Since its clinical introduction almost a decade ago, transesophageal echocardiography (TEE) has been widely embraced by the medical community. This test allows superior anatomic and functional assessment of the atria and aorta. The very high accuracy of TEE for the identification of atrial thrombi [4] has been used to justify use of the test (in concert with short-term anticoagulation) as a guide for early cardioversion of atrial fibrillation [5].

In this issue, data from two large prospective studies examining the prognostic value of TEE in patients with chronic or paroxysmal atrial fibrillation are reported [6, 7]. These studies provide insights into the pathophysiology of thromboembolism in the population with atrial fibrillation and the potential clinical role of TEE in risk stratification.

The Stroke Prevention in Atrial Fibrillation (SPAF-III) investigators [6] report on a 382-patient subset of the prematurely terminated SPAF-III study [8]. The parent study investigated 1044 patients with nonvalvular atrial fibrillation who were at high risk for thromboembolism. Included patients were women older than 75 years of age and patients of any age and either sex who had hypertension, previous thromboembolism, left ventricular dysfunction, or clinical congestive heart failure.

Transesophageal echocardiography was done within 3 months (mean, 21 days) after patients were randomly assigned to adjusted-dose warfarin (target international normalized ratio, 2.0 to 3.0) or combination therapy with low-dose warfarin (target international normalized ratio, 1.2 to 1.5) and aspirin. Dense spontaneous echocardiographic contrast, found in 20% of patients, almost tripled the risk for stroke. A trend toward benefit of adjusted-dose warfarin was seen (stroke rate, 4.5% per year compared with 18.2% per year in patients receiving combination therapy; P = 0.09).

Thrombi were noted in 10% of all patients, were strongly associated with dense spontaneous echocardiographic contrast, were confined to or involved the left atrial appendage, and tripled the risk for stroke. Thrombi were seen with similar frequency in both treatment groups when TEE was done within 2 weeks of randomization, but the incidence in the adjusted-dose warfarin group was much lower when TEE was done later (4% after 2 weeks of adjusted-dose warfarin therapy compared with 15% after 2 weeks of combination therapy; P = 0.004). Among patients receiving adjusted-dose warfarin, absence of thrombus at the time of TEE predicted a low rate of stroke (2.3% per year); in contrast, the rate was almost 18% per year when thrombus was present (P = 0.006). Transesophageal echocardiographic evidence of complex atherosclerotic aortic plaque was seen in 35% of patients, was equally prevalent in the two treatment groups, and was associated with an increased risk for stroke (15.8% compared with 4.0% in the adjusted-dose warfarin group; P = 0.02). Multivariate analysis identified dense spontaneous echocardiographic contrast and complex aortic plaque as independent TEE predictors of thromboembolism.

The underlying role of atrial thrombi in the pathophysiology of thromboembolism in the population with atrial fibrillation is highlighted by TEE studies showing that almost 45% of patients with thromboembolism and atrial fibrillation have residual left atrial thrombi [9]. Patients presenting with TEE evidence of atrial thrombi are at high risk for subsequent thromboembolism [10]. In 1960, Goldman [11] suggested the subsequently widely accepted hypothesis that the mechanism of warfarin's benefit in the prevention of thromboembolism is promotion of thrombus organization and endothelialization to the atrial wall. Among patients with nonvalvular atrial fibrillation who are referred for TEE-guided cardioversion, however, thrombus resolution has been reported in 50% to 80% of patients after 1 month of warfarin therapy [12]. The SPAF-III data, which show reduced incidence of atrial thrombi in patients undergoing TEE after 3 weeks of adjusted-dose warfarin therapy (compared with patients receiving combination therapy), extend these data by identifying resolution of atrial thrombus or inhibition of new thrombus formation as a primary beneficial mechanism of adjusted-dose warfarin for patients with chronic or recurrent atrial fibrillation.

Complex aortic plaque has been recognized as a risk factor for stroke in the general population [13], but its significant relation to thromboembolism in the population with atrial fibrillation is a major finding of this study. In this clinically defined high-risk population, the lack of complex plaque on TEE conveyed a favorable prognosis (stroke rate, 1.1% to 1.2% per year), regardless of treatment. However, the authors do not explain the mechanism of benefit of adjusted-dose warfarin in patients with complex plaque. Unlike atrial thrombi, complex plaque was equally prevalent in both treatment groups. This finding suggests that short-term resolution of plaque is unlikely, but it does not exclude the possible mechanism of "long"-term promotion of the resolution or inhibition of plaque. Serial TEE studies in this population may help elucidate the mechanism of warfarin's benefit. This study also casts doubt on the use of aggressive maintenance of sinus rhythm (by pharmacologic, mechanical [such as the Maze procedure [14], or electrical [such as implantable atrial defibrillators] means) or the removal or obliteration of atrial appendage [15] as primary therapy for prevention of thromboembolic complications because these strategies should not affect complex aortic plaque.

The SPAF-III substudy [6] also highlights the urgent need for improved strategies for patients at highest risk for thromboembolism. Even with use of adjusted-dose warfarin, patients with complex plaque have a risk for thromboembolism of 12% per year; the risk exceeds 20% per year with coexistent left atrial thrombi or spontaneous echocardiographic contrast. Conversely, the 32% of "clinically high-risk" patients who lack both of these TEE findings had a low risk for thromboembolism (1.3% per year) regardless of treatment. In this group, aspirin may be sufficient.

Also in this issue, Stollberger and colleagues [7] report observational data on 409 patients with nonrheumatic atrial fibrillation who underwent TEE and were followed for clinical thromboembolism for almost 5 years. This was not a randomized study of treatment strategies for atrial fibrillation. Warfarin was recommended only if a "definite" left atrial thrombus was identified by TEE, but 30% of patients did not receive the drug. These 399 patients presumably received single-dose aspirin. Although left atrial thrombi and an enlarged left atrial appendage were associated with stroke or embolism, no TEE index was identified as an independent risk factor for thromboembolism. This may have been because the authors' criteria for a definite left atrial thrombus (including length 15 mm and independent mobility) were relatively strict. Exclusion of the group with "probable" thrombus is also inconsistent with the clinical definitions used by others [4-6]. Stroke or embolism occurred in 30% of patients with definite left atrial thrombi and 20% of patients with probable thrombi. Unfortunately, data on aortic plaque were not collected.

How are we to reconcile the apparent discrepancies between these two studies? It is important to note the differences in the study samples and primary echocardiographic findings. The SPAF-III study specifically included patients at high risk for thromboembolism, whereas Stollberger and colleagues examined a relatively low-risk group (fewer of these patients were female, and the patients were younger, had more intermittent atrial fibrillation, and had not had stroke within 1 year of study entry). These differences may partly explain why Stollberger and colleagues reported a low incidence of atrial thrombi (2.5% compared with 10% in the SPAF-III substudy), a very low incidence of spontaneous echocardiographic contrast (12% compared with 63% in the SPAF-III substudy), and a lower overall event rate (3.0% per year compared with 7.8% per year in the combination therapy group in the SPAF-III substudy). Stollberger and coworkers also found no association between spontaneous echocardiographic contrast and left atrial thrombi, a result in stark contrast to those of many other TEE studies [5, 16-18]. Although the older TEE technology (monoplane and biplane imaging) and the stricter criteria used by Stollberger and colleagues may partly explain the lower incidence of atrial thrombus, these factors do not explain the fivefold lower incidence of spontaneous echocardiographic contrast.

Does everyone with atrial fibrillation need TEE, or is the test simply an expensive imaging toy? As TEE is increasingly being accepted as a safe [5] and cost-effective [19] clinical tool for guiding early cardioversion, information on dense spontaneous echocardiographic contrast and complex aortic plaque (both of which are independent of anticoagulation status [20]) are now readily available for many patients. The SPAF-III data suggest that risk shown by TEE may supersede clinical risk with regard to clinical management and prognosis. We interpret these data as suggesting that patients (regardless of clinical risk) with nonvalvular atrial fibrillation who lack dense spontaneous echocardiographic contrast, atrial thrombi, and complex aortic plaque have a good prognosis, can probably be treated with long-term aspirin therapy, and may benefit (from a thromboembolic perspective) from aggressive maintenance of sinus rhythm. All patients with dense spontaneous echocardiographic contrast, depressed atrial appendage function, complex aortic plaque, or atrial thrombi should be treated with adjusted-dose warfarin. Because patients who have high clinical risk and one or more adverse TEE findings remain at high risk for thromboembolism despite warfarin therapy, new, more optimal therapies are urgently needed. If these strategies can be validated in future clinical trials, atrial fibrillation will probably emerge as a leading indication for TEE and TEE will emerge as the standard of care for patients with atrial fibrillation.