Reflections by Contrarians on the Post-CREST Evaluation of Carotid Stenting for Stroke Prevention

There are few human vessels that have been studied as intensively as the carotid artery in relation to the diagnosis, prevention and treatment of ischaemic stroke. Since the beginning, controversies have raged about the best medical and surgical management. With the publication of the North American Symptomatic Carotid Endarterectomy trial, the validity of surgical carotid endarterectomy (CEA) for stroke prevention was firmly established (1, 2). The advent of less invasive percutaneous interventional procedures led to carotid angioplasty and stenting (CAS) for the treatment of carotid stenosis. Over the last 20-years, a large clinical experience with the procedure has led many to claim that it is a safe and effective alternative to CEA. As the technology of balloons, stents, and embolic protection devices has improved, the purported indications for CAS have broadened to include patients with lower risk, moderate stenoses (50–69%) and those with lowest risk, asymptomatic disease.

The early evidence in favour of CAS was based on large case series, often from single centres with heterogeneous patient populations. More recent randomized-controlled trials (RCTs) of CEA vs. CAS have provided contradictory evidence. The Carotid and Vertebral Artery Transluminal Angioplasty Study (CAVATAS) found no difference between CAS and CEA, although the 30-day stroke and death rates of approximately 10% were unacceptably high in both arms (3). The Stent-Protected Angioplasty Versus Carotid Endarterectomy (SPACE) trial showed comparable safety with both procedures, but failed to show noninferiority of CAS vs. CEA (4). The Stenting and Angioplasty with Protection in Patients at High Risk for Endarterectomy trial (SAPPHIRE), which led to the institutional approval of CAS in the United States, showed that CAS was safer than CEA in a high-risk group of predominantly asymptomatic patients (72%) with severe carotid stenoses (5). The difference in safety was due to the higher incidence of nondisabling myocardial infarcts (MIs) in the CEA group, but no difference in the risk of stroke and death between the groups was shown. The Endarterectomy Versus Angioplasty in Patients with Severe Symptomatic Carotid Stenosis (EVA-3S) and the International Carotid Stenting Study (ICSS) both found significantly increased 30-day stroke and death rates with CAS vs. CEA (9.6% vs. 3.9% in EVA-3S, 8.5% vs. 5.1% in ICSS (6, 7). This prompted some investigators to suggest that CAS is no better than medical therapy, and that a moratorium should be placed on its performance (8).

Clinicians have eagerly awaited the publication of the largest RCT, The Carotid Revascularization Endarterectomy versus Stenting trial (CREST). This North American study began recruiting symptomatic patients in 1999. Because recruitment was slow, the trial allowed entry of the substantially lower risk asymptomatic patients beginning in 2005. This was not the original design of the trial; asymptomatic subjects were added as an expediency to recruit the originally planned numbers albeit of the higher risk patients. The recently published results (9) in this evenly mixed high- and low-risk total of 2502 patients, with a mean follow-up of 2.5-years, showed no significant difference between CAS and CEA in the estimated 4-year rates of the primary end points of stroke, death and nonfatal MI (7.2% vs. 6.8%). There were more strokes in the CAS group (4.1% vs. 2.3%) and more MIs in the CEA group (2.3% vs. 1.1%). Of serious concern is the fact that the investigators violated the principle of Ceteris Paribus, essential to a credible trial, which requires ‘all other things being the same’ when comparing two entities (10). Symptomatic and asymptomatic carotid disease are not the same. It requires quite a leap of faith to believe that the <1500 asymptomatic subjects eventually included are an adequate sample size to demonstrate the equivalence of treatment strategies. Previous calculations have placed this number at >10 000 (M. Eliasziw, personal communication).

The popular press, some medical publications, and many clinicians in the United States have heralded the CREST results as decisively showing equivalence between CEA and CAS. Criticisms of CREST include the use of nonfatal MI as a primary endpoint, equating it with stroke, and the fact that 47% of the patients were asymptomatic.

Arguments have been made that asymptomatic carotid disease rarely, if ever, deserves either CAS or CEA. The major trials comparing medical management with surgery showed that complication rates from CEA must be <3% if they have no symptoms to achieve any clinical benefit and that the number needed to treat to prevent one stroke in 2-years is between 67 and 83. The largest trials of CEA in subjects with asymptomatic carotid stenosis found only a 5.9% absolute risk of stroke reduction at 5-years, compared with 15.6% absolute risk reduction in symptomatic patients. Medical management of asymptomatic carotid disease, particularly the widespread use of statins, has markedly reduced the annual stroke risk in these patients to <1%, well below the risk of stroke or death in asymptomatic patients in CREST of 4.5% for CAS and 2.7% for CEA (11).

There is relatively limited long-term follow-up information available regarding the durability of CAS. The report of the 10-year follow-up of CAVATAS patients showed a threefold increase in the risk of severe (>70%) restenosis in the CAS group compared with CEA, although the risk of new stroke was low.

So what is the concerned clinician to do? The first task is to temper the ‘hype’ for CAS with a critical appraisal of the evidence. CREST is generally consistent with prior studies, which showed that CAS is still associated with a higher peri-procedural risk of stroke or death than CEA, a difference maintained up to 4-years. In patients over 70-years of age, CEA is clearly superior to CAS. More contemporary comparative data are needed on the medical management of asymptomatic carotid disease before either CAS or CEA can be recommended in these patients. There is a danger that CREST will be interpreted as sanctioning the continuation of the already widespread treatment of asymptomatic carotid disease with CAS, which is not warranted by the available data. There are at least three trials of intervention vs. best medical management of asymptomatic carotid disease in progress. More data are needed regarding the durability and long-term outcomes of CAS. Although local equipoise in either CAS or CEA may play a role in individual decision making, the true equivalence of the procedures has not yet been proven, as no truly rigorous comparative trials are available to guide these decisions. CEA should remain the procedure of choice for stroke prevention in patients with severe, symptomatic carotid stenosis. CAS may be a good alternative in those symptomatic patients who have major medical comorbidities precluding CEA, patients who have a restenosis after previous CEA, and those with stenosis related to prior neck radiation. The increased risk of nonfatal MI in the CREST group subjected to CEA, and the same risk in the SAPPHIRE trial clearly suggest that patients being considered for CEA or CAS require a careful preliminary cardiac evaluation. Carotid procedures may trigger MIs but cannot prevent them. Thus, it does not make sense to include them in the primary outcome category.

In conclusion, the authors fail to accept that the evidence produced by CREST should lead to an increased use of CAS for the treatment of asymptomatic carotid disease. In this era of proposed need to reduce medical expenditures in the public sector, efforts in this direction should come to a halt.