|Home | About | Journals | Submit | Contact Us | Français|
The Carotid Revascularization, Endarterectomy vs. Stenting Trial (CREST) completed randomization on July 18, 2008. Sponsored by the National Institute of Neurological Disorders and Stroke (NINDS), the trial has enrolled 2,522 participants across North America and is the largest randomized clinical trial (RCT) comparing the efficacy of carotid artery stenting (CAS) to carotid endarterectomy (CEA). It is also the largest RCT to assess carotid revascularization in both symptomatic and asymptomatic patients with carotid artery stenosis. Conventional-risk patients with symptomatic carotid stenosis (≥50% by angiography, ≥70% by ultrasound) or asymptomatic carotid stenosis (≥60% by angiography, ≥70% by ultrasound) were randomized to both treatment arms in a 1:1 ratio. Eligibility criteria for CREST were similar to those of the previous NINDS-sponsored CEA RCTs. The investigational devices used in the CAS arm of the study are the RX Acculink® stent and the RX Accunet® embolic protection system, (Abbott Vascular). The primary aim is to contrast the efficacy of CAS versus CEA in preventing stroke, myocardial infarction, and all-cause mortality during a 30-day peri-procedural period, and ipsilateral stroke over the follow-up period (extending up to 4 years). The secondary aims are to contrast the efficacy of CAS and CEA in men and women, the restenosis rates of the two procedures, health-related quality of life, and cost effectiveness of CAS and CEA. The conclusion of enrollment in CREST marks the end of a long recruitment period from 117 community and academic hospital centers across the United States and Canada. Each surgeon and interventionalist underwent a rigorous credentialing process that included performance-assessment of prior CEA and CAS procedures. Credentialing of interventionalists also included a review of additional CAS procedures enrolled into a CREST Lead-in phase prior to entering patients into the randomized trial; 1564 patients were enrolled in the lead-in, the final pathway for the largest credentialing effort to date for any clinical trial. CREST will provide long-term follow-up after carotid revascularization based on systematic ultrasonographic and neurologic surveillance, and on quality of life and cost-effectiveness comparisons between CAS and CEA in the setting of a RCT. We present a brief description of the CREST protocol, impediments that were overcome during the trial, salient results from the lead-in phase of the trial, a summary of enrollment activities and characteristics of the final cohort, and a timeline for anticipated results from the randomized phase.
The Carotid Revascularization Endarterectomy vs. Stenting Trial (CREST), funded by the National Institute of Neurological Disorders and Stroke (NINDS), is a randomized clinical trial (RCT) comparing the efficacy of carotid artery stenting (CAS) versus carotid endarterectomy (CEA) in conventional-risk patients with extracranial carotid artery stenosis. Recent randomized trials have not definitively resolved whether the efficacy of CAS and CEA are different1–5. CREST is the largest study of its kind and adequately powered to detect clinically significant differences between the two procedures. It is the only study to include both symptomatic and asymptomatic patients in the same trial. Since the objective is to compare high quality CEA with high quality CAS, appropriate credentialing was performed for both groups of investigators, and the study incorporated accommodation for the learning curve associated with the new procedure. This resulted in one of the largest and most comprehensive lead-in processes ever performed prior to randomization. Finally, carotid ultrasound, cerebral angiography, and electrocardiography assessments were reviewed by their respective core laboratories and all potential endpoints were adjudicated.
The trial is therefore being conducted in the most optimal conditions achievable for a study of this magnitude and complexity. The credentialing process was concluded after the accrual of 1,564 patients in the CREST lead-in phase. The randomized trial successfully completed enrollment of its target cohort of 2,522 patients from 117 community and academic centers in North America on July 18, 2008. The protocol was approved by the Institutional Review Boards at participating sites and participants provided signed informed consent. This has been one of the most widely anticipated trials among physicians involved in the treatment of carotid disease, and the lead-in phase has already contributed important information regarding the management of carotid stenosis. This brief update reviews the available data from other trials comparing CAS and CEA, the design and some of the achievements of CREST, important conclusions drawn from the lead-in phase, characteristics of the final cohort, and a projected timeline of future results.
Published data from recent randomized studies comparing CEA and CAS have not been able to provide definitive results. Although CAS has been recommended in specialized subsets of patients6, 7 such as restenosis after CEA, radiation-induced carotid stenosis, anatomically high lesions, and in higher-risk patients, the appropriateness of its use in conventional-risk patients remains an unresolved matter. The first published multi-center trial, the Carotid And Vertebral Artery Transluminal Angioplasty Study (CAVATAS)8, recruited 504 patients. At enrollment 90% were symptomatic; 253 were randomized to surgical treatment; and 251 underwent endovascular therapy. The combined stroke and death rate at 30 days post-procedure was 9.9% for CEA and 10% for endovascular treatment, and death or disabling strokes were observed in 5.9% of CEA patients and 6.4% of endovascular patients. However, these results are not pertinent today because only 55 of the 251 patients in the endovascular group were treated with a stent, and an embolic protection device (EPD) was not used. The CAVATAS investigators have recently published results based on the long-term follow-up of the same cohort of 504 patients9. The 8-year incidence for ipsilateral non-perioperative stroke in the endovascular versus surgical groups was not significantly different (11·3% vs. 8·6%, HR 1·22, 95% CI 0·59–2·54). In a second report, the authors have compared the long-term restenosis rates in 413 patients from the original randomized cohort in whom follow-up information was available (200 had endovascular, and 213 received surgical treatment)10. They found that severe carotid restenosis (≥70%) or occlusion occurred significantly more often in patients in the endovascular versus the endarterectomy group over 5 years (30.7% vs. 10.5%, HR 3·17, 95% CI 1·89–5·32; p<0·0001). Of note, patients in the endovascular arm treated with a stent (n=50) had a lower restenosis rate versus those treated with angioplasty alone (16.6% vs. 36.2%, HR 0·43, 0·19–0·97; p=0·04). Therefore, the results may not be entirely generalizable to current clinical practice which incorporates mandatory stenting. Since duplex velocity criteria utilized to assess stented patients was the same as that for native carotid arteries, the authors cautioned that the restenosis rates may have been over-estimated in that cohort. These results underline the need for an adequately powered study comparing the results of CEA with endovascular therapy using currently established therapeutic protocols.
The Stenting and Angioplasty with Protection in Patients at High Risk for Endarterectomy (SAPPHIRE) trial1, 2 was a multi-center clinical trial that randomized 334 patients to CAS or CEA. A total of 68% of the patients were asymptomatic with stenoses >80% and 32% were symptomatic with stenoses >50%. Participants were considered high risk for CEA if they were over age 80 years or had clinically significant heart disease, severe pulmonary disease, contralateral carotid occlusion or laryngeal nerve palsy, prior radical neck surgery or radiotherapy, and post-CEA restenosis. The primary endpoint was a composite of stroke, death, or myocardial infarction (MI) within 30 days of the procedure and ipsilateral stroke or death from 31 days to 1 year post-procedure, which occurred in 12.2% of the CAS group and 20.1% of the CEA group. The two procedures were not significantly different (p=.053) and CAS was therefore deemed non-inferior to CEA. Of concern was the relatively high 30-day composite stroke and death rate for both procedures, which raised concerns regarding the expertise of operators in both arms of the study, patient selection, and whether medical therapy would have been superior in this high-risk group.
The Endarterectomy vs. Angioplasty in Patients with Symptomatic Severe Carotid Stenosis trial (EVA-3S)3 was limited to symptomatic patients with 60% to 99% stenosis. A total of 262 participants were randomized to endarterectomy and 265 were randomized to stenting. The 30-day combined stroke and death rate was higher in the CAS group at 9.6% compared to 3.9% for CEA. The results of this study have been criticized because of the potential inexperience of stent operators. Furthermore, 8.1% of the stenting procedures were performed without an embolic protection device, and significantly fewer adverse events occurred in those who had the stenting with embolic protection.
The Stent-Protected Angioplasty vs. Carotid Endarterectomy (SPACE) trial randomized 1,200 symptomatic patients with stenoses of 50% to 99% to stenting or endarterectomy. Stents were used universally, and embolic protection devices were used in 26% of the patients. The 30-day stroke or death rates were similar, 6.3% vs. 6.8% (CEA vs. CAS)5, while the 2-year stroke plus 30-day stroke and death rates were also similar 8.8 % vs. 9.5% (CEA vs. CAS)4. As an inferiority trial, however, randomization of patients was stopped because CAS was not proven to be non-inferior to CEA and possibly would not have been proven according to the trial’s authors, even with the recruitment of twice the sample size of 1,200 patients.
In this randomized trial (CREST-randomized), conventional-risk patients with symptomatic carotid stenosis (≥50% by angiography, ≥70% by ultrasound, or ≥70% by CTA/MRA if ultrasound is 50–69%) or asymptomatic carotid stenosis (≥60% by angiography, ≥70% by ultrasound, or ≥80% by CTA/ MRA if ultrasound is 50–69%) were randomized to CAS or CEA in a 1:1 ratio. Eligibility criteria for CREST were similar to those of the previous National Institutes of Health (NIH)-sponsored CEA RCTs, the North American Symptomatic Carotid Endarterectomy Trial (NASCET)11,12, and the Asymptomatic Carotid Atherosclerosis Study (ACAS)13. The CREST protocol has 90% power to detect an annual absolute difference of 1.2% in the primary endpoints between CAS and CEA. Depending on the event rate, CREST also has 80% power to detect a difference for symptomatic and asymptomatic patients separately. The investigational devices used in the CAS arm of the study are the RX Acculink® stent and the RX Accunet® embolic protection system (Figure 1), manufactured by Abbott Vascular, Santa Clara, Calif.
The primary aim in CREST is to compare the efficacy of CAS vs. CEA in preventing stroke, (MI), and all-cause mortality during a 30-day peri-procedural period, and ipsilateral stroke over the follow-up period (extending up to 4 years). Secondary aims are to: 1) describe the differential efficacy of CAS and CEA in men and women, 2) contrast 30-day peri-procedural and post-procedural morbidity and mortality, 3) contrast the restenosis rates of the two procedures, 4) evaluate differences in health-related quality of life and cost effectiveness, and 5) identify subgroups of participants at differential risk for procedural morbidity and mortality after CAS and CEA.
Prior to participating in the CREST randomized phase, each interventionalist was required to participate in the CREST lead-in credentialing phase. Interventional expertise could thus be developed and documented, and learning curve effects could be minimized. In the lead-in, patients with symptomatic or asymptomatic high-grade carotid stenosis that had conventional or high risk (as determined by the presence of medical co-morbidities, cervical radiation/surgery, high anatomic location, contralateral occlusion, or post CEA restenosis) were enrolled to undergo CAS by potential interventionalists. The same RX Acculink® / RX Accunet® devices approved for the randomized trial were used for the purposes of credentialing. Outcomes assessed were 30-day stroke, MI and death, and 1-year ipsilateral stroke. A multidisciplinary Interventional Management Committee reviewed the outcomes and credentialed prospective interventionalists. Only then could they participate in the randomized phase. Analysis of the data from the 1,564 CAS patients enrolled in the lead-in has yielded important results, some are summarized below.
All endpoints for the lead-in were reviewed by a Clinical Events Committee. For the randomized phase, all endpoints are evaluated by either the Stroke or the Myocardial Infarction Adjudication Committees, and members of these committees are blinded to the assigned treatment. Recurrent or new ischemic stroke was defined as an acute neurological ischemic event of at least 24 hours duration with focal signs and symptoms. One or both of the following could be used as confirmatory evidence but not necessary for the designation of stroke: a one-point increase on the NIH Stroke Scale or an appropriate new or extended abnormality seen on CT or MRI. Review for potential stroke was triggered by a positive transient ischemic attack/stroke questionnaire and NIH Stroke Scale as performed preoperatively and during follow-up by a study neurologist. Postoperative or procedural MI is determined by electrocardiography, enzyme abnormalities, or clinical presentation of MI. Post-procedural restenosis is determined by duplex ultrasonography with review by a core laboratory. Quality of Life and economic analyses were also carried out by a core laboratory.
CREST was funded as a multi-center RCT on January 15, 1999. The trial could not commence because despite previous assurances of coverage for CREST, HCFA, (currently CMS), denied reimbursement for CAS. In June, 2000, President Clinton issued an Executive Order directing HCFA to reimburse Medicare participants in NIH clinical trials. More than a year later, HCFA finally initiated payments in the Fall of 2001.
Around that time, cerebral EPDs began to gain wide acceptance among interventionalists. They expressed a reluctance to perform CAS without the option to use a protection device. Embolic protection was introduced into CREST, necessitating protocol amendments and new Institutional Review Board approvals. The FDA approved the incorporation of the ACCUNET™ device into CREST in August 2001. It is at this point that recruitment to CREST commenced and continued to increase from 4 to 20–40 subjects per month.
In February 2002, a product malfunction in the recovery catheter of ACCUNET™ occurred in a non-CREST case. While no adverse event occurred and the device was retrieved successfully, recruitment to CREST was halted for product investigation and recall. Once appropriate modifications to the recovery catheter and protocol were approved by the FDA, NIH, and IRBs, enrollment was resumed in June 2002.
Finally, while these delays were occurring, several competing FDA-approved trials began to affect randomization to CREST. While the inclusion/exclusion criteria of these industry-funded trials did not technically overlap, they had registry components and features that led to practical study overlap and enrollment competition (e.g. SAPPHIRE, ARCHeR, and CAPTURE).
In a trial in which two technical procedures are being compared with regard to immediate and long-term results, it was of paramount importance that the expertise of investigators be established prior to participation. Since the objective was to compare high quality CEA with high quality CAS, appropriate credentialing had to be performed for both groups of investigators. This would ensure that the outcome of the two procedures was compared under ideal circumstances, and would not be influenced by varied technical expertise in either group. The process of credentialing surgeons performing CEA was well-established through two prior NIH-sponsored clinical trials11, 13. A similar rigorous credentialing process for surgeons was successfully implemented in CREST and was described previously14.
The credentialing process for interventionalists involved a thorough evaluation of clinical facilities at a potential site by the Site Selection Committee, and an assessment of the clinical and technical expertise of the site’s interventionalists by the Interventional Management Committee (IMC) as outlined above15. The IMC consisted of a multidisciplinary group of physicians (from neurology, vascular surgery, interventional cardiology, interventional radiology, and neurosurgery). Credentialing of the interventionalists involved a unique two-step process. First, potential interventionalists were evaluated by the IMC based on their prior experience with CAS. Their patients’ evaluation, procedure reports, and outcomes were scrutinized. If deficiencies in knowledge-base or technical skills were identified and/or a high complication rate was noted, then those interventionalists were excluded from participation. The interventionalists who were deemed to have adequate prior experience were then allowed to enter the second step of the credentialing process consisting of the Lead-in phase. The lead-in phase was built into the study and was funded by the NIH. The approved interventionalists underwent training with the study devices, which included use of animal models, and performed up to 20 lead-in CAS cases. Final IMC approval to enter the randomization phase of CREST was based on safety and technical proficiency demonstrated during the lead-in phase of the trial. Once the randomized phase of the trial was completed, the credentialing process was concluded with 1,564 patients included in the CREST lead-in phase.
The CREST protocol included one of the largest credentialing efforts ever undertaken prior to initiating participation in a formal randomized trial. This resulted in the collection of demographic, procedural, and rigorous 1-year follow-up of the lead-in cohort. The analysis of this experience, which we briefly summarize below, has been presented in abstracts and manuscripts, and will be the subject of more upcoming manuscripts.
A rigorous credentialing process is feasible prior to randomization in a large RCT of CEA and CAS15. The IMC was fairly stringent in its review, insisting that the majority of operators demonstrate their expertise by performing from 5 to 20 CAS procedures in the closely-monitored lead-in phase. A minority were allowed to initiate randomization directly, based upon their high volume of CAS procedures, familiarity with the CREST study devices, and good procedural outcomes. The specialties from which interventionalists were derived are shown in Figure 2 and reflect the varied specialties involved in performing CAS procedures in the community.
The rigorous selection process and monitoring of outcomes in the lead-in phase was associated with an overall 30-day stroke and death rate of 4.4 ± 0.6%16, which compares favorably with the suggested Society for Vascular Surgery7 and American Heart Association17, guidelines for carotid revascularization. Detailed results of symptomatic and asymptomatic patients are presented in Table I.
The lead-in phase has been instrumental in highlighting subsets of patients that may be particularly high-risk for CAS. In the process, it has helped clinicians to improve patient-selection and therefore, outcomes after CAS. One of the earliest trends observed and reported from the lead-in phase was the relatively high morbidity obtained by CAS performed on elderly patients18. The results were counter to the expectations of some that the minimally invasive procedure would offer less cardiac morbidity in this cohort. The association of stroke with age is consistent and the adverse events were observed to rise with each decade (Figure 3). While there was not a surgical comparison in the lead-in phase, the safety concern specific to CAS was confirmed in the SPACE trial where the 30-day rates of stroke and death for stenting were significantly higher with advancing age, being 11.1% for stenting and 7.5% for endarterectomy in those aged 75 years or older. These results have instilled a cautionary note for CAS in the elderly. Several mechanisms are potential mediators of this effect, including carotid arterial tortuosity and calcification, as well as increased aortic arch angulation and atheromatous involvement associated with aging. It remains to be seen how age will interact with the outcomes between CEA and CAS in the randomized cohort of CREST patients since CEA may cause increased adverse events in the elderly from anesthetic stress.
Factors that may increase the risk of periprocedural complications associated with CAS include anatomic considerations such as a type III aortic arch, severe aortic or carotid calcification, arterial tortuosity, high-grade stenosis and intra-arterial thrombus. The potential relationship of these factors to clinical outcomes in the lead-in cohort, are being analyzed. These data emphasize the importance of appropriate lesion sizing and accuracy of stent deployment during carotid stenting.
Several CAS procedures have been associated with hemodynamic instability during or after the stenting. Since the long-term significance of these events was not well-defined, the CREST lead-in data were evaluated to determine the risk of stroke or death associated with hemodynamic and hemorrhagic events occurring within seven days of CAS19. Hemorrhagic complications included groin or retroperitoneal bleeds and excluded cerebral hemorrhage. Hemodynamic events were defined as having hypotension. There were 21 patients with either a bleed or hematoma, and their 30-day stroke and death rate was 19.6% vs. 4.3% in those without. Peri-procedural hypotension occurred in 71 patients; the event rate was 4.1 % for those without hypotension and 11.4% for those with hypotension (p=0.004). These data emphasize the importance of careful technique, hemodynamic monitoring, and aggressive management of hemodynamic events during CAS. While hemodynamic events are known to occur with both procedures, the CREST randomized study will be able to report whether they are significantly different from each other.
The incidence of luminal narrowing of the internal carotid artery as measured by Duplex ultrasound (DUS) following CAS has been controversial. There is concern that altered compliance as a result of the stent may result in alterations in the recorded velocities that may overestimate the actual degree of restenosis after CAS20, 21. The CREST investigators therefore analyzed the restenosis rates observed at one year of follow-up in the CREST lead-in cohort using the one-month DUS as the baseline22. This eliminated potential stent-generated artifacts in the assessment of re-stenosis. Thirteen percent of the lead-in patients demonstrated in-stent restenosis as determined by a >2SD increase in their peak systolic velocities from the 1-month to 1-year DUS. Only 1.2% of patients however, required repeat revascularization by 1 year of follow-up. The SPACE trial has subsequently reported that recurrent stenosis of ≥70% by duplex occurred in 4.6% of patients undergoing CEA and 10.7% of those undergoing CAS (p=.0009)4. The SAPPHIRE study had noted a restenosis ≥50% rate of 19% at one year1. The lead-in results highlight the importance of this endpoint, since early periprocedural benefit with one procedure may be offset by a higher recurrence and re-intervention rate associated with that procedure in the long-term. The results of the CREST randomized study are awaited, which is adequately powered to provide evidence comparing the restenosis rates associated with CAS vs. CEA.
In the Asymptomatic Carotid Atherosclerosis Study (ACAS)13, men with asymptomatic stenosis undergoing CEA had a 66% relative risk reduction in adverse events over 5 years but only a 17% reduction occurred for women. In the North American Symptomatic Carotid Endarterectomy Trial (NASCET), no differential gender effects were seen among patients with >70% stenosis11. However, men with 50–69% stenosis had greater benefit after CEA than women12. In EVA-3S the excess risk associated with stenting on univariate analysis, was greater for men (p=0.03)3 while in SPACE there were no differences noted after CAS between genders (n=338 women)4. The causes for a differential efficacy between genders are not well understood, but none of the previous randomized trials anticipated a differential gender effect and they were therefore underpowered to address the question. Based on these results, CREST was designed with plans to evaluate the possibility of a differential gender effect. Women comprised 37% of the CREST Lead-in cohort23 and were no different from men based on age, symptomatic status, or characteristics of the internal carotid artery. The 30-day stroke and death rate for women was 4.5% (26/579) compared to 4.2% (41/985) in men. The difference in stroke and death rate was not significant nor were there any significant differences by gender after adjustment for age, anatomic features, or risk factors.
Originally, CREST had FDA approval for only 40 clinical centers to recruit patients, but study leadership realized additional sites would be needed to meet enrollment goals in a timely fashion. Maximizing the number of CREST centers was crucial in overcoming the initial roadblocks to recruitment faced by the study (Figure 4). Once the issues related to CMS, FDA and other logistics were satisfied, efforts were concentrated on identifying and activating qualifying sites, and CREST recruitment increased dramatically from 2004 onwards from under 10 patients per month to over 50 per month. The average monthly rate of enrollment between January and June of 2008 was 54. With an additional 35 patients randomized in July, CREST enrollment was completed on July 18, 2008 (Figure 5).
While the randomized results will not be available for several months, some general information is presented here. The total patients randomized in CREST were 2,522 of which 53% (n=1326) were symptomatic patients and 47% (n=1196) were asymptomatic. As of July 18, 2008, 117 centers had enrolled at least one CREST participant. Enrollment of women was 37% of total patient randomizations and minority recruitment was 9.3%. The distribution of general risk factors among those recruited mirrors that of a typical population of patients with atherosclerotic occlusive disease, with a notably high incidence of hyperlipidemia and coronary heart disease (Table II). The mean age of this cohort is 69.1 years. The adverse event rates for the cohort are not available as yet. However, it is clear that adverse events in either group of patients (CAS or CEA) were not high enough to warrant a premature closure or advisory from the independent CREST Data Safety and Monitoring Board.
Exceptional efforts are required to accomplish and complete a multi-center RCT. The success of CREST rests primarily with the enrolling clinical centers. In addition to their accomplishments, a large coordinated effort by NINDS, the CREST Administrative Center, the Recruitment Center, the Statistical and Data Management Center, the core laboratories, FDA and CMS has gone into the study. The medical community is eager for results. A timeline of the prior and projected activities related to CREST is presented in Figure 6. It is anticipated that CREST will be reporting its stated outcomes by early 2010 with consequent major practice-changing implications regarding the care of patients with carotid disease, both symptomatic and asymptomatic.
Funding: NINDS-NIH RO1 NS 038384
Additional funding was provided by Abbott Vascular. We express our thanks to the CREST Publications and Presentations Committee for their review of the manuscript.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.