|Home | About | Journals | Submit | Contact Us | Français|
The prevalence of the C282Y homozygous HFE genotype is high, approximately 1 in 200 in populations of Anglo-Celtic descent, and most authorities assumed this mutation would have a high clinical penetrance. Recent studies report the clinical penetrance of C282Y homozygous hereditary haemochromatosis is much lower than its prevalence, with possibly less than 5% developing clinical disease, although there is lack of consensus on a precise estimate. This review discusses reasons for this paradigm shift, including controversy on various definitions of clinical penetrance.
It is inescapable that there are pronounced variations in clinical penetrance, and that certain C282Y homozygous individuals will not develop the clinical phenotype. This has prompted a search for modifier gene mutations amongst iron-metabolism genes, especially the known non- HFE haemochromatosis genes, and for possible environmental factors which might explain the observed variation in clinical penetrance.
Following discovery of the HFE gene and its main C282Y mutation in 1996,1 genotyping for this mutation was rapidly and widely instituted, especially in countries with predominantly Anglo-Celtic ethnicity. In Australia, HFE genotyping was the first DNA-based test funded by the healthcare system, provided prior testing had revealed persistently elevations in at least one of the serum iron indices.
Within 2–3 years of cloning of the HFE gene, it became apparent that where Anglo-Celtic ethnicity prevailed, 80 to 100% of patients who expressed the clinical signs of hereditary haemochromatosis were homozygous for the C282Y mutation.2 One in seven Australians of Anglo-Celtic descent are heterozygous for the C282Y mutation while 1 in 200 are homozygous,3 making HFE hereditary haemochromatosis the single most common autosomal recessive gene disorder yet described. It is worth noting that this mutation is less common in other population groups such as southern Europeans, much less common in Asian and the Asian sub-continent and is not found in indigenous Australians.
Although the definition of HFE-related haemochromatosis promulgated by the International Consensus Conference on Haemochromatosis includes both C282Y homozygous and compound heterozygous [C282Y / H63D] genotypes,4 this review discusses only C282Y homozygous HFE hereditary haemochromatosis. Most studies of penetrance only report on this genotype, which typically comprises approximately 60–70% of all referrals to tertiary centres for further investigation of possible haemochromatosis.5
The textbook definition of penetrance is apparently straightforward, it is the proportion of people who have the genotype that develop the phenotype. However the clinical symptoms and signs of C282Y homozygous haemochromatosis are highly variable and there is no consensus on which phenotypic features should be adopted in studies to investigate penetrance. In this review, preference is given to community based studies conducted in the post genotyping era, to avoid the possible problem of ascertainment bias introduced by cross-sectional case series or studies of apparently ‘unaffected’ relatives of haemochromatosis probands. Studies of blood donor populations are also not cited as the opposite bias may be operating, where periodic donations may delay or cancel expression of the disease phenotype.
Biochemical penetrance of the C282Y homozygous genotype may be defined as expression of the biochemical phenotype, usually by a raised serum transferrin saturation and/or ferritin concentration. Serum ferritin is often regarded as a surrogate marker of iron loading, but it is prone to false positives as it is an acute phase reactant and may be elevated for a variety of reasons unrelated to iron overload. Biochemical penetrance is usually at least 50%, for example Asberg reported on a large Norwegian community population and found 72% of 126 women and 91% of 171 men with newly diagnosed C282Y homozygous haemochromatosis had an elevated serum ferritin.6 In another large study of 152 C282Y homozygotes identified through membership of an integrated medical care system in California, 75% of men and 40% of women had raised transferrin saturation and 76% of men and 54% of women had raised serum ferritin.7 In Australia, genotyping of the Busselton community population revealed 12/12 new C282Y homozygotes had raised transferrin saturations and 7/12 had raised serum ferritin.3
The highly variable nature of the clinical symptoms and signs of C282Y homozygous haemochromatosis also evolve and change over a lifetime and various stages in the evolution of the disease can be used to define clinical phenotype. The early signs of haemochromatosis are notoriously difficult to score or define and they also occur commonly in the population at large, for example fatigue, arthritis or darkening of the skin. The extreme phenotype which includes cirrhosis, hepatocellular cancer and diabetes mellitus is thankfully now less frequently seen, because of increased awareness of iron overload disease and the advent of genotyping, especially when applied to family studies.
The Californian community study used a detailed health questionnaire followed by physician interview and assessed the frequency of symptoms most commonly associated with full-blown disease.7 They reported that C282Y homozygous individuals did not have any increase in the number of the following symptoms compared to age matched controls: limited general health, chronic fatigue, joint symptoms, skin darkening, impotence, diabetes, arrhythmias and aspartate aminotransferase concentrations >40 U/L. Liver fibrosis was not assessed, but collagen IV, a surrogate serum test for liver fibrosis in haemochromatosis was undertaken and 26% of homozygotes exceeded the normal range compared to 11% of normal controls.
An Australian community population study included hepatomegaly, arthritis and skin darkening as indicative of penetrance but the most important factor determining clinical penetrance were the liver biopsy findings, including quantitating iron overload by measuring liver biopsy iron.3 Although highly invasive and not without risk, liver biopsy defines the extent of fibrosis progression towards cirrhosis and liver biopsy iron allows staging of the degree of iron overload according to recommendations of the International Consensus Conference on Haemochromatosis: minimal iron overload, hepatic iron concentration 30–99; moderate, 100–199; and severe >200 μmol/g dry weight.4
Liver biopsy is being phased out of the clinical work-up of newly diagnosed C282Y homozygous cases and current recommendations reserve liver biopsy for C282Y homozygous individuals who fulfil at least one of the following criteria:8
It is useful to consider the evidence in favour of a high clinical penetrance for C282Y homozygous haemochromatosis. In the pre-genotyping era, observational studies from which clinical penetrance data were obtained usually recruited haemochromatosis patients with established clinical symptoms from tertiary referral centres. Ascertainment bias was therefore a consistent feature of early studies. Using such highly selected patients, an Australian study showed 94% of men and 68% of women expressed the iron-loaded phenotype and the remainder did not show clinical penetrance of overt disease.9 Using more stringent criteria for severe clinical penetrance collated from a 30 year database in the pre-genotyping era, Adams et al estimated in 1995 that 43% of male and 28% of female patients developed one or more of the following life threatening manifestations; heart failure, diabetes, cirrhosis, or hepatocellular carcinoma.10 The effect of this bias was not appreciated until population based genotyping work selected asymptomatic C282Y homozygous haemochromatosis subjects for study.
A few studies have documented the changes in serum iron indices which developed over time in asymptomatic C282Y homozygotes initially detected by population screening. In Australia, genotyping of about 3000 subjects from the Busselton community population revealed 12/12 new C282Y homozygotes had raised transferrin saturations and 7/12 had raised serum ferritin.3 A 17 year follow-up study using stored serum was conducted on 10 newly diagnosed C282Y homozygotes.11 Although the median transferrin saturation value increased from 42% to 76%, ferritin was much more variable. Ferritin levels increased in 4 subjects, remained constant in a further 4 who had initial levels below 200 ug/L and decreased in 2 subjects. Interestingly of the 4 subjects whose ferritin levels were persistently elevated over 500 ug/L for the 17 year period, 3 developed stage III or IV fibrosis based on the METAVIR scoring system, in which cirrhosis is stage IV.
A Canadian study recruited newly diagnosed, untreated asymptomatic C282Y homozygotes and selected 22 who initially had a normal ferritin level for follow-up over a median period of four years.12 These comprised 14 pre and 4 post-menopausal women and 4 men. 20 of 22 failed to show any significant increase in ferritin level. Of the 2 who increased, only one, a male, had an increase over a 3 year period which exceeded the upper limit of normal for ferritin, from 295 to 344 ug/L.
Longitudinal assessment of serum iron indices in 23 C282Y homozygotes identified in the Copenhagen City Heart Study was performed for up to 25 years.13 Once again, while transferrin saturation values uniformly increased, the behaviour of ferritin was variable. At the conclusion of the follow-up period, 10 of 16 C282Y homozygous women and 6 of 7 men had ferritin levels higher than 200 ug/L. The authors concluded that, on the basis of their observations, C282Y homozygotes identified during population screening at most need to be screened for manifestations every 10 to 20 years.13
The key study which alerted medical researchers to the possibility that clinical penetrance had previously been overestimated was published in Lancet in 2002.7 This study reported on about 41,000 subjects attending a health appraisal clinic in southern California. The key novel features of this study were:
The observed C282Y homozygote frequency was the same or slightly more than the calculated frequency expected for a population in Hardy-Weinberg equilibrium across all ages, suggesting that homozygotes were not under represented due to illness or death. Symptoms and signs of haemochromatosis were no more frequent in C282Y homozygotes than in matched controls. Only one homozygote had multiple clinical features of haemochromatosis, leading the authors to conclude that clinical penetrance was about 1%.
This very low clinical penetrance has been questioned by several groups with experience in the management of hereditary haemochromatosis.14 The main objections to the study are summarised here:
Most objections were adequately rebutted, however the potentially most serious objection centred on the absence of liver biopsy data to determine the extent of liver fibrosis in the C282Y homozygotes. Some liver biopsy data was available from another large community population study of about 65,000 Norwegians.6 Given the reported gene frequency of 0.078 in Norwegians, one would expect to find about 400 C282Y homozygotes among these subjects. Liver biopsy was performed on a sub-set of 179 C282Y homozygotes who had high ferritin levels and therefore met the revised criteria for recommending liver biopsy. Severe organ damage was found in 4 homozygotes with cirrhosis and moderate fibrosis in a further 8, a total of 3% of the 400 putative homozygotes. Of the 12 with either cirrhosis or moderate fibrosis, 11 were men. Fibrosis seen on liver biopsy is a solid quantitative definition of clinical penetrance and although the exact total number of homozygotes was not established, puts the penetrance at about 3%.
The absence of liver biopsy data in most community based studies has left lingering doubts on low estimates of clinical penetrance. Because modern criteria for assessing newly diagnosed C282Y homozygotes only specifies liver biopsy if certain criteria are fulfilled,8 it is highly unlikely that future studies could ethically be designed to include liver biopsy. However one thorough community based study which included liver biopsy was initiated by the University of Utah prior to the availability of HFE genotyping and subsequently followed up with positive identification of C282Y homozygotes.16 Family members, mainly siblings, were recruited on the basis of being HLA-identical to a proband with an established clinical diagnosis and who subsequently were shown to be C282Y homozygous. Almost all such family members underwent liver biopsy. Clinical penetrance was assessed with these objective criteria: cirrhosis, fibrosis, elevated aminotransferases with no causes other than iron overload, and radiographic evidence of haemochromatotic arthropathy. Of 52 C282Y homozygote men over the age of 40 years, 52% fulfilled at least one of these criteria; as did 16% of the 43 post-menopausal women.
However ascertainment bias may have been introduced by studying relatives of probands which showed clinical penetrance, and a statistical analysis revealed that this was indeed the case. Siblings of probands with liver disease were three times more likely to have liver biopsy abnormalities than siblings of probands without liver disease.15 This finding in itself argues strongly for the existence of putative modifier genes to be further discussed below. To minimize the effects of such modifier genes, an additional analysis was performed on C282Y homozygotes who were relatives of healthy probands and lower percentages for clinical penetrance were obtained; 29% of men over the age of 40 years and 11% of post-menopausal women fulfilled at least one of the criteria. These estimates of clinical penetrance are of more serious concern than the estimate of 1% obtained from study of the health appraisal clinic population in southern California.7
Magnetic resonance imaging has a role in the investigation of suspected liver iron overload. Recently, a non-invasive method of measuring and imaging liver iron concentration in vivo has been developed which utilises common clinical 1.5T magnetic resonance imaging units.16 The technique measures an MRI parameter known as the hydrogen proton transverse relaxation rate R2, which has a high degree of sensitivity and specificity to liver iron concentration over a wide range. A digital pathology model has been proposed to deliver the technology to the clinical community, whereby magnetic resonance imaging data are transmitted electronically to a centralised service facility.17
If we agree to define clinical penetrance of C282Y homozygosity in terms of becoming clinically ill or having a life expectancy compromised by iron overload, or both, the clinical penetrance haemochromatosis disease remains controversial and may never be precisely known. The current best estimate is that it is probably lies somewhere between the 1% obtained in California and the 29% of men over the age of 40 years and 11% of post-menopausal women found in Utah.
The ease with which genotyping for common single-gene disorders such as the C282Y mutation of HFE can be performed has emphasized and reminded us that patients with the same genotype can and do express different clinical penetrance. A significant future challenge is to determine the factors responsible for expression of the clinical phenotype. Whether iron overload eventually manifests as clinical disease depends on the interplay between environmental aspects such as diet and alcohol intake and genetic and epigenetic factors. While recognising this interplay, for simplicity we will consider environment and genes separately.
Dietary iron intake would be expected to exert an important influence on whether iron overload develops, but does not seem to be a critical factor. This may be because a normal diet provides a narrow range of iron, typically 10–15 mg per day. Many food components interfere with iron absorption, for example cereal brans, egg yolk and tannic acid in tea. The amount of iron absorbed is held within a narrow range of 1–2 mg per day with females absorbing slightly more than males, whereas in haemochromatosis patients, maximal absorption is 8–10 mg per day.18 Foods such as red meat which have a high haem content are common in the Western diet and provide iron in a highly bioavailable form. Haem has a specific receptor to aid absorption, and iron is released via the enzyme haem oxygenase. In addition to its high haem content, meat promotes the absorption of non-haem iron, possibly through the formation of stable amino acid iron complexes. The other important promoter of non-haem iron absorption is ascorbic acid, which may act in concert with duodenal cytochrome b, an iron-regulated ferric reductase.
Both inorganic iron and haem absorption are reportedly increased in haemochromatosis, however haem absorption was independent of iron stores, i.e. absorption was not reduced by high iron stores indicated by an elevated serum ferritin.19 There has not been enough work done specifically in C282Y homozygous haemochromatosis to properly define the role of diet in iron loading.
Alcohol consumption has been proven to contribute to liver iron loading and calculation of the ratio of hepatic iron to age, the hepatic iron index, was first proposed to differentiate between patients with hereditary haemochromatosis and alcoholic siderosis.20 Ferritin concentrations in subjects who are wild-type for C282Y show significant increases with increasing alcohol consumption.21 The mechanism for the effect of alcohol consumption on serum ferritin is poorly understood. Alcohol appears to have many modes of action which could affect serum ferritin concentrations. These include: an induction of an inflammatory response in the liver with resultant de-novo ferritin synthesis, causing ferritin release from liver cells and altering iron absorption by changing gut permeability.22 Two recent papers have reported the effect of excessive alcohol consumption [>60 g per day] on C282Y homozygous subjects. In 378 such patients from Brittany, France those who consumed excess alcohol had significantly increased serum ferritin and both alanine and aspartate aminotransferases and by inference, increased risk of cirrhosis.23 Australian C282Y homozygous patients who consumed excess alcohol and underwent liver biopsy were confirmed to have an increased risk of severe fibrosis.24 However in discussing the 152 C282Y homozygotes identified by genotyping a health appraisal clinic population, Beutler has reported a contrary finding, and reports no relationship at all between serum ferritin levels and frequency of drinking, although no data was given on amounts of alcohol consumed.25 This again raises the possibility of ascertainment bias, as both of the earlier reports were retrospective studies of symptomatic patients from tertiary care centres. It may be that alcohol behaves differently in C282Y homozygotes who do not display clinical penetrance, such as those identified in the health appraisal clinic population.
The concept of a mutation giving rise to clinical penetrance only in individuals with another mutation is a plausible explanation for the spectrum of clinical phenotypes seen in C282Y homozygotes. In an Australian study of a large community sample of twins, the role of HFE mutations were compared to other genetic factors in determining variations in iron stores.26 Variation in HFE C282Y and H63D accounted for less than 5% of the phenotypic variance of serum ferritin values, indicating highly significant effects of as yet unidentified genes on iron stores.
The most obvious candidate modifying genes which could accelerate iron loading in C282Y homozygous patients if they carried a second mutation are those already identified as causes of non-HFE haemochromatosis, a constantly growing list. At the time of writing, the main genes involved are as shown in the Table.27–30 Not all the genes that can give rise to iron overload in humans are listed, for example mutations in transferrin, caeruloplasmin and H [heavy chain] ferritin have all been described, but are either exceedingly rare or have been sequenced in sufficient numbers of C282Y homozygous patients to rule out an important role in producing the phenotype in humans. In addition some genes known to be involved in iron metabolism, but thus far associated with iron overload only in animals such as mice, have not been considered, examples include β 2 microglobulin and transferrin receptor 1.
Several studies have screened multiple non-HFE genes in C282Y homozygous patients which did not find mutations associated with iron loading. A Californian group has looked for such genes in two studies. The first study examined transferrin receptor-1, ferroportin, ceruloplasmin, ferritin light and heavy chains, iron regulatory proteins (IRP)-1 and -2, and hepcidin with negative results.31 A much wider selection of genes was examined for the second report, however only five C282Y homozygous patients with clinical disease were studied, and also with negative results: the coding sequence, exon-intron junctions, and promoters of each of these genes was sequenced, transferrin, transferrin receptor 1, transferrin receptor 2, ferritin-L, ferritin-H, IRP1, IRP2, HFE, beta(2) microglobulin, mobilferrin/calreticulin, ceruloplasmin, ferroportin, NRAMP1, NRAMP2 (DMT1), haptoglobin, heme oxygenase-1, heme oxygenase-2, hepcidin, USF2, ZIRTL, duodenal cytochrome b ferric reductase (DCYTB), TNF alpha, keratin 8, and keratin 18.32 Another group have developed denaturing high performance liquid chromatography (DHPLC) methods more suitable for rapid scanning of DNA, which could be used to examine larger cohorts.33
A UK study examined two kindreds and reported synergistic interactions between two newly described HAMP (hepcidin antimicrobial protein) mutations and the C282Y mutation.34 This led the authors to propose a digenic model of inheritance where the severity of the phenotype of either C282Y heterozygous or homozygous patients could be modified by heterozygosity for mutations disrupting the function of HAMP. A French group also obtained positive data on HAMP as a modifier gene by screening a large cohort of 392 C282Y homozygous patients for known HAMP mutations by DHPLC. 5 of the 392 patients or 1.3% of the total also carried a previously described HAMP mutation and were among the most severely iron-loaded patients.35 It is too early to speculate on whether HAMP mutations play a role in altering the penetrance of C282Y homozygous haemochromatosis.
Even less work has been carried out to see if the newly described hemojuvelin associated haemochromatosis30 may be a candidate modifier gene, and other non-HFE genes probably remain to be found.
Iron loading is affected by the common polymorphisms of haptoglobin, the haemoglobin binding protein of plasma; C282Y homozygous men carrying the Hp 2-2 polymorphism had higher iron stores assessed as serum ferritin than those with the other two polymorphisms, but not increased liver fibrosis.36 Several studies have also suggested that inheritance of two copies of the ancestral haplotype leads to higher iron overload than either one or no copies.37,38 The implication is that genes in the region of the Major Histocompatibility Complex act as genetic modifiers, but a detailed analysis has failed to reveal candidate loci.31
Modifying genes may exert an effect on certain aspects of iron-loading disease, for example liver fibrosis is related to polymorphisms in the promoter region of the tumour necrosis factor α gene.39 While this has not been directly tested by liver biopsy, one study failed to find any relationship between collagen IV, an indirect serum marker of fibrosis, and tumour necrosis factor genotype.40
Given that a C282Y homozygous genotype is a necessary but not sufficient condition for the development of clinical disease, the search must continue for the environmental and/ or genetic factors which determine whether any given C282Y homozygous individual develops iron loading disease.