Hereditary Hemochromatosis: Title and subTitle BreakGene Discovery and Its Implications for Population-Based Screening

HEREDITARY hemochromatosis is a disease of iron regulation that results in excessive iron absorption, and ultimately in iron overload that leads to hepatic cirrhosis, diabetes mellitus, cardiomyopathy, and other clinical complications. It is one of the few genetic diseases for which simple effective therapy exists: removal of iron by phlebotomy improves survival in symptomatic persons.^{1 - 2} When phlebotomy is begun prior to the development of cirrhosis or diabetes, affected people appear to have a normal life expectancy.^{1 - 3}

People with hemochromatosis can be identified before symptoms occur by abnormal serum iron measures. Because an effective therapy is available, population screening with serum iron measures has been advocated.^{3 - 9} Cost-effectiveness studies to date have supported this screening approach.^{4 - 5 ,7 - 8} Recently, a candidate gene for hemochromatosis, termed HFE (or HLA-H),^{10 - 11} has been identified and 2HFE mutations described.¹⁰ This discovery raises the possibility that genetic testing could provide earlier or more complete ascertainment of hemochromatosis-affected persons than screening with serum iron measures.

The Centers for Disease Control and Prevention and the National Human Genome Research Institute jointly sponsored a meeting—which was open to and attended by clinicians, scientists, public health officials, and members of advocacy groups—in March 1997 to explore the implications of the HFE gene discovery. A panel with expertise in epidemiology, genetics, hepatology, iron overload disorders, molecular biology, public health, and the ethical, legal, and social implications of genetic information reviewed the pathophysiology, epidemiology, and molecular genetics of hemochromatosis, and they assessed the clinical, ethical, legal, and social implications of genetic testing for hemochromatosis. Consensus was achieved by group discussion and confirmed by voice vote. The panel concluded unanimously that it would be premature to implement population screening using a genetic test for hemochromatosis. One member of the audience dissented. This article presents a summary of the discussions leading to the panel's conclusion and of the research agenda needed to resolve unanswered questions about the genetics of hemochromatosis.

Increased gastrointestinal iron absorption occurs in hemochromatosis and results in excessive uptake and persistent accumulation of iron in the body.^{12 - 15} When the body's storage capacity is surpassed, the iron is deposited in the tissues of multiple organs, causing tissue damage. Hemochromatosis is characterized by a broad array of clinical findings. These include cirrhosis, cardiomyopathy, diabetes, and symptoms such as arthralgia, generalized fatigue, bronzed skin, abdominal pain, impotence, and diminished menstrual cycles.^{1 - 3 ,12 - 15} Death is typically the result of liver failure, primary liver cancer, or the complications of diabetes or cardiomyopathy.^{1 - 3 ,12 - 15}

The diagnosis of hemochromatosis may be suggested by either clinical presentation or abnormalities in serum iron measures. A persistently elevated level of transferrin saturation (TS) (TS=serum iron/total iron binding capacity×100) is the most reliable measure for identifying affected persons^{4 - 9} and, in the absence of other causes, provides a presumptive diagnosis of hemochromatosis. An elevated serum ferritin level indicates increased iron stores and possible iron overload.^{12 - 15} The diagnosis of iron overload is confirmed by a liver biopsy specimen that documents elevated hepatic iron.^{12 - 16} Alternatively, the removal of 4 to 5 g of iron by serial phlebotomy confirms the presence of iron overload.^{12 - 15} Removal of stored iron by phlebotomy also represents the initial phase of treatment. The end point of this therapy is the reduction of iron stores as determined by hemoglobin level, reduction of serum ferritin level to a normal range, and improvement in clinical symptoms. In those with significant iron overload at the time of diagnosis, once- or twice-weekly phlebotomy may be required for more than a year to achieve these results.^{12 - 15} Subsequently, the disease is controlled by periodic phlebotomy, usually 3 to 4 times per year, to prevent reaccumulation of iron stores.

Symptoms of hemochromatosis are nonspecific, have a gradual onset, and mimic other common disorders. As a result, they may be attributed to other causes, and diagnosis is often delayed.^{3 ,13} Screening studies using serum iron measures indicate a prevalence of 1 in 200 to 400 persons, much higher than is recognized clinically.^{4 - 9 ,17 - 18} These data suggest that a substantial number of affected persons die of disease complications attributed to other causes or, alternatively, that screening identifies many people unlikely to develop clinical symptoms. Hemochromatosis is linked to the HLA region on chromosome 6.¹⁹ Studies using HLA linkage were instrumental in defining hemochromatosis as an autosomal recessive disease,²⁰ and based on the prevalence of the disease ascertained in screening trials, the carrier rate is estimated to be 1 in 10 in white populations.

The discrepancy between the prevalence of hemochromatosis estimated from TS screening and the number of people clinically diagnosed as having hemochromatosis raises questions of case definition. Asymptomatic people can be identified by an elevated TS level, but the sensitivity and specificity of this test depend on the cutoff used to define an elevated level.^{9 ,21 - 22} A low threshold increases false-positive results, including the identification of people who are heterozygous carriers of hemochromatosis mutations; a high threshold results in failure to identify some who are affected. HLA linkage studies can be used within families to identify affected persons, but this approach requires the presence of a clinically affected proband. Prospective follow-up of people identified through screening has been limited. As a result, it is not yet possible to determine the proportion of people with a hemochromatosis genotype who remain undetected in screening trials or would remain healthy even if untreated. Therefore the penetrance of hemochromatosis genotypes—or likelihood that clinical disease will occur in people with a given genotype—is unknown and requires further study.

Because hemochromatosis is an autosomal recessive disease, clinical symptoms are expected to occur with equal frequency in men and women. However, in all clinical series reported to date, men have outnumbered women, often by a substantial margin. In a German cohort only 11% of those affected were female² ; in a Canadian cohort, 35% of those presenting with symptoms were female.³ In a study from 2 tertiary referral centers, female patients with hemochromatosis were less likely to have cirrhosis than age-matched male patients and, on average, had lower serum ferritin concentrations and less iron overload, although the hepatic iron index was similar for both sexes.²³ Males and females are detected in equal numbers in those diagnosed by HLA-based family screening.^{3 ,24} In 1 HLA-based screening study of siblings of affected persons, sisters with the hemochromatosis genotype were less likely to have hepatic fibrosis or a high hepatic iron index than brothers; no women had cirrhosis, whereas 18% of men did.²⁴ Among those who had both iron overload and symptoms, the mean age was 51 years for men and 66 years for women.²⁴

A possible mechanism for the gender difference is women's loss of iron through menstruation and pregnancy. This hypothesis is supported by the observation of earlier onset of symptoms in women who have undergone premature menopause.^{1 ,23} However, the full range of hemochromatosis complications has been observed in affected women, and the age of onset of clinical symptoms was similar in male and female patients in a clinical cohort.²³ Among both men and women with hemochromatosis, marked variation occurs in the frequency of symptoms and fatal complications. Thus, gender differences in iron loss cannot fully account for variations in the severity of hemochromatosis. Intrafamilial variation has been reported, suggesting environmental modifiers.²⁵ Other factors that have been proposed to affect severity include differences in iron intake^{12 - 13} and alcohol use.²⁶ Genetic differences, either in hemochromatosis mutations or in polygenic modifying factors, could also be important contributors to differences in severity of hemochromatosis among affected people.

Data from Europe and North America suggest that hemochromatosis is more common in whites than in members of other races,¹⁴ and may be most common in people of Celtic origin.¹⁸ A recent study suggests that the prevalence of hemochromatosis in people of Hispanic descent may be similar to that in whites.²⁷ Hemochromatosis is reported to be less common in African Americans than in European Americans,²⁸ although a distinct non–HLA-linked primary iron overload syndrome has been described in African Americans and in sub-Saharan Africans.^{29 - 31} The usefulness of TS as a screening tool and appropriate screening threshold levels are likely to vary for people of different racial and ethnic groups.⁹

The identification of a candidate hemochromatosis gene, termed HFE (or HLA-H),^{10 - 11} provides a new way to approach questions of genetic heterogeneity. Identification of the HFE gene was based on its known linkage to the HLA region.¹⁰ DNA markers were used to identify a chromosomal segment within the region conserved in hemochromatosis-affected persons vs controls. Potential genes within this segment were identified by molecular techniques including direct selection, exon trapping, and genomic sequencing; each gene was then analyzed by DNA sequencing for presence of mutations in chromosomes from hemochromatosis-affected people. The only nucleotide change consistent with a hemochromatosis mutation was found in a major histocompatibility complex (MHC) class I–like gene now designated HFE . This mutation, designated C282Y (also termed 845A to indicate the position and nature of the base pair change within the DNA sequence), results in substitution of tyrosine for cysteine at amino acid position 282 of the protein product. Of the hemochromatosis-affected people in the study, 83% were homozygous for C282Y, and 5% were heterozygous. Additional HFE mutations were sought in the heterozygotes, leading to identification of a second mutation, termed H63D (also termed 187G ). Only 1 hemochromatosis patient was homozygous for the H63D mutation, but several hemochromatosis patients were compound heterozygotes for the C282Y and H63D mutations.¹⁰

The C282Y and H63D mutations have been identified in additional hemochromatosis populations in the United States, the United Kingdom, Canada, Australia, France, and Italy, as summarized in Table 1.^{32 - 40} In all series reported, most cases were homozygous for C282Y, with the percentage ranging from 60% to 100%.^{32 - 40} In 7 of the series, a small percentage (3%-7%) were compound heterozygotes carrying both mutations, and an even smaller percentage were homozygous for H63D (1%-4%, Table 1).^{32 - 39} The 2 mutations have not been observed on the same chromosome. A substantial minority of people with hemochromatosis (4%-21%) carry neither mutation. In 1 study, iron loading was substantially lower in affected people who had either 1 or no C282Y allele than in C282Y homozygotes.³³

Two control subjects, from 2 different studies, were found to be homozygous for C282Y^{33 ,40} ; 1 had evidence of iron overload⁴⁰ (Table 2). Family studies have documented C282Y homozygotes without evidence of iron overload, including some at elderly ages, identified because they were siblings of affected persons.^{39 ,41} In control subjects, heterozygosity for C282Y ranged from 2% to 14%, consistent with the estimated carrier rate of 10%. A small percentage (0%-4%) of control subjects was found to be homozygous for H63D. Heterozygosity for H63D in control subjects ranged from 16% to 23% (Table 2). In an international study of 2978 unaffected subjects, the C282Y allele was most common in persons of Northern European descent and was not found in African, Asian, or Australian subjects; the H63D allele was more common and more widespread.⁴²

Some affected people who lack HFE mutations may have a different disorder; those without affected family members may have a nongenetic cause of excessive iron accumulation. However, evidence for HLA linkage was found in some affected people negative for the C282Y allele, in populations from Italy and Alabama. The Italian cohort included C282Y-negative persons from families with disease linked to chromosome 6p, where the HLA region is located; mutation analysis, including direct sequencing in 3 cases, failed to reveal HFE mutations.³⁴ In the Alabama cohort, coexpression of HLA markers and iron overload was substantially higher in C282Y-negative probands than in control subjects.³³

These data are limited by lack of population-based case detection, use of different criteria for case definition in different studies (Table 1), and use of different populations for cases and controls in most studies. However, taken together, these observations suggest reduced penetrance for both hemochromatosis mutations, with substantially lower penetrance for H63D than for C282Y. The C282Y homozygotes appear to have the highest likelihood of developing iron overload, with a lower risk for compound heterozygotes (C282Y/H63D).^{33 ,43 - 44} It is not clear whether H63D homozygotes have an increased risk for iron overload.^{43 - 44} These data also raise the possibility of additional hemochromatosis genes being located in the same region as HFE. Though efforts to detect such genes have so far been unsuccessful,⁴⁵ the presence of additional linked hemochromatosis genes would be consistent with the duplication of genes of similar classes known to occur in the MHC region, where HFE is located.⁴⁶

Tests for hemochromatosis, whether they use serum iron measures or a DNA-based assay, could potentially be used in 2 settings: (1) as a screening tool to identify affected people in the general population, and (2) as a method for case finding within groups at increased risk, eg, people with clinical findings suggestive of hemochromatosis or family members of people affected with hemochromatosis. The deliberations of the expert panel reported in this article focused on the implications of genetic testing for general population screening.

Screening using serum iron measures (with threshold values for TS of 60% for men and 50% for women) is considered elsewhere.⁴⁷ A recommendation for population screening using serum iron measures has been made by the College of American Pathologists and others,^{4 - 9} based on results of screening trials.^{4 - 8 ,17 ,21 - 22} However, an evaluation of screening using the criteria of the US Preventive Services Task Force⁴⁸ suggests that there is currently insufficient evidence to recommend universal screening because of uncertainties about the natural history and burden of suffering of hemochromatosis.⁴⁷ The use of serum iron measures for case detection in high-risk groups, including patients with symptoms suggestive of hemochromatosis and relatives of hemochromatosis-affected people, is recommended.⁴⁷

With identification of the C282Y and H63D mutations, a DNA-based test for hemochromatosis is possible, and commercial tests for these mutations are available. Although the genetic test currently costs much more than TS screening, preliminary data suggest that a low-cost approach to mutation screening could be developed (Ernest Beutler, MD, PhD, unpublished data, 1997). Assuming a comparable cost and high sensitivity and specificity for the genetic test (ie, 0.85 and 1.0, respectively), Adams and Valberg⁴⁹ found that genetic screening compared favorably with screening with conventional iron measures. Key issues in considering the use of a DNA-based test for screening are the predictive value of the test and implications of a DNA-based diagnosis.

Uncertainties about genotype-phenotype correlations in hemochromatosis, including uncertainty about the percentage of homozygotes that will develop clinical disease and the effect of genetic or environmental modifiers on clinical presentation, indicate the need for caution in the use of genetic testing. In addition, genetic testing raises general concerns of stigmatization and discrimination and of possible breaches of privacy and confidentiality. Anecdotal information has already emerged concerning the loss of insurance coverage and employment when a person is diagnosed as having hemochromatosis.⁵⁰ Substantive concerns have been raised about these issues for other genetic conditions as well.^{51 - 52} Efforts to pass laws prohibiting discrimination in health insurance and employment are being made,⁵³ but they are not yet robust enough to ensure full protection from such discrimination. There is also concern about more subtle harms associated with genotype-based diagnosis, including diminished self-worth and family disruption. The complex nature of these concerns led the National Institutes of Health and the Department of Energy to form the Task Force on Genetic Testing, which has called for strict scrutiny of genetic tests when predictive value is poorly defined or treatment options are uncertain.⁵⁴

When an intervention to improve health is available, as it is for hemochromatosis, the adverse effects of genetic testing may be outweighed by the benefits. It can be argued that, among genetic diseases, hemochromatosis is most like phenylketonuria (PKU), a condition in which mental retardation is prevented by universal newborn screening and institution of early dietary therapy.⁵⁵ Hemochromatosis, as defined by TS screening, is 1 to 2 orders of magnitude more frequent than PKU (1:200-400 vs 1:10000), and morbidity and excess mortality appear to be preventable with a treatment likely to be viewed as less burdensome than the PKU diet.

The comparison with PKU is instructive in another way, however, in that newborn testing for PKU is based on measurement of phenylalanine level, a phenotypic assay. As detailed knowledge of the genetics of PKU has emerged, mutations of varying clinical consequence have been documented.^{54 ,57} Despite imperfections in the phenotypic assay, a test based on genotype would be far more problematic, because comprehensive genetic testing would be costly and identify persons with trivial abnormalities of phenylalanine metabolism, while testing for a limited set of genotypes would miss affected persons. Cystic fibrosis (CF) offers another example of a genetic disease with multiple mutations of varying severity, some of little or no apparent clinical significance.^{58 - 59} Many CF-affected persons carry rare CF mutations that would not be detected using a commercial screening test. Also, the variation in clinical severity seen in CF-affected people is only partially explained by genetic differences.^{59 - 60} As we acquire more knowledge about the molecular basis of genetic disease, it becomes increasingly clear that variable expressivity (ie, modification of a genetic trait by other genes or the environment) is the rule rather than the exception in genetic disease. Although genetic testing for hemochromatosis currently involves only 2 known mutations, the same issues of reduced penetrance and limited sensitivity of the genetic test arise.

A DNA-based diagnosis of hemochromatosis would increase identification of affected persons before symptom onset, and therefore at a time when phlebotomy is most likely effective. However, early diagnosis would also prolong the period during which a healthy person could be subjected to stigmatization and discrimination and would increase cost and burden of treatment. Newborn screening is needed for PKU because early treatment is vitally important, but the late onset of symptoms in hemochromatosis argues for screening at a later age. Also, the examples of PKU and CF suggest there may always be an inherent uncertainty in a hemochromatosis diagnosis made on the basis of genotype rather than on evidence of increased iron stores or iron overload. This uncertainty could make a genetic test for hemochromatosis a measure of genetic susceptibility, similar to genetic susceptibility tests for other late-onset diseases like cancer.

It can be argued that a phenotypic screening test is always preferable to a genotypic test, because phenotype is the clinical concern. In this view, the availability of a phenotypic test should lead to increased caution in the use of a DNA-based test. In hemochromatosis, convincing evidence of better outcomes from mutation testing would be needed before DNA-based screening would be favored over screening using serum iron measures. Although hemochromatosis meets many of the criteria for population screening adopted by the World Health Organization⁶¹ (Table 3), questions remain about the natural history of the disease and the predictive value of screening tests, and the degree of uncertainty is greater when the disease is defined by genotype. Until more is known, treatment decisions in hemochromatosis should continue to be based on iron status rather than genotype.

Testing for hemochromatosis mutations is simpler and more accurate than HLA-linkage testing and, thus, has a potential role in family-based screening. If 2 hemochromatosis mutations can be identified in an affected person, siblings at risk to develop hemochromatosis can be identified by testing for the presence of these mutations. This approach provides a reliable and inexpensive method for family-based case detection in most cases. If the affected person has only 1 identifiable hemochromatosis mutation, however, DNA-based testing will not distinguish him or her from heterozygous carriers, and if no mutation is found, this approach cannot be used.

Testing for hemochromatosis mutations may also provide a better confirmatory test for hemochromatosis than liver biopsy or quantitative phlebotomy in persons who undergo TS screening. As with family-based screening, however, hemochromatosis can be confirmed only when 2 hemochromatosis mutations can be identified. Also, until more is known about the penetrance of hemochromatosis genotypes, there will be uncertainty about the predictive value of a hemochromatosis genotype for future disease, when used in either family studies or screening follow-up.

The work of understanding the molecular genetics of hemochromatosis has just begun. The genotype studies accomplished to date indicate that hemochromatosis, similar to most other genetic diseases, is genetically complex. A careful search for additional mutations of the HFE gene is needed, as is a rigorous search for other hemochromatosis-associated genes, in both the MHC region on chromosome 6 and in other chromosomal regions. Clinical correlation will be needed to address the question of allelic and locus heterogeneity. If genetic studies confirm that genotype is a significant contributor to phenotypic variation in hemochromatosis, additional issues will need to be addressed. These include whether all hemochromatosis genotypes meet the threshold for identification by screening, whether the efficacy of phlebotomy is the same for people with different genotypes, and what genetic and environmental modifiers affect the phenotype associated with different hemochromatosis genotypes.

Population-based studies are needed to address genotype-phenotype relationships. Cross-sectional and case-control studies, with adequate numbers of older men and women, will help to determine the age- and sex-related penetrance of the various hemochromatosis genotypes. Prospective studies are also needed to assess the predictive value of serum iron measures, particularly the degree to which TS predicts clinical outcome and whether changes in serum ferritin predate iron overload. Using genetic data, researchers will be able to assess definitively whether there are people with hemochromatosis genotypes who demonstrate parenchymal iron overload without elevated serum ferritin levels. Studies of this kind will also provide more definitive information concerning the risk of end-organ disease in heterozygotes.

Many social and clinical concerns associated with genetic screening for hemochromatosis remain unresolved as well. Although there are anecdotal reports of discrimination based on genetic information, there are few empirical data on the scope of the problem. Psychological effects of genetic testing in asymptomatic people, efficacy of methods for providing counseling and education, and the outcome of protective legislation are still largely unknown.

The use of genetic testing for hemochromatosis, and by implication the definition of hemochromatosis by genotype, poses fundamental questions about disease and health. Current clinical, epidemiologic, and genetic data suggest that the relationship between the hemochromatosis genotype and iron overload is complex. Many questions about the etiology, molecular basis, and clinical course of hemochromatosis-associated iron overload remain unresolved. Thus, it was the consensus of participants in the meeting reported herein that genetic tests for HFE mutations should not be used in population screening at the present time. Genetic testing may have a role in family studies, if the genotype of the affected family member is known, and is likely to help in confirming diagnosis in some people whose serum iron measures suggest hemochromatosis. However, there is currently considerable uncertainty concerning the proportion of people with hemochromatosis who carry the C282Y mutation, the penetrance of both known hemochromatosis mutations, and the existence of other hemochromatosis genes or mutations. Ideally, use of genetic tests for hemochromatosis, even for these restricted purposes, should occur within a context of ongoing data collection so that these questions can be resolved.⁴⁹

As with other genetic diseases, a genetically based diagnosis of hemochromatosis poses the dilemma of potential psychological, social, and economic harm to persons identified as being affected. In population screening for hemochromatosis, DNA-based testing currently offers no advantage over phenotypic testing and may introduce unnecessary risk. As research goes forward, it will be important to ensure that advances in genetic knowledge are linked to an understanding of their social consequences. If future studies strengthen the case for DNA-based testing in hemochromatosis, strategies will be needed to prevent social harm resulting from such testing.