Abstract
Affecting nearly 50 million people globally, atrial fibrillation (AF) underpins almost one third of all ischaemic strokes, emanating primarily from the left atrial appendage (LAA). Current clinical scores such as CHA₂DS₂-VASc, while beneficial for high-risk AF patients, offer limited insight into the pro-thrombotic mechanisms of Virchow’s triad - blood stasis, endothelial damage, and hypercoagulability, essential for comprehending and mitigating thrombus formation risks. This Thesis aims to harness biophysical computational modelling to delve deeper into Virchow’s triad on a patient-spcific basis to improve our understanding of thrombogenesis under flow conditions and potentially supplement the CHA₂DS₂-VASc score.Through a series of studies combining reaction-diffusion-convection (RDC) equations for three key clotting proteins (thrombin, fibrinogen and fibrin) with computational fluid dynamics simulations, based on models from Cine magnetic resonance imaging (MRI) data, this work proposes a novel risk-stratification score, providing nuanced insights into patient thrombogenicity and laying the groundwork for personalised preventive strategies in AF- related stroke risk.
The first study involved assessing the numerical stability of RDC equations for thrombin, fibrinogen and fibrin using a semi-realistic 2D model of the left atrium (LA) and a finite- element solution approach. Three outcomes were obtained from this study: i) numerical stabilisation schemes were tested and integrated into the simulations to resolve instabilities, ii) a two-way viscosity coupling scheme was devised to mimic the jellification of fibrin as clotting occurs, and iii) the impact of LAA morphology was investigated using 2D LAA shapes derived from clinical literature, with results corroborating clinical findings of the “chicken wing” morphology having the lowest risk of thrombus formation, while the “broccoli” morphology having the highest risk.
The next investigation was presented in two parts. First a modelling pipeline was developed and presented in detail describing how patient Cine MRI data with high temporal resolution can be leveraged to generate 3D patient-specific model of the LA which conform to the realistic cardiac motion observed in the MRI data using a wall tracking algorithm. Boundary conditions were applied to temporally varying models to account for each aspect of Virchow’s triad, with CFD to assess blood stasis, the endothelial cell activation potential (ECAP) used to identify regions prone to thrombogenesis, and hypercoagulatbility and thrombus formation determined from solution of the RDC equations for the key clotting proteins. A novel risk stratification score based on clinical and modelling data (categorising risk as A = low, B = moderate and C = high) was proposed incorporating each aspect of Virchow’s triad with CHA₂DS₂-VASc. This methodology was then applied to a patient cohort (N=9) with image data acquired pre- and post-catheter ablation therapy for each patient (totalling 18 simulations). Key findings from this study suggested that simulation results corroborated clinical risk scores for some patients eg. Patient 1 with a CHA₂DS₂-VASc = 1 and low-risk profile of [A A A] and Patient 2 with CHA₂DS₂-VASc = 3 and a high-risk profile of [C C C]), while for others, such as Patient 3 and Patient 4, who both had a CHA₂DS₂-VASc = 0, indicated that there was a much higher risk of fibrin gel formation than the clinical risk score suggested with moderate to high-risk profiles of [C B C] and [B C C] determined from simulations, respectively. These results reaffirm that the simple CHA₂DS₂-VASc score struggles to provide quantifiable, mechanistic detail into patient thrombogenic risk. However, biophysical modelling shows potential in aiding to bridging this gap which may be a critical step towards improving treatment approaches for AF patients at risk of stroke.
The final study in this Thesis investigated the relationship between LA strain and fibrosis and if these metrics can be combined with the thrombus formation results. We hypothesise that deposition of rigid collagen (fibrosis) following catheter ablation therapy to restore regular rhythm comes at the expense of reduced myocardial contractility (strain). Through statistical analysis, we found i) an inverse relationship between strain and fibrosis which supports the hypothesis of low strain in regions of high fibrosis; ii) a decrease in strain after ablation in all patients imaged during regular sinus rhythm in both pre- and post-ablation scans (perhaps due to addition of fibrosis), while all patients imaged during AF in their pre- ablation scan saw an increase (due to restoration of regular rhythm); iii) a positive correlation of R2 = 0.578 between strain and LAA blood velocity, with higher strain (LA contractility) being associated with higher LAA velocities. iv) A low correlation was found between strain and thrombus formation (R2 below 0.3), despite evidence from literature that there may be a link between strain and reduction in stroke risk.
This Thesis provides a comprehensive analysis of literature, introducing and applying numerical tools for 2D and 3D flow and coagulation simulations, the development of an image-based pipeline, and a novel risk stratification scheme. Collectively, this foundational work helps to improve our understanding of intracardiac thrombus formation mechanisms in the expanding AF patient population, underscoring the potential of biophysical modelling in advancing personalised patient care.
| Date of Award | 1 Apr 2025 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Adelaide De Vecchi (Supervisor), Oleg Aslanidi (Supervisor) & Steven Williams (Supervisor) |