Background
Plasmodium falciparum, the most deadly species of
Plasmodium parasites that infect humans, remains a public health problem with the majority of cases and deaths occurring in sub-Saharan Africa [
1]. Anti-malarial drug resistance is a major public health problem that hinders the control of malaria.
P. falciparum resistance has been observed for all anti-malarial drugs used to date, including the artemisinin derivatives, where resistance has emerged in Asia [
2‐
4]. Continuous monitoring of the effectiveness of anti-malarial drugs both in vivo and in vitro plays a critical role in guiding treatment policy. Monitoring molecular markers of resistance is a quick and effective way to identify changes in drug resistance in real time. Malaria remains an important public health issue generally in Africa, and specifically in Sénégal and Tanzania, causing significant morbidity and mortality in infants and pregnant women [
2]. In Sénégal, the epidemiological profile is characterized by a stable endemic malaria, marked by a seasonal increase, with parasite prevalence trends having declined overall from 5.9% in 2008 to 1.2% in 2014 [
5]. However, malaria incidence remains elevated, especially in parts of the country where deaths attributable to malaria persist [
6]. In contrast, malaria transmission in Mlandizi, Tanzania is perennial [
7], with a high burden of malaria infection and clinical disease as indicated by the 678,207 reported cases of malaria in 2014 that resulted in 5368 deaths from malaria [
2].
Chloroquine (CQ) was the treatment of choice against the uncomplicated malaria in both Tanzania and Sénégal for decades. However, rising rates of CQ resistance led Tanzania to change its first-line treatment from CQ to sulfadoxine–pyrimethamine (SP) in 2001 and then to artemisinin-based combination therapy (ACT) in 2006 [
8]. Sénégal changed from CQ to SP-amodiaquine (AQ) in 2003 for use in seasonal malaria chemoprevention defined as the intermittent administration of full treatment courses of an anti-malarial medicine to children during the malaria season in areas of highly seasonal transmission (SMC) and then to ACT as first-line treatment of uncomplicated
P. falciparum in 2006. SP remains in use for intermittent pregnancy treatment (IPT) in both countries [
9‐
12].
SP is a combination of two antifolate compounds sulfadoxine that inhibits dihydropteroate synthetase (DHPS) and pyrimethamine that targets dihydrofolate reductase (DHFR). This combination acts synergistically against
P. falciparum, and SP resistance mutations have been well documented. Mutations resulting in the following amino acid changes N51I, C59R, S108N and I164L have been identified in the
dhfr gene associated with resistance to pyrimethamine [
13‐
19]. Mutations resulting in the following amino acid changes S436A, A437G, K540E and A613T/S in the
dhps locus have similarly been linked to sulfadoxine resistance [
20‐
25]. Despite high levels of resistance to SP in many countries, this drug combination is still widely used for treatment of uncomplicated malaria, for preventing malaria in pregnant women in the context of IPT [
2,
26], or in combination with artemisinin derivatives for SMC as recommended by the World Health Organization (WHO). Routine monitoring of genetic resistance mutations affecting SP efficacy is useful in determining whether the drugs should continue to be used for treatment of uncomplicated malaria or malaria pregnancy.
Different methods have been developed to evaluate the association of single nucleotide polymorphisms (SNPs) and specific phenotypes. Polymerase chain reaction (PCR) restriction fragment length polymorphism (PCR–RFLP), Taqman real-time PCR with allele-specific probes, and denaturing gradient gel electrophoresis (DGGE) are the most commonly used techniques that are suitable for these types of studies [
27,
28]. PCR/RFLP is time consuming and needs specific restriction enzymes for each SNP, as well as the ability to resolve and visualize the products using gel electrophoresis. DGGE requires extensive expertise that is not always available in disease endemic settings. Furthermore, the Taqman fluorescent probes are expensive and reagents expire rapidly. High-resolution melting (HRM) analysis is a post-PCR analysis method designed to investigate variance in nucleic acid sequences [
29]. Many studies have already published the accuracy, specificity and sensitivity of this technique, and its ability to detect minor alleles [
30‐
32], and identify new genetic variants that can be confirmed by sequencing [
29,
31]. HRM is a powerful analysis tool for large-scale genotyping as it is rapid, low cost and easy to deploy in the field.
The goals of this study was to: (1) compare the results of HRM to those using PCR–RFLP in the context of drug resistance marker surveillance in a malaria endemic country; and, (2) to determine the prevalence of mutations N51I, C59R, S108N in the dhfr gene and A437G, K540E, A581G, A613T/S in the dhps gene, across two malaria endemic settings with distinct frequencies of polyclonal infections (infections harbouring more than 1 parasite genome), as determined by MSP 1 and 2 genotyping.
Discussion
This study assessed the accuracy of HRM in comparison with PCR–RFLP for detecting infections of P. falciparum in two areas Mlandizi, Tanzania and Thiès, Sénégal two regions with variable endemicity and transmission intensity.
HRM analysis is comparable to PCR–RFLP for classifying SNPs; however, PCR–RFLP is laborious, time consuming, and requires a specific restriction enzyme for each SNP. This method also requires the separation of PCR products on a gel, which often takes hours to perform and increases the risk of contamination, making it difficult to genotype a large number of samples. Furthermore, interpretation of the digestion profiles can be subjective in cases of suboptimal digestion, low DNA yields, faint PCR products. Here, the results demonstrate that even when performed in a malaria-endemic laboratory setting, HRM is a rapid, accurate, powerful, economic, and a “closed-tube” mutation typing method that detects sequence variation within the PCR products, and can detect minor alleles in a mixed genotype population of parasite DNA. As described by previous studies, HRM can identify known and novel polymorphisms, detect multiple genotypes, and is both sensitive and specific [
29,
30,
38‐
41]. This study applied the HRM technology to type polymorphisms in mixed genotype infections in two African countries.
In Tanzania, more mixed genotypes were identified by HRM than PCR/RFLP at codon 51 (p = 0.005), 59 and 108. In Sénégal, a country with fewer polygenomic infections, no mixed infections was observed by PCR/RFLP, however several were detected by HRM, although the small number resulted in non-significant p-values (Table
1). These results demonstrate that HRM is more sensitive than PCR/RFLP and can easily detect mixed alleles. Since PCR–RFLP may not detect clones which are at low frequency in a mixed population, due to the qualitative nature of the assay, a minor allele could easily pass unnoticed. In contrast, HRM detected mixed infections at a higher frequency in both populations, suggesting that the technology of HRM to detect minor subpopulations is more sensitive than PCR–RFLP. While in countries like Senegal with few polygenomic infections and a low multiplicity of infection, there may not be a significant difference in the techniques; whereas, the improved sensitivity and ability to detect minor alleles is more pronounced in sample populations such as Tanzania with a high prevalence of polygenomic infections and a higher multiplicity of infection. This makes HRM a more attractive and accurate method for typing samples from both countries, but especially in countries like Tanzania, which are characterized by a high frequency of mixed infections. Furthermore, the ability to detect rare, low-frequency drug resistance alleles is likely important for surveillance of these markers as drug pressure is applied and likely to select for such variants.
As HRM was the most sensitive method evaluated, it was used exclusively for determining the genotype of
dhfr and
dhps genes to look at the drug resistance profiles in both countries. The frequency of mutant alleles at codon 437 in
dhps gene and at codons 51, 59 and 108 on
dhfr gene associated with in vivo and in vitro to SP resistance [
23,
42] was higher in Sénégal and Tanzania (Table
2). These high frequencies of mutation were observed in a study conducted in Dakar, Senegal [
43] and in Tanzania [
44]. The presence of mutations at codons 540, 581 on
dhps gene was not detected in either country.
In both countries, the high prevalence of mutations in
dhfr and
dhps could be explained by the use of SP as a second line treatment for malaria in Senegal and first line in Tanzania at the time of sample collection. In Sénégal, SP has been used since 2003 in combination with amodiaquine for use as SMC for children; whereas, in Tanzania, SP was introduced in 2001 as first line treatment for uncomplicated malaria but removed in 2006 due to the high level of resistance observed in vivo and in vitro. SP remains the mainstay drug regime for intermittent preventative treatment of pregnant women (IPTp) in both countries. It is very possible that the continued use of SP may favour stepwise selection of mutations in these areas, contributing to the high prevalence of mutant alleles observed in this study. The mutation A437G in the
dhps gene and N51I, C59R and S108N in
dhfr gene were more prevalent in Senegal than in Tanzania (Tables
2 and
3), which is interesting given that there has been longer term SP pressure in Tanzania compared to Sénégal. It has been observed in some studies that SP resistance emerges more rapidly in low-transmission compared to high-transmission areas [
45], and this is consistent with the results of this study.
One potential confounder to the more frequent resistant alleles in Sénégal compared to Tanzania is the difference in the MOI between the two sites. As many infections in Tanzania are polygenomic and contain a high MOI, it is possible that the number of mutant alleles circulating in the population is underestimated as both PCR-RFLP and HRM can classify alleles as mutant or wild-type, but cannot determine the number of alleles of each (just the total population profile: all wild-type, all mutant, or mixed). To address this challenge,
msp-
1 and
msp-
2 typing data was combined with drug resistance allele typing to determine the number of wild-type or mutant parasite genomes (Table
3). When accounting for the frequency on a parasite genome basis (rather than a per human basis), the results do not significantly change as mutant alleles are still higher in Sénégal than Tanzania.
Often, studies report combinations of mutations in both
dhfr and
dhps as a way to compare with WHO guidelines for continued SP use. When combining mutant alleles, the single mutation (
dhfr S108N) and the double mutation (
dhfr C59R/S108N) was more represented in Sénégal than in Tanzania with p = 0.01 and p = 0.005 respectively (Table
4). Triple and quadruple mutations were not significantly different between the two sites, although they were high for both sample sets. Encouragingly, the quintuple mutation N51I/C59R/S108N
dhfr and A437G/K540E
dhps gene, which is strongly associated with in vivo and in vitro SP resistance in East and Southern Africa [
46,
47] was not observed, consistent with findings from previous studies in Sénégal by Ndiaye et al. [
48,
49] and Wurtz et al. [
43]. However, a recent study conducted in Sénégal found a single sample with the quintuple mutation [
50], resulting in an overall population prevalence of 1.1%. In light of this result, continued and constant monitoring of drug resistance molecular markers is essential.
Authors’ contributions
YDN and CKD performed experiments and wrote the manuscript. ADA and DN conceived and designed the study. AKB conceived and designed the study, supervised the research, assisted with analysis, edited and reviewed the manuscript. BD, AM, NPM, RFD contributed materials and analysis tools and offered experimental advice. AG, MS, TN performed DNA extractions. ABD, ASB, MN and MD reviewed the manuscript. ZP, NF, DW, SM, SKV, ADA and DN supervised the research, wrote, read, and approved the final manuscript. All authors read and approved the final manuscript.