The US Food and Drug Administration (FDA) Circulatory System Devices Panel meeting to review the evidence for the Edwards Sapien transcatheter heart valve (Edwards Lifesciences, Irvine, CA) used in transcatheter aortic valve implantation (TAVI) took place in Gaithersburg, Maryland, on July 20, 2011. I was among a group of physicians who attended the meeting to answer specific questions from the panel on behalf of Edwards Lifesciences. My role was to give a European perspective on TAVI 3 years into the commercial rollout of this technology in Europe. The meeting gave me a fascinating insight into the workings of the FDA and the United States regulatory system.

The purpose of the day was for a group of clinicians, industry, and patient representatives (the panel) to consider a series of questions that were posed to the panel on behalf of the FDA and then to vote on whether they believed the Edwards Sapien transfemoral TAVI technology was effective and safe compared to medical therapy in patients who were deemed not suitable for open surgical aortic valve replacement. The result would be the recommendation of the panel to the FDA, but the final decision as to whether to allow the technology to be commercially introduced to the United States is ultimately that of the FDA.

THE MEETING

The day followed a very structured agenda. After some brief introductory remarks from the Chair of the panel, Edwards Lifesciences (termed throughout the day as the sponsor) presented for 75 minutes on the clinical data for the Edwards Sapien transcatheter heart valve. This principally revolved around data from the PARTNER trial, which were previously reported in the New England Journal of Medicine,1 discussions on postapproval studies, and arrangements for training of individuals and institutions during the rollout phase of the technology. Clinical data from the PARTNER trial were that of Cohort B, the so-called surgically inoperable patients, and were presented by Craig R. Smith, MD, and Martin Leon, MD. This was followed by a presentation by the FDA giving their view on the PARTNER trial and their proposals for postapproval studies.

Immediately after lunch, there was a 1-hour open public hearing, during which, those interested were allowed to present data, information, or views, orally or in writing, on the issue pending before the panel. During this time, two patients, two TAVI nurse coordinators, an individual clinician, the Society for Cardiovascular Angiography and Interventions (represented by Augusto Pichard, MD), the American College of Cardiology (represented by American College of Cardiology President David Holmes, MD), and the Society of Thoracic Surgeons (represented by the Society of Thoracic Surgeons President Michael Mack, MD) presented their evidence to the panel. The rest of the afternoon was taken up by panel discussions on specific questions posed to them by the FDA regarding the data, proposed indications for use, and the postapproval studies.

THE DEBATE

Much of the day was spent discussing the relative efficacy and safety of the procedure. The efficacy of TAVI is best assessed by mortality and quality-of-life (QOL) improvements in the TAVI arm. There was a general recognition that TAVI extended life compared to medical therapy in patients who were unsuitable for surgical aortic valve replacement (Figure 1). However, the panel appeared to have some misconceptions. One comment was that TAVI only extended life by 18 months to 2 years and whether this was “really important in a patient in their mid-80s.” My first thought was that maybe someone should ask such a patient. I think I know what the answer would be! The second comment was that although the therapy extended life, “nearly all the patients had died within 2 years,” and therefore, QOL is much more important. This statement is not true—66% of patients in the SOURCE registry were still alive at 2 years (Figure 2). Similar data were presented to the panel from the Canadian Registry.2 It was acknowledged that QOL is important, and impressive QOL data in favor of the TAVI arm were presented for Cohort B (Figure 3).

Patient Selection

The main concerns of the panel were patient selection in the PARTNER trial, the heterogeneous nature of the treatment in the control arm, and complications of the procedure, with particular focus on neurological events and vascular complications.

There was vigorous debate on how one defines an inoperable patient and which patients should not have TAVI. There was broad agreement that there was no definition of inoperable. The pragmatic definition presented by Edwards was that a surgeon should ask himself, “in the absence of TAVI, would I operate on this patient?” If the answer was “no,” the patient becomes a Cohort B-type patient. It was widely agreed that some patients have too many comorbidities and should be turned down for both surgical aortic valve replacement and TAVI; this type of patient seems to be increasingly called the Cohort C patient.

This is probably one of the most difficult aspects of TAVI (ie, optimal patient selection). In Europe, multiple registries are actively seeking to develop a TAVI risk score, which would facilitate patient selection and further improve 1-year mortality—it is well known that the majority of deaths between 30 days and 1 year relate to patient comorbidities rather than being caused by cardiac issues. My contribution to this section of the panel discussion was to discuss the patient populations that are being treated in Europe. There is obviously a major concern in the United States about potential risk creep (toward lower-risk surgical patients) in Europe for TAVI. I pointed out that the indication for TAVI in Europe was severe symptomatic aortic stenosis in highrisk surgical patients, generally with a logistic EuroSCORE of > 20. I also made the point that there had been very little obvious change in risk between Cohort 1 and Cohort 2 of the SOURCE registry and that the average EuroSCORE in European registries was well higher than 20 (Table 1).

On this basis, I could see no major evidence of risk creep in Europe. A crucial element in ensuring that the appropriate patients receive TAVI is to make sure that programs are delivered by heart teams, within which, surgeons play an important role. At my own institution, the surgeon is the gatekeeper of the technology, and any change in the demographics of TAVI patients will be principally driven by the cardiothoracic surgeon.

The PARTNER trial compared the use of transfemoral TAVI versus standard therapy in patients who were considered inoperable. Patients in the standard therapy arm received a heterogeneous variety of interventions, including balloon aortic valvuloplasty in 80% of cases. This was believed to be a high rate by the panel, and there was debate as to whether this really reflected a real-world standard of care. It was eventually agreed that there was (1) no evidence that balloon aortic valvuloplasty led to adverse events (some data were presented suggesting the opposite), and (2) that there is no real standard care for the inoperable symptomatic aortic stenosis patient.

Stroke

The debate regarding stroke was particularly challenging and robust. An important secondary endpoint was a composite safety endpoint of death, myocardial infarction, stroke, and renal failure. The FDA took important exception to the fact that, at the end of the study, stroke was changed to only major stroke, as classified by a retrospective assessment using the modified Rankin scale.3 The FDA maintained that there is no validation of a retrospective modified Rankin scale assessment, and analysis should be based on all stroke rather than just major stroke (as it was in the original protocol). This resulted in the stroke rate rising from 5% (major stroke) to 7.3% (all stroke) in the TAVI arm and 1.1% to 1.7% in the control arm at 30 days.

My own interpretation was that this endpoint should be harmonized across all TAVI trials using the Valve Academic Research Council definitions. What surely is important is the effect of any neurological event on QOL. Strokes will either result in death, debilitation, partial recovery, or full recovery. It is important to note that Edwards presented data that showed that the majority of patients who survived TAVI but had a stroke still had improved QOL at 1 year. Perhaps the concept of survival free of debilitating stroke should be introduced as an endpoint of TAVI trials.

There was some discussion as to whether there was a continued stroke risk in the TAVI arm that was higher than that of the control arm from 30 days to 1 year (2.8% in the control arm and 4.5% in the TAVI arm). This appeared to be refuted by an analysis that included the fact that more TAVI patients were alive and, therefore, available to have a stroke (ie, correcting the denominator). With this analysis, the stroke rate appeared to be similar in the TAVI and control groups after 30 days, with no excess ongoing risk in the TAVI arm.

One of my own contributions was to briefly discuss the potential causes of stroke and the potential ways in which stroke could be reduced in the future. A significant number of strokes occur more than 5 days after the procedure and, therefore, are obviously not procedural. Thus, it is likely that neurological events occurring during TAVI are multifactorial and that no single intervention is likely to solve the problem. Potential strategies to reduce stroke include minimizing the catheter French size to limit injury while traversing the aortic arch, careful control of blood pressure during anesthesia, limitation of rapid pacing, careful control of wires to avoid the carotid circulation, and standardization of pharmacotherapy during and after the procedure. This last point was discussed in some detail. Standardization of pharmacotherapy is likely to be part of the postapproval studies. It was acknowledged that although the recommendation during the study was for dual-antiplatelet therapy for 6 months, it was unclear how many investigators had rigidly followed this recommendation. The FDA was explicit that future FDA-regulated studies of TAVI will require more intense neurological evaluations.

Bleeding

The FDA combined the bleeding events and vascular complications and stated that 55.9% of patients in the TAVI arm underwent one of these complications. In the New England Journal of Medicine article,1 major vascular complications were reported in 16.2% of TAVI patients at 30 days, and major bleeding was reported in 16.8% of patients. It is well known that major vascular complications are associated with 30-day and 1-year mortality in TAVI, and these levels of vascular complications are unacceptably high. The device that will be introduced in the United States will be 22 and 24 F. In Europe, we now have access to 16- and 18-F devices, and it is hoped that this is one way that these vascular complications may be reduced, with a subsequent improvement in early and 1-year mortality. Discussion took place as to whether proper training and patient selection in the United States could mitigate against this complication. The FDA made it clear that it would favor a postapproval study that included compliance with training and an evaluation of their effectiveness in reducing vascular complications.

It should be noted that during the first rollout phase of the device, only the transfemoral system will be available in the United States, with no alternative vascular access site. This could be a potential problem, and clinicians will have to be disciplined to not “push the envelope” of vascular access in the absence of an alternative access site. In general, the availability of the transapical or transaortic approach improves the results of transfemoral TAVI because these access sites provide the clinician with an alternative that avoids transfemoral TAVI in small or calcified peripheral vessels.

Postapproval Studies

Two postapproval studies were discussed and labeled as PAS 1 and PAS 2. One panel member commented that it was interesting to be asked to discuss a postapproval study before a device was actually approved—a thought that had also struck me and one of the more amusing moments of the day. PAS 1 was essentially extending the follow-up of the Cohort B patients of the PARTNER trial to 5 years. The only area of controversy was the need for QOL information. Edwards argued that the major improvement in QOL at 1 year was sufficient to demonstrate the efficacy of the treatment and that 5-year data would be redundant. The panel unanimously disagreed and believed that these data should be collected.

PAS 2 was essentially a very rigorous registry (although many features, such as 100% monitoring, make it feel like the rigor of a clinical trial) that will measure adherence to indications, changes in patient populations, device durability, and QOL, as well as individually powered hypothesisgenerating endpoints at 30 days and 1 year, such as vascular complication rates and neurological events. The data that seemed to be required were the same as in the PARTNER trial but with a patient population between 750 and 1,000 patients, making it five times larger than PARTNER. This seemed to me to be a hugely expensive venture that was being asked of the company who understandably did not support the proposal.

The alternative put forward by the American College of Cardiology and Society of Thoracic Surgeons was a national registry that will follow these patients. This is exactly the system we use in the United Kingdom and would seem to be a very reasonable approach. In the United Kingdom, all patients have to be entered into the national registry, otherwise the procedures are not funded. In addition, there is central tracking of mortality via the Office of National Statistics. Centers can then be assessed both in terms of mortality rates and patient demographics (providing an assessment of the performance of the multidisciplinary teams and a potential methodology to measure risk creep) against other units in the country.

The most difficult and dramatic moments of the day occurred immediately after lunch when data were presented (on request from the panel) on the results of the continued access group of patients. This was a small group of 80 inoperable patients who were randomized after the end of the formal trial in a continued access protocol. These patients were randomized as the Cohort A trial (TAVI vs high-risk surgery) continued randomization, and therefore, it was believed that there could be an imbalance in randomization decision making if operators were aware that all patients in the Cohort B continued access group would receive TAVI treatment.

In the data presented for this small group of randomized patients, the TAVI patients appeared to have a higher 1-year mortality rate than the control group. This caused great consternation among the panel as well as a prolonged and anxious discussion, which included demands to see how the survival curves would change if these patients were added to the main trial results. One explanation put forward by the investigators was that five new centers were able to place patients in the continued access protocol, and that, in some way, this could have influenced the results. One panel member was even more concerned at this response and commented on the fact that 200 new centers were potentially about to be started up in the commercial rollout of the device, if it was approved, and wondered what kind of results we could expect from these centers. Eventually, the Chair of the panel, with guidance, steered the panel to consider their vote only on the basis of the main trial results, and the assumption was that the results of the continued access patients were unreliable because of low patient numbers. There were clearly a number of misgivings from at least two panel members, which were probably reflected in the subsequent voting.

THE VOTE

Having started at 8 AM, the big moment of voting arrived at just after 6 PM. Three questions were posed to the panel regarding (1) the device showing reasonable assurances of safety, (2) the device showing reasonable assurances of efficacy, and (3) that the benefits outweighed the risks of the device. A farce then ensued as the electronic voting system failed to work and the panel had to revert to paper. The final votes gave a resounding vote of confidence in TAVI from the panel to the FDA. The panel voted 7:3 for safety, 9:1 for efficacy, and 9:0 (with one abstention) that benefits outweighed the risks.

THE RECOMMENDATION

At the end of the meeting, the following was proposed as the indication for use of the Edwards Sapien device: “Indicated for transfemoral delivery in patients with severe symptomatic native aortic stenosis who have been determined by a cardiac surgeon to be inoperable for open aortic valve replacement and in whom existing comorbidities would not preclude the expected benefit from correction of the aortic stenosis.” The words symptomatic and native had been added during the deliberations of the day.

CONCLUSION AND A EUROPEAN PERSPECTIVE

This was an extraordinary event to witness. The drama was high, and it was like being in a courtroom. However, my thoughts kept returning to the patients. We were discussing the introduction of an early-generation, large-French system compared to the latest-generation, small-caliber device that we use in the Europe. In addition, we have multiple vascular access site options that improve the results of the transfemoral approach by limiting the need to “push the envelope” of transfemoral access. What would an American patient really think? This trial demonstrated a huge net clinical benefit despite the potential complications of the procedure. However, the discussions of the panel concentrated on the defensive elements for both the FDA and the panel. Is this process really protecting them from inappropriate therapy or limiting their access to the latest devices? I think I know what they would say.

It is anticipated to be 2013 before any citizen of the United States will be able to access the devices that are currently available in Europe. I really believe that TAVI is a breakthrough technology that will change the face of the treatment of valvular heart disease in the world. It seems tragic that systems that have been put in place to ensure that patients in the United States receive optimal health care are actually resulting in the opposite effect. I would propose providing patients the information and let them decide; they will get it right.

Martyn Thomas, MD, FRCP, is Clinical Director of Cardiovascular Services, St. Thomas' Hospital in London, United Kingdom. He has disclosed that he receives grant/research funding from Edwards Lifesciences. Dr. Thomas may be reached at 07885268035; mttwins@aol.com.

  1. Leon MB, Smith CR, Mack M, et al; for the PARTNER Trial Investigators. Transcatheter aortic valve implantation for aortic stenosis in patients who cannot undergo surgery. N Engl J Med. 2010;363:1597-1607.
  2. Rodés-Cabau J, Webb JG, Cheung A, et al. Transcatheter aortic valve implantation for the treatment of severe symptomatic aortic stenosis in patients at very high or prohibitive surgical risk: acute and late outcomes of the multicenter Canadian experience. J Am Coll Cardiol. 2010;55:1080-1090.
  3. Burn JP. Reliability of the modified Rankin scale. Stroke. 1992;23:438.