Ninety three per cent of the delegates attending the carotid sessions yesterday voted “no” to the question “CREST is telling us to swing towards carotid stenting, Should we?” at the great debate. The results of the voting gave Michael Jenkins and Michel Makaroun clear victory over Sumaira Macdonald and William Gray Sumaira Macdonald, Newcastle upon Tyne, UK, said that CREST was one of the largest contemporary randomised trials of carotid artery stenting versus carotid endarterectomy in standard risk patients.
“It forms part of the ‘big four’ representing the current evidence base for carotid intervention in ‘standard risk’ patients the remaining trials being EVA3S, SPACE, and ICSS,” she noted. “Whilst all are important in terms of current world knowledge, a number of process issues have arguably limited the generalisability of the other three. These ‘issues’ include failure to mandate the use of embolic protection devices and adequate baseline experience in the carotid artery stenting limb of the European trials, suboptimal myocardial infarction surveillance and myocardial infarction adjudication and for EVA3S and SPACE, and early termination.” Macdonald stated that CREST was exemplary by comparison: “The use of embolic protection devices was mandatory, the interventionists were properly trained, the incidence of periprocedural myocardial infarction was carefully screened and this outcome stringently adjudicated. It ran to completion. It is perhaps one of the more robust bodies of evidence, therefore and as a result, plays an important role in the provision of recommendations,” she said.
Siding with Macdonald, William Gray, New York, USA, told delegates that the CREST data clearly tells us to “swing” toward stenting as a reasonable alternative to endarterectomy. He noted that in CREST there are important findings beyond the primary endpoint. “There were slightly more minor strokes in the stenting group and slightly more myocardial infarctions in the endarterectomy group. The long-term consequences of these two disparities had distinctly different implications: by six months, the neurological status of both endarterectomy and stenting patients with minor stroke had equalised, and there was no mortality differential. At four years, one in four patients who sustained a myocardial infarction of any magnitude had died, as compared to an incidence of one in 10 deaths among the patients who did not have an event, a statistically significant difference,” Gray said.
He added that the age differential that was first published in the New England Journal of Medicine based on the intention-to-treat analysis and suggested that there was an advantage to endarterectomy over the age of 80 “was actually incorrect, as shown by the per-protocol analysis performed by the FDA in its assessment of the data”. “In point of fact the hazard ratio of the two techniques was not different (1.01) over the age of 80, and actually only showed a significant difference under the age of 60 with an advantage going to stenting in that age group,” he noted.
Gray also said that, although CREST included both symptomatic and asymptomatic patients, subsequent pre-planned analysis showed no differences for the primary outcome regardless of the symptom status. “There were non-primary outcome ‘nuisance’ events favouring stenting. There were no cranial nerve injuries in the stenting group, and over 5% in the endarterectomy group, more than 2% of which persisted at six months and almost of which involved a motor deficit. There was an approximately 8x greater incidence of access site complication for endarterectomy,” he said. “Based on these facts and analyses the unbiased observer, but more importantly the well-informed physician, cannot exclude stenting performed by an experienced operator using embolic protection in a well-selected patient as a reasonable, and in some instances, preferable as less invasive, therapy for patients requiring carotid intervention,” Gray said.
He added that “since there has been a decade-long controversy regarding the place of carotid artery stenting, with the vascular surgical specialty community largely opposed to the concept and practice, it is now fair to state that CREST causes us to ‘swing’ away from this nihilistic approach and toward parity between the two therapies. Used in such a complimentary fashion, we have the opportunity to provide the lowest risk alternatives to our patients, who are the ultimate beneficiaries of our ‘swing’.”
CREST has been misinterpreted
Representing Frank J Veith, New York, USA, who was not able to make CX 34, Michael Jenkins, London, UK, said that as a randomised controlled trial, CREST was in many ways “exemplary”. However, he added, like most randomised controlled trials, CREST had flaws and weaknesses. “It can also be argued that CREST results have been somewhat misinterpreted by its authors to reach unjustified conclusions. More importantly, CREST results have has been further misinterpreted or ‘spun’ by others to produce conclusions that are unjustified by the data from the trial,” Jenkins said. For example, he continued, CREST formed the basis for an important part of the American Heart Association Guideline on Management of Patients With Extracranial Carotid and Vertebral Artery Disease, which was also approved by 13 other organisations. “This important guideline document reached one conclusion that ‘carotid artery stenting is indicated as an alternative to carotid endarterectomy for symptomatic patients at average or low risk of complications associated with endovascular intervention…’”
Jenkins argued that CREST was originally designed to compare stenting and endarterectomy for the treatment of moderate and high grade carotid stenosis in recently symptomatic (six months) patients. However, he said, a large number of asymptomatic patients (1,180) were also included beginning in 2005. He criticised the inclusion of myocardial infarction in the primary endpoint. “There were substantially more deaths and strokes in the stenting treated patients than those treated by endarterectomy. Only when myocardial infarctions were included were the adverse events similar in the two groups,” he said. “CREST had several additional flaws. Adding asymptomatic patients to the study diluted its power and prevented significance from being reached in some of the adverse events. The composite endpoint weakened the trial. Myocardial infarctions are not the equivalent of strokes. This is borne out by the greater degree of disability after a stroke than a myocardial infarction. There is another possible flaw in CREST. The stenting treated patients received more intensive antiplatelet therapy during and after their procedure than did the carotid endarterectomy patients,” he added.
Jenkins also noted that in CREST “the stenting operators were only vetted to entre patients after they were shown to have a high level of experience and skill. Thus the stenting results in CREST may not be representative of those generally performing the procedure in the ‘real world’. This possibility is supported by the higher adverse event rates with stenting in all reported population based studies than in CREST.”
“What about the American Heart Association Guideline and its conclusion, largely based on CREST, that “carotid artery stenting is an alternative to carotid endarterectomy” in symptomatic average and low risk patients? A Webster’s dictionary definition of alternative is “choice between two things”. This implies equivalence. In view of the described considerations about CREST, its data details and the ICSS findings, it would seem that the American Heart Association Guideline’s conclusion that stenting is an alternative to endarterectomy is not yet justified except when there are clear contraindications to endarterectomy in a symptomatic carotid stenosis patient requiring invasive treatment. High or long lesions or infected, scarred or immobile necks represent examples of such contraindications.”
Agreeing with Jenkins, Michel Makaroun, Pittsburgh, USA, quipped whether Macdonald and Gray were “swinging for CREST or swinging for cash”.
He added that the CREST primary endpoint is flawed because it is a composite that combines periprocedural myocardial infarction with stroke and death. “CREST demonstrated that myocardial infarction has almost no bearing on patient quality of life, while both minor and major stroke clearly affect multiple quality of life domains. Without myocardial infarction, the CREST clearly demonstrates endarterectomy to be superior to stenting. Patients who suffer periprocedural myocardial infarction are indeed at a higher risk of post-operative death, but the CREST investigators clearly state that there is no cause-and-effect relationship, as the myocardial infarction is likely just a marker for more severe concomitant cardiac disease.”
In CREST, Makaroun continued, asymptomatic cardiac ischaemia was included as an endpoint, but asymptomatic neurologic events were not. “Data from ICSS suggests that nearly three times as many stenting patients as endarterectomy patients have new MR imaging lesions post-procedurally,” he said. “The CREST periprocedural results are very good, but the applicability of the stenting results to a wider population outside of the study is questionable. CREST appears to be telling us that symptomatic patients, older patients, and women are better served with endarterectomy. This leaves asymptomatic, younger male patients as having equivalent outcomes with stenting and endarterectomy. These young patients are probably not well served by a new technology with limited long term follow-up. CREST is telling us many things, but it is decidedly not telling us to swing towards carotid stenting.”