Shear-driven modelling of thrombus formation in type B aortic dissection

Background: Type B aortic dissection (TBAD) is a dangerous pathological condition with a high mortality rate. TBAD is initiated by an intimal tear that allows blood to flow between the aortic wall layers, causing them to separate. As a result, alongside the original aorta (true lumen), a false lumen (FL) develops. TBAD compromises the whole cardiovascular system, in the worst case resulting in complete aortic rupture. Clinical studies have shown that dilation and rupture of the FL are related to the failure of the FL to thrombose. Complete FL thrombosis has been found to improve the clinical outcomes of patients with chronic TBAD and is the desired outcome of any treatment. Partial FL thrombosis has been associated with late dissection-related deaths and the requirement for re-intervention, thus the level of FL thrombosis is dominant in classifying the risk of TBAD patients. Therefore, it is important to investigate and understand under which conditions complete thrombosis of the FL occurs. Method: Local FL hemodynamics play an essential role in thrombus formation and growth. In this study, we developed a simplified phenomenological model to predict FL thrombosis in TBAD under physiological flow conditions. Based on an existing shear-driven thrombosis model, a comprehensive model reduction study was performed to improve computational efficiency. The reduced model has been implemented in Ansys CFX and applied to a TBAD case following thoracic endovascular aortic repair (TEVAR) to test the model. Predicted thrombus formation based on post-TEVAR geometry at 1-month was compared to actual thrombus formation observed on a 3-year follow-up CT scan. Results: The predicted FL status is in excellent agreement with the 3-year follow-up scan, both in terms of thrombus location and total volume, thus validating the new model. The computational cost of the new model is significantly lower than the previous thrombus model, with an approximate 65% reduction in computational time. Such improvement means the new model is a significant step towards clinical applicability. Conclusion: The thrombosis model developed in this study is accurate and efficient at predicting FL thrombosis based on patient-specific data, and may assist clinicians in choosing individualized treatments in the future.