Joint analysis of in-silico haemodynamics and morphological parameters of the left atrial appendage

Abstract

Atrial fibrillation (AF), which is considered the most common human arrhythmia, drastically increases the risk of stroke. Around 99% of thrombi in non-valvular AF are formed in the left atrial appendage (LAA). The current guidelines for patients stratification and treatment depending on their thrombogenic risk is based on the CHA2DS2-VASc score that presents some limitations since only a small subset of factors are considered. Furthermore, the role of the LAA in the process of thrombus formation is not fully understood. Recent studies claimed that morphological variables and haemodynamics indices could be better predictors than the CHA2DS2-VASc score. The aim of this project is to perform a joint analysis with haemodynamics indices extracted from patient-specific simulations and morphological parameters of the left atrium (LA) and LAA in order to improve the stratification of those patients who suffer from AF and better tailor the treatment. A cohort of 71 non-valvular AF patients, provided by OLV Hospital, was analyzed. Geometries were obtained from 3D rotational angiography images and segmented. Computational fluid dynamics (CFD) of 38 geometries were carried out. Haemodynamics indices related with the risk of thrombus formation have been computed from CFD simulations such as endothelial cell activation potential (ECAP) or relative residence time (RRT), together with velocity and vorticity analysis. Moreover, morphologic and volumetric parameters, such as the length of the centreline, the maximum ostium diameter and the bending angle of the LAA, have been gather and analyzed. A statistical analysis was performed to analyse the differences between morphological and haemodynamics parameters between patients with and without transient ischemic attack (TIA) history. The results of the statistical tests and regression models of morphological parameters suggest that larger LAAs (e.g. larger ostium diameter), have a higher risk of thrombus formation. As for haemodynamics parameters, none of them could accurately identify TIA vs non-TIA patients. However, the Akaike information criterion (AIC) when combining morphological and haemodynamics parameters in a regression model reported a good predictor model for risk of thrombus formation was obtained.