Study Selection: English-language publications on atherosclerosis of the thoracic aorta were selected.

Data Synthesis: During the last 6 years, the increasing use of transesophageal echocardiography has shown that atherosclerotic plaque in the thoracic aorta is a source of otherwise unexplained embolic events, including stroke, transient ischemic attack, and peripheral emboli. Retrospective studies have documented a strong independent association between larger lesions (4 mm to 5 mm) and previous embolic disease, and prospective studies have shown that patients with these lesions have a high risk for future events (in one study, the risk for stroke was 12%; in another, the risk for cerebral or peripheral events was 33% in a follow-up period of just 14 months). These lesions also pose a serious risk for embolization caused by manipulation of the aorta during catheterization, intra-aortic balloon-pump placement, and cannulation of the aorta for heart surgery. Pathologic examination has shown atherosclerotic plaque, often with superimposed thrombi that account for the mobile components seen on transesophageal echocardiography. The management of patients who have atherosclerotic lesions in the thoracic aorta has not been determined prospectively. However, anticoagulation may help prevent emboli, as it does for patients who have thrombi in other locations, such as the left atrium and the left ventricle.

Conclusions: Protruding atherosclerotic lesions in the thoracic aorta, often with superimposed mobile thrombi, are an important cause of embolic disease. Transesophageal echocardiography should be considered in the work-up of patients who have unexplained embolic events.

Modern clinical experience has indicated that as many as 40% of strokes are still of undetermined cause. The routine evaluation of patients with stroke and peripheral emboli has included the inspection of the carotid arteries and the heart; however, transthoracic echocardiography has had frustratingly negative results in most of these patients. The advent of transesophageal echocardiography has made possible the relatively noninvasive, clear visualization of the aortic arch and the descending thoracic aorta. This procedure can safely be done at the bedside, while patients are awake, and with a very low risk for complications [4].

In the search for a source of embolization, transesophageal echocardiography has revealed dramatic abnormalities of the thoracic aorta (Figure 1, Figure 2, and Figure 3). In 1990, we reported a new finding in three patients who had embolic disease [5]. In these patients, transesophageal echocardiography showed striking, protruding atheromas in the aortic arch with components that were seen to move independently with blood flow. Since 1990, much has been learned about atherosclerotic disease of the thoracic aorta and its relation to embolic events. In this article, we review the developments that have been made in this area during the previous 6 years, including important advances in pathology, clinical epidemiology, diagnosis, and management.

View larger version (125K): [in this window] [in a new window]   Figure 1. Transesophageal echocardiogram (horizontal, 0-degree view) of a normal aortic arch. The smooth intimal surface is shown.

View larger version (113K): [in this window] [in a new window]   Figure 2. Transesophageal echocardiogram (horizontal, 0-degree view) of an aortic arch with moderate atherosclerotic plaque (white arrows).

View larger version (139K): [in this window] [in a new window]   Figure 3. Transesophageal echocardiogram (horizontal, 0-degree view) of an aortic arch with large, multi-lobed, ulcerated protruding atheromas (arrows).

Methods

Top Methods Author & Article Info References

Information for this review was obtained from published articles that were located through review of the bibliographies of published papers and through MEDLINE searches. Information was also obtained through consultation with persons working in the field.

Observational Studies

In 1990, Pop and colleagues [6] did transesophageal echocardiography in 72 patients who presented with transient neurologic events. Transesophageal echocardiography was superior to transthoracic echocardiography for diagnosing possible intracardiac sources of embolization. Forty-four percent of the 72 patients had aortic atherosclerosis; however, this study had no control group and did not classify the degree of atherosclerosis. Therefore, only an association and not a causal relation could be established between aortic atherosclerosis and transient ischemic attack. In 1991, Karalis and coworkers [7] found intra-aortic atherosclerotic debris on transesophageal echocardiography in 38 of 556 patients (7%). The incidence of a history of embolic events in these 38 patients was 31%, and the incidence of previous embolization in patients who had mobile components to their atheromas was even higher (8 of 11 patients [73%]).

Two years later, Toyoda and associates [8] reported on 62 patients with embolic stroke who were studied with transesophageal echocardiography. Twenty-six of these patients (42%) had protruding atheromas in the aortic arch, and 52 had possible embolic sources other than these atheromas. Of the 10 patients who did not have an additional possible source of the embolism, 3 had protruding atheromas.

To evaluate the frequency of aortic plaque or thrombi that could embolize to the brain, Amarenco and coworkers [9] studied 12 patients who had cerebral infarction of undetermined cause. Transesophageal echocardiography showed an echogenic mass in the aortic arch in 6 of these patients. Lesions were pedunculated in 1 patient or broad based in 5 patients; each lesion had an irregular surface and an intraluminal protrusion ranging from 3 mm to 15 mm. Simons and associates [10] studied 8 patients who had peripheral embolization. In each of these patients, severe aortic disease with protruding or mobile atheromas was seen on transesophageal echocardiography. Using transesophageal echocardiography, Horowitz and colleagues [11] found mobile aortic plaques in 7 of 183 patients (4%) with brain ischemia.

Case-Control Study

A case–control study design was used to evaluate 122 patients who, for the purpose of finding a source of embolization, were referred for transesophageal echocardiography [12]. All of these patients had a history of unexplained stroke, transient ischemic attack, or peripheral embolization. One hundred twenty-two age- and sex-matched controls were randomly chosen from among 385 patients who had transesophageal echocardiography during the same period for indications other than embolic symptoms. Of the 122 case-patients, 33 (27%) had protruding atheromas. In the blinded, randomized matching of case-patients with controls, 32 of these 33 case-patients (97%) were matched with controls who did not have atheromas.

The odds ratio for the occurrence of symptoms in the presence of protruding atheromas was 3.2 (95% CI, 1.6 to 6.5; P < 0.001). Eleven case-patients had mobile components to their protruding atheromas. Strikingly, no control had a mobile lesion. The 33 symptomatic patients with protruding atheromas had a total of 43 embolic episodes. Thirty-three of these episodes were neurologic (21 strokes and 12 transient ischemic attacks), and 10 involved the peripheral circulation (1 was in the kidney, 1 was in the arm, and 1 was in the leg [8]). It was hypothesized that right brain emboli would be less likely to occur than left brain or peripheral emboli because atheromas located in the aortic arch are more often distal to the innominate artery. This was confirmed by the finding that only 8 embolic episodes (18%) involved the right brain and 31 (72%) involved the left brain or the peripheral circulation (P < 0.005).

Prospective Studies

All of the studies discussed above suggest that a causal relation exists between aortic atheromas and cerebral or peripheral embolization. However, because these studies were retrospective and subject to referral bias, the data do not reflect the true risk that these atheromas present. To avoid the limitations of retrospective analysis, the significance of the protruding atheromas in the thoracic aorta as a predictor of future events was investigated in a prospective study [13]. The risk for future vascular events in patients with protruding atheromas in the thoracic aorta, as seen on transesophageal echocardiography, was evaluated. In 1 year, 42 patients had protruding atheromas and no other source of emboli. These patients were selected from a total of 521 patients who had transesophageal echocardiography during the 1-year period. They were followed for as long as 2 years (mean follow-up period, 14 months) and were compared with controls who did not have atheromas and were matched for age, sex, and blood pressure. Of the 42 patients with atheromas, 14 (33%) had a total of 19 vascular events during the follow-up period (5 in the brain, 2 in the eye, 4 in the kidney, 1 in the bowel, and 7 in a lower extremity). Of the 42 controls, only 3 (7%) had vascular events (2 in the brain and 1 in the eye). Univariate analysis showed that only protruding atheromas were significantly correlated with vascular events (P = 0.003); multivariate analysis showed that only protruding atheromas independently predicted events (P = 0.005; odds ratio, 4.3 [CI, 1.2 to 15.0]).

This prospective study showed a trend toward increased mortality in patients with protruding atheromas (9 deaths [21%]) compared with controls (6 deaths [14%]), but this difference was not statistically significant. Among the 9 patients with atheromas who died, death was related to vascular events in 5 patients: 2 who had fatal strokes and 3 who had multiple-organ ischemia. The other 4 patients with atheromas died of causes that had no clear relation to their aortic atheromas (1 had sudden death, 1 died of congestive heart failure, 1 died of pulmonary embolus, and 1 died of bleeding complications after heart surgery). Among the 6 controls who died, death was due to nonembolic complications of heart surgery in 2 controls, to stroke in 1 control, to myocardial infarction in 1 control, to ventricular fibrillation caused by sarcoid heart disease in 1 control, and to stomach carcinoma in 1 control. Thus, the incidence of death related to vascular events was higher in patients with atheroma (5 of 9 patients with atheroma [56%] compared with 1 of 6 controls [17%]), although this trend did not reach statistical significance (P = 0.09).

An important group comprised the 14 patients (33%) with protruding atheroma who did not have a history of embolization before having transesophageal echocardiography. These patients had been referred for transesophageal echocardiography because of valve disease (n = 8), possible endocarditis (n = 4) (the study results were negative for vegetations), and coronary artery disease (n = 1) and to rule out aortic atheroma before cardiopulmonary bypass surgery (n = 1). Vascular events occurred during the follow-up period in 6 of these 14 patients (43%). One patient died after ischemia of the bowel, kidney, and lower extremities. In other patients, two embolic events involved the legs, two involved the eye, and one involved the brain. Univariate analysis identified protruding atheromas (P = 0.002) and diabetes (P = 0.04) as significant correlates of these future vascular events; multivariate analysis showed that only protruding atheromas independently predicted future events. Thus, the incidence of future vascular events (33%) in the patients with a history of embolic events (which may in itself be considered a risk factor) was not greater than the incidence of these events (43%) in the group with no history of embolization.

Recently, 331 patients 60 years of age or older who were consecutively admitted to the hospital with brain infarction were evaluated by transesophageal echocardiography and followed for 2 to 4 years [14]. The incidence of recurrent brain infarction was 11.9 per 100 person-years of follow-up in patients who had aortic plaque thickness of at least 4 mm; the risk for any vascular event was 26%. In contrast, the incidence of recurrent brain infarction per 100 person-years was 3.5 in those with aortic wall thicknesses of 1 mm to 3.9 mm and 2.8 in those with aortic wall thicknesses less than 1 mm. Multivariate analysis showed that the larger aortic plaques were independent predictors of recurrent brain infarction (relative risk, 3.8 [CI, 1.8 to 7.8]; P = 0.001). Larger plaques were also significantly associated with all vascular events (relative risk, 3.5 [CI, 2.1 to 5.9]; P < 0.001).

Taken together, these studies indicate that although protruding atheromas in the thoracic aorta may be a marker for diffuse atherosclerosis, they are also likely to be a source of emboli in patients with and patients without other potential sources of embolization, such as demonstrable carotid stenosis. This conclusion is even more plausible in patients who have carotid lesions that are on the contralateral side and thus cannot be responsible for neurologic symptoms.

Association of Protruding Atheromas in the Aortic Arch with Carotid Artery Disease

Because the studies discussed above document the importance of protruding atheromas in the thoracic aorta as an embolic source in patients with no other demonstrable source of embolization ("cryptogenic stroke"), and because atherosclerosis is a diffuse disease, we postulated that these aortic lesions might also be important as potential sources of emboli in the subset of patients with documented carotid artery disease. Demopoulous and coworkers [15] studied 45 patients using transesophageal echocardiography. All of the patients had clinically significant carotid stenosis and had had stroke or transient ischemic attack within 6 weeks of having transesophageal echocardiography. These 45 case-patients were matched for age, sex, and history of hypertension with 45 controls who had also recently had a cerebral event but did not have significant carotid stenosis. Protruding atheromas were present in the aortic arches of 17 of 45 case-patients (38%) but only 7 of 45 controls (16%) (P = 0.02). In addition, mobile atheromas (which carry the highest embolic risk) were seen almost exclusively in case-patients (6 of 45 case-patients [13%] and only 1 of 45 controls [2%]; P = 0.05). All case-patients with mobile atheromas had the most severe carotid stenosis (diameter stenosis > 80%). Cerebral symptoms were discordant with the site of carotid stenosis in 10 case-patients; 4 of the 10 had atheromas in the aortic arch.

Clinical-Pathologic Correlations

A pathologic study done in 1945 [16] reported the presence of mural thrombi superimposed on atheromatous plaques in the aortas of 6 of 9 patients who had arterial occlusive embolization and severe, eroded aortic atherosclerotic plaques on autopsy. In a more recent study [17], Amarenco and associates reported on 500 consecutive autopsies done in patients with cerebrovascular and other neurologic diseases. They found that ulcerated plaques were present in the aortic arches of 26% of 239 patients with cerebrovascular disease but only 5% of 261 patients with other neurologic diseases (P < 0.001). Moreover, the prevalence of ulcerated plaques in the aortic arch was 61% among the 28 patients with no other known cause of cerebral infarction and 22% among the 155 patients for whom the cause of cerebral infarction was known (P < 0.001). The authors concluded that ulcerated plaque in the aortic arch may play a role in causing cerebral infarction, especially in patients with cryptogenic stroke. This group later used transesophageal echocardiography to study 250 consecutive patients with stroke and found atherosclerotic plaques at least 4 mm thick in 14.4% of patients with stroke but only 2% of controls [18].

Mobile components of protruding atheromas in the aortic arch were seen on transesophageal echocardiography in two patients who later had endarterectomy of the aortic arch [19, 20]. The material removed from the aortic arch consisted of atherosclerotic plaque with superimposed thrombi (Figure 4 and Figure 5). Nevertheless, the exact composition of protruding atheromas seen on transesophageal echocardiography has not been definitively established in any large group of patients. The echocardiographic image cannot distinguish among blood clots, fibrin, or flail segments of ruptured plaques. Given the experience of the two patients mentioned above, however, as well as the experiences of others who have subsequently had surgery at our institution, it seems likely that the mobile components seen on transesophageal echocardiography are thrombi. In addition, the two patients mentioned above had thrombi surgically removed from their femoral arteries after embolic events; this is further evidence that the mobile components seen in their aortic arches were thrombi. Nihoyannopoulos and associates [21] described 3 patients whose aortic lesions contained mobile components that proved to be thrombi at surgery.

View larger version (114K): [in this window] [in a new window]   Figure 5. Pathologic specimen of thrombus removed from an aortic arch during surgery for a protruding atheroma with mobile components. The specimen was taken from a 43-year-old man who had had an above-the-knee amputation and, despite the use of anticoagulation therapy, had emboli to the left arm and the leg that was not amputated. Reproduced from Tunick and colleagues [20] with permission of Mosby-Year Book.

Several reports have noted that the mobile component in the aortic arch disappeared after anticoagulation [22, 23] or thrombolytic therapy [24]. The instability of these lesions was further underlined in a recent transesophageal echocardiographic study of natural history in patients with protruding atheromas in the aorta [25]. The investigators for this study found that of 18 patients with protruding atheromas (defined as a lesion of 5 mm or mobile lesions or both) who had a follow-up study, 11 (61%) developed a new mobile lesion. Of the 10 patients who had a mobile lesion on the first study and had a follow-up study, 7 (70%) had resolution of a specific previously documented mobile lesion. Thus, it is common for patients with protruding aortic atheromas to develop superimposed mobile lesions (presumably thrombi). It is also common for these mobile components to disappear, possibly because of intrinsic thrombolysis, although they may also disappear because they break loose and embolize.

In an autopsy study by Amarenco and associates [17], atherosclerotic ulcers were common but super-imposed thrombi on atherosclerotic plaques were not seen. The reason for the discrepancy between these pathologic findings and surgical evidence or transesophageal echocardiographic images of protruding masses with mobile components has not been established. It is likely, however, that part or all of the mobile components disappeared after death as a result of clot lysis or during preparation of the aorta for pathologic examination, which included cleaning and flushing of the intimal surfaces. This hypothesis is supported by information in the recent autopsy series by Khatibzadeh and coworkers [26], who examined the thoracic aorta for atherosclerotic disease in 120 consecutive autopsies and found thrombi in 17 cases. Pathologic evidence of embolization was present in 40 patients, and 68% of these patients had complicated plaques (defined as plaques that included ulceration, debris, or thrombi) in the aorta. Complicated plaques were present in only 34% of persons without embolism. Logistic regression analysis found that embolization was associated with complicated plaques in the thoracic aorta, severe carotid disease, and atrial fibrillation.

An additional, different clinical picture-the classic atheroemboli syndrome-is associated with aortic atherosclerosis. This well-described syndrome consists of renal insufficiency, skin lesions, blue toes, and multisystem findings caused by embolic phenomena [27]. It can occur spontaneously but frequently results from arterial manipulation, such as cardiac catheterization and aortotomy during coronary artery bypass surgery. In this syndrome, cholesterol emboli occlude the lumina of small arteries; they can be seen as refractile bodies within the retinal arteries and can be diagnosed microscopically in biopsy specimens from the skin, muscle, or kidney. Protruding atheromas, frequently with mobile components, have recently been seen on transesophageal echocardiography in patients in whom cholesterol emboli syndromes occur spontaneously or after aortic manipulation. Characteristic microscopic findings were seen in biopsy specimens from the skin, muscle, and kidney [28, 29].

Embolization during Invasive Procedures

Instrumentation of the aorta during cardiac catheterization or intra-aortic balloon-pump placement may result in embolization in patients with protruding aortic atheromas. Karalis and colleagues [30] studied patients who were referred for transesophageal echocardiography after having had invasive aortic procedures [30]. Embolism occurred during femoral catheterization in 7 of 48 patients (15%) with protruding atheromas and only 2 of 70 patients without protruding atheromas. During intra-aortic balloon-pump placement, embolism occurred in 4 of 7 patients (57%) with protruding aortic atheromas and in none of 10 patients without protruding atheromas. The presence of mobile components of the protruding atheromas was also associated with a higher tendency toward embolization during invasive aortic procedures. Nine of 21 patients (43%) with mobile components but only 2 of 27 patients (7%) with protruding atheromas but no mobile components had embolic events; this difference was significant. Patients have also been studied with transesophageal echocardiography during intra-aortic balloon counterpulsation. In 1 patient who had an atheroma in the descending thoracic aorta, a mobile component developed during intra-aortic balloon pumping [31]. It is likely that trauma to the atherosclerotic lesion created plaque rupture with thrombus formation that was responsible for the appearance of the new mobile component. Because patients with protruding and mobile aortic atheromas are at high risk for embolization during catheterization, it is important to identify these patients before doing such procedures. The risk for embolization may be reduced by avoiding the invasive procedure altogether or by doing cardiac catheterization through the right brachial artery. Retrograde advancement of the catheter into the ascending aorta prevents potential damage to the intimal surfaces of the femoral and iliac arteries, the abdominal and descending aorta, and the aortic arch and may therefore prevent embolic complications.

Embolic Potential of Atherosclerotic Lesions during Open-Heart Surgery

Stroke, transient ischemic attack, and peripheral embolization are common complications of heart surgery, especially in elderly patients. In various series [32, 33], the incidence of these events ranges from 2% to 5%. The incidence of these events increases in patients older than 70 years of age, in whom the rate of stroke is as great as 7% [34]. The dislodgment of thrombi or atherosclerotic material during manipulation of the aorta is certainly a plausible cause of stroke during cardiopulmonary bypass surgery; several maneuvers that are done during this surgery could cause this complication. These include palpation, cross-clamping, anastomosis of the proximal vein graft to the ascending aorta, cannulation, and the "sandblast" effect of cannula flow against the aortic wall.

Intraoperative transesophageal echocardiography was used to study 130 patients older than 65 years of age who were having surgery that involved cardiopulmonary bypass [35]. Twenty-three patients (18%) had atheromas that protruded into the aortic arch. The incidence of stroke was significantly higher in patients with protruding atheromas (3 of 23 [15%]) than in those without protruding atheromas (2 of 107 [2%]) (P < 0.05). Although the numbers are small, it should be noted that the 2 patients without protruding atheromas who developed stroke also had prosthetic aortic valves. Furthermore, the neurologic symptoms of these patients were not present when the patients awoke from anesthesia; they appeared in the days after surgery. In contrast, the patients with protruding atheromas awoke with their neurologic symptoms already present. It is therefore likely that the protruding atheromas embolized during surgery. A convincing example was noted in one of these patients, in whom the aortic arch was imaged throughout the cannulation of the aortic arch for cardiopulmonary bypass. The cannula could be seen moving through the mobile component of the atheroma, which then disappeared. The patient had an intraoperative stroke.

The results of these studies led to a change in surgical technique that was designed to avoid dislodging the protruding atheroma. In a prospective comparison, strokes occurred in three of seven patients who had the unmodified procedure; no strokes were seen in the five patients who had the modified procedure. The numbers are again small, and the difference was not statistically significant, but it is possible that a change in technique may prevent embolic stroke or peripheral embolization in selected patients.

Several approaches to this problem have been developed. The most radical is to debride the aortic arch under profound hypothermic circulatory arrest before cardiopulmonary bypass. A less radical approach includes avoiding the atheroma by selecting a site for cannulation that is distal to the site of atheroma. The latter technique may prevent intraoperative embolic stroke, but it leaves the atheroma within the aorta and exposes the patient to a potentially higher risk for embolization in the future.

Wareing and coworkers [36] evaluated the ascending aorta in 500 patients older than 50 years of age who had various cardiac operations. Sixty-eight patients (13.6%) had significant atherosclerotic disease in the ascending aorta and were thought to be at increased risk for the development of embolic complications. In these patients, one or more of several modifications of the standard technique was used. These modifications included alterations in the site of aortic cannulation as well as in the techniques of aortic clamping, attachment of vein grafts, and cannulation for the infusion of cardioplegic solution. Ten patients with severe diffuse atheromatous disease had graft replacement of the ascending aorta with hypothermic circulatory arrest, without aortic clamping. Of 68 patients with significant atheromatous disease who had surgery that used a modified technique, only 1 had a reversible ischemic neurologic deficit and none had a permanent intraoperative stroke.

Management

Diagnosis

Transesophageal echocardiography has been shown to be superior to transthoracic echocardiography in locating and identifying embolic sources in patients with cardiogenic embolism. Conditions such as left atrial thrombus, spontaneous left atrial echocardiographic contrast, patent foramen ovale, atrial septal aneurysm, and valvular vegetations are diagnosed with increased sensitivity by transesophageal echocardiography [37]. The aortic arch and the descending thoracic aorta are not well visualized on transthoracic echocardiography and are seen with high resolution on transesophageal echocardiography. The sensitivity and specificity of transesophageal echocardiography for potentially embolic aortic lesions cannot be determined: Even pathologic specimens may not reflect the presence of thrombi in vivo, and no other gold standard exists.

Figure 6 shows an algorithm for the use of ultrasonography in the diagnosis of the source of cerebral and peripheral embolization. As can be seen in this figure, transesophageal echocardiography is indicated when a source of embolization has not been found on other vascular ultrasonographic studies or transthoracic echocardiography.

View larger version (15K): [in this window] [in a new window]   Figure 6. Ultrasonographic evaluation of patients with embolic events. TEE = transesophageal echocardiography; TTE = transthoracic echocardiography.

Therapy

The proper management of patients with protruding aortic atheromas is not yet known. It may be logical, however, to administer anticoagulation to patients whose atheromas have mobile components, because these components have been proven to be thrombi in a small number of patients and have been noted to disappear during anticoagulation therapy. Craig and colleagues [38] recently showed that the mobile elements of atheromas in the aortic arch disappeared or decreased in size while patients were receiving warfarin therapy and that no embolic events occurred during follow-up in a small number of patients. However, the risk–benefit ratio for warfarin therapy needs to be evaluated further. In addition to the widely known complications of anticoagulant therapy, bleeding into atherosclerotic plaques with plaque rupture and cholesterol embolization have been attributed to anticoagulation therapy in a few reports [39, 40]. The value of antithrombotic agents (aspirin or ticlopidine) in patients with aortic atherosclerosis is not known. Successful thrombolysis was described in a single case [24], but it is theoretically possible that thrombolytic agents could selectively lyse the stalk of pedunculated lesions, releasing the bulk of the lesions into the bloodstream as emboli.

Aortic endarterectomy has been done in selected patients, including those in whom cardiac surgery was planned and was not planned. In one study [41], 12 patients were treated surgically. In all of these patients, transesophageal echocardiography showed protruding atheromas, many of which had mobile components. Arterial cannulation was distal to the left carotid artery in 9 patients, in the femoral artery in 2 patients, and in the ascending aorta in 1 patient to avoid the atheromas in the aortic arch. Under circulatory arrest and with the tympanic membrane temperature at 15 °C, the aortic arch was opened and the atherosclerotic plaques were gently debrided. One patient died of intraoperative aortic dissection, and one had postoperative confusion (which cleared) and a small parietal infarction on computed tomography. The 11 survivors were followed for 1.5 to 24 months (mean, 12 months). One case of sudden death occurred at 8 months, and one peripheral embolus to the toes occurred at 3 months. No cerebral emboli occurred. An antiplatelet agent or warfarin was used in all patients.

Surgical removal of atheromas of the aortic arch may eventually become an option in patients with severe embolic symptoms and may also be considered if atheromas are discovered with intraoperative transesophageal echocardiography. Although aortic surgery for atherosclerotic disease is technically possible, the exact risks and long-term results of this procedure are not yet known. The indications for such surgery, as well as the short- and long-term effects of this treatment compared with those of anticoagulation therapy, are also not known. At our institution, a surgical approach to the clinical problem of embolization from the aorta is reserved for highly selected patients who have had recurrent events and are good surgical candidates.

Conclusions

Thoracic aortic atherosclerosis is a common cause of cerebral and peripheral embolization. Protruding atheroma, intimal ulceration, and superimposed mobile thrombi are risk factors. When present, these lesions may lead to vascular events in as many as one third of patients in a single year. The technique of choice for the diagnosis of this condition is transesophageal echocardiography, which is now being done routinely on awake patients, at the bedside, and with a very low risk for complications. The most important remaining question is that of therapy. Surgery is technically feasible but should be reserved for a very select group of good surgical candidates who have had recurrent events. The role of anticoagulant agents, antithrombotic drugs, and strategies to promote atheroma regression awaits systematic evaluation.