Electrocardiographic imaging of repolarisation abnormalities in structural heart disease for characterisation of arrhythmogenic phenotypes

Background Electrocardiographic Imaging (ECGI) enables non-invasive assessment of 3-dimensional activation and repolarisation sequences. Whether ECGI allows to characterise repolarisation abnormalities in relation to myocardial scar and differentiate arrhythmogenic phenotypes has not been systematically investigated. Objective To compare ECGI-derived activation time (AT) and activation-recovery interval (ARI) metrics in sinus rhythm in patients with structural heart disease (SHD) and evaluate their relationship to myocardial scar, left ventricular systolic dysfunction (LVSD) and re-entrant ventricular tachycardia (VT) status. Methods We retrospectively analysed 71 patients who underwent 252-electrode ECGI recording with anatomical CT in sinus rhythm and delayed-enhancement cardiac MRI. Patients were categorised into four cohorts according to presence of myocardial scar, ejection fraction (LVEF) and history of re-entrant VT: (1) scar+/VT+, (2) scar+/VT-, (3) LVSD+/VT-, (4) controls. For all cases, reconstructed epicardial unipolar activation, repolarisation and derived ARI maps during sinus were co-registered with 3D MRI scar models. ARI duration, ARI dispersion and AT dispersion were estimated at ventricular and, for the left ventricle, AHA segmental levels and compared across groups and tissue types using mixed-effects models. Dispersion was estimated as standard deviation of AT and ARI of all EGMs in the respective spatial domain. EGMs with >10mm distance from MRI scar-shell were excluded. Results 39 patients with scar+/VT+ (28.3% ischemic aetiology, LVEF 41.2±17.5%), 14 patients with scar+/VT- (37.5% ischemic, LVEF 33.3±10.6%), 9 patients with LVSD/VT- (LVEF 21.9±5.6%) and 9 controls (LVEF 59.2±7.1%) were included. LV scar burden in VT+ subjects was higher compared to cases without re-entrant VT (median 26.2g [IQR 25.5] vs 19.4g [IQR 12.1], p<0.001). In sinus rhythm, ARI duration and dispersion differed significantly between groups and tissue types (p<0.001). On average, ARI was longest over scar regions; ARI dispersion was higher in transmural (22.0ms [26.3]) and subepicardial scar (20.6ms [22.8]), compared to subendocardial (16.5ms [16.1]) and non-scarred myocardium (15.8ms [20.3], p<0.001). ARI dispersion associated with scar was significantly higher in scar VT+ compared to scar-VT- (b = -8.4ms, p=0.009). Activation dispersion was significantly increased in segments of scar compared to normal myocardium (b = -5.3ms, p<0.001) but did overall not allow to distinguish VT status. Severe LV dysfunction, even in the absence of overt myocardial scar, was also associated with significantly prolonged ARI and elevated ARI and AT dispersion metrics compared to controls. Conclusion ECGI-derived repolarisation abnormalities in SHD are associated with scar presence, transmurality and VT history. Segmental ARI dispersion in association with myocardial scar may potentially serve as a complementary non-invasive marker of arrhythmogenic risk.