Cardiac magnetic resonance-derived extracellular volume radiomics in reperfused ST-elevation myocardial infarction: long-term prognostic value and risk stratification

Abstract Aims Global extracellular volume (ECV) fraction independently predicts outcomes after ST-elevation myocardial infarction (STEMI), but microvascular injuries complicate its interpretation. This study aims to assess the prognostic value of ECV-derived radiomics [radiomics score (RadScore)] from cardiac magnetic resonance (CMR) for risk stratification in reperfused STEMI patients. Methods and results We retrospectively included 441 reperfused STEMI patients (mean age 60 ± 11 years; 371 males) from two centres, divided into development (n = 347) and validation cohort (n = 94) by centres. CMR scan was performed 1 week within the index event. ECV mapping was obtained from pre- and post-contrast T1 mappings. RadScore was developed through radiomics analysis on ECV mapping, and its predictive performance for major adverse cardiovascular events (MACE) was evaluated via logistic and Cox regression analyses. Over a median 3.1-year follow-up, MACE occurred in 81 patients (18.4%). RadScore showed strong predictive capability with an area under the curve (AUC) of 0.865 [95% confidence interval (CI): 0.768–0.962] and 0.821 (95% CI: 0.701–0.940) on the internal and external test cohort, respectively, reclassifying 44% and 32% of patients over clinical model, respectively. Patients with RadScore above the optimal threshold (RS+) experienced a more than 12-fold increase in MACE risk (95% CI: 2.96–47.71, P < 0.001) after adjusting for known risk factors. RS + provides incremental prognostic value beyond LVEF. Conclusion RadScore derived from ECV mapping was able to predict MACE and provided incremental prognostic value beyond established risk factors. Our findings underscore the potential long-term impact of acute changes in extracellular space patterns after myocardial infarction.