We investigated the relation between left ventricular diastolic dysfunction and left atrial appendage (LAA) thrombus in patients with atrial fibrillation (AF). We performed transesophageal echocardiography to examine LAA thrombus or spontaneous echo contrast (SEC) and to measure LAA emptying flow velocity in consecutive 376 patients with AF. We estimated diastolic filling pressure as the ratio of early transmitral flow velocity (E) to mitral annular velocity (e′) on transthoracic echocardiogram. E/e′ ratio in 28 patients (7.4%) with LAA thrombi was higher than that in patients without thrombus (18.3 ± 9.3 vs 11.4 ± 5.9, p <0.0001). The fourth quartile of E/e′ (>13.6) consisted of 19 patients with thrombi and had a higher prevalence of thrombi than the others (p <0.0001). Multivariate regression analysis selected E/e′ ≥13 as an independent predictor of LAA thrombus with an odds ratio of 3.50 (1.22 to 10.61) in addition to LA dimension and ejection fraction. Increased quartile of E/e′ was negatively associated with LAA flow velocity and positively with rate of SEC. In conclusion, increased diastolic filling pressure is associated with a higher rate of LAA thrombus in AF, partly through blood stasis or impaired LAA function.
Left ventricular (LV) diastolic dysfunction is a common feature of risk factors for stroke in patients with atrial fibrillation (AF), but its relation to left atrial appendage (LAA) thrombus is not well elucidated. The ratio of early transmitral flow velocity (E) to mitral annulus velocity during diastole (e′) correlates closely to LV diastolic filling pressures. It is the most reliable echocardiographic index to detect LV diastolic dysfunction even in AF. In the present study, we investigated the relation between E/e′ and LAA thrombus in patients with AF. We also investigated the relation between E/e′ and LA spontaneous echo contrast (SEC) and LAA emptying flow velocity to clarify the mechanism that exacerbates LAA thrombus formation.
Methods
We enrolled consecutive 376 patients with nonrheumatic, paroxysmals or persistent AF from January 2006 to December 2007 who underwent transthoracic echocardiography and transesophageal echocardiography within a 30-day period. We excluded patients with mitral stenosis or previous mitral valve surgery. We also excluded patients with biventricular pacemaker or with New York Heart Association class IV heart failure because E/e′ might not properly estimate LV filling pressure in these patients. We confirmed presence of AF in each patient by ≥1 AF episode documented with electrocardiography. If a patient underwent multiple transthoracic or transesophageal echocardiographic examinations, we chose the earliest examination as an index study. We reviewed medical records of all patients to determine clinical features including age, gender, history of hypertension, diabetes, stroke, structural heart diseases, congestive heart failure, and state of anticoagulant therapy at time of transesophageal echocardiography. Presence of heart failure was determined based on medical records, patients’ symptoms, physical examinations, or medical tests. We calculated a CHADS2 score to evaluate the risk of cardiogenic stroke. The study protocol was approved by the hospital’s ethics committee. One investigator obtained informed consent from each patient before transthoracic or transesophageal echocardiography.
We performed transesophageal echocardiography using a SSD-6500 system (Aloka, Mitaka, Tokyo, Japan) equipped with a 5.0-MHz monoplane transesophageal probe. LAA was viewed in midesophageal short-axis and 2-chamber images. LAA thrombus was defined as a well-circumscribed, echocardiographically reflective mass that was of different texture than the atrial wall and that had a uniform consistency. SEC was defined as a dynamic smoke-like signal that swirled slowly in a circular pattern within the left atrium. LAA flow velocity profiles were obtained by pulse-wave Doppler echocardiography with sample volume placed at the LAA orifice. LAA peak emptying velocities were averaged with each RR interval over ≥5 consecutive cardiac cycles.
We performed transthoracic echocardiographic examination using a SONOS 7500 or iE-33 system (Philips Healthcare, Andover, Massachusetts). We recorded tissue Doppler images from the apical 4-chamber view and measured septal e′ (centimeters per second) on the pulse-wave Doppler spectrum as an average of 5 consecutive beats. We measured transmitral E (centimeters per second) as an average of 5 consecutive cardiac cycles and calculated an E/e′ ratio. We also measured LA diameter, LV end-diastolic dimension, and LV ejection fraction. LV mass (grams) was calculated as 0.80 × (1.04 × [{septal wall thickness in diastole + LV end-diastolic dimension + posterior wall thickness in diastole} 3 − LV end-diastolic dimension 3 ]) + 0.6 and indexed to body surface area as LV mass index. Relative wall thickness was calculated as 2 × (posterior wall thickness in diastole)/LV end-diastolic dimension. Mitral regurgitation was qualitatively graded as none, mild, moderate, or severe based on regurgitant jet area and spatial distribution of regurgitant flow.
All continuous variables are expressed as mean ± SD and were compared by 1-way analysis of variance. Significance of difference was calculated with the Tukey Honestly Significant Difference test for factor analysis. Categorical variables were compared with Fisher’s exact test. We performed logistic multivariate regression analysis to determine clinical and transthoracic echocardiographic factors for predicting LAA thrombus formation; clinical factors used for analysis were age, gender, history of hypertension, diabetes mellitus, previous stroke, congestive heart failure, structural heart disease, type of AF (paroxysmal or persistent), and state of anticoagulant therapy. Echocardiographic parameters were LA diameter, LV end-diastolic dimension, ejection fraction, LV mass index, relative wall thickness, presence of at least moderate mitral regurgitation, and E/e′ ≥13. We performed receiver operator curve analysis to determine the optimal cut-off value of E/e′ for predicting LAA thrombus. Differences were considered statistically significant at a p value <0.05 (2-sided). JMP 5.0.1 (SAS Institute, Cary, North Carolina) was used for statistical analysis.
Results
LAA thrombi were observed in 28 patients (7.4%) on transesophageal echocardiogram. Patients with LAA thrombi were significantly older and had higher prevalence of persistent AF and of structural heart disease than those without thrombus ( Table 1 ). More patients with thrombi were in AF at transesophageal echocardiography than were those without thrombus. Numbers of patients receiving warfarin were comparable between those with and without thrombi. International normalized ratio of prothrombin time showed no differences between the 2 subsets of those receiving warfarin. Incidences of hypertension and diabetes showed no differences between the 2 subsets. Previous stroke and heart failure were more frequently observed in patients with LAA thrombi. Therefore, patients with thrombi had significantly higher CHADS2 scores than those without thrombus.
| Variables | All Patients (n = 375) | LAA Thrombus | p Value | |
|---|---|---|---|---|
| No (n = 347) | Yes (n = 28) | |||
| Age (years) | 64 ± 11 | 63 ± 11 | 68 ± 10 | 0.03 |
| Men/women | 293/80 | 269/76 | 24/4 | 0.34 |
| Paroxysmal/persistent atrial fibrillation | 246/130 | 236/112 | 10/18 | 0.0006 |
| Atrial fibrillation during transesophageal echocardiography | 197 (52.4%) | 175 (50.3%) | 22 (78.6%) | 0.004 |
| Structural heart disease | 95 (25.3) | 79 (22.8) | 16 (57.1) | <0.0001 |
| Anticoagulation therapy | 218 (58.1%) | 202 (58.1%) | 16 (57.1%) | 0.93 |
| International normalized ratio of prothrombin time ⁎ | 1.9 ± 0.7 | 2.2 ± 0.9 | 1.9 ± 0.5 | 0.08 |
| Heart failure | 206 (54.7%) | 182 (52.3%) | 24 (85.7%) | 0.0006 |
| Hypertension | 177 (47.1%) | 161 (46.3%) | 16 (57.1%) | 0.27 |
| Diabetes mellitus | 59 (15.7%) | 52 (14.9%) | 7 (25.0%) | 0.16 |
| Previous stroke | 38 (10.1%) | 32 (9.2%) | 6 (21.4%) | 0.04 |
| CHADS2 score | 1.5 ± 1.3 | 1.5 ± 1.2 | 2.4 ± 1.4 | 0.0001 |
⁎ International normalized ratio of prothrombin time was an average value in patients receiving warfarin.
Transthoracic echocardiography was performed before or after transesophageal echocardiography with median of 6-day interval (25% to 75% quartile, 3 to 9 days). E/e′ was significantly higher in patients with LAA thrombi than in those without thrombus (18.3 ± 9.3 vs 11.4 ± 5.9, p <0.0001). Patients with LAA thrombi also had larger LA diameters and LV end-diastolic dimension, lower ejection fraction, and higher prevalence of at least moderate mitral regurgitation than those without thrombus ( Table 2 ). There were no significant differences in relative wall thickness and LV mass index.
| Variables | All Patients (n = 375) | LAA Thrombus | p Value | |
|---|---|---|---|---|
| No (n = 347) | Yes (n = 28) | |||
| Left atrial diameter (cm) | 3.8 ± 0.6 | 3.7 ± 0.6 | 4.4 ± 0.8 | <0.0001 |
| Left ventricular end-diastolic diameter (cm) | 49 ± 6 | 49 ± 6 | 54 ± 9 | <0.0001 |
| Ejection fraction (%) | 61 ± 14 | 62 ± 13 | 48 ± 20 | 0.0009 |
| Relative wall thickness | 0.43 ± 0.09 | 0.43 ± 0.08 | 0.42 ± 0.09 | 0.51 |
| Left ventricular mass index (g/m 2 ) | 120 ± 39 | 118 ± 37 | 153 ± 45 | <0.0001 |
| Mitral regurgitation at least moderate grade | 126 (33.5%) | 110 (31.6%) | 16 (57.1%) | 0.006 |
| Early transmitral flow velocity/mitral annular velocity (E/e′) | 11.9 ± 6.4 | 11.4 ± 5.9 | 18.3 ± 9.3 | <0.0001 |
| Spontaneous echo contrast | 120 (31.9%) | 96 (27.6%) | 24 (85.7%) | <0.0001 |
| Left atrial appendage flow velocity (cm/s) | 44.0 ± 20.0 | 44.8 ± 20.1 | 30.7 ± 12.4 | 0.002 |
We divided the study patients into 4 groups based on quartile of E/e′. E/e′ >15, a widely used estimate of increased LV filling pressure, was observed in 75 of 94 patients in the fourth quartile (E/e′ >13.6). Nineteen of 28 patients with LAA thrombi were in quartile 4, and quartile 4 had more patients with LAA thrombi than the other 3 groups (p <0.0001; Figure 1 ). However, only 19 of 94 patients (20.2%) in quartile 4 had LAA thrombi. Receiver operator curve analysis demonstrated that 12.8 was the optimal cut-off value of E/e′ for predicting LAA thrombus (area under the curve 0.77; Figure 2 ). Based on an equation by Nagueh et al, E/e′ equal to 12.8 corresponds to 20 mm Hg of LV filling pressure. E/e′ >12.8 predicted presence of LAA thrombus with 75.0% sensitivity and 74.4% specificity.
Using clinical and transthoracic echocardiographic parameters, we performed logistic multivariate analysis to determine the predictors for LAA thrombi. We used E/e′ ≥13 as a factor for analysis. LA diameter, LV ejection fraction, and E/e′ ≥13 were selected as significant predictors for LAA thrombus among clinical and echocardiographic parameters ( Table 3 ). E/e′ ≥13 had an odds ratio of 3.50 (95% confidence interval 1.22 to 10.61) for LAA thrombus.
| Wald Chi-square | p Value | |
|---|---|---|
| Ejection fraction | 7.97 | 0.005 |
| Early transmitral flow velocity/mitral annular velocity ≥13 | 5.26 | 0.02 |
| Left atrial diameter | 4.32 | 0.04 |
| Persistent atrial fibrillation | 3.06 | 0.08 |
| Left ventricular end-diastolic dimension | 2.07 | 0.15 |
| Heart failure | 2.17 | 0.14 |
| Mitral regurgitation at least moderate grade | 2.09 | 0.15 |
| Age | 1.01 | 0.32 |
| Gender | 0.75 | 0.39 |
| Left ventricular mass index | 0.62 | 0.43 |
| Relative wall thickness | 0.42 | 0.52 |
| Anticoagulation therapy | 0.51 | 0.48 |
| Atrial fibrillation during transesophageal echocardiography | 0.72 | 0.70 |
| Diabetes mellitus | 0.18 | 0.67 |
| Structural heart disease | 0.06 | 0.80 |
| Previous stroke | 0.06 | 0.80 |
| Hypertension | 0.007 | 0.93 |
We investigated relations between E/e′ and LAA emptying flow velocity. Patients with LAA thrombi had lower LAA flow velocity than those without thrombus (30.7 ± 12.4 vs 44.8 ± 20.1 cm/s, p = 0.002). E/e′ ratio was weakly but significantly correlated with LAA flow velocity (r = 0.2, p = 0.0002). Increased E/e′ quartile was significantly associated with decreased LAA flow velocity (p = 0.0005 by analysis of variance). Quartile 4 had significantly lower velocity than quartiles 1 and 2 (quartile 4 vs 1 37.2 ± 17.5 vs 49.7 ± 22.8 cm/s, p = 0.0002, vs quartile 2 45.3 ± 19.2 cm/s, p = 0.03, vs quartile 3 44.7 ± 18.7 cm/s, p = 0.13), although there were no significant differences among quartiles 1, 2, and 3 ( Figure 3 ). LAA flow velocity was also weakly correlated with LA dimension (r = 0.36, p <0.0001).
