In-silico insights into personalised therapy selection of anti-arrhythmic drugs and ablation using patient-specific biatrial models for atrial fibrillation

Abstract

Listen

Translate article

Abstract Introduction Current treatment approaches for Atrial Fibrillation (AF) often lack personalisation, adopting a one-size-fits-all paradigm that disregards inter-patient diversity and its unique interplay with therapeutic responses. In-silico drug trials have demonstrated some value in the personalisation of care; however, their integration with patient-specific anatomical data and the synergistic effects of combined therapies remain unknown. Purpose To evaluate the patient-specific response to anti-arrhythmic drugs (AADs) and their efficacy both individually and in combination with pulmonary vein isolation (PVI) using personalised biatrial models created from imaging data. Methods Patient-specific biatrial models (n=70) were constructed from late gadolinium-enhanced MRI images by manual segmentation from the Atrial Segmentation Challenge (2018) dataset (Fig.1A). Models were pre-processed with landmark selection, regional assignment, and fibre allocation before simulations (Fig.1B). PVI was modelled as electrically inert rings around the pulmonary veins (Fig.1C) and AADs were administered via ionic modulation within CARPentry software (Fig.1D). Four commonly used AADs were evaluated: vernakalant, acute and chronic amiodarone, flecainide, and digoxin. Finite element simulations were performed with the addition of AADs, both individually and in combination with PVI, to evaluate their efficacy in terminating AF (Fig.2A). These simulations were post-processed to produce phase singularity (PS) density maps to evaluate changes in wavefront patterns and locations of re-entry and wavefront break-up (Fig.2B). Results The combination of digoxin with PVI and digoxin alone demonstrated the highest effectiveness in terminating AF, both achieving success in 68.4% of cases (95% CI: 67.5% to 89.3%, p≤0.0001). This efficacy significantly exceeded that of other AADs used either alone or in combination with PVI (15.8%-57.9%), whereas PVI alone demonstrated 0% efficacy (p≤0.05). The combined therapy exhibited a multi-faceted localisation pattern (Fig.2A), with AADs initially redistributing PSs to the pulmonary veins and left atrial appendage. This redistribution resulted in synergistic effects in terminating AF episodes in 52.6% more cases when combined with PVI (Fig.2B). Comparative analyses showed an inverse correlation between baseline PS density count and anatomical surface area with AF termination across all treatments, with significant correlations observed with vernakalant alone (rho=-0.55, p≤0.05) and combined with PVI (rho=-0.77, p≤0.05). Conclusions This study demonstrates the significance of personalised treatment selection in optimising patient outcomes. Combined therapeutic interventions have shown synergistic benefits in AF termination, surpassing individual therapies in this patient-specific model cohort. This shows the translational potential of using patient-specific stratification to inform treatment selection.Biatrial model generationAF therapeutic simulation and processing