The diversity of liver diseases among outpatient referrals for elevated serum ferritin

Department of Medicine, Division of Gastroenterology, University of Western Ontario, London, Ontario Correspondence: Dr Paul C Adams, Department of Medicine, University Hospital, 339 Windermere Road, London, Ontario N6A 5A5. Telephone 519-685-8500 ext 35375, fax 519-663-3649, e-mail padams@uwo.ca Received December 2, 2005. Accepted January 11, 2006 K Wong, PC Adams. The diversity of liver diseases among outpatient referrals for elevated serum ferritin. Can J Gastroenterol 2006;20(7):467-470.


METHODS
This was a retrospective study of outpatient referrals for ferritin elevation to the hemochromatosis and liver disease clinic at University Hospital in London, Ontario, between 1999 and 2005. All patients reviewed had ferritin concentrations greater than 300 µg/L (normal lab reference range 15 µg/L to 200 µg/L for women, and 30 µg/L to 300 µg/L for men). Patients were stratified into two groups based on their ferritin concentration at their first clinic visit: ferritin 300 µg/L to 1000 µg/L, and ferritin greater than 1000 µg/L. Those who were referred for reasons other than hyperferritinemia (eg, known hepatitis) with a coincident high ferritin concentration were excluded. The ferritin concentration, transferrin saturation, aspartate aminotransferase levels, alanine aminotransferase (ALT) levels, HFE mutations (C282Y and H63D), liver biopsy results, liver iron concentration values and final clinical diagnosis were collected. Individuals in the present study with potentially more than one factor contributing to a high ferritin concentration were placed into the 'uncertain diagnosis' group.
Liver biopsy reports were reviewed. The paraffin blocks of those patients with excess stainable iron were submitted for biochemical quantification of liver iron concentration. Liver iron concentration was also determined if requested by the ordering physician. Liver iron concentration was measured by atomic absorption spectrophotometry. Liver tissue was removed from paraffin-embedded blocks by washing in xylene and then drying to a constant weight before analysis (6).
Mean ferritin concentrations and liver iron concentrations were compared using Student's t test. A χ 2 test was used to compare proportions. Statistics were analyzed using Winstat 3.0 (Kalmia Co, USA).

RESULTS
A total of 482 charts were reviewed, of which 119 (25%) had serum ferritin concentrations greater than 1000 µg/L. There was a male predominance of approximately 3:1. The mean age among all groups was approximately 50 years, with a range of 49.5 to 55.3 years (Table 1). All patients with HFE-linked hemochromatosis (ie, C282Y or H63D mutations of the HFE gene) were Caucasian. The group of patients with iron overload from other causes included patients with Asian, East Indian, Caribbean and Arab ancestry.
In patients with a ferritin concentration greater than 1000 µg/L, 42% (50 of 119) had HFE-linked hemochromatosis, of which 86% (43 of 50) were C282Y homozygotes, 12% (six of 50) were C282Y heterozygotes and 2% (one of 50) were H63D homozygotes. Other diagnoses included chronic alcohol consumption, alcoholic liver disease, nonalcoholic steatohepatitis, other causes of iron overload and newly diagnosed chronic viral hepatitis ( Figure 1). In 9% (11 of 119) of the cases, the cause of the iron overload remained uncertain despite a thorough workup including liver biopsy. Of those with other causes for iron overload, diagnoses included multiple, red cell transfusions, intravenous iron therapy, aceruloplasminemia and juvenile hemochromatosis. There were six patients with iron overload, confirmed by liver biopsy, of unknown etiology. Three patients had hereditary hyperferritinemia cataract syndrome (7). This diagnosis was made based on a family history of bilateral, early-onset cataracts, elevated ferritin concentrations with no biochemical or tissue evidence of iron overload, and a mutation in the iron-responsive element of the L-ferritin gene. One per cent had miscellaneous diagnoses, including autoimmune hepatitis, hepatitis B and lymphoma (Figure 1).  Figure 1) Diagnoses in patients (n=119) with elevated ferritin (Fe) (greater than 1000 µg/L). EtOH Alcoholic liver disease; Hep C Hepatitis C; HH HFE-related hemochromatosis; Misc Miscellaneous; NAFLD Nonalcoholic fatty liver disease Overall, 66.3% (79 of 119) of these patients underwent a liver biopsy as part of their workup.
In the group with serum ferritin greater than 1000 µg/L, 71% (27 of 38) of those who had documented iron overload by biopsy (liver iron concentration greater than 35 µmol/g dry weight) had a transferrin saturation greater than 50% (normal range 20% to 55%). Of the C282Y homozygotes who had an elevated liver iron concentration, 88% (21 of 24) had an elevated transferrin saturation compared with 43% (six of 14) of the non-C282Y homozygotes. Liver iron concentration was measured only if there was evidence of significant stainable iron on liver biopsy, as determined by the pathologist or by request of the ordering physician. Forty-four of the patients with ferritin concentrations greater than 1000 µg/L had liver iron concentrations determined, of which 34 (77%) were elevated above 35 µmol/g. Among those with iron overload, the liver iron concentrations were higher in the C282Y homozygotes than in non-C282Y homozygotes, ie, all other patients including compound heterozygotes and H63D homozygotes, (mean 297 µmol/g, range 74.6 µmol/g to 632 µmol/g versus 134 µmol/g, range 36.1 µmol/g to 514 µmol/g, respectively; P=0.0028 [reference range 0 µmol/g to 35 µmol/g]). Thirty-two of 50 (64%) of the non-C282Y homozygotes and six of 25 (24%) of the C282Y homozygotes with a serum ferritin greater than 1000 µg/L did not have iron overload on liver biopsy.
In patients with ferritin 300 µg/L to 1000 µg/L, 33% (119 of 363) had HFE-related hemochromatosis, of which 59% (70 of 119) were C282Y homozygotes, 24% (28 of 119) were compound heterozygotes (C282Y/H63D) and 17% (20 of 119) were H63D homozygotes. Eighty-six per cent (313 of 363) of the patients in this group had genetic testing for HFE mutations. These patients were further stratified into four subgroups based on ferritin concentration (Table 1). HFE-linked hemochromatosis, and alcoholic and nonalcoholic steatohepatitis made up the majority of the diagnoses. Other causes of iron overload included porphyria cutanea tarda and ironloading anemias. Miscellaneous diagnoses included hepatitis B and C, autoimmune hepatitis, methotrexate-induced liver injury, granulomatous hepatitis and primary biliary cirrhosis ( Figure 2). The number of patients who underwent biopsies in this group was significantly less at 18.5% (67 of 363). Among the patients tested for HFE mutations with a ferritin concentration between 300 µg/L to 1000 µg/L, the mean (± SD) ferritin concentrations in C282Y homozygotes and in non-C282Y homozygotes were not statistically different (633±184 µg/L and 575±223 µg/L, respectively; P=0.057). There was no statistical difference in the percentage of patients diagnosed with hemochromatosis in the ferritin 300 µg/L to 1000 µg/L group versus the ferritin greater than 1000 µg/L group (P=0.067).

DISCUSSION
Although the diagnosis of HFE-linked hemochromatosis was a common diagnosis in outpatient referrals for elevated serum ferritin, more than 50% of these referred patients had an alternate explanation for their elevated ferritin. The clinic at the University of Western Ontario is known for the study of iron overload and this likely results in a referral bias, which may increase the percentage of patients with HFE-linked hemochromatosis. HFE-linked hemochromatosis, alcoholic liver disease and nonalcoholic fatty liver disease made up the majority of the diagnoses among all subgroups of patients (60% to 80%). The percentage of patients diagnosed with HFElinked hemochromatosis was similar, regardless of ferritin concentration (P=0.067). The only previous study that has examined diagnoses of high ferritin concentrations focused on inpatients with extreme elevations (greater than 1000 µg/L) of serum ferritin (8). In that study (8), the diagnoses, in order of decreasing frequency, included liver disease, renal disease, malignancy, HIV infection, non-HIV infection, sickle cell disease and red blood cell transfusions. Those with liver disease had a mean (± SD) serum ferritin concentration of 2753.1±656.8 µg/L. That study differs from the current study in its selection of patients, and this may explain the predominance of infectious and inflammatory etiologies in those patients.
A serum ferritin concentration between 300 µg/L and 1000 µg/L is common. The Hemochromatosis and Iron Overload Screening (HEIRS) (9) study (a recent, large, multicentre screening study) demonstrated that 5.9% of 43,453 Caucasian subjects had elevated ferritin in a primary care population, and 19% of Asian subjects had elevated ferritin. One of the limitations of the present study is the small percentage of liver biopsies performed on this group of patients. It will likely be difficult to circumvent this issue, given the invasive nature of liver biopsies and the risks outweighing the benefits in these patients. Previous studies have shown that patients with hemochromatosis and a ferritin concentration less than 1000 µg/L are unlikely to have significant fibrosis, and therefore, a liver biopsy is rarely recommended (10,11). Other noninvasive testing, such as magnetic resonance imaging studies, may help, although sensitivity is lower at low iron concentrations (12). In the present study, some patients had abdominal imaging demonstrating fatty liver. Even when liver biopsies are performed, such as in the group with serum  ferritin concentrations greater than 1000 µg/L, 9% of patients still had an unknown diagnosis, which is comparable with the other subgroups. A biopsy is particularly useful in non-C282Y homozygotes with a ferritin concentration greater than 1000 µg/L; in many of these, iron overload was not found (64%). Another diagnostic challenge is the lack of specificity of an elevated ferritin concentration as demonstrated by the variety of diagnoses. Genetic testing can be helpful in the diagnosis of HFE-linked hemochromatosis. Genetic testing for new iron genes (ferroportin, hemojuvelin, hepcidin and transferrin receptor 2) may never become commercially available because of the rarity of these conditions (13). Because obesity, diabetes mellitus and daily alcohol consumption are relatively common, it is difficult to determine the contribution of one or more of these conditions to the serum ferritin concentration. Even in patients who have an HFE mutation, it may be difficult to tease out the expressing phenotype with iron overload from the nonexpressing phenotype with coexistent fatty liver disease. Transferrin saturation is frequently used to screen for hemochromatosis, with the assumption that it reflects iron overload. In the HEIRS study (9), transferrin saturation was found to be greater than 50% in 84% of men and greater than 45% in 73% of women who were C282Y homozygotes. This is similar to our finding that 88% of C282Y homozygotes with biopsy-proven iron overload had an elevated transferrin saturation. In contrast, two other screening studies have found a low sensitivity (50% to 60%) of a transferrin saturation of 50% or more (14,15). These latter studies can be explained by the high prevalence of nonexpressing C282Y homozygotes, and elevated transferrin saturation remains a useful test in the detection of an iron-loaded C282Y homozygote. In the present study, only 43% of non-C282Y homozygotes with biopsyproven iron overload had a transferrin saturation of greater than 50%. Therefore, a normal transferrin saturation result in a non-C282Y homozygote with an elevated ferritin level does not exclude iron overload. The mean transaminase concentrations were significantly higher in non-C282Y homozygotes than in C282Y homozygotes. None of the homozygotes had a transaminase elevation more than three times the upper limit of normal, regardless of ferritin concentration. A previous study (16) from our centre demonstrated that 69% of typical patients had an elevated ALT concentration with the mean (± SD) ALT concentrations being 43.8±28.5 U/L in noncirrhotic patients and 84.4±99.3 U/L in cirrhotic patients.
The overall prevalences of C282Y heterozygotes and H63D heterozygotes was 10.3% and 18.8%, respectively. These values are consistent with that observed in large screening studies of the general population (9), so they are not over-represented in this referred population. These mutations are usually associated with normal iron studies, and therefore, we would not expect an increase in prevalence of heterozygotes in patients with hyperferritinemia.
Because the majority of patients referred for hyperferritinemia do not have HFE-linked hemochromatosis, it is prudent to search for an alternate diagnosis in such individuals. Obesity, alcohol and viral hepatitis (B and C) are the most common diagnoses. Transferrin saturation testing may be helpful, because a high percentage of C282Y homozygotes with iron overload have a transferrin saturation of greater than 50%. Transferrin saturation, however, may be less helpful in predicting iron overload in non-C282Y homozygotes. Mild ferritin elevations (300 µg/L to 1000 µg/L) may be more difficult to diagnose definitively in the absence of useful, noninvasive testing. Elevated liver enzymes appear to decrease the probability of a diagnosis of HFE-linked hemochromatosis. The clinical management of patients with mild hyperferritinemia has not been well studied. Observation over time or a trial as a voluntary blood donor are both useful approaches in this group.