P-R-O-C-E-E-D-I-N-G-S

9:01 a.m.

ACTING CHAIR MAISEL: Good morning. I would like to call to order this meeting of the Circulatory System Devices Panel. Today's topic is discussion of a pre-market application for the W.L. Gore and Associates GORE TAG Thoracic Endoprosthesis, P040043. I would like to ask Geretta Wood to read the Conflict of Interest statement.

EXEC. SEC. WOOD: The following announcement addresses Conflict of Interest issues associated with this meeting and is made a part of the record to prevent even the appearance of an impropriety. To determine if any conflict existed, the Agency reviewed the submitted agenda and all financial interest reported by the Committee participants. The Conflict of Interest statutes prohibit special Government employees from participating in matters that could affect their or their employer's financial interest.

However, the Agency has determined that participation of certain members and consultants, the need for whose services outweighs the potential conflict of interest involved is in the best interest of the Government. Therefore, waivers have been granted for Drs. Charles Bridges, L. Henry Edmunds, Thomas Ferguson, William Maisel, Clyde Yancy and a waiver was previously granted for Dr. Judah Weinberger for their interest in firms that could potentially be affected by the Panel's recommendation.

The waivers for Drs. Bridges, Edmunds, Ferguson and Maisel involve a grant to their institution for the sponsor's study. The panelists had no knowledge of the funding and had no involvement in the generation or analysis. Dr. Ferguson's waiver also involves his affiliation with a nonprofit organization that is the recipient of an unrelated educational grant from a competitor.

Funding to the organization is between $100,001 and $300,000 per year. Dr. Yancy's waiver involves unrelated consulting services with a competitor for which his fees have not yet been determined. Dr. Weinberger's waiver includes a stockholding in a competitor in which the value is between $50,001 and $100,000. The waivers allow these individuals to participate fully in today's deliberations.

Copies of these waivers may be obtained from the Agency's Freedom of Information Office, Room 112A-15 of the Parklawn Building. We would like to note for the record that the Agency took into consideration other matters involving Drs. Mitchell Krucoff, Joanne Lindenfeld and Clyde Yancy. These panelists reported past or current interest involving firms at issue, but in matters that are not related to today's agenda. The Agency has determined therefore that these individuals may participate fully in the Panel's deliberations.

The Agency also would like to note that in the event that the discussion involves any other products or firms not already on the agenda for which an FDA participant has a financial interest, the participant should excuse him or herself from such involvement and the exclusion will be noted for the record. With respect to all other participants, we ask in the interest of fairness that all persons making statements or presentations disclose any current or previous financial involvement with any firm whose products they may wish to comment on.

ACTING CHAIR MAISEL: Thank you, Geretta. My name is Dr. William Maisel. I'm a cardiologist at Brigham and Women's Hospital and I would like to invite the Panel Members to introduce themselves starting on my left with Dr. Zuckerman.

DR. ZUCKERMAN: Bram Zuckerman, Director, FDA Division of Cardiovascular Devices.

DR. FERGUSON: Tom Ferguson, Professor Emeritus, Washington University, Saint Louis.

DR. LINDENFELD: Joanne Lindenfeld. I'm a Cardiologist at the University of Colorado.

DR. KRUCOFF: Mitch Krucoff. I'm a Cardiologist at Duke University Medical Center and the Director of the Cardiovascular Devices Unit at the Duke Clinical Research Institute.

DR. NICHOLAS: Gary Nicholas, a vascular surgeon, Professor of Surgery, Penn State University, Lehigh Valley Hospital.

DR. BRIDGES: Charles Bridges, Cardiothoracic Surgeon, University of Pennsylvania.

EXEC. SEC. WOOD: Geretta Wood, Executive Secretary for the Advisory Panel.

DR. SOMBERG: I'm John Somberg. I'm a Professor of Medicine and Pharmacology at Rush University in Chicago, Illinois.

DR. KATO: Norman Kato, Cardiothoracic Surgery, private practice, Encino, California.

DR. NORMAND: Sharon-Lise Norman, Professor of Health Care Policy and Biostatistics at Harvard Medical School and Harvard School of Public Health.

DR. JOHNSTON: Wayne Johnston, Vascular Surgeon, Professor of Surgery at University of Toronto.

DR. WEINBERGER: Judah Weinberger, Interventional Cardiologist at Columbia University.

MR. MORTON: Michael Morton. I'm the industry representative. I'm employed by Medtronics.

MS. MOTTLE: Linda Mottle, Director and Faculty of the Clinical Research Program at Gateway Community College.

ACTING CHAIR MAISEL: Thank you. Geretta, if you would read the voting status statement, please?

EXEC. SEC. WOOD: Pursuant to the authority granted under the Medical Devices Advisory Committee Charter dated October 27, 1990 and as amended August 18, 1999, I appoint the following individuals as voting members of the Circulatory System Devices Panel for this meeting on January 13, 2005: Charles R. Bridges, M.D., L. Henry Edmunds, Jr., M.D., Thomas B. Ferguson, M.D., Kenneth W. Johnston, M.D., Joanne Lindenfeld, M.D., Norman S. Kato, M.D., Gary G. Nicholas, M.D., John C. Somberg, M.D., Clyde Yancy, M.D., Judah Z. Weinberger, M.D., Ph.D.

For the record, these individuals are special Government employees and are consultants to this Panel under the Medical Devices Advisory Committee. They have undergone the customary Conflict of Interest review and have reviewed the material to be considered at this meeting. The Agency would also like to note that Dr. William Maisel has consented to serve as Chair for the duration of this meeting. This is signed by Daniel G. Schultz, M.D., Director, Center for Devices and Radiological Health and signed January 11, 2005.

ACTING CHAIR MAISEL: Thank you. Before we begin this morning's discussion on this application, the FDA has two brief presentations. I would like to invite Dr. Binita Ashar, Acting Clinical Director, of the CDRH to talk about the Critical Path Initiative.

DR. ASHAR: Great. Thank you and good morning. I appreciate this opportunity to discuss with you the Agency's Critical Path Initiative from the CDRH perspective. Basically, what I'm going to do this morning is I'm going to identify some of the challenges in medical product development. Then I will define for you what the Critical Path Initiative is and then describe what our future efforts are for bringing this initiative further.

Basically, the present state of affairs is that there is a scientific challenge that we have a number of disease processes, Alzheimer's, AIDS, cardiovascular diseases that need better treatments and we not only need better treatments, but we need better preventative therapies. At the same time, we're faced with a societal challenge and that is the urgency for timely development of treatments for these diseases. And not only do we need these treatments to be timely, but we also need these treatments to be affordable.

In the present state of affairs, there is great optimism based on new biomedical discovery. We have sequenced the human genome. We have new genomic and proteomic technologies. There are advances in medical imaging. We have nanotechnology advances that potentially can offer the right treatment to the right patient in the right location with far fewer side effects than ever before. And at the same time, we have been investing to produce these basic biomedical advances.

There has been an increase in NIH funding of double over the past five years. And pharmaceutical research and development has also increased at the same rate. Overall, our society has provided major investments in basic biomedical technology research and this is a graphical representation demonstrating the increase in research spending, both from the pharmaceutical RND side as well as in the NIH budget.

Now, you would expect that this acceleration in development, this would have translated into increased medical product development. However, from the drugs and biologic side, in fact, there has been a decline in the number of FDA new products that have been submitted. Now, this necessarily hasn't been the case for medical devices, but the fact of the matter is we could be doing better. And this is a graphical representation demonstrating the 10-year trend in pre-market device application showing the number of original PMAs that have been submitted.

Now, at the same time, we are noticing that, at least on the drug and biologic side, the cost of bringing a new drug to market is estimated to be about $1.7 billion, and the reason for this is largely because there is a high failure rate of new drug candidates late in the clinical development process. Now, what is the cause of this problem? Well, some of these new technologies aren't at their full potential.

And what has been occurring or what we have noticed to occur is that industries have been focusing on easier targets and because of various business arrangements have focused on potentially the cash cows and not necessarily treatments that might affect smaller populations. They have found that the development process has become uncertain. Some of the additional challenges that I have mentioned already are that there is a failure late in the clinical development process, at least for drugs and biologics.

Now, I want to mention that the Critical Path is different for devices. Device development is different because of the device regulation process. We have a least burdensome provision of FDAMA, which is different than drugs and biologics. We are committed to finding a least burdensome path to market. We have quality systems and design controls that are not prescriptive, but are focused on what the end result is and has the product, indeed, met the expectations that we have requested.

The innovation process is different for devices. The small molecule issue is biocompatibility, not necessarily biometabolism. The process is an iterative process whereby sometimes during the clinical development phase, there might be minor changes in the device. There is a user learning curve that we face with the use of medical devices and performance and durability are also engineering issues. Different pharmaceuticals in the device industry is represented by small manufacturers that may not have the resources to put forth all of the time and effort and expenditures that they might need to to bring a product forward.

Other additional causative factors that have been shown as a hurdle in the medical product development is that some of the basic science investment and progress has surpassed what we are able to actually translate into new medical products. Essentially, we are using the evaluation tools and infrastructure of the last century to bring forth new medical products of this century. We are doing randomized controlled clinical trials like we have always done them.

We are not necessarily using our cumulative knowledge of the society to overcome some of the development hurdles, so that we can bring medical products to market faster without compromising our safety and effectiveness evaluations. And this has resulted in a bottleneck at the Critical Path for delivering new medical products to patients.

So the central Critical Path thesis is that there has been a great societal investment in research and development to improve medical product development. However, there has not been an investment in the tools necessary to translate this basic biomedical research into new medical products. So what do I mean by this? Well, tools that might be computer simulation tools or registries or new surrogate markers that have been validated or biomarkers that might be able to identify patient populations that might be most amenable to these treatments that would potentially cut down the size of these various clinical trials.

There has been a great investment in the basic research, but not investment and attention to the tools to bring this research to translate into medical products. And some of the problem is that academia is not adequately funded to perform the scientific investigations to develop new tools. This has generally not been conceptualized up until this point, at least, as being FDA's role. And any efforts to develop valuative tools in the private sector are proprietary and they are, therefore, not generalizable and available for use for the population at large.

So the FDA's Critical Path Initiative is an attempt to bring attention and focus to the need for targeted scientific efforts to modernize the techniques and methods used to evaluate the safety, effectiveness and quality of medical products as they move from product selection to design and mass manufacture. And this diagram demonstrates how Critical Path research differs from what is generally considered translational research.

You notice that basic scientific research is the type of research that is largely conducted by academic organizations and by our sister agencies like NIH. NIH is also quite interested in translational research of bringing this new basic research into the clinical arena. However, Critical Path research is really FDA's arena where this translational research is looked at from the perspective of mass manufacture and mass marketing.

Can these clinical trial results formed in a small population translate into the generalized patient populations that we are approving a device for? And in evaluating any sort of medical products, you look at three dimensions of Critical Path. You look at safety. Can this device adequately perform in a safe manner? Can this device demonstrate efficacy in the population? And can this device be mass produced to the point that it is generalizable to not only the premiere centers in the United States, but also to all of the community hospitals? And can these results that we see in these early clinical trials be ones that can be replicated over and over again in smaller areas?

And so if we had tools that might be able to help us in our assessment in deciding whether a product is safe or effective or can be industrialized, wouldn't it be great to be able to bring these products to development faster without compromising our safety and effectiveness evaluation? So basically, the Critical Path science basis is to be able to understand what types of tools we might be able to invest in.

And FDA potentially could take an organizational role in bringing groups together, consumer groups, patient groups, academia and industry groups to develop some of these scientific tools that might be then available in the public arena for use by all industry groups. And this is something that NIH and academia have generally not focused on and Critical Path is intended to be something that is supplementing what we already have learned in our translational research basis.

So the work ahead, basically, this is for scientific improvement. It is not to be confused with regulatory evolution or streamlining or making the paper pushing faster or easier. It is, essentially, using the science that we already know to develop tools that help in our evaluations. The regulatory process is a related effort and can assist with this, but it is not the focus of this initiative.

So what have we done so far with Critical Path? Well, basically, there was a Federal Register docket open describing a Critical Path that was open through the summer and we received a number of responses from industry groups, patient groups, professional organizations, individual industries. This initiative was also presented to the FDA Advisory Board, the Science Board, to receive some of their feedback. And we have had individual meetings with various scientists, companies, patient groups and many, many others just to get the word out, just to get feedback out.

This has also been presented at the FDA Science Forum and at many speeches and panel presentations. And, basically, we have received overwhelming support. In fact, they have asked FDA to embark on doing things that is well outside of FDA's resources. And we have heard this really from all of our patient groups and all of our industry groups.

Submitters actually, again, ask for us to work on a number of things intent actually outside of some of our range. And some of the things that they suggested is, you know, streamlining clinical trials if we had better biomarkers, if we had a process by which we knew that we could validate a surrogate endpoint and promote effective product development. What they wanted repeatedly was FDA feedback on particular endpoints and particular surrogate endpoints. And how we can harmonize internationally so that we could better do this so that a clinical trial in one area of the world would be applicable to the United States, how we could focus on cancer trials, combination products.

Some industry groups actually have commented on the use of proprietary data. You know, very tentatively, they mentioned that perhaps FDA might find a way to use some of this information with the consent of all involved parties to further medical product development. And so, basically, the bottom line is this is an initiative that is intended to use science to integrate into the regulatory process, so that we can make our safety and effectiveness evaluations faster and more cost effectively.

This is not just an FDA effort. We can't do this alone. We need to work with our stakeholders to make this a reality. And we need to focus on particular scientific areas that first and perhaps expand at a later date. So the next step on a Critical Path is from the docket we received a number of comments. We are identifying all of the possible proposals and prioritizing various opportunities for developing valuative tools. We will be putting forth a national Critical Path opportunities list that reflects all of the comments that we have received.

We need to find a mechanism by which we can continue to obtain such feedback and update this list so that not only FDA, but other interested parties, might embark on Critical Path research. Thank you very much for your attention.

ACTING CHAIR MAISEL: Thank you very much. Next, I would like to invite Megan Moynahan, who is the Branch Chief for Pacing, Defibrillator and Leads, to update the Panel on some recent decisions.

DR. MOYNAHAN: Good morning. Thank you very much. I would like to take a few minutes this morning to update you on the Panel meeting that occurred in July this past year in which the panelists discussed the Guidant Companion application, a labeling review, and the Philips Medical HeartStart Home Over-the-Counter AED.

Beginning with companion, the FDA raised a number of concerns to the Panel and I'm not going to represent them all here, but the primary one related to whether the Panel felt that the data were sufficient to support an expanded patient population for the Guidant CRT-D Device to include patients who did not have to have a requirement for an ICD. There was a lot of discussion about the change in definition of hospitalization and the FDA has some concern about how to interpret both the primary and secondary endpoints based on that.

We asked the Panel to comment on how the indication should be worded and we asked for some broad labeling recommendations on that application. The Panel recommended that the data supports expanding the indicated patient population, but they had some concerns about how that would be worded in the indication statement. They specifically asked us to avoid using the term "all-cause hospitalization" in the indication statement. And they wanted a special separate section of the labeling to call out the benefit with respect to the primary endpoint.

The approved indications appears as follows, and as you see, it only indicates the intended patient population, so it is indicated for patients with moderate to severe heart failure who remain symptomatic despite stable optimal heart failure drug therapy and have an LVEF of less than 35 percent and a QRS no greater than 120. And now, there is a new clinical outcome section that appears in the labeling just after the indication statement in which we go into more detail as to the clinical benefit to patients.

Here is where we identify the reduction in risk of all-cause mortality or first hospitalization, is how it is presented, and then we go onto define how hospitalization was used in that trial, including noting that the hospitalizations did not include the device implant attempt or any reattempts. We also identified the reduction and risk of all-cause mortality and we mentioned the reduction of heart failure symptoms.

The Panel also made a number of other recommendations with respect to the labeling and, in particular, they were interested to see how we were going to be presenting hospitalizations. While they agreed that the representation of the primary endpoint should not include the index or implant hospitalization, they felt that it would be important to present in clinically meaningful information to physicians and patients that describe hospitalizations in general.

And so this is an example of the approved labeling that we've worked with the company to develop. This is the original Kaplan-Meier curve for the primary endpoint of all-cause mortality or first heart failure hospitalization and this is the same as what you would see in the published literature. But the labeling also represents a number of hospitalizations per patient year and that was done to account for the difference in follow-up for the two different groups.

It presents this information comparing the OPT or the control group to the CRT-D group and it also gives you an idea of the relative contribution of the implant hospitalization in both cases. There is also a graph in the labeling that depicts the number of hospitalization days per patient year, again distinguishing between the OPT group and the CRT-D group and also indicating a relative contribution of the implant hospitalizations that occurred in both groups.

And finally, there is a representation of the number of heart failure hospitalization days per patient year comparing the OPT group to the CRT-D group. Based on their Panel recommendations, the FDA approved the companion submission on September 14, 2004.

Now, moving on to the Philips Medical Over-the-Counter AED, the FDA raised a number of concerns to the Panel that day including asking whether the data was sufficient to support over-the-counter availability of the device and, in particular, we wanted them to comment on the adequacy of the user testing and whether the sponsor had appropriately integrated CPR prompts and notifications to dial 911 or to notify the Emergency Medical Services.

We asked the Panel to comment on whether the data were sufficient to support over-the-counter availability of the pediatric pads. We asked broadly for labeling recommendations and to comment on the sponsor's methods for tracking devices in the event of a recall or adverse event reporting and whether they believe that a post-market study would be required. Because this was a 510(k) application, there was no vote. However, the Panel was felt to be generally in favor of over-the-counter availability of the device. They felt that the usability testing was adequate and that the voice prompts for CPR and the visual prompts for calling 911 were felt to be adequate.

There was no consensus, however, on whether the pediatric pads should be available over-the-counter. There were quite a number of specific labeling recommendations that were given to us and the Panel recommended a post-market study, but asked FDA to review the tracking and adverse event reporting methods. FDA ultimately concurred with the Panel that there was sufficient usability testing and we did not require additional testing on the part of the sponsor. We felt that the prompts for CPR were appropriate and with minor modifications to the 911 reminders we felt that they were appropriate as well.

Importantly, the pediatric pads were not included in our over-the-counter decision and they still remain available as a prescription accessory. And that was done for a number of reasons. We felt that ultimately this would simplify a very complex purchasing decision by not offering too many options or accessory products to the user. We felt that it sent an important message that underscores that sudden cardiac arrest is an adult public health concern, one that's not shared equally by the pediatric population.

We feel that this decision ensures safe and effective use on both adults and children. And we felt that this was not going to impact availability too detrimentally of families who have higher risk children, because those families should be well-integrated into the medical system and would easily be able to get a prescription for the products.

We made substantial labeling modifications based on Panel recommendations. I'm not going to go through all of them, but I'll give you one example. The Panel felt that the outer box should be designed to help customers make an informed purchase decision and these are some of the things that appear now on the outside of the box and also appear on websites that are offering this product for over-the-counter sale.

For example, it mentions that you should speak to your doctor and that a defibrillator does not take the place of seeking medical help, that you can't use the device on yourself, that users may need to perform CPR, that responding to cardiac arrest may require you to kneel, that voice instructions and materials are in English and that the HeartStart provides audible and visible indicators for maintenance.

FDA's clearance decision included acceptance of the sponsor's methods for post-market tracking and adverse event reporting and we're continuing to work with the sponsor on developing the Post-Market Study Plan. Ultimately, FDA cleared this product for over-the-counter use on September 16, 2004. Thank you very much.

ACTING CHAIR MAISEL: Thank you, Megan, for those updates. At this point, we will begin the open public session of this morning's meeting, both the Food and Drug Administration and the public believe in a transparent process for information gathering and decision making to ensure such transparency at the open public hearing session of the Advisory Committee meeting, FDA believes that it is important to understand the context of an individual's presentation.

For this reason, FDA encourages you, the open public hearing speaker, at the beginning of your written or oral statement to advise the Committee of any financial relationship that you may have with the sponsor, its product, and if known, its direct competitors. For example, this financial information may include the sponsor's payment of your travel, lodging or other expenses in connection with your attendance at this meeting.

Likewise, FDA encourages you at the beginning of your statement to advise the Committee if you do not have any such financial relationships. If you choose not to address this issue of financial relationships at the beginning of your statement, it will not preclude you from speaking. At this point, I would like to invite Dr. Rodney White to approach and address the Panel.

DR. WHITE: Thank you very much. It's a pleasure to be here today. I'm representing the Society for Vascular Surgery and a project that you have heard about before, I think, that we would like to update you on. The Lifeline Registry, which now is the SVS/American Vascular Association Outcomes Registry has been an effort that we have updated Panels on serially related endoluminal grafts.

At the beginning of this, I would like to tell you that my Conflicts of Interest are that I have no commercial interest in Gore. I'm not a Gore investigator. I am here representing the Society for Vascular Surgery as the Secretary and Chairman of the SVS/AVA Lifeline Registry Committee. I'm an academic surgeon. I make my living treating these kinds of patients and get promoted based on publishing papers, so I think my major conflict is I make a living doing this sort of stuff.

The Lifeline Registry was established in 1997 to look prospectively at post-approval of abdominal aortic aneurysms. The SVS has now recently, in association with the American Vascular Association, an expansion of our nonprofit foundation efforts, extended the SVS capability to look at outcomes analysis, not only to the endoluminal grafts, but to the technologies we're talking about today, thoracic grafts or in a concurrent effort to carotid stents and endarterectomy.

A unique aspect we offer is that we can look at both of these technologies concurrently and in that regard, we would like to emphasize from the beginning that the SVS is going to make a specific effort to make available operative data related to these technologies, so as we move into these other areas, it will be relevant.

Registry is unique in that the initial attempt was to do something that hadn't been done successfully previously and that was to look prospectively to establish a registry that would take scientific data, put it together over time and actually look at a number of stakeholders involved in this, including the societies and clinicians, the foundation that I mentioned, federal agencies, and we have been very fortunate to have both FDA and CMS active in these efforts with their input.

An industrial advisory committee made of, in this case, the following companies which have supported that effort, I will emphasize that of the group today W.L. Gore was a founding member of the registry, has been very proactive in supporting this effort and that this has been an important part in how we progress. The registry goals were to evaluate long-term and prospectively endoluminal graft function.

Because of the requirement for post-market surveillance, I think this is a topic that will come up later, we were able to establish with a central registry committee working beneath the foundation and again with the ex officio input of the FDA, NIH, CMS and this industrial advisory committee and our data center, New England Research Institute, a way to be able to collect this data, put it together and report it.

The funding mechanisms are by the foundation itself and industrial partners feeding this data and collecting it. The registry then initially was to look at the post-market IDE data, collect that and look at it over the five-year surveillance interval that was available. Now, if you think about that, that makes it a very high compliance audited data set. There are then two parts to this registry I'll tell you about briefly, because we've got now six-year results to look at, was to take this five- year PMA model, look at these patients over time and because of the post-approval market, we're able to look at the requirement of the Agency to have the manufacturers submit this data and work out a collaborative effort to collect this.

The long-term results of the FDA devices have then been evaluated and in a collective fashion you will see. Now, these are just some numbers to give you an idea of how these can be powered over time, but with four approved devices and some commercial site entries, you will see we're now nearly at 3,000 patients. A very important part of this is we also have a concurrent surgical control group of patients that can be compared for outcomes analysis. And in the kinds of considerations we're doing today, these are particularly important.

Primary and secondary endpoints that I won't list in detail were looked at and were able to, in almost all cases now, start to look at very important outcome issues and patient selection parameters and have reached statistical significance in many of these related to the outcome comparing endoluminal grafts to the conventional stent graft technologies. The same for hospital parameters, ICU stays, things that are routinely looked at and again we can compare these two groups.

What we have come up with then, now, these are six-year curves, are comparisons of morbidity, mortality, aneurysm-related mortality, freedom from rupture, gender analysis, this has been highlighted in other examples, freedom from surgical conversion. And again, just to highlight this very briefly, there is a data set that was presented at the SVS this year in a publication submitted to the Journal of Vascular Surgery that will summarize these six-year data outcomes and forward. But in general, it gives us a very good standard to be able to apply this to and have a surgical cohort group.

Now, the other important relevance to today is not only if we looked at the PMA data sets, but we've tried to extend this to clinical sites outside of the use and studies. There are currently 15 centers entering that data and we've also, in collaboration with the Canadian Vascular, have 22 Canadian sites that are submitting data to the registry. So this becomes very powerful. At some extent, we tried to simplify that and make it automated where the sites would, in an automated fashion, enter their data, the reports go then, and this sort of prototype fashion, to look at critical measurements related to an endoluminal graft.

I know these are hard to see, but it would be diameter, volume, distance from an anatomic landmark, anything that deviates over time, and these are sequential intervals highlighted in red. There's a possibility to put in images, so you can look at sequentially patient records. In a thoracic prototype relevant to today, we could do a similar sort of entry looking at fixed points, measurements, volumes and diameter, put in sequential records, collect these that can be given to the patients, put in their charts and used for data or in the carotid scenario that I mentioned as an outcomes, do a similar sort of thing where we can look at critical parameters in operative carotid stent patients, collect the data, have imaging analysis, in this case, the patient has had an event after that and be able to correlate all this data.

So that the summary would be then that the SVS has had now a six-year track record in collaboration with the Agency and industry to do this. We would like to present it as a future prototype and that the SVS is committed to making outcomes analysis related to temporary and new devices a priority for our society. Thanks for the opportunity to present this material.

ACTING CHAIR MAISEL: Thank you for your comments. The next invited speaker or public speaker is Dr. Greg Sicard.

DR. SICARD: Good morning. I'll be very brief, since Dr. White already presented a lot of the data that is supported by the Society for Vascular Surgery. But my name is Gregorio Sicard. I'm a practicing vascular surgeon from Washington University School of Medicine in Saint Louis and currently the President of the Society for Vascular Surgery. I do not have any financial interest in W.L. Gore or the specific product and my trip was paid by the Society for Vascular Surgery.

I come here today primarily representing the Society for Vascular Surgery and secondarily as a practicing vascular surgeon. The Society for Vascular Surgery has over 2,300 members, many of which care for patients that have thoracic aortic aneurysms and fully understand the impact that a less invasive approach for the treatment of this condition will have.

As a vascular surgeon who for many years has practiced both open and endoluminal treatment of the intrarenal abdominal aortic aneurysm and open repair of aortic abdominal aneurysms, I rise to comment on the benefits of making this technology available for patients with this disease. The introduction of endoluminal prosthesis for the treatment of intrarenal abdominal aortic aneurysm has had a significant impact in patient care because of the decrease perioperative mortality and morbidity associated with this less invasive technique.

This has been recently documented in AAA randomized trials from both the United Kingdom and Holland. The approval of a thoracic endograft device will add an important option for cardiothoracic and vascular surgeons who treat patients with these conditions. Even in experienced hands, the surgical treatment of thoracic aortic aneurysms carry a 5 to 10 percent perioperative mortality and a significant morbidity estimated between 50 and 75 percent that is, obviously, associated with prolonged hospitalization and increased health care cost.

This less invasive technology offers a new approach that is associated with lower operative and post-operative mortality and morbidity. Therefore, I strongly urge this Panel to approve the thoracic endograft device. This lower risk treatment modality will provide significant benefits to patients as well as expand the options that treating physicians can offer to this patient population. Thank you for your attention.

ACTING CHAIR MAISEL: Thank you. Is there anyone else in the audience who wishes to address today's Panel? Please, approach. Please, identify yourself and mention any financial conflicts.

MR. TINKER: My name is Bill Tinker. I'm a patient of Dr. Bavaria. I've had no contact until today, I've met some of the people from Gore and I want to let you know that I'm here because I wanted to be here. I survived. I had an attack where I went to surgery to repair an aneurysm, an abdominal aneurysm. Back in June of last year, I had a large aneurysm in my chest, told to be inoperable and I was given two weeks to live. So I opted for the stent and it worked and now I'm here.

With four or five years in between, I suffered with the aneurysm. You have to understand it's not easy. I couldn't pick up anything larger than 20 pounds. I drove in the right hand lane off the road, because I knew once it blew I would have 10 seconds to get off the road before I killed somebody. Now, I'm driving in the passing lane. That in itself is really nice. It's good to have this happen and I want you to know that it's traumatic what I went through in the open-heart.

Yes, and the open surgery was -- it was a month in the hospital. My kidney shut down three times. My valves shut down. My lungs shut down. They contacted my family on two or three occasions that I was dying and I just kept coming through and coming through and a month later they woke me up and it was a year later before I could walk 100 feet to my mailbox to pick up my newspaper and my mail. It was that long in recuperating.

Two days after I had this sort of massive aneurysm, after I received the stent, two days later I was ready to go home. And I was absolutely pain-free. There was no pain. And to this day I'm able to loft around 50 pound bags of cracked corn and salt and no problem at all. It's just amazing and it's still ticking and it actually looks like a barbed-wire fence, but it works. And I just wanted to let you know.

I had one other thing I wanted to make a point or two. I hadn't mentioned those people. Back in '99, I had a 2 inch patch put in my abdomen and I went through all that trauma and close to death several times, that bill came to $500,000 for a week stay in intensive care. Back this June when I had this done, well, the amount was $90,000, but it wouldn't have been that high if they had had that ready for me when I went in. I had to go through a lot and stay in the hospital, that ran it up.

But I'm estimating it would have been $50,000 tops. So we're talking 10-fold savings we're going to have with this. Patients are going to be able to have this without their insurance company going broke. And I want everybody to have it as well as I did and have a good as a life as I do. Thank you very much for listening to me.

ACTING CHAIR MAISEL: Thank you very much for your comments. Is there anyone else in the audience this morning who would like to approach? Yes?

DR. CAMBRIA: Good morning. My name is Richard Cambria. I am a Professor of Surgery at the Harvard Medical School and Chief of the Division of Vascular and Endovascular Surgery at the Massachusetts General Hospital in Boston.

I traveled here today at my own expense. I have no financial interest in the W.L. Gore Company or its products, although our group and our institution has received support in the form of the support for conducting clinical trials from the W.L. Gore Company and virtually every other device development pivotal trial in abdominal aortic aneurysms and thoracic aortic aneurysms. We currently participate in all three extant thoracic aortic aneurysm stent graft trials.

Our group at the Mass General has embraced stent graft repair of abdominal aortic aneurysm and thoracic aortic aneurysms. We currently treat, approximately, 350 aneurysms of all types annually at our hospital. We implanted the first stent graft for an abdominal aneurysm in New England in 1994 and the first thoracic stent graft in 1997.

To date, we have implanted some 900 abdominal aortic aneurysm stent grafts and over 100 thoracic aneurysm stent grafts. We have participated in virtually every pivotal trial to evaluate stent grafts including, as mentioned, all of those for the thoracic aorta.

My own practice has centered on thoracic and complex thoracoabdominal aortic aneurysms and, in that context, I have personally performed over 500 open aneurysm repairs of the thoracic and thoracoabdominal aorta. Thus, we speak from a position of, I think, some experience and I guess we would like to think expertise in this field.

Our practice has evolved to the point where some 65 percent of abdominal aneurysms are treated with stent grafts, and I personally treat every thoracic aneurysm where such treatment can be performed with a stent graft, as opposed to an open operation. We are, of course, currently limited in the application of stent grafts in the thoracic aorta to those patients who qualify for the available FDA- sponsored clinical trials since, of course, there is no commercially approved device.

I would just remind the Panel that today we focus on degenerative aneurysm of the thoracic aorta, but this is just the beginning. There is a whole host of thoracic aortic pathology, including traumatic lesions, traumatic tears and aortic dissections, and I am certain that we will see stent graft repair play a very important part in the treatment of all of these pathologies over the next few years.

In virtually every comparison of endovascular therapy, as opposed to conventional open surgery, issues of safety, efficacy and durability are, of course, prescient. None of these endpoints can be separated from what I refer to as the morbidity quotient of the procedure, namely, what is the risk of the treatment?

In certain vascular lesions that these Panels have heard, for example, the issue of carotid artery disease, there is, in fact, a very narrow margin, if any, between the morbidity quotient of open surgery as opposed to endovascular therapy. The opposite end of the spectrum is true for the treatment of thoracic aortic aneurysms.

Open surgical repair of thoracic aortic pathology, although refined at the moment to a high level of sophistication, is still accompanied, even in the hands of experts, by major morbidity in the form of death or paraplegia in some 10 percent of patients. Thus, the morbidity quotient of the pathology that we are talking about here today is extreme in the difference between endoluminal therapies and open surgical repair.

ACTING CHAIR MAISEL: Dr. Cambria, if you could conclude your remarks in the next minute, please.

DR. CAMBRIA: Yes, I will.

ACTING CHAIR MAISEL: Thank you.

DR. CAMBRIA: Yes, I will. Patients need this pathology, this technology. We who treat these patients absolutely require it. This will be the single significant advance in the treatment of thoracic aortic pathology in our lifetime. Thank you for your attention.

ACTING CHAIR MAISEL: Thank you very much. Are there other individuals who wish to approach? Yes?

DR. TUCHEK: Good morning. First of all, I want to let you know that I have no financial interest with Gore. I'm here on my own time and at my own expense. My name is Dr. Michael Tuchek. I am a cardiovascular and thoracic surgeon at Loyola University in Chicago, Illinois.

Loyola has got one of the largest open heart programs and largest aortic surgery programs in the city. I guess I'm one of the few people in the audience like Dr. Cambria, one of the old fashioned surgeons who still do a lot of open procedures.

I am also fortunate enough to be one of the busiest endovascular surgeons in the country. I did more Medtronic AneuRx devices than any surgeon in the country last year. I am one of the primary investigators in the Valor Trial for thoracic stent grafts and I am a leading enroller in that trial currently. So I am being blessed with being able to do both the open procedures and have a lot of experience with stent grafting also.

Obviously, because I am in the Medtronic Trial, I am not involved with Gore. I am not in their trial. I have never placed their device. I have seen it placed once and that is my total experience with Gore. I am sure in the audience there's a few Gore people who are a little concerned about a Medtronic guy being here talking about their device but, rest assured, I'm not here to torpedo their efforts. I am here to applaud them.

We have all looked at the Gore data and you are going to be looking at it in detail shortly. I'm not going to go over it, but I reviewed it. I think they did a great job in this trial. I think that when they had issues, they dealt with them. They never jeopardized their patients' care while the trial was ongoing and I applaud them for that. I hope that the Medtronic trial is doing just as well as this trial was.

When I started doing abdominal stent grafts in 1999, I thought the trial that was going on, and when it got approved, I thought the technology was slick, the word I used. When you look at open operations, they are fairly morbid. When I started doing thoracic stent grafts, I thought that technology was nothing short of astounding, truly astounding. My son calls it radical, but I use the word astounding. It is truly impressive technology.

When I do an open operation, for those of you who are not cardiac surgeons, we make incisions from the back of the neck all the way down to the naval. We break ribs or resect ribs. They go on a heart-lung machine frequently. We resect the aneurysm. There is bleeding. There is a lot of morbidity associated with the operation.

And in a week or two, if the patient does well, a lot of them go home in wheelchairs, paralyzed permanently, and that is when the operation goes perfectly. That is an issue. There is a lot of morbidity associated with the open operation and it's a fine operation. I love that operation, but there are still significant issues with it.

When I take a patient for a thoracic stent graft, I make a one inch incision and in a day or two, that patient walks home. I have had no paraplegic patients. I have had no strokes. Knock on wood, I hope I don't have any in the future. It's truly an astounding technology.

This is critically needed technology and is far and away, I believe, better than the open operation and it hurts me to say that, because I'm a surgeon who loves to do open surgery. We need an endovascular alternative to treat these sick patients, and I implore you to recommend that this device get approved.

If and when it does get approved, and I hope that's soon, I think it needs to be restricted to the busiest centers, those centers that do a lot of open surgery and that have a lot of experience in endovascular stent grafting. I don't want to see any learning curve disasters, I think, that may be unlike or maybe like the carotid post-market surveillance work that is going on.

We need to have something like that here for 18, 24 months, have the most experienced people doing this procedure, giving good long-term follow-up results. And I think, ultimately, we'll find that this technology far and away exceeds what we can do in doing an open operation. As a busy open and endovascular surgeon, I want this technology, but my patients desperately need this technology. Thank you for your attention.

ACTING CHAIR MAISEL: Thank you. Are there any other individuals that would like to address the Panel?

DR. KARMY-JONES: Good morning. Thank you. My name is Riyad Karmy-Jones. I am a cardiothoracic trauma and now an interventional radiologist and cardiothoracic trauma surgeon at Harborview Medical Center at the University of Washington, Seattle, and unfortunately I have no financial attachment with Gore.

I just wanted to speak from two relatively known perspectives. Harborview is the only Level I trauma center for the WAMI region. We are effectively the county hospital for Washington, Alaska, Montana, Idaho and so on. The bulk of what we see in thoracic vascular and particular others are emergencies usually coming in at night.

So the two things I would like to talk about are much of which has been alluded to, is these devices can be placed very quickly, even quicker than an open operation, which can be critical in a patient with a complex leaking thoracoabdominal or thoracic aortic aneurysm. It does reduce the stress in these patients many of whom are actively being resuscitated as they present over minutes or hours to our institution, and we believe that there is a marked benefit for an endovascular approach.

And then I would just like to flip to the other side just very briefly, is that to consider that most of these devices are compared to open repair, but there are significant complications associated with medical therapy for these lesions, end organ failure, rupture, renal failure. We see patients who are presenting with bowel ischemia and renal failure and stroke because of prolonged aggressive medical management and are not candidates for the open repair.

So I think that these devices ought to also be considered as offering potentially a significant advantage over medical management in many cases for some of these patients. Thank you very much.

ACTING CHAIR MAISEL: Thank you. Is there anyone else who would like to address the Panel? At this point, we will close the open public hearing.

DR. ZUCKERMAN: Dr. Maisel, can I ask one question to the Panel?

ACTING CHAIR MAISEL: Of course.

DR. ZUCKERMAN: We began this open session with a very nice presentation from Dr. Ashar of FDA talking about a general construct for the Critical Path Initiative and some ideas that the Agency has for streamlining the translational process. We then got more specific with the work that Dr. White and the SVS have done recently in the AAA area.

While it's not the intent of the Agency to specifically endorse a particular registry or approach, my question though to the Panel members is when data like these are accumulated, and this Panel has looked at multiple AAA sponsor submissions, could these data be utilized instead as a control data set? Is it worthwhile to really actively examine other options in the AAA area? Any general responses would be appreciated.

ACTING CHAIR MAISEL: Dr. Somberg?

DR. SOMBERG: It's always useful to have comparative data and I think it can be very helpful, but in the early stages it's also very important to have control trials and a randomized base, and I think while we move into this, it's the case in drugs, it's the case in devices, as we move into a field it's always more arduous for the first carriers of the spear than the people who come behind, the mopping up after operations and all, no pun intended.

So I think while it can be useful, there is the other side of the coin, which is that there is a responsibility for those who are initially introducing a device, a technique or another therapeutic entity to try to have that in the context at least of one control trial, preferably a randomized one.

ACTING CHAIR MAISEL: Mitch?

DR. KRUCOFF: Yes. Bram, I think the spirit of the question has its own answer. Of course, they can be useful. We just have to be smart about how useful and when and where. And the two things that I think will be important drivers of that, one is just the opportunity to take proprietarily owned data sets and compile them is itself an organization issue we have struggled with on other fronts like stent data.

The other is just to stay, as I know you would very clearly, aware of how long a time period these different data sets are aggregated over, so that data from the '80s or the '90s, how much of a time change there is in collecting some of these less numerous cases.

ACTING CHAIR MAISEL: Dr. Johnston?

DR. JOHNSTON: I would echo some of those concerns, but as a member of the Society for Vascular Surgery and having looked at the data set, it is now maturing nicely and it is much more sophisticated in terms of data collection and openness than it was a couple of years ago, for example, and so I think that you will find this data set extremely useful in the future.

ACTING CHAIR MAISEL: Dr. Nicholas?

DR. NICHOLAS: I think there is a real use for this information, and I would recommend that we consider not only the time interval of this control surgical group, but also use it only up until the point where we see that there are some major new changes in open surgical techniques. This operation for open intrarenal aortic aneurysm is pretty well standardized and until something new comes along, I think it would serve to be a control group.

What I would recommend is that the initial effort to do this would be to combine a look using the Lifeline data as well as a contemporary control group for the first study or two to see if the hypothesis holds.

ACTING CHAIR MAISEL: Dr. Normand?

DR. NORMAND: I would just agree with Mitch very strongly, with his answer, but I would echo that not only is it the quality of the data that's collected. I think it would be very clever about the analytical methods that you're using. There is a lot of selection issues that I think a straightforward simplistic analysis is not going to be worthwhile.

So I think it's a great idea to use more data, but I think many people are going to have to be more open-minded about the analytic strategies you will use because of the huge selection issues in a registry database.

ACTING CHAIR MAISEL: Very well. So at this point, we will close the open public hearing and I would like to invite the sponsor to begin their presentation.

MR. NILSON: Thank you for being here today. My name is Mike Nilson. I am the product specialist for the GORE TAG Thoracic Endoprosthesis. We will refer to this as the TAG device throughout the remainder of the day. Also presenting with me today are Dr. Scott Mitchell and Dr. Michel Makaroun. Both Dr. Mitchell and Makaroun were principal investigators in our clinical trial program.

Dr. Mitchell will present both etiology and current therapy. I will present device history and design. Dr. Makaroun will present the clinical data, and then Dr. Mitchell will present the risk/benefit profile for the TAG device.

We are here today to request a recommendation for approval of the TAG device for the indication of endovascular repair for aneurysms of the descending thoracic aorta. Currently, there is no FDA-approved thoracic endovascular device to meet this therapeutic void. In the next presentation, Dr. Mitchell, who is a professor of cardiothoracic surgery at Stanford University, will discuss the etiology and current therapy of aneurysms in the descending thoracic aorta.

DR. MITCHELL: Thank you, Mr. Nilson, and good morning. First, I would like to clarify that as one of the co-principal investigators, that I have served as a consultant to the Gore Company for the last several years. However, other than that, I have no financial relationships with Gore nor any proprietary patent royalties.

This morning we will be discussing aneurysms, specifically aneurysms of the thoracic aorta. An aneurysm may be defined as a local or a focal dilation and weakening of the aortic wall, which is secondary to many processes. Today we will discuss primarily degenerative aneurysms, those that occur as a result of the ravages of hypertension and arteriosclerosis.

An aneurysm or rupture of the aorta, both the thoracic and the abdominal aorta, is estimated to cause 32,000 deaths annually in the U.S. To put this in perspective, breast cancer accounts for 41,000 deaths annually.

After Juan Perotti first described an endovascular approach for abdominal aneurysms in the 1980s, Gore became involved with the development of endovascular repair for aneurysmal disease in 1994 and in 1997 initiated the clinical evaluation of abdominal and thoracic endoprostheses.

By 2002, the Gore EXCLUDER device was approved for the repair of abdominal aortic aneurysms and by 2004, there had been over 20,000 EXCLUDER implants worldwide. In 2005, we now have the opportunity to have the TAG device considered for FDA approval.

The area that we will address today is the descending thoracic aorta, an area uniquely suitable for endograft technology because of its relatively straight course with few side branches. It is bounded by the transverse arch superiorly and the diaphragm interiorly. Neighboring vessels of the distal arch include the left subclavian artery and just below the diaphragm, the celiac axis.

Thoracic aneurysms may be either focal or diffuse, but they share one critical natural history phenomena and that is that of continued dilation until they eventually rupture, and the goal of all our therapies, open or endovascular, is to prevent aneurysm rupture and death. Although most aneurysms are asymptomatic, some present with symptoms with pain, compression of the esophagus or traction on an adjacent nerve.

There are, approximately, 15,000 new cases of thoracic aortic aneurysms diagnosed yearly, which results in over 5,000 surgical repairs. Nevertheless, there are still an estimated 2,500 deaths annually from rupture in the U.S. We have good outcomes data from probably the most studied county in the U.S., that of Olmstead County in central Minnesota.

Over the 1960s and '70s, there was a fairly uniform incidence of thoracic aortic aneurysms of about three per 100,000, but in the two recent decades, we have seen an increase to now about 10 per 100,000. Whether this represents a true increase in incidence or it reflects our aging population or perhaps even our increased diagnostic capabilities is unknown.

However, we do know several things and that is that the risk of rupture is increased as the aneurysm grows in size. It's increased in older patients and in patients who have concomitant emphysema or COPD. Additionally, patients who present with pain over the rapid increase in size are at increased risk for rupture.

And so that we can see on the bottom line of this graph, if your aneurysm is less than 4 centimeters, your risk of rupture at five years is fairly nominal, 3 to 4 percent. But if your aneurysm exceeds 6 centimeters in size, there is about a 10 percent incidence per year rupture with over 30 percent having ruptured at the end of a three-year interval.

These slides depict a fairly typical mid descending thoracic aortic aneurysm with a magnetic resonance angiogram on the left, the surgical exposure of the same aneurysm through, as you can imagine, a fairly broad incision to get this type of anatomic exposure. You need room to operate on these patients safely and by these approaches, you inflict significant morbidity as has been referenced in some previous remarks. With adequate exposure and good technique, you can affect a very effective surgical repair as seen on the right.

One problem with surgical repairs is that they are morbid and the older the patient, the worse the morbidity. Increasing age and the frequently concomitant pulmonary disease puts these patients at highest risk, which presents surgeons with the dilemma that it is the older, sicker patient who most needs the operation, but it's that same patient who is most at risk for catastrophic complications.

The open surgical repair is effective and durable, but it does exact a significant toll. Mortality in this series was 6 percent. 14 percent of patients experienced paraplegia, and there was a 70 percent incidence of cumulative morbidity, that is the number of patients who incurred any major complication in the postoperative period and recovery is frequently protracted.

Patients, as we have heard, are fearful of this operation. They frequently have lived independently and now to suddenly be debilitated is a major problem for them and many refuse operation. We saw a fairly significant onslaught of these patients in the early 1990s at Stanford University and were impressed with the age and comorbidities.

We teamed up with our interventional colleagues and formed our own thoracic stent graft program, which was approved by our own IRB for high risk patients. We constructed a hybrid device from FDA-approved Z stents covered with an improved Dacron graft and began a high risk trial, which resulted in 103 patients being treated.

By our own estimates, 60 percent of these patients were absolute nonoperative candidates with either unstable coronary disease, very severe obstructive pulmonary disease or two or greater previous attempts at repair. Indeed, we expected a surgical mortality with open procedures exceeding 30 percent, and our 9 percent endovascular mortality we thought was quite respectable and prompted us to continue more investigations.

The fundamentals of endovascular repair are fairly straightforward. It's a minimally invasive procedure usually through an incision in the groin or a small flank incision. We now can reliably deliver and deploy endovascular devices whose hemostatic seal excludes the aneurysm from the circulation and, thus, prevents aneurysm rupture.

This is an angiogram of our very first TAG device patient, a very pleasant 72 year-old woman who had presented with a rather dramatic increase in size. And as you can see from the angiogram on the left, this is a pretty sizeable aneurysm, which is completely excluded by the fourth postoperative day when she left the hospital.

In summary, 6 centimeter aneurysms of a thoracic aorta have a rupture rate of, approximately, 10 percent per year. Open surgical repair is effective and durable, but the cumulative morbidity of 70 percent or greater and our own 6 percent mortality remains substantial and there are other limitations. The early results of thoracic endovascular repair showed potential patient benefit. I would like to return the podium to Mr. Nilson.

MR. NILSON: The TAG device has been thoroughly studied. We began implanting the device in the U.S. in 1998 beginning with a feasibility study, TAG 97-01, followed in 1999 by the pivotal study, TAG 99-01. Due to fractures in the deployment wire, Gore chose not to pursue FDA approval until we could modify the design to minimize the likelihood of these wire fractures. Consequently, in November of 2001, Gore voluntarily withdrew the device from commercial distribution and the device was modified. I will describe those modifications in a minute.

After the modifications were completed, Gore conducted a confirmatory study, TAG 03-03, designed to confirm the preclinical test results of the modified device. A treatment IDE, TAG 04-02, allows study centers access to the device pending approval and is currently ongoing. We have five-year follow-up data on patients in the feasibility study and two-year follow-up on those in the pivotal study. The confirmatory study finished follow-up in August of 2004, and we're continuing to follow all patients through five years.

Between 1998 and 2001, 2,800 original devices were implanted in over 2,100 patients, mostly in Europe. During the period of device modification from November of 2001 until November of 2003, the use of the original TAG device was limited to nonsurgical patients in three centers in the U.S. Since November of 2003, over 1,500 modified devices have been implanted in over 1,100 patients, again, mostly in Europe where the device has been commercially available since March of 2004.

Notice from the numbers that most patients received more than one device. Today we will focus on the U.S. clinical trial data. The picture you see is the original TAG device, which was designed with the deployment wire, also referred to as a spine, and is highlighted by the red box. The purpose of this wire was to provide longitudinal support during deployment. It counteracted forces due to blood flow during deployment until the device engaged in the aortic necks. Once the device was seated into the necks, the wire had no further design function.

As mentioned previously, this wire had a higher than anticipated fracture rate. The deployment wire fracture rate was 32 percent in our longest term test cohort patients. Only five out of the 44 patients who have been identified with fractures have clinical sequelae associated with these fractures. These reported sequelae are endoleaks, predominantly Type III. We used information gained from the clinical use of the original design to design tests to replicate the failure mode and ultimately leading to the modification of the TAG device.

At this time, I would like to hand out samples of both the modified and original TAG device and I will collect these samples after my portion of the presentation. For everybody in the audience, I'm going to allow the Panel two minutes to look at these devices before I start my presentation up again.

MR. MORTON: Mike, can we take them out of the bags?

MR. NILSON: Yes, you can remove the devices from the bag. They are in the bags, because there is a paired sample of an original and modified device in each bag.

ACTING CHAIR MAISEL: I think you can continue with your presentation while the Panel is looking at the devices.

MR. NILSON: Okay. Gore minimized design modifications in order to maintain device attributes and clinical performance while eliminating the deployment wire. The modifications did not change the device's fundamental design. To compensate for the loss of this deployment wire, the graft material was strengthened.

The original TAG device graft material was constructed from two fluoropolymer layers. The modified TAG device is constructed from three fluoropolymer layers. The additional layer, which is similar to that incorporated into the marketed EXCLUDER Bifurcated Endoprosthesis, is sandwiched between two original layers and this layer provides the longitudinal stiffness that was formerly provided by the deployment wire.

The TAG device is a symmetrical tube consisting of a nitinol self-expanding stent and a fluoropolymer liner. The stent is attached to the liner without sutures by trapping the wire between the liner and the attachment film. Flares are located on both ends of the device to aid in conforming to tortuous anatomy. Sealing cuffs on both ends of the device help exclude the aneurysm from circulation by eliminating endoleaks. At the base of the flares are two radiopaque gold bands, which aid in placement and follow-up.

There is a deployment sleeve, which constrains the device on the end of the delivery catheter and remains permanently attached to the device after deployment. The TAG device has a flexible 100 centimeter working length catheter to access the descending thoracic aorta from the groin. Radiopaque olives at both ends of the device protect the device during insertion and manipulation, as well as facilitate position.

A fluoropolymer deployment sleeve constrains the device on the leading end of the delivery catheter. There is a guidewire port that accommodates .035 inch guidewires and a flushing port that removes trapped air from the guidewire lumen.

A deployment knob is located at the control end of the catheter and has a deployment line that runs the entire length of the catheter connecting the deployment knob to the deployment sleeve. Pulling this knob releases the device from the catheter at its desired target. The delivery catheter and deployment method remains exactly the same as the original device.

The following animation shows how the TAG device is delivered to its desired location within the body. You will see an aneurysm in the descending thoracic aorta distal to the left subclavian. That is an .035 inch guidewire accessing across the aneurysm. The device is being advanced and positioned to its desired target and the deployment is initiated in the middle and extends to both ends, again in slow motion, and this facilitates a very accurate deployment.

The following video shows a real time deployment of a TAG device in a patient with an aneurysm in the descending thoracic aorta. Because the deployment is rapid, this video will repeat the deployment sequence several times in succession. On the top course of the screen, you will see the deployment of a TAG device. If you're having trouble seeing it, notice the curved shape the device has constrained on the end of the delivery catheter.

Once the device is released off the catheter, it conforms to the anatomy. Risk analysis was performed to determine potential effects of this device modification. This analysis was essential in determining testing requirements to verify that device modifications would not adversely effect device performance. The TAG device has been extensively tested in our Comprehensive Testing Program. This testing included newly developed durability tests that replicate the deployment wire fractures.

The requirements deemed appropriate were developed through a combination of established ISO standards and collaborative efforts between industry and the FDA. These tests assure the TAG device functions as intended, which is to exclude aneurysms from circulation and prevent aneurysmal rupture. This is an example of one of our many durability tests. This test was specifically developed to replicate the deployment wire fracture. Notice the extreme curvature of the spine in the original device in the lower left. To confirm these results related to the device functional performance, poor conducted or limited clinical trial, TAG 03-03, we'll even call the confirmatory study. We will now collect the samples.

In the next part of our presentation, Dr. Makaroun, who is a Professor and Chief of Vascular Surgery at the University of Pittsburgh, will present results from our Clinical Trial Program.

DR. MAKAROUN: Thank you, Mr. Nilson. Good morning. I would like to start by declaring that I do serve as consultant for W.L. Gore and I have received both educational and research grants from W.L. Gore as well as just about any other manufacturer in this field.

It really gives me great pleasure to be here today on behalf of all the investigators that participated in these clinical trials to share with you the results of the three phases of the TAG development that so far have spanned over seven years, over the first device being implanted in February of 1998. All the investigators, as well as myself, have looked forward to this day with much anticipation and the hope that we can finally bring this technology to our patients.

The first study was the feasibility study that started in 1998 and concluded enrollment in 1999. This study was carried out at two sites in the United States and in all enrolled 28 patients with descending thoracic aneurysms. The mortality at 30 days was only one patient or 3.6 percent. For one year, the mortality was 21 percent with no incidents of paraplegia or stroke. Renal failure and myocardial infarction was noted in only one patient each or 3.6 percent.

Through a five-year follow-up period two additional AEs long-term were reported between two and five years. All-cause mortality at five years was 25 percent. Endoleaks were noted at any time in 21 percent of the patients and was a growth in 18 percent, fractures in 32 percent. There was one conversion and two reinterventions over time to replace additional devices.

None of the following events occurred during the follow-up. There were no aneurysm ruptures, migration, extrusion, erosion, lumen obstruction or branch vessel occlusion over time. These encouraging results left the development of the pivotal Phase II Trial, the 99-01 Trial, that started enrolling patients in 1999 and completed enrollment in May of 2001. The pivotal study was a multicenter study that was carried out at 17 clinical sites in the United States and was designed to be non-randomized with a control arm.

The test subjects were all treated with a TAG device and were compared to post-subjects that were treated by the traditional open surgical repair. One-year clinical endpoints were used for the analysis, but all patients were to be followed for five years and the follow-up is still ongoing. The control group were all enrolled from the same sites as the patients undergoing the TAG device. They were in two groups, 44 patients were enrolled concurrently with the device during the study, in addition to 50 patients that had recently undergone open repair at the participating centers.

This strategy actually was quite successful in generating probably one of the largest series of isolated descending thoracic aneurysms available for comparison. To limit bias, the historical cohort was enrolled by working facts sequentially from the last patient that was treated prior to the initiation of the study. And a goal was set to have no more than five subject enrollment difference between the TAG and the control site.

As such, 82 percent of the surgical controls had their procedures between January of 1998 and May of 2001. The historical and concurrent groups were as such very similar in all major demographic and clinical variables. The primary safety hypothesis that was tested in this pivotal study was that the percentage of subjects with more than one major adverse event through one-year post-treatment will be lower than the TAG device group when compared to the surgical control group.

The primary efficacy hypothesis was that the percentage of subjects, three from major device- related events through one-year follow-up for the TAG device group will exceed 80 percent. The secondary hypotheses were also tested and they were the procedural blood loss, ICU and hospital stay as well as the time period to return to normal activities will be lower in the TAG device group compared to the surgical control group.

The term major as used in this trial was derived from the Sacks criteria published in 1997 and these were the only criteria available for classification at the time. A major adverse event was one that required therapy and post-hospitalization between 24 and 48 hours or required major therapy and unplanned increase in level of care or prolonged hospitalization resulted in permanent adverse sequelae or death.

A minor adverse event is one that requires no therapy and is of no consequence or requires nominal therapy and is of no consequence, including an overnight admission for observation. For the two separate adverse events to be tracked in the study, were PDP classified and are shown here in these two slides. To illustrate, a paraplegia resulting in permanent deficit or a groin psuedoaneurysm requiring repair would be classified as major adverse events in this trial, while a transient mental status change not prolonging hospitalization or a small groin hematoma and not requiring treatment would be classified as a minor adverse event.

The sample size estimate for the pivotal trial was based on the primary safety endpoint, which was the one year incidence of major adverse events. Allowing for a Type I error rate of 0.05 (two-sided) and a power of 0.8, the controlled incidence of major adverse events was estimated at 40 percent and the test incidents of major adverse events was estimated at 20 percent. This resulted in a sample size of 82 in both groups for comparison.

Allowing for two training cases per side and subject attrition due to loss to follow-up intent- to-treat failures and deaths, 140 test subjects were enrolled and treated with the TAG device and compared to the 94 controlled patients that were treated by open surgery, 44 were enrolled concurrently with a TAG and the 50 historical patients enrolled by the consecutive review of the most recent surgical patients in reverse chronological orders at the same institutions.

Key inclusion criteria for all patients in this trial included a descending thoracic aneurysm that necessitated surgical repair, defined as a fusiform aneurysm twice the size of the healthy aorta or any size saccular aneurysm. All patients have to have a life expectancy of more than two years, be surgical candidates and be aged more than 21 years. Exclusion criteria for all patients from this study were mycotic aneurysm and uncontained aneurysmal rupture, all patients with aortic dissections, both acute and chronic, were excluded.

We did not allow planned concomitant surgery or major surgery within 30 days of treatment, MI or stroke within six weeks of treatment, renal insufficiency and degenerative connective tissue disorders. Specific inclusion criteria for the TAG patients, obviously, required in aortic morphology that meets the IFU guidelines, namely aortic diameters between 23 and 37 millimeter and at least 2 centimeter healthy proximal and distal necks.

Specific exclusion criteria for the TAG patients include the patients that have a different size aorta above and below the aneurysm and the inability to compensate for that taper with multiple devices. Patients with significant thrombus of the proximal or distal landing zones, a planned occlusion of the left carotid or the celiac artery and respiratory insufficiency precluding thoracotomy.

Pre-operatively the patients underwent the standard physical examinations, blood tests and medical assessment. Additional imaging including an angiogram and CT scan. Angiography with a marker catheter was used to assess the length of the neck and the length of the aorta to be covered, the location of the aneurysm and the tortuosity associated with it, as well as the axis vessels required to reach that area.

The CT scan was used to get the size and the quality of the proximal neck, the size of the aneurysm, as well as to assess the distal size of the neck and the quality of the aorta at that level. The device was usually inserted through a small groin incision with a contralateral puncture for the angiographic catheter used in the deployment sequence to assess the exact location of the graft.

The procedure usually started with angiography followed by positioning of the graft to the desired location and concluded with a post-deployment angiography to ensure complete exclusion of the aneurysm. All surgical patients underwent the standard left thoracotomy and standard aneurysmorrhaphy.

Post-procedure, the patient had a full view chest X-ray with views specially designed to show the endograft at discharge. A follow-up schedule was at one month with a CT scan, then at 6 months, 12 months and yearly thereafter with both a chest X-ray and CT scan. This is an example of some of the views that are required to evaluate the graft, and the CT scan was used for the evaluation of endoleaks and the size of the aneurysmal sac over time.

The baseline demographics were very well- matched between both groups. The TAG device group was three years older than the surgical controls, but this was not significant. Of note is that the TAG device group and the surgical controls were very well-matched when it comes to gender, which is different than the previous trials that enrolled in the abdominal area. This is of particular importance as the aneurysms of the descending thoracic aorta do not show the same predilection to males as the abdominal aorta does.

Baseline aortic morphology was again well- matched between both groups, except for the smaller diameter of the proximal and distal necks in the TAG device, which is expected because of the requirements for sealing.

Of most importance is the aneurysm diameter, which is one of the most important predictors of rupture, as well as an independent predictor of major adverse events. The aneurysm diameter was very well-matched between both groups.

The baseline comorbidities were also quite similar between the TAG device and the surgical control group. Although coronary artery disease appeared to be more prevalent among the TAG device group, this difference was not significant. Symptomatic aneurysms, however, were significantly more prevalent in the surgical control groups compared to the TAG device.

The risk classifications of the patients in this trial was carried out based on the standard ASA classification and the SVS risk score, and there was no significant difference in either classifications. Both groups had the same risk classification.

There was, however, an imbalance at baseline for the New York Heart Association classification. This particular classification was used mostly to exclude patients with a New York Heart Class IV, which was an exclusion criterion in the study. The large number of patients who did not have classification noted makes the comparison in this category very difficult.

In all, 140 patients underwent the TAG device group in the pivotal trial. 137 of them or 98 percent had a successful implantation of the device. All three failures were due to poor iliac access. 77 patients or 55 percent required more than one device to bridge the aneurysm. 21 patients or 15 percent required an iliac conduit to the aortic iliac segment to access the aorta. This is an example of the most commonly performed conduit, which is a 10 millimeter Dacron graft to the common iliac artery through a small flank incision.

Operative results that are quite important to surgeons practicing in this field are presented here in the standard reporting format for the surgical area, which is 30 days or in-hospital event rate even after 30 days if the patient stayed in the hospital. If you use this formula, operative mortality is 2 percent for the TAG device and 6 percent for the surgical control. Paraplegia was 3 percent in the TAG device and 14 percent in the control and stroke were both 4 percent in both groups.

The primary safety endpoint in this trial was the percentage of subjects free from major adverse events through one year of follow-up, and the results show a marked reduction in the major adverse events in the TAG device group compared to the surgical control group that was highly significant. 42 percent of the TAG device patients had any major adverse event through one year while 77 percent of the surgical control group had major adverse events over the first year.

This therapeutic benefit was evident in the following categories. Both bleeding and pulmonary showed a significant reduction of the major adverse events compared to the surgical control group, and this was due to a high percentage of procedural bleeding in the surgical control group and the respiratory failure in the post-operative period.

Renal and wound complications also showed a significantly lower proportion in the TAG device group compared to the surgical control. Of particular note, the neurologic complications in the TAG device group were lower than the surgical control group. Although the patients who had cardiac events was lower in the TAG device group, this was not significant.

The only category that showed a higher major adverse event rate in the TAG device group was that of vascular events, and this was related to the large sheath that was required for the introduction of the device through the iliac system. 11 percent of those were due to vascular trauma.

The Kaplan-Meier estimates of the freedom from major adverse event over one year shows a substantial advantage to the TAG device group over the surgical controls that is highly significant through the first year. Actually, a 61 percent reduction of these major adverse events is evident by the first 14 days due to the high event rates in the surgical arm periprocedurally.

Of note is that 70 percent of all major adverse events noted in the first year in the TAG device group occurred in the first 30 days. This was also noticed previously in the 97-01 trial where 63 percent of all events over five years actually were noticed in the first 30 days.

Carried all the way through two years, you can see that the benefit to the TAG device group remains significant. All-cause mortality through two years was no different between the TAG device group and the surgical control group. The TAG device group over two years had 24 percent of the patients succumb and in the surgical control group, it was 26 percent. The causes of death are typical for this elderly population with the associated comorbidities.

Although there is an early numerical advantage to the TAG device group that is associated with the early mortality from the surgical procedure, the freedom from all-cause death through two years is no different for the two arms of the study.

Included in this all-cause mortality is the more relevant aneurysm-related mortality, which is defined as the death prior to hospital discharge or death within 30 days of the primary procedure or any secondary procedure to treat the original aneurysm, which also includes death from ruptures.

Freedom from aneurysm-related mortality through two years was 97 percent for the TAG device group and 90 percent for the open surgical controls. This difference is significant. And as you can note from the graph, there were no mortalities in either arm after the first year that was related to the aneurysm.

In summary, the primary safety endpoint of this pivotal study was met with a significantly lower proportion of TAG subjects experiencing major adverse events through one year of follow-up. It was 42 percent in the TAG group and 77 percent in the open surgical group.

The primary efficacy endpoint of this pivotal study was the percentage of subjects that were free from major device-related events through one-year follow-up for the TAG device group. The efficacy for the surgical procedure was assumed to be 100 percent. A predefined point estimate of 80 percent for the endovascular group was considered to be a reasonable efficacy outcome and since the device was expected to show a considerable improvement in safety profile.

The major category in the device-related events through one year was again derived from the Sacks criteria and included for the device-related events, endoleaks, migration and realignment, aneurysm enlargement, branch vessel occlusion, deployment failure, extrusion erosion, lumen obstruction and material failure.

To illustrate the definition, endoleaks requiring intervention, such as an additional device, will be classified as a major device-related event. However, endoleaks not requiring any intervention and being observed by serial imaging will be classified as minor device-related events.

Since endoleaks are the most frequent device-related events in most of these trials, they were further classified according to the same classification for the abdominal aneurysms. Type I was due to either proximal or distal attachment sites. Type II was due to retrograde flow from branches, Type III from a structural defect or junctional endoleak and Type IV from material porosity.

Freedom from major device-related events in this trial over one year was 94 percent for the TAG device. This freedom from major device-related events was significant when compared to the predefined limit of 80 percent. There were eight major device-related events during the first year.

Since 10 patients did not have their 12 month follow-up visit, a worst case analysis was performed assigning a major device-related event to all 10. If that is carried out, the freedom from major device-related events drops to 87 percent, but even at that level, the lower 95 percent confidence interval is 80.4 percent.

The freedom from major device-related events carried through two years show a stable line after the initial six months without additional major device-related events, especially during the second year. It continues to be 94 percent both at one and at two years and both are significant compared to the 80 percent predefined limit.

So in summary, the primary efficacy endpoint of this pivotal trial was also met with 94 percent of the test subjects free from a major device-related event through one year. This was statistically greater than the predefined limit of 80 percent. In addition, there were no aneurysm ruptures noted through these two years.

Secondary outcomes are reported here by the median value. Procedural blood loss was 250 ml in the TAG device group and 1,850 in the surgical controls. No p-value was reported because of the large number of missing data from the surgical control. ICU stay was one day for the TAG device and three for the surgical controls, and the hospital stay was three days for the TAG device group and 10 for the surgical control. Time to return to normal activities was reduced to 30 days in the TAG device compared to 78 days for the surgical control.

To summarize the pivotal trial, the TAG device was safe and effective for the treatment of aneurysms of the descending thoracic aorta. The primary safety endpoint was met with 42 percent of the TAG device group having major adverse events and 77 percent for the surgical controls. The primary efficacy endpoint was also met with 94 percent freedom from major device-related events. Both were highly significant. All secondary endpoints were met.

Despite these excellent results, the sponsor chose not to seek approval of the device to the dismay and chagrin of most of the investigators. They decided to proceed with a modification of the device to eliminate the failure mode that was identified in the longitudinal spine fractures.

Mr. Nilson has already discussed with you the preclinical testing that showed the device to be at least the same or better in most of the bench testing that were performed. A confirmatory study was initiated after the modification to ensure that the early deployment and early results are satisfactory as the spine had some function in the deployment of the device.

The confirmatory study was started in 2003 and finished enrollment in June of 2004 and, again, it was conducted to confirm the functional performance of the modified TAG device during deployment and through the first 30 days. A 30 day endpoint was chosen based on the TAG 99-01 Study, which showed that 70 percent of the major adverse events occurred within the first 30 days in the periprocedural period. That difference was also maintained from 30 days all the way through two years.

The study was carried out at 11 sites. All but one participated in the TAG 99-01 pivotal trial. It was designed as, obviously, a non-randomized prospective trial, all test subjects treated with the modified TAG device, and they were compared to the control data from the pivotal study, the TAG 99-01. 30 day study endpoints were used, although all patients are planned to have a follow-up of five years.

Identical inclusion and exclusion criteria were used in this study compared to the pivotal study to allow the comparison to the control data from the pivotal trial. The primary safety endpoint for this study was the percentage of subjects with more than one major adverse event through 30 days post-treatment in the TAG device group compared to the surgical control group from the TAG 99-01 Study.

The efficacy endpoint was the percentage of subjects with major device-related events in the TAG device group through 30 days of follow-up. These same secondary endpoints were used for this as the 99-01 and included the procedural blood loss, ICU and hospital stay and the time to return to normal activities.

The sample size estimate for the confirmatory study was again based on the primary safety endpoint, which was the 30 day incidence of major adverse events allowing for a 5 point error rate of 0.05 and the power of 0.86. The controlled incidence of major adverse events was assumed to be 63 percent and the expected incidence of the TAG device in major adverse events was 38 percent. This led to an estimate of a sample size of 40 requiring the new modified TAG device to be compared to the 94 patients that were in the control arm of the 99-01.

Again, allowing for some subject attrition, 51 patients were enrolled in this study and treated with the modified TAG device and they were compared with the same 94 control subjects that were derived from the TAG 99-01 Study.

The preoperative assessment was very similar to the 99-01 Study, including physical examination, blood test and imaging. The follow-up at discharge, again, included the chest X-ray and at the 30 day follow-up visit, both a chest X-ray and a CT scan were included. All the subjects will continue, obviously, to be followed up through the next five years.

Baseline demographics were, again, quite similar between the TAG device group and the surgical control group. There were some more male patients in the TAG device group in the 03-03, but this difference was not significant.

Baseline aortic morphology was, again, very similar to the 99-01 Study with well-matched groups between the TAG device and the surgical controls with the only difference being the smaller size of the proximal neck diameter and the distal neck diameter in the TAG device group because of the anatomic requirement of the procedure.

The aneurysm diameter was quite similar between the TAG device and the surgical control group. Baseline comorbidities were also well-matched between the TAG device group and the surgical control arm. In this comparison, the symptomatic aneurysm difference did not reach statistical significance. However, there were more prevalence of cancer or a history of cancer in the TAG device group compared to the surgical control.

Risk classification according to the ASA was very well-matched between the TAG and the surgical control. The SVS risk score was slightly higher in the TAG device group and this was significant. Again, for the New York Heart Association, there was a large number of patients that were not classified and in this particular case did not reach significance.

The safety endpoints through 30 days again were quite striking showing a significant advantage to the TAG device group compared to the surgical control, major adverse events were noted in 12 percent of the patients treated with the TAG device and 70 percent of the surgical controls. With several categories showing this therapeutic benefit of the TAG device over the surgical controls, including bleeding, pulmonary, cardiac, renal, and again neurologic complications.

The difference in major adverse events between the two groups was significant. Vascular complications were again noted in more patients with the TAG device group compared to the surgical control group and this difference in this particular case was not significant. The freedom from the major device -- from major adverse event through 30 days showed a significant advantage to the TAG device group compared to the surgical controlled with the P being less than 0.001.

Most of the advantage is noted very procedurally in the very early post-operative period. This slide illustrates all three groups and both trials showing that both the TAG device group from both studies delivered some therapeutic benefit compared to the surgical control.

In summary, the primary safety endpoint of this trial was met with significantly lower proportion of TAG device subjects experiencing major adverse events through 30 days compared to the TAG 99-01 surgical control. There were no TAG device deaths through 30 days. The efficacy endpoint of this confirmatory study was the freedom from major device- related events through 30 days post-treatment. And since no patient experienced a major device-related event in the confirmatory study, the efficacy was 100 percent. The lower 95 percent confidence interval was 93 percent.

In other worst case scenarios, analysis was performed assigning major device-related events to the two patients that did not complete a 30 day follow-up and this reduced the efficacy to 96 percent with a lower 95 percent confidence interval of 86 percent. The secondary outcomes measured in this confirmatory study are reported as a medium value. The procedural blood loss was 200 ml in the TAG device group and 1850 in the surgical control.

The ICU stay was again one day in the TAG device and three days in the surgical control. The hospital stay was three days for the TAG device and 10 in the surgical control and again the time to return to normal activities was shortened in the TAG device group to 15 days versus 78 for the surgical control.

In summary, the confirmatory study confirms the result of the peak in testing, that the modified design is equivalent or improved over the original design with the primary safety endpoint met, 12 percent TAG versus 70 percent major adverse events in the surgical control with the difference being highly significant and the primary efficacy endpoint being 100 percent freedom from major device-related events.

In conclusion, the studies of the TAG device show that for treatment of aneurysms of the descending thoracic aorta the TAG device is safer than open surgical repair. It provides effective treatment for aneurysms of the descending thoracic aorta and results in less blood loss, shorter hospital and ICU stay and a quicker return to normal activities compared to the open surgical repair.

Now, I yield the podium to Dr. Mitchell, who will discuss with you the risks and benefits of the TAG device.

DR. MITCHELL: Thank you, Dr. Makaroun. As we have seen, the open repair of descending aneurysms incur significant morbidity and mortality, specifically 6 percent mortality in our series and a 77 percent cumulative morbidity. Compared to the open procedure, the TAG device was able to dramatically lower 30 day mortality from 6 to 1 percent and total morbidity from 77 to 42 percent with a reduction in paraplegia from 14 percent to 3 percent and major pulmonary complications from 38 to 13 percent.

These translate into significantly better patient outcomes and recovery. However, when we consider endovascular repairs, we do introduce some new risks which are specific to endovascular repair and they are listed here. There are possibilities for deployment failure, branch vessel occlusion can occur from inadvertent coverage. There is the new possibility of injury to access vessels. We now have a new problem called endoleaks and aneurysms can enlarge afterwards, devices can migrate and there are instances of material failure.

However, as we've seen in this experience, the incidence of these events has all been relatively low. Less than 4 percent for the greatest problem and ranging around 1 percent for most of these complications. And additionally, most of these complications can be taken care of with subsequent endovascular procedures. The modifications to the TAG device eliminate the risk associated with fractures of the deployment wire and its mechanical and deployment properties have been extensively confirmed by mechanical testing, as well as by the TAG 03 evaluation.

We examined the embracement of endovascular repair for abdominally aortic aneurysms as reported in the State of New York through the years to 2000 and 2002. We can see a fairly significant increase in the utilization of these procedures in 24 to 60 hospitals. And even with a fairly large number of institutions performing this for the first time, there remains significant reductions in mortality from 4 to 1 percent for endovascular procedures, suggesting that this technology can be translated.

The benefits of the TAG device as we have demonstrated are the significant reduction of major adverse events, hospital stay can be reduced and patients return to normal activity more quickly. There is also a significant improvement in aneurysm- related mortality and to date we have seen no evidence for aneurysm rupture in the late follow-up.

We should note, however, that although we have made comparisons to endovascular repair of abdominal aortic aneurysms, make no mistake the operative repair of thoracic aortic aneurysms is a more severe procedure and the incurred morbidity increases 2 to 3 fold. The incremental benefit afforded to these patients by repair, endovascular repair of these descending aneurysms is dramatic. We feel this device offers patients and their physicians an important therapeutic alternative.

Speaking as a surgeon, we need this device and our patients want it. Thank you.

MR. NILSON: Thank you, Dr. Mitchell. The sponsor is proposing the following post-market program to evaluate long-term performance of our patients enrolled under the pivotal confirmatory and treatment IDE studies, we will follow approximately 250 patients through five years. This includes approximately 100 modified TAG device patients. A post-market study is being discussed at the Agency and could include up to 100 patients and up to 25 centers. The purpose of this study is to evaluate the performance of the procedure in a wider community.

The indication we propose for your consideration is the following: The GORE TAG Thoracic Endoprosthesis is indicated for endovascular repair of aneurysms of the descending thoracic aorta. The presentation is concluded. Thank you very much for your attention.

Before we start the Q&A session, I would like to introduce a few potential respondents for the sponsor. Dr. John Matsumura, who is a Professor of Surgery at Northwestern University and an investigator in the Clinical Trial Program. Dr. Joel Verter, who is a biostatistician, and Mr. Lou Smith is a medical products technical leader for Gore and co-chairs the AAMI Vascular Prosthesis Committee. I will be the moderator for the sponsor during the Q&A session and we welcome your questions at this time.

ACTING CHAIR MAISEL: Thank you very much for a very thorough and eloquent presentation and I would like to open the session now to questions from the Panel, reminding the Panel that we will have ample time to discuss these issues further this afternoon. Dr. Normand?

DR. NORMAND: I just have some questions of clarification. Could you define for me what you mean by intent-to-treat failures? I'm not sure what you mean by that.

MR. NILSON: The TAG 99-01 Study was an intent-to-treat study.

DR. NORMAND: Yes.

MR. NILSON: Which means any patient who has consented under either arm, even if they received the device, was still enrolled in that part of the trial.

DR. NORMAND: But it sounded like you said you accounted for attrition as well as intent-to-treat failures. So then the question is did you do an intent-to-treat analysis?

MR. NILSON: We included intent-to-treat failures in our worst case analysis.

DR. NORMAND: Okay. And then I just have two more quick clarifications. You had mentioned follow-up schedules like one month and 12 months. From what time point is that measured? From the time of the procedure?

MR. NILSON: From discharge.

DR. NORMAND: From discharge.

MR. NILSON: Yes.

DR. NORMAND: And was that just for the TAG patients?

MR. NILSON: I believe it was for both.

DR. NORMAND: So even for the surgical repair arm it was from discharge?

MR. NILSON: Yes, it was from discharge.

DR. NORMAND: Thank you.

ACTING CHAIR MAISEL: John?

DR. SOMBERG: A couple of questions. The first is for the confirmatory study, it's 30 days is the follow-up and I understand the rationale, but do you have additional follow-ups since the brief and what was presented, you know, probably put together information months ago? And I just wondered if there's additional information? Is there anything out of the ordinary in the follow-up of the confirmatory study?

MR. NILSON: We do have additional follow-up in preparation for our upcoming one-year follow-up. We have had one death and we have had one major device-related event, which was an aneurysm. And we have contacted 90 percent of the subjects who were enrolled in the trial. So 46 of the 51 patients enrolled have been contacted post-30 days.

DR. SOMBERG: You also in a narrative discussion, in one of the cases that had a mortality, something called post-implant syndrome for TAG. I just wondered what that was. And that was not mentioned or categorized in any other areas. Is that a problem with the -- is there such a thing as a post-implant syndrome?

MR. NILSON: I would defer to Dr. Makaroun to give the clinical perspective on post-implant syndrome.

DR. MAKAROUN: Early in the experience with endovascular treatment of aneurysms of the abdominal aorta, post-implant syndrome of fever and prostration and various systemic symptoms was described. It actually was relatively frequent with the hand-made devices in a variety of settings and that's why it is included, essentially, with most of these trials and continued to be follow-up with the advent of the commercial available devices with sterilization of product. This implant syndrome has been extremely rare, but it's customary to continue to include it as one of the possible adverse events.

DR. SOMBERG: I had one further.

ACTING CHAIR MAISEL: Dr. Johnston?

DR. JOHNSTON: Can you clarify in the censored data and in other parts of the submission, I could not get a sense of how many patients were actually lost to follow-up.

MR. NILSON: Dr. Verter will answer that.

DR. VERTER: My name is Joel Verter. I'm a statistical consultant along with other members of my group on this project. We have no equity interest in the company. Censoring is used in the sense of a statistical term. There are actually no patients that were really censored in the sense of not providing information. So, for example, in answer to previous questions, those patients who didn't have a 12 month visit when we did the one-year analysis, you get a worst case analysis and provided then, imputed to them an actual event.

If you're referring to the 03 Study, perhaps, in particular, where at the end of 30 days it is indicated that 60 subjects are in the censored column, is that what you are referring to?

DR. JOHNSTON: No, I'm not. I'm referring to the fact that when I went through the individual data and I'm familiar with what censoring is, I could not anywhere determine whether 100 percent of the patients were, indeed, followed or whether X number were lost to follow-up and therefore not included.

DR. VERTER: Okay. In --

DR. JOHNSTON: I'm simply looking for the number of patients.

DR. VERTER: Right. There were a number of patients who withdrew from the study at various times.

DR. JOHNSTON: Correct. What were those numbers?

DR. VERTER: We have a -- please, show this slide. This slide describes the status of the subjects in the 99-01 Study through 24 months. For example, at one year, 73 percent of the TAG and 55 percent of the total subjects had a follow-up visit, 20 and 23 percent, respectively, had died, 6 and 7 percent, respectively, withdrew, and 1 to 14 percent missed the visit.

DR. JOHNSTON: So looking at the line of lost to follow-up and we're only looking at TAG, is that a cumulative number 3 plus 6 plus 9?

DR. VERTER: No, no. At one year, it was 6 percent.

DR. JOHNSTON: I'm sorry. And so at two years 10 percent. Thank you.

DR. LINDENFELD: Could I just follow-up on that? Do you know the mortality for all the subjects at one year or are there some for whom you don't know mortality at one year?

DR. VERTER: The best estimate we have would be the Kaplan-Meier curve, which, of course, would censor those patients at their last known visit. So subject to withdrawing at three months, we would not know the mortality.

DR. LINDENFELD: Not. So we do not know. Okay. Maybe one of the presenters can clarify this for me. It was a very nice presentation. Thank you. The concurrent controls look very similar to the TAG patients. But some were chosen for surgery and I just would like to get a little bit better sense of -- can you give me a little bit better sense about, you know, the patients wanted this device, the surgeons are very enthusiastic about it. That means there have to be some differences between the TAG patients and the surgical patients.

Can one of you give me a little bit of insight into what those differences were? How you chose those? I know some was the size of the neck and maybe you can comment on the fact of how having an inadequate neck for this device impacts complication rate. But just give me a sense of how the patients were put into these two categories.

MR. NILSON: Dr. Mitchell can give you a clinical perspective on that.

DR. MITCHELL: Thank you for that question. Patients came to surgical procedures by various routes. The primary route was some anatomical constraints not related necessarily to the aneurysm size, but inadequate length of aneurysm necks, although they still had to be able to have a clamp applied or inadequate access vessels, that's predominant, that's a characteristic that doesn't have a cohort in the open repair.

Additionally, some patients didn't elect to have the procedure with endovascular repair because of the necessity for long-term follow-up and they might very well be living at very remote places from study sites and didn't want to return. So there were probably three incidents, three criteria that put people into the or out of a surgical group.

DR. LINDENFELD: Well, then can you comment for me the lack of an inadequate proximal neck, at least means that you are closer to the -- to some central vessels. And does that impact on the outcome of these patients, do you think? Is that a difference in these patients?

DR. MITCHELL: One of the criteria was still that they had to be an adequate neck to have a clamp safely placed by the judgment of the surgeon. So I think as long as you can clamp that neck, you have -- I think you still have comparable patients.

DR. LINDENFELD: Okay. And then in terms of the access, you're talking about peripheral vascular disease.

DR. MITCHELL: Correct.

DR. LINDENFELD: And extensive peripheral vascular disease not allowing access, do you think that makes a difference in these groups of patients and their outcomes?

DR. MITCHELL: I think the surgical control group could have the same distal vascular disease, but it didn't exclude them from the operating procedure. So it's a requirement only for the TAG group. I don't think it makes the groups incomparable.

DR. LINDENFELD: Okay. I don't know if we'll go back this way. Tell me about also thrombus is not allowed in the aneurysm for the TAG group? Does that impact on your difference in mortality between the two groups?

DR. MITCHELL: Probably all these necks have some thrombus, but we thought that there couldn't be a good seal if there was extensive thrombus in the proximal neck, but probably that same extensive thrombus would preclude clamping. So I suspect that had an equal effect on both groups.

DR. LINDENFELD: Do we know that? I mean, I'm concerned, because doesn't thrombus predict enlargement of the aneurysm? At least in some literature it does. You know, there are some fairly major differences here that it's difficult to see in this list, but exist and, you know, just to get back to whether or not these were fairly comparable groups.

DR. MITCHELL: I'm not aware of thrombus in the aneurysm neck being a risk factor to enlargement or rupture.

DR. LINDENFELD: Okay. And then I guess what I would like to see maybe, Bill, let me know if we'll come back to this later, but can you see a list of why the surgical, at least the concurrent controls chose to have surgery or why they were eliminated from the TAG group? Can we just get a sense of that? I mean, I would like to know if it was for anatomical differences, for choice?

DR. NORMAND: Yes, I would like to know what percent were excluded from the TAG that were in the concurrent.

DR. LINDENFELD: Right. Exactly.

DR. NORMAND: I mean, that's a really important number, in my mind.

DR. KRUCOFF: So while they are looking, can I just ask along this line one other just point of clarification? If you approached a patient who was in every other way a reasonable patient for the TAG device and the patient declined to participate in the study, were they considered and/or consented as a controlled patient?

MR. NILSON: If a patient was a candidate for the test arm, but chose not to be in the test arm?

DR. KRUCOFF: Right.

MR. NILSON: He could be enrolled in certain control arm.

DR. KRUCOFF: Okay. And so hopefully if we can get what John is asking for, there would be some population in your control arm who actually are anatomically truly comparable and just didn't want to participate in the treatment arm? Any idea how many such patients?

MR. NILSON: We don't have that information. With regard to the subject screening, could you show the slide, please? This is the information we have for the pivotal study. This includes both arms of the study and this shows that 28 percent or 90 subjects were rejected for either anatomical, including and excluding violations or insufficient screening documentations. We do not have a subset analysis of this information.

DR. NORMAND: Because it seems to me as you're saying, if you were to randomize, I guess, a certain percentage of your control group, it wouldn't be part of that study. And I just want to figure out, because in theory it doesn't make any sense to estimate an effect if you're not comparable. It would be really important to know that number.

ACTING CHAIR MAISEL: Dr. Edmunds?

DR. EDMUNDS: I would like to know, because I couldn't find it in the handout material, were the exclusion criteria applied to the study patients' neck by your cobbled, serial backward regression control group.

MR. NILSON: The inclusion and exclusion criteria were identical for both of the groups in question.

DR. EDMUNDS: So that none of the patients had an emergency surgery or had a stroke or heart attack within six weeks of the procedure in the control group.

MR. NILSON: Prior to six weeks that was the exclusion criteria.

DR. EDMUNDS: Well, the exclusion criteria was within six weeks, not prior.

MR. NILSON: Within six weeks to enrollment into the study.

DR. EDMUNDS: Well, I mean, someone would be pretty insane to operate on somebody within six weeks.

MR. NILSON: We agree.

DR. EDMUNDS: The second thing I would like to ask is, don't extrapolate, do you have any histologic data of this device in a patient who died for whatever reason?

MR. NILSON: The sponsor has received several explants throughout the course of the device. We have evaluated those explants for a number of attributes including histological evaluation and we do have histological results from patients.

DR. EDMUNDS: Do you have it for us today?

MR. NILSON: We do not have physical histological slides for you to view today.

DR. EDMUNDS: All right. The third thing and I hope -- I think Scott's presentation was cut a little bit, but I haven't seen any presentation of the deployment of the device.

MR. NILSON: During the device part of my presentation, I showed an animation which was a cartoon, for a lack of better words, to describe the deployment and then I actually showed an actual deployment that may have been difficult to see which is why we have the cartoon animation, because some of the radiographic images do not project very well.

DR. EDMUNDS: You don't have a video?

MR. NILSON: I will show you again. Would you like me to show you the cartoon or the actual video?

DR. EDMUNDS: Well, that's up to the Chairman.

ACTING CHAIR MAISEL: We already viewed the video. If you would like to view it again, why don't you pull that up and meanwhile, Mitch, why don't you ask your question?

DR. KRUCOFF: I'm sorry. Because you all have actually two control groups, I just want to make sure that I'm not getting confused. My understanding is there was one control group that was concomitant in time who were simply by and large not anatomically fit for the device, although as you've said if the patient chose not to have the device, they might end up in that group.

My understanding is the other control group was taken from participating centers, registries or whatever of previously operated patients, marched backward in time. Do you have a breakdown of that control group as to how many, in fact, might or would have been TAG candidates had the device been available or the study been enrolling, i.e., the comparability of that population to the actual TAG implant population?

MR. NILSON: Baseline morphology was collected on both groups and Dr. Makaroun showed that they were very comparable in specific attributes like aneurysm diameter, which is a predictor of risk rupture. But I have the aneurysm morphology data for 99-01.

DR. KRUCOFF: Okay. I don't want to drag this in the wrong direction. What I'm trying to ask is simply at the end of the day out of your backward in time control group, the patients identified within the participating sites before enrollment in this study had begun, how many of those patients in that control group would actually have been candidates in all features for enrollment if the trial had been running, at that time?

MR. NILSON: Dr. Makaroun?

DR. MAKAROUN: That's the same question.

DR. EDMUNDS: That's the same question I asked.

DR. KRUCOFF: I thought so too, but I just wanted to make sure we're talking about the same enrollment.

DR. MAKAROUN: Unfortunately, this particular data set is not very complete, because of imaging of the patients that were involved historically, it was, obviously, not obtained prospectively and not all of it is available to analyze that those particular patients could have been or would not have been candidates for the TAG device.

ACTING CHAIR MAISEL: Thank you. Why don't you show the video one more time, so Dr. Edmunds can see that?

MR. NILSON: Could we show the live video not the animation?

DR. EDMUNDS: If you all have seen it, it's fine.

ACTING CHAIR MAISEL: Well, I would like you to see it as well.

DR. NORMAND: We want you to see it.

MR. NILSON: So the device is constrained on the delivery catheter in the upper portion of this video. This device -- this video will loop in sequence. You can see the device constraining there to being deployed. This is real time, the actual deployment takes fractions of a second. Again, it's constrained, deployed.

ACTING CHAIR MAISEL: Thank you. I think at this point, I would like to take a break. We'll take a 10 minute break and regroup at 11:35.

(Whereupon, at 11:25 a.m. a recess until 11:37 a.m.)

ACTING CHAIR MAISEL: At this point, I would like to invite the FDA to give their presentation and remind the Panel that it will have ample opportunity this afternoon to ask questions in more depth.

MS. ABEL: All right. Thank you for the opportunity to present the FDA perspective on this application that you are discussing today. I am Dorothy Abel. I am the primary lead reviewer on this application and, as with every PMA, we do have a review team most of which are listed here, and you will see that Matthew Krueger was my co-team leader.

And I wanted to mention that a lot of the folks that are on this team were also on the AAA Endovascular Graft Teams that presented previously at this Panel, and many of the folks are here in the room, in case you have any specific questions with respect to their areas of expertise.

The proposed indication you have already seen is for treatment of endovascular repair of aneurysms of the descending thoracic aorta and this is the first endovascular graft for treatment of thoracic aneurysms considered for marketing approval by the FDA. The unique aspect of this PMA is the fact that we do have the two different device designs, and so I will spend some time discussing the modifications in the evaluation before I talk about the FDA review summary.

You have already heard about the clinical studies on this device. I did want to remind you that the evaluation of the original device design included the feasibility and the pivotal study with enrollment between February of '98 and May of '01.

After enrollment was complete in the 99-01 Study, fractures were observed in the longitudinal spines, as you have already heard about, and those spines were intended to provide longitudinal stiffness during the deployment of the device. There were also a few fractures in the supporting wire frame in the region where the wires were not bonded to the underlying graft material, so in this area right here those stent portions were able to move more and there were some that had fractures.

This table is included in your Panel package. I just wanted to point out that there were four patients who had both spine and apex fractures, one patient who had an apex fracture and the rest, the 39, were spine fractures and these are worldwide numbers.

The clinical sequelae associated with these spines were five cases worldwide of endoleak and one case of an enlarging aneurysm OUS. Despite these fractures, the clinical results of the original design in the pivotal study were favorable as compared to the surgical control. And as you have seen, there is a statistical improvement in the proportion of patients experiencing greater than one major adverse event through the one-year post-treatment.

The sponsor determined anyway that it was appropriate to redesign the device to minimize the potential for wire fractures, and they made the modifications with the intention of maintaining the clinical performance. There was no change in the fundamental design of the implant. Both versions were constructed of an expanded ePTFE tube, reinforced with ePTFE/FEP film with an external nitinol wire supporting structure bonded to the graft material with an external ePTFE/FEP bonding tape.

The differences between the designs, the original design had the longitudinal spines, which the company is referring to as deployment wires, and there was the unbonded portion of the wire frame that I mentioned, and that unbonded portion was intended to accommodate the spines. The modified design does not have the longitudinal spines and the wire frame is bonded in a uniform manner to the graft material.

There were graft material modifications to strengthen, to provide the longitudinal stiffness previously provided by the spine. The graft material strengthening was accomplished by replacing several layers of the original reinforcing film with layers of an additional stronger, less permeable ePTFE/FEP film.

There were no new materials incorporated into the device. The result was an axially stiffer and less permeable graft material. The same material as was incorporated into the commercially available EXCLUDER Bifurcated Endoprosthesis that's intended to treat AAA devices was done.

For all device modifications, whether implemented before or after marketing of the device, FDA considers the potential impact of the changes on device function when identifying the testing needed to verify that the changes have not adversely affected device performance. So in other words, we look at a risk assessment to determine what additional information is necessary. For the GORE TAG Thoracic EXCLUDER Endoprosthesis, this consisted of the preclinical testing and the confirmatory clinical data.

The mechanical and preclinical in vivo testing that we agreed or required from the risk assessment addressed the potential for changes in this list of attributes, including deployment accuracy, conformity to the vessel wall, migration resistance, durability, etcetera. The evaluation included a comparison to the original device design, and the modified device performed as well or better than the original device, including long-term implant durability testing. We also agreed that a clinical evaluation would be appropriate to confirm the favorable results of the preclinical testing.

I wanted to emphasize that although I have spent a little bit of time talking about modifications, that the primary data set for this PMA is the clinical data for the original device design out to one year. There is also five-year data for the feasibility study from the original device design and 24 month data provided for the pivotal study. Evidence to support approval of the current device designs includes preclinical testing on the modified device, as compared to the original device, and the confirmatory clinical data.

I will now speak to the review summary, although I will not be covering the clinical review as that will be discussed by Dr. Farb and Mr. Kamer.

The draft summary of safety and effectiveness data in the Panel package includes summaries of the preclinical test data provided in the PMA. A review of the biocompatibility, in vivo animal studies, manufacturing and sterilization information, including packaging and shelf life, have been completed and there are no outstanding issues regarding these parts of the PMA.

I want to talk a little bit more about the bench testing aspect of our review. A complete battery of testing results was provided for the modified design of the device and, as the sponsor mentioned, the testing platform was based on the ISO Standard for Endovascular Prosthesis. In addition, there was testing to further evaluate the performance of the device under conditions simulated in the clinical environment.

In all preclinical testing, the modified device performed as well or better than the original device, including long-term implant durability testing. And I just wanted to note that all of the preclinical testing was repeated for the final device design, so that we aren't counting on any of the testing of the original design for our consideration of the safety and effectiveness of the device.

Bench testing review observations. The testing was comprehensive and the results acceptable. There was some clarification requested and provided on the corrosion properties of the metallic components of the implant, but there are no outstanding concerns regarding the bench testing for this device.

You may be wondering why we don't have outstanding concerns with respect to device integrity, because the second study that was conducted is a 30 day study, which was not designed to evaluate device integrity. The thoracic endovascular grafts are subject to conditions that may result in the loss of device integrity, such as structural failures, as you saw with the original studies.

Depending on the location and type of the breach of integrity, there may or may not be an immediate or eventual clinical consequence. This was demonstrated by the information from the original design of the GORE TAG Thoracic Endoprosthesis. The fractures were associated with a relatively low rate of clinical sequelae, although there was a high rate of structural failures.

The results of the clinical study for the original design were favorable as compared to the surgical control. The parts of the original device that were prone to breaking, that is the longitudinal spines, were removed in a redesign of the product.

So as far as our assessment of the integrity for this device, the implant durability testing showed that the modified device was superior to the original design. All of the parameters measured were comparable or improved for the modified device. There is a risk of wire fractures in the modified device, though none have been observed in the clinical use of this device within the limited duration of follow-up.

Despite this, the clinical results for the original design of the device demonstrate that fractures are rarely associated with clinical sequelae, and that is why we believe that adequate information has been provided to assess safety and effectiveness with respect to the structural integrity of the device.

Additional considerations we would like to present for the Panel are the training for this product. The proposed training program is predicated on the training program utilized for the GORE EXCLUDER Bifurcated Endoprosthesis for treatment of AAA and the European release of the TAG device. The program includes a tiered approach based on prior endovascular experience.

The most intensive training will be provided to clinicians with experience using AAA endovascular grafts, but not thoracic endovascular grafts. The training program includes a Gore-sponsored training course, additional TAG case viewing and Gore-supervised training cases.

What we would like the Panel to consider is that this is the first thoracic endovascular graft that may be approved in the U.S., and the adequacy of the proposed physician training plan, as described in the Panel pack, should be discussed.

The sponsor already described the post-approval study plan and we just wanted to mention that the study to assess the performance of the device when used to treat other etiologies, such as dissections, transections, penetrating ulcers in addition to aneurysms in patients at high risk of morbidity and mortality associated with surgical repair is planned to begin in the near future, and that would be an aspect of post-market evaluation that we would be interested in making sure it covers the fact that the device can be used in different etiologies.

The sponsor is going to continue to follow the other patients up to five years in accordance with the original IDE protocols, and you have heard that they are also working with us to determine whether an additional 100 patient study would be appropriate.

So the post approval study considerations is that the plan includes a collection of longer term clinical data. The numbers of patients to be followed should be discussed. The plan does not include enrollment of new patients. It didn't before, we just started talking about it, and so the need for real-world data should also be discussed.

In summary, the clinical results were favorable for the original TAG design despite the wire fractures. In addition, the Circulatory System Devices Advisory Panel has recommended approval of endovascular AAA devices with wire breaks. However, the sponsor elected to modify the device to minimize the risk of fractures.

A 30 day confirmatory study was determined to be appropriate, because the risk analysis demonstrated that the modifications should only affect device delivery and not long-term efficacy. The results of preclinical testing were favorable, and the majority of the device-related events occurred within the first 30 days in the pivotal study.

Finally, the confirmatory study results for the modified device design satisfactorily addressed the device deployability and the short-term risk potentially associated with the design changes. With that, I will turn over the podium to Dr. Andrew Farb who will be covering the clinical aspects of our review.

DR. FARB: Thank you, Dorothy. I am Andrew Farb. I am the lead clinical reviewer and, like other speakers, I am going to make reference to the original TAG device, as well as the modified TAG device, the original being the other one above with the longitudinal spine and the spine removed in the modified device.

Here is a summary table of the various clinical studies that have been discussed and are also available in the Panel pack. I'm going to speak briefly about the feasibility study, but spend most of my remarks concentrated on the pivotal and confirmatory studies. Please, understand that some of the data that you are going to hear you have heard already this morning. I'm going to try to call out those data, which we feel are most important.

The feasibility study achieved initial clinical experience with the device. It was a single arm study of 28 patients using the original device with the longitudinal spine, and the purpose of the study was to establish preliminary device safety and to justify and aid in the design of the longer pivotal study. And at this point, this feasibility study provides five-year follow-up data to explore the durability of this treatment.

There were 28 patients that received the device. There were no endoprosthesis deployment failures and no procedural deaths. Through one-year post-treatment, 57 percent had at least one major adverse event, but I would like to emphasize the low event rate of clinically important events such as paraplegia, stroke, renal failure and myocardial infarction.

Looking long-term from the feasibility study, no adverse events were reported during the second and fifth years of the follow-up period and one adverse event was reported for each the third and fourth years in the follow-up period.

And now through a five year, 60 months, follow-up in 11 patients, there have been no aneurysm ruptures, no endoprostheses migrations. Endoprostheses fractures have been observed in nine subjects, endoleaks in six, aneurysm enlargement in five and of that group, two of the patients required intervention, one, a revision and, one, a convergence into an open surgical procedure.

With these data, the pivotal study was designed and implemented. It is a non-blinded, non-randomized control study using the original device design. And as you have already heard, 140 patients were in the GORE TAG group to be compared with a surgical control group consisting itself of two groups, a historical control group working retrospectively from the clinical centers enrolled in the trial and a concurrent surgical control group.

This was a multicenter study that was performed in the United States at seven clinical sites. Subjects were evaluated for adverse events and device-related events that occurred through the hospital discharge and these patients had follow-up visits at 30 days and six months post-treatment and annually thereafter. As you have already heard, these patients will continue to be treated, continue to be followed, for five years.

The GORE TAG subjects had chest X-rays performed at 6, 12 and 24 months with CT scans in these patients performed at 1, 6, 12 and 24 months. There were 140 patients in the GORE TAG group compared to 94 surgical controls.

All baseline demographic and clinical characteristics were similar between treatment groups, except for a higher prevalence of symptomatic aneurysms in the surgical control groups versus the GORE TAG group. Pretreatment aneurysm diameters were similar between the GORE TAG group and the control group.

The primary safety endpoint was the proportion of subjects who experienced greater than one major adverse event through one-year post-treatment, and I would like to just call out the alternative hypothesis at the last bullet, and that is the proportion of subjects who experience at least one major adverse event through one-year post-treatment would be less in the TAG group than in the control group.

Major adverse events were categorized as either major and minor and that has been covered previously, and so I will move on to the actual events. And the safety endpoint was the proportion of subjects who had had any of these following events during the follow-up period.

Looking at results. Safety. The proportion of patients with at least one major adverse event for the GORE TAG group was 42 percent versus 77 percent for the surgical controls, and with these data one can reject the null hypothesis with a p-value of less than 0.001. Looking at a worst case scenario, 10 GORE TAG patients had no 12 month visit.

Assuming that all of these 10 patients experienced at least one major adverse event through one-year post-treatment, the estimated one-year major event incidence increased from 42 percent to 49 percent. With these data, the significance level for the comparison to the surgical control group did remain significant at less than 0.001.

This table emphasizes the important, clinically relevant safety outcomes and for bleeding complications, pulmonary complications, renal, wound and especially neurologic complications, the Gore group had a lower incidence of events. Not unexpectedly, vascular complications were increased in the GORE TAG group versus the control group. Kaplan-Meier estimates showed importantly that out to two years, the freedom from major adverse events was superior in the GORE TAG group versus the surgical control group.

Moving next to mortality. There was no between group differences in all-cause mortality between the device treated patients who had a mortality of 24 percent through two years versus the control group, which had a mortality of 26 percent.

However, the company then chose to look at aneurysm-related mortality defined as death prior to hospital discharge, death within 30 days of the primary procedure or within 30 days of any secondary procedure to treat the original aneurysm or death due to aneurysm rupture. And here, aneurysm-related mortality was lower in the TAG group versus the control group through two years. Further, there were no device-related deaths noted through two years.

The efficacy endpoint was the proportion of subjects treated with the GORE TAG Endoprosthesis who are free from a major device-related event through the 12 month follow-up visit, and this efficacy endpoint was a composite of the proportion of subjects free from the complications listed below including aneurysm enlargement, rupture, deployment failure, branch vessel occlusion and lumen obstruction.

In designing the efficacy endpoints, several considerations were made. FDA and the sponsor agreed to an analysis plan where the device would need to show superior safety since the efficacy of the GORE TAG prosthesis was expected to be less than that of surgical repair with the efficacy of open surgical repair assumed to be 100 percent. A point estimate of 80 percent was judged to be a reasonable efficacy outcome for endovascular treatment in this study.

Looking at the outcome, freedom from a major device-related event for the GORE TAG group was 94 percent. For this 94 percent, eight subjects or 6 percent experienced at least one major device-related event through the 12 month follow-up visit with 75 percent of those events occurring within 30 days. And with those data, one can reject the null hypothesis with a significant p-value.

Once again, during a worst case scenario, 10 patients who had no follow-up visit and assume all those 10 patients experienced a major device-related event, now the estimate of the probability of not having a major device event decreases from 94 percent to 87 percent but, once again, still able to reject the null hypothesis that p equals 0.02.

A little difficult to see, but here are the clinically relevant device-related events in these eight patients, four endoleaks, zero aneurysm ruptures, two treatment-related device events, one unplanned occlusion of a branch vessel, one prosthesis migration and three aneurysm enlargements. If you look to the right, you can see that from 12 months to 24 months, there was only one additional device-related event and that is an aneurysm enlargement seen below right.

Secondary endpoints, as you have heard, were procedural blood loss, length of ICU and hospital stay and the time to return to normal activities. And as you have already heard, there was a decrease in the immediate length of ICU and hospital stay, as well as less blood loss and a quicker return to normal activities in the GORE TAG Endoprosthesis group.

The imaging core laboratory observed 19 patients with prostheses material failures and that is all wire fractures through 24 months. To date, one adverse event has been associated with breaks in the spine wires and that is a Type III endoleak requiring implantation of an additional GORE TAG Endoprosthesis.

As also you have heard, the device has been modified to minimize the risk of fractures through the elimination of the longitudinal spine. Uniform bonding has been added to the stent structure to aid in deployability.

The rationale for the 30 day confirmatory study, that's TAG 03-03, for the modified device design was based on a risk analysis demonstrated that only device delivery and not long-term efficacy would be affected by the modifications. The modified device performed as well or better than the original device in preclinical bench testing. The majority of device-related events occurred within the first 30 days in the pivotal study.

So this led to the confirmatory study, which was a non-blinded, non-randomized prospective single arm study using the modified device design. This was performed in 51 patients and the controls were the surgical controls in the original pivotal study, 99-01. The study was performed at 11 sites in the United States. It had the same inclusion and exclusion criteria, the same screening assessments, core laboratories and study data collection as the pivotal study.

There were 51 patients who were similar in age to the 94 surgical controls, as well as the 140 TAG patients in Study 99-01, the pivotal study, and baseline critical characteristics were generally similar among the various arms of the study, except for a higher incidence of cancer in one of the groups.

Further, the aorta and aneurysm measurements for the confirmatory patients, the 03-03 TAG subjects, did not differ from the pivotal study TAG patients with respect to aortic diameters, proximal and distal to the aneurysm, the aneurysm diameter itself and aneurysm length.

The safety endpoint was the proportion of subjects who experienced at least one major adverse event through 30 days post-treatment, and the important alternative hypothesis was the proportion of subjects who experienced at least one MAE through 30 days post-treatment would be less in the TAG subjects than in the surgical controls.

For the outcome, the proportion of patients with at least one major adverse event in the prosthesis group was 12 percent versus 70 percent for the surgical control group. We can reject the null hypothesis with a p less than 001 and doing a worst case scenario of the two patients who had no 30 day follow-up visit, the MAE incidence increases from 12 percent to 16 percent, which remains statistically significant.

And here are the clinically important outcomes and you can see for bleeding, pulmonary, cardiac, renal, wound and neurologic complications, a lower incidence of the major adverse events in the endoprosthesis group versus the surgical controls. None of the 51 patients died during the first 30 days versus 6 percent in the surgical group, and there have been no aneurysm ruptures reported with the modified TAG device.

For efficacy, this was defined as the proportion of subjects treated with the modified device who were free from a major device-related event through the 30 day follow-up visit and, as you have heard, no subjects in the device group experienced at least one major device-related event through 30 days. That corresponds to a 95 percent confidence interval of .9321.00.

There were two patients, as mentioned, who did not have a follow-up visit and if we take a worst case scenario and assume those patients did have a major device-related event, the confidence interval changes from 0.93 to 0.87 as a lower bound. There were no deployment-related adverse events and six patients had minor endoleaks.

The same secondary endpoints were looked at in the confirmatory study and, once again, the mean length of ICU stay, hospital stay, blood loss and time to return to normal activities were superior in the endoprosthesis group versus the surgical controls. To date, no wire fractures have been observed in the modified device.

Additional clinical data is and will be available, the out-of-U.S. Gore registry of 114 subjects, and there are three sponsor-investigator IDEs, two of which include patients of high surgical risk and one study is a study of patients with thoracic aortic emergencies.

And just to give you an appreciation of the different etiologies of the thoracic diseases that will be available from these studies, you can see that in the Sponsor-Investigator Study I and II and the European Registry, just over 50 percent of these patients will be treated for thoracic aneurysms with a smaller percentage of patients for the various other aortic pathologies listed.

So in conclusion, from the studies presented, all pre-specified safety and effectiveness hypotheses were met. Spine wire fractures in the original device design occurred, but were rarely associated with clinical sequelae. Device modifications, specifically removal of the spine wires, did not compromise device deployment or safety and efficacy through a 30 day endpoint.

There have been no aneurysm ruptures reported for patients treated with either of the TAG devices, and the reported product results are acceptable. With that, I will close and turn this over to Mr. Kamer.

MR. KAMER: Good afternoon. I'm Gary Kamer. I'm the statistician who reviewed this for the FDA, this submission. First I want to look at the statistical considerations. These are general, a very large overview. One has to do with the clinical study design that was non-randomized, and also the effectiveness of the endovascular repair is assessed somewhat differently from that of the surgical control in that the surgical control is assumed pretty much to be at 100 percent.

Secondly, two primary studies were performed, TAG 99-01 and 03-03, both with complimentary objectives showing that the device, the current device works. Another issue was hypothesis formulation. We'll discuss that a little bit more later. And then finally, safety and effectiveness conclusions, what can be said when we're finished with the study.

Now, turning to TAG 99-01, it's a non-blinded, non-randomized control study using the original device design. There were 140 GORE TAG subjects and then there were two groups that you heard already, of open surgical control patients, some historical controls and some concurrent, and there were 17 clinical sites.

Now, turning back to non-randomized studies, although this applies to both the 99-01 and 03-03 studies, I can introduce, I think, at this point the issues of selection bias and comparability considerations. First of all, selection bias may present itself as an observed baseline comparability issue. Physical adjustment techniques do exist for this type of a bias, dealing with this type of covariates.

Secondly, treatment arm imbalances in unobserved or unmeasured baseline variables cannot be adjusted via statistical techniques. So that remains. Thirdly, bias may also express itself in the exclusion of various types of patients or assignment decisions being based on unobserved criteria. Again, adjustment techniques, statistical techniques are not cable of adjusting for those.

Finally, treatment comparison may be improved via covariate adjustment or propensity score analysis. These are two of the statistical techniques that do exist for making adjustments in those things that can be adjusted for.

Now, going back to TAG 99-01, we're looking at the comparability of the treatment groups. The comparison of baseline covariates between TAG subjects and the surgical control groups appear to be reasonably well-matched. Covariate analysis pretty much have held that. And now, I'm going a little bit more into the propensity score analysis, which is one we want to stress in this case.

Propensity score analysis provides a post-randomization via calculation of each patient's probability of having been assigned to the treatment. All observed covariates should be or may be considered in propensity score analysis. It eliminates the issues of over-fitting or at least it mentions the issues of over-fitting of the model as common in covariate adjustment approach.

It emphasizes overall patient condition relevant to outcome. I think it does this more so than the covariate approach. Still, it cannot adjust for unobserved baseline errors. Propensity score analysis results showed the reasonable baseline covariate balance between the original TAG device arm and the surgery arm for the safety comparison. Propensity score analysis is used only for the evaluation of comparability of the treatment groups and not for the adjustment result p-values and estimates. This is speaking now of the analysis that we received.

Finally, model and results are still being evaluated by FDA due to recent receipt of the data. We actually received the raw data for analysis purposes on January 7, so we've had less than a week to look at them, at this point. The safety hypothesis, the null hypothesis was that the proportion of subjects who experienced at least one major event, adverse event through one-year post-treatment was equal to the control subjects and the TAG -- was equal between the control subjects and the TAG subjects.

The alternative is that the proportion of subjects or patients who experienced at least one major event, adverse event through one-year post-treatment was less for the TAG subjects than the controls. The effectiveness hypothesis was a different TAG sort, in that it was just showing simply the proportion of subjects free from any major device-related event through 12 months would be less than .8 or 80 percent. The alternative is that it would be greater. And I think also the sponsor also presented a cognizable approach to that also showing the same thing.

Safety and effectiveness conclusions for TAG 99-01, the first 12 month major adverse event rates for the original TAG device were statistically lower than that for surgery. Secondly, dealing with the effectiveness, the effectiveness rate of the original TAG device was greater than 80 percent. And then finally, what happened, which wasn't anticipated at the time, was the discovery that there were fractures resulted in major redesign of the original TAG device, which lead to the next study TAG 03-03.

Once again, it was a non-blinded and non-randomized study using the modified device designs you saw earlier. 51 TAG subjects, the safety control was going to be the surgical control enrolled in the TAG 99-01 Study. There were 11 clinical sites all in the U.S. for this study. The same inclusion/exclusion criteria, screening assessments, CEC scores, etcetera were used in the TAG 99-01 Study.

Now, dealing with the comparability of recruitment groups in TAG 03-03, the same comparability and selection bias considerations as existed for 99-01. The comparison of baseline covariates between the TAG 03-03 subjects and the surgical controls were pretty much well-matched again. The covariate analysis again showed that the groups were reasonably well-matched. There wasn't anything indicated.

Now, turning again to the propensity score analysis, we have a chart or picture which basically is a box plot and without going into any great detail, what you're looking for on there is we could have done one of any three that they presented analyses. We chose the one that probably looks the worst for the showing similarity between the patients or the populations or the randomization, fairness of the randomization. But this one still is not really bad, because what you're looking for is an overlap of the two boxes that you see. And as you notice, there is substantial overlap in this.

This was for the TAG 03-03 device versus the surgical control from TAG 99-01. And going further with the results, they do show the reasonable balance between a TAG device arm and the surgical control arm. And that's for the safety comparison. Propensity score analysis only used for the evaluation comparability to treatment groups again and not for the adjustment results. In other words, no p-values no confidence intervals were adjusted in the analysis.

Looking now, turning to the safety hypothesis for TAG 03-03, the modified device, the safety null hypothesis was the proportion of subjects who experienced at least one major adverse event through 30 days post-treatment was equal in the control subjects and the TAG subjects. The alternative is that the proportion through experience rarely one major adverse event through 30 days was less in the TAG subjects than in the control patients, an improvement.

Safety outcomes for that study, the proportion of patients with at least one major adverse event per TAG device was 12 percent versus 70 percent for the surgical control group of 30 days. Therefore, the null hypothesis was rejected at the .001 level. Secondly, you have 2 or 4 percent of the TAG subjects who had no 30 day follow-up, that's been mentioned before. Now, assuming that both of these patients experienced the major adverse through 30 days post-treatment, the estimated 30 day adverse, major adverse event incidence increases from 12 percent to 16 percent. Again, compared to the control, that is still significant. So it does not change the result.

Effectiveness endpoint, TAG 03-03, the proportion of subjects treated with the TAG modified device design who were free from a major device-related event, two 30 day post-follow-up visit. On the day it was presented descriptively both by the sponsor and by us, primarily because the original hypothesis was not appropriately stated to be tested and then used as a claim. No subjects experienced even one major -- greater than one -- even one major adverse-related event to 30 day follow-up visit for the device.

They gave a 95 percent confidence interval from 93 percent to 100 percent. But you did have, as we mentioned earlier, two subjects who had no 30 day follow-up. Assuming that both patients experienced a major device-related event, then the 95 percent confidence interval for the proportion of patients free from a device event is 87 percent to approximately 100 percent. So 87 percent is the lower confidence interval.

Looking further at effectiveness for that, for the TAG 03-03 Study, 2 out of 51 patients were not evaluated 30 days and what I'm going to do is show the -- what would happen if one or two of these patients had now been evaluated or had experienced at least one major adverse, a device-related adverse event. And that's just an extension of what you had before. As you see, at 2, the 86 or 87 percent to almost 100 and zero which was observed, you realize that the adverse event-free rate would be no lower than 93 percent based on the confidence interval.

So you can look at that. And, basically, what happens is you are dropping about 3 percent every time you add one other patient on. That's not only to show that these were missing, it's also to show somewhat sensitivity or volatility for so few patients. Volatility results. The limitations of effectiveness analysis for TAG 03-03, the 03-03 hypothesis was not formulated so as to statistically establish effectiveness in the modified TAG device, so nothing can be a first statistic concerning the effectiveness of the modified TAG device relevant to the original device.

The modified device effectiveness, therefore, is limited to the point estimate and confidence interval for that device. All right. Our closing, looking at the statistic prospective on this overall. We have non-randomization in TAG in both studies and it was appropriately addressed in the submission and the subsequent analysis. Safety hypothesis of both devices were met. Effectiveness assessment of TAG 99-01 was addressed in that also. The effectiveness assessment of the TAG 03-03, the modified device, is limited by the absence of an appropriately defined hypothesis comparing the two TAG treatment groups. Thank you.

ACTING CHAIR MAISEL: Thank you very much. At this point, I'll open the discussion to the Panel Members to ask questions of the FDA and I'll take the liberty of asking the first question for Dorothy or whoever wants to answer it.

I had a question about the structural integrity of the device, the modified device, in particular, and TAG 99-01 could be the poster child for why we need clinical studies and why preclinical testing is not in and of itself sufficient. They did all the appropriate testing, all the required testing, yet it wasn't until after the device was implanted that we detected these fractures.

While they developed a preclinical test that was able to identify the problem, there is now a new device which I would consider substantially modified from the original, and I'm concerned that preclinical testing might not identify all the potential problems with that device. And so I wonder if you just might address the guidelines for testing of endovascular stents, how they came about and how confident you are that all the long-term structural abnormalities can be identified by the preclinical testing?

MS. ABEL: I think, first of all, there are different reasons for doing preclinical testing. And certainly, one is to try to predict a longer term clinical performance. But also it's to compare device designs as we have here. Initially, when these parts were developed, both the thoracic and the AAA endovascular grafts, there were no standards or guidances that were applicable specifically to these devices.

So the stent guidances and the vascular graft guidances were used by the manufacturers who did their analysis to try to determine the appropriate testing. Each individual manufacturer came up with their own testing strategy, they conducted their test, they justified their results, and that's really all we had to go on. Since then, we have had quite a bit of advancement in the area of evaluation of endovascular grafts.

First of all, we have written a standard, the ISO standard for endovascular prosthesis and that covers both AAA and thoracic endovascular grafts as well as some other implant locations, and that standard is based no a risk assessment. So you look at what sort of characteristics does the product need in order to be able to perform appropriately clinically? And so let's say you have to be able to deliver the implant. And so what testing do you need to do in order to determine whether or not you can actually do that?

So I think we have come a long way with respect to the testing in general. I think the company has also come a long way even beyond the standard, because after they saw the failures, they were able to further advance the testing to incorporate what they had learned about the clinical environment. And that's one of the limitations we've had from the beginning with these devices is trying to figure out even if you know the type of testing that's applicable, how do you determine the parameters of test? What sort of stresses are these devices subject to in the in vivo environment?

And what the sponsor did was try to incorporate those sorts of things with some additional testing and also within their existing testing so that they do have more realistic testing. I don't think we can ever be certain that a device is not going to have failures, even if we looked at it in the preclinical setting. But what we can look at is we know the benchmark for the current design. We've modified that design and so now you are comparing to the current design, and in this case it was as good or better in all the aspects. So we're talking about specifically durability.

The other thing is that there certainly is the potential for an occasional wire fracture and we've seen that with the AAA devices, but also what we've seen is that there has not been clinical sequelae associated with the majority of those fractures. So we can't guarantee that there won't be any fractures with this device, but we feel that even if there are fractures, the potential of having clinical sequelae associated with those are fairly low. Does that help?

ACTING CHAIR MAISEL: Yes, it does very much. Thank you. Judah?

DR. WEINBERGER: This is a question regarding the composite endpoint but consists of about three dozen different adverse events. We've been educated in the coronary world to accept composite endpoints like major adverse cardiac events which are hard composite endpoints or fairly hard anyway. When I look at the list of composite endpoints that were thrown in to define a major adverse event, there are things in here that are rather soft and that are certainly dependent on other adverse events.

So for instance, angina, many of these patients had angina pre-op. They were Class II or Class III angina pre-op and pre-procedure. They then went on to have a procedure and certainly in the operative group would be expected to have more blood loss. And seeing more angina post-op is sort of a secondary effective that's more or less expected. So I'm troubled a little bit by seeing composite endpoints that are this large and this complex and you can't tease out the objective components that are really independent in trying to assess is this truly-- do we see a safety benefit?

So that if I could pull out prospectively, ideally, those hard endpoints that are clinically relevant, make sure the patients don't go home with stroke, paraplegia, a new change in ejection fraction or, you know, new QAs, I mean, really hard endpoints that would help a lot. What I see when I see the large composite endpoint like this is you can shove a lot of stuff under the rug with soft endpoints, especially since you are very well-aware of what the treatments of the patients have been exposed to are.

MS. ABEL: I'll let Dr. Farb answer that question. Though I just want to emphasize that we did present the results for those various endpoints separately also. And the reason that we do end up with these composite endpoints is so that we can design a study that's a reasonable size that can be conducted in a reasonable time frame.

DR. WEINBERGER: But we're giving equal weight to endpoints of varying degrees of hardness. So death counts the same thing as an angina episode or a bump in creatinine or a transfusion.

MS. ABEL: Well, that's why we have the differentiation between the major adverse events and the minor adverse events in an attempt to address that to some extent.

DR. FARB: I think your point is very well-taken in presentation and try to call out those, the sort of the meat of those endpoints that we are really interested in, and that is things like paraplegia, respiratory failure, certainly death, myocardial infarction and there are some statistical implications of doing multiple testing and I can defer to my statistical colleagues for that.

In their data analysis, while we don't have significant levels that are valid for each individual important endpoints that we're both sort of concentrating on, there are confidence intervals presented. And when the confidence intervals do not include zero, then we have some basis for saying that we think that there is a meaningful difference between the two groups.

DR. LINDENFELD: If I could just add to that comment? I think the concern about soft endpoints is not just soft endpoints. But the surgical patients were monitored for soft endpoints for a much longer period of time. So if we consider reversible angina without an MI or a lot of non-sustained V-tach, the surgical group is monitored for three days in the ICU and we're going to see every one of those, whereas we're going to see none of them or very few of them after the first day in the device group.

And I think that's another really important part of those kinds of soft endpoints. The observation period is enormously different between these two groups and I'm concerned about that as -- we'll come back later to the definition of some of these.

ACTING CHAIR MAISEL: Dr. Krucoff?

DR. KRUCOFF: Are you going to sit down, Andrew? Doctor, I'll let you pick the responder. I have, actually, these are somewhat statistically or random questions, although the interpretations may -- you might be the right person at the mike.

Gary in his presentation said several times that there is no statistical method that can address for unobserved baseline variables. And it's impressive to me that we have a control group, some of whom are anatomically individuals who might have been appropriate for the TAG device, but were done before the study started or declined or whatever and others who were not. Now, your conclusion in a propensity analysis is that these populations are comparable.

So is your real conclusion that these anatomic variations are meaningless?

DR. FARB: I don't know if we know that answer. There are a few anatomical differences that were present between the two groups. And they did do an analysis between the two control groups showing that the covariates were similar between those groups, independent of the comparison between the control and the TAG endoprosthesis. Propensity score, again, I'll, you know, defer to Gary, does help us in being able to conclude that the two groups were reasonably comparable in terms of the observed covariates.

DR. NORMAND: If I can interject?

DR. KRUCOFF: Okay. The observed --

DR. NORMAND: Weren't they missing more for the control group, some of that information, that would permit that comparison of what Mitch is asking?

DR. FARB: Right.

DR. NORMAND: So it was differentially missing in the control group?

DR. FARB: Yes, I think that's the limitation of using the retrospective surgical controls as alluded to that you don't have the opportunity to have the imaging, for example, in all those patients. That's right. That is the limitation.

DR. EDMUNDS: Do you really have enough numbers for propensity, Mitch?

DR. FARB: Gary?

MR. KAMER: I didn't hear the question.

DR. FARB: The question is do you have adequate numbers to do propensity or not?

MR. KAMER: Yes, I believe we do. It is getting a little small when you get down to 51 patients, with missing data and everything, but you can. I think it is possible to do. You end up doing probably instead doing deciles, you're going to do tertiles, you know, just three levels for matching, which makes it a little broad. Then adjustment may be an issue also. But it's much harder with 51 patients than it is when you're looking at the original 99-01, which has a larger number of patients.

DR. NORMAND: But you --

DR. KRUCOFF: So, Gary, did I miss it? Are you -- the covariates that I saw are almost exclusively demographic and/or clinical. Do you have anatomic characteristics in there somewhere?

MR. KAMER: I think the sponsor can -- we've had about five days to look at it and actually nobody in this group has looked at it extensively on the team right here.

DR. KRUCOFF: My understanding from --

MR. KAMER: But let me ask.

DR. KRUCOFF: -- the related question earlier, from a clarification point of view, is that these data do not exist.

ACTING CHAIR MAISEL: So why don't we ask this?

MR. KAMER: Okay. Then it would not be included in the model that doesn't exist, if they don't exist.

DR. KRUCOFF: Yes.

MR. KAMER: That would be the situation.

DR. KRUCOFF: Okay. Because the implicit conclusion that these are comparable populations only reaches to demographics and clinical studies, right?

MR. KAMER: Well, it only reaches those observed or recorded covariates, right.

DR. KRUCOFF: All right. So while I've got you, let me ask one last point of clarification that's probably fully unfair, because you haven't had these data long. But we were shown a slide in the sponsor's presentation that is a comparison of freedom from MAEs through 30 days.

MR. KAMER: Yes.

DR. KRUCOFF: It's plotted as a survival curve.

MR. KAMER: Yes.

DR. KRUCOFF: That includes the control population, the pivotal study and the confirmatory study.

MR. KAMER: Right.

DR. KRUCOFF: Okay. So my understanding of the pivotal study and the confirmatory study is they used the same enrollment criteria and they gathered the same data.

MR. KAMER: That's what I understand.

DR. KRUCOFF: Okay.

MR. KAMER: Yes.

DR. KRUCOFF: And, at least as I look at this graft, it looks to me like the confirmatory study population behaves significantly differently or at least differently.

MR. KAMER: Yes.

DR. KRUCOFF: Enough to raise a question of are we capturing or are we missing something here that's actually even capable of defining different outcomes in the populations treated with the device?

MR. KAMER: Yes, there may be. Again, we're dealing very -- we're at the very beginning of analyzing several different components, one of which would be, of course, the treatment patients in 99-01 and 03-03. There are some indications that that may be a more diverse comparison than some of the other ones we have looked at to this point. But is it diverse enough to cause a problem, to cause the difference? I couldn't say at this time. But there does appear to be some more difference between that than you would find say between the original, the 99-01 control and surgical groups.

DR. KRUCOFF: Okay.

MR. KAMER: Control surgical and treatment groups, there seems to be a little more difference in those.

DR. KRUCOFF: I guess what I'm trying to get at, Gary, is do you think that we are characterizing, even with propensity scores, similarities between these populations that are totally independent of whether or not or how the device behaves?

MR. KAMER: Similar?

DR. KRUCOFF: Well, we're calling these populations the same.

MR. KAMER: Right.

DR. KRUCOFF: Based on demographic and clinical descriptors that you have data for to model propensity profiles.

MR. KAMER: Yes.

DR. KRUCOFF: That may, in fact, be leaving out what actually makes these populations different or not.

MR. KAMER: Well, that's the argument for randomization all the time. When you randomize, you balance, at least in theory you balance, and hopefully you do.

DR. KRUCOFF: I'm just asking whether you think you have enough data in these data?

MR. KAMER: Observed and unobserved, and I agree with that. So I think it could be. Yes, of course, they could be left out, yes.

DR. KRUCOFF: Do you think you have enough data in these data sets to call these populations comparable for the purposes of this assessment?

MR. KAMER: I've heard it said before that the -- by a clinical trial expert who had a lot more experience than I had, it is actually designed to run, that you only account for about 25 percent when you do go back and adjust these covariate adjustment procedures or apparently propensity score analysis type procedures. So you can't account for all of them. You just start gathering, maybe it's a grant in a person's eye that makes one person be assigned a certain way or it's just really the only way to do it to feel comfortable that you have done it is randomization.

But these methods try to bring a non-randomized study back towards a randomized study with what information you have. Are they perfect? No.

DR. NORMAND: If I could interject for a second? It sounds to me, I mean, we already know already you can't adjust for what you didn't collect.

MR. KAMER: Right.

DR. NORMAND: And so we're concerned about the selection bias based on the unobservables. But I don't know if this is what you're getting at, but I think a secondary concern, not necessarily in the importance, is that for the data that are collected, you're missing data and you're missing data much more in the control group. So even if you believe they were comparable, which we don't have the data for, I guess my question would be adjusting for what you had knowing that, at least based on my read of this, that you have differential missing-ness for the control group versus the treated group.

Do you feel that you had enough or sufficient information, because now there's a selection bia on missing-ness, to look at the comparability of even the observed people?

MR. KAMER: Again, I think that's always a trouble within the study we run into. And quite often we do find that the treatment groups differ as far as missing data, follow-up, etcetera, in general. And the answer, I think, would be it's a problem. Do we have enough? I think it's way too early, after six days of looking at data.

DR. NORMAND: Sure.

MR. KAMER: To say yes or no to that.

ACTING CHAIR MAISEL: Dr. Somberg, did you have a question?

DR. SOMBERG: Yes, I think there's an important thread here that we've had in this question period and I agree with Dr. Weinberger that there are so many different endpoints in these composites, some of which are significant, some of which are less. So, therefore, I move to the mortality. With that said, I would like you to look at page 10 of my briefing book from the FDA's material here, the second paragraph. It goes on to state that even really the mortality endpoint, and I think the TAG 01 pivotal study is what we should focus on, because that's going to be the major -- have the major number of patients, the longest follow-up and really is what is pivotal for the, by its definition, approval of this device.

And it says there is a very major difference between the New York Heart Association and the symptomatology of the aneurysms, but at least half in the control group, half of the data is missing. You can't replace missing data. But my specific question, after all those statements, is was there any attempt made to adjust for the difference in severity Class III versus Class II, which has an impact on outcome for the data we have to try to account for that or has that not been done? Because I didn't see that in the sponsor's presentation and I didn't see it in the follow-up materials from the FDA.

MS. ABEL: I really think that that would be most appropriately addressed by the sponsor. I think there are several of these questions that would be most appropriately addressed by the sponsor. I think they have the clinical expertise to discuss what effect potential differences in anatomy would have on the two treatment groups. And I also think that they know better in terms of what they were using NYHA data for and how much of this information they have been able to obtain and what additional analysis they have done. That we have presented what we have seen.

ACTING CHAIR MAISEL: So we can address those questions to the sponsor after lunch. But just to directly address the FDA. So in your analysis, was New York Heart Association adjusted for in the propensity score analysis?

MR. KAMER: We took a look at that, at those. We noticed there was a large discrepancy in missing data between the two groups, treatment groups. And really a little surprised that that wasn't collected, whereby it's a very important normally covariate. However, upon using the data that were provided, we really found that it was not correlated to outcome as much as we had thought.

DR. NORMAND: Is that complete case then? You did a complete case analysis? How did you do it for those? How did you include it for those that didn't have it measured in the propensity score analysis?

MR. KAMER: Was that imputed? It was imputed. Okay. It was imputed also. I know most of the analysis we have performed have been using imputations.

DR. NORMAND: So you imputed the New York Heart score for those that were missing the data?

MR. KAMER: Yes. Again, this is preliminary, but we were -- but it was not then one of those covariates that would have been related to outcome and therefore was not different. I don't believe it was different. Well, it was somewhat different, but it was not included, I don't believe, in any of the other analysis.

DR. NORMAND: So you just used mean imputation for everybody? I'll stop.

ACTING CHAIR MAISEL: I mean, the bottom line is that there is a lot of data missing from the control group that we can never get back and we can do all kinds of analyses, but we just don't know what we don't know.

DR. LINDENFELD: I just have a specific question maybe for Bram or for some of the FDA representatives. It seems to me that not only are we missing data, but I'm not certain that we have collected the data that indicates the risks to the patients adequately. Just looking back before I came, there was a recent review of 1,100 patients with thoracoabdominal aneurysms and the four biggest risk factors for mortality were creatinine, which we don't have in here, age, which we do, symptoms, which were clearly worse in the surgical group, and the Crawford type of aneurysm, which we don't have.

So those in a multi-variate analysis were the top four. We have one not different, one very different and two we don't have. This would be an opportunity to look back at the SVS database and let them also tell us what are the top risks and do we have enough of that data to be sure that we can compare these groups? I mean, what are their risks as well? But I'm concerned that we're just missing a lot of baseline data which we can't correct for, because, as was said, we can't correct for data that we don't even have.

ACTING CHAIR MAISEL: Do you have any more specific questions for the FDA? Dr. Bridges and then Dr. Edmunds.

DR. BRIDGES: I have a question that also relates to Dr. Weinberger's point about the question about endoleaks. There is a significant incidence of minor endoleaks or minor adverse events that are endoleaks. It was 12 percent in the TAG 03-03 and 14 percent, approximately, in the TAG 99-01. And in terms of major adverse events related to endoleaks, it was only 3 percent. And the definitions of those two are somewhat soft, i.e., a minor endoleak is one that's being followed serially that doesn't require intervention. Whereas, a major endoleak is one that does require intervention.

So, first of all, I would expect that there should be an incidence of minor endoleaks that become converted to major endoleaks at some point, and I'm not sure looking through the data that I see evidence that that happens. And secondly, the issue of whether an intervention is required or not is also, you know, somewhat subject to observer, interobserver differences. Do you have any comment on that? Is that an issue? And what's the fate of all those 14 percent of patients that have minor endoleaks? Have those -- maybe someone could comment on that.

ACTING CHAIR MAISEL: That seems like a more appropriate question for the sponsor. Maybe you can work on getting that. Do you have that information available, at this time, regarding the endoleaks? Why don't you step up and just answer his question, please?

MR. NILSON: The definition of major and minor was determined by the Sacks criteria as you've already alluded. I would like to point out that we have not seen any ruptures in either of the arms. I will bring up Dr. Makaroun to give the clinical perspective on major versus minor endoleaks.

DR. MAKAROUN: Forgive me if I ask you to go back and give me the multiple questions.

DR. BRIDGES: Yes, sorry.

DR. MAKAROUN: If I remember your question correctly, one of them related to whether some minor endoleaks can turn into a major endoleak in the further follow-up.

DR. BRIDGES: Right.

DR. MAKAROUN: Obviously, by the definition that is used, an endoleak that does not -- let's say for two years and is classified as minor and is treated in the third year becomes major, but typically that essential endoleak becomes classified as major. The majority of endoleaks that you saw reported there are through a particular period. And at any time an endoleak was observed even if it went away, it gets reported in that percentage that you are seeing. Actually, the number of endoleaks that are still there and observed is much fewer than the number of endoleaks that are reported in this data. I can probably show you this.

ACTING CHAIR MAISEL: Were there any minor endoleaks that became major?

DR. MAKAROUN: One that retroactively became major. It was -- it was during year three.

ACTING CHAIR MAISEL: Okay. Maybe we can do this a little more in depth after lunch. And, Dr. Edmunds, if you could ask your final question and then we'll take a break.

DR. EDMUNDS: Has the FDA seen any histologic data about this endoprosthesis or any endoprosthesis?

MS. ABEL: We have not seen histological data for this particular endoprosthesis. We have seen it on other devices. We have seen it for the AAA device, from Gore and other AAA devices.

DR. EDMUNDS: How does it heal?

MS. ABEL: I'm really not qualified to answer that question and I don't think you've looked at the -- I would ask that the sponsor be able to address that question. I'm sorry.

ACTING CHAIR MAISEL: At this point, why don't we take a break and we can explore those and other issues after lunch. We will resume at 1:45.

(Whereupon, the hearing was recessed at 12:45 p.m. to reconvene at 1:46 p.m. this same day.)

A-F-T-E-R-N-O-O-N S-E-S-S-I-O-N

1:46 p.m.

ACTING CHAIR MAISEL: If everyone could, please, take their seat, we can begin. Please, be seated, so we can begin the afternoon session. I would like to start with our lead reviewers doing their reviews and questioning the sponsor, and we will begin with Dr. Edmunds. Actually, you can do it from your seat. That's okay.

DR. EDMUNDS: My review of this very thick, heavy, repetitious package can be consolidated down to, first, I don't think there is any material difference made with the modification. I think the modification is an improvement and I see no downside, and I did notice that when you bend it, holding the two ends like the curve of the aortic arch that is not involved in this protocol, there is no compromise of the lumen whatsoever.

I also think that the safety and the efficacy of the device is demonstrated. However, I must say that I think the statistical control group is a joke.

ACTING CHAIR MAISEL: Do you have any questions for the sponsor at this point?

DR. EDMUNDS: Well, I would really like to see some histologic data, but there is none here, so I don't think that needs to be brought up again.

ACTING CHAIR MAISEL: Okay. Thank you. Dr. Yancy?

DR. YANCY: Thank you, Bill. I apologize that I didn't hear the sponsor's presentation. Flight delays weren't avoidable. But nevertheless, I have reviewed the supplied PowerPoint file and gone through the same continued information, as so described.

I think it has already been identified what disease process it is that we are concerned about, and that is a fairly significant expression of peripheral atherosclerotic disease with significant morbidity and mortality. And as a cardiovascular specialist, I certainly have a great amount of respect for this disease, because I recognize how commonly it is companioned with significant coronary disease, thus increasing perioperative risk, so that a percutaneous treatment strategy, one that would minimize the morbidity and mortality of the given procedure would, in fact, be a significant advance.

So one looks at this platform from a clinician's perspective with hopes that the technology is at the point where it can be embraced and move forward in a reasonable way. Reviewing the technology, there are some immediate concerns that come to mind and, in reviewing the entire packet, there are more significant concerns that come to mind. Specifically, I will start with the last comment made.

There is, in fact, reference in our packet of animal data from which there are necropsy findings that simply give us global statements that the host response was normal and that the histology was unremarkable. In my judgment, that is insufficient information and it would be helpful to know what the integrity of the endothelialization is of the device, if that indeed is what occurs, and what the kind of scarring process might happen to be. So there are some qualitative references to that, but I agree that I think it would be helpful to have better information.

As I have gone through and looked at the three different trials starting with the feasibility study, I view that as a pilot study utilizing only 28 patients, a fairly reasonable 30-day mortality rate of 3.6 percent, but I remark that the one-year mortality rate was 21 percent. Now, that's only five patients but, nevertheless, that is of concern.

As well, the major adverse event rate in the feasibility study was 57 percent. The more worrisome events didn't occur, but I make the point of the 57 percent, because the statistical estimates for the pivotal trial were based on a 50 percent reduction of the major adverse event rate, so it would almost seem as if the suggestion is that for the surgical cohort, 100 percent of the patients would end up with a major adverse event.

It was interesting that the questions of lead fracture or wire fracture were identified during the feasibility study, and I find it somewhat curious that the adjustment in the technology was not made at that point, but rather was made after the pivotal study.

I think, as has been suggested earlier, that the pivotal study is the most significant database to look at, because it does, in fact, have a pretence of a comparative population. I say pretence, because I am very concerned that this was a non-randomized experience, and I am even more concerned that the control group or the reference population was quite heterogenous.

Approximately, half of the control group were consecutive retrospective cases that perhaps may have been candidates for the device or not, and then 44 percent were prospectively enrolled cases. But I am even more concerned about the prospectively enrolled cases, because in the text of the information provided by the sponsor, it's indicated that the main reason that those individuals were treated in the surgical cohort is that they failed the screen for the EXCLUDER with the most likely reason for failure being an inappropriate, unacceptable anatomical substrate.

Now, Mitch was getting at this earlier, and I fully support the direction of his questions. It is indicated in the table that the incidence of symptomatic aneurysms was 21 percent in the device group, 38 percent in the surgical group. That is a difference of nearly 100 percent, and I cannot appreciate how we could say the groups are similar when the anatomical substrate, that upon which the surgical methodology is based, appears to be so strikingly different at least in a qualitative way.

What further raises my concern is that not only is the anatomy more likely to be problematic in the surgical group, but the distribution of the NYHA class, even though data points were missing, was clearly disparate with more Class II individuals in the device group and more Class III individuals in the control group.

Thus, if you take those two observations, it seems to me that the surgical group or the control group was actually a higher risk group, and so when one compares an appearance of safety and a reduction of major adverse events with the device against what appears to be a higher risk group, I think that, to a certain extent, disqualifies the statements that based on those comparators, that the group treated with the device was, in fact, treated with a platform that was felt to be safe.

It is admitted that this statistical analysis does, in fact, demonstrate with reasonable power that the relative risk reduction of events was quite impressive, but that was a 14 day, 61 percent relative risk reduction. By two years, it was 37 percent and one wonders about the durability of the device and whether or not the difference between major adverse event rates would be neutralized with longer follow-up.

I am especially concerned looking at the pivotal trial that the issues that I, as a cardiovascular specialist, would be concerned about, specifically mortality in cardiac events, were totally unaffected by the use of what, at face value, appeared to be a lower risk platform. There was no difference in mortality probably related to the sample size, admittedly, and there was no difference in cardiac events.

And given the very high event rate of this group, the sample size should have been enough to detect a difference in cardiac events. And so that, in my judgment, means that we would have to look at other components of the major adverse event category that would have to be quite robust to overcome a lack of difference in those major outcomes.

Admittedly, there was a difference in the incidence of paraplegia, which is a large risk for this kind of surgery, but I don't think it reaches the bar and I would embrace what Dr. Weinberger said, that when you have this kind of surfeit of variables that are in your major adverse event category and you have a composite that is so broad, I think that greatly dilutes the implication of any single component of that composite. And certainly from cardiovascular clinical trials, we're very reluctant to do studies with a large number of composite endpoints, because it can disqualify our study design.

The confirmatory study is also one that I have great concern about, because if it represents the iteration of the platform that has been improved, that is to say that the wire has been removed and there is still longitudinal integrity through other manipulations in the engineering, then we would be going forward to market with an n of 51 observations in an uncontrolled observational experience that could not qualify as a study.

And basing that as our means of saying that this is reasonable, I believe that the reference to a control group in the confirmatory study is inadequate. I say that, because that control group is the historical control group from the pivotal trial, which I already believe is problematic, because it's heterogenous and it is, in my judgment, clinically different from the group upon which there was an intervention.

I note that there is a statement that over 2,000 such devices are commercially deployed in European countries. It may be inappropriate for us to see those data, but I would be curious to know what the market experience has been like in those European applications.

There are statements in the manual about the intent for a post-market analysis plan. I am troubled by the sponsor's design of an apparent registry. The total number of subjects engaged in that registry would be 250, the majority of which are individuals that are already in the stated clinical trials.

I think that one of the areas of great concern is in the perioperative experience, and to miss that opportunity in the registry would be problematic. Clearly, we need long-term data, but I don't think this registry would suffice and that would need to be strongly revisited.

There also are statements that the sponsor has suggested about the training of future implanters, and it strikes me that it is largely a technical plan and doesn't really capture the nature of the illness. And given what I believe are differences in the populations that were identified and cited in these experiences, I believe that that needs to be a part of any educational format.

So if I summarize my observations, these are the things that I would say are concerns for me. The total number of individuals evaluated in the three submitted trials is only 313. Given the sponsor's own data to suggest there are 15,000 new cases annually and another 5,000 procedures done per year, I don't see how we can likely move forward with data on only 51 patients using the most reasonable platform that is available.

I am also concerned that we don't have a good handle on major adverse events, because we have been so global in the inclusion. There are more than three dozen or actually over 40 major adverse events and they are of varying severity, and I don't believe that that's a reasonable way to look at those.

I think that we have to respect the fact that the possibility exists, at least from a clinical perspective, that dissimilar patient populations were studied in the pivotal trial and so, in my judgment, disqualifies the comparisons.

Finally, I believe that the plans to move forward are compromised. We have not yet dealt with one of the major issues that concern me going through the data, and that is that there are inherent risks associated with this procedure, particularly at the vascular level.

There is a statement in the FDA memo tab, page 10, that TAG subjects experience more vascular complications than control subjects, 18 percent versus 6 percent. That is a threefold increase, and the majority of those vascular complications were vascular trauma. And in the sponsor's supplied information, the comparison chart of the size of the aneurysm compared to the size of the introducer sheath to deploy the device suggests sizes as large as 20-24 French.

And for those of us that have been in the invasive suite, it may not intimidate a surgeon, but it certainly intimidates me. Balloon pumps usually go through 9.5 or 10 French. So I am concerned about taking people with overtly diseased blood vessels with a large atherosclerotic burden and applying such technology. So I think that that is a problematic issue. Thus, I have great concerns about the data that are presented to us.

ACTING CHAIR MAISEL: Thank you very much, Dr. Yancy. Did you have any specific questions that you wanted to ask the sponsor? You mentioned some outside of U.S. experience that they had mentioned or other specific questions?

DR. YANCY: I really just wanted to make comments. I don't know that there are any answers to questions that would change the perspective that I have.

ACTING CHAIR MAISEL: Very well. So why don't we go ahead and proceed with additional questions and comments from the Panel Members and we'll start with Dr. Weinberger, please.

DR. WEINBERGER: I only have a couple of questions by way of clarification. We were presented and a couple of slides highlighted groups of outcomes in terms of safety. I think Dr. Makaroun presented a couple of slides where you showed us what the risks were for the TAG versus the surgical groups broken down in categories, respiratory, renal, etcetera, where you had red highlights. Did any of those reach statistical significance?

DR. MAKAROUN: A confidence interval was constructed for every single one of them, except for the one I specifically mentioned did not reach statistical significance, which was the cardiac. All the others highlighted in red did and they were all in favor of the TAG control group, except for the vascular complication, which was in favor of the surgical control group and that also, the 95 percent confidence interval of the risk factor, did not cross zero, so it was significant.

DR. WEINBERGER: And in those patients who underwent surgical conversion at some point during their course, was the implant actually removed? Was the device removed?

DR. MAKAROUN: There was one conversion and that happened in month three. The particular patient was suspected to have an infection and that prosthesis was removed.

DR. WEINBERGER: And getting to what Dr. Edmunds has been asking about, was there any analysis done of that device?

DR. MAKAROUN: I will let the sponsor answer this, but typically after three months we would not expect a whole lot of changes.

DR. WEINBERGER: It's too early?

DR. MAKAROUN: Excuse me?

DR. WEINBERGER: It's too early in the course?

DR. MAKAROUN: Too early.

MR. NILSON: All devices that were explanted were not returned to Gore. In fact, we only received three U.S. devices that have been explanted over the course of all the pivotals and confirmatory and feasibility studies.

DR. WEINBERGER: Okay. And then I hate to harp on this control group issue, but it seems to me that there is a very distinct imbalance not just in terms of anatomy, but also in terms of the clinical setting. There were a much larger number of patients who are having symptoms, unstable symptoms, in the control group of impending aneurysm rupture. Is that the case?

MR. NILSON: Dr. Makaroun will address symptomatic versus non-symptomatic aneurysms.

DR. MAKAROUN: We actually feel that the control group was very well-balanced. In terms of the symptom, which is the only variable that was statistically significant between the two groups, I would like to preface by saying that all patients with ruptures were excluded from this trial and all patients with mycotic aneurysms were excluded from this trial.

The symptoms are not strictly pain symptoms. Some of them were pain, but some of them were also related to local pressure on phrenic nerve or on the trachea or on the esophagus, so some of them did not carry any of the characteristics that you may associate with the higher risk type of category that we discussed. Can you, please, show us the slide?

A subgroup analysis was actually performed for the symptomatic aneurysms, and on the left hand side you can see the classification between symptomatic and non-symptomatic aneurysm with the major adverse event rate through one year listed for both the TAG and the surgical controls, and you can see that the therapeutic benefit of the TAG over the surgical control is actually evident for the symptomatic aneurysm alone and for the asymptomatic aneurysms alone.

It's also of note that the major adverse event rate for symptomatic versus non-symptomatic is essentially the same whether the patient was treated by the TAG or by the surgical control. In addition, a Cox Regression Analysis Model that was performed did not show that symptomatic aneurysm was predictive of any major adverse events.

DR. WEINBERGER: That's it.

ACTING CHAIR MAISEL: Thank you. Dr. Johnston?

DR. JOHNSTON: I recognize the difficulty of developing a clinically safe and effective treatment for this very complex and important problem. I would like to ignore the comparative data for a moment and go to the complications, and I would like to start with the minor complications.

I understand the definition of minor complications, but I am not sure how fair some of the definitions might be and I can cite the examples if you want, Table 62, defining renal insufficiency as minor. Most clinicians, no matter how that occurred, would not feel that was minor.

Prosthesis thrombosis in Table 21 does not strike me as a minor complication. Late nerve injury does not strike me as minor. Table 27 listing prosthesis migration, prosthesis failure, is that minor? And one of the late endoleaks. And then further, there was one late case of paraplegia listed as minor, late.

I wonder if you can address the reasons why these complications might be listed as minor complications, because I am going to then come back to the major complications in trying to understand the true impact of this prosthesis.

MR. NILSON: Can you show the major/minor definition from the main presentation?

DR. JOHNSTON: I understand the major/minor definition. I'm trying to understand how, in these individual numbers that are in the tables, they could be listed as minor.

MR. NILSON: Dr. Makaroun will address the clinical relevance of the major versus minor definition.

DR. MAKAROUN: You listed several descriptive terms for the type of complication. We did discuss initially why some of them were classified as major, some of them were classified as minor. The type of the complication did not determine whether something was classified as major or minor, but the severity of the complication and its clinical importance.

You went through some examples. I'm blocking on the several that you had the chance to mention, one of them --

DR. JOHNSTON: For example, page 63, Table 21, prosthesis thrombosis.

DR. MAKAROUN: That particular event was a layering of thrombus on the inside of the prosthesis that did not affect the lumen and nothing was done about it, so it was classified as minor.

DR. JOHNSTON: Right.

DR. MAKAROUN: There are certain other --

DR. JOHNSTON: Prosthesis migration, page 72, Table 27, prosthesis migration.

DR. MAKAROUN: Again, the definition of migration was movement of more than 1 centimeter from this predefined level. If it did not compromise the exclusion of the aneurysm, especially with the change of morphology over time and nothing was done about it, it was recorded as an adverse device-related event. But obviously, since nothing was done about it because there was no associated problems, then it became a minor event.

DR. JOHNSTON: Right. Renal insufficiency, for example, Table 21, page 62.

DR. MAKAROUN: The definition of renal insufficiency, maybe I should point you to Tab B in the briefing book. There is predefined categories for all the events that are classified with what is considered significant and nonsignificant.

To be significant, it had to be more than 30 percent rise over the baseline creatinine and if it's more than 30 days, then it became renal failure. If it's less than 30 days, it's renal insufficiency. If it was a very minor transient rise of creatinine that went back to normal, it would have been classified as minor.

DR. JOHNSTON: Understand. Let me see if there's any others. And the same would apply to nerve injury, Table 22?

DR. MAKAROUN: Correct.

DR. JOHNSTON: All right. I think you see where my concern is here. We're dealing with definitions, but we have a relatively small patient population and perhaps not perfect comparative data, and so I just want to make sure that --

DR. MAKAROUN: Correct.

DR. JOHNSTON: -- I understand the classifications. So the ones I have listed, you would all still count as minor and, I guess, I would agree with most of them.

DR. MAKAROUN: Actually some of the predefined limits to some of these complications, in the minds of many investigators, tend to favor the surgical control arm. For example, it was some, let's say, an ileus. If it was less than 96 hours, it was not counted as an ileus. If respiratory failure lasted less than 24 hours, it was not counted as a major adverse event. So when they became major adverse events, they were truly a major adverse event, and some consideration was given to the expected major adverse event that may happen with the surgical control arm.

Something came up a little bit earlier also regarding angina and, in this same line, I would like to indicate that if a preexisting condition, for example angina, was there and the patient had angina afterwards, that was not counted as a major adverse event. That is also indicated in the protocol. And for angina, per se, there was only one incidence of angina in the postoperative period and it was in a test patient and not in a control.

DR. JOHNSTON: I think you'll want to answer some more questions in a second. I want to be clear that I am not dealing with the comparative data. I'm simply dealing with some of the complications that I think I required clarification were minor, because that is now going to affect my view on what was major.

Would the patients, by the way, have regarded those as minor complications since you saw --

DR. MAKAROUN: By and all, yes.

DR. JOHNSTON: Okay.

DR. MAKAROUN: When you tell the patient that the device has moved down, but it's still sealing his aneurysm and there is no evidence that it's going to affect the treatment, they were pretty satisfied.

DR. JOHNSTON: All right. Now, my major concern relates to what was a major complication, not comparing it to open surgery, but what was a major complication and in how many patients? And I have gone through the table and I have listed major complications, such as bleeding with the procedure, that is clearly major, bleeding post-procedure, respiratory failure, renal failure, renal insufficiency, thrombosis, paraplegia, re-operation. I get a total of 33 major complications and we would all agree those are major. Now, is that in 33 patients out of 140?

DR. MAKAROUN: No.

DR. JOHNSTON: Or would some have more than that?

DR. MAKAROUN: Several patients.

DR. JOHNSTON: More than one complication.

DR. MAKAROUN: Correct. Several patients had more than one complication.

DR. JOHNSTON: Right.

DR. MAKAROUN: More than one major adverse event and the definition is the time to the first major adverse event, but then some patients had more than one major adverse event.

DR. JOHNSTON: So how many patients out of the 140 had a major complication?

DR. MAKAROUN: In which group?

DR. JOHNSTON: Purely in the endoprosthesis group.

DR. MAKAROUN: In the TAG group?

DR. JOHNSTON: Correct.

DR. MAKAROUN: The major adverse events was 42 percent for the --

DR. JOHNSTON: That was total. I'm trying to eliminate some things like ileus and so on, because I'm coming from the fact that I am having trouble with the comparative groups and I think some of the statistical concerns have already been expressed.

So I'm trying to put myself in the position of evaluating the TAG group and understanding what percentage of people had a complication. And it wasn't 33 out of 140. It was less than that.

DR. MAKAROUN: The numbers --

DR. JOHNSTON: Complications attributable to --

DR. MAKAROUN: The numbers that were shown are not the number of complications. It's the number of the proportion of patients who had at least one adverse event. So the 42 and the 77, and no matter how you drop it down, are the percentage of patients who had one or more adverse event.

DR. JOHNSTON: I'm not getting the sense that this is a fair description, however, of the device. If I were going to implant one of these devices in the future, if approved, I wouldn't tell a patient that would be their complication rate, because I don't believe it's that high.

Do you understand where I'm coming from? I'm making myself clear? I'm having trouble with the comparative study and I'm trying to understand the complication rate.

DR. MAKAROUN: Collecting prospective data with a predefined set of events that you're going to track is typically what generates event rates that are higher than we typically would associate with certain procedures that we do all the time, and these are the numbers that came out from the prospective collection of the TAG data over two years.

ACTING CHAIR MAISEL: Dr. Johnston, are you saying that there are some events that are defined as major that you don't consider major?