Hemochromatosis in Italy in the last 30 years: Role of genetic and acquired factors

Authors

  • Anna Ludovica Fracanzani,

    1. Department of Internal Medicine, Policlinico Mangiagalli Regina Elena Hospital Foundation IRCCS, University of Milan, Italy and Center for the study of Metabolic Liver Disease, Policlinico Mangiagalli Regina Elena Hospital Foundation IRCCS, University of Milan, Italy
    Search for more papers by this author
  • Alberto Piperno,

    1. Department of Internal Medicine, S. Gerardo Hospital, Monza, University of Milan Bicocca, Italy
    Search for more papers by this author
  • Luca Valenti,

    1. Department of Internal Medicine, Policlinico Mangiagalli Regina Elena Hospital Foundation IRCCS, University of Milan, Italy and Center for the study of Metabolic Liver Disease, Policlinico Mangiagalli Regina Elena Hospital Foundation IRCCS, University of Milan, Italy
    Search for more papers by this author
  • Mirella Fraquelli,

    1. Department of Gastroenterology, Policlinico Mangiagalli Regina Elena Hospital Foundation IRCCS, University of Milan, Italy
    Search for more papers by this author
  • Sabina Coletti,

    1. Department of Internal Medicine, S. Gerardo Hospital, Monza, University of Milan Bicocca, Italy
    Search for more papers by this author
  • Alessandra Maraschi,

    1. Department of Internal Medicine, Policlinico Mangiagalli Regina Elena Hospital Foundation IRCCS, University of Milan, Italy and Center for the study of Metabolic Liver Disease, Policlinico Mangiagalli Regina Elena Hospital Foundation IRCCS, University of Milan, Italy
    Search for more papers by this author
  • Dario Consonni,

    1. Department of Epidemiology Unit, Policlinico Mangiagalli Regina Elena Hospital Foundation IRCCS, University of Milan, Italy
    Search for more papers by this author
  • Enzo Coviello,

    1. Statistics and Epidemiology Unit ASL Bari, Italy
    Search for more papers by this author
  • Dario Conte,

    1. Department of Gastroenterology, Policlinico Mangiagalli Regina Elena Hospital Foundation IRCCS, University of Milan, Italy
    Search for more papers by this author
  • Silvia Fargion

    Corresponding author
    1. Department of Internal Medicine, Policlinico Mangiagalli Regina Elena Hospital Foundation IRCCS, University of Milan, Italy and Center for the study of Metabolic Liver Disease, Policlinico Mangiagalli Regina Elena Hospital Foundation IRCCS, University of Milan, Italy
    • Dipartimento Medicina Interna, Fondazione Ospedale Policlinico Mangiagalli e Regina Elena IRCCS, Via F Sforza 35, 20122 Milan, Italy===

    Search for more papers by this author
    • fax: (39)-2-50320296


  • Potential conflict of interest: Nothing to report.

Abstract

The clinical presentation of hereditary hemochromatosis has changed markedly in recent years. The aim of this study was to analyze a large series of consecutive Italian patients with hemochromatosis diagnosed between 1976 and 2007 to determine whether the genetic background and the presence of acquired risk factors influenced the severity of iron overload and the natural history of the disease. A cohort of 452 Italian patients with iron overload—338 HFE-related (C282Y homozygotes or compound C82Y/H63D heterozygotes) and 114 non–HFE-related—were followed prospectively for a median of 112 months. Alcohol intake, smoking habits, and iron removed to depletion were similar in patients with and without HFE-related iron overload. Hepatitis B virus (4% and 9%; P = 0.04) and hepatitis C virus (6% and 19%; P = 0.002) infections were more frequent in patients with non–HFE-related iron overload. Seventy-three percent of patients with HFE and 61% of patients with non–HFE-related disease had no acquired risk factor. Cirrhosis was significantly more frequent in non-HFE patients independent of the presence of acquired risk factors (P = 0.02). Sex, alcohol intake, prevalence of smoking, hepatitis C virus infection, glucose, lipids, iron-related parameters, and prevalence of C282Y/H63D differed significantly over the years. At enrollment, cirrhosis was present in 145 cases and was significantly more frequent in the first decade (80%, 47%, and 13%; P = 0.001). Survival did not differ across the decades in cirrhotic patients; hepatocellular carcinoma occurred similarly in HFE and non-HFE patients. Conclusion: Patients with HFE and non–HFE-related iron overload have comparable iron overload and similar clinical history. Patients who were diagnosed during the last 10 years and were not identified as cirrhotic at enrollment have less severe disease and lower prevalence of acquired risk factors, independent of genetic background. (HEPATOLOGY 2010;51:501–510.)

Hereditary hemochromatosis is an autosomal recessive disease that is characterized by progressive iron overload due to increased iron absorption and is responsible for damage to the liver, heart, joints, pancreas, and endocrine glands. Mutations in the HFE gene are detected in the large majority of patients of northern European descent and cause reduced release of hepcidin, the hepatic hormone that inhibits iron absorption by interacting with ferroportin-1, which is responsible for iron export from the basal membrane of duodenocytes and from macrophages.1, 2

In recent years, other mutations in non–HFE-related genes, which are responsible for very rare forms of hereditary hemochromatosis, have been described,3, 4 but attempts to identify these mutations in patients with non–HFE-related hemochromatosis have proven unsuccessful in the large majority of patients of Mediterranean origin.5 In addition, it has become clear through epidemiological studies that penetrance of C282Y homozygosity, the typical genotype associated with hemochromatosis, is very low, suggesting that the C282Y/C282Y genotype is not sufficient to cause phenotypically expressed hereditary hemochromatosis.6–11 Thus, host-related factors (such as coinheritance of modifier genes) and acquired factors have been suggested to play a prominent role in determining the clinical manifestations and the natural history of hereditary hemochromatosis.12–15

Significant improvements in the diagnosis and management of hereditary hemochromatosis that have the potential to influence natural history by preventing complications (particularly liver cancer, the first cause of death in cirrhotic patients) have been made in recent years.16–21 Some studies22–24 have reported that better awareness of hereditary hemochromatosis by physicians and widely available genetic tests allowing identification of C282Y+/+ subjects at a preclinical stage leads to an earlier diagnosis with consequently milder organ damage (when present). However, no attention has been paid to the possible different role of cofactors throughout the years.

The aim of the present study was to analyze a large series of consecutive Italian patients with hemochromatosis diagnosed between 1976 and 2007 to define whether the genetic background and the presence of acquired risk factors influenced the severity of iron overload and the natural history of the disease across the years.

Abbreviations

CI, confidence interval; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; HR, hazard ratio.

Patients and Methods

We retrospectively reviewed a prospectively recruited clinical database of patients referred for iron overload to three reference centers for hereditary hemochromatosis in northern Italy from 1976 to 2007 (median, 96 months [range, 12-384 months]). We identified a total of 452 patients with clinical, biochemical, and/or histological evidence of hemochromatosis phenotype as defined by classical diagnostic criteria,25, 26 including hepatic iron concentration, transferrin saturation percentage, serum ferritin, and iron removed to reach depletion and the absence of other known causes of iron overload in subjects with or without C282Y/C282Y or C282Y/H63D HFE genotypes. Many of these patients had been included in a previous Italian study.5 The 269 patients with C282Y homozygotes and the 69 patients with compound heterozygotes (C282Y/H63D) were identified as having HFE-related hemochromatosis. Among this group, patients diagnosed before 1996 underwent liver biopsy, unless contraindicated, with determination of hepatic iron concentration and hepatic iron index, after 1996 liver biopsy was performed when the ferritin level was >1,000 ng/mL in the presence of increased alanine aminotransferase according to international guidelines.25–27 The remaining 114 patients were defined as non–HFE-related iron overload, and 90 underwent liver biopsy. The 24 patients who were not biopsied had clinical evidence of cirrhosis (n = 5), altered blood coagulation tests (n = 4), noncompliance to biopsy (n = 15), with 13 of the last group having ferritin levels >1,000 ng/mL. To support the diagnosis of hemochromatosis in the non–HFE-related patients, iron overload was analyzed in relation to the presence of acquired risk factors for liver disease and severity of the disease. Seventy-five percent of the patients originated from northern Italy; 25% originated from southern Italy. Familiarity for hemochromatosis was present in 8% of patients with non–HFE-related hemochromatosis and in 17% of patients with HFE-related hemochromatosis. Patients were evaluated according to three decades of enrollment (1977-1986, 1987-1996, and 1997-2007).

Complete medical history, routine biochemical and serological data, HFE genotype, liver ultrasonography, and physical examination reports at diagnosis were available in all cases, as well as a careful estimation of daily alcohol consumption and smoking habits in the previous 5 years. In all patients, alcohol intake was reported in grams per day, and alcohol abuse was defined as a daily alcohol intake >50 g in men and >40 g in women. Daily alcohol intake or binge drinking was recorded by interviewing all patients and/or their relatives, and a questionnaire was recorded with history of alcohol intake for at least 5 years. The information on alcohol intake beyond 5 years was also available in ≈35% of patients. All patients were submitted to an iron-depleting regimen by weekly phlebotomy up to iron depletion, and thereafter to maintenance phlebotomy.

Regular follow-up included physical examination, routine blood tests, ultrasonography performed yearly in patients without cirrhosis, and twice a year in subjects with cirrhosis at enrollment.

Hepatocellular Carcinoma Surveillance.

All patients with cirrhosis were prospectively evaluated at least every 6 months by way of physical examination, liver ultrasound, and serum alpha-fetoprotein measurements. When these investigations suggested a possible diagnosis of hepatocellular carcinoma (HCC), computed tomography, magnetic resonance imaging, and/or fine needle biopsy of the lesion were performed according to the recommendation of the Barcelona Conference.28 HCC was diagnosed based on one of the following criteria: histological evidence of hepatocellular carcinoma, evidence or convergent demonstration of a focal lesion >2 cm in size with arterial hypervascularization by two different imaging techniques, or the combination of one imaging technique showing this morphologic aspect with a serum alpha-fetoprotein level of ≥400 ng/mL.

Fifty-five patients, corresponding to 12% of the entire series (five of whom were cirrhotic) were lost at follow-up, and pertinent data were censored at the time of the last observation. All the other patients underwent regular controls. All patients had given informed written consent to data handling according to a protocol approved by the Senior Staff Committee of our institutions, a board comparable to an Institutional Review Board.

A search was conducted for anti–hepatitis B core, hepatits B surface antigen, and anti–hepatitis C virus (HCV) for all patients. When positive, hepatitis B virus (HBV) DNA and HCV RNA were also determined. Serum iron and total iron binding capacity were determined by standard colorimetric methods, and transferrin saturation percentage was calculated. Serum ferritin was evaluated by way of enzyme immunoassay (Enzymun Ferritin ES, Boehringer Mannheim, Mannhein, Germany).

HFE Mutations.

C282Y and H63D HFE mutations were searched for in genomic DNA extracted from peripheral leukocytes by way of restriction analysis29 in all patients. In patients without C282Y homozygotes or compound heterozygotes (C282Y/H63D), the S65C HFE mutation was searched for by way of restriction analysis,29 ferroportin (SLC40A1) and transferrin receptor 2 gene mutations were searched for by way of direct sequencing,30, 31 and hemojuvelin and hepcidin mutations were searched for by way of denaturing high-performance liquid chromatography followed by sequencing in case of positive results or direct sequencing according to the center protocol.

Liver Biopsy.

For baseline routine histology, liver biopsies were fixed in 10% formalin, embedded in paraffin, and stained with hematoxylin-eosin, Perls' Prussian blue for storage iron, and Masson's trichrome for connective tissue. Hepatic iron concentration was measured by way of atomic absorption spectroscopy. The hepatic iron index was calculated as iron concentration divided by age. In the presence of focal lesions, ultrasound-guided liver samples were obtained from both lesional and extralesional tissue.

Statistical Analysis.

Distribution of variables across decades was compared using the chi-square test and analysis of variance when appropriate, and P values were reported. Survival was evaluated with the product-limit estimator method as described by Kaplan and Meier.32 Follow-up was censored at January 2008. The observed survival curve was performed in patients stratified according to the three decades of diagnosis. In addition, the following subgroups of patients were compared with each other: (1) C282Y homozygotes and compounds versus others (heterozygous for C282Y, H63D, S65C, or patients carrying mutations for hemojuvelin, hepcidin, ferroportin and transferrin receptor 2 genes and patients wild-type for all mutations), defined as HFE versus non-HFE patients; (2) patients with or without acquired risk factors; and (3) cirrhotic versus noncirrhotic patients. Survival in these subgroups was compared using the log-rank test.33

The probability of being HCC-free was also evaluated with the product-limit estimator method as described by Kaplan and Meier32 by censoring follow-up when HCC was diagnosed. The hazard ratio (HR) of death and HCC development for the main clinical variables was estimated by way of Cox regression analysis, including disease covariates (viral infection, alcohol intake, smoking habits, iron removed to complete depletion, and age at diagnosis). Results were considered significant at P < 0.05. Analysis was performed by the JMP 6.0 statistical analysis software (SAS, Cary, NJ).

Expected survival was the survival in a reference population similar to the cohort of patients at the start of follow-up, where the matching factors were age, calendar time, and sex. This estimate was obtained through the population mortality rate tables that in this case has been represented by the mortality rates in the population of the region Lombardia.The expected survival has been first computed for each individual of the cohort and then combined in an overall estimate according to the Ederer method.34

Results

Characteristics of Patients with HFE and Non–HFE-Related Iron Overload.

The characteristics of the patients studied are detailed in Table 1. Alcohol intake, smoking habits, and amount of iron removed up to iron depletion were similar in patients with or without HFE-related iron overload, whereas a significantly higher prevalence of chronic HBV infection (13 [4%] versus 10 [9%]; P = 0.04) and HCV infection (20 [6%] versus 22 [19%]; P = 0.002) was observed in non–HFE-related hemochromatosis. Seventy-three percent of patients with HFE and 61% of patients with non–HFE-related disease had no detectable acquired risk factor. A significantly higher prevalence of cirrhosis (40% versus 26%; P = 0.001) was present, at diagnosis, in non–HFE-related hemochromatosis.

Table 1. Characteristics of Patients with HFE-Related and Non–HFE-Related Iron Overload
CharacteristicsHFE-Related (n = 338)Non–HFE-Related (n = 114)P Value
  • Abbreviations: ALT, alanine aminotransferase; BMI, body mass index; HDL, high-density lipoprotein; NS, not significant.

  • All data were acquired at the time of diagnosis. Values are expressed as n (%) or mean ± standard deviation.

  • *

    Available in 260 patients who underwent liver biopsy.

Male sex247 (73)90 (78)NS
Age, years47 ± 1349 ± 11NS
BMI, kg/m224 ± 323 ± 3NS
Alcohol, g/day29 ± 4129 ± 28NS
Smoker101 (29)45 (40)NS
ALT, UI/L49 ± 4760 ± 52NS
Transferrin saturation, %81 ± 2074 ± 260.01
Serum ferritin, ng/dL1,609 ± 1,7161,767 ± 1,331NS
Hepatic iron concentration,* age in years8.2 ± 146.1 ± 5NS
Iron removed, g8.9 ± 79.5 ± 6NS
Fasting glucose, mg/dL99 ± 29102 ± 37NS
Total cholesterol, mg/dL183 ± 45175 ± 54NS
HDL cholesterol, mg/dL52 ± 2040 ± 200.001
Tryglicerides, mg/dL131 ± 103123 ± 93NS
Acquired factors   
 Alcohol intake (>50 g/day in men; >40 g/day in women)53 (16)19 (17)NS
 HBV13 (4)10 (9)0.04
 HCV20 (6)22 (19)0.002
 HBV/HCV/ alcohol (either one)91 (27)45 (39)0.01
Clinical manifestations   
 Cirrhosis90 (26)50 (40)0.007
 Diabetes54 (16)15 (13)NS
 Arthropathy51 (15)18(16)NS
 Hypogonadism51(15)14(12)NS
 Cardiopathy40(12)10(8)NS

Influence of HBV, HCV, Alcohol Abuse and Cirrhosis on Iron Overload in Patients with or Without HFE-Related Iron Overload.

To support the diagnosis of hemochromatosis in the 114 (25%) patients with non–HFE-related iron overload, we compared iron status in relation to the presence of acquired risk factors and of cirrhosis in patients with or without HFE-related iron overload (Table 2). As expected, patients with cirrhosis either positive or negative for hemochromatosis-related genotypes had significantly more severe iron overload than noncirrhotic patients independent of the presence of acquired risk factors. Noncirrhotic patients with acquired risk factors had a similar degree of iron overload independent of the genetic background. A similar proportion of patients positive for acquired factors had cirrhosis in the HFE-related (50%) and non–HFE-related iron overload (58%) subjects, and a higher proportion of subjects with non–HFE-related iron overload with no acquired risk factor had cirrhosis (38 versus 23%; P = 0.02).

Table 2. Iron Status in HFE-Related and Non–HFE-Related Hemochromatosis Patients Divided According to the Presence of Cirrhosis and of Acquired Risk Factors
Iron StatusHFE-RelatedNon-HFE-Related
Cirrhosis Present (n = 90)Cirrhosis Absent (n = 248)Cirrhosis Present (n = 50)Cirrhosis Absent (n = 64)
NegativePositiveNegativePositiveNegativePositiveNegativePositive
  1. All data were acquired at the time of diagnosis.

  2. a versus c, P = 0.001.

  3. b versus c, P = 0.01.

  4. d versus e, P = 0.01.

Hepatic iron concentration13 ± 189 ± 45.5 ± 3.68.5 ± 68.8 ± 7.68.1 ± 25.6 ± 4.65.5 ± 1.9
Iron removed16 ± 9a14 ± 85.6 ± 3.5b9.8 ± 7c13 ± 8d10 ± 77.6 ± 5.4e7.6 ± 4

Characteristics of Patients Across the Three Decades of Study.

Table 3 outlines the characteristics of the patients with or without HFE-related iron overload according to the time interval of enrollment. Age at diagnosis did not differ across the three decades in HFE and non–HFE-related hemochromatosis patients (47 ± 11 versus 46 ± 12 years, 47 ± 14 versus 51 ± 10 years, and 47 ± 10 versus 49 ± 11 years).

Table 3. Demographic, Biochemical Viral, and Clinical Characteristics Significantly Different in 338 HFE-Related and 114 Non–HFE-Related Hemochromatosis Cases According to Decade of Enrollment
CharacteristicsHFE-Related Hemochromatosis (n = 338)Non–HFE-Related Hemochromatosis (n = 114)
1977–1986 (n = 30)1987–1996 (n = 102)1997–2007 (n = 206)P Value1977–1986 (n = 12)1987–1996 (n = 50)1997–2007 (n = 52)P Value
  1. All data were acquired at the time of diagnosis. Values are expressed as n (%) or mean ± standard deviation unless indicated otherwise.

Sex (M/F)26/477/25114/62NS8/445/537/150.03
Alcohol intake, g/day64 ± 7833 ± 4121 ± 270.00123 ± 2238 ± 3220 ± 210.008
Smoker53%27%26%0.0241%55%24%0.01
Fasting glucose, mg/dL114 ± 33104 ± 3595 ± 240.004115 ± 6197 ± 33104 ± 34NS
Total cholesterol, mg/dL155 ± 47176 ± 50191 ± 400.002156 ± 20170 ± 71186 ± 51NS
HBV-positive3 (10)4 (4)4 (2)NS2 (18)4 (8)4 (8)NS
HCV-positive4 (13)11 (11)5 (2)0.0066 (54)12 (25)1 (2)0.001
Serum ferritin, ng/mL2,427 ± 1,4702,380 ± 2,3601,094 ± 1,0400.0012,437 ± 1,3591,790 ± 1,5201,570 ± 1,0390.05
Iron removed, g18 ± 1111 ± 6.95.6 ± 40.00111 ± 5.79.6 ± 68.7 ± 60.08
Proportion of patients with:        
 Acquired risk factors43%30%15%0.00170%56%12%0.001
 Cirrhosis73%45%10%0.000183%52%23%0.0001
 Arthopathy33%22%9%0.00157%9%15%0.03
 Diabetes/oral glucose intolerance40%18%5%0.00128%17%13%NS
 Cardiopathy35%21%6%0.00125%0%12%0.02
 Hypogonadism56%22%8%0.000175%6%11%0.01

During the three decades, only 12% of the whole series of patients (five of whom were cirrhotics) were lost at follow-up; all the other patients underwent regular controls. Seventy patients died during follow-up, 14 without cirrhosis and 56 with cirrhosis. Liver cancer, which developed in 51 cirrhotics and corresponded with a rate of 3.9% per year considering the cirrhotic population, was the primary cause of death. Thirty-nine out of 70 (56%) patients died of HCC, with a similar distribution of death across the three decades.

The incidence of HCC was significantly higher in the first decade, but the proportion of cirrhotic patients who developed HCC did not differ significantly across the three decades. HCC developed only in patients with cirrhosis at enrollment, except in three HCV-infected patients who, despite iron depletion, developed cirrhosis at the time of HCC diagnosis.

Survival.

Observed and expected cumulative survival of the overall series of patients with hemochromatosis is shown in Fig. 1. Survival was significantly reduced in patients with hemochromatosis when compared with the expected survival rates for an age- and sex-matched normal population. Survival was significantly reduced in cirrhotic patients compared with noncirrhotic patients, and the mean age of patients who were cirrhotic at diagnosis was significantly higher than that of noncirrhotic patients (50 ± 11 versus 46 ± 13 years, P = 0.0002).

Figure 1.

Observed (solid line) and expected (dotted line) cumulative survival of the overall series of patients with hemochromatosis during 30 years of follow-up. Survival was significantly reduced in patients with hemochromatosis when compared with the expected survival rates for an age- and sex-matched normal population.

Estimated cumulative survival in cirrhotic patients with HFE- or non–HFE-related iron overload did not differ significantly (P = 1.14) (Fig. 2A). Fig. 2B and Fig. 2C show the cumulative survivals in patients with non–HFE- or HFE-related iron overload divided according to the presence or absence of viral infection and/or alcohol abuse. No significant difference was observed in non–HFE-related iron overload (P = 0.14) (Fig. 2B). The only significant difference was observed in the group of cirrhotic patients with HFE-related hemochromatosis; those negative for acquired risk factors had a better survival (P = 0.02).

Figure 2.

Kaplan–Meier estimates for survival in cirrhotic patients with hemochromatosis. (A) HFE-related (circles) and non–HFE-related (solid lines) hemochromatosis. (B) Non–HFE-related patients and (C) HFE-related patients divided according to the presence or absence of acquired factors. A significant difference was detected between HFE-related hemochromatosis with or without risk factors (P = 0.02). Figures at the bottom refer to the number of patients still under observation at each time interval.

Observed and expected survival for the entire cohort (452 cases) according to the decade of enrollment are shown in Fig. 3. The observed and expected survival in the third decade are similar. A significantly better survival was observed in patients diagnosed in the last decade compared with the other decades (P = 0.0001).

Figure 3.

Observed (solid lines) and expected (dotted lines) survival in the overall series of patients with hemochromatosis according to the decade of enrollment (red, 1977-1986; blue, 1987-1996; black, 1997-2007). A significant difference across the three decades in the overall series (P = 0.0001). The observed and expected survival of patients of the last decades are similar. Figures at the bottom refer to the number of patients still under observation at each time interval.

Fig. 4 shows the survival of cirrhotic patients in the three decade. No difference was observed when only cirrhotic patients were analyzed (P = 0.9). Forty-five out of the 258 patients (17%) diagnosed during the third decade had a follow-up longer than 10 years, and 10 of them had cirrhosis.

Figure 4.

Kaplan–Meier estimates for survival in cirrhotic patients according to the decade of enrollment (line, 1977-1986; square, 1987-1996; circle, 1997-2007). No significant differences was detected across the decades. Figures at the bottom refer to the number of patients still under observation at each time interval.

A Cox proportional hazards regression model was fitted to the data in order to identify those factors independently associated with survival. Variables significantly associated with survival were age (HR 1.03, 95% confidence interval [CI] 1.00-1.07 per year; P = 0.03) and smoking habits (HR 2.1, 95% CI 1.14-3.8; P = 0.02 for smokers versus nonsmokers).

Hepatocellular Carcinoma.

The cumulative probability of being HCC-free according to the decade of enrollment is shown in Fig. 5, in (A) the overall series of cirrhotic patients, (B) non–HFE-related hemochromatosis, and (C) HFE-related hemochromatosis. HCC developed at a similar rate in cirrhotic patients with HFE (P = 0.07) or non-HFE (P = 0.8) iron overload. No significant difference in the probability of being HCC-free was observed when patients were divided according to the presence or absence of viral infections and/or alcohol abuse. Cox proportional hazards regression model was fitted to the data in order to identify those factors independently associated with the cumulative probability of developing HCC. Variables significantly associated with the rate of HCC occurrence were age (HR 1.05, 95% CI 1.02-1.07 per year; P = 0.001) and smoking habits (HR 2.3, 95% CI 1.2-2.7; P = 0.01 for smokers versus nonsmokers).

Figure 5.

Cumulative probability of being HCC-free, as calculated by the Kaplan–Meier method, in cirrhotic patients with hemochromatosis. (A) Overall series, (B) non–HFE-related patients, and (C) HFE-related patients stratified according to the decade of enrollment (solid line, 1977-1986; circles, 1987-1996; squares, 1997-2007). No significant differences were observed between groups. Figures at the bottom refer to the number of patients still under observation at each time interval.

Discussion

This study, which was conducted in a prospectively collected cohort of consecutive patients with hemochromatosis diagnosed between 1976 and 2007 in three northern Italian reference centers, indicates that patients diagnosed in recent years have a disease characterized by a milder iron overload, a lower prevalence of cirrhosis, extrahepatic manifestations and acquired factors, and a similar observed and expected survival in the last decade, but that in cirrhotic patients the risk of developing HCC does not differ across the years. In addition, this study confirms and extends the results reported in a previous Italian multicenter study,5 showing that 25% of the patients were negative for HFE genotypes at risk for hereditary hemochromatosis. Interestingly, the amount of iron overload of non-HFE patients was comparable to that of HFE patients and was not influenced by the presence of cirrhosis and acquired risk factors.

The milder iron overload detected in our patients is consistent with the significant improvement in both diagnosis and management of hereditary hemochromatosis22–24 due to a better awareness of the disease by physicians and an easy access to HFE typing, leading to an earlier diagnosis and consequently a milder disease expression. Indeed, the amount of iron removed to reach depletion and prevalence of cirrhosis dramatically dropped in the last decades, accounting for a relevant improvement in survival, which reflects at least in part an ascertainment bias. Accordingly, prevalence of extrahepatic manifestations related to the severity of iron overload such as arthropathy, diabetes, cardiac involvement, and hypogonadism also decreased significantly, thus reducing the spectrum of the clinical manifestations of the disease at the time of diagnosis.15–17, 35 In line with our results is the recent report of a decline in diabetes prevalence associated with an improved survival of patients with hereditary hemochromatosis.36

Interestingly, patients diagnosed in recent years had a higher body mass index, as uniformly observed in the populations of western countries. Recently, Powell et al.37 reported a higher prevalence of obesity and/or hepatic steatosis in subjects with hemochromatosis and demonstrated that the coexistence of these metabolic alterations was associated with more advanced fibrosis. We do not have information on the presence of steatosis in the large majority of our patients; however, we were unable to find a similar correlation when we analyzed the relationship between body mass index and fibrosis, possibly because patients diagnosed during the last decade (for whom information was available) were those with a milder disease.

As shown in the previous multicenter Italian study,5 ≈25% of patients do not carry HFE genotypes at risk for hereditary hemochromatosis. The genetic background of these subjects remains unidentified,4 and the search for the recently reported non-HFE gene mutations was negative in almost all patients tested (data not shown) as reported in a previous study.38

The comparable iron overload as well as the prevalence of extrahepatic manifestations in patients with or without HFE genotypes at risk of disease reinforces the diagnosis of hereditary hemochromatosis also in HFE-negative patients who have repeatedly been described in Mediterranean countries.38–40

We observed a decreased prevalence of HBV and HCV infection and of alcohol intake, which parallels that observed in the Italian general population as a consequence of the improved behavior and the aggressive widespread campaign against alcohol abuse.41–43 The decreased prevalence of acquired factors was evident in both patients with and without the HFE genotypes at risk, although those that were negative had a slightly higher coexistence of exogenous risk factors compared with patients with HFE-related hemochromatosis.

The presence of acquired factors in patients with non–HFE-related iron overload is a possible limitation of this study since it can be suggested that these patients do not have hemochromatosis and that iron overload is secondary to the acquired factors. However, more than 60% of these patients had no detectable acquired risk factor, and the degree of iron overload was not influenced by the presence of coexistent hepatotoxic factors, suggesting a minor role of acquired risk factor in the natural history of the disease. A possible explanation is that in non–HFE-related hemochromatosis, undefined host factors (unknown genes) modify the clinical history of the disease.

However, it cannot be ruled out that, as reported in other non-hemochromatotic liver diseases, in subjects with still unknown genetic defects or with HFE genotypes usually not associated with severe iron overload (H63D+/+ or C282Y+/−), HCV and/or alcohol enhance the underlying defect of iron metabolism leading to phenotypically expressed hemochromatosis.44

As expected, the severity of iron overload at diagnosis decreased through the years more markedly in HFE-related than in non–HFE-related hemochromatosis, reflecting the availability of genetic markers for HFE-related hemochromatosis, leading to an earlier diagnosis. In line with these findings is the significant decrease of cirrhosis-related complications in HFE-related hemochromatosis and a similar survival of the patients compared with a matched population in the last decade. Interestingly, the prevalence of C282Y homozygosity did not vary across the decades and was consistent with the value reported in a previous large multicenter Italian study.5 On the contrary, the prevalence of compound heterozygotes—which usually have a mild iron overload with scant or absent phenotypic disease expression—significantly increased in the last decade, reflecting the improved diagnosis of hereditary hemochromatosis, which also accounts for the similar age of the patients at diagnosis in the three decades as well as for the increased proportion of females in the last decade. However, the influence of a possible referral bias cannot be ruled out given that the three centers participating in the study are reference centers for hereditary hemochromatosis. In addition, the improvement in survival due to healthier population as a whole might partly account for the similar age of patients diagnosed in recent years.

In the present cohort consisting of 452 patients with hemochromatosis, the median follow-up period was 9.3 years (range, 6 months to 32 years). The expected survival rates for an age- and sex-matched normal population differed significantly from the observed survival of patients with hemochromatosis. Survival was worse in patients with hemochromatosis than in those without cirrhosis (data not shown). As expected, survival gradually increased during this long-term follow-up due to the diagnosis of hemochromatosis at a nonsymptomatic stage with a consequent decrease in the amount and duration of iron overload as reported.15–18, 24, 35 Survival did not differ between non–HFE- and HFE-related patients either with or without cirrhosis, the last one with a survival comparable to that of normal subjects (data not shown). Interestingly, long-term survival did not improve in the recent years in the subset of patients with cirrhosis at enrollment. Thus, the finding of significant morbidity in hereditary hemochromatosis patients strengthens the need for an earlier diagnosis to prevent long-term complications, of which HCC is the most important.45 Indeed, although cirrhosis was significantly less frequent in patients diagnosed during the last years, the incidence of HCC and the probability of being HCC-free in cirrhotic patients did not vary across the three decades. In the present study, we were unable to define whether iron depletion in cirrhotics decreased the risk of HCC occurrence, because the large majority of our patients developed HCC after depletion and for ethical reasons we do not have a non–iron-depleted control group. In agreement with previous reports,46 smoking habits were significantly associated with HCC occurrence.

In conclusion, data from this long-term prospective study comprising a large series of Italian patients indicate that subjects with HFE- and non–HFE-related hemochromatosis have comparable iron overload and similar clinical history. Patients without cirrhosis at enrollment who are diagnosed during the last 10 years have less severe disease and lower prevalence of acquired risk factors, which influences survival more in HFE-related than in non–HFE-related patients.

Ancillary