Introduction: Atherosclerosis is the build-up of fatty deposits, cholesterol, cellular wastes, calcium and other substances as a “plaque” in the inner lining of arteries. These building plaques can rupture to form a blood clot, effectively blocking blood flow. Although the pathogenesis of atheromas as cell-mediated responses to inflammation have been postulated and shown in specific static models, the integration of this information into a model that captures the dynamics of the process is lacking. To develop effective treatment strategies and techniques, it is essential to characterize plaque formation through the integration of inflammatory responses, treatments, and predisposing factors. Agent-based modeling utilizes mechanistic basic science knowledge to develop a computational model that dynamically represents the inflammatory influences on atherosclerosis. Methods: A basic science literature review was conducted to identify the mechanisms of specific cell-cell interactions in inflammation and atheroma formation. This information was organized into influence diagrams of each cell type's behavior. Cells were treated as input-output devices, where the cell's local environment (molecules, mediators and neighboring cells) led to actions within the cells (expressed as mechanistic rules) to produce outputs back into the local environment. These cellular influence diagrams were converted into an abstract agent based model (ABM) using NetLogo. The resulting ABM included the multiple arterial layers: intima, media, and adventitia; and cellular agent classes: endothelial cells, smooth muscles cells, macrophages, foam cells, platelets, monocytes, t-lymphocytes and fibroblasts. Pattern oriented analysis was used to evaluate the model's behavior under differential shear forces associated with bifurcations of the vascular tree, where increased endothelial activation leading to atheroma formation occurs. Results: The arterial wall ABM successfully simulated general vascular wall cell interactions in macro-vessel endothelial activation, clot formation, intra-intimal inflammation and pre-atheroma formation. These behaviors matched those seen in histological examinations of atheroma formation. Furthermore, varying levels of predisposing factors, such as LDL and platelets, corresponded to respective changes in the likelihood and severity of atheroma development. When subjected to different abstract therapies for specified conditions, the ABM demonstrated behaviors similar to those seen in clinical situations. Conclusions: Combining atherosclerotic and generative inflammatory responses within an ABM provides a dynamic representation of the mechanistic hypothesis associated with atheroma formation. This model has multiple potential applications for modeling arterial wall responses to stent placements, angioplasty or surgical intervention. Agent based modeling, used as a means of dynamic knowledge representation, can be integrated into traditional biomedical research environments, where it can be the first step in the evaluation proposed mechanisms and interventions. Future work includes the incorporation of anti-platelet modalities, anti-nitric oxide therapies, anti-inflammatory therapies, and drug-eliciting stents.
Read full abstract