These two large cohort studies have confirmed the previous finding of genotypes and haplotypes in F7
having significant effects upon FVII levels. In accordance with other studies [17
] one FVII raising haplotype, and one FVII lowering haplotype was identified. The effects identified in baseline samples were confirmed in years 1–5 for FVIIc levels in NPHS-II. The FVII lowering haplotype, H5, was associated with a small reduction in FVIIc/FVIIag ratio, which perhaps suggests that H5 gives rise to a protein with reduced function
. The variant within the F7
promoter, I/D323, within H5, has been shown to be functional and to affect levels of transcription [19
] but would not be expected to be associated with reduced activity of FVII protein. The amino acid substitution (R353Q) within H5 is a good contender for a reduced functional activity of the FVII protein.
When each of the F7 haplotypes were compared to the most frequent haplotype, H1, which contained the common allele at all sites, no significant associations with CHD could be identified, although there was a tendency for both H2 and H5 to exhibit a higher risk of CHD. In NPHS-II when carriers of H2 were compared to non-carriers (ie all other haplotypes), a similar higher CHD-risk was identified to that seen with individual genotypes within each haplotype (>25% increase in risk was identified) and these effects reached statistical significance. A similar trend for CHD-risk was also observed for H2 in the four-SNP model (data not shown). The CHD-risk association for H2 in NPHS-II was, however, completely reversed in WH-II, such that a reduced CHD-risk was associated with H2. It is possible that because DNA was not available in WH-II until year 17 of the study that there had been a survival bias against the H2 allele in WH-II. However, while there was a statistically significant difference between the allele frequencies for H2 in WH-II and NPHS-II, the frequency of H2 in WH-II was actually higher than that observed for NPHS-II, suggesting that more, rather than less of those with the H2 allele had survived in WH-II. It is possible that differences in CHD-risk between the studies were due to differences in demographics, WH-II having a more favorable risk profile with lower age, less smoking, lower BMI, lower blood pressure, (Supplementary Table S1
). These finding require testing in further large prospective studies of heart disease.
It should be noted also that increased CHD-risk for H2 in NPHS-II remained in a model in which FVII levels were also included, suggesting that the risk shown in this study was not acting solely via the increased FVII levels. It is possible that the F7 allele is in LD with a risk allele outside the F7 gene, and further analysis to determine LD between the F7 gene and other SNPs in the surrounding region would be useful.
For H5 carriers, compared to non-carriers, a >25% increase in CHD-risk was identified in fully adjusted models, but the CHD-risk was no longer significant when FVIIc levels were included in the model, suggesting that risk is acting through the lower FVIIc levels identified for this haplotype. The findings of CHD-risk association were consistent in NPHS-II and WH-II, and a pooled estimate across the two studies was statistically significant. Interestingly, activated markers of coagulation downstream from FVII in NPHS-II were also significantly lower for this allele and the question arises as to what the pathological mechanism is for this lower, functionally less active FVII and lower levels of CAMs downstream from FVIIa. The findings point to reduced activation of the extrinsic coagulation pathway, resulting from a circulating FVII species with reduced coagulant activity. This observation is, however, counter-intuitive for an allele associated with increased risk of CHD. The processes of atherogenesis are complex, involving both plasma and vessel wall factors including proteins implicated in inflammation, dyslipidaemia, and thrombosis. However intra-plaque bleeding is also a recognized contributor to atherogenesis and it would be interesting to know, in future studies, whether this process is increased for the H5 haplotype. Interestingly there are reports of FVII deficiency associated with thrombotic events [51
]. While the mechanism for FVII deficiency in thrombotic conditions has not been elucidated, those reported to date are functional deficiencies (normal or near normal antigen levels, with reduced FVIIc levels) and associated with amino acid substitutions. In the study of Marty et al. [52
], 6 out of the 14 cases studied had the 353Q variant, associated with H5 in the current study. Reduced binding to tissue factor has been suggested as a contributory pathological mechanism in these cases [26
]. However, it should be stated also that the observation may also be by chance and needs confirmation in further studies.
The combining of SNPs to form haplotypes and comparison of specific haplotypes to non-carriers of that haplotype, therefore, provided more statistically significant CHD-risk results than risk analysis for individual SNPs, or for assessment of risk in each haplotype, compared to the common haplotype. Reduced ability to detect CHD-risk when each haplotype was compared to the common haplotype is likely to be due to the lower numbers within the comparison group.
While the associations between F7 haplotypes and FVIIc levels are consistent across the studies and confirm previous reports, we have only studied UK Caucasian men, and effect sizes and associations may be different in women and in other ethnic groups. Since FVIIc was measured using the same method in both studies, the lower levels seen in the in WH-II men are likely to be due mainly to their younger age and lower triglyceride levels, although other unmeasured dietary or lifestyle factors may contribute. There are also differences in the definition of the CHD endpoint used in the two studies, with WH-II but not NPHS-II including definite angina, but it seems unlikely that the lower CHD risk associated with haplotype 5 can be confounded by this, and overall the prevalence of CHD was similar in both studies.
In summary we have identified a FVII raising haplotype (H2) that had significant, but opposite effects upon CHD-risk in two prospective studies, and a FVII lowering haplotype (H5), particularly associated with lower coagulant activity, that was associated with increased CHD-risk in two prospective studies. The elimination of significant CHD-risk associated with H5 when FVII levels were included in the model suggests that CHD-risk is acting via the lower (particularly coagulant) circulating FVII. Baseline FVII levels have been previously shown in cohort studies to be lower in those who developed a CHD-event [5
], and in NPHS-II also, low FVIIa levels were previously shown to be associated with CHD-risk [50
]. In combination, these findings suggest a non-linear association of FVII with CHD-risk with genotypes associated with both lower and potentially with higher levels of FVII (or FVIIa) being implicated in CHD.