In silico predictions of action potential propagation in doxorubicin cardiotoxicity: A parametric study using preclinical 3D magnetic resonance imaging-based fibrotic left ventricle models.

Doxorubicin (DOX) is a widely used chemotherapeutic agent, but its cardiotoxic effects, including diffuse myocardial fibrosis, increase the risk of dangerous arrhythmias. There is a critical need for non-invasive tools to predict DOX-related ventricular arrhythmias in early chronic stages following chemotherapy. A computational study was performed using experimental data from three pigs: one control and two at 9 weeks following DOX. Customized 3D left ventricular (LV) models were generated from late gadolinium-enhanced magnetic resonance imaging and electro-anatomical maps, integrating tissue structure, electrical properties (healthy/fibrosis) and fibre directions. Action potential (AP) wave propagation was simulated using a high-performance numerical solver. A virtual programmed stimulation protocol was applied in 96 simulations to assess arrhythmia inducibility, varying the parameters corresponding to excitability and conduction velocity in fibrotic zones. Arrhythmias were inducible only in DOX-treated cases. Reentrant wave genesis depended on: excitability, conduction velocity, fibrosis distribution and AP duration heterogeneity. In one scenario, AP heterogeneities and a ≥70% reduction in diffusion coefficient were required to induce reentry despite unchanged excitability in fibrosis. This study presents the first computational simulation of DOX-induced cardiotoxicity in a realistic 3D LV model using a highly efficient, automated Lattice-Boltzmann approach. Our findings provide insights into arrhythmogenic mechanisms and may aid in developing strategies to prevent and treat DOX-related cardiotoxicity. KEY POINTS: We developed a novel semi-automated computational framework to construct high-resolution 3D magnetic resonance imaging-based left ventricular models designed to study via simulations the electrical activity after chemotherapy using a GPU-optimized Lattice-Boltzmann method solver. Our digital heart twins were directly calibrated and validated using measurements of conduction velocity and action potential wave features obtained via catheter-based electro-anatomical mapping after chemotherapy in preclinical swine models. This specific virtual parametric study demonstrates that both electrophysiological and structural alterations induced by diffuse fibrosis substantially modulate ventricular arrhythmias in the sub-chronic phase following doxorubicin therapy.