When all subgroups are considered, the results favored surgery over medical treatment alone; however, several important subgroups comprised too few patients to show statistical significance, thus creating potential misunderstanding and difficulty in clinical interpretation. For example, the 5-year risk for perioperative stroke or death plus major ipsilateral stroke was 6.0% for patients receiving medical treatment and 3.4% for patients having surgery, but the difference was not significant (P = 0.12). The lack of statistical significance does not mean that carotid endarterectomy does not prevent major stroke. The ACAS study design [2] was not intended, nor was the sample size large enough, to address prevention of subgroups of stroke such as major or minor stroke. Criticisms of the ACAS results suggesting that carotid endarterectomy does not prevent major stroke in patients with asymptomatic carotid artery stenosis are therefore inaccurate.

Another example of subgroup analysis without sufficient power involves the different outcomes for men and women. For the primary end point, the absolute and relative 5-year risk reductions were 8% and 66%, respectively, for men and 1.4% and 17%, respectively, for women. These differences are provocative but not statistically significant with regard to a sex difference in the benefit of carotid endarterectomy. Finally, in contrast to findings of other studies, the ACAS investigators found that an increasing degree of stenosis shown by arteriography did not predict increased risk in patients receiving medical treatment or those having surgery. However, the stenosis subgroups were small, particularly the group of patients with high-grade stenoses; 25% of the patients with angiograms had 80% to 89% stenosis and only 5% had 90% to 99% stenosis. Therefore, the ACAS trial does not preclude the possibility that, in a larger study population, increasing stenosis and risk may be related.

As anticipated [2], because the rate of stroke for the patients receiving medical treatment in ACAS was low (11%), the absolute 5-year risk reduction of 5.9% in the patients having surgery is modest. In contrast, the North American Symptomatic Carotid Endarterectomy Trial (NASCET) [3] showed that in patients with symptomatic hemodynamically significant carotid artery disease and stenosis of at least 70% who had surgery, the absolute risk for stroke was reduced by 17% over 2 years. This robust risk reduction was largely driven by the high rate of stroke in the patients receiving medical treatment (26%). Therefore, both ACAS and NASCET confirm the previously documented differences in risk for stroke between asymptomatic [4, 5] and symptomatic [6] patients and suggest different degrees of stroke prevention conferred by carotid endarterectomy. However, both groups of patients benefit from the procedure.

Recent studies of the treatment of hypertension in the elderly can provide perspective for the 5.9% absolute risk reduction shown for carotid endarterectomy in ACAS. The Medical Research Council (MRC) trial investigators [7] randomly assigned hypertensive patients aged 65 to 74 years to receive active antihypertensive treatment or placebo; in these patients, reduction of stroke over the mean 5.8-year follow-up was the greatest among the vascular end points measured. The absolute risk reduction was 1.4%. For patients 60 years of age and older with isolated systolic hypertension who were randomly assigned to receive active treatment or placebo in the Systolic Hypertension in the Elderly Program (SHEP) [8], the primary end point was stroke. Over the 4- to 5-year average follow-up, the absolute risk reduction was 3.7%. In ACAS, stroke-free survival was higher in the patients having surgery; however, the difference was not statistically significant (relative risk reduction, 20%; P = 0.08). In the MRC trial, mortality did not significantly differ in the active-treatment and placebo groups. In SHEP, mortality was lower in the active-treatment group (relative risk reduction, 13% [95% CI, –5% to 27%]).

The combined perioperative mortality and morbidity in ACAS for the patients having surgery was 2.3%, a dramatic improvement in safety compared with results from the previous large randomized trials of carotid endarterectomy [3, 6, 9-11]. Can the ACAS results be applied to the general medical community? Editorial comments and practice guidelines [12] include caveats limiting application of the procedures to patients in whom carotid endarterectomy is determined to be low risk. But how can referring physicians determine the track record of their surgical colleagues? In two population-based studies of all carotid endarterectomies done over 12 months in a large metropolitan area, investigators had difficulty detecting differences in perioperative outcomes among surgeons [13, 14]. The mean annual case-load was 11 surgeries. Thus, a single perioperative complication in 1 year would signify a spurious but unacceptably high complication rate (9%). Yet, if the surgeon continued to average 11 carotid endarterectomies and 1 major complication each year, 11 years (126 cases) would elapse before a significant difference from the recommended 3% complication rate [12] would be detectable (P < 0.05). Despite these difficulties with documentation, careful consideration must be given to this surgical risk [15]; looking to the future, we suggest that all surgeons keep a summary account of their cases of carotid endarterectomy, including indications and outcomes.

The ACAS results suggest another avenue by which surgical outcomes can be improved. Seventeen strokes occurred during the perioperative period, and 2 occurred after randomization but before carotid endarterectomy. Of the remaining 15 strokes, 5 occurred as a direct result of the carotid arteriography done on 414 of the patients having surgery. This arteriographic complication rate is higher than the 0.5% to 1% rate reported from studies based in single centers [16, 17]. Still, the ACAS results [17] may be more representative of the higher arteriographic risk in a population of largely elderly patients with severe atherosclerotic disease. The morbidity and mortality attributed to patients in the surgical arm could be decreased and the benefit of carotid endarterectomy improved by decreasing or eliminating angiographic complications. Substituting a combination of duplex ultrasonography [18] plus magnetic resonance angiography [19] could eliminate these serious complications in the evaluation of the degree of stenosis. Our data show an 8% false-positive rate for Doppler (not duplex) evaluation. We could not evaluate the false-negative rate. The literature indicates that magnetic resonance angiography tends to overestimate the degree of stenosis. However, this technique holds promise for providing a safe and relatively inexpensive way to evaluate flow through extracranial and intracranial vessels. With future improvements in these techniques, the availability of duplex ultrasonography and magnetic resonance angiography may help us to avoid invasive arteriography.

The ACAS results favoring surgery are consistent with those of the Veterans Affairs Cooperative randomized trial of carotid endarterectomy for asymptomatic carotid artery stenosis [10]. In 444 men with carotid stenosis of 50% or more shown by arteriography, the incidence of ipsilateral stroke alone was 9.4% in patients receiving medical treatment and 4.7% in those having surgery (P = 0.056). However, rates of combined stroke and death did not differ between the medical treatment and surgery groups. In the ACAS study, the Kaplan-Meier estimate of the 5-year risk for stroke or death was 31.9% in patients receiving medical treatment and 25.6% in those having surgery (P = 0.08). Two other randomized studies of carotid endarterectomy for asymptomatic carotid stenosis have been reported [11, 20], but design issues related to randomization, sample size, and duration of follow-up make comparison difficult.

Asymptomatic carotid artery disease is an important public health concern. The prevalence of stenosis of 50% or more in at least one carotid artery in patients 65 years of age or older may exceed 5% [5]. However, the overall risk for stroke is low. The European Carotid Surgery Trialists Collaborative group [5] prospectively studied 2295 patients and found a 3-year risk for ipsilateral stroke of 1.8% for asymptomatic stenosis of 0% to 29%, 2.1% for stenosis of 30% to 69%, and 5.7% for stenosis of 70% to 99%.

The benefits of surgery are modest and do not justify screening the general population. Carotid artery disease is much less prevalent in the general population than in patients with specific medical indications for carotid ultrasonography (such as those with carotid bruits [approximately 75% of the ACAS patients]; those who have had transient ischemic attack or stroke; and those with coronary and peripheral vascular disease). Although the false-positive rate for Doppler (not duplex) ultrasonography for hemodynamically significant carotid artery stenosis is low [21], the occurrence of false-positive results could preclude the effective use of this technique for screening in the general population.

Because the risk for stroke is low in asymptomatic patients with modest degrees of stenosis [4, 5] and because no randomized trials have assessed carotid endarterectomy for asymptomatic carotid stenosis of less than 50%, we do not recommend the procedure for patients with asymptomatic carotid stenosis of less than 60%. We do consider it for patients with carotid stenosis greater than 60% as shown by ultrasound or arteriography. Carotid endarterectomy should be recommended only when the combined arteriographic and surgical mortality and morbidity are estimated not to exceed 3% and the prognosis for healthy life expectancy is at least 5 years. However, for individual patients at excess risk for carotid artery disease, such as those evaluated and treated in ACAS, carotid endarterectomy can reduce the risk for subsequent stroke.