The prevalence of CAD is higher in Scotland than in England for largely unexplained reasons [4, 30]. This observation was also evident in the UK Biobank participants studied here. The traditional risk factors included in the FRS hardly explained the difference in CAD prevalence between the two countries. Out of 163 genome-wide significant risk alleles studied, 35 had higher RAF in Scotland whereas 37 had higher RAF in England. However, overall these differences appeared to neutralize each other since there was no significant difference in the means and distributions of both weighted and unweighted GRS based on 163 CAD SNPs.
According to the ancestral-complex disease susceptibility model, genetic variations existed before the human spreading out of Africa and evolved with an extremely slow speed [31, 32]. However, nowadays environment and lifestyle are remarkably different from that of our ancestors. A mismatch between the ancestral variants and current environment might contribute to the development of some of non-communicable, complex diseases [2, 33].
It is unclear as to whether differences in ancestral variants contributing to CAD risk explain regional differences in CAD prevalence. With respect to England and Scotland we observed that about 40 % of genomewide significant variants displayed significant differences in allele frequencies. It is remarkable to find that many significant differences in allele frequencies of disease relevant genes in such closely related populations. However, the balanced effect – 35 variants had higher RAF in Scotland and 37 had higher RAF in England – suggests that this is not driven by any selection pressure on these risk alleles, which is in line with findings of Keyue and Iftikhar, who did not observe significant differences in the distribution of Fst values at 158 CVD-associated SNPs compared to background SNPs [34]. In fact, the net effects of these differences at multiple loci seem to neutralize each other, since we observed no differences in the CAD risk based on polygenic risk scores.
Thus, genetic susceptibility to CAD – based to common risk alleles – appears to be rather similar in England and Scotland. The same applies to traditional risk factors for CAD since the present as well as previous studies failed to demonstrate profound differences between these two countries [35, 36]. In 1989, Carstairs and Morris reported that Scotland suffers from more severe deprivation than England and Wales [37], In 2011 the same pattern of deprivation was still observed between the countries of Scotland and England [38]. In 2013, Newton et al. reported that significant health inequalities remain between the poorest and most deprived areas [39]. Thus, social deprivation might be one of the explanations for Scotland´s higher CAD rates. In order to lower CAD rates in Scotland, it seems to be reasonable to intensify preventive measures delivered at the most deprived.
A limitation of our study may be the fact that the lead SNPs we used to represent risk at a given genomewide significant locus might not be the causal ones. However, these variants were associated with the strongest risk such that the causal variants are likely to be in very high linkage disequilibrium. Moreover, the estimation of risk based on polygenic risk scores is unlikely to be affected by lack of knowledge on the causal variant. Another limitation of our study could be that we did not explore rare variants, gene-gene interactions, gene-environment, and exposure to epigenetic factors. All of these can modulate genetic risk [2, 40, 41] but are challenging to investigate in a study like ours. As for the traditional factors analysis, we only included the major risk factors for CAD (sex, age, BMI, HDL-C, TC, SBP, antihypertensive medication, smoking status and diabetes), while other important factors such as physical activity, family history and socioeconomic status are not included in the Framingham risk model [42]. Finally, the UKB population has been considered to represent a relatively low risk such that the data may not be representative for the entire population spectrum [43]. Nevertheless, the repeatedly observed differences in CAD prevalence between Scotland and England were apparent in UKB as well.