UNITED STATES DEPARTMENT OF
HEALTH AND HUMAN SERVICES
+ + +
FOOD AND DRUG ADMINISTRATION
+ + +
MEDICAL DEVICES DISPUTE RESOLUTION PANEL
+ + +
SEPTEMBER 6, 2001
+ + +
The panel met in Salon E at 9751 Washingtonian Boulevard, Gaithersburg, Maryland, at 8:00 a.m., Scott Ramsey, M.D., Ph.D., Acting Panel Chairman, presiding.
SCOTT D. RAMSEY, M.D., Ph.D., Acting Chairman
MARK D. CARLSON, M.D., M.A., Standing Voting Member
RALPH B. D'AGOSTINO, Ph.D., Temporary Voting Member
GERALD J. SHIRK, M.D., Temporary Voting Member
KIM L. THORNTON, M.D., Temporary Voting Member
HECTOR HUGO GONZALEZ, R.N., Ph.D., Nonvoting Member
JUDY GORDON, D.V.M., Nonvoting Member
ROXOLANA HORBOWYJ, M.D., FDA
RICHARD KOTZ, FDA
DAVID KRAUS, Ph.D., FDA
CELIA WITTEN, M.D., FDA
KAREN M. BECKER, Ph.D., LifeCore
JAMES W. BRACKE, Ph.D., LifeCore
THEODORE COLTON, Sc.D., LifeCore
ALAN H. DeCHERNEY, M.D., LifeCore
SEBASTIAN FARO, M.D., Ph.D., LifeCore
DOUGLAS B. JOHNS, Ph.D., LifeCore
LUIGI MASTROIANNI, JR., M.D., LifeCore
STEVEN PIANTADOSI, M.D., Ph.D., LifeCore
DONALD B. RUBIN, Ph.D., LifeCore
RICHARD P. CHIACCHIERINI, Ph.D.
GERE diZEREGA, M.D.
LENA HOLMDAHL, M.D., Ph.D.
L. MICHAEL KETTLE, M.D.
L. RUSSELL MALINAK, M.D.
MARK G. MARTENS, M.D.
JOHN SEVER, M.D.
Introductory Remarks, Les Weinstein 11
Dr. Michael Kettel 16
Dr. Lena Holmdahl 19
Dr. Russell Malinak 24
Dr. Melvin Thornton 28, 262
James W. Bracke, Ph.D. 35
Karen M .Becker, Ph.D. 37
Douglas B. Johns, Ph.D. 59
Luigi Mastroianni, Jr., M.D. 79
Alan H. DeCherney, M.D. 85
Sebastian Faro, M.D., Ph.D. 89
Theodore Colton, Sc.D. 94, 107, 115
Steven Piantadosi, M.D., Ph.D. 98
Donald B. Rubin, Ph.D. 109
David Kraus, Ph.D. 124
Roxolana Horbowyj, M.D. 128
Richard Kotz 152
Response by Sponsor:
Karen M. Becker, Ph.D. 174
Steven Piantadosi, M.D., Ph.D. 178
Donald B. Rubin, Ph.D. 183
Theodore Colton, Sc.D. 187
Sebastian Faro, M.D. Ph.D. 190
Panel Discussion of the Question 194
Bess Weatherman 251
Mark G. Martens, M.D. 256
Augusta Sisler 265
Panel Deliberations and Vote 268
CHAIRMAN RAMSEY: I'd like to call to order the Medical Devices Dispute Resolution Panel.
I'm Scott Ramsey. I'm the Acting Chair of the panel.
This meeting is being held at the request of LifeCore Biomedical to resolve a scientific dispute between LifeCore, the sponsor of premarket approval application PMA 990015, as amended, for INTERGEL adhesion prevention solution, and the Office of Device Evaluation in FDA Center for Devices and Radiologic Health.
On November 15th, 2000, ODE sent LifeCore Biomedical a not approvable letter regarding its PMA, as amended, for INTERGEL adhesion prevention solution. The letter states that there is not sufficient information directly relating to the performance of this device to its indication for use to demonstrate reasonable assurance of safety and effectiveness.
The indication for use as described in LifeCore's amendment to the PMA, which is Amendment 11, dated June 2nd, 2000, is for use as an intraperitoneal instillate for reduction of adhesion formation following gynecologic pelvic surgery.
This amendment modified the indication for use that was proposed in the original PMA, which was for a general surgery indication.
LifeCore disagrees with ODE's decision to issue the not approvable letter and the reasons for issuing it as enumerated in their letter. It's LifeCore's opinion that the existing scientific data provides reasonable assurance of safety and effectiveness.
Specifically, LifeCore believes the PMA as amended should be approved because the available data shows that, first, there exists a statistically and clinically significant benefit in favor of INTERGEL solution as compared to control, lactated Ringer solution, in reducing adhesion formation following gynecologic pelvic surgery.
And, second, that the benefit is achieved without exposing the patient to any unacceptable risk, including infection.
Thus, this panel to whom LifeCore has appealed the not approval letter is charged to answering the following question and to make recommendations to the Director of the Center for Devices and Radiologic Health as to how this scientific dispute should be resolved.
In particular, the question is whether the PMA as amended provides reasonable assurance of the safety and effectiveness of INTERGEL for its intended use as an intraperitoneal instillate for reduction of adhesion formation following gynecology pelvic surgery.
In answering this question, the panel should determine whether there are statistically significant differences between INTERGEL solution and control, whether those differences can be considered clinically significant, and, second, whether the benefits of the product outweigh the potential risks, including potential risks of infection.
Our panel will discuss this question, and then when the panel votes on a recommendation to the Center Director on approvability of this PMA, as amended, that vote will constitute our answer to the question.
I'd like to note for the record that the voting members present constitute a quorum as required by 21 CFR, Part 14, and at this point I'd like each panel member to introduce him or herself, also designating their specialty, position title, and status on the panel.
And I'll start with Dr. Carlson.
DR. CARLSON: My name is Mark Carlson. I'm a cardiac electrophysiologist by training, and by trade I'm Professor of Medicine at Case Western Reserve University Medical School in Cleveland.
CHAIRMAN RAMSEY: And please also state whether you're a voting, a nonvoting, standing or temporary member.
DR. CARLSON: I am a voting standing member.
DR. D'AGOSTINO: Ralph D'Agostino, statistician from Boston University. I'm a temporary voting member.
DR. GORDON: Judy Gordon. I'm the industry representative to this panel. I'm a regulatory consultant. I've spent the last 20 years directing clinical trials of medical devices and pharmaceuticals and representing companies, more recently largely small device companies to the FDA.
DR. KIM THORNTON: Kim Thornton. I'm an Assistant Professor in the Division of Obstetrics, Gynecology and Reproductive Biology at Harvard Medical School and also reproductive endocrinologist at Boston IVA. I'm a voting non-temporary or temporary member.
DR. GONZALEZ: I'm Hector Gonzalez. I'm a registered nurse. I'm the consumer representative on the panel, nonvoting, and I'm the CEO for the San Antonio chapter of the Hispanic Nurse Association and Chairman and Professor Emeritus of Nursing at San Antonio College.
DR. SHIRK: I'm Gerry Shirk. I'm a private gynecologist in practice in Cedar Rapids, Iowa, and a clinical Associate Professor at the University of Iowa, and I am a temporary voting member.
CHAIRMAN RAMSEY: And I'm Scott Ramsey. I'm an Associate Professor at the University of Washington and the Fred Hutchinson Cancer Research Center. I'm a standing voting member, but I will be serving as the Acting Chair in this case voting only in case of a tie.
I'll now ask Les Weinstein, the CDRH ombudsman and Executive Secretary of the panel, to make a few comments.
MR. WEINSTEIN: Thank you, Dr. Ramsey.
The next item of business are statements that I will read into the record. The first is an appointment to temporary voting status.
Pursuant to the authority granted under the Medical Devices Advisory Committee charter, dated October 27th, 1990, as amended on August 18th, 1999 and November 16th, 1999, I appoint the following people as voting members of the Medical Device Dispute Resolution Panel for this meeting on September 6th, 2001:
Gerald J. Shirk, M.D.;
Kim L. Thornton, M.D.;
Ralph B. D'Agostino, Ph.D.
For the record, these people are special government employees and are consultants to other panels under the Medical Devices Advisory Committee. They have undergone the customary conflict of interest review and have reviewed the material to be considered at this meeting.
In addition, I appoint Scott D. Ramsey, M.D., Ph.D., to act as Temporary Chair for the duration of this meeting.
Signed David Feigal, M.D., M.P.H., Director, the Center for Devices and Radiological Health.
The second statement is a conflict of interest statement.
The following announcement addresses conflict of interest issues associated with this meeting and is made part of the record to preclude even the appearance of impropriety. To determine if any conflict exists, the agency reviewed the submitted agenda for this meeting 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 employer's financial interests. However, the agency has determined that the participation of certain members and consultants, the need for whose services outweighs the potential conflict of interest involved, is in the best interest of the government. Therefore, a waiver has been granted for Dr. Ralph D'Agostino for his interest in a firm that could potentially be affected by the panel's recommendations.
Copies of this waiver may be obtained from the agency's Freedom of Information Office in Room 12A15 of the Parklawn Building in Rockville.
We would like to note for the record that the agency took into consideration other matters regarding Drs. Mark Carlson and Ralph D'Agostino. These panelists reported interests in firms at issue, but in matters that are now concluded, unrelated to today's agenda or imputed from an employing institution.
The agency has determined, therefore, that they may participate fully in all discussions.
In the event that the discussions involve any other products or firms not already on the agenda for which an FDA participant has a financial interest, the participant should excuse him or herself from such involvement, and the exclusion will be noted for the record.
With respect to all other participants, we ask in the interest of fairness that all persons making statements or presentations disclose any current or previous financial involvement with any firm whose products they may wish to comment on.
Also, anyone who will be making a presentation to the panel today should provide copies of your remarks, including overheads unless you have already done so. They will be collected from you when you go up to the podium.
In addition, after the meeting, Dr. Ramsey and I will be available to respond to questions from the press.
CHAIRMAN RAMSEY: Before we have presentations from the sponsor and from the FDA, we're going to start with an open public hearing. So at this time we're going to open the floor to anyone from the public who would like to address the panel and present data, information, or views on the issues pending before our committee.
I'd ask that all of the persons who do address the panel come forward to the microphone and speak clearly for our transcriptionist.
We also request that persons making statements either during the open public hearing or any other portion of the meeting disclose whether they have any involvement, including and not limited to financial interests in any medical device company, including LifeCore or one of its competitors.
So before making your presentation to the panel, please state the nature of your interest, including such things as whether the company, LifeCore and other companies, paid for your expenses to attend the meeting and whether your organization receives funding from LifeCore or another device company.
I'll ask Mr. Weinstein to present the speakers who have beforehand requested a chance to address the panel in response to the Federal Register notice.
MR. WEINSTEIN: The following people have requested time to speak. I'll read their names, and that will be the order in which you may come up to the podium to make your presentation.
Dr. Michael Kettel.
Dr. Lena Holmdahl.
Dr. Russell Malinak.
Dr. Melvin Thornton.
And Ms. Bess Weatherman, who will be speaking this afternoon.
So if we can begin with Dr. Kettel, is he here?
CHAIRMAN RAMSEY: Because of the number of speakers we have at this open meeting, I'm going to ask if at all possible to try to limit your comments to five minutes. I'll raise my hand when you have one minute left just so you'll know approximately where we are.
DR. KETTEL: Thank you, Dr. Ramsey.
My name is Michael Kettel, and I am presenting today my views on the approvability of INTERGEL.
In introduction, I was a participant in the clinical trial that will be presented today and so I did receive some financial remuneration from LifeCore for that participation. However, they have not paid for my travel today.
I am a private practicing reproductive endocrinologist in San Diego, California, and have had extensive experience treating fertility patients with a variety of different disorders, including pelvic adhesions.
In our center, we treat over 2,000 patients a year between my partner and I with various causes of infertility. We perform over 400 surgical procedures per year, and have a variety of experiences in dealing with pelvic adhesive disease.
I, for the sake of my comments today, did a short analysis of our experience from last year just to give a sense to the panel of the significance of publications to the infertile population here in the United States.
We did 408 surgical procedures last year between two gynecologic surgeons. Of those 408 patients, there were over 200 of them that had pelvic adhesive disease, but there were 114 of them for which the pelvic adhesive disease was deemed probably the sole cause of their individual infertility.
That represents 28 percent of the surgical patients in our practice with infertility solely attributable to pelvic adhesive disease, certainly not a clinically insignificant problem.
Lastly, I would make a second point, and that has to do with experiences that I've gained over the years participating in clinical trials. I've had the opportunity and pleasure to participate in clinical trials for several devices which have been brought before the FAA and have gained approval for adhesion prevention.
There are currently three products that I'm aware of that have gained approval by the FDA for adhesion prevention and gynecologic health surgery. All of these three other devices have been proven to be effective, but limited, and tier limitation is based on their site of application. They're all devices which are placed on the surgical side exclusively and are, in fact, effective at producing adhesions at the surgical site.
However, for the members of the panel and others who are experienced with surgical procedures, one is -- and you will become, I think aware of later today -- aware of the fact that adhesions don't always just limit themselves to surgical site applications, and that INTERGEL is the only product thus far that's been brought before the FDA which has the potential applicability of also preventing adhesions at non-surgical sites.
Non-surgical site adhesions are clearly as important as surgical site adhesions, particularly as it pertains to disease around the fallopian tubes.
Thank you very much for allowing me this opportunity to speak.
CHAIRMAN RAMSEY: Thank you.
MR. WEINSTEIN: Ms. Holmdahl.
DR. HOLMDAHL: I have some overheads that I would like to show if possible.
CHAIRMAN RAMSEY: While we're getting started, would you state your institution and name and nature of interest?
DR. HOLMDAHL: My name is Lena Holmdahl. I am Associate Professor of Surgery at the University of Goteborg, with a special interest in information both from a scientific and clinical standpoint.
And the focus has been mainly on pathogenesis, but also on iteration of adhesion reduction produced in therapies, including design and production of such trials.
I have no financial interest in the sponsor, but the sponsor has admittedly reimbursed me for my time.
And what I'm going to say is in the very limited is I will try to show that clinical outcome studies when it comes to adhesion formation are either unsafe to prove efficacy or not feasible, at least not in the pre-market period.
The next one, please.
For the following reasons. The clinical outcomes that we would like to assess when it comes to adhesion relation is either small bowel obstruction, infertility, or positive pain.
And from a scientific standpoint there is an established relationship between small bowel obstruction and adhesions, and about 60 to 70 percent of small bowel obstruction is related to post surgical adhesions.
And when it comes to infertility that figure is flower, but there is an established relationship by the theorists, about 20 percent.
And when it comes to pain, from a scientific standpoint, it's very hard to get evidence that adhesions actually can cause pain. So I would say that there is no established relationship with the outcome variable.
The next one, please.
When we're assessing small bowel obstructions, we actually consider the assessment tools. That would be either abdominal surgery or clinical and radiological evidence of small bowel obstruction, but then we would have to relook at the other factors that possibly might cause small bowel obstruction.
And the incidence rate of small bowel obstruction after surgery is very low. From the literature it can be estimated to be 3.6 percent, and using the standard that is statistical in a sample size of consultations to test such a hypothesis.
Next one, please.
The same is true for gynecological surgery, and there the incidence is even lower. So the sample size increases, and it will be more like 40,000 patients.
Next one, please.
Even if we consider pain, the family would have the event rate, which is very poorly documented in the literature, and we would also have troubles in assessing pain after any clinical trial.
But if we assume that the event pain is intangibly a lot of major complications after surgery, the sample size would be very large and would be more like 36,000 patients.
The event tree is likely to be higher after pelvic surgery, and is likely to decrease the sample size, but still it would be very much.
Next one, please.
If we could consider pregnancy as a lifetime variable, the event rate is much higher, and that would then decrease the sample size. So we're approaching figures that we could handle, and the sample size would be close to 1,600 patients.
But the problem is that they will now -- they would like to have pregnancy as an outcome variable because there is an alternative treatment, which is ILEA (phonetic), and it's very likely that many of these patients would have benefitted from being referred to ILEA in the first place.
So to summarize, clinical outcome variables, additional models directed at fertility or pain are difficult to handle because of the subjectivity. There is delayed appearance of the outcome. We have large assessment tools. The incidence is poorly documented. They would require very large sample sizes, and there are alternative treatments available.
So the conclusion is that outcome -- clinical outcome studies are not feasible.
CHAIRMAN RAMSEY: Thanks for your comments and coming all the way from Sweden.
MR. WEINSTEIN: Dr. Malinak.
DR. MALINAK: I'm Russell Malinak. I have no financial interest in the sponsor. However, my expenses and trip have been paid and my professional time.
I'm Emeritus Professor of Obstetrics and Gynecology at Baylor College of Medicine. I recently retired from a gynecologic surgery practice of 35 years in that institution's related hospitals where my principal interest has been in reproductive surgery.
Thus the majority of operative procedures I have performed have been to restore or enhance fertility or to relieve chronic pelvic pain.
Included in my job as full-time faculty has been teaching of medical students, residents, and fellows in conducting research in women's health issues. A major focus of my teaching and research has been in the realm of post surgical peritoneal adhesion formation, in attempts to understand pathogenesis and, more importantly, to search for ways to reduce or eliminate these adhesions.
They remain an inevitable sequel to any operative procedure.
To that end, I have participated in multiple clinical trials of the application of new materials or methods to accomplish our goal. Of particular note, I participated in the pivotal trial of INTERCEED, the first adhesion prevention barrier approved by the FDA.
To my frustration, each promising new method has failed in one way or another. During my tenure at Baylor College of Medicine, I have observed and been privileged to participate in many phenomenal advanced in the science and art of obstetrics and gynecology.
Yet effective reduction in postoperative adhesions has eluded us and remains the largest unmet need in the entire realm of women's health care.
Several years ago I participated in the randomized clinical trial to evaluate the effectiveness and safety of INTERGEL. I was particularly pleased to join this study because it was the best designed clinical trial of an adhesion prevention product I had ever seen, and in contrast to the limitations of INTERCEED, it provided a method to reduce de novo adhesions, as well as surgical site adhesions.
A tremendous amount of time and resources were expended in executing this protocol, which typical of any randomized trial and surgery is among the most difficult to carry out in all of medicine.
I was pleased that the outcome measure was the adhesion itself rather than pregnancy or pelvic pain, both of which are multifactorial in origin and, therefore, provide less rigorous analysis.
Upon completion of the study, I was gratified to see clinically significant reduction in adhesions secondary to the product, which was also proven safe. In a patient population which had an uncharacteristically low incidence of adhesions, there was a five-found reduction in moderate to severe adhesions in the study group as compared to control, and a 31 percent reduction in reformed adhesions.
This reduction is clinically quite significant and would most likely be more impressive in a patient population at high risk for adhesion formation.
In the amended material, I was further pleased to see the focus. The fallopian tubes and ovaries are the organs most vulnerable to distortion by adhesions and, therefore, most likely to be associated in infertility and/or pelvic pain when significant adhesions form.
In the analysis of the data from this study, I have seen for the first time a simplified and clinically meaningful way to portray the results of adhesion studies in patients, not in animals, that is in the shift tables developed by the sponsor in conjunction with the FDA. This approach will aid in interpretation of future adhesion prevention trials.
In closing, it is my opinion that an adhesion reduction product has been identified by very good science, has been evaluated in a rigorous clinical trial, and has been shown by critical statistical analysis to be safe and effective.
It will be a disservice to women's health if this product is denied our patients in the United States of America.
Thank you for the opportunity to speak.
CHAIRMAN RAMSEY: Thank you, Dr. Malinak.
MR. WEINSTEIN: Dr. Thornton.
DR. MELVIN THORNTON: Good morning. My name is Dr. Melvin Thornton, and I'm Assistant Professor at Columbia University and currently the Medical Director of the Center for Women's Reproductive Care at Columbia University.
And I was a participant in the clinical trials for INTERGEL, and I did not receive financial assistance for being here today.
I came here today because I feel this is the most important meeting that is going to happen this year in women's health care because, as you heard from the previous speakers, adhesions are a major problem with surgery for women, and as a reproductive surgeon, it's difficult to train residents and teach them about adhesion prevention when there are no products out there that I can feel confident in giving the patient, knowing that when I leave the OR she's going to have a reduction in her adhesions.
But the reason for this, because the products that are approved are very difficult to use and take a lot of time to use, the majority of the physicians out there choose not to even attempt to use these products. So these patients are leaving the operating room with nothing to prevent adhesions from forming.
This INTERGEL has been shown to be effective and has been shown to be safe, and as a participant in clinical trials, I know for a fact that it works, and it's easy to use.
So this, if it's approved, the majority of physicians out there will definitely use this because it's easy for them to use, and using something is better than using nothing.
And I'd like to say thank you for the time.
CHAIRMAN RAMSEY: Thanks, Dr. Thornton.
Is there anyone else who would like to speak at this portion of the hearing? Anyone else?
CHAIRMAN RAMSEY: I'll turn it over to Les then.
MR. WEINSTEIN: I have two comments that were submitted before the meeting that was originally scheduled in June, and those comments were submitted for the June 4th meeting that was rescheduled to today, and I've been asked to read these portions of those two comments into the record.
One comment was submitted by the American Society for Reproductive Medicine.
The American Society for Reproductive Medicine and the Society of Reproductive Surgeons are pleased to submit comments to the Dispute Resolution Panel on the top of postoperative adhesions. Our intent is to emphasize to the committee the clinical importance of postoperative adhesions and devices and agents designed to reduce the development of these adhesions.
Postoperative pelvic and abdominal adhesions are a common problem. It is estimated that more than half of the more than three million women who undergo gynecologic surgery every year will develop some postoperative adhesions.
These postoperative adhesions can lead to significant medical complications, chronic pain, reduced quality of life, and increased cost of medical care.
Common examples of medical complications include bowel obstruction and infertility. In addition to these complications, postoperative adhesions may produce chronic pelvic pain, which may be mild and a nuisance for some, but severe and disabling for others.
Clearly, impaired fertility and chronic pelvic pain may reduce the quality of life experienced by those who develop adhesions. Furthermore, the diagnostic tests and surgical therapies performed to evaluate and treat the infertility and pain are costly and sometimes ineffective.
Finally, pelvic adhesions may also complicate and increase the morbidity of any subsequent pelvic surgeries. We see these clinical consequences of postoperative adhesions every week in our clinical practice. Good surgical techniques and the products that are currently on the market will not always prevent the development of postoperative adhesions in many of our patients.
As a result, there is a need to develop additional and effective methods to reduce the occurrence of adhesions after surgery. The development and use of new devices and agents that are capable of reducing the incidence, extent, and severity of adhesions are likely to reduce postoperative complications, pain, decreased quality of life, and the cost of subsequent medical and surgical care.
Signed by Dr. William R. Keye, spelled K-e-y-e, and Dr. R. Dale McClure. Dr. Keye is the president-elect of ASRM. Dr. McClure is the president of SRS.
The next comment was submitted by Dr. Barry Stewart of the Pacific Gynecology Specialists Group in Seattle, Washington.
As one of the principal investigators of INTERGEL prevention solution, I have had the opportunity to experience its characteristics and efficacy first hand. Its ease of application in gel form and the widespread surface coverage that it provided in confined spaces seemed particularly advantageous over other products available.
I also found it to be clearly helpful to my patients in preventing pelvic adhesions after myomectomy relative to Ringer's lactate.
I do not have the experience of the panel in analyzing the test data presented, but if safety and efficacy in the range of other such products can be demonstrated, the gel form in which INTERGEL is utilized would increase its application and the potential benefits derived for a large number of patients undergoing abdominal pelvic surgery.
I personally look forward to its approval and to my utilization of it in my surgical practice.
Signed Barry C. Stewart, M.D.
Those are the only comments that I've received.
CHAIRMAN RAMSEY: Let me ask the panel if they have any questions that they'd like to ask to any of the speakers during the open presentation.
DR. D'AGOSTINO: Can I?
CHAIRMAN RAMSEY: Yes.
DR. D'AGOSTINO: The speaker who said that you can't measure clinical outcomes in a particular premarket phase, is the suggestion then that we should rely on a surrogate? I mean, she didn't say that, but is that the implication?
DR. HOLMDAHL: Yes, I would propose that in the premarket period, to rely on adhesion information.
DR. D'AGOSTINO: Even though you don't know that pain relates, the surrogate relates to relieving pain?
DR. HOLMDAHL: That is true, but if you're looking into efficacy or an adhesion reducing, ending therapy, then I would suggest that it's better to look at the -- to measure adhesion in the postoperative period.
DR. D'AGOSTINO: Thanks.
CHAIRMAN RAMSEY: Any other questions?
CHAIRMAN RAMSEY: Okay. We're now going to move on. We're a little ahead of the time, which is great, but we're going to go ahead and move on to the presentations by the parties in dispute, and we're going to start with LifeCore's presentations.
As with the open hearing, everyone who comes up, I'd ask that you please speak clearly into the microphone for our transcriptionist, and also as before, before making your presentation, please state your name, affiliation, and any financial interest with the company.
Just to remind you, the definition of financial interest includes compensation for time and services or expenses of those and your assistance and staff in conducting a study, preparing a report, or appearing at the panel meeting on behalf of the sponsor, including paid travel. If you're an employee of the company, obviously you don't have to make those type of disclosures.
DR. BRACKE: Good morning, ladies and gentlemen. I am Jim Bracke. I am president and chief executive officer of LifeCore Biomedical, the sponsor organization.
I would like to start by thanking the FDA on behalf of LifeCore for making the Dispute Resolution Panel option available to sponsors of PMAs.
I would like to further thank Les Weinstein for a very extensive effort in coordinating this first Dispute Resolution Panel unit with the sponsor.
I would also like to thank all of the experts involved in this entire process who have put in considerable effort in preparing for today's meeting.
I would now like to introduce you to Dr. Karen Becker, who is representing LifeCore in this matter before the FDA. Dr. Becker is worldwide managing director of health care products at the Weinberg Group. She has worked for 15 years on the clinical evaluation of implanted medical devices, pre and post marketing and has published a textbook on the subject.
DR. BECKER: Thank you, Jim.
I have no financial interest in LifeCore, and I'm being compensated for my time.
May I ask for a clarification before I begin my remarks? The questions to be asked of the panel, would you repeat those?
MR. WEINSTEIN: I think that they were misread.
CHAIRMAN RAMSEY: Okay. Sorry. Let me try that again.
It says: "whether the PMA, as amended, provides reasonable assurance of the safety and effectiveness of INTERGEL for its intended use as an intraperitoneal instillate for reduction of adhesion formation following gynecologic pelvic surgery. In answering this question, the panel should determine whether a statistically significant difference between INTERGEL solution and control can be considered clinically significant; and, second, do the benefits of the product outweigh the potential risks, including any risk of infection?"
DR. BECKER: Thank you.
MR. WEINSTEIN: Does that clarify?
DR. BECKER: Yes.
Well, I want to first thank Les and CDRH for giving the sponsor the opportunity to talk to you today in a fair and open scientific forum about the scientific issues in dispute.
The sponsor's presentation today has four parts. I will provide a summation of the sponsor's position on the scientific issues with regard to the INTERGEL PMA as amended in June 2nd, 2000.
Dr. Johns will then present the results of the INTERGEL clinical trial. His presentation will be followed by a consensus comment on safety and effectiveness of INTERGEL by independent, well qualified clinical experts who are here today.
Lastly, a consensus statement will be provided to you by independent experts on the statistical issues.
First slide, please.
It is the sponsor's position that valid scientific evidence has been submitted to FDA sufficient to establish reasonable assurance of safety and effectiveness for INTERGEL adhesion prevention solution for the labeled indication in accordance with requirements enumerated in the federal Food, Drug and Cosmetic Act and applicable regulations.
We respectfully disagree with the determination by OBE that premarket approval requirements for this product have not been met.
This product is safe and effective. This submission has been thoroughly peer reviewed by well qualified experts. Today we will provide you with a summary of the data submitted as part of this project, the scientific basis for the sponsor's position, and independent expert opinion in support of the sponsor's conclusions.
Next slide, please.
As you have heard and as you are well aware, post surgical adhesions following gynecological surgery caused significant morbidity. Over three million gynecologic pelvic surgeries are performed annually in the United States according to the National Center for Health Care Statistics.
From 60 to 90 percent of these cases result in post surgical adhesions. Adhesions are responsible for infertility, bowel obstruction, complications of additional surgery, and in some cases chronic pelvic pain.
These outcomes are not self-limiting, do not resolve without intervention, and are difficult to manage. Over 400,000 women undergo adhesiolysis surgery every year in the United States. It has been reliably estimated that at least five percent of all hospital readmissions or due to complications from post surgical adhesions.
The rate of readmissions post surgery due to post surgical adhesions does not decline over time, but continues over ten years. This conclusion is based on the work of Dr. Harold Ellis and colleagues in the U.K. utilizing the Scottish National Health Service medical records and linkage database, which is the largest verifiable database on hospital readmissions due to adhesions. There were incorporated 30,000 patients followed for ten years.
The cost to the U.S. health care system of managing co-surgical adhesions is estimated to be $1.6 billion. This figure has been replicated on a per capita basis by data from the U.K.
I would point out though that this is a gross underestimate because it relies only on peritoneal adhesiolysis surgery.
Two products are available in the United States for reducing the risk of surgical site adhesions as adjuncts to good surgical technique. No adjuncts are available for reducing the risk of adhesions beyond the surgical site.
INTERGEL, the product under consideration today, is available to women in Canada, the European Union, and in many other countries throughout the world.
Next slide, please.
As you know, the clinical evaluation of post surgical adhesions is extremely difficult. Products approved for use by FDA have thus far been restricted to two site specific adhesions. Both of these approvals rely on adhesions as the endpoint.
Almost two years ago, FDA convened a meeting of the Obstetrics and Gynecology Devices Advisory Panel to consider the premarketing data requirements for adjuncts to reduce the risks of post surgical adhesions in gynecologic surgery. This meeting considered a draft guidance on barrier adhesion products. The focus of the expert testimony provided and the panel discussion which ensued was on the design of clinical trials for adhesion barrier products.
Some of you on the panel today were on that panel in January 2000: Dr. D'Agostino, I believe, and Dr. Shirk. Many other people present in the room today were on that panel. It was a very snowy day. Dr. Alan De Cherney provided testimony.
This meeting yielded three very important conclusions regarding the state of the clinical science and recommendations with regard to regulatory standards to be applied to these products in the United States.
First, the experts agreed that adhesions are an endpoint and not a surrogate. This position was expressed by leaders in the field, including Dr. De Cherney, Dr. Steven Schwaitzberg (phonetic), and the panel chair, Dr. George Blanco (phonetic), in his summation of the testimony and panel discussion.
Secondly, there was agreement that these studies, even with adhesions as the endpoint are extremely difficult to do. Outcome studies to assess fertility, bowel obstruction, and pain cannot be conducted pre-marketing.
Thirdly, there was a consensus that adhesion assessment in any given study should be uniform and systematic, considering such features as location, extent, incidence, and severity.
No single adhesion assessment tool is used by investigators uniformly, but the American Fertility Society scoring system for adnexal adhesions was cited as the method most commonly employed by gynecological surgeons.
Next slide, please.
Now, I will review for you how we ended up here today. The following is a synopsis of the regulatory history of the submission under consideration.
In 1995, following a pilot study to assess safety and to gauge sample size calculations, FDA cleared an IDE for the INTERGEL pivotal trial. This slide is the study hypothesis for the clear protocol developed in collaboration with FDA.
The objective of this multi-center trial is to assess the safety and efficacy of INTERGEL in preventing or reducing adhesions in patients undergoing peritoneal cavity surgery.
The study protocol cleared by the division as the basis for a pivotal trial in support of the PMA was clearly designed to measure adhesions. The INTERGEL clinical trial utilized the AFS scoring system to assess in a systematic and uniform manner the incidence, extent, and severity of adhesions at 24 anatomical sites, prospectively in all patients in a blinded manner at every center in the same way. A reduction in the mean score of all 24 sites evaluated was designated as the primary endpoint for this trial.
The INTERGEL trial was multi-center, prospective, randomized, controlled, blinded. It required two invasive procedures: a laparotomy followed by a second laparoscopy.
This was an extremely resource intensive trial. It required over two years to accrue subjects to this study. It could not be repeated today given the minimally invasive surgical techniques used and increasing reliance on in vitro fertilization.
Further, I must point out that investing in the development of adhesion prevention problem products is problematic when six years after the initiation of a study, such as this one, we are still discussing whether or not adhesions are an endpoint.
At least six products of which we are aware are in various stages of development for nonsurgical site adhesion prevention as adjuncts to good surgical technique. These development programs are on hold, pending a consensus by the regulatory -- pending action by the regulatory community that is in agreement with clinical experts on this matter.
Next slide, please.
In 1999, the INTERGEL PMA was filed with FDA and granted expedited approval status on the basis of an unmet public health need. The proposed intended use was supported by statistically significant differences in a prospectively identified primary endpoint and secondary endpoints. That proposed intended use was as follows.
As a single use IP instillate reduction of adhesions following gynecologic pelvic surgery. It has been shown to reduce the incident, extent, and severity of post surgical adhesions throughout the abdominal cavity when used as an adjunct to good surgical technique during laparotomy procedures.
In the course of the review of this PMA, FDA asked for additional information. In December of 1999, about one month before the General and Plastic Surgery Devices Advisory Committee was going to consider the approvability of this PMA, the sponsor received a letter from the agency requiring more information, including, number one, a supplemental infection potentiation study in rats, a larger study than the previous one that was negative.
Secondly, FDA required that the sponsor submit AFS scores. These were part of the conditions of approval.
Obviously with one month prior to a panel meeting, the animal safety study -- there was not enough time to conduct the animal safety study.
Next slide, please.
When the general Plastic Surgery Devices Panel convened to consider the original INTERGEL PMA in January of 2000, as I said, a supplemental infection potentiation study in rats was not yet completed. As you will note from having read the panel transcripts, FDA questioned the clinical significance of the study hypothesis developed in 1995, adhesion reduction.
At that panel meeting, the required analysis of AFS scores was not considered due to time constraints. That is also reflected in the record.
Finally, as noted in the panel transcripts, FDA had concluded that the rate of infection with INTERGEL was higher than that in control. That panel was told that among the four cases cited in the INTERGEL group -- I'm sorry -- the panel was not told that among the four cases of postoperative infection cited in the INTERGEL group, the panel was not told that one was a case of chicken pox and one was a head cold. This was an unfortunate misunderstanding that had a significant impact on the vote.
The panel voted five to two against approval of the PMA, citing concerns about safety, the need for the completion of an additional animal infection potentiation study, and questions about clinical utility.
Interestingly, just two weeks later, a different advisory panel convened by FDA, the Obstetrics and Gynecology Devices Panel that I've previously mentioned, met and agreed that adhesion reduction is, in fact, a suitable endpoint as an adjuvant for this intended use.
Next slide, please.
In June of 2000, the sponsor, following discussions with FDA, submitted an amended PMA to address each of the concerns raised by the GPS Panel at FDA. That is the submission before you today.
No FDA advisory panel has ever considered this submission or these data in the context of this intended use.
The product has a revised intended use narrowed to reflect the adhesion assessments in the pivotal trial that speak directly to clinical outcomes of accepted importance to gynecologic surgeons: namely, adnexal adhesions, reformed adhesions, and surgical site adhesions. This is an OB-GYN indication as opposed to a general surgery indication. The adhesion assessment data in support of this intended use were prospectively gathered and provided in the original PMA.
The PMA as amended provides a systematic review of the clinical literature validating that assessment of adnexal adhesion scores using the AFS score as a valid prognostic tool.INTERGEL
The PMA as amended also provides the new, larger infection potentiation study in rats, which was required by FDA. It was negative, as were the results of an earlier, smaller study.
Since this meeting almost two years ago, INTERGEL has been marketed all over the world, with approximately 35,000 units sold. So we now have the benefits of more clinical experience. This broad clinical experience confirms the safety and performance of the product. There is no data to suggest that there is an increased risk of infection.
The PMA as amended also provides statistical analyses and resolution of statistical issues raised by the advisory panel and FDA, including the issues of poolability and incomplete ascertainment, although we note that at the original panel, Dr. Metz concluded that the data can be pooled.
Today you have the benefit also of expert opinion. These data -- this peer review of the clinical trial was not available two years ago to FDA or to the original GPS Advisory Panel. Leaders in the field of reproductive medicine, obstetrics, and gynecology, adhesion pathophysiology management, and postoperative infection have reviewed this submission before you in depth and agree that the product is safe and effective for the intended use.
It has also been peer reviewed by a panel of experienced and well qualified statisticians who conclude that the sponsor's analysis is sound.
And finally, very importantly, you are the first panel to consider an adhesion prevention adjuvant following the consensus reached by the OB-GYN Advisory Panel on barrier adhesion products. At that panel it was agreed that adhesions are an endpoint, not a surrogate.
This is the first opportunity to act on that medical consensus in a forum that contributes directly to an improvement in public health.
Next slide, please.
There is consistent evidence, as you will hear, that INTERGEL is effective. The product has been shown to be effective in a well designed, controlled clinical trial. The primary and secondary outcomes of the study endpoints, all measures of adhesion incidents, it's extent and severity consistently demonstrate an improvement compared to control, which was lactated Ringer's solution.
The effectiveness established in the pivotal trial for INTERGEL confirm the findings of the pilot study in which reduction and adhesion formation was observed.
Review of these data by leaders in the field led them to conclude that the manner of assessing adhesions in this trial is, in fact, appropriate. The magnitude of the changes observed compared to control are clinically significant and meaningful. A substantial proportion of the population benefitted, and the benefits of the product outweigh the risks.
You have their consensus report in writing for the record, and they are here to speak with you today.
In addition to the clinical data, the effectiveness of INTERGEL in reducing adhesions is supported by data gathered in animal models and the mechanism of action, that is, as a resorbable barrier adhesive, a mechanism of action that shares in common with the two site specific products on the market.
Next slide, please.
The evidence is consistent that INTERGEL is safe. The product has been shown to be safe both in clinical use and in animal studies. The pilot study of INTERGEL was a safety study. It generated no data of concern with regard to infection or any other adverse event.
The pivotal trial of INTERGEL confirms the safety of the product. No adverse events, including postoperative pelvic infections occurred at a higher rate compared to control.
The post marketing experience with INTERGEL, approximately 35,000 units sold as Iaset (phonetic) confirms the safe profile and has over the last three years raised no issues of concern. The infection potentiation study in rats was negative the first time and the second time.
FDA has raised the concern that patients treated with INTERGEL may have increased risk of postoperative infections despite the data I have just summarized. As part of the clinical review of these data by experts, a complete and independent review of the patient records from the pivotal trial was undertaken to examine the safety results.
Dr. Sebastian Faro and Dr. John Sever developed a protocol and a criteria for the review of these data. The conclusion of their exhaustive review is that the postoperative pelvic infection rate between INTERGEL and control is not different, and differences in the absolute number of cases reported are not clinically meaningful.
You will have an opportunity to hear from them directly today.
Also, I will note what you are well aware of, that within our common adverse event occurring at less than five percent incidence, a clinical trial of approximately 12,000 subjects would be required to determined the true incidence of postoperative infection. Such a study could never be conducted premarketing and is not warranted by this body of evidence.
Next slide, please.
The statistical methods used in this trial are salient. FDA has questioned many aspects of the statistical method applied to the INTERGEL pivotal trial. Some questions were discussed at the advisory panel meeting and during the review process of the PMA.
The statistical methods applied to the analysis of this data set are entirely appropriate. The sponsor has responded thoughtfully and rigorously to every concern raised by FDA.
The sponsor requested a peer review of these data by three well qualified experts who confirmed that the analysis as presented in the PMA, the analyses, are appropriate, and the results of the trial robust.
Their opinion has been provided to you in writing for the record, and you will hear from them directly today.
The sponsor did adhere to the data analysis plan in the study protocol. All data analyses were conducted in accord with good statistical practices, ICH standards, and sound clinical and statistical rationale.
It is appropriate to utilize the data from all centers. The sponsor properly considered the impact of missing data at second look in complete ascertainment, including utilizing a worst case imputation methodology required by FDA in the protocol cleared in 1995.
You should note that this worst case imputation methodology has no sound clinical or statistical rationale, but was provided as required by FDA.
This is the misnamed ITT analysis that you have seen referred to in documents from the agency. You will note that even with application of this unscientific worst case imputation, the primary endpoint of the INTERGEL trial was still statistically significant compared to control.
We conclude that the statistical questions related to this trial have been resolved and welcome your experienced peer review of the methods utilized.
Next slide, please.
Finally, I have put up here the intended use for the submission under consideration today. It is very important to distinguish correctly between the intended use of a product and a study hypothesis. The data gathered in a trial or trials and all of the available data is what drives the statement of intended use of a product, and of course, it is the data that determines the labeling in its entirety, including adequate directions for use, warnings, precautions and claims.
The original intended use for this product is, in the opinion of the sponsor, well supported by the data in the original PMA. However, since the GPS Advisory Committee and FDA questioned the clinical utility of adhesion reduction, the sponsor proposed a more narrow intended use from the same data set.
Specifically a new intended use, a new product was provided to the agency and is the submission under review today. The intended use is as follows.
INTERGEL solution is a single use IP instillate indicated to reduce the likelihood of developing moderate or severe postoperative adenocele adhesions in patients undergoing adhesiolysis or myomectomy during conservative gynecologic pelvic surgery by laparotomy when used as an adjunct to good surgical technique.
INTERGEL solution was also shown to reduce adhesion reformation to sites in addition to the adnexa and adhesion formation at surgical sites, including the anterior abdominal incisions.
I want to emphasize that this is not a subset analysis. This is not a retrospective analysis, and this is not an inappropriate ad hoc analysis. This intended use reflects the data prospectively gathered in all study subjects. Ten of the 24 anatomical sites assessed were the tubes and the ovaries.
These data were provided in the original study cohort, and additional analyses were provided in the PMA as amended to support this label.
The submission you're considering today provides a scientifically sound resolution to questions of clinical utility, safety, and statistical methodology.
Last slide, please.
Finally, this slide summarizes for you the primary and secondary endpoints prospectively identified in the INTERGEL clinical trial. It also provides the endpoints relied upon in the approval of INTERCEED Seprafilm for comparison.
All of these products rely on adhesions as the endpoint. The magnitude of the effects reserved for these adjuvants is comparable. The data on INTERGEL is consistent and, in the sponsor's opinion and in the opinion of experts in the field, compelling.
The AFS scores listed here were provided in the original clinical study report and were required by FDA.
Adhesions following gynecologic surgery is an important problem for women that requires attention now. Your due consideration of this submission is very much appreciated.
I will now introduce you to Dr. Doug Johns from Gynecare/Ethicon R&D, who designed the INTERGEL pivotal trial with Dr. Gere diZerega.
DR. JOHNS: Good morning. My name is Doug Johns. I'm an employee of Gynecare and a consultant to LifeCore Biomedical.
I will be presenting an overview of the INTERGEL pivotal clinical trial design and the data.
A pivotal multi-center study was initiated to assess the safety and efficacy of INTERGEL in reducing adhesions in patients undergoing peritoneal cavity surgery by laparotomy, with a planned second look.
The study was a third party blinded, randomized, concurrent control design. It was carried out at 11 centers in the United States and give centers in Europe.
Female patients 18 to 45 years of age undergoing peritoneal cavity surgery by laparotomy with a planned second look received 300 milliliters of INTERGEL or lactated Ringers just prior to closure.
The analysis plan and the protocol specified all treated patients were to be evaluated for safety and all evaluable patients were to be evaluated for efficacy, and an evaluable patient was defined as all patients that completed the second look laparoscopy.
The protocol also included an FDA required intent to treat analysis, which required a worst case imputation for missing data at second look.
In addition, the protocol also defined a subset of this previous group which excluded subjects who refused second look for reasons unrelated to the device and informed censoring analysis.
It is important to note that the protocol did not identify the intent to treat population as the primary analysis; that this worst case was done because the agency insisted upon seeing it; and at the same time encouraged us to do other analyses, which we felt were more appropriate.
This presentation will focus on the evaluable population for efficacy and all treated patients for safety. The intent to treat analyses referred to will be discussed in detail by our statistical experts later.
Adhesions were assessed at 24 anatomical sites at the initial laparotomy procedure and, again, at second look. The 24 sites throughout the peritoneal cavity included 16 pelvic sites and eight abdominal sites. The 24th site, the interior abdominal incision, was the laparotomy incision from the first surgery. So there were 23 sites assessed at the first operation and 24 assessed at the second look.
An adhesion assessment chart was used by investigators to capture the data. If an adhesion was present for each site, it was marked yes or no. If an adhesion was present, the severity of that adhesion was then determined. Filming avascular adhesions were considered mild, while organized, cohesive, vascular and/or dense adhesions were considered severe.
Extent was also captured for each site where an adhesion was present. Adhesions were considered to be localized if less than one third of the site was involved with adhesions. It was moderate if between one third and two thirds of the site was involved with adhesions, and extensive if more than two thirds of the site was covered. This, in fact, is the AFS methodology.
After the characteristics of the adhesion were established, the surgeon would indicate whether or not the adhesion was lysed and, if so, by which method. The methods included sharp dissection, cautery, laser, or blunt dissection.
In addition to adhesiolysis, other surgical intervention at each of the anatomical sites was captured. The presence of endometriosis was specifically noted if it was surgically treated, and if so, how, which method.
If other surgery was done, that is, other than adhesiolysis or other than surgical treatment of endometriosis, this was also noted, and the use of sutures at each of the anatomical sites was also noted.
From this data, all of the primary and secondary endpoints prospectively identified in the protocol can be calculated.
Postoperatively, patients were monitored for adverse events and medications were tracked throughout the study. Laboratory evaluations and abdominal auscultation percussion were carried out at day three, and between days seven and 28, and finally a second look was carried out at six to 12 weeks following the initial surgical procedure.
The primary efficacy identified in the protocol, which was intended to support a general surgery label, was an adhesion score using the adhesion scoring method of the American fertility society applied to these 24 sites. We've termed this the modified AFS score, of mAFS.
The focus of the analysis for the June 2nd, 2000 amendment was changed to provide our results in a clinically relevant context, that is, the AFS score for adnexal adhesions. It is important to note that the AFS data was required by FDA. Data from all patients who had second look were included, and the data used to calculate the score was prospectively collected and presented in the original clinical trial report.
The June 2nd amendment also includes additional analyses of these data in response to the points raised by the panel and FDA and as Dr. Becker alluded to earlier.
Secondary endpoints included adhesion incidents which was expressed as a proportion of sites with adhesions as determined by the number of sites divided by the number of possible sites with adhesions.
Adhesion extent was determined using a three point scale, and adhesion severity was determined using a three point scale as well.
Other secondary analyses specified in the protocol included the different adhesion categories, of course, all adhesions, reformed adhesions, de novo adhesions.
The de novo category included surgical site and nonsurgical site de novo adhesions.
And finally, all surgical site adhesions, which includes both free formed adhesions and de novo surgical site adhesions.
We also did analysis by surgical procedure. This was not specified in the protocol, but was done at the request of FDA.
Turning to the study results, 281 of the 303 patients that were randomized for the study were treated. Two hundred and 65 of these patients completed the study which included a second look laparoscopy. That leaves 16 patients, which is less than six percent, who discontinued from the study after receiving treatment.
The reasons patients did not return for second look is shown for you here. One patient was pregnant. One patient had a failed laparoscopy due to obesity. Seven patients had no complaints, but simply did not want a second surgery. Six other patients had some complaints, but also did not want a second surgery, and one patient was truly lost to follow-up.
So if you exclude patients who did not return for reasons unrelated to the device, you're left with four treatment patients and three control, the ones on the bottom.
Surgical procedures. Myomectomy was by far the most common procedure performed. Approximately 70 percent of the patients in the study had a myomectomy procedure.
Adhesiolysis was second. Approximately 50 percent of the patients.
Ovarian procedures and tubal procedures were also fairly common, as was surgical treatment of endometriosis.
Dr. Becker has already mentioned pooling of the data, and our experts will be discussing that topic in more detail in a little while, but it's important to note some of the primary considerations, and to this end, I would like to share some of the European and U.S. data as it is appropriate.
As you'll see on this slide, myomectomy was the most common procedure in the United states, around 70 percent of the patients, followed by adhesiolysis, which was on the order of 35 to 40 percent.
This trend was reversed in Europe. Approximately 40 percent of the patients in Europe underwent myomectomy, and approximately 70 percent of the patients underwent adhesiolysis.
However, these surgical procedures carried out in the U.S. and in Europe were anticipated and allowed for under the protocol. They involved similar surgical technique and differed only in their overall percentages in the two populations.
Now, for the data. At baseline patients in the INTERGEL group were similar to patients in the lactated Ringer's control group. The number and proportion of adhesions and the modified AFS score, severity, extent, and AFS scores were all similar, indicated here by the nonsignificant p values.
Therefore, the amount of adhesions remaining after surgery was also similar following the surgical intervention shown for you here. Most of the adhesions were lysed. Similar amounts of additional surgery were done. So what was left behind was a small number of adhesions.
At second look, the original primary efficacy variable, the modified AFS score was reduced from 2.18 to 1.21. This result was highly significant and amounted to a 44 percent reduction.
The modified AFS score was also significantly reduced in the U.S. and European populations. That data is shown here. Note that the treatment group and control group baseline in the United States is similar. The treatment group and control group baseline data in Europe is similar.
However, the baselines in Europe were higher than that in the U.S. The different baseline starting points are a direct result of the difference in the percentages of the surgical procedures carried out.
Patients undergoing myomectomy procedures tend to start out with a very low baseline modified AFS score, and this score goes up at second look. INTERGEL significantly reduces adhesions, however, for the myomectomy population shown through here.
Now, considering the myomectomy patients in both the U.S. and Europe, you can see that similar trends are observed. In the U.S. and in Europe, both populations start off with a low baseline and increase at second look. In both populations INTERGEL reduces adhesions.
Patients undergoing adhesiolysis tend to start off with a higher baseline adhesion score, as shown for you here, and INTERGEL significantly reduces adhesions at second look compared to control.
As you would expect, baseline adhesion scores for the U.S. and European patients followed the same trend. Again, similar control and treatment groups at baseline and similar trends in the U.S. and Europe.
Now, to the AFS score. At baseline, the treatment and control groups were similar, as I told you, and at second look, the AFS score was reduced by 61 percent, from a mean score of 4.23 to a mean score of 1.67. This result was highly significant.
The secondary efficacy variables were also reduced by INTERGEL. The proportion, extent, and severity were reduced by 17 percent, 27 percent, and 32 percent respectively.
Reductions for each adhesion type were also observed. The proportion of reformed adhesions was reduced by 31 percent. Surgical site de novo adhesions, 23 percent.
The reduction in the proportion of de novo adhesions was not statistically significant, although the trend was in favor of INTERGEL. However, the extent and severity of each adhesion type as shown for you here was significantly reduced in all cases. These numbers ranged in the 30 percents for reduction across the board for the different adhesion categories.
The modified AFS score, the primary variable for all of these adhesion types, was also reduced for each of these adhesion types. These reductions ranged from 45 to 49 percent. These results were all statistically significant.
Now, while these averaging techniques can be used to compare treatment and control adhesion reduction, they tend to obscure individual patient benefits, and you're left with the question: what does it mean to reduce a mean score from a two to a one?
The means can be looked at to test for statistical difference between the treatment and control groups, and I've just shared that with you. to answer the question of clinical significance, one must look at the results on a patient-by-patient basis.
Individual patient results can be ascertained by evaluating the number of patients who shift from one AFS category to another. Now, recall the AFS score for adnexal adhesions was developed to provide a way of assessing the patient's adhesion status, but also to provide a prognostic indication.
The prognostic indication is shown for you here. Minimal adhesions are defined as AFS scores between zero and five; mild, AFS scores between six and ten; moderate, AFS scores between 11 and 20; and severe, AFS scores between 21 and 32.
Note in the INTERGEL group, 109 patients started off with minimal adhesions. Of these 109, 103 remained minimal; four became mild; one became moderate; and one became severe.
Similarly, 109 patients in the control group had minimal scores at baseline. Of these, 96 remained minimal; six became mild; three became moderate; and four became severe.
Overall, at second look there are more INTERGEL patients in the minimal category -- next slide, Gary -- compared to control. There were 121 patients with minimal compared to 105 in the control group, and there were fewer patients in each of the mild, moderate, and severe categories.
In fact, the number of patients with moderate and severe adhesions was reduced -- next slide, Gary -- from 17 in the control group to only three in the INTERGEL group, 17 moderate and severe to three moderate and severe.
Analysis using the Cochran-Mantel-Haenszel test controlling for baseline level indicates a highly significant p value.
Now, the published literature indicates that patients with minimal and mild adhesions tend to do well compared to patients with moderate to severe adhesions. Our clinical experts agree with this conclusion as well, and a s a result, we've combined these groups into what we've termed a binary analysis where patients with moderate to severe adhesions are considered to be treatment failures.
Only three of the 122 patients, INTERGEL patients, shifted from the minimal-mild category to the moderate-severe category compared to ten of the 117 control patients.
All nine patients in the INTERGEL group that started off in the moderate-severe category improved to the minimal-mild category. This is compared to only ten of the 17 control patients.
Overall, only three patients in the INTERGEL group had moderate or severe adhesions. This is two percent compared to 17 percent or 13 percent of the control population.
The reduction in treatment failures from 13 percent to two percent is not only statistically significant; it is clinically significant as well.
I think it's also important to note that ten, over half of the control group treatment failures, came from patients who started out with minimal or mild adhesions or patients who did not have an adhesion problem prior to their first surgery.
Analysis of the failure rate by surgical procedure was also carried out. In all cases the INTERGEL patients' risk of treatment failure was reduced compared to control. For the two most common procedures, myomectomy and adhesiolysis, this result was highly significant.
Now, as previously mentioned, the FDA was concerned about the continent enrollment, U.S. and Europe, and the adhesiolysis status, that is, whether the patient had no adhesions, minimal mild adhesions at baseline or moderate or severe adhesions.
To respond to this, a Cochran-Mantel-Haenszel analysis was performed that evaluates the percentage of patients with moderate or several AFS scores at second look stratified by continent of enrollment and adhesiolysis category. This table presents those results.
Note that the computation of the common relative risk is justified by the nonsignificant Breslow-Day (phonetic) test of homogeneity and after adjustment for continent and for adhesiolysis category, there remains a fivefold reduction in the risk of moderate or severe adhesions, and that result is highly significant.
This is consistent with the same CMH analysis stratified for an adhesiolysis subgroup only, as well as the unstratified analysis. The similarity and consistency of these results indicate that the adjustment for continent and/or adhesiolysis status had little impact on the risk of developing moderate or severe adhesions at second look.
Turning now to safety, the safety assessment of INTERGEL includes the preclinical studies, the adverse events, laboratory evaluations, medications, and gross observations from second look from the clinical studies, as well as our INTERGEL post marketing experience.
You've heard most of this previously from Dr. Becker, but I'd like to review the adverse event data from the pivotal study in a little more detail.
I apologize for the busy slide.
The incidence of commonly reported adverse events is shown for you here, presented by body system. This list includes all AEs that were reported for five percent or more of the patients in either group. The only event that reached statistical significance was allergic reaction.
However, it was the lactated Ringer's group that at a higher proportion of this event, upon closer scrutiny, however, you'll note that it was, in fact, seasonal allergies that were being coded under this heading and not actual allergic reactions to the product.
You'll also note that infection was not listed in this table because it was not observed in five percent of the patient population, and this is not surprising as our clinical experts have advised us. They would expect infection rates for the gynecologic surgery population in this study to range from between 2.4 to 3.3 percent.
At the previous GPS panel meeting, the agency reported a four to one risk of increased infection for the INTERGEL patients. The data that led to this statement was the adverse event code for infection, for the body as a whole system, for the U.S. population only.
In fact, there were four INTERGEL patients and one control patient in the patient listing. The event that led to two of the INTERGEL patients being coded as having an infection were chicken pox and a head cold.
Subsequently, we have reviewed the data in detail, and we have had independent clinical experts that you will hear review the data in detail. If you include all wound infections and all infections coded by the investigators as possibly being related to treatment, you will find ten INTERGEL patients and four control patients.
This still includes patients being coded as having an infection who had chicken pox and a head cold, as well as a patient with a bladder infection, a patient with a positive chlamydia culture at the time of surgery, and a patient with a vaginal fungal infection. Clearly some of these infections are unrelated to the device.
In fact, the investigators coded only four of these events as being possibly related, three treatment and one control. As you will hear from our clinical experts, when the data are considered appropriately there is no difference in the infection rate between INTERGEL and control.
In summary, I'd like to remind the panel that the primary and nearly all of the secondary variables were significantly reduced by INTERGEL and that the safety of INTERGEL in this surgical population was comparable to lactated Ringers.
The mAFS scores and the AFS scores were all reduced. The proportion of adhesions was reduced. The number of treatment failures was reduced, and different types of adhesions were all reduced. Clearly, the data obtained in this prospective, blinded, multi-center trial demonstrated that INTERGEL adhesion prevention solution reduced adhesions compared to good surgical technique plus Ringer's lactate.
That these reductions are clinically meaningful will now be assessed by our clinical experts, and thank you for your attention.
I'd now like to introduce Dr. Luigi Mastroianni, who is the William Goodell Professor of Obstetrics and Gynecology and the former chairman of obstetrics and gynecology from the University of Pennsylvania School of Medicine.
DR. MASTROIANNI: Good morning.
May I have the microphone here?
DR. JOHNS: It works.
DR. MASTROIANNI: Yes. As you've heard, I'm Luigi Mastroianni, and I'm Professor of OB-GYN at the University of Pennsylvania, where I've been working in the field of infertility and in basic aspects of reproductive biology for at least three decades.
My areas of interest are fallopian tube function and the function of the adnexas specifically, and clinically I've been vitally interested in the surgical approaches to infertility and to the reproductive tract.
My introduction to INTERGEL came when I was asked to join a group of clinical experts to consider issues related to INTERGEL this last year. I have no financial interest at all in INTERGEL or in the company, nor do I have any financial interest in any device companies. I'm being compensated for my time and being reimbursed for my expenses.
Well, we were asked to review voluminous data most of which has been reviewed in previous presentations, and we were asked to look at the clinical data on safety and effectiveness and to consider the study design and methodology for assessing adhesions and clinical significance of the findings and also the potential for increased risk of infection.
The clinical experts who were involved in this endeavor included Alan De Cherney, professor and chairman at the University of California in Los Angeles; Harold Ellis from the University of London; Sebastian Faro from the University of Texas Health Sciences Center; and Lena Holmdahl from Goteborg University in Sweden; and Russ Malinak from Baylor College of Medicine; Mark Martens, who's Chairman of Obstetrics and Gynecology of Franklin Square Hospital in Baltimore; and John Sever, Professor of Pediatrics, Obstetrics and Gynecology, Microbiology and Immunology at George Washington University in this city.
And each panel member independently reviewed all of the relevant data, and it was voluminous, many inches, and the data which we considered and the documents which we considered are listed in this slide, and I won't read them, but the panel met and jointly provided a consensus opinion, which you have, I think, in your packets.
And actually the consensus opinion found that there was a valid and reliable body of evidence which had been provided to support the conclusion that INTERGEL is an effective adjunct for the reduction of post surgical adhesions in gynecologic surgery, and the study design, execution and analysis provide valid data to support the proposed intended use.
And finally, the benefit of INTERGEL use for the labeled indication outweighs any potential risks.
The sponsor has provided valid scientific evidence to demonstrate that the product does not pose an unreasonable risk of injury or illness.
Well, today we are going to focus on four areas. We'll review the validity of the study design. I'll do that, and then Dr. DeCherney will discuss methodology for assessing adhesions, and the clinical significance of same.
And then Dr. Faro will speak specifically to the issue of increased infection or infection overall associated with the use of the product.
Now for the validity of the study design. The committee found that it was really exceptionally good. It was a multi-center, randomized, prospective, blinded, controlled trial in women undergoing pelvic gynecologic surgery by a laparotomy. The study included centers in the U.S. and Europe. European patient characteristics were reviewed, demographics, baseline adhesions, surgical procedures, results, and they do not confound or bias the study.
There was randomization at each study center. Adhesion assessment procedures were uniform. Baseline adhesions were comparable between INTERGEL and control group, and there was a single protocol which was used.
The design addresses clinical utility. Adhesions are a clinically valid endpoint, in our opinion, and not a surrogate. Adnexal adhesions do impair fertility. Reformed adhesions, de novo adhesions, adhesions at the surgical site, other anatomical sites are undesired. Adhesions are the enemy in reproductive surgery.
Other outcome studies premarketing are really not feasible, and this has been spoken to by earlier presenters. I mean, bowel obstruction is very rare. It's clearly a major complication of myomectomy, an operation which I've been involved in for many, many years. It's so important to give women an option of preserving their reproductive function even into the late reproductive years in the times, and myomectomy is an operation which is increasingly used for that purpose, and adhesions are still one of the major complications.
And bowel obstruction, although it does not occur very commonly, is a serious, serious issue.
Fertility is an endpoint. Well, it's multi-factorial, very long in duration. There are alternative treatments.
Pain can be transient, subjective, variable in its etiology and be very hard to use that as an endpoint, the improvement of pain.
And as for the logistics, the accrual rate, sample sizes, and duration of studies, using any of those endpoints would be just outrageously prolonged and inexpensive and, I think, impossible.
Well, at this point I think we ought to turn to the issue of the validity of the methodology, and you know, we all look at data as we ready articles and consider in each case the methodology which was used to acquire those data.
And so Dr. DeCherney will speak to the validity of the methodology for assessing adhesions.
DR. DeCHERNEY: Thank you.
I'm Alan DeCherney. I have no financial interest in INTERGEL or companies associated with INTERGEL.
And today I'll address the American Fertility Society evaluation of adhesions and effectiveness as far as adhesion reduction and INTERGEL as an adjunctive therapy.
Adhesions were evaluated in the INTERGEL trial from observation at three sites on the ovary, two sites on the fallopian tube, which is consistent with the AFS method. This was published in 1988. It is a scoring system which was a consensus statement by the American Fertility Society, now the American Society for Reproductive Medicine.
The data was prospectively gathered, uniformly in both groups, and was an anatomically correct calculation. Clinical literature and experience indicate that the extent and severity of adhesions correlates with fertility prognosis. This is pretty much based on the surgical literature where lysis of adhesions occurs and fertility is increased in patients. So it's based pretty much on the surgical literature of which there is extensive information.
The AFS scoring system provides a standardized, functionally sound basis for quantifying the impact of pelvic adhesions or fertility potential. There's only one study that looks at this precisely. This was published in Fertility and Sterility in 1994. This, too, was a consensus statement, and found that the interobservational correlation coefficient was 0.7 as far as the scoring system was concerned, and actually when you looked at the high end, patients with severe adhesions, the correlation coefficient for observation was even higher.
The next slide just looks at three groups of patients, all patients with myomectomy and adhesiolysis, the two most commonly reported procedures in this study utilizing INTERGEL and compared to lactated Ringers as a control, and as mentioned before, you can see the results are statistically significant, and certainly clinically significant as far as reduction of adhesions in those patients that have moderate and severe adhesions, which are those patients most -- who have the most compromised fertility because of the inability of the ovary in the tube to be juxtaposed as far as ovulation induction, ovulation and ovary capture is concerned.
The risk of moderate and severe adnexal adhesions was reduced fivefold in the INTERGEL group compared to the control group, as shown in the previously slide, 13 versus two percent. Patients with moderate to severe adnexal adhesions have little to no likelihood of natural conception because of this barrier that's presented between the two and the OS site.
Now, the changes observed in the modified AFS score, 61 percent reduction, for patients with baseline to second look the shift scores that we just demonstrated were statistically significant and clinically meaningful. Obvious, those patients with the most severe adhesions are those that have the most compromised in their fertility.
The reason for the modified AFS score in the original study was to make this a comparable study for surgeons, as well as gynecologists. Other sites included the bowel and omentum, things that were more pertinent to the general surgeon than to the gynecologic surgeon as far as postoperative adhesions, especially bowel adhesions and bowel obstruction in the future.
The effectiveness of INTERGEL in reducing the risk of adnexal adhesion formation was observed for all patients in subgroups of surgical procedures in the trial as demonstrated with the all patients, myomectomy and adnectal surgery.
So in conclusion, a consistent response in reduced adhesion incidence was seen in other endpoints as well. Certainly a 31 percent reduction in reformation adhesions is impressive, as are the de novo or surgical site adhesion reduction of 24 percent.
These, of course, are clinically important results, as mentioned by the other panelists and the speakers in the beginning of the program, as far as fertility is concerned. The results are comparable to previously approved adjuncts for site specific adhesion reduction.
And now Dr. Faro will discuss the risks.
DR. FARO: I'm Sebastian Faro, obstetrician, gynecologist with a special interest in infectious diseases, and I have no financial interest in LifeCore, and they are reimbursing me for expenses and paying for my professional time.
Dr. Sever and I independently reviewed the clinical report forms of all patients with a mention of infection in the report, and there were 30 INTERGEL and 20 in the Ringer's lactate group. We identified patients with possible treatment related pelvic infections using predetermined diagnostic criteria and clinical judgment, and we compared rates between groups and drew conclusions.
The diagnostic criteria for pelvic infection are well documented in our literature, and to document a postoperative pelvic infection is more subjective than it is on documented clinical findings, but we do use criteria that are available to us.
Elevation in temperature and co-elevation in pulse rate, elevated white count or a left shift with an increase in immature band forms, purulent drainage from the site, and cellulitis are the criteria that we use and will publish in our literature.
The consensus agreement of patients with possible treatment related postoperative pelvic infections, in the INTERGEL group we have three, one pelvic infection and two wound infections. In the Ringer's lactated group there were three, one wound infection -- correction -- two wound infections and one pelvic infection.
When we compare the infection rates between the study PI, the FDA auditor, and Dr. Sever and myself, you can see in this table relatively there were really no differences in the infection rates between any of the analysis and between the two groups in study.
The comparison of infection rate assessment, the difference in number of possible treated related pelvic infections between INTERGEL and the control group are not statistically significant. However, the conservative estimates for the gynecologic surgery were found, which ranged from 2.4 percent to 3.3 percent.
Febrile morbidity is a more common diagnosis that's made in our patients after surgery, and oftentimes misconstrued as infection. The phenomenon is observed in women undergoing pelvic surgery by laparotomy. Oftentimes this is due to the surgical procedure itself or accumulation of blood.
Some patients with fever may have been diagnosed with postoperative pelvic infection in the absence of sustained elevated white counts and clinical signs of infection, and these patients were probably in the febrile morbidity group.
This slide characterizes what I'm talking about and the difference. If you look at patients who have fever and elevated pulse rate in the infected patients, there is a parallel between the two. The two findings actually parallel each other throughout the course of the illness.
Where in the febrile morbidity group, there's an elevation in temperature, but the pulse rate tends to remain within the normal range.
When we looked at adverse effects not related to treatment, chicken pox is certainly not related to the surgical procedure or Ringer's lactate or INTERGEL. A head cold should not be construed as a complication of the surgical procedure, nor should cystitis or a urinary tract infection since most of these patients, I would assume all of these patients were catheterized at least once and had an ingrowing Foley. So that's the risk factor for that.
Potentially significant adverse effects that I would look for in using an agent instilled into the peritoneal cavity would be the occurrence of postoperative or interoperative pulmonary edema, which there was none. There was no electrolyte imbalances. There were no cases of anuria, and there were no cases of immunosuppression. All of these factors would be complications associated with infection.
So safety conclusions. Using accepted clinical criteria for the diagnosis of pelvic infection, we determined that the rate of treatment related pelvic infection was comparable among INTERGEL patients and controls. No other important adverse events were noted.
Concern over infection risk was not borne out in the post marketing experience, two events out of more than 35,000 units sold. The animal potentiation study in a large number of rats was negative and was not available to the GPS panel.
If you want to say that the rate model that was used was a high risk, intra abdominal sepsis model first described by Annie Honordunk (phonetic) and John Bartlett and Sherry Gorbach, and not to find an increase in that model with the INTERGEL group, I think, is really important because that is a definite significant high risk study model to be using for intra abdominal sepsis.
And with that I would conclude.
Oh, I have one more slide. The study design, execution, and analysis of the INTERGEL adhesion prevention solution provides valid, reliable data sufficient upon which to base the conclusions supported for the proposed intended use. The benefit of INTERGEL used for the label indications outweighs any probable risk and the sponsor had provided valid scientific evidence to demonstrate that the product does not pose an unreasonable risk of injury or illness.
I'd like to close with one statement in that. In the surgical procedures done in this study, they are all low risk for infection because of the nature of the surgery that was performed. The evidence provided is reasonable and sufficient to establish that the product is effective and that the nature, extent, and magnitude of the reductions observed are clinically significant.
There was no increase in infection among those with INTERGEL. The absolute number of patients who benefitted is a significant portion of the study population.
Thank you, and I will pass on.
DR. COLTON: My name is Ted Colton, and I have no vested interest for the sponsor, and I do hope that the sponsor will pay for my expenses for coming to this meeting.
I will give my credentials in just a minute. I just want to note I don't know how often statisticians are in the enviable position of being the last in a presentation, having the wind-up position at least with regard to the sponsor's presentation, but in the unenviable position unless things have changed of being what stands between a much needed break after we finish.
In any event, I want to summarize a consensus report that you all have, I believe, that I and my two colleagues who looked at the INTERGEL pivotal trial wrote after we had examined all the accompanying documentation.
There are three parts to our presentation. I want to describe briefly our credentials, again very quickly go over what we reviewed, and unlike our clinical colleagues, we identified six issues, six statistical issues. I have four clinical issues, and then each of us, each of the three of us will address each of these six specific statistical issues, and finally I will summarize our conclusions.
Next slide, please.
Okay. I'm Professor and Chair Emeritus, meaning a step-down chair -- I'm not retired though -- in the Department of Epidemiology and Biostatistics at Boston University School of Public Health.
My colleague to the right, Steve Piantadosi is Director of Biostatistics at the Comprehensive Cancer Center at Johns Hopkins School of Medicine.
And finally, last on my right is Don Rubin, who is Professor and Chair of the Department of Statistics at Harvard University.
I think you heard this before. I'll just say we reviewed what we believe are all of the relevant data of the PMA, as amended, et cetera. You've heard this before. And after we did our independent review, we convened and we jointly forged the consensus opinion that we wish to present to you.
Our opinion concludes, our report concludes that the trial provides valid scientific evidence to base conclusions regarding the effectiveness and safety of INTERGEL prior to marketing. We concur with the sponsor that the trial is well designed, and the analysis described in the PMA as amended is scientifically sound.
The analysis proceeded as described in the protocol. Additional analyses were carried out at the request of ODA, and we believe in accord with appropriate statistical practices.
The results are statistically robust and sufficient to support approval, and we did not identify any methodologic or statistical issues that were sufficiently critical or problematic that would undermine the validity of the results.
We looked at all of the methodology and data analysis. We looked at one numbers and our clinical colleagues did, and we looked at the questions that were posed by ODE on the statistical issues.
The six statistical issues that we identified and that we will discuss are the pivotal trial design, adherence to the protocol and pooling, which Dr. Piantadosi will handle. I'll say something about statistical power. Dr. Rubin will say something about incomplete ascertainment and the intention to treat analysis.
So let me turn now to Dr. Piantadosi for the first three issues.
DR. PIANTADOSI: Thank you, Dr. Colton.
I have no financial interest in the company, only an intellectual interest in the topic, and like nearly everyone else in the room, I'm not working for free.
DR. PIANTADOSI: I'm going to discuss the first statistical issue, which deals with the pivotal trial design, and I want to emphasize what you've already heard, that being the critical features that this study has incorporated to eliminate selection bias and control the precision of estimation of the treatment effect.
You've heard that the study was masked. This is unusual for a surgical trial. It's unusual for a device trial. The very size of this study and rigor with which it was done is unusual in this kind of setting.
The trial is what might be conventionally referred to as triple masked. The surgeon, the evaluator of adhesions postoperatively at the second look, and the patient herself are unaware of the treatment assignment.
The study was a multi-center study, the state of the art for providing a heterogeneous enough cohort to support the external validity of the findings. This is a strength, not a weakness of this study.
Within each center participating in the trial there was a blocked and stratified randomization which provides multiple independent and unbiased estimates of the treatment effect at each center, which can then be pooled across center under appropriate conditions that I'll discuss in a moment to provide an overall, unbiased, valid, and precise estimate of the treatment effect.
We found in our review that the study design was scientifically valid, and it meets the scientific and regulatory standards that might be applied in this setting.
The second issue that we addressed had to do with adherence to the protocol plans for the statistical analyses and concerns that the products of those analyses might be deemed post hoc or data dredging. This is not the case.
The PMA as amended presents all of the analyses that are set forth in the study protocol. The secondary endpoints about which you've heard were prospectively defined in the study protocol, and we have been able to identify no issues of multiplicity of analysis that would invalidate any of the type one error or significance levels that have been presented to you.
You've heard that the AFS is an appropriate endpoint. I think that whether we describe it as a surrogate or clinical endpoint is semantic. In this case it's clearly the appropriate one, and the data that comprise the score of the AFS were prospectively collected.
The indication that's being asked for is new, but the data set is not. It is the data set that was submitted as part of the original PMA.
There were additional analyses that were requested and performed by the sponsor, requested by the FDA and agreed upon. These analyses have been presented and discussed and labeled as such, but all of the analyses on which the claims are being made were those prospectively laid out in the protocol.
The next slide shows the completion of our conclusions regarding adherence to the protocol. There were three protocol specified cohorts for study in this trial. The first was an attempt to identify what might be called the intention to treat, that is, all patients treated on the study and categorized in the treatment group to which they were assigned. This is a respectable analytic strategy.
There was superimposed on the intention to treat cohort the additional requirement that worst score imputation be used to replace missing values for those patients who did not have second look data.
These two notions are independent and separate, that first being intention to treat and the second being a method for imputing missing data. Dr. Rubin will have more to say about that in a moment.
The second analysis that was specified in the protocol was to analyze those patients who had complete data at second look laparoscopy. This is called the evaluable patient analysis, and you've heard many of the results based on that protocol prespecified invalid analysis.
Here was a third analysis specified in the protocol. You heard that if patients are excluded based on reasons unrelated to the device that there's essential balance in the two treatment groups, making that analysis essentially consistent with the second one, and so statistics on that have not been presented.
All of those analyses, however, are prespecified, and we've reviewed the findings from all of them.
The third issue is on the next slide and deals with pooling of data. There is going to be some contention about this issue, but I want to provide you with my reasoning supporting by my colleagues about the rationale for using data from all the sites.
First of all, this was a common protocol. It was written and conducted as such. The inclusion and exclusion criteria were identical at all sites in both continents as you would expect from a multi-center study. The procedures used in the trial in evaluation were identical in the U.S. and Europe.
As I mentioned earlier, in addition, the randomization was blocked and stratified, which is really the strongest justification for pooling because the estimates of treatment effect from each of the centers regardless of where they are from provide unbiased estimates of the treatment effect.
The basic issue in deciding whether or not the data from the two continents should be pooled has to do with evidence of a common treatment effect. We like to pool when there's evidence that the treatment effect might be the same across the pooling units, and the protocol attempted to lay out some criteria under which we could reassure ourselves that there was, in fact, a common treatment effect across continents.
Those criteria are listed on the next slide, and quite honestly if I had written this section of the protocol, I don't think I would have written it this way, but nevertheless, I think these criteria are substantially in the ballpark and appropriate for the purpose.
The first criteria and probably the most meaningful one is that there should be no significant interaction between the treatment effect and the continent, indicating that there is a common treatment effect or no evidence that there is a difference in the treatment effect.
Secondly, we would be reassured if there are similar statistically as well as qualitatively demographic in pretreatment or baseline variables. The protocol states that lack of similarity might be a basis, but it's not required to be a basis for precluding the combining of the data from the two comments.
And then thirdly, the similarity on second look adhesion scores was just to be something that would provide reassurance that combining the data from the comments would be appropriate.
In fact, if we look at these criteria, there is no interaction between continent and treatment once you account for adhesiolysis category. The failure to account for adhesiolysis category is the sources of the differences in interpretation on this issue.
If you fail to account for it, it appears that there is quantitative interaction between treatment and continent, and I'm choosing my words very carefully here, quantitative meaning that the treatment effect is in the same direction on both continents, but has different magnitude.
However, the fact that the interaction is substantially in the same direction after accounting for adhesiolysis means that there is sense in talking about an overall treatment effect even though there are slightly different magnitudes on the two continents.
More appropriately, however, the proper thing to do is to account for the effect of adhesiolysis, and when you do that, you see that the apparent interaction between treatment and continent is, in fact, not an interaction at all. It is the effect of the uncontrolled predictor variable, adhesiolysis, which is different in the two continents as a matter of patient characteristics and medical care practices.
So, in fact, properly accounting for that factor shows that the treatment effect is common and satisfies the first pooling requirement.
Next, there are some differences in baseline variables, as you might expect, given the different demographics of the two continents. However, it's our opinion that these are inconsequential and are not an impediment to combining the data because none of these factors modify the effect of treatment, as I've just indicated in the reasoning about continent.
And in the case of differences in baseline factors, there are sound statistical procedures available for accounting for those differences and preserving lack of bias and precision in the overall estimate of the treatment effect.
And then finally, the data support that second look adhesion scores between the U.S. and Europe are similar, again, after accounting for the predictor factor, adhesiolysis. Thus, we have satisfied all of the three requirements laid out in the protocol for pooling of the data.
DR. COLTON: Let me deal with the issue number one -- next slide, please -- which is statistical power.
First of all, statistical power is a probability statement about a hypothetical treatment effect. In the pivotal trial, the observation of statistically significant differences between treatment and control for any endpoint is unaffected by the original power calculation in the study protocol.
The sample size in the pivotal trial was determined based on differences in the primary outcome, mAFS scores, as observed in a pilot study. The original power calculations were based on standard methods, and this key information gleaned from the pilot study.
Had there been a more precise estimate of the true standard deviation of the primary outcome, one probably would have planned a smaller trial. When the trial was complete, the difference in scores, in mAFS scores in the study, that is, the evaluable population, was smaller, namely, a one unit change than that anticipated when the study was designed based on the pilot study results, a 2.1 unit change.
However, it's important to note the standard deviation was also smaller, 1.5 for INTERGEL and 2.2 for control compared to the standard deviation of five, which was the basis for the sample size calculation.
Hence, in the actual pivotal trial there is greater precision of estimation than had been originally anticipated.
And another important point we feel is that there's no rationale to question the validity of a trial because the treatment effect and/or the variance observed are smaller than that hypothesized before the trial began. This trial assessed the incremental benefit provided by an adjunctive treatment and found that it was statistically significant.
It would have been preferable, of course, to have had a more precise measurement of the outcome of interest with less variation than a large effect with more variation.
Now, comments on the medical importance of this statistically significant results is clearly in the bailiwick of the judgment of clinical experts, and you're already heard the consensus panel on the clinical significance of the statistically significant findings.
DR. RUBIN: I'm Donald Rubin, and I, too, have no financial interest in LifeCore, although I, too, am assured that I'm not working for free.
I'm going to be talking about two closely related issues. They're incomplete ascertainment and ITT analysis, intention to treat analysis.
With respect to incomplete ascertainment, 16 of the 281 randomized and treated patients did not complete the study. That's less than six percent did not have second look data.
You already saw this in Dr. Johns' presentation and saw that most of the reasons for not having second look data are unrelated to the treatment.
The findings reported by the sponsor are based on the evaluable population, which includes the 265 patients, and this is not uncommon in such trials.
Now, in support of these evaluable population results which you have seen, these are analyses based on these 265 with second look data, there were four separate sensitivity analyses that were performed and submitted by the sponsor in support of the analysis of the evaluable population.
And these sensitivity analyses were based on various imputation methods for those people without second look data. The robustness and propriety of the evaluable population results as submitted by the sponsor were supported by these four sensitivity analyses.
Moreover, the propriety and robustness of the evaluable population analyses were supported by independent intention to treat analyses that we created based on scientifically imputed data.
Now, I'd like to talk about intention to treat analysis, and I'm going to spend a few slides on this because you'll see this phrase "intention to treat analysis" repeatedly, I believe, in the subsequent FDA presentation, and I think it requires some clarification.
The ITT population includes all those randomized and treated, including the 16 patients without second look data. So the intention to treat population is size 281.
Now, I think we're all sympathetic to the use of this population and understand the reasons for its use in many trials, and including this trial.
However, it is impossible to conduct an ITT analysis without imputation. In this case there are 16 people without second look data, and if you can conduct an analysis on all 281, you have to do some kind of imputation, either explicit or implicit, of this missing second look data for these 16.
What the FDA did and what the sponsor agreed to do as a sensitivity analysis was input the worst possible score. So it's worst possible score imputation. This is not an ITT analysis, but it's an analysis on the ITT population after imputing the worst possible score to that population.
But you'll see the phrase repeatedly "ITT analysis," and I want to be sure that we're clear about that.
It is our view that this imputation method is neither mainstream nor scientifically defensible in this trial. There are trials where it may be defensible, and perhaps I'll say something about that later during questions and answers, but we do not believe it's defensible nor scientifically reasonable in this trial.
As an example of that, there is a woman who refused second look data, second look examination because she was pregnant, and yet she was imputed to have the worst possible adhesion score on all outcomes. And I don't believe that's considered scientifically reasonable at all.
Now, we also connected our own independent analysis that I designed of the ITT population based on what I regard as scientific imputations. In this technology I was blinded to the results. I had no idea which treatment group was being imputed. I had no idea what effect it would have on the final analyses. There were no outcome, no second look data available at all in doing that, and I can describe that more in detail later if anybody is particularly interested in that.
But the important thing is that we had no idea, any of us, in what effect it would have on the resultant analysis.
Also, the method was entirely nonparametric. No model was used. It was based on matching the important baseline variables. They were the center, the randomization center, and baseline adhesion scores, and that's how the imputations were done.
It was independent in treatment group and continent. In other words, each treatment group and continent, those four categories, Treatment A, B, continent, Europe-U.S., were completely independently imputed. So there was no contamination of continent results or treatment results across treatment groups or across continents.
Moreover, the imputation incorporates uncertainty of the imputed values in order to allow for valid inferences. In other words, more than one imputed value was created, and the analyses of these multiply imputed data sets was combined according to standard combining rules.
The results of this independent analysis were as follows. INTERGEL was superior to lactated Ringer's solution for all subjects, regardless of the baseline adhesion rate. The relative risk, again, was greater than five, very significant results in support of the evaluable population results.
The superiority of INTERGEL over lactated Ringers was even true in the U.S. alone and even in the U.S. alone for those patients with no baseline adhesiolysis. So that the superiority of INTERGEL over lactated Ringers was clear even if you exclude the European centers and even if you exclude in the U.S. those with baseline adhesiolysis.
Moreover, the variability of the outcome, the variability of the primary outcomes was typically substantially less with INTERGEL than with control. That is, there was a difference in level of effect that was in favor of INTERGEL, and moreover, the variability of the second look variables was less, substantially less in many cases when using INTERGEL than when using control, indicating that surgically clinically you'd have more predictable, good outcomes with INTERGEL than with control.
In conclusion of this part on intention to treat analysis, these results based on the scientific imputed population strongly support the findings that have been reported on the evaluable population.
DR. COLTON: We have two more slides and which I will do.
First, where do we stand? Now, here's our conclusions. First, the pivotal trial results are appropriate and reliable. That is, we find them statistically valid. A statistically significant difference in AFS scores has been demonstrated with a high degree of certainty, very respectable p value, for INTERGEL conferred a fivefold lower risk of moderate to severe adhesions, and the results are supported by statistically significant secondary variables.
These results are consistent -- you've heard this already -- with pilot study and with animal studies that have been conducted. They are supported by four separate imputation analyses that the sponsor conducted. They are supported by additional independently conducted intention to treat analyses based upon scientifically imputed second look data.
And finally, last slide, please.
We find the data analyses presented by the sponsor are robust and they're adequate to support conclusions regarding clinical significance by appropriately qualified clinical experts.
One final word, also my personal view is I really feel to me at my age and stage of career development this has been a great learning experience for me, and one of the things, it's been really a wonderful experience to work with colleagues as talented and creative as Dr. Piantadosi and Dr. Rubin, and really apart from getting compensated for my time, I've really enjoyed working with these two colleagues on this.
I think this concludes the LifeCore presentation.
CHAIRMAN RAMSEY: Thanks.
Just a comment that you may not be working for free, but I'm nearly working for free.
DR. COLTON: We know.
CHAIRMAN RAMSEY: We are close to break. I do want to allow a couple of moments for the panel to ask questions of those who did presentations for LifeCore this morning.
DR. D'AGOSTINO: Yes. I enjoyed the presentations. Quite impressive.
When I look at the data, and tell me why I'm wrong, you keep saying fivefold increase and what have you, but it's like three versus 17. You did imputation for the missing data, but what if the 17 was 14? What would the significance be? How stable are even the evaluable patients? How stable are these analyses?
Do you follow my questions?
DR. RUBIN: I sort of do. We did not do an analysis where we actually changed the observed values of the data to see what would happen if actual observed second look data were something else. We always left those the same.
But when we did both the sensitivity analysis that were performed and submitted by the sponsor and our analyses where we did the scientific imputations, those imputations were changed. They were basically, including the sponsor's sensitivity analysis, they were all over the map to cover what we regard as anything reasonable that could --
DR. D'AGOSTINO: And was --
DR. RUBIN: -- the extremes.
DR. D'AGOSTINO: -- the test an exact test or was it the asymptotic version of the test? Because there must be zeros all over the place in these tables that are being pooled.
DR. RUBIN: That's correct, but, in fact, the tests were not done by a natural permutation test or randomization test. But if you look at the occurrence of zeros, the data are more benign than a binomial with the probability of .8.
And so these are more robust than the usual Wilcoxon tests, Mann-Whitney-Hume tests, T tests, would be completely supported by the underlying randomization based analysis that would be permutation tests.
DR. D'AGOSTINO: Jim, what I'm trying to get at is just how robust are even the evaluable. I mean there's five and so forth. I'm jumping into the middle.
CHAIRMAN RAMSEY: Yes.
DR. SHIRK: Dr. Shirk.
I've got a question for the statisticians. The initial PMA obviously was -- the statistics for the initial PMA was set up on an intent to treat, and I think that that would have -- if you look at the initial PMA, there would have been 303 or 304 patients that were evaluable.
There was obviously a calculated dropout rate in this so that the statistical model that they set up initially obviously was based on a 20 percent dropout rate for the INTERGEL group and also a ten percent, I think, dropout rate for the control group.
So that, you know, how does this analysis and the endpoints and your analysis of the endpoints change between the two things since obviously the endpoint agreed on was based on a much higher dropout rate and obviously applying worst case scenario to those patients who dropped out?
DR. BECKER: Dr. Piantadosi.
DR. PIANTADOSI: Yes. Thanks.
If I understand your question, the short answer is that only good things happen when the dropout rate and missing data rate is lower than that which you planned for in the trial. So ordinarily I would not expect that the findings would be incorrect, so to speak, because I had less of a dropout rate than what I planned for in the protocol.
The usual effect of dropouts, treatment crossovers and other kinds of imperfections in the data is to create a situation where the treatment effect is smaller and closer to the null value than it might otherwise be.
The additional precision, the addition observation generated by observations that you have but didn't expect to have improve the inference. So I think the fact that the trial as conducted did not quite measure up to the way that it was planned in the sense that the treatment effect was smaller, the variability in the treatment effect was smaller, and the compliance to the plans of the protocol was higher than that originally planned is actually a strength of the trial, not a weakness.
CHAIRMAN RAMSEY: Did you have a question?
DR. KIM THORNTON: I have one question.
CHAIRMAN RAMSEY: Okay. Go ahead.
DR. KIM THORNTON: This is for Dr. Piantadosi.
You had mentioned that you might have developed criteria for pooling of data that would have been different than what was included in this protocol. What might you have done differently?
DR. PIANTADOSI: Well, I think that the criteria laid out sort of captured the general sense of what provides reassurance that data are poolable, but I probably would have written the criteria a little more sharply and a little more clearly.
The main issue for me is whether or not you can generate evidence for or against a common treatment effect across the units that you want to pool. Now, what's very common and known when we design multi-center clinical trials is that we're going to see a fair amount of heterogeneity among the treatment centers that participate in the trial if we look at each center at the treatment effect or treatment difference that's provided by each center.
And sometimes that heterogeneity, if you look at it within each center, can be rather startling and bewildering, but because each of the estimates is guaranteed by the procedures, the randomization and the methodology of the trial to be free of bias, then we can pool those and average them and provide an overall treatment effect.
So ordinarily when we look around among centers to see whether there's heterogeneity in the treatment effect, sometimes we see it; sometimes we don't. And the real issue is: is there heterogeneity of such a large degree that you would disbelieve that you're really seeing the same treatment effect in the U.S. and Europe or in Center A and Center B.
And I think that can be stated more clearly than it was in the protocol, but I'm not trying to object to the way the protocol was written or the way those analyses were carried out. I'm simply telling the panel that when one evaluates the use of those criteria in the protocol, it's important to consider this predictor variable, adhesiolysis, because failure to account for it properly will make it look as though there's an interaction when, in fact, there is not.
CHAIRMAN RAMSEY: I'm going to take a chair privilege to cut us off. We are going to have a chance to ask the panel questions in the afternoon, ask LifeCore questions, and I'd like us to take a ten minute biology break here, and then we'll move on.
(Whereupon, the foregoing matter went off the record at 10:21 a.m. and went back on the record at 10:44 a.m.)
CHAIRMAN RAMSEY: We're going to get started in just a minute. The FDA person is coming back to start with their presentation.
I realize at the start that I didn't state for you my background. So I thought I'd just take 30 seconds to do that.
My background is that I'm, in addition to being an Associate Professor at Fred Hutchinson Cancer Center in the University of Washington, my background is I'm a general internist. I practice at the University of Washington, and I have expertise in health economics and evidence based medicine. So that should make everyone hate me.
CHAIRMAN RAMSEY: No, but those are my areas of expertise, and I just wanted to make sure everyone knew my background.
It should be just a moment and then -- okay. Here she comes. So we're now going to start with the FDA presentation.
DR. KRAUS: I'm six, three. I'm used to lifting mics.
I'd like to start this morning by saying good morning and welcome and thank you all for coming. I'd like to thank the Chairman and members of the panel. Mr. Weinstein has done a great job of organizing everything.
Representatives of LifeCore Biomedical, Dr. Feigal, members of FDA, and especially members of the public for showing such an interest in this regulatory process.
My name is David Kraus, and I am a reviewer in the Office of Device Evaluation, Division of General Restorative and Neurological Devices at the Center for Devices and Radiological Health. I'm a cell biologist by training and lead reviewer for the INTERGEL adhesion prevention solution PMA, which is being discussed here today.
Next slide, please.
I'd like to introduce the folks who did the reviewing and helped me immensely and actually did most of the work. Dr. Lisa Harvey reviewed the animal infection data. Dr. Roxy Horbowyj reviewed the clinical data as the lead clinical reviewer. Dr. David Kaplan was our manufacturing reviewer. Dr. Richard Kotz was our statistics reviewer.
I, as well as being the lead reviewer, did the preclinical toxicology, and Dr. Diane Mitchell was our OB-GYN clinical consultant.
Next slide, please.
Today you will be hearing a presentation on how the FDA review team interprets the data presented in the INTERGEL adhesion prevention solution PMA and the amendments which followed. I will be giving you a brief overview and an introduction.
Dr. Horbowyj will give a review of the pilot and clinical studies performed in support of approval of the PMA, and then she will be followed by Dr. Kotz, who will discuss the agency statistical analysis of the clinical data presented in the PMA.
I'd like to emphasize that the agency's presentation will focus solely on the clinical data and the statistical analyses that are presented in the PMA. At this time we do not feel that there are any preclinical issues that need to be discussed any further at this meeting.
Next slide, please.
This slide shows the indication for use proposed by LifeCore Biomedical in the original PMA submission, and I'm not going to read it. It's in your handouts.
Slide 5, please.
The following is the indication for use as proposed in the subsequent amendment, which followed the not approvable recommendation from the panel at the January 12, 2000 General and Plastic Surgery Devices Panel meeting. This was submitted in the amendment which was discussed previously by the sponsor, which was presented to the agency in June of 2000.
Again, it's been read, and I'm not going to read it again.
Can I have the next slide, please?
As you have already listened to the sponsor's presentation and interpretation of the study results and you're now prepared to hear the FDA presentation, please keep the following question in mind. During this afternoon's discussion period, you'll be asked to comment on this question.
The question is: does the PMA as actually was read by Dr. Ramsey this morning -- and I won't read it again.
I'd like to now introduce our clinical reviewer, Dr. Roxy Horbowyj.
DR. HORBOWYJ: Good morning. My name is Roxy Horbowyj. I'm a general and critical care surgeon and a clinical reviewer for this submission.
I will present the FDA clinical perspective of INTERGEL use in patients undergoing clean, open, gynecologic procedures for pain, infertility, or irregular bleeding.
Next slide, please.
I will briefly present background on adhesions and adhesion evaluation, as well as a summary of private study data and highlights of a pivotal study which apply to the currently proposed indications for use.
Adhesions -- next slide, please -- form as a protective response to localize a peritoneal insult. Most commonly adhesions are due to trauma, foreign bodies, or infection. Adhesions may cause or minimize morbidity.
For example, adhesions may prevent volvulus or contain a bowel leak. Adhesions, however, may also cause pain, bowel obstruction, and female infertility.
Adhesions may be characterized by their rotation, whether the adhesion occurs at a surgical site known as the direct adhesion or a remote site known as an indirect adhesion; by occurrence, that is, whether the adhesion is new, known as de novo, or previously lysed, that is, reformed.
Adhesions may also be characterized by their extent over which they cover in a particular anatomic site, and by severity, how difficult the adhesion is to lyse, how vascularized the adhesion is.
At this time there is no consensus as to how to best predict which peritoneal cavity adhesions or which adhesion characteristics specifically will cause or minimize morbidity.
Next slide, please.
Adhesion evaluation consensus on how best to correlate adhesion characteristics with clinical outcome has not been reached. To evaluate adhesions, the American Fertility Society has published a method of adnexal evaluation which included an adhesion scoring method called the American fertility society score, the AFS score.
For the INTERGEL pivotal study, the sponsor developed several scores to evaluate adhesions based on the AFS score, two scores specifically were the modified AFS score and a retrospected AFS score.
The AFS in INTERGEL clinical study scores differ in the number of anatomic sites evaluated, method of anatomic site evaluation, and method of determining the final score per patient.
Now I will go over the scores.
Next slide, please.
The AFS score was developed for evaluating adnexal adhesions in an effort to address the need for a standardized classification scheme for adhesions expected to be associated with infertility. In determining an AFS score for a patient, four anatomic sites are evaluated per patient: the right ovary, the right tube, the left ovary and the left tube.
The scores per side are summed. The final AFS score per patient is the score of the side with the lowest summed score. The higher score representing the side with the higher adhesion burden is dropped.
Next slide, please.
An AFS score, as published by the American Fertility Society, is based on the incident, extent, and severity of an adhesion at an anatomic site. A score of zero, one, two, four, eight or 16 is assigned to a tube or ovary depending on the severity of the adhesion, which may be mild or severe and the extent of the adhesion, which may be localized, moderate, or extensive.
The final AFS score range is zero to 32 as the tube and ovary per side are summed, and the higher score is dropped.
Next slide, please.
The published literature and its interpretation of the American Fertility Society scores are often limited by small sample sizes that are reported and by the use of variations of the published score. For example, different anatomic sites may be evaluated. Some studies report use of the score applied only to the fallopian tube, and sometimes different score assignments are used, a range of zero to 20 instead of zero to 32. So a direct comparison is difficult.
Interobserver reproducibility at the level of 0.7 has been reported as observed in less than one third of surgeon pairs studied, and the published reference is on the slide there.
The modified American Fertility Society score developed by the sponsor to evaluate adhesions throughout the peritoneal cavity used 24 prespecified anatomic sites. At each site the incidence, extent, and severity are evaluated according to the published AFS scoring scheme, except in the case of four sites: the small bowel, the omentum and the left and right colon, which were to be assigned an extent score of moderate for any adhesion noted at these sites.
The final modified AFS score, mAFS score, per patient is the average of 24 site scores. No scores are dropped, and then AFS score range is zero to 16.
When attempting to interpret a single mAF score per patient, it's notable that a single mAF score per patient in a graph of mAFS versus number of sites with adhesion may refer to a broad range of adhesions.
I'm sorry I don't have a pointer, but I think it may be useful to -- in looking at this chart, for example, if a patient has one adhesion, our colors have been changed here. So I will go through this more slowly.
If a patient has mAFS score, which is on the vertical axis, of one, and if you follow across, you can see that you can have a range of the number of adhesions with this one score.
So the consequences and potentials of evaluating just a single score alone may be varied because of the associated, potentially different number of adhesions that may be associated with a single score.
While I have this up, let me also -- you've now heard extensively about the study design very well put forth by the sponsor and their outcomes as well. But in the pivotal study, patients who were enrolled and inclusion/exclusion criteria was that patients were to have fewer than 12 adhesions.
So in this area of potential scores and adhesions that is possible with the scoring system, this is the area that was studied, and as the sponsor has said, it was less than one -- there was less than one change in number of sites of adhesions comparing INTERGEL and control and a change in approximately one in mAFS score.
So essentially the changes that were experienced in the study were about the size of one of these blocks.
Next slide, please.
Going back to the retrospective American Fertility score now, the mAFS score was designed prospectively for this study, but retrospective to the pivotal study to provide a score that addresses adnexal adhesions alone, in other words, an analogue to the AFS score, the sponsor calculated adhesion scores for each ovary, tube from the mAFS scores.
This required the use of ten mAFS score sites as in the mAFS score system. Each ovary was evaluated as three anatomic sites, and each tube was evaluated at two anatomic sites.
The extent per ovary or tube was calculated using the average of the extent to numeric values from all sites making up a tube or ovary. The severity per tube or ovary was assigned as the maximum severity of contributing sites.
The calculated extent and the assigned severity scores determined the overall tube or ovary adhesion score. The ovary and tube scores per side were summed, and the lower sum score became the rAFS score as was done with the published score.
The final retrospective AFS score range then was zero to 32, as is found in the published score.
Next slide, please.
The retrospective AFS score was also stratified in several ways, which are listed here. The retrospective AFS score had these strata as the basis for the current claims.
Next slide, please.
Overall, considering the modified AFS score and the retrospective AFS score, there are several limitations to score interpretation. For example, as I've tried to demonstrate with the graph, the mAFS score overlaps confounds interpretation of a single score alone. The correlation of clinical outcome with an mAFS score or change in the mAFS score is not known. The correlation between the standard or force site AFS score and the retrospective ten site AFS score is not know, and the correlation of clinical outcome with any AFS score stratification has not been established.
And now I will address the INTERGEL clinical studies.
Next slide, please.
INTERGEL, as you have heard, is a 0.5 percent ferric hyaluronate gel. It's an aqueous solution of sodium hyaluronate ionically cross-linked with ferric chloride.
Next slide, please.
The objectives of the projects that if we were to assess the methodology of use and to make a preliminary assessment of the safety of 300 cc's of INTERGEL compared to 300 cc's of lactated Ringer's solution in patients undergoing peritoneal cavity surgery by laparotomy, and with second looks laparoscopy.
Objectives of the pivotal study were to assess the safety and effectiveness of INTERGEL compared to lactated Ringer's solution in the same volumes and improve any or reducing adhesions in patients undergoing peritoneal cavity surgery.
The pivotal study design was based on product study outcome.
Next slide, please.
The pilot study was a prospective, randomized, single center, single investigator study of 21 patients undergoing laparotomy for infertility with six to 12 week follow-up for second look laparoscopy.
Three hundred cc's of INTERGEL or lactated Ringers were left in the peritoneal cavity at the end of surgery. The adhesion incidence, extent and severity were evaluated, evaluated 18 prespecified anatomic sites at baseline and at second look, and a modified AFS score was calculated retrospectively.
At second look differences between INTERGEL and control were 4.65 for the mean AFS score, 4.91 for the mean adhesion incidence.
At second look adjusted for baseline, differences between INTERGEL and control were 4.12 mean AFS score and 4.13 mean adhesion incidence. The difference between INTERGEL and control in the mean modified AFS score at second look adjusted for baseline, which was 4.12, was the basis for the pivotal study design.
Infection in the pilot study occurred in one out of 11 INTERGEL patients and zero out of ten control patients.
Next slide, please.
The pivotal study was a prospective, multi-center study to be undertaken at 12 U.S. and six European centers. The study was to enroll otherwise healthy 18 to 45 year old females with pain, bleeding, or infertility and adhesions at up to 11 of 24 prespecified anatomic sites.
Randomization to INTERGEL or control occurred preoperatively, that is, before all inclusion and exclusion criteria were evaluated.
The dose of 300 cc's per patient was not adjusted for patient weight. Follow-up was to occur at seven days postop. for laboratory evaluations and at six to 12 weeks for second look laparoscopy. A third party masked device application or evaluation was proposed.
Next slide, please.
Pivotal study endpoints were for safety and for effectiveness. Safety was evaluated by adverse events. Effectiveness was evaluated by primary and secondary endpoints as the sponsor has reported also.
The primary endpoint was the modified American fertility score based on 24 sites. Secondary endpoints included the proportion of adhesions: adhesion incidence, yes or no; adhesion extent, which was evaluated at zero, one, two, or three; and adhesion severity, which was evaluated as zero, one or three.
Next slide, please.
From pilot study data, the evaluable pivotal study cohort was expected to demonstrate a difference between INTERGEL control group mean AFS score of 4.1. This was the second look mean AFS -- mean mAFS score adjusted for baseline. Assumptions in the design of the pivotal study were that loss to follow-up would be 20 percent in the INTERGEL group and ten percent in the control group.
The sponsor also proposed a loss to follow-up patients would be assigned a mAFS score of 16, and it was planned that an intent to treat analysis was being performed with these assumptions.
Therefore, based on these assumptions, differences between INTERGEL and control mean AFS score, mean mAFS score adjusted for baseline decreased from 4.1 to 2.1.
The sample size to detect a difference of 2.1 in the mean mAFS score was then 180 patients. This was with a standard deviation of 5.0 for both groups and assuming the ITT analysis, a power of 80 percent and a significance level of 0.05.
Next slide, please.
The study was designed to evaluate 180 patients. Two hundred and eighty-one patients, however, were enrolled, 200 in the U.S. and 81 in Europe. Two hundred and sixty-five patients were evaluable, 188 in the U.S. and 77 in Europe.
A comparable number of patients enrolled into the INTERGEL and control groups in each continent. A comparable loss to follow-up occurred in each continent. However, only about half of the loss to follow-up that was expected and accounted for in study design occurred. The actual overall evaluable population was 85 patients larger than the prospectively calculated sample size.
Hence, the calculated sample size of N equals 180 was higher than necessary to detect a difference of 2.1, the mean mAFS score, with a standard deviation of 5.0 for both groups, which had been reduced from 4.1 by adjustment for expected loss to follow-up and planned intent to treat analysis.
Next slide, please.
It is notable that at baseline INTERGEL and control cohorts were clinically comparable within each continent. Comparing U.S. and Europe, cohorts were clinically comparable for distribution of age, weight, inclusion and exclusion criteria.
Comparing the U.s. and Europe, however, the cohorts were not clinically comparable for distribution of race, baseline adhesion incidence, mAFS score, or AFS scores. Procedure type also varied as the sponsor has demonstrated to you today.
Therefore, the U.S. and European cohorts were not considered combinable clinically or by prospective combinability criteria, which FDA statistician Richard Kotz will discuss.
Next slide, please.
Specifically differences in race between the U.S. and Europe were demonstrated here on this slide. Approximately 40 to 50 percent of the U.S. population was Caucasian, while approximately 80 to 90 percent of the European population was Caucasian.
Next slide, please.
This slide demonstrates differences between the European and the U.S. populations for mean adhesion evaluation at baseline. The incidence of adhesions in the U.S. patient cohorts was 2.49 and 2.27 for INTERGEL and control, respectively, while the incidence of adhesions in Europe was 6 and 6.4 for INTERGEL and control, respectively.
This similar trend and difference was noted for other adhesion evaluation parameters as well.
Next slide, please.
There were also differences in the number of percentages of patients who underwent adhesiolysis in Europe compared to the U.S. Approximately 40 percent of patients in the U.S. underwent adhesiolysis where up to 78 percent of patients in Europe underwent adhesiolysis.
As a result, 60 percent or so patients in the U.S. had non-adhesiolysis associated procedures; whereas in Europe only 22 to 30 percent of patients had non-adhesiolysis associated procedures.
The current indications for use that the sponsor has proposed is that the INTERGEL solution is a single use intraperitoneal instillate indicated to reduce the likelihood of developing moderate or severe postoperative adnexal adhesions in patients undergoing adhesiolysis or myomectomy during conservative gynecologic pelvic surgery by laparotomy, when used as an adjunct to good surgical technique.
INTERGEL solution was also shown to reduce adhesion formation to sites in addition to the adnexal and adhesion formation at surgical sites including the anterior abdominal wall.
I will go over the pivotal study data that is suggested to support these claims in terms of the retrospective scores and stratifications as provided by the sponsor and in terms of the combined U.S. and European evaluable patient population as this is the basis in which the sponsor proposes these claims.
Next slide, please.
As to the aspect of the claim referring to the likelihood of developing moderate or severe postoperative adnexal adhesions, that is, in the strata retrospective AFS score equal to 11 to 32, it is notable that at baseline the number of patients with moderate to severe adhesions was nine out of 131 patients in the INTERGEL group and 17 out of 134 patients in control. This is a difference of eight patients 5.8 percent. It's shown here on the first slide, first line.
For these patients with moderate to severe adhesions at baseline, at second look there were nine fewer patients with moderate to severe postoperative adhesions in the INTERGEL group and ten fewer patients in the control group, nine and ten.
This is a comparable number of patients, fewer patients in the moderate to severe adhesion group at second look.
The sponsor notes that this is 100 percent reduction in INTERGEL patients from nine to zero and a 59 percent reduction in control patients.
However, as the sample size of this subgroup of patients with moderate to severe adhesions is small, nine and 17 out of 265 total, it is not possible to know if the effect of the INTERGEL group is limited by the number of patients in the subgroup by device effect, by chance, or by other factors, such as surgical technique.
The sponsor also notes that the difference in the number and percentage of patients with moderate or severe adhesions is three out of 131 compared to 17 out of 134, a difference of 14 patients, which is 10.3.
This difference, however, does not account for the differences between INTERGEL control at baseline. Accounting for baseline, the difference between INTERGEL and control in the number of patients with moderate or severe postoperative adhesions is six, or 4.5 percent.
Next slide, please.
Referring to patients undergoing adhesiolysis or myomectomy during conservative gynecologic pelvic surgery by laparotomy, for this analysis the retrospective AFS score was stratified into two groups, zero to ten and 11 to 32.
The shift analysis was performed for procedure subgroups, myomectomy, no myomectomy, adhesiolysis, non-adhesiolysis, tubal procedures, ovarian procedures and endometrial ablation.
Nominal statistical significance of p less than 0.05 is reported for the myomectomy and adhesiolysis subgroups only. It is notable, however, that most patients were part of more than one subgroup.
Next slide, please.
As to the aspect of the claim referring to adhesion reformation to sites in addition to the adnexa on the prospective scale, in terms of incidence the difference between INTERGEL and control is 0.94. This difference is less than one adhesion.
In terms of proportion of sites with reformed adhesions, the difference between INTERGEL control is 0.2, which the sponsor notes is 31 percent reduction in reformed adhesions.
As to the extent and severity of reformed adhesions, the difference between INTERGEL and control is 0.5 and 0.53, respectively. This difference in extent and severity represents less than one category of change for extent and severity as these parameters were graded in the trial.
Next slide, please.
As to the aspect of the claim referring to adhesion formation at surgical sites on the prospective scales, in terms of incidence the difference between INTERGEL and control is 0.69 adhesions. This difference represents less than one adhesion per patient.
In terms of proportion of sites with reformed adhesions, the difference between INTERGEL and control is 0.11, which the sponsor notes is a 23 percent reduction in reformed adhesions.
As to the extent and severity of reformed adhesions, the difference between INTERGEL and control is 0.33 and 0.36, respectively. This difference of 0.33 in extent and 0.36 in severity represents a less than one category change for extent and for severity as these parameters were evaluated in this trial.
Next slide, please.
A notable adverse event is infection, and infection is notable here as infection is known to cause adhesions. The study population in this trial included only clean, nonmalignant cases in 18 to 45 year old, otherwise healthy and immune competent patients.
Patients who are at lowest risk for infection as clean, contaminated, contaminated and dirty cases were excluded interoperatively, and no malignant cases were enrolled.
Please note that the numbers that I'm reporting and have reported are as presented to FDA in our submissions.
The incidence of infection was reported to be 7.0 percent in INTERGEL treated patients and 2.9 percent for control overall. In terms of possibly related infection as per investigator and the sponsor's independent assessors, the incident of infection possibly related to the device used was 4.2 percent in INTERGEL treated patients and 2.2 percent in control, a difference of two.
These are all listed on this slide, but I'm not really sure how well it's projecting for everyone to see.
So assessed by the independent assessor and investigator, this is the overall infection rates of 4.2, six patients, and three patients for control overall.
The incidence of infection possibly related to the device use in U.S. INTERGEL patients was 4.9 percent, five patients, compared to two percent, two patients, in U.S. in control, a difference of 2.9 percent.
I have a listing of the patients and will gladly read off the diagnoses that were assigned and considered to be related to device use by the investigator and independent assessor, which I believe was Dr. Sever.
Continuing on this slide, the European incidence of infection was the same for INTERGEL and control cohorts.
Next slide, please.
In summary, device use was studied in 18.45 year old women with low baseline adhesion burden and otherwise good health undergoing clean, noncancer gynecologic procedures. Baseline evaluation differences between continent and treatment groups are greater than differences within a continent per treatment group for variables such as race, adhesion evaluation, and procedure type.
Revised indications for use are based on evaluable patients from the U.S. and Europe, the U.S. constituting or contributing 188 patients, Europe contributing 81.
The revised indications for use claims are based on binary stratification of retrospective AFS scores and shift analysis of seven procedure subgroups. The nominal statistical significance, p less than 0.5, noted for the myomectomy and adhesiolysis subgroups only.
Next slide, please.
Moderate to severe adhesions with respect to this, nine fewer INTERGEL and ten fewer control patients with moderate to severe adhesions at baseline had moderate to severe adhesions at second look.
Accounting for baseline, six, or 4.5 percent fewer INTERGEL patients, had moderate to severe adhesions at second look compared to control.
Reformed and surgical site adhesions at second look from this aspect, the incidence demonstrated a difference between INTERGEL and control that is less than one occasion, and from the aspect of extent and severity, the differences between INTERGEL and control were less than one category of change.
As far as infection rate, as possibly related to device use per investigator and independent assessor, 4.9 percent U.S. INTERGEL treated patients compared to 2.0 percent U.S. control patients were rated this way, and 2.5 percent of patients in the European cohorts both in INTERGEL and control were assessed to be possibly related to device use per the investigator and independent assessor.
Next slide, please.
I will now present our statistician, Richard Kotz, to present the statistical perspective.
MR. KOTZ: Thank you, Dr. Horbowyj.
I'm Richard Kotz. I'm a statistician at the FDA and will be presenting the statistical review of the INTERGEL adhesion barrier. I have been statistical reviewer for this product since the development of the pivotal study protocol.
I will first present the sponsor's study protocol and the results of their study, specifically the proposed sample size and analysis plan specified in the protocol. I will then show that it's inappropriate to pool the U.S. and European subjects, and that there is no difference between INTERGEL and the control when data is analyzed in the manner described in the protocol.
In the second part of my talk, I will discuss the sponsor's revised claim and show that the secondary endpoints, reformed and surgical site adhesions, and the retrospective defined endpoint AFS score do not demonstrate the superiority of INTERGEL over the controls.
The sample size was based on results from the sponsor's 20 subject pilot study. They observed a difference in modified AFS score of four between the two treatments. INTERGEL had a modified AFS score of 1.7 and a standard deviation of 1.4, and the control had a score of 5.7 with a standard deviation of 2.8. These were at second look.
This is the designated primary endpoint, modified AFS score as the designated primary endpoint in this study.
Much focus was also placed on the secondary endpoint, incidence of adhesions in which there was a difference of 4.9 at second look, 6.1 for INTERGEL and 11.0 for the control.
The sponsor expected a loss to follow-up of 20 percent for INTERGEL and ten percent for the control groups. Though we generally require an intent to treat study design for pivotal studies, the sponsor chose to assign the worst modified AFS score of 16 to those patients.
Furthermore, it should be noted that it is not the worst case analysis. The worst case analysis involved treating all of the INTERGEL loss to follow-up patients as failures and the control loss to follow-up patients as successes.
Note that by the design of the study with an unequal rate of patients lost to follow-up and assigning them the worst score, the sponsor has reduced the difference to be detected between the two treatment groups from 4.1 to 2.1. This treatment of loss to follow-up patients also increased the standard deviation to 5.0. This is important in the calculation of the sample size.
They calculated the sample size necessary to test for a difference of 2.1 with a standard deviation of 5.0 and a power of 80 percent for a two-sided test of five percent to be 90 subjects per arm.
To analyze the data, they plan to use an intent to treat analysis to assign the worst scores to subjects lost to follow-up. They also proposed using non-parametric statistics since the modified AFS scores were skewed to the right.
The sponsor planned to include 200 subjects in this study. Since they were conducting a concurrent European trial of 80 subjects, they would use only U.S. subjects unless they found that the U.S. and European subjects could be combined.
If combinable, they would stop the U.S. study at 120 subjects and combine them with the 80 European subjects to obtain the desired 200 subjects.
If not combinable, they would continue enrolling U.S. subjects until they had 200 and only use the U.S. subjects in their statistical analyses for effectiveness.
In their protocol, they specified three conditions which must be satisfied in order for it to be acceptable to combine patients across continents. You've already seen these, but, again, first, the baseline demographic pretreatment variables, including adhesion scores should be similar.
Second, there should be no significant interaction between continent and treatment effect.
And, thirdly, second scores should be similar.
In the following slides, I will demonstrate that the first two conditions do not hold and that, therefore, the U.S. patient group is the appropriate data set for statistical analysis.
I want you to first note that I changed this slide slightly from that in the slide you received for last week to make it easier to read. All of the numbers are the same. I just changed the order for ease of interpreting the results.
I'm not very good with these pointers, but I will give it a shot.
Note that the sample size is approximately 100 each treatment group in the U.S. and 40 per group in Europe.
(Pause in proceedings.)
MR. KOTZ: Okay. Thank you. We'll keep on going. Okay. Thank you.
First note -- oh, wait. Let's first look at the modified AFS score. Note that differences between continents for both groups are statistically significant -- forget that -- are statistically significant, and note that the baseline in Europe is about two to three times greater than in the U.S. So baseline in Europe is twice as large. In Europe it's twice as large as that in the U.S. for the INTERGEL patients, and the same is true for the control patients.
For the incidence of adhesions, the differences are also highly statistically significantly different, less than .001, for both treatment groups, and the baseline in Europe is also two to three times greater than that in the U.S., six versus 2.49, the INTERGEL patients, 6.4 versus 2.3 for the control group, control patients.
This second condition is that there should be no interaction between continent and the effect for each of the treatments, INTERGEL and the control. We must measure this as a change from baseline since it has been established that the baseline across treatments -- across continents are so very different.
This interaction is evaluated in the next four slides for each combination of treatment, INTERGEL and control, and for each endpoint, modified AFS and incidence.
First, let us look at the change in the modified AF score from baseline to second look across continents for the INTERGEL patients. Note that in the U.S. the second look is three and a half times greater than the baseline, 2.74 versus .78. In Europe it is only about 40 percent greater, 2.2 versus 1.6.
When we look at the graft, we see substantial interaction between the change from baseline in the U.S. and Europe. That is illustrated by the fact that the lines are not parallel. Rather, they intersect.
The difference between continents in change from baseline is even more pronounced for the control patients. In the U.S., the second look is more than four times greater than the baseline, 2.8 versus .68. In Europe it is only 25 percent greater.
When we look at the graph, again, we see substantial interaction between the change from baseline in the U.S. and Europe.
Now, let us turn to the same comparisons for incidence of adhesions. Again, we see for INTERGEL the increase over baseline was greater than threefold, .68 to 2.83, while less than 25 percent in Europe, and again, our graph shows substantial interaction, the continent and change from baseline.
And finally for the control we see a three and a half fold increase over baseline in incidence of adhesions for U.S. patients, while for Europe, the increase over baseline is only 30 percent, for three and a half-fold, jumps from 2.27 in the baseline; at second look in Europe, 6.4 to 8.2. That increase represents 30 percent. And, again, there is evidence of interaction.
In summary, the baseline values are highly statistically significantly. In fact, we have seen that the modified AFS score and number of adhesions are consistently two to three times greater in the European patients. The U.S. and European patients are very different with respect to baseline values.
We also saw substantial interaction between change from baseline and continent. At second look, the U.S. patients scored three to four times greater than baseline for both modified AFS and incidence of adhesions while the European patients show only a modest 25 percent to 40 percent increase in their scores.
Thus, the U.S. and European patients appear to respond very differently to effects of treatment. Therefore, it is clearly not appropriate to pool the U.S. and European patient data, and thus, the U.S. data is the appropriate data set to evaluate the device effectiveness.
Next slide.Now let us turn to the issue of comparing the effect of INTERGEL on adhesion formation to that of the control, lactated Ringers, for both the primary endpoint, modified AFS, and the secondary endpoint, incidence of adhesions.
We will use the statistical analysis plans specified in the protocol. That is an intent to treat analysis of the 200 U.S. patients.
Next slide, please.
First, let us look at the results for the modified AFS score. Both baseline and second look scores are provided. It is clear that there's very little difference between INTERGEL and the control for both baseline and second look scores, .78 -- .68 at baseline and approximately 2.74 to 2.83 at second look.
The magnitude of the difference at second look is only .1 in modified AFS score, and the difference is non-significant using Wilcoxon test. We see the difference in change from baseline was only about .2, 1.96 for the INTERGEL to 2.15 for the control.
Now let us look at the results for incidence of adhesions. Again, notice that there is very little difference between INTERGEL and the control for both baseline and second look. In fact, the magnitude of the difference is the same at baseline and second look, that is, .2 adhesions, and if you notice there's no change from baseline for the two -- in score for change in baseline in the two groups.
All differences were not significant using the Wilcoxon test.
Next slide, please.
Therefore, when using the analysis plan specified in the protocol, that is, an intent to treat analysis of the 200 U.S. patients, we found that there was no statistically significant difference between INTERGEL and the control for the modified AFS score and no statistically significant difference between INTERGEL and the control in the incidence of adhesions.
In fact, the observed difference of .1 for the modified AFS score was much smaller than the 2.1 difference that this intent to treat study was designed to detect.
Thus, it has been demonstrated that the data are not combinable across continents, and therefore, the complement of 200 U.S. patients is the appropriate one to analyze and has been demonstrated there is not a statistically significant difference between INTERGEL patients and control patients with respect to either modified AFS or incidence of adhesions at second look.
Now I would like to discuss problems with the sponsor's analysis of all of the patients. If you recall, the pilot study had a difference of 4.1 in the modified AFS score between INTERGEL and control at second look.
After this study was adjusted for an unequal loss to follow-up and the worst scores were given to patients lost to follow-up in both groups, the sponsor calculated that the difference in modified AFS score that they wanted to detect was now 2.1 and that the standard deviation would increase to 5.0.
As already mentioned, the resulting sample size necessary to detect this difference at 2.1 with 80 percent power based on this intent to treat analysis was 90 patients per treatment arm.
The sponsor's approach was to analyze the primary and secondary endpoints using the combined U.S. and European evaluable patients, ignoring loss to follow-up. This results in a study in which the clinical and statistical significance are no longer aligned.
In fact, the post hoc analysis plan now, the sponsor's post hoc analysis plan now has 80 percent power to detect a difference of only 0.75 instead of the 2.1.
Thus, using this approach can lead to analyses of the data that can result in misleading p values.
Now we will go on to Part 2 of my presentation. After the conclusion of the previous panel meeting, the sponsor submitted an amendment with a revised indication for use. The first part of the revised claim addresses adnexal adhesions.
To support a claim of effectiveness with respect to adnexal adhesions, the sponsor uses the AFS score as opposed to the modified AFS score.
The second part of the revised claim addresses pelvic and abdominal adhesion reformation, and the sponsor uses selected secondary endpoints performed on surgical site adhesions in an attempt to support this claim.
Let us first look at the second part of this revised claim. The sponsor uses the results for the secondary endpoints, reformed adhesions and surgical site adhesions, to support this part of the claim. Note that these two endpoints are subsets of the secondary endpoint incidence of adhesions. The sponsor claimed he found significant differences for these two endpoints, but only after departing from the original analysis plan using the combined U.S. and European evaluable patient group, excluding patients lost to follow-up.
The problem with this analysis plan is that it gives misleading p values for the reasons just discussed.
We should also note that since the sponsor failed to detect differences in total incidence of adhesions, the analysis of selected subsets of this endpoint can only be considered exploratory.
This slide presents averages for reform and surgical site adhesions for the U.S. intent to treat patient group. Also note that I have included results for de novo adhesions as well for purposes of completeness.
All of these form overlapping subsets of the total incidence of adhesions. Average incidence of baseline adhesions are also presented for purposes of comparison.
Now, let us look at the table for these endpoints. I think it is clear from this table that there's very little difference between the two products for any of the three endpoints listed above, reformed, surgical site, de novo adhesions. In fact, none of them even come close to approaching statistical significance.
In summary, not only were there no statistical differences between groups for the modified AFS and incidence of adhesions, but there were no statistical difference in any of the selected endpoints, reform, de novo, or surgical site adhesions, when analyzing the data as proposed in the study protocol.
Next we will review the data supporting the part of the revised claim dealing with adnexal adhesions. To evaluate this, the sponsor used an AFS score. Note that this score was specified, defined, and calculated after the study was completed. then exploratory or after the fact analyses were performed on the data.
The problem with exploratory analyses on post hoc endpoints is that if you look hard enough you can eventually find one with a small p value. That is, you will eventually find a difference that appears to be statistically significantly different.
Note that these types of analysis are best suited for exploring new hypotheses that can be tested in a new study. In the sponsor's last commitment, they presented the data for the AFS score stratified across three subgroups, though they have also previously presented the data partitioned in two, four, and five subgroups as well.
Because FDA didn't have sufficiently detailed data, we can only present an intent to treat table for the U.S. patients' data partitioned over two subgroups. This data is presented in the next slide.
Now let us look at the AFS data. Now in this table that the number of patients with the baseline status specified in the left is in the denominator. The number of patients, those patients having moderate or severe AFS scores at second look in the numerator, and I will illustrate this with the data in the first cell, that being this one.
Of the 79 patients with minimal and mild AFS score at baseline, 12 developed moderate or severe AFS scores at second look or we can take the next cell. There's five INTERGEL patients with minimal or mild -- I mean with moderate or severe adhesion score, AFS scores at baseline. Zero developed moderate or severe scores at second look.
As we can see from this table, there's no difference between INTERGEL and the control for the minimal-mild group, and we can reach no conclusion for the moderate or severe group as the sample sizes are so small.
In fact, there are only five INTERGEL patients with moderate-severe condition at baseline in the U.S.
Next slide. Oh, no, that slide.
In summary, when using the analysis plan specified in the protocol, the results show no difference between treatment groups for any of the subgroups, that is, post hoc endpoint.
In contrast, the sponsor's analysis used the smallest subgroup of patients with moderate and severe adhesions to derive their analysis, combined U.S. and European patients. But note that this subgroup accounts for less than ten percent of the patients in the study and includes on five U.S. INTERGEL patients.
Assess en route (phonetic).
MR. KOTZ: In summary, the sponsor has bought the design and intent to treat study to evaluate 180 to 200 patients, including an expected 20 percent INTERGEL and ten percent control loss to follow-up. The specified sample size was reached with 200 U.S. patients.
Since the U.S. and European data were clearly shown to be not combinable, the U.S. patient group comprises the appropriate patient group to analyze, clearly showing that using the analysis plan specified in the protocol, that is, an intent to treat analysis of the 200 U.S. patients, that there was no difference between INTERGEL and control for modified AFS score, the number of adhesions, reform and surgical site adhesions and AFS status. Thus, the product did not demonstrate superiority over the control for the primary endpoint or for any of the secondary endpoints the study was designed to evaluate.
CHAIRMAN RAMSEY: We're running well behind, and we have agreed to give LifeCore a chance to rebut, but I would like to ask the panel if they have any brief clarifying questions regarding the presentation for FDA.
DR. D'AGOSTINO: Richard, are you saying that it's inappropriate to pool because the protocol said that the rates have to be the same in the U.S. and Europe?
Where I'm going is that if the subjects are randomized within centers and within countries within centers, what difference does it make if the rates are higher in Europe versus U.S. on baseline?
MR. KOTZ: Well, not only are they higher on baseline, but we saw the treatment effect very different as well. The effect of treatment seemed to be very different, too.
It just makes it very difficult to (pause) --
DR. D'AGOSTINO: I don't want to take too much time. We can get into this, but I think this is a point that we want to get back to.
CHAIRMAN RAMSEY: A quick on hopefully?
DR. SHIRK: Yes. I guess I'd ask Richard obviously the same question that I asked the statisticians from LifeCore. I guess your opinion is that by not following the intent to treat protocol, that you significantly skewed the data towards success rather than failure. Is that sort of what you're saying, or that you've got a smaller --
MR. KOTZ: Yes. Well, they affect the p values of the statistical test, the statistical test using an evaluable analysis on the intent to treat study design.
Does that address your question, Dr. Shirk?
DR. GORDON: I have one comment here.
CHAIRMAN RAMSEY: Okay.
DR. GORDON: Just quickly, I guess just a general concern relative to how protocols go through the IDE PMA process. It was of concern to me in reviewing the information that the protocol -- and you know, there were copies of the statistical plan throughout these documents -- clearly identified three types of analyses, and yet FDA has identified the intent to treat as the only one the sponsor agreed to present and hasn't presented anything else, and that isn't clear to me from the documents.
So I'm assuming that the protocol implies that --
MR. KOTZ: For effectiveness, for statistical effectiveness the intent to treat analysis was the one that was specified. Yes, they specified several analysis plans, but we requested an intent to treat analysis plan for demonstration of effectiveness.
DR. GORDON: But it also shows or identifies an efficacy for efficacy purposes an evaluable -- I'm just making the point again relative to this process.
MR. KOTZ: Yes. If the panel is interested, I have all that data that I presented for the evaluable patients in the U.S., and I can present that to you, and you can evaluate the differences.
I don't think it's appropriate to statistically evaluate that data, but you can -- I'm perfectly happy to put that up.
CHAIRMAN RAMSEY: Let's -- okay. Go ahead and make one response if you'd like. I'm sorry. I don't know your name.
DR. WITTEN: Excuse me. This is Dr. Witten from FDA.
I just want to clarify in response to your question that the clinical perspective that was provided by Dr. Horbowyj was based on the same data set used by the sponsor, which was the evaluable data set of the combined cohort.
CHAIRMAN RAMSEY: We will have time for questions after this session, and we are running late. So I would like to give the sponsor their designated time for rebuttal. They get extra credit if they can do it in less than 15 minutes.
DR. BECKER: Thank you.
I would like to clarify a couple of things for the record, and then we have a statistical comments by Dr. Piantadosi and clinical comments by Dr. DeCherney.
First of all, I'm very concerned about the discussions regarding the analysis plans specified in the protocol. Everybody has a copy of the protocol in your panel pack. If you'll refer to the protocol on page 28, it clearly identifies three evaluable cohorts for the efficacy analysis.
It does not state that the worst case imputation, the so-called intent to treat, is the primary analysis of the study. So on page 28 of the protocol, you'll find that there's three cohorts described, and all of those analyses were done by the sponsor.
Secondly, on page 31 you will see a special paragraph inserted into the protocol, and it reads, "As requested by FDA, a worst case imputation will be used to deal with missing data on patients at second look."
In 1995, I know it's hard for everybody to remember, and people, you know, who were there maybe aren't there anymore, but at the time that was considered to be by FDA -- that was presented as a requirement for clearance of the IDE, and the sponsor was told, "Put it in there and then do other appropriate analyses when you get the data per standard statistical practice."
So the sponsor did not choose to do this worst case analysis that I think everybody here agrees is unscientific and not sound.
Finally, before I turn this over to Dr. Piantadosi, I want to address another apparent misconception that will help us maybe bridge this disconnect today between our two views of this data and the submission under consideration.
We've heard the word "retrospective" many, many, many times, and I think that it's important to recognize, first of all, that as we have also said many times all of these analyses were prospectively defined in the protocol.
But FDA did, in fact, not just ask for, but required these AFS scores. On December 7th, 1999, during the course of the review of this PMA, FDA issued a major deficiency letter, and if you've never had a major deficiency letter sent to you, let me tell you what it is.
It's a letter in which FDA informs the sponsor that the review of the PMA cannot continue unless additional information is provided, and in order to complete the review you are given a list of analyses, studies, whatever that need to be done.
And in this major deficiency letter of December 7th, 1999, Item No. 8 says, "Please provide shift tables for standard AFS scores. The tables should be presented for the change in baseline after surgery standard AFS score and the unadjusted second look standard AFS score. The shift tables should show the shifts of patients from the minimum-mild disease group into the moderate-severe disease group and vice versa. In addition, the tables should be for all patients, patients with no adhesiolysis and patients with adhesiolysis."
So I want to clarify for the record that this information was required by FDA.
Finally, I hope that after lunch you might give us an opportunity to explain why there seems to be a misunderstanding about the shift tables and this perception that only a few patients benefitted, when in fact all subgroups of patients benefitted in this trial.
DR. PIANTADOSI: Thank you.
I'd like to address a couple of points focused on the statistical review that I think are in error, and much of the difference that you've heard between the sponsor presentation and the FDA conclusions are a consequence of that error.
I'm going to focus on the issue of pooling because you've heard several different things about pooling, and I'd like to really clear that up.
The short answer to Dr. D'Agostino's question about why either baseline differences or treatment differences between continents should be consequential is that they are inconsequential. As I indicated in my remarks earlier, the issue is whether or not there is a common treatment effect across continents.
The information that the FDA has showed you regarding this point actually doesn't bear on the point. If I could show the first transparency, this is a reproduction of the slides that the FDA has used to argue that there are differences between the United States and Europe that render the data not poolable.
This is incorrect. Look at the first chart in the upper right-hand corner. What you see there is crossing lines within the INTERGEL group. You cannot learn about treatment by continent interactions by looking only within one of the treatment groups, and what you see here is actually very similar effects during the trial within the INTERGEL group.
And similarly, in the control group you cannot learn about interaction between treatment effect and continent by looking within the control group, and the whole series of such analyses presented by the FDA in support of their point that the data are not combinable are incorrect.
The issue is whether the differences between these two control groups is a function of continent or not, and in that regard, this notion, this footnote which is mentioned on the table is a very telling footnote, and the FDA knows, for example, that they have to put that footnote there because if you remove this condition, that is, if you look within adhesiolysis category, you will see that the data are, in fact, combinable.
The second transparency I'm going to show demonstrates this. What you're going to see on this transparency is the results of the study and the baseline. Let's look over on the left first.
In U.S. and Europe, broken down by patients in the two adhesiolysis categories, look at the baseline and look at the treatment effects. They are similar in the U.S. and Europe in both adhesiolysis groups.
What the FDA has done in arguing against poolability is to combine the adhesiolysis groups, which you can see are different. They're different as a matter of characteristic of the patient and medical practice on the two continents. They have combined those two groups and then said that U.S. and Europe are not combinable on that basis.
In fact, the U.S. and Europe are similar and are combinable. That does satisfy the criteria in the protocol once you account for adhesiolysis.
Furthermore, if you look over here and pool across adhesiolysis, so this part of the analysis is analogous to, but not identical to what the FDA has done. You can see that actually there's a fairly strong similarity with the exception of the baseline scores in the U.S. and Europe.
Now, why doesn't this look more like the FDA analysis? It's because this has been based on scores for the evaluable patient population.
So what happens is if you make the two mistakes that the FDA has made, if you fail to account for adhesiolysis and you use the worst case imputation, then this picture looks like you should not be combining U.S. and Europe.
But if you read the protocol carefully, you'll actually realize that the criterion for poolability that I'm referring to here refers to the evaluable patient population.
So, in fact, this is the correct way to decide whether or not the data are poolable across U.S. and Europe. They are. One needs to simply account for the adhesiolysis variable and then you see the homogeneity of treatment effect, and you see the absence of treatment by continent interaction. All three of the criteria listed in the protocol are satisfied and the data are poolable.
Finally, I'd like to address one point. I know Dr. Rubin and Dr. Colton are to have something to say. The notion of what is a subset analysis, and this has also been implicitly incorrectly defined by the statistical reviewer at FDA.
A subset analysis normally refers to a subset of patients on the clinical trial, and why we worry about subset analyses are that we're afraid somebody is going to fish around in the data until they find a smaller set of patients than those randomized in the trial that demonstrates the kind of treatment effect that they would like to use to their advantage.
That is not what's been done here, and referring to these analyses as a subset analysis is incorrect. What has been done here is that the outcome which was prospectively collected has been used, but not all of the points on the outcome have been used.
This is not a subset analysis. It's the same issue as if I do a study with survival as an outcome, and I also measure time to disease progression or response rate in a cancer trial, and I may not use all of those outcomes or I may use an outcome censored by one rather than another. That is not a subset analysis, provided I conduct it on all of the patients on the trial, which is what has been done here.
So referring to this as a subset analysis in an attempt to remove and denigrate its pedigree is incorrect, and I want to be very sure that the panel understands that this is not that type of subset analysis.
DR. RUBIN: Although I'll focus my comments on the imputation and the ITT analysis, I did want to make one comment that Dr. Colton addressed earlier.
This issue of the pilot study and the difference between a pilot study and the pivotal study, we have a pilot study of 21 subjects, and you have a pivotal study of 281, and the fact that there are some differences there are supposed to somehow impugn the p values and the analyses of the unbiasedness of the randomized pivotal trial of 281 subjects is really absurd. They don't affect the p values. They don't affect the obtained data. They just don't.
The pivotal trial is more than ten times as large. It's not at all a surprise that the results in the pivotal trial are somewhat different than the results in the pilot study, and in fact, I think if you did statistical tests, did some back of the envelope ones, they're not even significantly different between the pivotal trial and the pilot because the pilot is so small and it has such large variability relative to pivotal trial.
Ted may want to add to that later.
With respect to the ITT population and the ITT analysis, as I suggested earlier this morning, you would see lots of transparencies with ITT analysis on it, as if that meant that that was the ITT analysis.
I want to emphasize again ITT refers to the intention to treat population because all of those patients who were randomized and treated, there were 281 patients randomized and treated. There were 16 without second look data, and in order to do an ITT analysis on that population, you must somehow impute either explicitly or implicitly.
Now, the analysis, the imputation analysis that the FDA proposed and carried out was based on the worth possible value being imputed from each woman without a second look. The worst possible value of AFS, the worst possible value of modified AFS; so even a woman who was pregnant at the time for second look who refused second look because she was pregnant, she was imputed to have the worst possible values of adhesions.
I don't think that's reasonable. I don't think it's scientific at all.
The fact that FDA can come up with an even less scientific and even less reasonable method of imputation, that is, everybody who was exposed to INTERGEL gets the worst possible value and everybody who was exposed to Ringers gets the best possible value, the fact that that's even a less scientific and less reasonable analysis doesn't justify the analysis that they did as being reason nor scientific. It's just not.
This idea that the results are only shown for a small subpopulation in the U.S., I do have some transparencies that were based on the scientific imputation that we did do that was blinded, again, to outcome, blinded to results, and maybe it's appropriate. I just put up one transparency which shows result of those analyses in the U.S.
So this is only in the U.S. It does not even include the European patients, and it only is looking at those patients in the U.S. with no baseline adhesiolysis, and as noted by other people, in the U.S. more than half the patients had no baseline adhesiolysis.
And you'll see that in this subgroup that there are 63 INTERGEL, 59 controls, and with respect to all the outcomes, they're in the favorable direction, and for the modified AFS, which is the primary endpoint, both adjusted and unadjusted, adjusted for baseline adhesions, the results are significant, well beyond the traditional .05 level.
So if you want to take away some P values that you think are scientifically founded, there they are, and they're strongly in favor of INTERGEL over control.
DR. COLTON: I just want to add a few words with regard to the issue of statistical power, and in fact, actually in some of the correspondence from the FDA, there was a term used of a study being overpowered, and Dr. Kotz referred to misleading p values.
And first of all, if one believes in p values, and not all statisticians believe in p values, but I think certainly the FDA is among those who worship at the shrine of the p value --
DR. COLTON: -- the p value is, the power of the study, as I said, is a hypothetical statement. I think we've shown with reasonable back-up and evidence that the continental and the U.S. sites can be combined. There's no reason not to combine them.
Even though the study may have been designed to have a sample size of 180, here are the data. Here are the total data we have. Here's what we calculate, and putting it in terms of a p value, here is the p value.
It's not misleading. This is the statistical significance. Whether that p value, for whatever difference was found, is clinically meaningful, I think we've already said that that is really an issue that is to the clinical expertise.
But to me what has been done in this study, here are all of the data we have. We've analyzed all the data. We've stratified. We've looked at the continent versus the U.S.
When we look at all of the data together and we make reasonable assumptions about the missing data, these are the p values that we come up with, and I don't think they can be in any way said as being misleading or being overpowered. This is what we found, and the issue is: is this clinically important?
Okay. I just have three points. Number one, the fact that adhesions are not predictive of morbidity, I don't think that's true. As far as infertility is concerned, there are a fair number of prospective studies looking at the surgical approach where infertility is corrected because the only pick-up mechanism is altered.
Now, as far as the AFS score versus the modified AFS score, essentially the -- and you heard that this was required -- the AFS score is really a carve-out of the modified AFS score. all of the data points that are necessarily to calculate out the AFS score are present in the modified AFS score as designed in the study.
So although the score is retrospectively calculated, it's based on prospective data.
Now, the criticism that neither of these scores are perfect and predict outcome, I think that that's fair. There are no studies documenting that the AFS score predicts outcome, and basically what it's evolved to us just an observational score. There has to be some way to observe adhesion formation and adhesion reformation, and the AFS score just quantitates that.
Now, just looking at outcome, Dr. Horbowyj talked about the incidence and did quote the figures of a 31 percent decrease in reformation and a 23 percent decrease in surgical site adhesions, and as far as a clinical standpoint, that's impressive and comparable at least to other adjuncts of therapies that are available.
DR. FARO: In regards to their analysis on infection, it would be helpful for me to understand what criteria Dr. Horbowyj used if she reviewed these clinical records in deciding which patients actually met criteria for infectious postoperative morbidity.
In reviewing it on a limited case basis, the wound infection rates, if we just focus on that, was comparable, two patients in each group. I don't think that Ringer's lactate or lactated Ringers or INTERGEL contributed to wound infection in either case. I think the low rate of wound infection is acceptable in gynecologic abdominal surgery that we saw here.
If we look at a patient classified as having bladder pain and then classified as infected, it's rather difficult to determine a diagnosis of infection in this patient. Many patients who have Foley catheters inserted will have bladder pain postoperatively. They will have bladder spasm secondary to the indwelling catheter.
So we have a discrepancy here, but the bottom line is this compound in no way supported infection in the animal studies. There was no statistical difference between the two groups, and I would expect if this compound had a basis for inciting infection in the pelvic cavity or in the abdominal cavity, we would have seen a significant number of patients who developed a paralytic ileus following surgery, and this was not the case.
And these dissections that are done in this type of surgery often involve adhesions between the adnexa, the uterus and the bowel.
So I think this is, in my opinion, very safe agent with regard to infectious morbidity and the potential risk for infectious morbidity.
CHAIRMAN RAMSEY: Okay. We are 40 minutes behind schedule. Yeah, I think we're going to cut lunch to 45 minutes. That will be the penalty for over-going, and let us convene back at five minutes to one.
(Whereupon, at 12:11 p.m., the meeting was recessed for lunch, to reconvene at 12:55 p.m., the same day.)
CHAIRMAN RAMSEY: Welcome back. The meeting will reconvene with a panel discussion portion of our meeting.
Although this portion is open to the public, we ask the public attendees not participate except at the specific request of the panel. Okay?
As I stated at the beginning of today's meeting, the panel is charged to answer the following question and to make a recommendation to the center director as to how this dispute should be resolved, and I'm going to read the question and try to be accurate this time.
Does the PMA, as amended, provide reasonable assurance of the safety and effectiveness of INTERGEL for its intended use as an intraperitoneal instillate for reduction of adhesion formation following gynecologic pelvic surgery?
And in answering that question, we have two specific questions to address. One is whether the statistically significant differences between INTERGEL solution and control can be considered clinically significant, and second, whether the benefits of the product outweigh the potential risks, including any risk of infection.
And I think a couple of procedural things before I get started. Someone left a key that was found on the floor, and if this is anyone's we'll leave it out front for you to get.
And so let me go on. So what we're going to do now is the panel is going to discuss among itself and ask questions of the sponsor or the FDA.
One thing for the panel. Dr. Piantadosi has to leave at 1:30 to teach a class, and so if you have specific questions for him, now would be the time to do it.
But let me, again, to get things started and frame the question, let me start with the first question: whether the statistically significant differences between INTERGEL solution and control can be considered to be clinically significant.
And I'll open it to the panel to address that question and ask questions to the sponsor or FDA.
DR. D'AGOSTINO: Let me start maybe a little bit further back than where you are right now, but I think the question of surrogate versus endpoint is an important one. I mean, I think that the adhesions is an appropriate endpoint. If it isn't, then we're wasting our time.
Am I comfortable in making that assumption, that everybody thinks that the adhesions is all right?
Then the question that I have and want to begin the discussion is one of the things that bothers me, and I keep hearing, I think, different things from the FDA versus the sponsor, was the shift from the modified to the unmodified or whatever the R stood for, was that analysis not included in the package that went to the previous panel? It wasn't discussed, but was it included?
And when did that analysis come?
See, I'm bothered by the notion that you have an endpoint in a protocol and you direct your analysis to that, and then you later on -- not a subset -- but later on you shift to a new endpoint, and when did that shift come? Was it FDA suggested?
DR. KRAUS: That data was presented to the January 12, 2000 panel.
DR. D'AGOSTINO: So it's not completely --
DR. KRAUS: Slightly different than what was presented in the January or June 12 or June 2nd amendment, but basically the same data.
DR. D'AGOSTINO: So it's not a completely off-the-wall, brand new thing that we're facing.
DR. KRAUS: No.
DR. D'AGOSTINO: And then the other question, just to go on if you don't mind when I have this here, but I'd be happy to give it, with this pooling question I understand -- I think I understand -- what the FDA's concerns are, but I think also that good statistics practice would argue that you can, in fact, reasonably pool the data if you do the randomization within groups and so forth and you have an analysis that indicates that the treatment effects are pretty stable across the items you're pooling.
So I do, again, think I understand what the FDA's arguments are, but I do think that if I'm making sense and if people are agreeing with me, I think the U.S. and the European, pooling those together and looking at analysis does make sense to me, and again, I stand to be corrected, but I think that's -- from what I heard, I think that's a reasonable approach.
CHAIRMAN RAMSEY: Go ahead.
DR. GONZALEZ: My question, it's more of an observation. I just want to know if the FDA and the sponsors feel the same way on it, but, you know, in looking at the presentations this morning, there was a concern of prospective versus retrospective, and my question is from what I can gather -- and I'm just trying to make sure that that's my correct perception -- from what I can gather, the agency's definition of retrospective is the pulling out of the data from data that had been prospectively collected.
Am I correct on that assumption?
CHAIRMAN RAMSEY: Do you want the agency to respond to that?
DR. GONZALEZ: Yeah, whoever feels --
CHAIRMAN RAMSEY: Okay.
DR. WITTEN: The point that was made about the retrospective AFS score, yes, was pulling out those data points and then collapsing them into another score.
CHAIRMAN RAMSEY: Go ahead. Dr. Shirk.
DR. SHIRK: Well, I guess I've got several questions, but they all sort of revolve around the initial study design. It seems to me that a lot of the problem that we're seeing with the statistical analysis comes from the fact that the initial pilot study was basically based totally on adhesions, and that the scores were fairly high, and you could see a significant drop, you know, from the treatment group to the initial group.
So that the initial study was based and a statistical model was based on basically treating patients with adhesions and seeing what happens afterwards.
The study then went on. The final study basically incorporated a group of patients that include a large group of patients that had no adhesions to begin with, and then to see what those adhesion -- what kind of adhesions those procedures ended up with.
These are obviously procedures with known adhesive complications, but to me it obviously shows that we'd have had an easier time with just all adhesion patients and seeing how we drop off.
The problem is the number of parameters that are involved with this thing and how do we statistically handle the number of parameters? There's no control on surgical technique. There's no control on the procedures done or materials used in the procedures. Each of these have significant factors as far as adhesion formation.
And then you've got two groups of patients, one group that has adhesions and you're trying to see how many adhesions didn't reform, and then a group of patients that basically had no adhesions who then basically formed adhesions. Certainly the new indications are basically almost aimed at that small subgroup of patients that had dense adhesions to begin with and then didn't reform them, and how can we make a statistical statement on a small subgroup of patients in the study versus, you know, using the whole group of which, you know, two thirds of them didn't have any adhesions to begin with.
CHAIRMAN RAMSEY: Would you like either group to respond or just comment for us?
DR. SHIRK: Well, I think, yeah, I guess first of all I'd like Ralph to respond to that.
DR. D'AGOSTINO: No, I think that's --
DR. SHIRK: And then the group.
DR. D'AGOSTINO: I think that's important. I didn't want to go on and on, but that's where I was heading also. You have 80 percent with basically nothing at baseline. Some of them develop adhesions. You have I think it's 26 individuals, if I have the right numbers. Twenty-six individuals had moderate-severe adhesions to begin with, and is that the group of real interest?
And I did mention it earlier today, and I though the sponsor said they were going to respond to it later on. I don't find the numbers very compelling. As a statistician, I'm surprised there's so much statistical significance when I'm dealing with three versus 17, and then when I split it up such as you're doing, which I think is perfectly appropriate and correct, it's even less compelling.
And I think the sponsor or somebody should try to fill us in on that.
DR. GORDON: Judy Gordon.
I think the sponsor mentioned that there seems to be and I similarly was left with the impression of small numbers relative to the shift scores, and I think the shift scores serve at least in part to address some of the issues you're describing in terms of surgical factors.
But they didn't have a chance to present it. I don't know if everyone is interested in seeing it, but it was suggested that there was a larger effect than maybe we're left with the impression that there is. So if others are interested, maybe this is a good time for the sponsor to show this.
DR. SHIRK: But my question is if we're looking at shift scores or basically looking at a very small population, and is that population we're looking at specifically empowered to do or meet the question that we're asking.
CHAIRMAN RAMSEY: Any comments? Do you want to have some response?
DR. PIANTADOSI: Well, let me try to provide a sensible answer. I think it's a fair question. There's a lot of factors flying around here, and you asked the question how can we cope with those and account for and sort out those factors in dissecting out the treatment effect, and the answer is a word. It's very simple. And the answer is randomization.
That's why we do randomized studies, so that we don't have to make an explicit model based, quantitative accounting of all the factors that we think are affecting the outcome.
It's a fair point that maybe not all the patients stand to benefit from a given intervention. That's true of every clinical trial that's ever been done. It's particularly true in my home base, which are cancer studies, where it seems quite evident that when you apply a treatment it's not necessarily the case that everybody benefits a little bit or a known amount. Some people seem to benefit a lot; others don't, but that kind of finding, since we can't tell ahead of time who is going to benefit and who isn't going to benefit, that's why we use the rigorous methods for reduction of bias, estimation of the treatment effect, and that's why we use a larger study, so that we get enough of the kind of patients who are going to experience a benefit to detect.
There is nothing about the fact that a relatively small fraction of patients appeared to benefit, that invalidates the methodology of the trial or the result that you're seeing. This is part and parcel for why we do randomized trials.
DR. D'AGOSTINO: Well, you say, well, maybe there's some subjectivity in rating the lesions or adhesions.
DR. PIANTADOSI: There is.
DR. D'AGOSTINO: You know, then if you have three versus 17, if maybe the 17 wasn't really 17, but it was 14, and then maybe the baseline is nine versus 17; well, maybe if it was nine versus nine, you know, there's be something different.
So the --
DR. PIANTADOSI: Yes, if the data were different, the result would be different, but they're not. And, yes, there's subjectivity. That's why we have a control group.
And we can change the data and say hypothetically we would come to a different conclusion if the data were different, but we have a large, rigorously done trial. There's masking to remove the influence of the subjectivity. There's randomization, and the outcome is the outcome.
DR. RUBIN: I'd like to amplify that if I may. I'd like to amplify that.
Steve's point on the importance of the randomization really is critical here, and I'm sure many of you realize this, but if you have a randomized trial and you have some noisy data, there's a noisy baseline assessment perhaps and noisy second look assessment, that noise can only contribute to smaller estimated effects of treatment versus control, can contribute only to less significant effects of treatment versus control.
It can't create a significant difference. The assessments were blinded. No one knew whether people were taking INTERGEL or Ringer's solution. If you had found nonsignificance, you could say, "Well, that's because there was a lot of noise in the data. There was a lot of uncertainty in how you record this subjectivity and how you record this."
But once you find significance, once you find real effects that are consistent, that noise can't explain them away.
I've been a situation where we have an endpoint. It doesn't turn out to be good. We say, "Gee, look at the date." And you say, you know, "Maybe compliance is what's going on," and you get the data and you analyze it for compliance, and, boy, compliance was it.
And you analyze and no matter how you analyze it from that point on, compliance is it.
You run another study and compliance shows nothing, but severity does, and no matter how you analyze it, severity does.
And so we have this sort of shifting of endpoints and what have you with some of these numbers. So it's not as -- you know, in a nice world what you're saying would be very comforting, but I think that there's one study here. It's not going to be replicated. It can't be, $25 million or something like that, and so there's a lot of -- and the numbers are small, and so I think, you know, pushing to think these things out and just to say randomization handles it, it's a little discomforting for me.
DR. RUBIN: I agree with that, but I think that the characterization that there was a shifting of endpoints, looking around for the correct endpoints really isn't an accurate characterization.
DR. D'AGOSTINO: This is what I was trying to get at.
DR. RUBIN: Yes.
DR. D'AGOSTINO: When was it unclearly?
DR. RUBIN: I understand that. I think Karen may be more appropriate than I in describing the pathway that took place, but in my understanding, all of these endpoints were either specified in the initial protocol as approved by FDA or requested by FDA, and there weren't lots of them. There were just ones to clarify the conclusions. They weren't shifting around looking for significance. I think there always was significance, and there was just repeated significance with these different outcomes.
DR. BECKER: I'd just like to add that the original primary endpoint for the pivotal trial, which is the mean of the adhesion scores evaluated at the 24 anatomical sites was also statistically significant.
CHAIRMAN RAMSEY: Let me get the FDA. Do you guys want to chime in on this discussion?
DR. WITTEN: Well, I just want to clarify about the deficiency letter or deficiency letters in general, which is sometimes when we're looking at a trial and in discussion with the sponsor, they indicate some other information that might shed some light on the benefit of their product. We will ask for that analysis in a deficiency letter.
And then, of course, we would want to look at the results as we did here.
CHAIRMAN RAMSEY: Okay. Go ahead.
DR. KIM THORNTON: One of the questions we were asked to look at regarding the statistical significance, there seems to be a difference in terms of the intention to treat analysis in terms of how the decision is made as to use the worst case scenario in both the INTERGEL and control groups versus a more conservative estimate, and how do you rectify that because obviously you're coming up with a completely different conclusion.
CHAIRMAN RAMSEY: Go ahead if you'd like to respond first.
DR. WITTEN: The question that you're being asked, the first of those questions -- I mean, I'm sorry I don't have the right wording -- is whether or not the statistically significant results that were provided show clinical significance or something like that.
And so I think one question is if you -- and that question is if you take the analysis that's under discussion provided by the sponsor, and you accept that analysis taking into account how it was performed and what the study was and how it was designed.
Is there clinical significance of those results?
DR. PIANTADOSI: Could I provide a supplemental answer to Dr. Thornton's question?
The reason way in which that decision is made or how you evaluate evidence is to essentially do a sensitivity analysis. You try several different things, and you ask yourself are the results under different sets of assumptions and different procedures consistent with one another.
If I do the evaluable cohort analysis, if I do an all patients analysis with the worst possible score imputed, do they look the same?
If I use other methods for imputation, if I make other assumptions about how to replace those 16 missing data points, what consistency do I see across all of those analyses?
And actually what we've seen is a considerable degree of consistency. The one analysis that appears inconsistent with the other set is the one that makes the most possible conservative assumption, and that is the one that seems to generate the heat or the nonsignificant findings in the questions.
But the general answer to the question is let me try several different things and see if the finding holds up under any possible set of assumptions.
CHAIRMAN RAMSEY: Go ahead.
Actually since the FDA -- since that was a main issue, go ahead and address that, and then we'll hear your question.
MR. KOTZ: Yes. There are, you know, a lot of different ways to look at the data, and that's not the only way. The way the FDA looked at the data is not the only way to show nonstatistical significance. I'm just trying to show you that there are many ways. If you take the data and you look at all of the evaluable patients in the U.S., only evaluable patients, and if you weren't lost to follow-up, if you look at that data set, no statistical difference in the secondary endpoints.
So, you know, there is a question of, you know, it just has to do with how you -- it does have to do with how you look at the data and what the statistical analysis is, and you know, they think one way and we think the other.
So I don't know if we're ever going to, you know, totally resolve that issue.
CHAIRMAN RAMSEY: Let me have you ask your question now.
DR. SHIRK: Well, I guess I'm going to come back to the original question that I asked, and that's basically the study design from the original PMA and ask each of the groups to say obviously what was there, was that a statistically fair model to look at, given the situation, and if not, why not? Okay?
DR. RUBIN: Maybe you could clarify that a little bit. When you say was that a fair model to look at, are you talking about the pilot study outcomes as a model for designing the --
DR. SHIRK: Right. Using the pilot model and the spread that they wanted between the adhesion scores, and it seems to me like what we're arguing basically, the FDA is arguing that we set up this model, and obviously it didn't meet the model.
Now, you're saying, well, vote with the model. You know, we want to know whether it's statistically significant or not, and we're going to use whatever statistical analysis that we've got to get there.
And I think that, you know, my question basically is using the initial data that was presented and the way the study was set up and the way the statistical analysis was set up, was that a fair model. So --
DR. RUBIN: I think I understand.
DR. SHIRK: -- the noise is out of it. Okay?
DR. RUBIN: I think I understand the question and reorient me if I don't.
The pilot study in my opinion is completely irrelevant to understanding the results of this trial. We could have designed this trial, conducted this trial and analyzed this trial whether the pilot study was ever done or not. It is completely irrelevant.
The fact that we use pilot data and apparently the pilot data that were obtained were within the scope of variation of what was observed, but they were not a perfect match to what was subsequently observed in the trial. It's irrelevant.
The point is that the randomized study was rigorously done. There are no methodologic flaws in that study, and it demonstrates a particular result.
The fact that that result either incorporates heterogeneity within the study centers or is slightly quantitatively inconsistent with some earlier smaller pilot study in no way changes the inference from the randomized trial. It's completely and totally irrelevant.
This is a point that the FDA has missed. They're reasoning incorrectly about this. The fact that the power calculation was done, the variance estimate was taken from the pilot trial in no way changes the inference from the randomized trial.
DR. RUBIN: Just to reinforce that -- am I allowed to reinforce or may I just reinforce that?
I've been doing statistics for many, many years, and I'm traditionally trained. My advisor was Wayne Cochran, and the statistics here are fine, and the results in the pivotal trial are not affected by the results in the pilot study.
CHAIRMAN RAMSEY: Let's give the FDA a chance to discuss the issue.
DR. HORBOWYJ: I think one of our main issues, and looking at the pilot study as well as the pivotal study, is looking at it clinically and looking at clinical significance in addition to statistical significance because we are approving this for patients, and it's really whether or not something is statistically significant -- our most important part is whether it's clinically significant.
And in that point I think it is relevant somewhat to look at the pilot study where a small number of patients were treated and a certain result came up, and then at the pivotal study as well where a large number of patients were treated and the number of patients who are perceived to have had a benefit are of an order of magnitude that almost approximates that of the pilot study.
CHAIRMAN RAMSEY: So in other words, I want to make sure the panel is clear on this. I think the issue is whether the study was designed to detect a certain level of difference between treatment and control and what was found was about half of what it was originally designed, but it was still found to be statistically significant if you include Europe and the U.S.
DR. HORBOWYJ: And the first question asked: is it clinically significant? Are the results of this study clinically significant?
CHAIRMAN RAMSEY: And that's really the first question we've been asked to focus on.
DR. HORBOWYJ: That's really the first question, right. So if you have a change, a mAFS score of one and a change of adhesion incidence of one, is that clinically significant?
CHAIRMAN RAMSEY: Okay. So Dr. Carlson would like to ask a question.
DR. CARLSON: So first I'm going to try to clarify something for myself, and anybody please chime in and correct me if I'm off base.
Whether it's relevant or not, a lot of the discussion has had to do with this pilot study and the basis for which the study was designed and the expected results based on that pilot study. The prediction was that there would be a certain numerical difference. In fact, there wasn't. There was a lower incidence of these adhesions even in the control group such that it was almost impossible to demonstrate that degree of a numerical difference.
So the question then comes down to is the percentage difference that we're seeing clinically relevant because the numerical difference is not nearly what was expected when the study was designed.
Is that a fair way of assessing this, FDA? And then I'll --
DR. HORBOWYJ: I think that when you -- and this is my personal opinion as an FDA personnel -- that when you look at percentages, you do have to look at the same time at the numbers because you can have a change from two to one and it's 50 percent, or you can have a change, you know of much greater magnitude, and it somewhat has a different significance.
DR. CARLSON: But you have to look at the opportunity for improvement as well, correct?
DR. HORBOWYJ: So I think you have to look at everything, and see really what is significant.
DR. CARLSON: Okay.
DR. HORBOWYJ: I don't think it's appropriate to just choose one.
CHAIRMAN RAMSEY: Okay. Do you want to have a quick response?
DR. PIANTADOSI: Yeah, just a quick response.
And I agree that whether or not a treatment effect of a given magnitude is clinically significant or not is the fundamental question. However, that information or the answer to that question is not contained in the hypothetical that was constructed prior to the start of the trial.
Nobody in this room has done more power calculations than I have. I do them by the hour, and --
DR. PIANTADOSI: Okay, Ralph. I take that back.
And all I can tell you is that these are hypothetical constructs whether they are based on real pilot data or whether they're based on the literature or whether they're based on the gestalt of the investigator.
And that's fine as a way to tell yourself how big the study should be. Once the data are in hand, the hypothetical that you engaged in to pick the sample size is meaningless.
DR. CARLSON: I understand.
CHAIRMAN RAMSEY: Go ahead, Hector.
DR. GONZALEZ: My question doesn't have much to do with the study and so on because there's a voluminous amount of information, but in reading the materials one of the things that I picked up, and I think probably the sponsor might be able to help me with that question because they're the ones that raised it in one of their letters, but in the letter of January 4th, you know, that Dr. Becker sent to Les Weinstein, on page 3 at the top of it there's a statement that says that CDRH approved compassionate use of INTERGEL in two patients and then on three others on an emergency basis.
What was the outcome of those patients clinically?
DR. BECKER: Gere, would that be a question for you? Doug?
DR. JOHNS: The five patients in question, if you will, four of those patients had small bowel obstruction and were having surgery to clear the adhesions associated with that bowel obstruction. The fifth patient was a chronic pain patient. All five patients had surgery. All five patients received INTERGEL. All fine patients are doing fine, no second surgeries required.
CHAIRMAN RAMSEY: Dr. D'Agostino.
DR. D'AGOSTINO: Just in terms of the pilot study to hopefully put it to bed, I mean, I agree 100 percent with what was said that you need something to generate numbers, but once you have the data, don't go back to that. It's a story that's not interesting anymore.
And so I think if I'm following where we're heading, though the question starts off saying statistical significance, I think that I'm comfortable that there is statistical significance, and then question then for me becomes when I look at the data I'm basically looking at three versus 17.
The sponsor tells us this is a fivefold difference when you do the appropriate adjustments, but I still see three versus 17, and is that clinically relevant?
I'd love to give an answer to that.
CHAIRMAN RAMSEY: Let's actually have some medical --
DR. MASTROIANNI: But isn't that why we do statistical analysis? Because numbers are small, and it extrapolates the numbers to make clinical sense?
I mean, any study then is flawed because how many patients is enough to show effectiveness? Yes, 100 percent is okay, but is 50 percent okay? And is that repeatable?
So I think that we have to take the statistical analysis of the information, and if it is statistically significant, then it does make a difference, and it is clinically relevant.
DR. D'AGOSTINO: In a number of other fields we worry about that question, and we say that the study should run until we get enough events so that we don't end up with 20 events. We end up with 500 events, and then we split it up and see treatment versus control and so forth.
DR. MASTROIANNI: I mean, sure.
DR. D'AGOSTINO: So, you know, it's a real question.
DR. DeCHERNEY: Well, that's not necessarily true through. I mean, there are many --
DR. D'AGOSTINO: No. I say this is the issue.
DR. DeCHERNEY: You can always do a limited number of patients in any given study.
DR. D'AGOSTINO: Exactly, and that's the issue. Do we think the three and 17 has supplied enough information because of all of the statistical safeguards, the design of the study or do we want something else?
CHAIRMAN RAMSEY: Go ahead.
DR. SHIRK: But that was my question. I mean, basically we're looking at 20 percent of the study, you know. Forget those patients. I mean, obviously that becomes a much smaller group than the 200 you were talking about before. Okay?
I mean, are we to make our decision on essentially 40 patients? I mean, that's --
DR. DeCHERNEY: Well, because that's why you take the events and you apply statistical analysis to it.
DR. SHIRK: What can you say on 40 patients? Do you know what I mean?
CHAIRMAN RAMSEY: Go ahead. Please talk to the microphone.
DR. HORBOWYJ: When looking at three and 17, I would just like to remind you also that there was a difference of eight patients at baseline. So I think that is something to be taken into account. There were eight more patients in the control group who started out in that group. So then the difference really is 6.5 or six patients. I'm sorry.
CHAIRMAN RAMSEY: You know, it seems --
DR. BECKER: That's inaccurate.
CHAIRMAN RAMSEY: Okay.
DR. BECKER: Or we're misunderstanding Dr. Horbowyj's slide.
CHAIRMAN RAMSEY: Does the panel want to have clarification on that issue?
DR. SHIRK: Yes, I would like to see.
CHAIRMAN RAMSEY: Okay. Let's let LifeCore present, and then I'd like FDA to comment on whether they agree with the accuracy. So go ahead.
DR. BECKER: Okay. I think there is a misunderstanding about the shift tables, and they are difficult to grasp when you only get to see it for 15 seconds. And patients benefitted both ways, in both directions.
So actually Doug Johns can put that slide up again and explain the shift tables again, and then we would like to comment on FDA's analysis of that data because some calculations were done that it looks as if the second look scores were adjusted at baseline twice.
DR. FARO: Can I add something?
CHAIRMAN RAMSEY: Okay. Quickly, while we're putting things up.
DR. FARO: While we're setting up the overhead.
The comment about making an inference on a small number of patients in a study make me think of I was chair of the Data Monitoring Committee for the breast cancer prevention trial, which involved 13,000 women, and the decision, the ultimate conclusion that Tamoxifen prevents breast cancer is based on only a few hundred women who have breast cancer.
So this notion that a small proportion --
DR. D'AGOSTINO: But if you got 13,000 people in this study, you would have a lot of events.
DR. FARO: Oh, right, yes.
DR. D'AGOSTINO: You know, so it gets circular.
DR. FARO: Oh, no. It's just the idea.
DR. D'AGOSTINO: Here we only have 400 people in the study.
CHAIRMAN RAMSEY: Okay. Well, let's go ahead and have you present. Well, go ahead.
DR. JOHNS: Okay. If you focus on, first of all, the number of patients, and we'll use the binary analysis because it's simpler. It puts it in two categories. So patients with minimal and mild adhesions at baseline.
Look first at row one in the INTERGEL group. There were 122 of these patients. One hundred and nineteen of that 122 remained in the minimal category. Three became moderate and severe.
In contrast, there's 117 patients in the minimal and mild category in the control population. Of those, ten became moderate and severe.
Now, if you look at the moderate and severe patients at baseline, yes, there's fewer patients there to begin with. There were nine in the INTERGEL group, all nine of which improved. There were 17 in the control group. Slightly more than half improved.
At the nd of the day, if you define moderate and severe adhesions as a treatment failure, and we'll have to ask the clinicians if they agree with that, and I think you've heard that, you've got a 13 percent failure rate in your treatment -- excuse me -- in your control population, and you can influence that to the level of two percent by use of INTERGEL.
Now, over half of those treatment failures in the control population came from patients who did not have a problem with adhesions at baseline. So it's those 117 patients from which the failures came from that constitute the majority in the control population.
CHAIRMAN RAMSEY: What I'd like to do, and, panel, you can disagree with me on this and I will retract, but one problem we're having, I think, is the two parties are presenting two different tables with slightly different analysis of the same data set, and it would be useful to me, and you can tell me no, but to have FDA comment on this table relative to what they presented to us to tell us whether they agree or disagree with what was just said.
And if you'd like to also present your table, sort of this summary argument and contrast it to it, I think that also would be helpful to us.
DR. HORBOWYJ: My table goes with sponsor's table, which is Table 3.1, which was given to the panel in like review.
CHAIRMAN RAMSEY: Can you give us a moment to find that table?
DR. HORBOWYJ: Yes.
PARTICIPANT: Dr. Horbowyj, was it in your presentation packet?
DR. HORBOWYJ: In my presentation packet, which you now have, of the overheads with notes on page 32 is the information that I presented this morning.
MR. WEINSTEIN: The new overheads are overheads that were in the panel pack.
CHAIRMAN RAMSEY: The ones that we just --
DR. HORBOWYJ: They were the overheads that were handed out just now.
CHAIRMAN RAMSEY: You know, did we actually get -- I don't think we got the -- no, we didn't get the slides that you --
DR. HORBOWYJ: Apparently some people did.
CHAIRMAN RAMSEY: Get that, yes. Okay. I hear what -- okay. Go ahead.
DR. HORBOWYJ: In any case, as the sponsor just said, it's somewhat difficult for me to read from here, but there are nine patients out of 131 in the INTERGEL group who had moderate to severe adhesions at baseline and 17 in the control group out of 134, when then nine minus 17 is eight, and that is where my comment as to the difference of eight comes from.
DR. CARLSON: You're saying at baseline there's a difference of eight patients --
DR. HORBOWYJ: Right.
DR. CARLSON: -- in terms of who had this particular severity of adhesions.
DR. HORBOWYJ: Right. Eight more control patients had moderate to severe adhesions at baseline compared to cohort of INTERGEL.
DR. CARLSON: Is that a significant difference?
DR. HORBOWYJ: I don't know.
DR. RUBIN: Well, it certainly shouldn't be because that's randomized. It's just noise. You're looking at noise.
CHAIRMAN RAMSEY: No, I think the issue is though the --
DR. HORBOWYJ: I'm looking at the magnitude of the numbers.
DR. D'AGOSTINO: The magnitude. I don't think -- the statistical significance argument, I guess, is over, and now the question is: is there enough in the numbers to make us happy that there's something clinically going on?
DR. HORBOWYJ: Right. And so at the end, at second look in the evaluable population as the sponsor presents -- and I agree. I'm not trying to be at all critical. I'm just presenting the information as it was presented to us -- and as you have said, there were three INTERGEL patients and 17 control patients who had moderate to severe adhesions, a difference of 14, and the only thing that I'm saying is that at baseline, the difference was that eight more patients had moderate to severe adhesions in the control group, and at second look there were 14.
But then you're comparing eight and 14, and if we're saying that the groups were comparable and eight was not important statistically, was it clinically significant?
And then if we're saying 14 now is statistically significant, well, clinically is it significant?
I'm just asking to look at the whole picture at the beginning and at the end.
And so then that difference, eight and 14, is six.
DR. DeCHERNEY: If there's an absolute difference in the numbers and it's statistically significant, which means you're magnifying the numbers in some mathematical way, that makes it clinically significant.
DR. CHIACCHIERINI: Yes. I'm Richard Chiacchierini. I am Senior Vice President for Statistics for C.L. McIntosh. I'm a consultant to LifeCore. I have no financial interest in the company other than my fee for service basis, and I have been assisting LifeCore since January of 2000.
I also might add that I was the former Director of the Division of Biostatistics at CDRH and FDA for 12 years.
The way that the FDA has evaluated that data is totally incorrect, and the reason it's totally incorrect is because the net change from beginning the study to the end of the study is irrelevant because every patient had an opportunity to experience adhesions. So the fact that those nine didn't develop adhesions doesn't mean we should subtract them from the 17 who developed them.
And so that total difference, the fact that everyone independently had an opportunity to develop an adhesion, means that every patient who entered the trial had a risk, a potential risk of developing adhesions, and so making the difference of the before and after numbers is irrelevant.
DR. D'AGOSTINO: I get a lot of comfort in these discussions because it's nice to know that statistical significance is equivalent to clinical significance. That's what we oftentimes end these discussions with.
DR. FARO: Only if they move together.
DR. D'AGOSTINO: We have statistical significance. What you just said makes perfect sense for the statistics argument. The numerical is what is bothering some people.
DR. CHIACCHIERINI: But I'd like to add -- this is Dr. Chiacchierini again -- Dr. D'Agostino, in the June 2nd amendment, in response to the failed to follow up in the 16 patients, a number of sensitivity analyses were done which did exactly what you suggested in a sense that the first of which was a true intent to treat analysis, a rather conservative intent to treat analysis in which randomly the success rate for the control group was imputed to anybody who didn't have a second look, and it's conservative because the success rate for the treated group was higher.
When we did that, we did that 1,000 times to see what the frequency of times that we would observe a nonsignificant result, and the results are as follows.
The median p value for those 1,000 imputations was .006, and the upper 95 percent confidence interval was .0 -- .03. What that implies is nearly the entire population was below .05, and so no matter which way we slice the three and the 17 and change those numbers, we're still going to be statistically significant the vast proportion of the time.
DR. D'AGOSTINO: That was the question I was asking earlier today when I said what if it was 14, three and 14. I was not under the impression that you did a sensitivity analysis that really addressed that. I was under the impression that you did a sensitivity analysis in terms of different imputation schemes.
DR. CHIACCHIERINI: You are correct, but the opportunity was for all of those 16 patients to have shifted those numbers, and those numbers in some instances did come closer together, but out of the 1,000 imputations, the number of circumstances which resulted in imputations that resulted in non-statistical significance was a handful.
DR. D'AGOSTINO: I don't think there's a problem with the statistical significance.
CHAIRMAN RAMSEY: It seems to me, just to use an evidence based medicine term that I'm familiar with, is that what we're arguing with here is relative risk reduction versus absolute risk reduction as a result of the treatment, and I think we're clear on the relative risk reduction that's been presented, and I think we all understand.
Has anyone computed an absolute risk reduction? And then the corollary of that would be the number needed to treat.
The reason I'm asking that is that it might clarify for the panel what level of difference we're talking about here.
DR. CHIACCHIERINI: In the revised analysis in June 2000, there was an analysis that was done that used the Cochran-Mantel-Haenszel adjustment procedure that proceeded to compute an odds ratio or in this case a relative risk. The relative risk is a relative computation of risk of the ratio of the control group to the treated group. That relative risk was a fivefold reduction. The relative risk was .19.
And whether we adjust it for continent, whether we adjust it for adhesiolysis or whether we left it unadjusted, the relative risk from that analysis was a fivefold reduction from .19 to .18.
CHAIRMAN RAMSEY: Again, I think we understand that the relative risk is great, but the absolute risk, I think, is what we've been asking, the panel has been asking about, and I don't --
DR. BECKER: Would you like to hear from Dr. Gere diZerega, who --
CHAIRMAN RAMSEY: I'm sorry. We can't hear you.
DR. BECKER: I'm sorry. If you would like to hear from Dr. Gere diZerega, who designed the clinical trial, he is available.
DR. DeCHERNEY: In terms of absolute risk, those are absent changes that take place, and they're not in terms of relative risk.
CHAIRMAN RAMSEY: All right. Well, let me ask the panel. We focused on Question 1 which you could find here, which is whether the statistically significant differences between INTERGEL solution and control can be considered to be clinically significant.
If you don't have more questions on that issue, I want to move on to the second issue the second issue.
Is there someone you'd like to comment? No?
DR. BECKER: I'm sorry. I thought that there was still the opportunity to ask FDA to comment on clinical significance, and I thought that we were -- you were asking us to do the same.
CHAIRMAN RAMSEY: Let me ask the panel. Would you like to hear this individual? I'm sorry?
DR. KIM THORNTON: I don't have any reservation in hearing his comment.
CHAIRMAN RAMSEY: Okay. Why don't you go ahead.
DR. diZEREGA: My name is Gere diZerega. I'm a Professor of OB-GYN at the University of Southern California. I've been involved with this clinical trial since 1995.
It was at my hospital. Prior to 1995 we performed the pilot clinical trial which generated so much discussion.
I've been a consultant to the sponsor since the initiation of the clinical study.
And what I would like to do is directly comment on what I think is a very important part of the clinical significance, and I think one of the reasons that the shift table is informative to me as a clinician in talking to other clinicians, nothing about statistics, and what I'd like to draw your attention to is the far right-hand side of this slide.
If you notice, from the standpoint of view of someone who does reproductive surgery, what I want to do is make patients better and reduce failures. As Dr. Mastroianni, who was one of the authors of the AFS system, reported to us earlier, patients who received moderate to severe scores are failures from the standpoint of view of the likelihood of reproducing.
This is a fairly routine result in this kind of patient population. If there were fewer myelomectomies, as Dr. Shirk has suggested, this number would even be higher.
So the 13 percent failures with this kind of a mixed population is what we would expect, and I've been involved with these types of clinical trials since 1980, with the first clinical trial for adhesion prevention. This is an expected result.
What is special and unique here to me as a health care provider is the two percent. This is the reduction that I focus on as someone taking care of women. I can reduce that failure rate, that 13 percent, down to two percent with the use of this device, which certainly we found to be very safe in our clinical participation.
CHAIRMAN RAMSEY: Before we move on to the second question of safety, let me ask the FDA if they want to make any closing comments on this particular issue. No?
DR. HORBOWYJ: No.
CHAIRMAN RAMSEY: Okay. So let's move on to the second question, which is really safety. Do the benefits of the product outweigh the potential risks, including any risk of infection?
And I'll open it to the panel for questions.
DR. CARLSON: I have a couple of questions regarding the infection risk, which has been discussed ad nauseam, so I will try to make them brief. We've heard there are a lot of ways to look at the infections or the possible infections or the probable infections, and the numbers we've heard from the experts from the company are that there were three in the control group and three in the treatment group.
FDA tells us that there were possibly six in the treatment group and three in the control group.
Are those two numbers fundamentally different? Are they statistically significant?
And I'd be interested in both clinical and statistical feedback from any of our colleagues here.
DR. SHIRK: I don't think they're statistically significant.
DR. RUBIN: No, they're not.
DR. SHIRK: Thank you.
DR. SHIRK: That took care of that. What was the second part of it? That's it?
DR. CARLSON: Well, if they're not statistically significantly different, they're not likely to be clinically significantly different.
DR. SHIRK: Absolutely.
CHAIRMAN RAMSEY: I guess I'd like to know what the p value is from both the FDA on their calculations and from LifeCore on the infection rate.
DR. SHIRK: On the six versus three?
CHAIRMAN RAMSEY: In the tables, I think you're referring to p equals not significant, but the actual number isn't given. I mean, is it .06? Is it .5?
DR. BECKER: I can find it.
DR. FARO: I think the important point to understand in this type of surgery, to begin with, is a very low risk operation with regard to infectious potential. Most pelvic infections that occur in female patients undergoing pelvic surgery occur in those women where the vagina is opened to the peritoneal cavity because the microbes involved are usually endogenous microbes from the patient. It's not from the environment.
So I think that's why we're seeing a low number of infections, to begin with, and this material doesn't seem to be conducive to enhancing growth of bacteria, as was demonstrated in the rat study, which I think is an excellent study to look at for infection potential.
DR. CHIACCHIERINI: Mr. Chairman, we have the p values.
CHAIRMAN RAMSEY: Okay.
DR. CHIACCHIERINI: For the crude analysis of the ten versus four, a chi square with each correction gave a p value of .19. Had we used a Fisher's exact test, the p value would even be higher because it's a less sensitive test.
For the FDA's numbers of six and three, the p there is .5 using the same test.
CHAIRMAN RAMSEY: Okay.
DR. CHIACCHIERINI: Point, five.
CHAIRMAN RAMSEY: Yeah. Other questions from the panel or discussion? Sure, go ahead.
DR. SHIRK: I had one question, and that has to do with the statement of use and what we're really voting. That's basically to LifeCore basically how -- you know, since it's a labeling issue -- what patients would qualify for use of this, you know, device. Are we talking about only those patients who either have severe or moderate adhesions at the time of surgical findings or undergoing of a procedure like myomectomy? Are you talking about general use for everybody who has gynecologic surgery for, you know, for essentially reproductive stuff?
I mean, obviously you could open end this thing and infer that basically if it works for moderate and severe adhesions, it works for mild adhesions. So let's give it to 100 percent of the patients. Okay?
So I guess my question would be basically from a marketing standpoint and from a labeling standpoint, what are the indications for use?
DR. DeCHERNEY: I can answer that. Certainly anyone whose potential reproduction, you know, from a gynecologic standpoint is an issue. Myomectomies, if they're done in patients that want further reproduction, and there's kind of a trend in America now to do myomectomies on 44 year olds, which perhaps is passing; those patients I don't think would be candidates, but certainly the 34 year old who has a myomectomy.
Tubal surgery patients; lysis of adhesion patients, and the only group that I would add are women that have ovarian surgery, ovarian cysts removed early on in their life, also to preserve reproduction, and if you look at the INTERCEED studies, it's one of the most effective areas, is the use of wrapping the ovary and preventing adhesions since those patients have up to 40 percent chance of forming periovarian adhesions.
So that's pretty much the group that I would reserve use of this agent for.
DR. SHIRK: But that almost encompasses everybody we operate on in the 35 to 20 year old range, who wants to preserve fertility.
DR. DeCHERNEY: Well, that's true, but the majority of the three million cases, certainly 60 percent of them are hysterectomies, and if you went further on and had cancer surgery, the majority of the three million cases are not patients that are requiring future fertility.
DR. SHIRK: The question was just so the panel understood what the indications for use were going to be. Do you know what I mean?
DR. GORDON: Maybe I could add to that. I mean, the indication for use statement is very clear or the proposed indication for use statement, and so I think what you're talking about is what is the real clinical usage going to be.
But I think in our assessment we need to limit that judgment to what the sponsor has proposed. I mean, obviously it may or may not lend itself to other applications, but there won't be an opportunity to market other applications. Other studies would have to be done to support that.
So I think we're getting a little bit off target there.
DR. SHIRK: I don't know if we're getting off target because labeling and use become a major issue in the question. Do you know what I mean? In any panel meeting for, you know -- that you look at a PMA, you obviously look at the labeling issue.
So I think it's something that, you know, as a panel we have to address. I mean, it's something -- I mean, are we totally overlooking those issues? And since we basically are serving two functions today, basically just dissolving a dispute obviously, but secondly, looking at a brand new essentially PMA. Okay?
I mean, so it's no different than bringing a new PMA. A PMA obviously, since the indications have changed, you've not gone through the same process that we would go through if the general surgery plastics people would go through it or the OB-GYN panel goes through it. I mean, a lot of those issues are basically issues that the panel goes through.
And I don't think that as a panel we need to totally circumvent those issues. Do you know what I mean? Because we obviously have two jobs here, I think, you know, one, just dissolving the dispute on safety and effectiveness, and two, then basically looking at the device in total as basically okaying an entire PMA.
CHAIRMAN RAMSEY: Okay. You're free to respond if you'd like.
DR. GORDON: No, no. I agree, and I guess then the issue is: is this indication for use statement appropriate or should it be modified?
CHAIRMAN RAMSEY: Does anybody have any comments on changing the indication for use statement or the appropriateness of it?
CHAIRMAN RAMSEY: Okay. Let me open it up to just general questions or comments among ourselves about any aspect of what we've heard today unless you have another question you'd like to focus or comment on, point two, the safety issue. Any general questions or comments for ourselves or for FDA or the sponsor group?
DR. D'AGOSTINO: Can I go back to one? As the statistician on the panel, I laid out quite nicely, I think, or quite clearly hopefully what I think about the statistics. I'd love to hear what the panel members think about the clinical significance.
We've asked a lot of questions. Are we moving quickly to that discussion?
CHAIRMAN RAMSEY: Well, let me just say as a procedural note that we are going to -- I'm going to close this off momentarily and we're going to have an open hearing again, and then we will move on to the voting procedures.
DR. SHIRK: Yeah, from a clinical standpoint, I think, you know, again, I've got some questions as to how great the clinical value is. I don't think that there's any question that there's probably some clinical value in this thing, and it's statistically significant.
But as to is this magic stuff where you throw it in the belly and all the adhesions float away? I don't think that's true. Okay?
DR. DeCHERNEY: But that's nobody's contention.
DR. SHIRK: No, and so that the public -- but you know, it's only going to be physicians who use it and tell patients, "I put this stuff in and now all the adhesions we took down shouldn't come back." I mean, I don't think that's going to happen.
Obviously like in the other adhesive device that we use, there's going to be some improvement, but it's certainly, you know, not a magic cure for adhesions, and so I think, I mean, from a clinician's standpoint that's the way I look at it.
Like, yeah, there's an advantage, but how great an advantage, I think, is still yet to be demonstrated, and that will come from other studies.
DR. GONZALEZ: If I could just follow up on that because in one of the reports I was reading, there was a question. I don't know if it was FDA that raised it. It must have been FDA. I found it very fascinating because it said, "Well, does the INTERGEL prevent adhesions or does it delay the formation of them?"
And I can't remember who raised that. I want to say it was an FDA report, and so that question is really almost tangential to your issue as to whether long term, you know, what does it do one way or the other.
DR. KIM THORNTON: Well, personally, I think that if you have an agent that could have any clinical effect on reducing adhesions and you can show some sort of significance both statistically and clinically, then it may be of benefit to patients that are going to be using it.
I think that you need to make sure that from a labeling perspective, if -- I mean, you know, we can certainly set up guidelines where we can be restrictive, given to the population it's been studied in. Other studies will expand those indications if appropriate and if, you know, they demonstrate that it's clinically significant in other areas.
DR. DeCHERNEY: Actually Dr. diZerega's work and others have shown that the adhesions are practically formed within 48 to 72 hours. So if it deters adhesion formation immediately, it probably is a longstanding effect.
CHAIRMAN RAMSEY: Is there any more discussion among the panel about the two subquestions we have here or anything else before I move on to the public hearing?
It seems that the winds of time have changed and now we're doing well.
CHAIRMAN RAMSEY: But I don't want to cut us off if there's more discussion to be had.
Shall I reread the question? Okay. Let me reread the main question again for the panel because this is, in essence, what we're voting for.
Does the PMA, as amended, provide reasonable assurance of the safety and effectiveness of INTERGEL for its intended use as an intraperitoneal instillate for reduction of adhesion formation following gynecologic pelvic surgery?
Let me close with that question the section of panel discussion and now move on to the second open public meeting, and at this session interested persons can have an opportunity to address issues specific to the matter before the committee.
As we've been doing all day today, I would ask people addressing the panel to come forward and speak clearly into the microphone and also to state whether they have any involvement, including, but not limited to, financial interest in any medical device company, including LifeCore or any of its competitors, and also please state the nature of your interest, such as, for example, whether the company has paid you for your time or travel to appear here today.
MR. WEINSTEIN: Now, one person did respect to speak in the afternoon. It's Ms. Weatherman. Is she here?
CHAIRMAN RAMSEY: Let me just have a show of hands for all those who would like to comment. One other person.
Okay. There was an individual who presented this morning who wanted to show a video, Dr. Thornton. He's not here? Oh, there he is. Okay.
How long is your video?
DR. MELVIN THORNTON: About three and a half minutes.
CHAIRMAN RAMSEY: Okay. Very good. Go ahead.
MS. WEATHERMAN: Good afternoon. My name is Bess Weatherman. I'm a partner at Warburg Pincus, which is a venture capital firm based in New York, and I'm also the Vice Chair of the National Venture Capital Association Medical Industry Group.
Neither Warburg Pincus nor myself personally nor any of our portfolio companies, to my knowledge, have any stock ownership in the sponsor or are receiving any remuneration whatsoever for speaking here today.
Warburg Pincus is the largest independent venture capital firm in the world with assets under management of ten billion, and I'm the partner responsible for investing a portion of those assets in medical device companies.
The National Venture Capital Association is the largest venture capital association in the world and represents approximately 450 of the largest venture capital firms in the U.S., the collected assets under management of over 300 billion, and annual venture investments of over 50 billion.
The NVCA, the National Venture Capital Association, members who actively invest in drugs, medical devices, and biotechnology have invested over six billion in these areas in the last year alone and over 40 billion over the last ten years.
We are the largest single source of development capital in the United States for biotechnology and medical devices. At any given time, the venture capital industry is responsible for the management of over 400 medical device and biotechnology companies, and over the past 20 years have founded more than 3,000 such companies.
In turn, these companies were responsible for developing or introducing virtually all of the drugs derived from biotechnology, and most of the groundbreaking new medical devices introduced during the past 20 years to the benefit of millions of patients in the United States and around the world.
I have followed the INTERGEL dispute over the last year and appreciate the opportunity to offer my views.
First, the venture capital community's willingness to fund medical device and drug development ventures depends critically, if not primarily, on consistency and predictability in the regulatory process. Regulatory decisions that are unpredictable or surprising add substantial additional risk to our investments, which are already extremely risky.
An increase in this level of risk reduced the flow of investments into this area. A significant operating assumption of those of us who invest in medical technology has been that statistically significant results of blinded, multi-center, randomized, placebo controlled trials was the regulatory gold standard and supported the strongest assumption of approvability in the absence of extraordinarily persuasive evidence of lack of efficacy or clear safety problems.
In the past, the clear legislative and regulatory policy has always been to approve such technology and let the medical community decide, the physicians, whether or not its ultimate fate, whether to use the device, therefore insuring a strong flow of new venture technology to the U.S. public.
The position of the FDA in this dispute represents a marked departure from this policy. By rejecting the successful results of a level one clinical trial without the clear, contrary, scientific evidence historically required to do so, the FDA is setting a very negative precedent.
In the past, similar adverse changes in regulatory practice have caused substantial reductions in investment in new mechanical technology and, therefore, their proliferation for the patients who need them.
Another important issue today is the effectiveness, fairness, and independence of this dispute resolution process. The establishment of the medical device dispute resolution panel was a long sought goal of the medical venture capital community, which was fully achieved in the FDA Reform Act passed in 1997.
While it is obvious that such a forum must exist in the interest of basic fairness and due process, how it is actually implemented and whether its judgments are perceived as reasonable and independent will be the acid test of whether it fulfills its charter.
These panels must be more than mere advisory panels whose role is largely controlled by the FDA. They must fulfill their clear legislative mandate and independently decide whether the probable benefit to health outweighs the probable risk for a subject device or drug.
Therefore, the independence of your decision here today and the procedures you follow will be widely examined and may significantly affect the continued development of many new medical devices.
CHAIRMAN RAMSEY: Thanks for your comment.
Can we have the next presenter, please?
DR. MARTENS: Hello. My name is Mark Martens. I'm Chairman of OB-GYN at Franklin Square Hospital Center and Director of Women and Children's Services for MedStar in Baltimore.
I wasn't planning on speaking today. So I hope I don't --
CHAIRMAN RAMSEY: Please state affiliations.
DR. MARTENS: Oh, yeah. I feel embarrassed after the last presentation. I bought 100 shares of stock three or four years ago before I even heard about INTERGEL.
I drove myself down here. I paid my hotel bill a little while ago. I didn't receive any money for the consultant meeting they had before, and in the last FDA presentation, as promised I donated my fee to an education company in Minnesota.
So I'm here as a patient advocate, and I hope the comments of the first physicians who were up here, whom I've never met before, aren't forgotten.
As Chairman, and some of my colleagues on the panel know, as Chairman, especially as previous Director of Gynecology, they think that we're magical surgeons. So I get the most difficult cases. They come to me. They're always adhesion cases. They're always chronic pelvic pain, and no one wants to do anything because the patients -- you know when you operate on them and do adhesiolysis that the adhesions are going to come up unless you have something.
I've also done 125 studies. I believe I was told I was the largest enroller on the last product that was approved by the FDA, and I didn't come here for that product. I'm here today because if you're going to use this pilot study, I've enrolled more patients in double blind, randomized trials for adhesiolysis than in that pilot study, and I saw a clinical difference with this product.
I currently -- the reason I'm here is as a patient advocate. What I currently do is I tell my patients that the best way to have this problem treated, severe adhesions and prevention of adhesions, is to go to Canada. I do that right now.
And if they can't afford to go to Canada, I get compassionate use, and I've done that. I feel that strongly about this.
Unfortunately, compassionate use may not be available much longer as mentioned with the previous conversation, and I'm now setting up a hierarchy, and I'm doing it, not you all, but I'm setting up a hierarchy where rich patients get care and poor patients don't.
I'm sorry to get a little sappy, but you know, I graduated from George Washington University 20 years ago, and we sat through the Hippocratic oath and I didn't take it seriously. I'm taking it seriously now.
Patients who have it don't have a life. They have pain. They have hysterectomies when they're 20 years old, and we need to do something about it. So I take this very seriously.
Now, clinical data, let me get focused again. I don't want to do harm to patients. So I will not operate on a patient if I'm going to tell them, "Adhesions may come back. You may have problems again," and so that's a problem, but I do sometimes.
The clinical data here I think is very strong. I thought it was all solved, that I wasn't going to get up and talk because I thought clinical significance is as plain as English. It seems like statistics is multi-lingual, and we didn't get very clear today.
But I'm trained to clinically evaluate patients and publications. I've written over 100 publications. I'm on the editorial board.
This paper, these data have been published twice in peer reviewed journals. So I think the data are strong.
As a clinical investigator, again, I've seen these trials over and over again. Dr. Shirk, these are very difficult studies to do. I hate doing these studies because I'm telling patients that I'm going to do a second surgery on them. The only way I can get it through my mind is to say that there's good data saying that if you want a baby and you have adhesions, lysing of adhesions at a second look improves your outcome, and that barely gets me by my ethics to say, "Do these studies."
You can't do these studies any better than this study was done, and my patients who felt better, I didn't do a second look on if they didn't want to because their pain was gone, and I didn't want to subject them to that pain.
These patients are given a score in the ITT that says they did horribly, and it's costing them and it's cost against the product. The whole idea that we're trying to get these patients pregnant, the only reason I do these studies is because they way they want to get pregnant. Having their pregnancy in the trial would cost against the product. So this ITT analysis really disturbed me.
And, again, six out of the seven patients didn't have a second look were in the treatment group. So I mean, I think that proves something to me.
So ethics, the ethics and the clinical values, its clinical efficacy, hopefully with statistical efficacy, the clinical efficacy is what doctors do to patients, and what I do because I've seen this product; I've used this product; because I've seen patients like this, but unfortunately I'm forced to set up pelvic pain centers. A pelvic pain center is about as profitable, as happy for the medical staff as PMS centers. I mean, they're difficult patients. We don't have cures for them.
But if you take the worst case scenario where someone said a change in the baseline score from 2.-something down to something and it was .19 or two; if you take ten percent out of 400,000 patients that are being operated on every year, two million that have these type of surgeries, I'll take that ten percent any time.
So I think there's a lot of clinical significance to that, and I know you, too, in your practices see patients like this.
So basically the concerns about infection, I think, have been addressed. I would be vigilant about infections. I think labeling should address the possibility of infections, but the numbers weren't significant. I hope significance is important.
So thank you for allowing me to speak at this meeting today.
CHAIRMAN RAMSEY: Thank you. Thank you.
Well, I understand there's one -- yes? Okay, great. So you are going to present a video, yes?
DR. MELVIN THORNTON: Yes. First off, thank you for the second opportunity. My video wasn't working earlier.
But what I wanted to do is there's a lot of talk about statistical significance. We talk about is it clinically significant, and what I'd like to do is to show you what moderate-severe adhesions are in patients who are just presenting for complaint of infertility, and their only history is significant for prior surgery.
So may I have the first slide?
What you're going to see here, this is a young woman who presented to our hospital. Her chief complaint was infertility. She had a history of ovarian surgery. She had a cyst removed, and what you can see here is a lot of adhesions between the bowel and the anterior abdominal wall, and this pointer is not working, but that is her right ovary where she had her surgery from.
That is moderate to severe adhesions. If you can give me a reduction, if you can give me, although out of my patients only two percent of those will have that, I'll take that any day. Okay?
But 13 percent of the patients it has been shown will have these moderate-severe adhesions. If you can reduce it down to two percent, like I say, I'll take it any day.
And this is another common surgery that women undergo. This is a myomectomy, and what you're seeing here with the myomectomy, once again, you're going to see bowel attached to anterior abdominal wall, and right here, inside of the adhesion, severe adhesions here, is her right ovary and her right fallopian tube.
Once again, her chief complaint was infertility. Her only history was a previous myomectomy. This is what we're talking about when we talk about clinically significant or when we take patients from the moderate to severe category down to the mild to minimum. To me that is very clinically significant, and it's what our goal of being a physician is, is to take care of our patients.
CHAIRMAN RAMSEY: Thank you.
Again, we are well ahead of time, and what I think I'm going to do is have us take a ten minute break. I would like the panel to collect their thoughts. We won't be talking among ourselves, but just for a moment to collect their thoughts before we come back and vote.
Okay. So, please, ten minutes only. Come back at 2:30.
(Whereupon, the foregoing matter went off the record at 2:19 p.m. and went back on the record at 2:31 p.m.)
CHAIRMAN RAMSEY: Those of us who live on the left coast have long plane rides that we're hoping to catch.
During the break I was informed that a member of the public who did raise her hand and wished to speak, we just didn't see her. So --
MR. WEINSTEIN: I think it's Augusta Sisler.
CHAIRMAN RAMSEY: Yes. So I'd like to give her a moment before we move on to the deliberation and vote.
And please state your name and your interest.
MS. SISLER: Hi. My name is Augusta Sisler. I'm a volunteer with the International (inaudible) Society.
I'm also a patient who has had five surgeries for adhesions, and I haven't worked in a couple of years due to it. And I'm here to promote awareness of this issue or issues.
So if you have any questions to ask me, how it feels, I can tell you for a fact.
The doctor that just got up recently, I could hardly -- as he was talking, I could hardly hold the tears back because I know what he feels, the way he feels for his patients.
This affects thousands of people. It's not only men -- I mean lots of women. When the adhesions hit your bowel or, you know, the small bowel and your reproductive system and you can't eat because of the adhesions are restricting and pulling, there's got to be something out there.
I'm not here to support, you know, any of the groups. I'm just here for patient awareness, and I'll tell you it's really scary when you can't work. You can't eat.
And I applaud all of you doctors out there that are helping us, and I hope that the government will also help. So let's get to work.
CHAIRMAN RAMSEY: One question. Thank you for commenting.
CHAIRMAN RAMSEY: I'm sorry to ask you this, but it's just for a point of record, whether you have any financial interest in the company or --
MS. SISLER: No.
CHAIRMAN RAMSEY: -- were paid to come here by any --
MS. SISLER: No, sir.
CHAIRMAN RAMSEY: Okay.
MS. SISLER: Nope, I have never been paid to do anything and am just here to promote a witness.
CHAIRMAN RAMSEY: Thank you for your comments. We appreciate them.
MS. SISLER: Thank you.
CHAIRMAN RAMSEY: Okay. I am now going to close the public hearing and move on to the panel deliberations and vote portion of the meeting.
I'm going to begin by asking any panel members have any comments or summary statements they would like to make before we move on to the next part.
CHAIRMAN RAMSEY: No. Okay. In that case I'm going to turn it over to Les who's going to -- I'm sorry. Back up.
I want to give the sponsor and the FDA a chance to have a closing argument, and we'll begin with FDA.
DR. WITTEN: We don't have any additional comments.
CHAIRMAN RAMSEY: No additional comments. Okay.
Would the sponsor like to have any closing comments?
DR. BECKER: Yes, just briefly to say that we are gratified with the fair forum and the conclusions regarding the statistical significance of the study. We feel that the testimony you've heard is quite compelling that the product is safe and that the benefits outweigh the risks.
CHAIRMAN RAMSEY: Okay. Well, I'll turn it over to Les Weinstein who will give us the options for voting.
MR. WEINSTEIN: These are the panel voting options.
Medical device premarket approval application, PMA, must stand on its own merits, and the panel's recommendation must be supported by safety and effectiveness data in the application or by applicable publicly available information.
Safety is defined in the federal Food, Drug, and Cosmetic Act as reasonable assurance based on valid scientific evidence that the probable benefits to health on the conditions of intended use outweigh any probable risks.
Effectiveness is defined as reasonable assurance that in a significant portion of the population the use of the device for its intended uses and conditions of use when labeled will provide clinically significant results.
The panel's recommendation options for the vote are as follows:
One, approval. There are no conditions attached.
Two, approvable with conditions. The panel may recommend that the PMA be found approvable subject to specified conditions. Prior to voting each of the conditions should be discussed and voted on by the panel.
Three, not approvable. The panel may recommend that the PMA is not approvable if the data do not provide reasonable assurance that the device is safe or, if a reasonable assurance has not been given, that the device is effective under the conditions of use prescribed, recommended or suggested in the proposed labeling.
Following the voting, the Chair will ask each panel member to present a brief statement outlining the reasons for his or her vote.
This statement that I just ready, by the way, panel, is in the information that I gave to you last night. So you do have a copy of it for your reference.
CHAIRMAN RAMSEY: Okay. Now we will vote on a recommendation to FDA's Director of the Center of the Devices and Radiological Health as to how the scientific dispute regarding the approvability of this PMA should be resolved.
The recommendation, as Mr. Weinstein said, was approval, approvable with conditions, or not approvable. Would one of the panel members like to make a motion?
Someone has to.
DR. D'AGOSTINO: I move for approval.
CHAIRMAN RAMSEY: Do we have a second to that motion?
DR. KIM THORNTON: I'll second that.
CHAIRMAN RAMSEY: In that case, I'm going to go and ask each voting member to make their individual vote --
MR. WEINSTEIN: Excuse me. You need to discuss the motion first, right?
CHAIRMAN RAMSEY: We need to discuss?
CHAIRMAN RAMSEY: Okay. I'm sorry. Procedural mistake. We'll move on to discussion of the motion from the panel.
Do you have a comment?
DR. GONZALEZ: Well, I don't know if it's a comment or if it's a condition or how maybe the panel can help me with it, but in sifting through all the information there were two things that concerned me. And I'm speaking now as a consumer representative.
And one, you know, how to do with that question I raised earlier about the prevention versus the delay of adhesions. So somewhere that troubled me about whether we should have some condition, some caveat that could be worked out between the sponsor and the agency to ferret that information out as a possibility.
The other part is the infectious issue, and for that part, I had the question about the option of either tracking the possibility of infection in those cases or something of that type. As I say, those are just the two larger issues. I really don't know how to couch them in terms of a final recommendation, however. Maybe the panel can help me with that.
CHAIRMAN RAMSEY: Are you suggesting that we attach conditions to --
DR. GONZALEZ: That's basically what I'm trying to do, I think.
CHAIRMAN RAMSEY: Okay.
DR. GORDON: Could I ask a question? Is the condition that you want definition or separation of the data sets? Because I think that if you look at the indication for use, it's quite clear in saying reduce the likelihood of developing moderate or severe, which was reflected in the shift data, and reduce adhesion reformation.
I'm wondering if you want additional clarity around that or I think it will be helpful to FDA and the sponsor if we can be more specific in what's needed because to put a condition on a recommendation, I think we need to give some specific -- I think we need better clarity around what you would like that condition to be.
CHAIRMAN RAMSEY: I have a procedural question here. Dr. Gonzalez is not a voting member.
DR. GONZALEZ: No.
CHAIRMAN RAMSEY: And I'll go around and ask whether someone would entertain a motion to approve with conditions. Do I need to ask a voting member whether they would?
DR. SHIRK: Nancy, do you want to weigh in?
CHAIRMAN RAMSEY: Right now there's a motion on the floor.
DR. GONZALEZ: You have a main motion, and I'm just raising the question. I was talking, you know, about those two issues.
DR. GORDON: And I'm asking for additional clarification before you go to add a condition, meaning what exactly is the condition.
CHAIRMAN RAMSEY: I think what we need to do is to vote, and then if it is not voted as approved, as the motion stands, then we will go back and see if there is a motion for approvable with conditions and then discuss that. Is that acceptable?
Because we have a motion that we have to resolve with a vote right now.
DR. CARLSON: On the other hand, the implication of his point might affect the vote. So personally I would like to hear if he has anything more to say about the conditions because it affects the current vote, not because --
CHAIRMAN RAMSEY: And that's fine, and if you have more comments you'd like to make along the line --
DR. GONZALEZ: The only other piece I would add, and I don't know how to phrase it except that I made the comment earlier. I said that maybe we could have something that one of the conditions would be that the panel recommends with the approval part that the agency and the sponsor may develop protocols to track the delay and the prevention of adhesions and the instances of infection.
CHAIRMAN RAMSEY: Okay. Let me move on and vote on the original motion, and if it is not approved as the motion is stated, then we'll go back to that.
So I'm just going to go around the room starting with Dr. Thornton to ask to vote on the motion for approvable. In favor or opposed?
DR. KIM THORNTON: In favor.
CHAIRMAN RAMSEY: Okay. Dr. Shirk?
DR. SHIRK: Yes.
CHAIRMAN RAMSEY: Dr. Carlson?
DR. CARLSON: Approve.
DR. D'AGOSTINO: Yes.
CHAIRMAN RAMSEY: Dr. D'Agostino.
So we have a quorum.
CHAIRMAN RAMSEY: We have a majority. So we have a unanimous vote in favor of approval, and the next step in the procedure is to ask each voting member to give a reason for their voting for approval.
So, again, I'll start with Dr. Thornton.
DR. KIM THORNTON: I think in answer to the questions that were posed, first, it's my opinion that in terms of demonstrating statistical significance in the study that the sponsor did that, and that the statistical significance was also indicative of clinical significance.
The safety data with regards to infection, I think it was clear that there wasn't an increased incidence in infection in the study group as compared to the control group. So as a result of that I think that the data that was presented here answered the questions that were posed and warranted approval.
CHAIRMAN RAMSEY: Okay. Dr. Shirk.
DR. SHIRK: Well, I guess I voted yes because I think it was statistically significant, I thought. How clinically significant it is, I think, is still in question, and the magnitude of the clinical significance, but certain clinical significance follows statistical significance.
I do think it's safe. I think the study could have been much more well managed from the beginning. It was obviously confused by both a pilot study and also by the fact that you had multiple parameters involved with this, and I think that's created some fairly marked confusion on the review of this process.
CHAIRMAN RAMSEY: Dr. Carlson.
DR. CARLSON: I'll be frank and brief in that coming into this based on the data I had up till this morning I was not inclined to approve even with conditions. I would congratulate the FDA and the sponsor on their very excellent and clear presentations.
I had specific questions each of which were answered to my satisfaction regarding the endpoint, which I think was agreed to prior to the study and which, though not perfect and though intermediate, appears to be the best available endpoint we have now.
The combinability I had questions about, and I was satisfied with the answer there. The risk for infection was answered appropriately, and in regards to the analysis, I thank all of the statisticians, particularly Dr. D'Agostino, regarding the, quote, unquote, retrospective analysis and the, quote, unquote, subgroup analysis.
That left clinical significance, and although the absolute numbers aren't as great as I think anyone other than maybe the sponsor's competitors would like to see, I think we're facing a situation in medicine today where we've made tremendous strides in the past century, and the strides that we make are going to be incremental. It makes it much more difficult to assess what's effective and what's not, and I congratulate everybody on their analysis.
DR. D'AGOSTINO: I also want to thank the sponsor and the FDA for the thought process and the time they spent and the wonderful presentations they made.
I was quite a virgin, you know, to the particular problem and read through the materials, and the discussion and the transcript of the previous meeting left me completely confounded in terms of how I should go, but I think as we saw materials develop, sent to us, and then the sponsor's material, the FDA's material, and the unfolding today, I think what we have is a very powerful study. Forget the expense of it. That's probably not what we should be worrying about, but a multi-national study that really did have to me a lot of consistency.
And I understand where the FDA's objections came from, but I think that they were addressed quite correctly and substantially by the sponsor, and I think the data from a statistics point of view, given all of the rigor, the imputation methods and so forth, that there's a lot of robustness there, and we can feel very comfortable about it.
With the clinical significance, as I said, I've lived through more and more of these situations where statistical significance becomes equivalent to clinical significance because we don't seem to want to bite the bullet on that, but even though the numbers are small, I mean, this was a pool of subjects that was evidently quite hard to deal with and so forth, and there is substantialness to this.
And I was playing some of the games and I was asking what if that 17 became 15 and so forth, and even doing some approximate analysis. They would have to come down pretty far before you'd lose the significance.
So I hope the sponsor doesn't have on television after the Claritin ads that there's a fivefold improvement.
DR. D'AGOSTINO: But there is an improvement, which I buy and really believe in.
CHAIRMAN RAMSEY: Okay. Now, the voting members have had a chance to have their say, but we also would like the consumer and industry representatives, if they are so inclined, to comment. So, Dr. Gordon, would you like to comment first?
DR. GORDON: I would like to comment because I think these forums are so important. I've spent most of my professional career because I believe in this process, the regulatory review and the quality of that review, and I've had consistently good experiences with this, and again, I commend everyone's effort in seeking a balanced approach.
But I'm really encouraged that this forum has played out the way it has because I think it could be an excellent resource for companies, although certainly one of last resort because there are many opportunities for interaction within the agency before one undertakes something that's this resource intensive.
But I think that this has been a really, really thoughtful forum, and everyone had really great comments. And I think if companies are considering it, then it's certainly worth undertaking.
CHAIRMAN RAMSEY: Thank you.
And Dr. Gonzalez.
DR. GONZALEZ: From a personal perspective, one of the things that I always enjoy about coming to these meetings is that, you know, we have always had hard working folks on whatever side of the table they are, whatever side of the aisle they are, whatever you want to call it, and that that hard work is only, you know, reflected in making -- in my case as a panel member, making me a much better person, a much more knowledgeable individual with some of the issues that come forth.
So we also get a tremendous amount of education, plus in addition to the hard work we put in.
So second, I do want to congratulate both FDA and the sponsoring agency for their hard work in making this presentation on both sides of the fence.
CHAIRMAN RAMSEY: Okay. I have some privilege as Chairman to make a closing comment, and I've strived mightily not to show any partiality to one party or the other, and I'm going to continue to do that in the closing comments, but just one comment on this panel and the process.
It is, I think, a panel that will be used as last resort, and I also should comment that this is merely a recommendation to the Director, not a binding rule, but I do hope that the Director will take the strength of the consensus very seriously in their decision ultimately on this device.
As I said, I think one of the most important parts of this panel is to have impartiality and to give both sides a chance to present their views, and I hope both parties feel that they had an adequate chance to represent their views and be heard.
If that's not the case, I personally would like to hear those opinions.
So let me just move on and close and say that this concludes the Medical Devices Dispute Resolutions Panel review of scientific issues in dispute between LifeCore Biomedical, Incorporated, and FDA regarding PMA 990015 as amended for INTERGEL adhesion gel prevention solution.
On behalf of the panel or on behalf of the FDA I'd like to thank the panelists, and I would also like to thank the sponsor and the FDA for what I know was a tremendous amount of effort on their behalf to come to this meeting.
We're now adjourned.
(Whereupon, at 2:54 p.m., the meeting was concluded.)