3D substrate complexity analysis using cardiac MRI predicts ICD therapy in post-infarct ventricular tachycardia

Abstract

Funding Acknowledgements Type of funding sources: Foundation. Main funding source(s): EACVI, Netherlands Heart Institute. Background Implantable cardiac defibrillator (ICD) implantation can protect against sudden cardiac death (SCD) after a myocardial infarction. However, relatively few patients with an ICD experience a life-threatening arrhythmic event. Imaging studies have proposed metrics based on 2D analysis of late gadolinium enhancement (LGE) characteristics to predict post-infarct malignant arrhythmias and improve SCD risk assessment. However, given the intrinsic 3D nature of the electrical pathways through the infarcted regions, 3D reconstructions of the scar substrate from LGE imaging may be required to fully characterize the pro-arrhythmic nature of the scar substrate. Aim To evaluate the accuracy of LGE based 3D metrics such as conduction corridors (regions of borderzone (BZ) surrounded by scar core) and 3D interface surfaces (boundaries between scar and myocardium) towards predicting ICD therapy. Methods ADAS LV and custom-made software was used to generate 3D patient-specific ventricular models in a prospective cohort of post-infarct patients (n=40) having undergone LGE imaging pre-ICD implantation. The extent of variation in scar-characteristics was evaluated in ADAS by quantifying the BZ, scar core, the number and weight of conduction corridors i.e. BZ surrounded by scar core. Custom-written scripts were used to calculate metrics describing the 3D topology of the scar substrate, specifically the interface area between myocardium and total enhancement (BZ+core), and the interface between BZ and core. These metrics were compared with ICD therapy during follow-up. Results Total corridors were comparable between both groups (6.53 ± 7.9 vs. 4.6 ± 4, p = .38). Corridor weight demonstrated a trend towards higher mass in the event group (2.7 ± 2.1g vs. 1.6 ± 1.4g, p = .06). Patients with an event (n=17) had higher myocardium-total enhancement interface (103.8±35.1cm2 vs. 77.4±33.7cm2, p = .021) and BZ-core interface (76±27.5cm2 vs. 55.2±27.6cm2, p = .024). Cox-regression demonstrated a significant independent association of myocardium-total enhancement interface with an event (HR 2.79; 1.44–5.44, p < .01). Kaplan-Meier analysis (Figure 1) showed a significantly higher event rate in patients with an interface area between myocardium-total enhancement of more than 72cm2 and BZ-core more than 42.3cm2. Conclusion These results demonstrate that patients with appropriate device therapy had larger myocardium-total enhancement and BZ-core surface interface areas. Conceptually, the BZ-core interface could be considered to be related to the reentrant circuit path-length whilst the myocardium-total enhancement interface reflects the surfaces most-likely to initiate unidirectional block, both of which can be consider pro-arrhythmic substrates. These findings emphasize the importance of visualizing and thereby characterizing substrate as a 3D entity instead of the currently applied 2D approach to facilitate early identification of high-risk patients.