We performed TEE in patients with nonvalvular atrial fibrillation who were at high intrinsic risk for thromboembolism and were participating in the Stroke Prevention in Atrial Fibrillation (SPAF-III) study, a randomized clinical trial. In that study, adjusted-dose warfarin (target international normalized ratio [INR], 2 to 3) was compared with combination therapy consisting of low-intensity warfarin (INR, 1.2 to 1.5) plus aspirin, 325 mg/d. The trial was stopped after an interim analysis indicated that the efficacy of adjusted-dose warfarin therapy was superior to that of combination therapy. Low anticoagulation intensity in the combination therapy group yielded high stroke rates that were similar to those expected during treatment with aspirin alone [14, 15]. In the current analysis, we 1) correlate TEE observations among patients with atrial fibrillation at high risk for thromboembolism with rate of subsequent stroke and 2) compare treatment responses.

Methods

Top Methods Results Discussion Author & Article Info References

Patients

Patients were recruited from the SPAF-III multicenter trial, which was designed to compare two antithrombotic medication regimens for prevention of ischemic stroke and systemic embolism (primary events). The study, whose design is described else-where in detail [14], involved patients with nonvalvular atrial fibrillation who were stratified for risk on the basis of clinical and transthoracic echocardiographic criteria derived from previous analyses [15]. Women older than 75 years of age and patients of any age and either sex with systolic hypertension (systolic blood pressure >160 mm Hg), previous thromboembolism, or impaired left ventricular function (M-mode echocardiographic fractional shortening of 0.25 or recent clinical congestive heart failure) were randomly assigned to receive adjusted-dose warfarin (target INR, 2.0 to 3.0) or a combination of low-intensity warfarin (0.5 to 3.0 mg/d; dose was initially adjusted to cause deviation of the INR to 1.2 to 1.5 and was then fixed) and aspirin. Seventy-three percent of the patients were receiving warfarin at various dose intensities before trial entry; these patients were equally distributed between assigned treatment groups. Antithrombotic treatment was assigned before performance of TEE and was not influenced by TEE findings. Patients were followed monthly by telephone and in clinic every 3 months, and a standard questionnaire on stroke symptoms was administered annually. All patients whose responses to the questionnaire suggested the occurrence of stroke were referred for clinical examination by a local affiliated neurologist who was blinded to treatment, and all events were adjudicated by a blinded event verification committee [14].

This project was designed to supplement the SPAF-III clinical trial on the basis of the hypothesis that spontaneous echocardiographic contrast detected by TEE identifies patients at high risk for thromboembolism for whom adjusted-dose warfarin would prove superior to combination therapy. All participants in the parent study were encouraged, but not required, to provide separate, written informed consent for TEE after random assignment to antithrombotic treatment. This was done to prevent the results of TEE from influencing recruitment of patients into the trial or from altering treatment. The intended sample size of 412 TEE examinations was selected to detect an absolute difference in event rates of 3.9% per year (based on projected event rates of 5.4% per year in patients with spontaneous echocardiographic contrast and 1.5% per year in patients without contrast) over a mean observation period of 2.5 years. Secondary objectives were to correlate clinical thromboembolic events with function of the left atrial appendage, thrombus of the left atrium and left atrial appendage, and atherosclerotic lesions in the thoracic aorta imaged by TEE.

Transesophageal Echocardiography

Transesophageal echocardiography was performed within 3 months of enrollment with patients under light sedation and topical anesthesia, according to specific protocols for acquisition and interpretation [16]. Commercially available ultrasonography equipment incorporated biplane or multiplane transducers with at least 48 crystals operating at a standard frequency of 5 MHz with gray-scale imaging. Signal processing, log compression, and persistence values were established for each manufacturer. Images were interpreted locally, and data were entered into standardized forms by echocardiographers who were blinded to antithrombotic treatment assignment but not to patient demographic or clinical features. Original videotaped recordings were collected in a central echocardiographic registry for validation and quality control. To assure generalizability of the results to laboratories used in clinical practice, the analyses reported here are based on image interpretation at the local laboratories.

The following features were specifically assessed: left atrial appendage size, peak antegrade (emptying) velocity of blood flow in the left atrial appendage, spontaneous echocardiographic contrast in the left atrium or left atrial appendage, thrombus of the left atrium or left atrial appendage, and atherosclerotic disease of the thoracic aorta. Spontaneous echocardiographic contrast was defined as a pattern of slowly swirling intracavitary echodensities imaged with gain settings adjusted to distinguish background noise. Contrast was classified as faint when visualized intermittently during the cardiac cycle or when enhanced gain settings were required for detection and was classified as dense when continuously present at standard gain. Aortic plaque was classified by location and as either simple or complex according to mobility, ulceration, pedunculation, or thickness greater than 4 mm [16]. We postulated that reduced emptying velocity from the left atrial appendage, which reflects stasis, would be associated with dense spontaneous echocardiographic contrast and thrombus formation, whereas complex aortic plaque would be a consequence of atherosclerotic vascular abnormality. We also postulated that these findings would predispose patients to ischemic stroke through separate pathogenic mechanisms.

Statistical Analysis

The analyses reported here were limited to variables accessible only by TEE and were hypothesized a priori to have potential predictive relations to clinical thromboembolism. Baseline comparisons between groups were done by using the chi-square test for categorical data and the Student t-test for continuous variables. The Cox proportional-hazards regression model was used to estimate the univariate relative risk associated with TEE characteristics. The primary analysis compared rates of cerebral ischemic events and systemic embolism among patients who had dense spontaneous echocardiographic contrast detected in the left atrium or left atrial appendage with rates in patients who did not have dense contrast; the log-rank statistic was used to compare the times to events between these groups. Secondary analyses, conducted similarly, compared event rates in groups that were identified by left atrial appendage size, flow velocity, thrombus, and aortic plaque. For these comparisons, the area of the left atrial appendage was dichotomized at the upper quartile; flow velocity was dichotomized at the lower quartile. When interaction was seen between treatment and TEE characteristics, analysis of the relation of TEE findings to events was confined to the combination therapy group and treatment effect was described separately; when no interaction was seen, the analysis was stratified by therapy. The relative reduction in event rate associated with adjusted-dose warfarin therapy was computed as the difference between unity and the rate ratio determined by a Poisson regression model [17]. A Cox proportional-hazards regression model was fit to confirm the independence of TEE characteristics identified as univariate thromboembolic risk factors. Confidence intervals for event rates were calculated by using a Poisson regression model; tests of significance were two-sided and were not adjusted for multiple comparisons. All analyses were done according to the intention-to-treat principle.

Results

Top Methods Results Discussion Author & Article Info References

Of 1044 high-risk patients with atrial fibrillation enrolled in the randomized trial, 382 (37%) volunteered for TEE (range, 10% to 80% of SPAF-III participants recruited at 18 clinical centers offering TEE). Salient features of these 382 patients were similar to those of the 662 participants who did not undergo transesophageal imaging (Table 1). The mean patient age was 71 years, 36% of patients had clinical ischemic heart disease, and almost half had a history of congestive heart failure. Balanced proportions of patients were assigned to therapy with adjusted-dose warfarin (n = 190) or fixed-dose warfarin and aspirin (n = 192). Time from random assignment of treatment to TEE was similar in the two groups (mean, 23 days). The trial was stopped after a mean observation period of 1.1 years because an interim analysis indicated the lack of efficacy of the combination regimen [18]. Up to that point, 23 primary events (all of which were ischemic strokes) occurred in the subset of participants who had undergone TEE (event rate, 3.1% per year in patients assigned to adjusted-dose warfarin compared with 7.8% per year in those assigned to combination therapy; P = 0.04). In the subset of patients who underwent TEE, the median INR during follow-up was 2.3 in patients receiving adjusted-dose warfarin and 1.2 in those receiving combination therapy.

Abnormalities of the Left Atrial Appendage

The peak antegrade flow velocity in the left atrial appendage, measured by the Doppler signal, averaged 33 cm/s and did not differ significantly between treatment groups. In 37% of patients, the antegrade (emptying) flow rate was reduced to 20 cm/s or less. Under these conditions, the prevalence of spontaneous echocardiographic contrast was 75% (compared with 58% with higher velocity) and the prevalence of thrombus was 17% (compared with 5% with higher velocity) (P < 0.001 for both comparisons). Spontaneous echocardiographic contrast was detected in the left atrium or left atrial appendage in 63% of the patients examined by TEE, was classified as dense in 20%, and had similar prevalence in the two treatment groups (Table 2). Spontaneous echocardiographic contrast occurred in 38 patients (89%) with thrombus, and thrombus formation was related to contrast density. Thrombus of the atrial appendage was identified in 24% of patients with dense contrast (18 of 76 patients), 10% of patients with faint contrast (16 of 164 patients) and 3% of patients without contrast (4 of 140 patients) (P < 0.001). Although thrombus was more prevalent in the combination therapy group (Table 2), the association between thrombus and dense spontaneous echocardiographic contrast was seen in both treatment groups. The prevalence of dense spontaneous echocardiographic contrast was similar in each category of high risk (24% in patients with previous thromboembolism, 22% in patients with poor left ventricular function, 22% in patients with systolic hypertension, and 20% in women older than 75 years of age).

Thromboembolism rates were inversely related to peak left atrial appendage flow velocity (Table 3). Among patients with peak antegrade velocity less than 20 cm/s, the relative risk for ischemic stroke was 2.6 compared with patients whose velocity exceeded 20 cm/s (P = 0.02). Patients with dense spontaneous echocardiographic contrast assigned to combination therapy had an event rate of 18.2% per year (95% CI, 7.6% to 44% per year) and a relative risk of 2.7 compared with patients without dense contrast (P = 0.06). Patients in the adjusted-dose warfarin group who had dense echocardiographic contrast had an event rate of 4.5% per year (CI, 1.1% to 18% per year); the relative rate reduction was 75% compared with the combination therapy group (P = 0.09) (Table 3). Among patients receiving combination therapy who had dense echocardiographic contrast but no atrial thrombus, the event rate was 14.4% per year (CI, 4.7% to 45% per year).

Thrombus was detected in the left atrial appendage in 38 patients (10%) and was simultaneously present in the atrial cavity in three patients. None had thrombus in the atrial cavity alone. Imaging with TEE, performed an average of 23 days after initiation of randomized treatment, disclosed left atrial thrombus more frequently in patients assigned to combination therapy (26 patients [14%]) than in those assigned to adjusted-dose warfarin (12 patients [6%]) (P = 0.02). Among patients undergoing TEE within 2 weeks of study entry, the prevalence of thrombus was similar in the two treatment groups (10% to 12%). After the first 2 weeks, however, fewer patients assigned to dose-adjusted warfarin than those assigned to combination therapy had thrombus (4% compared with 15%; P = 0.004) (Figure 1). The median INR within 1 week of TEE was 2.2 in patients assigned to adjusted-dose warfarin and 1.2 in those assigned to combination therapy. During the entire observation period, median INRs did not significantly differ between patients with and those without thrombus in either the adjusted-dose warfarin group (2.3) or the combination therapy group (1.2).

View larger version (22K): [in this window] [in a new window]   Figure 1. Prevalence of thrombus of the left atrial appendage (LAA) in high-risk patients with atrial fibrillation according to study group assignment. Striped bars represent combination therapy with low-intensity warfarin and aspirin, 325 mg/d; white bars represent adjusted-dose warfarin. Error bars represent 95% CIs.

Thrombus of the left atrial appendage was associated with increased risk for primary events (relative risk, 2.7; P = 0.04). Among patients with thrombus of the left atrial appendage at study entry who were assigned to the combination regimen, the rate of cerebral ischemic events was 12.9% per year (CI, 4.2% to 40% per year). In patients treated with adjusted-dose warfarin, absence of thrombus at the time of TEE predicted a low rate of ischemic events (2.3% per year [CI, 1.0% to 5.5% per year]); the rate was high (17.9% per year [CI, 4.5% to 71%]) when thrombus was present (P = 0.006) (Table 3).

Atherosclerotic Aortic Plaque

Atherosclerotic lesions in the thoracic aorta were detected in 63% of patients, and prevalence increased progressively from the ascending (14%) to the transverse (37%) and the descending (58%) segments of the aorta. Fifty-six percent of plaque was classified as morphologically complex; 35% of all patients had this type of plaque. Complex plaque was equally prevalent in the two treatment groups (Table 2) and was present in 46% of patients with previous thromboembolism, 32% of patients with poor left ventricular function, 42% of patients with systolic hypertension, and 34% of women older than 75 years of age. The prevalence of complex plaque was similar in patients with dense spontaneous echocardiographic contrast and those without contrast (38% compared with 34%; P > 0.2).

Complex aortic plaque was strongly associated with an increased risk for thromboembolism in patients assigned to the combination regimen (relative risk, 4.0 compared with patients without complex plaque; P = 0.005); the event rate was 15.8% per year in these patients compared with 4.0% per year in patients treated with adjusted-dose warfarin (P = 0.02). The absolute rate reduction associated with adjusted-dose warfarin therapy was greatest (9.0% per year [CI, 0.9% to 17% per year]) when plaque was identified in the descending aorta, distal to the origins of the major cerebral arteries. The difference in rates of cerebral ischemic events between patients in the combination therapy group who had plaque in the ascending aorta (15.6% per year [CI, 5.9% to 42% per year]) and those whose lesions were confined to the descending aorta (17.0% per year [CI, 7.6% to 38% per year]) was not statistically significant. Patients with no plaque had a low event rate (1.1% to 1.2% per year) regardless of treatment, although the CIs in the individual treatment groups were wide (Table 3).

Transesophageal Echocardiographic Risk Factors for Ischemic Events

Because of the small number of events, multivariate analysis was only done to evaluate the independence of TEE characteristics. When assessed together in a multivariate model involving patients assigned to combination therapy, dense spontaneous echocardiographic contrast (P = 0.06) and complex atherosclerotic plaque in the thoracic aorta (P = 0.01) were independently predictive of thromboembolism. Seven percent of patients had both dense spontaneous echocardiographic contrast and complex plaque, 11% had dense contrast, 30% had complex plaque, and 51% had neither. No other baseline TEE findings were independently predictive of thromboembolism.

A constellation of TEE abnormalities related to stasis in the left atrial appendage (consisting of reduced emptying velocity, dense echocardiographic contrast, or thrombus), together with complex aortic plaque, identified patients at extremely high risk for stroke during combination therapy (event rate, 20.5% per year [CI, 9.8% to 43% per year]) (Figure 2). Conversely, the event rate seemed lower in patients with neither complex aortic plaque nor these left atrial abnormalities (1.3% per year [CI, 0.2% to 9.5% per year]).

View larger version (17K): [in this window] [in a new window]   Figure 2. Event rates during combination therapy in subsets defined according to findings on transesophageal echocardiography. Ratios in parentheses relate events to patients with the specified criteria. * = Left atrial abnormality refers to thrombus of the left atrial appendage (14% of patients), dense spontaneous echocardiographic contrast (18% of patients), or peak antegrade (emptying) velocity of blood flow from the left atrial appendage less than 20 cm/s (35% of patients). The 95% CIs for the events are as follows: left atrial abnormality, 2.9% to 21.0% per year; complex aortic plaque, 4.5% to 32.0% per year; both, 9.8% to 43% per year; and neither, 0.2% to 9.5% per year.

Patient Tolerance of Transesophageal Echocardiography

In this elderly cohort of high-risk patients with atrial fibrillation, 4 patients (1.1%) had complications of TEE and 2 (0.6%) required hospitalization because of aspiration, exacerbation of angina, and congestive heart failure without permanent sequelae. Hypopharyngeal hematoma in 1 patient and vomiting without aspiration in another resolved spontaneously. Minor complications, such as severe throat soreness and inability to introduce the esophageal transducer, occurred in fewer than 2% of cases.

Discussion

Top Methods Results Discussion Author & Article Info References

Transesophageal echocardiographic findings predicted subsequent thromboembolic events in patients with atrial fibrillation who were at high risk for thromboembolism, all of whom received antithrombotic therapy. The SPAF-III trial ended before the planned duration of exposure was completed in patients examined by TEE because anticoagulation with adjusted-dose warfarin (INR, 2 to 3) was found to be far superior for prevention of primary events than the fixed-dose combination of low-intensity warfarin and aspirin [18]. The high event rate seen with combination therapy was similar to that projected from earlier studies that used aspirin alone in patient subgroups with similar clinical features [14, 15]. Although early termination impaired the statistical power of our study, we identified TEE features associated with thromboembolism that shed light on the pathogenesis of ischemic events and the mechanisms underlying the response to anticoagulation.

These data confirm links between atrial appendage function, dense spontaneous echocardiographic contrast, thrombus of the atrial appendage, and ischemic stroke in patients with atrial fibrillation [18]. Spontaneous echocardiographic contrast was related to reduced blood flow velocity in the left atrial appendage (reduced velocity indicates stasis) and was detected in most of these patients, who were included because of their high thromboembolic risk [18-22]. The prevalence of thrombus was related to the duration of exposure to antithrombotic therapy and was lower in patients with dense echocardiographic contrast who received adjusted-dose warfarin than in those who received combination therapy. Despite therapy with adjusted-dose warfarin, risk for ischemic events increased substantially when thrombus was detected. This observation supports the view that reduction of thrombus is one mechanism by which warfarin in the intensity used in the adjusted-dose group proved effective for preventing thromboembolism and implies that detection of thrombus in high-risk patients with atrial fibrillation receiving warfarin may call for more intensive anticoagulation.

Atherosclerotic plaque in the ascending and transverse (arch) segments of the thoracic aorta are potential sources of cerebral ischemic events, particularly when ulceration, pedunculation, mobility, or thickness of the lesions increases the potential for embolism [23, 24]; such plaque may reflect generalized atherosclerosis involving arteries that supply the brain. Plaque in the thoracic aorta was identified in most high-risk patients with atrial fibrillation and was seen more frequently in the descending than in the ascending segments of the thoracic aorta. This may have occurred because of the distribution of vascular abnormality or because the descending aorta is more easily imaged (patients with plaque in the ascending aorta invariably also had lesions in the distal thoracic aorta).

The rate of ischemic events in patients without aortic plaque was relatively low, even in patients assigned to the less effective antithrombotic treatment (1.2% per year [CI, 0.2% to 8.7% per year]). It is not clear, however, whether patients without detectable plaque can be safely managed without long-term anticoagulation or whether treatment aimed specifically at controlling atherosclerosis in patients with plaque could reduce the risk for stroke. Although the rate of cerebral ischemic events associated with aortic plaque was lower in patients assigned to adjusted-dose warfarin than in those assigned to combination therapy, the origin of these events in individual patients could not be determined. Furthermore, patients with atrial fibrillation and complex plaque structure who received adjusted-dose warfarin still had a high rate of thromboembolism (4.0% per year [CI, 1.3% to 12% per year]). Thus, more effective treatments must be sought for patients with extensive aortic atherosclerosis.

An important limitation of multivariate models derives from their dependence on the prevalence of the variables in the cohort and the strength of their association with outcome events. The low stroke rate in patients treated with adjusted-dose warfarin limits our ability to reliably discern TEE findings that can identify patients at higher risk for stroke in that group. However, the effects of all TEE risk factors were qualitatively similar in both treatment groups, suggesting that our findings in the combination therapy group can be extrapolated to those in the adjusted-dose group, despite the overall lower risk in the latter group.

A constellation of TEE abnormalities, including stasis of blood in the left atrial appendage, spontaneous echocardiographic contrast, and thrombus formation, may be pathophysiologically related to cardiogenic embolism in patients with atrial fibrillation. Complex plaque in the thoracic aorta, on the other hand, reflects arterial abnormalities that contribute to stroke by other mechanisms. If this theory is valid, the presence of TEE abnormalities involving both the left atrial appendage and aortic plaque should compound the risk for stroke during combination therapy; the absence of either should identify patients at lower risk. Figure 2 supports this theory within broad CIs. Similar mechanisms are probably operative in patients receiving adjusted-dose warfarin, although the effects of these mechanisms were less apparent because of the lower event rate in this group. In addition, because all patients were treated with varying doses of warfarin with or without aspirin, we cannot reliably identify a low-risk subgroup according to TEE findings.

The high-risk patients included in our trial were elderly and had substantial associated cardiovascular disease. The value of TEE for delineating stroke risk among patients with less severe comorbid conditions is being investigated in an ongoing component of the SPAF-III study; in this part of the study, patients with atrial fibrillation who are at lower intrinsic risk for stroke are receiving aspirin alone [25].

Other limitations of our study include early termination of the trial, which occurred because an interim analysis indicated the superior efficacy of adjusted-dose warfarin. This impaired the statistical power and contributed to the broad CIs surrounding many of these observations. Interpretation of TEE images at the local clinical laboratories rather than at the core echocardiography laboratory introduced interobserver variability but enhanced the generalizability of the findings to medical practice and predisposed to underestimation of the strength of the correlations. In addition, because TEE was done an average of 23 days after randomization, patients were exposed to assigned therapy for different intervals before examination.

Despite these limitations, the correlation of TEE findings with clinical thromboembolic events yielded robust relative risks and has pathophysiologic implications. Dense spontaneous echocardiographic contrast in the left atrium and left atrial appendage arises when blood flow velocity is low, and adequate anticoagulation reduces the association of contrast with formation of thrombus of the atrial appendage. Thrombus is directly related to risk for stroke, which is substantial when thrombus persists despite warfarin therapy. Finally, extracardiac pathologic abnormalities, such as atherosclerotic disease of the thoracic aorta, seem to contribute independently to stroke in patients with atrial fibrillation who have a high risk for thromboembolism.

Appendix: Stroke Prevention in Atrial Fibrillation Investigators Committee on Echocardiography

Investigators are listed by center in order of recruitment; numbers of patients are given in parentheses.

University of Texas Health Science Center, San Antonio, Texas (49): Miguel Zabalgoitia, MD, and Anne Leonard, RN, BSN. Mayo Clinic, Jacksonville, Florida (42): Joseph L. Blackshear, MD; Robert Safford, MD; and Vickie S. Baker, RN, BSN. University Medical Center, Tucson, Arizona (39): Paul Fenster, MD; Gregory D. Pennock, MD; Janet Ohm, RN; and Bobbi J. Huerta, RN, BSN. Kaiser Sunnyside Medical Center, Clackamas, Oregon (36): Richard Strauss, MD; Marlene McKenzie, RN; Patricia Hart-McArthur, RN; Mindy Gramberg, RN; and Heather Houston, RN. University of Ottawa Heart Institute, Ottawa, Ontario, Canada (28): Kwan-Leung Chan, MD; Jennifer Biggs, RN; and Anne Ives, RN. Moses Cone Memorial Hospital, Greensboro, North Carolina (26): Robert Rothbart, MD, and Diana Murray, RN, BSN. Mayo Clinic, Scottsdale, Arizona (23): Hassan Loutfi, MD; John J. Lynch, MD; and Lori Carlson, RN. St. Louis University Medical Center, St. Louis, Missouri (22): Arthur J. Labovitz, MD, and Anne Vigil, RN, MSN. Hennepin County Medical Center, Minneapolis, Minnesota (20): Richard W. Asinger, MD; Candace D. Dick, MD; Susan M. Newburg, RN, BSN; and Jeanne Fifield, RN. Mayo Clinic, Rochester, Minnesota (19): Bijoy K. Khandheria, MD; Anne E. Holland, RN, BSN; and Kathy R. Tucker, RN, BSN. Mount Sinai Medical Center, New York, New York (17): Martin E. Goldman, MD, and Elizabeth B. Rothlauf, RN, MS. University Hospital and Clinics of Missouri, Columbia, Missouri (15): Darla Hess, MD, and Lori Young, RN, BSN. Brooke Army Medical Center, San Antonio, Texas (14): David M. Mego, MD, and Sheri Boyd, MD. Kansas University Medical Center, Kansas City, Kansas (13): David Wilson, MD; Barbie Nolte, RN, BSN; and Cari Edwards, RN. St. John's Mercy Hospital, St. Louis, Missouri (10): Leslie Mezei, MD, and Sherry Riggio, RN. Oregon Health Science University, Portland, Oregon (7): George Giraud, MD; Christy Marchant, RN, MBA; and Julie Timberg, RN. University of California, San Diego, San Diego, California (7): Howard C. Dittrich, MD; William Keen, MD; and Carol Kerridge, RN, MPH. Boston University Medical Center, Boston, Massachusetts (7): Sheila Bernard, MD, and Nancy Battinelli, RN. Stanford University School of Medicine, Palo Alto, California (3): John Edwin Atwood, MD, and Susan Quaglietti, RN, MSN. University of South Alabama, Mobile, Alabama (3): Martin A. Alpert, MD; Clara V. Massey, MD; and Carolyn Hupp, RN, MSN.

Data Management and Biostatistics: Jodi Koehler, MS, and Lesly A. Pearce, MS (Statistics and Epidemiology Research Corp., Seattle, Washington).

Clinical Coordinators: Jonathan L. Halperin, MD (Mount Sinai Medical Center, New York, New York), and Robert G. Hart, MD (University of Texas Health Science Center, San Antonio, Texas).

Federal Project Officer: John R. Marler, MD (National Institute of Neurological Disorders and Stroke, Bethesda, Maryland).

Advisory Committee: Anthony N. DeMaria, MD (University of California Medical Center, San Diego, California); A. Jamil Tajik, MD (Mayo Clinic, Rochester, Minnesota); and Arthur Weyman, MD (Massachusetts General Hospital, Boston, Massachusetts).

Writing Committee: Kwan-Leung Chan, MD (Chair); Richard W. Asinger, MD; Joseph L. Blackshear, MD; Paul Fenster, MD; Martin E. Goldman, MD; Jodi Koehler, MS; Jonathan L. Halperin, MD; and Miguel Zabalgoitia, MD.

Corresponding Author Address: Kwan-Leung Chan, MD, University of Ottawa Heart Institute, Ottawa Civic Hospital, 1053 Carling Avenue, Ottawa K1Y 4E9, Canada.