Asymptomatic Carotid Disease and the Fallacy of Preventive Surgery

John Mandrola


September 30, 2015

We have to talk about disease in the carotid arteries and the common practice of performing "preventive" carotid surgery in patients who complain of nothing. I did not know the evidence for this well-accepted treatment strategy was so weak—and outdated.

You know the default: A blockage is bad and a blockage can be fixed. So it goes that hundreds of thousands of asymptomatic patients have undergone carotid endarterectomy (CEA) and carotid artery stenting (CAS). Billions have been spent, and hardly a whisper of dissent has been heard.

To be clear, my words here pertain to the treatment of asymptomatic patients with abnormal shadows on a neck ultrasound. There is no debate about intervening in patients with symptoms.

Let's start with a contrast in countries. In the US, approximately 90% of carotid intervention in the Medicare population is done in patients without complaints.[1] The rate in Denmark is 0%.[2] Before you dismiss the Danish way, note that Denmark has lower rates of death from heart disease than the US.[3]

Natural Bypass

A patient recently asked me to review his brain-scan report. His eyes were drawn to the words "complete occlusion" of a carotid artery. That sounded ominous, but he was neurologically perfect.

When I voiced my surprise about the occluded carotid to the neuroradiologist, he looked at me as if I was joking. "John, remember the circle of Willis acts as a natural bypass; the brain has ample collaterals."

This made me wonder if this is like the coronary story where intervening on asymptomatic lesions achieves little. Why then do we screen for carotid disease? Why do we perform so many surgeries and stents on asymptomatic lesions?

That's the thing with doctors. It's easy to fall into a routine. You have patients to see and procedures to do. Noticing the evidence for something as intuitive as "fixing" blockages in the neck can merge into our collective blind spot.

Then I read a provocative study[4] (and editorial[5]) published last week in JAMA Neurology. The results of this single-center retrospective study from a stroke-prevention center in Ontario led me to the larger story of carotid intervention as a preventive measure.

The JAMA Neurology Study:

Dr David Spence and colleagues (Western University, Ontario) set out to determine the risk of progression to complete occlusion of asymptomatic carotid lesions and whether total occlusion increased the risk of stroke.

This group has followed more than 3500 patients for over 20 years. They did yearly ultrasound exams and used this data to study 316 patients who were asymptomatic at the time of their total carotid artery occlusion. Around 2002, they intensified medical management based on plaque measurement.

Their results were striking. Most of the new occlusions (80%) occurred before 2002. Only one patient (0.3%) had ipsilateral stroke at the time of total occlusion, and three patients (0.9%) had ipsilateral stroke during follow-up. Neither the percent stenosis nor previous contralateral occlusion predicted events.

In the discussion section, the authors compared their observed stroke risk with that of carotid interventions. In the Carotid Revascularization Endarterectomy vs Stenting Trial (CREST), 30-day risk of stroke or death was 2.5% for CAS and 1.4% for CEA in asymptomatic patients.[6] A study[1] of real-world Medicare patients revealed a 1-year rate of stroke or death of 16.7% for CAS and 11.0% for CEA. The point is obvious: contemporary medical management of asymptomatic carotid disease provides a very low risk of stroke and should be preferred over surgery or stents.

In a Medscape Medical News story, Dr Spence called the practice of carotid stenting in asymptomatic patients "worse than unethical." Describing the situation in the US, where most carotid interventions are in asymptomatic patients, Spence used the word "deplorable."


Strong words, but here are six reasons to rethink the idea of preventive carotid intervention in asymptomatic patients.

First, a big caveat: Everyone agrees, including the AHA and ASA guideline writers,[7] that intervention in a "highly selected" group of asymptomatic patients is reasonable. This group, however, is far less than 90% of all interventions.

CEA benefits are small. Two large multicenter trials support the (American) preference for carotid endarterectomy over medical management in asymptomatic patients with high-grade carotid disease. The Asymptomatic Carotid Atherosclerosis Study[8] (ACAS) compared the two strategies in 1662 patients from 1987 to 1993 and the Asymptomatic Carotid Surgery Trial (ACST)[9] did likewise in 3120 patients from 1993 to 2003. These trials favored CEA over medical therapy, but absolute risk reductions in ipsilateral stroke were only 6% and 3%, respectively, over 5 years.

Medical management has dramatically improved. In the 1980s-vintage ACAS trial, best medical therapy included aspirin, advice to stop smoking, and control of blood pressure. That's it. In the ACST trial, medical therapy improved a little; it's estimated that 40% of the patients were on statins, but at that time, the dose of statins was much lower than current practice.

The intensity of medical therapy is no small thing. In a review paper, Naylor et al highlighted the decline in stroke rates in these two trials over 15 years.[10]In the first 5 years of ACAS (published in 1995), the risk of any stroke was 17.5%, for ACST (published in 2004) it was 11.8%, falling to 7.2% by the time the 10-year results were published (2010). In 2013, Dutch investigators reported a series of patients with asymptomatic disease followed for 6 years without intervention.[11] The rate of ischemic stroke was 0.5% per year in the 193 patients with 70% to 99% stenoses.

Given the small benefits of surgery and the temporal decline in baseline stroke rates, it's little surprise that a 2008 decision analysis estimated that CEA vs medical therapy conferred a mere 4 additional stroke-free days over 5 years in asymptomatic disease.[12] In this hypothetical model, if annual stroke rates dropped below 1.1%, CEA became detrimental.

Carotid stenting data weaker. The evidence for placing carotid stents in asymptomatic patients is weaker than CEA. The CREST trial,[6] which changed entry criteria during recruitment to include asymptomatic patients, was not powered to detect a difference between CEA and CAS in that subgroup. Both perioperative and 30-day stroke events were numerically higher in the CAS group, but the difference did not reach statistical significance. CAS made it into the guidelines, and volumes rose sharply.

Then came sobering results from the real world. Multiple large registries and audits have shown consistently higher rates of 30-day death and stroke with CAS compared with CEA.[13–15] What's more, in one audit,[16] the median annual physician CAS volume was only 1.5 procedures, and hospitals that did more CAS reported higher death and stroke rates. A vascular surgeon at my hospital told me CAS is covered (locally) only for symptomatic high-risk patients.

Stenosis progression is not predictive. This same vascular surgeon said you can't just watch a stenosis progress from 60% to 90% without intervening, a view that is supported more by eminence than evidence. Dr Spence's report of low event rates with total occlusion is not an outlier. Although two substudies[17,18] derived from the ACST trial revealed small relative increases in risk of ipsilateral strokes with stenosis progression, the actual number of events is low. Using data from one of these studies, Naylor et al write: "In practical terms, this means that for every 700 asymptomatic patients with a 70% to 99% asymptomatic stenosis treated medically, seven will occlude their carotid artery each year, but only one of the 700 will suffer an ipsilateral stroke, while one further patient will suffer an ipsilateral stroke at a later date following the occlusion."[17]

No benefit in women? Neither the ACAS nor ACST trials demonstrated a benefit for CEA in women.[19] In ACST, benefit was noted in women only when operative risk was removed. (That seems a dubious way to look at results.) A Cochrane Review was also unable to establish a benefit in women.[20] I'd say women are an important subgroup to consider.

It makes sense that preventive CEA is not superior. Atherosclerosis is a systemic disease. A carotid endarterectomy (or stent) is a focal treatment. Focal treatments work well when that lesion is causing trouble (STEMI), but those benefits do not necessarily translate to a dormant lesion—however stenotic it may be. In cardiology, we knew this to be true, but we ignored it until regulators forced us to pay attention to the COURAGE trial.[21] Now we place stents in lesions proven to cause symptoms. One of the many reasons focal treatments don't work well in dormant lesions is collateral circulation. Given the brain's built-in collateral blood supply, the data on CEA should be no surprise.


It is now time to see the evidence and embrace a contemporary view of the disease atherosclerosis when it presents in the carotid artery.

It doesn't mean we avoid carotid intervention in all asymptomatic patients. More than ever, we need skilled specialists—not only to operate but to identify the select group of patients who will benefit. It also doesn't mean we should stop doing ultrasound exams of the neck. When we find asymptomatic disease, we should see a systemic disease and prescribe aggressive systemic treatments.

Perhaps the most important reason to favor medical therapy for patients with asymptomatic carotid disease is that the number-one reason they die is heart disease—not stroke. In favoring intense medical therapy, therefore, we not only help our patients avoid strokes, but we also help them live longer.



Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: