Towards a Multiscale Computational Framework for Cardiac Allograft Vasculopathy: Design and Development of a Lymphocyte Transport Model in Murine Coronary Arteries

Cardiac Allograft Vasculopathy (CAV) is a leading cause of chronic graft rejection in heart transplant (HTx) recipients, with an incidence of 50% within 10 years post-HTx and an associated 30% mortality rate. CAV is characterized by progressive intimal thickening and luminal narrowing of the coronary arteries. Its multiscale nature is reflected in the interplay between immunological (e.g., lymphocyte (LYM) and macrophage (MP) infiltration), biomechanical (e.g., disturbed flow patterns like low wall shear stress (WSS)), and geometrical (e.g., irregular arterial wall shape profiles) factors. Current therapies are limited, with re-transplantation being the only definitive solution. Moreover, diagnosis remains challenging and typically possible only at late stages, due to an incomplete understanding of the mechanisms underlying the disease early onset. While traditional studies on CAV have relied on animal models (mainly murine), these approaches are constrained by ethical concerns and the limited availability of longitudinal time points. In this context, computational (in silico) models have emerged as valuable complementary tools to in vivo research. Recently, two computational models have been developed in the context of CAV: (i) an agent-based model (ABM) capable of replicating coronary artery remodeling in mice, and (ii) a computational fluid dynamic (CFD) model of murine left coronary artery (LCA), which paves the way for a multiscale approach in CAV research. Given the central role of LYM infiltration in initiating the cascade of events leading to CAV, we hypothesize that a computational model predicting LYM transport can identify regions of potential accumulation, thereby revealing areas at risk for disease progression. In parallel, we hypothesize that embedding this model within a multiscale in silico framework, which integrates ABM with inputs from CFD and LYM transport simulations, will enable the investigation of CAV mechanisms across molecular, cellular, and tissue scales. Accordingly, steady-state CFD simulations were performed on n=6 3D murine LCA geometries. A convection-diffusion model was developed to simulate LYM distribution along the vessel, incorporating mouse-specific parameters and boundary conditions. The outputs from both the CFD and LYM transport models subsequently served as inputs for the ABM within a multiscale framework. In turn, the ABM-predicted lumen remodeling was used to reconstruct updated 3D LCA geometries, enabling an iterative coupling between CFD and ABM over a four-week follow-up period. The LYM transport model revealed preferential accumulation in the outer wall of bifurcations and in vessel segments with elliptical rather than circular profile. In these areas, spatial correlations between low-WSS and elevated LYM concentrations were observed, underscoring the influence of hemodynamics on immune responses. Moreover, the observation that accumulation also occurred in other regions suggested that additional factors, such as wall permeability and immune cell recruitment, may play a critical role in driving accumulation. When coupled with the ABM, the integrated multiscale model successfully reproduced key pathological features of CAV, including heterogeneous LYM infiltration, MP activation, and luminal narrowing. In conclusion, once validated against in vivo data, this in silico framework will hold significant potential for advancing the understanding of CAV pathogenesis and identifying novel biomarkers for early detection.