Introduction

Despite overall global progress made towards disease control and even elimination in a few countries over the past two decades, malaria remains a leading cause of morbidity and mortality, particularly among young children and pregnant women in Africa [1]. Effective disease control interventions include the use of rapid diagnostic tests for malaria, artemisinin-based combination therapies (ACT) for treatment, preventive antimalarial drug administration in African children and pregnant women, antimalarial vaccine in African children, distribution of bed nets to prevent mosquito bites, and various chemical means to kill mosquitoes. These disease control measures are directed against Plasmodium falciparum [2]. On the global scale, these interventions have also been beneficial in reducing the malaria burden due to Plasmodium vivax, another human malaria species with a wider geographical distribution than P. falciparum, leading to a considerable decrease in the number of P. vivax cases over the past two decades [1]. Plasmodium vivax malaria occurs in Asia, in South and Central America, in the Near East, in the Oceania region, as well as in some African countries [3–10].

As many countries adopt measures to control and eliminate malaria, concerns have been raised about P. vivax malaria, often considered a neglected disease [3–7]. These concerns are warranted, based on established knowledge and recent findings. First, contrary to long-held beliefs, P. vivax is widespread in Africa, including areas where it was thought to be non-existent [8–10]. The misconception that P. vivax is absent from sub-Saharan Africa stems from the “Duffy-malaria hypothesis:” most individuals of black African origin are resistant to P. vivax infection due to the absence of the Duffy antigen, also known as the “Duffy antigen receptor for chemokines” (DARC), on the surface of erythrocytes [11,12]. In contrast, populations of non-black African origin generally have a Duffy-positive phenotype, making them susceptible to P. vivax infection. Duffy antigen has been considered the major receptor for the interaction between Duffy binding protein (DBP) of P. vivax merozoites and host reticulocytes during erythrocyte invasion [13], as supported by the general observation that Duffy-negative phenotype offers relative protection from P. vivax infection. Second, in Asian and South American countries where P. falciparum and P. vivax co-exist, the prevalence of P. falciparum has been decreasing faster than that of P. vivax, leading to an increasing proportion of P. vivax cases [14,15]. Third, the now outdated notion of P. vivax infection as a “mild” disease has been overcome. It is now recognized that P. vivax malaria can cause severe and complicated illness, potentially leading to a fatal outcome [16–19]. Fourth, P. vivax (and also P. ovale) is distinguished by unique biological features, the most significant for malaria control being its ability to produce hypnozoites during the hepatic stage [20,21]. These dormant hepatic forms can reactivate weeks, months, or even years after the primo-infection, causing new blood infections called relapse [20,21]. Fifth, the only effective drugs that can kill hypnozoites, 8-aminoquinolines (e.g., primaquine and tafenoquine), are still largely underused and/or misused in treating P. vivax and P. ovale infections. This is either due to unavailability of these drugs in many African pharmacies or poor patient compliance with the standard 14-day treatment regimen [22–26]. As a consequence, many patients with laboratory-confirmed P. vivax malaria are not receiving the full treatment recommended by the World Health Organization (WHO), which includes both blood schizonticides (chloroquine or ACT) and anti-hypnozoite drug, to achieve “radical cure,” i.e., the elimination of both blood-stage and latent liver-stage parasites [2,27]. After incomplete treatment, these patients become potential reservoirs and sources of new infections, thereby hampering efforts to interrupt disease transmission.

The major problem with the administration of 8-aminoquinolines as anti-hypnozoite drugs is their potentially serious adverse effects, in particular acute hemolytic anemia, that may occur in patients with glucose-6-phosphate dehydrogenase (G6PD) deficiency [28]. Plasmodium vivax-infected patients should be screened and assessed for the level of G6PD enzymatic activity before prescribing primaquine or adjusting primaquine dose in individuals with mild to moderate G6PD deficiency. To promote, facilitate, and enhance the feasibility of primaquine administration in point-of-care settings, several on-the-spot field diagnostic tests for G6PD activity have been developed and assessed in the field [29–33].

G6PD deficiency is an X-linked genetic disorder and is therefore more commonly expressed in men who are hemizygous [34]. In addition to sex, ethnic origin is a major factor that determines not only G6PD deficiency, but also Duffy-negative genotype and phenotype. Malaria has been the selective force for G6PD deficiency (for both P. falciparum and P. vivax) and Duffy-negative phenotype (for P. vivax) to provide protection from these malaria species or, at least, from severe malaria [35,36]. Due to the presence of several ethnic groups belonging to Arab-Berber or black African ancestry, as well as P. vivax [37–39], molecular studies on G6PD and Duffy antigen genotypes in Mauritania may provide insight into how these genotypes/phenotypes interact to determine the most appropriate and safest clinical management of P. vivax-infected patients with 8-aminoquinolines.

Recent studies have demonstrated that G6PD deficiency occurs much more frequently in Mauritanians of black African descent [33,40–42]. As for Duffy phenotype or genotype, only very limited epidemiological data are available in Mauritania. In the first known published study on Duffy antigen in Mauritania [43], it was reported that 54% (28/52) of white Moors were Duffy-positive, and 46% (24/52) were Duffy-negative, based on blood group phenotyping. The “black Moors” (defined in that study as those belonging to all other ethnic groups, including Haratins or black Moors, Wolofs, Poulars, and Soninkés) were mostly Duffy-negative (54/55, 98.2%). In another more recent study based on molecular genotyping of blood samples from P. vivax-infected patients matched with those who were not malaria-infected [44], most white Moors (186/226, 82.3%) were Duffy-positive, and all cases of P. vivax infection (n = 77) were diagnosed exclusively in Duffy-positive white Moors. Among the few febrile Mauritanian patients of black African descent included in that study (n = 14), only two (2/14; 14.3%) were Duffy-positive. Among 12 Duffy-negative patients (12/14, 85.7%), one Duffy-negative patient belonging to Wolof ethnic group was infected with P. vivax, supporting the increasing number of recent reports from sub-Saharan African countries that few Duffy-negative Africans can be infected with this malaria species [45–50].

Previous studies conducted in Mauritania have compartmentalized different aspects of human genetics (i.e., Duffy antigens and G6PD activity) and P. vivax infection. Intuitively, we hypothesize that Duffy phenotype and G6PD deficiency have an “opposite” effect in Africa [35,36]: non-black Africans are usually Duffy-positive and the African-type G6PD A^– genotype is rarely encountered, while most black Africans are Duffy-negative, some of whom may have G6PD A^– genotype. Although non-black Africans along the Mediterranean coast may be affected by the Mediterranean-type G6PD B^– genotype, we have shown that this genotype is absent in different ethnic groups residing in Mauritania [40]. Based on this hypothesis, Duffy-positive Mauritanians are likely to be white Moors who are less affected with G6PD deficiency, which implies that, in general, they are susceptible to P. vivax infection and are likely to have a normal G6PD activity, which allows the administration of the standard dose of primaquine for radical cure. By contrast, Duffy-negative Mauritanians tend to be non-white Moors who are relatively protected from P. vivax infection but have a higher probability of being affected by G6PD deficiency. However, in a relatively “rare” case of a Duffy-negative individual being infected with P. vivax [44], we may be confronted with the problem of identifying such individuals and prescribing a modified primaquine dose or even withholding anti-hypnozoite treatment with primaquine. In a country like Mauritania where a multitude of peoples of different ethnic origins co-exist, the probability of encountering such a scenario needs to be assessed, and a relevant guideline is required for health practitioners to ensure a safe administration of primaquine. It is in this context that the present epidemiological study was conducted to analyze the allelic frequencies of mutations in Duffy and G6PD genes and predict the possibility of administering 8-aminoquinoline drugs in case of P. vivax infection. This scenario extends well beyond the borders of Mauritania and is relevant in other sub-Saharan African countries in continental Africa that have reported P. vivax malaria in Duffy-negative patients [45–50]. Other continents, as in Central and South America, where descendants of black African ancestry are present, are also concerned [51–53].

Materials and methods

Results

Duffy blood group genotypes and predicted phenotypes

Genotypes of Duffy blood group system were determined by multiplex amplicon-based sequencing in a total of 1,101 blood samples (427 males and 674 females) (Fig 1). Interpretable sequence data on Duffy antigen polymorphisms at all 11 codons (or specific nucleotides in the promoter region) were available from 954 (86.6%) samples. PCR amplification failed either partially at one or several targeted SNPs or completely for all 11 SNPs in 147 (13.4%) samples most likely due to an inadequate amount of blood spots and/or possible DNA degradation associated with poor conservation conditions before transporting the samples from the field to the laboratory in France (note: the archived samples were collected in 2015–2020 and exposed to high ambient temperature in Mauritania). Because of the limited amount of blood spots available for each sample, no attempt was made to repeat PCR sequencing in case of failure. A similar PCR success rate was found using the same samples in an earlier study on G6PD genotyping [40]. Among 954 successfully sequenced samples with complete results at 11 SNP markers, the mean sequencing depth across individual samples was 7000 X (25^th percentile, 485 X; 75th percentile, 9060 X) for the amplicon harboring the promoter region, and 652 X (25^th percentile, 36 X; 75^th percentile, 808 X) for the second amplicon flanking the 5’-terminus of the exon 2 (S1 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow chart ([77]).

https://doi.org/10.1371/journal.pntd.0013882.g001

Eleven major mutations known to be associated with Duffy blood group system and its isoforms were identified in blood samples obtained from Mauritanian patients (Table 1). Fixed wild-type alleles were observed in all patients at seven SNPs: T-69C, G145T, G266A, G287A, G395A, G407A, and G408A. Co-dominant FY*01 and FY*02 alleles (also commonly referred to as the FYA and FYB alleles, respectively), which differ by a SNP at nucleotide 125, were used to define the phenotypes, based on the criteria presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Mutations in the atypical chemokine receptor 1 gene (ACKR1 gene coding for the Duffy blood group system) analyzed by multiplex amplicon-based DNA sequencing.

https://doi.org/10.1371/journal.pntd.0013882.t001

Nineteen individuals (19/954; 1.9%) were homozygous for the FY*01 allele, and 681 individuals (681/954; 71.4%) were homozygous for the FY*02 allele (Table 2). Among these 681 individuals homozygous for the FY*02 allele, 165 (17.3%) carried the T > C mutation at nucleotide position T-67C which prevents Fyb antigen expression only in the red blood cells in one of the chromosome pairs (i.e., genotype FY*02/FY*02N.01 and phenotype Fy(a + b^–)) and 430 (45.1%) in both chromosome pairs (i.e., genotype FY*02N.01/FY*02N.01 with a Duffy null or erythrocyte silent (ES) phenotype Fy(a^–b^–)).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Duffy antigen genotypes and predicted phenotypes in Mauritanian patients.

https://doi.org/10.1371/journal.pntd.0013882.t002

A total of 254 (254/954, 26.6%) individuals were heterozygous for the FY*01 and FY*02 alleles, among whom 116 (12.2%) were predicted to express both Fya and Fyb antigens normally (i.e., genotype FY*01/FY*02 and phenotype Fy(a + b+)). A total of 119 (12.5%) carried the FY*02 allele with the T > C mutation at nucleotide position T-67C which prevents Fyb antigen expression (i.e., genotype FY*01/FY*02N.01 and phenotype Fy(a + b^–)). Ten patients (1%) had FY*01N.01/FY*02N.01 genotype, which corresponds to a Duffy null phenotype Fy(a^–b^–) (Table 2).

Two other mutations were reported to prevent Fyb antigen expression: G407A and G781A. The nucleotide G407 was wild-type in all samples. The amplicon-based method used in the present study was not designed to determine the SNP at nucleotide 781 in exon 2.

The weak expression of both Fya and Fyb antigens was determined by the co-occurrence of mutations C265T and G298A (genotype FY*02W.01). Only two individuals (0.2% of all samples) carried the FY*02W.01 (C265) allele. These two individuals were, however, not predicted to have a weak Fyb antigen expression because they also carried a FY*02 allele (i.e., genotype FY*02/FY*02W.01) [78]. All other alleles known to be associated with the weak expression of Fya or Fyb antigen (i.e., G145T and G266A) were either fixed in all samples or not included in the present study (i.e., C901T).

Five patients carried a non-referenced G > C mutation at position 125 at the heterozygous state with allele FY*02. Of these five cases, four were homozygous C at position T-67C, which would probably result in Duffy null phenotype Fy(a^–b^–), while one patient carried both T and C nucleotides at this position. Four additional individuals carried a non-referenced G > T mutation at position 125 at the heterozygous state with allele FY*02, two of whom had the T > C mutation at nucleotide position -67 at the homozygous state. These mutations are non-synonymous and encode an alanine (G125C) or a valine (G125T). We did not predict the Duffy blood group phenotypes for these individuals, including those with missing genotypes at nucleotide positions 125 and/or -67.

Duffy genotypes were not equally distributed between the sexes and among ethnic groups (Figs 2 and 3). The predicted Duffy blood phenotypes were significantly associated with sex, with females (295/579, 50.9%) being more affected with Duffy-negative phenotype than in males (145/366, 39.6%) (Pearson’s chi-squared test with simulated p-value, p = 0.001) (Fig 2). This unexpected, slightly skewed distribution of Duffy genotypes between sexes was probably due to sampling bias in our hospital-based studies which included patients enrolled in the Mother and Child Hospital. The genotype FY*01N.01/FY*02N.01 was exclusively represented in 10 females, while FY*02W.01 allele was only found in two males.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Proportions of different Duffy genotypes with respect to sex.

Duffy genotypes in green and orange correspond to Duffy phenotype + and –, respectively.

https://doi.org/10.1371/journal.pntd.0013882.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Proportions of different Duffy genotypes with respect to ethnic background.

Duffy genotypes in green and orange correspond to Duffy phenotype + and –, respectively. Individuals who declared a double ethnic origin (n = 2) were excluded from these figures. Ethnic background: Black Moors (BM), White Moors (WM), Pular (PL), Wolof (WF), and Soninké (SNK).

https://doi.org/10.1371/journal.pntd.0013882.g003

Genotypes conferring a non-expression of the Duffy antigen on red blood cells (i.e., Duffy null phenotype associated with the genotypes FY*01N.01/FY*02N.01 and FY*02N.01/FY*02N.01) were overrepresented in individuals of Black African descent (ranging from 65% to 88% in Black Moors, Pular, Wolof, and Soninke, as compared to 16% in White Moors) (Fig 3). The predicted Duffy blood phenotypes were significantly associated with ethnic origins (Fisher’s exact test with Monte Carlo p-values simulation, p < 0.001).

Predicted Duffy blood group expression and G6PD deficiency in different ethnic groups

Overall, G6PD-deficient genotypes patients represented 9.8% of cases included in the present study (n = 943), with 16.6%, 5.6%, 2.9%, and 1.7% of individuals with Fy(a^–b^–), Fy(a^–b+), Fy(a + b^–), and Fy(a + b+) phenotypes, respectively (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Relationship between G6PD genotypes and predicted Duffy phenotypes in Mauritanians.

https://doi.org/10.1371/journal.pntd.0013882.t003

Analysis of G6PD deficiency genotypes by sex and ethnic groups did not show any statistically significant association with Duffy phenotypes (chi-squared test, p > 0.1). Among 432 white Moors with both Duffy and G6PD genotype data, 356 (82.4%) were Duffy-positive, of whom only 7 (2.0%) carried the African-type G6PD A^– mutations. Of the remaining 76 (17.6%) white Moors, 8 (10.5%) were Duffy-negative and affected with G6PD A^– type deficiency, mostly in heterozygous females (n = 6). Of 511 individuals of black African ancestry with both Duffy and G6PD genotype data, 140 (27.4%) were predicted to be Duffy-positive, i.e., Fy(a^–b+), Fy(a + b^–), or Fy(a + b+), and 13 (9.3%, including heterozygous females) were carriers of mutations associated with African-type G6PD A^– deficiency. Among Duffy-negative Fy(a^–b^–) individuals of black African ancestry, 65/371 (17.5%) were affected by African-type G6PD A^– deficiency, mostly (n = 44) in heterozygous females.

Discussion

The updated epidemiological data on Duffy antigen in Mauritanian populations confirm our initial hypothesis that white Moors, who are descendants of Arab and Berber ethnic groups genetically and historically related to peoples in North Africa, are predominantly Duffy-positive (84%). In a study conducted by genotyping (PCR sequencing by Sanger method) in 2007–2009 exclusively in samples collected in Nouakchott, a closely similar proportion of Duffy-positive white Moors (82.3%; total number of white Moors included, 221) was reported [44]. By contrast, another earlier study based on blood group phenotyping in Nouakchott in 1984 reported a much lower proportion of Duffy-positive white Moors in a small sample size (54%, n = 52) [43].

Mauritania straddles between the Maghreb region to the north and the Sahel to the south, and a large part of its territory is occupied by the Sahara desert. Global mapping of the distribution of FY*B allele generated from modeling has suggested that at the periphery of the region where Fy*B^ES allele predominates, as in sub-Saharan Africa, Fy*B tends to be prevalent [79]. Although this predicted model tends to agree with the results obtained in the present study, 16% of white Moors were Duffy-negative. This is a relatively high proportion of Duffy-negative phenotype in an ethnic group that is supposedly “homogeneous” and may suggest the limits of modeling at the country-level without evidence-based data and historical considerations. The populations in Africa have undergone various interactions and migrations over the past centuries, resulting in a complex human genetic structure, but the extent of population movement along the east-west and north-south axes within the African continent and also into Africa (e.g., from the current Middle East), as well as the roles played by autochthonous peoples, consanguinity, and centuries of Arab slave trade, is still being hotly debated [80–83]. Nonetheless, the cultural and linguistic ties and geographic proximity to the Maghreb are obvious factors to be taken into consideration to explain our observations concerning white Moors. For example, in Algeria, a neighboring country to the northeast of Mauritania, in sharp contrast to the peoples of Arab descent (8.9–10.4% Duffy-negative), a Berber population residing in central region of Algeria has been characterized as predominantly (i.e., 62%) Duffy-negative [84]. Similarly, in Rabat and several cities in south-central Morocco where both populations of Arab and Berber origin co-exist, the prevalence of Duffy-negative subjects, as determined by either phenotyping or genotyping, ranged from 11.1% to 24.8% [85,86]. In Tunisia, where a minority Berber ethnic group also exists, it was reported that 11.3% (13/115) of the general population is Duffy-negative [87]. It has been advanced as an hypothesis that genotypes associated with Duffy-negative phenotype in peoples of Arab descent may have originated, at least in part, from the Middle East, where high prevalence of Duffy-negative phenotype (61–78%) has been reported, instead from sub-Saharan Africa [84,88,89]. This hypothesis is supported by the historical fact regarding the Arab-Islamic conquest of the Maghreb during the seventh and eighth centuries A.D.

Data on Duffy genotypes generated in the present study need to be confronted with G6PD genotypes which were also evaluated for the same samples [40]. Only nine of 423 (2.1%) white Moors were characterized to have the African-type G6PD A^– genotype, and only two of them (0.5%), both in males, were also Duffy-negative. The source of African-type G6PD A^– genotype in white Moors living in Mauritania has not been elucidated. Although it can reasonably be argued that the African-type G6PD A^– genotype most likely arrived to Mauritania with peoples originating from sub-Saharan Africa, an alternative hypothesis that it was derived through persons in the Maghreb and in the Middle East cannot be totally refuted at present. For example, in Algeria and Morocco, two neighboring countries of Mauritania to the north, both the Mediterranean-type G6PD B^– and African-type G6PD A^– deficiencies co-exist in the patient population with clinical symptoms (anemia, favism, neonatal jaundice), with about 50% of the cases being due to the African-type G6PD A^– deficiency [90,91]. In Arab countries in the Middle East, the Mediterranean-type G6PD B^– deficiency predominates over other types of G6PD deficiencies, but in a small minority of G6PD deficient Arab patients (generally < 2%, similar to the proportion found in Mauritanian white Moors), the deficiency is associated with the African-type G6PD A^– [92–94]. However, the fact that the Mediterranean-type G6PD B^– deficiency has not been found in Mauritania argues against the hypothetical north-to-south (i.e., from the Maghreb) and east-to-west (i.e., from the Middle East) genetic flow and seems to favor the south-to-north axis, i.e., from sub-Saharan Africa. Further surveillance of Mediterranean-type G6PD B^– deficiency in Mauritania and comparative genome analysis of other genes would be required for a more conclusive evidence supporting the latter hypothesis on gene flow.

The available epidemiological data on malaria in Mauritania have indicated that P. vivax malaria is largely a disease that affects white Moors [38,39,44]. These data are in line with the findings that the presence of Duffy antigen on reticulocyte surface is one of the major requirements for P. vivax infection. Since 2014, the Mauritanian Ministry of Health recommends ACT and primaquine to treat P. vivax infection [95]. The results of the present study suggest not only that a large majority of white Moors are Duffy-positive and susceptible to P. vivax infection, but also that they are eligible for anti-hypnozoite therapy with primaquine at the standard dose. It would not be probably cost-effective to systematically screen for G6PD activity in P. vivax-infected white Moors, as in northern Mauritania where they are largely predominant. In the rare case when gross hematuria or other signs of hemolytic anemia are encountered by the white Moor patient within the first three days of primaquine treatment due to mild to moderate G6PD deficiency associated with G6PD A^– genotype (class B according to the 2024 revised WHO classification of G6PD variants) [96], drug treatment can be immediately suspended while awaiting for further laboratory examinations to adapt the treatment. The question on whether Duffy-negative white Moors are relatively more protected from P. vivax infection than Duffy-positive white Moors has not been addressed in our studies.

As for individuals belonging to ethnic groups of black African descent, who are genetically and culturally related to populations living in sub-Saharan Africa, the present study confirmed our hypothesis that a large majority of them (65–88%, depending on the ethnic background) are Duffy-negative. A previous molecular study conducted in Mauritania demonstrated that P. vivax infection occurs rarely in these populations, reinforcing the argument that Duffy antigen remains the major receptor and pathway for parasite invasion into reticulocytes [44]. Despite the mounting evidence of the occurrence of P. vivax malaria in Duffy-negative African patients elsewhere in the African continent [45–50], this phenomenon appears to be more of an exception, rather than the rule, at least in Mauritania at present. Moreover, it remains unknown whether the parasite invades reticulocytes through alternative pathways [97,98]. A recent experimental study has reported that P. vivax can invade Duffy-negative erythroblasts, some of which express functional DARC transiently during erythropoiesis (i.e., the mean of 1.0–3.2% of Duffy-negative erythroblasts between day 0 and day 12, compared to 24.6–78.3% in Duffy-positive erythroblasts) [99]. If the results of these experiments are confirmed, it can be deduced that alternative pathways are not necessary to explain the widespread presence of P. vivax in sub-Saharan Africa. These experimental findings are also in agreement with the clinical observation that P. vivax occurs at low parasitemia due, not only to the low number of circulating reticulocytes under physiological conditions in the human host, but also to the low number of erythroblasts which transiently express DARC in Duffy-negative individuals [99].

About one-fifth (65/337; 19.3%) of Duffy-negative individuals of black African ancestry were affected by G6PD deficiency, including heterozygous females whose G6PD phenotype cannot be predicted with accuracy from the genotype due to random X chromosome inactivation unless the enzymatic activity is measured by spectrophotometry or an alternative, reliable field-compatible tool [31,33,100,101]. However, these Duffy-negative, G6PD-deficient individuals are likely to be relatively protected from P. vivax malaria and are expected to require primaquine therapy only on rare occasions. The patients who may potentially encounter a problem with primaquine treatment for P. vivax infection are those who are both Duffy-positive and G6PD-deficient. Our results suggested that 7 of 432 (1.6%) white Moors and 13 of 511 (2.5%) Mauritanians of black African ancestry, i.e., a total of 20 of 943 patients (2.1% of our patient population, including both sexes), fall into this at-risk category. Faced with a laboratory-confirmed diagnosis of P. vivax, systematic screening and determination of the level of G6PD enzymatic activity are required before administering primaquine for ethical and safety reasons [2,29–33].

In addition to Duffy antigen and G6PD, the hepatic cytochrome P-450 isozyme 2D6 (CYP2D6) plays a crucial role in the expected efficacy of anti-relapse therapy. Primaquine is a pro-drug which requires hepatic transformation to its active metabolite, hydroxy-primaquine. Some CYP2D6 gene polymorphisms result in null or impaired drug metabolism, leading to low plasma concentration of the active metabolite and failure to prevent relapse in “poor metabolizers” [102–105].

About 40% of the population exposed to the risk of P. vivax infection in southeast Asia, where relevant data are available, is not eligible for the standard anti-relapse treatment with primaquine due to G6PD deficiency and may face treatment failure due to poor drug metabolism [103]. There are no data on CYP2D6 polymorphisms in Mauritanian populations. Data in the literature suggest that the lowest proportion of populations with a normal CYP2D6 activity is found in East/Southeast Asia and Africa [106]. Among the studied populations, 56% in West Africa (49–60% in sub-Saharan Africa) and 32% in North Africa (30–32% in Algeria and Morocco, two neighboring countries to the north of Mauritania) were characterized to be either null or impaired metabolizers, of whom 15–21% of the individuals in West and North Africa (15% in Algeria and 21% in Morocco) had null metabolism [106–109]. If we assume that Mauritanian white Moors are genetically closely related to Algerian and Moroccan populations, about 69% of them are expected to have a normal CYP2D6 activity. Based on this assumption and our data on Duffy and G6PD, it can be predicted that, after G6PD screening is performed, the proportion of P. vivax-infected white Moor patients eligible for the standard anti-hypnozoite primaquine therapy is 56.6% in males and 56.8% in females. Our data are in agreement with the earlier prediction made in Asian populations that about 40% of the population potentially exposed to the risk of P. vivax infection (43% of male and female white Moors) would not benefit from the standard anti-relapse treatment with primaquine due, in part, to G6PD deficiency and, more importantly, to poor drug metabolism [103].

Several limitations of the present study should be noted. First, only a limited number of mutant alleles were assessed. Duffy genotypes are highly complex [62,63], and many “minor” mutations in the ACKR1 gene that can occur and lead to silent Duffy phenotype were not explored. Likewise, more than 200 mutations in the gene coding for G6PD have been described, some of which can result in different levels of enzyme deficiency [34]. Only the major G6PD genotypes associated with either African-type or Mediterranean-type deficiency were analyzed in our previous studies [33,40]. Second, PCR sequencing failed in a number of dried blood spots, most likely due to DNA degradation over time. The small quantity of available samples did not allow repetition of PCR amplification in case of PCR failure. However, the exclusion of some samples most probably did not introduce an important bias in sampling as all samples were subject to similar conservation conditions in the field. Third, data generated from the present study are not representative of the general population. Many of our blood samples were obtained from patients with symptoms suggestive of malaria. In the study site such as Atar, where white Moors predominate, patients infected with malaria or presenting with signs and symptoms associated with malaria are more likely to be white Moors, who tend to be overrepresented in the study area. In addition, no attempt was made to obtain a representative sample of each ethnic group because the Mauritanian government does not allow for a census identifying ethnic groups and the current ethnic composition of the country’s population is unknown. Fourth, data analysis by ethnic origin and study sites was not performed due to the tendency of some ethnic groups to assemble in certain regions of the country, with the exception of the capital city, the cosmopolitan melting pot of Mauritania. Fifth, at present, the original source of Duffy-negative and African-type G6PD A^– genotypes found in Mauritania today cannot be determined. Analysis of other gene markers, together with a more complete understanding of the history of human migration and interactions between human populations, would be necessary to gain a better knowledge of gene flow in Africa. Lastly, perhaps the most important limitation of the study from the clinical and epidemiological viewpoint is our inability to answer the question concerning the level of susceptibility of Duffy-negative white Moors and Duffy-positive individuals of black African ancestry to P. vivax malaria. Future studies designed to specifically answer these questions would be required.

Conclusions

The novel multiplex amplicon-based DNA sequencing proved to be a rapid, high-throughput method to characterize Duffy genotype based on 11 major mutations in the present study, compared to the more labor-intensive PCR-RFLP or PCR sequencing using the Sanger dideoxy method to characterize G6PD genotype in our earlier works [33,40]. The novel methodological approach is useful for epidemiological studies. Further improvement of the multiplex assay for genotyping simultaneously Duffy and G6PD variants would enhance diagnostic efficacy. The results of the study showed two major populations: (1) mostly Duffy-positive white Moors with normal G6PD activity on one side of the spectrum and (2) mostly Duffy-negative ethnic groups of black African ancestry, some of whom presenting the African-type G6PD A^– deficiency, on the other side of the spectrum. The former is susceptible to P. vivax infection due to Duffy-positive phenotype, but most of them can be treated safely with primaquine; the latter is “resistant” to P. vivax malaria but, in case of infection, requires screening for G6PD activity before primaquine therapy. In between these two extremes are few individuals (Duffy-negative white Moors and Duffy-positive individuals of black African origin) who also require G6PD screening to ensure the safe administration of primaquine therapy. From the epidemiological viewpoint, the data presented here may be useful to identify priority areas in Mauritania to implement targeted interventions with primaquine, particularly in the southern Sahelian and Sahelian-Saharan transition zones.

Supporting information

Acknowledgments

We thank Igor Filipović and Gordana Rašić for their help in the implementation of the multiplexing scheme. They designed the six nucleotide barcoding used in this study that was optimized to be informative even in the case of trimming in the first few bases of the sequence.