Abstract 16247: Personalized MRI-Based Modeling Predicts Ventricular Tachycardia Vulnerability in Patients Receiving Primary Prevention ICDs

Introduction: Patients with ischemic cardiomyopathy and low ejection fraction ( Methods: Subject-specific models of the ventricles, including post-infarct scar and peri-infarct regions, were constructed from late-gadolinium enhanced cardiac MRIs for subjects (n=16) receiving an ICD. In the models, programmed stimulation was performed to establish the inducibility and exit sites of VT. The models were created blinded to clinical outcome. During the ICD placement, programmed stimulation was performed to validate model predictions. The VT was recorded with a 12 lead ECG and evaluated as right vs left bundle branch block, superior vs inferior axis, and the polarity of lead I. The inducibility and morphology of VT from computational models and clinical studies were compared to evaluate the ability of the model to predict clinical VTs. For morphology comparisons, we correlated each episode of the clinically induced VT to the model predicted exit site based on these three criteria. Results: Monomorphic VT was clinically induced in 8 of 16 patients. The models correctly predicted the inducibility of VT or lack thereof in all 16 cases. In the 8 cases in which the model correctly predicted VT, 21 of 24 morphology parameters matched the exit site location in the models. Discussion: The results demonstrate the feasibility of subject-specific modeling to identify patients at risk of SCD from monomorphic VT and encourage further testing in a larger patient cohort.