UNITED STATES OF AMERICA

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

8:06 a.m.

CHAIRMAN MAISEL: Good morning.

I'd like to call to order this meeting of the Circulatory System Devices Panel. Today's topic is discussion of a premarket notification for the Alsius Corporation CoolGard 3000/Icy Catheter system, K040429.

Geretta?

EXECUTIVE SECRETARY WOOD: Before we begin this morning, I have a couple of announcements.

Due to an emergency, Dr. Hallstrom was not able to join us today. We have Dr. Brent Blumenstein on the phone filling in for this morning.

Now I would like to read the conflict of interest.

The following announcement addresses Conflict of Interest issues associated with this meeting and is made part of the record to preclude even the appearance of an impropriety.

To determine if any conflict existed, the Agency reviewed the submitted agenda for this meeting and all financial interest reported by the Committee participants. The Conflict of Interest statutes prohibit special Government employees from participating in matters that could affect their or their employer's financial 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, a limited waiver has been granted for Mr. Halperin for his interests related to the issues before the Panel that could potentially be affected by the Panel's recommendations. The limited waiver allows him to participate in the review and discussion, but excludes him from voting. Copies of this waiver may be obtained from the Agency's Freedom of Information Office, Room 12A-15 of the Parklawn Building.

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 him or herself from such involvement and the exclusion will be noted for the record.

With respect to all other participants, we ask in the interest of fairness that all persons making statements or presentations disclose any current or previous financial involvement with any firm whose products they may wish to comment upon.

ACTING CHAIR MAISEL: Thank you.

At this point I'd like to have the Panel members introduce themselves. I am William Maisel, cardiologist at Brigham and Women's Hospital in Boston.

And why don't we start on my right with Dr. Yustein?

DR. YUSTEIN: I'm Ron Yustein. I'm the Acting Critical Deputy Director for FDA's Office of Device Evaluation.

MS. MOTTLE: Linda Mottle, Director of Clinical Research at GateWay in Phoenix. Consumer rep.

DR. MARLER: John Marler, a neurologist with the National Institute of Neurological Disorders and Stroke.

DR. BRINKER: Jeff Brinker, Interventional cardiologist, Johns Hopkins.

DR. PAGE: Rick Page, cardiologist, University of Washington.

EXECUTIVE SECRETARY WOOD: Geretta Wood, Exec Sec.

DR. BROTT: Thomas Brott, neurologist, Mayo Clinic.

DR. SOMBERG: John Somberg, Professor of Medicine and Pharmacology, Rush University, Chicago.

DR. HALPERIN: Henry Halperin, I'm an electrophysiologist at Johns Hopkins Hospital.

DR. KATO: Norman Kato, cardiothoracic surgery, private practice, Encino, California.

DR. WEISFELDT: Myron Weisfeldt. I'm Chair of the Department of Medicine at Johns Hopkins. My background is in cardiology.

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

CHAIRMAN MAISEL: Dr. Blumenstein? Dr. Blumenstein, can you hear us okay? I'll take that as a no.

DR. BLUMENSTEIN: I'm having a very hard time hearing. I heard some better than others.

CHAIRMAN MAISEL: Can you introduce yourself, please?

DR. BLUMENSTEIN: I'm sorry. Was I asked a question?

CHAIRMAN MAISEL: Can you introduce yourself?

DR. BLUMENSTEIN: Yes. My name is Blumenstein. I'm a consultant working independently living in Seattle, Washington.

CHAIRMAN MAISEL: Thank you.

At this point I'd like to open the public hearing session of today's meeting. Both the Food and Drug Administration and the public believe in a transparent process for information gathering and decision making. To assure such transparency at the open public hearing session of the advisory committee meeting, FDA believes that it is important to understand 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, it's product, or if know 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.

Is there anyone in the audience who wishes to address the Panel this morning? Yes, sir?

DR. VANDEN HOEK: Hi. Good morning. My name is Dr. Terry Vanden Hoek. I'm an Associate Professor of Medicine at University of Chicago and in practice as an emergency medicine attending there.

My conflicts of interests that I should declare today is Alsius is paying for my expenses today. I get no other salary support from Alsius or any other cooling device funding.

My research funding is from the NIH and Department of Defense for study of mechanism of post-resuscitation injury after cardiac arrest and hemorrhagic shock.

I was asked to speak today regarding the ILCOR recommendations that were published in 2003 in Circulation. And these recommendations were the result of consensus meetings by professionals from six continents, seven professional organizations that included cardiologist, neurologist, critical care, anesthesiologists, emergency medicine, physicians as well as nursing staff that practice in the critical care and emergency medicine setting.

And what's remarkable is that there were worksheet presentations that were done and a number of forums represented. All of the weight of the evidence for the use of therapeutic hypothermia for selected patients after cardiac arrest.

And what the conclusion was that the American Heart Association had already recognized that hypothermia may play some role after cardiac arrest. And if you read the guidelines in 2000 what they say is that hemodynamically stable patients who develop a degree of hypothermia spontaneously after cardiac arrest should not be actively warmed. And there was already a recognition back then that hypothermia could be tolerated and, in fact, may be beneficial but at that time we didn't have the evidence to suggest that they should be actively cooled in certain groups of patients.

After the evidence had been presented at multiple forums in which specialists could participate, critique the evidence, the decision was made to issue an advisory statement between the guidelines, which were published in 2000 and will be coming out this year, which is an unusual event and I think reflects the consensus among the specialists that we needed to say something because the weight of the evidence suggested that there was actually something we could do for cardiac arrest patients that improved survival. And that was to cool subsets of patients to 32 to 34 degrees for 12 to 24 hours.

And we issued that advisory statement in the summer of 2003.

The thing I want to emphasize to the Panel is how remarkable it is that we actually have anything that can improve survival from cardiac arrest. It's an extremely lethal disease. The mortality rate is 95 percent. It is remarkable that we have even one randomized controlled trial that shows any evidence of improved survival benefit, not only to hospital discharge but actually to six months survival.

The fact that those patients can actually go back and work at last part time and live independently is extraordinary.

I think that the fact that you can develop a consensus amount that many physicians from that many continents is also quite remarkable.

The other thing I want to emphasize is that we think there is great promise in the use of hypothermia but we have a long ways to go. Surface cooling has its own limitations. It could take eight hours to cool a patient. Based on the animal data that's available and some human trials, we think that if we can cool faster, the benefit may be even more profound. And certainly we don't have all the answers yet. Cooling studies such as has been done by Clifton in the study of traumatic brain injury, the onset of cooling was much longer, took longer to cool his patients than it did in the HACA trial.

The Cool MI trials, the cooling certainly was not done for 12 to 24 hours. And those are actually what would be considered Level 7 studies when we present our worksheet. Because those were studies of cooling for other diseases.

So I would like to take any questions. Right now I have to get back to a study section for the NIH.

CHAIRMAN MAISEL: We're going to go through all the public speakers and we'll do a question session at the end if any of the Panel members have questions.

Thank you very much.

DR. VANDEN HOEK: Okay. Thank you.

CHAIRMAN MAISEL: Is there anyone else who wishes to address the Panel this morning?

DR. PEBERDY: Good morning. My name is Mary Ann Peberdy. I'm an Associate Professor of Medicine and Emergency Medicine at Virginia Commonwealth University in Richmond. And I'm a cardiologist by background.

I receive grant support from Medivance for the participation in the RESCUE trial. And my expenses to attend today's meeting will be paid by Alsius.

Intellectual disclosures are that I am an unpaid volunteer for the American Heart Association and a member and past char of the Science Advisory Board for its National Registry of Cardiopulmonary Resuscitation, which is an in-hospital database that currently houses data on over 60,000 in-hospital cardiac arrest resuscitation events. As the current vice chair Research for NRCPR, I am responsible for developing a dataset for post resuscitation care that will include the induction of mild hypothermia in that dataset.

The purpose of my comments today is to provide this Panel with a description of the current practice of providing mild hypothermia to survivors of cardiac arrest. My comments are based, in part, from the data on NRCPR o n information obtained from its participants, from users groups and focus groups, on communications with other clinicians and scientists both individually and as part of the Guidelines 2005 Conference, and on my personal experience as Chairperson of my hospital's Resuscitation Committee, and the primary physician responsible for providing this therapy at my hospital.

The national and international scientific community strongly recommend treatment with mild hypothermia for subsets of comatose survivors of out-of-hospital cardiac arrest. Despite the science and the current guidelines, hypothermia is under utilized in the United States.

Dr. Abella and his colleagues at the University of Chicago recently published an article in Resuscitation citing the survey results of the practice of providing post-resuscitation hypothermia. Out of 265 responders, only 13 percent had admitted to using hypothermia following a cardiac arrest.

While some of this may be due to the fact that it often takes a decade or more to translate science into clinical practice, a third of the responders stated that the reason for not using hypothermia was that the current cooling methods were technically too difficult or too slow and imprecise. Less than five percent of U.S. hospitals participating in NRCPR currently provide hypothermia. Many admit to believing that the therapy should be provided, but find the current protocols to burdensome to perform.

A few hospitals have even reported that after developing their protocols and implementing the hypothermia system, that after doing their first patient the case was so difficult, messy and burdensome that they abandoned this therapy completely.

My personal experience mirrors that of others. Our hospital has been performing hypothermia for over a year, and I am intimately familiar with our cases.

The induction, maintenance and reversal of mild hypothermia is not only difficult to perform accurately and consistently, but is tremendously time and resource consuming. Without a device specifically to provide and control mild hypothermia clinicians are currently left with whatever is cold and available to them for quickly decreasing body temperature. This may include infusing large amounts of icy saline, lavaging gastric contents with ice, packing the patient in ice bags which can range anything from traditional ice packs to large hefty trash bags filled with ice cubes and placing the patient on a cooling blanket traditionally used for the purposes of reducing fever rather than the induction and maintenance of mild hypothermia.

Although these methods of treatment are quite basic and without need for regulation, that does not necessarily equate with them being safe and the quickest and a most effective way to provide this therapy.

Clinicians often use a combination of these methods to induce the quickest time to target temperature. Thus, cooling from inside out and outside in.

Almost all of the patients are packed in ice that moves and leaks and creates an inconsistent temperature control and winds up with the patient being in a soaking wet bed. The inability to control the depth of hypothermia within a reasonable degree of predictability leads to overshoot with respect to the target temperature. Well over half of the patients with induced hypothermia in our institution, a place where that pays particular attention to how we deliver this therapy, have patients with temperatures that fall below 33 to 34 degrees at sometime during their therapy, which could place the patient at risk for further complications from imprecise treatment. The resulting actions to try and rewarm the patient to a more acceptable level of mild hypothermia leads to a ping-pong effect with the patient having temperature curves that swing up and down requiring cooling and reheating for almost the entire duration of the protocol. Rather than a consistent predictable temperature curve for cooling, maintenance and reversal, the typical patient is subject to jagged fluctuations in body temperature.

In addition to this, the ice packs or cooling blankets invariably leak leading to a very electrical unstable patient lying in a puddle of water. Many of us have commented that we have been quite lucky to not have patients refibrillate and require defibrillation in this setting. The dangers of defibrillation are obvious.

The multiple staff carrying for these critically ill patients often have their hands full in providing immediately necessary medical therapies and appropriately find it unreasonable to be expected to change a bed one, two or even times during duration of this protocol.

Attempting to provide this therapy by these means is, for lack of a better term, a thermalregulatory nightmare and potentially even electrically dangerous.

As both scientists and clinicians we can all agree that there is still much to be learned about how to best deliver this therapy, the time frame to initiate cooling, the depth and duration of hypothermia, the optimal use of sedative and paralytics, the different methods of cooling and need for CNS monitoring, just to name a few, are all areas that will clearly benefit from further experience and evaluation. But despite these unknowns post-resuscitation hypothermia makes a difference.

Our hospital has seen noticeable improvements in survival to discharge from out-of-hospital cardiac arrest since starting our program. Other programs also anecdotally report similar results.

In both Europe and the United many EMS systems are now seeing striking differences in survival to discharge after initial ROSC when delivering patients to hospitals performing aggressive hypothermia compared to hospitals that do not.

There was a frank discussion at a recent EMS Director's meeting specifically talking about whether or not it is unethical to deliver patients with out-of-hospital cardiac arrest to hospitals that do not provide hypothermia and whether or not appropriate patients should be diverted to those hospitals known to provide this therapy.

The future focus of survival from cardiac arrest will clearly have post-resuscitation care as a prominent feature. To help spotlight the importance of this currently missing link in the chain of survival, the NRCPR is developing a dataset specifically geared for delivery of care after ROSC and collection of data in this area. Specific attention will be paid to the delivery of mild hypothermia so that many of the current unknowns will be able to be evaluated and tweaked and cared for most carefully. Data on this resuscitation outcomes module will be used by individual hospitals to improve their process of resuscitation as well as by the scientific community as a longitudinal mechanism to track trends in treatment and outcome variables. So a mechanism is already in place for patients who receive hypothermia to be tracked by an independent board of scientists and clinicians.

In summary, I have three roles pertinent to the therapy being discussed today. I care for patients who have suffered a cardiac arrest. I teach others to care for patients who have suffered a cardiac arrest. And I do research to find better ways to care for these patients.

Everything that I know indicates that this therapy should be applied to certain comatose survivors of sudden death. During the time that we are in this room meeting today over 400 people will die from cardiac arrest and only 20 of them will ultimately survive without improve post-resuscitation therapies such as hypothermia.

We are dealing with a disease that carries with it an almost certain death. And although we acknowledge that we do not yet know all the intricacies of how to best deliver this therapy, we do know that despite this hypothermia makes a difference.

Please don't sacrifice better for best in considering devices that may help improve the delivery of this therapy. Clinicians need choices and better options to deliver mild hypothermia and patients need a better chance to survive.

Thank you.

CHAIRMAN MAISEL: Thank you.

Is there anyone else who wishes to address the Panel this morning?

Is there anyone on the Panel that has a question for either of our speakers? Jeff?

DR. BRINKER: I'd just like the last speaker, who is very impassioned, to tell me about whether she uses this device now to do away with all the hassle of leaking ice bags and things?

DR. PEBERDY: No.

DR. BRINKER: Why?

EXECUTIVE SECRETARY WOOD: Please come back to the podium.

DR. PEBERDY: The device is currently not FDA approved for this indication and our hospital was quite reluctant to use devices off label.

DR. BRINKER: There is no device, including ice bags, approved for this indication.

DR. PEBERDY: That is correct.

DR. BRINKER: But you use ice bags?

DR. PEBERDY: But we don't have to purchase them and they are not regulated.

DR. BRINKER: Okay.

DR. PEBERDY: This device is approved for hypothermia for other indications. So to our hospital that makes a difference to them.

DR. BRINKER: Okay.

CHAIRMAN MAISEL: Dr. Vanden Hoek, you can have a seat. No one from the Panel has asked a question to you.

DR. VANDEN HOEK: It is possible that, because we do use the catheter in our institution, we also use cooling blankets and ice bags. Our intensive care unit made a decision a year ago to try to implement a cooling pathway as well. And because of the cooling blanket issues with temperature drops, our ICE had decided to implement a cooling catheter and try to control that temperature better.

CHAIRMAN MAISEL: Okay. Thank you for your comments.

Anyone else on the Panel have a question? John?

DR. MARLER: Well, I was just curious if this second speaker could say why in her hospital it wouldn't be possible. I don't know what numbers were involved, but it sounded from the energy of her presentation like there were quite a few patients in the registry. Why would it be so difficult to do a comparison of ice bags to cooling catheters?

DR. PEBERDY: I don't think that we have said that there would be a difficulty to do that. There are literally just a handful of hospitals now that are performing this therapy. Many are in the stages of developing protocols to implement hypothermia programs which, in a typical hospital, takes almost a year to work through the entire process and do the education and get the protocol development. So there are not really enough hospitals with patients that we would be able to get a meaning comparison at this particular time, although we are continuing to look at data that gets entered on hypothermia patients.

CHAIRMAN MAISEL: Mike?

DR. WEISFELDT: To either of the speakers.

Were you involved in the recent deliberations of American Heart and ILCOR CBR standards dealing with cooling? And if you were, was there any significant effort to vacate the ILCOR statement that was previously adopted?

DR. VANDEN HOEK: In answer to your question, the therapeutic hypothermia worksheets were presented at a plenary session in Dallas at the end of January. And I think that the consensus was that those guidelines will stand.

CHAIRMAN MAISEL: Any other speakers that wish to address the Panel this morning?

At this point I'd like to close the open public hearing and we will move onto the sponsor's presentation.

I'd like to remind the speakers to introduce themselves, to state their conflict of interest as well. Thank you.

DR. BLUMENSTEIN: This is Brent Blumenstein. I'm not hearing anything speaking. Is there a reason for that?

CHAIRMAN MAISEL: Nor are we. We're just getting started here in a moment.

DR. COLLINS: It's the sound of one hand clapping. We will start now.

Good morning. Thank you. My name is Dr. Ken Collins.

CHAIRMAN MAISEL: Could you speak into the microphone a little more.

DR. COLLINS: Is that better?

CHAIRMAN MAISEL: Yes.

DR. COLLINS: My name is Dr. Ken Collins.

Welcome here on St. Patrick's Day.

I'm Executive Vice President of Alsius Corporation. Clearly, my conflicts of interests are that I'm a full time employee of Alsius Corporation.

Speaking also today will be Dr. Sterz from Vienna. Dr. Sterz is the Associate Professor of Internal Medicine, University of Vienna.

And Dr. Risto Roine, he's the Associate Professor of Neurology at the University of Helsinki.

Because the order of slides will change if I don't now delete that slide since I've introduced everybody. And now your slide numbers will match.

Why we are here. Alsius submitted a 510(k) for the CoolGard/Icy catheter system for the induction of hypothermia after cardiac arrest in certain patients. The FDA requested that we conduct a randomized controlled trial for this 510(k). And we appealed this request to the FDA.

In response to our appeal, the FDA offered this Panel. Alsius would like to thank the FDA for this flexibility. And for the purposes of this presentation, we will constrict ourselves to the analysis from the 510(k) dataset.

This is a 510(k) notification.

Next slide.

Substantial equivalence to a predicate device is the standard for FDA clearance of a 510(k). The statutory provisions for 510(k)s refers to clinical and scientific data, and there is no specific requirement for an randomized controlled trial. Alsius believes that it has provided the FDA with the data to support substantial equivalence, and therefore clearance of this 510(k).

For the purpose of determining substantial equivalence for a device, the 510(k) should include according to the statute as quoted "appropriate clinical or scientific data." And the intent is to show that the device is as safe and effective as a legally marketed device.

You previously heard from Dr. Vanden Hoek, who is as you'll see, one of the primary authors on the ILCOR recommendations. ILCOR is the International Liaison Committee on Resuscitation. It is made up of multiple representatives from multiple countries. Typically, for example, in Dr. Vanden Hoek's position in representing the AHA or the European Resuscitation Council, or the Australian Resuscitation Council, the appropriate medical bodies.

The recommendations were first reached in October of 2002 and then published as per the reference at the bottom of the slide in 2003.

Of note, "unconscious adult patients with spontaneous circulation after out-of-hospital cardiac arrest should be cooled to 32 to 34 degrees for 12 to 24 hours with the initial rhythm has been ventricular fibrillation." They also commented that such cooling may be beneficial for other reasons or in-hospital cardiac arrest.

It was the statement that was bolded in that slide that became the indication for use for the Alsius 510(k).

In terms of predicate devices, the Alsius CoolGard system is cleared for marketing. I will give you a description of the device in following slides.

It is cleared for the induction, maintenance and reversal of mild hypothermia in neurosurgery, for rewarming in cardiac surgery and for fever control in the cerebral infarction/intracerebral hemorrhage. The other predicate device was a blanket system, the Thermorite Model HC-83.

The device provides controlled hypothermia. This device is already approved and widely used in the European Union and now approved in Canada.

In terms of the clinical data that was submitted, and which we will review in subsequent slides, Alsius provided in the 510(k) a review of the literature including summaries of some of the compelling animal models demonstrating the mechanism of the benefits of hypothermia, a meta-analysis, Alsius IDE Feasibility Study and an analysis of cardiac arrest patient registry at the Allegemeines Krankenhaus in Vienna, the university hospital in Vienna. To avoid me stumbling that, I will call it the university hospital in Vienna.

Cardiac arrest is one of the most complicated clinical environments. It is an intensive care environment with multiple interventions required simultaneously. Providing hypothermia via a central line is efficient.

The existing cooling methods, there are several. Fluids, limited by the amount of volume you can infuse. Ice/lavage, a nurse intensive and the wet surfaces pose risk. Surface cooling limits access to the body and patient and can increase shivering. As you heard today, it's associated with less control of the desired patient temperature.

Due to these disadvantages, and as explained by Dr. Peberdy, there is a need for better tool to provide hypothermia. The Alsius' device is a tool that allows controlled convenient hypothermia.

The CoolGard 3000 system, it's the picture of the external unit pictured here. It's a heat exchange system. It pumps saline to and from the catheter in a closed loop with temperature control. It cools in a controlled manner at rates between 0.05 to 1.5 degree C per hour.

On the projection now is a picture of the icy catheter in situ. It's a femoral vein insertion. It's a 8.2 French catheter that ends up with the cooling elements predominately in the inferior vena cava.

The close up picture shows the balloons axially mounted on the shaft of the catheter. And I do have and we'll pass it around later, catheters for you to play with, to look at.

Cardiac arrest has significant implications for public health. It effects a large number of people with a terrible outcome. Rhea et al reviewed 35 U.S. community hospitals. In terms of the survival for any rhythm it was 8.4 percent overall, and for primary VT/VF it was 17.7 percent. And larger cities, depending upon the emergency care available, the rates varied enormously. Annually in the United States, at least 155,000 people experience EMS-treated cardiac arrest of any rhythm recorded, and 60,000 experience EMS-treated ventricular fibrillation-rhythm as a cardiac arrest. It's a significant public health issue.

In the 510(k) we presented a meta-analysis. This is subsequently being published as a peer reviewed article in the well respected Critical Care Medicine. I believe you received a copy of that in your Panel pack.

Three studies were looked at to establish an expected result. The HACA study, which you've heard about before and which Mr. Sterz and Dr. Roine were both intimately involved, the Bernard study by my countrymen and the Hachimi-Idrissi study. The end result is a clear benefit of hypothermia in the comatose survivors of cardiac arrest.

This graph shows survival. It is a meta-analysis presenting the risk difference. You can see the three trials, the HACA, Bernard and Idrissi. And overall -- I'm going to use the lethal-looking laser pointer -- there's a clear benefit in terms of the risk difference and its confidence intervals.

Similarly, in terms of survival with good outcome, the three trials are presented. The overall 95 percent confidence intervals and the risk difference is significant.

Included in our 510(k) were the results of the feasibility IDE. Let me put this in historic context for you. We started with a view to understanding a perspective randomized controlled trial. We experienced very slow enroll despite our best efforts. We started in 2001 in March.

In February of 2002 the New England Journal of Medicine articles, the HACA study, the Bernard study were published. ILCOR met in October of 2002 and published its recommendations in the mid-2003.

At this stage we believed that there was established of hypothermia in cardiac arrest. The feasibility study enrolled 13 patients with inclusion criteria that were tied in line with what we expected would be required in a randomized trial. The 30 day survival was 69 percent and the adverse event profile was consistent with the published studies that I've just mentioned.

At the Allegemeines Krankenhaus in Vienna university hospital there is an ongoing data collection activity. They're interested in the use of mild hypothermia and the post-resuscitative care of the comatose survivors in cardiac arrest. It contains data from the randomized controlled trial, patient registry data and includes information on this device and several others. Because of confidentiality, Alsius does not have access to the measurement on other devices.

From this dataset selected controls and device data was analyzed and submitted in our original 510(k) notification. Over the past year there has been ongoing data collection in the registry. This includes additional patients in both the control and device cohorts. And the analysis that I present today are from the 510(k) dataset.

The selection criteria the patients had to be the comatose survivors of cardiac arrest with primary successful cardiac pulmonary resuscitation resulting in return of spontaneous circulation on arrival to the emergency department. This was consecutive series of adults who had survived at least 24 hours and who had had non-trivial resuscitation times.

This table presents the baseline data. There are differences between the device and control groups, as is made clear in this slide. I would like to point out that the GSC score on admission being equal to three is more common in the device group. And that the average ROSC time is also longer.

The imbalance would suggest that the device group where a patient population would grow to risk difference.

This is a crude survival analysis. The hypothermia group, 43 of 62 patients survived. And the controlled group 695 out of approximately 1200.

The risk difference unadjusted was 11 percent with a confidence interval being near significant.

Further analyses were done. This is a 30 day survival multivariate model with values related to outcome. The univariate analysis that I previously showed you, the survival is on the top line. They present results to both survival and survival with good neurological outcome.

If you look at the confidence intervals, you'll see the risk differences at 11 percent for survival unadjusted, 17.7 percent for survival adjusted. Approximately 11 percent for univariate survival with good neurological outcome and 23 percent for adjusted. The confidence intervals are significant.

Of clinical relevance, the number needed to treat the patients using hypothermia for the survival is between 6 and 9. And for survival with good neurological outcome is between 4 and 9.

It's a remarkable benefit, and it implies that for every six or so patients treated with the Alsius device, a life will be saved.

This analysis included propensity scores, and we did do a Hosmer-Lemeshow model of fit for this.

In addition, we did other analyses of propensity scores. Propensity scores are a method to attempt to create better match comparisons between cohorts.

This slide shows the distribution propensity scores of the device an the control cohorts. The lack of good overlap could suggest the groups are not well matched. However, the model which includes the scores and other based on variable that's displayed in this slide, provides an odds ratio I would suggest significant benefit to the device.

On this slide, the dataset, that's the internal code for the CoolGard system. We are looking at all baseline variables.

For the code variate set, set number one is the treatment group and the propensity score. Set number two is the treatment group, the propensity score and all other variables used in the propensity score analysis.

I draw your attention to the odds ratio. Note the highly significant confidence intervals.

In addition, we present logistic regressions of survival with good neurological outcome for both the unadjusted and adjusted and death -- I'm sorry for switching axis for you, less than discharge. And, again, looking at the confidence intervals around the odds ratios, these are significant results.

In summary, may I first of all point out that the efficacy outcome is also the primary safety outcome. Efficacy is survival.

Vienna registry data shows significant improvement in survival and survival with good neurological outcome with the Alsius CoolGard system.

Dr. Roine will later discuss the meaning of the term "good neurological outcome" and the use of the CPC scale.

The effect as seen is consistent across all the different analyses with that reported in the New England Journal of Medicine published randomized controlled trial and the meta-analysis based upon them. These data are appropriate and adequate to demonstrate that the device is as safe and effective as the legally marketed predicate.

I will now hand you to or hand over the microphone to Dr. Fritz Sterz.

DR. STERZ: Good morning, ladies and gentlemen. I hope you will understand my Viennese English. My teacher, whom I want to give credit to, always said it's Vienglish. My teacher was Peter Safar at the University of Pittsburgh. And without him, I wouldn't have been able to stay here and talk to you, which I am very proud about.

So Peter Safar 15 years ago in his lab has taught me how to apply hypothermia in animals.

For this meeting I am paid Alsius with regards to the travel expenses. And with regards to the soldiers working at home, I get both paid a fellow by Alsius. Unfortunately, that's all what I get by Alsius. I would like much, much more.

We opened our department in 1995 at the Vienna General Hospital. And since this time I'm collecting cardiac arrest data according to the recommendations which was initially coming from Peter Safar's PRCP trials and later on from the Utstein styled recommendation. And since this time I have a committee approval HACA registry running at our department. At the moment we are counting 2,500 patients in this registry, which have been admitted to our department.

We have also initiated together with Dr. Roine of the University of Helsinki, the so called hypothermia cardiac arrest trial, which was a multisample trial in Europe and somehow in New England Journal has been published and proven that hypothermia seems to work.

We continued after HACA cooling patients, not only with Alsius, also with other companies which I'm grateful for because they paid the other soldiers which I needed to do this whole work. So I'm paid by several companies, and I might not be biased to what's Alsius in this regard.

So we are entering patients in this registry who had cardiac arrest either in the hospital, in other departments, admitted to our department either outside the hospital or even if they had their arrest in the emergency department. The criteria for getting into this registry was they had to be in our department.

And finally but not least, we have gotten where we are grateful to the data from Bernard and Hachimi-Idrissi to do a so called meta-analysis together with our HACA data.

Next slide.

To give you an idea, this is our hospital. It's a 2,500 bed hospital in the middle of Vienna. It's one of the major university hospitals in Austria. And it has a Seattle-type emergency department where I am based. It has nothing to do with emergency medical service, which was created by our Emperor and which was established already 120 years ago. And since then is run by emergency physicians.

We have physician staffed ambulances in Vienna. We have also a helicopter. And these ambulances are running 24 hours a day. Beside this we a lot of EMT-staffed ambulances. This is only to give you a picture about our department and about my surrounding.

The department is treating about 300 patients per day.

Next slide.

And this is how it looks in our department. Maybe a little bit different than in American states from what I have seen.

We have a so called acute care unit within our department where we can treat patients like in an intensive care unit, and we are approved as an intensive care unit. And this is how it looks like if a patient is treated with the Alsius device and what kind of establishment is there, I don't have to describe in detail.

What I want to point out in recent times we have added to the Alsius device giving these patients in the beginning when they have been admitted, cold fluids to be faster in initiating the cooling or initiating the lower temperature because it always takes a little bit of time to get the Alsius device established.

Next slide.

This was the reason for changing to Alsius and the changing to the cold fluids given in the initiation because it took us a very long time in the HACA trial with the cold air blowing on these patients to get the temperature down to our, between Risto and myself and the other partners of our HACA trial decided target temperature which was based mainly on the animal trials which I did with Peter Safar. And I had a continuous conversation with Peter Safar how we should do it, and he was somehow a consultant for this study.

And we kept these humans for 24 hours in this kind of range, and then rewarmed them, at this time, not actively unfortunately, which we should have done.

Next slide.

This is now the same picture with the Alsius device where we are much faster, much better in controlling this temperature and much better in controlling the rewarming. And this has been only on those patients where no cold fluid has been used. And what you can see, why we now switched to cold fluids in the beginning is this increase of temperature in the beginning of the admission time.

Next slide.

And these patients now and patients from our registry which hadn't been used for testing other devices for cooling the HACA patients, we used to analyze the safety between the device and the controlled group. However, we decided it wouldn't be fair to put all these patients into the analyses, and therefore we selected between both groups a certain group of patients which had ventricular fibrillation on presentation, presumed cardiac cause of cardiac arrest. So we made before we delivered the data to our statistician a selection of our patients to have the group somehow the same in both groups. We, of course, not included in this analyses of patients who had, for example, stroke as their reason for cardiac arrest.

In the controlled group you can see the mortality was still very, very high, as we have seen it in the HACA trial.

And here you can see the green, the normothermia group where we have selected now 81 patients, where the hypothermia group the so called safety data with regards to what is always discussed bleeding, pneumonia, sepsis, pancreatitis, rental failure, bradycardia and all the other arrhythmias which could occur during cooling. What you can see is that we have tiny little bit more renal failures or much more renal failures. This was due to the definition of our renal failure. Our renal failure definition was an increase of the serum creatinine of about 50 percent. However, no one of these patients needed hemodialysis.

Bradycardia was expected to be a little bit more in the cooling group because that's due to the cooling. But there was no more bleeding, no more pneumonia and the only thing which is different is the pancreatitis issue. But if you look at the number it's here 4 in the hypothermia and zero in the controlled group. I don't know if we can really count on this, however this pancreatitis was also only defined by the serum amylase level which was increased and resolved within one week.

Next slide.

So again, we had no major problems with regards to bleeding to hematologic situation. We had no significant differences in lethal arrhythmias. WE had more bradycardias. We had this renal serum creatinine increase but it was solved within rewarming.

You can see these data presented in a paper which was just recently published in Resuscitation that it resolves by itself. And we had this observation that the serum amylase increased, however also resolved within 24 hours.

We had pneumonia in both groups, but no significant differences.

This compares somehow to the HACA trial, but we also prospectively looked at these side effects where we have in the protocol given definitions for all these kinds of side effects or safety issues. However, we haven't been in the presentation of the data of the HACA trial, such rigid on the arrhythmias which could be presented because we have this data available. But comparing this data to our data with the CoolGard device doesn't show a major difference.

Okay. You go ahead.

Once more, thank you very much. And I hope I clear a little bit of this picture.

DR. COLLINS: In summary, these are critically ill comatose patients. This is an intensive care environment with multiple interventions required and there's a high rate of adverse events in both groups.

There were no cardiac and hematologic compromise from cooling. And the increased rates of renal insufficiency and chemical pancreatitis resolved with rewarming.

The use of the device provides the desired cooling with an acceptable safety profile.

I'd now like to ask Dr. Roine, Risto Roine to address the Panel.

DR. ROINE: Mr. Chairman, dear Panel members, my name is Risto Roine and I'm a stroke neurologist from Helsinki. I am the Associate Professor of Neurology at the University of Helsinki, Finland and the chief physician of the Acute Stroke Unit which is one of the largest in Europe.

I have been involved in resuscitation research since '85. And I was one of the founding members of the HACA trial, as Fritz Sterz told you. And I'm also involved in the European Resuscitation Council HACA registry.

And my financial disclosure is that I have no financial interest in Alsius, any of their products or any other product in the market. But Alsius is taking care of my expenses to attend this meeting.

I was asked to comment briefly on the outcome scale, cerebral performance categories scale used in the HACA trial.

As you know, it reminds very much the other outcome scales used in stroke research like Glasgow Outcome Score and the modified Rankin score. And the main thing is that there is a cut off point between CPC 2 and CPC 3 defining independent or dependent outcome, which was also used as the dichotomous outcome in the HACA trial.

Next.

The CPC Outcome Scale is currently being recommended by the Utstein style uniform reporting guidelines ever since 1991, and it still is. It's recommended by most, if not all, reviews written on this topic. And it's used in most major resuscitation research trials. I found 52 papers since 1990.

It's always used in the conjunction with the OPC scale, the overall performance categories. And it's considered to be a more relevant measurement for cognitive outcome than the Glasgow Outcome Scale or modified Rankin score used in stroke research. But it's easy to derive the Glasgow Outcome Score Scale and the modified Rankin scale from the CPC and OPC scores.

And, as I said, the limit between independent and independent categories between CPC 2 and CPC 3, this is a very, very clear cut definition. CPC is a valid scale for cardiac arrest outcome studies.

Next please.

Then I very briefly described the situation in Finland. As you know, Finland is a cold Nordic country with a population of 5 million people. But we quite heavily use therapeutic hypothermia in Finland. It's the hypothermia for cardiac arrest is adopted in all national and local guidelines. It's being routinely used by all five universities hospitals, and all these university hospitals, those have the Alsius product that we are talking about today for endovascular cooling. And three university hospitals do also have devices for external cooling, mainly the Allon 2001 device.

My Helsinki University Central Hospital has four units of this Alsius device and two units of the KCI TeraKool device for external cooling. More than 90 percent of our patients receive endovascular cooling instead of the external cooling at this point.

We just did a survey in Finland by the Critical Care Society of Finland just a few days ago showing that six of our 15 central hospitals have the Alsius CoolGard device only, and eight of the 15 have the Allon device only. One central hospital did not reply.

Next please.

So in Helsinki, more than 90 percent of all eligible out-of-hospital VF cardiac arrest patients are routinely cooled, that was 48 patients in 2004. And of all patients who were entered into the European Resuscitation Council HACA registry in 2004, 72 percent are now alive and at home. This figure is considerably higher than the general outcome of VF cardiac arrest patients at the time when I did thesis about this very topic in 1993.

So I finish here. Thank you very much for your attention. Thank you.

DR. COLLINS: In conclusion then, this is a 510(k). The product is already clear and on the market for three different temperature management indications. Of course, hypothermia and normothermia applications.

There is clear independent evidence of the survival benefits of hypothermia in cardiac arrest patients. The ILCOR recommendations are clear.

We have presented data to show that the Alsius CoolGard device achieves the desired result effectively and with a favorable safety profile.

CHAIRMAN MAISEL: Thank you very much for a very concise presentation.

I'd like to open the discussion up to the Panel to ask any questions of the sponsor, of course reminding the Panel that we'll have several hours to ask questions later.

Rick, why don't we start with you?

DR. PAGE: Thank you.

The feasibility study enrolled only 13 patients in 18 months with several centers. Can you explain that to me, because I'm hearing of the zeal for this technology and the need for it, and that enrollment is surprisingly low?

DR. COLLINS: I think the comments about the historical context. This technology is evolving very rapidly. At the time we were used an informed consent, which is very difficult to do when someone arrives basically dead. So we were unable to obtain informed consent, is the largest reason for not being able to enroll.

And the community resistance in two of the areas when we started to examine community consent around the Duke University at that time, there was considerable community resistance to involvement in clinical trials under community consent. And we did not go down that path.

DR. PAGE: Thank you.

CHAIRMAN MAISEL: Jeff?

DR. BRINKER: Ken, you mentioned the lack of shivering -- or actually you had mentioned the lack of shivering, you said external devices can produce shivering. This device doesn't produce shivering?

DR. COLLINS: All devices produce shivering. Shivering is a -- both a spinal mechanism with a central control and the degree to which you see temperature responses to which you see shivering depends upon the patient's level of consciousness, the drugs on board. But the initiation is highest if the areas of skin around the neck, the armpits, the exposed vessels have high numbers of sensors for the response to be simulated.

Once you get any patient's temperature down to somewhere 35.5 and 36 you will get shivering in patients. But the skin contact alone can potentiate the degree of shivering experience.

DR. BRINKER: There's a trial of a intravascular device for the performance of interventional cardiology. And that produced a huge amount of shivering. Those patients, of course, were more conscious and require drugs to control this.

So the question is do patients need to be treated with sedatives and/or blocker?

DR. COLLINS: Well, I'll make one comment and then I'll pass this to the clinicians.

This is indicated for the comatose survivors. And in the comatose survivors, at least initially, it's not such an issue. But I'll pass on to Dr. Sterz.

DR. STERZ: If patients are lucky, they're found by bystanders. Then the ambulance service is called and tries to resuscitate them. Then after establishing restoration of spontaneous circulation, they either receive sedative or analgesic drugs by the ambulance service, depending on the awareness or depending on how they are doing. But most of these patients come unconsciousness without any sedation and drugs to our department. And they're immediately assessed on admission about their Glasgow Coma Score Scale and Pittsburgh Brain Stern Score and so on.

For then further treating them we had a standard order of procedure that sedation and analgesia plus paralyzers has to be used in these patients according to a protocol for 24 hours.

So the assessment of the patient was either done by the ambulance physicians or mostly by ourself with regards to being comatose for being included in the study and then immediately afterwards we started with sedation, analgesia and paralyzers for the duration of cooling until they reached 36 degrees again.

DR. BRINKER: Okay. And the patients in the studies that were normothermic, did they get the same blocking agents and sedatives?

DR. STERZ: This is a routine protocol regardless without or with cooling. And the data in the HACA trial have shown that the same drug amounts of paralyzers and sedation and analgesia was used. Right now in this series most of the control patients came out of the HACA trial, therefore the same regiment was used. And in addition this regiment is used anyhow if they get cooled or not.

DR. COLLINS: I think you're actually referring to Alsius as normothermia trials were you?

DR. BRINKER: Yes.

DR. COLLINS: Yes. Many of those patients were ventilated and therefore under some form of agents. In the normothermia patients we tend not to get shivering provided the skin around the groin is protected from the cold of the catheter.

DR. BRINKER: So when you give these drugs, do you actually -- in the hypothermia patients, you do not see shivering? Do you see physically shivering? What I'm getting at is could the shivering account for something like the renal problems that I've been seeing, either by myoglobin or something like that?

DR. STERZ: Well, for example in the HACA trial we didn't see these renal problems. This with regards to the renal problems, I cannot report about if our patient shiver or not, because they have both groups, control and hypothermia groups, on this sedation and analgesia and paralyzing drugs right from the beginning, before they reach the temperature.

We have experienced this in some myocardial infarction patients which have been conscious using this device and trying to cool them. And there we got incredible problems because of shivering and discomfort of our patients. But these were conscious patients and not cardiac arrest survivors.

DR. BRINKER: Fine. Thank you.

DR. COLLINS: And one final comment. The company, Alsius has done a feasibility trial in acute myocardial infarction and the results are published, and therefore in the public record.

In those patients in which adequate paradigm was present, adequate being they were still breathing but were not shivering, there did not seem to be any ECG evidence of myoclonic activity suggesting that there wasn't actually sort of subobvious shivering going on. It can be suppressed.

CHAIRMAN MAISEL: Dr. Brott and Dr. Somberg.

DR. BROTT: Two, hopefully, brief questions.

We were provided the HACA paper, and it states to prevent shivering paralysis was induced by the intravenous administration of pancuronium for two hours.

It's a little ambiguous. Was every patient in normothermia paralyzed for 32 hours?

DR. STERZ: Yes. Clear, straight answer. Yes.

DR. BROTT: Okay. Second question. Thank you very much for coming, Dr. Roine from Finland. And I thought your information was very interesting.

I'm wondering if you could give some raw numbers? You know, we had percentages. You know, greater than 90 percent of patients had endovascular cooling. It said four hospitals. It's at your university hospital. But do you have numbers in terms of total admissions of patients with coma after cardiac arrest? How many received cooling of any kind? How many received cooling with this device? How many had this SAE or didn't? You know, some hard numbers.

DR. ROINE: Yes. Thank you.

In Helsinki all cardiac arrest patients are being treated at one hospital. They have the central hospital. And between 90 and 95 percent of those patients will be treated using hypothermia. Only if it's found out that the resuscitation was not indicated or the patient is in terminal condition, or there is a contraindication for intensive care, then hypothermia was not used. And nine patients out of ten will receive endovascular cooling because we have actually three units available. The fourth unit is in another hospital.

DR. BROTT: I understand that. What I was wondering the raw numbers. Not percentages.

DR. ROINE: Yes. Okay.

DR. BROTT: The absolute raw numbers of patients who are admitted with out-of-hospital cardiac arrest.

DR. ROINE: Yes.

DR. BROTT: How many with coma, how many were treated. You know, the raw numbers with hypothermia of any kind.

DR. ROINE: All right. 2004 and 48 patients were treated with endovascular cooling at our hospital. They were all VF cardiac arrest patients. Forty-eight.

DR. BROTT: And how many with this device?

DR. ROINE: All of them were treated with this device.

DR. BROTT: Thank you.

CHAIRMAN MAISEL: John?

DR. SOMBERG: A couple of quick questions relating to your current presentation. One is I'm most interested in the Austrian registry and what you've presented. Could you tell me the time from event to resuscitation and the time from event to cooling? One of the things that came up in an earlier Panel meeting was the difference between the European experience and the United States experience.

DR. STERZ: Due to having this acute care unit in the emergency department, which is an intensive care unit, we have this tool of making this chain of survival very short and not having these problems within the hospital to get patients to the ICUs. And, therefore, our lack or the time which was missing was the time of the ambulance drive to the hospital, which was in an average 20 minutes.

So these patients got cool within an average to 33 degrees, as you can see on these temperatures curves which is always counting from restoration of spontaneous circulation until reaching the target temperature, of approximately if I recall it right four to five hours. I don't have it now in my mind, but reaching the target temperature was in the endovascular cooling between four and five hours. And in the HACA trial around eight hours. I don't know how it is --

DR. SOMBERG: But from the time of the event to the time of initiation of this process?

DR. STERZ: Initiation? How long it takes from admission to initiate --

DR. SOMBERG: No, no. Let's say someone has an arrest and they are discovered. From the point they're down they receive -- you know, from the time of the arrest to the time they are down, it's all you've had in Europe, the time is 2, 3, 4 minutes up to 6 minutes; it's different than the States. I just wanted what the time was for this subset you have there.

Do you have rapid resuscitation in Austria?

DR. STERZ: Yes.

DR. SOMBERG: Okay. So you do. So you're in these very low numbers. And then from the time that they receive the resuscitation effort to the time they initiate cooling?

DR. STERZ: It's approximately 20 to 30 minutes.

DR. SOMBERG: Twenty to 30 minutes. Okay.

Also, can you tell me in this registry I was confused by the material booklet in that many of the patients who received the CoolGard system also received in the subanalyses rapid infusion of cold saline. Was this the case in the material you presented?

DR. STERZ: This was the case in the last series of patients, which we did now. Not in the first series of patients where we didn't do ice cold fluid cooling. The percentage of patients, I think the first 50 patients hadn't been cooled with additional ice cold fluids. Then due to this short delay of 50 minutes until getting endovascular cooling started, we decided it have to been earlier and to take out the fluids from the fridge and initiate cooling right when they are coming in the door.

DR. SOMBERG: So you would say that the fluid is an integral part of the process of --

DR. STERZ: Right now, yes.

DR. SOMBERG: Yes. Okay. And can you also in the registry clarify how the controlled group is constituted? Is this done by a committee that doesn't know the results of the intervention group? You know, is it a blinded selection? Is it prespecified criteria before the controls are selected, etcetera?

DR. STERZ: The endovascular cooling or the cooling at all, which device whatever we take, the patient have to have certain inclusion criterias. And unfortunately right now the controlled group shouldn't have these inclusion criterias. So the controlled group are different.

And the inclusion criterias I have shown in this one slide are very general right now as compared to the HACA trial where we have been much more strict. I'm hearing now that they're even much, much more generous than we are. We take, for example, no patients which had very short resuscitation times.

Whether they're comatose? They have to be, they have to have primary successful restoration of spontaneous circulation, they have to be over 18 years and they have to have patients who have at least survived 24 hours, but this was done for the analyzers retrospectively. And for the analyzers retrospectively we took patients only which resuscitation times of more than one minute.

DR. SOMBERG: But I thought I understand you is that you an overall control group, which was very broad in general, and then you clarified that down or distilled it down to a smaller group. I just don't understand the process of how that was done.

DR. STERZ: The majority of the overall control group was historic group. Was the control group before 2003 when we started with endovascular cooling. The majority of these control group patients is coming from this patient population. And much of them or many of them are coming either from the HACA trial where we did the prospective randomized trial and where we had the controlled group selected.

DR. COLLINS: I think to answer your question, because I do know the answer, Alsius approached the Professor for access to his data. There is a statistician on staff, Dr. Marcus Mullner at that time at the university. He wrote a prospective selection criteria, and then that is applied. The set was extracted and the analyses was done.

Does that answer your question?

DR. SOMBERG: Well, it tries to.

And I have one just final question, very quickly. What is the cost of this procedure of cooling?

CHAIRMAN MAISEL: We're not here to consider the financial issues. You're welcome to answer the question if you wish, but you're not obligated to.

Okay. John?

DR. MARLER: Yes. We're looking at a device that, as I understand it, is put in through the groin into a major vein. Left there for a long period of time, many hours and then withdrawn. And yet in all of the complications, I looked through the material and perhaps I missed it, but I don't see any complications like bleeding like hitting an artery or a nerve instead of the vein or injuring them in the process. I don't see any evidence of any pulmonary emboli. I don't see any evidence of having to withdraw the catheter early or to use more than one device to obtain the successful cooling. And that just kind of goes against my common sense that the nature of this kind of device. And I was wondering if there's any such events that happened or have they just not been reported? How is this possible?

DR. STERZ: With regards to pulmonary embolism and deep vein thrombosis, all these patients are heparinized. A lot of these patients receive thrombolysis or go to the cath lab because they have been myocardial infarction patients with cardiac arrest. And we haven't seen such kind of emboli or some deep vein thrombosis. We have seen bleeding, but this data have been presented.

We haven't any major problems as usually what you have with puncturing a vein. If you are in the artery, you withdraw and then you try to hit the vein beside it. But that's routine clinically doing if you use hemodialysis or if you cannot a vein in the subclavian vein, then no major problems with regards to puncture or no major problems with regards to placing the catheter. I don't know if Risto has the same experience.

DR. ROINE: Yes. I have nothing to add to this.

DR. MARLER: Is anticoagulation a routine a routine part of external cooling?

DR. STERZ: Not of external cooling, but of these patients. All these patients are intensive care unit patients and they receive somehow anticoagulants.

DR. COLLINS: The company has also submitted to the FDA in a previous application the randomized control trial of normothermia patients. And as a central line, the catheter performs as a central line with the usual complications of a central line.

CHAIRMAN MAISEL: Henry?

DR. HALPERIN: A few questions.

There was a substantial difference in the statistical significance between the analyses that were done unadjusted and adjusted. Could you comment on the particular variables that made the biggest difference in the adjustment?

DR. COLLINS: I'm going to ask Joel to come and help me with this one.

DR. VERTER: Good morning. My name is Joel Verter. I'm a statistician with Statistics Collaborative. And I helped the company with some of its analysis. I did not do the original analysis, as Dr. Collins noted. That was done by Dr. Mullner in Vienna.

The variables that we used were pretty much -- there were many models, first let me say. But the variables that we used were the baseline variables that were available to us. And I believe what you're asking did we do some kind of stepwise analysis --

DR. HALPERIN: Well, what particular variables made the biggest difference is really what I'd like to know.

DR. VERTER: Right. Yes. And I believe that the only answer to that would be if we did some sort of stepwise, and we did not do that. We did the univariate models, we did bivariate, we did group in each individual variable and then we did the complete set.

DR. HALPERIN: So how many variables were used in the adjustment? Was all those, there were like variables or something like that?

DR. VERTER: That's about correct, yes.

DR. HALPERIN: Because you have like 60 patients or something like that?

DR. VERTER: There was 62 coolings.

DR. HALPERIN: And did you do an analysis to see if you could use that many variables to adjust?

DR. VERTER: The issue for us was that, as Dr. Collins noted, we were limited by the 510(k) application. And that model which was done in Vienna used all those variables. I would agree that ten, but 62 may be pushing the limits.

DR. HALPERIN: So you didn't really look for one particular thing like time to resuscitation or age or anything like that that was actually a particularly significant variable?

DR. VERTER: The only analysis I could speak to you about that is the bivariate ones which we did here which included group and each one of those individually. And I'd have to go check my notes, but on an individual basis none of them seemed to have a major impact. When you put them altogether, they had the biggest impact.

DR. HALPERIN: Okay. In the Helsinki data you mentioned there were 40 patients or so that met your inclusion criteria over some period of time. Could you comment on many total cardiac arrests there were in Helsinki?

DR. ROINE: I didn't bring the actual numbers with me. And, unfortunately, to match everything else the cardiac arrest cases are VF in Helsinki. And of those patients with restore spontaneous circulation and admission to hospital, almost all will receive hypothermia at our institution.

DR. HALPERIN: But these 40 patients, what percentage --

DR. ROINE: Forty-eight.

DR. HALPERIN: Forty-eight. What percentage of the total cardiac arrest population does that represent?

DR. ROINE: It's more than 100 in Helsinki, in the city of Helsinki.

DR. HALPERIN: So there's a 100 cardiac arrests in Helsinki over that period of time?

DR. ROINE: Sorry not to bring the actual numbers with me. I just don't have the actual numbers.

DR. HALPERIN: Okay. And from historically do you know what the survival was in those patients before they got cooling?

DR. ROINE: Yes. The secondary survival rate in the patient material in my thesis was around 40.

DR. HALPERIN: Forty. So you went from 40 to 70 in that limited population.

DR. ROINE: Yes.

DR. HALPERIN: Okay. And then one other thing is there was an issue of informed consent. You said that that limited the number of patients enrolled early. And this would be consent in patients who were successfully resuscitated and then were in the hospital. So the pressure of time probably really wasn't really there like what it would be during trying to do a new CPR intervention during actual resuscitation.

So what were the issues in obtaining informed consent?

DR. COLLINS: From memory the way it was structured, at last the patients who were comatose, you had to get family member consent. And that was the time --

DR. HALPERIN: And then what's changed in allowing this to become generally used? Because the same issues would be there?

DR. COLLINS: I'm sorry, I don't understand the question. The device is released for this indication in the United States.

DR. HALPERIN: So you're talking about this would be just in the U.S., not in Europe?

DR. COLLINS: Yes.

DR. HALPERIN: Okay.

CHAIRMAN MAISEL: I wondered if you could address an issue of surface cooling versus endovascular cooling? The only randomized data you're presenting today is on surface cooling. We've heard from a number of speakers this morning about how the time course for the endovascular cooling is different from surface cooling. We heard from a public speaker about overshooting or not reaching the temperatures fast enough. We heard from Dr. Sterz about how the endovascular catheter cools much more rapidly. So I have several questions related to that.

Number one, do we have any data that cooling faster is better?

Number two, with regard to the registry, maybe you could just be a little more explicit and explain why some patients were cooled and some were not, because that seems that the issue of interpreting the data you've presented.

And I also wonder if you could comment on why some patients were endovascularly cooled in the registry and some apparently were surface cooled or cooled by other means?

DR. STERZ: So when we finished HACA, we were approached by Alsius and they were asking us to use the Alsius device. So in the first period, I don't know when we started exactly, we used the Alsius device only. Then later on Allon came with this surface cooling device. And a little bit later on Medivance came with their surface cooling device. And just recently we have been approached by KCI to use their new surface cooling device, which we have formally used in the HACA trial.

Data of all these surface cooling devices are available, but I have undersigned with the companies that are not allowed to present them. Therefore, I cannot present them. Maybe, I don't know -- but I have them. I have them on my laptop. If it's really needed and if it's permitted.

With regards to which patient is getting what when, we have a list in our department where boxes are empty below the certain surface cooling or endovascular cooling methods. And to be fair to every method, this list has always to be in a straight line that all these kind of devices are used in the same amount of numbers than the others. However, due to be Alsius first with us, they had an advantage of being used in one year only their device. And therefore, in the beginning mostly the Alsius device used. Just recently we used all of the other devices.

With regards to the inclusion and exclusion --

CHAIRMAN MAISEL: Can I interrupt for one second. Have you compared the endovascular cooling to the surface cooling in your registry data base?

DR. STERZ: Yes, we have, for us.

CHAIRMAN MAISEL: Without violating any of your confidentiality agreements about specific companies, can you tell us whether they are comparable?

DR. STERZ: They are comparable.

CHAIRMAN MAISEL: So with regard to a mortality and safety issues --

DR. STERZ: With regard to the efficacy. With regard to the outcome, I have too less data in these patients with surface cooling yet.

CHAIRMAN MAISEL: So there is no statistically significant difference in survival or survival with good neurologic outcome between endovascular cooling and surface cooling in your database?

DR. STERZ: At the moment it doesn't look like this, no.

CHAIRMAN MAISEL: Okay.

DR. STERZ: Of course we don't have the data.

CHAIRMAN MAISEL: Right. I understand.

DR. STERZ: With regard to the inclusion and exclusion, I want to point one thing out. When we stopped HACA we were so convinced that cooling helps these patients, that at the beginning we stated this SOPs with the inclusion very general. However, to get a team like our team moving in this direction, in the beginning of such a new treatment you lose a lot of patients which are not cooled at the end, because in the night shift there's a fellow who maybe doesn't like it.

So in the beginning we had maybe lost a lot of patients which should have been cooled. Right now we don't lose any patient anymore because the people are used to it now and we have the experience of 120 endovascular cooling patients now. And I think that's quite a lot. And this proves that it's in general use.

CHAIRMAN MAISEL: Okay. I'm a little troubled that there is data that would potentially be very useful for this Panel to review with regard to the surface cooling and your database that we aren't allowed to look at.

Mike?

DR. WEISFELDT: Yes. I want to pursue the subject of the acute renal failures statistically significant difference in part as an extension of Dr. Marler's comments about the potential complications of putting a catheter in through the vasculature. And I want to confirm what I think I've heard, I guess first, which is that although the word acute renal failure is the descriptor, the definition that was used is a 0.5 milligram percent increase in creatinine. Is that --

DR. STERZ: Yes. This was taken from the literature.

DR. WEISFELDT: Okay.

DR. STERZ: I think this is a very tough definition. This definition is not acute renal failure which needed hemodialysis in these patients, or which needed some type of specific treatment. And if you look at our recent publication about the renal problem in cooling such patients published by Andrea Zeiner in Resuscitation it resolves within seven days. You don't see it anymore. You don't have to do anything except that you have to do the fluid management according to -- but that's what you have to do anyhow in such patients.

DR. WEISFELDT: I missed your last statement, but I want to go on with my line of questioning. You've answered the first question.

The second question is I want to confirm again that this is not seen with surface cooling as in the HACA trial? So this is -- the increase in creatinine to that level which is statistical significant was seen in the database study and it was not seen with surface cooling as in the HACA trial.

DR. STERZ: That's right.

DR. WEISFELDT: Okay. And can you give me the mean or the variance or do you have a plot of the actual highest creatinine in the patients who have had acute renal failure by that definition?

DR. STERZ: I have it in this paper. There is a very graph where you have the mean value over the time.

DR. WEISFELDT: In these patients?

DR. STERZ: In these patients, yes.

DR. WEISFELDT: Okay. And where is that?

DR. STERZ: It's in Resuscitation just 2004, Zeiner was the first author. Andrea Zeiner.

DR. WEISFELDT: Do you remember how many had a creatinine over two or three?

DR. STERZ: Very few.

DR. COLLINS: I just please respond to Dr. Maisel. The company cannot coerce other companies to submit their data. And although I understand your discouragement, we cannot as a company nor can the Panel force a company to participate who is not a Panel presenter.

CHAIRMAN MAISEL: I understand. And I did not mean to imply that the company in anyway was withholding data. I guess I should have better underscored that. We're talking now about registry data where we don't actually have full access to the data.

Anyone else on the Panel have questions for the sponsor?

Norm?

DR. KATO: Do you have any other information about the performance of this catheter in other clinical situations? And one of the questions we're asked is to measure this device against itself. And while apparently it is indicated for use in cardio thoracic surgery, as a practicing heart surgeon I haven't seen this device ever used. So I'm kind of curious to see what other -- in order to make this leap for your 510(k) submission, we have to have enough data in order to have some basis for that next step. So do you have anything to present along those lines?

DR. COLLINS: The company has three indications in the United States, all three of which will be, I guess, normothermia in cerebrovascular hemorrhage. A 300 patient randomized controlled trial was done and the data was submitted.

The company is involved in Europe in cardiothoracic surgery indications and has just published, and I will try to obtain for you if I can-- I didn't bring them with me -- information related to the use of the product in the post-surgical recovery of patients. The catheter's purpose there is simply to prevent the after drop decrease in core temperature delays extubation in patients after cardiothoracic surgery.

Our company is involved in the HACA-ERC registry. The registry is a several hundred patient now study in Europe that looks at cardiac arrest. I may not under European law share that information with the Panel until such time as it has been made public. That's their privacy directive. I can't change that. But we have a large ongoing post-market surveillance for that in Europe.

In terms of the normothermia application post-market, we have an ongoing post-market surveillance of the normothermia population for which I think you received a copy of our latest report.

And the report of the 300 patient randomized controlled trial was available to you by the 510(k) submission if you wish to see it.

DR. KATO: But, I mean, you don't have any data or slides --

DR. COLLINS: I didn't come prepared with other than this Panel presentation.

DR. KATO: Okay. Thank you.

CHAIRMAN MAISEL: Mike?

DR. WEISFELDT: In those patient groups how long is the catheter in place, and this is not the cardiac arrest patients for whom the device is approved, how long is the device used in neurosurgery or other, obviously in comparison to the 24 hours here?

DR. COLLINS: The period post-surgery is typically the 24 hour recovery period. In the normothermia condition it's up to seven days. The catheter in this case is labeled not to be used for more than four days.

CHAIRMAN MAISEL: We'll have additional time later today to ask questions.

Why don't we take a 15 break at this point and reconvene at 10:00 a.m. for the FDA presentation.

(Whereupon, at 9:42 a.m. off the record until 10:02 a.m.)

MR. FELTEN: I'm Richard Felten. I was the team leader reviewer for the Alsius 510(k). The other members of the review team and who will be presenting are here: Dr. Swain, who will be doing the clinical review; Dr. Zhao who is our statistician giving the statistical review, and; Dr. Lazar will give you a short presentation on neurological endpoints as we've used those to interpret this data here.

As Dr. Collins has already mentioned to you the Alsius CoolGard 3000 is already a legally marketed product through the pre-market identification process. And the device that is the subject of this discussion is identical to the device that has already been cleared through pre-market identification for three specific indications for use.

The first indication for use, as you notice, is for fever reduction as adjunct to antipyretic therapy in patients with cerebral infarcts and intracerebral hemorrhage who require access to central venous catheter circulation and who are intubated and sedated.

Next slide.

This indication was actually based on a clinical trial data from Alsius. And one of the things our division routinely does when we have a very new use or a very new indication for use in the pre-market notification is ask the companies to include some of their clinical trial data. Alsius has done this, as you notice with this warning, because in their actual clinical trial they actually looked at four different groups of patients. Our plan was limited to only two of those because there is some suggestion in this data that there is absolute change in mortality, although those were not significantly different.

The package that you received does contain the most recent summary of the post-market surveillance data that they are now conducting in this area of fever reduction.

Next slide.

The second indication for use they have is for use in cardiac surgery patients to achieve and/or maintain normothermia during surgery and recover/intensive care. And as Dr. Collins also indicated, this means that the device actually was used probably up to 24 hours after the completion of surgery. And the intent was, as he mentioned, to stabilize the rewarming process as a way to avoid the downside.

Next slide, please.

And the third indication is to induce, maintain and reverse mild hypothermia in neurosurgery patients in surgery and recovery/intensive care.

You should also be aware that all three of these indications for use were granted to Alsius and several other companies based on clinical trial data. So that all of these studies do have clinical data to support indication use they have. And the indications for use are very specific because the clinical trial data is very specific for that use.

Next slide.

As you also are aware now, Alsius is now requesting a new indication for use. And this indication for use, again, is very specific and it's based on the ILCOR recommendation. And that is for use in the induction, maintenance and reversal of mild hypothermia in the treatment of unconscious adult patients with spontaneous circulation after out-of-hospital cardiac arrest when the initial rhythms was ventricular fibrillation.

And to support this request -- next slide -- what Alsius has presented to us is the meta-analysis with the three surface cooling studies, a 13-patient prospective uncontrolled U.S. feasibility study which was done under an approved IDE. And then the prospective, non-randomized single-site observational registry with the matched controls from AKH Registry Data. And this clinical trial data will be discussed by Dr. Swain.

Thank you.

DR. SWAIN: Let's see if I can find the presentation here. Okay.

Hello. I'm Julie Swain, cardiovascular surgeon advisor to the FDA. And prior to being in my present position, I was Chief of Cardiac Surgery at three institutions, Professor of Surgery and a senior investigator at NIH.

My career has really been spent in studying hypothermia especially relating to ventricular fibrillation thresholds and brain function.

So with that introduction we'll go over the data.

As Richard just said, we're looking at the data presented by the sponsor. And it's been a pleasure to work with the sponsor. They've been very responsive in sending us information.

We have clinical issues, though, with several of the pieces of data that have been submitted. The meta-analysis, specific data from Alsius, some safety data and a question about the clinical issue of applicability of the general date to the Alsius device.

The three studies, as you know, included the Bernard Austrian study. We have the public no patient-level data, HACA patient-level data, and the Belgium study no patient-level data. And the AKH Registry. I just wanted to go over a few minutes the comparisons between these.

Bernard study, Australia, HACA, Europe and the Brussels study. And I think there's always questions about applicability of out of U.S. data to U.S. data. And I think I learned from the Princess Diana story about the differences in delivery of health care. And I think that's one of the questions that the Panel just asked. So we need to know about the specifics of that.

Bernard study, they're all relatively small studies. Forty-three hypothermia patients out of 77 total. 137 in the HACA study of hypothermia. And 16 in the Belgium study. And the AKH Registry which contains a lot of patients with various protocols and various types of patients.

Inclusion criteria are somewhat different between all of these studies. Bernard was VF, not VT, VF. Men over 18 but women over 50. They're excluded if there are no ICU bed, cardiac arrest coma.

HACA study has the inclusion criteria listed here. And they actually looked at the denominator, which I think Dr. Somberg was asking about, and found that 8 percent of the patients screened without a -- cardiac arrest were eligible for these study with these inclusion criteria.

The Belgium study had the inclusion criteria as listed, hemodynamically stable, things of that sort.

AKH, because it's a registry and from '91 to 2003 we received one protocol from 1995 which was the HACA protocol which clearly was not used with this device, and a lot of the others in the external cooling devices I think that were mentioned. And before '95 wasn't used for the data from '91 to '95. So we're not sure of the inclusion criteria in the registry other than cardiac arrest patients.

And most importantly, this included in-house or out-of-hospital cardiac arrest patients. And the question is that in-hospital may in fact reduce time to resuscitation, although that's a very different patient group than out-of-hospital cardiac arrest.

The duration of cooling, 12 hours in one study, 24 hours, less than 4 hours in another. We don't know in the AKH Registry because there are various protocols used the exact conditions. How they were cooled, very much different between the three different studies that were published. Everything from a helmet to cool ice packs and all that.

They were started in the field in Australia. The cooling was started in the field in the ambulance. They're started in the hospital in HACA. In the hospital in the Belgium study. And I would assume in the hospital in the AKH Registry, although that is not clear to me.

Also rewarming, there was active rewarming in the Australian study, passive rewarming in the other two. I'm not sure about the rewarming methods in all of the patients in the AKH Registry. It's not data that we received.

Prespecified study endpoints. Survival to hospital discharge with neuro function allowing home or rehab for Australia. That's different from a CPC good or moderate at six months for the HACA study. There were really no prespecified endpoints for the Belgium Idrissi feasibility study. And since the AKH Registry was really a chart review from '91 to '94, apparently, and then a registry after that time, there are really no prespecified endpoints.

What's very interesting is something that actually I read the Bernard paper probably ten times over two years before I finally saw that it in small print in the statistical analysis section, it was a planned sample size of 62 and they completed the study. At the end of the study the primary endpoint was nonsignificant. A failed study. So they added more patients. And they added 12 hypothermia and three controlled. They didn't say they were added randomized, and I just had Dr. Irony, our statistician do a quick calculation. And I believe there's something like a 1.4 percent chance that this was a random occurrence. So I don't know how the 12 and 3 were added. And the paper then gives a p-value of .046 without talking about this interim analysis or adding patients, or anything of that sort. So we have a very difficult time interpreting that p-value. It could be interpreted that this really a hypothesis generating study. It was a negative study that's hypothesis generating. But there may be some question about that.

The HACA study we don't know how many patients were planned prospectively in this study. All it says it that the study was stopped prematurely because of low enrollment and the end of funding. And we don't know if an analysis was done then or what, when and how more patients or less patients got added or what. So it's very difficult to interpret how this was done. And there was sort of marginal significance. The data is not strongly positive. So that is, I think, a question that we have.

The Idrissi study, the Belgium study was an observational feasibility study that ended four hours. The official study ended four hours after the patient's admission.

AKH, again, we don't -- it's not a prospective study, so we really don't have for this '91 to 2003 an endpoint that was specified at that time.

Well, what about the results of these studies? So we have three different success endpoints. And when you look at the Bernard study, we don't know how to interpret that p-value because of the problems I've mentioned.

The HACA study if you look is a confidence interval of 1.08 to 1.81. So barely significant at that view.

The Idrissi study, since there were no endpoints, we don't know success.

When we look at mortality, and that's been mentioned here, we have in the Bernard study a nonsignificant p-value for mortality.

The HACA study it was a confidence limit to up to .95. Again, close. So there's a marginal benefit of mortality there.

The Idrissi study had an 87 percent mortality in-hospital mortality with both groups.

The issues we have with the registry is that it's kind of retrospective prospective. I believe medical students abstracted the data in parts of it. Observational. It's a chart review versus a protocol, versus other protocols that we don't know about. And then the possibility of patient selection or treatment bias. And I think Dr. Sterz mentioned that as well.

It's an institution with Dr. Sterz who is a ardent proponent of hypothermia. And the standard of care at the institution was hypothermia, but some patients didn't get hypothermia. And we don't know why they didn't. We know a certain group didn't. It was a group that needed cardiac or neurological imagining, and that's a very different patient population from people who come with cardiac arrest who don't need cardiac or neuroimaging. And one propensity analysis item that I wanted was who the attending physician was, but we don't have that data collected. And as Dr. Sterz was saying, that maybe the fellow that was working at night and may be a far different experience and abilities than Dr. Sterz, didn't do hypothermia. Therefore, that's a control.

So there's a lot of question about how comparable the control groups are. Dr. Zhao will be talking about that.

We also have historical patients early 1990s didn't get hypothermia at all, and they're included.

So we have unknown covariates, known covariates and unknown covariates that we cannot compare.

One of the very most important issues is that patients who died within 24 hours were excluded. And having done work in looking at ventricular fibrillation from a threshold and hypothermia, that's a group I would be very interested in seeing. But all of those patients were excluded from the registry.

For safety, they compared the AKH Registry device with core cooling treatment group versus those that didn't get hypothermia. Again, a comparison that may be problematic. And compared the AKH Registry core cooling with the HACA trial surface cooling from data at one institution.

It's important to mention that none of these trials were powered to -- and many of the meta-analysis trials were powered to detect differences in individual adverse events. So therefore saying something is not significant is that we don't have the power to really determine that.

And there were no prespecified safety endpoints in any of these studies.

The HACA study looked at complications only during the first seven days. And I think that goes to Dr. Marler's comment about pulmonary embolism and DVT and things of that sort.

So there were no prespecified endpoints. The end total was 137 for hypothermia and 138 in the control. And you can see that there is some difference, again not significant, in each of these complications; bleeding, pneumonia, sepsis, pulmonary edema and arrythmia. Again, in the first seven days.

Three patients had hypothermia stopped because of arrhythmias or hemodynamic instability.

The Bernard study looked at only at complications at hospital discharge, and 77 total. There were no specified endpoints. There's a statement no significant hemorrhagic complications, but there's no other mention of safety events. But there was -- excuse me. The wrong slides.

There wasn't an 87 percent mortality. I believe there was like 30 percent mortality in that group. Sorry.

The next in the Hachimi-Idrissi study from Belgium. That one had an 87 percent mortality. Again, no prespecified safety endpoints.

They mentioned that they monitored the various complications, and then they say no other significant complications occurred in any patients. But 87 percent died in the hospital. And I think they must have meant in the four hours that the study was being conducted.

So these studies are not helpful for us in determining the safety of hypothermia.

AKH Registry, arrhythmias and bleeding only had a follow up up to 32 hours. Pneumonia, sepsis, renal failure, pancreatitis up to seven days. So we really don't have long term data. And again when you're putting a device in thermal vein or instrumenting any parts of the vascular system, I'm unfamiliar with studies that have no complications, although we have no complications in this registry listed after the seven days or studied after the seven days.

And then when you look at the AKH Registry of noncardiac adverse events given, I think, the slide that the sponsor just showed from the first submission, we see differences but again nothing is powered as an individual safety endpoint for any of these complications. And the question is the definitions of renal failure and things of that sort. So it's difficult to understand these complications or to be reassured by a lack of complications.

Well, what else do we have? We've got a couple of studies, two other studies I'll mention today that are actually well conducted prospective randomized controlled study looking at cooling. And one I think Dr. Becker just mentioned, data presented by Bill O'Neill at TCT in 2003 looking at left ventricular infarct size and the effect of po st-event cooling on that. And that was a negative study with no difference between the LB infarct size endpoints. So a hypothesis generating study.

And when you look at the safety results, again, it was not powered to look at individual safety complications. But you can see that there are differences in the safety profiles of the two ways. Again, no significant difference. It wasn't powered to detect a significance difference.

We look at Guy Clifton's study of traumatic brain injury published in the New England Journal in 2001. And their endpoint was percent of patients with poor outcome by their definition. Very sophisticated neurological tests for these patients again found no difference in the primary endpoint and negative study.

What was interesting even though the study was not powered to detect differences in individual events, we did find in this study a significant difference in the amount of hypotension, bradycardia with hypotension. And hypotension is significant hypotension with end organ dysfunction. Bradycardia with hypotension and percent of hospital days with complications were all significant.

So what we have is a body of data submitted to us that shows us that there is actually burden of proof on the sponsor to show us that the method of cooling and all of these do not effect outcomes either by efficacy or safety.

And in summary, we have questions remaining regarding the safety and effectiveness of surface cooling for the proposed indication. It's unclear whether data on surface cooling can be used to support the safety and effectiveness of endovascular cooling for the proposed indication. We think there are limitations in the AKH Registry data. And the risk benefit profile of the Alsius device for the proposed indication is unclear.

Thank you.

DR. ZHAO: My name is Yihua Zhao. I was the statistical reviewer for this submission.

I will present our statistical review for this submission now.

During my presentation I will first give an overview of the examination. I will talk about the two study objectives and the data sources, as well as sponsor's statistical analyses used in this submission.

Following that I will review the sponsor's meta-analysis and a propensity score analysis And after that I will give a brief comments on the safety analysis. And will conclude my presentation with our statistical summary.

There are two study objectives. The first one is to assess the effectiveness of mild therapeutic hypothermia delivered via any methods in improving survival and a neurological recovery in primary survivors of cardiac arrest.

The second objective is to assess the effectiveness and safety of mild endovascular hypothermia delivered by Alsius CoolGard system in improving survival and a neurological recovery in primary survivors of cardiac arrest.

The main datasets used in the submission is the AKH Cardiac Arrest Registry. This was an observational study conducted in the emergency department in Vienna General Hospital in Austria. The data used in their analysis was collected between 1991 and 2003. There were 67 patients receiving hypothermia via CoolGard system. One patient died within 24 hours and four patients had missing covariates.

There were 1536 normothermia patients, among them 205 patients died within 24 hours and 140 patients had missing covariates.

In the sponsor's analysis patients who died within 24 hours were excluded and patients who had missing covariates were also excluded.

Additional data sources used in the submission come from three published studies. The first one is the HACA study conducted between 1996 and 2001. It was a randomized controlled trial contained in nine centers in five European countries. There were 137 hypothermia patients received hypothermia through cool air and there were 138 normothermia patients in this study.

The second study is the Bernard study conducted between 1996 to 1999. It was also a randomized controlled trial conducted in four centers in Melbourne, Australia. There were 43 hypothermia patients and 34 normothermia patients in this study.

The third study is the Idrissi study conducted over six months in 2000 and 2001. It was randomized controlled feasibility study conducted in a single center in Belgium. There were 15 hypothermia patients and fourteen normothermia patients.

Two assess of effectiveness of mild therapeutic hypothermia in improving survival and neurological recovery the sponsor performed a meta-analysis using aggregated data from HACA Bernard and Idrissi studies. And they used the fixed effects model to combine point estimates from the three randomized controlled trials.

To assess the safety and effectiveness of mild endovascular therapeutic hypothermia applied via the CoolGard system in improving survival and neurological recovery, the sponsor used the AKH Cardiac Arrest Registry study. Because data from this registry did not come from a randomized controlled study, the sponsor conducted a propensity score analysis to reduce spires due to imbalance in observed baseline characteristics.

Now I will review sponsor's meta- analysis.

The three studies used meta-analysis have different survival and neurological endpoint.

In HACA study, as pointed out by Dr. Swain, in HACA study the survival endpoint was the overall mortality at six months and the good neurological recovery was defined using Pittsburgh Cerebral-Performance Category of 1 or 2.

In the Bernard study the survival endpoint was survival to hospital discharge. And the good neurological recovery was defined as having sufficiently good neurological function to be sent home or to a rehabilitation facility.

In the Idrissi study the survival endpoint was survival to the end of study, which is four hours after applying hypothermia. And the good neurological recovery was not explicitly described in the Idrissi study.

The three studies also have different patient cohorts because they use different inclusion/exclusion criteria and they were conducted at different locations.

The three studies also used different surface cooling methods. In HACA study patients who are assigned to the hypothermia group were put in a mattress with a cover that delivers cold air over the entire body. If the target temperature was not reached, patients will be applied ice pack around their head and heck.

In the Bernard study patients receive ice packs around head, neck, pulse and limbs.

In Idrissi study hypothermia patients receiving cooling helmet around their head and neck.

Those three studies used different target temperatures. In HACA study cooling was initiated in the hospital and the target temperature was 32 to 34 degrees celsius. And the cooling was sustained for 24 hours followed by passive rewarming.

In the Bernard study cooling was initiated in ambulance and the target temperature was 33 celsius for 12 hours followed by active rewarming.

And in Idrissi study cooling was initiated in hospital. The target temperature was 34 celsius for four hours followed by passive rewarming.

The issues with the sponsor's meta-analysis include the following:

Combining the three randomized controlled trials is questionable due to:

Different inclusion/exclusion criteria;

Different times, places for initiating cooling;

Different target cooling temperatures;

Different cooling methods, and different cooling durations;

Different rewarming techniques, and;

Different neurological endpoints employed in these three studies.

There were no CoolGard or endovascular cooling patients in any of the three studies.

Recently the sponsor proposed retrospective Bayesian meta-analysis in which a sequential updating approach was used. This approach is equivalent to pulling old data from the three randomized controlled approach. The proposed approach can only be used when patients in the three studies are exchangeable; in other words when the patients in the three studies can be clinically considered as coming from the same population. And, again, exchangeability remains questionable due to the differences among the three studies.

Even if the three studies were similar, the agency would recommend the following steps whenever a Bayesian analysis is performed.

The first use of prior information should be agreed upon in advance by sponsor and FDA.

Second, comparability among studies should be agreed upon in advance by clinicians.

And third, to guard against unforeseen differences among studies, Bayesian hierarchical models are recommended.

And lastly, simulations are strongly recommended in advance to assess operational characteristics of study design and to control Type 1 error.

Now I will review the statistical inference derived from AKH Registry study.

In the effective analysis, the sponsor considered the following true endpoints. The first one is survival to 30 days or discharge.

The second one is survival to 30 days or discharge and good neurological recovery, which was defined as CPC equals 1 or 2.

In the effective analysis the sponsor did not explicitly explain how to use these two effectiveness endpoints to define study success. And also patients who died with 24 hours and patients having missing covariates were excluded from the effectiveness analysis.

So there was CoolGard patients and 1191 normothermia patients in their analysis.

In the safety analysis the sponsor measured arrhythmias and the bleeding up to 32 hours and also measured the pancreatitis, pneumonia, sepsis and rental failure up to seven days.

Because the AKH Registry study was not a randomized controlled trial, the baseline characteristics between CoolGard and the normothermia groups are very different.

The following two slides will show the baseline covariates that are significantly different between the CoolGard and normothermia groups. From this slide we can see that CoolGard patients were younger and had a longer duration of cardiac arrest than the normothermia patients.

And this slide shows that CoolGard had less often history of diabetes. Milder NYHA Score on admission. Had more comatose patients on admission. Were more often out-of-hospital cardiac arrest of presumed cardiac cause and had more often VT or VF and received more often basic life support than the normothermia patients.

This bar plot shows the patient frequency by year in the AKH Registry between 1991 and 2003. And we can see that CoolGard patients were only present in the year 2002 and 2003.

So CoolGard and the normothermia groups are different and may not be comparable at baseline because there are CoolGard patients only in 2002 and 2003. And there is imbalance in several observed baseline covariates between the two groups. As a result, any direct treatment comparisons on effectiveness endpoint are inappropriate.

We may not know the extent of imbalance between the two groups regarding variables that were not measured and may be important in predicting treatment allocation and outcomes. Such variables may include cardiac neuro/imagining, attending physicians, etcetera.

And the treatment comparison may be improved via propensity score analysis.

Now I will briefly talk about -- introduce the propensity score analysis.

Propensity score analysis provides approximate balance in observed baseline covariates. And all important covariates should be considered in the propensity score model.

Propensity score analysis cannot adjust for unobserved covariates which could be important in predicting treatment assignment and outcomes. If there are important baseline covariates which influence the treatment assignment but were not collected, propensity score analysis can be seriously degraded.

Due to limitations of observational studies, randomized controlled trials are strongly encourage when they are possible, because randomized controlled trials can balance both observed and unobserved imbalance between the two treatment groups.

When propensity scores are balanced across the CoolGard and the normothermia groups the distributions of all observed covariates which are included in the propensity score model are balanced in expectation across the two groups. As a result propensity scores can be used as a diagnostic tool to measure comparability between the CoolGard and the normothermia groups.

If the two treatment groups overlap well enough in terms of the propensity scores, we can compare them adjusting for propensity scores.

In the sponsor's propensity score analysis they include the following covariates in the propensity score model, and this is also what is included in their propensity score model. And this slide shows the covariates that were collected but were not included in their propensity score model.

Note that the following covariates were not collected and, therefore, were not included in the propensity score model used by the sponsor: Cardiac neuro/imaging, treating team physician and the reason for not receiving hypothermia in the hospital where hypothermia was standard of care. And there's no way to know if there is still imbalance between the two groups regarding these covariates.

The following table shows the distribution of patients in propensity score strata. Patients in the same stratum have similar propensity scores and we can see that in the first stratum there are 251 normothermia patients. But there are no CoolGard patients have similar propensity score in this strata.

So after we got the propensity scores, how could we use them in our analysis? We can match the control patients to the CoolGard patients using the propensity score. We can also perform stratification analysis using the propensity score strata. And we can also use regression adjustment by including propensity score as the continuous covariates in the regression model.

The sponsor used matching and their method is not clear to us.

What we used is first excluded patients in the first propensity score strata and then we performed the stratification analysis by doing separate analysis within each strata and then combining the results across the strata.

And we also did a logistic regression adjustment by including propensity score as the continuous covariates in the logistic regression model.

Also included in this logistic regression model are all covariates which were used in the propensity score model.

Here is our result from the stratification analysis. I used odds ratio in our analysis. If odds ratio is greater than one, then CoolGard system is better than normothermia regarding the specific effectiveness endpoint. Because the sponsor did not explicitly specify the success criteria, we used Bonferroni correction and we reported 97.5 confidence interval here. If the lower limit of the 97.5 confidence interval of the odds ratio is greater than one, then CoolGard system is statistically significantly better than normothermia regarding the specific effectiveness endpoint. And if the 97.5 percent confidence interval including one, then CoolGard patients is not statistically different between normothermia patient.

And for survival to 30 days or discharge, the odds ratio between CoolGard and the normothermia is 1.69. And the 97.5 percent confidence interval is .85 to 3.49. And the lower limits, this confidence interval include one. So CoolGard is not significantly different between normothermia regarding survival.

For survival and a good neurological recovery, the odds ratio is 2.56. And again, the corresponding 97.5 confidence interval does not include one, so the CoolGard is significantly better than normothermia in survival and the good neurological recovery.

And this is our result from regression adjustment. And regarding survival endpoint, the odds ratio is 1.73 and again, the 07.5 confidence interval does include one. So the CoolGard is not significantly different between the normothermia.

And the for the survival and a good neurological recovery endpoint, the odds ratio is 3.02. And there 97.5 percent confidence interval does not include one, so CoolGard is significantly better than normothermia regarding survival and good neurological recovery.

So because date of cardiac arrest may be an important variable that is not included in the propensity score analysis, and there were only CoolGard patients in 2002 and 2002, we performed an analysis considering only patients treated in 2002 and 2003. And here is the distribution of patients in the propensity score strata if we only using patients enrolled in 2002 and 2003. And here in the first two strata there were no CoolGard patients having similar propensity scores as the normothermia patients in the same strata. And I want to point out in that in the last strata there were more CoolGard patients than the normothermia patients have similar propensity scores.

So what we did is we also excluded the patients in the first and second strata and we performed the stratification analysis and the logistic regression adjustment. And this is our result from stratification analysis.

For survival to 30 days and a discharge, the odds ratio is 1.14. And the 97.5 percent confidence interval does include one, which suggests that CoolGard is not significantly better than normothermia.

For survival and a good neurological recovery the odds ratio is 2.91. And the 97.5 confidence interval also include one, so the CoolGard patient is not significantly different than the normothermia patient.

And this is our result from the regression adjustment. For the survival endpoint the odds ratio is .89 and the 97.5 percent confidence interval does include does include one. So the CoolGard is not significantly different than normothermia.

And for survival and a good neurological recovery the odds ratio is 3.98. And this time the 97.5 confidence interval is barely above one. So we can see the CoolGard device is marginally significantly different -- better than the normothermia patients.

In the first safety analysis the sponsor used 66 patients in the AKH Registry study. And they selected 81 normothermia patients from the AKH Registry to match the inclusion/exclusion criteria.

I want to point out that in this safety analysis normothermia patients were selected differently from the propensity score analysis. They were not selected based on the propensity score, but they were selected to match the inclusion and exclusion criteria.

And in the second safety analysis the sponsor used the 66 CoolGard patients in the AKH Registry and they used 134 hypothermia patients treated with cool air in the HACA study.

And also I want to point out that patients in these groups may not be comparable at baseline.

So in summary, combing HACA, Bernard and Idrissi studies in the meta-analysis may be questionable due to the several differences among the three studies.

Not all important covariates were included in the propensity score analysis. And the propensity score analysis is just an effort to reduce bias due to non-randomization, but it cannot balance two treatment groups on uncollected covariates.

Using the current propensity score model, only survival to 30 days and a good neurological recovery is marginally significantly different between the CoolGard and normothermia groups. And normothermia patients in the safety analysis are selected differently from those in the propensity score analysis.

Thank you.

DR. LAZAR: My name is Ronald Lazar. I'm a Professor of Clinical Neuropsychology in Department of Neurology and Neurological Surgery at Columbia University College of Physicians and Surgeons, and the Director of Levine Cerebral Localization Laboratory.

I've been working in clinical trials for almost 20 years involving physical neurological and neurocognitive outcomes. And most recently involved in mechanical circulatory devices for end stage heart failure. And I served on this Panel a couple of years ago.

What I'm going to address this morning really comes to the notion of what constitutes a good neurological outcome. And some of the material I'm going to talk about this morning was presented to the Panel back in September, and I'm also going to give some additional considerations.

As most know, the effects of postanoxic encephalopathy following cardiac arrest follow a continuum that ranges anywhere from mild to severe. So that at the lighter end of the spectrum patients who have brief periods of anoxic from arrest can have inattentiveness, weakening of judgment and motor incoordination. Sometimes the mild deficits are transient, and sometimes they're permanent.

At a great degree of severity you have memory impairment, apathy, disinhibition and poor judgment.

And at the most severe levels of encephalopathy short of coma, at the physical level there's spasticity, paresis, ataxia, pseudobulbar palsy. And from a mental status point of view there's language disorders, disorders of orientation in space, purposeful movement of motor function, memory disorders and so forth.

As was pointed out earlier, probably the most commonly use outcome instruments to take a look at the brain integrity following cardiac arrest is the Cerebral Performance Categories, which actually got its origin from Jennett and Bond's work in head injury in Scotland back in 1975, then was used in cardiac arrest beginning in the 1980s. And as outlined earlier, there are five levels of function.

There's good cerebral performance where the patient is essentially conscious with mild deficits and presumably can lead a normal life. They might have some minor neurological problems.

At a more severe level there's moderate cerebral performance where the patient is also conscious but presumed capable of part time work in a sheltered environment or independent activities of daily living and may have neurological impairment including hemiplegia, dystaxia, dysphagia, dysarthria, seizures or permanent memory impairment. And I'll come back to that a little later on.

And then there is severe cerebral performance where patients are conscious but presumably are dependent on others for daily support because of impaired brain function.

Then there's coma and finally death.

So let's take a look at out-of-hospital cardiac arrest. And these are the studies that have been cited this morning as support for the sponsor for this device. And we're going to use these three studies as a frame of reference, especially the Bernard study and the Holzer meta-analysis, because for them a good neurological recovery is defined as a CPC score of 1, good cerebral performance and a CPC score of 2, which is moderate cerebral performance. And I'll come back to these in just a minute.

Now I think one of the best studies ever published in the field about neurocognitive outcomes following cardiac arrest is Dr. Roine's study from 1993 that appeared in JAMA. And what they did was they did an analysis in a placebo controlled study of nimodipine versus placebo for cardiac arrest. And so out of 677 possible patients that could have been studied, they were able to follow 68. And what they did as major outcome measures, they looked at neurocognitive scores at 3 and at 12 months.

And what they found was that at 12 months 48 percent of the patients of the 68 had moderate to serious deficits in the areas of delayed recall, manual dexterity, calculations, skilled motor movement, planning, attention and motivation. And I think equally significantly about half the patients had significant depression. And I think this goes to one of the points that was stressed in the sponsor's material submitted for our review this morning, and that is it was felt that these patients are glad merely to be alive. And, in fact, these data indicate that half of the patients don't necessarily agree with that point.

So let's take a look at good cerebral performance. Again, this is the conscious patients which are presumed able to have the best kind of life afterwards. And what's important to point out, Dr. Roine mentioned that there were 52 studies in the literature in which a CPC was used, but in fact I could only find one study that was a validation study about the CPC. And that was Hsu published in the Annals of Emergency Medicine in 1996.

And what they noted was that in their view the CPC was subjective and categories poorly defined, and frequently used only at hospital discharge. Although we know since 1996 it's been used for 30 day and 60 day and longer term outcomes. But most importantly, it had never been validated or compared with any other measures in the past, which is not true in the case of the scales used in stroke, such as the modified Rankin scale where there's a lot of validity data. And so that's what Dr. Hsu did.

And what she found was in essence that if you compare the outcome just for CPC scores of 1, the best presumed neurological recovery, in fact there was very little specificity of the CPC compared to the Functional Status Questionnaire which is a commonly used standardized instrument of function in a variety of emergency medical situations. And in fact, the correlation of CPC at discharge and at follow-up, was only 3.2 which is not significant. So other than Hsu's data, which was essentially a negative study, there are no validation studies for the CPC, even at the best level of function.

But what about a poorer level of function? Let's take a look at the moderate cerebral performance. And here patients are presumed to be capable of part time work and sheltered environment or independent activities of daily living.

And if you look at Holzer's paper he stated favorable long term neurological outcome was defined as good neurologic recovery and being alive at six months ...sufficient cerebral function for activities of daily living would include dressing, travel by public transportation, food preparation and so forth.

And if you just take a look at really the physical specification of a moderate cerebral outcome, patients who are hemiplegic, that is they do not have use of one side of their body, cannot dress themselves. They cannot take public transportation. And they cannot prepare their own food.

Well, what about at a cognitive level of function? Well, if you look at Dr. Roine's data, 48 percent of his patients at 12 months who had moderate to severe impairment would be eligible for Social Security disability in the United States with a CPC score of 2.

Well, the argument is well the sponsor said well these patients are older anyway. You know, they're not going to go back to work because these are presumed patients who are beyond that point. But, in fact, if you look at the Public Access Defibrillator Trial which was published in the New England Journal last year, one-third of those treated by EMS were in fact lower than 65 years of age. And in the CoolGard presentation today their average age is 51. So these people are capable of work and would be disabled with a CPC score of 2.

So, is a more standardized evaluation available? Can we actually do something that is a little bit more extensive with regard to neurological function.

Well, let's take a look at this from a number of different dimensions. First, from the viewpoint of the disease, that is out-of-hospital cardiac arrest we can point to Lim's work published last year in the Journal of Neurology. From the viewpoint of treatment, hypothermia for brain injury, which Dr. Swain talked about before, there was the Clifton study published a few years ago in the New England Journal.

And if you look at the patients of comparable age and symptom severity, we could take a look at the patients with end stage heart failure who have to go on permanent circulatory device support, and we could take a look at the HeartMate trial, the MicroMed trials currently underway. And what all these trials have in common, all these studies is that they all evaluated memory, mental speed, decision making, language perception and motor function.

So it is possible to do something more standardized. And I'm not necessarily advocating that we do all these tests, but there are tests available and they can be done. And, in fact, if in the Roine study if they had correlated the outcome on CPC with his outcomes, we would have had validation of that measure. But, unfortunately to my knowledge, those data don't exist.

So based on the existing evidence that we have, the physiologic impact of cerebral anoxia following cardiac arrest is well documented with effects that can be transient or permanent, mild to severe. And that the field is already moving toward objective validated measure of brain functions that include physical and cognitive outcomes.

The measurement of brain function in a clinical trial should be performed by clinical neuroscience specialists who are blinded to treatment or at least not involved in the treatment to minimize potential bias.

And finally, the neural endpoints need to be obtained in the acute period, at discharge and at longer term follow-up in order to assure patient meaningful outcomes.

Thank you.

CHAIRMAN MAISEL: Thank you.

Do any members of the Panel any questions they'd like to address to the FDA? Dr. Brott?

DR. BROTT: Thank you, Dr. Lazar

Could you put Slide 7 up?

I was interpreting (2) as independent activities of daily life. And I noticed that it's not uncommon for neurologists, even, to sometimes be loose with the term "hemiplegia." Is that parenthetical statement in all forms of this examination? Because hemiplegic contradicts independent activities of daily life. Somebody who is hemiplegic cannot do that. And so there's an internal contradiction. And I'm wondering if it's a semitic problem?

DR. LAZAR: Well, I can't answer that since these are the published criteria for the CPC.

DR. BROTT: So that parenthesis is always there in a written form?

DR. LAZAR: That is correct.

DR. BROTT: Okay. Okay.

DR. LAZAR: These are the quotations from that literature.

DR. BROTT: That answers that question.

I guess I'd ask Dr. Sterz --

CHAIRMAN MAISEL: Why don't we -- okay. Is that all right?

DR. BROTT: It has to do with this.

Were patients in category 2 both hemiplegic and judged to be independent in activities of daily living in the trial that you were in, the HACA trial?

DR. STERZ: Straight answer, yes. This was -- the criteria for being 2 or 1. However, I have learned the Ulstein style by heart since he is now. I don't remember that his practice are there. What is written in the practice I don't remember that it is written there. It's a very clear stated in the Ulstein style that they have to be independent activities of daily life. I don't recall this. I have to throw the Ulstein style again. Sorry.

CHAIRMAN MAISEL: Any additional questions for the FDA? Yes, Jeff?

DR. BRINKER: Julie, your discussion puts us in a little bit of a quandary because we have to make two decisions. One decision I thought up front would be easy. There are published guidelines suggesting that hypothermia in fact should be and actually in publication is the standard of care for treating patients with cardiac arrest caused by ventricular fibrillation. And you're suggesting that perhaps this is wishful thinking based on a few less than optimal studies.

So could you be perhaps more specific in saying that you reject the recommendations by AHA in this regard?

DR. SWAIN: Well, talking from the FDA standpoint today, you know we evaluate data. Not what someone thinks about data. And you and I, and most of the members here, have been in society guidelines creating and government guidelines creating jobs. I've done that for the unstable angina guidelines for health care, and various other society guidelines. And there are various considerations taken in guidelines. I've been an associate editor of several journals and on the editorial board of a whole lot. And there's publications. We make mistakes sometimes. And there are other reasons we may publish something.

You know, we're just looking at the data here. And when I look at the data, and I must say after the last Panel meeting two Panel members came up to me and said, gee, I didn't see that stuff about the Australian trial being a failed trial and then adding more patients. I didn't see it until I was on the airplane last time. I had read that thing for two years.

So when I look at the level of evidence that we normally require, especially in the cardiovascular division, those studies don't match the level of evidence both the first, which was a true failed trial until patients were added and then we don't know why it was stopped. And the second one that didn't meet the enrollment endpoint and we don't know why. There's also publication bias, things of that.

So simply looking at the data there is not, I would say, two good randomized trials showing the efficacy of post-event hypothermia. My personal opinion.

DR. BRINKER: That's why you, therefore, personally think there would be no ethical qualms about doing a randomized controlled trial against normothermia?

DR. SWAIN: No.

CHAIRMAN MAISEL: John?

DR. SOMBERG: I'm not a statistician, which everyone probably realizes. I was very much helped by your analysis. And I just wanted to go through my understanding, if it's correct, that the reason you required a higher level of confidence, 92. something percent, which is very high; that was necessitated by using the Bonferroni. And that was necessitated by the lack of comparability of the different groups? Am I correct in that?

DR. ZHAO: Using a wider confidence interval will be more conservative because the sponsor used the two effective endpoints and we don't know how they defined the study success if, like, both should be like significant or either one should be significant if it can be defined as success. And so we choose the more conservative way and we reported a wider confidence interval.

DR. SOMBERG: Okay. So you say you choose the more conservative way, but for I guess a layperson, which I am here, is what does that mean less conservative, more conservative? I think if we're talking about probabilities of success, can you put that into some sort of number of knowing what you're doing?

DR. ZHAO: More conservative means that if, like if we use the 95 percent confidency interval and it's marginally significant and now we're using a wider confidency interval and it's no longer significant. So we will just -- by using conservative means that we won't approve a device that is falsely significantly better than the normothermia patients.

DR. SOMBERG: What I'm trying to get at it is we usually take a nominal p-value of, you know, we're willing to take a 1 in 20 mistake or something. Why are you changing the levels of acceptance of mistakes?

DR. ZHAO: The p-value is another side of the story, which if you go to the 95 percent confidence interval, which does not include one, then the p-value is less than one. Then the p-value less than .05. And here just simply reporting p-value it doesn't mean anything. Because, like, you have the point estimate and you have the p-value. And you don't know what's the range of these estimate, like how precisely estimate.

So I think by providing confidence interval you will have more information with the possible range of estimate and with the precision of this estimate. And which we choose cost precision because we don't know the study success criteria.

MS. IRONY: Hi. I'm Telba Irony. I'm the Branch Chief of General and Surgical Devices Branch, Division of Biostatistics.

I just wanted to add another bit of information of why we did that correction.

You wanted to connect that with the possibility of making a mistake and controlling to make it up to five percent.

When we received the submission, we didn't know what the endpoint was. If it was only survival or survival plus neurological recovery. And that might be if you get involved in points, median points, some will be significant by chance. So because we had these two and we didn't know if the success of the trial will be both are successful, either one are successful or maybe a third one that was not successful and was not even presented there. In order to control the Type 1 error and make an overall type error at maximal at five percent we had to make this correction.

I don't know if that -- it has nothing to do with the variables.

Does it help?

DR. SWAIN: Yes. And if you look at the ultimate, as mentioned, safety as mortality, then there really are two endpoints. And so one does a correction normally for multiple endpoints.

CHAIRMAN MAISEL: Henry?

DR. HALPERIN: Julie, it's a very nice presentation.

I had a question about the Clifton study because I didn't see it in the packet, and maybe you could clarify.

What was the patient population there? Were they patients with strokes or --

DR. SWAIN: No. It was traumatic brain injury, as I put on a slide. It's a TBI study that was, I believe, multiple institutional.

DR. HALPERIN: But they were never hypotensive.

DR. SWAIN: Hypotensive? I don't know, you know. As I said, it has different amounts of cooling, different types of cooling. It's just yet another post-event cooling study actually well done with the endpoints that Ron Lazar --

DR. HALPERIN: But in that particular brain injury is there adequate cerebral profusion, typically?

DR. SWAIN: You could probably answer that.

DR. LAZAR: I can't remember the details of the study per se, typically there is. And we're really talking about a brain concussive event that puts people into the hospital and need to be treated. And they try to use hypothermia in these cases to reduce the amount of swelling and so forth in the brain. So it's not a profusion area defect, it's rather more of a global, you know, a trauma to the brain as a whole.

But if there was a cardiac problem, then I don't know the answer to that associated with the head trauma, per se.

DR. MARLER: The whole field of head injury, many of the treatments are focused on the fact that a lot of the injury is due to alternation and profusion of the injured portion of the brain. And a lot of the treatment is directed toward a -- a lot of the neuroprotective treatment is the same as that, say, for stroke where there is decreased profusion.

DR. LAZAR: But it's profusion within the brain itself as opposed to a source, let's say, from the heart.

DR. MARLER: Right. But in either case, the target we're talking about is the brain function, correct?

DR. LAZAR: Yes.

DR. HALPERIN: Because, I mean the issue would be how applicable is that to a post-cardiac arrest situation where theoretically, at least -- I mean I don't know about this delayed reprofusion issues, but at least the artery should be open and there should be the possibility of having adequate profusion. Where if there were some situation where you were comparing whether there was actual ischemia ongoing, brain ischemia, then that actually might not be a comparable kind of a situation to extrapolate the data to.

DR. SWAIN: Yes. Same problem of, you know, heterogeneity of patient population like an in-and-out hospital. Cardiac arrest patients are also heterogeneous.

DR. SOMBERG: Can I ask a follow-up on that?

CHAIRMAN MAISEL: Sure.

DR. SOMBERG: Just be specific, so a concussed injury and on injury do to hypoperfusion are we saying that is a different pathophysiology or a similar in terms of -- and you can sort of say that an intervention would be similarly effecting that and maybe it's the variance of the populations, or is it totally different or we don't know?

DR. LAZAR: That's a complicated question. And there's probably some overlap, but fundamentally they're different pathophysiologies there.

CHAIRMAN MAISEL: Dr. Brott, did you want to comment?

DR. BROTT: I think that both basic and clinical neurologists and neurosurgeons would agree that this is apples and oranges.

CHAIRMAN MAISEL: Rick?

DR. PAGE: Dr. Zhao, thank you for your explanation. I just want to make sure I got something right in terms of slide 34 where you commented there was marginally significant improvement in survival to 30 days or discharge and good neurological recovery every using a confidence interval of 97.5 percent, it's marginally significant as I interpret it and part related to relatively low numbers. But the actual effect, there is a fourfold increase?

DR. ZHAO: Yes.

DR. PAGE: And the result that we would be hoping for, which is indeed both discharge and what's considered good neurologic activity, is that correct?

DR. ZHAO: The point estimate, the odds ratio is four -- almost four.

DR. PAGE: So the thing that's marginal is the statistical test, but it did meet your level of statistical significance. But the effect is actually a fourfold increase in survival with what's considered good neurologic function?

DR. ZHAO: Yes. Yes.

DR. MARLER: What would the absolute numbers be on that?

DR. ZHAO: What do you mean absolute numbers?

DR. MARLER: You gave us a ratio of 2 percentages, but are they low percentages?

DR. ZHAO: I cannot say from the top of my head. Sorry.

DR. MARLER: Okay.

DR. ZHAO: But I think the odds ratio, it's more informative here.

DR. MARLER: But not if it's one patient in one group and four in the other?

DR. ZHAO: I cannot say. Sorry.

DR. SWAIN: The question is the groups. And, again, the questions we have about comparability of patients who got hypothermia in a hospital were normally -- that's the treatment of care and who didn't get it. So it's not just a statistical point that Mary made. It's also is the control group comparable for the reasons she listed.

CHAIRMAN MAISEL: Any other questions for the FDA?

Okay. I'd like to thank the FDA for their presentation.

And at this point I would like to open the discussion.

I think at this point since we have already covered a lot of the topics, why don't we move on to our primary reviews. And I'll start with Dr. Somberg, if you would please do your --

DR. SOMBERG: Well, thank you, Mr. Chairman.

I was asked to review this material and submit, which I will for the record, a written review. I will read parts of my written review of the material. I might also say that I reviewed the first material approximately two weeks ago. And approximately six days ago received the supplemental packet and reviewed that as well. So there are two basic material packets that are in review here.

As background, I would like to say that a Circulatory System Device Panel meeting was held on September 21, 2004 to discuss trial design issues for cardiac arrest studies as well as the role of hypothermia as a treatment for post-arrest. Dr. Swann presented a good review of the studies today. The MI study found no difference in outcome with or without cooling and was a perspective randomized controlled study comparing cooling and normothermia, but in an MI situation.

Clifton, et.al. performed a prospective randomized controlled trial of normothermia to hypothermia in patients following brain injury and also found no benefit.

Dr. Swann pointed out two studies of hypothermia following cardiac arrest, Bernard which is the Australian study and an European study. Both studies were performed outside the U.S. Their response times are average or two or three minutes different from the U.S. system.

The summary of the problems with the two studies is to be found in the submission pack tab 2.

The Panel on September 21, 2004 discussed the topic extensively and concluded that the data are interesting and promising, but would not support an application for a particular device or a therapeutic approach. The Panel did not feel that the standard of care was hypothermia in the U.S. and that the studies were small, different from the types of patients seen in the U.S. and used different types of cooling approaches.

"The Panel went on to discuss if surface-induced hypothermia is the same as core-induced hypothermia. The Panel felt that endovascular technique raised a different set of safety issues from surface-induced hypothermia." And I'm quoting this from the Panel summary.

Many felt that a randomized controlled trial would be needed to evaluate an intervention.

The proposal that the sponsor's requesting approval for the endovascular cooling system based on the similarity of device to a predicate device and the assertion on page 11 of the sponsor's material in the briefing package of the benefits of hypothermia in treatment of comatose survivors of sudden cardiac arrest due to ventricular fibrillation have been established in randomized controlled trials.

My review. One. I do not believe that a predicate device exists. The initial device was a heat exchange blanket. A catheter was not been used for the indication requested, and does not serve as a predicate. A thermal blanket has not been approved for post-arrest ventricular fibrillation patients.

Two. As reported in the September '04 Panel meeting a consensus did not exist for the acceptance of hypothermia as the treatment of choice for post-VF arrest patients. The data was felt very promising, but insufficient to establish hypothermia as the standard of care in the U.S. or determine that endovascular catheter is the appropriate way to establish hypothermia. In fact, reviewing all hypothermia studies, one finds that the Cool MI study which was controlled and randomized found a higher mortality hypothermia group. The Clifton study found no improvement with hypothermia. An additionally, in the Hackett study, the Bernard study and the Australian study, none of the patients were treated with the CoolGard system, the system we're reviewing today.

Three. Even without the lack of predicate device and hypothermia controversy, an evaluation of potential toxicity endovascular cooling needs to be done in terms of hypercoagulability, live toxicity effects on creatinine levels, and incident of infection. All of these areas are suggested as problematic from studies and none have been systematically evaluated with the CoolGard system.

Additional information on how to deal with shivering neurovascular blocking drugs and the optimum protocol need to be established and placed in the guidelines for use of this device.

Four. A key study is the Alsius study. That was an observational study with a retrospective control group that was treated differently than a more recent treatment group. The study used various methods to obtain hypothermia. The sponsor also provided a reanalysis of data comparing the AKH Cardiac Arrest Registry to the results of the CoolGard 3000 system.

One disturbing part of the analysis is that patients who died within 24 hours of treatment were excluded, which could bias the study. I would have liked to have seen an analysis with these patients included.

Additionally, the monitoring was inadequate since the records were in German. And it was disturbing to note that the person who did monitor the study, as noted in the materials provided, did not read German.

The comparability of the groups being compared cannot be evaluated from the data available, and thus it's impossible to reach a conclusion on the validity of the data presented.

The supplemental materials provided additional data from a meta-analysis of three studies in 96 patients with selected and matched controls. The data is put together on a retrospective data and no basis for matching the patients controlled versus intervention is reported, and no justification for the selection of the patients over groups is made.

In fact, all newly added patients seem also to have concomitant cold fluids administered, making this group noncomparable to a controlled group which do not receive fluids. And the fluids would have to be of a normal temperature.

An attached study by Kliegel from Resuscitation highlights the potential problem of two liters of iced fluids causing cooling induction.

Additionally, selection of 26 out of the 167 patients of the study, basis not explained, a time of 38 minutes from return of circulation to administration of fluids an 53 minutes from start of resuscitation to cold fluid administration are times not comparable to the U.S. population.

All these differences highlight the significant difficulties of making comparison. Descriptions, possibly identifying the differences in the 96 are not available from the sponsor's report. But are probably no less than those seen in the 26 patient subset in the Kliegel study.

In terms of efficacy, the additional information on these patients is not helpful due to possible selection bias and potential noncomparability of the control group.

Also, the toxicity noted on table 2 presented by the sponsor in the supplemental information is disturbing, given the high incidence of toxicity and the inability to exclude a selection bias in the choice of controls in table 3 of the supplemental information provided. The comparison to controls is supposed to place in perspective the toxicity associated with cooling. However, the selection of controls retrospectively fails to do this since the method of selection of the controls remains unknown and questionable.

That concludes my advance review.

CHAIRMAN MAISEL: Do you have any questions for the sponsor before we move on or the FDA?

DR. SOMBERG: Not at this time.

CHAIRMAN MAISEL: Okay. Dr. Yustein, maybe I could ask you to comment regarding the Panel involvement's in the determination of whether a predicate device exists, which is what Dr. Somberg brought up. My understanding would be that that is the purview of the FDA and not this Panel?

DR. SOMBERG: I just wanted to interject that I did not suggest that I'm making the decision. I'm just suggesting that if I in my review thinking, which I wanted present was that if a predicate device existed, it would be a different set of risk benefit analyses. And if one does not feel that there was approval of a device for resuscitation, one would ask for a different risk benefit assessment of the device. And that's all I was suggesting in my review. But I'm certainly on the legal person to decide that.

DR. YUSTEIN: We certainly appreciate Dr. Somberg's comments, but what you said, Dr. Maisel, is correct.

CHAIRMAN MAISEL: So why don't we move on to our second review, which is Dr. Brott.

DR. BROTT: Thank you.

I'm a neurologist who hangs out in intensive care units. I'm a hospitalist/neurologist. Was the principal designer of the NIH Stroke Scale and had the original IND for tPA for stroke. And also have an interest in brain hemorrhage. So brain injury and urgent intervention and trying to measure effects of interventions are things I've been working on for a while.

And I had the opportunity to review this material in February. At the time I reviewed it with regard to the draft questions that were provided, and that's how I'll respond today.

I submitted a written report. But I must say I keep learning about this topic as well every day and with each presentation, as Dr. Swain pointed out.

First of all, the importance of cardiac arrest coma and of its treatment is very well described and justified within the materials. I'm not certain, though that the physiology is well understood either experimentally or in humans. The basic references provided relate to therapeutic hypothermia. Information regarding depth of hypothermia and duration and therapeutic hypothermia or the complications of experimental hypothermia are provided in the submission, but there was not a great deal of depth.

We've heard about the clinical data, and I'll try not to repeat what we've heard today. But first with the Bernard study.

I noted that randomization imbalance. And, actually, it was 31/31 at the end of the trial and then ended up 43/34. But what hasn't been brought up today is that good outcome was being discharged home or to a rehab facility. And it was based on the evaluation of a specialist in rehabilitation medicine. But there were no objective criteria provided as to how this rating was to be done. And this rating was the only determinate of whether the patient fulfilled the primary outcome.

The authors actually conclude at the end of their New England Journal article, the last two sentences are "However, treatment assignment was not blinded and there is the possibility that some aspects of care differed between groups. Therefore, further studies are required to confirm these findings and determine the optimal duration of hypothermia."

The second study, of course, is larger. I think it was carried out in more rigorous fashion than the first. I did have some concerns. There was a difference in the two groups which was impressive to me, but the therapy was delivered on the average about eight hours after circulation had ceased. And as a neurologist who deals with focal brian injury, this raised questions with me. And then today I did learn, and this was very troubling, that category 2 and category 3 which I missed, one could be hemiplegic and independent in all activities of daily living or be hemiplegic and be dependent in all activities of daily living. And that disturbs me in terms of the cut point used in this trial.

I do recognize that the assessment was blinded at six months. And, hopefully, that decreased the likelihood for investigator bias given the weakness that we learned about today specifically with the scale.

The third study, a randomized trial, was small as you know and did not explicitly describe good neurological recovery.

So these are the randomized trials upon which some of us might address predicate device or not. And these studies and the advisory statements of the Advanced Life Support Task Force of the International Liaison Committee on Resuscitation published in the July 8, 2003 issue of Circulation, they certainly posed a dilemma.

Alsius states in their submission that mild hypothermia is standard of care. My colleagues at the Mayo Clinic informed me that this treatment is almost never used in our intensive care units. A cardiology colleague who is chair of our Internal Medicine Department at Mayo is not aware of any general usage.

A cardiology college at Cincinnati whose group covers every hospital on the Ohio side of the Cincinnati, northern Kentucky metropolitan area states that this treatment is not used at any of those hospitals treating over one million people.

I spoke with the Director of Intensive Care Units for Kaiser in San Diego. To the best of his knowledge hypothermia is not used within the Kaiser system for post-arrest coma.

Therefore, one must question why this data has apparently not resulted in a change in care in these examples of standard of care within the United States.

With regard to safety and effectiveness of the CoolGard system itself, the randomized control data referred to support the safety and effectiveness of surface cooling with a temperature drop of approximately .9 degrees per hour maintained for 12 to 24 hours, these data were also reviewed as we all heard at the advisory Panel September 21st. At that time the Panel concurred that the randomized trials were interesting, but as was just stated, would not be sufficient support for an application for a particular device.

Quote from that hearing. "One cannot say that any cooling apparatus capable of dropping body temperature to a certain point should be labeled as indicated for cardiac arrest." The Panel quote concurred that endovascular techniques raise a different set of safety issues from surface induced hypothermia. And they went on to discuss that which I will not repeat.

We were given evidence from the 13 patient feasibility trial. We also requested and I was provided, as all Panel members were, additional surveillance data on 20 patients who were treated. And we also received some additional information with regard to patients with intracerebral hemorrhage and cerebral infarction.

And first the AKH, we've heard about that study. And I would just conclude that I tend to agree with the specific limitations that we heard about in that trial. And just summarize by saying, you know the medical students did review the data without written procedures. I was under the impression the records were primarily in English. I agree, the auditor didn't speak German. But what troubled me most were that table 3 shows significant differences at baseline with regard to the history of diabetes, the New York Heart Association Heart Failure Score, the frequency of out-of-hospital cardiac arrest, evidence for a presumed cardiac cause, ventricular tachycardia or V-fib on the first EKG, whether nor not basic life support measures were being given, Glasgow Outcome Scale on admission and patient age. And with statistically significant differences on those variables, I really was unable to make a confident inference from this dataset myself.

With regard to the feasibility study of 13 patients, which were studied over a prolonged period of time, as we've heard, four of these patients died. All were classified as non-device related. One patient died of a cardiac cause and three died after withdrawal of the life support. Of the remaining nine, two were in a persistent vegetative state, two were severely disabled, five had a good neurological recovery with a Glasgow Outcome Scale of 1 to 2.

Alsius, this submission actually states, "this was a feasibility IDE of limited scope. It demonstrated that the Icy catheter system was able to provide reliable and controllable thermal manipulation." No statements were made with regard to safety and efficacy.

So that's the dataset that we were given. And then we were given the additional information, as I mentioned. And we had a table with regard to complaints. And it had to do with technical failures of the device. And I'm co-PI for the CREST trial, which is a trial comparing carotid stenting to carotid endarterectomy. And in the course of that trial with over a 1,000 patients we have had two technical complications, which of course were immediately reported to FDA. But we have a much greater responsibility than to just look at technical complications amongst the patients that we study. And this was referred to by Mr. Marler and also several of the Panel members with regard to other aspects of safety that might relate to endovascular manipulation. And this was not really provided to us.

I will not summarize my conclusions with regard to the first four questions, but just head on to the fifth question, and one related to appropriate endpoints.

Up until today I thought the CPC was perhaps an acceptable scale. Dr. Roine pointed it has many similarities to scales that we use, the modified Rankin scale, the Glasgow Outcome Scale. And I think that if that ambiguity or inconsistency where you can have a patient with complete hemiplegia who is deemed independent in activities of daily living, and Dr. Sterz stated there were such patients put into category 2, if I understood your answer -- no?

DR. ROINE: No.

DR. BROTT: So there were no hemiplegic patients in category 2?

DR. ROINE: Not a single one.

DR. BROTT: Not a single one. Well, that's encouraging. Because I do feel that a categorical scale like the CPC is appropriate. But I think it can be improved if you do a detailed inventory of activities of daily living first. This is what we did in the NINDS IPA Stroke trial, as Dr. Roine knows. The Barthel Index is one such inventory where one has to ask questions with regard can you use -- can you get dressed, can you walk a certain distance, can you go up and down steps, can you completely independently use the toilet. It goes through these basics. And it takes a little while. But then when you have that score, which itself may not be that particularly useful, you have gone through and been forced to inventory what activities the patient can and cannot do. Then performing a more limited categorical scale has more objectivity and is less open to investigator bias.

And as a primary endpoint, such an ordinal scale with a few categories I think is still appropriate.

I learned again today from the presentation of Dr. Lazar that, certainly, I think it would be worth our while to add some simple, easily tolerated measures of cognitive function. And certainly we do have several of those that could be carried out within 2 or 3 minutes. He and I discussed the Folstein, which is not the best scale by any means, but given that this treatment has not been accepted nationally and if it is an appropriate treatment, clearly we need to persuade our physician colleagues around the United States the Folstein or some very widely recognized scale that can relate to all neurologist, all cardiologist, I think would be an appropriate secondary endpoint.

And other brief tests that we discussed -- where is Dr. Lazar?

Trials A and B, I think those three could be done very easily, would be well tolerated and I think could enhance such a trial.

The second part of question 5 mentioned randomized controlled trial. My primary concern with this specific device is whether or not it is substantially equivalent in safety and efficacy to the cooling methods used in the randomized controlled trials that we've discussed today. The new questions of safety have been mentioned by the other Panel members. Such a trial could have two arms comparing surface cooling to endovascular cooling.

I personally think that the data that we have had reviewed for us today with regard to hypothermia is not so strong as to make a third arm of the trial, a normothermia group, unethical in the United States.

And at that point, I'd close.

CHAIRMAN MAISEL: Thank you, Dr. Brott, for your comments. And at this point why don't we start having the Panel members make their additional comments.

We'll start with Dr. Brinker.

DR. BRINKER: Can I ask a couple of questions or is that --

CHAIRMAN MAISEL: No. You can make comments or ask the FDA or the sponsor questions.

DR. MARLER: Ken, maybe you can delegate who should answer this question. But it seems to me when you have a patient who suffered a cardiac arrest and is in coma, they very often have a lot of other things done to them during the hospitalization. And sometimes, not unoften, not infrequently, they have a great impact themselves on what happens to the patient.

So what percentage of patients are we looking at that have actually had cath intervention, bypass therapy, etcetera? And were they in some way censored if they had an adverse event consequent to one of those procedures?

DR. COLLINS: I think this is a question to hand to Fritz Sterz.

Just to make sure I understand, you're asking about the control group that was presented, were they --

DR. BRINKER: Well, I'd like to know if either group and whether in fact that would influence their placement in the control group or the active group, or whether they were censored. For instance, if they had another cardiac arrest during their stay in the hospital during one of these procedures, etcetera?

DR. STERZ: With regards to HACA, we have been asked all these questions and we have resubmitted these data in our revisions.

There have been no differences in using thrombolysis in either groups. No differences in using cath lab procedures. And no difference in using either other procedures like you said like you said before.

And we have also submitted regarding HACA that we were shooting for 500 patients. We did no interim analysis. I don't think that this is written in the paper, but this data would be unavailable.

With regard to the CoolGard data, this data would be available. WE didn't present them today because in the prospective trial they didn't have any influence on the outcome.

DR. BRINKER: So the other half of my question was no patient was censored for any reason once they were admitted to the trial as far as outcome goes?

DR. STERZ: With regards to you mean censored --

DR. BRINKER: Well, "censored" meaning excluded for analysis because they had an arrest or died during a CABG operation or an angioplasty or something like that.

DR. STERZ: Then he would fall in the mortality group.

DR. BRINKER: So he wasn't censored then? Okay. That's the question. Thank you.

My general comments are that I tend to agree with the two reviewers, that I came here actually willing to accept a benefit for hypothermia in patients suffering VT cardiac arrest. I think, however, that the information upon which this recommendation has been based, that is the efficacy of hypothermia, is on less firm grounds than I would like to have it seen. And then once we say that, once we say that there is some equipoise about whether hypothermia is the standard of care, a lot of other things then fall out of that.

So my feeling is that I would have loved to have seen a truly randomized trial comparing this device with normothermia care of these patients.

CHAIRMAN MAISEL: Thank you, Jeff.

Rick?

DR. PAGE: I'm a cardiac physiologist and have cared for patients post-cardiac arrest for 15 years or more now. And I must say that as I was thinking of the patients I've cared for, and some of the most tragic are the ones who actually have survived with significant impairment, the nonindependent livers, and hemiplegia is not something I typically see. So that seemed to be from my standpoint not a major issue, although we've made something of it. And I think it's interesting to hear that among your class 2s there weren't patient that you're recalling that were truly hemiplegic, and that would certainly be consistent with my clinical experience.

I, too, came here with probably more zeal for hypothermia and its potential and for patients post-arrest. I'm a little bit less reluctant to give up some remaining enthusiasm for this.

I'm impressed by the fact that the international bodies got together and examined these data and as a group recommended that this be standard of care. Now the question is why isn't it standard of care. And I think the obvious answer is that there are no tools. Hefty bags full of ice is not a way to make a major change in how you're dealing with post-arrest patients. And so I think that accounts for what might appear to be skepticism, and I think has in part, just an issue that there is no tool to administer this practically.

In terms of the safety issues we're talking about, I'm concerned that I don't have as much data in terms of safety as I wish. On the other hand, as I think about who are these patients we're dealing with, we're dealing with patients who most of the time expire and are left when they don't expire with persistent profound neurologic disability that leave them dependent for the rest of their lives.

And when I'm looking at data that suggests a four time increase in survival to functional neurologic status, I find that compelling even with the problems that I wholeheartedly agree we can find in each individual trial.

So I suppose I am warmer to this still; no pun intended. But if this group were to recommend and if the FDA were to approve this, I think I'd really want to see the data and have collection of data to know whether the right decision had been made.

CHAIRMAN MAISEL: Thank you, Rick.

Henry?

DR. HALPERIN: I'm a Professor of Medicine, Radiology and Biomedical Engineering at Hopkins. I've been doing CPR research for more than 20 years. I'm very familiar with this area. I've done a little bit of hypothermia work myself. And I have been a past Chairman of the American Heart Association's Advanced Cardiovascular Life Support Subcommittee wherein some of that time we did deal with this issue. So I'm actually very familiar with a lot of the aspects of this issue.

Our own hospital has implemented hypothermia in our coronary care unit ever since Dr. Sterz and others, and Dr. Bernard's studies were published. And, actually, you know ancedotally since some anecdotes were already presented by Panel members, our own anecdotal experience has actually been very positive with hypothermia. There were some patients that we thought would never survive that actually did. This experience has been duplicated at a couple of other universities that I'm aware of that I actually won't mentioned, because I actually haven't asked for their permission to do that. With similar very positive anecdotes. I don't think any of those places actually used this particular device, though.

So then I think there's a couple of major issues here, one of which is it hypothermia per se that causes the benefit or is the way you achieve hypothermia that causes the benefit? Personally I think it's the hypothermia itself rather than the way it's achieved. Because I think the major downside potentially of the device under question is in fact the complication profile. And since this device is approved for a couple of other indications and the adverse events probably have been pretty well documented, the issue is just cardiac arrest per se, add a new covariable that would make the adverse event profile different in a post-cardiac arrest situation versus post-surgery or whatever else it's being used for. And I don't have an answer for that, although I'd be surprised if that was an issue.

So I think then at looking at the data, I mean certainly the data has been very rigorously looked at in many different ways. And to me the bottom line is that even under the most rigorous scrutiny of this data, there's still seems to be a benefit, although marginally statistically significant when viewed at least by the FDA's point of view. I would be interested in to how often 97.5 percent confidence intervals versus 95 confidence intervals are actually used. But that's probably less of an issue. Because in fact there's been no intervention other than hypothermia that when started after a cardiac arrest occurred that showed any amount of benefit at all.

And I think that, you know, we can argue and will over how good an intervention hypothermia probably is, but nothing else has even shown any glimmer of hope. So I'm a bit more hopeful I think about the use of hypothermia in cardiac arrest still.

CHAIRMAN MAISEL: Thank you.

Dr. Blumenstein, are you still there?

DR. BLUMENSTEIN: Yes, I'm here. I have some brief comments. Is it appropriate to give them now?

CHAIRMAN MAISEL: Yes. That would be perfect.

DR. BLUMENSTEIN: Sorry, I didn't hear the response.

CHAIRMAN MAISEL: Yes. Please give your comments now.

DR. BLUMENSTEIN: All right. Well, the sponsor has used four techniques in trying to resuscitate the data. They've used Bayesian analysis, propensity score, meta-analysis. These are all techniques that if you add a poor quality randomized clinical trial, that you could call these things Hamburger Helper.

In this case where you have a mishmash of convenience data, I think that you could say that these things would be called "dog food helper."

A randomized clinical trial is essential in this situation. And I don't know how one could come to a conclusion that there has been a demonstration that efficacy or safety without a randomized clinical trial.

There was one question that was asked about if there is a requirement for randomized clinical trial what sort of endpoint serves. And my suggestion would be something like a bad thing-free survival where bad thing is determined when a patient enters a state for which there's no substantial chance of improvement, such as a neurological deficit or something along those lines. And I'm suggesting doing a survival analysis here because I think that you'll get some greater sensitivity and a greater understanding of what's going on if one designs the trial for an assessment of a endpoint, a failure time endpoint.

That's all I have to say.

CHAIRMAN MAISEL: Thank you for your comments.

And we'll move on to Norm.

DR. KATO: My comments are really directed at two separate issues. The first being on the issue of hypothermia for cardiac arrest. Looking at the experience of cardiothoracic surgery, we have for a long time believed that hypothermia was necessary for the performance of -- and even mild hypothermia was necessary for the performance of cardiovascular surgery. However, in looking over the historical information that actually came about because we never, I don't believe there was ever any randomized study that looked at hypothermia per se. But it actually was because of the fear of actually keeping the patient on the cardiopulmonary bypass machine for a long period of time. Hypothermia was thought to be neuroprotective, thought to be myocardial protective. But the reality is in the 1990s when we began doing warm cardiac surgery and keeping the patient basically warm and doing warm continuous mycardiopreservation techniques, we basically realized that hypothermia wasn't necessary at all.

As a matter of fact, we learned that our biggest problem in the operating room was keeping the patient warm. It is actually very common that hypothermia below 34 degree centigrade occurs very, very quickly in the operating room setting without any need for ice or any other intervention, but just being naked in a air condition room suffices quite well.

I've heard today that there have been a lot of anecdotal reports about hypothermia in cardiac arrest from many distinguished universities and distinguished colleagues here. And I think that before a panel such as this or any other group basically rules on or establishes a standard of hypothermia as a standard of care in cardiac arrest, that the randomized prospective study does need to be done as suggested by many of the Panel members before me.

That being said, that hypothermia is at least -- we do not have sufficient data to claim that hypothermia is a standard of care or is the standard of care in this situation, I'm a little big disappointed with the sponsor for not taking the -- or not assessing the scientific field correctly by not performing that randomized study with the device.

I am somewhat fearful that on occasion the 510(k) predicate device pathway sometimes allows sponsors to avoid the -- or attempts to avoid the issue of performing that necessary clinical trial which will prove that efficacy issue once and for all.

So I would challenge the sponsor that while this device looks like it may have a role in this area, to proceed with that clinical trial. Seek the advice of experts such as Dr. Brott and others in the neurological field for their knowledge about outcome data. And seek out other experts in terms of clinical trials who have much more experience than I do. And do that study and show once and for all whether this device does work, and really take the high road. Because I think the other thing that's becoming clear also with the increased scrutiny of both devices as well as drug products is we definitely need to have more and more data available to us.

CHAIRMAN MAISEL: Thanks, Norm.

I would just simply comment that the 510(k) process does not exclude the possibility of a randomized trial being required. That's not -- doesn't necessarily make it a PMA.

DR. KATO: Yes. And I didn't want to say that this was not an issue -- that the 510(k) process did not require that trial. But I was just saying that I think that the -- I wanted to challenge the sponsor to take the high road and do the trial.

CHAIRMAN MAISEL: Okay. Thanks, Norm.

Mike?

DR. WEISFELDT: As will relate to some of my answer, let me just -- with regard to the overall proposition of the benefit or lack of benefit of hypothermia, I have I think a special relationship to the issue which I'll comment on.

Again, I'm a cardiologist by background. I'm in an administrative position now at Hopkins. But about nine months ago I became the study chair for what is called the Resuscitation Outcomes Consortium which is a five year NIH Department of Defense and Canadian similar group sponsored study of resuscitation from cardiac arrest and from traumatic injury. The network is to do definitive clinical trials in those two disorders in eight U.S. cities or communities and three Canadian with the potential, at least, based upon numbers of patients of entering, 3,000 patients per year in studies of trauma and 10,000 patients per year in studies of cardiac resuscitation. So if you will, the potential for doing a study actually exists.

So let me make my focus comments and then with that as a background orientation statement.

The comment on the -- I'm going to make a brief comment on the second of the two issues, which is this device relative to other devices. And my opinion is that the device needs a prospective randomized trial for safety relative to surface cooling performed in some way.

So finished with that comment, let me now go to the really to me extremely difficult situation of having the data we have presented and discussed about the core issue of the value of hypothermia in this subset of patients versus no hypothermia.

Resuscitation from cardiac arrest there is no drug, there is no procedure, there is no device other than the defibrillator that's ever been shown in a randomized study to provide benefit, no matter what the drug, device is. The only device is the automatic implantable defibrillator, where I was to some degree involved in the study that was published in the New England Journal of Medicine, which as in passing, a rather difficult study that is not beyond criticism itself.

In this arena of cardiac arrest and the performance of research without informed consent or with the proper term is an exemption from the informed consent, we have no national body that has any official capacity relative to what it has equipoise and appropriateness for clinical research. We rely, in fact, upon local IRBs to provide such guidance, such decision making. And I believe that even with the critiques of the studies in place and analyzed, that the degree of difficulty of performing a study such as we've been talking about would indeed, be extraordinarily difficult in any setting because of the lack of -- your opinions may be that it's equipoised. And the FDA's opinions may be that it's equipoised. But nobody really has the singular ability to define. They have no official stance on where there is equipoise. And still you have -- and what you are telling a local IRB is there are national and international standards that this works, but we don't accept the data and therefore we think you as an individual IRB with, if you will, 20 IRBs per study site, you make the decision to do this study.

The data and the critiques, one wonders whether further inquiry upon by the FDA or the statisticians into the HACA study and the Bernard study would not allow better analysis of groups from that study that might in fact become more convincing or settle some disputes.

The HACA study critique, I would just read back, the study stopped prematurely because of low enrollment and ending of funding. Plan study number not known. Unknown whether interim analysis was performed. Well, those are a set of factual information that could be obtained.

The Bernard study some further questioning might clarify a part of that study that would be includable in a meta-analysis between those two studies. The third study seems to me to be trivial to the question.

Since there are animal studies that seem to show a benefit done by Dr. Safar and his group, and if we could get more confidence about these studies from delving into them further, we might be able to draw a better conclusion about the indication and the recommendations which are now in place about the use of this treatment.

There was one other point I was going to make, but I forgot it, so I'll close.

CHAIRMAN MAISEL: Thank you.

I'd just like to make a couple of comments with my background as a clinical cardiac electrophysiologist who sees cardiac arrest survivors.

I think the ILCOR recommendations are noteworthy. I think a number of smart people reviewed the data and I think their recommendation, while based on the data that we've seen that is somewhat flawed, I think were thoughtful certainly in their review.

I do not think it is the standard of care, at least in this country, for people to receive hypothermia following cardiac arrest. That being said, it certainly is a reasonable treatment option for these patients, at least the surface cooling. And the reasons why it is not the standard of care are a little unclear to me, although I suspect as mentioned earlier, I think it's more the reality of trying to actually do surface cooling on a patient. It's extremely difficult and challenging, even if you decide that it's something that you want to do.

And so I do see the potential for the device that we are reviewing today to be very useful, meaning that it's a lot "cleaner," maybe easier to actually implement in an emergency situation and not interfere quite as much with patient care.

That being said, I have a little trouble making the leap from surface cooling to endovascular cooling for a number of reasons, including most of which have been addressed already this morning including the safety profile. I'm not comfortable that we have enough clarification of the safety issues with the endovascular cooling.

The design of a randomized trial I think is a challenge. I do think surface cooling versus endovascular cooling would be an ethical study to conduct. I think the problem that could arise is that the endovascular cooling may have a higher complication rate. And without a normothermia control group we may have difficulty assessing the benefit of endovascular cooling. And if we look at the data of surface cooling versus endovascular cooling, we may decide that the complication rate is higher and is therefore, not a reasonable option. But if surface cooling is not being performed because it's efficient and can't be performed, then a higher complication rate may be acceptable.

One question I had as far as clarification regarding the registry, I'm a little bit confused as to why patients were excluded if they died within 24 hours. I mean, that seems -- you know, if I'm sitting here in front of a patient that's 38 minutes into their return of spontaneous circulation and I need to make a decision about whether or not to cool them, that decision, I can't wait 24 hours to make that decision.

The comparison, to my knowledge, that exclusion was not applied to the published New England Journal papers. And so the comparison we're making of numbers of survival and good neurologic recovery seem a little bit apples and oranges to me. And maybe the sponsor can just address that 24 hour cut off for me.

DR. STERZ: They've excluded from analysis but not from treatment.

CHAIRMAN MAISEL: Were they enrolled in the registry?

DR. STERZ: Yes, of course. But they were then, when we took the data out of the registry, excluded for the analysis.

CHAIRMAN MAISEL: Right. And I'm asking why you choose to do that. To me the analysis would have been treatment versus -- you know, normothermia versus endovascular cooling.

DR. STERZ: We thought if we do these for both -- you can have it with them. It doesn't make a difference. We looked at it. If we include it's the same results show up. But this is not in information. But we thought being fair to this treatment or being fair to these patients, or whatever, we wanted to say we want to have them at least surviving the certain period of time because they died of cardiac shock or whatever -- of the severity of the cardiac arrest itself. And our -- can answer you the question if the period of no flow gets a certain period of time, you have an estimate, yes. But the real estimate how long this no flow or damage to the heart was if you cannot stabilize them.

CHAIRMAN MAISEL: So do you have a --

DR. STERZ: If you look at our data in the CoolGard group, there was only one excluded and then the controlled group, there were 250 excluded. So --

CHAIRMAN MAISEL: Can you put those numbers back? It's sort of like intent-to-treat an on-treatment analysis given your explanation of why you took them out. Can you put that data back in and give us the complication and the mortality rates and neurologic benefits.

DR. STERZ: Sure. Of course.

CHAIRMAN MAISEL: And then I believe you said there's no difference, but it would be nice for us to be able --

DR. STERZ: Of course.

CHAIRMAN MAISEL: Because it would be an important analysis point.

DR. STERZ: Sure.

CHAIRMAN MAISEL: And maybe after lunch we could see that data?

DR. STERZ: Yes, and maybe I will leave.

DR. HALPERIN: It's in your computer, right?

DR. STERZ: But I think I have them ready, but I have to look it up. I can do over lunch.

CHAIRMAN MAISEL: Okay. We'll ask you again after lunch for that information.

The final comment I would like to make is simply that where you have presented a number of analysis, and I realize the data is limited, we've essentially looked at the same data in four different ways or three different ways. I mean, we saw the New England Journal studies, and then they were represented. At least some of the data was represented in a meta-analysis and some of the data is represented in the registry. So, you know, these are not independent datasets that we're looking at from three different sources. There is certainly an overlap there, and that also I think limits the interpretability of these studies.

So at I think at this point -- oh, yes. Henry?

DR. HALPERIN: I'm sorry. One last comment. Because this notion of the standard of care has been thrown around a lot, and I just wanted to address that very briefly. Because the ILCOR statement on hypothermia has been shown. And, in fact, I don't think as august a body as ILCOR is, I don't think most U.S. or cardiac arrest treatment and training facilities would actually take that as the standard of care. Because they actually look to the American Heart Association, which is kind of the self-proclaimed keeper of the guidelines, if you will. Guidelines are published every five to seven years which basically present the current state-of-the-art of what cardiac arrest is. And although the Heart Association is very clear to say it's not the standard of clear, in truth it becomes these standard of care. At least it's what most people wind up doing.

And the Heart Association itself has not published a statement definitively on this subject yet. It will come out, though, at the end of this year in the 2005 guidelines that have just been developed are in the process of being developed. The conference was at the end of January, just this past January. And so the various committees are now kind of deciding what level of recommendation to give different things. And one of the those different things is going to be hypothermia.

The previous guidelines were in the year 2000. And I think these studies did come out after that. So then the Heart Association really has not had an opportunity to really publish formal guidelines on this particular subject. So I think that is one particular reason why this may not have been adopted more than it has been.

CHAIRMAN MAISEL: Thanks, Henry.

I think at that point we will break for lunch. And let's convene at 1:15, please.

(Whereupon, the meeting at 12:08 was adjourned, to reconvene this same day at 1:20 p.m.)

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

1:20 p.m.

CHAIRMAN MAISEL: Good afternoon.

Why don't we begin the afternoon session? We're somewhat ahead of schedule and had the Panel members, we've done the lead reviews and have each of the Panel members question the sponsor and the FDA and make their comments.

I was asked by the sponsor if they could address some of the Panel questions from this morning, specifically issues related to the exclusion of patients who died with 24 hours and some corrections to some statements that were made this morning. So you may address the Panel at this time, if you wish.

EXECUTIVE SECRETARY WOOD: I would ask the sponsor not to sit at the table, but rather whoever is addressing the issues stand at the podium. We like to leave it free so that the Panel can ask questions of either the FDA or the sponsor.

DR. COLLINS: There were questions asked this morning, in particular for the patients that did not live 24 hours. The starting number of patients for this analysis was 1882, that's including all CoolGard and control patients.

Now the control patients, of course, were control patients. They weren't cooled by any mechanism.

They were recruited, treated and went to other centers were 82. So they didn't complete their treatment at AKH.

Other methods of cooling were 92.

And then within the first 24 hours there were 206 deaths, one CoolGard and 205 normothermia.

The odds ratio with the deaths under 24 hours excluded, we reported to you, as 1.61. If we add in all the patients who died under 24 hours, the odds ratio is 2.17.

With the deaths less than 24 hours removed, the simple survival outcomes are as posted; 43 of 62, 695 of 1,191. And with deaths left in for CoolGard, 34 out of 63 and for the normothermia 695 out of 1,396. It was conservative for us to produce data with the deaths under 24 hours removed. It made the picture less appealing for the device. Putting it back in again, as you can see, significantly improves the odds ratio.

The question was asked about the 2002 to 2003 datasets, what the actual numbers were. The 30 day survival normothermia versus device, 56 and 119 for 47 percent, the device 43 per 63, 68 percent survival with an unadjusted odds ratio of 2.43.

Just one comment about ILCOR recommendations and also about HACA. The advisory statement was endorsed by the AHA. It was an interim advisory statement in July 2003 published in Circulation.

I believe in the public session this afternoon Dr. Vanden Hoek will be available to talk to that. I think there have been a lot of statements made about AHA has or has not said. That was an AHA advisory statement. And the rather exhaustive mechanism for assessing the data, which I'm told was both bloody and prolonged, was the standard AHA mechanism for doing so.

Fritz, did you want to address the HACA?

DR. STERZ: Well, with regards to HACA study, again, the sample size we were shooting for was 500. And there was no interim analysis done. It was really only given up because we were running out of money after four years. It was a real four year undertaking that we could do it.

Regarding the HACA, the safety data have been prospectively defined and prospectively collected. There was no retrospective chart review. It was all prospective in the HACA trial.

With regard to the registry, the data on outcome, the data on the baselines have been prospectively input in a database by myself -- controlling all the charts which have been filled out according to a specific defined protocol by our team. It was a team of ten, sometimes 15, physicians who were trained by myself how to do it, how to make the interviews, how to get this paperwork filled out. Then I got the paperwork, if I haven't done it myself, on my desk. Reviewed all charts myself. All the data which are in the database, 2,500 times 500 I put myself into the database and quality controlled them.

In addition, I had a physician after myself controlling me if I did the right job and controlling again, if the data were put in an appropriate manner into this database.

I would be glad or whatever I can tell you if you want this data and use them for your analysis, I would provide them to you immediately. I'm not a statistician, I'm giving you what I have recorded.

And this is the same thing with the meta-analysis. For the meta-analysis we have received all the data from the participating centers, which was the Bernard study and which was the Idrissi study. We have it raw in our hand, and I'm sure they will give me the permission to hand it over to you to reanalyze this meta-analysis because in this database for the meta-analysis we have the follow-ups, and this was published now in Critical Care Medicine, until discharge. Not only for survival, but also but for neurologic outcomes. I would be happy to pass them over to you if you want to reanalyze this.

In the HACA trial we had certain time points of follow-ups after cardiac arrest. One day, two day, one week, one month, discharge and six months. And we interviewed the patients by phone, and not only this we called them in to come back to our department and to check them up by blinded investigators.

DR. COLLINS: The chart review was our monitoring of their charts, not --

DR. STERZ: The chart review --

DR. COLLINS: And, yes, I used a translator from the German to help me to do that.

DR. STERZ: The chart review was done only in the historic control patients used for the safety issues. Not for the HACA. In the '90/'91 patients we didn't look at the safety points because there was no reason to look at it. So in these control patients I was forced to do a retrospective chart review for the safety data only. These were about, I think, 20 to 30 percent of the data which were done by chart review and only regarding the safety data.

All the other data were done prospectively.

CHAIRMAN MAISEL: Okay. I appreciate that clarification. I can't speak for everyone on the Panel, but I don't have any issues with regard to the quality of the data that was presented or the diligence with which you submitted your data.

John?

DR. SOMBERG: I still need clarification to understand that if I'm correct, that there is a subset of patients from your registry or your data collection -- I won't even use the word "registry" -- from your data collection that have received the device under or have received a therapy with the device under consideration today. And this was a subset analysis called the AKH Cardiac Arrest Registry. Am I correct in stating that?

And you're saying that the controls in this group compared to those who got the drug were all done at the same time and they are not retrospective controls versus perspective collection of data? So just take me through. Because I'm not interested in proving a lot of different points. But I'm interested in your device, and I'm interested in is there a comparative group that's done contemporaneous with the same intake criteria and not that they couldn't get the therapy and this group could get the therapy.

DR. COLLINS: I think in terms of the HACA, while the HACA study was running -- correct me if I'm wrong, Fritz -- the patients that were recruited into the HACA trial were recruited into the HACA trial. But hypothermia was not the standard of care in your institution, is that correct? It's not correct? Would you please talk to that?

DR. STERZ: The HACA trial was running until 2002 or 2001, I don't recall it now anymore. But the HACA trial was -- I didn't interfere with anything else in the HACA trial. The HACA trial was on its own from 1999 until 2002 or 2003. It was on its own.

Afterwards, we immediately switched over to Alsius CoolGard device because we wanted to test it and we wanted to see. And this started 2002 where you have here the concurrent analyzed controlled data by the FDA.

DR. SOMBERG: And how is the groups go from device to normothermia? What determines that selection process? It's not a randomized controlled study. So how do you --

DR. STERZ: This selection process was done on the standard order, which I have given my fellows. I told my fellows this and this, and this inclusion criteria you have to meet, then you put the patients into cooling procedures.

DR. SOMBERG: And if they didn't meet that, they went into the controlled group?

DR. STERZ: Therefore, the propensity score scores are so different between these two groups.

DR. SOMBERG: Okay.

DR. STERZ: Sorry, but that's how it is. Otherwise I wouldn't have given this standard order for procedures in the beginning.

DR. COLLINS: There was also a question asked about the relationship to neuropsychological data on neuropsychological testing on the CPC. Dr. Roine has that data.

CHAIRMAN MAISEL: So maybe I can invite the sponsor to have a seat. We'll open up the discussion to the Panel and they can ask whoever they would like to ask.

So why don't I start with Dr. Marler and see if you have any comments since you didn't get an opportunity to speak this morning?

DR. MARLER: Well, I just wanted to say that what I've heard from the applicants and the public speakers is that there's really been a remarkably good and sincere effort to look at the data that's available and trying to make a conclusion about what is effective and safe to use.

Instead of the randomized clinical trial, though, the applicants are presenting kind of a series of logical inferences. And each little step in the logic to me introduces a degree of uncertainty which at the end of reading through this and at the end of, at least where I am right now, it leaves me unsure whether patients will truly benefit from the procedure that we're talking about or really unsure that we can be confident that they're likely to benefit. And I just can remember quite a few times when professional judgment and professional standards are developed pretty much through the same pattern of thinking looking at observational studies or just clinical experience to develop standards based on professional judgment and consensus that have been eventually proven quite wrong when actually tested in a objective clinical trial. And I feel that there's the situation that we're there, could possibly be there in this question.

CHAIRMAN MAISEL: Thank you.

And at this point I would like to ask the Panel if they have any more comments or questions for either the FDA or the sponsor before we move onto the FDA questions.

Henry?

DR. HALPERIN: I just wanted to clarify one issue about AHA endorsement. I know ILCOR gave it a Class I recommendation is my understanding, which would make it the standard of care. And the AHA, ACLS committee at least that I was on at the time did not sustain that Class I recommendation.

I don't know if it's appropriate to maybe call on Terry, if you could clarify that? Because weren't you involved in that AHA process?

DR. VANDEN HOEK: Yes, I was. And I've been on the ACLS Subcommittee for I guess almost ten years now. But the ILCOR group, the Europeans put together, have wanted to give it a Class I. I think that there's going to be some finalized discussion by American Heart. But I think that their overall very favorable. I can't say whether it would be a I or a IIa, but I think that there would be some sense that this would be a reasonable thing to in a number of circumstances.

DR. HALPERIN: But there's not currently a class recommendation by AHA?

DR. VANDEN HOEK: That's correct.

DR. HALPERIN: Yes.

CHAIRMAN MAISEL: Any other Panel comments?

Why don't we move on to the questions. And if I can ask Geretta to read the first question.

EXECUTIVE SECRETARY WOOD: Alsius is requesting to add the following new indication for use to the CoolGard 3000 system:

"For use in the induction, maintenance and reversal of mild hypothermia in the treatment of unconscious adult patients with spontaneous circulation after out-of-hospital cardiac arrest when the initial rhythm was ventricular fibrillation."

In support of this request the sponsor has submitted a meta-analysis of several surface cooling studies and clinical data with their specific device from two different sources:

A 13 patient prospective uncontrolled U.S. feasibility study, and;

A prospective nonrandomized, single-site observational registry with "matched" controls. The AKH Registry Data.

Question number one: Please discuss whether you believe the data provides reasonable assurance of safety for the proper indication. In your discussion, please specifically address whether:

(a) The manner in which the data was collected (prospective registry) provides adequate assurance that the rates of adverse events noted in the submission are representative of what might be expected in actual clinical practice.

(b) The data adequately addresses the risks and potential concerns of intravascular cooling mentioned at the September 2004 Panel meeting, including bleeding, clotting, DIC and ventricular fibrillation.

CHAIRMAN MAISEL: So we had quite a bit of discussion about some of these issues this morning. And, Mike, you asked a few questions on this topic. Do you want to summarize your thoughts regarding the safety issues?

DR. WEISFELDT: I think the safety issue remains unclear. You have some information to suggest renal and perhaps pancreatic complications of long term catheter residence in the inferior vena cava with cooling of that catheter to, obviously, a temperature that may or may not have an effect on the blood that goes by. And it certainly does concern me as to whether there is any database for a duration of cooling that mirrors the duration that is recommended currently for cooling with the catheter system, even in the approved indications, that is the ILCOR statement at least is 12 to 24 hours of cooling. And at least so far I haven't heard anything to the notion of safety of this catheter in cooling in any human being for that period of time in the approved indications. And, therefore, we have to look at only the data that's available in the cardiac arrest situation. And there are, certainly, some evidence of renal and pancreatic problems that may not be present with other forms of cooling. So I remain concerned about safety and believe that a randomized trial with safety endpoints for surface cooling versus endovascular cooling would be a highly desirable piece of data.

DR. BLUMENSTEIN: Yes. Brent Blumenstein. I'm not hearing.

CHAIRMAN MAISEL: We'll try to fix that.

Did you have some comments or you just couldn't hear well?

DR. BLUMENSTEIN: I was not hearing. The previous speaker just faded away.

CHAIRMAN MAISEL: Okay. Anyone else on the Panel have safety comments? John?

DR. SOMBERG: Well, the more I think of this, the more it becomes a conundrum because I sympathize with the sponsor here because they're bearing, as the first one, is bring this cooling to the regulatory for it, they're carrying the burden of trying to prove the issue of hypothermia and also prove the issue of safety. And it may have been much better to have initially focused in what we heard today is reanalysis of the AKH Registry.

We had two groups receiving different therapy, cooling/noncooling , but contemporaneous at a same time and there is a suggestion, as there is all through the database, that there's a benefit here. The trouble is are the two groups comparable and have they been looked at to be comparable enough. And I'm getting a little bit more comfortable that they may be. But it was not clear from the data packet or any information.

So I'm not sure are we suffering from multiple misdata inputs, if you will, or are we just having inadequate information. And with that said, I still do not, as I hear others in the Panel as well who are more experience in the resuscitation area than I am, say they're not comfortable with the safety. But I think the cup is half filled or even more so, and I do not think it's a binary decision yes or forever no in this area.

So I think it would be useful just from my advantage point to: (1) have comparable groups exposed or not exposed to the intervention. And one has to be clear what the intervention is. Because it's more than just the catheter, it's also the medications and it's also, I believe in the latter group, this ice saline rapid induction technique as well. So when you put it all together that's what you're going to be looking at as a safety issue. And then to expose it to that. That could still be done.

Now the second issue I have is does cooling really work, and there are some questions about that. And I would like to see some data. I believe it does in the European experience. I believe it would be useful to have some data with, unfortunately like the city of Chicago where it takes an hour to get anyplace you know the marked delays and that sort of thing, does it really have a benefit in there. And I'm not sure if the sponsor can answer that issue. And it may be something that, you know even before I even heard about this NIH consortium, it's something that the NIH could need to help as a public health issue. But now there is a consortium and it may be able to answer that sort of thing.

So I think the committee's asked for opinions in a lot of areas that are very hard. And I can't speak for the committee, but my opinion is that I'm leaning towards feeling that there is a signal here both in terms of hypothermia efficacy and in efficacy and safety of this device. But it's kind of murky and someone has to, not in this forum, but has to be able to go through in much greater care and probably get some additional information.

CHAIRMAN MAISEL: Are there any Panel members who feel we have adequate data in front of us to say that the device is safe?

DR. PAGE: I'll address that. And I would first off say that we don't have enough data to say for sure it's safe, hypothermia therapy that is. And, indeed, there have been practice patterns that have gone on for years that have come around by consensus.

I gave a lot of Lidocaine to acute infarcts a while back before we figured out that it wasn't really a good thing to do.

On the other hand, I don't know if we're going to get any better data than what we have here. And I'm seeing a strong enough signal in terms of safety and efficacy of this of the concept of the hypothermia that I lean in the direction toward getting all the information we can from the data that have been collected. Maybe I'm a pessimist, but I don't believe we'll have the randomized trial that we all wished we had before us right now. And that being lacking, I can give an endorsement to the concept of hypothermia.

CHAIRMAN MAISEL: For those members who do not feel that safety has been adequately expressed, what specific endpoints do you feel need to be obtained, and are there any study designs short of a randomized trial that can specifically address this safety issue. Not necessarily effectiveness, but the safety issue. For example, would a prospective registry of another 50 patients or a 100 patients supply you with what you need?

John?

DR. SOMBERG: And I want to repeat. I'll just be very brief. Is that randomized trial is preferable. But if one for reasons cannot do that, then the reanalysis of the AKH Registry raises the issue are the two groups comparable? Because why didn't one group in an institution where there's a dynamic leader who says hypothermia is the way to go, why didn't those people get it? So I'm worried about that.

So if one can assure the comparability of the two groups, if they're not randomized, maybe it's just one doctor does it one way on Monday, Wednesday and Friday and the other one does it Tuesday, Thursday and Saturday; that's fine as well. But some sort of way to make sure that one group is not much sicker than the other, the control group, therefore making the intervention look more effective and look safer because the sicker that have more problems.

CHAIRMAN MAISEL: Jeff?

DR. BRINKER: I would just like to question one piece of this, and that is the concept of the randomized trial cannot be done anymore. In a country where we exist, and I would venture to say that less than 10 percent of patients having cardiac arrest or defib out-of-hospital get hypothermia now. So it seems to me that a randomized trial could be done, if not to normothermia certainly to some form of hypothermia other than this device. And compare the two in terms of a number of endpoints. One, of course, is safety and for the sponsor it might be good to see for them if there were other issues in terms of utility and ease of patient care what device offered advantage.

So I think that there is room for that kind of information that would give us some information on safety.

The other issue that boggles my mind but in the FDA logic of this probably makes sense, if we accept all the data that shows from the European and Australian studies that hypothermia is good for this situation, basically none of that was achieved with this device, the randomized trials. So we're in a position of potentially recommending an approval of one device on the basis that it is roughly equivalent in action to other devices which are not approved for this, which sort of boggles my mind a little bit. But I assume it's not unreasonable to think that way.

But if you do, then it should be equally reasonable to think that you could a randomized trial between those two, this device and other devices, and to see if there's a difference in safety, utility, etcetera.

CHAIRMAN MAISEL: As this Panel often encounters a relatively small randomized trial that addresses uncommon safety issues, it's going to be a difficult design to identify rare safety issues.

Tom?

DR. BROTT: In that regard, I would say that, John Marler's here he's heard this before, but patients like this in terms of a study are like diamonds. They're rare. They will be difficult to enroll. And I think they need to be studied extremely carefully so that a Panel such as ours can be assured we have eight different ways of looking at the comparisons. We have the primary endpoint, but then we have other things.

These patients, they all ventricular fibrillation, arrest. I'm not a cardiologist, but they must have coronary artery disease. Many of them, if not 80 to 90 percent. Does ice cold water or more rapid cooling do something to recurrence of coronary artery disease or their infarct? What happens to the EKG?

Whatever the design, I think that the surveillance has to be at a different level than what we've been presented, which is why I think we've struggled.

DR. BRINKER: As far as being diamonds, we might be more like South Africa than you might think. There are over 100,000 people a year that might qualify for this if it were done in a reasonable way. And Mike was suggesting that they could enroll up to 10,000 patients in a three year study. So I think that you could do this if it was done appropriately. But I agree that there are other pieces of information that would be extremely important to have.

But the bottom line for our concerns now given the data that we do have, would be simply whether this device is as safe as a comparative device. And while we might not get the very rare things, if there's a general consensus that at least we do away with issues like renal failure and pancreatic problems and maybe infection, then those are big things that I'd like to see settled.

CHAIRMAN MAISEL: So if I could try to summarize the Panel's thoughts regarding safety. Most, but not all of the Panel members seem to feel that we need more data with regard to the safety endpoints, specifically with regard to some of the things that we've seen with pancreatic and renal injury as well as bleeding complications with catheter insertion, etcetera. That a randomized trial design would be the best way to establish these issues, but not necessarily absolutely the only way.

Does anyone take issue with anything I just said? Okay.

Why don't we move on to question number two.

DR. YUSTEIN: Dr. Maisel, can I ask you one question. Does the Panel have any concerns about any other specific safety items? You mentioned pancreatitis and renal insufficiency. Dr. Brott was getting at possible recardiac events that occurred secondary. Does this Panel have any other items that particularly stuck out at them or will stick out for them?

CHAIRMAN MAISEL: I think certainly anything related to catheter insertion, so bleeding vascular complications would certainly be relevant.

DR. KATO: The traditional ones, thrombosis, bleeding, you know coagulopathy.

DR. BRINKER: Infection, since that's a high overall rate.

DR. KATO: Oh, infections, sure.

DR. SOMBERG: Sepsis was something that came out specifically from the cooling system, although not in the latest studies with this device.

DR. KATO: I mean there are, obviously, a lot of mechanical issues that go along with this catheter. You know if the water pressure builds up, balloon rupture, inadvertent administration of the fluid into the patient. You know, the usual mechanical things that would be associated with catheter designs such as this.

CHAIRMAN MAISEL: Number two.

EXECUTIVE SECRETARY WOOD: Please discuss whether you believe the data provides reasonable assurance of effectiveness for the proposed indication. In your discussion, please specifically comment on whether the issue of nonrandomized data was adequately addressed by the propensity analysis?

CHAIRMAN MAISEL: So many of our comments in earlier discussion have addressed this topic. I think many panelist have issues with interpreting the data. The issue of effectiveness, I don't know that we have the data to establish that endovascular cooling is equivalent to surface cooling as was presented in many of the New England Journal studies, etcetera.

So I think that my sense is the consensus is that nonrandomized data is not adequate.

I would invite other panelists to comment as well.

DR. SOMBERG: Well, I'd just make the same point earlier for safety, is that it is conceivable and we have -- I think you've said this yourself, Bill. That sometimes concomitant controls that are comparable could help establish that point as well. So I don't think we have to necessarily demand, although it's preferable to have a randomized placebo controlled trial.

DR. MARLER: Well, I guess in response to that, I think that the labor and the effort that goes into doing a registry or a nonrandomized study of sufficient quality to approach being convincing is not that much different from actually introducing one more step of randomization. And I think a clinical trial, I haven't heard anything today that convinces me that it's not possible to do a trial. It may be difficult, but most all of them all. And a registry well done is difficult as well.

CHAIRMAN MAISEL: I would also comment that I think they did about as good a job as you can do with a registry and the problems that we're dealing with are just what registries are about, which is they're not randomized and it's difficult to account for what you don't know.

Anyone else have any comments on effectiveness? I'm not sure we've absolutely gotten a flavor for -- let me ask this question. Are there Panel members who agree with John that randomized trials are not necessarily necessary?

DR. SOMBERG: I'm not sure I said that.

CHAIRMAN MAISEL: Didn't you say that?

DR. SOMBERG: I said requisite, required.

CHAIRMAN MAISEL: Well, that's what we're asking. We're trying to give guidance about whether they must do a randomized trial or whether there is another way out.

DR. SOMBERG: Okay.

CHAIRMAN MAISEL: Norm? Or Mike, go ahead.

DR. WEISFELDT: There are devices that have been -- treatments that have been approved by the FDA without a randomized trial. I mean automatic defibrillators, pacemakers; a lot of them.

CHAIRMAN MAISEL: Right. So it's specifically for this device, what nonrandomized information would you need to approve the device?

DR. BLUMENSTEIN: This is Brent Blumenstein.

A much larger effect size.

DR. WEISFELDT: Yes, right. You know, uniform survival. People are unconscious an hour after a cardiac arrest. That would do it, wouldn't it?

CHAIRMAN MAISEL: Okay.

DR. YUSTEIN: Dr. Page?

DR. PAGE: Let me just be clear. Are we talking about -- because my comments in, maybe my pessimism about having the great randomized trial and having to just deal with the data we have, my comments in terms of safety and efficacy were having to do with hypothermia conceptually. Are we now talking about the device?

CHAIRMAN MAISEL: Yes. I'm trying to be pragmatic, and we're going to finish this Panel meeting and the FDA is going to need to give advice to the sponsor about if we decide that we're not going to approve it, not that we're voting. They're going to need to give very specific recommendations about the least amount of data that is going to need to be presented to reach the bar that we're setting. So I'm trying to understand.

We certainly all would love to have a randomized trial here. I'm asking whether we absolutely need an randomized trial or whether there is some other form that is easier to achieve that would satisfy us.

DR. YUSTEIN: Dr. Page, all these questions are written with respect to this particular advice and the data that you've seen before you. If you need to take other things into account in order to answer those questions, that's fine. But all these questions are written specifically for this product or this device for this application.

DR. PAGE: Thank you.

DR. BRINKER: I think we should also be aware of the fact that sometimes even if you had a large number of patients in a nonrandomized situation, you'll find yourself no closer to the answer than you are now. So, for instance, what if they did a 100 patient trial, registry and found that the incidence of renal failure measured by a half of milligram percent elevation in creatinine was 6 percent? What would we say at the end of that line unless we had a comparison?

DR. BROTT: One thing. It may not be that a randomized trial would have to be strictly one-to-one randomization if there were two arms. I think some of our safety concerns could be compatible with the different randomization ratio that would provide a panel in the future the information that they would need.

CHAIRMAN MAISEL: Henry?

DR. HALPERIN: I'm still a little perplexed on the safety issue. Because I'm still trying to get it clear in my own mind what the differences in the post-cardiac arrest state are from just using the device for its other already approved indications.

For instance, all the mechanical issues with the catheter in terms of catheter integrity and replacement, it's hard to believe there's a difference in the post-cardiac arrest state than there is in general use of the catheter. However, with the renal failure there actually might be differences because the added insult of ischemia during cardiac arrest might be a covariable that would not necessarily be taken into account by just general use of the device.

So I guess part of the question, and we can still ask the sponsor's questions, but is there any renal failure data on just inducing hypothermia in general with this catheter? And is it any different in the cardiac arrest state?

DR. COLLINS: There's no randomized controlled trial on that question.

In terms of the effects of hypothermia on renal clearance of that level of creatinine, no. We have examined vessel on post-mortem samples and not seen any obvious indication that there's a problem. And within the normothermia treatment in a controlled trial where the catheter is too cold, but often there was a small subgroup -- most of them are actually -- there was no effect there, although there were daily bloods drawn. So it seems to be isolated.

DR. HALPERIN: The other comment is is that I think we should all say or realize once again that these trials are very difficult to do and this particular disease has a very high lethality. So I'm not certain that this scrutiny should be exactly identical to every other situations because of the enormous lethality of this particular disease.

CHAIRMAN MAISEL: Bill? John?

DR. SOMBERG: With that said, I think it's important to measure the risk versus the benefit. First you're going to introduce cooling to patients. We would like to know that cooling works. And then you're going to introduce cooling by the catheter, and there are modalities of cooling. Now we've heard that some of them are inconvenient. But what happens in a controlled trial and it could be a randomized or nonrandomized and we can discuss that, but in a controlled trial there was cost associated with this device. And I'm not saying an economic cost. But a safety cost. You might say that maybe it's better to go back to basics and have the interns and residents carry the ice bags.

And I'm not demanding a large randomized trial with a very sensitive. But this committee has for many lesser devices asked for some sort of demonstration of the risk; and that's the third question they're going to ask in a minute. Is the risk/benefit ratio. And just because something works, doesn't mean you're going to use because it may have tremendous risk. And because it has risk, it may not work. So I just don't think it's been shown, although the data from this reanalysis is more moving than I first thought. The trouble is I'm still not sure if the control group because of the selection process was a lot sicker, therefore it makes the efficacy look better and the toxicity look less. If you can show me some sort of comparable control group or go out and do that and show that the efficacy is greater and the toxicity is minimal, and I know it's a very lethal disease, then you would say but go ahead and do this. And I think that's the FDA is going to have to make this balancing judgment. And the data is half there, but not sufficient for -- obviously there's a lot of difference here. So it's demonstrating it's not sufficient.

CHAIRMAN MAISEL: Norm?

DR. KATO: I have to agree with Dr. Somberg. And I think that the randomized perspective trial, as difficult as it's going to be to do this and given the lethality of this disease and I don't think anybody's questioning that, but what we've seen recently in some of the more higher profile issues facing the FDA, I mean we've learned okay, even simple drugs like ibuprofen have the same cardiovascular risk as, let's say, as Vioxx or Celebrex which have been -- one of those drugs have now been pulled from the market. But you wouldn't have know that unless you actually did that trial.

So I think the comparison between the catheter and alternative forms of cooling I think is important. I think randomization, some type of clinical trial, some type of rigorous trial is important in order to satisfy not only the Panel, but also the public's demand for safety and efficacy.

CHAIRMAN MAISEL: Why don't we move on to question three.

Geretta?

EXECUTIVE SECRETARY WOOD: Taking into account all pertinent clinical information available as well as your responses to the above questions, please comment on whether you believe the data provides an overall risk/benefit ratio which supports marketing clearance of the device in the United States for the proposed indication.

CHAIRMAN MAISEL: So we're not taking a formal vote today, however I would like to hear all Panel members on this question, and I'll start with Dr. Marler?

DR. MARLER: No. I don't think that the data provided gives an adequate estimate of the overall risk and benefit for the reasons that we've delineated.

CHAIRMAN MAISEL: Dr. Brinker?

DR. BRINKER: I'm optimistic about the device, but I don't think it meets the level of evidence that I feel comfortable with approving myself at this time.

CHAIRMAN MAISEL: Dr. Page?

DR. PAGE: And I think we're all wrestling with the same problem with the data.

Looking at it from a slightly different perspective, I tip over just slightly to saying yes.

CHAIRMAN MAISEL: Dr. Brott?

DR. BROTT: I'm optimistic about the device as well. And the data that we've looked at today is not new. And physicians in this country have not accepted it for the reasons that we've proposed. And I think that this study that has the rigorous study that's been proposed might not just help the Panel, but if the therapy is in fact effective with this device and shown to be that, I think our peers will be much more likely to accept it and use it.

CHAIRMAN MAISEL: Dr. Somberg?

DR. SOMBERG: I'm not yet there but moving in the right direction.

CHAIRMAN MAISEL: Henry?

DR. HALPERIN: I'm very optimistic about hypothermia and I think the data for the device is borderline. And I would give a slight positive bent.

CHAIRMAN MAISEL: Dr. Blumenstein?

DR. BLUMENSTEIN: No.

CHAIRMAN MAISEL: Dr. Kato?

DR. KATO: I would vote no. Again, I think that the data on hypothermia is interesting, but not -- but I really would like to see -- you know, if it's so obvious that this is going to be a benefit, then it should be easy to run the trial and you're going to have obvious positive results.

CHAIRMAN MAISEL: Mike?

DR. WEISFELDT: No, on the count of safety. But whether the safety concern was met, then I would be in favor.

CHAIRMAN MAISEL: I, too, feel that we're moving in the right direction but don't have everything that we need.

So why don't we move on the next question, Geretta?

EXECUTIVE SECRETARY WOOD: If you believe that the data currently submitted is adequate and sufficient to support marketing clearance:

(a) Please comment on what specific elements should be included in the labeling to accurately reflect the risks, benefits, and proper use of the device including any modifications to the proposed:

I. Indications for use statement.

II. Contradictions.

III. Warnings/precautions, and

IV. instructions for use.

For the latter, please comment on what specific rates of cooling, duration of cooling, optimal target temperatures, rewarming rates, and optimal time to initiation of therapy are supported by the data or whether the general treatment guidelines proposed by the sponsor (32 to 34 degree centigrade for 12 to 24 hours) are sufficient for labeling purposes.

((b) Please comment on whether you believe a post-market study should be required and if so, what the critical components and design of that study should be.

CHAIRMAN MAISEL: So it's difficult to comment on the labeling issues without knowing exactly what data or the study design, which will obviously have implications for instructions for use.

My comment about the labeling is that I found it very verbose. I mean, there were dozen of pages before you got to any clinical data regarding the device.

I think it needs to be explicitly stated when surface cooling rather endovascular cooling for hypothermia is used.

I think the term "hypothermia" was used liberally without clarifying what type of hypothermia. So I think that in whatever instructions for use eventually come of this, that will need to be explicitly spelled out.

It's also difficult to comment to comment on a post-market study at this time.

Dr. Yustein, do you have any other comments?

DR. YUSTEIN: Maybe if Dr. Page and Dr. Halperin, since they were more on the favorable risk/benefit ratio item, if they can add any comments to that if they were to see anything as far as the labeling that would help us your view of the risk/benefit question.

DR. PAGE: There's a lot there. And, of course, I'd rather especially since it's not going to have happen immediately, I'd like to think about that a little bit and get back to you.

DR. YUSTEIN: These questions were of course written before we knew what you were going to responding to the first three.

DR. PAGE: Right. Right.

CHAIRMAN MAISEL: Dr. Halperin, do you want to address the --

DR. HALPERIN: I don't really have anything more to say, except that I think that clearly if one were to fall slightly on the positive side, a post-market surveillance of any potential complications would be critical.

CHAIRMAN MAISEL: John?

DR. SOMBERG: I also think that in any labeling there has to be some mention of fluids, sedatives, paralytics, paralyzing agents and such because I don't think that this is a therapy that's really taken out of context. And a physician making this choice or a team making this choice may not have not studied as thoroughly as the expert groups that have been applying it so far. So I think you have to have an little bit of -- and I agree with Dr. Maisel that it has to be a user friendly explanation for it. Because the only reason I read it was that you guys made me do it. But otherwise, you know, like all docs you just read the dose and the PRD, and that's about it and you go for it. And, you know, I hate to think of somebody -- you know, if fluids were essential, if the sedatives were essential and which ones was essential. And the person wasn't suffering. So it's sort of tied in together.

DR. BRINKER: I would hope that if a further study is agreed upon that some of these issues actually would be prospectively addressed in that study to make it less of a guess work as to what kind of ancillary cocktails the patients need and how they're rewarmed, and things of that nature.

CHAIRMAN MAISEL: Next question, Geretta?

EXECUTIVE SECRETARY WOOD: If you do not believe that the data presented today met the threshold for marketing clearance, please discuss what additional type and amount of clinical data would be required to meet this level of assurance. In your discussion, please comment on:

(a) The appropriate endpoints including assessment scales and timing of assessments which should be used to evaluate the effectiveness of endovascular cooling catheters for this indication.

CHAIRMAN MAISEL: Tom, do you want to try to tackle that one?

DR. BROTT: As far as the time, we do have the HACA study which I think was six months. And it's always nice to be able to compare. But with brain injury we do have evidence that with focal brain injury, anyway, measures of three months are almost identical to measures at a year. And so I would say that in terms of time, I think three or six months in terms of the time of the assessment would be appropriate.

I do think that we require an ordinal scale at this point in time. But I would think that the ordinal scale will be more objective if a detailed inventory of activities of daily living is performed first at the time of the blinded evaluation.

Because the trial will not be huge and because Panel members and physician peers need to be persuaded of this concept and this device, I think that secondary endpoints would be beneficial be beneficial, including cognitive endpoints that we've discussed before that would be widely accepted by our peers, such as a standardized test of some sort that's brief in patients who don't tolerate long neurological psychological investigations.

I would not require brain imaging, but the standard of care is shifting, I believe, in my talks and in my institution where we are obtaining more brain imaging in these patients than we were obtaining before. To require brain imaging in a study such as this could be overly burdensome and expensive for the sponsor. But if it were possible to do so, it could supplement the other endpoints.

In the brain imaging literature that we have thus far, there are some surprises. And, again, I think that could be useful.

CHAIRMAN MAISEL: Any other comments regarding endpoints? Jeff?

DR. BRINKER: Well, I think that somewhere in the discussions between the FDA and the sponsor there needs to be resolution of something that we haven't come to resolution with here, and that is after all is said and done is the concept of hypothermia accepted and unquestionable? So that if that's the case, then much of the endpoints that you are suggesting might be overkill and that the real issue is the safety, which would have to be done with a comparator of another cooling device.

So I think conceptually there needs to be some decision and it's not going to be resolved at this Panel meeting about whether the FDA feels that that hypothermia for out-of-hospital cardiac arrest with defib is a done deal.

DR. BROTT: I would agree with those points with regard to the importance of safety.

CHAIRMAN MAISEL: John?

DR. SOMBERG: I just would say the optimum study would be a three-on study where there is a comparative of, say, topical cooling versus a noncool group. I think the experience in Minnesota, the experience in Illinois is such that so few people get this procedure of cooling by the device, that I think there's still room at this juncture -- there may not be room in a few weeks in a few months if people -- if all societies make this the standard of care. And I'd like to hear from some of the experts in resuscitation on that issue.

But the optimum protocol would be to have two comparative groups. And it also might be optimal for the sponsor in that if you only have to treat six patients to show a survival benefit of one, that is very very profound. And if I knew that, you know, I would make a lot of noise in my institution if there was a patient missed that. I mean, ACE inhibitors, ARBs, beta blocks don't really approach those numbers in most instances. And we make a lot of noise about not giving those.

So that could be very impressive or it may work out that in the United States given all the response times, there's no benefit from almost anything. And it'll be troubling to do this sort of thing.

So I would hope that type of ideal study could be done.

CHAIRMAN MAISEL: Now you've touched on questions 5(b) and 5(c). So why don't we put them on the table.

EXECUTIVE SECRETARY WOOD: Okay.

(b) Whether a randomized controlled trial would specifically be required and if so, what the appropriate control groups would be. If not, please comment on what other types of trial design would be adequate.

(c) Whether, due to the potential differences in standard of care between the international community and the United States, data collected in the U.S. would be required.

CHAIRMAN MAISEL: Okay. I think we've discussed 5(b) to a certain degree. And it seems that most of the Panel feels that a randomized controlled trial would be required, although that's not an absolute.

Any one have any other comments regarding the question 5(b) regarding the trial design?

DR. BRINKER: I make a proposal that probably won't be accepted, but I would think that the people that wrote today's presentation should have a discussion with the people responsible for the writing of the AHA and ILCOR guidelines and see if there's something being missed between the diametrically opposed opinions of what the data show. And maybe that could help resolve part of this issue.

CHAIRMAN MAISEL: I'm not sure I understand your comment. Can you --

DR. BRINKER: Well, I think the analysis of the same data by the FDA suggests that there's no assurance of efficacy with hypothermia in general. On the other hand, the same data while interpreted by people with expertise in this field suggests that that has been demonstrated to a degree in which no other study is necessary.

And the points brought out by the FDA seemed very convincing as to whether this data should stand or not. And I'd like to know maybe they're missing something that these experts have resolved and maybe it would make our job and their job a lot easier to have that kind of interaction.

CHAIRMAN MAISEL: I understand your point. Now I think it's an excellent point. I think clearly the bar by which the FDA measures things is different than the international community than the AHA and they're asking different questions and answering different questions.

I do think that the proposal, actually, is a good one of having a meeting similar to what has been done with any other topics where representatives of the AHA, ILCOR, industry, the FDA could get together and discuss in an open public forum some of these issues that we've struggled with.

DR. YUSTEIN: Dr. Maisel, can I ask just one question? Throughout the discussion today various different control groups were suggested. People suggested endovascular versus surface, endovascular versus normothermia and then also some people have actually even brought up the topic of a three-armed with each of those.

Does the Panel feel strongly or suggest one over the other?

DR. BROTT: I would just say that, and this is repetition, that I believe that many IRBs would allow a three-arm trial to go forward. And as Dr. Somberg mentioned, and as the sponsor has presented, the NNT for this therapy in some of these trials was suggested to be less than ten. And if that's the case, not only would such a three-arm trial be very persuasive to the FDA, but also the American public and our peers in intensive care units and in cardiology.

DR. PAGE: And I'd agree. It would be great to have a three-arm trial. My only concern is whether it can or will be done.

The second best would be an active comparator with just the external and internal cooling, I suppose.

And somehow fitting in the ongoing question of duration and depth of cooling. And then suddenly the trial gets a little bit bigger, but if we had the three-arm trial, we'd be done already.

DR. HALPERIN: But if this would be a three-arm trial and powered enough to show relatively small potential incidences of complications, it's going to be an enormous trial. And you know, an enormous difficult trial of a very lethal disease is going to take a long time.

DR. SOMBERG: The FDA is -- well, I should talk for the FDA. I apologize.

My study of the subject is that they've never asked to power a trial toxicity. You're powering the trial for the predominant efficacy endpoint. And there's a very potent -- as your research and as people have suggested say. But what you do is at least have a sample size to see if there are any obvious differences. And sometimes surprises come out.

So we don't want to see something that comes one in 10,000 or one in 1,000. But you want to see a problem that would come in maybe one in 50. So if there's a host of different problems with the catheter, nothing with topical cooling. And topical cooling is better. That's not necessarily the reasons to withhold approval for the device. But it's something that the general medical community can know about and make risk/benefit decisions.

DR. HALPERIN: AHA is very, very enthusiastic about working with the FDA on this as well as other issues. It's actually come from the AHA side as well.

Also when AHA reviews stuff, animal data and cellular data can be a major part of that which could in some ways explain some of the differences and points of view on this subject.

DR. MARLER: I guess I'd like to add that I think -- I don't know, I'm always a person who wants to rip the bandage off quickly; it hurts less long.

I think that by encouraging the sponsor and the community to think that there's some nonrandomized easy way to get around answering the question is to do a disservice to the patients and the community. I think that if you pitch in, get this trial done, that you'll have the answer that you want, particularly if your estimates of the effect of this are accurate or anywhere close to accurate. This is not going to be this large extremely expensive trial that you seem to be imagining. And I don't think you're going to have to convince the community until this actual comparison is made.

The three-arm trial, I mean is nice because it tests two different modalities. But I think the basic trial that you may not be able to do in five years, which you still won't have the convincing data to look at if you continue with registries or some nontrial methods. They may become progressively more difficult to perform. But right now I think you have the opportunity to have a randomized controlled trial where you have essentially a placebo arm.

CHAIRMAN MAISEL: And can we comment on question (c) the differences in standard of care between the international community and the United States and is U.S. data required?

DR. BRINKER: Yes, I think that that's been addressed. I think because of differences in rapidity with which people get resuscitated as well as different forms of up front therapy that's administered to patients, I think U.S. data is necessary for this.

CHAIRMAN MAISEL: Norm?

DR. KATO: Yes, I would agree with that. I think that the chain of survival parameters can be different.

The other advantage of having multiple centers in the United States is pretty obvious. I mean, you get thought-leaders at various institutions trying this device, drug out, whatever it is protocol. If it works great, then you have that many more people on board when you're facing the FDA again.

DR. MARLER: I don't think it's easy to generalize about international centers, having just visited the center in Helsinki a couple of weeks ago where the door-to-needle time for acute stroke is 12 minutes they're working on right now.

I think you could probably include some international centers if you set up criteria. I think you should probably also be sure that -- there may be more variability I'm saying between U.S. centers than there are between high quality U.S. centers and high quality international centers.

DR. YUSTEIN: Yes. I think this question wasn't meant to say do all the sites need to be in the U.S., but would U.S. data be required in addition to any international sites.

DR. SOMBERG: But it's not really U.S. versus Europe or U.S. versus anyplace else. It's really comparability.

DR. YUSTEIN: Right.

DR. SOMBERG: And generally speaking there's a very long latency between downtime, resuscitation and resuscitation to intervention. And that may play a role in the efficacy of the therapy.

So if you took not good centers of Europe, but bad centers in Europe and good centers potentially given our logistics and all, you may get a better comparability.

I'm saying you need some information about the times. And maybe you recommendation might say is that if you resuscitate in such a time and if you're able to cool them in such a -- you know, like tPA, you have a three hour window, you're able to do that fine. But if someone comes to you and the resident gets an idea 24 hours later, let's cool this person. This is totally inappropriate and only doing harm. So I think that's maybe in the labeling.

CHAIRMAN MAISEL: Any other Panel comments?

Dr. Yustein, any unanswered questions from the FDA questions.

DR. YUSTEIN: No.

CHAIRMAN MAISEL: Okay. So at this point we'll move on to the open public hearing session for the afternoon. Is there anyone who wishes to address the Panel in the audience.

Seeing none, we will close the open public hearing.

I'll once again ask Mr. Yustein any other questions from the FDA?

DR. YUSTEIN: No. I think the team is satisfied. Thank you.

CHAIRMAN MAISEL: Does the sponsor have any additional comments or questions that they would like to address to the Panel?

DR. COLLINS: No.

CHAIRMAN MAISEL: I'd like to ask Michael Morton, our industry representative to comment.

MR. MORTON: Thank you. No comments about the particular sponsor that we've heard from today. But just in general, it's important that the industry and the FDA work together on utilizing the 510(k) process to bring devices to patients. And I certainly appreciate the attention that the Panel has given today and the focus on the 510(k) process, because it's not typical of what this Panel reviews.

Also, appreciate the efforts of the sponsor and the fact that there were physicians who traveled trans-Atlantic flights to come here.

And also note that the FDA did a very concise and good review today.

Thanks.

CHAIRMAN MAISEL: Thank you.

And I'll ask our consumer rep Linda Mottle to comment, please.

MS. MOTTLE: Thank you, Dr. Maisel.

I concur that it would be wonderful to have a device that will increase survivability and neurological outcome for this devastating condition.

I also concur with various Panel members that we don't quite have the data yet to prove that to us. But I really hope that the sponsor will endeavor to conduct these clinical trials with the input from the FDA and the Panel here today so that we can move forward in this arena.

And having worked with IRBs in acute emergency type clinical trials, I don't see it as the impediment that has been presented. And I do think that, hopefully, the FDA will be able to get that 2000 ER research guidance revision done soon to facilitate that process.

I would also like to see a little bit more instruction in the operational manuals. Coming from a clinician standpoint I saw some inconsistencies in that at one point it says 5-cc irrigation syringe and other 10. I mean, just little inconsistencies like that. And I would hope that the FDA would look that over.

Thank you.

CHAIRMAN MAISEL: Thank you very much.

Are there any other comments from any of the Panel members? John?

DR. SOMBERG: I just have one more thing to add, and I'm not sure this is something the FDA or the Panel can fix. But the burden is very high for the sponsor because if they do all these studies we recommend, do everything well, the next sponsor will have a much easier pathway. And therefore, the incentive for them to bear or their role just to be there just a hair ahead of someone else is very difficult.

So I think to do a studies like we're recommending, and that's one of the things that I agonized with when I review this, you know, it's not a patent, it's not a unique approach. Cooling, there are many ways to do that, etcetera. So how do you incentivize the process. And I heard that the NIH is working this area has developed these centers. And I think it's most important to try to do something like with advocating piggyback on a larger project in some way. Because I do not think that all in the meetings that the FDA has with this sponsor, would be able to overcome some of these problems of the lack of incentive to do the proper pathway to get a whole series of which devices, only one of which the sponsor has renumeration from.

It's a real conundrum.

CHAIRMAN MAISEL: Additional comments?

DR. YUSTEIN: Dr. Maisel, before you close out, I just wanted to thank all the Panel members for assembling today on relatively short notice. Usually this is much shorter than our usual pre-panel preparation.

And also Dr. Blumenstein for being able to fill in for our statisticians on a very very short notice. We appreciate that.

CHAIRMAN MAISEL: Does the sponsor have any final comments?

Okay. This concludes the -- oh, I'm sorry. Go ahead.

DR. COLLINS: Just to say thank you.

CHAIRMAN MAISEL: You're welcome.

This concludes the recommendations of the Panel regarding the Alsius Corporation CoolGard 3000 Icy Catheter System K040429.

We are adjourned.

(Whereupon, at 2:33 p.m. the meeting was adjourned.)