Introduction

Anaemia (haemoglobin [Hb] < 110 g/l) is a serious public health problem affecting more than half of children less than 5 y of age in malaria-endemic countries of Africa; in a survey in The Gambia 76% of children were anaemic [1]. Anaemia has multiple causes, including malaria and micronutrient deficiencies [2]. In sub-Saharan Africa 20% to 40% of children have undetectable levels of haptoglobin (Hp) due to haemolysis, and low levels of haptoglobin are strongly associated with malaria infection [3]. Hp, an acute-phase plasma protein, is characterised by its strong binding affinity (>10¹⁰ mol⁻¹) for free haemoglobin released following haemolysis. The Hp¹ and Hp² alleles are encoded by a single gene on Chromosome 16; the Hp² allele was formed from an intragenic duplication originating from a nonhomologous crossing-over of two Hp¹ alleles [4]. Haptoglobin exists in three common phenotypes: the homodimer Hp1-1, the linear polymer Hp1-2, and the large circular polymer Hp2-2 [5]. Clear functional differences exist between the phenotypes, including differences in modulation of oxidant stress, recycling of haem-iron, and immune function [5]. Haemolytic stress is likely to accentuate differences between the phenotypes, as has been found in haptoglobin knock-out mice compared to wild-type mice [6].

The Hp2-2 polymer has very different biochemical and biophysical properties compared to Hp1-1 and Hp1-2 [5]. The Hp2-2 protein is present in lower concentrations [7] and binds less efficiently to free haemoglobin; Hp1-1 and Hp1-2 have higher binding affinities [8]. These differences are reflected in vivo by altered oxidant defence and iron handling. Vitamin C concentrations were significantly lower in the Hp2-2 phenotype (49.9 μmol/l), but did not differ between Hp1-1 and Hp1-2 individuals (61.5 μmol/l and 63.7 μmol/l respectively) [9]. Furthermore, in Hp2-2 individuals iron is delocalised into poorly exchangeable storage compartments of the mononuclear phagocytic system; excess monocyte iron was found in the Hp2-2 phenotype (687 μg/g L-ferritin) compared to the Hp1-1 and Hp1-2 phenotypes (326 μg/g and 366 μg/g L-ferritin, respectively) [10]. However, it is not known whether Hp2-2 phenotype might be a risk factor for anaemia in an environment of malaria-induced haemolysis and limited dietary iron availability.

The pathogenesis of malarial anaemia is complex and includes haemolysis, accelerated erythrophagocytosis, and cytokine-induced dyserythropoiesis [11, 12]; reduced antioxidant defence [13, 14] and a shift in iron distribution from functional to storage compartments [15, 16] have also been found. The association between haptoglobin and susceptibility to malaria infection is controversial, and published reports are conflicting. The Hp2-2 phenotype was associated with protection from severe malaria and placental infection in a number of case control studies [17–20]. However, a large study analyzing haptoglobin genotypes and a recent case control study did not find an association [21, 22].

We hypothesised that the Hp2-2 phenotype, due to known functional differences, may be a risk factor for anaemia in children in an environment of malaria and relative iron deficiency. We thus investigated the possibility that children homozygous for the Hp² allele would have a lower haemoglobin level after the malaria season. To control for the multifactorial aetiology of anaemia we measured baseline haemoglobin levels in the same children prior to the malaria season, and other factors that might influence haemoglobin level. We explored the interaction with plasma iron markers and compared the impact of haptoglobin with other genetic polymorphisms that might influence haemoglobin levels over the malaria season; namely, HbS and the G6PD A and A− variants common to sub-Saharan Africa [23].

Methods

Patients and Methods

Ethical permission for the study was granted by the Gambian Government and Medical Research Council Ethics Committee, and Gambian National DNA Collection Guidelines were followed regarding the handling of genetic material and information. Parental written informed consent was obtained for all study participants.

A cohort of 780 children aged from 2 to 6 y was recruited from ten rural villages in the West Kiang region of The Gambia at the start of the malaria season, July 2001, with follow-up to December 2001/January 2002. All children were eligible except those with serious chronic illness or those enrolled in another study. Figure 1 provides an overview of the study population. Ethnic groups were Mandinka (nine villages, 700 children) and Fulani (one village, 80 children). Children had anthropometric measurements taken and were examined by the study clinician. All children received a 3-d course of mebendazole at the start of the study for possible hookworm infection. A blood sample was collected for full blood count, malaria slide, iron status assays, haptoglobin concentration, α-1-antichymotrypsin (a marker of inflammation), and DNA extraction. Children with a temperature over 37.5 °C had a malaria blood film, appropriate clinical treatment, and a blood sample 2 wk later after recovery from illness. Children with malaria parasites on blood film were treated with chloroquine and pyrimethamine-sulfadoxine (Fansidar) according to Gambian Government guidelines. This procedure was repeated at the end of the malaria season for each child.

Download:

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

Figure 1. Sample Construction

At the start of the malaria season, 780 children (aged 2–6 y) were recruited from ten rural Gambian villages, of these, 61 children were lost to follow-up (of whom four died). At the end of the malaria season, 707 children were surveyed. A total of 671 children had complete haemoglobin, malaria blood film, and haptoglobin genotype data. After biochemical assays and further genotyping, 565 children had complete data for multivariate regression analysis.

https://doi.org/10.1371/journal.pmed.0030172.g001

Malaria incidence is highly seasonal in The Gambia with the majority of malaria cases occurring between September and December [24]. Haemoglobin levels in children from the study area were previously found to be highest in July and lowest in November [25]. We thus sampled at the start and end of the malaria season to assess the effect of haptoglobin genotype on haemoglobin levels in the malaria season compared to baseline levels. Children were followed across the malaria season to control for multiple individual factors that may influence haemoglobin levels.

Results

Characteristics of Study Population

Haptoglobin genotypes were in Hardy-Weinberg equilibrium. Figure 1 shows numbers of children lost to follow-up and those with pyrexia who were sampled 2 wk later following clinical management; haptoglobin genotype distribution did not differ in either of these groups of children. Four children died; cause of death was severe malarial anaemia in two (one of whom had sickle cell disease), cerebral malaria in one, and the cause was unknown in the fourth. Binomial regression analysis did not reveal significant geographic clustering of the Hp² allele by village, and no association was found between haptoglobin genotype and either sickle genotype or G6PD deficiency. Table 1 summarises the characteristics of the sample by haptoglobin genotype.

Download:

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

Table 1.

Characteristics of Study Population by Haptoglobin Genotype

https://doi.org/10.1371/journal.pmed.0030172.t001

Univariate Analysis

Overall, mean haemoglobin levels fell across the malaria season; from 106.9 g/l (95% confidence interval [CI_95%] 105.9,108.0) to 101.2 g/l (CI_95% 99.9,102.4) by the end of the malaria season; a mean drop of 5.8 g/l (CI_95% 4.5,7.0) ( p = 0.0001). In primary analysis, children carrying the Hp^2/2 genotype had a mean drop in haemoglobin of 8.9 g/l (CI_95% 5.7,12.1) compared with 5.1 g/l (CI_95% 3.8,6.4) for the other genotypes ( p = 0.02 by Student's t-test). The drop in haemoglobin was similar for the Hp^1/1 and Hp^1/2 genotypes (5.9 g/l and 4.6 g/l respectively). In secondary analysis an ANOVA for the three genotypes also indicated a greater drop in mean haemoglobin level in children with the Hp^2/2 genotype ( p = 0.045). By comparison, sickle genotype and G6PD deficiency did not significantly influence drop in haemoglobin level over the malaria season. Drop in haemoglobin level over the malaria season for haptoglobin, HbAS, and G6PD A type deficiency is shown in Figure 2. Baseline haemoglobin levels did not differ in the haptoglobin genotypes, but were lower by the end of the malaria season in children carrying the Hp^2/2 genotype ( p = 0.05 in univariate analysis, Hp^2/2 versus Hp^1/2 and Hp^1/1 combined). As found previously [7], children carrying the Hp^2/2 genotype had lower mean haptoglobin concentrations. We also found an increased number of asymptomatic children with malaria parasitaemia in the Hp^2/2 group at the end of the malaria season ( p = 0.02 for Hp^2/2 versus Hp^1/1 and Hp^1/2 combined, p = 0.06 for heterogeneity among the three genotypes).

Download:

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

Figure 2. Drop in Haemoglobin Level during the Malaria Season according to HbAS, G6PD Deficiency, and Haptoglobin Genotype

Haemoglobin levels at the start and end of the malaria season are shown by haptoglobin genotype ( Hp^2/2 versus Hp^1/1 and Hp^1/2 combined), HbAS compared to HbAA and G6PD (A type) deficiency, wild-type, heterozygotes, and homozygotes/hemizygotes. Error bars denote standard error of the mean.

https://doi.org/10.1371/journal.pmed.0030172.g002

Multivariate Regression Analysis

Multiple factors influenced haemoglobin level. A multivariate regression analysis was performed with haemoglobin level at the end of the malaria season as dependent variable. In this model, baseline haemoglobin level, village, age group, malaria parasites on blood film, iron status, haptoglobin concentration (a measure of haemolysis in the last 10 d), G6PD A type deficiency, and haptoglobin genotype ( p = 0.0009, coefficient = −4.2) emerged as significant predictors of haemoglobin level at the end of the malaria season (Table 2). Sex, baseline iron markers (log ZnPP and log soluble transferrin receptor), weight for height Z-scores, α₁-antichymotrypsin levels, and sickle genotype were not retained in the model, as the p-value exceeded 0.1. In a secondary analysis, p = 0.002 for heterogeneity among the three haptoglobin genotypes. The Hp^1/1 and Hp^1/2 genotypes did not differ (1 g/l difference, p = 0.32). A further multivariate analysis with change in Hb as the dependent variable found a similar effect for haptoglobin genotype ( p = 0.007, coefficient = −4.1). Iron status (as measured by ferritin, zinc protoporphyrin, transferrin saturation, and soluble transferrin receptor concentration) strongly influenced haemoglobin level, but was not significantly altered by haptoglobin genotype.

Download:

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

Table 2.

Multivariate Regression Analysis for Haemoglobin Level at the End of the Malaria Season

https://doi.org/10.1371/journal.pmed.0030172.t002

Discussion

Multiple factors contributed to the complex aetiology of anaemia in young rural Gambian children; both iron status and malaria infection emerged as strong predictors of haemoglobin level. We speculated that haptoglobin, encoded by a gene that regulates haemoglobin-iron metabolism after haemolysis, might critically influence haemoglobin levels in an environment of malaria-induced haemolytic stress and iron deficiency. Haemoglobin levels fell generally over the malaria season, but this effect was concentrated in children carrying the Hp^2/2 genotype (who had approximately twice the drop in Hb of children carrying the Hp^1/2 genotype). Baseline haemoglobin levels were similar by Hp genotype, but by the end of the malaria season differed significantly in the Hp^2/2 genotype (17% of children) compared to the Hp^1/1 and Hp^1/2 genotypes (coefficient = −4.2, p = 0.0009 in a multivariate regression analysis). The effect of Hp^2/2 was greater than that of previously described genetic variants such as G6PD A type deficiency (3 g/l difference, p = 0.03) and HbAS (2 g/l difference, p = 0.68). The difference by haptoglobin genotype was robust to adjustments for other factors influencing haemoglobin level and is also supported by our retrospective analysis of a cross-sectional study of haptoglobin phenotypes conducted in this population in the 1970s [29]. This found that the Hp2-2 phenotype was associated with haemoglobin levels of less than 120 g/l (odds ratio = 1.92, p = 0.028, n = 825). A study from Melanesia has also reported similar differences in the 0- to 4-y age group in a malaria-endemic area (albeit based on a small sample size, n = 41) [30].

So how might the Hp^2/2 genotype result in anaemia under conditions of malaria-induced haemolysis and iron deficiency? We propose three possible explanations for impaired haematological recovery from malaria infection in the Hp^2/2 genotype. These mechanisms need not be mutually exclusive.

Firstly, impaired iron recycling in the Hp^2/2 genotype due to delocalisation of iron into poorly exchangeable macrophage storage compartments may result in iron-deficient erythropoiesis. The haemoglobin-scavenging macrophage receptor CD163 has greater functional affinity for the Hp2-2-haemoglobin complex compared to the Hp1-1-Hb complex [31] resulting in significantly increased monocyte iron trapping in Hp2-2 patients [10]. We therefore hypothesized that prolonged or recurrent episodes of malaria would result in less efficient recycling of haemoglobin iron and iron-deficient erythropoiesis. However, we did not find an association between haptoglobin genotype and markers of iron status; it is possible, though, that monocyte iron levels may have differed between the genotypes. The association between iron homeostasis and haptoglobin genotype is controversial; some studies have found that the Hp^2/2 genotype is associated with iron loading [10, 32], and others have not found an association [33–35]. Further studies are in progress to assess the associations between haptoglobin genotype, monocyte-iron levels, and malaria infection.

Secondly, the Hp^2/2 genotype may influence haematological status by failing to quench haemoglobin-iron-mediated oxidant stress. In vitro work found that the Hp 2-2 protein was associated with reduced inhibition of oxidation of low-density lipoprotein (LDL), increased redox active iron and increased oxidant stress levels compared to the Hp1-1 protein [36, 37]. These effects have been confirmed in vivo; Hp2-2 individuals have reduced vitamin C levels [9, 38], reduced ferroxidase activity if smokers [39], and significantly higher oxLDL/LDL ratios if male [40]. Malarial anaemia is associated with a reduction in red cell membrane antioxidants [13] and increased markers of oxidant stress [41]. Moreover, evidence suggests that oxidant damage to the red cell membrane leads to accelerated erythrophagocytosis via aggregation of band 3 proteins and binding of autologous IgG and complement [42, 43].

A third possibility is modulation of inflammatory cytokine balance. Haptoglobin promotes Th1 over Th2 activation in mice experiments [44] and suppresses monocyte production of tumour necrosis factor-α, IL-10, and IL-12 in response to lipopolysaccharide [45]. It is speculated that cross-linking of the macrophage CD163 receptor by the multimeric Hp2-2-haemoglobin complex mimics antibody binding and triggers a signalling cascade resulting in increased secretion of anti-inflammatory cytokines [31, 46]. In support of this idea, cross-linking of the CD163 receptor by haptoglobin-haemoglobin complexes was found to strongly stimulate IL-10 secretion in cultured macrophages [47]. Additionally, macrophage iron loading, as might be seen in the Hp^2/2 genotype, is also associated with a reduced respiratory burst and decreased nitric oxide, IFNγ, and TNF production [48]. The significance of these possible effects remains to be clearly elucidated but it is possible that by dampening the inflammatory response Hp2–2 individuals may increase the chronicity of malaria infections.

The Hp² allele is thought to have spread under strong genetic pressure, and haptoglobin disease associations are reviewed in Langlois and Delanghe (1996) [5]. The strongest disease pressure in this age group in rural Africa would be from malaria. Malaria infects more than 50 million children each year, and in many parts of Africa the average child has several malaria infections a year and parasites in the blood almost continuously [49]. Furthermore, the Hp^2/2 genotype, present in 17% of children in our study, is also common. Whether there is an association between haptoglobin polymorphisms and susceptibility to malaria infection remains controversial; studies indicate either a protective or neutral effect of Hp^2/2 genotype [17–22]. We found an increase in asymptomatic malaria parasitaemia in the Hp^2/2 genotype at the end of the malaria season ( p = 0.02 for Hp^2/2 versus Hp^1/2 and Hp^1/1 combined). A further possibility is that the Hp^2/2 genotype might be associated with asymptomatic, chronic parasitaemia that contributes to anaemia, but may also provide protection against severe malaria.

In summary, our results suggest that the Hp^2/2 genotype is a risk factor for childhood anaemia in malaria-endemic countries. The high prevalence of a potentially detrimental allele might be explained by balancing selection pressures. An intriguing possibility is that the Hp^2/2 genotype may protect against life-threatening malaria [18, 19] at the expense of impaired haematological recovery from mild and asymptomatic malaria. The effect of haptoglobin genotype in other haemolytic diseases, such as sickle cell, is yet to be elucidated. Further research is underway to reconfirm this finding and to investigate the potential mechanisms of anaemia.

Supporting Information

Acknowledgments

The Medical Research Council was not involved in study design or preparation of the manuscript. We would like to thank the children who took part in this study and their parents. We would also like to thank Baba Jobarteh, Sosseh Sanyang, Kabiru Ceesay, Musa Colley, Khalilu Sanneh, and Ibrima Camera for their assistance in the field and laboratory. We thank Christophe Aucan and Chris Bates for helpful discussions and assistance regarding the haptoglobin and iron assays, respectively.