- Top of page
- Subjects and methods
People with genetic haemochromatosis (GH) accumulate iron from excessive dietary absorption. In populations of northern European origin, over 90% of patients are homozygous for the C282Y mutation of the HFE gene. While about 1 in 200 people in the general population have this genotype the proportion who develop clinical haemochromatosis is not known. The influence of HFE genotype on iron status was investigated in 10 556 blood donors. The allele frequencies of the C282Y and H63D mutations were 8·23% and 15·3% respectively. Heterozygosity for C282Y occurred in 1 in 7·9 donors, for H63D in 1 in 4·2 donors, and 1 in 42 were compound heterozygotes. Homozygosity for H63D occurred in 1 in 42 donors and 1 in 147 (72) were homozygous for C282Y. Mean values increased for transferrin saturation (TS) and serum ferritin (sFn), and decreased for unsaturated iron binding capacity (UIBC) in the order: donors lacking the mutations, H63D heterozygotes, C282Y heterozygotes, H63D homozygotes, compound heterozygotes and C282Y homozygotes, but serum ferritin (sFn) concentrations were no higher in H63D heterozygotes and C282Y heterozygous women than in donors lacking mutations. The percentage of donors failing the screening test for anaemia or of those with sFn < 15 µg/l did not differ among the genotype groups. C282Y and H63D heterozygotes and donors homozygous for H63D were at no greater risk of iron accumulation than donors lacking mutations, of whom 1 in 1200 had both a raised TS and sFn. The risk was higher for compound heterozygotes (1 in 80, P = 0·003) and for C282Y homozygotes (1 in 5, P < 0·0001). There was no correlation between sFn and either age or donation frequency in C282Y homozygotes. None of the 63 C282Y homozygous donors interviewed showed physical signs of overload or were aware of relatives with haemochromatosis. The Welsh Blood Service collects blood from about 140 000 people each year including an estimated 950 who are homozygous for HFE C282Y. They are probably healthy and unaware of any family history of iron overload.
Genetic haemochromatosis (GH) is an autosomal recessive disorder in which those affected accumulate iron as a result of increased absorption from food (Bothwell & MacPhail, 1998). GH, defined in terms of iron accumulation, is common in countries in which the population is largely of northern European origin (Worwood, 1994a). If patients are diagnosed in the precirrhotic stage and treated by phlebotomy to remove the excess iron, life expectancy is normal (Niederau et al, 1996). However, once cirrhosis has developed patients have a shortened life expectancy and a high risk of liver cancers, even when iron depletion has been achieved.
In 1996, the HFE gene was located on the short arm of chromosome 6 about 5 Mb telomeric to the human leucocyte antigen (HLA) class I region (Feder et al, 1996). It codes for a HLA class I-like protein and over 90% of chromosomes from patients with GH carry a single mutation, C282Y, on this gene (Feder et al, 1996). There is convincing functional evidence for the involvement of the HFE protein in iron transport across cell membranes, as the HFE protein forms a complex with the transferrin receptor and decreases the affinity of the receptor for transferrin (Feder et al, 1998; Lebron et al, 1998). HFE knockout mice develop a similar form of iron overload to humans (Zhou et al, 1998). The C282Y mutation abolishes a disulphide bridge, which prevents binding to β2 microglobulin and cell-surface expression (Feder et al, 1997). A second mutation, H63D, is common in the general population, but may cause iron accumulation if present with the C282Y mutation (Feder et al, 1996). In Europe, Australia and the USA, from 60% to 100% of patients with GH are homozygous for the C282Y mutation of the HFE gene (Worwood, 1999). In the general population, the C282Y mutation frequency varies from 0·5% in Italy to over 9% in Ireland and parts of Brittany. The mutation is not found in non-European populations. The H63D mutation is more widespread and in Europe the frequency is 12–15% (Merryweather-Clarke et al, 2000). In the UK, over 90% of patients with GH are homozygous for the C282Y mutation of the HFE gene and another 4% are compound heterozygotes (The UK Haemochromatosis Consortium & Worwood, 1997). Other forms of non-HFE-related haemochromatosis include those patients with clinical haemochromatosis who lack detectable mutations in the HFE gene and any HLA haplotype association (Pietrangelo et al, 1999; Roetto et al, 2001).
On first investigation, most homozygous relatives of a haemochromatosis patient show evidence of iron accumulation (a raised transferrin saturation; TS) and about 50% of men and a lower proportion of women display at least one of the clinical manifestations of haemochromatosis: hepatomegaly, abdominal pain, skin pigmentation, weight loss, fatigue, arthropathy, hypogonadism, impotence, liver disease and cirrhosis (Bradley et al, 1996).
There are now several reports about clinical penetrance in subjects homozygous for C282Y detected by genetic testing. Olynyk et al (1999) found similar figures for penetrance in a population-based study of 3011 unrelated, white adults in Busselton, Australia, where homozygotes for haemochromatosis were detected by genetic testing. However, other population-based studies of both blood donors (Adams et al, 2000) and non-blood donors (Beutler et al, 2000) have revealed a lower clinical penetrance for homozygosity for C282Y.
Early population studies suggested an incidence of 1 in 5000 to 1 in 10 000 for clinical haemochromatosis (Sheldon, 1935; Finch & Finch, 1955). In the island of Jersey, where the C282Y allele frequency is 8·4% (Merryweather-Clarke et al, 1998), about 1 in 140 people would be homozygous for C282Y yet the population prevalence for clinical haemochromatosis was 1 in 4700. Haemochromatosis or iron overload was mentioned in about 1 in 6000 death certificates in the USA from 1979 to 1992 (Yang et al, 1998). The frequency of heterozygosity and homozygosity for C282Y was not significantly increased in patients with type II diabetes (Braun et al, 1998; Frayling et al, 1998; Florkowski et al, 1999) and liver disease (Bhavnani et al, 2000; Willis et al, 2000). It is therefore already apparent that the majority of C282Y homozygotes are not being detected clinically.
Motulsky (1979) suggested that the haemochromatosis gene may have spread because of a selective advantage – protection of heterozygotes against iron deficiency. About 25% of heterozygotes, identified by HLA typing of family members of haemochromatosis patients, have either a raised TS or a raised serum ferritin (sFn) (Powell et al, 1990; Adams, 1994; Bulaj et al, 1996). Slightly but significantly higher values for serum iron (sFe) and TS have been found in heterozygotes for either C282Y (Burt et al, 1998; Datz et al, 1998; Beutler et al, 2000) or H63D (Burt et al, 1998; Beutler et al, 2000) compared with subjects lacking these mutations. The differences in ferritin levels were smaller and not significant. In compound heterozygotes, there are greater differences (Beutler et al, 2000; Rossi et al, 2000). In heterozygotes for C282Y (Datz et al, 1998; Beutler et al, 2000) and H63D (Beutler et al, 2000), Hb levels were slightly higher than in subjects lacking mutations. Beutler et al (2000) have noted a lower prevalence of anaemia among women carrying either mutation, but the differences are small and only significant if all subjects carrying mutations were compared with those lacking mutations. Such changes may indicate an increase in iron absorption that might permit C282Y heterozygotes to donate blood more frequently than donors without the mutation.
We have studied the effect of HFE mutations on iron status by examining HFE genotypes and measures of iron status in blood donors. Blood donors are relatively young and fit and this largely removes the influence of disease on measures of iron status. By studying large numbers, the dietary influences on genotype are equalized. Blood donors also have greater iron losses as a result of donation of blood. However, this additional iron loss can be quantified along with the number of times that a donor has not been permitted to give blood because of possible anaemia.
- Top of page
- Subjects and methods
About 1 in 7 donors carried the C282Y mutation and 1 in 4 the H63D mutation. These frequencies are among the highest in Europe. While donors were not asked about ethnic origin, the ethnic communities are known to be under-represented. Within the area included in the survey, about 3·0% belong to ethnic minorities and, of these, nearly 67% are Asian, Black or of Mediterranean origin (Hogg & Modell, 1998; personal communication Dr A. May). The C282Y mutation of the HFE gene is rarely found in non-European populations and is uncommon in Southern Italy and Greece. The H63D allele is found in Mediterranean populations at a similar frequency to that among northern Europeans, but elsewhere its frequency is lower although variable (Merryweather-Clarke et al, 2000). As less than 3% of blood donors were from the ethnic minorities, the frequencies given above relate to the population of Northern European origin.
Serum iron and ferritin concentrations are influenced by age, sex, diet and disease, as well as by biological variation within an individual and by methodological variation (Worwood, 1994b). In the case of blood donors, the frequency of blood donation also needs to be considered (Milman & Sondergaard, 1984). Although there is diurnal variation of serum iron concentration and blood was taken between 10·00 and 18·00 h, samples were assayed in order of collection and not according to sex or genotype, so methodological and biological variation should be similar for all genotypes.
There were no differences in age or donation frequency between the genotype groups except among the male, C282Y homozygotes who had given significantly more blood than other donors. Generally sFe, TS and sFn increased, and UIBC and TIBC decreased, in the order: wild type, H63D heterozygotes, C282Y heterozygotes, H63D homozygotes, compound heterozygotes and C282Y homozygotes. Despite the statistical significance of these differences in iron status, they are small and only the mean UIBC and TS in the male C282Y homozygotes lie outside the normal range established for first-time blood donors.
For female donors, sFn did not differ between wild type, C282Y or H63D heterozygotes and donors homozygous for H63D. Although this may indicate that there is little change in total body iron stores, it may also reflect a failure of sFn to respond to an increase in iron concentration in hepatic parenchymal cells in the early stages of iron accumulation, when iron concentrations in reticuloendothelial cells are not increased (Worwood, 1990). Edwards et al (2000) summarized reports of liver iron concentration in subjects heterozygous for GH. Despite increases in liver iron concentrations to about the upper limit of normal, serum ferritin levels remained within the normal range.
Storage iron levels are influenced more by the frequency of donation rather than by the total number of units given (Milman & Søndergaard, 1984) and, in both men and women, sFn decreased with the number of units of blood given in the 3 years prior to testing. Storage iron levels were calculated by assuming that 1 µg of sFn is equivalent to 8 mg of storage iron in the body (Walters et al, 1973). Median iron stores for all donors were 0·65 g for men and 0·35 g for women. Among men with only wild-type alleles, giving 3 units of blood, equivalent to 720 mg of Fe, in the previous 3 years caused a reduction in the level of storage iron of about 200 mg compared with those who had not donated during the period. This is a result of the increased iron absorption that is stimulated by blood donation (Garry et al, 1995). The effect of blood donation on sFn concentration was similar in donors lacking HFE mutations and heterozygous donors.
If heterozygosity for C282Y provides young women with protection against iron deficiency anaemia then it was not apparent in our study. Serum ferritin concentrations did not vary between heterozygotes and wild-type subjects and the proportion of donors with sFn concentrations < 15 µg/l did not differ. The proportion of donors failing the copper sulphate screening test for anaemia was not decreased by heterozygosity for C282Y. Thus, there was little indication that heterozygotes for C282Y will be able to give blood more frequently than donors not carrying this mutation. However, this finding is not in agreement with Beutler et al (2000), who noted a small but significant reduction in the prevalence of anaemia among subjects carrying HFE mutations compared with those lacking mutations.
If storage iron levels are not increased in C282Y heterozygotes, any association between heterozygosity for HFE C282Y and disease (Dorak et al, 1999; Maclean et al, 1999; Roest et al, 1999; Tuomainen et al, 1999) may be related to the subtle changes in plasma iron and iron binding capacity. These changes may also explain the strong association between both heterozygosity and homozygosity for H63D in Italian patients with sporadic porphyria cutanea tarda (Sampietro et al, 1998) in which the C282Y mutation is rare. They may also be related to the increase in non-transferrin-bound iron recently reported in subjects heterozygous for C282Y (de Valk et al, 2000).
Approximately 1 in 1200 (90% CI 1 in 690 to 1 in 4545) donors lacking mutations and H63D heterozygotes had both a raised TS and sFn (based on the thresholds established for first-time donors lacking mutations). For C282Y heterozygotes, 1 in 437 (90% CI 1 in 225 to 1 in 7692) had raised levels. For donors homozygous for H63D, 1 in 200 (1 in 90 to > 1 in 10 000) had raised values for TS and sFn. These donors are not at significant risk of iron overload compared with donors lacking mutations (Fisher's exact test, P > 0·05). For compound heterozygotes, about 1 in 80 (90% CI 1 in 42 to 1429) had raised levels (P = 0·003). For C282Y homozygosity, about 1 in 4 male donors and 1 in 5 female donors had raised levels (1 in 4·6 90% CI 3·3 to 7·4 P < 0·0001). It should also be noted that the threshold for sFn was not the usual upper limit of normal taken for diagnostic purposes and even fewer donors exceeded 300 µg/l (male) and 200 µg/l (female). Heterozygotes and donors homozygous for H63D appear to be at little risk from iron overload and, for compound heterozygotes, the risk is an order of magnitude less than for C282Y homozygotes.
The UIBC has been proposed as a cheap and rapid screening test for iron overload (Pintar et al, 1981; Adams et al, 2000). In this survey the UIBC appears to have a similar sensitivity and specificity to TS.
The mean age of the C282Y homozygous men was 38 years, reflecting that for all donors, only 8 years younger than the mean age of 46 years at diagnosis for 251 patients (90% men) reported by (Niederau et al, 1996). In a group of 200 patients (10% of whom were women), responding to a questionnaire, the mean age of onset of symptoms was only 41 years, even though a diagnosis of GH was not usually made for several more years (McDonnell et al, 1999). Bassett et al (1984) studied 34 relatives of patients with GH who shared the same HLA haplotypes. All were under 35 years of age and were classified by HLA typing as homozygous for haemochromatosis. For men and women the median TS was 87% and 67%, respectively, and was 450 µg/l and 320 µg/l for sFn. A median of 7 g of iron was removed by phlebotomy. Initial sFn correlated with iron stores determined by phlebotomy (r = 0·92) with the same relationship between storage iron (determined by quantitative phlebotomy) and sFn as that found in normal subjects by Walters et al (1973). At the time of diagnosis, the patients studied by Niederau et al (1996) had accumulated 21·2 ± 1·1 g (SEM) of iron. If storage iron is estimated by assuming that 1 µg of serum ferritin is equivalent to 8 mg of storage iron (Walters et al, 1973), the median level of storage iron in our study was 1·23 g for male blood donors homozygous for C282Y and 0·52 g for female donors. Median iron stores, calculated in the same way for all donors, were 0·65 and 0·35 g respectively. There was thus an order of magnitude difference in the level of iron stores between blood donors homozygous for C282Y and patients presenting with clinical symptoms of GH or homozygotes found by screening the families of relatives. This is unlikely to be solely as a result of the age difference.
Are the low iron stores a result of blood donation? While regular blood donation reduces the level of storage iron, there is partial compensation because of the increase in iron absorption (Garry et al, 1995). Among men lacking HFE mutations, donating 3 units of blood in 3 years caused a mean reduction of 200 mg in iron stores, although donation losses were about 700 mg. However, among those homozygous for C282Y, there was no relationship between donation frequency and sFn. An enhanced rate of iron absorption in C282Y homozygotes may ensure that iron losses as a result of donation are more fully compensated (Lynch et al, 1989), thus explaining this lack of relationship. An intriguing finding of ours is that male donors homozygous for C282Y had donated significantly more blood in the 3 years before testing than all donors (3·7 units and 2·8 units). While this observation is of interest, particularly because there may be an underlying biological explanation, it should be noted that Barton et al (1999) found that voluntary blood donation before diagnosis did not decrease the severity of iron overload.
Subjects homozygous for haemochromatosis might be excluded as blood donors either because they do not feel well or because they are receiving medical treatment. There is no evidence to support exclusion of C282Y homozygotes. Based on overall allele frequencies for men and women, we expected to find 34 men and 42 women homozygous for C282Y. There were 30 men and 42 women discovered by genetic testing. Furthermore, the age distribution was the same as that for all donors.
The C282Y genotype is present in 90% of patients with haemochromatosis and is strongly associated with iron accumulation in blood donors. Among first-degree relatives of patients with clinical haemochromatosis, approximately 50% of subjects homozygous for HFE C282Y eventually develop clinical complications of haemochromatosis (Bradley et al, 1996). A similar penetrance was reported by Olynyk et al (1999) among randomly selected subjects discovered by genetic testing of the adult population of Busselton, Australia. In other surveys, however, in which subjects have been identified as homozygous for C282Y by genetic testing, the penetrance has been lower (Adams et al, 2000; Beutler et al, 2000). Within the region served by the Welsh Blood Service, about 140 000 people give blood each year, of whom about 950 will be homozygous for C282Y. Our data suggest that the great majority of these are probably healthy subjects unaware of any family history of iron overload. Locally, the clinical burden appears to be low. Cardiff and the Vale of Glamorgan has a population of 434 000 and is part of the region covered by our survey. The data from the current survey reveals that 1 in 150 people are homozygous for C282Y. If all men over 50 years of age have clinical manifestations of iron overload, there should be 480 men with GH. We used the ICD10 classification (haemochromatosis E83·1) to identify patients from hospital records and validated the list of patients from local knowledge and laboratory records. There were 31 patients (23 men) resident in the area and being treated in the two district general hospitals for the years 1998–2000 (unpublished observations).
Prior to the discovery of the HFE gene, Edwards et al (1988) surveyed 11 065 blood donors from Utah, USA. The Utah donors were of similar mean age to the donors from South Wales and they had similar mean values for non-fasting TS (6·2% of Utah donors with TS > 50% compared with 4·2% for South Wales). Within this Utah population of donors, 18 men and eight women were classified as homozygous for GH on the basis of a fasting TS (> 62%), liver biopsy and family studies. As women were clearly under-represented, the male data were used to calculate the frequency of the GH genotype to be 0·0045 (1 in 222). Unfortunately, data for the frequency of HFE mutations are not yet available for the Utah population. In South Wales, the frequency of homozygosity for HFE C282Y was 0·0068. At the time of testing, 25/29 men and 18/41 women had a TS > 50%. This confirms that most men homozygous for C282Y would have been identified in the Utah study and that less than 50% of the women would have been detected. Measuring TS would fail to detect 20% of young men and over 50% of women. In female, first-time donors homozygous for C282Y we found that TS decreased with age, and in people presenting at a health appraisal clinic, average age 58 years, Beutler et al (2000) found that only a minority of those homozygous for C282Y had a raised TS. This suggests that measuring TS may be less effective for detecting GH in older people than in people aged 30–50 years, but numbers in both studies were small.
While an alternative detection strategy for haemochromatosis is population screening by genetic testing, it is apparent from our data and that of others that this will identify large numbers of healthy people who are at risk from iron accumulation, yet who may never suffer from the consequences of iron overload. All those detected in this way will need to be counselled and offered a family investigation. Those not requiring immediate treatment require regular testing to detect the onset of iron overload. There is an urgent need to identify the cofactors that convert a relatively benign process of iron accumulation into iron overload, tissue damage and disease in a proportion of those homozygous for HFE C282Y. Until these additional risk factors are identified, it is likely that the costs to the individual and community will outweigh the benefits. Population screening for haemochromatosis has been strongly advocated, as the condition is common and preventable by regular blood donation (Allen & Williamson, 2000). However, our study lends support to those advocating a more cautious approach (Haddow & Bradley, 1999).