Combining Polygenic with Clinical Risk Scores in Atrial Fibrillation Risk Prediction: Implications for Population Screening.

Background: AF development is determined by clinical risk factors and genetic predisposition. Few studies have explored whether incorporating polygenic risk scores (PRS) improves clinical risk prediction beyond existing models.

Objective: We evaluated the interaction between AF-PRS and the HARMS2-AF and CHARGE-AF clinical risk scores on incident AF risk among the UK Biobank.

Methods: AF-PRS was examined in those with and without incident AF based on ICD-10 coding and divided into tertiles defined as low, intermediate and high risk categories. Regression analysis examined the impact of AF PRS combined with the HARMS2-AF and CHARGE-AF risk scores and AF risk.

Results: Among 285,734 participants with available whole genome sequencing data (52% female, age 57 years [50-63], 84.6% caucasian), AF incidence was 6.6% with a median time to AF 8.5 [5.0-11.2] over median 12.9 years follow up. High AF-PRS tertile was independently associated with incident AF risk, after adjustment for clinical risk factors (HR 2.75, 95% CI 2.62-2.89, p<0.001). AF-PRS enhanced AF risk prediction when combined with the HARMS2-AF risk model (AUC 0.828 improved to 0.839 with the addition of AF-PRS (DeLong p<0.001) with overall NRI 13.5% [12.8%-14.1%] and the CHARGE-AF risk model (AUC 0.808 improved to 0.828 with the addition of AF-PRS (DeLong p<0.001) with overall NRI 7.3% [6.7%-7.9%]).

Conclusions: Combining genetic and clinical risk using the HARMS2-AF and CHARGE-AF risk scores significantly improved AF risk prediction. Incorporating polygenic to clinical risk scores may enhance population screening and promote targeted interventions to reduce the incidence of AF.