Identification of Novel Genes Associated with Atrial Fibrillation and Development of Atrial Fibrillation Predictive Models by Incorporating Polygenic Risk Scores and PheWAS-Derived Risk Factors

Abstract

AbstractBACKGROUND:Atrial fibrillation (AF) is the most common atrial arrhythmia and is subcategorized into numerous clinical phenotypes. Previous studies demonstrated that early-onset AF was associated with genetic loci among the certain populations.OBJECTIVES:The objective of this study was to develop AF predictive models using AF-associated single-nucleotide polymorphisms (SNPs) selected from the Genome-Wide Association Study (GWAS) of a large cohort of Taiwanese and explore whether the models posed the prediction power for AF.METHODS:75,121 total subjects including 5,694 AF patients and 69,427 normal controls with the GWAS data were included in this study. The polygenic risk scores based on AF-associated SNPs were determined and then integrated with Phenome-wide association study (PheWAS)-derived risk factors including clinical and demographic variables. The robust AF predictive models were developed through advanced statistical and machine learning techniques and then were evaluated in terms of discrimination, calibration, and clinical utility.RESULTS:The results demonstrated that the top 30 significant SNPs associated with AF were located on chromosomes 10 and 16, which involvedNEURL1,SH3PXD2A,INA,NT5C2,STN1, andZFHX3genes withINA,NT5C2, andSTN1being new discoveries in association with AF. The GWAS predictive power for AF had an area under the curve (AUC) of 0.626 (P< 0.001) and 0.851 (P< 0.001) before and after adjusting with age and gender, respectively. The results of PheWAS analysis showed that the top 10 diseases associated with discovered genes were all circulatory system diseases. The results of this study suggested that AF could be predicted by genetic information alone with moderate accuracy. The GWAS could be a robust and useful tool for detecting polygenic diseases by capturing the cumulative effects and genetic interactions of moderately associated but statistically significant SNPs.CONCLUSIONS:By integrating genetic and phenotypic data, the accuracy and clinical relevance of predictive models for AF were improved. The results of this study might improve AF risk classification, enable personalized interventions, and ultimately reduce the burden of AF-related morbidity and mortality.Abstract FigureGraphic Abstract