Article Text
Statistics from Altmetric.com
Editor,—The UK Haemochromatosis Consortium’s report (Gut1997;41:841–4) questions the importance of the H63D allele in hereditary haemochromatosis (HHC). In their commentary, Goldwurm and Powell (Gut1997;41:855–6) also doubt the relevance of H63D. However, there is compelling evidence that the H63D allele is associated with HHC.1-12
To evaluate the association of H63D with haemochromatosis, one must first recognise that H63D and C282Y are independent mutations of theHFE gene. Because the C282Y mutation is so common and is highly penetrant, homozygous C282Y predominates among cases of haemochromatosis, and H63D is not observed in C282Y homozygotes. If one excludes from analysis the C282Y homozygotes, then the strong association of H63D with HHC becomes apparent. For five large series from North America1-5 that have been analysed in this manner, we found a highly significant (p=0.000001) twofold “enrichment” of HHC cases with H63D, compared with random control cases. For studies reported from Europe,6-10 we also found a highly significant “enrichment”, in which groups with HHC have roughly twice the frequency of H63D compared with control groups: Borot et al, p=0.003; Jouanolleet al, p=0.00004; Carellaet al, p= 0.073; Martinezet al, p=0.000002; and Gottschalket al, p=0.02. When all the data reported from Europe and from North America are combined, including those of the Consortium, the association of H63D with HHC is quite significant (p < 0.000001).
The strong association of H63D with haemochromatosis might depend on the fact that patients who are compound heterozygotes—that is, C282Y/H63D, are at risk of developing clinically significant iron overload. Therefore, one may question whether H63D itself is associated with haemochromatosis. We have tested this possibility by re-analysing all the published data (combining data from 10 studies, including those of the UK Haemochromatosis Consortium). In this re-analysis, we have compared the frequency of homozygous H63D, of which approximately 50 cases have now been described (a third with clinically significant iron overload), with the combined frequencies of homozygous C282C (that is, homozygous normal) and heterozygous H63D (that is, C282C/H63D), in random controls and patients with haemochromatosis, after exclusion of all cases that have C282Y. Thus, we have tested the effect of H63D independently of any effect that might be observed in cases that also have C282Y. In this analysis of 1093 control subjects and 163 patients with HHC, we found that the proportion of H63D homozygotes (6%) in the latter group was nearly three times higher than the proportion of H63D homozygotes in the control group (2.2%), χ2=8.4, p= 0.0038. This significant enrichment of HHC cases with homozygous H63D is slightly greater than that which we and others have observed when comparing the proportion of H63D alleles in controls and HHC cases prior to exclusion of all C282Y heterozygotes from the analysis.
These analyses confirm that H63D is an independent risk factor for HHC and not just a polymorphism of the HFE gene. Others independently performed similar analyses of subsets of the data we analysed, and came to the same conclusion.11 12However, although there is compelling evidence for this association, it is also important to note that the H63D allele has low penetrance.1 11 12 Analysis of the genotype distributions in the various studies suggests that H63D homozygotes have only a two- to fivefold increased risk for developing HHC compared with normal controls.
Is a single test for C282Y sufficient for ascertainment of HHC? In Italy and in Alabama, USA, fewer than 70% of patients with HHC are homozygous for C282Y. Even in the UK, where a very high frequency of C282Y has been observed, 10% of HHC cases cannot be detected if one uses a test for C282Y only, and more than 30% in Alabama and in Italy and France (excluding Brittany). Clinicians may be misled to believe that patients who are not homozygous for C282Y do not have HHC. Testing for H63D helps to close this gap. Even if tests were available for both HHC alleles, too many cases would not be correctly ascertained if clinicians rely solely on the DNA tests of theHFE gene to make this diagnosis. With addition of the test for the H63D mutation the interpretation of results must take into account the low penetrance of the C282Y/H63D and H63D/H63D genotypes. However, this is true also for interpretation of C282Y test results as the genotype C282Y/C282Y also shows incomplete penetrance. The DNA test is a useful diagnostic method but, as with most laboratory tests, results must be interpreted in the context of other clinical and laboratory findings.
References
Reply
Editor,—We thank Fairbanks et al for their comments and agree with their conclusion that the H63D mutation may be associated with iron accumulation. This is supported by the recent finding that the HFE protein with the H63D mutation does not reduce the affinity of transferrin for its receptor in the same way as the wild type protein.1-1 However, we have reservations about combining data from various groups of patients and controls where the frequency of the mutations in the general population varies greatly.
We should, however, point out that our study of the UK patients was completed before November 1996 and submitted in December 1996. Most of the studies referred to by Fairbanks et alhave been published since that time.1-2-1-5 We found four “compound heterozygotes” with iron overload and two others who had a diagnosis of haemochromatosis but did not have sufficient iron accumulation to satisfy our criteria for iron overload. Since then two of us (MW (http://www.uwcm.ac.uk/uwcm/hg/worwood/) and WMCR) have been providing a diagnostic service for haemochromatosis. We test for both mutations and regard testing for transferrin saturation and ferritin as essential in making a diagnosis. The H63D mutation is valuable not only because compound heterozygotes must be regarded as “at risk for iron accumulation”1-6 but also because the test provides a check on the results for the C282Y mutation. We have never seen an example of a subject homozygous for C282Y who also has the H63D mutation.
Table 1 shows the genotypes for 423 samples received in the haematology department of the University Hospital of Wales and 42 samples received at the Wessex Regional Genetics Laboratory for testing. These are either family members of patients with haemochromatosis or patients suspected of having the condition. Note the increased frequency of the compound heterozygotes compared with control subjects (blood donors from South Wales studied by MW and KJHR and colleagues). However, there is a decreased frequency for H63D homozygotes in Cardiff but not in Wessex, suggesting regional variation. Although expressing frequencies in terms of non-C282Y chromosomes may demonstrate an enrichment in haemochromatosis, we have seen only one further case of clinical iron overload in people homozygous for H63D and that was in Wessex.
That the clinical penetrance of the H63D mutation is very low is supported by studies of blood donors from Jersey.1-7Subjects having either the homozygous C282Y genotype or both mutations (H63D/C282Y) had increased transferrin saturation and serum ferritin concentrations compared with subjects with other genotypes. There was no evidence that heterozygotes or subjects homozygous for H63D had raised iron stores. However, much larger studies will be needed to establish any relations between genotype and iron status in normal subjects.
We would also point out that in a study published last April describing the global prevalence of the H63D and C282Y mutations,1-8the C282Y mutation was almost exclusively found in peoples of Northwest European origin, whereas the H63D mutation had a more global distribution although was still highest in parts of Europe. The belief that genetic haemochromatosis was a disease limited to peoples of European extraction may explain why it has only been observed in these populations due to selection bias. However, the problem arises as to why in places such as India and Saudi Arabia the H63D allele frequency seems to be over 5% and haemochromatosis has not been reported or recognised in these populations. Is the H63D allele conferring protection against anaemia due to diseases such as malaria or hookworm, or anaemia due to multiple pregnancies thus producing a clinical ascertainment bias against diagnosing haemochromatosis? Clearly even in the developed world the penetrance of disease in H63D homozygotes is variable so it might be expected that in conjunction with factors such as malnutrition the H63D allele may be playing a protective rather than disease causing role. It is also important to point out that both C282Y and H63D mutations have been implicated in iron overload in sporadic porphyria cutanea tarda1-9 1-10 and therefore these mutations should also be considered when studying patients with this condition.
We therefore agree with Fairbanks et al that the diagnosis of haemochromatosis requires testing for both the common mutations as well as careful assessment of iron status and other clinical and laboratory findings, and welcome their analysis. The accumulation of both genetic and biological data following early studies of genotype, including our own, underline the rapidity with which the field is developing and illustrate the need for regular revision of diagnostic, screening, and management guidelines. We believe that appropriate management of patients with mutations inHFE or clinical evidence of iron overload in the absence of mutations requires the integration of diagnostic and clinical genetics, and clinical hepatology services as provided in Southampton. We urgently need reliable information about the clinical penetrance of the H63D mutation. There are a number of studies in progress which will provide information about this, including a study of 10 000 blood donors in South Wales funded by the Wales Office of Research and Development for Health and Social Care (MW).
Acknowledgments
We thank John Harvey of the Wessex Regional Genetics Laboratory, and Martin Howell (Department of Immunology) and Diana Eccles (Department of Clinical Genetics) of the Southampton University Hospital Trust.
References
Reply
Editor,—We appreciate the detailed response of Dr Fairbanks and colleagues. As we stated in our commentary, knowledge of hereditary haemochromatosis (HHC) and iron metabolism has advanced very rapidly in the short time since the discovery of theHFE gene. We agree that whereas the initial evidence suggested than any role of the second (H63D) mutation inHFE was minor and possibly confined to the compound heterozygous (C282Y/H63D) state,2-1-2-3 recent emerging evidence has indicated that H63D has a significant independent role in iron overload. Indeed, in the very short time since Fairbankset al’s letter was written additional clinical evidence confirming such an independent role has emerged2-4 2-5 and the molecular basis has been partly elucidated.2-6 In cultured cells, the C282Y mutation prevents the association of the mutant HFE protein with the transferrin receptor (TfR). In contrast, wild type HFE protein decreases the affinity of the TfR for transferrin and over expressed H63D protein lacks this effect.2-6 Thus, H63D has an independent, more subtle, effect on iron metabolism than C282Y, presumably involving the affinity of TfR for transferrin on the cell surface.
The key clinical question now relates to the degree of iron overload and severity of disease resulting from the H63D mutation. The emerging data would suggest that compound (C282Y/H63D) heterozygotes, while developing iron overload of a degree previously recognised as homozygous HLA associated HHC, as a group do not have as severe iron overload as C282Y homozygotes2-4 and that iron stores in H63D homozygotes are even less—that is, H63D has both low penetrance and comparatively low expressivity.