Study protocol

We performed transthoracic echocardiography (TTE) and transesopageal echocardiography (TEE) in 300 consecutive patients with warfarin therapy that had persistent AF for more than one month to evaluate intracardiac thrombus, the severity of aortic or mitral valve stenosis or regurgitation or congenital heart disease. Persistent AF was documented electrocardiographically by Holter monitor. There were 37 patients excluded from the study because of mitral stenosis or moderate-to-severe mitral regurgitation or previous mitral valve operation. Three patients were excluded because of poor echocardiographic recordings. The final analysis included 260 patients with nonvalvular persistent AF that were already receiving warfarin. We divided those patients into two groups according to the presence or absence of LAA thrombus. Moreover, each group was divided into subgroups according to a CHADS2 score ≤1 (low CHADS2 subgroup with LAA thrombus and without LAA thrombus)[5]. All patients were on warfarin as an anticoagulant to prevent thromboembolism according to the Japanese Circulation Society guidelines for pharmacotherapy of AF[12]. The PT-INR in all patients was measured one month before and at the time of TTE and TEE. The average of the two PT-INR measurements was used for analysis. LAA thrombus was defined as an echo-dense mass of more than 2 mm in diameter attached to the LAA wall that could be distinguished from the surrounding endocardium or pectinate muscles. The experimental protocol was approved by the ethics committee of our institution and informed consent was obtained from all patients.

Echocardiography

TTE and TEE were performed using an ACUSON Sequoia 512 ultrasound system (Siemens, Mountain View, CA) to evaluate LAA structure and contractile fraction. We examined the differences in echocardiographic parameters between patients with and without LAA thrombus. The parameters were obtained according to the standards of the American Society of Echocardiography[13, 14]. For TEE, a 4-7 MHz multi-plane transducer with a 7.4 mm diameter pediatric probe was used to diminish patient discomfort as previously described[15]. A time-LAA volume curve and a time-LAA strain curve can be automatically and promptly provided during a single cardiac cycle by VVI with feature-tracking echocardiography using off-line software (Syngo Velocity Vector Imaging, Siemens, Mountain View, CA) (Figure1). We measured LAA volume by two-dimensional TEE three times at three different angles (50, 70 and 90 degrees) and used average values for further analysis to improve the accuracy of the LAA volume measurement, as we previously reported in detail[11]. Average values of three cardiac cycles [(maximum volume – minimum volume)/maximum volume × 100] were used as LAA emptying fraction (EF). LAA volumes were determined using Simpson’s method. Although an autopsy study reported that the LAA is usually a multilobed structure[16], Cardiac computed tomography or magnetic resonance tomography revealed that in 97% of patients, the LAA in vivo was shaped like a cactus (30%), chicken wing (48%) or windsock (19%)[6]. Since three-dimensional TEE showed that the shape of LAA was round throughout the cardiac cycle in the present study, it was considered reasonable to measure LAA volume using the disk summation method (Figure2). Furthermore, LAAPV, SEC score (grade 0 - 4), LA dimension (LAD) and left ventricular (LV) ejection fraction were measured by the same methods as previously reported[5, 11, 17]. Excellent reproducibility and reliability of the VVI measurements were demonstrated in previous studies[9–11].

Statistical analyses

Data for continuous variables were expressed as the mean ± one standard deviation. Categorical data were summarized as percentages and compared using a Chi-square test or a Fisher’s exact test as appropriate. Comparisons of echocardiographic parameters between main groups and between subgroups were performed by an unpaired Student’s t-test. Comparisons of SEC, CHADS2 score between groups were performed by a Mann–Whitney U-test. The optimal cutoff values for the determination of sensitivity and specificity of the echocardiographic parameters to determine LAA thrombus were obtained from receiver operating characteristic (ROC) curve analysis. Multivariate logistic regression analysis was performed to identify the independent determinants of LAA thrombus. Multivariate logistic regression analysis was performed using clinical variables with a p value ≤0.10 in univariate analysis to determine the independent predictors of LAA thrombus. A p-value <0.05 was considered significant. Statistical analyses were performed using Stat View version 5.0 (SAS Institute Inc, Cary, NC, USA).

Patient characteristics

Echocardiographic parameters

Determinants of LAA thrombus

Using an LAAEF of 21% as the cutoff value, ROC curve analysis showed that the sensitivity was 93%, the specificity was 96% and the area under the curve (AUC) was 0.97. Using an LAAPV cutoff value of 24 cm/sec, the sensitivity was 73%, the specificity was 75% and the AUC was 0.73 (Figure3). The sensitivities, specificities, positive predictive values and negative predictive values at the optimal cutoff values are listed in Table4.

LAA thrombus in patients with a low CHADS2 score

Thromboembolic events related to AF result in significant morbidity and mortality. However, the risk of thromboembolic events is quite variable[18–21]. A higher CHADS2 score[3–6] is associated with an increased risk of systemic embolism, bleeding and death in patients with AF receiving oral anticoagulants[22]. However, the CHADS2 score may not always be a good predictor of stroke risk, especially in intermediate risk patients and patients with relatively low scores[13]. A previous study reported that 10% of persistent AF patients had LAA thrombus confirmed by TEE prior to catheter ablation despite an average PT-INR of 1.9 ± 0.5, and 77% of the patients with LAA thrombus had a CHADS2 score ≤2[23]. In the present study, LAA thrombus was found in 15/140 (10.7%) patients with a CHADS2 score ≤1. Among all of the patients, LAA thrombus was found in 16.5%. This may be due to the fact that target PT-INR level was achieved only in the two thirds of the patients. However, this high rate of LAA thrombus in the present study was concordant with that in a previous study (15.6%)[24]. Some patients with AF that receiving warfarin and had a CHADS2 score ≤1 suffered from cerebral embolism because warfarin was not totally effective in preventing LAA thrombus, despite a PT-INR in the target range. Therefore, better risk stratification is needed to improve the selection of AF patients who require strong anticoagulant therapy to prevent stroke.

TEE is known as the most sensitive and specific tool to assess LAA contractile fraction and detect LAA thrombus[17, 23]. Daniel et al. suggested that TEE was associated with an acceptable low risk when performed by experienced operators under safe conditions[25]. In addition, we used a pediatric probe to minimize patient discomfort under safe conditions, as previously described[15]. Multivariate analysis in the present study showed that the association between LAA thrombus and LAA contractility was stronger than the association between LAA thrombus and age, hypertension and prior stroke. A recent TEE study demonstrated that increased LA volume and lower LV ejection fraction were significant predictors of LAA thrombus[24]. However, LAA motion was not considered in this recent study. The present study suggested that an LAAEF cutoff value of 21%, as measured by VVI echocardiography, should be used in a future prospective study to predict AF-related thrombus. Performing TEE in patients with a low CHADS2 score is controversial. However, our results suggest that evaluation of LAAEF by TEE is important, despite discomfort during the procedure because of the risk of cerebral infarction due to thromboembolism.

Study limitations

There are several limitations of the present study. First, our findings are based on an observational study with a relatively small number of patients, particularly in the patients with both AF and LAA thrombus. We cannot draw conclusions regarding long-term outcomes, because the present study was a cross-sectional study. A prospective study in a lager patient population is needed to validate the predictive value of LAAEF and LAA peak strain on long-term outcomes. The predictive accuracy of the selected cutoff value for LAAEF and strain needs to be tested prospectively in an independent population to confirm its ability to predict LAA thrombus. Second, it is possible that the incidence of LAA thrombus does not always predict the occurrence of thromboembolic events because thromboembolism may be affected by other factors such as age, hypertension and diabetes mellitus. Finally, we measured the PT-INR levels one month before and at the time of TTE and TEE and averaged the two measurements. However, we recognized that there could be disparity between the measurements, and average value may not provide an accurate index of blood coagulability during this one-month period. In general, accurate evaluation of the blood coagulability was not possible.