Shadows on the Cave Wall: The Role of Transesophageal Echocardiography in Atrial Fibrillation

Atrial fibrillation, the most common arrhythmia requiring therapy, affects 0.4% of the general population and 2% to 5% of persons older than the age of 60 years [1-3]. Moreover, atrial fibrillation has become the focus of renewed interest and investigation. Current management strategies include cardioversion and antiarrhythmic drugs to maintain sinus rhythm and the alternative approach of rate control. In the latter, the ventricular response is modified by the use of dromotropic drugs, catheter ablation, or radiofrequency modification of the atrioventricular node. In both strategies, anticoagulation or platelet inhibitor therapy is used for preventing embolic stroke. The use of anticoagulation to manage patients having cardioversion is controversial and is germane to the role of transesophageal echocardiography in evaluating patients with atrial fibrillation and to the study by Manning and associates [4] in this issue. It is well established that a thromboembolic event may follow cardioversion within hours or days of the restoration of sinus rhythm [5, 6]. Nonrandomized studies suggest that the risk for embolism is reduced by the administration of anticoagulant agents for at least 3 weeks before cardioversion but that this therapy is necessary only in patients with atrial fibrillation lasting 72 hours or more [5, 6]. Manning and colleagues [7] previously found that among patients in whom no thrombus was seen on transesophageal echocardiography at hospital admission, cardioversion can be done safely without previous anticoagulation. A subsequent study [8] confirmed the anecdotal experience of many physicians: Emboli can occur in patients who have not received anticoagulants even though no thrombus had developed before cardioversion. A landmark Australian study by Fatkin and colleagues [9] provided a logical explanation for this finding: A delay may exist between the restoration of electrical sinus rhythm and the return of mechanical contraction, that is, atrial stunning. During this period, the stunned atrium may be the source of stasis, thrombus formation, and embolization after the resumption of atrial contractions [9]. What about the patient with atrial fibrillation lasting less than 72 hours? A study from Beth Israel Hospital showed that the duration of stunning is a function of the duration of atrial fibrillation before cardioversion [10]. Presumably, therefore, patients with atrial fibrillation of less than 72 hours' duration might be at less risk for stunning and for developing emboli after cardioversion. Nonetheless, a recent publication from Stoddard and colleagues [11] showed an appreciable incidence of thrombus (14%) in patients with atrial fibrillation lasting 48 hours or less. Given the above findings, is it reasonable to expect that transesophageal echocardiography will help guide the use of anticoagulation before and after cardioversion? The data on atrial stunning and the potential for thromboembolism after cardioversion strongly support the use of anticoagulation after successful cardioversion in all patients without contraindications. The optimal duration of anticoagulation is unknown; however, a recent study showed that after a median 4-week duration of warfarin therapy, 89% of atrial thrombi had completely resolved [12]. In one study [13], 95% of 233 cardioverted patients in whom atrial thrombi were not seen on transesophageal echocardiography and who were not receiving anticoagulants were successfully cardioverted to sinus rhythm without clinical thromboembolism [13]. However, this study did not prove that a transesophageal echocardiography-guided approach, when used in conjunction with systemic heparin, is safer than the conventional strategy of delayed cardioversion after several weeks of anticoagulation therapy. The authors pointed out that a series of more than 7000 patients would be required to directly compare the two approaches with reasonable statistical validity. Although anticoagulation after cardioversion is unavoidable, the use of transesophageal echocardiography to look for thrombi may prevent a second hospital admission for cardioversion, as is conventionally done. Moreover, the shorter the duration of atrial fibrillation, the more likely cardioversion will be successfulthat is, atrial fibrillation begets atrial fibrillation [14]. One approach [13] is to initiate heparin therapy on hospital admission, do transesophageal echocardiography, and, if no thrombus is seen, proceed with cardioversion within the next 24 hours, followed by anticoagulation for an as-yet undetermined period. In patients with a thrombus, at least 4 weeks of anticoagulation is required; a second transesophageal echocardiogram may be needed to assess thrombus resolution. This may be the safest approach, but at what cost? Crucial to the role of transesophageal echocardiography is unequivocal documentation of the accuracy of the technique for detecting thrombi. In this respect, the study by Manning and colleagues [4] in this issue is important and encouraging. Although conventional transthoracic two-dimensional echocardiography has been useful for detecting thrombi in the left ventricle, imaging of thrombi in the left atrium has been problematic [15]. Early reports [16, 17] noted the presence of a thrombus in some patients, but the sensitivity of transthoracic echocardiography for detecting thrombi in the body of the left atrium (75%) and particularly the appendage (approximately 0%) has been disappointing [18]. Unfortunately, thrombi in the left atrial appendage are most likely to account for embolic stroke [11, 19, 20]. Although transesophageal echocardiography was first used in 1976 [20], its popularity has grown enormously in the past decade with the introduction of high-resolution, reliable instrumentation. The ability of this technique to visualize thrombi in the left atrium and the left atrial appendage [22] contributed to its burgeoning use for excluding a source of embolus. The metaphor of shadows on the cave wall, used by Plato to describe the separation of a presumed reality from that which can be perceived by the senses, is particularly appropriate to cardiac ultrasound. Echocardiography is a highly operator-dependent technique for which substantial experience is required to obtain high-quality images and to distinguish artifact from reality. Moreover, continuous translation and rotation of the heart throughout the cardiac cycle may convert this technical advantage of real-time tomographic imaging into a liability. This may be particularly likely in an anxious patient during esophageal intubation if heart rate and inotropy are increased. Even substantial thrombi may be easily missed unless the ultrasound beam is successfully directed through the diagnostic target. The left atrial appendix is a narrow tubular structure, and its size, shape, and orientation to the body of the left atrium vary. Therefore, complete ultrasound examination of this structure can be difficult, and failure can result in false-negative diagnoses. In contrast, imaging of longitudinal muscular ridges (pectinate muscles) in the appendage perpendicular to their long axis results in the appearance of circular structures that can be misinterpreted as small clots. Other potential sources of false-positive diagnoses include beam-width artifact, confounding images produced by normal structures as they rotate into the ultrasound beam, and spontaneous echo contrast (so-called swirl or smoke), which is often found in the presence of blood flow stasis and is a common epiphenomenon of intracardiac thrombus. Given these considerations, validation of transesophageal echocardiography for detecting thrombi in the left atrium and the left atrial appendage is particularly important. The report by Manning and colleagues [4] in this issue directly verifies the high sensitivity and specificity of transesophageal echocardiography for detecting thrombi in the left atrium and left atrial appendage. In their study, planned inspection of the atrium during surgery disclosed an atrial thrombus in 14 of 231 patients. Manning and colleagues found that intraoperative (preoperative in two patients) transesophageal echocardiography had a sensitivity of 100%, a specificity of 99%, and a positive predictive value of 86%. Their study confirmed an association among the presence of a left atrial thrombus, previous thromboembolism, use of warfarin in the month before surgery, and atrial fibrillation. However, mitral regurgitation was less severe in patients with thrombi; this finding is consistent with the protective effect provided by mitral regurgitation, which flushes out the left atrium and appendage with each systole, thereby preventing flow stasis. Some points that might affect the applicability of transesophageal echocardiography in widespread clinical practice should be considered. First, Manning and colleagues are not only highly skilled echocardiographers but also have a major investigative interest in left atrial thrombi and are particularly experienced in the use of transesophageal echocardiography for detecting such thrombi. Given some of the pitfalls of obtaining accurate echocardiographic images, it remains to be determined whether physicians in clinical practice can reproduce the exceptional accuracy of Manning and colleagues. Second, instrumentation may be important. The authors used multiplane or biplane probes for most patients and one type of ultrasonograph for all. Multiplane and biplane probes allow more thorough examination and possibly improved detection of thrombi [23] in the left atrium and the left atrial appendage than do monoplane probes, and it is not certain that the imaging characteristics of all types of ultrasonograph are equivalent. Third, all but two examinations were done under general anesthesia. Most patients in whom transesophageal echocardiography is done to detect atrial thrombi are sedated but conscious. The accuracy of t