UNITED STATES OF AMERICA
FOOD AND DRUG ADMINISTRATION
+ + + + +
CENTER FOR DEVICES AND RADIOLOGICAL HEALTH
MEDICAL DEVICES ADVISORY COMMITTEE
+ + + + +
CIRCULATORY SYSTEM DEVICES PANEL
+ + + + +
+ + + + +
SEPTEMBER 21, 2004
+ + + + +
The Panel met at 9:00 a.m. in Salons A, B, and C of the Hilton Gaithersburg Washington, D.C./North, 620 Perry Parkway, Gaithersburg, Maryland, Dr. William H. Maisel, Acting Chairperson, presiding.
WILLIAM H. MAISEL, M.D., M.P.H., Acting Chairperson
LANCE BECKER, M.D., Consultant
JEFFREY A. BRINKER, M.D., Consultant
THOMAS G. BROTT, M.D., Consultant
ALFRED HALLSTROM, Ph.D., Consultant
HENRY HALPERIN, M.D., Consultant
NORMAN S. KATO, M.D., Consultant
JOHN MARLER, M.D., Consultant
MICHAEL C. MORTON, Industry Representative
LINDA MOTTLE, Consumer Representative
SHARON-LISE NORMAND, Ph.D., Voting Member
JOSEPH P. ORNATO, M.D., Consultant
JOHN C. SOMBERG, M.D., Consultant
JUDAH Z. WEINBERGER, M.D., Ph.D., Consultant
MYRON WEISFELDT, M.D., Consultant
CLYDE YANCY, M.D., Consultant
GERETTA WOOD, Executive Secretary
BRAM ZUCKERMAN, M.D., Cardiovascular Devices
RANDALL BROCKMAN, M.D.
RICHARD FELTEN, M.S.
ELISA HARVEY, D.V.M.
RONALD M. LAZAR, Ph.D., Consultant
NEIL OGDEN, M.S.
JULIE A. SWAIN, M.D., Consultant
ELIZABETH J. TRITSCHLER, M.S.E.
CELIA WITTEN, Ph.D., M.D., General Restorative and
I. Call to Order
‑‑ Comments by Dr. William Maisel 4
‑‑ Conflict of Interest Statement 4
‑‑ Introduction of Panel Members 7
II. FDA Presentation - Data Requirements
for CPR Devices
‑‑ Randall Brockman 9
‑‑ Elizabeth Tritschler 20
‑‑ Ronald Lazar 27
‑‑ Elisa Harvey 38
‑‑ Questions 46
III. Open Public Session
‑‑ Kenneth Collins 71
‑‑ Keith Lurie 77
IV. Open Committee Discussion
‑‑ FDA Questions and Panel 91
V. FDA Presentation - Hypothermia 176
‑‑ Questions 212
VI. Open Public Session 216
VII. Open Committee Discussion 217
‑‑ FDA Questions and Panel
ACTING CHAIRPERSON MAISEL: Good morning. Why don't we get started?
I'd like to call to order this meeting of the Circulatory System Devices Panel. Today's topic is discussion of the type of data required to effectively evaluate the performance of CPR and hypothermia devices.
And I'll ask Geretta Wood to read the conflict of interest statement, please.
MS. WOOD: The following announcement addresses conflict of interest issues associated with this meeting and is made a part of the record to preclude even the appearance of an impropriety. To determine if any conflict existed, the agency reviewed the submitted agenda and all financial interests reported by the committee participants.
The conflict of interest statutes prohibit special government employees from participating in matters that could affect their or their employers' financial interests. However, the agency has determined that participation of certain members and consultants, the need for whose services outweighs the potential conflict of interest involved, is in the best interest of the government.
Therefore, waivers have been granted for Drs. Lance Becker, Alfred Hallstrom, Normand Kato, Joseph Ornato, Judah Weinberger, and Clyde Yancy, for their interest in firms that could potentially be affected by the Panel's recommendations. The waivers allow these individuals to participate fully in today's deliberations.
A limited waiver has been granted to Dr. Henry Halperin for his interest in a firm. The limited waiver allows him to participate in the review and discussion, but excludes him from voting. A copy of the waivers may be obtained from the agency's Freedom of Information Office, Room 12A-15, of the Parklawn Building.
We would like to note for the record that the agency took into consideration other matters regarding Drs. Becker, Brinker, Halperin, Ornato, and Yancy. These panelists reported past or current interest involving firms at issue, but in matters that are unrelated to today's agenda.
Drs. Weisfeldt and Halperin also reported past and or current interest in firms at issue. The agency has determined that these individuals may participate in the Panel discussions.
The agency also would like to note that Dr. William Maisel has consented to serve as the Chair for the duration of this meeting.
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 firms whose products they may wish to comment upon.
ACTING CHAIRPERSON MAISEL: I'd now like to have the Panel members introduce themselves. I'm William Maisel, a Cardiologist at Brigham & Women's Hospital.
And why don't we start with Dr. Zuckerman on my left, please.
DR. ZUCKERMAN: Bram Zuckerman, Director, FDA Division of Cardiovascular Devices.
DR. BECKER: I'm Lance Becker, an Emergency Medicine Physician, at the University of Chicago.
DR. HALPERIN: Henry Halperin. I'm a Clinical Electrophysiologist at Johns Hopkins.
DR. WEISFELDT: I'm Myron Weisfeldt. I'm Chair of the Department of Medicine at Johns Hopkins.
DR. BROTT: Tom Brott. I'm a Neurologist at Mayo Clinic.
MR. MARLER: John Marler. I'm a Neurologist at the National Institute of Neurological Disorders and Stroke, NIH, and head the Clinical Trial Group there.
DR. HALLSTROM: Al Hallstrom. I'm a Professor of Biostatistics at the University of Washington.
DR. KATO: Norman Kato, Cardiac and Thoracic Surgery, Encino, California.
MS. WOOD: Geretta Wood, Executive Secretary.
DR. ORNATO: Dr. Joe Ornato, Chairman of Emergency Medicine and also a Cardiologist at Virginia Commonwealth University Medical Center, Richmond, Virginia.
DR. NORMAND: Thanks. I'm Sharon-Lise Normand, Professor of Health Care Policy and Biostatistics at Harvard Medical School and Harvard School of Public Health.
DR. SOMBERG: I'm John Somberg, Professor of Medicine and Pharmacology at Rush University in Chicago.
DR. BRINKER: Jeff Brinker, Johns Hopkins.
DR. YANCY: Clyde Yancy, UT Southwestern, and Cardiologist and Professor of Medicine.
DR. WEINBERGER: Judah Weinberger, Director of Interventional Cardiology, Columbia, New York.
MS. MOTTLE: Linda Mottle, Director of Clinical Research Program, Gateway College in Phoenix, Consumer Rep.
MR. MORTON: Michael Morton. I'm the Industry Rep. I'm employed by Sorin Group.
ACTING CHAIRPERSON MAISEL: Thank you.
At this point, I'd like to invite the FDA to give their presentation.
DR. BROCKMAN: Good morning. I'd like to outline the order of the FDA's presentations this morning. I'm Randall Brockman. I'll give a brief clinical history of CPR devices. Elizabeth Tritschler will provide a regulatory history of CPR devices. Dr. Ronald Lazar will discuss neural events and outcomes in cardiac arrest clinical trials. And Dr. Elisa Harvey will discuss exception from informed consent in CPR device trials. Geretta Wood will then present the FDA's questions to the panel.
Well, I'm Randy Brockman. I'm a Cardiac Electrophysiologist with the FDA. I'd like to address some important issues in clinical trial design for new CPR devices, and I'd like to provide a clinical summary of the history of CPR and its devices to assist with the first goal. There will be a separate session on post-arrest hypothermia this afternoon.
Well, there's ample evidence of the important impact of chain of survival function on survival of out-of-hospital cardiac arrest. Early defibrillation, in particular, has emerged as an important intervention. We've seen numerous interventions at various points along this chain.
And while some have resulted in improvements in short-term success, such as return of spontaneous circulation and short-term survival, a few interventions have resulted in improvements in hospital discharge and improvements in neurologic outcome.
It'll be important for future trials to evaluate appropriate success endpoints. How should we define those endpoints? Should a study of a new investigational device have to demonstrate improvement in hospital discharge rates and neurologic outcome when this encompasses the entire chain of survival? Alternatively, could such trials be designed to evaluate a short-term endpoint with additional trials adding to the database?
Following is a brief history of some of the published treatment interventions in cardiac arrest. My goal is to highlight the results of these reports and to use these results as a framework to decide on appropriate endpoints for future CPR trials.
Just from a historical perspective, resuscitation of patients who have experienced a cardiopulmonary arrest has been attempted for over a century. In the 1950s, Safar and Elam sort of rediscovered, if you will, a mouth-to-mouth ventilation by reading how midwives use the technique to revive newborns. But until 1960 no successful resuscitation was limited to victims of respiratory arrest.
In 1960, Kouwenhoven described that forceful chest compressions, closed chest cardiac massage, produced respectable arterial pulses. Combined, these two techniques form the critical steps of modern CPR, and they've been practiced for more than 40 years.
The success rates following in-hospital cardiac arrest have remained essentially unchanged over the last three to four decades, with return of spontaneous circulation in about 30 percent of patients and approximately 15 percent of patients being discharged neurologically intact.
In a randomized control trial of in-hospital cardiac arrest, interposed abdominal counterpulsation demonstrated improvement in the rate of return of spontaneous circulation with about 51 percent in the IAC group versus about 27 percent in the standard CPR group.
At-hospital discharge, a significantly greater proportion of patients was alive in the IAC group versus the hospital discharge ‑‑ excuse me, versus the standard CPR group. That was 25 percent versus about 7 percent.
The rate of patients discharged neurologically intact was not statistically significantly different in the IAC CPR group compared to the standard CPR group. While there was a trend, it was 17 percent versus 6 percent. That was not statistically significantly different.
Patients who suffer an out-of-hospital cardiac arrest have even worse outcomes than those who are resuscitated in the hospital, with hospital admission rates between 8 and 22 percent, and between 1 and 8 percent being discharged neurologically intact.
This has been largely unchanged despite additions to the basic components of CPR, such as high-dose epinephrine, transcutaneous pacing, and vest CPR. Techniques such as active compression-decompression CPR, with or without inspiratory impedance threshold devices, have demonstrated mixed findings. And AEDs have demonstrated improved survival.
I'm going to briefly go through some of this data. In one study of high-dose epinephrine ‑‑ this was an unblinded, randomized control trial of over 3,300 patients ‑‑ high-dose epinephrine compared to standard-dose epinephrine resulted in a higher rate of return of spontaneous circulation, about 40 percent versus about 36 percent, and survival to hospital admission about 26 percent versus about 23 percent. But there was no difference in the rate of survival to hospital discharge or neurologic status.
In two other trials, both double-blinded, randomized control trials, totalling over 1,900 patients, high-dose epinephrine failed to demonstrate any substantial improvement in neurologic outcome or survival.
Vest CPR includes a pneumatically-cycled, circumferential, thoracic vest system, which is used to augment intrathoracic pressure during CPR. In a small, unblinded, randomized control trial ‑‑ this was in-hospital cardiac arrest ‑‑ there was a trend towards increase in return of spontaneous circulation and 24-hour survival, but there was no difference in survival to hospital discharge.
And then, in an unblinded, concurrent controlled trial, which evaluated the effect of transcutaneous pacing and out-of-hospital asystolic cardiac arrest, no improvement was found in the rate of survival to hospital admission or the rate of survival to hospital discharge.
Active compression-decompression CPR uses a suction-like device applied to the sternum to allow active chestwall decompression in order to enhance negative intrathoracic pressure during the decompression phase. The goal is to enhance venous blood return.
Active compression-decompression CPR compared to standard CPR has demonstrated mixed findings. Two studies ‑‑ by the way, the numbers here correlate to my references in the Panel pack. Two studies, both unblinded, group crossover control trials of out-of-hospital cardiac arrest, totalling over 1,400 patients, demonstrated no difference in survival to hospital admission, survival to hospital discharge, or neurologic outcomes.
However, a different study ‑‑ this was an outside U.S., unblinded, parallel group crossover design, with 750 victims of out-of-hospital cardiac arrest. The study compared ACD-CPR to standard CPR, and demonstrated an improvement in return of spontaneous circulation. It was about 45 percent versus 30 percent in the standard CPR group.
Improvement in 24-hour survival was 26 percent versus about 13 percent, and hospital discharge without neurologic impairment ‑‑ and this was about 5-1/2 percent versus 2 percent. This latter study, which demonstrated improved outcomes, differed from the other two in that a physician was present on the scene of the arrest to guide ACLS therapy. And, in addition, the EMS personnel involved had been using the ACD-CPR techniques for several years, raising the possibility of a learning curve effect.
Inspiratory impedance threshold devices have been combined with ACD-CPR devices. Inspiratory impedance threshold devices are designed to help maintain the increased negative intrathoracic pressure generated during active decompression in order to augment venous return.
Comparing ACD-CPR plus the ITD, the inspiratory impedance threshold device, to standard CPR, two studies ‑‑ both were randomized control trials involving over 600 patients ‑‑ demonstrated these devices to increase the 24-hour survival rates. In the first trial it was 37 percent versus about 22 percent. In the second trial it was 32 percent versus 22 percent. But were not found to change the survival to hospital discharge rates.
In the first trial it was 18 versus 13 percent. In the second one it was 5 versus 4. The first trial excluded subjects for whom the known time from collapse to initiation of CPR was greater than 15 minutes. The second one excluded patients for whom the known time from collapse to initiation of CPR was greater than 30 minutes. I think that difference likely explains the difference in survival to hospital discharge rates.
I present most of this just to demonstrate the notion that short-term survival does not necessarily predict long-term survival.
On the other hand, AED seemed to improve more than short-term survival and out-of-hospital cardiac arrest. The two trials I present here are both single-arm, unblinded trials of out-of-hospital arrest. The first one is the CASINO study, and when compared to published survival rates patients who received early defibrillation from an AED had an improved survival-to-hospital discharge of about 53 percent for VF arrest patients and 38 percent for all-cause arrest patients.
And then, in the second trial ‑‑ this is the long-term outcomes of out-of-hospital cardiac arrest after successful early defibrillation with an AED study ‑‑ when compared to published rates of about 1 to 8 percent, there was an improvement in the rate of hospital discharge with intact neurologic function of 40 percent. I'd note this trial evaluated VF arrest only, and the published rates are for all cardiac arrest.
A recently-published study of public access defibrillation, the PAD trial, demonstrated improvement in survival to hospital discharge ‑‑ 23 percent versus 14 percent for the standard CPR. The survivors had similar functional status.
So, in summary, survival rates with intact neurologic function have changed little over the past 30 to 40 years. Recent medical devices, such as AEDs and possibly ACD-CPR, plus or minus the impedance threshold devices, appear to be capable of having an impact. Choosing appropriate endpoints for clinical trials will be important to determine which devices will facilitate improvement in long-term outcome.
Will additional improvements in the chain of survival also lead to additional quality of life benefit in those who survive cardiac arrest? And, more importantly, can we accept short-term improvement survival as a marker for long-term improvement?
Conversely, in light of the chain of survival concept, is it reasonable to expect an individual medical device to lead to long-term improvement, or can we accept improvements in each step along the chain with the ultimate goal of improving long-term outcomes when each step along the chain is strengthened?
And, finally, fostering an environment to enhance clinical research in this field will be important.
Thank you. And now I'd like to introduce Elizabeth Tritschler, who will give you a regulatory history of CPR devices.
MS. TRITSCHLER: Hi. My name is Elizabeth Tritschler, and I'm an Engineering Reviewer in the Division of Cardiovascular Devices. Today I will brief you on the regulatory history of CPR devices.
The regulation of CPR devices can be broken down into three categories. The first category, which has been regulated since the 1970s, contains devices that mechanically assist the rescuer in chest compressions. Then, we have a new generation of devices in the 1980s, and these devices provide the rescuer with feedback regarding the compression depth and frequency.
And the 1990s brought a third generation of CPR devices, which are significantly different than the first two generations in that they are intended to enhance CPR hemodynamics. Now that we've seen this overview of the three types of devices that we have reviewed, I'm going to go into details about how the FDA has reviewed these types of devices.
And, first, I will start with external cardiac compression devices. The Medical Devices Amendment was passed in 1976, and a few months later we saw the first marketing clearance for an external cardiac compression device. Many more submissions for external cardiac compression devices have been reviewed and cleared for marketing since 1976. These devices all contain some form of chestpiece. Some also contain a backboard.
The manual external cardiac compression devices require the rescuer to determine the rate of compression as in standard CPR. And then we have some external cardiac compression devices that are automated and provide compressions at a fixed rate.
These devices are intended to assist the rescuer by reducing the work required to compress the victim's chest and/or by distributing the compression force more evenly over the sternum. By reducing the work required to compress the victim's chest, these devices reduce the potential for rescuer fatigue.
External cardiac compression devices are Class III products and are reviewed through the 510(k) pre-market notification process in which the sponsor demonstrates substantial equivalence to a pre-amendment or previously cleared predicate device.
Generally, external cardiac compression device submissions do not contain clinical data due to the similarities in design and technological characteristics to predicate devices.
And now we have the second generation of CPR devices. These were introduced a decade later in the 1980s with the first marketing clearance for a cardiopulmonary resuscitation aid device in 1984. CPR aid devices provide audible indicators of compression rate and/or visual indicators of compression depth.
It should be noted that in reviewing these devices the agency has worked with the sponsors to ensure that the device specifications are consistent with the AHA guidelines. These guidelines put out by the American Heart Association suggest appropriate rates and depths of compression for different patient populations.
These devices are designed with force gauges and a corresponding display. However, achieving a specific depth of compression can require different amounts of force in different patients due to variations in patient size and chest wall compliance.
Therefore, the device labeling for CPR aid devices instructs the rescuer to perform the first chest compression per standard CPR ‑‑ in other words, to just eyeball the appropriate chest compression depth. And then, to note the force displayed on the device when the depth is achieved.
Then, for subsequent compressions on that patient, the rescuer can just watch the force gauge to make sure he or she is providing compressions with the appropriate amount of force to compress the patient's chest to the depth specified in the AHA guidelines.
These devices are intended to assist rescuers simply by providing feedback to help them maintain compliance with AHA guidelines for CPR. This feedback is especially helpful to fatigued rescuers who might otherwise be providing weakened compressions without even realizing they're doing so.
Like external cardiac compression devices, CPR aid devices are Class III products and are also regulated through the 510(k) pre-market notification process. Generally, 510(k) submissions for CPR aid devices do not contain clinical data due to the similarities in design and technological characteristics to predicate devices.
And in the early 1990s, we saw the emergence of a third generation of CPR devices ‑‑ those devices intended to enhance CPR hemodynamics. Some examples of these types of devices ‑‑ and some of these Randy has already spoken about ‑‑ include interposed abdominal compression devices, active compression and decompression devices, circumferential chest compression devices, and minimally invasive open chest cardiac massage.
The agency made some precedent-setting regulatory decisions in the early 1990s regarding these devices. First, the agency determined that no pre-amendment or previously-cleared predicate device exists for CPR devices intended to enhance hemodynamics. Second, the agency determined that submissions for devices capable of enhancing CPR hemodynamics would require clinical data to support such claims.
So clinical studies for CPR devices have evaluated various primary and secondary endpoints, such as survival to admission to the ICU, survival to 24 hours, end tidal carbon dioxide, presence of a pulse during CPR, and various neurological evaluations at different time points ranging from 30 days to one year.
And some of these evaluations are based on CPC ‑‑ cerebral performance categories ‑‑ the Glasgow Coma Score, and other quality of life assessments. And Dr. Lazar will be going into more details regarding the neurological endpoint shortly.
On June 29th ‑‑ I know there's a typo in the slide. It should be June 29, 1998 ‑‑ this Panel met regarding a PMA for an active compression and decompression device. The device was intended to increase negative intrathoracic pressure thereby causing increased ventricular filling, increased cardiac output, and increased coronary artery and cerebral circulation.
The Panel recommended the submission be found not approvable due to problems with the clinical data such as lack of randomization at all sites and substantial OUS data used to support success. OUS data is problematic in that the treatment methods and outcomes are affected by variations in the EMS systems in other countries compared to the United States.
And six years later we're here with a meeting of the Circulatory System Panel to discuss CPR devices. And where are we now? Well, we have over 30 cardiac compression devices cleared for marketing. We have a handful of cardiopulmonary resuscitation aid devices cleared for marketing, and there are no devices intended to enhance CPR hemodynamics approved for marketing in the United States.
So today we're asking the help of the Panel in identifying appropriate clinical trial endpoints and a scientifically sound and feasible clinical trial design in order to advance the science and medical therapies for this patient population.
And now Dr. Ron Lazar, who is a neuropsychologist at Columbia University in the Stroke and Critical Care Division, will discuss neurological and functional endpoints.
DR. LAZAR: Thank you. Of the many end organ effects of cardiac arrest, I think few would doubt that the impact on the brain is something of extreme significance. And what I'd like to do this morning is spend a little bit of the time I have talking a little bit about the pathophysiology, the functional impact of cardiac arrest neurologically, and some issues regarding the measurement of cerebella outcomes.
I think to start off the process I think we need to discuss a little bit about the cascade of events as they occur in the brain. This is the ‑‑ a very brief description of what happens during the course of cerebral ischemia. If we start at the left, at the time of the cardiac event, shortly thereafter, in an effort to maintain cerebral blood flow, arterials expand in order to maintain cerebral profusion. And they will continue expanding until at the point they're maximally dilated and are no longer able to expand further.
At this point, cerebral blood flow diminishes, and in order to maintain oxygen metabolism, noted by the line here, the neurones demand increased oxygen, and you have an increase in the oxygen extraction fraction.
At the point that all the oxygen has been extracted from the blood, profusion has dropped, we go from a point of aerobic metabolism of the neurones to anaerobic metabolism. And during this period of time denoted as ischemia, the cell begins to degenerate in a very systematic fashion. And over the course of time enough of the elements degenerate until eventually infarction occurs.
It is this period of time during cerebral ischemia where the critical period for CPR exists. So that the longer we traverse this interval the more extensive and more permanent the injury is going to be.
This is a CAT scan of a case reported in The New England Journal last year of a 50-year-old female with a sudden loss of consciousness with no measurable pulse or blood pressure, and breathing and circulation returned spontaneously reportedly within a few minutes.
And one hour after the ER presentation this scan of the brain on the left shows the early signs of the cerebral injury. But you will note that the sulci are still apparent, and you have fairly normal ventricular size. But after about four hours from this cardiac arrest the ventricles are now compressed, the sulci have been effaced, secondary to the cytotoxic edema arising from infarction. And this patient obviously did very poorly.
Going from the anatomy described in a CAT scan to the physiology in a PET scan, here we see the case of a patient who regained consciousness on ‑‑ and this is day two ‑‑ where at the top we have cerebral blood flow, on the bottom we have oxygen metabolism.
And you can see here that although there is blood flow going to the brain adequate to ordinary support function, because of the cardiac arrest, the blue areas denoted here indicate poor oxygen metabolism. And as a result, the brain is not functioning properly.
So what is the functional impact? Obviously, it varies from mild to severe. And at the mild stages of postanoxic encephalopathy, we have inattentiveness, weakening of judgment, and motor coordination. At a greater level of severity we have memory impairments, apathy, disinhibition, and poor judgment.
And at the severest outcomes in postanoxic encephalopathy, in the otherwise awake patients, you have patients who have language disturbances, inability to recognize their environment, inability to use their hands in purposeful ways, amnestic disturbances, difficulty in calculations, and impaired reasoning.
In the physical spectrum, you have spasticity, paresis, ataxia, pseudobulbar palsy, and other kinds of effects there.
When we get down to this level of dysfunction ‑‑ and these are patients that I, unfortunately, have to see in my own clinical practice ‑‑ for such disabled individuals alive is not necessarily the better alternative.
Well, how do we know some of these outcomes? Well, in a study published by Roine and colleagues in JAMA about a decade ago, they looked at a placebo, controlled, randomized double-blind trial of nimodipine versus placebo, looking at 68 survivors of out-of-hospital cardiac arrest who were evaluated over a two and a half year period.
And they took a look at neurocognitive outcomes in three and 12 months after discharge from the hospital. And by "neurocognitive outcome" I'm referring to functions such as language, memory, cognition, perception, and so forth.
They defined a priority ‑‑ a priori the abnormality as a score at or below the second percentile when compared to the normal population on that particular test, and what they found was the following. There was no statistical difference between nimodipine and placebo groups, which was bad news from the point of view of the clinical trial, but good news in the sense that we could collapse the groups and analyze the total outcomes.
The general intelligence scores were essentially within normal limits. There were no or mild deficits in about half of the survivors at one year. There was relatively mild impact on language and visual perception, and deficits were slightly less frequent at 12 months than they were at three months.
So from the point of view of people in the emergency room, these people were walking, they were talking, and at a very superficial level they seemed to be functioning well. But if you looked at the cognitive outcomes, half the patients had a moderate to severe abnormality in memory, manual dexterity, calculations, skilled motor movement, planning, initiation attention, motivation, and depression. And I'm going to come back to these Roine data later on in another context.
When you look at an MRI scan of a patient who suffered at a hospital cardiac arrest with memory intact versus impaired memory, it's not easily seen on the slide here, but if you compare these two slices of the brain ‑‑ and this is a front view of the brain where this is left, this is right, and this is the top, and this is the bottom ‑‑ you can see the increase in ventricular size. You see cortical loss here in the medial temple region, and you see here the sulcal enlargement.
So the point that was made here in this study was that the cardiac arrest is not a focal problem; it's a whole brain problem.
Now, with all of this knowledge, it's kind of surprising that if you look at the emergency medical literature, most of their neurological outcomes have relied on something called the Cerebral Performance Categories, which I will describe briefly.
The highest level of performance, a good level of performance, involves patients who are conscious, alert, able to work, and lead a normal life. Then, we have minor psychological and neurological deficits. A moderate cerebral performance is a conscious patient who is capable of part-time work in a sheltered environment or independent activities of daily living, with some residual neurological deficits.
Severe cerebral performance are patients - involve patients who are conscious but fully dependent on others for their activities of daily living, coma and vegetative state, and down below there is death.
In most of the cardiac resuscitative literature, an intact neurological patient is one who falls into either of these two categories. And so the question is: how sensitive are these measures to neurological function?
Well, if we take a look, first, at the highest level of outcome, of good cerebral performance, Hsu and her colleagues reported ‑‑ or discussed the fact that the CPC is subjective and its categories are poorly defined. It is frequently used only at hospital discharge, and it has never been validated or compared to other measures.
And so what they did was they compared the CPC with an instrument called the Functional Status Questionnaire at discharge and at followup, and the Functional Status Questionnaire is this well-validated study having been used in a variety of medical environments to take a look at patient outcomes.
And what they found was that a CPC score of one on discharge had a sensitivity of 78 percent, but a specificity of only 43 percent for same or better subjective quality of life than before the arrest. The ability of the CPC to predict abnormal performances on the Functional Status Questionnaire had a sensitivity of only 32 percent and specificity of 43 percent.
And if you looked at the predictive ability of the CPC, the correlation of the CPC at discharge and at followup was only .32.
If you look now at the moderate cerebral performance category, this is now the part-time work. What I did was I took the liberty of taking a glance at the Social Security Act and what constituted someone who is eligible for disability benefits, and found under impaired organic mental disorders that the spheres of disability would occur in activities of daily living, social functioning, concentration, and deterioration in work and work settings.
If we now go back to the Roine data that I presented to you earlier, the 48 percent who have a moderate to severe impairment, they would be eligible for total and complete disability with a CPC score of two. Is this an attack neurological outcome? And I would suggest not.
So based on the existing evidence, the physical ‑‑ physiological, rather, of cerebral inoxia following cardiac arrest is well documented with effects that can be both transient and permanent. And I can also tell you from my own clinical practice that even mild deficits can be permanent.
And, therefore, the teacher in the classroom or the attorney in court or the bond trader on Wall Street or the parent raising children ‑‑ a mild deficit can actually make a difference between competence and futility.
We need an objective, validated measure of brain function that will include physical and cognitive outcomes, and that these outcomes need to be specified in advance with operational definitions that take into consideration contemporary views of neurological function and imaging, and that the clinical performance scales lack the sensitivity and specificity needed to serve this role.
The measurement of brain function in a clinical trial should be performed by clinical neuroscience specialists who are blind to treatment, not the emergency room physician.
And, finally, neural endpoints need to be obtained in the acute period, at discharge, and at longer term followup to ensure meaningful patient outcomes.
Okay. I'm now going to introduce Dr. Elisa Harvey, who will talk about exceptions in informed consent.
DR. HARVEY: Good morning. I'm Elisa Harvey. I'm representing the Investigational Device Exemption staff in ODE at FDA. And I'm here to provide a little bit of an overview regarding the regulations as they currently exist with respect to exception from informed consent for these kinds of device trials.
As we know, informed consent is a fundamental element of all human subject research, and these are outlined through the Declaration of Helsinki and the 1979 Belmont Report, which both identify the basic principles that are a part of informed consent.
It has long been recognized that there is an appropriate place for consent by a legally authorized representative or proxy for patients and populations that are incapable of providing their own informed consent, such as the pediatric population or individuals that are cognitively impaired.
But prior to 1996, there was no mechanism in our regulations for prospectively waiving consent altogether for research. There were case-by-case waivers of consent but not a mechanism for a prospective waiver.
We recognize that there are obviously emergency situations where medical intervention is urgently needed, but the patient is unable to provide consent for whatever reason. And yet the urgency of the situation precludes obtaining consent by proxy, and, in particular, we recognize that research into this kind of area is also urgently needed.
So in order to try and address those issues, in 1996 a new FDA regulation was promulgated, not just for devices but at the agency level for all kinds of trials for emergency research where a waiver of consent might be an important element of the research.
And the regulation was intended to address this need to permit exception from informed consent in very specific circumstances, which I'll go through. It recognized the need, though, that there should be some additional protections of patient's rights when research is undertaken in this fashion without consent.
The regulation was developed with significant input from the medical community through a series of open meetings, and also a draft regulation which was published in 1995 allowing for a comment period after which the final regulation was implemented in 1996.
The regulation is found here in the Code of Federal Regulations, 21 CFR 50.24. It identifies the specific criteria or circumstances for these kinds of studies and establishes the requirements for the study conduct. And it also identifies some additional steps that sponsors and IRBs must take to assure adequate patient protections.
The criteria are as follows. The subjects must be in a life-threatening situation. The current treatments that are available for treatment of that patient are both either unproven and/or unsatisfactory. Participation in the study should hold at least the potential for direct benefit to that individual patient in that circumstance, not just an indirect benefit over the long term to a different population.
And the study could not feasibly be conducted without this exception. And what we mean by "feasibility" is that there would either be too few patients who would be able to provide the consent out of the total population in a study or who would have an acceptable proxy that would be available in the appropriate time period to provide that consent for them.
In addition, it wouldn't be ‑‑ the population must be such that it wouldn't be possible to prospectively identify the population from which those study patients would likely be drawn and able to provide consent ahead of time.
As far as the study conduct goes, the regulation stipulates that investigators must make every attempt to obtain consent from a legal or authorized representative within some specified ‑‑ within the protocol time interval before proceeding to go ahead and enter the patient in a study.
And if consent is not able to be gotten prospectively, the investigators must inform the patient and/or their legally authorized representative about their participation in the study as soon as possible.
The additional protections that are outlined in the regulation are that a separate IDE, or investigational device exemption application, must be submitted to and approved by the FDA ahead of time for all such studies. And the IRBs must consult ‑‑ this is an important aspect of it I know which has been the subject of much discussion, but IRBs must consult with the individual communities where this kind of study would be conducted.
The study must be publicly disclosed to those communities before initiation of the study, and the results must be publicly disclosed when the study is completed, either in the form of publications in peer review journals and potentially also in other venues that are more accessible to the lay population.
The study must be overseen by an Independent Data Safety Monitoring Board. And the IRBs for studies in which multiple study sites are involved must be notified of the concerns raised by other IRBs that are participating in that study, or approving that study.
In order to help assist in the clinical community in understanding what this regulation meant and how to appropriately meet the requirements, an FDA guidance document was issued in the year 2000. Again, this is not just for device trials, but for all kinds of studies involving unapproved products that would be a part of these kinds of emergency research trials.
The guidance can be found at this website, and what it does is attempt to clarify the requirements in the reg. And it was ‑‑ the content of the guidance has been informed by some of the initial experiences that were conducted under this reg following the 1996 publication. There was also a period of public comment for the draft guidance document, which identified the need for some further clarifications, and these clarifications and revisions are currently underway.
As far as experience with the reg since it has been implemented in 1996, the most ‑‑ probably the one that's most cited is the Public Access Defibrillation trial, which was recently published in The New England Journal.
Some investigators have described their approaches to the regulatory requirements in detail, and I think these ‑‑ it's clear that these reports are very helpful in assisting the entire clinical community in developing approaches that are optimal for both the patients and the investigators in getting these studies done.
So as far as the current status of the regulation and these kinds of studies, the draft ‑‑ like I said, the draft guidance is currently being revised to incorporate some of the public comments and provide more clarification.
The past experience that has been out there thus far should facilitate some increased efficiency in some of the future studies that are done in accordance with these requirements. And we should recognize that sponsors, investigators, IRBs, and FDA are all still in learning mode with respect to how to best implement this regulation and make sure we're providing adequate patient protections.
If there are any questions or comments about the regulation or the guidance, I'd be happy to take questions, either now or following the meeting. I can be reached by e-mail or phone, and I'd be happy to take the questions.
And I guess Geretta is going to ‑‑ or Geretta is going to read the questions?
ACTING CHAIRPERSON MAISEL: We'll do the questions later. So I'd like to thank the FDA for their presentation and for providing an excellent foundation for this morning's discussion, and at this point invite the Panel to ask any questions of the FDA, reminding, of course, the Panel that they will have ample opportunity to discuss these issues later.
Yes, Dr. Brott.
DR. BROTT: Dr. Lazar's group at Columbia has a long history of investigation of patients' cognitive function in association with brain injury, particularly with stroke. And I'm wondering if Dr. Lazar could make some suggestions on what neurological endpoints he would either recommend or recommend for study.
DR. LAZAR: I think that's a good question, and I think that, for example, having a neurological physical outcome scales, like the NIH Stroke Scale, for example. Let's separate the physical and the cognitive outcomes. I think that for physical outcomes we can look at things like the NIH Stroke Scale, for example, which has more quantitation than merely observation of what people can do.
I also think that scales like the Barthel and the Modified Rankin Scale can also be used to measure some aspects of the impact of physical disability. With regard to cognitive outcomes, as you know, being a stroke neurologist, that it takes longer to evaluate that. And I think that we could target the nature of the test to the kinds of dysfunction we would expect.
So that there has to be measurement of memory and language and psychomotor speed, and so forth, and that there are really I think well-recognized tests that can be used in a reasonable amount of time to measure these outcomes. It doesn't have to be a five-hour battery of neurocognitive tests. They can be done in a much shorter interval than that. If you want names of specific tests, we can do that also, but I'm not sure this is the venue for that.
Did I answer your question for you?
DR. BROTT: Yes.
ACTING CHAIRPERSON MAISEL: Dr. Yancy.
DR. SOMBERG: I just wanted to followup. Do you have one for him? No.
If you ‑‑ you sort of ‑‑ well, you didn't sort of, you did suggest that the CPC test was inadequate, both in sensitivity and specificity. Could you be specific about what is adequate?
DR. LAZAR: I think that what is adequate is that it's not only what you test, it's also when you test it. And typically the CPC is used at the time of discharge, and there are not many published studies on long-term outcomes with the CPC. And the CPC was never validated against other measurements.
So it would mean that at the time that the patient is admitted that a neurologist, for example, would do an initial thorough neurologic exam and put that into something like the NIH Stroke Scale, or something like that, to measure neurological disability. Cognitive function is not necessarily assessable at that point in time.
As you get to discharge, you repeat the neurological exam with an outcome measure like the NIH Stroke Scale, and then you use measures of word retrieval and of memory and of perception, and so forth, that could be used, let's say, at discharge. And a battery of tests can be anywhere from 45 minutes to an hour, to measure those outcomes. Some patients will do well on them, and some patients will not.
And then, you can measure them in 30 days, and then you can measure them at six months and at one year, and, therefore, get serial measurement of these functions over time. You could also look at other outcomes such as ‑‑ of cerebral blood flow. You could look at Doppler. You could look at diffusion-weighted MRI, which I ‑‑ data I didn't present, as surrogate measures of neurological outcome.
DR. SOMBERG: I would just say that, you know, I hear what you're saying, but in a clinical trial you want to maintain ‑‑ you want to be effective, but you also want to be simple. And you seem to be implying that there's no simple instrument. Having a neurologist spend an hour two or three times with each patient makes ‑‑
DR. LAZAR: Well, I ‑‑
DR. SOMBERG: ‑‑ makes for greater complexity.
DR. LAZAR: Well, I think you're right. Unfortunately, the brain is not a simple organ. And it does ‑‑ it does a lot of things, and it ‑‑ we have physical outcomes, we have cognitive outcomes, and if ‑‑ and you also have to think about the burden to the patient having incurred a cardiac arrest and what happens to them outside and what the costs are to them.
There may be costs of testing these functions, but there's also a cost to patients who are struggling out there who are led to believe that they're surviving well when, in fact, they're not. And I think we need to know what it is that happens to them as a result of intervention, and it's something that's not approached in a trivial kind of way.
DR. ZUCKERMAN: Dr. Somberg, your points are well taken. But I would make the analogy to what we found, for example, with the LVAD development program, where neurological function is a key aspect of the effectiveness that we're trying to determine here.
Again, the key concept is to make the neurological testing user-friendly, and we do abide by those principles. But I do think in terms of the overall effectiveness question we can't forget about Dr. Lazar's points, and certainly we have great neurological input here today.
ACTING CHAIRPERSON MAISEL: Dr. Yancy.
DR. YANCY: Along a different line of questioning for the FDA, I was struck to see over 30 devices that have an approval for the indication for CPR. And I'm wondering if there's any post-marketing data on the outcomes, since those devices have been approved.
In large measure, the methodology is woefully unacceptable, and I would think that we would have some rationale to continue to collect data to see if a learning curve is present, so that outcomes are better, or, if as it's widely distributed, the results are even less good because the operator variability increases.
Are there any post-marketing data?
MS. TRITSCHLER: The short answer is no. And a little bit of explanation along with that is the 30 cleared devices are cleared and not approved. They're cleared through the 510(k) process, and that's different from the PMA process. So we don't have the same ability to request a post-approval study.
DR. YANCY: So how many devices are on the market? Or maybe some of our emergency consultants can tell us that. How many devices ‑‑
MS. TRITSCHLER: I'm sorry. How many of which type of device?
DR. YANCY: How many of the resuscitation devices are actually on the market and being utilized?
MS. TRITSCHLER: There's about ‑‑ probably about 40 total that have 510(k) clearance. I don't know how many are currently marketed, and those devices that have the 510(k) clearance are just intended to assist the rescuer. They aren't intended to enhance any kind of clinical outcome of CPR.
Does that answer ‑‑
DR. YANCY: That does. But I think that one of the things we should consider in any trial design is the requirement for ongoing longitudinal data collection.
ACTING CHAIRPERSON MAISEL: Dr. Brinker.
DR. BRINKER: I was wondering about your category of those CPR devices that bring about hemodynamic improvement. It seems to me that you're not only looking for hemodynamic improvement, because hemodynamic improvement, at least during CPR, would be easily surrogated to relatively simply measured entities.
But what you're really looking for is an endpoint to the hemodynamic improvement. So at the end of the day, aren't you always looking for better survival as a final common denominator rather than hemodynamic improvement during the application of the device?
DR. ZUCKERMAN: Perhaps. I think that's one of the reasons why this Panel has convened. CPR trials, for a variety of reasons, are extremely difficult to perform, and this Panel will deliberate on many aspects of trial design.
But, you know, certainly in an ideal world, perhaps we would like to be able to point to a surrogate that would both correlate and fully capture the endpoint of interest, which you've mentioned, Dr. Brinker.
The real question, though, is: is there a surrogate for the one you've mentioned, or even some endpoint that comes close that could be utilized for trial design in this field?
DR. BRINKER: So, Bram, let's say I had a device that could unequivocally, during CPR, give me higher blood pressure, greater cardiac output, and perhaps even ‑‑ well, let's stop there. But as a subcategory, perhaps increase cerebral blood flow.
And I studied this device, and I confirmed all those observations, yet there was no increase in survival to hospital ‑‑ end of hospitalization, nor neurologically intact survival. Would that device be approved as a ‑‑ because it could deliver hemodynamic improvement over other available devices?
ACTING CHAIRPERSON MAISEL: I'm going to interrupt and simply comment that we'll have plenty of time to discuss the appropriate endpoints and whether ‑‑ you know, we can decide what the appropriate endpoints are.
Dr. Normand, did you have a question?
DR. NORMAND: Yes, I have a question completely unrelated to that. I was wondering if the FDA could comment on the issues with the studies conducted outside the U.S. And, specifically, you mentioned variations in EMS, and I am wondering whether or not data could be collected that one could adjust for such differences and things of that nature.
DR. HARVEY: Well, I'm not sure I specifically can answer that question, but I did want to make a clarification about two aspects of the regulation that I didn't mention before. One has to do with the acceptability of OUS data when it hasn't explicitly followed the U.S. regulation.
And the answer to the question of whether that kind of data would be acceptable is that it's not obligated to follow the reg, since it's conducted outside the country. What it is obligated to do is follow either that individual country's regulations and laws, or the Declaration of Helsinki, whichever affords the greater protections.
The other clarification I wanted to make had to do with how pediatric populations in studies are intended to be included or not in ‑‑ within the context of this reg. And they're not specifically addressed in this Reg 50.24, but they're not excluded either. We recognize that a large number of these studies might potentially involve pediatric populations, and they are intended to be a part of this regulation as well.
Pediatric consent and research is also covered a little further down in that regulation in 50.55. And they don't supersede or trump one another; both of those parts of the regulation should be followed with respect to pediatric consent.
DR. NORMAND: But if I ‑‑
DR. HARVEY: As far as the rest of your question, probably somebody else is better suited to answer that.
DR. ZUCKERMAN: Dr. Normand, so I think, if I interpret your question correctly, if we do operate within the regs, can we do a global CPR trial?
DR. NORMAND: Yes. In other words ‑‑
DR. ZUCKERMAN: Yes.
DR. NORMAND: ‑‑ yes.
DR. ZUCKERMAN: Okay. And the answer is: perhaps. Certainly, when we look at outside U.S. data, even when it is collected within our ‑‑ the proper regulatory framework for OUS data, we want to make sure that the data can be extrapolated to the American population.
The FDA presenters gave one example where we had trouble making that extrapolation in this particular device field, and we've had other examples in other device fields. However, I do think if one prospectively considers the appropriate questions, as you seem to be doing, then the potential is there for more of a global drop.
ACTING CHAIRPERSON MAISEL: Dr. Somberg.
DR. SOMBERG: Well, just specifically about this point, you bring up one of the problems of the regulations, in that the Declaration of Helsinki, to my understanding and from what I've been told in this area, does not provide for investigation without informed consent.
So since that is considered the highest form of protection outside the U.S., there's really not a provision for this type of investigation, as I see it. It is often done, but truly there isn't ‑‑ unless you tell me that's been changed in some wy.
DR. HARVEY: Well, duly noted. I mean, this is the regulation as it currently exists. It's my understanding that there are currently efforts or activities underway at the agency to relook at how we express our interest in what kinds of patient protections are afforded in studies that are done OUS, and it may be that we are going to approach it from a different perspective than just the Declaration of Helsinki in the future. But these are the circumstances we have right now, so ‑‑
ACTING CHAIRPERSON MAISEL: Dr. Halperin.
DR. HALPERIN: Yes. I ‑‑
DR. BROCKMAN: Can I just make a ‑‑ can I make a comment? I'm sorry. Bram alluded to the comment I'm going to make, but in dealing with the OUS data ‑‑ this is going back to one of the points I made.
We occasionally have trouble taking OUS data when the EMS system in the region of interest is substantially different from the EMS system here, and the example I cited was a study by Plaisance where a physician was present on the scene of all outside hospital arrests. They respond with the EMS system, so a physician ‑‑ a critical care or emergency specialist was present to guide ACLS therapy on the scene.
Well, so was the improvement in survival due to the fact that there was a physician on the scene? Or was it due to the device? We just don't know the answer to that, and it's difficult to port that, then, into our EMS system here.
DR. NORMAND: My question was more in the spirit of prospectively, if ‑‑ I just wanted to understand, if it's retrospectively, it's not fixable. But prospectively, if ‑‑
ACTING CHAIRPERSON MAISEL: Dr. Halperin.
DR. HALPERIN: Yes. I'd just like a clarification on regulations versus science, and the way this ‑‑ one of the devices are classified, because, in fact, there's apparently 30 or 40 devices that have been approved to ‑‑ as external cardiac compressors or aids in external cardiac compression, but none have been improved for improving hemodynamics.
But, in fact, it's been well documented that properly performed CPR generates substantially better hemodynamics than improperly performed CPR. I mean, this is from many different laboratories.
So, in fact, then, devices that assure the correct performance of CPR are, in fact, improving the hemodynamics. But yet if one claims apparently that a device is being approved for improved hemodynamics, then that's a different process.
Can clarification be made about that?
MS. FLEISCHER: Just for the record, I'm Dina Fleischer from the FDA. Yes. I don't want to get into a big discussion on the difference between 510(k) and PMA. However, when it's a 510(k), you're basically saying you're equivalent to something already cleared on the market.
And so the claims that are being made would have to be the same indication for use in sort of the same sort of claims. And so that's why they sort of ‑‑ those 30 devices all are ‑‑ are aids in CPR.
Now, if the claim that you want to make with their device, for instance, is that it improves the hemodynamics, etcetera, that would be a new indication for use. Now, what route that would take hasn't been ‑‑ we haven't really clearly defined in FDA.
But just to say that it is a new indication, and so it would be different data that would need to be provided, perhaps a PMA with an IDE, and that's a route that can be taken. But up until now, the data that we've given you is what has been submitted to the FDA, and the data that has been submitted.
Is that clear, or ‑‑
DR. HALPERIN: There still seems to be a disconnect between the science and the regulations in that respect.
MS. FLEISCHER: That's why we're hoping that this Panel will help us sort of streamline the process and get clearer points for the indications for use.
ACTING CHAIRPERSON MAISEL: Dr. Marler, and then Dr. Becker.
DR. MARLER: Yes. I wanted to ask the FDA about more specific information about the timing of two processes. I guess one is: how long can it be before the heart is essentially restarted or CPR is started? How long does the brain survive? And how long do you have to perform an intervention? To me, it seems to be a critical time.
And then, how many patients are in trials of devices that actually are within the time that the brain actually can respond to any treatment before that infarction is the predominant place where it is? Because it seems to me you have two independent processes, and we're not directly thinking about it and hooking them together. But you'd have to start the recirculation in a time that the brain can respond.
In other words, what was the time scale on your ‑‑ on the slide? And what was the time scale on the trials that have been done?
DR. LAZAR: I don't have an answer on the trials that were done, because most of the ‑‑ most of the more in-depth studies are not done right at that moment when they're admitted. Most patients, as you know, don't survive.
I think it's about four to six minutes following the cardiac arrest when the brain really begins to fail. And there are other factors, as you know, that impact upon that ‑‑ the age of the patient, how much, you know, intercranial disease might already exist, and so forth. So it's a very, very brief interval, and ‑‑
DR. MARLER: So if you don't get some blood-carrying oxygen and glucose to the brain within four to six minutes, you're not likely to have much impact on neurological outcome, is that correct?
DR. LAZAR: Not much beyond that. That's to my knowledge. Those are the studies that I'm aware of.
DR. MARLER: It seems to me that clears a lot of the air, but ‑‑
ACTING CHAIRPERSON MAISEL: Did you have a comment on ‑‑ Dr. Weisfeldt, did you have a comment on this?
DR. WEISFELDT: Well, Dr. Lazar, just I'm concerned about the lack of control for the nimodipine placebo data. I'm concerned not only about age adjustment but disease adjustment. Patients who undergo cardiac arrest clearly have cardiovascular disease that often affects brain function itself. Can you give us any notion about what a disease- and age-adjusted population might show in the same testing?
DR. LAZAR: I don't have a good answer for that, and the studies that I've read haven't explored that in depth. I mean, one of the things is ‑‑ well, let's say the patient had a stroke prior to the cardiac arrest. What would be the implication for that patient, for example? And so in some of the literature that make a distinction between the CPC, and then there's another scale that tries to factor in how the patient was functioning prior to the cardiac arrest.
And they've tried to do some interviewing of a patient who was in a nursing home, for example, or living independently at home, and so forth, and trying to factor that in. But there the outcomes have always been the CPC and nothing more fine-grained.
And so we really don't have the answers to the questions that you're really asking, but I think they're very important ones.
DR. MARLER: So there's no answer to me on the time interval for patients in existing trials?
DR. LAZAR: That's correct. To my knowledge, with ‑‑ with fine-grain measurement, that would satisfy conventional neurologic criteria.
ACTING CHAIRPERSON MAISEL: Any other burning questions for the FDA before we move on? Why don't we ‑‑ we'll take two more questions ‑‑ Dr. Becker and Dr. Hallstrom ‑‑ and then we'll move on.
DR. BECKER: Yes. I'd like to thank the Panel, and I'd like to ask the question in terms of can you give us a little more explicit information in terms of labeling? We've heard about device categorization, but isn't labeling and the request for labeling from a sponsor an important piece of the burden, if you will, that needs to be presented?
And so my question is: as you talked about the different generations of devices with ‑‑ it was quite notable that sort of the third generation has almost no approved device at this point. Could you comment on whether that's really a labeling issue, meaning the claim of superiority, or is that something intrinsic to the device itself?
MS. TRITSCHLER: I don't think that it's really a labeling issue. The FDA kind of made a regulatory decision that devices that have this ability or capacity to enhance clinical outcomes, even if they're not going to label that claim, if they have the ability to do that, they still need to have clinical data to support that.
DR. BECKER: But so just to clarify, so if a device that improved hemodynamics, for example, said that it simply was equivalent to standard CPR, would the burden of science based on that be different than, you know, a much lesser device that would make the same claim?
DR. ZUCKERMAN: Okay. Those are very important and difficult questions to answer, and that's why we have a whole question set. But I think what you're getting at, Dr. Becker, is an important point, in that the third-generation devices have looked for a superiority claim. And you're suggesting, could the agency consider an equivalence claim?
We're looking to the Panel experts to help us out on that particular question, and we're very interested in hearing your responses, number one. So I'm not going to bias the Panel.
But, number two, I think it will be very important to hear from the Panel members and our statisticians as to what an equivalence claim actually implies, etcetera. Sometimes equivalence trials are harder to do than superiority trials.
ACTING CHAIRPERSON MAISEL: Dr. Hallstrom?
DR. HALLSTROM: Yes, I had a question again for Dr. Lazar. I'm sorry I didn't get it in there when you were standing up. I'm concerned about what you're doing with missing data when you have these repeated measures long term. You're going to have a substantial amount of dropout and refusals.
DR. LAZAR: You're absolutely right, and how we deal with the missing data is an important statistical matter. I know that when I was working with Sharon-Lise on heart failure, and looking at LVADs and other mechanical circulatory support, this same matter came up. And it's a very complex statistical issue, and we need statisticians to help us. But perhaps we could take a look at the characteristics of the patients up until the time that they're lost. That's certainly one starting point.
And to see whether or not there are any predictive factors about who it is that drops out of the system, and to see whether or not that is analytically helpful to us. But I appreciate your point, and the survival analysis is very complicated in dealing with the dropouts. I understand your point.
ACTING CHAIRPERSON MAISEL: Thank you.
I'd like to move on at this point. We'll have opportunity to discuss these issues further and to question the FDA, if desired, later. At this point, I'd like to open the public hearing session of the meeting. Both the Food and Drug Administration and the public believe in a transparent process for information-gathering and decision-making.
To ensure such transparency at the open public hearing session of the Advisory Committee meeting, FDA believes that it is important to understand the context of an individual's presentation.
For this reason, FDA encourages you, the open public hearing speaker, at the beginning of your written or oral statement to advise the committee of any financial relationship that you may have with the sponsor, its product ‑‑ we don't have a sponsor today, but products, if known, its direct competitors.
For example, this financial information may include the sponsor's payment of your travel, lodging, or other expenses in connection with your attendance at the meeting. Likewise, FDA encourages you at the beginning of your statement to advise the committee if you do not have any financial ‑‑ 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.
So at this point, I'd like to call the first speaker, Kenneth Collins, to the podium.
MR. COLLINS: Good morning. My name is Kenneth Collins. I'm the Executive Vice President at Alsius Corporation. I'm a full-time employee. They're my financial interest.
Alsius is a manufacturer of medical devices, including devices marketed currently in the United States for fever reduction and for the induction, maintenance, and reversal of mild hypothermia, in specific patient populations not under discussion today.
Alsius does have before the FDA a 510(k) notification pending for clearance that relates to an existing marketed endovascular heat exchange system for use in the induction, maintenance, and reversal of mild hypothermia, in the treatment of adult patients after out-of-hospital cardiac arrest where the initial rhythm was ventricular fibrillation.
As stated in the FDA's Register notice, we're here to discuss and make recommendations regarding clinical trial design and the evaluation of CPR-enhancing devices and therapies for cardiac arrest patients.
As a manufacturer of medical devices, we have sought to present today, relating specifically to the session this afternoon on hypothermia in the post-recovery phase of resuscitation care. Cardiac arrest is not, in itself, a disease. It's a potentially reversal plunge from life to death.
Successful resuscitation returns the patient to life, but there are consequences to face in connection with the treatment and outcomes of the precipitating disease state, and the additional effects of the hypoxic insult associated with the arrest.
Successful treatment of sudden cardiac arrest, its predecessor conditions, and sequelae, requires interventions applied across multiple providers, often across several clinical settings ‑‑ the so-called chain of life.
These interventions make it difficult to assess the contribution of any single link in the chain. Even so, multiple interventions, including hypothermia, have been subject to complex multi-year trials and have been shown to be effective in reproving morbidity and/or mortality in this devastating state.
The focus of the comments today from ‑‑ my comments today are on therapeutic hypothermia, controlled or mild hypothermia, and sudden cardiac arrest.
This has been a topic for nearly 50 years. There is now persuasive data demonstrating the benefit in humans. In fact, the therapeutic value of hypothermia in the immediate treatment of the patient suffering out-of-hospital cardiac arrest has been recognized and included in professional guidelines.
The American Heart Association, American College of Cardiology, as part of their membership in the International Liaison Committee on Resuscitation, have recommended that the 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 when the initial rhythm was ventricular fibrillation.
This recommendation is based upon two randomized controlled trials ‑‑ the so-called HACA, or Hypothermia After Cardiac Arrest study in Europe, and the study by Bernard, et al. in Australia.
Significant improvements in morbidity and mortality were obtained. If you look at the data as a whole, if you treat seven patients, an additional one goes home, there are several methods for inducing mild hypothermia achieved in the HACA and Bernard clinical trials.
External means, such as ice packs, cold blankets, and forced-air cooling have been most commonly used to date. Other methods of inducing comparable hypothermia are variable, including endovascular heat exchange catheters. These products are already released and on the market for other indications, for uses that include both normothermia applications but also the induction, maintenance, and reversal of mild hypothermia.
Each of these devices serves as a tool for inducing mild hypothermia. Alsius believes that in the light of the pre-clinical and clinical data already available there is no reasonable, scientific basis to require each individual device to bear the full burden of another randomized controlled trial to prove the clinical utility of inducing mild hypothermia in sudden cardiac arrest patients.
I'm being told to sum up.
The question to the FDA review of individual devices in this particular setting should not be whether each individual device can, once again, be shown to improve survival, but, rather, where the device introduces new questions of safety or efficacy that are different from the predicate devices.
If the clinical data are required, the FDA and the sponsor can and should be feasible in choosing the most appropriate data types and study methods consistent with the statutory least burdensome approach. The FDA has shown clear leadership in its use of post-market studies.
I do wish to press one small point made by a previous speaker. The FDA does issue post-market surveillance orders in relation to 510(k) product. Indeed, it has recently done so in respect to temperature regulation devices.
The ability to use post-market studies after clearance, in conjunction with such agencies as the National Registry of Cardiopulmonary Resuscitation, offers real public value, particularly since there are provisions within the Hospital Insurance Portability and Accountability Act that allow efficient data collection under the FDA's tight and appropriate oversight.
Thank you for allowing the presentation.
ACTING CHAIRPERSON MAISEL: Thank you.
The next speaker is Dr. Keith Lurie.
DR. LURIE: Good morning. My name is Keith Lurie. I am a practicing cardiac electrophysiologist, an inventor of the active compression-decompression, and co-inventor of the impedance threshold device. And I founded a company, Advanced Circulatory Systems, to try to get this technology onto the streets.
I'm also a professor of medicine and emergency medicine at the University of Minnesota, and I'm pleased to be able to speak to this committee this morning.
I'd really like to thank you, the FDA, for having this Panel meeting today. It's a very important step in helping to evaluate new CPR technologies.
Despite the widespread practice of CPR, its inherent inefficiencies contribute to the extraordinarily high death rates for patients with cardiac arrest. Greater than 1,000 Americans will die today from cardiac arrest. That's more than all the losses of Americans in Iraq to date. Half of those patients, or less actually, present with ventricular fibrillation, the most favorable rhythm we've heard about.
And even after surviving to the hospital, nearly 75 percent of them will die before hospital discharge. This problem is enormous. It's been underrecognized, and it must be recognized before this Panel can logically proceed with ways to look at the questions at hand.
We are very pleased that the FDA is taking a fresh look at this problem of CPR device evaluation.
My first point relates to the need to define minimal essential requirements for safety and effectiveness of new CPR devices. Safety and effectiveness, as this is a disease process where nearly all people die, are certainly relative.
Even in cities like Seattle, survival rates are only 17 percent for all patients who receive CPR. We can do better. By defining the essential minimal requirements for safety and efficacy for CPR devices, by using the current standard of care, a pair of hands for comparison, we will make a big step forward. Very few people use those 30-plus cleared devices that we heard about.
My second and most important point focuses on the question of endpoints for studies of new CPR technology. They must be consistent with the chain of survival approach recommended by the experts at the AHA. Each new technology should only be evaluated foremost to demonstrate safety and effectiveness for what it was designed to do.
For example, if a defibrillator is being developed to terminate ventricular fibrillation, and studies show that it can safely and effectively accomplish this task, such studies should be sufficient for a new device clearance.
Given the non-standardized care of patients once they are admitted to the hospital, it is difficult, if not impossible at present, to control for the large number of critical variables associated within hospital care that impact the potential value of a CPR device.
What is, therefore, critically needed is that each device that strengthens each link in the chain of survival is evaluated by itself to make sure that it is able to safely and effectively strengthen that given link in the chain, whether it's an improved way to call for help, whether it's an improved way to move blood during CPR, to ventilate and provide oxygen without lowering blood pressure, to defibrillate at the right time with the right kind of energy without damaging the heart, or to provide cooling.
Each new technology must be evaluated to determine if it is as safe and effective as whatever is being used today. If the standard of care is a pair of hands, that should be the standard to which the alternative therapies will be tested. Not some other device or technology that either does not work or is no longer being used in the care of patients.
We all strive for longer-term patient outcomes, such as increased hospital discharge or one-year survival. However, if such endpoints are required prior to the initial clearance of new CPR technologies that were developed to strengthen each individual link in the chain of survival, there will be little or no progress.
For example, no biphasic defibrillator has ever been shown to improve long-term survival. But such devices are the standard of care as they defibrillate more effectively than earlier versions.
Demanding long-term endpoints prior to clearing products for use would be unfair to the technology, deny care to the patients who desperately need them. And, moreover, the long-term survival endpoints cannot be achievable without an enormous number of patients, large, more adequately powered studies, not to mention the tremendous expense, and, most importantly, the opportunity cost in terms of the lives lost along the way prior to the device clearance.
If we use AEDs as an example, they were introduced in the mid-'80s in Seattle by Dr. Leonard Cobb. Twenty years later, $25 million later, and with a barely statistically significant study, The New England Journal most recently described the results of the PAD trial by members of this Panel. That's a great trial, but think of all the lives that would have been lost had we not had the AEDs out there in advance.
My final and third point is that the control group is critical for CPR studies. The control group study should be the current standard of care recommended by the AHA. Technologies and approaches that are speculative and not based on conclusive results with patients should not serve as a control. The gold standard for CPR is conventional manual CPR, not a device. Conventional CPR should serve as a control group until there is another gold standard.
We are at a crossroads in CPR research. To impact the extraordinarily high current mortality rates, it would require more rapid, nimble, and creative thinking about this technology, a lowering of regulatory barriers, and a commitment by all parties involved to remain engaged in developing and testing these new technologies.
The FDA can continue to play a leadership role by first recognizing the regulatory barriers, that they have prevented progress, and, second, developing creative ways to remove these barriers. This meeting is a real step forward in this regard.
While I praise the recent efforts of the FDA to, for example, allow defibrillators to be sold without prescription, there will be no real progress in CPR until we move more blood.
MS. WOOD: Please, please complete your statement.
DR. LURIE: I shall. Thank you. Until we move more blood during CPR. Many of the devices that strengthen the links in the chain of survival are already developed. With an improved understanding of what is needed to demonstrate their safety and effectiveness in clinical trials, we can pick the right road forward, so that our loved ones, our friends, our neighbors, really have a second chance after cardiac arrest.
ACTING CHAIRPERSON MAISEL: Thank you.
The next speaker is Dr. Joe Putnam. Is Dr. Putnam here? Is there another representative of the Society of Thoracic Surgeons here? Okay. Very well.
Geretta will now read a statement into the record.
MS. WOOD: This letter is dated September 6, 2004. "Thank you for this opportunity to address you. I would like to make a few comments about both the need to develop and implement studies of new devices for the treatment of sudden out-of-hospital cardiac arrest, OOHCA, and the ethical challenges related to conducting these studies.
"Primarily, this is a plea to further study the process of protecting human subjects while moving forward with well-designed studies. Of course, sound science is an integral part of protecting subjects, since it is only reasonable to ask subjects to accept the possible risks of a study if there is real hope that the study will provide the answers to a scientific question that will then benefit future patients.
"It is estimated that between 250- to 450,000 Americans over the age of 35 die from sudden cardiac death annually. Despite advances in health care, there has been little improvement in survival from OOHCA, which is estimated to be 5 percent nationally. In fact, the proportion of cardiac deaths attributable to OOHCA increased by 23.5 percent between 1989 and 1998. Thus, well-designed studies testing new treatment interventions in cardiac arrest are critical.
"However, for treatments to be effective, they must be administered early. This often makes it impossible to obtain informed consent from the patients before enrolling them in the studies of new, potentially beneficial treatments. Surrogates are not commonly available at the scene, and when they are the emotional nature of the situation often makes obtaining consent from them impossible.
"This dilemma can be summarized as: consent of human subjects for participation in research requires that they fully understand their role and risk, not be coerced, and be allowed to withdraw at any time without penalty.
"In an emergency situation, informed consent is not always possible, but the need for good research data is very high. Here is the ethnical difficulty and a real conflict of values. A population that might ultimately benefit from research cannot consent to the research, and are, thus, excluded from the potential of therapeutic advances.
"Patients at high risk of morbidity or death with cardiac arrest, shock, head injury, or altered mental status are evidently incapable of providing an adequate consent, but, nevertheless, are often in the greatest need of innovative therapy and might be willing to assume some risk for potential benefit.
"To help address this dilemma, in 1996 the Department of Health and Human Services and the Food and Drug Administration jointly published regulations known as the Final Rule for performing studies when obtaining prospective informed consent is impossible because of the patient's acute medical condition.
"These regulations create two new safeguards to protect human subjects ‑‑ community consultation and community notification. Limited information is known about the effectiveness of the community consultation and notification process.
"Researchers have raised concerns that the rules hinder their ability to perform resuscitation research. At the same time, there is also little known about subjects' actual experience in these studies, and whether they are adequately protected. While studies using these rules have the potential to find new treatments that may save lives, the burdens and risks of these studies fall to the subjects enrolled in the studies.
"While challenging studies have successfully used exception to informed consent, a recent abstract reporting on a survey of United States medical school IRBs found that a significant number of IRBs at medical schools have reviewed at least one study under the final rule, and that the more funding a site receives from NIH the more likely it is to have reviewed a study.
"On the other hand, another recent study suggests that the new rules may be limiting the ability of United States researchers to perform resuscitation research. They found a decrease in cardiac arrest trials in the past decade and suggest that this may be due to the regulations.
"Surveys of public willingness to be involved in research without consent has shown that willingness depended on income and the perceived risk of harm. These studies also found many respondents had concerns about studies performed without consent, but most subjects would personally be willing to be enrolled in such a study.
"No studies to date have evaluated the experience of subjects that have been enrolled in a study using exception to informed consent. We do not know whether or not these subjects believe that the process protected their rights. Such studies may help determine better means of community consultation and notification.
"We do know that researchers report that complying with the rules is complex. For example, the public access to defibrillation trial, PAD trial, found that the study was reviewed by a total of 101 IRBs, and median interval from submission to approval was 108 days.
"They were unable to report on total cost, because this data was not collected prospectively. One study found that the disclosure process required in excess of 80 hours. Another found that the process leading to waiver added $5,600 to a study that was terminated after four persons were enrolled.
"Calls have been made for modifications to the statutes. However, those who advocate rewriting the regulations most carefully assess what the ‑‑ must carefully assess what the real barriers to resuscitation research are.
"A lack of understanding of the regulations may exist, and the final rule was not written to make research without consent easy, but to protect patients. As the dialogue continues, and as we learn more, the time may come to approach you, the policymakers, to modify the laws.
"However, before that can or should happen, we need objective data about how the rules are affecting both the ability to perform the research and the subjects they are meant to protect. If the guidelines are to continue, there is a need to determine if patients enrolled in such studies believe that their rights have been protected.
"At the present, we need to look for novel ways to implement the rules. A study of 16 IRBs from the institutions participating in a multi-center trial found variability in several areas. One IRB waived the requirement for informed consent, five IRBs permitted telephone consent, and three IRBs allowed prisoners to be enrolled.
"Because multi-center trials require the approval of so many IRBs, some have suggested the establishment of a central IRB. Such an IRB could be composed of ethicists with expertise in the regulations surrounding exemption from informed consent research, resuscitation researchers, and a diverse spectrum of community representatives.
"Thank you for your time. Terri Schmidt, M.D., M.S., Professor and Vice Chair, Department of Emergency Medicine, Oregon Health and Sciences University, Chair, Ethics Committee, Society for Academic Emergency Medicine.
ACTING CHAIRPERSON MAISEL: Thank you, Geretta.
Is there anyone else in the audience that wishes to address the Panel today on today's topic or any other topic? Seeing none, at this point I would like to close the open public hearing.
It is now ‑‑ I have 10:40. Why don't we take a 15-minute break and reconvene at 10:55.
(Whereupon, the proceedings in the foregoing matter went off the record at 10:40 a.m. and went back on the record at 10:58 a.m.)
ACTING CHAIRPERSON MAISEL: So we'll begin our open committee discussion at this point, and we will use the FDA questions as a guide. There are three main topics, maybe four, within the FDA questions. And what we'll do is discuss each topic and try to confine our comments to the topic at hand, and then answer the questions.
So, for example, the first one is inclusion/exclusion criteria for the CPR-enhancing devices. And so why don't we open the discussion on who should be included in these trials, you know, witness/non-witness arrests, type of rhythm ‑‑ VF or other rhythms ‑‑ etcetera. So why don't we have discussion on those topics.
DR. ORNATO: Thank you for giving us all an opportunity to put our minds together. Specials thanks to the FDA.
In response to your question, I think it really, to some extent, of course depends on precisely what you're looking at for a device or a drug. If you're looking at biphasic versus monophasic, for example, obviously you're just going to be looking at VF patients.
That said, I think for many of the broader trials, unless there's some specific niche that's being targeted, as in the defibrillator issue, it's awfully difficult to really be sure what rhythm you're dealing with initially.
Now, if it's clearly pulseless electrical activity, and you've unorganized rhythm but no pulse, no signs of life, that's fairly easy. But the differentiation between coarse, medium, fine, particularly fine VF, and asystole, is very, very murky.
And in our EMS system ‑‑ I'm Medical Director for the City of Richmond ‑‑ we regularly show our paramedics tracings that they've recorded from the field with five- or 10-second snippets of rhythm. And they will raise their hand, how many think it's V-fib, how many think it's fine V-fib, how many think it's asystole.
And what I'm getting at is they'll disagree, we'll have sort of a bell-shaped curve. We'll show the next rhythm. They don't realize they're coming from contiguous five- to 10-second strips of the exact same patient. Because VF has a direction, has a vector, it's very difficult to know in any tiny snippet whether you're really dealing with VF or you're just 90 degrees off the vector.
So I think most of us are becoming more convinced that the rhythm itself initially is maybe a little less important. And it's a lot easier to design trials when you throw out the broad net and take all the rest, or at least all the rest that are witnessed, where you've got some belief that it's been a relatively short downtime interval, and then do you subanalyses afterwards.
It gets you out of a lot of the problematic areas, again with the exception of interventions that are very specific to the rhythm, like ventricular failure.
Hopefully, that will get us started.
ACTING CHAIRPERSON MAISEL: Dr. Somberg.
DR. SOMBERG: Well, I hear what Dr. Ornato says, and I have a concern in that, yes, the rhythm may not be the most critical aspect, except it depends on what device you're developing, of course, if it has a relation directly to conversion of a rhythm.
But isn't a rhythm a good surrogate for time down? And, I mean, you know, it's a classic thing. You run to an arrest. The only experience I have is the hospitals. If you run into arrest in a hospital, you turn around and you say ‑‑ you know, four walls, "How long has this patient been down?"
If you don't get an answer, or if, you know, you have an assistant or some nurse's aid, or what have you, they discover this. I mean, they just look and say, you know, "This happened." So, therefore, I think it ‑‑ you know, if you have ventricular fibrillation, not always, but it may be more likely that you have a latency that's diminished.
And I think one ‑‑ you know, there are several key considerations in our discussion today, and I think the first one is the latency. And that is the time from the initial occurrence to when you lose perfusion, and I think that relates to a lot of determinants of outcome. And if you have a very long period, I'm not sure there's anything you're going to do.
In fact, we heard this four to six minutes. Let's say that's true. Let's say six minutes, or we'll give it eight minutes. That means most cardiac arrests in the United States cannot be addressed effectively, because no one is going to get to people in six or eight minutes. I mean, you know, if we drove out here, went to 270 and back, it would take longer. So I don't know how an ambulance could possibly get to someone.
So with all that said, I think the rhythm may not be ‑‑ you know, we don't want to approve a device for hemodynamic CPR augmentation because you have VF, fine VF, tosade de pointes, you know, multi-form ventricular tachycardia, etcetera. But it may be an index.
And if I was doing a study ‑‑ I mean, you know, all of these are going to be recommendations to someone who is sitting there, the Panel to set up a study, but if I was doing a study, I would want to pick the most viable patients to address first and then maybe address people who have prolonged resuscitation.
So I think rhythm, while I agree with you is not ‑‑ first of all, you can't always say, "What is the rhythm?" because it may change momentarily and you only have one snapshot. But let's say you do have some inclination.
I would be more inclined to pick a rhythm, and it's my understanding that fibrillation or ventricular tachycardia that may be pulseless is even more of an earlier antecedent, is the appropriate consideration, because those patients in those trials, or those patients who entered those trials, have a greater, I think, propensity to have some sort of benefit.
ACTING CHAIRPERSON MAISEL: Dr. Becker.
DR. BECKER: Yes. I'd just like to make a comment that it seems to me there's another important aspect to inclusion/exclusion that we need to consider, which is sort of a new paradigm in terms of the timing of cardiac arrest and when we're providing therapies.
And a recent paper that I'd like to highlight by Dr. Weisfeldt talks about the three phases of cardiac arrest. And the notion there being that in the early phase of cardiac arrest you may have one therapy that's most appropriate, but that shortly after that there may be a totally different therapy that becomes the critical initial therapy, in the circulatory phase or in the metabolic phase.
And so I don't think it's possible to collapse that whole audience of patients into a single study any longer with what we know in terms of the physiology of cardiac arrest. And so I think that when we think about inclusion/exclusion criteria and the communities that we're studying, you know, we have to be very careful that studies, if you will, are designed to answer the question that they seek to answer, by using the most appropriate population.
And I would just suggest that, for example, if someone were studying the metabolic phase of cardiac arrest, that it would not be appropriate to subject that therapy to all-comers in cardiac arrest, because we know that early defibrillation would be the most appropriate thing for the very early patients.
So I think that the science ultimately guides inclusion/exclusion, and I think that as the new paradigm shift occurs with the appreciation of the phases of cardiac arrest therapy, much like we would not treat a Stage I cancer protocol the same as a Stage as a Stage III cancer ‑‑ no one would do that ‑‑ you would say there would be different therapies. I think that we will have to adjust the way we look at these clinical studies as well.
ACTING CHAIRPERSON MAISEL: Dr. Halperin.
DR. HALPERIN: Yes. Cardiac arrest obviously is a ‑‑ can be a very complicated disease process ‑‑ has been pointed out. And it has a number of unique aspects that really differentiate it from other disease processes, and, in fact, the clinical trials then that are going to be designed and executed and analyzed to deal with cardiac arrest have some inherent differences.
And one of those differences is is that the ‑‑ the inclusion and exclusion criteria may not be obvious, or obtainable, at the time when patients need to be enrolled, because, in fact, data on how long the downtime is, and exactly what comorbidities may be present, which would normally be used in exclusion criteria in other studies, actually that information may not be available in a timely fashion.
And I think that, then, scientists and regulators who deal with these studies I think have to be cognizant of those issues, and take those issues into account, so that the classic criteria that we use for designing and judging studies, maybe there should be some leeway taken to take into account the special considerations for cardiac arrest trials.
And this may include things like actually prospective criteria for inclusion or exclusion criteria that could be applied even after patients are enrolled, because, in fact, we don't want to study people who are not viable necessarily, because any intervention, as has been pointed out, would not work in patients who are completely non-viable and dead.
And, again, those are more complicated features that should be taken into account in cardiothoracic trials.
And one last comment at this point is is that, although ventricular fibrillation may be a surrogate for time in some situations, at least half, if not more, cardiac arrests that occur these days are not due to primary ventricular fibrillation. And studies of those rhythms are probably very important, so we certainly shouldn't exclude non-ventricular fibrillation arrest trials.
And, in fact, blood flow devices actually may be more effective in those kinds of arrests than in ventricular fibrillation arrests.
ACTING CHAIRPERSON MAISEL: Dr. Normand.
DR. NORMAND: I realize we're not talking about the design right now, but it's difficult for me not to think about the design when thinking about the inclusion/exclusion criteria.
And with that in mind, and reading the material that was handed out, it seems to me that one needs to think about the latency time differently if you were to randomize. And pretend that you weren't randomizing, and I think we would think about things a little bit differently, because clearly the distributions of the populations in the various arms would be more subject to confounding.
So I think it's important obviously ‑‑ this is an obvious fact, but I think if we're talking about the inclusion/exclusion criteria, we need to think about the type of design. I realize that's further down, but I want to put that out.
It also relates to in terms of the type of data that are collected and the feasibility of including and excluding the right populations of people. So I actually would have different suggestions depending on whether or not we go a randomized trial route or if we're going down perhaps, let's say, an observational ‑‑ prospectively, well-designed observational study.
And then I'll just add one more comment, and that is related to the question about ‑‑ the latency question about time elapsed between arrest and arrival. And my simplistic understanding of the literature says either you have no idea ‑‑ if it's witnessed, you might have a single report, or you may have multiple reports.
And just, again, in terms of inclusion/exclusion criteria, I think it would be important for ‑‑ at least to get a handle on who is included and excluded is to figure out how often it's no idea, how often it's a single report, how often it's a multiple report.
And when there's no idea, that's a different set of issues. When there are more reports, then statistically we can minimize the error, if we have multiple reports. And there are ways to refine that, but ‑‑ but, again, I think it's at least difficult for me to think operationally about inclusion and exclusion criteria without thinking about the design.
ACTING CHAIRPERSON MAISEL: Dr. Yancy.
DR. YANCY: I would concur that the design does, in large measure, dictate the population, or vice versa. But let me throw out another possibility with regard to inclusion/exclusion criteria.
I think that the data that we've been given a chance to review reflects how heterogenous the patient population is that's affected by cardiac arrest. And if we are to move this paradigm forward, we probably need to find a way to have a more uniform patient population.
There are some contradictions here in thought process, because you're talking about an immediate circumstance and emergency, so it doesn't give you the flexibility of lots of thought for going through the process of inclusions and exclusions.
But one patient population that I do think merits a bit of thought is the hospitalized patient in a CC or critical care environment. I have the privilege of sitting in oversight of a large registry in heart failure, and I can tell you that, of over 100,000 patient episodes, there's a 1.5 percent incidence of CPR being administered. That's 1,500 patients. That's decidedly more than any of the studies we've seen.
Now, that incidence may be higher or lower for other cardiovascular illnesses, but my point is that, in an ICU setting, you can overcome the informed consent issues, because, as a matter of fact, upon admission to the ICU, these issues can be discussed. So you have that opportunity.
You may be doing a lot of prep work for low incidence, but at least you'd get around that, because in my judgment the informed consent is the most difficult part of this whole problem.
Secondly, you have the chance to learn more. I don't think that this area is as well defined as it should be. So you get to understand what the pre-existing clinical circumstances are and what the precipitating factors might be. You understand the latency quite well because you can chronicle almost everything that happens.
So I would think that that would be a patient population which you could start your investigation, and then move from the ICU setting to the broader hospitalized patient population and then to the outpatient setting. Usually the focus of these kinds of efforts is to start in the out-of-hospital scenario. It's very dramatic. It's very immediate. But it's so heterogenous that the efficacy is hard to demonstrate.
I think that ‑‑ and I don't disagree with the notion of a broad net, but for this paradigm maybe you start with a narrow funnel and then widen it as you identify in whom it works best and how so.
ACTING CHAIRPERSON MAISEL: Dr. Ornato.
DR. ORNATO: Dr. Yancy, I have to agree with you on your comments with respect to the population that you've just described. But I have to also I think take issue from a slightly different perspective, because when one looks at the literature on heart failure patients who arrest in an ICU, coronary patients with ischemic disease who arrest in a CCU, floor patients in a hospital, both on telemetry and in non-telemetry as well as other general areas of the hospital, as we see in the National Registry of CPR, which has over 25,000 in-hospital cardiac arrests now from 400 different hospitals in the U.S., and we contrast each of those venues and the patients and their presentations and outcomes with out of hospital, they're totally different animals. Totally different animals.
The pathophysiology is different. The initial substrate in the myocardium in terms of high-energy phosphates, for example, are wildly different. The outcomes are quite a bit different, and that's one of the problems that I think faces us in trying to figure out what the right model is.
And that's perhaps the reason I took the initial stab at answering the question, by saying there are so many of these other confounders, often we find it's just easier to throw the broad net. Then, as we get more information, subset it and do our secondary analyses on the subsets.
Second ‑‑ and I need to let other panelists get in, but I want to reiterate the heterogeneous etiology of cardiac arrest. The V-fib going to ‑‑ to fine V-fib to asystole, Dr. Somberg I think very properly and correctly pointed out, we've all lived with, we all understand it, and it's a correct model.
It doesn't necessarily apply, as we all know, to other etiologies of cardiac arrest. For example, pulmonary embolus. It's now estimated that 5 to 10 percent of out-of-hospital cardiac arrests occur due to sudden, unexpected massive pulmonary embolus. It may even be higher in certain hospitalized patient groups.
Those patients tend to present, as documented in the literature, by decision rules that have been prospectively obtained and then validated in secondary data sets. A high percent of those patients present with shortness of breath followed rapidly by cardiac arrest with pulseless electrical activity as their initial presentation. That's a special presentation.
There's a trial underway in Europe called TROICA. I sit on ‑‑ I chair the DSMB. It's looking at thrombolytic therapy for out-of-hospital cardiac arrest.
If we were to restrict trials to the VF model, it would be hugely problematic, because we would a priori be excluding innocently, and not realizing it, perhaps one of the most attractive groups that might have at least a theoretical opportunity to respond to that form of therapy.
So I guess my final point, in summary, is that the more you get into this can of worms, the more you realize that unless it's a very specific device for a specific scenario, like VF, you're often better off throwing the broad net and then leading up to your design and secondary analyses, trying to filter out, and using criteria for whom you exclude from analysis as the way of filtering down to get a homogeneous group.
ACTING CHAIRPERSON MAISEL: So at this point, why don't I ask Geretta to start reading the questions, so we can make sure we cover all the topics that are of interest.
MS. WOOD: "The following are considerations when defining the cardiac arrest trial patient population. The defined trial population can support a reasonable enrollment rate. The population defined has the potential for providing scientifically valid data. The population defined, and, thus, the results obtained from these patients may be able to support a wider population of cardiac arrest patients.
"Please discuss the following regarding the inclusion/exclusion criteria. The first question: should the study exclude non-witnessed arrest, or should the study include both witnessed as well as non-witnessed arrest?"
ACTING CHAIRPERSON MAISEL: Well, if I could just summarize our thoughts regarding general inclusion/exclusion criteria, I think the consensus is that we certainly don't want to exclude important patient populations from being studied. It seems that both witnessed and non-witnessed arrests could be and should be studied. However, it may not be appropriate to study them in the same trial.
And identifying more uniform or well-defined patient populations for individual studies may be helpful. And that probably ‑‑ that may apply to the documented ventricular fibrillation in the next question, although we may argue over the ability to diagnose the rhythm accurately.
DR. SOMBERG: Bill?
ACTING CHAIRPERSON MAISEL: Yes. Dr. Somberg.
DR. SOMBERG: I think that was well put, and I just want to say I think we're trying to develop a general process here. And, you know, there are very specific ‑‑ it's like two things. You know, a suit that fits all size 40s, or one that only does something for a unique population.
So, you know, it is not to exclude anybody or any groups, but I think this ‑‑ that the only way you can have this meeting is to have some sort of general idea if you had a general treatment for a general cardiac arrest.
ACTING CHAIRPERSON MAISEL: Al, did you have something you wanted to add?
DR. HALLSTROM: Well, I would just like to strongly support both Joe's and Henry's comments. I think in out-of-hospital situation, it's going to be very difficult to ‑‑ to put in place very complicated inclusion and exclusion criteria, even the simple issue of witness versus non-witness.
But I think that data should be collected, and I think one should be able to define your primary comparison group based on data that is collected prior to the intervention. So I would strongly encourage that trial design allow more patients to be enrolled than are actually defined for the primary comparison.
ACTING CHAIRPERSON MAISEL: Okay. Next? Judah.
DR. WEINBERGER: I have a comment I'd like to really direct more at the biostatisticians, and that is if we throw the net wider, don't we have the possibility of decreasing the power, or increasing the number of patients we'll have to enroll to see an effect by confounding it with lots of other patients who have either underlying problems that are not going to be treatable, or other diseases that are not going to be allowed in principle to receive a primary positive endpoint?
And by opening up the trial, we're going to vastly increase the cost to the sponsor of coming to a positive endpoint.
ACTING CHAIRPERSON MAISEL: Okay. I think, in general, I think what we're ‑‑
DR. NORMAND: Should I say "perhaps" like ‑‑
‑‑ other people? No. But yes. I mean, and I think I also heard a statement that the efficaciousness is very different. And so, obviously, you could have some ‑‑ so, yes, you will, and you don't want to be ‑‑ have such a large population that you could have a big effect and no effect and a negative effect. So ‑‑
ACTING CHAIRPERSON MAISEL: I think we all recognize the difficulties with these populations. I think answering the question of: should we exclude non-witnessed arrests, or exclude certain patient populations, I think we're saying, no, we should not necessary exclude patient populations. Is that accurate?
DR. MARLER: It's been my observation that exclusion criteria don't generally make the trial that ‑‑ as much easier to do as you might think. You still have to be there. You still have to do the ‑‑ actually see the patient and determine the criteria. And by that time you've done a lot of the work.
By excluding a significant number of patients, you discourage people participating in the trial. So I don't think there's that much to be gained in terms of either power or in terms of ease of execution or cost as you might think in excluding a lot of patients.
ACTING CHAIRPERSON MAISEL: Dr. Zuckerman, did you want to comment?
DR. ZUCKERMAN: Yes, if I could ask Dr. Hallstrom to elaborate a little bit. What I took from Dr. Halperin's comment is that he's looking to do a trial with a wide net. And then, at the end of the day, instead of doing classic intent-to-treat analysis, he wants to do some filtering such that the comparison may not preserve the randomized component of the trial fully.
You know, there are problems in this area, but do you have any practical considerations that you can provide us with? You know, classically, we do like to use intent to treat and not focus on subanalyses.
DR. HALLSTROM: Yes. I think ‑‑ what's the reason for intent to treat? It's to be sure you don't have bias. So I think the intent to treat is kind of ‑‑ has gotten kind of carried away. If you have data that is collected before you actually put in place the intervention, and that data is collected ‑‑ therefore, is not affected by the intervention, you can use that data to select a subset of the patients who were treated by other intervention. And you will have an unbiased comparison group.
And I think it addresses the issue that was raised over here. For example, if you mix ventricular fibrillation on asystole patients with a common treatment, and you dilute the power because perhaps in the asystole patients you have expectation of very little increase in survival ‑‑ or none ‑‑ yet the complications of trying to separate the two at the time of the intervention may make it very difficult.
You can still restrict your analysis to the VF patients, provided that data was collected before the intervention.
ACTING CHAIRPERSON MAISEL: Dr. Somberg.
DR. HALLSTROM: If it would be unbiased.
DR. MARLER: And the subset was prespecified, you said?
DR. HALLSTROM: Well, I think you always have to pre-specify. Otherwise, you're playing around with your alpha and beta a lot.
DR. SOMBERG: I would just make the point that that's great for hypothesis-generating, but it wouldn't be for a definitive trial for approval. If you had a device that only worked in a subset, say the people with the VF, and they were 20 percent, and 80 percent had asystole in the study, but you had a fantastic signal if you did that, and I was sitting on the Panel two years from now or five years from now, I would recommend that they do another study to take people with VF and prove that point.
Because so often ‑‑ and I can give you ‑‑ you know, examples right now where we see subsets that work. They do the study, and the study comes out negative. So it's ‑‑ it's not clear-cut that the one variable that you've picked out is the one that's operant in the determination. So I think that's a very slippery slope to make a decision on.
ACTING CHAIRPERSON MAISEL: Dr. Weisfeldt.
DR. WEISFELDT: In a sense, we're starting this discussion with entrance criteria when we haven't really decided what the outcome is that we're going to accept. And if we look at this issue from the point of view of other major cardiovascular diseases, as to what has really influenced devices and drugs, it's survival to some reasonable quality.
And if we say that that is the ultimate endpoint variable that we're going to go toward, and our purview is devices in CPR in a large group of people with substantial mortality, then the notion of having entrance criteria that optimize the possibility that that device is going to save lives, and that group is a substantial group of folks among the many who die, that to me seems to be the optimal approach to the entrance criteria.
It does go a little bit to Lance Becker, who is really a co-conspirator of mine, and the phases of resuscitation. But, clearly, if you're studying a better defibrillator, you have to have ventricular fibrillation as the entrance criteria. If it's a pacemaker, you have to have asystole.
If it's somebody ‑‑ and, arguably, there are some data to suggest that there's a phase in which defibrillation has not worked or other measures have not worked, where very clearly there is lots of data to suggest that better circulation is going to work. But it's a subgroup that's identified somehow related to downtime, and arguably response to initial defibrillation.
So you could see a device for circulation if we were going to demand an outcome of improved survival in which there would be a certain set of entrance criteria that reflected the group that needs circulation.
And then, if you go toward, if you will, a metabolic intervention ‑‑ hypothermia ‑‑ it's unlikely that hypothermia in somebody who, because of defibrillation or CPR is alive and alert and heart is functioning fine, and has no problems at five minutes after ROSC, that probably is a patient you would want to exclude from a study of a metabolic intervention hypothermia, because they ‑‑ it's not likely that you're going to improve survival in that group.
So my answer is the entrance criteria should reflect the place in the resuscitation population where that device may have some survival value, and the endpoint that we're seeking is improved survival.
ACTING CHAIRPERSON MAISEL: Why don't we move on to Part C. I think we've addressed A and B.
MS. WOOD: Okay. "Based on the literature regarding early intervention and survival outcomes, should there be a limit on the time that has elapsed between arrest and initiation of CPR?" If so, can you suggest how best to obtain accurate, unbiased time estimates?"
ACTING CHAIRPERSON MAISEL: Dr. Ornato.
DR. ORNATO: Once again, I'm going to start it. Yes, there should be, but that's Nobel Prize material for someone to figure out a way to do it. All joking aside, it's a wonderful question, and there is quite a bit of interesting science in the last decade and a half or so, since Chuck Brown and others looked at such things as median frequency for the VF patients.
And now there are a whole family of different measures for VF patients looking at the relationship between time and markers such as the characteristics of the electrical rhythm. The problem is that, unless we have a device on the patient as in Dr. Yancy's population, which I think is a wonderful example, where they are in an intensive care unit, where you've got them on a monitor, where many of our monitors, hopefully all of them, most of them, have a time stamp, you know, where you know where the event begins, then for such patients, absolutely, we ought to be paying attention to that.
But for the majority of our studies, particularly for the in-hospital but non-telemetry-monitored patients, and certainly for the out-of-hospital patients, they are usually not on any device that's giving us an atomic clock synchronized time stamp.
And so we just technologically, at the moment, simply don't have the means to really start the time at the beginning. And short of that, even such things as when they collapsed or when they passed out, aren't always particularly active, because ‑‑ or accurate, rather, because usually it's a layperson who is the witness, and they may not realize that the person may be underperfusing but may not be totally without perfusion, at least for a brief period, a la they go into V-tach but have some perfusion. Then, when we see them they're in V-fib.
Final comment is that, remember, there's a practical side of this. A lot of these trials are being done in the field, where the major of arrests occur. Often the people "enrolling them," are firefighters, EMTs, paramedics, especially if we're talking about first responder interventions, some of the airway devices or CPR devices.
These are not research nurses. These aren't people who are going to be able to methodically go through a hospital chart for 30 minutes before going in to see the patient to decide whether they've got exclusion or inclusion criteria. These are people whose focus is on trying to get that patient to survive.
That's I think why you're hearing the sentiment expressed from some of us who have worked in this area and stepped on all the landmines and realized that it's scientifically great to be able to get the pure group. It certainly makes your power much simpler and much smaller. But the practical perspective on this makes it impossible to do that in many cases.
ACTING CHAIRPERSON MAISEL: Anyone with ‑‑ sure. Dr. Brott.
DR. BROTT: Well, we have the experience of trying to treat patients with stroke within 90 minutes, and we actually treated somebody off a Delta airliner. But the nice thing was is that you had electronic verification with that patient. You know, the pilot knew what had happened, and so forth, and we ‑‑ and so I would say I agree. I mean, I know how difficult it is to get the time.
But I think somebody made the comment operational, and I think that in this electronic age that trials need to be designed so that if they're in the ICU that time is definitely part of the protocol. I haven't really seen that.
In publications that I read as a neurologist, you know, I'm interested and I'm looking. I really haven't noticed this. And I'm a hospitalist, and I see the cardiac arrest charts, and I really don't see too often in the note monitor documents V-fib started at 11:21 a.m. In fact, I can't think of a progress note that stated that in the last few months, and I've seen, you know, a number of in-hospital cardiac arrests.
So my suggestion would be that any trial have operational times such as, when did the call come into the life squad? When was the first rhythm recorded? Some of these papers have that, but most of them don't. And I think that five years from now, in 2010, we should have a lot of electronic time markers that we can apply, so that we don't have to speculate about four to six minutes with as much uncertainty as we have today.
DR. ORNATO: Could I respond to that, Bill, briefly?
ACTING CHAIRPERSON MAISEL: Yes.
DR. ORNATO: You're exactly right. The problem is there are three, four, or maybe five papers in literature on timing in resuscitation, both out of hospital and in hospital. I'm co-author on two of them.
They've all shown the same thing and that is that people don't synchronize their watches to the atomic clock. And there's so much heterogeneity, for example, in the National Registry of CPR, the largest repository of in-hospital cardiac arrest data in the world, something like I think it's 5 or 10 percent of our data have interventions like defibrillation or CPR or drugs being charted five or 10 minutes before the cardiac arrest supposedly begin, because there are different people and different machines using different timepieces.
And the result is that it ‑‑ you know, if you average it all out, it looks great. But if you look at an individual event, and you try to figure out what happened, it just doesn't work.
ACTING CHAIRPERSON MAISEL: Dr. Somberg.
DR. SOMBERG: Another consideration is the duration of the cardiac arrest treatment, and that's been ongoing, and when you put the intervention in. And while this question asks, how long should you be down before the treatment was initiated, many studies don't look into this, and it is a major factor.
If someone is receiving first responder CPR, then more advanced ACLS, and then you finally get an investigator in his office, you know, an hour later, that's totally different than if it's applied within five minutes or something.
So while I don't have a magic number, I think it has to be specified, and the earlier that the intervention is applied the more likely it relates to the outcome.
ACTING CHAIRPERSON MAISEL: So I think just to summarize, I think we're saying documenting time as accurately as possible, using medical records if possible, and first responders and call times and things like that, to provide the best estimate.
So let's move on to D.
MS. WOOD: "Should the study patients be limited to patients who have arrested in the field, or should the study also include in-hospital cardiac arrest patients?"
ACTING CHAIRPERSON MAISEL: I think we already addressed this. We said both can and should be included, although probably not mixed in the same study, and that they are ‑‑ there are important differences in these populations and should probably be studied separately.
MS. WOOD: "If field patients are to be included, should CPR be initiated by professionals only, or can patients be enrolled if timing is recorded by and CPR is initiated by bystanders?"
ACTING CHAIRPERSON MAISEL: Again, just on the general theme of inclusion, I would think we would want to include both these patient populations, recognizing that there may be important differences in outcome, and that these data should be carefully collected and perhaps pre-specified subgroup analysis.
MS. WOOD: "Do you have any other suggestions regarding the inclusion/exclusion criteria?"
ACTING CHAIRPERSON MAISEL: Dr. Somberg.
DR. SOMBERG: Well, I think certain conditions impact on outcome later. Let's say someone falls on their head, and that initiates the cardiac arrest. Obviously, that will impact on neurologic outcome. Someone who is exsanguinating due to a large laceration of the femoral artery, or something of that nature, hemodynamic support. So there are certain logical things that have to be taken into account.
ACTING CHAIRPERSON MAISEL: Any other comments on the inclusion/exclusion criteria before we move on?
DR. YANCY: Just one observation, if I might. If you think about what we've just talked about, we've basically said we would include all etiologies. We said that we would include in hospital, out of hospital. We would include CPR by professionals and lay individuals. And it's relatively time-insensitive because it's so hard to quantify.
I really think that that is a very, very heterogenous grouping, and that represents a huge challenge.
ACTING CHAIRPERSON MAISEL: I agree with your statement. We also, I think, commented that we would like to aim for more uniform, well-defined studies, so that individual studies would study more well-defined populations, but that we didn't want to exclude any of these patients from future study.
So why don't we move on to the study endpoints, and why don't we open by just having a general discussion.
DR. BRINKER: I think that one of the problems with past clinical studies in resuscitation has been the issue about what is an acceptable endpoint. If the FDA, for instance, takes the position that survival is the ultimate and only acceptable endpoint, then this reflects dramatically, actually, on both the inclusion criteria and the study design.
I think that there should be some surrogates that can be used when you're looking at specific issues. So that if you wanted a ‑‑ if you designed a device, for instance, that could maintain profusion of vital organs without effective cardiac contractility, you might be able to use a surrogate endpoint, such as blood flow, and maybe cerebral blood flow, which I think would be important because it's a conveyor to other therapies.
And except for the patient who has sudden ventricular fibrillation, is immediately defibrillated, it seems to me that the ultimate save of patients, both physically and physically with a neurologically intact system, is to use a variety of devices, or intervene in a variety of different areas ‑‑ maintain cardiovascular function, improve the utilization of vital nutrients by the brain by downgrading metabolism, things ‑‑ a whole bunch of things.
So that I would ‑‑ and it would be very difficult to attack all of these things in one study. So I would ‑‑ I would like to think that one could piecemeal the specific device using surrogate endpoints rather than necessitating an ultimate survival benefit to show efficacy.
ACTING CHAIRPERSON MAISEL: That seemed to get the Panel riled up.
Why don't we start with Dr. Halperin.
DR. HALPERIN: You know, I think that to really get a good answer to what is the correct endpoint, if there is a correct answer, we really have to understand what are we doing during CPR. Because I think it becomes much clearer, if you really understand what your particular device or approach is doing.
For instance, if you have a device, since that's my area of expertise, that it augments blood flow during CPR, there is many, many studies that have shown that if you have viable animals or viable patients, if you augment blood flow during CPR you'll improve survival, even long-term survival.
However, if, in fact, you augment blood flow in somebody who is not viable, you might get improved short-term survival, but then they are brain dead, even though the heart may still be viable, and you may get only short-term survival.
So I think that, then, the ‑‑ what the ‑‑ the physiology of the device, as well as the inclusion/exclusion criteria actually interact very strongly here, because if we had only viable patients that were enrolled in a study we might get improved survival at discharge if we had augmented blood flow. But, in fact, if we had a lot of non-viable patients enrolled in the study, we might get improved short-term survival, but not improved long-term survival.
So I think, then, we have to take into account the inclusion/exclusion criteria, as well as the physiology of what's happening in deciding what is the appropriate endpoint.
ACTING CHAIRPERSON MAISEL: Dr. Kato.
DR. KATO: Well, initially, when I first read over the packet, I was very concerned about device ‑‑ specifically about devices and their ability to try to look at only one part of the chain of survival, or chain of survival concept, much like the recently-approved over-the-counter AEDs.
I was actually fearful that, you know, the over-the-counter ‑‑ using the over-the-counter AEDs as an example, while we have considered it to be effective and safe for the treatment of ventricular fibrillation, there are a number of assumptions that go along with it.
For example, you have to speak ‑‑ you have to be able to understand English. You have to have another person there. In the absence of a secondary person to apply the paddles to you, and if they are unable to understand English, then you're basically out of luck even though the device will work.
And I then thought about what we have done in the cardiovascular surgery field. And much like looking at CPR and the CPR success, it's really ‑‑ you're really looking at a system approach or ‑‑ and a system outcome. However, devices per se are going to be helping at every step of the way.
And so while one of the problems is that, in the cardiovascular area, for example, particularly in cardiovascular surgery, we use devices such as aortic cannula, cardiopulmonary bypass machines, but ‑‑ but all of this technology and all of these devices, including monitoring in the Intensive Care Unit, you know, required technology which occurred at different steps or at different time intervals during the past 20 or 30 years, and only recently has all of that technology come together to create a system of care which has been highly effective.
And so, therefore, I've had to change my ‑‑ in the past 30 minutes of this discussion, change my idea about that ‑‑ that I think that these devices, for better or for worse, are going to have to be evaluated on their own intervention in the area of the chain that they are going to work.
And that the ‑‑ that looking at CPR ‑‑ cannot be looked at CPR per se but as a step in that entire system of care, which then is really a different issue that the FDA doesn't necessarily address, because the FDA doesn't regulate, you know, a process of care. And that's where the device, again, has to be looked at very specifically at one point in that chain.
ACTING CHAIRPERSON MAISEL: Dr. Becker.
DR. BECKER: Yes. I'd like to sort of echo some of those sentiments, and to suggest something perhaps radical. That what we really need to be concerned with is the labeling, and I think that is what we're supposed to be taking care of, as a matter of fact.
And so what I'd like to really highlight is that it would seem to me that we need to have a system in which ‑‑ where we have truth in advertising, where devices can be labeled for what they do, even if what they do is a small part within the chain of survival.
So an example might be a device that simply allows you to call 911 faster some way. One would hope that it would be labeled as a device that allows you to call 911 faster. I don't think it would be appropriate that it should be held to a standard that one would have to do a trial to show that there's better survival based on the whole system that Dr. Kato has just described so well ‑‑ a very complex system as to whether or not there's really survival at the end of the day.
And that, to me, I think the important thing is really the labeling. That a device, for example, that claims that it produces superior survival at the end of the day I think really needs to demonstrate that in an unequivocal way. But I think there has to be the opportunity that a device that clearly labels exactly what it does, that physicians are allowed, if you will, to do what physicians do, which is to utilize some judgment and to have access to those kinds of devices.
ACTING CHAIRPERSON MAISEL: Dr. Somberg.
DR. SOMBERG: This is a very complex area, and, you know, you can sort of think about things carefully one day, come to one conclusion, think about things the next day and maybe even come and ‑‑ well, you did it in 30 minutes; it takes me 24 hours.
But I think these devices are a little bit like obscenity. You know, it's hard to define it. But when you see it, you know what it is. That was paraphrasing of Justice Black, I believe, who stated that.
But if it was a device to call 911, I have no problem. But let's say it's a device to change the cardiac rhythm in a way that could be beneficial or harmful. It gives you pacing support. It cools the body. It warms the body. It thumps twice as fast as any man could do. I mean, you ‑‑ and it costs an awful lot of money.
I think you have to deliver more than a transient benefit, and, therefore, I think there's always going to be a need for a Device Division in the FDA, to go through each application and see how it applies.
So I think my personal feeling is that it's not going to be feasible to come up with an endpoint. But if I had one of the major devices before me, such as a defibrillator, a cooler, a supporter, you would want to see some definitive benefit.
If I have something that gives me a better grip on the chest wall, I would have a different criteria. And maybe that's a simplistic approach, but I don't see how ‑‑ thinking about this for multiple 24 hours while this meeting was planned, I don't see another approach to the problem.
ACTING CHAIRPERSON MAISEL: Dr. Zuckerman.
DR. ZUCKERMAN: Okay. You know, the Advisory Panel is trying to get to the heart of the matter, which is understanding the choice of the primary endpoint and what might be acceptable for FDA approval. And while I agree with Dr. Becker that appropriate labeling and indications for use is part of the equation, I do think that it might be more instructive to go back to Dr. Weisfeldt's original paradigm and challenge.
The first thing is the agency, for more complex CPR devices, is interested in the assessment of a reasonable assurance of safety and effectiveness and defining clinical utility for that device. In clinical language, that usually translates into a reasonable risk-benefit profile for the device.
As Dr. Weisfeldt indicated, I would agree with his choice of the ultimate primary clinical endpoint of interest. And, certainly, if a trial was designed that way, it might have a very easy path through FDA. But is that the only path to show reasonable assurance of safety and effectiveness in clinical utility?
The answer may be no, with the following caveats. You know, people talk about the chain of survival. They talk about surrogates. But from our perspective, we're interested in this Advisory Panel perhaps giving us some more information about, one, why these surrogates might be appropriate. It's important to recognize that we're just now looking for endpoints that correlate with the primary endpoint that Dr. Weisfeldt has suggested.
It's more correlation and capture of the ultimate treatment effect, and how you define that is essential, because certainly we've seen many examples in the cardiovascular literature where so-called surrogates turn out to be very problematic, and we don't find clinical utility, etcetera.
So, you know, one trial plus use of external data for an intermediate endpoint may be inappropriate, but I think the Panel needs to flesh out what are some of those endpoints and how you can get to the goal line that way, rather than just saying that there is something out there right now.
ACTING CHAIRPERSON MAISEL: Dr. Yancy.
DR. YANCY: Bill, I would tend to strongly support that Bram just said. I think there's already evidence in our packet that there is an approach that increases the return of circulatory stability, but with no downstream effect on the number of lives that are discharged in a meaningful, productive way.
So the inclination towards surrogates, even though it makes the clinical trial design at the outset seem to be a little bit easier, I think at the end of the day for this circumstance it doesn't affect the outcome we're looking for. And so I would adhere to what I think Dr. Brinker said earlier, that being discharged alive and functional is topmost.
And obviously there are some iterations of technology that do have very specific approaches or very specific purposes, and that would be intuitively clear that that's all that they do. But for a device that is ostensibly designed to improve the outcome, then we should insist that a meaningful outcome is demonstrated.
ACTING CHAIRPERSON MAISEL: Dr. Ornato.
DR. ORNATO: Okay. Here's the problem. You've got time-dependent physiology. The first three to four minutes, as Drs. Weisfeldt and Becker have pointed out, are totally different, roughly, than the next three or four, and roughly from that point forward, as best we know today, physiologically.
When you've got a very powerful intervention, like early defibrillation, and you can apply it very early, as we did in the PAD trial, it makes total sense ‑‑ in fact, in my opinion, it makes absolutely no sense to look at anything other than the longest reasonable interval that you can power a trial for. At least in PAD we powered it to survival to discharge, looking at neurologic outcome as a secondary outcome.
The problem is for many of the other interventions you're beyond that first three minutes in terms of where they're kicking in. And so you've got the problem from the get-go that the cardiac surgeon had back in the '50s and early '60s.
And, Norman, I think you're absolutely right on the money. You've had a revelation that in the CPR world of research has been talked about for years, which is that if we were all back in the '50s and talking about how to design trials that relate to the performance of open heart surgery, either to correct congenital heart problems, VSDs, tetralogy, or to fix valvular heart disease or coronary disease, it would take multiple advances ‑‑ anesthesia monitoring, in some cases pressor management, management of fluids, shock, blood, blood products, grafting materials, valves, etcetera, etcetera, sutures, not to mention all the technological things that relate to us humans.
It would take multiple of those to be able to get your first person through that procedure alive, neurologically intact. And so once we get beyond that first easy three minutes of the VF patient where defibrillation is the therapy, we're now into the demand where we have to, we believe, correct at least a minimal number of things in order to have anything likely result in neurologically intact outcome.
If we fail to correct those things, we may be overlooking very powerful interventions that we will never be able to determine have value, because they are being studied in the traditional way one intervention at a time.
I think that's really the sentiment that I think Dr. Lurie got us started on, and why we're beginning to give you what seems to be schizophrenic answer to your question, because I think we all agree that what really matters unquestionably in the end is survival to at least discharge and probably beyond, clearly with better neurologic measures. I don't think any one of us disagrees with that ethically, morally, economically, and all the other ways.
The problem is: how do we get there?
ACTING CHAIRPERSON MAISEL: Since it seems there may be a little division among the Panel on this aspect, maybe for some people who feel that surrogate markers are acceptable, could you talk about the specific endpoints or markers that you think would be useful?
DR. HALPERIN: Sure. That's a great distinction there, and I think that when we've grappled with this problem in the past the issue is ‑‑ is, again, what is the purpose of this CPR intervention that you're actually studying?
And for at least blood flow devices, the purpose of the blood flow devices, you know, is to restore oxygenated blood flow to the brain and the heart mainly, to keep the brain viable and to restore viability of the heart, so it will beat again.
Certainly, if the brain is ‑‑ you know, too much time has elapsed, so that there actually has been brain damage, but the heart is still viable, the device will actually ‑‑ could do its specified and desired effect and actually bring the heart back by generating increased blood flow.
If there is not brain damage, though, then the patient would be viable. So in that situation, then, to test how good that particular device is working, short-term survival might be more than adequate, because then it would show that, in fact, the device is efficacious or not, if there was no improved short-term survival, and then that ‑‑ then, trials or some ways of screening out patients who are non-viable could be studied, or the device could be used on them in the future, in fact, with the great prospect that, in fact, the device would be useful for improving survival to discharge, and, you know, long-term survival.
So in that situation, then, some measure of short-term survival would probably be more than adequate. So, then, the endpoints that are used, then, are improved hemodynamics, return to spontaneous circulation, say 24-hour survival, and then survival to discharge. These are the main endpoints that are used.
And I would, then, propose that useful surrogates could be some number of hour survival to show the hemodynamic efficacy of a device. That's what was being proposed by the sponsor, and then, depending on the particular inclusion/exclusion criteria, longer-term survival could be appropriate.
ACTING CHAIRPERSON MAISEL: Dr. Weisfeldt.
DR. WEISFELDT: I don't often disagree with Dr. Halperin, but I'm going to try.
And I'm going to try to defend the notion that you can't get away from reasonable-term mortality. And that is to say that ‑‑ that if it's hemodynamics that we're talking about, there are interventions that improve brain blood flow and decrease myocardial blood flow.
The heart may well be viable, get ROSC as a result of the device and its use, but, in fact, myocardial damage may have occurred that is greater as a result of using the device that, let's say, improved brain flow but diminished myocardial blood flow by increasing right atrial pressure.
There is certainly some hemodynamics that one can think about that would likely increase brain blood flow but reduce myocardial blood flow. And maybe the other way around as well.
So you get ‑‑ to some degree, you get back to the issue of: how confident are we of the intervening variable? We have no MACE. There is no summated endpoint that predicts survival like there is in acute/subacute ACS. There is no real variable except survival.
And if the issue is the need for multiplicity of interventions in order to improve survival, then somebody has got to step up to the batter's box and say, "Okay. These are the three. Here's a drug, a device, a methodology in this group of patients, where we need to intervene with a whole cocktail in order to improve survival."
So you look at the shock study, look at the study of patients with cardiogenic shock and acute myocardial infarction. Very close to the group of CPR.
All kinds of speculation about what might be beneficial, what might not, like intra-aortic balloon pumps, bypass surgery, angioplasty. So you do a randomized study in which you use reperfusion, basically, as the endpoint.
You randomize patients. You do everything in the intervention arm to have successful revascularization. You improve survival, and that changes the standard of care, and that justifies, obviously, the huge expense and approach to survival.
So I ‑‑ I think it's the challenge to put together what is the right cocktail to test it, but I still think the endpoint is survival.
ACTING CHAIRPERSON MAISEL: Dr. Weinberger.
DR. WEINBERGER: I agree with the sentiment expressed by Dr. Weisfeldt, but I think that we ‑‑ the bar is being set very, very high. I think that the best CPR intervention generically could possibly do, the very best it could do, would be to return the patient to the pre-CPR state. That's the best you could possibly expect from a CPR intervention.
So that if the pre-CPR state made it unlikely for the patient to leave the hospital, the post-CPR state, with the best possible intervention in the world, is not going to make it any more likely for the patient to leave the hospital.
I think that asking for a device to return you to the pre-CPR state is too high a bar as well. So, obviously, some of these patients are going to come out with renal failure. Some of these patients are going to come out with some cognitive impairment, and the question is, is what do we find as an acceptable endpoint for that ‑‑ for pre-CPR state, plus or minus a certain amount.
And the question for us to deal with as clinicians is: at what point do we expect return to the pre-CPR state to be defined? And not continue monitoring the patients for the natural history of the rest of their underlying illnesses. So that we are dealing with a broad slice of medicine here, on top of which an event comes along and drops the patient down.
And I think that what we have to decide on clinically, as clinicians, is at what point do we assess return to pre-CPR state and decide that we've gotten close enough to make that a useful therapy?
ACTING CHAIRPERSON MAISEL: Dr. Marler.
DR. MARLER: I just wanted to comment that we've been talking about a chain of physiological events. Also, there's a chain of development, and, I mean, the usual way to proceed with your so-called surrogates is to see ‑‑ you may be interested initially to find out if a device actually does increase blood flow. But the reality of it is, if I understand what has been said, is that sometime or other, if it's going to ‑‑ you're going to improve patient outcome, you're going to have to get there in the first few minutes.
And so that if you first establish that the device can increase blood flow, then it seems to me that it's going to have to be able to be used early on anyway, if there is ever going to be anything ‑‑ any treatment. Regardless of how many different interventions there are, it's going to have to be used early on.
Now, if the device doesn't have any potential effect on brain function, that might be one thing. But increasing blood flow ‑‑ most of the devices we've talked about would seem to have an immediate effect on outcome, both neurologically and cardiologically. So the development can also be put in a chain or in stages.
ACTING CHAIRPERSON MAISEL: Why don't we go to the questions to make sure that we answer everything that is asked of us, and we'll have another chance to talk about the surrogate endpoints.
DR. BROTT: May I just make a ‑‑ just a very quick comment?
ACTING CHAIRPERSON MAISEL: Sure.
DR. BROTT: Being a neurologist, every endpoint I've heard is the same ones that I read about 30 years ago. And so I guess when we come back to it, between now and then it would be nice to hear from people who know about this more than I do. Is there anything new? I was very intrigued on this BNP and heart failure and, you know, as a brain doctor I thought this was pretty neat.
But are there new non-neurologic endpoints? And my comment was for neurology there are no endpoints. With MRI, there are now series looking at post-cardiac arrest MRI changes. And for those of you who take care of patients with stroke, I mean, you look at an MRI scan, and basically you can tell from 10 feet whether or not the patient has had a stroke, as small as a pea you can tell from 10 feet.
And it seems to me that, again, looking forward to 2010, we need neurologic measures of brain injury, ideally that could be expressed as a number, ideally that could be added and subtracted, ideally that wouldn't be dependent too much on differences in culture, language, pre-arrest neurological function. And so I have my concerns about the cognitive measures.
We've been working on this for a while, and it's tough because of those problems. But I think tissue-based measures of brain injury have real promise, and could be incorporated into the studies that are being done now as secondary outcome measures, and perhaps be ready for primetime in five or 10 years.
ACTING CHAIRPERSON MAISEL: Thank you.
MS. WOOD: "What would appropriate clinical endpoints be for efficacy in a cardiac arrest study? Survival. If so, how would survival be defined? Would functioning in a vegetative state be considered surviving?"
ACTING CHAIRPERSON MAISEL: Dr. Somberg.
DR. SOMBERG: Well, I knew at some point I'd bring this in. I was ‑‑ I'm involved in a number of drug studies, and there are certain benefits that are certainly not as powerful as defibrillators, we heard. And I've thought about this long and hard, and I think survival, immediate survival, termination of ‑‑ you know, was one endpoint, the arrythmia.
Survival at one hour, survival to hospital, survival in 24 hours, were all along the chain, but it's useful because each one of these points gives you access to the next part moving along. If you don't survive approximately 24 hours, you're not going to get revascularized. You're not going to have these things. If you don't survive one hour, you're not going to have a balloon inserted, and you're not going to be ‑‑ you know, to these other points.
So I think there is merit to those endpoints as opposed to an endpoint of hemodynamic ‑‑ you know, I feel a pulse, the FIO2 is better, the end-tidal CO2 is better. We may have prognostic indications, but I don't think those are substantial enough to warrant the prolongation of the person's terminal event to ‑‑ for such at the time.
So I think certain indicators ‑‑ you know, for certain devices that's indicated. But we have to say what we're studying, and if we're studying a device, and to ‑‑ to cool someone to save neurologic function, then I think survival at one hour and 24 hours has no meaning.
So just saying survival at one hour and 24 hours has to take into account ‑‑ or making your choice of endpoint has to take into account what you hope to accomplish and what sort of prolonged intervention you apply.
ACTING CHAIRPERSON MAISEL: I'm not sure we can tackle the "would functioning in a vegetative state be considered survival." It sounds ‑‑ religious is one comment.
Bigger than all of us. But certainly survival alone may not be adequate. Obviously, if everyone survived but they're all in a vegetative state, that may not be a worthwhile device. So that kind of brings us to Part B, which is the neurologic issues.
MS. WOOD: "What are meaningful neurological endpoints? And how should they be evaluated? Is a composite endpoint acceptable?"
ACTING CHAIRPERSON MAISEL: So we heard mention of the MRI. Dr. Brott or Dr. Marler, do you want to try to tackle this one?
DR. MARLER: MRI or composite endpoints?
ACTING CHAIRPERSON MAISEL: The entire question.
DR. MARLER: Maybe I could begin by just a comment. I think that whether you analyze it statistically as a composite endpoint or not, any endpoint in neurology is usually a composite of several separate measures. And the problem with the fine points of cognition is you need finer and finer ‑‑ more and more components in your assessment as you look for smaller changes in brain function.
My understanding in limited contact with patients who suffered from cardiac arrest is that there is some small component of change, but what we're really concerned about is a pretty obvious, serious, and major loss of neurological function.
ACTING CHAIRPERSON MAISEL: Dr. Brott.
DR. BROTT: I would, again, stress the importance for kind of a look at cognition as a continuous variable. I note ‑‑ and the discussion was made, you know, are we all there, or are we not? And, of course, the people around this table, we value our cognition so much that we might be attracted to this dichotomous view. But once the brain is injured, of course, the brain is different. It's not like being quadriplegic where your brain is the same, and you're quadriplegic.
In this case, your brain is injured, and so your standards of what you consider to be meaningful, everyday activity may be different. Your standards in terms of what's acceptable movement may be different. Your taste in food, your taste in companions, can all be affected by that brain injury.
And so it's very unique in terms of injury to the brain, and what classifies as acceptable everyday life. And that's why I think you need a ‑‑ if the ideal, we don't have it, and we heard that we don't have it, is something that tries to assess, in a numeric fashion, cognitive function, so we can at least get some ordering of patients and their outcome, not are they bad, are they good, but to be able to sort them along some kind of scale from a cognitive point of view.
The NIH stroke scale was mentioned. That's nice. It gives a number from zero to 42. It correlates with the size of the infarc. It tells you, with pretty good sensitivity, whether or not your middle cerebral artery is occluded. And it also correlates with functional outcome.
Something like that would be very nice. And, of course, that's for, in a way, functions we share with our pets. You know, can we move, can we see, and so forth. So the challenge is much greater for cognition, but I really think that we have to do that.
And then for the future, MRI ‑‑ I was mentioning, for those of you who have had a crick in the neck or a crick in the back, and you get your MRI scan and it's negative, because you've got changes on both sides, as so many of us do, and is it causing a problem or not, well, now there's T2 fat saturation, which can show inflammation of the issue around an inflamed facet and how the synovia. And this is within the last six months.
DWI with stroke, look what it's been able to do. And so I ‑‑ I think that MRI is an area which might provide that continuum of brain injury that we could express numerically to use as a meaningful surrogate marker in future studies, again, thinking ahead to 2010.
DR. MARLER: I think in the meantime, kind of the consequences of what I was saying, is that if there is a neuropsychological outcome, it might begin as a more or less ‑‑ something like a six-point scale. And then, at the upper end of functioning, it might have to break up into many smaller points, depending upon the outcome that you're looking at.
But I still think that you'd get a lot of information from a relatively simple, straightforward brain function scale. I don't know, I'm thinking of a Rankin scale or any similar scale that assesses a person's ability to function.
And I think what's more important for clinical trials is the reliability of those scales. The ability of different people evaluating the same patient at the same time to get the same scale value is incredibly important as to how well ‑‑ how much power you get out of that particular outcome.
ACTING CHAIRPERSON MAISEL: So it seems we've discussed a number of different scales, all of which have shortcomings, many of which may be acceptable under the right circumstances, such as with the goal being that they be accurate and reliable in the patient population that's being studied, if that's possible, although it sounds like that may not exist right now.
DR. MARLER: Dr. Brott and I are both familiar with the statistical methodology of global outcome measure that combines several different measures. You could combine, you know, the presence of lesions on an MRI scan with a Rankin scale with a Trail Making B Test. And it gives you the ability to kind of increase your power rather remarkably.
ACTING CHAIRPERSON MAISEL: Dr. Normand?
DR. NORMAND: I did bring up the statistics first, so I'm going to interject right now in terms of the analysis of using the scale measures. And yes, indeed, there are novel methods, statistical methods, to include multiple measures, either serial measurements combined with survival as well as combined with more biological measures.
I think, you know, this is all important. However, I do think it poses a challenge to the sponsor of such a trial to do the power calculation for that analysis. And while it could potentially provide a more powerful study, I do think that the FDA needs to be open-minded about looking at such calculations in order to: a) identify a clinically meaningful improvement when you have a multivariate type outcome.
And it's not that it's not doable; it's doable, but I don't ‑‑ I've not seen such calculations done where people around the Panel will say, "Well, I understand that," because actually it's enormously complex. It's right, but yet it's complex.
The other thing that I'd like to mention ‑‑ and Dr. Lazar mentioned this when he was speaking earlier ‑‑ and that has to do with the fact that when we have ‑‑ a question was raised by one of the panelists about the missing data. We do have an issue with Items A and B, and that is survival and the outcomes. They are related. Obviously, if you're dead, you can't be measured on a particular outcome. And so we have to worry about informative censoring.
So there are other statistical issues that are raised when you consider both A and B. I actually vote for having both A and B. I think you should consider items in addition to survival when you're doing such a trial. I'm just recommending that we need to be more open-minded about the type of analyses that are done when you have such data available.
And, moreover, I think there needs to be a little more broad-mindedness regarding the power and sample size calculations that are provided in such a trial.
ACTING CHAIRPERSON MAISEL: Dr. Somberg.
DR. SOMBERG: Often these trials are small, and, in fact, until recently I think one of the problems in resuscitation work was that the small numbers of patients in many of the critical trials ‑‑ I think it is to be encouraged that you have larger n's in this area of study.
With that said, I think that emphasizes the need to have the secondary endpoint evaluation simple. So, you know, it's nice to do some of this work, but I think it may be up to the NIH to do multiple MRIs, etcetera, in a circumstance, while to develop certain devices one has to come up with simpler measures of neurologic function. Otherwise, these studies may end up with 20, and then you'll be faulted for the number, and you really have to have ‑‑ the number is what I think is most critical.
DR. BROTT: I would agree with that. And I was actually thinking just before you spoke that ‑‑ can you justify MRI after an arrest today as standard of care? And I think we're getting close to that. And in terms of the companies needing to pay for it, for example, but I think we're getting close. And I certainly predict by 2010 MRI would be standard of care, and so would not be ‑‑ necessarily be part of the research budget.
It might be tricky, but I think that it's a technology we could still use, like a lot of the standard of care things that you're getting already.
ACTING CHAIRPERSON MAISEL: Why don't we move on to Part C.
MS. WOOD: "Are there clinically acceptable surrogate endpoints that can be used? For example, return of spontaneous circulation, 24-hour survival, hemodynamic improvement, quality of life, work status, functional status, etcetera. Should these surrogates be primary or secondary endpoints?"
ACTING CHAIRPERSON MAISEL: While much of our discussion focused on this issue, it seems like we're a little bit divided. Certainly, at the one extreme, Dr. Becker's example of a device that calls 911, I think we could all agree would be acceptable, although I doubt there would be a PMA for that device.
I sense that we're split over the things such as hemodynamic surrogate endpoints and, Bram, I wonder if you have what you need, or do you need more specific comments?
DR. ZUCKERMAN: No. The agency can always consider surrogate endpoints, but the real question is, at the end of the day, for a particular device with labeled indication: have you shown in your package reasonable assurance of safety and effectiveness and clinical utility?
So let's go back to the Weisfeldt example, the last example that Dr. Weisfeldt gave, where he utilized a mid-level surrogate, but then the real outcome of interest went in the reverse direction.
So that's a problem for the FDA at the end of the day. And certainly if a sponsor were to proposal some type of surrogate endpoint, due to the argument that some have posed that we're trying to treat one portion of the chain of survival, I think still the agency would need to see how the chain finishes up at discharge from hospital, hospitalization discharge rates, what's happening, what power we have for some of those secondary endpoints to look at negative trends, etcetera.
So it's not a simple question that you're asking us.
DR. ORNATO: I have a question.
DR. ZUCKERMAN: Yes.
DR. ORNATO: I have a question for you, Bram. What I think Mike elegantly put forth as a hypothesis is this whole issue of a strategy trial versus a single-drug or a single-device trial. And that's a hot topic, and resuscitation, as most of you know, these days.
And we look at AIDS as an example, where multi-drug cocktails are used or Hodgkins' Disease or childhood leukemia or even some of the carcinomas ‑‑ breast ‑‑ where strategies that use in some cases surgical intervention followed by X-ray therapy and/or chemo in some combination are now in many cases appropriate therapies that are advances.
Here's my question. Non-FDA ‑‑ already non-FDA approved interventions, like a new CPR device, in combination with a drug, or perhaps two drugs, may be required for a sensible strategy going after the physiology. Is that something that could be handled by the existing regulatory process?
In other words, suppose a trial is done, and in the end a cocktail with three totally unapproved things ‑‑ a) would it be allowed to go forward? Is there a way? And b) in the end, could you give each of those three things labeling? Obviously, it would have to be in concert with the other two. But do you see the problem?
DR. ZUCKERMAN: Yes.
DR. ORNATO: Is it solvable?
DR. ZUCKERMAN: Fortunately, we have an office of combination products, and we're already starting down that road with the drug-eluting stent approvals. And certainly we'd be interested in talking with any potential sponsor that wants a device and drug approved simultaneously, and all of the, you know, labeling quandaries.
But I think ‑‑ let's perhaps keep it a little bit simpler and get back to the way you started your response to my comments. I'm glad you mentioned the AIDS example. It's a great example where an appropriate surrogate was developed for trying to improve the efficiency of treatments.
However, I think one needs to recognize that there was a lot of work developed by NIH and others of putting data sets together to really validate the surrogates, etcetera. One problem perhaps I still see is in our discussion here of the literature, the quality of these surrogates, is still questionable. And that becomes a question, then, for FDA at the end of the day, and it goes back to Dr. Weisfeldt's points.
DR. HALPERIN: Let me just make one ‑‑
ACTING CHAIRPERSON MAISEL: Dr. Halperin, and then Dr. Becker.
DR. HALPERIN: One quick point on that, because although I know cost is not supposed to be an issue with the FDA, but, in fact, cardiology trials have been all over the map, anywhere from a few patients to many, many tens of thousands of patients.
And cardiac arrest trials can be very difficult to do, and, in fact, a number of trials that have shown improved return to spontaneous circulation and 24-hour survival probably would have shown improved long-term survival if the studies were much, much larger, because, in fact, some of the differences in the survival to discharge group have been much larger, in fact, in differences that have been shown in other areas like in survival using thrombolytics.
So I guess part of this issue is that, is it the agency's mission to really make megatrials happen in cardiac arrest research where nobody is actually going to do them? And then, are we going to then potentially be in the situation of denying potentially very useful advices simply because there's not enough money to actually do a trial that's big enough?
DR. BECKER: I'm wondering if I can just ask a question. And it follows up on something that Dr. Yancy raised, which is just, have we used the post-market data to sort of our full ability to use? And I'm just wondering how the agency feels in terms of our current use of post-market.
In other words, we're very, very concerned that sort of this ‑‑ this whole demonstration of effectiveness and, you know, as we move into sort of the next decade with more and more devices and possibly combinations, and very complex types of potential physiology, you know, is the post-market, you know, an opportunity to provide additional input to the agency on a device?
DR. ZUCKERMAN: The agency is always concerned with finding the right pre-/post-market balance. And certainly in response to Dr. Halperin's concerns the agency is also ‑‑ Center for Devices and Radiological Health is also obligated to find a least burdensome way for device manufacturers, etcetera.
But by the same token, let's take the example that you've posed, Dr. Becker, of, can post-market be a substitute? No, it can't be a substitute if up front we haven't met our specified goals of reasonable assurance of safety and effectiveness and clinical utility.
But certainly that benchmark may not be perhaps, you know, an improvement in the rate of neurologically intact patients who are in the hospital. And, yes, I think there definitely is a role for better development of post-market study of these devices to appreciate the iterative way that these devices need to be developed.
Certainly, our experience in the past has been that once a product is approved, generally the manufacturers may lose interest quickly in doing post-market studies. But that's why we're here today about trying to change the paradigm.
ACTING CHAIRPERSON MAISEL: Why don't we move on to question 3, please, Geretta.
MS. WOOD: "Based on the clinical endpoints discussed above, what length of followup should be considered for the efficacy endpoints? If a device was associated with a 24-hour survival advantage, but did not improve hospital mortality, would this device be considered efficacious?"
ACTING CHAIRPERSON MAISEL: It seems we already touched on this topic quite a bit. There can be numerous endpoints along the way ‑‑ resuscitation, and with return of spontaneous circulation, hospital admission, hospital discharge, etcetera. It sounds like from the FDA perspective, and perhaps from the Panel perspective, hospital discharge alive is an important endpoint.
Do people feel that's something preceding hospital discharge as an acceptable endpoint? Hospital discharge alive?
DR. SOMBERG: Well, what I said before, I think under certain circumstances for certain types of devices we might ask less. You know, survival one hour and 24 hours might be a consideration when other things are to be followed. But you wouldn't want to base it on just some sort of a priori surrogate, which sounds physiologic.
But I think to ask for, you know, what happens 30 days later, and it's often 30 days to get out of ‑‑ 10 days to get out of the hospital is ‑‑ there's a tremendous chain of events that's occurring. And I don't know if you want to have cardiac surgery and all other considerations confound that.
But that said, it's important to find that data out, because if it doesn't do any difference, or markedly negative, that may impact on the use of the device.
ACTING CHAIRPERSON MAISEL: So how do you feel about the statement here, "If a device was associated with a 24-hour survival advantage, but did not improve hospital mortality, would this device be considered efficacious?"
DR. SOMBERG: I would say with certain panels with certain devices with a PMA, it may be efficacious. But not if it's aimed towards an intervention once you get to the hospital for long-term outcomes.
Let's say you had something to improve neurologic function, whichever way you would do it, and you gave it the ‑‑ it made no difference on how people: a) got out of the hospital, and b) how they neurologically functioned, the small subset that got out had no difference in neurologic ‑‑ it would be a negative study, even though they may show some changes like MRI or brain scan, reduced infarc, reduced ‑‑ whatever.
So it's very hard. You know, you're asking to be pinned down in advance, and to write sort of like a check without filling in the amount.
ACTING CHAIRPERSON MAISEL: Dr. Brinker.
DR. BRINKER: One might get the impression that what we really are looking for is a magic bullet that we can utilize in any scenario where a patient has an arrest in a hospital, out of hospital, untimed or anything else.
And that magic bullet would bring about a prolonged survival or at least post-hospitalization survival, in a neurologically intact person. And, theoretically, anything short of that would be not acceptable.
And I think what we do when we think in those terms is to exclude the building blocks, the iterations that usually go into eventually the development of something close to a magic bullet or combination bullet, that can achieve this. And I would hate to think that we're frustrating progress or evolution by insisting on too high a threshold for acceptance, even though that threshold would not seem to be ideal or even satisfactory.
Again, this brings up the issue of efficacious versus clinical utility. Something that keeps you alive while you're doing it may be efficacious, but it's not clinically useful unless it keeps you alive way after you stop doing it. And if that's going to be a significant obstacle, we may be delaying the evolution of the ultimate magic bullet, if such exists.
ACTING CHAIRPERSON MAISEL: I think it also leads nicely into our next question, which is the balance of effectiveness with safety, which may also have an impact on what are acceptable endpoints and time endpoints.
MS. WOOD: "What should the primary composite safety endpoint include?"
ACTING CHAIRPERSON MAISEL: Jeff.
DR. BRINKER: I'd like to just make a shot at some things. I mean, clearly, it shouldn't cause ‑‑ a study drug or device shouldn't cause increased damage in a system over the control. I think that would be the primary safety endpoint.
Assuming that there is no other ‑‑ assuming that there is either equal efficacy ‑‑ that there is equal efficacy, the two modalities, or superior efficacy in the study arm, the safety part of it would be an issue that the study arm shouldn't cause more damage than the accepted or standard arm.
ACTING CHAIRPERSON MAISEL: Dr. Weisfeldt.
DR. WEISFELDT: I'd also argue that from a safety point of view the ‑‑ having data on safety through the period of hospitalization as a reasonable standard. Thinking about the possibility that there may be a device, may be a drug ‑‑ epinephrine ‑‑ where ROSCs may be improved because you improve myocardial blood flow, but you diminish gut blood flow, it's well documented that Gram-negative sepsis is a common accompaniment in the first week or so after resuscitation with GI organisms.
So it could be that the effect of a drug or device may be to improve 24-hour survival, but yet from a safety point of view or a complication point of view, serious complications may occur in that period between 24 hours and discharge from the hospital that would be undetected if we weren't looking for or insisting on safety in that period of time. So I ‑‑ it just goes back to the same theme that safety issues are more than 24 hours.
ACTING CHAIRPERSON MAISEL: Joe.
DR. ORNATO: I think that Dr. Lazar has enlightened all of us, and I'm personally very appreciative because I've learned a lot from his comments and summary this morning. Many of us previously ‑‑ and I'll speak for myself ‑‑ thought OPC/CPCs were perhaps a little better than they really are.
But I think the concept of neurologic outcome as a safety measure is clearly at the top of what I would think my list would indicate. I think the idea of having some measure ‑‑ obviously, we need a better one ‑‑ of neurologic outcome, as well as overall performance measure return to some kind of functional status has in the CPC/OPC ‑‑ is still probably valid, although I think I've been convinced this morning that we clearly need better tools.
The final piece of this, though, is that in this really unusual area of clinical medicine and emergency critical care, I think one could make an argument that death is not the worst outcome. In fact, I would like to make that argument.
I think a severely neurologically impaired person who survives for a significant period of time, but without what I think most of us would consider even to the naked eye meaningful existence to themselves or those around them, is arguably far worse than a merciful, rapid death.
And so I think as the agency looks at this complex question, one of the paradigm shifts that explicitly I think is obvious to me is that perhaps some kind of a scoring system, as the question is begging might be worth researchers considering and worth the agency considering looking at. And that it may very well have a rank that's very different from most other clinical trials.
ACTING CHAIRPERSON MAISEL: Dr. Yancy.
DR. YANCY: The safety issues are largely device-specific, whether it's a resuscitative device, a defibrillatory device, a chest compression device, etcetera. But I think along the same lines of the safety referable to the device is the safety of the application. That is to say, there is a learning curve. There is a teaching phenomenon that has to occur.
The greatest example would be the minimally-invasive open chest massage. I mean, the technique might be perfectly appropriate and may do a great job getting your surrogates met. But if the learning curve is so steep, then it becomes a problem, because any device or platform that is developed will be widely distributed to people with widely disparate skill levels and aptitude levels.
So I think any safety equation is not only the safety of the device per se but the safety of the application.
ACTING CHAIRPERSON MAISEL: I think that's an excellent point. So, in general, it sounds like our primary safety endpoints are also our effectiveness endpoints and mortality neurologic outcome. There are some device-specific endpoints such as bleeding or infection for hypothermia devices, which we'll discuss later, etcetera.
So we discussed ‑‑ well, you can read question 5, if you want.
MS. WOOD: Okay. "What length of followup should be considered for the safety endpoint?"
ACTING CHAIRPERSON MAISEL: And we've discussed this as well. It sounds like at least through the hospitalization would be a reasonable endpoint for that. So now why don't we try to wrap things up pre-lunch and go quickly through the study design. And I think I'll just have Geretta read the questions, and we'll go right into them.
MS. WOOD: "Can a scientific study be performed using a single-arm study with historical controls, U.S. and/or OUS? If so, should the historical controls meet the same inclusion/ exclusion criteria? Do you have any suggestions on how to reduce bias if historical controls and/or OUS data are used?"
ACTING CHAIRPERSON MAISEL: John.
DR. SOMBERG: I don't think historical controls could be used in this area very effectively, and it would be nice to say so, but I think things are changing. There are differences especially with different data sets from Europe versus the United States who were, we heard, who participates.
So I really don't think that's the case. I really think you need a randomized group. A company would be risking too much on differences, and I think it's more risky for the company versus cost than it is even for the ‑‑ you know, that's ‑‑ that would be even worse than the bad scientific data that's collected. So I strongly recommend randomized control trials.
ACTING CHAIRPERSON MAISEL: Dr. Normand.
DR. NORMAND: I heard today that time is a big confounder, that things are changing rapidly. And if that is, indeed, the case, then I don't see how you could use only historical control data. So it seems to rule out the use of historical control data.
I would say, though, that Dr. Somberg was using the word "randomized" and "historical control." I think there's a difference. You could have prospective non-randomized studies. It's a difference between historical control and ‑‑ they're two different things. So let me just add that correction.
DR. SOMBERG: I'm advocating both, prospective and ‑‑
ACTING CHAIRPERSON MAISEL: Anyone disagree? Okay. 7?
MS. WOOD: "If the study is unblinded, do you expect any substantial positive or negative placebo effect, or an effective investigator bias on patient selection or endpoint evaluation? If so, how can these problems be minimized?"
ACTING CHAIRPERSON MAISEL: Dr. Brinker.
DR. BRINKER: Well, it seems like 90-plus percent of these studies have to be unblinded to the operator. It's going to be hard to do any kind of CPR study that's double-blinded. So I'm not sure how that exists.
But I do think in non-randomized studies, which gets back to John's comments, as opposed to contemporary non-randomized studies, if you have a randomized study it sort of helps lessen the impact of operator and investigator bias than if you have contemporary non-randomized trials.
ACTING CHAIRPERSON MAISEL: Dr. Normand.
DR. NORMAND: Do you have a comment?
DR. MARLER: I was just going to say one common way to deal with reducing the bias, which can be immense, is to have an observer of the outcome later on who has no way of knowing how the patient was treated as completely independent from the study, and in which someone monitors the ‑‑ you know, is sure that the patient isn't ‑‑ understands they can't tell ‑‑ if the patient happens to know how they were treated, can't discuss that.
So a blinded independent observer in some ways affects this, but the ‑‑ you have to worry, though, in that case that there's biased ancillary or secondary or followup treatment introduced in the care of the patient by physicians who know how the original treatment was given.
DR. NORMAND: That was the point I was going to make. And an additional point I was going to add is that I think that most trials do this, and even observational studies do this, is to have a good data collection system in hand in order to measure sort of the degree or the intensity of treatment that is indeed applied. So at the very least, you have it measured.
ACTING CHAIRPERSON MAISEL: Dr. Somberg.
DR. SOMBERG: I think with all this said, attempts should be made to try to intervene in both arms of the study to blind ‑‑ for instance, if one had a device for cooling, it may be important to hook up that device and to do everything but to bring the body's temperature down four degrees.
It may be important to put something ‑‑ you know, to take ‑‑ to stop whatever you're doing, put something on, and provide some sort of hemodynamic augmentation, even though it doesn't provide the right augmentation, because all that may certainly confound the study, and the less confounding you have the better off you are.
ACTING CHAIRPERSON MAISEL: Henry.
DR. HALPERIN: Yes. This is an issue where disinterested third parties should probably be the people that are actually doing the CPR. This is one of the disadvantages that have occurred or at least been perceived to have occurred in in-hospital studies where the actual investigators actually doing the CPR.
I think there is inherent propensity for bias in that situation, where people who have really no particular interest in the study ‑‑ financially, intellectually, or anything ‑‑ would be the appropriate people to then do this. And I think that mitigates some of the bias that could be otherwise potentially there.
ACTING CHAIRPERSON MAISEL: So I think the general message is blind whenever possible, certainly blinding the outcome assessments such as neurologic evaluation.
Dr. Zuckerman, did you have a comment?
DR. ZUCKERMAN: Yes. I'd like the CPR experts to respond to this point. Certainly, in most areas of device trials, we see negative placebo effect, meaning if you are randomized to the control arm it's a problem for the patient and the doc, even if there is no involvement of the doc in the device company, etcetera.
With CPR trials in the CPR arm, how do you assess the quality of CPR, the control?
DR. SOMBERG: Well, there may be a play of previous data, historical controls. If your control group is so much worse than what everyone else has done to show a signal, some ‑‑ there's a problem awry.
ACTING CHAIRPERSON MAISEL: Dr. Halperin.
DR. HALPERIN: Yes. There are some precedents for this, and some trials have actually had an independent third party who is actually observing the CPR, like an ACLS instructor, or what have you. And how good that is is questionable, but there were two really nice studies, fairly large ‑‑ one 500, one about 900 patients ‑‑ where ACLS instructors actually observed people doing CPR, and actually the people that did better CPR did better. So it was actually consistent.
But for most trials, there is no way to do it with the current technologies. That may change with newer technologies, but it really is a very difficult question that really is still open today.
ACTING CHAIRPERSON MAISEL: Why don't we move on to question 8, please.
MS. WOOD: "Should the trial design be a non-inferiority (equivalence) or superiority study for safety and efficacy? Under what conditions should an equivalence trial be acceptable? Limited labeling claims.
"If not inferiority, what equivalence deltas would be clinically acceptable for the safety and effectiveness endpoints discussed above? If superiority, such as new technology, would need to demonstrate improvement over current technology, are there new clinical trial designs that can be considered such as superiority for surrogate endpoint and equivalent hospital discharge rates? Or additional post-market studies on devices technologies to supplement initial pre-market approval safety and effectiveness data?"
ACTING CHAIRPERSON MAISEL: Jeff.
DR. BRINKER: Bram, I just want to get the thrust of this question correctly, because if we're dealing ‑‑ let's say a CPR device for hemodynamic benefit, there are no equivalency ‑‑ there is no equivalent, right? So it would be impossible to do that study for this category of device.
DR. ZUCKERMAN: Right. But we're talking more generally here. For example, suppose there was a CPR device that was proposed to not improve survival, but just to make CPR easier for the operator, something like that. You know, what do you need in an equivalence trial?
DR. BRINKER: Well, that would be a PMA application, because otherwise it could be cleared through a 510(k), if it was just making it easier for the operator. And the only difference is that it would be a PMA because the labeling would say, "This makes it easier for you."
I'm trying to put this in the context of, really, what we're trying to settle here. But ‑‑ go ahead.
DR. ZUCKERMAN: You know, I think in a general context there could be CPR devices that are so-called equivalent to standard CPR, and the agency would need to consider that application, regardless of whether it's 510(k) or PMA. I think that we would need to see clinical data, and it's important to recognize that, you know, 10 percent of our 510(k) applications, say, need appropriate clinical data.
So let's pose the question, you know, how do you define "equivalence" in a CPR trial?
ACTING CHAIRPERSON MAISEL: Dr. Yancy.
DR. YANCY: Well, I think the only way that we can approach that, Bram, is that we have to look at the limited metrics that are available. And so if we know what the best outcomes are for what return of circulatory stability, or if we know what the global outcomes are for either being discharged, or discharged intact, then as meager as they may be they become the benchmarks.
And so that would be the reference point for designing a non-inferiority approach, but that's ‑‑ I think that's all we can do in that regard. And there may be methodologies for which there is no benchmark, and then you've got to take a totally different approach.
ACTING CHAIRPERSON MAISEL: Dr. Normand.
DR. NORMAND: I have a question about the equivalence and the suggestion that was just made by Dr. Yancy. I guess in terms of if it was going to be equivalence, then I think understanding ‑‑ and maybe I'm jumping ahead in terms of how you actually determine equivalence, because I think collecting the data, and having the data available that tells you how much uncertainty is attached to the, let's say, devices already in the market, I think that's a very important problem.
And, hence, I would argue that to define "equivalence," either clinical or even beyond that, I think we need more than just one number. And so I'm worried about equivalence trials in terms of being able to have the information available to say what the size of the delta would be, because I'm not sure about the line in the sand and where that line is.
So we need information. We need access to information to say sort of what the current state is, and not that it's 20 percent, or whatever it is.
ACTING CHAIRPERSON MAISEL: That gets back a little bit to our discussion earlier about what type of trials would be acceptable. Historical controls I think we decided would not be acceptable. So, for example, a randomized trial might ‑‑ would certainly answer that type of question.
DR. NORMAND: Would answer that, yes.
ACTING CHAIRPERSON MAISEL: Why don't we move on to question 9, please.
MS. WOOD: "Discuss the possibility of developing a registry and using the data for future studies. What are some of the necessary data points that should be collected, keeping in mind the use of this data as historical controls for future studies?"
ACTING CHAIRPERSON MAISEL: Sharon.
DR. NORMAND: Okay. I know we all want to go for lunch, but I actually sat down and thought about this just because we do a lot with ‑‑ as I assume everybody here does, with registry data.
So I thought I would take the first crack at it and sort of giving some suggestions in terms of key items that I thought should be in there. But let me throw out the first one. I think it needs to be auditable. You have to be able to audit some of the elements in such a database. And I'm going to say that as the first point.
We could talk about I'm just sort of saying in an ideal world. I think the other thing that would be important to be in there, given that it's been mentioned about confounding over time, is I think that the information that needs to be in such a registry have to be collected in detail that we had information with regard to potential confounders that relate to time of entry, time on the market, when it's done, things of that nature.
And in addition to the ‑‑ sort of the usual character ‑‑ and people are going to say the usual things in terms of the characteristics of the patients, the characteristics of the outcomes that are measured.
I do think that in having a registry, in order to tap into some of the information that's going to be needed to supplement either a new design, a new trial, or to understand what's happening currently, I do think a lot of information with regard to the operators of the device and information about the timeframe in which the treatment is undertaken.
And then I think, lastly, I ‑‑ I know this is a little bit technical, but I think if we're going ‑‑ if one is going to use a registry in order to infer, let's say, effectiveness, then I think one needs to think about including elements that perhaps one wouldn't normally think of including.
And what I mean by that are I think we need to find data that would be related to who got what treatments, but not related to particular outcomes. And for those of you that know what I'm talking about, we need to find an instrument. And so there needs to be the usual cast of information that is typically in a registry, but I guess I'm arguing for it to include probably much more detail about the operators, about the site where it's being ‑‑ where the data are being collected.
And then, I guess lastly I'll just emphasize that I do think you need to have the opportunity to be able to verify the elements in the registry.
ACTING CHAIRPERSON MAISEL: Dr. Becker.
DR. BECKER: Yes, if I can just maybe add something. It seems to me that one of the potential useful roles for a registry would be to look at adverse events that are not obvious initially, and that there may be a real role for registries in the specific sort of demonstrating or describing adverse sorts of events that are not clear and in early trials.
And the other thing that ‑‑ where I think registries may be very useful is for the setting where there's a very ‑‑ if you will, a very narrow indication for some type of a device. And the example that comes to mind is sort of like the pediatric defibrillation pads, that ‑‑ you know, that was ‑‑ that was a device that has been approved, and we knew that there would just be a very small number of patients that would really be appropriate for the use of those types of pads.
And I think it is to the agency's credit that what ‑‑ as I understand, was really what came out of that was the notion that there was initial safety data that was presented, and then the idea of an ongoing registry to follow what ‑‑ how that data would accumulate, because if we required ‑‑ you know, if the bar was a demonstration of survival, you know, we wouldn't have pediatric pads for another 100 years basically. I mean, and that's ‑‑ and I mean that in terms of 100 years.
So I think it's very important that we think and we think creatively in terms of how we can sort of optimize this balance of, if you will, pre-market/post-market type of data, so that we can ultimately do the best that we can for all of the citizens.
ACTING CHAIRPERSON MAISEL: Joe.
DR. ORNATO: Thanks, Bill. Firstly ‑‑ two points. One is, of course, registries do exist. NR-CPR is the one that has been mentioned a couple of times today ‑‑ in hospital, fairly large. Out of hospital was driven initially by the so-called Goodstein ‑‑ first set of Goodstein guidelines. There are now pediatric Goodstein guidelines and a number of other derivatives.
NHTSA developed a DEEDS database, and that, to some degree, has influenced an out-of-hospital project that National Association of EMS Physicians have underway.
But if you look critically at these registries, the problem has not so much been defining the data points. They need to be improved ‑‑ everyone knows that ‑‑ and they are being improved. But we have a really good starting point.
The problem is they're all biased. They're all inadequately numbered in terms of numbers of cases, and they tend to be a fairly small, fairly biased, select group of sources. For example, NR-CPR has only 400 hospitals in it. The out-of-hospital project that's underway is still really in a pilot phase.
And particularly for the out-of-hospital registries, we're still struggling with this whole issue of: does HIPAA apply? And if so, when and how? And out-of-hospital registries are struggling trying to figure out if there's a lawful way for them to get ‑‑ even simply survive to hospital discharge data. Forget any kind of neurologic outcome.
So the first issue is registries do exist. The methodologic problems do exist, but I think they are solvable. The broader issue is one of society at large really trying to figure out whether we want cardiac arrest to move forward like cancer and trauma have.
They've both been driven greatly by the presence of registries, certainly cancer has, and certainly I think we would all have to accept trauma has as well with the need for trauma registries that are a requirement of our trauma centers nationally. So that's the first point.
The second point is related to it, which is this whole issue of the philosophic, perhaps governmental choice, and that is whether an agency like CDC, which I think is looking at this right now from our conversations with them, should really champion, perhaps with the help of others, this notion that cardiac arrest should, in some form, perhaps certain categories of it, be reportable.
That would tear down a lot of the HIPAA problems from the get-go, if a governmental entity required it. It would also deal with a lot of the political and operational issues that hospitals are struggling with whether they want to pay the price tag for the data collection and the filling out of forms, etcetera, etcetera.
So that's my two cents. I think it's a very important question you're asking, and I think if it's possible to create a registry by whatever number of years, but certainly by the Healthy People 2010, that would ‑‑ that would be a major jumpstart to a lot of things. And it would certainly, I think, help the FDA's effort.
ACTING CHAIRPERSON MAISEL: Why don't we move on to the final question.
MS. WOOD: "Can uniform definitions of adverse events be created? If yes, by whom?"
ACTING CHAIRPERSON MAISEL: Should we take that as a no, they cannot be created?
DR. SOMBERG: Take it as a yes. I don't know why I'm ‑‑ but take it as a yes, they probably can be created. And it relates to the very interesting discussion the last five minutes. Is there some registry or group ‑‑ maybe the CDC or someone else would need to do some standardization, because it's going to be important. If you have a registry, but you don't have definitions, what are you recording?
ACTING CHAIRPERSON MAISEL: Any other comments? Yes, Al.
DR. HALLSTROM: This actually goes back to 8.c. I had a comment here, but I forgot it. The question was whether you could use new trial designs such as a surrogate for the endpoint with equivalent hospital discharge. And I think that sort of thing should definitely be explored.
I've just looked at some of that recently, and, for example, a number of trials that were showing up here with non-significant hospital survival, but significant hospital admission, had you used a bivariate endpoint so that in ‑‑ which is essentially what's being advocated here, those probably would have been significant bivariate endpoints in the right direction.
The problem, of course, when you go to a bivariate endpoint is if you end up somewhere off ‑‑ not on that diagonal where you have the same proportional hospital effect, you don't exactly know how to interpret what you're going to do with it. But in terms of sample size, for example ‑‑
DR. ZUCKERMAN: If you don't know how to interpret it, perhaps another trial.
DR. HALLSTROM: Right. Maybe another trial. But in terms of sample size, if you're planning for that proportional hospital survival, one of the studies I looked at you had to ‑‑ to look at hospital survival per se, you had to have almost a three-fold increase in sample size over hospital admission.
To look at the bivariate endpoint, you needed about a 20 percent increase in sample size. So I think there are considerable savings, and these kinds of things should be looked at carefully.
ACTING CHAIRPERSON MAISEL: Well, thank you very much.
Why don't we adjourn for lunch, and we'll reconvene in one hour at 2:05.
(Whereupon, at 1:05 p.m., the proceedings in the foregoing matter recessed for lunch.)
ACTING CHAIRPERSON MAISEL: Good afternoon. I'd like to begin this afternoon's session in which we'll be discussing CPR and hypothermia. I invite the FDA to begin their presentation.
MR. FELTEN: My name is Richard Felten and I'm a reviewer in the General Surgical Devices Branch, and what I'm going to do is just to do a very brief overview of the devices that are now available for induction of hypothermia. Next slide.
The most common device, obviously, that everybody's really familiar with of course are the cooling blankets. And these devices basically include blankets or even the more complex things like chest vests or leg wraps and so on. And their intended purpose is to simply provide surface cooling using the circulating air or water. And these devices have indications for use that are basically to induce hyper- or hypothermia, or for localized temperature therapy.
Next type of similar product are the cooling surfaces. These are actually cooling tables on which patients are simply laid. And in this case, the device simply uses circulating cold air to make the table and the surface cool. And these devices are marketed simply for use in treatment of patients who need to be cooled.
Another device that can be used although it is not usually used of course are -- next slide please -- simply cold packs. These of course would be ice bags, or the standard kinds of cold packs that we usually use for localized cooling. And these are actually marketed simply for cold therapy for body services. And these also could be used.
The next device actually is the external heat exchange systems. And in this case you have a series of individual items that are put together to provide the hypothermia. You have the heat exchanger which is intended to cool or warm blood. You have circulating pumps that circulate the blood from the body externally to the external heat exchanger, and then back into the body. And there is a system controller that regulates the whole process which is involved in regulating the blood flow, monitors the temperature, and maintains either the cooling or the warming, which can be done as required.
The most recent and the newest product are the endovascular cooling systems. These systems are essentially the same as the heat exchange system in that again you have an external heat exchange system, you have circulating pumps, and you have central controllers. However, in this case, the cooling is actually done internally in the body. In this case, the catheter is used, placed usually in the vena cava. Cold fluid is circulated through these catheters. There is a heat exchange system at the very tip. And the blood is actually cooled within the body itself.
When these were introduced or brought to our attention, we were somewhat concerned in this case that there may be some differences in safety profiles between internal cooling of the blood versus the external cooling through the external heat exchangers. And in this case, the agency determined that we did have some questions about safety, and therefore, as you'll notice up here, instead of having just general plans for hypothermia or maintaining temperature, these devices actually have very specific indications for use. You have a cardiac surgery plan for achieving or maintaining normothermia during surgery, recovery, intensive care. You have an induce, maintain and reverse mild hypothermia in neurosurgical patients during surgery, peri-intensive care. And you have a fever reduction in a specific patient population as adjunct to antipyretic therapy in patients with cerebral infarcts and intracerebral hemorrhages, which require access to central venous circulation.
And all three of these indications were all based on clinical trial data, and which we did look both at the effectiveness of the device to achieve or maintain the production of the hypothermia, or maintaining temperature, and looked at safety. And it's this issue, or this is one of the issues that we're asking the panel to discuss as part of their discussions today on whether or not the endovascular cooling systems do or do not present different questions compared to the external cooling systems, and is the mechanism of action of producing hypothermia the same or different between these two systems.
And I'd like to now introduce Dr. Julie Swain who will actually give a presentation on some of the clinical information about hypothermia. Thank you.
DR. SWAIN: We're going to switch slides here for a second. Pleasure to talk to the panel today. And I'm going to talk about the topic of post-event hypothermia because some of the devices that we're looking at, perhaps there's been interest in the community of looking at these devices for post cardiac arrest hypothermia.
And I'll introduce the topic by talking about some trials, and myocardial infarction, and acute head trauma, then look at post-event hypothermia and resuscitation. Talk about the ILCOR recommendations that Dr. Collins talked about earlier this morning, the clinical studies upon which those are based, and then kind of an executive summary of the questions for the panel. Next slide.
Well, in myocardial infarction, there was the COOL-MI study. And this is all from data presented at TCT last year, and it can be found on the Web under the Web address listed here. This was a prospective randomized trial, reasonable-sized, and it was cooling with one of the endovascular IVC catheters, versus normothermia during percutaneous coronary intervention. They had to have a myocardial infarction less than six hours prior to it. It was normothermia versus 33 degrees for I believe it was 24 hours, not three hours. That's a mistake. Or no, excuse me, it was for three hours, sorry. And it had a relatively quantitative endpoint, a surrogate, which was infarct size at 30 days by spec. Next slide.
And what this showed was that there was really no difference in the endpoint between these two studies. Notice that the N for hypothermia in blue on all these slides, and normothermia in red. The N was 177, which is a reasonably sized device trial. But again, no statistical difference. Next slide.
What is very interesting to us is that this study was not powered to detect individual safety events. But when you looked at the main serious adverse events, there was a trend towards a higher death rate, vascular bleeding, DVT, shock, pulmonary edema in this, although none of these were significant. But in a reasonably sized trial, this was sort of an interesting trend. Next slide.
Well, what about the brain injury studies? And Guy Clifton at UT did a reasonably large study on acute brain injury. This was again prospective randomized trial, normothermia versus hypothermia, 33 degrees for 48 hours, injury less than six hours old. They used surface cooling and GI cooling. Endpoint was Glasgow Outcome Score at six months. And then they had a series of secondary endpoints. Next slide. Which were many of the tests that Dr. Lazar spoke about this morning. Next slide.
And the results of the trial, absolutely no difference. And again, an N of 199 on the hypothermia group, a fairly large trial. And non-significant for all secondary endpoints and the primary endpoint. Of course, there's a lot of Monday-morning quarterbacking, or today it would be Tuesday morning for Minnesota, but whatever. Yes you can design a trial that's different, and maybe these two trials, the COOL-MI and brain injury could be used to set up a hypothesis for subgroup testing in another randomized control trial to see if in some subgroups it might be advantageous. But again, the two largest trials in these two organ systems find no difference. Next slide.
Again, in Clifton's trial, it wasn't powered for individual adverse events, but they had a statistically significant difference in critical hypotension, more in the hypothermia group, as well as bradycardia with hypotension. And the percent of hospital days with complications were all higher in the hypothermia group. So it brings to some of the safety questions. All three of these trials bring to some safety questions that we may well have in the future. Next slide.
Well, what about post-event hypothermia and cardiac arrest? And as Dr. Collins said, there's two randomized control trials that were published in New England Journal two and a half years ago. One I'll call the Australian trial, and the other which was a multi-centered trial in Europe primarily from five countries. Next slide. Based on these two studies ILCOR had the following recommendations about unconscious adult patients, spontaneous circulation after out-of-hospital cardiac arrest, cooled to 32-34, 12 to 24 hours when the initial rhythm was VF. And then a level 4 recommendation based really totally on anecdotes such cooling may also be beneficial for other rhythms or in-hospital cardiac arrest.
And what I'd like to do now is look at the data upon which these recommendations are based. Next slide. Well, let's compare these two studies. The location, Australia and Europe. And as Dr. Zuckerman said this morning, we've had multiple instances where European or out-of-U.S. data is really not consistent with in-U.S. data. You have to look are these the same kind of patients, are the EMS systems the same? In one of these studies, I believe it was an average of two minutes from the 911 or 911-equivalent call to the EMS service getting on scene. I can tell you in San Diego you can add a zero to that very often because it's a different health care system than in other places.
Inclusion criteria in these two studies. If you're going to use two studies as essentially a meta analysis to support a recommendation, well, you can look at the inclusion criteria, and they're very different in these two studies, especially in the Australian study. Women greater than 50 years old were the only ones included, the only females included in the study. And much more strict criteria in the European study. It's interesting in the European study that 91 percent of the patients screened with out-of-hospital cardiac arrest were not eligible by these inclusion criteria. Next slide.
Well, what about the actual act of cooling? Well, in Australia it started in the field with cool packs. In Europe, it started on hospital admission. Cooling method was cool packs, then ice packs in Australia; air-cooled mattress and ice packs in Europe. Target duration: 12 hours in one study, 24 for the other study. Temperature: 33 for one, 32 to 34 for the other. Re-warming: active in one study, passive in the other. And you can look at some of the standard deviations and some of the curves in one of the studies, and you can see that the standard deviation goes above 38 degrees, which indicates to me that there was hyperthermia present in some of the patients. And this is in the control group. And I think we all know for sure the effect of hyperthermia on neurological function. There's a great deal of cardiac surgery literature relating to that. Next slide.
Primary endpoint. Australia, it was survival to hospital discharge with essentially this CPC-type score. And I don't know what the requirements are to go home or to rehabilitation in Australia versus what happens in the United States. It's so variable among communities in the United States. In Europe, it was the CPC score good or moderate, as Dr. Lazar spoke about this morning. Next slide.
Well, what about the results? And you can see the success endpoint of either on the left side Australia discharge home, or at six months on the right side. And when you look at the difference in control groups, which looks like different populations being studied, or different processes of care. And a p-value of 0.046 on the Australian study, which was an N of 43 I believe for the number in the hypothermia group. And if you look at the confidence limits on the bottom for the Europe study, you know, identity 1.0 is fairly close there. Next slide.
Well, you also look at mortality. And not a statistically different numbers in the Australian group. And again, when you look at the confidence limits of the European group, you know, up to 0.95. So perhaps there's a trend here, but there's also a couple of questions on how these studies were performed. Next slide.
In the Australian study, originally it was designed to be a sample of 31 patients in each group. After the study was completed, the primary endpoint was not significant. Therefore, they enrolled more patients. I can say that for the agency, the FDA, that's probably not a study design that we would think is rigorous enough. The final p of 0.046 is just that, 0.046. There was no alpha spending penalty assigned even though -- you could say there was an interim analysis, but actually it was a final analysis and then add more patients. So the question is is this really a study that shows a difference between the two groups with any rigorous statistical design. Also, seven of the patients, about 10 percent, were randomized and treated, and then dropped from the study. And you worry about selection bias in the patients dropped. Next slide.
European study also. We don't know how many patients were designed to be in that study, but they stated that they stopped the study early because of low enrollment and end of funding. We don't know if data analysis was performed and then the study was stopped, or what the status of that was. And we don't know the planned number of patients in the study. And we don't know if interim analyses were performed. Next slide.
Well, what about the adverse events? The Australian study simply says that there were no clinically significant infections in either group, which would be far different from any other post-event hypothermia study done or published. There was really no mention of any other adverse events. European study, three patients had hypothermia stop because of arrhythmia or hemodynamic instability. And we don't know the complications during hospitalization. All they reported -- next slide -- were the complications during the first seven days after resuscitation from cardiac arrest. And you can see that in every complication studied that I've listed on the top, bleeding, pneumonia, sepsis, pulmonary edema, arrhythmias, there were more, but not statistically significantly more, in the hypothermia group. And on the bottom are listed several complications that were either equal or less than one percent difference in those. Next slide.
And as Richard Felten just said, there are multiple cooling methods that the agency is looking at to develop hypothermia, from head cooling, neck cooling devices, all kinds of surface cooling devices, GI lavage, which as Dr. Witten told me is really not regulated by the FDA. There's no devices specifically for that. And endovascular, meaning extracorporeal circulation, and IVC catheters. As cardiac surgeons, we have a long 50-year history of surface hypothermia and then endovascular hypothermia, and feel that there are a great number of differences between initiation of hypothermia in those groups. And the question is would the efficacy be the same if you have a study of a surface method that shows improved liver function or something, and an endovascular catheter. Because you want to assume that the results are the same for both safety and efficacy. Final slide, please. Second to the last slide.
Well, the ILCOR made the recommendations that I showed you at the beginning based on these two studies. And you really need to judge whether you think that one would base blanket recommendations on those two studies. But finally they stated that future research is needed to determine optimal duration, and target temperature, and rates of cooling and re-warming. As you can see from the two studies that were looked at that they used different durations of hypothermia, methods of getting there, and all kinds of things. And so the question is do those two studies really give you something of which you can base objective performance criteria, things like that. Next slide, which is the final slide I think.
So, we have a series of questions for you that Dr. Ogden will read for you. And essentially there are two kinds of questions. Do you think that post-event hypothermia is the standard for treating out-of-hospital cardiac arrest patients in the United States? I did a little survey in the San Diego area, and it's not the standard in any of the hospitals that I surveyed there. And I believe we've done a survey here in the Washington area and found one hospital that routinely does that. So do you think that the data, essentially these two randomized studies, lead you to believe that one accepts the principle of post-event hypothermia in cardiac arrest patients. And then second, if you do accept post-event hypothermia, would surface induced hypothermia be equivalent to -- in safety and efficacy -- to endovascular induced hypothermia. So these are the questions that we would like you to discuss this afternoon. Thank you.
ACTING CHAIRPERSON MAISEL: So before we get to the questions for the panel, first I'd like to ask Dr. Witten to introduce herself since she has joined this afternoon's panel.
DR. WITTEN: Yes, thank you. I'm Dr. Celia Witten. I'm the division director of the Division of General Restorative and Neurological Devices in the Office of Device Evaluation at FDA, which would be the reviewing division for cooling devices with labeled indication for post-arrest hypothermia.
ACTING CHAIRPERSON MAISEL: And at this point I'd like to ask the panel members if they have any questions for the FDA.
DR. SOMBERG: Is it appropriate to ask have there been any -- are there any trials currently under way for these devices that have resulted from discussions with FDA?
DR. WITTEN: Unfortunately because of confidentiality we wouldn't be able to answer any questions about ongoing studies.
ACTING CHAIRPERSON MAISEL: Dr. Becker.
DR. BECKER: Yes. Again, I appreciate the really thoughtful analysis on these studies. Could I just ask a little clarification in terms of the timing. Because it seems to me that one of the things that has not really been addressed is the very important consideration on timing, and the induction of hypothermia which may explain a lot of the kinds of things that you've been describing.
DR. SWAIN: I agree with that. I've sort of spent my career studying hypothermia, and especially cardiac surgery and brain protection. And you're right, I have no doubt whatsoever that the minute you take a ligation band off a coronary artery or carotid on rats, or whatever we're studying in the laboratory, and induce hypothermia right after that it does work. To some degree. Whether it's persistent is also a question. And when we look at the Monday-morning quarterbacking of all these studies, when you get a negative study, which is essentially that Australia study is a negative study, is what could have been done better. And oh, it must be because it was too long, in the head bump patients, Clifton's patients and all that. Well, then one would need to design a study where it's less long. And that's the challenge in any clinical medicine is developing a treatment that is based in reality of when you actually see these patients. These patients, cardiac arrest patients, you end up seeing a lot sooner than you'll see the myocardial infarction. I've got chest pain after lunch, and it must be the burrito I ate. Whereas cardiac arrest, you pretty much know, even though you're off by a few minutes, but you pretty much know when that occurred. So I think that is a problem, and it demands a good trial design.
ACTING CHAIRPERSON MAISEL: Dr. Yancy.
DR. YANCY: Judith, as you reviewed the COOL-MI trial, what was in our brochures suggested that for the subgroup with the anterior injury pattern that there may have been by a retrospective subgroup analysis a signal, a hypothesis that can be generated. I have a sense you might want to clarify that or refute that.
And secondly, looking at the adverse event table, it's certainly was a powerful statement you made looking at the direction of the numbers, but looking at the p-values it was only one variable that really got close to what we might traditionally describe as a trend. So I thought you might just develop both of those again.
DR. SWAIN: Yes. And as I said, there are no statistical difference in those, the COOL-MI adverse events. And the study was not powered to detect that. So no problem, we're looking at the trend.
As far as the Monday-morning quarterbacking, which is post hoc retrospective data mining analysis, then I think one can do that in COOL-MI, in Clifton's study. And Clifton's also done it, saying is there a subgroup that might benefit, and then let's design a study to test that subgroup. And I think that's very real. But you know, the problem you have on subgroup analysis is you don't know how many analyses were performed to find the one group that benefited. And Sharon-Lise will be able to comment on that I think a great deal. We very often see, and forget about cardiovascular neuro devices, let's talk about orthopedic devices or something. We very often see a failed study, and then go for subgroup analysis and find something that might be beneficial. And what we need to do is test the hypothesis then that anterior MIs will benefit from post-event hypothermia.
ACTING CHAIRPERSON MAISEL: Any other questions for the FDA? I'd like to move on to the afternoon portion of the open public hearing and ask if there's anyone in the audience who would like to address the panel on this topic today? Seeing none, we will close the open public hearing. It seems a little premature to take a break, so why don't we move on to the FDA questions.
MR. OGDEN: My name is Neil Ogden. I'm the branch chief for the General Surgery Devices Branch. And we have three questions. The first one has three parts.
There have been two randomized controlled studies reported in the literature describing the beneficial effects of mild hypothermia in a select group of patients who are comatose after cardiac arrest and have spontaneous returned circulation. In these studies hypothermia was achieved by various methods of surface cooling.
Part (a). Do you believe that the existing data in the literature are adequate to support the safety and effectiveness of surface cooling for achieving mild hypothermia in unconscious adult patients with spontaneous circulation after out-of-hospital cardiac arrest.
Would you like me to read all parts?
ACTING CHAIRPERSON MAISEL: Why don't you read (a) and (b), because they go together.
MR. OGDEN: Part (b). If you believe the existing data are adequate to support such a labeling indication for blankets and other surface cooling devices, please discuss any recommendations for the instructions for use. For example, temperature, length of treatment, et cetera.
ACTING CHAIRPERSON MAISEL: Discussion from the panel. Dr. Somberg?
DR. SOMBERG: I'll be glad to see if I can get someone excited. It's always more difficult to be an evaluator on the basis of some data, as opposed to having two sides make an argument and then try to judge that. But from what I saw, I thought there was interesting data. And it certainly wouldn't be an area I would tell someone interested in becoming involved in investigative studies to avoid because there's nothing that looks positive here. But at the same time, I do not think it rises to the level of enough information to support an application for a particular device or therapeutic approach. I was struck by the small number of patients involved given the size of the problem, and the difficulty of demonstrating success. And we've heard this morning about the different neurologic scales, et cetera, and the difficulties there.
So I think you have to have a larger study. You have to look at composite endpoints. You have to have a good validated means of assessing benefit, and probably you have to look at patients who are more likely to do better than to take all comers, especially with a small sample. But I think if this was a 1,500-patient study, or a 500-patient study, and it was appropriately designed, some of the data supports that cooling might be a benefit. But I don't think that paper has been published or that presentation's been made.
ACTING CHAIRPERSON MAISEL: Joe.
DR. ORNATO: You know, John, you raise a great point, and I can't disagree with you. You're absolutely right from a scientific standpoint. It'd be great if we could have a 500, 400, 1,000 patient study. And it would certainly be great if we had more than just VF out-of-hospital patients in the larger of the two. But I think the challenge that we all have as we've I think shared with one another this morning is how difficult an area it is for clinical research to be done in this area. And frankly, from I guess my own perspective, when I saw both of these studies come out I was almost as amazed that they were able to get the number of cases that they were able to get. Because remember, you're only talking about people who survive that initial resuscitation event. And so we've already filtered out a very large percent. And that's why, Dr. Brockman, you're absolutely right. We're winding up with a very small percent of the total number of patients. But that's almost in part inescapable because of the poor initial ROSC. And you add on top of that the fact that in both of these studies I think to some degree they made an attempt to not take all comers.
The challenge for most of us trying to apply this information clinically is not whether to do it but to whom. The problem is even more complex in that someone, I think several of you pointed out that one of the open questions is is it now the standard of care. If you look at it from the perspective of the standard of care being defined in the normal legal sense, it's not, because very clearly most communities are not broadly applying this. So in a court of law, the usual legal definition would not apply. In National Registry of CPR, of the 400-plus hospitals, as of not this last quarter but the quarter before, which are the last data that we've looked at, we have only 10 or 11 hospitals who have admitted to doing one or more hypothermia cases post-resuscitation. Now, that sounds dreadful, and it arguably might be if you take the perspective that we should be applying the ILCOR guidelines. But I want to caution everyone that these are in-hospital cardiac arrest patients in NRCPR. It excludes those patients who have arrested out of hospital. And of course, the two studies are primarily pre-hospital cardiac arrest. So in essence, it's an extrapolation of the existing data for hospitals to be reporting in NRCPR that they're applying it on in-hospital arrest patients. So that's the reality of what little data we have. But I think the bottom line is it's a very small number of hospitals, and a very small number of patients who yet are getting this form of therapy, rightly or wrongly.
Which now brings the flip side of it, which is that from a medical-legal, and a medical and perhaps an ethical standpoint it pushes us into a very interesting discussion of not the legal definition of the standard of care but to some degree what obligation we clinicians have to apply evidence-based expert recommendations that are made in this case not just on a national or an international basis. We are all well aware of the fact that translation of evidence-based guidelines into clinical practice has been a problem across the board with the 200 or 300 major papers that have been written on that topic in the last 10 years. I just recently reviewed the literature. It's a very large body of information, and with very little exception there's been a huge translational problem. Everything from as simple as giving people with ST elevation MIs aspirin to initially use of therapy and beta blockers and cholesterol-lowering agents and the like.
Take the flip side of the issue of what is the standard to which we are being expected to comply even in the legal sense. Lots of folks get into trouble when there's been a bad outcome in courts of law when there are national consensus guidelines that urge a certain therapy when they're evidence-based and meet the kinds of criteria that ILCOR's document now provides. So I know I'm going around in a circle here, but what I'm trying to do is to as best I can provide somewhat of a balanced perspective that I think there are two different ways of looking at this. They both have merit. I think, John, you've articulated properly one of the sides, and I think there is another side. I'm a bit hard-pressed to know, because there are only two trials, because they have some problems, you know, where the FDA really ultimately ought to come down at this moment where we know we'd love to have more data. I can tell you that many of us clinicians and researchers who work in this area feel compelled to apply hypothermia, at least as narrowly defined as it is in the two studies that were the basis of ILCOR. And I think the reason we're somewhat persuaded, even though we'd love to see more evidence, is that it does have reasonable science behind it from animal models. It wasn't a great surprise that this outcome occurred. Peter Safar, the late Dr. Safar, I think led us in this direction for a couple of decades now. And so it's somewhat consistent, I think, with what the animal models suggest. I'll stop there.
ACTING CHAIRPERSON MAISEL: Dr. Normand.
DR. NORMAND: I have two questions which may be very naïve. So the first question is if you have an out-of-hospital arrest, and this procedure is applied, and then presumably you're brought to a hospital, is that part -- no. Lots of people are shaking their head no.
UNIDENTIFIED PARTICIPANT: This is mostly applied after you got to the hospital.
DR. NORMAND: After, that was my question. So in theory, that's in hospital billing data. I'm just trying to think of another data source is what I'm going at. And in terms of --
UNIDENTIFIED PARTICIPANT: No, because they can't bill it.
DR. NORMAND: They can't bill it. So that's the question. I don't know if that's ethical or unethical to ask, but I was wondering whether or not if you're billing for the data, the hospital's going to be reimbursed for it. In any event, my question was is that going to -- is there another data source that may be potentially available to look at either via hospital discharge claims or via Medicare or something like that. But people are shaking their head no.
DR. ORNATO: Right now there is no billing code for it.
DR. NORMAND: There's no code.
DR. ORNATO: We've actually reviewed that. There is no billing code for induced hypothermia.
DR. NORMAND: So it's free.
DR. ORNATO: It's free care.
DR. SOMBERG: The ice is free. Not all hotels, but some.
DR. KATO: Well, I don't think you bill for the hypothermia per se, but you bill for the cooling blanket. If that's what you're going to do. And that would definitely have a code within the hospital. If you use the blanket, right.
ACTING CHAIRPERSON MAISEL: Dr. Brott.
DR. BROTT: I think these two studies are kind of an example of studies that were very simple, and could be done very easily, which many people probably in the audience and on the panel would love to be able to do. But here we are, because the results were not a slam dunk we're looking for other clues, and I think that for instance from the neurologic side if we had some other neurologic endpoints, they weren't primary, but they were looked at, and they were all consistent with these primary endpoints, I think we'd be feeling much more comfortable. So these two trials may be examples where simplicity actually has ended up to be a disadvantage.
ACTING CHAIRPERSON MAISEL: Dr. Becker.
DR. BECKER: I think it's important to sort of keep in mind if you will the relative strength of this compared to many other things that we have accepted in practice. And so while it may seem perhaps odd to the panel that based on these two studies there's an international recommendation, I think it's important to keep in mind that if one went through the ACLS algorithm and looked at drugs like epinephrine, you couldn't find two randomized trials for epinephrine right now. You could not find two randomized trials for lidocaine right now. And those are absolutely, absolutely accepted types of intervention so that when the international groups that have looked at this have really done what I think is a very thorough and admirable evidence-based evaluation that would involve an evaluation of all the animal literature that seemed relevant, the human literature that seemed relevant, the associated types of studies. And when they then lay that out on a grid which is actually done and is available to anyone who's interested in it, I think you do begin to see a picture that is compatible with the kind of conclusions that they've drawn.
And I just do want to highlight, because maybe it wasn't clear from Dr. Swain's comments, that the international recommendation was really for a very limited group of patients. It was sort of represented as a blanket statement. That's not really true. The indication was for comatose survivors of out-of-hospital ventricular fibrillation, witnesses ventricular fibrillation. I mean, it's a very narrow indication. And so I think as a clinician, my take on it is to have two randomized clinical trials in the New England Journal of Medicine is way better than just about any other aspect of the ACLS algorithm that I'm familiar with. So, I think the thoughtful approach is that whether, you know, they're definitely not perfect. They're not perfect studies. But they certainly in terms of guidance for clinicians, they have to be very powerful kinds of evidence that we take into account in trying to both come up with international guidelines, and in terms of guiding our own therapies. And I just think it's important to sort of keep that in mind as we judge these studies.
ACTING CHAIRPERSON MAISEL: Dr. Halperin.
DR. HALPERIN: It's been mentioned that hypothermia is clearly not the standard of care today because it's not practiced in many hospitals. And that kind of was put in the context of ILCOR guidelines being published. But I think that most U.S. facilities actually use the American Heart Association's guidelines to at least guide them, if you will, as to what advanced life support therapies will be used. And in fact, there is no AHA guideline on hypothermia that's been published to date. It's being considered, because the hypothermia studies were actually done after the year 2000 when the last guidelines were published. And the next guidelines are not going to be published until probably late 2005 or 2006. They're actually being worked on now. And certainly the ACLS subcommittee is pondering that with ILCOR, and there will be some guidelines that will come out. And I suspect that given what Dr. Becker said about the two randomized clinical trials, plus all the other data, and the fact that the hypothermia data far exceeds the quality and the quantity of the data that exists for most ACLS recommendations, that in fact it will get a fairly strong recommendation from the American Heart Association, which will be consistent with the ILCOR guidelines. So then that'll be published I guess in 2006.
So then the issue is, you know, then will it be adopted and become a standard of care at that time. I don't know, but I suspect it'll be used a lot more than it is now. So it'll be an interesting situation where in fact the clinical guidelines recommend that hypothermia be used in that situation, and then it'll be up to the regulatory boards to decide what is appropriate to do for the devices that would actually allow that to occur.
ACTING CHAIRPERSON MAISEL: Dr. Ornato.
DR. ORNATO: Henry, that's a really good point. AEDs came on the market in the late `70s, I guess, early `80s, if I recall correctly. And the number of sales as somewhat of an index of clinical use was, I don't know what the numbers are because I'm not obviously in that industry, but as an end user, it was pretty small. And this was called to our attention back in '87 or '88 as I recall when I was I believe on the AHA ACLS subcommittee. And we issued an -- after reviewing the data that then existed, we issued an interim ACLS guideline on the use of AEDs, and we actually added a module to the ACLS text. And if you look at the proliferation of the use of AEDs, it suddenly shot up after that. So Henry, I think your point is exactly right, that an AHA guideline, although it doesn't legally meet the definition in a courtroom of the standard of care, it certainly drives this whole topic.
The other point is a practical one. The current inexpensive ice, you know, just things that are readily available that aren't billable in a hospital environment, techniques for lowering core body temperature, seem like they're pretty easy to do and straightforward, but at least at our own institution and at the other like institutions that we communicate with on a regular basis on post-resuscitation care, we've all pretty much seen the same thing which is that it's very hard to get our physicians and nurses to really embrace this form of therapy. Not as much from the philosophic standpoint, although that's a vital part of our discussion right now, but from the standpoint of actually doing it. It's not particularly easy to use the more crude methods. The patient is wet, they're sedated, paralyzed, on a ventilator for a period of time. You have no way of tracking during that period of time what's happening to them neurologically. It makes for a great deal of discomfort in terms of the clinicians. There are questions that we get all the time about can I do this, can I do that, can I use it post-fibroembolytic therapy, can I use it with heparinization, and so on and so forth. There's concern about giving medication, and what it does to prolong the half-lives of medication. It's a very, very challenging series of clinical questions. And I think as we've been asking our colleagues around the country what's going on at their institution with this specific therapy, what we're finding is it isn't as simple as do people buy into embracing either the ILCOR guidelines or the two studies, and the animal studies that are behind them. But there's a translational, operational, piece of this that is not trivial. I think it's a huge chunk of this. And that may or may not be helped by devices, if they are to become more readily available in this area. I think a lot of the nurses would love to have a prettier, easier method to use, but that's perhaps not, obviously, the gist of what our scientific focus ought to be.
ACTING CHAIRPERSON MAISEL: Clyde.
DR. YANCY: Bill, to get back to the question before us. I respect the opinions from Drs. Becker and Ornato, and I definitely respect the process that the ILCOR effort represents because I know those processes pretty well. So I think that we would have to acknowledge that there are sufficient data, albeit a little bit soft, that we can in fact say yes, that the existing data adequate support the safety and limited effectiveness of surface cooling for achieving mild hypothermia in such patients as have been described.
I guess the real dilemma we have here has to do with taking the next step. That is, what's on the board, in which patient, and under what circumstances. And all we can do is steal a page from the typical cardiovascular trial and say it has to be in those patients that were studied that meet the exact same profile, and the therapy has to be given the same way. And then that becomes the push point. Because the question is do we take data that are already different, although I admit they've been vetted through our highest tier review, and say we can extrapolate the paradigm to an approach other than surface cooling, that is endovascular cooling, or do we require the endovascular cooling to go through the same sort of process. That is, I think, where the rest of our conversation needs to reside, because I have to accept that surface cooling has some benefit. But where do we go from here?
ACTING CHAIRPERSON MAISEL: Before we get to the endovascular cooling, I think something that I'd be interested in further discussing is obviously not all cooling is the same, so if someone brings in a product to the FDA, whether it's an ice pack or a cooling blanket, what standard does it need to meet to get approved, to show safety and effectiveness. Is it enough to cool as fast and as long as one of these other trials, or is there some other standard?
DR. YANCY: Yes, I couldn't agree with you more. Remember that the ILCOR also puts a proviso on its own statement that as soon as it says it's safe and effective immediately, more research is needed to address these very issues. So I think we have to be very careful how this is positioned.
ACTING CHAIRPERSON MAISEL: Jeff.
DR. BRINKER: Just from a regulatory point of view, it's not clear to me that these devices that already exist for cooling need to be labeled for cardiac resuscitation in any way. What is the agency's feeling about that?
DR. WITTEN: I'll answer that, and then I will really be happy if someone will answer Dr. Maisel's question that he asked. And the answer is that if a sponsor came in with an application for a specific clinical indication, even if the device was already on the market for the general indication, we would need to evaluate it and make a decision.
DR. BRINKER: Well, that's the point. I think that it's -- with a guideline that suggests cooling is potentially good and probably should be used, perhaps in the absence of any other reasonable alternative. There's a strong likelihood that without any regulatory prodding, that there would be no further controlled studies thought to be even ethically justified. And it seems to me that there needs to be some thought about this, especially in view of the fact that there are alternative cooling methods, and the data upon which this rests for out-of-hospital cardiac arrest, no matter how significant and discrete the patient populations were, the implication is that it's going to be used for all cardiac arrests. And I don't know whether the data's there to really support that.
So do we really want to assume that it's unethical to do any more studies comparing non-cooling to cooling?
ACTING CHAIRPERSON MAISEL: John.
DR. SOMBERG: I'm going to be the bete noire of the group here, because I looked over these studies before I came, I've looked over them again after listening, and I think on the basis of less than 200 patients with one study negative on survival, another study that meets the endpoint of survival but has we're talking about a 10-patient, 12-patient difference, they're very small numbers. I thought if -- knowing that I have sat on this committee before, if someone came with a device that you hooked up to a patient and you got these results. Forget about they're being in the New England Journal. There's all sorts of politics here. But if you had these results, I'd be very much surprised if that device had passed. So I think to set a standard of care based on this -- that's what you're talking about, a standard of care -- based on this data, and therefore you could just say you have an equivalent device to cooling packs. Maybe it does it fast, and maybe it does it with no one gets wet and electrocuted or what have you, and therefore we should get approved based on this, and we can't study it any further, is just going to stifle science and maybe, just maybe, base everything on a pack of poorly balanced cards. I just do not believe these two studies meet any FDA advisory panel standard of approval.
I can't speak for the groups that went around making -- and I'll just say this. When people make standards based on practice, it's what best out there now. You feel pressured to do that. That's different than evaluating a particular device that's before you. So I think we should not say because it's been one body, and another body may do this, so that it's become the standard, therefore we should say that's appropriate for the approval of a device or a drug. And drugs in this area, you say epinephrine, it's sort of a grandfathered agent. But amniodarone would have never been supplanted, and lidocaine, if the results were based on this amount of information.
DR. BECKER: Amniodarone was worse.
DR. SOMBERG: No, I disagree with you on that. I think the data from the arrest trial are more substantial.
ACTING CHAIRPERSON MAISEL: Dr. Hallstrom. Let's talk about hypothermia devices. Dr. Hallstrom.
DR. HALLSTROM: My concern with accepting these two studies as defining the standard of care is the difficulty of doing the next study. Because if I were on an IRB and this was the standard of care, and now you bring another cooling device which is going to cool a little faster or some such thing like that, what does my sample size have to be to do a comparison? I'm all of a sudden into the 1,000-patient realm instead of the 100-patient realm. And I think -- I just have enough questions about this data that I would like to see a few more 100-patient trials. And indeed, if the effect is as great as these two papers suggest, that is all you need is another 100- or 200-patient trial.
ACTING CHAIRPERSON MAISEL: Norm.
DR. KATO: I think I have to agree with John. I'm beginning to feel now that there is I guess some key words in this question. Because they talk about various methods of surface cooling, and we're talking about a labeling indication for a device which we don't see in front of us. The way this panel works is that a sponsor comes to us with a device and with their data in hand, presents it, and then we make a decision based on the data at hand. I would feel very uncomfortable trying to prospectively grant anybody who comes along with a device in the future who can cool the body using various methods of surface cooling, sorry for the pun, but a blanket okay to achieve a labeling indication as such. Cooling devices are on the market already. They can be used as off-label devices. They currently have two articles which they can use to support the use as an off-label device. But I feel very uncomfortable trying to make a future statement about what device could be accepted now, given the fact that there is no device ahead of us.
DR. WITTEN: Thank you for that comment. And may I ask a follow-on question, which is what type of information would you want to see. If someone came in with a surface cooling device application, what type of information would you want that application to contain in support of that indication for the device?
DR. KATO: I assume that's for the panel.
DR. WITTEN: Generally for the panel.
DR. KATO: Because that's really Question (c), right?
DR. WITTEN: That is -- yes. Question (c) isn't quite phrased like that, but yes, that is Question (c). So we can wait till we get there.
ACTING CHAIRPERSON MAISEL: Why don't you read Question (c) now and we can move on to that.
MR. OGDEN: I'd be glad to, thank you. If you do not believe the literature supports an indication in the labeling for surface cooling for achieving mild hypothermia, please discuss an adequate study design to demonstrate that these are safe and effective for achieving mild hypothermia in patients with cardiac arrest. Please discuss possible control groups, endpoints, and time of measurement of endpoints, keeping in mind as you refer to your own experiences whether induction of mild hypothermia is a commonly accepted practice procedure in your geographic region such that it would be unethical to study surface hypothermia versus a control of standard of care that does not include cooling.
DR. WITTEN: And before you start, may I just provide some clarification of this question, which is this specifically relates to surface cooling. And I know in the last part of the discussion there was some discussion of endovascular and of surface cooling. But we have another question that's about -- or our next two questions after this have to do with other methods of hypothermia. So this specifically relates to surface cooling devices.
ACTING CHAIRPERSON MAISEL: Henry.
DR. HALPERIN: Yes, I just wanted to make a comment about the issue of standard of care, and stifling further research and what have you. Because I think we're a little bit off the subject on that. Because although I've stated already that I think the two studies on hypothermia are important studies, they're hardly definitive for the role of hypothermia in cardiac arrest, because they apply to actually only at most 10 percent of the cardiac arrest population. I mean, these are comatose survivors of out-of-hospital witnessed ventricular fibrillation arrest. I mean, there's a lot of qualifiers to it. So all of the other cardiac arrest population are not covered at all by these studies. And that research is certainly wide open. So even if one accepted this as the standard of care, it would be a tiny patient population that it would apply to. And the study of hypothermia for cardiac arrest in general is still wide open.
ACTING CHAIRPERSON MAISEL: I'm not sure we've reached any consensus on this point. So maybe I can ask if we have any consensus regarding whether we feel comfortable with a blanket support of safety and effectiveness of surface cooling in general. There were a few people who seemed potentially in support of that concept. If so, can you speak up and maybe clarify your position? Are there people who feel that a blanket support of surface cooling is appropriate?
DR. YANCY: I think it has to be in the context of the clinical circumstances. The sentiment that I've perceived, at least, is that for patients that reflect the ones in which there are signals of benefit, it would be hard to say the answer is no, even if they aren't definitive trials. But to say that it is a global approach equivalent to restoring circulation and creating an airway, I don't think any of us can say that.
ACTING CHAIRPERSON MAISEL: What if we were more specific and said support -- are there people who feel comfortable supporting the safety and effectiveness of surface cooling for achieving mild hypothermia in the unconscious patient with spontaneous circulation after ventricular fibrillation. Are there people who would feel comfortable with that? I see a few. Are there people who are uncomfortable with that?
DR. NORMAND: Hello.
ACTING CHAIRPERSON MAISEL: Dr. Normand.
DR. NORMAND: No, I'm sorry to do this, but I think you're talking about the data based on the two trials, right?
ACTING CHAIRPERSON MAISEL: Correct.
DR. NORMAND: And I'm uncomfortable with the information in the two trials. Obviously I'm not a clinician, but based on the data that are indicated in those two trials, with trending and with the way the trials were designed, I think the data would not make me feel comfortable with agreeing with that.
ACTING CHAIRPERSON MAISEL: Okay. Well I think we have a good sense of what the panel feels. Dr. Weisfeldt?
DR. WEISFELDT: I was going to try to go back to the question we were just asked about the trial design, what trial would you design. To be honest, I think I'd design the two studies that we just heard reviewed. Because you're talking about a metabolic intervention that is only going to improve the outcome of an organ that is recoverable from the point of view of the ischemic insult that occurred during the cardiac arrest. That's the intent. If you were to apply hypothermia to a broader population, then you're looking at mechanisms of failure to survival that have little or nothing probably to do, but have more to do with the intrinsic heart disease, or injury that's occurred somewhere else. So you come rather quickly, if you think about a study design where you have a half a chance of showing a favorable outcome from a metabolic intervention that's going to improve tissue survival, to a similar study, at least very similar, to this one.
I then go to the comments of several people that compared to other types of interventions and CPR, where we have no data, and I would want to correct for the record the amniodarone comment because it's not correct. There was no survival benefit to out-of-hospital cardiac arrest for two amniodarone studies. Here we have survival to discharge from the hospital in two studies, reviewed, claimed. And like so many other studies, if you dissect out details, you can find criticisms. And unlike a panel who is, let's say they were reviewing a surface device for cooling, there would be an advocate for the device that would have presented hopefully equally articulately and convincingly to the FDA representative what the deficiencies are. So I'm personally persuaded that the studies are not half bad, that there are better data, that one would certainly hope to see further data produced, but a judgment could easily be made here that in the subgroup of patients we're talking about, that this treatment is, by virtue of two randomized prospective controlled studies, safe and effective. In that population.
ACTING CHAIRPERSON MAISEL: Do we have other comments regarding the trial design, study endpoints, control groups? Dr. Brinker.
DR. BRINKER: Well, for these not half bad studies, I would at least, if they are to be labeled -- I mean, the issue is nobody's preventing them from being used. They could be used whenever they want. The issue is should that be a labeled indication based on the two studies. And I have my -- I still have my doubts that we need to label this so, because it sets up a straw man for other kinds of technology that would then warrant a less vigorous scientific study. All they would have to do is mimic their -- potentially, at least -- their ability to cool to a similar temperature, and do it without introducing any excess adverse event.
So, but if one were to say that any cooling apparatus capable of dropping the temperature to X amount, that based on these studies should be labeled as indicated for cardiac arrest, which includes presumably bags of ice water, I would suggest that all the other issues that were controlled for in the two studies, namely temperature, core temperature -- assuming they measured core temperature -- but degree of temperature reduction, be cited as a goal, and the co-administration of paralyzing and sedative drugs, which may in themselves have a beneficial effect, be included as part of the labeling indications.
ACTING CHAIRPERSON MAISEL: Dr. Brott.
DR. BROTT: In terms of (c), as best I can tell these could be viewed as coma studies. And they're coma studies of small numbers of patients. And I think that if you have coma studies of small numbers of patients, you have to make sure that your endpoints are as unbiased as possible. And in the first study, there was a tremendous opening for bias. And I can't imagine that the person doing the assessments as to whether or not they could go home or not, or to rehab, had no idea as to which treatment group they were in.
The second study, we're not really told anything about how the CPC was carried out. We don't know if there was any Barthel, or any kind of assessment. So the assessments of small numbers of patients were really not described in detail enough for us to have confidence that they were unbiased. And so with regard to Question (c), I would think that since it's a coma study, the endpoints would have to be very carefully considered, to have not only a primary outcome that had to do with neurologic outcome, but then some secondary measures, since there would be small numbers of patients, to at least provide some support, or at least consistency, with the primary endpoints. So I could not accept (a), and I think the reasons are the suggestions for (c).
ACTING CHAIRPERSON MAISEL: Any other comments on surface cooling? Why don't we take a break for 15 minutes, and come back and finish up with endovascular cooling. So let's resume at 3:30, please.
(Whereupon, the foregoing matter went off the record at 3:14 p.m. and went back on the record at 3:32 p.m.)
ACTING CHAIRPERSON MAISEL: So we'll turn our attention now to endovascular cooling devices. And why don't we read Questions 2 and 3, please.
MR. OGDEN: Thank you, Mr. Chairman. Question 2. Endovascular cooling catheters represent a new technology for achieving hypothermia. Please discuss whether or not you believe that surface-induced hypothermia is comparable to core-induced hypothermia in relation to safety and effectiveness measures. Is there literature to show that core- and surface-induced hypothermia are physiologically equivalent?
Question 3. Please discuss an appropriate study design to evaluate safety and effectiveness of endovascular cooling catheters for patients following cardiac arrest. For example, please address the appropriate control group and endpoints for this study.
ACTING CHAIRPERSON MAISEL: Discussion. Dr. Weinberger.
DR. WEINBERGER: Alright. I think that we couldn't even all agree on whether we thought surface cooling was appropriate therapy for a more general class of patients. So I think asking whether endovascular catheters might be equivalent is sort of premature.
But on a more serious note, I think I'd have to know quite a bit more about the nature of the device in order to specify what sort of controls I would want. For instance, are we talking about an endovascular device that requires a large hole in a vein or a small hole? Is the patient heparinized during this process or not? There are potentially interesting complications that one can envision from this process that are very unique to an endovascular location that wouldn't be expected to be seen in a surface cooling methodology. So without the appropriate details of the device, and some consideration of the expected complications, including local bleeding, including hypercoagulable states, because you're putting a device of unknown size into the IVC where flow is sluggish, and you know, to be cooling the blood, presumably we don't know what the surface temperature of the coil is going to be. We're presumably shooting for a core temperature of 33, but what's the temperature on the surface of the coil? Is that at 30 or below, and what is that doing to the coagulation system? Certainly in the COOL-MI trial there were some interesting events that happened in relation to people who were cooled. There were events of pulmonary edema, and pulmonary emboli. I think that it really deserves a much more careful kind of thought than -- and treating this like a convention PMA-type trial. That is, a randomized control trial which might potentially require even three arms if the device itself is problematic even without its use. So I think that the discussion is predicated on knowing a bit more about the device, and at least seeing some of the animal data that is motivating use of this particular device.
ACTING CHAIRPERSON MAISEL: Everyone agrees with Judah? Dr. Becker.
DR. BECKER: I pretty much do. I guess my major concern is really one of safety more so than the cooling aspect of it. That it would seem to me that there would be different safety issues, and that that would sensibly be the focus of a review from my standpoint; that I think they are not the same in terms of potential safety. And certainly we haven't seen any data here at least to indicate that there is safety. There may be data out there, but you know, we've not seen that today.
ACTING CHAIRPERSON MAISEL: Could you be a little more specific about what sort of safety endpoints, you know, what time you would like those safety endpoints?
DR. BECKER: Well, I agree with many of the things that were said in terms of the temperature effects, local effects, bleeding kinds of effects, you know, just trauma to the vessel, incidence of if you poke something in the femoral vein, how often do you go through the vein, end up in the retroperitoneal space. You know, there really are safety kinds of concerns when you become invasive that I think are potentially important. And in the absence of seeing that, I think that those would have to be satisfied. But I guess I'd say then beyond that, I think that if the safety concerns were satisfied that they appear to be as safe, I would tend to think that then cooling is cooling at some point; that effects on the brain in terms of neurologic long-term recovery, I have no reason to think that they would be substantially different between surface cooling and endovascular cooling beyond the safety issues.
ACTING CHAIRPERSON MAISEL: Clyde.
DR. YANCY: Well, that would be with one caveat. I mean, we do recognize the clinical syndrome of hypothermia that in some circumstances can be an important clinical situation that has to be addressed. And we don't know that there's one threshold above which everyone is safe and below which people at risk. There may be a continuum that turns on gender, age, body mass, et cetera. So understanding so little about the implications of endovascular cooling and how that affects core temperature, the rapidity to which you are cooled, all of these dynamics are, at least in my mind, so unclear that I think the overwhelming answer to the question is that we cannot believe that surface and core are identical until we see more data. And the safety issues I think have to go beyond the procedural things which admittedly are a concern, and have to deal with the very issue of generating hypothermia systemically and what's associated with that.
ACTING CHAIRPERSON MAISEL: Lance.
DR. BECKER: Just one follow-up on that, because I think Clyde raised an important point. And maybe we should've mentioned this with surface cooling, that I have some concerns over safety with surface cooling. I think that in terms of warnings and things like that on the labeling, I think there has to be very early on attention to sort of what is the monitoring so that you don't over-cool an individual. That is, too much cooling we know can be lethal. Now, that's true in many things that we do in medicine. Too much of many things can ultimately be lethal. But I think that needs to be -- I think we need, actually, a little better handle on that.
Because I can tell you now that we are doing surface cooling that issues with respect to the actual control of an individual patient are not trivial. What I mean by that is it's actually, over-shoot is very common in trying to cool an individual down. And I have heard at least at one national meeting of at least one death that's been attributed to over-shoot in the attempt to cool someone. This is with surface cooling. And so I think that the issue of monitoring of the therapy is something that probably does need to be addressed.
And it's sort of interesting to come back to how the endovascular may or may not just fit into this. It may turn out that endovascular has an advantage because it has the ability to both cool and to warm, for example. I mean, it may ultimately turn out to be a safer device. But again, I think we need to see that data in order to really make that judgment.
ACTING CHAIRPERSON MAISEL: And what would the control group be for these studies of endovascular devices?
DR. BECKER: My thought would be that -- this is really a tough one. Because what you're really asking is could you ethically have a non-cooled group, I think, or at least that's what I struggle with. And I'm not sure that I know the answer. I think that if you are using one of the groups -- you know, we have a very, very narrow group for surface cooling that we -- or at least that some of us believe that there's a clear indication for. I think that if you're outside that group, there would be no question that an appropriate control group would be a control group that was normothermic. And I don't think anyone would have too much difficulty with that. So if you were looking at asystole, or PEA, or some of the conditions where we really don't have any data, my thought would be that an appropriate group would be a normothermic group.
ACTING CHAIRPERSON MAISEL: Dr. Brinker.
DR. BRINKER: Lance, I think it would be hard to get a sponsor to take a group that -- basically they would have to establish superiority under no cooling, and the group that is arguably at higher risk of not benefiting from any type of cooling. So it'd be a high-risk study for anybody to undertake, although it could be an important one in terms of knowledge because it fills in the gap of what we don't have with the surface cooling studies. On the other hand, if you don't mandate that, that is all you have to do is to show you're as good as the surface cooling study, then you have to compare it to surface cooling, and you'd get some interesting data about the ease of achieving cooling and maintenance of that degree of hypothermia. But what you wouldn't get is any difference in, presumably at least, it's unanticipated that you would get any difference in mortality, or I should switch it around, in survival. And in fact, the survival rates since it's not quite clear what they would be -- what's the normally anticipated survival rate with cooling. It could be anywhere over the place. And it'll be hard. It'll be hard no matter what the sponsor of these devices do. But I don't think that, unless they had some out-of-U.S. experience leading them in a certain way that they could afford to do anything but compare it to surface cooling. Which wouldn't give us all the information we'd like to see from such a study.
ACTING CHAIRPERSON MAISEL: Norm.
DR. KATO: From what I've read, I still have a problem with what the definition of normothermia is. And the reason why I bring that up is because our experience in the operating room with surgical patients is that within about an hour, hour and a half, core temperature drops from normal down to 34, sometimes 33 degrees Centigrade. Given that many of our ICUs are air conditioned, and we like to work in air conditioned settings, you know, is it reasonable just to allow a patient to equilibrate to ambient temperature as a control? Although, I have to tell you, they're going to cool off anyway. Is that what normothermia should be? Because I think the converse, which is to try to maintain a body temperature of 37-38 degrees, in the cardiovascular field we've also learned that when you're warming somebody up from even 33-34 degrees, you basically have to stop at 35 because they will, you know, much like a turkey coming out of the oven, they will continue to cook until they hit 37-38 degrees. So that's the other part of the problem with that. So I echo the comments about over-shoot. But again, I'd have to look at some more data to get a comfort level with that definition of normothermia.
ACTING CHAIRPERSON MAISEL: John.
DR. SOMBERG: Well, when comparing surface to core cooling, I think it's important to how you measure the temperature of the body as well. There wasn't any material in the handout. I didn't do any research on this, but in both studies they used first tympanic membrane temperature and then they used bladder temperature. And in fact one did used a Swan, I think one Australian one used a Swan temperature as well of the heart. But there are different ways to measure temperature. And if one was comparing surface with catheter cooling, I think one would first want to see if you'd get similar readings on that. And I didn't see anything in the literature, but maybe there is literature on this.
ACTING CHAIRPERSON MAISEL: Clyde?
DR. YANCY: Yes, one practical issue is if you talk a little bit more about trial design and think this through. I'm not sure how you stand at the bedside with a patient who's comatose and present three options to a family: normothermia, surface cooling, and endovascular. And do that without some major conflict internally. You're talking about a desperate situation where you feel compelled to do as much as possible. And I think even though we can sketch out an ideal design, I think making it happen and overcoming the informed consent barrier, I think I really struggle with that.
ACTING CHAIRPERSON MAISEL: There was a comment from Dr. Becker earlier that, quote, "Cooling is cooling." And I just wanted to get a sense of whether we feel that endovascular -- recognizing the safety issues, that we feel that endovascular cooling is likely to be the same and have the same effects, or whether we think it's different and needs to be studied with separate clinical endpoints, et cetera.
DR. SOMBERG: Can anyone answer my question, or maybe the FDA who's looked at this issue more, about the measuring the cooling before we vote on is cooling the same? In other words, if you cool using those two methods, do you get similar core temperature, brain temperature, cardiac temperatures?
DR. WITTEN: Are you asking whether if you cool with surface cooling or endovascular cooling you get similar temperatures?
DR. SOMBERG: Yes.
DR. WITTEN: Well, it's hard to give a simple answer to that, because usually in the kinds of studies that we've seen, the sponsors try to achieve a certain temperature. And you could achieve it. But there are the issues that were discussed a little bit earlier about how long it takes, and whether there's a problem with over-shoot, whether there's a problem with re-warming. And so I think those are all questions. With all these technologies, it's exactly how well you're able to control what you're doing. It depends on the specific device and the specific technology, certainly, that's true.
DR. SOMBERG: But there are different ways to measure the temperature. You know, we're saying cooling is cooling, but what's the point of measurement? Where are you getting it?
DR. WITTEN: Yes, well we've had different measurements we've seen. I mean, the answer's the same in terms of what you get, but bladder temperature, tympanic membrane temperature, those would be two of the places that we've seen measurements taken during studies. Rectal temperature.
ACTING CHAIRPERSON MAISEL: So it sounds like we're saying endovascular cooling techniques are different. They certainly raise their own safety issues. I don't think it's clear that brain cooling -- I don't know that we know whether brain cooling is the same based on a temperature in IVC versus surface versus bladder, what have you. I think we'd ideally like to see randomized trials. I don't know how realistic that is. The comparison group, certainly in groups that have already been studies it sounds like should be surface cooling. And for groups that have not been studied it could probably be either control groups or surface cooling, perhaps. Anyone have any additional comments on endovascular cooling?
DR. SOMBERG: I just want to stand corrected. He just gave this to me, that it has been looked at, and there doesn't seem to be variance for the two methods in terms of the core temperature one reaches. That's in that summary article.
DR. NORMAND: It was summarized in the article. I just couldn't interpret it.
DR. WITTEN: May I ask one follow-up question, which is I think there was a nice discussion on both local and systemic adverse events that one might conceivably be concerned about. And I'd like to know if there are any specific adverse events in this particular population that you'd want to pay attention to if you were doing a study of endovascular cooling. In other words, we have the general systemic and local safety concerns, but in this particular population, is there a specific category or type of adverse events that we particularly would want to note in a study?
ACTING CHAIRPERSON MAISEL: Judah?
DR. WEINBERGER: I was having a discussion with Dr. Weisfeldt during the break. I think patients who have low flow states, particularly in the IVC, people who survive cardiac arrest, even if they have the circulation, some of them have poor pump function and low cardiac outputs. Those are people if you put a device into the IVC, and you have slow flow, are more likely to thrombose. And thrombosis might appear as an IVC thrombosis, it might appear as pulmonary embolus. And I think that class of complications is one you're probably want to be sensitized to.
DR. WITTEN: Thank you.
DR. KATO: The other general complications from cooling can be ventricular fibrillation, DIC, bleeding problems. Basically just talk to any cardiovascular surgeon who performs deep hypothermic circulatory arrest and they'll tell you it can be virtually anything. And the re-warming process can be a disaster too. You know, gas can come out of liquid. You can get air emboli. A whole host of metabolic derangements, not to mention differences in splenic blood flow causing sudden acidosis as circulations are restored in terms of profusion. So there can be a whole host of problems with endovascular cooling.
ACTING CHAIRPERSON MAISEL: Any additional panel comments? Dr. Witten, any other comments or questions?
DR. WITTEN: No. I'd like to thank the panel for this really helpful discussion this afternoon. Thank you.
ACTING CHAIRPERSON MAISEL: Dr. Zuckerman, any comments or questions for the panel?
DR. ZUCKERMAN: Again, on behalf of the agency we found this whole discussion to be extremely productive and thank the panel members.
ACTING CHAIRPERSON MAISEL: I'd like to invite our industry representative Michael Morton to make any comments.
MR. MORTON: Thanks, just a couple of quick comments here. I appreciate many of the comments that the panel has made today regarding the size and design of studies, acknowledging the challenges of this patient population, and informed consent, realizing that if the expectations become too high for these studies the cost of the studies could preclude a sponsor from bringing them to market. So thank you.
ACTING CHAIRPERSON MAISEL: Thank you. And I'd like to invite our consume representative Linda Mottle to make any comments.
DR. MOTTLE: Thank you, Dr. Maisel. I'd like to echo some of that same thought. The deliberations have been very forthcoming. Some of the things that I still am concerned about are some of the ethics dealing with our implementation of new technologies into our public health system, and that we do not stifle those developments. We've heard many comments that many of our ACLS algorithms now don't have a lot of wonderful clinical studies behind them, and yet they are standards of care. We have new technologies emerging with some quasi-studies to back them up, and yet we have hundreds of thousands of deaths. We also have other precedents in the clinical trial arenas, such as with cancer, AIDS, where not the tightest of scientific study results are often used, and yet the progress continues to develop and implement new treatment modalities. And I'd like those to be considered also in this arena.
ACTING CHAIRPERSON MAISEL: Thank you. Any other final comments or recommendations from the panel? Dr. Weisfeldt.
DR. WEISFELDT: The last comment, Dr. Mottle, and the letter from Dr. Schmidt point to the issue that I think everybody is aware of, that the FDA-initiated regulations on the waiver of informed consent does create significant impediment to doing studies. We've seen several publications that have documented the decline in resuscitation research in this country, and there are European issues that are coming to the fore that are parallel.
But there's one comment that at least to me has come up a couple of times when the waiver has been discussed, and that is whether -- you cannot have a national IRB because IRBs are regulated in local fashion. But the agency, the FDA, could decide to create a national advisory board on resuscitation research that would constitute, if you will, advisors to the agency, much like if you will the gene therapy advisory board creates advice to the NIH and the NIH director about clinical trials. If there were such a body that advised the FDA on the appropriateness and ethics of resuscitation research, testing of specific devices, my sense is that local IRBs would find some cover, some support for making difficult decisions that they now have to make essentially individually based upon whatever expertise they may have, which is oftentimes not very much in the area that we're talking about.
The IDE, we understand, is the agency's major way with devices for giving approval for research. But in truth, the IDE is a technical issue, and it doesn't come with a lot of deliberation about the appropriateness of the research. So, I mean some of the problems we have here in the area we're talking about clearly deal with the sample size, ability to do research in this arena. And one of the major perceived impediments to more and better research, I believe, is the waiver. And at least that suggestion might be something the agency could consider, and in its wisdom see whether they believe that this might help. And obviously, any panel member, I would love to hear comments.
ACTING CHAIRPERSON MAISEL: Any other additional comments from the panel? Dr. Becker.
DR. BECKER: Yes, I'd like to just sort of thank the agency for holding this panel. I've had sort of the opportunity to be involved in this research for a long time, and this is one of the first times that I've been aware of a panel with no sponsor where sort of a real airing of a number of difficult issues could take place in this kind of a setting. And I'd like to compliment the directors and the individuals who have put this together and pulled all the people together.
And I guess my one recommendation would be that, you know, a venue like this would be considered in the future. Because I think this is a very different dynamic area of science. I think that what we have today, I hope, will not be what we have in five years, and not what we have in five years after that. And so I think the notion of sort of revisiting this very dynamic issues around the appropriate controls and consents and devices is really one that warrants sort of this kind of attention. And I would like to just personally thank the agency for their efforts to have this be an open and very welcoming sort of venue.
ACTING CHAIRPERSON MAISEL: Seeing no additional comments, this concludes the recommendations of the panel regarding the type of data required to effectively evaluate the performance of CPR in hypothermia devices, and I'd like to thank the panel members for attending.
(Whereupon, the foregoing matter went off the record at 4:01 p.m.)