From systemic adiposity to local epicardial adipose tissue targets: uncovering novel mechanisms and biomarkers in atrial fibrillation

Background Accumulating evidence highlights epicardial adipose tissue (EAT), the heart’s own visceral fat depot, as a key driver of atrial fibrillation (AF) pathogenesis. However, a comprehensive evaluation spanning population genetics, single-nucleus RNA sequencing (snRNA-seq), and circulating biomarkers is needed to establish causal pathways and identify actionable therapeutic targets. Methods We began with population-level observational analyses using The National Health and Nutrition Examination Survey (NHANES) data to examine the association between visceral adipose tissue (VAT) and AF risk, followed by two-sample Mendelian randomization (MR) using GWAS summary data to assess causality. Focusing on EAT as the visceral fat depot anatomically closest to the myocardium, we performed snRNA-seq on EAT specimens from 9 individuals. This cohort included 6 patients with nonvalvular AF and severe coronary artery disease (CAD), alongside 3 controls who had severe CAD but no history of AF, enabling characterization of cellular heterogeneity and transcriptomic alterations. Leveraging the single-cell data, we applied drug2cell to screen for adverse effects of anti-AF drugs across EAT cell types at single-cell resolution, while Summary-data-based MR (SMR) integrated EAT gene expression and GWAS data to identify novel therapeutic targets. Finally, using two cohorts (UK Biobank and our institutional plasma data), we found circulating predictor for future AF occurrence. Results Integration of large-scale NHANES clinical data with genome-wide MR demonstrates that increased VAT mass is causally associated with higher AF risk, independent of traditional risk factors and subcutaneous adipose tissue. Genetic analyses incorporating VAT-related GWAS from UK Biobank and AF data from FinnGen and the Million Veteran Program further corroborate this relationship. We present the inaugural single-nucleus transcriptomic profile of human EAT in AF, revealing a previously uncharacterized cellular landscape and providing a foundational resource for understanding local cardiac adipose mechanisms in arrhythmogenesis. Through multi-omics integration and computational drug-cell interaction mapping, we identify pathogenic neuronal clusters marked by impaired myelination signatures and nominate PTPRN2 as a potential target for pharmacological intervention. Large-scale prospective cohort analyses establish elevated circulating IL-17 as a predictor of future AF onset, supporting its utility as a clinically accessible biomarker for risk stratification. Conclusions Integrating the first snRNA-seq atlas of human EAT in AF with multi-omics and computational strategies, we also unveil a previously unrecognized cellular landscape that clarifies how local cardiac adipose tissue drives arrhythmogenesis. Additionally, plasma analyses indicate that elevated plasma interleukin-17 (IL-17) serves as a readily measurable biomarker for AF prediction.