Hemoglobin C associated with protection from severe malaria in the Dogon of Mali, a West African population with a low prevalence of hemoglobin S

Aarti Agarwal, Aldiouma Guindo, Yacouba Cissoko, James G. Taylor, Drissa Coulibaly, Abdoulaye Koné, Kassoum Kayentao, Abdoulaye Djimde, Christopher V. Plowe, Ogobara Doumbo, Thomas E. Wellems and Dapa Diallo


The malaria hypothesis proposes a survival advantage for individuals with hemoglobin variants in areas of endemicPlasmodium falciparum malaria. Hemoglobin C (HbC) is a possible example in West Africa, where this hemoglobin has a centric distribution with high frequencies among certain populations including the Dogon ethnic group. To test whether HbC is associated with protection from malaria, we performed a case-control study in the Dogon of Bandiagara, Mali. HbC was present in 68 of 391 (17.4%) of uncomplicated malaria control cases, whereas it was detected in only 3 of 67 cases (4.5%) of severe malaria (odds ratio [OR], 0.22;P = .01). Further, HbC was present in only 1 of 34 cases (2.9%) with cerebral manifestations, the most common presentation of severe malaria in this population (OR, 0.14; P = .03). Episodes of uncomplicated malaria and parasitemias (4800-205 050/μL) were identified in cases of homozygous HbC (HbCC), which indicates thatP falciparum parasites are able to efficiently replicate within HbCC erythrocytes in vivo. These findings suggest that HbC does not protect against infection or uncomplicated malaria but can protect against severe malaria in the Dogon population of Bandiagara, Mali. The data also suggest that the protective effect associated with HbC may be greater than that of HbS in this population.


More than 50 years ago, Haldane1suggested that protection against malaria was conferred to individuals by a form of thalassemia. Since that time, the observed overlaps in the geographic distributions of malaria with hemoglobinopathies and other red blood cell disorders have been cited in support for the hypothesis that malaria has been an important evolutionary force in selection of these variants.2 Epidemiologic and in vitro support for the malaria hypothesis is best documented for the thalassemias3-5 and sickle hemoglobin (HbS),6-10 which in different regions of Africa, the Middle East, and Asia are maintained in a state of balanced polymorphism because of the protective effect of the heterozygous state against severe disease. Hemoglobin E (HbE) has also been associated with a reduced prevalence of severe Plasmodium falciparummalaria, with the odds of severe complications being 15%-20% of those found for homozygous hemoglobin A (HbA).11

The malaria hypothesis has been invoked to explain the observed frequency of HbC in certain populations of West Africa, although epidemiological evidence has been contradictory.12-16 One study suggested an increased immunoglobulin response in individuals heterozygous for HbC.17 The presence of heterozygous HbC, however, was not found to correlate with reductions of P falciparum infections or parasite densities.12-14 17 18 Further, 2 clinical studies involving small numbers of HbC-trait individuals in some ethnic groups did not show significant evidence of reduced rates of severe disease.15 16 Although individuals homozygous for HbC (HbCC) were not observed in these studies, thereby leaving the possibility of protection from malaria by the HbCC state unresolved, these findings have not supported the possibility of a balanced polymorphism in which positive selection by malaria for HbC may counter negative selective pressures from the morbidity of the HbSC state,19 the occasional pathologic developments in the HbC homozygous state,20 and a possible reduced reproductive fitness of the βCgene.18 21

The HbC and HbS mutations both occur in the 6th position in the β-globin amino acid sequence (β-6), where glutamic acid is substituted by lysine or valine, respectively. Although there are distinct pathologic consequences, the remarkable fact that both substitutions occur at β-6 may reflect a common influence of position on the effects of these different mutations in vivo. To test the hypothesis that HbC protects against severe malaria, we conducted an unmatched case-control study in the predominantly Dogon population of Bandiagara, Mali, where the gene frequency of HbC is high (0.087) and that of HbS is low (0.016).22 23 Here we report data that provide evidence for an association between HbC and protection against severe malaria in this Dogon population. The data also suggest less selection for the HbAS state in this group than for HbAC.

Materials and methods


Bandiagara, a village situated on a rocky plain above the Dogon escarpment in Mali, has a population of about 12 000 and mesoendemicP falciparum malaria with an intense seasonal occurrence from July to December. The present study was conducted during the malaria transmission seasons of 1997 and 1998 and was carried out in collaboration with local village leaders and traditional medical healers. Ethical approval for the study was obtained from the institutional review boards of the National Institute of Allergy and Infectious Diseases, Bethesda, MD; the University of Maryland, Baltimore, MD; and the University of Mali, Bamako, Mali. The patients were evaluated and enrolled at the Bandiagara Traditional Medicine Center.

Diagnosis and classification of cases

Inclusion and classification of each case were based on the symptoms, physical signs, and laboratory findings of malaria at the time of presentation. Severe malaria was established by microscopic confirmation of asexual P falciparum parasites in the peripheral blood and the following modification of established World Health Organization (WHO) criteria24: evidence for cerebral compromise (coma, obtundation, or witnessed convulsions without other explanation); severe anemia (hematocrit less than 15%); hyperparasitemia ofEp  > 5 × 105/μL; hypoglycemia (serum glucose less than 2.2 mmol/L (40 mg/dL); and respiratory distress, clinical shock, or severe prostration, as detailed by Marsh et al.25 Uncomplicated malaria was established by microscopically confirmed parasitemia ofEp  < 5 × 105/μL with fever, headache, myalgias, or gastrointestinal symptoms and without any findings of severe malaria. Splenomegaly in this study was defined as a Hackett score of at least 1.26

Patients with uncomplicated malaria and parasitemia ofEp  < 1 × 105/μL (control cases) were treated with chloroquine in an outpatient oral drug efficacy study (C. V. Plowe et al, unpublished data, 2000). Patients with severe malaria or parasitemia ofEp  > 1 × 105/μL were placed under intensive care with treatment that included parenteral fluids, quinine, glucose, and antibiotics, as judged appropriate.

Hemoglobin typing

Hemoglobin types were determined by analysis of red blood cell lysates using cellulose acetate electrophoresis at an alkaline pH. Cellulose acetate plates (Helena Laboratories, Beaumont, TX) were stained with Ponceau red (Geimsa, St Louis, MO), yielding clear discrimination of hemoglobins A, F, S, and C/A2. Although HbC comigrates with HbA2 during electrophoresis, the presence of the HbC-trait was evident by its relative abundance in heterozygous and homozygous individuals. Homozygosity for HbC was also confirmed retrospectively for 1 of 7 cases from a frozen sample of preserved erythrocytes by citrate-agar electrophoresis (Helena Laboratories) and quantitative measurements of hemoglobins F, A2, and C by cation-exchange high-precision liquid chromatography (HPLC) using the variant β-thalassemia short program (Bio-Rad Laboratories, Hercules, CA). HbAS samples identified by electrophoresis were confirmed with metabisulfite sickling tests.

Statistical analysis

Demographic, clinical, and treatment profiles were recorded on case record forms, transferred to computer files, and analyzed by standard software including EpiInfo 6.0 (the Centers for Disease Control and Prevention, Atlanta, GA), Instat 2.03 (GraphPad Software, San Diego, CA), and Microsoft Excel (Microsoft, Seattle, WA). Parasitemias were log-transformed for statistical analysis. The controls for all comparisons were defined as uncomplicated malaria cases with parasitemia ofEp  < 1 × 105/μL. We used the 2-tailed Fisher exact test, 2-tailed Student and Welch ttests, Mann-Whitney U (Wilcoxon rank sum) test, one-way analysis of variance (ANOVA), or Kruskal-Wallis nonparametric test as appropriate. Comparison statistics included the Mantel-Haenszel weighted odds ratio (OR), exact 95% confidence limit (CL), Mantel-Haenszel summary chi-square, and P values.P < .05 was considered statistically significant. Statistical power for comparisons was calculated using the observed OR and a 1-sided type I (α) error rate of 0.05.


During the 2 transmission seasons of 1997 and 1998, the study team evaluated 3645 patient presentations for symptoms of illness. Of these presentations, 789 cases fit the definition of uncomplicated malaria, and 89 cases met the criteria for severe malaria. The ethnic representation among all malaria cases was 63% Dogon, 13% Peulh, and 24% other ethnic groups including Bambara, Ouolof, and Sonrhai. Approximately equal percentages of the uncomplicated and severe malaria cases had a record of self-treatment with antimalarial remedies prior to presentation to the Medicine Center (44% and 48%, respectively, with this information available for 213 cases). Of the 789 uncomplicated malaria cases, 50% were male and 50% were female, whereas 55% of the 89 severe cases were male and 45% were female.

Among the 89 severe malaria cases, 67 occurred in the Dogon (Table1). The 22 cases in other ethnic groups were not of sufficient number to independently test for a significant association between HbC and a reduced prevalence of severe malaria, and these cases are not detailed in this report. As expected, patients with severe malaria were significantly younger and more anemic than controls with uncomplicated malaria. Of the 67 severe Dogon cases, 34 cases (51%) presented with cerebral manifestations, 12 (18%) with severe anemia, 18 (27%) with hyperparasitemia (Ep  > 5 × 105/μL), 6 (9%) with respiratory distress or shock, and 6 (9%) with severe prostration. Different manifestations of severe malaria thus overlapped at presentation in 9 of the 67 cases. The severe malaria cases in the Dogon were more likely to be from rural areas surrounding Bandiagara than were the uncomplicated malaria control cases (28% vs 6%, respectively; P < .000 000 1) (Table 1). Complete residential information was not available for 16% of the study population.

Table 1.

Clinical and laboratory features of the Dogon malaria cases

As a baseline for analysis of hemoglobin type prevalences in this study, we first compared the distribution of hemoglobin types in the Dogon cases of uncomplicated malaria with those reported from an extensive survey of hemoglobinopathies in the Dogon country.22 23 The results of this comparison showed that the reported prevalences of HbAA, HbAC, HbCC, and HbAS were indistinguishable from the prevalences of 80.0% HbAA, 15.9% HbAC, 1.5% HbCC, and 2.6% HbAS found in the Dogon cases of uncomplicated malaria (chi-square test with 3 degrees of freedom,P = .95) (Table 2). Estimated gene frequencies from the above observations in the uncomplicated malaria controls (βA  = 0.892,βC  = 0.095, andβS  = 0.013) and in the overall study population among the Dogon (βA  = 0.901,βC  = 0.084, andβS  = 0.015) were calculated to be consistent with frequencies of the Hardy-Weinberg equilibrium27(chi-square test with 5 degrees of freedom, P = .80 for controls; chi-square test with 5 degrees of freedom,P = .61 for all Dogon cases).

Table 2.

Distribution of hemoglobin types in uncomplicated malaria control cases among the Dogon

Table 3 presents the distribution of hemoglobin types among the severe and uncomplicated cases of malaria in the Dogon. The percentage of cases with HbC (ie, HbAC and HbCC) in cases of severe malaria (4.5%) was significantly lower than the percentage in uncomplicated control cases (17.4% = 15.9% + 1.5%; OR, 0.22; 95% CL, 0.04-0.74;P = .01). The 7 homozygous HbCC cases all presented with uncomplicated malaria.

Table 3.

Distribution by hemoglobin type in cases of uncomplicated and severe malaria

The prevalence of HbAS was not found to be significantly different in the Dogon groups with severe and uncomplicated malaria (OR, 1.91; 95% CL, 0.31-8.92; P = .59) (Table 3). Cerebral manifestations were identified in 2 of the 3 severe malaria HbAS cases.

Comparisons of mean parasitemia by hemoglobin type relative to HbAA among the Dogon population are presented in Table4. Among the uncomplicated malaria control cases, uncomplicated cases withEp > 1 × 105/μL, or severe malaria cases, there was no significant difference in mean parasitemia between the HbAC or HbAS cases and the HbAA cases.

Table 4.

Comparison of mean parasitemias by hemoglobin type among the Dogon

Table 5 presents the results of the calculated ORs and P values for the prevalences of HbC and HbS in the severe malaria syndromes among the Dogon cases listed in Table 3. In this analysis, the total prevalence of HbC (HbAC plus HbCC presentations) was found to be significantly reduced in the cerebral manifestations group (OR, 0.14; 95% CL, 0.00-0.89;P = .03). The numbers of cases were not sufficient to determine whether the prevalence of HbC differed significantly in severe anemia, hyperparasitemia, respiratory distress or shock, or severe prostration.

Table 5.

ORs from the prevalences of hemoglobins C and S in severe malaria syndromes among the Dogon

The Dogon presenting for this study showed a somewhat higher prevalence of severe malaria than the Peulh and other ethnic groups (data not shown). Residence data indicated a possible catchment effect because the proportion of Dogon severe cases from rural areas (28%) was significantly higher than the proportion of uncomplicated control cases from rural areas (6%; P < .000 000 1) (Table 1). To reduce a possible bias resulting from the selective referral of very ill children from remote, predominately Dogon villages outside Bandiagara, we compared the hemoglobin prevalences in urban severe malaria cases against those in randomly matched urban uncomplicated malaria control cases. Although this smaller sample size reduced the statistical power for this comparison from 73% (type II error of 0.27) (Table 3) to 53% (type II error of 0.47), the results remained consistent with protection against severe malaria by HbC (OR, 0.22; 95% CL, 0.02-1.00; P = .06) (Table6).

Table 6.

Age-matched comparison of severe malaria among urban Dogon cases

We also analyzed the prevalence of splenomegaly in uncomplicated malaria controls and severe malaria cases by hemoglobin type (Table7). Splenomegaly was found to be significantly more frequent in cases of severe malaria in the Dogon (P < .001) (Table 7). No instances of splenomegaly were detected in the 7 CC cases presenting in this study.

Table 7.

Prevalence of splenomegaly by hemoglobin type among the Dogon cases


The results of this study indicate that HbC is associated with protection against the severe form of P falciparum malaria in the Dogon of Bandiagara, Mali. In this ethnic group, the prevalence of HbC was significantly lower among cases of severe malaria than among cases of uncomplicated malaria. The protective effect indicated by the calculated OR of 0.22 would correspond roughly to an 80% reduction in the risk of severe malaria. As 16% of the Dogon population in Bandiagara carry at least 1 βC allele, the protective potential of this association (approximately 13% potential severe cases prevented28) would suggest an effect in the Dogon that is comparable to that associated with HbS in other populations.8

HbCC individuals with P falciparum infections were readily found, including 1 case withEp  = 205 050/μL, a parasitemia that corresponds to approximately 5% infected erythrocytes in the bloodstream. These findings show that P falciparum can infect and proliferate within the bloodstream of the HbCC homozygote even though P falciparum parasites have been reported not to proliferate in HbCC erythrocytes under in vitro conditions.29 30 None of the 7 HbCC cases manifested severe or life-threatening complications of malaria.

This study was not designed to determine whether HbC was associated with a protection against uncomplicated malaria. However, the fact that the prevalence of HbC in the malaria control group is similar to that observed in the Dogon community does suggest lack of a major effect from HbC on parasite infection and uncomplicated malaria. Prevalences of 15.9% HbAC and 1.5% HbCC in the uncomplicated malaria control cases compare closely to reported average rates of 14.9% HbAC and 1.1% HbCC in the Dogon population.22 23 These findings further suggest a parallel between HbC in the Dogon and HbS in other ethnic groups, which in the heterozygous state protects not so much against the occurrence of P falciparum infection and uncomplicated malaria as against the severe forms of the disease.8 In the cases of uncomplicated malaria among the Dogon, mean parasitemias among the HbAC, HbCC, or HbAS hemoglobin types did not significantly differ relative to the HbAA type (Table 4), which is consistent with previous studies.12-14 17 18

The relatively high proportion of Dogon cases presenting with severe malaria in this study may reflect selection bias from a catchment effect and may thus indicate a limitation in the study design with regard to residential criteria for enrollment.31 Separate analysis of this catchment effect and incorporation of 2 matched uncomplicated controls for each urban severe malaria case confirmed the OR for a protective effect of HbC against severe malaria. Although this strategy was used to minimize possible confounding variables, we recognize that differences in such factors as the prevalence of the HbC trait, entomologic inoculation rates, or self-treatment with antimalarial remedies could have affected prevalences of severe malaria in this analysis.

Severe anemia was found to be relatively infrequent in our study, so the number of cases was insufficient to test whether the HbC trait is associated with protection against this manifestation of severe malaria. In contrast to reports from other regions of Africa, where severe anemia (defined as a hemoglobin of less than 50 g/L [5 g/dL]) was present in 30%-80% of severe malaria cases,8 32 33 only 18% of the severe cases in Bandiagara presented with a comparable degree of anemia. Possible roles of other genetic loci affecting malaria susceptibility, parasite transmission intensity, prior antimalarial treatment, nutritional differences, and other factors in anemia have yet to be explored. In The Gambia in West Africa, we note that a reduced prevalence of anemia but not of cerebral malaria was linked to the human leukocyte antigen (HLA)-DRB1*1302 haplotype, and the HLA-B53 allele was associated with protection from both forms of the severe disease.8

Of the 16 HbAS Dogon cases in the study, 3 cases presented with manifestations of severe malaria, and 2 of these cases presented with convulsions. While additional studies will be required to establish the risk of severe malaria in Dogon children with HbAS, these cases raise the question whether there is a less relative selection for HbAS by malaria in Dogons than for HbAC and HbCC. Such a possibility would be consistent with the fact that incidences of HbS and HbC are 3% and 16%, respectively, in the Dogons, which is different from the respective incidences of 20% HbS and 3% HbC in the Malinke group34 from which the Dogons are thought to have migrated 500-800 years ago.35 Genetic differences between these populations or regional environment factors affecting protection by HbS or HbC may account for the contrast in the current findings from our previous study of a predominantly Bambara study group in Bamako, Mali,15 and those from another large study in The Gambia.8 Linkage disequilibrium in these different populations may also be present between theβ S orβ C genes and other loci that affect susceptibility to severe malaria.

The possibility that protective effects associated with different hemoglobin mutations may vary among different human populations is consistent with other reported findings. Early or vigorous development of malaria immunity has been associated with protection of sickle cell trait and thalassemia,36-40 and ethnic variations in the immunological response to P falciparum infection and rates of the clinical episodes of malaria are described from clinical studies in West Africa.41 42 Differences in the prevalence of splenomegaly from malaria infection have been found among West African populations.43 44 Specific environmental influences and diets (eg, the effect of fava beans in the case of HbE)45may also contribute to the selection of hemoglobin variants. Investigations of questions raised by these observations should provide fundamental information on factors that affect the distribution of hemoglobinopathies and determine the disease manifestations of severe malaria.


We thank the residents of Bandiagara, Mali; the Bandiagara Traditional Healers group; Dr Chiaka Diakite and the staff of the Traditional Medical Center in Bandiagara for their help and encouragement; Boureima Ouologuem and Akouni Dougnon for their assistance; the Regional Malaria Control Program at Mopti, Mali; and Robert W. Gwadz and Richard K. Sakai for their efforts in the support of this work.


  • Thomas E. Wellems, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bldg 4, Rm 126, 4 Center Dr, MSC 0425, Bethesda, MD 20892-0425; e-mail:tew{at}

  • Supported in part by the US Agency for International Development (USAID) through the Health and Human Resources Analysis for Africa (HHRAA) Program and through the USAID Mission, Bamako, Mali.

  • The publication costs of this article were defrayed in part by page charge payment. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 U.S.C. section 1734.

  • Submitted February 7, 2000.
  • Accepted May 30, 2000.


View Abstract