Use of the Sentinel (Boston Scientific) embolic protection device is highly variable and its ability to limit cerebral injury modest at best, a TVT Registry study suggests.
The primary analysis, using an instrumental variable approach, showed no significant reduction in in-hospital or 30-day stroke with the device among more than 120,000 transcatheter aortic valve replacements (TAVRs) performed in the United States since its approval.
“Nevertheless, the secondary propensity-weighted analysis and the confidence interval for the primary analysis are consistent with a possible modest relative risk reduction in stroke of about 20% and a number-needed-to-treat of about 300 to prevent one stroke,” David J. Cohen, MD, University of Missouri Kansas City School of Medicine.
“These findings support clinical equipoise and provide a strong rationale for ongoing large-scale randomized trials to test whether embolic protection devices provide meaningful clinical benefit,” he said during a late-breaking trial session at TCT Connect, the virtual meeting of Transcatheter Cardiovascular Therapeutics (TCT) 2020.
Stroke continues to occur in 2% to 5% of patients undergoing TAVR, however, embolic protection devices (EPDs) have only been tested in small randomized trials using primary imaging endpoints, Cohen observed. “Thus, there remains considerable uncertainty regarding the clinical benefits associated with these devices.”
The study included 123,186 elective or urgent transfemoral TAVR procedures between January 2018 and December 2019 in the Society of Thoracic Surgeons/American College of Cardiology TVT Registry. Of these, 12,409 were performed with an EPD and 110,777 were not. Sites performing less than 20 TAVRs/year and emergent, alternative access, or concurrent mitral/TAVR procedures were excluded.
The proportion of hospitals using EPD rose steadily from 7% in the first quarter of 2018 to 28% in the fourth quarter of 2019. The proportion of patients receiving the device increased from 5% to 13% over the same period.
Usage varied widely, with 66% of hospitals never using an EPD over the study period and only 5% using it more than 50% of the time, Cohen observed. “By Q4 2019, this had increased to 8% but was still an extreme minority of hospitals in the United States.”
In the instrumental variable analysis, the adjusted rate of the primary endpoint of in-hospital stroke was in 1.39% in patients treated with EPD and 1.54% in those with no EPD (relative risk [RR], 0.90; 95% CI, 0.68 – 1.13; P = .41).
Rates were also similar for in-hospital death or stroke (2.4% vs 2.6%), major bleeding (4.0% vs 4.4%), device success (97% vs 97.2%), 30-day stroke (2.0% vs 2.1%), and 30-day mortality (1.9% vs 2.2%).
In the propensity-weighted analysis, which accounted for 30 demographic, clinical, and hospital-level characteristics, however, in-hospital stroke occurred in 1.30% of patients with EPD and 1.58% without (RR, 0.82; 95% CI, 0.69 – 0.97; P = .02).
Significant differences were also observed in in-hospital death or stroke (2.1% vs 2.5%) and 30-day stroke (1.9% vs 2.2%) as well as mortality (1.7% vs 2.2%).
Accounting for Patient/Site Differences
EPD was more likely to be used in patients with bicuspid valves (6.8%), and less often in patients on dialysis (2.1%), likely because of the need for right radial access to place the Sentinel device, Cohen said.
Otherwise, no other patient characteristics were significantly different between the two groups including age, sex, prior stroke, or surgical risk.
In contrast, hospital characteristics differed substantially by EPD use. Devices were used more often at teaching hospitals (82.4%), hospitals with a higher annualized volume, in an urban location (75%), and in the Northeast and Midwest (both 30.6%).
During a press briefing on the study, Susheel Kodali, MD, who led the trial leading to the Sentinel device approval, expressed concern that the analysis was able account for all confounders.
“For example, the sites that use Sentinel may be concerned about stroke and may be more focused on the neurologic analysis and do neurologic checks,” said Kodali, of New York-Presbyterian/Columbia University Medical Center in New York City. “So how do you account for something like that when those sites may be over-reporting stroke compared to sites not using Sentinel and maybe not focused on strokes?”
Cohen said that’s something the investigators wrestled with and, indeed, the data showed that sites that used more embolic protection diagnosed, slightly but significantly, more strokes in 2017 before Sentinel was introduced (1.9% vs 1.7%). As a result, the 2017 site stroke rate was adjusted for in both analyses.
Both men agreed that the question of whether embolic protection is being overused without a real clinical benefit will require randomized clinical trials. The ongoing, randomized PROTECTED TAVR trial, comparing TAVR with and without the Sentinel device in 3000 patients on the hard endpoint of stroke 72 hours post-TAVR, should report results in mid-2022.
“If I were to take something home from this study, first of all it’s the equipoise,” Cohen said. “Because I’ve heard some people say: I can’t ethically do this, we have this approved device, and I believe that it protects against stroke. I would say these data certainly suggest that equipoise is reasonable here and that nobody should be so convinced that they’re so correct they can’t randomize in this trial.”
“Second, what I do think is sobering to me is that, even if there’s benefit here, the benefit may be less than we think,” Cohen said. “I don’t know what the PROTECTED TAVR trial is powered to detect but we saw, kind of generously, if we look at the most favorable analysis about a 20% to 25% risk reduction. And that’s a little bit sobering for a device that’s supposed to protect three out of the four great vessels.”
Gregg Stone, MD, who moderated the formal presentation, agreed the study supports clinical equipoise but said the most important finding is the number-needed-to-treat of 300.
“If the truth was somewhere in between your confidence intervals, it’s more likely closer to 400 or 500,” said Stone, of the Icahn School of Medicine at Mount Sinai, New York City. “So that clearly wouldn’t justify embolic protection device use in all patients. So the real question is, who are the high-risk patients from your own data that may have benefited more by EPD use?”
Cohen said they haven’t looked specifically at that question thus far, but that patients with prior stroke and those with more bulky calcification may benefit more. “I would also say a risk factor for stroke is a lot of peripheral artery disease and from people who use a lot of this device — I don’t — those are the cases where it can be most difficult to place or where it isn’t placed at all.”
Panelist Megan Coylewright, MD, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire, said, “Even if the number-needed-to-treat is high, a lot of strokes perhaps could be prevented.
“I will say in the low-risk TAVR, patients are asking me more about this and there is significant variation in who is offering it,” she said. “So there’s a lot of different factors that might influence uptake and the scientific data in your very complex analysis.”
The study was limited by the observational design and reliance on site-reported strokes, rather than formal neurologic assessment, “therefore under-reporting is likely, particularly in the case of less disabling strokes,” Cohen said. Also, cognitive decline was not assessed and the instrumental variable analysis was somewhat underpowered.
The study was supported by the American College of Cardiology Foundation’s National Cardiovascular Data Registry (NCDR) and the Society of Thoracic Surgeons National Database. Cohen reported receiving research grant support and consulting income from Edwards Lifesciences, Medtronic, Boston Scientific, and Abbott.
Transcatheter Cardiovascular Therapeutics (TCT) 2020. Presented October 16, 2020.