111/Non-invasive 3D mapping of premature ventricular complexes – The Brompton experience

Background: A novel non-invasive three-dimensional (3D) mapping software, and its upgraded version, shows the unique feature of combining 3D DICOM data with 12-lead electrocardiographic (ECG) data to precisely localize ventricular arrhythmias. This allows to “move” the localisation diagnosis to the ward prior to the invasive cath lab procedure and is especially helpful in patients with polymorphic PVCs. Its accuracy to localise premature ventricular complexes (PVCs) and guide catheter ablation has been already demonstrated in previous studies. However, recruited patients had structurally normal hearts (<10% of scar burden).
Purpose: We aimed to challenge the non-invasive 3D mapping software accuracy in a different and more challenging patient population, and to compare it with its upgraded version within the same cohort.
Methods: Eleven consecutive patients (7F, mean age 46 ± 17 yrs) were studied. Underlying cardiac conditions were: four (36.4%) patients with congenital heart disease of moderate severity (Bethesda 2), two patients (18.2%) with dilated cardiomyopathy, one patient (9%) with Brugada syndrome, three patients (27.3%) with structurally normal heart and
<10% of scar burden, and lastly one (9%) with structurally normal heart but severe ischaemic disease. For all, a 3D personalised model of the heart and torso were generated from either cardiac magnetic resonance imaging or computed tomography scans. This was merged with a 3D picture of the ECG-leads position. Finally, digital ECG files of the PVC or
paced beats derived from either 12-lead Holter or electrophysiological (EP) recording system were imported into the 3D non-invasive mapping software for analysis. Finally, the location was compared to the site of ablation on electro-anatomical mapping system.
Results: The prior non-invasive 3D mapping software showed an accuracy of 63.6% in the 22 ECG locations collected: eight right ventricular outflow tract (RVOT), four right ventricle (RV), seven endocardial left ventricle (LV), three epicardial LV; overall, four paced beats and 18 PVCs. However, when locations where re-analysed with the upgraded software version, the accuracy showed to improve up to 90.9%. Two patients (18.2%) underwent only an EP study. However, among the paced morphologies (two from RVOT and two from RV apex) only one was correctly localised at the RVOT using the former version, whereas the upgraded software could correct the second RVOT localisation.
Conclusions: Accurate non-invasive 3D localisation of PVCs’ origin was achieved in 90.9% of analysed samples in a diverse cohort of patients including patients with structural heart disease, cardiomyopathy and a scar burden beyond 10%. Challenges are still encountered with paced beats, and a prospective trial is currently in preparation.