Background: Clinically identifying the subset of persistent AF patients for whom pulmonary vein isolation (PVI) ablation is a sufficient treatment is challenging. Virtual patient cohorts allow mechanistic investigation into the individual contribution of the electrical and structural substrate to AF ablation and testing of metrics for predicting outcome.
Objective: To investigate the effects of AF cycle length (CL), electrical driver location, left atrial (LA) surface area, and fibrosis burden on PVI outcome using a virtual cohort.
Methods: We generated 100 patient-specific left atrial bilayer models incorporating fibrotic remodeling from persistent AF late-gadolinium enhancement (LGE) MRI, together with atrial fibers from a novel DTMRI atlas. AF was simulated and post processed to determine electrical driver locations over 15s. PVI ablation outcome was classified as responder (termination or macro-reentry) or non-responder (AF continues).
Results: PVI outcome for the cohort was 48% responder. AF CL was no different between the groups (206ms vs 206ms, p=0.26). The ratio of electrical drivers in the PV compared to the whole LA was significantly higher in PVI responders (0.34 vs 0.23, p=0.004, see Fig). LA area was significantly smaller in PVI responders (99cm² vs 117cm², p=0.003). Fibrosis area was not significantly different (10.4cm² vs 7.7cm², p=0.1).
Conclusion: LA surface area and pulmonary vein driver density are predictive of PVI outcome, whereas fibrosis area and AF cycle length are not predictive. We present a large personalised virtual patient cohort and utilize it to predict the density of electrical drivers in the PV region for predicting persistent AF PVI outcome.