No Benefit of Sentinel Device During TAVI: Updated Meta-analysis

NEW YORK, NY—Yet again, use of the Sentinel cerebral protection system (Boston Scientific) has not been shown to significantly improve clinical outcomes for patients undergoing TAVI, according to the results of an updated meta-analysis.

Risks of stroke, TIA, and all-cause mortality were not lower in an unselected population of patients undergoing TAVI with versus without protection, Sunny Goel, MD (Mount Sinai South Nassau, Oceanside, NY), reported last week at New York Valves 2025. The lack of difference was seen for both disabling and nondisabling strokes and for events occurring either shortly after the procedure or up to 30 days later.

Subgroup analyses were limited by the lack of patient-level data from the five included trials, but Goel noted that prior efforts to identify a specific subset that would have better outcomes with cerebral protection haven’t been successful.

“We’re still struggling to find a patient population that will benefit from it,” he said. “I don’t know which patient it is, but we all believe it works. But we haven’t seen anything in the clinical trials to prove it.” 

Presentation NY Valves 2025

Sentinel Cerebral Embolic Protection During Transcatheter Aortic Valve Replacement: An Updated Meta-Analysis of Randomized Controlled Trials

Prior studies have clearly shown that cerebral embolic protection devices capture debris that would have traveled to the brain in nearly all patients undergoing TAVI. Despite this visual documentation, embolic protection hasn’t translated into a clinical benefit in multiple randomized trials. In the past few years, the results of two large trials—PROTECTED TAVR and BHF PROTECT-TAVI—continued that trend, dampening enthusiasm for protection devices during the procedure.

For the new meta-analysis, published simultaneously online in JACC: Cardiovascular Interventions, Goel and his colleagues pooled the results of five trials—CLEAN-TAVI, MISTRAL-C, SENTINEL, PROTECTED TAVR, and BHF PROTECT-TAVI. MISTRAL-C and SENTINEL reported subacute outcomes up to 30 days after TAVI, whereas the other three trials reported acute outcomes at 72 hours or discharge. In total, there were 11,097 patients randomized to the Sentinel device or no cerebral embolic protection during TAVI.

There were no significant differences between groups in any of the outcomes examined:

• Any stroke (2.37% vs 2.58%; risk ratio 0.88; 95% CI 0.70-1.12)
• Disabling stroke (1.01% vs 1.38%; RR 0.68; 95% CI 0.41-1.14)
• Nondisabling stroke (1.35% vs 1.20%; RR 1.03; 95% CI 0.75-1.43)
• TIA (0.35% vs 0.27%; RR 1.28; 95% CI 0.66-2.50)
• All-cause mortality (0.73% vs 0.65%; RR 1.09; 95% CI 0.70-1.72)

A sensitivity analysis examining acute and subacute events also revealed no differences between the Sentinel and control groups.

CLEAN-TAVI and MISTRAL-C provided data on neurocognitive endpoints, although they were underpowered and exploratory, Goel et al note in their paper. The trials showed, however, a reduction in new ischemic lesions on brain MRI and “trends toward improved neurocognition with cerebral embolic protection device use, suggesting a potential subclinical benefit,” they write.

What Comes Next?

Results of PROTECTED TAVR led to a decline in the use of the Sentinel device in the United States, and Goel said it changed his practice in a similar way. He was using the Sentinel in almost all of his TAVI cases until 2022 but has since become more selective, deploying it in about 10% of procedures.

“I don’t think the device is bad. I think we’re just not able to find the right population for it,” he said, pointing to patients with heavily calcified bicuspid valves, those undergoing valve-in-valve procedures, or those with a prior history of ischemic stroke as potential groups that might benefit more from protection. “Teasing out those high-risk populations probably will be the way to go.”

Goel noted that the increased risk of stroke isn’t confined to the periprocedural period, with some occurring up to 30 days after TAVI. Some of those events could be related to interrupted anticoagulation around the time of the procedure in patients with atrial fibrillation and a high CHA₂DS₂-VASc score, he suggested. For that reason, his team has adopted an approach that involves starting high-risk patients on IV heparin a day before TAVI, maintaining treatment until just before the procedure, and then resuming IV heparin after about 4 hours once good vascular closure has been achieved. Treatment with a direct oral anticoagulant is initiated the next day.

Asked if a different embolic protection device is needed, Goel pointed out that others like TriGuard and Embol-X have been tested but also haven’t had a major impact on clinical outcomes. Perhaps, he suggested, a mechanism other than embolism might be at play.

Teasing out those high-risk populations probably will be the way to go. Sunny Goel

Goel also indicated that there could be potential neurocognitive benefits of preventing debris from going to the brain that wouldn’t become apparent until patients were followed over longer periods of time.

Dustin Kliner, MD (UPMC, Pittsburgh, PA), who moderated the session, agreed with that possibility.

“The filters are full. The MRIs light up like a Christmas tree when you see them afterwards, and I think it’s very important to not let this technology die because of the cognitive impairment piece that you point out,” he said. “There has to be some benefit from that perspective, which we just don’t know yet. And I’m not sure who will need this to prevent some sort of future decline, but as we get into younger and lower-risk patients, we certainly have to think about that.”

Alan Yeung, MD (Stanford University School of Medicine, CA), one of the discussants, indicated that it will be difficult to detect any neurocognitive impact of using cerebral embolic protection devices, particularly in older patients with age-related declines in function. “I think it still makes sense to put [the devices] in in a hundred percent [of patients] if you are comfortable with it, because it probably does something, but can you prove it? I think it’s going to be difficult.”