Electrical Substrates Driving Response to Cardiac Resynchronization Therapy: A Combined Clinical–Computational Evaluation

BACKGROUND: The predictive value of interventricular versus intraventricular dyssynchrony for response to cardiac resynchronization therapy (CRT) remains unclear. We investigated the relative importance of both ventricular electrical substrate components for left ventricular (LV) hemodynamic function. METHODS AND RESULTS: First, we used the cardiovascular computational model CircAdapt to characterize the isolated effect of intrinsic interventricular and intraventricular activation on CRT response (Delta LVdP/dt(max)). Simulated Delta LVdP/dt(max) (range: 1.3%-26.5%) increased considerably with increasing interventricular dyssynchrony. In contrast, the isolated effect of intraventricular dyssynchrony in either the LV or right ventricle was limited (Delta LVdP/dt(max) range: 12.3%-18.3% and 14.1%-15.7%, respectively). Effects of activation during biventricular pacing on Delta LVdP/dt(max) were small. Second, electrocardiographic imaging-derived activation characteristics of 51 CRT candidates were used to personalize ventricular activation in CircAdapt. The individualized models were subsequently used to assess the accuracy of Delta LVdP/dt(max) prediction based on the electrical data. The model-predicted Delta LVdP/dt(max) was close to the actual value in patients with left bundle branch block (measured-simulated: 2.7 +/- 9.0%) when only intrinsic interventricular dyssynchrony was personalized. Among patients without left bundle branch block, Delta LVdP/dt(max) was systematically overpredicted by CircAdapt (measured-simulated: 9.2 +/- 7.1%). Adding intraventricular activation to the model did not improve the accuracy of the response prediction. CONCLUSIONS: Computer simulations revealed that intrinsic interventricular dyssynchrony is the dominant component of the electrical substrate driving the response to CRT. Intrinsic intraventricular dyssynchrony and any dyssynchrony during biventricular pacing play a minor role in this respect. This may facilitate patient-specific modeling for prediction of CRT response.