Background: Left atrial appendage (LAA) morphology, investigated by computed tomography and magnetic resonance imaging, has proved to relate to the risk of cerebrovascular events in patients with atrial fibrillation (AF). The aim of the present study was to assess reproducibility of transesophageal echocardiography (TEE) imaging in describing LAA morphology.
Methods: Two-hundred consecutive patients referred for TEE were enrolled. In the first group of 47 (23.5%) patients LAA morphology was analyzed by conventional TEE and described as ChickenWing, Windsock, Cactus or Cauliflower. In the second group of 153 (76.5%) patients, instead, a 3D-Xplane diagnostic algorithm was performed to stratify LAA morphology as linear (ChickenWing) or complex (Windsock/Cactus and Cauliflower). Interobserver variability within three independent readers was assessed in both groups of patients and stratified by operator's experience and training. In a subgroup of 19 (12.4%) patients, the agreement of LAA morphology description by 3D-Xplane diagnostic algorithm was compared to cardiac magnetic resonance.
Results: By conventional TEE the agreement among operators on LAA morphology classification was poor (ρ<0.13). The 3D-XPlane diagnostic algorithm, significantly increased interobserver agreement up to ρ=0.32 within all readers and up to ρ=0.82 among the experienced and specifically trained operators. LAA morphology description in this latter group provided strong agreement with cardiac magnetic resonance (up to ρ=0.77).
Conclusions: LAA morphology assessment is challenging by conventional TEE. To improve reproducibility, the use of the 3D-Xplane technique combined with a specific diagnostic algorithm and training of the operators is fundamental.