challenge for policy makers who have responsibility for setting drug use policies on a national scale for both SMC and IPTp. In any case, regular monitoring of anti-malarial drug resistance should remain a key activity in the sites where SMC and SP IPTp are implemented. Given the sometimes large differences between maker prevalence, even between nearby sites, determination of the molecular marker prevalences in the actual study site should be a high priority to monitor anti-malarial resistance.