Development of an In Silico Simulation Tool for Personalized Stent Interventions

Introduction: In-stent restenosis (ISR) remains a challenge that hinders long-term treatment success. Stents and implantation procedures are vital factors that determine clinical outcomes. The mechanical impact of a stent on the arterial wall significantly influences ISR development. Therefore, careful consideration must be given to the key parameters such as inflating pressure, stent type, and size. It is also important to account for patient-specific factors (e.g. plaque and vessel morphology, immune characteristics). At present, the assessment for optimal implantation procedures primarily relies on clinicians' experiences. No tool is available to rapidly estimate the risk of ISR while incorporating patient-specific factors and assisting clinicians with stent selection and procedural adjustments during the intervention.
Aim: This multidisciplinary project aims to develop a simulation tool that accounts for patient-specific factors, providing reliable and rapid estimates of ISR, and guiding clinicians for optimal implantation procedures.

Comprehensive Data Acquisition for Modeling In vivo and ex vivo image processing modalities, along with Lugol pre-treatment technique for enhancing ex vivo micro-CT were developed. These methods provided patient-specific data for the modeling. An atherogenic apolipoprotein E knockout rat model was established, in which human stents were implanted and assessed for stent-induced injury. Using these data, we evaluated and confirmed a positive correlation between stent strut indentation, vessel wall injury, and neointima growth, proposing indentation as a novel tool for predicting ISR. In addition, we explored the relationship between indentation and hemodynamic indicators using an in silico model, identifying an association between high indentation and abnormal hemodynamic indicator values. This knowledge will be incorporated into the computational model. 

Computational Modeling Multi-layered strut configuration (metallic core and polymeric coating) of drug-eluting stents and its deployment were successfully modeled using reduced-integration solid and solid beam elements. This approach solved the computational intensity while improving efficiency and robustness. Interaction between a stent, blood flow, and arterial tissue was modeled. The hemodynamics and drug release in the lumen were simulated using a proprietary XNS code, while drug release and cell dynamics in the arterial wall were simulated using the FEAP method. This framework includes simulation of blood flow and extracts hemodynamic indicators (e.g. time-averaged wall shear stress, oscillatory shear index), drug influx from the polymeric layer into the arterial wall, drug diffusion into the lumen, the influence of hemodynamic indicators on ISR, and the flux of PDGF and TGF based on hemodynamic indicators values. Moreover, a 3D surrogate model was developed using physics-informed neural networks (PINNs), allowing for the integration of patient-specific data and tracking key factors such as PDGF, TGF-beta, ECM, SMC density, and drug concentration for ISR prediction. The results from this model complement the finite element modeling. 

Conclusion: This multidimensional model has significant potential for replicating the mechanisms of ISR while accounting for patient-specific factors. In the long run, it is expected to guide clinicians in developing tailored treatment plans and enhancing the effectiveness of clinical interventions.