file:///C|/Storage/1121orth.

txt

1 DEPARTMENT OF HEALTH AND HUMAN SERVICES FOOD AND DRUG ADMINISTRATION CENTER FOR DEVICES AND RADIOLOGICAL HEALTH

MEETING OF THE ORTHOPEDICS AND REHABILITATION DEVICES ADVISORY PANEL P000054 INDUCTOS (RHBMP-2/ACS) FOR ORTHOPEDIC TRAUMA WYETH RESEARCH Thursday, November 21, 2002 9:30 a.m.

Walker/Whetstone Room Gaithersburg Holiday Inn 2 Montgomery Avenue Gaithersburg, Maryland

file:///C|/Storage/1121orth.txt (1 of 309) [12/2/02 8:59:10 AM]

file:///C|/Storage/1121orth.txt

2 C O N T E N T S PAGE MORNING SESSION: Call to Order Dr. Michael J. Yaszemski, Chairperson Open Public Hearing (No Speakers) Sponsor Presentation Dr. F. Owen Fields, Worldwide Regulatory Affairs, Wyeth Research Dr. Rod Riedel (Preclinical Data) Dr. Alex Valentin (Clinical Data) 8 9 4

12 19 25

Dr. Marc F. Swiontkowski (Relevance to U.S. Practice) FDA Presentation Mr. Aric Kaiser (PMA Lead Reviewer) Dr. Barbara Buch (Preclinical) Dr. Chang S. Lao (Statistical) Panel Presentations Dr. Sanjiv Naidu (Preclinical) Dr. Maureen Finnegan (Clinical) Dr. Kinley Larntz (Statistical) Open Public Hearing Bill Christiansen, on behalf of Orthopedic Surgical Manufacturers Association (OSMA) AFTERNOON SESSION Panel Questions and General Discussion Final Sponsor Comments Vote

58

74 78 109

120 127 139

149

156 208 210

file:///C|/Storage/1121orth.txt (2 of 309) [12/2/02 8:59:10 AM]

file:///C|/Storage/1121orth.txt

3 C O N T E N T S (Continued) General Panel Discussion of Spinal Devices Open Public Hearing: Brenda Seidman (Seidman Toxicology Services) Diane Johnson (Medtronic Spinal Dynamics) Dr. Bailey Lipscomb (Medtronic Sophomore Danik) Dr. Rich Jansen (Disc Dynamics, Incorporated) Dr. Britt Norton (Raymedica, Inc.) FDA Presentation Dr. Barbara Buch Panel Reviews Dr. John Kirkpatrick (Clinical/Preclinical) Dr. John Doull (Preclinical Toxicology/Safety) Discussion of FDA Questions to Panel Adjourn 268 279 281 309 253 233 238 240 244 248 232

file:///C|/Storage/1121orth.txt (3 of 309) [12/2/02 8:59:10 AM]

file:///C|/Storage/1121orth.txt

4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 P R O C E E D I N G S DR. YASZEMSKI: Good morning, everybody.

May I ask that everybody take their seats, and we'll go ahead and get started. Welcome. MR. DEMIAN: like to welcome you. We are ready to begin this meeting of the Orthopedic and Rehabilitation Devices Panel. My name is Hany Demian, and I am the Executive Secretary for this panel. I would like to remind everyone that you are requested to sign in on the attendance sheets which are available outside the doors. You may Good morning. I would first

pick up an agenda and information about today's meeting, including how to find out about future meeting dates through the Advisory Panel phone line and how to obtain meeting minutes and transcripts. I will now read two statements that are required to be read into the record. The first one

is an Appointment to Temporary Voting Status Statement, and the second one is the Conflict of Interest Statement. "Appointment to Temporary Voting Status. Pursuant to the authority granted under the Medical

file:///C|/Storage/1121orth.txt (4 of 309) [12/2/02 8:59:10 AM]

file:///C|/Storage/1121orth.txt

5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 21, 2002." "The following announcement addresses conflict of interest issues associated with this meeting and is made part of the record to preclude even the appearance of any impropriety. To Device Advisory Committee Charter dated October 27, 1990 and as amended August 18, 1999, I appoint the following individuals as voting members of the Orthopedic and Rehabilitation Devices Panel for this meeting on November 21, 2002: John Doull, and Albert Aboulafia. Andrew Schmidt, For the record,

these individuals are Special Government Employees and consultants to this panel or other panels under the Medical Device Advisory Committee. They have

undergone the customary conflict of interest review and have reviewed the material to be considered at this meeting." This is signed by the Director of the Center for Devices and Radiological Health, Dr. David Feigal. "Conflict of Interest Statement, November

determine if any conflict existed, the Agency reviewed the submitted agenda for this meeting and all financial interests reported by the Committee's participants. The conflict of interest statutes

file:///C|/Storage/1121orth.txt (5 of 309) [12/2/02 8:59:10 AM]

file:///C|/Storage/1121orth.txt

6 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 prohibit Special Government Employees from participating in matters that could affect their or their employers' financial interest. However, the

Agency has determined that the participation of certain members and consultants, the need for whose services outweigh the potential conflict of interest involved, is in the best interest of the Government." "Therefore, waivers have been granted to Drs. Albert Aboulafia, John Kirkpatrick, Kinley Larntz, and Andrew Schmidt for their interest in firms that could potentially be affected by the panel's recommendation. The waivers allow them to

participate fully in today's deliberations." "Dr. Aboulafia's waiver involves a consulting arrangement with a competing technology firm. For this unrelated consulting service he

receives less than $10,001 a year." "Dr. Kirkpatrick's waiver involves stockholdings valued between $5,001 and $50,000 in competing firms and the parent company of several competing manufacturers." "Dr. Larntz' waiver involves a consulting agreement with a competing technology firm. For

his consulting services, he receives less than

file:///C|/Storage/1121orth.txt (6 of 309) [12/2/02 8:59:10 AM]

file:///C|/Storage/1121orth.txt

7 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 $10,001." "Dr. Andrew Schmidt's waiver involves a consulting arrangement with a competing technology firm. For this consulting service, he receives

less than $10,001 a year." "Copies of these waivers may be obtained from the Agency's Freedom of Information Office, Room 12A-15 of the Parklawn Building." "We would like to note for the record that the Agency took into consideration other matters regarding Drs. Maureen Finnegan, John Kirkpatrick, Andrew Schmidt, Michael Yaszemski, and Ms. Karen Rue. They reported interest in firms at issue but The

in matters not related to today's agenda.

Agency has determined therefore that they may participate fully in all discussions. In the event

that the discussions involve any other products or firms not related to today's agenda for which an FDA participant has a financial interest, the participant should excuse him or herself from such involvement, and the exclusion will be noted for the record." "With respect to all other participants, we ask in the interest of fairness that all persons making statements or presentations disclose any

file:///C|/Storage/1121orth.txt (7 of 309) [12/2/02 8:59:10 AM]

file:///C|/Storage/1121orth.txt

8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 current or previous financial involvement with any firm whose product they may wish to comment upon." Before turning this meeting over to Dr. Michael Yaszemski, I would like to introduce our distinguished panel members who have generously given their time to help FDA in matters being discussed today and other FDA staff seated at the table, so we'll just go around the room and introduce ourselves, our affiliation, and our areas of interest. Mike? DR. YASZEMSKI: Rochester, Minnesota. Michael Yaszemski,

I am an orthopedic surgeon,

and on the research side, I am a chemical engineer. DR. NAIDU: College of Medicine. Sanjiv Naidu, Penn State I am an orthopedic surgeon

and a materials scientist. DR. LARNTZ: Kinley Larntz, Professor

Emeritus, University of Minnesota School of Statistics. I also work as an independent

statistical consultant, and my interest is in biostatistics, clinical design. DR. SCHMIDT: Minneapolis. of Minnesota. Dr. Andrew Schmidt from

I am on the faculty of the University I practice primarily in trauma and

file:///C|/Storage/1121orth.txt (8 of 309) [12/2/02 8:59:10 AM]

file:///C|/Storage/1121orth.txt

9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Aboulafia. adult reconstructive surgery, and I am interested in biologic applications to improve the human factors. DR. ABOULAFIA: My name is Albert

I am an orthopedic oncologist at Sinai

Hospital in Baltimore and the University of Maryland. MS. WITTEN: Celia Witten. I am with the

Food and Drug Administration.

I am the Division

Director of the division reviewing this product, DGRND. MS. MAHER: Sally Maher. I am the Senior

Director of Regulatory and Clinical, Smith & Nephew Endoscopy, and I am the industry rep. MS. RUE: Karen Rue. I am a registered

nurse at Lafayette General Medical Center and Acadian Health Care Alliance, and I am the consumer representative. DR. DOULL: John Doull. I am a clinical

toxicologist from the University of Kansas Medical Center. DR. KIRKPATRICK: John Kirkpatrick,

Associate Professor of Orthopedic Surgery at the University of Alabama Birmingham. DR. FINNEGAN: Maureen Finnegan, at UT

file:///C|/Storage/1121orth.txt (9 of 309) [12/2/02 8:59:10 AM]

file:///C|/Storage/1121orth.txt

10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Southwestern in Dallas, with a research interest in fracture healing. MR. DEMIAN: Thank you.

At this time, I would like to turn the meeting over to our chairman, Dr. Michael Yaszemski. DR. YASZEMSKI: good morning again. I am Dr. Michael Yaszemski, and I will be the chair for this meeting. Today, we the panel will be making recommendations to the Food and Drug Administration regarding a Pre-Market Approval application for a growth factor soaked in collagen sponge to treat tibial fractures. I would like to note for the record that the voting members present constitute a quorum as required by 21 CFR Part 14. We will now proceed with the open public hearing session of the meeting. I would ask at Thanks, everybody, and

this time that all persons addressing the panel come forward and speak clearly into the microphone, as the transcriptionist is dependent on this means of providing an accurate record of the meeting. And I will ask as a protocol for

file:///C|/Storage/1121orth.txt (10 of 309) [12/2/02 8:59:10 AM]

file:///C|/Storage/1121orth.txt

11 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 today--and please bear with me; I know as the discussion goes back and forth, we frequently forget--but I will apologize in advance for interrupting you if you don't introduce yourself so we can have our transcriptionist know who everybody is when they talk. We are requesting that all persons making statements during the open public hearing of the meeting disclose whether they have financial interest in a medical device company. Before

making your presentation to the panel, please state your name, your affiliation, and the nature of your financial interest, if any. At this time, is there anyone wishing to address the panel? [No response.] DR. YASZEMSKI: asked to be scheduled. here? [No response.] DR. YASZEMSKI: We will now proceed to There was a person who Mr. Christiansen, are you

consider the Pre-Market Approval application for Wyeth Pharmaceuticals, Incorporated InductOs rhBMP-2 Absorbable Collagen Sponge. I would like to remind the public

file:///C|/Storage/1121orth.txt (11 of 309) [12/2/02 8:59:10 AM]

file:///C|/Storage/1121orth.txt

12 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 observers at this meeting that while this portion of the meeting is open to public observation, public attendees may not participate except at the specific request of the panel. We are now ready to begin with the sponsor's presentation. presentation by FDA. We'll follow that with the

I would like to ask that each

speaker state his/her name and affiliation with the firm before beginning the presentation. Wyeth, please. Sponsor Presentation DR. FIELDS: Owen Fields. Good morning. My name is

I am in Regulatory Affairs at Wyeth.

I would like to thank FDA for this opportunity to present before the panel, and I would also like to thank the panel members for their time and attention this morning. [Slide.] I will begin by describing the product. The product is a combination of a biologically active protein with a matrix that serves to deliver the protein to its site of action. The protein is

known as recombinant human bone morphogenetic protein-2. Understandably, this is often

abbreviated as RHBMP-2 or simply as BMP-2.

file:///C|/Storage/1121orth.txt (12 of 309) [12/2/02 8:59:11 AM]

file:///C|/Storage/1121orth.txt

13 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 protein. BMP-2 is an osteoinductive protein, that is, it induces bone formation. And the protein is

produced by what are now fairly standard biological production techniques. [Slide.] There is a matrix used to deliver the The matrix used to deliver the protein is This product is It was

an absorbable collagen sponge.

manufactured by Integra Life Sciences.

originally approved in 1981 as an implantable hemostatic agent, and when it is placed to the use of the hemostatic agent, it is known by the trade name Helistat. [Slide.] To prepare the BMP-2/ACS product, the vial of protein is first reconstituted with water for injection, and this solution is then applied to the absorbable collagen sponge. For the use that we

propose, the combined product is then surgically implanted at the fracture site in this form. Two strengths of BMP-2,.75 and 1.5 mg/ml, have been clinically tested, as you have read in your packages, but only the 1.5 mg dosage form is proposed for marketing, and this is due to the clinical results and the differentiation between

file:///C|/Storage/1121orth.txt (13 of 309) [12/2/02 8:59:11 AM]

file:///C|/Storage/1121orth.txt

14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 the efficacy of the two doses which you will hear some detail about in subsequent presentations. [Slide.] The BMP-2/ACS product is what is known as a "combination product" in that it shares characteristics with both biologicals and devices. It is like a device, and then it is surgically implanted. It is generally used only for a single

application, and it acts only at the site of implantation. However, it is also like a biological or drug product in that we had to establish the proper delivery context and the proper dose, and in that it has pharmacological activity. As a result of this, both drug and device development paradigms had to be applied to the product, and in fact globally, the product is regulated as both, specifically. In the U.S., the

product is regulated as a device, whereas in Europe, the product is actually regulated as a drug. And as you will see, we conducted a preclinical program involving both drug and device studies whereas we conducted a clinical program more similar to that of the device. And as you

file:///C|/Storage/1121orth.txt (14 of 309) [12/2/02 8:59:11 AM]

file:///C|/Storage/1121orth.txt

15 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 might imagine, the manufacturing characterization and work that was done was very typical of a biological product and in fact was fully reviewed by the Center for Biologics. [Slide.] The BMP-2/ACS product has been simultaneously developed in several medical areas. Starting on the left, the product has been developed for spine fusion in combination with a spine fusion cage. Our commercial partner,

Medtronic Sophomore Danik, has carried out this development, and the BMP-2/ACS product when it is used for this indication is known by the trade name INFUSE. In addition, the product has been developed by Wyeth for use in orthopedic trauma, specifically for long-bone fractures. The product

when it is placed for this use is known by the trade name InductOs. It is important to note here that the rhBMP-2 and the ACS for all of these uses are biochemically identical, and the ratio at which these two components are combined are also identical. [Slide.]

file:///C|/Storage/1121orth.txt (15 of 309) [12/2/02 8:59:11 AM]

file:///C|/Storage/1121orth.txt

16 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Products containing BMP-2/ACS have previously been approved in two different contexts globally. Specifically, in January 2002, this

panel recommended approval of the INFUSE/LT-CAGE product, which is a combination of BMP-2/ACS with an already approved spine fusion cage. This PMA

was sponsored by Medtronic Sophomore Danik, our corporate partner. The spine fusion product was formally approved by FDA in July of this year, and given this, in our presentation, we will not focus on the manufacturer of BMP-2 and ACS; we will also not focus on the preclinical safety program in that the preclinical safety program supports both uses of the product and was discussed in some detail at the January 2002 panel meeting. [Slide.] BMP-2/ACS, that is, the product without the fusion cage for the use that we propose, has also been approved in the European Union for orthopedic trauma. Specifically InductOs was

approved as a drug in the European Union in September of this year. Wyeth was the sponsor of

the medicines application that allowed this approval.

file:///C|/Storage/1121orth.txt (16 of 309) [12/2/02 8:59:12 AM]

file:///C|/Storage/1121orth.txt

17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 As is typical for drug approvals, this approval included some post-approval commitments. First, we have preclinical post-approval commitments, which are essentially the same as some of those that have been made for the INFUSE product in the U.S., so the same set of studies that were subject to post-approval commitments for the INFUSE product are also post-approval commitments in the European approval. [Slide.] Second, we will also be conducting a tibia fracture clinical trial to expand experience with the product when it is used with reamed IM nails. Obviously, the approval of InductOs for the orthopedic trauma indication in the U.S. is the focus of the remainder of our presentation. [Slide.] The proposed orthopedic trauma indication is as follows. InductOs is indicated for the

treatment of acute long-bone fractures that require open surgical management. InductOs increases the probability of fracture healing, accelerates fracture healing, and decreases the frequency and invasiveness of interventions for delayed union or nonunion.

file:///C|/Storage/1121orth.txt (17 of 309) [12/2/02 8:59:12 AM]

file:///C|/Storage/1121orth.txt

18 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 The mechanistic rationale for this indication is fairly simple to understand. We

believe that the data indicate that osteoinduction at the fracture site improves the probably and speed of healing and reduces the need for secondary interventions. [Slide.] FDA has placed some specific questions before the panel, and our objective today is to address these questions. We believe we can address

each and every question that has been raised by FDA. Toward this end, I will be followed by three

speakers. First, you will hear a brief review of the relevant preclinical data from Dr. Rod Riedel.

Second, Dr. Alex Valentin will review the clinical data available. Third, Dr. Marc Swiontkowski will review the relevance of the clinical data to U.S. practice. And finally, later today, we will address any questions that the panel may have. [Slide.] In order to help address questions from the panel, we have a wide range of resources

file:///C|/Storage/1121orth.txt (18 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 employee. Fields. Dr. Riedel? DR. RIEDEL: Good morning. available as listed on this slide, and next, you will hear from Dr. Rod Riedel regarding the relevant preclinical data. Thank you. DR. YASZEMSKI: Thanks very much, Dr.

My name is Rod Riedel, and I am a Wyeth I will review several preclinical

proof-of-concept studies and preclinical safety results. These data establish a rationale for

evaluating rhBMP-2 as a treatment for long-bone fractures. [Slide.] This slide shows the signature biological activity of rhBMP-2, its ability to induce bone formation de novo. In this classic rat ectopic

implant assay, BMP-2 is implanted at a non-bony site. The photograph on the left of the slide

shows a bone ossicle induced by BMP-2 14 days following implantation. Histological analysis of this new bony tissue, as shown in the photograph on the right, shows extensive formation of trabecular bone,

file:///C|/Storage/1121orth.txt (19 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 corresponding to the dark pink regions in the photograph on the right, as well as a complete complement of normal, bone-associated cells such as osteoblasts, osteoclasts, stromal cells, and all other bone marrow elements. This bone-inducing activity has been labeled "osteoinduction." It is unique to BMP-2 ments.

and several other members of the BMP protein family. [Slide.] rhBMP-2 can also stimulate bone formation at an orthotopic site, as illustrated by this radiograph of a rabbit ulnar osteotomy treated with rhBMP-2. The following slide provides additional histological information about the effect of BMP-2 at an orthotopic site. [Slide.] As shown at the yellow arrows in the slide, BMP-2 treatment yields increased bone formation and an acceleration of that bone's maturation. [Slide.] BMP-2's ability to induce bone de novo and to stimulate bone formation at an orthotopic site

file:///C|/Storage/1121orth.txt (20 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 provides rationales for two different types of clinical applications. BMP-2 could first be used

to replace a component of standard orthopedic treatment such as autogenous bone graft. Last

January, this panel reviewed this specific use of rhBMP-2/ACS in spine surgery and recommended its approval. Alternatively, BMP-2 could be used to augment standard therapy in an effort to improve its outcome. For example, rhBMP-2/ACS could be

used as an adjunct to standard fracture fixation in certain fractures. We decided to explore this

potential application in a series of preclinical studies. [Slide.] We established a rabbit model to study the effect of BMP-2 as an adjunct treatment. The model

uses surgically-generated, mid-diafaceal ulnar osteotomies to assess bone repair. The osteotomies

do not require instrumented fixation and typically heal within 6 to 8 weeks. It is feasible to

perform bilateral osteotomies, allowing the use of within animal controls which greatly decreases study variability. The model allows the use of plain view

file:///C|/Storage/1121orth.txt (21 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 radiographs, biomechanical analyses, and bone histology as well as the measurement of the local pharmacokinetics of BMP-2. This model does not stimulate the bone and soft tissue damage that occurs in fractures but rather serves as a proof-of-concept system for evaluating bony healing under controlled conditions. [Slide.] We used this model to determine whether BMP-2 could accelerate the process of normal bony healing. In the experiment shown in this slide,

normal animals received bilateral ulnar osteotomies and were treated with either BMP-2/ACS, depicted in the yellow bars, or ACS alone, depicted in the orange bars. Each animal received standard

surgical treatment on the contralateral ulna, depicted by the gray bars. Normal ulnar biomechanical strength was determined by testing age-matched, unoperated limbs, depicted by the blue bar on the right. The slide shows the results of torsional biomechanical testing to failure at various time points following surgery. BMP-2 treatment

accelerated the return to normal strength in this

file:///C|/Storage/1121orth.txt (22 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 model, yielding significant differences from controls at several time points. Importantly,

rhBMP-2 treatment reduced the time to return to fully normal strength by one-third, in this model, from 6 weeks to 4 weeks. The results of this study have been published and form part of the rationale for testing whether BMP-2 could accelerate normal fracture healing in patients. [Slide.] We also studied whether BMP-2 treatment could provide a beneficial effect under conditions of impaired bone healing. To address this

question, we modified the rabbit ulna osteotomy model. We administered systemic prednisolone to

all animals at doses of this glucocorticosteroid that were sufficient to impair bone healing. This slide shows the result from this study in which the effect of rhBMP-2/ACS treatment can be clearly observed in a plain view radiograph. The results of this study have also been published and form part of our rationale for testing BMP-2 in patients with difficult-to-heal fractures. [Slide.]

file:///C|/Storage/1121orth.txt (23 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 We have conducted a comprehensive nonclinical safety evaluation of rhBMP-2 and rhBMP-2/ACS as listed on this slide. All these

studies were reviewed by the panel last January and established the nonclinical safety profile for BMP-2. In these evaluations, no dose-limiting toxicity was detected in any study at exposure levels that greatly exceed anticipated patient exposure. Furthermore, no systemic effect of BMP-2 We attribute this

was observed in any study.

finding to the extremely rapid clearance of BMP-2 from the systemic circulation, resulting in very low, if any, systemic exposure. In summary, we have established a preclinical rationale for evaluating rhBMP-2 as a treatment for long-bone fractures. Our studies

showed that BMP-2 can accelerate bony healing under normal conditions and can additionally increase the probability of bony healing in impaired healing conditions. These effects were consistently observed, radiographically, biomechanically, and histologically. The bone formed by BMP-2 contained

all the components of normal bone and modeled and

file:///C|/Storage/1121orth.txt (24 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

25 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Riedel. Dr. Valentin? DR. VALENTIN: Alex Valentin. Good morning. My name is remodeled in a manner similar to that of host bone. Additionally, the nonclinical safety profile of rhBMP-2/ACS has been established. In conclusion, our preclinical data support the clinical use of InductOs in fracture patients. I now turn the presentation over to Dr. Alex Valentin, who will review the results of the InductOs pivotal clinical study in open tibia fracture patients. DR. YASZEMSKI: Thanks very much, Dr.

I am an employee of Wyeth Research,

and I have been directing BMP-2 clinical research since 1993. [Slide.] I am here today to present the results of our clinical study in patients presenting with open tibia shaft fractures, because the study met its four stated objectives. The pivotal study has shown that InductOs reduced the rate of reintervention for delayed union, accelerated clinical fracture healing,

file:///C|/Storage/1121orth.txt (25 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 demonstrated an appropriate safety profile, and these results were also corroborated by the independent radiology panel. [Slide.] Before presenting the results of this pivotal study, I would like to say a word here about our clinical program, because this pivotal study actually concludes 10 years of our clinical research. During these years, Wyeth sponsored a number of clinical trials, and many were designed and conducted in collaboration with lead orthopedic surgeons in the United States and abroad. including all the studies conducted by our commercial partner, Medtronic Sophomore Danik, Wyeth has enrolled over 1,000 patients, of whom more than 675 were treated with BMP-2. These studies helped us gain important insights into the mechanism of action of BMP-2 in patients and also in its safe use. This Not

information is summarized in the package that we have shared with the panel. [Slide.] Of these studies, the three studies listed here at the bottom of the slide stand out, because

file:///C|/Storage/1121orth.txt (26 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 they helped us under the U.S. standard clinical practice in patients with open tibia shaft fractures. [Slide.] These studies yielded four important conclusions. First, we learned that about 40

percent of the open tibia shaft fractures surveyed in the United States required some form of reintervention to promote fracture healing; second, that surgeons make decisions to reintervene before they can diagnose fracture healing; third, with respect to the definition of study endpoints, we learned that the diagnosis of delayed union was multifactorial. Surgeons made their decision

weighing the progression of clinical symptoms and signs of radiographic fracture healing. And

finally, we concluded that given the high frequency of interventions to promote fracture healing, their avoidance was a relevant endpoint for the orthopedic trauma patients. Today this panel will discuss the relevance of our study primary endpoint, and therefore I would like to emphasize that this endpoint was selected on the basis of this preliminary work. U.S.

file:///C|/Storage/1121orth.txt (27 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 [Slide.] I will now review our pivotal study starting with the study rationale and study design. I plan to describe patients' baseline characteristics, the treatment they received. will then review the key efficacy and safety results and our conclusions supporting the efficacy and safety of InductOs in this patient population. Throughout this presentation, I also plan to address two questions raised by the FDA: Did I

the protocol design allow the measurement of an objective treatment effect, and are these effects clinically relevant? [Slide.] We have chosen open tibia fractures knowing well that we would be facing a challenging indication. Unlike preclinical experimental models

in rabbits, tibia fractures are associated with contaminated wounds and poor vascular supply. Also, the condition requires urgent care adjusted to patients' needs, not leaving surgeons much room to conduct clinical trials. We chose this study nonetheless because open fractures affect the tibia more often than any other bone and represent a very serious condition.

file:///C|/Storage/1121orth.txt (28 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 For example, despite all the progress made in the treatment of open tibia fractures, 62 percent of ulna unions reported in the United States affect the tibia shaft region. We confirmed our preliminary studies that not only do these patients have a great reduction in functional ability, but they suffer also a dramatic and sustained reduction in their quality of life. [Slide.] So we decided to test if BMP-2 could improve the success of standard treatment in these fractures, and as illustrated by the preclinical results presented earlier, BMP-2 has been shown to induce bone in a variety of preclinical models. We hypothesized, therefore, that BMP-2 may improve fracture healing by stimulating bone formation at the site of a recent fracture, therefore reducing the number of reinterventions required. Let me now describe the study design. [Slide.] The following objectives were selected. The primary efficacy objective was to demonstrate by the end of the study an increased probability of

file:///C|/Storage/1121orth.txt (29 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 fracture healing reintervention for delayed union. We acknowledge the challenges raised by this choice. The primary objective depends on

investigator's decision, and investigators could be biased by their treatment of treatment allocation. I would like to stress here that by choosing to evaluate reinterventions, we selected also an endpoint that required to take the patient back to the OR. And no surgeon or patient makes We felt, therefore, that

this decision lightly.

the consequences of this decision was a warranty of very careful consideration. In addition, the study had two secondary efficacy objectives. They evaluate what is

commonly called clinical fracture healing and radiographic fracture healing. For the purpose of

this study and to avoid any confusion, we call radiographic fracture healing fracture union. These two endpoints were evaluated separately. Clinical fracture healing was

evaluated by surgeons, whereas radiographic fracture union was evaluated by an independent radiologic panel blinded to treatment allocation. Finally, the study also addressed the safety of BMP-2 use in these patients.

file:///C|/Storage/1121orth.txt (30 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

31 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 [Slide.] The study was a prospective, randomized, controlled evaluation of BMP-2 in patients presenting open tibia shaft fractures. Patient

randomization was stratified based on the Gustilo I severity to assure a balanced distribution of this important factor among treatment groups. [Slide.] The study was single blind. not aware of the treatment allocation. Patients were And in

addition, a panel of three radiologists reviewed radiographs to assess fracture union. For the purpose of the panel evaluation, union was prospectively defined as being at least three or four cortices being breached on orthogonal viewed. We validated the methods of radiographic review and trained the radiologists prior to the study initiation. X-rays were reviewed by the radiologists of the Osteoporosis and Arthritis Research Group at UCSF under the direction of Professor Harry Genent. To clarify a question raised by the FDA, no x-ray was withheld from the panel interpretation on grounds that the patient was deemed a treatment

file:///C|/Storage/1121orth.txt (31 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 failure. The panel and the surgeons have reviewed the same patients and the same x-rays. To facilitate outcome evaluation, no more than one limb could be treated per patient, no more than one BMP-2/ACS could be used, and repeat treatments were not allowed. Patients were

followed for a total of 12 months postoperatively, at seven prescheduled visits. [Slide.] In the interest of time, I will not review here all the patient inclusion and exclusion criteria. They are detailed in the panel's I will focus, however, on our

briefing package.

exclusion criteria which are relevant to the FDA comment on prophylactic use of bone graft. Such prophylactic treatment was indeed excluded from this study. This exclusion criterion

is justified by our prior findings as well as current experience that prophylactic bone grafting is very rarely prescribed in patients with open tibia shaft fractures treated with an IM nail. Accordingly, should this product be approved, patients requiring prophylactic bone grafting should be excluded by the label.

file:///C|/Storage/1121orth.txt (32 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

33 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 [Slide.] Having addressed the exclusion criteria, I would like now to review the actual standard of care administered in this study. This is an

important point since the panel will review its adequacy and its relevance to the U.S. practice. Wound irrigation and debridement had to be promptly performed and repeated as many times as required. The initial fracture reduction and

stabilization were to be conducted within 24 hours. Definitive wound closure and fixation of the fracture was conducted no later than 14 days after the injury, and definitive fracture stabilization required the use of an IM nail.

Reflecting the absence of national or international consensus on the type of nail to use in open tibia shaft fractures, investigators were authorized to use either reamed or unreamed nails of a brand familiar to them. [Slide.] Let me now review the definition of the treatment groups. Patients were randomly allocated And please note that in

to three treatment groups.

this study, BMP-2 was not tested as a replacement of bone graft but as an adjuvant to the standard

file:///C|/Storage/1121orth.txt (33 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

34 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 treatment. The study was designed to demonstrate superiority of at least one of the two BMP-2 concentrations tested as compared with control patients. So patients randomized to the control

group received only standard care, while patients randomized to BMP-2/ACS received the standard care and the test article. Two concentrations were tested, either .75 mg/ml or 1.5 mg/ml. [Slide.] BMP-2 was implanted at the time of definitive closure of the fracture-related wound, and like in this diagrammatic example, the product was positioned so that it breached the region of comminution, making good contact with the major proximal and distal tibial fragments. [Slide.] I will conclude the review of the study design with comments on the standardization of investigator assessments, which was one of FDA's concerns. We believe the study has been as standardized as possible considering the context of this orthopedic trauma condition.

file:///C|/Storage/1121orth.txt (34 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

35 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Investigators were trained with the same protocol and the same study guidelines. The

protocol included a common definition of fracture healing. There is no commonly-accepted definition

of delayed fracture union, but to the extent a definition was available, that definition was provided. And the protocol also had a definition

of non-union. Now, more importantly, investigators were prospectively required to report at each follow-up visit the presence or absence of fracture tenderness upon palpation, weight-bearing status, and fracture union. These are the cardinal signs

used everywhere to diagnose fracture healing, and as with other clinical conditions, there is no set of signs or symptoms that is always present in each and every patient defining delayed union or healing. The intent of prospectively collecting

this information was not to force what we considered an artificial definition of fracture healing or delayed union, but to use a common set of symptoms and document the soundness of the investigator's decision. [Slide.] A word about patient enrollment. The

file:///C|/Storage/1121orth.txt (35 of 309) [12/2/02 8:59:13 AM]

file:///C|/Storage/1121orth.txt

36 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 study was to our knowledge the largest clinical study in this patient population. centers in 11 countries. [Slide.] Enrolling 450 patients achieved the study objective, and as pointed out on this slide, we noted that during the study conduct, three patients died, one in each treatment group, and seven others withdrew at their request or that of the investigators. Of the 440 remaining patients, 93 to 97 percent were actually evaluated at the last follow-up visit at 12 months. Furthermore, more It involved 49

than 90 percent of all patients were evaluated at all visits, and this number includes the evaluation of all protocol-required efficacy and safety outcomes and in our opinion is a strong indication of good compliance with the clinical protocol. [Slide.] Let me now describe the main baseline conditions and the standard care. We have in total

assessed seven demographic variables and 29 other covariables such as baseline comorbidities, fracture and injury characteristics, treatment administered before and after definitive wound

file:///C|/Storage/1121orth.txt (36 of 309) [12/2/02 8:59:15 AM]

file:///C|/Storage/1121orth.txt

37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 closure, rehabilitation prescribed for the region under study. This comprehensive review was conducted to ensure that study results were sufficiently comparable among all sites to justify pooling them together and compare outcomes across treatment groups. I will now summarize some of the findings in my next slides. [Slide.] Overall, patient demographics were very comparable across treatment groups. The small

difference in median age is not clinically meaningful. And in all respects, study subjects

reflected well the general patient population with open tibia fractures--mostly young males who suffered a high-energy trauma such as motor vehicle accident. Smoking, which is a non-risk for delayed union, was noted equally in all treatment groups. [Slide.] The distribution of the fracture characteristics was also very comparable among treatment groups. Although the majority of

fractures were isolated to the tibia under study, a

file:///C|/Storage/1121orth.txt (37 of 309) [12/2/02 8:59:15 AM]

file:///C|/Storage/1121orth.txt

38 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 substantial proportion of patients in all three treatment groups presented multiple fractures. This is typical of the presentation of patients with open tibia fractures. And fracture severity categorized according to the AO classification, which is very comparable to the OTA classification, was comparable among treatment groups. So was wound

severity according to the Gustilo-Anderson classification. For example, Gustilo IIIB

fractures, which have a very poor fracture prognosis, were observed in 11 to 18 percent of the cases involved. [Slide.] With respect to the mode of fracture reduction and fixation, we noted that all but one percent of the patients enrolled were actually treated using IM nail, and most patients received statically locked nails. In agreement with the

protocol design, a few patients equally distributed across treatment groups received dynamically locked nails or unlocked nails. About two-thirds of the

patients received unreamed nails, while the rest of the patients received reamed nails. We found an imbalance in the distribution

file:///C|/Storage/1121orth.txt (38 of 309) [12/2/02 8:59:15 AM]

file:///C|/Storage/1121orth.txt

39 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 of reamed and unreamed nails which approached statistical significance. We have investigated the

cause of this imbalance, and our analysis, which was included in the panel's briefing package, shows this imbalance was a consequence of a centralized randomization procedure. So, to summarize, with two exceptions, demographic characteristics, risk factors, fracture presentation, and fracture management were found sufficiently similar across all centers to justify the pooling of the data and among treatment groups to justify the treatment effect analysis. [Slide.] Let me now review the study results. The primary efficacy objective of this study was to demonstrate an increased probability of fracture healing. At study completion,

treatment success was based on the number of fractures being healed by the investigator in the absence of other intervention to promote fracture healing. Secondary procedures to promote fracture healing and also reported screw breakage leading to fracture self-dynamization were deemed indicative of treatment failure.

file:///C|/Storage/1121orth.txt (39 of 309) [12/2/02 8:59:15 AM]

file:///C|/Storage/1121orth.txt

40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 [Slide.] The primary efficacy objective of this study was met. BMP-2 increased treatment success,

and in patients receiving BMP-2/ACS 1.5, the success rate was increased by one-third. Conversely, there was a 41 percent decrease in failure rate. This is a clinically substantial

improvement, and the overall difference among treatment groups was highly significant. Furthermore, BMP-2 was dose-dependent, and we believe that the dose-dependency of the treatment effect further supports our conclusion that this effect was unbiased. [Slide.] As I pointed out on the study endpoints, treatment failures account for two categories of event--secondary interventions and self-dynamizations. With respect to secondary

interventions and in the spirit of an intent-to-treat analysis, all procedures, whether prescribed or preformed, were counted. In

addition, in some cases, the inadvertent breakage of locking screw led to fracture self-dynamization. These occurrences were counted as treatment failures because self-dynamization affects fracture

file:///C|/Storage/1121orth.txt (40 of 309) [12/2/02 8:59:15 AM]

file:///C|/Storage/1121orth.txt

41 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 healing. It is noteworthy that self-dynamization appeared to be treated more frequently in control patients and in BMP-2 treated patients. difference approached significance. These hardware failures are observed in the presence of delayed union. Self-dynamizations This

are not subject to investigative bias because they occur without intervention from the investigator. In this case, the difference observed in self-dynamizations independently suggests that patients treated with BMP-2 healed faster and reduced their exposure to this risk. [Slide.] We have also assessed the number of interventions to identify interventions that may have been done in the same patients either concurrently or sequentially. This analysis

complements the previous one, which counted the number of patients with secondary intervention. The number of procedures was counted only in patients who actually received the treatment they were randomized to and had evaluable outcomes. call these patients the evaluable population. Supporting the results observed in the We

file:///C|/Storage/1121orth.txt (41 of 309) [12/2/02 8:59:15 AM]

file:///C|/Storage/1121orth.txt

42 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 primary efficacy endpoint, in this patient population, the number of interventions was found significantly decreased in the 1.5 mg/ml treatment group as compared to the control group. [Slide.] Accounting for the number of interventions also allows us to characterize and evaluate their invasiveness. Noninvasive procedures represent ultrasound, electrical stimulation, or magnetic field stimulation. Less invasive procedures

represent nail dynamization or addition of a functional brace. [Slide.] And finally, the most invasive procedures represent procedures such as bone graft, exchange nail, tibial osteotomy or bone transport. Half of the procedures conducted in control patients were categorized as most invasive, whereas invasive procedures represented only 37 percent of the procedures undertaken in patients treated with BMP-2 1.5. Consequently, in these evaluable patients receiving InductOs, the number of the most invasive interventions was found significantly decreased as

file:///C|/Storage/1121orth.txt (42 of 309) [12/2/02 8:59:16 AM]

file:///C|/Storage/1121orth.txt

43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 compared to control patients. [Slide.] A secondary efficacy objective of this study was to demonstrate an accelerated fracture healing. This endpoint was the number of fractures

clinically healed without a secondary intervention as determined by the investigator 26 weeks post-injury. As I mentioned earlier, to determine fracture healing, investigators assessed clinical parameters such as tenderness upon fracture palpation, patient's weight-bearing status, and reviewed the orthogonal radiographic views of the fracture. The 6-month time point for this evaluation was suggested to us during the 1996 review of the protocol by FDA. We have adopted it for this

study, and in addition, fracture healing was also reported at all other preceding and following visits. [Slide.] As illustrated on this slide, the cumulative rate of fracture healing indicates that the secondary efficacy objective was also met. Twenty-six weeks post-injury, InductOs increased

file:///C|/Storage/1121orth.txt (43 of 309) [12/2/02 8:59:16 AM]

file:///C|/Storage/1121orth.txt

44 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 the number of patients healed without secondary intervention by 53 percent as compared to control patients, thus suggesting an accelerated fracture healing. This increase was highly statistically

significant. Similarly, the number of patients healed after treatment with InductOs was significantly increased at every other visit starting at the Week 14 follow-up and lasting until the end of the study. [Slide.] This accelerated healing was confirmed by this Kaplan-Meyer evaluation of probability of healing as a function of time. According to this

evaluation, patients treated with InductOs healed significantly faster than control patients. The

median time to healing for InductOs patients was 39 days shorter than for control patients, a clinically and statistically significant difference. [Slide.] Another secondary efficacy objective of this study was to demonstrate accelerated fracture union. We are referring to this outcome as

"fracture union" rather than "fracture healing"

file:///C|/Storage/1121orth.txt (44 of 309) [12/2/02 8:59:16 AM]

file:///C|/Storage/1121orth.txt

45 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 because the endpoint was based solely on radiographs. For this objective, the endpoint was the number of fractures with radiographic healing assessed 6 months after the fracture, and the result was also reported at all the other visits. For the purpose of this endpoint, patients united after a second intervention were counted as treatment failures. This time, the endpoint was measured independently by three trained radiologists from UCSF who were blinded to patients' treatment allocation and patients' clinical presentation. And as I mentioned already, the radiographic assessments used the same x-rays that were used by the investigators, and all radiographs from all patients that were reviewed by the surgeons were also reviewed by the panel. [Slide.] As indicated on this chart, this efficacy objective was also met. The radiology panel

blinded to treatment allocation and patient symptoms independently confirmed that BMP-2 treatment effect. At 26 weeks after the fracture

in the InductOs group, the number of patients with

file:///C|/Storage/1121orth.txt (45 of 309) [12/2/02 8:59:16 AM]

file:///C|/Storage/1121orth.txt

46 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 fractures united was 65 percent higher than in the control group. This clinically significant

increase was also statistically significant. The difference in union rate between InductOs and control patients was still significant at the 50-week visits, when 34 percent more patients had their fractures united in the InductOs group as compared to the standard care. [Slide.] As pointed out by FDA, the Kaplan-Meyer evaluation of radiographic union failed to identify the difference between treatment groups. We

believe in this case that radiographic fracture union is best evaluated using the cumulative rate of union by visit rather than the Kaplan-Meyer display. And in my next slides, I will attempt to

explain why. [Slide.] As illustrated on my previous slides, the radiology panel diagnosis of fracture union occurred at a later time point than investigators' assessment of clinical fracture healing. time is not unusual. In the absence of guidance provided by patients' clinical science, it reflects the This lag

file:///C|/Storage/1121orth.txt (46 of 309) [12/2/02 8:59:16 AM]

file:///C|/Storage/1121orth.txt

47 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 difficulties of transforming the progressive callous mineralization into a dichotomous decision of success and failure. We verified this explanation in a separate study where a group of orthopedic surgeons were requested to determine fracture union without clinical information. Furthermore, similar

observations were made and published by many other authors. [Slide.] So, as an example of lag time observed sometimes between investigators and radiologists, this slide illustrates the case of a 29-year-old male who suffered a Gustilo IIIB left tibia fracture following a motorcycle accident. His

fracture was treated with an unreamed nail and BMP-2/ACS 1.5. As illustrated on these AP views, the fracture progressed toward healing, and at 20 weeks post-injury, the patient had no tenderness upon palpation, was full-weight-bearing, and was deemed healed by the investigator. In this case, the

diagnosis of fracture union was confirmed at the following visit by the independent radiology panel, but the delay varied from patient and patient.

file:///C|/Storage/1121orth.txt (47 of 309) [12/2/02 8:59:16 AM]

file:///C|/Storage/1121orth.txt

48 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 [Slide.] Overall, the delay observed between investigators and radiologists is illustrated on this slide which shows side-by-side the two evaluations for the same patients. For the purpose

of this example, only the control patients were selected. The study was designed to maximize clinical evaluation in the first 6 months of patient follow-up. The delay between clinicians

and radiologists resulted in pushing the evaluation of radiographic union to the second half of the study, in the last 6 months, where only two follow-up visits were planned. [Slide.] So it is important to understand that for most patients, clinical fracture healing was established by the investigators in the first 6 months of the study. In contract, the independent

panel observed radiographic union about 3 months later, in the second half of the study. As a result, clinical fracture healing could be assessed by the investigators at one of four follow-up visits planned in the first 6 months, and in contrast, the second half of the

file:///C|/Storage/1121orth.txt (48 of 309) [12/2/02 8:59:16 AM]

file:///C|/Storage/1121orth.txt

49 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 study, radiographic union could be assessed by the panel only at two follow-up visits. The paucity of time points available in the second half of the study has hindered the Kaplan-Meyer evaluation of fracture union, and it resulted in what can be called an "interval censoring" of data. It is for this reason that we prefer instead to express the radiographic union results as a cumulative proportion of patients united by visit. [Slide.] We believe that the independent of fracture union by the radiology panel is most appropriate to corroborate the investigators' findings in a different manner. In cooperation

with the FDA, we constructed a composite index called Combined Clinical and Radiographic Endpoint, or CCRE. As shown on this slide, this composite index assigns patients to four groups reflecting the clinical assessment of healed/not healed, and the radiographic assessment of united or not united. It can be viewed as a way of separating

true positive results, shown here on the first

file:///C|/Storage/1121orth.txt (49 of 309) [12/2/02 8:59:16 AM]

file:///C|/Storage/1121orth.txt

50 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 line, from false positive, false negative, and true negative results shown here on the following three lines. If you would allow me to draw your attention to the second line, showing patients healed but not united, I would like to make the If the study results were

following observation.

affected by investigator bias, one would expect to see the number of patients healed without radiographic substantiation of union being different among treatment groups; but that is not what the data show. In the control group, there were 16 such patient, in the BMP groups, there were 21 and 19. This small numeric excess is not only statistically insignificant, but also very small compared to the size of the treatment effect. So one must concluded that the healed but not united patients are best explained by a lag in the radiographic findings happening equally in all three treatment groups and not due to investigator bias. [Slide.] By using this composite index, we further analyzed BMP-2 treatment effects. This slide

file:///C|/Storage/1121orth.txt (50 of 309) [12/2/02 8:59:16 AM]

file:///C|/Storage/1121orth.txt

51 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 presents the results of this analysis. Conservatively, we scored as treatment successes only patients clinically healed and radiographically united. Even by using this

conservative assessment, we still found that 36 percent more patients in the InductOs group were healed and united as compared to the control group. This difference remains clinically and statistically significant. [Slide.] In their review, the FDA has expressed the concern about the heterogeneity of the study population which may have founded study results. As I have pointed out in my presentation, the review of seven demographic criteria and 29 covariables led us to conclude that the randomization process was generally successful in generating comparable treatment groups. To further address the question, we conducted additional analysis using prospectively defined patient categories. These categories

reflect fracture severity such as isolated versus fractures are multiple skeletal locations, Gustilo classification, risk factors such as smoking, patient enrollment at different research centers or

file:///C|/Storage/1121orth.txt (51 of 309) [12/2/02 8:59:17 AM]

file:///C|/Storage/1121orth.txt

52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 countries, and the type of nail used. Our conclusion has been that no specific patient category as defined above influenced the overall efficacy results. justify the data-pooling. [Slide.] Another question in front of the panel today concerns the potential investigator bias made possible by the knowledge of treatment allocation. To address this issue, we conducted several post-hoc evaluations to determine the consistency of treatment failure or success assessments across centers and treatment groups. We concluded that These findings further

success and failure criteria were applied consistently. For example, investigator prescription of secondary intervention with a diagnosis of fracture healing were concordant with patients' clinical status. Additionally, reintervention decisions were made at comparable time points across treatment groups. That means that no patient was

given extra time to achieve healing before a reintervention was decided. The time from injury

to prescription of reintervention was consistent

file:///C|/Storage/1121orth.txt (52 of 309) [12/2/02 8:59:17 AM]

file:///C|/Storage/1121orth.txt

53 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 with patients' condition and, most importantly, comparable across all three treatment groups. Also, the clinical status of patients diagnosed as healed did not deteriorate. These findings made us conclude that no evidence of investigator bias could be detected and that the study was conducted with appropriate standardization to earn the review of its results. [Slide.] Let me now review the BMP-2/ACS safety. This objective was based on clinical, radiographic, histologic, and lab evaluations and on the detection of antibody titers to BMP-2, bovine, and Human Type 1 collagen. As a background to this safety review, let me remind you that we have generated a database including over 1,000 patients from all Wyeth BMP-2 studies. These results were summarized in our

investigator brochure which was included in the panel's briefing package. The safety data collected in this pivotal study support our general conclusion that BMP-2/ACS can be safely used in patients. We found that most

adverse events were consistent with the trauma setting. Generally, the treatment groups have the

file:///C|/Storage/1121orth.txt (53 of 309) [12/2/02 8:59:17 AM]

file:///C|/Storage/1121orth.txt

54 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 same frequency and the same severity of adverse events. [Slide.] Let me now review some of these parameters, in particular those of interest to orthopedic surgeons. I will start with the

incidence of infection. The introduction of a foreign body in this potentially contaminated wound always poses the question of an increased infectious risk, so we monitored very carefully the occurrence of infections in this pivotal study, and we have conservatively recorded all declared infections, whether superficial or deep, whether confirmed or not by bacteriologic analysis. And this reporting

rule explains a relatively high incidence of infectious events reported here. However, the

incidence of infections was found comparable among all three treatment groups. And, as expected, most infections occurred in patients who suffered the most severe fractures, the Gustilo IIIA and IIIB, and furthermore, to track the most severe infections, we identified all the cases of deep-bone infection requiring the surgical procedure. And this is the infectious

file:///C|/Storage/1121orth.txt (54 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

55 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 event typically reported in orthopedic research publications. Again, the incidence of deep-bone infections was comparable with data previously published in the U.S. and is strictly comparable across treatment groups, suggesting that BMP-2 did not increase patients' exposure to infection. [Slide.] Another safety variable important to orthopedic surgeons is the occurrence of hardware failure. In this study, hardware failure includes

most broken or bent locking screws and in two cases, a broken nail. Hardware failures occur with decreasing frequency across the three treatment groups--21 percent, 17 percent, and 11 percent. These

decreased incidences of hardware failure in BMP-2-treated patients was observed whether patients were treated with a reamed or an unreamed nail. [Slide.] Because of the osteoinductive properties of BMP-2, we have monitored all signs of exaggerated or abnormal bone formation. Such

events include calcification of remote sites, which

file:///C|/Storage/1121orth.txt (55 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

56 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 we called ectopic, and calcification in the fracture vicinity, which we have called heterotopic. Conservatively, heterotopic callouses in tibia/fibula synostosis were added to the heterotopic calcifications. No ectopic calcification due to BMP-2 was reported, and overall, heterotopic calcification, hypertrophic callouses, and tibia/fibula synostosis included, were rarely observed in this study. While not statistically significant, treated patients with BMP-2 had an increase in the number of such reports compared to control patients--respectively, eight, four, and four. [Slide.] Patients enrolled in this pivotal study were also monitored for the development of antibodies to BMP-2 and to Type I collagen, and serum samples were collected at baseline 6 and 20 weeks post-treatment. We have detected anti-BMP-2 antibodies in a small number of patients. This human response

was always marked by a low and transient titer. Anti-bovine Type I collagen antibodies were also detected in nine of the control patients

file:///C|/Storage/1121orth.txt (56 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

57 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 and in 22 and 29 BMP-2/ACS patients. There was no

cross-reactivity, and none of these patients presented an immune reaction to Human Type I collagen. The presence of antibodies to BMP-2 or to bovine collagen was not associated with lower treatment efficacy or associated with any accompanying symptoms of immune reaction or allergy. [Slide.] The other safety evaluations are summarized on this slide. [Slide.] So, to summarize my presentation, we found that InductOs may offer new therapeutic options in the treatment of open tibia fractures which are serious orthopedic trauma conditions. In this study, the product has consistently demonstrated safety and efficacy irrespective of fracture severity, documented risk factors, type of nail used, investigational center, or country where patients were treated. [Slide.] BMP-2 effective and safety sue was supported by results observed in regard to all four

file:///C|/Storage/1121orth.txt (57 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

58 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Valentin. Dr. Swiontkowski, welcome. DR. SWIONTKOWSKI: Good morning. My name study objectives. First, BMP-2/ACS reduced the rate of reintervention for delayed union; in addition, BMP-2 accelerated clinical fracture healing; the evidence of efficacy was corroborated by the independent panel evaluating radiographic fracture union; and finally, InductOs safety profile was deemed appropriate. I thank you for your attention and wish not to turn the presentation Professor Marc Swiontkowski, who will discuss the relevance of these results to the U.S. trauma population. DR. YASZEMSKI: Thanks very much, Dr.

is Marc Swiontkowski, and I am from the University of Minnesota Department of Orthopedic Surgery, where I have been privileged to serve as chairman of that department since 1997. I practice in two Level I trauma centers in the Twin Cities. I also co-direct the Clinical

Outcomes Research Center for the University of Minnesota School of Medicine. Prior to my arrival in Minneapolis, I was

file:///C|/Storage/1121orth.txt (58 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

59 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 the Chief of Orthopedic Surgery at Harbor View Medical Center in Seattle, Washington, also a Level I trauma center. I have spent my entire 17 years of clinical practice in the management of severely injured patients, and based on my research interests have a particular interest in improving the knowledge base on which we have managed these patients. I have been involved with the rhBMP-2 series of studies since 1991. I have participated

in the initial discussions regarding the selection of the clinical indication to study through the conduct of the 12-patient safety study in the United States, the 60-patient trial in the United States, and as an advisor to the pivotal trial which we are discussing today. I have no financial interest or ethical conflicts regarding the study we are discussing today, although I am an active paid consultant with Wyeth, who has paid my way here today. [Slide.] This morning, I will address some of the questions that FDA has raised to the panel. I will

review the relevance of the study data to clinical

file:///C|/Storage/1121orth.txt (59 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

60 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 practice both in the United States and internationally. I will comment on the selection of the clinical indication and the primary efficacy endpoint secondary intervention to promote fracture healing. I will also comment on the study design and conduct and how the study findings are relevant for patients with tibial fractures in the United States. I will compare the data collected from the

pivotal study with that of an identically-designed FDA-approved trial of 60 patients conducted in the United States. Finally, I will address how the pivotal study findings are of utility to surgeons managing patients with long-bone fractures. [Slide.] Open tibia shaft fractures were selected as the clinical indication because of the high medical need for improving outcomes. These

fractures have a high rate of delayed union and, among common fractures, the highest rate of non-union. As a stringent model, results from

these fractures can be readily applied to other less severe long-bone fractures with more biologically friendly tissue conditions.

file:///C|/Storage/1121orth.txt (60 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

61 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 In other words, if the protein works in this setting of severe muscle damage from high-energy blunt injury, it will work anywhere. As part of the panel discussion today, you have been asked to evaluate the definition of standard care described by the sponsor for this patient population. The question is what role bone

grafting plays in the management of these fractures. As a practicing orthopedic traumatologist, I will reiterate the sponsor's previous statements that prophylactic bone grafting is not standard care, either in the U.S. or internationally, for the management of the fracture type studied in this trial. [Slide.] I base my statements about the standard of care not only on my experience in clinical practice but also from investigation of worldwide standards, including data collected by Wyeth and information in the literature. Wyeth conducted two separate U.S.-based investigations of patients with tibial shaft fractures. These include a retrospective chart

review of 484 patients conducted in 1993 and a

file:///C|/Storage/1121orth.txt (61 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

62 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 prospective natural history study of 86 patients conducted in 1994. Both studies demonstrated a low

incidence of bone grafting, revealing that early prophylactic bone grafting is not commonly practiced in the management of these fractures. Regarding the literature, when early prophylactic bone grafting has been advocated, its use has been strictly limited to those fractures with substantial bone loss. Patients with

substantial bone loss are a different type of patient population and were excluded from this pivotal study. In general, surgeons seek to avoid bone grafting whenever possible, because there is a 9 to 10 percent incidence of major complications and 20 to 30 percent incidence of persistent pain and disability. [Slide.] This pivotal trial was designed in consultation with over 20 of my peers who staff Level I trauma centers around the United States and Canada, in four separate meetings. During these

meetings, we discussed the merits of individual skeletal injuries and debated the aspects of the study design amongst ourselves and with Wyeth.

file:///C|/Storage/1121orth.txt (62 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

63 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Ultimately, we recommended the open tibial fracture model. In addition to medical need, its merits include that the ACS could be implanted at the time of wound closure, which allowed adequate time for patient and family consent. This is a key

consideration in the trauma population. We felt that these fractures were consistently managed with staged wound closure. argued the pros and cons of a blinded trial and ultimately decided upon an open-label trial design as the majority of practicing participating surgeons stated with confidence that they could not ethically place a plain collagen sponge in an open fracture wound, as it would have little potential benefit to the patient and could expose the patient to the risk of infection in a potentially contaminated wound environment. Further, they believed that their ethics committees would not approve such a trial-blinding strategy. [Slide.] After reaching agreement on the open tibial fracture model and open label study design, this group of orthopedic traumatologists We

file:///C|/Storage/1121orth.txt (63 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

64 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 extensively debated the pros and cons of the primary efficacy endpoint. We felt that revision surgery--secondary intervention--was a measurable endpoint that held the most relevance from a clinical perspective. Failure was defined as the need for any secondary intervention with a stimulatory effect on healing. This is a highly conservative definition which includes not only surgery but also other less invasive interventions. The endpoint used in this trial is consistent with other control trials, both past and present, conducted within the orthopedic trauma community. [Slide.] The panel has been asked to evaluate the clinical relevance of the primary efficacy endpoint used in this trial. In making the pivotal trial as

directly applicable to clinical practice as possible, the clinician's assessment of fracture healing, based on a combined assessment of clinical and radiographic findings, was favored over a purely radiographic assessment of efficacy. This is consistent with clinical practice where treatment decisions are routinely made based

file:///C|/Storage/1121orth.txt (64 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

65 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 on the combined clinical presentation of the patient and radiographic appearance of the fracture. As the sponsor has previously described, the key variables in the assessment of fracture healing are lack of fracture site tenderness upon manual palpation, ability to be full-weight-bearing, and radiographic fracture union. [Slide projector malfunction.] DR. SWIONTKOWSKI; the slides. DR. YASZEMSKI: That will be up to you, I'm not sure we need

Dr. Swiontkowski, if you want to proceed. DR. SWIONTKOWSKI: Yes; I don't mind. I'm

really not sure you need them, anyway. This trial was hypothesis-driven. Specifically, an expected rate of secondary intervention was established from the prospective natural history study conducted by Wyeth, and these data were used to support the hypothesis and sample size for this pivotal study. Because of the practical nature of conducting a trial within busy trauma centers, the study was designed to minimize the disruption of

file:///C|/Storage/1121orth.txt (65 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

66 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 care and to be compatible with day-to-day clinical practice. The design was acceptable to the

clinical investigators in the United States, and then, as the sponsoring agency went international, the trial design, which had been piloted in the United States, was acceptable to investigators in numerous countries. This study designed passed all human subjects committee reviews, which spoke to our consideration of patients and their families in the trial design. [Slide.] What we have here in this three-arm, 450-patient, randomized control trial of patients with open tibial fractures is the largest orthopedic trauma clinical trial that I am aware of. It was conducted at high-quality centers and,

as a part of my advising role to Wyeth, I helped identify these centers. As a part of my academic

career, I have traveled to many centers in the UK, South Africa, and Germany. majority of these sites. I have visited the I knew that the quality

of care in the operating room and postoperatively, as well as the implants available and the surgeon training, was basically equivalent to United

file:///C|/Storage/1121orth.txt (66 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

67 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 States' standards. Many of these surgeons had in

fact trained at our center in Seattle or in other U.S. or Canadian centers with which we are all familiar. Over 90 percent of the centers involved in this trial have published their clinical outcomes for treatment of long-bone injuries in the peer-reviewed literature. This further confirms

their commitment to delivering a high standard of care. The fact that we chose the right centers is revealed in the fact that we had over 90 percent patient follow-up at one year across all centers and countries. The quality of this study has been

certified by high-profile podium presentations at the Orthopedic Trauma Association, CCOT, and the American Academy of Orthopedic Surgeons. It has

also been accepted for publication by the Journal of Bone and Joint Surgery, which is widely considered to be the premier peer-reviewed publication in the field of orthopedic surgery. [Slide.] As the sponsor has previously described, definitions for key study outcomes were provided to investigators participating in this trial. These

file:///C|/Storage/1121orth.txt (67 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

68 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 include definitions for delayed union, non-union, and secondary intervention for delayed union or non-union. Recognizing the need to define these

key variables, the sponsor provided radiographic and clinical definition that not only conformed to the assessment methods commonly described in the orthopedic literature, but also to governmental guidelines including HCFA and FDA. The definitions provided are appropriate in that they are consistent with clinical practice. [Slide.] To address the relevance of the pivotal study data to the U.S. population, I will review a number of variables collected as part of the FDA-approved 60-patient U.S. study in patients with open tibia fracture, compare it to that of the pivotal study. It is important to note that the U.S. study was a randomized controlled trial that used the same endpoints, decisions for healing, and radiographic assessment as the international pivotal study. Among these two studies, the U.S. and the pivotal trial, the population of patients included is comparable overall. As you can see, the

file:///C|/Storage/1121orth.txt (68 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

69 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 demographic and baseline profile of the patients in both trials in terms of age, gender, smoking history, and extent of injury are very similar, with no significant differences. Similarly, there were no differences in the distribution of fracture patterns by AO/OTA classification, Gustilo wound type, or mechanism of injury. Over 90 percent of patients in both the

U.S. and international cohort were injured by a blunt mechanism. [Slide.] Again, the relevance of this pivotal study to the U.S. population is demonstrated by the consistent standards of fracture and wound treatment that were applied among the centers involved in the two trials. State-of-the-art

implants were used and, with few exceptions, were manufactured by one of four major implant companies, all of which are FDA-approved for use in the United States. The patients had repeat irrigation and debridement with delayed wound closure in many cases, and consistent muscle flap/skin graft coverage for larger wounds, the Gustilo IIIB classification.

file:///C|/Storage/1121orth.txt (69 of 309) [12/2/02 8:59:18 AM]

file:///C|/Storage/1121orth.txt

70 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 I should note that there was a small number of Type IIIB fractures in the international cohort which were treated with skin grafts, whereas in the United States cohort, all patients with IIIB fractures received muscle flaps. In each case that

a muscle flap was not used, the investigator verified that the fracture grade was indeed Gustilo Type IIIB. [Slide.] The pivotal study design did not randomize by reamed versus unreamed technique of nail insertion. At the time the study was conducted,

and even today, the use of reamed versus unreamed nails in the management of open tibial shaft fractures remains an unresolved controversy. A meta-analysis published in the Journal of Orthopedic Trauma by Bhandari et al. has failed to confirm an advantage or disadvantage for reamed or unreamed nailing in this setting. Given that

this issue remains unresolved, the sponsor's decision to include both types of nail insertion techniques in the pivotal trial was appropriate. In comparing the U.S. and international studies, one observes differences in use. The

issue of reamed versus unreamed nailing is now

file:///C|/Storage/1121orth.txt (70 of 309) [12/2/02 8:59:19 AM]

file:///C|/Storage/1121orth.txt

71 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 being specifically addressed through a randomized control trial of 900 patients in the U.S. and Canada, funded collaboratively between the National Institute of Arthritis and Musculoskeletal Skin Diseases at the NIH and the Canadian Institute for Health. This trial, the Sprint trial, for which I am the principal investigator, utilizes the same clinically-relevant endpoint of secondary intervention as judged with the surgeon, with patient exam and radiographic review, as was used in this pivotal trial. As sometimes happens, the

study data show an imbalance for nail insertion technique. [Slide.] Finally, the most important outcome--the rate of secondary intervention. Again referencing

our 60-patient U.S. trial, the data from the international pivotal study supports similar decisionmaking used in these different settings. The rates of secondary intervention in the standard of care arm verified the hypothesis that this is not an uncommon outcome and confirmed what we learned from the natural history study conducted by Wyeth, which revealed a 41 percent rate of

file:///C|/Storage/1121orth.txt (71 of 309) [12/2/02 8:59:20 AM]

file:///C|/Storage/1121orth.txt

72 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 secondary intervention among patients with open tibial shaft fractures treated with IM nailing. The types of secondary interventions were not significantly different between trials, with IM nail dynamization being most common, followed by exchange nailing and bone grafting procedures. An important measure of quality of care--infection--utilizing the strict definition outlined by Dr. Valentin was 18 percent in the U.S. cohort and 22 percent in the international cohort. Another important measure--the time to decision for reintervention--was also comparable across the two trials, peaking between 10 and 20 weeks post-injury, time points consistent with those reported in the orthopedic literature. The decision as to when to intervene was within 2 days for patients in the three trial treatment arms in both studies, confirming a lack of bias in this decisionmaking. [Slide.] Last, I will comment on the utility of the pivotal study findings to other surgeons managing patients with long-bone fractures. As we have seen today, the trial findings are consistent with the preliminary data collected

file:///C|/Storage/1121orth.txt (72 of 309) [12/2/02 8:59:20 AM]

file:///C|/Storage/1121orth.txt

73 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 in the natural history and pilot studies conducted in the U.S. The standards applied in the

international study in terms of definitions utilized and treatments administered are compatible with U.S. practice. The primary efficacy endpoint of secondary intervention was a clinically meaningful measure and being consistent with U.S. clinical practice is directly applicable to the care of patients here in the U.S. Most importantly, through the significant reduction in secondary intervention, the study results represent a substantial clinical benefit to patients with these types of severe injuries. Finally, I must reiterate that this fracture model represents the most harsh biologic environment in which to test the impact of rhBMP-2. I am confident and comfortable with the conclusion that the protein will definitely work in other settings. [Slide.] In conclusion, to my knowledge, this is a controlled trial of the highest quality yet performed in the orthopedic trauma community. The

study shows a significant decrease in the rates of

file:///C|/Storage/1121orth.txt (73 of 309) [12/2/02 8:59:20 AM]

file:///C|/Storage/1121orth.txt

74 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 MR. KAISER: secondary intervention which are highly relevant to patients with these injuries, as well as the physicians treating them. As an orthopedic traumatologist, I am comfortable with the conclusion that rhBMP-2 with a collagen sponge is safe, that it accelerates healing in long-bone fractures, and will be a useful adjunct to my care of these patients. I am proud to have played a small role in the conduct of this trial and, along with the Wyeth representatives, would be pleased to answer any questions that the panel may have. Thank you. DR. YASZEMSKI: Swiontkowski. We're going to proceed now with the presentations by FDA. Aric Kaiser, the lead Thanks very much, Dr.

reviewer, will present first. FDA Presentation Good morning. I am Aric

Kaiser, the lead reviewer for this PMA. We are going to be discussing InductOs, the rhBMP-2/ACS device from Wyeth Pharmaceuticals. [Slide.] I would also like to introduce the

file:///C|/Storage/1121orth.txt (74 of 309) [12/2/02 8:59:20 AM]

file:///C|/Storage/1121orth.txt

75 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 collaborator for this project, Peter Hudson, who was the lead preclinical reviewer; Barbara Buch, the clinical reviewer; and Chang Lao, our statistical reviewer. [Slide.] In addition to those three, we also had a number of people from CDRH as well as CBER and CDER who provided valuable input on reviewing the information provided by the sponsor. [Slide.] As described by the company, this is a two-component combination product consisting of a growth factor, rhBMP-2, and a carrier for that protein, the ACS absorbable collagen sponge. [Slide.] One of the things I want to point out is that in the clinical presentation that we will be giving, we are reviewing only the tibia data that was generated as a result of the clinical trial. We did not review and we are not presenting any of the data that was summarized in the study that related to the use of the product either in other orthopedic applications or in dental applications. This leads into a discussion that we are going to have you participate in this afternoon,

file:///C|/Storage/1121orth.txt (75 of 309) [12/2/02 8:59:20 AM]

file:///C|/Storage/1121orth.txt

76 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 looking at the indication for product, whether it is the indication that was actually studied by the company, the acute open tibia fracture stabilized with IM nails, compared to the proposed indication which is listed up here that includes both an indication for the product as well as some claims about the behavior of the product. [Slide.] A little history about this PMA. The

original PMA was issued a not approvable letter, and the basis for this letter was our belief that there were some critical deficiencies in the design of the trial that would have prevented the company from developing data to fall in the category of valid scientific evidence. The sponsor, after reviewing that information, believed they could address our concerns and submitted a response to that letter where they re-analyzed their data and then tried to address what we had identified as the critical issues as well as some other issues of a clinical and statistical nature that had been identified in that letter. [Slide.] We are not going to go into a discussion

file:///C|/Storage/1121orth.txt (76 of 309) [12/2/02 8:59:20 AM]

file:///C|/Storage/1121orth.txt

77 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 of the preclinical information. As the sponsor

mentioned, this was previously discussed by the panel at the January meeting. All I want to identify from a preclinical standpoint is that as a result of that meeting, there were some issues that were identified as being outstanding preclinical issues, and as the company has said, they are conducting some additional in vitro and in vivo evaluations to address those concerns, having to do with the potential ability of the protein to promote tumor formation as well as the impact--the potential impact--of the protein on fetal development. The company is also working on some new ELISA assays that are more sensitive. We are going to focus today on clinical and statistical issues that were outstanding from our review of the resubmission of the product. That would lead us to the end of this discussion, where we are going to have you provide some comment on four questions that we are going to pose--one having to do with the study design, some questions having to do with the ability of the data to demonstrate safety and effectiveness, and then, finally, this labeling question that I alluded to

file:///C|/Storage/1121orth.txt (77 of 309) [12/2/02 8:59:20 AM]

file:///C|/Storage/1121orth.txt

78 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 having to do with the indications. With that, I would like to introduce Dr. Barbara Buch, who was the clinical reviewer for this product and will present some issues having to do with the clinical trial design as well as the resulting clinical data. DR. YASZEMSKI: Dr. Buch? DR. BUCH: Good morning. Thanks, Mr. Kaiser.

My name is Dr. Barbara Buch, and I am an orthopedic surgeon in the Orthopedic Device Branch of the FDA, and I thank you for your attention and endurance today. The sponsor has already presented the summary of the international investigation quite nicely. It is not my intent to repeat the summary

of safety and effectiveness which has already been so well-presented by the sponsor. What I do want to do is bring your attention to some major issues that may have impacted on the study outcomes and made our evaluation and interpretation of this study and its results quite challenging. The issues involve primarily aspects of study design, general clinical relevance, but it

file:///C|/Storage/1121orth.txt (78 of 309) [12/2/02 8:59:20 AM]

file:///C|/Storage/1121orth.txt

79 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 also involves aspects of interpretation and analysis and questions raised when evaluating the safety of this device. Before we can even look at study results, we should evaluate the strength of the study itself--what are the confounding variables, how were the patients assessed, what data was collected, and how was the data analyzed. Clinical trials often differ from clinical practice, as they are intended to specifically define the therapeutic effect and safety profile of a specific treatment without question. Ideally, when comparing two or three groups, it is preferable to have them as much alike as possible except for the treatment variable being studied. This allows us the most confidence that a

difference in outcome is due to that treatment. [Slide.] As has already been mentioned, this is a difficult task to accomplish, especially in a trauma population, as there are multiple confounding variables that influence the final outcome--variables such as the energy of the initial injury, the environment in which the initial injury occurs, and the contamination of the

file:///C|/Storage/1121orth.txt (79 of 309) [12/2/02 8:59:20 AM]

file:///C|/Storage/1121orth.txt

80 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 fracture, prior medical health and nutritional status, the number of associated injuries, the number and type of fractures, and the amount of bone and soft tissue loss, especially after irrigation and debridement, all affect fracture healing. In this study, there were multiple confounding variables. The most striking is the

choice and type of IM nail technique chosen for definitive fracture fixation. The choice of an

unreamed or reamed nail was left up to the surgeon and not specified in the protocol. While the effect of this is an ongoing debate, there are varying thoughts as to the effect of this on fracture healing and the risk of infection depending on the nail technique used. In this study, there was a statistically significant difference between the standard of care group and the groups receiving the rhBMP-2 treatments. The two treatment groups had greater

numbers of patients receiving reamed nails as compared to the standard of care group. Although

it is not statistically different, it may be clinically significant that patients in the larger-dose rhBMP-2 group received a slightly

file:///C|/Storage/1121orth.txt (80 of 309) [12/2/02 8:59:20 AM]

file:///C|/Storage/1121orth.txt

81 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 higher percentage of larger-diameter, unlocked nails. There were other more subtle but possibly confounding variables associated with this study. For example, all fracture types, fracture locations, and patterns were considered collectively. It is well-known that the purpose

for using classification systems is to direct treatments so that as patients fall into these categories would be expected by experience to have different outcomes and risks of adverse events. Yet in this study, all fracture types and patterns were considered together. This is exemplified by

the fact that patients with Gustilo Types I, II, and IIIA were considered as one group. This

contradicts the concept of the classification system. In addition, patients with isolated fractures of the tibia were grouped with patients with multiple and varied other injuries. Both of

the rhBMP-2 groups had lower numbers of patients with associated multiple injuries and more patients with isolated tibia injuries. As a minor point, the protocol allowed for less than a full sponge to be implanted if the

file:///C|/Storage/1121orth.txt (81 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

82 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 fracture defect or configuration did not allow the implantation of the full sponge. Therefore, not

all the patients received the same dose of rhBMP, although a majority did receive a full sponge. Whether this had any effect on the rate of healing is not known. Finally, trauma physicians have different philosophies for the treatment for different fracture types, which may differ significantly across different parts of the world and across the United States, East and West Coasts. Cultural and

geographic differences across the world influence patient expectations for treatment outcomes, which also may affect study outcomes. It is also well-known that experience can play a role in the decisionmaking process and the patient outcome in trauma. In this study, larger

centers with greater numbers of patients may have a different experience than smaller centers with fewer trauma patients. Although the differences for these variables discussed were not large enough to establish statistical significance, such as in the case of the choice of nailing techniques, when considered collectively, there remains a concern

file:///C|/Storage/1121orth.txt (82 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

83 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 that many variables impacted clinically on the final outcome. [Slide.] Because of the divers cultural expectations, treatment philosophies and geographic locations of the investigational sites, the poolabilities of these patients and the applicability of this population to the U.S. population was an issue. As seen in this chart, approximately 50 percent of the patients were enrolled in two of the countries. South Africa contributed 138 patients, The remainder of the countries

and Germany, 83.

contributed fewer than 40 patients, with several site contributing fewer than 10 patients each. Centers with few patients were pooled with centers contributing large numbers of patients. As has been stated, often, larger trauma centers have different experiences in outcomes as compared to centers who do not have large numbers of trauma patients. [Slide.] Patient characteristics actually did differ from country to country. For example, the

proportions of patients who smoked varied in each

file:///C|/Storage/1121orth.txt (83 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

84 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 of the countries, ranging from 31 percent in the Scandinavian countries to 64 percent in South Africa. Fewer patients with isolated fractures The numbers

were enrolled in Australia and Israel.

of isolated fractures ranged from 49 percent in Germany to 75 percent in South Africa. The types of fractures treated in each country were also varied. no For example, there were

Gustilo IIIB fractures in France, but there was

a great incidence, about 38 percent, of Gustilo Type IIIB fractures enrolled in the Scandinavian countries. Some of the centers used reamed nails, This difference

and others used unreamed nails.

was also noted among the treatment groups. When results are reviewed by investigational sites, differences in outcome are apparent from site to site. The ability to

extrapolate the patient population to the U.S. population was not clearly validated in this submission. Although the sponsor did not consider the differences in these populations to be clinically relevant, the multiple differences bring into question the ability to pool these patients and distinguish between the confounding factors to

file:///C|/Storage/1121orth.txt (84 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

85 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 used. Patients were assessed by investigators both clinically and radiographically, and their x-rays were assessed by an independent radiology panel. These were then combined in the CCRE or the establish a treatment effect. [Slide.] Now let's look at the assessment methods

Combined Clinical and Radiological Endpoint. Now let's consider this aspect of study design in detail. Clinical assessments by the

investigators included an assessment of fracture site tenderness, an assessment of healing status by x-ray, and the patient's weigh-bearing status. [Slide.] Our concern about the pain assessment is as follows. Pain was assessed by each investigator

who had to interpret the patient's subjective report of pain upon direct palpation of the fracture site by the investigators. There was no

standardized or objective measurement scale used to quantify this parameter. In fact, there were no

standardized objective methods to define the intensity of pain and compare it to initial fracture site tenderness or differentiate fracture

file:///C|/Storage/1121orth.txt (85 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

86 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 site pain from referred pain from other injuries or soft tissue injury. [Slide.] This is an example of the case report form that investigators filled out for each patient. you can see in the top portion of the slide, fracture site tenderness was documented as either "absent," "not evaluated" or "present." You can see in the bottom portion of the slide that weight-bearing status was also determined by the investigator, but the progression of status from non-weight-bearing to touch-down to partial weigh-bearing and then to full weight-bearing was not based on any objective or standardized criteria. [Slide.] This slide shows the criteria for radiographic union that the independent radiology panel used to determine whether a fracture was united. The objective criteria as seen in the As

slide consisted of three or four cortices demonstrated cortical bridging and/or the complete disappearance of fracture lines and was delineated in detail in the independent radiology case report form along with other radiographic determinations

file:///C|/Storage/1121orth.txt (86 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

87 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 including hardware failure. [Slide.] However, on this slide, as can be seen in this case report form the investigators filled out, there is no indication of how the decision of union was made and whether the decision of radiographic healing was based on any objective criteria. In a subsequent submission, the definition for radiographic healing was provided in the second submission. [Slide.] The difficulty in interpreting results without precise criteria for all evaluators is demonstrated by looking at this case study provided by the sponsor in the submission of a Grade IIIA tibia fracture which was treated with a sponge soaked in the .75 dose of rhBMP-2. There was a

clear and significant discrepancy between the determination by the investigator and the independent radiology panel, as seen in this slide. [Slide.] You will note that the investigator considered this Grade III fracture united at only 10 weeks, while the independent radiology panel considered this fracture healed at 26 weeks. You

file:///C|/Storage/1121orth.txt (87 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

88 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 will note that the postoperative radiograph and the 6-week radiograph are not provided here. [Slide.] The definition of a healed fracture was specified in the protocol, as it is here. However,

whether all three criteria were expected in order to consider a fracture healed was not. Although

provisions were made for patients with different weight-bearing status due to other injuries, it is not specified whether all three criteria had to be satisfied to be considered a healed fracture. The distinctions between delayed union based on time or other criteria and a healing fracture were not defined well in the original protocol. A post-hoc analysis states that every

patient recommended for secondary intervention failed to meet at least one of three criteria defining fracture healing. [Slide.] This is the definition of delayed union that was provided in the protocol: "A fracture is

considered a delayed union if insufficient fracture healing was observed as determined by the investigator's radiographic and clinical assessment."

file:///C|/Storage/1121orth.txt (88 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

89 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 This definition of delayed union is vague. There are no objective criteria specified in the protocol that defined what was to be used to make the decision of delayed union. For example, there

is no delineation of the time or interval course to establish that a fracture was delayed in healing, nor are there clinical and radiographic criteria to make this decision provided. This is not clearly differentiated from fracture non-union. How the decision was made to

recommend a secondary intervention based on this definition is not specified in the protocol. [Slide.] It is not clear what criteria the decision by investigators to recommend a secondary intervention was actually based on. The ambiguity

can be seen in this table which shows that secondary interventions were based on the failure of one or two or three of the clinical criteria. In addition, the sponsor provided data tables that showed that patients may be considered healed based on one or two or three of these criteria. [Slide.] So the question still remains how was the

file:///C|/Storage/1121orth.txt (89 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

90 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 decision to recommend a secondary intervention made. The imprecise guidance of the protocol and

lack of standard criteria for assessing patients, coupled with the fact that investigators were unblinded to the treatment in determining which patients should receive this secondary intervention, are subject to potentially significant investigator bias. [Slide.] Next, the choice of control group should be considered. All the different types of

fractures and different types of wounds were treated with the same treatment protocol. Fractures are graded differently because different types are expected to have different prognostic rates for healing and adverse events. Therefore,

they may need different treatment regimens. For example, the incidence of infection differs between Gustilo Type I and Gustilo Type III in the literature. In this investigation, the

population of fractures was comprised of 40 to 45 percent Gustilo Type IIIA and B fractures. Whether

the standard of care is applicable to all grades of fractures and wound types is not clearly justified. As we have already heard, the use of

file:///C|/Storage/1121orth.txt (90 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

91 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 prophylactic bone graft may be considered controversial. There are authors who do recommend

the addition of autologous bone graft, recommending evaluation for this at 2 to 6 weeks following wound closure. This literature review was provided for

you in the panel pack. The use of prophylactic bone grafting has been shown to enhance healing and may be associated with a reduced risk of infection in some studies. The inclusion criteria as stated by the sponsor stated that patients with anticipated treatment plans which included additional procedures to promote fracture healing, such as bone graft, were excluded. This decision, however, was only relevant to that considered at the time of definitive wound closure, when in practice, this may often not be made until the wound is closed and the initial fracture progress has been assessed. Regardless of this controversy, I would like you to consider that while the treatment groups in this investigation received a substance that would potentially increase bone formation or enhance bone healing--namely, rhBMP-2--the control group did not.

file:///C|/Storage/1121orth.txt (91 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

92 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 All of this raises the question of whether the standard of care group was treated comparably to the rhBMP-2 treatment groups and therefore whether the study results between the treatment arms can effectively be compared. [Slide.] The clinical relevance of the endpoints was also considered. These are placed on this These are

slide for your convenience and memory.

the primary and secondary endpoints that were specified in the protocol. Realize that the

primary endpoint consists of four subgroups. [Slide.] What is important in fracture healing studies, though? the primary focus. Fracture healing. This should be

A review of the literature

shows that the commonly determined outcomes include the incidence of union and nonunion as a function of time, the incidence of infection, and the time to healing, usually comparing one method of fixation to another, or between fracture types. The importance of some of these effectiveness endpoints are secondary in this investigation. [Slide.]

file:///C|/Storage/1121orth.txt (92 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

93 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 The combined clinical and radiographic endpoint was intended to support the observations made by the treating surgeons with those made by the independent radiology panel. However, this

combination has not been validated as an analytical method. There are several potential problems with using this method of corroborating investigators' assessment. First, the CCRE is composed of

potentially biased subjective assessment compared with a more objective assessment. Second, the CCRE combines two dissimilar assessments--a clinical assessment with a radiologic assessment--instead of comparing two radiologic assessments. Lastly, patients with secondary interventions were not treated the same as patients who did not have secondary interventions. Patients

with secondary interventions were paired with the results of independent radiology review at the time the decision for secondary intervention was made, while the patients with no secondary interventions were paired with the independent radiology evaluations at 12 months. [Slide.]

file:///C|/Storage/1121orth.txt (93 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

94 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 The treatment of missing data related to the success and failure was inconsistent in some cases, and this may also affect the result. The number of patients who met the criteria for treatment and follow-up defined in the evaluable patients, was 404 patients out of the total 450 patients, or 89 percent. When evaluating results, missing patients or missing data were treated differently in three cases that I will give examples of. For the data

analysis of the rate of secondary interventions, patients with no data or with no outcome were considered in the category of patients with no secondary interventions. This may falsely elevate

the number of patients and the rates of success. In another case, in the analysis of the independent radiology panel results, 44 patients or 10 percent with missing 6-month data for united fracture assessment had data carried forward from previous visits. About two-thirds of the patients This practice may

were in rhBMP-2 treated groups.

have overestimated the not united rates in this and subsequent visits. In the third case, when reporting the summary of combined clinical and radiographic

file:///C|/Storage/1121orth.txt (94 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

95 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 endpoint results, missing patients with no clinical outcome or no radiographic outcome were included in the category of no secondary intervention/not united. Again, the failures may have been

overestimated in this particular case. [Slide.] Therein lies the dilemma. All of these

factors related to study design and protocol make evaluation of the results of the study and the confidence that the results are truly an effective treatment quite a challenge. The fundamental issues include the fact that investigators were unblinded; they made the assessments related to the determination of the primary endpoint with imprecise, not specifically spelled-out guidelines for making their decisions. While this may be acceptable for clinical practice, it may not be appropriate for a controlled clinical trial. Even the attempt to support the investigators' assessment, the CCRE, is partially based on the investigators' assessment as well. Additional design issues of the protocol that led to uncertainty that the protocol provided rigorous and explicit guidelines were provided to

file:///C|/Storage/1121orth.txt (95 of 309) [12/2/02 8:59:21 AM]

file:///C|/Storage/1121orth.txt

96 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 all investigators are listed here. These include

the fact that no time course was defined to distinguish delayed healing from aggressively healing fractures. There were no precise

radiographic or clinical criteria to separate the healing fracture from one that was considered to have delayed healing. How the patients with delayed healing fractures were recommended for secondary interventions is imprecise. The sponsors did not

provide adequate evidence to provide a confidence that the extent to which patients were recommended for secondary intervention were guided by the same criteria across all investigational sites. [Slide.] Looking at the effectiveness results has also raised some issues to bear in mind if one gets past the issue of study design. Depending on which

analysis is reviewed, different results may be reported. These analyses include the primary

endpoint, the rate of fracture healing, the time to event analysis, and the 50 percent probability of healing, as well as the incidence of non-union. Each of these will be considered. [Slide.]

file:///C|/Storage/1121orth.txt (96 of 309) [12/2/02 8:59:22 AM]

file:///C|/Storage/1121orth.txt

97 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 The overall rate of secondary interventions is highest in the control group. Recall, however, that in addition to the number of interventions recommended and performed, the overall rate of secondary interventions included the number of self-dynamizations or screw breakages, the number of secondary interventions recommended but not performed, and the number of secondary interventions not recommended but performed anyway. The 0.75 mg/ml dose treatment group had the most patients in the latter category. self-dynamizations or hardware failures are excluded, the differences between the groups is reduced. It is not clear why patients were If the

included who had secondary interventions recommended but not performed, or those who had secondary interventions with no recommendation were included. This quandary further underlines the

problem with this primary endpoint. [Slide.] If we look at the rate of fracture healing at 6 months or 26 weeks, this was also a secondary endpoint in the study. This rate differs depending

on whether the investigator or the independent

file:///C|/Storage/1121orth.txt (97 of 309) [12/2/02 8:59:22 AM]

file:///C|/Storage/1121orth.txt

98 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 result. radiology panel made the determination, as seen in this slide. The rate determined by the radiology panel was significantly lower, and the differences between groups was lower than by the investigators' assessments. The low success rate indicates that a

choice of 6 months may not have been an appropriate endpoint for this study. The discrepancies at 39

weeks and 50 weeks are progressively less, as has already been discussed. [Slide.] The time to healing yields yet another On this graph, the red line represents the

course of the 1.5 mg/ml dose rhBMP; the yellow line represents the 0.75 mg/ml group, and the black line represents the standard of care group. It is obvious that there is little difference between the control and the lower dose of rhBMP throughout the time course of the study. However, we should look at two example probability points. The sponsor has already reported that the 50 percent point of probability of healing represents that lower line that traverses horizontally across the graph. There is a

file:///C|/Storage/1121orth.txt (98 of 309) [12/2/02 8:59:22 AM]

file:///C|/Storage/1121orth.txt

99 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 chart. difference between the groups at this point. However, this may not be a point that clinicians will consider relevant. If we look at the 75 percent probability of healing or the 80 percent probability of healing which is represented by the horizontal line that is above that, there are no differences between the groups. [Slide.] This is clearly demonstrated on this Although there is a difference between the

standard of care and the higher doses of rhBMP at the 50 percent probability of healing, there is little difference in the days to healing at a point where there is 75 percent probability of healing. They are virtually the same between the groups. [Slide.] Now let's look at the same type of graph as shown by the determination of the independent radiology panel. If there were indeed a difference

between the treatment groups, this would be expected to be carried out by the detection of the independent radiology panel. I would like to draw your attention to the 180-day mark on the lower asymtope [phonetic] of

file:///C|/Storage/1121orth.txt (99 of 309) [12/2/02 8:59:22 AM]

file:///C|/Storage/1121orth.txt

100 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 the graph, which corresponds to the 25 percent probability of fracture healing. At this time

point, there was no difference between the groups. [Slide.] This is clearly evident on this table, which shows that the days to healing at either the 25 percent, 50 percent, or 75 percent probability are almost identical for each group, regardless of the probability chosen. [Slide.] Lastly, I would like to discuss the issue of nonunion. Overall, at 12 months, the nonunion

rates as determined by the independent radiology panel were 53 percent of the patients in the standard of care group, 48 percent in the 0.75 mg/ml group, and 38 percent in the 1.5 mg/ml group. These rates are slightly lower by the investigators' assessment. By either assessment, these rates appear high, and perhaps one explanation may be that the results reflect the diverse fractures treated in this study or other multiple factors previously mentioned. Patients with secondary interventions were an interesting group to review. These patients who

file:///C|/Storage/1121orth.txt (100 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

101 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 graph. actually had a secondary intervention to promote healing had similar nonunion rates whether they were treated with the rhBMP-2-impregnated sponge or not. [Slide.] This is clearly demonstrated on this It is of interest that the rate of nonunion

in patients with secondary interventions is clearly demonstrated by--again, the black line is standard of care, and the red is the 1.5 mg/ml dose. Patients treated with rhBMP-2-impregnated collagen sponge who required a secondary intervention were considered healed later than patients who did not. This is shown in this graph

and noted in the chart previously. [Slide.] Let's move on to the safety issues that concerned us. There were no apparently However, there

life-threatening safety concerns.

are some issues which are not explained and whose clinical significance remains unclear. I will

briefly touch on each of these issues and the questions that were raised. [Slide.] The first issue deals with serology

file:///C|/Storage/1121orth.txt (101 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

102 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 results. The rates of authentic antibody response

to rhBMP-2 and Type I bovine collagen are higher in this study than in published data in other rhBMP-2 in collagen sponge investigations. The rate of

authentic antibody responses to rhBMP-2 in treated patients in a recently published study using rhBMP-2 and the collagen sponge in an anterior lumbar spine fusion study was 0.7 percent. study, 12 patients had an authentic immune response. In the 1.5 mg/ml rhBMP-2 group, the rate In this

of authentic responses was 6 percent and one percent in the control population. This is an

increased rate as compared to the previous study. What the contribution of the trauma setting is to these results is not clearly known. The fact that there are antibody responses is the concern, especially since we don't know what the clinical significance of these results is. The

numbers of patients may also be too small to make any sound conclusions, however. There were 60 patients who developed antibodies to Type I bovine collagen and approximately half of the patients had persistently elevated antibody titers 20 weeks or longer. the rhBMP-2 lumbar fusion study, the In

file:///C|/Storage/1121orth.txt (102 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

103 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 investigational patients had a similar rate of authentic positive immune responses to Type I bovine collagen. Again, the clinical significance of this is difficult because the small respective numbers and the respective significance is unknown. In

both cases, however, the analysis of the adverse events did not seem to correlate with the presence or absence of immune response. [Slide.] The rate of hardware failure was relatively high in this study. There were 48 screw

breakages or bending noted in the standard of care group, 31 in the middle dose, with two nail fractures in this category, and 24 screw breakages in the 1.5 mg/ml dose. slide; sorry. Hardware failure consisted mostly of bending or breaking of locking screws. Whether There is a typo in this

this hardware failure is a function of the hardware used to treat the fracture, the type and size of nail used and thus the screw size used, or a function of the high degree of ununited or non-healed fractures is unclear. [Slide.]

file:///C|/Storage/1121orth.txt (103 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

104 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 The one adverse event listing in which there was a statistically significant difference between the control and the investigational groups is the rate of abnormal lab values related to liver and pancreatic function--specifically, elevated amylase and hypomagnesemia. There were 30 patients in the rhBMP-2 treated groups with elevated amylase as compared to five in the SOC group. There were two patients in

each of the standard of care group and the higher-dose rhBMP group compared to 11 patients in the middle-dose rhBMP group with a finding of hypomagnesemia. Although the sponsor states that the patients had no overt manifestations of pancreatitis, the concern stems from published preclinical data that showed pancreatic cell lines, showed increased cellular proliferation when exposed to rhBMP-2. The significance or

association to this population is not clear. Coincidentally, one patient in the previously-conducted study using rhBMP-2 in the spine was diagnosed with pancreatic cancer during the duration of the study. The remaining liver functions were

file:///C|/Storage/1121orth.txt (104 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

105 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 elevated in a large number of patients, but this could possibly be explained by soft tissue trauma and not due to the biologic that was added. Whether any of this is clinically significant due to trauma or the exposure of rhBMP-2 is unclear, but it should not be dismissed. [Slide.] A total of 17 patients experienced at least one event classified as heterotopic or ectopic ossification or callous.s The patients in

the 1.5 mg/ml rhBMP-2 group had the highest number and percentage of patients with heterotopic ossification, but no action was required to treat any of these heterotopic ossification-related events. I would like to point out, however, that according to the sponsor, heterotopic and ectopic ossification was not a significant concern. Although the sponsor stated that no ectopic ossification was reported, the table in the submission includes a patient in the category titled, "Ectopic ossification/other sites." Additionally, it is not clear how the sponsor ruled out heterotopic ossification in other areas of the tibia. Sponsor had told us that no

file:///C|/Storage/1121orth.txt (105 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

106 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 additional radiographic physical exam or serum testing was done to observe heterotopic bone formation in other areas away from the fracture site studied. We believe it would be difficult to

determine the presence of heterotopic bone or even distinguish it from myecitis ossificans secondary to trauma on history alone. It would be difficult

to determine whether the ectopic ossification was due to trauma to the soft tissues or bone in the area was an effect of increased bone formation in response to rhBMP-2. With the limited surveillance that was done in the study, one cannot comment on the existence of distant ectopic ossification in other sites of the body. [Slide.] The overall infection rates were comparable between the three groups when the denominator is the intend-to-treat population, with the standard of care group experiencing slightly more infections both overall and in the region under study. The figures for the leg tibia category do not include skin infections noted as wound dehisants, gangrene, inflammation or necrosis.

file:///C|/Storage/1121orth.txt (106 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

107 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Overall, these rates seem somewhat higher than in other multiple fracture and single fracture comparison studies reviewed in the literature. When broken down by Gustilo classification, the rates are also high. Especially concerning are the

rates in the Gustilo Types I and II fracture categories. are so high. Although the small sample size may be a factor in these rates, one should consider what other factors may also play a part in these higher rates. Although some of these safety uncertainties It is not really clear why these rates

may be consistent with the trauma population studied or may be in too low of a sample size to be of statistical significance, these issues should still be considered when evaluating this device. [Slide.] In conclusion, we agree that trauma patients represent a difficult population to study even in a well-controlled trial. the multiple confounding factors. no exception. It is difficult to make comparisons between this study and the literature because of the multiple varied factors that confound or This is due to This trial was

file:///C|/Storage/1121orth.txt (107 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

108 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 influence the outcome. Other additional factors

contribute to the challenge of evaluating this device in this study. On our part, this is mainly

centered on factors related to study design. Objective criteria for making assessments were no clearly defined, and therefore, the judgments that were made were not clearly defined. The protocol was not specific in instructions to the investigators, especially regarding the distinction between a healing fracture and delayed union. There is some debate

as to whether the control group represents the standard of care for all the fracture and wound types described in the patients in the study. Endpoints commonly found in the literature were not considered to be primary endpoints. For the

primary endpoint chosen, how the investigator came to the determination to perform an intervention was not specific. In the study, more importantly, there was no clear substantiation that all investigators used the same criteria for making the decision for secondary interventions and that pooling across so many different sites was justified. When looking past the design issues at the

file:///C|/Storage/1121orth.txt (108 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

109 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 effectiveness results and some aspects of the safety profile, further questions were raised. All

these factors caused considerable uncertainty as to the actual treatment effect and the safety profile of this device. Thank you for your attention. DR. YASZEMSKI: Dr. Lao? DR. LAO: Good morning. My name is Chang I have been Thanks, Dr. Buch.

Lao, and I am a statistician at FDA.

working at FDA for the Center for Drugs and Devices for a long time. Today, my presentation will concentrate on the statistics evaluation of the primary endpoint, which is the proportion or rate of secondary intervention. [Slide.] I will concentrate on three major issues for this PMA. The first issue is pooling of the

multi-clinic data or multi-national data; the second issue is the reproducibility study, intraand inter-observer agreement; the final issue is the survival analysis, time-specific, versus crude event, which is secondary intervention, probability.

file:///C|/Storage/1121orth.txt (109 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

110 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 [Slide.] As you can see, on the first issue, pooling of data, I have three charts. The first

one shows ideal conditions--the two lines are parallel--between Center 1 and Center 2. In both

of them, the high dose is better than control in terms of the percent of secondary intervention success. The one in the middle is acceptable--we call it quantitative center by treating interaction. The distance is not parallel, but at

least they go in the same direction between the two sites. The last one, the crossover, which is questionable, we call qualitative interaction. The

primary endpoint varies from the opposite direction between Center 1 and Center 2. [Slide.] This is a hypothetical example of wrong pooling of the data. In Site 1, you can see two

success and failure proportions between high dose and control are identical--33 percent each. In

Site 2, they are also the same--67 percent--and the P value in both of them equals one. equals zero. Chi-square

So if the P value equals one, it is

file:///C|/Storage/1121orth.txt (110 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

111 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 not significant. But what happens if we pool the other numbers together or the observations together? see the last table, where percent success is 58 percent for high dose and 45 percent for control. Chi-square equals 5.56, and P value is 0.018, which is significant. So the question here is can you add all the numbers together and do the pooling of the data. So this kind of pooling, adding some of the You

observations together, is statistically invalid. We should use a better approach here. [Slide.] This is [inaudible] pyramid data, which is the estimated difference, high dose minus control in percent success if we have secondary intervention, and the exact 95 confidence interval by regrouped investigator sites. of about 30 regrouped sites here. We have a total By definition,

"regrouped" means any country in which the number of patients is less than nine patients across regrouped were pooled into a combined site. As you can see, in this chart, the confidence interval, which was based on exact binomial distribution because of smaller sample

file:///C|/Storage/1121orth.txt (111 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

112 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 size with each regrouped site-about seven or

eight of them, the point estimate, which is the dot on the left of the zero, which means the point estimate of difference between the two success proportions in favor of the control. There are 22

or 23 in favor of the high dose--but this is just point estimate based on the data. We have to

consider the uncertainty of the estimate, because we have to deal with the sample size. That is what

the 95 percent confidence interval tells you. You can see the width of the confidence interval is wide in most cases because of the smaller sample size. include zero. Almost 29 out of 30 of them

So statistically speaking, none of

them is significantly different from zero between control and high dose. [Slide.] The second chart is a different data base. It is the combined clinical radiological endpoint or CCRE patients. pattern. Pretty much you see a similar

About 10 or 11 of them go in the negative

direction, to the left side of zero, and about 20 of them go to the right side of zero. But the

confidence interval, again, about 29 of them include zero.

file:///C|/Storage/1121orth.txt (112 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

113 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 So if you were to predict success by site, none of them can stand alone in favor of the device. Now, what happens--can we combine them? [Slide.] Okay. There are two statistical models The

used here, what we call the meta-analysis.

meta-analysis is a very big issue in clinical trials. If you do a MedLine search, there are

probably over 1,000 articles dealing with meta-analysis. Here, I present two models. fixed model. One is the

The assumption in the fixed model is

that each center has the same clinical effect, considering within-center variability only. Random effect is assumption of very diverse nature in study design, methods among sites and heterogeneity among sites, so we consider both within-center and between-center variability. So which ones are not appropriate in here? From the previous two charts, you can see some things, but we can do a formal statistical test. [Slide.] Here, the meta-analysis is of 30 regrouped investigator sites. In our analysis, we excluded

file:///C|/Storage/1121orth.txt (113 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

114 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 effect? 0.04. P value is significant, less than 0.01. We just divided by the mean difference divided by standard error, which is a T statistic with 26 degree of freedom. And the 95 percent confidence interval, which does not include zero, is in favor of the high dose, 0.039, 0.194. But what happens if we allow the random We consider not only within-center three regrouped sites because they had zero variance in the difference of the two proportions. It means you have both zero percent success or 100 percent success in the control and high dose. So our analysis included 27 regrouped sites, and we used the method of moments suggested by DerSimonian and Laird in their Statistics in Medicine 1986 paper. [Slide.] Here are the results of the meta-analysis for 27 regrouped sites. For the fixed effect, the

mean difference--meaning difference being proportion success--which is high dose minus control, equals about 11 percent difference. Standard error of estimated difference is

file:///C|/Storage/1121orth.txt (114 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

115 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 variability but also between-center variability, which is a mean difference of about 12 percent. Standard error is larger because we include between-center variability here. The P value is

not significant, and the 95 percent confidence interval was inclusive, from minus 0.04 percent to plus 0.28 percent. And at Chi-square equal 97, which is highly significant, less than 0.005 P value, what that means is reject the null hypothesis, which is between-center variance equals zero. So by this test, it means there is significant variability from center to center. would prefer the random effect model in this example. [Slide.] The second chart for the CCRE data pretty much has the same conclusions as the regrouped investigator sites, as you can see here. The random effect in the 95 percent confidence interval does include zero also. The We

Chi-square test is very highly significant and rejects the null hypothesis of no difference in terms of variability among sites. [Slide.]

file:///C|/Storage/1121orth.txt (115 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

116 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 The second issue here that I am going to discuss a little bit is the reproducibility study intra- and inter-observer agreement. In the PMA, we have a total of 60 patients from 10 U.S. trauma centers. Please remember that

this is not from PMA data, not from the pivotal study; this is from a different protocol, from a previous protocol study. So the two teams of multiple raters--three raters, actually--fracture union, between observer, inter-observer, Kappa statistics--I am going to explain what Kappa means. Kappa says that the

higher the number, the better; 1.0 is perfect agreement. Zero is no agreement at all. So 0.87

for the inter-observer agreement and 95 confidence interval of 0.7 to 1.0, based on 20 patients. The last two were intra-observer, with the same observer, the first study and the second study, about 0.5, but the confidence interval, the lower and upper bounds are quite wide. The reason

is because of a smaller sample size, 20 patients, and some observed predicted proportion of agreement. [Slide.] This chart just gives reference to general

file:///C|/Storage/1121orth.txt (116 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

117 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 important. here. guidance about what Kappa statistics mean. This is

from Professor Gary Koch, 1977 Biometrics paper. He is a professor at the University of North Carolina Chapel Hill. This is general guidance. The one in the

middle, 0.4 to 0.6, is moderate; clinical interpretation is moderate agreement. For the

intra-, with the same observer, in the last table, we have two of them at 0.5. So the [inaudible] is So this is

like flipping a coin, 50 percent each.

just to give you general information here. [Slide.] The general question for the reproducibility study here is on sample patient selection. Twenty patients out of a total of 60

patients, which is not from the PMA data. And the random sample, is it representative to the reproducibility study, to the PMA data, or are they masking? And for time comparability--time is And you have multiple raters per team

I think that between-observer Kappa 0.87,

that sounds higher than the intra-observer Kappa 0.50. I think the reason is because it is based on If you had two out of three raters

majority rule.

file:///C|/Storage/1121orth.txt (117 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

118 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 agree, that is considered agreement; that is what I think. [Slide.] The final topic is survival analysis versus crude secondary intervention event probability. The life-table or Kaplan-Meyer analysis, we just recently received for the final three analyses--first, fracture healing assessed by investigator; second, radiographic assessment of fracture union by independent radiology panel; and the last was the CCRE study. [Slide.] On survival analysis in this PMA, for a long time, we hadn't seen the patient follow-up data and the survival analysis, so we did not know what patients were censored, lost to follow-up, missing, or at risk of secondary intervention at different time points. Also, we had to assume censoring independent of treatment. And if you look at the proportion of success to failure, we always consider what time you are talking about. Are you talking about 12

months, are you talking about 6 months, 3 months,

file:///C|/Storage/1121orth.txt (118 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

119 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 whatever? The crude secondary intervention event probability is always less than time-specific cumulative SI event probability by survival analysis, unless all patients had completed the entire follow-up study, or all patients with secondary intervention. This is not a very

practical situation in clinical trials. [Slide.] In conclusion, first, with study design. Heterogeneity among centers--I would think that random effect for combined analysis is not appropriate; and that direct adding up all numbers and the corresponding analyses is not valid; survival analysis is required due to patients censored or lost to follow-up; and questionable reproducibility studies. Finally, in the survival analysis, we think the assumption here sounds like the whole dataset comes from one center, not adjusted for center differences. But including a center in the

model is also not easy, because we have too many regrouped sites here--30 of them. You have to cut

down that number to include in the model. So the survival analysis here assumes data

file:///C|/Storage/1121orth.txt (119 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

120 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 very much. DR. YASZEMSKI: Thanks very much, Dr. Lao. from one site, all the numbers together. This is the end of my talk. Thank you

Right now, we're going to take a break for 5 minutes, and we're going to come back after the break and have the presentations from the panel reviewers, which will begin with Dr. Naidu's assessment of the preclinical studies. We'll see everybody in 5 minutes. [Break.] Panel Presentations DR. YASZEMSKI: I'll ask Dr. Naidu to give

his presentation now on the preclinical studies. Dr. Naidu? DR. NAIDU: Thank you.

My charge was to provide the panel and the audience here with a summary of preclinical results. I derived most of my information from the first volume issued by the sponsor. The sponsor

has already gone over some of the preclinical results, especially in an animal model, with the New Zealand adult rabbits. However, in the volume provided, there are

file:///C|/Storage/1121orth.txt (120 of 309) [12/2/02 8:59:23 AM]

file:///C|/Storage/1121orth.txt

121 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 various other animal models in which the BMP was tested, which included the Rhesus monkey model, the canine model, in addition to the rabbit osteotomy model, and also a goat tibia fracture study was performed. So I will touch on the fracture models

that were not addressed by the sponsor in their presentation initially. Of course, this is in a long-bone fracture healing model, and with regard to the Rhesus radial defect studies, unilateral radial defects 3.5 cm in size, four times the diameter of the bone, were created. These were internally fixed with a

fixation plate. From this, the sponsor reaches the conclusion that the bone-bridging was highly variable, inconsistent, no dose response was discernible. The only thing that one could

conclude from this Rhesus radial defect study was that the antibody responses to rhBMP-2 were higher in the treated group, whereas none of the control animals showed any of these antibody titers. These were low antibody titers, and they were detected in 35 percent of the treated animals. This was with the ELISA assay. With the Rhesus ulna defect study, the

file:///C|/Storage/1121orth.txt (121 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

122 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 size of the defect was not clearly defined in the material provided. Presumably, the ACS rhBMP-2

sponge was placed at the site of the defect, and the negative control was buffer with the ACS sponge. What the sponsor concluded was that the BMP-2/ACS sponge was replaced with a dense population of spindle-shaped cells, presumably capable of continued bone formation. However,

considerable compression of muscle into the center of the defect was noted, limiting the volume of potential bone formation. My conclusion--nothing happened. Therefore, in the primate study and the preclinical studies, as designed by the sponsor, I could not conclude much except for the fact that the antibody titers to rhBMP-2 were apparent in 35 percent of the Rhesus monkeys. The next study that I want to touch on is the canine radial defect study. Bilateral radial

osteotomies in mature hound-type dogs with 2.5 cm diafacile [phonetic] critical-size segmented defects were stabilized with external fixation. One limb received the rhBMP-2/ACS or the ACS and buffer as the negative control. The contralateral

file:///C|/Storage/1121orth.txt (122 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

123 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 limb received autogenous bone graft harvested from the humeral head. The rhBMP-2 concentration studied with 0.05, 0.2, 0.8 mg/cc. In all concentrations of

rhBMP-2, radiographic union was achieved at 12-week time point. Trigonal failure of all three rhBMP-2 dose groups were equal to the contralateral limb treated with the autologous bone graft. It was

superior in the 0.05 and 0.8 mg group, but there was no difference in the 0.2 mg group. In conclusion, the radial density in the period from 4 to 12 weeks was essentially equivalent in the rhBMP-2 treated defects and defects treated were autographed. The total

energy-to-torsional failure of three dose groups was equal to in one group and superior to the contralateral limbs treated with auto bone graft in two of the dose groups, especially in the 0.05 and 0.8 mg. There was a longer-term follow-up for 24 weeks in some of these animals, and at 24 weeks, biomechanical testing demonstrated no significant differences between the rhBMP-2 group and the autogenous bone graft group. Of course, the

buffer, the negative control group, did not heal.

file:///C|/Storage/1121orth.txt (123 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

124 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Therefore, the rhBMP-2 is equivalent to the autogenous bone graft group in this canine radial defect study. Going on to the canine femoral intercalary allograft incorporation study, the goal was to compare the effect of augmentation of the host bone allograft junctions with rhBMP-2 and compare this to the augmentation with autogenous bone graft. Six-centimeter midshaft femoral defects in 21 mixed-breed dogs were fixed with frozen allograft stabilized with an interlocking nail. Host

allograft junction was augmented with rhBMP-2/ACS at a dose of 1.15 mg total dose or buffer ACS, and of course, the last group was the autogenous cantelas bone graft. This augmentation was done at

the proximal site and also at the distal site. Animals were euthanized at 24 weeks post surgery. At the proximal junction, the BMP/ACS torsional strength was significantly greater than the cantelas bone graft group or the buffer group. However, at the distal junction, there was no difference between the BMP group and the cantelas graft group. Both the BMP and cantelas bone graft groups were torsionally better than the negative

file:///C|/Storage/1121orth.txt (124 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

125 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 study. control group at both the proximal and distal junctions. These are all mechanical studies, and these are the conclusions reached from the mechanical studies. The next is the goat tibial fracture This was a closed fracture model.

Bilateral closed midshaft tibial fractures were created using a three-point bending jig. rhBMP-2/ACS combination or buffer/ACS combination was placed at the fracture site in one limb, either by wrapping it or by doing as an onlay. And the

contralateral limb served as a surgical control. The tibia was stabilized with an ex-fix. a closed fracture model again. X-rays at 6 weeks showed that all three groups had healed, and differences are hard to discern by x-ray. Biomechanical testing showed This was

that fractures that were treated with rhBMP-2/ACS had greater torsional toughness relative to the untreated controls at P equals 0.017. However, fractures treated with buffer/ACS group, the so-called negative control group, also showed a trend toward increased torsional toughness, albeit with a slightly lower P value.

file:///C|/Storage/1121orth.txt (125 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

126 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 So you can't conclude anything. On the other hand, [inaudible] showed that torsional toughness in the rhBMP-2/ACS fractures wrapped with the sponge was significantly higher than those treated with an onlay. Whether they

contained BMP or not did not make a difference. Who knows what one can conclude from this? I'll let Dr. Kinley Larntz comment with regard to stats. Lastly, the rabbit ulna osteotomy model, which was shown by Dr. Riedel from the sponsor--he showed a beautiful radiograph which showed that it had healed; he showed a nice histology slide which showed bridging callous albeit not complete healing in the standard treatment group. But here are the numbers. Of course, the

mechanical testing of this fracture is the gold standard. On Volume 1, page 35, the bar graphs are

quite clear with regard to torsional strength testing. In this bilateral mid-ulna osteotomies

group, in 72 male rabbits with one millimeter defect, there were three treatment groups--rhBMP-2/ACS onlay; buffer/ACS onlay; and no treatment. Torsional loading was performed in treated

file:///C|/Storage/1121orth.txt (126 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

127 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Naidu. We're now going to proceed to Dr. Finnegan and her clinical review. Dr. Finnegan? DR. FINNEGAN: I have been asked to do the limbs at 2 weeks, 3 weeks, 4 weeks, and 6 weeks. Failure torque was no different at 2 weeks for any of the groups--of course, that is reasonable. It

was significantly higher in the BMP-2/ACS group at 3 and 4 weeks. But at 6 weeks, again, there was no

difference between any of the three groups tested, and at 6 weeks, in fact, no matter what you did to it, it was similar to the intact group. That is my summary on the preclinical animal trials. In conclusion, the preclinical

animal trial studies are highly variable, mixed. My gut feeling is that the rhBMP-2 does enhance bone formation, but practical clinical use is going to be very tough to show, based on the preclinical data. Thank you. DR. YASZEMSKI: Thanks very much, Dr.

clinical review, and as I started into my homework, I realized two things--one, I needed to figure out whom I had annoyed, and two, I needed a large

file:///C|/Storage/1121orth.txt (127 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

128 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 bottle of aspirin. You are going to hear some repetition, but I think some of that repetition is important. Reviewing the material for clinical trials, the initial trial was 12 patients who were open-label, as was discussed. The dose there was

0.43, which is much lower than was in the pivotal trial. This was done for safety and efficacy, and healing and no

there was no difference between adverse effects.

The next study, actually, that is mentioned, which was not presented by the sponsors was a study using open tibial shaft fractures and an external fixeter, with the same doses that they presented in their pivotal study, and this was terminated because of patient recruitment problems. That brought them to the pivotal study, which is classed as a Phase 3 multi-center prospective, randomized but single-blind, that being the patient only, and all of the Gustilo grades that you have heard. nailing. The control was labeled standard of care, and this was limited to wound coverage and an IM nail. These were treated definitively with IM

file:///C|/Storage/1121orth.txt (128 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

129 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 The experimental--and from now on, I am going to call this the "composite"--is the Helistat absorbable collagen sponge with the recombinant BMP in the two doses that have been described. According to the sponsors, the BMP has left the site at 4 weeks, and the sponge is gone at 8 weeks. They define the material as

osteoinductive, stating that it produces trabecular bone, and I think you saw that fairly nicely on their histology slides. However, this is not the normal bone for this diafaceal [phonetic] region, and there does not appear to be much work done on the process of remodeling afterward. In the material that I received, the sponsors outlined very nicely five objectives, and what I would like to do is outline those objectives, and at the end of my talk, I will discuss whether I think they were met or not met. The first objective was to increase the likelihood of healing with the composite. The

second was to define the safety of the composite. The third was to document that this increased healing was actually present at 6 months. The

fourth was to document that radiographic union was

file:///C|/Storage/1121orth.txt (129 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

130 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 observed earlier in the composite group. And the

fifth was to evaluate the potential economic benefit of treatment with the composite as compared to the control. Really interesting for the exclusions to their patient group which they did not bring up in their presentation was that they excluded any history of exposure to silicon or injectable collagen as well as hypersensitivity to monoclonal antibodies or gamma globulin. Their protocol as they have discussed had some interesting components to it. The first is

that although at initial assessment, a Gustilo grade was given the final grade was not done until their definitive wound closure, which needed to be carried out within 14 days. As has been previously discussed, intermedullary rodding was allowed to be picked by the investigator whether it was reamed or not reamed, and this was not controlled. Also what is interesting is that a very small percentage of the patients received their IM rodding at their first intervention, although the largest number of the patients were Grade I and Grade II Gustilo class. And I respect Dr.

file:///C|/Storage/1121orth.txt (130 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

131 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Swiontkowski immensely, and I have known him for a long time, but I actually do not think this is standard of care in the United States. As well, I did not find any definition of acceptable bone loss for patients who would be allowed in this study. Failure was described as requiring a second intervention, and that included self-dynamization, and I think that's a significant problem with trying to define the results of this study. The nonunion and delayed union definitions have already been covered. Follow-up was defined as at the 12-month or one-year period and involved both the clinical evaluation, the chemistry, the serology, the radiographs, and health resources consumption. As we have discussed, this is a multi-center study. Initially, there were actually

59 investigators, of whom 10 evidently contributed no patients whatsoever; and this occurred in 11 countries. Nineteen of these principal

investigators are said to have had a portion of their training in the U.S. This portion is not

defined, and given the licensing difficulties in

file:///C|/Storage/1121orth.txt (131 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

132 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 most of the States, my concern is that a good portion of these would have been either in a research lab or in observation only. The sponsors state actually several times that there are basically no treating differences between any of the countries, and my first problem with that statement is the fact that there were no Grade III fractures in France--and with apologies to anybody here who actually might be from France,

I have first-hand seen their driving, and I do not believe that there are no Grade III tibial fractures in France. So the question is what happened to these fractures, and it would suggest to me that the protocol was not the standard of care for the French for treating Grade IIIB fractures. both the CCOT organization and the combined English-speaking organizations have meetings on a regular basis, and I think that it is fairly obvious that the standard of care aground the world is different. Some of this is based on culture, As well,

some of it is based on health care resources and economy, and some of it is based on training. The German and South African groups, as was discussed, had the largest number of patients

file:///C|/Storage/1121orth.txt (132 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

133 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 contributing. What I found very disturbing was

that there were some very capable countries that had very few patients contributed, and I think this calls to several questions. The first one is was this study protocol actually not standard of care for the various countries that were involved. Did the patients

actually self-select themselves out, suggesting that perhaps there were other reasons why they did not get involved in the study? Or do these centers

actually truly see very few open tibial fractures, and if that's the case, one would be concerned about the patients that they did contribute as far as their experience in both judgment and technique is concerned. When you look at the results, there are actually a couple of areas that have been touched on but not really elaborated. First is that the

standard of care group or the control group actually had a larger number of unreamed nails. And if you look at the failure rate in the first 30 to 60 days, you will see that there is a very high number in the standard of care. nail in an A 9 mm unreamed

otherwise healthy young male has a

fairly high incidence, which has been reported in

file:///C|/Storage/1121orth.txt (133 of 309) [12/2/02 8:59:25 AM]

file:///C|/Storage/1121orth.txt

134 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 the literature, of self-dynamizing by fracturing the screws. So one would wonder if in fact these

patients went on to heal very nicely, and that the implant used was not appropriate for study conditions--it would be appropriate for treatment, accepting that the screw would break, but perhaps not appropriate for the study. As well, although not statistically significant, there were more patients with multiple fractures in the standard of care group than there were in the other group. For the 1.5 composite, there was a higher number of younger patients, and there was also a higher number of reamed rods. And I was very

delighted to see that the European Union has asked that this be addressed, because as I was reading through this, this is one of the areas where I said, "Hello?" There is no question that--and you can see this with the American study that Marc showed when they used exchange nails--that the concept of the philosophy is that you are putting stem cells or some osteogenic potential at the fracture site when you do this process. affects the results. I think that that definitely

file:///C|/Storage/1121orth.txt (134 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

135 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 And actually, to expand on that, if you look at the research or the philosophy of promoting fracture healing, you will see that most people consider three concepts--osteoinduction, osteoconduction, and osteogenic potential. Although the sponsors state that this is osteoinductive, if you read their investigators brochure, they actually state that the collagen sponge needs to contact both the distal and the proximal ends of the fracture, suggesting that probably there is some osteoconduction, and if you then add osteogenic cells, you have a fairly potent fracture promotion. If you go to their evaluation, I agree with both the statistician and the clinician from the FDA in that the CCRE has no validation associated with it whatsoever. In spite of that,

if you look at the 75 percent of patients healed, it is within 48 hours for all thee groups, and if you look at the radiological evaluation, it is exactly the same. There is some fallout at the 50

percent healing results, but I don't think it is significant. What is really interesting is that at the 12-month or one-year level, all the groups had more

file:///C|/Storage/1121orth.txt (135 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

136 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 than 25 percent of their fractures not united, and this is using either the sponsor's numbers, of which the lowest is 26, or the FDA numbers, where the lowest is 38. The other thing--and this is probably not totally fair--but if you take all of the secondary procedures in the control group and you take all of the secondary procedures in the BMP group, in fact, there is a larger number of patients who received BMP--there is also a larger number of patients who received BMP, but there is a larger number of patients who underwent a secondary procedure. Looking at side effects, the local side effects are really minimal. There is 17 percent

reported hypesthesia with the 1.5 composite as compared to single digits for the other two groups. My biggest concern is the antibodies to the recombinant BMP. There is one in the control As has been

and nine in the 1.5 composite.

previously discussed, this is significantly higher than previously noted. And the sponsors actually

pulled out the numbers for antibodies for all of their experimental groups, and it rounds out to just slightly more than 3 percent if you put all of their studies together.

file:///C|/Storage/1121orth.txt (136 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

137 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 My problem with this is that nobody understands what this actually means. There has

been no long-term data, although there have been more than 1,000 patients receiving BMP, as noted by the sponsor. The other thing that concerns me is that the investigator brochure states that the safety and efficacy of repeat use is unknown, and it probably is not in the public interest if a key trauma surgeon puts the sponge on a Grade I tibial fracture in a 32-year-old motorcycle rider who, years later at the age of 47 or 50, needs titanium cage and is unable to get healing promotion because he has antibodies. So if we go back to reviewing the objectives which they discussed in the beginning, the increased likelihood of healing with the composite, I think this study has way too many variables, and you can't come to a conclusion, but the answer is probably borderline. Defining the safety, I do think they have shown that it appears to be safe at least in the short term. Documenting increased healing with the composite at 6 months--I agree with the FDA

file:///C|/Storage/1121orth.txt (137 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

138 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 evaluator; there is too much subjective data in the clinical investigation to be able to state that. I think the documentation that the radiographs healed earlier is a definite "No". And the potential economic benefits--if 100 of the 300 BMP patients had to have a second procedure, then it is unlikely there are significant economic benefits to this. So in conclusion, the studies suggest that it is safe. This pivotal study is such a potpourri

that it is unable to tell us much except that it does not appear that this material interferes with healing, nor does it appear that it has a deleterious effect. I have several questions for the panel. What was the scientific reasoning behind having such an eclectic group studies where the variables were going to be so multiple? Can you explain the Why did you

lack of Grade III entrances in France?

not have a more specific protocol for the investigators, including not only the reaming versus unreaming, but also using scales to measure weight-bearing and other more objective means of clinical outcome? The fourth question is why did you include

file:///C|/Storage/1121orth.txt (138 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

139 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Finnegan. We're going to move now to Dr. Larntz for his statistical presentation. Dr. Larntz? DR. LARNTZ: I have a number of comments, Grade I's at all, as most people consider these to be pretty close to closed fractures. My last question--and I may actually be wrong in interpreting the data, because it was a little hard to pull some of the wheat out of some of the chaff in this--but it appeared to me that in some of the patients--I think it is three--with BMP antibodies, actually, three of them had delayed or nonunions or some kind of intervention. That's the end of my discussion. DR. YASZEMSKI: Thanks very much, Dr.

and they are a mixed bag, my comments, and I'm not sure I have them very well-organized, but I'll try to tell you a few things that I found out and what I learned by doing this, looking at this, playing with this, enjoying the data very much. First of all, I think it is very interesting that the rate of secondary intervention, at least as measured by the sponsors, does show an effect. It does; there is no

file:///C|/Storage/1121orth.txt (139 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

140 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 question. What they measured shows an effect, the By showing an

rate of secondary intervention.

effect, I mean the 1.5 dose has a lower rate of measured secondary intervention in numbers of patients to the control. there. There is a question about poolability of that, and I appreciated Dr. Lao's analysis, and I appreciate--actually, the sponsor did a similar analysis in materials that they provided that they did not report, at least, not in the same detail. What is interesting is that the analysis of poolability depends a lot on scale. Now, let's So that seems to be

see--how many of you want to know about logid versus probability scales--you can all raise your hands. Whoops--I don't see any. Well, that's too

bad, because I'm going to tell you a little bit about it. What we are looking at is probabilities which are proportions, numbers of successes over numbers of total attempted, or numbers of failures over numbers of total attempted. What is interesting about the probability scale, which you all remember from your first course in statistics, is that the variability of

file:///C|/Storage/1121orth.txt (140 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

141 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 that scale depends upon where you are in the scale, and in fact, when you are at very low proportions, you have very small variability, or if you are at very high proportions, you have very small variability. If you are in the middle, you have

higher variability. The long and short of that is that statisticians like me prefer to analyze the data for the most part in the logid scalae, which is the log odds, whatever that is. Odds are proportion

fore over proportion--well, I guess everyone knows about odds. Don't they have lotteries in this Okay. So

country now--not lotteries, but casinos.

it is proportion for versus proportion against, and then, to make that work statistically, mathematically, you take the log of that. the logid scale. Now, if you do the poolability analysis in logid scale, which the sponsors actually did do in the materials, it turns out that you need to think about some of the same issues that Dr. Lao talked about, which are random effects versus fixed effects. It is very clear--it is very clear--that What does that mean? That's

random effects are necessary.

That means the rate of secondary

file:///C|/Storage/1121orth.txt (141 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

142 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 intervention differed considerably by site. is no question about that. There

The rate of secondary That

interventions differed considerably by site. doesn't mean you have anything wrong. expect that. I'm a statistician. I would

If they didn't

vary by site, I wouldn't have enough to do, right? So they vary by site. And if you account for that and take account of that variability by site in the probability scale, which is what Dr. Lao did, you would adjust out and find that in fact there was no significant difference, no significant effect, in terms of secondary intervention. However, if you do that in the logid scale, which in my opinion is the correct scale to use, if you do that in the logid scale--the sponsors tried to do that and used some statistical package from what they called "SAS"--is that the one you used--I don't know if anyone has ever heard of that package or not, but it's a very major package. That package has difficulty handling the And Dr. Lao had some He omitted three

kind of data that they had.

difficulty handling that, too.

sites because they had zero failures, I think. Well, those sites might be useful, don't you think?

file:///C|/Storage/1121orth.txt (142 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

143 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 analysis? Zero failures is a pretty good number, do you agree? Should you throw out data like that? Well,

I wouldn't. And I wouldn't use the kind of package that SAS does. I would do a Bayesian

analysis--some of you who have been here before have heard me talk about that. if I do a Bayesian

analysis in the logid scale--sorry for the long shaggy dog story here--if you do a Bayesian analysis in the logid scale, it turns out that you don't have total poolability of sites, but what you do have is a significant effect with respect to numbers of reinterventions. So in fact in the Bayesian analysis, you do random effects, you do get a statistically significant effect, so I'll just say that now. That's the first story. By the way, I am going to comment slightly about those Kappas and reliability. The sponsor

provided that data--Kappas based on 20 evaluations. And again, this is technical--20 is too few to talk about, so I'll stop talking. The next point--do we need survival Do we need to use those Kaplan-Meyers?

There is a difference in the sponsors saying, well,

file:///C|/Storage/1121orth.txt (143 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

144 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 maybe the crude rates are better. Maybe the And of

Kaplan-Meyers are not the right thing.

course, the Kaplan-Meyers, depending on where you draw the line, as was shown by the FDA, you might get different answers with respect to conclusions. An important aspect of the Kaplan-Meyers is a censoring of data--that is, the number of patients for whom you don't have a data after a certain point. One of the assumptions, basic

assumptions, is that in fact that censoring does not depend on eventual outcome. on eventual outcome. It doesn't depend

And you would think that that

might not depend much on treatment group as well. Let me tell you what I learned from data that actually arrived Saturday by FedEx--how is that for the way we got our information? In that

data that arrived on Saturday by FedEx, I found out that there actually was censoring. I sort of heard

that there was not much missing data, but then, the censoring with respect to radiologic assessment or with respect to even investor assessment is considerable. For instance, with respect to radiologic assessment--remember, we are trying to go out 6, 9, 12 months--it was easy, because there was a table,

file:///C|/Storage/1121orth.txt (144 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

145 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 and I took it out of the table, to look at it at 50 days. That's not quite the 6 months, I think--you

have to realize that I am a statistician, and the calendar maybe different for me--but at 150 days, the rate of censoring for the control group was 34 percent. That means that there were no data for

radiologic assessment--no data that could be used for radiologic assessment for 34 percent of the patients after 150 days. Remember, I said it should be the same for the different groups, approximately, if things seem okay, if there is not something different about they way they are doing things? For this 0.75 BMP group, the rate of censoring was 25 percent. saying? Do you see what I'm

And for the BMP 1.5 group, the one that we

have thinking would have an effect, the rate of censoring was 16 percent. So in fact, there is more radiologic assessment missing for the control group. There is more radiologic assessment missing for the control group, at least if I understand the information provided to me on Saturday, okay?

The same thing goes with respect to fracture healing by the investigator--37 percent

file:///C|/Storage/1121orth.txt (145 of 309) [12/2/02 8:59:26 AM]

file:///C|/Storage/1121orth.txt

146 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 number. have no investigator assessment of fracture healing in the control group after 150 days. That is,

within 150 days, 37 percent of the individuals are censored within the first 150 days. Excuse me--that sounds like a large What is interesting is if you--I'll skip

to the 1.5 BMP group--in the 1.5 BMP group, the rate is 19 percent. So there is a vast difference

in the censoring rates for these measures that we are worried about--a vast difference in the censoring rates. me. So, even if we make adjustments for that--I assume there is nothing funny about the amount of censoring--if we make adjustments, if we look at the radiologic assessment of time to union, there is no difference between the three groups if we make adjustments for the amount of censoring. As I said, with respect to secondary interventions, there does seem to be a difference; it is clear. That is the primary endpoint, and That concerns me, that concerns

there does seem to be a difference. What else do I want to say? On the IM

nail issue, there are differences in the rate of reamed versus unreamed, if that's the right term.

file:///C|/Storage/1121orth.txt (146 of 309) [12/2/02 8:59:27 AM]

file:///C|/Storage/1121orth.txt

147 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 If you in fact make an adjustment--and buried in one of the reports is an adjusted analysis for just that--the effect of that on secondary intervention--actually, the effect that we see for the 1.5 versus control--is decreased considerably. It is still statistically significant. does seem to be some effect. So there

If you make a

covariate adjustment for the IM nail--I didn't have the data for that; I just found the analysis--if you have the data on the IM nail, it looks like if you make a statistical adjustment for that, you reduce the effect that you see. It is still

significant, but it is not--what was it, .0036 as the P value? It is not that big; it is smaller.

Finally--maybe not finally, but finally on this page, anyway--I am confused by something that was said about the fact that the investigators and the radiologic panel had access to the same x-rays. That's fine, but what you didn't tell me was did they have access to all of the x-rays that were taken. out. And I don't know that. I can't figure that

I would have expected assessment to be made If that is true, that

of all x-rays by the panel.

they had access to all of them, that's fine, but I still worry about the amount of missing data that

file:///C|/Storage/1121orth.txt (147 of 309) [12/2/02 8:59:27 AM]

file:///C|/Storage/1121orth.txt

148 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Larntz. What I'd like to do now is go over the order for the rest of the meeting. We're going to is implied in the Kaplan-Meyer analyses. And if you will just bear with me for a second--I'll stop there. Thank you. DR. YASZEMSKI: Thanks very much, Dr.

proceed with the second of three open public sessions now, and I'll call for folks who might want to speak in just a moment. Then, we're going to take lunch. I'll ask

the panel to stay here for a working lunch, and everyone else is free to have lunch where they would like to. We're going to take a half an hour

for lunch from the time that we break, and we'll come back at that time and ask the FDA to put up the questions that they have for us, and we'll have a discussion of those questions and a vote. So now, we're going to proceed with another open public session hearing. I would ask again as I did before that persons addressing the panel come forward, speak clearly into the microphone, and state their name and affiliation.

file:///C|/Storage/1121orth.txt (148 of 309) [12/2/02 8:59:27 AM]

file:///C|/Storage/1121orth.txt

149 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Witten. Open Public Hearing MR. CHRISTIANSEN: Thank you, Dr. shortly. DR. YASZEMSKI: Yes. Thank you, Dr. We are requesting all persons making a statement during this open public hearing to disclose whether they have financial interest in any medical device company. Is there anyone at this time who would like to address the panel? one listed presenter. I'll ask again--we had

Mr. Christiansen, if you are

here and would like to speak, now is an okay time. Good. Welcome. By the way, Dr. Yaszemski,

MS. WITTEN: can I just clarify?

DR. YASZEMSKI: MS. WITTEN:

Yes, ma'am.

We are actually having a

nonworking lunch; is that right? DR. YASZEMSKI: not do that? MS. WITTEN: nonworking lunch. DR. YASZEMSKI: MS. WITTEN: Okay, fine. That's fine. Yes. It's actually a Is that correct? Did we

And then we'll reconvene

file:///C|/Storage/1121orth.txt (149 of 309) [12/2/02 8:59:27 AM]

file:///C|/Storage/1121orth.txt

150 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Yaszemski, and I'm sorry I wasn't here for the morning session. I misjudged the traffic

conditions here in the D.C. area. My name is Bill Christiansen. I am a

full-time employee of Depuy Acromad, and I am speaking here today representing the Orthopedic Surgical Manufacturers Association, otherwise known as OSMA. OSMA is a trade association with over 30 member companies, and we welcome the opportunity to provide general comments at today's Orthopedic Advisory Panel meeting. OSMA's comments should not

be taken as an endorsement of the products being discussed here today. We ask instead that our

comments be considered during today's panel deliberations. These comments represent the

careful compilation of the member companies' views. OSMA was formed over 45 years ago and has worked cooperatively with the FDA, the American Academy of Orthopedic Surgeons, the American Society for Testing Materials, and other professional medical societies and standards development bodies. This collaboration has helped

to ensure that orthopedic medical products are safe, of uniform high quality, and supplied in

file:///C|/Storage/1121orth.txt (150 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

151 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 quantities sufficient to meet national needs. Association membership currently includes over 30 companies who produce over 85 percent of the orthopedic products in clinical use in the United States today. OSMA has a strong and vested interest in ensuring the ongoing availability of safe and effective medical devices. The deliberation of the panel today and the panel's recommendation to the FDA will have a direct bearing on the availability of new products. We make these comments to remind the panel of the regulatory burden that must be met today. the panel to focus its deliberations on the product's safety and effectiveness based on the data provided. The FDA is responsible for protecting the American public from drugs, devices, foods, and cosmetics that are either adulterated or unsafe or ineffective. However, FDA does have another role The Orthopedic Devices We urge

to foster innovation.

Branch is fortunate to have available a staff of qualified reviewers, including certified orthopedic surgeons, to evaluate the types of applications brought before this panel.

file:///C|/Storage/1121orth.txt (151 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

152 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 The role of this panel is also very important to the analysis of the data in the manufacturer's application and to determine the availability of new and innovative products in the U.S. marketplace. Those of you on the panel have

been selected based on your training and experience. You also bring the view of practicing

clinicians who treat patients with commercially available products. OSMA is aware that you have received training from FDA on the law and the regulation, and we do not intend to repeat that information today. We do, however, want to emphasize two

points that may have a bearing on today's deliberations. One is reasonable assurance of safety and effectiveness, and two, valid scientific evidence. The first point--reasonable assurance of safety and effectiveness. The definition in the

law is that if there is a reasonable assurance that a device is safe when it can be determined that the probable benefits outweigh the probable risks. Some important caveats associated with this oversimplified statement include valid scientific evidence and proper labeling and that safety data

file:///C|/Storage/1121orth.txt (152 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

153 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 job today. evidence. may be generated in the laboratory, in humans, or in animals. There is a reasonable assurance that a device is effective when it provides a clinically significant result. Again, labeling and valid

scientific evidence play important roles in this determination. The regulation and the law clearly state that the standard to be met is a reasonable assurance of safety and effectiveness. "Reasonable" is defined as "moderate, fair, and inexpensive." The second point is valid scientific The regulation states that while

controlled investigations shall be the principal means to generate data that are used in the effectiveness determination, the following principles are cited in the regulation as being recognized by the scientific community as essentials in a well-controlled investigation--a study protocol, a method of selecting subjects, methods of observation and recording results, and a comparison of the results with the control. To conclude, the panel has an important You must listen to the data presented

file:///C|/Storage/1121orth.txt (153 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

154 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 by the sponsor, evaluate the FDA presentations, and make a recommendation about the approvability of the sponsor's application. We speak for many

applicants when we ask for your careful consideration. Please keep in mind that the

standard is a "reasonable assurance" balancing the benefits with the risks. The regulatory standard Please

is not proof beyond the shadow of a doubt. be thoughtful in weighing the evidence.

Today OSMA thanks the FDA and the panel for this opportunity to speak. Our association

trusts that its comments will be taken in the spirit offered--to help the FDA decide whether to make a new product available for use in the U.S. marketplace. OSMA members are present in the audience to answer any questions during the deliberations today. Thank you. DR. YASZEMSKI: Christiansen. We're going to break for lunch now. have 20 minutes to one; we'll start up again at 1:10. Thanks everybody. I Thanks very much, Mr.

file:///C|/Storage/1121orth.txt (154 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

155 1 2 3 [Whereupon, at 12:40 p.m., the proceedings were recessed, to reconvene at 1:15 p.m. this same day.]

file:///C|/Storage/1121orth.txt (155 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

156 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Panel Questions & General Discussion DR. YASZEMSKI: Thanks, everybody. AFTERNOON SESSION [1:15 p.m.]

I'll ask Mr. Kaiser if he would put the FDA questions up, we'll read them and then have a general panel discussion. While Mr. Kaiser is getting the question up on the screen, I'm going to go ahead and read the first question. labeling. The FDA has asked: "Discuss the adequacy The first question concerns

of data from experience treating acute open tibia fractures stabilized with IM nails to support a more general indication: 'The treatment of acute

long-bone fractures that require open surgical management. InductOs increases the probability of

fracture healing, accelerates fracture healing, and decreases the frequency and invasiveness of interventions for delayed union or nonunion....'" What we'll do is--Dr. Witten? MS. WITTEN: I'm sorry, but it might be

good to go through these in order. DR. YASZEMSKI: MS. WITTEN: We'll do as you ask. The first one is the

Okay.

file:///C|/Storage/1121orth.txt (156 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

157 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 sponsors. clockwise. DR. FINNEGAN: I actually agree with the Witten? MS. WITTEN: Oh, absolutely. If that's study design. DR. FINNEGAN: Actually, we had asked for

this one first, because we think it can be dealt with quickly. MS. WITTEN: Oh, okay. That's fine.

DR. YASZEMSKI:

Is that okay with FDA, Dr.

your intention, that's fine. DR. YASZEMSKI: Okay. Thank you.

What we'll do is go around the table and ask every panel member, voting and nonvoting, if they have a comment on this question. The panel members are free if they would like to request information from the sponsors; they may request, and then I'll ask the sponsors to come up and answer the specific questions posed to them. Dr. Finnegan, let's start with you and go

I do not think that there is a

significant difference between the tibia and any other long bones, and in fact I would think that the labeling is more critical related to the grade of open fracture than it is to the specific long

file:///C|/Storage/1121orth.txt (157 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

158 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 point. DR. YASZEMSKI: DR. ABOULAFIA: Dr. Aboulafia? Yes, I hate to dissent, sponsors. bone that is involved. DR. YASZEMSKI: Dr. Kirkpatrick? DR. KIRKPATRICK: I'm sorry, I thought we Thank you, Dr. Finnegan.

were going counter-clockwise. At any rate, I also agree with the The question that I would raise is do we

need to expand indications where it may not be needed at all. DR. YASZEMSKI: Dr. Doull? DR. DOULL: I have no comment on that. Thank you. Thank you.

DR. YASZEMSKI: Ms. Rue? MS. RUE:

I have no comment. Ms. Maher?

DR. YASZEMSKI: MS. MAHER:

I have no comment at this

but I think, in keeping with Dr. Kirkpatrick, I still think it is a broad jump to say that one situation begets another and that while the data, which we will get to in the study design, may or may not support its use in open tibial fractures,

file:///C|/Storage/1121orth.txt (158 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

159 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 broad. DR. YASZEMSKI: Dr. Schmidt? DR. SCHMIDT: I generally agree with the Thanks, Dr. Aboulafia. so to extrapolate it to other long-bone fractures may be also problematic and dependent on how those fractures are healed, whether it is primary or secondary healing, whether it is a plate versus a rod. So I think it becomes a little bit too

sponsors on this issue. I agree that they really have chosen the most severe model for fracture healing, which would be an open tibia fracture. A

femur typically does not have trouble healing, but even in that regard there can be problems. So in

general, I am comfortable with the more broad indication, but I think that we do need to see some data at some point to demonstrate that. DR. YASZEMSKI: Dr. Larntz? DR. LARNTZ: No comment. Thank you. Thank you.

DR. YASZEMSKI: Dr. Naidu? DR. NAIDU: far as the labeling.

I concur with the sponsor as

file:///C|/Storage/1121orth.txt (159 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

160 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 design. Ms. Witten? MS. WITTEN: Actually, let me just ask a Thank you. DR. YASZEMSKI: Thank you, Dr. Naidu.

Dr. Witten, the panel feels that in general, the sponsor's decision about labeling is appropriate. There is some concern that it may not

directly extrapolate and that more data down the line will be necessary, but there don't seem to be any major disagreements with the labeling. Have we adequately discussed FDA's questions regarding the labeling? MS. WITTEN: Yes. Thanks very much.

DR. YASZEMSKI:

We are now going to move on to the study

follow-on question, although you may be planning to discuss it when you talk about effectiveness. The comments from the panel have related to the first part of it, that is, the treatment of acute long-bone fractures that require open surgical management. The rest of it--probability

of fracture healing, accelerates fracture healing, decreases frequency and invasiveness of interventions--I think we would appreciate a

file:///C|/Storage/1121orth.txt (160 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

161 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 comment on that specifically either now or when you discuss the question on effectiveness. DR. YASZEMSKI: If it would be acceptable

to FDA, I think we'll do it with the effectiveness question. Mr. Kaiser, we'll go to the study design. "Study Design. Discuss the impact of the

following on the ability of the study to collect clinically valid data: 1) definition of standard

of care in view of the multiple confounding factors; 2) clinical relevance of rate of secondary interventions required to promote healing as a primary endpoint; 3) reliability of interpretation of the terms 'union', 'healing', 'delayed union', and 'delayed healing' at various sites." Please start again, Dr. Finnegan. DR. FINNEGAN: Well, this is sort of

beating a dead horse, but anyway, I think that we have talked about the standard of care. The

standard of care that was used in the study is one that is an acceptable standard of care worldwide. By that, I mean that patients were not denied treatment that would have been appropriate, and patients were not given treatment that was inappropriate.

file:///C|/Storage/1121orth.txt (161 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

162 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 However, I am not sure that it is up-to-date standard of care, but I think it is probably acceptable. Then, I would go to the last point on the definitions of "union", "healing", "delayed union", and "delayed healing". I think there are a number

of problems with the study design, both subjectively for the investigators as well as the fact that the blinding was not double and that, as Dr. Larntz has shown very nicely, there does appear from statistics to probably have been some investigator enthusiasm for the study material. On the clinical relevance of the rate of secondary interventions, my problem with this is the use of the small, unreamed nails which will have hardware problems, and I think that that is not a sign of a delayed union at the fracture site, it is a sign of mechanical forces on the implant that the implant probably can't handle. So I think

that that does have relevance as far as the study design and the implications of the sponsor are concerned. DR. YASZEMSKI: Dr. Kirkpatrick? DR. KIRKPATRICK: Mr. Chairman, would it Thanks, Dr. Finnegan.

file:///C|/Storage/1121orth.txt (162 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

163 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 question. be appropriate to ask a question of the sponsor at this point about a specific issue within the study? DR. YASZEMSKI: DR. KIRKPATRICK: Yes. Thank you.

I would like the sponsor to tell me, because I don't recall seeing the data separated out, how many of the fractures in each group had the fibula fracture fixed. DR. VALENTIN: This is Alex Valentin.

I would like to get back to you on this We have this number, but we need to

check exactly the listing, and I think we can get back to you on this. DR. YASZEMSKI: Dr. Kirkpatrick, would

that be okay to come back to you with that when they have it? DR. KIRKPATRICK: DR. YASZEMSKI: That would be fine. Do you have other

questions or comments regarding the study design? DR. KIRKPATRICK: DR. YASZEMSKI: DR. DOULL: No. Dr. Doull?

I also had a question and it

had to do with using the worldwide data as a predictive basis for the U.S. population. When Dr. Finnegan was discussing

file:///C|/Storage/1121orth.txt (163 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

164 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 confounders there, I wondered--I didn't find in the data that you gave us specific information about diet and smoking and ethnicity and so on, about those specific confounders in that worldwide study. I am wondering whether I missed that, or if that is not available. DR. VALENTIN: We have collected a certain

number of variables at baseline and provided this information in the package. For example, we

provided the ethnicity, we provided the percent of patients with a history of smoking, which is a known risk of delayed union. We have in total

identified 27 covariables at baseline in addition to seven demographic characteristics . In choosing our covariables, we were intent to select those that were identified as having an effect on the outcome, and we are satisfied that we did that. We have no knowledge

of other covariables that might have affected the outcome. Have I answered your question? DR. DOULL: Yes. I guess my question

would be whether, if you just used the U.S., for example, and lost all that worldwide data, the EU would still have pretty much the same database.

file:///C|/Storage/1121orth.txt (164 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

165 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 you. DR. YASZEMSKI: MS. MAHER: Ms. Maher? Thank you. DR. YASZEMSKI: Ms. Rue? MS. RUE: I don't have anything. Thank Thank you, Dr. Doull. DR. SWIONTKOWSKI: This is Marc

Swiontkowski from the University of Minnesota. What I tried to allude to in m presentation was that those covariates were not distinctly different between the U.S. population and the worldwide population. DR. DOULL: Okay; that's reassuring.

I actually have a follow-up

question to Dr. Doull's question, and that is questions were raised about whether the standard of care is applicable to the U.S. and whether their data was applicable. I was wondering if you could

comment on that in a more general term. DR. SWIONTKOWSKI: Swiontkowski again. I appreciate Dr. Finnegan's analysis. I This is Marc

think there was some confusion about whether or not the nail was placed at the time of the initial procedure. You mentioned that the majority of

file:///C|/Storage/1121orth.txt (165 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

166 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 patients did not have the nail placed at the original--that's incorrect. I think what you are reading is a table that looked at supplemental cast or splint fixation. That is in addition to the IM nail. The

only patients who did not get an IM nail at the time of the original presentation had an ex-fix, and that was less than 10 percent in which is was exchanged at the time. misinterpretation. So in summary, the standard of care, as I said in my comments, is very close, if not equivalent, to the care rendered in the United States. DR. YASZEMSKI: Dr. Aboulafia? DR. ABOULAFIA: I think there is a pattern Thanks, Dr. Swiontkowski. So I think you have a

emerging of what the concerns are from panel and from FDA, and a lot of my concerns have already been expressed by Dr. Finnegan and Dr. Buch. I'll start with the positive. Before

being critical, you have to also think what could be done better, and for things like radiographic analysis and clinical analysis, although it may be subject to some criticism, if it is the best we can

file:///C|/Storage/1121orth.txt (166 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

167 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 do, that criticism goes away. The same thing with standard of care issues, and I want to make clear that I don't have any issues with either standard of care or clinical or radiographic assessment, excepting the fact that there is a margin of error and that it is the best we can do. So I'll take those out of the equation. The other things to some extent are repetition and may be just saying the same thing over again, but in my own words. The issue of

reamed versus unreamed nails is not only relevant as it relates to mechanical failures but also the effect on decision for secondary surgery or secondary interventions. Are surgeons more likely

to go back and do an exchange nailing in a patient who was initially treated with a small-diameter nail, who may be heavyset, and they are reluctant to allow that patient to weigh-bear because of risk of mechanical failure? Also, one of the monitors or measures of secondary outcomes was the effect on weight-bearing or what the weight-bearing status is. Weight-bearing status is determined not necessarily by fracture healing, but also the diameter of the nail. Even though some of the smaller,

file:///C|/Storage/1121orth.txt (167 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

168 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 solid-diameter nails biomechanically have allegedly the same strength as a larger, non-solid nail or hollow nail, the screws that are used to lock those nails are different sizes, and they are subject to different mechanical failures as a result. So what

effect did the choice of whether it was a reamed or unreamed nail have on those secondary things such as weight-bearing status? The decision--and this obviously relates to study design and was brought up by Dr. Buch and FDA--the decision of how secondary interventions were determined was very, very subjective, and that is reflected again by the diameter of the nail. Also, there is no mention of whether secondary procedures were related to nail reduction or errant screw placement. In one of the example

in Volume 2, I think, on page 346, one of the distal interlocks is going into the fibula, and you can see some radiolucency around the fibula. While

that patient didn't have secondary intervention, errant screw placement or problems with the screw may have influenced the decision to go ahead with an exchange nailing; how was that separated out? It has already been mentioned that the BMP group had larger-diameter nails, and many of those

file:///C|/Storage/1121orth.txt (168 of 309) [12/2/02 8:59:28 AM]

file:///C|/Storage/1121orth.txt

169 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 were unlocked. You obviously cannot have

mechanical failure of the interlock if there is no interlock, so it biases that secondary variable because the "N" is smaller. The fact that censoring differences took place, as pointed out by Dr. Larntz, raises the issue whether physicians were treating patients differently in the two groups. What I mean by that

is why would there be such a difference between the group that received BMP and the control group in terms of censoring. Were physicians more interested or had a certain level of enthusiasm for the BMP patients? The reason that becomes an important issue is again if it leads to secondary intervention. Some people put a small-diameter unreamed nail, get over as an initial internal splint, sort of with the idea of going back and doing an exchange nailing if there is no evidence of radiographic healing in a 6-week period of time. Would a physician who knew that the patient had BMP be less inclined to take that patient back for an exchange nailing, saying, "Well, maybe the BMP will start working"? So I think that physicians who were not

file:///C|/Storage/1121orth.txt (169 of 309) [12/2/02 8:59:29 AM]

file:///C|/Storage/1121orth.txt

170 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 blinded did handle those patients differently. I said in the beginning that if you can't find a better way to do it, you can't be too critical. I think you can find a better way of

doing it and have very defined criteria of what constitutes secondary intervention, whether someone goes back for an exchange nailing or not. And while the sponsor says that prophylactic bone grafting was not allowed, exchange nailing might be considered a type of bone grafting. It is not the typical post- [inaudible]

bone grafting or open autogenous bone grafting that we normally talk about, but secondary interventions were done at an average of 108 days or in that range, between 100 and 110, and that is not the definition of a delayed union. Certainly there are

high-risk fractures, and no one would say that that is a violation of standard of care, but it is a form of prophylactic intervention used to encourage bone healing. Then, there was also a little bit of an issue--and I don't know what the magnitude of this was--but there was variation in the dose, and some patients didn't get the dose that are listed under their labeling of 1.5 or not.

file:///C|/Storage/1121orth.txt (170 of 309) [12/2/02 8:59:29 AM]

file:///C|/Storage/1121orth.txt

171 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 So those were issues in the study design that I think are confounding variables that actually can be controlled and make for a better scientific experiment. DR. YASZEMSKI: Dr. Schmidt? DR. SCHMIDT: I agree with most of the Thanks, Dr. Aboulafia.

comments that Dr. Aboulafia just made and have a few more of my own to add. There are significant differences in the standard of care from country to country, and that has been alluded to. For instance, in Germany,

there are still a number of surgeons who will basically use small-diameter unreamed nails in most open tibia fractures, and in some areas of the United Kingdom, for instance, a reamed tibial nail will be used in every, single patient. And that

local standard of care is going to influence management. Surgeons who use reamed nails are

going to have a far less dramatic problem with hardware failure than the surgeons in Germany who may be accustomed to seeing a lot of their interlocking screws break. So there is going to be

a bias from country to country or center to center in the threshold for recommending a secondary

file:///C|/Storage/1121orth.txt (171 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

172 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 intervention, and I think that that is a significant problem with how things worked out with this particular study, where we have centers that are dramatically different in their practice, and this may reflect some of the variation that was seen in the data. I wish there had been a little more uniformity among the centers that were studied in terms of their underlying treatment protocols. One

question that came up was is the control group an appropriate control group, and no one has really discussed that. I think what the study is looking at is whether this growth factor accelerates fracture healing, which is different from using it as a bone graft to fill in a defect. There are other methods available right now to accelerate bone healing, for instance, ultrasonic or electrical bone stimulation. Those,

even though they are available, I would say that they are not the typical standard of care. I don't

use them much in my own practice, yet those methods are available. I think perhaps it would have been

nice to have had another group that had a noninvasive method of bone healing acceleration

file:///C|/Storage/1121orth.txt (172 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

173 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 studied. But given those factors, I agree with what Dr. Finnegan said, that the global standard of care, that appropriate treatment was provided and inappropriate treatment was withheld, was met by this study. Are we going on to discuss all of these bullet points now, or just the-DR. YASZEMSKI: under study design. DR. SCHMIDT: I also have some comments All the bullet points

about the secondary intervention, and this has been brought up as well. The breakage of interlocking screws is very common, and it is a marker that a fracture has not yet healed. Obviously, a healed fracture is But the point

not going to have hardware failure.

is that a broken screw is also a method of treatment. For instance, when a screw breaks, the

fracture becomes dynamized, and more often than not, it is a clinically irrelevant incidental finding that you see after a fracture has healed. So to count that as one of the determinants of the primary endpoint I think needs a little bit of further discussion.

file:///C|/Storage/1121orth.txt (173 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

174 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 I would be interested to see what the analysis is if you threw out that group of patients and really just looked at secondary interventions that were invasive to the patient--for instance, the need to exchange the nail or to perform a later bone graft. Another question I had--and this relates a little bit to safety--was the high infection rate in all three cohorts of patients. That seems to

indicate that there may have been a systematic problem compared to what we have in the United States. Our infection rate--we did a study at my

hospital of tibial nails, and we had an infection rate of about 5 percent. I know that the sponsor

was trying to be very careful to include every possible infection, and that is going to make it a little bit higher, but it still seems higher than it should be. One question I had--and this may be a ticklish point to bring up in front of the FDA--but were antibiotic beads typically used in this study? I know that they are very commonly used in Europe, and I would submit that they represent a standard of care. DR. YASZEMSKI: Dr. Valentin?

file:///C|/Storage/1121orth.txt (174 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

175 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DR. VALENTIN: The antibiotic beads were

used when necessary during the initial phase of one treatment, and they had to be removed at the time of definitive wound closure upon implantation of BMP-2/ACS. DR. SCHMIDT: That raises another question

that I have, which is is there any animal data that suggests there might be a difference in efficacy when this composite is used immediately after there has been an antibiotic bead in the same defect. DR. YASZEMSKI: DR. VALENTIN: Dr. Valentin? It is our experience that

there is no apparent interaction between BMP-2 and antibiotics. As a matter of fact, all animal

studies have systematic antibiotic treatment at the same time as BMP-2 treatments. deleterious effect there. DR. YASZEMSKI: Dr. Larntz? DR. LARNTZ: I have a question. I would Thanks, Dr. Schmidt. We haven't seen any

just like to understand a little better the decision to have a radiologic evaluation. If I

understand right--maybe I didn't understand--it wasn't done at all visits for all patients. that correct? Is

You can answer "yes" or "no" for

file:///C|/Storage/1121orth.txt (175 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

176 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 visits. visits. DR. VALENTIN: DR. LARNTZ: by your panel? DR. VALENTIN: DR. LARNTZ: That is correct. Okay--all patients, all Correct. And all those were evaluated visits. DR. LARNTZ: For all patients at all correct. DR. LARNTZ: evaluation? DR. VALENTIN: Of all patients at all So you did do radiologic that. DR. VALENTIN: The answer is no, it is not

There wasn't any exclusion based on-DR. VALENTIN: No--except for x-ray lost

in the mail-DR. LARNTZ: understand that. DR. YASZEMSKI: For the transcriptionist, I understand that, I

that discussion was between Dr. Valentin and Dr. Larntz. DR. LARNTZ: That's fine. Thank yo.

DR. YASZEMSKI: Dr. Naidu?

file:///C|/Storage/1121orth.txt (176 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

177 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DR. NAIDU: I have several comments and

also several questions. First of all, with regard to standard of care, Dr. Schmidt has clearly stated that standard care is different from country to country. My

question to Dr. Swiontkowski is were you dictating this fracture care, or was the surgeon calling it? I mean, you stated that you trained all of these people. Could you clarify that? DR. SWIONTKOWSKI: Yes. The protocol

defined the standard treatment, which was irrigation and debridement, and then, prophylactic antibiotics and insertion of an IM nail. DR. NAIDU: Okay. So it appears as if the

standard of care is variable, and I do concur with Dr. Schmidt with regard to that. DR. SWIONTKOWSKI: DR. NAIDU: Which aspect?

With regard to the reaming,

unreaming, and the treatment in various countries that Dr. Schmidt clearly cited, that the cultural differences are there. So therefore, in my

opinion, the standard of care was different, and that is a big issue. The second issue is the clinical relevance of the primary endpoint and the rate of secondary

file:///C|/Storage/1121orth.txt (177 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

178 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 interventions required to promote healing. You

know, we are looking at the efficacy of a drug. This is classified as a drug in Europe. classified as a device. Even though Dr. It is

Swiontkowski clearly states to us that there is an NIH-sponsored trial of reamed versus unreamed nailing underway with secondary intervention as the final endpoint, we are not testing a new product in the sense that this is a different drug. Like the

European Union--EU stated that this is a drug. Therefore, I have a problem with this clinical relevance of the primary endpoint being the rate of secondary interventions, even though there may be an NIH-related protocol that is approved. When one judges the safety and efficacy

of a drug/device, I'm not sure this is a good enough primary endpoint. Finally, the reliability of the interpretation of the terms "union", "healing", "delayed union", "delayed healing" at various sites is completely nebulous. Thank you. DR. YASZEMSKI: DR. FINNEGAN: Dr. Finnegan? Dr. Swiontkowski, in Volume

IB, page 59, Table 4.2.4-1, it says the number of

file:///C|/Storage/1121orth.txt (178 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

179 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 patients who received preliminary external fixation or intramedullary nail fixation by treatment group, and there are significantly small numbers receiving external fixation IM rod, and the patients without preliminary fixation number at least 130 in each group. DR. VALENTIN: This is Alex Valentin

taking that question, if you don't mind. DR. YASZEMSKI: DR. VALENTIN: Go ahead. I would like to clarify

this table, and I agree it may be a little bit confusing in that table. What we meant to say is that initially, the patients had casts or splints in addition to their primary treatment by an IM nail. The

external fixiter was used, as indicated there, in about 7 percent of the patients, and I believe there were two patients--one in the standard of care and one in the 1.5--who initially received an IM nail, and there was an exchange of that IM nail. The other patients, the 7 percent, received an external fixiter, which was then changed to an IM nail. And the remaining patients all received an In addition, they

IM nail as their primary care.

received a cast and other sorts of treatment that

file:///C|/Storage/1121orth.txt (179 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

180 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 are indicated there. DR. FINNEGAN: So the following page has

the following table which says "Days from injury to definitive fracture fixation with IM nail," and the standard is 2.6, 2.7, 2.9, and the maximum is 13 or 14 days. DR. YASZEMSKI: correct as well. information. This information is

It does not contradict the first

These patients were initially treated

for the vast majority with an IM nail, in a few cases by external fixiter, and then moved to an IM nail; and in two cases, an IM nail exchange. DR. FINNEGAN: DR. YASZEMSKI: DR. ABOULAFIA: Thank you. Dr. Aboulafia? I was just going to add,

again related to study design, two small comments. One of the things that is missing is Winquist [phonetic] classification, which impacts on weight-bearing status. There are patients whom you

allow immediate weigh-bearing--"on the way to the recovery room" is the euphemism I use--and it has to do more with fracture pattern than it does anything else. So when you look at that secondary

endpoint, weight-bearing status, and don't include data related to degree of comminution and cortical

file:///C|/Storage/1121orth.txt (180 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

181 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 it there. DR. YASZEMSKI: Thank you. contact and the things associated with Winquist classification, it does a disservice. And whether

that is corrected by randomizing them, you would want to know if treatment groups were different in the allocation of those with severe comminution versus those without severe comminution. And there was one other thing--I'll leave

Dr. Witten, we have had a discussion now about the study design. There are some strengths In

and weaknesses that have been identified.

general, the standard of care was deemed acceptable with some caveats, and those caveats included that local standard of care, specifically in other countries, may have affected the choice of the initial nail, and also the differences in different countries perhaps had effect on the infection rate. It was brought up that other noninvasive means of accelerated fracture healing such as electrical or ultrasound means may have been appropriate additions. It was also brought up that

since some of the patients didn't get the full sponge that perhaps there was some dose variability among the three groups.

file:///C|/Storage/1121orth.txt (181 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

182 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 with you? DR. SCHMIDT: Yes. As we have heard The small nails were considered to perhaps be a factor, but in general, given the strengths and weaknesses as discussed, there didn't appear to be any discussion of any critical flaws in the study design as brought up by the members of the panel. Have we adequately addressed your questions on this issue? MS. WITTEN: Yes. Thank you.

DR. YASZEMSKI:

Mr. Kaiser, could we move on, please, to effectiveness? For effectiveness, the FDA is asking us: "Accounting for trial design, resulting data and statistical analyses, discuss the adequacy of effectiveness in terms of the decrease in the number of secondary interventions required to promote fracture healing, and accelerated fracture healing determined by the fracture healing at 6 months assessed by the investigator and the radiographic evidence of fracture union assessed by the independent radiologist. Dr. Schmidt, could I ask that we start

file:///C|/Storage/1121orth.txt (182 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

183 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 discussed this morning, this is a very complicated study and very difficult to just answer those in a "yes" or "no" manner. I think I should state my personal bias, which is that I know that these compounds are effective. There have been ample animal models in

the literature, some of which were reviewed today, that BMPs to promote bone healing. I guess the question, though, is does it accelerate bone healing in human beings with an open tibia fracture. I have to defer to the statistical analysis that was presented by Dr. Larntz, and it sounds like that answer is true, although it is still a little bit difficult and perhaps a bit nebulous and depends to some degree on how you analyze those statistics, and even that may not be perfectly clear. I don't really think they have truly shown that there is an increased rate of union at 6 months, and I think you could argue about is the 6-month important or some different time frame, and it appears that the results are different at the different time frames, so I am less comfortable with that assertion. And I am also concerned about

file:///C|/Storage/1121orth.txt (183 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

184 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 the issues with the radiographic analysis that we heard discussed. DR. YASZEMSKI: Thank you.

Dr. Larntz, can we come around to you? DR. LARNTZ: Sure. I believe that the way

secondary interventions were measured, there is clearly a significant effect in that variable with respect to whether or not there was a secondary intervention. And I think that that is true

accounting for the diversity of the sites, which I understand some people might think that's a disadvantage. As a statistician, I think diversity of sites is actually quite good and useful--but I may be in a minority with respect to pure study design. With respect to time to healing, we are in a more problematic state, and clearly, the radiologic assessments, however that was done--and if you look at that Kaplan-Meyer analysis, the curves are on top of each other, so what can you say? I once had someone try to tell me there was a

difference in a subset based on that, and I walked out of the room. other. With respect to the investigator They are just on top of each

file:///C|/Storage/1121orth.txt (184 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

185 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 evaluation of time to healing, it seems like there probably is a slight advantage for the 1.5 BMP with respect to the investigator time to healing. But

again, I worry about--well, I'll just say that as recorded, I would say that that is probably there. The P values in the Kaplan-Meyer are slightly different for the log rank Wilcoxin [phonetic] test; they give slightly different indications--but I think there probably is something in the time to healing with respect to the investigator version of that. With respect to the radiologic evaluation by the panel, there is nothing there. DR. YASZEMSKI: Dr. Naidu? DR. NAIDU: With regard to the first Thanks, Dr. Larntz.

bullet point, whether the sponsor has shown a decrease in number of secondary interventions required to promote fracture healing, I will rely some of my conclusions on Dr. Larntz' statistical analysis, and it seems as if they have shown that. But as far as accelerated fracture healing as determined by fracture healing at 6 months, Dr. Larntz clearly went through the numbers in detail--in the standard of care group without any

file:///C|/Storage/1121orth.txt (185 of 309) [12/2/02 8:59:30 AM]

file:///C|/Storage/1121orth.txt

186 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 BMP, 37 percent of the x-rays at 150 days were censored; in the 1.5 group, only 19 percent censoring was there. Based on all these numbers

presented, I am not sure that accelerated fracture healing can be concluded at 6 months as assessed by the investigator. Thanks. DR. YASZEMSKI: Dr. Finnegan? DR. FINNEGAN: I would concur--do you want Thanks, Dr. Naidu.

us to address labeling at the same time, or no--the question that Dr. Witten had? DR. YASZEMSKI: I think that, with the

permission of FDA, since they asked for additional comment on labeling after our discussion of that question, if you have something to add to it at this point, please do so. DR. FINNEGAN: I would suggest that the

label probably should not contain "accelerates fracture healing" and "decreases the frequency and invasiveness of interventions," the reason being that I really do think that there are so many confounding variables that certainly, there is no deleterious effect, but I really don't think that even for secondary interventions, given the

file:///C|/Storage/1121orth.txt (186 of 309) [12/2/02 8:59:31 AM]

file:///C|/Storage/1121orth.txt

187 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 mechanical properties of the small nail, that we can draw any large conclusions. DR. YASZEMSKI: Dr. Kirkpatrick? DR. KIRKPATRICK: Has the sponsor been Thanks, Dr. Finnegan.

able to get the fibula data for me? DR. VALENTIN: Yes. The fibula data were

collected on 437 patients, and there were 17 patients in the standard of care, 10 patients in the 0.75 and 10 patients in the 1.5, who received additional treatment of their fibula. DR. KIRKPATRICK: So roughly 10 percent of

your control group had the fibula fixed, and less than that--somewhere around 5 or 6 percent--of the other two groups. statistics as well? DR. VALENTIN: From what standpoint--if it Was that looked at with

is comparable across treatment groups? DR. KIRKPATRICK: Did you do any

statistical analysis on whether that had an impact on the results? DR. VALENTIN: No. I would just like to

DR. KIRKPATRICK:

point out one concern with regard to effectiveness. If the control group, which had a lower rate of

file:///C|/Storage/1121orth.txt (187 of 309) [12/2/02 8:59:31 AM]

file:///C|/Storage/1121orth.txt

188 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Thank yo. DR. YASZEMSKI: Dr. Doull? DR. DOULL: definitions. I am a little confused about Thanks, Dr. Kirkpatrick. healing, so to speak, had a higher rate of fibula fixation which may act as a distraction device on the fracture, then the normal biology of the fracture healing may have been disrupted somewhat, so it may not be a true indicator of the actual effectiveness. That's all the comment I have on that.

It seems to me that when you defined

heterotopic and ectopic, you defined that as local and distant, and your rate for ectopic was zero. But what would you define as increased bone growth further up on the tibia? DR. VALENTIN: Is that heterotopic? Thank you. I

We would define this as heterotopic. would like first of all the clarify that we have reported two ectopic calcification. In my

presentation, I stated that no calcification related to BMP-2 was reported. We had two

calcification observed, one in the femur and one at another side I can't remember now, and neither case was attributable to the BMP treatment.

file:///C|/Storage/1121orth.txt (188 of 309) [12/2/02 8:59:31 AM]

file:///C|/Storage/1121orth.txt

189 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 both. With respect to heterotopic ossifications, I show the numbers. We had four, four, and eight That

patients with heterotopic ossification.

assessment was for the region under study, which was the full limb we followed for the fracture. DR. DOULL: Okay, thanks. Thanks, Dr. Doull.

DR. YASZEMSKI: Ms. Rue? MS. RUE:

No further comment. Ms. Maher?

DR. YASZEMSKI: MS. MAHER:

I actually have two questions.

One, there was a question raised earlier, and maybe it was answered, and I didn't get it, in which case I apologize. That is, were all x-rays reviewed by

both the panel and the investigators? DR. VALENTIN: Yes. The answer is yes to

We addressed this question earlier. MS. MAHER: Okay. And my second question

is that I heard questions raised by Dr. Larntz regarding the censoring and radiologic assessment, and I was wondering if you all could comment on the questions he raised. DR. VALENTIN: This is a very important

question, and thank you very much for asking this. We would like to clarify here the process

file:///C|/Storage/1121orth.txt (189 of 309) [12/2/02 8:59:31 AM]

file:///C|/Storage/1121orth.txt

190 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 we have followed. The first question we had for

the primary endpoint was the avoidance of secondary intervention. The second question we had for the

secondary endpoints was for the patients who do not have a second intervention, do we see further acceleration of fracture healing or fracture union. So by definition, the patients who had a second intervention had to be removed from that analysis. So this censoring was done on purpose,

and the different numbers pointed to earlier in this analysis reflect in fact the number who had second interventions. So in order to answer the question was there an acceleration of fracture healing in the other patients, those who did not require second intervention, we had first of all to remove the patients who had one. This is dealt with in

different ways, with different types of analysis. For instance, in the Kaplan-Meyer display, this is handled by way of censoring, and we actually agree with your analysis that this weakens the Kaplan-Meyer analysis in that respect. That is why

we also pointed out in my presentation that we believe the rate of healing or of union by visit where the patients second interventions were

file:///C|/Storage/1121orth.txt (190 of 309) [12/2/02 8:59:31 AM]

file:///C|/Storage/1121orth.txt

191 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 accounted as failures, not healed or not united, was a more powerful evaluation of these data. DR. YASZEMSKI: Thank you.

Ms. Maher, does that answer your question--Dr. Valentin, if you are not done, please go ahead. DR. VALENTIN: I'm sorry. I just wanted

to add, therefore, that the censoring of the data in each treatment group, therefore, just to restate it, is a reflection of the success of the treatment. More patients were censored in the

standard of care because more patients required a second intervention; less patients were censored in the 1.5 group because less patients required second intervention. Thank you. MS. MAHER: Thank you. I actually have

one follow-up question.

I was wondering if you all

could comment on your opinion on the appropriateness of pooling the data from these various centers. DR. VALENTIN: I would like to take the

first stab at this question; this is clearly one that was very hotly debated, and many people have given their opinion here.

file:///C|/Storage/1121orth.txt (191 of 309) [12/2/02 8:59:32 AM]

file:///C|/Storage/1121orth.txt

192 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 I would like to give you the perspective of a clinician. We have selected 59 centers, of First of

which 49 have participated with patients. all, I would like to point out here that we

selected these 49 centers out of 400 that were screened. So we were very careful to make sure

that the centers participating in the study had common criteria--using, for example, the same types of surgical procedures, the same type of practice--so that we could compare the centers. was not done out of the Yellow Pages. The second part of my answer is that we have monitored quite a large number of variables to make sure that indeed these centers have contributed patients treated in the same manner, and as I was pointing out, seven demographic criteria, 27 covariables were checked, and we didn't find a difference between them. We have also run statistical analyses to look at center by treatment interaction, and in our analysis, there was no interaction, and that has justified the treatment analysis that we have conducted. At the end of the day, I'm saying that we want to have a population that reflects the general It

file:///C|/Storage/1121orth.txt (192 of 309) [12/2/02 8:59:32 AM]

file:///C|/Storage/1121orth.txt

193 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 trauma population. If we are too selective, we

will finish with all patients having had exactly the same fracture, the same treatment, the same nail, and that may be a very narrow application than other treatments. So it was our intent to

have a balanced application. DR. YASZEMSKI: Thank you.

Does that answer your question, Ms. Maher? MS. MAHER: Yes. Thank you. Dr. Aboulafia? Addressing the issue of

DR. YASZEMSKI: DR. ABOULAFIA:

effectiveness, the OSMA spokesperson made the point that our task here is to look at a reasonable assurance of safety and efficacy and sponsor innovation, and I would clearly support that. And

I certainly share Dr. Schmidt's enthusiasm about biological modifiers and recombinant BMPs as a group and other things to promote fracture healing, and I think it is a noble cause. Having said all that, I think some of that enthusiasm is based on both preclinical and clinical data in sites other than the tibia, of course. So then, the question or the weighing on

the other side of that teeter-totter is are the conclusions supported by the data, and that relates

file:///C|/Storage/1121orth.txt (193 of 309) [12/2/02 8:59:32 AM]

file:///C|/Storage/1121orth.txt

194 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 to study design, and there are issues with that. One of the things I neglected to mention about study design is whether subsets of patients' proximal, distal and middle third fractures were equally spread out. We know that healing

potential, rates, complications, depending on the anatomic site, is significant within the tibia itself and that the proximal tibia and the distal tibia are not the same bone in many respects. So I have less enthusiasm for whether the sponsor has proven effectiveness based on the data presented as it relates to the tibia in this study. DR. YASZEMSKI: Dr. Naidu? DR. NAIDU: I do have an additional Thanks, Dr. Aboulafia.

comment about the censoring business, because Dr. Larntz clearly made it clear that at 6-month time point, 34 percent of x-rays at the 6-month group was missing. Now, if you go back to the definitions of heal fracture, the three criteria--absence of tenderness upon manual palpation at the fracture site; radiographic fracture union as assessed by the investigator; and full weight-bearing status--and if you look at the number of patients

file:///C|/Storage/1121orth.txt (194 of 309) [12/2/02 8:59:32 AM]

file:///C|/Storage/1121orth.txt

195 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 with secondary intervention recommended and patients meeting the criteria of delayed union, all three criteria were used in only 26 percent of the patient population; two criteria were used in 52 percent; one criterion was used in 23 percent. Therefore, this issue of censoring does become important. We should not dismiss that. Thanks, Dr. Naidu.

DR. YASZEMSKI:

Are there further comments? [No response.] DR. YASZEMSKI: Dr. Witten, we have

discussed the issue of effectiveness and the FDA's questions regarding it. We previously discussed

the study design and offered you our opinions about the strengths and weaknesses of it, so these comments will be issues of effectiveness for the study design as presented in the application by the sponsors. Dr. Larntz indicated that with respect to secondary interventions, there clearly was an effect demonstrated by these data. The time to

healing, the investigator evaluation, and the radiologist evaluation were less positive in their strength with respect to demonstrating an effect. Although there could have been an effective with

file:///C|/Storage/1121orth.txt (195 of 309) [12/2/02 8:59:32 AM]

file:///C|/Storage/1121orth.txt

196 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 question? MS. WITTEN: Yes, but I do have a the time to healing, it was most weak with the radiologist interpretation. There were questions with respect to the two points that FDA has asked, that although the decrease in the number of secondary interventions is shown by the data, there is some question as to whether the accelerated fracture healing has been shown by this data, and there are arguments on either side of that, as you have just heard. Have we adequately discussed this

follow-on question based on the answers to this question and the answers to the previous question. DR. YASZEMSKI: MS. WITTEN: Please go ahead.

I would like to know whether

anybody on the panel would like to comment on the relevance of the effectiveness shown by these data on the study performed outside of the U.S. to effectiveness in U.S. patients. DR. YASZEMSKI: Panel members, the

question is does the data gathered on patients outside of the U.S. apply to the care of patients within the U.S. that? Would anyone like to comment on

file:///C|/Storage/1121orth.txt (196 of 309) [12/2/02 8:59:32 AM]

file:///C|/Storage/1121orth.txt

197 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Dr. Finnegan? DR. FINNEGAN: I think that one of the

nice parts about working in the melting pot of the world is that, yes, the world will produce the same patients that you have in your patient population. DR. YASZEMSKI: Dr. Kirkpatrick? DR. KIRKPATRICK: I would agree. I have Other comments?

even been in Third World countries that would manage tibias in a similar fashion and get them acutely debrided, irrigated and stabilized with an IM nail. DR. YASZEMSKI: DR. SCHMIDT: Dr. Schmidt? I agree, and a lot of the

practice variation that I alluded to earlier, we see within our own country, even in the State of Minnesota, perhaps. concerns about that. DR. YASZEMSKI: adequately addressed this? MS. WITTEN: Yes. Thank you. Dr. Witten, have we So I don't have any real

DR. YASZEMSKI:

Thanks so much.

Mr. Kaiser, could we go to the next question, please? The next question from the FDA regards

file:///C|/Storage/1121orth.txt (197 of 309) [12/2/02 8:59:32 AM]

file:///C|/Storage/1121orth.txt

198 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 safety: "Accounting for trial design and resulting

data, discuss whether or not the sponsor has provided a reasonable assurance of device safety in view of the rate of authentic antibody response to rhBMP-2 and to bovine Type 1 collagen; rate of hardware failure; rate of infection; rate of abnormal liver function lab values." Dr. Doull, can we start with you, please? DR. DOULL: Yes. There are four

questions--the antibody response, the rate of hardware failure and infection, and the abnormal lab values. I wasn't really too concerned about the antibody until Dr. Finnegan raised the possibility that those antibodies might in fact influence the subsequent response of that patient to another episode. It seems to me that that is a question

which could be resolved by animal experimentation, and I would think that's certainly something worthwhile doing. I found the data regarding the hardware failure and rate of infection lacking a causal relationship to some adverse effect of rhBMP or the sponge. And finally, then,the abnormal liver

file:///C|/Storage/1121orth.txt (198 of 309) [12/2/02 8:59:32 AM]

file:///C|/Storage/1121orth.txt

199 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 function values--what you really didn't tell us there was how long they persisted and whether that was the same in your control population and in the two treatment groups that you talked about. DR. YASZEMSKI: Dr. Doull, shall we ask

the sponsors to comment on that? DR. DOULL: Yes, could you tell us? Dr. Valentin? So if I understand the

DR. YASZEMSKI: DR. VALENTIN:

question, you would like to know if the incidence of liver function abnormalities was comparable across treatment groups? DR. DOULL: and for the magnesium. DR. VALENTIN: With respect to amylase, we Right, both for the amylase

have seen an increase in the number of patients reporting anomalies in the 0.75 mg group as compared to the control group, and an elevation at 1.5 compared to control group. In other words, you

have--I am citing from memory here--I think four patients in the control group, 10 patients in the 1.75, and 6, I think, in the 1.5--and please bear with me; I am trying to remember the exact numbers. But we didn't find an exact dose response, and we think that by just having the battery of 200 tests,

file:///C|/Storage/1121orth.txt (199 of 309) [12/2/02 8:59:32 AM]

file:///C|/Storage/1121orth.txt

200 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 with every adverse event tested, it was likely that some would be elevated and irregularly elevated between the three treatment groups. I personally don't think there is a relationship between BMP and increased amylase. think it translates the status of trauma of these patients. However, we stated the numbers as they I

are, and there is a statistical difference between 0.75 and control. DR. DOULL: Dose response is a hallmark of

toxicity, and I would agree with you that if you don't see dose response, you have to ask that question. DR. VALENTIN: And I would like to add

that for hypomagnesemia that the same thing was true--the 0.75 was more elevated at 1.5, and both were slightly more elevated than control. Again,

we couldn't see a clearly relationship between these elevations in very few patients, by the way, and the treatment with BMP. DR. DOULL: I would like to point out that

we have previously reviewed the tox database for this material and agreed that it was adequate. am impressed with the fact that in this presentation, you are using the term "safety" more I

file:///C|/Storage/1121orth.txt (200 of 309) [12/2/02 8:59:32 AM]

file:///C|/Storage/1121orth.txt

201 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 than you are using "toxicity." The tox database

which you use to support this application are really safety studies. You are showing that at

various levels--1,000 times anticipated dose and so on--you are seeing no adverse effects, but you do not characterize toxicity with safety studies. you do is show that at that dose, it is safe. order to characterize toxicity, you have to demonstrate adverse effects, and you have to show the dose required to produce those kinds of adverse effects, and in all the studies you did--the acute tox, the 28-day tox, the teratology, the reproduction--those are all safety studies. So that I can't help but wonder what is the true toxicity of rhBMP, and I guess we haven't answered that question, and we may not need to, because in terms of the safety, if you do a range of safety or a margin of exposure, you have huge factors here. Thank you. DR. YASZEMSKI: Dr. Kirkpatrick? DR. KIRKPATRICK: I also would concur with Thanks, Dr. Doull. All In

Dr. Finnegan's concern about a secondary challenge from an immune-converted or antibody-positive

file:///C|/Storage/1121orth.txt (201 of 309) [12/2/02 8:59:32 AM]

file:///C|/Storage/1121orth.txt

202 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 animal model, so to speak, a secondary challenge sometime in the future. I don't think we can I think if we

answer that in humans, obviously.

could come up with a reasonable way with an immune study, which I have no expertise in, it would be a reasonable assurance--and fairly simply, I would hope--that a future challenge would not result in a catastrophic immune response or autoimmune response. DR. YASZEMSKI: Dr. Finnegan? DR. FINNEGAN: The only thing I would add Thanks, Dr. Kirkpatrick.

to that is that you have a huge number of patients out there, and you could do a natural history on the antibodies to see if they persisted or not for the patients you have given this to over "x" number of years. I have no concern about the hardware failure, but that just means I do this operation more than once a week. Dr. Schmidt's comments on infection are interesting; I did not pick that up, and I'm not exactly sure what the effect of that is. And I

would defer to Dr. Doull for the liver function. DR. YASZEMSKI: Thank you.

file:///C|/Storage/1121orth.txt (202 of 309) [12/2/02 8:59:33 AM]

file:///C|/Storage/1121orth.txt

203 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Dr. Naidu? DR. NAIDU: I do have a problem with the Doctor, as

authentic antibody response to rhBMP-2.

the FDA clinician presented--and I sat here at the previous INFUSE meeting--the previous spine approval, the antibody response was only 0.7 percent. This is almost 10 times as much. I don't

know exactly what the clinical relevance is, but it is significant, and therefore, I would concur with Dr. Finnegan with regard to this. The rate of hardware failure, I'm not too concerned--it breaks. And high rate of breakage is That is not

noted in the unreamed group. surprising.

The rate of infection is high, but these are high-energy open tibia fractures, so you're going to expect it to be high. But I did find Dr.

Schmidt's comments very interesting in that in his institution, it is about 5 percent. But the rate

of abnormal lab values with regard to liver function, I will defer to our toxicologist on the panel. DR. YASZEMSKI: Dr. Larntz? DR. LARNTZ: I have nothing to add. Thanks, Dr. Naidu.

file:///C|/Storage/1121orth.txt (203 of 309) [12/2/02 8:59:33 AM]

file:///C|/Storage/1121orth.txt

204 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DR. YASZEMSKI: Dr. Schmidt? DR. SCHMIDT: I don't know what more to I do Thanks.

add about the issue of the immune response.

have a question about why it may be higher in this group than it was in the spine group. It may be a

reflection of the different immune response for a traumatized patient as opposed to someone undergoing surgery. I do think that is something

that probably ought to be looked at. The comments I made earlier about the infections stem from my knowledge that previous studies looking at acute bone grafting in open tibia fractures have shown a high infection rate. There was one study--the most recent one I am aware of was published over 10 years ago, and granted, things have changed since then in management of these cases--but acute bone grafting of open tibia fractures has a dramatically higher infection rate, and what we know from clinical experience is that you need to wait until the soft tissues have healed, typically, 6 to 8 weeks, before you add an autogenous bone graft to a patient who has had an open tibia fracture. If you do it before then, the

infection rate is high, and the graft often just

file:///C|/Storage/1121orth.txt (204 of 309) [12/2/02 8:59:33 AM]

file:///C|/Storage/1121orth.txt

205 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 resorbs and isn't effective. That raises the issue of is this type of composite device going to be effective before the soft tissues have really restored themselves and regained their vascular supply and that sort of stuff, and that is what the study was looking at. In terms of the other issues, I'm not too worried about the hardware failure; I think that may just represent the small-diameter nails. One

point I would like to make, though, is that the term "BMP" is really a misnomer. These are There are

generic morphogenetic proteins.

cell-[inaudible] models that are active throughout development; they affect the whole body. They

affect DNA transcription, and I think it is a mistake to think that these are just a simple protein that is only going to act on the bone. They have potential, far-reaching side effects, and I think that patients who receive this should be followed. I think there is data that some tumors do respond to these prostate and pancreatic cancers. Tumor lines, I think, have shown responsive to BMP-2. I just think that's a small but theoretical

concern that needs to be addressed with long-term

file:///C|/Storage/1121orth.txt (205 of 309) [12/2/02 8:59:33 AM]

file:///C|/Storage/1121orth.txt

206 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 studies. DR. YASZEMSKI: Dr. Aboulafia? DR. ABOULAFIA: Briefly, I do not have any Thanks, Dr. Schmidt.

concerns related to safety of the product and actually feel that the development of antibodies is probably less of an issue than some members have expressed. DR. YASZEMSKI: Ms. Maher? MS. MAHER: I have nothing further to add. Thank you. Thanks, Dr. Aboulafia.

DR. YASZEMSKI: Ms. Rue? MS. RUE:

I have nothing further to add. Thank you.

DR. YASZEMSKI:

Dr. Witten, we have discussed safety in terms of the questions that the FDA has posed to us. There seem to be no issues with respect to

hardware or infections, and the infections seem to be accounted for by the nature of this injury. There have been a few comments made, however, about the nature of the antibody response, that it did greater than previously documented and perhaps warrants looking at these patients over a longer period of time.

file:///C|/Storage/1121orth.txt (206 of 309) [12/2/02 8:59:33 AM]

file:///C|/Storage/1121orth.txt

207 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 this time? [No response.] DR. YASZEMSKI: Seeing no one, I would It was noted that the trauma may in fact be an effect and hence this be a manifestation of the injury that these people have undergone. There was also a question that we should be cautious about the unknown but potentially theoretic effect of a tumor in the future. However, in general, I think the tenor of the conversation was that this device is safe, as shown in the data from the study presented by the sponsor. Have we adequately discussed this from FDA's perspective? MS. WITTEN: Yes. Thank you.

DR. YASZEMSKI:

Thanks, Dr. Witten. We are

That will conclude our discussion. going to proceed now with another open public

hearing session, and I would ask again if anyone in the audience would like to address the panel, please come forward to the microphone, state clearly your name, affiliation, and financial considerations. Would anybody like to address the panel at

file:///C|/Storage/1121orth.txt (207 of 309) [12/2/02 8:59:33 AM]

file:///C|/Storage/1121orth.txt

208 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 like now to ask the sponsor if any of your members have any final comments and would like to address the panel in any way before we proceed with voting. Dr. Swiontkowski? Final Sponsor Comments DR. SWIONTKOWSKI: that I'd like to comment on. There are two areas One is on Dr. I tried

Schmidt's concern regarding the infection.

to point out in my third-to-last slide that using the strict definition in this trial where basically, any redness, whether culture-proven or not, was infected, by that definition in the U.S. 60-patient trial, it was 18 percent, and in this, it was 22 percent. So I think it is related to the strict definition, not really a different outcome as a result of the care. The other area I would like to comment on is the study design and the endpoint selection. have to beg the panel's indulgence for a minute, because I think that the decision here could potentially affect the ability to study trauma patients in general for any condition in the future. When we discussed the whole issue of which I

file:///C|/Storage/1121orth.txt (208 of 309) [12/2/02 8:59:33 AM]

file:///C|/Storage/1121orth.txt

209 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 model to use with our potential clinician investigators, they were adamant about the fact of not having a control, and some of the panelists commented that it would be nice to have a control. They really felt that it was an ethical issue of not putting a sponge into a potentially infected environment. And as Dr. Schmidt just pointed out,

maybe that environment where you have acutely injured soft tissue affected he basic ability of the protein to stimulate healing, which I agree with. I think this is the most severe model. Second, regarding the endpoint, I appreciate Dr. Naidu's comment about we are using the same endpoint in a randomized controlled trial, but this is different. drug. We are talking about a

But I would submit that this is a decision It

to take a patient back to the operating room.

is not to discontinue a drug therapy or something like that. And if there were a bias, you should be

able to see it in the days to decision to intervene, and they were within 2 days in all three of the groups. So there was no bias there.

I would also point out that in the randomized control trial that we are conducting now, we have a rule that you can't intervene within

file:///C|/Storage/1121orth.txt (209 of 309) [12/2/02 8:59:33 AM]

file:///C|/Storage/1121orth.txt

210 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 MR. DEMIAN: 6 months, and 25 percent of the surgeons are violating that rule. That is because as surgeons,

we are always going to act in what we believe is the best interest of the patient, and I think that that is why we have the design that we have here today, because we are dealing with surgeons who are making real decisions that are major decisions to intervene, and it is not a trivial intervention. Thank you. DR. YASZEMSKI: Swiontkowski. I would now like to ask Mr. Demian to read the voting instructions for the panel. Vote I will now provide you with Thanks very much, Dr.

the panel recommendation options for premarket approval applications. "The Medical Device Amendments to the Federal Food, Drug, and Cosmetic Act require that the Food and Drug Administration obtain a recommendation from an outside advisory expert panel on designated medical device pre-market approval applications that are filed with the agency." "The PMA must stand on its own merits, and

file:///C|/Storage/1121orth.txt (210 of 309) [12/2/02 8:59:33 AM]

file:///C|/Storage/1121orth.txt

211 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 the recommendations must be supported by safety and effectiveness data in the application or by applicable, publicly available information." "Safety is defined in the Act as reasonable assurance, based on valid scientific evidence, that the probable benefits to health under the conditions of use outweigh any probable risks." "Effectiveness is defined as reasonable assurance that in a significant portion of the population, the use of the device for its intended uses and conditions of use when labeled will provide clinically significant results." "Your recommendation options for the vote are as follow: 1) Approval. There are no

conditions attached. conditions.

2) Approvable with

You may recommend that the PMA be

found approvable subject to specified conditions such as a resolution of clearly identified deficiencies which have been cited by you, the panel, or FDA staff. All the conditions are

discussed by the panel and listed by the panel chair and then voted on one by one. For example,

you may specify what type of follow-up information the panel or FDA should evaluate prior to or after

file:///C|/Storage/1121orth.txt (211 of 309) [12/2/02 8:59:34 AM]

file:///C|/Storage/1121orth.txt

212 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 approval. Panel follow-up is usually done through

homework assignments by one or two panel primary reviewers or to other specified members of this panel. A formal discussion of the application at a

future panel meeting is not usually held." "If you recommend a post-approval requirement to be imposed as a condition of approval, then your recommendation should address the following points: The purpose of the

requirement, the number of subjects to be evaluated, and the types of reports that should be submitted." The third option is not approvable. Of

the five reasons the Act specifies for denial of approval, the following three reasons are applicable to your panel deliberations: The data

to not provide reasonable assurance that the device is safe under the conditions of use prescribed, recommended, or suggested in the proposed labeling; reasonable assurance has not been given that the device is effective under the conditions of prescribed use recommended or suggested in the labeling; and based on a fair evaluation of all material facts in your discussions, you believe the proposed labeling to be false or misleading."

file:///C|/Storage/1121orth.txt (212 of 309) [12/2/02 8:59:34 AM]

file:///C|/Storage/1121orth.txt

213 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Demian. Before beginning the voting process, I would like to make a point of procedure that was brought up to me, that the sponsors are not supposed to be at the presenters' table during the vote. I recognize the chairs are all tight, but "If you recommend that the application is not approvable for any of these stated reasons, then we ask that you identify the measures that you think are necessary for the application to be placed in approvable form. Traditionally, the

consumer representative and the industry representative do not vote, and Dr. Yaszemski, as panel chairman, votes only in the case of a tie." Dr. Yaszemski? DR. YASZEMSKI: Thank you very much, Mr.

could I just ask you to perhaps back up a little bit or move the table up a little bit? If that's

okay with FDA, we'll just do it that way. Thanks very much. The other point is that I'd like to mention both for the panel's benefit and for the record that votes taken are votes for or against the motion made by the panel. Votes are not votes

in favor of or against the product.

file:///C|/Storage/1121orth.txt (213 of 309) [12/2/02 8:59:34 AM]

file:///C|/Storage/1121orth.txt

214 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 a drug. Ms. Maher. MS. MAHER: I would just like to clarify At this time, I would like to ask if there is a motion to be made. MS. MAHER: Just before the motion. Okay, before the motion,

DR. YASZEMSKI:

one thing that I have heard twice now and I'm a little concerned with. Dr. Naidu brought up the fact that this is In fact, it is not a drug. Under U.S.

law, it is regulated as a device, which means we are looking for reasonable assurance of safety and effectiveness, as opposed to the drug standard. As

we are moving forward to the vote, I would like the panel to remember that. DR. YASZEMSKI: Thank you for that point

of clarification, Ms. Maher. Dr. Finnegan? DR. FINNEGAN: looking for a motion. DR. YASZEMSKI: DR. FINNEGAN: Yes, ma'am. Actually, I think this is This was fairly Mr. Chairman, you were

part of my job description.

difficult, but taking into account what Hany has just outlined, I think there is reasonable

file:///C|/Storage/1121orth.txt (214 of 309) [12/2/02 8:59:34 AM]

file:///C|/Storage/1121orth.txt

215 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 seconded. Before we go to discussion, I'm going to mention another point of protocol. We will assurance that this is safe. I am less sure that

there is reasonable assurance that it is very effective, but it certainly is not deleterious. So my motion is that it be approved with a boatload of conditions. DR. YASZEMSKI: DR. LARNTZ: Thank you.

Second. The motion has been

DR. YASZEMSKI:

entertain, after Dr. Finnegan's conditions which she will read, a discussion regarding addition of conditions or deletion of conditions; we will discuss and vote on each of them independently, and after we have had that discussion, we will re-read the motion as it stands with conditions that have already been voted for inclusion or exclusion and then vote on that motion. Dr. Finnegan, may we hear your conditions? DR. FINNEGAN: one at a time? DR. YASZEMSKI: Let's hear all of them, Then, we'll Do you want all of them, or

because they constitute your motion.

ask others if they want to add or subtract.

file:///C|/Storage/1121orth.txt (215 of 309) [12/2/02 8:59:34 AM]

file:///C|/Storage/1121orth.txt

216 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DR. FINNEGAN: All right. My first

condition is that this should be limited to Grade III open fractures which have been stabilized and the material placed at definitive wound closure. My second condition is that the users should be educated on the potential benefit of adding osteogenic material to this composite. My third condition is that labeling in very large letters needs to deal with unknown factor of repeat use and that perhaps this also be education for the users. My fourth condition is that there need to be two studies on the antibodies just so we have the knowledge. I think part of the problem is that

we don't have any knowledge. I think there could be a natural history study done on the 1,096 patients who have already received this material, and then a prospective study, either animal or human or perhaps both, and I would leave that up to the sponsor and FDA to work out. And my last condition is that I think it is mandatory that there be post-market surveillance, and what are the things they need to answer for that? DR. YASZEMSKI: Thank you.

file:///C|/Storage/1121orth.txt (216 of 309) [12/2/02 8:59:34 AM]

file:///C|/Storage/1121orth.txt

217 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Okay. The purpose of the requirement is to further define the efficacy of the implant. The Those are the five conditions. Dr. Witten? MS. WITTEN: I was going to answer Dr.

Finnegan's question, but perhaps Hany already did. DR. FINNEGAN: My right hand already did.

reason for choosing the Grade III opens is that it will limit undue exposure to the antibodies until we know the antibody history, and these are also the patients who require the maximum help to get their fractures to heal. The number of subjects to be evaluated I think is a number that would produce statistically useful information, and again I would leave that up to the sponsor and the FDA. And the types of

reports that should be submitted I would think would be a composite of this study perhaps with the design improved per the recommendations of the panel and the FDA clinical reviewer. You are looking rather perplexed. MS. WITTEN: Can you just explain the

objective of the study again? DR. FINNEGAN: The purpose of the study?

file:///C|/Storage/1121orth.txt (217 of 309) [12/2/02 8:59:34 AM]

file:///C|/Storage/1121orth.txt

218 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 deletions? Dr. Aboulafia? DR. ABOULAFIA: I think to limit it to stands. Discussion for modifications, additions, MS. WITTEN: DR. FINNEGAN: Yes. It would be to--how am I

going to word this--to further elucidate the potential effectiveness of this composite. you don't like that. MS. WITTEN: DR. FINNEGAN: it--that's the problem. MS. WITTEN: DR. FINNEGAN: It's your motion. Thank you. It's your motion. But I want you to buy Okay,

Dr. Kirkpatrick has it--to clarify the effectiveness of this material as it has been somewhat muddy to the panel. DR. YASZEMSKI: That's the motion as

just Grade III open fractures may have a theoretical advantage that you are going to try to help most those who need it the most. But I think

the truth of the matter is that a problem fracture or a high-risk fracture is not simply just a Grade III fracture. Again, we talked about issues of

file:///C|/Storage/1121orth.txt (218 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

219 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 yes. comorbidities, diabetics, steroid users, distal third fractures versus proximal third fractures, and I think we need to leave that up to the judgment of the treating physician to define what he or she considers to be a potential problem fracture or a patient who may benefit from this project if we're saying there is a theoretical advantage to using this product to promote fracture healing in long bones treated with intramedullary nails. DR. YASZEMSKI: Is your motion to modify

or delete one of Dr. Finnegan's conditions? DR. ABOULAFIA: I would modify it with

saying that the indications would be those patients for whom intervention is thought to be beneficial. DR. YASZEMSKI: There is a motion to

change Dr. Finnegan's condition of limitation to Grade III open fractures--and I'm going to paraphrase, Dr. Aboulafia, so you tell me if I say it right--to those open fractures of any grade that in the opinion of the treating surgeon represent a problem fracture that could benefit from the device. DR. ABOULAFIA: Dr. Aboulafia is nodding

file:///C|/Storage/1121orth.txt (219 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

220 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 motion? DR. LARNTZ: I'll second. Okay. Dr. Finnegan-DR. YASZEMSKI: Is there a second to that

DR. YASZEMSKI: DR. KIRKPATRICK: DR. YASZEMSKI: Okay.

Is that tibia only? This is tibias, yes, sir.

We're going to go around and vote

on Dr. Aboulafia's modification, Dr. Finnegan, to your motion. Do you have commentary on that? DR. FINNEGAN: Well, yes. Actually, I

think we already previously said for labeling that this could be any fracture, so this is not just the tibia--Grade III open fracture of any long bone. DR. YASZEMSKI: Dr. Aboulafia, would that

be okay with your motion--any Grade III open fracture that the clinician considers a problem? DR. ABOULAFIA: I don't know. Does the

panel want to extrapolate the data? DR. YASZEMSKI: we'll vote on it. DR. FINNEGAN: it to tibias. My motion was not to limit If you make that motion,

My motion was to limit it to Grade

III open fractures. DR. ABOULAFIA: Okay, yes.

file:///C|/Storage/1121orth.txt (220 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

221 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DR. YASZEMSKI: around the vote. Okay. We're going to go

We're going to vote on a change Her first

in Dr. Finnegan's first condition.

condition was that this device approval, condition of approval, be that the device be limited to Grade III open fractures. Dr. Aboulafia has made a motion to change that condition to any open fracture that in the opinion of a treating surgeon is a problem fracture that could benefit from use of the device. Dr. Doull, yes or no? DR. DOULL: Yes. Dr. Kirkpatrick? Yes. Dr. Finnegan? No. Dr. Naidu?

DR. YASZEMSKI: DR. KIRKPATRICK: DR. YASZEMSKI: DR. FINNEGAN: DR. YASZEMSKI: DR. NAIDU: No.

DR. YASZEMSKI: DR. LARNTZ:

Dr. Larntz?

Yes. Dr. Schmidt? Yes. Dr. Aboulafia? Yes. The motion passes. So

DR. YASZEMSKI: DR. SCHMIDT: DR. YASZEMSKI: DR. ABOULAFIA: DR. YASZEMSKI:

file:///C|/Storage/1121orth.txt (221 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

222 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Condition 1 of Dr. Finnegan's motion for approval with conditions is that this device be limited to any open fracture that in the opinion of the treating surgeon is a problem fracture and would benefit from the use of the device. Are there any other motions for additions, deletions, or changes to Dr. Finnegan's conditions? DR. KIRKPATRICK: clarification. DR. YASZEMSKI: DR. KIRKPATRICK: Go ahead, Dr. Kirkpatrick. When you talk about I have a question of

safety on the immune response, are you talking about a secondary challenge once someone has converted positive, or an animal has converted positive? DR. FINNEGAN: That is why I said I would I think it

leave the study up to the sponsor.

probably needs to be a combination of an animal and perhaps a person over a period of time, especially if the natural history shows that the antibodies seem to disappear fairly quickly. DR. KIRKPATRICK: So basically, if the FDA

consults with an immunologist and feels that a secondary challenge study is not necessary, we would go with that.

file:///C|/Storage/1121orth.txt (222 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

223 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DR. FINNEGAN: DR. YASZEMSKI: DR. DOULL: Yes. Dr. Doull?

Well, in that case, then, if

you do that study, will you do the history of the antibody response and you determine in people and in animals what that means, then you don't really need the labeling requirement-MR. DEMIAN: DR. DOULL: Speak into the mike, please. The labeling requirement which

says do not repeat--once you have that information, you would no longer need that labeling requirement? DR. FINNEGAN: That is correct, but that

would be withdrawn down the road, because this is going to take some time. DR. DOULL: Okay. Thank you.

DR. YASZEMSKI:

Other motions for additions, modifications, or deletions to Dr. Finnegan's conditions? DR. KIRKPATRICK: DR. YASZEMSKI: DR. KIRKPATRICK: Yes. Dr. Kirkpatrick? One further condition.

I would like the sponsor and the FDA to work out assurances of a statistical nature that the fibula fracture fixation had no effect on either the

file:///C|/Storage/1121orth.txt (223 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

224 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 primary or secondary endpoints and also was not utilized in more fractures that were more severely comminuted, for example, in other words, the AOCs as opposed to the AOAs. If they can provide that

information to the FDA, and everybody feels that that is statistically fine, I would say that we could proceed with approval. DR. YASZEMSKI: This is a motion for the

addition of a condition regarding looking at the presence of the fibula fractures in the study and that the sponsor would get this data together for the FDA. Is there a second, first, before we discuss it? DR. LARNTZ: I'll second. There is a second.

DR. YASZEMSKI:

Discussion, Dr. Larntz. DR. LARNTZ: If it would be all right with

Dr. Kirkpatrick, I would like to expand that to the use of reamed and unreamed nails also, as a covariate to the study. DR. YASZEMSKI: If everybody is okay--Dr. If you are okay

Kirkpatrick, it is your motion. with adding that to it-DR. KIRKPATRICK:

I would prefer the two

file:///C|/Storage/1121orth.txt (224 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

225 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 issues remain separate. DR. YASZEMSKI: it separate. Discussion? [No response.] DR. YASZEMSKI: Dr. Aboulafia? DR. ABOULAFIA: Kirkpatrick's motion. DR. YASZEMSKI: DR. SCHMIDT: DR. YASZEMSKI: DR. LARNTZ: Dr. Schmidt? In favor. Dr. Larntz? In favor of Dr. Let's vote. Okay. We're going to keep

Yes. Dr. Naidu?

DR. YASZEMSKI: DR. NAIDU:

Can I abstain? Yes, you may abstain.

DR. YASZEMSKI: DR. NAIDU:

I abstain. Abstention.

DR. YASZEMSKI: Dr. Finnegan? DR. FINNEGAN: DR. YASZEMSKI: DR. KIRKPATRICK: DR. YASZEMSKI: DR. DOULL: Yes.

Yes. Dr. Kirkpatrick? Yes. Dr. Doull?

DR. YASZEMSKI:

The motion passes.

file:///C|/Storage/1121orth.txt (225 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

226 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 deletions? DR. ABOULAFIA: DR. YASZEMSKI: Can you repeat each one? Seeing none, we're going There is now a sixth condition. Any other discussion, additions,

to do this. I'm going to go over Dr. Finnegan's motion as it is now with the conditions and then call for a vote, unless there is any further discussion that anybody would like to bring up. DR. KIRKPATRICK: Excuse me. Just a point

of order. There was a suggestion about looking at nonreamed and reamed nails. DR. YASZEMSKI: and heard none. DR. KIRKPATRICK: DR. YASZEMSKI: Okay. Thank you. I asked for another motion

The motion is for approval with conditions. There are six conditions--number one,

that the use of the device be limited to open fractures that in the opinion of the treating surgeon represent a problem fracture that would benefit in his or her clinical judgment from the use of the device; number two, that there be user education regarding this device; number three, that the labeling include a statement that the factor of

file:///C|/Storage/1121orth.txt (226 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

227 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 repeated use is at present unknown with respect to antibodies; number four, that there be two studies on antibodies, one, a natural history study of the 1,096 patients who already have been in the study, and number two, either an animal or a human study as determined by the sponsor and the FDA; number five, that there be post-market surveillance to clarify the issue of acceleration of fracture healing; number six, that the sponsor and the FDA will work out the statistics regarding the presence of a fibula fracture and whether, when accounted for, that fibula fracture had any effect on the results as presented. This is the motion. discuss it further? DR. KIRKPATRICK: fracture fixation. DR. YASZEMSKI: Thank you for that Fibula fracture The motion was fibula Would anybody like to

clarification, Dr. Kirkpatrick. fixation. Thank you. Further discussion? [No response.] DR. YASZEMSKI: Dr. Aboulafia? DR. ABOULAFIA:

We're going to vote.

In favor.

file:///C|/Storage/1121orth.txt (227 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

228 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DR. YASZEMSKI: DR. SCHMIDT: DR. YASZEMSKI: DR. LARNTZ: Dr. Schmidt? In favor. Dr. Larntz?

Yes. Dr. Naidu?

DR. YASZEMSKI: DR. NAIDU: No.

DR. YASZEMSKI: DR. FINNEGAN: DR. YASZEMSKI: DR. KIRKPATRICK: DR. YASZEMSKI: DR. DOULL: Yes.

Dr. Finnegan? Yes. Dr. Kirkpatrick? Yes. Dr. Doull?

DR. YASZEMSKI: Thank you. Mr. Demian? MR. DEMIAN:

The motion passes.

We are going to go around the

room and ask people why they voted the way they did. DR. YASZEMSKI: Thank you.

Dr. Witten, thank you for reminding us of this important function. What we are going to do for the benefit of the FDA and the sponsor and the public at-large is poll the panel and ask each of them why they voted the way they did and what they consider the

file:///C|/Storage/1121orth.txt (228 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

229 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 positives and negatives that affected their vote. Dr. Aboulafia? DR. ABOULAFIA: I think this is one of

those things where sponsor may not appreciate my help. I think a large portion of this decision was

based on the safety, which I think it is a safe product. I think the issue of effectiveness is at best weakly demonstrated. And my hope is that with

the power of some of the coinvestigators who have national and international reputations both in fracture management and in study design, they may be able to better design a study that will clearly demonstrate the effectiveness of this product, and that what I am really allowing is fair market or free society to determine whether the cost-benefit analysis is worthwhile or not. DR. YASZEMSKI: Dr. Schmidt? DR. SCHMIDT: I agree with Dr. Aboulafia's Thanks, Dr. Aboulafia.

comments whole-heartedly and have no other ones to add. DR. YASZEMSKI: Dr. Larntz? DR. LARNTZ: I believe this product is Thank you.

file:///C|/Storage/1121orth.txt (229 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

230 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 safe. I believe they have shown effectiveness with With respect to

respect to secondary intervention.

other endpoints, I think it is quite questionable. DR. YASZEMSKI: Dr. Naidu? DR. NAIDU: I said "no" mainly because in Thanks, Dr. Larntz.

my opinion, the preclinical data is mixed; the clinical data has too many confounding factors, as previously presenters have clearly demonstrated; x-ray data was not complete. Healing criteria was

varied for secondary intervention groups, and I just did not feel right voting "yes" for it. Thank you. DR. YASZEMSKI: Dr. Finnegan? DR. FINNEGAN: I have to sort of agree Thanks, Dr. Naidu.

with Dr. Naidu in that I do think that both the sponsor and some of the investigators are capable of a much better study, and I agree with Dr. Aboulafia that I hope that will in fact occur. But I do think it is safe, and I do think there are patients who will actually benefit from its use. DR. YASZEMSKI: Dr. Kirkpatrick? Thanks, Dr. Finnegan.

file:///C|/Storage/1121orth.txt (230 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

231 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Witten. Thanks, everybody on the panel. We're going to take a 5-minute break now, and then reconvene. DR. KIRKPATRICK: I think the

effectiveness data is borderline. I hope that the fibula does not cause any change in my opinion that I think it is marginally effective, or does offer some improvement. I think with the condition that we have on the safety issue with regard to the antibody response, if that is satisfactory, then I think it is a reasonable decision to go approval. DR. YASZEMSKI: Dr. Doull? DR. DOULL: The issues are safety and Thanks, Dr. Kirkpatrick.

effectiveness, and as a clinical toxicologist, I am of course influenced primarily by the safety issue. But I think the arguments presented for efficacy, weakly efficacious, are satisfactory to me. DR. YASZEMSKI: Thanks, Dr. Doull.

Dr. Witten, have we adequately discussed this to FDA's satisfaction? MS. WITTEN: Yes. Thank you. Thanks very much, Dr.

DR. YASZEMSKI:

file:///C|/Storage/1121orth.txt (231 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

232 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 [Break.] General Panel Discussion of Spinal Devices DR. YASZEMSKI: I'll ask everyone to

please take their seats, and we're going to get started with the spinal portion of today's meeting. We are going to have a general panel discussion regarding spinal devices this afternoon. FDA has provided us with a list of preclinical and clinical questions related to the evaluation of fusion and nonfusion spinal devices. We are going to have an open public hearing session regarding this general discussion on spinal devices, and we have had six persons who have requested speaking time. I will mention to the folks in the audience that these folks have put handouts outside the door, if anybody is interested in getting them. We are going to ask the speakers, please, in the interest of keeping on time, to limit your comments to 5 minutes. Each of the six speakers

will have 5 minutes to speak. I would ask that all persons addressing the panel come forward and speak directly into the microphone, as the transcriptionist is dependent on this means for providing an accurate record of the

file:///C|/Storage/1121orth.txt (232 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

233 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 meeting. And we request again that all persons

making statements during the open public hearing disclose whether they have financial interest in any medical device company. Before making your presentation to the panel, please state your name, your affiliation, and the nature of your financial interest, if any. The first person will be Mr. Antonio Valduvit [phonetic]. Are you here, sir? [No response.] DR. YASZEMSKI: Not in attendance.

The next person who has asked to speak is Ms. Brenda Seidman. Is she here? Thank you. Welcome.

Open Public Hearing DR. SEIDMAN: My name is Dr. Brenda Services. I am

Seidman, of Seidman Toxicology

both a general and neurotoxicologist and an active participant in six ISO 10993 committees. ISO 10993

is the international standard on the biological evaluation of devices. I have served as a consultant to orthopedics manufacturers over the last several years. Today I represent myself.

file:///C|/Storage/1121orth.txt (233 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

234 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 My purpose in speaking today is to respond to FDA's proposal to include a requirement for particle injection studies in its spinal implant guidance documents. As a participant in the ISO 10993 standard-setting process, I suspect the agency may be confusing foreign body reactions with chemically-induced toxicity. Foreign body

reactions are inflammatory reactions that appear to be related to the loosening of orthopedic implants. Such reactions are responses to the physical properties of a material, such as size, shape, and surface properties. Chemical toxicity related to a device is a biological response to its chemical leachates. ISO 10993 currently addresses chemical but not foreign body reactions. As such, ISO 10993 is

the most appropriate mechanism for addressing chemical toxicity. For those unfamiliar with the

standard, it addresses the chemical toxicity of materials from finished devices by several means, including chemical characterization, clinical history of use, the scientific literature, and in vitro and in vivo testing on both extracts of the material and on the material itself.

file:///C|/Storage/1121orth.txt (234 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

235 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Potential chemical toxicity related to wear debris can and should be addressed using ISO 10993. The agency may be concerned that fatigued materials might be chemically different than non-fatigued materials. Although I am personally

unaware of such changes, manufacturers would need only to demonstrate that chemical differences do not exist in order to rely on their ISO 10993 evaluations of their devices' non-fatigued materials from their finished devices. Next transparency, please. the sake of time. Now with regard to testing for foreign body reactions. The agency appears to have made I'll go on for

the assumption that in vivo testing is the only or most suitable means of testing for foreign body inflammation. To my knowledge, neither ISO nor

ASTM have developed or validated a test method in vitro or in vivo for an evaluation of foreign body effects. Therefore, it makes sense to consider the possible use of in vitro test methods. In vitro

tests have the potential be more sensitive and focused, raise no animal welfare issues, could be

file:///C|/Storage/1121orth.txt (235 of 309) [12/2/02 8:59:35 AM]

file:///C|/Storage/1121orth.txt

236 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 performed in a shorter period of time, and would be less costly and burdensome to industry. Furthermore, it may not be reasonable to assume that a device's wear debris from an animal model will be sufficiently similar to that generated after implantation in humans. Wouldn't there be

different wear and load scenarios that would be difficult if not impossible to replicate using an in vivo model? Second, with all due respect to FDA and the panel, isn't there value in developing proposed biological testing methods within larger consensus groups, such as relevant ISO and ASTM working groups? In the interim, is there value in

requiring manufacturers to develop and perform nonvalidated tests with objectives not carefully articulated by the agency? Third, should the agency's goal be to use the medical device industry as a methods incubator? While methods need to be developed to characterize the risks associated with wear debris, it is unlikely that forcing the development of methods into the submissions process is scientifically justifiable, effective, consistent with animal welfare regulations, or otherwise ethical.

file:///C|/Storage/1121orth.txt (236 of 309) [12/2/02 8:59:36 AM]

file:///C|/Storage/1121orth.txt

237 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 I have provided the panel with numerous technical questions regarding proposed testing, and those are made available outside on the table. Unfortunately, these questions are impossible to fully present within my allotted time. To summarize, however, my questions relate to the following: The rationale for performing

foreign body evaluations when no wear debris is generated under physiological conditions; the characteristics of wear debris we consider relevant to foreign body reaction; the dynamic aspects of particle generation; the possible role of pre-existing pathology; and the considerations for selecting the most appropriate model for either an in vitro or in vivo test. Last, given the need for basic research--not research as part of a submissions process; they are two different things--I would like the panel to come up with suggestions on what the agency and industry can do to satisfy debris safety concerns while validated methods are being developed. to there? Thank you. DR. YASZEMSKI: Thanks very much, Dr. In other words, how do we get from here

file:///C|/Storage/1121orth.txt (237 of 309) [12/2/02 8:59:36 AM]

file:///C|/Storage/1121orth.txt

238 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Seidman. Next, Diane Johnson. MS. JOHNSON: My name is Diana Johnson. I

am a full-time employee of Medtronic Spinal Dynamics. With respect to one of the questions posed by FDA, Spinal Dynamics is requesting that the advisory panel specifically consider FDA's position on durability testing and its relationship to wear particulate testing. In the questions, FDA suggests durability testing should be conducted and the loads and motions utilized in the testing should be justified. In a related question, FDA is also requesting information related to the biologic effects of particulate that is generated by devices following implantation. Spinal Dynamics is requesting that the panel consider the following specific questions. Simulation testing for hips and knees is conducted utilizing loads and motion representative of activities of daily living, as opposed to maximum loads and motions. This approach is based

on 20 years of research which shows that explanted

file:///C|/Storage/1121orth.txt (238 of 309) [12/2/02 8:59:36 AM]

file:///C|/Storage/1121orth.txt

239 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 devices show wear similar to that produced during simulation testing conducted at the ADL load and motion profile. This testing methodology is currently recommended in ASTM testing methodologies for hips and knees, which indicate that loads and motions should be those associated with walking. In light of the correlation that is clearly established for other joints, is it appropriate to perform simulation testing for functional devices at activities of daily living or at maximum loads and motions at some time suggested by FDA? If the biologic effect of particulate is approximated using simulator-generated particulate, should the particle size and distribution and quantity be determined using a load and motion profile associated with the activities of daily living? If the biologic effect of particulate is approximated using particulate generated in an animal model, there is likely to be some level of nonphysiologic loading, especially if a quadruped is utilized. This may result in the production of

particles that would not be generated in humans under physiologic loading.

file:///C|/Storage/1121orth.txt (239 of 309) [12/2/02 8:59:36 AM]

file:///C|/Storage/1121orth.txt

240 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Does the generation of particulate due to nonphysiologic loading necessarily invalidate the model in terms of the evaluation of effects of particulate that is wear-generated in the model? Thank you. DR. YASZEMSKI: Thank you very much.

Next, Dr. Bailey Lipscomb. DR. LIPSCOMB: Members of the panel, my

name is Bailey Lipscomb, and I am Vice President of Clinical Affairs at Medtronic Sophomore Danik in Memphis, Tennessee. We appreciate the opportunity

to make a few comments concerning issues that affect our IDE clinical studies both for nonfusion and for fusion spinal implants. First, after having sponsored numerous clinical trials on spinal implants, we believe that oswestri [phonetic] pain success should be based on a percent improvement from baseline rather than the current FDA-mandated 15-point improvement. recognizes that preoperative scores have a substantial range and that it is easier for a patient with a preoperative score of 80 to improve 15 points as opposed to a preoperative score of 40. We recommend a success for oswestri [phonetic] or its cervical counterpart, the MDI, be This

file:///C|/Storage/1121orth.txt (240 of 309) [12/2/02 8:59:36 AM]

file:///C|/Storage/1121orth.txt

241 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 a 20 percent improvement form baseline. Second, we are concerned about FDA's recent condition on assessing neurological status in IDE clinical study patients. This is a very

important consideration, since neurological status is one of the components of the overall success criteria which is the primary outcome variable. Along with neurological success, you have fusion, you have pain success, and you have no serious device-related safety issues. the components of overall success. These are

Therefore,

anything affecting neurological success directly impacts the overall success rate and ultimate conclusions from the study. Heretofore, the premise for classifying a patient as a neurological success is that their overall neurological condition after surgery is no worse than it was before surgery, and there were means for summarizing the approximately 40-plus assessments of sensory, motor function, and reflexes to make this determination. Recently, FDA has required that neurological success be based on no worsening of any single measurement in the entire neurological assessment. Stated another way, if any one of the

file:///C|/Storage/1121orth.txt (241 of 309) [12/2/02 8:59:36 AM]

file:///C|/Storage/1121orth.txt

242 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 40 post-operative measurements is worse than the preoperative measurement, then the person is a neurological failure and therefore an overall success failure for the study. Stated another way, a person could be fused, have dramatic pain relief, no adverse event, and intact both in terms of sensory and motor function, but could be a neurological and an overall failure because one reflex measurement after surgery was worse than the preoperative measurement. And that is in light of the fact that

other reflex measurements could have improved. We believe that FDA's current requirement for interpreting neurological results exceeds the prior intent of discerning whether a patient is as neurologically intact after surgery as before. believe this new directive may misrepresent the true neurological status of study patients and will inappropriately lower study success rates. The third point pertains to unnecessarily large clinical study sizes. As you are aware, many We

spinal implant studies are noninferiority trials in which investigational treatment is compared to a standard of care control. Sample sizes for these

studies are primarily impacted by two factors--one,

file:///C|/Storage/1121orth.txt (242 of 309) [12/2/02 8:59:36 AM]

file:///C|/Storage/1121orth.txt

243 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 the overall success rate for the study, and two, the noninferiority margin. I will not dwell on the

overall success rates, even though their multi-component nature drives their values toward 50 percent in higher sample sizes. Rather, I want

to focus on the selection of the noninferiority margin, or delta. The impact of delta selection is dramatic. A difference of one percentage point can add 100 patients to an overall study size. Presently, FDA

advises that delta not exceed 10 percent regardless of the success rate. We believe that this

one-size-fits-all approach is inappropriate and that delta should vary with the success rate. For

example, in a range of success rates between 50 and 95 percent, the delta could vary between 12-1/2 and 7-1/2 percent, respectively. Statistical

literature supports our proposal. The deltas described above will yield a sample size of about 450 patients in a two-arm study, adequate enough to characterize the safety and effectiveness of a spinal implant. Otherwise,

the sample sizes can approach 700 patients. Clinical studies of this magnitude are very burdensome, delay the availability of new

file:///C|/Storage/1121orth.txt (243 of 309) [12/2/02 8:59:36 AM]

file:///C|/Storage/1121orth.txt

244 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 technologies to surgeons, discourage the pursuit of new treatment modalities, and most importantly, are unnecessary. If you are concerned that you may approve a device that is observed to be more than 10 percent worse than the control, the reality is that this won't happen. Even with a delta of 12-1/2

percent and a sample size of 450 patients, noninferiority could not be claimed if the observed control success rate were 60 percent versus a 55 percent rate for the investigational group--only 5 points different. In conclusion, we appreciate the panel considering these three points and would like to thank FDA for the opportunity to make these comments. Thank you. DR. YASZEMSKI: Dr. Jansen? DR. JANSEN: My name is Rich Jansen. I am Thanks very much.

Vice President of Regulatory and Clinical Affairs at Disc Dynamics, Incorporated. Disc Dynamics is an early-stage medical device company in the Minneapolis area, developing a disc nucleus prosthesis.

file:///C|/Storage/1121orth.txt (244 of 309) [12/2/02 8:59:36 AM]

file:///C|/Storage/1121orth.txt

245 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 I would like to comment on three issues being discussed here today. First, I would like to

suggest that the radiographic endpoint of measuring fusion or motion on flexion and extension films should be a secondary success criterion. I have

talked with many surgeons who have argued that patients do not come to their offices requesting a fusion. Their concern is that they have

intolerable pain, they cannot go to work, they cannot pick up their kids, and other activities of daily living. Study success should be measured as clinical improvement such as pain and function. should measure and report range of motion using flexion and extension films, but this should be a secondary endpoint. Regardless of the final radiographic results, it is the patients that we should be concerned about, and patients will consider their surgery a success if they have manageable pain, can go back to work and function in their usual daily activities. Secondly, I would like to point out that we have been unable to find an acceptable model for a disc nucleus prosthesis. With regard to the We

file:///C|/Storage/1121orth.txt (245 of 309) [12/2/02 8:59:36 AM]

file:///C|/Storage/1121orth.txt

246 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 baboon, which is the most frequently suggested model, this may be a good model for fusion devices or for total disc replacements, but not for disc nucleus replacements. We conducted the study using a baboon model and found several limitations with this model, with less than desirable results. The disc

space is so narrow that even in the hands of a very good and experienced surgeon doing animal research with this model, many of the endplates were damaged during surgery in both the sham-operated level and the nucleus-implanted level, indicating that this was a result of the discectomy procedure, not the device. In addition, the nucleus cavity in the baboon is so small that the portal of entry to gain access to the nucleus is about the same size as the implant, leading to a high rate of extrusion. This

is not at all the case in humans, where the nucleus cavity and implant size are many times larger than the access port needed to implant the Disc Dynamics device. Finally, we found extensive heterotopic ossification in both the sham-operated levels and the implanted disc levels at 3 months. These

file:///C|/Storage/1121orth.txt (246 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

247 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 issues make this model unsuitable for mechanical evaluations. We have also tried the mini pig model and found the same limitations with that as we did with the baboon model. So unfortunately, we do not believe that there is an adequate model to address disc nucleus prostheses at this time. The third issue I would like to address is a statement made in the draft questions for this meeting. 11-12-02. these are the draft questions dated There is one sentence in here that I

would like to read. "Because devices not intended for fusion are intended to stabilize the spinal motion segment and retain functional motion, they must be designed to last the lifetime of the individual rather than until fusion has occurred." Total disc replacements and disc nucleus replacements are frequently referred to as spine arthroplasty devices. If we look at our orthopedic

counterparts at hip and knee arthroplasty, we know that they are not expected to last the lifetime of all patients. Based on the type of spine

arthroplasty device, there are some devices that do

file:///C|/Storage/1121orth.txt (247 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

248 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 not preclude reasonable and appropriate follow-up surgical procedures if required. I would like to suggest that we look at each type of device before concluding that all spine arthroplasty devices must be designed as lifetime implants. Thank you for the opportunity to consider these issues. DR. YASZEMSKI: Dr. Norton? DR. NORTON: My name is Britt Norton, and Thank you.

I am Vice President of Research and Development with Raymedica in Minneapolis. Mr. Chairman, distinguished members of the panel, thank you for the opportunity to speak with you. We agree with the objectives and many of the concepts in the spinal guidance document and draft questions, but there are a few areas that we feel merit further discussion. [Slide.] In developing test methods for prosthetic disc nucleuses, which I am going to focus on today, the fundamental differences between the nucleus replacement and fusion must be recognized.

file:///C|/Storage/1121orth.txt (248 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

249 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Specifically, nucleus devices are intended to maintain segmental motion and perform their stabilization in that manner. Fusion devices, of

course, are intended to eliminate segmental motion. Nucleus devices are intended to function in concern with the surrounding tissues of the endplates, the annulus, the ligaments, and the facette joints. Fusion devices, of course,

generally render the surrounding tissues more nonfunctional. [Slide.] These differences in approach to stabilization are reflected in device function and the component materials as well. Nucleus devices are intended to mimic load-deformation behavior of a normal disc, while fusion devices replace that normal disc with solid bone mass. Current designs for nucleus devices utilize elastomeric polymers, which have a high strain capability and high energy absorption capability. Of course, fusion materials are rigid;

the metals are polymer composites possessing low strain and are more energy-transmitting than absorbing. Elastomeric materials used in nucleus

file:///C|/Storage/1121orth.txt (249 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

250 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 devices have high fatigue durability, more metals, and fusion devices are prone to embrittlement following fatigue. With these things in mind, I would like to very briefly talk about the tests described in the questions proposed. Compression fatigue,

durability/shear testing, I will lump together; migration, expulsion, creep and stress, relaxation, and potential for generating wear debris. [Slide.] With regard to compression fatigue testing, the proposed test is more appropriate for metallic than elastomeric constructs. Specifically, an asymptotic endurance limit as proposed is typically used to describe strain hardening and embrittlement of metals due to fatigue, whereas elastomers can accommodate relatively large strains, thereby reducing the value of this test for testing nucleus devices. It is also mentioned that the proposed requirement for an appropriate control device should be reconsidered as for this type of device, one does not currently exist. [Slide.] With regard to durability testing, the

file:///C|/Storage/1121orth.txt (250 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

251 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 proposed test is more appropriate for total disc replacement, especially regarding wear debris. With the total disc replacement, segmental motion results in relative movement of the rigid device components offering a large shear component within the device. With nucleus devices, segmental motion

results in more focal compression of the elastomeric device, with much smaller amount of shear compression. [Slide.] With regard to shear testing, we suggest that a single fatigue compressive shear test could satisfy the intent of compression fatigue, durability and shear test for this kind of device. Compressive testing using angled platens can provide both compressive and shear forces similar to the lordotic geometry of the lumbar spine. We propose evaluating device functionality

following fatigue rather than determining asymptotic endurance limits, and also propose that tests be performed in simulated physiologic environment, in saline solution, and the solution can then be evaluated for particulates following the test. [Slide.]

file:///C|/Storage/1121orth.txt (251 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

252 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 With regard to migration and expulsion testing, there is a basic concern here that the mechanism responsible for migrations and expulsions seen clinically in some devices has not been documented. Development, then, of a bench test to

evaluate this would not be validateable. Should animal testing be considered, the differences in loads and disc geometries between human and animal lumbar discs would prevent the use of this test data to predict migration and expulsion in humans. Should simulated use testing in cadaver tissue be considered, for instance, using cyclic complex motion, the natural degeneration at even room temperature of this type of material would really limit the applicability of this type of test. So the best evaluation for migration and expulsion is controlled clinical study. The risks

associated with clinical migration and expulsion so far appear to be no greater than reherniation following discectomy. [Slide.] With regard to creep and stress relaxation testing, it is important to note that the choice of

file:///C|/Storage/1121orth.txt (252 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

253 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 test parameters such as load or load duration must be made based on the objectives of the test. Physiologically relevant test conditions should be used when evaluating finished devices. Nonphysiologically relevant test conditions can be used to evaluate component materials, but cannot be used then to predict in vivo device performance. The value of such a test would then become questionable. [Slide.] Lastly, potential for generating wear debris, and specifically to the question of using animal models. The viability of a functional

animal model is unproven. Animal models are not adequately functional with regard to a prosthetic nucleus, again, a device designed to maintain motion and biomechanical function of the disc. Specifically, there are significant disparities between humans and animals with regard to load-generating activities, posture, and ranges of motion. Specifically for quadrupeds, their

spines generally lack biconcave endplate shapes in humans. Combining these provides a great potential

for device expulsion, requiring the use of an unacceptable number of animals, to then gain

file:///C|/Storage/1121orth.txt (253 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

254 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 questionable results. [Slide.] Animals that possess some bipedal ability, such as baboons, have small disc spaces that will require miniaturized devices. The problem with

this is that proportional scaling of these devices, especially those that are composite in nature, with expansion-limiting components, the proportional scaling will likely have a nontrivial effect on device performance due to material and geometric nonlinearities. To develop such a device for a baboon, say, the test would effectively become an evaluation of a device designed for that animal rather than one designed for humans. [Slide.] In summary, prosthetic nucleus devices, total disc replacements, and fusion devices all aim to stabilize the spinal segment but do so in completely different ways. The tests used to

evaluate these devices must recognize these differences and be specifically designed for each type of device. I also want to mention that the appropriate patient groups for each type of device

file:///C|/Storage/1121orth.txt (254 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

255 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DR. BUCH: here? [No response.] DR. YASZEMSKI: No. Before we conclude may not be the same, and as such, any one type of device will likely not be an appropriate test control for another type of device. Thank you. DR. YASZEMSKI: Thank you.

May I ask one more time if Mr. Valduvit is

the open public session, would anyone else like to address the panel? [No response.] DR. YASZEMSKI: Seeing none, we will

proceed with Dr. Buch and the FDA lead presentation. FDA Presentation Hello again. My name is Dr.

Barbara Buch, and I am an orthopedic surgeon and member of the FDA's Orthopedic Devices Branch. At the outset, I'd like to stress that this discussion is not related to any specific device, nor is it related to any one specific preclinical or clinical trial related to any specific device. It is my fervent hope that this discussion

file:///C|/Storage/1121orth.txt (255 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

256 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 will endeavor to provide the panel with provocative questions to stimulate a discussion which may ultimately provide information to aid and update our current thinking regarding guidelines for clinical trials which strive to prove the safety and effectiveness information for many spinal devices. This will include both preclinical and clinical aspects of clinical trials and will focus on the study of emerging spinal technologies. Just by way of background, in January of 2000, the FDA issued the Guidance Document for the Preparation of IDEs for Spinal Systems. Prior to

its issuance, ORDB presented a preliminary background document to the Orthopedic and Rehabilitation Devices Panel. During the October

8, 1998 panel meeting, input was received from panel members and the public which resulted in the current guidance document which is available to the public. At that time, the FDA requested some input on nonfusion devices which are not intended to facilitate fusion of the spine. Unlike fusion

devices, these devices allow some functional motion through various levels of the spine. These include

file:///C|/Storage/1121orth.txt (256 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

257 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 device that provide stability while continuing to allow some percentage of normal or functional motion, devices which allow motion and growth, and devices which stabilize vertebral body and spinal fractures. Examples of these devices were included in the references enclosed in the panel package. The current spinal guidance focuses primarily on spinal fusion devices for various etiologies, with brief guidance on such nonfusion devices as vertebral body replacements and disc replacements. While the FDA guidance for spinal implant 510(k)s issued September 27, 2000 outlined in detail devices intended for fusion, and there is a voluntary testing standard available for pedicle screw systems and intervertebral body fusion devices, these being ASTM 1717 and ASTM 2077, respectively, many sponsors have chosen to modify versions of these two testing standards to address different types of spinal systems. Because there a\re currently testing standards in development, the FDA has asked sponsors to contact appropriate standards bodies, including ASTM and ISO, for additional information.

file:///C|/Storage/1121orth.txt (257 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

258 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 As the scope of spinal devices expands, the FDA recognizes that the need to update the spinal guidance to include additional clarification and suggestions for preclinical testing, clinical assessments, endpoints, and success determinations related to emerging spinal technologies is necessary. Therefore, I will begin with what is typically our current recommendations to companies. Currently, the FDA guidance for spinal implant 510(k)s, which extensively covers devices intended for fusion, recommends various static and fatigue testing for spinal devices. Because

devices not intended for fusion are intended to stabilize the spinal motion segment and retain functional motion, they must be designed to last the lifetime of the individual rather than until fusion occurs. This corresponds to the definition

of a spinal implant as stated in the guidance. Therefore, the current testing typically requested for devices intended for nonfusion may not be adequate. In addition, the current testing

for fusion devices made of other materials than stainless steel and titanium may also not be adequate.

file:///C|/Storage/1121orth.txt (258 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

259 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 The FDA currently requests the following testing for spinal devices. For fusion devices,

for example, devices intended to stabilize by fusing motion segments, those being pedicle screw systems, intervertebral fusion devices, and vertebral body fusion devices, fatigue testing is requested. This should involve a minimum of six samples of the worst-case construct to generate

stress or load versus the number of cycles in a curve that characterizes the asymptotic endurance limit compared to an appropriate control device. The rationale for components chosen as the worst-case scenario should be provided by the sponsor. The interconnection mechanisms or systems may be tested in the same set of constructs, or each in a separate set of constructs. Each

interconnection mechanism should be tested, or an adequate rationale for not testing the interconnections is asked to be provided. Additionally, testing should be performed out to a minimum run of 10 million cycles for intervertebral body replacement devices intended for tumor patients, because these patients may

file:///C|/Storage/1121orth.txt (259 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

260 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 represent a great difficulty in achieving fusion. Therefore, this device is acting more like a stabilizer in this condition. The second test that is required is a static test. This should involve a minimum of five As with

samples of the worst-case construct.

fatigue testing, the components tested and the loading mode should be justified. Examples of these types of construct testing typically performed for a given type of spinal system in order to establish relative safety are as follows. For lumbar and thoracic pedicle

screw systems that are intended for fusion, both static and fatigue testing as well as bending testing should be provided, this in accordance to ASTM Standard 1717. For cervical, pedicle, or lateral mass systems intended for fusion, static and fatigue testing should also be provided. The loading mode,

either torsional or bending, is dependent on the design and the material. For intervertebral body fusion devices, static and fatigue testing again should be provided. The loading mode, which may be axial,

torsional, bending, or shear, is dependent on the

file:///C|/Storage/1121orth.txt (260 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

261 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 design, the material, and the levels and number of levels of use. Finally, for vertebral body replacement devices, static and fatigue testing in bending and torsional loading modes should be provided. Now let's look at nonfusion devices. These are devices intended to stabilize the spine yet retain some kind of functional motion over a wide range. These might be things like disc

nucleus replacements, intervertebral disc prostheses, and screw- or hook-based stabilization spinal systems that do not attempt to afford a fusion. These are some of the items that the FDA believes might be appropriate to consider. first is compression fatigue. The

This fatigue testing

should involve a minimum of six samples of the worst-case construct to generate a stress or load versus the number of cycles curve that characterizations the asymptotic endurance limit, which is then compared to an appropriate control. The rationale for the components chosen as worst case again should be provided. The next testing is durability testing. Durability testing should involve cyclical loading

file:///C|/Storage/1121orth.txt (261 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

262 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 testing in several loading modes--for example, in flexion-extension, lateral bending, and axial rotation. And it should involve a minimum of six

samples of the worst-case construct carried out to 10 million cycles. This test can either be combined to incorporate all testing directions to one test or separated into each loading mode. Durability testing establishes loading direction, the stability of the device, and wear generation potential. Clinical justification for

the loads and angles chosen are asked to be provided. Static compression testing should involve a minimum of five samples of the worst-case construct. As with fatigue testing, the components

tested and the loading mode should be justified. Other potential tests that are not as clearly defined might be: migration and expulsion

testing, static and dynamic shear testing, creep, and stress-relaxation testing. Examples of the types of construct testing typically performed for a given type of spinal system in order to establish relative safety are as follows for these nonfusion-type devices.

file:///C|/Storage/1121orth.txt (262 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

263 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 For vertebral disc replacement, static and fatigue testing in multiple-load modes should be provided out to 10 million cycles. For stabilization pedicle screw systems intended for nonfusion, dynamic shear testing and torsion testing would be provided. For nucleus replacements, expulsion testing is requested. For nucleus replacements as

well, fatigue compression testing on new and aged devices should be provided. For devices with polymer components, creep and/or stress-relaxation testing should be provided. Depending on the design of the system, the sponsor may need to perform different tests in lieu of those identified above, perform additional tests in different testing modes, and provide testing on individual components of the subject system. While there is a voluntary standard available for pedicle screw systems intended for fusion and for intervertebral body fusion devices, many sponsors have used modified versions to

address different types of spinal systems relative to their device. Because there are testing

standards in development for these devices as well,

file:///C|/Storage/1121orth.txt (263 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

264 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 issues. sponsors are advised to contact appropriate standard bodies for information regarding test setups and parameters for their specific device. Now, having this background, we'll get to the questions at hand. The first question deals with preclinical We would like the panel to please comment

on the currently-recommended preclinical mechanical debris and wear testing to evaluate new materials, device properties and integrity, and the wear debris for fusion and nonfusion devices. Within

this, we would like you to discuss what additional testing, if any, should be added to current testing recommendations for the following devices, and we would like to ask you to make your comments for each of the following subcategories of nonfusion devices--intervertebral disc or joint replacements that can be placed in the cervical or thoraco-lumbar areas; stabilization devices for nonfusion; intervertebral disc nucleus replacements, and devices manufactured out of new materials. The second preclinical issue is this. FDA is currently requesting information for any device used in the area of the spinal cord and The

file:///C|/Storage/1121orth.txt (264 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

265 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 nerve roots that has the potential to generate debris regarding local and systemic effects. For

those incorporating new materials such as polymers or composites or other designs for both fusion and nonfusion, the FDA currently recommends that manufacturers perform wear simulations and fatigue tests to evaluate the potential for the device to generate wear debris. The FDA believes that the wear debris generated from these tests should be collected and characterized. For those devices where this may be

an issue, the FDA has suggested two options based on current literature and methods employed in spinal research studies. These include an

injection study of various-size particles into the spinal cord area of small animals and functional animal models. Because of the limitations of the current testing methods and models, should devices made of new materials and/or those intended to retain motion be tested for local and systemic effects independent of the type of material or the amount of wear debris generated? If you suggest that testing be performed, please describe the testing that you would

file:///C|/Storage/1121orth.txt (265 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

266 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 recommend. For example, discuss the viability and

usefulness of injection animal studies, including the amount and distribution of sizes and shapes of wear debris that should be injected into the animal, which can then predict what may occur clinically for the life of the implant. Second, discuss recommendations and the viability and usefulness of a functional animal model in predicting what may occur clinically for the life of the implant, or discuss any alternatives you may have. Next, I would like to get to the clinical issues and some questions that we would have for the panel. For spinal assemblies not intended to fuse motion segments, as I have just delineated, the goals of treatment may be to stabilize the spine, maintain normal or functional motion, or treat disease early in its course to prevent further progression and to conserve motion instead of fusing segments of the spine to alleviate pain and restore function. These types of devices provide challenges in choosing the best methods to evaluate safety and effectiveness.

file:///C|/Storage/1121orth.txt (266 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

267 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Our current spinal guidance describes methods to assure that data collected provide adequate characterization of the safety and effectiveness of devices. A copy of the spinal guidance was provided to all panel members, and as I have stated before, is available to the public on the Web, and these sections will not be repeated here. However, I

would direct your attention to the appropriate patient inclusion and exclusion criteria, the effectiveness evaluations, safety evaluations, and patient and study success criteria. The FDA believes that the populations and goals of treatment may be different for devices that maintain functional motion. Therefore, we

will ask you to please discuss study designs which may be better-suited to evaluate nonfusion spinal devices. In your discussion, we would like you to

please comment on enrollment criteria, patient populations, controls, success criteria, and goals of the study, each of these that would be suitable for these types of nonfusion spinal devices. Devices intended to stabilize the spine yet retain functional motion are expected to have an upper limit of motion beyond which one would

file:///C|/Storage/1121orth.txt (267 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

268 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Buch. We are now going to begin the pane lead reviews, and I'll ask Dr. Kirkpatrick to start with his preclinical and clinical reviews, and then we'll ask Dr. Doull to follow on with the toxicology review. Dr. Kirkpatrick? Panel Reviews DR. KIRKPATRICK: Thank you. consider the device to be unstable and a lower one below which one would consider the device to have inadequate motion or possibly even consider the segment to be fused. Therefore, we are going to ask you to please discuss the amount of motion and on what scale to define a patient as a functional and clinical success--for example, a clinically significant improvement in the condition for each of cervical, thoracic and lumbar levels for nonfusion spinal devices. Those are all the questions we are putting before you. DR. YASZEMSKI: Thanks very much, Dr.

My distinguished colleagues on the panel, our FDA friends have invited the guy from Alabama

file:///C|/Storage/1121orth.txt (268 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

269 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 to address the preclinical and clinical issues, so I need to paraphrase the questions to better understand them. In essence, we are being asked to recommend mechanical testing of unknown devices when there is no current validated or consensus method to test them. We are being asked to recommend test methods for devices for a wide range of designs and intended use with no specifics known to us yet. We are asked to recommend toxicology and biocompatibility for unknown materials and debris with no validated test methods. We are asked to recommend clinical evaluations where the indications, intended use, controls, and safety concerns are not specific. Now, if that's not enough for us, I'm worried about the time that we have. At any rate,

I would like to also give a brief comment on some of the concerns that were raised by industry and independent representatives from the public, and that is from the standpoint of the panel, my discussion is going to be what we would like to see, not necessarily what we would accept. In

other words, we always work from compromise, so

file:///C|/Storage/1121orth.txt (269 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

270 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 from the standpoint of being an incubator for testing, yes, we will ask for the companies to satisfy the burden of some of the questions we have with regard to safety and effectiveness, and there may not be current test methods to do so. We also may recommend that the extensive nature of the tests may be pretty high, but that is where compromise comes in in working with the FDA reviewer panel. With regard to preclinical questions, I believe that these devices require mechanical characterization; they should be loaded in all intended and anticipated modes of load and motion to mechanical failure to characterize what amount of load will bring a failure. They should have

durability testing at a physiologic load and motion with anticipated length of service--I am concerned that the 10 million cycles currently recommended is a little bit low if we think of a 40-year-old, like my neighbor, who rose every morning, 5 days a week, and rode a couple of miles, that is a lot of load, a lot of motion, and a lot of repetition of that motion,a nd I don't think 10 million cycles would represent a very long period of time for him if we are looking at a 40-year-old with hopefully a 10-

file:///C|/Storage/1121orth.txt (270 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

271 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 to 15-year life span of the disc replacement, for example, similar to what a joint replacement would be expected to do. I do think that physiologic load needs to be considered very seriously. Physiology for a

grandmother as far as loading her lumbar disc is very different from the physiology of my neighbor loading his lumbar disc. In addition, I think that mechanical characterization should include device changes after durability testing, and that includes looking at the wear, the characterization of debris, plastic deformation, geometry changes, and any mechanical changes of polymeric or other type materials that we may not be reviewing yet. As far as potential tests, I agree that migration and expulsion are important to consider with nuclear replacements. I think that static and

dynamic shear testing may also be important, depending on the design and intended use of the implant. Creep and stress relaxation for viscal I think we

elastic designs are important as well.

all know that when we stand, our disc spaces shrink during the day. And I also think that evaluation

of bone implant interface may be required,

file:///C|/Storage/1121orth.txt (271 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

272 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 depending on the specific design of the implant. Overall, I think that each of these devices that is intended to preserve motion, we should define those motion limits, and I think the best way to do that in current literature is stability testing. We should compare to a quote

"normal" and a quote "expected" based upon the device. We should characterize the neutral zone

and elastic zone of these devices and characterize the failure at the extremes of motion for these devices, as many times, a patient may be in an accept and have to extend more or flex more than they would do under physiologic loading. With regard to new materials, I think biocompatibility and toxicology are important, and I will defer that to our other panel members. We do need to characterize any corrosion, wear, or biologic response to such debris, and I think shelf life and in vivo degradation are important in some of the polymeric devices and materials that may be coming down the pike. Nuclear replacements, I think they require mechanical characterization as well, and I would include creep and stress-relaxation, expulsion testing, stability testing. If simpler tests can

file:///C|/Storage/1121orth.txt (272 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

273 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 be validated, I think it is appropriate to use them. I think that the insertion site repair of disc nuclear replacements should also be tested. In other words, to get something into the disc site, into the nucleus, you have to go through the annulus. Whether they are using the place where

the disc herniated already or if they are doing it in a degenerative condition, there may be some type of repair of the annulus, and that needs to be tested as well. We also need to look again at degradation and deterioration of properties over time if they are nonmetallic implants. With regard to particulate debris, it is unknown for the spine. I think we could follow the

total joint arthroplasty lead, and perhaps we could add somebody with that expertise to the panel who has a particular interest in particulate debris and testing. Animal models do appear to provide us with the best guess at present as to the effect of these particulates. Particulates in joints may differ Unless we are using a

from the non-joint sites.

facette joint replacement, the spine may react

file:///C|/Storage/1121orth.txt (273 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

274 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 completely differently than the total joint arthroplasty. guess. As far as debris around the dura and whether there is a neurologic effect, we don't know that, either, and I think it warrants some investigation. Perhaps an animal model would be But right now, it again is a best

the most appropriate first guess at that. Basically, sa far as particulate debris in the spine, there are no established methods. Moving on to clinical issues, with regard to inclusion and exclusion, I think the criteria placed for fusion are appropriate. However, as we

noted in the earlier presentation today, I would like to see the indications refined and specific as opposed to fairly broad. We talked about--for

those not available for most of the day--we had four different fracture classifications of severity of the fracture combined with four different open wound classifications, and as such, that creates a very complex and heterogeneous dataset. I would urge investigators to consider making it as refined as possible for disc replacements or nonfusion devices. I would also

suggest that we consider the philosophy of the

file:///C|/Storage/1121orth.txt (274 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

275 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 implant in picking the populations. When I say

that, there are some issues with regard to disc replacement, such as the adjacent segment degeneration. That has been proposed as a very

serious problem, and that is why we are developing these disc replacements. I think that that is not

exactly an easy thing to say, that the degeneration occurs at the adjacent segment just because of the fusion. It may actually be happening because of

the natural history of the spine. Biomechanics has shown that your load on the adjacent segment is similar to the load at the affected segment, and other segments farther away from that adjacent segment are also subjected to the same increases in load and motion, biomechanically. So the question is is that really

making a big difference. As far as primary measures on what to look at for clinical success, radiographic measures are challenging at best. Validated pain measures and

specific function measures I think are appropriate. The challenge of course remains as to how much change do we need to see in those to call it a success. Specifically on radiographic criteria, the

file:///C|/Storage/1121orth.txt (275 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

276 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 range of motion--again, flexion-extension views give us very poor reproducibility and are very challenging to interpret. It may involve having to

add extra implants to the patient such as beads to be able to actually do good motion measurements. I think the absence of bridging bone is a reasonable thing; if we are trying to preserve motion, we should not have bridging bone result. The bone implant interface needs to be evaluated in those that are supposed to be fixated. not have radiolucencies develop. The implant position should be evaluated radiographically. migration. The implant geometry, to detect where; the disc height, again to detect where and add to the implant position issue; and adjacent segment--if we are agreeing with the philosophy that we are sparing the adjacent segment from degeneration, we can't see progression of the adjacent segment degeneration in the clinical trial. As far as safety, I think some key things clinically we need to evaluate include what are the revision options for these implants. Is That would detect subsidence and We should

replacement of the device appropriate, or is the

file:///C|/Storage/1121orth.txt (276 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

277 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 only option fusion after removal of a failed device? Are there other procedures that could be

done following this? If fusion after failure is the intended treatment once an implant fails, is that success equal to a primary fusion? That should be included

in our evaluation of the success of the implant. And of course, we have also heard about neurologic effects. I personally believe that all

neurologic effects should be reported, and the panel should be given the opportunity to determine whether they feel it is caused by the device itself or whether it is a surgical complication. Patient success measures--I think specific radiographic criteria should be developed. going to be implant-dependent. Specific It is

improvement in pain and function--I don't know how much improvement we should look for. 50 percent improvement? Seventy-five percent? to pin down. From the reading of the FDA regulations, it sounds like if we have 5 percent improvement and can demonstrate it, that is enough to satisfy the efficacy. I am not so sure, clinically, that I Should it be

Twenty-five percent? It is a very difficult thing

file:///C|/Storage/1121orth.txt (277 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

278 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 look at? benefit? would put somebody through a major operation for a 5 percent improvement. And then, what risk is acceptable for what That's what the bottom line is on that

discussion. On study success, what are the controls going to be? Should we look at nonoperative

controls to demonstrate an improvement over the natural history? That would seem to be a logical

measure, especially in a nonfusion device that is preserving motion. Should we look at traditional operative management to compare pain and function scales? obviously cannot compare radiographs there, but pain and function may be appropriate. And then, what sort of follow-up should we Is 2 years adequate? Personally, I don't We

feel that 2 years is adequate for some of these studies. I think 5 to 10 would be better if we are

trying to look at a longevity-producing implant. And then, how much improvement relative to other treatments, as I mentioned before, is a very difficult question to answer. In general as a summary, I would suggest that nonfusion devices require more extensive and

file:///C|/Storage/1121orth.txt (278 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

279 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 complex evaluations, both preclinical and clinical; longer in vitro testing; different endpoints; and environmental exposure should be evaluated in the in vitro testing, as well as stability testing. Debris and particulate matter should be characterized and determined whether it is toxic or induces an immune reaction. evaluated as well. Degradation should be

And we should have extensive

clinical data to be able to make our judgments. In short, like my teenage daughter around Christmas time, she'll ask for the world--but as far as the others in the household are concerned, we will adjust to some specific requests, look for rational and justifiable compromise, and then work together as a team to serve our patients. Thank you very much. DR. YASZEMSKI: Thanks, Dr. Kirkpatrick.

Dr. Doull, can we ask you to present your preclinical toxicology review? DR. DOULL: Well, I'm pleased that Dr.

Kirkpatrick included the mechanical testing as part of the efficacy. That means I don't have to deal

with that since I am going to deal with safety. The safety issue, as one of the commenters mentioned, boils down to whether the wear and tear

file:///C|/Storage/1121orth.txt (279 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

280 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 debris has a chemical or a toxic effect or whether it has a physical adverse effect, and how one sorts that out. There is no real basis that I know of for concluding that simply because you reduce the size of the material, that you induce some special kind of toxicity. It is a question of dose, and as you

reduce the dose, you reduce the toxicity. Whether the physical state of the material can in fact induce some special kind of toxicity I think is something that needs to be explored, and certainly that is one area where animal testing would help. I think we have a strong background in that area, however, and that comes from our studies with solid-state tumor agenesis. We know a lot

about elastomers, silicone elastomers, asbestos, metals, and so on, about solid-state tumor agenesis, and I think that gives us a good head start on how we might approach the area here. By and large with solid-state tumor agenesis, as one reduces the size of the particle, you reduce the propensity of that material to produce tumor, and hopefully, one would have similar effects here.

file:///C|/Storage/1121orth.txt (280 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

281 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Doull. What I would like to do now if it would be acceptable to FDA, Dr. Witten, is to put the questions up in the order that Dr. Buch presented them to us, and let's discuss them one at a time. Thanks, Mr. Melkerson. The first slide discussed the preclinical issues, and I think there are probably a few questions embedded in here, so we'll have to separate them out a little bit, and I am open to discussion of my method of separation if somebody thinks they can do it better. The first bullet point, perhaps we can take as a question: "Please comment on the The specific kind of neurological testing that one might do to look for particle size effects I think is a challenge, and I am not aware of really good animal models in fact that would help us with that, and I think clearly, we are going to have to develop some good models in order to get a handle on this. And I'll leave the clinical areas to my colleagues. Discussion of FDA Questions to Panel DR. YASZEMSKI: Thanks very much, Dr.

file:///C|/Storage/1121orth.txt (281 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

282 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 flavors. currently recommended preclinical mechanical, debris, or wear testing to evaluate new materials, device properties integrity and wear debris for both fusion and nonfusion devices." We all have the list that Dr. Buch gave us of three slides of current recommendations. Would anybody care to start, and then we'll go around and discuss the current recommendations? DR. NAIDU: Sure, I can start. Dr. Naidu.

DR. YASZEMSKI: DR. NAIDU:

The current recommendations

for fusion devices and nonfusion devices have been specified by Barbara pretty well. The problem is

that major area where it lacks is nonfusion devices. The nonfusion devices can come in many They could be thermoplastic elastomers,

they could be lightly crosslinked elastomers, they could be highly densely crosslinked elastomers, they could be crystalline polymers, they could be amorphous polymers. They could be many things.

The problem is that depending on the material that you choose, they will age, and obviously, what you can't do is just base yourself

file:///C|/Storage/1121orth.txt (282 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

283 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 topic. on all these plain, old, simple mechanical testing. You are going to have to do some sort of viscal elastic testing, whether it involves generating enthol-B [phonetic[ curves and studying crystallinity, whether you are going to use DMA or thermal analysis, depending on the mode of loading that you want to do, and depending on whether you want to use torsion pendulum tests. There are many varieties of tests. sit down and discuss this issue forever here. not exactly sure that we can reach a consensus, because these materials are so variable. Like I We can I am

said, it ranges all the way from thermoplastic elastomers to crosslinked to non-crosslinked to crystalline to semi-crystalline to amorphous. So I think that this is a very involved I'm not sure that we are going to reach a

consensus at this point within the next hour or two, and therefore, I think that all these issues should be addressed individually with the people who are actually producing these devices. When you start talking about things like peak, you have got to start thinking about whether it is pure peak, neat peak, is it a composite peak, is it peak-on-peak. There are many things. So

file:///C|/Storage/1121orth.txt (283 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

284 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 suffice it to say that there is a lot to be discussed. DR. YASZEMSKI: Thank you. Dr. Finnegan,

Let's come around this way.

do you have anything that you would like to add to Dr. Naidu's comments? DR. FINNEGAN: The only two things I would

add are that I would like to reinforce the fact that the longevity issue and the properties after a length of time on the shelf do need to be tested, because the total joint supported demonstrated that that is a problem. The other issue I would bring up is the method of sterilization, because again, if it is gamma radiation, and you have those kinds of polymers, you will also have an effect. also needs to be tested. DR. YASZEMSKI: Dr. Kirkpatrick? DR. KIRKPATRICK: I think conceptually, I would Thanks. So that

most of the things have been addressed. suggest once again that wear testing and characterization of wear debris would be appropriate. DR. YASZEMSKI: Thank you.

file:///C|/Storage/1121orth.txt (284 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

285 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 comments. Dr. Doull? DR. DOULL: Nothing more. Thanks.

DR. YASZEMSKI: Dr. Aboulafia? DR. ABOULAFIA: DR. YASZEMSKI: Dr. Schmidt? DR. SCHMIDT:

Nothing specific to add. Thanks.

I don't have a lot to add.

I am not a spine surgeon and not too up-to-date on a lot of the testing methods we're talking about, but I do know a little bit about wear particles in total joints, and I know that a combination of wear debris and motion has been shown to be a bad combination. So I think that that is something

that needs to be studied. I think we need to work with the regulatory agencies to define standards, and it is going to need to be a long, involved process with all the different players here to co me up with something. DR. YASZEMSKI: Dr. Larntz? DR. LARNTZ: Just a couple of specific Thank you.

In the current guidelines, there are I

some numbers, like "N" equals 6 for SN curve.

file:///C|/Storage/1121orth.txt (285 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

286 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 have actually done a little bit of work with those, and boy, I can't imagine what we find out with N equals 6. I am a statistician--I am not an engineer, where everything is perfect and where everything works perfectly--but then, when I come in and see engineers do more than 6, they say, "Wow, I didn't expect that to break out there." So we have to be very careful with specific numbers, very careful, and I think we have pretty inadequate characterization from some of the standards that are set by so-called standards agencies--excuse me for being a heretic on that. And as far as things like cycles, actually, I decided to do a little calculation. I

worry about my knees because I am a bicyclist, and I just figured out that in 2 years, I do 10 million cycles on my bicycle. I have nothing else to do,

so that's nice, but I would like my knees to last a long time, and if I have to do something else, I would like those other things to last a long time. So we have to be very careful when we set upper limits. comment. I agree with Dr. Kirkpatrick's

Many of the bounds that we are talking

about are low--I'm sorry, I'm no speaking as a

file:///C|/Storage/1121orth.txt (286 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

287 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 statistician now; I am speaking as a potential consumer, okay--but as a statistician, I remember doing this stuff with wire or raw material, and the engineer said, "It never breaks, it never breaks," and I said, "Let's just try it." He put the thing

over the weekend, and he was surprised to come back and find half the specimens broken that he thought were going to be perfect. So things to happen, and we have to be very careful. As far as other materials, obviously, innovation is important, and testing is critical, and people have to have incredible intimate knowledge of these materials to develop the appropriate tests. And I do believe there are

going to be a lot of very specific, very different kinds of tests developed, and I do think we have to not underestimate--I'm going to put a plug in for my own field--the variability in the results of tests that we often do. Thank you. DR. YASZEMSKI: Thanks, Dr. Larntz.

Dr. Witten, as we go over this round of discussion, I'm going to note that although the first bullet of the first question is what I

file:///C|/Storage/1121orth.txt (287 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

288 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 started out with, I think that the discussion extended into additional testing, so if it is acceptable to FDA, I think I'll summarize on the first slide both current and new testing. I think we probably will lend more heterogeneity to the slide than currently exists, because there really is no consensus as to where this is all going to go. And I think what I am

hearing from the panel members is that that is probably okay right now, because these are new devices in a new field, and we can perhaps learn some things from problems that have occurred with existing devices from other fields. They include that there will be a variety of materials as there were for joints when joints came up, and some commonalities were that wear debris has been a problem, so that we should probably look at wear debris in these new materials. The sterilization effect will be an issue, especially in light of the fact that some of these new materials perhaps will be combination products in which the effect of sterilization on biologics associated with devices will really have to be taken into account.

file:///C|/Storage/1121orth.txt (288 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

289 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Longevity and shelf life is important, because we will have to know how long in advance and what the demand is for these, and we'll have to get some idea for how long they will last on the shelf before we put them in and maintain those properties that we are talking about testing. Several references have been made to the voluntary standards organizations, the ASTM and the ISO, and perhaps as per Dr. Naidu's comment, we can learn from the testing that exists under ASTM, usually Section (b), for polymeric materials used for mostly nonbiologics, non-human devices, and apply some of those, the thermal tests and the dynamic mechanical analysis that he mentioned, to the various polymeric devices that will occur here. And I guess as a summary, what I have heard from everybody is that each different material may need a different set of tests, and all the people who are part of this--the FDA, the investigators, the companies, and the patients--are going to have to work together and work this out. It's not a very clear answer, but have we addressed the issue to your satisfaction? MS. WITTEN: yes. Thank you.

DR. YASZEMSKI:

Thank you.

file:///C|/Storage/1121orth.txt (289 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

290 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 We're going to move on to the next question, then. Dr. Buch, thank you.

Question 2 concerns preclinical issues. "For those incorporating new materials--polymers, composites--or designs, both fusion and nonfusion, the FDA recommends that manufactures perform wear simulations and fatigue tests to evaluate the potential for the device to generate wear debris. The FDA believes the wear

debris generated from these tests should be collected and characterized. For those devices

where this may be an issue, the FDA suggests two options--injection study of various-sized particles into the spinal cord area of small animals, and functional animal models." If I may start, I think we have already discussed in the last question that we believe wear debris testing is important, so we'll say that that should be included. I would ask the panel members please to comment now on the type of wear debris testing. The two that are mentioned here, spinal cord particulate and injection, and what would functional animal models be--would anyone like to comment on the animal models?

file:///C|/Storage/1121orth.txt (290 of 309) [12/2/02 8:59:37 AM]

file:///C|/Storage/1121orth.txt

291 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Dr. Kirkpatrick? DR. KIRKPATRICK: My comment on the animal

model was this talks about into the spinal cord area. I don't think that's very specific. I would

suggest that the particles or particulates that are produced through wear testing of the device that is made should be placed in the area of intended use adjacent to the dura. In other words, if we are

looking at a ligament replacement that is going to be posterior for lumbar spine only application, it should just be posterior on the dura. If we are

looking at a disc replacement that produces wear debris, it should be placed anterior to the dura. If it is for cervical disc, it should be in the cervical spine of whatever the animal is, so we could represent whether there is a root effect or a cord effect or both. DR. YASZEMSKI: Ms. Maher? Ms. MAHER: My only comment on this is Thank you.

that if we are going to be generating wear testing on wear debris, the wear testing should be under physiological loads, not under the maximum load. DR. YASZEMSKI: Dr. Doull? Thanks.

file:///C|/Storage/1121orth.txt (291 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

292 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DR. DOULL: Let me add to that that the focus should be on the physical averse effects rather than the chemical adverse effects, which can be tested in the conventional approaches. If we

are developing new technology, it ought to be for physical adverse effects. DR. YASZEMSKI: Dr. Kirkpatrick? DR. KIRKPATRICK: If I could just add two Thank you.

caveats to what was just said, one is the physiologic loading of the most vigorous patient anticipated for the device would be appropriate as opposed to maximal loading. We also should look at

the immune response to the particulate debris, because that is the main problem in total joints. DR. YASZEMSKI: Dr. Larntz? DR. LARNTZ: If I can just comment a Thank you.

little bit further, the testing at physiological load or even at maximal load is if you do it right and think about it is near impossible. Accelerated

testing is absolutely necessary, and modeling of those accelerated tests is the key to making sure that they apply to the physiological load or the maximal physiological load. Testing at

file:///C|/Storage/1121orth.txt (292 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

293 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 physiological load means that you will never get your test done. You don't want that. You want to

be able to get your test done, you want to be able to get it done quickly, and accelerated testing is the way to do that. DR. YASZEMSKI: Dr. Naidu? DR. NAIDU: The particle size is also Thank you.

important, and just to reflect Dr. Larntz' suggestion that accelerated testing is needed, along the same lines, a barrage of particles, you have to put those particles at the site, like Dr. Kirkpatrick suggested, at the site of implantation, and reaction to such particles should be studied, because as suggested, before accelerated testing, these debris particles are going to be small, and we don't know what the response is going to be. It is the same thing in total joint literature with polyethylene particles--you really can't predict, but there is definitely a correlation with size. So you have to do a

dose-response type study, implanting such particles in intended areas of use. So I do concur with all the other panel members as far as that goes.

file:///C|/Storage/1121orth.txt (293 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

294 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Thank you. DR. YASZEMSKI: Dr. Kirkpatrick? DR. KIRKPATRICK: on particle size. Just one further comment Thank you.

I don't think that the

manufacturer should be required to produce particles of a size that is not produced by wear under physiologic conditions. In other words, I

don't think we should arbitrarily ask for certain size of particles. It should be justified by the

particles produced by that device. DR. YASZEMSKI: Ms. Maher? MS. MAHER: I also want to emphasize that we need to be careful not to place the burden so high on the manufacturers that they decide not to develop in this area, because while wear debris in joints is a problem, prior to having total joint, it was a bigger problem. So I think we need to Thank you.

balance the risk and benefit of what we are asking for. I understand all the science that we are bringing out in everything we have learned, but if people say, fine, we aren't going to develop any artificial discs because the burden is too high,

file:///C|/Storage/1121orth.txt (294 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

295 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 then I think that as a whole, the public has lost out. DR. YASZEMSKI: DR. NAIDU: comment on that? DR. YASZEMSKI: DR. NAIDU: Dr. Naidu, go ahead. Thank you.

Can I just make an additional

I understand that the burden

may be high, but we are also talking about high stakes. These are neurological tissues--the spine.

Total joint, you have the bone, and you have to consider that, too. Thank you. DR. YASZEMSKI: Dr. Aboulafia? DR. ABOULAFIA: DR. YASZEMSKI: I have nothing to add. Dr. Doull, may I ask you Thank you.

with respect to toxicology--the FDA has asked us about alternative animal models, and they mentioned functional small animal models. From a toxicology

perspective, are studies performed in certain types of animals directly expandable to humans? What do we need to know from a toxicology perspective about the type of animal we recommend to FDA and to industry? DR. DOULL: There are two cardinal rules

file:///C|/Storage/1121orth.txt (295 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

296 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 in toxicology. The first is the dose makes the

poison, and the second is results in one species are predictive for another species when properly qualified. So all we have to do to use animal studies is properly qualify them. In terms of the neurological testing, Food and Drug, EPA, and a number of agencies have put together some guidelines for testing for neurological effects, and for new substances, new polymers, new elastomers and so on, clearly, those are the kinds of things one ought to do first, because that is going to characterize the chemical toxicity of the material, and we can do that before we do any wear and tear debris studies up front, in a sense, and much simpler. DR. YASZEMSKI: Thank you.

Would anybody else like to make a comment on this issue, which will include both of these slides--thank you, Mr. Melkerson, for putting them up. [No response.] DR. YASZEMSKI: Seeing none, Dr. Witten,

we have had a discussion that I think has tended toward suggesting we balance the risk and benefit

file:///C|/Storage/1121orth.txt (296 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

297 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Dr. Buch. "FDA believes that the populations and goals of treatment may be different for devices to patients and to industry's interest in pursuing this. Particles, once generated, will probably

show a spectrum of particle size distribution, and the recommendation has been that we ask industry only to test those particle sizes that will occur in a specific device. Dr. Doull has let us know that if someone recommends a certain device that it would be prudent to test the material first, before putting it in a device, to assess its chemical toxicity and then to check its physical effects after it has been put in the device. We should use physiologic

loads and accelerated testing, and again, be flexible given the new and novel nature of these devices. We have also heard that we need to be concerned about the immune response to these particles. Have we discussed this adequately? MS. WITTEN: Yes. Thank you.

DR. YASZEMSKI:

Thanks, Dr. Witten.

Let's go on to clinical issues, please,

file:///C|/Storage/1121orth.txt (297 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

298 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 that maintain functional motion. Therefore, we are

asked to discuss study designs which may be better-suited to evaluate nonfusion spinal devices." We are asked to comment on enrollment criteria, patient populations, controls, success criteria, and study goals that would be suitable for nonfusion spinal devices. These are devices, on the next slide, "intended to stabilize the spine yet retain functional motion. They are expected to have an

upper limit of motion beyond which one would consider it to be unstable, a lower limit beyond which one would consider it to have inadequate motion or possibly consider the segment to be fused." We are also asked to discuss the amount of motion and on what scale to define a patient as a functional and clinical success. Comments from the panel members on this--nonfusion devices, clinical studies, exclusion/inclusion criteria, ranges of motion. Dr. Kirkpatrick? DR. KIRKPATRICK: DR. ABOULAFIA: Go ahead. I was going to sort of

file:///C|/Storage/1121orth.txt (298 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

299 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 echo what Dr. Kirkpatrick said earlier. One of the ideas behind using these nonfusion devices is that you may alter degenerative changes at segment. adjacent

And hopefully, any study design would

compare possibly a level and similar cohort of patients, groups who have been fused at a particular level, to see if there is a difference in the natural history at the adjacent segment by not fusing the treatment level. DR. YASZEMSKI: DR. KIRKPATRICK: Dr. Kirkpatrick? Just again reiterating

what I said earlier about inclusion criteria, I think what is there is fine; however, the investigators or the sponsors should seriously consider what indications they want applied to their device. I don't think it would be

appropriate for them to consider studying a nuclear replacement, for example, in a Grade II spondylolisthesis unless they felt it was distracting and reducing that particular disorder. Otherwise, there are going to be too many conflicting issues. Similarly, for a disc replacement, they need to consider whether they are looking at deformity to correct. I think that would unduly

file:///C|/Storage/1121orth.txt (299 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

300 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 follow up. compound their analysis and may alter their results. better. DR. YASZEMSKI: MS. RUE: Ms. Rue? So the cleaner the indication, the

From a consumer's perspective, I

think that success criteria should definitely include improvement of pain and functional capacity and not just no digression in it, and to what degree will have to be determined. DR. YASZEMSKI: Dr. Larntz? DR. LARNTZ: Just a couple comments to Thank you.

I think I hear something about I think

homogeneity of population for indication.

devices are typically used across a broad range of patients. I think we have to make sure that we are

not afraid of testing them across a broad range of patients. I hear that a lot. People are worried

about--I have heard that recently--sites being different. I think that's actually an advantage.

We want to find out if things work across a range, because they are used across a range. Even if it

is past recommended, labeled with an indication, we have got to recognize that that indication may not be totally limiting.

file:///C|/Storage/1121orth.txt (300 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

301 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 With respect to patient success, I think we have had a lot of trouble at different times defining "patient success," and I think that should be--I understand the reason for that--I think that patient success, of course, is a multivariate--multivariate--construct, for want of a better term--heaven forbid, I thought I'd never use that word. At any rate, what we need to do is think hard about comparing patients, which patient is doing better than another. I think that's a great

way to do things, and I have argued that for many years, that what we need to do is try to get scales that compare patients and who does better than another one, and I would argue that that kind of thinking has been missing. What we are doing now

is success/failure, and our successes have been a broad range of patients, and our failures have been an even broader range of patients. So we are

lumping them into a dichotomous--we are losing tons of information by doing that. We are making sample I think it

sizes larger than necessary to do that.

is very important when we do comparative studies, which I think most of these should be comparative, although sometimes we have to think about--the

file:///C|/Storage/1121orth.txt (301 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

302 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 question up there about controls is difficult, right, because in some of these things, we aren't sure what the alternative is; particularly for some new therapies, there may not be much of an alternative, so that is interesting to think about. But I think it is important to think about ways of doing these so that we can reduce sample size to get the information out that we need. think the old dichotomy of looking at patient success/failure and just doing that is very dangerous. doing that. DR. YASZEMSKI: Other comments? Dr. Kirkpatrick? DR. KIRKPATRICK: Getting back to the Thanks, Dr. Larntz. I think we lose tons of information by I

second slide on clinical issues which talked about the different motions which would be appropriate, basically, I don't think we should set a level of motion that is required for this, because if we have a patient who gets a nonfusion device, and they end up doing, functionally and painwise, much better than they would have without it, I think it is reasonable that it is an efficacious device much like our colleague just mentioned a few moments

file:///C|/Storage/1121orth.txt (302 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

303 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 ago. If we put arbitrary definitions on must have so much motion at that segment to be considered a success, we are not going to be able to do it. If you want to have a guide for how much

motion is what is there, or should be there, I think the White and Punjabi [phonetic] data is pretty well-accepted as being fairly close to what we would expect to be a physiologic range of motion, and whether you can duplicate that with a device remains to be seen. And--there was one other thing that just escapes my mind. DR. YASZEMSKI: As you are thinking about

it, I'm going to ask some others to think about two additional issues. Please offer commentary if you

have it on the devices that are primarily nonfusion devices in the anterior portion of the spine--that is, either a disc replacement or a nucleus replacement--would there be any consideration regarding the number of levels for an ideal patient? We are asked about enrollment criteria.

Would an ideal patient for this study be one, for example, with single-level disease, or would it be just as appropriate to consider persons with

file:///C|/Storage/1121orth.txt (303 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

304 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 multilevel disease? Comments? Dr. Kirkpatrick? DR. KIRKPATRICK: to my earlier comment. I think that gets back

If you are looking at using

a disc replacement because it spares the adjacent segment, then multiple levels are going to eliminate a rational analysis of that. It depends

on their rationale for recommending the device. DR. YASZEMSKI: DR. KIRKPATRICK: So again, study-dependent. However, if we are

looking at being able to serve patients with multilevel degenerative disease or multilevel pathology of similar circumstances, I think one or two levels would be reasonable, although again, it would take away from the cleanliness of the data to be able to analyze it. I did remember the other issue I wanted to bring up, and that is measurement of motion on flexion-extension views, as I mentioned in my earlier presentation, is very difficult to do as far as a standardized method. It is probably going

to require the addition of beads in there so we can do stereophotogrametry techniques, basically, and look at whether the motion is real or parallax.

file:///C|/Storage/1121orth.txt (304 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

305 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 So I think we are going to have to consider whether that is something that is going to be critical or not, because that does add some added implant risk and surgical morbidity. DR. YASZEMSKI: Thank you.

Another thing I'd like to ask the panel members for comment on for those devices, i.e., disc replacement and nucleus replacement, that are anterior, is what if any information should we know and/or ask about the status of arthritis in the facette joints at the level we are putting the anterior device in when evaluating the anterior device? Comments? Dr. Aboulafia? DR. ABOULAFIA: I wonder if that won't be

addressed by the indications for the product. DR. YASZEMSKI: DR. KIRKPATRICK: Dr. Kirkpatrick? I would suggest we

ignore the condition of the facettes from the standpoint of the testing, because if the companies find that if they have significant facette pathology, and the patients continue with significant pain and limited function, they are going to eliminate that indication on their own.

file:///C|/Storage/1121orth.txt (305 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

306 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DR. YASZEMSKI: Thank you.

Dr. Witten, we have talked about clinical issues, the inclusion and exclusion criteria, and I think the discussion indicated that we should study a wide range of disease diagnoses, that the indications for replacement versus fusion would be an important aspect of any study design. We would

need to assess the adjacent segments and decide as one of the outcome variables whether nonfusion interventions would have any effect on adjacent segment progression of degeneration and arthritis. Success criteria should include pain and functional improvements as reported by the patient, and with respect to the ranges of motion question asked, that the White and Punjabi data regarding neutral zone, physiologic zone, and trauma zone would be good guidelines, but that range of motion in and of itself should not exclude--range of motion considerations should not exclude projects that otherwise have merit and could benefit patients. Have we discussed this adequately from FDA's perspective? MS. WITTEN: You have, except that I have one little follow-on question.

file:///C|/Storage/1121orth.txt (306 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

307 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 DR. YASZEMSKI: MS. WITTEN: Please go ahead.

Basically, you have discussed

the range of motion question with respect, I would say, to effectiveness, that is, if there is limited range, the goal of the range shouldn't be considered to be the goal of treatment success. But is there an upper limit beyond which there might be a safety concern--or, perhaps you have answered that by referring to-DR. YASZEMSKI: Yes. I think that

actually, Dr. Kirkpatrick covered that, and I am going to ask him to comment on the White and Punjabi transition between physiologic zone and trauma zone. I think you covered that; would you care to comment again? DR. KIRKPATRICK: I think it will be

addressed partly in the preclinical issues, because they are going to have to characterize how much motion is there. If you find out that it is 50

percent over what the White and Punjabi data is, you are going to question it seriously, and so will we as a panel. If it is within 10 percent of it,

we may not question it, because once it is implanted, the scar tissue may correct for that.

file:///C|/Storage/1121orth.txt (307 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

308 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Clinically, if you are going to measure that, again, if it is going to fail, it is going to demonstrate radiographic failure with either motion of the component, a deformity developing in the spine, that sort of thing--assuming the preclinical stuff looks okay. Does that address what you are wondering? MS. WITTEN: Yes. Thank you.

DR. YASZEMSKI: Dr. Finnegan? DR. FINNEGAN:

As he was discussing the

materials and the stiffness of the material, I am wondering if the concern about a created spinal stenosis by retropulsion of any of the materials and how the stiffness of the material will affect the area of the spinal canal and also the effect on the cord and how much it will react to that. might be something that needs to be added. DR. YASZEMSKI: Mr. Demian? MR. DEMIAN: I would like to thank the Thank you. That

panel for their time and effort and energy today in reviewing the material and for their participation on this panel. At this time, I would like to remind all

file:///C|/Storage/1121orth.txt (308 of 309) [12/2/02 8:59:39 AM]

file:///C|/Storage/1121orth.txt

309 1 2 3 4 5 6 7 8 adjourned. Thank you. [Whereupon, at 4:23 p.m., the proceedings were concluded.] - - panel members that if you want the material destroyed, just leave it in front of you. And the last thing--this meeting is

file:///C|/Storage/1121orth.txt (309 of 309) [12/2/02 8:59:39 AM]