Epicardial adipose tissue thickness on transthoracic echocardiography predicts 2-year incident atrial fibrillation in elderly hypertensive patients.

Background: Epicardial adipose tissue (EAT) promotes atrial remodeling, yet prospective data on whether a single transthoracic-echocardiographic measurement of EAT can identify elderly hypertensive patients at short-term risk of atrial fibrillation (AF) are limited.

Methods: In this single-center cohort study (March 2021-June 2024), 460 hypertensive adults aged ≥65 years in sinus rhythm were enrolled; epicardial adipose tissue thickness was measured on the right-ventricular free wall, and participants underwent intensive multimodal rhythm surveillance for 24 months. Cox models were adjusted for age, body mass index, systolic blood pressure, diabetes, left-atrial (LA) volume index, and β-blocker use; performance was optimism-corrected with 200 bootstraps.

Results: During 902 person-years of follow-up, 55 participants (12.0%; 6.1 events per 100 person-years) developed incident AF. Baseline EAT was greater in cases than in controls (7.9 ± 1.4 vs. 5.7 ± 1.2 mm; p < 0.001). Each 1 mm increase in EAT independently conferred a 62% higher AF hazard [hazard ratio (HR): 1.62, 95% CI: 1.29-2.04]; the optimism-corrected HR was 1.56. The findings were consistent in those with treated obstructive sleep apnea (OSA) (HR: 1.60) and in those without OSA (HR: 1.59; interaction p = 0.93) and after additional adjustment for high-sensitivity C-reactive protein (HR: 1.55 in 410 participants with biomarker data). Adding continuous EAT to a clinical model improved the C-index from 0.74 to 0.79 (optimism-corrected 0.78), reduced the Akaike information criterion by 16 points, and yielded a continuous net reclassification improvement of 0.25 (95% CI: 0.09-0.39) and an integrated discrimination improvement gain of 0.05. Time-specific area under the receiver-operating-characteristic curves (AUCs) remained ≥0.76 and calibration was preserved (Grønnesby-Borgan p ≥ 0.60). A receiver-operating-characteristic analysis identified 6.5 mm as the optimal EAT threshold (80% sensitivity, 68% specificity); 24-month AF incidence rate was 24.7% above vs. 4.1% below this cut point (log-rank p < 0.001). The EAT-AF association was robust in Fine-Gray competing-risk models and consistent across sex, obesity, diabetes, and LA-size strata (all interaction p > 0.20).

Conclusions: Echocardiographic EAT thickness is a reproducible and incrementally informative predictor of 2-year incident AF in elderly hypertensive patients. Incorporating this simple metric into routine scans could refine risk stratification and guide targeted rhythm surveillance.