Rethinking post-infarction remodelling: identification and validation of data-driven cardiac MRI phenotypes for risk stratification
- 22 May 2026
- journal article
- Published by Oxford University Press (OUP) in European Heart Journal - Cardiovascular Imaging
Abstract
Aims: Conventional definitions of adverse left ventricular remodelling (ALVR) following ST-elevation myocardial infarction (STEMI), which are based on longitudinal changes in left ventricular end-diastolic volume index and ejection fraction, have demonstrated limited prognostic value. This study aimed to identify prognostic relevant remodelling phenotypes using a data-driven approach on longitudinal cardiac magnetic resonance (CMR) data. Methods and results: In this prospective longitudinal study, a derivation cohort (n = 337) and an independent validation cohort (n = 190) of patients with reperfused STEMI were analysed. Patients underwent CMR at <7 days and 6-month follow-up. An unsupervised clustering algorithm was applied using baseline biventricular, biatrial, and infarct parameters, plus their longitudinal changes. Major adverse cardiovascular events (MACE) included all-cause mortality, recurrent myocardial infarction (MI), and heart failure. Three reproducible post-MI phenotypes were identified: Low Risk (n = 175), Early Remodelling (n = 96, severe acute injury with minimal longitudinal change), and Atrial-dominant Remodelling (n = 66, bi-atrial and ventricular enlargement). Phenotype-stratified MACE-free survival differed significantly in both cohorts (derivation P < 0.001; validation P = 0.002). In the validation cohort, both Early Remodelling (hazard ratio [HR] 4.73; 95% confidence interval [CI]: 1.77–12.64) and Atrial-dominant Remodelling (HR 4.02; 95% CI: 1.49–10.88) were associated with elevated MACE risk vs. the Low-Risk group; the Atrial-dominant Phenotype was further associated with heart failure hospitalization (subdistribution HR: 5.92; 95% CI: 1.04–33.56). Adding the derived phenotypes to baseline CMR parameters improved MACE discrimination (C-index 0.755 vs. 0.714; ΔC 0.041, 95% CI: 0.006–0.074), whereas conventional ALVR did not. Conclusion: Three reproducible post-MI remodelling phenotypes were identified, which provide superior risk stratification for MACE over conventional ALVR criteria and insights into post-MI heart failure.Keywords
Funding Information
- National Natural Science Foundation of China (82471931)
- National Natural Science Foundation of China (82421001)
- Shanghai Municipal Commission of Science and Technology Medical Innovation Research Special Project (23Y11906900)
- Shanghai Municipal Commission of Science and Technology Medical Innovation Research Special Project (20244Z0003)
- Renji Hospital National High-Level Leading Talents Cultivation Program
- Science and Technology Commission of Shanghai Municipality (25TS1405400)
- Shanghai Jiao Tong University AI for Engineering Initiative (WH410263001/004)
This publication has 45 references indexed in Scilit:
- Incidence and Predictors of Heart Failure with Reduced and Preserved Ejection Fraction After ST-Elevation Myocardial Infarction in the Contemporary Era of Early Percutaneous Coronary InterventionEuropean Journal of Heart Failure, 2024
- Intramyocardial Hemorrhage and the “Wave Front” of Reperfusion Injury Compromising Myocardial SalvageJACC, 2022
- Predominant Subtype of Heart Failure After Acute Myocardial Infarction is Heart Failure with Non-Reduced Ejection FractionESC Heart Failure, 2020
- Definition of left ventricular remodelling following ST-elevation myocardial infarction: a systematic review of cardiac magnetic resonance studies in the past decadeHeart Failure Reviews, 2020
- Left Ventricular Post-Infarct RemodelingJACC: Heart Failure, 2020
- Practical recommendations for reportingFine‐Gray model analyses for competing risk dataStatistics in Medicine, 2017
- Pathophysiology of LV Remodeling in Survivors of STEMIJACC: Cardiovascular Imaging, 2015
- Effect of Metformin on Left Ventricular Function After Acute Myocardial Infarction in Patients Without DiabetesJAMA, 2014
- Extensions of net reclassification improvement calculations to measure usefulness of new biomarkersStatistics in Medicine, 2010
- Evaluating the added predictive ability of a new marker: From area under the ROC curve to reclassification and beyondStatistics in Medicine, 2007