(P135) PROTEOGENOMIC PREDICTION OF ATRIAL FIBRILLATION IN NON-ISCHEMIC CARDIOMYOPATHY

Background: The development of atrial fibrillation (AF) in patients with non-ischemic cardiomyopathy (NICM), a precursor for heart failure with prominent genetic underpinnings, is associated with poor prognosis. Tools to risk stratify NICM patients for AF remain lacking. With the rise of scientific advancements enabling personalized risk stratification, such as polygenic risk scores and serum proteomics, we sought to employ a proteogenomics approach to improve prediction of AF in patients with NICM.

METHODS AND RESULTS: Using the UK Biobank, we constructed an NICM cohort, defined by presence of left ventricular failure or dilated cardiomyopathy and absence of ischemic etiology or hypertrophic cardiomyopathy, and excluded AF cases which developed before NICM. The final cohort comprised 2,661 patients, with a mean (SD) age of NICM diagnosis of 66 (9.8), 55.4% male sex, and 84.4% European genetic ancestry. The prevalence of AF after NICM diagnosis was 23.9% (636 cases), and those who developed AF had greater all-cause mortality (Hazard Ratio HR 2.69, 95% CI 2.24,3.23), ischemic stroke (HR 1.87, 95% CI 1.06,3.29), heart-failure related outcomes (HR 2.47, 95% CI 1.80,3.38), and the composite of lethal arrhythmias, sudden cardiac arrest, and ICD implantation (HR 2.94, 95% CI 2.16,4.00) over 20 years of follow-up. Application of a polygenic risk score for AF (PRS-AF) in Europeans (limited due to few cases in non-Europeans) was found to robustly predict AF (HR per 1-standard deviation (SD) 1.30, 95% CI 1.18,1.42), with NICM patients in the top 10th percentile of PRS-AF risk being 1.91 times more likely to develop AF compared to intermediate (10-89th percentile) risk participants (HR 1.91, 95% CI 1.42,2.60) over 20 years of follow-up. Importantly, a high PRS-AF predicted risk across all clinical risk categories, defined using the CHARGE-AF score, and improved predictive performance (net reclassification index 0.246, 95% CI 0.107,0.470; P< 0.001) compared to risk stratification using only the clinical CHARGE-AF risk score + sex. In 329 patients who also had serum protein measurements, we found high levels (top tertile) of NT-proBNP to predict AF with (HR 2.33, 95% CI 1.33,4.10) or without (HR 2.23, 95% CI 0.77,2.60) adjustment for traditional risk factors. PRS-AF remained predictive (HR per 1-SD 1.34, 95% CI 1.09,1.70) after additional adjustment for NT-proBNP levels, suggesting that the integration of protein biomarkers and genomics into risk stratification can improve prediction of AF.

Conclusion: Accounting for the proteogenomic profiles of patients with NICM can improve prediction of AF, enabling personalized management and treatment in this population.