Background
The efficacy of anti-malarial drugs in endemic areas is assessed over a follow-up period of 28 to 63 days, depending of the drug's residence time in the organism, following initiation of treatment. While a longer follow-up allows capturing more late failures [
1] the likelihood of re-infection increases in a way that is dependent upon the intensity of transmission in the study area. Comparing molecular genotypic pattern of pre-treatment (baseline) and recurrent infections provides a means to help characterize the recurrent parasites as a recrudescence, i.e. a true failure, or a new infection (either from pre-existing liver infection or a newly established infection from an infected mosquito bite), i.e. a successful treatment.
Several
Plasmodium falciparum genes show extensive genetic polymorphism. This phenomenon is exploited for genetic finger printing and for assessing parasite population dynamics. For instance, high polymorphism has been shown in
msp1,
msp2 and
glurp genes in different geographical locations in malaria endemic areas [
2‐
7]. The probability of a patient, particularly in areas of intense transmission, to be newly infected with a parasite possessing an identical genotype to the former infection is low [
8]. Therefore, comparing the genotypes of the three loci at baseline and at the time of parasite recurrence would be expected to discriminate between recrudescent and new infections [
8,
9]. Numerous clinical drug trials have applied this approach to correct the outcomes of drug efficacy studies [
10‐
17].
However, the discriminating power of different markers is dependent on the extent of allelic diversity and on the frequency of each allele within the population under study. Indeed, recurrent episodes after treatment can be reliably classified as recrudescence or re-infections if the frequencies of the msp2, msp1, and glurp alleles, as detected by the genotyping protocol employed, are known. It is obvious that a second infection appearing during follow-up after the first has been apparently cleared can be erroneously classified as recrudescent if some alleles predominate in the population or if heterozygosity is low, because under these circumstances a new infection would be more likely to share the same genotype as the baseline infection. This will lead to an over-estimation of treatment failures and consequently unnecessary treatment policy changes.
The allelic diversity of
msp2 has been observed to be high in some areas, such as in the Kilombero valley in Tanzania (82
msp2 alleles) [
2], Papua New Guinea (42
msp2 alleles) [
18], Ghana (164
msp2 alleles) [
19] and Côte d'Ivoire (50
msp2 genotypes) [
20]. Conversely, there is very limited information on
msp1 or
glurp diversity across sub-Saharan Africa. This study presents the
msp1, msp2 and
glurp genetic diversity and allele frequencies in five Sub-Saharan African countries with different transmission intensities namely, Malawi, Tanzania, Uganda, Burkina Faso and São Tomé.
Discussion
Polymorphic regions from of the
P. falciparum msp1, msp2 and
glurp loci have been selected as the recommended markers for parasite genotyping in anti-malarial drug trials and efficacy studies [
23]. However, the parasites' genetic profile has not been systematically documented in many malaria endemic countries. A large number of archived
P. falciparum positive pre-treatment infections were genotyped in order to compare the diversity and allelic frequencies for these three markers among five geographical areas with different transmission intensities across sub-Saharan Africa; namely Malawi, Tanzania, Uganda, Burkina Faso and São Tomé. These blood samples were collected in the region during two clinical trials (WHO/TDR and IHI/Swiss TPH) of artemisinin-based combination therapy that were conducted between 1996 and 2000. The findings from these studies have been published elsewhere [
22,
13]. The aim of this study was to determine whether the genetic diversity of the markers or their suitability for PCR-correction of drug efficacy trials in endemic countries, varied between countries. In addition, since the PCR-corrected treatment failures in both of the above trials were >10%, the study also aimed at validating these corrections by assessing the Day 0 genotypic profile as recommended [
23].
Although Isocode stix were stored at room temperature over nine years, the majority (53 - 95%) of the DNA samples could be amplified; with some unexplained variation unrelated to amplified fragment size across the geographical sites. This molecular genotyping study shows that on average the majority of the patients were infected with more than one parasite genotype on the day of admission. The mean MOI values were heterogenous across the different sites, being lowest in Uganda, and highest in Burkina-Faso and Tanzania. Mean MOI was highest for
msp2 (2.24), followed by
msp1 (1.48) and was lowest for
glurp (1.40) and the allelic diversity followed a similar trend recording 116, 17 and 14 alleles, respectively. The allelic variants were spatially distributed across the five Sub-Saharan African countries. The differences in allele diversity between
msp2 on the one hand and
msp1 and
glurp on the other are clearly attributable to the method used for DNA fragment sizing. Indeed capillary electrophoresis used for
msp2 has a much higher power of resolution than gel elecrophoresis and digitalized fragment sizing used for
msp1 and
glurp[
24]. However, since the genotyping methods used were the same for all geographic sites, it is possible to compare diversity of a given marker between countries. Tanzania recorded the highest genetic diversity, while Uganda recorded the lowest diversity in all the three markers.
In the WHO/TDR trials [
15], the crude post day-14 parasitological recurrence rate for all treatment groups was 22% and was 89.7% for chloroquine alone and 13.6% for artemether-lumefantrine for the IHI-Swiss TPH trial [
21]. According to the recently adopted recommendations for malaria genotyping [
23], PCR-corrected failure rates exceeding 10% would require determining MOI and allelic frequencies in order to confirm the validity of the PCR-correction. Thus observation of high MOI, high allele diversity and low allele frequencies for all markers (especially
msp2 whose fragments were sized by the most accurate method) and study sites are strongly indicative of the high discriminatory power of this three-marker genotyping strategy. These findings further validate the PCR-adjusted outcomes recorded previously [
15,
21].
High genetic diversity and low allelic frequencies have been reported previously from other sites that differ substantially in transmission intensity: Gabon [
7], Uganda [
10], Senegal [
25], Burkina Faso [
26] and Honduras [
27]. Nonetheless, three allelic variants of
msp1 (151-176 bp, 176-201 bp and 201-226 bp), four of
msp2 (wos12, wos3, K1-long and FC27-455) and five of
glurp (857-907 bp, 907-957 bp, 957-1007 bp, 1007-1057 and 1057-1107 were recorded at frequencies exceeding 5% in some of the study sites. Such frequently occurring genotypes might lead to misclassification of recurrent parasites in clinical trials and drug efficacy studies. However, from the current data even the most abundant genotype of
glurp with a recorded allelic frequency of 20%, the probability of acquiring a new independent infection of the same genotype by chance equals the square of the allelic frequency. This probability of 4% for a reinfection with the same genotype is not very high for a single genotyping marker; it would be even much lower when all three markers were genotyped.
It should be noted that the blood samples analysed were collected more than nine years ago, hence the genetic profiles described might no longer accurately represent the current situation. Nonetheless, this study provides important genetic background data in these areas. The strength of the present study hinges on four major factors: (i) the large number of baseline samples that, (ii) were collected from five countries with different levels of transmission intensity, (iii) use of automated DNA sizing methods to remove investigator bias and error in assigning molecular weights of the PCR products, and (iv) comparative analyses across countries could be conducted, because the data was obtained using the same amplification protocol.
One of the factors that directly impinge on the utility of genotyping protocols in drug efficacy studies is that of bin size selection. For this study, as for all previous studies, the bin sizes for
msp1 or
glurp fragments have been set rather arbitrarily. The lengths of the repeat units, whose number varies between the different allelic variants, was taken into account to set bin width of 25 bp and 50 bp for
msp1 and
glurp, respectively. In contrast,
msp2 fragments were sized by capillary electrophoresis with 3 bp bin width that quite obviously represents the smallest size difference possible in a coding region. In previous studies, different bin widths were used, for example, [
28] used a conservative bin width of 40 bp for
msp1 and
msp2, whilst [
10] used bin sizes of 10 bp for
msp1 and
msp2 and 20 bp for
glurp. In addition to these variations in bin width, the different fragment sizing methods employed by different researchers make it difficult to compare data for a particular marker between studies.
Another drawback of this genotyping protocol is the variability in the electrophoretic migration of a given DNA fragment between gels, and indeed between different regions of the same gel. This is clearly illustrated in our observations of variable RO33 fragment sizes by digital gel documentation, despite the fact that these fragments were confirmed to be of the same size (215 bp) by sequencing. It is most likely that such spurious variability will also occur for variants of the other msp1 allelic families, and maybe even to a greater extent in the much larger glurp fragments. Thus, the quality and value of the data obtained from binning of alleles across different gels depends largely on gel quality (e.g. "smiling effect" or unequal loading of gel slot) and on the accuracy of image digitization. Electrophoretic variability is a potential problem when one wishes to establish the frequencies of each allelic variant. When sizing is done across different gels, binning through an image analyzer might be more prone to error than doing it by eye, especially when samples are run side by side. This error observed when comparing fragments from different gels, is highly unlikely to affect the validity of the genotype pattern comparisons of baseline and recurrent infections, on which PCR correction is based, because the amplified products from these samples are usually migrated in the gels side-by-side and, therefore, with a much reduced chances of variability. For such comparisons, detection of size differences can either be done visually or through an image analysis programme. Admittedly, visual analysis also poses some degree of subjectivity: at what level of difference in migration does one say that two bands are different, especially when they differ in quantity and consequently intensity and thickness in the gel.
Ultimately, the ability to distinguish between two allelic variants depends directly on the resolution of the method used to analyse the amplified fragments. At present capillary electrophoresis offers the best solution, because it provides accurate and reproducible estimates of DNA fragment lengths with a resolution power down to a few base pairs difference. It is highly likely that the
msp1 and
glurp fragments amplified in the course of this study in fact encompassed a larger repertoire of distinct allelic variants than those resolved by simple agarose gel electrophoresis. By the time this study was done, only
msp2 capillary electrophoresis protocol was described [
19]. Now protocols for
msp1[
29,
30] and
glurp[
24] have been developed and all three markers can be analyzed and allele diversity/frequency compared more accurately.
Authors' contributions
Study conceptualization and design was done by KM, HPB and PO. KM and IF did study monitoring and supervision in Tanzania and Switzerland, respectively. IF and GS provided technical advisory support in genotyping and data interpretation. FM, DS, SS and KM carried out the laboratory work. FM and GN performed statistical analysis. FM, GN, and KM composed the primary version of the manuscript and all other authors contributed modifications. All authors read and approved the final manuscript.