Decreased iron burden in overweight C282Y homozygous women: Putative role of increased hepcidin production


  • Potential conflict of interest: Nothing to report.


An excess of visceral adipose tissue could be involved as a modulator of the penetrance of HFE hemochromatosis since fat mass is associated with overexpression of hepcidin and low transferrin saturation was found to be associated with being overweight in women. This study was aimed at assessing the relationship between body mass index (BMI), a surrogate marker of insulin resistance, and iron burden in HFE hemochromatosis. In all, 877 patients from a cohort of C282Y homozygotes were included in the study when BMI at diagnosis and amount of iron removed (AIR) by phlebotomy were available. No relationship between AIR and BMI was found in men, whereas 15.1% (52/345) of women with AIR <6 g had BMI ≥28 versus 3.9% (2/51) of women with AIR ≥6 g (P = 0.03). At multivariate analysis, BMI was an independent factor negatively associated with AIR (odds ratio: 0.13; 95% confidence interval [CI]: 0.03-0.71) together with serum ferritin, serum transferrin, transferrin saturation, hemoglobin, and alanine aminotransferase. In a control group of 30 C282Y homozygous women, serum hepcidin was significantly higher in overweight (14.3 mmoL/L ± 7.1) than in lean (7.9 mmoL/L ± 4.3) women (P = 0.0005). Conclusion: In C282Y homozygous women, BMI ≥28 kg/m2 is independently associated with a lower amount of iron removed by phlebotomy. BMI is likely a modulator factor of the phenotypic expression of C282Y homozygosity, likely through an increase of circulating levels of hepcidin. (HEPATOLOGY 2013;)

HFE-hemochromatosis is, by far, the most common form of genetic iron-overload disease in Caucasians.1 It is transmitted as an autosomal recessive trait and mainly related to the C282Y mutation on the HFE gene involved in the regulation of hepcidin synthesis.2 C282Y homozygosity results in a hepdicin-deficient state3 responsible for increased transferrin saturation and, then, parenchymal iron overload. However, its penetrance is incomplete, and according to Allen et al.,4 only 1% of C282Y homozygous women and 28% of C282Y homozygous men develop a clinical disease. Thus, C282Y homozygosity is a necessary but not a sufficient condition to promote an overt disorder. This implies that genetic and environmental factors may modulate either iron burden and/or organ damage in HFE-hemochromatosis.5 Gender,6, 7 alcohol,8 regimen,9, 10 drugs,11 and polymorphisms in genes involved in iron metabolism12-14 have been identified as such putative modulating factors.

Based on the study of Laine et al.,15 who reported that, in a large screening program conducted in the general population of Brittany, France, transferrin saturation (TS) was lower in overweight than in lean C282Y homozygous women, we hypothesized that overweight might be associated with underexpression of HFE-hemochromatosis and that this might be related to the production of hepcidin by the visceral adipose tissue.16, 17 The present retrospective study was aimed at assessing, in a large cohort of C282Y homozygotes, the amount of iron removed (AIR) by phlebotomy according to body mass index (BMI). The correlation of serum hepcidin level with BMI was studied in an independent group of patients with frozen blood samples available.


AIR, amount of iron removed; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; GGT, gamma-glutamyl-transferase; Hb, hemoglobin; MCV, mean corpuscular volume; mRNA, messenger RNA; ROC, receiver operating characteristic; TG, triglycerides; TS, transferrin saturation.

Patients and Methods

C282Y homozygotes were selected from a cohort initiated in 1990 for the management of family screening procedures. All probands followed for genetic hemochromatosis since 1989 at the University Hospital of Rennes, France and their relatives are collected in the database which records initial and follow-up biometric, clinical, biological, imaging, histological, and therapeutic data. This database was declared to the French Committee “Informatics and Freedom” (CNIL) and patients gave informed written consent. Data are recorded in patient files by a technical research assistant, keyboarded twice by two secretaries, checked by a clinical research assistant (M.P.) and analyzed by a biostatistician (J.M.) devoted to the database. At the time of the study, 1,985 C282Y homozygotes had been recorded.

C282Y homozygotes were included in the study when they were older than 18 years, had BMI recorded at the time of diagnosis, and had follow-up data available to allow a reliable calculation of AIR. AIR was defined as the amount of iron removed since the beginning of phlebotomies until the achievement of low body iron stores, corresponding usually to serum ferritin levels lower than 80 μg/L. It was calculated assuming that the removal of 1 L of blood corresponds to 0.5 g of depleted iron.18

The following data were recorded when available in the database at the time of diagnosis: (1) biological data: serum iron (μmol/L), serum transferrin (g/L), transferrin saturation (TS; percent), serum aspartate aminotransferase (AST; IU/L), alanine aminotransferase (ALT; IU/L), gamma-glutamyl-transferase (GGT; IU/L), hemoglobin (Hb; g/dL), mean corpuscular volume (MCV), HDL-cholesterol (mmoL/L), serum triglycerides (TG; mmol/L); (2) clinical data: hypertension (blood pressure ≥140/90 mmHg or antihypertensive therapy), tobacco and alcohol consumption, diabetes (fasting blood glucose ≥1.26 g/L or antidiabetic therapy) and, in women, number of pregnancies and menopause status; (3) existence of frozen blood samples drawn at the time of diagnosis.

Serum hepcidin was measured by an immune-enzymatic assay (EIA Bachem, Bubendorf, Switzerland) without preliminary extraction. Due to a technical incident (defrosting during transport), frozen samples available from the study group were rendered unusable. Then a second set of 30 frozen samples, drawn at the time of diagnosis, on the morning in fasting subjects, before initiation of therapy and stored at −80°C, was constituted from C282Y homozygous patients, of whom 4/30 did not fulfill the criteria of inclusion due to the unavailability of AIR.

A Pearson correlation test was used to evaluate the relationship between BMI and AIR in men and women separately. To determine classes of BMI, receiver operating characteristic (ROC) curve analysis according to low and high AIR was performed and the value of BMI associated with the highest Youden index was chosen to separate patients into two classes of BMI. Then univariate analysis of AIR and BMI as categorical variables was performed using Student test or the Wilcoxon test for quantitative data, and χ2 or exact Fisher's test for qualitative data. All variables for which statistical significance was <0.2 were introduced into a generalized linear regression multivariate model with AIR as the independent variable (SAS 9.2, Cary, NC). To analyze the relationship between serum hepcidin and BMI, a Wilcoxon test was used. Statistical significance was considered as P < 0.05. Results are expressed as mean ± standard deviation (SD).


Among the 1,985 C282Y homozygotes recorded at the time of inclusion, 1,108 patients were excluded because of age <18 years and/or absence of AIR and/or absence of BMI at diagnosis. The study population consisted then of 877 patients (396 women and 481 men) whose main characteristics are presented in Table 1. No linear correlation (Pearson's test) was found between AIR and BMI either in women (Fig. 1) or in men (Fig. 2). However, in women a BMI threshold beyond which AIR was <6 g (85th percentile) could be suspected at examination (Fig. 1) was determined as 28 kg/m2 by ROC curve analysis. Then univariate and multivariate analysis was carried out in women only.

Table 1. Main Clinical and Biochemical Characteristics of the C282Y Homozygous Patients and of the 30 Control Women
 Women (n=396)Control Women (n=30)Men (n=481)
 Mean (SD)Mean (SD)Mean (SD)
Age (years)47.7 (14.1)48.7 (14.6)43.4 (11.4)
BMI (kg/m2)23.9 (4.5)23.9 (4.6)25.1 (3.7)
AIR (g)3.5 (3.7)2.8 (2.1)8.2 (6.8)
Serum ferritin (μg/L)689.4 (973.8)594.6 (444.8)1613.6 (1395.8)
Serum iron (μmol/L)31.4 (7.9)30.9 (6.5)35.8 (8.1)
Serum transferrin (g/L)1.83 (0.34)1.71 (0.32)1.83 (0.31)
Transferrin saturation (%)70.8 (18.1)74.2 (17.2)78.8 (17.1)
AST (UI/L)22.8 (15.7)25.0 (12.6)31.6 (19.3)
ALT (UI/L)28.7 (24.5)31.9 (24.0)53.7 (36.9)
GGT (UI/L)28.5 (28.3)33.8 (25.5)46.9 (47.8)
Hemoglobin (g/dL)13.7 (0.9)13.7 (0.8)15.0 (1.1)
MCV94.6 (5.1)92.7 (12.1)95.0 (4.3)
HDL-cholesterol (mmol/L)1.64 (0.42)1.63 (0.33)1.30 (0.27)
Serum triglycerides (mmol/L)1.08 (0.54)0.95 (0.51)1.43 (0.80)
 n (%)n (%)n (%)
BMI (kg/m2)   
< 28342 (86.4)26 (86.7)381 (79.2)
≥ 2854 (13.6)4 (13.3)100 (20.8)
Antihypertensive therapy   
Yes46 (14.6)3 (10.0)48 (12.8)
No269 (85.4)27 (90.0)326 (87.2)
Yes10 (3.0)1 (3.5)32 (7.9)
No320 (97.0)28 (96.5)371 (92.1)
Alcohol consumption   
Nonexcessive285 (95.6)25 (96.2)320 (79.6)
Excessive13 (4.4)1 (3.8)82 (20.4)
Tobacco consumption   
Current smoker60 (24.0)7 (26.9)111 (33.6)
Exsmoker23 (9.2)1 (3.9)68 (20.6)
Nonsmoker167 (66.8)18 (69.2)151 (45.8)
Yes172 (49.7)13 (44.8)
No174 (51.3)16 (55.2)
065 (20.0)4 (13.8)
1 or 2150 (46.2)16 (55.2)
≥ 3110 (33.8)9 (31.0)
Figure 1.

Despite the absence of linear correlation between AIR (g) and BMI (kg/m2) in C282Y homozygous women, distribution of AIR < and ≥6 g (horizontal dotted line) was significantly different (χ2 test, P = 0.03) and BMI was < or ≥28 kg/m2 (vertical dotted line).

Figure 2.

Absence of correlation between AIR (g) and BMI (kg/m2) in C282Y homozygous men.

Univariate Analysis.

Among the 396 women, 12.9% (51/396) had AIR ≥6 g. As expected, AIR ≥6 g was positively correlated with age at diagnosis and markers of both iron burden (serum ferritin and transferrin saturation) and organ damage (AST, ALT, GGT, and diabetes) (Table 2). Although the mean BMI was not significantly different between women with low and high AIR (24.0 kg/m2 ± 4.5 versus 23.9 kg/m2 ± 4.9; P = 0.94), the distribution of AIR clearly differed according to the two classes of BMI: among women with AIR ≥6 g, 3.9% (2/51) had BMI ≥28 kg/m2 versus 15.1% (52/345) in women with AIR < 6 g (P = 0.03) (Fig. 1). AIR was positively associated with menopause but not with the number of pregnancies (Table 2).

Table 2. Main Clinical and Biochemical Characteristics of C282Y Homozygous Women According to AIR Classes
  AIR < 6 g n=345AIR ≥ 6 g n=51 
 nMeans (SD)Means (SD)P
Age (years)39647.0 (14.5)52.1 (10.2)0.02
BMI (kg/m2)39624.0 (4.5)23.9 (4.9)0.94
Serum ferritin (μg/L)390450.3 (293.6)2353.5 (1959.7)<0.0001
Serum iron (μmol/L)38631.2 (8.1)32.9 (6.3)0.16
Serum transferrin (g/L)3751.86 (0.33)1.59 (0.34)<0.0001
Transferrin saturation (%)38368.9 (17.9)85.1 (11.8)<0.0001
AST (UI/L)33521.0 (11.7)35.6 (29.0)<0.0001
ALT (UI/L)34326.0 (19.3)48.2 (42.6)<0.0001
GGT (UI/L)30226.7 (24.5)40.0 (44.1)0.005
Hemoglobin (g/dL)30213.8 (0.9)13.2 (1.1)0.0002
MCV29694.3 (5.2)96.7 (4.0)0.009
HDL-cholesterol (mmol/L)1381.64 (0.42)1.64 (0.47)0.96
Serum triglycerides (mmol/L)1971.08 (0.56)1.09 (0.43)0.93
  n (%)n (%) 
BMI (kg/m2)396   
< 28 293 (84.9)49 (96.1)0.03
≥ 28 52 (15.1)2 (3.9) 
Antihypertensive therapy315   
Yes 39 (14.1)7 (18)0.53
No 237 (85.9)32 (82) 
Yes 6 (2.1)4 (9.8)0.025
No 283 (97.9)37 (90.2) 
Alcohol consumption298   
Nonexcessive 251 (96.5)34 (89.5)0.07
Excessive 9 (3.5)4(10.5) 
Tobacco consumption250   
Current smoker 54 (24.3)6 (21.4)0.85
Exsmoker 21 (9.5)2 (7.1) 
No smoking 147 (66.2)20 (71.4) 
Yes 144 (47.5)28 (65.1)0.0006
No 159 (52.5)15 (34.9) 
0 61 (21.4)4 (10)0.23
1 or 2 130 (45.6)20 (50) 
≥ 3 94 (33)16 (40) 

When compared with the 342 women with BMI <28 kg/m2, the 54 women with BMI ≥28 kg/m2 were older and had lower transferrin saturation (63.2% ± 18.9 versus 72.0% ± 17.7), although AIR and serum ferritin levels did not differ (Table 3).

Table 3. Main Clinical and Biochemical Characteristics of C282Y Homozygous Women According to BMI Classes
  BMI < 28 kg/m2 n=342BMI ≥ 28 kg/m2 n=54 
 nMeans (SD)Means (SD)P
Age (years)39646.9 (14.2)52.4 (12.7)0.008
AIR (g)3963.6 (3.9)2.9 (1.6)0.23
Serum ferritin (μg/L)390707.5 (1036.0)574.4 (378.3)0.36
Serum iron (μmol/L)38631.8 (7.8)28.6 (8.1)0.006
Serum transferrin (g/L)3751.82 (0.34)1.87 (0.38)0.30
Transferrin saturation (%)38372.0 (17.7)63.2 (18.9)0.0009
AST (UI/L)33522.8 (16.3)22.8 (12.0)0.99
ALT (UI/L)34328.2 (24.6)31.3 (23.7)0.4
GGT (UI/L)30226.7 (24.5)40.0 (44.1)0.005
Hemoglobin (g/dL)30213.7 (0.9)14.0 (0.8)0.036
MCV29696.6 (5.4)94.1 (3.3)0.55
HDL-Cholesterol (mmol/L)1381.69 (0.43)1.40 (0.22)0.002
Serum triglycerides (mmol/L)1971.03 (0.50)1.39 (0.58)0.001
  n (%)n (%) 
Antihypertensive therapy315   
Yes 39 (14.1)7 (18)0.53
No 237 (85.9)32 (82) 
Yes 7 (2.4)3 (6.8)0.14
No 279 (97.6)41 (93.2) 
Alcohol consumption298   
Nonexcessive 245 (95.3)40 (97.6)1.00
Excessive 12 (4.7)1 (2.4) 
Tobacco consumption250   
Current smoker 53 (25.2)7 (17.5)0.58
Exsmoker 19 (9.1)4 (10) 
No smoking 138 (65.7)29 (72.5) 
Yes 21 (11.5)15 (33.3)0.004
No 239 (88.5)30 (66.7) 
0 61 (21.7)4 (9.1)0.15
1 or 2 127 (45.2)23 (52.3) 
≥ 3 93 (33.1)17 (38.6) 

Multivariate Analysis.

Serum ferritin, TS, and ALT were positively associated, and serum transferrin, hemoglobin, and BMI ≥28 kg/m2 were negatively associated with AIR ≥6 g (Table 4).

Table 4. Multivariate Analysis in Women. Variable to be Explained: Amount of Iron Removed
VariablesnOdds Ratio95% CI
BMI (kg/m2)396  
< 283421 
≥ 28540.130.03-0.71
Serum ferritin (μg/L)290  
< 5002231 
≥ 50016744.789.60-208.80
Serum transferrin (g/L)375  
≤ 1.81681 
> 1.82070.380.16-0.92
Transferrin saturation (%)375  
< 752021 
≥ 751817.872.76-22.43
Hemoglobin (g/dL)302  
≤ 13681 
< 132340.150.05-0.44
ALT (UI/L)343  
≤ 211721 
> 211713.781.29-11.09

Serum Hepcidin According to BMI.

The 30 women of the control group did not differ significantly from the study group with respect to age and iron burden (Table 1). In this group, hepcidin was significantly higher in overweight (BMI ≥25 kg/m2) than in lean women (BMI <25 kg/m2; P = 0.0005) (Fig. 3). The mean serum hepcidin of the 10 overweight women (29.1 kg/m2 ± 3.8) was 14.3 mmoL/L (± 7.1) compared to 7.9 mmoL/L (± 4.3) in lean women (21.2 kg/m2 ± 2).

Figure 3.

Serum hepcidin level in lean (BMI <25 kg/m2; n = 20) and overweight (BMI ≥25 kg/m2; n = 10) C282Y homozygous women. Large horizontal bar = median; upper and lower box lines = 25% and 75% percentiles, respectively.


The present data demonstrate that, in C282Y homozygous women, but not in men, overweight defined as BMI ≥28 kg/m2 is independently associated with lower iron burden and suggests that this could be related to an increase of hepcidin production.

Due to the retrospective design of the study over a period of 30 years, parameters allowing for a strict definition of the metabolic syndrome, i.e., waist circumference, blood pressure, serum HDL cholesterol, serum triglycerides, and serum glucose, were often missing in the database. Thus, BMI, which was documented in most cases, was chosen as a surrogate marker. This is certainly a limitation of the study since increased waist circumference—which is considered a more reliable clinical marker of insulin resistance than BMI19—was shown to be more closely associated with low transferrin saturation than BMI in a preliminary study of C282Y homozygous women detected through a systematic genotyping in the general population of Brittany, France.15

AIR was chosen as the main marker of iron burden because, when correctly calculated, it is the method of reference to assess total body iron stores.20 Unlike HIC determined by histology, biochemistry,21, 22 or magnetic resonance imaging (MRI),23 AIR is not influenced by tissue sampling and takes into account both hepatic and extrahepatic mobilizable iron. It is also much more reliable than serum ferritin levels, which may be increased far beyond the real amount of iron excess in several common conditions such as metabolic abnormalities,24 excessive alcohol consumption,25 inflammatory syndrome, or increased serum transaminases levels.

In the present study there was no statistical linear correlation between AIR and BMI in either sex, but in women we found a BMI threshold of 28 kg/m2 beyond which almost all AIR values remained lower than 6 g. Despite this, when compared to women with BMI <28 kg/m2, women with BMI ≥28 kg/m2 were older, were more often postmenopausal, and did not have more pregnancies, all conditions associated with increased body iron stores. Moreover, when taking into account these potential confounding factors and others in a multivariate model, we found that BMI remained as an independent explanatory variable of AIR together with expected variables including parameters of iron metabolism, ALT, and hemoglobin.

Laine et al.15 demonstrated lower C282Y homozygosity expression, defined as TS <45%, in overweight women when compared to normal and lean women in a Britton general population. In a much larger sample, the present study confirms that both serum iron and transferrin saturation are significantly lower in women with BMI ≥28 than in women with BMI <28 kg/m2, which suggests a lower bioavailability of systemic iron in case of significant overweight, and then a lower rate of iron loading. Such data could support the role of an increased production of hepcidin in lowering iron burden in overweight C282Y women. Indeed, metabolic syndrome is associated with a minimal chronic inflammatory state related to the synthesis of proinflammatory cytokines and adipokines,26 in particular interleukin (IL)-627 and leptin.28 These cytokines are known to promote hepcidin gene transcription through the STAT3 pathway.29 Moreover, several studies have shown that there is an overexpression of hepcidin in obesity. Bekri et al.16 reported that subcutaneous and visceral adipose hepcidin messenger RNA (mRNA) expression was significantly higher in obese compared to lean women, while liver mRNA expression was similar. Tussing-Humphreys et al.30 confirmed this result in obese women, but reported that hepatic hepcidin mRNA expression was strongly correlated with serum hepcidin, but not adipose hepcidin mRNA.16, 30 In a small sample of patients, a study reported that there was no oversecretion of hepcidin by subcutaneous adipose tissue whether the patient was obese or lean.31 Thus, the relationship that we found between serum hepcidin level and BMI in C282Y homozygous women suggests that the overproduction of hepcidin could be responsible for lower disease penetrance in the overweight cases. However, the mechanism of such an overproduction remains unclear, especially with respect to the tissular origin of hepcidin.

The relationship between BMI and AIR was found only in C282Y homozygous women. It is interesting to note that the association of iron deficiency with obesity was also described in women only (and in children)32, 33 and attributed to increased hepcidin synthesis secondary to an overweight-related chronic inflammatory state,30 possibly from extrahepatic sites.16 On the other hand, the dysmetabolic iron overload syndrome (DIOS), with associated features of insulin resistance and moderate iron overload, is mainly described in men.34 Altogether, these data strongly suggests that there is crosstalk between iron metabolism, insulin-resistance, and hormonal environment. In women, after the cessation of menstruation, the incidence of metabolic syndrome progressively increases up to that of men.35 This is commonly attributed to loss of the estrogen-related protective effect against insulin resistance.36 Thus, we speculate that the decrease in estrogen production could lead, through an increase of fat mass, to an overweight-related chronic inflammatory state resulting in increased hepcidin expression. Estrogen exposure of fish was found to result in a decrease in hepatic hepcidin expression, strengthening this hypothesis.37 However, such a link is likely more complex, since visceral adipose tissue is also a site of estrogen synthesis by aromatization of androgens from suprarenal glands,38 proportionally to fat mass.39 Furthermore, the aromatase expression is enhanced by proinflammatory cytokines whose expression is increased in metabolic syndrome.40 In premenopausal women, even in overweight cases, extraovarian estrogen synthesis would have not enough influence because it is overtaken by ovarian synthesis. In postmenopausal women, fat mass represents the only site of estrogen synthesis, which could influence iron metabolism. Moreover, a possible link between estrogen and the BMP6 pathway, the main pathway of hepcidin regulation, could also exist since estrogen decrease is associated with a decrease of BMP6 expression in bones, and partly explains osteoporosis in postmenopausal women.41

In conclusion, in C282Y homozygous women, BMI values greater than 28 kg/m2 are associated with a decrease of the amount of iron removed and of both serum iron and transferrin saturation levels, which supports an increased production of hepcidin. Thus, being overweight is likely a modulating factor of iron burden in women with HFE hemochromatosis. The fact that this effect was exclusively demonstrated in women suggests a link between metabolic syndrome, hepcidin metabolism, and sex hormones.


The authors thank colleagues from the Liver Unit of Rennes for allowing the use of patient charts, the nursing staff for performing phlebotomy programs and for follow-up of patients, and Béatrice Leclerc for oversight of the administrative and family screening procedures. We thank the Centre de Ressources Biologiques of Rennes for managing patient samples.