Stenting or surgery for carotid stenosis: All in the interpretation?

The conclusions of the International Carotid Stenting Study (ICSS) favor surgery over stenting, reflecting those of previous trials but clashing with those of the Carotid Revascularization Endarterectomy vs Stenting Trial (CREST).

MedWire reporter Eleanor McDermid spoke to ICSS chief investigator Professor Martin Brown about the ongoing debate over how best to treat patients with severe carotid stenosis.

Martin Brown is Professor of Stroke Medicine at the Institute of Neurology, University College London, UK. He was the principal investigator of the Carotid and Vertebral Artery Transluminal Angioplasty Study (CAVATAS) and is the chief investigator of ICSS and the coordinator of the European Carotid Stenting Network.

After aborted trials and controversy over stenting operator experience, has carotid artery stenting (CAS) now proved itself against carotid endarterectomy (CEA)?

It depends on who you ask…

The devil’s in the detail

Two major randomized trials of CAS versus CEA have just released their primary findings: the American-led CREST, reported at the International Stroke Conference in San Antonio, Texas, USA, and the British-led ICSS, which appears in The Lancet.¹

On the face of it, the trials’ findings are very similar. Yet the CREST investigators conclude that CAS and CEA are broadly equivalent in terms of safety, whereas the ICSS team favors surgery.

“The trend of the [CREST] results is very similar to ours and the other studies,” says ICSS chief investigator Martin Brown (Institute of Neurology, London, UK). “They’re quite similar, but they put a very different interpretation on the results than we have.”

CREST randomly assigned 1321 symptomatic and 1181 asymptomatic patients to undergo CAS or CEA, while ICSS included 1713 patients, all symptomatic.

Periprocedural stroke, death, and myocardial infarction (MI) rates between the CAS and CEA groups were similar in CREST, at 90-day rates of 5.2% and 4.5%, respectively. But in ICSS, the corresponding rates between randomization and 120 days were significantly higher in the stenting group, at 8.5% versus 5.2% with surgery.

Both trials reported significantly higher stroke rates after stenting than after surgery, at 4.1% versus 3.2%, representing a 79% increased risk, in CREST, and 7.7% versus 4.1%, amounting to a 92% relative increase, in ICSS.

But the CREST investigators report an excess of MIs in the CEA group, at 2.3% compared with 1.1% in the CAS group. MI rates in ICSS were very low (less than 1%) and similar in both groups.

This discrepancy is probably because all the CREST patients had post-intervention electrocardiograms and cardiac enzyme measurements, whereas only clinical events were recorded in ICSS. Brown thinks that, had the trials used the same approach to detecting MIs, they would have probably achieved similar results.

It is not clear why MIs would be more frequent in the surgical than stenting group. Brown suggests use of local anesthesia may contribute: the amount required for CEA may depress the heart. “I don’t think the stenting uses enough local anesthesia to affect the heart – probably why it saves some heart attacks.”

But he says: “It’s not fair to say… there were more strokes with stenting but there were less heart attacks and therefore they cancel each other out. They probably don’t.”

Indeed CREST, which followed-up patients for 4 years, found a poorer quality of life among patients who suffered a stroke than among those who suffered MI.

“It’s not fair to say… there were more strokes with stenting but there were less heart attacks and therefore they cancel each other out. They probably don’t.”

Subclinical evidence

The main purpose of carotid intervention is to prevent stroke, so the ICSS team took a thorough approach to detecting periprocedural strokes. In a magnetic resonance imaging (MRI) substudy, 231 patients underwent imaging before and after intervention. The results backed the main trial findings of an increased stroke risk with stenting.²

Brown explains that this substudy helped to counterbalance some of the inherent weakness of a trial of surgery versus endovascular treatment. “Unlike a drug trial, it’s very difficult to do these trials in a double-blind fashion,” he says.

Blinding the investigators is near-impossible, “because obviously the patient who’s had the operation has a scar in their neck,” says Brown.

“And more importantly we couldn’t keep the patients blinded either. If they decided they didn’t like the surgeon, or they didn’t like the radiologist who did the stenting, they could say, you know, they’d had a bad time.”

This creates a risk for ascertainment bias, where an investigator’s diagnosis of clinical events may be influenced by their knowledge of which treatment a patient has had. MRI scans, in contrast, can be reviewed by investigators blinded to the identity of the patient and to which treatment they have had. So the MRI substudy served to support the clinical findings.

And Brown adds: “The other important reason for the MRI study [is that] it allowed us to pick up damage to the brain that hadn’t caused any symptoms.”

This proved considerable. Among patients in the substudy, 8% of the CAS group and 5% of the CEA group suffered clinical strokes (symptoms lasting at least 24 hours), but 50% versus 17%, respectively, had new cerebral lesions on post-intervention diffusion-weighted imaging scans.

Some study centers had a policy of using embolic protection devices. But new lesions appeared in 73% of CAS versus 17% of CEA patients at these centers, whereas the corresponding rates at centers that did not use the devices were 34% versus 16%.

To position an embolic protection device, the stenting operator has to guide it up the aorta and then into the carotid artery and across the target lesion of a patient who likely has systemic atherosclerosis. Thus, there is a risk for dislodging particles of plaque, which can then find their way into the cerebral circulation and cause a stroke.

In contrast, the surgeon immediately clamps the carotid artery above the lesion, barring anything he might dislodge from reaching the brain.

“I think it’s just that the surgical protection is much more effective than the way in which the stenters can protect the brain during the procedure by putting filters or other things across, because they still have to cross the lesion before they’ve got their protection device in place,” says Brown.

The human factor

To date, there have been five large randomized trials of CAS versus CEA. The Stenting and Angioplasty with Protection in Patients at High Risk for Endarterectomy (SAPPHIRE) trial, demonstrated noninferiority of CAS to CEA, but was restricted to patients at high surgical risk.³

The Stent-supported Percutaneous Angioplasty of the Carotid artery versus Endarterectomy (SPACE) trial failed to demonstrate noninferiority of CAS to CEA in a lower-risk population, but did not quite close its case, because while the absolute difference between the two groups was within the investigators’ predefined limits for noninferiority, the upper confidence interval exceeded them, leaving room for argument.^4,5

The Endarterectomy versus Angioplasty in Patients with Symptomatic Severe Carotid Stenosis (EVA-3S) trial was halted prematurely because of stroke or death rates of 9.6% in the stenting group compared with 3.9% in the CEA group.⁶ But the trial was criticized, among other things, for requiring its stenting operators to have experience of just two procedures.⁷

Operator experience has been a hot topic in the CAS versus CEA debate, with SPACE reporting that CAS outcomes improved with increasing patient enrollment, whereas CEA outcomes did not.⁸ Recently, a multidisciplinary group of Italian physicians agreed that stenting operators need to have completed 150 procedures before they can be considered safe.⁹

Brown is somewhat skeptical of this view. “There is no point in developing a technique that requires people to experiment on 150 patients before they get good at it,” he says.

“I mean, it may be so, but if it takes 150 procedures before you’re safe at it, nobody should ever start, should they? And what it’s showing is those people who have done 150 cannot transfer their skills.”

He adds: “One of the issues is that if you do a lot of patients, you’re probably doing a lot that didn’t need treatment, and their risks may be lower.”

Still, both ICSS and CREST enforced stringent measures to ensure high levels of expertise among their operators. ICSS centers with an operator who had performed at least 50 stenting procedures, with at least 10 in the carotid artery, were considered experienced. Those that did not meet these criteria were supervised by an outside operator.

“We can be pretty sure that operator experience didn’t affect our trial,” says Brown. “But of course what we can’t measure is operator skills, which is a different thing.”

ICSS did, in fact, suspend two investigators (both stenters) who were deemed to have substandard outcomes. But Brown stresses that the people involved may have just been unlucky.

“We can be pretty sure that operator experience didn’t affect our trial. But of course what we can’t measure is operator skills, which is a different thing.”

CREST took an even more rigorous approach to ensuring the quality of its stenting operators, requiring them to perform 50 carotid interventions in a run-in phase that was not included in the trial.

“I think they were taking the top people and comparing them against surgery,” says Brown. “We took the average person and compared them against surgery.”

This could be one reason for the at-odds conclusions of the two trials, he says. “I think it’s a difference of approach, the kind of British style of trial tends to say, well let’s look at NHS [National Health Service] practice and get a result that’s relevant to the average NHS practice.”

And he stresses that the ICSS findings represent the average for 50 centers. “The idea of averaging 50 centers is we say this is a result you can apply to the average hospital center.

Although I wouldn’t say they were, if you like, the ‘average’ hospital center, because people who take part in trials are usually higher quality than actually in your average.”

This throws up a potential problem with the CREST approach.

“If you have very strict criteria for who can join a trial, then you will end up having a result that’s applied outside the very expert centers, and therefore it will actually be wrong for the average center,” says Brown.

So the CREST results have not swayed him. “I still think surgery is still the treatment of choice for suitable patients.”

Even if the two procedures did have similar clinical risks, “there’s still this problem of the MRI study showing more subclinical damage to the brain,” he notes.

However, Brown does stress that stenting for symptomatic stenosis is proven to be better than best medical treatment. “Despite the fact we’ve known the results of [ICSS] for a little bit of time we still do stent some patients for particular reasons – they can’t have surgery, or some types of lesions will be more suitable for stenting. We know for example that if someone’s got narrowing from radiotherapy, stenting seems to be much safer than surgery.”

“If you have very strict criteria for who can join a trial, then you will end up having a result that’s applied outside the very expert centers, and therefore it will actually be wrong for the average center.”

But Brown concedes that CREST has shown that stenting can be done at a very low overall complication rate. “I think we put it as a challenge to people who still think stenting is an option to prove it, and to show that they can train people up like the CREST investigators.”

In the real world, he notes, some centers may have a far better stenter than surgeon, while in others the reverse may be true.

“But it looks that the best stenter is never going to be better than the best surgeon. And I think that’s kind of the bottom line of that argument.”