CREST Trial Published, but Debate Continues on Stenting vs Surgery for Symptomatic Carotid Disease

May 26, 2010

May 26, 2010 — Final results of the Carotid Revascularization Endarterectomy versus Stenting Trial (CREST), showing similar net outcomes with carotid artery stenting (CAS) and carotid endarterectomy (CEA) for the treatment of carotid stenosis, are now published, appearing online May 26 in the New England Journal of Medicine.

CREST was the largest prospective randomized trial to date comparing these 2 interventions, enrolling 2502 patients from 117 US and Canadian centers.

For the composite primary endpoint of any stroke, myocardial infarction (MI), or death during the periprocedural period or ipsilateral stroke on follow-up, stenting was associated with a 7.2% rate of these events vs 6.8% with surgery, a nonsignificant difference.

However, individual risks varied. At 30 days, the rate of stroke was significantly higher with stenting at 4.1%, vs 2.3% with surgery, although major stroke was not different at less than 1% in both groups.

Conversely, MI was higher with CEA at 2.3% vs 1.1% with stenting, again a statistical difference. Patients who had an MI, though, reported a better quality of life after recovery than those who had a stroke, the study authors noted.

Rates of ipsilateral stroke during a mean follow-up of 2.5 years were equal between groups at 2.0% for stenting and 2.4% with surgery. The investigators also report an effect of age, with younger patients having slightly fewer events with CAS than CEA and older patients having fewer events with surgery.

The study was supported by the National Institute of Neurological Disorders and Stroke, with supplemental funding by Abbott. Preliminary results were presented at the American Stroke Association's International Stroke Conference (ISC) 2010 in February 2010 and reported by Medscape Neurology at that time.

"I think these CREST results show that physicians now have 2 ways to treat carotid artery disease, the tried and true surgical approach and the new kid on the block carotid stenting," principal investigator Thomas G. Brott, MD, professor of neurology and director for research at the Mayo Clinic in Jacksonville, Florida, told Medscape Neurology.

For symptomatic patients, he noted, "results with both procedures are excellent with regard to safety and durability out to 4 years." Similarly for asymptomatic patients, "both procedures are exceptionally safe and, again, exceptionally durable."

Sticking With Surgery in Symptomatic Patients

However, in an editorial accompanying the new publication, Stephen M. Davis, MD, from the Department of Neurology at Royal Melbourne Hospital and the University of Melbourne, and Geoffrey A. Donnan, MD, also from the University of Melbourne as well as Florey Neuroscience Institutes and Austin Hospital, all in Melbourne, Australia, conclude that for now surgery is their treatment of choice, at least for patients with symptomatic carotid stenosis.

Despite being among the largest of the randomized trials, with "impressively low complication rates," results of CREST in the periprocedural period are similar to those seen in previous trials of symptomatic patients, the Endarterectomy versus Angioplasty in Patients with Symptomatic Severe Carotid Stenosis (EVA-3S), the Stent-Protected Angioplasty versus Carotid Endarterectomy (SPACE) trial, and, most recently, the International Carotid Stenting Study (ICSS).

"Namely, carotid-artery stenting is associated with a higher periprocedural risk of stroke or death, a difference that was still significant at 4 years," Dr. Davis and Dr. Donnan write. This finding was echoed by another recent meta-analysis of 11 randomized trials, not including CREST, that found endarterectomy was superior to stenting for short-term but not long-term outcomes.

"The crux of the debate about the CREST results is this: Can periprocedural stroke and myocardial infarction be considered equivalent events in terms of longer-term health implications? We think not." Post hoc analyses of health status at 1 year using the 36-Item Short-Form Health Survey summary scales confirmed that major and minor stroke had effects on physical and mental health; MI did not.

However, they acknowledge that "the intriguing effect of age" may "indicate greater equipoise in the choice of interventions in younger patients."

In the end, the risk-benefit issue is complex, they write, and should be discussed with patients. "We conclude that until more data are available, carotid endarterectomy remains the preferred treatment for most patients with symptomatic carotid stenosis; treatment for asymptomatic stenosis remains controversial. However, given the lack of significant difference in the rate of long-term outcomes, the individualization of treatment choices is appropriate."


The CREST compared CAS and CEA for stroke prevention in patients with both asymptomatic and symptomatic extracranial carotid stenosis. The trial was preceded by a National Institute of Health–funded lead-in phase meant to credential interventionalists in what was at the time the fairly new and evolving technology of carotid stenting.

The trial included 1321 symptomatic and 1181 asymptomatic patients who were randomized to receive either stenting using the same stent and distal protection devices (Acculink and Accunet devices) or CEA.

Patients with symptomatic stenosis had 50% or greater stenosis by angiography, 70% or greater by ultrasonography, or 70% or greater by computed tomography (CT) or magnetic resonance angiography (MRA), a newer imaging technique that was used later in the trial for enrollment purposes. Asymptomatic patients had lesions identified by any of angiography (≥60%), ultrasonography (≥70%), or CT/MRA (≥80%).

The primary endpoint was a composite, including any clinical stroke, MI, or death during the periprocedural period, plus ipsilateral stroke on the vessel that was treated with patients followed up out to 4 years. Endpoints were blindly adjudicated by 2 independent neurologists.

No significant difference between the 2 treatment groups was seen on the primary endpoint, and Kaplan-Meier curves confirmed most events were periprocedural.

Table 1. CREST: Primary, Secondary, and Safety Endpoints

Endpoint CAS CEA Hazard Ratio (95% CI) P Value
Primary endpoint ≤4 years 7.2 6.8 1.11 (0.81 – 1.51) .51
Primary endpoint: periprocedural components 5.2 4.5 1.18 (0.82 – 1.68) .38
Periprocedural stroke and MI        
Any periprocedural stroke 4.1 2.3 1.79 (1.14 – 2.82) .01
Periprocedural major stroke 0.9 0.6 1.35 (0.54 – 3.36) .52
Periprocedural MI 1.1 2.3 0.50 (0.26 – 0.94) .03
Cranial nerve palsies 0.3 4.7 0.07 (0.02 – 0.18) <.0001
Ipsilateral stroke after periprocedural period ≤4 years 2.0 2.4 0.94 (0.50 – 1.76) .85

CAS = carotid artery stenting; CEA = carotid endarterectomy; CI = confidence interval; CREST = Carotid Revascularization Endarterectomy Versus Stenting Trial; MI = myocardial infarction

Enrollment in CREST began with only symptomatic patients but was slow at the outset, and about midway through the trial, asymptomatic patients were also added.

The new publication includes more precise data on outcomes in these groups not released initially at the ISC meeting; at that time, CREST investigators said only that the 2 groups had similar outcomes to the overall findings.

Table 2. CREST: Outcomes in Patients With Symptomatic vs Asymptomatic Carotid Stenosis in Periprocedural Period

Endpoint Absolute Treatment Effect of CAS vs CEA, % (95% CI) Hazard Ratio (95% CI) P Value
Primary endpoint      
Asymptomatic 0.0 (−2.2 to 2.1) 1.02 (0.55 – 1.86) .96
Symptomatic 1.4 (−1.2 to 3.9) 1.26 (0.81 – 1.96) .30
Any periprocedural stroke or postprocedural ipsilateral stroke      
Asymptomatic 1.2 (−0.4 to 2.7) 1.88 (0.79 – 4.42) .15
Symptomatic 2.3 (0.1 to 4.5) 1.74 (1.02 – 2.98) .04
Myocardial infarction      
Asymptomatic −1.0 (−2.5 to 0.4) 0.55 (0.22 – 1.38) .20
Symptomatic −1.2 (−2.6 to 0.1) 0.45 (0.18 – 1.11) .08

CAS = carotid artery stenting; CEA = carotid endarterectomy; CI = confidence interval; CREST = Carotid Revascularization Endarterectomy Versus Stenting Trial

Table 3. CREST: Outcomes in Patients With Symptomatic vs Asymptomatic Carotid Stenosis at 4 Years

Endpoint Absolute Treatment Effect of CAS vs CEA, % (95% CI) Hazard Ratio (95% CI) P Value
Primary endpoint      
Asymptomatic 0.7 (−2.1 to 3.4) 1.17 (0.69 – 1.98) .56
Symptomatic 0.2 (−3.0 to 3.4) 1.08 (0.74 – 1.59) .69
Any periprocedural stroke or postprocedural ipsilateral stroke      
Asymptomatic 1.9 (−0.5 to 4.3) 1.86 (0.95 – 3.66) .07
Symptomatic 1.2 (−1.8 to 4.1) 1.29 (0.84 – 1.98) .25

CAS = carotid artery stenting; CEA = carotid endarterectomy; CI = confidence interval; CREST = Carotid Revascularization Endarterectomy Versus Stenting Trial

At the European Stroke Conference 2010 here, Ariane Mackey, MD, a neuroradiologist at the Quebec Heart Institute, Laval Hospital, Quebec, Canada, the third-leading recruiting center, will present these results for symptomatic vs asymptomatic patients in more detail on Friday morning.

Stenting or Surgery?

Interpretation of stenting vs surgery trials for carotid stenosis has become something of a hot button issue in the stroke community. Immediately after publication of ICSS and presentation of CREST, there was a heated exchange of correspondence on interpretation of results across trials.

The upshot is many are still not convinced that the 2 approaches are essentially comparable or have been at least reserving their judgment until the full publication of CREST became available.

Some, like Dr. Davis and Dr. Donnan, feel that the results in aggregate favor surgery, at least for symptomatic patients, and others have raised the issue of whether with best medical therapy and given the risks of intervention, asymptomatic patients should undergo revascularization at all.

Interventionalists, on the other hand, are criticizing EVA-3S, SPACE, and ICSS on the basis of the inexperience of operators allowed to participate in the stenting arms and the spotty use of distal protection.

The night before the scheduled presentation by CREST investigators in February, the ICSS interim 120-day safety results were published online in The Lancet and appeared in the March 20-26 issue, showing higher rates of stroke, death, or periprocedural MI in patients treated with stenting relative to surgery. The primary result of ICSS, the 3-year rate of fatal or disabling stroke in any territory, is expected in 2012.

A commentary accompanying publication of the ICSS results in The Lancet waited until data from CREST were available, allowing its author, Peter M. Rothwell, MD, from the University Department of Clinical Neurology at John Radcliffe Hospital, in Oxford, United Kingdom, to discuss results of both trials, as well as the ICSS substudy showing an excess of acute ischemic lesions on diffusion-weighted imaging after carotid stenting (Bonati et al. Lancet Neurol. Published online February 25, 2010).

"What are the clinical implications of these new data?" Dr. Rothwell writes. "Pending more detailed reporting of the CREST results and meta-analysis with ICSS and previous trials, routine stenting in symptomatic patients must now be difficult to justify, in view of the highly consistent excess absolute procedural risk of stroke (around 3%)," he noted. "Indeed, stenting might be no better than medical treatment alone in patients with moderate symptomatic stenosis...for which endarterectomy was of only modest benefit compared with medical treatment in the major trials, particularly the risk of disabling stroke and death" (Rothwell. Lancet. 2010;375:985-997).

In a previous editorial after release of the Carotid and Vertebral Artery Transluminal Angioplasty Study (CAVATAS) registry results, Dr. Rothwell proposed that given poor outcomes seen with endovascular treatment in patients with symptomatic carotid stenosis, that it was "time for a moratorium" ( Lancet Neurol. 2009;10:871-873).

Crossing Editorial Swords on Credentialing

Simultaneous to the online publication of ICSS on the evening of February 25, but before presentation the next day of the CREST results, correspondence critical of EVA-3S, SPACE, and ICSS from well-known interventionalists appeared online at Lancet Neurology, with first study author Marco Roffi, MD, from the University Hospital and faculty of medicine, Geneva, Switzerland, and coauthored by William Gray, MD, Jay Yadav, MD, and Horst Sievert, MD, among others.

In their letter, Dr. Roffi and colleagues question whether EVA-3S, SPACE, and ICSS constitute adequate comparisons of stenting and endarterectomy. These trials, they assert, "should be considered not just scientifically, but ethically questionable because the endovascular experience required for interventionalists to be eligible for the studies was minimal. As a consequence, patients allocated to CAS might have been exposed to unnecessary risk and the assessment of safety and efficacy of the endovascular approach confounded," they write.

For example, in EVA-3S, operators were required to have performed 12 or more CAS procedures or 5 CAS procedures and 30 endovascular procedures in the supra-aortic trunk. In SPACE, a minimum of 25 procedures were required, although the protocol was amended, they note, to allow for tutoring of operators who did not have enough experience. In ICSS, that number was 50 total stenting procedures, of which only 10 were required to be carotid procedures.

The rationale for these "limited" requirements "might have been the absence of procedural insight or concerns about funding or slow enrollment," Dr. Roffi and colleagues speculate. "Regardless, no surgeon would have ever allowed a peer with a total lifetime experience of just 10 CEA procedures to participate in a randomized trial," they note, nor do they meet current minimum recommendations.

Dr. Roffi and colleagues concluded that the 3 trials provide an inadequate comparison of these treatments. "Their findings run counter to our experience and the outcomes of thousands of patients observed in independently monitored and adjudicated CAS registries closely monitored by the US Food and Drug Administration," they write. "Therefore, although we agree on a moratorium of randomized clinical testing as has been done so far, we encourage clinical investigation to be pursued in both surgical and endovascular high-volume centers" (Lancet Neurol. 2010;9:339-341).

The Authors' Reply

For their part, authors, including Martin M. Brown, principal investigator of ICSS, and Werner Hacke, principal investigator of the SPACE trial, called the assertion by Dr. Roffi and colleagues that their trials were ethically questionable a "bold step, considering that the trials received research ethics approval in about 20 countries."

They conclude that Dr. Roffi and colleagues "seem to misunderstand the goal of randomized trials."

"EVA-3S, SPACE, and ICSS were not designed to test the very best interventionalists against the very best surgeons," they write. "Instead, they answered the question more relevant to patients and health service providers: are the results equivalent when the average interventionalist treats the patient compared with the average surgeon? The trials have convincingly shown that, on average, carotid surgeons do revascularization in symptomatic patients more safely than do interventionalists.

The trials have convincingly shown that, on average, carotid surgeons do revascularization in symptomatic patients more safely than do interventionalists.

"This does not mean that the best interventionalist could not do revascularization more safely than the less good surgeon, but it does make it unlikely that the best interventionalist would do the procedure more safely than the best surgeon," they concluded (Lancet Neurol. 2010;9:341-342).

Interpretation for Practice?

Finally, Medscape Neurology asked other stroke neurologists about how they interpret the CREST results.

Ralph Sacco, MD, professor and Olemberg Family Chair in Neurological Disorders at the University of Miami Miller School of Medicine in Florida, says that in his view, unfortunately, more information is still needed to make these decisions.

"Although we hoped that a single, large, well-designed trial such as CREST could answer all the questions regarding the interventional treatment of carotid stenosis, that is seldom the case in medicine," Dr. Sacco told Medscape Neurology

"We already have the time-honored, evidence-based treatment of carotid endarterectomy and increasing evidence that carotid angioplasty and stenting in the right patient with the best operators will offer some benefits particularly in younger patients," he said. "More data from all of these trials will be needed to help us guide the care of our patients," he added.

Philip B. Gorelick, MD, from the University of Illinois College of Medicine in Chicago, said that CREST was a rigorous trial, using a single device and "excellent medical management" that should make it the benchmark for angioplasty and stenting trials for carotid stenosis.

"The findings, under these very rigid and carefully done circumstances show similarity," Dr. Gorelick told Medscape Neurology. "There's a little bit of trading that's going on here in terms of risk, so when you look at net clinical benefit on the one hand, you have fewer strokes with carotid endarterectomy by almost a factor of 2 — getting close to that, but not quite — but then we're trading off more myocardial infarctions with carotid endarterectomy."

He said he was a bit surprised at more MIs seen in the surgery group and speculated that this might relate to an increased sensitivity of assays to detect infarctions. More information on the type and severity of MIs would be helpful.

"The good news is that they found similar results, which means we have a new alternative for therapy," Dr. Gorelick said. "I think people need to pay careful attention to this age cutoff, because I think that's going to drive a lot what we do. I'm not as of yet totally convinced about the durability issue because they went about 2.5 years on average, and they went up to 4 years, but endarterectomy has been followed out to 8 years or more, so I think we have a little more ways to go."

"In my own personal practice, I've got patients who've had angioplasty and stents for 10 years now, and they've been doing well, so I have reason to believe it will be durable, but we still need to make sure," he added.

Other good news from the findings are that both women and men respond, as well as both symptomatic and asymptomatic patients, he noted. "However, I would say in the asymptomatic group, I don't think we should be running to do angioplasty and stents in these patients. We learned from the ACAS study [Asymptomatic Carotid Atherosclerosis Study] and the other asymptomatic trials for endarterectomy, that there's a very modest benefit."

Disclosure documents for all authors and editorialists are available at

N Engl J Med. Published online May 26, 2010.

XIX European Stroke Conference: Presenting Friday, May 28, 2010, Large Clinical Trials B Session.