The TriGUARD 3 Cerebral Embolic Protection Device is designed to minimize the risk of cerebral damage by deflecting embolic debris away from cerebral circulation during transcatheter aortic valve replacement (TAVR), but there is no evidence that it actually accomplishes that, according to the consensus of a U.S. Food and Drug Administration (FDA) advisory panel.
But it is unclear whether the TriGUARD 3 even needs to meet that bar to receive approval under what is known a 510(k) premarket application.
The one recommendation that the FDA asked the Circulatory System Devices Panel of the Medical Devices Advisory Committee to make during its meeting Tuesday concerned the proposed indications for use of the TriGUARD 3 (Keystone Heart Ltd.), although the agency did ask the expert panel for insights on several other issues concerning the device.
Unlike the more-stringent premarket approval, which is reserved for high-risk devices and relies on performance data to demonstrate safety and efficacy, the 510(k) compare a new, moderate-risk device to a similar, previously approved device, known as a “predicate” device.
In the case of TriGUARD 3, that predicate device is the Sentinel Cerebral Protection System (Boston Scientific), which the FDA approved in 2017 for commercial use. Sentinel is the only commercially approved cerebral embolic protection device for TAVR in the U.S.
As Karen Jaffe, MS, MBA, RAC, a regulatory consultant for Keystone Heart, pointed out to the advisory panel Tuesday, this means that all TriGUARD 3 has to show is whether it is “substantially equivalent” to Sentinel and that it meets a set of FDA requirements known as “special controls.”
Sentinel is a temporary embolic protection device that is used during TAVR. It has two filters with baskets designed to catch embolic debris that breaks away during the TAVR procedure to keep it from reaching the brain and reducing the chances of stroke or transient ischemic attack. The filter baskets are inserted into the brachiocephalic and left common carotid arteries – two of the three arteries that connect the aortic arch to the cerebral circulation. Sentinel is inserted via transradial access.
TriGUARD 3, also a temporary device inserted during TAVR, has one large filter designed to cover all three of the arteries that connect the aortic arch to the cerebral circulation: brachiocephalic, left common carotid and left subclavian. It is inserted via transfemoral access with an 8-French delivery sheath.
Keystone Heart representatives presented data making the case that the TriGUARD 3 is “substantially equivalent” to the Sentinel. Both devices met the primary safety endpoint but did not meet the primary effectiveness endpoints in their respective pivotal trials.
When the Circulatory System Devices Panel considered Sentinel in 2017, it ultimately recommended that device be approved because it met its safety endpoint and, while it did not meet its effectiveness endpoint, it showed no risk of harm and did show substantial amounts of debris caught in the filter baskets, panel members pointed out Tuesday.
What troubled several members about TriGUARD 3 was that, while not statistically significant, the difference in stroke rate actually favored the unprotected control group, indicating a possible signal of harm from the device. But Keystone Heart representatives pointed out, and even the most troubled panel members agreed, that the phase II REFLECT trial was not adequately powered to show a true difference, meaning that play of chance cannot be ruled out as a reason for the device showing numerically higher risk of stroke.
As with the Sentinel panel meeting in 2017, the FDA did not ask the panel members to vote on their response to questions about TriGUARD 3. Rather, they asked for insights and consensus answers to a series of questions.
The first group of questions concerned the panel’s clinical interpretation of the safety results of both Sentinel and TriGUARD 3 in their respective pivotal trials. Richard Lange, MD, MBA, the president of Texas Tech University Health Sciences Center at El Paso and chair of the advisory panel, summarized the members’ responses by noting that no one felt comfortable with the performance goal (upper bound of 95% confidence interval of a composite event rate of 34.4%), which several said was too permissive. (TriGUARD 3 easily cleared that goal with a 21.3% composite rate.) Panelists also expressed concerns that the data Keystone Heart presented included roll-in subjects who received TriGUARD 3, not just those who were randomized to the device arm, and that the mixing of different groups of controls from different phases of the REFLECT trials was not appropriate for analysis.
The second group of questions focused on the effectiveness endpoints. Lange noted that panel members felt that the per-treatment group in REFLECT, which was defined as patients who had verifiable three-vessel coverage at two of the three measured timepoints (before, during and post-TAVR), was “highly selected” and “cherry-picked” and that analysis should not be based on it. Several members also expressed concern that there was discordance between magnetic resonance imaging brain scans showing fewer large lesions in TriGUARD 3 patients than in control patients, yet that TriGUARD 3 patients had numerically higher stroke and worsening National Institutes of Health Stroke Scale than control patients.
The third group of questions focused on whether the study data show that TriGUARD 3 was able to show secure and stable positioning throughout the TAVR procedure as well as the strengths and limitations of real-world data on device positioning. Several panel members noted that the “real world data” were not truly real-world but were instead the experience of a single center in the Netherlands with three skilled operators – an experience that might not be generalizable to other TAVR operators or centers. They also noted that the device in the REFLECT study is not the same as that in the Netherlands data, which was presented by Pieter Stella, MD, PhD, of Utrecht Medical Center, and showed 95% coverage of all three cerebral vessels from the aortic arch. The device used in the Netherlands has a translucent crimper, while that in the REFLECT study had an opaque covering.
In response to the fourth question, concerning the panel’s clinical interpretation of bleeding and vascular complications among TriGUARD 3 patients in the REFLECT trial and any possible relationship to the 8-French introducer sheath, panel members said that larger sheaths will increase bleeding risk but that this can be minimized with proper technique. Lange noted that an 8-French sheath is not much larger than the 6-French imaging catheter typically used during TAVR procedures.
Indications for use
The final question was about indications for use. Keystone Heart’s proposed indication for use for TriGUARD 3 is: “The TriGUARD 3 Cerebral Embolic Protection Device is designed to minimize the risk of cerebral damage by deflecting embolic debris away from the cerebral circulation during trans-catheter aortic valve replacement (TAVR).”
Joaquin E. Cigarroa, MD, a professor of medicine at Oregon Health and Science University, said he found no fault with the statement as written, that the TriGUARD 3 is designed to accomplish those goals. However, he and other panel members say the data do not show that the device can actually accomplish what it is designed to do.
John C. Somberg, MD, a professor at Rush Medical College, Lake Bluff, Illinois, said that while this may be beyond the scope of the advisory panel, he would like to see a phase III trial with adequately powered data to show whether TriGUARD 3 could, in fact, meet its stated design goal.
The FDA will consider the panel’s input as it decides whether to clear the TriGUARD 3 the 510(k) application. The device was granted CE mark approval in Europe in March 2020.