Review article: targeted screening for hereditary haemochromatosis in high-risk groups

Transcription

1 Aliment Pharmacol Ther 2004; 20: doi: /j x Review article: targeted screening for hereditary haemochromatosis in high-risk groups S. DUBOIS* & K. V. KOWDLEY *Senior Fellow, Division of Gastroenterology, University of Washington, WA; Professor of Medicine, Director, Iron Overload Clinic, Division of Gastroenterology/Hepatology, University of Washington School of Medicine, Seattle, WA, USA Accepted for publication 2 May 2004 SUMMARY Patients with hereditary haemochromatosis are at risk for significant morbidity from iron overload as well as reduced life-expectancy once cirrhosis is established. Although inexpensive, sensitive screening tests and effective therapy are available, there is continued debate regarding the utility of screening for this condition because of recent data suggesting that the homozygous haemochromatosis mutation (C282Y) is associated with low penetrance and mild expressivity when identified in population screening studies. In this review, we examine the published data related to general population screening for haemochromatosis, as well as the evidence for screening selected high-risk populations. We also suggest possible screening strategies based on the available evidence. INTRODUCTION Hereditary haemochromatosis, also known as human leucocyte antigen (HLA)-linked or HFE-associated haemochromatosis, is an inherited disorder of iron metabolism leading to excessive absorption of iron from the gastrointestinal tract. Consequently, there is progressive iron loading in the parenchyma of many organs, leading to diabetes, arthropathy, cardiac disease, liver failure, and premature mortality among patients with cirrhosis. 1, 2 Screening studies have shown haemochromatosis to be one of the most common inherited disorders, with an estimated prevalence of approximately 1:200 1:400, 3 11 depending on whether genotypic or phenotypic criteria are employed. The haemochromatosis candidate gene was identified in 1996; it is a novel major histocompatibility complex class I-like gene, designated HFE. 12 Two common mutations have been described: a mutation resulting Correspondence to: Dr K. V. Kowdley, University of Washington, Box , 1959 NE Pacific Street, Seattle, WA 98195, USA. kkowdley@u.washington.edu in a cysteine to tyrosine substitution at amino acid 282 (C282Y), and a histidine to aspartate substitution at amino acid 63 (H63D). Approximately 85 90% of those with typical clinical manifestations of hereditary haemochromatosis are homozygous for C282Y, although compound heterozygosity (C282Y/H63D) may also result in increased disease susceptibility. 2 This review primarily focuses on a description of the clinical features of haemochromatosis and an examination of the published literature on screening populations with clinical features associated with iron overload ( highrisk groups) when compared with general population screening for haemochromatosis. CLINICAL FEATURES The clinical manifestations of haemochromatosis are varied and are a result of the deposition of iron in multiple organs. The classical description of the clinical presentation of haemochromatosis is the triad of cirrhosis, skin bronzing, and diabetes. However, a number of other associated features have been noted, including weakness, arthralgia or frank arthritis, Ó 2004 Blackwell Publishing Ltd 1

2 2 S. DUBOIS & K. V. KOWDLEY hepatomegaly, abdominal pain, impotence, increased skin pigmentation, arrhythmia, and congestive heart failure. 13 Prior to the 1970s, patients with haemochromatosis were identified largely by presentation with severe end organ damage as a result of iron overload. More recent studies suggest that increasingly, more patients are being diagnosed early in the course of the disease with non-specific or no symptoms. Niederau et al. described 163 patients with hereditary haemochromatosis in 1985; 83% presented with weakness and lethargy, 55% with diabetes, 62% with abnormal liver enzymes, and 69% of patients were found to have cirrhosis. 14 By contrast, Bacon et al. reported a series of 40 patients in 1997; 83% had come to medical attention either because of family screening (18%) or by incidental serum blood chemistry testing (65%). The majority of patients (82%) were asymptomatic at the time of diagnosis. 15 As the liver is the primary site of iron storage in the body, it is usually the first organ affected in this disorder. 13 Ferritin and haemosiderin deposits are mainly found in hepatocytes, rather than in the reticuloendothelial cells such as Kupffer cells. 16 Liver disease may be associated with mild-to-moderate increase in serum aminotransferases. 13 In one retrospective study, mild abnormalities in AST and ALT were found in more than 65% of 100 patients with haemochromatosis at initial presentation. 17 Importantly, however, many patients may have normal liver enzymes even in the setting of cirrhosis. Hepatomegaly and associated epigastric or right upper quadrant abdominal pain may be present and may be related to the severity of iron deposition. As iron overload progresses, cirrhosis and subsequently hepatocellular carcinoma may develop. 13 Unfortunately, once cirrhosis is established, it has generally not been found to reverse with phlebotomy treatment. 14 Diabetes mellitus may be associated with haemochromatosis, and has been described in 10 30% of patients. 1 The exact nature and pathogenesis of the relationship between diabetes and iron overload is still not fully understood. One factor is impaired insulin secretion caused by accumulation of iron in the beta cells of the pancreas. 18 Another consideration in the pathogenesis of glucose intolerance is insulin resistance secondary to liver disease. 19 Inappropriately high levels of serum insulin after an oral glucose load can be found even in patients with non-cirrhotic haemochomatosis. This finding may be due to impaired insulin degradation secondary to hepatocyte iron deposition. 18 Carbohydrate metabolism has been shown to improve with phlebotomy in non-cirrhotic patients with impaired glucose tolerance. 19 Arthropathy may be observed in up to 30 40% of patients with hereditary haemochromatosis, and arthralgia without frank arthritis may be an early symptom. 1 In one series, arthropathy was by far the most common symptom reported prior to the diagnosis of haemochromatosis. 20 Arthropathy tends to involve the small joints of the hands, particularly the second and third metacarpal phalangeal (MCP) joints. It can extend to involve the wrists, elbows, shoulders, hips, and other joints, causing significant disability. Some patients may have episodes of acute joint pain and inflammation, likely secondary to calcium pyrophosphate dehydrate crystal deposition (pseudogout). Radiological changes include a loss of space at the MCP joints and/or intercarpal joints, subchondral cysts, squaring of the metacarpal heads, hook-like osteophytes, chondrocalcinosis and osteopenia. 21 Phlebotomy may not improve arthropathy, and may even aggravate joint symptoms, thus suggesting that perhaps factors in addition to iron deposition may be involved in this clinical manifestation. 22 Cardiac abnormalities in haemochromatosis may lead to both morbidity and mortality. Manifestations include cardiomyopathy presenting as congestive heart failure, and arrhythmias. In an analysis of multiple-cause mortality data, cardiac disorders were reported in 52% of persons who died with haemochromatosis. 23 Endomyocardial iron deposits are found predominantly in the ventricles, and less frequently in the atria or conduction tissue. 24 The most common arrhythmias observed are atrial tachycardias. Cardiomyopathy may be dilated or restrictive in nature. Dilated cardiomyopathy as a result of haemochromatosis is associated with increased ventricular dimensions and decreased systolic function. The restrictive type is characterized by normal ventricular dimensions with normal or increased wall thickness, and impaired ventricular filling. 25 Phlebotomy treatment, particularly in the non-dilated type, has been shown to 24, 25 improve ventricular dysfunction and symptoms. Endocrine abnormalities are observed in haemochromatosis, commonly manifesting as impotence and decreased libido in males. These changes have been described in patients with and without cirrhosis. Hypogonadotrophic hypogonadism is the most common of these symptoms, and has been attributed to the

3 REVIEW: HEREDITARY HAEMOCHROMATOSIS IN HIGH-RISK GROUP 3 selective accumulation of iron predominantly in the gonadotrophic cells of the pituitary gland, leading to impaired secretion of LH and FSH. 19 There have also been reports of amenorrhoea in females with haemochromatosis. 16 Abnormal hypothalamic-pituitary function in hereditary haemochromatosis may not improve with phlebotomy. Thyroid dysfunction is reported less commonly, but both hypothyroidism and hyperthyroidism have been described, and autopsy studies have revealed iron deposition in the thyroid gland. In one study of 49 hereditary haemochromatosis patients, 11.8% were found to have abnormal thyroid function, the majority being hypothyroid. The authors postulated that iron may have caused direct injury to the thyroid, followed by immune activation and subsequent hypothyroidism. 26 Skin pigmentation in haemochromatosis is part of the original descriptive triad. It has been found to reflect both melanin deposition (bronzed coloration) and iron deposition (slate-grey coloration) in the basal layers of the epidermis and the lining cells of sweat glands. 13 Because of the potential for these significant complications of untreated hereditary haemochromatosis, many have advocated screening for the disorder. Widespread population screening as well as screening in potentially high-risk groups (those with diabetes, abnormal liver tests, arthritis) have been proposed. The section below outlines the arguments and controversies surrounding population screening; the following sections will focus on presenting the evidence for screening in selected groups. POPULATION SCREENING In the last decade, much attention has been devoted to the issues surrounding population screening for hereditary haemochromatosis. The primary objective of screening is to identify and treat individuals with hereditary haemochromatosis prior to development of end organ complications. Screening has been recommended because haemochromatosis is a common inherited disorder with associated morbidity and mortality. There are sensitive screening tests, and effective treatment for the disorder is safe and readily available in the form of phlebotomy. Furthermore, it has been shown that individuals identified in the precirrhotic stage and treated by phlebotomy have survival rates equivalent to the average population. 14, 27 Therefore, haemochromatosis appears to satisfy most United States Preventive Services Task Force criteria and World Health Organization criteria for a condition in which population screening ought to be implemented, 28, 29 including relatively high prevalence and significant morbidity of the disorder, availability of accurate screening tests, a recognized early stage of the condition, and treatment proven to be safe and effective. Some have argued that there are insufficient data to justify screening because of the lack of natural history studies demonstrating that patients with increased ferritin levels but without end organ damage will develop these complications over time. It is our opinion that given the potentially irreversible sequelae of some end organ manifestations such as cirrhosis, and the safety of phlebotomy treatment, such observational studies of haemochromatosis patients with elevated serum ferritin levels are unlikely to be considered ethical. A number of studies have found screening for hereditary haemochromatosis to be a cost-effective strategy. Even in models assuming a probability of C282Y homozygotes developing symptomatic disease at a rate as low as 20%, the cost per life-year saved has been estimated at significantly less than $ This figure is substantially lower than other common medical prevention practices such as mammography for breast cancer detection ($22 000/life-year saved), 31 and screening for colorectal cancer ($ / life-year saved) 32 or cervical cancer ($ /life-year saved). 33 Furthermore, HFE gene testing has been shown to be even more cost-effective in screening first degree relatives of an affected proband because of the high prevalence and high penetrance. 34 There are several concerns regarding the implementation of population screening for haemochromatosis. The first issue, which must be addressed, is case definition. Should haemochromatosis be defined based on genotyping or on the basis of phenotype? Both approaches have inherent advantages and disadvantages. Adhering to the phenotypic definition may result in need for repeat testing as serum iron values may increase over time. Furthermore, phenotypic testing does not identify all individuals who may be at risk for complications of iron overload based on genotype, but rather only subjects that are expressing biochemical abnormalities. On the contrary, the advantages of phenotypic screening are that testing is readily available, fairly inexpensive, and has been employed in numerous population-based screening studies with reasonable sensitivity and specificity. Additionally, phenotypic-

4 4 S. DUBOIS & K. V. KOWDLEY based testing would avoid the pitfalls associated with identifying non-expressing homozygotes, such as stress, anxiety and possible discrimination in employment and insurance. Nevertheless, there are several unresolved issues that limit use of phenotypic-based screening. The optimal screening test [serum transferrin-iron saturation (TS), ferritin, or unbound iron-binding capacity (UIBC)] and the appropriate threshold values for hereditary haemochromatosis remain controversial. The most widely used screening test is serum TS with recommended thresholds ranging from 45 to 70%. Serum TS has been reported to be 94% sensitive and equally specific in detecting homozygosity for C282Y mutation. 6, 35 However, other studies report sensitivities of 52 75% at a threshold value of >45%. 3, 4, 7 Serum ferritin is often used in conjunction with TS for initial screening, but has been unreliable as sole marker. The serum ferritin can be beneficial in classifying patients with regard to the presence or absence of cirrhosis, and in monitoring the effectiveness of phlebotomy therapy. The serum unbound IBC (UIBC), a single-step, automated colorimetric test, has been shown in some studies to be as reliable as TS. 4, 36 The UIBC assay is more economical than TS at an estimated cost of only $1, has a favourable cost-effectiveness profile estimated at $5500 per homozygote detected. However, optimal thresholds for UIBC are yet to be clarified. Several population screening studies using serum TS, UIBC, and ferritin as screening tests and liver biopsy or quantitative phlebotomy as confirmatory tests have demonstrated a high prevalence of haemochromatosis, particularly amongst Caucasians (Table 1). In a large study in the United States of over asymptomatic subjects using TS and ferritin, 138 patients had a serum ferritin >200 and TS > 45%; the estimated prevalence of hereditary haemochromatosis in this group based on liver biopsy was 0.45%. 37 Niederau et al. screened 6039 healthy German subjects with TS; 42 subjects had repeatedly elevated TS and ferritin, 28 of whom had iron overload consistent with haemochromatosis on liver biopsy or by quantitative phlebotomy. 9 Six HLA identical siblings also were found to have haemochromatosis by this method, for a total prevalence of approximately 0.56%. Similarly, an Australian study of over 1000 healthy subjects with TS > 45% found a prevalence of haemochromatosis of 1:300 based on liver biopsy. 11 In the United States, Baer et al. screened approximately 4000 asymptomatic men in an HMO setting in California using serum TS and ferritin. 10 Fourteen subjects with a persistently elevated TS 62% and elevated ferritin were referred for liver biopsy; eight of 12 subjects who underwent biopsy had the diagnosis of haemochromatosis confirmed by hepatic iron concentration, hepatic iron index and/or quantitative phlebotomy, a prevalence of one in 497 in the population screened. Notably, only 50% of subjects in this study were Caucasian. Similarly, a prevalence of 1:600 was found in a screening study of 1968 healthy employees in North Carolina using TS. 38 However, only three of 16 subjects with elevated TS and ferritin underwent liver biopsy in this study; therefore, the prevalence may have been underestimated. The other proposed method for population screening is HFE genotyping. Genotypic screening is highly specific, although more costly than serum TS. Importantly, genotypic screening has the potential disadvantage of detecting asymptomatic persons with normal iron parameters, many of whom may never develop clinical disease. Detecting asymptomatic patients via genetic testing can potentially lead to discrimination affecting an individual s health insurance or employment, or cause considerable anxiety and social stress. 39 Another difficulty with genotyping is that testing may not Author N Population Screening test Prevalence of HHC Stave USA TS > 50% women, 1:600 >60% men, SF Phatak USA TS > 45%, SF 1:225 Niederau Germany TS > 50% women, 1:178 >60% men, SF Baer USA TS 62%, SF 1:497 Wiggers Denmark TS 50 70%, SF 1:430 Leggett Australia TS 45%, SF 1:275 Edwards USA TS > 62%, SF 1:225 Table 1. Population screening studies using phenotypic tests SF, serum ferritin; TS, transferrin saturation; HHC, hereditary haemochromatosis.

5 REVIEW: HEREDITARY HAEMOCHROMATOSIS IN HIGH-RISK GROUP 5 identify up to 20% of Caucasians and most non- Caucasians with non-hfe-associated iron overload. 2 Other primary iron overload disorders include mutations in ferroportin, transferrin receptor 2, and hepcidin, in which no readily available testing yet exists. 40 A potential secondary benefit of identifying homozygotes through genetic testing is in the subsequent testing of their relatives, many of whom may have iron overload and may benefit from treatment. 34 HFE genotyping also has the advantage of being a simple, one-time testing strategy. Possibly the most important factor in influencing the use of HFE gene testing for population screening is the clinical penetrance of the C282Y homozygous genotype. The published cross-sectional genotype phenotype correlation studies that have examined the penetrance of this genotype have found varying results, partly because of lack of uniform criteria for phenotypic expression (Table 2). Many studies have shown that a significant proportion of C282Y homozygotes have serum biochemical evidence of iron overload. Burt et al. screened 1064 healthy subjects in New Zealand, revealing five C282Y homozygotes; all five had TS 55%. 7 Three of the five had elevated serum ferritin levels and underwent liver biopsy demonstrating hepatic iron index >1.9. In the USA study, 5 six of 1450 participants were found to be C282Y homozygous (0.4%); four of six had elevated TS ( 50% in women, 60% in men). Olynyk et al. found TS 45% to be 94% sensitive in detecting C282Y homozygotes in a study of 3011 adults in Brusselton, Australia. 6 Sixteen C282Y homozygotes (0.5%) were detected; all had either elevated serum TS (n ¼ 15) or elevated serum ferritin level. Phatak et al. 30 found TS 45% in 11 of 12 C282Y homozygotes unidentified by screening; 100% of men and 50% of women homozygous for C282Y had ferritin levels >200 lg/l. In contrast, only 7% with genotypes C282Y/H63D or H63D/H63D had a serum ferritin >300 lg/l, suggesting a lower risk of iron overload in subjects with these genotypes compared with C282Y homozygotes. 41 Adams et al. screened 5211 volunteer blood donors in London, Ontario with UIBC, TS, and HFE genotyping; 16 C282Y homozygotes were identified, a prevalence of one in Only three of the homozygotes had elevated serum ferritin level; the sensitivity of TS > 45% for detecting C282Y homozygosity was 75%. Beutler et al. 3 screened subjects in an HMO setting in San Diego, California. The authors found 43 C282Y homozygotes; the sensitivity of a TS 50% for detecting homozygosity was only 52%. In a recent Norwegian study, participants were screened for haemochomatosis as part of a health survey programme. Serum TS was measured as the initial screening study; HFE genotyping was obtained if the elevated TS level was confirmed with repeat testing. The authors found that 297 of 559 subjects (53%) were homozygous for C282Y; 72% of females and 91% of males had elevated serum ferritin levels. Only a small percentage of subjects with elevated TS and ferritin were found to have cirrhosis; however, only about half of the patients in this group underwent liver biopsy. Recent studies have raised further questions regarding the penetrance of the C282Y homozygous C282Y mutation. Beutler et al. extended their previous study by screening subjects attending a health appraisal clinic for HFE mutations along with clinical and serum biochemical correlation. 43 They estimated that fewer than 1% of C282Y homozygotes develop overt clinical features of haemochromatosis, as only one of the 152 homozygotes studied fit the clinical criteria usually applied for the diagnosis of haemochromatosis. Symptoms and findings commonly associated with Table 2. Population screening studies using HFE genotyping Author N Population Screening test Prevalence C282Y/C282Y Homozygotes with increased ferritin (%) Beutler USA TS, SF 1 in Phatak USA TS, SF 1 in Asberg Norway TS, SF 1 in Beutler USA TS, SF 1 in Adams England blood donors UIBC, TS 1 in McDonnell USA TS, SF 1 in Olynyk Australia TS, SF 1 in Burt New Zealand TS, SF 1 in SF, serum ferritin; TS, transferrin saturation; UIBC, unbound iron-binding capacity.

6 6 S. DUBOIS & K. V. KOWDLEY haemochromatosis such as fatigue, arthropathy, abdominal pain, impotence, and diabetes were not significantly more prevalent among C282Y homozygotes or C282Y/H63D compound heterozygotes than among controls as assessed by a self-administered questionnaire. The only clinical abnormalities found in C282Y homozygotes compared with controls were abnormal liver enzymes or a history of liver problems. Among homozygotes, pre-treatment serum transferrin was >50% in 75% of men and 40% of women. Median values of plasma collagen IV concentration, a surrogate marker for hepatic fibrosis, was found to be higher in C282Y homozygotes than in controls; however, liver biopsies were not performed to assess for significant liver disease. A limitation of this study is that the authors excluded 28 C282Y homozygotes from their analysis who had been previously diagnosed with haemochromatosis. It is likely that many of these patients may have had clinical signs and symptoms of haemochromatosis, and excluding this data could have led to an underestimation of penetrance in this study. Another potential factor affecting the estimate of penetrance may have been a higher proportion of the screened population with Spanish ancestry, which could have led to reporting to different gene gene interactions than observed among C282Y homozygotes of predominantly Northern European descent. 44 Similarly, in a Danish cohort of 9174 individuals, 23 C282Y homozygotes were followed for 25 years; none developed clinically overt haemochromatosis despite the majority having biochemical evidence of iron overload. However, liver biopsy was not performed to exclude fibrosis or cirrhosis in any of the patients. 45 A more recent study evaluated biochemical and clinical features retrospectively of 10 C282Y homozygotes during a 17-year period prior to diagnosis. Median serum TS levels increased; four patients had persistently elevated ferritin, and three of these developed bridging fibrosis or cirrhosis on liver biopsy. 46 In summary, populations screening studies based on TS have found a prevalence of 1:300 1:500. However, studies utilizing the HFE mutation analysis for screening have found a higher prevalence of C282Y homozygotes, ranging from 1:150 to 1:300. Both genetic and environmental factors and interactions may modify disease expression and influence the penetrance of the haemochromatosis genotype. It is likely that differences in study design, geographic location, and heterogeneity of patient populations screened with regard to diet, alcohol use, and gene gene interactions have led to the widely variable estimates of penetrance. Based on these difficulties, it can be concluded that although HFE gene testing is an attractive method for population screening, phenotypic-based strategies are more likely to be accepted at the present time, with HFE genotyping reserved as a second line or confirmatory test. Certainly, there is currently no consensus on general population screening for haemochromatosis, and more information is required from future well-designed screening programmes. The HEIRS screening project of over primary care patients and similar investigations are in progress to provide further information regarding the prevalence and penetrance of HFE mutations, and hopefully will provide insight into the various genetic and environmental determinants of disease expression. 47 Additional questions worth considering in population screening for hereditary haemochromatosis are: at what age should screening begin? Should screening be done in the general population or in targeted high-risk populations? It is generally agreed that men should be screened earlier in life than women; some have suggested screening men after the age of 30, and women >40 years of age since significant end organ damage is unlikely prior to that time. 1 A potentially cost-effective strategy is to perform screening based on ethnicity. Individuals of Northern European descent may be targeted, as this population will likely have a higher prevalence of phenotypic disease. Another proposed strategy is to target screening to populations demonstrating damage in end organs affected by haemochromatosis. These include individuals with diabetes, liver disease, cardiac failure, arthritis, and endocrinopathies. The published literature on the prevalence of hereditary haemochromatosis in these groups is reviewed in the following section. SCREENING IN HIGH-RISK POPULATIONS Diabetics Diabetes mellitus has been reported to be present in 20 80% of patients with hereditary haemochromatosis. Therefore, screening for this disorder in diabetic patients may be a cost-effective strategy. However, there are conflicting results of the utility of screening this population from studies conducted prior to the availability of HFE genotyping; the prevalence of haemochromatosis defined by biochemical studies and liver biopsy

7 REVIEW: HEREDITARY HAEMOCHROMATOSIS IN HIGH-RISK GROUP 7 Table 3. Screening studies in patients with type 2 diabetes using iron studies Author N Population Screening test Turnbull England TS, SF 1:727 George UK TS 1:199 Czink Hungary TS 1:134 Obrien Ireland TS. SF 1:191 Phelps Australia TS, SF 1:250 HHC prevalence* * Based on liver biopsy. SF, serum ferritin; TS, transferrin saturation; HHC, hereditary haemochromatosis. among diabetics has been reported to be as low as 0.1% or as high as 9.5% The most convincing data in support of screening a diabetic population for haemochromatosis prior to the era of HFE gene testing were published by Phelps et al. 49 Their screening strategy using TS in 418 type 1 and type 2 diabetic patients in Australia revealed four biopsy-proven cases of hereditary haemochromatosis, a prevalence rate of 9.6%. All cases had a TS > 75% and an elevated ferritin of >300 lg/l. A study using TS in 1194 patients attending a diabetic clinic in the United Kingdom found six patients with biopsy-proven haemochromatosis, a prevalence of 0.5%. 48 Turnbull et al. screened 727 type 1 and type 2 diabetics in England; 22 patients had abnormal iron indices on repeat testing, but only one of 10 subjects that underwent liver biopsy had haemochromatosis based on biopsy criteria. 51 As only 0.13% of the population studied was confirmed to have haemochromatosis, the authors concluded that targeted screening among diabetics was not more effective than screening the general population. O Brien et al. identified three of 572 patients with haemochromatosis among diabetic patients screened with serum iron studies. 52 The positive predictive value of serum TS > 62% was 66%, and much lower for an elevated ferritin (17%). Based on these findings, the authors did not recommend routine screening among diabetics for haemochromatosis. An autopsy study of hepatic iron stores in 15 patients with non-insulin-dependent diabetes demonstrated no difference in the hepatic iron content or distribution compared with 17 age-matched controls, concluding that non-insulin-dependent diabetes was not associated with substantial iron overload. 53 Subsequent to the availability of HFE genotyping, several case control studies have examined the prevalence of HFE mutations among diabetic patients compared with sex- and age-matched non-diabetic subjects (Table 4). The majority of these investigations, to be reviewed below, have shown no difference in the frequency of C282Y homozygosity in diabetics compared with controls, with some important exceptions. In a study comparing 184 non-insulin-dependent diabetic patients with 87 blood-donor control subjects in France, 54 no differences were found in the frequency of C282Y or H63D mutations between the two groups; only one diabetic patient was found to be a C282Y homozygote. However, serum iron, TS, and ferritin were significantly increased in patients with HFE mutations compared with those who were wild type. A study in Table 4. Screening studies of HFE mutations in diabetics compared with controls Author N Population HFE mutations Sampson 56 Fernandez-Real 58 DuBois-Laforgue 54 Frayling 55 Kwan 60 Florkowski 59 Braun NS, 220 DM2 108 NS, 170 DM2 87 NS, 184 DM2 215 NS, 238 DM2 103 DM1, 105 DM NS, 230 DM2 355 NS, 401 DM2 England Spain France England Canada New Zealand Germany No difference No difference in C282Y; increased frequency of H63D allele in DM2 No difference No difference Increased C282Y frequency in DM2: (21.9% DM2, 11.7% DM1) No difference No difference NS, normal subjects; DM2, type 2 diabetes mellitus; DM1, type 1 diabetes mellitus.

8 8 S. DUBOIS & K. V. KOWDLEY England did not find a higher prevalence of C282Y homozygotes among diabetics (0.42%) compared with age-matched controls (0.46%). 55 In another English study, Sampson et al. compared HFE genotypes of 220 normal Caucasian male subjects to 220 Caucasian males with type 2 diabetes. 56 The prevalence of HFE mutations did not differ significantly between patients and controls; one C282Y homozygote (0.45%) and three C282Y/H63D compound heterozgotes (1.3%) were found in the diabetic group. Similar studies in Germany and Spain found no significant differences in the prevalence of the C282Y mutation between patients and controls. 57, 58 However, the Spanish study did find a significantly increased prevalence of the H63D allele among diabetics; serum ferritin levels were also significantly higher in type 2 diabetics who carried HFE mutations. Other studies, however, have found targeted screening for haemochromatosis in the diabetic population to be an effective strategy. Kwan et al. studied 105 patients with type 2 diabetes compared with 105 patients with type 1 diabetes; all were of European ancestry. 60 They found an increased prevalence of the C282Y mutation in type 2 diabetics compared with type 1 patients (21.9% and 11.7%, respectively); one type 2 diabetic patient was identified as a C282Y homozygote. A recent prospective study in Finland conducted a 4 year follow-up of 555 healthy men and found that 11% of subjects carrying the C282Y allele developed diabetes compared with 5% of non-carriers (P ¼ 0.047). 61 Moczulski et al. 62 examined 563 Caucasian patients with type 2 diabetes and 196 healthy controls; they found a significantly increased frequency of the C282Y mutation in diabetics compared with controls (2% vs. 7%; P ¼ ). Mutation analysis revealed one C282Y homozygote and nine C282Y/H63D compound heterozygotes among the diabetic group, but neither of these genotypes were observed among controls. Interestingly, an increased frequency of the H63D mutation was observed in diabetics with microalbuminuria, suggesting a possible role of the mutation in the susceptibility to diabetic nephropathy. A recent, large retrospective study 63 conducted in Copenhagen specifically studied individuals with lateonset type 1 diabetes (after the age of 30 years). The HFE genotypes of 716 such patients were compared with 9174 controls; the proportion of C282Y homozygotes among patients with late-onset type 1 diabetes (1.26%) was significantly higher than in the general population (0.25%). All nine homozygotes in the diabetic group had serum TS > 50%; thus serum TS had excellent sensitivity and specificity (100% and 96%, respectively) for predicting C282Y homozygosity. The age range among these homozygotes was years, and all male subjects (five) had serum 14 levels >2000 lg/ml. Six of the seven patients who underwent liver biopsy were found to have cirrhosis, including all male subjects. The authors concluded that all individuals with late-onset type 1 diabetes mellitus should be screened for hereditary haemochromatosis based on these findings. Arthritis Arthropathy is a clinical feature of haemochromatosis in 13 43% of patients. 14, 15 A small number of studies have examined the yield of screening for hereditary haemochromatosis among patients with arthritis. Li et al. screened 92 adult patients with a diagnosis of rheumatoid arthritis and 87 unrelated controls for HFE mutations. 64 There was no difference in the prevalence of the HFE mutations between patients with rheumatoid arthritis and control subjects; no C282Y homozygotes were identified. However, the frequency of the H63D mutation was significantly higher in patients with RA compared with the control group (40% vs.20%, P < 0.006). In a recent study, in the United Kingdom, patients with inflammatory arthritis and 1000 controls were screened for HFE mutations. Five of the arthritis patients were found to be C282Y homozygous. The frequency of C282Y homozygosity in this population (0.005) was thus not significantly greater than the expected rate in the general population; therefore, the authors concluded that screening patients with inflammatory arthritis for haemochromatosis was not efficacious. By contrast, Olynyk et al. proposed that screening arthritis patients may indeed be an effective approach among patients attending a rheumatology clinic in Australia. 66 A total of 339 arthritis patients were screened over a 12-month period with serum TS and ferritin levels. Twelve of the patients had persistently elevated values and eight underwent liver biopsy. Five of the 339 patients screened (1.48%) were confirmed as having haemochromatosis by biopsy criteria (hepatic iron index >1.9). All patients with haemochromatosis had serum TS > 55%. However, only two of the five patients with haemochromatosis were noted to have the classic findings of chondrocalcinosis and arthropathy of the MCP joints. The authors recommended screening for haemochromatosis using serum TS in all patients with peripheral arthropathy.

9 REVIEW: HEREDITARY HAEMOCHROMATOSIS IN HIGH-RISK GROUP 9 Heart disease Cardiac involvement is another significant cause of morbidity and mortality in hereditary haemochromatosis. However, there is limited published literature on screening for haemochromatosis among patients with heart disease. Mahon et al. investigated the frequencies of HFE mutations in 207 Caucasian patients with dilated cardiomyopathy at a referral cardiac centre in the United Kingdom compared with 200 healthy Caucasian controls. 67 The authors found the C282Y mutation was not more frequent in patients with dilated cardiomyopathy compared with controls (15% and 12%, respectively; OR ¼ 1.2, 95% confidence interval: ). However, the H63D allele was significantly more common in patients with dilated cardiomyopathy. Additionally, there was a significantly increased frequency of C282Y in the subset of male patients with dilated cardiomyopathy. The authors noted a progressive increase in the serum iron and TS among patients in the following order: patients wild type for both mutations, H63D heterozygotes, C282Y heterozygotes, H63D homozygotes, C282Y/H63D compound heterozygotes, followed by C282Y homozygotes. However, no associations were found between serum TS or ferritin and left ventricular dimensions or function, peak oxygen consumption, or functional class of heart failure. In another investigation, Rosenqvist and Hultcrantz 68 examined the prevalence of haemochromatosis based on phenotype in 232 male patients with bradyarrhythmias (second- or third-degree AV block) requiring a pacemaker. Serum ferritin and TS were measured; liver biopsy was performed if ferritin was elevated to more than twice the upper limit of normal on two determinations. Hepatic iron overload was present in three of four patients who underwent liver biopsy, representing a prevalence of iron overload of 1.3% in this cohort. All three had ferritin levels >500 lg/l; two patients had other clinical signs of haemochromatosis including congestive heart failure and cirrhosis. Based on their results, the authors recommend selective screening for haemochromatosis in male patients with symptomatic high-grade AV block. Liver disease The prevalence of haemochromatosis has been reported to be as high as 3% in patients with asymptomatic elevations in serum aminotransferases. 69 Screening for haemochromatosis using serum TS and ferritin levels has therefore been recommended in patients with 17, 69 unexplained abnormalities in serum liver tests. There are only a limited number of published studies specifically addressing the yield of screening for hereditary haemochromatosis among patients with elevated serum liver enzymes. Bhavnani et al. 70 in the United Kingdom studied patients with elevated serum ALT found on routine blood testing. Among patients screened, 56 were found to have increased serum TS > 60%; nine were C282Y homozygotes and three were C282Y/H63D compound heterozygotes. These investigators conclude that TS screening followed by HFE genotyping in patients with elevated ALT is costeffective. A second study, in the United Kingdom examined the HFE genotypes of 427 patients being investigated in a liver clinic for abnormal liver tests. 71 The population studied was ethnically diverse, mostly male, and carried various diagnoses including alcoholic liver disease, chronic hepatitis C, primary biliary cirrhosis, and chronic hepatitis B. Eleven patients were found to be C282Y homozygous; the C282Y mutation was found almost exclusively in patients of northern European descent (98%). TS > 60% had a sensitivity and specificity of 91% and 93%, respectively, for C282Y homozygosity, while serum ferritin >200 lg/l had a sensitivity and specificity of 73% and 70%, respectively. The authors concluded that routine testing for iron overload using serum TS and ferritin in all patients with liver disease is justified, and advocated initial testing with TS followed by HFE genotyping if TS is elevated. In a study by Bacon et al., 72 HFE genotyping was performed on 132 patients with liver disease of various aetiologies. The authors found that 5% of these individuals were C282Y homozygotes and 6% were C282Y/H63D compound heterozygotes. Serum TS was >45% in 98% of C282Y homozygotes. Although hepatic iron concentration was elevated in all C282Y homozygotes, 15% did not meet the proposed biopsy criteria of a hepatic iron index 1.9. The authors concluded that HFE genotyping in patients with liver disease and suspected iron overload is clinically useful to identify otherwise unsuspected C282Y homozygotes. This study may have been limited by referral bias, overrepresenting the proportion with hereditary haemochromatosis.

10 10 S. DUBOIS & K. V. KOWDLEY Additionally, the role of HFE mutations and their contribution to the severity of liver disease has been studied in viral hepatitis, porphyria cutanea tarda (PCT), non-alcoholic steatohepatitis (NASH), and alcoholic liver disease. However, many of these studies are limited by referral or selection bias, and have not controlled for other factors such as alcohol use, male gender, or duration of disease. The results in studies involving chronic viral hepatitis have been variable. Some studies have reported no association between HFE mutations and iron stores in hepatitis C patients. A French study of 209 patients with chronic hepatitis C found no association between hepatic iron accumulation and the presence of HFE mutations. 73 Similarly, a study of 164 Scottish patients with chronic hepatitis C also found no association between HFE mutations and serum biochemical or hepatic iron overload, as well as no association with fibrosis stage. 74 It is, however, noteworthy that three of 164 subjects were found to be C282Y homozygotes (1.8%). Kazemi-Shirazi et al. also assessed HFE mutations in 184 patients with chronic hepatitis C, as well as liver biopsy specimens the majority of subjects. The authors found that the degree of hepatic fibrosis did not differ between patients with or without HFE gene mutations, and allele frequencies for the C282Y and H63D mutations did not significantly differ in patients with chronic hepatitis C compared with healthy controls. 75 Another study of liver histology and HFE gene mutations in patients with chronic hepatitis C found no difference in H63D and C282Y mutation frequency between siderotic and non-siderotic patients. 76 Other studies, however, have shown an association between HFE mutations and iron overload and/or hepatic fibrosis in patients with chronic hepatitis C. Smith et al. reported no difference between the prevalence of C282Y mutation between patients with chronic hepatitis C and controls. 77 However, the presence of C282Y mutation in patients with chronic hepatitis C was significantly correlated with liver fibrosis and the degree of hepatocyte iron staining. Similarly, Piperno et al. studied 110 patients with chronic viral hepatitis (B and C) and 139 controls; they found no difference in the prevalence of HFE mutations between the two groups. 78 However, male patients heterozygous for the C282Y mutation had increased hepatic iron stores, and a significant correlation was found between the H63D mutation and hepatic iron overload. A previously mentioned European study found no significant differences in the allele frequencies of HFE mutations in hepatitis C patients compared with controls, but identified five C282Y homozygotes (2.7%). 79 Although the presence of HFE mutations was associated with elevated serum iron markers, most patients with elevated serum TS or ferritin did not have HFE mutations. Finally, in the USA study of 316 patients with chronic hepatitis C, the presence of HFE mutations was found to be independently associated with iron loading and advanced fibrosis, especially after controlling for disease duration. 80 In sum, the current evidence suggests that HFE mutations may contribute to iron overload in patients with chronic hepatitis C, although potentially other mechanisms of iron accumulation are yet to be defined. Hepatitis C has also been associated with PCT; several studies have examined the potential link between PCT and HFE mutations. Bonkovsky et al. found that up to 73% of PCT patients carried an HFE mutation; 15% were C282Y homozygotes. 81 The authors recommend that all patients with PCT be tested for hepatitis C as well as for HFE mutations. The prevalence of HFE mutations has also been studied in NASH. George et al., found the prevalence of the C282Y mutation was significantly greater in NASH subjects than controls; 7.9% of NASH patients were C282Y homozygotes. 82 There was a significant association between the presence of C282Y mutation and advanced hepatic fibrosis. Similar results were demonstrated in a study of 57 NASH patients in the United States; 2.8% were C282Y homozygotes and 16.7% were C282Y heterozygotes, compared with 0% and 11.2%, respectively, of 348 historical controls. 83 The prevalence of heterozygosity for either the C282Y or H63D mutation (61%) was significantly higher in NASH subjects compared with historical controls. Furthermore, levels of serum iron, TS, ferritin and degree of iron staining were significantly higher in NASH subjects with HFE mutations, although there were no significant differences in hepatic iron concentrations or hepatic iron indices in NASH patients with and without HFE mutations. Patients with C282Y mutations also had significantly higher hepatic fibrosis stage on liver biopsy compared to those without (P < 0.05). However, Chitturi et al. found that hepatic iron was not linked to hepatic fibrogenesis in their multivariate analysis of 93 patients with NASH. No C282Y homozygotes were found in their cohort, and there were no differences in histological grades of steatosis, inflammation or fibrosis

11 REVIEW: HEREDITARY HAEMOCHROMATOSIS IN HIGH-RISK GROUP 11 between subjects with or without C282Y. 84 In a study of 263 patients with NASH, the prevalence of C282Y and H63D mutations was similar to the general population. The study found that insulin resistance was the major factor for advanced fibrosis in these patients, and iron burden and HFE mutations did not contribute significantly to hepatic fibrosis. 85 The role of HFE gene mutations in alcoholic liver disease has not been well-studied. Grove et al. performed HFE genotyping in 257 patients with alcoholic liver disease and found no difference in the prevalence of HFE mutations compared with controls; furthermore, the presence of HFE mutations had no effect on hepatic iron stores in alcoholic liver disease. 86 Another recent study confirmed that the prevalence of HFE mutations is not increased among patients with alcoholic cirrhosis compared with controls. 87 However, limitations of both these studies is that they included predominantly cirrhotic patients, and that they did not assess hepatic iron concentrations. Family members Since hereditary haemochromatosis is characterized by an autosomal recessive pattern of inheritance, family members of affected individuals have a greater risk of having the disorder compared with the general population, and siblings of an affected proband are at greatest risk of being homozygotes (25%). Therefore, screening relatives (particularly first-degree relatives) of patients with hereditary haemochromatosis using HFE genotype testing, is recommended. 88 Assuming 90% penetrance of the C282Y homozygous genotype in males, and 50% penetrance in females, Adams et al. have estimated the risk of iron overload to be 24% in male siblings of an affected proband and 13% in female siblings. 89 Prior to the availability of HFE genotyping, a retrospective study of HLA typing and iron studies in siblings of probands classified 37 of 105 siblings as putative homozygotes, 25 of whom had elevated ferritin levels. 90 Powell et al. prospectively studied 209 subjects in 40 families with confirmed haemochromatosis. 91 Forty-eight of 51 subjects predicted to be homozygous showed increased hepatic iron stores as assessed by liver biopsy and quantitative phlebotomy. Bulaj et al. identified 214 homozygous relatives of 291 putative, clinically affected probands with haemochromatosis using HLA typing and HFE genotyping. 92 The authors found that 96% of men and 69% of women had iron overload as diagnosed by excess hepatic iron or elevated ferritin levels. Furthermore, disease-related conditions (liver disease, arthropathy, diabetes, hypogonadism) were found in 52% of the men over 40 years of age. The most frequent abnormalities were hepatic fibrosis and cirrhosis, which were related to the degree of elevation of serum ferritin levels. Children of homozygotes are thought to have a lower risk of being affected since homozygous heterozygous mating is required; the estimated prevalence is approximately one in 20. In a screening study, in children of putative homozygotes, 11 of 255 children were identified as homozygotes based on a TS > 55% and presence of excess hepatic iron, or by being an HLA-identical sibling of a child identified with these features. 93 Symptoms of joint pain and stiffness were present in 27% of homozygous children; one child had cirrhosis on liver biopsy. In a cost-effectiveness analysis, HFE gene testing for the C282Y mutation was found to be cost-effective in screening relatives of patients with hereditary haemochromatosis, with an incremental cost-effectiveness ratio of $3665/life-year saved. 34 CONCLUSION Hereditary haemochromatosis is a common genetic disorder that can lead to significant morbidity and mortality via deposition of iron throughout the body. There are adequate data to support the assertion that the clinical complications of iron overload in this condition may be avoided through screening and early treatment. The practice guidelines of the American Association for the Study of Liver Diseases recommend screening for haemochromatosis in the following target populations: patients with unexplained manifestations of liver disease or known liver disease with elevated serum iron markers; type 2 diabetics; first degree relatives of haemochomatosis patients; and patients with early onset atypical arthropathy, cardiac disease, or male sexual dysfunction. 88 We believe the current data support screening using serum TS followed by confirmatory HFE gene testing in family members of individuals with haemochromatosis, patients with lateonset type 1 diabetes mellitus, and patients with liver disease, especially those with chronic hepatitis C and NASH. Screening is also reasonable in individuals with elevated serum liver enzymes. At this time, there are inadequate data to warrant screening for haemochro-