Meta-analysis Quashes Hopes for Routine Embolic Protection in TAVI

PARIS, France—A new analysis of more than 10,000 patients likely slams the door on routinely using cerebral embolic protection for stroke prevention in patients with aortic stenosis undergoing TAVI.

In a patient-level look at the PROTECTED TAVR and British Heart Foundation’s BHF PROTECT-TAVI randomized trials, there were no significant differences in stroke or disabling stroke between those undergoing TAVI with and without embolic protection.

When investigators focused solely on patients who received a device (Sentinel; Boston Scientific) in the per-protocol analysis, there appeared to be a benefit, particularly when it came to reducing the risk of disabling stroke. That finding should be considered hypothesis-generating, they say.

“At the moment, there’s no evidence that we should be using [embolic protection],” lead investigator Rajesh Kharbanda, MBChB, PhD (Oxford University Hospitals NHS Foundation Trust, England), told TCTMD. “That’s not to say there’ll never be evidence. I think we probably need to get data for a more stratified approach to better understand which patients are at higher risk for stroke. There may well be a benefit, perhaps, in that [high-risk] group, particularly on disabling stroke, but we don’t have the granularity of data yet.”

In general, Kharbanda said that “as an evidence-based practitioner, it’s difficult for me say, ‘Use this device.’”

As an evidence-based practitioner, it’s difficult for me say, ‘Use this device.’ Rajesh Kharbanda

The new study, which was presented Wednesday as a late-breaking clinical trial at EuroPCR 2025, likely doesn’t surprise too many clinicians given that PROTECTED TAVR, which was led by Samir Kapadia, MD (Cleveland Clinic, OH), and BHF PROTECT-TAVI, which was led by Kharbanda, both were neutral trials when it came to stroke prevention.  

Neel Butala, MD (University of Colorado Anschutz School of Medicine, Aurora), who published an analysis from the STS/ACC TVT Registry showing embolic protection was associated with a small, borderline significant reduction in disabling stroke, said that stroke risk in TAVI is low overall but remains persistent and is tough to predict for individual patients.

“I really want there to be a safe and effective device to prevent stroke for TAVR,” he told TCTMD. However, based on the randomized trials and other studies, he does not routinely use embolic protection during his cases. “That being said, there are some subgroups where I would definitely consider using it, such as those with prior stroke,” Butala added. “This is based on our prior registry work.”

Still, Butala believes there is a need for continued innovation in this space, citing stroke prevention during TAVI as a large “unmet need.”

Kharbanda made a similar point, noting that stroke is a devastating complication for patients who want to maintain their independence. “They’re older and many of them rate stroke worse than death as an outcome,” he said.

Two Large, Randomized Trials

PROTECTED TAVR, which was published in 2022, was the first to deal a blow to the idea that preventing dislodged emboli from traveling to the brain would lead to a reduction in stroke. The study included 3,000 patients randomized to embolic protection with Sentinel or usual care when undergoing TAVI. The primary endpoint—stroke within 72 hours or prior to discharge—was not significantly different between the two groups, but there was a significant reduction in the risk of disabling stroke, a secondary endpoint.

With BHF PROTECT-TAVI, published only in March, there were more than 7,600 patients randomized to TAVI with and without Sentinel for cerebral embolic protection. Again, there was no difference in the risk of stroke within 72 hours or prior to discharge or in the risk of disabling stroke.

In combining both studies—a plan that was agreed upon prior to the results of either trial being known—the risk of stroke in the intention-to-treat analysis was 2.2% and 2.3% in TAVI-treated patients with and without embolic protection, respectively (P = 0.641). The risk of disabling stroke was 1.0% with embolic protection and 1.3% in patients undergoing TAVI without the device (P = 0.090).

In the per-protocol analysis, there was a significant difference in the risk of all stroke—1.7% and 2.3% with and without embolic protection (P = 0.023)—though statistical significance was no longer there after adjusting for age and sex. With disabling stroke, there was a 0.5%-lower risk with embolic protection in the per-protocol analysis (0.8% and 1.3% with and without embolic protection, respectively; P = 0.007). The edge favoring embolic protection remained after adjusting for age and sex.  

“There’s a suggestion that disabling stroke may be reduced, but we really need further analysis,” said Kharbanda. “The per-protocol analysis is prone to postrandomization error and selection bias, and we need to understand what that signal means.”

Butala also cautioned against making too much of the per-protocol analysis.

“It is possible that patients who were randomized to cerebral embolic protection but did not have both filters deployed also had abnormal or calcified vasculature, which would also be associated with a higher risk of stroke, and which would confound the results,” he said.

The researchers conducted a complier average causal effect (CACE) analysis, which is a regression analysis that attempts to protect randomization by assuming patients in the control group are similar in compliance to the treatment group. In the unadjusted and adjusted CACE analyses, embolic protection was not associated with a reduction in stroke or disabling stroke.

What Happens Next?

Stephan Windecker, MD (Bern University Hospital, Switzerland), one of the panelists during the late-breaking session, said his high-volume TAVI center had routinely used cerebral embolic protection but stopped after PROTECTED TAVR.

While the two trials were similar, there were some notable differences, said Windecker. In PROTECTED TAVR, patients were excluded if the anatomy was deemed unsuitable for embolic protection, whereas BHF PROTECT-TAVI was a trial without systematic screening. As a result, there was a difference in the rates of successful device deployment: both filters were successfully deployed in 81.2% of patients in PROTECT-TAVI versus more than 95% in PROTECTED TAVR.

“If you want to prove the point of a protective mechanism of a device, shouldn’t you limit that to patients where the anatomy is suitable?” asked Windecker.

Kharbanda stressed that BHF PROTECT-TAVI was a pragmatic trial that recruited one-third of all patients undergoing TAVI in the United Kingdom. PROTECTED TAVR, on the other hand, was a selective strategy trial. There are subtle distinctions between asking whether the device works at all and whether it works in general everyday practice, he said.

“I think it terms of a routine strategy, we can say it’s difficult to find evidence [of benefit],” Kharbanda said. “But, it doesn’t rule out the argument for a selective strategy or in patient subgroups where you can safely deliver the device.”

To TCTMD, Kharbanda said the next step involves figuring out if there is a type of high-risk patient who might particularly benefit from embolic protection. However, the researchers have not yet had the chance to dig into subgroups.

Butala noted there are several other embolic-protection devices in development and he’s hopeful they can show a benefit. However, given the evidence to date, he doesn’t think it will be enough to show noninferiority to the Sentinel device in order to justify their use.  

David Moliterno, MD (University of Kentucky, Lexington), another panelist, also urged the field forward, and called on clinicians to participate in future clinical trials.

“If, for example, 250,000 TAVRs are done globally this year and we really believe what we’ve seen of this 0.5% reduction in major disabling strokes, that’s well over a thousand cases,” he said. “I’m not saying [embolic protection] should be done routinely, and I believe we do need to figure out who gets the most benefit, but I think that’s going to require future studies and require more than a third of patients participating.”