LONGITUDINAL ANALYSIS OF LEFT VENTRICULAR REMODELING IN POST-MYOCARDIAL INFARCTION PATIENTS USING MRI

Abstract

Left ventricular (LV) remodeling following acute myocardial infarction (AMI) is a critical determinant of patient outcomes, often leading to heart failure and increased mortality. This study utilized serial cardiac magnetic resonance imaging (MRI) to evaluate the temporal progression of LV remodeling in critically ill post-AMI patients. A prospective cohort of patients with confirmed AMI underwent cardiac MRI at baseline (within 7 days), 6 months, and 12 months. Parameters including LV end-diastolic volume (LVEDV), end-systolic volume (LVESV), ejection fraction (LVEF), infarct size, transmurality, and myocardial strain were assessed. Late gadolinium enhancement (LGE) was used to quantify myocardial scarring. Adverse remodeling was defined as a >15% increase in LVEDV from baseline. Clinical variables such as hemoglobin glycation index (HGI) and stress hyperglycemia ratio (SHR) were also evaluated. Machine learning algorithms were applied to explore predictive modeling based on imaging and clinical data. Significant reductions in LVEDV and LVEF occurred only in individuals who showed unfavorable remodeling, as the mean LVEF fell from 49.8% at the start to 42.6% after 12 months (p<0.01).  There was more infarct area and transmurality in this group which went along with reduced function.  The results showed that it was the myocardial strain test and not the volumetric study, that picked up the early regional decline seen in myocardial function.  Unfavorable remodeling in the arteries was related to high SHR and HGI levels, suggesting a role for metabolic factors.  Forecasts of remodeling were accurate for initial imaging and clinical data, with a model AUROC of 0.87.  The results suggest that both cardiac MRI and AI are valuable in identifying who requires special treatment after a heart attack.  Cardiac MRI tests done over time reveal important changes in the heart of patients who have had a heart attack.  Combining imaging biomarkers with medical and metabolic information, using AI tools, may help to make risk assessment better and adjust personalized treatment decisions.