Cardiovascular magnetic resonance-based arrhythmogenic substrate assessment improves sudden cardiac death risk stratification in heart failure with mildly reduced ejection fraction: a multicenter retrospective observational study

Abstract

Summary Background The risk stratification of sudden cardiac death (SCD) in patients with heart failure (HF) with mildly reduced ejection fraction (HFmrEF) remains suboptimal. This study aims to evaluate the role of late gadolinium enhancement (LGE) and T1 mapping in predicting SCD in patients with HFmrEF and to provide an improved clinical risk stratification algorithm. Methods A total of 1855 patients with HFmrEF from five tertiary medical centers were enrolled in this study, all of whom underwent echocardiography and cardiovascular magnetic resonance (CMR). Data analysis utilized a multicenter development cohort (n = 1417) and a multicenter external validation cohort (n = 438). The primary endpoints were sudden cardiac death, appropriate ICD shocks, and resuscitated cardiac arrest. Secondary endpoints were a composite of HF death, unplanned HF hospitalization, or cardiac transplantation. Based on validated MRI predictors, a risk algorithm and corresponding clinical workflow were developed for SCD risk assessment. Findings During a median follow-up of 42 months in the development cohort, 142 and 140 patients experienced primary and secondary endpoints, respectively. Adjusted analyses identified the following as significant predictors of SCD-related events: LGE ≥6.3% (HR 5.33, 95% CI: 3.74–7.60; P < 0.001), ECV ≥28.4% (HR 3.44, 95% CI: 2.32–5.11; P < 0.001), and a native T1 z-score ≥2.7 (HR 2.40, 95% CI 1.67–3.48; P < 0.001). Patients with ECV >28.4% and no LGE showed higher SCD risk compared to those with ECV <28.4% and LGE <6.3% or mid-wall LGE. Conversely, patients with ln (NT-proBNP) <6.2, LGE <6.3%, and ECV <28.4% had a lower SCD risk, with an annual event rate of 0.4%. Notably, LGE ≥6.3% was associated with a high annual SCD event rate of 7.9%, irrespective of ECV, NT-proBNP levels, and LGE distribution. Interpretation In HFmrEF, at least 6.3% of LGE serves as a strong predictor of SCD risk, irrespective of its distribution. ECV significantly enhances risk stratification, particularly in patients with negative or mid-wall LGE. Funding This study was supported by the National Natural Science Foundation of China (Grant Nos. 81871354, 81571672) and Key R&D Program of Shandong Province, China (2025CXPT106, 2025CXGC020305).