Featured Article

Introduction
Carotid artery stenosis and atrial fibrillation are the two major sources of atheroembolic stroke and peripheral emboli, accounting for nearly 60% of the cases of stroke or cerebrovascular accident (CVA) and emboli.¹ “Cryptogenic stroke” occurs in about 40% of patients in whom no etiology is identified. ¹

In the last decade transesophageal echocardiography (TEE) has gained increasing acceptance for determining possible etiologic factors of stroke and peripheral emboli. In 1990, Tunick and Kronzon reported for the first time aortic atheroma as a new finding by TEE, and suggested it was an etiologic factor of stroke and peripheral emboli.² Assessment of the thoracic aorta for atheroma has become a routine component of TEE studies. Though no causal relationship has been identified, the association with atheroembolic stroke and peripheral emboli is well established.^3–7 This association is independent of other risk factors for stroke.⁸ Many case-control and retrospective studies have reported a consistent prevalence of aortic atheroma of 21–27% in patients with an atheroembolic disease.^8,9

Anticoagulation with warfarin, anti-platelet agents (aspirin, ticlopidine) and hydroxymethyl glutaryl coenzyme A reductase inhibitors (statins) are currently used to manage patients with protruding aortic atheroma. Serial TEE studies have shown regression of aortic atheroma with lipid-lowering therapy and warfarin.^10,11 The mobile elements in atheromatous lesions have been shown to be thrombi- superimposed on atherosclerotic plaque.¹² Thus, the use of warfarin to treat this condition seems intuitive, and is supported by a few studies.^13–15 To date, there have been no large randomized trials of warfarin therapy in patients with aortic atheroma.³³ There are several unanswered questions about the safety of chronic anticoagulation in these patients, especially due to reported cases of adverse outcomes related to peripheral embolic events — “blue-toe syndrome”.^16–21 Our aims of this study were: (1) to identify high-risk features of aortic atheroma in stroke patients by TEE and (2) to assess whether warfarin use increased the risk of peripheral embolic, recurrent stroke and vascular endpoint.

Methods
Data from 280 consecutive patients referred for TEE over a four-year period following an ischemic CVA were retrospectively reviewed using electronic charts, computerized database and echocardiography laboratory records. Mean follow-up was 26 months, ranging from 64 to 5 months; 72% of patients having completed >/- 12-month follow-up.

Aortic atheroma was identified in 68 patients. Clinical characteristics, size, location in the aorta and mobility of aortic atheroma and predetermined end-points (peripheral embolic events, recurrent stroke and vascular endpoint) were recorded. Peripheral embolic events were defined as distal emboli to extremities and end organs, except CVA; recurrent stroke (CVA) was treated as a separate endpoint and vascular endpoint defined as a composite endpoint of recurrent CVA and/or peripheral embolic events. The presence of prosthetic valves and intra-cardiac thrombi were also noted. Data concerning patient baseline characteristics are expressed as a percentage of the study patient group (n = 68).

Baseline clinical characteristics of patients are shown in Table 1(a). The presenting CVA was considered to be the index vascular event. Recurrent CVA or peripheral embolic events were defined as vascular events during the follow-up period. Size of aortic atheroma was measured at its greatest transverse diameter (small atheroma were < 4 mm and large atheroma >/- 4 mm). Localization of aortic atheroma as proximal or distal was in reference to take-off of the left subclavian artery from the aortic arch. Mobility of aortic atheroma was characterized on the basis of visual identification of mobile components.

The choice of antiplatelet agents or warfarin therapy was at the discretion of the referring physician. The international normalized ratio (INR) in the warfarin treated group was between 2–3.

Statistics
Descriptive statistics were calculated for the study sample using standard univariate method for both continuous and categorical variables. Comparisons between the mobile and non-mobile aortic atheroma groups were evaluated by either a t-test for continuous variables or a chi-square analysis for categorical variables. Outcome measures were first evaluated using chi-square analyses; follow-up analyses for conditional independence were performed utilizing the Cochran-Mantel-Haenszel Statistic (CMH). A CMH p ? 0.05 suggests conditional independence. Estimation of risk (Odds ratios, OR), expressed with 95% confidence intervals (CI) was done using multivariate logistic regression analysis. Discrimination of the model was determined using the c statistic, and calibration of the risk-adjustment model was assessed using the Hosmer-Lemeshow statistic. All analyses were performed using SAS for Windows version 7.0.

Results
The study population was comprised of 68 patients (mean age of 73 years) with an ischemic CVA and aortic atheroma detected by TEE. Hypertension was present in 88% of patients, a history of smoking in 85%, hyperlipidemia in 48% and diabetes in 30%. A history of coronary disease was present in 58%; 41% had history of prior stroke; and 25% had atrial fibrillation. Large aortic atheroma were seen in 44 patients (64.7%), and small aortic atheroma in 24 (35.3%); Proximal atheroma were noted in 30 patients (44.1%), and distal in 60 (88.2%); 22 patients (32.3%) had mobile, and 46 (67.3%) had non-mobile aortic atheroma. Baseline characteristics of patients are shown in Table 1(a). A bivariate analysis of patients with mobile versus non-mobile and on warfarin versus antiplatelet therapy is shown in Table 1(b). Analysis of endpoints by TEE characteristics of aortic atheroma is shown in Table 2. Mobility was a high-risk feature of aortic atheroma and was associated with increased vascular endpoint (p = 0.05), and reaching a combined endpoint of all cause death and vascular endpoint (p = 0.001). The size and location of aortic atheroma were not associated with increase in peripheral embolic events, recurrent stroke and vascular endpoint. There were no significant differences between the mobile and non-mobile atheroma groups, except for hyperlipidemia and distal location of aortic atheroma, which were significantly higher in the group with non-mobile atheroma group.

Endpoint analysis in patients with mobile and non-mobile atheroma is shown in Table 3a. Of the 37 patients on warfarin, 7 (18.9%) had embolic events (“blue-toe syndrome” in 2 patients, acute lower extremity thromboembolic event in 1 patient, emboli to the eye in 1 patient, acute upper extremity thromboembolic event in 1 patient, bowel infarction in 1 patient and acute renal failure in 1 patient), compared to 2 (6.4%) embolic events (both patients had lower extremity thromboembolic events) out of 31 patients on antiplatelet agents (p = 0.14) (Table 3b). Eleven patients (29.7%) out of 37 in the warfarin-treated group, and 4 (12.9%) of 31 in the antiplatelet group had recurrent CVA (p = 0.09). In the analysis of warfarin use and vascular endpoint, 17 (45.9%) out of 37 warfarin-treated patients reached a vascular endpoint compared with 6 (19.3%) of 31 on antiplatelet therapy (p = 0.02) (Table 3b). Increased vascular endpoint on warfarin therapy was seen even after adjusting for mobility of aortic atheroma (p = 0.03). There were no significant differences between the warfarin and antiplatelet therapy groups, except for diabetes. To control the high-risk profile of the patients and warfarin use in atrial fibrillation, which is the most common indication for its use in stroke patients, we conducted a multivariate regression analysis (Table 4). Warfarin use in patients with mobile aortic atheroma was associated with an increased risk of reaching a vascular endpoint, OR = 10.9, CI = 1.5–76.3; p = 0.01, independent of traditional risk factors (age > 70 years, hypertension, diabetes, hyperlipidemia) and warfarin use in atrial fibrillation. The wide CI is reflective of the small study size. As expected, warfarin use in patients in atrial fibrillation did not have an increased vascular endpoint (OR = 0.05, CI = 0.005–0.568, p = 0.01). More importantly, our results confirm the beneficial effect of warfarin in stroke patients with atrial fibrillation and aortic atheroma.

Discussion
Introduction: The presence of aortic atheroma is related to the risk factors for atherosclerosis; age above 70 years, hypertension, hyperlipidemia, smoking, and markers of hypercoaguability.^22,23 Embolic complications occur at a rate of 20–30% per year in patients with atheromatous plaques in the aorta.²⁴ The presence of aortic atheroma incrementally adds to the risk of stroke and emboli from carotid artery stenosis 25 and/or atrial fibrillation.²⁶ Aortic plaque thickness of > 4 mm³ and presence of mobile component 13 are TEE predictors of an increased embolic risk. The same is true for a ruptured, non-calcific plaque with or without superimposed thrombi.²⁷ An autopsy study reported a higher prevalence of ulcerated plaques in patients with “cryptogenic stroke”.²⁸ A higher atheroembolic complication rate has been reported in patients with aortic atheroma undergoing carotid endartectomy, coronary catheterization and cross clamping of the aorta.^29,30 TEE identification of aortic atheroma is a marker for the presence of occlusive coronary disease.³¹

Pathogenesis. The pathogenesis of atherosclerosis involves a complex series of events, similar to a chronic inflammatory process, with the formation atherosclerotic plaque. Symptoms occur when advanced lesions are complicated by plaque rupture, or hemorrhage into the plaque, emboli, or thrombosis. There is evidence to suggest that thrombin constitutes mobile elements of aortic atherosclerotic plaques, and it is a marker of adverse outcomes.

Thus, the use of chronic anticoagulation with warfarin seems logical. On the other hand, the pathogenesis of plaque rupture and intraplaque hemorrhage is distinct, though one may lead to the other. Atherosclerotic plaques can ulcerate, develop mobile projections and be a source of embolism. Cholesterol microembolization is also considered a source of peripheral emboli or “blue-toe syndrome”. Many anecdotal reports have implicated warfarin as a precipitating factor in the pathogenesis of “blue-toe syndrome”.^16–21 Warfarin can result in intraplaque hemorrhage and result in distal embolization; it can also prevent the formation of thrombi over ulcerated atheromatous plaques, allowing exposed cholesterol crystals to be swept into the circulation.^19,20 Warfarin use was determined to be the second most common precipitating factor for cholesterol crystal embolization after angiographic procedures, in a review of 221 cases.¹⁸ Similarly, anecdotal evidence also exists for digital embolization from plaque-related thrombus in the thoracic aorta, its identification with TEE and resolution with warfarin therapy.¹¹ TEE identification of heterogeneous and irregular plaque morphology is associated with intraplaque hemorrhage and neurologic events.²¹ Thus, in the absence of large randomized studies of warfarin use in stroke patients with aortic atheroma, the peripheral embolic risk of warfarin therapy remains undefined.

Therapeutic Options. Lipid-lowering therapy and systemic anticoagulation are presently favored in stroke patients with aortic atheroma, on the basis of few observational studies and retrospective analyses.^13,14 There is evidence that lipid-laden plaques are more prone to ulcerate, rupture and undergo thrombosis²⁷ and thus lipid lowering with statins may stabilize the plaque and reduce the embolic risk. Statins also reduce platelet thrombin generation.³²

Only a few studies have demonstrated a decrease in the incidence of embolic events in patients with large and mobile aortic atheroma treated with warfarin.^5,14,15 Regression of mobile atheromatous plaques on warfarin has been shown by TEE studies.¹⁰ Dressler et al reported a higher incidence of vascular events (stroke and peripheral emboli) in patients with mobile aortic atheroma not on warfarin compared to those treated with warfarin (45% versus 5%).¹³ In an observational non-randomized study of warfarin, aspirin and ticlopidine given to patients with large aortic atheroma, there was a significant reduction in embolic events in the group treated with warfarin.¹⁴ In the cohort with mobile atheroma, there was a significant reduction in mortality and a trend towards reduction in embolic events with warfarin therapy. Of note, the above studies included 31 and 129 patients respectively and that the absolute number of atheroembolic episodes was small, 3 and 5, respectively.^13,14 In the Stroke Prevention in Atrial Fibrillation (SPAF) trial, patients with atrial fibrillation were assigned to therapeutic anticoagulation (INR = 2.0–3.0), as opposed to low-fixed dose warfarin (INR = 1.2–1.5). There was a 75% reduction in risk of stroke in patients with atheroma on therapeutic anticoagulation. Only one of 134 patients on warfarin had an atheroembolic episode.²⁶ However, randomization to either treatment arm was not on the basis of TEE characteristics of aortic atheroma. Systemic anticoagulation remains controversial because of the potential of inducing plaque hemorrhage and subsequent rupture and embolization.^16,17

Our Study. In our study, we observed that mobility and not location and size was a high-risk feature of aortic atheroma and was associated with increased vascular endpoint. A higher number of patients reached vascular endpoint on warfarin therapy and this association was preserved even after adjusting for mobility and traditional risk factors. Also, in contrast with other studies with small numbers of peripheral embolic events, we found an increase in the absolute number of peripheral embolic events. As seen in the SPAF trial, warfarin use in patients in atrial fibrillation was not associated with increased vascular endpoint.

Potential limitations of our study include the retrospective nature of the analysis, small number of patients, limited follow-up duration and potential for a selection bias in an elderly population of patients at a higher risk for atherosclerotic disease and stroke. This risk factor profile places our patient group in a very high risk category, compared to contemporary studies. These limitations do not allow us to implicate warfarin as the sole determinant of peripheral embolic and vascular events in our patient population and should not dissuade us from using it in conditions where its benefit has been clearly demonstrated in large randomized trials. Our analysis reaffirms this fact, as the use of warfarin was clearly beneficial in elderly high-risk patients with atrial fibrillation. Despite the limitations of our study, our findings are very relevant for management of elderly high-risk population with stroke, for whom warfarin therapy for aortic atheroma cannot be claimed to be unequivocally beneficial. The presence of mobile aortic atheroma on TEE identifies patients at a higher risk for vascular events and portends an adverse outcome in this patient population. In our study, the use of systemic anticoagulation in such patients did not show a clear benefit. This study may provide an insight into unanswered questions regarding warfarin therapy in a high-risk patient population with stroke and aortic atheroma. A prospective randomized trial is needed to answer these questions conclusively.

Conclusion
(1) Mobility of aortic atheroma is a high-risk feature and is a predictor of vascular endpoint. (2) Warfarin was associated with higher vascular endpoints in elderly stroke patients with aortic atheroma and this association was preserved even after adjusting for mobility. Warfarin use in patients with mobile aortic atheroma was associated with increased vascular endpoint, independent of traditional risk factors. Though not statistically significant, warfarin use was associated with higher peripheral embolic events.