This article is more than 5 years old. Information may no longer be current.
FDA Debates Embolic Protection During TAVR
Experts say use of devices warranted, but more data needed.
Transcatheter aortic valve replacement has brought many benefits to patients who have severe aortic stenosis but may not be able to undergo open surgery. However, a risk of TAVR is stroke, if the procedure causes emboli to reach the brain.
While the newest generations of TAVR systems are associated with a lower risk for stroke compared with earlier generations, the problem remains prominent enough that significant research and development efforts have been made for embolic protection devices that can be used during TAVR.
One device, Sentinel Cerebral Protection System (Claret Medical), was found to be safe but did not meet its primary efficacy endpoint in the SENTINEL trial, results of which were presented at the TCT Scientific Symposium and published in the Journal of the American College of Cardiology in November 2016.
Nonetheless, during a hearing in February to advise on Claret Medical’s request for de novo classification of the device, members of the FDA’s Circulatory System Devices Panel said despite flaws in current data, clearance of the device for the U.S. market may be warranted. A second cerebral protection device (TriGuard, Keystone Heart), is currently being studied in the REFLECT U.S. pivotal trial.
A Case for Clearance
FDA advisory panel members stated that current efficacy data on the embolic protection device, based on findings from MRI scans, are not strong enough to assess how well the device might protect against stroke. However, many said the device serves an important need and may be worth allowing on the market because it is safe and appears to prevent disastrous outcomes.
No votes were taken by the panel as part of Claret Medical’s request for de novo classification of the device; rather, the FDA asked the panel for insights to help the agency make its decision.
Claret Medical asked the FDA for de novo classification of the device based on the results of SENTINEL, which found that use of the device in 363 patients undergoing TAVR was safe, but was not statistically significant for efficacy, defined as reduction in median new lesion volume in protected brain territories on MRI scans between 2 and 7 days vs. use of no embolic protection device.
The primary safety outcome was MACCE, defined as death, stroke or acute kidney injury at 30 days. Patients assigned the device had a MACCE rate of 7.3%, lower than the historical performance goal of 18.3% (P for noninferiority < .001) and not significantly different from the control group (9.9%; P = .41).
While median new lesion volume in protected brain territories on MRI scans between 2 and 7 days was lower in the device group than the control group, the difference was not significant (178 mm3 vs. 102 mm3; reduction, 42%; P = .33). The 30-day stroke rate was numerically lower in the device group compared with controls (5.6% vs. 9.1%; P = .25), but the study was not powered for clinical endpoints.
The problem with the data “is more about the imperfection of the device than a flaw in the concept of the device,” panelist David D. Yuh, MD, from Stamford Hospital, Stamford, Connecticut, said during the meeting.
Such flaws may not prevent the device from being allowed on the market, however. A de novo classification is a designation the FDA may grant enabling clearance for a novel medical device of low or moderate risk for which there is not a substantially equivalent product on the market. Devices granted de novo classification are not eligible for 510(k) clearance, which requires substantial equivalence to an existing product, but their applications are not subject to the same level of scrutiny as higher-risk devices without substantial equivalence to an existing product, which must go through the premarket approval process.
Outcome of Hearing
During the hearing, the FDA asked panel members whether the device was safe, whether diffusion-weighted MRI results from the SENTINEL trial were an acceptable surrogate for clinical stroke, how future trials of the device should be designed, how to report the effectiveness outcomes of the device in the labeling, what the clinical significance of debris capture might be, how to interpret the neurocognitive outcomes from the SENTINEL trial, whether the proposed indications and labeling were appropriate, whether benefit–risk considerations were adequate and what data should be collected postmarket if de novo classification is granted.
The panel reached consensus that the device was safe.
“I’m not concerned about a signal that relates to potential harm” from use of the device, said panel member Joaquin E. Cigarroa, MD, from Oregon Health and Science University, Portland.
Panel members were less certain, however, about how to interpret the efficacy data.
“Intuitively, it seems like a very good thing to me that these debris that are released during TAVR are taken out of the circulation pretty effectively by this filter,” said panelist Jeffrey S. Borer, MD, from State University of New York Downstate Medical Center. “Even though I can’t interpret what it really means, I intuitively believe that’s it’s better that the images that were done looked better after a filter was used than when a filter wasn’t used. What troubles me is that I can’t relate those results clearly to clinical benefit. There’s a suggestion, but it’s not as clear as I would like it to be. However, if I had to have TAVR, I would rather have this [device] in than not have it in.”
Cigarroa agreed. “I have a concern about efficacy, but I don’t know of a good clinical trial that is doable to overcome what I thought was a very well-conducted and well-designed trial [SENTINEL],” he said. “Because I think preventing potential for any stroke or cognitive dysfunction in a patient population that is already full of comorbidities and is moving toward lower [surgical] risk, as an interventionalist, I would use this device. I can’t tell you it is effective in reducing clinical endpoints, but I believe the totality of the data suggests that it might. The outcome, when it occurs, is so devastating that I would want to prevent it.”
Panel Chair
“While we are all frustrated with the outcome of the study, I’m not sure it could have been done better and I don’t think it ever will be done better, so we have the data we have,” he said. “I think it shows a signal in terms of capture of debris, and we’re in agreement that we’d much rather have that debris coming out of the body than going to our brains. For that reason, I would favor this.”
Future Directions
If the Sentinel device is approved, more research and development are needed, panel members said.
Future trials should “not just have imaging outcomes, but have neurological outcomes in a systematic way,” John C. Somberg, MD, from Rush University Medical Center, Lake Bluff, Illinois, one of several panelists who expressed dissatisfaction with diffusion-weighted MRI endpoints as a surrogate for stroke, said during deliberations.
A long-term registry measuring stroke and neurocognitive function of people who received the device might be a good idea, Somberg said.
In addition to a registry, follow-up of the SENTINEL trial population should also be conducted if possible, said George W. Vetrovec, MD, MACC, MSCAI, from Medical College of Virginia and Virginia Commonwealth University.
“If anything could be obtained from late follow-up on these patients, I think we could learn a lot, even if it were neurocognitive information instead of imaging,” he said. “Assuming approval, it would be wonderful to see a registry that went out 2 or 3 years with clinical follow-up. We could learn a lot just by knowing what happens to these patients.” — by Erik Swain
- Reference:
- Kapadia SR, et al. J Am Coll Cardiol. 2016;doi:10.1016/j.jacc.2016.10.023.
Disclosure: The members of the FDA’s Circulatory System Devices Panel report no relevant financial disclosures.
Perspective
Back to Top
It’s not a surprise that this device was said to be safe. It has been shown that we did not start causing strokes while we were trying to prevent them. It was disappointing that the device did not affect clinical stroke, but not hugely surprising. Clinical stroke during the procedure is such an uncommon occurrence now. The debate about the MRI-derived data has been ongoing. We know the diffusion-weighted images we are seeing represent small areas of cell necrosis in the brain. Most of us think that can’t be good, but we don’t know exactly how bad it is. The same images are seen in surgical AVR, albeit slightly less often than in TAVR, and we know there is no real difference in clinical stroke between TAVR and surgery; in fact, TAVR has been on the lower side of surgery in the most recent randomized trials.
The panel’s conclusion that this device is safe is correct. The panel’s conclusion that the device may prevent catastrophic stroke in uncommon cases is correct. The counter to that is, if we use this device in all TAVR procedures, we are going to add a lot of cost to a procedure that is already hard to break even on for many institutions. Then the question is, will operators be allowed to use the device at their discretion, or, as with carotid stenting, will they not be paid for the procedure unless they use it? To me, that would be a step too far, as there is no real evidence to support it. Approval of the device is reasonable, but then it will fall to clinicians to determine if they should add it to every commercial case performed. If that is done, the institution might lose money on the procedures. Another consideration, however, is what happens if the device is not used, and the patient has a stroke and sues, claiming it could have been prevented?
As a practitioner in this field, I am very concerned about neurologic injury, particularly as TAVR moves to lower-risk patients. If a 70-year-old patient after TAVR has diffusion-weighted lesions that go away after 3 months, do they later get dementia earlier than they would have otherwise? It’s hard to tell, and it’s something we all worry about.
If the device gets approved, I would like to see its use captured in the Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy registry data. I would also require that if it is going to be used, an NIH Stroke Scale score be calculated before and after the procedure, and if it changes, a neurologist should see the patient. Long-term registry follow-up is needed because strokes are uncommon events and the only way to capture uncommon events is through large registries. I would require that participation in TVT be required if reimbursement is to be given for the device, and that neurologic screening be performed.