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

9:06 a.m.

CHAIRMAN LASKEY: On the record. If we can have everybody take their seats please. This is a good sign. Everyone listens. Good morning. I=m Warren Laskey. I have the pleasure of calling this morning session to order. The topic this morning will be a discussion of the PMA for the Cordis PRECISE Nitinol Stent System P030047. I=d like to start with our Executive Secretary reading the conflict of interest statement.

MS. WOOD: AThe following announcement addresses conflict of interest issues associated with this meeting and is made a part of the record to preclude even the appearance of an impropriety. To determine if any conflict existed, the Agency reviewed the submitted agenda and all financial interests reported by the Committee participants.

The conflict of interest statutes prohibit special government employees from participating in manners that could affect their or their employers= financial interests. 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. Mitchell Krucoff, Christopher White, and a waiver was previously granted for Dr. Judah Weinberger for their interests in firms that could potentially be affected by the panel=s recommendations. Dr. Krucoff=s waiver involves consulting with a competing firm on unrelated manners for which he receives an annual fee of less than $10,001.

Dr. White=s waiver involves grants to his institution for studies of the sponsor and several competing firms in which he had no involvement in data generation or analysis. Funding to the institution for the sponsor=s study was less than $100,000 per year. The total amount of funding for the competitors= studies was less than $100,000.

Dr. Weinberger=s waiver involves stock holdings in competing firms in which the values are between $25,001 and $50,000. 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 12A-15 of the Parklawn Building.

We would like to note for the record that the Agency took into consideration other matters regarding Drs. Andrew Comerota, Mitchell Krucoff, Kenneth Najarian, Michael Pentecost, Cynthia Tracy, and Judah Weinberger. 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 that these individuals may participate fully in the panel=s deliberations.

In the event that the discussions involve any other products or firms not already on the agenda for which an FDA participant has a financial interest, the participant should excuse himself 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 upon.@

CHAIRMAN LASKEY: Thank you. If we can just go around the table and have everyone introduce themselves.

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

DR. AZIZ: Salim Aziz, Clinical Associate Professor at GW and private practice in Washington.

DR. KRUCOFF: Mitch Krucoff, Cardiology Division at Duke University and the Director of Devices Clinical Trials at the Duke Clinical Research Institute.

DR. TRACY: Cindy Tracy, the Director of Electrophysiology at George Washington University, Associate Director of the Division of Cardiology.

DR. COMEROTA: Anthony Comerota, Vascular Surgeon, Jobst Vascular Center in Toledo, Ohio.

DR. NICHOLAS: Gary Nicholas, Lehigh Valley Hospital, Professor of Surgery, Penn State.

DR. PENTECOST: Michael Pentecost, Chairman of Radiology at Georgetown.

MS. WOOD: Geretta Wood, Executive Secretary.

CHAIRMAN LASKEY: Warren Laskey, Interventional Cardiologist, Uniformed Services University.

DR. ABRAMS: Gary Abrams, Associate Professor of Neurology, University of California - San Francisco.

DR. WHITE: Chris White, Interventional Cardiology, Ochsner Clinic in New Orleans.

DR. WEINBERGER: Judah Weinberger, Director of Interventional Cardiology, Columbia, New York.

DR. MAISEL: William Maisel, Electrophysiologist, Cardiovascular Division at Brigham and Women=s Hospital.

DR. NAJARIAN: Ken Najarian, Interventional Radiologist, University of Vermont.

DR. HUGHES: Allen Hughes, Assistant Professor of MIS at George Mason University, the consumer representative.

MR. MORTON: Michael Morton, I=m the industry representative. I=m employed by Carbomedics.

CHAIRMAN LASKEY: And Geretta, if you could please read the voting status statement.

MS. WOOD: APursuant 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 April 21, 2004: Judah Z. Weinberger, M.D., Ph.D.; Kenneth E. Najarian, M.D.; Michael J. Pentecost, M.D.; Anthony J. Comerota, M.D.; Gary M. Abrams, M.D.; Gary Nicholas, M.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, signed by David W. Feigal, Jr., M.D., M.P.H., Director, Center for Devices and Radiological Health and dated April 16, 2004.@

CHAIRMAN LASKEY: I=d like to begin this morning with the open public hearing portion of our session today. Prior to having the invited speakers come to the podium, I just want to read the following paragraph if I might.

ABoth 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 the 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.@ That being said, I would like to call our first speaker this morning for the open public session. That would be Dr. Janette Durham.

DR. DURHAM: Good morning. I am Dr. Janette Durham, a Professor of Radiology and an Interventional Radiologist from the University of Colorado Health Sciences Center. I have nothing to disclose or a conflict of interest. I am also the President of the Society of Interventional Radiology.

SIR is a non-profit, national, scientific organization of more than 4,000 physicians and Allied Health professionals committed to improving health and the quality of life through the practice of vascular and interventional radiology. This society promotes education, research, and communication while providing strong leadership in the development of health care policy.

SIR members have undergone training and cervico-cerebral angiography as part of our ACGME-approved residency program. Our members have extensive experience placing endovascular stents in multiple vascular beds. SIR recognizes the importance of carotid atherosclerosis and its appropriate management.

In a recent SIR member survey, 22 percent of respondents reported having performed 25 or more carotid stent cases and the collective total of carotid stent experience was over 5,000 cases performed. Of those surveyed, 90 percent responded that they are interested in training to perform carotid stenting.

SIR supports carotid stenting as an effective and beneficial new technology for appropriately selected patients. We believe that there is sufficient evidence to warrant approval of this technology. SIR has had an opportunity to review in a preliminary fashion the training program put forth by the sponsor. We feel it=s a sound program for device training.

We intend to participate as needed to provide educational content and proctors. Procedural safety and effectiveness will be equally as important to device safety and effectiveness. As a physician, I am involved in the diagnosis, prevention, and treatment of stroke.

In my practice, I recognize that stroke is one of the most devastating events a person can experience. Those who survive stroke are often disabled and have extensive health care needs. It is important that appropriately trained and skilled physicians treat patients who are being treated with a device to prevent stroke so that stroke is not the result of treatment.

It is important that labeling include the endovascular skills necessary to ensure high quality outcomes. Physicians are responsible for having undergone the necessary procedural training in addition to device training to qualify them to perform invasive procedures and utilize new technologies.

Hospitals are responsible for overseeing that physicians in fact have appropriate credentials to perform procedures safely. Industry need not share the responsibility for procedural training. To do this would unreasonably burden industry and add to the cost of advancing technology.

SIR has provided CME training and education on carotid stenting at our national meeting the past two years. We plan to continue this effort locally in the next year. In addition, SIR participated in the development of guidelines for the performance of carotid arteriography and most recently has developed a multi-society document for the appropriate quality and performance criteria for carotid artery stent placement which was published last September in The Journal of Vascular and Interventional Radiology and The American Society of Neuroradiology.

These guidelines are based on published science which recognizes a learning curve in the performance of carotid arteriography and carotid stent placement. In respect to stroke, SIR has also participated in developing a multi-society reporting standard for product stent technology assessment and uniformity of reporting in the literature. This will be published this May in Stroke and The Journal of Vascular and Interventional Radiology.

We recognize that carotid stenting is a dynamic area. As additional peer reviewed studies are published, SIR looks forward to working with all specialities involved in carotid stenting to refine these guidelines and further improve patient care. In closing, I thank the panel for the opportunity to provide comments. I am pleased to be available for any questions that you may have.

CHAIRMAN LASKEY: Thank you much, Dr. Durham. We=re going to try and minimize the Q and A, so I=m going to limit this to one question per speaker. Dr. Krucoff.

DR. KRUCOFF: Just a quick question. I=m sorry if I missed this. Is this a formal consensus or position statement on behalf of the society or is this an individual statement?

DR. DURHAM: It is on behalf of the society.

CHAIRMAN LASKEY: The next speaker who has requested time is Dr. Ken Rosenfield representing the ACC and SCA&I. Dr. Rosenfield. Please forgive the presence of the timer. We=re limiting these comments to ten minutes.

DR. ROSENFIELD: My understanding is that there is a shared presentation, SCA&I and ACC, is that correct?

CHAIRMAN LASKEY: That is correct. You will precede Dr. Gray.

DR. ROSENFIELD: Okay, members of the panel, FDA staff, and guests, my name is Dr. Kenneth Rosenfield. I am the Director of Cardiac and Vascular Services at Massachusetts General Hospital. I have the pleasure of standing along side Dr. William Gray who is the Director of Endovascular Interventions at Swedish Medical Center in Seattle.

Dr. Gray and I very much appreciate the opportunity to speak on behalf of two prominent organizations, the American College of Cardiology and the Society for Cardiovascular Angiography and Intervention or the SCA&I. As we embark on our comments, we disclose that we each have served in a consulting role for several companies, Cordis amongst them, whose products may be used for carotid stenting.

We have received modest compensation for time spent away from our practices while serving those consulting roles. In addition, Dr. Gray and I both have served actively as enrolling investigators in the SAPPHIRE trial. Our participation in this and several other trials of carotid stenting for high risk surgical patients as well as our role as busy and experienced cardiovascular clinicians caring for large numbers of patients with a high burden of atherosclerotic disease enables us to comment from an informed and seasoned perspective.

While we are formally here to represent physicians in our respective organizations, we believe that we are ultimately here to represent the patients we all treat. On behalf of those patients, many of whom are at risk for disabling stroke and who will benefit from the lowest risk carotid revascularization available, we, our college, and our society, come today in the strongest support for carotid stenting.

The position that we represent today is that of the ACC and SCA&I. The American College of Cardiology is a 30,890 member non-profit professional medical society and teaching institution whose mission is to advocate for quality cardiovascular care through education, research, promotion, development, and application of standards and guidelines and to influence health care policy. The college represents more than 90 percent of cardiovascular specialists practicing in the United States.

The SCA&I is a 3,150 member non-profit sub-speciality professional medical organization comprised of cardiovascular and vascular interventionalists from several specialities who care for patients with vascular disease and perform both cardiac and extra-cardiac invasive procedures. SCA&I=s mission is to promote excellence in catherization and angiography through physician education and representation, clinical guidelines, and quality assurance to enhance patient care.

On behalf of their members and the millions of patients for whom their members deliver care, the ACC and SCA&I both support treatments and approaches that promise to optimize and/or improve care while minimizing the negative effects and degree of invasiveness for patients. Furthermore, the approach for our organizations and their members has not necessarily been to accept the status quo but rather to pursue advances in treatment in order to accomplish our shared mission.

The ACC and SCA&I are here today in strong support of carotid angioplasty and stenting as an example of innovation and opportunity for less invasive treatment options for our patients. It is perhaps for this reason that more than any other speciality cardiologists have championed this new approach to carotid revascularization and stroke prevention.

There are numerous patients in every cardiology practice who are burdened with comorbid conditions that render conventional endarterectomy higher risk. Perhaps more than any other specialty, it is the patients cared for by cardiologists who have the most to gain if less invasive stroke prevention therapies are available which simultaneously offer reduction in peri-procedure MI and other surgical-related complications while providing for equivalent stroke prevention.

Conversely, it is these same higher risk patients who will suffer most if effective new therapies are withheld or stymied. The college and the society believe that the results of the SAPPHIRE trial along with other data now emerging provide the evidence base to support approval of carotid stenting with this protection for the subset of patients identified by the inclusion criteria for the trial. The ACC and SCA&I organizations strongly support that approval. We would like to focus on several specific areas in our comments to follow.

These include the role of the cardiovascular specialist in carotid artery disease management, secondly, the current gap in care and the lack of evidence base for patients with high risk features undergoing carotid vascularization, thirdly, our society=s interpretation of the SAPPHIRE and other data regarding carotid stenting, and fourthly, the ACC and SCA&I position regarding carotid stenting as an alternative for revascularization including the importance of training and post-market surveillance. A longer written version of our comments has been provided for the panel, the FDA staff, and the Register. With this, I=ll hand the podium over to Dr. Gray.

DR. GRAY: Thanks, Ken. Members of the panel, atherosclerotic disease states our core clinical competency of our two societies and of the more than 30,000 specialists that they represent. Our broad view of cardiovascular patients includes the critical recognition that atherosclerosis is a systemic disease and that the longitudinal clinical care and education of the patient and not episodic intervention is the key to effective reduction of morbid, life altering, and costly events such as myocardial infarction, sudden cardiac death, ischemic cardiomyopathy, renal failure, stroke, et cetera.

Specific to carotid stenting with embolic protection, cardiovascular specialists have been dominant among the vanguard of this new and promising technology for almost ten years and account for roughly 70 percent of all carotid stent procedures performed worldwide to date. In trials now before the panel as well as others to come, cardiologists form the important core of principal investigators and produce nothing short of spectacular results often in hostile, local, regulatory, and reimbursement environments but always in consideration of expanding the safety and effectiveness of the options available to the patient with extracranial carotid artery disease.

The cardiology community prides itself on practicing evidence-based medicine. It is in that spirit that we participate with our peers from other specialties to complete trials such as SAPPHIRE which are designed to clarify the role of carotid stenting vis a vie the existing standard of care endarterectomy.

The cardiology community has gone to great lengths to define the learning curve associated with carotid stenting so as to minimize the chances of causing harm to patients by indiscriminate performance of these procedures by unqualified interventionalists. It is on the background of this dedication to the evidence-based treatment, education, and research of cardiac and vascular diseases for our patients in general and stroke prevention specifically that ACC and SCA&I come before the panel today.

In practice for five decades, carotid endarterectomy for stroke prevention in a patient with extracranial bifurcation disease is an elegant and effective operation. However, not until 1991 with the publication of NASCET was endarterectomy shown to be effective in symptomatic patients versus medical therapy. The results of asymptomatic carotid trial, the ACAS trial, in 1995 extended surgical efficacy to the asymptomatic trial with severe carotid stenosis.

Based largely on these two trials, carotid endarterectomy is performed in over 150,000 patients every year in the United States. It is estimated that approximately two-thirds of these are asymptomatic. While the NASCET and ACAS landmark trials established surgical interventions effective in managing carotid stenosis, these studies excluded patients with significant comorbidities likely to increase their surgical risk.

Indeed, over 80 percent of the patients screened in NASCET and the majority of patients screened in ACAS were excluded mostly on the basis of one or more criteria which may have placed the patient at a higher risk of peri or post-operative procedural events. In high surgical risk patients, there are no randomized data comparing surgery to any alternative therapy. There are however data for multiple high risk surgical registries demonstrating that stroke and death rates are on average at least twice that of the aforementioned trials.

In spite of this lack of randomized control data, endarterectomy continues to be performed in these patients almost with a higher morbidity, mortality, and cost. In short, this patient cohort with endarterectomy has not been shown to be safe nor effective. This represents a significant national gap in the ability to offer these patients a proven therapy.

Endarterectomy has been clearly shown to vary widely with experience and volume. Even at NASCET investigational sites, outcomes are not as robust as those that were seen in the trial. This variability also represents a further gap in assuring predictable, quality outcomes for our high risk patients.

There is then, after review of available information, a clear and worrisome diversions between the clinical data available regarding the benefit of endarterectomy in patients without surgical risk and the current clinical practice of endarterectomy in patients with significant comorbidities in this country. It is on this background and with this gap in mind that we now consider the data in carotid stenting with embolic protection.

The panel is currently considering data from the SAPPHIRE trial, among other sources, in its deliberation regarding the application of Cordis Johnson & Johnson for premarket approval of its carotid stent and embolic protection device for the treatment of high risk patients in extracranial carotid artery disease. A presentation of the SAPPHIRE data has allowed several important observations.

This is the first randomized trial ever to explore any alternative to carotid endarterectomy in high risk patients. Although designed as a non-inferiority trial and in spite of its early stoppage, it appears to have demonstrated a significant advantage of stenting over surgery.

Late neurologic events after 30 days occur infrequently and demonstrate effective stroke prevention which is the goal of any effective carotid therapy. Repeated restorization rates for stenting are meaningfully lower then that for surgery, almost reaching statistical significance in this trial. These results, as sound as they are in and of themselves, are further supported by results already presented in print from other completed trials.

The results from those other trials, investigational carotid stenting in the U.S., demonstrate a remarkable uniformity in nearly 2,000 patients across devices, operators, and sites and endorse the results of SAPPHIRE as consistent and reproducible. It is useful noting that compared to the aforementioned trials ratifying endarterectomy as a standard of care in this country studies reporting outcomes in carotid stenting now total almost four times the number of patients in NASCET and several hundred more than the number of asymptomatic patients study in ACAS.

After reviewing these data, the college and society believe there is strong evidence that rigorous testing of carotid stenting has demonstrated comparable results and even superiority in some cases to carotid surgery in several important categories and in a significant number of patients to draw such a conclusion. I finish comments with Kenny.

DR. ROSENFIELD: Based on the current data available, the college and the society believe that carotid stenting with embolic protection should be made available as an option to patients with clinical or anatomical comorbidities as defined in the SAPPHIRE inclusion criteria in order that they may take advantage of this lower risk alternative to surgery and improve their outcomes. To deny these patients a clearly beneficial alternative to endarterectomy is neither in the best interest of the patient nor society as a whole.

The remarkable results from stenting, achieved in a fraction of the time that it took carotid surgery to mature, will only be replicated through continued expert application of the technology and procedure and with careful patient selection. The necessary skills transfer therefore is important once systems are available out of an IDE setting. Both the ACC and the SCA&I are committed to training and credentialing as a critical component of device and procedural approval.

Competency in carotid stenting requires acquisition of certain skill sets. These include cognitive, clinical, and technical skills. There is clearly a learning curve associated with achieving competence in carotid stenting.

The ACC and SCA&I are in favor of establishing rigorous but not prohibitive training and credentialing requirements. Specifically, we propose that training and certification be obtained within a rigorous, well-defined program which is based on thresholds for achievement of competence but does not present unreasonable barriers.

Several documents are currently under preparation by multi-specialty groups such as the AHA/ACC competency document as well as the AHA guidelines documents for cerebrovascular imaging. These documents will aid in identifying the requisite skills and numbers of procedures to achieve competence.

The college and society also understand the critical need for and support the implementation of careful ongoing tracking of outcomes post-PMA follow up using standardized definitions and measures. This ongoing surveillance will assure the adequacy of training and appropriateness of patient care.

Indeed, the ACC and the SCA&I have been at the forefront of developing standardized and systematized mechanisms by which key clinical and procedural data elements can be collected and analyzed to create new benchmarks and compare to existing benchmarks. As an example, the ACC NCDR, National Cardiovascular Data Registry, in conjunction with the Cardiothoracic Surgical Database represents the largest such effort to date.

ACC NCDR is already conducting two studies on behalf of the FDA. These are underway. We look forward to the opportunity to collaborate further in this regard. In spite of the robust nature of the SAPPHIRE and the other data at hand and the benefits already realized by the thousands of patients who have been treated thus far with carotid stenting in the United States and worldwide, there will be those who will be opposed to carotid stenting approval or critical of the trial design.

I would refer you to the longer version of our comments here, the written document, which would express our feelings about these various issues. Specifically the longer version addresses the issue of MI as an inclusion criteria in this trial, the issue of MI as an endpoint in this trial, the possible requirement for pre-approval by a surgeon before undergoing carotid stenting, and the absence of a medical arm for this trial.

Time precludes us from describing these sentiments in detail, but we would refer you to the written documents that we provided for the panel. We would like to focus on one key element or issue --

CHAIRMAN LASKEY: Dr. Rosenfield, excuse me, you have one minute remaining.

DR. ROSENFIELD: Okay, I=ll wrap up my comments then. The other issue that we would like to refer you to is the issue of the inclusion of asymptomatic patients in this trial and whether this should be applied to asymptomatic patients. ACC and SCA&I believe at this point that the focus should be on implementation and careful roll-out of the technique by ensuring that the procedure is made available to the appropriate patients and while at the same time making certain that its use is not overextended to those who are not high risk as defined in the trial and also recognizing the need for appropriate threshold criteria without creating barriers for talented operators of any specialty to ensure proper training for interventionalists.

Finally, the focus should be on instituting systems to enable meticulous monitoring results in the post-market phase to ensure compliance, proper patient selection, and integrity of the results as well as to provide a mechanism for continued quality improvement. Most importantly, we would like to reiterate that the ACC and SCA&I position regarding this procedure in the current era and as demonstrated by the SAPPHIRE trial is that this can provide a real and meaningful benefit for patients in this country who are at high risk for CEA or endarterectomy.

It is in the best interest of these patients, whose options are quite limited, to make the procedure available. We have been honored to be here today to represent our professional organization. We also are humbled by the opportunity to speak on behalf of the patients who have participated in carotid stent research and future patients who will benefit from its approval. Thank you very much.

CHAIRMAN LASKEY: Thank you both very much. It is clearly a full plate. In the interest of time, again, which is a precious commodity this morning, we will move on. The next speaker requesting time is Dr. Bacarach.

DR. BACARACH: Good morning, ladies and gentlemen of the panel. My name is Dr. Michael Bacarach. I=m very pleased to have the opportunity to present to you today on behalf of the Society of Vascular Medicine and Biology.

I=m a practicing interventional vascular medicine specialist. I=m currently the Director of the Heart Hospital in Sioux Falls, South Dakota. I=m also an Associate Professor of Clinical Medicine at the University of South Dakota. I=m the Treasurer of the Society of Vascular Medicine and Biology.

It=s my goal this morning to briefly describe the Society of Vascular Medicine and Biology, to present our society=s position regarding carotid stent support angioplasty, and the SAPPHIRE trial before you today. I wish to disclose that I did serve as an investigator for the SAPPHIRE trial. I have been an investigator in three additional carotid stent trials.

I have no financial relationship or conflict of interest with Cordis or Johnson & Johnson. I have received no compensation for my appearance today. I am here as an officer of the Society of Vascular Medicine and Biology to present our society=s position.

The Society of Vascular Medicine and Biology is an non-profit professional medical organization. It was founded in 1989 to foster a broad mission of patient care, education, and research in the field of vascular medicine. Our goal is to maintain a high standard of clinical practice and patient advocacy in vascular medicine.

The Society of Vascular Medicine and Biology is the only national professional medical society representing physicians with expertise in medical, surgical, and endovascular strategies for the treatment of these complex patients. Our membership includes individuals with expertise in vascular medicine, cardiology, vascular surgery, radiology, vascular nursing, vascular technology, and vascular biological research.

Extracranial carotid artery disease is an area of expertise of the physician members of the society. The development of endovascular therapy for vascular disease has been profound and has led to many advances which have improved the care of our patients with vascular disease. Specific use of less invasive therapies and strategies for revascularization have made treatment for many of my complex patients deemed suboptimal candidates for surgical revascularization life saving.

Carotid stent support angioplasty using cerebral embolic protection devices is one example of such innovation and advantage to our patients. My colleagues and I see many patients with carotid lesions that are inaccessible to standard endarterectomy or have prohibitive surgical risk from serious comorbid conditions making treatment difficult and risky.

Carotid stent support angioplasty represents a major advance in my ability to care for these patients. The SAPPHIRE trial was performed with sufficient scientific rigor and oversight to demonstrate convincingly that carotid stent support angioplasty with embolic protection is an appropriate first line therapy for high risk symptomatic and asymptomatic patients.

The society was impressed with the results of the SAPPHIRE trial. Our society strongly supports approval of carotid stent support angioplasty with embolic protection for high risk patients deemed to be in need of revascularization for the prevention of stroke.

The benefits of carotid stent support angioplasty by appropriately skilled, trained, and experienced operators and interventionalists are established. We do not support however broad adoption of this technology and technique without responsible and adequate training. As a national, professional medical society, the Society of Vascular Medicine and Biology urges you to approve carotid stent support angioplasty with embolic protection for high risk patients.

We urge you to assure that the proper training and experience is required prior to the adoption of this technique. Physician thought leaders must be involved in the development of this treatment breakthrough so that responsible, skilled, and experienced physicians treat our ill patients in the best, safe, and most appropriate manner. I thank you very much for the privilege of representing the society before you today. Thank you.

CHAIRMAN LASKEY: Thank you, sir. Any questions from the panel? Is there anyone else who wishes to come forth and address the panel on today=s topic or any other topic? Yes, sir, please come forward. Just identify yourself.

DR. HANLEY: Sure, I=m Daniel Hanley. I represent the American Academy of Neurology.

MS. WOOD: Do you have any financial disclosures?

DR. HANLEY: Certainly. I represent the American Academy of Neurology. They have paid for my transportation here. I have previous relationships with Jansen as a medical consultant. This is a Johnson & Johnson company. I have no relationship with Cordis.

I am a former board member of the National Institute of Health, American Academy of Neurology, and a current board member of the National Stroke Association. I=m a board member of a for-profit public company, NMT, which makes cardiologic devices which do not compete with this device.

Representing the academy, I wish to address the panel today. I bring to my comments 25 years of experience as a stroke physician and neurologist with an emphasis on acute care neurology, interventional procedures, their complications, and post-procedural care and recovery of stroke patients. I bring one decade of public advocacy for improved stroke care on the part of the American Academy of Neurology, American Heart Association, and the National Stroke Association.

I wish to comment in three areas: (1) to enforce the importance of the entire process today, (2) to make the panel aware of an academy white paper regarding training, and (3) to make a simple comment regarding the standards by which comparisons should be made.

The first issue, I=m pleased to be here while the FDA deliberates on a new industry sponsored trial data set that could lead to reduction in stroke events and the improvement or the addition to the armamentarium of interventions for Americans with stroke risks. The AAN, American Academy of Neurology, doesn=t presume to predict the outcome of today=s deliberation. Rather, we hope that patient safety and benefit are enhance by today=s outcome.

My second comment, we wish to make the Committee aware of the last three decades effort to improve stroke and stroke care by the systematic use of practitioner training pathways. The academy has not had the opportunity to comment on the training pathway suggested for this application but welcomes that opportunity now and hopes to submit its comments in the near future.

The specific neurovascular stroke coalition pathway has been developed and is brain-specific. It is to this that I wish to speak. Despite this pathway=s sponsorship by organized radiology, neurology, and neurosurgery, it is not as well known as similar heart-based pathways for coronary angiography and coronary procedures.

The pathway is articulated in the American Academy of Neurology=s white paper, a copy of which will be left today with this panel. The academy wishes to ask that the details of this pathway for training and competency in cranial-cervical angiography be incorporated into the decision-making today regarding the overall use of stenting devices and the protocol to place stents in patients with stroke.

The essence of the white paper is that patient safety is only protected when we apply to cranial-cervical angiography and carotid stenting the lessons we have learned in coronary angiography. These lessons have lead to improved heart outcomes. The deliberation today must consider how we can achieve a different goal, improved brain outcomes.

The lessons we believe are quite simple. (1) The procedure in question must be performed by practitioners with prolonged training times specific to diseases of the brain because patient selection, pre and post-procedure management, and procedure performance are all directed at brain processes. (2) The proceduralist must demonstrate both technical and cognitive competence prior to credentialing to select patients, perform carotid stenting, and organize the care of these patients after the procedure.

(3) Because stroke is the most feared medical complication, the standards for performance of brain vascular procedures should be at least as stringent and at least as specific as the standards for coronary angiography. Specifically, these are a minimum of experience of 100 procedures for technical competence and a minimum training period of one year in brain stroke patient care in an ACGME credentialed neurovascular program for cognitive training.

The issue of non-neurologically trained specialists is addressed in this white paper. We believe that these requirements should apply to all practitioners whether they are neurologically trained or not. We do not believe that training in coronary disease and coronary angiography alone prepare the practitioner for treatment of stroke.

We do not believe that short, CME courses, whether industry sponsored or otherwise, or simulation of procedures, not on patients, substitute for organized, credentialed training in brain vascular angiography. We make this recommendation because it is evidence-based and has been demonstrated in multiple brain angiographic domains to produce optimal patient safety.

We ask that the decision-making today regarding stroke, a brain disease, and this carotid stent device reflect our extensive knowledge about training and competency for brain angiography in the indications, in the labeling, and the instructions regarding competency of physicians who will perform this procedure. My third comment is directed towards the standards that should be applied today.

We suggest that the standard that protects patient well being be the current established medical therapy for stroke and that comparisons of the event rates for patients who are risk matched with medical treatments not requiring angiography or stent placement be considered in today=s deliberations. I thank you for your patience and I=m willing to answer any questions. We will provide you with a copy of the white paper which has been endorsed by all of the neuro-societies and radiology.

CHAIRMAN LASKEY: Thank you, sir. Any panel questions?

DR. HANLEY: Thank you.

CHAIRMAN LASKEY: Thank you again. Anyone else? Then at this point, I would like to close the open public hearing and move on to the sponsor presentation.

DR. COHEN: Mr. Chairperson, Committee Members, Dr. Zuckerman, representatives of the FDA, and representatives of the public, good morning. My name is Dr. Sidney Cohen. I=m Group Director of Clinical Research at Cordis Corporation. I=ll be presenting on behalf of Cordis this morning. I=m also an Adjunct Associate Professor of Medicine at the University of Pennsylvania.

In the next hour and 15 minutes, I would like to cover the following topics. I=d like to provide an overview of this project, go over some background information on stroke and carotid endarterectomy, provide a brief description of the devices that were studied, and provide an overview of the PMA clinical data which encompasses a total of 1,619 patients.

This includes both non-randomized carotid artery stent supportive data from two trials, the CASCADE study and a FEASIBILITY study done predominantly in the United States as well as the pivotal trial data from the SAPPHIRE trial which will be presented by Dr. Ken Ouriel. I will then briefly provide an overview of the training program that we have developed and discuss our plans for post-market surveillance study.

The requested indication is detailed here. I=m not going to read it for the sake of time. But to summarize, the Cordis PRECISE Nitinol Stent System used in conjunction with the ANGIOGUARD XP Emboli Capture Guidewire is indicated for the treatment of carotid artery disease in high risk patients. High risk is defined as at least 50 percent stenosis in patients with symptoms and at least 80 percent stenosis in patients without symptoms.

In addition patients both symptomatic and asymptomatic must have more than one condition or at least one condition that places them at high risk for carotid endarterectomy. We=ll go into what those risk factors are in the course of this presentation.

These studies started with a U.S. study called the U.S. FEASIBILITY study which was begun in September 1998. The SAPPHIRE pivotal study was begun in August 2000. The PMA was filed in October 2003. There are three conclusions from these studies that we plan to prove this morning, and that is (1) that we achieved our primary end point of non-inferiority of carotid artery stenting to carotid endarterectomy at one year for the major end point of major adverse events, (2) that carotid artery stenting provides improved outcomes in terms of reducing myocardial infarction, reducing the need for reinterventions and producing a statistically significant decrease in cranial nerve injuries, and (3) that the benefit of carotid artery stenting is sustained, and we will provide data up to three years from our studies.

Finally, the PMA was granted expedited review status in November 2003 being considered a significant therapeutic advance. You may be aware that Cordis was issued a warning letter on April 1. Cordis continues to work with the FDA on GMP and quality systems issues.

I have some background information on stroke and carotid disease. There are over 700,000 strokes that occur annually in the United States. Stroke is the third leading cause of death with an estimated 164,000 deaths per year. Up to 30 percent of strokes are caused by carotid artery disease. It=s the number one cause of disability in the United States.

The costs to take care of patients with stroke are in excess of $53 billion per year. If you are under the age of 65 and you have a stroke, you have an over 50 percent chance of dying within eight years. But by enlarge, this is a disease that affects the elderly and particularly those with comorbid medical conditions.

Carotid endarterectomy has a 50 year history of development and refinement to its present status. It=s currently the interventional standard of care in treating patients with carotid disease with the purpose of reducing stroke. There are up to 200,000 carotid endarterectomies performed each year in the United States.

It=s estimated that at least 20 percent of carotid endarterectomies are performed on high surgical risk patients annually in the United States with high surgical risk defined based on anatomic and medical comorbidities where the anatomic issues increase the risk of the procedure and the medical comorbidities increase the risk of having a myocardial infarction and death. There are a number of randomized clinical studies which have supported the superiority of carotid endarterectomy over best medical therapy that was available at the time the studies were undertaken.

These studies have led to carotid endarterectomy again being considered the standard of care for the interventional treatment of both symptomatic and asymptomatic carotid artery disease. It=s clear, however, that the current treatment of patient with carotid disease using carotid endarterectomy extends beyond the NASCET and ACAS inclusion criteria.

By enlarge, NASCET and ACAS studied a relatively healthy subset of patients. For example, ACAS screened 25 patients in order to enroll one patient whereas NASCET only enrolled one out of every three patients who underwent carotid endarterectomy at the participating institutions.

Patients considered at high risk for carotid endarterectomy, as defined by ineligibility, comprise up to 50 percent of patients in different published series. A study from the Ochsner Clinic encompassing 366 patients yielded 46.2 percent being trial ineligible. A study from the Cleveland Clinic encompassing over 3,000 patients indicated that just under 20 percent of patients were trial ineligible.

From a database for the Agency for Health Care Research and Quality, which encompasses over 7.5 million admission during the year 2001, there were 30,000 patients in that database who underwent carotid endarterectomy. And 35.1 percent of those had features that would have made them ineligible for NASCET and ACAS being considered them high risk.

The specific criteria that we=re talking about include anatomic and medical comorbidities. The anatomic risks include tandem lesions, previous carotid endarterectomy, previous radiation therapy to the neck, and status post-radical neck dissection. The medical comorbidities include age greater than 79, a previous stroke, a previous myocardial infarction, unstable angina, atrial fibrillation, symptomatic heart failure, valvular heart disease, cancer with a less than 50 percent five year survival, and renal pulmonary and liver failure.

The data on the next several slides will provide evidence in two regards; first, that outcomes in patients undergoing carotid endarterectomy do not match what is in the literature and in addition that there are patients that are at high risk that are undergoing carotid endarterectomy. This is a study published by Wennberg in which mortality in patients in a Medicare database of 113,000 patients treated with carotid endarterectomy from 1992 and 1993 was investigated.

On the left side, you can see the mortality rates from the ACAS study. For the NASCET study, you see that the mortality for patients undergoing carotid endarterectomy in the same hospitals that participated in ACAS and NASCET are more than two-fold higher than the mortalities reported in the literature for those two studies. And non-trial hospitals had somewhat higher mortality.

In addition, non-trial data from a number of centers that includes both single center, Ochsner Clinic, Ohio Registry which is a composite of Medicare database from that state, and New York Registry both symptomatic and asymptomatic patients, a composite of six hospitals, yielded incidents of rates of death of up to one percent, rates of stroke between two and a half and four and a half percent, giving rates of stroke and death between two and a half and five and a half percent.

Another study of academic medical centers in a retrospective analysis of 1,160 patients at 12 centers in the United States for patients undergoing carotid endarterectomy in the years 1988 through 1990, using an end point of in-hospital death, myocardial infarction, and stroke, and an end point that=s similar to that used in the SAPPHIRE trial, yielded an overall outcome of 6.9 percent. Patients who were over 75 who were symptomatic or had angina had higher event rates than those overall.

If we break out that 6.9 percent overall rate into its individual components, we see a death rate of 1.4 percent, non-fatal stroke rate of 3.4 percent yielding a combined death/non-fatal stroke rate of 4.8 percent and a MI rate of 2.1 percent. Certainly this study as well as the previous studies suggest both that patients currently undergoing carotid endarterectomy have risk factors that lead to outcomes that are not quite what is published in ACAS and NASCET.

In addition, there=s data that the patients currently undergoing therapy are actually comprised mostly of asymptomatic patients. Again, data from the same registries mentioned before or single site data indicates that a low of 25 percent or anywhere between 60 and 75 percent of patients currently going carotid endarterectomy in the United States are asymptomatic.

While there is no contemporary data that would allow us to understand the outcomes with medical therapy of patients who have carotid stenosis and who are asymptomatic, there is data that is more historical in nature that could be brought to bear on this. This is a study of asymptomatic patients totaling 1,196 which indicates that the stroke rate is fairly flat until you get to the 80 percent level where the stroke rate increases rapidly from one percent up to over five and a half percent.

This value of 80 percent to 99 percent actually is supported by data published from the European Carotid Surgery Trialists paper of asymptomatic patients which indicated that the three year stroke rate for the same cohort of patients for the 80 to 89 percent was 9.8 percent and for the 90 to 99 percent was 14.4 percent. In addition, I would remind you that of the patients enrolled in the ACAS trial, only one-third of those had an 80 percent or greater stenosis.

In fact, this data led to the choice of 80 percent as the minimum stenosis for asymptomatic patients in the SAPPHIRE trial. Thus, in the United States, the standard indications for surgical treatment of carotid disease include both NASCET and ACAS eligible as well as ineligible patients, symptomatic and asymptomatic patients, and higher risk patients with high risk being defined on anatomic and medical comorbidities and thus, SAPPHIRE trial study patients who currently are referred for treatment of their carotid disease.

We chose to study high surgical risk patients because in the initial evaluation of the new technology, it was decided to study it in a cohort of patients where carotid endarterectomy is technically demanding. It=s demanding based on anatomic factors which difficult access surgically may lead to increased local tissue and nerve injury as well as for the presence of medical comorbidities where patients would be less tolerant of general anesthesia and surgery. Thus, carotid artery stenting is studied as an alternative and less invasive method of therapy.

I=d like to move on now to a brief description of the devices used in these studies. The carotid artery stenting system consists of two devices; a stent delivery system and emboli protection device. The stent delivery system is comprised of a stent and a delivery catheter.

The Cordis PRECISE Nitinol Stent comes in two french sizes; 5.5 french and 6 french. The 5.5 french comes in diameters of 5, 6, 7, and 8 millimeters with lengths of 20, 30, and 40. The 6 french system has sizes of 9 and 10 millimeters diameter by 20, 30, and 40 millimeters in length.

In addition, tapered stents were studied. In the 5.5 system, that=s a 6 to 8 millimeter taper diameter by 30 millimeter length. For the 6 french system, 7 to 9 and 7 to 9 millimeter diameters with a 30 millimeter length. The stent delivery system has a usable length of 135 centimeters with a guidewire lumen of 0.018 inch.

Emboli protection is provided by the ANGIOGUARD XP Emboli Capture Guidewire. This is a polyurethane filter on a Nitinol frame. Basket diameters range from 4 to 8 millimeters. We oversize the basket in use by anywhere from 0.5 to 1.5 millimeters versus the reference vessel diameter. The pore size of the filter is 100 microns. The crossing profile is 3.5 french. The wire diameter again is 0.014.

I=d like to show an animation of the system in use. What you will see is, first, the inside view of the artery. That=s not good. What you would have seen is the inside view of the artery with first the ANGIOGUARD device being deployed past the lesion, the sheath being withdrawn, deploying the umbrella-shaped ANGIOGUARD. That would be followed by a balloon dilatation with release of material from the lesion being captured by the ANGIOGUARD which is distill to the lesion, the placement of the stent which is a Nitinol stent which self-expands upon withdrawal of the sheath, and then finally capture of the ANGIOGUARD device and then retrieval of that device from the body.

I=d like to move on now to an overview of the PMA clinical data which encompasses a total of 1,619 patients. Again, this is provided as supportive data from the CASCADE study done in Europe and the FEASIBILITY study done predominantly in the United States. The purpose of these two studies were to gain experience with the carotid stent system and provide a learning curve for investigators.

It allowed us to refine the stent delivery system and to evaluate the advantage of adding the ANGIOGUARD device. Two studies will be described. The CASCADE study done entirely in Europe was a non-randomized study of carotid artery stenting encompassing 121 patients. Even though the primary end point was 30 days, we have a one year follow up in those patients.

The FEASIBILITY study was done predominantly, again, in the United States. It=s a non-randomized study of carotid artery stenting. A total of 261 patients were enrolled. That has a three year follow up even though the primary end point was not at three years.

Let=s move on to the CASCADE study. The objective here was to evaluate the safety and performance of the SMART stent with and without ANGIOGUARD Emboli Capture in patients with high grade carotid artery stenosis. The primary end point was ipsilateral stroke or procedure-related death within 30 days of stent implantation.

This is a multi-center, prospective, non-randomized study in nine centers in Europe using the 7 french SMART system which is a predecessor to the PRECISE system, identical stent just a slightly different delivery system. There were 121 patients enrolled, 31 with ANGIOGUARD. It was conducted from September >98 through May 2002. It included symptomatic patients with greater than 70 percent stenosis, asymptomatic patients greater than 85 percent stenosis with the stenosis occurring between the origin of the origin of the common carotid and the extra-cranial segment of the internal carotid artery.

The primary end point is shown here. (Indicating.) There were no procedure-related deaths. Ipsilateral stroke occurred at a rate of 7.4 percent. If we divide the data between the patients who were treated with stent alone in blue and stent with an ANGIOGUARD in red, we see a reduction of events in the patients we used with ANGIOGUARD with ipsilateral stroke rate of 3.2 percent and no major ipsilateral strokes.

Conclusions. From the CASCADE study, which is carotid artery stenting, was found to be feasible for the treatment of carotid stenosis. The ANGIOGUARD distal protection device functioned well and appeared to reduce the risk of distal embolization resulting in fewer strokes such that use with the ANGIOGUARD the 30 day stroke rate was 3.2 percent with no major strokes.

The U.S. FEASIBILITY study=s objective was to assess the feasibility of carotid artery stenting in the treatment of obstructive carotid artery disease. It=s also to assess and standardize optimal operator techniques as this also served as the run-in phase for the clinical trial. It was designed as a non-randomized prospective study of 33 centers using the 6 and 7 french SMART system, again predecessors to the PRECISE system, and the 5.5 french PRECISE stent delivery system.

There were 261 patients enrolled, 85 of whom received stenting with the ANGIOGUARD device. They were enrolled from September >98 through July 2001. We have follow up out to three years. Inclusion criteria included patients who were symptomatic, needed to have at least 60 percent stenosis. Patients who were asymptomatic had to have at least 80 percent stenosis by ultrasound or angiography with again disease of the native common or internal carotid arteries.

Inclusion criteria here were somewhat different. They included anatomic risk factors which made the patients at somewhat higher risk for surgical endarterectomy. This included restenosis after carotid endarterectomy, a history of radical neck dissection, a history of contralateral carotid artery occlusion, a history of an ostial lesion of the common carotid, and a high take off carotid bifurcation disease.

The primary end point was 30 day major adverse events, MAE, defined as death, any stroke, and/or myocardial infarction. Key secondary end points included major clinical events at six months and yearly to three years, patency defined as less than 50 percent restenosis by carotid ultrasound at 48 hours, 30 days, six months, and yearly to three years, and neurologic assessments that were performed by an independent neurologist at 28 hours, 30 days, six months, and yearly to three years.

The end points are depicted here with a death rate of 0.8, MI of 1.1, stroke of 6.1 yielding a major adverse event rate of 6.9. Again, if we take the data and separate it out between the patients in blue who received a stent only versus patients in red who were treated with a stent and the ANGIOGUARD, you see that the stroke rate with ANGIOGUARD is 2.4. Once again, there were no major ipsilateral strokes.

We have here the cumulative incidents of major adverse events. I=d like to take a second to review this slide as you will be seeing this cumulative incidents curve several times during this presentation. At the very bottom of the curve - and I=m sorry, I don=t want to hit the gentleman=s head with the back of the pointer here - but you see the table that indicates the number of patients at risk at the different time periods.

On the Y axis is the cumulative percentage of major adverse events. The X axis is the time after the procedure. The end points are indicated at 30 days here by the numbers, one year, two years, and three years. Error bars are 1.5 times the standard error.

What you see here is an increase in the rate of adverse events over the three year follow up. When we look to see what the components of this increase in curve are, first, we look at the cumulative percentage of all stroke to 30 days and ipsilateral stroke from days 31 through three years. What you see is a rate at 30 days of 6.1 which increases to 8.7 at three years. That an increase of just under one percent per year.

On the other hand, if you look at the cumulative incidents rate percentage of death to three years, you see an increase in the curve over the course of this time period. It is this increased death that contributes to the increased rate of major adverse events. This increase of death rate or the deaths are likely due to the elderly age and the significant medical comorbidities of these patients.

In conclusion, for the U.S. FEASIBILITY study, we were able to demonstrate the feasibility of carotid stenting with the Cordis PRECISE Nitinol stent system. The ANGIOGUARD Emboli protection device appeared to reduce the incidents of stroke. Again, with use of the device, the stroke rate at 30 days was 2.4 percent and there were no major strokes. This also provided a run in to the pivotal SAPPHIRE study.

Because the number of patients in the FEASIBILITY study and the CASCADE study were small, we did an exploratory analysis to see whether combining the data from those two trials would yield significance. So on the right side of the slide here is the combined incidents of stroke without ANGIOGUARD and the combined incidents of stroke with ANGIOGUARD. You see the difference here, from 8.6 to 2.6, does reach statistical significance at the p = 0.02 level.

From these two studies, we were able to refine the carotid artery stent delivery system with a reduction in profile from 7 french to 5.5 french. That allowed us to improve the design of the delivery system. The data supports the benefits of the ANGIOGUARD Emboli protection device in reducing stroke. It has demonstrated the feasibility of performing carotid artery stenting. At this time, I would like to ask Dr. Ken Ouriel to come to the podium to present the pivotal SAPPHIRE trial data.

DR. OURIEL: Thank you, Sid. Good morning. I=m Dr. Kenneth Ouriel. I=m Chairman of the Division of Surgery at the Cleveland Clinic Foundation and Professor of Surgery at the Cleveland Clinic Lerner College of Medicine at Case Western Reserve University.

I=m one of the members of the executive committee of SAPPHIRE. I=m going to present the methodology and results of this pivotal trial. I=d like to disclose that my lodging for one night was paid by Cordis. My travel here was paid for by the Cleveland Clinic. I have no other conflicts to disclose at this time.

The objective of the SAPPHIRE study was to compare the safety and effectiveness of carotid stenting with emboli protection to endarterectomy in the treatment of carotid artery disease in high risk patients. There were a total of 2,294 patients screened for eligibility for the SAPPHIRE trial. Among these, roughly one-third or 747 patients met the inclusion and exclusion criteria as determined by concurrence between and interventionalist, a surgeon, and a neurologist at each site.

Within this cohort of 747 patients, both the surgeon and the interventionalist felt that either carotid stenting or endarterectomy were feasible in 334 patients. This group underwent randomization to stent treatment in exactly one-half or 167 patients and to endarterectomy in the other one-half.

There were 406 patients who the surgeons thought were unacceptable for carotid endarterectomy. These patients were not randomized. Rather, they were entered into a non-randomized stent treatment arm. There were seven patients who the interventionalists thought were at unacceptable risk for stenting. These patients were entered into a small, non-randomized endarterectomy treatment arm.

The primary end point of this trial was death (all cause), any stroke, and myocardial infarction to 30 days post-procedure plus death (all cause) and ipsilateral stroke between days 31 and 360 post-procedure. There are real differences between SAPPHIRE and previous surgical trials.

First, the primary end point of SAPPHIRE included all cause mortality rather than just peri-procedural or neurologic-related deaths. The composite end point of major adverse events included myocardial infarction in addition to death and stroke. The 24 hour post-procedure stroke evaluation was performed by a neurologist.

Stroke scales were utilized in addition to physical examination in the classification of stroke. Vessel restenosis and patency was documented by duplex ultrasound. Lastly, a multi-disciplinary team provided input on the treatment strategy including eligibility and appropriateness for randomization.

Some have asked why myocardial infarction was included in the primary end point of SAPPHIRE. Myocardial infarction leads to disability, death, prolonged hospitalization, and health care costs and as such is thought to be a key safety end point. In patients undergoing vascular surgery who sustain a perioperative non-Q wave MI, there is a six-fold increase in mortality over the subsequent six months.

Perioperative myocardial infarction predicts mortality at one year. There is a 27-fold increase in the risk of another myocardial infarction over the next six months. Therefore, perioperative myocardial infarction is a strong surrogate for long-term mortality after vascular surgical procedures. Lastly, perioperative myocardial infarction is part of the primary end point for other carotid artery stenting trials such as CREST and ARCHeR.

Myocardial infarction was defined as either Q-wave or non-Q-wave. The definition of Q-wave MI was relatively standard requiring acute symptoms and new pathologic Q-waves. Non-Q-wave MIs were defined using the WHO definition of a CK ratio of greater than two times the upper limit of normal and a CK-MB fraction greater than normal in the absence of new Q-waves.

The definition of stroke was standard requiring a focal deficit of abrupt onset lasting more than one day. While the presence or absence of a stroke was not determined using stroke scales, when a stroke did occur, it was classified as major or minor using the NIH, Rankin, and Barthel scales.

The SAPPHIRE study was designed as an equivalence or in statistical parlance non-inferiority trial. The design was based on the following parameters. This was a high risk study. The majority of events were expected to occur within 30 days for an overall one year event rate of 14 percent. The delta was chosen to be three percent, a definition that was agreeable to the clinicians and the Agency.

The statistical power was set at 90 percent. The one-sided type I error rate was set at 0.025 which is conventional. What this means is that we would expect the results to be equivalent if we could be 97.5 percent certain that stenting was no more than three percent worse than endarterectomy.

We employed an interim analysis plan so that we could terminate the trial early if we could demonstrate either non-inferiority or superiority. Given the fact that this was the first randomized FDA IDE trial and had a potential for slow enrollment, the triangular method that we employed allowed for flexibility in choosing the timing of sequential testing during enrollment.

Our initial plan was to allow the performance of interim analyses every 100 patients. This statistical plan was also flexible to allow enrollment of up to 2,400 patients, if needed. For example, this is roughly the sample size of CREST. Based on conservative efforts of the stent=s performance, we anticipated that a sample size of 600 to 800 patients would result in a decision to stop the trial for non-inferiority.

As the FDA has pointed out, the initial analysis plan was changed. All changes were done in accordance with the flexibility allowed with the triangular method. We decided to omit the first interim analyses since a sample size of anything less than 300 patients was thought to be unconvincing. Before the revised planned analysis in the fall 2001, it was clear that enrollment was proceeding so slowly that we were unlikely to reach 400 patients.

Therefore, with an expectation of the trial terminating for slow enrollment between 300 and 350 patients, we decided to omit all interim analyses and perform a single final analysis when enrollment was inevitably terminated. It is important to point out that this change in interim testing was permitted under the triangular method. Since interim analysis was not performed in the study, the first analysis was the final analysis. Therefore, standard testing without correction for interim looks was appropriate.

This is a graphical representation of the rate of enrollment. (Indicating.) Enrollment was robust for the first 12 months of the study. At that point, enrollment diminished concurrent with the availability of competing stenting registries from other companies. There were now outlets for patients to be treated with stenting outside of the randomized SAPPHIRE trial. In fact, the Cordis site IDEs began after the termination of randomization.

Importantly, all patients enrolled in SAPPHIRE were referred for treatment of their carotid disease. All randomized patients would have been treated likely with endarterectomy if not for the trial. Symptomatic patients were required to have a 50 percent or greater stenosis by duplex or angiography. Asymptomatic patients had to have a stenosis of 80 percent or greater.

Disease had to be located in the native common or internal carotid artery. Importantly, consensus agreement by a multi-disciplinary team was required which included an interventionalist, a neurologist, and a surgeon. A patient had to have at least one comorbid condition which increase the risk of endarterectomy. These comorbid conditions could be anatomic or medical.

Key anatomic inclusion criteria that assured a high risk subset included contralateral carotid occlusion, contralateral recurrent laryngeal nerve palsies, previous radiation therapy to the next, previous endarterectomy with the presence of a recurrent stenosis, difficult surgical access such as a high internal carotid artery lesion, or severe tandem lesions.

Key medical comorbidities that assured a high risk subset included the following: congestive heart failure, open heart surgery within six weeks, a recent myocardial infarction, angina at a low workload or unstable angina, severe COPD, or age greater than 80 years.

At this point, I would like to present the results of the randomized portion of the SAPPHIRE trial. Table 1 of any randomized trial is always a comparison of the demographics and comorbidities of the two treatment groups. The randomized stent and randomized endarterectomy arms of SAPPHIRE were similar with respect to all baseline variables except three: coronary artery disease, previous coronary bypass, and previous PTCA. These characteristics were more frequent in the stenting arm. So if anything, the randomized stent arm was slightly more ill than the randomized endarterectomy arm.

There was a high degree of procedural success in the stented patients. The stent was successfully delivered to its intended location more than 99 percent of the time. Deployment of the stent resulted in less than a 30 percent residual stenosis in approximately 90 percent of the cases. The 30 percent threshold is currently used for coronary stent trials however.

Using a 50 percent threshold, possibly more appropriate for a peripheral trial, approximately 99 percent of the patients were successfully treated. The ANGIOGUARD filter was deployed on the first attempt and retrieved successfully in over 95 percent of the subjects in the randomized stent arm and in over 91 percent of the patients in the non-randomized stent arm. Ultimately, 98 percent of the randomized stent and 95 percent of the non-randomized stent subjects had successful deployment and retrieval of the ANGIOGUARD device.

Let=s move on to study outcome presenting data on an intent to treat basis unless otherwise specified. Among the 167 patients randomized to stent, one year compliance was achieved with respect to clinical criteria in 93.5 percent of cases and with respect to duplex ultrasound in 80.6 percent of the cases.

In the endarterectomy group, complete clinical follow up was available at one year in 85.6 percent of the cases and duplex ultrasound in about 69 percent of the cases. To remind you, all clinical events were adjudicated by an independent clinical events committee and all angiograms and duplex studies by independent core laboratories.

This slide depicts 30 day data in the two randomized groups; endarterectomy in red and stenting in blue. There were no statistically significant differences in the rate of death, stroke, myocardial infarction, or the composite of major adverse events. At one year, again, there were no statistically significant differences in the frequency of death, stroke, myocardial infarction, or major adverse events. In each case, however, the data trended in favor of stenting over endarterectomy.

This is probably the most important slide that we=re going to show you today. This is the primary end point analysis. The percent difference in one year MAE is along the abscissa with a dotted red line demonstrating the target delta of three percent. The horizontal line is the point estimate for the MAE difference with a raw value of 7.2 percent in favor of stenting over endarterectomy.

As you can see, the 95 percent confidence interval is to the left of the margin of non-inferiority. In other words, the primary goal of the study was achieved. We were more than 95 percent certain that stenting was no more than three percent worse than endarterectomy. In fact and importantly, we were certain that non-inferiority was achieved with a p-value of 0.0035. In fact, with this particular test, had the 95 percent confidence bar been slightly to the left of zero rather than slightly to the right, stenting would actually have been statistically superior to endarterectomy with regard to the primary end point.

The FDA statisticians asked us to perform the analysis as if we had performed interim testing at 100, 200, 300, and 334 patients. This table displays the results of that retrospective interim analysis. There would have been three interim analyses and one final analysis. The recommendations are listed in the last column and would have been as follows.

We would have chosen to continue the trial after 100 patients. We would have chosen to continue the trial after 200 patients. We would have decided to stop the trial at 300 patients. There would have been some additional run on patients. We probably would have ended up with somewhere between 300 and 350 patients. The final analysis would have included the run ons.

The p-values for superiority would have been 0.066. Importantly, the p-value for non-inferiority would have been 0.003, well below our threshold of 0.025. So with interim analyses and with corrections for multiple sequential testing, our conclusion would have been exactly the same. Stenting is equivalent to endarterectomy.

Having demonstrated non-inferiority in the primary end point of one year major adverse events, it makes sense to look at the individual end points at one year. There were no statistically significant differences between the randomized groups. But again, as this slide demonstrates, all trends were in favor of stenting over endarterectomy.

Again, there was no statistically significant difference in the rate of stroke at one year, 7.2 percent in the endarterectomy arm and 6 percent in the stented arm. Diving strokes into major and minor ipsilateral events, it appeared as though the strokes that occurred in the endarterectomy patients were more often major, and the strokes that occurred in the stented patients were more often minor. But these differences did not attain statistical significance.

These two Kaplan-Meier curves represent the cumulative percentage of subjects experiencing a major adverse event over one year of follow up. The MAE rate was 20.1 percent in the endarterectomy group and 12.2 percent in the stented group. While the trial was designed to be a non-inferiority trial, stenting almost hit statistical significance for superiority. The p-value was 0.053 with a log rank test.

Data out to two years is displayed here. The trends continued through 720 days of follow up with an MAE rate of 26.7 percent in the endarterectomy group and 19.2 percent in the stent group.

When the composite adverse event rate is split out by its components, we see that the rate of perioperative stroke was relatively low at just over three percent in both treatment arms. Importantly, the rate of stroke remained relatively flat thereafter with roughly a one percent annual risk of subsequent stroke over the next two years.

These two Kaplan-Meier curves depict the risk of death over two years of follow up. The risk of perioperative death was relatively low at 2.5 percent in the endarterectomy group and 1.2 percent in the stent group. Over the next two years however, mortality increased to 20.9 percent in the endarterectomy group and 14.4 percent in the stent group, a rate representative of the comorbid conditions of the subjects enrolled in the trial.

Of note, the median survival for the stented patients was 8.5 years and for the endarterectomy patients was 5.0 years. The cause of death is broken out here. There were 33 total deaths over the first year of follow up; 21 in the endarterectomy group and 12 in the stent group. Only four of the 33 deaths were tied to a neurological event; three in the endarterectomy group and one in the stented group.

By far, non-neurologic deaths predominated. Twenty of the 33 deaths occurred as a result of other causes. Those other causes are broken down here. At the bottom of the slide, cardiac causes were the most common occurring in 18 of the 29 cases of non-neurologic death. Other causes are listed here without significant differences between the two treatment arms.

The complications in the randomized stent and endarterectomy subjects are listed here. Target lesion revascularization was performed in 0.6 percent of the stent group and 3.6 percent in the endarterectomy group, a difference that did not attain statistical difference. Vessel thrombosis, defined in the protocol as angiographically confirmed occlusion, was not documented in either group.

Major bleeding occurred in similar numbers of the stented and endarterectomy patients, nine and ten percent respectively. There was a greater number of cranial nerve injuries in the endarterectomy group; 4.9 percent and not unexpectedly zero in the stented patients, a difference that was significant at the 0.01 level.

The rate of restenosis, defined in the protocol as 50 percent or greater, was 19.7 percent in the stented group and 31.3 percent in the endarterectomy group, a difference that just missed statistical significance. But using more clinically applicable definitions of greater than 70 or 80 percent diameter reduction, the rate of restenosis was much lower. Using the 80 percent threshold, the rate of restenosis was 0.8 percent in the stent group and 4.2 percent in the endarterectomy group, again, a difference that did not attain statistical significance.

Clinically driven target lesion revascularization, which for all intensive purposes represents a result of critical restenosis, this occurred with very similar frequency to the presence of an 80 percent or greater stenosis. Well, we showed you an intent to treat analysis. But a small number of patients never underwent treatment.

Therefore, it=s interesting to present the outcome of the patients who were actually treated with a specified modality. The reasons subjects did not receive specified treatment included ineligibility found after the patient had been randomized, withdrawal of consent prior to treatment, and deterioration in the patient=s condition prior to treatment.

Interestingly in the treated patients, the frequency of major ipsilateral stroke and MI was significantly higher in the endarterectomy treatment arm. In the treated patients, by Kaplan-Meier analysis, the one year major adverse event rate was 20.1 percent in the endarterectomy group versus 12.0 percent in the stented group, a difference that was statistically significant by the log rank test with a p-value of 0.048.

We=ll move on to data from the 406 patients in the non-randomized stent arm, patients that met the criteria for inclusion but for whom the surgeon felt open surgical repair carried an unacceptably high risk. Initially, the intent was to compare data from the non-randomized stent arm to an objective performance criteria or OPC from the literature.

The pre-specified OPC was 12.94 percent. This was not met. In fact, from an evaluation of the data from the SAPPHIRE randomized carotid endarterectomy arm, it had been underestimated. The true MAE was 19.2 percent. The Agency was consulted in March of last year. A supplemental non-inferiority was suggested using data from the SAPPHIRE endarterectomy group and adjusting for differences in baseline demographics.

A propensity analysis was necessary because of the higher rate of comorbidities in the non-randomized stent group compared to the endarterectomy group with a statistically high rate of Class 3 or 4 CCS patients, previous neck radiation therapy, high cervical lesions, prior endarterectomy, and prior stroke.

These three Kaplan-Meier curves demonstrate the rate of MAE up to 360 days. Despite a higher severity of illness in the non-randomized stent group, outcome was as good or possibly better than that of the randomized endarterectomy treatment arm. In fact, the curve fell midway between the randomized stent and the randomized endarterectomy outcomes.

This is the analysis the Agency suggested. The outcome of the non-randomized stent group was non-inferior to that of the randomized endarterectomy group. The confidence interval falls just below the three percent delta that was pre-specified with a p-value of 0.05.

Looking at complications, the rate of target lesion revascularization and cranial nerve injury was significantly lower in the non-randomized stent arm. The rates of vessel thrombosis and major bleeding were similar in the two groups. Given the small number of patients in the non-randomized endarterectomy arm, data will not be covered.

While we will present data from subgroup analyses, the study was not powered for such analyses. I will now present data from the symptomatic and asymptomatic cohorts numbering 96 and 237 respectively. The 30 day MAE rate in the asymptomatic endarterectomy in red and asymptomatic stent patients in blue is illustrated here. There were no significant differences in any of the individual end points or in the composite MAE rate.

Corresponding data at one year is illustrated here. Again, there were no differences in the rate of the individual end points or in the rate of the composite end point. In each case, however, there were trends in favor of stent over endarterectomy. The p-value for the difference in the MAE rate by Fisher=s Exact high-score test was 0.07.

With Kaplan-Meier analyses of MAE to one year, asymptomatic patients randomized to stent did better than those randomized to endarterectomy, 10.5 percent versus 20.3 percent with a p-value by the log rank test of 0.04. The median survival of the stented asymptomatic patients was 12 years. The median survival of the endarterectomy asymptomatic patients was six years.

Moving on to symptomatic patients, the rates of the individual end points at 30 days were not statistically different in the two treatment groups. Point estimates favored the stented patients for all end points. At one year, similar results were observed without significant differences in any of the end points but with trends towards improvement in the stented groups for each of the end points.

These two Kaplan-Meier curves display the frequency of major adverse events in the symptomatic cohort estimated at 20 percent in the endarterectomy arm and 16.3 percent in the stent arm, a difference that was not statistically significant. Of note, the median survival for the symptomatic stent patients was five years and for the endarterectomy patients 3.5 years.

To assure the technical expertise of the surgeons in the SAPPHIRE trial and to convince ourselves that it was representative of surgeons throughout the United States, we evaluated volume and outcome. The 53 SAPPHIRE surgeons were high volume operators reporting a pre-trial experience averaging 36 carotid endarterectomies per year with a median of 28 endarterectomies per year.

This histogram depicts Medicare data from Wennberg published in JAMA about six years ago. Dividing the number of endarterectomies a surgeon performs into terciles, the lowest tercile performed, the cut off, was less than six carotid endarterectomies per year. The middle tercile was defined as between seven and 21 endarterectomies per year. The highest tercile was more than 21 endarterectomies per year.

As you can see from Wennberg=s data, the mortality rate for carotid endarterectomy decreased from 2.5 percent for surgeons performing less than seven cases annually to just over 1.5 percent for those Medicare surgeons performing more than 21 cases annually. Same data here but now adding the pre-trial volumes of the SAPPHIRE surgeons below the X axis.

With few exceptions, the SAPPHIRE surgeons= prior volume placed them in the highest tercile of experience. One index of surgical expertise is the rate of cranial nerve injuries. Despite the inclusion of re-do endarterectomies in the SAPPHIRE data set, the rate of cranial nerve injury was similar to both NASCET and the VA cooperative studies, studies that did not include repeat carotid endarterectomies.

To evaluate the SAPPHIRE surgeons= outcomes, the rate of 30 day ipsilateral stroke was used since this was one of the few end points available from each of the trials. Overall in SAPPHIRE, this rate was 1.8 percent. The SAPPHIRE symptomatic endarterectomy patients were compared with NASCET patients. While the numbers are small, the SAPPHIRE rate of zero is certainly no worse than the NASCET rate of 5.5 percent.

Comparing SAPPHIRE asymptomatic endarterectomy patients with ACAS, the rates were also very close, 2.5 percent versus 1.8 percent. These observations suggest that the surgical outcome for SAPPHIRE was quite similar to NASCET and ACAS for the end point of perioperative stroke despite the greater frequency of comorbidities in the SAPPHIRE data set.

We also compared the results of carotid stenting in SAPPHIRE to the outcomes of previously published surgical data. For symptomatic patients, there were no significant differences in the rate of ipsilateral stroke at 30 days between the SAPPHIRE randomized stent group, the non-randomized stent group, and the endarterectomy arm of the NASCET trial. For asymptomatic patients, there were no significant differences in the 30 day risk of ipsilateral stroke in the SAPPHIRE randomized stent arm, the SAPPHIRE non-randomized stent arm, and ACAS.

In symptomatic SAPPHIRE patients, the 30 day rate for all cause mortality was zero in the randomized stent arm and 0.8 percent in the non-randomized stent arm. For asymptomatic SAPPHIRE patients, the 30 day rate of all cause mortality was 1.7 percent in the randomized stent arm and 2.8 percent in the non-randomized stent arm. These data compare favorably with corresponding data from NASCET and ACAS.

In conclusion, the primary end point of the SAPPHIRE trial was achieved. Carotid artery stenting clearly was non-inferior to carotid endarterectomy in high risk patients. In fact, there were trends favoring stenting over endarterectomy with respect to major ipsilateral stroke, myocardial infarction, target lesion revascularization, and restenosis.

Further, there was a significant decrease in the rate of cranial nerve injuries in the stented group. In the symptomatic and asymptomatic subset analyses, there was significant improvement at 360 days in favor of stenting over endarterectomy in asymptomatic patients with a 50 percent reduction in the rate of major adverse events.

The MAE rate was similar in the symptomatic patients treated with stenting or endarterectomy. The risk of ipsilateral stroke in stented patients overlapped the risks from the NASCET and ACAS trials. In other words, the results of the SAPPHIRE trial was in keeping with previously published data.

With respect to the non-randomized carotid stent arm, there appeared to be risk factors that identified patients that may be at too high risk for endarterectomy. These risk factors were anatomic, medical, or both.

Interestingly, the patients entered into the non-randomized stent arm because the surgeon considered them to be at too high risk for endarterectomy had outcomes that were not inferior to the randomized endarterectomy patients even though the stented group had significantly more comorbidities. This was true for both the symptomatic and the asymptomatic patients. I would now like to reintroduce Dr. Sid Cohen to continue with training and post-marketing surveillance.

DR. COHEN: Thank you, Ken. I=d like to take the next couple of minutes just providing an overview of the training program that we=re proposing to undertake as well as the post-marketing surveillance study and finish with conclusions. The carotid artery stent training system is intended to build upon existing catheter-based expertise to develop the physician=s knowledge and technical abilities in performing carotid artery stenting.

The system was developed using a variety of experts including SAPPHIRE investigators, experts in Internet-based training, experts in simulator modeling, and experts in proficiency measurements. The process of this education encompasses five steps that are pretty traditional but with some modernization.

It includes an online didactic session, observation of actual cases, simulation using a simulator, a proctoring system, as well as training of adjunctive staff in performing the procedure. These trainings occur for the didactic at Internet delivery, for observation and simulation using regional education centers, for the proctoring network and staff training on-site training at the physician=s facility.

What=s unique here is that we have included very importantly a measurement of proficiency that occurs at each step to ensure that high quality patient outcomes would be generated from physicians trained in this system. For the online didactic training, the goal is to transfer expert knowledge through doing and decision-making as opposed to just reading. The goal is to ensure procedural success, providing a detailed understanding of carotid anatomy and brain anatomy, appropriate selection of cases, and high performance in terms of technical execution of the procedure.

Training at the regional educational center occurs in a small group setting where four modules are reviewed over two days. This includes both didactic presentations, observation of actual cases, simulation lab using a simulator, and a product lab to gain familiarity with the products used in carotid artery stenting. The physicians interact with realistic graphical simulations. Their task performance is formally assessed. The understanding of learning objectives is demonstrated.

On-site training at the physician=s facility by physician proctors utilizes a network of physicians who are experienced in performing carotid artery stenting using the Cordis system. These people act as proctors. The proctors either sign off the training and experience an application is adequate or suggest additional training recommendations in order to meet minimal proficiency standards.

The training program encompasses a total of 34 hours of training with exposure to a minimum of 15 cases. This serves as the foundation for hospital credentialing. In order to demonstrate outcomes of this training system in an earlier form, I would like to present outcomes from investigator IDE studies that were performed independent of Cordis but whose investigators were trained using an earlier version of this training system.

These investigator IDEs occurred at 36 centers, 30 of whom were non-SAPPHIRE investigators. All the investigators were trained and proctored on the use of the stent and the emboli protection system. Patient selection criteria was similar to that of the U.S. FEASIBILITY study. The neurologists evaluated the patients at 24 hours and at 30 days post-procedure. The data that I will be showing you is site-reported and unadjudicated.

Thirty day event rates, again site-reported, included a rate of death of 0.6 percent, stroke 2.6 percent, MI 1.4 percent yielding a major adverse event rate of 4.3 in 491 patients. Comparison of these outcomes with the data previously presented from CASCADE study in green, FEASIBILITY in yellow, SAPPHIRE in blue with the institutional IDEs in red shows that outcomes for both stroke as well as for death are very similar.

I=d like to move on now to the post-marketing surveillance study that we=re proposing to undertake. The goal here is to compare clinical outcomes with historical control data from SAPPHIRE in the early time period following approval and assess the effectiveness of the training program. It=s designed as a multi-center, prospective, non-randomized, open label study with a 30 day composite end point where major adverse events are defined as all death and all stroke.

Patients included will be those at high risk with de novo or restenotic lesions. We plan to enroll at least 1,000 patients with the inclusion criteria matching the labeled indications. Follow up will include neurologic exams at discharge and at 30 days performed by a neurologist and clinical events tracking through discharge by a 30 day office visit and a nine month telephone contact. There also will be monitoring with a stopping rule to ensure safety with electronic data capture to expedite review of outcomes.

I=d like to provide a summary and conclusions to this presentation. What we have discussed is that stroke is a disease that has significant morbidity and mortality. It=s due to carotid disease in up to 30 percent of patients. The goal is to prevent stroke and improve the quality of life.

Carotid endarterectomy is the current interventional standard of care for NASCET and ACAS eligible and ineligible patients, for symptomatic and asymptomatic patients, as well as for low, intermediate, and high risk patients. We acknowledge that there are no multi-center randomized studies that define outcomes in high risk medical or surgical risk patients. However, SAPPHIRE is intended as an objective comparison of carotid endarterectomy, the current interventional standard of care, with carotid artery stenting, a less invasive approach to therapy.

Again, Cordis is seeking an indication - I will not read this but summarize it - for use of the PRECISE Nitinol Stent System in conjunction with the ANGIOGUARD XP Emboli Capture Guidewire for use in the treatment of carotid artery disease in high risk patients with symptomatic patients having at least 50 percent atherosclerosis stenosis, asymptomatic at least 80 percent atherosclerosis stenosis with the symptomatic and asymptomatic patients having at least one of the conditions, either anatomic or medical comorbidities that place them at high risk.

This indication is supported by data that we=ve presented from the SAPPHIRE trial where we achieved our primary end point of non-inferiority of carotid artery stenting to carotid endarterectomy for the end point of major adverse events at one year with carotid artery stenting, improving outcomes in terms of reducing myocardial infarctions, reducing the need for reinterventions, and providing a statistically significant decrease, actually an absence, of cranial nerve injuries. We also have provided data in the supportive studies that the benefit of treatment is durable with data that we=ve presented with up to three year follow up.

Cordis will institute a training program to ensure that the outcomes of carotid stenting in the non-trial setting replicates the safety and effectiveness demonstrated in the SAPPHIRE trial. We will conduct a post-marketing surveillance study with the goal of quantifying patient outcomes and confirming the adequacy of physician training. Thank you very much. I would be happy to answer any questions.

CHAIRMAN LASKEY: Well, first of all, bravo for staying within the dreaded yellow and red lights. That was an excellent presentation from both of you. Realizing that each panel member will have an opportunity to query again this afternoon and that we=re coming up to a short break, are there particular areas of clarification that we can try and resolve now? Dr. Aziz.

DR. AZIZ: Just for clarification, once the stenosis was diagnosed by ultrasound, did the patient have an angiogram as well before surgery was done?

DR. COHEN: For the patients who received carotid stenting, obviously an angiogram was undertaken. For the patients who underwent carotid endarterectomy, no angiogram was required. A minority of patients actually underwent angiography because of the dangers of angiography.

DR. AZIZ: Interesting.

CHAIRMAN LASKEY: Tony.

DR. COMEROTA: Dr. Cohen, that was a very elegant presentation. Both you and Dr. Ouriel did it beautifully and very convincingly. In the FEASIBILITY study, could you tell us how many patients were symptomatic and how many were asymptomatic and how many had atherosclerotic disease and how many had recurrent stenosis?

DR. COHEN: I would need to check the data tables to be sure. My memory is that over 60 percent were symptomatic. I do not know that we gathered data on how many were native de novo lesions versus restenotic, but we can check on that and get back to you.

CHAIRMAN LASKEY: One question I had for Dr. Ouriel I guess. With respect to the surgical arm, was there a standardization of the surgical approach, i.e. general versus local? How was that decided? What was the standard surgical approach?

DR. OURIEL: Well, actually, it was left up to the surgeons. So we did not dictate that a surgeon had to use a patch or not use a patch, a shunt or no shunt, or general versus local anesthesia. I can tell you that most procedures were done with a patch and under general anesthesia.

DR. AZIZ: So none of them had an eversion endarterectomy. They had the standard endarterectomy.

DR. OURIEL: No, that=s not necessarily true. I don=t have those numbers, but again, it was left up to the surgeon. In fact, there were some cases that had vein patches, some prosthetic patches. Some re-do endarterectomies had a saphenous vein short interposition graph. So it was left up to the discretion of the operating surgeon.

DR. COHEN: If I could answer the question that was asked before for previous carotid endarterectomy with recurrent stenosis, that occurred in 22.4 percent in the patients in the FEASIBILITY study.

CHAIRMAN LASKEY: Okay, well, thank you both again. Let=s take a rigorous ten minute break. We=ll see you back in ten minutes. Off the record.

(Whereupon, the foregoing matter went off the record at 11:03 a.m. and went back on the record at 11:26 a.m.)

CHAIRMAN LASKEY: On the record. If we can all regroup again please. Thank you all very much for your compliance, another watchword. We would now like to proceed with the Agency=s presentation.

MS. KENNELL: Good morning, panel members and audience. Our FDA presentation --

MS. WOOD: Lisa, pull the mic a little closer.

MS. KENNELL: Thank you. I=m trying to juggle the laptop as well. Our FDA presentation will involve three presenters. I will be presenting some background information and comments about the non-clinical information in the file.

Our statistician Heng Li will present several slides detailing statistical issues and conclusions. Dr. Ronald Weintraub, a consultant to FDA on this project, will discuss issues relating to the clinical study. We have a substantial number of difficult questions for panel discussion, so I want to move through our presentation as quickly as possible.

I would like to acknowledge the people who helped me on this project. I had three engineers and three clinicians who provided input as well as Dr. Li, the statistician. I reviewed the remainder of the information in the submission as well as coordinating the reviews from the team members.

The next several slides detail configurations and sizes of the stent and embolic protection device that the sponsor proposes to offer for sale. The OTW, over the wire, configuration will be offered in either 6 or 5.5 french profile with the larger profile being for the larger stent diameters. Stent diameters in the OTW configuration will range from 5 to 10 millimeters in both tapered and straight configurations.

The sponsor also makes an RX, rapid exchange, configuration that is compatible with a 0.14 inch guidewire rather than the 0.18 inch needed for the OTW version in the same sizes as the OTW minus the tapered. However, due to a recent development, we are not considering this configuration today.

Similar to the stent, the ANGIOGUARD XP Emboli Capture Guidewire will also be made in both an OTW and an RX configuration. Filter sizes in both configurations will range from 4 to 8 millimeters. Again, the RX configuration will not be considered today.

There have been some recent developments relating to the RX configurations. The sponsor submitted an unsolicited amendment to the PMA just two weeks ago which proposed a change in the Instructions for Use for these devices. What prompted this submission were complaints received by Cordis relating to air being entrained in the RX configuration when used off-label in carotid and other indications.

While many of these instances resulted in no injury to the patient, there were a few that resulted in adverse events from air embolism. This rate has been increasing and is not up to an estimated 0.14 percent.

Cordis investigated these events to try to determine the root cause followed by some testing on the bench to try to simulate this problem and correct it. The problem seems to occur in the RX configuration because of the tolerance and the length of the pod in the RX. We are concerned that the bench testing performed by the sponsor to date is not optimal because saline was used in the testing and the viscosity of saline is different than that of blood.

We believe that additional animal and possibly clinical testing may need to be performed. After this slide was finalized, Cordis called to indicate that animal testing had been performed but it was not included in the amendment for review. Based on the bench and animal testing, the sponsor has proposed stipulating larger guiding catheters for introducer sheaths and more detailed instructions for preparing the delivery system.

FDA will continue to work with the sponsor to resolve this issue since we believe that further testing is probably warranted and the results of the animal studies will need to be submitted and reviewed. The indication sought by the sponsor is provided on this slide and provides options for patients with and without symptoms and stipulations relating to degree of stenosis as well as comorbidities making them a high risk for surgery.

The IDE for this device has had a long history beginning in 1998. Since the first submission, there have been many changes made to the device, materials of construction, sizes, and profile, with the most significant being the introduction of the ANGIOGUARD Embolic protection device during the latter part of the FEASIBILITY study, lowering the profile of the device, and the development of the RX configuration.

The sponsor terminated their randomized study early and gave the reasons detailed in this slide. Regarding the first bullet, these competing studies were facilitated by Cordis in that Cordis provided each investigator with a letter of authorization to allow FDA to access the Cordis file for background information. Cordis also supplied most, if not all, of the single investigators a copy of their FEASIBILITY protocol, which was a registry design, and the case report forms and consent that was developed for that study.

Most opted to follow this protocol with little modification, but Cordis was not privy to interactions between these single investigators and FDA. Most investigators were approved to perform somewhere between 50 and 100 carotid artery stenting cases.

While there was no contractual relationship between Cordis and these single investigators, a sponsor is required under the PMA regulation to provide FDA with all data known to them or that should be known to them. So Cordis coordinated with most of these investigators to obtain their 30 day data for inclusion into the PMA. It should be noted that all of these studies stipulate a minimum of 12 month follow up.

As stated earlier, there have been many design changes made throughout the history of the file. For each change the sponsor made, they provided testing appropriate to the specific change. Testing has included fatigue, simulated use, device specification, and integrity testing sometimes on the bench, sometimes in animals, sometimes in both.

We met with the firm just prior to the submission of the PMA and agreed that the RX configuration of the stent and ANGIOGUARD could be approved without clinical data since the working ends have not been modified. In the original animal testing results submitted prior to the proposed fix of using larger guiding catheters or sheaths, one comment in the report was that a larger guide would be needed for the 10 by 40 centimeter size.

While we did not suspect a problem with this statement, the new developments in humans, with devices being used off-label, may warrant further evaluation clinically and we may have to reconsider our agreement. To date, the engineering reviews have been completed, at least those performed prior to the proposal for larger guiding catheters and sheaths, as have biocompatibility. These are considered adequate and complete.

The review of the sterilization validation is not yet complete, but no major issues are anticipated from that review. As noted earlier, each additional bench, animal, and possible clinical data may be needed to fully validate the RX configuration.

Another non-clinical issue has arisen recently. The FDA issued Cordis a corporate warning letter on April 1, 2004. Our investigators noted many serious non-compliance issues with respect to the current Good Manufacturing Practices requirements. The letter quoted that there were systemic problems noted at many facilities.

Obviously these are going to need to be rectified before approval can be granted for the PRECISE and the ANGIOGUARD. I=m going to now turn over the podium to Dr. Heng Li, the FDA statistician for a brief discussion of the statistical issues.

DR. LI: In my presentation, I will make a few comments on the SAPPHIRE randomized trial, the SAPPHIRE stent registry, and the statistical procedure of propensities for analysis followed by some concluding remarks. First, let me talk about the SAPPHIRE randomized trial.

This randomized clinical trial, whose objective is to compare stenting with carotid endarterectomy, had a group sequential statistical plan in which the sequential triangular test was used. Unlike a fixed sample sized plan, the group sequential design does not pre-specify the sample size of the trial. Instead, the design establishes a set of stopping rules and schedules a series of interim analyses.

At each interim analyses, the available data are examined and the trial is either stopped or continued according to the stopping rules. The stopping rules and the schedule of interim analyses are specified so as to control the frequent error probabilities at their intended or specified levels. For the SAPPHIRE randomized trial, the interim analyses are scheduled at intervals of every 100 patients.

The maximum sample size was specified to be 2,400. In the original protocol, it was expected that the trial would be stopped at a sample size of at least 600. I will come back to discuss the original group sequential trial in some more detail for the SAPPHIRE randomized trial in a minute.

But before I do that, let me point out that the sponsor made no claims that the original group sequential protocol had been followed. As a matter of fact, in the current PMA submission, the message appeared to be that the original group sequential statistical plan had not been followed. In particular, the scheduled interim analyses had not been conducted. However, FDA was not aware of any change in the statistical plan.

In the current submission, data from SAPPHIRE randomized trial were used to make the declaration that stenting is non-inferior to CEA. This declaration is based upon statistical inference. We know that statistical inferences for design studies need to be made according to the statistical plan in the current study protocol. Otherwise, they are unplanned.

Therefore, the statistical inferences in the current PMA submission that led to the declaration of non-inferiority of stenting relative to CEA based on the SAPPHIRE randomized trial is unplanned since it made reference to a statistical plan that is not in the current study protocol, namely a fixed sample size design based on 334 patients, the number at which the trial happened to be discontinued. We all know that statistical inference based on unplanned analyses are less reliable. So we don=t think it=s necessary that the inference be based on an unreliable, unplanned analysis.

Now, let us describe the original group sequential protocol in a little more detail using a picture. In this picture, the label of the horizontal axis V represents the amount of information that has accumulated at a given time or before a given time. The label of the vertical axis Z represents the difference in treatment effect as reflected in the data at that time.

At any stage of the clinical trial, the values of V and Z can be calculated from the available data. So we can imagine that as the trial progresses, it traces out a sample path on the V-Z plane. Because the amount of information increases as more data became available, the sample path goes from left to right starting at V equal to zero.

As it moves to the right, it may wander up and down. The triangle on the graph defines the stopping rule in the original group sequential plan for the SAPPHIRE randomized trial. If the trial is continuously monitored, then when the sample paths cross one of the triangular boundaries the trial is stopped.

If the sample path crosses the upper triangular boundary, then the trial is stopped and the non-inferiority tested. When the sample path crosses the lower triangular boundary, the trial is stopped and no non-inferiority can be claimed. As long as the sample path is within the triangular region, the trial continues.

The inner boundaries are called the Christmas tree boundaries because of their shape. These reflect the adjustment necessary for discreet monitoring. Since the scheduled monitoring is discreet with interim analysis planned for every 100 patients, the Christmas tree boundaries would have been used for the SAPPHIRE randomized trial.

This slide summarizes the stopping rules described earlier according to the original protocol. For the SAPPHIRE randomized trial, based on the data contained in the PMA submission, we can calculate Z and V as mentioned before. So we can plot a point in the Z-V plane. Of course, this single point wouldn=t tell us what would have happened had the original group sequential protocol been followed.

In the sponsor=s presentation, there is one slide that contains very valuable information which is the attempt of reconstructing the group sequential plan or a reconstruction of what would have happened had the original protocol been followed to the best approximation. As far as I=m aware, this information wasn=t contained in the PMA submission. As far as I know, it wasn=t submitted. If it was submitted, it wasn=t submitted before 4:00 p.m. yesterday. I=m looking forward to reviewing this very valuable information in the near future.

Let me now turn to the SAPPHIRE stent registry. For the SAPPHIRE stent registry, the predefined objective performance criteria is to reject a null hypothesis that the 360 day major adverse event rate is greater than 16.94 percent. The observed 360 day major adverse event rate is 15.76 percent.

A 95 confidence interval for a 360 day major adverse event rate has a lower bound of 12.36 percent and an upper bound of 19.68 percent. The upper confidence limit, 19.68 percent, exceeds the OPC of 16.94 percent. The pre-specified OPC has not been met.

After realizing that the OPC is not met, the sponsor made unplanned comparisons between the stent registry and CEA arm of the randomized study. Since the patient characteristics of those two groups by the nature of how they are assigned are not necessarily the same, a straightforward comparison as was conducted in the current PMA submission is not appropriate.

To address this issue, the sponsor used a propensity score method to compare the two groups attempting to make post-hoc claim of non-inferiority of stent registry to randomized CEA. The phrase Apropensity score method,@ as it is commonly used, refers to a class of statistical procedures that can help evaluate difference in treatment effect when the treatment groups are not necessarily comparable by balancing a set of chosen covariates.

It works by first introducing a model for the probability of a subject being assigned to one of the treatment groups given the values of the covariates. This probability is called the propensity score, hence the name propensity score method. The issue of missing data in the modeling, the issue of missing covariates in propensity score modeling could be addressed by multiple imputation.

The result of propensity score modeling is that each subject is assigned a propensity score. One way of using propensity score analysis to compare treatment effects after a propensity score is calculated for each subject is to divide the patients into five strata according to their propensity scores.

The first stratum consists of patients with propensity scores in the top twentieth percentile and so on all the way down to the last stratum consisting of patients with propensity scores in the lowest twentieth percentile. It turns out that all the covariates included in the propensity score model could be simultaneously balanced to a great extent within each stratum. Therefore, bias due to imbalance of those covariates could be removed to a great extent when treatment comparison is made within each stratum.

The potential of being able to simultaneously balance a large number of covariates to a great extent is a very attractive feature of the propensity score method. However, the sponsor may not have taken full advantage of the propensity score analysis in carrying out their propensity score analysis.

The potential issues include not all observed, clinically relevant covariates were included in the propensity score model, not all patients are included in the treatment comparison, and of course the analysis itself is unplanned. Given the sensitivity of the results to any improvement of methodology, this is all the more of a concern.

Now, let me get to the concluding remarks. In conclusion, original group sequential protocol was not followed and FDA was not informed of any change in protocol for a new statistical plan. Evidence of non-inferiority under original group sequential protocol was not supplied in the current PMA submission.

For the stent registry, it fails to meet a pre-specified operating OPC, objective performance criteria. Any non-inferiority claim based on the sponsor=s post-hoc propensity score analysis is problematic for the reason mentioned above. Now, let me turn the podium to Dr. Weintraub.

DR. WEINTRAUB: Good morning. The SAPPHIRE pivotal clinical study was designed as a multi-center randomized group sequential study studied by intention to treat as a comparison between patients undergoing open operative carotid endarterectomy and those being treated with the carotid angioplasty in the Cordis PRECISE stent system. In addition to the 334 patients randomized to the stent and endarterectomy arms, there was an additional stent registry cohort.

This presentation will examine the comparative results of the randomized arms, compare subgroups within the randomized arms, examine the results of the non-randomized stent registry, and look at the relative clinical effectiveness of the stent and endarterectomy techniques. Finally, a brief survey of historical randomized trials comparing endarterectomy with medical management of carotid stenosis will be introduced as a frame of reference.

In order to be considered for enrollment in the SAPPHIRE study, patients were required to be considered high risk by a neurologic or anatomic criteria in addition to having one or more technical or medical comorbid features considered to present high risk for carotid endarterectomy. These include the following and are defined in greater detail in your panel packs. The sponsor also introduced these in detail earlier this morning.

There were a number of exclusion criteria which are also enumerated in these next two slides and in detail in your panel packs and again were discussed by the sponsor earlier. I=ll just take a second so you can look at them.

There were 167 patients in each arm. These serve as the basis of comparison in the pivotal trial. In the non-randomized registry, 406 patients met inclusion criteria but were determined by the surgeon at each site to be at too high a risk for carotid endarterectomy and inappropriate therefore for randomization.

First of all, I skipped a page obviously so I=m going to have to go back. The names of the various laboratories are provided in your panel packs. Data analysis was performed by the Harvard Clinical Research Institute which also provided the adjudication committee.

The primary end points. Please note that the composite of major adverse events at 30 days post-procedure includes myocardial infarctions. These are not included unless they were fatal in the historical randomized trials comparing endarterectomy and medical therapy. The second primary end point consists of the composite 30 day major adverse events plus death and/or ipsilateral stroke at one month to a year.

Secondary end points are listed in the next two slides. Again, they were detailed in your panel packs. For the pivotal randomized trial, 29 centers enrolled patients. A total of 334 patients were enrolled equally divided between stent and endarterectomy. Five centers however enrolled the majority of patients. The one year major adverse event rates are listed for those five centers in the slide shown here.

Let=s look at the primary events. The primary end point of 30 day adverse event rates for the randomized stent and endarterectomy arms are displayed here. Please note the confidence limits in the far right columns. Let me review for the panel as well as for myself that if the limits embraced by the brackets encompass zero, the values of the two arms are not considered to be statistically different. As shown here, several pairs of data points approach but do not reach statistical significance.

Here are the one year or 360 day major adverse event rates. There are no statistically significant differences between the two groups. Moving forward, we look at the two year major adverse event rates and again no statistical difference between the two groups.

In this slide, the data are presented in several ways. First, both randomized and registry data are presented. They are also divided into neurologically symptomatic and asymptomatic cohorts. Finally, the 30 day major adverse events are scrubbed of their non-fatal myocardial infarctions making them more comparable to historical randomized control trials. That=s done in this third column.

Again, there=s no significant difference between the randomized arms. The incidents and severity of myocardial infarctions in the randomized stent and endarterectomy arms was examined. It did not differ significantly. The same data for the registry are displayed though no formal statistical comparison can be made. Again, no differences.

As previously mentioned, a total of 406 patients were entered into the stent registry. The sponsor states that they were entered at the choice of the surgeon investigators who felt they were too high risk for randomization. On this slide are represented the reasons stated for surgical turndown. Note that the reasons were not enumerated in 50 percent of the patients. Please also note that approximately 70 percent of the patients were neurologically asymptomatic.

Represented here are the 30, 360, and 720 day adverse event rates among the registry patients. Take note especially, if you would, of this figure of 15.8 percent for major adverse event rates at one year. To briefly review Dr. Heng Li=s analysis, objective performance criteria were set at 12.94 percent.

This figure was determined by calculations derived from review of the literature of randomized, controlled, endarterectomy trials as well as site reviews of the relevant study populations of the SAPPHIRE trial. Since the original delta of four percent was specified, the null hypothesis assumed a 360 day major adverse event rate of the 12.94 percent plus four percent of 16.94 percent or higher.

The observed rate in the trial however was 15.76 percent as presented in the previous slide. Therefore, the sponsor could not reject the null hypothesis. In other words, the criterion for non-inferiority was not met. It=s clear that the propensity score method has not been thoroughly explored. Questions remain about the adequacy of analysis.

Let=s return to the randomized control study. The following slides show the patients subdivided into neurologically symptomatic and asymptomatic cohorts. Examination of the 30 day major adverse events demonstrated non-inferiority of the stent with respect to endarterectomy. I direct your attention to the number of patients in the symptomatic trials. Here=s the 30 day adverse event rates. Subgroup analysis at 360 days also showed non-inferiority of the stent.

Turning to the asymptomatic randomized patients, they were also compared with respect to 30 day major adverse event rates. Also note the rather larger number in these cohorts. There were over twice as many asymptomatic as symptomatic patients in randomized pivotal trial. Once again, there were no significant differences between randomized stent and randomized endarterectomy groups at the 30 day mark.

Results at one year were similar. There were no differences between the randomized stent patients and those who underwent endarterectomy, although the superiority of stent approached significance - you see the zero - at about the 0.07 level.

Subgroups other than symptomatic and asymptomatic were examined. Here are displayed male and female sex and diabetes in the randomized stent and endarterectomy arms and in the registry at 30 and 360 days. At 30 days, major adverse events in the diabetics occurred more frequently in endarterectomized patients actually reaching statistical significance with a p-value of 0.03.

One year major adverse events occurred more frequently in males almost reaching statistical significance. The subgroup of elderly patients was examined but the numbers were relatively small. No differences were noted. Recurrent stenosis occurred with similar frequency in all the groups.

Secondary end points were reviewed. In this table, lesion, procedure, and device success and protection for all patients randomized or selected for stent are displayed. Success defined by these various parameters ranged between 88 and 96 percent. The column to the far right represents the individual investigator sponsored trials.

Other secondary end points are shown in here in the different cohorts; trapped material, freedom from lesion restenosis, and freedom from major adverse events at one year. There=s probably no other surgically treated disease entity that has been studied so thoroughly in randomized control trials than carotid stenosis. Thus far, carotid endarterectomy has been compared with optimal medical therapy in a series of trials over the period of a decade or more.

Because the SAPPHIRE study is itself a randomized study comparing a new technology with carotid endarterectomy as the gold standard, it would be appropriate at this point to consider conclusions derived from those historical randomized studies which compare to endarterectomy with the then standard medical therapy. The best known studies are the asymptomatic carotid artery atherosclerosis study called ACAS which looked at patients with asymptomatic stenosis greater than 60 percent and the Veterans Administration study of asymptomatic males with greater than 50 percent stenosis.

The ECTS European study examined symptomatic patients. The NASCET, North American Symptomatic Carotid Endarterectomy Trial, also studied symptomatic patients. In the interest of time, I have condensed the conclusions of these several excellent studies into three slides. If the panel were to find it germane, we have more detailed slides available for the discussion period.

Here are the conclusions. For symptomatic patients, high grade stenosis carotid endarterectomy was very effective, greater than 50 percent reduction in the risk of stroke and any death at two years. Not only that but risk reduction varied with stenosis, that is, the greater the stenosis, the greater the risk reduction of operation.

For moderate stenosis, success was less certain. It was calculated that 23 operations were required to prevent each severe ipsilateral stroke at five years. Not only that but each two percent increase in 30 day perioperative event rate reduced the five year benefit by 20 percent.

In the asymptomatic patients with greater than 60 percent stenosis, endarterectomy was very effective with approximately a 50 percent reduction in the risk of ipsilateral stroke or perioperative stroke or death if the procedure could be performed with a perioperative, major adverse event rate of less than three percent. Not only that but angiography alone entailed a risk of stroke of 1.2 percent.

These conclusions are consonant with the American Heart Association guidelines for stroke prevention and for guidance for the appropriate use of endarterectomy published in 2001 and 1998 respectively. Finally, there are data that caution us about the indiscriminate employment of endarterectomy.

First, the risk of stroke in asymptomatic patients is statistically low. Second, in a study of NASCET patients who had asymptomatic stenosis in the artery contralateral to the symptomatic side, 45 percent of subsequent neurologic events were of lacunar or cardioembolic etiology. Both of these diagnoses are most common in those patients who are elderly or who have major medical comorbidities.

Finally, the application of mechanical technologies in patients with a limited life expectancy should be approached quite cautiously. In the following four slides, the respective symptomatic and asymptomatic historical randomized control data for endarterectomy cohorts are appended to the corresponding SAPPHIRE cohort.

Because the historical studies excluded non-fatal myocardial infarction in the definition of major adverse events, as pointed out by the sponsor, these have been scrubbed from the SAPPHIRE data or a range given where it was not entirely possible to get the exact numbers. I gave a range.

Some historical study data have been estimated for Kaplan-Meier curves. As you look at these tables, please understand that they are being offered merely as a frame of reference and not as a scientific comparison. Consider also the relative size of the cohorts.

For symptomatic patients, the NACSET cohort illustrated here is for the patients with high grade stenosis. The recruitment of these patients was discontinued at a mean follow up of 2.7 years because of the demonstrated superiority of endarterectomy compared to medical management. These are the same cohorts presented at the one year point. Remember these are symptomatic patients in the randomized control study stent arm and the carotid endarterectomy arm.

Moving to the asymptomatic patients, here the asymptomatic SAPPHIRE patients are displayed and juxtaposed to the ACAS asymptomatic trial patients. The same cohort groups are followed to one year. This completes the data presentation.

What are the limitations of the sponsor=s study? The pre-specified enrollment plan and study analysis was not carried to completion in the SAPPHIRE randomized study. This resulted in a smaller size study with small sample sizes in important subsets of carotid populations.

But what can we conclude? The randomized study suggests non-inferiority of the stent to the carotid endarterectomy. Registry cohort failed to meet the OPC pre-specified criteria. The comparability of the registry to the control endarterectomy patients has not been optimally defined or conducted. Thank you.

CHAIRMAN LASKEY: Are there questions from the panel to any of these three presentations? Mitch.

DR. KRUCOFF: A question for Dr. Li. I just want to make sure that I understand what you said with regard to your triangular method, Christmas tree slide that went onto a picture with a point with a dot on it. That was your sixth slide. So is that dot your data assessment from the randomized cohort at 334 patients that lie within the boundaries?

DR. LI: Right, it=s based on a Z and V value calculated from the 364 patients= data included in the PMA submission.

DR. KRUCOFF: Right, and so the next slide, you have a based on the above graph comment. Is what I understood that you are saying based on the information in the sponsor=s presentation today that was not present by 4:00 p.m. yesterday, would you change this conclusion? Is that what I=m understanding you to say?

DR. LI: Basically the short answer is yes. The sponsor, based on what I understand of the one slide presentation, is trying, to the best possible extent, to reconstruct what might have happened had the original group sequential protocol been followed.

My impression is that, based on the sponsor=s presentation, their conclusion is that had the group sequential protocol been followed then the trial would have been stopped at 300 patients and then subsequently non-inferiority can be declared. But it is important to make a distinction that in actuality the protocol hasn=t been followed as was implied by the presentation.

So it=s impossible to know what would actually have happened, but it=s to a best approximation. For example, had the protocol been followed to every detail, then enrollment, recruitment would have been stopped at the 300 patients, when 300 patients have had the 30 day data, which might not have been the case. So it=s not possible to repeat all the detail. But it=s only a rough approximation and this approximation needs to be verified.

DR. KRUCOFF: I take it you are talking about Dr. Cohen=s slide that had the outcome columns conclusion at 100 and --

DR. LI: Right, 100, 200, and 300.

DR. KRUCOFF: It was continue, continue, stop.

DR. LI: Continue, continue, stop, right.

DR. KRUCOFF: But you at 300 have a dot that=s still within the boundaries which would not imply stop. So I guess what I=m asking is - and understanding you haven=t had the chance to review this yet - do you have any sense of why one retrospective reconstruction of this would say stop and 300 where it appears your retrospective reconstruction of this would say continue at 300?

DR. LI: Okay, so remember when I described the group sequential protocol, I mentioned that the sample path may wander up and down. Although it always go from left to right, it wanders up and down. That dot in the previous slide is calculated at 334 patients. Apparently at 300 patients, if the sponsor=s calculation was right, then the boundary would have been crossed at that time, and it=s completely possible.

DR. KRUCOFF: Okay, thank you.

CHAIRMAN LASKEY: Other questions from the panel?

DR. ABRAMS: I have a question for Dr. Weintraub. In your additional slides, do you have data on the ACAS study for higher grade stenosis of asymptomatic?

DR. WEINTRAUB: I=m having trouble with a touchy mouse. I=m sorry, you asked for NACSET.

DR. ABRAMS: ACAS.

DR. WEINTRAUB: Oh, for ACAS. It should be pointed out in the European trial. The problem with the European and North American data are that the Europeans measured stenosis differently from the North Americans. But in an excellent editorial published at about the time that the final NASCET trial came out, the writer actually compared the two.

The take away message was that at moderate stenoses, which were about equivalent to about 40 to 50 percent stenosis in the American system, that results particularly in asymptomatic patients probably did not favor endarterectomy. Whether that would be different because endarterectomy is not the same as stent, we just don=t know. But certainly on the basis of the historical record, one has to approach moderate to less than 50 percent stenosis with a great deal of caution particularly in asymptomatic patients.

DR. ABRAMS: Is it actually possible from this data to do a subgroup analysis on the greater than 80 percent --

DR. WEINTRAUB: Yes, this was done. Unfortunately, it=s in there somewhere. To answer your question - and we can find that slide at the lunch break for you - in an NASCET, the degree of stenosis was broken down into three subsets.

Interestingly, the greatest effectiveness was in the middle subset. When it got up to 85 to 99 percent, it dropped off a little bit. But there was clearly a separation depending on the degree of stenosis. That was also certainly true in the European study. I think it was true in the ACAS but I would have to look at it again.

DR. COMEROTA: One question. I hate to belabor this point, Dr. Weintraub. And you don=t need to pull up the slide.

DR. WEINTRAUB: Very wise.

DR. COMEROTA: In your analysis, you compared the symptomatic patients with the NASCET group. Obviously we all know that there=s two publications for NASCET; the 70 to 99 percent stenosis and then the less than 70 percent stenosis. You compared the NASCET group greater than 70 percent. They all had arteriographic documentation of that degree of stenosis.

DR. WEINTRAUB: That is correct to my understanding, yes.

DR. COMEROTA: All of the patients in the SAPPHIRE had arteriographic documentation to the degree of stenosis. Was that compared on your slide, the definition of the arteriographic stenosis?

DR. WEINTRAUB: This is a question that I had raised in my original evaluation. I was told by the sponsors that angiography was not used routinely in those patients undergoing carotid endarterectomy.

DR. COMEROTA: Right, I=m talking about the registry and the randomized stent patients.

DR. WEINTRAUB: The randomized stent patients, of course, all had angiographic analysis preoperatively.

DR. COMEROTA: All patients having a stent had an arteriogram.

DR. WEINTRAUB: That=s correct.

DR. COMEROTA: But you are not sure what the distribution of the degree of stenosis was in the stented patients, correct?

DR. WEINTRAUB: I would really have to ask the sponsor about that. It=s in the panel pack because the degree of stenosis was broken down quite exactly. But in terms of grouping them, I can=t tell you that. The sponsor might be able to.

CHAIRMAN LASKEY: Well, if there are no further questions, since my hypoglycemia has taken hold, let=s break for an hour lunch. I have 12:25 p.m. Let=s resume at 1:25 p.m. Thank you. Off the record.

(Whereupon, at 12:22 p.m., the above-entitled matter recessed to reconvene at 1:34 p.m. the same day.)

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

1:34 p.m.

CHAIRMAN LASKEY: On the record. All right. I=d like to thank everybody for coming back. We=ll resume this afternoon=s session starting with Dr. Judah Weinberger=s queries. Judah.

DR. WEINBERGER: Thanks very much. First of all a comment to the sponsor. I think that the data presented is interesting, thought provoking and hopefully we can sort through it and figure out precisely what to do with it.

First an administrative question. I would just like to ask the sponsor to explain why the FDA wasn=t aware of the latest changes in terms of statistical analysis that was pointed out. It=s really hard for me to interpret a study when there=s a disagreement between the FDA and the sponsor as to statistical validity.

DR. COHEN: The answer to your question is that we were made aware that there was a question about the statistical methods being applied last Friday. We really did not have time to prepare materials to send to the Committee beforehand.

DR. WEINBERGER: So it is your position that there was no deviation from the original plan.

DR. COHEN: That=s correct.

DR. WEINBERGER: And it=s the FDA=s position that there was a significant deviation from the original plan. Is Dr. Li here?

DR. ZUCKERMAN: Ms. Kennell, would you come to the podium?

MS. KENNELL: You=re talking to the wrong person when you=re talking statistics, but I believe that that is our understanding about the original plan. I looked up the protocol last night just so I would make sure that it was fresh in my mind. The original protocol said that they were going to do interim analysis at every 100 patients and that was not done.

DR. WEINBERGER: All right. I=m not a statistician, but maybe you can explain to me how not doing an analysis prejudices the interpretation of the study. Rather I would imagine the less you look the more powerful the statistic is because we=re taught that repeated analyses actually require higher B-

MS. KENNELL: I don=t know that I would be comfortable answering that question. I don=t think I have enough statistical expertise. I can try to find out for our statistical session.

DR. ZUCKERMAN: Dr. Weinberger, let=s try to answer your questions. 1. The FDA and the sponsor had a protocol with a prespecified statistical analysis plan that has been reviewed by FDA. That prespecified statistical analysis plan was changed at some point. FDA was unaware of when it was changed and what went into making that change.

Usually we would expect a major supplement and meeting with FDA to discuss a major change in a clinical trial like that. As such because of the events that transpired, it becomes somewhat problematic to interpret the statistics as presented by the sponsor. It doesn=t mean it=s impossible, but it does throw another question mark. Dr. Greg Campbell, our Chief of Biostatistics, is also here and he may want to add a point.

DR. CAMPBELL: Greg Campbell, FDA. Dr. Weinberger, your question is a good question. It=s always a good idea to follow the original plan that a company has to analyze data. When they do not follow that plan or we at the FDA do not follow that plan, we=re doing post hoc analyses and those are not recommended. The issue as to how often you look is a very complicated one in this case and I think I would prefer not to try to comment on it only because it=s different if one looks all the time versus one looks at particular intervals.

DR. WEINBERGER: Okay. I=d like to dig around a little bit in the data if we might and deal with a couple of issues of trying to understand who=s getting the biggest bang for the buck or who is really benefitting in this trial compared to standard therapy. So if we understand the trial, the patients have to have been previously identified by their referring physician as people who need carotid revascularization. This comment is directed at the sponsors. Is that correct? So these are patients who are previously identified by their referring physician as needing revascularization.

DR. COHEN: That is correct.

DR. WEINBERGER: All right. So this is not anyone in the trial identifying the patient. Then that patient gets referred in for revascularization, has an anatomic study, has a clinical stratification and based upon that is eligible for the study.

DR. COHEN: That=s correct.

DR. WEINBERGER: All right. In that group of patients that are the symptomatic ones and the asymptomatic ones, you=ve done stratification of those two groups in your analysis as has the FDA. I=d like to focus on a different sort of a stratification that is among the patients who ultimately ended up in the registry, people who were not operated on because the vascular surgeon decided that this was inappropriate. Some of those patients ended up for anatomic reasons and some ended up for medical comorbidities.

One of the things that=s really counter-intuitive here is a statement that you have in Panel Pack where you said that the people ended in the registry group for anatomic reasons did better with surgery than they did with stenting. Let me read it to you correctly. AThere was significant differences in major adverse events at 360 days when patients were stratified by anatomic and medical criteria.@ I=m sorry. This is in the randomized group. AThe anatomic, high risk patients had major adverse events of 10.3 percent for stent patients and 5.6 percent for carotid endarterectomy patients.@ Now the reason that people are high risk if for anatomic reasons, how is it that they do better with surgery than they do with stenting?

DR. COHEN: One of the figures of the design of this trial if I understand your question correctly was that it allowed the surgeons to take patients that they felt were at high risk and not operate on them. Those patients then went into the non-randomized stent registry. That would be expected then to have the high risk surgery patients in the stent registry as opposed to the surgical arm.

DR. WEINBERGER: Let=s go back to the randomized trial.

DR. COHEN: Okay.

DR. WEINBERGER: In the randomized trial, you get into the randomized trial with having a high risk anatomic feature or high risk medical features.

DR. COHEN: That is correct.

DR. WEINBERGER: Okay. If you have high risk anatomic features, I would expect that you would do better with stenting than you would do with surgery. If you have high risk clinical features, it might be that you would do B- That=s unclear.

But certainly if you had high risk anatomic features, it would appear to be that you would be more likely to expect a good result from stenting. It=s rather counter-intuitive that patients with high risk anatomic features at least not appear to do significantly better, not in a statistical sense, but do much better in terms of major adverse events when they have a surgical approach rather than a percutaneous approach.

DR. COHEN: I=m sorry. Could I ask you to tell me where you=re looking at in the Panel Pack?

DR. WEINBERGER: Sure. This would be in the summary of FDA Review Memo and this is with the marked up memo that we got last week on page 12. That includes your comments.

DR. COHEN: I=m sorry. This is not in the Panel Pack.

DR. WEINBERGER: Geratta, do they have that in their Panel Pack now? That=s what I had Fed Ex=ed last week.

MS. WOOD: They should have that. It was the FDA Memo as edited per the company=s request.

DR. COHEN: I=d like to ask Dr. Ken Ouriel to answer this question.

DR. OURIEL: Okay. If I understood the question.

DR. WEINBERGER: All right. Let=s go through it slowly so it=s a point that=s at least important to me. I don=t know about the rest of the panel. If you take patients who are randomized in this study, they get into the study because they are either symptomatic or they are asymptomatic with high grade stenosis plus they have to have another feature that puts them at high risk.

DR. OURIEL: Correct.

DR. WEINBERGER: And that other feature can be either medical comorbidity or anatomic problems or radiation to the neck or both.

DR. OURIEL: Right.

DR. WEINBERGER: All right. So if you look among patients who get into the trial and you look at those people who get in for anatomic comorbidities primarily, those are the people you would expect would do worse with surgery because the anatomic comorbidities are defined as comorbidities that are important to a surgeon. That is radiation to the neck. That=s important to a surgeon, not to an endovascular therapy.

DR. OURIEL: Well, we know that it=s important for surgery. We didn=t know whether it was important or not for stenting. It may be that radiation arteritis is a high risk for stenting.

DR. WEINBERGER: Okay. Or a contralateral recurrent laryngeal nerve palsy, etc., all those things that push you anatomically to worry about doing that.

DR. OURIEL: Correct.

DR. WEINBERGER: All right. So in that anatomic high risk group, you report a major adverse event rate at 10.3 percent for stented patients and 5.6 percent for carotid endarterectomy. I=m not being critical of the surgeons. In fact, I=m congratulating the surgeons. I=m trying to understand how it is that stenting interacts negatively with anatomic comorbidities.

DR. OURIEL: I don=t have that data in front of me, but I wonder what the confidence intervals are there. The numbers get very small when you get down to the subgroups. I=m just wondering if those two numbers really are similar.

DR. COHEN: Can you tell us where this is again?

DR. WEINBERGER: Sure. This is page 12 of the section that=s labeled Section 4, Summary of FDA Review Memos. It=s in the revision of that section that came by last week. Do you have that? I don=t know what it was in the original document because the page numbers have changed.

DR. COHEN: Let me make a couple comments. First of all, only about 20 percent of the patients in the randomized portion of the trial had anatomic comorbidities. That means that the number of patients we=re talking about is very small. So when you=re then making a subgroup of that subgroup, a standard error on those endpoints alone, I=m sure it overlaps significantly. As you stated, there was no statistical significant difference. If you could point us to the exact sentence that you=re looking at.

DR. WEINBERGER: All right. There=s a page in my book that=s entitled AMinor Deficiencies (FDA Questions).@ Do you have that page?

DR. COHEN: Yes, this is page 12?

DR. WEINBERGER: This is page 12.

DR. COHEN: Questions 1, 2 and 3?

DR. WEINBERGER: Questions 1, 2 and 3. If you go down to the next to the last paragraph.

DR. COHEN: Yes.

DR. WEINBERGER: There were significant differences in major adverse events at 360 days. The flip side of that is actually a very complimentary finding for stenting and that is in patients with significant medical illnesses. There seems to be a very decided benefit to the stenting arm over the surgical arm. What I=m trying to tease out here is the statistics at least to a non-statistician are quite borderline. I=m trying to figure out who=s going to benefit from this approach.

What it seems to me is that based upon the data admittedly the subgroups may contain smaller number of patients. The patients who are more likely to benefit at least based upon what you=re reporting are the patients with medical comorbidities rather than anatomic problems.

DR. COHEN: I don=t think that=s a fair comment.

DR. WEINBERGER: You don=t think that=s a fair comment.

DR. COHEN: No.

DR. WEINBERGER: All right. Then let me ask you this. Why do surgeons refuse to operate on patients? You told us that 50 percent of the patients who were put in the registry got into the registry for reasons you could identify. Fifty percent had no identifiable reason for being in the registry.

DR. COHEN: Actually if I might just correct that statement. It=s not that we didn=t have information. We were asked this question three weeks ago to provide information on what were the factors that led to patients being entered into the non-randomized stent registry. That data is not on the case report forms. That=s on the screening logs and we had no database containing that data. In order to answer the question that was posed, we accumulated as many screening logs as we could. That=s why there is incomplete information.

What I would point out is that the items on that list are nearly identical to the list of demographic features which were at increased frequency in the patients in the non-randomized stent arm. So even though the data is incomplete which is more because of lack of time as opposed to not having the information necessarily, the factors that were identified were similar.

DR. WEINBERGER: So if you=re a vascular surgeon and you get a patient you is 75 years old with two or three medical comorbidities, is that sufficient in your practice not to operate on them?

DR. OURIEL: That would probably be a patient that would be eligible for the randomized portion of that study. Now if that patient had a lesion at the C1 vertebral body or if the patient had an MI within the last four weeks or other compounding things, then you might decide that this patient needed treatment but they were just too high a risk to have an endarterectomy. It really varied from site to site and surgeon to surgeon. Since some of the things are subjective, it=s very difficult to capture. Certainly you can=t capture it in a case report form.

DR. WEINBERGER: I think what I=m struggling with is that everything about this is very much interpretative. In other words, the decision to put someone in for revascularization is interpretative. The decision whether or not to push towards surgery is interpretative. Although we have some rigorous reasons that when patients accumulate enough of objective reasons to push them towards revascularization, that=s when the vascular surgeon will make some comment as to whether or not he wants to do it. Somehow they were eligible for the study without a vascular surgeon agreeing up front that they would do the case.

I was wondering why that was not a required part of the protocol. Why are patients eligible for enrollment and in fact enrolled before a vascular surgeon agrees that they would do the procedure if the patient is randomized?

DR. OURIEL: Well, in fact, they were not enrolled. They were actually eligible B- There was a certain number if you remember the slide was eligible for randomization and in those patients, you needed the surgeon and the interventionist to agree that they could have either form of therapy. Now if they were eligible by the criteria for enrollment, but the surgeon said that they did not want to do an endarterectomy then they ended up in the non-randomized stenting arm.

DR. WEINBERGER: All right. And the outcomes among those patients, those are patients who if we believe met the bulk of them, two-thirds of them are asymptomatic. Is that right?

DR. OURIEL: Correct. Roughly.

DR. WEINBERGER: Roughly. And ACAS would say that those asymptomatic patients would benefit from having an operation would have a revascularization under normal circumstances. But ACAS doesn=t report stroke frequencies that are anything close to what you are seeing with this patient population. So help me understand that particular conundrum.

DR. OURIEL: I think I can do that. ACAS, first of all, is a different set of patients because they didn=t have the medical comorbidities and probably didn=t have many of the anatomic comorbidities that this trial did. That said, if you remember some of the slides and you really look at the stroke rate which I think is what you mentioned despite the higher comorbidities in SAPPHIRE and the higher comorbidities for sure in the non-randomized stent arm, the 30-day stroke rates aren=t all that different.

DR. WEINBERGER: Okay. Then the last issue that I wanted to raise before I turn this over to somebody else is in terms of looking for subtle neurological deficits that occur post-procedure whether they be whatever the method of revascularization the nature of the neurological examination involved cognitive testing as well.

DR. COHEN: First of all, there=s an independent exam by an neurologist. Second, there are three tests that were used which are tests for deficiencies due to stroke, the NIH Stroke Scale, the Rankin and the Barthel. I think it would be best to have Dr. Pierre Fayad comment specifically on the components of those tests.

DR. FAYAD: Good afternoon. I=m Dr. Pierre Fayad. I=m a circ neurologist and professor and chairman of neurological sciences at the University of Nebraska. I am paid for my expenses today and for my time by Cordis. I am on the executive committee for the SAPPHIRE study. To answer the question, there were no specific neuropsychological testing that was requested as part of the SAPPHIRE trial. However, the neurologic exam would assess some basic neuropsychological functions and the NIH Stroke Scale would just direct the basic orientation and speech functions and so on. But there were no specific testing for neuropsychological function.

DR. WEINBERGER: Okay. Thanks very much.

DR. COHEN: May I? Just one little response. You had asked about the differences in outcomes on whether or not they were helping guide whether or not you should or should not enroll patients if this were approved. I would just like to point out that the numbers that you are referring to for anatomic reasons there were 35 patients who got carotid endarterectomy for anatomic reasons, 36 who received them who received the stent. The number of patients who actually had events in those two groups was two patients and four patients. So that=s why I say the difference between those are statistically meaningless. The numbers are so small.

DR. WEINBERGER: Thank you.

CHAIRMAN LASKEY: Thank you, Judah. Dr. Comerota.

DR. COMEROTA: Well, thank you, Dr. Laskey. I=m going to go about this a little bit differently. First of all, I think the sponsors are to be commended. They submitted detailed analyses from the studies that they have performed in bulk to the panel. They laid out a logical plan of investigation beginning with the FEASIBILITY study, moving to a randomized trial and then moving on to the registry for the reasons previously stated and then giving additional supporting documentation from IDE and CASCADE.

The pivotal clinical study as we saw was the SAPPHIRE trial which was performed under the IDE as you know. It did present supporting and safety data and the presentations were very elegant. At the end of those presentations, I thought this was going to be a very short day because the decisions would be quite obvious.

But let=s look at the studies a little bit and let=s look at some of the information. First of all, this began with the FEASIBILITY study which evaluated device and procedure safety and provided SAPPHIRE investigators experience with the angioplasty and the stent system. I would say in reviewing the investigators, the interventionalists, involved in this study, they are to be commended because these represent the top interventionalists in the country and I suspect if we were to rank them, they may fall within the top one to two percent of interventionalists in the world.

So the FEASIBILITY study was performed at 33 sites in the United States and it included 262 patients and 177 underwent angioplasty and stent and 85 patients had angioplasty and stenting with distal protection. They were followed for a year and then longer term follow-up was presented.

No demographic profile however was reported to us. Therefore, differentiation between symptomatic and asymptomatic patients as may have surfaced from this morning=s discussion was not presented. Subsequently we learned what the data were on atherosclerotic disease versus neoinitimal fibroplasia in this patient group.

Now it=s interesting that the stopping rules for the FEASIBILITY study were determined and were projected on the basis of the NASCET trial data of a major adverse events rate of 6.7 percent which was the second study from NASCET. The first study had a death and 30-day stroke and death rate of 5.8 percent. So that was chosen as the higher and perhaps the better number would have been the mean of the two if we=re dealing with symptomatic patients alone.

However, when we look at the data in the FEASIBILITY trial, major adverse events at 30 days was 6.9 percent and at one year, 10.7 percent. Death rate at 30 days was 0.8 percent and at one year was 3.8 percent. Stroke at 30 days was 6.1 percent and 8.4 percent at one year.

Now we did subsequently receive a differentiation of atherosclerotic patients versus recurrent stenosis or neoinitimal fibroplasic lesions. The major adverse event rate in the atherosclerotic patients was 10.8 percent. It was 8.5 percent in the recurrent stenosis group. The death rate was the same in both groups, 3.6 and 3.4 percents. Ipsilateral stroke was 7.7 percent in atherosclerotic lesions and 3.4 percent with recurrent stenosis. I think this is a trend and an appreciation that most of us have had that patients who have a recurrent carotid artery stenosis are a lesser risk for percutaneous interventions than patients with atherosclerotic disease. The mean pretreatment stenosis in the FEASIBILITY study was 66 percent and patients who had 50 percent or more restenosis at the end of the year numbered 24 percent of those treated with angioplasty and stenting.

So as I mentioned above, the distribution of patients was not given up front in the FEASIBILITY study. If, however, the distribution of the run-in patients were similar to SAPPHIRE patients in terms of disease and symptoms status, it can be assumed that nearly 70 percent of the patients would have been asymptomatic and about 25 percent would have had a recurrent carotid stenosis. Both of these groups are relatively low risk for ipsilateral stroke.

So if the stopping rule was actually calculated based upon lesion risk and if it was apportioned to the distribution of symptomatic versus asymptomatic patients, then the 30-day major adverse event rate should have calculated at about 2.9 to 3.0 percent rather than the 6.7 percent. If this were actually used as the guideline, the stopping rule would have been invoked and the FEASIBILITY study would have been terminated based upon that calculation.

When we look at the actual disease distribution in the patients in the FEASIBILITY study according to the angiographic description of the diameter percent stenosis, less than 10 percent of the patients had an 80 to 99 percent stenosis. About 38 percent of the patients had a 70 to 99 percent stenosis. The balance had less than 70 percent stenosis. So the magnitude of the severity of the disease was not too severe in terms of an angiographic diameter reduction stenosis.

As we move into the SAPPHIRE study, obviously these are the principal data supporting the submission for this IDE. The primary objective of the SAPPHIRE trial was to compare the safety and effectiveness of carotid stenting with distal protection using these devices versus carotid endarterectomy in the treatment of patients who are considered at high risk for carotid endarterectomy. It was an one-to-one randomization, multi-center trial.

The diagnosis initially was made on the basis of a duplex and I think the velocity criteria was certainly very appropriate. The primary endpoints which I think are very important to consider are the composite endpoints as clearly elucidated earlier including death and stroke and myocardial infarction at 30 days post procedure and then those additional data up to 12 months and beyond. The high risk criteria were well defined.

As we know, the SAPPHIRE trial was a randomized study and it was targeted for 600 to 900 patients, the thought with the interim analysis and that discussion has already proceeded so I won=t get into that. And we=ve seen the results in the randomized trial. We know that the death and stroke rate at 30 days in the stented patients was 4.2 percent and the death and stroke rate in the carotid endarterectomy patients was 4.8 percent. The overall stroke rate in the carotid angioplasty and stent patients was 3.6 percent. It was 3.0 percent in the carotid endarterectomy patients.

There were other associated events with procedures that didn=t surface in the discussion today as yet. A severe hypotension occurring during the procedure was 17.4 percent occurring in the carotid angioplasty and stent group and 3.0 percent in carotid endarterectomy patients. Bardycardia and/or asystole occurred in 8.4 percent in the carotid angioplasty and stent patients and 3.0 percent in the carotid endarterectomy patients. Those numbers barely reached statistical significance, not quite. That was 0.6 percent but the severe hypotension was very significant. Cranial nerve injury obviously 4.2 percent in the CEA patients and none in the angioplasty and stent patients. Distal vasospasm at the time of intervention occurred in 22 percent of the patients having carotid angioplasty and stenting. The analyses were presented and I think they were presented very well. I just supplemented some of those data with what I=ve reviewed.

Now when we look at the degree of stenosis of the lesion and I think we would all agree that a patient who has a high grade stenosis especially a patient who is symptomatic with a high grade stenosis is one who needs to have that carotid lesion corrected. If you look at the patients that we have data on that were submitted to us in the carotid angioplasty and stent group who had an 80 to 99 percent stenosis represented 22 percent of the patients in the randomized trial undergoing carotid angioplasty and stenting. Fifty-five percent had 70 to 99 percent stenosis. Therefore 45 percent of the patients who were randomized and received the stent had less than 70 percent stenosis.

Looking at the available data for carotid endarterectomy patients, 45 percent of the patients had an 80 to 99 percent stenosis. Eighty-five percent had an 70 to 99 percent stenosis. If we look at the registry data, 19 percent of the registry patients had an 80 to 99 percent stenosis. So we=re dealing with patients who underwent a percutaneous procedure that did not have relatively high grade stenosis in general.

A comparison was submitted to us regarding the registry patients and that comparison which was performed by the company compared the registry patients versus the carotid endarterectomy patients. They included of course death and stroke. When we look at the 30-day results of death and stroke, 5.9 percent in the stent patients versus 4.8 percent in the carotid endarterectomy patients.

If we look at overall registry patients versus carotid endarterectomy patients, ipsilateral stroke at 30 days 4.2 percent in the registry stented patients, 1.8 percent in the carotid endarterectomy patients. Then if we look at all strokes to 30 days, eight percent for stent and 5.8 percent for carotid endarterectomy.

If we look at symptomatic patients because this is real crux, I think, of what we=re dealing with, symptomatic patients with high grade stenosis, a death and stroke rate at 30 days is 8.1 percent in the stent patients and 6.5 percent in carotid endarterectomy patients. Ipsilateral stroke at 30 days, 6.5 percent in the stent patients and zero percent in carotid endarterectomy patients. And death or ipsilateral stroke 7.3 percent and 6.5 percent obviously. Then if we look at all strokes to 30 days, 8.1 percent in the carotid angioplasty and stent patients, two percent in the carotid endarterectomy and that one patient had a contralateral stroke.

Then if we move on to the asymptomatic patients, ipsilateral stroke was really no different, 3.2 percent and 2.5 percent respectively. Death and ipsilateral stroke at 30 days, 6.0 percent in the stented group and 3.3 percent in the carotid endarterectomy group. Death to 30 days in the stented group was 2.8 percent, 0.8 percent in the carotid endarterectomy group. So I think once we begin to look at the data, things do become clarified.

The CASCADE results were presented to us and 121 patients were reported in the CASCADE trial which was a European study. Ninety of those patients underwent angioplasty and stenting with no ANGIOGUARD protection and 31 of those patients had ANGIOGUARD protection. There were no deaths in the CASCADE study. There was an 8.2 percent stroke rate, ten percent in those patients who had no ANGIOGUARD protection and 3.2 percent in the 31 patients who had protection with ANGIOGUARD. Obviously we see a trend developing here that protection seemed to have been effective in that group. There were many more TIAs in the patients who had no ANGIOGUARD protection versus those who did have ANGIOGUARD protection.

Then the IDE data were presented. But I think since none of the IDE data were adjudicated, I think we have to look at it just as that. I don=t think we can accept the IDE data with very much vigor since it was not adjudicated. The FDA has asked us to address a number of questions. Warren, should I answer those questions from my perspective or should we wait until later?

CHAIRMAN LASKEY: We=ll have a turn at them as we go around. Do you have other queries or clarifications?

DR. COMEROTA: Well, yes. I have one other major point that I would like to address and it was asked by the FDA, but I would like to make it part of my preliminary comments. It has to do with the issue of myocardial infarction as an endpoint. I would just like to find my comments here.

The question was raised whether myocardial infarction either non-Q-wave or Q-wave myocardial infarction was a valid endpoint for this study. I think we just need to take a step back and look at what are we doing and why. Obviously the purpose of any procedure to treat carotid artery atherosclerosis is to reduce the risk of stroke, predominantly reduce the risk of ipsilateral stroke assuming that the procedure that we=re performing does not put the patient at risk for contralateral stroke.

Carotid endarterectomy has demonstrated its effectiveness in reducing stroke and death due to stroke in very large randomized trials and every one of them has been adjudicated by neurologists and compared to best medical care. Now this endpoint was achieved in reasonable risk patients.

If high risk patients were to have been included in these trials, the operation itself may not have proven beneficial compared to best medical care. If that were the case, carotid endarterectomy would not be available today for comparison to carotid angioplasty and stenting. The comparator would be best medical care.

In order to achieve equivalence with carotid endarterectomy in patients considered at high risk for operation, the SAPPHIRE included myocardial infarction as a component with major adverse events. None of us want our patients to have an MI and there is a substantial associated subsequent morbidity/mortality associated with anyone who suffers a myocardial infarction be it Q-wave or non-Q-wave.

However, if one looks at that as a endpoint obviously MI inherently favors a percutaneous procedure compared to an operative procedure in high risk patients. Furthermore, the premature termination of this study appears to bias this outcome in favor of carotid angioplasty and stenting. The reason why I say that is because it is well established that patients who have had a coronary artery bypass graft and who subsequently undergo noncardiac surgery have approximately a 50 percent risk reduction of a mortality associated with that operation. They have a 70 percent risk reduction of a nonfatal myocardial infarction.

In the SAPPHIRE trial in the randomized trial, 43 percent of the stented patients had a prior coronary artery bypass graft. 30.8 percent of the carotid endarterectomy patients had a prior coronary artery bypass graft. This difference is statistically significant. Furthermore, 35 percent of the stented patients had a prior percutaneous transluminal coronary angioplasty (PTCA) versus 23 percent of the carotid endarterectomy patients. This difference is statistically significant. The sum of that is that 80 percent of the stented patients had prior coronary revascularization versus 54 percent of the carotid endarterectomy patients.

So in the SAPPHIRE trial if coronary revascularization was equivalent and therefore the carotid endarterectomy patients were protected to the same degree as carotid angioplasty and stent patients were protected, would the difference in cardiac events have been observed? I think there=s a real chance that those difference would not have been observed.

The bias of prior coronary revascularization in favor of carotid angioplasty and stenting patients deflates the importance of the difference in myocardial infarction outcome between those two groups in the SAPPHIRE trial. Furthermore, carotid angioplasty and stent patients were treated with Clopidogrel in addition to aspirin. We all know that the combination of aspirin and Clopidogrel protects patients at high risk from coronary events.

This pharmacologic protection was not offered to patients undergoing carotid endarterectomy and perhaps for good reason. But there is now a revascularization bias and a pharmacotherapy bias in favor of the reduction of myocardial events in patients undergoing carotid angioplasty and stenting compared to carotid endarterectomy. I think that will conclude my comments for now. Thank you.

CHAIRMAN LASKEY: Did you wish the sponsor to respond to any of that or all of that?

DR. COMEROTA: Well these are observations based upon all the information that was presented in the Panel Pack. I had no additional information than what everybody else had.

CHAIRMAN LASKEY: Fair enough. Okay. Well, thank you. Let us go around the table for thoughts comments and queries. That=s right. Dr. Aziz. Thank you.

DR. AZIZ: Thank you. I would also like to commend the sponsor on an excellent presentation. Let me go straight to the topic of coronary artery disease in some of these patients and this is a question for the sponsors and maybe one of you could answer. Once a patient is identified as having carotid stenosis, what investigations are done to rule out underlying coronary artery disease particularly in patients who are going to go for a carotid endarterectomy? Do they have any noninvasive tests done or do they go straight to surgery?

DR. COHEN: Again, patients in this trial would have otherwise received the same type of preoperative evaluation that any patient undergoing either surgery or an interventional procedure would have undertaken, so whatever was appropriate given the individual patient=s history and physical exam, the EKG findings, whatever.

Having said that, I=d like to take the opportunity to mention that the distribution of patients with coronary disease in fact was equal between the different arms of the trial. Second of all, there was data captured in terms of patients who had a positive exercise test. So yes, that was obtained and those were equally distributed for patients who had positive exercise tests.

The other thing is that first you can=t simply sum the percentage of patients who have had bypass and who have had percutaneous coronary interventions because obviously those two groups overlap significantly. The other thing is that we actually looked to see whether the presence of coronary disease, the presence of bypass surgery or whatever played an important role in any of the individual outcomes as well as the composite outcomes and they did not.

DR. AZIZ: Let me see if I understand you. If a patient was found to have coronary artery disease or suggestive on the testing, would that patient have that corrected first before he had the carotid endarterectomy?

DR. COHEN: Yes, and it was explicitly stated in the protocol that if there was coexisting coronary disease or if another surgical procedure needed to be undertaken, be it, a carotid procedure or cardiac or otherwise, it could not occur with 30 days of the endarterectomy procedure.

DR. AZIZ: And then if a patient had, let=s say, carotid angioplasty and stenting done, obviously the patient would be put on Plavix I would presume for a long period of time. That would clearly delay him having coronary artery surgery for a number of months.

DR. COHEN: Actually the duration of Plavix was two weeks as mandated in the protocol.

DR. AZIZ: That=s one. Were there any patients who during the course of the carotid stenting develop carotid section?

DR. COHEN: I=ll ask Dr. Ouriel to answer this question.

DR. OURIEL: A long walk for a short answer. No.

DR. AZIZ: Now in terms of the emboli protection, I see these two groups of patients, clearly patients who are having intervention done on the carotid artery. The danger is obviously having emboli going in and you=ve obviously included an emboli protection device. I=m sure that the emboli protection device captures some of the emboli. But I guess we really had no way of knowing what percentage of the emboli it captures.

DR. COHEN: Actually there is data in the Panel Pack that speaks to that and I can summarize that. The percentage of emboli-capture devices, ANGIOGUARDs, actually had debris in it. It varied somewhat between the trials, but it was easily between 50 and 80 percent amongst the trials that we presented today. The average number of particles was six to eight particles per filter in filters that had debris, although the number ranged all the way up to, I believe, 20 particles per filter. The particles could be as large as 1 X 1-1/2 millimeters in size. The composition was basically what you would expect of an atherosclerotic plaque. There were smooth muscle cells, foam cells, cholesterol crystals, necrotic core, collagen, elastin and clot basically.

DR. AZIZ: Are you aware of any studies that were done, not necessarily for this trial, but patients may have had MRIs pre- and post-carotid extending procedure with or without protection devices?

DR. COHEN: No. There have been no studies that I=m aware of that have completed, although, I believe there are studies underway directed at neuropsychiatric changes as well as perhaps imaging studies.

DR. AZIZ: Okay. And you aren=t aware of any studies where TCD monitoring was being done at the time. Maybe you might be able to answer that question.

DR. OURIEL: You=re not talking about studies related to this particular panel.

DR. AZIZ: No.

DR. OURIEL: I=m sure there are TCD studies that are available. Of course, it=s a surrogate endpoint. There are studies that show that if you have protection that you get fewer hits, but it=s not part of this analysis.

DR. AZIZ: All right. I think I have one. Now I think it was earlier stated that there was a 17 percent incidence of hypotensive episodes during the placement of the stent. Do you have any ideas as to why that happened or what could be done to prevent that?

DR. OURIEL: Sure. I think of that as part and parcel of the stenting process just like for those of us who are surgeons what we know when the anesthesiologist puts our patient to sleep we get hypotension frequency. Now it=s not recorded in many cases, but we know that it occurs when the patient is put to sleep. We know that in a patient who has a carotid lesion that you stent especially a tight stenosis, maybe one with a lot of calcium, they are going to get a vagal impulse and it=s not surprising that you get hypotension and Bardycardia.

DR. AZIZ: Nothing else for the time being.

CHAIRMAN LASKEY: Dr. Krucoff.

DR. KRUCOFF: Dr. Cohen, you may as well hang around. Have a seat. Obviously one of the things we=re all wrestling with here is the statistical analysis plan, its origin and then what=s actually eventuated. Certainly I=ll echo everybody. You guys have done a great job in taking what=s a very complex dataset and at least presenting it in a very cogent and understandable fashion. In general, studies in this realm are conducted with Data Safety and Monitoring Board. Was there a DSMB for this trial?

DR. COHEN: Yes, there was.

DR. KRUCOFF: And did they have a role in the original triangular analysis plan or can you help me understand who was going to look at the data along the way in the original plan for these 100-patient cohorts and how was that originally envisioned from your perspective?

DR. COHEN: Perhaps what would be useful would be to have a little bit better understanding of exactly what this data analysis, interim analysis, plan was and how it played out. I think I would like to ask Dr. Rick Kuntz to come up and provide an explanation.

DR. KUNTZ: My name is Rick Kuntz. I=m a cardiologist in Boston. I=m mostly the chief of the Division of Clinical Biometrics at the Brigham Women=s Hospital. I functioned as the chief scientific officer and CRO run by Harvard which ran this trial.

The statistical interim analysis was contracted with a group in England called the Whitehead group. We have a representative from here who developed the triangular test that was used. We performed the analysis using that methodology. The sponsor and the group in England worked on the analysis plan and actually conducted the analysis plan.

Let me just explain my perspective on this, but being a clinician and having a little bit of background in statistics on this. It was clear from the beginning in the design of this trial that we didn=t know what the final sample size would be. There were a lot of unknown variables. An interim analysis was actually quite an effective way of potentially looking at this study. The study was going to be as large as 2400 patients if in fact our estimates were off and possibly as small as 300 or 400 patients if we had really good results. We estimated from our best analysis of the literature that probably 600 to 800 patients would result in a final analysis demonstrating non-inferiority. Maybe if there was a bang-up job done by the stents there would be superiority at some point, but our main goal was to look at non-inferiority.

In this analysis plan, the triangular test follows most interim analysis theory. That is in fact the more you look the more alpha there that you have to spend because you are rolling the dice each time to look for a positive result. In this study, the triangular test allows you to alter the times when you look during the conduct of the study. That is that you can actually use other cues to determine whether or not you want to look or not look and that=s part of the analysis plan and part of the textbooks that are written, part of the theory that=s been published in biostatistical literature and part of the program.

There were non-data driven reasons to not look at the first 200 patient intervals. That was because this was a new break-through therapy and it was very unlikely that regardless of the result of 100 or 200 patients that this would result in anything convincing to anybody because you just don=t have enough patients. The first reasonable time to look would be at 300 patients and that was decided at the beginning of this study. There was no data reviewed at all.

When it came to the point where the first planned interim analysis now, 300 patients, was going to be done, it was very clear on those curves that this study wasn=t going to be enrolling much more than 350 or 400 patients. So a decision was made to do the first look ever as the final look. The notion about not using the monitoring portion of the Whitehead test was omitted under the complete allowances of this Whitehead triangular test.

This was not communicated by the sponsor to the FDA. That probably was a mistake. They should have communicated that with the update and tell them what they were doing. But technically speaking as far as I understand it and as far as the program goes, this was all allowable in the analysis.

There was only one analysis done during the study and that was the final analysis. There were no decisions made anywhere based on any of the data and I can verify that. Nobody got any of the data except Data Safety and Monitoring Committee during this study.

So in the end, this study which started to peter out quickly at around 280 patients enrolled did peter out at about 350 patients, 334, when the enrollment was so slow that it wasn=t worth the money and the resources to continue and it was clear that it wouldn=t continue any further because many studies were having problems randomizing and there were the registries available that were chipping into the abilities randomized. So because of that, all analysis were presenting as a first time and final analysis as is appropriate.

There were irregularities with respect to the communication with the Food and Drug Administration about exactly what was going on, but I think the communications dealt with the fact that they didn=t know that this was appropriate to communication since it was all within the design of the triangular test. Looking back at it since this was a not typical traditional test that was used although very valid, there should have been better communication.

There is no doubt about it, but the statistical analysis we feel stands as is. It=s a one-time analysis. There were no increased chances to look at a positive result. Nobody rolled the dice more than once to understand whether results came up positive or not and these are the final results of the study. That is the best way that I think we can clarify on a clinical level what actually happened in this study with the interim analysis.

DR. KRUCOFF: Thank you. That at least for me clarifies some really key things. So what I=m hearing you say, Rick, is that nobody peaked at these data along the way. The original Whitehead triangular, somebody, not including the FDA, decided up front that the options for earlier interims which were within the statistical plan were going to be delayed until the 300 patient point. The decision was made before enrollment had begun.

DR. KUNTZ: No, actually I think that decision was made after enrollment was done, but within the allowances of the play of chance of the Whitehead triangular. That method can be explained theoretically by our statistician from England on this and he would be more than happy to explain that. But I think the key is that there were people looking at the data. We still use the 100-patient interval for the Data Safety and Monitoring Committee to review the data. They did see it at 100, 200, 300 patients to make sure that there were no safety concerns.

DR. KRUCOFF: But then just to make sure that I=m following you. Data and Safety who saw the data along the way, were they involved in the decision to go to 300 as a first formal Whitehead triangular look or were they independent of that decision?

DR. KUNTZ: They were independent. The decision to do the first interim analysis at 300 patients was a decision made by the sponsor.

DR. KRUCOFF: Okay. And last, was there anyone from the sponsor involved in the Data and Safety Board?

DR. KUNTZ: No. Nobody from the sponsor saw any of the data.

DR. KRUCOFF: Okay. Thank you. I=d like to shift gears a little bit and understand a little bit more about where angiograms were used and where they weren=t. As I understand for the randomized SAPPHIRE patients, Dr. Cohen, the patients would come in through whatever clinical evaluation recommended for carotid revascularization.

DR. COHEN: Correct.

DR. KRUCOFF: It sounds like there is a cohort of patients who were operated on based on ultrasound Doppler without a concomitant angiogram.

DR. COHEN: That=s correct.

DR. KRUCOFF: All of the patients on the percutaneous side had angiographic information. Do you know how many of them had a diagnostic angiogram independently from the interventional procedure? My questions are twofold. One is just how you knew whether a patient was a candidate for a stent or not in order to decide whether or not they were randomizable or should be put in the register without an angiogram.

DR. COHEN: My understanding is that for the surgical intervention or surgical treatment that because of the risk of doing angiography in patients who have significant carotid disease that it is usually omitted. So you go directly from ultrasound to surgery. I believe the majority of patients in this trial - we can conject to make sure that this is correct - their diagnostic angiogram was done at the time that they received their treatment with the stent and distal protection.

DR. KRUCOFF: So my only question is then how did you know you could randomize them if you didn=t have angiographic criteria to know whether they were tortuous of calcified or some of the things that would make you think stenting was not an option?

DR. OURIEL: Sure. In fact, some of the difference between the intent-to-treat and the treated is because you did an angiogram and you found out that this patient should not have been treated with the assigned therapy.

DR. KRUCOFF: Okay. So then if I understand, patients came in clinically. If they randomized to surgery, they could get operated on on the basis of the Doppler ultrasound data.

DR. OURIEL: Or could have an angiogram in the minority of patients.

DR. KRUCOFF: Okay. Or if they randomized to stenting, then most all of these patients had their first angiographic delineation of the anatomy at the time of their interventional procedure. Is that it?

DR. OURIEL: I can=t give you the numbers on that? But what I can tell you is that if a patient had an angiogram in preparation for a carotid stent and let=s say you found the stenosis was only 30 percent, then they would still stay in that arm intent-to-treat but they would not have been treated.

DR. KRUCOFF: Okay. And importantly then, patients who got angiograms with the 1.3 or so percent risk of a complication from that angiogram would already be in the intention to treat analysis. So complications from a diagnostic angiogram are in fact in the intention to treat.

DR. OURIEL: Sure. Once they are randomized. Although the 1.3 percent risk of stroke with angiography hopefully is no longer the case.

DR. KRUCOFF: Right. Understood. Then if I can, Dr. Cohen, ask you a little bit about the use of the ANGIOGUARD. As I look through the CASCADE and FEASIBILITY data at least my interpretation of what=s there is that patients who did not have the ANGIOGUARD didn=t have it because it wasn=t available.

DR. OURIEL: That=s correct.

DR. KRUCOFF: And then the percentage of the denominators that did receive it was presumably because it was available. Is that it?

DR. COHEN: That=s right. It became available later on in those trials.

DR. KRUCOFF: Okay. And I guess in looking at these data, one question that will come up that I=m just going to go ahead and ask is obviously the ANGIOGUARD seems to have an important function in all of this. What I=m wondering is whether it=s your inclination in the instructions for use to suggest that since that=s deployed first - you put the wire up, the distal protection system first - if you are for whatever reason unable to deploy the ANGIOGUARD, are you going to suggest that the procedure be terminated?

DR. COHEN: No, I don=t believe that=s what occurred in the trial. That would not be our suggestion.

DR. KRUCOFF: Okay. And yet is it fair to say that the data on stenting alone compared to stenting with distal protection looks like there=s a significant role for distal protection?

DR. COHEN: But I=d also offer that this was early in the learning curve for some of the physicians participating in the trials, so just the learning of doing coronary stenting. Second, this was with earlier generation devices. Third, I would say that there=s a cohort of physicians who do carotid stenting who do not believe in the benefit of distal protection.

The reason we did do exploratory analysis of combining both CASCADE and FEASIBILITY was to provide data that there is benefit to having distal protection. The reason that we collected the filters and had a pathology lab evaluate them was to demonstrate that there is material captured on them. However, the question I assume you=re getting at is if you are unable to pass it, should you not do the procedure.

I don=t know the percentage - and we can look that up - of patients who had to have predilatation before the ANGIOGUARD was actually placed, but that is a component of the dataset. My understanding is that those patients had outcomes that were not different than the overall trial outcomes.

DR. KRUCOFF: Okay. We can come back to this in discussions about labeling. At least the best that I could get out of the three sets of data, the CASCADE, FEASIBILITY and SAPPHIRE, SAPPHIRE, it seemed like, you probably had a more advanced iteration of the ANGIOGUARD end of the system because the deployment rate was fairly high.

DR. COHEN: That=s correct.

DR. KRUCOFF: And again I think you=ve shown some of it particularly in the randomized patients the noninferiority data for primary endpoint with regard to the 334 patients you have in intention to treat analysis. About 90 some percent of those had distal protection. What I=m concerned about is whether the data would look the same if the 334 patients had been randomized to unprotected stent versus not and what that ought to mean in terms of the real recommendations. But we can come back to that.

Do you know offhand -- And I don=t do carotid so I=m going to have to extrapolate from coronaries. In self-expanding platforms in the coronary arteries, there are times when all it takes is the structural strength of the Nitinol to dilate the lesion? There are times when you have to predilate the lesion to get the stent across and there are times when after deploying the self-expanding platform you have to post-dilate. Do you have - or if at least it was there, I=m sorry because I missed it - what percentage of the time was predilatation required and what percentage of the time was post-dilatation required?

DR. COHEN: I don=t have those percentages off the top of my head. I would need to look them up and we can see if we can get that. But both predilatation and post-dilatation were done in some cases.

DR. OURIEL: Well, I can tell you our practice at the Cleveland Clinic is to almost always predilate and always post-dilate because you do get a significant stenosis if you don=t post-dilate in almost every case.

DR. KRUCOFF: Okay. Because while I=m not going to go back to the marked hypertension and Bardycardia, in cases I=ve actually observed certainly the baroceptors in the neck is a pretty richly enervated territory managing patients. After carotid endarterectomy, you sure learn that lesson in patients. Obviously in a significant proportion of these patients how much manipulation is involved may also relate to that assault.

DR. OURIEL: And it=s always on the post-dil that it occurs.

DR. KRUCOFF: That is goes. Yes. But at Cleveland at least, you think that was most of the time you ended up post-dilating.

DR. OURIEL: Yes.

DR. KRUCOFF: Okay. Last two quick questions. One, it was pointed out in the distribution that out of the range of devices that you all manufactured the five millimeter device and the tapered seven to ten millimeter device was rarely used. Do you feel like you have sufficient data? Do you think a five millimeter flow channel to the brain, do we understand enough about that to say that=s going to behave identically to larger caliber vessels and do we have enough information to understand the tapered seven to ten?

DR. COHEN: I think it might be useful to understand why the five millimeters and the tapered are used. I would like to ask Dr. Nick Hopkins to make some comments.

DR. HOPKINS: Hi, I=m Nick Hopkins, neurosurgeon from Buffalo, chairman of the Department of Neurosurgery there and professor of radiology there. I=m kind of a hybrid. I=ve doing neurosurgery and carotid endarterectomy since 1979 and doing carotid stenting since the mid >90s. We have a large experience. Cortis did pay my way here and paid for my transportation and my lodging.

The question about the five millimeters. First of all, almost every carotid artery narrows down to somewhere near five millimeters somewhere near the skull base and a five millimeter opening, if I understand your question correctly, is plenty to provide normal flow. We don=t see significant reductions in flow until we have stenosis somewhere close to 60 percent. So a five millimeter opening is plenty large enough for that.

The reason for the seven to ten I think is the rare patient where you have a common carotid that is so large that you wouldn=t have good stent apposition if you didn=t have a ten millimeter stent. So a seven to ten taper takes care of that situation which is unusual, but it happens. So to not have those two sizes available I think would put us at a disadvantage. Does that help?

DR. KRUCOFF: I guess the conundrum for me is that five millimeters is the small end of carotid which would be the huge end for coronaries and again in my limited world of coronaries though smaller, longer term durability of stenting interventions is certainly different in smaller vessels than in larger vessels. As we get out into larger vessels, carotids and peripheral, there is some relatedness albeit the total flow volume is different. I guess one of the impressions I=ve gotten here is for ethical reasons more than anything. We don=t have a lot of angiographic follow-up or detailed follow-up in the whole cohort much less in how five millimeter stents behave in carotids. Is that fair?

DR. HOPKINS: I think that=s fair. You made the point that the average carotid carries 250 millimeters of flow per minute. It=s a huge difference in terms of the flow volume to the brain and the heart. A five millimeter stent, I think, is critically important in a situation where for example you have a lesion confined to the internal carotid. You want to limit your stent placement to the internal carotid and you have an internal carotid that may be relatively small. So you don=t want to greatly oversize, if you can avoid it, more than you have to. We like to oversize a millimeter or two, but I would hate to put a six or seven millimeter stent in three millimeter carotid. I think a five millimeter is extremely important for that situation.

DR. KRUCOFF: Okay. I lied. I actually have two last quick questions, one really quick one just to ask, your interpretation of the slide that you put up and when I asked the FDA, Dr. Li, earlier. You put up a slide for an FDA simulation of the V-Z approach that had 100 patients continue it, 200 patients continue it, 300 patients terminate. Dr. Li=s impression was that at 300 patients in that sawtooth, sort of, Christmas tree you were at a point that you would say would reach the recommendation to stop, whereas at 334, there was dot on his slide which was still within decision matrix boundaries which would be to continue. As far as you can tell, is that still appropriate? Did you all get your continue option at 300 and did you look in the same way at 334 as to whether you would have been inside or outside?

DR. COHEN: I think what I=d really like to do is have the expert here that we have available answer the question. I think if you don=t mind, Nigel, to come up. The short answer of that is that we=re trying to interpret this using our background which is not applicable to this specific analysis method.

DR. KRUCOFF: Maybe while he=s setting that up, can I ask you my last question? Why self-expanding, not balloon-expandable and is Nitinol a reasonable platform for a drug-eluding future?

DR. OURIEL: Well, we actually used to use balloon-expandable stents in the carotid, but as soon as the patient put their neck on their hand and rested their neck against some pressure, it would crush. So it=s clear that balloon-expandable stents aren=t going to be good in superficial locations. Nitinol right now is the best we have.

DR. COHEN: And just so people are aware of this, Nitinol has the ability to self-expand. Plus if you crush it, it will expand back to a predetermined size. That=s it=s major advantage and why we use it in areas that are compressible.

DR. KRUCOFF: Okay. So while they are setting up, have you all begun to explore whether Nitinol is a reasonable platform for polymer and drug-eluding configurations?

DR. COHEN: I=m not sure that=s an appropriate answer to give in this forum.

DR. KRUCOFF: Okay. Never mind. It might get you small vessel interest.

DR. STALLARD: Okay. I=m Dr. Nigel Stallard. I=m a principal research fellow at the Medical and Pharmaceutical Statistics Research Unit at the University of Reading. I=ve been paid as a consultant by Cortis and they=ve paid for me to come to this meeting and paid for my time?

MS. WOOD: Could you please pull the mike a little closer?

DR. STALLARD: I=m sorry. Okay. So just to briefly talk about the analysis that we would do had we performed those interim analyses at 100, 200 and 300 patients. But I want to start by just outlining how we analyze at the end of a sequential trial like this. Here=s just a picture of the triangular region and the triangular region is designed so that the probability of crossing the upper boundary during monitoring of the trial is controlled to be 0.025 when the treatment difference is actually the delta is the lower limit of the non-inferiority region that we=re interested in detecting. So that=s what we=re controlling when we do this test.

As has been explained, you can monitor at any time you wish so long as that time is chosen independently of any observed treatment difference. And the triangular region is calculated based on the assumption that you=re going to monitor continuously. As soon as you monitor the number of discrete points, the chance of stopping any one of those points is reduced or the chance of stopping in its total trial is reduced. So to control that to be at the 0.025 level, you need to bring in the boundaries further. So the solid lines that you just saw are, if you like, the most stringent ones. We actually adjust those bringing them in to allow for the analyses that we did. The more analyses you do, the less you have to bring them in because you=re looking more often.

When you stop the trial at that point, you can calculate a P-value and it=s helpful just to remember the definition of what we mean by a P-value as statisticians. What we mean in the case of testing for non-inferiority is it=s the probability that if we had a true difference of delta that we would see data as strongly supporting non-inferiority as we=ve observed or more strongly supporting non-inferiority. So that=s what we mean by a P-value. It=s that probability which we need to calculate based on our inter-monitoring to be able to perform a valid analysis at the end of our test.

In order to work out the probability of data that might have been more supportive, we need to address the question AWhat do we mean by more supportive in terms of non-inferiority now that we have a sequential test?@ For a standard analysis when you just do one test, that=s fairly clear. It just means a larger test is at stake. For a sequential analysis, it=s more complicated. It means either stopping at the same interim with a larger test - that=s clearly more supportive evidence - or stopping on the upper boundary earlier on because that also would have been more supportive of non-inferiority.

Here=s just a picture, not of the real data, but just of some hypothetical trial. You can just about see the dotted lines corresponding to the Christmas tree boundary and you actually use these inner points of that Christmas boundary. I=m sorry that it=s so faint. So the probability of observing data more extreme is the probability of either having more extreme data at this look so that the solid line above the plotted point or the probability of stopping at either of the two previous looks, the two solid lines which are drawn in there. It=s the probability of those three parts that we work out to get our P-value if that was our sample path.

That=s if we stop exactly when we cross the boundary. If we don=t do that, but cross the boundary and then take another look, then we do what=s called an over-running analysis. The P-value here is calculated ignoring the point which led to stopping and basing instead using our final value here of looking at the probability of having a larger value here. So in this case, we would sum up the probability of being here, here or here. (Indicating.)

One thing to say is that this can be much less than 0.05 or 0.025 in this case because the whole probability of stopping on the upper boundary is 0.025. So by stopping very early on, we can get a much more significant P-value. There is the real data just taking that one look at 334 patients. You can see that because we were above the inner point of the Christmas tree boundary, we in fact would have stopped the trial there if we would have the position to stop the trial not already been taken. That leads to the analysis which has been presented.

Here=s the one reproducing the three looks at 100, 200 and 300 and then treating the look at 334 patients as over-running data and allowing for that over-running analysis I=ve just described, then we get the analysis results which have been presented. So we get a P-value for testing for non-inferiority which in fact is almost exactly the same as the P-value ignoring those three that we=ve seen.

DR. KRUCOFF: Okay. Thank you. That was again very helpful. So again if I understand what you=re saying, your analysis in Dr. Cohen=s table is based on 334 using the over-run appropriate to the model.

DR. STALLARD: That=s correct.

DR. KRUCOFF: And we then would have to resolve why the FDA would have put up a slide suggesting that at 334 you would be reaching a decision to make stop. We=re using the same denominator.

DR. STALLARD: We=re using the same slide here as the FDA because the last point is in fact back inside the boundary.

DR. KRUCOFF: But it=s the over-run analysis.

DR. STALLARD: But nevertheless it=s because as you see here the over-run analysis that=s appropriate and that leads to the conclusion.

DR. KRUCOFF: Which is what you were representing in your slide.

DR. STALLARD: That=s correct. So the conclusion is exactly the same whether you include those extra interim analyses or not.

DR. KRUCOFF: Yes.

DR. COHEN: Just to answer, Dr. Krucoff had asked two questions. One, we do have data in our case report forms concerning the percent predilatation, but we don=t have that available. We can supply that to the FDA later. In response to your other question about patients who do have an ANGIOGUARD, in the randomized portion, there were six patients who did not have an ANGIOGUARD. In the non-randomized stent registry, there were 25 patients who did not have an ANGIOGUARD. Reading through the descriptors, there were no strokes albeit it is a small number.

DR. KRUCOFF: Okay. What I=m going to come back to later is whether the instructions for use should have any advice on preventing hypotension and whether we have any sort of clue as to how you might do that.

DR. TRACY: Thanks. I=m very confused by Christmas trees, but I think I just would like to ask Dr. Li. If I understand, it depends when you look at the data. Even the sponsor is indicating that it would fall under the line if you had done the analysis the way it was initially intended to be done. So my question is that somebody made that decision somewhere along the line to not do it the way that FDA and everybody else including IRBs and if the patients had any reason to understand the analysis that was going to be performed, the patients would not have understood the analysis would be performed this way.

Does the sponsor see any ethical dilemma with not following their originally outlined program or do you feel that you in fact did follow your originally outlined program? Then I guess I would ask the FDA also, the statistician, Dr. Li, if he concurs with the analysis that was just presented? Maybe I=ll ask you that first, Dr. Li. Are you happy that the dot falls outside the boundary if the analysis was performed the way the sponsor indicated that it was performed?

DR. LI: You mean the final dot being inside the boundary?

DR. TRACY: Yes. Is it in or is it out?

DR. ZUCKERMAN: Dr. Tracy, you are asking an extremely difficult question.

DR. TRACY: I understand that.

DR. ZUCKERMAN: Let me take a first crack at it. As you know, that analysis is not provided in your PMA Panel Pack. Dr. Lee, I believe, saw that analysis for the first time yesterday. These are very complex analyses that cannot be done on the back of a handkerchief or napkin. So all we can say at the present time is this is very interesting and certainly there=s a lot to this, but these analyses that were just shown need to be appreciated in the context that they have not been thoroughly reviewed by FDA statistics. Dr. Li.

DR. LI: I think if that satisfies your question, then I have nothing further to add.

DR. TRACY: Okay. I think it does. The answer is that the conundrum comes up by the fact that the originally outlined program per the best of my ability to understand was not followed or it was so darn complex nobody understood exactly what the program was that was going to be followed. So I think there=s some kind of false there with setting something up that is so difficult to adhere to. I wonder about the ethics of the situation that we=re in right now. I wonder what the patient thinks. I wonder what IRBs think.

DR. COHEN: I think I would like to make two points clear. First, the method itself allows for discretion in terms of whether or not to perform an analysis. That=s within the confines of this analysis method. So there is nothing that=s an ethical issue in terms of response or non-response communication or non-communication. What we failed to do was communicate how we decided to enact this whole method with the FDA. Does that help?

DR. TRACY: Yes, it helps and I agree. You did fail to communicate. It puts this panel in a bit of a difficult situation, but we=ll let that go.

DR. COHEN: Well, let me make sure we understand. The analysis, the triangular method, has to do basically with the stopping rule. It has nothing to do with the actual separate analysis of the primary endpoint as to whether or not it meets non-inferiority. Those are two separate things. The only relationship between the two is if you do a preliminary analysis. Then you basically enact an alpha penalty on your final analysis. Otherwise the two are distinct.

DR. TRACY: Okay. There was a group of patients who were not treated. That, I believe, was slide 85 in your presentation. In our little packet here, it=s page 29. It=s a total of 24 patients or so who were not treated but who had been randomized. Do you have any data on what happened to those patients?

DR. COHEN: Actually that was presented in the main presentation. If you=ll give me a second, I can find the slide. But we gave the reasons why patients - it was eight and 16 patients - were not enrolled.

DR. TRACY: Right.

DR. COHEN: There was a mixture of reasons as I=m looking for this. One was obviously that we=d already discussed upon angiography the patient was found not to meet the inclusion or exclusion criteria. I=m sorry.

DR. TRACY: That was slide 85 in your original presentation. My question was what happened to those 24 patients.

DR. COHEN: They were included on an intent to treat analysis. Are you asking what happened to them in terms of treatment or were they included in the analysis? They were included.

DR. TRACY: They were included in the analysis?

DR. COHEN: Yes, in the intent-to-treat because they were randomized.

DR. TRACY: Okay. So you have no separate information on what happened to those patients clinically.

DR. COHEN: We have the information. I don=t have it.

DR. TRACY: But you don=t have any analyzed. Okay. One of the things I=m struggling with is trying to figure out what the indications for the device would be. I was hoping that that group of patients would provide some type of analysis or a control group.

But I guess if we relied back on the historic controls that were presented, in asymptomatic patients even with high grade stenosis at about a year it looks like the risk of stroke is about five percent in historic controls. It looks like the stroke rate is around 7.3 percent at a year with the stent treated patients in asymptomatic patients, 7.7 percent stroke at one year in asymptomatic patients. So I=m wondering how I would convince a patient who is asymptomatic and faces about a five percent one year stroke risk to undergo a procedure that will give them a seven percent stroke risk.

DR. COHEN: I=m not sure that I would agree with the five percent stroke risk. What we had presented in our major presentation is - the first statement was - that there is no contemporary data that allows us to understand what the risk of stroke is in these patients. There is no study that=s followed medical therapy in these patients. That was not the goal of this trial. The goal of this trial was to determine what the outcomes were in these patients who are treated either with carotid endarterectomy or with stenting. There is an absence of historical data and we acknowledge that.

DR. TRACY: Yes. Even if you compare within the asymptomatic versus symptomatic patients, your 237, the outcomes in the 237 asymptomatics versus the 96 symptomatic patients, it looks as though the asymptomatic patients fare worse than do the symptomatic patients which is of some concern.

DR. COHEN: Perhaps you could point to exactly what you=re looking at.

DR. TRACY: Slide 96.

DR. OURIEL: I think while Dr. Cohen is looking for that what this trial shows is that if you decide that a patient needs treatment and they are asymptomatic, high risk and if you were going to perform a carotid endarterectomy, they will do as well with a carotid stent.

DR. TRACY: Okay. Then it becomes a clinical issue of whether or not you want to subject somebody to that risk either of stent or surgery.

DR. OURIEL: Yes. You see with this trial the decision was already made that they were going to get treatment or they wouldn=t have entered the trial.

DR. TRACY: Right. It just makes me wonder what would go through my mind to get me to a point of making a recommendation like that.

DR. OURIEL: Personally as a vascular surgeon if I had a patient with an 80 or greater percent carotid stenosis, asymptomatic, that fell into this category, then I would probably recommend carotid endarterectomy in the absence of a stent. If a stent were available after this data, I would consider stent equally with carotid endarterectomy.

DR. COHEN: And if I could just respond to the other thing if I=m looking at the slide that you are indicating with the stroke rate of 3.3 percent CEA and 5.1 percent with stent, let me point out that the total number of patients in each group is only 120 and 117. So you=re talking about the difference in outcomes of one or two patients. I would not suggest making clinical decisions based on such a small number.

DR. TRACY: And I think that it=s comparing the outcomes ultimately in symptomatic versus asymptomatic patients. There seems to be a discrepancy.

DR. COHEN: Actually with regard to the overall event rate of major adverse events which was the endpoint of the trial, the asymptomatic patients actually had numerically the lowest major adverse event rates.

DR. TRACY: Who had the higher stroke rater, asymptomatic or symptomatic?

DR. COHEN: I=m sorry. At what time and what type of stroke? Ipsilateral stroke?

DR. TRACY: Stroke.

DR. COHEN: All strokes?

DR. TRACY: All stroke.

DR. COHEN: I=m sorry. At 360 days if you look on slide 97 which is asymptomatic and slide 100 which is symptomatic at 360 days, the overall stroke rate is 7.5. 7.7 for asymptomatic. 6.5 and 2.0 for symptomatic.

DR. TRACY: So it is the asymptomatic that have a higher stroke rate.

DR. COHEN: Numerically, yes, but not statistically.

CHAIRMAN LASKEY: Well, about that, we keep going back. I think we need to be very clear about the lack of utility of post hoc subgroup analyses. I don=t think it belongs on the docket.

DR. COHEN: I agree.

CHAIRMAN LASKEY: It should be qualified as such and I think we need to be very careful about playing this game.

DR. TRACY: I=ll try to wrap up here very quickly. There appeared to be a higher rate of TIAs in the stent group at 360 days. Why was that true and did it correlate the degree of restenosis or initial stenosis or was there some predictor for that?

DR. COHEN: The observation is there that there were an increased frequency of TIAs occurring. We have no mechanistic explanation. We can make guesses, but I don=t think that=s really useful in this forum.

DR. TRACY: Okay. That=s all. Thank you.

DR. COMEROTA: I made the majority of my comments. I have three quick questions.

CHAIRMAN LASKEY: Can you hold on? I wanted to make one pass around the table.

DR. COMEROTA: Pardon me.

CHAIRMAN LASKEY: Sorry. Thank you.

DR. NICHOLAS: First I would like to thank the sponsors and Dr. Ouriel for an excellent presentation that clarified a great deal of the information for me. I have a couple of questions and I think the first one of why didn=t you follow the protocol has been answered several times now. I think I have a grasp of that. Not that I have a grasp of the Christmas tree, mind you, but at least I understand that part.

The question I have was the inclusion criteria of high risk patients. One, who decided they were high risk? Was it done by at a committee, by individual who then submitted the patient to the protocol or was it all patients were selected at the time of a group meeting when you had a new patient?

DR. OURIEL: Well, if you mean the eligibility criteria for the trial?

DR. NICHOLAS: Right.

DR. OURIEL: Those were pretty clear cut and they were determined by this three-membered group, surgeon interventionalists and neurologists.

DR. NICHOLAS: But my question was did all three decide or did one of three decide?

DR. OURIEL: All three needed to decide that the patient was eligible. Then there was this secondary decision. The surgeon had to be willing to do an endarterectomy. The interventionalist had to be willing to do a stent for them to get randomized.

DR. NICHOLAS: Okay. The next question I have is again related to inclusion criteria in that you included high risk patients, people who have unstable angina and a high degree heart failure, people who I wouldn=t normally consider for an elective procedure. Did these patients have intervention for their heart disease if we just pick out that group of high risk people before they had their carotid addressed?

DR. OURIEL: Sure. There were a whole subset of patients who would have been treated before they entered this trial and also there=s a subset of patients who would never have entered the trial at all, the 2200 patients screened getting down to 700 patients that were entered for instance. I think some of the patients you=re talking about really never ended up in this analysis because maybe they had no treatment at all.

DR. NICHOLAS: Okay. But for those who did enter with high cardiac risk factors, I assume those problems were treated before you went on to either stenting or endarterectomy for, let=s take an asymptomatic lesion for instance.

DR. OURIEL: If the surgeon and the interventionalist were unwilling to treat the patient because of those factors, then they would not have been in the trial until those were addressed. Does that answer your question?

DR. NICHOLAS: Okay. Thanks.

DR. PENTECOST: Was there a correlation with the size of the stent in the post-intervention stenosis, i.e. were the patients with smaller stents have a higher degree of problems afterwards?

DR. OURIEL: I understand your question. It=s a good question. It=s certainly been shown for instance in renal artery stenosis and restenosis. Do we have an answer to that now? I think it=s not an analysis that we=ve yet done but a good question.

DR. PENTECOST: What about patients that had collateral imaging? I would think a lot of these patients just before they came into the trial had CT angiography and particularly MR angiography, also post-intervention. Were you blinded to that information or how to do that factoring into your decision?

DR. OURIEL: Well, the decisions were made based on duplex ultrasound first irrespective of other tests.

DR. PENTECOST: So if you had a patient with a duplex ultrasound and an MR angiogram, you didn=t look at the MR angiogram.

DR. OURIEL: It wasn=t used in the decision-making process.

DR. PENTECOST: Okay.

DR. OURIEL: But in all respects, the angiogram would have overridden the duplex if it were performed.

DR. PENTECOST: Even MR angiogram?

DR. OURIEL: No, contrast.

DR. PENTECOST: If you used MR angiography or CT angiography at all.

DR. OURIEL: The contrast angiogram or the time of the stent would override the duplex in the decision-making process.

DR. PENTECOST: Okay. That=s all.

DR. ABRAMS: I had some questions focusing on the strokes and particularly on the minor strokes. If I understood correctly, the decision about whether somebody had a minor stroke is they had to have a change on their NIH Stroke Scale essentially. There was no standardization of neurological examinations. Is that correct? Every neurologist just performed whatever his standard neurologic exam was.

DR. COHEN: I=ll ask Dr. Fayad to answer.

DR. FAYAD: The NIH Stroke Scale is a standardized neurological exam that tries as best as possible essentially to make an examination or deficits comparable between different patients. Along with the Rankin Scale and the Barthel Index which the Rankin Scale is a global disability score or outcome score and the Barthel Index which is an activities of daily living scale, all of these three were configured to try to come out with the degree of disability related to the stroke. So the stroke was determined as minor when there were no disabilities related to the NIH Stroke Scale or to the Rankin Scale or the Barthel Index.

DR. ABRAMS: Let me follow up on that with a second. The neurologists were blinded, right, to the procedure that was done and how did you do that?

DR. FAYAD: I don=t think they were blinded.

DR. ABRAMS: So the neurologists knew whether the patient had a stent or a CEA. Okay. Now if somebody woke up after a procedure and said they were a little dizzy or they were a bit foggy, it didn=t necessary qualify for a change on the NIH Stroke Scale. Was anything done? Were they imaged or was it decided that no, this wasn=t a stroke?

DR. FAYAD: It was left up to the neurologist at the center. They had to have neurology. If there were any neurologic symptoms, they had to have the neurologist involved in the study evaluate them and it was left to his discretion for further work-up. If he or she determined that this was a stroke, it was sent to the adjudication committee.

DR. ABRAMS: So basically he had to trigger, initiate, things on the stroke. If he decided that something wasn=t a stroke, it was post-anesthesia.

DR. FAYAD: That is correct. In fact, all patients were examined by the neurologist before they were discharged from the hospital. It was within 24 to 48 hours.

DR. ABRAMS: Yes, it=s a little concerning because I understand at least a large number of patients are going to have five or six minutes perhaps of significant hypotension, Bardycardia. They may have some subtle watershed infarction or watershed changes that might not necessarily show up on the NIH Stroke Scale. If these are looked into further, we could be missing a fair number of ischemic vascular events.

DR. FAYAD: There is always a risk of missing a few things, but there was nothing major missed as they had to standardly be evaluated by the neurologists and the NIH Stroke Scale was only used as a standard measurement. But it did not replace the neurologic evaluation.

DR. ABRAMS: Now there appear to be about 25 or 30 strokes that did occur during the study. These strokes, I presume, were B- These individuals had imaging studies, had modern imaging with diffusion studies. Is that correct?

DR. FAYAD: I don=t have the answer to that. I assume that it was left up to what was the decision of the neurologist, but it was not mandated by the study. So it was part of the care of the patient.

DR. ABRAMS: So the Data Safety and Monitoring Board did not actually review the clinical evaluations of these individuals at any particular point.

DR. FAYAD: Well, it was adjudicated if it was declared again an event. It was adjudicated by the events committee. Then if it was adjudicated, then it always went to the safety committee of course.

DR. ABRAMS: And one last question, post surgery, the individuals with the stents received Ticlopidine or Colpidogrel for two weeks. Was there any standardization of medical treatment going beyond that two week period and did you analyze that or do any subgroup analyses to see if how many people took Colpidogrel, how many people took aspirin or any other stroke prophylactic agent?

DR. COHEN: Yes, we have limited data on medications that were taken at the six month and one year time point, but it=s very limited information. We can get the exact numbers for you. I know that there were more patients in the stenting arms that continued on oral anti-platelet agents, for example, at the six month and one year time point. We did collect data on other cardiovascular medications, but not in a careful fashion that would allow us to make detailed comments.

DR. ABRAMS: Did you look, say, at atrial fibrillation? I didn=t see it as analyzed among the two groups. Did you look at atrial fibrillation between the two groups to see whether there was a difference and whether they were warfarin therapy or things like that?

DR. COHEN: You=re talking about on follow-up?

DR. ABRAMS: Yes.

DR. COHEN: Not on presentation, but on follow-up?

DR. ABRAMS: Yes.

DR. COHEN: No, I don=t believe so.

DR. ABRAMS: Okay. Thank you.

DR. WHITE: Thank you, Warren. I=d like to congratulate the sponsor and the investigators on completing such a trial and I ask these questions in the context of having been a carotid stentor for more than ten years now. We placed our first stent in January of 1994. I=d like to actually make sure I heard what Dr. Ouriel said clearly and that is I think the questions that I=ve heard some of the panel members asking and I think many people would like to know is who are the people that should be treated, but that was not the question that was intended to be answered by this trial. As you=ve said, I think I heard people were committed to revascularization and the question to be answered was how the stent compared to endarterectomy in patients who were committed to have revascularization.

DR. OURIEL: Yes.

DR. WHITE: So it doesn=t answer the question about who should get revascularized.

DR. OURIEL: Correct.

DR. WHITE: I=d like to just respectfully as I can disagree with Dr. Comerota about what he said about MIs. I think MIs are extremely important in the management of patients particularly non-Q MIs. I think they=re an element of all current major cardiovascular trials. You can=t do one today without looking at the instance of non-Q MIs. The fact that they weren=t measured in NASCET and ACAS I think is unfortunate, but in that time period, we weren=t aware of the importance. It wasn=t that somebody said we don=t care about MIs then. It just wasn=t an issue.

It currently is now for the CREST trial. It certainly is important. I think it=s recognized to be an important part of this. Dr. Comerota mentioned something about the Colpidogrel bias in the treatment group and I want to make sure we clarify that. The first point is the Colpidogrel advantage absolute numbers is in single digits for cardiac patients. It=s a risk reduction from nine percent to seven percent, a very small number. It=s not a issue.

The second thing is that patients in this trial were only treated for two weeks or mandated to be treated for two weeks with Colpidogrel so it would be a short interval. The third thing was the stroke rate for surgery as I understand it something that happened peri-procedurally. So it=s wasn=t something so much that happened nine or ten months later that might have benefitted from anti-platelet regime, but something that happened in the 30 day window. Is that true?

DR. OURIEL: Yes, the stroke rate really plateaued after the initial events. Both groups.

DR. WHITE: I wanted to ask you if you had any idea about the issue of debris and the filter and stratified that particularly redo patients because the issue about filters and the necessity of having a filter would obviously depend upon the yield of debris. It would seem to me that an intimal proliferative disease such as failed redos endarterectomy would be a smooth muscle disease as opposed to an atherosclerotic process. Were you able to stratify the debris in the filters by those patients?

DR. OURIEL: We can find out if we have the data. I don=t have the data, but I can tell you anecdotally as, Chris, you already know that it=s not just an intimal hyperplastic lesion on the redo. You also have this patchulous patch with a lot of very friable debris. I=ve been surprised more than once on a redo to find debris in the filter.

DR. WHITE: And I was also impressed by the high rate of Arestenosis by ultrasound.@ We=ve not seen restenosis at these rates clinically. I just wondered. I see Dr. Popma in the audience. I know you did an angiographic follow-up. Was there an correlation between angiographic restenosis and ultrasound restenosis? Was there an over estimation because of increased flow?

DR. OURIEL: While either by Jeff or B Jeff is coming up. I=ll answer that yes, at 50 percent threshold there=s a high rate of restenosis. But if you look really at the clinically-relevant 70 or 80 percent, it=s really very low.

DR. WHITE: Do you believe though, Michael, that you really did have 50 percent restenosis or do you think you were looking at increased velocities?

DR. JAFF: My name is Michael Jaff. I am an vascular medicine specialist in New York City. I=m the medical director of the vascular core lab that was contracted by Cordis for the SAPPHIRE trial. In that capacity, I=m a paid consultant to Cordis.

I=m actually going to show a very small number of data that will answer Dr. White=s question. At the time that the SAPPHIRE trial was being planned and these criteria for initial stenosis prior to treatment, prior to randomization, and follow-up after treatment was designed, we didn=t know the impact of a stent on carotid artery and the duplex velocities that would develop. So we=ve now learned some very interesting data that was just published in January of this year that I think sheds a lot of light on this.

But let me just show you a couple of quick slides that I think will answer Dr. White=s question. This is data from the recently published late last year Society for Radiology and Ultrasound Consensus Conference on Carotid Duplex Ultrasonography. What this slide shows you is that the initial important criteria for stenosis in a non-treated or native carotid artery is the peak systolic velocity (PSV). Without going into the physics of this, as many of you know, the faster the peak systolic velocity or the faster the speed of blood flow the more severe the degree of stenosis. This is shown very nicely on this slide.

However, what you=ll also show is there is a significant amount of overlap between these categories. Carotid ultrasound cannot and has never been touted to be able to identify 51 percent versus 55 percent versus 57 percent stenosis. It categorizes ranges of stenosis severity. In fact, there have been a number of studies that have demonstrated correlations between carotid ultrasound and angiography.

Let me show you the data on restenosis. As you=ve already seen in your Panel Pack, these are the predetermined duplex velocity criteria that were defined for enrollment in the SAPPHIRE trial. You can see here again that the greater the peak systolic velocity the more severe the stenosis and the determinant that=s separated out moderate to severe degrees of stenosis was an additional increase in end diastolic velocity (EDV). These criteria were chosen to be quite conservative to make sure that we identify the patients who had truly 80 percent or greater degrees of stenosis.

When we followed these patients and reviewed this data what we found was that fewer than five percent of patients in the SAPPHIRE trial who were randomized to carotid artery stenting had a diameter stenosis less than 50 percent which demonstrates the excellent accuracy of duplex ultrasound to screen out carotid disease which would not benefit potentially from revascularization based on previously published data.

Now to answer the question about restenosis, there have been some recent elegant studies published from New Jersey from Hobson and others that have demonstrated that once a stent is placed in an internal carotid artery, the compliance of the vessel decreases so that the velocities increase artifactually. In fact, the data that we saw, this slide which you=ve already seen which demonstrated that the duplex defined restenosis of greater than 50 percent was 19.7 percent in the stent group, 31.3 percent in the endarterectomy group demonstrates the high sensitivity of carotid duplex ultrasound, but the likely overestimated degree of stenosis based on loss of compliance and in addition in carotid endarterectomy especially in nonpatched carotid endarterectomy, the same phenomenon has been known to occur.

However, what I would like to again present to the panel is that the area in which carotid ultrasound excels in its accuracy is in the higher degrees of stenosis. In fact, you can see here that the greater than 80 percent diameter stenosis as identified by duplex ultrasound was 0.8 percent in the stent group, 4.2 percent in the endarterectomy group which parallels quite impressively with the actual target lesion revascularization rates.

DR. WHITE: Thank you. I have some questions about the Patient Brochure. Warren, is this the time or do you want to wait?

CHAIRMAN LASKEY: Now is the time.

DR. WHITE: Could you get your Patient Brochure out so I can look at pages and talk to you about some of the crazy things you=re telling our patients? I would like you to look at page seven under section 3.1 of the Panel Pack. It=s section 3.1, Patient Brochure and it=s page seven of that brochure. That top paragraph the last sentence says AYou=ll be asked to take aspirin for one to two days prior to the procedure.@ These are the instructions in general for the patient. I would like to make sure that you add or edit that sentence to include that they will likely be asked also to take Plavix or Ticlopidine before the procedure because at the end on the discharge instructions, you do refer to that. I think you should be consistent.

The next issue is on page nine. You talk about after your procedure. You fail to mention anything about closure devices. In fact in our laboratory, closure devices are used in 80 to 90 percent of our patients. In fact, we use them in virtually all of the carotid stent patients to avoid hypotension from bleeding. So I think at least some verbiage to prepare the patient that they might have a closure device as opposed to the standard reaction would be appropriate.

On the next page 10 in the second paragraph, you reassure patients that MIs are not contraindicated. But later in this document, you mention that it=s only been tested to a testor of 1.5. I think bigger (PH) magnets are now being widely available and so you perhaps want to be more cautious in your reassurance. You don=t want a patient walking in and saying ANo problem. I can have this MR if somebody has a 3-Tesla mag.@

CHAIRMAN LASKEY: Is that true for Nitinol or is it a non-issue?

DR. WHITE: Well my problem is on page 29 of the next section you say that you have not tested the precise stent. I=m sorry. This is section 3.2, the last page of the next section. You say the precise stent has not been evaluated above Tesla 1.5. That=s 12.0 on page 26. So if you=re going to say you haven=t tested above 1.5, I think it=s hard to give a blanket reassurance. It may be true but I think the information ought to be consistent.

CHAIRMAN LASKEY: Or just take it out. It=s irrelevant. Nitinol, non-ferromagnetic, not even close.

DR. WHITE: In the next paragraph under ALifestyle Changes@ this is the interesting sentence. In the middle of that paragraph, you say AThose patients who are able to reduce fats and cholesterol in their diets are less likely to redevelop blockages in the stent.@ That=s fabulous. It just isn=t true. Fix that.

I=d like to go to the next section which is 3.2 APrecise Over the Wire IFU.@ It=s the instructions for use, page number six. You talk again in this section about the indications for use for your procedure, but you don=t actually specify the anatomic or comorbidity issues here for the operator. I believe a table should be put in that actually specifically lists the comorbid and the anatomic criteria for the SAPPHIRE trial so that the operator can follow those carefully.

At the bottom of that page, the last bullet point says AStent placement is not recommended for patients@ and the first one is Awith poor renal function who in the physician=s opinion may be at risk for an reaction to contrast.@ I have a problem with that sentence because we cannot obviously predict contrast reactions. So I would like to modify the phrase in some way in these days of medical liability. I would like for you to give me a way to have physician judgment, perhaps even inserting the word Ahigh@ in front of Arisk@ so that I would not be putting the stent in a patient at Ahigh risk@ for example because everybody is at some risk for renal insufficiency.

Then on the top of the next page you go on in the sub-bullets and you say that AAneurysmal dilatation immediately proximal or distal to the lesion is not recommended.@ In fact, most of my patients have some element of ectasia, either proximally or distally to these bifurcation lesions. So if you want me not to treat aneurisms you have to define what an aneurism is or take it out.

DR. OURIEL: Okay. We can define that.

DR. WHITE: The next line says AIn patients in whom femoral or brachial access is not possible.@ I=m trying to think. Are you telling me that I can=t do this with a direct carotid puncture? Am I in trouble if I do that?

DR. OURIEL: Well, I don=t know if you=re in trouble if you did that.

DR. WHITE: Well, you=re telling me it=s not B- Again I would be careful about the wording because I=m going to get hung by somebody I do direct carotid puncture and has some sort of problem and some plaintiff attorney is going to get their hands on this and say AWe told you not to do that.@ I just want you to think about what you=re telling me not to do.

DR. OURIEL: Sure.

DR. WHITE: You say down under 6.0 on that same page to Aavoid stent placement that would obstruct access to a vital side branch.@ Do you see that under that bullet? My concern is that I routinely drop this stents across external carotid arteries. Is that not a vital side branch?

DR. OURIEL: I would not consider that a vital side branch and we=ll address that line as well.

DR. WHITE: Okay. And then further down, the third bullet down, you say AVenous access should be available.@ Do you see that one?

DR. OURIEL: Yes.

DR. WHITE: ADuring carotid stenting in order to manage the Bardycardia and hypotension.@ When you say that, are you talking about peripheral venous access or are you telling me I have to have a femoral venous sheath?

DR. OURIEL: Peripheral venous access.

DR. WHITE: Perhaps you could say that because I=ll tell you that it=s going to come up. Mitch has already asked about Bardycardia and hypotension. We=ve really gone away from the routine temporary pacemakers. We have more complications from the sticks than we do from the hypotension. So we really don=t want to promote the idea that everybody has to have a femoral venous stick because you don=t want hypotension two hours after you put a carotid stent in somebody from a hematoma. That=s a safety issue.

DR. OURIEL: Sure.

DR. WHITE: That=s all I have. Thank you.

CHAIRMAN LASKEY: Dr. Maisel.

DR. MAISEL: I wanted to focus for a couple minutes on some of the described device problems that occurred during this study. Specifically it=s mentioned that there are eight patients that had failure of stent delivery, although most of these ultimately had successful delivery. Another eight patients had difficulty in passing or retrieving the ANGIOGUARD. In the registry, an additional 25 patients had ANGIOGUARD delivery failure. Could you comment and provide a little more detail about what were the reasons? Particularly with the ANGIOGUARD delivery failure, what actually was limiting the delivery of the system?

DR. OURIEL: Well, I can=t tell you exactly on those particular patients, but I can tell you that in general especially with previous versions of the ANGIOGUARD, sometimes a lesion could not be accessed very easily especially in the early portions of this trial. That=s basically why an ANGIOGUARD may be difficult. If you ding up the wire, then you=re going to pull it out and get another ANGIOGUARD and that would be an instance where initially the deployment of the ANGIOGUARD was not possible, but eventually it was.

DR. MAISEL: When you say Acan=t be accessed easily@ do you mean it=s stenotic, it=s narrowed, tortuous? What specifically are you referring to?

DR. OURIEL: All of the above. Angulations of the internal, tight lesions. It can be difficult to get any wire through a tight carotid lesion.

DR. MAISEL: And so were most of the ANGIOGUARD delivery failures then early in the registry and early in the randomized trial?

DR. OURIEL: We can look at that slide again and see exactly comparing the FEASIBILITY to the randomized trial. Let=s see if you have that on one of your B-

DR. MAISEL: Simply because you=re suggesting it may have been related to inexperience or trouble with the actual device.

DR. OURIEL: Well, I think that was subsequently modified.

DR. COHEN: There are two aspects to the answer. The first one is the patient=s anatomy with access toward tortuosity, lesion, severity, etc. The second one is that as we have mentioned early in the presentation there have been several generations of the device and the major improvements have been to improve deliverability and lower the profile of the device.

DR. MAISEL: Understanding that subgroup analysis is fraught with danger, I=m going to delve into it a little bit, seeing as Dr. Laskey is several seats away. I=m particularly interested in talking about the asymptomatic patients. I am struck as was Dr. Tracy the high stroke rate particularly at 30 days for these patients that were asymptomatic.

I certainly recognize as you=ve mentioned several times that there=s contemporary data to know what that rate would be in these higher risk patients. It relates a little bit to the ANGIOGUARD delivery system question. I=m struck by the fact that the asymptomatic stroke rate is higher than the symptomatic stroke rate.

I wonder whether that might be due to delivery of the system in more stenotic vessels or to stent delivery in an 80 or 90 percent stenosis rather than a 50 or 60 percent stenosis. That might result in more distal embolization. Do you have any data either from the pathology, from the ANGIOGUARD devices or from analysis of the relative stenosis and complication rate?

DR. COHEN: The first part of the answer is we obviously have expressed our disagreement with any suggestion that the event rates are different. The number of patients in each group is low. One or two patients one way or the other can have marked effects on the rates when you=re looking at subgroup analyses given the size of the samples. The second part to your question is no, we don=t have any data.

DR. MAISEL: I would also simply comment that the design of the study I think could have anticipated some of these issues. Grouping symptomatic and asymptomatic patients which have clearly different risks and clearly the acceptable safety margin would be different in these populations could have been anticipated. To my knowledge there was no stratification of symptomatic or asymptomatic at the time of randomization that might have helped address some of these issues.

DR. COHEN: That=s actually not true. The patients at the time of randomization were stratified as to their symptom status. Just again, the goal of this trial was not to provide detailed information on subgroups. It was to analyze the primary endpoint.

DR. MAISEL: The other comment I=ll make and I simply would like to say that I certainly recognize that it appears that the stent group did at least as well as the CEA group overall, although we can debate the statistics. What=s not clear to me is how these patients would compare to best medical therapy. We were not provided really any data on the medical therapy that these patients were receiving which I think is a critical aspect of the care of these patients.

Maybe you can just clarify. You mentioned that you have some six and 12 month data. Can you show us anything that shows what percentage of patients were receiving anticoagulants which can reduce stroke rates by maybe as much as 20 percent? How many were on statins or other lipid-lowering therapy that can reduce stroke rates by 20 or 30 percent in these patients?

DR. COHEN: I can give you a limited amount of data and unfortunately it will have to be verbal. I don=t have it in a slide. At the time of discharge in terms of the category of anticoagulants, in all this data I will give the stent data first and then the carotid endarterectomy or the surgery second. This is only for the randomized portion of the trial. For anticoagulants, 31.6, 43.2 percent. For antiplatelet agents, 98.1, 71.6 percent. For beta blockers, 41.3, 59.0 percent. For lipid-lowering agents, 69.9 percent, 65.5 percent. That=s at discharge. At 30-day follow-up, it was 94.3 percent. I=m sorry. This is for clopidogrel specifically. 28.8 percent.

DR. MAISEL: Thank you. That=s very helpful.

DR. NAJARIAN: Hi, just a few questions.

DR. MAISEL: I=m sorry. Can I just ask one more question? During the presentation, I can=t remember which of you mentioned it, but while showing some of the Kaplan-Meier you, on numerous occasions, mentioned Amedian survival.@ Sometimes that was three, four, five, six years. Can you explain to me how you had a median survival of that long in a trial that was this short?

DR. OURIEL: How it was calculated?

DR. MAISEL: Yes.

DR. COHEN: I=d like to as Joe Massaro to come up.

DR. MASSARO: I=m Joe Massaro. I=m with Harvard Clinical Research Institute, managing director of Biostatistics and Data Management. I also am an assistant professor of biostatistics at Boston University. Cordis paid for my travel down here and my lodging. Other than that, I have no financial interest in Cordis.

Basically it=s all extrapolated. We took the one-year survival data and got an estimate of the one-year maze rate for each group and then just extrapolated that over the course of time until we came up with an estimate of 50 percent of the patients would be alive.

DR. MAISEL: So it=s safe to say that those really are B-

DR. MASSARO: It=s estimated.

DR. MAISEL: Okay. Thanks.

DR. NAJARIAN: My turn? Just a few questions. In both the carotid endarterectomy cases and the stent cases, what percentage of each group had general anesthesia versus local anesthesia?

DR. OURIEL: For carotid endarterectomy, about 91 percent of the patients had general anesthetic.

DR. NAJARIAN: And for the carotid stent arm?

DR. OURIEL: It was done under local.

DR. NAJARIAN: All under local. Including the 406 patients in the registry?

DR. OURIEL: Yes.

DR. NAJARIAN: Great. Another question. I was just looking at in-hospital complications of both groups. The randomized arm for the carotid stenting, there were five out of 159 patients who had a stroke. In the stent registry, ten out of 406 patients had a stroke in hospital. Do you have any data regarding those strokes like what percentage were hemorrhagic and what percentage were ischemic? Were any of the strokes since they did occur in hospital treated interventionally with tPA or thrombolysis?

DR. COHEN: If I could ask you to divide your question up. There were two parts to it. The first part again?

DR. NAJARIAN: Two parts. I=m sorry. The first question is of the 15 strokes from all stent patients what percentage were hemorrhagic versus bland.

DR. COHEN: Yes. I know that there was, I believe, a total - now this is in the entire trial over the first year - of three hemorrhagic strokes. We=re going to check right now how many were in hospital.

DR. NAJARIAN: I=m sorry. How many were hemorrhagic?

DR. OURIEL: Four or five were hemorrhagic. I can=t tell you which treatment arms they were in. None were treated to our knowledge with thrombolysis.

DR. NAJARIAN: So approximately there were five or six bland, non-hemorrhagic strokes then that weren=t treated.

DR. OURIEL: Approximately, but those numbers aren=t exact.

DR. NAJARIAN: Okay. This will come up at some point. I=m not sure this is the appropriate time to mention it, but in your training criteria, it says AThe sponsor has proposed a training program called CASES. This program must be completed prior to shipment of any device to each center. This program will be tailored to meet the needs of each physician with more intensive training for those with little or no experience.@ The Amore intensive training@ that you=re referring to, is that the third major bullet down?

DR. OURIEL: I=m sorry. Could you tell me what page you=re looking at?

DR. NAJARIAN: I=m sorry. Page seven. Actually I=m in the FDA packet, AIntroduction FDA Questions No. 1.@

DR. OURIEL: I=m sorry. Is this in the Panel Pack you=re talking about?

DR. NAJARIAN: Yes, the Panel Pack.

DR. OURIEL: The Panel Pack. Is this the printed page number or the stamped page number on the page?

DR. NAJARIAN: The one that refers to training, page seven. It=s also probably in your pack as well.

DR. OURIEL: I have a page seven that has a diagram on it. It starts out AA Qualified Physician@ with an arrow down to AOne online didactic session.@ Is that what you=re referring to?

DR. NAJARIAN: Maybe we can=t find the same page, but regarding training.

DR. OURIEL: Okay. I believe we have it.

DR. ZUCKERMAN: You=re talking about the last question.

DR. NAJARIAN: Yes, regarding the training that you specified here. I was just wondering. It says with patients Awith little or not experience.@ Is that with stenting?

DR. COHEN: No, specifically with carotid stenting. Perhaps the explanation that will make this clear is to turn it around the other way. What we did in this program was to provide less intensive training for those physicians who were experienced in performing carotid stenting either with or without the Cordis devices. If they had experience of 25 cases of stenting and ten with the Cordis devices, then we considered there was a need for minimal training in those individuals. If there were people who were experienced in carotid stenting but specifically not in the Cordis devices, then they needed to have training in the Cordis devices. For other physicians, they needed to have the full program.

DR. NAJARIAN: Okay. Now there is no experience here for documented cerebral arteriography. I was just wondering. Does that mean that a physician at a hospital, any physician, could go through this training course and have devices shipped to them?

DR. COHEN: Again, it=s up to the hospital to decide whether or not a physician is allowed to perform a procedure. We=re not involved in the credentialing process. All we are doing is assuring that people who would be using our devices have been adequately trained in that procedure.

DR. NAJARIAN: In that device.

DR. COHEN: Right.

DR. NAJARIAN: I=ll just ask. Do you think that minimal training in carotid access should be necessary?

DR. OURIEL: On a local level?

DR. NAJARIAN: On a local level.

DR. OURIEL: Outside of the scope of this, I think yes, someone needs training. In carotid access, I don=t think diagnostic angiography should be encouraged however as a mechanism to get training if you weren=t going to do it anyway.

DR. NAJARIAN: That=s all of my questions. Thank you.

CHAIRMAN LASKEY: Are you sure?

DR. NAJARIAN: Yes.

CHAIRMAN LASKEY: Okay. We=re two minutes past an official break time. Did you want to requery? You had some additional quick question and I had one quick question and then I think we should break before the rounding third, heading home.

DR. COMEROTA: Sure. Hopefully these will be quick. Dr. Cohen, if the FEASIBILITY study results exceeded the projected major adverse events for carotid endarterectomy which were calculated into the trial - so if you had a projected event rate more than two times that of CEA - would you have proceeded with the randomized trial?

DR. COHEN: I=m sorry. You=re asking a hypothetical question.

DR. COMEROTA: Yes.

DR. COHEN: And this is based on the U.S. FEASIBILITY Study?

DR. COMEROTA: Correct.

DR. COHEN: Okay. There was this formal stopping rule which was a twofold rule. I forget the formal name of the rule.

DR. COMEROTA: Correct.

DR. COHEN: And that rule was followed. So we followed the rules of what was agreed upon by both the FDA and us as being appropriate.

DR. COMEROTA: Okay. These are high risk patients. We=ve more or less agreed. At least, they have high risk characteristics. Do you think these are high risk carotid lesions?

DR. OURIEL: Do you mean are they high risk carotid lesions for stroke?

DR. COMEROTA: Correct.

DR. OURIEL: I=m not sure I understand.

DR. COMEROTA: Correct.

DR. OURIEL: Well, I really don=t know the answer to that, Tony, but I can tell you I just don=t know that answer to that. But I know what you=re thinking and we=ve thought the same thing that a patient with diffuse atherosclerosis with a 70 percent carotid lesion may be more likely to have an event than a patient without diffuse atherosclerosis and a 70 percent carotid lesion, but no data.

DR. COMEROTA: Okay. The results as we=ve mentioned before, these were top notched physicians involved in this trial. We see what the data are. When this gets out, where do you think the real world event rate is going to lie when all interventionalists begin to use this?

DR. COHEN: I think the best answer to that is to refer you to the details of the post-marketing surveillance study. We=ve communicated with the agency and the desire here is actually to obtain data in a variety of settings including both academic and non-academic centers, a geographically diverse number of hospitals as well as low volume, intermediate volume and high volume operators so that we will gain insight into whether the training we are doing is adequate and whether or not there is a threshold as I believe what you=re alluding to that would have an effect on safety.

DR. COMEROTA: Do you have information on the results of brain imaging that was performed on the patients in both arms of the trial post-procedure?

DR. COHEN: No, that was not part of the formal trial.

DR. COMEROTA: But when they were performed, you don=t have that information.

DR. COHEN: We don=t have it. No.

DR. COMEROTA: Okay. I know that with all due respect, Warren, subgroup analyses and so forth you did say that these were stratified up front asymptomatic versus symptomatic. I mean we make our clinical decisions in very large part on whether the patient is symptomatic or asymptomatic and the behavior of those lesions and the degree of stenosis drives what we do. It has a direct impact on the benefit that the patient receives from the procedure or lack of benefit.

The absolute reason why any procedure is deemed beneficial is when it=s performed at a very risk of an event. That defines how these patients are going to be identified and who will benefit from that technique. Just let me get your thoughts on these high risk patients with potentially low risk lesions, why they weren=t considered to be put into a medical management arm?

DR. COHEN: Then I would refer back to what the major goal of this trial was. It was not an NIH trial looking at trying to understand a disease process, event rates, etc. It was a trial looking at patients who are treated in the United States and I think we=ve provided sufficient information from the literature and from different databases to support that. These patients are being treated today in the United States and what we were studying specifically was an alternate form of treatment that we believe from the data is less invasive and in some regards safer. It offers an alternative therapy for these patients.

DR. COMEROTA: Well, I understand. I understand exactly. But our panel is charged with identifying whether this is appropriate or if the global idea isn=t appropriate, what=s the niche for the technique. Without that information, it=s difficult to get to that answer. This is a discussion that I=m sure we=re going to have after the break.

What we do know is the stroke rate in a very large study, unwritten national trial, medical treatment from 10 to 15 years ago or 15 and more years ago - we know that medical treatment is better today - that 30 day stroke rate in those patients treated medically is 0.3 percent. 0.3. We know what the 30-day stroke rate in symptomatic patients is. Well, all the cerebral vascular events was 3.3 percent. Stroke and death, neurologic deficit and death, is around 1.0 percent in symptomatic patients. The only way that any technique can equal that over time is to have a very, very low event rate. So I would give you across the board that these patients should not be operated. They should not have a carotid endarterectomy.

DR. COHEN: Well, I can understand and respectfully I understand your opinion in that regard. However, again what we have shown is that these patients are being intervened upon in the United States. Our goal here wasn=t to determine whether that=s appropriate or inappropriate, but to point out that they are being treated and to allow an alternate therapy to be compared to the current standard of care.

DR. COMEROTA: And if we approve that, then you are asking us to put a stamp of approval on this therapy.

DR. COHEN: As an alternative to the procedure B-

DR. COMEROTA: As an alternative to another inappropriate form of therapy.

DR. COHEN: Sorry.

CHAIRMAN LASKEY: I think (a) it=s time for break. Let=s cool off. (b) Thank you very much. We realize that this last ten minutes has been speculative. We appreciate your patience. Thank you, Tony. I have a quick question for Nigel. Can I catch you on the break? Thank you. I have 3:55 p.m. Can we reconvene in ten minutes please for the remainder. Off the record.

(Whereupon, the foregoing matter went off the record at 3:56 p.m. and went back on the record at 4:14 p.m.)

CHAIRMAN LASKEY: Thank you again.

Before we finish up here and move to the questions and the panel vote, we do need to reopen the open public hearing portion of this afternoon's meeting and prior to that, Ms. Wood wants to read a statement into the record.

MS. WOOD: This is in addition to the conflict of interest statement. We would like to note for the record that a waiver has been granted for Dr. William Maisel. His imputed waiver involves a contract to his institution for the sponsor's study in which he had no involvement in data generation or analysis. The waiver allows Dr. Maisel to participate fully in today's deliberations.

A copy of this waiver may be obtained from the agency's Freedom of Information Office, Room 12A-15 of the Parklawn Building.

CHAIRMAN LASKEY: And the first speaker requesting time for the open public session is Dr. Robert Hobson.

Dr. Hobson.

I will remind this afternoon's speakers that they're limited to ten minutes on the clock as well, just like this morning.

DR. HOBSON: Thanks very much, Dr. Laskey.

It's a privilege for me to make a few comments on behalf of the CREST investigators, a group that is studying the efficacy of endarterectomy versus stenting in a good risk sample of patients. That's supported by the National Institutes of Health.

I have no relationship with the Cordis J&J Company. I have received modest support from the Guidant Corporation in association with the CREST trial.

These are data on the impact of clinical trials on a number of carotid endarterectomies performed in the United States annually. On the vertical axis is the number of endarterectomies in thousands, and on the horizontal axis, the year of interest. These data are from the Dartmouth Atlas of Vascular Health.

And notice that in 1985, the number of carotid endarterectomies performed in the United States was about 100,000. With the publication of a trial that had little to do with extracranial carotid stenosis, the EC-IC bypass trial, there was a substantial decrease in the number of cases just by association.

In other words, this was a trial primarily interested in internal carotid occlusion in which a branch of the external carotid, the superficial temporal, was anastomosed to a temporal branch of the middle cerebral through a temporary trephine, and even though there was no difference demonstrated between that bypass and even though it stayed open in over 90 percent of cases, it showed no difference with medical therapy, and as a result, there was a substantial and significant decrease in the number of endarterectomies performed until publication of the symptomatic trials in '91, the North American symptomatic carotid endarterectomy trial supported by NIH, the European carotid surgery trial in Europe, and the VA symptomatic trail.

And we are back up to about 100,000 operations in 1994-5, with the publication of ACAS and the VAA symptomatic trial, up to the '96 data of about 140,000 operations, which has been sustained out to 2000 now.

Now, the CREST trial is supported by the Neurology Institute of the NIH and looks at this same question being asked by the SAPPHIRE trial. What is the efficacy of endarterectomy versus stenting in a conventional risk group of patients, that is, the 80 percent of patients treated by surgeons who do carotid endarterectomy.

This particular trial then looks at the better risk patients and uses devices supplied by the Guidant Corporation, the ACCULINK nitinol self-expanding stent, and the ACCUNET anti-embolic device.

The CREST Executive Committee is populated by many of the speakers who were also involved in the SAPPHIRE trial. These are the experts in the field. Tom Brott, the co-PI for CREST, is a neurologist. Gary Roubin is co-PI for intervention in cardiology; Bob Ferguson for intervention. Nick Hopkins is the neurosurgeon and Wes Moore the vascular surgeon. Rick Kuntz has done the data management in the past at ACRI, and George Howard is the biostatistician. Jeff Popma runs the core lab in angiography, and Kirk Beach has taken Gene Strandness' position as the co-PI for ultrasound.

The trial is being conducted at 70 centers in the United States and Canada and is wrestling with the issue of recruiting patients to a trial that is randomizing symptomatic patients only. However, progress is being made, and during the last month we had our most productive month with over 30 patients randomized in the treatment categories.

However, I recognize that on my own service, a group of vascular surgeons that perform carotid endarterectomy as well as stenting, that these are our current indications in higher risk patients for the performance of carotid artery stenting. Carotid restenosis was our first subset of patients treated because it is acknowledged at low risk for neurological events, and therefore, it's the ideal training case for a new interventionalist.

High risk patients, radiation induced stenosis, and there are very few anatomically inaccessible lesions at the C2 or above.

On our service we initiated a program in carotid artery stenting in September of 1996, and we've done 204 cases now over that seven-year period. During that same period we have done 885 carotid endarterectomies. So this constitutes about 20 percent of a vascular surgeon's work load currently. So we are sympathetic to the clearance of indications for carotid artery stenting.

As part of our work-up through CREST, we wanted to look at the published data on the prospective analysis, that is, the randomized clinical trial data comparing carotid artery stenting and endarterectomy, and these are the published reports. I know the focus of our discussion today is SAPPHIRE, and I'd like to make a few comments on that.

I would suggest, however, that the data from Schneider and SAPPHIRE are only available to us today by abstract format, and although the other three trials are published, none have had much in the way of outreach one way or the other in terms of preference for stenting.

Now, as an observation, sitting in the audience today, I think the SAPPHIRE investigators right over here did exactly what this group told them to do: go out and do a noninferiority trial.

And the dilemma you must have on this panel is the results are an extraordinarily small sample, and I wouldn't want to be in your position. And although we will live with your decision, it's a problem.

As abstracted originally after the original report at the American Heart, there were 307 patients. Thirty-two percent were symptomatic. That's 96 patients. Now, recall my first slide that had its data on the impact of a clinical trial. It does change clinical practice. What we're suggesting is that we change a clinical algorithm established over the last two decades based on the analysis of 96 symptomatic patients, and I would submit that a great deal of the discussion this morning regarding things that I certainly didn't understand in terms of triangular biostatistics come down to a very small number of events.

If less than five events swung one way or the other it would change the result of this very small trial, and my lament to you is that we haven't provided a higher standard for the SAPPHIRE investigators to follow. Superiority trials have been used in the impact on clinical trial algorithms, and the smallest sample size was over 1,700 patients. So it gives you an opportunity to focus in on stroke.

After all, we're interested in stroke prevention, and there was no significant difference between those two treatments in terms of stroke prevention. It's very possible, I think, that the SAPPHIRE trialists have identified a subset of patients that should be treated by neither endarterectomy nor carotid stenting.

What's the impact of medical therapy? It would have been magnificent. This trial would have been a ground breaker if you had included a medical therapy arm, and I know that wasn't your goal, and I can understand that you wish you had done that.

Five-year survival data with such a large number of asymptomatic patients would have been nice, too.

CHAIRMAN LASKEY: Sir, you have one minute.

DR. HOBSON: I've already commented on the restenosis issue.

So with regard to the future of carotid stenting, the CREST investigators can live with approval of a device provided the data driven introduction of carotid stenting is confined to SAPPHIRE-like patients.

The challenges, I think, have been covered very nicely. One has been an optimal trial where you've done what the federal government asked you to do. There's no data on medical therapy which is unfortunate.

I was reassured by Dr. Cohen's presentation that the FDA can probably define post PMA surveillance, that the FDA can monitor the results of carotid artery stenting at trial and registry centers, as well as centers just introducing carotid artery stenting, and I'm pleased that the interventionists will be trained in a way that is essentially comparable to the recommendations made by many experts in the CREST trial.

And if these things are followed, if these challenges are met, then I think that we can proceed with a randomized trial on conventional risk patients. My concern, in conclusion, would be that approval of this device based on a very small number of patients might in any way interfere with the proof of purpose trial, the CREST trial.

Thanks very much.

CHAIRMAN LASKEY: Thank you, sir.

The next individual requesting time is Dr. Ku.

DR. KU: Thank you.

My name is Andrew Ku. I'm representing myself as an individual. I'm an interventional neuroradiologist in practice since 1991.

Okay. Disclosure. No current ownership or shares in the company being discussed or competitors at the present time. I have had prior ownership and probably will consider them in the future, but mainly as an investor.

Currently my travel reimbursement is paid for by myself. I am not representing anybody. There is a possibility I may be reimbursed by ASITN, as I'm also an observer for them.

The current data that has been presented shows that the PRECISE stent and the ANGIOGUARD XP distal protection device may be as safe as surgery in high risk surgical patients. I don't believe that the current data shows that the precise stent and ANGIOGUARD XP distal protection device is safer than medical therapy in asymptomatic high risk patients, as defined in the SAPPHIRE study. In fact, most studies that have covered asymptomatic patients probably show that this option is potentially inferior.

So that's my major concern because the study does include asymptomatic patients.

This is just a review of the data that was presented earlier. Of interest is even though it's not statistically significant is the information that's highlighted in red, which covers the combined major mortality/morbidity rate, comparing stents in symptomatic patients versus endarterectomy patients. It is significant lowers. Well, it's not significantly. It is lower, but not statistically significant in stent patients.

So for patients who are symptomatic, i.e., having TIA, had recent prior stroke in that ipsilateral territory, it seems like it works pretty well.

Asymptomatic patients, same data. The numbers are pretty similar. So you've got to wonder, you know. Is there a higher risk for these asymptomatic patients when you use the device? Is it worth it?

The information that was presented by the FDA, reanalyzing the data that was submitted basically shows the same thing. So I will go through that very quickly.

But basically it shows that the patients who are symptomatic were significantly benefitted by the device, and the patients who were asymptomatic didn't really do much better than endarterectomy.

There are trials. There's the University Medical Center, and then there is the real world. Most of these studies, including all of the symptomatic and asymptomatic studies have been performed in major medical centers with the best trained physicians in the world.

For example, carotid endarterectomy in NASCET, ACAS, SAPPHIRE. They were all done by physicians who were very experienced. You didn't have somebody who was doing three endarterectomies a year doing these, in general.

Review of endarterectomy results from Medicare data shows that the outcome is not as good in patients who are sampled from Medicare data, probably because there are a lot of patients being treated by physicians who don't do many procedures, or there's a lot more variability in, you know, the patients' clinical conditions.

So I think we must be very cautious about comparing all of these trials versus the real world. I would hope that you would be very conservative in your analysis and allow the needed margin of safety that real world conditions demand.

There are two major causes of stroke, either embolic or nonembolic. Nonembolic I mean by thrombosis of the vessel. So basically if you have a diseased blood vessel, there's plaque potentially. There could be a piece of stuff that forms and that breaks off.

The other possibility is that due to flow restriction, there's in situ thrombosis of the vessel, and I think that most strokes fall into these two categories.

The NASCET study, which is a study that covered symptomatic patients, did show benefit from surgical therapy over medical therapy.

Studies in asymptomatic patients, however, are somewhat different. There have been at four major randomized trials covering medical therapy to endarterectomy. Three out of the four showed no positive benefit from endarterectomy. The Mayo Clinic asymptomatic carotid endarterectomy trial was prematurely halted due to safety issues from the surgical arm. Only ACAS was positive in showing benefit from surgery.

However, if you look at the actual data, the five-year medical history of ipsilateral minor stroke was 11 percent or 2.2 percent annually. The surgical arm reduced that risk, but if you analyze how many endarterectomies are needed to prevent one minor stroke, it was about 83 endarterectomies, and the data didn't show any significant prevention of major strokes.

So if you look at it after you subtract out the natural medical portion of the complication rate or stroke rate, surgery gave basically a one percent per year improvement over those five years.

Eighteen-year follow-up, they had a 17 percent stroke rate, of which the contralateral non-stenotic stroke was nine percent. So if you take out that nine percent and divide it by 18 years, the benefit was about half a percent per year.

And the Canadian Stroke Consortium in their analysis found that they did not have any indication for endarterectomy for any level of asymptomatic stenosis.

Part of the NASCET trial reanalysis, one of the comments was most individuals with asymptomatic disease fared better with medical therapy. So here we have a SAPPHIRE trial where two-thirds of the patients were asymptomatic, and they were being treated. This was presented, I believe, at the American Stroke Association meeting. It's not published.

They did a reanalysis of ACAS and the reanalysis showed no statistical improvement with endarterectomy. So maybe there is improvement. Two, point, three percent, you have to do better than 2.3 percent complication rate to show benefit. Maybe not.

I have a major concern with the current indication for use with this particular device. As I stated in your packet, my major concern is in the patients without neurologic symptoms and with greater than or equal to 80 percent stenosis. It would make common sense that if something is really narrowed, you would have a high risk of stroke, but it's not proven by the data.

So I feel that if you're going to operate on these patients or stent on these patients, you're taking a lot of risk for very little or negative benefit, and I think that we would do a disservice.

So since I do use carotid stents, I remember one of my patients in 1991 died because she had bilateral carotid dissections from a car accident. She had multiple injuries, and they could not anticoagulate her. I did an angiogram, made the diagnosis. She was neurologically normal. She was dead the next day from bilateral infarcts.

So I think that there is a very strong indication for a device for certain patients, but I would like the labeling to contraindicate the use of this device in asymptomatic carotid disease because I don't think the data that is available justifies it, and I think that if we allow people to use it as a physician driven device, in this particular instance it may be a severe mistake.

I know that the FDA doesn't like to regulate the practice of medicine, but if that language is in there, the lawyer certainly will.

I would like to see a PMA showing safety and effectiveness prior to use in an asymptomatic device. I think that device may have potential. Things are improving every single day. Our devices are getting better. The techniques are getting better. So there probably will be a point where our complication rate is low enough that this device will be useful.

So we as physicians or I as a physician feel that I must treat the patient and not the angiographic procedure or angiographic picture, and I think that most clinicians will agree with that.

If you want any comment on training, I could do that, but that's a separate issue.

Thank you very much.

CHAIRMAN LASKEY: Thank you, Dr. Ku.

The remaining speaker who has elected time is Dr. Rodney White.

DR. WHITE: Thank you very much.

My name is Rod White. I'm a vascular surgeon from Torrance, California. I'm here today representing the Society for Vascular Surgery in lieu of a letter that Dr. Green submitted.

My own conflicts, I paid my way to this meeting as an observer. My greatest conflict is I'm a practicing vascular surgeon who does both open carotid endarterectomies and carotid stents, and I make my living doing this. So any of us who are here to tell you that that isn't a conflict, I think it is probably from both directions.

Now, Dr. Green had submitted a letter to be read here today, and then was unable to attend. So I am going to read this letter for him as the Secretary for the Society for Vascular Surgery. This is his letter from the SVS, and as I read it, this will obviously be him speaking rather than me.

"I am current President of the Society for Vascular Surgery. I am also a practicing vascular surgeon and Chairman of the Department of Surgery at Lennox Hill Hospital in New York City.

"I am formally requesting an opportunity to speak to the FDA panel concerning approval of the Cordis ANGIOGUARD and PRECIS stent systems for carotid angioplasty and stenting. I do so representing a group with vast experience in the management of patients with cerebral vascular disease.

"The Cordis Company and principal investigators are to be congratulated for designing and conducting a randomized trial comparing carotid stenting with embolic protection to endarterectomy in a selected group of patients considered at high risk for endarterectomy. They hypothesized an equivalence between stenting and endarterectomy in this defined subset of patients, and their data appear to support their contention.

"While the definition of high risk use in the SAPPHIRE trial is not uniformly accepted by all vascular surgeons, we do agree that certain patients are likely to benefit from carotid angioplasty and stenting when performed at a level of expertise similar to that of the trialists. These patients include those with contralateral laryngeal nerve palsy, a history of radiation therapy to the neck, previous carotid endarterectomy with recurrent stenosis, and those with medical co-morbidities that might adversely affect the outcome and the opinion of surgeons, interventionalists, and anesthesiologists responsible for the patient.

"We believe, however, that this cohort of patients in SAPPHIRE represents a small percentage of those in the general population currently undergoing carotid endarterectomy and that this study is not reflective of current national practice.

"We cannot overstate how important we regard trials with expanded indications powered sufficiently to allow the data to determine any subsequent expansion of indications for usage. Pending results of large scale experience from a single arm registry or dual arm randomized trial with independently adjudicated one-year outcome data, there are little data to support the use of carotid stenting in lower risk patients. We are concerned that because the differentiation between high and low risk is not always clear and the monitoring and usage nearly impossible that the procedure will be utilized in patients not adequately studied.

"The adjudication of high risk is best done by a collaborative decision making process, including multiple physicians and a surgeon that performs carotid endarterectomy.

"If approved, carotid stenting should be performed by those operators with expertise not just on technical aspects of delivering a stent to a target, but on all of the pre and post procedural components carotid endarterectomy requires.

"This means that a thorough knowledge of the natural history of carotid bifurcation disease, medical co-morbidities, possible neurologic consequences of both stroke and reperfusion and the ability to provide post procedural care necessary in addition to the requisite technical skills.

"We are concerned that no one interested group of physicians has expertise in all of these areas. Multi-specialty coordination and cooperation will be required to achieve outstanding outcomes that we all deserve.

"Because carotid stenting is a new procedure to the majority of vascular surgeons, interventional cardiologists and interventional radiologists, training and credentialing presents a unique challenge. Each of the vested subspecialties has a different skill set and knowledge base. Specifically, there are groups that have more expertise in catheter aspects of carotid stenting, groups that have more expertise in diagnostic components, namely cerebral angiography, and those with more expertise in the management of patients.

"No one of these ingredients is more or less critical to successful outcome. We are encouraged that many interested processional societies are working collaboratively for the creation of CPT and ICD-9 codes and to address national noncoverage decisions for carotid stenting over the past nine months.

"We are discouraged that similar collaboration has not occurred regarding training, competencing and credentialing standards. The tendency of each group to emphasize its strengths and minimize its weaknesses relative to carotid stenting is self-serving and not in the best interest of patient care.

"It is critical that each of the representative societies establish it own set of responsible guidelines for credentialing requirements with the understanding that the final decision will be made locally.

"We believe that anyone who wishes to perform carotid stenting should possess a minimum of skills associated with the advanced interventionalist regardless of the target lesion treated. Certainly a familiarity of the anatomy and behavior, the cerebral vessels is essential, no less so than coronary, renal, or lower extremity vascular anatomy. Many of the skills and tools required to perform renal and superficial femoral artery angioplasty and stenting are transferrable to the extracranial circulation.

"Other variables being equal, practitioners experienced in coronary, renal and lower extremity and subclavian interventions will require fewer procedures to become proficient in carotid stenting. Those without such experience will require many more procedures.

"Vascular surgery training requirements have not in the past included a minimum number of cases as a requisite for certification. Rather, a curriculum is approved. Training programs are reviewed for competence, and individuals are certified and then qualified.

"We believe that the case numbers are less relevant than demonstrated competence. We agree conceptually with the certification process developed by CREST investigators whereby performance parameters are included in the determination of competence.

"Lastly, I would like to comment on the proposition that an arbitrary number of diagnostic cerebral angiograms be a prerequisite credentialing requirement for carotid stenting. The panel should be aware that there remain relatively few indications for diagnostic angiograms performed as sole procedures. To create any threshold for training on this basis creates an unacceptable risk to patients, and there is a definite instance of stroke from the diagnostic procedure alone, irrespective of an intervention.

"Further, any diagnostic procedures do not provide experience in the more complex techniques, such as guide/sheath cannulation of the common carotid artery, use of embolic protection devices and stent deployment. We would hope that the neurointerventionalists would agree to collaborate in the care of these patients rather than to create artificial and potentially dangerous barriers.

"In conclusion, the society for vascular surgery is supportive of the efforts bringing this new technology forward. While we still believe that carotid endarterectomy is appropriate in the majority of patients with carotid artery stenosis and indications for intervention, we support the judicious use of carotid artery stenting in bona fide high risk patients.

"We recognize the challenge of introducing this technology in the larger community and will continue to work with our medical colleagues in industry to achieve our goals to improve patient care."

This is submitted, "Sincerely, Richard M. Green, President, Society for Vascular Surgery."

Thank you.

CHAIRMAN LASKEY: Thank you, sir.

Is there anyone else that wishes to come forth? Yes, sir.

DR. DALL'OLMO: Thank you.

My name is Carlo Dall'olmo. I am a community based vascular surgeon in Flint, Michigan, member of a ten-man vascular group, and we have performed up to 400, 450 carotid endarterectomies a year, and over the past several years have been involved in the four carotid stent training trials.

I believe that carotid stenting is an exciting new therapy with several important questions which remain to be answered because the applicability of this procedure will depend on the answers.

The first is the two questions that carotid stenting initially raised. Sine carotid stenting hit the scene, there were always two questions that had to be answered. The first was: could it be done as effectively and safely as a carotid endarterectomy? And the second was: what is its durability? Is it as durable as a carotid endarterectomy?

I think that referable to the first question, what we heard today is that, yes, it can be done safely. So the initial step is there.

However, what is the longevity of this procedure? What is its durability? Is three years enough follow-up to answer this question?

I don't believe that's enough. I don't believe that data is in. Carotid endarterectomy is a durable procedure, and it's not unusual to see patients who are ten, 12, 15, and 18 years post carotid endarterectomy walking into your office. Certainly in the community we see them in our churches. We see them in our stores. We see them on the streets.

The second question is whether or not the data on 334 patients in the SAPPHIRE in a randomized arm is enough to extrapolate to an entire population on a broad label. I believe more data is needed.

Finally, the issue of high risk. The current criteria seemed to be loose and apply to too many patients. Prior to the year 2000 when we weren't involved in any carotid stent trials, we operated on the high risk patients routinely. As a matter of fact, high risk was almost used to exclude a patient because it often meant that their longevity was so short that it wasn't worth intervening.

I believe that the definition of high risk, with the exception of anatomical lesions, such as an irradiated neck or a high lesion or perhaps a recurrent carotid stenosis, I believe the definition of high risk needs to be stringently defined, and that hasn't been done yet.

Until these questions are more definitively answered with either larger randomized studies or more data, I speak in favor of a limited scope of applicability. There's nothing lost by doing this until better data is available. If it's good today, it will be good a couple of years from now when we get more data.

I think one of the risks that we really run, and this is a serious problem, is applying this on a broad label at this time and then having it go out and having a number of problems develop at all levels, which is a real risk.

Better, in my opinion, that we start slowly and expand, and I thank you for the opportunity to comment.

CHAIRMAN LASKEY: Thank you, sir.

Are there any other folks who wish to come forth?

(No response.)

CHAIRMAN LASKEY: If not, we'll close this open public hearing.

Geretta will read one more letter into the record.

MS. WOOD: I received this statement from Colin P. Derdeyne, M.D., Associate Professor at Mallinckrodt Institute of Radiology in the Department of Neurology and Neurological Surgery at Washington University School of Medicine.

"Dear Panel Members:

"The labeled indication for this device should not include patients with asymptomatic carotid stenosis. The data from the SAPPHIRE trial that has been presented to date do not support safety and efficacy for protected angioplasty and stenting in this population. While outcome was significantly better in the endovascular group than those who underwent surgical endarterectomy, it is possible and, indeed, highly likely that these patients would have done better on medical therapy alone.

"We cannot conclude that this device is safe and effective for asymptomatic patients. In the SAPPHIRE data presented at the AHA scientific sessions in Chicago in 2002, the 30-day rate for stroke, myocardial infarction, and death was 6.7 percent for the asymptomatic endovascular group. One year of follow-up data reportedly are over ten percent for these adverse outcomes in previously asymptomatic patients.

"I do not know how many patients in the trial were asymptomatic, nor the co-morbidities that led them to be categorized as high risk. Asymptomatic patients were required to have greater than 80 percent stenosis of the target carotid artery. This categorization was used to select patients perceived as high risk for revascularization, not high risk for stroke.

"We do not know the natural history for these patients. The best data we can have comes from the asymptomatic carotid atherosclerosis study, or the ACAS, published in 1995 in the Journal of the American Medical Association. Sixteen hundred fifty-nine patients were randomized to best medical therapy, aspirin or surgical endarterectomy. A relatively young, health, good surgical risk population was studied. Overall, a very slight, but statistically significant benefit was found with surgery. Annualized event rates for ipsilateral stroke from five-year projected data were approximately two percent for ipsilateral stroke in the medical group, and one percent for the surgical group. Thirty-day periprocedural surgical complication rates were 2.1 percent.

"Importantly, there was no stratification of increased risk with higher degrees of stenosis. Eighty-eight medically treated patients had greater than 80 percent stenosis. Three, or 3.7 percent, suffered a stroke during follow-up.

"Another group of patients that might be thought to be high risk was not. The five-year risk of stroke for 77 medically treated patients with contralateral complete carotid artery occlusion was 3.5 percent while the five-year stroke risk in the 85 surgical patients was 5.5 percent, Baker, Stroke, 2002.

"Finally, women do not appear to benefit from intervention, although the study was not powered to make this determination. It is important to note that since publication of the ACAS study, several improvements have been made and medical therapy that may further reduce the risk of stroke in asymptomatic patients.

"Statin agents have been shown to lower the risk of stroke in patients with atherosclerotic diseases published in Lancet 2004. Better anti-hypertensive agents are also now available.

"In summary, surgical revascularization for patients with asymptomatic carotid stenosis is of marginal benefit even if the most healthy of patients and with extremely low periprocedural complication rates. This benefit may not be present in women or in patients with greater than 80 percent stenosis or contralateral complete carotid occlusion. The event rates reported in the SAPPHIRE trial for both surgical and endovascular treatment of asymptomatic patients are extremely concerning.

"Even with the lower rates seen in the endovascular group, the most rational conclusion that can be drawn at present is that these patients should be treated medically. A trial of stenting versus best medical therapy is needed to determine if these devices are safe and effective in asymptomatic people. The label for this device should not include asymptomatic patients.

"Thank you for your time and consideration.

"Sincerely, Dr. Colin P. Derdeyne."

CHAIRMAN LASKEY: And now the open public forum is closed.

We need to consider the questions put to the panel. In view of the hour and the fact that we do need to vote and we cannot lose any of the members up here who need to leave for the airport, I would just request that we hold the discussion in response to these questions to a minimum, panel. Let's try and develop a consensus for the agency concisely.

So with that, can we have the questions, please?

MS. WOOD: I'm going to go ahead and start reading the first question while they're bringing up the projector.

Can the data from the investigator-sponsor studies be considered in the evaluation of high risk carotid stenting given the differences in trial conduct for the high risk investigator sponsor registry?

CHAIRMAN LASKEY: Well, I'll take point on these and then please feel free to modify, amend, chime in or correct.

I think that these IND studies which were part of the panel packet are in no way representative of the patient population that we are asked to consider today for the randomized portion of the trial. I think the event rates are rather low, strikingly low, which speak to at least one problem with the generalizability of these studies. There's a lack of adjudication of these endpoints.

So in summary, I don't think that these data from the IND studies can be used for today's discussion.

Good. All right. Next.

MS. WOOD: Number two, how does the large enrollment in the registry CAS arm affect interpretation of results?

CHAIRMAN LASKEY: Interpretation of results of what?

DR. ZUCKERMAN: The randomized trial.

CHAIRMAN LASKEY: All right. Well, point number one, I think we went through great pains this morning to speak to the fact that these patients wound up in the carotid stent registry for certain specific reasons, some which were clearly identified and some which were yet to be identified.

I believe Dr. Cohen alluded to the fact that the data collection forms describing all of the clinical characteristics in covariates really haven't been culled yet to ascertain differences or similarities that would allow us to draw conclusions of same or different behavior.

Number three, the outcome of the propensity score analysis remains in abeyance in part because of this, the lack of sufficient number of covariates examined, and I think those analyses are still under study.

So without answers to those three, I'm not sure that we can bring information about the applicability or generalizability of the carotid stent patients into this realm. We can certainly speak to the randomized trial data separately and the stent registry patients separately, but I think making comparisons is premature and perhaps hazardous.

DR. COMEROTA: Comment. Would it be accurate then to assume that this would be a real world type of scenario because these are patients that are evaluated for an operative procedure deemed not to be of adequate acceptance for the operation and, therefore, directed for carotid angioplasty and stenting?

CHAIRMAN LASKEY: Well, you're asking a question in a question. We're supposed to --

DR. COMEROTA: Well, see, Warren, you said they're not applicable, and I think that they may be applicable in light of what is the role of carotid angioplasty and stenting in these potentially high risk patients.

CHAIRMAN LASKEY: In those patients.

DR. COMEROTA: Right.

CHAIRMAN LASKEY: These are different patients than the ones in the randomized trial. Agreed?

DR. COMEROTA: Some of them were, yes.

CHAIRMAN LASKEY: In ways that are both identified and yet to be identified.

DR. COMEROTA: Correct.

CHAIRMAN LASKEY: And therein lies part of the problem of doing some of these additional analyses to establish comparability of results across these different patient subsets.

DR. ZUCKERMAN: Okay. The point of the question though was that there was a large exit of patients from the original enrolled trials to this registry. Does it invalidate the results of the randomized trial in any way?

CHAIRMAN LASKEY: No, I don't think I heard that today. It doesn't invalidate it.

DR. ZUCKERMAN: Okay.

CHAIRMAN LASKEY: It certainly qualifies it.

Judah, any comment? Okay.

Number three. How does premature termination of the pivotal randomized study affect the conclusions derived from this study?

Well, we went back and forth about whether it was premature termination or not, whether it was part of the up front decision to do sequential analyses, that that was built into the study design, that perhaps it was not prematurely terminated. I'm not sure from my standpoint whether this is just a semantic issue between the agency and the sponsor, and I'm not sure we should spend much more time on this. We've really beat it up pretty well this morning.

DR. KRUCOFF: I would basically agree to just turn it the other way. I think it's clear it was administratively prematurely terminated. I think the question then becomes of the range of statistical expertise on the data available, how legitimate are the conclusions from the randomized segment, and I do think there are some discussion between the expertise from London and Harvard and the agency would probably be worthwhile.

CHAIRMAN LASKEY: That being said, the goal and the endpoint was addressed and was found to be adequately powered to reject the noninferiority null hypothesis despite the fact that it was before the target.

DR. KRUCOFF: I think the one other thing I come away with, Warren, is that at least the premature termination was not the result of looks at the data that would be inappropriate or biased in that regard. That to me would be a much more fatal kind of problem.

CHAIRMAN LASKEY: Absolutely.

Okay. the next question is?

MS. WOOD: How does premature termination of the pivotal randomized study affect conclusions derived from this study?

CHAIRMAN LASKEY: We just did that.

MS. WOOD: Please discuss how data from previous carotid treatment trials NASCET, ACAS can be used to analyze the current perioperative 30-day data set with regard to safety.

CHAIRMAN LASKEY: Well, we just heard some information during the open public session, data to that effect. Do the surgeons wish to speak to these, quote, control rates?

DR. COMEROTA: I think the basic issues have been raised. The reference to NASCET and ACAS have always been what is the operative morbidity/mortality, but then you take that operative morbidity/mortality and then select instead of a good risk group a high risk group of patients and then say, "Okay. We're going to compare the operation to a new form of intervention."

And if you look at the new form of intervention, if we specifically focus on ACAS and the 30-day result rate, since the majority of the patients in SAPPHIRE were asymptomatic, that 30-day result rate is ten to 15 times higher what it was in the medically treated patients in ACAS.

If you look at the 30-day result rate, it's somewhere between -- it's some factor higher than NASCET. If you look at the death and stroke rate at the end of 30 days, it's probably a factor of two higher.

DR. WHITE: I disagree with that. I think that's not quite true, and I think the heart of this issue is that NASCET nor ACAS provide a proper comparator, and if we had had a sponsor here come and try to use ACAS or NASCET to justify their response, we wouldn't have heard it because they were not properly controlled.

It doesn't mean you can't look at NASCET. It doesn't mean you can't use that as a --

CHAIRMAN LASKEY: We're just answering the question.

DR. WHITE: -- as setting some limits, but you cannot use ACAS or NASCET as a proper comparator. they're just different populations of patients. There's nothing to make you think they were the same.

CHAIRMAN LASKEY: Not only that, but the ground breaking aspect of this trial was that they included myocardial infarction as well, which was not part of the composite endpoint in either NASCET or ACAS.

So I think that's another variable that was discussed this morning in terms of its relevance as well, and it's probably a very key variable.

So to answer the question, it can't very well properly be used in the context of redefining your MAE rate.

DR. ZUCKERMAN: Okay, but, Dr. Laskey, I just heard two opinions as to how usable the prior randomized carotid trials are, one person saying yes, one person saying no, that there are major differences.

What are the opinions of other panel members? Is there a consensus one way or the other? How generalizable are these trials?

CHAIRMAN LASKEY: Well, but before we do that, we have to make sure we're talking about the same thing. The event rates that were reported in NASCET and ACAS are not the event rates discussed today unless we throw out the myocardial infarction data. So that's one problem.

But Mitch.

DR. KRUCOFF: I think a fundamental issue here is around the issue of intention to treat. Intention to treat is to try and capture a real clinical scenario where you see a patient and would intend to treat them in some way, and I think it has been very clearly and repeatedly stated that SAPPHIRE was focused on an intention to revascularize, and what that patient population is and how it varies in the practice of medicine is a whole series of discussions, but I think to take studies that include an intention to randomize between revascularization or not is a fundamentally different starting point than taking a population in whom the intention is to revascularize.

And I think that to me is the biggest chasm to leap here in trying to use historical controls where the intention was to randomize between revascularization or not in comparison to this set of data.

CHAIRMAN LASKEY: Well, you're quite correct, but the question refers to using the periop data with respect to safety. So that we need to make sure we're looking at the same variables here.

CHAIRMAN LASKEY: Gary.

DR. NICHOLAS: If I can, I think that the applicability to the SAPPHIRE study and to try to utilize it to cross-validate the study, I don't think it can be done, but I do think that both NASCET and ACAS can set a guideline for stroke and not looking at myocardial infarction obviously or even death rate which might be altered because of the population.

But if we look at stroke in both the symptomatic and asymptomatic patients, I think they do set a standard, that pure and simply endpoint, the one the patient worries about and the one we worry about.

So I do think they have validity in assessing this type of protocol.

CHAIRMAN LASKEY: Even in the era of Plavix and so forth, which was not --

DR. NICHOLAS: Yes, sir.

DR. WEINBERGER: I think that the older studies don't seek out to enroll patients with high risk co-morbid backgrounds, and I think that for that reason alone you could not compare outcomes even if they included MIs. I think that even if MIs were included as endpoints in NASCET and ACAS, they still would not provide valid comparators.

CHAIRMAN LASKEY: Suffice to say the use of historical controls as controls is dangerous.

Next.

MS. WOOD: There were multiple ways for higher risk patients to be entered into the SAPPHIRE trial. Please discuss the impact of these various patient subgroups on ability to generalize safety and effectiveness results.

CHAIRMAN LASKEY: Well, I'm not sure how many multiple ways there were for patients to wander into the study. Once it had been decided that they needed something done was the very first step, and then things evolved or devolved from there, but that's a different issue than how they got to that point, and we don't know how they got to that point.

The next sentence though speaks to various patient subgroups which is kind of a disconnect from multiple ways for high risk patients to get into the trial. So there's two ideologically distinct concepts being addressed here. I don't think we can speak to how patients got into the study in the first place because we're not privy to that data, but in terms of the impact of the various patient subgroups, that's a terribly important issue that I think we grappled with all afternoon if we just used the symptomatic versus asymptomatic division to begin with.

That clearly affects the data on safety and effectiveness because it applies in the one group but doesn't apply in the other group. So it's confusing to us.

Folks?

DR. TRACY: Warren, I think that's the crux of the problem, is that both the subgroups, symptomatic versus asymptomatic, and I think that's where maybe the NASCET and the ACAS data have to come into play to some extent, to look back at that and what societies have regarded those results in terms of their recommendations regarding endarterectomy or stenting in asymptomatic patients.

I think the problem comes when you're lumping these two groups of patients together.

CHAIRMAN LASKEY: Which precludes our ability to generalize. I guess the other hooker here is that we've heard a number of pleas for a medical control arm, and I guess Dr. Zuckerman, you want to address the nuances of that. That would help to put some perspective into safety and efficacy. I understand it's not --

DR. ZUCKERMAN: Okay. I'm not sure what that has to do with Question 5 though. I'm sorry.

CHAIRMAN LASKEY: Well, we need a comparator. We need a valid comparator.

DR. ZUCKERMAN: Yeah.

CHAIRMAN LASKEY: And so comparing symptomatic to asymptomatic I think is not, I think, the end of the day for us, that there are other patient subsets out there, if you will, many of them in the medical treated arms that, again, we've heard a great deal in favor of that this afternoon.

But why was that not -- why is that not likely to work its way into today's discussion? Why is it inapplicable or perhaps it is applicable. So can you just speak to the inclusion of the medical control arm?

DR. ZUCKERMAN: Okay. We're talking about two important subsets, symptomatic and asymptomatic, and the question for the asymptomatics was perhaps a better trial design would have been a three-arm trial with a medical control group.

However, from a legal, FDA regulatory perspective, that sort of trial design is not necessarily required if surgical endarterectomy for asymptomatic patients is an acceptable standard of care and the sponsor shows that compared to surgical endarterectomy in the asymptomatic group, there is a reasonable risk benefit profile.

DR. TRACY: The problem is that there is not agreement on that. I mean, the American Society of Interventional and Therapeutic Neuroradiology and the American Society of Neuroradiology and the Society of International Radiology have a paper that was given to us which indicates that asymptomatic endarterectomy is at best controversial, not indicated in the Canadian publications, sort of marginally indicated in the U.S. publications, and that carotid stenting of asymptomatic patients is listed as a relative contraindication.

So it's a very difficult position for us to be in. We're in a place where we may be asked to approve something, approve a device for something that it's not indicated for.

DR. WHITE: Wait though, Cynthia, wait, wait. I mean, I agree there's some debate about asymptomatic revascularization patients. There is, but there's an AHA consensus statement about the appropriateness of that revascularization that we've been practicing for a long time.

DR. TRACY: For endarterectomy or for stenting?

DR. WHITE: Yes, for endarterectomy, for endarterectomy greater than 80 percent.

DR. TRACY: Right.

DR. WHITE: The AHA consensus document.

DR. TRACY: Exactly.

DR. WHITE: The same group that was put into this trial.

DR. TRACY: But not for stenting.

DR. WHITE: The question we have to ask is is the stenting as safe as or equal to the surgery, not whether the surgery is appropriate for greater than 80 percent endarterectomy or greater than 80 percent lesion. Surgery is appropriate for greater than 80 percent lesions. that's an established fact.

It's not that we can't debate it, but it's an established medical fact. I do it every day. It's standard of practice. We shouldn't go backtrack there. We shouldn't get confused in that morass, and Tony will tell you that. That is standard practice in the United States today.

Asymptomatic patient, greater than 80 percent stenosis, endarterectomy is indicated for that patient. Now, you may not operate on every single one of them, but it's an appropriate thing to do, and what we have to decide is in that population of patients, which were who were randomized in that trial, was the stent as good as surgery or better or worse, but not whether revascularizing 80 percent or greater lesions is not what we're being asked.

I mean, I think if we go there we're never going to get out of here.

DR. KRUCOFF: Yeah, I have to agree with Chris. I think if the question is what's the optimal treatment for asymptomatic carotid artery disease, we're hosed. We have nothing to go on that's new.

But I think if you say in a population of patients who are asymptomatic or scheduled for carotid endarterectomy, is this a safe and effective, reasonable alternative, I think that's what the SAPPHIRE study a data set to think of.

DR. ABRAMS: Yeah, I would also say that several of the following questions also address the same thing. I think really I agree with Chris. We have to decide whether we're looking at safety or we're looking at effectiveness and then maybe take the issue of limitations of indications for things somewhat later or at least open it to discussion, but I don't think we want to discuss whether or not or what the indications are for endarterectomy.

DR. TRACY: But even still there is a higher risk of TIA in those treated with stent. Is that a safety issue?

DR. WHITE: First of all, I don't believe that's true. I don't think the data says that.

DR. TRACY: That's what the data says. I mean it --

DR. WHITE: No, it was not a statistically ‑-

DR. TRACY: -- either says it or doesn't say it.

DR. WHITE: It doesn't say they were worse. It says they are the same. There was no statistical difference that I saw. Was there a statistically different group?

The numbers were small, and the variability was wide, and so it wasn't a higher risk.

CHAIRMAN LASKEY: Well, 6.6 is not small in one year.

DR. WHITE: No, I'm sorry. the numbers of the patients enrolled in the trial, the numbers compared in the event rates were small, and so it leads to difficulty in understanding the differences, but there is no reason for me to believe that TIAs occur more often after stenting than after endarterectomy. Nothing. There's nothing that I know that would indicate that.

DR. KRUCOFF: So maybe one other piece of the last part of this question, Dr. Zuckerman, would be that patient subgroups are tough to grapple at as one result of premature stoppage of the trial, which is very small, and it makes them very ambiguous.

DR. ZUCKERMAN: So you've looked at Question 5(b), but Question 5(a) refers more to Cordis' Slide 11, where you have the mechanisms by way of higher risk patients were defined anatomic risks, medical co-morbidities. There were multiple pathways.

Is this an acceptable type of schema? Are there any comments?

CHAIRMAN LASKEY: I think it is acceptable, but they're still -- that's just the part above water. I think, as with icebergs, there's so much more beneath the surface. How do patients who are asymptomatic with 80 percent stenoses wander into the system and attract attention?

I think that is part of the crux of this in terms of defining the appropriate patient population for this device. Yes, it's clear if you are in Class 3-4 CHF or you've had a recent acute coronary syndrome, et cetera, et cetera, you're certainly at high risk, but that's not the majority of these folks.

And one does wonder how you get into the system if you're asymptomatic and then discovered to have an 80 percent stenosis. I think that's more to the heart of how we practice and more to the heart of who benefits from these interventions as well, but we can't address that.

I think the categorization of high risk is certainly defensible. I haven't heard any arguments about the classification scheme for what's high risk today, but I think there's more to it than that, particularly with a study in which two thirds of the patients are asymptomatic.

How did they get into this study?

DR. COMEROTA: Warren, what responsibility do we have? I mean, I agree with the comments that if you take it on the surface, did you prove equivalence? The answer is yes, but one way to prove equivalence is to do any study with very, very small numbers, and you're going to get equivalence no matter how different the outcome is going to be.

Now, is --

CHAIRMAN LASKEY: I wouldn't say that in a room with the statistical fire power that's sitting here.

DR. COMEROTA: No, no, that's right.

(Laughter.)

DR. COMEROTA: But I think that it was admitted that you need a fair amount of difference to have small numbers be statistically significant, correct? I think we'd all agree on that.

And if our charge is only to look at the difference between these two groups and the randomized trial, the answer is clear. Do we have a little bit more responsibility than that?

And I think that is what we're grappling with. That's what I'm grappling with because we know that the question addresses this variable way. So you have recurrent carotid stenosis. You have asymptomatic lesions. You have symptomatic lesions, and the majority that were entered, the majority that were treated with carotid angioplasty and stent had less than 80 percent stenosis, the overwhelming majority.

So we're looking at patients as I see them who have relatively low risk lesions who are high risk for intervention, and now we're choosing a high risk intervention such as carotid endarterectomy, and we as vascular surgeons are not proud of these results. And I think to a person we would say these patients should not be operated.

But now we're asked to approve another procedure that has equivalent outcome to outcomes that we're embarrassed about, and that is the fundamental disconnect of what we're being asked to do. That's what I'm having problems with.

DR. WEINBERGER: But the reality is that vascular surgeons are not turning those people away.

DR. COMEROTA: Well, look at the registry. Vascular surgeons turned them away, and they were intervened with.

DR. WHITE: No. That's not what happened.

DR. COMEROTA: Is that not correct?

DR. WHITE: No. Those patients were referred for surgery. The decision for surgery was made. The surgeon then said, "I can't operate," and then they got into the registry.

DR. COMEROTA: Okay.

DR. WHITE: It has been said about 15 times today.

DR. COMEROTA: Well, I can't or I shouldn't, or is there a difference? Is that not good surgical judgment?

DR. KRUCOFF: I have to take a little exception at one other comment. Proving equivalence, statistical proof of equivalence is not accomplished just by doing small numbers. It's not the same as seeing no difference because your numbers are too small.

Now, we can question whether we actually accept the methods used that would have terminated the trial at 300 patients, but presuming that, in fact, that is a legitimate statistic, and I think probably when it's reviewed by FDA, it probably will be from what we heard today, we're not just looking at a casual finding from a small number of patients. This is a significant noninferiority statistic that is a little more than just something you get by doing too little work.

DR. TRACY: Regardless of the question of noninferiority, and I think to my mind it has been satisfied as not being inferior to carotid artery endarterectomy; however, the AHA guidelines indicate it is approvable or recommended for 60 percent stenosis if the stroke mortality rate is less than three percent.

That is what the AHA guidelines state. That is what it states.

DR. WHITE: Well, I think you're just misunderstanding just a little bit, and that is that the consensus statement that was written by -- I'm trying to think who the first author was -- there was a consensus statement from the HA that said that not 60 percent was the indication for endarterectomy, but 80 percent.

DR. TRACY: With what mortality or stroke rate?

DR. WHITE: Again, using the ACAS data, but again, as you look, it gets a little complicated, but the numbers you're looking at were guidelines for programs to study safety. The indications for operating on asymptomatic patients is greater than 80 percent, which is why it was used in this trial.

DR. TRACY: I think though that however you cut this is controversial. I mean no matter how many ways you look at this, and maybe this is what's happening, but you're taking a carotid surgery and extending it to something that's even more --

DR. WHITE: Cindy, I'm telling you it's standard of practice. It happens every day in the hospital.

DR. TRACY: Sure, it happens every day, but should it?

DR. WHITE: It's standard of practice. It's not controversial.

DR. TRACY: Surgery happens every day.

DR. WHITE: Yes, greater than 80 percent for asymptomatic patients. No question about that.

DR. TRACY: All right.

CHAIRMAN LASKEY: Therefore?

DR. WEINBERGER: One further comment that speaks to this question. I think that the numbers are too small to break out anatomic versus medical co-morbidities, to break out symptomatic versus asymptomatic, and I lament that very much together with Dr. Comerota.

The fact is that surgeons in most institutions operate on these patients, period. If our surgical colleagues would turn them away routinely, then we'd have a very good argument here as to whether or not we should be carotid stenting them.

But the truth is that the standard of current surgical care is to treat people who are asymptomatic, who have 80 percent stenoses with and without medical co-morbidities. That is the reality of what's happening in most institutions.

DR. COMEROTA: All we're saying is that there are very, very few 80 percent stenoses and more in this trial by angiographic description of diameter reduction stenosis.

When you look at the symptomatic patients that were randomized, there were probably ten or 12 patients with an 80 to 99 percent stenosis.

DR. WHITE: Are you looking at core lab data? To get into the trial you had to have more than 80 percent lesions. I don't where this is coming from.

DR. COMEROTA: Chris, they gave it to you in your packet.

DR. WHITE: But I mean, are you telling me that --

DR. COMEROTA: Every patient had an angiogram.

DR. WHITE: When I look at a patient to get into this trial, he had to have more than 80 percent to get into the trial. Now, if two months later --

CHAIRMAN LASKEY: That's asymptomatic. If you're --

DR. WHITE: Asymptomatic that had greater than 80 percent at trial. One hundred percent of the patients had to have that. Now, two months later if the core lab goes behind me and says, "You know, that wasn't 80 percent. That was only 65 percent, Dr. White," then that's what you're talking about.

DR. COMEROTA: Chris.

DR. WHITE: But to get into the trial 100 percent of the --

DR. COMEROTA: Look at the data, the angiographic data in the panel pack.

DR. WHITE: Are you suggesting that patients with asymptomatic lesions were enrolled with less than 80 percent lesions?

DR. COMEROTA: I'm suggesting that the --

DR. WHITE: Is that your criteria that I don't know about?

DR. COMEROTA: I'm suggesting that the overwhelming majority were less than 80 percent. If you want the specific numbers --

DR. WHITE: But that was core lab. That was post hoc data. That was analyzed after --

DR. COMEROTA: This is arteriographic analysis.

DR. WHITE: After the fact.

DR. COMEROTA: This is at the time of arteriogram before treatment.

DR. WHITE: You're wrong.

DR. COMEROTA: Chris, this is submitted to us in the panel pack.

DR. WHITE: Maybe we could ask the sponsor to clarify that for us, whether they would admit asymptomatic patients to the trial with less than an 80 percent stenosis.

CHAIRMAN LASKEY: Well, the duplex was 80, but then there's this carotid --

DR. WHITE: Well, no. For stenting they had to have an angiogram.

CHAIRMAN LASKEY: Right.

DR. WHITE: And at the time of that angiogram if the investigator could not say it was 80 percent, they could not enroll that patient.

CHAIRMAN LASKEY: Well, let's put that on the table.

DR. POPMA: Can I help?

DR. WHITE: Yes.

CHAIRMAN LASKEY: Yes.

DR. POPMA: We're going to pull up some very quick slide. The hour is very late.

I'm Jeff Popma. We directed the angiograph, the core lab analysis for this.

Very insightful comments. I had my travel paid to come down here and also I'm on a coronary stent advisory board.

That was an excellent question about the disparity between the clinical site readings. The average visual clinical site reading, which you can see on this slide is perfect, was 85.2 percent. It's way too late in the afternoon to go to the differences between NASCET and the ECST criteria, but when we reapplied the ECST criteria from core lab analysis, the stenosis was about 82 percent, and the NASCET appropriately pointed out was a 68 percent stenosis.

And it's a huge issue about where investigators actually take the reference vessel diameter.

Having said all of this, the patients got into the study based on the predefined Doppler criteria for an 80 percent stenosis, which is in clinical practice. Then we got the angiogram, as Chris points out, months or two later. We've got very detailed analyses about how we compare NASCET with ECST, with visual readings that were performed and provided in this.

Chris is exactly correct that the site reported visual analyses, was an 85 percent NASCET based visual analysis, but all of the patients got into the study based on the Doppler criteria which were predefined and standardized for this trial.

So I don't believe -- it's a long discussion which I can take care of later -- I don't believe that inappropriate patients were put in the trial. I believe that this trial reflects the type of patients that are treated every day in clinical practice and very comparable to what we've seen with other clinical trials that are ongoing right now.

DR. COMEROTA: Could you answer then the information on Table No. -- for every study, the feasibility study, the randomized trial, the registry, there's a patient-by-patient printout.

DR. POPMA: Correct.

DR. COMEROTA: And there's this column that says "pre-procedure DS (percent)." Does that mean pre-procedure diameter stenosis in percentage?

DR. POPMA: Your readings, which were retrospectively late on down the line, yes, that's the baseline percent stenosis.

DR. COMEROTA: Based upon an arteriogram.

DR. POPMA: Based upon our independent core lab reading of the arteriogram.

DR. COMEROTA: Okay.

DR. POPMA: There's a subtlety to this question.

DR. COMEROTA: I'm not quarreling that the patients didn't have your velocity criteria to get into the trial. That is not the issue. All I'm saying is by arteriographic analyses the number of patients by angiogram defined 80 percent stenosis were very, very few. There were about 19 percent, 80 to 99 percent stenosis, 19 percent in the randomized trial and less than ten percent in the -- well, I have the numbers, but there are very few relatively speaking by arteriogram.

DR. POPMA: Using the NASCET criteria, the NASCET criteria which takes the parallel portion of the internal carotid, as we all know. You're correct that the mean percent stenosis was 65 percent.

Having said that, it's a bit issue that we can talk about now or later. The Doppler criteria that were predefined and used for inclusion for patients in the study were really initially validated against the ECST criteria. We redid that based on an imputed ECST criteria, and the Doppler readings are very close to what they were initially validated with Gene Strandness and otherwise.

And having said all of this, most patients who go off to carotid endarterectomy don't pass through the catheterization laboratory before they go to endarterectomy. This trial used standard Doppler criteria for the 80 percent diameter stenosis.

We're in the process of writing the series of validation papers, but basically I believe that the patients who are enrolled in this trial met the Doppler criteria, correlated with the ECST, and if we were going to readjust everything, we would go back and say that the NASCET criteria was a bit lower, but that's what we would expect for every trial.

Having said that, one last slide which I'll show here -- unfortunately I'm not going to be able to blow it up to show it.

The real data that we should have been talking about as we talk about the applicability of NASCET criteria for this trial was a meta analysis that was performed by Peter Rothwell along with Alan Fox. We have had multiple discussions with Alan Fox in our core laboratory over the last several months that really took all of the ECST criteria and went back and reanalyzed them using the NASCET criteria.

All of the angiograms in ECST were re-read, and the publication of this meta analysis says two things to me. Without question, those patients that had a greater than 70 percent -- in symptomatic patients -- those that had a greater than 70 percent stenosis did great with endarterectomy, but in addition of an appropriately defined ECST and NASCET meta analysis, there still is a statistical benefit associated with anybody with more than a 50 percent symptomatic stenosis.

That's coupled with this finding from ACAS, and I know that we have focused on a greater than 80 percent stenosis using a NASCET criteria and ACAS, and Dr. Hobson is here and has spoken, and there are a lot of subtleties to this, but this is the data as was published in the JAMA article written for ACAS.

Even in those patients that had a greater than 60 percent stenosis, there was still a benefit with carotic endarterectomy. Absolutely we have selected as a guideline for all the angiographic reasons we've talked about 80 percent, but if we really drill in hard on the actual NASCET criteria itself, there was still a statistical benefit at the lower percent stenosis.

So having said all of this, I mean, if you have questions I'll try to answer them, but I believe that both in symptomatic patients and in an asymptomatic patient this is the standard in clinical practice today with respect to carotid endarterectomy, and I believe that our results support the fact that we're actually benefitting patients by revascularization rather than with medical therapy, understanding all of the limitations that we have with the NASCET criteria.

I don't know if that helps at all, but that's --

DR. COMEROTA: Well, I won't get into your last comment, but very simply, we're just looking at an angiographic definition of diameter reduction stenosis. I'm not quarreling with the criteria. As a matter of fact, I agreed with your criteria for your noninvasive studies 100 percent.

But when I look at the data that I received for the angiographic analysis, it's what I said, you know.

DR. POPMA: That's correct. And as we talked --

DR. COMEROTA: And there's very few people with a high grade stenosis that we would agree angiographically had demonstrated.

DR. POPMA: I don't know that I agree with that. By NASCET criteria, 95 percent of the patients had more -- 95 percent of the patients had more than a 50 percent stenosis using the NASCET criteria. I think virtually every lesion subset that we look at there's a benefit with revascularization therapy.

Your point is well taken that we said 80 percent in the protocol based on the Doppler, and what I'm trying to express is the Doppler was based on the ECST criteria.

DR. COMEROTA: I know. You're trying to get back to the Doppler. All I'm saying is the arteriogram. That's all.

CHAIRMAN LASKEY: Jeff, your point is well taken, and I think this would be a hell of a time for us to get back in to study inclusion criteria. So, I mean, thank you. that's been very helpful.

And it's not the first time that actually what is actually in a study is less than what you think is there. That's lessons learned from QCA.

We have wandered far from the discussion of the patient subgroups, but I think some important territory has been aired out.

Number six?

MS. WOOD: Effectiveness of stroke prophylaxis has historically required two to five years' monitoring with safety outcomes generally accessible within the lesser period of one year. Please discuss whether chronic data presented in the SAPPHIRE trial for the OTW configuration provide evidence of sustained effectiveness of CAS in preventing stroke in patients at high risk for CEA.

CHAIRMAN LASKEY: Well, panel members, correct me if I'm wrong, but all I heard was that there were extrapolations of the Kaplan-Meier curves from the one year out to three and four years. So I think that's --

DR. WHITE: Well, I think they showed us two-year data, three-year data.

CHAIRMAN LASKEY: The data on the median survivals out to --

DR. WHITE: No, the stroke and death rate, right? Was it stroke and death?

PARTICIPANT: There was stroke and death for three years.

DR. WHITE: Was it for the complete -- it was not for the whole data set. You don't have three-year follow-up on the whole data set. Is that right?

DR. COMEROTA: If the question is did it prevent stroke as well as CEA, the answer is yes. If the question is did it prevent stroke in high risk patients that are not going to undergo CEA, we don't have that information.

DR. COHEN: Three-year data was presented from the U.S. feasibility study, and the five-year life expectancy for half of the patients was extrapolated from the SAPPHIRE randomized data.

CHAIRMAN LASKEY: Right. Thank you.

So there is no good, long-term data.

Number seven.

MS. WOOD: Is it appropriate for the sponsor to employ OPCs developed from NASCET and ACAS outcomes to assess outcomes for both symptomatic and asymptomatic patients in the SAPPHIRE trial or should the ACAS rates from the symptomatic trial be used for comparison?

CHAIRMAN LASKEY: Well, we just had an extended discussion about the study inclusion criteria and how perhaps they were erroneous to start, and if you re-look at them, the conclusions needed to be modified certainly for the NASCET data.

So I mean, this is a very rapidly moving target when you continue to reanalyze the data. I'm not -- fellow members, is it appropriate for the sponsor to employ OPCs?

DR. KRUCOFF: Well, Warren, you know, we can maybe suggest to the agency that, on the one hand, you have a SAPPHIRE cohort who are randomized, and you have an hypothesis that you can test within that population unto itself.

Then there's all of the registries and feasibility and the long-term data from the feasibility, and how far you could extrapolate to use historical point information from NASCET to ACAS might be helpful there, but I think we've spent a lot of time going over how many limitations you're going to find in that particular question.

CHAIRMAN LASKEY: All right. So not appropriate.

Number eight.

MS. WOOD: The ACAS and NASCET's did not include myocardial infarction as an endpoint. The SAPPHIRE trial included MI as a component of MAE. The actual distribution of non-QA MIs are provided under Tab 8 addendum of the panel pack. Please comment on the sponsor's choice of this composite endpoint.

DR. COMEROTA: I think it's appropriate.

CHAIRMAN LASKEY: Well, it's appropriate because it's contemporary, but is it appropriate when you're looking at surgical outcomes versus non-surgical outcomes, knowing that with surgical patients you're stirring up the pro-thrombotic milieu. You're stirring up the pro-platelet milieu. You're stirring up the flammatory milieu. You're stirring up a lot of factors which go into postop surgical morbidity and mortality, which are not applicable in the stent patients.

Now, that's perhaps an advantage of this study, but surgical patients were at a decided disadvantage in this study because of the risk of postoperative badness that happens when you give general anesthesia to patients who are at high risk for bad things.

I don't know what else there is to say.

DR. KRUCOFF: I happen to think that's quite appropriate in the same way in thrombolytic trials we look at stroke. You know, ultimately it's a net clinical benefit concept, and I think it's pretty clear that all current trials are looking in terms of ultimately the net clinical benefit to patients in this area.

CHAIRMAN LASKEY: Right, but we need to tease out, especially when we're making comparisons to studies that don't have MI as part of the event rate. We need to understand that this is a terribly important difference about this trial and perhaps its strong point, but there's no question when you look at the Kaplan-Meier curves or mortality. Just look at mortality out to two years. There is a clear and distinct, early, sustained, and persistent -- I don't care whether the log rank is significant or not -- but the carotid surgical patients fare less well in terms of long-term mortality, and I think that's telling us something about their underlying risk not related to the surgical procedure itself.

DR. COMEROTA: Does the fact that there were significantly more carotid angioplasty stent patients having had coronary revascularization, should that come into our consideration in putting this into perspective?

PARTICIPANT: No.

DR. AZIZ: I think it should because I think you're already protecting against the event like a myocardial infarct in the surgical patients. If they have an underlying coronary artery disease, you haven't tackled it, you know.

DR. COMEROTA: I think if you look at those curves that you're referring to Dr. Laskey, I mean they continue to separate, and they separate a great deal at about six -- start to separate at about six or eight months and then really begin to diverge as time goes by.

CHAIRMAN LASKEY: My point exactly. So there's more than just the early hazard. There's another hazard which is kicking in which we don't have a good handle on, and certainly the long-term follow-up data will be critical.

DR. ZUCKERMAN: Okay. Dr. Laskey, can you give some advice for trials going forward? The point of the question: is MI as significant as death and stroke in these questions?

So that should be routinely in the primary composite endpoint.

CHAIRMAN LASKEY: Is that a question?

DR. ZUCKERMAN: Yeah.

CHAIRMAN LASKEY: Yes, it certainly belongs as an appropriate endpoint. I think we need to understand its pathogenesis and its behavior and the risk factors IV. I mean there's a lot that goes into this.

It's aggravating to sit here and hear that 50 percent of these patients were not treated with beta blockers. They all had extensive coronary disease. They all had extensive vascular disease, and to be so under treated is, I think, a shame.

And we all know that if you intensely monitor in the perioperative period and perhaps get beta blockers on board preop, postop, there's this whole area which has not been addressed, which is the pre and postop management of the patients in this study. That may certainly impact beneficially on the postop infarction rates.

I'm astonished by the world class cardiovascular caregivers who cannot see their way to giving an adequate beta blockade to these patients at high risk.

I'm done.

(Laughter.)

CHAIRMAN LASKEY: Nine.

MS. WOOD: The indications for carotid artery extending in the registry arm were largely dictated by hazards of surgical exposure. The ability to deploy a stent should not be affected by these criteria or the outcomes achieved in this registry, i.e., ten percent stroke and TIA at 30 days and an additional 16 percent a one year acceptable.

DR. WEINBERGER: I don't know whether the premise of this question is true. We've been trying to ferret out how they got into the registry, and it's true that the surgeons didn't want to operate on them, but the analysis of those patients for the reasons that thy got into the registry is not heavily weighted towards anatomic exclusions. It's also weighted towards clinical exclusions. Am I wrong?

DR. ZUCKERMAN: Not all of the reasons for entry are known, but for those patients who did have anatomic reasons as the primary reason, those are the results.

DR. COMEROTA: The registry followed the randomized trial, correct?

DR. WEINBERGER: No, it was concurrent.

CHAIRMAN LASKEY: Parallel with.

DR. COMEROTA: Parallel with? It did not continue to enter patients into the registry after the randomized trial was completed? No?

DR. WEINBERGER: No.

DR. COMEROTA: Okay.

DR. WHITE: I think you have to ask what you compare the ten percent to. I think that, you know, for example, we know that for stroke and death if it's a reoperation on an endarterectomy, the HA has told us the ten percent is an acceptable level.

So I think you have to say what are you comparing it to because ten percent is high, one in ten, and I think that I saw the sponsor present the data for the registry compared to the surgery data, which was questionable statistically in terms of its honesty or the appropriateness of doing that.

But when I saw that comparison, the registry data actually fell between the surgery arm and the randomized arm. So it didn't appear to be worse than the surgery arm.

So in that context of comparison, it seems to be appropriate or acceptable.

DR. KRUCOFF: If you will really take this full circle, the conundrum becomes that patients who are referred for endarterectomy for whatever reason, symptomatic or not, who a surgeon evaluates and says, "Huh-un, this is too high a risk," that's the group who you probably ought to compare to medical therapy, but that's a study, you know, that gets totally beyond the pale here.

So in fact, we have data on sort of the middle risk category, patients who are acceptable for carotid and referred for carotid endarterectomy, acceptable for carotid endarterectomy, who are randomized, we have a data set that I think we might be able to get to a conclusion on.

But those who are actually too sick for surgery, we have these numbers, and really I think we'd have to step back to say, "Compared to what?" right to Chris' point.

CHAIRMAN LASKEY: And finally, because they're only estimates and because the numbers really are small, it would be helpful to just look at the confidence intervals around these two. They may be rather unacceptable. I don't think we should just look at the ra estimate.

Okay, ten.

MS. WOOD: Please comment on whether the incidence of ipsilateral stroke is acceptable.

CHAIRMAN LASKEY: There's that word again.

(Laughter.)

CHAIRMAN LASKEY: And we need a comparator here.

DR. KRUCOFF: So compared to endarterectomy in a randomized cohort, it's equivalent.

CHAIRMAN LASKEY: Not acceptable but equivalent.

DR. WHITE: Well, I mean, again, compared to what we understand. I think the largest single subset of patients were redo endarterectomy patients. Is that true, in the randomized trial? It was the largest single indication, repeat endarterectomy. Is that true, the largest?

I can't remember. Anyway, if ten percent stroke and death is okay for those people, then we're well within -- if that's the accepted indication, then we're well within the ballpark of that number, and because it wasn't inferior, I think it is acceptable.

DR. COMEROTA: Would it be appropriate to look at the registry patients as a comparator to answer this question since they were deemed unacceptable for operations or is that not appropriate?

CHAIRMAN LASKEY: Well, it would be appropriate, but we don't have the statistical -- we don't have the methodology to do that. We're missing a fair amount of data from my understanding. There's a bunch of covariates there which haven't worked their way into this analysis that may yet.

DR. COMEROTA: Well, we have the data on ipsilateral stroke.

CHAIRMAN LASKEY: Yeah.

DR. COMEROTA: The data on ipsilateral stroke for all patients in the registry trial, in the registry, was 4.2 percent, and if you look at ipsilateral stroke in the carotid endarterectomy patients from SAPPHIRE, it's 1.8 percent.

If you look at symptomatic patients, ipsilateral stroke in the registry was 6.5 percent, and the carotid endarterectomy patient was zero.

If you look at asymptomatic patients, it was too high in both. It was 3.2 percent in the registry. It was 2.5 percent in the operative group.

So those are the data. That's why I asked if it's appropriate or not.

CHAIRMAN LASKEY: Well, you just can't compare them because they really are A and B. You just can't compare the registry group in any way easily to the randomized trial. They are different and hence this effort to do at least a propensity score if not more, but it's very difficult to just look at those two numbers and say whether they're different or not.

I think you need to drill down a little further.

Number 11.

MS. WOOD: The various studies employed a total of only four sized five millimeter stents. Does the panel believe that there are adequate safety and effectiveness information for this size?

CHAIRMAN LASKEY: Tony, Judah? Not a lot of data on this.

DR. WEINBERGER: Not a lot of data.

CHAIRMAN LASKEY: But needed; agreed?

DR. WHITE: I think that in the packet the size, the iteration of these stents, the five to eight is the same, essentially the same stent with the same surface area, the same shortening.

I don't think this is the same issue as a balloon expandable stent. I think it's a much less of an issue than it is in a balloon expandable stent, and I think that it would be -- I don't see why the five millimeter stent would inherently be troublesome. It would be used in a smaller artery.

And I think to not have a five if we're going to do this would be putting some operators at a disadvantage in a smaller area because then they'd be using a six. So I think if a five is the right size we ought to be able to use a five. It just doesn't happen very often.

CHAIRMAN LASKEY: Okay.

MS. WOOD: Has the totality of data presented for the OTW configuration in the carotid stent PMA shown reasonable assurance of safety and effectiveness?

If not, what niche indications have been shown to be safe and effective for carotid stenting?

CHAIRMAN LASKEY: Well, this is pretty much why we're here today.

(Laughter.)

CHAIRMAN LASKEY: So -- what's that?

PARTICIPANT: This is.

CHAIRMAN LASKEY: Yes. So in terms of safety, I think that the investigators have demonstrated safety from the noninferiority standpoint. I didn't hear anything to the contrary today.

In terms of efficacy, I think there is a lot of controversial material here in terms of the reduction in the rates of stroke down the road, and there's some unfortunate discordant outcome data with respect to symptomatic/asymptomatic, men/women, other subgroups which are annoying. You like to see all of the point estimates line up to the left of the hazard ratio, but that's not the case here.

So that's my crack at this. Do you want to round this out, folks?

DR. ABRAMS: I'd agree very simply. I think we do agree that it's probably safe, but we have real qualms about effectiveness, which is not a question we're going to be able to answer today. So I guess that's sort of what the agency is looking for. I think that's kind of a black line answer to the question.

DR. WHITE: But is the answer about efficacy really noninferiority? I mean, there was no superiority intended here. So efficacy, I think sometimes we think of efficacy in terms of superiority, but do we believe that not only did we not cause harm, but are we not inferior to stenting?

And I think that's what the randomized data set shows. So efficacy makes it sound stronger than that, but I think the truth is that it's not inferior.

DR. KRUCOFF: I think it's not a niche indication, but given the enrollment criteria for the SAPPHIRE randomized cohort, and as Chris says, you really want to sit not to the left of zero, but to the left of the boundary that's appropriate for non-inferiority.

But in that population, I don't know if we'd call it a niche. Patients who are recommended for revascularization or a carotid, symptomatic or non, who are candidates for surgery despite having one additional high risk characteristic; that in that population, I think the equivalence of safety and effectiveness out to the year of follow-up that's been reported is pretty straightforward.

What goes beyond that year I think would be a place we could think about.

DR. MAISEL: I respectfully disagree, and I have particular concerns about the asymptomatic group, and what do you call a stroke, a safety issue or an effectiveness issue, I think, is a little murky. But the 30-day rate for asymptomatic patients of about five percent is very concerning to me.

I certainly recognize the troubles we've been having with the comparator group and that CEA is performed in a lot of patients that have 80 percent stenosis, but there's no CEA data on patients like this that were enrolled in this trial. Many of the data that we've been referring to are extrapolated from much lower risk patients.

I'm particularly concerned about approving this device for asymptomatic patients when we really don't know whether it's the right thing to do for patients who present better high risk with an 780 percent stenosis. We do not know what the best treatment is for those patients.

DR. WHITE: You're right, but we've backslid again. We just backslid again into the indications for revascularization.

DR. MAISEL: But if you look at the indications for the study, this study entry criteria was not "you are going to be revascularized."

DR. WHITE: Yes, it is.

DR. MAISEL: No, it's not. It was "you are referred for revascularization."

DR. WHITE: Bill, if you don't give me a stent, then these patients will be operated on with their 5.3 percent --

DR. MAISEL: I disagree with that. I think some of them will not be operated on. I think some of them will receive medical therapy. I think more than half of them will receive medical therapy as evidenced by the registry data.

And so if a stent is not available, it's conceivable that many of these patients will receive medical therapy. I agree that the data suggests that the asymptomatic patients are at very high risk, better than carotid endarterectomy. I agree to that.

What I don't agree is that stenting these patients is the best therapy for them.

DR. WHITE: I don't think they've demonstrated superiority, and you used the word "best." I think that what you have to say is that you do not agree that it's not inferior to surgery.

DR. MAISEL: I agree that it is not inferior to surgery.

DR. WHITE: But you're not sure that --

DR. MAISEL: I agree --

DR. WHITE: -- the right thing to do is to operate on these patients at all.

DR. MAISEL: I am saying I do not believe that stenting these patients -- that there's evidence here that stenting these patients is the right thing to do.

DR. WHITE: But there's evidence that stenting these patients it not inferior to surgery.

DR. MAISEL: I agree with that, but the question is safety, and I am not sure that this device is the right thing to do. I am not sure it is the safest thing for the patients to receive this device.

DR. COMEROTA: Is it not true that in the registry patients that if they did not have -- I mean, seven patients were operated upon because it was evident, at least the opinions were that it was more appropriate to operate than not.

But if they were allocated to the registry patients, if there were not a registry, these patients would not have been intervened with; is that correct? It's not correct?

DR. COHEN: Of the approximately 2,200 patients who were screened, one third of them were actually enrolled, and that compares favorably with the trials that we are deciding whether or not are appropriate comparisons. NASCET only enrolled one patient for every three that received carotid endarterectomy.

DR. COMEROTA: So we have a high risk for an operation that we randomize to angioplasty and stent versus operation. Now, how did they get to the registry?

DR. COHEN: Basically the patient met entry criteria, okay, and the surgeon decided that they did not want to take the patient for surgery.

DR. COMEROTA: So we don't want to operate on this patient. So you go ahead and put in a stent.

DR. COHEN: That's correct.

DR. COMEROTA: Okay, and then by virtue of radiologists or the interventionalists saying in seven patients, "We don't want to intervene on this patient. You go ahead and operate," then they were operated.

DR. COHEN: Yes. I think a good analogy to make here is a cardiology analogy. We have bypass surgery and we have multi-vessel stenting. There are issues with both of them, advantages and disadvantages. In assessing individual patients from their history, their laboratory findings, other tests that are obtained, we make our best clinical judgment as to what the most appropriate therapy is, and that's what we're talking about here, whether there should be an alternative to what's already utilized today in the United States.

DR. OURIEL: Tony, I was just going answer your previous question about the asymptomatic side of this, and I'll try to make it quick. But the data is the data, and to split out asymptomatic patients is really beyond the scope of the trial.

And that said, as we know, 70 percent of our patients in this country are getting operated on for asymptomatic disease, and they're not the low risk patients. The New York study, seven centers circled around New York City, many of these patients are high risk.

So what do we do this procedure for? We do it to prevent major ipsilateral stroke, and what was the major ipsilateral stroke in the treated patients? Well, in the symptomatic treated patients it was zero in the stent arm, and in the asymptomatic treated patients it was zero, and that was at one year. That was at one year.

So the results are obviously very good, and let's look at a couple of other studies very quickly. the first is ECST, and 9.8 percent risk of stroke at three years in 80 to 90 percent stenoses. It wasn't ACAS that split them out by stenosis. It was ECST.

And in the 90 to 99 percent it was 14.4 percent stroke at three years, and then recently presented data in London, about two weeks ago at the Charing Cross study meeting ACST data, 1,500 patients with severely stenotic asymptomatic disease, and at five years, a 12 percent incidence of stroke, 2.5 percent per year.

So the goal of the SAPPHIRE trial was to demonstrate non-inferiority of stenting and endarterectomy in patients who we all treat, like the patients at high risk for surgery, and this was conclusively demonstrated with a P value of 0.0035, noninferiority. It's not that there just weren't enough patients. It was well powered.

So the trial succeeded in proving in my mind beyond any doubts that stenting is a safe, effective, and appropriate alternative to endarterectomy in symptomatic patients, in asymptomatic patients that are at high risk.

DR. COMEROTA: You're very convincing, but we're being asked -- I'm being asked; you're asking us -- to change our entire paradigm of the management of symptomatic patients on the basis of 50 symptomatic patients being treated with the carotid angioplasty and stent and 39 patients with atherosclerosis who are symptomatic. That is what we're being asked to do.

DR. OURIEL: Well, I don't think so. Respectfully, I think what we're asking you to do is that if you have a patient that you are going to treat that fit into this high risk criteria, that you ought to be able to consider stenting in addition to endarterectomy.

CHAIRMAN LASKEY: You may not get the answer now, but you're going to get it on the vote.

DR. ZUCKERMAN: That's fine.

CHAIRMAN LASKEY: Okay. Labeling.

MS. WOOD: Are the indications and contraindications for the OTW configuration clear and supported by the SAPPHIRE study findings?

If not, please identify the indication you believe is supported by the sponsor's data. Specifically, is stenting of asymptomatic patients supported? Should any criteria stipulating when stenting of asymptomatic patients is appropriate be included in the labeling?

CHAIRMAN LASKEY: Well, we really just addressed that. I'm not sure we're going to get any further by any more dialogue, but I think people will vote with their feet. Hopefully before they vote with their feet, they'll vote with their hands about this issue and whether the labeling should be so targeted.

So can we answer that in another way?

DR. ZUCKERMAN: Okay. Are there any comments on Dr. Krucoff's prior suggestion that if you can't cross with the distal protection device, this procedure should not be done?

DR. WHITE: You mean absolutely or relatively?

DR. ZUCKERMAN: There was a comment before that this should be a contraindication for --

DR. WHITE: I think it should be a relative contraindication. I think you can use words like "discourage" or "rethink it," but I think you have to consider what your other options are. I think the operator at that time needs to weigh the risks and benefits knowing full well that -- I mean, there are still people out there in the world who are not convinced that protection devices are absolutely required. So I don't know if we want to legislate that, but I do think that we should encourage their use as was done in this trial.

So I like the words "discourage" or "warn" or "concern me," but I don't like the absolute contraindication that says if I can't get a protection device across I can't do it.

DR. MAISEL: I think if the label included the data we were shown today with and without distal protection device and the risk of the procedure, that would be very helpful.

CHAIRMAN LASKEY: Is that all the data there is?

DR. TRACY: Warren, I don't want to belabor this. Can I ask: is there such a thing in anybody's mind then as a low risk high grade stenosis?

Okay.

CHAIRMAN LASKEY: Again, low risk for what? Low risk for intervention or low risk for subsequent events?

DR. TRACY: Well, since our concern is stroke rate primarily in the asymptomatic patients, is there a low risk, asymptomatic patient who should not be exposed to this high risk procedure?

I mean if there was a way to state that in the indications, really specifying that these are high risk asymptomatic patients, it might make me a teeny bit happier about it.

DR. KRUCOFF: Well, I think it would be very reasonable to consider a phrase in the indication labeling that the population of cohort for whom this is intended are patients who are considered good candidates for benefit from carotid endarterectomy. That's really where we have randomized data, symptomatic and asymptomatic, whatever. That's the group.

And if they're not candidates for surgery, the registry arm of SAPPHIRE is actually a group who are not candidates for surgery, and whether this device is indicated in them or not, I think that's a separate question, but at least to me the most clear indication for use here would be in patients who are candidates for surgery who would be anatomically commensurate with stenting.

To touch on Bram's question, I don't know to what degree a label or a condition can be applied, but I think just based actually on the feasibility data and the data where before the ANGIOGUARD wire was available, there's clearly or appears to be a different outcome with this stent than with the distal protection.

For all of the different religions amongst interventionalists, Chris, I think the data that's available would suggest at least a warning that the risk of distal embolization appears to be higher based on the data.

DR. WHITE: I believe that. I just don't think it should be an absolute contraindication. I don't like absolutes.

DR. KRUCOFF: Yeah, but some way to warn operators that if you can't get the distal protection system across --

DR. WHITE: Rethink whether you want to --

DR. KRUCOFF: -- this patient is a candidate for carotid endarterectomy, and at least to warn them that they should think that through.

MR. MORTON: Dr. Laskey, my only comment about a warning, I'm not sure that the data from this study showed that there was more danger in not using the distal protection device.

DR. WHITE: No, but they presented cumulative data over several -- they combined the data for several of the trials that showed that the patients who didn't have the device had a higher stroke rate.

DR. KRUCOFF: Yeah, the good news is that they made the new version so good that 95 percent of the time you're going to be able with a skilled operator to get the distal protection system. It is more flexible. It's more deliverable. That's the good news, I think, that SAPPHIRE does sort of look like is the expectable direction.

But I do think the meta analysis ultimately with this stent, with and without the angioguard at least operators should be warned if you can't get into a distal protection position, be aware that there may be a down side. And these are candidates for carotid endarterectomy.

CHAIRMAN LASKEY: Okay.

MS. WOOD: Patients with complex atherosclerotic disease of the aorta or highly tortuous carotid arteries are not optimal candidates for carotid stenting. Please comment on the adequacy of the labeling with regard to patients with these anatomic characteristics.

If there are candidate that are not optimal that should be added, please also identify them.

DR. WHITE: I think the sponsor in the PMA pretty well sets out the contraindications in terms of thrombus and heavily calcified and tortuous lesions. I think they're pretty well known, and I'm not sure why the agency is asking this question. Do you think there were additional things besides what were listed?

DR. ZUCKERMAN: We're not talking about the clinical trial inclusion/exclusion criteria. We're talking about how the warnings and precautions presently read with respect to that factor. Is there any other statement that you would put in regarding lesion complexity that would make operators think again about doing a carotid stent procedure or give them pause to think?

DR. WHITE: It's very difficult to replace good judgment at the table. Things change. Angiograms look a certain way. You put a guiding catheter or a sheath in. The carotids shift and kinds appear where they weren't before. So it's a moving target. It's very difficult to know before you get into the cath. lab what the anatomy will actually be.

So I think the warnings of the tortuosity, the calcification, the thrombus, I think those things are all up front, but I think it's just good judgment and training that's going to teach operators when they need to not be doing these things.

MS. WOOD: Should any other warnings and/or precautions be stipulated in the labeling for the over-the-wire configuration in addition to those found in the proposed labeling?

DR. KRUCOFF: This is where I was going to mention the distal, the ANGIOGUARD thing. I do think that based on the data available that a warning to the operator that of you are unable to position distal protection, that the outcomes or risk of embolization may be different should be clearly stated.

DR. WHITE: Can I ask the sponsor what was the basis for the intracranial contraindication? I'm trying to figure out why that's a bad thing to do.

DR. COHEN: Are you talking about aneurysms?

DR. WHITE: No. You say here that patients -- stenting of intracranial arteries is a relative contraindication. Just carotid stenting; that if you're going to treat an intracranial lesion, you shouldn't treat an extra cranial lesion.

DR. COHEN: Yeah.

DR. WHITE: Do you know why that's in there, Jay?

DR. OURIEL: It's merely a carryover from the exclusion criteria for SAPPHIRE.

DR. WHITE: Well, it's a good exclusion criteria because you don't want to confound your outcomes, but in reality sometimes we need to treat outflow lesions to make the stent be latent and work.

DR. OURIEL: Sure.

DR. WHITE: So it probably shouldn't be there.

DR. OURIEL: Point well taken.

CHAIRMAN LASKEY: But then again, wasn't there something about tandem lesions?

DR. WHITE: The tandem is an indication, but these are lesions up in the --

CHAIRMAN LASKEY: Yeah, okay. All right.

MS. WOOD: Please comment on whether the sponsor's post approval study plan is adequate. If not, what additional information do you believe should be collected post approval?

Specifically, do you recommend that an independent neurologist make the neurological assessments at each follow-up?

DR. ABRAMS: Yeah, I'd like to comment. I definitely think so. I think it is a mistake or perhaps it could have been thought out. There should have been independent neurologists making the evaluations during the current study. I'm sure there were difficulties in doing it, and that's why it wasn't done, but I think if the opportunity arises to do it post marketing, I think it definitely should be done, particularly if you're going to use minor strokes as an adverse event as an important outcome.

CHAIRMAN LASKEY: And, yes, we think that their post approval study plan is adequate, although one always raises one's eyebrows at the nice round number of 1,000. So you just might want to pursue that.

DR. ZUCKERMAN: Okay. The agency needs clarification on where the neurologist is needed. There is follow-up now planned at discharge 30 days, nine months, et cetera. Is the independent neurologist needed in the peri-stenting arena or is there some reason for nine months to have an independent neurologist?

What are we getting at here? Do we want to assure in a post approval study that we can replicate and generalize the acute carotid stenting results?

DR. WHITE: How long will you ask the sponsor to carry out the study of the randomized trial patients?

DR. ZUCKERMAN: That's our next question for an advisory panel.

DR. WHITE: Well, my point is that if you want to know if events are occurring, you have to have an independent neurologist. You cannot rely on an operator. So if you want the data and you want to know who had a stroke or who had an event, then I don't care whether it's nine months or two years. It has to be someone that didn't stand at the table to put that stent in.

DR. ZUCKERMAN: Okay, but there are two different questions here. For a post approval study, what is the main point that the panel wants to see? Is it that when this procedure is generalized to multiple hospitals in the U.S. we can replicate the peri-stenting rates? That's one question.

The second question is in the IDE cohort, if you want longer term, well documented follow-up, you can ask the sponsor for a neurology examination of those patients, but for this next 1,000 patients, where is the neurologist critical? What is the point of having a neurologist?

DR. WHITE: I think it's a 30-day endpoint. A 30-day endpoint is going to be the critical element here.

DR. ABRAMS: I would add one more, too, as an endpoint in addition to the 30 days because I think there's still the question about the low grade embolization from devices that are placed in the vasculature, and I think if it could be two points, it would be 30 days and one year.

DR. WHITE: Let me just say that I don't necessarily disagree with that, but I think we ought to be intensively studying the cohort of the subject of the study intensively to determine that. I think as we distribute this in post market surveillance, the lion's share of the events are going to be peri- procedural, and quality assurance issues and safety issues are going to be all settled in 30 days.

So I think we should get emboli late outcomes out of the cohort here followed for an appropriate length of time, but as we go forward with post market, I'm not sure that more than 30 days is going to be much bang for our buck.

DR. COMEROTA: Is there going to be a threshold above which there will be critical re-review of the technique if there are major adverse events occurring that exceed, substantially exceed, what were observed in the SAPPHIRE?

DR. ZUCKERMAN: Yeah, the sponsor has a data monitoring plan built into the post approval study.

DR. COMEROTA: And would that include centers that were not part of the submission, or would that be inclusive of the subsequent patients that these investigators are entering?

DR. ZUCKERMAN: Your advice would be very helpful here.

DR. COMEROTA: I would suggest that that subsequent 1,000 be separate from the current investigators. These individuals are very talented, and their outcomes, I think, are going to set a standard, and I think we need to insure that others can match that outcome.

DR. KRUCOFF: I would, I hope, agree that it should be both, and I think continuing to let people who already have expertise use this in a post market environment and making sure that new centers, which I think they pretty explicitly made the plan to include smaller hospitals, new operators. I think you really need both, and that appears to me to be what they have committed to.

The only other thing I wanted to mention in this post approval is that I would at least put special emphasis on or attention on areas where we have the least data from the pivotal trial, and that would be the five millimeter stents and the cases where distal protection was not able to be deployed, but they were stented; that those are populations that I would flag.

DR. ZUCKERMAN: So Part B of Dr. White's response was the follow-up of the ID cohort. The sponsor presently plans three-year follow-up. Is that long enough to show durability of the procedure?

DR. ABRAMS: Yes, I'd think so.

CHAIRMAN LASKEY: Training?

MS. WOOD: Please comment on whether the sponsor's training plan is adequate. If not, what additional requirements do you believe should be added to the training program?

CHAIRMAN LASKEY: Well, I guess this morning's comments were it was accepted by the panel that this was a reasonable training program, and that's to be distinguished from qualification, certification, confidence, and credentialing. Those are all issues which are not within the purview of the FDA or the panel, but are critically important.

We've certainly heard a lot of testimony in favor of those things today in the open public session, but the training program seemed appropriate. You know, I think we all have concerns about fly-by carotid angiography in terms of acquisition of skills, but I think looking at the best case scenario, people will be well intentioned, will try and gain sufficient experience.

The Cordis will adhere to a fairly rigorous program, but it's really within the governance of the local institutions to maintain some quality assurance and some credentialing.

But having said that, there's not much more that we can recommend. Any other thoughts on training?

DR. KRUCOFF: Warren, I don't know if this quite fits into a training program mold, but I guess one of the things I've been sitting here thinking is whether, given the controversy with regard particularly to asymptomatic patients, whether just the definition of a patient as a clinically indicated surgical candidate would be something that should be maintained in a multi-disciplinary mode and whether discussing how you identify the appropriate patients belongs in a training program or not.

You know, one option would be almost like the old brachytherapy approach, to involve several disciplines in order to execute the procedure. It does make things a lot more cumbersome, and I'm not sure it really belongs in a training program.

But one thing might be to have surgical and medical opinion that the patient is a candidate for carotid endarterectomy as a way of describing the patient population, who ultimately we know something about the safety and effectiveness of the device.

CHAIRMAN LASKEY: Yeah, well, that's certainly the high road. We would espouse that, but that's probably not where most people travel unfortunately, and as Chris said, you know, you let people use their judgment, and hopefully they'll use good judgment. There's no way to mandate that. There's no way to inculcate that. You either have it or -- so anyway, I think with respect to the device, the training programs outlined is adequate, but the panel certainly has concerns with respect to confidence and certification.

Okay? Before we move on to the vote, I just want to as Dr. Zuckerman and the agency if they have any additional comments or questions prior to.

DR. ZUCKERMAN: No. We would just ask that before any vote is taken that the panel members pay particular attention to the regulatory definitions of safety and effectiveness as read by Ms. Geretta Wood.

MS. WOOD: I have one call for some clarification. I had a question regarding who the members are that would be voting on this device. If you'd look at your attendees list, the voting members and the consultants that are listed will all be voting. The consultants were deputized, all of the consultants here at the table.

CHAIRMAN LASKEY: Anything else?

I'd like to ask the sponsor if they have any additional comments or questions.

DR. COHEN: No. thank you very much.

CHAIRMAN LASKEY: Okay. And, Dr. Hughes, do you have any comments prior to the vote?

DR. HUGHES: Yes. Thank you, Dr. Laskey.

I hope I can get my points across briefly without loss of meaning. My comments are more along the lines of general risk management as opposed to specific medical and surgical criteria.

First of all, I want to commend the FDA's staff and the sponsor for their presentations and, of course, the panel for its review of the device. It's as thorough a review as can be at this point.

And I also want to commend the presenters in the open public session who, in general, I think, did a fine job of illuminating the issues.

It's uncomfortable though to place the panel in the position of either recommending approval or non-approval with data and information, as well as changes in protocol, you know, coming along, you know, so close to the panel meeting, as well as some of the apparent shortcomings in the pre and postoperative medical therapies as has, you know, been brought out.

But anyway, I feel like the panel and the FDA are at a point where there is a need for a balance of technological innovation, which I think that this device and therapy represent an acceptable risk to the ultimate consumer, certainly, you know, the patient.

Let's see. With the lack of statistical analysis of the device, you know, in accordance with strict adherence to FDA protocols, risk management for this device if approved I think would require a tremendous amount of cooperation among the participants in the overall risk management system, that is, you know, FDA constituents, the manufacturer, as well as professional societies.

So input in the risk management system from all of those parties would, indeed, be needed and very aggressive, I think, and it would require the aggressive enhancement of several attributes of the medical device risk management system, in particular, you know, those concerning labeling and post market surveillance.

Of course, they have been addressed here, particularly the labeling in terms of what's been said by panel members and what was read into the record from one of the parties in the public open session.

Some of those aggressive measures, I think, in post market surveillance would require what I would consider some, you know, voluntary actions on the part of the manufacturer, possibly beyond what the FDA would be prepared to mandate, and I'm thinking about such things as an aggressive device retrieval program. I've noted some concerns for device durability, and given the somewhat limited follow-up time for the studies that we see here.

So I feel like there would need to be a program set in place for aggressive device retrieval. If we're talking about that, what also may need to be considered is some form of an autopsy program coupled with device retrieval.

Of course, if we're talking without device retrieval it still may be possible; I'm not sure; it may be possible to glean some insight in the form of noninvasive autopsy, MRI, CT scans, and I'm thinking in terms of, you know, whether or not that each individual faces that ultimate ending point, whether they're a patient or not, and so it's a matter of not an issue of whether death is caused by some failure of the device, but there might be some information gleaned from some form of autopsy program.

And then finally, I believe that the manufacturer should also consider some form of emergency compensation funds if it's determined that there was something overlooked in this review, the pre-market review and evaluation.

Part of risk management is to accept that there will be some level of adverse events, and if this level is higher than we've anticipated, I feel that the manufacturer should be prepared to compensate for any later intervention that might be necessary to, you know, correct the device therapy, the device situation, if that would, indeed, be possible.

And, you know, thank you for your time.

CHAIRMAN LASKEY: Michael.

MR. MORTON: I would acknowledge that there have been some very intelligent and heartfelt statements made by each member of the panel today. I would also acknowledge Dr. Zuckerman's comments that the role of the panel is to evaluate and make a decision based upon the data that have been presented today and make a judgment based upon evidence of safety and effectiveness.

And I understand that the Executive Secretary is going to read that formal definition in just a bit.

So thank you very much.

CHAIRMAN LASKEY: Thanks, Mike.

Geretta, if you can please read the voting options.

MS. WOOD: Before I read the voting options, just a point of clarification. Dr. Salim Aziz, who is also a voting member, his name was omitted from the attendees list. He also will be voting today.

The medical device amendments to the Federal Food, Drug, and Cosmetic Act as amended by the Safe Medical Devices Act of 1990 allows the Food and Drug Administration to obtain a recommendation from an expert advisory panel on designated medical device pre-market approval applications, PMAs, that are filed with the agency. The PMA must stand on its own merits, and your recommendation must be supported by safety and effectiveness data in the application or by applicable publicly available information.

Safety is defined in the act as reasonable assurance based on valid scientific evidence that the probable benefits to health under conditions on intended use outweigh any probable risks.

Effectiveness is defined as reasonable assurance that in a significant portion of the population, the use of the device for its intended uses and conditions of use when labeled will provide clinically significant results.

Your recommendation options for the vote are as follows:

Approval, if there are no conditions attached;

Approvable with conditions. The panel may recommend that the PMA be found approvable subject to specified conditions, such as physician or patient education, labeling changes, or a further analysis of existing data.

Prior to voting all of the conditions should be discussed by the panel.

The third option is not approvable. The panel may recommend that the PMA is not approvable if the data do not provide a reasonable assurance that the device is safe or if a reasonable assurance has not been given that the device is effective under the conditions for use prescribed, recommended or suggested in the proposed labeling.

Following the vote, the Chair will ask each panel to present a brief statement outlining the reasons for their vote.

CHAIRMAN LASKEY: I'd like to ask for a motion on the PMA. Anyone?

DR. NAJARIAN: I make a motion that the PMA be approved, but proof for the treatment of symptomatic carotid disease in high risk patients.

CHAIRMAN LASKEY: We probably should back that up, and then we will discuss the conditions to be applied, but --

MS. WOOD: Excuse me. Dr. Najarian, are you suggesting that you would like for the device to be approved or are you --

DR. NAJARIAN: Approved with conditions.

MS. WOOD: -- would like approvable with conditions?

DR. NAJARIAN: Approvable with conditions.

MS. WOOD: Thank you.

CHAIRMAN LASKEY: Is there a second?

DR. WEINBERGER: I'll second.

CHAIRMAN LASKEY: Okay. It has been moved and seconded that the PMA is approvable with conditions, and let's take these conditions.

DR. TRACY: Warren, let me clarify this. I believe he just voted for or he just recommended approval for patients with symptomatic carotid disease.

DR. WEINBERGER: No, a condition.

CHAIRMAN LASKEY: No, that will be a condition that I'm about to ask for. It's approvable with conditions is the motion which has been seconded.

DR. TRACY: I believe he stated though -- so you're making the distinction, just so we're clear. What do you think we're moving for? what do you think we're approving here?

DR. NAJARIAN: I will back up and simplify. My motion is vote for approval for conditions, which we can now discuss.

DR. TRACY: Okay.

CHAIRMAN LASKEY: Which has been seconded. The first condition of approval then is?

DR. NAJARIAN: That the use be limited to symptomatic patients -- excuse me -- symptomatic high risk patients.

DR. COMEROTA: Would that be with moderate or high grade carotid stenosis? Symptomatic patients with moderate or high grade carotid stenosis --

DR. NAJARIAN: Correct. This is --

DR. COMEROTA: -- carotid endarterectomy.

DR. NAJARIAN: Yeah.

DR. WEINBERGER: Do you mean limited or do you mean indicated for? I don't think we're putting limitations here. We're stating indications. The indications should be indicated for patients with symptomatic carotid disease with high grade anatomic features.

CHAIRMAN LASKEY: Well, on the table right now is the condition that they be symptomatic and at high risk.

DR. NAJARIAN: High risk, correct, not high anatomic.

CHAIRMAN LASKEY: And the discussion that ensued also put some quantification in as well?

DR. COMEROTA: I raised a point of clarification with Dr. Najarian, did he mean to include moderate to high grade carotid stenosis. In other words, the NASCET terminology.

DR. NAJARIAN: Yes.

DR. COMEROTA: In patients at high risk for carotid endarterectomy.

DR. WEINBERGER: And we're talking about an indication recommendation or limitation?

CHAIRMAN LASKEY: An indication.

DR. NAJARIAN: I need some help here.

CHAIRMAN LASKEY: Yeah, an indication. I think the process here should go through the generic approvable with conditions, approved or disapproved. We now have on the table the motion which has been seconded, and we're moving forward with it. It's approvable with several conditions, I'm sure, the first of which is that its procedure be applied to symptomatic high risk patients, and that is high risk for carotid endarterectomy. Is that accurate?

DR. COMEROTA: I asked for clarification from Dr. Najarian. I think he clarified that was accurate.

DR. MAISEL: Do you mean symptomatic patients only or symptomatic patients?

DR. NAJARIAN: Only, only. I guess I'd have to look back. Symptomatic patients --

DR. MAISEL: Are you trying to exclude the ‑-

DR. NAJARIAN: I'm trying to exclude the asymptomatic group.

DR. KRUCOFF: So is this open for discussion or do we need a second?

CHAIRMAN LASKEY: We're discussing this condition.

DR. KRUCOFF: Right. Okay. Because I think we're going to face a real conundrum here. The only data that we really have that's interpretable is for patients in whom carotid endarterectomy was indicated, symptomatic and asymptomatic.

As soon as you drop out asymptomatic patients, you don't have enough data left to know what you're doing. So I think the panel has got to face the conundrum that for all of the miasma of our poor, hapless asymptomatic patients going to get intervened on where medical therapy might actually do just fine.

The data set that we have gives its clearest look at the use of this device in patients in whom the clinical consideration that they would be better off revascularized has already been made whether they're symptomatic or not, and as soon as we drop the asymptomatic patients from that, there is no data left to make database decisions.

DR. WHITE: I would support that, and I really do think that two-thirds of this data was asymptomatic, and I think we really run the risk of inappropriately parsing this data to leave out that population.

I think if you can't accept the asymptomatic patients, then you probably can't accept it on the whole.

DR. KRUCOFF: Now, Bill, what you said before, Bill, the registry patients here are not the issue. It's the randomized patients.

DR. MAISEL: But the registry patients are an issue in that the patients who would come to the table in the real world are the randomized patients plus the registry patients because those were the patients that got sent in to be enrolled in the study. So over 50 percent of those patients ended up not being surgical candidates even though whoever referred them in thought they were surgical candidates.

DR. KRUCOFF: Okay. So what we