Bloodletting Ameliorates Insulin Sensitivity and Secretion in Parallel to Reducing Liver Iron in Carriers of HFE Gene Mutations

  1. Francesco Equitani, MD1,
  2. Josè Manuel Fernandez-Real, MD2,
  3. Giacomo Menichella, MD1,
  4. Maurizio Koch, MD3,
  5. Menotti Calvani, MD4,
  6. Valerio Nobili, MD5,
  7. Geltrude Mingrone, MD, PHD4 and
  8. Melania Manco, MD, PHD5
  1. 1Transfusion Medicine, SanFilippo Neri Hospital, Rome, Italy
  2. 2Section of Diabetes, Endocrinology, and Nutrition, University Hospital, Girona, Spain
  3. 3Liver Unit, SanFilippo Neri Hospital, Rome, Italy
  4. 4Department of Internal Medicine, Catholic University, Rome, Italy
  5. 5Department of Hepato-Gastroenterology and Nutrition, Bambino Gesù Hospital and Research Institute, Rome, Italy
  1. Address correspondence and reprint requests to Melania Manco, MD, PhD, Department of Hepato-Gastroenterology and Nutrition, Bambino Gesù Hospital and Research Institute, S. Onofrio 4 square, 00165 Rome, Italy, E-mail: melaniamanco{at}


OBJECTIVE—To clarify the pathogenesis of diabetes associated with mutations of the hemochromatosis (HFE) gene, 17 carriers, 9 normal glucose tolerant (NGT) and 8 diabetic, were evaluated in an interventional trial.

RESEARCH DESIGN AND METHODS—At enrollment and after a 2-year bloodletting period, euglycemic-hyperinsulinemic clamp, oral glucose tolerance test (OGTT), liver histology (nonalcoholic fatty liver disease activity score [NAS]), and liver iron content (LIC) were assessed.

RESULTS—NGT subjects had significantly higher baseline insulin sensitivity (P ≤ 0.001), secretion, and insulinogenic index (calculated from the OGTT) (P ≤ 0.0001 for both) and lower LIC (P = 0.004) and NAS (P = 0.02) than diabetic patients. Baseline LIC correlated negatively with insulin secretion (NGT r0 = −0.676, P ≤ 0.0001; diabetes r0 = −0.589, P = 0.02) and insulin sensitivity (M value) (NGT r0 = −0.597, P = 0.009; diabetes r0 = −0.535, P = 0.03) and positively with NAS (diabetes r0 = 0.649, P = 0.007) and triglycerides (NGT r0 = 0.563, P = 0.015). At month 24, circulating iron was reduced by 179 ± 26% in NGT and 284 ± 54% in diabetic subjects. Insulin secretion (NGT 20 ± 4%; diabetes 33 ± 7%) and insulin sensitivity (NGT 25 ± 5%; diabetes 18 ± 3%) increased. LIC decreased in both groups (NGT 126 ± 42%; diabetes 61 ± 13%), and NAS ameliorated (NGT 65.1 ± 6.5 vs. 38.1 ± 6.83; P ≤ 0.0001; diabetes 2.1 ± 10.7 vs. 69.9 ± 10; P ≤ 0.0001).

CONCLUSIONS—Iron depletion ameliorates insulin secretion and sensitivity in NGT and diabetic carriers of HFE gene mutations. This amelioration occurs in parallel with decreased LIC and improved NAS. These results justify glucose tolerance testing and prophylactic iron depletion in asymptomatic carriers as well.

The most common iron overload disorder in the general population is type 1 hereditary hemochromatosis (HH). The disease occurs in ∼5 per 1,000 Caucasian people of northern European descent (1). It is an autosomal recessive disorder mostly due to a homozygous mutation within the hemochromatosis (HFE) gene on chromosome 6, the Cys282Tyr mutation, or, less frequently, the His63Asp mutation. The clinical significance of heterozygosis for these mutations is still controversial (1).

HH is associated with an increased rate of diabetes. Iron exerts a toxic effect on β-cells, which causes cell apoptosis and death (2,3). Iron depletion ameliorates β-cell function in patients with (4,5) and without iron overload (6,7). In HH, iron depletion has also been associated with the amelioration of insulin sensitivity in patients who have not developed diabetes (8). In liver, iron depot contributes to impaired insulin metabolism by reducing the insulin-extracting capacity (9), leading to hepatic insulin resistance (10).

Increased insulin resistance favors the progression from nonalcoholic fatty liver disease (NAFLD) to nonalcoholic steatohepatitis, fibrosis, and cirrhosis and further increasing both peripheral and hepatic insulin resistance in a vicious cycle (11). In muscle, iron interferes with glucose uptake (2). Bloodletting enhances insulin sensitivity in healthy donors (4) and patients with iron-induced insulin resistance (1214). Nonhomozygous HFE gene mutations are generally associated with a mild variant of hemochromatosis. When the disease is clinically evident, it is probably due to the occurrence of conditions such as the metabolic syndrome (12).

Therefore, the pathogenesis of diabetes in homozygous or heterozygous carriers of HFE gene mutations remains poorly understood, with both decreased insulin secretion and sensitivity being potential contributing factors. Thus, to add new insight into the pathogenesis of this topic, we studied prospectively the effects of bloodletting on insulin sensitivity and secretion, serum biochemical parameters, liver iron content (LIC), and histology in nine normal glucose tolerant (NGT) subjects and eight subjects with newly diagnosed diabetes with either HH or heterozygous HFE gene mutations before and after 2 years of bloodletting procedures.


The study participants were recruited from all consecutive male newly diagnosed HFE homozygous or heterozygous mutation carrier referrals to the Hemochromatosis Unit at the SanFilippo Neri General Hospital from January 2003 to December 2006. Healthy male subjects (n = 10), enrolled in a contemporaneous study at Catholic University (15), and healthy blood donors (n = 23) at SanFilippo Neri General Hospital served as control subjects. They were not smokers; were matched for age, BMI, and body composition; did not have biochemical signs of iron overload (ferritin <300 μg/l and transferrin saturation <45%); and had no evidence of ultrasound liver brightness.

Inclusion criteria were male sex, age 40–60 years, BMI ≤30 kg/m2, no serious associated illness (heart failure, cirrhosis, or panhypopituitarism), and no consumption of alcohol or use of medications. Only NGT and diabetic subjects were enrolled (16), whereas those with impaired glucose tolerance (three subjects) were excluded. All subjects received general nutritional counseling at baseline with prescription of a balanced diet regimen (carbohydrate 50–60%, fat 23–30%, and protein 15–20%) but no extra physical activity. A1C >7.5% at baseline or during the follow-up was considered a reason for dropping out of the study.

Diagnosis of HFE mutations (Cys282Tyr, His63Asp, and Ser65Cys) was genetically confirmed by real-time PCR. Non-HFE gene mutations (TFR2, c-HAMP, and HJV) were tested. Conversely, HFE and non-HFE gene mutations were ruled out in healthy subjects (17).

Bloodletting was performed every 2 weeks by phlebotomy or erythrocytapheresis (Hemonetics MCS-plus Blood Cell Processor; Hemonetics, Braintree, MA), and blood volume was restored to normal by hemodilution using a normal saline solution. In phlebotomy, 450 ml of blood was removed for each procedure.

Patients were admitted to the unit for clinical and iron balance evaluation every 2 weeks. In the beginning, treatment was done every 2 weeks to achieve the target value of ferritin ≤100 ng/ml and saturation ≤45% in patients with nuclear magnetic resonance (NMR) negative for iron depots in liver or ≤50 ng/ml and saturation ≤30% in patients with NMR positive for iron accumulation; thereafter, timing of procedures was based on the individual need to maintain these values. At baseline and at 24 months, body composition, insulin secretion and sensitivity, liver NMR and heart ultrasound results, liver histology, and iron content were evaluated.

The study protocol conformed to the principles of the Declaration of Helsinki and to the recommendations of the Ethics Committee at the SanFilippo Neri General Hospital. The nature and the purpose of the study were carefully explained before informed consent was requested from each patient.

Insulin secretion was calculated as the corrected insulin response from 30-min insulin and glucose levels: {insulin 30/[glucose 30 × (glucose 30–3.9)}] (milli-International Units × millimoles squared) (18). The insulinogenic index (IGI) was calculated as (insulin 30 − insulin 0)/(glucose 30 − glucose 0) (19).

Insulin sensitivity was estimated by a euglycemic-hyperinsulinemic clamp as described previously (20). Whole-body glucose uptake (M value in micromoles per kilogram free fat mass [FFM] per minute) was determined during a primed constant infusion of insulin (at the rate of 6 pmol · min−1 · kg−1). It was normalized for kilograms of FFM.

Body composition was estimated by the isotopic dilution method. FFM (in kilograms) was obtained by dividing total body water by 0.73 (21).

Histological features of steatosis, inflammation (portal and lobular), hepatocyte ballooning, and fibrosis were scored by using the scoring system for NAFLD (22). Features of steatosis, lobular inflammation, and hepatocyte ballooning were combined in a score going from 0 to 8, named the NAFLD activity score (NAS). NAS ≥5 is diagnostic of nonalcoholic steatohepatitis, NAS ≤2 is diagnostic of simple steatosis, and values in between are considered indeterminate.

Liver iron content (LIC) was measured in all subjects by means of atomic absorption spectroscopy according to the method of Barry and Sherlock (23). Liver tissue samples of at least 0.5 mg dry weight that had been frozen immediately at −20°C after collection were used.

Serum was stored at −80°C for later assays. Plasma insulin and C-peptide were measured by a specific radioimmunoassay (MYRIA Technogenetics, Milan, Italy). A1C was measured by high-performance liquid chromatography (L-9100; Hitachi, Rahway, NJ).


Data are presented as means ± SEM. Before statistical analysis, normal distribution and homogeneity of the variances were tested. Parameters that did not fulfill tests were log transformed before analysis. To evaluate the effect of 24-month bloodletting on continuous variables, a Wilcoxon signed-ranks test was used. To compare the effect of bloodletting in NGT and diabetic subjects separately at baseline and at the end of the follow-up, a Mann-Whitney U test was performed. To compare HFE gene mutation carriers versus healthy subjects and to evaluate the effect of blood withdrawn throughout the follow-up, it was necessary to perform repeated ANOVA and post hoc analysis (Bonferroni post hoc test), whenever appropriate.

Relationships between variables were determined by linear correlation analysis (Spearman's r), and regression analysis was performed by standard techniques. Levels of statistical significance were set at P < 0.05. Data analyses were performed with SPSS statistical software (V12.0; SPSS, Chicago, IL).


Comparison of carriers of HFE gene mutations versus control subjects

Control subjects were matched with both diabetic and NGT groups of HFE gene mutation carriers in age, BMI, and FFM. Fasting glucose and insulin did not differ. Control subjects had significantly lower total cholesterol, triglycerides, liver enzymes, serum iron, and transferrin compared with NGT and diabetic patients at baseline and after 2 years (statistical significance shown in Table 1). Insulin secretion was significantly lower in diabetic subjects (P ≤ 0.05, both at baseline and after 2 years). Insulin sensitivity was greater in control subjects than in diabetic subjects before (P ≤ 0.0001) and after iron depletion (P ≤ 0.0001).

Baseline values in groups of carriers of HFE gene mutations

At baseline, NGT and diabetic subjects differed significantly in fasting glucose (P ≤ 0.0001), A1C (P ≤ 0.0001), triglycerides (P = 0.05), insulin sensitivity (P = 0.001) and secretion, and IGI (P ≤ 0.0001 for both) (Table 1). No significant differences were found in aminotransferases, whereas LIC (P = 0.004) and NAS (P = 0.02) were significantly higher in diabetic subjects (Table 2).

Effect of the treatment in carriers of HFE gene mutations

No subject with diabetes dropped out the study because of A1C >7.5%, and no subjects discontinued the study. The goal values of ferritin and saturation were obtained in all subjects by month 5. To achieve these values, 8 ± 3 procedures in NGT vs. 9 ± 2 procedures in diabetic subjects were needed. Throughout the whole study, NGT subjects underwent a total of 21 ± 2 vs. 20 ± 1 procedures in diabetic subjects.

A significant improvement in several metabolic parameters was seen as soon as the treatment was initiated. At month 4, ferritin (1,039.4 ± 129.8 vs. 264.1 ± 27.8 ng/ml; P ≤ 0.0001), iron (130.3 ± 8.9 vs. 67.5 ± 5.8 μg/dl; P ≤ 0.0001), and saturation (67 ± 3 vs. 42 ± 2%; P ≤ 0.001) significantly decreased. Cholesterol (243.9 ± 9.0 vs. 209.1 ± 7.5 mg/dl; P = 0.008), triglycerides (210.8 ± 10.2 vs. 168.35 ± 9.1 mg/dl; P = 0.005), fasting glycemia (115.1 ± 6.1 vs. 92.1 ± 4.6 mg/dl; P = 0.01), lactate dehydrogenase (329.8 ± 18.1 vs. 282.7 ± 11.5 IU/l; P = 0.05), aspartate aminotransferase (AST) (43.1 ± 2.7 vs. 32.1 ± 2.3 IU/l; P = 0.01), alanine aminotransferase (ALT) (67.1 ± 4.1 vs. 45.2 ± 2.4 IU/l; P = 0.001), and γ-glutamyltransferase (γ-GT) (62.9 ± 4.3 vs. 39.8 ± 4.1 IU/l; P ≤ 0.0001) significantly ameliorated. Time courses of the most representative variables are depicted separately for NGT and diabetic subjects in Fig. 1.

At the end of the study, bloodletting was associated with an improved metabolic pathway. Compared with baseline, ferritin decreased to 58.1 ± 4.2 ng/ml (P ≤ 0.0001) and serum iron to 41.0 ± 1.5 μg/dl (P ≤ 0.0001). Cholesterol (201.2 ± 4.3 mg/dl; P ≤ 0.0001), lactate dehydrogenase (295.8 ± 13.5 IU/l; P = 0.004), AST (30.8 ± 2.0 IU/l; P ≤ 0.0001), ALT (43.8 ± 3.2 IU/l; P ≤ 0.0001), γ-GT (41.6 ± 3.9 IU/l; P ≤ 0.0001), triglycerides (146.5 ± 8.3 mg/dl P ≤ 0.0001), and fasting glycemia (94.9 ± 4.6 mg/dl; P = 0.01) ameliorated in all subjects. Table 1 shows mean ± SEM values separately for NGT and diabetic subjects.

For insulin metabolism (Table 1), after 2 years, insulin secretion increased by 20 ± 4% in NGT and 33 ± 7% in diabetic subjects (NS); IGI increased by 16 ± 5 and 47 ± 19%, respectively (NS); and glucose uptake increased by 25 ± 5 and 18 ± 3% (NS). Diabetic subjects had significantly lower insulin secretion, IGI, and insulin sensitivity than NGT subjects (P ≤ 0.0001 for all parameters before and after treatment). LIC decreased by 46 ± 2 in NGT and by 35 ± 2% in diabetic subjects. Figure 2 shows changes in insulin sensitivity, secretion, IGI, and LIC for each subject from both groups.

Table 2 shows histological findings, LIC, and NAS in NGT and diabetic subjects at baseline and after 24 months. No subject had biopsy-proven cirrhosis. In diabetic subjects, liver steatosis (P = 0.02) and inflammation (P = 0.03) were significantly reduced with respect to baseline.

Statistical correlations

Levels of circulating iron correlated significantly with ALT (r0 = 0.703; P ≤ 0.0001), AST (r0 = 0.765; P ≤ 0.0001), γ-GT (r0 = 0.771; P ≤ 0.0001), C-peptide (r0 = −0.361; P = 0.015), LIC (r0 = 0.484; P = 0.004), glucose uptake (r0 = −0.360; P = 0.004), NAS (r0 = 0.364; P = 0.03), cholesterol (r0 = 0.573; P ≤ 0.0001), triglycerides (r0 = 0.679; P ≤ 0.0001), IGI (r0 = 0.337; P = 0.002), and HbA2 (r0 = 0.301; P = 0.03). Significant correlations were found between ferritin and AST (r0 = 0.778; P ≤ 0.0001), ALT (r0 = 0.592; P ≤ 0.0001), γ-GT (r0 = 0.735; P ≤ 0.0001), C-peptide (r0 = −0.314; P = 0.04), LIC (r0 = 0.40; P = 0.02), NAS (r0 = 0.393; P = 0.02), cholesterol (r0 = 0.418; P ≤ 0.0001), triglycerides (r0 = 0.629; P ≤ 0.0001), IGI (r0 = 0.255; P = 0.02), and HbA2 (r0 = 0.289; P = 0.04).

In NGT (y = −0.2376x + 50.286; R2 = 0.40, P ≤ 0.0001) and diabetic subjects (y = −0.2905x + 55.425; R2 = 0.41, P ≤ 0.0001), LIC correlated with glucose uptake. In NGT subjects, LIC correlated also with triglycerides (r0 = 0.563; P = 0.015) and negatively with insulin secretion (r0 = −0.676; P ≤ 0.0001).

In diabetic subjects, LIC correlated with NAS (r0 = 0.649; P = 0.007). IGI (r0 = −0.589; P = 0.016) and triglycerides (r0 = 0.775; P ≤ 0.0001) were related to ALT levels.

In a multistep linear regression analysis, the best predictors of whole-body glucose uptake in HFE gene mutation carriers (R2 = 0.55; P ≤ 0.0001) were LIC (β = −0.549; P ≤ 0.0001) and triglycerides (β = −0.307; P = 0.03). BMI, ferritin, iron, ALT, insulin secretion, IGI, and NAS were excluded variables.


In patients with phenotypic appearance of HFE gene mutations, iron overload is associated with impaired insulin metabolism and features of metabolic syndrome. β-Cell function deteriorates to a larger extent than insulin sensitivity in the early development of diabetes related to the HFE gene mutations. Insulin resistance occurs when glucose tolerance is still preserved and becomes more severe later when glucose tolerance is impaired. Thus, both aspects contribute significantly to the pathogenesis of diabetes in these patients. Constant iron depletion ameliorates all biochemical and histological features associated with metabolic syndrome. β-Cell activity ameliorates as well as glucose uptake. Subjects with the latter keep showing lower levels of secretion and sensitivity compared with NGT patients even after achieving a normalization of iron balance.

These findings confirm recent data from McClain's group (7,8), but they also add significant insights to our knowledge of the pathophysiology of diabetes associated with HFE gene mutations. In a small cohort of selected homozygous patients, McClain et al. found NGT subjects having normal insulin secretion and sensitivity, whereas subjects with impaired glucose tolerance compensated for diminished insulin secretion by increased insulin sensitivity. Subjects in whom this compensatory mechanism failed seemed to be those who developed overt diabetes, showing lower insulin secretion rates but normal glucose uptake (8).

In the present cohort of homozygous and heterozygous subjects, we observed values of glucose uptake in NGT subjects overlapping with those of healthy volunteers, thus confirming a previous report in nondiabetic, noncirrhotic carriers of heterogeneous HFE gene mutations (24). In diabetic subjects both insulin sensitivity and secretion are significantly impaired. It is conceivable that the lower value of glucose uptake that we observed in our diabetic subjects might be due to the genetic heterogeneity of our cohort compared with that of Klein's series.

Furthermore, values of glucose uptake, apparently within a normal range in NGT subjects, are further enhanced after iron depletion, as already described in healthy frequent blood donors (4), suggesting that high levels of circulating iron can reduce the potential insulin sensitivity of a subject. In healthy volunteers, 75% of glucose uptake occurs in muscle and the remaining 25% in liver. It has been hypothesized that these proportions may be modified in presence of iron overload (25). Accordingly, we found that the liver iron depot correlates with systemic glucose uptake and its reduction is associated with the amelioration of histological grading and NAS (Table 2). In this regard, NGT carriers of HFE gene mutations have a mean value of liver iron content that is below the threshold commonly considered as diagnostic for hemochromatosis, and this value is significantly reduced after treatment. In diabetic subjects, who have higher levels of LIC than NGT subjects, iron depletion is associated with the improvement of all parameters related to insulin metabolism, liver function, and histology. The decrease in liver iron content is associated with the significant amelioration of steatosis and inflammation, perhaps because iron is able to participate in the formation of powerful oxidant species (26). These differences in outcomes may be due to the multifactorial nature of the transition from normal glucose tolerance to impaired glucose tolerance and diabetes in carriers of HFE gene mutations. Overweight/obesity-related steatosis, high levels of triglycerides, or the presence of obesity-related morbidities may play a role as cofactors.

In summary, we have demonstrated that the loss of insulin secretory capacity can be the primary event leading to HFE gene mutation–related diabetes, but the role for decreased insulin sensitivity cannot be disregarded, with liver acting as a primary site of insulin resistance and contributing to the ultimate expression of diabetes. Increased body fatness, environmental factors, and risk factors for metabolic syndrome can work together. Thus, obesity and related morbidities should be actively addressed in the management of carriers of HFE gene mutations. Especially important for the prevention of diabetes-related morbidities that can occur very early, this would justify glucose tolerance testing of carriers of HFE gene mutations as well as prophylactic iron depletion in unaffected individuals with diagnosed disease. The reversibility of iron-related abnormalities in insulin metabolism warrants further investigation.

Figure 1—

Time courses of plasma triglycerides (A) and glucose (B), serum ferritin (C), iron (D), and levels of ALT (E) in NGT (dotted line) and diabetic (solid line) subjects with HFE gene mutations. Levels of significance at post hoc analysis: °P ≤ 0.01; §P ≤ 0.001; +P ≤ 0.0001.

Figure 2—

Individual changes in insulin secretion (A), IGI (B), whole-body insulin uptake (M value) (C), and LIC (D) in NGT (dotted line) and diabetic (solid line) subjects carrying HFE gene mutations. Levels of significance at post hoc analysis: ‡P ≤ 0.05; °P ≤ 0.01; §P ≤ 0.001; +P ≤ 0.0001.

Table 1—

Anthropometric and biochemical variables in healthy control subjects and subjects carrying HFE gene mutations according to glucose tolerance

Table 2—

Histological findings, NAS, and LIC before and 24 months after bloodletting in NGT and diabetic subjects carrying HFE gene mutations


  • Published ahead of print at on 24 October 2007. DOI: 10.2337/dc07-0939. Clinical trial reg. no. NCT00440986,

    A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.

    The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C Section 1734 solely to indicate this fact.

    • Accepted September 16, 2007.
    • Received May 17, 2007.


| Table of Contents

This Article

  1. Diabetes Care vol. 31 no. 1 3-8
  1. All Versions of this Article:
    1. dc07-0939v1
    2. 31/1/3 most recent