Cited 33 times since 2012 (2.7 per year) source: EuropePMC Circulation. Cardiovascular genetics, Volume 5, Issue 2, 28 4 2012, Pages 202-209 Literature-based genetic risk scores for coronary heart disease: the Cardiovascular Registry Maastricht (CAREMA) prospective cohort study. Vaarhorst AA, Lu Y, Heijmans BT, Dollé ME, Böhringer S, Putter H, Imholz S, Merry AH, van Greevenbroek MM, Jukema JW, Gorgels AP, van den Brandt PA, Müller M, Schouten LJ, Feskens EJ, Boer JM, Slagboom PE
Background
Genome-wide association studies (GWAS) have identified many single-nucleotide polymorphisms (SNPs) associated with coronary heart disease (CHD) or CHD risk factors (RF). Using a case-cohort study within the prospective Cardiovascular Registry Maastricht (CAREMA) cohort, we tested if genetic risk scores (GRS) based on GWAS-identified SNPs are associated with and predictive for future CHD.
Methods and results
Incident cases (n=742), that is, participants who developed CHD during a median follow-up of 12.1 years (range, 0.0-16.9 years), were compared with a randomly selected subcohort of 2221 participants selected from the total cohort (n=21 148). We genotyped 179 SNPs previously associated with CHD or CHD RF in GWAS as published up to May 2, 2011. The allele-count GRS, composed of all SNPs, the 153 RF SNPs, or the 29 CHD SNPs were not associated with CHD independent of CHD RF. The weighted 29 CHD SNP GRS, with weights obtained from GWAS for every SNP, were associated with CHD independent of CHD RF (hazard ratio, 1.12 per weighted risk allele; 95% confidence interval, 1.04-1.21) and improved risk reclassification with 2.8% (P=0.031). As an exploratory approach to achieve weighting, we performed least absolute shrinkage and selection operator (LASSO) regression analysis on all SNPs and the CHD SNPs. The CHD LASSO GRS performed equal to the weighted CHD GRS, whereas the Overall LASSO GRS performed slightly better than the weighted CHD GRS.
Conclusions
A GRS composed of CHD SNPs improves risk prediction when adjusted for the effect sizes of the SNPs. Alternatively LASSO regression analysis may be used to achieve weighting; however, validation in independent populations is required.
