The results of the CASANOVA (Carotid Artery Stenosis with Asymptomatic Narrowing: Operation versus Aspirin) [4] and Mayo Asymptomatic Carotid Endarterectomy [5] randomized trials were negative. Both trials had design problems and unacceptably high rates of perioperative complications. The Veterans Affairs trial found benefit only when transient ischemic attacks were included with stroke as a primary outcome event [6]. When the transient events were omitted from the analyses, endarterectomy did not improve the occurrence of perioperative stroke and death or long-term stroke-free survival. Reduction in the number of transient ischemic events is not an adequate exchange for the more serious perioperative risk for stroke and death.

The Asymptomatic Carotid Artery Study (ACAS), a multicenter trial conducted in 39 centers in the United States and Canada, randomly assigned half of the 1662 asymptomatic patients to receive the best medical care plus endarterectomy and the other half to receive only the best medical care [7]. In this issue of Annals, two of the ACAS investigators comment on this study [8]. On the basis of the statistical results of this study, Drs. Brott and Toole urge physicians to consider endarterectomy in asymptomatic persons with carotid artery stenosis of 60% or greater. In our view, a closer look at the ACAS results yields a much more cautious interpretation of their data.

The reported 53% relative risk reduction is much less impressive when converted into an absolute risk reduction of 5.9% over 5 years. This is an average stroke reduction of slightly more than 1% per year. If this positive, albeit slight, benefit could be accomplished with a smaller associated surgical complication rate, there would be no concern that its statistical significance was being interpreted as a matter of clinical importance. Even in ACAS, however, the dramatically low perioperative rate of 2.3% is the same as the annual risk for stroke in the persons in the control group. Asymptomatic patients for whom carotid endarterectomy is recommended must be advised that they face a higher risk for stroke in the first year than if they continue to receive medical treatment alone. Several more years may pass before these patients can be shown to benefit from the procedure. In other reported studies of asymptomatic patients, the average perioperative rate was 4.5% [4, 5, 9, 10]. An acceptable risk-benefit ratio is not achieved by such an "average" perioperative rate. The 3% rate recommended by an American Heart Association Committee as the maximum perioperative risk in asymptomatic patients is three times higher than the average annualized risk difference of ACAS [11].

According to the ACAS results, the numbers of patients needed to be treated by endarterectomy to prevent 1 stroke in 2 and 5 years are high: 67 and 17, respectively. By contrast, in the symptomatic patients studied by the North American Symptomatic Carotid Endarterectomy Trial (NASCET) [12], 10 patients needed to be treated by endarterectomy to prevent 1 stroke in 1 year and 6 patients to prevent 1 stroke in 2 years. For patients in NASCET with the highest degree of stenosis (90% to 99%), only 3 patients needed to have endarterectomy to prevent 1 stroke in 2 years [12].

See also the preceding Perspective as well as pp 649-55 and the editorial note on p 729.

In the ACAS trial, women were not shown to benefit from endarterectomy. The facts that only 565 women were studied in the trial and that the reported perioperative complication rate (3.6%) was more than double the rate in men (1.7%) probably account for this anomalous finding.

It is particularly disappointing that no difference was seen in the occurrence of disabling strokes between the surgical and medical arms of ACAS. Patients must be advised that major disability faces them just as often when they elect endarterectomy as when they elect medical therapy and that major stroke is an immediate risk. Physicians who would recommend carotid endarterectomy for the asymptomatic patient must recognize that the economic burden and the personal and family anguish of a disabling stroke will not be diminished.

Many previous observations have affirmed that patients with the highest degree of asymptomatic stenosis ( 80%) are more vulnerable to stroke than those with lesser degrees ( 75% to 80%) [13-16]. It is disappointing that ACAS included too few patients to support this theory. Only 85 outcome events of ipsilateral stroke or perioperative death occurred in the median 2.7 years of follow-up. A larger sample size for this type of study, on the order of 7000 persons, has been recommended on the basis of observed outcome events in population case series [17]. In addition, because ACAS depended on Doppler ultrasound as the primary imaging technique, the patients could not be stratified by deciles of stenosis. The beneficial results for endarterectomy from ACAS are too modest to be applied to all persons with stenosis greater than 60%. The published data cannot be used to formulate guidelines to advise which asymptomatic persons with what amount of stenosis and what combination of risk factors should have the procedure.

None of the studies of asymptomatic patients that were done before ACAS depended on ultrasound except to screen potential candidates for angiography. Even when ultrasound is perfected to the point at which the degree of extracranial stenosis can be determined with complete accuracy, it will not identify many important intracranial pathologic processes (such as stenosis, occlusion, and aneurysm). Ultrasound is ineffective in the presence of extensive calcification and tortuosity, and it does not image lesions that extend beyond the carotid bulb or identify near-occlusion and soft thrombi within the arteries. All these features add to the risk for stroke and of endarterectomy [18].

Another caveat to the use of ultrasound relates to a technical point: The measurements from Doppler studies are commonly reported as a percentage that reflects an area rather than a linear measurement. A 60% area stenosis seen by ultrasound is only a 45% to 55% linear stenosis seen by angiography. Consequently, the description of a 60% stenosis from a Doppler report may be misinterpreted in clinical practice. In ACAS, 60% is meant to reflect a conversion to linear stenosis by a comparison of peak frequencies on Doppler with arteriographic linear measurements. The 60% value is only equivalent to a linear measurement on an arteriogram if the cut-point formula of ACAS is applied [19]. Because no data confirm the benefit of endarterectomy at levels of 45% to 55% linear stenosis, area measurement by Doppler must not be equated with linear measurement by angiography.

The published results from the studies in asymptomatic patients are not sturdy enough to lead to the abandonment of conventional arteriography and its small inherent risk. It is hoped that the perfection of ultrasound and magnetic resonance angiography will make conventional arteriography unnecessary; however, these refinements are still ahead of us.

The results from all trials of asymptomatic carotid stenosis are disappointingly inconclusive. With further study, patients with the most severe stenosis will probably be shown to benefit from carotid endarterectomy. When this benefit is known, the procedure should be done by the most skilled surgeons whose results have been validated by independent audit. We and others [20, 21] worry that premature uncritical acceptance of the reports will lead to many inappropriate endarterectomies. A result that is exalted for its statistical significance must also have compelling clinical importance before patients are urged to seek investigation leading to a treatment that imposes risks for major stroke or death. We do not recommend mass screening of populations to detect asymptomatic carotid artery lesions.

In the Asymptomatic Carotid Surgery Trial, a much larger randomized controlled trial of this problem [22], investigators plan to have enough patients in all deciles of asymptomatic stenosis to confirm or deny that there are subgroups of patients who will clearly benefit from carotid endarterectomy. The results of this trial are not expected for 3 or more years. In the meantime, we and our patients must wait for clearer guidelines of the kind demanded by evidence-based medicine.