Atrial fibrillation (AF) is one of the most common arrhythmias encountered in practice. According to the Global Burden of Disease Study 2019, the estimated prevalence of AF in adults is between 2% and 4%. Fifty million people worldwide are estimated to be affected (Joglar et al., 2024).

Thromboembolic events are a primary complication of AF, with nearly one-quarter of ischemic strokes potentially attributable to AF (Ding et al., 2023). Thromboembolic complications in AF patients arise because AF is associated with multiple thromboembolic risk factors, such as advanced age, hypertension, diabetes, coronary artery disease, and heart failure (Joglar et al., 2024; Kamel et al., 2016; Mac Grory et al., 2022; Reddy et al., 2022). More importantly, AF itself can lead to cardiogenic thrombus formation (Singer et al., 2021), with 91%−100% of these thrombi located in the left atrial appendage (LAA) (Di Biase et al., 2018). LAA thrombus (LAAT) significantly increases the risk of thromboembolism and is strongly correlated with outcomes of thromboembolic events and overall mortality (Heo et al., 2023; Hautmann et al., 2023; Zhang et al., 2024). Current practice guidelines place significant emphasis on assessing the risk of thromboembolic events in AF patients, particularly the risk of stroke, often overlooking the evaluation of the risk factors of LAAT (Joglar et al., 2024). As previously mentioned, not all thromboembolic events in AF patients are caused by cardiogenic thrombi; there may be differences in the prevention and treatment of thromboembolic events depending on the underlying mechanisms (Mac Grory et al., 2022). Conversely, assessing LAAT risk aids in more precise risk stratification, enhancing individualized treatment and research for AF patients.

As a primary source of thrombi in AF patients, the morphology of the LAA is closely related to the formation of LAAT. On the one hand, the morphology of the LAA can serve as a biomarker for atrial myopathy and various complications associated with AF, and it is thus related to LAAT. On the other hand, the morphology of the LAA itself may directly or indirectly contribute to thrombus formation. To better understand the relationship between LAA morphology and LAAT formation in AF patients, this paper will first provide a brief introduction to LAA anatomy, LAAT epidemiology, and LAAT diagnosis; then, a detailed discussion will be presented on the role of LAA morphology in the formation of LAAT and its ability to predict the risk of LAAT.

The LAA exhibits considerable variability in size, shape, and anatomical relationships with surrounding structures (Gonzalez-Casal et al., 2022). It is located laterally to the LA and anterior to the left pulmonary veins, separated by Warthin's ridge. It is positioned superiorly and anteriorly adjacent to the pulmonary artery and inferiorly adjacent to the left atrioventricular groove and the left circumflex branch. The left phrenic nerve may traverse it laterally.

The definitions of the LAA orifice opening vary, including the anatomical definition, the narrowest part of the LAA body, and Warthin's ridge-circumflex branch plane (Fang et al., 2022). Using the ridge-circumflex plane to define the LAA orifice may aid in standardizing the morphological research on the LAA (Hozawa et al., 2018). The shape of the LAA orifice can be classified as oval (68.9%), tubular (10%), triangular (7.7%), teardrop (7.7%), and round (5.7%) (Madaffari et al., 2020). The shape and size of the LAA orifice are influenced by various factors such as LA filling status and heart rhythm (Hozawa et al., 2018; Spencer et al., 2015).

Typically, the body of the LAA extends anteriorly and superiorly, with some patients exhibiting lateral or posterior extension and a few extending to the pericardial sinus. The average length of the LAA body is 45 mm (Madaffari et al., 2020). The endocardial thickness is uneven, and the atrial wall between the pectinate muscles is relatively thin (less than 1 mm). Thicker pectinate muscles seen on imaging might be misdiagnosed as LAAT (Gonzalez-Casal et al., 2022; Madaffari et al., 2020). The morphology of the LAA body is diverse, with multiple lobes possible. In an autopsy study of nearly 500 cases, 54% of patients had two lobes, 23% had three lobes, 20% had one lobe, and 3% had four or more lobes (Madaffari et al., 2020). Various classification methods have been proposed for LAA morphology based on lobes and characteristics. Due to its ability to predict stroke risk effectively and its guiding significance for LAA closure therapy, the classification method proposed by Wang et al. is currently commonly used, including chicken wing, windbag, cactus, and cauliflower types (Di Biase et al., 2012).

In addition to lobes and pectinate muscles, trabeculae are vital anatomical structures within the LAA that influence its function and local hemodynamics. Though often underappreciated in clinical imaging, recent computational studies have demonstrated their significant role in modulating intra-appendage blood flow and shear stress. Vella et al. showed that trabecular architecture affects flow dynamics and thrombogenic potential in AF (Vella et al., 2024). Likewise, Musotto et al. reported that trabeculae contribute to blood stasis by altering shear gradients and vortex formation (Musotto et al., 2024). These findings suggest that trabeculae should not be regarded as minor structural features but as relevant morphological factors impacting thromboembolic risk. Incorporating trabecular analysis may thus enhance the morphological assessment of the LAA.

In summary, the LAA is one of the most variable parts of cardiac anatomy, and it is highly diverse and irregular. Therefore, a unified, comprehensive, and objective description of LAA morphology is currently needed.

Due to the often-asymptomatic nature of LAAT before thrombus detachment, LAAT is frequently discovered incidentally during LAAT screening before AF electrical cardioversion or catheter treatment. As a result, the accurate prevalence of LAAT is challenging to estimate. The prevalence of LAAT varies significantly across different reports due to variations in patient characteristics. A meta-analysis focused on non-anticoagulated patients found that the prevalence of LAAT in individuals with non-valvular AF ranges from 5% to 27% (Azzalini et al., 2018). Other related studies have shown that despite receiving guideline-recommended anticoagulant therapy, the prevalence of LAAT remains as high as 2.73% to 7.2% (Mortensen et al., 2021; Pieszko et al., 2024). The studies mentioned above consider trans-esophageal echocardiography (TEE) the gold standard for detecting LAAT. However, due to its semi-invasive nature, large-scale screening is not feasible.

Additionally, these studies mainly included AF patients undergoing evaluation for catheter ablation or cardioversion, which introduces a selection bias. Additionally, the covert nature of LAAT and the invasiveness of TEE make it challenging to conduct prospective studies focused on LAAT as an endpoint. Therefore, current epidemiological research on LAAT primarily relies on cross-sectional or retrospective studies, which can provide prevalence data but need help to provide incidence information, posing a significant challenge for future research on LAAT.

TEE is currently the primary method for diagnosing LAAT. With the intraoperative confirmation of LAAT as the gold standard, TEE has a sensitivity of 93%−100% and specificity of 99%−100% for detecting LAAT (Azzalini et al., 2018). TEE can directly assess the presence of thrombus in the LAA and evaluate LAA morphology, blood flow velocity, emptying fraction, contraction ability, and spontaneous echo contrast, all closely related to LAAT risk. Spontaneous echo contrast is mainly considered a “pre-thrombotic state” for LAAT and, in some studies, has even been used as a surrogate marker for LAAT (Wang et al., 2023). Despite its numerous advantages, TEE is semi-invasive, which can lead to poor patient experience and potential complications such as esophageal perforation. Additionally, TEE has some relative contraindications, such as a history of swallowing difficulties, restricted neck mobility, esophageal varices, and coagulation disorders (Puchalski et al., 2019).

Due to the limitations of TEE, guidelines recommend cardiac-enhanced computed tomography (CT) with biphasic scanning as an alternative to TEE for ruling out LAAT before AF cardioversion or catheter treatment (Joglar et al., 2024). Using TEE results as the gold standard, cardiac-enhanced CT has a specificity and sensitivity of 89% and 95%, respectively, with single-phase scanning. In the biphasic CT, specificity and sensitivity improve to 100% and 99% (Yu et al., 2021). However, the significant radiation exposure associated with biphasic CT has led to studies suggesting that radiomics analysis based on single-phase CT images significantly improves the diagnostic accuracy for LAAT (Chun et al., 2021; Li et al., 2023). Nonetheless, current radiomics research has limited sample sizes and numerous influencing factors, such as scanning parameters, equipment, and parameter extraction algorithms, requiring validation (Yip and Aerts, 2016). Compared to TEE, cardiac-enhanced CT offers lower invasiveness, higher patient acceptance, and more comprehensive anatomical information but lacks assessment of LAA function and carries higher radiation risks and potential contrast-related complications.

Current research indicates that the sensitivity and specificity of cardiac MRI in diagnosing LAAT are comparable to those of enhanced CT. Additionally, advancements in new technologies are expected to improve diagnostic accuracy while reducing patient impact (Vira et al., 2019; Orbán et al., 2023). Intracardiac echocardiography provides a close-range and multi-angle examination of the LAA, achieves a 97% concordance with TEE in diagnosing LAAT, and can detect LAATs that TEE might miss (Srivatsa et al., 2014). It is an alternative for patients unsuitable for TEE or cardiac CT (Jingquan et al., 2022). However, due to its highly invasive nature and high cost, intracardiac echocardiography is primarily used as an adjunct tool during catheter-based treatments (de Leon et al., 2023).

Although the association between AF and thrombus formation has long been recognized, the mechanisms underlying the formation of LAAT are not fully understood. According to Virchow's triad for thrombus formation—blood stasis, morphological abnormality, and hypercoagulable states—this section will elucidate the mechanisms of LAAT formation and analyze the roles of LAA and LA morphology in this process.

Numerous studies have confirmed the association between the morphology of the LAA and stroke, as detailed in other reviews (Fang et al., 2022; Sun et al., 2023). However, studies examining the relationship between LAA morphology and LAAT are relatively limited, as summarized in Table 2. Unlike stroke and systemic embolic events, LAAT is asymptomatic and commonly diagnosed through invasive techniques such as TEE and CT scans, which involve X-ray radiation, making prospective studies challenging. All the studies listed in Table 2 are single-center, retrospective analyses. These studies predominantly include patients who underwent AF cardioversion or catheter ablation, leading to significant selection bias. Therefore, caution is required when interpreting these study results.

The findings from Table 2 indicate that indicators reflecting LAA size, including volume, LAA orifice diameter, LAA orifice area, and LAA depth, are the most common morphological parameters associated with LAAT. Notably, the correlation between LAA orifice size and LAAT remains contentious (Wang et al., 2023; Miki et al., 2022; Kiskaddon and Decker, 2023; Chen et al., 2017). This phenomenon may occur because LAA orifice size results from LA and LAA remodeling, which is positively correlated with the severity of remodeling and, consequently, with the risk of LAAT. Conversely, an enlarged LAA orifice may facilitate blood exchange between the LAA and LA, potentially alleviating blood stasis and acting as a protective factor against LAAT formation.

In addition to LAA size parameters, another category of LAAT-related parameters can be summarized as the complexity of the LAA, including LAA morphological classification, the number of LAA lobes, and the angle of the pectinate muscles (错误!未找到引用源。). As mentioned, the anatomical morphology of the LAA is highly variable, and these parameters are defined based on specific LAA characteristics, making it challenging to evaluate LAA comprehensively. Moreover, these parameters can be highly subjective; for instance, definitions of LAA morphological classification vary across studies (Chen et al., 2017; Suehiro et al., 2021; He et al., 2020). Even with standardized definitions, there are significant inter- and intra-researcher variations in LAA morphological classification and lobe count (Taina et al., 2014). Despite these limitations, research summarized in Table 2 generally suggests that more complex LAA morphology correlates with a higher risk of LAAT, such as cauliflower-shaped LAA, increased lobation, and larger pectinate muscle angles relative to the main lobe of the LAA. Recently, Lei and colleagues employed fractal analysis to quantitatively assess the complexity of LAA morphology based on surface curvature and roughness—referred to as fractal dimension—and found it to be significantly correlated with LAAT and stroke, with predictive power superior to the CHA2DS2-VASc score (Lei et al., 2022). Although the fractal dimension provides a more objective and comprehensive assessment of LAA morphology compared to previous parameters, it only evaluates the complexity of LAA morphology, not its overall shape.

Additionally, this study only assessed the fractal dimension, overlooking the role of the LA, while the interaction between the LA and LAA is an essential factor in LAAT formation (Fang et al., 2021). Recently, SSM has gained wide application and achieved significant results in the medical field (Bonsel et al., 2022; Corral Acero et al., 2022). SSM constructs shape space vectors through a series of landmark points, mathematically describing shapes, thereby converting shape data into statistical models. This method offers a comprehensive, objective, and quantitative description of shape. Bieging and colleagues utilized cardiac MRI data to create an SSM for the LA and LAA, enhancing stroke prediction capabilities when combined with the CHA2DS2-VASc score (Bieging et al., 2021). Therefore, SSM holds promise as a new method for comprehensive and objective evaluation of the overall morphology of the LA and LAA, offering new insights into the role of LA and LAA morphology in LAAT formation. The LAA is the primary source for thromboembolism in AF. The morphology of the LAA plays a significant role in blood stasis, morphological changes, and hypercoagulable states, which are directly or indirectly related to thrombus formation. Accurate evaluation of the morphology of the LAA can assist with risk stratification in patients with AF. Currently, the commonly used LAA morphological evaluation indicators are insufficiently comprehensive and objective. Recently, new imaging protocols have allowed for LA morphological remodeling and fibrosis assessment, which has been demonstrated to correlate with assessing the individual's risks of thromboembolic events and the practical imaging of patients with LAAT.