Assessment of the clinical utility of adding common single nucleotide polymorphism genetic scores to classical risk factor algorithms in coronary heart disease risk prediction in UK men.

Background: Risk prediction algorithms for coronary heart disease (CHD) are recommended for clinical use. However, their predictive ability remains modest and the inclusion of genetic risk may improve their performance. Methods: QRISK2 was used to assess CHD risk using conventional risk factors (CRFs). The performance of a 19 single nucleotide polymorphism (SNP) gene score (GS) for CHD including variants identified by genome-wide association study and candidate gene studies (weighted using the results from the CARDIoGRAMplusC4D meta-analysis) was assessed using the second Northwick Park Heart Study (NPHSII) of 2775 healthy UK men (284 cases). To improve the GS, five SNPs with weak evidence of an association with CHD were removed and replaced with seven robustly associated SNPs - giving a 21-SNP GS. Results: The weighted 19 SNP GS was associated with lipid traits (p < 0.05) and CHD after adjustment for CRFs, (OR = 1.31 per standard deviation, p = 0.03). Addition of the 19 SNP GS to QRISK2 showed improved discrimination (area under the receiver operator characteristic curve 0.68 vs. 0.70 p = 0.02), a positive net reclassification index (0.07, p = 0.04) compared to QRISK2 alone and maintained good calibration (p = 0.17). The 21-SNP GS was also associated with CHD after adjustment for CRFs (OR = 1.39 per standard deviation, 1.42 x 10(-3)), but the combined QRISK2 plus GS score was poorly calibrated (p = 0.03) and showed no improvement in discrimination (p = 0.55) or reclassification (p = 0.10) compared to QRISK2 alone. Conclusions: The 19-SNP GS is robustly associated with CHD and showed potential clinical utility in the UK population.