Methods

Top Methods Results Discussion Author & Article Info References

Study Design

Outpatients attending diabetes clinics of four hospitals in northern Italy were recruited consecutively and screened for hemochromatosis. Standardization of the data collection procedures ensured data homogeneity. Criteria for the diagnosis of diabetes were as follows: fasting plasma glucose concentrations greater than 7.7 mmol/L (>140 mg/dL) on at least two occasions and plasma glucose concentrations greater than 11.1 mmol/L (>200 mg/dL) at 2 hours on an oral glucose tolerance test and on at least one other occasion during the 2-hour test after ingestion of 75 g of glucose. Patients with impaired glucose tolerance were excluded from the study.

Outpatients with nonulcer dyspepsia or the irritable bowel syndrome served as the control group and were subdivided according to decades of age. Within each decade, the male-to-female ratio was similar to that in the diabetic group. All participants gave written consent to participate in the study, which was approved by the ethics committee of IRCCS, Ospedale Maggiore, Milano, Italy.

Laboratory Methods

Serum aspartate and alanine aminotransferase, alkaline phosphatase, -glutamyltranspeptidase, and bilirubin levels were measured by using an Olympus AU510 analyzer (Eppendorf-Netheler, Hamburg, Germany); hepatitis B surface antigen was determined by using radioimmunoassay (Abbott Laboratories, North Chicago, Illinois); and antibody to hepatitis C virus was determined by using recombinant immunoblot assay II (Ortho Diagnostic System, Milan, Italy). The colorimetric method (Ferro-Check II, Hyland, Milan, Italy) was used to determine serum iron levels (reference value, 11 to 29 µmol/L [60 to 160 µg/dL]) and serum transferrin levels (reference value, 2.00 to 3.00 g/L [200 to 300 mg/dL]). Transferrin saturation was derived from the following formula: serum iron/serum transferrin x 0.80 (reference value 40). Serum ferritin levels (reference value, 30 to 300 µg/L in men and 20 to 160 µg/L in women) were measured by radioimmunoassay (Liso-Phase, Lepetit, Milan, Italy).

Complete clinical and biochemical evaluation to exclude secondary iron overload was performed in patients and controls with a transferrin saturation greater than 55% and a serum ferritin level greater than 300 µg/L (in women) or greater than 400 µg/L (in men). These measurements were documented twice in samples obtained 1 month apart. Information about alcohol intake (g/d) was obtained, and random blood alcohol levels were determined to confidently exclude alcohol abuse. In the absence of contraindications, liver biopsy was done in patients with biochemical evidence of hemochromatosis, and siderosis was estimated semiquantitatively [12] and quantitatively (reference value, <150 µg/100 mg of dry weight) [13]. A hepatic iron index (hepatic iron concentration/age) [14] greater than 1.9 was considered diagnostic for hemochromatosis [15].

All but one patient with hemochromatosis underwent venesections (220 mg of iron was removed each time). The total iron removed (in grams) was calculated by multiplying the number of venesections by 220.

Statistical Analysis

The Fisher exact test was used to compare diabetic patients with controls.

Role of Funding Source

Our funding source had no role in the collection, analysis, or interpretation of the data or in submission of the paper for publication.

Results

Top Methods Results Discussion Author & Article Info References

Of the 894 patients (418 men and 476 women), 117 had type 1 diabetes (13%; 56 men and 61 women) and 777 had type 2 diabetes (87%; 362 men and 415 women). The mean duration of diabetes was 11 ± 9 years (range, 1 to 42 years). The mean age of patients with diabetes was 62 ± 9 years (range, 18 to 70 years). The control group (467 patients) consisted of 234 men and 233 women (mean age, 60 ± 11 years [range, 18 to 70 years]). Fifteen diabetic patients (1.7% of 894) and 2 controls (0.4% of 467) had persistently elevated transferrin saturation and serum ferritin levels. Four patients were subsequently excluded from the study because of long-term alcohol abuse (2 diabetic patients and 1 control) and recurrent pulmonary infections (1 diabetic patient). The remaining 12 diabetic patients (11 men and 1 woman) had type 2 diabetes. Transferrin saturation in these patients ranged from 70% to 100%, and serum ferritin levels ranged from 620 to 3200 µg/L (Table 1).

Liver biopsy was performed in nine diabetic patients and one control. It was not performed in three of the diabetic patients because of refusal by one patient, a prolonged prothrombin time in one patient, and a low platelet count in one patient. Histologic evaluation of the liver revealed precirrhotic hemochromatosis in four patients and cirrhotic hemochromatosis in five patients. The single control with iron overload had liver fibrosis, with a hepatic iron index of 5.0. Iron depletion (that is, transferrin saturation <20% and serum ferritin level <50 µg/L in the presence of mild anemia) was achieved in seven diabetic patients, and venesections are still being done in the remaining four patients. Patient 3 did not receive treatment because of liver failure and died 3 months after the diagnosis of hemochromatosis.

The overall prevalence of hemochromatosis in this series of diabetic patients was 1.34% (95% CI, 0.7% to 2.3%) (12 of 894), compared with 0.2% (CI, 0.01% to 1.4%) in controls (1 of 467) (P = 0.032). The odds ratio of hemochromatosis in association with diabetes was 6.3 (CI, 1.1 to 37.7). When only patients with type 2 diabetes were considered, the prevalence of hemochromatosis was 1.54% (12 of 777 [95% confidence limits, 0.8 to 2.6]) (P = 0.019) and the odds ratio of hemochromatosis with type 2 diabetes was 7.3 (CI, 1.3 to 41.9).

Table 2 gives a breakdown of the number of patients with elevated iron indexes and the final diagnoses in diabetic patients and controls who underwent complete clinical evaluation, including liver biopsy in patients with elevated transferrin saturation and serum ferritin levels. The initial transferrin saturation had a positive predictive value for hemochromatosis of 39% (CI, 22% to 58%); serum ferritin level had a positive predictive value of 18% (CI, 10% to 30%). The positive predictive values of a second test were 80% for transferrin saturation (CI, 52% to 95%) and 20% for serum ferritin level (CI, 11% to 32%). The positive predictive value for both indexes at first and second testing was 80% (CI, 52% to 95%).

Discussion

Top Methods Results Discussion Author & Article Info References

In our series of diabetic patients, the prevalence of previously undiagnosed genetic hemochromatosis was statistically significantly higher than that in sex- and age-matched controls. Transferrin saturation was found to be the best screening tool for hemochromatosis. The overall prevalence (1.34%) of hemochromatosis in this series of diabetic patients is similar to the prevalence (0.96%) reported in an Australian series that used similarly strict diagnostic criteria [11]. The difference in the prevalence between our control group and our diabetic group, compared with the difference reported in the Australian study, could be attributable to the inclusion of a general population in the Australian study and the use of age- and sex-matched controls in our series. With regard to two European series of diabetic patients [9, 10], differences in the prevalence of genetic hemochromatosis between our study and the two European studies (1.34% compared with 0.49% and 0.52%, respectively) could be attributed to the use of less strict diagnostic criteria in the two European studies (for example, liver biopsy was not done for definitive diagnosis of hemochromatosis and no elevated cutpoint for iron-related indexes was used) and to the lack of a true control group.

The type of diabetes was not specified in previous series [10, 11]. Of note, all of our diabetic patients with hemochromatosis had type 2 diabetes. This finding is not surprising because in hemochromatosis-related diabetes, as in type 2 diabetes, the insulin response to a meal is impaired and the glucagon response is intact or even exaggerated.

That transferrin saturation accurately predicts hemochromatosis has been noted in the general population [16, 17] and in our study. However, serum ferritin levels can be influenced by alcohol abuse, chronic hepatitis C, tumors, and inflammatory conditions. This could be a particularly important limitation. In our series of diabetic patients, 16% had chronic infections and 54% had hepatic steatosis, with concomitant necrosis in some cases. This explains the high rate at which serum ferritin levels falsely indicated hemochromatosis.

The observation that genetic hemochromatosis had not been previously diagnosed in the 12 diabetic patients, despite the long duration of follow-up after the diagnosis of diabetes, could be attributed both to unawareness of the frequency of hemochromatosis in Italy until recent years and to the relative paucity (40%) of characteristic clinical manifestations.

Our results, which are from a limited geographic area of northern Italy, cannot be extrapolated to other geographic areas, where the prevalence of diabetes and hemochromatosis might be different. In addition, the severity of clinical manifestations of hemochromatosis may be markedly influenced by differences in genotypic and phenotypic expression [18]. For instance, the presence of ancestral haplotype, more frequent in Celtic populations, is associated with a more severe phenotype and a substantially higher degree of iron overload [19].

The severity and social costs of late complications of genetic hemochromatosis, mainly cirrhosis and hepatocellular carcinoma, make diagnosis and treatment of this disease an important goal and underscore the need for screening of patients at increased risk for the disease, such as diabetic patients. Overall, our findings indicate that the frequency of genetic hemochromatosis is underestimated, even in select higher-risk groups such as diabetic patients. The findings suggest that screening may be valuable, especially in patients with type 2 diabetes.

Drs. Guala and Manachino: Dipartimento Medicina Interna, Ospedale Civile di Vercelli, Italy.

Dr. Colli: Dipartimento Medicina Interna, Ospedale Civile di Morbegno, (SO), Italy.

Dr. Aimo: Dipartimento Medicina Interna, Ospedale di Salo, (BS), Italy.

Dr. Andreoletti: Dipartimento di Medicina Interna, Ospedale di Cernusco sul Naviglio, (MI), Italy.