Transesophageal echocardiography is a recently introduced, moderately invasive diagnostic imaging technique that allows superior visualization of posterior structures such as the left atrium and the atrial appendage. Transesophageal echocardiography is frequently done in the search for atrial thrombi among patients presenting with acute thromboembolism [8, 9]. It has also been used to exclude atrial thrombi before procedures that are considered to be associated with a higher risk for thromboembolism if atrial thrombi are present, such as percutaneous balloon mitral valvuloplasty [10, 11] and cardioversion [12, 13]. Although small surgical series [7, 14] have suggested that the positive predictive value of transesophageal echocardiography is high, no large, consecutive, intraoperative series have been reported. We determined the accuracy of transesophageal echocardiography for identifying and excluding left atrial thrombi by doing intraoperative transesophageal studies in patients who were having cardiac surgery and in whom direct visualization of the left atrial contents was planned.

Methods

Top Methods Results Discussion Conclusion Author & Article Info References

Patients

The study sample consisted of 231 consecutive adult patients who had transesophageal echocardiographic examination before elective cardiac surgery between 1 October 1989 and 25 August 1995. Visualization of the left atrial contents was planned as part of the surgical procedure. Patients included 128 women and 103 men (mean age ±SD, 64 ± 13 years). The indications for repair or replacement of the mitral valve included rheumatic mitral valve disease (86 patients), mitral valve prolapse (71 patients), ischemic mitral regurgitation (30 patients), prosthetic valve dysfunction (22 patients), endocarditis (11 patients), cardiomyopathy (3 patients), and radiation-induced mitral valve injury (1 patient). Seven additional patients were having excision of a left atrial myxoma. Eighty-three patients (36%) were experiencing atrial fibrillation at the time of transesophageal study, and 46 additional patients (20%) had a history of atrial fibrillation. Four patients were experiencing a paced rhythm and 144 patients, a sinus rhythm. Thirty-nine patients (17%) had a history of thromboembolism, and 106 patients (46%) were receiving warfarin during at least part of the month before surgery.

Transesophageal Echocardiography

Transesophageal echocardiography was done after induction of general anesthesia and sternotomy using a commercial 5.0-MHz monoplane (48 patients) or biplane (69 patients) probe or a 5.0/3.5-MHz multiplane probe (114 patients) and a Sonos 1000, 1500, or 2500 echocardiograph unit (Hewlett Packard Co., Medical Products Division, Andover, Massachusetts). The left atrial appendage was initially viewed in the horizontal (0-degree) plane with the tip of the probe slightly flexed; the probe was gently withdrawn until the bifurcation of the pulmonary artery was visualized. Initial imaging of the left atrial appendage in the vertical (90-degree) plane was followed by posterior and anterior rotation of the probe until the coronary sinus and aorta were visualized, respectively. Multiplane imaging was done at 0 degrees and 90 degrees as described above, followed by imaging of the left atrial appendage in the horizontal (0-degree) plane and 5- to 10-degree stepwise rotation of the imaging sector from 0 degrees to 180 degrees during continuous visualization of the appendage. An atrial thrombus was reported when at least one observer identified a well-circumscribed, echo-reflective mass that was of different texture than the atrial wall and that had a uniform consistency [15]. Spontaneous echo contrast (a marker of blood stasis) was defined as dynamic "smokelike" echoes within the atrium [16]. The extent of mitral regurgitation was assessed by color Doppler and graded on a scale of 0 (none) to 3 (severe). Images and Doppler data were recorded on videotape; independent decisions on the presence of a thrombus were made by an observer performing the study in the operating room and by a second observer who reviewed the videotape at a later date without knowledge of the first observer's conclusions. Both observers made decisions without knowledge of the patient's clinical history.

All observers were attending physicians with extensive training and experience in transesophageal echocardiography (> 150 studies) [17]. Nine observers were used, although five observers read 91% of the studies (WJM, 34%; CAW, 21%; PS, 14%; JMH, 14%; ADM, 8%). Transesophageal echocardiographic data on the presence or absence of a thrombus were initially withheld from the surgeon unless they were considered to affect patient care (for example, a mobile thrombus in the body of the left atrium). Two hundred twenty-nine patients were studied during surgery immediately before the initiation of cardiopulmonary bypass. Two patients had elective transesophageal echocardiographic study that showed left atrial appendage thrombus and that was followed within 1 week by mitral valve surgery.

Transthoracic Echocardiography

Data from conventional transthoracic echocardiographic examinations done within 30 days of surgery were available on 143 patients (62%). Two-dimensional imaging had been done using a Hewlett Packard Sonos 500, 1000, or 1500 combined imaging unit and a Doppler echocardiograph unit equipped with a 2.0- or 2.5-MHz phased-array transducer. The M-mode left atrial dimension was measured in the parasternal long-axis view. We considered ventricular systolic function to be abnormal if evidence suggested global or regional hypokinesis.

Surgical Confirmation

All patients had cardiotomy under cardiopulmonary bypass, which was done using moderate hypothermia (28 °C) and aortic arterial and bicaval venous cannulation under one period of continuous aortic cross-clamping. Cold crystalloid or blood-potassium cardioplegia was administered antegrade, retrograde, or in combination every 15 to 20 minutes. In most patients, the left atrium was then entered through an incision in the interatrial groove. In patients with left atrial myxomas, the left atrium was approached either through the right atrium and atrial septum or through a biatrial incision [18]. The surgeon then visually inspected the body of the atrium and the appendage. If a thrombus was identified, its size and site were recorded.

Statistical Analysis

All data are expressed as the mean ±SD. We compared categorical variables using the chi-square test with continuity (Yates) correction for small numbers (Statview II, Abacus Concepts, Berkeley, California). We compared continuous variables using analysis of variance. In all analyses, tests were two-tailed, and a P value of less than 0.05 was considered to be statistically significant. We calculated the sensitivity, specificity, positive and negative predictive values, and accuracy for the identification of left atrial thrombi by transesophageal echocardiography; we used surgical identification or exclusion as the reference standard.

The hospital Investigational Review Board approved the study.

Results

Top Methods Results Discussion Conclusion Author & Article Info References

Transesophageal Echocardiography

Transesophageal imaging of the body of the left atrium was done in all patients without complication. The left atrial appendage was evaluated in 229 patients. Two patients had previously had cardiac surgery, during which the appendage had been amputated. Transesophageal echocardiography identified 14 left atrial thrombi. Thrombus size ranged from 3 to 80 mm (7 patients had thrombi ranging from 3 to 10 mm, 4 had thrombi ranging from 11 to 20 mm, and 3 had thrombi that were greater than 20 mm). Eleven thrombi were confined to the atrial appendage (Figure 1, top left), 1 thrombus was located within the body of the atrium and extended into the appendage (Figure 1, top right), 1 thrombus was located entirely within the body of the left atrium (Figure 1 bottom left), and 1 thrombus was identified across the interatrial septum (Figure 1 bottom right). Three thrombi were identified with the use of the single-plane transesophageal probe. Five thrombi were identified during biplane probe examination, and 6 thrombi were seen with the use of the multiplane transesophageal probe. Ten of 11 thrombi identified with the biplane or multiplane probe were seen in both the horizontal (0-degree) imaging plane and in the vertical (90-degree) and intermediate imaging planes. One thrombus imaged during biplane examination was visualized only in the vertical plane Figure 2 and was not apparent in the horizontal (0-degree) plane. Two of 14 left atrial thrombi were identified by only one observer. In both of these cases, thrombi were apparent on transesophageal echocardiography in the distal appendage in patients with rheumatic mitral valve disease, atrial fibrillation, and severe left atrial spontaneous contrast. These thrombi included 1 thrombus observed during monoplane study and another during biplane study.

View larger version (120K): [in this window] [in a new window]   Figure 2. Transesophageal echocardiograms obtained during biplane examination. Panel A. No thrombus visualized in the horizontal (0-degree) imaging plane. Panel B. In the same patient, an 8-mm left atrial appendage thrombus was identified (white arrow) in the vertical (90-degree) imaging plane and was confirmed during direct atrial inspection. AO equals ascending aorta; LAA equals left atrial appendage.

View larger version (107K): [in this window] [in a new window]   Figure 1. Transesophageal echocardiograms. Panel A. A 20-mm thrombus (white arrow) confined to the atrial appendage. Panel B. A 14-mm thrombus (white arrow) along the lateral wall of the left atrium and extending into the atrial appendage in a patient with extensive spontaneous echo contrast. Panel C. A 20-mm thrombus (white arrow) confined to the posterior left atrium. Panel D. An 80-mm linear thrombus (white arrows) extending from the right atrium across the interatrial septum into the left atrium. LA = left atrium; LAA = left atrial appendage; RA = right atrium.

Anatomical Findings

Direct inspection of the left atrium and appendage confirmed atrial thrombi in 12 patients. All thrombi identified by both observers were confirmed at surgery, with the thrombus site and size corresponding to those seen on transesophageal examination. No additional thrombi were seen on direct atrial inspection. Neither thrombus seen by only a single observer on transesophageal echocardiography was confirmed during direct inspection of the atrium.

When compared with direct inspection of the left atrium, transesophageal echocardiography had a sensitivity of 100% (95% CI, 74% to 100%) and a specificity of 99% (CI, 97% to 99.9%). In this cohort, in which the prevalence of atrial thrombi was 5.2%, transesophageal echocardiographic study for identifying atrial thrombus had a positive predictive value of 86% (12/14), a negative predictive value of 100% (217/217), and an overall diagnostic accuracy of 99%. The clinical characteristics of the patients with surgically confirmed left atrial thrombi within the body and appendage are summarized in Table 1. All patients with atrial thrombi had a history of atrial fibrillation (either current or previous) and had been receiving warfarin during the month before surgery. Rheumatic mitral valve disease or prosthetic valve dysfunction was also more common in these patients than in those who did not have left atrial thrombi (P = 0.03). According to analysis of echocardiographic indices Table 1, patients with atrial thrombi had less severe mitral regurgitation, but no significant difference was seen in left atrial dimension, the prevalence of abnormal left ventricular systolic function, history of thromboembolism, or type (monoplane, biplane, or multiplane) of transesophageal probe used (P = 0.5).

We categorized patients according to whether they had a relatively high risk for atrial thrombi because of a history of clinical thromboembolism or the presence of atrial fibrillation and either rheumatic mitral valve disease or prosthetic valve dysfunction. On the basis of this subgrouping, 96 patients (42%) were identified as being at "high" clinical risk. All 11 thrombi confined to the left atrium were included in this high-risk group, for an overall prevalence of atrial thrombus of 11%. Both false-positives were also in this group. The sensitivity and specificity of transesophageal echocardiography for identifying left atrial thrombi in this subgroup were 100% and 98% (CI, 92% to 99.7%), respectively. The positive predictive value was 84% (11/13), and the negative predictive value was 100% (83/83). The clinical and echocardiographic characteristics among this high-risk group are shown in Table 1. When we compared patients with and those without atrial thrombi in the high-risk group, mitral regurgitation was significantly lower among the subset with atrial thrombi and the prevalence of left atrial spontaneous contrast was significantly higher. Heart rhythm, warfarin use, size of the left atrium, and abnormal left ventricular systolic function did not significantly differ between these two patient groups.

Discussion

Top Methods Results Discussion Conclusion Author & Article Info References

Transesophageal echocardiography offers superior visualization of posterior cardiac structures. Although transesophageal echocardiography has a presumably high accuracy for identifying left atrial thrombi and left atrial appendage thrombi, our large, prospective, near-simultaneous intraoperative series directly validates the high sensitivity and specificity of transesophageal echocardiography for the noninvasive identification or exclusion of left atrial thrombi. No thrombi were identified through direct inspection of the atrium in any patient in whom near-simultaneous transesophageal echocardiography had excluded a thrombus. All thrombi identified by both echocardiographers were confirmed on direct inspection of the atrium, whereas two appendage thrombi identified by only one independent observer were not confirmed during direct inspection. Both of these latter cases were seen in patients with atrial fibrillation, rheumatic mitral valve disease, and extensive spontaneous left atrial contrast.

Clinical use of transesophageal echocardiography has continued to expand as studies showing its utility in the diagnosis or exclusion of cardiovascular pathologic abnormalities have been reported. Among patients presenting with thromboembolism, transesophageal echocardiography is frequently done in the search for atrial thrombi [8, 9]. Transesophageal echocardiography has also been advocated as a tool to exclude atrial thrombi before procedures that are known to be associated with a higher risk for thromboembolism, such as percutaneous balloon mitral valvuloplasty [10, 11] and cardioversion [12, 13]. Although patients in these series are followed for clinical events, these conditions rarely lead to death or urgent surgery. Thus, confirmation of transesophageal findings with any "gold standard," such as direct visualization, has been lacking; in addition, such visualization will probably not be done because of ethical considerations.

Previous Transesophageal Echocardiographic Studies

Previous studies examining the ability of transesophageal echocardiography to identify left atrial thrombi have included small patient series or nonsimultaneous studies. Mugge and colleagues [14] reported their findings from a series of 75 patients in whom transesophageal echocardiography identified thrombi; only 12 of these patients were subsequently referred for cardiac surgery. Thrombi were confirmed in all 12 patients having surgery. Previous studies by Kronzon and colleagues [10] and by researchers from our hospital [11], in which transesophageal echocardiography was used to exclude atrial thrombi among patients referred for percutaneous balloon mitral valvuloplasty, also confirmed the presence of thrombi in the small subset of patients with thrombi identified by transesophageal echocardiography who were referred for surgery. These data show the high positive predictive value of transesophageal study in these populations but do not confirm sensitivity or specificity. Although specificity could be inferred because of the absence of thromboembolic events among patients without transesophageal echocardiographic evidence of thrombi, this assumption was not validated by direct inspection of atrial contents.

Olson and coworkers [19] described a series of 20 patients who had monoplane transesophageal echocardiographic study immediately before replacement or repair of the mitral valve. They identified atrial thrombi in 3 patients: 2 with a thrombus in the left atrium and 1 with a thrombus in the right atrium. All thrombi were confirmed at surgery, and no additional thrombi were identified. Aschenberg and associates [7] did nonoperative monoplane transesophageal studies in 21 consecutive patients who had presented for routine mitral valve replacement. Six patients had transesophageal echocardiographic evidence of left atrial thrombi, and all six thrombi were confirmed at surgery. No additional left atrial thrombi were seen on direct inspection of the atrium. More recently, Hwang and coworkers [20] reported the findings in a large group of patients with rheumatic mitral valve disease who had 3.75-MHz monoplane transesophageal echocardiographic study while awake 3 or fewer days before mitral valve surgery. No intraoperative studies were done, and most patients had atrial fibrillation and substantial mitral stenosis. Monoplane transesophageal echocardiography had a sensitivity of 93.3% and a specificity of 100% for identifying left atrial thrombi. One of two false-negative studies in their series involved an 8-mm thrombus located along the posterior wall of the left atrium. These investigators hypothesized that this thrombus may have been "newly formed" and thus developed during the interval between transesophageal study and surgery [20]. The second false-negative case was a mural thrombus that was seen on retrospective review of the videotape. The monoplane probe used by these investigators has been supplanted by biplane and multiplane technology, with resulting improved visualization of the left atrial appendage. In our series, one thrombus imaged during biplane examination was only seen during imaging at 90 degrees (vertical plane) and was not identified during horizontal (monoplane or 0-degree) scanning (Figure 2). In addition, the rheumatic population represents only a small percentage of the patients currently referred for transesophageal echocardiographic examination, and data from this population may not be applicable to all groups.

Transesophageal Echocardiography in the Present Study

Our study shows the high sensitivity and specificity of transesophageal echocardiography for the noninvasive identification of left atrial thrombi: Left atrial thrombi were excluded by transesophageal examination in 217 patients, none of whom were found to have evidence of left atrial thrombi on direct inspection. These data are particularly relevant because transesophageal echocardiography is used to exclude left atrial thrombi among patients referred for interventions in which thromboembolic complications are closely associated with the presence of atrial thrombi before the planned intervention [7, 11-13]. The many consecutive near-simultaneous studies in this series and the high specificity of transesophageal echocardiography, particularly in the high-risk subgroup, validated the use of this imaging tool for this purpose.

The 5.2% incidence of left atrial thrombi in our patients is much lower than the 29% reported by Aschenberg and colleagues [7] and the 14% reported by Hwang and coworkers [20]. This discrepancy is probably related to differences in the patient samples. Aschenberg and colleagues [7] and Hwang and coworkers [20] restricted their studies to patients with rheumatic mitral valve disease, a population previously described as having the highest risk for atrial thrombi [21]. Eleven percent of our patients with rheumatic mitral valve disease had evidence of left atrial thrombi, a value similar to that found by Hwang and coworkers [20]. The use of warfarin anticoagulation, previous thromboembolism, and cardiac rhythm also probably differed between the studies.

In our series, patients with left atrial thrombi were more likely to have a history of atrial fibrillation, a condition in which the lack of organized atrial mechanical activity leads to blood stasis and the development of atrial thrombi. The patients with left atrial thrombi also had an increased prevalence of previous thromboembolism and had received warfarin during the month before surgery. These factors are probably related because patients presenting with thromboembolism and atrial fibrillation are often treated with prolonged warfarin therapy. In our series, warfarin was routinely withheld 4 to 7 days before surgery.

The severity of mitral regurgitation was found to be significantly lower among the patients with atrial thrombi. This "protective" effect of mitral regurgitation (against the formation of left atrial thrombi) has been hypothesized as being related to the "mixing" of atrial contents as blood from the left ventricle is ejected into the atrium at high velocity during ventricular systole. This finding is similar to transesophageal echocardiographic data reported by Black and colleagues [22] and Castello and coworkers [16], although these investigators did not intraoperatively confirm atrial thrombi and the patient samples differed from ours. Increased severity of mitral regurgitation has also been found to protect against the subsequent development of thromboembolic events among patients with atrial fibrillation [23].

Our study was not designed to compare the accuracy of monoplane, biplane, and multiplane transesophageal echocardiography for identifying left atrial thrombi. Various probes were used in our study because of the evolution of transesophageal technology that occurred during the study period. The additional viewing planes afforded by biplane and multiplane imaging are generally considered to permit superior visualization of atrial structures [24], and we recommend the use of multiplane imaging whenever available. Because monoplane imaging was initially done in all cases, and one thrombus Figure 2 was not seen in this imaging plane but was identified later in the vertical (90-degree) imaging plane portion of the study, we can estimate the sensitivity and specificity for monoplane transesophageal echocardiography as 92% and 99%, respectively. These data are similar to those reported by Hwang and colleagues [20] for nonsimultaneous monoplane studies in the rheumatic population. Although we have not had the opportunity to image a left atrial thrombus only during acquisition of images in the intermediate imaging planes, others have reported this experience [24].

Study Limitations

Our surgeons were aware of an individual patient's clinical history of thromboembolism, mitral valve structure, history of warfarin use, and cardiac rhythm. Except for the patient with the mobile thrombus across the interatrial septum, the surgeons were unaware of the transesophageal echocardiographic data. We cannot, however, exclude an investigational bias due to the surgeons' belief or the previous probability that a thrombus would be present. Our surgeons took great care to examine the left atria and appendages for thrombi, but we cannot exclude the possibility that a small thrombus was removed during drainage of the left atrial contents or was not visualized on direct inspection. Our data may not be directly applicable to patients having transesophageal echocardiographic study for nonvalvular indications, such as screening before cardioversion of atrial fibrillation, or to patients with a suspected cardiac source of embolism. However, a direct validation study such as ours could probably not even be done in such a group because of ethical considerations. A significant minority of our cohort, however, had a relatively high risk for atrial thrombi, and the specificity of transesophageal echocardiography for this group was excellent. Our data are also not directly applicable for the evaluation of right atrial thrombi.

Conclusion

Top Methods Results Discussion Conclusion Author & Article Info References

In summary, this large, consecutive intraoperative series documents the high accuracy of transesophageal echocardiography for the noninvasive assessment of the left atrium for thrombi. These data strongly support its clinical use for both identifying and excluding left atrial thrombi.

Presented in part at the 1994 Annual Meeting of the American Heart Association, Dallas, Texas, 15 November 1994.