The Spinning of CREST

Frank J. Veith, MD


June 14, 2010

Level 1 evidence from well-conducted, randomized, controlled trials is supposed to be the holy grail of scientific data on which to base good medical practice. The recent CREST (Carotid Revascularization Endarterectomy versus Stent Trial), which compared the outcomes of carotid stenting (CAS) and carotid endarterectomy (CEA), is one such study that was designed and conducted in an exemplary manner. It has attracted a great deal of attention, and its findings should be part of a solid and definitive basis for determining medical practice.

Yet, when the findings of CREST have been presented, the conclusions differ greatly depending on the bias of the presenter. If a vascular specialist with a bias toward CEA interprets the CREST findings, the lower stroke and mortality rates for CEA justifies the conclusion that CEA is superior and should be used preferentially to treat patients who require an invasive procedure for carotid stenosis. On the other hand, if a vascular specialist with a bias toward CAS interprets the findings, he or she will regard them as definitive and will conclude that the equivalent and low overall adverse event rates for the 2 procedures justifies the widespread and increased use of CAS to treat patients who have symptomatic or asymptomatic carotid stenosis.

The extent to which the preliminary results of CREST are being spun should be recognized; the truth lies somewhere in between these 2 diametrically opposed views. Moreover, the true significance of CREST awaits a careful analysis and subanalysis of the trial's detailed results, most of which have yet to be revealed in peer-reviewed publications.

Before determining how CREST should influence practice patterns and utilization of CEA and CAS, we need to see the data relating to the costs for the 2 procedures. More important, we need to see how much long-term disability occurred in patients who had adverse events from the 2 procedures. Was there long-standing disability or impaired function from the strokes, myocardial infarctions, or cranial nerve defects in the 2 groups of patients? What were the effects of age and other factors on the incidence of stroke, myocardial infarction, and mortality rates?

In addition, we must consider that CREST, although well designed and conducted, has its flaws. The techniques used for CAS are already outdated. For example, flow reversal and cessation methods for cerebral embolic protection, already clearly shown to be useful in some circumstances, were never used. The case could be made that current CAS results would be better than they were in CREST. Moreover, even better CAS technology, such as membrane-covered stents, is on the horizon and may further improve CAS results. Offsetting these possibilities, the CAS operators in CREST were unusually vetted and skilled. This raises the following question: Will the CAS results in CREST be generalizable if the procedure is widely adopted by vascular specialists at large? Population-based studies suggest that CAS will have substantially higher adverse event rates than those in CREST.

Another issue with CREST is that investigators included both symptomatic and asymptomatic patients. This was necessary to accumulate an adequate number of patients. In the asymptomatic patient cohort, the possibility remains that current best medical therapy, particularly with high-dose statins, could produce equal or better outcomes than both CEA and CAS, because the stroke rate in such medically treated patients ranges between only 0.4% and 0.8% per year. This possibility will remain until appropriate prospective trials are completed, such as SPACE II (Stent-Supported Percutaneous Angioplasty of the Carotid Artery versus Endarterectomy II) and TACIT (Transatlantic Asymptomatic Carotid Intervention Trial), which compare both CEA and CAS with best medical treatment. The completion of these trials will take more than 5 years, and TACIT has yet to be funded.

So how should we cut through the spin surrounding CREST and use the trial to help us as soon as possible? We should await its formal publication and the analyses and subanalyses of the data contained in these articles. We should recognize the intrinsic flaws in this well-conducted trial and view it in the perspective of other randomized prospective trials that also have flaws. Most important, we should demand other trials in this interesting and controversial area. Other trials are needed to evaluate evolving improvements in patient selection and technology for CAS and to compare both CAS and CEA with best medical therapy in asymptomatic patients. Other work is needed to develop methods to detect vulnerable but asymptomatic plaques at high risk of causing a stroke so they can be treated invasively without subjecting many patients with low-risk plaques to unneeded procedures at a great monetary cost to society.

Although CREST is an important trial, we must remember that no trial is perfect or definitive or has timeless value. Beware the spinning -- in any direction -- that purports to make it so.