1
DEPARTMENT OF HEALTH AND HUMAN SERVICES
FOOD AND DRUG ADMINISTRATION
CENTER FOR DRUG EVALUATION AND RESEARCH
ONCOLOGIC DRUGS ADVISORY COMMITTEE
70TH MEETING
Thursday, January 31, 2002
8:00 a.m.
Advisors and Consultants Staff Conference Room
5630 Fishers Lane
Rockville, Maryland
2
PARTICIPANTS
Stacy Nerenstone, M.D., Chairperson
Karen M. Templeton-Somers, Ph.D., Executive
Secretary
MEMBERS
Kathy Albain, M.D. (By Videoconference)
Otis W. Brawley, M.D.
Stephen L. George, Ph.D.
David P. Kelsen, M.D. (By Teleconference)
Scott Lippman, M.D.
Jody L. Pelusi, F.N.P., Ph.D.
(Consumer Representative)
Donna Przepiorka, M.D., Ph.D.
Sara A. Taylor, M.D. (By Videoconference)
VOTING CONSULTANT
Derek Raghavan, M.D., Ph.D.
NON-VOTING GUEST SPEAKERS
Philip Bonomi, M.D.
Patrick J. Loehrer, M.D.
VOTING PATIENT REPRESENTATIVE
Eugene J. Kazmierczak
FDA
Amna Ibrahim, M.D.
Richard Pazdur, M.D.
Nancy Scher, M.D.
Robert B. Temple, M.D.
Grant Williams, M.D.
3
C O N T E N T S
Call to Order:
Stacy Nerenstone, M.D. 4
Conflict of Interest Statement:
Karen M. Templeton-Somers, Ph.D. 6
Open Public Hearing:
Ann E. Fonfa (letter) 8
NDA 21-386, Zometa (zoledronic acid for injection)
Novartis Pharmaceuticals Corporation
Sponsor Presentation
Introduction:
Burkhard Daldrup, Ph.D. 10
Pathophysiology of Metastatic Bone Disease
and the Role of Bisphosphonates:
Robert Coleman, M.D., FRCP 16
Zometa in Breast Cancer and Multiple Myeloma:
James Berenson, M.D. 38
Paul Gallo, Ph.D. 48
James Berenson, M.D. 52
FDA Presentation
Zometa in Breast Cancer and Multiple Myeloma:
(Study 010)
Grant Williams, M.D. 63
Questions to the Committee 78
Sponsor Presentation
Zometa in Prostate Cancer and Solid Tumors
Other than Prostate Cancer and Breast Cancer:
Matthew Smith, M.D. 102
Robert Coleman, M.D., FRCP 117
Conclusions:
David Parkinson 129
FDA Presentation
Zometa in Prostate Cancer and Solid Tumors
Other than Prostate Cancer and Breast Cancer
(Studies 039 and 011):
Amna Ibrahim, M.D. 134
Safety Review (Studies 010, 011 and 039)
Nancy Scher, M.D. 144
Questions from the Committee 149
Introduction to the Issues 171
Committee Discussion and Vote 174
4
1 P R O C E E D I N G S
2 Call to Order
3 DR. NERENSTONE: I would like to welcome
4 everyone to ODAC, our 70th meeting. We still start
5 with our usual introduction of the committee.
6 Kathy, can you hear us?
7 DR. ALBAIN: Yes; I can. Good morning.
8 DR. NERENSTONE: Why don't you start with
9 the introductions. We will go around the table.
10 You are first.
11 DR. ALBAIN: Kathy Albain, medical
12 oncology, Loyola University, Chicago.
13 DR. LOEHRER: I am Pat Loehrer from
14 Indiana University.
15 DR. BONOMI: Phil Bonomi, medical
16 oncology, Rush University in Chicago.
17 DR. RAGHAVAN: Derek Raghavan, U.S.C. in
18 Los Angeles.
19 DR. GEORGE: Stephen George, Duke
20 University.
21 DR. LIPPMAN: Scott Lippman, M.D. Anderson
22 Cancer Center.
23 MR. KAZMIERCZAK: Gene Kazmierczak,
24 patient representative.
25 DR. PRZEPIORKA: Donna Przepiorka, Baylor
5
1 at Houston.
2 DR. TEMPLETON-SOMERS: Karen Somers,
3 Executive Secretary to the committee, FDA.
4 DR. NERENSTONE: Stacy Nerenstone, medical
5 oncology, Hartford, Connecticut.
6 DR. BRAWLEY: Otis Brawley, medical
7 oncology, Emory University, Atlanta.
8 DR. PELUSI: Jody Pelusi, oncology nurse
9 practitioner, Phoenix Indian Medical Center and
10 consumer rep.
11 DR. SCHER: Nancy Scher, medical oncology,
12 FDA.
13 DR. IBRAHIM: Amna Ibrahim, medical
14 officer, FDA.
15 DR. WILLIAMS: Grant Williams, medical
16 team leader, FDA.
17 DR. PAZDUR: Richard Pazdur, Division
18 Director, Oncology Drugs, FDA.
19 MS. TEMPLETON-SOMERS: I would like to
20 welcome everyone to our conference room here in our
21 office. It is unusual for us to hold ODAC here but
22 this is sort of an unusual ODAC, a meeting with a
23 late addition to our schedule and so our hotel
24 choice was very limited. We apologize in advance
25 for any crowding, for the crowd, mostly, but we
6
1 thought it was important to get the meeting
2 scheduled in January rather than waiting for our
3 regularly scheduled meeting in late February.
4 This meeting is also ground-breaking in
5 that it represents our first steps into the world
6 of having member participation by electronic means.
7 Dr. Kathy Albain and, hopefully, Dr. Sarah Taylor
8 are participating by videoconferencing from their
9 home bases in the Midwest. Sarah may not be
10 joining us because we have heard news that Kansas
11 City has no power due to the snow storm. So even
12 electronic means are not going to get around that.
13 Kathy was saying that she might be grounded in the
14 airport, too.
15 Dr. Kelsen will be joining us from New
16 York by a combination of webcasting, by which he is
17 going to watch the proceedings, and
18 teleconferencing.
19 I am going to go right into the conflict
20 of interest.
21 Conflict of Interest Statement
22 MS. TEMPLETON-SOMERS: The following
23 announcement addresses the issue of conflict of
24 interest with respect to this meeting and is made a
25 part of record to preclude even the appearance of
7
1 such at this meeting.
2 Based on the submitted agenda an
3 information provided by the participants, the
4 agency has determined that all reported interests
5 in firms regulated by the Center for Drug
6 Evaluation and Research present no potential for a
7 conflict of interest at this meeting with the
8 following exceptions.
9 In accordance with 18 USC 208(b)(1), Jody
10 Pelusi, R.N., Ph.D., has been granted a waiver for
11 serving on an advisory board for a competitor and
12 for her speaking for a competitor. She receives
13 less than $10,000 a year for her participation on
14 the advisory board and from $5,000 to $10,000 a
15 year for her speaking.
16 In addition, Scott Lippman, M.D., has been
17 granted a waiver under 18 USC 208(b)(3) for his
18 consulting for a competitor on unrelated matters.
19 He receives from $10,000 to $50,000 a year for his
20 consulting.
21 A copy of these waiver statements may be
22 obtained by submitting a written request to the
23 agency's Freedom of Information Office, Room 12A-30, at the
24 Parklawn Building.
25 With respect to FDA's invited guests, Dr.
8
1 Philip Bonomi has a reported interest that we
2 believe should be made public to allow the
3 participants to objectively evaluate his comments.
4 Dr. Bonomi is a scientific advisor for Genentech
5 and OSI.
6 In the event that the discussions involve
7 any other products or firms not already on the
8 agenda for which FDA participants have a financial
9 interest, the participants are aware of the need to
10 exclude themselves from such involvement and their
11 exclusion will be noted for the record.
12 With respect to all other participants, we
13 ask, in the interest of fairness, that they address
14 any current or previous financial involvement with
15 any firm whose product they may wish to comment
16 upon.
17 Thank you.
18 DR. NERENSTONE: We will go now to the
19 open public hearing part and Ann E. Fonfa from the
20 Annie Appleseed Project. The letter will be read.
21 Open Public Hearing
22 MS. TEMPLETON-SOMERS: This letter is a
23 statement from Ann Fonfa of the Annie Appleseed
24 Project. "I am Ann Fonfa, a breast-cancer survivor
25 and activist, founder of the Annie Appleseed
9
1 Project which educates, informs, advocates and
2 raises awareness for those cancer patients, family
3 and friends interested in or using complementary
4 alternative natural therapies.
5 I have just finished giving NCI input on
6 their consumer guide for clinical trials in that
7 they show that most drugs take about fourteen years
8 to reach the approval stage. My question to this
9 body is why are we spending thousands of human
10 subject hours in constantly approving drugs that
11 are little better than the ones we cancer patients
12 already have access to.
13 There is something wrong with this entire
14 system when the best that we can do is offer a drug
15 that has just about the same safety profile, just
16 about the same response results but differs in a
17 very minor way. We patients are, therefore,
18 condemned to live out our lives, however long that
19 may be, with no real advances in treatments.
20 I resent this and I am taking this
21 opportunity to say so. Aim higher. We are all
22 tired of crawling on our hands and knees through a
23 field of broken glass. We want to leap over it
24 and, for that, we need new drugs that are different
25 and that make a real difference in our lives.
10
1 Thank you for your attention. If you
2 would like to find out more about our organization,
3 please go to our website,
4 www.annieappleseedproject.org.
5 FYI, I completely support the idea of
6 bisphosphonates for treatment of metastatic bone
7 disease."
8 DR. NERENSTONE: Are there any other
9 people for the open public hearing?
10 Seeing no one, then we will turn now to
11 the supplemental NDA for Zometa indicated for the
12 treatment of bone metastases in patients with
13 multiple myeloma, breast cancer, prostate cancer
14 and other solid tumors. Novartis will start with
15 their sponsor presentation.
16 NDA 21-386, Zometa (zoledronic acid for injection)
17 Novartis Pharmaceuticals Corporation
18 Introduction
19 DR. DALDRUP: Dr. Nerenstone, Dr. Pazdur,
20 Dr. Williams, members of the advisory committee,
21 FDA and guests, good morning.
22 [Slide.]
23 My name is Burkhard Daldrup. I am Global
24 Head of Drug Regulatory Affairs for Novartis
25 Oncology. On behalf of Novartis, I would like to
11
1 thank you for the opportunity this morning to
2 present and review our Zometa data for a new
3 indication in the treatment of bone metastases.
4 [Slide.]
5 Zometa belongs to a new class of highly
6 potent bisphosphonates. In August 2001, Zometa was
7 approved by FDA for its first indication of therapy
8 for the treatment of hypercalcemia of malignancy.
9 Zometa is currently approved for this indication in
10 more than sixty countries around the world.
11 A dossier for the treatment of bone
12 metastases was filed in July 2001 in Europe and a
13 supplemental application was also submitted in
14 August 2001 in the U.S. as well as in many other
15 countries.
16 The recommended dose is 4 milligrams
17 infused over 15 minutes administered every three to
18 four weeks. Novartis is also evaluating
19 nononcologic indications for Zometa including, at
20 this time, Paget's disease, osteoporosis and
21 rheumatoid arthritis.
22 [Slide.]
23 Specifically, we are seeking FDA approval
24 for the following proposed indication. Zometa is
25 indicated for the treatment of osteolytic,
12
1 osteoblastic and mixed bone metastases of solid
2 tumors and osteolytic lesions of multiple myeloma
3 in conjunction with standard antineoplastic
4 therapy.
5 [Slide.]
6 Three phase III trials form the basis of
7 this supplemental NDA. These trials are the
8 largest randomized studies ever conducted in the
9 treatment of bone metastases. Study 010 is a
10 pivotal, randomized, double-blind, double-dummy
11 study comparing Zometa to pamidronate in patients
12 with multiple myeloma and breast cancer. In this
13 study, patients were treated for thirteen months.
14 Studies 039 and 011 are pivotal,
15 randomized, double-blind, placebo-controlled
16 trials. Study 039 was conducted in patients with
17 prostate cancer over fifteen months. Study 011 was
18 conducted in patients with non-small-cell lung
19 cancer and other solid tumors over nine months.
20 The clinical program was discussed with
21 the FDA and other major health authorities from
22 around the world.
23 [Slide.]
24 The data derived from these three large
25 pivotal studies support the following clinical
13
1 profile for Zometa. Zometa, given at a dose of 4
2 milligrams every three to four weeks is bone
3 specific not tumor specific as Zometa shows
4 effectiveness in a broad variety of different tumor
5 types studied.
6 The clinical trials involved patients with
7 breast cancer and multiple myeloma as well as
8 patients with prostate cancer and other solid
9 tumors. Other bisphosphonates have not
10 demonstrated efficacy in these latter tumor types
11 to date.
12 [Slide.]
13 So Zometa is the first bisphosphonate
14 shown to be effective for the treatment of bone
15 metastases over a wide variety of tumor types.
16 Cumulative safety experience from all trials in the
17 treatment of bone metastases indicates that the
18 safety of Zometa at a dose of 4 milligrams infused
19 over 15 minutes is comparable with that of i.v.
20 pamidronate 90 milligrams, the current standard of
21 care for patients with multiple myeloma and breast
22 cancer.
23 The overall safety profile of Zometa is
24 supported by data from more than 3,000 patients
25 treated to date.
14
1 [Slide.]
2 This morning, we would like to present to
3 you detailed data on the safety and efficacy of
4 Zometa in the treatment of bone metastases. First,
5 Dr. Robert Coleman will give an overview of the
6 pathophysiology of metastatic bone disease and the
7 role of bisphosphonates. Dr. Coleman is Professor
8 of Medical Oncology at the Cancer Research Center
9 at Weston Park Hospital in Sheffield, U.K.
10 Dr. James Berenson will then present the
11 data on Zometa in the treatment of breast cancer
12 and multiple myeloma, study 010. Dr. Berenson is
13 Director of the Multiple Myeloma and Bone
14 Metastases Programs at Cedar Sinai Medical Center
15 in Los Angeles.
16 Dr. Paul Gallo, Assistant Director of
17 Biostatistics, Novartis, will provide also some
18 clarification on the statistical analysis for
19 study 010.
20 Then, after FDA's presentation and
21 discussion by the committee, Dr. Matthew Smith will
22 continue with a discussion of the role of Zometa in
23 prostate cancer. Dr. Smith is Professor of
24 Medicine at Massachusetts General Hospital, Boston,
25 and was an investigator for study 039.
15
1 Dr. Coleman will then return to discuss
2 the role of Zometa in the treatment of solid tumors
3 other than breast and prostate cancer, study 011.
4 Finally, Dr. David Parkinson will present
5 the overall summary and conclusions. Dr. Parkinson
6 is Vice President and Global Head for Clinical
7 Research at Novartis Oncology.
8 [Slide.]
9 In addition to the presenters for today,
10 we also have several clinical experts and advisors
11 with us who are available to help answer specific
12 questions the committee may have. Dr. John Seaman
13 will field responses to the committee's questions
14 and provide background regarding the Zometa
15 Clinical Research and Development Program. Dr.
16 Seaman in the International Team Leader for Zometa
17 at Novartis Oncology.
18 For biostatistical aspects, we have two
19 consultants in attendance, Dr. Richard Cook who is
20 an Associate Professor at the University of
21 Waterloo in Ontario, and Dr. Thomas Fleming, who is
22 Professor and Chair of the Department of
23 Biostatistics at the University of Washington in
24 Seattle.
25 Clinical experts with us today are Dr.
16
1 Pierre Major, medical oncologist at the Hamilton
2 Regional Cancer Center in Ontario and Associate
3 Professor at McMaster University; Dr. Joseph
4 Simeone, Professor of Radiology at Harvard Medical
5 School; and, representing the Renal Advisory Board
6 which has closely monitored the renal safety of
7 Zometa during development, Dr. Raimund Hirschberg,
8 nephrologist and Professor of Medicine at the
9 Harbor UCLA Medical Center in Torrence, California.
10 I would now like to turn the podium over
11 to Dr. Robert Coleman for an overview of the
12 pathophysiology of metastatic bone disease and the
13 role of bisphosphonates.
14 Dr. Coleman, please.
15 Pathophysiology of Metastatic Bone Diseases
16 and the Role of Bisphosphonates
17 DR. COLEMAN: Good morning.
18 [Slide.]
19 Dr. Nerenstone, members of the ODAC panel,
20 ladies and gentlemen. This morning, to provide the
21 background information to today's presentations,
22 there are really four aspects that I would like to
23 get across to the panel over the next twenty to
24 twenty-five minutes.
25 These are the clinical importance and
17
1 consequences of metastatic bone disease. The
2 second is the underlying pathophysiology and some
3 of the similarities that exist across the range of
4 tumor types that affect patients; thirdly, the
5 experience with previous bisphosphonates, notably
6 pamidronate, in the management of metastatic bone
7 disease; fourthly, the background information of
8 zoledronic acid in terms of its pharmacology and
9 the rationale for its dose and schedule in the
10 trials you are going to hear about.
11 [Slide.]
12 Turning first to the clinical importance
13 and prognosis of bone metastases, this slide shows
14 a number of tumors that commonly spread to bone.
15 They are listed in the order that you might
16 associate with the radiographic spectrum of disease
17 that we see on plane X-rays. In other words, at
18 the top is myeloma, typically a very lytic
19 condition. At the bottom is prostate cancer which
20 we associate more with a blastic condition. In
21 between are tumors that have a varied appearance of
22 lytic mixed and blastic.
23 This slide also shows the disease
24 prevalence in the United States and makes the point
25 that, particular for breast cancer and prostate
18
1 cancer, we have an enormous clinical burden to deal
2 with.
3 That is made doubly important when you
4 look at the incidence of bone metastases that
5 typically complicates advanced disease with,
6 perhaps, three-quarters or even four-fifths of
7 patients with breast cancer and prostate cancer
8 developing bone metastases during the course of
9 their illness.
10 The right-hand part of the slide shows the
11 median survival after development of bone
12 metastases and makes the point that, for many of
13 these conditions, particular breast and prostate
14 cancer, the median survival is measurable more in
15 years than in months. So this is a chronic
16 condition requiring long-term palliative therapy.
17 [Slide.]
18 The disease causes a number of very
19 important complications, very important to the
20 patient and very important to our healthcare
21 resources. The complications that we see from bone
22 metastases are shown in this slide and include
23 radiation therapy to bone, pathological fractures,
24 either of long bones or vertebral bodies,
25 hypercalcemia and malignancy, surgery to bone and,
19
1 in some cases, spinal-cord compression.
2 This slide is designed to try and give you
3 a feel for the proportion of patients that
4 experience these events on standard therapy. They
5 are taken from the placebo arms of randomized
6 trials that were assessing bisphosphonate use. But
7 these figures related to the placebo arms of either
8 pamidronate or Zometa trials.
9 I think they show that, across the board
10 of disease, breast, prostate and other tumors, that
11 these complications are common, perhaps three or
12 four occurring per year in a typical patient.
13 About one-third of patients with breast cancer have
14 relatively similar proportions with other diseases
15 and will require radiotherapy and a similar number
16 would experience a pathological fracture.
17 Obviously, some of the other events are
18 less common but sometimes more clinically
19 significant.
20 [Slide.]
21 Having outlined that clinical importance,
22 I want to move on to the pathophysiology. This
23 first slide is a very simplistic view of the
24 relationship between tumor cells and bone cells.
25 But it serves to make the point that osteoclast
20
1 activation--in other words, the acceleration of
2 bone resorption--is very important in the crosstalk
3 between tumor cells and bone.
4 Tumor cells, as most of us will know,
5 reach the target organ through the circulation and
6 are attracted to bone surface by a variety of
7 cytokines and growth factors which are probably
8 released from bone.
9 If the tumor cell possesses the right
10 machinery to produce relevant cytokines and growth
11 factors, it is able to stimulate osteoclast
12 activity, either directly or through bystander
13 cells, to resorb bone. That resorption of bone, as
14 you will see, is responsible for most of the
15 complications.
16 There is also a feedback loop where bone
17 cytokines and bone growth factors may stimulate the
18 proliferation and growth of tumor cells in the
19 biomicro environment.
20 The third mechanism that is illustrated on
21 that slide is is there a direct effect of cancer
22 cells on bone which is, perhaps, independent of the
23 osteoclast. That is an area under research but, to
24 date, it has been extremely difficult to
25 demonstrate any direct destruction of bone by tumor
21
1 cells in either the clinic or in animal-tumor
2 models.
3 So, to the best of our knowledge, by far,
4 the major pathway is through osteoclast activation.
5 [Slide.]
6 How does this pathway differ between what
7 you see on x-rays in osteolytic disease and what
8 you appreciate as an osteoblastic osteosclerotic
9 lesion, typical, perhaps, of prostate cancer of
10 some breast patients.
11 This slide shows that same loop of
12 osteoclast activity, both for lytic and for blastic
13 disease. The osteoclast loop is very similar for
14 both ends of the spectrum. Of course, osteoclast
15 disease is associated with excessive new bone
16 formation and there are growth factors and
17 cytokines produced by prostate cells and other
18 blastic-inducing tumors that stimulate bone
19 formation.
20 But that bone formation is probably not of
21 huge clinical importance. It is not contributing
22 greatly to the structure of the underlying bone.
23 So, across the spectrum, that osteoclast process
24 appears to be very important.
25 [Slide.]
22
1 That disturbance of bone-cell function
2 leads to changes in bone remodeling. Bone
3 remodeling is essential. it is going on in all of
4 us. It is essential to maintain the structural
5 integrity of bone. It is necessary to replace old
6 and fatiguing bone with replacement by new and
7 healthy bone. Usually, that process is coupled and
8 balanced; in other words, areas of bone resorption
9 are repaired in the right quantity and in the right
10 place by the various coupling signals that exist in
11 the bone microenvironment.
12 Cancer disturbs that process in a number
13 of ways. This diagram shows what we associate with
14 osteoclast disease. There are excessive numbers of
15 resorption cavities and the skeleton is unable to
16 repair that damage at a rate that maintains
17 structure integrity in either trabecular or
18 cortical bone. So, gradually, the bone thins and
19 fractures.
20 In mixed lesions, this is the appearance
21 that appears to be happening from histomorphometric
22 studies in that, yes, there is new bone formation
23 but it is in the wrong place and there is still
24 unopposed bone resorption. This process is even
25 more marked in osteosclerotic disease where there
23
1 are excessive amounts of new bone formation. But,
2 again, in the majority of resorption cavities, it
3 is not applied in the right place. It is laid down
4 on creascent bone surfaces and is not really
5 contributing to bone strength.
6 [Slide.]
7 How do we know that? There have been a
8 number of publications, which I don't have time to
9 go into, of histomorphometric studies of bone
10 metastases showing the importance of osteoclast
11 activity across the range of conditions. This
12 slide shows one of many studies that have looked at
13 bone markers, or bone formation and bone
14 resorption, and shows the effect of either lytic or
15 mixed or sclerotic disease on a well-known marker
16 of bone formation, alkaline phosphatase or on a
17 marker of bone resorption, the N-telopeptide, which
18 is a collagen fraction, in this case measured in
19 urine.
20 On the left is the bone formation. Of
21 course, as you would expect, bone formation is
22 massively increased in osteoblastic disease and
23 normal, or even subnormal, in lytic and mixed
24 disease.
25 What is of interest in this slide is that,
24
1 when we look at bone resorption, bone resorption
2 rates are massively increased in osteoblastic
3 disease. Of course, they are increased in lytic
4 disease as well but, if anything, they are even
5 more greatly increased in the osteoblastic
6 patients. So I think that gives you some
7 biochemical evidence for the importance of bone
8 resorption across the range of conditions that
9 might affect bone.
10 [Slide.]
11 As I have hinted, by and large, increased
12 bone resorption is what is responsible for the
13 problems that the patient complains of in the
14 clinic, some aspects of the pain, certainly the
15 fractures and the hypercalcemia that patients may
16 different.
17 [Slide.]
18 So, turning now to treatment. Of course,
19 we all recognize there are many treatments out
20 there for the management of these patients. Most
21 of these treatments are going to remain important
22 for the foreseeable future. Bisphosphonates, I
23 think by all of us, are seen as a complementary
24 approach. In other words, they are usually used in
25 addition to standard therapy, either endocrine or
25
1 chemotherapy as appropriate.
2 But, as you will see, they may reduce the
3 requirements for some other modalities, such as
4 radiation to bone, surgical intervention and,
5 perhaps, some aspects of analgesic use.
6 [Slide.]
7 The bisphosphonates are quite a large
8 class of agents and they have relatively similar
9 pharmacology in terms of their effects on bone
10 cells. This cartoon summarizes the three principal
11 mechanisms that we are aware of. Firstly, all
12 bisphosphonates bind very avidly to calcium, so
13 they bind to the bone surface and the
14 hydroxyapatite and they make it very difficult for
15 the osteoclast to adhere to that bone and resorb
16 it.
17 Secondly, they have direct effects on
18 osteoclast function and activity through
19 biochemical pathways and the indication of
20 apoptosis. Thirdly, at least myobisphosphonates
21 have the ability to affect production and
22 maturation of osteoclasts.
23 [Slide.]
24 So what is the use of bisphosphonates at
25 the moment in the conditions that we are discussing
26
1 today. There is a lot of uncontrolled data but I
2 am going to concentrate on the phase III
3 pamidronate studies in breast cancer, myeloma and
4 prostate cancer that are available to us.
5 These have looked at various endpoints but
6 predominantly at prevention of skeletal-related
7 events. Firstly, looking at breast cancer and
8 myeloma, there are five important studies of note,
9 four in breast cancer and one in myeloma.
10 In terms of the breast-cancer patients,
11 they really fall into two groups. The first were
12 two studies performed in Europe with doses of 45 or
13 60 milligrams of pamidronate--in other words, below
14 the current recommended dosage, which used, as an
15 endpoint, time to progression in bone assessed by
16 radiologists not involved in the study.
17 These studies did show an improvement of
18 time to progression in bone of three to four months
19 but did not show significant effects on skeletal-related
20 events at the doses used. They were then
21 followed by the better-known international trials
22 published by Theriault and colleagues and
23 Hortobagyi and colleagues which led to the
24 registration of pamidronate in this country and
25 worldwide.
27
1 These trials recruited breast-cancer
2 patients with at least one predominantly lytic
3 lesion plus or minus mixed or blastic lesions as
4 typically occurs in our patient population and
5 included either patients on endocrine therapy or
6 patients on chemotherapy at study entry and I think
7 showed quite clearly that pamidronate was able to
8 reduce the frequency of skeletal events, the number
9 of events per unit time and also the time to
10 skeletal events by about 10 to 13 percent.
11 In multiple myeloma, the study from
12 Berenson et al., showed similar results and led to
13 the use of pamidronate worldwide in multiple
14 myeloma again showing a significant reduction in
15 the proportion of skeletal-related events. So I
16 think there is little doubt that pamidronate works
17 in these two conditions.
18 [Slide.]
19 Let's just look at those data in a little
20 more detail, first the two pivotal breast-cancer
21 studies which have been amalgamated here as part of
22 publication from Anna Lipton and colleagues. It is
23 showing the results both at twelve months, which is
24 close to the analysis you will hear for Zometa
25 later, and at 24 months, a later follow-up
28
1 analysis.
2 It shows an 11 percent reduction in the
3 proportion of patients experiencing a skeletal
4 event at twelve months which increases slightly
5 more at 24 months. It is very clear from this
6 slide that the effect is quite marked in terms of
7 effects on radiation to bone and pathological
8 fractures and is maintained, or even increases, as
9 time goes by.
10 [Slide.]
11 This slides shows a similar analysis but
12 for multiple myeloma, again, a short-term analysis
13 at nine months and a follow up analysis at 21
14 months, again showing an improvement in absolute
15 terms of around 17 percent at nine months in terms
16 of the proportion of patients with skeletal events.
17 This effect is maintained out to two years
18 and includes both effects on radiation requirements
19 and fractures.
20 [Slide.]
21 As a clinician, I sometimes find it
22 difficult to assimilate proportions of patients
23 experiencing events and so I include this slide
24 which gives a flavor of the totality of events that
25 occurred in these studies and shows that about 40
29
1 percent of events were abolished by the addition of
2 pamidronate to underlying systemic therapy whether
3 you look at either the breast-cancer protocols or
4 the multiple-myeloma protocol.
5 You can see that it affected important
6 events like requiring surgery, requiring
7 radiotherapy and nonvertebral long-bone fractures.
8 [Slide.]
9 In addition, pamidronate and intravenous
10 bisphosphonates in general can have beneficial
11 effects on pain and analgesic consumption. This
12 slide shows this pain and analgesic results from
13 the three pivotal pamidronate trials. Of course,
14 as patients live with their cancer over a period of
15 two years or so, by and large, they deteriorate.
16 Their performance status deteriorates. They
17 require more analgesics and they have more pain.
18 But what this study shows was that the
19 addition of pamidronate slowed that deterioration
20 and reduced in increase that most patients would
21 experience in analgesic requirements. For multiple
22 myeloma, at least at the nine-month analysis, there
23 was actually a reduction in analgesia requirements
24 and pain compared to the placebo-treated patients.
25 [Slide.]
30
1 What about other tumors? Well, here we
2 have far fewer data with pamidronate and, indeed,
3 any other bisphosphonate prior to Zometa. As I
4 have indicated, there is biochemical and
5 histomorphometric evidence of increased bone
6 resorption with osteosclerotic metastases and there
7 are reports in the literature, somewhat anecdotal,
8 perhaps, but, nevertheless, of useful pain relief
9 from acute high-dose bisphosphonate treatment for
10 sclerotic metastases or other tumors.
11 But no previous randomized trial evidence
12 exists to date that shows a beneficial effect of
13 bisphosphonates on skeletal events.
14 [Slide.]
15 The only study that is really available to
16 present to you today is a trial that was conducted
17 with pamidronate and which has been presented in
18 abstract form which was a relatively short-term
19 study of only six months duration with a primary
20 endpoint, actually, of pain rather than skeletal
21 events. But, as far as this study can show,
22 pamidronate was unable to influence the pattern of
23 skeletal events, the number of patients who
24 experienced a skeletal event or the skeletal
25 morbidity rate.
31
1 So we don't have any evidence in prostate
2 cancer to date that pamidronate or any other
3 bisphosphonate is particularly useful apart from,
4 perhaps, treating pain.
5 [Slide.]
6 What about adverse-event profile of
7 bisphosphonates? In general, these are very well-tolerated
8 compounds compared to many of the things
9 that we use in oncology. Intravenous
10 bisphosphonates are associated with the acute-phase
11 response classically comprised of fever, myalgia,
12 arthralgia. There is an increased incidence of
13 anemia for uncertain reasons with bisphosphonates
14 and occasional mineral disorders such as
15 hypercalcemia and hyperphosphatemia.
16 Very importantly, there are renal effects
17 of bisphosphonates which are seen as a class effect
18 and are very much related to the dose given and
19 particularly the infusion time over which the dose
20 is administered. That is seen with cadrinate*,
21 pamidronate and almost any other intravenous
22 bisphosphonate.
23 [Slide.]
24 How does zoledronic acid differ? Well, it
25 is more potent, at least in the laboratory
32
1 situation, and it is more potent because of this
2 unique structure where, on the site of the PCP
3 backbone, or bone hook, is this imidazole side
4 chain with these two nitrogen atoms which increases
5 its potency above any other bisphosphonate
6 currently in development.
7 [Slide.]
8 There are a number of key preclinical
9 properties of Zometa. First, in vitro, it has been
10 shown to potently inhibit osteoclast formation and
11 bone resorption really regardless of the underlying
12 pathogenic stimulus and, in vivo, is able to
13 inhibit bone resorption in a variety of benign and
14 malignant bone-disease models, again irrespective
15 of tumor types.
16 It does this without deleterious effect on
17 bone structure in that it preserves bone
18 architecture and strength and does not inhibit bone
19 formation.
20 Interestingly and, perhaps, not of direct
21 relevance today, but Zometa also has novel effects
22 of angiogenesis and on pain and neurotransmitter
23 production. Lastly, in a number of animal models
24 where tumor cells have been inoculated into
25 animals, it has been shown that treatment with
33
1 bisphosphonates such as Zometa is able to reduce
2 the number and size of bone metastases and inhibit
3 much of the tumor-induced osteolysis associated
4 with those animal models.
5 [Slide.]
6 In terms of the pharmacology, Zometa is
7 very similar to other bisphosphonates. There is
8 very little protein binding or uptake by red blood
9 cells and no significant interaction with
10 cytochrome P450 metabolizing enzymes.
11 In vivo, there are similar
12 pharmacokinetics in that, after intravenous
13 administration, there is a rapid disappearance of
14 the drug from circulation and the plasma drug
15 concentrations are dose-proportional. Most of an
16 infused dose goes to bone, perhaps about 60
17 percent, and the rest is rapidly eliminated by the
18 kidney over approximately 24 hours.
19 [Slide.]
20 There have been studies of Zometa in
21 patients with renal dysfunction. This slide
22 summarizes the area under the concentration curve
23 for 24 hours of Zometa given on three occasions to
24 three different groups of patients, either with
25 normal renal function, mild impairment or moderate
34
1 renal dysfunction.
2 It shows that, at least down to a
3 creatinine clearance of 30, that although this is
4 associated with a small increase in AUC, it has no
5 effect on urine excretion and the increase in AUC
6 is not affected by repeated dosages. There is no
7 accumulation of the compound with time.
8 On the basis of these studies, there is no
9 indication as, indeed, there is no indication for
10 pamidronate, to dose reduce in renal impairment at
11 least down to clearance of 30 mls per minute.
12 [Slide.]
13 What about the dose and schedule for use
14 in oncology patients? Well, a lot of data has been
15 generated on phase I and phase II trials. This
16 slide summarizes findings from a phase II trial of
17 some 270 or more patients, protocol 007, where
18 Zometa was given in addition to standard therapy to
19 a population of breast cancer and myeloma patients
20 and doses of 4, 2 and 0.4 milligrams were compared
21 to pamidronate.
22 This study showed that, on a three- to
23 four-weekly schedule, that Zometa, at 2 and 4
24 milligrams, was able to produce sustained effects
25 on serum and urinary markers of bone resorption but
35
1 the 4 milligram dose was the most effective dose
2 tested. Skeletal events and pathological fractures
3 were reduced most by the 4 milligram dose of Zometa
4 and the 0.4 milligram dose was clearly ineffective
5 compared to the other Zometa doses or, indeed,
6 pamidronate.
7 Lastly, the time to first skeletal event
8 in the breast-cancer patients was almost two months
9 longer in the 4 milligram arm versus the 2
10 milligram dose although, being a phase II study, of
11 course it was not powered to show this as a
12 significant difference.
13 [Slide.]
14 Exploring the schedule and dose in a
15 little more detail using bone markers, this slide
16 illustrates a couple of key points. Firstly, this
17 is taken from protocol 007. It shows the
18 biochemical profiles of the three Zometa dosages,
19 0.4, 2 and 4 milligrams, and pamidronate 90 and
20 shows that there is a rapid inhibition of bone
21 resorption which reaches a maximum at around a week
22 after infusion.
23 But the red line, the 4 milligram Zometa
24 group, shows a more sustained effect with
25 persisting complete inhibition of bone resorption
36
1 at four weeks whereas pamidronate and the other
2 dosages are beginning to wear off by that time
3 point.
4 The right-hand side of the slide shows the
5 effects of chronic dosing where that effect of
6 inhibiting bone resorption is maintained and,
7 again, is more marked at the 4 milligram dosage
8 than it is at other dosages tested in that study
9 or, again, more marked than pamidronate.
10 [Slide.]
11 You have already heard in the introductory
12 talk that Zometa is already licensed for the
13 treatment of hypercalcemia of malignancy. This
14 slide just reminds us of the data that led to that
15 approval. This was a study in patients with
16 moderate to severe hypercalcemia of malignancy and
17 compared pamidronate 90 milligrams with Zometa at 4
18 or 8 milligrams and showed quite clearly that
19 Zometa was superior with nearly 90 percent of
20 patients achieving normocalcemia by ten days which
21 was the primary endpoint of the study, compared to
22 70 percent with pamidronate.
23 In addition, Zometa worked more quickly
24 and, of interest, there was no difference between
25 the 4 and 8 milligram dosages in this trial.
37
1 Clearly, these data were not available when the
2 phase III studies were generated and the protocols
3 designed. But it suggests that Zometa is superior
4 to pamidronate but there is no dose response
5 between 4 and 8 milligrams.
6 [Slide.]
7 So, to conclude my presentation, I think
8 we can all agree that metastatic bone disease is an
9 important healthcare problem in oncology and there
10 are many unmet needs for our patients in terms of
11 skeletal complications. Although the underlying
12 tumor biology is very different at the primary
13 site, when the disease gets to bone, the
14 pathophysiology is very similar in that it is
15 mediated through osteoclasts and, as far as we can
16 tell, osteoclast activation accompanies all bone
17 metastases with or without bone formation.
18 The currently available bisphosphonates
19 have a limited range of activity which I think we
20 have to say is, at the present, confined to breast
21 cancer and myeloma. Zometa is a more potent
22 inhibitor of osteoclast activity which, we believe,
23 provides a bone-specific treatment which is
24 applicable across a range of tumor types.
25 I thank you for your attention and would
38
1 now like to turn the podium over to Dr. Jim
2 Berenson who is going to present the first of the
3 three phase III trials, protocol 010.
4 Pathophysiology of Metastatic Bone Diseases
5 and the Role of Bisphosphonates
6 DR. BERENSON: Good morning all. Good
7 morning Dr. Nerenstone, members of the ODAC, those
8 of you who are on teleconference and
9 videoconference this morning.
10 [Slide.]
11 It is my pleasure to present to you the
12 first of three randomized double-blind studies
13 evaluating Zometa at two doses. This first study
14 involves patients with breast cancer metastatic to
15 bone and multiple myeloma with at least one lytic
16 lesion. Unlike the other studies, in this study,
17 the comparator is pamidronate since, as you heard
18 from Dr. Coleman, this has shown to be effective in
19 reducing skeletal complications in these types of
20 patients.
21 [Slide.]
22 This was a double-blind, double-dummy,
23 study involving 1648 patients who were stratified
24 prior to assignment to which drug based on *Duriei-Salmon
25 stage III patients with at least one lytic
39
1 lesion with myeloma and then breast-cancer patients
2 either on hormonal therapy or chemotherapy. This
3 patients had to have stage IV breast with at least
4 one either lytic, blastic or mixed lesion.
5 They could be on appropriate
6 antineoplastic therapy at baseline and this could
7 be changed at the discretion of the treating
8 physician during the trial. ECOG performance
9 status could be 0, 1 or 2 and the serum creatinine
10 at the time of study entry had to be less than or
11 equal to 3 milligrams per deciliter.
12 [Slide.]
13 The study design is shown here. This was
14 a 12-month dosing study. Pamidronate was given as
15 90 milligrams every three to four weeks over two
16 hours and then Zometa as a 4-milligram and
17 initially 8-milligram, and you will see why some of
18 those patients were changed to 4 milligrams, every
19 three to four weeks initially over five minutes and
20 amended to increase the infusion time to fifteen
21 minutes, and you will see the reasons for that
22 momentarily.
23 The duration of the study was one month
24 beyond the dosing regimen; that is, thirteen
25 months. Patients during the trial also received
40
1 oral vitamin D and calcium daily. This was to
2 reduce the incidence of hypocalcemia which can
3 occur with bisphosphonate use and, indeed, in this
4 study, those patients receiving pamidronate had a
5 lower incidence of hypocalcemia than that that had
6 been observed in the trials previously presented by
7 Dr. Coleman.
8 A second reason for using these two
9 medications in these patients was to reduce the
10 potential for microfractures which can occur in
11 animals as a result of bisphosphonate use with
12 increases in parathyroid hormone. This has not
13 been observed clinically, but it is a potential
14 reason to use vitamin D and calcium as well.
15 [Slide.]
16 The primary objective of the trial is
17 shown here. It was to demonstrate the efficacy of
18 Zometa through the noninferiority comparison to
19 pamidronate for the treatment of bone metastases.
20 In this trial, a margin of 8 percent with two-sided, 95-
21 percent, confidence intervals was used
22 based on the results of the prior trials that Dr.
23 Coleman presented comparing pamidronate to placebo
24 in breast-cancer patients and myeloma patients with
25 lytic bone disease.
41
1 We also wanted to demonstrate that the
2 safety profile of this new bisphosphonate given
3 chronically was comparable to chronic-used
4 pamidronate.
5 [Slide.]
6 The primary study endpoint is shown here.
7 It was to determine the proportion or the percent
8 of patients experiencing at least one skeletal-related event
9 not counting hypercalcemia of
10 malignancy as an event for this primary endpoint.
11 [Slide.]
12 A number of secondary endpoints were also
13 analyzed including time to first skeletal-related
14 event, the skeletal morbidity rate, and Andersen-Gill
15 multiple-event analysis which I will describe
16 momentarily. In addition, the time to first
17 skeletal-related event, skeletal morbidity rate and
18 the number of patients experiencing any skeletal
19 event was also determined, this time counting
20 hypercalcemia of malignancy as a skeletal event.
21 The pain and analgesic scores were
22 analyzed, a bone-lesion response as well. The
23 time to the progression of the patient's overall
24 disease as well as progression of their bony
25 metastasis. Then safety was analyzed including an
42
1 analysis of survival. Importantly, beyond the
2 thirteen months of the trial, an additional six
3 months of safety data is available and will be
4 presented for overall survival as well as changes
5 in serum creatinine in these patients.
6 [Slide.]
7 The definition of a skeletal-related event
8 is similar to that that was employed in the prior
9 trials comparing pamidronate to placebo in breast
10 and myeloma defined as the development of any new
11 pathological fracture, the development of spinal-cord
12 compression, the requirement for radiation
13 therapy for either bone pain or to treat actual or
14 impending pathological fractures or spinal-cord
15 compression, surgery to bone and, as mentioned
16 previously in some of the analyses, hypercalcemia
17 of malignancy of defined as an event.
18 [Slide.]
19 The preplanned analysis from this trial
20 was to determine, first of all, the proportion or
21 the percent of patients with at least one skeletal
22 event. This was defined as the number of patients
23 with at least one skeletal event divided by the
24 number of patients in that treatment group. The
25 time to first skeletal-related event was defined as
43
1 the time the patient was randomized to the
2 development of the first skeletal event and this
3 was determined in days.
4 The skeletal morbidity rate looks at the
5 number of events over time; that is, the number of
6 skeletal events divided by the time the patient is
7 on trial determined in years. The Anderson-Gill
8 method allows one to analyze multiple events over
9 time and it uses a more general model and takes
10 into account not only the time of the number of
11 events in a given period of time but the time
12 between events as well and, thus, is a more robust
13 analysis of multiple events over time in a given
14 patient.
15 [Slide.]
16 The history of the trial is shown here.
17 The initial design was to employ two doses of
18 Zometa, either 4 or 8 milligrams, initially given
19 over five minutes versus standard-dose pamidronate
20 as a 90-milligram infusion given over two hours
21 every three to four weeks for twelve months. In
22 June of 1999, because of concerns of changes in
23 creatinine in patients receiving Zometa, the
24 infusion time for Zometa was increased from five to
25 fifteen minutes.
44
1 In order to assure that was occurring, the
2 infusion volume was also increased from 50
3 milliliters to 100 milliliters. The following
4 year, in June of 2000, a second renal amendment was
5 made because of continuing concerns in the 8-milligram dose
6 of changes in creatinine, so all
7 those patients subsequently received 4 milligrams
8 and subsequently will be known as the 8/4 milligram
9 group.
10 In addition, renal-function monitoring was
11 begun so that, prior to each dose of Zometa within
12 a two-week period of time, the serum creatinine was
13 checked. A statistical amendment was also made
14 because of the mix in the 8-milligram dose group so
15 that the primary efficacy analysis was based on
16 comparing Zometa at 4 milligrams versus pamidronate
17 at 90 milligrams.
18 [Slide.]
19 The trial started in, as you see, the fall
20 of 1998 and concluded approximately one year ago.
21 Approximately half of the 1648 patients were
22 accrued prior to the change in the infusion time
23 from five to fifteen minutes and, indeed, nearly
24 half of these patients only received 8 milligrams
25 during the entire trial in the 8-milligram dosing
45
1 group.
2 The rest of the patients were accrued
3 during the subsequent six-month period of time,
4 about half of the patients. As you see, in June of
5 2000, a second renal amendment was made to change
6 8-milligram dosing patients to only receive 4. In
7 fact, those patients who were actually entered
8 after renal amendment 1, about a quarter of those
9 patients only received the 8-milligram dose.
10 The rest of the trial was completed so
11 that the last study visit occurred approximately
12 one year ago.
13 [Slide.]
14 The demographics and prognostic factors
15 among the approximately 1650 patients on this trial
16 are shown here. One can see that the mean age was
17 in the upper 50s. As you would guess, given that
18 this involved a lot of breast-cancer patients,
19 about 80 percent of the patients entered were
20 female.
21 About 85 percent were Caucasian. The
22 performance status in most cases was 0 or 1. There
23 was an approximately equal distribution of patients
24 who had breast cancer with metastatic bone disease
25 who were on chemotherapy, hormonal therapy as well
46
1 as those patients with myeloma who had at least one
2 lytic lesion. So, similar numbers of patients and
3 similar numbers in each arm.
4 [Slide.]
5 As one can see here, the number of
6 patients who were able to complete the study
7 therapy through twelve months of treatment and
8 additional months of follow up was about 60 percent
9 and there was an equal distribution in those
10 receiving either 4 or 8/4 Zometa versus pamidronate
11 at 90 milligrams.
12 [Slide.]
13 The reasons for early discontinuation
14 amongst approximately the 40 percent of patients
15 who did not complete the study, as you would
16 probably guess, death occurring in about 10 to 11
17 percent of patients and a similar proportion
18 throughout all three arms.
19 Adverse events also occurring with a
20 similar proportion amongst patients on Zometa 4,
21 8/4 or pamidronate at 90. Withdrawal of consent, a
22 similar proportion. And the other causes for early
23 discontinuation, again, nothing stands out. Small
24 numbers and not much difference between the arms.
25 [Slide.]
47
1 The primary endpoint results are shown
2 here. We are looking at the percentage of patients
3 who had at least one skeletal event by thirteen
4 months; that is, twelve months of treatment, an
5 additional month of follow up. As one can see from
6 the data, the percentage of patients having at
7 least one skeletal event by this time point is
8 quite similar across all three arms approximating
9 45 percent.
10 The red is Zometa at 4 milligrams, Zometa
11 8/4 is the blue, and the purple, here, is the
12 pamidronate arm. Importantly, as one can also see,
13 the Zometa at 4, and well remembering many in the
14 8/4 group only received 8, the percentage of
15 patients having an event is quite similar in both
16 of these two arms.
17 Now, comparison of the data shows that,
18 indeed, the primary endpoint comparing Zometa at 4
19 versus pamidronate at 90, within the 95 percent
20 confidence interval, remembering what told I said
21 earlier, we are looking for an 8 percent margin, we
22 can see here that Zometa may be as much as 7.9
23 percent better than pamidronate and as worse as 3.7
24 percent inferior. Indeed, this is well within the
25 8 percent mark that we looked at as our primary
48
1 endpoint in the trial.
2 Now, in order to clarify these issues
3 better in terms of interpreting a noninferiority
4 trial, from the statistical perspective, I now want
5 to bring up Dr. Paul Gallo from Novartis
6 Biostatistics.
7 DR. GALLO: Good morning.
8 [Slide.]
9 I would like to very briefly discuss some
10 issues in noninferiority trials which is a very
11 complex and quickly evolving area and their
12 implications for the interpretation of the data
13 that you have just seen.
14 A basic rationale for a noninferiority
15 trial is that if we can show sufficient
16 comparability between a new treatment and a
17 standard treatment and use historical information
18 that demonstrated how superior that standard is
19 relative to placebo, then we can infer that the new
20 treatment would have beaten placebo had there been
21 a placebo arm in the current trial.
22 In this case, the relevant historical
23 information comes from the pamidronate registration
24 trials 218 and 19. I will point to this screen to
25 the extent that I am able. I am going to summarize
49
1 an analysis that was done by the FDA statistical
2 reviewer and was included in the briefing document.
3 In designing study 010, we had actually
4 used a slightly different analysis, but the results
5 are not practically different so I will just
6 illustrate with this analysis to try to keep things
7 a little bit simpler.
8 In a combined analysis of data from these
9 trials, we can see that pamidronate had 13 percent
10 fewer events than placebo and the lower limit for
11 the magnitude of benefit was 7.3 percent. By
12 recent conventional practice if, in the current
13 study, we can exclude the possibility that Zometa
14 is 7.3 percent worse than pamidronate, then we can
15 claim that Zometa is effective. The number, 7.3,
16 would be called the noninferiority margin. It is
17 the potential disadvantage of Zometa that we have
18 to disprove.
19 Again, we had done a slightly different
20 analysis that led to the 8 percent margin in the
21 protocol that was previously mentioned by Dr.
22 Berenson.
23 This is a graphical illustration of what I
24 have just described, an estimated advantage for
25 pamidronate relative to placebo of 13.1 percent and
50
1 its confidence limits.
2 Recently, FDA personnel have proposed that
3 it may be appropriately conservative to consider
4 using an even smaller margin, in particular, half
5 of the lower confidence limit. The rationale is
6 that the current practice makes a constancy
7 assumption, namely that the standard is equally
8 effective in the current trial as it was in the
9 historical trials. In practice, we can't guarantee
10 that there are not subtle differences between the
11 historical and current trials so that the standard
12 possibly might not be as effective in the current
13 trial.
14 Achieving a stricter noninferiority
15 criterion would provide more assurance that an
16 effect is real, even allowing for some violation of
17 the constancy assumption. In our current case, the
18 conservative criterion would be a margin equal to
19 half of 7.3 percent, or 3.65 percent.
20 Now we look at the protocol 010 data,
21 specifically the primary 4-milligram dose results
22 and tie it together with the historical data by
23 lining up the pamidronate effect. We see a 2
24 percent difference advantage for Zometa with a
25 confidence limit that, in a worse case, goes as far
51
1 as a 3.7 percent advantage for pamidronate.
2 We can see that the conventional standard
3 of excluding 7.3 percent is safely reached. In
4 fact, these results just about achieve the proposed
5 conservative standard even though the trial was not
6 designed to be powered for such a standard which
7 was really not in existence at that time.
8 So this provides somewhat of a higher
9 level of comfort that the Zometa effect is real
10 even allowing for the possibility that, for some
11 reason, pamidronate might not be quite as effective
12 here as it was in the historical trials.
13 So, to summarize, that is the evidence for
14 the effectiveness of Zometa in this trial. We
15 estimate from the data that Zometa is just a bit
16 better than pamidronate and we know, from
17 historical trials, that pamidronate seems to be
18 quite a bit better than placebo and, in a worse
19 case, both for Zometa and for pamidronate, the
20 confidence intervals don't overlap.
21 Allowing for some violation of the
22 constancy assumption by using the newer
23 conservative standard, they still essentially don't
24 overlap.
25 Now I will turn things back to Dr.
52
1 Berenson to continue with the presentation of the
2 trial results.
3 DR. BERENSON: Thank you.
4 [Slide.]
5 I will continue to present the results
6 from the trial. As you see here, we are looking at
7 the specific types of skeletal events that occur
8 during the trial and the proportions, actually, are
9 quite consistent to what we observed from our
10 earlier trials comparing pamidronate to placebo.
11 Indeed, fracture and radiotherapy to bone
12 are the most frequent events. As you see here,
13 looking at all four types of events, the proportion
14 of patients having an event are quite similar in
15 the pamidronate, the Zometa 4 and 8/4 group with
16 the exception there are a smaller proportion of
17 patients who receive Zometa at 4 milligrams who are
18 actually experiencing the requirement for
19 radiotherapy to bone but the 8/4 group looks a lot
20 like pamidronate here.
21 [Slide.]
22 Indeed, the time to first skeletal-related
23 event, a secondary endpoint in the trial, as you
24 can see, these three curves are quite
25 superimposable upon one another; that is, the time
53
1 to first event here is approximately one year in
2 the Zometa 4, the 8/4 as well as the pamidronate
3 arm and, as you see, the hazard ratios here are
4 slightly less than 1, but do cross 1.
5 [Slide.]
6 When one looks at the number of skeletal
7 events per year, the MSR, or skeletal morbidity
8 rate, the red, again, is Zometa. The blue is
9 Zometa 8/4 and the purple is pamidronate 90. The
10 number of events per year is about 1 in the two
11 Zometa groups, a little higher with pamidronate at
12 1.4 but this is not significantly different, again
13 quite equivalent across these three arms in terms
14 of number of events per year.
15 [Slide.]
16 Now, as I mentioned earlier, another type
17 of analysis was done to look at the number of
18 events per year, not just time to first event or
19 proportional analysis of a single event. This
20 analyzes the type to each count in skeletal-related
21 event and, therefore, takes into account the
22 interval between subsequent skeletal-related events
23 in addition to just the time to the first event or
24 the number of events over time as the SMR has done.
25 So we believe this is a more robust
54
1 analysis. As you can see from the hazard ratios
2 here, which are less than 1, actually, the relative
3 reduction or benefit in Zometa is in the range of 9
4 to 11 percent here in terms of the Andersen-Gill
5 analysis. But, again, these do cross 1 but
6 certainly suggest equivalence of pamidronate to
7 Zometa not only in terms of time to first event and
8 number of events per year, but multiple events over
9 time as well.
10 [Slide.]
11 This slide is actually not in the handout
12 for those of you who are off-site, Drs. Taylor,
13 Albain and Kelsen, but we wanted to include it.
14 This is an exploratory analysis to determine the
15 proportion of breast-cancer patients who had an SRE
16 based on the type of bone lesion they had at the
17 time of study entry.
18 So those patients who had lytic bone
19 disease had at least one lytic lesion. The blastic
20 group only contained blastic lesions and those that
21 could not be placed in either group were considered
22 other.
23 As you see, in the lytic group, the red
24 bar, again, Zometa at 4, and the blue bar, Zometa
25 8/4, the number of patients having at least one
55
1 event is actually less than in pamidronate. In the
2 middle of the other group, there is a trend in the
3 other direction. The pamidronate is slightly less
4 than the other two groups.
5 On the right side, the blastic group,
6 although the Zometa 4 is lower, the Zometa 8/4
7 looks quite similar to pamidronate. So I think it
8 is very difficult to conclude from this single
9 slide whether the drug works, as you well could
10 guess, in blastic disease because the comparator
11 here is pamidronate and we have not previously
12 considered pamidronate in patients with blastic
13 disease from the trials in the mid-90's. I will
14 leave that for subsequent discussion, but I did
15 want to show that slide.
16 [Slide.]
17 Looking at disease and quality-of-life-related
18 endpoints, I will just summarize and I will
19 show you some of the data. That was very
20 comparable disease and quality-of-life-related
21 measures and changes that were demonstrated in the
22 two Zometa arms compared to the pamidronate arm;
23 that is, the time to progression of bone
24 metastasis, the time to progression of any type of
25 malignant process going on in the patients, the
56
1 pain analgesic score changes during the trial,
2 changes in both ECOG performance status and the
3 FACT-G global quality-of-life were quite similar
4 across these three arms.
5 [Slide.]
6 Now, first looking at the data for
7 disease-related endpoints here, in the top part
8 here, we are looking at the time to the progression
9 of the bone disease measured in days. You can see,
10 it is quite similar across all three arms
11 approximating six months. Now, secondly, looking
12 at the time to the progression of the patient's
13 malignancy, again, the numbers are quite similar
14 although slightly shorter for pamidronate but
15 reassuredly show in terms of the two Zometa arms.
16 The Zometa, it looks quite similar. It
17 certainly doesn't look worse than pamidronate in
18 terms of the progression of the patient's
19 underlying malignancy.
20 [Slide.]
21 Here we are showing the data for quality-of-life
22 endpoints and I will summarize by telling
23 you that brief pain inventory score changes,
24 analgesic scores, ECOG performance changes, FACT-G
25 total-score changes and, again, here we are looking
57
1 at the change from the baseline to either end of
2 study or last measurement, were not significantly
3 different from Zometa at 4, Zometa at 8/4, compared
4 to pamidronate although those patients, over time,
5 in terms of pain, did improve pain within each arm,
6 just not different between the arms.
7 [Slide.]
8 So here is a summary of the efficacy data
9 from the trial. On the left side, we are looking,
10 again, at the percent of patients who had at least
11 one skeletal event by thirteen months, twelve
12 months of therapy, one month of additional follow
13 up, and the numbers are strikingly quite similar,
14 44, 46 and 46.
15 The time to first skeletal-related event,
16 hazard ratio less than 1, crosses 1, again
17 consistent with equivalence of this drug in terms
18 of efficacy to standard-dose pamidronate. The mean
19 SMR--as I mentioned earlier, the SMRs were around 1
20 for the two Zometa arms, slightly higher in the
21 pamidronate arm, 1.4, but not significantly
22 different.
23 Looking at multiple events over time in a
24 little different matter, as I mentioned earlier, by
25 Andersen-Gill analysis, the hazard ratios, in fact,
58
1 are less than 1, again cross 1. So, overall, the
2 results of this trial in terms of efficacy
3 certainly suggest comparable efficacy of this new
4 drug to pamidronate in reducing skeletal
5 complications in these patients.
6 [Slide.]
7 Now let's turn our attention to the safety
8 data from this trial. The primary cause of death
9 during the trial or within one month after study
10 drug termination is shown here. As you could
11 guess, the most common cause, of course, would be
12 the underlying malignancy in approximately 8 to 10
13 percent of patients throughout all three of these
14 arms.
15 Other causes which were less common,
16 including respiratory, cardiac and infectious
17 causes, certainly are similar across all three arms
18 and the other less common causes, as you can see,
19 the numbers quite small and quite similar across
20 three arms, very reassuring.
21 [Slide.]
22 Now, overall survival was obviously looked
23 at as well. As you can see, in this trial, in
24 those patients again with breast cancer metastatic
25 to bone and myeloma with at least one lytic lesion,
59
1 the survival was approximately 800 days in all
2 three arms, strikingly similar in the Zometa 4, 8/4
3 and pamidronate arms with p-values that are very
4 high and much above 0.05.
5 [Slide.]
6 The incidence of adverse events that
7 occurred in at least 15 percent of patients in one
8 of the arms is shown here regardless of whether we
9 believed it was related to the study drug. You see
10 a lot of white here indicating that there were very
11 similar proportions of patients in all three arms
12 having these events with the exception of the
13 yellow, pyrexia, a known side effect with the first
14 or second administration of bisphosphonates, about
15 5 percent more common in the Zometa arms and,
16 indeed, anorexia, really for unknown reasons, a
17 little bit more than 5 percent than the pamidronate
18 in the 4-milligram arm. But there are a lot, as
19 you see, of adverse events we looked at.
20 [Slide.]
21 Now, importantly, the incidence of anemia
22 has been shown to be higher with chronic
23 administration of pamidronate in the previously
24 done trials comparing pamidronate to placebo in
25 breast-cancer on chemotherapy and myeloma patients
60
1 with at least one lytic lesion.
2 Although, there was some incidence of
3 anemia, it was less than 6 percent in all treatment
4 groups and was not different between the treatment
5 groups as well. Of probably more importance, the
6 use of packed red blood-cell transfusions and
7 exogenous erythropoietin support was not different
8 between the treatment arms as well.
9 Electrolyte and mineral adverse events,
10 mentioned previously in Dr. Coleman's talk which
11 occurs with pamidronate occasionally, this
12 incidence was quite uncommon in all treatment
13 groups. The most common events with the Zometa at
14 the 4-milligram dose were hypophosphatemia, low
15 potassium or high potassium, but these were
16 uncommon.
17 [Slide.]
18 Now, the definition of serum creatinine
19 used to monitor these patients following renal
20 amendment 2 was a serum creatinine that was normal
21 at baseline and then increased by 0.5 milligrams
22 per deciliter. A serum creatinine that was
23 abnormal at baseline and increased by more than 1
24 milligram per deciliter or a doubling of serum
25 creatinine from the number that was available at
61
1 study entry.
2 If the serum creatinine was increased by
3 any of these three determinations, it was decided
4 to hold the dose until it returned to 10 percent of
5 the baseline serum level.
6 [Slide.]
7 This is a rather complex slide but an
8 important one. Turning our attention, first of
9 all, to the top half of the slide, this is the
10 number of patients who had no increase of serum
11 creatinine and, again, the definitions I showed you
12 in the previous slide, before we changed the
13 infusion time from five to fifteen minutes.
14 The purple line represents patients on
15 pamidronate. The two Zometa curves in red and
16 light blue, as you can see, are lower lines
17 suggesting that, indeed, before the amendment
18 change was made, a slowing of infusion time, there
19 was an increased risk of creatinines going up in
20 patients on Zometa in the 4 or 8/4 group.
21 Following the amendment change, shown in
22 the bottom half of the slide, and, again, about
23 half of the patients accrued at that time, one can
24 see a disappearance of the increased risk of renal
25 problems or creatinine rises in those patients who
62
1 were receiving Zometa at 4 over fifteen minutes;
2 that is, the dark red and the purple curves are
3 superimposable, hazard ratio there at 1.012.
4 But we continue to observe problems in the
5 8-milligram dose when given over fifteen minutes in
6 the light blue with a hazard ratio over 2 and a
7 quite significant p-value suggesting continued
8 problems. Importantly, as I mentioned earlier,
9 nearly half of the patients who were on the 8-milligram dose
10 prior to the fifteen-minute infusion
11 time change only received that dose, about 21
12 percent of patients following that change.
13 [Slide.]
14 Now, let's look at the data globally in
15 terms of significant NCI grade 3 and 4 serum
16 creatinine changes. After the fifteen-minute
17 amendment was changed, only those patients
18 enrolled, again that was about half of the 1648,
19 803, and one can see grade 3 problems only in one
20 patient at Zometa at 4, no grade 4s, and, as you
21 can see, slightly more in the Zometa 8/4s at six
22 grade 3s and one grade 4, again, necessitating the
23 change in those patients from 8 to 4.
24 In the pamidronate, in fact, there were
25 more events here of grade 3 and grade 4 than Zometa
63
1 at 4.
2 [Slide.]
3 So the safety summary that I have just
4 outlined tells that when you give Zometa at a 4-milligram
5 i.v. dose over fifteen minutes every
6 three to four weeks, the safety profile, including
7 changes in creatinine, is quite comparable to
8 intravenously administered pamidronate at
9 90 milligrams over two hours.
10 [Slide.]
11 Indeed, an overall summary of this trial
12 would tell us that Zometa at 4 milligrams over
13 fifteen minutes given every three to four weeks
14 demonstrates comparable safety and efficacy as
15 noninferiority design in this trial to standard
16 dose pamidronate at 90 milligrams over 120 minutes
17 in treating bone metastasis of all types in
18 patients in breast cancer, in lytic bone disease,
19 in multiple myeloma.
20 DR. NERENSTONE: Thank you very much. We
21 are going to go right to the FDA presentation for
22 review of the study 010. Then we are going to
23 leave the clarifying questions to both the sponsor
24 and the FDA until after the FDA presentation.
25 FDA Presentation
64
1 Zometa in Breast Cancer and Multiple Myeloma
2 (Study 010)
3 DR. WILLIAMS: Dr. Nerenstone, members of
4 ODAC, it is my pleasure to begin the FDA
5 presentation of this application.
6 [Slide.]
7 I want to break from protocol a little.
8 You are welcome to interrupt me if I am on wrong
9 slide. Otherwise, we will wait until the end.
10 [Slide.]
11 First, I would like to recognize the FDA
12 review team. The FDA clinical review team worked
13 hard to review this large complicated application.
14 This slide recognizes the members of that team.
15 Debra Vause led us as project manager. Five
16 medical and statistical reviewers helped prepare
17 the information for this meeting. Medical
18 reviewers included myself for study 010, Dr. Amna
19 Ibrahim for studies 011 and 039, Dr. Nancy Scher
20 for the safety data from these studies and Dr.
21 Richard Pazdur, our Division Director.
22 The statistical review of study 010 was
23 performed by Dr. Ning Li for study 010. For 011
24 and 039 was Dr. Raji Sridhara. And the statistical
25 team leader is Dr. Gang Chen. All of the reviewers
65
1 will be available for questions and discussion.
2 [Slide.]
3 This is the outline of my presentation.
4 First, I will present an overview and then a
5 regulatory framework followed by discussion of the
6 FDA findings from study 010 and myeloma and breast
7 cancer. The latter will include a discussion of
8 the noninferiority trial design, results and
9 examination of assumptions.
10 [Slide.]
11 As a reminder, this is the proposed Zometa
12 indication; treatment of osteolytic, osteoblastic
13 and mixed bone metastases of solid tumors and
14 osteolytic of multiple myeloma. These are in
15 conjunction with standard antineoplastic therapy.
16 [Slide.]
17 The FDA reviews the Zometa from a
18 regulatory perspective as well as from a scientific
19 perspective. The FDA requirement to demonstrate
20 efficacy dates to a 1962 amendment to the Federal
21 Food, Drug and Cosmetic Act which required
22 substantial evidence of efficacy from adequate and
23 well-controlled investigations.
24 Usually, this means evidence from multiple
25 clinical trials but very impressive and robust
66
1 results from a single multicenter trial has
2 sometimes been accepted. An important question you
3 will be considering during your deliberations will
4 be whether these trials meet the regulatory
5 efficacy requirement.
6 You will be asked this for each of the
7 different indications that correspond to the
8 different trials. You might find that one of the
9 trials is so impressive that it supports approval
10 withput any support from the other trials or you
11 may find that, while a single trial is not
12 convincing enough to stand alone, you find the
13 results of one of the other trials to be supportive
14 also.
15 It is a matter of both science and of
16 judgment whether the results of the different
17 trials support each other. This is a fitting topic
18 for ODAC and we invite your advice.
19 [Slide.]
20 At the planning stage, both FDA and
21 Novartis assumed that some sharing of information
22 between tumor types was reasonable. For example,
23 breast cancer and myeloma were combined in a single
24 study. For each indication, only single studies
25 were planned. Finally, multiple different solid-tumor types
67
1 were lumped together in the single
2 study 011.
3 Of course, these general assumptions were
4 made without having the results in hand. The
5 questions before ODAC concern specifics. Now, with
6 the result in hand, there may be unanticipated
7 questions. This often seems to happen in clinical
8 trials in the real world. So we need your advice
9 whether these assumptions are still reasonable in
10 light of the clinical-trial data.
11 [Slide.]
12 In the next two slides, I want to discuss
13 the skeletal-related event analyses used in these
14 studies. Evaluating efficacy by measuring cancer-patient
15 morbidity is difficult and there is no
16 perfect endpoint. Patients suffering from cancer
17 often drop out early from studies either from
18 death, side effects of treatment of from other
19 cancer problems not captured by the endpoint.
20 Historically, in evaluating bisphosphonate
21 treatment of cancer metastatic to bone, FDA has
22 emphasized two very closely related endpoints;
23 first, the proportion of patients entering the
24 study who have a documented event and, second, the
25 time to skeletal-related event.
68
1 In studies of bisphosphonate, the results
2 from these two analyses usually go hand in hand and
3 this should not be surprising. Both use exactly
4 the same skeletal events. The first analysis
5 describes the number of patients with at least one
6 event and the second uses the additional
7 information of event timing.
8 During FDA analysis of these studies,
9 questions arose regarding which of the two
10 different endpoints was preferred.
11 [Slide.]
12 However, both endpoints share serious
13 problems. That problem is dropouts. Some patients
14 who drop out may subsequently have an event and
15 this is not counted. This would lead to an
16 underestimation of the event frequency in the
17 proportions analysis.
18 Furthermore, patients may drop out because
19 of symptoms of an impending event and that event is
20 not documented as such. This would lead to a
21 phenomenon known as informative censoring,
22 something that would violate assumptions of the
23 time-to-event analysis. This analysis assumes that
24 censoring occurs in a random manner.
25 These analyses share the problem of
69
1 competing risks, simultaneous risk of cancer-related
2 problems such as death and the risk of
3 having a skeletal event. With both endpoints,
4 there is likely to be bias in estimating the true
5 effect of the drug. This bias could affect how
6 accurately we describe the benefit of the drug but
7 this bias should apply equally and in the same
8 direction to both arms.
9 Because this study is blinded, neither
10 analysis should introduce bias between the study
11 arms. So a significant p-value for either endpoint
12 might be viewed as valid evidence that efficacy has
13 been demonstrated. I am sure that ODAC's Dr.
14 George and the FDA statisticians will be glad to
15 discuss this point further during this afternoon's
16 discussion.
17 So I believe we should view these two
18 closely-related endpoints with a bit more
19 flexibility than we might otherwise in evaluating
20 primary and secondary endpoints. Novartis'
21 prespecified endpoint was the proportion of
22 patients having an event. The FDA's statisticians
23 preferred analysis, this time to SRE.
24 As you have observed in these trials,
25 results were sometimes statistically significant
70
1 for one endpoint but not for the other.
2 [Slide.]
3 Each of the clinical trials raised one or
4 more important questions which I will summarize in
5 the following three slides. For study 010 in
6 breast cancer and myeloma, we note that it was a
7 single study of noninferiority design. Due to the
8 inherent weakness of noninferiority studies, we
9 usually expect results from two such studies or
10 additional evidence from studies of a different
11 design.
12 The question for ODAC, do the totality of
13 the data in the NDA provide support for this
14 indication. We believe they do and we invite your
15 comments.
16 [Slide.]
17 In prostate cancer, the 4-milligram Zometa
18 arm shows convincing results for both the primary
19 and secondary event endpoints. The 8-milligram arm
20 shows no statistical difference from placebo. Two
21 questions arise; first, considering both arms of
22 the study, how convincing are these data that
23 Zometa, 4 milligrams, is effective?
24 Second, prostate cancer produces blastic
25 bone metastasis. In considering the efficacy of
71
1 Zometa in prostate cancer, is it reasonable to also
2 consider evidence of Zometa efficacy from studies
3 of lytic bone metastases from other cancers?
4 Finally, after you consider all of the
5 evidence in the NDA, we will ask you whether the
6 regulatory efficacy requirement has been satisfied
7 for Zometa and prostate cancer; that is, is there
8 substantial evidence of efficacy from adequate and
9 well-controlled investigations?
10 [Slide.]
11 In patients with other solid tumors, the
12 4-milligram arm was statistically better than
13 placebo in time to skeletal-related event but not
14 by the proportions analysis. The 8-milligram
15 Zometa arm was statistically better than placebo in
16 both analyses. FDA believes the data are
17 convincing, that the population studied in this
18 trial received a benefit from Zometa.
19 However, the population studied was
20 heterogeneous composed of patients with many
21 different tumor types. The question we want you to
22 consider is whether these data support approval for
23 treatment of all individual patients with all types
24 of other solid tumors.
25 [Slide.]
72
1 Now, we will turn to study 010. As you
2 have heard, study 010 was an international
3 multicenter stratified double-blind study that
4 randomized patients to Zometa 4 milligrams, Zometa
5 8 milligrams or Aredia 90 milligrams i.v. every
6 three to four weeks for twelve months.
7 Randomization was stratified by center and
8 three disease strata; myeloma, breast cancer
9 treated with hormones and breast cancer treated
10 with chemotherapy. In this trial, the efficacy of
11 Zometa was determined by a noninferiority
12 comparison to Aredia.
13 [Slide.]
14 First, a few comments about design.
15 First, the design of this study assumes that data
16 from the bisphosphonate efficacy in bone metastases
17 and myeloma and breast cancer can be combined and
18 analyzed together.
19 Then a few comments about the inferiority
20 design. Here are listed several different points
21 or several different steps in the noninferiority
22 analysis. I think Novartis did a nice job of
23 presenting that. First, you estimate the
24 historical Aredia effect size versus placebo. Then
25 you compare Zometa and Aredia in the current study.
73
1 You determine Zometa efficacy by showing that a
2 reasonable fraction of the Aredia effect size is
3 proven for Zometa using statistical methodology.
4 Finally, you evaluate the constancy assumption.
5 There is much discussion and research
6 regarding the best methodology for performing
7 noninferiority analyses. In this case, because
8 only a single noninferiority trial was submitted,
9 FDA used a conservative method, the two 95 percent
10 confidence interval method which I will describe.
11 I think, again, Novartis has touched on this so I
12 will be brief.
13 [Slide.]
14 As you have seen, the slide summarized the
15 efficacy of Aredia showing the 13.1 percent
16 difference from historical trials. In the
17 submission, Novartis had centered on the 13.1
18 percent as the effect size. In the FDA's more
19 conservative analysis, we have used the 7.3 percent
20 lower confidence interval.
21 [Slide.]
22 Then, there is a comparison between Aredia
23 and Zometa, as you see, a 44 percent event rate on
24 the Zometa arm versus 46 percent on the Aredia arm.
25 [Slide.]
74
1 Although the estimate from these data
2 favors Aredia by 2 percent, again, the method used
3 a conservative limit of the confidence interval to
4 estimate the Zometa effect at 3.7 percent. By the
5 FDA analysis, using the effect size of historical
6 estimate of 7.3 percent and the worst-case estimate
7 of Zometa effect at 3.7 percent, we calculated 49
8 percent retention of the Aredia effect versus
9 placebo in sort of a worst-case analysis.
10 [Slide.]
11 A critical assumption of making
12 conclusions from noninferiority trials is this
13 constancy assumption. This requires a
14 determination that active control drug, Aredia,
15 would have shown efficacy versus placebo in the
16 current clinical-trial setting.
17 Potential problems that could challenge
18 this constancy assumption include different study
19 populations in the historical study compared to the
20 current study, changes in supportive care and also
21 study conduct could affect this assumption. Sloppy
22 trial conduct could obscure differences and make it
23 easier to win than a noninferiority analysis.
24 [Slide.]
25 Important differences were found between
75
1 the current and historical studies. Compared to
2 the Aredia versus placebo studies, more patients on
3 study 010 had a short time since diagnosis of bone
4 metastases. More patients had a history of a
5 previous skeletal-related events. In breast
6 cancer, more patients had no lytic bone lesions.
7 [Slide.]
8 So FDA used a couple of approaches to
9 evaluate the importance of these historical
10 differences. First, retrospective analysis of the
11 estimates of Aredia versus placebo data showed that
12 the Aredia effect appeared even greater in patients
13 with a short time since diagnosis of bone
14 metastases and in patients with a history of
15 previous skeletal-related event.
16 Therefore, enrichment of the study
17 population with these patients should, if anything,
18 increase the sensitivity of the study.
19 [Slide.]
20 The question of whether the active control
21 Aredia is effective in breast-cancer patients
22 without lytic lesions, however, cannot be directly
23 examined in by an historical Aredia versus placebo
24 study because only patients with lytic lesions were
25 entered. One can examine Zometa efficacy in the
76
1 subgroup of study 010 who had lytic disease. Such
2 a subgroup analysis of study 010 comparing Zometa
3 versus Aredia and breast-cancer patients with lytic
4 bone lesions did not suggest a lack of Zometa
5 efficacy here.
6 As this slide shows, it suggested a
7 noninferiority trend in favor of Zometa.
8 [Slide.]
9 Another concern was that Aredia might have
10 been effective in decreasing certain types of
11 skeletal-related events such as fractures but that
12 the current study had few such events. This
13 possibility was evaluated and the types of
14 skeletal-related events found most commonly in the
15 historical Aredia trials were similar to those seen
16 in study 010. Most were fractures and radiation
17 therapy to bone.
18 [Slide.]
19 So these are the FDA conclusions from the
20 study. Study 010 is a single study of
21 noninferiority design demonstrating efficacy of
22 Zometa for patients with bone lesions of myeloma
23 and breast cancer. We believe other data from
24 studies 011 and 039 are supportive and they
25 collectively meet the regulatory requirement for
77
1 substantial evidence of efficacy for well-controlled
2 investigations for the treatment of
3 myeloma and breast cancer.
4 We look forward to your discussion and
5 your advice.
6 Than you. Both the medical reviewers and
7 the statistical group will be available for
8 clarifying questions in the table along with the
9 company and then, this afternoon, we will also be
10 here for the discussion.
11 DR. TEMPLETON-SOMERS: We would like to
12 check and see if Dr. Albain and Dr. Kelsen and
13 maybe Dr. Taylor are on line.
14 DR. KELSEN: David Kelsen in New York.
15 DR. TEMPLETON-SOMERS: Kathy, are you
16 there, too?
17 DR. ALBAIN: Yes; I am here.
18 DR. TEMPLETON-SOMERS: Is there any chance
19 that Sarah made it through the snow storm?
20 DR. TAYLOR: I'm here.
21 DR. TEMPLETON-SOMERS: Good. We are glad
22 to see you. How is the weather there?
23 DR. TAYLOR: There are a lot of trees down
24 and a lot of ice.
25 DR. NERENSTONE: Now we see you, Sarah.
78
1 Welcome. We were afraid we wouldn't get to you
2 today, or you wouldn't get to us.
3 I am going to open in now for questions
4 from the committee.
5 Questions from the Committee
6 DR. NERENSTONE: I want to remind
7 everybody that we really want to stay focused on
8 just factual questions and don't want to open up
9 for discussion. We will have lots of time for that
10 later.
11 Questions from the committee for either
12 Novartis or FDA? Dr. George?
13 DR. GEORGE: I have a question about the
14 original design, the rationale behind having two
15 doses and what the objective of the study was of
16 that. Was there a feeling that there might be a
17 dose-response kind of effect or was it simply to
18 have two different doses, both of which would be
19 compared independently to the placebo--well, all of
20 these studies had the same question.
21 DR. NERENSTONE: Would whoever comes up to
22 the podium please identify himself.
23 DR. SEAMAN: Good morning. I am Dr. John
24 Seaman from Novartis. I worked on all the Aredia
25 submissions and also the Zometa submission. Going
79
1 back to the historical reason for the 8-milligram
2 and the 4-milligram being included in this trial,
3 it was basically done because, when we did our
4 phase I trial, we saw markers of bone resorption
5 being suppressed to a great extent with the 8-milligram dose
6 versus the 4.
7 We had gone all the way up to a 16-milligram dose
8 and saw no increase in the
9 suppression so we felt that, using that surrogate
10 marker as and endpoint, we should go a bit higher
11 to see if we had better efficacy.
12 Unfortunately, we didn't have the data
13 from the hypercalcemia trials to show there was an
14 equivalent efficacy in terms of 8 and 4 when we got
15 that trial started.
16 DR. NERENSTONE: Dr. George
17 DR. GEORGE: Just a quick follow up. The
18 reason I am asking is to try to determine what your
19 mind set was. Did you anticipate that you could
20 have had a situation where there is a reversal;
21 that is, that the 4 looks effective not the 8 and
22 vice-versa. How would you have interpreted that?
23 In other words, were you looking at a dose-response
24 kind of thing or were you really just going to
25 compare them as if they are two treatments, two
80
1 placebo, and see--
2 DR. SEAMAN: We were looking to see if we
3 were going to get a dose response, we would have a
4 better response rate.
5 DR. NERENSTONE: Dr. Raghavan?
6 DR. RAGHAVAN: I really do understand all
7 the flaws of retrospective subset analysis but one
8 of the things that has puzzled me is some of the
9 inconsistencies between the 4 and the 8/4. So,
10 recognizing all the flaws of doing a subset
11 analysis post hoc, do you have any information you
12 can share with us about the absolutely pure 8-milligram
13 group dissecting out the post-fiddling
14 dose reduction time change? Do you have any data
15 that give the same sort of endpoints that you
16 looked at the for the global group taking out just
17 the 8-milligram doses, pure 8-milligram doses?
18 DR. SEAMAN: Unfortunately, we don't have
19 any 8-milligram-dose-alone efficacy.
20 DR. NERENSTONE: I have a question along
21 those same lines. It seems to me that, according
22 to the presentation that the 8-milligram dose had a
23 higher incidence of renal problems and, therefore,
24 the doses were held until the creatinine came back
25 down to normal.
81
1 Can you give us a feeling for how many
2 doses were held and how long and is there a
3 possibility that what happened is that, instead of
4 8-milligram dosing every three to four weeks, you
5 had a number of patients who got 8-milligram dosing
6 every six to eight weeks?
7 I agree with Derek that there is some
8 inconsistency in the data and I was just wondering
9 if you have any other thoughts about that?
10 DR. SEAMAN: Why don't I turn this over to
11 Dr. Raimund Hirschberg who is from UCLA and was a
12 member of our renal advisory board that actually
13 looked at every piece of this data and let him
14 answer that for you.
15 DR. HIRSCHBERG: At the time, in June of
16 2000 when this second amendment which included this
17 creatinine sort of flag point, at this same time,
18 the 8-milligram dose was discontinued so this could
19 not have kicked in with the 8-milligram dose. It
20 excluded each other.
21 DR. TEMPLE: I didn't see a noninferiority
22 analysis of the 8-milligram, 8/4, dose. Did you do
23 that? I mean, in some sense, it is a replication
24 of the other or could be taken as. Obviously, we
25 know the rates were equal, but did you do a similar
82
1 analysis?
2 DR. SEAMAN: I will have Dr. Bee-Lian
3 Chen, our statistician, answer that. We did do a
4 8/4 analysis. This is Dr. Bee-Lian Chen from
5 Novartis.
6 DR. CHEN: My name is Bee-Lian Chen. I am
7 the project statistician for the project. I may
8 answer the question too quick in terms of
9 noninferiority. It is based on extensive analysis
10 as done by FDA and internally by Novartis. But
11 when we performed the analysis for the study
12 report, we pretty much paralleled the comparison,
13 the 4-milligram versus Aredia 90 milligrams and 8-milligram
14 versus Aredia 90 milligrams.
15 Basically, we have the confidence interval
16 based on the study 010 results. But, in terms of
17 extensive analysis performed with the
18 noninferiority details, we didn't do that.
19 DR. TEMPLE: The main result you report is
20 that you have pretty close to 50 percent retention
21 of a conservatively estimated effect of
22 pamidronate. It certainly is conservative to say 8
23 percent is the lower bound. That is the lower
24 bound of the 95 percent confidence interval. That
25 is not the typical result. That is an extremely
83
1 low result for pamidronate.
2 So, granting that it is conservative, you
3 then do your analysis and you show the difference
4 between the two treatments has a worst-case bound
5 of almost the same as half of that effect. Fine.
6 So you must have some idea what the upper bound for
7 the 8/4 group is and be able to say with some idea
8 what percent retention of the pamidronate effect
9 you have in that group.
10 DR. CHEN: The FDA statistician--
11 DR. TEMPLE: Or we do; right.
12 DR. NERENSTONE: Dr. Li of FDA?
13 DR. LI: Ning Li, FDA. Actually we did
14 the 930 analysis for the 8-milligram arm under--the
15 committee members who got the FDA briefing package
16 can find the analysis on page 68, in the
17 statistical review.
18 Essentially, the result is comparable to
19 the 4 milligrams for study 010. The only
20 difference is the percentage of retention is lower.
21 It is about 20, 20.5, percent rather than the 49
22 percent as in the 4-milligram arm, so still
23 demonstrating the 20 percent efficacy compared to
24 placebo according to our analysis.
25 DR. TEMPLE: Can I just make one
84
1 observation? Dr. Gallo said that the reason that
2 you do this 50 percent is to support the constancy
3 assumption. This will be discussed later but that
4 is incorrect. You evaluate the constancy on its
5 own. You make your best guess. There is no way to
6 really do it.
7 The preservation of effect is a clinical
8 judgment; that is, how much of the effect of the
9 control agent must you have. After all, this is a
10 situation in which you are so sure the control
11 agent has an effect that you are not allowing
12 yourself to use a placebo.
13 So we customarily ask--this is well-established--
14 how reassured are we that the effect
15 of the control is present. Arbitrarily, completely
16 arbitrarily, we have often said something like 50
17 percent, but we don't stick to that. So it is of
18 some interest that the second group didn't have as
19 strong support that it retained at least 50
20 percent. But, again, this is a very conservative
21 analysis and we know that. We have used much less
22 conservative analyses and presented them to this
23 committee.
24 DR. NERENSTONE: Dr. Przepiorka?
25 DR. PRZEPIORKA: A question about the
85
1 study population, if I can. What percentage of the
2 patients had hypercalcemia at the time of
3 enrollment in the study and what percentage
4 actually received vitamin D and calcium, as Dr.
5 Berenson indicated in one of his slides, throughout
6 the study?
7 DR. SEAMAN: Can I get a little
8 clarification what percentage in terms of entry
9 into the trial or occurred during the course of
10 trial?
11 DR. PRZEPIORKA: At the time of entry.
12 DR. SEAMAN: No one had hypercalcemia.
13 That was an exclusion criteria for entering the
14 trial and everyone was to receive calcium and oral
15 vitamin D during the course of the trial.
16 DR. PRZEPIORKA: 10 percent of the
17 patients withdraw consent. Is there any sense for
18 why?
19 DR. SEAMAN: It probably has mostly to do
20 with the patient population. We looked at that and
21 went back and looked at our old Aredia trials and
22 determined that basically these patients are
23 progressing and they just don't want to come back
24 and talk to some of our investigators, and they
25 felt the same thing. They just withdrew consent.
86
1 They were terminal and just stopped coming. It is
2 much higher in our subsequent trials you will see
3 today.
4 DR. NERENSTONE: Dr. George
5 DR. GEORGE: I have a question about the
6 constancy assumption in the 010 trial. In
7 particular, most of what was presented--I guess all
8 that was presented today had to do with breast
9 cancer and myeloma sort of together, but one sort
10 of piece of indirect evidence about the constancy
11 assumption would be to look at the Aredia, just the
12 proportion who had SREs historically compared to
13 what happened in this trial.
14 In the myeloma group, it looked much lower
15 in the past. The percentage of patients with SREs
16 in the NDA studies was much lower than on the 010
17 trial in contrast with the breast cancer which was
18 very similar, almost identical, in fact, remarkably
19 similar. Do you have any explanation for this
20 little tidbit because the issue might be, of
21 course, that the effect on this new trial is you
22 are doing a noninferiority to something that is not
23 any different than placebo. Do you have any reason
24 for that?
25 DR. SEAMAN: You are correct, and Dr. Dirk
87
1 Reitsma, who is a clinical research physician and
2 has been working with me for the last fifteen years
3 with Aredia and Zometa will answer that question.
4 DR. REITSMA: Good morning. My name is
5 Dirk Reitsma. I am a research physician with
6 Novartis Oncology Development. It is actually a
7 very interesting observation that you brought up.
8 It appears, as you say, in contrast to the breast-cancer
9 trials that, in the multiple-myeloma trials,
10 there are more events.
11 I will show you that.
12 [Slide.]
13 This slide shows you the proportion of
14 patients having any SRE contrasting the previous
15 trial in pamidronate with the multiple-myeloma
16 patients on top, study 012, and Zometa study 010 on
17 the bottom. The two lines to look at are the top
18 line in the old study 012, the pamidronate 90
19 milligrams, and the bottom line, pamidronate 90
20 milligrams in study 010.
21 What you see is the difference at the very
22 beginning between 010 and 025. What you also see
23 is that these programs behave in much the same way.
24 If you go across from three to twelve months, it is
25 10 to 21 to 24 to 38. In the bottom line, it is,
88
1 again, a big jump in the beginning, 25 to 40, and a
2 leveling off of the accrual of events.
3 Now, that is odd and I think the clue to
4 that is on the next slide.
5 [Slide.]
6 What you see here is the demographic and
7 some prognostic variables, specifically age, which
8 is the same all the way across for both trials.
9 The old pamidronate trial is on the left and the
10 study 010 is on the right.
11 If you look at the bottom line, I think that is the
12 clue to what was happening and it was shown already
13 by Dr. Berenson,
14 I believe, where you see the longer median
15 time from diagnosis to visit 2 in the previous
16 trial compared to now, what happened was, in the
17 meantime, Aredia was approved for treatment.
18 People realized that once these people had a bone
19 disease, they needed to get treatment.
20 The puzzling bit about that is why would
21 they have more fractures at that point. So the
22 hypothesis becomes that the event rate in patients
23 being diagnosed with multiple myeloma is initially
24 driven by that cohort of lesions they have when
25 they come untreated into the trial, bearing in mind
89
1 that a lot of the breast-cancer patients at that
2 point would already have been on treatment for
3 other metastases unlike multiple myeloma which
4 present with bone disease.
5 So, if I can have the next slide.
6 [Slide.]
7 The way to look at that would be to look
8 at time between diagnosis and getting into the
9 trial and how that affected the event rate. Here
10 you see, on the top again, the old pamidronate
11 trial of multiple myeloma and, on the right, that
12 column that says, "with pathologic fractures," you
13 see that split out by patients that had had their
14 diagnosis within six months and patients that had
15 their diagnosis longer than that.
16 You see that it is true that those
17 patients do, indeed, have a higher event rate with
18 a recent diagnosis. So that would say, indeed,
19 before the treatment kicks in and the chemotherapy,
20 they have events. So you see, basically, the same
21 thing in a slightly different cut of the data from
22 the current trial 010 on the bottom of the slide.
23 Again, if you look across, starting at
24 three months, you see the patients that just came
25 into the trial had a higher rate of 21 compared to
90
1 15 if they had been diagnosed previously by which
2 time chemotherapy would have kicked in in a lot of
3 patients. After that, you see a fairly similar
4 behavior in both groups.
5 DR. NERENSTONE: Dr. Loehrer?
6 DR. LOEHRER: I just have a couple of
7 questions because I can't remember. In the
8 original pamidronate studies, did everyone get
9 vitamin D and calcium?
10 DR. SEAMAN: No. In the original
11 pamidronate trials, not everyone got vitamin D and
12 calcium.
13 DR. LOEHRER: That impact, in terms of
14 this therapy, then, it is a couple of different
15 variables compared to the historical controls; is
16 that right?
17 DR. SEAMAN: I think it had a major impact
18 on particularly electrolyte and mineral imbalances
19 in terms of the amount of vitamin D and calcium
20 that was given. It wasn't even the minimum daily
21 requirement that we would take on a daily basis.
22 DR. LOEHRER: One of the things Dr.
23 Berenson mentioned, and I kind of skipped through
24 this and I wasn't sure, is, in the skeletal-related
25 events, it was said, actually, that hypercalcemia
91
1 was included in terms of their criteria yet he said
2 it was used sparingly. It is not clear how this
3 hypercalcemia was actually included in these
4 incidents of skeletal-related events.
5 DR. SEAMAN: The analysis that we have
6 done and presented here today did not included
7 hypercalcemia of malignancy. Either Grant or
8 myself can answer the question regarding why we
9 don't include it, but, in essence, we didn't
10 include it because we know bisphosphonates work in
11 hypercalcemia and we thought that that was not an
12 important endpoint and not including it was
13 probably appropriate.
14 DR. LOEHRER: Just two more questions.
15 One is, there were four renal deaths in the 8-milligram
16 dosage. Can you explain, were those
17 part--were they drug related or incidental, do you
18 think?
19 DR. SEAMAN: There were renal deaths in
20 this trial. Mainly, they were myeloma patients.
21 In every case, there was underlying pathophysiology
22 and drugs that were on board that you couldn't sort
23 out whether it was being caused by the drug or the
24 disease.
25 DR. LOEHRER: Just finally, just again a
92
1 comment, on Dr. Coleman's slide because there are
2 some people in the audience who are not necessarily
3 sophisticated and who may be more stockholders, but
4 when one looks at, for example, the incidence of
5 the various malignancies on a second slide for
6 bladder cancer, for example, one might suspect, if
7 you would be real excited and multiple 40 percent
8 by 582,000 and think this is how many patients are
9 going to be candidates for receiving this drug, the
10 fallacy of that, actually, is that, at autopsy, for
11 example, that 40 percent of the patients have it.
12 8 percent of bladder cancer patients, for example,
13 are going to survive without having metastatic
14 disease. So, I think, just to clarify those
15 issues. The same with the other malignancies.
16 DR. SEAMAN: Okay.
17 DR. NERENSTONE: I would like to ask Dr.
18 Kelsen, Dr. Albain or Dr. Taylor if they have any
19 questions for either our sponsor or the FDA.
20 DR. TAYLOR: I don't at this point.
21 DR. ALBAIN: I have a question or two.
22 This is Kathy. I have a question regarding the
23 systemic therapy status at the time of study entry.
24 In particular, were these patients already on some
25 stable type of chemotherapy or hormonal therapy or
93
1 were they newly progressing and just started on a
2 new systemic therapy because, certainly, in breast
3 cancer and, to some degree, in myeloma, the
4 systemic therapy could significantly reduce
5 skeletal-related as well as visceral events.
6 So, could you comment on that first and
7 then I have a follow-up to that.
8 DR. SEAMAN: The answer to that first
9 question is that patients entered the trial on
10 appropriate antineoplastic therapy and that meant
11 that they could have started anywhere from the last
12 year to the last two to three weeks prior to
13 entering the trial. So it is a whole host of
14 patients that are coming in at a variety of times
15 during the course of the disease.
16 DR. ALBAIN: Do you have any data on how
17 often the therapy was changed during the course of
18 the trial for each of the subsets, the chemotherapy
19 and the hormonal subset?
20 DR. SEAMAN: Yes. Just a second and we
21 will find that. Can I have this slide up? This is
22 for multiple myeloma, how many regimen changes were
23 done during the course of the trial.
24 [Slide.]
25 You can see that around 6 percent stayed
94
1 on the same therapy they entered and 6 percent of
2 the patients went to greater than five changes.
3 The vast majority had between one and two changes
4 during the course of the study.
5 DR. ALBAIN: I can't read that.
6 DR. SEAMAN: Oh; I'm sorry. I apologize.
7 DR. ALBAIN: The difference between the
8 three arms; was there any difference among the
9 arms?
10 DR. SEAMAN: Not really when you look at
11 the proportion of patients having changes in their
12 chemotherapy in the myeloma, it is very similar
13 depending on whether they had one, two or three
14 changes or up to five changes.
15 DR. ALBAIN: Do you have this data for
16 breast-cancer chemotherapy?
17 DR. SEAMAN: Yes.
18 [Slide.]
19 The next slide displayed on this slide--I
20 know you can't see it. I apologize--the breast-cancer
21 chemotherapy changes and the breast-cancer
22 hormonal changes. In the course of the trial for
23 the breast-cancer chemotherapy changes, there were
24 no changes in terms of entering the trial, only
25 around 1 to 3 percent of the patients.
95
1 The vast majority had one to two changes--around
2 30 percent had one to two changes in their
3 chemotherapy regimen and they were equivalent
4 across the treatment groups. In terms of their
5 hormonal therapies, the vast majority here, in
6 terms of changes, occurred only once. Around 40 to
7 50 percent of the patients had a change in their
8 hormonal therapy during the course of the trial and
9 twice in between, 25 and 32 percent of patients
10 having a change in their hormonal therapy.
11 DR. ALBAIN: One last question. Could you
12 comment on the study-treatment duration of twelve
13 months. Do you have any data, perhaps not from
14 this trial, but from the pamidronate trials on
15 longer durations and toxicity beyond for those
16 patients doing well at twelve months?
17 DR. SEAMAN: This trial was initially
18 designed to not only have a twelve-month core in
19 terms of looking at the overall efficacy and safety
20 of Zometa versus pamidronate but also has an
21 extension which will close the last patient, last
22 visit, within the next few months and will be
23 subject to another supplemental NDA next year for
24 long-term data.
25 In the pamidronate trials, we have data up
96
1 to 24 months for breast cancer and 21 months for
2 myeloma. There is no difference in the overall
3 safety profiles that we could see that would occur
4 at a later date. One of the things that was of
5 concern, there may have been a few more renal
6 events in the myeloma patient population in
7 protocol 012, the original Aredia trial, but it
8 wasn't clear to us, and still is not clear to us,
9 if that was probably the disease and not the drug.
10 DR. ALBAIN: Thank you very much.
11 DR. NERENSTONE: Dr. Taylor, did you have
12 a question for us?
13 DR. TAYLOR: Just to clarify, then, we
14 don't really know how many patients were on second-
15 or third-line chemotherapy when they came into the
16 trial?
17 DR. SEAMAN: I don't know, at trial entry,
18 how many were on second-line or third-line
19 chemotherapy. I know how many changes were taking
20 place during the course of the trial.
21 DR. TAYLOR: In the pamidronate data, over
22 that two-year period, can you tell if there was
23 continued reduction in skeletal events?
24 DR. SEAMAN: I'm sorry; I couldn't hear
25 you.
97
1 DR. TAYLOR: Was there continued reduction
2 in skeletal events over the 24-month period of time
3 with the pamidronate?
4 DR. SEAMAN: Yes. In the pamidronate
5 trials, you saw a continued effect on skeletal-related
6 events for the 24-month breast-cancer data
7 and the 21-month myeloma data.
8 DR. TAYLOR: Thank you.
9 DR. NERENSTONE: Dr. Przepiorka?
10 DR. PRZEPIORKA: The difference between
11 the pamidronate and placebo was actually, according
12 to the data here, somewhat smaller than at the
13 twelve-month mark. Do you have any preliminary
14 data from the current trial regarding time to
15 skeletal-related events after twelve months?
16 DR. SEAMAN: No. That is the subject of
17 the studies that we are closing down now in terms
18 of the extension. We will have that data within
19 the next year.
20 DR. COLEMAN: Dr. Pelusi?
21 DR. PELUSI: Two questions. One is,
22 either in your historical data or in your current
23 studies, do you see any difference in ethnic
24 minorities in terms of response to bisphosphonates
25 at all because it seemed like the majority of
98
1 people accrued to the current studies were mostly
2 Caucasian patients?
3 DR. SEAMAN: That is even more true for
4 the original Aredia trials but I will share with
5 you what we have from these trials so you can make
6 your own judgment. As you said, the sample sizes
7 here are small.
8 Could I have the slide up, 43, please?
9 [Slide.]
10 As I said, the sample sizes are small in
11 terms of other types of races, whether it be black
12 or other, as we captured them. But you can see, in
13 protocol 010, again with pamidronate control trial,
14 that around 29 percent of the patients in the
15 Zometa 4-milligram treatment arm had an SRE and
16 around 30 percent of the pamidronate arm.
17 In protocol 011, around 27 percent had an
18 SRE in the Zometa treatment arm and 33 percent in
19 the placebo arm. More importantly, in protocol
20 039, the prostate-cancer patient population which
21 has a problem in terms of the number of blacks
22 having it and now progressive seems to be in the
23 black patient population.
24 There is a 17 percentage of the patients
25 having an SRE in the 4-milligram arm for Zometa and
99
1 a 42 percent for placebo, which is quite high.
2 But, again, remembering the n's are small in all
3 these trials.
4 DR. PELUSI: My last question would relate
5 to your quality-of-life assessments. Were the
6 quality-of-life assessments done in general for
7 quality of life or was it specifically looking at
8 quality of life as related to skeletal events?
9 DR. SEAMAN: Unfortunately, there are no
10 quality-of-life tools for skeletal-related events.
11 The ones we used were, as you saw, FACT-G and ECOG
12 performance status. That is the best we could do
13 at that time. There still is nothing that I am
14 aware of.
15 DR. PELUSI: I don't think there is but I
16 guess the point that I was trying to just kind of
17 bring up to our awareness is, as we are looking at
18 some of these endpoints, is do we really ask the
19 question, what do these skeletal-related events do
20 in terms of function and being able to look at what
21 are the goals for some of the patients.
22 Thank you.
23 DR. SEAMAN: I am in agreement.
24 DR. NERENSTONE: Dr. Raghavan?
25 DR. RAGHAVAN: This is directed, actually,
100
1 to Dr. Williams. It may be that I misunderstood
2 your statistician in the reference to pages 67 and
3 68. Did any of the twelve inquiring minds in your
4 team with the n equals 300 patients that had 8
5 milligrams to look at the early phase of the trial?
6 300 patients got through the phase I part of the
7 trial. That is not a subset selection, really,
8 because they were just being left alone until they
9 ran into renal problems.
10 It is actually kind of interesting to look
11 at the whole global day's presentation, stuff we
12 have had in advance, this dichotomy between 8, 8/4
13 and 4. It would expect, Grant, you or one of the
14 group would have played with the numbers.
15 DR. WILLIAMS: Thanks. Now, who all do
16 you include in this?
17 DR. RAGHAVAN: I would say the twelve
18 inquiring minds that you listed. I wouldn't leave
19 anybody out, so I counted very carefully.
20 DR. WILLIAMS: I think our approach
21 basically was to consider that the 8-milligram arm
22 really was an 8-milligram arm. If you start
23 looking at the times when they were accrued and how
24 many doses they received, there is a very small
25 number of doses and many patients received only 8
101
1 milligrams.
2 So, from an efficacy standpoint, I think
3 it is a complete study. Dr. Ibrahim has done
4 analyses in her studies and in mine that they
5 basically received the same numbers of doses of
6 something and most of them were 8 milligrams. We
7 looked carefully to see if we could find the
8 evidence that there was a problem with that arm and
9 we just didn't see it.
10 I think if you accept the results, I think
11 we would feel that they would have to be more by
12 chance than anything else.
13 DR. SRIDHARA: I am Rajeshwari Sridhara.
14 The study was reviewed by Ling Li, but if you refer
15 to the statistical review on page 19, you get some
16 sense of what was going on. As Grant said, it was
17 not possible for us to look at how many doses each
18 received or whether they received exactly 8 or 4,
19 but it tells you, over a period of time, how the
20 events were occurring. So that gives you some
21 sense of what was happening between 4 and 8.
22 It gives you about the 4. Everything
23 about 8 is in the appendix. It is on page 19.
24 That is for the 4 milligrams, how the events were
25 happening.
102
1 DR. NERENSTONE: Are there any other
2 questions?
3 What I would like to do then is for us to
4 take a break. I would like to be back at 10:45,
5 please.
6 [Break.]
7 DR. NERENSTONE: If the sponsor would like
8 to start their presentation on Zometa in the
9 prostate cancer and solid tumors other than
10 prostate cancer and breast cancer.
11 Sponsor Presentation
12 Zometa in Prostate Cancer and Solid Tumors
13 Other than Prostate Cancer and Breast Cancer
14 DR. SMITH: Thank you. I am Matthew
15 Smith. I am an assistant professor of medicine at
16 Harvard Medical School and a medical oncologist at
17 Massachusetts General Hospital. I was a
18 participant in study 039. Good morning.
19 [Slide.]
20 My first task is to introduce two double-blind
21 placebo-controlled randomized trials of
22 Zometa in patients with bone metastases, protocol
23 039 for men with metastatic prostate cancer and
24 protocol 011 for patients with solid tumors other
25 than prostate cancer or breast cancer.
103
1 After introducing all studies, I will
2 present the efficacy and safety data for protocol
3 039. Dr. Robert Coleman will then present the
4 efficacy and safety data for protocol 011.
5 [Slide.]
6 The objective of the protocols was to
7 demonstrate that Zometa is superior to placebo for
8 the treatment of bone metastases.
9 [Slide.]
10 The primary endpoint for each study was
11 defined as the proportion of patients experiencing
12 any skeletal-related event, or SRE, not including
13 hypercalcemia of malignancy.
14 [Slide.]
15 The secondary study endpoints included
16 time to first SRE, skeletal morbidity rate and
17 Andersen-Gill multiple-event analysis. Secondary
18 analyses were also performed considering these
19 outcomes including hypercalcemia of malignancy.
20 Other secondary endpoints include pain and
21 analgesic scores, bone-lesion response, time to
22 progression of disease and safety including
23 survival.
24 Six months of additional survival in
25 serum-creatinine data were included in the 120-day
104
1 safety update.
2 [Slide.]
3 SREs were defined as pathological
4 fractures, spinal-cord compression, radiation
5 therapy to treat bone pain or to treat or prevent
6 pathological fractures or spinal-cord compression
7 or surgery to bone. In protocol 039, for men with
8 prostate cancer, the definition of SREs also
9 included change in antineoplastic therapy for bone
10 pain.
11 Hypercalcemia of malignancy was not
12 included in the definition of SREs for the primary
13 efficacy analyses but was included for some of the
14 secondary analyses.
15 [Slide.]
16 These are the preplanned analyses terms.
17 They were defined in the same manner as protocol
18 010.
19 [Slide.]
20 The original study design randomly
21 assigned patients to treatment with Zometa 4 or 8
22 milligrams or placebo administered as a five-minute
23 infusion. Two renal safety amendments address
24 concerns about renal safety. In June, 1999, renal
25 amendment 1 increased the infusion time from five
105
1 to fifteen minutes and increased the volume from 50
2 to 100 milliliters.
3 In June, 2000, renal amendment 2 switched
4 the 8-milligram dose to 4 milligrams and all
5 subjects assigned to the 8-milligram dose will be
6 termed the 8/4 group to indicate this change.
7 Renal amendment 2 also introduced monitoring of
8 renal function with measurement of serum creatinine
9 within two weeks before each dose.
10 Before unblinding of the data, a
11 statistical amendment defined the primary efficacy
12 analysis based on the comparison of the Zometa 4-milligram
13 group versus placebo.
14 [Slide.]
15 This figure illustrates the time lines for
16 the renal amendments in patient accrual. In
17 protocol 039, for men with prostate cancer, 368 of
18 648 men were accrued before renal amendment 1. In
19 protocol 011, for patients with solid tumors other
20 than breast cancer or prostate cancer, 195 of 773
21 patients were accrued before renal amendment 1.
22 Both studies completed accrual before
23 renal amendment 2 and treatment and follow up
24 continued through January, 2001. Notably, most
25 patients completed or discontinued the study before
106
1 renal amendment 2 and, as a result, as you have
2 already heard, three-quarters of the patients
3 assigned to the Zometa 8/4 group received only the
4 8-milligram dose.
5 [Slide.]
6 Next, I will summarize the design,
7 efficacy and safety data for protocol 039.
8 [Slide.]
9 Protocol 039 included men with progressive
10 metastatic prostate cancer. Requirements for study
11 entry included radiographic documentation of bone
12 metastases, rising serum PSA, baseline serum
13 testosterone concentration in the castrate range,
14 no strong opiate analgesics, ECOG performance
15 status of 0, 1 or 2, serum creatinine less than 3
16 and appropriate neoplastic therapy at study entry.
17 [Slide.]
18 Subjects were stratified according to the
19 presence or absence of distant metastases at the
20 time of initial diagnosis with prostate cancer. As
21 in the prior studies, all patients received
22 supplemental vitamin D and calcium and Zometa was
23 administered every three weeks for fifteen months.
24 [Slide.]
25 The groups were well balanced in most
107
1 baseline demographic and prognostic factors. The
2 mean age was 71 to 72 years. Approximately 10
3 percent of the men were black. Most men had an
4 ECOG performance status of 0 or 1. Median PSA at
5 study entry was 61 to 89 and both Zometa groups had
6 higher median PSA values than the placebo group.
7 [Slide.]
8 Patient dispositions by group are shown
9 here. As expected for this population of older men
10 with metastatic prostate cancer, about one-third of
11 men in each group completed fifteen months of
12 treatment. The Zometa 4-milligram group had the
13 highest rate of study completion.
14 [Slide.]
15 This table shows the reasons for early
16 discontinuation. Discontinuation due to an
17 unsatisfactory therapeutic effect was more common
18 in the placebo group than the Zometa groups. Rates
19 of early discontinuation for other reasons were
20 similar for all groups. These data are similar to
21 the historical results of placebo-controlled trials
22 of pamidronate in breast cancer.
23 [Slide.]
24 The primary efficacy analysis is shown in
25 this figure. 44 percent of men in the placebo
108
1 group experienced one or more SREs by fifteen
2 months. In both Zometa groups, the proportion of
3 men with an SRE was less than the placebo group.
4 The primary efficacy analysis was positive.
5 33 percent of men in the Zometa 4-milligram group
6 experienced an SRE. This improvement was
7 statistically significant compared to placebo and
8 the p-value for that comparison is 0.021.
9 Notably, the improvement in the Zometa 4-milligram
10 group remained significant even when
11 fractures were excluded as an SRE. 38 percent of
12 men in the Zometa 8/4 milligram groups experienced
13 and SRE although this improvement compared to
14 placebo did not reach statistical significance.
15 Thus, while the primary efficacy analysis of the
16 Zometa 4-milligram group was positive, the results
17 from the 8/4 group raised two important questions.
18 [Slide.]
19 First, why was no dose effect observed?
20 The doses of Zometa, as you have already heard,
21 were chosen based on early dose-finding studies
22 and, as Dr. Robert Coleman has nicely introduced,
23 Zometa targets the osteoclasts. NTX is a
24 biochemical marker of osteoclast function and
25 approximately 70 percent inhibition was achieved in
109
1 both Zometa groups, 4 and 8/4, and this inhibition
2 was maintained throughout the duration of the
3 study.
4 So, with maximum target inhibition in the
5 Zometa 4-milligram group, it is not surprising that
6 no dose effect was observed.
7 [Slide.]
8 The second question, how should the
9 results of the Zometa 8/4 milligram group be
10 interpreted relative to the positive primary
11 efficacy analysis in the 4-milligram group? This
12 is a revised slide.
13 We attempted to address this issue in a
14 combined analysis of both Zometa groups compared to
15 placebo. As you recall, the proportion of men with
16 an SRE in each Zometa group was less than the
17 placebo group. In this combined analysis, SREs
18 were reduced from 44 percent in the placebo-treated
19 group to 36 percent in the combined Zometa groups.
20 This risk reduction was significant compared to
21 placebo. The p-value for this comparison was
22 0.041.
23 I would also like to add that the
24 treatment effect observed, even in this combined
25 analysis, compares quite favorably with the
110
1 original pivotal placebo-controlled trials of
2 pamidronate in breast cancer.
3 The efficacy of Zometa across a spectrum
4 of skeletal-related events further supports the
5 effectiveness of Zometa in metastatic prostate
6 cancer.
7 [Slide.]
8 This figure shows SREs by type and
9 treatment group. For all groups, the most common
10 types of SREs were radiation to bone and fractures.
11 The risk of these and other events were
12 consistently lower in the Zometa group than in the
13 placebo group.
14 [Slide.]
15 Kaplan-Meier estimates of time to first
16 SRE are shown here. The median time to first SRE
17 was 321 days in the placebo group. The median
18 time to first SRE was longer in each of the Zometa
19 groups. After 420 days, the median time to first
20 SRE was not yet reached in the Zometa 4-milligram
21 group and this improvement in time to first SRE is
22 statistically significant. The p-value is 0.011.
23 [Slide.]
24 SREs were also analyzed as events per year
25 or skeletal morbidity rate. The SMR was lower in
111
1 both Zometa groups than in the placebo group.
2 Compared to placebo, the SMR in the Zometa 4-milligram was
3 decreased by 46 percent. This
4 improvement, again, was significant.
5 [Slide.]
6 Andersen-Gill multiple-event analyses were
7 performed to provide a robust evaluation of the
8 changes in event rate over time. Compared to
9 placebo, the hazard ratios for each of the Zometa
10 groups were less than 1. The risk reduction in the
11 Zometa 8/4 group was 15 percent. The risk
12 reduction in the Zometa 4-milligram group was
13 36 percent and significant compared to placebo.
14 [Slide.]
15 This is a new slide. As an exploratory
16 analysis, we also evaluated the proportion of men
17 with an SRE based on the radiographic
18 classification of bone lesions. Bone lesions were
19 classified as lytic, blastic or mixed. For this
20 analysis, men were defined as members of the lytic
21 subset if they had one or more lytic lesion
22 regardless of whether they had many other blastic
23 metastases.
24 They were classified as the blastic subset
25 if they had exclusively blastic lesions and other
112
1 if they could not be defined in either the lytic or
2 blastic subset. For all, the lytic, blastic and
3 other subsets, the Zometa groups had fewer events
4 than the placebo group. This consistent treatment
5 effect can be clearly seen in the blastic and other
6 subsets.
7 The result in the lytic subset is a bit
8 more varied where you do a more dramatic treatment
9 effect in the 4-milligram group but I would also
10 point out that few patients were in this subset and
11 this is subject to more random variation.
12 Again, while this subset analysis has
13 limitations, it suggests that Zometa is effective
14 in prostate cancer across the spectrum of
15 radiographic classifications of lesions.
16 [Slide.]
17 Here we see disease-related endpoints.
18 The time to progression of bone lesions and time to
19 disease progression were similar for all the
20 groups.
21 [Slide.]
22 At study completion, changes in analgesic
23 scores, ECOG performance status and FACT-G total
24 scores were similar for all groups. Pain scores
25 increased from baseline for all groups. The
113
1 increases in pain scores were attenuated in both
2 Zometa groups. Compared to the placebo group, the
3 relative improvements in pain scores were
4 significant at all time points for the Zometa 8/4
5 group and at three and nine months for the Zometa
6 4-milligram group.
7 [Slide.]
8 This table summarizes the efficacy data
9 for Zometa in men with prostate cancer. Compared
10 to placebo, both Zometa groups showed improvements
11 in the proportion of men with any SRE, time to
12 first SRE, mean SMR and multiple-event analysis
13 hazard ratios.
14 For the Zometa 4-milligram group, the
15 improvements in each of these outcomes were
16 statistically significant. Collectively, the
17 efficacy data showed that Zometa decreases skeletal
18 complications in men with metastatic prostate
19 cancer.
20 [Slide.]
21 The safety data for protocol 039 is
22 summarized in the next few slide.
23 [Slide.]
24 Causes of death during the trial or within
25 28 days of drug termination were similar for all
114
1 groups. As expected for this patient population,
2 chance of progression was the most common cause of
3 death for all groups.
4 [Slide.]
5 There were no statistically significant
6 differences in overall survival between the groups.
7 The median survival in the Zometa 4-milligram group
8 was about three months longer than the placebo
9 group. The p-value for this comparison was 0.087.
10 [Slide.]
11 This slide summarizes the common adverse
12 events. Events that occurred at least 5 percent
13 more often than placebo are highlighted in yellow.
14 Events that occurred at least 5 percent less often
15 than placebo are highlighted in green. Adverse
16 events related to intravenous bisphosphonates
17 including pyrexia and myalgias were more common in
18 the Zometa groups. Bone pain was less common in
19 the Zometa 4-milligram group than in the other
20 groups.
21 [Slide.]
22 Grade 3 or grade 4 anemia occurred in less
23 than 10 percent of men in all the groups. Grade 3
24 or 4 anemia was more common in the Zometa 8/4 group
25 than in the other two groups. Blood transfusion
115
1 and treatment with erythropoietin were somewhat
2 more common in the Zometa groups than placebo.
3 Grade 3 or 4 hypocalcemia was observed in
4 less than 2 percent of patients in all groups.
5 Grade 3 or 4 hypermagnesemia and hypophosphatemia
6 were more common in the Zometa groups than placebo
7 although no patient experienced symptoms related to
8 these mineral changes.
9 [Slide.]
10 This slide shows NCI grade 3 and 4 serum
11 creatinine changes after renal amendment 1. For
12 men randomized to the study after renal amendment
13 1, there were no grade 4 changes in serum
14 creatinine in any of the groups. Grade 3 changes
15 in serum creatinine were uncommon but a few more
16 events were observed in the Zometa groups than in
17 the placebo group.
18 [Slide.]
19 This slide shows Kaplan-Meier estimates of
20 the first increase in serum creatinine. It follows
21 the same format at Dr. Berenson's talk and serum
22 creatinine increase was defined in the same manner
23 as protocol 011.
24 Patients randomized prior to renal
25 amendment 1 are shown in the upper panel. Patients
116
1 randomized after renal amendment 1 are shown in the
2 lower panel. Before renal amendment 1 the risk of
3 serum creatinine increase was significantly higher
4 in the Zometa groups. Compared to placebo, the
5 hazard ratios for the Zometa 4-milligram group and
6 8/4 groups were 2.0 and 4.0 respectively.
7 After renal amendment 1, excess risk of
8 serum creatinine increase was markedly reduced in
9 the Zometa 8/4 group. The excess risk was nearly
10 eliminated in the Zometa 4-milligram group with a
11 hazard ratio of 1.1. These results highlight the
12 success of fifteen-minute infusion time in
13 improving the renal-safety profile.
14 [Slide.]
15 The safety data indicate that Zometa is
16 well tolerated. Adverse events associated with
17 bisphosphonates were more common in the Zometa
18 groups than placebo. The renal-safety profile of
19 Zometa 4-milligrams over fifteen minutes is similar
20 to placebo in men with prostate cancer.
21 [Slide.]
22 Collectively, the protocol 039 data
23 indicate that Zometa decreases complications in men
24 with prostate cancer and bone metastases.
25 Thank you for your attention. Dr. Robert
117
1 Coleman will now present the efficacy and safety
2 data for protocol 011.
3 DR. COLEMAN: Good morning again.
4 [Slide.]
5 It is my pleasure to present the third of
6 these randomized clinical trials and to focus on
7 the data in protocol 011 which included solid
8 tumors other than prostate cancer or breast cancer
9 in a placebo-controlled trial.
10 Dr. Matthew Smith has already given you
11 the definitions for the endpoints. He has also
12 highlighted the renal safety changes and how that
13 affected the recruitment times. So I am going to
14 go straight into the trial design which is shown on
15 this slide.
16 [Slide.]
17 These patients had to have histological
18 confirmation of advanced malignancy from a tumor
19 other than prostate or breast cancer and had to
20 have radiographic evidence of at least one bone
21 metastasis. On entry into the study, they were
22 allowed to be on appropriate antineoplastic therapy
23 and this therapy could be changed as was
24 appropriate during the study period.
25 They had to have reasonable renal function
118
1 with a serum creatinine below 3 milligrams per
2 deciliter and they had to be of ECOG performance
3 status 0, 1 or 2.
4 [Slide.]
5 Prior to randomization, patients were
6 stratified into two groups, either non-small-cell
7 lung cancer or all other solid tumors which
8 included some 20 different primary-tumor types.
9 The most common are shown here, being renal-cell
10 cancer, small-cell lung cancer, carcinomas of
11 unknown primary type, bladder and colorectal.
12 As for protocol 011 and protocol 039, all
13 patients received supplemental vitamin D and
14 calcium. The dose and dosing regimens were as
15 defined in the protocol 039 study, namely patients
16 received Zometa at 4 or 8, as you have heard
17 subsequently reduced to 4 milligrams, or placebo as
18 a five-minute infusion initially subsequently
19 amended to fifteen minutes. This was given on a
20 three-weekly basis.
21 Because of the short survival prospects of
22 these patients, the endpoint was chosen to be at
23 nine months after eight infusions.
24 [Slide.]
25 The demographics and prognostic factors
119
1 for protocol 011 are shown on this slide. The mean
2 age was similar at around 60 in all three treatment
3 groups. About two thirds were male. Some 90
4 percent were Caucasian with other ethnic groups
5 relatively infrequently represented. Around 80
6 percent were of apparently good performance status,
7 with a performance status of 0 or 1 on the ECOG
8 scale and had reasonable quality of life with a
9 FACT-G score of around 70 with 100 being
10 performance quality.
11 You can see that the stratification
12 resulted in about one-half of patients entered into
13 the study having a diagnosis of non-small-cell lung
14 cancer and the other half being the other solid
15 tumors. Because of the stratification, they were
16 well-balanced between the three treatment arms.
17 [Slide.]
18 As a reflection of the poor prognosis of
19 these patients, only 25 percent of patients
20 completed study therapy out to nine months which
21 means that three-quarters withdrew from the study
22 therapy due to various reasons which will be shown
23 on a subsequent slide. However, where possible,
24 patients were followed after therapy
25 discontinuation until the nine-month endpoint.
120
1 [Slide.]
2 This slide summarizes those reasons for
3 early discontinuation. The most common cause was
4 death, nearly always due to the underlying
5 malignancy. Next were adverse events. Again, most
6 of these were adverse events associated with that
7 malignancy rather than the treatments being
8 administered. And just under 20 percent withdrew
9 their consent. Again, this was usually due to
10 deteriorating performance status and the
11 difficulties of attending for regular infusion
12 therapies.
13 The other reasons for withdrawal are
14 relatively uncommon and are listed on this slide
15 and show no differences between the three treatment
16 groups.
17 [Slide.]
18 I will now take you to the results and the
19 primary efficacy analysis which you will recall was
20 the proportion of patients experiencing and SRE.
21 Also, on this slide, is the time to first SRE for
22 the three treatment groups.
23 On the left-hand panel, you see the
24 proportion of patients experiencing one or more
25 events. It was 44 percent in the placebo arm which
121
1 was reduced to 38 percent in the Zometa 4
2 milligrams but with a p-value which did not reach
3 significance of 0.127. For the 8-milligram dose,
4 it was 35 percent which did reach significance with
5 a p-value of 0.023.
6 However, on the right-hand side is the
7 time-to-first-event analysis which, as we have
8 already heard from Dr. Williams, is the preferred--or the
9 statistician's preferred analysis of these
10 data and takes into account the fact that many
11 patients drop out due to death and other reasons,
12 and that is not well-reflected in the time to first
13 SRE analysis--I'm sorry; in the percent of patients
14 analysis.
15 In the time to first SRE analysis, it is
16 clear that both treatments are working with an
17 extension in time to first SRE for approximately
18 two to three months and a difference of about 10
19 percent which appears three months into therapy and
20 persists out to at least eight months, as shown on
21 this graph.
22 So, in this analysis, both dosages show
23 significant improvements over placebo with p-values
24 of 0.023 and 0.034 for the 4-milligram and the 8-milligram
25 groups respectively.
122
1 [Slide.]
2 As you have seen before, here is a slide
3 showing the individual components which may cut the
4 SREs. There is a reduction for both Zometa groups
5 in terms of radiation therapy to bone, a reduction,
6 although less marked, in fractures and, as you
7 would expect with infrequent events, little change
8 in terms of surgery to bone or spinal-cord
9 compression rates.
10 [Slide.]
11 As with the other studies, skeletal
12 morbidity rates were calculated and analyzed. This
13 slide shows that the skeletal morbidity rate for
14 placebo was 2.5 and was reduced to 2.24 in the
15 Zometa 4-milligram arm which did not quite reach
16 significance at 0.069 and was significantly reduced
17 to 1.55 in the 8-milligram treatment arm.
18 [Slide.]
19 As for the other protocols, an Andersen-Gill
20 multiple-event analysis was performed. The
21 main reason for this is to look at the possible
22 differences between the strata as well as the
23 differences between the treatment arms.
24 You can see that the hazard ratio, looking
25 at the 4-milligram data, is very similar for non-small-cell
123
1 lung-cancer patients as it is for other
2 solid tumors with approximately 27 percent risk
3 reduction. For the Zometa 8/4 milligrams, overall,
4 there is a 32 percent risk reduction with an
5 apparent increased efficacy in lung cancer than in
6 other solid tumors but the numbers, obviously, are
7 relatively small in this subset analysis.
8 [Slide.]
9 This slide, like in the previous
10 presentations, has been added and I am sorry for
11 the people who are not on site who don't have
12 access to this slide, but it shows the breakdown of
13 SREs in terms of the radiographic appearances of
14 their lesions on study entry. So, as before, there
15 is a group of patients with at least one lytic
16 lesion plus-or-minus other types throughout the
17 skeleton. They are labeled as lytic, a group with
18 only blastic disease, labeled as blastic, and a
19 group that fall between those two extremes, labeled
20 other.
21 The slides shows that in the lytic
22 metastases, there is a reduction in favor of both
23 treatment arms compared to placebo. There is also
24 a reduction in the blastic patients. The group in
25 the middle, the other patients, there is no obvious
124
1 change and we would suggest that that is more to do
2 with classification of radiographic subtypes than a
3 biological underlying reason for why that group
4 show no difference from placebo.
5 [Slide.]
6 Turning now to disease-related endpoints,
7 time to progression in bone and time to progression
8 of the overall disease is shown here in days. In
9 comparing the 4-milligram arm to placebo, there was
10 no difference in time to progression in bone or
11 overall progression. There is a suggestion with
12 the 8-milligram dose that bone progression is
13 delayed from 109 to 238 days.
14 [Slide.]
15 As with the other studies, a number of
16 quality-of-life-related issues were assessed.
17 These included the brief pain inventory score,
18 analgesic scores, ECOG performance status and the
19 FACT-G quality-of-life assessment.
20 Between the three treatment arms, there
21 were no significant differences in these quality-of-life
22 endpoints. In other words, pain was little
23 changed through the nine-month period. Analgesia
24 increased slightly. There was approximately a 1.0
25 increase in performance status and a small decline
125
1 in quality of life, but no discernible or
2 statistically significant differences between the
3 Zometa arms and placebo.
4 [Slide.]
5 So, to summarize the efficacy, as we have
6 seen before, the four different analyses, the
7 proportion with a skeletal-related event, the time
8 to first event, the skeletal-morbidity rate and the
9 Andersen-Gill analysis, the results are positive
10 for all analyses for the 8-milligram dosage and are
11 statistically significant for the time to first SRE
12 in the 4-milligram dosage group.
13 We would, therefore, conclude that Zometa
14 is the first bisphosphonate to demonstrate efficacy
15 in decreasing skeletal complications in this broad
16 range of solid tumors.
17 [Slide.]
18 Finally, the safety analysis from protocol
19 011.
20 [Slide.]
21 This slide shows the primary cause of
22 death during the trial or within 28 days after
23 study-drug termination. The most frequent cause of
24 death was the underlying malignancy followed by
25 respiratory complications which is, perhaps, not
126
1 surprising for a population of non-small-cell lung-cancer
2 patients.
3 The other causes of death were infrequent
4 and are very similar between the treatment groups
5 with renal and urinary causes of death being very
6 unusual.
7 [Slide.]
8 Here is the Kaplan-Meier plot for
9 survival. There is no significant difference
10 between any of the three treatment arms with a
11 median survival of just six months.
12 [Slide.]
13 Here is a similar slide to the one you
14 have seen before of adverse events occurring in
15 more than 15 percent of patients. There are very
16 few events that are more frequent in the Zometa-treated
17 arms. In particular, in this trial, the
18 acute-phase reactions were unusual and did not
19 appear on this slide. There was a slight reduction
20 in bone pain as an adverse event in the Zometa-treated
21 patients and odd increases in nausea,
22 dyspnea and headache of uncertain reasons in the
23 Zometa 4-milligram treatment arm.
24 [Slide.]
25 In terms of hematology, electrolyte and
127
1 mineral changes, anemia incidence was low at less
2 than 5 percent for all treatment groups but was
3 slightly higher in the Zometa-treated cohorts.
4 However, the use of red blood cells and
5 erythropoietin was similar for all treatment
6 groups.
7 Electrolyte and mineral adverse events
8 were uncommon with an instance of hypercalcemia of
9 less than 2 percent for all treatment groups, but
10 there was a higher incidence of hyperphosphatemia
11 in Zometa-treated groups although this was not of
12 any clinical significance and related symptoms were
13 not reported.
14 [Slide.]
15 This slide shows the NCI grade 3 and 4
16 serum-creatinine changes in the patients enrolled
17 after the fifteen minutes of measurement. It shows
18 firstly that grade 3 and 4 changes are rare, at
19 around 1 percent, 1 to 2 percent, and there was no
20 difference in these severe events between the
21 Zometa arms and the placebo arm.
22 [Slide.]
23 You have seen this kind of analysis
24 before. It is the Kaplan-Meier estimate of first
25 serum-creatinine increase with the top panel
128
1 showing the Kaplan-Meier estimate before the
2 infusion-time amendment and the bottom panel the
3 same sort of plot after the fifteen-minute infusion
4 change.
5 The top part of the panel shows that both
6 Zometa arms were associated with an increased risk
7 of renal dysfunction with a hazard ratio of 3.8 for
8 Zometa 4 and 2.9 for Zometa 8. After the
9 amendment, there is a substantial reduction in the
10 hazard ratio for both treatment arms although some
11 increased risk persists over and above placebo with
12 the hazard ratio for the 4-milligram arm being 1.6
13 with a p-value of 0.0228.
14 [Slide.]
15 So, to summarize the safety, adverse
16 events commonly associated with bisphosphonates
17 such as hyperphosphatemia and anemia were reported
18 more frequently in the Zometa-treated groups. The
19 risk of renal deterioration was moderately higher
20 in the Zometa 4-milligram treatment group even at
21 the fifteen-minute infusion than it was in the
22 placebo group.
23 [Slide.]
24 I would, therefore, conclude, in overall
25 summary, that Zometa is the first bisphosphonate to
129
1 demonstrate efficacy in decreasing skeletal
2 complications across this broad range of solid
3 tumors affecting bone and that Zometa, when given
4 at a dose of 4 milligrams over fifteen minutes has
5 an acceptable safety profile which is probably very
6 similar to the safety profile of intravenous
7 pamidronate, 90 milligrams.
8 Thank you very much for your attention. I
9 will now pass over the podium to the next speaker,
10 Dr. David Parkinson.
11 Conclusions
12 DR. PARKINSON: Thank you, Rod. I am
13 David Parkinson from Novartis Clinical Research.
14 [Slide.]
15 Before I summarize and conclude the
16 Novartis part of the presentation today, we wanted
17 to take this opportunity to thank the hundreds of
18 investigators and research staff at literally
19 hundreds of sites in thirty countries around the
20 world. We also wanted to express our appreciation
21 to the more than 3000 patients who have contributed
22 to these studies.
23 [Slide.]
24 We believe that the problems which are
25 being addressed here today are, in fact, extremely
130
1 important to cancer patients. As you have heard
2 from Dr. Coleman earlier, the consequences of bone
3 metastases are very serious events in the lives of
4 cancer patients. Current therapy is clearly
5 inadequate to meet to clinical needs of this broad
6 population.
7 The reason we began this very large and
8 complex clinical program is that we had significant
9 preclinical data suggesting the superior potency of
10 Zometa in inhibiting osteoclast activity when one
11 compared that activity with the entire range of
12 other bisphosphonates, pamidronate included.
13 The current treatment program, therefore,
14 was designed to test the efficacy of Zometa across
15 a broad range of tumors beyond the areas where we
16 knew the pamidronate was effective.
17 [Slide.]
18 We have presented here, this morning, data
19 from three large international controlled double-blind and,
20 in the case of protocol 010, double-dummy randomized
21 clinical trials. As I have
22 indicated, these trials have included more than
23 3000 patients with breast cancer, myeloma, prostate
24 cancer and the range of other solid tumors.
25 To reemphasize what you have heard, this
131
1 is the largest clinical-trial program ever
2 conducted to evaluate the efficacy and safety of
3 bisphosphonates in patients with cancer metastatic
4 to bone.
5 [Slide.]
6 Our initial experience with this agent
7 showed that, with higher doses and/or with shorter
8 infusion times, renal events occurred which were
9 characteristic of those associated with all other
10 intravenous bisphosphonates. But we believe we
11 have shown you that, by increasing the infusion
12 time to fifteen minutes and by using the 4-milligram dose of
13 Zometa, the renal-safety profile
14 does not differ from that of pamidronate.
15 Furthermore, the safety profile is
16 consistent for other adverse events, other
17 toxicities, with that similarly typically seen with
18 other intravenous bisphosphonates.
19 [Slide.]
20 With respect to efficacy, going across the
21 range of tumors, we believe that we have shown that
22 the effectiveness of Zometa as compared to
23 pamidronate has been reliably established in
24 preplanned analyses, as you have heard, of the
25 range of the skeletal-related events with this
132
1 noninferiority design that we have talked so much
2 about.
3 [Slide.]
4 Furthermore, in the range of solid tumors,
5 we believe that there is consistent benefit across
6 the range of skeletal-related-event analysis with a
7 relative reduction of 14 percent in the proportion
8 of patients having skeletal-related events.
9 Importantly, we see an extension of the median time
10 to this first event by more than two months. That
11 represents a 27 percent reduction in relative risk
12 in a very poor-prognosis patient population, as you
13 have just seen.
14 This is the first clinical trial
15 demonstration of bisphosphonate benefit in these
16 patient populations.
17 [Slide.]
18 Prostate cancer, protocol 039, we believe
19 represents a clear demonstration of efficacy, both
20 in terms of the 25 percent relative reduction of
21 SREs as well as with the extension of the time to
22 first SRE by, in this case, more than 100 days
23 representing a 33 percent relative risk reduction
24 as compared to placebo.
25 Again, this is the first demonstration of
133
1 such benefit in patients with prostate cancer and
2 an important addition, we submit, to the therapy of
3 prostate-cancer patients.
4 [Slide.]
5 To conclude, we have confirmed the
6 activity, we believe, of Zometa in breast cancer
7 and myeloma. We have demonstrated extension of
8 that clinical benefit to the range of other tumors.
9 We believe that the consistency of this efficacy is
10 an important characteristic of the drug, that it
11 extends across all three trials, across multiple
12 tumor types and with multiple endpoints.
13 Furthermore, the efficacy is observed in
14 patients with bone lesions ranging from osteolytic
15 to osteoblastic in radiological appearance.
16 Importantly, this efficacy is observed with a
17 safety profile similar to that to pamidronate and
18 with a much more convenient administration time.
19 We thank you for your attention this
20 afternoon--still this morning--and look forward to
21 further discussions.
22 Thank you.
23 DR. NERENSTONE: Thank you. We will now
24 go on to the FDA presentation.
25 FDA Presentation
134
1 Zometa in Prostate Cancer and Solid Tumors
2 Other than Prostate Cancer and Breast Cancer
3 (Studies 010, 011 and 039)
4 DR. IBRAHIM: Good morning. I am Amna
5 Ibrahim and I will be discussing the issues related
6 to the efficacy of two trials. The first trial to
7 be discussed with be in prostate-cancer patients.
8 This will be followed by a discussion of the solid-tumor
9 trial.
10 [Slide.]
11 This is the first indication for a
12 bisphosphonate for prostate-cancer patients. As
13 already pointed out by Dr. Williams, the main issue
14 for the prostate-cancer study is the lack of
15 concordance in the 4- and 8-milligram arms of the
16 prostate-cancer trial.
17 The critical questions for the study are
18 considering both the 4-milligram and the 8-milligram arms,
19 how convincing is the prostate-cancer trial. Can the data
20 from other studies
21 provide support?
22 [Slide.]
23 The overview of my presentation on
24 efficacy of the prostate-cancer study will be as
25 follows. We will go through the study results and
135
1 some of the exploratory analysis. Then there can
2 be several reasons to explain the lack of
3 concordance in the two Zometa arms such as baseline
4 imbalances and large numbers of early
5 discontinuations.
6 I will represent to you the division's
7 conclusions regarding the possibility of baseline
8 imbalances. No impact of early discontinuations
9 could be found on the result of the primary
10 endpoint. At the end, a summary of the issues will
11 be presented.
12 [Slide.]
13 This slide illustrates the protocol-specified
14 primary endpoint; that is, proportion of
15 patients with at least one SRE. The second column
16 from the left shows the proportion of patients with
17 at least one SRE. The third and fourth columns
18 give the p-value and confidence intervals.
19 33 percent of patients in the 4-milligram
20 arm had at least one SRE. This was statistically
21 better than placebo with a p-value of 0.021. 38
22 percent and 44 percent of the patients in the 8-milligram
23 and placebo arm, respectively, had at
24 least one SRE. These were not statistically
25 different from each other. The p-value was 0.222.
136
1 [Slide.]
2 We had similar results for time to first
3 event. The 4-milligram arm is better than placebo
4 statistically with a p-value of 0.009 whereas the
5 8-milligram arm is no different statistically from
6 placebo with a p-value of 0.541. The median time
7 to first event was not reached for the 4-milligram
8 arm. There was a trend towards improvement for
9 8 milligrams. This is lack of concordance between
10 the results of the two Zometa arms. The lower dose
11 of the 4-milligram arm appears to be effective
12 where as the higher dose of 8-milligram arm does
13 not demonstrate efficacy.
14 [Slide.]
15 This slide shows hazard ratios of the
16 comparison of each Zometa arm versus placebo in the
17 two studies I am presenting. Study 011, in red,
18 illustrates the hazard ratio in the solid-tumor
19 trial the discussion of which will follow the
20 prostate-cancer trial. Then, in green, you can see
21 the hazard ratios for the prostate-cancer trial,
22 that is trial 039.
23 The point estimates for the hazard ratios
24 and the 95 percent confidence interval were less
25 than 1 for the 4-milligram arm. The point estimate
137
1 for 8 milligrams was 0.912. The upper end of the
2 95 percent confidence interval of hazard ratio was
3 over 1 at 1.226.
4 [Slide.]
5 The secondary endpoints shown here showed
6 no statistical difference across the three arms of
7 the study.
8 [Slide.]
9 The next three slides present to you
10 analyses that were not prespecified and are
11 exploratory in nature.
12 [Slide.]
13 When the results of the two Zometa arms
14 are pooled together, the p-value for time to first
15 SRE was borderline at 0.06. The point estimate of
16 the hazard ratio was 0.781 with the upper end of
17 the 95 percent confidence interval as 1.01.
18 Looking at the proportion of patients with any
19 first event, the p-value was 0.04. The point
20 estimate for the difference of proportions was -0.08.
21 [Slide.]
22 Individual SREs may be evaluated in
23 several ways with pros and cons existing for each
24 method. This graph has been reproduced from the
138
1 sponsor's briefing package. It represents the
2 proportion of patients with the individual types of
3 SRE. This was not the prespecified analysis.
4 Each type of SRE such as radiation
5 fracture or any other SRE was counted the first
6 time it occurred in that patient regardless of the
7 number of times it occurred subsequently. In this
8 method, an event of a pathological fracture which
9 resulted in surgery and radiation would show up in
10 three different categories.
11 [Slide.]
12 Can we rely on evidence from other trials
13 in the NDA for blastic metastases? Can we draw
14 support from the results of the breast-cancer and
15 myeloma study or from the other solid-tumor study?
16 This graph presents the findings in a subgroup of
17 patients in the solid-tumor trial. It gives the
18 proportion of patients with any SRE in patients in
19 whom metastasis was blastic at baseline.
20 There were a total of 133 patients in the
21 solid-tumor study with blastic only met at
22 baseline. 42 of them were in the 4-milligram arm,
23 51 in the 8-milligram arm and 40 in the placebo
24 arm. Eleven and fifteen patients in the 4-milligram and 8-
25 milligram arms and 14 patients in
139
1 the placebo had any SREs. No conclusions can be
2 drawn regarding effect of Zometa from the solid-tumor study
3 due to the subgroup analysis and the
4 relatively small number of patients.
5 Does literature provide support for the
6 efficacy of a bisphosphonate in prostate cancer?
7 There were no large randomized trials published for
8 the effect of bisphosphonate in prostate cancer for
9 SREs.
10 Does any other study provide support of a
11 bisphosphonate in prostate cancer? Novartis showed
12 a slide on the effect of Aredia on proportions of
13 patients with SREs in Aredia and placebo. Their
14 conclusion was that no effect was demonstrated.
15 This data has not been submitted to FDA for
16 analysis.
17 [Slide.]
18 We did not detect any baseline imbalances
19 in the three arms. Prior SREs, baseline PSAs and
20 the analgesic scores were important prognostic
21 factors for both arms. In the multivariate Cox
22 regression model, however, these factors did not
23 alter the overall time to first SRE results. The
24 4-milligram arm remained significantly better than
25 placebo. The 8-milligram arm was, again, not
140
1 statistical different from placebo.
2 [Slide.]
3 Early discontinuations were not the reason
4 for the discordant results of the two treatment
5 arms. Early dropouts ranged from 62 to 72 percent
6 in the three arms by the end of the fifteen months.
7 The number of infusions administered were similar
8 at three months implying an equal number of
9 patients treated. However, the number of SREs were
10 already diverging by three months.
11 [Slide.]
12 Both proportions of patients with SRE and
13 time to first SRE for the 4-milligram arm were
14 significantly better than placebo. There was no
15 difference statistically for both these endpoints
16 between the 8-milligram arm and placebo.
17 [Slide.]
18 As you have heard, the 8-milligram arm was
19 dropped from the trial due to safety reasons. It
20 may be argued that the 8-milligram arm should be
21 ignored completely. This guidance states that,
22 when considering a single multicenter trial, all
23 available data should be examined to either support
24 or undercut reliance on a single multicenter trial.
25 [Slide.]
141
1 Another guidance states that support may
2 be drawn from another trial if the other trial was
3 conducted in a disease considered to be
4 biologically similar to the trial in question.
5 Since this is a first indication for an
6 osteoblastic tumor, we will be interested in your
7 opinion to this question; are osteoblastic lesions
8 biologically similar to osteolytic lesions.
9 [Slide.]
10 I will conclude the presentation on the
11 prostate-cancer trial by the summary of issues.
12 Considering both the 4-milligram and 8-milligram
13 arms, how convincing is study 039? This is the
14 first indication of a bisphosphonate for a
15 predominantly osteoblastic disease. Can support be
16 drawn from other trials? Is there substantial
17 evidence to support efficacy of the 4-milligram
18 arm?
19 [Slide.]
20 The next discussion will be on the solid-tumor
21 trial.
22 [Slide.]
23 As with the previous presentation, primary
24 endpoint results will be presented. This will be
25 followed by issues raised because of the
142
1 heterogeneity of the patient population. SREs may
2 be affected by the concurrent therapy and issues
3 dealing with chemotherapy will be presented. At
4 the end, the summary of issues of this trial will
5 be discussed.
6 [Slide.]
7 38 percent of patients in the 4-milligram
8 arm of Zometa had at least one SRE which was
9 statistical no different to the 44 percent of
10 patients in the placebo arm. The 8-milligram arm
11 showed a statistical improvement over placebo. The
12 4-milligram arm, in this study, did not prove
13 statistically significant superiority over placebo
14 in the protocol-specified primary endpoint.
15 [Slide.]
16 This slide presents the FDA preferred
17 endpoint of time to first SRE. In these patients,
18 the was a 67-day improvement over placebo in time
19 to first SRE in the 4-milligram arm of Zometa. It
20 should be noted that this improvement occurred in a
21 group of patients who had a median survival of less
22 than seven months.
23 [Slide.]
24 This slide shows an exploratory analysis
25 where the results of both Zometa arms were pooled
143
1 together. The p-value for time to first event was
2 0.01. The hazard ratios were less than 1 for point
3 estimate as well as 95 percent confidence interval.
4 The p-value for the proportions of patients with
5 any SRE was 0.03.
6 [Slide.]
7 The population included in this trial is
8 heterogeneous. Different tumor types have a
9 varying predilection for the metastases to bone.
10 The different tumor types may have a variable
11 behavior in the bone. Lastly, there may be a
12 potentially variable response to Zometa in the
13 diverse tumor types in the study.
14 [Slide.]
15 This slide is meant to show the tumor
16 types included in the study. They were fairly
17 evenly balanced except for the renal-cell-cancer
18 patients that was slightly more in the 4-milligram
19 arm.
20 [Slide.]
21 SREs may be affected by response to
22 chemotherapy. Prior chemotherapy treatment was not
23 recorded. However, the study was blinded and
24 randomized and it is likely that it does not impact
25 on the study results.
144
1 [Slide.]
2 Summarizing the other solid-tumor trial,
3 there was no statistical difference for the 4-milligram arm
4 for the protocol-specified endpoint.
5 There was substantial evidence for the 4-milligram
6 arm for time to first SRE and there was substantial
7 efficacy for the 8-milligram arm in both endpoints;
8 that is, proportion of patients with any SRE and
9 for time to first event.
10 [Slide.]
11 Issues of the other solid-tumor trial.
12 There was a heterogenous population. Is there
13 substantial evidence to support efficacy of the 4-milligram
14 arm? If yes, should Zometa be approved
15 for all solid tumors?
16 Thank you. Dr. Nancy Scher will now
17 present to you the safety data.
18 Safety Data
19 (Studies 010, 011 and 039)
20 DR. SCHER: Good morning.
21 [Slide.]
22 I am Dr. Nancy Scher and I will discuss
23 the safety analysis of the three trials.
24 [Slide.]
25 Early in the course of the bone-metastases
145
1 trials, renal safety became a concern when three
2 patients were reported with acute renal failure.
3 The protocol was amended to improve safety. The
4 infusion time was increased from five to fifteen
5 minutes. The Zometa 8-milligram dose was decreased
6 to 4 milligrams and, as you heard, this became the
7 8/4 milligram arm of each study.
8 Serum-creatinine monitoring was required
9 before each dose. Zometa was to be held for renal
10 deterioration as previously defined and resumed
11 when the creatinine was within 10 percent of
12 baseline.
13 [Slide.]
14 This table shows renal deterioration by
15 baseline creatinine for the breast and myeloma
16 patients who were randomized after fifteen-minute
17 infusion amendment. The first row shows the number
18 of percentage of patients with normal baseline
19 creatinine who developed renal deterioration
20 according to treatment arm. The second row shows
21 patients with abnormal baseline creatinine. The
22 third row shows the outcome for all patients.
23 The percent renal deterioration was
24 similar for Zometa 4-milligram and Aredia. Renal
25 deterioration occurred in patients with normal and
146
1 with abnormal baseline creatinine. The renal
2 effects seem dose-dependent for Zometa 8 arm
3 compared to the 4-milligram arm.
4 [Slide.]
5 For this Aredia-controlled study of
6 patients with multiple myeloma and breast cancer,
7 the incidence of adverse events was similar for
8 Zometa and Aredia. Slightly more patients in both
9 Zometa arms had a greater than 25 percent from
10 baseline decrease in hemoglobin. The incidence of
11 renal-function deterioration was greater for Zometa
12 4 than for Aredia prior to the fifteen-minute
13 infusion amendment, as you have heard.
14 Post-amendment, the incidence was similar
15 for Zometa 4 and Aredia. The time to first renal
16 deterioration was similar by Kaplan Meier analysis.
17 [Slide.]
18 This table shows renal deterioration by
19 baseline creatinine for prostate-cancer patients
20 who were randomized following the fifteen-minute
21 infusion amendment. The incidence of renal
22 deterioration was slightly higher for Zometa 4 than
23 placebo. For the entire group and for patients
24 with normal creatinine, rows 1 and 3, the effect
25 was greater with Zometa 8.
147
1 Perhaps the small number of patients with
2 abnormal baseline creatinine in the Zometa 4 group
3 exaggerates the adverse effect in this arm compared
4 with placebo and even compared with Zometa 8.
5 [Slide.]
6 This table shows renal deterioration by
7 baseline creatinine for patients with solid tumors
8 excluding prostate and breast cancer who were
9 randomized following the fifteen-minute infusion
10 amendment. The incidence of renal deterioration
11 was greater for Zometa 4 than placebo, both for
12 patients with normal and abnormal renal function.
13 In this study, the effect was similar for
14 the 4- and 8-milligram treatment arms.
15 [Slide.]
16 For the two placebo-controlled studies,
17 adverse events previously reported to be associated
18 with bisphosphonates such as fever, arthralgias,
19 electrolyte and mineral abnormalities were more
20 common with Zometa than placebo, as was anemia. As
21 you heard, there was no increase in grade 3 or 4
22 hematologic events. The incidence of renal-function
23 deterioration was greater for Zometa 4
24 than for placebo. This incidence tended to
25 increase over time with duration of therapy.
148
1 [Slide.]
2 Our overall conclusions are that Zometa 4-
3 milligrams when given intravenously over fifteen
4 minutes every three or four weeks has an acceptable
5 safety profile. It is more toxic than placebo but
6 comparable in safety profile to Aredia.
7 Renal events occurred in patients with
8 normal and abnormal renal function. Particular
9 caution is indicated for patients with abnormal
10 renal function. Patients with creatinine greater
11 than 3 were excluded from the current trials.
12 Zometa is excreted unchanged by the kidneys
13 resulting in an effectively higher exposure for
14 patients with renal dysfunction. Serum creatinine
15 monitoring is appropriate in patients with normal
16 as well as abnormal renal function.
17 I would like to thank you for your
18 attention. We will be available at the table for
19 questions of clarification now and we will be
20 present for the discussion after lunch.
21 DR. NERENSTONE: Thank you very much. I
22 will open it up to questions for the committee. I
23 want to remind everyone that we are going to have
24 time afterwards for discussion of the issues.
25 Right now, I really wanted it to be specifically
149
1 questions and clarifications.
2 Questions from the Committee
3 DR. NERENSTONE: I would like to start by
4 asking Dr. Scher, when you say that you recommend
5 monitored serum creatinine in all patients, how
6 often do you think that that is necessary; before
7 every treatment, every two or three treatments?
8 DR. SCHER: I feel that the conservative
9 answer would be to model the conduct of the trials
10 after the renal amendments. So that would be to
11 monitor the creatinine prior to each dose.
12 DR. NERENSTONE: Other questions? Dr.
13 George?
14 DR. GEORGE: I have a question about the
15 intent-to-treat analysis or approach taken. There
16 were, in both of these studies, a large number of
17 discontinuations prior to the scheduled end of the
18 study. There was one brief mention that an attempt
19 was made to follow these patients for the primary
20 endpoints. Is there evidence on that point? How
21 many of these that were discontinued, in fact, were
22 followed through the period and how many were just--did that
23 discontinuation mean that no further
24 observation was possible?
25 DR. SEAMAN: As in any oncology trial,
150
1 there were a lot of dropouts, as you said, as you
2 might expect. We have taken into consideration and
3 have done analysis in terms of the three trials to
4 take into consideration whether they dropped out
5 from death or dropped out otherwise. I will gladly
6 show you that information.
7 Let's take a look at 080.
8 [Slide.]
9 If you take into consideration--let's take
10 a look at protocol 011 first and you take into
11 consideration both death and dropouts and look at
12 the time to the first skeletal-related event, you
13 can see the median time still reaches significance
14 for the Zometa 4-milligram group versus placebo
15 with the median time around 127 days for the Zometa
16 4 group and 85 for placebo.
17 Can I have the next slide, please, 081.
18 [Slide.]
19 Let's look at protocol 010 with the same
20 sort of scrutiny in terms of taking deaths and
21 dropouts. You can see, again, the median time to
22 the first SRE is around 312 days for Zometa 4-milligram and
23 252 days for the pamidronate 90
24 milligram. The p-value here is 0.099.
25 [Slide.]
151
1 Finally, for protocol 039, which is the
2 next slide, protocol 042, you can see here, again,
3 taking dropouts and deaths into consideration and
4 looking at the time to the first SRE, we still
5 maintain the significance of 4-milligram treatment
6 group over placebo with the median time being
7 around 337 days for the Zometa 4-milligram group
8 and placebo, 221.
9 Grant?
10 DR. WILLIAMS: This occurred to me, too,
11 and I know I asked for an analysis from Novartis
12 and they supplied one about how many--you were
13 supposed to monitor events after going off the drug
14 but it wasn't clear how many of those events
15 actually happened. It was really relatively few.
16 So if I were to redesign the study, I would
17 probably not try to do that because it was a very
18 sketchy collection of data thereafter, I think.
19 But looking with and without, it didn't
20 seem to make a difference. There were so few
21 events that were collected after going off drug
22 that it didn't seem to make a difference whether
23 you included or excluded them.
24 DR. NERENSTONE: Dr. George, was your
25 question answered?
152
1 DR. GEORGE: Not completely. Let me just
2 state it. We will come back to in the discussion
3 because this is an important issue with respect to
4 what these estimates are of percentage of patients
5 with these events. But, just to be clear, what you
6 showed me there was you assumed any dropout was the
7 same as an SRE at that time.
8 What I was really asking about was, among
9 those dropouts, and I think, Grant, maybe had
10 answered it, it was very rare after a dropout to
11 actually be able to observe what happened between
12 then and the end of the study.
13 DR. SEAMAN: That is correct, because of
14 the patient population.
15 DR. NERENSTONE: Dr. Przepiorka?
16 DR. PRZEPIORKA: Can you clarify, please,
17 for patients who did develop an SRE, did they stay
18 on study drug?
19 DR. SEAMAN: Yes; they did. They stayed
20 on study drug and we continued to follow them and
21 they continued to be treated.
22 DR. PRZEPIORKA: Was there any difference
23 in second, third, fourth or multiple SREs between
24 the treatment arms?
25 DR. SEAMAN: That was the Andersen-Gill
153
1 multiple-event analysis. Basically, you could see
2 from protocols 011 and 039, they were positive.
3 DR. PRZEPIORKA: Has anyone looked at a
4 hazard plot? We have looked at the rate per year,
5 but patients were restaged at multiple times during
6 the year and the Kaplan-Meier plots seem to drop at
7 three months and a little at six and more at nine
8 months. I was just wondering if there was any
9 point where the hazard for SREs actually plateaued
10 out for all three lines and you have lost the
11 effect of the drug.
12 DR. SEAMAN: No. I am looking at my
13 biostatistician. We didn't do a hazard plot like
14 that.
15 DR. PRZEPIORKA: Do you know how long
16 bisphosphonates stay in the bone?
17 DR. SEAMAN: Yes. From preclinical animal
18 models, they stay in the bone for at least a year
19 after they had received one dose. But that doesn't
20 mean they are pharmacologically active. What
21 happens is the bone is remodeled. It is covered
22 over and an osteoclast buries in exactly that same
23 site where the bisphosphonate is present. It is
24 not reactivated.
25 DR. NERENSTONE: Mr. Kazmierczak?
154
1 MR. KAZMIERCZAK: Gene Kazmierczak. I am
2 a prostate-cancer patient. Both the sponsor and
3 the FDA seem to agree that, when you consider time
4 to the first SRE, that the results in study 011, in
5 both arms of the study, the 4 and 8/4, show a
6 significant improvement from the standpoint of time
7 to the first SRE.
8 When you look at 039, the 4-milligram arm
9 doesn't show any significance with regard to
10 improvement in time--or does; pardon me--but the 8
11 doesn't. You did do an analysis that lumped the 4
12 and the 8 together with regard to numbers of SREs
13 but you didn't do that for time to the first SRE.
14 I noticed by your chart that, when you look at time
15 to the first SRE in the 8/4 arm, it wasn't
16 significant.
17 Maybe you could explain why.
18 DR. SEAMAN: I think the FDA medical
19 reviewer did do an analysis of the time to the
20 first SRE and maybe she can answer that. I think
21 it is significant but let her explain.
22 DR. SRIDHARA: It is barely significant at
23 0.06. But, you know, this is exploratory and how
24 we interpret this p-value is questionable. We are
25 doing multiple analyses and we are not adjusting
155
1 for all of this multiple testing. So it is not
2 correct to be comparing it to 0.05 and say that
3 this is borderline or any of those. It was just an
4 exploratory analysis.
5 The other thing is, your question, if I
6 understand correctly, we saw that there was no
7 difference in 8 milligrams versus placebo but how
8 come, when we combine together, we saw some
9 difference. That is simply a matter of power and
10 you are putting the two together and, therefore,
11 even a smaller difference you can detect by doing a
12 larger study.
13 DR. NERENSTONE: Dr. Pelusi.
14 DR. PELUSI: When Dr. Smith was talking, I
15 just needed some clarification in terms of
16 inclusion into the prostate study. You had on your
17 slide, on page 5, that individuals who were not on
18 strong opioids could be included. I guess I have a
19 few questions about that since many of our patients
20 would be using strong opioids and why the decision
21 was made to--what the definition is of a strong
22 opiate--why the decision was made not to include
23 them and if, in the course of their disease, they
24 required them, did they go off study or were they
25 allowed to take strong opioids?
156
1 DR. SMITH: So the definition of a strong
2 opioid was anything stronger than codeine. So that
3 part is easy. Why they were excluded if they
4 needed more pain medicines, that was simply an
5 attempt to define an homogenous patient population.
6 There can be, certainly, variability. That was the
7 basis for doing so.
8 Patients were not removed from the study
9 if they subsequently required narcotics. As you
10 can imagine, many of these men did.
11 DR. NERENSTONE: Dr. Bonomi?
12 DR. BONOMI: Did you collect serial PSA
13 levels in the prostate study?
14 DR. SEAMAN: Yes; we did. Would you like
15 to see them? Could we have that?
16 [Slide.]
17 These are the median PSA levels for the 4,
18 8/4 and placebo treatment groups over the time
19 course of the study, over the fifteen-month time
20 course of the study. As you can see, they start at
21 the baseline like you saw in the '80s or '60s and,
22 as they progressed, so did their PSA values. We
23 also looked to see was it preceded in terms of were
24 their PSA values elevated prior to their overall
25 progression and the answer is yes.
157
1 DR. BONOMI: One other follow-up on that.
2 The baselines are different, too. Are those
3 differences significant?
4 DR. SEAMAN: We did go and take a look at
5 that information to see if it had an impact on
6 whether or not they were having more progressive
7 disease in terms of their bone disease, and that
8 was not significant when we looked at that.
9 DR. NERENSTONE: Dr. Lippman?
10 DR. LIPPMAN: I think that the question
11 Mr. Kazmierczak was getting at and maybe we can
12 discuss this more later, is do you have any
13 suggestions why, in terms of biologic plausibility
14 or statistical, why the higher dose would be less
15 effective than the 4, why the 8/4 was not
16 significant and the 4 was.
17 DR. NERENSTONE: I think I am going to
18 take the chair's prerogative and say that we are
19 going to leave that for discussion and stick to the
20 questions of specific--for clarification for the
21 FDA and the sponsor because I think that will come
22 up in the discussion. I think it is going to be a
23 question.
24 Dr. Loehrer?
25 DR. LOEHRER: Actually, in September last
158
1 year, there was a New England Journal of Medicine
2 article that looked at pamidronate in patients with
3 prostate cancer increasing bone-mineral density. I
4 guess I would challenge you, and maybe the
5 inquiring minds that Derek had, in the prostate-cancer
6 population, the challenge was does Zometa
7 work in terms of blastic metastases.
8 In reality, is it actually a population of
9 patients with osteoporosis. So, putting it, framed
10 in that way, particularly in the patients with
11 prostate cancer, there was no control or at least
12 could you give me the analysis of the time on
13 hormonal therapy prior to going on study and the
14 analysis from that. Similarly, I guess we can go
15 back to the myeloma patients in terms of
16 corticosteroid use and the duration of
17 corticosteroid use prior to going on study.
18 Or, thirdly, was there a subgroup of
19 patients that had bone-mineral density done that
20 would show us that there weren't any imbalances in
21 any of the arms here?
22 DR. SEAMAN: I am going to turn this over
23 to Dr. Matthew Smith who actually wrote the paper
24 in NEJM.
25 DR. SMITH: Thank you for the question.
159
1 It can be answered in several ways. First, the
2 time from diagnosis to study entry was similar
3 across all the groups. I will see if we can pull
4 up the data on prior hormonal therapy. I am not
5 sure of that, but if we have it, we will certainly
6 present it to you.
7 Bone-mineral density cannot be reliably
8 measured in men with metastatic prostate cancer,
9 period. So the bone-mineral-density measurements
10 would be unreliable in this setting.
11 The question you raise though is have we
12 done something useful in addition to preventing
13 osteoporosis or have we done more than that and
14 prevented disease-related complications, as I
15 understand your question. So the way I asked this
16 question to be looked at and I presented in my
17 talk, but it was a bullet point, was if you look at
18 the primary efficacy analysis for the 4 group and
19 you take out all the fractures, the comparison with
20 placebo remains significant. So I think this is
21 the best way to address the issue of have you done
22 more than treat osteoporosis.
23 I think that analysis says you have. If
24 you look, also, at the other analysis, looking at
25 the type of events that you prevent, you prevent
160
1 radiation therapy to bone. You prevent other kinds
2 of problems that are separate from osteoporosis.
3 DR. LOEHRER: Could you, again, assure me,
4 then? I am looking for imbalances between the
5 groups and duration of hormonal therapy which we
6 know can predict for more problems with
7 osteoporosis. So what were the differences between
8 the groups with respect to duration of hormonal
9 therapy prior to going on study?
10 DR. SMITH: John, do we have that data?
11 DR. SEAMAN: No; we don't. Unfortunately,
12 we don't have the data.
13 DR. SMITH: I can tell you that the time
14 from diagnosis was similar. We don't have the data
15 for duration of prior hormonal therapy. But I
16 think, again, looking at the clinical endpoint,
17 fractures, if you take them out, the primary
18 efficacy analysis for the 4 group remains
19 significant compared to placebo.
20 DR. LOEHRER: Just to clarify. Can you
21 get fractures from osteoporosis?
22 DR. SMITH: That is the concern, that
23 osteoporosis leads to fractures. Osteoporosis is
24 typically defined by a bone-mineral-density
25 criteria. But the reason we are concerned about
161
1 that is that it increases the risk of fractures.
2 So, if you take the clinical outcome due to
3 osteoporosis out of the equation, you still have
4 benefit in the Zometa 4 group in the primary
5 efficacy analysis.
6 DR. NERENSTONE: Dr. Kelsen, Dr. Albain or
7 Dr. Taylor, any questions for the sponsor or for
8 the FDA?
9 Hearing none, Dr. Brawley?
10 DR. BRAWLEY: With Dr. Smith there, can we
11 have the last slide that was shown, the one with
12 the PSAs?
13 [Slide.]
14 I just want to clarify. The PSA rise over
15 time for the 8/4 group whereas it was relatively
16 stable for the placebo and 4 group. Am I reading
17 that correctly?
18 DR. SMITH: I will ask for a clarification
19 on how the analyses were done but you have to
20 realize that, as you got out further, you don't
21 have repeated measurements. So the further you go
22 out on this line, the fewer measurements you have
23 maintained. Remember, only about a third of the
24 patients completed treatment at fifteen months.
25 So, once you get out--the second half of this
162
1 panel, there is not a lot of time measurement.
2 So I would interpret the large separation
3 that you see in the right-hand side of this figure
4 very cautiously.
5 Does that answer your question, Dr.
6 Brawley?
7 DR. BRAWLEY: Yes. I think it is about
8 the best answer I am going to get right now.
9 DR. NERENSTONE: Dr. Raghavan?
10 DR. RAGHAVAN: Matt, coming back to the
11 point that I think Pat Loeher was trying to get at,
12 one of the issues that I think we do need to try to
13 get a handle on is the level of selection of the
14 patient population. When you look at the patients
15 with prostate cancer, the mean or median weights
16 were around 82 kilograms. These are, by and large,
17 relatively chubby patients with prostate cancer.
18 The length of time from presentation to
19 hitting the study is relatively long. I think it
20 is perfectly sound to have a homogenous population
21 so that is not the question. I think you have
22 achieved that. But my question is do you have a
23 concern that you have subselected out the best
24 patients so that you are not answering the question
25 in the context of killing bad, aggressive prostate
163
1 cancer.
2 These look like pretty good actors. I
3 just wondered, does this tell us something about
4 the group of patients where this may have a role to
5 play?
6 DR. SMITH: That is a great question. I
7 want to make a couple of comments. First, their
8 survival would argue against being great actors.
9 The medical survival was about one year. It is
10 pretty bad. It is actually worse than most
11 published phase II studies of chemotherapy for
12 hormone-refractory disease.
13 The second issue of weight I think
14 probably reflects the fact that many of these men
15 have been on hormone therapy for a long time, so
16 you lose lean body mass but you gain fat mass and
17 increase weight. So I think their apparent weights
18 overrepresented their vigor.
19 DR. NERENSTONE: Dr. Raghavan, go ahead.
20 DR. RAGHAVAN: Just a point of
21 clarification. Isn't it true, though, that you
22 have got heterogeneity of chemotherapy patients. I
23 think if I read the entry criteria correctly, you
24 have people who are and who are not on
25 chemotherapy. So that would argue against what you
164
1 just said for duration.
2 DR. SMITH: Let me clarify that. Prior
3 chemotherapy was excluded. So, if you had prior
4 chemotherapy, you were excluded. You could go onto
5 chemotherapy during the course of the trial but if
6 you had been on chemotherapy, you couldn't, then,
7 enter the trial.
8 DR. NERENSTONE: Any further questions
9 from the committee? Dr. Albain?
10 DR. ALBAIN: I have a question and also
11 Dr. Kelsen did. There was something wrong with our
12 audio feed. It has now been corrected. My
13 question has to do with the pathophysiology of bone
14 metastases and non-small-cell lung cancer in
15 particular and wondering if there is any data that
16 would indicate that the dose response for the
17 bisphosphonate might differ than in the hormone-dependent
18 malignancies of breast or prostate,
19 getting at this issue in the other solid tumors and
20 in lung, in particular, the higher dose seemed to
21 better.
22 So is there anything in the
23 pathophysiology of the process?
24 DR. SEAMAN: I will bring Dr. Robert
25 Coleman up to talk to that.
165
1 DR. COLEMAN: Robert Coleman. I am not
2 aware of any specific data that have looked at
3 marker changes in lung-cancer patients and compared
4 them to prostate-cancer patients, so it is a good
5 question but I am not aware of any data in the
6 literature to answer it and I don't think we have
7 an analysis of marker changes by disease type. We
8 have marker changes cut in other ways, but not by
9 disease type.
10 So I am sorry, I don't think I can answer
11 your question.
12 DR. ALBAIN: Hopefully, we can get that
13 data down the road. To follow up regarding the
14 lung patients, is there any further analysis of use
15 of systemic chemotherapy in this group? We saw a
16 bullet go by that probably there were no imbalances
17 due to the randomization, but how many of these
18 patients were not being treated with systemic
19 chemotherapy versus were? Do you have that data?
20 DR. SEAMAN: We are looking it up right
21 now. I think we do have that data. Just a second.
22 We have the one prior to. How about, do we have
23 the one after?
24 [Slide.]
25 These are the types of chemotherapy. I
166
1 don't think you can see these so I will read them
2 to you. The vast majority of patients were
3 receiving chemotherapy, around 75 percent,
4 carboplatin being the number-one agent, paclitaxel
5 second, gemcitabine and cisplatin followed by that.
6 So they were receiving, in 75 percent of the cases,
7 some sort of chemotherapy for solid tumors.
8 You need to know about non-small-cell lung
9 cancer also; correct?
10 DR. ALBAIN: Correct. If you have it.
11 DR. SEAMAN: That was for everything. I'm
12 sorry. I will take that back. If we could go back
13 to the previous slide. That was for all patients.
14 That includes not only non-small-cell lung cancer
15 but it also includes all the other cell and tumors
16 during the course of the trial. I don't think we
17 have it broken down specifically for non-small-cell
18 lung cancer.
19 DR. ALBAIN: Do you have it by subsequent
20 change in regimen like you did for the other
21 studies?
22 DR. SEAMAN: Yes; I do. Slide 17.
23 [Slide.]
24 In the non-small-cell lung cancer, we
25 broke it down by strata in this. About 16 to 20
167
1 percent of the patients stayed on their original
2 regimen. Between 30 and 40 percent of the
3 patients, up to 50 percent of the patients, changed
4 at least one time up to twice for the non-small-cell lung
5 cancer.
6 In other solid tumors, it is about 30
7 percent of the patients stayed on their original
8 regimen and then 40 percent changed once and around
9 12 to 20 percent of the patients changed at least
10 twice.
11 DR. ALBAIN: Could the camera be moved so
12 I could see the placebo column, please?
13 DR. SEAMAN: Yes.
14 DR. ALBAIN: Thank you.
15 DR. NERENSTONE: Kathy, your question was
16 answered?
17 DR. ALBAIN: Yes. I just couldn't read
18 the placebo column. It was off-screen. I can see
19 it now. Thank you.
20 DR. NERENSTONE: Dr. Kelsen?
21 DR. KELSEN: Thank you. This may have
22 been answered. I may have missed it when some of
23 the audio was lost, but, in trying to get a better
24 understanding as to why, in study 039, there was
25 benefit to the lower dose of Zometa but not as
168
1 apparent benefit to the higher dose.
2 In the briefing book, it indicates that
3 Aredia was also studied in prostate cancer and
4 failed to show benefit. Could you give us any
5 data? Was there a similar trend toward benefit
6 that didn't reach statistical significance? Was
7 there a complete absence of benefit? Was there any
8 reason to suspect that osteoblastic lesions really
9 are more resistant to bisphosphonate therapy?
10 DR. SEAMAN: In the study that was
11 conducted in the prostate-cancer trial with
12 pamidronate, when we looked at the data from there,
13 we could not detect a benefit in skeletal-related
14 episodes. However, that was not the primary-efficacy
15 endpoint in this trial. The primary
16 efficacy endpoint was sized for pain.
17 It may be a reflection of the design of
18 the trial in terms of the inclusion criteria, but
19 we certainly did not see a benefit from the 90
20 milligram treatment group compared to placebo in
21 that trial.
22 DR. KELSEN: Thank you.
23 DR. NERENSTONE: Any other questions from
24 the committee? Dr. Przepiorka?
25 DR. PRZEPIORKA: Just to clarify about the
169
1 quality-of-life indicators, please. If you could
2 just summarize--if what I am summarizing, let me
3 say, it is true that there is no difference in the
4 quality-of-life indicator, or the change in
5 quality-of-life indicators, between the 4-milligram
6 arm and placebo for the prostate-cancer trial and
7 the solid-tumor trial?
8 DR. SEAMAN: That's correct.
9 DR. PRZEPIORKA: There is no difference in
10 three of the four indicators for the breast cancer
11 myeloma trial and, actually, the Aredia arm fared
12 better in some of the subscales for the FACT-G
13 trial?
14 DR. SEAMAN: That's correct. But you
15 should make sure you look at that FACT-G scale with
16 some concern because that change, when you look at
17 the literature, is probably not clinically relevant
18 in terms of the numbers you are seeing there. It
19 also should be remembered that we would not expect
20 to see a difference between the three active
21 controls in this sort of study in protocol 010.
22 DR. NERENSTONE: Dr. Ibrahim?
23 DR. IBRAHIM: Just a clarification for
24 study 011. It will be difficult to look at any one
25 individual strata. It would be difficult because
170
1 there were small-cell patients in the non-small-cell
2 stratum. There were several small-cell
3 patients, small-cell lung-cancer patients, who were
4 there. So it will be difficult to--an intent-to-treat
5 analysis will be different from the actual
6 fact. It could be.
7 DR. NERENSTONE: Thank you.
8 Right now, it is 12:15, for those of us in
9 Washington. We are going to break for forty-five
10 minutes and return at 1 o'clock. So, our people at
11 remote areas, it will be a forty-five minute break.
12 Thank you.
13 [Whereupon, at 12:15 p.m., the proceedings
14 were recessed, to reconvene at 1:00 p.m., this same
15 day.]
171
1 A F T E R N O O N P R O C E E D I N G S
2 [1:00 p.m.]
3 DR. NERENSTONE: Dr. Williams, if you
4 would like to start us off.
5 Introduction to the Issues
6 DR. WILLIAMS: To begin this afternoon's
7 session, I will briefly introduce the questions
8 that we have prepared. You should have a copy of
9 the questions in your packet. You will probably
10 want to refer to them.
11 On the first page, you will find two
12 tables that summarize the Zometa efficacy findings.
13 These will be useful as a reference as we proceed.
14 The first table summarizes the
15 noninferiority study of Zometa versus Aredia in
16 breast cancer and myeloma. The second table
17 summarizes the two placebo-controlled studies, one
18 in prostate cancer and the other in solid tumors.
19 [Slide.]
20 Results from both of the critical
21 skeletal-related events analyses are listed here,
22 the proportions analysis, as presented in the third
23 column, followed by the estimated treatment
24 difference and the 95 percent confidence intervals
25 within that difference.
172
1 A negative number here is an estimate that
2 favors the Zometa arm. The sixth column provides
3 the hazard ratio for time to first skeletal-related
4 event and then the confidence intervals. In this
5 analysis, a number less than 1 is an estimate that
6 favors Zometa. P-values are included for each
7 analysis.
8 As I discussed earlier, each of these
9 clinical trials suggests one or more important
10 questions which I will summarize again in the next
11 three slides. For study 010 in breast cancer and
12 myeloma, we note that it is a single study of
13 noninferiority design. The question for ODAC is do
14 the totality of the data in the NDA provide support
15 for this indication.
16 [Slide.]
17 In prostate cancer, the 4-milligram Zometa
18 arm shows convincing results for both the primary
19 and secondary skeletal-related endpoints. The 8-milligram
20 arm shows no statistical difference from
21 placebo. Again, two questions arise. First,
22 considering both of the arms of the study, how
23 convincing are these data that Zometa 4 milligrams
24 is effective.
25 Second, prostate cancer produces blastic
173
1 bone metastases. When considering the efficacy of
2 Zometa in prostate cancer, is it reasonable to also
3 consider evidence of Zometa efficacy from studies
4 of lytic bone metastases from other cancers?
5 [Slide.]
6 In patients with other solid tumors, the
7 4-milligram Zometa arm was statistical better than
8 placebo in time to skeletal-related event but not
9 by the proportions analysis. The 8-milligram
10 Zometa arm was statistical better than placebo in
11 both analyses. FDA believes that these data are
12 convincing for the populations studied and received
13 benefit. However, the population was
14 heterogeneous. We want you to consider whether
15 these data support approval for the treatment of
16 all individual patients with all types of other
17 solid tumors.
18 [Slide.]
19 For each of these studies, we want you,
20 then, to put on the regulatory hat with us--if we
21 have one in your size. You have to check with
22 Stacy--and determine whether the data collectively
23 meet the regulatory standard for demonstrating
24 efficacy, substantial evidence from adequate and
25 well-controlled investigations.
174
1 As I noted earlier, this means evidence
2 from multiple clinical trials--it usually means
3 evidence from multiple clinical trials--but very
4 impressive and robust results from a single
5 multicenter trial sometimes have been accepted.
6 You may find that one of these trials is
7 so impressive that it supports approval without any
8 support from another trial or you may find that,
9 while a single trial is not convincing alone, you
10 find the results of another study to be supportive.
11 We look forward to your advice on these
12 matters.
13 DR. NERENSTONE: Thank you.
14 I am going to open the discussion now to
15 the committee members. Just to check, Dr. Albain,
16 can you hear us?
17 DR. ALBAIN: Yes; I can.
18 DR. NERENSTONE: Dr. Taylor, are you
19 there?
20 DR. TAYLOR: Yes; I can hear you.
21 DR. NERENSTONE: Dr. Kelsen, are you with
22 us?
23 DR. KELSEN: Yes; I am.
24 Committee Discussion and Vote
25 DR. NERENSTONE: Does anyone on the
175
1 committee have any questions? Actually, let's have
2 some comments now. We are opening it up for
3 discussion.
4 Dr. Lippman, I know that you had some
5 comments that I think are very important to open up
6 the discussion.
7 DR. LIPPMAN: I think that my comments
8 were asked by subsequent people on the committee
9 and answered so I don't have anything else to add
10 to that. I think the issue of the 8/4 versus 4, we
11 have heard all that we will know about that issue.
12 DR. NERENSTONE: I think that is an issue
13 that is still a little problematic. Thinking about
14 that, which is why is there not consistency in all
15 the arms, I agree with Dr. Lippman that it seems
16 like the sponsor has not really been able to
17 adequately give us data about why, or explain why,
18 it might be not consistent. It may be that they
19 don't have the data.
20 Does the FDA have any thoughts about why
21 there might be some inconsistency?
22 DR. WILLIAMS: We certainly looked to find
23 some source of the problem. In my opinion, if you
24 are looking for an excuse to explain, say, the drug
25 works and why we find these things, the trials were
176
1 designed at 80 percent power. You expect one in
2 five studies to be negative. I think the power of
3 these studies was probably even less than that,
4 getting the final effect size.
5 Depending on which analysis you see, you
6 see trends in favor, and we don't see anything
7 against. So my opinion is that, if the drug works
8 and we have these results, it is probably chance.
9 DR. PAZDUR: One of the things I would
10 like to bear in on this is we should look at the
11 totality of data that is coming and not simply the
12 clinical trials. One of the things that I urged
13 the company to reiterate at this time, or people in
14 their own minds to think of, is the pathophysiology
15 of the disease, any potential differences in
16 osteoblastic or osteolytic lesions, especially with
17 the interaction of the bisphosphonates.
18 Would one expect from the underlying
19 disease process and the mechanism of action of the
20 bisphosphonates that one should see a difference in
21 osteoblastic and osteolytic disease. Perhaps the
22 company could address this again for the committee
23 because, really, it isn't just the clinical trials.
24 It is the totality of evidence that we must look
25 at.
177
1 DR. SEAMAN: I am Dr. John Seaman from
2 Novartis. I would like the slide with the
3 information regarding all the trials that we have
4 done with Zometa for hypercalcemia in bone
5 metastases, et cetera.
6 [Slide.]
7 This slide depicts all the results of
8 trials done in terms of Zometa in a malignant
9 indication. On the left-hand side, you will see
10 the Zometa hypercalcemia data in regards to the
11 response rates for Zometa, 4-milligrams and 8
12 milligrams, versus pamidronate. These are in
13 patients with bone metastases because they are
14 probably a more reasonable patient population to
15 look at because humoral is somewhat different.
16 As you can see, the proportion of patients
17 having a response to Zometa in these treatment
18 groups are somewhat variable. You see a 90 percent
19 response for the Zometa 4-milligram treatment
20 group, and 84 percent response for the 8/4 and a 80
21 percent response for pamidronate.
22 That is somewhat consistent when you look
23 across the three trials that we have done in terms
24 of bone metastases. There is a reflection that you
25 see responses in terms of efficacy for prostate
178
1 cancer and for other solid tumors include non-small-cell
2 lung cancer that is somewhat different
3 when you look at the treatment groups.
4 If you look at the 4-milligram treatment
5 group, for example, in the prostate-cancer patient
6 population, there is a 33 percentage of patients
7 having an SRE. For the 4-milligram group, it is
8 around 38 percent. They are both reducing the
9 portion of patients having an SRE although the 8/4
10 does not reach significance.
11 Somewhat of a different pattern sort of
12 shows up in terms of the protocol 010 where the 8-milligram
13 group is a bit more successful in having
14 less SREs, 35 percent, and 38 percent in the 4-milligram
15 treatment group.
16 Then you see, even in the other trials
17 where we have seen noninferiority, you see
18 variability in the data that you are looking at.
19 Our feeling is the totality of data supports the
20 fact that Zometa works across all these tumor types
21 in terms of what is going on.
22 I think this is the first time we have an
23 opportunity to look at all the results for all the
24 trials we have done in a malignant setting with
25 patients who have bone metastases.
179
1 DR. NERENSTONE: Thank you.
2 Dr. George?
3 DR. GEORGE: With respect to this issue of
4 whether the results are contradictory, or the 4 and
5 8 business, it seems to me, in looking at all this,
6 to try to synthesize it some, the obvious thing
7 that I would do is just pool those results. If you
8 had a situation--if your model ahead of time is
9 that you can't come up with the plausible
10 biological reasons why the 8 would produce a worse
11 control in terms of the primary endpoint, then a
12 sound statistical procedure is to sort of, in these
13 restricted inference procedures--is to pool the
14 data, to pool them and say, "This is the best
15 estimate we can do of what the effect is."
16 I would do that on all these studies, if
17 you want to try to get a handle on what that effect
18 size really is. I think that is important because
19 we have been focusing on a lot of the--we look at
20 these confidence intervals but the effect size is
21 really a key issue.
22 I like to think in terms of the number-needed-to-
23 treat analyses. If the effect size is
24 really, say, 8 percent like it seems to be in one
25 of these, then the number needed to treat is 12,
180
1 12.5, to be precise, or something like you have to
2 treat twelve or thirteen patients to reduce one SRE
3 in one patient. The other eleven or twelve are
4 treated for nothing--well, maybe not nothing, but
5 at least with respect to the primary endpoint, that
6 is what happens.
7 So then you have to assess is it worth it
8 in that sense. So that is the way I tend to think
9 of it. In other words, whether you do a formal
10 pooling, just add them together, or pool them some
11 other way, I think it is important to do that.
12 That is the only way you can make sense of it.
13 But I have a fundamental point I wanted to
14 address concerning--it is sort of a technical point
15 but I think it is important with respect to what it
16 is we are estimating. There have been two
17 analyses, types of analyses, presented here with
18 respect to the primary endpoint in the first event
19 that occurs.
20 One is just simply the percentage of SREs
21 that are observed in the specified time frame. The
22 other is the time to that event. Now, there is a
23 problem with both of these. It boils down to the
24 fact--or it relates to the fact that you are in an
25 area of a lot of competing risks. So what I would
181
1 have really like to have seen is something like
2 cumulative-incidence kinds of curves.
3 Just to be clear, if you look at the
4 percentage of events, the percentage of SREs that
5 occur in this time frame, if you had no censoring
6 at all, then those percentages would give you the
7 same as the cumulative incidence. But, when you
8 have censoring, they give you the wrong answer.
9 They give you something different.
10 The answer you are getting when you look
11 at those percentages of SREs is too low. It is too
12 low. The cumulative incidence would actually be
13 higher. On the other hand, if you do the time to
14 the event and you censor the competing events,
15 which I think is what was done on both the FDA's
16 and the sponsor's analysis, then you get something
17 that is too high.
18 So that is why you get this discrepancy.
19 Maybe it is true that, if you did something else,
20 if you did something like this cumulative
21 incidence, that you would get the same qualitative
22 answer but it is not guaranteed. So I don't know
23 what that answer is. I would have liked to have
24 seen that.
25 If you have already done it--
182
1 DR. SEAMAN: We have done it
2 DR. GEORGE: Then that would be
3 interesting to see.
4 [Slide.]
5 DR. SEAMAN: Here is the cumulative
6 incidence rate of SREs in all patients for the 4,
7 8/4 and pamidronate treatment group in terms of
8 median time to first SRE. You can see that it is
9 very similar in terms of the median time being
10 around 397 days with the 4, 373 for the 8/4 and 370
11 for the 90 milligram.
12 DR. WILLIAMS: Could you clarify? Is that
13 patients or all events, patients within event?
14 When you say cumulative incidence rate, does that
15 mean to the first event in a patient or all events?
16 DR. SEAMAN: It is the first-
17 DR. GEORGE: And do you have the same
18 thing for the other--
19 DR. SEAMAN: Can I see the next slide,
20 Slide 11?
21 [Slide.]
22 You can see that we have maintained the
23 significance there with the median time, 314 not
24 reached for the 8/4 and 185.
25 DR. GEORGE: Remind me; where was the time
183
1 point? If you do something vertically here--I just
2 forgot the time that you looked at.
3 DR. SEAMAN: Nine months is the endpoint
4 DR. GEORGE: Nine months?
5 DR. SEAMAN: Yes.
6 [Slide.]
7 And then for 039, again, the cumulative
8 incidence for 4-milligram treatment.
9 DR. GEORGE: So, trying to assimilate this
10 quickly, qualitatively, you get the same results;
11 is that correct?
12 DR. SEAMAN: Correct.
13 DR. SCHER: Excuse me. To me, it looks
14 like the top two, the 8 and the placebo, seem to be
15 running together.
16 DR. SEAMAN: It looks like exactly what we
17 had if you look at the time to first SRE in that
18 039 data. They are running together the opposite
19 way.
20 DR. BRAWLEY: That is consistent with the
21 PSA data that was shown earlier as well for the
22 8/4.
23 DR. SEAMAN: Matt wants to address that a
24 bit more. We talked about it a bit during the
25 break and maybe he could say a little bit more
184
1 about that because we were talking about the 8/4
2 data a bit during the break, too, when we looked at
3 the data and talked a bit.
4 DR. SMITH: I wish I could, with
5 certainty, explain the difference between the 4 and
6 8/4 group in the prostate study. I don't think it
7 is possible. So I think an honest attempt has been
8 made and a thoughtful review by the FDA
9 statisticians have done so.
10 I think about it in the following way. It
11 is not just about reaching significance. It is
12 also about the treatment effect. I like what Dr.
13 George has said which is, if you think there is the
14 possibility of imbalances or chance effect, then
15 hold it against the drug and lump the less
16 favorable arm in with the more favorable arm.
17 If you do that analysis, you have a 20
18 percent relative reduction--not absolute reduction,
19 but 20 percent relative reduction in SREs in the
20 combined 4 and 8/4 group compared to placebo. That
21 is statistically significant.
22 By the way, that is exactly the same
23 treatment effect you see in the pivotal placebo-controlled
24 trials of pamidronate in breast cancer
25 with 24 months follow up. You lump 18 and 19. So
185
1 breast-cancer patients with bone metastases treated
2 with either chemotherapy or endocrine therapy, if
3 you lump the studies together, at 24 months, you
4 have about a 20 percent reduction in risk, relative
5 reduction in risk.
6 So I think the treatment effect in
7 prostate cancer is quite comparable to that you see
8 in metastatic breast cancer. In the pooled
9 analysis, trying to correct for any possible
10 imbalances, you do maintain statistical
11 significance with the primary effectiveness
12 analysis.
13 I think there have been a lot of
14 thoughtful attempts to provide other explanations,
15 possible imbalances and such. When you do that, it
16 does move the 8/4 group closer to statistical
17 significance. It doesn't reach it. I am not sure
18 what else could be done.
19 DR. NERENSTONE: Dr. Lippman?
20 DR. LIPPMAN: I would just like to
21 clarify, Stacy, your comment. My question wasn't
22 meant to be negative. It wasn't meant to be that
23 the company couldn't explain it. I actually asked
24 when the FDA presented. People that have really
25 pored over this data and tortured it, I was just
186
1 trying to see if they had any sense of some of
2 imbalances or thoughts that made it kind of--or
3 people that know more about bone metastases than I
4 do, if there was biological plausibility for this.
5 Certainly, in other areas, higher doses
6 are not always better. Clearly, the 4-milligram is
7 the cleanest arm and that data was convincing. The
8 8/4 shows the same trend, although it is not
9 significant. So, again, I would just like to
10 reiterate what Grant says. I can accept the fact
11 that this was a chance finding. The totality of
12 the data are very consistent.
13 I didn't mean--I thought the way you
14 phrased it might have a sense that I was not happy
15 with the answer or negative in any sense. I just
16 was trying to see if people who looked at the data
17 more closely had any other insight.
18 DR. BONOMI: Maybe I am missing something,
19 but when we looked at the curves, when you showed
20 the PSA levels, it was higher at baseline and it
21 went higher throughout the study and that would
22 suggest that their disease wasn't being controlled
23 and that they were a worse group of patients.
24 DR. SMITH: I incompletely answered Dr.
25 Raghavan's question about survival.
187
1 If you could bring up the overall survival
2 slide. I think it is 28.
3 I misspoke by saying their median survival
4 was about a year. It is, in fact, a bit longer
5 than that. There were differences none of which
6 reached statistical significance but there were
7 differences in their survival.
8 I need the Kaplan-Meier survival curves,
9 Slide 28 from my presentation.
10 [Slide.]
11 So the median survival in placebo was
12 about fifteen months, 469 days. It was worse in
13 Zometa, 8/4, 418, and best in Zometa 4. So there
14 was trend towards better survival in 4 and the
15 worst arm was the 8/4.
16 Now, in all of the other studies, we have
17 looked at survival basically overlapped. So, if
18 you look at overall survival, maybe as another way
19 of thinking about potential imbalances, and
20 recognize that we don't have a better explanation.
21 I am trying to provide a reasonable explanation
22 here.
23 If you look at that and say, this is an
24 integrated way of looking at prognostic factors,
25 perhaps this is the explanation. My guess is if
188
1 you pooled the survival for the 4 and 8/4, it
2 looked exactly like placebo.
3 DR. NERENSTONE: Dr. Albain, did you have
4 a question?
5 DR. ALBAIN: No.
6 DR. NERENSTONE: Dr. Temple?
7 DR. TEMPLE: I wanted to ask Steve how
8 much explanation of these kinds of differences is
9 needed. The power of these studies isn't infinite.
10 The p-values for the most favorable results are
11 fairly close to 0.05. If the true difference is
12 what it was in that study, your chance of
13 replicating it on another study is only modest, at
14 best.
15 So is there anything here to explain
16 except that the effect isn't very big?
17 DR. GEORGE: I would say no. I think the
18 results here are pretty clear, if you are trying to
19 get my conclusion ahead of time. It is pretty
20 clear. The way I get them is to pool those results
21 together. I see attenuation of the effect of the
22 most extreme one because I think it is probably
23 random variation. I haven't heard any other
24 plausible reason.
25 I have heard some things thrown out that
189
1 might have contributed but there is no real good
2 evidence that there is anything there. So it is
3 fairly straightforward for me in that sense.
4 DR. NERENSTONE: I just wanted to ask--or
5 talk to Dr. Williams. I think that three of the
6 studies are pretty--to my mind, the answers are
7 pretty--or two of the studies-- are pretty clear.
8 But, going back to the solid tumor, which is a
9 little bit more of a hodgepodge of studies, and
10 some concern about the heterogeneity of the patient
11 population, I guess, from a clinical point of view,
12 I would look at the statistical numbers there.
13 There, of interest, is that the p-values
14 are significant; that is, the p-values for the
15 Zometa 4 milligrams in the time to first SRE, which
16 is your preferred analysis, was significant at the
17 0.02 level and both the difference in the
18 proportion as well as the time to first SRE are
19 significant for the 8-milligram.
20 The argument about heterogeneity of the
21 patient population, I think, is important. Again,
22 if you postulated that perhaps there would be a
23 difference of what it is doing or how tumor types
24 are going to act on the bone, and if they didn't
25 have a significant p-value, that would be raised.
190
1 The fact that these are significant, in my
2 mind, suggests that the mechanism either is the
3 same or is affected in the same positive way. To
4 my thinking, I don't have a problem with that. I
5 think, from a clinical point of view, to have a
6 phase III trial in metastatic cervix cancer to the
7 bone is not going to happen and, in fact, this data
8 suggests that those patients would benefit from
9 this kind of treatment.
10 I just wanted to open this up to the
11 committee because I think is the most problematic
12 study.
13 Dr. Lippman?
14 DR. LIPPMAN: Again, this gets at what
15 Rick raised. From what I have heard, and not being
16 an expert on bone physiology, it seems as thought
17 the pathophysiology of the lytic and blastic
18 lesions, the aspect is similar. Unless there is
19 really a biological reason to think that it is
20 different, I think it may be reasonable to include
21 them all as one for the reasons you mentioned, and
22 Stacy, that it is really not going to be practical
23 or possible to do single-site definitive trials.
24 So there would have to be some real
25 biologic reason not to, and I haven't heard that.
191
1 DR. NERENSTONE: Dr. Temple.
2 DR. TEMPLE: As Grant said, we always
3 wonder how quickly to generalize any putative
4 common mechanism. Anybody can make up a mechanism,
5 you know, and it will always be plausible because
6 they wouldn't offer anything silly. So the
7 question is how much replication, how much
8 verification, do you need.
9 As Grant said, the division thought, well,
10 you have got prostate, you have got breast, you
11 have got a variety of solid tumors. That is
12 getting to the point where you might consider this
13 to be a generalizable finding. That is obviously
14 one of the things we want to ask everybody, how
15 plausible that seems, plus all the other data from
16 drugs in the class. You bring all of that.
17 DR. NERENSTONE: Dr. Przepiorka?
18 DR. PRZEPIORKA: I think there is a lot of
19 data to suggest that statistically we are seeing
20 something. But I guess the question on the table
21 is is there substantial evidence to support
22 approvals for these indications. We are clearly
23 not treating the primary cancer, and the survival
24 statistics pretty much show that we are not
25 treating the primary cancer.
192
1 The quality-of-life indicators demonstrate
2 no benefit for the drug. The median time to SREs
3 are the same in the breast-cancer trials so you
4 can't really say anything since that is an
5 equivalency trial. But, for the solid-tumor trial,
6 the reduction of median time to SRE is only two
7 months. So the question is, in the clinic, is that
8 going to make a difference.
9 The point reduction in SREs at nine months
10 or one year, I have heard conflicting data and it
11 wasn't on the slide, is 8 percent. Is 8 percent
12 enough to really say that this is substantial
13 clinical evidence for a benefit for these patients?
14 I would like to hear anybody else's
15 comments about that, too.
16 DR. NERENSTONE: Dr. George
17 DR. GEORGE: I am not going to address
18 clinically whether it is, but it relates to what I
19 mentioned earlier about the--if it is true. Let's
20 just assume, for the moment, that 8 percent is the
21 true difference. I mean, really, if you give this
22 agent, you will get 8 percent less SREs in the time
23 frame specified. That is the main thing you are
24 looking at. We are not considering all the other
25 endpoints right now.
193
1 That means that you have to treat twelve
2 or so patients in order to get that one benefit.
3 So eleven are not getting any benefit, in that
4 sense. They might be getting something else if
5 there is some other benefit, but I didn't see any
6 other benefit.
7 So what you are getting is those eleven
8 patients or so were getting treated with something
9 that is not doing them any good. So, is it worth
10 it? I don't know regulatory--
11 DR. PAZDUR: Could I clarify something,
12 Donna? When you are talking about significance
13 here or substantial evidence, what you are really
14 talking about is is this of clinical importance or
15 clinical significance.
16 The way, really, the questions were raised
17 and what we were thinking of is more substantial
18 evidence of a drug effect. There are, really, two
19 kind of separate issues, perhaps. One is does it
20 make sense from a clinical point of view this two
21 months or whatever, X months, may be. That is a
22 clinical question.
23 The other one is substantial evidence, is
24 that reproducible, is it a reliable statistical
25 point of view. So there are two kind of different
194
1 issues come into play here, I think.
2 DR. NERENSTONE: Dr. Przepiorka?
3 DR. PRZEPIORKA: I was actually thinking
4 about that in the same way as well because,
5 clearly, if the effect is so marginal, a second
6 study may not show a benefit at all.
7 DR. NERENSTONE: The sponsor would like to
8 say something.
9 DR. SEAMAN: I think if you take into
10 consideration the 8 percent, that is probably
11 correctly, if you are just counting the first
12 event. But, if you wouldn't mind, let me show you
13 all the events, if you count all the events. The
14 only way you can look at that is in a multiple-event
15 analysis.
16 It is not in any of your slides, but let
17 me just give you an idea how many events occurred
18 in this patient population over the course of the
19 trial and just give you the numerical numbers.
20 May I have those slides?
21 If you remember back to Dr. Coleman's
22 slide where he showed the hundreds and hundreds of
23 events that took over the 24 month period. Let's
24 take a look at what we saw, too. That 8 percent is
25 just a first event, whether it be proportion or--
195
1 [Slide.]
2 This is for protocol 010. During the
3 course of that thirteen-month evaluation, there
4 were 808 SREs in terms of Zometa 4 and pamidronate
5 at 849. These are the types of things that took
6 place during that time frame. So it is
7 substantial. It is pretty similar to what we are
8 seeing. So we are just talking about an 8 percent
9 reduction.
10 Now, we don't have a placebo group in this
11 but let's look at the next set of slides for 011
12 and 039.
13 [Slide.]
14 You can see, again, here, that, over that
15 nine-month time frame, there are a number of these
16 events occurring in the placebo treatment group,
17 quite a few, 275, if you included hypercalcemia.
18 You were just counting that first event. In each
19 treatment group, you are having an impact. In the
20 treatment group, you are seeing an impact on that.
21 [Slide.]
22 Finally, in protocol 039, again you see
23 that there are a substantial number of events we
24 are not counting. So there are a whole host of
25 other events that we are taking into consideration.
196
1 If you look at the multiple-event analysis
2 in terms of protocol 011 and protocol 039, it is a
3 25 to 27 percent reduction in these events rates.
4 DR. NERENSTONE: I think, Donna, one of
5 your questions is more a more philosophical one,
6 which is how meaningful clinically is this drug. I
7 think no one who treats patients is going to say
8 this is a home run but it probably is a small
9 improvement.
10 Certainly, for those of us who treat a lot
11 of breast-cancer patients, showing that Zometa is
12 not worse--and you might want to debate that--but
13 is probably not worse than pamidronate, getting
14 patients out in fifteen minutes rather than two
15 hours definitely is going to add to their quality
16 of life and the patients will tell you that. And
17 that is a big deal.
18 So that study, in itself, I think, as much
19 as we don't like "me, too" drugs, does provide an
20 advantage, just on the basis of that alone.
21 MR. KAZMIERCZAK: I would agree with that.
22 Fifteen minutes instead of two hours certainly does
23 make a difference when you are a patient. I agree.
24 DR. NERENSTONE: Dr. Temple?
25 DR. TEMPLE: Of course, if the effect
197
1 isn't worth anything, you can avoid the fifteen
2 minutes and the two hours.
3 It is worth remembering that if you look
4 at the absolute percent difference--I mean, there
5 is a constant debate about this and Steve is trying
6 to do this, how many people do you have to treat to
7 benefit anybody. In some settings, hypertension
8 and things like that, we are accustomed to looking
9 at the percent reduction in bad events when the
10 actual reduction might be only 1 or 2 percent per
11 year. But that is because the events are
12 considered so very bad.
13 In these, I guess not all of the events
14 are really horrible, but some of them probably are.
15 So it is worth at least considering the hazard-ratio part of
16 it, too, which gives you some sense
17 of the relative reduction in the nasty events and
18 which shows a larger effect, obviously, than the
19 absolute reduction of roughly double.
20 DR. NERENSTONE: Dr. George?
21 DR. GEORGE: Just as follow up. Actually,
22 I don't know if it was addressed here. Maybe
23 someone knows it. Grant, I think, mentioned
24 something about it earlier. The endpoint, the SRE,
25 is a very heterogeneous collection of events, some
198
1 of which are treatment driven and some of which are
2 biologically based.
3 I have been just assuming that this is a
4 widely accepted endpoint in this area, but what is
5 the genesis of this endpoint?
6 DR. WILLIAMS: I am sure Dr. Seaman could
7 comment on this too, but, as I recall, the company
8 developed this endpoint for the Aredia trials and
9 certainly, we were involved. The basic idea was to
10 try to capture the very different kinds of
11 morbidity that seemed to be about of the same
12 significance. There was a big debate, do you put
13 hypercalcemia on here, and we steadfastly said no,
14 that this is something different, of a different
15 nature, that you could treat later.
16 I think that, as it happened, the bone
17 events ended up being a little different than we
18 expected. I think we were probably thinking about
19 long-bone fractures and it ended up being a lot of
20 compression fractures, of less significance,
21 perhaps, than we initially anticipated.
22 But I think we really have come to think
23 of the radiation therapy to bone to be a reasonable
24 sort of an integrator of what--a physician, I don't
25 think, just willy-nilly, goes out and puts the
199
1 patient in for a course of radiation therapy that
2 ended in a blinded trial where they don't know
3 which arm is getting which.
4 They are making that commitment usually
5 for something significant. So, even though they
6 are different in nature, we felt like they were
7 about of the same significance and I think we felt
8 comfortable over the past seven or eight years,
9 perhaps, with this event.
10 I don't know that it has been used
11 anywhere outside of the regulatory environment.
12 DR. NERENSTONE: Dr. George
13 DR. GEORGE: Along those lines, the
14 consistency of these studies, if you look at this
15 first page you gave us, if you do that pooling that
16 I was talking about across the treatments, it is
17 really remarkable that the proportion of patients
18 with an SRE, it would 36 percent in the treated
19 group pooled in the prostate-cancer study, in the
20 placebo, 44 percent.
21 In the solid tumors, the lung cancer and
22 others, exactly the same thing, with a different
23 time period but in different kinds of events and
24 patterns. But, overall, you get precisely the same
25 point estimates. That is with that 8 percent
200
1 effect.
2 DR. NERENSTONE: Dr. Bonomi?
3 DR. BONOMI: From a clinical perspective,
4 if you prevent one patient out of twelve from
5 having severe pain or from having a pathologic
6 fracture, I think that is very clinically relevant.
7 Even though we can't pick it up in the quality of
8 life, all the patients are in one pool, that really
9 has a huge impact on the quality of life, if you
10 have terrible pain. Certainly, if you have a
11 fracture, it is much worse.
12 DR. NERENSTONE: Dr. Raghavan?
13 DR. RAGHAVAN: I would like to change
14 gears just a little bit. It comes back to Matt
15 Smith's late response to my question. You know,
16 one of the things that I think is important--because one
17 senses there is a consensus around the
18 table that these are relatively straightforward
19 decisions. It is the fine tuning that we are
20 spending the time on.
21 So I, with that in mind, took the liberty
22 of looking at the package insert. It does strike
23 me that, in the context of prostate cancer, for
24 someone who does see a lot of prostate cancer and
25 bad prostate cancer that is aggressive, as a
201
1 clinician, I am not sure that I know where this
2 drug should find its place.
3 I think everything that has been said
4 about reducing the pain for patients is true. I
5 think it is good to do that, but I am not sure that
6 what I have heard today will tell me how to apply
7 it in clinical practice. So, if it does happen
8 that this gets approval, I would say to the FDA,
9 you want to spend some time with the company
10 looking at the development of the package insert
11 and getting the wording right.
12 We know that people tend to telescope from
13 meetings such as this. I could imagine that there
14 could be tens of thousands of patients who have
15 radical prostatectomies who go on this straight
16 away "to prevent bone complications."
17 We know that people use drugs in strange
18 ways. Now, even within the context of what we have
19 heard, I am absolutely convinced that a very good
20 thing to be is--I mean, it is not a good thing to
21 be stuck with prostate cancer, but it seems to get
22 into this trial is a good thing because every arm
23 has good survival.
24 I think what that really tells us is it is
25 a very heterogenous group. In the real world, the
202
1 median survival for prostate cancer is short.
2 Skeletal-related events occur more commonly than
3 even in the placebo-treated group here. It
4 probably means the doctors have managed them well.
5 I don't take that away from the treating team, but
6 I do think there is case selection, as I said
7 before.
8 Median weight seems to be--well, half the
9 patients have a weight between 80 and 130
10 kilograms. That is pretty big for a patient with
11 prostate cancer. There is not a lot of anemia.
12 There is not a lot of hydronephrosis. There are
13 not a lot of the things that you would expect to
14 see. So what I think we have got in the definition
15 of patients going into the trial is a mix of people
16 who have had a bit of chemotherapy for a range of
17 reasons.
18 People are on hormones, off hormones, and
19 so on. So, while I don't think that detracts from
20 the importance of the product, I think the FDA is
21 going to need to spend some time on the package
22 insert. It may well behoove the company to look at
23 sets of patients like those who present with more
24 significant symptoms than require codeine only.
25 The vast majority of patients with
203
1 prostate cancer and bone metastases have severe
2 pain and major problems. So I think they have
3 picked a good group, which is fine, but in the
4 further development of the drug, it might be nice
5 to encourage them to look at some of the tougher
6 cases.
7 DR. NERENSTONE: Dr. Loeher?
8 DR. LOEHER: I just want to echo what
9 Derek said. I mean, a nice clean study which would
10 have answered a lot of questions is primary
11 treatment of hormonal-sensitive patients to do
12 orchi-activity or do LHR-antagonist with or without
13 this. Then you would be able to get removal out of
14 these variables and find out exactly how to do
15 that. So I would encourage the company to pursue
16 that.
17 The other question I had, I guess along
18 that line, and picking up on what Donna said, is
19 what is the subset of patients that are going to
20 benefit. Do we need to treat all of these people
21 or can we identify them? The question I have, and
22 I don't understand curves and stuff, but you had
23 this urinary telopeptide-creatinine ratio. Is this
24 useful in terms of predicting who will respond and
25 who will not respond in terms of outcome?
204
1 DR. SEAMAN: I will have Rob Coleman
2 answer that who is very familiar with a lot of the
3 things that have been done over the years and it is
4 somewhat in its infancy.
5 DR. COLEMAN: Rob Coleman. Most of the
6 data with N-telopeptides is actually from the
7 breast-cancer literature. It is a small series so
8 it is difficult to add a lot of weight to it, but
9 it seems that what we should be trying to do is
10 normalize bone resorption. The way we measure that
11 is, obviously, with one of these bone markers.
12 There are some patients who you won't
13 normalize bone resorption even with a
14 bisphosphonate for reasons that are not very well
15 explained. So I think we do have to look at trying
16 to use the markers to pick out those patients who
17 are unresponsive because, clearly, bisphosphonates
18 are not a panacea. They are an improvement but
19 they are not a panacea.
20 Trying to use the markers to predict who
21 is going to respond in the first place--for
22 instance, if you have absolutely normal bone
23 resorption, why add in the bone-resorption-inhibitor on top?
24 So there may be people who you
25 don't need to treat or there may be people who are
205
1 resistant and our hope is that the markers will
2 sort this out
3 Obviously, there is as huge database of
4 information that will be analyzed over the coming
5 months to try and dissect that out, but that is the
6 theory behind it. But there isn't a sort of
7 internationally approved set of response criteria
8 that we could use as of today.
9 DR. NERENSTONE: Other comments from the
10 committee? Dr. Kelsen?
11 DR. KELSEN: No; I have no more comments
12 at this point.
13 DR. NERENSTONE: Dr. Albain?
14 DR. ALBAIN: Yes. I wondered if we could
15 have a little bit of discussion about the labeling
16 in renal function and what--
17 DR. NERENSTONE: We are losing you, Kathy.
18 Hold on just a minute. Try again.
19 DR. ALBAIN: I wondered if we could have a
20 little discussion about the labeling and the
21 pretreatment renal function as well as monitoring.
22 Are we going to recommend that this not be given
23 for creatinine clearances less than 30 or could we
24 have a little more discussion about that?
25 DR. WILLIAMS: We are having discussions
206
1 internally on that matter. I don't think we have
2 made up our mind what to recommend. Certainly,
3 this is renally excreted and there seems to be a
4 relationship between the creatinine and toxicity
5 and AUC and toxicity. But whether or not you would
6 recommend treatment outside of the range of the
7 study, that is still a point for debate, I think.
8 We probably won't have that debate here.
9 DR. NERENSTONE: Any other questions, Dr.
10 Albain?
11 DR. ALBAIN: Just thinking, when we give
12 pamidronate, as a rule, in practice, we don't
13 routinely--or many places do not get a serum
14 creatinine before each dose. What we have seen
15 today would seem to indicate that you probably
16 should as much as you should with this agent.
17 Is that going to be a strong suggestion
18 that it be done? It was alluded to during the
19 presentations. How necessary is that?
20 DR. SCHER: I think you are right that
21 initially the nephrotoxic nature of pamidronate was
22 not that obvious as imparted to us clinicians,
23 which it was at the time. I can't make a comment
24 on that label at this point but, as Grant said, the
25 exact labeling of this drug will be under
207
1 discussion. Clearly, we will be recommending, if
2 Zometa is approved, that creatinine be closely
3 monitored and there will be some comment on level
4 of renal function. But that has to be discussed
5 further.
6 DR. ALBAIN: Just to follow that up a
7 little bit further because these patients routinely
8 will be just coming in including visits for the
9 agent without physician input on each and every
10 dose. If we follow the guidelines of the clinical
11 trial for Zometa, it is a lot more involved.
12 You hold it. You wait until it decreases
13 to a certain fraction, et cetera. So it may change
14 the national practice standards of how we give
15 these agents based on what you decide.
16 DR. NERENSTONE: To follow that up, are
17 you going to recommend the vitamin D and calcium
18 which, I guess, all patients on these studies go on
19 and was sort of news to me.
20 DR. WILLIAMS: We generally do put in the
21 label the way it is done in the study. We have a
22 noninferiority study which seemed very similar with
23 the amount of toxicity to Aredia. We just happen
24 to have the sponsor for Aredia here in the same
25 room. I would think that we would generally want
208
1 that label to be pretty similar regarding this.
2 DR. NERENSTONE: Dr. Kelsen, did you have
3 any questions?
4 DR. KELSEN: No additional questions.
5 DR. NERENSTONE: Dr. Taylor, do you have
6 any questions?
7 DR. TAYLOR: Not right now.
8 DR. NERENSTONE: I will take that as a no.
9 Dr. Lippman?
10 DR. LIPPMAN: I just wanted to bring up
11 again, just to make sure we brought some closure,
12 albeit with, maybe, not the perfect answers, but
13 FDA, Bob, everyone raised the issue of the blastic
14 versus lytic. It came up again in Pat's questions
15 about can we dissect who would respond or not.
16 Since one might have thought that the biggest
17 predictor might have been whether it is blastic or
18 lytic and that doesn't seem to be the case, is
19 there any more discussion? I think Bob left it
20 with, well, you can always come up with biologic
21 plausibility.
22 I think we have heard one mechanism. We
23 haven't heard any others that would dispute it and
24 we see results in the two tables, Grant, that you
25 put on this page which show very similar results in
209
1 prostate and the others.
2 So I don't know that we have any
3 information or anything else to say that it
4 shouldn't be used in both. I just don't know.
5 Have we resolved that? Because it is one of the
6 major questions that you raised and one of the
7 questions on the sheet. Is there anything that you
8 want to discuss?
9 DR. WILLIAMS: You have heard the
10 evidence, preclinical and clinical. It is your
11 discussion.
12 DR. PAZDUR: Perhaps you will resolve it
13 with your vote.
14 DR. NERENSTONE: Dr. George
15 DR. GEORGE: Along those lines, one other
16 issue I haven't completely resolved in my own mind
17 is the myeloma data with respect to the first study
18 because the main thing that concerns me about that
19 still is that difference with the Aredia effect--not the
20 effect because we don't have a placebo, but
21 the Aredia results--in the new study and the old.
22 It is sort of indirect evidence that there is not a
23 constancy here but--maybe very indirect, but there
24 were some prognostic factors that seemed to be
25 different.
210
1 But I am still a little worried about
2 that. After saying I favor pooling generally, now
3 I am sort of backtracking a little bit, got a
4 little worried about the fact that breast cancer
5 seems very clear, clear-cut. That is nice. And
6 the myeloma, I am wondering whether that evidence
7 is really there strong enough.
8 [Slide.]
9 The only thing I could think to do was to
10 try to find patients that were similar within the
11 old trial and see what estimate effect we had. It
12 seemed to be, if anything, greater. So that is all
13 I could think to do to examine that question. It
14 seemed to be clear that they were a more aggressive
15 disease and one could wonder is it so aggressive
16 that we are going to have no effect on it.
17 But, in looking back at similar patients
18 with what appeared to be more aggressive disease,
19 earlier time since diagnosis, the effect seemed, if
20 anything, more, the Aredia versus placebo effect.
21 So I think that is about all we could do.
22 DR. NERENSTONE: Dr. Raghavan?
23 DR. RAGHAVAN: I would like to respond to
24 Scott Lippman's question. I think that it is only
25 one trial but the differences are actually quite
211
1 compelling, not withstanding the caveats that I put
2 in, in prostate cancer. That really is an
3 absolutely blastic-metastasis-dominated disease. I
4 think they showed it somewhere in the stuff we have
5 looked at today that there were a tiny proportion
6 of mixed sclerotic and lytic.
7 But, for practical purposes, prostate
8 cancer, in the board questions, comes up. You see
9 a picture of blastic metastasis. So I was pretty
10 impressed with the prostate as a paradigm of
11 blastic disease so it didn't leave me feeling too
12 uncomfortable about that issue.
13 DR. NERENSTONE: Dr. Lippman?
14 DR. LIPPMAN: I think that was sort of the
15 point, that we have sort of the biologic
16 plausibility, the only mechanism that we have. And
17 we have consistent clinical results that also
18 support efficacy. So, again, I don't know that we
19 have anything else to resolve it, but there is no
20 other data to have to suggest that isn't active in
21 that setting.
22 DR. NERENSTONE: Mr. Kazmierczak?
23 MR. KAZMIERCZAK: I guess I would have
24 been a participant in your clinical trial if I
25 could because I am a patient with a rising PSA
212
1 right now. I hope I am one of the one in twelve.
2 One in twelve is certainly better than none in
3 twelve. Thank you.
4 DR. NERENSTONE: If we could turn now to
5 the questions. We have all seen the first page. I
6 am not going to go over that.
7 For new drug approval, substantial
8 evidence of efficacy from adequate and well-controlled
9 investigations is required. Evidence
10 from multiple clinical trials is usually submitted
11 but robust results from a single multicenter trial
12 have been accepted.
13 We are going to consider whether the
14 results from trials fulfill the regulatory
15 requirement. So the first study we are going to
16 look at is study 010 in breast cancer and myeloma.
17 In that study, 44 percent of Aredia patients had an
18 SRE on study versus 46 percent of Zometa patients.
19 Using the conservative two-95-percent-confidence-
20 interval method, the FDA calculates that
21 Zometa retains at least 49 percent of Aredia's
22 efficacy demonstrated historically in comparison to
23 placebo.
24 The first question, do other studies, 011
25 and 039, provide supportive evidence for Zometa's
213
1 efficacy in breast cancer and myeloma.
2 DR. SEAMAN: The numbers are reversed, 44
3 for Zometa, for Aredia 46 percent.
4 DR. NERENSTONE: Okay. Thank you. So it
5 is 46 percent of the Aredia and 44 percent for
6 Zometa. I am going to have to see a show of hands
7 and then I will ask for our participants at the
8 remote locations by name. So, first we are going
9 to go around the table. Do the other studies
10 provide supportive evidence for Zometa's efficacy
11 in breast cancer and myeloma?
12 Dr. Kelsen?
13 DR. KELSEN: Yes.
14 DR. NERENSTONE: Dr. Albain?
15 DR. ALBAIN: Yes.
16 DR. NERENSTONE: Dr. Taylor?
17 DR. TAYLOR: Yes.
18 DR. NERENSTONE: Dr. Raghavan?
19 DR. RAGHAVAN: Yes.
20 DR. NERENSTONE: Dr. George
21 DR. GEORGE: Yes.
22 DR. LIPPMAN: Yes.
23 MR. KAZMIERCZAK: Not being a clinician, I
24 will abstain.
25 DR. PRZEPIORKA: Yes.
214
1 DR. NERENSTONE: Yes.
2 DR. BRAWLEY: Yes.
3 DR. PELUSI: Yes.
4 DR. NERENSTONE: It is ten yes and one
5 abstention.
6 Part b; is there substantial evidence from
7 adequate and well-controlled investigations of
8 Zometa, 4 milligrams, efficacy in breast cancer and
9 myeloma?
10 Dr. Kelsen?
11 DR. KELSEN: Yes.
12 DR. NERENSTONE: I'm sorry; first of all,
13 are there any other comments from the committee?
14 Okay.
15 Dr. Kelsen?
16 DR. KELSEN: Yes.
17 DR. NERENSTONE: Dr. Albain?
18 DR. ALBAIN: Yes.
19 DR. NERENSTONE: Dr. Taylor?
20 DR. TAYLOR: Yes.
21 DR. NERENSTONE: Dr. Raghavan?
22 DR. RAGHAVAN: Yes.
23 DR. NERENSTONE: Dr. George
24 DR. GEORGE: Yes.
25 DR. NERENSTONE: You can now go around the
215
1 table. I don't have to call your name.
2 DR. LIPPMAN: Yes.
3 MR. KAZMIERCZAK: Yes.
4 DR. PRZEPIORKA: Yes.
5 DR. NERENSTONE: Yes.
6 DR. BRAWLEY: Yes.
7 DR. PELUSI: Yes.
8 DR. NERENSTONE: Eleven yes.
9 The next page. Study 039 in prostate
10 cancer. Zometa studies 010 and 011 have evaluated
11 Zometa efficacy in predominantly lytic metastases.
12 Can results from these studies provide supportive
13 evidence for Zometa's efficacy in prostate cancer
14 which produces predominantly blastic bone
15 metastases? Further comments from the committee?
16 I think we have talked about this.
17 Dr. Kelsen, do you want to start us?
18 DR. KELSEN: Yes; I will say yes.
19 DR. NERENSTONE: Dr. Albain?
20 DR. ALBAIN: Yes.
21 DR. NERENSTONE: Dr. Taylor?
22 DR. TAYLOR: Yes.
23 DR. RAGHAVAN: Yes.
24 DR. GEORGE: Yes.
25 DR. LIPPMAN: Yes.
216
1 MR. KAZMIERCZAK: Yes.
2 DR. PRZEPIORKA: Yes.
3 DR. NERENSTONE: Yes.
4 DR. BRAWLEY: Yes.
5 DR. PELUSI: Yes.
6 DR. NERENSTONE: Eleven yes.
7 Part b; is there substantial evidence of
8 Zometa, 4 milligrams, efficacy in prostate cancer
9 from adequate and well-controlled investigations?
10 Any further comments?
11 Dr. Kelsen, you are doing such a good job.
12 Would you like to start?
13 DR. KELSEN: Yes.
14 DR. NERENSTONE: Dr. Albain?
15 DR. ALBAIN: Yes.
16 DR. NERENSTONE: Dr. Taylor?
17 DR. TAYLOR: Yes.
18 DR. RAGHAVAN: Yes.
19 DR. GEORGE: Yes.
20 DR. LIPPMAN: Yes.
21 MR. KAZMIERCZAK: Yes.
22 DR. PRZEPIORKA: Unfortunately, I have to
23 say no. The reason I have to say no is because, as
24 was pointed out, there may be a difference between
25 the patient groups because of the survival
217
1 difference in the Zometa, 4-milligram, dose so I am
2 not sure I could actually draw this conclusion as
3 much as I would like to.
4 DR. NERENSTONE: Yes.
5 DR. BRAWLEY: For 4 milligrams, yes.
6 DR. PELUSI: Yes.
7 DR. NERENSTONE: The vote is ten yes and
8 one no.
9 For study 011 in other solid tumors.
10 Analysis from both the 4-milligram and 8-milligram
11 Zometa arms of study 011 support the efficacy of
12 Zometa. Do you agree with FDA that these results
13 provide substantial evidence of Zometa, 4
14 milligrams, efficacy in the population studied?
15 Any comments?
16 Dr. Kelsen?
17 DR. KELSEN: Yes.
18 DR. NERENSTONE: Dr. Albain?
19 DR. ALBAIN: Yes.
20 DR. NERENSTONE: Dr. Taylor?
21 DR. TAYLOR: Yes.
22 DR. RAGHAVAN: Yes.
23 DR. GEORGE: Yes.
24 DR. LIPPMAN: Yes.
25 MR. KAZMIERCZAK: I abstain.
218
1 DR. PRZEPIORKA: Yes.
2 DR. NERENSTONE: Yes.
3 DR. BRAWLEY: Yes.
4 DR. PELUSI: Yes.
5 DR. NERENSTONE: Ten yes and one
6 abstention.
7 The sponsor's proposed indication includes
8 "treatment of osteolytic, osteoblastic and mixed
9 bone metastases of solid tumors." This indication
10 infers treatment as indicated for patients with
11 bone metastases from all solid tumors irrespective
12 of the primary tumor. Do you agree with this
13 proposed indication?
14 Further comments? Dr. Loehrer?
15 DR. LOEHRER: I have some concerns about
16 how this may be used for prostate cancer, patients
17 who are not hormone refractory who have bone mets
18 may get primary hormonal therapy and do extremely
19 well. The way this indication is listed, it goes
20 beyond what the studies were designed.
21 DR. WILLIAMS: This is actually referring
22 just to the other solid-tumor group, but we will
23 take that into consideration when we are thinking
24 about the prostate indication.
25 DR. NERENSTONE: Other comments? For the
219
1 vote, Dr. Kelsen?
2 DR. KELSEN: Yes.
3 DR. NERENSTONE: Dr. Albain?
4 DR. ALBAIN: Yes.
5 DR. NERENSTONE: Dr. Taylor?
6 DR. TAYLOR: Yes.
7 DR. RAGHAVAN: Yes.
8 DR. GEORGE: Yes.
9 DR. LIPPMAN: Yes.
10 MR. KAZMIERCZAK: Yes.
11 DR. PRZEPIORKA: Yes.
12 DR. NERENSTONE: Yes.
13 DR. BRAWLEY: Yes.
14 DR. PELUSI: Yes.
15 DR. NERENSTONE: Eleven yes.
16 There is a last sentence. Please provide
17 suggestions for wording of the indication section
18 or the clinical-trials section of the Zometa
19 labeling with regard to this issue.
20 Do you have enough discussion now?
21 DR. WILLIAMS: Yes. It wasn't clear where
22 the discussion would go, whether you were going to
23 say, "all solid tumors except," this or that. But
24 I think you have addressed that.
25 DR. NERENSTONE: I would like to thank
220
1 everybody for their attendance. We will meet again
2 at the end of February, February 27.
3 Thank you.
4 DR. TEMPLETON-SOMERS: Thank you Kathy,
5 Sarah and David for pioneering for us. Thanks.
6 [Whereupon, at 2:00 p.m., the meeting was
7 adjourned.]