Epicardial adipose tissue native T1 and LVEF improvement across BMI categories in HFrEF: an exploratory CMR study

Objective To explore whether the association between regional epicardial adipose tissue (EAT) native T1 and improvement in left ventricular ejection fraction (LVEF) differs according to body mass index (BMI) category in patients with heart failure with reduced ejection fraction (HFrEF). Methods This single-center retrospective exploratory study included 119 patients with clinically diagnosed HFrEF and a baseline cardiovascular magnetic resonance (CMR) LVEF < 40%. Improvement was defined as ≥10 percentage points higher follow-up than baseline echocardiographic LVEF, using the last eligible examination at 3–18 months. BMI was categorized using a cutoff of 24 kg / m 2 . Four BMI–EAT native T1 categories based on the cohort median T1 were used only descriptively. The primary logistic model included a BMI category × standardized continuous EAT native T1 interaction, adjusting for age, sex, baseline echocardiographic LVEF, EAT volume index (EATVi), and the CMR-to-baseline echocardiography interval. BMI-specific effects came from the same model; sensitivity analyses used Firth regression and alternative follow-up, exposure, and outcome specifications. Results Overall, 53 patients met the criterion for LVEF improvement, including 16 of 52 patients (30.8%) with BMI < 24 kg / m 2 and 37 of 67 (55.2%) with BMI ≥ 24 kg / m 2 . The interaction term yielded an odds ratio of 0.396 (95% CI, 0.161–0.975; Wald P = 0.044), and inclusion of the interaction improved model fit (LR P = 0.037). For each 1-SD higher EAT native T1, the BMI-specific ORs were 1.407 (95% CI, 0.776–2.551; P = 0.261) for BMI < 24 kg / m 2 and 0.557 (95% CI, 0.284–1.092; P = 0.088) for BMI ≥ 24 kg/m 2 . Separate sensitivity analyses excluding patients with recorded follow-up CRT/CRT-D or jointly adjusting for baseline AF diagnosis and continuous eGFR, including Firth estimation of the joint model, yielded interaction point estimates close to the primary estimate, with CIs including 1. Uncertainty remained in the approximately 6-month window and continuous Δ LVEF analyses. Conclusion The association between regional EAT native T1 and LVEF improvement may vary according to BMI category in HFrEF. However, the BMI-specific associations were imprecisely estimated and sensitivity analyses showed variable precision. These findings are hypothesis generating and require independent confirmation in larger prospective cohorts using standardized imaging methods.