Non-invasive electrocardiographic imaging of torsades de pointes in an ovine model of drug-induced QT prolongation: recovery dispersion sets the stage, and abnormal impulse formation pulls the trigger

Background Torsades de pointes (TdP) arise from a dynamic spatiotemporal interplay between arrhythmogenic substrate and triggering events. Most existing in-vivo models investigating TdP rely on invasive surgical cardiac access, which alters electromechanical function of the heart and autonomic influences, thus limiting their translational relevance to spontaneous arrhythmogenesis in patients. Purpose To non-invasively characterize regional and temporal TdP substrate-trigger interaction using electrocardiographic imaging (ECGI) in a large-animal model of drug-induced QT prolongation with spontaneous TdP. Methods Female Welsh Mountain sheep (N=7, ≈18 months, 45±6 kg) were anesthetized and shaved. After placement of 128 electrodes for body-surface potential mapping, a contrast-enhanced computed tomography scan was performed to acquire heart-to-torso geometry and precise electrode positions. Subsequently, escalating IKr blockade was induced by i.v. dofetilide (N=7/7; up to 21-42 µg/kg), followed by IKs blockade by i.v. HMR1556 (N=4/7; up to 2.25 mg/kg), until spontaneous TdP occurred. ECGI-derived epicardial unipolar electrograms were reconstructed, and electrophysiological characteristics during sinus rhythm, spontaneous premature ventricular complexes (PVC), and TdP induction were mapped (Fig. A). Results Dofetilide increased Fridericia-corrected QT (QTcF) by +47±27% to 581±105 ms (p=0.008). Corresponding ECGI-derived recovery times (RTs; +63±49% at 21 µg/kg; p=0.01) and RT gradients (RTG; +79±117% at 21 µg/kg; p=0.04) increased dose-dependently (Fig. B). Activation time (AT) did not change significantly upon IKr block. Interestingly, the dofetilide-mediated increase in RTGs was not due to amplification of pre-existing gradients but arose from newly-formed regions of heterogeneous RT prolongation (38±21% spatial RT variability, p=0.007) (Fig. C). RTGs were spatiotemporally dynamic, as exemplified by beat-to-beat shifts in RTG locations upon isoproterenol (3/7 animals, 1.25-5 µg/kg). After a median (range) dose of 21 (21-42) µg/kg dofetilide and 1.61 (0.69-2.25) mg/kg HMR1556, 4/4 animals with combined IKr and IKs block developed PVCs and (non-)sustained (ns)TdPs. Per animal, a median (range) of 8 (3-11) isolated PVCs and 5 (3-8) (ns)TdPs were mapped, showing beat-to-beat augmentation of RTGs momentary to PVC. Whereas arrhythmia coupling intervals did not differ between (ns)TdP and isolated PVCs, (ns)TdP-initiating beats exhibited longer mean ATs (+83±125%, p=0.0001), longer activation durations (+71±141%, p=0.004), higher AT dispersion (+87±182%, p=0.004), and higher AT gradients (+80±162%, p=0.002; Fig. D). Conclusion Spatiotemporally heterogeneous RT prolongation creates a vulnerable TdP substrate, whereas abnormally activating PVCs act as the immediate trigger for TdP onset, highlighting the translational potential of ECGI for in-vivo investigations of arrhythmogenic mechanisms related to prolonged repolarization.