Cardiovascular magnetic resonance determinants of ventricular arrhythmic events after myocardial infarction

Abstract

Abstract Funding Acknowledgements Type of funding sources: None. Background Late gadolinium enhancement cardiac magnetic resonance (LGE-CMR) permits to identify the arrhythmogenic substrate in chronic post-myocardial infarction (MI) patients. It is unknown why a minority of chronic post-MI patients develop sustained ventricular tachycardias (VT) over follow-up, regardless of their left ventricular ejection fraction (LVEF). Objectives To noninvasively characterize scar differences and potential predictors of VT occurrence in chronic post-MI patients. Methods A case-control study was designed through retrospective LGE-CMR data analysis of chronic post-MI patients i) consecutively referred for VT substrate ablation after a first VT episode (n = 66), and ii) from a control group (n = 84) with no arrhythmia evidence. The myocardium was characterized differentiating core, border zone (BZ) and BZ channels (BZC) using the ADAS 3D post-processing imaging platform. Clinical and scar characteristics, including a novel parameter, the BZC mass, were compared between both groups. Results 150 post-MI patients were included for analysis. Four multivariate Cox proportional hazards regression models were created for total scar mass (model 1), BZ mass (model 2), core mass (model 3), and BZC mass (model 4, see table). All of them were adjusted by age, sex, and LVEF. In the corresponding models, only total scar mass, BZ mass, core mass, and BZC mass were independent variables associated with the development of VT. BZC mass showed the best performance: a cut-off of 5.15 g identified the cases with 92.4% sensitivity and 86.9% specificity [AUC 0.93 (0.89–0.97); p < 0.001], with a significant increase in the AUC compared to the other scar parameters (p < 0.001 for all pairwise comparisons using the De Long’s test). By using BZC mass as a risk stratification parameter together with LVEF, the net reclassification improvement (NRI) was 33.3% for the cases, and 39.3% for the controls. The net proportion of patients reclassified correctly was 36.7% Conclusions The mass of BZC, automatically obtained with a commercially available CMR post-processing software, is the strongest independent variable associated with the occurrence of clinical SMVT in post-MI patients after covariate adjustment for age, sex, and LVEF. The measurement of BZC mass could permit a more accurate arrhythmia risk stratification than LVEF in chronic post-MI patients (NRI 36.7%). Scar characteristics analyzed from LGE-CMR imaging should be taken into consideration to better stratify ventricular arrhythmia risk in chronic post-MI patients. Abstract Figure. ROC curves for predicting VT occurrence