Anaemia, iron deficiency and a common polymorphism of iron-regulation, TMPRSS6 rs855791, in Rwandan children


  • Ina Danquah,

    1. Department of Molecular Epidemiology, German Institute of Human Nutrition Potsdam-Rehbruecke, Nuthetal, Germany
    Search for more papers by this author
  • Jean-Bosco Gahutu,

    1. Faculty of Medicine, Butare University Teaching Hospital, National University of Rwanda, Butare, Rwanda
    Search for more papers by this author
  • Irene Zeile,

    1. Institute of Tropical Medicine and International Health, Charité – University Medicine Berlin, Berlin, Germany
    Search for more papers by this author
  • Andre Musemakweri,

    1. Faculty of Medicine, Butare University Teaching Hospital, National University of Rwanda, Butare, Rwanda
    Search for more papers by this author
  • Frank P. Mockenhaupt

    Corresponding author
    1. Institute of Tropical Medicine and International Health, Charité – University Medicine Berlin, Berlin, Germany
    • Corresponding Author Frank P. Mockenhaupt, Institute of Tropical Medicine and International Health, Charité – Universitätsmedizin Berlin, Spandauer Damm 130, 14050 Berlin, Germany. Tel.: +49 30 30116 815; Fax: +49 30 30116 888; E-mail:

    Search for more papers by this author


Anaemia in children living in sub-Saharan Africa is common, but its causes are diverse. In 545 children below 5 years of age from rural southern Rwanda, we assessed the role of iron deficiency (ID) and of the TMPRSS6 736(V) (rs855791) allele, known to reduce iron status and haemoglobin (Hb) levels, in anaemia and Hb concentrations. Anaemia (Hb <11 g/dl) was present in 34.4% of the children and ID (ferritin <12 ng/ml) in 17.6%. The TMPRSS6 736(V) allele was uncommon (allele frequency, 0.096) and not associated with ID. In multivariate analysis, ID was positively associated with anaemia (adjusted odds ratio, 1.67) to an extent comparable with α+-thalassaemia, breastfeeding, inflammation and low household income, but the odds were substantially higher in Plasmodium falciparum infection (adjusted odds ratio, 10.3). These findings were verified in a multivariate analysis of Hb concentrations. The TMPRSS6 736(V) allele only tended to be associated with low Hb levels. TMPRSS6 736(V) is comparatively rare among Rwandan children and may only slightly contribute to low Hb concentrations. Preventable causes of anaemia, notably ID and P. falciparum infection, largely outweigh its impact and need to be addressed to improve the haematological status of children in the study area.


L'anémie chez les enfants vivant en Afrique subsaharienne est commune, mais ses causes sont diverses. Chez 545 enfants de moins de cinq ans en zone rurale dans le sud du Rwanda, nous avons évalué le rôle de la carence en fer et de l'allèle TMPRSS6 736(V) (rs855791), connue pour réduire la teneur en fer et les taux d’Hb, dans l'anémie et dans les concentrations en hémoglobine (Hb). Une anémie (Hb < 11 g/dl) était présente chez 34.4% des enfants et une carence en fer (ferritine < 12 ng/ml) chez 17.6% d'entre eux. L'allèle TMPRSS6 736(V) était rare (fréquence de l'allèle: 0.096) et non associé à la carence en fer. En analyse multivariée, la carence en fer était positivement associée à une anémie (odds ratio ajusté: 1.67) dans une mesure comparable avec α-thalassémie, l'allaitement, l'inflammation et le faible revenu du ménage, mais les chances étaient sensiblement plus élevées dans les cas d'infection par Plasmodium falciparum (odds ratio ajusté: 10.3). Ces résultats ont été vérifiés dans une analyse multivariée des concentrations d’Hb. L'allèle TMPRSS6 736(V) tendait seulement à être associé à de faibles taux d’Hb. TMPRSS6 736(V) est relativement rare chez les enfants rwandais et peut à peine contribuer à de faibles concentrations d’Hb. Les causes évitables de l'anémie, notamment la carence en fer et l'infection par P. falciparum, l'emportent largement sur l'impact de l'allèle et, doivent être adressées dans le but d'améliorer le statut hématologique des enfants dans la zone d’étude.


La anemia es común entre los niños que viven en África subsahariana, pero sus causas son diversas. Hemos evaluado el papel de la deficiencia de hierro (DH) y el papel del alelo TMPRSS6 736(V) (rs855791) – que se sabe reduce el estatus de hierro y los niveles de Hb – sobre la anemia y las concentraciones de Hb de 545 niños menores de cinco años de edad en zonas rurales del sur de Ruanda. La anemia (Hb < 11 g/dl) estaba presente en 34.4% de los niños y la DH (ferritina < 12 ng/ml) en 17.6%. El alelo TMPRSS6 736(V) era poco común (frecuencia del alelo, 0.096) y no estaba asociado con DH. En un análisis multivariado, la DH estaba asociada de forma positiva con la anemia (odds ratio ajustado, 1.67) hasta cierto punto comparable con la +-talasemia, la lactancia materna, la inflamación y unos bajos ingresos del hogar, pero la probabilidad era sustancialmente mayor en la infección por Plasmodium falciparum (odds ratio ajustado, 10.3). Estos hallazgos fueron verificados en un análisis multivariado de concentraciones de Hb. El alelo TMPRSS6 736(V) solo tendía a estar asociado con bajos niveles de Hb. TMPRSS6 736(V) es comparativamente raro entre niños de Ruanda y podría contribuir solo parcialmente a las bajas concentraciones de Hb. Las causas prevenibles de anemia, principalmente DH y la infección por P. falciparum, sobrepasan con creces su impacto y deberían abordarse para mejorar el estatus hematológico de los niños en el área de estudio.


Anaemia is present in two of three pre-school children in sub-Saharan Africa (SSA) (World Health Organization 2008). Iron deficiency (ID) – largely due to low intake, poor absorption or increased requirements (e. g., pregnancy, growth) – accounts for roughly half of anaemia worldwide, but its relative contribution varies with, for example, age and region. Severe anaemia is a major cause of childhood mortality in SSA, whereas milder and often chronic anaemia, notably ID anaemia, substantially impairs cognitive and physical development (World Health Organization 2001, 2008).

In Rwanda, a national survey in 2005 revealed 56% of pre-school children to be anaemic (haemoglobin (Hb) <11 g/dl) with a slightly lower prevalence (47%) in the South province (Institut National de la Statistique du Rwanda, ORC Macro 2006). In a rural area of that province in 2010, we found anaemia in 34% of 545 pre-school children, and associations with Plasmodium falciparum infection and α+-thalassaemia (Gahutu et al. 2011, 2012). Here, we re-assessed risk factors for anaemia focusing on ID. Specifically in this respect, we examined the role of a common mutation (rs855791, valin (V) to alanin (A) change at codon 736, V736A) in the transmembrane serine protease 6 gene (TMPRSS6). The product of this gene, matriptase-2, influences the transcriptional regulation of hepcidin, the key regulator of iron homoeostasis. High levels of hepcidin cause ID (Hentze et al. 2010). The TMPRSS6 736(V) allele is associated with comparatively high hepcidin serum levels (Nai et al. 2011) and, in Caucasians and Asians, with low values for serum iron, transferrin saturation, mean cell volume (Benyamin et al. 2009) and Hb (average difference between homozygotes, 0.2 g/dl) (Chambers et al. 2009). However, recent work suggests other, yet unknown, serum hepcidin independent mechanisms, which play a role in the association of TMPRSS6 variants with serum iron parameters (Galesloot et al. 2013). In SSA, common infections and (secondary) inflammatory processes complicate the assessment of ID and of its contribution to anaemia (Mockenhaupt et al. 1999). Against this background, we explored whether TMPRSS6 736(V) is associated with anaemia and ID among African children.

Subjects and methods

We conducted a cluster-sampled survey on common diseases in children under 5 years of age in Butare and its rural surroundings in 2010. The study procedures and details on malaria and erythrocyte variants have been published elsewhere (Gahutu et al. 2011, 2012). The present report focuses on 545 children from the rural but densely populated farmland hills of Huye subdistrict (1700–1800 m asl). Briefly, 25 households each were randomly chosen in 24 randomly selected villages, and one child was randomly selected per household. Informed written consent was obtained from the children's parents, and the study was approved by the National Ethics Committee, Republic of Rwanda.

All children were clinically examined, medical and socio-demographic data documented, and a venous blood sample was collected. Fever (axillary temperature ≥37.5 °C), underweight (weight-for-age z-score <−2) and anaemia (Hb <11 g/dl; HemoCue, Angelholm, Sweden) were documented. Malaria parasite density was estimated microscopically (Gahutu et al. 2011), and the species of Plasmodium as well as submicroscopic infections (i.e. below the detection threshold of microscopy) were identified by polymerase chain reaction (PCR) assays (Rubio et al. 2002) after DNA extraction (QIAamp; Qiagen, Germany). TMPRSS6 rs855791 was typed by melting curve analysis using commercially available primers and probes (TIB Mobiol, Berlin, Germany). Erythrocyte variants were identified as previously described (Gahutu et al. 2012). Plasma concentrations of ferritin and C-reactive protein (CRP) were measured by ELISA (Assaypro, St. Charles, MO, USA). ID was defined as ferritin <12 ng/ml (Cook & Skikne 1989), and inflammation as CRP >5 ng/ml (Erhardt et al. 2004). Children were treated according to Rwandan health authority guidelines.

Data were analysed using the survey data analysis module in Stata 9.0 (Stata Corporation, College Station, TX, USA). Due to the cluster sampling and the non-normal distribution of continuous variables, weighted nonparametric tests for survey data were used to compare means (95% confidence intervals [CIs]) and proportions (95% CIs). As for factors associated with anaemia, odds ratios (ORs) and 95% CIs were calculated applying logistic regression, weighting for the population size of each sampling cluster cell, that is, residence. Therefore, prevalences (%) do not necessarily correspond to absolute numbers. Stepwise backward removal of exposure variables was performed to identify independently associated variables. Weighted linear regression was used to estimate factors associated with normalised Hb concentrations (to the three transformation). A P-value ≤0.05 was considered statistically significant.


Selected characteristics of the children are shown in Table 1. Most parents were farmers, many had no formal education, and income and asset ownership were low. One in four children was underweight. P. falciparum infection (predominately asymptomatic) and inflammation were each seen in 1 in 6 children, and anaemia in one-third of the children. The TMPRSS6 736(V) allele was observed in 17.8% (allele frequency 0.096).

Table 1. Characteristics of 545 children from Huye subdistrict, South Province, Rwanda
ParameterPer cent (95% CI)n/N
  1. Values for metric variables are presented as means (95% CIs) and for categorical variables as per cent (95% CIs).

  2. Values were calculated using nonparametric tests for survey data weighted for the population size of the sampling cluster.

  3. a

    Defined as microscopically visible P. falciparum infection plus fever (>37.4 °C) or history of fever (last 48 h).

  4. b

    Using alternative definitions, the prevalences were 16.9% (95% CI, 13.6–20.2; n = 93) (ferritin <12 ng/ml, or <30 ng/ml if CRP >5 mg/dl) and 21.1% (95% CI, 17.5–24.6; n = 118) (ferritin <30 ng/ml).

N 545 
Proportion girls44.6 (39.8–49.3)247/544
Age (months), n = 545 [mean (95% CI)]31.4 (30.1–32.7) 
Proportion breastfed51.5 (46.8–56.3)285/544
Monthly household income (<5000 RwF)37.3 (32.8–41.8)218/544
Lacking formal education, mother30.1 (25.7–34.4)165/543
Mother's occupation: farmer/laborer98.7 (97.6–99.8)533/540
Lacking formal education, father39.1 (34.4–43.8)195/534
Father's occupation: farmer/laborer88.4 (85.2–91.7)472/543
Absence of assessed household assets52.1 (47.3–56.9)185/540
No. of people/household, n = 544 [mean (95% CI)]5.5 (5.4–5.7) 
No. of siblings, n = 544 [mean (95% CI)]2.1 (1.9–2.2) 
Underweight (weight-for-age z-score <−2)24.4 (20.3–28.5)133/543
Fever (>37.4 °C axillary)3.1 (1.3–4.9)18/543
History of fever (last 48 h)9.4 (6.7–12.2)49/502
Plasmodium falciparum infection (PCR)16.2 (12.7–19.7)88/545
Malaria parasites on blood film11.7 (8.8–14.6)61/545
Malariaa2.5 (1.0–4.0)16/545
Proportion anaemic (Hb <11 g/dl)34.4 (29.8–38.9)88/545
Proportion moderately severe anaemia (Hb <7 g/dl)1.9 (0.7–3.1)10/545
Hb (g/dl), n = 545 [mean (95% CI)]11.2 (11.1–11.4) 
Iron deficiency (Ferritin <12 ng/ml)b16.8 (13.5–20.0)92/545
Inflammation (CRP >5 mg/l)16.3 (12.9–19.6)82/545
None86.0 (82.9–89.0)467/545
Heterozygous13.4 (10.4–16.3)75/545
Homozygous0.7 (0.0–1.5)3/545
Glucose-6-phosphate dehydrogenase deficiency
None90.4 (87.8–92.9)489/544
Heterozygous5.0 (3.2–6.8)30/544
Homo-/hemizygous4.6 (2.7–6.5)25/544
TMPRSS6 rs855791
AA82.2 (78.8–85.6)450/545
AV15.6 (12.4–18.8)85/545
VV2.2 (0.8–3.6)10/545

Iron deficiency was present in 17.5%. This figure was similar when accounting for the role of inflammation in the definition of ID (Table 1; 15.1% (95% CI, 11.6–18.7%) of children with ferritin ≥12 ng/ml showed inflammation). Ferritin levels were comparatively low in children ≤2 years of age (mean, 67.1; 95% CI, 56.4–77.8 ng/ml) and higher thereafter (mean, 87.2; 95% CI, 77.6–96.7 ng/ml, = 0.006). Correspondingly, ID was more common in children ≤2 years of age (22.6%) than in the older children (13.6%, = 0.001). ID was not associated with sex, residence, CRP levels, underweight or socio-economic indicators (data not shown). Nor was the TMPRSS6 736 variant associated with ID: in children with AA, VA and VV genotypes, ID was observed in 16.5% (95% CI, 12.9–20.0%; 74/450), 17.5% (95% CI, 9.3–25.6%; 16/85) and 23.7% (95% CI, −5.3 to 52.7%; 2/10), respectively (= 0.68). In these groups, mean (95% CI) plasma ferritin concentrations were 79.0 (71.2–86.8), 88.8 (67.3–110.3) and 58.8 (27.6–90.0) ng/mlL, respectively (= 0.57).

Table 2 displays factors associated with anaemia. Anaemia was observed in 51.5% of children with ID and in 31.2% of children without (OR, 2.3). 25.0% (95% CI, 18.5–31.4%; 47/186) of anaemic children had ID, the proportion being virtually identical in children ≤2 years of age (25.3%; 95% CI, 16.1–34.4%; 24/94) and in older children (24.7%; 95% CI, 15.5–33.8; 23/92). In comparison, 34.6% (95% CI, 27.5–41.8%; 64/186) of anaemic children were P. falciparum infected, with a lower proportion in younger (23.6%; 95% CI, 14.6–32.7%; 22/94) than older children (45.2%; 95% CI, 34.6–55.9%; 42/92; = 0.004). The slight overrepresentation of anaemia in children carrying the TMPRSS6 736(V) allele did not reach statistical significance, irrespective of age. In multivariate analysis, age and P. falciparum infection were the strongest independent predictors of anaemia. Moreover, the odds were increased by 67% in iron deficient children. Similar estimates were seen for α-thalassaemia, breastfeeding, inflammation and low household income. In this model, TMPRSS6 736(V) only tended to be associated with anaemia (Table 2).

Table 2. Univariate and multivariate analyses of factors associated with anaemia
FactorAnaemia (%)OR (95% CI) P aOR (95% CI)a P
  1. a

    A logistic regression model with weighted estimates for residence by population size of the sample cluster was adjusted for sex, and the other variables remaining significant after stepwise backward removal.

  2. b

    <−2 weight-for-age z-score.

  3. c

    If remaining in this model, AV, aOR, 1.45 (95% CI, 0.78–2.70), = 0.24; VV, aOR, 2.90 (95% CI, 0.86–9.80), = 0.086; AV + VV, aOR, 1.57 (95% CI, 0.88–2.80), = 0.125.

Age (years)
0 < 164.81   
1 < 240.80.37 (0.19–0.73)0.0040.26 (0.13–0.54)<0.0001
2 < 330.00.23 (0.12–0.46)<0.00010.12 (0.06–0.27)<0.0001
3 < 426.00.19 (0.09–0.39)<0.00010.07 (0.03–0.18)<0.0001
4 < 524.70.18 (0.08–0.38)<0.00010.05 (0.02–0.13)<0.0001
No30.61 1 
Yes38.51.42 (0.98–2.07)0.0662.03 (1.09–3.77)0.026
History of fever (last 48 h)
Yes53.42.38 (1.28–4.42)0.006  
Monthly household income
≥5000 RwF30.31   
<5000 RwF41.21.62 (1.11–2.35)0.0131.62 (1.04–2.52)0.032
Household assets
None43.12.03 (1.39–2.96)<0.0001  
Present44.01.90 (1.20–3.02)0.0072.02 (1.13–3.61)0.017
P. falciparum infection
Present73.77.57 (4.42–12.95)<0.000110.29 (5.77–18.36)<0.0001
CRP >5 mg/dl
Yes50.22.19 (1.33–3.60)0.0021.87 (1.08–3.26)0.026
Ferritin <12 ng/ml
Yes51.52.34 (1.46–3.76)<0.00011.67 (1.00–2.79)0.050
Present45.91.92 (1.27–2.92)0.002  
TMPRSS6 rs855791
AV37.01.15 (0.70–1.91)0.58 c  
VV49.01.89 (0.50–7.09)0.34  

Lastly, factors influencing (normalised) Hb concentrations per se were analysed. Mean (95% CI) Hb concentrations in children with and without ID were 10.5 (95% CI, 10.0–10.9) and 11.4 (95% CI, 11.3–11.6) g/dl, respectively (< 0.0001). In contrast, the TMPRSS6 variant was not significantly associated with Hb concentrations (means [95% CIs]; AA, 11.3 [11.1–11.4]; AV, 11.1 [10.7–11.4]; VV, 11.0 [9.3–12.7]; = 0.37). In a multiple linear regression model replicating the above logistic regression model (all univariately associated parameters plus the TMPRSS6 polymorphism), P. falciparum infection was the strongest factor independently associated with low Hb concentrations (regression coefficient β = −8.66 [standard error = 3.87]; < 0.0001) followed by α+-thalassaemia (no/yes, β = −5.63 [3.63], < 0.0001), ID (β = −5.20 [3.87], = 0.016), inflammation (β = −5.05 [3.78], = 0.017), breastfeeding (β = −5.02 [3.80], = 0.02) and a household income below the median (β = −4.78, [3.43], = 0.007). The TMPRSS6 736(V) allele was nominally associated with low Hb concentrations (no/yes, β = −4.48 [3.72], = 0.08). Age was the only parameter positively associated with Hb levels in this model (years, β = 5.64 [2.79], < 0.0001).


Contrasting its significance, data on anaemia and ID in children in SSA are remarkably scarce. Here, we show that in the southern highlands of Rwanda, every fourth child with anaemia had ID, and the Hb difference between children with and without iron deficiency was in the range of 1 g/dl. ID was more prevalent in young children but not associated with underweight or socio-economic parameters, suggesting that the actual risk factors for ID, for example, nutrients intake or micronutrient deficiencies, were not assessed in the present study. In addition to ID, anaemia, present in 34%, was found to be associated with low age, breastfeeding, low household income, α+-thalassaemia, inflammation and, most strongly, P. falciparum infection. This illustrates the complex aetiology of anaemia in African children (World Health Organization 2008). Also, in the present study, these factors clearly outweighed the influence of the TMPRSS6 polymorphism which, in addition, was not significantly associated with anaemia. Although a larger sample size and a more focused study design would have been desirable, our findings are biologically plausible and the Hb difference between homozygotes resembled the one previously reported (Chambers et al. 2009). Ferritin concentrations also did not correlate with the TMPRSS6 genotypes. Apart from sample size reasons, abundant infectious and inflammatory processes affecting children in SSA may on the one hand influence the validity of ferritin as a marker of iron status (Mockenhaupt et al. 1999). On the other hand, the influence of these processes likely overrides the impact of the TMPRSS6 polymorphism on hepcidin concentrations and other iron-regulatory mechanisms which secondarily influence macrophage iron release, iron absorption and thus plasma iron levels and anaemia (Hentze et al. 2010). Notable, however, is the low allele frequency of TMPRSS6 736(V) of 0.096 in the present study population. While this accords with genetic data from other African populations, the frequency is considerably higher in European (≥0.4) and Asian (≥0.50) populations (Nai et al. 2011). Whether the unfavourable TMPRSS6 736(V) allele is subject to evolutionary selection still needs to be explored.

In conclusion, anaemia among pre-school children in the southern highlands of Rwanda is frequent. One in four anaemic children has iron deficiency. The TMPRSS6 736(V) allele previously reported to be associated with ID and anaemia is less common in Rwanda than in non-African populations and seems to contribute only slightly to low haemoglobin levels. Preventable conditions, notably P. falciparum infection but also poverty, are common factors strongly associated with anaemia and need to be addressed to improve child health in this area.


The authors thank the children's families for participation in this study, and the staff at Sovu health centre, Kabutare district hospital and Butare University Teaching Hospital central laboratory for assistance and cooperation. The active assistance by the Rwanda office of the German Development Cooperation (GIZ) is gratefully acknowledged. This study was supported by the German Federal Ministry for Economic Cooperation and Development via the ESTHER programme (Ensemble pour une Solidarité Thérapeutique Hospitalière En Réseau). This work forms part of the doctoral thesis of IZ.