METHODS

The subjects included were men aged 45–64 years in 1989 who had blood samples as part of the WOSCOPS protocol. Clinical data were available in 580 cases who subsequently had a coronary event (fatal or non-fatal myocardial infarction in 503, revascularisation procedure in 77), and 1160 matched controls who had not had a coronary event, over a mean follow up period of 4.9 years. Specimens for genetic analysis had been stored in a Biobank at −70°C. For the current study, specimens were missing or insufficient to allow both C282Y and H63D polymorphism analysis from 98 patients who had a coronary event (16.9%) and from 56 subjects who had no coronary event (4.8%). The reason for the higher frequency of missing specimens in the cases is not known, but was not regarded as a significant source of bias.

On specimens from the remaining 482 cases and 1104 controls, C282Y and H63D polymorphisms of the HFE gene were tested for by polymerase chain reaction and restriction enzyme digest⁶ followed by separation of the fragments on a polyacrylamide gel.

Approval for the project was granted by the Lanarkshire and North Glasgow Trust ethics and research committees.

RESULTS

The baseline characteristics of the cases and controls in the present study are shown in table 1⇓. Subjects who experienced a coronary event during the 4.9 years of trial follow up had a higher body mass index, blood pressure, LDL cholesterol, fibrinogen, white cell count, and C reactive protein, a lower HDL cholesterol, and were more likely to have hypertension.

In the 1104 control samples analysed, 922 (83.5%) did not carry the C282Y mutation, 179 (16.2%) were heterozygotes for C282Y, and three (0.3%) were C282Y homozygotes. Twenty five subjects (2.3%) were compound heterozygotes for C282Y/H63D. The frequency of C282Y homozygotes is less than would be expected given the frequency of the C282Y heterozygotes (χ², p  =  0.063). The likely explanation for this is that some C282Y homozygotes may have been excluded from entry into the WOSCOPS study because of abnormalities of liver function tests.

Table 2⇓ shows the frequency of HFE genotypes in cases and controls. The distributions were nearly identical. Table 3⇓ summarises hazard ratios and their 95% confidence intervals (CI) in prespecified subgroup analyses. Again there were no significant differences between cases and controls. Repeating the analyses after adjusting for possible confounding factors (body mass index, blood pressure, LDL and HDL cholesterol, white cell count, fibrinogen, and C reactive protein) gave an odds ratio of 0.87 (95% CI 0.63 to 1.19) for C282Y heterozygotes and 0.39 (95% CI 0.13 to 1.17) for C282Y/H63D compound heterozygotes. These were non-significant differences, although the trend is towards a protective effect of the C282Y mutation for coronary heart disease.

DISCUSSION

The C282Y mutation in the HFE gene was present in heterozygous state in 16.2% of control subjects in this sample of Scottish middle aged men with moderate hypercholesterolaemia. This is consistent with the high prevalence of the mutation previously noted in healthy northern European populations.

Our results, using a nested case–control design drawing samples from a cohort study, do not show evidence of a link between C282Y heterozygosity and coronary events. This finding is consistent with the reports of several point case–control studies which show no such association.^9–,¹⁵

Three published cohort studies have suggested an increased risk of coronary events in C282Y heterozygotes. The Finnish Kuopio ischaemic risk factor study (KIHD)¹⁶ reported on men aged 42–60 years with no coronary heart disease at baseline. Over a mean follow up of nine years there were 68 subjects with acute myocardial infarction, eight (11.8%) of whom were C282Y heterozygotes. Seventy seven (6.7%) of 1150 subjects who did not experience acute myocardial infarction were C282Y heterozygotes. The crude relative risk of acute myocardial infarction was 2.0 (95% CI 0.9 to 4.1) and the relative risk adjusted for other risk factors was 2.3 (95% CI 1.1 to 4.8).

In the second study,¹⁷ a cohort of 12239 Dutch women aged 51–69 years at baseline was followed for 16–18 years; C282Y status was assessed among 531 women who died of cardiovascular disease and 555 randomly selected women who did not die from this cause. The number of deaths from acute myocardial infarction is not stated, but a 1.5-fold increased risk (95% CI 0.9 to 2.5) was reported for C282Y heterozygotes.

The third study was the US ARIC (atherosclerosis risk in communities) study,¹⁸ which recruited 14 215 subjects aged 45–64 years at baseline (43% male, 57% female; 25% black, 75% white) and followed them up for a mean of 2.9 years. Over this period, there were 243 incident cases of coronary heart disease (157 probable or definite myocardial infarctions, 20 silent myocardial infarctions, 26 definite fatal coronary heart disease, 40 revascularisation procedures), and a reference cohort of 535 subjects was identified. The crude relative risk of coronary heart disease in C282Y heterozygotes was 1.6 (95% CI 0.9 to 3.0), and 2.7 (95% CI 1.2 to 6.0) after controlling for other risk factors such as lower prevalence of smoking and lower mean LDL cholesterol in the C282Y group.

The reason why three cohort studies have suggested that the C282Y mutation is a risk factor for coronary heart disease, while our study and several case–control studies have shown no evidence of a link, is not immediately clear. The results in the positive cohort studies were of marginal statistical significance, with confidence intervals crossing the line of unity only after adjustment for possible confounding factors. In the case–control studies, it is possible that survival bias contributed to the negative findings, but the three prospective studies did not specifically report a stronger impact of C282Y heterozygosity on fatal coronary heart disease. In our study there was a follow up of 4.9 years, and it might be argued that this was not long enough to allow iron to accumulate in sufficient amounts to have an adverse effect on free radical mediated oxidative stress. However, the age group we studied was similar to that of the other three prospective studies, and our population was exclusively male (men are known to have higher serum iron concentrations than women). It is possible that the hypercholesterolaemic nature of our study group resulted in a masking of a true association between C282Y and coronary heart disease. However, if the oxidative stress mechanism proposed for a true association were correct, one might expect a greater likelihood of finding a significant relative risk in subjects with high LDL cholesterol values.

We did not have access to serum samples to measure iron, transferrin, and ferritin concentrations in our subjects. In published work, two of eight prospective cohort studies have reported an association of serum ferritin with risk of coronary heart disease or atherosclerosis, and none of five cohort studies assessing transferrin saturation has reported an association with increased risk of coronary heart disease.^20,21

Taken as a whole, the evidence linking C282Y heterozygosity as a risk factor for coronary events is tenuous. Our study provides further evidence that no such link exists.