The predictive value of a concise classification of left atrial appendage morphology to thrombosis in non‐valvular atrial fibrillation patients

Abstract Background The complexity of left atrial appendage (LAA) in patients with nonvalvular atrial fibrillation (NVAF) is closely related to LAA thrombosis and stroke incidence. But the classification of LAA morphology is not uniform and controversial. Hypothesis This study divided the LAA into two categories according to the LAA morphology to investigate the risk of thrombosis related to the LAA structural complexity in NVAF patients. Methods A total of 336 NVAF patients were enrolled continuously in this study. The patients were divided into thrombosis group and non‐thrombosis group according to whether the thrombus presence in LAA. Through computer LAA three‐dimensional reconstruction, LAA morphology was divided into the complex type and simple type according to with or without the clearly lobulated structure judged by imaging experts. The relationship between LAA thrombosis and various potential risk factors was analyzed. Results A total of 19 potential risk factors for LAA thrombosis in NVAF patients were enrolled into statistical analysis. The coincidence rate of LAA morphology classification was 96.4% (324/336) between two imaging experts. Multivariate logistic regression analysis showed that complex LAA morphology (OR 4.168, 95% CI 1.871‐9.288, P < .001) was associated with the presence of LAA thrombus, independently of other enrolled risks. Conclusions It is a concise and reliable method to divide the LAA morphology into complex type and simple type according to whether with the clearly lobulated structure. The complex LAA is an independent risk factor for LAA thrombosis in NVAF patients.


| General data
Patients were recruited through our institution's Computerized Patient Record System. The following patient characteristics were documented upon participant recruitment: gender; age; type of AF; duration of AF (accurate to the month); oral anticoagulant use before admission; blood pressure during hospitalization; the presence of coronary heart disease, hypertension, diabetes, or vascular disease (old myocardial infarction, peripheral arterial disease, or aortic plaque); a history of stroke/transient ischemic attack (TIA)/thromboembolism (TE); and presenting either with or without symptoms and signs of congestive heart failure (CHF).

| Biochemical criterion
The following biochemical criteria were recorded: fasting serum glucose, brain natriuretic peptide (BNP), creatinine (Scr), plasma fibrinogen (Fbg) concentration, and prothrombin time international normalized ratio (INR) of every patient. All of which were measured for the first time after admission to the hospital. Moreover, chest X-rays or computed tomography (CT) scans were used to determine whether aortic sclerosis was present. In addition, head CT or magnetic resonance imaging (MRI) was used to determine the presence or absence of cerebral infarction. Finally, coronary artery CT imaging or angiography was used to determine the degree of coronary artery stenosis, which aided in the diagnosis of coronary heart disease.

| Clinical grouping
Based on the results of TEE and left atrial CTA, which were employed to assess whether a patient had a thrombus, the patients were subsequently divided into one of two groups: the thrombus group or nonthrombus group. TEE is used as a "gold standard" for the diagnosis of LAA in patients with thrombus. 10

| Acquisition and classification of LAA morphology
The three-dimensional reconstruction of the left atrial CTA image was performed by CT image post-processing system to obtain the LAA morphology of each patient; the image was classified based on the LAA's morphological characteristics. In this study, the LAA morphology was divided into two categories: simple-LAA and complex-LAA.

| Statistical analysis
Statistical analysis was performed using SPSS 19.0 software. Each group of variables was tested for normality and homogeneity of variance. The measurement data conformed to be the normal distribution as mean ± SD; the univariate analysis used two independent sample t tests. If the measurement data did not conform to be the normal distribution, the median was used. The rank-sum test was used for comparison between groups. Counting data use cases and percentage (%), the comparison between groups using χ 2 tests. Multivariate logistic regression analysis was used to investigate independent risk factors for LAA thrombosis. The raw data of AF course and BNP does not conform to the normal distribution, the natural logarithm transformation for these parameters were performed before logistic regression analysis.

| Classification of LAA morphology
Among 336 patients, 162 (48.2%) cases had complex LAA morphology. The coincidence rate of LAA morphology classification was 96.4% (324/336) between two imaging experts, only 12 (3.6%) cases need the third expert to participate in the judgment.
According to Di Biase's LAA morphological categories, 5 the proportion of chicken wing-like ( Figure 1A (Table 1).
Among the four types LAA patients, the lowest incidence of LAA thrombus is the chicken wing-like LAA group (9.0%), and that the highest is the cauliflower-like LAA group (22.0%; Table 2). The incidence of LAA thrombus in chicken wing-like LAA patients is markedly lower than that in non-chicken wing-like LAA patients (9.0% vs 17.7,

| Comparison among groups
As shown in Table 3, complex-LAA accounted for 65.2% and 45.5%, respectively, in thrombus group and non-thrombus group (P = .013).

| Risk factors associated with thrombosis in LAA
Univariate analysis showed that there were three factors related to LAA thrombosis, such as AF course, LAd, NPAF, complex-LAA, and LVEF (Table S1) This study divided the LAA into two categories (simple-LAA and complex-LAA), and compared with the four classifications of Di Biase et al. 5 As shown in Table 1  In previous studies, cauliflower-like LAA was regarded as the most complex LAA structure. This form of LAA has a short overall length, a large number of lobulated structures, and has been confirmed as an independent risk factor for stroke/TIA and LAA thrombosis by several studies. 5,7 In comparison, chicken wing-like LAA is the simplest structure, and the risk of stroke/TIA and LAA thrombosis is low. 5 If the analysis is based on the number of LAA lobes, then it can be said that the higher the number of lobes, the greater the proportion of LAA thrombosis. 6 In this study, simple-LAA was defined as an The reason why complex-LAA is more likely to form a thrombus is related to LAA's lobulated structure, as well as its abundance of trabecular and lower LAA blood-flow rate. Khurram et al 12 found that cauliflower-like LAA has extensive trabecular formation, and extensive LAA trabeculae are independently associated with a history of TIA/stroke in AF patients. LAA morphology is an independent determinant of LAA flow rate, and LAA flow rates in patients with chicken wing-like LAA are significantly higher than those with cactus-like or cauliflower-like LAA. 13 The caecum and lobulated structure of the LAA, as well as the uneven trabecular inside the LAA can slow down the blood-flow rate and create a vortex, which may easily lead to blood deposition and thrombosis development.
During AF attack, the active systolic and diastolic functions of LAA were inhibited, and the emptying of LAA was reduced, which further promoted the occurrence of hypercoagulable state and thrombosis.
The results of this study also showed that the course of AF among those patients in the thrombosis group was significantly longer than that in the non-thrombosis group [30 (1-180) months vs 22 (1-120) months, P < .001), and the proportion of NPAF was significantly higher than that in the non-thrombotic group (76.1% vs 20.3%, P < .01). The two factors were independent risk factors for thrombosis in LAA, which was consistent with the findings of previous studies. 14-16

| STUDY LIMITATIONS
In this study, of all 19 potential risk factors, only AF course, NPAF, and complex-LAA were independent predictors of LAA thrombosis.
However, several high-risk independent risk factors of stroke proved by a large number of previous studies, such as heart failure, high-risk CHA2DS2-VASc score, and shock/TIA/TE history, were not risk factors of LAA thrombosis according to the results of this study. The main reason may be that the patients included in this study are limited to these with radiofrequency ablation indications (except for LAA thrombosis) in a single center, and only 336 patients enrolled, which cannot enough to represent the whole NVAF population.

| CONCLUSIONS
In conclusion, it is a concise and reliable method to divide the LAA morphology into complex type and simple type according to with or without the clearly lobulated structure. The complex LAA is an independent risk factor for LAA thrombosis in NVAF patients.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of this article.