A Major Medical Reversal in Vascular Surgery

The CREST 2 trial provided big surprises in the treatment of patients with asymptomatic carotid artery disease

Last week, investigators presented and published (NEJM) results of the CREST 2 trial of medical therapy vs revascularization of asymptomatic carotid stenosis. The findings surprised many in the vascular surgery world and represent a major medical reversal.

Here is the scenario: ultrasound machines are everywhere. Patients often get a scan of their neck arteries—sometimes for screening purposes, sometimes because a doctor hears a noise in the neck.

When a stenosis is found, everyone gets worried. The patient gets sent to a vascular surgeon.

What happens next depends on your country. In the US, it’s highly likely that the stenosis gets fixed—either with a stent (CAS) or what’s called endarterectomy (CEA). In Denmark or Australia, the first treatment is medical therapy (cholesterol lowering). There is a huge geographic variation in the approach to asymptomatic disease.

You might wonder why you would not fix a major blockage in the main artery to one side of the brain. Well, the group of David Spence have followed thousands of patients in his clinic ant Western University and shown that even when the artery occludes totally, stroke is rare (less than 1%). That’s because the brain has a built in collateral supply through the circle of Willis. Spence has long advocated intensive medical therapy. The other reason not to fix a carotid lesion is that it is an uncommon cause of stroke. This author calculated a population attributable risk of stroke from an asymptomatic carotid stenosis as less than 1%. (PAR here means that if you removed all asymptomatic lesions from a population you’d reduce stroke by 1%; HTN on the other hand has a PAR of 95%).

The Trial

CREST 2 enrolled patients with asymptomatic carotid stenosis >70% to two parallel arms, one was intensive medical management (IMM) + CAS vs IMM alone and the other arm was IMM + CEA vs IMM alone. About 1200 patients were in each arm.

Intensive medical management including blood pressure, glucose and lipid reduction via medications. Stenting and surgeries were done by experts in expert centers. Operators were well vetted.

The primary endpoint was a composite of any stroke or death in the first 44 days or ipsilateral (same side) ischemic stroke during the remaining days after 44 days.

First the CAS arm:

In the stenting trial, the rate of the primary endpoint was lower in the stenting arm. 2.5 vs 6.0% with an absolute risk reduction of 3.2% and a p-value = .02. The relative risk increase was 2.13 (95% CI 1.15-4.39)

The breakdown of the two components of the endpoint goes like this: in the first 44 days, there 0 events in the IMM arm and 7 strokes/1 death in the carotid arm—0% vs 1.3%

In the follow-up period beyond 44 days there was 28 pts with ipsilateral stroke in the IMM arm vs 7 in the CAS arm (1.7% vs 0.4%).

In the CEA arm:

In the CEA trial, the rate of the primary endpoint was lower in the CEA group, but it did not reach statistical significance 3.7% vs 5.3% p =0.24%

Similar breakdown in the components. More events in the CEA vs IMM arms in the periprocedural period (9 vs 3) and slightly less in the subsequent 4 years (0.5% vs 1.3%) but neither reached significance.

Other important results:

Another fact, pointed out in the accompanying editorial, was that medical therapy was less than intense. Only 60 to 70% of the patients had a SBP level within the target range (<130 mm Hg), less than 80% had a LDL-c level within the target range of less than 70 mg per deciliter and only approximately 50% of the patients with diabetes had a HbA1c level within the target range.

Disabling Stroke Numbers in the CAS vs IMM trial:

In table S4b in the supplement, we see that disabling stroke rates were very low, and similar. In the CAS vs IMM, there were 8 vs 6 disabling strokes over the 4 year period.

My Comments

Let’s first place this trial into context with two smaller trials that were recently published. The SPACE 2 trial had a similar design of CAS vs CEA vs medical therapy and found no significant differences. It was however underpowered. Similarly, the ECST-2 trial also found no benefit of carotid revascularization (either CEA or CAS) over medical therapy, though it too was underpowered and included lower risk patients.

I start with the two negative trials, Dr Spence’s work, and the low PAR, to emphasize that an asymptomatic carotid lesion is quite low risk. So our priors going into CREST 2 should be pessimistic that revascularization would be beneficial.

The easiest conclusion to make from CREST 2 is that carotid endarterectomy did not come close to meeting the statistical threshold for benefit over medical therapy alone. Purists might say that the 95% confidence intervals were wide, and the best case scenario allows for a 4% lower rate of stroke. My counter would be that the reason for the uncertainty is the low numbers of stroke in both arms, which is a good thing.

The harder conclusion is what to say about carotid stenting. On the one hand, it outperformed medical therapy by reaching a statistically significantly lower rate of the primary endpoint. And if that’s the case, should we recommend patients with carotid lesions be referred for stenting?

I would say no, not routinely. Here are four reasons why I would be cautious about routine stenting:

• Medical therapy was suboptimal: CREST 2 shows that, even in a trial, it’s hard to get to goal levels of metabolic risk factors. Patients with carotid disease need to have aggressive management of BP, lipids and glucose. Doing so would likely drive down the rate of stroke. But crucially, go back to the Spence study I linked above; the most common cause of death in patients with carotid lesions is CV disease. So by increasing medical therapy, you likely improve overall survival.

• Patient selection was serious: The trial protocol lists at least a dozen exclusions to carotid stenting—mostly anatomic. CREST 2, like all trials, makes sure to enroll ideal patients. In addition, the trial vetted all operators. So you can’t translate this easily to low-volume centers or to all carotid lesions.

• Effect Size was small: When you look at disabling stroke, it was 8 vs 6. That’s essentially the same. While the statistical threshold was met the difference is tiny. And if that is the case, the default should be to avoid a procedure.

• Statistical fragility: The authors provide a tipping point analysis and it’s only 3 events in the CAS arm would have rendered the trial non-significant.

In Sum:

My Neutral Martian view of CREST 2 is that if you have a patient with an asymptomatic carotid lesion, you get serious about medical therapy (including lifestyle advice) and follow the patient closely. You inform the patient what symptoms may look like and intervene when and if symptoms occur.

There is little to no benefit to routine intervention. The caveat of course is that you have to consider the patients who were not enrolled in CREST 2. There may be lesions that are especially severe that are not like those enrolled in the trial.

Finally, CREST 2 shows the value of trials. As a profession we must embrace the value of randomization and testing our beliefs. Kudos to the investigators and the NIH for funding such an important trial.