AN AGENT-BASED MODEL OF HEART TRANSPLANT ALLOGRAFT VASCULOPATHY

Cardiac allograft vasculopathy (CAV) is a major cause of heart transplant rejection. CAV is an accelerated form of coronary artery disease that affects the allograft. It is characterized by progressive intimal thickening due to a complex interplay between immunological and non-immunological events. The underlying mechanisms of CAV are largely unknown. To gain a better knowledge of its dynamics, many pre-clinical studies (e.g., in vivo models) are needed, but they are often associated with high costs, long protocol timelines and ethical concerns. Moreover, access to crucial time points across the experimental follow-up and measurements of biomechanical quantities are critically difficult. Consequently, the use of in vivo models remains somehow constrained, fostering the research to exploit alternative methodologies. Computational modeling (in silico) can overcome some of these bottlenecks by providing a digital replica of pathology development. This approach can provide insights into the multi-scale biological processes governing the post-transplant period. It has the capability to estimate the effect of quantities hardly measurable in vivo, simulate long observation periods and access unlimited number of time points with low computational time. In this view, biologists and clinicians can take advantage of versatile and cost-effective platforms to pre-test therapeutic treatments and surgical procedures in silico, thus directing in vivo research on the most promising outcomes. In this thesis, a 2D computational model simulating CAV in mice coronary arteries was developed to elucidate the pivotal events involved in CAV pathology development. A stochastic 2D agent-based model (ABM) of an idealized mouse coronary artery was implemented at the cell-tissue scale and populated with cell and extracellular matrix (ECM) agents. CAV initiation and progression were simulated by detailing: i) behavior of each agent, ii) mutual interactions, and iii) environmental conditions. The model was integrated with literature data from mice models, which include coronary artery geometry, immune cell trends, and hemodynamic distributions. The following three major cascades of events leading to CAV at the vessel wall-blood interface were simulated: i) macrophage infiltration, ii) chemoattractant diffusion, and iii) arterial wall remodeling, driven by inflammatory and hemodynamic stimuli. Phenomena were implemented by defining probabilistic rules and using a Monte Carlo method. A large sensitivity analysis, based on latin hypercube sampling and partial rank correlation coefficient evaluation, was conducted to measure the relation between each driving parameter and the ABM outputs of interest. The model was able to capture relevant features of CAV as a function of inflammation and hemodynamic stimuli, namely: i) lumen area decrease over time and progressive intimal thickening, ii) asymmetric intimal growth at the most susceptible regions of heterogenous stimuli, reflecting a focal CAV, and iii) concentric and symmetric intimal growth, under homogeneous stimuli, mirroring a diffuse CAV. Sensitivity analysis confirmed the experimental evidences, identifying in the immune response factors, the most decisive role in CAV lesion formation. In conclusion, the model proved the potentiality of the implemented approach to simulate CAV development, thus constituting a proof-of-concept and laying the basis for the implementation of a 3D and mouse-specific model to better investigate the role of the local hemodynamics.