FOOD AND DRUG ADMINISTRATION
ONCOLOGIC DRUGS ADVISORY COMMITTEE
- - -
Monday, September 10, 2001
Holiday Inn - Bethesda
8120 Wisconsin Avenue
STACY NERENSTONE, M.D., Chair
Associate Clinical Professor
Oncology Associates, P.C.
Helen & Harry Gray Cancer Center
85 Retreat Avenue
Hartford, Connecticut 06106
KAREN M. TEMPLETON-SOMERS, PH.D.
Advisors and Consultants Staff, HFD-21
Food and Drug Administration
5600 Fishers Lane
Rockville, Maryland 20857
KATHY ALBAIN, M.D.
Professor of Medicine
Division of Hematology/Oncology
Loyola University Medical Center
Cancer Center, Room 109
2160 South First Avenue
Maywood, Illinois 60153
DOUGLAS BLAYNEY, M.D.
Wilshire Oncology Medical Group, Inc.
50 Bellefontaine Street, Suite 304
Pasadena, California 91105
JOHN T. CARPENTER, JR., M.D.
Professor of Medicine
Division of Hematology and Oncology
University of Alabama at Birmingham
1530 3rd Avenue South
Birmingham, Alabama 35294-3280
STEPHEN L. GEORGE, PH.D.
Professor of Biostatistics
Department of Biostatistics and Bioinformatics
Hanes House, Room 219
Trent Drive at Erwin Road
Duke University Medical Center
Durham, North Carolina 27710
COMMITTEE MEMBERS: (Continued)
DAVID P. KELSEN, M.D.
Chief, Gastrointestinal Oncology Service
Memorial Sloan-Kettering Cancer Center
1275 York Avenue
New York, New York 10021
SCOTT M. LIPPMAN, M.D.
Professor of Medicine and Cancer Prevention
The University of Texas M.D. Anderson Cancer Center
Department of Clinical Cancer Prevention
1515 Holcombe Boulevard, HMB 11.192c, Box 236
Houston, Texas 77030
JODY L. PELUSI, F.N.P., PH.D., Consumer Representative
Phoenix Indian Medical Center
4212 North 16th Street
Phoenix, Arizona 85016
DONNA PRZEPIORKA, M.D., PH.D.
Stem Cell Transplant Program
Center for Cell and Gene Therapy
Baylor College of Medicine
6565 Fannin Street, M964
Houston, Texas 77030
BRUCE G. REDMAN, D.O.
Associate Professor of Internal Medicine
Division of Hematology/Oncology
University of Michigan Comprehensive Cancer Center
7216 Cancer Center
1500 East Medical Center Drive
Ann Arbor, Michigan 48109-0948
GEORGE W. SLEDGE, JR., M.D.
Departments of Medicine and Pathology
Indiana University School of Medicine
Indiana Cancer Pavilion
535 Barnhill Drive, Room 473
Indianapolis, Indiana 46202
MARION EVERETT COUCH, M.D., PH.D.
Department of Otolaryngology, Head and Neck Surgery
Johns Hopkins Medical Center, JHOC
601 North Caroline Street, Suite 6252
Baltimore, Maryland 21287
BONNIE GLISSON, M.D.
Section Chief, Head and Neck Medical Oncology
Department of Thoracic and Head and Neck Medical Oncology
University of Texas M.D. Anderson Cancer Center
1515 Holcombe Boulevard, Box 80
Houston, Texas 77030-4009
LAWRENCE V. RUBINSTEIN, PH.D.
Head, Developmental Clinical Trials and
Preclinical Studies Section
Biometric Research Branch
National Cancer Institute
6130 Executive Boulevard, EPN Room 8130
Rockville, Maryland 20852
VOTING PATIENT REPRESENTATIVES:
NON-VOTING INDUSTRY REPRESENTATIVE:
GEORGE H. OHYE
3 Heritage Hills Court
Skillman, New Jersey 08558-2340
FOOD AND DRUG ADMINISTRATION STAFF:
GREGORY FRYKMAN, M.D.
RICHARD PAZDUR, M.D.
RAJESHWARI SRIDHARA, PH.D.
ROBERT TEMPLE, M.D.
GRANT WILLIAMS, M.D.
MATRIX PHARMACEUTICAL, INC. REPRESENTATIVES:
LAURENCE ELIAS, M.D.
STEPHEN B. HOWELL, M.D.
GLENN MILLS, M.D.
RICHARD D. LEAVITT, M.D.
JOHN MACKOWIAK, PH.D.
MORGAN E. STEWART, PH.D.
ROBERT TRESSLER, PH.D.
EVERETT E. VOKES, M.D.
BARRY L. WENIG, M.D., M.P.H.
RICHARD W. CURRY
EDWARD F. McCARTAN
C O N T E N T S
IntraDose (cisplatin/epinephrine) Injectable Gel
Matrix Pharmaceutical, Inc.
AGENDA ITEM PAGE
CONFLICT OF INTEREST STATEMENT
By Dr. Templeton-Somers 9
OPEN PUBLIC HEARING
By Edward F. McCartan 10
By Richard W. Curry 13
By Kim Thiboldeaux 17
By Ian Stewart Findlay (video) 20
MATRIX PHARMACEUTICAL PRESENTATION
By Dr. Stephen Howell 23
Current Management of Head and Neck Cancer
By Dr. Glenn Mills 25
Pharmacologic Rationale and Challenges Associated
with Demonstration of Clinical Benefit
By Dr. Stephen Howell 28
Clinical Study Results, Efficacy, and Safety
By Dr. Richard Leavitt 36
By Dr. Stephen Howell 53
By Dr. Glenn Mills 25
Questions from the Committee 60
C O N T E N T S (Continued)
AGENDA ITEM PAGE
By Dr. Grant Williams 123
By Dr. Gregory Frykman 129
By Dr. Rajeshwari Sridhara 143
By Dr. Grant Williams 157
Questions from the Committee 160
COMMITTEE DISCUSSION AND VOTE 188
DR. NERENSTONE: Good afternoon. I think we're going to get started in the afternoon session. From our agenda it's going to be discussion of the cisplatin/epinephrine injectable gel.
We'd like to start by going around the table, if everyone could introduce themselves and tell us where they're from. Dr. Glisson, if you'd like to start.
DR. GLISSON: Bonnie Glisson, M.D. Anderson Cancer Center in head and neck medical oncology.
DR. KELSEN: David Kelsen, Sloan-Kettering.
DR. ALBAN: Kathy Alban, medical oncology, Loyola University, Chicago.
MR. GRUETT: Glenn Gruett, patient representative from Appleton, Wisconsin.
DR. LIPPMANN: Scott Lippmann, M.D. Anderson Cancer Center.
DR. CARPENTER: John Carpenter from the University of Alabama at Birmingham.
DR. PRZEPIORKA: Donna Przepiorka, Baylor College of Medicine, Houston.
DR. NERENSTONE: Stacy Nerenstone, medical oncologist, Hartford, Connecticut.
DR. SLEDGE: George Sledge, medical oncologist, Indiana University.
DR. PELUSI: Jody Pelusi, oncology nurse practitioner, Phoenix Indian Medical Center, and I'm the consumer rep.
DR. RUBENSTEIN: Larry Rubenstein, biostatistician, National Cancer Institute.
DR. REDMAN: Bruce Redman, University of Michigan Cancer Center.
DR. COUCH: Marion Couch, head and neck surgery, Johns Hopkins Hospital.
DR. BLAYNEY: Doug Blayney, medical oncologist, Wilshire Oncology Medical Group, Pasadena, California.
DR. SRIDHARA: Raje Sridhara, FDA.
DR. FRYKMAN: Gregory Frykman, medical officer, FDA.
DR. WILLIAMS: Grant Williams, medical team leader.
DR. PAZDUR: Richard Pazdur, division director, FDA.
MR. TEMPLE: Bob Temple, office director, FDA.
* DR. TEMPLETON-SOMERS: The following announcement addresses the issue of conflict of interest with respect to this meeting and is made a part of the record to preclude even the appearance of such at this meeting.
Based on the submitted agenda and the information provided by the participants, the agency has determined that all reported interests in firms regulated by the Center for Drug Evaluation and Research present no potential for a conflict of interest at this meeting with the following exceptions. Stephen George, Ph.D., and Sarah Taylor, M.D., are recused from participating in the discussions and vote concerning IntraDose.
In the event that the discussions involve any other products or firms not already on the agenda for which FDA participants have a financial interest, the participants are aware of the need to exclude themselves from such involvement and their exclusion will be noted for the record.
With respect to all other participants, we ask in the interest of fairness that they address any current or previous financial involvement with any firm whose product they may wish to comment upon.
DR. NERENSTONE: We're going to turn to our open public hearing. Mr. McCartan.
* MR. McCARTAN: Good afternoon. Can I be heard in the back? I assume so. My name is Ed McCartan. First I want to thank the FDA for giving me the opportunity to appear before this distinguished panel.
I'm here as a 15-year survivor of head and neck cancer and to advocate for non-invasive or less invasive treatment of head and neck and oral cancer in order to minimize the damaging effects of the current treatments and also to provide palliative care for those in extreme circumstances.
Matrix Pharmaceutical has made a start in this direction with IntraDose, which as I understand it, is a gel which can be injected directly into the cancer.
Matrix, incidentally, has provided my transportation expenses to and from New York. But I'm not representing any particular cancer-related organization. I am here as an individual to speak for survivors and for those facing treatment who will become survivors.
My credentials come from a long-time association with support groups in New York City mostly. I have been associated with the National Coalition for Cancer Survivorship, with Cancer Care. I have been on advisory panels with the Cancer Information Service, and have volunteered for 13 years at the post-treatment resource program at Memorial Sloan-Kettering. So, my outlook and ideas have been formed by that experience and from listening to, talking to, and reading about hundreds of survivors. So, I think I can speak for them.
The current treatment for head and neck cancer is, of course, radiation, surgery, and chemotherapy in various combinations and degrees. The same treatments, of course, are used for other cancers, but the problem for us as survivors is the damage that is done to the head and neck, organs, nerves, and bones as a result of the treatment. I think most of us have gone through dry mouths, a variety of pain, loss of taste and smell, loss of hair, nausea, depression, and difficulty in chewing and swallowing. These results can last for months and for even years, and beyond that there are permanent problems, such as loss of speech when associated with cancer of the larynx. Damage to hearing and sight and to the teeth and to the jaw and sometimes the necessity to take nourishment through a tube in the stomach.
There have been many advances in treatment over the last decade. There is more precise surgery, more focused radiation, and more tolerable chemotherapy, but still the damage occurs. It would be wonderful if the pain and damage could be minimized or avoided by treating with non-invasive items such as drugs and medication. As it stands now, many survivors have told me in sincerity that if they had known what was going to happen to them after the treatment, they would have refused treatment, and I know people who have refused treatment because of what they feared the results would be. This is what we hope to avoid.
You may wonder why I'm speaking for survivors when we've already been through the after-effects. Well, the obvious answer to that is that we all face recurrence, and certainly I wouldn't want to go through the experience again. And knowing what we do, we certainly wouldn't want those undergoing treatment to face the same experience if there is another way. It is that other way that I hope the panel will help to bring about. So, thank you.
Are there any questions?
DR. NERENSTONE: Thank you very much.
Mr. Curry is our next speaker.
* MR. CURRY: Thank you for inviting me here. My name is Richard Curry and I want to talk to you about my father and how IntraDose helped to give him three or four good months of life.
I'd like to say first that I have no financial interest in Matrix Pharmaceutical, although Matrix did provide for travel expenses to come to this meeting.
My father's name is Anderson Rudolph Curry. He spent all of his life as a carpenter until he retired. He took a great deal of pleasure in building things. He was a strong man. He worked in the sun. He smoked cigarettes. He liked to play golf. And he enjoyed being with people.
He was a member of the American Legion and the VFW, with a lifetime membership in both. He was a veteran of World War II. And he liked to have a drink every now and then. He'd often cook for the American Legion, and got involved in many charitable events in his community.
He was 72 years old when he was diagnosed with cancer of the larynx in 1992. He underwent radiation therapy and had a complete response.
Two years later I remember taking him to the hospital after it was determined that the cancer had returned. He had to have his larynx removed and he had a tracheotomy. He was in the hospital for weeks. The surgery devastated him, both mentally and physically.
After surgery, the amount of time he was able to spend with his friends and communicate with them declined dramatically. He didn't like to leave the house, and it was painful for him to play golf. I think he was very self-conscious because he couldn't speak well.
Dad had to return to the hospital every few months for a procedure that would dilate his throat so he could swallow. In 1996 the doctors found recurrent head and neck cancer. They said he couldn't dilate his throat anymore and they put a feeding tube in his stomach.
In September of 1996, they really didn't expect him to live for three or four more months, but he used that feeding tube for more than eight months. He did not want to have chemotherapy. He didn't want anything to do with chemotherapy because of the side effects, and he was already devastated from the surgery.
In September of 1996, his doctor at the VA hospital told him about a clinical study with IntraDose that was going on at the Tucson VA hospital, and Dad wanted to become a part of this study. The doctor involved in the study in Tucson, Dr. Gerwald, is a medical oncologist. He said it was a double-blind study, where even the doctor would not know if he was getting the drug or the placebo.
Dad was looking forward to possibly feeling better, not for a cure. He was hoping that there would be something he could do to improve the quality of life, or maybe just extend it a little bit. He also wanted to be able to help someone else by what was being learned by his taking part in the study.
The first thing that had to be considered was how he was going to get to Tucson and back. I agreed to drive him to all of his treatments. I looked forward to the time that we were able to spend together, and I wanted to be a part of what he was experiencing and share with him the time that I could. I was glad to see that he had chosen to do something that could help someone else, and to that end, I wanted to help him as much as possible.
He received a total of six treatments with IntraDose over about two months. With the first treatment, there was some swelling, but I don't recall there being any major reaction to the injection. Of course, it was uncomfortable with somebody sticking a needle in his neck repeatedly, but not debilitating. Later, after the second or third treatment, the pain was worse, so they gave him morphine. He seemed to get some relief from that, and he indicated that he felt pretty uncomfortable the first day or two after the treatment, but he was always ready to go back for the next treatment.
Dad said at one point, it feels uncomfortable but it feels like it's healing. You know what I mean? It doesn't feel good but more of a healing pain.
I never heard him complain about nausea during the treatment with IntraDose, and after the third injection the tumor changed color. It turned black. It developed a big scab, and it eventually healed. It was like the whole tumor had died.
Dr. Gerwald had been concerned that the tumor would block his tracheotomy opening and make it hard for him to breathe, but the treatment seemed to prevent that.
By the time they were into the fifth week, I think he was feeling better. His attitude had changed and he knew that he had been given a little more time. He was going out to play golf, and he started fishing again, and he was visiting his friends.
Before the treatment with IntraDose, my father's quality of life had gotten worse, and as the cancer had progressed over four years, Dr. Gerwald and his staff were not waiting for him to die. They were waiting for him to respond, and that was a big difference. When he did respond, there was joy for everyone at that point.
If IntraDose had been available when the tumor was discovered, I'm sure he would have used it. It had given him three or four good months of life, and they were good times for me and the people who knew my father. I really hope that this treatment will be available to people like my father. He got better during those treatments. And if it could help someone else, it really should be on the market.
Are there any questions? Thank you.
DR. NERENSTONE: Thank you very much, Mr. Curry.
Our next speaker is Kim Thiboldeaux.
* MS. THIBOLDEAUX: Good afternoon. Thank you for the opportunity to be here today. My name is Kim Thiboldeaux and I am the president and CEO of a national nonprofit organization called the Wellness Community.
For the record, the Wellness Community receives no funding from Matrix Pharmaceutical, and Matrix did not pay for any travel or expenses related to my presence here today.
By way of background, the Wellness Community provides education, support, and hope to people with cancer and their loved ones. We currently have 20 facilities nationwide, four facilities in development in the U.S., and two facilities abroad. Our program includes support groups, educational seminars, nutritional workshops, exercise programs, and mind/body programs. We served an estimated 18,000 people with cancer in the year 2000, and these individuals made over 150,000 visits to our facilities.
At the Wellness Community, we serve people with all cancers at any stage of the disease. We see a wide range of diagnoses, and have had the opportunity to provide direct services to people with head and neck cancer.
While there are currently more than 160,000 people living with head and neck cancer in the U.S., the prognosis for this patient population has not significantly improved over the past 30 years. In addition, as you know, head and neck cancer can be a particularly devastating diagnosis, oftentimes causing facial and other deformities and interfering with basic functions such as breathing, talking, and swallowing. The psychological and emotional impact of this disease can be quite distressing, leaving little hope for the future.
We are in great need of improved treatment options and disease management tools for people with head and neck cancer. It is critical that new treatments not only fight the cancer, but also allow patients to experience a meaningful quality of life, whether that means continuing to work, traveling, enjoying time with family, or just taking a stroll in the park.
We are also in need of treatment options that are more targeted and can become an alternative to the disfiguring surgery that often accompanies the diagnosis of head and neck cancer. With rapid advances in cancer treatment, we are optimistic that the experimental therapies of today will quickly become the standard of care of tomorrow. We are also optimistic that physicians will engage cancer patients in an open dialogue about goals of treatment, lifestyle concerns, quality of life, side effect management, and other supportive care issues related to a cancer diagnosis.
I would ask today that you carefully consider the plight of patients with head and neck cancer and endeavor to understand the range of both medical and psychosocial issues these patients confront on a daily basis. I would ask today that you seriously consider the need these patients have for a broader range of treatment options and better tools to manage their disease. And I would ask today that you take a leadership role in encouraging patients to be educated, empowered, and optimistic about the cancer community's commitment to improving the lives of all people with cancer.
DR. NERENSTONE: Thank you very much.
Our next presenter, Mr. Findlay. It's a video.
MR. FINDLAY: My name is Ian Stewart Findlay. I am 59 years old.
I have no financial interest in Matrix Pharmaceutical. Matrix was, however, kind enough to provide resources to videotape my testimony to be given before the FDA.
I am an engineer for Boeing in Huntington Beach, California. I work on the Space Station and on the Delta 4 program.
Today I want to talk to you about my experience with IntraDose. I discovered I had cancer in 1992. I was rock climbing in Yosemite. I didn't seem to have the energy that I normally have. I went to UCLA and the head and neck clinic took six spine needle extracts and told me a week later that I had squamous cell carcinoma in the large lymph nodes in the left neck.
Within a month I had radical neck surgery. They took out 49 lymph nodes. They recommended I follow up the surgery with either radiation or chemotherapy. I decided I didn't like the effects of either of these two treatments, so I said, no, thank you.
I was okay for about two years, and then I started getting some more lumps in the same left neck area. So, I went to Hope Presbyterian Hospital in Newport Beach, California, and I received regular systematic carboplatinum and 5-FU chemotherapy for about two months.
Again, two years -þ well, actually it was about two years ago the tumors started coming back in the same left neck area. Just about that same time, I heard that Dr. Dan Castro, a head and neck surgeon at UCLA that was involved with administering a drug in a trial that meant direct injection into tumors. I heard it didn't knock your immunity system the way normal chemical therapy does, and the shrinkage of the tumors, in comparison to regular chemo, is fairly quick. So, I thought, well, I don't have anything to lose. I might as well try this new treatment.
My tumors were fairly good size. They had started restricting my mobility and interfering with all the sports that I was doing, so I had to stopp the sports.
Dr. Castro treated me once a week with the Matrix drug on the left side of the face and the neck. He said that if we could decrease the size of the tumors with the IntraDose, then it would be easier for him to follow up with surgery and/or radiation.
I kept working during the treatments. I didn't expect the treatments to totally eradicate the cancer, but I did expect it to reduce the size of the tumors, and it did. The tumors were always going down, and this made me pretty happy. This was great. My attitude started to improve.
The treatments caused quite a bit of discomfort for about two days. I would feel a mild stinging and warmness, but that was tolerable. I also experienced a little nausea -þ I would say on a scale of 1 to 10 a 3 -þ but only a couple of times. I don't ever remember my appetite being affected. I eat like a horse and I didn't lose any weight, unfortunately. My treatment did bother me some. I felt nausea and lightheadedness for about a day, but that was only one time. That time I experienced a stinging sensation at the injection site, but when I took pain pills, they reduced the pain to minor discomfort.
Both the radiation and normal systemic chemotherapy affected me much more than IntraDose in terms of my ability to work and function. I was certainly miserable enough during the radiation and the chemo to need to work part-time, and a nap in the afternoon was necessary. During the IntraDose injection treatment, I didn't have to do that. The treatment was less disruptive and not as discomforting in terms of weight loss, nausea and fever. Also, the shrinkage of the tumors was dramatically visible, so that was very satisfying. For these reasons I would think it would be a good option for cancer patients.
If I could speak to the FDA directly, I would say that of all the treatment options I've tried, the IntraDose definitely had the most dramatic and quick results, and it didn't knock down my immunity system like the systemic chemotherapy. If I had the opportunity, I probably would not have done the radical surgery. I would have tried the direct injection of IntraDose. I was very pleased when Dr. Castro started the treatments and I began to see these large tumors diminish.
In closing, I would like to urge the FDA to approve this drug in the hope that it can help other head and neck cancer patients.
DR. NERENSTONE: We'll turn now to the sponsor's presentation.
* DR. HOWELL: Good afternoon, ladies and gentlemen. It's my pleasure to open the presentation of NDA 21-236, cisplatin/epinephrine gel, for the treatment of squamous cell carcinoma of the head and neck.
My name is Stephen Howell. I'm a professor of medicine and medical oncologist at the University of California, San Diego, where I run the cancer pharmacology program for the UCSD Cancer Center.
I'm here today because of a longstanding interest in regional therapy and because I've been working with a team at Matrix Pharmaceutical for a number of years on the concepts underlying the product that we will hear about today.
The presentation today will consist of a discussion of the current management of recurrent head and neck cancer by Dr. Glenn Mills, who is a medical oncologist and professor of medicine, also head of the aerodigestive malignancy program at Louisiana State University. I will return to discuss the pharmacologic rationale and some of the challenges associated with the assessment of clinical benefit in these patients. Then Dr. Richard Leavitt from Matrix will present the clinical study results, and I will return again to discuss some of the clinical benefit issues. And finally Dr. Glenn Mills will finish up the presentation by discussing the risks and benefits of this product.
We are accompanied today by a number of independent experts who are available to answer questions about the disease. Dr. Everett Vokes, who is head of hematology/oncology at the University of Chicago, and sub-Chair of the Head and Neck Cancer Committee for RTOG. Dr. Barry Wenig, who is professor of otolaryngology and Chief of the Division of Head and Neck Surgery at Northwestern. Dr. John Mackowiak, Director of Research at the Center for Outcomes Research in Chapel Hill. Dr. John Durant, former Director of both Fox Chase and the University of Alabama Cancer Center, and Chairman of Clinical Cooperative Group, a past chairman and executive at ASCO. And Dr. Robert Woolson, professor and past Chair of Biostatistics at University of Iowa.
We are also accompanied by other staff from Matrix Pharmaceutical, Dr. Laurence Elias, who is the Nedical Director who has handled the safety analysis of this product. Dr. Morgan Stewart, Senior Director of Biostatistics, who has handled the biostatistical analysis of this product. And Dr. Robert Tressler, who has handled the preclinical studies.
I'll now turn the podium to Dr. Mills.
* DR. MILLS: Thank you, Dr. Howell. As Dr. Howell told you, I'm Glenn Mills from LSU-Shreveport, where I head up the head and neck program at Shreveport. Also, I'm a PI for SWOG.
What is the scope of the problem we're talking about today? It's estimated in this year, total, there will be about 50,000 new cases of head and neck cancer that we're going to see, at all sites, today. Of this group of patients, we estimate that there will be about 15,000 deaths from this disease this year. Of those people that die, approximately 50 to 65 percent will have local recurrence as a component of their disease at that time.
As you know, the risk factor for head and neck cancer include tobacco and alcohol, and this is important, particularly the tobacco use in this patient population, because of the concomitant diseases that we face, much like lung cancer. A lot of vascular disease, a lot of COPD, malnutrition from difficulty eating.
Early-stage disease is best managed with radiation and surgery. Relapses are still seen not uncommonly. Late-stage disease, really, we're talking about radiation and chemotherapy, with perhaps some form of surgery for some of these patients, but relapse remains a problem, and local relapse is still a problem in this disease.
The current chemotherapy standard that is recognized by the groups is cisplatinum/5-FU, which has been around for a while, and in the phase III setting, gives us response rates of about 30 to 35 percent, mostly partial remissions. Several new regimens are being explored in the primary treatment of this disease, but have yet to be compared to platinum/5-FU. We should remember that most all of the studies have shown that when you have locally recurrent disease in an irradiated and surgical field, the responsiveness to chemotherapy is diminished.
Let's now concentrate on locally recurrent head and neck cancer, the topic today. This is a highly debilitating disease, as you've heard from our patients that spoke earlier. Intractable pain is not infrequent. Compromised airway from the tumor obstructing the trachea, swallowing difficulties, frequently these patients have ulcerated wounds that are quite noticeable when you first walk in the room to see them and keep them from being around people. Local problems in this disease may predominate, even in those patients with systemic metastatic disease. Median survival in this group of patients is short and their quality of life is poor.
What are unmet needs now in locally recurrent disease? In patients who have failed radiation therapy and surgery, re-irradiation is now being explored in some patients, and indeed, some promising results are being seen, but this is not an option that's open for every patient. In chemotherapy-failed patients, in primarily cisplatinum-filed patients, we don't have any approved drugs. There are multiple drugs that have activity, and multiple combinations that have been reported in the literature. Little impact, however, overall on survival in this setting.
We need new agents with better or unique activity, with reduced toxicity, improving our palliative goals for these patients, and we need to be able to reduce the risk of catastrophic events -- airway compromise, swallowing difficulties, bleeding -- in this group of patients.
What are we talking about today? Let me show an example of a few patients, and these are patients you will hear about on this trial.
This patient had an 8 cubic centimeter tumor, lateral to the tracheostomy, and you can see the tumor right here, a small part of it, pushing into the tracheostomy, impinging his airway. He was no longer able to get his tracheostomy tube in place. His airway potentially is going to be compromised.
Here is a patient with a 4 cubic centimeter tumor at the base of the tongue, barely able to see the uvula in this patient. It is beginning to cause a significant oral problem. This is the type of problem that we're talking about today.
Dr. Howell will now talk about the pharmacological rationale.
* DR. HOWELL: Let me start by being clear about the patient population which we think is appropriate for treatment with CDDP/epi gel. And note that the indication has been refined since the NDA filing.
First, when head and neck carcinoma recurs, all patients should be considered for additional surgery, systemic chemotherapy, or re-irradiation. CDDP/epi gel is indicated for patients with locally dominant problematic lesions, who are not surgical candidates, because lesions are not resectable, resection would destroy function of a critical organ or the surgical risk is too high; who are not candidates for systemic chemotherapy because they failed prior regimens or have co-morbid disease that prohibit it; are not candidates for re-irradiation because the risk is too high, or they don't have access to appropriate radiation expertise or facilities; or who have refused all other modalities.
I would point out that this is a very small subset of all patients with head and neck cancer, and that this is an orphan indication and orphan status has already been granted for this product.
Now, the product consists of a viscous injectable gel containing cisplatin and epinephrine. The cisplatin is present as an insoluble suspension at 4 milligrams per milliliter, and of course this drug already has an established role in the treatment of head and neck carcinoma. The epinephrine is present at .1 milligram per milliliter, and it provides local vasoconstriction in and around the tumor. The gel ensures physically stable dispersion of the cisplatin, and facilitates accurate placement of the drug.
Now, when we give cisplatin intravenously, we produce very high concentrations in the plasma compartment, and quite large overall exposures for this compartment. Some of that drug crosses into the tumor compartment, but the levels that we achieve in the tumor are quite modest, and the overall exposure for the tumor is quite limited.
What they're attempting to do with intratumoral therapy is produce very much higher concentrations and exposures for the tumor and, at the same time, decreasing the exposures for the systemic circulation. So, when we inject CDDP/epi gel at this extremely high concentration, that portion of the tumor accessed by the injection has a very high exposure. Because of the vasoconstriction, the rate at which the drug leaks out of the tumor is markedly reduced, and reasonably matches the rate at which it's cleared from the systemic circulation, so the peak concentrations in the plasma are never very high, and neither is overall exposure.
The median dose of cisplatin administered intratumorally in these studies was only 10 milligrams per meter squared. This contrasts with the standard dose of cisplatin of anywhere between 70 and 100 milligrams per meter squared if given intravenously.
Now, there is large body of preclinical data from experimental models indicating that, in terms of local control, one can do far better by injecting this material directly into the tumor than one can do with any dose of cisplatin given systemically, even maximum tolerated or lethal doses. And if you inject intravenous cisplatin into a mouse who has a tumor here in its flank and image the radioactive cisplatin externally, you can see that there's a very small accumulation of cisplatin, and it washes out of the tumor very quickly by 1 hour. If you inject cisplatin solution directly into the tumor, you get higher concentrations, but again, it washes out of the tumor quite rapidly.
If you inject cisplatin in the form of CDDP/epi gel into the tumor, you get very much higher local concentrations, and the drug washes out of the tumor very much more slowly. This is shown in this graph. The blue line is free cisplatin injected into the tumor, short half-life. The red line is CDDP/epi gel injected into the tumor. Much higher peak concentrations, and a much, much longer half-life of the drug in the tumor.
Now, I want to point out that one of the important versatilities of this technology is that you do not have to get good drug distribution on any one injection. This is depicted here with the tumor being shown in light blue, the gel being shown in the platinum color, and the portion of the tumor successfully accessed by drug exiting from the gel being shown in yellow.
On the first treatment, you might very well get only a portion of that tumor covered. When you treat the patient again, that tumor has undergone some necrosis and reduction in size. There is a proportional reduction in dose, but now overall you get better drug distribution. By the third, fourth and fifth treatments, that distribution has progressively improved. So, I want to point out again, you do not have to get excellent drug distribution on any one treatment for the program to be successful.
The recommended dose is .25 milliliters per cubic centimeters of tumor volume, with a maximum amount at any one treatment setting being 40 milligrams, so this is a volume per volume dosing scheme.
Now, the company faced a number of challenges in designing and executing these trials. The study was originally designed with the primary endpoint being the response rate of the most troublesome tumor. The company was fully aware from the very beginning that it was important to demonstrate clinical benefit of this product, and in an effort to do so, clinical benefit information was collected by measuring improvement in symptoms and by looking at prevention of catastrophic anticipated complications.
The trials were powered on the MTT response rate because back in 1994 when these were designed, there was no validated method for assessing the anticipated clinical benefits of a local control.
In the years since then, all of us have paid a lot more attention to clinical benefit in terms of the importance for drug approval, and the FDA eventually asked the company to analyze for the variable patient benefit as an additional primary endpoint to this trial.
Now, that posed a problem because the trials have been powered on the basis of MTT response rate, and as noted by the medical reviewer himself, these patients have multiple different kinds of symptoms so that it's impossible to approve enough patients with any one type of symptom to properly power a trial.
And therefore, at the time this request came through, it was clear to the company that an integrated analysis of the two trials together was going to be necessary to respond to this challenge, and that analysis you will see today.
Now, there are some real problems in trying to assess clinical benefit in this patient population. One of the problems is the enormous heterogeneity of symptoms. These patients have many different kinds of symptoms. One patient will have pain as the predominant problem, another patient will have ulceration of a wound as the predominant problem.
There is variation in the number of symptoms per patient. Some patients have a single dominant symptom, other patients have four or five problems depending on the location, size of the tumor.
Some symptoms are more important to the patient than other symptoms.
We have the problem of assessing palliation versus prevention. In the management of this disease, both palliation of the patient symptoms and prevention of anticipated devastating complications, such as invasion of the carotid, invasion of the trachea, the orbit, are important aspects of patient management.
It would be preferable to have a dichotomous variable that gave a yes/no answer to the question of whether the patient benefit had been attained, but how do you combine together measures of palliation and prevention? These are different in nature. They are measured on different scales.
How do you deal with a situation where the most critical symptom gets better but others worsen?
How do you deal with a situation where you need to adjust palliative scores for differences in the importance of that symptom to the patient?
Well, let me remind you of clinical reality. The clinical reality is that it is very hard to achieve any kind of improvement in these refractory, recurrent, far advanced patients, as is shown here. And virtually any degree of symptom improvement is something that we in the medical community ought to celebrate.
Well, what approaches were taken in these trials to assess clinical benefit? Well, one of the things that was looked to was tumor shrinkage itself. Tumor shrinkage is often an obvious benefit, both to the patient and the physician, particularly when the lesion is an obstructing lesion. I would submit that the value of shrinking these kinds of tumor masses is fundamentally different from the value of shrinking a distal skin metastasis due to, say, melanoma or breast cancer, and that shrinkage of a tumor in these kinds of patients is a direct measure of clinical benefit.
The second measure of clinical benefit in these trials was palliation, and this was approached by identifying the patient's most troublesome tumor, and then the thing that was the most important symptom being generated by that tumor, and then using 4-point scales to track progress toward the goals prospectively and independently selected by the physician and the patient. And I want to emphasize that both the patient and the physician selected palliative goals.
Finally, prevention was looked to. The protocol identified the critical structures that were most threatened and then measured success in avoiding the anticipated complication.
And finally, a patient benefit algorithm was developed in an attempt to try to provide a yes/no answer as to whether clinical benefit was obtained when considering both palliative and preventive goals together.
I want to be clear that every time in this presentation that we refer to patient benefit, we are talking about the calculated product of the algorithm. When we talk about clinical benefit, we're talking about all the elements that might be construed as indicating that the patient had improved with therapy.
Let me turn the podium back to Dr. Richard Leavitt, who will present the clinical trial results.
* DR. LEAVITT: Good afternoon. I'm Richard Leavitt. I'm here to represent these clinical studies for Matrix Pharmaceutical, and it's my pleasure to present these results to you.
These studies were two adequate, well-controlled, double-blind and placebo-controlled trials done in patients with advanced cancer of the head and neck. The design of these studies was to randomize patients between receiving a blinded treatment with cisplatinum/epinephrine gel or placebo. It was an unbalanced randomization with twice as many patients receiving cisplatinum/epinephrine gel.
Patients were treated weekly for 6 weeks in an 8-week period. They were evaluated for response. Patients who had persistent tumor at the end of that period, or patients who had progressive tumor at any time during the therapy following three treatments had the opportunity to switch over to an unblinded therapy with cisplatinum/epinephrine gel. I would emphasize that at no time during this study was the identity of the therapy revealed to the patient, the investigator, or to the staff at Matrix Pharmaceutical.
These studies were done, one in North America and one in Europe and Israel. The studies were of identical design, followed identical protocols, and used identical patient and data collection instruments. All analyses that we will show you are intent-to-treat analyses, and the studies were simultaneously unblinded, so there was no opportunity for the results of one study to influence the conduct of the other.
I will first present to you the efficacy analyses, including the prior treatments that patients had before receiving therapy, and then return to the question of patient benefit. And finally, to consider safety issues.
In these studies, these were patients with advanced disease, and at the time of relapse, they were all considered first for standard therapy. You've heard that we've designated one tumor, the MTT, or most troublesome tumor, as the tumor that was either most symptomatic or most threatening, and for these patients, 89 percent of those tumors had occurred and recurred in a previously radiated field. This limits opportunity for repeat irradiation, which would be the other treatment modality for local control in previously unirradiated tumors.
I'd also emphasize that 89 percent of these patients had received multiple previous therapies, including surgery, radiation, and chemotherapy, in various combinations, many at the time of relapse following primary therapy.
These are the results of the trial. The North America study, the Europe study, and the combined results. The response rates of these tumors were gratifyingly high. In the North America study, the complete and partial response rate, again durable for a minimum of 28 days, was 34 percent. In the Europe study, 25 percent; combined results, 29 percent. In each of these trials, this result was statistically significantly different from the response in the placebo arm that was conducted simultaneously.
I would also point out that in each of these studies in the combined analysis, complete response of the tumors was nearly twice as frequent as partial response.
Even the partial responses in the study were clear partial responses. At their maximum regression, these tumors regressed from 79 to 99 percent in the group that were classified as partial responders.
Responses were prompt, and they were durable. The median time to response on this study was 21 days. The duration of response was 78 days. And I would remind you that the way we analyzed these data is that responses were censored for duration anytime the patient went on to receive any potentially confounding therapy. 33 of the 35 responders that we're discussing remained in local response at the treated tumor at the time that they went on to receive any confounding therapy, or left study for palliative care of another sort.
Time to progression is shown here, comparing the patients who were randomized to receive cisplatinum/epinephrine gel, and those patients randomized to receive placebo. The median time to progression was prolonged in patients who received cisplatinum/epinephrine gel, 149 days.
I explained to you that this is a placebo-controlled trial, and patients, at the time that they had progressive disease or failure to respond to therapy, had the opportunity to then cross over to open-labeled cisplatinum epinephrine gel. I'll also emphasize that at that time the blind still remained unbroken, so neither the patient nor the physician nor the sponsor had any knowledge of what treatment the patient had received.
In the group that crossed over, after having failed placebo therapy, 27 percent of these patients went on to have a response. This is nearly identical to the response rate in the combined analysis from the blinded phase. This response rate was obtained in spite of the fact that these tumors had increased during the placebo treatment from a mean size of 5.7 to 10.8 cubic centimeters at the time of crossover. Again, complete responses were more frequent than partial responses.
It's also important to examine the effect of previous therapy on the occurrence of these responses. Many of these patients had received previous platinum-based therapy with cisplatinum or carboplatin, and we asked the question, was there an effect of previous experience with platinum-based therapies on the response rate. In the 48 patients who had received either prior cisplatinum or carboplatinum, the response rate was 29 percent, and this is identical to the group of patients who were platinum-naive.
I'm now going to turn to patient benefit and the critical importance of the patient benefit outcomes in evaluating the response to this drug and the value of this drug to patients with advanced recurrent and refractory head and neck cancer. I will discuss the instrument used to collect these data, the treatment goal questionnaire, how these data were analyzed to come up with a single clear endpoint that declared patient benefit or no patient benefit, and then finally turn to the result.
The treatment goal questionnaire is designed to assess the direct effects of benefit from local therapy. What I mean by this is that patients and investigators, prior to beginning therapy, chose prospectively their goals for treatment.
Palliative goals were frequently chosen, and these were graded on a very clear 4-point scale. The differences between levels in the scale are quite clinically distinct from one another. In order to declare a benefit or achievement of a treatment goal, it must be durable for 28 days. On the other hand, failure of a treatment goal requires simply a worsening in the score that lasts for 7 days.
Preventive goals were also assessed in this trial, and these were important to both physicians and patients, but only physicians were given the opportunity of choosing an important goal of preventing an event that they felt was clinically imminent, and clinically important.
This approach in the treatment goals questionnaire was independently validated by the Center for Outcomes Research, and Dr. John Mackowiak, who conducted that validation, is available here for questions later if you would like more detail.
I would just like to speak to the distinctness of the different levels in these treatment goals that were put before patients, and pain control was frequently chosen as an objective. We tried to make the levels so distinct that there was not a great deal of influence from subjective factors. For a patient to go from level 4 to level 3, he must have had pain that was uncontrollable and now became controllable with strong pain medicines. The most difficult step perhaps is from level 3 to level 2. Patients who were pain dependant and needed narcotics or prescription pain medicines for relief had to be able to go from prescription medicines to over-the-counter, simple analgesics. And obviously level 1 is no longer a requirement for pain medications.
Putting together the results of these treatment goals needed a simple approach to end up with a single dichotomous clear endpoint and judgment of patient benefit. What we did is we put together these results and looked at the physicians' and the patients' responses. And the approach was very straightforward. We only scored a patient benefit if the goal was met by both the patient and the physician, or if either the patient and the physician met a goal, at least the other investigator or patient then said, my goal at least has not worsened in any way. If either the patient or the investigator said my goal for treatment is getting worse, we counted that as no patient benefit, no matter what other palliative benefits were noted by patient or investigator on other goals noted before study. The primary goal is the key to determining patient benefit in the data I will show you.
Turning now to the achievement of patient benefit. In the studies combined, 27 percent of the patients on cisplatinum/epinephrine gel, IntraDose, achieved benefit; only 12 percent of those patients on placebo. This reached statistical significance.
In the individual studies, which I will remind you we knew were not sized sufficiently to detect a statistically significant difference in benefit, there was nevertheless a strong trend in both studies. In the North America study, 34 percent of patients achieved patient benefit by this very strict definition; only 17 percent on placebo. Similarly, in the ex-U.S. study in Europe and Israel, 19 percent versus 9.
It's also reassuring looking at these data that when you look to patients who crossed over to receive active therapy in an open label phase, the patient benefit achieved for these patients was 41 percent, despite the fact that they had now crossed over to open label because they had not achieved a response during blinded therapy. This number is nearly identical to the combined benefit rate in patients originally randomized to blind therapy.
It's important to look at the components of the treatment goal algorithm and treatment benefit algorithm in order to gain some insight into these data. This is an analysis that we did in order to look into these data in more detail.
I'm sorry. I would quickly mention that response and patient benefit were highly correlated, and although this does not prove that there was patient benefit, again it does give us confidence that the measurement of tumor response is an important measurement of the outcome of this therapy.
Again, looking now just at the palliative component of this study and this benefit, and looking only at the primary palliative goal, there are 13 percent of patients in active, 4 percent in placebo who achieved goal, a trend that is not statistically significant. However, if we look at any of the palliative goals prospectively chosen by the patient or the physician, this difference is statistically significant, 18 percent versus 6 percent.
We also encouraged patients and physicians to be alert for other benefits that occur during therapy, even if they involve non-prospectively chosen palliative goals. And when we look at patients who reported on case report forms during therapy, while study was blinded, any other palliative benefits, we see that overall, including these previously unforseen benefits, the overall benefit rate for palliative goals was 34 percent. This is also statistically significant.
This is associated with patient response, and if you look at either the palliative goals, any palliative goal, and these unexpected and reported benefits, all of these are highly correlated with tumor response.
I'd now like to turn to the element of the patient benefit algorithm and determination which involves prevention. In this disease, advanced head and neck cancer, prevention of serious complications is an important part of the objectives of therapy. And we collected data on these prospectively by looking for prevention of such things as invasion of a vital structure, where this can be devastating. Airway obstruction also directly threatens life, and certainly impaired swallowing. All of these were frequently chosen preventive goals, and we believe that success in achieving these goals can be very clinically meaningful.
The organs that were chosen and specified by the investigator as the organs that were threatened and that he wished to prevent complications are listed here. For those 26 instances in which the investigator chose prevention of obstruction, it was the trachea or the airway that was most frequently chosen as the organ that was threatened. For prevention of invasion, in 31 of 50 cases it was a major blood vessel.
If we now look at the prospectively selected primary preventive goal, this is also statistically significantly associated with therapy. I will mention that with regard to preventive goals, the patients on placebo were counted as failing a preventive goal if they did not have at least 28 days of prevention.
Now, there are certainly challenges in evaluating preventive goals. Most importantly, FDA has pointed out that it is difficult to make a direct comparison between the rates in the placebo group because the patients frequently did not complete 28 days of blinded therapy. The reason that these patients did not complete therapy is the tumors were rapidly progressing, and during the time that they were on blinded therapy, the mean size of tumors nearly doubled from 5.7 to 10.8 cubic centimeters, and it is true that there were few patients available to remain in therapy at the end of 28 days. So, we must estimate what the preventive rate goal failure would be by a combination of overt failure of the goal and recognizing that these tumors were advancing and patients could not remain on blinded therapy.
Finally, we would propose that including preventive goals is important in assessing patient benefit. There was a single patient benefit outcome that was prespecified for both palliative and preventive goals. Physicians do believe that prevention is crucial in this disease, and in one respect, the ability to actually complete the 8-week blinded therapy is implicit evidence of important attainment of prevention. All of this was part of a prospectively planned analysis. It is important, once the blind is broken, to look at all of the components that might contribute to the palliation and the palliative benefit, but these kinds of analyses by both the sponsor and by FDA should be secondary and help to explain the data, but should not replace the primary analysis.
I would like to turn to other data that are supportive of these results that we have found in the clinical trial, and I would speak specifically about two open-label studies in other solid tumors, mostly in patients with tumors such as chest wall recurrence of breast cancer, malignant melanoma, sarcomas, and other tumors. The efficacy endpoints in these studies are identical to those that I showed you for the head and neck cancer trials.
Looking at the combined response rate overall, the response rate was 35 percent; 31 percent in a U.S.-North America study, 41 percent in an ex-U.S. study. Complete and partial responses were frequent.
Now, let me take you through this slide. We looked at patient benefit in these studies as well, but we did not have a simultaneous placebo control. The patient benefit rate for these studies was 37 percent and 25 percent, and again, we have confirmation from the association of benefit and response that local disease response is a meaningful outcome to measure and examine in these studies. Amongst responders, 55 percent were benefitters in the North America study, 50 percent in the ex-U.S. study. These differences are either statistically significant, or nearly so.
I'll quickly turn to the safety profile, talking about dosing questions and the ability to deliver the expected and projected dose, selected adverse events, and specifically local cytotoxic effects, the things that we expect to happen at the site of the treated tumor that frequently accompany response. And finally, I'll briefly discuss selected clinically important adverse events.
FDA has pointed out that patients did not receive the full 0.25 milliliters per cubic centimeter of tumor determined by the original treatment volume. However, true dosing errors were actually very infrequent in this trial. In most cases there was no dosing discrepancy, and in those cases where there were changes from the ideal dose, these were most frequently due to things that were pre-specified directions in the protocol for changing the dose. For example, if injecting locally into the tumor could simply not be accommodated, then dosing was supposed to have been stopped, and it was appropriately in 11 percent of the cases that received active gel.
Another 2 percent of cases had an incomplete dose delivered because there was some of the drug that actually refluxed from the tumor.
Other dosing deviations included such things as stopping treatment when the tumor responded. Again, specified by protocol.
Reasons such as adverse events were rare. Patient refusal of therapy was rare. And there were true dosing calculation errors in only 4 percent of the cases.
Now, this is a very busy slide, but I'd like to take you through some of the adverse events that occurred during these trials. In these two columns are data for cisplatinum/epinephrine gel. Here are data for placebo. And we have divided these between mild and moderate reactions and those considered severe.
For immediate injection effects, those things that are part of the injection procedure, mild and moderate pain were of identical incidence between the patients treated with cisplatinum/epinephrine gel and those treated with placebo. Severe pain was noted more frequently, however, in patients receiving cisplatinum/epinephrine gel, 10 percent versus 4 percent in the placebo group.
Otherwise, the incidence of side effects is close in these studies, but I will point out certain substantial differences. Again, at the site of treatment, mild to moderate or severe pain that occurs during the reaction and response of the tumor was more frequent and active than placebo. And similarly, when you look at distant effects such as pain, these were equally frequent in active and placebo. It is only the local condition where there is a substantial difference in pain between the two treatment arms.
I will also point to nausea and vomiting, which would be expected frequent complications of systemic cisplatinum therapy. These were seen somewhat more frequently, 14 percent, 14 percent, in the active, and only 5 percent and 1 percent in the placebo. This difference may have been due to low systemic levels of cisplatinum, or perhaps more likely the more frequent use of systemic narcotics.
Lastly, there are local conditions that develop at the site of recurrent tumor and at baseline, before these tumors retreated. We carefully reported all of the local conditions surrounding the tumor, and you see these listed here. And the bars are for active group and those that received placebo. And you can see these are about equal at baseline before therapy.
If we now turn our attention to any worsening of these conditions, either the developing of a new condition or the worsening in degree of any of these, there was an increase in these conditions for those patients receiving cisplatinum/epinephrine gel. Most particularly, erosion and ulceration occurred more frequently in cisplatinum/epinephrine gel, as did necrosis. However, the occurrence of eschar, which one can see as part of the healing process as tumors and local cytotoxic conditions resolve, were virtually only seen in the patients who received cisplatinum/epinephrine gel.
Finally, there were other clinically important adverse events that occurred. We saw six patients in these studies who developed cerebrovascular events. Five of these occurred in cisplatinum-epinephrine gel group, one in the placebo group. These happened early in the trials. We carefully analyzed each of these patients and concluded, although not conclusively, that these were most likely due to carotid artery vasospasm, perhaps from needle trauma to the carotid artery or from irritation of the artery.
We changed the protocols. We excluded tumors that directly involved the carotid artery, and since doing that we have treated most of the patients in the study, and we have not seen another treatment-related cerebrovascular event.
There were some cardiovascular changes that were noted during these studies, mostly blood pressure and pulse elevations, which were prospectively measured and assessed for each patient in the study. These were transient. They were not associated with any serious adverse events. A single patient had an apparent loss of consciousness that was a possible cardiopulmonary arrest. The patient was hospitalized overnight and released the next day without sequelae.
In summary, I've shown you two adequate, well-controlled, placebo-controlled trials randomized in patients with advanced head and neck cancer. The results in these trials stand on their own, but they are also confirmatory and complementary of one another. We've shown you that effective local control can be achieved in patients with advanced recurrent head and neck cancer. These were associated with real patient benefit: palliation of symptoms, and prevention of complications. And the patient benefit is associated with tumor response. The supportive trials had high response rates in patient benefits, and overall the safety profile is well managed.
I'd like to invite Dr. Howell to return.
* DR. HOWELL: As you heard, the patient benefit algorithm was an attempt to provide an assessment of both palliative and preventive goals within the same patient, but perhaps the simplest, cleanest way of looking at palliative benefit is to take the population of patients who had particular symptoms and ask what fraction of those patients got better.
This slide shows the three palliative goals that were most frequently selected, for which the numbers were large enough to make any reasonable analysis. And what you see is that on the CDDP/epi gel arm there was a modest, admittedly so, but very consistent difference in all of the goals selected by the patient and all of the goals selected by the physician. Recall that the instruments used to measure progress toward these palliative benefits had big jumps between one level and the next, so modest attainment of improvement was expected.
Now, the FDA has performed an analysis of the attainment of primary palliative goals, and has presented the data in this table for your consideration this afternoon, for a vote on whether they provide substantial evidence of clinical benefit. And the data show that some patients get better and some patients, and a different fraction of patients, get worse. This is what we expect from the use of chemotherapy in this patient population. It's part of the natural history of the disease. This product does not cause a response rate in greater than 50 percent of the patients, so one doesn't expect a shift in the median, we don't expect all patients to improve.
Now, there are a couple of things about how this data was calculated that are important for you to understand. In order to score as better, the improvement had to last for a full 28 days. In order to be scored as worse, the worsening only had to last for 7 days. This was a purposefully conservative scoring system. If an improvement doesn't last 28 days, you know, its value is not so clear. But if something gets worse, even for 7 days, that can really impact on a patient's well-being.
Now, the default is that we expect more patients to worsen in this situation than to get better, and we also expect, because we know that CDDP/epi gel causes transient local symptomatology due to tumor necrosis, to produce some transient worsening in some patients, particularly patients in whom the tumor invades the skin overlying the tumor mass.
Now, the second thing that's important for you to understand about this data is that patients stayed on the CDDP/epi gel longer than on the placebo arm, so there is a greater chance of worsening in the CDDP/epi gel arm.
Now, the FDA folks have raised the question as to whether treatment with CDDP/epi gel makes these patients worse. And if you take exactly the same data and now you look at just the first 28 days, because that's the period when the largest number of patients on both arms of the study were still on study, and you score improvement and worsening on the same time interval, 7 days, then you get this set of data, and there's a consistent small effect of CDDP/epi gel over placebo in both studies and in the combined data.
Now, some patients are still getting worse. But if you look at the data over a time period of the full six treatments and the 1 month of follow-up, over the time period when we expect the local effects to have largely resolved, you see absolutely no evidence that there's a difference in the worsening rate between the two arms of the study. In fact, if you look at the obstructive symptomatology, there's a substantial concern that the patients on the placebo arm are getting into trouble with obstruction at a higher rate than the patients on the CDDP/epi gel arm.
But now, maybe the most important way to look at clinical benefit in this patient population is to ask, look, if a patient attains some benefit, is there any possibility that that benefit is offset by something else going wrong in the same patient? So, 12 percent of the patients in these studies attained either the patient's or the physician's primary palliative goal. Now, if we take away from that any patients whose other primary goal worsened, we wind up with a net primary palliative goal rate.
Now, there were no such patients in these trials, so we see a small positive event. This is a patient population where we can be really pretty sure that clinical benefit was attained because of a good correspondence between the physician's evaluation and the patient's evaluation.
Now, if we do exactly the same thing with respect to all palliative goals, 18 percent of patients attained þ I'm sorry any palliative goal, either primary or secondary palliative goal, and take away those in which in the same patient something worsened -- that occurred in 4 of these patients -- we wind up with a benefit in terms of palliative goals of 14 percent. Again, a pretty high confidence level that this is a population of patients who really have benefitted from this treatment.
So, what are the pieces of evidence that speak to the issue of clinical benefit from these trials? Well, first is the fact that there was a statistically significant difference in response rate in both studies, and in some of these patients, particularly those with obstruction, this represents an obvious clinical benefit. And such a response rate is, as a separate issue, reasonably likely to predict patient benefit.
There's a positive trend for the patient benefit variable calculated by the algorithm in both studies that reached statistical significance in the prospective integrative analysis. When examining palliation, there's improvement, small, but there for each type of symptom when examined individually.
In the integrated analysis of all primary and secondary palliative goals þ- these are the palliative goals only -þ there was statistical significance. And if you add in primary, secondary and the unforeseen benefits that reached statistical significance in each trial individually, studies 403 and 503, the phase II trials, provide supportive evidence that this drug is active and that it provides patient benefit.
Now, let me pose a rhetorical question. What level of certainty do we need on clinical benefit in a population of patients who are very narrowly defined, have a devastating problem, and no other therapeutic options? I would submit that the evidence available from these trials is reasonably strong, provides a reasonable body of evidence that this product is effective.
* DR. MILLS: Thank you, Dr. Howell.
I'll be brief in the summary.
I think, looking at risk/benefit at the end, I'd first like to say, what are the risks and benefits of the current therapy we have for this patient population? And obviously no therapy is going to be an option for many of these patients, and I think we do know what will happen. These patients will get worse. Local problems will progress with bleeding or airway obstruction. There will be a decline in quality of life, and some of these critical local tumors may shorten the patient's life.
Current therapies, radiation therapy, re-irradiation can be tried, but frequently we have an ineffective dose in this patient population. Surgery is usually not an option. Chemotherapy, we have to be careful. I think if we recall from Dr. Forastiere's paper in April JCO, this is a particularly fragile population, prone to toxicity. And there's little improvement in survival with any of our options at the present time.
Cisplatinum/epi gel I believe has been shown to have few serious side effects. They are usually local wound care and can be managed. Systemic effects were uncommon. I think strokes did occur, but with appropriate patient selection to avoid tumors that involve the carotid artery, this can be avoided.
The benefits. Well, we do have a high complete response rate in this trial, and that I think is an intrinsic benefit for many of these patients.
Clinical benefit was seen, both palliative and preventive, and one good thing is these responses were prompt, 21 days, meaning you could use this product and move on to other treatments in a relatively short order of time if you needed to. My patients, I gave it outpatient to all of them. It was an outpatient procedure and it was not difficult.
What about the patients we discussed earlier? Our first patient had an obstructing tracheal lesion, right here. Here you can see after therapy with cisplatinum/epinephrine gel, the tumor has responded. He eventually did have a tumor response and you see the eschar formation. I think you've heard from his son that he benefitted.
Our patient with the oral tumor had a complete response, a complete resolution of their tumor with therapy.
I do believe that this gives these patients a third form of local therapy to be considered in their management. It gives me a needed addition that I need in the clinic to help these people when local problems are the predominant problem and we have nothing else to offer. It is an effective and beneficial therapy for local disease.
* DR. NERENSTONE: We will open up the questions to the sponsor from the committee. Dr. Kelsen?
DR. KELSEN: Could you describe to us how you validated this quality of life instrument, and in particular, for example, for pain control, could you describe how you determined that moving from level 1 to 2 was significant. Was this compared to MPAC or to other instruments which are felt to be validated?
DR. MACKOWIAK: Yes. My name is John Mackowiak, Director of Research at the Center for Outcomes Research in Chapel Hill, and I personally conducted that validation.
Slide 287 on, please. We did a number of things to validate that instrument, and the results were that we found that it had excellent validity, good reliability, and clinical meaningfulness, and I'll summarize those.
In the validity aspect, I interviewed patients with head and neck cancer, as well as investigators who participated in the trial. I learned that all of them agreed that the instrument had excellent content validity. We had the right items in the questionnaire. And from the study results, we know there was high association between tumor response and the benefit endpoints. We also know there was good reliability, but most important, there were important clinical differences.
Can I have slide 614, please? In interviewing patients and investigators both, I showed them these four different levels of pain control, which you saw earlier. And they had to sort them in the correct order first. They had their own independent cards. After sorting them in the correct order, I asked them the simple question, is it clinically meaningful to move from level 4 to level 3, is it clinically meaningful to move from level 3 to level 2, and on?
There were eight different questions, three questions on each one, and from the physicians, almost 100 percent of physicians agree that all these levels were clinically meaningful. Some of them were slightly less meaningful. Patients, all of them, agreed 100 percent of the time that the levels were clinically meaningful. So, that was how the validation process was done.
DR. NERENSTONE: Dr. Redman.
DR. REDMAN: Several questions again on the endpoints. Looking at your scale and specifically pain, if a patient required increasing narcotics for pain control, they wouldn't leave level 3. How did you score that?
DR. HOWELL: You're correct. If they did not have a dramatic reduction -þ
DR. REDMAN: I'm not talking reduction. I'm talking increase. Negative benefit. Patients on narcotics for pain control while on study requires increasing doses of narcotics. They're still level three.
DR. HOWELL: Yes, that is exactly true. You identified one of the issues with this instrument. If the patient required increasing levels of narcotics but still managed their pain, they stayed on level 3. In addition, if their narcotics were cut in half, 50 percent reduction, as was true with gemcitabine with their clinical benefit response claim, they still stayed on level 3. Even if they stopped using narcotics but switched to another prescription medication, they stayed on level 3. So, we know that the instrument is very valid.
If they do achieve a change in one level, we know there's a clinical benefit, but we also know now after using it, that it's not sensitive to some changes. Even though they may be achieving benefit, the instrument doesn't pick that up.
DR. REDMAN: A similar question for clarification. The patients identified their palliative endpoints on day 0, before treatment. If one of those endpoints was not pain control -- in other words, the patient put a level 1, I'm not having any pain, so you didn't consider that an endpoint. The patients on follow-up were only asked to assess those palliative points that they identified at the beginning, so if some other new symptom developed but they didn't identify it at the beginning, it was not recorded? This is for clarification.
DR. LEAVITT: Yes. I want to make it clear that on quality of life data, those were all prospectively chosen goals. However, anything that happened that made the patient worsen would have been reported as an adverse event. And for example, I showed you very briefly the adverse event data on the pain associated with the procedure in the immediate post-injection period. So, all of those would show up in our adverse event recording, but might not have had an impact on moving somebody up if they were already at the maximum goal. Maximum level.
Does that answer your question?
DR. REDMAN: Not quite. I think it's just clarification. A patient identifies four palliative factors that have a numerical value greater than 1 to them, but while on treatment, a factor that they didn't identify at the beginning becomes important to them. I think, at least the way I read it, they were asked to evaluate only the points they picked out at the beginning.
DR. LEAVITT: That's correct.
DR. NERENSTONE: Dr. Blayney.
DR. BLAYNEY: Yes, thank you. I have three questions.
How consistent, and what evidence do you have that your tumor volume measurements between injections and between centers, and which is really quite critical for the pharmacokinetics and the dosing, were consistent?
DR. HOWELL: The volume measurements were reasonably consistent. Recall that all the patients who were candidates for this trial had easily measurable lesions. They were preselected to be patients whose lesions could be easily measured. And there was a high reliability index in terms of being able to monitor changes in tumor volume, particularly under circumstances where a very large fraction of these patients actually attained not only CR but also a very, very good PR.
CT-scans were used often to look at the anatomy of the tumor, and sometimes to provide the depth measurement, but you recall that highly accurate measurements of tumor volume during this treatment phase are not really required for treatment success. You get good drug distribution, variable but good drug distribution, in the tumor by virtue of the opportunity to treat again and again, and precise dosing is neither required nor operationally clinically feasible in a wide variety of tumors. There's so much tumor heterogeneity that you basically have to do the best you can, as you would if you were infiltrating with lidocaine or another local anesthetic.
DR. BLAYNEY: And that is the second question. One of the photos you presented in your briefing document here, the poor fellow with the enucleation of his eye, you measured an MTT, most troubling tumor, that has an arrow pointing to it. A lesion in that eye socket that's superior looks like it waxes and wanes. If you had pointed the arrow toward that tumor, that could have been a complete response, is the way I view these photographs.
DR. HOWELL: That's a difficult photograph. Let me ask Dr. Elias to point that out exactly to you.
DR. ELIAS: I'm Dr. Laurence Elias, Medical Director with Matrix Pharmaceutical.
Why don't we go ahead and look at the slide of the patient you're referring to. Let me just walk you through this. This patient's most troublesome tumor was in this area here, and on placebo this grew. Then when the patient was crossed over to active treatment and had a good response of the MTT. Later there was a tumor superiorly, but the protocol permitted a treatment of tumors other than the MTT, and this was also treated and responded.
DR. BLAYNEY: You're talking about the lesion at the level of his helix there?
DR. ELIAS: Excuse me?
DR. BLAYNEY: The lesion at the level of his helix.
DR. ELIAS: I don't believe so. These are difficult lesions to photograph. We did not use these photographs for evaluating response, but are presenting them illustratively and use them to identify the MTT for the benefit of the investigators.
DR. BLAYNEY: My point is, though, it's very difficult to þ even the MTT tumors, if you'd have picked some that weren't necessarily injected, that you could have had responses just on the basis of happenstance, perhaps, or blood growth, or tumors falling off. I agree with you, these are difficult to measure.
DR. ELIAS: If we could have the next slide on. Again, the measurements were all done in the clinic by experienced investigators, and this is what really determined the responses we're reporting. And these pictures are illustrative.
But here is another patient who had an MTT that you can clearly see at the base of the neck, clearly in a threatening position. Now, this is a 44-year-old man who had originally a primary tumor at the base of the tongue, subsequently had multiple surgeries, radiation therapy, had several courses of cisplatinum/5-FU. If you go through his history, he was a convincingly refractory.
This patient went through the typical sequence of tumor necrosis, shown here at day 43 on treatment with cisplatinum/epinephrine gel, and then went on to have a very nice response with very nice healing at this point.
DR. BLAYNEY: And perhaps my last question is, it's very difficult for me to understand or to say that the gel is not a way of delivering epinephrine locally in a low dose of weekly cisplatinum and may have gotten the same tumor response, since you didn't do a control with epinephrine and your matrix material, your collagen material.
DR. ELIAS: Well, I think you can appreciate the difficulty of clinically doing multiple different types of placebo treatment, but I think the answer to your question is well addressed by some of the preclinical data we can show you.
DR. HOWELL: Actually, it sounds like your concern was the question of whether if one had just treated with epinephrine gel, whether one would have seen these same kinds of dramatic --
DR. BLAYNEY: Local epinephrine is painful, produces necrosis.
DR. HOWELL: Let me ask, would you expect that the injection of epinephrine to be able to manage this kind of lesion, substantial lesion here in the throat? Would epinephrine have been capable of winding up with a tumor reduction? I think most of us would think probably epinephrine alone couldn't have done this. Your point is well taken, and there was a very substantial debate about whether epinephrine should be included in the gel.
One has to be careful in selecting a placebo, that it doesn't cause patient problems. And epinephrine had the potential to do that.
DR. NERENSTONE: Dr. Lippman.
DR. LIPPMAN: I just had a few questions to get at the issue of the magnitude of the response and things that can affect this because as I understand it, most of the tumors were measured -- and it would be nice to see the data -- by clinical measurements. It would be good to see the data in terms of response by CT.
But the question I was getting at is the issue of being able to maintain the blind. In the document you indicate that the color of the active gel was different than the color of the placebo gel, and that it was put in a syringe to try to mask the difference. I guess the concern I have hearing the presentation is that the major problem that you couldn't get the drug to people is because the drug leaked, for lack of a better word. So, the drug came out. That was the major problem for not giving adequate therapy. If the gel is really a different color, it would be difficult to maintain the blind. And if the blind can't be maintained and the measurements are subjective, it makes it more difficult to determine the magnitude of the activity.
DR. HOWELL: Two points on that. You are right, there is a slight difference in color, and when the drug refluxes from a tumor that you've injected, it's usually mixed with blood, and that completely eliminates the color difference.
Let me ask Dr. Glenn Mills, who's had a lot of personal experience with this, to address that issue.
DR. MILLS: Yes, Dr. Lippman, there is a slight color difference between the gel and the placebo. I think if you put this on a white piece of paper you can clearly see the difference. But you know, with these tumors like you've seen, when you're injecting them and you get a little reflux, I found it very difficult to tell a color difference, if any, in my patients because of that admixture, some of the necrotic tumor, as well as some blood in it. I don't really feel like I could tell a difference.
DR. LIPPMAN: It didn't indicate what color it was. I just wanted to see how confident you were about maintaining the blind.
The other issue I have reflects sort of the dose response data. So, there was no difference in response rate at the higher versus the lower dose, but in fact I guess one aspect about that is that I was a little surprised that there was no difference in toxicity. I was having trouble finding the table, but there was significantly more nausea, I think, in the treatment group overall, and one might have expected that nausea would have been higher with the higher dose. So, I'd like your thoughts on that.
And the other issue of dose response in terms of activity -- again, there's no difference in the two doses that were used, but in fact, when one looks at the two pivotal trials, the one that looks most promising is the 414, which had a 34 percent response rate, a median duration of 85 days. The other study had a 25 percent response rate and a median duration of response of 64 days.
Yet, the 414 with the best results had the lowest compliance. In fact, that was really one of the major differences between the two studies, as I saw it, that particularly the group randomized to the IntraDose, 47 percent were able to take 80 percent or more of the drug. That's table 29, I believe, on page 46.
So, I'd appreciate your thoughts about these issues with dose response and dose that actually was received.
DR. HOWELL: Let me make just one point to start the answer to you. Remember that in local tumor therapy doubling the dose does not double the response rate, and you wouldn't expect it to. Doubling the dose does not get twice as much drug distribution within a local tumor nodule. So, the kinds of relationships between dose and biologic effect that we're used to dealing with in the intravenous world are different in a tumor, where we don't have such a rigorous and tight relationship between the dose actually gotten into the tumor and the response, because we don't always get the same distribution.
Let me ask Dr. Leavitt to address that point further.
DR. LEAVITT: That's correct. The question is, was there a change in response rate? The benefit rate and the response rate was maintained before and after the change in dose, and what you can see is in the 414 study the benefit rate was 42 percent. It dropped to 29 percent afterwards. The 514 is 24, 17 percent.
DR. LIPPMAN: So, there was a lowering of response rate, but not significant with those small numbers?
DR. LEAVITT: That's correct, and if you look here at response rate, you can see that the overall response rate went from 29 to 37 in the North America study, 29 to 22 in the ex-U.S. study. Overall, the 29 percent was maintained, and there's no difference statistically between these.
DR. LIPPMAN: And I wondered, do you have the response data from the combined studies with CT measured tumors? Do you have that available?
DR. HOWELL: No. CT scans were not used to assess the response in these studies. I would just point out that some of the cooperative groups have now ceased and desisted using imaging technology to assess the tumor response because of the complications of trying to image in irradiated fields and so forth.
DR. LIPPMAN: Just one final question. Do you have the data available, in terms of response data and the combined studies, on patients who failed cisplatin for recurrent disease?
DR. HOWELL: Yes, we do. Dr. Leavitt?
DR. LEAVITT: What we have are response rates in patients who have had previous exposure to cisplatin or carboplatin. Those are not always patients who had an immediate proximate failure of platinum. Some of those patients had had cisplatinum as part of initial management, and if you consider recurrence after initial management with adjuvant chemotherapy, then that's a failure.
Turning here, you can see amongst those patients who had either cisplatin or carboplatin -- and most of these, by the way, are cisplatinum -- 29 percent, 30 percent for those who are platinum naive.
DR. LIPPMAN: Sorry if I didn't clarify the question. I was looking specifically at patients who had failed cisplatinum for the management of recurrent disease, since there may be a difference in patients who receive, for instance, neo-adjuvant therapy had prolonged disease-free intervals and then recurred. Do you have the data by use of cisplatinum for the management of recurrent disease, then going on to this study?
DR. LEAVITT: I can get those data for you, I think, but I don't have those at my fingertips.
DR. NERENSTONE: Dr. Sledge.
DR. SLEDGE: A number of questions. First, with regard to inclusion and exclusion criteria, which are with regard to the universe of patients that we're talking about, I'm trying to get some sense of who actually is being treated here amongst all patients with recurrent head and neck cancer. On slide 6, I think it says that there is 7,500 to 10,000 patients a year with local disease who go on to die of head and neck cancer.
What's the real universe here, though, if we exclude patients with tumors greater than 20 centimeters squared, anyone with carotid involvement, anyone with carotid vascular disease, which I've got to imagine is reasonably common in this patient population, and anyone with systemic disease? What actual numbers are we talking about here?
DR. HOWELL: Let me ask Dr. Everett Vokes to address that issue.
DR. VOKES: I think that this would be a small number of patients, since those who have recurrent disease would first be considered for radiation; if they haven't had that, for surgery or chemotherapy. And you are excluding those patients with large bulky masses or those where carotid involvement cannot be excluded. So, I think it is a small number of patients. I could not give you a number for this nationally. I would estimate that it's maybe 2,000, 3,000.
DR. SLEDGE: So, 20 to 30 percent of the whole, roughly speaking?
DR. VOKES: At some point during their treatment.
DR. SLEDGE: When I look at the end of the briefing book, where it talks about the indications for the drug, most of these exclusion criteria are not mentioned. Does the company intend to ask for an indication that includes all the exclusion criteria used here?
DR. HOWELL: Yes. As I indicated, the indication has been refined since the NDA was submitted, so it's refined from what is printed on your question sheet today, and that is, we're talking about the patient population who are not candidates for surgery, not candidates for systemic chemotherapy, not candidates for re-irradiation, or simply have refused all of these things. So, it is a very small, narrowly defined patient population.
DR. SLEDGE: On slide 50, where you have the adverse events reported in greater than 8 percent of patients, for those of us who don't add very well, can you give me some sense of the total percentage of patients having grade 3 or 4 toxicity on the treatment arm versus the control arm? Total percentage. Not total number of patients, but total percentage of patients with grade 3 or 4 toxicity.
DR. ELIAS: The severe toxicities were relatively rare. Not seeing the data totaled in exactly the way you're asking for, but please note we can go ahead and look at that table again, that severe toxicities were relatively rare.
DR. SLEDGE: I ask because they seem to be more common in the treatment arm than in the control arm.
DR. ELIAS: They certainly are. The most common toxicity is pain, and all pain in all categories -- I think your question is about summing across several categories. Pain in all of these categories comes up to roughly 60 percent of patients, all grade.
DR. SLEDGE: It seems to me to be more than just pain.
DR. ELIAS: Any episode of pain during the entire course on observation, on study in blinded phase.
DR. SLEDGE: It seems to me that in just about every category there's more in the treatment arm than in the control arm. So, I'm trying to get some sense of the total number of patients, total percentage of patients who experience a severe side effect.
DR. ELIAS: We don't have it summed in exactly that way, but we do acknowledge that there are side effects and toxicity with this, as with any other treatment or medication, and that as I've just pointed out, the most common toxicity is pain, which could occur as a local immediate injection pain, as a pain in the local area, or a systemic pain in some other area. But this occurred more frequently in the treated group than in the control group, but it was not terribly common.
DR. HOWELL: Dr. Sledge, maybe it would be useful to hear directly from one of the investigators about whether pain or any of the other symptoms were particularly severe. Let me ask Dr. Glenn Mills again to address that issue.
DR. SLEDGE: You don't have to do that. I'm just saying, using what you call severe.
DR. HOWELL: Sure. The numbers are slightly different but they're not big differences, and we know that this is a product that causes -- and purposely we want it to cause -- local necrosis and some pain in that area.
DR. SLEDGE: But a lot of the side effects listed as severe appear to be systemic side effects to me.
DR. HOWELL: I think if you look down the list of systemic side effects, the event rates are low and the differences in the event rates are pretty modest.
DR. SLEDGE: But I'm asking in toto, since in just about every case, it appears to be more in one than in the other.
DR. HOWELL: I apologize. We simply don't have it available here now for that analysis.
DR. SLEDGE: Finally, for those of us who live in a centimeter squared universe, as opposed to the centimeters cubed universe, how do you measure centimeters cubed in these patients?
DR. MILLS: Well, I'm not a mathematician either, but the formula that Matrix supplied for us to calculate they tell me was based on a spheroid, which is length, width, height, times one-half. That's basically how you determine the volume determination.
DR. SLEDGE: When this gets into the clinic with the general medical oncologist, do you think that will be an easy switch?
DR. MILLS: I think for this product, yes, because it's a volume calculation and it's a volume per volume dosing. It's not a dosing based on creatinine or white count. So, you have to figure the approximate volume that you need. And I think that's good because when you treat these tumors, you do get an idea when you're injecting them whether you're getting good coverage because the gel does swell the tumor a little bit and you can see where you've injected.
DR. SLEDGE: I'm not saying it's bad. It may be the right way to do it for all tumors. What I'm asking is that if you were a clinician who sees 5 or 10 of these patients a year and are not used to the measurement method and you're giving a dose which is based upon the volume of tumor, what I'm asking is, do you think this will be an easy switch.
DR. MILLS: I think it would be an easy switch. It wasn't that difficult.
DR. NERENSTONE: I'm going to take the chair's prerogative for two quick follow-up questions to Dr. Sledge. Because you are basing a lot of this on clinical benefit, what about the duration of the toxicity? Do you have a slide about that?
DR. HOWELL: Let me give you a quick answer to that while they're getting the material. The local necrosis affects the swelling. The erythema resolves usually over a period of 20 to 40 days, sometimes taking slightly longer in some patients, but it is a fairly predictable, clear process of erythema inflammation followed by healing.
DR. NERENSTONE: At what point then do you calculate your duration of response? Because your duration of response median is 70 days. So, do you have the duration of response after the 40 days, or is it at the time of first change of the tumor?
DR. LEAVITT: Response duration was calculated very directly from the time of first onset of response, at least a 50 percent decrease in tumor, to the time of relapse.
DR. NERENSTONE: So, part of the 70 days duration could be at a time when patients are having more pain because it takes 40 days for the tumor pain to go away.
DR. LEAVITT: That's correct.
DR. NERENSTONE: And the second question, just a clarification. Dr. Sledge said something about systemic disease. My understanding is that systemic disease did not preclude enrollment on this trial. Is that correct?
DR. LEAVITT: That's correct. Patients had to have local dominant disease, and patients had other distant metastases that were symptomatically dominant, they should have gone on to receive other chemotherapy if they were otherwise good candidates for it.
DR. NERENSTONE: Thank you.
DR. KELSEN: You described in the briefing book why you decided to develop your own palliation scales. Did you compare them to scales, for example, for pain, which are well described by other investigators, such as MPAC, or any other instrument that's been validated. That's question one.
And question two is, how many patients using your scale moved two levels of improvement? That is, not from I guess going down from one to two, but from two to three steps down.
DR. LEAVITT: Let ask Dr. Morgan Stewart from Matrix Pharmaceutical to address that.
DR. STEWART: I'm Morgan Stewart. I'm the Director of Biostatistics and Data Management.
To take the second question first, there actually were not a lot of patients who had a two-grade increase, or actually it was a decrease, corresponding to an improvement. As was pointed out earlier, we purposely, when we designed the instrument, made these grades distinct from each other, and it would have been very difficult, although a few patients did do it, to have a sustained for 28 days or more displacement of two grades from baseline.
DR. KELSEN: Why would it have been difficult? I would have thought a patient who has level 3 pain -- that is, has to take a prescription drug, then going to either Tylenol or aspirin or no pain. How many patients actually were able to do that? Why do you think that's so difficult?
DR. STEWART: I believe there were three in the two studies. I'd have to look at the data to be sure. Remember that these were advanced patients.
DR. KELSEN: And did you compare this pain instrument, your pain scale, to other pain scales?
DR. STEWART: Yes. One of the other instruments that we used on these studies was the FACT head and neck scale, and that includes a pain question, asking -- I think it's a 6- or 7-point scale -- about current pain status. We looked at our patients who had selected pain as one of their goals versus what they had scored on the FACT.
Now, unfortunately, we've had a lot of problems getting patients to fill out the FACT. The compliance was low. And so we've been told by the developer of the FACT that we should interpret any data having to do with the FACT -- it's almost worthless because we had less than 50 percent of the patients who had a FACT score recorded after baseline.
So, while we did see some degree of association between patients who said that they were getting better on the treatment goal questionnaire for pain, also getting better on the FACT, it's difficult to interpret those data because of the low compliance on the FACT.
DR. KELSEN: So, could I just follow that? That means that they did do a FACT-head and neck, as well as this instrument.
DR. STEWART: Yes.
DR. KELSEN: But you don't have data from the FACT-head and neck.
DR. STEWART: We don't have reliable data because of the low compliance.
DR. NERENSTONE: Mr. Gruett.
MR. GRUETT: Looking at your background document, the mixing of the three chemistries involved, I have a follow-up question to this also. Why can't this be done ahead of time? Why does it have to be done just prior to injection?
DR. HOWELL: I'm sorry, I'm not entirely sure. Why does the mixing have to be done?
MR. GRUETT: Have to be done ahead of time? Why does it have to be done just prior to injection?
DR. HOWELL: This product doesn't contain any preservatives of any sort, and it's not necessarily as stable as a formulation with all three components, the epinephrine, the cisplatin, and the gel mixed together, for a long period of time. So, since the drugs are easily mixed together in the syringe immediately before injection, the approach was to do it that way rather than trying to develop a product that had a shelf life where all three things were mixed together.
Am I answering the question?
MR. GRUETT: Yes. That brings up a concern about the shelf life within the tumor and the potential of toxic breakdown of the drug within the tumor. Do you have a half-life or table showing the length of longevity the drug has before it breaks down?
DR. HOWELL: With cisplatin the pharmacokinetics are a little bit different. The product that this drug breaks down to is something that is exited from the tumor very rapidly. This drug has to get into the tumor cell, has to undergo activation inside the cell, and then reacts with the DNA. We don't expect in any clinical circumstance to see a lot of "breakdown" products that are toxic in the tumor or the plasma. There are some metabolites that have been inactivated, but we have strong evidence to indicate that the drug stays in its active form in the tumor for quite a long period of time relative to when the drug is injected just as a free solution.
MR. GRUETT: I didn't see any studies at all in your background information on this. Do you have those?
DR. HOWELL: Let me ask Dr. Robert Tressler to show you an example of the effect of the formulation on retaining the drug in the tumor.
DR. TRESSLER: If I understand your question correctly, you want to know how the gel product was broken down?
MR. GRUETT: Yes, and what happens in the tumor. How long does it exist in its present property?
DR. TRESSLER: I don't actually have a histographic slide prepared for that, but what I can tell you is we did look at that and did do a series of preclinical studies and histologically assessed the time course of absorption and dissolution or breakdown of the collagen gel product containing cisplatin. What we showed was that intratumorally, and also in normal tissues, collagen breakdown started to occur within 7 to 14 days. And by day 30 to 60 we could no longer detect any presence of the collagen matrix in the tissue by histological examination. And we looked at a variety of tumor types over a time course.
So, we see very nice bio-absorption, if you will, without significant changes.
DR. NERENSTONE: Dr. Pelusi.
DR. PELUSI: I believe my questions will go to Dr. Mills, since I think he's had the most experience. Dr. Mills, could you describe for me the patients as they came in. Were they all done as outpatients, and did they have to stay in the town where they were treated?
DR. MILLS: All the patients that I've treated on Matrix trials were done as an outpatient. I have a healthy respect for cisplatinum since when, I was a fellow it was the pre-dansetron days, so every one of my patients was pre-med and I had no nausea and vomiting. I gave them Demerol.
The pain that we're talking about in this procedure is the pain like in our clinic if we do a bone marrow aspirate and biopsy. It's similar to that pain. It's a transient pain, lasts a day or two, and then it's gone. It's not a real long-lasting pain.
If the patients came from very far away -- I had patients come as far as 250 miles -- I did ask them to stay overnight initially because this was very early on in the trial and I really wanted to keep them around for 24 hours. Subsequent patients and some that I'm treating right now on another study I let go home at the end of the day now because I feel comfortable with that.
DR. PELUSI: The reason that I ask that is many of the patients that I particularly deal with come from very long distances.
The other question also becomes, when you look at the criteria for patients that would utilize this, is there a concern for you in those that refuse other modalities because maybe other modalities would ensure that they have to stay in another town for six weeks, seven weeks in terms of other treatments? Do you see that actually bringing more patients to utilize this medication versus utilizing other forms of treatment that may be indicated?
DR. MILLS: I guess that's a difficult question for me to answer. You know, when I counsel a patient I give them their treatment options, and I really let the patient make their informed decision on that. You know, if the patients have systemic metastatic disease in this setting, I would really push them for systemic therapy if they're a candidate. In fact, all the patients I treated failed at least one or, in some cases, two or three types of chemotherapy. And I think that still is the number one management tool that we use in this patient population
But there are patients where the local disease itself -- it's just like a patient with a cord compression. Maybe they've got breast cancer but now they've got a cord compression. Well, we're going to take care of that local problem in a brief period of time and then put them back on their treatment. That's sort of how I view this product if I get to use it in the clinic. The patient has an obstructing lesion, I'll treat it with this product, and then go on to another therapy or another therapeutic option in the future.
DR. PELUSI: And just one last question. In terms of symptom management, as you ran both of these trials, were there standard protocols for the different types of symptoms that were experienced, so we know if indeed this comes on the market how best, if there is a best way to treat some of these symptoms?
DR. MILLS: I don't recall right offhand. It was just recommended good medical management in these patients for their pain. All my patients took an oral opioid, oxycodone, something like that.
It brings up another point, though, that I wanted to make, and I think Dr. Kelsen was really concerned about this earlier. Even though these patients may have had pain locally here, they had other symptoms and other things going on elsewhere, which can sometimes confound things, as we know.
I had patients that did benefit from pain relief and a significant decrease in narcotic consumption, but couldn't score a hit on this fourth step because even though they said, yeah, I'm better, my pain's better, I'm not taking breakthrough medication, they're still taking medication, and maybe taking it because they have other problems elsewhere as well.
DR. PELUSI: Thank you.
DR. NERENSTONE: Dr. Przepiorka.
DR. PRZEPIORKA: Three questions, if I may. I believe I heard that patients who had multiple tumors could be treated with multiple tumors at the same time, but you only followed the MTT for response. How many patients were treated for multiple tumors, and did you see any responses in the non-MTT that did not occur in the MTT, or vice versa?
DR. HOWELL: You are correct. The protocol permitted multiple tumors to be treated, and we do have data on that available.
DR. LEAVITT: Yes, we do have those data. The response rate in all treated tumors is very similar to the response rate for the MTT. In fact, it did prove that the tumors that were chosen as the MTT were also the most difficult in which to obtain a response.
DR. PRZEPIORKA: While you're waiting for the slide, just the second half to that question. I also believe I heard that patients had systemic disease and would be treated with chemotherapy. Was that the same time they were on this protocol, or was systemic chemotherapy held until the end of this treatment?
DR. LEAVITT: In no case during this protocol did we give concurrent systemic chemotherapy and cisplatinum/epinephrine gel. If patients were candidates for systemic chemotherapy and they needed system chemotherapy, they were not and should not have been entered on this protocol. We did have patients who completed this protocol and then subsequently developed more problematic distant disease, or first had noted systemic metastases, then went on for chemotherapy, but no, there was no delaying tactic here.
I do want to answer your question about all treated tumors. Now, these are patients who were stratified according to the most troublesome tumor size. And now looking at all of the tumors that were treated, not just the most troublesome tumor, the overall response rate is 30 percent for stratum 1 and 2. Stratum 1 I patients had 36 percent, 18 percent overall. And you can see the total number of tumors treated is 227.
DR. PRZEPIORKA: The draft package insert indicates that you would recommend a maximum of 10 mls of the gel per treatment. But your table indicates that patients received a median of 1 to 2 mls per treatment, with a maximum, in some of the studies, of 8 mls. How much data do you have in the 7-10 range to really support its safety?
DR. HOWELL: The vast majority of the patients were dosed with a median of 10 milligrams per meter squared of total platinum dose. The choice of 40 milligrams total dose as the recommended upper limit for any one treatment session simply was based on systemic toxicity safety concerns, and we didn't want to get into a situation where we were exposing the systemic circulation to a large volume of tumor. The principle on which this whole approach was based was to try to decrease systemic exposure while at the same time increasing tumor exposure.
Does that address the issue, or would you like to see some data?
DR. PRZEPIORKA: Well, I guess you just indicated that most of your patients were treated with 25 percent of what you are recommending as the maximum dose, so I wanted to know how much data do you actually have to support the safety of going up higher than that to the maximum dose that you recommend.
DR. HOWELL: Let me ask. Do we have data on that point?
DR. ELIAS: Well, the toxicity and the AEs we reported are across the range on an intent-to-treat basis, before the amendment, after the amendment, larger tumors, and included larger tumors that would not be included in the current labeling indication, included tumors that were larger than 20 centimeters cubed, as well as included patients treated at the original 0.5 dosage level. The systemic toxicities in any case were very modest and not at all comparable to what's seen with systemic intravenous cisplatinum.
Why don't we go ahead and look at slide 238. This shows adverse events by cumulative dose, which perhaps goes the most directly to your question. Now, in understanding the slide you need to remember that this includes a factor of time. In other words, it's cumulative dose over time. So, patients who had the larger doses cumulatively also may have included patients who are on study longer.
Nonetheless, the differences are there but are relatively modest. I believe you were mainly concerned about the nausea and vomiting, and in patients who had the larger dose range, this goes up to a maximum of about 33 percent. Again, this is all grades, and it needs to be compared to the placebo of 10 percent.
So, the dose-response effect is probably there but is relatively modest. I think the data well supports dosing within the range that we would intend to be included in the labeled indication.
DR. PRZEPIORKA: And my final question is, what are your plans for educating physicians on how to administer this?
DR. HOWELL: Dr. Leavitt?
DR. LEAVITT: We think that this is a unique form of therapy, and that with the availability of IntraDose for the treatment of patients with advanced head and neck cancer, we think that there should be an education program. This should involve both medical meetings and presentations and should involve medical grand rounds, surgical grand rounds and similar kinds of programs. We are committed to making sure that physicians, be they medical oncologists, otolaryngologists, head and neck surgeons, have thorough understanding of the treatment methodology, the patient selections that are appropriate based on the studies that we've shown you, and the appropriate use of the product, and any of the side effects to expect and how they should be managed. Matrix will support this product in the marketplace.
I'm sorry that I couldn't give you a direct answer to an earlier question about previous chemotherapy, and I don't have the numbers of patients who had had previous cisplatinum for relapse. I do have some information on the patients who had had any chemotherapy at the time of relapse. Much of that was systemic cisplatinum. Would that be helpful to you?
DR. LIPPMAN: No. I was really interested specifically in the platinum.
DR. NERENSTONE: Dr. Albain.
DR. ALBAIN: Thank you. I have two questions.
First, perhaps for the study statistician. Could you comment on the rationale for having two very small trials going on in parallel with a 2 to 1 randomization? What were your thoughts on that, since most of the analyses that have been presented are pooled analyses? And what were your early stopping criteria for those trials?
DR. STEWART: Well, to take the first question first. We don't consider that these were small trials. This is a rare disease. It's an orphan indication, and I'd like to point out that it took about six years to fully enroll each of these trials.
The reason for the sizing of the trials, the 90-patient total sample size in each trial with the 2 to 1 randomization was based, as was mentioned in the presentation, on ability to detect a difference in most troublesome tumor response rate of about 30 percent between the placebo, which of course we didn't expect to respond at all, and the CDDP/epi gel treated tumors.
DR. ALBAIN: Why did you not just do one trial? You had two trials going on. Since you presented pooled analyses.
DR. STEWART: I believe that this has been touched on earlier, but maybe not in enough detail. We wanted to do two randomized placebo-controlled trials to meet the regulations for product registration. Because of the shift in the importance of patient benefit as an endpoint, we were well into the trials before it became apparent that we were going to be required to be statistically significant. The logical thing to do in a case like that is to pool the data, especially when you have identical trials.
DR. ALBAIN: Did you have early stopping rules? I may have missed it in your briefing book.
DR. STEWART: Yes. I'm glad you brought that up. We did not have a formal stopping rule for either of the trials, and there was a reason for this. We had a data safety monitoring board which regularly reviewed the data. Every six months they reviewed the data from both trials, and they reviewed the data in an unblinded fashion. We weren't allowed to be there when they were looking at the unblinded data. And all of the members of the data safety monitoring board had no affiliation with Matrix, other than being our consultants to be on the data safety monitoring board.
It was at their recommendation, actually at the recommendation of Dr. Steve George, when he agreed to serve as the statistician on our DSMB, that we not have a formal stopping rule because he felt that, in some ways, it tied the committee's hands. They wanted to be able to assess the data on an ongoing basis and be flexible with regard to the recommendations they made.
DR. ALBAIN : Thank you.
My other question has to do with dose again. We heard that there probably is not so much a need to consider dose response in the way we usually think of it for higher doses, but what about lower dose? Is there a lower dose boundary for efficacy? Are we concerned at all because of these issues of leakage that we've heard about, this nonsignificant trend that a lower dose may not have as high a response rate in those tables we saw earlier, I think in Dr. Lippman's questions? Is there any data to reassure us that with learning curves, with some dose leakage, and at this lower dose after the amendment, that there still is reasonable efficacy?
DR. HOWELL: Let me start answering that question by just saying that remember that if some of the drug leaks on the first injection, you get the opportunity to come back a week later, after the tumor has reduced in some volume, because you killed some part of the tumor and you get a chance to give it a go again. And that's just the practicality of the clinical reality. Some of these tumors accept the full planned dose and some of them don't. Because of the heterogeneity of tumor size, consistency, location, that isn't invariable; it's easily controlled.
But the bottom line is that in the end you have a pretty good response rate, even given the limitations of trying to be sure that the drug is getting into each tumor on each injection.
DR. ALBAIN: But do we know that we even need this lower dose? I guess I'm trying to get at, have we done some studies, perhaps some small phase II studies, that you need at least a certain dose?
DR. HOWELL: No. We do have phase I clinical trial data, and I'll ask Dr. Leavitt to address that. But from the basic pharmacology of cisplatinum, when you're introducing drug at this kind of concentration, any part of the tumor that is accessed by 4 milligrams per milliliter cisplatinum is going to be injured, even if it's not a large fraction of the tumor. That part of the tumor that's accessed is going to be injured.
A phase I dose-ranging trial was performed. Would you like to see the data on that?
DR. ALBAIN: Summarize it, perhaps.
DR. HOWELL: The summary is that dose-ranging was done over quite a wide range of things, and that trial included a variety of different tumor types, as well as patients with head and neck carcinoma, and the dose range of .25 milliliters per cubic centimeter of tumor volume simply turned out to be something that in head and neck tumors, with their slightly more scarce qualities, was on the average reasonably well accepted.
DR. NERENSTONE: Dr. Couch.
DR. COUCH: My question is regarding patient selection, which is always a critical issue when you have these novel therapies. In the introduction it discussed that the patients that had obstruction, especially airway obstruction, would be potential candidates. According to your background package, the majority of tumors were in the neck, and these were the ones that I would think would be most likely to obstruct the airway.
My concern is that I'm worried that you're not going to be able to define the proximity of injecting this near the carotid arteries, which is certainly in the neck. For instance, after total laryngectomies, the carotid arteries are medialized and there's not much soft tissue there.
And then also the other issue is the wound. There is a worsening of the wound with eschar, a necrosis and erosion, so you don't want to have wound problems near the great vessels either.
Is there going to be a way that you can help physicians best select patients that will keep them out of trouble, especially in these cervical lesions, which unfortunately don't seem to respond as much as the facial and oral lesions?
DR. HOWELL: Let me ask Dr. Mills to give you a specific example of that situation and then Dr. Wenig to comment as a surgeon on how the anatomy --
DR. COUCH: I guess what I'm really getting down to is in your exclusion you say in close proximity to carotid artery. I really think it might be important to better define that.
DR. HOWELL: Let me address that issue exactly.
DR. LEAVITT: I'll comment from the medical oncologist's perspective, and also Dr. Wenig from the surgical perspective. He was also involved in these trials.
You are right. When these tumors do recur in the neck, involvement of the carotid artery can be a major problem. This is obviously a patient who would not be a candidate -- not be a candidate -- for therapy. And this is a large necrotic tumor, a stratum 3 tumor, if you would, with obviously carotid involvement at this point in time. I think when I screened patients with cervical disease, I obtained to CT scan, and that was my first step.
The carotid does tend to migrate medially in these patients who've had a lot of neck surgery. Here is a tracheostomy and here is a peristomal recurrence, and the carotid is right there, but there is a definite strike between the tumor and the carotid so there is clear separation. I like to see a very good separation from the carotid in any tumor I would consider treating in the neck, and I think this is where the CT is very important. I use the CT to not only help me define the anatomy, but also on your physical exam, the length and width were very easy. Sometimes the depth or height was confusing, and the CT could help us right there on our initial planned treatment dose, to make sure that our physical exam was correct.
I'll ask Dr. Wenig if he'd like to expand on that from the surgical perspective.
DR. WENIG: I'm Barry Wenig from Northwestern University.
By way of background, I treated 4 patients on the 414 North American study, and I treated 1 subsequent patient on the follow-up study. So, I have a total of 5 patients with hands-on experiences.
This is an example of a patient treated from Europe. It's not my patient. But I think it's fairly representative because it points out several factors. Your question about the neck wound care is illustrated here. In this first slide, I think in my experience certainly, and I'm sure in yours, when tumors recur in the neck, there is often breakdown of the skin, and if not the skin, then certainly there is wound breakdown in some way, shape, or form. So, we're obligated to treat those patients or take care of the wounds in some manner. And there's really no good way to control that that I know of, other than just the local wound care that we're offering those patients.
Following treatment with IntraDose, there is a very predictable pattern to the wound healing process. The eschar forms first. The necrosis comes next, and then the re-epithelialization comes third. It's fairly typical and standard, and it doesn't differ from patient to patient. I think that it's easy to take care of those wounds and you really don't have to worry about the depth of your injection causing the effect at the deepest recesses of the wound because it generally happens in a very superficial manner that all this occurs. So, potential damage to the carotid, which you would think would happen from deep wound necrosis, for lack of another term, really doesn't happen so much.
DR. NERENSTONE: Dr. Couch.
DR. COUCH: What defines close proximity? The answer of a plane -- is there going to be more of a definition than that for what constitutes a patient that is appropriate for this therapy?
DR. ELIAS: Well, I think we can clearly contraindicate the dangerous situations based upon our large experience in this disease in this phase III trial. Recall, when the amendment was instituted, directions were given to the physicians and these directions were remarkably effective at reducing the incidence of related events and the subsequent part of the trial, which was more than half the trial.
Specifically, the occurrence of these events were reduced to 0 zero in more than 100 patients, and this is unlikely to have happened by chance alone. So, we would use in our indication wording very similar, if not identical, to the amendment wording, which clearly was meaningful to the physicians and useful guidance, and could be interpreted by physicians, as Dr. Mills indicated.
DR. COUCH: I don't mean to beat this, but it's such an important point. One last question. So, once you made the amendment, patients that had cervical disease, the same percentage was treated. In other words, what types of patients were then excluded by these physicians so that they did not have complications?
DR. ELIAS: The amendment excluded patients who had tumors directly involving or adjacent to the carotid artery.
DR. COUCH: So, a decrease in the percentage of patients with cervical disease was then treated in the protocol after that?
DR. ELIAS: I don't know if the percentage of patients with cervical disease changed after the amendment. The amendment also excluded patients with larger tumors, and we would continue to not recommend treating larger tumors, tumors greater than 20 centimeters cubed in the head and neck region.
DR. HOWELL: You are correct. There was a slight decrease in the absolute number of cervical patients, appropriately so. Once you take that subset of patients out, there are going to be fewer cervical patients.
DR. RUBENSTEIN: We understand that patient benefit was made a co-primary endpoint in mid-trial, and it also appears that it was defined as a single endpoint in mid-trial too, that up to that point there were multiple endpoints and they were defined as a single one in mid-trial.
What other aspects of patient benefit were defined in mid-trial? Was the 28-day requirement introduced in mid-trial? Was the 1-point scale difference introduced in mid-trial?
DR. HOWELL: I apologize for being unclear. There was initially one single primary endpoint. There weren't multiple endpoints that were collapsed together in some way. In an attempt to find a way of quantitating clinical benefit, and working with the FDA, struggling with this, we tried to come up with the patient benefit algorithm, and eventually the FDA asked us to analyze that as a co-primary endpoint in the trial. Dr. Morgan can certainly speak to that in more detail if you care to.
DR. RUBENSTEIN: No, no. So, the endpoint itself was defined from the beginning. It's just that it became a co-primary endpoint in mid-trial.
DR. HOWELL: That is correct.
DR. NERENSTONE: Dr. Lippman.
DR. LIPPMAN: Just to follow up on the amendment and the toxicity issue that was discussed in terms of cerebrovascular events. Excluding stratum 3, greater than 20 centimeter tumors, I had the impression from the briefing document on page 31, that it was done because of a low response rate. Was it done because of a low response rate, or concern about increased toxicity, or both?
DR. HOWELL: Was the dose reduced because of a low response rate?
DR. LIPPMAN: Well, no. There were several amendments made at the same time, I guess.
DR. HOWELL: Right.
DR. LIPPMAN: The question raised here about the concern about proximity of the carotid and so on, that in response to that answer, and why you're not as concerned that this major side effect would occur with the new dosing, I had the impression that the main issue there was that the dose was reduced. But as part of the last answer, there was a comment also that the stratum 3 patients were reduced.
So, the question I have is, what was the basis for removing the stratum 3 patients? Was it based on low response rate, or increased toxicity, or both? And I'd also, if you have it, I'd like to see the data on those patients' response to toxicity.
DR. HOWELL: Sure. I think there are several parts to that question.
One is -- and I'll ask Dr. Leavitt to address it. One is there were two changes that were made simultaneously in amendment five. One was a change in patient eligibility that excluded the patients with carotid tumor involvement, no matter what the size of the tumor was. And the second was the change in the recommended dose. And the change in the recommended dose came about simply based on the experience in head and neck tumors, where it was discovered that on the average they just weren't accommodating quite as much gel as many of the other types of tumors that have been studied.
DR. LIPPMAN: Wasn't there a third major aspect to that, and that's excluding the large tumors? I thought there were three major changes.
DR. HOWELL: Yes, you're right. That was a third element in the amendment. Let me let Dr. Leavitt address that.
DR. LEAVITT: Specifically at that point in the trial, we examined all the data on all of the patients. We looked at the patients in stratum 3 and removed them from further entry in the trial because in advanced relapse cancer, it was clear at that time that stratum 3 patients had both a lower response rate and less opportunity to benefit. And because of that, that was the reason that they were removed from trial.
We looked at the response rate in stratum 3, and overall this was a 13 percent response rate and because of that we simply felt that this was too low a response rate to justify any increased toxicity in this group. When you also look at the benefit rate, which I'm not showing you -- well, let me do that.
Now focus just on stratum 3, looking at the benefit rate. This is also lower. It was 16 percent. And given the low response rate, the lower benefit rate, it seemed that it was not wise to keep these patients in trial.
I'll also add that this study was designed initially only for patients that we call stratum 1 and 2, and it was at the insistence of investigators that they also had an additional group of patients for whom they essentially didn't have good alternatives for therapy. They felt that it was important to explore whether this new therapy might be an important breakthrough for such patients. I hope in the future we can find a way to deal with patients who have such large tumors, perhaps further studies that look at the different treatment paradigms using IntraDose, either alone or with other therapies, might bring some benefit to these patients.
DR. LIPPMAN: The reason that I'm focusing on this group for one is that, as you know, this is going to be the most tempting group to treat, and that's I think why the investigators asked for you to open it up to that group. And I think when you talk about how these tumors are measured, I don't know that precisely we can distinguish 19 centimeters from 21. So, patients will be treated with these big tumors for both reasons. They have nothing left. It's a large group of patients. So, that's why I'm sort of getting at that.
Along the same lines, I would like, if you have it, to see the toxicity that was observed with these patients with large tumors, the stratum 3 patients.
DR. ELIAS: Again, as I mentioned before, we've attempted to report the toxicities comprehensively across the entire clinical experience. Here we do have it broken out as you request, according to stratum 3 versus stratum 1 and 2, and stratum 1 and 2 are further subdivided into the pre-amendment dosing and the post-amendment dosing.
This shows the most common toxicities in the moderate to severe grade. And the immediate injection pain is slightly less in the stratum 1 and 2 with the lower dosing. The moderate to severe local pain, which is the pain that can occur after the immediate injection, is also slightly less as you move to the current dosing indications, but it's not usually different.
Nausea and vomiting is slightly more severe in the stratum 3 with the initial dosing, and in terms of the local cytotoxic effects, the effects of local tumor breakdown, they change modestly with the stratum and the dosing change, particularly erosion and ulceration is a little bit higher with stratum 3. Necrosis, which is statistically correlated with response actually remains roughly the same across these different categories. So, there's a bit of a shift but not any big surprises I don't think.
DR. LIPPMAN: Thank you for showing those data.
I just have two quick follow-ups to what was just asked. When the issue of hospitalization versus outpatient therapy was brought up, Dr. Mills said that he treats all patients as an outpatient, and of course, he's very experienced with this and one would expect that would be the best case scenario. Of the data of the entire population of patients, how many required hospitalization or were hospitalized? Do you have those data?
DR. LEAVITT: We don't have those data because most often these patients were treated in an outpatient setting, but sometimes they were in hospital as a place to get treatment. In Europe frequently it is easier to have patients admitted to hospital rather than treat them as an outpatient. There are various cultural reasons for that. And even in the U.S. patients who are traveling from a distance to come for therapy did get hospitalized, so that we don't really have clear data that show you how frequently this could have been done as an outpatient.
DR. LIPPMAN: I think in the briefing document you indicated that they were treated as an outpatient or limited hospitalization.
DR. LEAVITT: That's true.
DR. LIPPMAN: So, your answer is that most of the time the hospitalization was because they were there already or cultural.
DR. LEAVITT: That's correct. These limited hospitalizations are really overnight admissions, sometimes 23-hour admission.
DR. LIPPMAN: Then one final question to follow up on Dr. Przepiorka's question. I thought, when she asked the question, you were going to show data on the whole issue of mixed responses, not the overall. I've seen the overall data that the response rate is about the same in the non-most troubling tumors and the most troubling tumors.
The question is, did you have cases, and what was your data on the discordance where you had, as Dr. Blayney I think brought up, a response in a non-MTT? I guess more of a concern to me would be a response in the MTT and a progression in the non-MTT. Do you have the data presented that way? Not overall, but discordance of response.
DR. LEAVITT: I'm sorry, I don't have those summary data, but would you accept an overall view? We never saw a distant progression of a treated tumor when an MTT was responding. The MTT in each and every individual patient was the least likely to respond of any treated tumor. Now, some of that may be the nature of the tumor, and some of that may simply be the treatment regimen, where we asked the investigator to start and make sure that he treated the MTT preferentially, so that in fact those tumors may have been most thoroughly treated.
DR. LIPPMAN: And a related question. Was the MTT always the largest tumor that the patient had?
DR. LEAVITT: Usually, but not always. A critically placed smaller tumor, for example, near an airway might have been designated as the most troublesome tumor.
DR. NERENSTONE: Dr. Temple?
DR. TEMPLE: I have a few questions about what the benefit that has been shown was. I guess my first one is that in the middle of the study, you changed from a primary endpoint of response rate, which you thought you had some idea about, to a co-primary endpoint, which I don't know what that means exactly, but it means you have to win on it, I suppose, of clinical benefit. But you didn't adjust the study size.
Now, it's perfectly obvious that responses are going to be more obvious and show a bigger difference than clinical benefit. Can you say something about why you didn't change the study size?
DR. HOWELL: I'll ask Dr. Morgan to address that.
DR. TEMPLE: The result of that is that no single study showed any clinical benefit, and you're going to ask people to look at it pooled. But why did you make that necessary?
DR. STEWART: Well, I think you have to look at the full evolution of the studies from 1994 to the unblinding in the year 2000. I used the word "evolution" earlier, and I think it really is the best word to use to describe what happened to the study design over that period of time.
I realize that there is some disagreement that's come through in the briefing books about when patient benefit became the primary endpoint, but certainly our perception was, although a clinical benefit endpoint was discussed, was included in the studies, and was known to be very important from the time the protocols were written, that it wasn't clear that this was something that the agency wanted as a primary endpoint until 1997. And it wasn't formally designated as "a co-primary endpoint" until actually last year.
So, as the studies evolved, we had reached a point where it would have been, I think, very awkward to increase the study size. We already had some data on board. We were halfway done with the studies. There's a lot of statistical issues that come into play when you have part of the data, even though you haven't unblinded yet, and you're restating your power and recalculating the study size.
I'd also like to point out that the issue of resizing the studies, based on another or new or newly recognized primary endpoint, was never an item of discussion between the agency and Matrix. It literally never came up. So, we felt there wasn't any need to do that.
DR. TEMPLE: Actually adjusting sample size under those circumstances with people blind has been addressed by many statisticians, and you actually don't pay much of a price. But I understand there may have been confusion.
So, let me be sure that we understand. On the patient benefit -- this is your slide 35 -- which includes both palliation and prevention, which appears to be the primary endpoint, you are not saying that either study showed a significant difference, but you do say that the combined studies showed a difference with nominal statistical significance. Is that fair?
DR. HOWELL: That is correct.
DR. TEMPLE: And it's not easy to tell because you don't quite do it that way, but if I look at the next slide, like slide 37, it looks as though 13 of the people who benefitted, roughly, had a palliation endpoint and most of the rest had a prevention endpoint. There seems to be some double counting and I can't quite sort it out, but is that more or less right? Because the prevention endpoints are described in figure 42, and there seem to be 26 versus 6, so that seems like a big part of the difference in the overall patient benefit. Is that true?
DR. STEWART: Well, there was not double counting in the sense that patients had a double shot.
DR. TEMPLE: No, I understand. But they may have had both things in some cases, but only one would count in your primary analysis.
DR. STEWART: Right. There were patients in whom the primary patient endpoint was palliative, and the primary physician goal, treatment goal was preventive. And in the original algorithm, both the patient's primary goal and the physician's primary goal were put into the algorithm to contribute to patient benefit. So, if one failed and the other was met, no benefit was ascribed.
DR. TEMPLE: But patients could never have a prevention goal. That had to be generated by the physician.
DR. STEWART: That's correct.
DR. TEMPLE: And to some extent that seems to be driving the result. That's not a critical comment. I mean, prevention is good.
DR. STEWART: Yes. We feel that prevention was important, as has been discussed, but we also recognize the agency's concern about prevention and palliation being different things, and that's why we presented some of our results for palliative goals only.
DR. TEMPLE: Okay, and then prevention is presented on page 42. Not everybody had a prevention goal so your numbers are down, but you're asserting a nominal significance of .027 on that thing.
This is my last. If you then go to number 58, where you talk about the FDA analysis of the palliative goals, which doesn't apply to prevention goals, I'm curious. I don't understand what 58 conveys to you.
But I first have to say, it looks as if, when the endpoint was defined, nobody took into account -- and we didn't suggest it to you either -- nobody considered the possibility that people would get worse. So, that wasn't part of the endpoint at all.
DR. STEWART: I think that's a fair statement, except that failure of a goal negated a benefit. So, at the time that the patient benefit algorithm was put together in 1997, worsening was definitely considered as part of the algorithm. If you had a worsening of either primary goal, you couldn't be a benefitter, and that's where the worsening comes in, and that is the genesis of our definition of worsening as 7 or 8 days with two measurements showing a worsening.
DR. TEMPLE: Okay, but you did on page 58 look at 28-day worsening, too. You might want to look at that one because at first glance, it looks very adverse, and you didn't find it so adverse and I couldn't tell why. This seems to say that in both studies about as many people got more worse on the drug as got more better on the drug. And in both cases now, they're defined with the same 28-day period of time. So, why isn't that very bad?
DR. STEWART: Well, we don't think it's bad, and I'm going to ask Dr. Leavitt to address this also. Our point here was, once we leveled the playing field, by having an equivalent definition of better and worse, that the great excess of worsening which was seen in the agency's analysis essentially went away, and now we've got a balance between better and worse.
I'd like to ask Dr. Leavitt to address the issue of patients of this type, when they're on treatment and what patterns of worsening and improvement are seen.
DR. TEMPLE: Okay. Just to be sure we understand, this is on the same scales that palliation is looked at.
DR. STEWART: Right.
DR. TEMPLE: So, that means worse means they went up a point, and better means they went down a point.
DR. STEWART: For the same period of time.
DR. TEMPLE: And they're both for a full 28 days.
DR. STEWART: No. We set the goalposts at the 7-day rule. But now we're evaluating both better and worse using the same rule.
DR. TEMPLE: It says first 28 days.
DR. STEWART: No. This is only the first 28 days on study. Since placebo patients tended to migrate off more quickly, as has been pointed out, there are two ways in which there could be an excess of worsening as done in the agency's analysis. First of all, it was easier to get worse than to get better because worsening only required 7 days and improvement only required 28 days.
DR. TEMPLE: Understood.
DR. STEWART: And the other way that there was a bias was that patients who were on CDDP/epi gel, because they were responding and benefitting, tended to stay on study longer, so they had a longer opportunity to have a worsening of one of their goals.
So, by limiting the analysis to the first 28 days, when we had pretty equal numbers of original placebo patients and original CDDP/epi gel patients on study, and leveling the goalposts as to the definition of worse and better, we were able to get rid of the large excess of worsening, which appeared in the original table. When I say get rid of it, I don't mean we waved our hands and made it go away. We showed that when you do the proper adjustment, that it's not there anymore.
DR. TEMPLE: Maybe everybody understands this, and if so, Stacy, tell me to shut up. But this doesn't mean someone was better for 28 days.
DR. STEWART: No.
DR. TEMPLE: That's not the better criterion that you used initially. It means they were better, what?
DR. STEWART: To be better in this analysis, you had to be better for 7 days, and to be worse you had to be worse for 7 days.
And I'd like to point out that the reason that we did this analysis was to show that there was some bias in the original table, not because we're claiming that a 7-day benefit is clinically meaningful. We stand by our claim that 28 days of improvement is what's really needed.
DR. TEMPLE: And you don't have people who were worse for 28 days?
DR. STEWART: No, because typically if they got worse, it meant they were progressing, and then they went off study.
DR. HOWELL: Let me just make one last comment on it. It's important to understand and be clear that when you say worse, you're talking about a fraction of the patients getting worse and another fraction, a different fraction, getting better. This is not a within-patient assessment.
DR. TEMPLE: No, I understand, but if it were really true, which it apparently isn't, that twice as many people had longstanding worsening on the drug than on placebo, that wouldn't be so good. That would suggest that the injections make people as much worse as they make them better. But you have explained that this is a transient associated with the injection, perhaps.
DR. HOWELL: We're just trying to get a grip on the issue of, does this drug cause patients to get worse, as well as some patients to get better. And the answer is, no.
DR. TEMPLE: Well, the answer is yes, briefly.
DR. NERENSTONE: My question also is a little bit about symptomatology. Depending on which graph you look at, anywhere from 15 to 30 percent of patients had nausea and vomiting, moderate to severe. Was everyone pre-treated with antiemetics?
DR. HOWELL: No. In fact, my understanding is that the vast majority did not receive any þ
DR. NERENSTONE: But, Dr. Mills, didn't you say that all of yours were?
DR. MILLS: The protocol did not specify the investigator had to pre-med. I pre-medicated all my patients, though, because like I said, platinum to me is a drug you need an antiemetic for.
DR. NERENSTONE: Well, it's sort of an important question because this drug is being touted as being available for people who don't want the side effects of chemotherapy. And I think certainly low-dose weekly systemic platinum may have a similar episode of nausea and vomiting. So, that's my question pertaining to that.
DR. MILLS: I don't think this is anywhere near that. Even in this trial, if you looked at severe nausea and vomiting, I think it was only 3 to 4 percent. That was all grades of nausea and vomiting as an AE, which is not the same as the platinum nausea and vomiting that we deal with.
DR. NERENSTONE: Do you have any data about the duration of the nausea and vomiting? I believe that you described it as moderate to severe.
DR. MILLS: I'd have to ask the company for that data.
Dr. Wenig, I think, could comment on his patients that he treated.
DR. WENIG: I can tell you that the patients that I treated and the other patients treated by the population of treating physicians who come from a surgical background did not pre-medicate their patients at all. I certainly didn't see anywhere near any levels of nausea and vomiting in any of the patients that are described.
DR. NERENSTONE: Another question I have about toxicity, and I understand that there are low levels, but there are some levels of platinum. Did anyone look at pre-treatment creatinines, or creatinine clearances as a preliminary for allowing patients on trial? And do you have any data in people who you would assume would have very limited creatinine clearances in terms of toxicity?
DR. HOWELL: Yes. Let me let Dr. Leavitt address the issue of whether that was prospectively called out in the protocol. But let me also comment that the area under the curve for exposure for the systemic circulation is in the range of 4 percent of what you get when you give a standard dose intravenously. So, issues of renal function really become pretty minor unless the renal function is severely impaired. You're just not getting that much systemic exposure.
Now, it doesn't take a whole lot of platinum in your systemic circulation to give you some nausea and vomiting, but when you look at the area under the concentration times time curve, the overall exposure is way below what you would expect to threaten the kidney in any way.
DR. NERENSTONE: Right. I'm just worried about what Dr. Lippman pointed out, which is that at the higher level of doses, with the bigger tumors, when you open this up to the general population, those patients perhaps, not at your recommendation, but may in fact go on trial. They really are going to see higher doses than perhaps the optimal patient would receive. And the people who are going to be put on these trials are going to be the elderly debilitated patients who have very limited creatinine clearances. I'm just wondering if you had a level of creatinine clearance that you would not want to see patients on trial, and whether any data has been generated about the toxicity.
DR. ELIAS: Yes. The protocol eligibility criteria excluded patients with serum creatinines or creatinine clearances more than 1.5 above the upper limits of normal, and we would adhere to those recommendations in the future.
With respect to changes in creatinine on protocol, these were tracked, and actually numerically there were more patients whose creatinines got better after treatment than got worse.
With respect to antiemetic medications, a subpopulation of patients at some point during the study did get antiemetics. It's not possible to discern a pattern with respect to the AEs that were reported. In other words, in patients who received antiemetics, some of them didn't have nausea and vomiting reported, some of them did. It's hard to ferret out because some of the patients were previously exposed to chemotherapy, and in some instances it was the physician's routine to give antiemetics, and in some instances antiemetics were given in response to an episode. So, it's hard to exactly ferret out a sequence across the whole study, but only a subpopulation of the patients actually received antiemetics while on treatment.
DR. NERENSTONE: And one last question, which is, for patients who required chemotherapy for non-indicator lesions, how was the duration of response determined for the lesions that were injected? In other words, at what point are they considered no longer to be responding to the intralesional treatment?
DR. LEAVITT: I'll give you a quick answer to that. We censored the duration of all responses when there was any confounding therapy, such as an exposure to systemic chemotherapy or any other kind of therapy that might have confounded the duration of response. That's when we cut it.
DR. NERENSTONE: I think what we'd like to do now is take a very brief 5-minute break. I'd like everybody back at 4:40.
* DR. WILLIAMS: Madam Chairman, members of the committee, ladies and gentlemen.
This slide presents the outline of the FDA presentation. I will start off the presentation with the regulatory background. Then Dr. Frykman will present the medical officer findings, followed by Dr. Sridhara's statistical comments. Finally, I will summarize the FDA findings and introduce the questions to the committee.
First, I'd like to commend Matrix for undertaking these studies -- randomized placebo-controlled trials in head and neck cancer is certainly an unusual phenomenon -- and also for having the courage to try to explore new endpoints to define clinical benefit in head and neck cancer. We've had a very good working relationship, I think, both through the years and also throughout this NDA review.
This slide recognizes the FDA review team. This includes reviewers from a number of disciplines, and is led by project manager Dianne Spillman.
Drug approval usually requires two adequate and well-controlled studies demonstrating the drug is effective and also safe for its intended use. The efficacy requirement is from a 1962 law that required substantial evidence of efficacy and stated that this evidence must come from adequate and well-controlled investigations.
FDA subsequently interpreted the 1962 amendment to mean that changes defined as efficacy must have clinical meaning. That is, they must represent clinical benefit. The laws and regulations give no firm definition of efficacy or clinical benefit. This judgment is left to the FDA.
As the ODAC is faced with giving FDA advice today on the meaning of clinical benefit in head and neck cancer, it seems appropriate for us to review the FDA's approach to clinical benefit and evaluation of oncology drugs in recent years.
In the early 1980s, FDA approved cancer drugs based on response rate. In the mid-1980s, upon the advice of ODAC, the FDA determined that response rates should not generally be the sole basis for approval. The possible benefit associated with partial response did not necessarily outweigh the substantial toxicity of cancer drugs. And correlation between tumor response and survival was not well established. The new FDA position required an improvement in survival, or in patient symptoms for approval.
Subsequently, however, the FDA did on some occasions base approval on other endpoints. Their acceptability was determined on a case-by-case basis, by FDA oncologists, with advice from ODAC.
The FDA stated that under selected circumstances impressive tumor-related outcomes could be considered clinical benefit. For instance, an improvement in disease-free survival can be a valid endpoint for an adjuvant setting if a large fraction of recurrences are symptomatic. Complete responses of reasonable duration may represent effectiveness in some diseases such as leukemia. The appropriateness of reliance on response rate should take into consideration the duration of response and the toxicity of treatment. Finally, the legitimacy of response rate as an endpoint is enhanced by correlation with improvement in tumor-related symptoms.
Drug approvals for some indications are of particular interest. For instance, cutaneous responses were the basis of approval for drugs for Kaposi's sarcoma, and cutaneous T-cell lymphoma and were considered clinical benefit because the lesions are visible to patients and responses were thought to be of palliative benefit, at least partially based on cosmesis.
Over the past 15 years, FDA has encouraged development of primary and secondary endpoints to evaluate patient symptoms, and several cancer drugs have been approved based on pain and morbidity endpoints. In 1995 Photofrin was approved for photodynamic therapy in completely esophageal cancer. In 1998 approval was extended to completely or partially obstructing endobronchial non-small cell lung cancer. FDA reviewers relied on patient-reported improvements in symptoms of obstruction to determine that luminal responses were clinically meaningful. In 1996 mitoxantrone was approved for treatment of advanced prostate cancer, based primarily on improvement in pain demonstrated in randomized controlled trials.
So, when FDA met with the applicant in 1994 and 1995, the FDA position was that shrinkage of tumor from local injection of drug into a head and neck cancer did not necessarily represent clinical benefit. Randomized controlled trials were recommended to demonstrate the responses were associated with patient benefit.
Because patients with head and neck cancer have such a variety of problems, FDA suggested identifying the one primary problem in each patient and documenting whether it got better. This would address a couple of problems in most quality of life analyses. First, because only one main problem is specified, one need not worry about the problem of multiple endpoints. Second, this design requires that all patients entered into the study actually have the problem that is being assessed.
FDA also suggested that objective tumor responses should correlate strongly with measures of clinical benefit. This correlation would help support the legitimacy of a clinical benefit endpoint and would support the clinical relevance of tumor responses.
As the sponsor finalized the protocol, FDA reviewers communicated a couple of points that are important for us to consider today. First, FDA was skeptical about the preventive goals. The reviewer stated the sponsor would have to provide convincing evidence that without treatment these events would indeed have happened within 28 days. FDA reviewers have not received such assurance, and furthermore, as you will hear during the FDA's statistical presentation, the NDA data show that differences between the study arms, the preventive goals are entirely due to differences in dropout patterns between the study arms; that is, due to patients dropping out more often on the placebo arm before data 28 than on cis gel. If you look at the events that were to be prevented, there were actually more events documented in the cisplatin gel arm. For this reason, we have not accepted preventive goals in our analysis. Our analysis is limited to palliative goals.
FDA reviewers also communicated to the sponsor their uncertainty whether a 1-point change on the proposed palliative scale represented a clinically significant change. The sponsor has provided information purporting to support the claim, and we seek ODAC's opinion on whether a 1-point change on the palliative scale is clinically significant. In our analyses, we present clinical benefit data, including the 1-point change, pending ODAC's advice on this point.
That concludes my introductive comments. Dr. Frykman will now present the FDA clinical review, and Dr. Sridhara will provide the FDA clinical analysis. Then, finally, I will return to summarize and to introduce the questions.
* DR. FRYKMAN: There were two key objectives around which both studies, 414 and 514, were designed. One, to compare the objective response rate, active versus placebo drug, and two, to assess the achievement of a primary treatment goal.
The key features of the design included stratification based on the most troublesome tumor size, block randomization, double-blinding, and placebo-controlled. The sample size was based on an appropriately powered design to detect a difference in objective response rate between the active and placebo gel.
Patients were enrolled, as you previously heard, on three strata, and I think I won't continue on with that.
This slide lists six studies that were submitted in support of IntraDose for head and neck cancer. I will review briefly the first two, with a few comments about the pharmacokinetic findings at the end.
Dr. Williams previously referred to correspondence with the company in 1994 when the initial discussion of clinical benefit first arose. Partway during enrollment into studies 414 and 514, the applicant and the division met to clarify the division's views on the endpoints for both studies. Symptomatic response was strongly recommended as a primary efficacy endpoint. Despite the studies being designed to detect a difference in objective response rate, the applicant and the division reached agreement that the primary efficacy analysis would be symptom improvement. Tumor responses would play a supportive role.
As the studies closed, additional clarification was made at the request of the applicant. The FDA clarified that a strong correlation would be required between a patient's tumor objective response and any palliative benefit claimed.
Finally, a 1-point improvement in palliative benefit, as measured by the treatment goal questionnaire, would not necessarily provide sufficient clinical evidence for clinical benefit.
During the initial accrual to both studies, an appreciation for the difficulty of administering a fixed dose of gel into certain tumors arose. The problem that was noted was the inability of some tumors to accommodate to protocol-specified volume. At times the correct volume could not be injected, and when it was completely instilled inside the tumor nodule, a fraction would leak out or reflux back out of the needle track.
The applicant, therefore, amended the dosing regimen and technique. Prior to amendment 5, as you've heard, the protocol-specified dose was .5 ml of gel per cc of tumor volume. This dose was based on the volume of the tumor at the first visit. The dose was administered by a single injection and bolus administration into the tumor. A total of 62 patients were enrolled at the time of the study amendment.
Following amendment 5, however, the dose was reduced by 50 percent to .25 ml of gel per cc of tumor volume. This time, however, the volume was recalculated at each visit just prior to injection. In addition, the injection technique was changed from a single injection to a fanning or a grid technique utilizing multiple needle tracks. 163 patients were enrolled following this amendment.
The eligibility criteria have been previously discussed, and I will go over just a few of the more important ones. The patient population included by refractory and recurrent squamous cell carcinoma of the head and neck, which had been treated by as few as one of the following modalities: surgery, chemotherapy, radiotherapy, or alpha interferon. Either primary or metastatic lesions were allowed, although systemic disease was excluded. Patients with a known history of cardiac arrythmias were also excluded owing to the epinephrine component of the gel.
Each study was identical in design and was comprised of three phases: a treatment phase lasting 10 to 12 weeks, a follow-up phase that lasted 5 months, and an extended follow-up phase for an unspecified duration.
The treatment phase was comprised of two periods: a treatment period and an evaluation period. The treatment period lasted 6 to 8 weeks, and it was during this time that the patients received six injections of blinded drug. The evaluation period was for assessment only, and although patients were seen weekly, no study drug was administered.
Following the treatment phase, patients entered into a follow-up phase in which they were seen monthly without injections. If local progression occurred, patients would become eligible for entering into extended follow-up, in which a higher dose of the open-label drug could be administered or concomitant therapy as well. The FDA based no efficacy considerations on any responses or clinical benefit that occurred during this phase.
As you previously heard, study 414 was a North American study, which was conducted in 44 centers in the U.S. and Canada over an approximately 5-year period. Study 514 was a predominantly European study in 28 centers over approximately the same 5-year time period.
With regard to baseline demographics of the enrolled population, the arms of each individual study were reasonably well balanced in terms of age, Karnofsky performance status, histological grade, prior therapy and ethnicity. This table and the following one reveal the degree to which patients remained in the treatment period of the blinded phase. A differential dropout is noted at treatments four, five, and six, with no patients remaining on the placebo arm at the end of six injections, versus 16 percent on the active arm. This finding has implications regarding blinding and forms some of the basis upon which the division discounted the preventive goal as suggesting a clinical benefit. Dr. Sridhara will address this issue in her remarks.
Similar results are seen in study 514 where the treatment conformity decreases to 42 percent in the active arm and 17 percent in the placebo arm at the end of six injections.
This table and the next table present the reasons why patients terminated enrollment in the study. Approximately one-quarter of the patients did so for systemic progression of their disease. This is an important number to keep in mind, as Dr. Sridhara will briefly mention the median survival of this patient population.
Another 20 percent terminated their enrollment due to progressive disease of the target tumor. What is not included in this table is the number of patients who progressed locally, not systemically, in the form of worsening, co-existing lesions, or by the appearance of new lesions not present at the initial visit.
In study 514, more than half of the patients, the top two, were terminated from the study for progression either systemically or locally, again raising the same issue of local regional progression in the presence of a remitting lesion that has received serial injections.
The key efficacy results are shown here and include objective response only, clinical benefit, and clinical benefit in the presence of an objective response. Of the 62 patients randomized to the active arm in study 414 in strata 1 and 2, there were 20 who achieved an objective response which was required to be maintained for 28 days with at least a 50 percent volume decrease.
This finding must be viewed in the context of four important factors. The first one is frequent dosing errors that were noted by the agency in its review. The second is that this response rate is a local response rate and may not be analogous to what we're used to with systemic objective response rates. Thirdly, in some cases a tumor response was noted in a field of newly appearing lesions. And fourth, responding lesions tended to be on the small side.
The dosing errors which arose from several sources were so numerous that only 3 of the 20 patients who responded actually received IntraDose at the protocol-specified dose and manner. A description of these errors will follow two slides after the same display for study 514.
Caution must also be exercised in interpreting this rate of objective response. Again, this is a local response and may not follow analogously to systemic response to the same degree. Of the 20 responding lesions, 12 were in stratum 1, with a median baseline tumor volume of 1.6 cc's.
Total clinical benefit was noted in 3 of the 51 patients for yielding a rate of 5.9 percent. The number of patients receiving clinical benefit in the presence of objective response was 2 out of the population of 51 who had chosen a palliative benefit treatment goal for a clinical benefit and objective response. The rate of this is 4 percent, and this represents the primary efficacy endpoint of this study.
The efficacy results from the European study showed objective responses in 13 of the 57 patients, for a response rate of approximately 23 percent. Only 6 followed the protocol-specified dose and schedule. Again, this represents a local tumor objective response rate in only the tumor that was injected. Of the 13 responders, 9 came from stratum 1, with a median baseline tumor volume of 2 cc's.
A total clinical benefit rate of 10 of 54, or 19 percent, was noted in this study. There were 5 patients in this population of 54 whom the division confirmed as having achieved both an objective response and palliative clinical benefit.
The applicant and the division, as early as 1994, wrestled with the differences and the color between the clear placebo gel and the yellow-colored active gel. Dr. Leavitt has referred to the efforts that the applicant undertook to maintain the blinding. Agreement with the division was reached that a yellow colored film or sleeve would be wrapped around the barrel of the administration syringe by the investigational pharmacy to maintain the blind of the injecting physician.
During the review of the study, there were several findings that independently led to question about the adequacy of this blind. Each of the findings alone does not necessarily raise concerns, but taken as a whole suggests that there is a potential for the study to have not maintained its double-blind feature. Concerns we noted during the review include differential local toxicity, differential dropout, local hair loss, and a yellow discolored eschar.
There were also a number of concerns that arose with the method of blinding. Could the yellow color be detected as the gel is injected through the tip of the syringe? Could the yellow color of the gel be detected upon refluxing back out of the tumor and then wiping it with a clean, sterile white gauze? Could the investigator readily observe a yellow color if a small amount of the material was expressed into the gauze prior to injecting? And how tightly was the yellow plastic sleeve wrapped around the syringe? Might it be removed by the investigator to improve tactile sense and dexterity immediately prior to the injection, and thereby unblind the investigator? In each of these scenarios we do not know how often, if ever, they occurred.
As mentioned a few moments ago, there are a number of sources of dosing error that were noted in the study arising from errors in measurement, errors in calculation, and errors in administration. The errors in measurement are similar to those that we confront when trying to determine if a lesion on CT has regressed to the requisite amount to be declared a partial response. This uncertainty increases with small lesions and is confounded by local tissue disruption, such as seen in this study, by the gel, either placebo or active gel, in terms of local in duration, necrosis, erosion, and so forth.
Calculation errors arose from using the incorrect gel-tumor ratio, missing tumor dimensions, and even injections in the absence of any tumor measurements.
Finally, administration errors resulting from the reflux of the gel out of the tumor, and/or the PI's discretion to use other than the protocol-specified dose, further confounded attempts to administer the required doses correctly.
This histogram represents the relative frequency that dosing errors occurred, and the magnitude of these errors. On the y axis is the number of doses. The x axis shows varying ranges of dosing errors in the percent of the planned dose. Doses smaller than the protocol-specified dose are on the left. The central four bars represent the number of doses that were within 25 percent of the expected dose, based on the available tumor dimension. This slide represents all doses in study 414. The large bar on the left represents patients who should have received an injection, but did not because of investigator discretion, missed appointments, or unacceptable local toxicity.
The bar on the far right represents the injections in which the dose was calculated incorrectly from either a mathematical error or the use of an incorrect gel-tumor ratio, which was changed with amendment five. Note that there is substantially more underdosing than overdosing, and this is mostly the result of reflux, the tumor being unable to accommodate the specified volume, and the investigator's discretion in using a smaller dose than planned.
This histogram for study 514 similarly shows the deviations from the protocol-specified or planned dose in the same manner as displayed in the previous slide for study 414. There was one notable difference from the previous histogram. A larger portion of patients in the central four bars appear to have received close to or exactly the correct dose. There remained, however, a still sizeable fraction of administrations that were incorrect or did not occur at all. As noted in study 414, more patients were underdosed than overdosed.
Within these studies the reviewers noted a herculean effort by the applicant and the investigators to collect serial local toxicity and palliative benefit data at each visit. We have, therefore, a detailed record over time of each responding or benefitting patient with regard to their unique palliative benefit, tumor size, and local toxicity assessments. Because of these extensive data about the local tumor, we could retrospectively evaluate the totality of each patient's data for integrity and internal consistency.
Examples of the inconsistencies shown on this slide were noted and speak to the questionable integrity of some of the clinical data in several claimed cases of clinical benefit. These data will not be shown at this time. They are available if needed.
The findings limit our confidence that, number one, a 1-point change in the specified treatment goal was reliably detected in these studies, and number two, that a 1-point change on the clinical benefit scale actually represents meaningful palliative benefit.
The applicant collected data as requested by the division on the nature, severity and duration of the local toxicity. You all have seen this data presented previously and I will skip the rest of this slide.
The final aspect of the safety portion of this presentation is a brief review of the systemic adverse events. The applicant has presented information in the briefing document about the six cases of cerebrovascular accidents that occurred in study 414. Additionally, one incident of complete blindness occurred in study 514. The investigators' attribution in this case was that it was directly related to IntraDose administration. We conclude that inadvertent direct injection into vital organs such as the carotid artery, the eye, or the optic nerve cannot be excluded.
The final issue that I would like to briefly address is the degree that the cisplatinum remained in the local tumor following injection. No specific assays of intratumoral cisplatinum levels were performed. However, we can glean a sense of how localized the cisplatinum remained by assaying for systemic exposure, and this was the point of study of the pharmacokinetic study.
Our clinical pharmacology colleagues made two important observations and drew one important conclusion from this study. The important observation is that the pharmacokinetics of the cisplatinum is highly variable, up to 100 percent coefficient of variation for the AUCs after intratumoral administration.
Secondly, the time to peak plasma concentration, or the Tmax, in the 16 patients range from 5 minutes to 24 hours, with a median of 1.5 hours. Their conclusion was that dose normalized exposure to cisplatinum after intratumoral administration was similar to intravenous exposure.
In summary, I would like to highlight some of the key issues that arose from a review of the clinical data derived from studies 414 and 514. There were numerous errors and deviations from the protocol-specified dose and schedule that raise questions about the adequacy of the conduct of both studies, especially in study 414.
Differential toxicity and dropout and other incidental findings in the presence of two medications with known color differences raise questions about how tightly the blind was maintained.
Additional inconsistencies were noted and raised questions about the integrity of the clinical data.
The response rates were modest, in the 20 to 30 percent range, and objective responses, tightly correlated with palliative benefit, was seen in 9 percent in the European study and a rate of 4 percent in the U.S. study. Keep in mind that these were local, not systemic responses from a locally injected cytotoxic drug and that smaller lesions appeared to respond better.
Local toxicity, as expected, was generally mild and moderate, but occasionally severe and increased in severity and frequency with additional treatments compared to placebo gel. Seven devastating adverse events were noted in the safety database limited to these two studies. Direct injection into vital structures cannot be excluded and may represent a substantial safety risk.
Finally, we ask the committee during their deliberations to consider the overall value of IntraDose for shrinking a target tumor in a field of newly appearing and/or progressing lesions in the local and/or regional area.
I would like to thank you for your attention and will turn the podium over to Dr. Sridhara, who will cover the FDA's statistical findings.
* DR. SRIDHARA: Thank you, Dr. Williams and Dr. Frykman.
This is a joint statistical review by Ms. Choi and myself.
There are two major areas of concern.
The two double-blinded, randomized studies were required to demonstrate efficacy with respect to two co-primary endpoints, namely objective tumor response and clinical patient benefit. The understanding was that both endpoints should demonstrate efficacy and thus not require adjustment of type 1 error for multiple endpoints. Both the studies failed to demonstrate clinical patient benefit.
The second concern is regarding the association between objective tumor response and patient benefit in terms of prediction of benefit from response and validating benefit measure. The association between tumor response and patient benefit is weak.
In the next couple of slides, I will focus on the primary endpoint of objective most troublesome tumor, or MTT, response. It is to be noted that both studies were sized based on MTT response as the endpoint. However, we never expect any substantial tumor response with placebo and, therefore, we are most likely to find significant tumor response with any active cytotoxic drug when compared to placebo.
This slide shows the sponsor and FDA analysis of tumor response in the U.S. study 414. Per FDA analysis, the tumor response in the IntraDose arm was 32 percent. Furthermore, 12 of the 20 objective responses in the IntraDose arm were in stratum 1, or in patients with smaller lesions.
This slide describes the sponsor and FDA analysis of study 514. Per FDA analysis, the response rate was 23 percent in the IntraDose treatment arm. 9 of the 13 objective responses in the IntraDose arm were in stratum 1 and in patients with less than or equal 5 cubic centimeter lesions.
This is the survival graph of the IntraDose versus placebo with the combined data of both the studies. There is no difference between the two arms and the estimate of median survival was about 3 months in both the arms. As expected, the local treatment response does not seem to translate to survival benefit.
I will now present analysis of the primary endpoint of clinical patient benefit.
The patient benefit was measured on a preselected treatment goal using the questionnaire designed by the sponsor. This questionnaire has only been used in the two studies under consideration here. The validation of this instrument was conducted by the sponsor with only 15 patients and cannot be considered as adequate.
The treatment goal, a palliative goal or preventive goal, was selected by the investigator prior to the randomization for each patient. The patients were also encouraged but not required to select a goal.
Furthermore, per protocol and sponsor analysis, the patient benefit is based on the investigator selected treatment goals, palliative or preventive. The palliative goals were wound care, pain control, abilities to see, hear or smell, physical appearance, obstructive symptom, and mobility. The preventive goals were prevention of invasion of vital structure or blood vessels, prevention of obstruction, or prevention of tumor breaking through the skin.
A post hoc patient benefit algorithm has been used by the sponsor to include patient assessment. However, it should be noted that in the case of the preventive goal, the patient could not contribute to the assessment of this benefit.
Furthermore, the two treatment goals, palliative and preventive goals, were measured on different scales: palliative goal on a 4-point scale, whereas preventive goal, on a 2-point scale. Combining these two scales implies that a change of score from 4 to 3 or 2 to 1 on the palliative scale is the same as met in the preventive scale. This combining of two different scales, one measuring symptom improvement and the other measuring progression, is questionable.
The sponsor has also defined a 1-point decrease for a duration of 4 weeks to be a benefit for the patient, and as mentioned earlier by Dr. Williams, this is an issue that needs to be discussed.
Lastly, there was no common treatment goal among the patients.
This is the algorithm the sponsor has used to combine the investigator and patient assessments. If both the patient and the investigator recorded a benefit, or if either of them recorded a benefit and the other recorded a no change, then the patient was counted as a benefitter. Note that particularly in the case of preventive goals, in most cases the patient's assessment was missing. Thus, it was essentially the investigator's call on the preventive goal assessment.
These are the results of the sponsor analysis, which includes palliative and preventive goals in assessing patient benefit. Clearly both studies individually did not demonstrate significant benefit over the placebo. In the pooled analysis of both studies, there is borderline significance.
However, pooled analysis is not acceptable when both studies have failed to demonstrate clinical benefit.
Pooling also inflates type 1 error.
Furthermore, even though in both the studies the plan was to accrue patients in a 2 to 1 ratio of IntraDose to placebo, in fact in the U.S. study, the ratio was 2.6 to 1 and in the Europe study, it was 1.5 to 1. Therefore, pooling these two studies will cause imbalance in randomization.
There is evidence that the patient populations were not the same in the two studies, for example, in terms of prior therapy, age, and performance status.
The selection pattern of treatment goals by the investigators varies between the two studies with more preventive goal as the primary selected goal in the U.S. study compared to the Europe study.
Pooled analysis can only be used as supportive evidence and not as primary evidence. At best, this can only be considered as one body of evidence.
This chart describes the number of treatments received in both treatment arms in study 414. The blue bars represent the IntraDose arm and the red bars represent the placebo arm. Clearly the dropout pattern in the two arms are different with no patients receiving six treatments in the placebo arm.
This chart describes the number of treatments received in both the treatment arms in the Europe study 514. As in the U.S. study, the dropout pattern in the two arms are different with only 17 percent of the patients receiving all six treatments in the placebo arm versus 42 percent of patients receiving all six treatments in the IntraDose arm.
In the U.S. study, the investigators selected in 50 percent of the patients preventive goal as the primary treatment goal, that is, in 31 of the 62 patients. Here not met is the occurrence of an event, that is, the event where the patient's tumor has invaded a vital structure of a blood vessel or the tumor is obstructing a structure or the tumor has broken through the skin. In the U.S. study, per investigator assessment, in 13 percent of patients, there as an event or failure, whereas there were none in the placebo arm.
Furthermore, the category "same" is when patients could not be assessed as either met or not met probably because of dropout. In other words, almost 50 percent of the patients in both the studies in the placebo arm could not be assessed for the achievement of preventive goal because of dropouts.
Thus, the preventive benefit assessment was discredited by the reviewers, as mentioned earlier by Dr. Williams, because of the differential pattern of dropout between the two treatment arms which could potentially bias investigator assessment. This could also potentially cause unblinding, which was addressed earlier by Dr. Frykman. Essentially these preventive scores are not interpretable. In fact, 13 percent of the patients in the IntraDose treated arm in the U.S. study had an event versus none in the placebo arm.
Also, we do not have a baseline estimate of the incidence of these events in an 8- to 12-week time period in this patient population to compare it to a baseline incidence rate.
As seen above, in the U.S. study in 50 percent of the patients, the investigator selected preventive goal as the primary goal.
Further discussions will, therefore, focus only on palliative goal assessment. In study 414, the investigator selected in 31 of the 62 IntraDose-treated patients palliative goal as the primary treatment goal. A negative score here indicates improvement or benefit in symptom, and a positive score indicates worsening of the symptom.
In this study only 1 patient, or 3 percent, had a benefit, if a change in score by 1 point is considered as a benefit, versus 6 patients, or 19 percent, who got worse on the very endpoint of interest. However, most of the patients, or 74 percent of the patients, did not have any change in their symptoms in both the arms.
In study 514, the investigator selected in 46 of the 57 IntraDose-treated patients palliative goal as the primary treatment goal. In this study 7 patients, or 15 percent, in the IntraDose-treated arm appear to have benefit versus 8 patients, or 17 percent, who got worse on the very endpoint of interest. Again, the majority of the patients, or 67 percent of the patients, did not have any change in their symptoms in both the arms.
The sponsor did talk about the worsening, that it was only a 7-day period versus an improvement for 28 days. However, it should be kept in mind that these were two consecutive measurements that were made for the worsening, and it's understandable that if somebody is worsening, that you cannot keep them for 28 days in the worsening score itself.
This is the FDA analysis of palliative benefit. Since we have discredited the preventive goal assessment, we analyzed the palliative goal benefit using both the patient or investigator assessments. A decrease in palliative goal score per patient or investigator assessment was scored as a decrease in score leading to a least conservative analysis of palliative goal assessment.
In this table, the first set of analyses is assuming that a change in score by a scale by 1 point or more to be clinically meaningful. Even in this least conservative analysis, it is of concern that in the U.S. study patients got worse four times more than those who felt better in the IntraDose-treated arm. In contrast, in the placebo arm, patients got worse two times more than those who felt better. Most of the patients by this analysis also did not have any change in symptom score.
The appropriate tests, Wilcoxon rank sum test or JT test, were conducted to test the difference between the two treatment arms which accounted for the categorical classification of patients into three categories: better, worse, and no change. There was no significant difference between IntraDose and placebo with respect to patient palliative benefit in this U.S. study by both of these tests.
In the sponsor's analysis, benefit is treated as a binary outcome not accounting for patients with worsening of symptoms and used Fisher's exact test to test the difference between the two arms. This is not an appropriate analysis. For purposes of illustration only, the p value using this method is also presented in this table.
Furthermore, the second set of results presented here is assuming that a change in score by a scale of 2 points or more to be clinically meaningful. In this case there were only 4 percent who felt better in the IntraDose, and no patients had benefit in the placebo arm.
As in the previous slide, in this table the first set of analyses is assuming that a change in score by a scale of 1 point or more to be clinically meaningful. In this least conservative analysis, as observed in the U.S. study, in this Europe also 22 percent got worse compared to 19 percent who felt better in the IntraDose-treated arm. Again, the majority of the patients did not have any change in symptom score.
There was no significant difference between IntraDose and placebo with respect to patient palliative benefit in the Europe study using the appropriate tests, Wilcoxon rank sum test or the JT test. For purposes of illustration only, the p value using the Fisher's exact test and treating benefit as a binary outcome is presented in this table. This is not an appropriate test as it does not take into account a third category of worsening of symptoms.
Furthermore, when the two studies were pooled and data analyzed using the Wilcoxon rank sum test or the JT test, there was no significant difference between the IntraDose and placebo arms.
The second set of results presented in this table is assuming that a change in score by a scale of 2 points or more to be clinically meaningful. In this case there were only 4 percent who felt better in the IntraDose, and it is exactly the same percentage of patients as in the U.S. study and no patients had benefit in the placebo arm.
This is a pictorial representation of the FDA palliative goal analysis. The green bars represent the percentage of patients who felt better. The red bar represents the percentage of patients who felt worse, and the blue bar represents the percentage of patients who did not have any change in their symptom score. In both studies, the majority of the patients did not have any change in symptom score, and in both studies, particularly in the U.S. study, more patients got worse than feeling better in the IntraDose arm.
In conclusion, regarding the primary efficacy endpoint of patient benefit, both studies failed to demonstrate clinical patient benefit of IntraDose versus placebo by both the sponsor's and FDA's analysis. Whether a change in score of 1 point is clinically meaningful needs to be discussed. If a 1-point change is excluded from counting as a benefit, then less than 5 percent of patients had any palliative benefit in both the studies. Even if a 1-point change is considered as a benefit, only 6 percent of the patients appeared to have any palliative benefit versus 25 percent who got worse in the U.S. study, and 19 percent appeared to have benefit versus 22 percent who got worse in the Europe study.
Furthermore, per the sponsor's analysis, only 5 percent of IntraDose-treated patients obtained investigator and patient-specified primary treatment goals.
The second major statistical issue is regarding association between objective tumor response and patient benefit in the IntraDose-treated patients. This is the sponsor analysis of the association between tumor response and patient benefit in the U.S. study. In this analysis, patient benefit includes both palliative and preventive benefit. Note that only 10 of the 62, or 16 percent, had both patient benefit and tumor response per this analysis. The majority of the patients had neither benefit nor response. This analysis does not provide a quantitative measure of association such as a correlation coefficient in the case of continuous variables. The p value is not meaningful since the association is driven by nonresponders and nonbenefitters.
A preferred measure of association is sensitivity which gives the probability of a patient having a benefit given that the patient has a tumor response. In this analysis, the sensitivity was 48 percent; that is, there is less than a 50 percent chance that a patient will have benefit if the patient has tumor response.
This is the sponsor analysis of the association between tumor response and patient benefit in the Europe study in the IntraDose-treated patients. As in the previous slide, in this analysis patient benefit includes both palliative and preventive benefit. Note that only 6 of the 57, or 11 percent, had both patient benefit and tumor response per this analysis. The majority of the patients again had neither benefit nor response. The p value again is not meaningful since the association is driven by nonresponders and nonbenefitters.
In this study also the sensitivity was 43 percent; that is, there is less than a 50 percent chance that a patient will have benefit if the patient has tumor response.
This is the FDA analysis of the association between tumor response and patient benefit in the U.S. study in the IntraDose-treated patients. In this analysis patient benefit includes only palliative benefit assessed by the patient or investigator. Note that only 2 of the 51, or 4 percent, had both patient benefit and tumor response per this analysis.
In this analysis the sensitivity was only 13 percent; that is, there is only a 13 percent chance that a patient will have benefit if the patient has tumor response.
This is the FDA analysis of the association between tumor response and patient benefit in the Europe study in the IntraDose-treated patients. Again, in this analysis patient benefit includes only palliative benefit assessed by the patient or investigator. Only 5 of the 54, or 9 percent, had both patient benefit and tumor response per this analysis.
In this analysis, the sensitivity was only 42 percent; that is, there is a 42 percent chance that a patient will have benefit if the patient has tumor response.
In conclusion, regarding association between tumor response and patient benefit, the p values presented by the sponsor are not interpretable, and the association is weak and driven by a large number of patients classified as nonresponders and nonbenefitters. There is less than a 50 percent chance that a patient will have benefit if he or she has tumor response. In other words, tumor response does not predict patient benefit.
In summary, both randomized studies failed to demonstrate statistically significant clinical patient benefit of IntraDose when compared to placebo. It is also not evident that the objective tumor response translates into clinical benefit.
* DR. WILLIAMS: Thank you, Dr. Sridhara and Dr Frykman.
My job is to summarize, and I'll try to be merciful and brief.
To summarize, FDA found that there was a reasonable rate of local tumor response. They found little evidence of clinical benefit as defined by improvement and prospectively defined palliative goals. This slide shows the response rate of 32 percent and 22 percent on the cisplatin gel arms and basically no responses, just 1 in one trial, on placebo. So, it seems clear that the cisplatin gel causes some tumors to become smaller.
These tables summarize the FDA's analysis of clinical benefit, as you've seen a couple of times. As you'll recall, the FDA analysis excludes preventive goals and includes palliative goals by either the investigator or the patient. In study 414, there is no suggestion of clinical benefit beyond what was shown by placebo. However, in study 514, if you just look at improvement or better, there appears to be some suggestion of more benefit on the cisplatin gel arm, but still at a fairly low rate of 19 percent versus 3 percent on placebo.
However, as we've been discussing, there's a bothersome phenomenon noted in both analyses. In both studies, there's more worsening than improving of the primary endpoints in the cisplatin gel arm. In addition, there is more worsening than is noted in the placebo arm. And I think this would disturb me probably the most.
As Dr. Sridhara noted, if you include worsening in statistical analyses and do a Wilcoxon rank sum test, there's no statistical significant between the arms and not even a trend.
And finally, when palliative benefit was compared to objective tumor response, there was certainly no strong correlation found.
So, the FDA presents these data and analyses to ODAC for assessment of whether cisplatin gel provides clinical benefit to patients with head and neck cancer, benefit that is greater than toxicity. When these studies were designed, the assumption of the FDA and the sponsor was that tumor responses do not necessarily equate with clinical benefit in the local treatment of head and neck cancer.
Approval considerations were to be based primarily on prospectively defined palliative benefit.
Questions for you to consider include whether a 1-point change on the 4-point palliative scale is meaningful within the context of this trial and, if so, whether the rate of benefit is acceptable in view of the toxicity of cisplatin gel.
One needs also to seriously consider that more cisplatin gel patients showed worsening of their main problem than showed improvement. Certainly we've had some discussion of that and I'm sure we'll have more.
Other issues that you may consider are the clinical meaning of the tumor responses in these trials considering the rate of the response, the nature and their duration, and the size of the tumors that are responding in view of the toxicity of treatment. In some settings, durable impressive responses have occasionally been supportive of approval. The sponsor has also collected additional data on clinical benefit in more of an anecdotal fashion.
Lastly, I want to address one issue that I think is likely to come up in deliberations. An approach to approval that we do not believe should be entertained is accelerated approval. Accelerated approval allows for approval based on a surrogate endpoint, such as response rate, that is reasonably likely to predict clinical benefit. After NDA approval, the sponsor must then do controlled randomized trials to show clinical benefit.
However, in the case of cisplatin gel, the randomized trials to evaluate clinical benefit have already been done. In fact, the sponsor must be commended on performing perhaps the only randomized, blinded trials in head and neck cancer. If these trials have failed to document clinical benefit, it seems doubtful that a later trial, a phase IV trial, after accelerated approval would succeed in doing so. Therefore, approval by the accelerated approval mechanism does not seem to be a reasonable option.
So, that concludes the FDA presentation. We'll be glad to take questions from our seats.
* DR. NERENSTONE: Why don't we have questions to the FDA first. Dr. Lippman.
DR. LIPPMAN: Grant, I had a couple of questions to clarify some of the issues.
In 1994, there was a meeting with the FDA which indicated there needed to be some sort of patient benefit information. What changed between that and the amendment? I don't know when patient benefit was put in during the trial, but what changed between what would have been a prespecified primary endpoint before the trial started to one of the things that I'm wrestling with and that's changing a primary endpoint during the trial?
DR. WILLIAMS: The situation was that in 1994 and 1995 we had meetings with the company. I was there. We recommended the concept of this kind of an endpoint, and clearly our intent was that there would be no approval unless we saw such a significant effect of the kind we were talking about. And whether you call that a primary or secondary endpoint really didn't matter to me and still doesn't matter to me. It's what's the p value for what we consider to be the most important analysis.
We send comments to the company and the company do what they will do with it, and I guess they did not change the primary endpoint. Then later on statistical analysis is kind of -- maybe an amendment, probably the later amendment, would say, look, your primary analysis isn't addressing what we said it should be. We may not have expressed it as primary analysis before, but as our most important concern, and said that it should be co-primary.
So, I don't think there's really been any difference of opinion. In fact, what I was hearing over the years I think from the company was more the fact that they didn't think they could get the patients to do it anyway, that the accrual was really a major problem and that was limiting them. So, I don't think there was really any change in our intent, and I think it's really more of a technical nature what we're hearing debated.
DR. LIPPMAN: Well, I'm not so sure it's technical because whether it's a prespecified secondary or primary endpoint I don't feel as strongly about. But it does bother me that it changed in the middle of the trial. So, I just wondered what information was conveyed between FDA and the company to have them change in 1997 when, if the same information was relayed to them in 1994 -- but I agree with you about the issue of having co-primary endpoints. I don't know what that means really. One is used to base sample size and the other one could be a secondary. But I would have preferred prespecified, and so I was getting at why in 1997 and not in 1994.
DR. NERENSTONE: Dr. Temple.
DR. TEMPLE: Well, in this case, whether you think of the clinical endpoint as one to be carried out only if the response rate endpoint wins or whether you think of them as something where both have to win, it doesn't really affect the statistical analysis. We all agreed no correction was needed. And it's obvious the whole design of the study was intended, at least in part, to look at the clinical endpoint. That's why they did those scales and all those things. So, we think it was clear.
But I still don't know the answer to my previous question. When, even after several years you finally figure out the importance of the clinical endpoint and you've calculated your sample size based on response rate, someone has to address the question of why you don't increase the study size because you have almost no chance.
DR. LIPPMAN: Now, I have another question. I may have missed this during the presentation or in the book. But I thought, Grant, it was you who said that the primary endpoint of patient benefit, the co-primary endpoint, which was based on the algorithm -- is that correct -- in terms of the design, was a post hoc analysis. That didn't come through in the sponsor's presentation. I wanted to clarify that because, again, whether you design this up front in the trial or during the trial -- obviously, ideally you'd like to have it up front -- but if the co-primary endpoint that we're talking about here, patient benefit, was a post hoc analysis, which includes really a primarily investigator-driven endpoint of prevention, that to me is a more serious concern. So, I just wanted to clarify that. Was it post hoc?
DR. WILLIAMS: Actually what is post hoc depends on who's looking at it.
The FDA involvement can be very tight and close or it can be a little more -- I don't want to say -- loose. But I think in recent years we are much more involved at every stage in making sure that everything we say is addressed. We have more meetings, et cetera. In 1994 and 1995, we were doing some of that, but in general we give advice and it's the responsibility of the sponsor to apply it. The advice was given. The analysis plan was not even formulated till later. So, it's hard to comment on an analysis plan.
DR. LIPPMAN: What I meant by post hoc -- just to clarify what I meant by post hoc, I meant that the study is done, the data is there, and you start looking at it. It's unblinded. You're looking at it. That's what I consider a post hoc analysis, and I'm wondering if that was what you meant when you said that the patient benefit endpoint was a post hoc analysis.
DR. WILLIAMS: I didn't say that.
DR. SRIDHARA: Can I address your question? The data on clinical benefit was collected prospectively. The questionnaire was already there right from the beginning, and on every patient the data was collected. However, there was no hypothesis set up prior to the starting as to what should be the difference that we are looking for or how it's going to be analyzed or how even the patient assessment was going to be combined together. So, all those things were as the study was going on, our post hoc definition of what is a benefit or how do you define a benefit or how can you combine the patient and investigator assessment, et cetera.
DR. LIPPMAN: So, when you said post hoc analysis, you meant the 1-point difference was not prespecified.
DR. SRIDHARA: No. Neither the 1-point difference nor a hypothesis regarding patient benefit itself as to what is the difference that we are looking for was not defined at the start of the study. However, as I said, the benefit scores themselves or the treatment goals themselves were there right from the beginning and on every patient it was collected.
DR. TEMPLE: Was the analysis planned before the study was unblinded or just in the course of the trial? Post hoc is not a fair term to use if everybody is still blind while they're doing it. Which do you mean?
DR. SRIDHARA: As the study was going on, all this was evolving as the statistician commented on this. Of course, I was not in the planning part. What we saw was the final analysis plan which came that they were going to use this patient algorithm.
I was trying to point out that this patient algorithm -- I think even in our minutes we have recorded that we didn't approve of combining the preventive and palliative goals. And this was something that was defined later on to combine the two goals together and also the patient and investigator assessments together.
DR. LIPPMAN: Just to rephrase the question, because one of the issues is the 1-point issue. That's one of them that you asked us to address. Was this 1-point issue prespecified before the study was unblinded?
DR. SRIDHARA: They had mentioned about it, but we had expressed our concerns regarding whether it would be meaningful or not. We were unsure.
DR. HOWELL: Madam Chairwoman, a point of clarification?
DR. TEMPLE: The answer to your question is yes, it was prespecified before they unblinded it. We just said maybe we're not going to buy it. That's a different question.
DR. LIPPMAN: No. That's different than post hoc, and I wanted to clarify that. So, it was not a post hoc analysis the way we would normally think of it.
Then two other issues. You put up, Grant, as your two major concerns, concerns with blinding and concerns with internal consistency.
Now, I raised the blinding issue particularly when you're talking about a 1-point change, all these things. So, I was concerned when I saw the different solutions. And I asked the sponsors about that, and they were confident that it still could be maintained.
What leads you to believe -- maybe I missed it -- that that may not be the case, that the investigator may have some idea about what was being given?
DR. FRYKMAN: First of all, it was very clear from the toxicity data that there was a difference between the placebo and the active drug. That's to be expected. For a single investigator who treats one patient, he may never be able to tell the difference, but for an investigator who may treat a half a dozen patients, he would ultimately acquire a sense of whether this was placebo or active drug. Ultimately, the drug would be unblinded.
The second issue that came up was that in the adverse event database, we just incidentally happen to find a couple of things that suggested that there was a way to tell, and one of the issues was specifically a yellow-colored eschar. The only way that eschar could have become yellow-colored, at least in Europe, we were told by the applicant that there was a topical antibiotic that was used. That's not available here in the United States. So, at least the coloration of the eschar, just an incidental finding in the AE database, suggested that there was a color difference. How often that occurred, we don't know, but it appeared.
The other issue had to do with local hair loss. We know that cisplatinum can cause this, and in fact you would not necessarily see hair loss with just injection of a collagen or saline into the local area.
DR. LIPPMAN: Again, I'm not as concerned that there may have been toxicities that would lead someone to believe of what drug was there, but I am concerned if this lesion turned yellow and that's the color of the solution. Again, I don't know how great of a problem it was. The sponsors felt that this was not a big problem, and I just wanted to see if you feel that it was based on your comment about the blinding.
DR. FRYKMAN: Yes. Again, I would get back to and reiterate the same comment that you made, that if this study had a very low level of noise, the curves were all smooth, and they appeared to respond, for example, the palliative benefit would definitely increase with the shrinking size of the tumor, and everything was all working together, then I'd frankly have a lot of faith in it.
What happens in this case, though, is that we've got some data that's noisy, and on top of that, we've got some question about blinding. It's not clear to me that you're as able to detect or as sensitive -- the sensitivity to detecting a 1-point difference is the same when the trial is conducted impeccably, which is probably impossible, or close to impeccably as opposed to where it was not conducted impeccably, especially in the case of unblinding.
DR. LIPPMAN: Do you have any idea how common it is for these lesions to turn yellow when they inject it? Is this a fairly common thing?
DR. FRYKMAN: I don't know that. I don't know have a specific number. Again, it cropped up in the AE database. That wouldn't necessarily even need to be in there, but one of the investigators apparently put this in. I suspect it happened more often than was in the AE database because it wouldn't be something that you'd report. Did it happen 5 percent, 20 percent of the time? I'm not able to say.
DR. NERENSTONE: Dr. Howell, could you briefly comment?
DR. HOWELL: Yes. Two points of clarification.
First, all the analyses were done on blinded data before the blind of either study was broken.
Secondly, the sponsor has explained to the agency that one investigator in Europe used a yellow-colored antibiotic solution and painted that on the eschars from that one study site. And that is the source of the yellow in one study site from one investigator.
The color difference is not sufficient to be able, once it is in the tissue, to cause any serious change in color of the tissue under any circumstances.
DR. NERENSTONE: Other questions for FDA?
DR. LIPPMAN: Can I just follow up, Stacy?
You made a comment about internal consistency and integrity of the data. That was a concern. Can you elucidate why you had that concern?
DR. FRYKMAN: Yes, I can explain to you briefly.
Again, we had an advantage in this trial in that we had lots and lots and lots of data that was collected over time on a visit-by-visit basis. Where I, reviewing the data, had some trouble with, frankly, sometimes believing what I was actually being told in a specific patient was in the case where a tumor lesion would shrink away completely. There are many cases, as you've seen in the data, in each study where a complete response was achieved. This is exciting and we hope something good comes out of it.
The assumption, when the trial was designed, was that if a tumor shrunk away, that the problem associated with it, whether it was wound care, whether it was pain, whether it was obstruction, or whatever the case was, that that problem would also remit.
What was frustrating is that it did not happen. There are cases where the tumor would remit completely and the patient's pain level would be identical. That's sort of one problem. There was a complete dissociation between tumor shrinking and a symptom going away.
The other issue that I noticed was that on the occasion where the patient and the physician would choose the same endpoint -- it wasn't necessarily the primary one, but it was the same endpoint, such as pain control -- there would sometimes be a disparity between what the physician was saying -- he might rate it as a 3, and the patient might rate it as a 4. Well, again, a 1-point change you could argue really is important or that really is not that important. But when there's internal inconsistency between the same objective findings with regard to a patient, that was a bit troubling.
The third factor has to do with the local toxicity. The sponsor, again, is to be congratulated on the absolutely superb job that was done by them and their investigators in collecting serial toxicity data. Keep in mind that there was no objective scale that the investigators had to go by, and they sort of winged it and said, well, this looks like a mild one, this looks like a moderate one, or this is gone completely.
What happens in the case of a severe necrosis, for example, as opposed to no necrosis when the patient starts receiving the injections, what that does to wound care was somewhat of a question. One would expect that if there were severe necrosis that appeared after three injections, that local wound care, which had been rated as a 2 would get at least up to a 3. Again, you can look at the scale and decide if the scale was sensitive enough or not for that problem. But there was a disparity between what you'd expect to see with local wound care and what you actually saw rated by the patient and/or the physician.
DR. NERENSTONE: Dr. Temple.
DR. TEMPLE: It's worth noting the first of those concerns maybe, on further reflection, is not so much of a concern. The allegation would be that a tumor disappeared and the investigators were too stupid to attribute clinical benefit to that. Now, to the extent that's true, it undermines the results of the study and makes them look weaker. I usually don't believe people contrive to make their data look weaker. So, I don't dispute the observation, but I'm not sure that comes under the heading of bad behavior or something. That doesn't, I have to say, seem like a major worry to me because it cuts against the study.
I also want to observe that I think there are many troubles with these data, but I want to distinguish some of the ones that I don't think are.
The sensitivity argument does not seem persuasive to me. We have no standard for what fraction of people whose tumor shrinks ought to get a clinical benefit out of that. There's no track record, no data. I think 50 percent would be pretty good if you believed all of them. Now, you also heard we don't believe some of them, which is a different question. But I don't think those are the main problems.
I think the lack of persuasive findings might be a real problem, but I think it's important to focus on the ones that really are worrisome.
DR. NERENSTONE: If there are no further questions for FDA -- Dr. Albain.
DR. ALBAIN: A few times it was mentioned that you discounted some of the analyses of the sponsor because the dropouts were greater in the placebo group, if I heard you correctly. To me, though, you'd expect that. If the agent is working, you are going to have patients go off study and they're going to progress much quicker on the placebo group.
The sponsor had showed -- and I hadn't fully understood your analysis -- a rebuttal to your analysis earlier on, and I wondered if you can comment on that rebuttal explanation.
DR. WILLIAMS: I think the only time when we really became insistent about the dropout issue is the preventive. It's not just dropout. It's the fact that there are no events to substantiate the difference between the two arms. It's all drop out by 28 or not drop out by day 28. The sponsor's analysis included as a failure someone who didn't make it to day 28. So, that's kind of like an event. You would think it would be driven by events you're trying to prevent rather than not making it to day 28. That's a case where the data are totally driven by differential dropout, and that's the case where we discounted those kind of data.
DR. SRIDHARA: Can I add to that? Basically the duration of the clinical benefit had to be for 28 days, and if the placebo patients were there only for 28 days and they were removed from that study for palliative benefit, they were changed over or crossed over to the treatment arm because of progression of the tumor, not because of the clinical benefit, so they could not be assessed for clinical benefit beyond 28 days if they did not get beyond 28 days of treatment. The requirement was that they have to have 4 weeks of this benefit. So, they couldn't assess this.
DR. TEMPLE: Could you just explain the prevention endpoint a little further? That's very crucial because we've just thrown out the major source of endpoints. So, if someone has gone 2 weeks on placebo and then progresses or something happens and haven't had the endpoint you're worried about preventing it, then the endpoint is never attributed to them.
DR. SRIDHARA: Yes.
DR. TEMPLE: Right? No.
DR. WILLIAMS: That is the endpoint. They are given a negative endpoint for that. It's not that they're inevaluable. It's they have failed.
DR. TEMPLE: So, they're said to have had the endpoint --
DR. WILLIAMS: It's equally bad to do that as to go 3 weeks and then have it break through the skin. They're lumped together.
DR. TEMPLE: So that when you actually count the events that you were trying to prevent, you don't see that difference. The whole thing is based on not making it.
DR. WILLIAMS: In fact, it's the other way. It's worse on cis gel.
DR. TEMPLE: Yes, that's what you said.
DR. NERENSTONE: Dr. Rubinstein.
DR. RUBINSTEIN: You argue that the preventive endpoint was invalid because of the difference in dropout, but the difference in dropout you demonstrated was primarily a result of earlier progression on the part of the placebo arm. It seems like that should have been anticipated at the beginning of the trial, and it also seems as if that could have been incorporated into the measurement of benefit. For example, one could even say that if a placebo patient is forced to drop out because of progression, that they didn't see the benefit by definition because they progressed. I'm not saying that that's the definition to use, but that certainly would be a possibility.
It seems as if this issue wasn't addressed at the beginning and it seems like this is certainly related to potential benefit of the agent. You even said that in certain cases you would approve agents on the basis of reducing time to progression.
DR. WILLIAMS: I agree with you that you would be showing something, and it basically would be early progression. But what were trying to do was a different type of clinical benefit or a different kind of endpoint. For showing clinical benefit according to patient symptoms or patient preference, this didn't fall within that category.
Now, if you want to describe that as being for the discussion of should progression be acceptable for head and neck cancer, I would agree with you. But for saying should we lump this phenomenon, which is early progression of tumor, into these data that have to do with patient symptoms, then I don't think that that would be acceptable.
DR. NERENSTONE: One of the questions I have just for clarification from a nonstatistical viewpoint, I'm not sure I have your same questions about the blinding because if you look at the placebo arm and the achievement of patient benefit according to the sponsor, there's a remarkably high benefit that they felt. In fact, it was on the placebo arm. In fact, isn't that one of the problems, that the study is so underpowered because there was such a big placebo benefit?
I think that one of the things that we should take away from that is the importance of a placebo-controlled arm because I think if you look at the problems with the randomization and the unbalanced randomization, you could say that perhaps all of the effect is completely due to a placebo effect but randomization is the problem.
DR. SRIDHARA: Yes. When you have imbalanced randomization, it is a problem. That's true. It was planned to be 2 to 1, and you see most of the benefits that you're talking about in the U.S. study which had 2.6 to 1. So, in fact, on the IntraDose there were even more patients.
With respect to preventive benefit that we are talking about, in fact, of the 21 benefits that the sponsor has claimed in the U.S. study, 20 of them were preventive goals, and there was only 1 patient with palliative goal benefit. So, all these 20 patients got benefit from preventive goal, and when we don't have an incidence rate to compare what would be the baseline, we are talking of a very small time period. So, we don't know if nothing might happen during that period or not.
The concern with the preventive goal is then that in fact we observed in 13 percent of the patients that an event happened, a failure happened, they had a breaking through or incidents of some of these adverse things that we were supposed to prevent happened; whereas, there were none in the placebo group. We are giving the benefit of doubt whether this was, in fact, because of the dropout rather than that we did not observe in placebo. In other words, we are saying that this is uninterpretable and therefore we can't use it.
DR. NERENSTONE: Dr. Lippman.
DR. LIPPMAN: A very quick clarification.
When you focused on the dropouts, I think you pointed us to the top two figures, and one of them was patients who terminated for systemic progression, which I think was about 25 percent of the population. Can you clarify that? Because we knew a number of patients had systemic disease. It wasn't ineligibility criteria and it's not being adequately treated. You'd expect it to progress. So, when you say they were taken off for progression, is that really because they had symptomatic progression and needed to be treated, or just that the tumor grew greater than 25 percent, which you would expect with this kind of approach?
DR. FRYKMAN: To be honest with you, I don't know the proportion of those. The way this table was derived was basically to go back through the database of what the sponsor had and sort of lump it into categories that seem to make sense. So, they didn't specify in that database, for example, whether the tumor had grown huge amounts or whatever the case was.
I will say that my sense out of the protocol was, though, that systemic progression was noted usually on the basis of a substantially worsening KPS, substantially worsening weight loss, or the patient complaining about something. There wasn't continual serial monitoring by CT or specific physical exam to pick that up early. In fact, that was one of the concerns.
It's debatable, but there was at least 1 patient there where it appeared she kept receiving additional treatments with IntraDose. She was, I think, randomized to the active arm and kept receiving the gel and it kept her tumor small. You could slowly see her fall off KPS and weight-wise, and then down the line she was eventually declared to be, I think, systemic progressing.
DR. NERENSTONE: If there are no more questions for FDA -- sorry. Dr. Rubinstein.
DR. RUBINSTEIN: I don't understand why you say there was an imbalance in the randomization simply because the 2 to 1 ratio was not maintained exactly and, in fact, differed between the two groups. It wouldn't be uncommon for a targeted ratio not to be maintained exactly. That to me doesn't create an imbalance in the randomization.
DR. SRIDHARA: It's only when we are trying to pool, but you have even lesser placebo and you are getting more patients from the U.S. study where they're claiming more responses, whether it is tumor response or benefit. If both were 2.6 to 1, then it was different. But one was 2.6 to 1 and the other was 1.5 to 1. So, that's where it's possible that it could cause imbalance.
DR. TEMPLE: If there really is an international difference, then pooling them would exaggerate the study, since they had more of the patients getting treated.
DR. SRIDHARA: Of course, it's evident that in the U.S. study, the preventive goal was picked at the primary goal more often than in the Europe study.
DR. NERENSTONE: Dr. Blayney.
DR. BLAYNEY: I'm trying to think where I'm wrong here. In spite of my concerns that this preparation is an elaborate way of giving epinephrine into a tumor and a low dose weekly cisplatin, no one disputes the fact that it shrinks tumors 25 to 30 percent of the time. And the statistical manipulations and machinations, rather, that you've done really revolve around an unverified pretty crummy instrument for trying to assess what's happening with symptoms in the head and neck. I think that's a very difficult area to assess symptoms. Getting people completely off of narcotics in 3 weeks is a big jump, and I think if an investigator says if didn't break through the skin or erode into the carotid artery or obstruct an airway, you sort of have to give the guy who's sitting there watching that happen or not happen the benefit of the doubt.
DR. FRYKMAN: I guess about all I can comment on that is that is exactly what the thinking was when the agency and the applicant worked together to sort out exactly that question. This was a trial that was designed to help sort that out. The results are the results. But I think that goes to the core of exactly what the intentions were. And if you had polled a half a dozen people at that time, I think people would have said, yes, it makes sense. You got a problem. You make the tumor shrink. The problem should get better.
DR. SRIDHARA: I just want to add regarding the analysis part itself. It is comparing to placebo and response rate, this is local response rate, this is not systemic response rate. I think you have to keep that in mind. We have already seen the survival graph that there's no difference between them. So, if the intention is to give some clinical benefit to the patient, then we are not seeing it in survival, so we ought to see it in some other clinical benefit. Even if we give all the preventive benefit that the sponsors are claiming, even by their own analysis, individually each study has not demonstrated a significant effect between placebo and the treatment arm.
DR. BLAYNEY: And I guess I'd go back to the statement that the wrong -- I don't think we should penalize the sponsor for trying to mount a randomized controlled trial in this very difficult illness. We just don't have the tool and today you wouldn't design a study using this tool to get clinical benefit. And nobody in 1997 or 1994 was smart enough to come up with something that showed a clinical benefit other than it shrank tumors.
DR. NERENSTONE: Mr. Gruett.
MR. GRUETT: As a patient representative, most of my throat was removed. I was advised before the operation that I had a 15 percent greater chance of death through this procedure, but the quality of life thereafter would be wonderful. I elected to go with that option.
Looking at the time to progression, if cancer would have come back in myself after 6 months, I had options of radiation and other things. But prior to the 6 months, I had no option.
Could this drug have given me the option increasing the time to progression to where additional help after that 6 month period I could have received radiation? If that's the case, it would have some value for someone like myself. I ask that. Dr. Frykman I think would be qualified.
DR. FRYKMAN: Yes. I don't know that I'd be able to answer that. The sponsor may actually want to answer that some. But remember that the intent behind developing this therapy was to improve a single benefit in patients who could not have received other therapy. It's not intended to necessarily be a preventive modality in a case where there's not an identifiable and preventable problem. So, in your case, as I understand it, if you had recurred, again the other options would have been available to you, but this agent I don't think has ever been intended -- I don't think the sponsor intends it -- to be something that would be used to prevent a recurrence specifically.
DR. NERENSTONE: Dr. Temple.
DR. TEMPLE: I don't think we had going in a sure idea of what the best way to assess clinical benefit is. A lot of people in situations like this use visual analog scales, and you could have a visual analog scale for each of the symptoms too. Whether that would work better we don't know until somebody succeeds in showing something. The real test would be whether they prove useful when you actually find a drug that affects them.
But whether that's the reason for any difference would be, I think, extremely hard to know. But it may be these steps were too far apart for anything to have a reasonable chance of making a big difference. That is possible because there weren't very many big differences. As people have been saying, the effect is, to the extent it was there, driven by the prevention claims, which you've heard a discussion of, and when you actually get down to the palliation score, the effect in each study can't be shown. But whether a visual analog scale would work better we don't know. We certainly wouldn't object to it. But in the absence of a success history, it's hard to know.
DR. NERENSTONE: Dr. Glisson.
DR. GLISSON: Yes. I was actually just going to make a comment on Dr. Blayney's comments, and then he sort of went round robin and said it himself.
Just to help clarify this issue about preventive goals, I don't think they're unworthy, but the problem with the data as they stand is that you're unable to determine in the placebo arm what that baseline rate is of the event happening because there was such short follow-up in the placebo patients. We don't know, when we start out with somebody who's got a cervical mass near the carotid space, when they're going to have a carotid rupture or if they're going to have a carotid rupture. We might worry about it, but they might never have it before they die, or they might have it in 6 months because they live longer than we thought. So, it's just incredibly difficult without that data in the placebo group to know about the importance of these preventive endpoints.
DR. NERENSTONE: Dr. Lippman.
DR. LIPPMAN: And also to pick up on one of Dr. Blayney's other points is that I don't have any doubt in my mind -- and I wouldn't think anyone here would -- that this IntraDose is more active than placebo and it produces responses. So, if that were the question, it would be an easy one for me.
Obviously, the issue is the big word, "clinical benefit," and that's where this all rides. Is that degree of response, a 20 to 30 percent response, with a 2- to 3-month response duration in the two trials, a benefit over the toxicities and other issues?
Even the fact that it wasn't a validated tool and you think it's crummy and we have better ones now, which I'm sure you're right, I would feel better about what's there if it was done up front and, even more than that, if it was powered, if there was some way that each study could actually have the power to look at it.
DR. NERENSTONE: Dr. Albain.
DR. ALBAIN: Grant, could you expand a little more on why this could not qualify for accelerated approval on response data alone?
DR. WILLIAMS: The point I made was that two randomized, controlled clinical trials had been done to look at clinical benefit, and depending on what you think, let's say we didn't see it. That's usually what you do after accelerated approval. So, if you're going to do accelerated approval, first of all you have to have the right population and basically have no other options, and then you have to have some benefit that seems to be better than what's out there. And then you have to be able to design a trial that will show that it provides clinical benefit. So, I think the latter point is the problem here. If you've already done two trials and they didn't do it, is it reasonable to think you're going to show clinical benefit?
DR. ALBAIN: If we're all acknowledging that the tool may not be what we would choose right now, if we accept the response, can we not then go back to the sponsor and request clinical benefit data post accelerated approval that's designed a bit differently? I'm just asking.
DR. WILLIAMS: I think it's certainly something that ODAC could discuss.
DR. TEMPLE: Grant is saying he's not sure it meets the test for reasonably likely to predict clinical benefit when you've had a trial that, even pooled, doesn't really show that. But that's a judgment call. People could disagree about that obviously.
DR. NERENSTONE: Dr. Lippman.
DR. LIPPMAN: Yes. Again, the issue of accelerated approval. I never thought of this application in that context, but if you're asking do I think this is promising enough and is there room for another trial, given the issues we've talked about with the patient benefit, my answer would be yes.
It's remarkable the two studies that were done actually. These are extremely difficult to do. There has been data from intra-arterial platinum for a long time that's more anecdotal that shows you can get responses in patients like this, but nothing that's been as well studied, randomized with all the bells and whistles. So, there has been a tremendous job and it's sort of being undermined by this patient benefit, which I'd like to see.
So, although I don't think it's an accelerated approval issue, I don't think the issue has been resolved, and it would be nice to go on and do a larger study powered to look at the patient benefit, and I think it would be reviewed much differently here.
* DR. NERENSTONE: Maybe we should go on to an open conversation, if we've finished with questions with the FDA. Would anyone like to start the discussion? Dr. Glisson, did you have anything else you wanted to add?
DR. GLISSON: I guess I'll just say from sort of a global perspective, I'm actually one of the strong advocates for looking at palliation in patients with recurrent head and neck cancer. It's one of the things I always talk about when I give a lecture on this subject.
Unfortunately, what the sponsor of this trial has presented to us really just does not get at that issue. They've tried, but there are a number of issues that have been touched on by the committee, and I won't reiterate. For a lot of the reasons, we just fall short of the mark here in terms of showing that this is actually palliative.
It does produce some responses. That's clear, but in fact, it's a tree we've cut down and we have a forest to worry about. It's an injectable technique when we know that, if it's not clinically evident, we have subclinical disease, microscopic disease, lots of other places besides the area that's being treated, and it's doomed to fail.
I understand that it may be reasonable for a very small segment of the population of patients with recurrent disease where you really have nothing else to do and you're going to try to use it simply to reduce the possibility that they're going to have a horrible complication. But in fact, I think even that has not been shown.
So, I'll say right out I'm a paid consultant to the FDA and I, of course, helped them come to many of their conclusions today. So, what they've presented is pretty much what I felt.
I'll agree with Dr. Lippman that there are certainly some intriguing pieces of data here that would suggest the therapy might have value, especially in concert with systemic treatments.
DR. NERENSTONE: Dr. Couch had to leave early but asked that I make her comments. She is a head and neck surgeon.
She had serious concerns about this product. She said that she felt that the sponsor's discussion about who should not be eligible, in terms of affecting the carotid with bleeds and actual blindness in patients with the incidence of stroke, was unacceptably toxic and that they really could not differentiate who might be eligible and who should not be eligible. She felt that it's a drug that she would not be able to use and she would not recommend approval based on toxicity as well.
DR. PAZDUR: Could people discuss the concept of a response rate with a local injectable disease versus a response rate when we give systemic therapy? Because I think that this is kind of mixing apples and oranges here. Conventionally when we hear the term "response rate," we think of response rates associated with systemic therapies where you're eradicating, for example, micrometastases, tumors all over the body. You have a systemic effect with multiple tumors going down with lack of progression. It has a different concept to me than simply injecting something into a tumor that could be relatively small and in many cases surgery could handle this quite aptly in a very small surgical procedure in some cases.
So, could people discuss this concept? Because it's one that is coming in other applications to us. To me there is a fundamental difference here which I can't get a handle on in my own mind, so to speak. So, I'd like some discussion on this point.
DR. NERENSTONE: Dr. Redman.
DR. REDMAN: Yes. I made this comment at lunch to some of my associates. It would be something to see radiation therapy brought before this group to claim clinical benefit. I might be shot by my dean when I go back.
DR. REDMAN: I think what is inherently wrong here is biologically something is happening. It's the tool, and I don't know what the right tool is. There is something going on. I don't know what global measure to use for a local effect other than it shrunk.
Not all patients are going to have pain. The tool you used couldn't even predict those nonresponders. They all stayed stable with your tool, the vast majority of them. They didn't get worse with the tool. They didn't have more pain. They didn't have negative changes.
So, I sort of agree with Dr. Pazdur. I don't know what tool to use for a local effect other than something shrunk. If you restrict yourself to everybody who's having symptoms from that one local site, you're never going to close your trial. It would just stay open continually. So, I'm somewhat conflicted. I think biologically something is going on here, some potential benefit. I just don't know, other than tumors shrinking, how to measure it. I just don't.
DR. NERENSTONE: Dr. Lippman.
DR. LIPPMAN: Rick, I think a lot of us, being medical oncologists, have a concern with local treatment of a systemic disease just at first. So, once you get beyond that, which you had to do for this, I think the idea of being able to palliate big tumors that are symptomatic -- local failure is a huge problem in head and neck cancer. There's no question about it.
Again, Dr. Glisson raised the issue of having good control rates to be able to analyze the positive benefit in a preventive way. And that's why I asked the issue about systemic failure because I would suspect systemic failure, and if someone is benefiting locally from a big tumor and it has a small nodule metastatic to the lung, I wouldn't abandon treatment on this program because you would expect it.
I think you sort of answered the question in a sense when you said what does it mean when you have a 1.5 centimeter tumor, which is I guess the stratum 1. It would be much more compelling if we saw this kind of thing in the big tumors because then you have more of a clinical problem that's easier to justify that you really are benefiting someone even though you know that they're going to fail systemically later. So, those are at least the issues that I think about in terms of a local treatment for this disease.
DR. NERENSTONE: Dr. Sledge.
DR. SLEDGE: I must say it doesn't bother me at all that you have an objective response in a small area when things are growing elsewhere.
What does concern me is the agency's analysis that this wasn't a surrogate for anything, as far as we could tell. It wasn't a surrogate for any measurable clinical benefit. If there was some significant linkage between the two, I would find it an imminently reasonable endpoint. The problem is I don't see that in the analysis here.
DR. NERENSTONE: Dr. Kelsen.
DR. KELSEN: I agree with Bruce. The problem with this analysis that I'm wrestling with is the instrument to -- because the endpoint is palliation, but local control mechanisms are not limited to head and neck, as we all know. Radiofrequency ablation is used all the time for hepatic metastases in asymptomatic patients because they're small lesions. I think it is an issue for us to deal with in a real global way. It's, whatever the local treatment is, radiation included, how are we going to measure outcome, and I don't get a real sense today that we have a feel for that.
DR. NERENSTONE: Dr. Blayney.
DR. BLAYNEY: I would make the analogy to radiation to spinal lesions. They all hurt. You radiate them and after a while you can see healing with sclerosis, but very commonly people remain on narcotics. Occasionally some people will have paresis and you'll prevent progression to permanent paralysis.
I think the issue here, analogous to this issue, there's a lot of dilution of the people who really need it. If this study was confined just to people with spinal cord compression, you'd see a lot greater clinical benefit. No question. Half of them or two-thirds would walk afterwards. Whereas, here maybe the 3 out of 100 or so that had spinal cord compression are the ones that had the dramatic clinical benefit and they're lost in the noise.
DR. NERENSTONE: Dr. Temple.
DR. TEMPLE: The thrust of Rick's question I think was do you agree with basically what we've been telling people, that if you're going to do a local therapy, you really need to show that corresponds to some beneficial outcome, that it shouldn't be presumed. There might be some presumption about systemic responses, but here that doesn't make sense.
What I'm hearing people say is that they generally agree with that. Not that it's always easy to do because sometimes what you think is caused by the tumor is not really or whatever it is, and it proved difficult here or maybe the instrument was no good, but that we should keep pressing people to find an instrument or find a way or select the right people so that when their tumors in fact shrink, they feel better.
DR. PAZDUR: My feeling was is perhaps the response rate is somewhat a proof of concept, that yes, you can cause tumor shrinkage. But here again, there has to be that leap to the clinical benefit issue of symptoms. Especially if you're identifying these symptoms prospectively as the most bothersome symptom, why aren't you seeing an improvement and a correlation with tumor reduction?
DR. NERENSTONE: I guess I'm going to play devil's advocate a little bit because I don't think that the instrument they used was so bad. I think that maybe the instrument they used was okay, and the whole concept may not be as active as they had wanted. So, I applaud them for using that, but not every trial is going to be positive, even if you have the right questions and you have the right instruments.
Yes, they can come back and do it again, but I just think it may be a limitation of the drugs that are being tested.
DR. PAZDUR: The other point that I wanted to get at is when we talk about a response rate of 30 percent for a systemic therapy, obviously we're feeling there are other tumors shrinking. Should we expect more from something that we're injecting right into the tumor? If we were taking a look at radiation therapy, for example, many times people get a very high complete response rate there. What is the magnitude of benefit that one would want from something that is being directly injected?
DR. NERENSTONE: I think that might be a conversation that we shouldn't have at 6:30 today.
DR. PAZDUR: Agreed.
DR. NERENSTONE: Dr. Lippman.
DR. LIPPMAN: I can try to answer it.
DR. LIPPMAN: I don't have a flight to catch.
I think it depends on the population. We have the world's expert in re-irradiation with chemotherapy in Dr. Vokes here who can comment on that. I think if these patients can't tolerate re-irradiation and radiation is no longer an option, I think 30 percent in tumors that are symptomatic is meaningful.
Again, I agree with Stacy on this. We have better tools, but if this were powered differently, I think these kind of differences might have been significant in the individual studies and we'd be having a different discussion. That's why I think another trial that's powered may provide light.
DR. NERENSTONE: Dr. Carpenter.
DR. CARPENTER: If what I hear is true, I think the one thing most everybody agrees on is if there's a clinical benefit, you can't use this methodology to show it. But it may be that if there's a clinical benefit, you can't use this methodology equally well to show that it's not there.
So, I think what we know is there's a small, but definite response rate and that we don't know how to get a clinical benefit or to measure it in this group of people. Part of our problems have to do with the methodology used, and part of our troubles have to do with the lack of baseline knowledge about what to expect in this clinical situation. Both are important.
I think we're going to end up coming down to the line about a measurable benefit as far as shrinkage and really just not knowing if there's, in fact, a clinical benefit associated with this therapy given the information we have now. And the decision is going to hinge on what do we do with not knowing.
DR. NERENSTONE: Well, I think we're really directed to the data that we have here. We're really sort of limited.
I want to get to the questions. Two quick questions, Dr. Albain.
DR. ALBAIN: I just wanted to go back, though, if we are going to have a chance to consider the accelerated question on response alone since there were co-primary endpoints. You asked that we come back to it in the discussion. To me, that's another way to look at what we're all saying here.
DR. NERENSTONE: If we could turn our attention to the questions.
DR. ALBAIN: Wait. I didn't get an answer to my question.
DR. NERENSTONE: I'm sorry.
DR. ALBAIN: I asked if we are going to be able to either write another question or have some discussion about that. Rick?
DR. WILLIAMS: Why don't you finish these and we'll see if you want to have another.
DR. ALBAIN: Okay.
* DR. NERENSTONE: The first question. Do 1-point changes on the palliative scale developed by the applicant represent significant clinical benefit within the context of the clinical trials 414 and 514? And if so, do the data in the charts presented from the primary palliative goals represent significant evidence of clinical benefit that outweighs the toxicity of treatment with the cisplatin/epinephrine gel in patients with symptomatic recurrent head and neck cancer?
People's comments specifically to this question?
DR. NERENSTONE: Okay. Then we'll take a vote on the first one. Does the 1-point change represent significant clinical benefit? We have to go around the table. Dr. Glisson?
DR. GLISSON: I'm not a voting member.
DR. NERENSTONE: Okay. Dr. Kelsen?
DR. KELSEN: No.
DR. NERENSTONE: I'm told you are a voting member for this. You are allowed to pass and reconsider.
DR. GLISSON: I'm sorry. I was kind of ignoring you because I didn't think I could vote.
DR. TEMPLE: This really asks whether the scale they used is reasonable.
DR. GLISSON: I think the 1-point difference is significant, as long as blindedness is maintained.
DR. NERENSTONE: So, that's a yes.
DR. KELSEN: No.
DR. ALBAIN: Yes.
MR. GRUETT: No.
DR. LIPPMAN: Yes, with the caveat that I know nothing about these scales or what's valid or not. I really would almost like to abstain. But my issue is more concerned with what Dr. Glisson raised, the issue of a control group and others, and less the 1-point issue.
DR. NERENSTONE: Is that a yes or a no or an abstention?
DR. LIPPMAN: Yes.
DR. CARPENTER: No.
DR. PRZEPIORKA: Yes.
DR. NERENSTONE: Yes.
DR. SLEDGE: Yes
DR. PELUSI: Yes.
DR. RUBINSTEIN: Yes.
DR. REDMAN: Yes, but it's still a poor instrument.
DR. BLAYNEY: Yes.
DR. NERENSTONE: So then, the question is, do the data represent significant evidence of clinical benefit that outweighs the toxicity of treatment?
Sorry. 10 yes, 3 no.
The second part. Do the data represent significant evidence of clinical benefit that outweighs the toxicity of treatment?
Is this really the approval question? Not yet, okay.
DR. TEMPLE: No. But because it includes both evidence of benefit and toxicity, it's very close. Maybe we should have written it differently.
DR. NERENSTONE: Dr. Blayney?
DR. BLAYNEY: Yes.
DR. REDMAN: Abstain.
DR. RUBINSTEIN: No, but I'd like to say that this begs the question of whether the preventive goals have been addressed adequately and whether the FDA's objections to using the preventive data are something that we agree with.
DR. PAZDUR: We can come back to that question.
DR. NERENSTONE: Okay, we'll come back to the preventive as an endpoint.
So, you're a no. Correct?
DR. RUBINSTEIN: I'm a no.
DR. TEMPLE: Stacy, before you leave that, it's worth remembering that even the company's analysis for each study doesn't show a benefit, including the preventive. So, keep that in mind too.
DR. PELUSI: No.
DR. SLEDGE: No.
DR. NERENSTONE: No.
DR. PRZEPIORKA: Abstain.
DR. CARPENTER: No.
DR. LIPPMAN: No, again for the reasons that Dr. Glisson reiterated. We just don't have the data. We don't have the data of the control rates in the placebo. I think that's why I say no.
MR. GRUETT: No.
DR. ALBAIN: Abstain.
DR. KELSEN: No.
DR. GLISSON: No.
DR. NERENSTONE: There are 3 abstentions, 1 yes, and 9 noes.
Question 2. Do the following data on response rate independently represent clinical benefit of treatment with the CEG that outweighs its toxicity in the treatment symptomatic of recurrent head and neck cancer? Does the response rate represent clinical benefit?
They remind you that 65 percent of the patients that were responding were from stratum 1, or tumors less than 5 centimeters cubed.
DR. GLISSON: No.
DR. KELSEN: I'm going to interpret this to mean does shrinkage alone mean something clinically, and my answer is no.
DR. ALBAIN: I don't think we know. I'm going to abstain again. I've never abstained in three years. It's two in one vote.
MR. GRUETT: No.
DR. LIPPMAN: No.
DR. CARPENTER: No.
DR. PRZEPIORKA: No.
DR. NERENSTONE: No.
DR. SLEDGE: No.
DR. PELUSI: No.
DR. RUBINSTEIN: I'm going to interpret to this to mean that whatever clinical benefit may or may not pertain to the response rate outweighs the toxicity, and on that basis, I say yes.
DR. REDMAN: Again, this is the bona fide response rate, not the clinical benefit. This is the clinical tumor shrinkage response rate.
Based on the question, I'd have to say yes.
DR. BLAYNEY: Yes.
DR. NERENSTONE: 1 abstention, 3 yes, 9 noes.
Number 3. Please discuss the clinical value of local treatments for head and neck cancer in patients with systemic disease or patients with locoregional progression.
DR. PAZDUR: You've kind of done this already, for the sake of time.
DR. NERENSTONE: Right. Do you want a vote?
DR. PAZDUR: There's no vote possible. It's a discussion point and I think we've discussed this already.
DR. NERENSTONE: Number 4. Do these trials provide substantial evidence that the cisplatin gel is safe and effective in the treatment of symptomatic recurrent head and neck cancer?
DR. TEMPLE: I think before you do that, you have to do Dr. Albain's question because you can provide substantial evidence for a surrogate approval too and then your standards are different. So, until you grapple with that, at least a little bit, you can't really answer this.
DR. WILLIAMS: One issue that would have to be addressed before you grapple with accelerated approval would be that the population that was studied had no alternative therapies. That was not the whole population. So, there's some difficulty in discussing that. Do you have any suggestions?
DR. TEMPLE: Well, that's a reminder of what accelerated approval refers to. It means there has to be no alternative or no good alternative or this is better than existing alternatives, and it has to be for a serious or life-threatening disease. I guess that one is fairly easy.
DR. WILLIAMS: I guess what I'm saying is I guess before we really grappled with it as an agency, we'd have to do some more review of the application to see how many patients there were that had no available therapy and what the response rate was. So, whatever you discuss begins with the caveat that we may then decide that there's not sufficient data.
DR. PAZDUR: Here again, the actual ruling on this is that the response rate or the surrogate endpoint must reasonably likely be predictive of clinical benefit. We'd just like to remind people that there has to be this link. It's just not that it's there. The size of the lesions, the location of the lesions might come into play here. What is the magnitude of a local response rate since we do not have a lot of experience in dealing in this area?
DR. WILLIAMS: I think it sort of goes back to Dr. Carpenter's comment. If you believe that this trial was designed with a reasonable chance of finding clinical benefit and didn't, then I think that's a problem for going forward with those discussions. If you believe that the trial was so insufficiently designed or powered to detect it, so that you believe that it's reasonably likely that the response does predict clinical benefit, I think that would have to be the circumstance under which you made further discussion of the issue.
DR. NERENSTONE: Just one comment about this. I'll open up the discussion. Talking to Dr. Couch who, as I said, is a head and neck surgeon, we talked about, well, what about these big tumors where you're worried. And I think that gets back to Dr. Lippman's concern. We saw the high response rate, but it's really in the small tumors that really are not causing life-threatening problems, at least at the time they're being treated. The whole issue of prevention is another whole issue.
But at least in terms of response rate, her feeling was that this drug is extraordinarily unsafe for large tumors in the neck especially because of the prior surgeries. Her feeling was that was exactly the patient population who you want to treat because you have very little options, but in whom the toxicity is much too high, and that the worst thing you want to do with these patients is hasten their demise by a product that has questionable benefit.
DR. LIPPMAN: I don't know how often this happens, but I assume there are some cases where you do reflect and we reflect to the sponsor that there's some very provocative findings here that we'd like to see followed on and evaluated along the lines we've indicated. To me that isn't synonymous with accelerated approval.
DR. WILLIAMS: Right. I agree.
DR. TEMPLE: Accelerated approval means it gets marketed while that happens, and a good suggestion means it doesn't.
DR. LIPPMAN: Right. That's where, as I indicated before, I draw the line. I'd like to see some of this followed up, but I don't think we have the data now to make an accelerated approval.
DR. WILLIAMS: Why don't I go ahead and suggest a question? Based on these data, do you think this tumor response rate is reasonably likely to predict clinical benefit after all that we've been through here? That's the first question that I think is reasonable to ask.
DR. NERENSTONE: Do you want a vote?
DR. PAZDUR: That's the essence of the accelerated approval. So, basically would the response rate presented in small tumors -- in fact, many of these occurred in stratum 1 or the majority occurred in stratum 1. Would this be reasonably likely to predict for clinical benefit?
DR. NERENSTONE: Grant, do you want to repeat the exact wording so that Karen can get it down?
DR. WILLIAMS: Okay. Would the response rate which has been seen in these trials be reasonably likely to predict clinical benefit? I mean, it's in these trials. These are the trials that we have. So, we don't need to specify. It's what these trials were.
DR. FRYKMAN: I'd like to add one comment, and that has to do again with accelerated approval. The presumption with accelerated approval is that there is an association between the surrogate and the actual clinical benefit. Again, we'll go back to the data that was presented here. Even under the sponsor's analysis, let alone the FDA's analysis, that linkage or what we're calling a linkage is actually very weak. In our more conservative analysis, it was 13 percent in the 414 study and 38 percent in the European study, which was conducted better. So, it seems to me that there is good evidence that there is no association.
DR. NERENSTONE: Dr. Temple.
DR. TEMPLE: If you believe the numbers, a 40 or 50 percent sensitivity, I don't believe it's reasonable to call that very weak. That would be very impressive if it were true. What you've heard is discussion of why our review doesn't think that it's true. So, I don't think that's the problem.
I think the problem is what you're talking about. Is the observed response rate, given the clinical data -- you're allowed to look at that -- something that makes you think that these responses are likely to lead to clinical benefit, with all the reservations people have said, that they're small, that to the extent people have been able to look at this endpoint, you didn't see that much, but you didn't see absolutely nothing. All of those things. How does that add up?
And then later on, you'd weigh that against the observed toxicity, but that's a different question.
DR. WILLIAMS: I think basically the credibility of the data from this instrument -- if we believe that the data are credible -- we believe that there was no result. If you don't believe the data are credible, then I think there could have been a result there that we didn't see. And it would fall back to your original suspicion or reasonable likeliness that there is a correlation.
DR. NERENSTONE: So, do you want us to vote on this? Okay.
DR. RUBINSTEIN: Could I just make one statement? The response rate for stratum 2 is 21 percent. Now, that may be considered low, but it's there. The response rate for stratum 1 is 37 percent. So, there definitely is a difference, but it's not that all the responses are occurring in stratum 1.
DR. SRIDHARA: Overall, of the responses that are there, two-thirds of them are coming from stratum 1, and one-third are coming from stratum 2.
DR. NERENSTONE: The question before us, will the response rate presented in these trials be reasonably likely to predict clinical benefit? Dr. Glisson, why don't we start with you.
DR. GLISSON: Yes. I believe we actually already voted on this issue. I'll say no again.
DR. KELSEN: No.
DR. ALBAIN: Possibly.
DR. NERENSTONE: You only have three choices: yes, no, or abstain.
DR. ALBAIN: All right. I will do it again. Yes.
MR. GRUETT: No.
DR. LIPPMAN: Again, this is related to question 2, which I think is what Dr. Glisson was referring to. I had my hand up because it would be nice to qualify the question a little bit just to indicate response rate and response duration because that's really what I'm looking at. There was an overwhelming percentage of CRs here and they were very durable. So, the response rate itself for the CR rate -- I don't know how I'd interpret that, but the fact that the median durations were 2 to 3 months influences my answer of no.
DR. WILLIAMS: It's meant to reflect the CR, PR, and duration.
DR. LIPPMAN: So, your question reflects duration also.
DR. WILLIAMS: Right.
DR. LIPPMAN: No.
DR. CARPENTER: Yes.
DR. PRZEPIORKA: Not to upstage Scott, but as Dr. Rubinstein pointed out, the CR rate is pretty impressive. If you actually go through the breakdown of who responds the most, it's people with oropharyngeal lesions which may be less than 5 centimeters, but they're in a very small space unless they're a Steelers' fan or something.
Time to progression, as he pointed out, is better with the gel.
There is a higher dropout rate in the placebo because of greater progression in the placebo. So, people on the treatment arm aren't dropping out because of toxicity, and in fact, if you look at dropouts for toxicity, they aren't very much different.
I think the tool that they used to look for clinical benefit is absolutely crummy. There are all sorts of other indicators that there's probably clinical benefit there, but somebody has to figure out what it is.
So, the answer to the question is yes.
DR. NERENSTONE: No.
DR. SLEDGE: No.
DR. PELUSI: No.
DR. RUBINSTEIN: Yes.
DR. REDMAN: Yes.
DR. BLAYNEY: My answer is yes. I think that this is going to get applied early in the clinical course before tumors get huge, and they can probably pick some critical areas to get at the clinical benefit.
DR. NERENSTONE: And the results are 6 yes, 7 noes. Pretty evenly divided there.
Do we hit number 4?
DR. TEMPLE: But you do have to divide the last into conventional and --
DR. PAZDUR: Well, that's the next question. The last question is basically safe and effective and demonstrating clinical benefit.
DR. TEMPLE: We'll take the previous vote as meaning the people who thought yes would probably agree that it should be approved under accelerated approval. And this is the question about not under accelerated approval.
DR. NERENSTONE: So, number 4 is for standard approval.
DR. PAZDUR: Yes, demonstration of clinical benefit.
DR. NERENSTONE: Do these trials provide substantial evidence that the cisplatin gel is safe and effective in the treatment of symptomatic recurrent head and neck cancer? So, the application as it stands now for regular approval.
DR. BLAYNEY: No.
DR. REDMAN: This is for regular approval.
DR. NERENSTONE: This is for the application we have now.
DR. REDMAN: No.
DR. RUBINSTEIN: No.
DR. PELUSI: No.
DR. SLEDGE: No.
DR. NERENSTONE: No.
DR. PRZEPIORKA: No.
DR. CARPENTER: No.
DR. LIPPMAN: My answer is no, but I again think I'm not sure the vote would have been the same if it was listed as accelerated approval. I don't know how important that is to you, but I think --
DR. TEMPLE: We're assuming that. We're assuming the previous vote represents accelerated approval. 7 noes, 6 yes.
DR. LIPPMAN: I'm not sure that's the case. I'd be interested to see what the vote is if accelerated approval is put on the table. So, if that's important to you, it might be worth voting on it.
But the answer to this question is no.
MR. GRUETT: No.
DR. ALBAIN: No.
DR. KELSEN: No.
DR. GLISSON: No.
DR. NERENSTONE: Do you want an accelerated approval vote?
DR. PAZDUR: If would like to. We considered the previous question reasonably likely to demonstrate clinical benefit equivalent to an accelerated approval because that is the criteria that we would use to make that decision.
DR. NERENSTONE: Do you want it to be explicit, committee, so that you understand that's what we're now talking about? So, if this application were an accelerated approval, meaning that there would be phase IV commitments to do other studies to show clinical benefit, because that's the only way it could get full approval is if clinical benefit were shown under the rules of accelerated approval, so that we would approve this application with those further commitments.
DR. PAZDUR: And the drug would be fully marketed here also, commercially available.
DR. WILLIAMS: As opposed to the previous question, your vote would indicate that you do think that the response rate is reasonably likely to predict clinical benefit, and for that reason, it's reasonable for us to approve the drug and then demonstrate the clinical benefit. So, by voting for accelerated approval, you'd also be voting the previous question.
DR. TEMPLE: It implies that the benefit suggested by the response rate outweighs whatever risks there are too. So, that is a little different from the previous question.
DR. LIPPMAN: And the drug would be marketed and then used.
DR. PAZDUR: Correct.
DR. NERENSTONE: But there would have to be a commitment for a phase IV trial. Is that likely to get done?
DR. PAZDUR: That's what you have to consider also in making a determination here because the negotiation with the company is that these trials would be done with due diligence.
DR. NERENSTONE: Okay. To restate the question, should this be approved under the accelerated approval mechanism? Dr. Blayney, would you like to start?
DR. BLAYNEY: I'm going to --
DR. HOWELL: Madam Chairman, a point of information. The company has indicated that it is more than happy to make that phase IV commitment.
DR. NERENSTONE: A wise business decision.
DR. PAZDUR: They haven't heard what the phase IV commitment is, though.
DR. NERENSTONE: We'll have some discussion. I think this is going to be a very important vote. Dr. Albain.
DR. ALBAIN: I think it would be absolutely critical that a panel of head and neck surgeons, ENT oncologists, medical oncologists that major in head and neck cancer be involved in a detailed prescription for the labeling that goes beyond just the fine print, but a box portion that would address all of Dr. Couch's concerns and many of the rest of us about in which patients this should be applied once it's on the market.
DR. NERENSTONE: Dr. Sledge.
DR. SLEDGE: I think we need a little reality testing here. This trial took six years to accrue. I've heard several people around the table, including several of those who think this is a wonderful idea for accelerated approval, say that the instruments used in this trial were crummy. I heard that word from more than one person. So, what we're talking about is developing a new instrument, validating it, and then doing a larger phase III trial? I mean, whose grandchildren are going to approve this drug?
DR. NERENSTONE: Dr. Rubinstein.
DR. RUBINSTEIN: Yes, those concerns aside --
DR. RUBINSTEIN: What is the usual procedure? Is a window set of some number of years?
DR. PAZDUR: Well, we follow these commitments, but the actual wording is -- and it is one for the interpretation of the division and the agency -- with due diligence basically. We need to have a realistic expectation that this trial can be done vis-a-vis Dr. Sledge's comments.
DR. NERENSTONE: In most cases, though, as a point of information, we don't usually do it this way. That is, an application comes where there is a response rate, and it's in a situation where there are no other options available. Then because you want the drug on the market, you say, we think that this response rate is going to translate into clinical benefit as we define it in the future for a phase IV commitment. So, this is a little bit different interpretation in the process of drug development.
DR. PAZDUR: Here again, there are some caveats here. We would have to take a look at the number of patients that truly represent an unmet medical need here.
Please remember also that this would be precedent setting in the sense that other people that fail their clinical trials could make claims that they have underpowered trials, that they used inappropriate scales to measure clinical benefit, et cetera, and they could come back and say, well, can we now have accelerated approval. So, this is somewhat a precedent-setting situation.
I don't know, Bob, if this has happened before.
DR. TEMPLE: Undoubtedly.
What's unusual about it is that usually you have a trial that was only looking at response rate. So, you didn't not find a clinical benefit. There was never any chance that you were going to find a clinical benefit. Here there was an attempt to look at a clinical benefit.
But people have said two opposite things. Some have said it was a good shot, used a good endpoint. Well, it was a little too small but it wasn't that small. So, that in some sense, if you believe that, argues that they've already tested that question.
If, in contrast, you believe that the endpoint was crummy and the size of the study gave them almost no hope, then you might believe that, even though they tried, the clinical endpoint has not been assessed and that there is a reasonable possibility that sometime in the next 30 years they would succeed in doing it.
So, it is a little unusual that way.
DR. NERENSTONE: I do want to remind the committee that we had a similar problem with oxaliplatin, if you remember. The endpoints that were in the application were not fulfilled, and there we encouraged the investigators to reapply with a better endpoint. We did not go for accelerated approval. So, that's just to remind people we have been in this situation before where we think there is a drug with a response rate that might be significant.
DR. PELUSI: I guess one of the things that bothers me is that I think we're all struggling with the fact that you see some response, and we all want that to be a good response. But when we look at the numbers in the community, and you're asking physicians to do a procedure that they may not do a lot of, the question is, where is it the safest done?
So, if we really start to look at that, I think that that puts something else on the table in terms of is there a subset of patients that maybe should be treated at a certain setting where people actually have that type of experience versus trying to do one every one or two years and not have the skills that some of the people --
DR. PAZDUR: Part of accelerated approval could allow us to restrict access to this drug and have it be used at only specialized centers if that is what you are suggesting.
DR. PELUSI: Well, I guess I'm struggling with that because if it just goes on the market, the question is then is it a free-for-all, and I don't think any of us would intend that to be so.
DR. NERENSTONE: Mr. Gruett.
MR. GRUETT: In an accelerated approval, aren't you offering the patients hope? And I don't believe there's enough hope generated here to offer that feeling within the patients. I would more look at going back into a phase III trial then going into an accelerated approval trial.
DR. NERENSTONE: Dr. Przepiorka.
DR. PRZEPIORKA: Two comments. I think if there is a phase IV commitment to do a clinical benefit analysis in another trial, there better be some urgency in getting the tool done and validated, and if it is done successfully, it will make a major contribution to oncology because there are other situations in which such a tool would be valuable.
With regard to the caveat that Rick brought up, I would hope that in the future, if clinical benefit is going to be a primary outcome, that the FDA has to slam the door and put clinical trials on hold rather than just give some advice and hope that the companies will follow it and that it is not going to let the company go ahead and do a trial where the statistical section is not completed and then turn around and --
DR. PAZDUR: Donna, things have changed since then. Obviously, there are special protocol assessments which we've implemented basically for all of our protocols going through that lock us into agreements, and those protocols are reviewed. The provisions for special protocol assessment have occurred after these protocols were initiated. So, unfortunately, there's a lag time here that the party was not able to benefit from that.
DR. NERENSTONE: Dr. Howell.
DR. HOWELL: Just a point of clarification for the committee. A subsequent trial combining IntraDose with systemic chemotherapy is well underway with a variety of other clinical benefit endpoints.
As you all noted, there are a lot of problems with this instrument. That doesn't mean that an appropriate instrument cannot be found and cannot be used to make a good association between tumor response rate, which we all agree is there, and clinical benefit. I think we would ask for your consideration in giving the sponsor the opportunity to try to nail down exactly this kind of issue for the purpose of broadly helping the whole field.
DR. NERENSTONE: But that can be done without granting accelerated approval. That study is ongoing.
DR. HOWELL: I remind you that this is a tiny patient population who have no other therapeutic options. I would doubt very seriously that this randomized trial would ever be done if there isn't encouragement from a regulatory strategy moving forward in this disease.
DR. NERENSTONE: Dr. Frykman.
DR. FRYKMAN: Yes. I just wanted to make one comment from the FDA perspective. If accelerated approval is considered and in fact it's granted, then we still have to write a label for this drug. As you recall, in the trial there were a number of dosing errors. In fact, frankly, the truth of the matter is we don't know what dose to have in the label. The sponsor proposes .25 ml per cc, and there's some evidence that was in the briefing document that that may, in fact, not be the right dose. It might actually provide an inferior response rate compared to the .5 ml per cc of tumor.
So, I would like to see if it's possible for the phase IV commitment, if that is even being considered, that in fact a consideration of reformulation and perhaps increasing the concentration, while keeping the volume of the gel down while still providing the same amount of platinum per tumor, might be a useful way to go.
DR. HOWELL: A point of information: That cannot be done, sir. You're at maximum concentration of drug in this gel.
DR. NERENSTONE: Last comment, because otherwise we're going to be here till midnight. Dr. Lippman.
DR. LIPPMAN: The issue of sending encouragement to the sponsor to follow that -- I mean, I think there must be other ways other than accelerated approval to do that. We've had this discussion. People are intrigued. There's definitely clinical activity here in terms of tumor responses. I think we've suggested some ways to address that and that we do think -- at least the sentiment of the committee -- a number of us do think it should be followed up. And that's not the kind of encouragement we give every application, as you know. It hardly ever comes up.
DR. PAZDUR: You should vote for this because it is reasonably likely to predict clinical benefit, not on the basis of sending messages to sponsors.
DR. TEMPLE: Because you believe it is. He's not telling you what to vote.
DR. NERENSTONE: Let's take a vote and everybody can have their last comment as we go around. The question is, should this be approved under the accelerated approval mechanism? Dr. Glisson.
DR. GLISSON: No.
DR. KELSEN: No.
DR. ALBAIN: Yes.
MR. GRUETT: No, but I'd like to see more phase III study.
DR. LIPPMAN: No. Ditto to that last comment.
DR. CARPENTER: Let me think a minute.
DR. PRZEPIORKA: Yes.
DR. NERENSTONE: No.
DR. SLEDGE: No.
DR. PELUSI: No.
DR. RUBINSTEIN: Abstain because I'm concerned that if we did go for accelerated approval, we would actually make the trial harder to do if the drug were commercially available.
DR. REDMAN: I think clinically something is going on, but I can't agree to letting the drug out on the market.
DR. NERENSTONE: I take it that's a no.
DR. REDMAN: No.
DR. BLAYNEY: I'm going to vote yes to be consistent with my previous votes, although I'm wary of setting a precedent here.
DR. NERENSTONE: Dr. Carpenter.
DR. CARPENTER: No.
DR. NERENSTONE: 1 abstention; yes, 3; 9 noes.
Thank you very much for this marathon session. We will begin tomorrow at 8 o'clock.
(Whereupon, at 7:03 p.m., the committee was recessed, to reconvene at 8:00 a.m., Tuesday, September 11, 2001.)_