Abstract 15071: Sensitivity to Significance Threshold in ’Real World’ Application of a Polygenic Risk Score for Resting QT Interval
Intro: Polygenic risk scores (PGS) provide the opportunity for direct application of genome wide association studies (GWAS) to clinical populations. While conventionally, genome-wide significance (i.e., p<5x10-8) has been used as the level to determine variants for use in PGS, it is unknown if this level is optimal for clinical application. In this study, we examined PRS for resting QT interval in two ‘real world’ cohorts, with use of several significance cut-points from the prior GWAS.
Methods: 2915 participants of European and 366 of African ancestry in the MGH Cardiology and Metabolic Patient (CAMP) study had genotypes imputed to 1000Genomes. For validation, an additional 820 European and 57 African participants of the Partners Biobank had genotypes imputed. To create the PRS, effect estimates for genetic variants associated with QT interval were obtained from the QT-IGC GWAS of ~70k individuals (Europeans), and from the CARe-COGENT GWAS (Africans). For each individual, we calculated PGS based on the negative log p value cut-point of 2 through 12 (i.e., 10-12) from the original GWAS, after pruning for linkage disequilibrium. Using multivariable linear regression, adjusted for age, sex, height, BMI, medication use, heart failure, and hypertension, we examined the association with heart rate-corrected QT interval.
Results: In Europeans from both cohorts, the PGS significantly increased the variation in QTc explained for almost all of the chosen cut-points (p<0.001), with a trend toward increased variance explained at lower p value cut-points (Figure). In Africans, this trend was not apparent in either cohort, which may have been due to lower enrollment numbers, or fewer significant variants.
Conclusion: For individuals of European descent, PGS provided a significant increase in variance in QT interval explained at nearly every p value cut-point; however, we did not detect any apparent benefit gained by including a greater number of variants of lower association.
Author Disclosures: M.A. Rosenberg: None. S.A. Lubitz: None. H. Lin: None. G. Kosova: None. P. Huang: None. R. Perlis: None. C. Newton-Cheh: None.
- © 2016 by American Heart Association, Inc.