The LAA is a finger-like extension originating from the main body of LA . It has active systolic and diastolic functions, and its mechanical dysfunction may lead to blood flow stagnation and thrombosis . At present, the LAAFV measured by TEE is the most commonly used method to evaluate the mechanical function of the LAA. A large number of studies have shown that the risk of thrombosis is steadily increased with the decrease of LAAFV [19,20,21]. Handke et al.  conducted a TEE study on 500 patients with cerebral ischemia and found that the measurement of the LAAFV may be an important quantitative substitute parameter for evaluating the risk of left atrial thromboembolism. However, TEE is semi-invasive and requires higher personal experience of the operator and may cause complications, such as bleeding and perforation . There are certain checkups, such as combined esophageal stenosis. Therefore, looking for noninvasive examination indicators that can effectively predict the mechanical function of the LAA has important clinical significance. Especially during the period of COVID-19.
At present, research on the anatomy of the LAA is still rare, and there is no uniform standard for the classification of the LAA. With the rapid development of multi-slice spiral CT and 3D reconstruction technology, it is possible to observe heart anatomy and vascular structure in detail. According to CT or MRI images of the heart, Wang  et al. divided LAA into four categories: chicken wing type, weathercock type, cauliflower type and cactus type. Among them, the chicken wing type is the most common form. However, due to the complex structure of LAA, sometimes it will show different morphological characteristics when viewed from different angles.It has been reported in the literature that this classification method is subjective . Therefore, we refer to the relevant literature to divide LAA morphology into chicken wing type and non-chicken wing type. This classification can significantly reduce subjectivity. Previous studies have shown that LAA morphology is related to the LAAFV [25, 26]. Fukushima et al.  reported that compared with the cactus type and cauliflower type LAA, the chicken-wing type LAA had a significantly higher flow velocity, but there was no statistical difference compared with the weathercock type. However, the study only included patients with PaAF. Our study found that whether it is in PaAF group or PeAF group, the LAA morphology is closely related to its mechanical function. The LAAFV of chicken-wing AF is higher than that of non-chicken wing AF. The possible mechanisms are as follows: First, chicken-wing patients may have greater muscle mass to contract the LAA . Second, high left atrial pressure or low left atrial compliance may change LAA morphology.
Compared with the direct measurement after CT three-dimensional reconstruction of the LAA, the measurement method used in this study judges that the LAA orifice is relatively less subjective and the measurement repeatability is good . At present, the research on LAA orifice area and the mechanical function of LAA in patients with AF has not been reported in the literature. Our study showed that the increase of LAA orifice area in patients with NVAF is closely related to the decrease of LAAFV. Agmon et al.  reported that the emptying speed of the LAA of the normal population was negatively correlated with the diameter of the LAA orifice measured by TEE (r=-0.29, p-value = 0.002). According to the continuity equation, as the cross-sectional area increases, the flow velocity becomes slower. In the case of constant flow, as the cross-sectional area decreases, the flow velocity becomes faster. Our data showed that the relationship between the LAA orifice area and its flow velocity in NVAF patients also conforms to the same law, so LAA orifice area is an important factor in determining LAAFV. Studies have shown that LAAFV was significantly negatively correlated with the LAD [29, 30]. Similar to the results of these studies, we found that LAD is negatively correlated with LAAFV. The reasons may be as follows: With the progress of AF, LA gradually undergoes structural remodeling, which in turn leads to an increase in the inner diameter and pressure of LA. The compliance of LAA is greater than that of the LA and can regulate left atrial pressure. However, the increase in left atrial pressure can lead to an increase in the afterload of LAA and lead to the decrease of LAAFV . This can also explain the view that some scholars believe that LAD can predict the risk of stroke. However, similar to the study of Harada M et al. , the multiple linear regression analysis did not prove that LAD is an independent factor of the LAAFV. Therefore, this needs more research to confirm.
The current guidelines recommend that AF be divided into five categories: primary AF, PaAF, PeAF, long-term PeAF and permanent AF . Petersen et al.  found that from PaAF group, to PeAF, and then to long-term PeAF group, the LAAFV gradually decreased (51.4 ± 25.1 cm/s vs. 40.9 ± 16.3 cm / s vs. 29.7 ± 15.1 cm/s, p-value < 0.001); Multiple linear regression analysis found that AF type was an independent predictor of LAAFV. It is consistent with our research results, and the reasons may be as follows: First, during the TEE and TTE examinations in this study, patients in the PaAF group were required to maintain sinus rhythm. When the heart rhythm is AF, the emptying time of LA and LAA is shortened, so the volume of LA is relatively increased. The rapid and irregular electrical activity will weaken the contractility of LAA, resulting in the decrease of LAAFV. Secondly, PeAF usually has a longer course than PaAF, the electrical remodeling and fibrosis of LA are more serious, which is more likely to cause an increase in the load of the LAA and cause systolic dysfunction, leading to the LAAFV was lower in PeAF group than that in PaAF group. LVEF is a common indicator reflecting left ventricular function. Our study found that LAAFV is positively correlated with LVEF. However, the multiple linear regression analysis did not prove that LVEF is an independent predictor of the LAAFV. This is similar to the study of Kishima et al. . The reason may be that the population included in our study is mainly NVAF patients with normal LVEF. So far, studys on the relationship between LVMI and LAA are rare, and there seems to be some controversy [33, 35]. Our study showed that LAAFV was negatively correlated with LVMI, and the multiple linear regression analysis proved that LVMI is an independent factor of the LAAFV, which is similar to the study of Harada M et al. .
At present, the CHA2DS2-VASc score is the most commonly used index for clinical assessment of stroke risk stratification in NVAF patients, and is used to guide anticoagulation therapy . The 2016 ESC guidelines recommended that patients with CHA2DS2-VASc score ≥ 1 can choose anticoagulation therapy . However, previous studies have found that patients with a score of 0 are still at risk of ischemic stroke. Gage et al. found that NVAF patients with a CHA2DS2-VASc score of 0 have an annual stroke risk as high as 1.9% . Moreover, the score mainly focuses on clinical indicators, and does not pay attention to the influence of heart structure and function on thrombosis. Therefore, the scoring system is not sufficient to comprehensively assess the risk of stroke in patients. Our study showed that non-chicken wing LAA in patients with NVAF can cause the decrease of LAAFV, which may increase the risk of thrombosis. Our research on the morphology and mechanical function of LAA may provide additional clinical significance for stroke risk stratification in NVAF patients. However, this study still has some limitations. First, this study is a retrospective, single-center, small-sample study. As a retrospective study, it was regretted that we did not discuss LA volume in this study. It is hoped that there will be a larger cohort for further prospective studies in the future. Second, there is still no uniform standardthe for the LAA morphology. Some clinical studies divide it into four types, this study only classifies it into two types. This classification can minimize the difference between observers, but it may not be precise enough.