1
1 FOOD AND DRUG ADMINISTRATION
2 CENTER FOR DRUG EVALUATION AND RESEARCH
3
4
5
6 JOINT MEETING OF THE PSYCHOPHARMACOLOGIC DRUGS AND
7 DRUG AND RISK MANAGEMENT ADVISORY COMMITTEES
8 (PDAC and DSaRM)
9
10
11
12
13 Thursday, November 1, 2018
14 8:00 a.m. to 5:01 p.m.
15
16
17
18
19 FDA White Oak Campus
20 Building 31, the Great Room
21 10903 New Hampshire Avenue
22 Silver Spring, Maryland
A Matter of Record (301) 890-4188 2
1 Meeting Roster
2 DESIGNATED FEDERAL OFFICER (Non-Voting)
3 Kalyani Bhatt, BS, MS
4 Division of Advisory Committee and Consultant
5 Management
6 Office of Executive Programs, CDER, FDA
7 PSYCHOPHARMACOLOGIC DRUGS ADVISORY COMMITTEE
8 MEMBERS (Voting)
9
10 Walter S. Dunn, MD, PhD
11 Staff Psychiatrist and Assistant Clinical Professor
12 West Los Angeles Veterans Administration Medical
13 Center
14 University of California Los Angeles
15 Department of Psychiatry
16 Los Angeles, California
17
18
19
20
21
22
A Matter of Record (301) 890-4188 3
1 Jess G. Fiedorowicz, MD, PhD
2 Associate Professor
3 Departments of Psychiatry, Epidemiology and
4 Internal Medicine
5 University of Iowa Carver College of Medicine
6 Iowa City, Iowa
7
8 Satish Iyengar, PhD
9 Chair and Professor of Statistics
10 Department of Statistics
11 University of Pittsburgh
12 Pittsburgh, Pennsylvania
13
14 Felipe A. Jain, MD
15 Assistant Clinical Professor of Psychiatry
16 Department of Psychiatry
17 University of California, San Francisco
18 401 Parnassus Avenue, Langley Porter
19 San Francisco, California
20
21
22
A Matter of Record (301) 890-4188 4
1 Rajesh Narendran, MD
2 (Chairperson)
3 Attending Psychiatrist
4 Re:solve Crisis Network
5 Western Psychiatric Institute and Clinics
6 Associate Professor in Radiology and Psychiatry
7 Psychiatric Molecular Imaging Program
8 University of Pittsburgh
9 Pittsburgh, Pennsylvania
10
11 Kim O. Wictzak
12 (Consumer Representative)
13 Co-Founder, Executive Director
14 Woodymatters
15 Minneapolis, Minnesota
16
17
18
19
20
21
22
A Matter of Record (301) 890-4188 5
1 PSYCHOPHARMACOLOGIC DRUGS ADVISORY COMMITTEE MEMBER
2 (Non-Voting)
3 Robert R. Conley, MD
4 (Industry Representative)
5 Global Development Leader
6 Pain and Core Therapeutic
7 Team and Distinguished Scholar
8 Eli Lilly and Company
9 Lilly Corporate Center
10 Indianapolis, Indiana
11
12 DRUG SAFETY AND RISK MANAGEMENT ADVISORY COMMITTEE
13 MEMBERS (Voting)
14 Kelly Besco, PharmD, FISMP, CPPS
15 Health-System Medication Safety Officer
16 OhioHealth Pharmacy Services
17 Dublin, Ohio
18
19
20
21
22
A Matter of Record (301) 890-4188 6
1 Marie R. Griffin, MD, MPH
2 Professor, Health Policy and Medicine
3 Director, Vanderbilt MPH Program
4 Department of Health Policy
5 Vanderbilt University Medical Center
6 Nashville, Tennessee
7
8 Laurel A. Habel, MPH, PhD
9 Associate Director, Cancer Research
10 Division of Research
11 Kaiser Permanente Northern California
12 Oakland, California
13
14 Sonia Hernandez-Diaz, MD, MPH, DrPH
15 Professor of Epidemiology
16 Department of Epidemiology
17 Harvard T.H. Chan School of Public Health
18 Boston, Massachusetts
19
20
21
22
A Matter of Record (301) 890-4188 7
1 Martin Kulldorff, PhD
2 Professor of Medicine and Biostatistician
3 Division of Pharmacoepidemiology and
4 Pharmacoeconomics
5 Department of Medicine
6 Harvard Medical School and Brigham & Women's
7 Hospital
8 Boston, Massachusetts
9
10 Steven B. Meisel, PharmD
11 System Director of Patient Safety
12 Fairview Health Services
13 Minneapolis, Minnesota
14
15 Anne-Michelle Ruha, MD, FACMT
16 Director, Medical Toxicology Fellowship Program
17 Department of Medical Toxicology
18 Banner University Medical Center
19 Clinical Associate Professor of Emergency Medicine
20 University of Arizona College of Medicine
21 Phoenix, Arizona
22
A Matter of Record (301) 890-4188 8
1 Terri L. Warholak, PhD, RPh, CPHQ, FAPhA
2 Professor and Assistant Dean
3 Academic Affairs and Assessment
4 College of Pharmacy
5 University of Arizona
6 Tucson, Arizona
7
8 TEMPORARY MEMBERS (Voting)
9 Jane B. Acri, PhD
10 Chief, Medication Discovery & Toxicology
11 Branch
12 Division of Therapeutics and Medical
13 Consequences
14 National Institute on Drug Abuse
15 National Institutes of Health (NIH)
16 Bethesda, Maryland
17
18
19
20
21
22
A Matter of Record (301) 890-4188 9
1 David Cella, PhD
2 Professor Chairperson
3 Medical Social Sciences
4 Northwestern University Feinberg School of
5 Medicine
6 Chicago, Illinois
7
8 Stephanie Y. Crawford, PhD, MPH
9 Department of Pharmacy Systems
10 Outcomes and Policy
11 University of Chicago
12 Chicago, Illinois
13
14 Harriet de Wit, PhD
15 Department of Psychiatry and Behavioral
16 Neuroscience
17 The University of Chicago
18 Chicago, Illinois
19
20
21
22
A Matter of Record (301) 890-4188 10
1 Kathryn E. Flynn, PhD
2 Associate Professor of Medicine
3 Division of Hematology and Oncology
4 Senior Scientific Director for Patient-Reported
5 Outcomes
6 Center for International Blood & Marrow Transplant
7 Research
8 Milwaukee, Wisconsin
9
10 Roxanne E. Jensen, PhD
11 Program Director
12 Outcomes Research Branch
13 Health Care Delivery Research Program
14 Division of Cancer Control and Population Sciences
15 National Cancer Institute, NIH
16 Bethesda, Maryland
17
18 Elizabeth Joniak-Grant
19 (Patient Representative)
20 Holly Springs, North Carolina
21
22
A Matter of Record (301) 890-4188 11
1 Brandon D.L. Marshall, PhD
2 Associate Professor
3 Department of Epidemiology
4 Brown University School of Public Health
5 Providence, Rhode Island
6
7 William T. Riley, PhD
8 Director
9 Office of Behavioral and Social Sciences Research
10 NIH
11 Bethesda, Maryland
12
13 FDA PARTICIPANTS (Non-Voting)
14 Robert Temple, MD
15 Deputy Director for Clinical Science
16 CDER FDA
17 Deputy Director (Acting)
18 Office of Drug Evaluation I (ODE I)
19 Office of New Drugs (OND), CDER, FDA
20
21
22
A Matter of Record (301) 890-4188 12
1 Ellis Unger, MD
2 Director
3 ODE I, OND, CDER, FDA
4
5 Mitchell Mathis, MD
6 Division Director
7 Division of Psychiatry Products (DPP)
8 ODE I, OND, CDER, FDA
9
10 Judy Staffa, PhD, RPh
11 Associate Director for Public Health Initiatives
12 Division of Epidemiology II (DEPI-II)
13 Office of Pharmacovigilance and Epidemiology
14 Office of Surveillance and Epidemiology
15 CDER, FDA
16
17 Tiffany Farchione, MD
18 Deputy Director
19 DPP, ODE I, OND, CDER, FDA
20
21
22
A Matter of Record (301) 890-4188 13
1 Dominic Chiapperino, PhD
2 Director
3 Controlled Substance Staff
4 Office of the Center Director, CDER, FDA
5
6 Daniel J. Lee, MD
7 Clinical Reviewer
8 DPP, ODE I, OND, CDER, FDA
9
10 Semhar Ogbagaber, PhD
11 Statistician
12 Division of Biometrics I
13 Office of Biostatistics
14 Office of Translational Sciences, CDER, FDA
15
16
17
18
19
20
21
22
A Matter of Record (301) 890-4188 14
1 C O N T E N T S
2 AGENDA ITEM PAGE
3 Call to Order and Introduction of Committee
4 Raj Narendran, MD 17
5 Conflict of Interest Statement
6 Kalyani Bhatt, BS, MS 23
7 FDA Opening Remarks
8 Mitchell Mathis, MD 27
9 Applicant Presentations - Alkermes, Inc.
10 Introduction
11 Lisa von Moltke, MD 40
12 The Unmet Need in MDD and Challenges in
13 MDD Clinical Trials
14 George Papakostas, MD 51
15 Clinical Efficacy
16 Jerald Schindler, DrPH 65
17 Clinical Safety, and Risk Mitigation
18 Strategies
19 Gary Bloomgren, MD 97
20
21
22
A Matter of Record (301) 890-4188 15
1 C O N T E N T S (continued)
2 AGENDA ITEM PAGE
3 Clinical Perspective and Benefit-Risk
4 Profile
5 Sanjay Mathew, MD 116
6 Conclusion
7 Lisa von Moltke, MD 122
8 Clarifying Questions to Applicant 123
9 FDA Presentations
10 Regulatory History
11 Tiffany Farchione, MD 155
12 Clinical Efficacy and Safety Overview
13 Semhar Ogbagaber, PhD 166
14 Daniel Lee, MD 181
15 Abuse Potential of
16 Buprenorphine/Samidorphan (BUP/SAM)
17 Edward Hawkins, PhD 187
18 FDA Review of the Epidemiologic and
19 Surveillance Data
20 Celeste Mallama, PhD, MPH 197
21
22
A Matter of Record (301) 890-4188 16
1 C O N T E N T S (continued)
2 AGENDA ITEM PAGE
3 University of Washington Medical School
4 Department of Psychiatry and
5 Behavioral Sciences Presentation
6 Depression Effects on Long-Term
7 Prescription Opioid Use, Abuse and
8 Addiction
9 Mark Sullivan, MD, PhD 207
10 FDA Presentations (continued)
11 Risk Management for
12 Buprenorphine/Samidorphan
13 Somya Dunn, MD 221
14 Clarifying Questions to FDA and 228
15 Dr. Sullivan
16 Open Public Hearing 268
17 Clarifying Questions (continued) 332
18 Charge to the Committee
19 Mitchell Mathis, MD 335
20 Questions to the Committee and Discussion 336
21 Adjournment 446
22
A Matter of Record (301) 890-4188 17
1 P R O C E E D I N G S
2 (8:01 a.m.)
3 Call to Order
4 Introduction of Committee
5 DR. NARENDRAN: I would first like to remind
6 everyone to please silence your cell phones,
7 smartphones, other devices if you have not already
8 done so. I would also like to identify the FDA
9 press contact, Sandy Walsh. If you're there,
10 please stand. She's over there.
11 My name is Raj Narendran. I'm the
12 chairperson for today's meeting. I will now call
13 the Joint Meeting of the Psychopharmacology Drug
14 Advisory Committee and the Drug Safety and Risk
15 Management Advisory Committee to order.
16 We'll start by going around the table and
17 introduce ourselves. We will start with the FDA to
18 my left and go around the table.
19 DR. UNGER: Good morning. I'm Ellis Unger.
20 I'm director of Office of Drug Evaluation I in the
21 Office of New Drugs, CDER.
22 DR. MATHIS: Mitchell Mathis, director of
A Matter of Record (301) 890-4188 18
1 the Division of Psychiatry Products.
2 DR. FARCHIONE: Tiffany Farchione, deputy
3 director of the Division of Psychiatry Products.
4 DR. OGBAGABER: Dr. Semhar Ogbagaber,
5 Division of Psychiatry, Office of Biostatistics
6 Division I
7 DR. STAFFA: Good morning. I'm Judy Staffa.
8 I'm the associate director for public health
9 initiatives in the Office of Surveillance and
10 Epidemiology.
11 DR. CHIAPPERINO: Good morning. I'm Dominic
12 Chiapperino. I'm the director in the controlled
13 substance staff.
14 DR. RUHA: Hi. I'm Michelle Ruha. I'm a
15 medical toxicologist and professor at the
16 University of Arizona College of Medicine, Phoenix.
17 DR. WARHOLAK: Hi. I'm Terri Warholak, and
18 I'm a professor and assistant dean at the
19 University of Arizona College of Pharmacy, Tucson.
20 DR. KULLDORFF: Good morning. My name is
21 Martin Kulldorff. I'm a biostatistician at the
22 Brigham and Women's Hospital and Harvard Medical
A Matter of Record (301) 890-4188 19
1 School.
2 DR. HABEL: Hi. I'm Laurel Habel. I'm an
3 epidemiologist and associate director for cancer
4 research at Kaiser Permanente Northern California.
5 DR. HERNANDEZ-DIAZ: Sonia Hernandez-Diaz,
6 pharmacoepidemiologist, Harvard Chan School of
7 Public Health.
8 DR. IYENGAR: I'm Satish Iyengar. I'm a
9 professor of statistics at the University of
10 Pittsburgh. I also have a joint appointment in
11 psychiatry at Western Psych.
12 DR. DUNN: Walter Dunn, psychiatrist,
13 assistant professor at UCLA in the West Los Angeles
14 VA.
15 MS. BHATT: Good morning. I'm Kalyani
16 Bhatt. I'm the designated federal officer for this
17 committee.
18 DR. NARENDRAN: Raj Narendran, psychiatrist,
19 University of Pittsburgh Medical Center.
20 DR. FIEDOROWICZ: Jess Fiedorowicz. I'm a
21 psychiatrist, associate professor of psychiatry,
22 epidemiology, and internal medicine at the
A Matter of Record (301) 890-4188 20
1 University of Iowa.
2 MS. WITCZAK: Good morning. Kim Witczak.
3 I'm consumer representative and founder of
4 Woodymatters.
5 MS. JONIAK-GRANT: Hello. I'm Elizabeth
6 Joniak-Grant. I'm the patient representative. I'm
7 also a sociologist, and I'm affiliated with the
8 National Coalition of Independent Scholars.
9 DR. BESCO: Good morning. I'm Kelly Besco.
10 I'm a pharmacist by background, and I currently
11 serve as the medication safety officer for the Ohio
12 Health Hospital System in Columbus, Ohio.
13 DR. GRIFFIN: Good morning. Marie Griffin,
14 pharmacoepidemiologist and general internist at
15 Vanderbilt University.
16 DR. MEISEL: Steve Meisel, director of
17 medication safety, Fairview Health Services in
18 Minneapolis.
19 DR. ACRI: Good morning. I'm Jane Acri.
20 I'm chief of the Medication Discovery and
21 Toxicology Branch at the National Institute on Drug
22 Abuse.
A Matter of Record (301) 890-4188 21
1 DR. DE WIT: My name is Harriet de Wit. I'm
2 at the University of Chicago, Department of
3 Psychiatry and Behavioral Neuroscience.
4 DR. JENSEN: Hi. My name is Roxanne Jensen.
5 I'm a program director in the outcomes research
6 branch at the National Cancer Institute.
7 DR. FLYNN: Hi. I'm Kathryn Flynn. I'm an
8 associate professor of medicine at the Medical
9 College of Wisconsin and an outcomes researcher.
10 DR. MARSHALL: Good morning. I'm Brandon
11 Marshall. I'm an associate professor of
12 epidemiology at the Brown School of Public Health
13 in Providence, Rhode Island.
14 DR. CRAWFORD: Good morning. My name is
15 Stephanie Crawford. I'm professor at the
16 University of Illinois at Chicago, Colleges of
17 Pharmacy and Medicine, and I'm also executive
18 associate dean.
19 DR. RILEY: Bill Riley. I'm the associate
20 director for behavioral and social sciences at the
21 National Institutes of Health.
22 DR. CELLA: Hi. I'm David Cella, also from
A Matter of Record (301) 890-4188 22
1 Chicago, but Northwestern University at Chicago.
2 I'm a clinical psychologist and outcomes
3 researcher.
4 DR. CONLEY: Good morning. I'm Rob Conley.
5 I'm a psychiatrist. I'm the distinguished scholar
6 for neuroscience at Eli Lilly and a professor of
7 psychiatry and pharmacy science at the University
8 of Maryland School of Medicine. I'm here as the
9 industry representative.
10 DR. NARENDRAN: Dr. Jain, if you could
11 introduce yourself.
12 DR. JAIN: Hi. I'm Felipe Jain, assistant
13 professor of psychiatry at Harvard Medical School
14 and a psychiatrist at the Massachusetts General
15 Hospital.
16 DR. NARENDRAN: Thank you. For topics such
17 as those being discussed at today's meeting, there
18 are often a variety of opinions, some of which are
19 quite strongly held. Our goal is that today's
20 meeting will be a fair and open forum for
21 discussion of these issues and those individuals
22 can express their views without interruption.
A Matter of Record (301) 890-4188 23
1 Thus, as a gentle reminder, individuals will
2 be allowed to speak into the record only if
3 recognized by the chairperson. We look forward to
4 a productive meeting.
5 In the spirit of the Federal Advisory
6 Committee Act and the Government in the Sunshine
7 Act, we ask that advisory committee members take
8 care that their conversations about the topic at
9 hand take place in the open forum of the meeting.
10 We are aware that members of the media are
11 anxious to speak with the FDA about these
12 proceedings. However, FDA will refrain from
13 discussing the details of this meeting with the
14 media until its conclusion. Also, the committee is
15 reminded to please refrain from discussing the
16 meeting topic during breaks or lunch. Thank you.
17 Now, I will pass it to Kalyani Bhatt, who
18 will read the conflict of interest statement.
19 Conflict of Interest Statement
20 MS. BHATT: Good morning. The Food and Drug
21 Administration is convening today's Joint Meeting
22 of the Psychopharmacologic Drugs Advisory Committee
A Matter of Record (301) 890-4188 24
1 and the Drug Safety and Risk Management Advisory
2 Committee under the authority of the Federal
3 Advisory Committee Act, FACA, of 1972.
4 With the exception of the industry
5 representative, all members and temporary voting
6 members of the committees are special government
7 employees, SGEs, or regular federal employees from
8 other agencies and are subject to federal conflict
9 of interest laws and regulations.
10 The following information on the status of
11 the committee's compliance with federal ethics and
12 conflict of interest laws, covered by but not
13 limited to those found at 18 U.S.C. Section 208, is
14 being provided to participants in today's meeting
15 and to the public.
16 FDA has determined that members and
17 temporary voting members of these committees are in
18 compliance with the federal ethics and conflict of
19 interest laws. Under 18 U.S.C. Section 208,
20 Congress has authorized FDA to grant waivers to
21 special government employees and regular federal
22 employees who have potential financial conflicts,
A Matter of Record (301) 890-4188 25
1 when it is determined that the agency's need for a
2 special government employee's services outweighs
3 his or her potential financial conflict of interest
4 or when the interest of a regular federal employee
5 is not so substantial as to be deemed likely to
6 affect the integrity of the services which the
7 government may expect from the employee.
8 Related to the discussion of today's
9 meeting, members and temporary voting members of
10 these committees have been screened for potential
11 financial conflicts of interest of their own as
12 well as those imputed to them, including those of
13 their spouses or minor children, and for purposes
14 of 18 U.S.C. Section 208, their employers.
15 These interests may include investments,
16 consulting, expert witness testimony, contracts,
17 grants, CRADAs, teaching, speaking, writing,
18 patents and royalties, and primary employment.
19 Today's agenda involves discussion of the
20 efficacy, safety, and risk-benefit profile of new
21 drug application NDA210417 for buprenorphine and
22 samidorphan sublingual tablets, submitted by
A Matter of Record (301) 890-4188 26
1 Alkermes, Incorporated for adjunctive treatment of
2 major depressive disorder.
3 This is a particular matters meeting during
4 which specific matters related to Alkermes' NDA
5 will be discussed. Based on the agenda for today's
6 meeting and all financial interests reported by the
7 committee members and temporary voting members, no
8 conflict of interest waivers have been issued in
9 connection with this meeting.
10 To ensure transparency, we encourage all
11 standing committee members and temporary voting
12 members to disclose any public statements that they
13 have made concerning the product at issue.
14 With respect to FDA's invited industry
15 representative, we would like to disclose that
16 Dr. Robert Conley is participating in this meeting
17 as a non-voting industry representative, acting on
18 behalf of regulated industry. Dr. Conley's role at
19 this meeting is to represent industry in general
20 and not any particular company. Dr. Conley is
21 employed by Eli Lilly and Company.
22 We'd like to remind members and temporary
A Matter of Record (301) 890-4188 27
1 voting members that if the discussions involve any
2 other products or firms not already on the agenda
3 for which an FDA participant has a personal or
4 imputed financial interest, the participants need
5 to exclude themselves from such involvement, and
6 their exclusion will be noted for the record.
7 FDA encourages all other participants to
8 advise the committee of any financial relationships
9 that they may have with the firm at issue. Thank
10 you.
11 DR. NARENDRAN: We will now proceed with the
12 FDA's introductory remarks, presented by
13 Dr. Mitchell Mathis, division director.
14 DR. MATHIS: I think that's the wrong
15 presentation.
16 MS. BHATT: We're having some issues,
17 Dr. Mathis. Just give us a minute.
18 FDA Opening Remarks - Mitchell Mathis
19 DR. MATHIS: Thank you.
20 My name is Dr. Mitchell Mathis, and I'm the
21 director of psychiatry products in the Center for
22 Drugs here at FDA. I'd like to thank you all for
A Matter of Record (301) 890-4188 28
1 coming. It's good to see so many people so early.
2 Thank you for being here.
3 What I would like to do is give you an
4 overview of what we're going to talk about today,
5 to try and set the stage for the issues that my
6 team and I think are important to talk about.
7 So by way of introduction, I will introduce
8 the product, buprenorphine samidorphan; introduce
9 the parts of the disease that the company is
10 developing this drug to treat, the form of the
11 disease; talk about current available treatments
12 that we have; and we have to spend some time
13 discussing the definition of substantial evidence.
14 I think you will see that that will be important to
15 today's discussion.
16 I'd also like to engage the committee for
17 their thoughts on the management of placebo
18 response. And you'll see as we proceed today that
19 the company has employed some strategies to address
20 placebo response. I'll go over the agenda for
21 today and then the core questions for the
22 committee.
A Matter of Record (301) 890-4188 29
1 The product is a combination product of
2 buprenorphine and samidorphan. Buprenorphine is an
3 approved drug. It's a partial agonist at the
4 mu opioid receptor, and it's approved for the
5 treatment of medication-assisted treatment in
6 opioid dependence disorders, and it's approved to
7 treat pain.
8 Samidorphan is the new molecular entity
9 here. Samidorphan is an antagonist, a blocker at
10 that same opioid receptor. And the pharmacologic
11 idea between the combination of these drugs is to
12 provide the buprenorphine and the samidorphan
13 together in a bioavailable formulation.
14 Samidorphan binds at mu opioid receptor and
15 prevents the opioid effects of the buprenorphine,
16 whereas the other effects of buprenorphine,
17 presumably which include treating depression, can
18 be present without the dependency and abuse risk.
19 That's how it was designed.
20 It's important to note that the indication
21 that's being sought is the adjunctive treatment of
22 major depressive disorder. Now, adjunctive
A Matter of Record (301) 890-4188 30
1 treatment of major depressive disorder, briefly,
2 means this. The patient has been exposed to a
3 monotherapy agent and has gotten better, but not
4 good enough. Your choices then, as a clinician, are
5 to stop the medication, start another, and see if
6 you can get a single medication to work.
7 If you've done that one or two times and the
8 patient is better but not good enough, then the
9 technique employed is adjunctive treatment. So you
10 can add a second drug -- this drug was developed
11 for that reason -- to make the patient completely
12 better, closer to remitted. And that's the idea
13 behind adjunctive treatment of major depressive
14 disorder.
15 The disease major depressive disorder is of
16 course a debilitating and a chronic illness. There
17 are millions of Americans right now suffering from
18 major depressive disorder. It's a leading cause of
19 disability worldwide. And unfortunately, partial
20 response is more the rule than the exception.
21 The sequence treatment alternatives to
22 relieve depression, study STAR*D, about a third of
A Matter of Record (301) 890-4188 31
1 patients get better the first time you try to make
2 them better with a monotherapy treatment. Now,
3 that means get better to a remitted state, and you
4 can define remission in different ways, but to
5 remission.
6 So there is a public health need for
7 effective add-on medications for the rest of the
8 patients who don't remit.
9 In terms of approved medications, we have
10 several drugs in different classes for monotherapy,
11 but there are only three drugs approved for this
12 indication: the adjunctive treatment of partially
13 responsive depression, quetiapine, extended
14 release; aripiprazole; and brexpiprazole.
15 I'm sure many of you have noticed that those
16 medications are all in the same class of drugs,
17 atypical or second-generation antipsychotics, which
18 means that they have safety risks in common.
19 Movement disorders, acute and chronic movement
20 disorders are a possibility with those drugs and
21 the metabolism syndrome: weight gain,
22 hypercholesterolemia, hyperglycemia. All three of
A Matter of Record (301) 890-4188 32
1 those are risk factors, cardiovascular risk
2 factors.
3 So buprenorphine/samidorphan, should it be
4 approved, would be in a different class to treat
5 this partially responsive depression.
6 Buprenorphine/samidorphan, however, has its
7 own safety risks, which we'll talk about today, and
8 they're different than the other drugs, but
9 remember that buprenorphine is an opiate, and
10 samidorphan has reduced the dependency risk and the
11 problems of the opiate, but not completely remove
12 them.
13 Substantial evidence; this concept is
14 important because we're going to focus on this a
15 great deal today. In 1962, the Food, Drug, and
16 Cosmetic Act was amended to require for the first
17 time that drug manufacturers establish a drug's
18 effectiveness by this concept of substantial
19 evidence.
20 Substantial evidence is defined as evidence
21 consisting of adequate and well-controlled
22 investigations, and it's long been FDA's position
A Matter of Record (301) 890-4188 33
1 that Congress intended for investigations to be
2 plural in the sense that at least two adequate and
3 well-controlled trials, each convincing on its own,
4 should be submitted to support a drug's
5 effectiveness.
6 There are situations where a single study
7 can be used to approve a drug. In 1997, the FDA
8 Modernization Act, FDAMA, amended that same section
9 of the Act to make it clear that FDA may rely,
10 quote, "on data from one adequate and well-
11 controlled clinical investigation and confirmatory
12 evidence" end quote, to constitute substantial
13 evidence if FDA determines that these data and
14 evidence are sufficient to establish effectiveness.
15 So practically speaking, this single study
16 with confirmatory evidence is reserved for
17 situations where an important clinical benefit like
18 a direct effect on survival is clearly -- when I
19 say clearly, I mean statistically clearly and
20 clinically clearly -- so that a confirmatory study,
21 would be either hard to repeat because there, for
22 instance, aren't enough patients, or there would be
A Matter of Record (301) 890-4188 34
1 an ethical reason not to repeat the study, to take
2 the time or the exposure to repeat the study.
3 That's not the case with adjunctive
4 treatment of major depressive disorder. As I've
5 already mentioned, it's unfortunate, but we have
6 many patients who have partially responsive
7 depression, and the other three drugs, including
8 this drug, have done multiple studies to assess
9 efficacy.
10 It's important to note that FDA and the
11 applicant do not agree on whether this drug has met
12 the standard for substantial evidence of
13 effectiveness. The applicant will argue that they
14 have two positive studies, study 202 and 207. To
15 provide substantial evidence of effectiveness for
16 buprenorphine/samidorphan, FDA will argue that
17 substantial evidence has not been provided, and my
18 team will provide our arguments to support this
19 case. We'll be discussing the efficacy in depth,
20 and I ask that you listen to both sides of the
21 argument and form your own opinion, which we will
22 ask you for later.
A Matter of Record (301) 890-4188 35
1 In terms of safety, the applicant will argue
2 that the opiate effects of buprenorphine are
3 largely but not completely blocked by samidorphan,
4 and my team and I will agree with that. There are
5 some opioid effects, but samidorphan seems to have
6 largely done whatever it was designed to do.
7 Here are the trials. The applicant has
8 conducted four trials designed to demonstrate
9 efficacy, two they consider positive and two they
10 consider negative. Since we agree on the negative
11 trials and since it's not usual to see up to half
12 of trials in MDD development programs fail, we need
13 not discuss these negative trials further with
14 regard to efficacy in my view.
15 I've asked my team to carefully discuss why
16 they have concluded that studies 202 and 207 are
17 not positive because there's disagreement with the
18 applicant on this point and because this will be of
19 obvious importance when we're asking you your
20 opinion about substantial evidence.
21 Briefly, 202 was originally designed as a
22 phase 2 proof-of-concept study, so it was designed
A Matter of Record (301) 890-4188 36
1 without a multiplicity plan to adjust for
2 evaluating the efficacy of multiple doses in the
3 same trial.
4 It's noteworthy to note that the study is
5 negative at the higher dose, the 8/8 dose; 8/8
6 means 8 milligrams of buprenorphine, 8 milligrams
7 of samidorphan. And it's negative if the 8/8 and
8 the 2/2 doses are combined. It's also negative and
9 it's important to point out if a single patient is
10 removed from the analysis, and we'll talk more
11 about that later.
12 The lack of an effect at a higher dose is
13 very concerning to us. It undercuts whatever
14 evidence there might be that a lower dose could be
15 effective here. It's easy to propose reasons why a
16 lower dose might work and a higher dose doesn't,
17 but in our experience, we've never seen this.
18 If we were to approve this drug on the basis
19 of two positive studies, with study 202 considered
20 as one of the two, it's important to recognize that
21 we'd be approving the first drug of an entirely new
22 class, knowing that evidence of effectiveness
A Matter of Record (301) 890-4188 37
1 hinges on data from a single patient. Remove this
2 patient and the nominal p-value is not
3 statistically significant anymore. There will be
4 more about this later.
5 Study 207 is positive only by an average of
6 endpoints over multiple visits. FDA advised
7 against this approach. And it is negative by the
8 usual end-of-treatment endpoint on the depression
9 rating scales.
10 It might not be irrational to average rating
11 scales in depression over multiple visits,
12 consecutive visits, to perhaps decrease
13 variability, but we've never taken a close look at
14 this, and we've explicitly told the applicant not
15 to do it here.
16 In closing, both studies, 202 and 207, are
17 of the SPCD design, Sequential Parallel Comparison
18 Design. We'll talk more about this later. It's a
19 design, however, that has not yet been determined
20 to be statistically acceptable to FDA.
21 I'd like to talk a little bit about the
22 management of placebo response. The sponsor has
A Matter of Record (301) 890-4188 38
1 done SPCD trials and a placebo run-in trial to
2 manage the placebo response. The placebo response
3 is, of course, a non-specific response to treatment
4 that's not related to active drug. And while this
5 is useful in the clinic in taking care of patients,
6 it is destructive to psychiatric drug development
7 trials.
8 It's a long-term problem in psych trials.
9 The Sequential Parallel Comparison Design was set
10 up to address this problem by putting more weight
11 on placebo non-responder data, and we'll talk to
12 you about that, but again, that's never been
13 accepted in the division.
14 So today, the applicant will discuss the
15 need for new treatments, the efficacy, safety, and
16 the risk-benefit of their drug product. FDA and
17 our guest speaker will then present our views of
18 efficacy and safety. We'll talk about the abuse
19 potential, the epidemiology of misuse and abuse,
20 the relationship of opioid use disorders to major
21 depressive disorder and how our risk evaluation and
22 mitigation strategy, REMS, could help to manage the
A Matter of Record (301) 890-4188 39
1 risk of buprenorphine/samidorphan, should it be
2 approved to treat MDD.
3 The committee will then discuss and vote on
4 several questions. I put the core questions here
5 so that you can see them before we get started.
6 There will be other questions, but these are the
7 ones that we'll need votes on.
8 Has substantial evidence of efficacy of
9 buprenorphine/samidorphan been presented by the
10 applicant? Has the applicant adequately
11 characterized the safety of buprenorphine/
12 samidorphan in treating major depressive disorder?
13 And do the available data support a favorable
14 benefit-risk profile of buprenorphine/samidorphan
15 to support approval.
16 That's all I have. I hand the meeting back
17 to the chair.
18 DR. NARENDRAN: Thank you, Dr. Mathis.
19 Both the FDA and the public believe in a
20 transparent process for information gathering and
21 decision making. To ensure such transparency at
22 the advisory committee meeting, FDA believes that
A Matter of Record (301) 890-4188 40
1 it is important to understand the context of an
2 individual's presentation.
3 For this reason, FDA encourages all
4 participants, including the sponsor's non-employee
5 presenters, to advise the committee of any
6 financial relationships that they may have with the
7 firm at issue, such as consulting fees, travel
8 expenses, honoraria, and interest in the sponsor,
9 including equity interests and those based upon the
10 outcome of the meeting.
11 Likewise, FDA encourages you, at the
12 beginning of your presentation, to advise the
13 committee if you do not have any such financial
14 relationships.
15 If you choose not to address this issue of
16 the financial relationships at the beginning of
17 your presentation, it will not preclude you from
18 speaking. We will now proceed with Alkermes'
19 presentations.
20 Applicant Presentation - Lisa von Moltke
21 DR. VON MOLTKE: Thank you, Mr. Chairman,
22 and good morning. Good morning to the other
A Matter of Record (301) 890-4188 41
1 members of the committee, representatives from the
2 agency, and other attendees. I'm Lisa von Moltke,
3 and I head up drug development at Alkermes, and I'm
4 going to be starting this morning with the
5 sponsor's presentation. We appreciate this
6 opportunity to present to you the buprenorphine/
7 samidorphan program for adjunctive treatment of
8 major depressive disorder.
9 Following my introduction, the rest of the
10 sponsor's presentation will proceed as outlined
11 here. Dr. George Papakostas from Harvard Medical
12 School will take us through the unmet need in MDD.
13 He will also outline the challenges in MDD clinical
14 trials.
15 Dr. Jerry Schindler from Alkermes will take
16 you through the clinical efficacy data. Dr. Gary
17 Bloomgren will take you through the clinical safety
18 data, and he'll also present the risk mitigation
19 strategies that Alkermes will be proposing.
20 Dr. Sanjay Mathew from Baylor College of
21 Medicine will give his clinical perspective and the
22 benefit-risk profile, and then I will close the
A Matter of Record (301) 890-4188 42
1 sponsor's presentation.
2 We also have with us an external
3 participant. Dr. Aparna Anderson is with us from
4 Statistics Collaborative. Alkermes also has a
5 number of participants, and they are listed here,
6 and they will introduce themselves as they answer
7 questions for the committee.
8 Now, there will be some key topics under
9 discussion today that I want to outline before we
10 get started. First, there is a major significant
11 unmet need for patients with major depressive
12 disorder. Specifically, there is a need for
13 therapies that work by new mechanisms of action.
14 There are significant development challenges
15 in studying MDD, including the high placebo
16 response rates, which you've already heard about.
17 And this warrants the use of advances in
18 methodology, including the SPCD trial design.
19 You're going to see data indicating that
20 buprenorphine/samidorphan has a positive benefit-
21 risk profile. The data include substantial
22 evidence of efficacy. It's comparable to other
A Matter of Record (301) 890-4188 43
1 adjunctive treatments, but with a completely
2 different mechanism of action. The safety profile
3 has been well characterized, and it's manageable,
4 and the abuse potential is low.
5 Finally, Alkermes realizes that it is
6 proposing a new opioid modulator in the midst of an
7 opioid crisis. Even with a profile that indicates
8 that the risk for abuse and dependence has been
9 mitigated by the addition of samidorphan, this will
10 call for a full commitment by Alkermes to the
11 obligations of education and monitoring, and we
12 will be presenting those to you.
13 I want to highlight some specific points
14 about the program to serve as a framework for
15 information that you're going to receive this
16 morning. And one of the first things is that I'm
17 going to be referring to buprenorphine/samidorphan
18 as BUP/SAM for the sake of efficiency.
19 The first point is that this program had its
20 genesis in the clinical observations that opioids,
21 and buprenorphine in particular, may have
22 antidepressant effects.
A Matter of Record (301) 890-4188 44
1 Samidorphan is being added as a way to
2 mitigate the abuse and dependence potential that we
3 know exists with buprenorphine. And as you've
4 already heard, samidorphan is a mu opioid
5 antagonist.
6 This therapy is being proposed for
7 adjunctive treatment of major depressive disorder,
8 so this is for patients who are not responding
9 adequately to available therapies, and they're
10 continuing to suffer with depressive symptoms.
11 The proposed dose is 2 milligrams of BUP and
12 2 milligrams of SAM, following a 1-week initial
13 titration. And there will also be a 1:1 dose for
14 special populations, and that's the patients that
15 have renal insufficiency or hepatic insufficiency.
16 The BUP/SAM development regulatory history
17 began with a pre-IND meeting in 2011. And during
18 that meeting, there were initial conversations
19 around study designs, including SPCD. There was an
20 end-of-phase 2 meeting in October of 2013, and
21 during that time, there were again discussions
22 around SPCD study designs.
A Matter of Record (301) 890-4188 45
1 The clinical review that came out of that
2 meeting indicated that, from a clinical
3 perspective, the trials appeared acceptable. That
4 same year, the program received fast-track
5 designation since BUP/SAM has the potential to
6 address an unmet medical need that serious. And
7 this was based on the phase 2 study, and that's the
8 202 study that you saw presented by Dr. Mathis.
9 There have been two scientific exchange
10 meetings, one in 2016 and one in 2017, and during
11 that time, the results of phase 3 studies were
12 shared. There was then a pre-NDA meeting in July
13 of 2017, and the NDA submission content was
14 discussed.
15 The applicant then filed in January of 2018,
16 and Alkermes was very surprised to get a refuse-to-
17 file in March. This was based on some factual
18 inaccuracies in the agency's assessment, and the
19 refuse-to-file was rescinded just 2 weeks later.
20 Before I leave the regulatory history, I
21 want to note a few points here that are not on this
22 slide. This process has been an ongoing multi-year
A Matter of Record (301) 890-4188 46
1 interaction with FDA with multiple points of
2 contact. Throughout the entire program, we have
3 been clear and FDA has acknowledged that our
4 evidence of efficacy was going to rely upon SPCD
5 studies.
6 At the pre-NDA meeting last summer, we
7 presented our plan to submit based on evidence from
8 studies 202, 205, and 207, and this was confirmed
9 with FDA. And as you've already heard, today
10 you're going to hear a couple areas of
11 disagreement, and these are regarding the utility
12 of the 202 study and the importance of the data
13 from the 205 study.
14 The 202 study was robustly designed and met
15 its prespecified primary endpoint. Now, FDA is
16 going to assert that a particular patient cannot be
17 included due to lack of eligibility. Alkermes has
18 sent a team and conducted an extensive
19 investigation and audit, and we were able to
20 substantiate multiple sources documenting the
21 eligibility, and these have been submitted to the
22 agency.
A Matter of Record (301) 890-4188 47
1 The 205 study is an important supportive
2 piece of evidence, and there is regulatory
3 precedent in MDD for considering it as such. And
4 we will be presenting the details of these studies,
5 both 202 and 205, to you this morning.
6 Alkermes undertook the development of
7 BUP/SAM because MDD is a major source of morbidity
8 and disability, and this illness carries a risk of
9 suicide. A significant percentage of patients do
10 not achieve adequate symptom relief with standard
11 therapies. All the approved antidepressants work
12 via monoaminergic mechanisms, so they're all
13 working on serotonin, norepinephrine, or dopamine,
14 and only antipsychotics are currently approved as
15 adjunctive therapies, and these can have
16 significant side effects. They can be serious and
17 they can be permanent.
18 So new approaches to treat depression, and
19 adjunctive treatment in particular, are urgently
20 needed. These options are needed by patients, and
21 the physicians, and other healthcare professionals
22 who care for them.
A Matter of Record (301) 890-4188 48
1 The BUP/SAM program is founded on evidence
2 in the literature that exists over decades that
3 opioids, and particularly buprenorphine, may have
4 antidepressant effects. We know that opioid
5 receptors are highly expressed in brain regions
6 that are associated with emotional regulation. But
7 a broader use in this indication has been limited
8 by the concerns around abuse and dependence that we
9 know exists with buprenorphine.
10 Samidorphan has been added to address this
11 abuse and dependence liability, and it's an
12 integral part of the resulting pharmacologic
13 profile for this therapy. And that's in contrast
14 to other antagonists that are added simply to deter
15 parenteral use, for example.
16 So the resulting therapy would consist of
17 buprenorphine, which is a partial mu opioid agonist
18 and a kappa opioid antagonist, and it has
19 bioavailability by the sublingual route.
20 Samidorphan, as you've heard, is the new molecular
21 entity here. And it's a potent mu opioid
22 antagonist. It's been optimized for that high
A Matter of Record (301) 890-4188 49
1 potency at the receptor. It's also been optimized
2 to have high bioavailability by the oral and
3 sublingual routes.
4 Again, this contrasts to other antagonists
5 that are combined with buprenorphine that do not
6 have bioavailability by the intended routes. So
7 the combination is co-formulated in a single
8 sublingual tablet, and it would be packaged in
9 individual blisters, as you see here.
10 This has been a large program with an
11 extensive safety and efficacy evaluation of
12 BUP/SAM. There have been 34 clinical studies, 19
13 conducted with BUP/SAM together and 15 with
14 samidorphan alone. There have been 4 placebo-
15 controlled studies in the MDD patient population,
16 and then there's been a long-term safety study.
17 Over 2,000 patients, so our subjects have
18 received BUP/SAM together, and over 1500 MDD
19 patients have been treated with the specific 2/2
20 dose.
21 The study population for the development
22 program is comprised of patients diagnosed with
A Matter of Record (301) 890-4188 50
1 MDD. They had a mean lifetime number of major
2 depressive episodes that ranged from 4 to 7
3 depending on the trial. They had all cycled
4 through multiple therapies, and the median duration
5 of the current episode was anywhere from 9 to
6 10 months.
7 They had had 1 to 2 inadequate responses to
8 an antidepressant therapy in the current cycle and
9 episode. So they also continued on their
10 background ADT, so they were all also on an SSRI,
11 SNRI, or bupropion.
12 So this morning, you're going to see the
13 evidence of efficacy and safety. Dr. Schindler is
14 going to present to you the two studies that met
15 their prespecified primary endpoints. These are
16 202 and 207. He's also going to present to you the
17 supportive evidence of efficacy from a third study,
18 and that's study 205.
19 The efficacy seen in these studies is
20 comparable to that seen with the other approved
21 adjunctive treatments for MDD, and it achieves this
22 efficacy by a totally new mechanism of action.
A Matter of Record (301) 890-4188 51
1 Dr. Gary Bloomgren is then going to present
2 to you the safety data, and you will see that this
3 therapy is a generally well-tolerated therapy.
4 Common AEs are GI and sedation related, and the
5 abuse potential is low. The overall benefit-risk
6 profile is positive.
7 Finally, because of the presence of
8 buprenorphine, even in the setting where this abuse
9 and dependence has been mitigated by the presence
10 of samidorphan, Alkermes is going to commit to
11 education and monitoring to ensure that the use of
12 this therapy is appropriate, and you're going to
13 hear about those propositions as well.
14 With that, I'm going to turn the
15 presentation over to Dr. George Papakostas, who
16 will take us through the unmet need in MDD.
17 Dr. Papakostas?
18 Applicant Presentation - George Papakostas
19 DR. PAPAKOSTAS: Thank you very much,
20 Dr. von Moltke.
21 Good morning, everyone. My name is George
22 Papakostas, and I am associate professor of
A Matter of Record (301) 890-4188 52
1 psychiatry at Harvard Medical School, director of
2 treatment-resistant depression studies at
3 Massachusetts General Hospital, and also a
4 scientific director at MGHCTNI. And essentially,
5 I'm up here this morning to convey two important
6 ideas that are clearly interlinked. So there's
7 going to be two components to this brief
8 presentation.
9 The first essentially echoes what the first
10 two discussions, the first two talks covered this
11 morning, is that major depressive disorder is a
12 serious and common medical disorder affecting
13 humanity. And in my point of view, and many would
14 agree, we fall terribly short in terms of
15 treatments for major depressive disorder. We have
16 a long way ahead of us to develop more treatments
17 for major depressive disorder.
18 It's a common medical illness. The lifetime
19 prevalence in adults in the United States is 16.6
20 percent as per 1 estimate. Because it involves so
21 many symptoms, it significantly impairs home, work,
22 life relationships, and social life, such that
A Matter of Record (301) 890-4188 53
1 almost 60 percent of patients with depression over
2 12 months report severe or very severe role
3 impairment. So it affects functioning and quality
4 of life at its core.
5 It's, as such, the number one contributor of
6 disability worldwide. It's a very disabling
7 illness. And being a medical illness, it increases
8 the risk of other medical illnesses such as heart
9 disease, diabetes mellitus, and cancer. And
10 equally importantly, is a major risk factor for
11 suicide. Suicide is climbing as a threat to
12 humanity. It results in more deaths than terrorism
13 and war put together. It is a major health threat.
14 So in my view, there are two gaps, there are
15 two big gaps in our development of treatments for
16 major depressive disorder. The first gap is that
17 over the last 50 years, we focused extensively on
18 developing monotherapies for depression.
19 Monotherapy is a clinical term that essentially
20 means giving a sole drug to treat major depressive
21 disorder.
22 We didn't find out until late, until 2006,
A Matter of Record (301) 890-4188 54
1 how much that strategy falls short of patients'
2 needs. Here is a slide from STAR*D. STAR*D is the
3 largest clinical trial ever conducted in
4 psychiatry, and it essentially looked at the
5 following question.
6 If you took patients with depression who
7 were treatment naïve and gave them different
8 options in succession, how likely was it that
9 patients would get better over these sequential
10 treatments?
11 So as we heard from Dr. Mathis'
12 presentation, treatment-naïve patients, adults with
13 MDD, you give the first treatment -- in this case,
14 it was an SSRI, citalopram -- one-third achieved
15 remission; two-thirds remain symptomatic.
16 You then switch, which is a clinical term
17 meaning you stop one therapy and you start another,
18 second-line antidepressant monotherapy; three-
19 quarters remain symptomatic. Then you go to a
20 third-line antidepressant therapy; 80 to 90 percent
21 remain symptomatic. You go to a fourth-line
22 antidepressant monotherapy; again, almost
A Matter of Record (301) 890-4188 55
1 90 percent remain symptomatic.
2 So the message from STAR*D here is clear.
3 If you rely on antidepressant monotherapy,
4 eventually it will become a futile effort.
5 The challenge, then, is that if there's
6 partial or non-response in patients with
7 depression, you get additional bad outcomes on top
8 of the outcomes of MDD. Patients who are partial
9 or non-responders, versus remitters, are at higher
10 risk for their illness coming back. As you can see
11 in the survival analysis on the left-hand side of
12 this slide, in blue you have no persistent
13 symptoms; in orange, you have persistent symptoms;
14 and the further down you are means that less
15 patients remain illness free.
16 There's a clear separation. The more you
17 get through remission, the safer you are from this
18 illness coming back. Patients who are incomplete,
19 insufficient responders, are 2 times more likely to
20 be hospitalized, are twice more likely to attempt
21 suicide, and are 3 times more likely to use
22 additional psychotropic medications compared to the
A Matter of Record (301) 890-4188 56
1 overall MDD population. So we need to get as many
2 patients better as quickly as possible to avoid
3 additional bad outcomes.
4 The second limitation of the development of
5 antidepressants is that for the past 50 years,
6 we've relied on essentially the permutation of a
7 single mechanism of action; that is, blocking the
8 reuptake of monoamines, predominantly serotonin,
9 sometimes norepinephrine, on rare occasions
10 dopamine, a very unimaginative approach and very
11 limited approach.
12 Of course, clinicians have caught on to the
13 idea that antidepressant monotherapies have become
14 progressively futile and they use adjuncts. Up
15 until 10 years ago, 15 years ago, there were no
16 FDA-approved adjuncts. If you used something,
17 there was limited evidence telling you it would
18 work, even though that is the most common approach
19 after 1 or 2 treatments fail, as we heard from the
20 earlier presenters.
21 Fortunately, we have three FDA-approved
22 adjuncts today. The limitation is that they're all
A Matter of Record (301) 890-4188 57
1 atypical antipsychotics and they're all
2 monoaminergic. So essentially, if you rely on the
3 evidence base for adjuncts, you have to have the
4 difficult discussion with the patient that, if you
5 try this medication, which has been proven to be
6 effective in treating depression, depending on what
7 you use, you may have to have a difficult talk
8 about weight gain, the chances of weight gain,
9 dyslipidemia, glucose dysregulation on rare
10 occasion, but very important, new onset diabetes.
11 You have to talk about akathisia, the fact
12 that they may experience restlessness, which will
13 be apparent to others, Parkinsonism, stiffness.
14 You have to talk about tardive dyskinesia.
15 Patients with depression often need long-term
16 therapy. You don't just get them better and take
17 them off their meds. The illness will recur.
18 The chances of tardive dyskinesia, which is
19 essentially motor ticks that are involuntary and
20 permanent, increase over time. This is a difficult
21 discussion to have with patients. This is a big
22 burden for patients, and this drives patients away
A Matter of Record (301) 890-4188 58
1 from a potentially effective therapy, shunting them
2 to adjuncts that are not really supported by
3 substantial evidence, and of course the difficult
4 discussion about potentially life-threatening
5 neuroleptic malignant syndrome that's been observed
6 with the antipsychotics, including the atypicals,
7 that in some patients, body temperature can rise
8 and this is potentially fatal; useful agents from
9 an efficacy standpoint; a lot of difficult
10 conversations, and risks of the safety side of
11 things.
12 So this is where we stand today. We have a
13 difficult and serious illness that's negatively
14 impacting humanity more and more each year. We
15 have a number of therapies. Patients are still
16 symptomatic. Many patients are still symptomatic.
17 Symptomatic patients have a higher illness burden
18 than the general MDD population and especially
19 those that are partial non-responders.
20 What we have today work on a single
21 mechanism of action, so we need more treatments
22 that replicate the efficacy of previous treatments,
A Matter of Record (301) 890-4188 59
1 are not associated with side effects, can be used
2 adjunctively, and hopefully are more imaginative in
3 terms of their mechanism of action, which brings me
4 to my second talk and my second portion of my
5 presentation, what is the challenge in getting
6 there?
7 The major challenge in getting there is that
8 the clinical research arena for major depressive
9 disorder is an extremely difficult and technical
10 arena. This is one of my favorite presentations,
11 one of my favorite papers to cite when trying to
12 illustrate this problem.
13 Essentially, this is an analysis from the
14 FDA, from Khin and colleagues, looking at the
15 following question. If you look at FDA-approved
16 agents, how many studies in the registry are
17 positive and how many are negative? Almost half of
18 studies in FDA-approved drugs are negative. So
19 that doesn't really count the number of drugs that
20 we've completely lost that could have helped
21 because of negative studies. So if you only look
22 at the FDA-approved drugs, 50 percent of trials
A Matter of Record (301) 890-4188 60
1 fail. That is a very bad statistic. That is a
2 sobering statistic.
3 So what is the response? We can't just
4 watch this happen. What is the response? What's
5 the reaction of the clinical research community to
6 try to stay one step ahead, to go through
7 thoughtful, carefully planned innovation?
8 Just to give you an example of how things
9 have changed in the adjunctive treatment world,
10 we've gone from simple randomization and that gave
11 us quetiapine XR. In simple randomization, you
12 essentially take patients who you know have
13 depression, who you know are on an antidepressant,
14 and you know that that antidepressant has not
15 worked by history.
16 So they report, or their doctors report,
17 that this hasn't worked in these patients. That is
18 what we call a retrospective assessment or
19 confirmation of treatment, non-response or partial
20 response.
21 So we've gone from that to a better design,
22 which is a misnomer. It's called the placebo
A Matter of Record (301) 890-4188 61
1 run-in, even though patients at all times of the
2 study are on an antidepressant and a dummy pill.
3 This design gave us aripiprazole and brexpiprazole,
4 two very useful agents, two atypical antipsychotics
5 like quetiapine.
6 This was novel and considered revolutionary
7 10, 15 years ago. This is something we accept now
8 as being part of the standard. So this gave us
9 aripiprazole and brexpiprazole. Essentially, what
10 happens is you have patients with depression. You
11 treat them in the study with an antidepressant to
12 confirm whether that works or not and, if it
13 doesn't, you randomize them to aripiprazole,
14 brexpiprazole, or placebo in these cases. This is
15 what's called prospective confirmation of
16 treatment, non-response or partial response.
17 A step ahead is what's called the SPCD
18 design, which is kind of a blend of the two
19 previous, and I'll explain how it refers and how it
20 stands as a necessary improvement to its
21 predecessors.
22 Let me put this in perspective. On the
A Matter of Record (301) 890-4188 62
1 slide here, you can see the predecessor design.
2 This is the antidepressant plus placebo lead-in
3 design. So essentially, you have patients. You
4 take them into the study. You put them on an
5 antidepressant and adjunctive placebo. Those that
6 do not respond are randomized to have the
7 experimental treatment added -- back in the days
8 aripiprazole or, more recently, brexpiprazole -- or
9 adjunctive placebo.
10 This is a great design, but has a serious
11 limitation. The limitation here is that, depending
12 on what study you look at, half of patients that
13 come in or even up to three-quarters do not
14 generate any data about the compound's efficacy,
15 safety, or tolerability.
16 So you have patients coming in. They sign
17 consent with the understanding that they're going
18 to contribute to a potential new therapy, and their
19 data is tossed out. This is an inefficient way of
20 doing things. This really makes things more
21 difficult from a practical standpoint. And one
22 could argue if we can get away with -- if we can
A Matter of Record (301) 890-4188 63
1 evolve to use all patients' data, it would be a
2 more ethical, in my opinion, study design because
3 you would need fewer patients, and you would use
4 the data from every patient who's consented with
5 the intention of generating information that would
6 generate a new treatment.
7 The solution to this is a simple tweak.
8 Essentially, what you do is from the beginning, you
9 add the possibility, then in addition to the
10 antidepressant, patients get the experimental
11 treatment as well.
12 Doing this, every single patient that comes
13 into the study and randomized generates data about
14 efficacy, safety, and tolerability. The patients
15 that start in the study are retrospective partial
16 non-responders, more similar to the quetiapine
17 study. In clinical terms, this is similar to
18 someone being referred to your practice who is on
19 an antidepressant, who's depressed, and who hasn't
20 gotten better.
21 On the right-hand side, you then move on to
22 the more traditional design of 10 years ago, where
A Matter of Record (301) 890-4188 64
1 you have patients continuing through the study on
2 their antidepressant placebo and they are
3 prospectively confirmed as being partial non-
4 responders.
5 In clinical terms, this is more similar to
6 someone coming to your office, you treat them with
7 an antidepressant, you see what happens, and then
8 you decide on an adjunct. That is essentially the
9 change. The advantage here is that every patient
10 counts.
11 This is not a new approach. This approach,
12 first proposed in 2003, has been used not only in
13 major depressive disorder, in Alzheimer's, in
14 generalized anxiety disorder, in ADD and
15 schizophrenia. It has been used as a monotherapy
16 or an adjunct. It has been sponsored by companies,
17 by foundations like the Stanley Foundation, by the
18 NIMH several times. In fact, Dr. Mathews
19 conducted -- one of the speakers -- one of the
20 first SPCD NIMH studies and then many more. It's
21 recognized as a way of doing business.
22 So clinical trial challenges, we need to
A Matter of Record (301) 890-4188 65
1 move things along to get treatments for patients,
2 demonstrating efficacy. The way we're doing things
3 now is difficult, inefficient. We have to improve
4 the way we do things to deliver treatments to
5 patients who need them.
6 SPCD, which is essentially a blend of its
7 two predecessors that have been used in approval
8 process, in my mind is a more efficient way of
9 doing things and allows for every patient and
10 studies to count.
11 So with that, I'd like to conclude and
12 invite the next speaker, Jerry Schindler from
13 Alkermes, who is going to discuss the efficacy
14 dataset.
15 Jerry?
16 Applicant Presentation - Jerald Schindler
17 DR. SCHINDLER: Thank you, Dr. Papakostas.
18 I'm Jerry Schindler, and I lead the
19 biostatistics group at Alkermes. And what I'd like
20 to do is present the results of our four pivotal
21 trials of BUP/SAM for the treatment of major
22 depressive disorder.
A Matter of Record (301) 890-4188 66
1 These four pivotal trials were randomized,
2 double-blind, and placebo-controlled, and I'll
3 present data that show that 2 of the 4 trials met
4 their primary endpoint. One study just missed the
5 primary endpoint, but it still provided supportive
6 evidence of efficacy. And 1 of the 4 pivotal
7 trials did fail to meet its primary endpoint.
8 So first, an overview of the 4 pivotal
9 trials. The program consisted of 1 phase 2 and 3
10 phase 3 trials. There were 3 SPCD designs and
11 1 placebo run-in design, and a 2/2 dose of BUP/SAM
12 was evaluated in all studies. The primary
13 assessments were based on either the HAM-D 17 scale
14 or the MADRS scale.
15 Here's the SPCD design again, which you just
16 saw. The highlighted areas showed the two
17 randomizations, and the randomization in stage 1
18 was 2 to 2 to 9, with more patients being
19 randomized to placebo to allow enough patients from
20 stage 1 to roll over into stage 2. The
21 randomization in stage 2 was even, 1 to 1 to 1, to
22 evaluate the 3 groups in stage 2.
A Matter of Record (301) 890-4188 67
1 Remember, the formal statistical analyses
2 are based on the randomizations. So the formal
3 statistical analysis are the comparisons of the
4 groups that are randomized in stage 1 and stage 2
5 combine together.
6 Now, one advantage of the SPCD design is
7 that we can follow patients from stage 1 who
8 received BUP/SAM into stage 2. For these patients,
9 there's no formal comparison group, but we can
10 still follow them anyway and see what happens, and
11 I'll present some of the data for those patients as
12 well a little bit later.
13 First, a brief mention of the statistical
14 analysis. All of these studies were longitudinal
15 trials with weekly efficacy assessments. To
16 analyze the data, we used the MMRM model, which
17 uses all available longitudinal data without
18 imputation.
19 At each time point, the way this model works
20 is that each time point calculates the mean change
21 from baseline by treatment group and then compares
22 this change from baseline between the two groups,
A Matter of Record (301) 890-4188 68
1 between BUP/SAM and placebo. And the statistical
2 test is to determine if this difference is equal to
3 zero or not.
4 Now, remember many of these, 3 of the 4
5 studies, are SPCD designs. So for the SPCD design,
6 we calculate the difference from stage 1 and the
7 difference from stage 2, and these prespecified
8 weights to combine the differences across both
9 stages for the trial. And the SPCD design, as
10 mentioned before, uses data from all the subjects
11 in the study.
12 One way to look at the data from the SPCD
13 design is to look at the end of treatment, and
14 that's a single time point, but you can look at the
15 end of treatment, and that's often done. And you
16 look at the end of treatment at the end of stage 1
17 and compare the difference between BUP/SAM and
18 placebo at the end of stage 1. And then look at
19 the difference between BUP/SAM and placebo at the
20 end of stage 2, and then combine these two
21 differences using prespecified weights.
22 Now, this is looking at the data at a single
A Matter of Record (301) 890-4188 69
1 time point, at the end of therapy. But we're
2 really trying to improve the symptoms of depression
3 in these patients, and what we'd really like to do
4 is measure the effect over multiple time points.
5 We want to see what the benefit is for the
6 patient over the course of treatment rather than
7 just at that one single time point, so another way
8 to look at the data, which is actually a little bit
9 better way to look at the data, is to look at the
10 data over multiple time points. And in this way,
11 you do almost the same thing, but you do it over
12 multiple time points.
13 So at stage 1, you look at the difference
14 between BUP/SAM and placebo at multiple time
15 points, at different weeks, and calculate that
16 difference from the model to get the group
17 different estimate from that model, and then
18 combine those differences across stage 1 to get one
19 estimate for stage 1 and do the same thing for
20 stage 2.
21 Look at the differences between BUP/SAM and
22 placebo in stage 2 and then combine those estimates
A Matter of Record (301) 890-4188 70
1 at the group level, not at the patient level, but
2 at the group level from the model, combine those
3 estimates and then get one estimate for stage 2.
4 And now you have one estimate for stage 1 and one
5 estimate for stage 2, and then you can combine the
6 data the same way as before.
7 So statistically, the math is the same for
8 either way to do it, but what this does is it looks
9 at multiple time points and looks at the treatment
10 effect over multiple time points. And it really
11 provides a better estimate of the effect of the
12 drug as perceived by the patient over the course of
13 treatment. It also has the benefit of reducing
14 some of the week-to-week variability in the data so
15 that we get a more precise estimate of the
16 treatment effect.
17 Now, let's look a little bit at the overall
18 results. First, to orient you a little bit to this
19 slide, this slide shows the difference between
20 BUP/SAM 2/2 and placebo and their confidence
21 intervals for the primary endpoints for each of the
22 four pivotal trials. The zero vertical line shows
A Matter of Record (301) 890-4188 71
1 when the difference is equal to zero, when there's
2 no difference between the two groups. Data on the
3 left would favor BUP/SAM and data on the right
4 would favor placebo.
5 You can see that all the mean values are on
6 the left, which is the side that favors BUP/SAM,
7 and for two of the confidence intervals, for the
8 202 study and for the 207 study, the entire
9 confidence intervals are on the left-hand side and
10 do not cross zero. These p-values are significant,
11 so these two studies, 2 out of the 4 studies, met
12 their primary endpoint, 202 and 207.
13 Study 205, the one in the middle, even
14 though the mean value is on the left-hand side, the
15 confidence interval just barely crosses zero, and
16 the p-value is 0.1. So it was not significant,
17 didn't meet its primary endpoint, although there's
18 a suggestion that there might be something worth
19 looking at for that study. Study 206, the placebo
20 run-ins trial, clearly you can see the confidence
21 interval crosses zero; the p-value was not
22 significant 0.72.
A Matter of Record (301) 890-4188 72
1 What I'd like to do now is present the data
2 for each study one at a time. Study 202, as you've
3 heard already, was our first study. It was
4 originally a phase 2 study because it was the first
5 efficacy study, and it was an SPCD design. And it
6 was a randomized, double-blind, placebo-controlled
7 study, which was conducted under the same rigor as
8 any phase 3 design. It evaluated 2 doses of
9 BUP/SAM, 2/2 and 8/8. And the primary endpoint
10 here was HAM-D 17, but we also collected MADRS-10
11 as a secondary endpoint.
12 So here are the results for study 202. On
13 the top of this slide, just to orient you a little
14 bit to the slide, are the changes from baseline.
15 Gray is placebo and green is the group that
16 received BUP/SAM 2/2. And we show stage 1. It's
17 an SPCD design, so we show stage 1 and stage 2, and
18 the evaluation of treatment effect was at week 4.
19 Then below this, just so you know what's on
20 the slide, below this are the confidence intervals
21 and the mean estimates for the comparison between
22 each treatment group and placebo. You can see that
A Matter of Record (301) 890-4188 73
1 there's clear separation between the patients who
2 receive BUP/SAM, the green line, and the patients
3 who receive placebo, the gray line, in both stage 1
4 and stage 2, pretty much starting by week 3 and
5 week 4 in both groups.
6 When you look at the forest plot below, you
7 see the statistical comparison of the difference
8 between BUP/SAM 2/2 and placebo. And you see here
9 the p-value is 0.014, so this is clearly a
10 significant result.
11 When you look at the comparison between
12 BUP/SAM 8/8 and placebo, the p-value is 0.7, 0.699,
13 and the confidence interval clearly crosses zero,
14 so this was not significant.
15 As you've heard, the FDA has raised two
16 issues about this study, and I'd like to discuss
17 them now. The first is that there was not a
18 prespecified multiplicity adjustment, and the
19 second was that one single patient might be driving
20 these results. So what I'd like to do now is
21 address some of these issues.
22 One note about study 202 is that there was
A Matter of Record (301) 890-4188 74
1 no prespecified multiplicity adjustment. Just to
2 remind you what the p-values were from the previous
3 slide, when you look at the comparison between
4 BUP/SAM and placebo, the 2/2 dose, the p-value was
5 significant, 0.014, and the 8/8 dose, the p-value
6 is 0.699.
7 Now, there are only two treatment groups, so
8 to apply a multiplicity adjustment, we're just
9 doing multiplicity adjustment among two treatment
10 groups. One way to do this is a post hoc
11 Bonferroni multiplicity adjustment, which has the
12 effect of splitting the alpha in half. So that
13 changes the decision rule. It doesn't affect the
14 p-value. It changes the decision rule from 0.05 to
15 0.025.
16 The advantage of a Bonferroni adjustment is
17 that it provides two independent tests, so it
18 splits the alpha error in half, so you have
19 one half for one comparison, one half for the other
20 comparison. It's very conservative, which we all
21 know, because it splits the alpha in half, but it
22 also requires no prior assumptions.
A Matter of Record (301) 890-4188 75
1 So it's assumption free and provides these
2 two independent tests. And when we use a
3 Bonferroni multiplicity adjustment here, the
4 p-value of 0.014 is still significant, even with a
5 much more strict decision rule, using 0.025 in the
6 multiplicity adjustment.
7 Now, the FDA has suggested that potentially
8 a hierarchical multiplicity adjustment would be of
9 use here. Remember, this was the first efficacy
10 study in BUP/SAM, so we didn't have evidence that
11 either dose was better than the other dose at this
12 point. Usually, when you apply a hierarchical
13 adjustment, you usually have some information
14 before you do the study that one dose might be
15 better than the other.
16 We're also aware that there's a complex
17 pharmacology of BUP/SAM because we're combining an
18 agonist and an antagonist, and we had uncertain
19 dose response. We might have an inverted U-shaped
20 dose-response curve, where when you put the two
21 together you might actually see that the dose
22 response, rather than being increasing, as you
A Matter of Record (301) 890-4188 76
1 might expect in general, it may not be increasing.
2 It may be an umbrella shape or an inverted U-shaped
3 dose-response curve, in which case a hierarchical
4 adjustment would definitely not be useful because
5 we wouldn't know which one to put first.
6 So I don't see that there's any
7 justification for wanting to use a hierarchical
8 adjustment for this study, for the first efficacy
9 study of BUP/SAM.
10 Now, the other comment that FDA made was
11 that a single patient drives the results. And when
12 we look at the data, what we see is that the
13 patient that was selected as that single patient
14 was the best responder from the study. So what FDA
15 did is, looking at the data after the blind has
16 been broken, the analysis has been done, and you've
17 looked at the data, they've identified that patient
18 that was the best responder from study 202 and
19 taken that patient out.
20 Now you know when you take out the best
21 responder, two things are going to happen. One is,
22 the mean of the overall study is going to decrease
A Matter of Record (301) 890-4188 77
1 a little bit because you know the best responder is
2 going to be contributing a bit to that mean. So
3 the mean's going to decrease a little bit. You're
4 also going to lose power if you take that patient
5 out.
6 We also have looked into whether there's any
7 legitimate reason to take that patient out, and the
8 patient was appropriately randomized, appropriately
9 selected for the trial, and a pertinent study was
10 conducted in an appropriate, blinded, and unbiased
11 manner. And we've gone back, and reviewed, and
12 sent that documentation to the FDA. So there's no
13 legitimate reason to take that patient out.
14 Removing a patient, as we all know, would be
15 contrary to the intent-to-treat principle, so you
16 wouldn't do that in general as a way to analyze the
17 data. But it might be interesting to see what
18 happens when you do take a patient out, and you
19 look at the best responder, and do it as a type of
20 sensitivity analysis.
21 So what I'd like to do is talk about it in
22 terms of a sensitivity analysis, not so much as the
A Matter of Record (301) 890-4188 78
1 primary analysis, but talk about it as a
2 sensitivity analysis. So down below, I have a
3 little bit of the confidence intervals that I've
4 shown before, the primary analysis, which I just
5 showed, which is the mean value and the confidence
6 interval of the 2/2 dose versus placebo where the
7 p-value was 0.014. That was on the previous slide.
8 Now, from the FDA briefing book, their
9 information is they took the patient out, which
10 obviously has an effect on the mean, and you can
11 see the mean gets a little bit smaller, gets closer
12 to zero, and the confidence interval stretches a
13 little bit because you're reducing the power of the
14 study. All things you would expect; when you take
15 the biggest responder out of a study, totally
16 things that you'd expect, you'd reduce the power
17 and you'd affect the mean.
18 But that's not the appropriate way to do a
19 sensitivity analysis for this. If you want to see
20 what is the effect of the best responder, the best
21 way to do it is to take out the best responder and
22 the worst responder, take out the top and the
A Matter of Record (301) 890-4188 79
1 bottom on both ends, and then replace that person
2 that's missing, not decrease the size of the study,
3 keep the study the same size so you have the same
4 power, but replace those people with their nearest
5 neighbor.
6 So you're bringing it in a little bit, but
7 you're still doing an appropriate sensitivity
8 analysis. And that's what the trim-and-replace one
9 is, where you take the top and the bottom, the best
10 responder and the worst responder, and then replace
11 them with their nearest neighbor, and then do the
12 analysis again.
13 You see, even when you replace the data with
14 the nearest neighbors, you see basically the same
15 thing. The result doesn't change very much. So
16 the data really are not being driven by that one
17 patient, but they'll be driven by the other
18 patients in the trial. And the nearest neighbors
19 are equal substitutes for that one patient.
20 Another way to look at the data is to take
21 that patient out of the study, keep the sample size
22 the same, but eliminate all of their data. If you
A Matter of Record (301) 890-4188 80
1 want to do a sensitivity analysis of that data, you
2 eliminate all of their data and use multiple
3 imputation to fill in for the missing data. But
4 you still keep that person in the randomization
5 scheme because they're properly randomized, and you
6 want to make sure that you have the same amount of
7 power.
8 So when you look at it as multiple
9 imputation question -- you also do this as a
10 sensitivity analysis -- you still see the same
11 result. So the trim-and-replace method and the
12 multiple imputation method, two sensitivity
13 analyses, give essentially the same result as the
14 primary analysis, which we presented earlier.
15 What that means is that that one single
16 patient is not driving the result, that all of the
17 patients together are contributing to the result,
18 and that, overall, the result is robust.
19 So now, I'd like to look at the next study,
20 study 205. Study 205 was another SPCD design. It
21 looked at 2 doses, 2/2 and 0.5 and 0.5, and the
22 primary endpoint here was MADRS-10.
A Matter of Record (301) 890-4188 81
1 Here are the results for study 205, same
2 display as before with the changes from baseline in
3 stage 1 and stage 2 at the top, and green is the
4 BUP/SAM 2/2 dose, and below are the mean values and
5 the confidence intervals for the comparison for the
6 difference between BUP/SAM 2/2 and placebo and
7 BUP/SAM 0.5/0.5 and placebo at the bottom.
8 What you see here when you look at that is
9 that you see clear separation between the curves
10 across the time points really starting after week 2
11 in stage 1 as before and pretty much throughout all
12 of stage 2.
13 So you see clear separation between the
14 curves, that there's greater reduction in
15 depression severity for the group that receives
16 BUP/SAM 2/2 relative to placebo.
17 Now, our comparison for this study, our
18 prespecified primary endpoint, was the difference
19 at week 5. And if you look at stage 2, you see a
20 little bit of variability in the data, which you
21 know we're going to see in this type of trial. And
22 in week 5, the two curves comes together just a
A Matter of Record (301) 890-4188 82
1 little bit, especially in stage 2, and you can see
2 that on the graph.
3 When you do the analysis, you see exactly
4 the same thing. The primary endpoint is the change
5 at week 5. And you can see that the difference
6 between the two treatment groups, between BUP/SAM
7 2/2 and placebo, even though the mean value is on
8 the left-hand side, which favors BUP/SAM, the
9 confidence interval just crosses zero. So the
10 p-value is 0.1 and not significant.
11 You see the same thing, only it's even more
12 extreme. When you look at the 0.5/0.5 comparison,
13 the mean value is around zero of the difference and
14 the confidence interval clearly is on both sides,
15 and the p-value is 0.9. It's clearly not
16 significant for the 0.5/0.5 dose.
17 But the 2/2 dose just barely missed, and the
18 reason it missed, you see it's some variability in
19 the data. Especially you see the variability at
20 week 5, where the points come together.
21 Now, the FDA briefing book ignored study 205
22 completely because it didn't meet the primary
A Matter of Record (301) 890-4188 83
1 endpoint. Even though the difference was not
2 significant at that single time point at week 5,
3 the data clearly show improvement in depression
4 symptoms across all of both stage 1 and stage 2,
5 and the 2/2 dose clearly looks like there's some
6 difference from placebo in this graph.
7 So I believe that this study really does
8 provide supportive evidence of efficacy. It didn't
9 meet its primary endpoint, but it provides
10 supportive evidence of efficacy.
11 Now, we learned some things after we did
12 study 205. One of the things that we learned, when
13 we look at 205 and 202 together, we learned that we
14 saw a clear signal that there was some activity for
15 BUP/SAM 2/2. So we saw early indications of
16 efficacy for BUP/SAM 2/2.
17 We also realized that we had variability in
18 our data, and we wanted to address that, and we
19 also wanted to be able to address the patient's
20 experience going through the trials.
21 So there are two ways to do it, two ways to
22 look at the data. One way is to continue to look
A Matter of Record (301) 890-4188 84
1 at that single time point, and then we are at the
2 mercy of the variability of the data. The other
3 way to do it is to average the difference between
4 the two groups at multiple time points.
5 The advantage, as I mentioned earlier, when
6 you average the differences together, the advantage
7 you get is that you reduce some of the influence of
8 the week-to-week variability. You also get a more
9 precise estimate of the treatment effect. And most
10 importantly, you get the patient's experience over
11 time, so you see what happens from the experience
12 of the patient over multiple weeks rather than at
13 one single time point. We wanted to pursue that in
14 our future studies.
15 We also saw in the literature that there was
16 some question that MADRS-6 might be a more
17 sensitive measure of depression and might measure
18 the core symptoms of depression severity better
19 than MADRS-10. And MADRS-6 is just a subset of
20 MADRS-10, so if we collect MADRS-10 data, we get
21 MADRS-6 also. So we can look at both. We can look
22 at MADRS-10 because we have the data for that, and
A Matter of Record (301) 890-4188 85
1 then we can look at the subset of MADRS-10,
2 MADRS-6.
3 So we wanted to explore both MADRS-6 and
4 MADRS-10 in our future trial, so we did that in
5 study 207.
6 Here's study 207. It's another SPCD design.
7 It looked at 2 doses of BUP/SAM, 2/2 and 1/1, and
8 it collected MADRS-10, which also gave us MADRS-6.
9 So we had the ability to look at both MADRS-10 and
10 MADRS-6.
11 For this study, the basis for concluding
12 efficacy was based on MADRS-6, where the difference
13 was averaged over multiple time points, week 3
14 through end of treatment. That was the basis for
15 concluding efficacy, which was prespecified in the
16 protocol. But we also included MADRS-10 because we
17 collected that data, and we averaged those
18 differences over multiple weeks, the same weeks,
19 week 3 to the end of treatment. So we average the
20 differences from MADRS-10, and we included that as
21 an endpoint.
22 We also had MADRS-10, so we could look at
A Matter of Record (301) 890-4188 86
1 the MADRS-10 at that single time point at the end
2 of treatment. So we now had three endpoints that
3 we could look at, although in our protocol, we
4 prespecified the basis for concluding efficacy
5 would be MADRS-6, where the difference was averaged
6 over these multiple time points.
7 We did that in a hierarchical manner,
8 looking at BUP/SAM 2/2 first, and then looking at
9 the difference between 1/1 and placebo second. And
10 we did a hierarchical. We used a hierarchical
11 multiplicity adjustment here because we already had
12 evidence that the 2/2 dose seemed to be effective,
13 and we wanted to get more evidence of that.
14 So here, a hierarchical multiplicity made
15 sense where it did not make sense in the 202 study.
16 Here, it does make sense, so we used a hierarchical
17 multiplicity adjustment.
18 Here's the data for study 207, same format
19 as before, where you see the data, the change from
20 baseline at the top for stage 1 and stage 2. And
21 here, you see clear separation between the two
22 groups. The green is BUP/SAM 2/2. The gray is
A Matter of Record (301) 890-4188 87
1 placebo. And you see clear separation across all
2 of the time points in stage 1, across all the time
3 points in stage 2. The highlighted areas are the
4 weeks that we incorporated into the analysis. We
5 averaged the difference at those time points and
6 included those in the analysis.
7 At the bottom, you see the hierarchy of the
8 endpoints. The top one is MADRS-6, which was
9 averaged over multiple time points, and the p-value
10 for that one was 0.015. So that was the basis for
11 concluding efficacy, so right away we know that
12 this was a positive trial. It met its primary
13 endpoint.
14 But we also wanted to continue the path down
15 the hierarchy, so the next step in the hierarchy is
16 MADRS-10, looking at the average difference for
17 multiple time points, so MADRS-10, the average over
18 multiple time points. And the p-value for that was
19 also significant, 0.026. So both of these showed a
20 significant difference from placebo for the 2/2
21 dose.
22 When we keep stepping through the hierarchy,
A Matter of Record (301) 890-4188 88
1 MADRS-10 at the end of treatment just barely
2 missed, .07, and the p-value there extended a
3 little bit over zero, and the p-value was not
4 significant. So that stops our pathway through the
5 hierarchy.
6 When we continue to look at the other data,
7 because we can do it and we have it on the slide,
8 the BUP/SAM 1/1 dose would not have been
9 significant anyway, but we never got to that in the
10 hierarchy. The only ones we got to were the top 3,
11 and then we stopped when we got to MADRS-10 end of
12 treatment. But the top 2 were significant, so this
13 study clearly met its primary endpoint.
14 Now, when you look at the data, especially
15 look at stage 1, we know we have variability in the
16 data. And if you look at the data in stage 1 at
17 week 5, you see a slight blip in the data,
18 especially for the BUP/SAM group, where the curves
19 start to come together a little bit. Also, the
20 BUP/SAM group starts to look more like placebo.
21 You might have the question, is this
22 indicative of potential waning of effect? The
A Matter of Record (301) 890-4188 89
1 treatment effect comes down for a few weeks and
2 then starts to go back up. And that was something
3 that we wanted to explore, and we can do it with
4 the SPCD design.
5 In this slide, we actually followed these
6 same patients into stage 2. What we're looking
7 at -- and just to remind you what you're looking
8 at -- the gray line is the placebo; the green line
9 is the patients who receive BUP/SAM 2/2, and
10 followed them from stage 1 to stage 2.
11 Now remember, the people who received
12 placebo in stage 1, many of them are re-randomized
13 into stage 2. So we're only showing the data for
14 placebo up until the end of stage 1. But the
15 patients who received BUP/SAM 2/2 in stage 1
16 continued to receive BUP/SAM in stage 2.
17 This is all blinded. The patients aren't
18 aware of their treatment, and they're not aware of
19 where they are in the study. So this is all
20 completely blind data. It's unblinded now, but it
21 was unblinded during the course of the study.
22 You can see that there is that little blip
A Matter of Record (301) 890-4188 90
1 at week 5, but then that data then come back and
2 return, and you can see a consistent reduction in
3 depression severity all the way through stage 2.
4 So this blip was just a little bit of statistical
5 wobble, and then it returned back to the path that
6 it was seeing before. And it is not any indication
7 of any waning of effect.
8 Now, we did another study as well.
9 Study 206 was a placebo run-in design, different
10 design, placebo run-in design, and it looked at
11 1 dose of BUP/SAM, the 2/2 dose versus placebo.
12 The primary endpoint for this study was MADRS-10,
13 evaluated at the end of treatment.
14 Here are the results for study 206. Here,
15 you can see that the two lines are almost exactly
16 on top of each other, and the difference between
17 BUP/SAM and placebo is almost zero. And the
18 confidence interval clearly crosses zero. So this
19 study was not significant. This study did not meet
20 its primary endpoint.
21 Now, one way to compare the data across all
22 four studies is to use a common endpoint, and what
A Matter of Record (301) 890-4188 91
1 I'd like to do is compare these studies and look at
2 all four studies. The advantage that we have is
3 that all four studies included weekly MADRS-10
4 assessments. So we collected MADRS-10 in every
5 study, and we can use MADRS-10 to look at the data
6 across all four studies.
7 Now, as I mentioned before, there are two
8 ways to look at MADRS-10 in a statistical way. The
9 one way, which we prefer and really is probably the
10 better way, is to look at MADRS-10 based on the
11 difference between the two groups at multiple time
12 points.
13 As I mentioned before, you calculate the
14 difference at multiple time points from the model
15 and average them together to get the overall
16 estimate of the mean effect. What this does is it
17 reduces the impact of the week-to-week variability
18 of the data and also better reflects the patient's
19 experience over time.
20 The other way to look at the data is the
21 more traditional way, to just look at the end of
22 treatment, just that single time point. And we can
A Matter of Record (301) 890-4188 92
1 do both. We've got the data. We've got MADRS-10
2 data, and we can look at both. And the end of
3 treatment, as I mentioned, is a more conventional
4 way to look at the data, but it has some of these
5 other issues.
6 So just to look at the data again, this is
7 MADRS-10 data for these three studies; study 202,
8 205, and 207; stage 1 first and then stage 2 for
9 each study, so you see 6 randomizations.
10 What you see is that all three studies show
11 clear separation between the treatment groups, and
12 they show a clear improvement in depression
13 symptoms for the BUP/SAM 2/2 dose relative to
14 placebo. And you can see that in every single
15 study, in every single randomization.
16 It's remarkably consistent across all six
17 randomizations. And this consistency is really
18 what's driving the results. It's not driven by
19 just one patient, as you may have heard in the
20 earlier presentation. It's not driven by one
21 patient. It's driven by all the patients. It's
22 driven by many of the patients. It's driven by the
A Matter of Record (301) 890-4188 93
1 patients in all of these trials. And we see
2 similar effects in each stage and for each study
3 across the entire program.
4 So when we look at the statistical analyses
5 of the data -- what you saw before was the data,
6 looking at MADRS-10. So here's the statistical
7 analysis of the MADRS-10 data. And this analysis
8 looks at MADRS-10, where we average the difference
9 over multiple time points.
10 Here, we see that 3 of the 4 studies show
11 confidence intervals that don't cross zero, so here
12 now, 202, 205, and 207, three, when we look at
13 MADRS-10, average over multiple time points, and
14 confidence intervals don't cross zero.
15 The 206 study, which I showed you that
16 didn't meet its primary endpoint, gives you the
17 same result here. When you look at MADRS-10
18 average over multiple time points, 206 was also not
19 significant, but 3 of the 4 were.
20 One way that is interesting to look at the
21 data is to do a meta-analysis of the data. We
22 recognize that three of the studies are SPCD
A Matter of Record (301) 890-4188 94
1 design. One's a placebo run-in. So it's not a
2 perfect meta-analysis. And we're not using this as
3 the primary endpoint of our trial, but it's still
4 interesting because we have the data. It's
5 interesting to look at what you get when you put
6 all the data together and what kind of pattern do
7 you see. So we have that for comparison.
8 What you see when you look at the
9 meta-analysis, looking at MADRS-10 average, what
10 you see is that the mean value is clearly on the
11 side that favors BUP/SAM, and the confidence
12 interval for the meta-analysis also doesn't cross
13 zero, also clearly it doesn't cross zero, and also
14 is on the side that favors BUP/SAM relative to
15 placebo.
16 So all of these taken together, the MADRS-10
17 average, looked for in each of these studies, and
18 the meta-analysis show that the 2/2 dose is better
19 than placebo. But now, there's that other way to
20 look at the data, looking just at the end of
21 treatment, so we have that as well.
22 When you look at the data just at the end of
A Matter of Record (301) 890-4188 95
1 treatment, and here are the confidence intervals
2 for that with the mean values and the confidence
3 intervals, here you see that 2 of the 4 studies
4 show confidence intervals that don't cross zero.
5 Study 202 and 205, the confidence intervals don't
6 cross zero. 207, the confidence interval just
7 barely crosses zero.
8 Remember, we saw before that there's a
9 little blip at week 5, and that's what's
10 responsible here for that confidence interval
11 crossing zero; and 206, again, the placebo run-in
12 design that failed to meet its primary endpoint is
13 not significant here, too.
14 The same story with a meta-analysis is used
15 as a type of sensitivity or supportive analysis.
16 It's really not the primary analysis, but we have
17 four studies, and we want to show what happens when
18 you combine them all into one statistic. So the
19 meta-analysis is not our primary analysis for this
20 program at all, but it really is a comparison of
21 how do we bring all the data together, even though
22 we have SPCD designs in parallel and a placebo
A Matter of Record (301) 890-4188 96
1 run-in design.
2 So we don't have a perfect meta-analysis,
3 but it's still something that's interesting to take
4 a look at. We see the same result as we saw
5 before, that the mean value is on the side that
6 favors BUP/SAM, and the confidence interval here
7 doesn't cross zero. So here's another indication
8 that the BUP/SAM 2/2 dose is better than placebo.
9 So overall, what did we learn from our
10 program? Overall, these trials showed that the
11 BUP/SAM 2/2 dose is effective for the adjunctive
12 treatment of depression. I've just shown you these
13 slides before, the data before. What I'd like to
14 show is that we've looked at the data many
15 different ways.
16 First, at the top, we looked at the
17 prespecified primary endpoints, and these were the
18 endpoints that were prespecified in the protocol.
19 And when we look at the prespecified primary
20 endpoints, 2 of the 4 trials met their primary
21 endpoint. And we see strong evidence of efficacy
22 from a third study, study 205, which we mentioned
A Matter of Record (301) 890-4188 97
1 earlier.
2 When we look at MADRS-10 difference, which
3 is averaged across multiple time points, we see
4 that 3 of the 4 studies demonstrate evidence of
5 efficacy. And then, when we look at MADRS-10 at
6 the end of treatment, at that single time point, we
7 see that 2 of the 4 studies show evidence of
8 efficacy favoring BUP/SAM over placebo.
9 Taken together from every direction, we
10 presented data that showed that the 2/2 dose of
11 BUP/SAM is effective for the adjunctive treatment
12 of major depressive disorder.
13 Now, I'd like to introduce Dr. Gary
14 Bloomgren, who will talk about drug safety.
15 Applicant Presentation - Gary Bloomgren
16 DR. BLOOMGREN: Thank you, Dr. Schindler.
17 Good morning, everyone. I'm Gary Bloomgren,
18 head of drug safety at Alkermes, and I'm going to
19 be providing the overview of the clinical safety
20 program of BUP/SAM.
21 Our integrated assessment of the safety
22 profile of BUP/SAM is based on data that we
A Matter of Record (301) 890-4188 98
1 collected from over 2,100 subjects treated with at
2 least 1 dose of BUP/SAM. And of that, more than
3 1500 included patients with major depressive
4 disorder treated on the 2/2 dose. And of those,
5 more than 700 received treatment for in excess of
6 12 months.
7 This totals greater than 1100 patient-years
8 of MDD patient exposure, which I'll be sharing
9 details on as we go through the presentation.
10 Now, the safety data was pooled across
11 4 placebo-controlled studies, with each study
12 having 2 randomizations, the first stage 1 at the
13 beginning of treatment. And the second mid-study,
14 when placebo non-responders were re-randomized for
15 stage 2, the data that I'll be sharing with you on
16 a number of tables that follow are summarized by
17 those randomization time points.
18 Treatment-emergent adverse events across the
19 development program were generally tolerability
20 related. And what you see in this table is the
21 treatment-emergent adverse events for BUP/SAM 2/2
22 that occurred at an incidence greater than or equal
A Matter of Record (301) 890-4188 99
1 to 5 percent and greater than placebo. And what
2 you see is the stage 1 to the left and then the
3 stage 2 randomizations and the data.
4 What you see is the rates of the TEAs were
5 less in stage 2 than in stage 1, but the events
6 were generally similar sorts of events. The TEAs
7 that we saw were generally gastrointestinal related
8 or associated with sedation.
9 Across the program, the majority of adverse
10 events were mild to moderate in severity and tended
11 to occur with treatment initiation. We saw no
12 meaningful differences by gender, age, race,
13 concomitant antidepressant used, or benzodiazepine
14 use subgroup. And there are no new findings that
15 we saw within the long-term study that went over a
16 year duration.
17 There were a few adverse events that led to
18 discontinuation across the program. And in this
19 table, you see those treatment-related adverse
20 events that led to discontinuation in an incidence
21 of greater than 2 percent. And again, with
22 stage 1, the incidence was a little higher with
A Matter of Record (301) 890-4188 100
1 BUP/SAM, 13.6 percent of subjects, compared with 2
2 percent on placebo, a significantly lower incidence
3 of 3.8 and 1.4, respectively, for BUP/SAM and
4 placebo in stage 2.
5 There were only 3 adverse events that
6 occurred in an incidence of greater than 2 percent
7 that led to discontinuation, and that included
8 nausea, vomiting, and dizziness shown here, and
9 that was only in the stage 1. We saw similar
10 findings in the long-term study, and that appeared
11 very similar to what we see here with stage 1.
12 There are few serious adverse events across
13 the development program. In the placebo-controlled
14 studies, 1.9 percent of the BUP/SAM versus 0.5 of
15 the placebo patients had an SAE. There was no
16 pattern to these SAEs and there were no deaths.
17 In the 1-year long-term study, 3.2 percent
18 of patients had an SAE. The most common events
19 were depression and suicidal ideation, each which
20 occurred at an incidence of 0.2 percent.
21 There were 2 deaths in the long-term study,
22 both assessed by the principal investigator as not
A Matter of Record (301) 890-4188 101
1 related. The first was a patient with chronic
2 obstructive pulmonary disease, who died of
3 respiratory arrest 47 days after the last dose of
4 drug, and the second was a patient with
5 hypertension and congestive heart failure who died
6 of a cerebral hemorrhage on day 87 of study
7 treatment.
8 There were no clinically relevant laboratory
9 vital sign, weight or ECG changes in either the
10 placebo-controlled or long-term studies. We saw no
11 meaningful post-baseline changes or outliers of
12 major clinical significance. And there was no
13 evidence of risk of QT prolongation associated with
14 BUP/SAM in a dedicated TQT study.
15 I'm going to talk now about some topics of
16 special interest, and they really come from three
17 areas; one either safety topics that were
18 identified within the BUP/SAM program itself, those
19 that are associated with buprenorphine as a safety
20 concern by itself, or by other antidepressant
21 treatments.
22 First, as it relates to the topics of
A Matter of Record (301) 890-4188 102
1 special interest associated with the BUP/SAM
2 program, we saw CNS sedation as a major finding.
3 This was mild to moderate in nature, typically was
4 associated with treatment initiation, and resolved
5 with continued use.
6 The other was one case of acute opioid
7 withdrawal, which was precipitated with the first
8 dose of BUP/SAM in a patient who had undisclosed
9 pre-existing opioid dependence. This event was
10 serious and assessed as related to treatment, and
11 attributed to samidorphan, the mu opioid antagonist
12 that's part of BUP/SAM, and is the reason why we
13 are contraindicating its use in any population of
14 patients who would either be chronically using
15 opioids or are opioid dependent.
16 Additional topics of special interest
17 included those topics that are associated with
18 buprenorphine alone, and that includes respiratory
19 depression, hypotension, orthostatic hypotension,
20 and hepatic injury. And we saw no evidence by
21 review of adverse events, changes in vital signs
22 from baseline, or laboratory tests as appropriate
A Matter of Record (301) 890-4188 103
1 that there was any change in this group relative to
2 the placebo plus ADT cohort.
3 Additionally, when we looked for items that
4 are potential safety concerns with other
5 antidepressant agents, including hypomania, mania,
6 sexual dysfunction, or suicidal ideation or
7 behavior, we again saw no difference in terms of
8 the AE profiles that we saw on treatment with
9 BUP/SAM plus ADT versus what we saw with placebo
10 plus ADT.
11 We also did a very thorough review looking
12 for suicidal ideation and behavior using the
13 Columbia-Suicide Severity Rating Scale or the
14 CSSRS, and what you see here is we saw less
15 suicidal ideation and behavior with BUP/SAM
16 compared with placebo. And this occurred with both
17 randomizations and had a similar finding in a
18 long-term study.
19 What you see in this table are the post-
20 baseline CSSRS events on treatment for each of the
21 stages for the suicidal ideation, suicidal
22 behavior, and the self-injurious behavior without
A Matter of Record (301) 890-4188 104
1 suicidal intent.
2 What you see is, numerically, these events
3 are less on the BUP/SAM 2/2 than on placebo in both
4 randomizations and that the prevalence of these
5 events in a 1-year long-term study were very
6 similar to what we saw with the BUP/SAM in the
7 controlled studies.
8 Because we added samidorphan specifically to
9 buprenorphine to mitigate the abuse potential, this
10 was an assessment of abuse potential. It was a
11 critical part of the development program, and I'll
12 be sharing the data that we have collected across
13 that as a part of our integrated assessment.
14 This included a dedicated human abuse
15 potential study, which I'll be going into more
16 detail shortly, and it also included, over 1500
17 patients with major depressive disorder that we
18 studied, an evaluation of adverse events of special
19 interest that were related to abuse potential,
20 dependence, and withdrawal. We also did an
21 objective assessment of withdrawal using the COWS
22 or Clinical Opioid Withdrawal Scale following
A Matter of Record (301) 890-4188 105
1 abrupt discontinuation of BUP/SAM.
2 The dedicated human abuse potential study;
3 our assessment was that the abuse potential is low.
4 And what you see here in the human abuse potential
5 study is a double-blinded, cross-over, 6-way study
6 using non-dependent recreational opioid users and
7 assessed treatments with placebo: the 2/2 dose,
8 the indicated dose, supratherapeutic doses of
9 BUP/SAM 4 and 8 times greater than the daily dose
10 for BUP/SAM, as well as buprenorphine alone, 8 and
11 16 milligrams, which were used as positive
12 controls.
13 The primary endpoint of this study is, at
14 the moment, drug liking difference from placebo.
15 And what you see with the dotted line at the 11
16 within the figure is the margin of clinical
17 significance for BUP/SAM versus placebo.
18 This is established through multiple HAP
19 studies looking at the placebo treatment groups and
20 looking at what the upper margin of significance
21 for placebo was across those studies. So anything
22 to the left of that dotted line is consistent with
A Matter of Record (301) 890-4188 106
1 what can be seen with placebo.
2 What you see is the 2/2 dose is consistent
3 with that of placebo. And even the
4 supratherapeutic doses, the 8/8 and 16/16, were
5 slightly greater than placebo but substantially
6 less than that of the buprenorphine corresponding
7 dose alone that was part of a positive control.
8 We also had a number of secondary endpoints
9 that were part of this study, which included
10 overall drug-liking and take-drug-again
11 assessments. Both of these endpoints are highly
12 predictive of real-world abuse liability, and it
13 shows that BUP/SAM is similar to placebo. And
14 that's not only with the 2/2 dose, but even the
15 supratherapeutic voices. And this is substantially
16 different from what we see with the active positive
17 control of BUP alone, further supporting our
18 assessment that there's a low abuse potential with
19 BUP/SAM.
20 We also saw consistent evidence of low abuse
21 potential across the MDD studies, and this, again,
22 is with over 1500 patients with MDD. We curated
A Matter of Record (301) 890-4188 107
1 that dataset for abuse potential terms. And the
2 majority of the events were nonspecific to abuse
3 potential. This included dizziness, somnolence,
4 sedation.
5 Although these can be seen with abuse
6 potential drugs, they're commonly seen with drugs
7 that have no abuse potential at all. What's more
8 important is there was a low incidence of euphoria-
9 related events. With BUP/SAM 2/2, the incidence
10 was 1.6 percent versus placebo at 0.2 percent, and
11 that's across stage 1 and stage 2 combined.
12 The incidence in the long-term study -- now,
13 this is a year-duration study. The incidence was
14 very similar, 1.2 percent. The majority of these
15 events were associated with the first dose of
16 treatment, and none of them reoccurred.
17 The other important thing to note is that we
18 did not see a dose effect within the MDD
19 population. Half of the events of euphoria
20 occurred with the initiation with a 0.5/0.5 dose
21 with titration, the other half with a 2/2 dose.
22 And there were none on the 8/8 dose. We saw no
A Matter of Record (301) 890-4188 108
1 abuse behavior and no evidence of dependence across
2 the development program.
3 We also did a thorough assessment for
4 withdrawal and saw little evidence of withdrawal.
5 And in these analyses, we took all patients that
6 had a minimum of 4 weeks being on treatment, and
7 then looked at the 2 weeks following abrupt
8 withdrawal of treatment to see if we saw any
9 evidence of withdrawal, and this slide summarizes
10 those findings.
11 Our mean post-discontinuation scores in that
12 2-weeks following discontinuation in that
13 population, was less than or equal to 1 on the COWS
14 scores. COWS scores, no withdraw, are from 0 to 4,
15 and that occurred in all treatment groups, whether
16 it be placebo or active. And that was also seen in
17 the long-term study.
18 If we look at the shift and mainly at the
19 outliers that are really summarized, then, within
20 the COWS scores table below, you see that in the
21 placebo-controlled studies, BUP/SAM 2/2 and
22 placebo, that 96 and 97 percent of patients had no
A Matter of Record (301) 890-4188 109
1 withdrawal; 2.7 of both BUP/SAM 2/2 and placebo had
2 mild withdrawal. And there was 1 patient which
3 made up the 0.9 percent in the controlled studies
4 that had moderate withdrawal by COWS score.
5 In the long-term study, we saw similar
6 findings. 94 percent of patients had no withdrawal
7 by COWS score, and 4.9 and 0.7, respectively, had
8 mild or moderate withdrawal by COWS scores.
9 Now, we looked at these 5.6 percent, the 4.9
10 plus the 0.7 patients, to see what we could assess
11 clinically as to the meaningfulness of this. And
12 what we found is that this accounts for 47 patients
13 within the study. And within these 47 patients, 44
14 of those 47 had no or either required any
15 prescription medication treatment whatsoever.
16 There were 3 patients out of the 47 that
17 made up that 5.6 percent, and those 2 had anxiety;
18 1 had insomnia. Each of them required a
19 benzodiazepine prescription, and I think puts in
20 some context, at least, with what we did see in
21 terms of findings as it relates to potential
22 withdrawals, that this is easily medically managed.
A Matter of Record (301) 890-4188 110
1 In summary, we feel that we've done a very
2 thorough assessment of the safety profile of
3 BUP/SAM. The common AEs we saw were
4 gastrointestinal and sedation related, mild to
5 moderate in severity, and they typically occurred
6 with treatment initiation.
7 There was no clinically meaningful changes
8 in laboratory vital signs, weight, or ECG changes
9 across the program, and we saw no evidence of
10 increased treatment-emergent suicidal ideation or
11 behavior.
12 Additionally, as it relates to our
13 assessments across many facets for abuse potential,
14 our assessment is the abuse potential of BUP/SAM is
15 low. In the human abuse potential study, the abuse
16 potential of BUP/SAM 2/2 is similar to placebo, and
17 even supratherapeutic doses were slightly greater
18 than placebo, but significantly less than the
19 equivalent dose of the positive control for BUP
20 alone.
21 Across the 1500 patients with MDD that we
22 studied, the data was consistent with these
A Matter of Record (301) 890-4188 111
1 findings. There was a low incidence of euphoria
2 typically with the first dose, none of which were
3 recurrent, and there was no evidence of dependence
4 during treatment and little evidence of withdrawal
5 even with abrupt discontinuation, and that that
6 discontinuation was well tolerated.
7 Now, as I've shared with you over the last
8 several slides, our assessment of the abuse
9 potential of BUP/SAM is low. But that said,
10 BUP/SAM contains buprenorphine, and particularly
11 during an opioid crisis, we want to ensure and we
12 are committed to make every effort that BUP/SAM is
13 used in an informed and appropriate manner.
14 This starts with how we've developed the
15 product, and by selecting the lowest effective dose
16 of BUP 2 milligrams and added samidorphan to
17 mitigate that abuse potential, we've co-formulated
18 the tablet as a monolayered tablet that's
19 micronized to a homogenous mixture of BUP/SAM that
20 cannot be mechanically separated.
21 The tablets are packaged in cards in
22 individual blisters, as Dr. von Moltke had shared
A Matter of Record (301) 890-4188 112
1 with you a photo of earlier in the presentation.
2 And this is using F1 packaging, which is the
3 highest quality packaging, to limit unintentional
4 pediatric exposure.
5 We're committed to continuous monitoring of
6 the distribution system from manufacturer, to
7 wholesaler to pharmacy, to monitor for any evidence
8 off any suspicious activity that might suggest
9 diversion.
10 We're also very much committed to educating
11 healthcare providers and patients to the
12 appropriate use of this product in the marketplace.
13 And we're also committed to implementing additional
14 post-marketing safety initiatives, which includes a
15 REMS, which I'll be speaking to briefly.
16 A centerpiece to our risk mitigation is
17 education. The objective of this is to alert
18 healthcare professionals and patients to important
19 safety information and appropriate use. This
20 includes the risks and precautions associated with
21 concomitant use of opioids, either prior to
22 considering initiating treatment as well as during
A Matter of Record (301) 890-4188 113
1 treatment.
2 It includes instructions as it relates to
3 safe use, storage, and disposal, and because
4 BUP/SAM contains buprenorphine, also alerting to
5 the risks of abuse, misuse, diversion, addiction,
6 overdose, and death that can occur with opioids.
7 As it relates to information to go to
8 healthcare professionals on the left panel, that
9 information will be predicated on the details that
10 are part of the product label as well as the
11 product REMS. And it will include selection of
12 appropriate patients for this sort of therapy, but
13 also educational materials that include patient
14 counseling and a placebo training tool, since this
15 would be the first sublingual delivered medication
16 in this sort of a population.
17 It includes a commitment for trained medical
18 staff available to respond to queries via a call
19 center or in person and a website and interactive
20 web-based training for physicians, whether they be
21 in rural or urban areas. And as it relates to
22 patients, a medication guide and a medication I.D.
A Matter of Record (301) 890-4188 114
1 wallet card that ensures that they have the
2 appropriate information so that they can share this
3 not only with family members, but also with other
4 healthcare professionals, who may be treating them
5 with other medications.
6 It also includes additional product website
7 materials to help educate patients around the
8 important risks and precautions that are associated
9 with this therapy.
10 As it relates to a risk evaluation and
11 mitigation strategy or REMS program, we've done a
12 lot of thinking about this, and we've modeled this
13 after other buprenorphine-containing therapies.
14 The objectives of this REMS would be to mitigate
15 the risk of misuse, which includes abuse and
16 accidental exposure.
17 It includes a medication guide for patients,
18 a REMS website, and call center for healthcare
19 professionals, communication materials, which
20 include the healthcare HCP letter which would go
21 out at the time of a product launch, to ensure that
22 we get out in front of prescribers with this
A Matter of Record (301) 890-4188 115
1 important information, HCP brochure and appropriate
2 use checklist.
3 We're also committing to product-specific
4 active monitoring for misuse, abuse, dependence, or
5 diversion using RADARS. This is a nationally
6 accepted, multifaceted approach for monitoring
7 that's used by FDA as well as many other
8 manufacturers in this space to assess for any
9 evidence of end user inappropriate use of BUP/SAM
10 specifically.
11 We're committed to identifying new
12 healthcare professionals on an annual basis and
13 ensuring that they have the information that I've
14 shared with you above. And of course, with any
15 REMS, we will be providing periodic assessments of
16 the effectiveness of the REMS. And based on new
17 data, if there is new information that suggests
18 that we need to take a different stance, that will
19 be a data-driven discussion and decision that we
20 will be making with the FDA to make sure that the
21 product is used appropriately.
22 Now, I'd like to turn the next stage of the
A Matter of Record (301) 890-4188 116
1 presentation over to Dr. Sanjay Mathew.
2 Applicant Presentation - Sanjay Mathew
3 DR. MATHEW: Thank you, Dr. Bloomgren.
4 My name is Sanjay Mathew. I'm a professor
5 of psychiatry at Baylor College of Medicine, and I
6 direct the mood and anxiety disorders program. And
7 I'm here today as a consultant to Alkermes. I've
8 been compensated for time and travel.
9 I'll be speaking as a practicing
10 psychiatrist from a clinical perspective and give
11 my assessment of the benefit-risk profile for
12 BUP/SAM, and in more general terms, who would be
13 an appropriate candidate for an adjunctive therapy.
14 These would be patients usually with
15 longstanding depression. They may have had
16 multiple episodes. They've had little or no
17 success with successive monotherapy
18 antidepressants, perhaps from a variety of classes,
19 SSRIs, SNRIs, bupropion, despite being dosed at
20 adequate doses and for adequate durations.
21 The current antidepressant that they're on
22 has helped them, but certainly has not enabled them
A Matter of Record (301) 890-4188 117
1 to achieve remission. And they have persistent
2 symptoms that go beyond the 9 symptoms of a major
3 depressive episode, to include significant
4 impairment in their social, occupational, and just
5 in their family functioning.
6 Finally, a patient has to be willing to
7 consider adding on another therapy. They've tried
8 perhaps several monotherapies, but now it's time
9 perhaps to add something on to the existing
10 antidepressant.
11 So a typical case for a person where BUP/SAM
12 may be considered would be a patient like J.D.
13 She's in her mid-40s. She's experiencing her
14 second major depressive episode. Her first episode
15 was in her 20s, and she responded at that time to
16 an SSRI, fluoxetine.
17 She's been in this episode for approximately
18 9 months. The first trial she had was an SSRI,
19 escitalopram, which was pushed to the FDA max dose
20 of 20 milligrams for about 3 months. And at
21 baseline, she had a high very severe score on the
22 Quick Inventory of Depressive Symptoms scale, which
A Matter of Record (301) 890-4188 118
1 is a self-report instrument for depression, a score
2 of 21.
3 So with the escitalopram, she did have some
4 improvement. Scores came down to 16. But that's
5 still a severe score. So at that point, she was
6 switched to an SNRI, venlafaxine, and that dose was
7 pushed to a fairly high dose, 300 milligrams, for
8 about 3 months, and her scores hovered at the 15-16
9 range, so there was no significant improvement, and
10 she had some side effects with the venlafaxine.
11 We discussed what would be the options at
12 this point, given the fact she had persistent
13 depressive symptoms that continued to impact her
14 ability to function. She's a mother of two
15 school-aged children. She really became socially
16 isolated, stopped going to church. She stopped
17 preparing their lunches, and really stopped
18 engaging with family and friends in her community.
19 So beyond the depressive symptoms, there was this
20 marked social withdrawal.
21 When we think about what is available for
22 patients like J.D., we think of the atypical
A Matter of Record (301) 890-4188 119
1 antipsychotics. These are the on-label options.
2 The most recently approved one was brexpiprazole,
3 and here is a graph just looking at the comparison
4 of the effect sizes between BUP/SAM across the
5 4 studies that were presented this morning and
6 brexpiprazole, essentially a comparable effect
7 size. And BUP/SAM does so through an entirely
8 different mechanism, so it's not a "me too"
9 approach. It's certainly not a monoaminergic
10 approach.
11 So we would consider initially augmentation
12 with what's available; aripiprazole, quetiapine,
13 perhaps brexpiprazole. However, in this case, one
14 of the main concerns for J.D. is she absolutely
15 does not want to have an agent that could
16 contribute to weight gain. She's struggled with
17 weight most of her life. Her current BMI is 30,
18 and she's a borderline diabetic. She was also
19 worried about daytime sleepiness and somnolence, as
20 one of the symptoms is residual fatigue.
21 So looking overall at the safety summary
22 from these studies that were presented, it appeared
A Matter of Record (301) 890-4188 120
1 the common adverse events were GI or sedation. The
2 sedation appeared to be limited and usually upon
3 treatment initiation, and was not a reason for
4 drop-out. And these were generally mild or
5 moderate in severity.
6 The low potential for abuse, of course, is
7 an important consideration, and this was evidence
8 from both the human abuse potential study and from
9 the patients with MDD, both short-term as well as
10 the long-term 12-month study.
11 So overall, my assessment of the benefit-
12 risk profile in terms of what are the risks; I
13 mean, clearly, with buprenorphine, has that issue
14 been mitigated with the presence of samidorphan,
15 the mu antagonist, and it appears it has. The
16 abuse liability significantly mitigated.
17 The adverse effects that were reported in
18 the studies, mainly G.I., appeared to be
19 manageable, transient, and did not lead to
20 significant drop-outs. There was a lack of weight
21 gain, metabolism impact, and movement side effects,
22 which certainly would present a contrast to the
A Matter of Record (301) 890-4188 121
1 existing therapies.
2 Then with any antidepressant therapy, you're
3 worried about induction of hypomania and
4 suicidality, and that did not appear to be the case
5 in the studies.
6 In the benefit considerations, of course you
7 want something that's effective robustly, and the
8 studies here showed consistency of those effects
9 across 3 of the 4 studies, clinically meaningful
10 efficacy in these difficult-to-treat patients,
11 recognizing these patients have had multiple
12 episodes, have been in this episode for many
13 months, and have high baseline scores on the MADRS,
14 and then the efficacy consistent as mentioned with
15 what's available.
16 So in conclusion, then, in my view, this
17 drug has a positive benefit-risk profile. Its
18 efficacy is comparable to what is available in
19 terms of adjunctive therapies, but offers a new
20 distinct mechanism of action. It has a favorable
21 safety profile, and in my opinion, patients like
22 J.D. need these options. And as clinicians, we
A Matter of Record (301) 890-4188 122
1 need to have additional options for our patients.
2 Thank you.
3 Applicant Presentation - Lisa von Moltke
4 DR. VON MOLTKE: Thank you, and we
5 appreciate the attention to the presentations. And
6 I want to close by briefly summarizing some key
7 points from this morning.
8 First, there is a significant unmet need for
9 patients with MDD, and there's a specific need for
10 therapies that are working via a new mechanism.
11 BUP/SAM works via opioid system modulation and
12 would bring a new mechanism of action to this
13 indication. This therapy maintains the
14 antidepressant activity of BUP while mitigating for
15 risk of abuse and dependence.
16 You've heard about the significant
17 challenges in studying MDD and the utilization of
18 SPCD to meet those challenges, and you've seen data
19 today from three important studies indicating that
20 there is a positive benefit-risk profile for
21 BUP/SAM. There is substantial evidence of efficacy
22 comparable to other adjunctive therapies, and
A Matter of Record (301) 890-4188 123
1 there's a well-characterized and manageable safety
2 profile.
3 Finally, Alkermes is committed to extensive
4 education and monitoring to ensure appropriate use,
5 and with that, we'll close the sponsor's
6 presentation. Thank you.
7 DR. NARENDRAN: Thank you.
8 If we could have the people who came to the
9 table introduce yourselves, Dr. Temple and Dr. Lee.
10 DR. LEE: Daniel Lee, clinical reviewer.
11 DR. TEMPLE: Bob Temple, deputy director of
12 ODE I.
13 Clarifying Questions to the Applicant
14 DR. NARENDRAN: If we could now open the
15 session for clarifying questions. We're already
16 running a little bit late, so I'd like to remind
17 the panel to be very focused in your questions.
18 It's really just clarifying questions, not to
19 engage in a discussion. Be cognizant of the other
20 panel members who may also want to ask questions.
21 So clarifying questions, very brief, try to
22 get the answer; please do not engage in a
A Matter of Record (301) 890-4188 124
1 discussion. There will be plenty of time for a
2 discussion later.
3 Dr. Warholak, your question?
4 DR. WARHOLAK: So my first question, I'm not
5 really sure who to address to because I have a
6 question about the clinical outcomes assessment
7 used as an endpoint. So I would address that
8 usually to a psychometrician, but I don't know if
9 there's one over there.
10 My question is, why did you decide to use
11 the MADRS-6, and what evidence of content validity
12 does that have compared to the 10? Also, can you
13 also give us some evidence about using the average
14 over the duration of the study as opposed to the
15 endpoint?
16 DR. VON MOLTKE: Sure. I'll ask two
17 different colleagues to come up and address that.
18 The MADRS-6 was evident in the literature, and
19 that's where we got our interest in it. I'll ask
20 Dr. Pathak to come up and walk you through some of
21 the differences. And then, with regard to
22 averaging, I'll ask one of our statisticians as
A Matter of Record (301) 890-4188 125
1 well as one of our clinicians to come up.
2 DR. PATHAK: Sanjeev Pathak, psychiatrist,
3 Alkermes. Thanks for that question and the
4 opportunity to clarify. We picked up MADRS-6 based
5 on emerging literature, and the literature
6 demonstrates that it is sensitive in evaluation of
7 depression symptoms and improvement. And also
8 recent literature suggests that it may be a good
9 measure for assessing adjunct treatment when every
10 patient is receiving a background antidepressant,
11 where some of the symptoms may have been addressed.
12 And the validation was published way back in 2002
13 by Beck [ph] and colleagues.
14 DR. SCHINDLER: Could I ask you just to
15 clarify? I'm not sure I totally understood your
16 question of evidence for averaging.
17 DR. WARHOLAK: Yes. It appears that the
18 usual way to look at the -- it is the MADRS-10, and
19 it's at the end of the therapeutic evaluation as
20 opposed to averaging every week over the study
21 period. So what comparative evidence do you have
22 that that's a valid approach?
A Matter of Record (301) 890-4188 126
1 DR. MATHEW: Just to clarify, we're not
2 averaging at the patient level. We're averaging at
3 the group mean level, so we're using the MMRM model
4 to estimate the treatment effect at each week, and
5 then averaging that treatment effect among multiple
6 weeks.
7 That's a method that's been used before.
8 Coming as a statistician, not a psychometrician,
9 when you calculate data using an MMRM model that is
10 longitudinal data over multiple weeks, one method
11 is to look at the difference over multiple weeks,
12 and then average those estimates together to
13 stabilize the estimate of treatment effect. And
14 that's a method that's been used before.
15 DR. SCHINDLER: I'll just add, from a
16 clinical perspective, looking at the average over
17 time makes perfect sense because that's exactly
18 what we're doing when we make treatment decisions.
19 We're not basing it on an end of treatment because
20 there really is no end of treatment. We're
21 continuing to assess patients longitudinally over
22 time.
A Matter of Record (301) 890-4188 127
1 DR. NARENDRAN: Next question, Ms. Grant?
2 MS. JONIAK-GRANT: Elizabeth Joniak-Grant.
3 There's a lot of discussion here in changes from
4 baseline or the difference with the BUP/SAM versus,
5 for example, placebo. So there's lots of talks of
6 minus 1, minus 2, minus 3. And I can't help but
7 think, what does this mean practically. What does
8 this mean for the actual person? When you say
9 there's this change of minus 2, what does that mean
10 for an individual's life, so to speak?
11 Does that mean they're sleeping 30 minutes
12 less a day? What's the clinical significance of
13 it?
14 DR. VON MOLTKE: So I would ask
15 Dr. Papakostas to give us his perspective on how
16 that translates.
17 DR. PAPAKOSTAS: Thank you for your
18 question. George Papakostas, Massachusetts General
19 Hospital, Harvard Medical School. I'm here as a
20 consultant to Alkermes, but not an employee of the
21 company.
22 It's a great question because studies vary
A Matter of Record (301) 890-4188 128
1 in terms of the patients that they have, and some
2 studies have more severe patients or less severe
3 patients. So how do you extract the clinical
4 significance of a statistically significant
5 finding?
6 The best way to do that is with a
7 standardized effect size, preferably by pooling the
8 data for both positive and negative studies, not
9 just cherry-picking.
10 So if that's applied to the current dataset
11 in one of the slides that was shown -- slide up,
12 please. This is actually essentially that. It's a
13 standardized effect size of all of the studies of
14 BUP/SAM, both positive and negative, and all of the
15 studies of brexpiprazole, both positive and
16 negative, that essentially show that the efficacy
17 of this program is similar to the efficacy of its
18 predecessor, that is, a typical antipsychotic
19 agent.
20 So it's pretty much comparable to the effect
21 that the atypicals have in clinical practice. So
22 depending on the patient that you use, you can
A Matter of Record (301) 890-4188 129
1 expect response rates of 25, 30 percent in clinical
2 practice. Thank you.
3 DR. NARENDRAN: Next question,
4 Dr. Fiedorowicz?
5 DR. FIEDOROWICZ: Yes. My question is, was
6 the integrity of the blind assessed in any of the
7 4 studies.
8 DR. VON MOLTKE: Yes, it was. And to speak
9 more specifically to that, I'll have Dr. Pathak
10 come up. I can tell you that, most recently in the
11 202 study, because we had teams going back out to
12 assess looking for documentation, et cetera, we
13 reconfirmed the integrity of that blind even a
14 second time.
15 DR. PATHAK: Could you please clarify what
16 would you be looking for in regards to
17 interrogating the integrity of the blind?
18 DR. FIEDOROWICZ: Were participants
19 specifically asked what treatment they believed
20 they were receiving?
21 DR. PATHAK: We did not ask the question.
22 Overall, though, we want to emphasize that all the
A Matter of Record (301) 890-4188 130
1 evidence suggests that the blind was maintained.
2 DR. NARENDRAN: Next question, Dr. Meisel?
3 DR. MEISEL: Thank you. Steve Meisel with
4 Fairview in Minneapolis; a couple of questions. I
5 think they'll all be brief. First of all, just to
6 clarify, is the postulate that the kappa and
7 antagonist quality of buprenorphine is responsible
8 for the efficacy? Because otherwise, you're
9 combining matter and antimatter with a mu agonist
10 and antagonist. Is that the postulate?
11 DR. VON MOLTKE: To some degree, that is
12 part of it. We certainly have the kappa antagonism
13 still going on. The amount of mu that's been
14 eradicated, so to speak, has been empirically
15 driven down such that we don't see on vast scales,
16 et cetera. So whether there is some residual
17 amount or nothing, we really can't say. But
18 clearly, the kappa antagonism remains.
19 DR. MEISEL: If I were a person who was
20 intent on diverting and abusing this product, have
21 you tested whether or not it is possible to
22 chemically or physically separate the two
A Matter of Record (301) 890-4188 131
1 chemicals? If I were to get a pile of these pills,
2 and there's no doubt that there will be some
3 internet-based recipes to separate them out, have
4 you tested whether that's possible or not?
5 DR. VON MOLTKE: Yes. Our chemists did
6 conduct a limited set of conditions, but those
7 conditions were specifically picked because they
8 felt that they were optimally able to likely
9 separate those two based on physiochemical
10 properties.
11 So they picked the conditions where they
12 thought it would be easiest, and they did find a
13 couple of conditions where you could separate
14 buprenorphine, but the yield was extremely low and
15 it was variable.
16 So starting from a 2-milligram pill or even
17 multiple pills, the highest yields were 23 percent,
18 and that was on a good day. It was very variable.
19 DR. MEISEL: Twenty-three percent sounds
20 high to me, but okay. And then the last clarifying
21 question; can you explain or postulate why the 8/8
22 dose is less effective than the 2/2 dose?
A Matter of Record (301) 890-4188 132
1 DR. VON MOLTKE: Sure. I think there are
2 two reasons in this case. First, with
3 buprenorphine, there's pretty good literature, both
4 in the preclinical and then in vitro, that it has a
5 U-shaped curve, especially on the nociceptor
6 effects, but even in some of the other paradigms
7 that has been looked at. So it's a complicated
8 pharmacology just for BUP.
9 In addition, in psychiatry, using these
10 symptom-based scales for efficacy, if your side
11 effect profile starts to bleed into affecting those
12 efficacy measurements, you're going to see a hit on
13 your efficacy.
14 So for example, we have nausea, vomiting,
15 things like that. That's going to impact the HAM-D
16 assessment of things like appetite. So that's been
17 well described, going back to the antipsychotics.
18 It's a known phenomenon.
19 DR. NARENDRAN: Next question, Dr. Iyengar?
20 DR. IYENGAR: I have two questions about the
21 sensitivity analysis that you've done. One is, did
22 you check the assumptions about the lack of
A Matter of Record (301) 890-4188 133
1 correlation between the two stages in the SPCD
2 design? And the second question relates to the
3 missingness, the missing data issue.
4 No matter what you do, there's some
5 assumption about the missing data. It sounds to me
6 like MAR is the assumption that has been used here.
7 Is that correct? And also, did you do any
8 sensitivity analyses to check to see if some non-
9 ignorable mechanism might be operating?
10 DR. VON MOLTKE: I'm going to ask
11 Dr. Schindler to come up and walk you through those
12 questions.
13 DR. SCHINDLER: Yes, we did assume that when
14 we do the MMRM model, that we have missing and
15 random MAR. And as you know, when you have missing
16 at random, the estimates from the MMRM model are
17 unbiased. We do have some sensitivity analyses.
18 The other question was about the correlation
19 of the estimates between the two groups. And even
20 though you have patients who were on placebo in
21 stage 1 and they rolled into stage 2, but the
22 advantage of the SPCD design is the differences
A Matter of Record (301) 890-4188 134
1 that you calculate are actually not correlated
2 between stage 1 and stage 2.
3 So at the difference level, the correlation
4 goes down to zero, and that's been demonstrated.
5 There was actually a paper by Yeh Fong Chen from
6 FDA, et al., who have shown that there isn't a
7 correlation between the difference.
8 I can show a little bit of some sensitivity
9 analyses for the missing data. Slide up.
10 Here's the study 202, and what you see here
11 is the HAM-D data and the MADRS-10 data at end of
12 treatment. The primary analysis for HAM-D is what
13 we presented earlier. MADRS-10 was also presented.
14 And then when you use the MMRM model, that's the
15 top line where we see the primary
16 endpoint -- there's other ways to deal with missing
17 data, and those other ways include multiple
18 imputation.
19 So when we do multiple imputation using the
20 other members of that same treatment group to fill
21 in the data for the people who are missing, that's
22 the second line, and you see for the HAM-D 17 at
A Matter of Record (301) 890-4188 135
1 end of treatment, that's virtually the same;
2 MADRS-10, end of treatment, also very similar for
3 multiple imputation.
4 But there's an even more conservative
5 approach to multiple imputation, as you know, and
6 that's where you replace the missingness with the
7 placebo. So you flip the missingness and use the
8 placebo means to replace the data in the multiple
9 imputation, which is much more conservative, which
10 would mean that the members who are missing are
11 having the same effect as placebo.
12 We know that's much more conservative. The
13 HAM-D 17 at end of treatment, the mean value is
14 still on the side that favors BUP/SAM. The
15 confidence interval spreads out a little bit across
16 the zero, but we know that's a very conservative
17 multiplicity, very conservative multiple imputation
18 method.
19 For MADRS-10 end of treatment, we see the
20 same result as before, so it doesn't cross zero.
21 So this is reassuring that even with our
22 sensitivity analyses, we still see a robust effect.
A Matter of Record (301) 890-4188 136
1 DR. IYENGAR: Just one quick follow-up;
2 roughly how much missingness was there?
3 DR. SCHINDLER: Well, these are longitudinal
4 studies, so the missingness -- for people who are
5 lost to follow-up, some may be lost over time. So
6 the most extreme would be towards the very end of
7 the study, the very last visit. So the MMRM model
8 uses all the data that's there at the earlier
9 visits. So the missingness is not the same at each
10 visit. The most extreme is about 20 percent at the
11 last visit.
12 DR. NARENDRAN: Next question, Dr. Jain?
13 DR. JAIN: Thank you. I have two questions.
14 One pertains to efficacy and the other to side
15 effects. In terms of efficacy, at the individual
16 subject level, have you calculated response rates
17 and remission rates from the placebo-controlled
18 trials?
19 On the safety side, you presented some data
20 related to the COWS in terms of withdrawal scores
21 in which participants were categorized into mild,
22 moderate, and severe categories.
A Matter of Record (301) 890-4188 137
1 Do you have that data available in a summary
2 fashion, a continuous fashion, such that we could
3 understand what symptoms participants were
4 experiencing and at what level subjects within the
5 mild category were experiencing withdrawal?
6 DR. VON MOLTKE: Yes. I will have
7 Dr. Pathak take your first question about response
8 and remission, and then we'll have one of our
9 safety people walk you through the COWS data that
10 we have.
11 DR. PATHAK: Sanjeev Pathak, psychiatrist,
12 Alkermes. Yes, we did explore or evaluate a
13 response in remission in the placebo-controlled
14 trials. I do want to emphasize that the duration
15 of the trials was 4, 5, or 6 weeks.
16 What we found in the overall patient
17 population; that the median time to remission was
18 approximately 3 months or so, and this we can glean
19 from following the patients as they rolled over to
20 the longer-term study. And overall, there was
21 clinically meaningful benefit.
22 DR. CONLEY: Next question?
A Matter of Record (301) 890-4188 138
1 DR. STANFORD: Hi. Dr. Arielle Stanford.
2 I'm a psychiatrist. I work at Alkermes. We looked
3 at the item levels for the COWS scores, and there
4 was no particular pattern of which items were
5 particularly common across the different subjects.
6 The items were generally G.I. in nature, which is
7 not uncommon for what we see following
8 discontinuation. But that said, there was no
9 pattern, and the individual item scores were low
10 for any particular item.
11 DR. NARENDRAN: Next question, Dr. Dunn?
12 DR. DUNN: Hi. Walter Dunn. One follow-up
13 and two quick questions. Follow-up; in terms of
14 the averaging of the MADRS, what was the rationale
15 for averaging weeks 3 through 5 as opposed to 4 and
16 5 or 2 through 5?
17 Second follow-up or second question; do you
18 have the individual data for that patient who the
19 FDA is proposing to exclude from study 202,
20 specifically their week-by-week response? And the
21 third question is, do we have any evidence that,
22 for patients who are on naltrexone, would BUP/SAM
A Matter of Record (301) 890-4188 139
1 still be effective for antidepressant?
2 DR. VON MOLTKE: We don't have data on the
3 naltrexone, but we would assume that being on a
4 complete blocker to start with might be something
5 that would not be something we would recommend.
6 I'm going to then march up to your second
7 question on the individual data for the patient
8 that was proposed to be excluded, and then I'll ask
9 Dr. Schindler to come up and address your first
10 question.
11 So we do have the single assessments for the
12 single patient. Slide up, please. So you can see
13 here, we've got both the HAM-D, which was the
14 primary endpoint as well as the MADRS-10. The
15 patient came in and had a pretty robust response
16 following both of those indices.
17 The patient then started a taper, which was
18 part of this particular study at week 4. You can
19 start to see the loss of response, and then
20 continued on placebo following that. So it looks
21 pretty consistent to us. And in addition to that,
22 the patient did meet all eligible criteria.
A Matter of Record (301) 890-4188 140
1 There had been some question as to how long
2 the patient had been on fluoxetine to give you an
3 idea of what the question was. And we found
4 actually 5 separate independent corroborators that
5 the patient had been on fluoxetine for more than
6 8 weeks, which was the defining line.
7 Slide down, please. Dr. Schindler?
8 DR. SCHINDLER: The earlier question about
9 averaging, why did we pick the weeks that we
10 picked; from the protocol, we used titration at the
11 beginning of the study, so we did want to average
12 during the time period of titration. But we did
13 look at different ways of averaging over different
14 weeks.
15 Slide up. This shows that what we wanted to
16 do was to really see what was the treatment effect
17 during the duration that there really was a
18 treatment effect to measure, but we looked at
19 different time points.
20 So this is the study 207, which I presented,
21 and this is MADRS-10 data. We looked at week 2
22 through end of treatment, week 3 through end of
A Matter of Record (301) 890-4188 141
1 treatment, which was the primary endpoint. We
2 looked at 4 through end of treatment. They give
3 essentially the same picture. I think you could
4 pick any of these if you wanted to.
5 We chose to look at the time point that
6 included the largest duration after titration.
7 DR. NARENDRAN: Ms. Witczak?
8 MS. WITCZAK: Kim Witczak, consumer
9 representative. I'm curious a little bit about the
10 ages of the people that were in the clinical trial
11 and were any of them seniors, just because that's
12 going to have real-world consequences.
13 DR. VON MOLTKE: Sure. The upper end of the
14 age range, I believe, was 70.
15 MS. WITCZAK: How much weighted, I mean --
16 DR. VON MOLTKE: Yes. There were not many
17 that were over age 65, for sure, and we would be
18 advising caution if a patient had underlying
19 concomitant serious illnesses, which is what was
20 done in the trial as well.
21 MS. WITCZAK: With multiple medications?
22 DR. VON MOLTKE: Yes, and things like heart
A Matter of Record (301) 890-4188 142
1 disease, lung disease. There'd have to be some
2 caution and some extra consultation with their
3 physicians.
4 MS. WITCZAK: Great. Thank you.
5 DR. NARENDRAN: Dr. Hernandez-Diaz?
6 DR. HERNANDEZ-DIAZ: Hi. Sorry for going
7 back to the question of what is clinically
8 meaningful, probably for Dr. Papakostas. The
9 question is not about the statistically significant
10 or standard deviations, but clinically, what would
11 you consider getting better?
12 Or in the plot, you put a remission. In the
13 scales that you use clinically and the ones that
14 have been used for research, is decreasing
15 15 points in the scale, 10 points in the scale, is
16 that the kind of question? I would like to
17 understand.
18 DR. PAPAKOSTAS: So historically,
19 antidepressants, the traditional antidepressants,
20 if you use them as monotherapy in a regular
21 population, not selected for treatment failures, if
22 you use the MADRS, you tend to see differences from
A Matter of Record (301) 890-4188 143
1 placebo, 3 to 4 points.
2 In augmentation studies, depending on how
3 many failures there have been, how difficult to
4 treat the population is, typically you see between
5 1 and a half and 2 and a half MADRS points of a
6 difference. This seems to fall between those two.
7 DR. HERNANDEZ-DIAZ: So Dr. Mathis presented
8 a case for patient J.D., presenting it as a patient
9 that had been on treatment and went from a scale of
10 21 to 16, so 5 points different, and that patient
11 was considered to be not responding, so a treatment
12 failure. So 5 points difference in clinical
13 practice is considered failure, but a difference of
14 1 in a trial is considered effectiveness?
15 DR. PAPAKOSTAS: I got it. I wasn't clear
16 enough in your question before. Let me try again.
17 So when I talk about 3 or 4 for monotherapy and
18 1 and a half to 2 and a half augmentation, I don't
19 mean improvement. I mean improvement in drug,
20 minus improvement for placebo.
21 So if you treat a patient, and they're very
22 severe, and their improvement is 5 points, that's
A Matter of Record (301) 890-4188 144
1 not enough. But if you have a decrease in
2 depressive symptoms that's greater than placebo in
3 augmentation studies of 1 and a half to 2 and a
4 half points, that is a strong signal.
5 DR. HERNANDEZ-DIAZ: Yes. But since in
6 clinical practice, we are not going to leave the
7 placebo, then we are going to have only the one
8 point unless we are going to leave placebo, and
9 placebo meaning in this context we're only taking
10 care of a patient being closer or whatever we are
11 doing in the trial. We are not going to do that
12 when we prescribe. Right? So we are going to have
13 only the one point.
14 DR. PAPAKOSTAS: Let me try a different way.
15 Is there a slide that shows how much of a change in
16 MADRS there was on the drug during a study?
17 DR. HERNANDEZ-DIAZ: Slide 46, I think.
18 (Laughter.)
19 DR. PAPAKOSTAS: Thank you. Slide up.
20 Excellent. Slide up, please.
21 So a change from baseline here, the average
22 here in stage 1 is 7 points and the average in
A Matter of Record (301) 890-4188 145
1 stage 2 is 3 points with a very low placebo
2 response. But that's the average. That's looking
3 at patients where this didn't work in patients and
4 where this did work. That's meaningful, and the
5 difference is also meaningful. Thank you.
6 DR. NARENDRAN: I think we're clear on that.
7 DR. MATHEW: I'm sorry. Just a quick
8 clarification. In the case, it was a QIDS scale,
9 so that's a self-report instrument, so we weren't
10 addressing the MADRS.
11 DR. NARENDRAN: Thank you, Sanjay.
12 Dr. Kulldorff?
13 DR. KULLDORFF: Thank you. My name is
14 Martin Kulldorff. I'm a biostatistician at Harvard
15 Medical School. I have questions for
16 Dr. Papakostas and for Dr. Schindler. Maybe we can
17 start with Dr. Papakostas.
18 On slide 16, if you can get it on the screen
19 also, I understand the first step there after the
20 first treatment, 63 percent are still symptomatic
21 and 37 percent we had success. The subsequent one,
22 is that 75 percent of those 63 that are still
A Matter of Record (301) 890-4188 146
1 symptomatic?
2 DR. PAPAKOSTAS: Correct. So if the first
3 doesn't work, the chances of the second working in
4 the patient where the first did not work goes down
5 to 25 percent. If you give another one in those
6 patients where it did not work, the success goes
7 down even further.
8 DR. KULLDORFF: So using the Excel
9 spreadsheet, I did a quick calculation to see what
10 is then the overall success rate after the four
11 steps. And my number was that the success rate is
12 at 66.1 percent and there are 33.9 that are still
13 symptomatic.
14 Would you agree that that would be a
15 reasonable number based on your experience?
16 DR. PAPAKOSTAS: I didn't do the
17 calculations, but I think a better way of orienting
18 this -- because I think I see what you mean. A
19 better way of orienting this is what do we do in
20 the clinic if you have a patient where the first or
21 the second treatment didn't work?
22 So let's say that you start from the end of
A Matter of Record (301) 890-4188 147
1 step 2. If you give them a monotherapy, the
2 chances that the person will remain symptomatic
3 after you optimize the dose duration is going to be
4 close to 90 percent if you use a monotherapy. So
5 you need to use an adjunct, and that's where we
6 need to grow. Thank you.
7 DR. KULLDORFF: Thank you.
8 DR. NARENDRAN: Next question, Dr. Crawford?
9 DR. KULLDORFF: I am confused if we can
10 compare slides 43 and 54. The 205 study, if we
11 look at slide 54, if we can get that on the
12 screen -- thank you. The 205 study there has an
13 effect size of about minus 2 and a half and is
14 statistically significant. But if you do it on 43,
15 there's an effect size of about minus 1.8, and it's
16 not statistically significant.
17 Why is there a difference there? I'm
18 missing something.
19 DR. SCHINDLER: It's the time point that's
20 different. There's two slides. On this slide,
21 we're looking at MADRS-10. Can you do the last
22 slide? This slide, just to remember, is for 205,
A Matter of Record (301) 890-4188 148
1 looking at MADRS-10 at the end of treatment. So
2 it's hard to keep track of the excellent time
3 points that we're making the comparison. So this
4 is a comparison at the end of treatment.
5 Now, the other slide, 43, if you bring that
6 slide up, that's at week 5. And you notice the
7 dilemma for this -- and this is exactly the reason
8 why we want to use the average difference across
9 multiple time points. At week 5, see at the top of
10 the slide, it says MADRS-10 evaluated at week 5.
11 At week 5, there's variability in the data,
12 and the curves -- especially you notice it in
13 stage 2. The curves start to come together a
14 little bit, and that difference is a little bit
15 smaller than it is at different time points. At
16 week 6, you can see the week right next to it, the
17 curves separate even more.
18 So then the question would be, when you look
19 at end of treatment, the end of treatment is
20 looking at week 6, but this evaluation was looking
21 at week 5.
22 So if your goal is to understand what is the
A Matter of Record (301) 890-4188 149
1 treatment effect, which week you picked, if you
2 only pick one time point, which week you pick is
3 important If you average some of the differences
4 together, then it becomes less important, and
5 you're really measuring the treatment effect.
6 You're not measuring the variability of the data.
7 I think that's what you've uncovered, that
8 if you look at just week 5 versus the end of
9 treatment, you see a slightly different picture.
10 And that's just due to the variability of the data.
11 DR. KULLDORFF: So on slide 54, you looked
12 at week 6 only then?
13 DR. SCHINDLER: Week 6, and you notice -- if
14 you look at this picture, you see stage 1 is 5
15 weeks and stage 2 is 6 weeks. So the end of
16 treatment for stage 1 is that 5-week period, but
17 the end of treatment for stage 2 is 6 weeks because
18 they're one week different.
19 DR. KULLDORFF: So slide 54 is a combination
20 of weeks 5 and 6.
21 DR. SCHINDLER: It's a combination. Right.
22 It's the end of each stage. It's the last week in
A Matter of Record (301) 890-4188 150
1 each stage.
2 DR. KULLDORFF: That leads to my second
3 question, which --
4 DR. NARENDRAN: Can we get one second? I
5 just want to -- we have two more questions, so
6 we're running out of time. I think we're clear on
7 that point.
8 Dr. Crawford?
9 DR. CRAWFORD: Thank you. The applicant has
10 provided extensive information and data on
11 buprenorphine and samidorphan as fixed-dose
12 combinations. The true new molecular entity is
13 samidorphan, and prior to the agency's
14 presentation, because I do see something there, I
15 wanted to ask the applicant, do you have any
16 comment of your findings you might have of the
17 efficacy and adverse effects of the NMEs,
18 samidorphan alone?
19 DR. VON MOLTKE: Yes, we've had an extensive
20 characterization of samidorphan. It actually has a
21 place at another development program for us as
22 well. It has some similarities in its AE profile,
A Matter of Record (301) 890-4188 151
1 things like nausea, G.I. upset, some constipation.
2 Its efficacy wasn't tested partly because
3 its purpose here was primarily to mitigate the mu
4 agonist activity, but it also is part of a group of
5 antagonists that we know from other studies have
6 not shown to have antidepressant activities.
7 So we really had no reason to think that
8 samidorphan on its own would have efficacy.
9 DR. NARENDRAN: Next question, Dr. Riley?
10 DR. RILEY: So as I understand it, what you
11 are doing is a mean or means on the MADRS average.
12 Did you also consider or do analyses on the patient
13 level instead of the group level? And if so, did
14 you get comparable results?
15 Related to that, I'm more familiar with
16 mixed models for repeated measures over time and
17 treatment effects. Did you consider those or do
18 those as well?
19 DR. VON MOLTKE: I'll ask Dr. Schindler to
20 come up.
21 DR. SCHINDLER: Just to answer the endpoint,
22 the last point, I think the MMRM model is a mixed
A Matter of Record (301) 890-4188 152
1 model, so I think that, yes, we use the MMRM model.
2 And recognize that what the MMRM model does is it
3 looks at all of the visits, all of the data from
4 all of the visits, and calculates a visit mean for
5 each treatment group as part of the model.
6 And that's part of what you estimate in the model,
7 and then you look at the contrast between the visit
8 mean for one group and the visit mean for another
9 group. And that's part of the way we analyze the
10 data.
11 What we don't do is analyze individual
12 patient data over multiple time points, and
13 recognize that there's a certain amount of missing
14 data, not a lot, but a certain amount of missing
15 data.
16 So if we average 2, 3, 4, 5, and 6 weeks
17 over individual patient data, some of those might
18 be missing, and so we didn't do that. That
19 analysis, I don't think, would be the fair analysis
20 for this type of data. The MMRRM model uses the
21 available data to estimate that, what the treatment
22 effect is for the study at each time point, and
A Matter of Record (301) 890-4188 153
1 then uses sort of a variance/covariance structure
2 to really estimate the trajectory of the change
3 from baseline, and that trajectory is what you see
4 as part of the model.
5 So that's the way we did it. We did not
6 average individual patient data over time. We just
7 averaged the treatment effects estimated from the
8 model at multiple over time at multiple time
9 points.
10 DR. NARENDRAN: I have two focused
11 questions. Do you have in vivo receptor occupancy
12 data for the kappa and the mu opioid receptor,
13 either in human or primates, for the 2-by-2 dose?
14 DR. VON MOLTKE: We do have in vitro data.
15 DR. NARENDRAN: In vivo, in vivo, in PET
16 imaging.
17 DR. VON MOLTKE: No, no. We do not. I'm
18 sorry.
19 DR. NARENDRAN: The second question is, the
20 human abuse potential study, you looked at in
21 controls and also you showed in major depressive
22 disorder. The drug liking scores, were they done
A Matter of Record (301) 890-4188 154
1 at week 1 or were they done at -- were they done
2 after 7 days when the metabolite accumulates?
3 Because it said samidorphan is a metabolite that is
4 a full mu agonist. What time points were they?
5 DR. VON MOLTKE: I'll have Dr. Pathak walk
6 you through that.
7 DR. PATHAK: Dr. Narendran, to clarify, your
8 question is about the human abuse potential study.
9 Right? These are single administration, and the
10 evaluation of drug liking is done at multiple time
11 points for a prolonged period of time. So in this
12 study, it was 48 hours and it was done repeatedly.
13 DR. NARENDRAN: So just to clarify, you did
14 not do multiple dosing and you did not get it at
15 7 days, where the mu agonist would have accumulated
16 the metabolite?
17 DR. VON MOLTKE: Right. The half studies
18 were single dose.
19 DR. NARENDRAN: Thank you.
20 I think we'll take a 15-minute break. We're
21 running a little bit over time. We'll try to make
22 up that 15 minutes during the lunch break if that's
A Matter of Record (301) 890-4188 155
1 okay.
2 Just to remind panel members, please
3 remember there can be no discussion in the break
4 amongst yourselves. So we'll meet back at 10:45.
5 Thank you.
6 (Whereupon, at 10:32 a.m., a recess was
7 taken.)
8 DR. NARENDRAN: Thank you. We will now
9 proceed with the FDA presentations, starting with
10 Dr. Tiffany Farchione, deputy director's
11 presentations.
12 FDA Presentation - Tiffany Farchione
13 DR. FARCHIONE: Hi. Good morning,
14 everybody. I am going to begin the FDA
15 presentations this morning by providing the
16 regulatory history of this development program and
17 the interactions that the company had with the
18 agency.
19 I plan to highlight a few key interactions
20 between the agency and the applicant. So beginning
21 with the pre-IND meeting in February of 2011;
22 moving on to the end-of-phase-2 meeting request
A Matter of Record (301) 890-4188 156
1 that was scheduled for October 2013; the written
2 guidance that we provided on the statistical
3 analysis plan in July of 2015; a guidance meeting
4 related to the completed studies 205 and 206 in
5 September of 2016; and guidance on the revised
6 statistical analysis plan and endpoints in February
7 of 2017; the pre-NDA meeting that included feedback
8 on the MADRS-6 in July of 2017; and culminating
9 with the NDA submission in January of this year.
10 I'm also going to highlight a few other
11 interactions in the year or so preceding NDA
12 submission, specifically a receipt of the amended
13 protocol and statistical analysis plan for
14 study 207 about a week before that September
15 meeting in 2016; the breakthrough therapy
16 designation advice that we provided in March of
17 2017; and the applicant submission of their dossier
18 for the MADRS-6 that they submitted in April of
19 2017.
20 The first interaction that we had with the
21 applicant related to this program was the pre-IND
22 meeting in February of 2011. And at that time,
A Matter of Record (301) 890-4188 157
1 they described their plan to develop buprenorphine
2 and samidorphan for the adjunctive treatment of
3 MDD.
4 We discussed the general design elements
5 necessary to support that indication, like
6 selecting patients with inadequate response to
7 antidepressants and ensuring that the patients
8 continued their antidepressant on which they had
9 suboptimal response while enrolled in the double-
10 blind treatment phase. So in other words, the
11 studies were to involve an add-on design, which
12 they did.
13 We also discussed the support needed for a
14 combination drug and the difficulties involved with
15 conducting a full factorial study for this
16 particular product. So normally, if you have a
17 combination drug, we expect that the company would
18 provide evidence of efficacy for both components
19 separately as well as together, but in this case,
20 the samidorphan was added for safety and not for
21 efficacy.
22 We also noted the potential to produce
A Matter of Record (301) 890-4188 158
1 opioid toxicity in opioid-naïve patients with MDD
2 or potentially opioid dependence if patients were
3 treated with buprenorphine alone. The applicant
4 did follow our advice on study design in this
5 regard and did not conduct the full factorial study
6 for the safety reasons noted.
7 The phase 2 study that's been referred to
8 today as study 202, was discussed during this
9 meeting as well. At that time, the applicant
10 specifically described it as a proof-of-concept
11 study. There was no a priori hypothesis as to
12 which dose would be more likely to be effective,
13 and there was no plan for multiplicity adjustment
14 to control for type 1 error or the chance of a
15 false-positive finding.
16 So because of that, we said that we had no
17 objection to using sequential parallel comparison
18 design in the proof-of-concept trial, and we
19 encouraged the applicant to provide a detailed
20 statistical analysis plan and seek feedback prior
21 to starting the study if they intended to use this
22 study to support an efficacy claim.
A Matter of Record (301) 890-4188 159
1 We received the statistical analysis plan
2 for this study in August of 2013 along with the
3 study report.
4 Following the completion of study 202, the
5 applicant requested an end-of-phase-2 meeting that
6 was scheduled for October of 2013. In their
7 background package for that meeting, they described
8 their planned phase 3 studies, and, as is typical,
9 submitted questions in advance related to that
10 plan.
11 The agency provided preliminary comments a
12 few days ahead of the scheduled meetings, and in
13 one of those comments, we did express our concern
14 related to the SPCD analyses. We noted that, from
15 a statistical perspective, even though the design
16 appeared to be reasonable, there hadn't been any
17 analytical proof for the validity of the associated
18 analyses, particularly when they were missing data.
19 Following the receipt of our comments, the
20 applicant cancelled the face-to-face meeting,
21 noting that our preliminary responses were
22 sufficient and they did not require further
A Matter of Record (301) 890-4188 160
1 discussion.
2 Our next notable interaction was in July of
3 2015. The applicant requested a guidance meeting
4 to discuss the statistical analysis plan for the
5 phase 3 studies, and we provided written responses
6 to that request. And at that time, we again noted
7 our concerns related to using SPCD in the pivotal
8 trials.
9 Again, we reiterated that we hadn't endorsed
10 any analytical method for SPCD in a confirmatory
11 setting. We were actively engaged with the
12 academic community to try to understand the pros
13 and cons of that design from a regulatory
14 perspective.
15 So we actually encouraged the applicants to
16 collect efficacy data from both stages because,
17 then, they would have sort of a back-up plan.
18 Right? So if the primary endpoint at the end of
19 the study using SPCD was not positive, we said that
20 we could use the data from stage 1, which was a
21 more traditional design if we needed to, if the
22 analyses were still unsettled at the time of the
A Matter of Record (301) 890-4188 161
1 NDA filing.
2 The applicant continued their phase 3
3 program, and when 2 of the 3 planned short-term
4 trials were completed, they requested another
5 meeting to discuss the results and to plan their
6 next steps.
7 During the face-to-face meeting, this was in
8 September of 2016, we discussed studies 205 and 206
9 at that time. So again, 205 was another SPCD study
10 and 206 was the one with the placebo lead-in that
11 you heard about earlier.
12 The applicant acknowledged that neither
13 study met its prespecified primary endpoint and
14 they presented some additional analyses that they
15 said would inform modifications to the other
16 ongoing short-term phase 3 study, which was
17 study 207.
18 Actually, a week prior to the face-to-face
19 meeting, the applicant did submit their modified
20 protocol and statistical analysis plan, but that
21 wasn't enough time for us to review it prior to the
22 meeting. So they ended up requesting a follow-up
A Matter of Record (301) 890-4188 162
1 meeting and that meeting was held in February of
2 2017.
3 So at that meeting, we specifically focused
4 on the changes in the primary endpoint and the
5 statistical analysis plan. So at this time, they
6 changed the primary endpoint from the end of
7 treatment MADRS-10 to the three endpoints that you
8 heard about earlier, to be evaluated in a
9 hierarchical fashion.
10 So the MADRS-6 average, using the change
11 from baseline to week 3 through the end of the
12 efficacy period; the MADRS-10 average change from
13 baseline, week 3 to end of efficacy; and then the
14 traditional change in MADRS-10 from baseline to end
15 of treatment, and again noting that the end of
16 treatment was week 5 for stage 1 and week 6 for
17 stage 2.
18 So in response, we noted that we have not
19 previously accepted the MADRS-6. This was a novel
20 endpoint to us. So we recommended that the
21 applicant submit a dossier for us to evaluate so
22 that we could actually look at the MADRS-6 and
A Matter of Record (301) 890-4188 163
1 determine whether it was fit for purpose for an
2 antidepressant trial.
3 We did not agree to the averaging strategy.
4 We actually didn't say that it was a review issue.
5 We said that, "We do not agree with your averaging
6 strategy." We also noted that the protocol
7 amendment introduced additional complexity into the
8 SPCD analyses because the stage 1 and stage 2 now
9 have different durations, so you had 5 weeks versus
10 6 weeks.
11 Then at that time, the applicant also
12 proposed pooling studies 205 and 207, but the
13 suggestion was made with the data in hand and not
14 prospectively. And we stated that the pooled
15 analyses could only be considered exploratory.
16 So despite this feedback, there was actually
17 little that the applicant could do to address our
18 concerns because the database lock for study 207
19 occurred in October of 2016.
20 The applicant also requested informal
21 breakthrough advice. This is a relatively new
22 option, which allows companies to ask the Division
A Matter of Record (301) 890-4188 164
1 leadership whether a product might qualify for a
2 breakthrough therapy designation or what additional
3 information they might need to support a
4 breakthrough request.
5 Now, they did mention earlier that they had
6 a fast-track designation, so that's one of the
7 other programs that we have for promising
8 applications, but the bar is higher for the
9 breakthrough therapy designation. So we actually
10 advised the applicant that it would be difficult to
11 grant a breakthrough designation for this program
12 because, at that time, we still had not yet
13 determined the acceptability of the MADRS-6, and we
14 noted that any statistical significance in phase 3
15 relied on post hoc analyses.
16 The applicant did then submit the dossier
17 for the MADRS-6 in April of 2017, and we consulted
18 our clinical outcome assessment staff at that time.
19 At the pre-NDA meeting that was held in July of
20 2017, we did discuss the COA review of the MADRS-6.
21 We noted that the MADRS-6 was not fit for purpose.
22 We said that it could not replace the MADRS-10 for
A Matter of Record (301) 890-4188 165
1 use as a primary endpoint because it excludes
2 concepts that are relative and important in major
3 depressive disorder.
4 So the missing items in the MADRS-6 are
5 related to reduced sleep, reduced appetite,
6 concentration difficulties, and suicidal thoughts.
7 These are core features of depression and can't be
8 excluded from an endpoint in a trial that is
9 designed to assess antidepressant efficacy. We
10 informed the applicant that any analyses of MADRS-6
11 would be considered exploratory; again, not a
12 review issue. We said they would be considered
13 exploratory.
14 That brings us to the January 31, 2018 NDA
15 submission, and as has been widely reported, we did
16 initially refuse to file the application. The
17 sponsor noted and we relayed in our background
18 package as well that there were some factual
19 inaccuracies in the letter in which we refused to
20 file, so at that point, we did have a conversation
21 with the applicant. They clarified the analyses
22 that they used that were intended to support the
A Matter of Record (301) 890-4188 166
1 application, and at that point, we agreed to file
2 the application, which brings us here today.
3 At this point, I'm going to hand things over
4 to Semhar so that he can talk to you about our
5 evaluation of the efficacy analyses. Thank you.
6 FDA Presentation - Semhar Ogbagaber
7 DR. OGBAGABER: Good morning. I'm Semhar
8 Ogbagaber from the Division of Biometrics I. I
9 will be presenting the FDA review of efficacy data
10 for buprenorphine samidorphan, which we call
11 BUP/SAM throughout the presentation.
12 Four studies were submitted in this NDA,
13 which investigated efficacy of BUP/SAM for
14 adjunctive treatment of major depressive disorder.
15 Both applicant and FDA agreed that studies 205 and
16 206 were negative. So in this talk, I will focus
17 on the other two studies, 202 and 207, that had
18 some concerns. Study 207 would be considered
19 positive if MADRS-10 average were accepted as
20 primary endpoint.
21 Since both studies 202 and 207 utilized SPCD
22 or the Sequential Parallel Comparison Design, I
A Matter of Record (301) 890-4188 167
1 will first briefly introduce SPCD before going
2 through the efficacy results, then present general
3 questions or concerns related to SPCD, and in the
4 end, touch on meta-analysis.
5 The SPCD was initially proposed to address
6 high placebo response observed in major depressive
7 disorder trials. Unlike the conventional parallel
8 comparison design, which has only 1 treatment
9 period, SPCD evidence is based on efficacy data
10 from 2 stages. More patients are allocated to
11 placebo group in stage 1, so more placebo
12 non-responders will be randomized in stage 2.
13 The SPCD estimated treatment effect is
14 essentially a weighted average of estimated
15 treatment effects from both stage 1 and stage 2.
16 Within each stage, you compare drug versus placebo
17 as in any conventional design. Stage 1 consists of
18 all incoming patients and stage 2, placebo
19 non-responders from stage 1.
20 Here is a hypothetical illustration of how
21 SPCD treatment effect is computed. As shown in the
22 previous slide, it's a weighted average of
A Matter of Record (301) 890-4188 168
1 treatment effects from both stage 1 and stage 2.
2 As illustrated in the table, supposed weight
3 allocation of 60 percent in stage 1 and 40 percent
4 in stage 2; suppose in the first stage we have 150
5 patients, which many of them are randomized into
6 placebo arm, the estimated treatment effect for
7 stage 1 would be 2 units.
8 At the end of stage 1, 70 patients were
9 identified as placebo non-responders. In stage 2,
10 the treatment effect is 3 units based on the
11 70 placebo non-responders, and the SPCD estimated
12 treatment effect is computed to be 2.4 as indicated
13 in the numbers in red.
14 This is the estimated treatment effect when
15 the two stages are combined. First, we calculate
16 the SPCD estimated treatment effect. To derive the
17 p-value, we need to calculate the variance
18 associated through this estimate.
19 From this illustration, it is apparent that
20 different weight allocation affects SPCD-estimated
21 treatment effect. If we expect a larger treatment
22 effect in stage 2, it would probably make sense to
A Matter of Record (301) 890-4188 169
1 assign more weight to stage 2. However, from a
2 statistical perspective, you might want to do it
3 the other way around because the sample size in
4 stage 2 is relatively smaller in stage 2 than
5 stage 1.
6 But clinical relevance should play a key
7 role in weight allocation and also which population
8 is relevant may deserve some attention.
9 Regardless, it's uncertain whether it is reasonable
10 to combine estimates from both stages because
11 stage 1 enrolls the usual patient population and
12 stage 2 an enriched population.
13 In addition, from this illustration on this
14 slide, there are 70 placebo non-responders entering
15 stage 2. These 70 patients also contributed data
16 in estimating treatment effect in stage 1. So
17 these patients appear to have more influence in the
18 overall estimation of the SPCD treatment effect.
19 Now, let's look at study 202. Study 202 was
20 a phase 2 proof-of-concept, multicenter, randomized
21 trial using SPCD that investigated BUP/SAM 2/2 and
22 8/8. The primary endpoint was changed from
A Matter of Record (301) 890-4188 170
1 baseline to week 4 in HAM-D total score.
2 In the applicant's primary analysis, there
3 was no multiplicity adjustment prespecified to
4 control the overall type 1 error rate. This was an
5 exploratory study to generate hypotheses for
6 study 207. It's uncommon to draw a final inference
7 based on exploratory objectives.
8 Results for study 202 are summarized in this
9 table. For the low dose, 2/2 compared to placebo
10 estimated treatment effect was minus 2.8, and for
11 the high dose, it was essentially neutral.
12 Although the nominal p-value for the 2/2
13 dose was 0.014, we do not consider the efficacy was
14 demonstrated because there were 2 doses compared
15 against placebo and multiplicity adjustment was not
16 prespecified.
17 Also, in a post hoc exploration, conclusion
18 becomes inconsistent when using Bonferroni or fixed
19 sequence testing procedure. If we use Bonferroni
20 procedure by splitting alpha, the 2/2 dose would
21 have made it. But if we use fixed sequence testing
22 procedures, starting with the high dose, which is
A Matter of Record (301) 890-4188 171
1 the prespecified procedure in the other studies, no
2 dose would have made it.
3 Typically, we would expect a larger
4 treatment effect on the higher dose, but in this
5 trial, estimated treatment effect was essentially
6 neutral and p-value was really high. If the 2/2
7 dose and 8/8 treatment groups were combined and
8 compared against placebo, result was not
9 statistically significant. In FDA's view, without
10 the prespecifications, study 202 is not an adequate
11 and well-controlled trial and statistical
12 interpretability is undermined.
13 In this exploratory study, site 124 was
14 identified as influential by the Office of
15 Scientific Investigations at FDA. This site
16 enrolled a single patient, randomized in the 2/2
17 dose group in stage 1 and had an extreme result.
18 This quote here has been amended, and I want
19 to emphasize that there is a slightly different
20 quote here. The quote on this slide, according to
21 the Office of Scientific Investigations, should be
22 the subject's eligibility for the study cannot be
A Matter of Record (301) 890-4188 172
1 determined due to incomplete and contradictory
2 source information. Therefore, we recommend DPP to
3 conduct sensitivity analysis for patient 124001 in
4 study 202.
5 To explore the impact of this patient, we
6 compared the response profile of this patient
7 against the treatment groups. This plot displays
8 the change from baseline in HAM-D 17 score at each
9 visit. The red curve is the single patient who was
10 randomized to the 2/2 dose group in stage 1. The
11 three curves are the group means for the three
12 treatment arms.
13 This patient was randomized to the 2/2 dose
14 group. The mean response profile for this dose
15 group was represented by the solid gray curve.
16 This clearly shows that the patient is quite far
17 from the mean plot of the assigned treatment arm.
18 We also explored the potential impact of
19 this site on efficacy results by removing the site.
20 By excluding this site, the magnitude of estimated
21 treatment effect dropped from 2.8 to 2.2 and
22 nominal p-value goes up from 0.014 to 0.057. The
A Matter of Record (301) 890-4188 173
1 strength of evidence for the 2/2 dose diminished.
2 In FDA's conclusion, study 202 was not an
3 adequate and well-controlled study. There was no
4 prospective plan for type 1 error control rate.
5 The nominally positive result for the 2/2 dose
6 depended on a single patient. This further
7 undermined the strength of evidence.
8 If study 202 had been adequate and well
9 controlled and if statistical significance of
10 results for the 2/2 dose had not been sensitive to
11 the contribution of a single patient, the neutral
12 results with the 8/8 dose would undercut the
13 persuasiveness of the results for the 2/2 dose.
14 This is study 207 that was a phase 3,
15 multicenter, randomized, and investigated BUP/SAM
16 dose 1/1 and 2/2. It also used SPCD design that
17 had two different periods in the first stage and
18 second stage. The first stage had 5 weeks and the
19 second stage had 6 weeks.
20 The proposed primary endpoints were a change
21 in MADRS-6 total score using average of change from
22 baseline to week 3 through end of efficacy period,
A Matter of Record (301) 890-4188 174
1 week 5 for stage 1 and week 6 for stage 2, change
2 in MADRS-10 total score using average of change
3 from baseline to week 3 through end of efficacy
4 period, change in MADRS-10 total score from
5 baseline to end of treatment period.
6 So initially, there was only one primary
7 endpoint specified, which was the third one.
8 Toward the end of the study, the applicant added
9 the two top endpoints and proposed to test them
10 sequentially in the order it appears, and both
11 added primary endpoints measured average
12 improvement over several weeks, and the top one is
13 based on MADRS-6.
14 The clinical outcome assessment staff at the
15 FDA had a concern with the MADRS-6 total score as a
16 primary endpoint because of the omission of
17 clinically important symptoms included in MADRS-10
18 and those symptoms were reduced sleep, reduced
19 appetite, concentration difficulties, and suicidal
20 thoughts.
21 This is MADRS-10 total score at each week or
22 visit, and visit zero corresponds to the baseline
A Matter of Record (301) 890-4188 175
1 visit for stage 1 and stage 2. There's a general
2 improvement in depression in both stages, but the
3 improvement seems to level off or worsen after week
4 4 in stage 1 or week 5 in stage 2.
5 In addition, comparing the two stages, the
6 baseline scores were generally lower in stage 2,
7 with an average of 32 in stage 1 and 27 in stage 2.
8 The observed improvement in stage 2 seems very tiny
9 compared to stage 1. Recall that stage 2 are all
10 placebo non-responders from stage 1.
11 This is the mean change from baseline in
12 MADRS-10 total score. At each week, we see similar
13 patterns to the previous slide; that is, there was
14 a general improvement in depression in both stages,
15 but improvement seems to level off or worsen after
16 week 4, stage 1 and week 5 in stage 2. Also,
17 larger improvement is observed in stage 1 than in
18 stage 2.
19 This is the applicant's efficacy result
20 within each stage and combined SPCD results. The
21 top portion is for the MADRS-10 average endpoint
22 and the bottom portion is for MADRS-10 end-of-
A Matter of Record (301) 890-4188 176
1 treatment endpoint. Among those comparisons, there
2 was only one comparison leading to statistical
3 significance, as shown in the cell with red
4 numbers. This was the 2/2 dose group on MADRS-10
5 average and only when the two stages were combined.
6 The estimated treatment effect was minus 1.9 with a
7 nominal p-value of 0.026.
8 In conclusion, efficacy on the 2/2 dose was
9 supported based on MADRS-10 average and not
10 MADRS-10 end of treatment, which was a conventional
11 primary endpoint used in most MDD trials. For the
12 MADRS-10 average, the magnitude of estimated
13 treatment effect was 1.9 on a 60-point MADRS scale.
14 FDA had a concern with MADRS-10 average as a
15 primary endpoint. In particular, averaging the
16 change in MADRS-6 or MADRS-10 tends to obscure
17 possible drop-off in drug efficacy after the first
18 few weeks of treatment. This is a clinical call
19 regarding acceptability and/or interpretability of
20 MADRS-10 average, SPCD in unequal durations in both
21 stages.
22 The SPCD is a novel design. There is
A Matter of Record (301) 890-4188 177
1 ongoing research. Its pros and cons have not been
2 fully recognized. We would like to shed some
3 insights to help further development in this area,
4 both clinical and statistical.
5 One clinical question is with regards to
6 mixed patient population associated with SPCD
7 design. Which stage deserves more weight? Stage 1
8 had the usual patient population, a trial
9 population, and stage 2 has placebo non-responders
10 from stage 1, an enriched population selected to
11 increase signal over noise, but less similar to the
12 patient population for whom drug would be
13 prescribed.
14 For an effective drug, one would expect to
15 see larger treatment effect in stage 2. However,
16 sample size is smalle. Is there a concern with
17 drop-outs when drop-outs are substantial? Placebo
18 non-responders who stay through end of stage 1 may
19 be intrinsically different from placebo drop-outs.
20 The second clinical question is with regards
21 to prespecified weight allocation. Is there a
22 clinically sensible weight allocation? If so, what
A Matter of Record (301) 890-4188 178
1 is sensible? Different weight allocation leads to
2 different treatment effect and may affect the
3 strength of statistical significance. And placebo
4 non-responders are reused so they have more
5 influence.
6 Question 3 is with regards to unequal
7 durations between stages. Is it clinically
8 relevant to combine estimated treatment effect over
9 unequal durations between stage 1 and stage 2?
10 Stage 1 has 5 weeks and stage 2 has 6 weeks in
11 study 207.
12 Question 4 is with regards to labeling. If
13 a drug is to be approved based on SPCD results,
14 even if neither stage demonstrates efficacy, how
15 does one describe the estimated treatment effect?
16 Regarding statistical analysis, the analysis
17 associated with SPCD tends to be complex because
18 estimates are from a combination of two stages, and
19 some patients are used in both stages. It's not
20 clear whether the correlation between the two
21 estimates from the two stages can be reasonably
22 assumed to be zero in any analysis.
A Matter of Record (301) 890-4188 179
1 To our knowledge, analysis proposed in
2 literature has been used based on zero correlation.
3 This is also an assumption made in applicant's
4 analysis. What are potential impacts of a wrong
5 assumption? If the drug is not effective, the
6 wrong assumption may lead to type 1 error
7 inflation.
8 If drug is effective, it may lead to bias in
9 estimating treatment effect; that is, the claimed
10 95 percent confidence interval of treatment effect
11 may not actually provide the 95 percent coverage.
12 To assess the impact of possible wrong
13 assumption about zero correlation, we performed a
14 bootstrap sampling-based statistical inference
15 without such rigorous assumption of normality. And
16 our conclusion based on the empirical 95 percent
17 confidence interval is consistent with applicant's
18 normal based conclusion. So for this trial, this
19 concern can be dropped. However, in another trial,
20 concern may not be dropped.
21 The second concern was with regards to
22 substantial drop-outs. Potential impact would be
A Matter of Record (301) 890-4188 180
1 on type 1 error inflation, still under research.
2 In this application, however, the drop-out rates
3 were not substantial.
4 Meta-analysis was included in the
5 applicant's package. We should be aware that
6 meta-analysis does not meet the usual standard for
7 substantial evidence. Two adequate and
8 well-controlled trials provide substantiation and
9 protection against false-positive finding. In this
10 application, the meta-analysis was not
11 prospectively planned or agreed as a means to
12 provide evidence of effectiveness.
13 In FDA's view, combinability of SPCD and
14 non-SPCD studies using meta-analysis does not
15 appear to be sound due to incompatibility of
16 populations in stage 1 and 2. Also, the
17 prespecified primary endpoint was not the same
18 across all studies. Also, the rationale behind
19 combining effects based on primary endpoints that
20 were not specified in SAP or protocol is
21 questionable.
22 In summary, for study 202, efficacy was not
A Matter of Record (301) 890-4188 181
1 demonstrated. Study is deemed to be negative
2 study. No prospective plan for multiple testing.
3 Site 124 further undermined strength of evidence.
4 Neutral results for 8/8 dose would undercut any
5 finding of efficacy for the 2/2 dose.
6 For study 207, efficacy on BUP/SAM 2/2 dose
7 was supported based on MADRS-10 average and not
8 MADRS-10 end of treatment. Acceptability or
9 interpretability of MADRS-10 or average in the user
10 of SPCD and unequal durations between stages is
11 still unclear. Meta-analysis does not meet the
12 usual standard of adequate and well-controlled
13 trials for substantial evidence.
14 Thank you, and now Danny is coming.
15 FDA Presentation - Daniel Lee
16 DR. LEE: Good morning, everyone. My name
17 is Daniel Lee, and I'm the clinical reviewer for
18 this application. I'll be presenting the safety
19 portion of the review.
20 I'd like to start by reiterating Dr. Mathis'
21 earlier statement. Based upon our review of the
22 available safety data, FDA believes that most of
A Matter of Record (301) 890-4188 182
1 the risk ascribed to other opiates are present to a
2 lesser degree with BUP/SAM. While samidorphan
3 appears to reduce the frequency of adverse events
4 associated with mu opiate receptor agonism, none of
5 the risks inherent to opiates appear to be
6 completely mitigated by its presence.
7 We analyzed the submitted safety data in
8 several different ways while attempting to
9 determine the best way to analyze the data. As we
10 compared the different analyses, we found that
11 pooling of safety data across the trials appeared
12 to produce misleading results. This was due to the
13 influence of within- and between-trial
14 heterogeneity, differences in randomization ratios,
15 and Simpson's paradox.
16 When we compared pooled results against
17 results obtained in the individual trials, many of
18 the safety signals observed in the individual
19 trials disappeared. We also noted the emergence of
20 new safety signals, which were not observed in any
21 of the submitted trials. For this reason, all
22 safety data I present today will be from individual
A Matter of Record (301) 890-4188 183
1 trials.
2 Overall, BUP/SAM appears to be relatively
3 safe and well tolerated outside of being an opiate.
4 No deaths were reported, and only 11 serious
5 adverse events were reported in the four efficacy
6 trials. While these were divided relatively evenly
7 between drug and placebo, it is important to
8 remember that significantly more individuals were
9 randomized to placebo than randomized to drug.
10 One of the 11 serious adverse events is
11 notable because it demonstrates that BUP/SAM
12 retains the ability to displace other opiate
13 molecules from the mu opiate receptor and
14 precipitate withdrawal. The last two points on the
15 slide will be the focus of attention for the
16 remainder of the presentation.
17 As we take a closer look at adverse events
18 associated with opiate-induced increase in gut
19 motility, I'd like to draw your attention to the
20 significant imbalance in population percentages
21 between groups. This imbalance is observed in
22 nearly all adverse events in the highlighted
A Matter of Record (301) 890-4188 184
1 columns.
2 Due to the 2:2:9 randomization ratio in
3 stage 1 of most trials, comparisons based on
4 absolute number of cases are not terribly helpful.
5 The same number of cases may represent
6 significantly different population percentages,
7 particularly in the initial stages of each trial.
8 I'd also like to point out the imbalance of
9 adverse events remains large in stage 2, despite
10 the fact that many of these individuals started out
11 as placebo non-responders in stage 1. These
12 findings largely align with the applicant's
13 presentation.
14 Nearly everything I've stated about G.I.
15 adverse events also holds true for nervous system
16 adverse events in stage 1. Curiously, the placebo
17 non-responder status does not appear to play a
18 significant role for adverse events occurring in
19 stage 2.
20 While far fewer cases occur in stage 2, the
21 same trends observed regarding adverse event
22 imbalance remain. These findings also largely
A Matter of Record (301) 890-4188 185
1 align with the applicant's presentation.
2 Drop-outs due to adverse events were
3 relatively low in the 4 trials overall. However,
4 most drop-outs reported occurred in the BUP/SAM 2/2
5 group. The majority of these drop-outs, which are
6 summarized in this table, are attributable to
7 adverse events suggestive of opiate activity in the
8 gut or central nervous system. Once again, these
9 findings largely align with the applicant's
10 presentation.
11 Our final area of focus pertaining to
12 potential retention of BUP/SAM opiate activity is
13 withdrawal. Withdrawal is measured using the
14 clinical opiate withdrawal scale, otherwise known
15 as the COWS. Unfortunately, conclusions that can
16 be drawn from the COWS data are somewhat limited
17 because only half of the participants in the
18 efficacy trials received a COWS, despite the
19 requirement that it be obtained for all
20 participants.
21 The COWS data collected in trials 202 and
22 205 were relatively complete. Missing data were
A Matter of Record (301) 890-4188 186
1 concentrated in the later trials, and the missing
2 data appeared to be non-random in distribution
3 between groups.
4 For trial 202, COWS data is not particularly
5 informative. I'm unsure what to make of the two
6 cases of mild opiate withdrawal noted in the
7 population exposed to drug in stage 1, then
8 transitioned to placebo in stage 2. I don't
9 consider them members of the drug group, nor do I
10 consider them members of the placebo group. The
11 single case of mild opiate withdrawal noted in the
12 2/2 group isn't particularly convincing.
13 Trial 205 is slightly more persuasive, but
14 the numbers remain quite small. At visit 13,
15 BUP/SAM provides 5 cases of mild opiate withdrawal
16 compared to placebo's 1 case of mild opiate
17 withdrawal. At visit 14, BUP/SAM provides 9 cases
18 of mild opiate withdrawal plus 1 case of moderate
19 opiate withdrawal compared to placebo's 4 cases of
20 mild opiate withdrawal.
21 Here is where the problem of missing data
22 rears its head. Notice that the majority of each
A Matter of Record (301) 890-4188 187
1 line represents missing data. At visit 12, BUP/SAM
2 provides 2 cases of mild opiate withdrawal compared
3 to placebo's zero cases of mild opiate withdrawal.
4 No cases of opiate withdrawal were noted at
5 visit 13.
6 Once again, take note of how much data are
7 missing. Despite the missing data, we observe an
8 upswing in mild opiate withdrawal cases in the 2/2
9 drug group. At visit 13, BUP/SAM provides 5 cases
10 of mild opiate withdrawal compared to placebo's 2
11 cases of mild opiate withdrawal. No cases of
12 opiate withdrawal were noted at visit 14.
13 While none of the COWS data is particularly
14 strong on its own, the trends observed in the
15 50 percent of COWS data submitted suggest that mild
16 opiate withdrawal occurs in a subset of
17 participants exposed to BUP/SAM.
18 I'd now like to hand the presentation off to
19 Dr. Edward Hawkins. He will be providing the
20 controlled substance portion of FDA's presentation.
21 FDA Presentation - Edward Hawkins
22 DR. HAWKINS: Good morning. My name's Ed
A Matter of Record (301) 890-4188 188
1 Hawkins, and I'm a reviewer with the controlled
2 substance staff. Today, I'll be speaking to you
3 about the abuse potential of buprenorphine/
4 samidorphan, also known as BUP/SAM, and of
5 samidorphan alone.
6 The applicant is proposing a new drug
7 product that combines the mu partial agonist,
8 buprenorphine, with samidorphan, a drug with mu
9 opioid antagonist properties, formulated as
10 sublingual tablets, also known as BUP/SAM.
11 In vitro studies indicate that it is
12 possible to separate to some degree the
13 buprenorphine from samidorphan and in BUP/SAM. No
14 studies were conducted to determine the safety or
15 abuse liability of BUP/SAM after manipulation.
16 This includes studies regarding the different
17 methods of administration that are typically
18 associated with abuse, for example intranasal.
19 Now, as I mentioned, the sponsor is required
20 to assess the abuse potential of samidorphan, which
21 is a new molecular entity that is chemically
22 synthesized from thebaine-derived naltrexone.
A Matter of Record (301) 890-4188 189
1 Under the Controlled Substances Act, all
2 derivatives of thebaine are Schedule II substances
3 until such time as they are down-scheduled or
4 de-controlled following an abuse potential
5 assessment. Thus, it was necessary to conduct an
6 abuse potential assessment for both samidorphan
7 alone and in combination with buprenorphine.
8 In vitro receptor binding and functional
9 studies show that samidorphan acts at mu opioid
10 receptors as an antagonist. Although samidorphan
11 is derived from an opioid, thebaine, administration
12 of samidorphan alone does not produce analgesia in
13 animals.
14 However, it can reduce the analgesic effects
15 of the mu opioid receptor agonist, morphine, and
16 can reverse the cardiac and respiratory depressant
17 effects of the mu opioid agonist, fentanyl. These
18 data demonstrate that in whole animals, samidorphan
19 has activity as a mu opioid antagonist.
20 A briefing to drugs or introduction to drug
21 discrimination studies; drug discrimination is an
22 experimental method of determining whether a test
A Matter of Record (301) 890-4188 190
1 drug produces pharmacological effects, which elicit
2 physical and behavioral responses in the animal
3 that are similar to their responses to a training
4 drug, typically a known drug of abuse. Test drugs
5 that produce a response similar to a training drug
6 with known abuse potential are also likely to be
7 abused by humans.
8 In this study, rats trained to discriminate
9 morphine from vehicle indicated that morphine
10 produced full generalization of 97 percent to the
11 morphine queue. However, samidorphan produced no
12 generalization, 5.4 percent, to the morphine queue.
13 These data show that samidorphan did not
14 produce sensations that are similar to those
15 produced by morphine. This was expected since
16 samidorphan is a mu opioid antagonist.
17 Now, a brief introduction to self-
18 administration studies, which is a method that
19 assesses whether a test drug produces rewarding
20 effects that increase the likelihood of behavioral
21 responses in order to obtain additional drug, also
22 known as positive reinforcement.
A Matter of Record (301) 890-4188 191
1 Drugs that are self-administered by animals
2 are likely to produce rewarding effects in humans.
3 The ability of a test drug to produce self-
4 administration is indicative that the drug has
5 abuse potential.
6 In this study, rats learned to lever press
7 for intravenous heroin as the training drug, after
8 self-administration of heroin was stable, and also
9 were allowed IV access to the following substances,
10 which produced varying degrees of self-
11 administration measured as infusions per session.
12 Samidorphan produced 9.2 infusions. Heroin
13 produced double that, at 18.8 infusions, naltrexone
14 produced the same as samidorphan, 8.1, and placebo
15 produced less than 5 infusions.
16 These data show that samidorphan does not
17 produce rewarding properties that sustain positive
18 reinforcement, similar to naltrexone.
19 Now, I'll speak about the human abuse
20 potential studies, also known as HAP studies, which
21 are used to evaluate the ability of a test drug to
22 produce positive subjective responses in subjects
A Matter of Record (301) 890-4188 192
1 compared to a known drug of abuse or a positive
2 control and to placebo.
3 Subjects in HAP studies are individuals with
4 a history of recreational drug use, but they are
5 not drug dependent. When the test drug produces
6 consistently large responses on positive subjective
7 scales that are far outside the acceptable placebo
8 range, it is likely that the test drug has abuse
9 potential.
10 The first of two HAP studies evaluated the
11 oral abuse potential of samidorphan alone at 2.5,
12 10, and 20 milligrams; oxycodone at 15 and 30
13 milligrams; and placebo. This study used a
14 randomized double-blind placebo-controlled
15 crossover design in healthy, non-dependent,
16 recreational opioid users.
17 The primary measure of drug liking using the
18 Visual Analog Scale as a bipolar scale from zero,
19 which is extreme disliking, to 100 of extreme
20 liking, with 50 as neutral, indicated that the
21 positive control drug, oxycodone, at both doses
22 produced statistically significantly higher mean
A Matter of Record (301) 890-4188 193
1 drug liking scores compared to placebo, which
2 establishes assay sensitivity and validates the
3 study.
4 Samidorphan at all 3 doses; 2.5, 10, and 20
5 milligrams, produced mean drug liking scores of 57,
6 58, and 60, respectively, that are within the
7 placebo range from 40 to 60. The secondary
8 measures of overall drug liking, high, good drug
9 effects, and take drug again for oxycodone produced
10 mean drug scores on each of these positive
11 subjective measures that were statistically
12 significantly greater than placebo.
13 Samidorphan at all 3 doses; 2.5, 10, and
14 20 milligrams, produced mean scores on each of
15 these positive subjective measures that were within
16 the placebo range.
17 The second of two HAP studies evaluated the
18 abuse potential of samidorphan alone, again at
19 higher doses, 10 and 30 milligrams, oxycodone,
20 pentazocine, naltrexone, and placebo. This study
21 used a randomized, double-blind, double-dummy
22 placebo-controlled crossover design in healthy,
A Matter of Record (301) 890-4188 194
1 non-dependent, recreational opioid users.
2 Again, the primary measure of drug liking on
3 a bipolar VAS scale from 0 to 100 with 50 as
4 neutral indicated that the positive control drugs,
5 oxycodone and pentazocine, produced statistically
6 significantly higher mean drug scores compared to
7 placebo, which establishes assay sensitivity and
8 validates the study.
9 Samidorphan at both doses, 10 and
10 30 milligrams, produced mean drug liking scores of
11 59 and 61, respectively, that were statistically
12 within the placebo range of 40 to 60. This was
13 also observed for naltrexone, which produced a
14 score of 58.
15 The secondary measures, measured by using a
16 VAS of overall drug liking, high, good drug
17 effects, and take drug again, indicate that
18 oxycodone and pentazocine produce mean scores on
19 each of these positive subjective measures that
20 were statistically significantly greater than
21 placebo. Samidorphan at both doses produced mean
22 scores on each of these positive subjective
A Matter of Record (301) 890-4188 195
1 measures that were within the placebo range. This
2 was also observed for naltrexone.
3 A third HAP study evaluated the abuse
4 potential of sublingual tablets of buprenorphine
5 and samidorphan at 2 and 2 milligrams, 8 and 8
6 milligrams, and 16 and 16 milligrams, buprenorphine
7 alone at 8 and 16 milligrams, and placebo. This
8 study used a randomized, double-blind, placebo-
9 controlled crossover design in healthy, non-
10 dependent recreational opioid users.
11 The primary measure of drug liking measured
12 on a bipolar VAS from 0 to 100 with 50 as neutral
13 indicated that the positive control drugs,
14 buprenorphine, at both doses, 8 and 16 milligrams,
15 produce statistically significantly higher mean
16 drug scores of 76 and 82 compared to placebo at 52,
17 which establishes assay sensitivity and validates
18 the study.
19 Samidorphan plus buprenorphine at all 3
20 doses; 2 and 2, 8 and 8, and 16 and 16 milligrams;
21 produced mean drug liking scores of 60, 61, and 64,
22 respectively, that were barely outside of the
A Matter of Record (301) 890-4188 196
1 placebo range.
2 The secondary measures of overall drug
3 liking, high, good drug effects, and take drug
4 again for buprenorphine produced mean scores on
5 each of these positive subjective measures that
6 were statistically significantly greater than
7 placebo.
8 Samidorphan plus buprenorphine, again at all
9 3 doses, 2/2, 8/8, and 16/16 milligrams, produced
10 mean scores on each of these positive subjective
11 measures that were either within the placebo range
12 or barely outside of this range.
13 As a result, the conclusions from the HAP
14 studies indicate that two single-dose HAP studies
15 conducted with samidorphan indicate that the drug
16 does not produce positive subjective responses. In
17 a third HAP study, the combination of samidorphan
18 and buprenorphine produced positive subjective
19 responses that were much less than those produced
20 by buprenorphine alone. However, they were
21 slightly outside of the placebo range.
22 Our final conclusions indicate that animal
A Matter of Record (301) 890-4188 197
1 and human studies consistently show that
2 samidorphan is a mu opioid antagonist with no
3 meaningful abuse potential, similar to the mu
4 opioid antagonist naltrexone.
5 BUP/SAM produces subjective responses that
6 are lower than those from administration of the
7 same dose of BUP alone, but does retain some
8 clinically significant abuse potential. Overall,
9 the positive subjective responses reported after
10 administration of BUP/SAM showed that the
11 combination has a low potential for abuse.
12 I will now pass it on to the next speaker,
13 Celeste Mallama.
14 FDA Presentation - Celeste Mallama
15 DR. MALLAMA: Good morning. My name is
16 Celesta Mallama. I'm a reviewer for the Division
17 of Epidemiology, and I'll be presenting the FDA
18 review of the epidemiologic and surveillance data
19 related to the misuse and abuse of buprenorphine
20 products.
21 In July of 2017, the National Academies of
22 Sciences, Engineering, and Medicine issued the
A Matter of Record (301) 890-4188 198
1 report, Pain Management and the Opioid Epidemic:
2 Balancing Societal and Individual Benefits and
3 Risks of Prescription Opioid Use. The NASEM
4 committee's charge was to help the FDA develop a
5 framework for opioid review, approval, and
6 monitoring that balances the individual patient
7 needs with the broader public health consequence of
8 opioid misuse.
9 The report suggests that such an approach to
10 evaluation of benefit-risk should include assessing
11 evidence of a product's potential for misuse and
12 diversion and predicted risks to family members and
13 society.
14 The purpose of this presentation is to
15 inform the broad consideration of the benefit-risk
16 balance for buprenorphine samidorphan by addressing
17 the following topics. First, I will present the
18 use, misuse, and abuse of currently marketed
19 buprenorphine and buprenorphine naloxone products,
20 after which Dr. Mark Sullivan will discuss the
21 complex relationships between depression, pain, and
22 substance use disorders.
A Matter of Record (301) 890-4188 199
1 The goal of this presentation is to provide
2 contextual information for discussion of the
3 overall benefit-risk balance on the product under
4 review. This presentation of real-world
5 buprenorphine abuse patterns is not intended as a
6 prediction of expected abuse of buprenorphine
7 samidorphan or it to be approved and marketed.
8 This graph, generated by FDA using IQVIA
9 national prescription audit data shows national
10 projections for the number of prescriptions for
11 buprenorphine-containing oral solid formulations
12 dispensed from U.S. outpatient retail pharmacies.
13 On the Y-axis are number of prescriptions in
14 millions and on the X-axis are years from 2013 to
15 2017.
16 These numbers demonstrate that the vast
17 majority of the market for oral solid
18 buprenorphine-containing products consist of
19 products indicated for opioid dependence as part of
20 medication-assisted treatment as opposed to pain,
21 and that dispensings of buprenorphine-naloxone
22 combination products, shown on the graph, in the
A Matter of Record (301) 890-4188 200
1 solid and dotted black lines, greatly outnumber
2 buprenorphine single-entity products shown on the
3 graph in the solid and dotted gray lines.
4 Buprenorphine transdermal patches, injectable, and
5 implant products are not included in these data.
6 This slide shows the proportion of
7 respondents reporting past-year misuse of
8 prescription pain relievers in the National Survey
9 on Drug Use and Health. This is a nationally
10 represented population survey of individuals 12 and
11 older in the United States.
12 Of note, the survey defines misuse as any
13 use other than that as directed by a healthcare
14 provider, and therefore includes non-medical use
15 for a therapeutic purpose such as taking someone
16 else's medication or more than directed to treat
17 pain, anxiety, or withdrawal, or to achieve a high.
18 The graph on the left shows the percentage
19 misusing each opioid among the total population.
20 As you can see, buprenorphine was less frequently
21 misused in the general population overall than more
22 commonly prescribed opioids like hydrocodone and
A Matter of Record (301) 890-4188 201
1 oxycodone.
2 The graph on the right shows the proportion
3 of respondents reporting misuse in people who used
4 that opioid in the last year, including use as
5 directed. Buprenorphine has the second highest
6 proportion of users reporting misuse within the
7 population of people using the specified drug.
8 An important point to keep in mind when
9 interpreting the data on the previous slide is that
10 the majority of buprenorphine dispensed in the U.S.
11 is for evidence-based treatment for opioid use
12 disorder.
13 Therefore, compared to the general
14 population who use other prescription opioid
15 analgesics, buprenorphine is disproportionately
16 dispensed to those with existing opioid use
17 disorder who are at elevated risk for misusing and
18 abusing opioids, including through injection or
19 other non-oral routes; or diverting it within
20 social networks with individuals with opioid use
21 disorder.
22 Therefore, directly comparing levels of
A Matter of Record (301) 890-4188 202
1 buprenorphine misuse and abuse to other opioids can
2 be challenging and potentially misleading. The
3 same caveat applies to the data on routes of
4 buprenorphine abuse that I will present next.
5 Several different data sources indicate that
6 abuse of marketed buprenorphine products through
7 non-oral routes is fairly common in cases that come
8 to medical attention among those abusing the
9 products, even when buprenorphine is in combination
10 with naloxone.
11 This graph, generated by FDA, contains data
12 extracted from the National Poison Data System, a
13 centralized data source maintained by the American
14 Association of Poison Control Centers that captures
15 information on a near real-time basis from a
16 national network of poison centers receiving calls
17 from the public or healthcare workers.
18 On the Y-axis is the percent of abuse cases
19 reporting exposure by each route and on the X-axis
20 are bars for single-ingredient buprenorphine
21 products and buprenorphine naloxone products.
22 Injection route is shown in black and inhalation
A Matter of Record (301) 890-4188 203
1 nasal route is shown in gray.
2 About a quarter of the poison control
3 centers' calls from 2013 to 2017 involving
4 buprenorphine abuse reported injection of the drug,
5 and this was similar for single-ingredient and
6 buprenorphine-naloxone combination products. You
7 do also see some inhalation nasal abuse, but it is
8 less frequent than injection.
9 This graph, also generated by FDA, contains
10 information from the National Electronic Injury
11 Surveillance System, cooperative adverse drug
12 events surveillance, a nationally representative
13 sample of emergency department visits in the U.S.,
14 drug from manual abstraction of clinical records by
15 trained coders.
16 The graph shows percentage of abuse cases
17 reporting injection route on the Y-axis and on the
18 X-axis are bars for single-ingredient buprenorphine
19 and buprenorphine-naloxone combination products.
20 The NEISS-CADES data source was updated in
21 2016 to include ED visits related to abuse. In
22 these emergency department data from 2016 to 2017,
A Matter of Record (301) 890-4188 204
1 more than 40 percent of cases involving abuse of
2 buprenorphine were related to injection of the
3 drug. Many of these were for infectious
4 complications of injection drug use. National
5 estimates for cases involving inhalation abuse did
6 not meet criteria for data precision.
7 In the 2014 article, Cicero, et al. reported
8 that, in a sample of individuals entering substance
9 use disorder treatment, 34.4 percent of those
10 reporting past-month buprenorphine use to get high
11 indicated that they had injected it in the month
12 prior to treatment.
13 Among respondents entering treatment centers
14 who indicated buprenorphine injection in the past
15 month, 43.6 percent injected buprenorphine-naloxone
16 tablets. Participants reported a number of simple
17 methods which they believed separated buprenorphine
18 from naloxone, resulting in what they believed to
19 be pure buprenorphine for injection; again, keeping
20 in mind that the population abusing buprenorphine
21 may be particularly enriched with people with
22 advanced substance use disorder who are more likely
A Matter of Record (301) 890-4188 205
1 to be experienced injection drug users.
2 To summarize the epidemiological data, the
3 vast majority of the market for oral solid
4 buprenorphine-containing products consist of
5 products indicated for opioid dependence as part of
6 evidence-based medication-assisted treatment as
7 opposed to pain. Abuse of both single-ingredient
8 buprenorphine and buprenorphine-naloxone
9 combination products is common and occurs through
10 both oral and non-oral routes, including injection.
11 In a survey of patients entering treatment
12 center with previous experience with buprenorphine,
13 respondents reported a number of methods that they
14 believe separated buprenorphine from naloxone,
15 resulting in what they termed to be pure
16 buprenorphine for injection.
17 However, comparing buprenorphine abuse
18 patterns with those other opioids and using these
19 patterns to try to predict what might happen in
20 patients being treated for depression is
21 challenging due to the high-risk nature of the
22 populations being treated with buprenorphine for
A Matter of Record (301) 890-4188 206
1 medication-assisted treatment.
2 The buprenorphine samidorphan under review
3 has a new orally bioavailable antagonist and is
4 indicated for a new population. Therefore, it is
5 not known whether similar misuse and abuse patterns
6 will be seen.
7 Considering the benefit-risk balance of a
8 new opioid-containing product, it is valuable to
9 understand the population in which it is likely to
10 be used and any safety concerns that may arise from
11 its use in a real-world setting.
12 In the epidemiologic review, in the FDA
13 background package, we included some information
14 from the published literature describing the
15 complex relationships between depression, pain, and
16 substance use disorders. We have asked Dr. Mark
17 Sullivan, an expert in this area, to speak on this
18 topic, again to inform the discussion on the
19 overall benefit-risk balance for the product under
20 review.
21 Dr. Mark Sullivan is a psychiatrist at the
22 University of Washington Medical Center for Pain
A Matter of Record (301) 890-4188 207
1 Relief and Regional Heart Center. Dr. Sullivan is
2 a University of Washington professor of psychiatry
3 and behavioral sciences, adjunct professor of
4 anesthesiology and pain medicine, and adjunct
5 professor of bioethics and humanities.
6 DR. SULLIVAN: Hello This is Mark Sullivan.
7 Can people hear me?
8 MS. BHATT: Yes, we can hear you,
9 Dr. Sullivan.
10 Presentation - Mark Sullivan
11 DR. SULLIVAN: So I will begin my slides.
12 I've been asked to talk about depression effects on
13 long-term prescription opioid use, abuse, and
14 addiction, and I begin my presentation with a
15 picture here of a PET scan showing mu opioid
16 receptors throughout areas of the brain thought to
17 be primarily related to depression.
18 Just a few disclosures, I do have some
19 grants from drug companies, from Pfizer to develop
20 a pain self-management support tool; from Purdue to
21 look at the effectiveness of primary care opioid
22 taper plans; and I have a number of federal grants
A Matter of Record (301) 890-4188 208
1 in the opioid area as well as some consulting with
2 Aetna about pharmacy benefits; Chrono Therapeutics
3 about an opioid taper device; and with the State of
4 Washington about opioid policy.
5 So let me begin with an introduction about
6 the epidemiology of chronic pain and depression.
7 There's a great deal of overlap between these two
8 conditions with a greater than 50 percent of
9 prevalence of major depression among patients
10 seeking care in chronic pain specialty settings.
11 Similarly, there's more than 50 percent
12 prevalence of chronic pain in patients seeking care
13 for depression. This has been examined thoroughly
14 over the years. In a multi-site WHO study, the
15 conclusion was that pain and depression,
16 particularly chronic pain, that it's activity
17 limiting and is mutually reinforcing with
18 depression such that each causes the other.
19 It has been shown that depression,
20 unfortunately, although it increases the likelihood
21 of opioid therapy, which I will show you data
22 about, it actually decreases the responsiveness of
A Matter of Record (301) 890-4188 209
1 pain to opioids. That's been shown by I.J. Wasan
2 in experimental study 2005 and in a clinical study
3 in 2015.
4 Pain also decreases responsiveness to
5 antidepressants and even to combined pharmaco and
6 psychodepression therapy, shown by Matt Bair and
7 Steven Thielke. Chronic pain and depression have
8 quite a similar neuroscience. They show similar
9 patterns of neural activation on functional scans
10 such as fMRIs and PET scans, and they respond to
11 many of the same medications, such as
12 antidepressants.
13 There is interesting interaction also
14 between chronic pain and substance use disorders.
15 We know, from many studies, that chronic pain is
16 common in patients with substance use disorders.
17 This has been reported between 27 to 87 percent
18 prevalence in chronic pain population, and
19 substance use populations will have chronic pain.
20 Patients with chronic pain are also thought
21 to be 2 to 3 times more likely to develop substance
22 use disorders. There are not as many studies in
A Matter of Record (301) 890-4188 210
1 this area. I list a couple of studies here.
2 Recent reviews of the neuroscience of chronic pain
3 and addiction have pointed towards the reward
4 deficiency as a link between chronic pain and
5 substance use disorders.
6 Chronic pain is characterized by low
7 dopamine turnover in reward centers, the experience
8 of anhedonia or lack of pleasure, and a set of what
9 are called anti-reward adaptations, where normally
10 rewarding activities no longer are rewarding. This
11 results in high salience of pain relief, but low
12 salience for other rewards, and I list a couple of
13 papers here that discuss this hypothesis.
14 So what about opioid therapy for chronic
15 pain and depression? It's first important to
16 remember that there are no treatment guidelines
17 recommending long-term opioid therapy for pain and
18 depression. There are in fact no RCTs of opioid
19 therapy for chronic pain in depressed patients. In
20 fact, virtually all of the RCTs of opioid therapy
21 for chronic pain have excluded depressed patients,
22 so in fact we have virtually no evidence base for
A Matter of Record (301) 890-4188 211
1 that practice, which is common.
2 Opioid therapy for depressed patients is
3 very common. One study that we were involved in
4 showed that, among HMO patients who had recent
5 depression within the past 2 years, opioid therapy
6 was 3 times more common than patients without a
7 history of depression. In addition to being more
8 common, depressed patients received higher daily
9 opioid doses, more days supplied, and were more
10 likely to get Schedule II opioids.
11 These findings have been replicated in other
12 samples of commercially insured patients, Medicaid
13 patients, and veteran patients. All of these
14 studies have shown that mental health disorders are
15 especially common in high-dose long-term opioid
16 patients.
17 My interpretation of this data is that
18 there's a pattern of adverse selection rather than
19 careful selection. Opioid treatment guidelines
20 have generally suggested that clinicians carefully
21 select chronic pain patients without substance use
22 and mental health disorders for opioid therapy, but
A Matter of Record (301) 890-4188 212
1 what's happening in actual practice is the opposite
2 of this.
3 So what I have called adverse selection is
4 defined as the selection of high-risk patients with
5 substance abuse and mental health disorders for a
6 high-risk opioid regimen. So what we see is that,
7 among these patients with substance abuse and
8 mental health disorders, they have higher rates of
9 opioid use, especially long term. They have higher
10 daily doses of opioids. They tend to have a longer
11 duration of therapy in days or years. They are
12 more likely to receive multiple opioids at the same
13 time, and they are more likely to receive
14 concurrent sedatives such as benzodiazepines or
15 muscle relaxants.
16 I recently spoke at the NIH Pain
17 Collaboratory meeting, where I argued that this
18 essentially results in a reverse treatment
19 disparity with vulnerable patients overtreated with
20 opioids, and I've written about this in a number of
21 papers.
22 Other investigators have also identified
A Matter of Record (301) 890-4188 213
1 patterns of opioid prescription and receipt
2 consistent with adverse selection hypothesis.
3 Karen Seal from UCSF studied 140,000 Iraq or
4 Afghanistan veterans with chronic pain and found
5 that opioids were received by 6 percent of those
6 veterans without mental health disorders,
7 12 percent of those with non-PTSD mental health
8 disorders, mostly depression, and 18 percent of
9 those with PTSD. So we get a doubling of the
10 opioid use rate in those with depression and then a
11 tripling among those with PTSD.
12 These mental health groups were also
13 characterized by receiving higher-dose opioids,
14 multiple opioid sedatives, early refills, and a
15 higher rate of adverse events from the opioid.
16 Davis, et al, published a study explaining that the
17 16 percent of Americans with mental health
18 disorders receive over half of all the opioids
19 prescribed in the U.S. Odds ratio for receiving
20 opioids for mental health patients was about 2.
21 A very large study done by Quinn, et al. in
22 2017 showed that among 10 million commercially
A Matter of Record (301) 890-4188 214
1 insured patients, mental health and substance use
2 disorders were associated with opioid therapy. The
3 likelihood of receiving opioids was about double
4 for those with an anxiety or mood disorder, and it
5 was about triple for those with a non-opioid
6 substance use disorder, and a 9-fold increase among
7 those with an opioid substance use disorder.
8 Some interesting recent studies have
9 revealed adverse selection of patients with
10 depression into long-term high-dose opioid therapy
11 as a process of self-selection. Halbert, et al.
12 showed that while depressed patients initiate
13 opioid therapy for chronic pain, only slightly more
14 often than non-depressed patients with chronic
15 pain, but they are twice as likely to transition
16 from short-term use to long-term use, and they're
17 self-selecting because they don't quit opioids,
18 which is what most patients do.
19 Jenna Goesling at Michigan has shown that
20 depressed patients appear to continue opioid use at
21 lower pain intensity levels and higher levels of
22 physical function than do non-depressed patients.
A Matter of Record (301) 890-4188 215
1 Alicia Grattan with me showed that depressed
2 patients tend to overuse opioids because they use
3 them to treat insomnia and stress.
4 Unfortunately, opioids are not an effective
5 treatment for depression or anxiety. They have
6 been used for this. One must remember that opioids
7 are the first psychotropic medication really ever
8 discovered and used. And as far back as the Greek
9 physician Hippocrates or the Roman physician,
10 Galen, opioids have been used to treat both mania
11 and melancholia. In fact, if you look in 19th and
12 20th century psychiatric textbooks, opioids are
13 recommended for these purposes. However, as I
14 mentioned before, there are no controlled studies
15 showing lasting relief of opioids for depression or
16 anxiety.
17 They may provide partial relief from anxiety
18 and insomnia, but particularly in patients with
19 PTSD, they appear to deepen avoidance and
20 deactivation and prolong the course of PTSD.
21 Now, it's important to note that
22 buprenorphine may be an exception in this regard
A Matter of Record (301) 890-4188 216
1 due to kappa antagonism, which has been written
2 about as a potential antidepressant mechanism by a
3 variety of investigators that I list here.
4 It's important to also know that mu opioids
5 may also cause depression. Generally, it's been
6 shown that opioid therapy that extends beyond
7 90 days or has a daily dose over 50 milligrams of
8 morphine or equivalent may increase the risk of
9 depression, according to a series of retrospective
10 cohort studies.
11 In general, rapid dose increase seems to
12 have the highest depression risk. Most of these
13 studies have been done by Scherrer and colleagues,
14 and they've also shown that recurrent incident and
15 treatment-resistant depression are all more likely
16 in opioid-treated patients.
17 Other investigators have shown that the most
18 vulnerable patients to opioid-induced depression
19 are those with low-pain self-efficacy; that is, low
20 confidence to continue with daily activities; poor
21 social support, and who began using their opioids
22 at a younger age.
A Matter of Record (301) 890-4188 217
1 Depression also appears to increase the
2 rates at which patients proceed from use to abuse
3 and addiction. Depression or a psychological
4 behavior called catastrophizing, characterized by
5 hopeless and helpless cognition, increases the risk
6 of misuse, non-medical use, and abuse of
7 prescription opioids among adults and adolescents.
8 Among adults with non-medical use,
9 depression doubles the risk of progressing to full-
10 fledged opioid use disorder. This in fact may be
11 the path by which depression increases the risk of
12 opioid use disorder among patients with chronic
13 pain, as it pushes them down this path from use to
14 misuse to abuse to addiction or opioid use
15 disorder.
16 Adverse selection is prominent among opioid-
17 treated patients, but it's even more marked among
18 those patients receiving both opioids and
19 benzodiazepines. We know from national data that
20 opioid prescribing has declined since 2012, but not
21 the prescribing of opioids with benzos.
22 Between 2001 and 2013, concurrent opioid-
A Matter of Record (301) 890-4188 218
1 benzo use doubled among privately insured patients,
2 and its estimated increased risk of opioid overdose
3 is between 2 and 10, depending on how well
4 confounding is controlled for. We know that about
5 10 percent of patients initiating antidepressants
6 also initiate benzodiazepines. That's sometimes
7 due to sooth anxiety or insomnia before the
8 antidepressants kick in. However, about 12 percent
9 of these people who initiate benzodiazepines with
10 their antidepressants become long-term
11 benzodiazepine users.
12 As we look for recent trends, there's
13 increasing rate of simultaneous antidepressant and
14 benzodiazepine use. Six percent of antidepressant
15 starts included benzodiazepine, also in 2001 up to
16 12 percent by 2012.
17 There is evidence of opioid dysregulation in
18 major depression. We see reduced opioid receptor
19 availability in major depressive disorder, and this
20 is particularly marked after sadness induction, as
21 mentioned by these two studies. This reduced
22 opioid receptor availability is associated with
A Matter of Record (301) 890-4188 219
1 reduced response to SSRI antidepressants and
2 increased adrenocorticotropic hormone levels.
3 Greater opioid release after social
4 rejection is associated with resiliency and lower
5 affect, so that's a basic release that's adaptive,
6 and what we see in major depression is a higher
7 tonic level of opioids, and that's what seems to be
8 associated with opioid problems.
9 If we look at the opioid receptor gene,
10 particularly the G-allele, that is associated with
11 the kind of psychiatric problems such as higher
12 neuroticism or a tendency to experience negative
13 emotional states and higher rates of depression.
14 We see in these patients with the G-allele,
15 they have greater reactivity to social rejection,
16 and this is manifested in altered responsivity of
17 the anterior cingulate cortex and the insula
18 cortex, as summarized by Pecina in a recent paper.
19 So there are some important and relevant
20 non-pain functions of exogenous opioids. Here, I'm
21 specifically speaking of mu opioid effects. We've
22 known for quite some time that mu opioids don't
A Matter of Record (301) 890-4188 220
1 simply address physical pain, but suppress
2 suppression distress, separation distress, and
3 that's found widely in mammals, whether they be
4 rodents, primates, or human.
5 We've known and confirmed recently that
6 opioids reduce the affective, emotional, or
7 affective aspect of pain more than the sensory
8 aspect. And as I mentioned, in terms of opioid use
9 among pain patients, it provides more relief of
10 stress, anxiety, and insomnia than mood problems,
11 which may worsen.
12 Interestingly, in our recent opioid taper
13 study, a number of patients reported that they no
14 longer felt like zombies, and their spouses
15 confirmed they have returned to their pre-opioid
16 personalities, and there have been other evidence
17 that opioids, specifically mu opioids, impair the
18 human capacity for emotion perception and social
19 inferences. This is one Australian study that I
20 cite here.
21 So in conclusion, the relationship between
22 opioids and depression is close, complex, and
A Matter of Record (301) 890-4188 221
1 multifaceted. Depression is associated with
2 endogenous opioid dysfunction. Depression is
3 associated with opioid misuse, non-medical use,
4 abuse, and opioid use disorder, and endogenous mu
5 opioids prescribed long term may increase the risk
6 of depression and social cognition deficits.
7 Thank you. That's the end of my remarks.
8 FDA Presentation - Sonya Dunn
9 DR. DUNN: Thank you, Dr. Sullivan.
10 My name is Somya Dunn, and good morning.
11 I'm from the Division of Risk Management. I'm
12 going to present a discussion on risk management
13 for buprenorphine samidorphan.
14 First, I'm going to present the background
15 on risk evaluation and mitigation strategies or
16 REMS. I'll present a summary of the REMS for
17 buprenorphine-containing products, potential safety
18 concerns associated with the use of buprenorphine
19 samidorphan or BUP/SAM, and possible risk
20 management strategies for BUP/SAM.
21 I will begin with REMS. A REMS is a drug
22 safety program that can be required by the FDA for
A Matter of Record (301) 890-4188 222
1 certain drugs. A REMS is designed to mitigate
2 risks associated with drug use and includes
3 strategies beyond labeling to ensure benefits
4 outweigh the risks of the drug.
5 The FDA Amendments Act of 2007 gave the FDA
6 authorization to require applicants and application
7 holders to develop and comply with REMS program if
8 determined necessary. The FDA has the authority to
9 require a REMS pre- or post-approval.
10 A REMS can include a number of components
11 such as a medication guide, a communication plan,
12 elements to assure safe use or ETASU, an
13 implementation system, and must include a time
14 table for submission of assessments.
15 If determined a necessary component of a
16 REMS, the elements to assure safe use can include
17 the following: certification and/or specialized
18 training of healthcare providers that prescribe the
19 drug; certification of pharmacies or other
20 dispensers of the drug; limited settings for
21 dispensing or administration of the drug; having
22 each patient using the drug subject to certain
A Matter of Record (301) 890-4188 223
1 monitoring;, the drug is dispensed, administered
2 only with evidence of safe use conditions, for
3 example a pregnancy test or liver function test or
4 enrollment of treated patients in a registry.
5 Additionally, an ETASU must align with the
6 serious risks listed in the labeling. They cannot
7 cause undue burden on patient access to the drug,
8 considering in particular patients with serious or
9 life-threatening diseases or conditions and
10 patients who have difficulty accessing healthcare.
11 I will continue on to a summary of relevant
12 REMS. All the buprenorphine-containing products
13 approved for either treatment of opioid dependence
14 or pain and that are intended for use in an
15 outpatient setting are covered by a REMS.
16 These are the four types of REMS for the
17 buprenorphine-containing products. The first REMS
18 is actually two different programs due to two
19 different sponsor groups, however, the program
20 requirements are the same. The four types of
21 programs are the BTODs, suboxone/Subutex program,
22 the probuphine program, the sublocade program, and
A Matter of Record (301) 890-4188 224
1 the opioid analgesic REMS program.
2 The first three REMS are for products
3 indicated for opioid dependence. These products
4 have required SAMHSA training, which I will discuss
5 shortly. This training is not part of the REMS.
6 The suboxone/Subutex and BTOD REMS as well as the
7 opioid analgesic REMS, the circled programs,
8 mitigate risks associated with opioids such as
9 accidental overdose, misuse, abuse, and addiction.
10 The middle two programs, probuphine and the
11 sublocade REMS, are designed to mitigate risks
12 associated with the formulation of the product.
13 The circled programs are most relevant to our
14 discussion today.
15 The suboxone, Subutex, and BTOD REMS consist
16 of provider and pharmacy educational materials and
17 an appropriate use checklist. Training is not
18 required as part of the REMS. However, there is
19 SAMHSA training required for providers prescribing
20 for treatment of opioid use disorder.
21 Buprenorphine-containing products used for
22 opioid use disorder can be prescribed outside of an
A Matter of Record (301) 890-4188 225
1 opioid treatment program, or OTP, if the provider
2 obtains a DATA 2000 waiver. The Substance Abuse
3 and Mental Health Services Administration, or
4 SAMHSA, manages the Drug Addiction Treatment Act of
5 2000. They set eligibility and certification
6 requirements for the DATA 2000 waiver, which
7 includes required training. Physicians that hold
8 the DATA 2000 waiver can prescribe and/or dispense
9 buprenorphine products for opioid use disorder in
10 settings other than OTPs.
11 Buprenorphine-containing products approved
12 for treatment of pain are covered in the opioid
13 analgesic REMS. This REMS requires manufacturers
14 to make training available to healthcare providers
15 involved in the management of patients with pain.
16 Training is not required to prescribe or dispense,
17 and the training focuses on pain management,
18 identifying risk factors for abuse and addiction,
19 how to counsel patients and their families on the
20 safe use of opioids, and fundamentals of addiction
21 medicine.
22 I will move on to the potential safety
A Matter of Record (301) 890-4188 226
1 concerns of BUP/SAM. The agency has concerns
2 regarding BUP/SAM for depression. In the U.S., the
3 majority of patients with depression are treated by
4 primary care clinicians. It is important for
5 prescribers to understand that BUP/SAM contains an
6 opioid. Use of BUP/SAM for depression will be
7 chronic, and this could lead to addiction,
8 dependence, or withdrawal.
9 In addition, there are safety concerns
10 regarding concomitant opioid and benzodiazepine
11 use. Mental illness, including depression, is
12 associated with comorbidities, including pain
13 conditions, opioid use disorder, and anxiety.
14 Concomitant use of an opioid agonist or
15 benzodiazepine could put patients at risk for
16 respiratory depression. Labeling for approved
17 opioid products does address this risk.
18 Mental illness, including depression, is
19 associated with misuse and abuse of prescription
20 opioids and higher doses of opioids may be needed
21 for analgesia if patients are taking BUP/SAM.
22 There are safety concerns in women of
A Matter of Record (301) 890-4188 227
1 childbearing potential. The prevalence of moderate
2 to severe major depressive disorder is 9.3 percent
3 in women ages 18 to 39. Women who are pregnant may
4 be putting their unborn infants at risk for
5 development of neonatal opiate withdrawal syndrome
6 or NOWS. This was not evaluated in the clinical
7 program for BUP/SAM. Labeling for approved
8 buprenorphine provides addresses the risk of NOWS.
9 The agency is also considering that patients
10 often struggle with adherence to therapy for many
11 chronic conditions, including depression. Given
12 that BUP is a partial agonist and SAM is an opioid
13 antagonist, it is unclear what impact lack of
14 adherence to BUP/SAM could have if concomitant
15 opioid agonists are used.
16 There is a possibility that changing mu
17 agonist and antagonist effects from inconsistent
18 use of BUP/SAM could increase the risk of
19 withdrawal or potentially increase the risk of
20 overdose and respiratory depression, particularly
21 if higher doses of opioids are used to overcome
22 antagonist effects.
A Matter of Record (301) 890-4188 228
1 Now, I will discuss risk management. The
2 applicant's proposed REMS mitigated the risks of
3 misuse and accidental exposure. Training is made
4 available to healthcare providers. Training is not
5 required to prescribe or dispense.
6 The agency is concerned about risks for
7 BUP/SAM, including those associated with opioids as
8 well as other potential safety concerns.
9 Prescribers of BUP/SAM for depression will not be
10 required to take training under DATA 2000 because
11 it is not indicated for opioid use disorder.
12 In conclusion, if approved, FDA will likely
13 require a REMS for BUP/SAM and is considering how
14 REMS can best address the potential concerns with
15 BUP/SAM in the indicated population and whether
16 training for potential prescribers is necessary as
17 part of the REMS.
18 Clarifying Questions to
19 FDA and Guest Speaker
20 DR. NARENDRAN: We'll now move on to
21 clarifying questions. If people could specify if
22 they have clarifying questions for the agency or
A Matter of Record (301) 890-4188 229
1 for Dr. Sullivan, that would be good. And before
2 you speak, please state your name for the record.
3 And if you can and you have a specific presenter,
4 please address that person as well.
5 I'm going to start with Dr. de Wit.
6 DR. DE WIT: This is Harriet de Wit,
7 University of Chicago. I have a question about the
8 FDA presentation. On slide number 8, this is the
9 abuse liability, could you just clarify?
10 It looks like the percent of abuse cases
11 reporting injection route are the same for
12 buprenorphine with and without naloxone. Is that
13 correct? So the naloxone is not serving as a
14 deterrent for intravenous use?
15 Do you need me to tell you who I'm
16 addressing?
17 DR. NARENDRAN: Who to address it to.
18 DR. MALLAMA: I can address the question.
19 That's correct. We saw, in the percentage of abuse
20 cases reporting, injection route, that it was
21 similar for buprenorphine, single ingredient, and
22 buprenorphine-naloxone, in the NEISS-CADES data and
A Matter of Record (301) 890-4188 230
1 the emergency department that's being displayed.
2 DR. DE WIT: Does that raise an issue with
3 the combination of buprenorphine and samidorphan if
4 it were to be used intravenously, that maybe the
5 samidorphan wouldn't block the mu agonist effect?
6 DR. MALLAMA: So this presentation on
7 buprenorphine-naloxone combination products can't
8 necessarily speak to what will be seen with
9 samidorphan. It's just to give context of the
10 real-world setting of what we're seeing in terms of
11 buprenorphine abuse, although they are similar in
12 terms of the fact that they both are combination
13 products with an antagonist included.
14 DR. STAFFA: This is Judy Staffa. I'd also
15 just like to add to that. What we were trying to
16 do was to show that in the real world, many people
17 assume that products that have naloxone added, that
18 that is a deterrent to non-oral routes of abuse,
19 but that's not what we're seeing.
20 So the other study I think from Cicero
21 suggests that people have developed methods for
22 separating, but we really don't understand fully
A Matter of Record (301) 890-4188 231
1 what's happening. We just wanted to introduce that
2 as something to consider. We really don't know
3 whether the same thing would happen with this
4 product.
5 DR. CHIAPPERINO: Hi. This is Dominic
6 Chiapperino, controlled substance staff. So
7 samidorphan would still have an antagonist effect
8 by the IV route. We don't have data by the IN
9 route.
10 DR. NARENDRAN: Dr. Kulldorff?
11 DR. KULLDORFF: Thank you, Martin Kulldorff,
12 biostatistician at Harvard Medical School. I had
13 one additional question for the applicant that I
14 didn't get through last time, and it's for Dr.
15 Schindler.
16 It was stated that the clinically important
17 effect size was somewhere between 1.5- and 2-point
18 reduction on the scale that you're using. But if
19 one looks at the point estimates and the confidence
20 intervals for the three studies, 205, 206, and 207,
21 it's clear that the effect size that the study was
22 powered for -- and I assume you did a power
A Matter of Record (301) 890-4188 232
1 calculation beforehand, is around 3 or so.
2 So if a clinically important effect size is
3 somewhere between 1.5 and 2, why did you power the
4 study for an effect size around 3?
5 DR. NARENDRAN: Does the sponsor want to
6 answer that question?
7 DR. SCHINDLER: I can just say you're
8 correct that the studies were powered for an effect
9 size of about 3, but we actually observed a smaller
10 effect that showed a significant difference. So
11 the power is really, as you know, our ability to
12 detect a difference if it's there, and we were able
13 to detect a difference.
14 The second part of your question, I think,
15 is also what is a clinically meaningful difference,
16 and I would defer to my clinical colleagues to
17 discuss what a clinically meaningful difference
18 would be. Maybe one of my statistical colleagues
19 has some additional comment to add.
20 DR. NARENDRAN: Very brief, very brief.
21 DR. MEMISOGLU: So you asked about a
22 rationale behind powering it based on a delta of 3,
A Matter of Record (301) 890-4188 233
1 so that was based on our experience in the 202
2 study and what we observed from MADRS-10.
3 DR. NARENDRAN: I think that addresses it.
4 I think I'm going to go back to the agency's
5 questions because it's their time.
6 Dr. Jain next. One second. If anybody else
7 has a question for Dr. Sullivan, he's going to have
8 to run in 15 minutes. He's on the phone. So does
9 anybody have a question for Dr. Sullivan of all the
10 lists I have here?
11 (No response.)
12 DR. NARENDRAN: No? Okay. Thank you.
13 Dr. Jain?
14 DR. JAIN: Felipe Jain, Harvard Medical
15 School. One of the two trials, the 207 trial that
16 this applicant has presented as one of their
17 pivotal efficacy trials, utilized a late change in
18 the primary outcome measure prior to unblinding.
19 How unprecedented is a request for such a
20 change? Is the FDA aware of other trials within
21 the context of treatment-resistant depression that
22 have requested a change such as this. And did the
A Matter of Record (301) 890-4188 234
1 applicant present any published literature
2 supporting their use of the change to this measure?
3 DR. TEMPLE: This is Bob Temple. I will try
4 to answer. I don't think we keep a list of these
5 things, but sometimes, other data, other studies,
6 and things like that convince people that the
7 endpoint they were using isn't the smartest one to
8 use. And as long as we're absolutely positive
9 there's been no breakdown of the data and nobody
10 knows what it is, we don't have any systematic
11 objection to that. It's not common in trials.
12 Trials are ongoing. You don't usually change the
13 endpoint.
14 I wish I could think of them, but there have
15 been other cases where people have changed the
16 endpoint. I don't know if -- Tiffany, you're
17 looking like you know some.
18 DR. FARCHIONE: Not that I know something
19 about a specific example, but I think that the
20 concern here is that it's not an issue that the
21 change was late because, again, it was made while
22 the study was still blinded, but with it being that
A Matter of Record (301) 890-4188 235
1 late, there really wasn't an opportunity for us to
2 provide feedback on those changes.
3 DR. TEMPLE: That's certainly another
4 critical question. You plan the study. We've
5 usually seen the protocol. We've said something
6 about the endpoint, and how they change it; they
7 don't know what we're going to think. But that's a
8 somewhat different question. And we worry
9 tremendously about whether there was leakage of the
10 interim data. That would invalidate any change
11 that anybody wanted to make, and it's not always
12 easy to know.
13 DR. NARENDRAN: I've been told Dr. Sullivan
14 may have left.
15 DR. CELLA: I do have a question for
16 Dr. Sullivan. Are you still there?
17 (No response.)
18 DR. NARENDRAN: Sounds like he's not
19 available. I don't know if the agency -
20 DR. CELLA: Maybe I'll ask it anyway.
21 (Laughter.)
22 DR. CELLA: Well, it's a question that maybe
A Matter of Record (301) 890-4188 236
1 the sponsor can answer or maybe FDA can answer. So
2 David Cella from Northwestern University in
3 Chicago. As far as I know -- is that Dr. Sullivan?
4 No. As far as I know, we don't know about any
5 exclusion criteria related to pain, and pain was
6 not assessed in the trials.
7 If that's correct, I was going to ask
8 Dr. Sullivan to what extent he thought this
9 combination might be treating pain and not
10 depression. But maybe other reviewers could
11 comment.
12 DR. NARENDRAN: I don't know if the agency
13 can comment or applicant. Was pain addressed?
14 DR. VON MOLTKE: We can comment on
15 inclusion/exclusion. Go ahead, Dr. Stanford.
16 DR. STANFORD: Pain and pain disorders were
17 not exclusionary to be enrolled in the program.
18 DR. NARENDRAN: Thank you.
19 FEMALE VOICE: But concomitant opioid use
20 was, right?
21 DR. VON MOLTKE: Yes.
22 DR. NARENDRAN: Next question, Dr. Dunn?
A Matter of Record (301) 890-4188 237
1 DR. DUNN: Walter Dunn, psychiatry, UCLA.
2 Two questions about the actual trial, so for
3 study 202, coming back to the question of
4 potentially throwing out that site 124 patient, if
5 the sponsors had prespecified the primary outcome
6 as only stage 2 outcomes, is that something the FDA
7 would have possibly considered? And I'm assuming
8 that if that patient was thrown out, that wouldn't
9 have affected just purely stage 2 outcomes.
10 Second question regarding study 207, my
11 understanding is that the longer duration of the
12 second stage made the analysis somewhat
13 complicated. So was there an analysis done just on
14 week 5 for 207, and would the outcomes have been
15 the same if we looked at 5 weeks for stage 1 and
16 5 weeks for stage 2?
17 Then the third question is regarding the
18 withdrawals. So for the assessment of the COWS, it
19 was a little unclear to me what the time frame was.
20 So my understanding was that the patients were
21 tapered for about a week, and then the COWS were
22 assessed at the end of 2 weeks.
A Matter of Record (301) 890-4188 238
1 Is that correct? And then is there any
2 sense of how long the withdrawal persisted or maybe
3 these mild withdrawals persisted, or is it only a
4 single time point?
5 DR. OGBAGABER: So for study 202, the
6 patient that was singled out and excluded, and we
7 did sensitivity analyses, he was or she was only in
8 the first stage and was not enrolled in the second
9 stage because he was dosed to the 2/2 arm. And
10 only placebo non-responders would go on to the
11 second stage.
12 DR. DUNN: If it was prespecified that the
13 primary endpoint was only stage 2 participants, is
14 that something the FDA would have possibly
15 considered as a valid endpoint or would it have to
16 be a combination of both stage 1 and stage 2?
17 DR. OGBAGABER: Yes. I mean, the SPCD would
18 have to combine and weight the stage to kind of
19 make inference.
20 DR. DUNN: Regarding study 207, maybe
21 redoing the analysis with just 5 weeks of stage 2?
22 DR. OGBAGABER: For the MADRS average, I
A Matter of Record (301) 890-4188 239
1 think they still went, but I'm not sure. The
2 sponsor can speak on this, if they did the 5/5 for
3 study 207 for on MADRS endpoint.
4 DR. VON MOLTKE: We'll address that. Go
5 ahead, Dr. Memisoglu.
6 DR. MEMISOGLU: Asli Memisoglu,
7 biostatistician, Alkermes. Can I get E-107? So
8 the question was whether having an equal duration
9 in the number of weeks included in the average gave
10 a consistent result, and it did.
11 Can I have the slide up, please? So what
12 you see here is the MADRS-10 analysis. The top
13 line is the average endpoint to end of treatment
14 that we've already presented, and the bottom line
15 is just limiting the analysis to weeks 3 to week 5
16 in both stages, and you can see that the results
17 are consistent.
18 DR. FARCHIONE: I think he was also asking,
19 though, about if you use week 5 as the end of
20 treatment. Right?
21 DR. DUNN: Actually, yes, they addressed
22 that question, yes. It's just limiting the
A Matter of Record (301) 890-4188 240
1 averages to 3 and 5 for both stage 1 and stage 2.
2 DR. FARCHIONE: So you only wanted to know
3 about the average, not if they just chose week 5 as
4 end of treatment?
5 DR. DUNN: Actually, that's an excellent
6 point. Yes. If you just ended it at week 5 for
7 stage 2, what would that have looked like?
8 DR. MEMISOGLU: I'd also like to comment on
9 the first part of your question about the stage 2
10 and study 202. And in that case, the stage 2
11 analysis was positive on its own. I think the
12 p-value was --
13 DR. DUNN: Right. And then in terms of just
14 the endpoint, at week 5 for stage 2, for 207, not
15 the average, but actually just the endpoint?
16 DR. MEMISOGLU: Right, and that did not meet
17 the threshold.
18 DR. DUNN: That did not meet criteria. And
19 the final question was about the withdrawal and the
20 COWS assessments, 1 week of taper and then
21 withdrawal assessment at 2 weeks. Is that correct?
22 DR. LEE: I would invite the applicant
A Matter of Record (301) 890-4188 241
1 to --
2 DR. VON MOLTKE: For the sponsor, I'll ask
3 Dr. Stanford to come on up.
4 DR. STANFORD: To answer your question, the
5 taper only existed in 202, so the main analysis for
6 withdrawal was in the pooled 205, 206, and 207
7 studies, where there was no taper and it was abrupt
8 discontinuation. And then for 205, there was one
9 week follow-up, and in 206 and 207, it was a 2-week
10 follow-up.
11 We also did the same analyses in our
12 long-term data, where there was three time points;
13 1 day, 1 week, and 2 weeks after discontinuation of
14 study drug. In that long-term study, the exposure
15 was up to a year.
16 DR. NARENDRAN: Next question, Dr.
17 Joniak-Grant?
18 MS. JONIAK-GRANT: Elizabeth Joniak-Grant.
19 Was a dossier for the validity of using the MADRS-6
20 submitted as was recommended by the FDA?
21 DR. FARCHIONE: Yes, it was. I noted that
22 it was submitted shortly after the breakthrough
A Matter of Record (301) 890-4188 242
1 therapy advice.
2 MS. JONIAK-GRANT: Can you give a little
3 more detail as to why it was still deemed, in the
4 FDA's view, as not valid?
5 DR. FARCHIONE: The main reason was because
6 of the four missing items, so without assessing
7 four key symptoms of depression, we didn't feel
8 like it accurately reflected an improvement in
9 depression overall. And I think, yes, we have Yeh
10 Fong Chen here from our clinical outcomes staff, so
11 he can speak more to that.
12 DR. CHEN: They submit a proposal to use
13 MADRS-6. And we said, well, please submit the
14 evidence to support MADRS-6 is as we have
15 [indiscernible] MADRS-10. And what they submitted
16 is actually quantitative study of effective
17 analyses in some correlations, basically to
18 show -- the main argument is that MADRS-6, the 6
19 items is unidimensional and the 10 items is not
20 unidimensional.
21 So based on the unidimensional assumption,
22 then the total score for the 6 is actually more
A Matter of Record (301) 890-4188 243
1 appropriate in that they show -- actually the
2 result that is submitted does not confirm whether
3 it's unidimensional. It confirms MADRS-6 is
4 unidimensional, but did neither confirm or not
5 confirm MADRS-10 is unidimensional or not
6 dimensional.
7 But as Dr. Farchione mentioned, the most
8 important thing is lack of four items, and that is
9 very important in terms of diagnosis of depression
10 and then also the recovery of depression symptoms,
11 and it's also in the DSM V.
12 So we evaluate the content validity first.
13 We don't think that MADRS-6 covers all the content
14 that is important to the patients. And then given
15 the quantitative [indiscernible], it did not fully
16 support that MADRS-6, and MADRS-10 is not
17 unidimensional.
18 So basically, it's just that it might be
19 actually all the symptoms. The 10 symptoms is a
20 continuance of the severity of depressions, and
21 then it lacks the more severe -- the MADRS-6 lacks
22 the more severe side of symptoms.
A Matter of Record (301) 890-4188 244
1 Then we see a lot of studies -- actually
2 they found MADRS-10 is unidimensional and some
3 MADRS-10 is not unidimensional. And I think that's
4 because the inclusion criteria of whether the
5 patient have more severe or less severe. When they
6 have less severe patients, there is less report the
7 appetite, the sleep problem, the suicidal thoughts,
8 and so it doesn't show up as unidimensional, but I
9 think it's a continuum.
10 DR. NARENDRAN: I do want to reiterate one
11 person, one question. Let's go ahead.
12 Dr. Conley? We still have a lot of
13 questions to go.
14 DR. CONLEY: Well, actually, I do have a
15 couple of concerns about this. It's more than one
16 question, so sorry, but I'll do one.
17 DR. NARENDRAN: Maybe we can come back in
18 line after that --
19 DR. CONLEY: At the end.
20 DR. NARENDRAN: -- if we have more time.
21 DR. CONLEY: So I'll start with the one that
22 I think is probably the most important to the
A Matter of Record (301) 890-4188 245
1 sponsor. In Dr. Dunn's presentation, you gave a
2 good presentation about, just in general, what REMS
3 are, but not really anything that I picked up
4 because the sponsor did propose a REMS of whether
5 what they were proposing was acceptable to the FDA
6 or what your thoughts were about it, so I assume
7 that's going to be a committee charge.
8 DR. DUNN: Can you guys hear me? This is a
9 little tall for me. We did meet several times and
10 discuss the REMS and what type of program we would
11 be putting together or having the sponsor put
12 together. But at this point, we still are not
13 decided exactly what the program would consist of.
14 So I don't think we're putting it out to you
15 as an exact question. It's more of like a
16 discussion point of what do you think. But where
17 we were focused was these programs that have
18 required training versus programs that don't. I
19 was trying to make that point in my presentation.
20 So that's something that, in particular, we
21 want you to think about, because of the risks that
22 we know of with opioids or opioid products, in this
A Matter of Record (301) 890-4188 246
1 particular product, are some of those concerns more
2 concerning?
3 That was the point that I was trying to
4 make, such as women of childbearing potential, and
5 NOWS, and concurrent benzodiazepine use, just
6 things that we already know about, but are we more
7 concerned in this particular population, and if we
8 are, then do we want to require training or not.
9 It's just something to think about in terms of the
10 program.
11 DR. NARENDRAN: Ms. Witczak?
12 MS. WITCZAK: Kim Witczak, consumer rep.
13 This is in line with REMS. What is the agency in
14 terms of direct to consumer advertising? What are
15 the guidelines there, as well as communicating out
16 to the patient organizations like National Alliance
17 for Mentally Ill, who could be out there as
18 promoting this as a possible treatment for their
19 patients.
20 Is that going to be part of it or is that
21 part of the discussion? Because as somebody has
22 spent her entire career in advertising and
A Matter of Record (301) 890-4188 247
1 communications, I saw in the presentation about
2 J.D. being the ideal candidate, I could see all
3 kinds of things from a marketing point of view,
4 that I would be out, saying you're not going to
5 gain weight; all of those things that are going to
6 be very attractive to people who may not want to go
7 on a antipsychotic.
8 DR. DUNN: I'm sorry. Specifically what's
9 the question?
10 MS. WITCZAK: Well, with direct-to-consumer
11 advertising, are there any kind of guidelines or
12 parameters around communicating to patient
13 organizations that will be out there as well?
14 DR. FARCHIONE: That wouldn't be part of
15 that. The REMS would be more related to the
16 physicians and the prescribers. We do have a
17 separate group that reviews the marketing materials
18 to make sure that there's nothing false and
19 misleading in there.
20 MS. WITCZAK: So I guess my question is, can
21 advertising happen right away when the drug got
22 approved? Can it be out there, being communicated
A Matter of Record (301) 890-4188 248
1 to the general public at the same time that there's
2 still the REMS that's going to the prescribers? So
3 that might be more for you guys internally.
4 DR. FARCHIONE: The company would submit
5 their initial marketing proposal to us for review.
6 MS. WITCZAK: Okay. So because it has a
7 REMS, you can't advertise it for a year? So those
8 are some of the things that I think are important
9 to the general public.
10 DR. DUNN: The REMS programs have nothing to
11 do with advertising. The communications that are
12 under the REMS programs are the letters that we
13 send to healthcare providers or healthcare provider
14 organizations to give them information about the
15 REMS program specifically and that discusses the
16 risks and requirements of the program or whatever
17 training is available through the program.
18 That's all the communication materials that
19 would be under the REMS, and for patients as well.
20 DR. UNGER: This is Ellis Unger. There's no
21 difference between a drug with and without a REMS
22 in terms of advertising if that's your question.
A Matter of Record (301) 890-4188 249
1 DR. NARENDRAN: Next question, Dr. Ruha?
2 DR. RUHA: Michelle Ruha. My question is
3 for the FDA. I just want to clarify something
4 about the MADRS-10 average. I think I understood
5 from the sponsor that using the average at each
6 assessment is more reflective of possibly
7 real-world treatment because there's really no end
8 of treatment. It's really the end of the study,
9 but in the real-world, it would be the end of the
10 treatment.
11 I believe the FDA did not necessarily agree
12 with using the MADRS-10 average, and I just want to
13 clarify that and clarify why. Is it because it
14 wasn't previously used in any studies, or it's not
15 validated, or did I misunderstand that?
16 DR. FARCHIONE: There were a few concerns
17 that came up during the course of discussions with
18 the applicant. One of the issues is we've never
19 used an average before. Now, it's not completely
20 out of the question that that might be a way to
21 increase power when you have variability and all of
22 those things. But every other drug that's been
A Matter of Record (301) 890-4188 250
1 approved for this indication has managed to win at
2 the end of treatment, so there was that concern.
3 Then there was also the issue that with this
4 being an opioid, is there some tolerance that
5 develops, and then you have a decrease in efficacy
6 that would be masked by an averaging strategy.
7 They did show the data on the one side where
8 the patients who did continue on the combination
9 continued to be okay, so that relieved our concern
10 on that issue a little bit. But the difficult
11 thing here is that we've got a study design that's
12 intended to reduce the impact of the placebo
13 response on the study, and then you have this still
14 difficulty separating, and then you throw in this
15 averaging strategy, which now also kind of
16 mitigates things.
17 There are just so many different ways of
18 slicing up the data. We wanted to minimize the
19 number of bells and whistles, I guess you could
20 say.
21 DR. NARENDRAN: I'll just ask my related
22 question. In the other medications that have been
A Matter of Record (301) 890-4188 251
1 approved for adjunct treatment, what was the
2 primary outcome measure? How long did those trials
3 last?
4 DR. FARCHIONE: Six to eight. It was either
5 6 or 8. I can double check.
6 DR. NARENDRAN: It's fine. You can let me
7 know. You can let us know later. The next
8 question is Dr. Iyengar?
9 DR. IYENGAR: Satish Iyengar from the
10 University of Pittsburgh. The FDA has brought up
11 two rather general issues of how to deal with
12 missing data and also the possible correlations.
13 What I also saw, though, in the presentation was
14 that the bootstrap verification and also the
15 careful depiction of the drop-outs indicated that
16 although there may be a violation of assumptions,
17 it was not a serious issue.
18 To what extent are these general issues
19 applicable to the studies that these people have
20 done?
21 DR. OGBAGABER: Regarding the MAR assumption
22 for study 207, actually, if we can go to backup
A Matter of Record (301) 890-4188 252
1 slide 13, 13 and 14, what we did was a graphical
2 depiction of these spaghetti plots for the drop-out
3 reasons and not really analytical.
4 And there is no method to check MAR using an
5 analytical method as you might know. It's all
6 graphical.
7 From here, drop-outs who behave -- within
8 each treatment group, we are assuming that they
9 behaved the same; each completer in the treatment
10 groups as if they hadn't dropped out. That's the
11 assumption and it's reasonable.
12 What was the second question?
13 DR. IYENGAR: The second question dealt with
14 the assumptions that you were checking with the
15 bootstrap. I think it was differences in variances
16 and also the correlation issue.
17 DR. OGBAGABER: Right. The bootstrap, slide
18 number 9 and 10, number 9. So for the bootstrap,
19 because the assumption for the applicant that they
20 use, using the method mentioned, was based on
21 normality assumption. So the only way we could
22 think of was a bootstrap to kind of check and go
A Matter of Record (301) 890-4188 253
1 over if that assumption of the covariance between
2 the two treatment stage effects being none, zero,
3 we used bootstrap, and we didn't make any of the
4 normal assumptions, and the conclusion remained the
5 same.
6 DR. NARENDRAN: Next question, Dr. Besco?
7 DR. BESCO: Thank you. Kelly Besco, Ohio
8 Health. Knowing that the current buprenorphine
9 products are being misused as often injected, it's
10 possible to separate these two products with
11 certain solvents. And in order to evaluate if this
12 product is safe, I think it would be helpful to
13 know what those solvents are and if they're common
14 household products.
15 I recognize that this is a public forum and
16 that we probably can't disclose that because of the
17 current buprenorphine products that are on the
18 market, but I just wanted to note for the record, I
19 think that would be helpful for members of the
20 committee to know what those products are for our
21 evaluation.
22 DR. CHIAPPERINO: This is Dominic
A Matter of Record (301) 890-4188 254
1 Chiapperino. You are correct that we can't speak
2 to those issues during the committee meeting.
3 Thanks.
4 DR. NARENDRAN: Next question, Dr. Crawford?
5 DR. CRAWFORD: Thank you. Stephanie
6 Crawford from the University of Illinois at
7 Chicago. This question is directed to
8 Dr. Ogbagaber.
9 In your presentation, because I think it's
10 pretty salient for our discussion later, going back
11 to your slide 8 regarding study 202, the single
12 subject from site 124 who had more extreme results,
13 the FDA Office of Scientific Investigations
14 questions the subject's eligibility from your slide
15 based on incomplete and/or contradictory source
16 information.
17 Is it possible for you to share more to
18 clarify for us what does that contradictory source
19 information and/or is it something very major that
20 was incomplete data?
21 DR. OGBAGABER: A staff from OSI is coming
22 to get to that question.
A Matter of Record (301) 890-4188 255
1 DR. KRONSTEIN: My name is Phillip Kronstein
2 from the Office of Scientific Investigations,
3 letting you know that we conducted an inspection of
4 that site, and yet incomplete means that for this
5 subject, there was no collateral information that
6 was not required by protocol. But if you have no
7 collateral information, then you are relying simply
8 on what the subject reports.
9 The subject in one part reported that they'd
10 been -- according to the sub-investigator who saw
11 the subject, reported the subject had been on
12 fluoxetine for 10 weeks prior to enrollment in the
13 study.
14 However, as part of the study, it was called
15 the failsafe, a safer interview was done. This
16 safe interview, as many may be aware, is a totally
17 independent interview, I think out of MGH, if I
18 remember correctly, and we reviewed the safer
19 interview. On two places on the safer interview,
20 it is written that the subject was only on
21 fluoxetine for 1 week.
22 Then there was an addendum on the safe
A Matter of Record (301) 890-4188 256
1 interview. This addendum was written 2 days later,
2 where we said, spoke to the clinic, and the subject
3 was on fluoxetine for 10 weeks.
4 We presumed that the clinic was not the
5 subject's psychiatric clinic, but rather the clinic
6 of the investigator. And the reason we presumed
7 that is because the subject indicated on a form,
8 they did not want their primary care physician or
9 their psychiatrist, if any, they didn't indicate,
10 contacted.
11 So there was no explanation for this
12 discrepancy. In fact, the safe interview is
13 supposed to be an independent interview. And if
14 indeed they did contact the clinical investigator
15 site, it would no longer be an independent
16 interview.
17 So this raised concerns about whether the
18 subject was truly eligible. And the fact is, based
19 on, again, these contradictions, you could not
20 determine basically two possibilities. Number one,
21 the subject was eligible; number two, the subject
22 was on fluoxetine for only 1 week. If the subject
A Matter of Record (301) 890-4188 257
1 had been on fluoxetine for only 1 week, of course
2 some of the results you may see may have been due
3 to fluoxetine and not the product.
4 So there's no additional information, and we
5 say we cannot determine whether the subject was
6 eligible.
7 DR. NARENDRAN: We have three more questions
8 for the agency that I want to get, and then after
9 that, we'll start the open public hearing at 1:15,
10 and maybe give extra time before the charge is
11 issued to respond to this.
12 Is that okay?
13 DR. VON MOLTKE: May I speak to the chair
14 after to be recognized later?
15 DR. NARENDRAN: After the open public
16 hearing, we can bring you guys back to address
17 that.
18 DR. VON MOLTKE: Okay, because we really
19 would like to clarify this since we've sent
20 information to the agency on this. Thank you.
21 DR. NARENDRAN: Thank you. Dr. Acri, your
22 question?
A Matter of Record (301) 890-4188 258
1 DR. ACRI: Thank you. Jane Acri from NIDA.
2 And my question has to do with the degree to which
3 the effects of buprenorphine are blocked by
4 samidorphan. In the briefing materials, there was
5 one graph that showed that the effects on pupillary
6 constriction was completely blocked by an equal
7 dose of samidorphan with a dose of buprenorphine.
8 And in the human abuse liability study, also, the 1
9 to 1 ratio blocked the liking effects of
10 buprenorphine, which was important.
11 But then there was something in the briefing
12 materials also that showed an in vitro functional
13 study showed that maximal stimulation of mu opioid
14 receptors produced by buprenorphine was only halved
15 by co-administration or by samidorphan. It
16 wouldn't be administration in vitro study. So it
17 looked like the effects of buprenorphine were not
18 completely blocked by samidorphan as shown in the
19 other two figures.
20 In addition to that, I think we haven't
21 really discussed the effects of this active
22 metabolite that is a mu agonist, that shows
A Matter of Record (301) 890-4188 259
1 accumulation over a period of 7 days. We've looked
2 at the abuse liability study that appeared to have
3 been done with acute dosing, but it seems to me
4 that with chronic dosing, you might get
5 accumulation of this metabolite, plus you have
6 incomplete blockade of the mu receptor.
7 So I'm just wondering if we've done a full
8 enough evaluation of the mu opioid effects?
9 DR. CHIAPPERINO: Those are excellent
10 questions. Regarding the studies that we have, we
11 don't have studies looking at buprenorphine/
12 samidorphan effects in ratios other than 1 to 1,
13 and that's an important point, particularly with
14 Dr. Meisel's points earlier about the potential to
15 manipulate the product.
16 As far as the mu opioid agonist metabolite,
17 we don't have data speaking directly to that other
18 than the adverse event profile in the chronic
19 studies. Of course, those are done in the patient
20 population.
21 So we don't see a higher incidence of
22 euphoric events in those studies. We did not see
A Matter of Record (301) 890-4188 260
1 them in the phase 1 studies to any noticeable
2 degree. So whatever the implications of there
3 being a metabolite that is a mu agonist, it's not
4 showing up in pharmacodynamic effects. So there's
5 an absence of data that may make us more
6 comfortable for sure.
7 DR. ACRI: Right. I'm also wondering about
8 its contribution to the withdrawal effects. The
9 COWS data was not gathered in a consistent way from
10 one day to the next, so I think it was the FDA that
11 presented the fact that it was sort of sporadic.
12 You couldn't see how much withdrawal people were
13 having over time. And the fact that you're getting
14 any withdrawal at all is kind of a concern and
15 makes me think there's more mu receptor stimulation
16 there than meets the eye.
17 DR. CHIAPPERINO: That's a good possibility.
18 DR. NARENDRAN: Next question,
19 Dr. Hernandez-Diaz?
20 DR. HERNANDEZ-DIAZ: Sonia Hernandez-Diaz
21 from Harvard Chan School of Public Health. I have
22 a question about how the FDA interprets this
A Matter of Record (301) 890-4188 261
1 stage 2 part of the study or any other study that
2 requires taking a placebo for some weeks, and among
3 those who do not respond, then go ahead and
4 randomize them.
5 With the interpretation of those results for
6 labeling or for reporting the results, would you
7 say then that the results will affect a population
8 that after being exposed to something in the
9 placebo or sugar pill do not respond, those are
10 ineligible for that medication?
11 It seems to me that is such a selective
12 group that the results from that part of the trial
13 or any trial, that that reflects a completely
14 different population.
15 So first, how do you then report and
16 translate that into clinical practice when
17 recommending the medication? And also, how are you
18 going to combine that stage 2 with stage 1 when
19 there are two different populations? We would not
20 pool that data. It's like they are coming from two
21 different populations, the second one being
22 selected in post-randomization.
A Matter of Record (301) 890-4188 262
1 So I don't understand why we are spending
2 time talking about the type 1 error in that design
3 when it's a different question, a different
4 population, what we are doing in stage 2.
5 So I guess two questions. Why are we
6 discussing type 1 error rather than questioning
7 whether they can be combined, and two, how could
8 you translate the results from stage 2 type of
9 population into recommendations for treatment to a
10 population that after being exposed for weeks
11 doesn't respond to placebo?
12 DR. TEMPLE: It's a good question, but I
13 would distinguish two aspects of it. One is
14 whether the drug has an effect in somebody, which
15 is a very important part of our conclusions about
16 whether a drug should be approved. The second is
17 how do you label it accurately?
18 Just to give an even worse example, from
19 your point of view, probably, we accept the idea of
20 a randomized withdrawal study; that is, you take
21 people, put them on drug. You then take the
22 responders and randomize to continued therapy or
A Matter of Record (301) 890-4188 263
1 placebo. We've described that in a number of
2 places. It's an enrichment strategy, if you like.
3 But then, once you've established that the drug
4 works by that, how do you decide how to label it?
5 Who's it for? What can you expect?
6 The answer is not easy, but at least you
7 know the drug works. So we tend to live with some
8 of those uncertainties. Usually, there's more than
9 one trial, not just the randomized withdrawal
10 study.
11 But in our labeling, we would always say how
12 the study was done. We would say only a hundred
13 people were given the drug; only 20 appeared to
14 respond, and the randomized withdrawal study was
15 done in those. So you'd have some idea of what the
16 likely response rate is.
17 But your question is a good one. If you use
18 any of these maneuvers, how do you label it exactly
19 in a way that tells people what would happen in
20 ordinary life? And my short answer is, we live
21 with that because establishing that something works
22 is really important even if you don't exactly know
A Matter of Record (301) 890-4188 264
1 how to tell people how to use it.
2 Is that enough?
3 DR. HERNANDEZ-DIAZ: Yes. Thank you.
4 DR. NARENDRAN: Thank you. Last question,
5 Dr. Meisel?
6 DR. MEISEL: Steve Meisel from Fairview in
7 Minneapolis. We've spent an awful lot of time
8 today talking about studies 202, 205, 206, and 207,
9 but not study 208. 208 is the open-label long-term
10 study, and I realize it's open label. But I wonder
11 if the agency would comment on the findings and any
12 conclusions that we can draw from 208.
13 My look at this, half of the patients
14 dropped out over the course of the year. And of
15 those, there was a 60 percent response rate,
16 meaning that there was only 30 percent. So how do
17 we take the short 5-week study and then look at the
18 52-week open-label study, and make something of
19 that? Are there any conclusions we can take from
20 that study 208?
21 DR. FARCHIONE: Typically, what we are using
22 those longer term open-label studies for is just
A Matter of Record (301) 890-4188 265
1 for long-term safety data. We're not really
2 looking to them for support of efficacy at all.
3 DR. MEISEL: So we should take no
4 conclusions whatsoever about efficacy based on 208?
5 DR. FARCHIONE: We aren't taking conclusions
6 on efficacy from 208.
7 DR. MEISEL: So all we have are the 5- or 6-
8 week trials. So the corollary to that; are the 5-
9 week trials sufficient for an antidepressant?
10 DR. FARCHIONE: So I did go back to look at
11 the other three that are approved for adjunctive
12 treatment of depression, and all three of them were
13 6-week trials.
14 DR. MATHIS: This is Mitch Mathis. We don't
15 think that short-term trials adequately describe
16 any of the diseases that we approve medications
17 for, so we always as a post-marketing commitment
18 will then ask for some longer term data.
19 We in the past have taken this issue to a
20 different body and AC, and asked them should we
21 have maintenance data before we approve a drug
22 acutely, and the answer was no; it will take too
A Matter of Record (301) 890-4188 266
1 long. We need new therapies, et cetera.
2 So our tradition is to do the short term,
3 get the approval based on short-term data, and then
4 we have some long-term safety data that goes with
5 that. That's what this 208 is. And then look for
6 a randomized withdrawal trial post-marketing,
7 should the drug be approved.
8 DR. NARENDRAN: I think we could stop here.
9 I'd like to do a 40-minute break if that's okay,
10 shorten the break. After the open public hearing,
11 if it's okay with the division, people who have
12 extra questions can probably ask it before the
13 charge. I'll let the division decide on that, if
14 they're agreeable to that. So 40 minutes from now,
15 we'll meet. That's 1:35.
16 We'll now break for lunch. We'll reconvene
17 in this room 40 minutes from now at 1:35. Please
18 take any personal belongings you may want with you
19 at this time. Panel members, please remember that
20 there should be no discussion of the meeting topic
21 during lunch amongst yourselves or amongst members
22 of the audience. Thanks.
A Matter of Record (301) 890-4188 267
1 (Whereupon, at 12:53 p.m., a lunch recess
2 was taken.)
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
A Matter of Record (301) 890-4188 268
1 A F T E R N O O N S E S S I O N
2 (1:35 p.m.)
3 Open Public Hearing
4 DR. NARENDRAN: We're going to start again.
5 Both the FDA and the public believe in a
6 transparent process for information gathering and
7 decision making. To ensure such transparency at
8 the open public hearing session of the advisory
9 committee meeting, FDA believes that it is
10 important to understand the context of an
11 individual's presentation.
12 For this reason, FDA encourages you, the
13 open public hearing speaker, at the beginning of
14 your written or oral statement, to advise the
15 committee of any financial relationship that you
16 may have with a sponsor, its product, and if known,
17 its direct competitors.
18 For example, this financial information may
19 include the sponsor's payment of your travel,
20 lodging, and any other expenses in connection with
21 your attendance at the meeting. Likewise, FDA
22 encourages you, at the beginning of your statement,
A Matter of Record (301) 890-4188 269
1 to advise the committee if you do not have any
2 financial relationships as well. If you choose not
3 to address this issue of financial relationships at
4 the beginning of your statement, it will not
5 preclude you from speaking.
6 The FDA and this committee place great
7 importance in the open public hearing process. The
8 insights and comments provided can help the agency
9 and this committee in their consideration of the
10 issues before them.
11 That said, in many instances and for many
12 topics, there will be a variety of opinions. One
13 of our goals today is for this open public hearing
14 to be conducted in a fair and open way, where every
15 participant is listened to carefully, and treated
16 with dignity, courtesy, and respect. Therefore,
17 please speak only when recognized by the
18 chairperson. Thank you for your cooperation.
19 Will speaker number 1 step up to the podium
20 and introduce yourself? Please state your name and
21 any organization you are representing for the
22 record.
A Matter of Record (301) 890-4188 270
1 DR. FAVA: I am Maurizio Fava. I am the
2 vice chair of the Department of Psychiatry at Mass
3 General, director of the Division of Clinical
4 Research at the MGH Research Institute, Director of
5 Clinical Trials Network and Institute, and
6 associate dean of clinical translational science at
7 Harvard Medical School.
8 I have been involved in the research that
9 has been presented today. I've helped Alkermes
10 design their studies, but I don't have any equity
11 on Alkermes. I have never received any personal
12 honoraria for my work. And I've always worked on
13 behalf of Mass General. And I paid for my own
14 travel to come here to make a comment on this.
15 My conflict of interest is that I'm the
16 co-inventor with David Schoenfeld from Harvard
17 School of Public Health and Mass General of the
18 Sequential Parallel Comparison Design.
19 My first statement is that I think, as
20 clinicians, I'm a clinician, we struggle with
21 treatment-resistant depression. We have over
22 40,000 people who commit suicide each year, and a
A Matter of Record (301) 890-4188 271
1 substantial proportion -- I think someone was
2 calculating STAR*D. We calculated that, at the end
3 of the four stages or levels of STAR*D, about
4 40 percent of the patients, 35 to 40, had not
5 really achieved remission despite multiple shots on
6 goal.
7 So treatment resistance is a big problem,
8 and what we have thus far clearly does not address
9 a significant portion of our patients, so new
10 tools, new therapies are critical.
11 Dr. Sullivan mentioned that opioid
12 dysregulation happens in depression. And
13 depression is very heterogeneous. Not everybody
14 will have an opioid dysregulation, but certainly, a
15 significant proportion of patients do have opioid
16 dysregulations.
17 That is why, in my opinion, depression is a
18 risk factor for opioid use disorder. Any risk
19 factor for opioid overdose is intentional and
20 unintentional. So if you add the 40,000 suicides
21 with the 40,000 lethal overdoses, this is a big
22 problem that we're facing as clinicians.
A Matter of Record (301) 890-4188 272
1 One of the first patients that our
2 depression program lost to suicide was someone with
3 treatment-resistant depression, was not responding
4 to monoamine therapies, and became an opioid addict
5 and committed suicide. So we know these patients
6 exist.
7 Dr. Sullivan mentioned something that is not
8 quite correct. There is tianeptine, which is a mu
9 agonist. It has been proven extensively by
10 Rennehan and others. It is a proven antidepressant
11 in a number of European countries and Asian
12 countries.
13 So the opioid mechanism has been exploited.
14 The problem with tianeptine as a mu agonist is the
15 potential for abuse. There have been many cases of
16 abuses in eastern Europe of tianeptine. So this
17 drug with the combination of a mu antagonist really
18 creates a deterrent for that, to the point that I
19 like to point out that the one patient who had
20 opioid use disorder had went into a significant
21 withdrawal the moment the patient was started on
22 the study, suggesting the fact it is clearly a
A Matter of Record (301) 890-4188 273
1 deterrent.
2 I think Dr. Dunn's observation is absolutely
3 correct with respect to the phase 2 study, that
4 even if you remove the subject in that academic
5 side -- and by the way, someone said -- because
6 we're responsible for the independent interviews.
7 When we look at patient-reported history,
8 patients are sometimes not great historians, so we
9 do call the site to get documentation information
10 about it. So it's still independent, but we want
11 to get additional information.
12 But in the phase 2 study, I like to point
13 out that there was a 4-point difference in stage 2,
14 a very robust 4-point difference in stage 2 without
15 that subject. So I think it was an absolutely
16 correct observation, and the study remains positive
17 whether or not -- in fact the statistical
18 elimination of the worst and the best patient still
19 maintains statistical significance.
20 So in my mind, I feel that the evidence is
21 there. You have two positive studies. And yes,
22 with innovations embraced by alchemists, the
A Matter of Record (301) 890-4188 274
1 sequential study, the averaging -- by the way, the
2 averaging of multiple visits is common in pain
3 studies.
4 So I think there were innovations, but I
5 think that embracing innovation is better than
6 having a 50 percent failure rate, which is common
7 in CNS trials. Thank you.
8 DR. NARENDRAN: Thank you. Will speaker
9 number 2 step up to the podium and introduce
10 yourself? Please state your name and any
11 organization you are representing for the record.
12 MR. COUNTS: Hi. Nathaniel Counts, Mental
13 Health America, and Mental Health America paid for
14 my time and travel to be here today.
15 Good morning. I'd really like to thank the
16 committee for its time and effort in considering
17 this for this full day. We really appreciate it.
18 I'm the senior policy director at Mental Health
19 America, which is the nation's leading community-
20 based nonprofit dedicated to addressing the needs
21 of those living with mental illness and promoting
22 overall mental health of Americans.
A Matter of Record (301) 890-4188 275
1 Our work is driven by our commitment to
2 promote mental health as a critical part of overall
3 wellness, including prevention services for all
4 early identification and intervention for those at
5 risk, integrated services, and support for those
6 who need it with recovery as the goal.
7 I was hoping to really underscore some of
8 the need because I know you've heard lots of
9 prevalence data today, but I wanted to kind of dive
10 a little bit deeper on that. So as just alluded
11 to, depression is one of the leading causes of
12 disability and a key contributor to two ongoing
13 epidemics, opioids and suicide.
14 I think the thing we wanted to underscore is
15 even though there is lots of therapies out there
16 already, lots of antidepressants, and options, and
17 research, these problems do persist. So our view
18 from MHA is that it isn't going to be just about
19 identifying needs as early as possible and
20 connecting people to the medications that already
21 exist, but there is actually, we think, a real need
22 for additional innovation.
A Matter of Record (301) 890-4188 276
1 Well, I'll kind of dive into it. I know
2 we've heard about STAR*D this morning, but the
3 Lancet in 2018 published the large network
4 meta-analysis of antidepressants and their
5 effectiveness. And it was exciting because it
6 finds odds ratios going from 1.37 to 2.13 across
7 all antidepressants.
8 So all of them improved depressive symptoms,
9 which I honestly always as a lay person have
10 trouble interpreting odds ratios. But a BMJ
11 article then goes on to say -- and I'll quote this
12 so I don't mess this up as well -- "All
13 antidepressants in the meta-analysis worked. They
14 all significantly increase the odds ratio of a
15 HAM-D score reduction by 50 percent, and across all
16 of them, the odds ratio was about 1.66, reducing it
17 by 50 percent."
18 To put the odds ratios, though, in a more
19 clinically relevant context, we need to know what
20 proportion get better in the placebo group. This
21 information was not provided, but other researchers
22 suggest that 30 to 40 percent of the placebo group
A Matter of Record (301) 890-4188 277
1 participants report improvement in remission.
2 I actually saw in Psychological Medicine
3 there was a meta-analysis suggesting about
4 53 percent, even as high as -- they're using this
5 typical placebo response; when you convert the odds
6 ratio of 1.6 to 10, it ends up being 10 to
7 12 percent more people in the treatment group
8 benefit compared to the placebo group.
9 So this raises like a very real issue with
10 the effectiveness of existing monotherapies and the
11 need for additional innovation in this space. And
12 I felt this from personal experience as well. I
13 struggle with the depression of taking multiple
14 antidepressants at all different doses, and levels,
15 and everything to absolutely no effect.
16 So I'm one of the other 50 percent for which
17 I didn't achieve spontaneous remission and for
18 which medications weren't effective, and this can
19 be really hard to have ongoing depression. But I'm
20 extremely lucky and have way more protective
21 factors than most.
22 I've actually never shared this before.
A Matter of Record (301) 890-4188 278
1 This is like a totally bizarre context to be
2 sharing these things, and especially after hearing
3 all this epidemiologic data about the different
4 constructs involved in depression. So not
5 everyone's as lucky as me, and I think having
6 additional information will really be critical.
7 I think 10 to 12 percent just can't be
8 acceptable for how effective our existing
9 medication is. For every patient population, we
10 can't just have 10 to 12 percent of people
11 achieving response of existing therapies.
12 I think through methods of existing
13 adjunctive therapies and pairing them with other
14 things, we can probably get -- let's see if we even
15 double or triple it, and that will still be
16 20-30 percent of additional effectiveness.
17 But I think we're really far from where we
18 need to end up being. And I think it's even more
19 horrifying when you really think about what
20 depression is as a condition. The diagnostic
21 criteria itself are feelings of worthlessness or
22 guilt almost every day, marked diminished interest
A Matter of Record (301) 890-4188 279
1 for pleasure in almost all activities nearly every
2 day, and recurring thoughts of death or suicide.
3 I mean, thinking about what that's like to
4 experience that level of pain every day is
5 extremely intense. It's even worse if you think
6 about 16 million adults every year do experience
7 this.
8 I think the most tragic thing that I've been
9 seeing is CDC data from BRFSS is showing that it's
10 trending in the wrong direction for adolescents.
11 All these risk factors are going down. Depression
12 is increasing over the past 3 years, though, and
13 suicidal ideation is increasing. And these
14 adolescents are going to become adults, and then
15 we're going to have a larger population of
16 depressed adults.
17 So I think that we do have a need for
18 innovation and really thinking about how can the
19 pipeline lead to better, newer therapies that do
20 help people and begin to alleviate the crisis we're
21 facing. So I really appreciate the time all of you
22 are taking today to think this through. Thank you
A Matter of Record (301) 890-4188 280
1 so much.
2 DR. NARENDRAN: Thank you. Will speaker
3 number 3 step up to the podium and introduce
4 yourself? Please state your name and any
5 organization.
6 MS. WALKER: Hi. My name is Ali Walker. I
7 am here on behalf of the American Foundation for
8 Suicide Prevention as a volunteer. My time is not
9 being paid for, and I have no financial disclosures
10 to provide to you.
11 I am by trade a critical care physician
12 assistant. I'm also the chair for the National
13 Capital Area Chapters local chapter of the American
14 Foundation for Suicide Prevention. And most
15 importantly, I'm a suicide loss survivor.
16 As of 8 years this past Sunday, I lost my
17 friend, Roma, to suicide. And I hope to elucidate
18 that I agree with what's been said so far about the
19 extreme importance of innovation as far as treating
20 major depressive disorder because it is a huge risk
21 factor for suicide.
22 Also, on a systemic level, worldwide, it is
A Matter of Record (301) 890-4188 281
1 the 11th leading cause of disability in the world.
2 In the United States, it's the second leading cause
3 of disability, which I found to be quite
4 unfortunate. Not only is it common, but it's
5 persistent. Over a 2-year period, more than
6 40 percent risk of recurrence for those who have a
7 single major depressive episode. And for those who
8 have a second episode, their risk of recurrence for
9 subsequent major depressive episodes is greater
10 than 75 percent.
11 I did some research on the history of
12 antidepressant development over the years, and I
13 was quite fascinated to find that the first
14 antidepressant medication was actually invented by
15 mistake in the late 1950s, upon the discovery of
16 euphoric side effects of iproniazid, which is being
17 used to treat tuberculosis.
18 Unfortunately, given the side effects of a
19 lot of drugs in this class, interactions with
20 tyramine, which is found in delicious things like
21 wine and cheese, those medications really were not
22 ideal, so there was an ongoing pursuit to develop
A Matter of Record (301) 890-4188 282
1 more antidepressant medications.
2 We saw a lot of innovations coming out from
3 the 1980s, late 1980s into the 1990s. In 1987,
4 fluoxetine, the first SSRI, was released, and
5 really was ground breaking at the time.
6 Unfortunately, though, since the early 2000s,
7 really, it seems there has not been a whole lot of
8 new discoveries. We've seen a number of drugs come
9 on the market that are enantiomers or have
10 different salt compounds in the composition of
11 those medications, but ultimately are not
12 significantly different from things that are
13 already existing.
14 Also interesting to me was the efficacy
15 pattern over the year. So the early iproniazid,
16 drug makers actually reported about a 75 percent or
17 70 percent response rate for patients taking that
18 for major depressive disorder.
19 I heard the mention of the STAR*D trial. I
20 actually did look at that study by the NIMH from
21 2001 to 2006, and only saw about 30 percent
22 response rate in first rounds of treatment for 8 to
A Matter of Record (301) 890-4188 283
1 12 weeks for standard therapy, which is awful, not
2 acceptable. As a clinician, for all of the other
3 health situations that we treat, we would never be
4 satisfied with that. Even after multiple rounds of
5 therapy, the most resistant patients, only about 67
6 percent, achieved a significant response.
7 So to put this in a more personal context, I
8 wanted you to consider a scenario for a moment.
9 The debilitating nature of depression is very real,
10 and I'll give you the example of my grandmother,
11 who over the summer was faced with the realization
12 of her own mortality after my grandfather was
13 admitted to the neuro ICU under the assumption that
14 he had weeks to live from a leptomeningeal disease.
15 In fact, he just had a very unusual
16 presentation of Lyme disease. Fortunately, he was
17 very easily treated. But after his release from
18 the hospital, my grandmother still persisted to
19 have significant mood changes. She was by all
20 definition depressed, lost interest in engaging
21 with her friends. She's a person that's out on the
22 golf course every day, organizes fundraisers, very
A Matter of Record (301) 890-4188 284
1 active individual, and she was having a hard time
2 getting out of bed in the morning, was not taking
3 phone calls when I would call her, which is highly
4 unusual for her.
5 Eventually, with some encouragement, we got
6 her to go see a physician, who recommended that she
7 start an antidepressant. So they did take her and
8 start her on mirtazapine, but after a couple of
9 weeks, actually, into this medication, she was
10 noticing a lot of abdominal cramping, diarrhea,
11 vomiting, the typical G.I. symptoms that are very
12 commonly associated with antidepressants.
13 So she got fed up because her mood had not
14 improved and now she has all of these uncomfortable
15 side effects and tossed the bottle in the trash,
16 and that was it. And that's not an effective way
17 to manage a condition that has completely turned
18 around your life and interfered with your ability
19 to enjoy the things that you do.
20 This scenario is very common, and I have
21 many friends and family members who struggle with
22 depression. I myself have struggled with
A Matter of Record (301) 890-4188 285
1 depression upon the loss of my friend to suicide.
2 And it's unfortunate to me that I have to, on a
3 regular basis, really fight to push people to
4 comply with medications being prescribed to them in
5 the hopes that they might improve on antidepressant
6 medications, knowing that they'll wait 6 to
7 12 weeks on average to have a response in their
8 mood, and knowing that in that time frame, they're
9 subject to a number of side effects from G.I.
10 discomforts and sexual dysfunction, and other
11 things that really do not contribute to positive
12 living.
13 So I will leave you with that, but I just
14 urge you to recognize that there is so much need
15 for innovation in the field of mental health and in
16 treatment of major depressive disorder.
17 DR. NARENDRAN: Thank you. Will speaker
18 number 4 step up to the podium and introduce
19 yourself? Please state your name and affiliation
20 for the record.
21 MR. CORSETTI: Thank you. My name is Gordon
22 Corsetti. I'm here on behalf of the American
A Matter of Record (301) 890-4188 286
1 Foundation for Suicide Prevention, and I have no
2 financial relationships to disclose.
3 I'm a depressive. I suffered for years
4 until I learned how to attack my depressed thoughts
5 and gain mastery over them. I deal with anxiety.
6 I'm naturally fearful in unstructured social
7 situations and very bright strobing lights, not
8 these, can trigger panic attacks that paralyze my
9 body and terrify my mind.
10 I live with suicidal thoughts, but I do not
11 act on them. I remain alive despite my brain's
12 best attempts to kill me. But here's the rub. I
13 never would have learned how to attack my depressed
14 thoughts, how to breathe through my anxiety, and
15 how to be okay with a suicidal thought without my
16 antidepressants.
17 I live with my mental illnesses now, but
18 years ago, I truly suffered. My first experience
19 with an antidepressant came during my freshman year
20 of college in 2006. An emergency room doctor
21 prescribed me the lowest dose of Paxil. And I
22 swallowed that pill every morning, and 2 weeks
A Matter of Record (301) 890-4188 287
1 later while walking across campus, my thoughts just
2 stopped.
3 Standing in the middle of campus, I waited
4 for an intrusive thought to pop into my head, but
5 no thought came. I didn't have a single horrible
6 thought in my head. And this wonderful sensation
7 of peace settled over me, and for the first time, I
8 walked forward without fear in years.
9 Since that day, I've been on SSRIs, SNRIs,
10 atypical antidepressants, and even a souped-up
11 version of vitamin D called Deplin. All of these
12 medications came with various side effects. My
13 weight went from 150 pounds to 195 pounds and back.
14 My desire for food and intimacy were drastically
15 reduced. Worst of all, though, was a persistent
16 sense of emotional numbness.
17 I accepted these side effects because the
18 net gain that I received gave me a much greater
19 quality of life. Still, I hated needing these
20 pills to feel or act normal. So I did what many
21 others have done. I stopped taking my medication.
22 I thought I could manage my symptoms through sheer
A Matter of Record (301) 890-4188 288
1 willpower.
2 Over the next several years, I started and
3 stopped many different medication regimens, and
4 each time I stopped, I spiraled into a
5 progressively worse depression. Three times, I
6 tried to take my own life.
7 Now, some would argue that antidepressants
8 cause me to attempt suicide. That is an incorrect
9 argument. Withdrawal from antidepressants
10 contributed to my attempts to kill myself. I stand
11 here today and proudly declare that while I have
12 experienced some of the worst consequences of
13 antidepressants, through their proper use, I have
14 experienced more joy than I ever thought possible.
15 These medications gave my mind space to
16 learn positive coping strategies, to build a strong
17 support network of family and friends, and
18 ultimately to gain agency over my illness. If
19 several hundred years ago, I went to a medical
20 professional and told him my symptoms, he would sit
21 me down and drill a hole into my head to release
22 the obvious demons from their imprisonment within
A Matter of Record (301) 890-4188 289
1 my skull. Two-hundred years ago, I would have been
2 strapped to a table so that the black bile from my
3 veins could be let out through bloodletting, and of
4 course, I would feel better as a result of that.
5 In a psychiatric hospital maybe just
6 60 years ago, I might have been lobotomized or
7 given this extreme electroshock therapy if only to
8 make me more compliant with the staff.
9 In 1987, a year before I was born, SSRIs
10 were introduced. SNRIs followed just a few years
11 later. Today, as a 30-year-old man, I am being
12 treated with medications that were discovered while
13 I was wearing diapers.
14 Now, I'm grateful that I live in an era that
15 favors medications over drilling into a skull to
16 treat the mentally ill, but my hope is that a few
17 years from now, we'll look back on the last
18 30 years of antidepressant use and think, this is
19 how we treated depression. We have so many more
20 options now.
21 It is time that we take the next step in
22 treating mental illness in this country. It is
A Matter of Record (301) 890-4188 290
1 time to invest in research to widen available
2 treatments. It is time for a new generation of
3 antidepressants. Thank you.
4 DR. NARENDRAN: Thank you. Will speaker
5 number 5 step up to the podium and introduce
6 yourself? Please state your name and organization
7 for the record.
8 DR. BODKIN: Hi there, all. This has been
9 fairly inaudible up until now. I hope that's
10 audible. I'm Alec Bodkin, J. Alexander Bodkin.
11 I'm a psychiatrist on staff at McLean Hospital. I
12 run the clinical psychopharmacology research
13 program and probably know more about the use of
14 buprenorphine as an antidepressant than any living
15 person, and we'll get to that.
16 I'm here on my own dime, but have consulted
17 Alkermes. In fact, I gratis-consulted to them as
18 they were first thinking about this possible
19 project because it seemed like an important way of
20 making safer a very effective treatment that had
21 big problems associated with it.
22 But in any case, I've been prescribing to
A Matter of Record (301) 890-4188 291
1 appropriate patients, i.e., very treatment-
2 refractory patients for whom ordinary monoaminergic
3 antidepressants just don't work, buprenorphine, and
4 other clever things, but buprenorphine specifically
5 for the past 30 years, when the dear departed
6 Jonathan Cole and I put together the first American
7 study of buprenorphine in depression.
8 Since then, I have consulted on countless
9 cases. I've always had a handful of buprenorphine-
10 treated patients in my own practice. And it has
11 serious problems, but it also has awesome efficacy
12 when it is needed.
13 One of the things that's been utterly missed
14 in this discussion, that's always missed, is that
15 there are responders who do extraordinarily well
16 and then there are ordinary patients who maybe do
17 so-so, but there may be side effects. They're all
18 blended together as though they have the same
19 problem, but depression is a multifaceted, many
20 different syndrome thing, really requiring many
21 different alternative treatments.
22 In any case, I'd like to get through my
A Matter of Record (301) 890-4188 292
1 written little things. I put so much work into it
2 and, if there's any time at the end, I'll speak my
3 mind beyond that.
4 Recently, I had the good fortune of being
5 able to treat 7 subjects in Alkermes' very long-
6 term, year-long study of safety, efficacy, and
7 tolerability at 2 milligrams of -- , I used to call
8 it Alks 5461, but going forward, it'll be BUP/SAM.
9 These were not very treatment-refractory
10 patients, as I customarily had treated with
11 buprenorphine, but they were treatment refractory
12 enough that SSRIs and related medicines didn't
13 really do the job adequately or didn't do it at
14 all.
15 So of our N of 7, 3 of them didn't do
16 particularly well. I mean, that's the way it goes,
17 and 4 of them did splendidly. So 57 percent of my
18 little N experienced remarkable degrees of
19 recovery, the kind that one rarely sees, that
20 persisted after treatment was completed.
21 Anyway, all of them, I will note, were
22 astonished at how well they felt, not at the
A Matter of Record (301) 890-4188 293
1 beginning of treatment. Unlike buprenorphine,
2 which causes a little bit of euphoria with dose
3 number 1, BUP/SAM causes nothing at dose number 1.
4 I mean, it's a little nausea sometimes, but it's
5 invisible. It creeps up on you.
6 But over time -- and it creeps up rather
7 slowly, faster than SSRIs, faster than MAOIs,
8 faster than tricyclics, but it creeps up slowly so
9 that it really wasn't until about week 6 that my
10 patients began to note that they felt better than
11 they ever had in their lives. And this really
12 persisted the remainder of the year.
13 We had one guy, a very troubled man, an
14 artist, who really didn't get well for 6 months,
15 but at that point was transformed. His life is
16 completely changed. His art has risen to high new
17 levels. He's gotten married. And he feels he
18 could not have done that without this help.
19 So anyway, when this stuff was stopped,
20 nobody had a withdrawal syndrome, unlike
21 buprenorphine, where that's the terrible problem
22 with buprenorphine. You can't get off it. If
A Matter of Record (301) 890-4188 294
1 you've been well for a year and a half, you want to
2 get off it, you get so miserable as you get down to
3 that last quarter milligram that you just say,
4 "Forget it; ill just stay on it," and that's what
5 people do.
6 There's no such problem with this compound
7 at all. It's invisible. You stop it and nothing
8 changes. You don't even lose the mood benefit.
9 And in fact, I spoke to my patients within the last
10 week getting ready for this and people still feel a
11 year later like their lives are simply better than
12 they had been, worth noting.
13 Anyway, let me briefly spell out some
14 important clinical properties of the Alks 5461 just
15 for the assembled. The therapeutic effects of the
16 drug in a significant portion of treatment-
17 refractory patients, or resistant patients with
18 depression, a significant proportion have extremely
19 robust benefits that persist long after the drug
20 treatment has been concluded.
21 This compound causes no acute euphoria. It
22 really causes no euphoria, and it has no more abuse
A Matter of Record (301) 890-4188 295
1 potential than an SSRI, and that is no
2 exaggeration. It has no immediate effects on mood
3 at all, and in fact, everyone had to have first
4 patients come in for their first -- well, they came
5 in for this anyway, but we stayed with them until
6 it had taken effect, and nobody felt anything other
7 than a little nausea sometimes.
8 This stuff differs from buprenorphine, which
9 causes euphoria, though it fades soon. And the
10 only real problem with that is that, in
11 irresponsible patients, they want to hang on to
12 that fading euphoria and they push the dose and
13 they push the dose.
14 So you have to be very careful who you bring
15 into treatment with buprenorphine and stay away
16 from people who have a tendency toward drug abuse.
17 That's just a requirement. People who are going to
18 pursue that buzz will not do well, nor will their
19 providers.
20 But unlike buprenorphine, its effect rose
21 over months. As I said, it takes about 6 weeks;
22 one patient took 6 months. And this would
A Matter of Record (301) 890-4188 296
1 certainly make the drug unattractive to the
2 potential drug abuser who wants to feel a quick
3 lift, because there is no quick lift.
4 What else have we've got here? Just that it
5 is extremely important, about this drug, that it
6 causes no withdrawal syndrome at all, even after an
7 entire year of daily use at 2 milligrams, which is
8 a non-trivial dose, and there was no dosage
9 tapering in the study, and wellness continued
10 indefinitely.
11 So the great limitation on buprenorphine,
12 apart from its abuse liability, is the chain that
13 it puts around people who are on it can't get off
14 it, which is not the end of the world if you need
15 it, but it is a burden that this compound
16 [indiscernible].
17 But I will only say in closing -- don't want
18 to take up too much time here -- is that we really
19 need to attend to the reality that in depressive
20 illness, there are robust responders to different
21 treatments. Occasionally, people will flower on
22 Prozac, not very frequently. Occasionally, people
A Matter of Record (301) 890-4188 297
1 will flower on Nardil, on the MAO inhibitors, which
2 were felt for a long time not to be very good
3 because most people didn't flower on them, but
4 people who got well got extremely well.
5 Well, people who get well on this stuff get
6 extremely well, and it's important to keep that in
7 mind. It's quite harmless. If it isn't going to
8 have any street value, buprenorphine is cheap
9 stuff, so no one's going to spend a lot of money
10 and time trying to extract a little bit from this.
11 It's just widely available. It would be silly and
12 a waste of money.
13 So this is not going to add to the drug
14 problems of our nation, and it really will give
15 people who otherwise can't get well, because their
16 problem relates to basically the dynorphin system
17 created by kappa receptors.
18 DR. NARENDRAN: I think your time's up.
19 DR. BODKIN: By fighting those receptors,
20 people get well. Anyway, I should go. Thank you
21 all so much.
22 DR. NARENDRAN: Thank you. Will speaker
A Matter of Record (301) 890-4188 298
1 number 6 step up to the podium? Please state your
2 name and institution for the record.
3 MS. WICKMAN: My name is Kathryn Wickman,
4 and I live with depression. I'm speaking today on
5 behalf of the Depression and Bipolar Support
6 Alliance. I have not received any compensation
7 from DSBA or Alkermes.
8 You've heard a lot of science and data so
9 far about mental illness and possible treatments.
10 I would like to put a face to those statistics. I
11 am here to talk about the anguish and emotional
12 pain that is my depression and why we need more
13 treatment options.
14 I have been living with the disease of
15 depression for 36 years from the age of 14. On my
16 worst days, depression meant lonely hours, deciding
17 if I had the strength to continue living, with the
18 seemingly never-ending unbearable burden of anguish
19 and hopelessness. No one should need to live with
20 a pain so strong that death seems to be the only
21 escape.
22 Other times, depression meant crying with
A Matter of Record (301) 890-4188 299
1 gasping tears in the stairwell at work, then
2 scraping myself together to get to my desk, where I
3 would lie and say my red eyes were due to
4 allergies, not tears. I would tell people the
5 reason I couldn't think clearly enough to function
6 at work was due to the allergy medications, not due
7 to my depression and the antidepressants that were
8 clearly not working.
9 My brain was so foggy, I could accomplish
10 nothing more than watch the clock count down to the
11 end of the day, when I could finally crawl back
12 into bed. I needed then and need now reliable,
13 affordable treatments to prevent these down days.
14 In my 30s, it took 5 tormented years to find
15 a treatment that would subdue my depression. I am
16 the poster child for treatment-resistant
17 depression. During that time, I tried over
18 40 medications and supplements. Not only did they
19 not work, but I had to live with unacceptable side
20 effects. I needed more treatment options.
21 Finally, at the age of 38, after living with
22 depression for 24 years, I found a medication that
A Matter of Record (301) 890-4188 300
1 worked. I had my Wizard of Oz moment, where the
2 movie starts in black and white and switches to
3 color. I realized everything had been black and
4 white in the past. I did not know what it meant to
5 live without the burden of depression.
6 With a medication, suddenly, I saw color.
7 Though the depression was not completely gone, I
8 then had about 11 years of roughly stable moods.
9 Then in 2017, I was diagnosed with breast cancer.
10 My oncologist told me, on top of everything else, I
11 might have to change one of my antidepressants
12 because it conflicted with a cancer treatment. I
13 warned her I was scared of what her treatments
14 might do to my moods. I was not afraid of the
15 cancer that might kill me. I was terrified of the
16 depression that might make me want to kill myself.
17 As I feared, my depression came back during
18 the cancer treatments. Ultimately, I had about
19 10 months of a severe depression, 3 months of which
20 were debilitating. At the start of this
21 depression, I thought, at least this time it will
22 be easier to find a solution. This time, they must
A Matter of Record (301) 890-4188 301
1 know more about mental illness. They must have
2 more treatment options for me now.
3 I was disappointed to learn this was not the
4 case. Between having already tried and failed with
5 most of the medications the decade earlier and now
6 needing treatments that were acceptable to both my
7 psychiatrist and oncologist, I had very few
8 options.
9 My psychiatrist finally said we were down to
10 my last alternative. I needed to do
11 electroconvulsive therapy. I was indignant. This
12 was 12 years after my last major depression. That
13 was 12 years to do research, 12 years for testing,
14 but I was in the exact same place of no viable
15 treatments.
16 The best news my psychiatrist could offer me
17 was that the methodology of pulsing electric
18 current through my brain was now less severe than
19 it used to be.
20 We needed better treatment options when I
21 was living with severe depression from the ages of
22 14 to 38, and we needed better treatment options
A Matter of Record (301) 890-4188 302
1 8 months ago, when I had another incapacitating
2 depression as a result of the cancer treatments.
3 Unfortunately, I suspect I will have at
4 least one more debilitating depression in my life.
5 I ask that you work to make sure there are better
6 options when that happens. I must have better
7 treatment options by them because the failure of
8 mental illness treatments could mean life or death
9 to me. Thank you.
10 DR. NARENDRAN: Thank you. Will speaker
11 number 7 step up to the podium? Please state your
12 name and organization for the record.
13 MR. SCHARF: Good afternoon. My name's Eric
14 Scharf. I am the advocacy advisor for the
15 Depression and Bipolar Alliance. However, I am
16 providing testimony on behalf of an individual who
17 was planning to be here today and could not attend
18 at the last minute. His name is Allan Sweet. He
19 lives with depression. I am speaking on behalf of
20 the Depression and Bipolar Alliance and Alliance of
21 Disabled Veterans. And I am speaking on behalf of
22 him as not received any compensation from DBSA or
A Matter of Record (301) 890-4188 303
1 Alkermes for his appearance.
2 "Over the years, I have faced many
3 challenges for my attaining and maintaining
4 wellness, as I have worked on being stable with my
5 major depressive disorder. I have a variety of
6 symptoms, but I believe that these 7 have plagued
7 me the most: low self-esteem, disturbed sleep,
8 irritability, anxiety, guilt, lack of motivation,
9 and suicidal thoughts.
10 "Being a veteran has allowed me access to
11 mental health services that have been helpful most
12 of the time. I have been prescribed various
13 medications over the years for my depression and
14 sleep. Though I was unclear when the depression
15 was first onset, I do know that after a
16 parotidectomy I had in 2007 at the Milwaukee VAMC,
17 my depression symptoms increased.
18 "Since that time, I have participated in
19 support groups, seen a number of mental health
20 professionals, one-on-one counseling with a social
21 worker, and, again, a variety of medications.
22 During this time of year, I'm also plagued with
A Matter of Record (301) 890-4188 304
1 seasonal affective disorder, where the days seem
2 dark most of the time and I find myself irritable
3 and full of guilt.
4 "My irritation comes from the lack of good
5 sleep, which leads to guilt, where I believe I am
6 not fulfilling my responsibilities. Though I
7 attempt to practice good sleep hygiene, eat
8 nutritionally, and exercise, I sometimes lack
9 motivation fulfilling as part of my wellness
10 routine. I try to see my mental health
11 professional every 90 days and often cancel, only
12 to reschedule appointments again due to lack of
13 motivation and not practicing my wellness routine.
14 "During this time of year, I use a sun lamp
15 in the morning for 30 minutes or so as suggested by
16 the social worker I was seeing. And though this
17 helps, I am not consistent, leading me to entertain
18 suicidal thoughts primarily surrounding my self-
19 worth and, again, self-esteem.
20 "Depression impacts my daily living, keeping
21 me from making good decisions in the area of
22 employment and following a wellness routine. I
A Matter of Record (301) 890-4188 305
1 find myself withdrawn and isolating myself from
2 others. The medication helps others, but I have
3 yet to use the right combination that helps me get
4 started in the morning, which leads to long days
5 and even longer nights from lack of sleep and
6 following my wellness routine. Living with
7 depression is a challenge and I will not give up,
8 as I know the person I can be and become."
9 Thank you on behalf of Mr. Sweet.
10 DR. NARENDRAN: Thank you. Will speaker
11 number 8 step up to the podium? Please state your
12 name and organization for the record.
13 MR. SPERLING: Good afternoon. My name is
14 Andrew Sperling. I'm the director of legislative
15 advocacy for the National Alliance on Mental
16 Illness, and as an employee of NAMI, they are
17 paying my time to be here today. I have no other
18 disclosures.
19 First, let me start with NAMI's perspective
20 on depression. The story you just heard from
21 Kathleen [sic], I'm sad to tell you, is rather the
22 rule rather than the exception for people living
A Matter of Record (301) 890-4188 306
1 with treatment-resistant depression. It is
2 enormously frustrating to be told to take a
3 medication and deal with monotherapy over and over,
4 and not see yourself get better.
5 It's an enormous challenge. We know that 16
6 million Americans experienced depression within the
7 past 12 months, 7 percent of our nation's
8 population. You've heard already about the public
9 health burden. I'm going to particularly focus on
10 the burden associated with suicide.
11 In the United States today, mortality from
12 suicide now exceeds that of both breast cancer and
13 prostate cancer. Those lines crossed just a few
14 years ago. You look at the advances we've made in
15 early identification and more effective treatments
16 for breast cancer and prostate cancer, but yet
17 suicide continues to go along, like over $40,000 a
18 year.
19 With treatment-resistant depression in
20 particular, it's very, very challenging. As the
21 STAR*D study demonstrated nearly a decade ago, only
22 one-third of patients actually get better on
A Matter of Record (301) 890-4188 307
1 monotherapy. So what happens the other two-thirds?
2 They need adjunctive therapy. Unfortunately, only
3 three medications with an on-label indication for
4 adjunctive therapy are antipsychotics.
5 We know about the side effects associated
6 with these medications, including weight gain,
7 metabolic syndrome, are extremely challenging and
8 make it very, very difficult for patients.
9 Monotherapy is futile for the majority of patients
10 living with treatment-resistant depression, so we
11 need better therapies.
12 I also want to note for the record that I
13 was on the patient stakeholder advisory group for
14 the last three prescription drug user-free
15 agreements; 4, 5, and 6. As part of that, we work
16 with many of our colleagues in the patient advocate
17 community across all disease states to try and
18 bring some changes to the FDA around patient-
19 focused drug development, to get this agency to
20 look more at surrogate endpoints, adaptive clinical
21 trial design, other things to get the voice of the
22 patient into the process of scrutinizing new drug
A Matter of Record (301) 890-4188 308
1 applications in the work that the FDA does.
2 Unfortunately, that doesn't seem to have
3 really seeped in here in the psychiatric drug
4 division office, and it's quite frankly telling in
5 some of the back and forth between the sponsor and
6 the agency that's taken place here today.
7 In August, NAMI submitted comments to the
8 agency on draft guidance for industry on major
9 depression, and we had some challenges there, and
10 I'd be happy to share that testimony with the
11 committee, particularly around some things that FDA
12 needs to be doing to address placebo effect.
13 You've heard some of this here today, why we
14 have a higher incidence of placebo effect with all
15 psychiatric disorders, but particularly with
16 depression, and investing and adopting things such
17 as Sequential Parallel Comparison Design, and other
18 types of trial designs that can actually address
19 placebo effect would be an advance forward.
20 Just as we did in our submission on the
21 comments on the guidance for industry, we will
22 again urge the agency to look at ways to use
A Matter of Record (301) 890-4188 309
1 adaptive clinical trial design to get around
2 placebo effect, to really spur development of new
3 medications, and also surrogate endpoints.
4 We need to get beyond depression scales that
5 have existed for the better part of 60, 70 years
6 and have better -- look at clinical endpoints, not
7 only that matter to patients, but modernize the way
8 we're measuring depression and measuring the
9 symptoms of depression.
10 We at NAMI strongly support innovation. The
11 treatments we have available, particularly for
12 treatment-resistant depression, simply aren't good
13 enough. Patients are demanding better treatments
14 and the opportunity to live a full life in the
15 community. Thank you very much.
16 DR. NARENDRAN: Thank you. Will speaker
17 number 9 step up to the podium? Please state your
18 name and organization for the record.
19 MR. POLLOCK: My name is Michael Pollock. I
20 serve as the chief executive officer for the
21 Depression and Bipolar Support Alliance. DBSA has
22 received funding from Alkermes to support our
A Matter of Record (301) 890-4188 310
1 education and outreach programs. DBSA is the
2 leading peer-directed national organization
3 focusing on mood disorders, depression and bipolar.
4 Unlike any other organization of its kind,
5 DBSA is created for and led by individuals who
6 themselves have a mood disorder diagnosis. This
7 first-person lived experience informs everything we
8 do.
9 In the 60 years that have passed since the
10 first antidepressant medication was approved by the
11 FDA, there have been significant advances in the
12 scientific understanding of depression, yet
13 treatment options that support individual
14 definitions of wellness remain elusive for many.
15 Information has been incremental. People electing
16 such treatment are consequently frustrated by and
17 losing hope of a pharmacologic solution.
18 The first priority for treatment is ensuring
19 that a person living with depression or bipolar
20 disorder is provided a pathway out of crisis and
21 onto stability. However, all too often, this
22 baseline stability is also the end game established
A Matter of Record (301) 890-4188 311
1 for successful long-term care.
2 In a DBSA-distributed survey this past
3 August that realized over 6,000 responses, nearly a
4 third reported having 10 or more discrete periods
5 of severe depression, and 36 percent indicated that
6 its impact is persistent.
7 DBSA believes that every person deserves the
8 opportunity not just to survive, but to thrive.
9 And to do that, we need to ensure true wellness as
10 the end goal for mental health treatment.
11 Additionally, the idea of wellness cannot be
12 embraced without considering the whole health of
13 the individual. The comorbidities associated with
14 depression are not insignificant. The prevalence
15 of depression among individuals living with heart
16 disease, diabetes, polycystic ovary syndrome, CPOD,
17 movement disorders, and Alzheimer's, just to name a
18 few, is well known, and the effect depression can
19 have on the positive outcomes of comorbid
20 conditions is significant.
21 Choosing between effective treatment for a
22 comorbidity in mental health is counterproductive.
A Matter of Record (301) 890-4188 312
1 Individuals living with mental health conditions on
2 average die 25 years sooner, and as a result of
3 co-occurring conditions that can be either
4 exacerbated by depression or could exacerbate the
5 depression, the cost of settling for reduced
6 symptoms is simply too great. And for many, it can
7 be a matter of life and death.
8 Even more challenging than understanding the
9 whole health ramifications of pharmacologic
10 interventions associated with comorbidity is the
11 realization that no one medication typically
12 provides an entire range of symptom relief for
13 depression.
14 Additionally, these interventions have
15 differing risk-benefit tolerances for each
16 individual. Further, the considerations around
17 medication risks and benefits can often be
18 different. The prescriber may approach the
19 challenge from a clinical perspective, symptom
20 relief, while the patient on the other hand, may be
21 seeking well-being outcomes.
22 When seeking solutions, patients weigh the
A Matter of Record (301) 890-4188 313
1 risks and benefits of that intervention against
2 symptom relief, functional considerations, and
3 those well-being outcomes. Added to this decision
4 is the fact that an intervention may not be
5 consistent in both its symptom relief and side
6 effects among the patient population.
7 This often results in a frustrating trial
8 and error period for both prescribers, who want to
9 help their patients, and the patient who's looking
10 for improvement. Unfortunately, during the trial
11 and error period, many patients reach a point where
12 they abandon hope and not just a pharmacological
13 intervention, but in any type of treatment.
14 The mood disorder community feels abandoned.
15 Many researchers consider depression a problem
16 solved due to the number of pharmaceutical
17 interventions currently available, but nothing
18 could be further from the truth.
19 If I've communicated anything today to this
20 committee, it's first that patients' voices count.
21 Patients want and need solutions that support a
22 pathway to wellness. One size does not fit all.
A Matter of Record (301) 890-4188 314
1 Solutions are complex as the individuals seeking
2 them. And individuals will evaluate the risks and
3 benefits of those solutions based on their own life
4 circumstances.
5 When considering this application, I urge
6 this committee to not abandon the patient
7 community. Thank you.
8 DR. NARENDRAN: Thank you. Will speaker
9 number 10 step up to the podium and state your name
10 and organization for the record?
11 MR. MADIGAN: Good afternoon. My name is
12 John Madigan. I'm the senior vice president and
13 chief public policy officer for the American
14 Foundation for Suicide Prevention, which paid my
15 way here today. And I've lived in D.C. for
16 41 years, and my Uber driver took me around the
17 Beltway on a trip that I could not repeat, but
18 that's a fun note.
19 I've been in my position now for 10 years.
20 My work is both professional and personal. I lost
21 my sister to suicide 21 years ago. As many of you
22 know, AFSP funds research to improve interventions,
A Matter of Record (301) 890-4188 315
1 train clinicians in suicide prevention, and
2 advocates for policy that will save lives.
3 AFSP is creating a culture that's smart
4 about mental health. AFSP brings people who have
5 been affected by suicide out of the darkness and
6 give them opportunities to help others like many of
7 my volunteers here today.
8 Many of you know there's no single cause for
9 suicide. Suicide most often occurs when stressors
10 and health issue coverage create an experience of
11 hopelessness and despair. Depression is the most
12 common condition associated with suicide, and it's
13 often undiagnosed and untreated.
14 Conditions like depression, anxiety, and
15 substance problems, especially when unaddressed,
16 increased risk for suicide, yet it's important to
17 note that most people who actively manage their
18 mental health conditions go on to a full and
19 engaging life.
20 This is why I'm here today. The National
21 Center for Health Statistics issued a report
22 February 13th of this year, indicating that 1 in 12
A Matter of Record (301) 890-4188 316
1 U.S. adults reporting having depression. Fifty
2 percent report some degree of difficulty with work,
3 home, or social activities; 30 percent reported
4 moderate or extreme difficulty.
5 The World Health Organization reports that
6 depression has risen to be the current leading
7 cause of medical disability worldwide. This is
8 likely the result of several facts because of the
9 disabling nature and symptoms of this serious
10 condition because depression is so highly
11 prevalent, because stigma keeps more than half the
12 people with depression from pursuing treatment, and
13 importantly -- this is really important -- because
14 treatment options are currently limited in
15 effectiveness and in options.
16 As the former lead lobbyist for the American
17 Cancer Society in the 1980s, I was pleased to read
18 FDA Commissioner Scott Gottlieb's public comments
19 on September 13, 2018 at a Friends of Cancer
20 research conference. Much like cancer, a diagnosis
21 of depression can for some people present a grave
22 situation. For many patients with depression,
A Matter of Record (301) 890-4188 317
1 their mental suffering is as painful, and in some
2 cases, more painful than physical health
3 conditions.
4 At that September conference, Dr. Gottlieb
5 said, I quote, "We at FDA want to take every chance
6 we have to foster and maximize the kinds of
7 innovations that will make these opportunities
8 available to patients." He was of course referring
9 to cancer.
10 I ask today, why not apply the same
11 opportunities for innovations and opportunities for
12 patients suffering with depression or other mental
13 health conditions. The FDA should do everything
14 possible to create an environment where companies
15 are willing to make important investments in a new
16 generation of antidepressants that are safe and
17 effective.
18 In 1970, the American people made clear
19 their desire to battle cancer deaths in the United
20 States. President Nixon responded during his
21 January 1971 State of the Union address, "The time
22 has come in America when the same kind of
A Matter of Record (301) 890-4188 318
1 concentrated effort that split the atom and took
2 man to the moon should be turned towards conquering
3 this dreaded disease. Let us make a total national
4 commitment to achieve this goal."
5 The Food and Drug Administration in 2018 and
6 beyond can lead a battle to create new
7 opportunities and innovations for depression drug
8 treatments that just might help us win the war
9 against suicide and mental illness. Thank you.
10 DR. NARENDRAN: Thank you. Will speaker
11 number 11 step up to the podium? Please state your
12 name and organization for the record.
13 MS. KENNEY: Good afternoon. My name is
14 Lauren Kenney, and I'm here on behalf of the
15 American Foundation for Suicide Prevention, and I
16 have no financial benefits to disclose.
17 I have been surrounded by mental illness and
18 depression as long as I can remember. I lost my
19 dad to suicide when I was just 9 years old. In the
20 20 years since he has been gone, I have struggled
21 with my own mental health, and in 2016, I was given
22 an official diagnosis of moderately severe
A Matter of Record (301) 890-4188 319
1 depression.
2 After meeting with my treatment team, it was
3 decided that medication was going to be a good
4 option for me. I had tried talk therapy, and
5 although it was helping, I needed additional
6 assistance in order to function on a daily basis.
7 Finding the right medication to help me
8 manage my depression symptoms was a trial and error
9 process. It took me and my doctor time to figure
10 out the right medications as well as the proper
11 dosages. This process came with side effects and
12 many, many frustrations.
13 That begin said, we were able to find
14 something that worked for me, and slowly, the
15 relief started to come. After being at a point
16 where I was unable to sleep, unable to get out of
17 bed, and at times unable to physically move,
18 shower, or get dressed, the medication allowed the
19 fog in my head to clear. Instead of constantly
20 being on the verge of tears, my eyes cleared and I
21 was able to function. Eventually, I was even able
22 to function again at a high level.
A Matter of Record (301) 890-4188 320
1 Earlier this year, I changed my full-time
2 career and started working for a mental health
3 awareness nonprofit. Every day, I hear from people
4 who are living successful, happy lives while living
5 with mental illness. They share their stories with
6 me, and many of their stories include their
7 treatment plan, which more often than not include
8 medication to help manage their symptoms. It's
9 clear to me that proper medication is crucial in
10 the management and treatment of depression.
11 For me personally and many folks I know that
12 have shared their stories, medications have helped
13 them live a more fulfilling satisfying life.
14 However, I also know that I'm a lucky one. I was
15 able to find medication that worked for me, and the
16 results outweighed the side effects.
17 Along with all the people who have shared
18 their story of a successful treatment plan, there
19 are those who have shared their story of all the
20 different medications they have tried, that have
21 not given them relief from their depression. And
22 there are even more stories that we have yet to
A Matter of Record (301) 890-4188 321
1 here because right now, their struggle is too much;
2 their pain is too debilitating.
3 These folks are some of the strongest people
4 I know because despite finding little to no
5 reprieve, they keep fighting. It is for those
6 people who are continuously fighting to find relief
7 for their indescribable pain and depression, that I
8 ask you to vote yes to new medications to treat
9 depression. They deserve a chance to live a life
10 with depression that is manageable and not all
11 consuming.
12 We all deserve to live a fulfilling life
13 without having to battle the agony of depression,
14 and I know that new medication could help make that
15 possible for those who continue to suffer.
16 As a suicide loss survivor, I know how
17 devastating mental illness can be and how it can
18 affect family and friends of those we lose. We owe
19 it to those fighting through the pain every minute
20 of every day and to their family and friends who do
21 everything they can to help them to allocate our
22 resources to properly treat depression. Not only
A Matter of Record (301) 890-4188 322
1 could new medications alleviate the daily struggle
2 for those with depression, but new medication could
3 also help us save lives.
4 Over the years, we have developed many new
5 ground-breaking treatments for other diseases, and
6 now it is time we focused our research, energy, and
7 resources to fighting depression.
8 My struggle matters. The struggle of those
9 who are unable to speak to you today matter. My
10 depression matters, and the depression of those who
11 have been unable to find the proper treatment
12 matters. And I along with others who are brave
13 enough and strong enough to live with depression
14 every day should have the opportunity to live
15 better with the help of new, more innovative
16 medications. Thank you.
17 DR. NARENDRAN: Thank you. Will speaker
18 number 12 step up to the podium? Please state your
19 name organization for the record.
20 DR. SAMBUNARIS: I'm Dr. Angelo Sambunaris.
21 I'm a psychiatrist. I practice clinical medicine
22 in Atlanta, Georgia. I'm also a clinical
A Matter of Record (301) 890-4188 323
1 investigator for various pharmaceutical clinical
2 trials and also investigator-initiated trials with
3 universities, currently Duke University, Medical
4 College of Georgia, studies that we design
5 ourselves. And I'm an adjunct professor at the
6 Mercer University College of Pharmacy.
7 My disclosures, I'm not a compensated
8 speaker here today. I am not a consultant or
9 advisor to Alkermes. I am currently investigating
10 studies with Allergan, Alkermes, Sunovion, and
11 Tonix Pharmaceuticals. And in full disclosure, I
12 was notified yesterday morning that I was permitted
13 to show up here today, and Alkermes did help me
14 with my travel expenses. Airlines didn't think
15 this was urgent enough to give me a discount.
16 The FDA is tasked with two questions. One,
17 is this compound safe? The data says yes. The
18 rest of the discussion is speculation and what-ifs.
19 Second question is, is this compound efficacious,
20 and the data tells us yes, but because of issues of
21 placebo response going on around the world,
22 especially in areas of anxiety and depression, a
A Matter of Record (301) 890-4188 324
1 new study design was introduced, and that has
2 clouded that picture of whether this drug is
3 efficacious.
4 We have a unique new chemical entity here.
5 It addresses a societal need. It addresses
6 depression symptoms. If we look at patients with
7 depression, over 90 percent experience symptoms of
8 anxiety. This medication is anxiolytic. It's the
9 first thing that I see as a clinical investigator,
10 as I'm evaluating these patients.
11 Patients with depression, over 90 percent
12 have sleep disturbance. This medication is
13 sedating and it helps with sleep, again helping
14 individuals with that impact on quality of life.
15 Something that we saw in our clinical
16 investigation over the last few years in the open-
17 label treatment studies, when I worked in the
18 pharmaceutical industry many years ago, our
19 statisticians told us that 1-year studies typically
20 resulted in 50 percent of the patients dropping
21 out. They dropped out for various reasons; the
22 burden of participating in a clinical trial or even
A Matter of Record (301) 890-4188 325
1 that they had improved and decided that they didn't
2 need to be in the study anymore, only to drop out
3 and find out later that they needed to get back
4 into that study, but were not permitted to do so.
5 When we were running the 208 study, we had
6 over 80 percent patients in retention, and in
7 addition, knowing that we have depression as a
8 chronic illness with episodes of remission and then
9 a return, a recurrence.
10 At our clinic in Atlanta, we have decided to
11 focus on how to improve the methodology of medical
12 research. One of the ways we do that is by not
13 enticing people to participate for financial
14 compensation. We actually give them free after-
15 study care.
16 Typically, in a study, when there's
17 improvement, after the study is over, we will see
18 in augmentation trials that within a matter of
19 weeks, the patient will, on monotherapy, lose their
20 effect. Once they lose that effect, we need to try
21 and figure out what augmentation strategy to use.
22 With the Alkermes compound, we did not see
A Matter of Record (301) 890-4188 326
1 that. We actually saw that some type of modulation
2 was occurring in the system where, a month out,
3 2 months out, 6 months out, they still were
4 improved on monotherapy, and the longest that we
5 went with a patient was 9 months. And we decided
6 to follow him just to see how long before there was
7 a return of symptoms; 9 months, and at that point,
8 he said, "I need something new to be added to my
9 treatment."
10 Current treatments are atypical
11 antipsychotics, and you've all heard about the side
12 effect profiles with them. There are many off-
13 label treatments out there and insurance does not
14 cover them. So patients are relegated to paying
15 out of pocket, which is a burden to them, and
16 unacceptable. We don't do that in any other
17 therapeutic area.
18 I think that the information from all the
19 speakers that went before me, those that have
20 personally experienced the symptoms, is compelling
21 enough to say, "Let's really consider this
22 medication with a vote of yes." I believe that the
A Matter of Record (301) 890-4188 327
1 advocacy groups have also done the same thing.
2 They've told you that there's a societal need. And
3 I believe that scientifically and clinically, and
4 from the experience of an investigator that has
5 seen these patients and treats these patients that
6 we have, again, there should be a yes vote on this
7 compound. Thank you.
8 DR. NARENDRAN: Thank you. Will speaker
9 number 13 step up to the podium please? Please
10 state your name and organization for the record.
11 DR. FOX-RAWLINGS: Thank you for the chance
12 to speak today on behalf of the National Center for
13 Health Research. I am Dr. Stephanie Fox-Rawlings.
14 Our center analyzes scientific and medical data to
15 provide objective health information to patients,
16 health professionals, and policymakers. We do not
17 accept funding from drug and medical device
18 companies, so I have no conflicts of interest.
19 As we've heard today, depression is a
20 serious health issue. However, FDA requires proof
21 that new drugs are safe and effective. New drugs
22 have to work in order to actually help patients.
A Matter of Record (301) 890-4188 328
1 I want to focus on efficacy first. The
2 evidence presented for BUP/SAM does not provide
3 adequate evidence that it reduces depression more
4 than placebo. Two of the three trials designed to
5 provide evidence of efficacy did not find any
6 statistical benefit of the drug compared to
7 placebo.
8 The third trial, study 207, had a
9 statistically significant reduction in depression
10 on the MADRS for the 2/2 dose, however, the trial
11 had shortcomings. For example, it used the
12 MADRS-6, which lacks important aspects of
13 depression and therefore can't prove efficacy.
14 Using the full MADRS, the only time that the
15 drug was more effective than placebo was for just a
16 few weeks in the middle of the short trial. By the
17 end of the trial, the placebo group was doing just
18 as well as the treatment group. This does not
19 demonstrate a meaningful benefit for patients, and
20 though statistically significant for that short
21 time, the treated patients improved less than 2
22 points more than placebo on a 60-point scale. This
A Matter of Record (301) 890-4188 329
1 small difference seems too small to be clinically
2 meaningful for patients.
3 This test has clear standards for
4 improvement. Responders are defined as those whose
5 symptoms improve by at least 50 percent, and
6 remission is defined by scores that are less than
7 10 or below.
8 Patients taking the drug were not more
9 likely to be a responder or to go into remission
10 than the patients taking placebo for any of these
11 efficacy trials. This again raises questions about
12 whether the result in study 207 was clinically
13 meaningful compared to other studies of
14 antidepressants. These trials may have just been
15 too short to demonstrate this, but there needs to
16 be some sort of confirmatory evidence for this
17 result.
18 Study 202 was designed as a proof-of-concept
19 study, and FDA points out that it can't prove
20 efficacy because it lacks the statistical controls
21 to make sure that the difference did not occur by
22 chance. Also, the relatively high drop-out rate
A Matter of Record (301) 890-4188 330
1 for patients in the drug arms could have had a
2 large effect on the results.
3 We agree with the FDA reviewers that the
4 lack of a dose response also raises red flags. If
5 the drug is effective, the higher dose should at
6 least show a statistical trend towards
7 significance, but it didn't.
8 As we know, in studies of depression, the
9 placebo groups often do quite well. Placebo
10 controls are essential because they help control
11 for the natural ebb and flow of depression
12 episodes. The sponsor tries to eliminate evidence
13 of a placebo effect. However, without a long-term
14 comparison of the placebo drug to drug arms, it is
15 not possible to determine whether natural
16 fluctuations in depression or treatment are
17 affecting the results.
18 There are also concerns about safety. The
19 clinical trials included very few older patients,
20 and older patients metabolize drugs more slowly and
21 are more likely to have adverse reactions. They're
22 more likely to be taking other drugs that can
A Matter of Record (301) 890-4188 331
1 interact with this drug.
2 Like all opioids, even ones that are
3 designed to be abuse deterrent, this drug has the
4 potential for misuse and abuse. This is a major
5 concern because depressed patients are more likely
6 to have substance abuse disorders and are at
7 increased risk for opioid overdose.
8 In summary, the clinical trial does not
9 provide adequate evidence that this drug reduces
10 symptoms of depression. There are concerns about
11 the potential for long-term harms to patients and
12 others who might use or abuse it.
13 This drug needs to provide strong evidence
14 of efficacy before approval. Although refractory
15 depression is a serious condition, prescribing new
16 treatment with unproven benefits and unknown risks
17 is dangerous. As you know, new drugs for
18 depression tend to be more widely prescribed than
19 the narrower indications that FDA approves.
20 Approving the drug for even treatment-resistant
21 depression could easily contribute to the opioid
22 epidemic. Thank you.
A Matter of Record (301) 890-4188 332
1 Clarifying Questions (continued)
2 DR. NARENDRAN: Thank you.
3 The open public hearing portion of this
4 meeting is now concluded and we will no longer take
5 any comments from the audience. Before we go to the
6 next section, I do want to provide the sponsor an
7 opportunity to answer the question that was raised
8 in the morning about the single subject.
9 DR. VON MOLTKE: Thank you. We wanted to
10 address the patient from study 202. This is the
11 patient that there was a request from FDA to
12 exclude. This patient is a 56-year-old African-
13 American male who had a very strong response to the
14 medication, and we've investigated, and there's no
15 reason at all to exclude this gentleman's response.
16 I'm going to ask Dr. Pathak to very quickly
17 and succinctly just list to you the pieces of
18 evidence that we made copies of and sent back to
19 the agency. So they have these.
20 DR. PATHAK: Sanjeev Pathak, psychiatrist at
21 Alkermes. We went to the site, and we confirmed
22 that the patient was eligible, had long-standing
A Matter of Record (301) 890-4188 333
1 depression, had cycled through many therapies, and
2 the fluoxetine was for an adequate dose and
3 adequate duration. Slide up, please.
4 We provided copies of 5 original documents
5 substantiating that the patient was indeed the kind
6 of patient who would be a candidate for a therapy
7 like this. There was a clinical note signed by the
8 physician listing the prior treatment as well as
9 the initiation of fluoxetine.
10 There was also the ATRQ done at the site.
11 There was the independent safer interview with an
12 additional note. There was a concomitant
13 medication log. So in totality, the patient was
14 appropriate for this study.
15 DR. VON MOLTKE: May I also ask if we can
16 clarify a comment made on the COWS scores that were
17 missing?
18 DR. NARENDRAN: Sure.
19 DR. VON MOLTKE: Thank you. I'll ask
20 Dr. Stanford to comment. On a number of the
21 trials, there was no requirement that the drug be
22 stopped if they were going to roll over into the
A Matter of Record (301) 890-4188 334
1 208 long-term study. So in that case, there was no
2 COWS because patients didn't stop.
3 So maybe you can just address that.
4 DR. STANFORD: Yes. So for the 202 study,
5 patients completed participation and were done.
6 However, for the 205, 206, and 207 studies,
7 patients either had a follow-up visit or were
8 allowed to roll over into the long-term safety
9 study. For 205, they were required to complete the
10 follow-up visit, and that's where the majority of
11 COWS data came from.
12 For 206 and 207, there was no reason to stop
13 medication if they were going to roll over directly
14 into the long-term study. So there was no missing
15 data. This was all per protocol. And I just
16 wanted to clarify that 266 patients were in the
17 placebo-controlled dataset and 839 in the long-term
18 safety study.
19 DR. NARENDRAN: Thank you. I don 't know if
20 the agency has a comment on COWS now.
21 I think we'll move to the next section. Dr.
22 Mathis will provide us with a charge to the
A Matter of Record (301) 890-4188 335
1 committee and present the questions for discussion.
2 Charge to Committee - Mitchell Mathis
3 DR. MATHIS: I know I took too much time
4 this morning, so I'll give it back to the group
5 this afternoon. The charge to the committee is
6 pretty simple. It's to think about these data,
7 think about the questions we have to answer from a
8 regulatory point of view if you can do that.
9 The next set of questions we have to
10 answer -- and we need your help with this -- are
11 have substantial evidence as we defined it earlier
12 of efficacy been provided for this drug product?
13 If it has, then, or even if it hasn't, you can
14 consider the next question about has the safety
15 been adequately characterized?
16 Then we will ask you, once those questions
17 have been answered, to do what we do with these
18 type of data, decide the synthesis of that, do the
19 benefits outweigh the risks. And I know that's a
20 lot to ask, and I'd like to spend the rest of the
21 afternoon, the time that we have this afternoon, to
22 answer your questions so that you can get to a
A Matter of Record (301) 890-4188 336
1 point where you can help us answer those questions.
2 Thank you.
3 Questions to the Committee and Discussion
4 DR. NARENDRAN: We will now proceed with the
5 questions to the committee and the panel
6 discussions. I would like to remind public
7 observers that while this meeting is open for
8 public observation, public attendees may not
9 participate except at the specific request of the
10 panel.
11 Question number 1, has substantial evidence
12 been presented by the applicant to support the
13 effectiveness of buprenorphine/samidorphan for the
14 adjunctive treatment of major depressive disorder?
15 This is a voting question.
16 DR. MATHIS: We have some time built into
17 the schedule -- this is Mitch Mathis -- to discuss
18 this, and I'm sure there's some leftover issues
19 that you had had.
20 DR. NARENDRAN: So if anybody has any
21 questions, this is the opportunity to raise.
22 Dr. de Wit?
A Matter of Record (301) 890-4188 337
1 DR. DE WIT: I have a leftover issue, and
2 that concerns the trial 206, that we all accepted
3 was negative. It had quite a large number of
4 subjects, I think 280, if I'm not mistaken. And in
5 some of our materials, it was said that it might
6 have been negative because of a large placebo
7 response, but judging by the figures that are
8 presented here, describing the two graphs for the
9 placebo versus drug, it doesn't look like the
10 placebo response is much greater than it was in the
11 other three trials.
12 So I wonder if someone could comment on the
13 negative trial and also if the FDA could say to
14 what extent can you -- what's your level of
15 accepting 2 out of 4 trials or 3 out of 4? How do
16 you judge what's acceptable in this kind of case?
17 DR. FARCHIONE: I think that this is a good
18 opportunity to talk a little bit about -- to just
19 acknowledge the idea that the rate of placebo
20 response in antidepressant trials is a problem. I
21 think we heard that a lot from the public comments
22 period as well. And it's increasing over time. So
A Matter of Record (301) 890-4188 338
1 that means that it's only getting more difficult to
2 see a difference between drug and placebo in
3 clinical trials.
4 So in the best-case scenario, we would have
5 better drugs with bigger treatment effects, and
6 then it'd be easier to see the difference between
7 drug and placebo, but that has proven difficult to
8 say the least. So it's not really unreasonable to
9 look at trial designs as a way to manage the impact
10 of placebo on the trials.
11 So earlier, somebody had asked about looking
12 at just stage 2 in study 202, and you guys, the
13 applicant, noted that just looking at stage 2 alone
14 would have been positive, and that's great. So
15 because none of the patients treated with the
16 buprenorphine/samidorphan combination in stage 1 or
17 that study had continued on drug in stage 2, it's
18 an enriched population. It's essentially similar
19 to a placebo lead-in, which is similar to study
20 206.
21 So even if we were to say look at that now,
22 and say, yes, so stage 2 of that study looks like
A Matter of Record (301) 890-4188 339
1 maybe you've got a positive response, it's still
2 fraught with all those issues of the post hoc
3 comparisons and whatnot.
4 So we would still want some substantiation
5 of that result, which then we would look at
6 study 206, and study 206 was negative, so you don't
7 get that substantiation there.
8 I think it's also a good chance for me to
9 mention that with regard to SPCD, conceptually the
10 design is appealing. From a clinical perspective,
11 it makes sense to me that you would be able to try
12 to mitigate the placebo response in your trials
13 with a design like that.
14 The stats are admittedly very complicated,
15 but if we could agree on what the analyses should
16 look like and everything, it's not unreasonable to
17 say that maybe this could help with this problem
18 that we have that's very troublesome.
19 But with SPCD, you'd expect the drug placebo
20 difference to be amplified in stage 2. So it
21 should be easier to achieve statistical
22 significance at the end of treatment, which didn't
A Matter of Record (301) 890-4188 340
1 happen here, so that pretty much leaves one of two
2 possible conclusions. Either SPCD doesn't work or
3 the drug doesn't work.
4 We don't know in this case what the right
5 answer there is because we don't have just your
6 basic standard, like drug versus placebo from the
7 get-go, design, whatever.
8 With regard to the number of negative
9 trials, that actually has come up in front of the
10 committee before. The consensus among the
11 committee at that particular meeting was, well,
12 you've got two positive trials. The negative
13 trials were not as impressive. You had the two
14 positive trials. It's hard to get two positive
15 trials, and so they said you've got two; that
16 should count.
17 DR. NARENDRAN: Next question,
18 Dr. Hernandez-Diaz?
19 DR. TEMPLE: Can I just add something? As
20 people have told you, something like 50 percent of
21 the apparently well-designed trials of
22 antidepressants fail. So if you do four trials,
A Matter of Record (301) 890-4188 341
1 it's not unusual to have two failures. The
2 question here is whether they have positive trials.
3 That's the question.
4 DR. UNGER: This is Ellis Unger. Just in
5 the interests of being completely transparent, in
6 study 202, let's assume that the patient from
7 site 124, there's no question about their
8 qualification for the study. Nevertheless, we get
9 very concerned when a study result is basically
10 hinging on a single patient.
11 Now, you don't have to be a statistician to
12 understand that if your p-value is 0.049 and you
13 remove a patient with a good response, that you're
14 now over 0.05, and you've taken a study from a
15 positive study to a negative study. We understand
16 that.
17 But here, the result was so extreme, we find
18 it unsettling. And I'm saying that to be
19 transparent. I want the committee to think about
20 it. You can convince us that we shouldn't be
21 concerned about it, but we are concerned about it.
22 And I want to let you know that we're concerned
A Matter of Record (301) 890-4188 342
1 about it and give you the opportunity to disagree
2 if you wish. But it does make us very concerned.
3 DR. VON MOLTKE: May the sponsor add
4 anything?
5 DR. NARENDRAN: Sure.
6 DR. VON MOLTKE: Dr. Anderson?
7 DR. ANDERSON: Aparna Anderson. I'm a
8 biostatistics consultant, and I've been compensated
9 for my time and travel. So if you take that
10 question of influential data points in the analysis
11 and you do a sensitivity analysis to probe that,
12 the methodologically sound approach to take is to
13 look at it in a symmetric way, not to just take the
14 best data point from the xml group and see what
15 happens.
16 If you want to look at that, you have to
17 treat it symmetrically because you're affecting the
18 distribution that way. And so that's why the
19 sponsor conducted an analysis that trimmed both
20 sides of the distribution and replaced it.
21 So it's not just trimming because you lose
22 variability just by virtue of reducing your sample
A Matter of Record (301) 890-4188 343
1 size. You also have to replace those data and do
2 it in a rational way.
3 The sponsor did it in two ways. They
4 replaced it with respect to the adjacent data
5 point. They also did a multiple imputational
6 analysis where they drew on the overall data
7 observed from the rest of the trial.
8 I think that's the methodologically sound
9 way to do it. Otherwise, it's a very, very
10 one-sided, lop-sided way to do it.
11 DR. NARENDRAN: Thank you. I think we get
12 the disagreement there.
13 Move to the next question. Dr. Hernandez-
14 Diaz?
15 DR. HERNANDEZ-DIAZ: Sonia Hernandez-Diaz
16 from Harvard Chan School of Public Health. My
17 question is to understand the efficacy to answer
18 that question. And through the discussion, very
19 interesting points from the participants.
20 I think we are getting the idea that
21 depression is complex, that maybe there are many
22 things going on, that maybe the treatments can
A Matter of Record (301) 890-4188 344
1 target specific things that can work for some
2 patients and not from others.
3 We also got an answer from this morning that
4 we are looking at our average. So I guess there
5 might be patients in the trials that might be
6 responding very well and others that might not be
7 responding.
8 So I wonder if, through the trials, it would
9 be possible to look at the data in a way that we
10 can identify where there are some patients that are
11 really dropping 20 points in the scale and others
12 are not, or is everybody dropping 1 or 2 points,
13 because that would I think point to the possibility
14 that maybe there may be a way to identify patients
15 that might benefit from this adjuvant therapy a
16 lot. And that would be, I think, a good outcome
17 from the results.
18 So I don't know if you have looked at the
19 data in a way to identify whether there is
20 potentially a group. I understand it's not the
21 primary endpoint.
22 DR. VON MOLTKE: That's a great question,
A Matter of Record (301) 890-4188 345
1 and we have looked at it that way. Dr. Schindler
2 will go through that with you.
3 DR. NARENDRAN: The agency wants to comment
4 back on that?
5 DR. OGBAGABER: Can you go to slide 2 of
6 back-up slides? We just want to show the spaghetti
7 plot of all the patients in study treatment group
8 2/2 for study 202. And this is the subject we are
9 talking about, how it defers from the rest of the
10 patients. And clearly, it's an outlier, and that's
11 why we did the sensitivity analysis.
12 There is no one good robust sensitivity
13 analysis method, but what the sponsor showed us, we
14 don't disagree with. It's not uncommon for us to
15 look at extreme results or extreme subjects and
16 re-do the analysis.
17 DR. VON MOLTKE: We would just say he's a
18 great responder, but anyway, here's Dr. Schindler.
19 DR. SCHINDLER: Yes. I can show another
20 view of the data from study 202, if we have slide
21 up. This is the individual patient data for
22 stage 1 and stage 2, split by BUP/SAM and placebo.
A Matter of Record (301) 890-4188 346
1 Green is BUP/SAM. Gray is placebo. And you see in
2 stage 1, the responder that we're talking about is
3 that individual; each line is an individual
4 patient.
5 So that patient at the far right-hand side
6 of the top left is the patient that was removed.
7 So it was the largest responder, the best
8 responder. And what we've heard over the course of
9 the day is that some patients respond more and some
10 patients respond less.
11 You can see, when you look at the placebo
12 groups, that there's some placebo responders, too.
13 So the patient was the best responder, and you
14 know, in any analysis, when you take out the best
15 responder, you're going to have an effect on the
16 analysis.
17 You see that there is intermediate response
18 over the course of stage 1. When you look at
19 stage 2, you also see a very different picture that
20 there's a response in stage 2, but these are the
21 placebo patients who have rolled over onto stage 2.
22 So this shows the individual patient responses.
A Matter of Record (301) 890-4188 347
1 And you can see that that big responder is not
2 really such an outlier, but just the best responder
3 in the dataset.
4 DR. NARENDRAN: I'll let the agency respond.
5 DR. UNGER: This is Dr. Unger. It almost
6 looks like a cumulative distribution. We did a
7 histogram. Unfortunately, it was too late to get
8 in the slide set, but if you simply put patients in
9 bins by their response, this patient is way off the
10 chart. And basically, the spaghetti plot showed
11 that he started out with the worst HAM-D in the
12 study and ended up with the best HAM-D in the
13 study. So his delta is huge compared to anyone
14 else. This I think is a little bit misleading.
15 DR. NARENDRAN: Thank you. We'll move to
16 the next question. Dr. Kulldorff?
17 DR. KULLDORFF: Thank you. My name is
18 Martin Kulldorff. I'm a biostatistician at Harvard
19 Medical School. The key two questions is, is this
20 drug really working or not? And then if it is
21 working, what's the clinical amount or clinical
22 effectiveness of it. How much does it work? I
A Matter of Record (301) 890-4188 348
1 wanted to try to synthesize these questions from a
2 statistical point of view.
3 I'm not concerned about this one patient.
4 I'm not concerned about SPCD. I think this has a
5 very marginal effect on the big picture. But one
6 thing that is concerning is the issue of multiple
7 testing and not using necessarily the lack of
8 clarity on what was the pre-defined outcomes to
9 measure.
10 So the first study, 202, was an exploratory
11 study, so I think the interesting thing is to look
12 at the 205, 206, 207, and I do think that all of
13 them are relevant for our evaluation, the effect
14 sizes.
15 Two of them were not statistically
16 significant. One was or was not depending on which
17 outcome we use, so let's say we used the one where
18 it was statistically significant. We then have
19 point estimates of a benefit of 1.9, 1.7 -- sorry,
20 1.9, 1.8, and 0.2. So the question is, then that
21 together, that amount, enough evidence to see that
22 it's effective?
A Matter of Record (301) 890-4188 349
1 Now, I like meta-analysis, but I don't like
2 at all the way it was done by the applicant here
3 because, when they did the meta-analysis, they
4 didn't use the prespecified outcomes. They used
5 other outcomes. So they found an effect size in
6 the meta-analysis was 2.0, 1.8, but those are
7 actually higher than the average of these three
8 studies that was not exploratory.
9 So if we take the average -- approximately
10 based on the sample sizes and the point estimates.
11 But if we take the 205, 206, and 207 together,
12 their average effect size, using the average for
13 the 207 rather than the 5-week, we get a combined
14 of 1.3 effect size. If we also include the 202,
15 the effect size will be about 1.5.
16 So my assessment is that the point estimates
17 for the combined effect size is approximately 1.3
18 and definitely less than 1.5 rather than in the 2
19 range.
20 So then the second question is, is this an
21 important clinical effect size that is important
22 clinically to be able to provide the patient or is
A Matter of Record (301) 890-4188 350
1 it too small for that? And we heard that 1.5 to 2
2 was clinically important according to the
3 applicant, but they had powered the study for 3.
4 So let's say 1.5. Is that clinically
5 important effect size? Is it useful for a patient
6 with a 1.5 reduction in this score? If the answer
7 is no, then, well, that's pretty much what this
8 study says, 1.5. So then there's no point in
9 approving the drug because it doesn't have
10 sufficient efficacy.
11 If the question is yes, that 1.5 is useful
12 clinical -- as a biostatistician, I can't judge
13 that. But if the answer is yes, then what should
14 have been done was to do a study that's powered for
15 1.5 instead of 3.
16 It's a very common mistake done both in
17 terms of companies, but certainly also in academia,
18 that one does a preliminary study, kind of a small
19 sample size, and one gets an effect size, and then
20 one powers the subsequent true study with that
21 observed effect size from the sort of trial study.
22 And that was sort of what's done here because the
A Matter of Record (301) 890-4188 351
1 trial study had an effect size of 2.8 and then they
2 powered it for 3.
3 Now, that's very, very dangerous to do that
4 because, by chance, the true will be half the time
5 more than 2.8 and half the time less than 2.8. So
6 by doing that process of powering a study, you're
7 almost guaranteeing a very large risk of having an
8 inconclusive study.
9 In this case, it's even worse because there
10 were actually two doses in the exploratory study,
11 and one was 0.5 and one was 2.8 benefit. So then
12 if you pick the biggest one, well, that's more
13 likely to have been sort of biased towards the
14 upper side.
15 So the reason why we have a very
16 inconclusive study here, it's because, if 1.5 is
17 truly important clinically, then the power should
18 have been studied for that.
19 So if that is the case, then I think it
20 would make sense for the applicant to do a study
21 which actually has that type of power to be able
22 to, with 80 percent or 90 percent power, detect the
A Matter of Record (301) 890-4188 352
1 clinically important effect size of 1.5.
2 That will answer the question if this drug
3 is actually useful or not for the patients. We
4 have heard that it's important to have new drugs,
5 but obviously we don't want to give drugs to
6 patients if it's not working. That's just wasting
7 precious time and increasing suffering.
8 So if the applicant does have a study, then
9 the question is what's the design. One of the
10 issues that were discussed was weighted over
11 3 weeks or the end of time point. And I think
12 there's a lot of good reasons for doing a weighted
13 one because there's a lot of variability in the
14 measurements because these are surveys.
15 So by doing the average of three
16 measurements instead of taking only one, one
17 reduces the variability, increases the power, and
18 reduces sample size needed.
19 So I think that's a good call to do a
20 weighted one. The only thing that wasn't quite
21 clear here is it should be agreed upon by the
22 applicant and the FDA ahead of time, that that will
A Matter of Record (301) 890-4188 353
1 be the main outcome to the study. So I would
2 certainly encourage to do that weighted outcome.
3 With the SPCD, there are pros and cons, but
4 I think there are a lot of good things with the
5 SPCD. So I would not have any objections to doing
6 that in a follow-up study as long as the power is
7 sufficient to detect what's clinically important.
8 The third one is in terms of the outcome
9 measure, it's sort of strange. At least in one of
10 these SPCDs, the outcome was different for stage 1
11 and stage 2. I think there was week 5 in one and
12 week 6 in the other, or some other thing like that.
13 And to me, that doesn't make any sense. You might
14 as well use the same one. That will be much more
15 clear and that wouldn't have any drawback for using
16 the same one.
17 So that's basically my sort of synthesis.
18 It's unclear at this point whether this drug has
19 efficacy. If it does, the point estimate is 1.5,
20 and whether that is clinically important or not, I
21 cannot judge, but if it is, I hope there will be a
22 new randomized study to be conducted to find out if
A Matter of Record (301) 890-4188 354
1 we truly have such an effect from this drug. Thank
2 you.
3 DR. NARENDRAN: Thank you. Those are good
4 comments.
5 DR. VON MOLTKE: A comment from the sponsor
6 regarding the clinical efficacy of the effects
7 seen?
8 DR. NARENDRAN: You can comment very
9 briefly, but I think it was more of a synthesis of
10 his data, I mean, his read on the data.
11 DR. VON MOLTKE: I also just wanted to
12 caution. I continue to hear the concern about the
13 8 milligrams of buprenorphine-containing being
14 bigger than the 2-milligram dose in the
15 exploratory. And I would really caution, with a
16 complex pharmacology, of something like
17 buprenorphine, there is a lot of literature out
18 there that that is not the case to be expected. So
19 I would just caution on making that inference
20 around 8 being better than 2.
21 DR. NARENDRAN: I'd like to take the
22 opportunity -- I mean, one of the things that you
A Matter of Record (301) 890-4188 355
1 do mention,. there is a lot of pharmacology, but we
2 haven't seen a lot of pharmacology of what in vivo
3 occupancy is.
4 You threw out nociceptor and you mentioned
5 kappa opiate receptors, but we don't even know if
6 this compound really has what level of occupancy.
7 Looking at the in vitro occupancy, it's about 0.08
8 nanomolar at the mu opiate receptors, and it's
9 about 1.5 for kappa in rodents, and it's about 4
10 for kappa in primates.
11 So that would suggest you have very modest
12 occupancy at 2 milligrams. And if you were at
13 8 milligrams, I would think you would fully
14 saturate and block mu opioid receptor occupancy,
15 and you would have more kappa antagonism, but you
16 don't see that.
17 So to me, the pharmacology of it is still
18 not fully defined. You guys talked a lot about
19 SPCD and placebo lead-in, but you're still in sort
20 of '70s and '80s kind of understanding of the
21 molecule. That also leads into some of the opioid
22 withdrawal effects that the doctor from NIDA, a
A Matter of Record (301) 890-4188 356
1 person mentioned. And again, I feel there might be
2 some metabolite accumulation and mu opiate
3 saturation that's overwriting samidorphan's
4 effects, and maybe that's what you're seeing.
5 So I think those would also be important
6 studies to do to understand the pharmacology, and
7 then power your study accordingly, is my thought.
8 [Inaudible - off mic].
9 DR. VON MOLTKE: Sorry about that.
10 One thing you are correct about is that the
11 program really did have its genesis sort of
12 backwards from the clinic, starting from an
13 observation, and then again empirically adding
14 samidorphan and bringing it down until we saw, with
15 the visual analog scales, et cetera, that we were
16 extinguishing drug liking and all those effects.
17 So from that standpoint, that is true. I
18 can tell you, though, that back at the time when
19 the 202 study was done, even at that time, there
20 would not have been the presumption that the 8 dose
21 would have been better than the 2/2 dose, just
22 given the history that we had, and that we were
A Matter of Record (301) 890-4188 357
1 also still in the process of adding that antagonist
2 and trying to figure out what ratio we were going
3 to end up with to take forward. And we did end up
4 with our anchor dose out of that study.
5 So I would agree with you it's not as nice
6 and clean as maybe having had the PET imaging and
7 all of those things, but it was empirically
8 determined in a way that made sense and got us a
9 dose that worked in phase 3.
10 DR. NARENDRAN: The Jordan Carp [ph] data, I
11 think which is done with just the buprenorphine, I
12 think it was effective at 0.5 milligrams without
13 samidorphan, which again highlights to me that a
14 very minimal mu opioid receptor occupancy is what
15 you're going in terms of therapeutic effect. And
16 maybe your 2 and 2 is just allowing for that, so
17 maybe 0.25 of buprenorphine in itself might be
18 effective.
19 DR. VON MOLTKE: Right. There's certainly
20 been discussion about that. I think the company
21 really had as its goal something that was not pure
22 buprenorphine. Really, the goal was to put out
A Matter of Record (301) 890-4188 358
1 something that had buprenorphine with something to
2 mitigate it. And while I've heard some
3 conversation about naloxone not fully preventing
4 misuse in terms of parenteral use, certainly
5 samidorphan is quite a bit different than naloxone,
6 which is notoriously short lived, for one thing.
7 Samidorphan is also more potent.
8 So we really wanted to have something that
9 was not pure buprenorphine.
10 DR. NARENDRAN: Thank you. Next question,
11 Dr. Cella?
12 DR. CELLA: Thank you. Mr. Chairman, can I
13 first clarify something with Professor Kulldorff,
14 and then make a comment and question?
15 DR. NARENDRAN: Sure
16 DR. CELLA: I just want to get the term
17 "effect size" because that's the core of my
18 concern.
19 When you were using that term "effect size"
20 I think you were referring to it like a difference
21 in the score on the MADRS, as an effect?
22 DR. KULLDORFF: The difference in the score
A Matter of Record (301) 890-4188 359
1 on the MADRS between the placebo and the --
2 DR. CELLA: Right, the 1.53, 2.0. But the
3 effect size that the sponsor showed overall
4 averaged around 0.2 between groups, and that's an
5 expression of the difference between groups over
6 the variant, the standard deviation or some measure
7 of variability; very small, maybe meaningful, but
8 very small.
9 So that's what I'm struggling with, is that
10 they had a range across the different studies of
11 zero to around 0.6, but most of them in the 0.2 to
12 0.3 range. And we're being asked this question, is
13 this substantial evidence?
14 I'm not as concerned with the one patient or
15 with the design, and I was very moved by the public
16 comment period, very impressed. The distinguished
17 clinicians that spoke talked about responding
18 patients.
19 If you put up the sponsor's slide E-177,
20 which actually gets at what I'm struggling with,
21 the last speaker, who does the systematic reviews,
22 made the comment that there were no differences in
A Matter of Record (301) 890-4188 360
1 response rate between groups, between placebo and
2 the combined therapy.
3 Dr. Jain asked a question earlier that I
4 don't think actually was really answered. You
5 asked about responders in placebo versus -- so I'd
6 like to give the sponsor another opportunity to
7 answer the question of the proportion of
8 benefitting individual patients -- which is shown
9 on this waterfall plot, and to me looks pretty
10 similar. In fact, if anything, it kind of looks
11 like the placebo patients in some ways might do
12 better; at least, there's a lot of them that do
13 quite well.
14 With an effect size of 0.2 across a mixture
15 of negative and positive studies, I find myself
16 really interested in knowing how many people as
17 individuals are really getting a big benefit
18 relative to those on placebo?
19 I'm impressed by the clinicians who spoke in
20 the public session that it may be that this drug,
21 in the right hands, with patient selection
22 carefully done and monitoring done very well, that
A Matter of Record (301) 890-4188 361
1 you can see this benefit that doesn't exist if you
2 just gave a placebo. But I don't see it in this
3 waterfall plot, and I'd like to give the sponsors
4 another chance to answer Dr. Jain's earlier
5 question.
6 DR. PAPAKOSTAS: Thank you. So let me try
7 again to address the excellent question of effect
8 size. We're looking at numbers, and we want to
9 know what does it look like in the real world.
10 The best way that explains it to me
11 personally as a clinician, as a researcher, is to
12 compare it with the precedent; to standardize it,
13 compare it with a precedent. Slide up.
14 This is a standardized effect size for the
15 positive and negative studies of BUP/SAM pooled, as
16 well as the positive and negative studies of the
17 most recently approved adjunct brexpiprazole
18 pooled, and they map on to each other. So in the
19 studies, BUP/SAM has the same efficacy as the most
20 recent approval, which is brexpiprazole. And
21 brexpiprazole, as we know -- I talked a lot about
22 the limitations of the atypicals, but they are also
A Matter of Record (301) 890-4188 362
1 very effective augmentation strategies.
2 Ten years ago, when we did not have
3 atypicals, we were in an even more difficult
4 situation. We need to keep moving forward,
5 replicating the efficacy of the atypicals, as done
6 here, and trying to mitigate all the side effect
7 profiles. So that's how I view it personally.
8 Thank you.
9 DR. VON MOLTKE: We'll have one of our
10 statisticians talk to you about more outlying
11 responses.
12 DR. MEMISOGLU: Hi. Asli Memisoglu,
13 biostatistician, Alkermes. Can I have E-178?
14 Slide up.
15 This is a similar waterfall plot, but rather
16 than MADRS change from baseline, this is showing
17 percent change from baseline, which gives you a
18 clearer view of subjects that meet the criteria of
19 response.
20 This isn't a cumulative distribution. Each
21 bar represents an individual patient. And the
22 dotted vertical line that's towards the middle of
A Matter of Record (301) 890-4188 363
1 the graph shows the point at which patients meet
2 the criteria of response. And you can see that in
3 both stage 1 and stage 2, there's a larger
4 proportion of subjects who were given BUP/SAM 2/2
5 that met the criteria of response.
6 But there are a couple of other interesting
7 points I want to make on this presentation. If you
8 look at stage 2, those bars that occur above the
9 zero line are individuals that are getting worse,
10 and you can see that we have a large proportion of
11 patients that were given placebo that are getting
12 worse. You can also see that the depth of the
13 individuals who exhibit response, so they would be
14 in the lower right-hand corner on BUP/SAM, is
15 greater than those that were on placebo.
16 You can see by the difference of the
17 vertical lines that the greater proportion of
18 subjects on BUP/SAM 2/2 showed a good response
19 compared to those on placebo and provide support
20 for what we're observing with the primary efficacy
21 analysis. Thank you.
22 DR. CELLA: Could you just clarify? It's a
A Matter of Record (301) 890-4188 364
1 minor thing, but it says HAM-D 17 in the title, but
2 MADRS-10 on the side. Is this HAM-D because it's
3 202?
4 DR. MEMISOGLU: It is HAM-D. I apologize
5 for that.
6 DR. NARENDRAN: Thank you. Dr. Conley?
7 DR. CONLEY: Thanks. Two things, and partly
8 about the fairness to industry comment here. I'm
9 really going to talk off of Dr. Farchione and
10 Dr. Ogbagaber's presentation, but just as a general
11 thing. I'm stuck a little bit on the refuse-to-
12 file on the reversal.
13 What I've heard so far today is the thorough
14 presentation by the FDA about your concerns about
15 the data, and I do hear that. But I'm struck on
16 the sentence that the applicant clarified analyses,
17 and then that was okay to submit.
18 I realize all of these things are hard
19 whenever you wind up saying no, but I was also
20 struck by everyone's comments and the efficiency of
21 our whole process here. And I'm wondering if we've
22 gone through the most efficient process, if you're
A Matter of Record (301) 890-4188 365
1 as concerned as you really are, should this have
2 been accepted. Because what I saw from
3 Dr. Ogbagaber's presentation is, actually, you
4 disagreed on all the studies because, in the slide
5 number 2, you disagreed on the two that they
6 thought were positive. And we heard from the
7 sponsor that they feel that there's positive data
8 from 205, which you said you agreed was negative,
9 so that's actually also a disagreement.
10 That seems like a lot of disagreement, and
11 it feels like I'm just a little concerned about
12 that process.
13 The second point, maybe to help this some,
14 about SPCD. On Dr. Ogbagaber's presentation,
15 slides 20 through 22, there's actually a lot of
16 good thoughts about what might help and what your
17 concerns are about this, but I wonder if it's
18 time -- and I know it's hard to issue guidance, but
19 if there's some method to come out, because we
20 can't argue that this is new. I mean, this SPCD
21 was published in, I think, 2003. So it's
22 unacceptable to say 15 years is innovative stuff
A Matter of Record (301) 890-4188 366
1 anymore.
2 Should sponsors think that we need a special
3 protocol assessment or something if you want this
4 or what might be a good method? So those are my
5 two things, a bit about the process, about
6 acceptance, and the other about how to do SPCD.
7 DR. FARCHIONE: So part of the issue with
8 the process, like I mentioned, the back and forth
9 during the development program, there was a lot of
10 very tight overlap between getting things and
11 giving comments and whatnot.
12 At the end of the day, we got the NDA
13 application, when we sent the letter back, the way
14 that we described the analyses for 207 was actually
15 wrong. That was the factual error, so we had the
16 teleconference. And there was the question
17 of -- we had said that we disagreed with the
18 MADRS-6 and we disagreed with the averaging
19 strategy, and yet, there it is; it's still in the
20 plan.
21 But in the strictest sense, they had their
22 prespecified primary analysis because it was
A Matter of Record (301) 890-4188 367
1 prespecified prior to unblinding, so then
2 technically it becomes a review issue at that
3 point, which is why we ultimately ended up filing
4 it.
5 As far as SPCD, one of the conundrums that
6 we keep beating our heads against the wall at this
7 point, a lot of the proposals for analyses are
8 based on simulations. And we had this big meeting,
9 2, 3 years ago or something, where we talked about
10 SPCD in just a dedicated forum just for that, and
11 tried to encourage sponsors, like just send us your
12 data. If you've done an SPCD trial, send us your
13 data so that way, we can start to try to look at
14 those assumptions and the simulation to see if the
15 assumptions in the simulations line up so that we
16 can actually settle on what the best analyses are.
17 Honestly, this is really the first
18 real-world set of data that we've seen. So that is
19 why a lot of it is still unsettled and up in the
20 air. We just don't have the information in order
21 to answer to the questions.
22 DR. CONLEY: Thanks for both.
A Matter of Record (301) 890-4188 368
1 DR. OGBAGABER: Semhar here. So with
2 regards to SPCD design, we've always kept an open
3 mind about it, and also this is the only applicant
4 that sent this SPCD design to DPP, and DPP has
5 never received a SPCD design before. We are open
6 to learning and we are actually encouraging
7 applicants to submit more SPCD designs so we learn
8 more and collect data.
9 So with regards to some of the simulation-
10 based analysis that's been done with regards to
11 weights, weights have been selected from
12 simulations basically, and we can learn more about
13 what weights to pick for stage 1 and stage 2 if we
14 had real data from trials.
15 DR. CONLEY: One thing just in a quick
16 response here -- I know with timing to be very
17 quick -- I hear you, but it is important that if
18 sponsors are actually in the process of trying to
19 get something approved and with a clock ticking and
20 timing, it would be challenging to send in data
21 just at risk to let you collect more data to figure
22 out where your heads are at.
A Matter of Record (301) 890-4188 369
1 So that's why I'm saying, I don't know if
2 guidance can answer it. I realize this is not an
3 easy thing to answer, but I'm putting it up right
4 now. It is a real challenge to give back to the
5 sponsor, to just give us more data when your actual
6 program's at risk when you give that data out.
7 DR. HARRIS: Could I just make two quick
8 points about this? The first was, at our end of
9 phase 2 meeting, we asked about the acceptability
10 of SPCD for our trial designs, and we were told
11 that from a clinical perspective, they were
12 acceptable. We also submitted the 202 data
13 multiple years ago, so they've been at FDA for
14 several years.
15 DR. NARENDRAN: Dr. Iyengar, do you want to
16 comment, being a statistician?
17 DR. IYENGAR: Yes. This is in part in
18 response to your question about SPCD. When I first
19 saw the two treatment-resistant depression and
20 SPCD, I thought something didn't quite match
21 because you're interested in SPCD when there's a
22 large placebo response, and I thought, how could
A Matter of Record (301) 890-4188 370
1 you have a large placebo response with something
2 that's resistant? What does resistance mean in
3 that case?
4 The other thing concerning the weighs, that
5 bothered me because the weights were always pitched
6 as pre-determined, determined a priori, and the 0.6
7 and 0.4 are very common. So I did a little bit of
8 Google searching, and it's only last year, 2017,
9 that a PhD thesis this was written at UNC, which
10 dealt with adaptive determination of the weights,
11 based on the data using a particular model.
12 So although the procedure is very
13 old -- it's about 15 years old now -- there's not a
14 whole lot that's known about it in terms of its
15 properties, not only theoretically, but also how it
16 works in practice. It's not like linear
17 regression, where we have hundreds of years of
18 experience. It's quite limited.
19 DR. NARENDRAN: Dr. Dunn?
20 DR. DUNN: Walter Dunn, psychiatry, UCLA.
21 Some questions and comments on two issues regarding
22 the safety and abuse potential for the product and
A Matter of Record (301) 890-4188 371
1 then also the proposed risk mitigation strategy.
2 So I agree with Raj about the pharmacology
3 in terms of our understanding of buprenorphine and
4 samidorphan and how they relate to abuse potential
5 and withdrawal. It's not entirely clear based off
6 of what I've heard today.
7 So I'm wondering if we can maybe look at one
8 of the other studies. It was mentioned that 208
9 was not designed to look at efficacy, but it was
10 mentioned that it had an usually high retention
11 rate after a year.
12 So I'm wondering from the FDA's perspective,
13 in terms of all the studies that you have, for
14 these kind of one-year type studies, what is the
15 average retention rate, and is this an outlier.
16 Because I'm wondering -- there are a couple
17 possibilities.
18 One, the patient decided to stay in because
19 there are minimal side effects from the medication,
20 which is great. Alternatively, there could be some
21 type of mild euphoria that's not being picked up by
22 the visual analog scale, or there is some slight
A Matter of Record (301) 890-4188 372
1 withdrawal effect when they come off the medication
2 for a couple days, that we know encourages them to
3 remain in the study so they can get the drug.
4 I'm asking this from the perspective of a
5 clinician, because in the real world, patients miss
6 doses. Patients will double or triple up on doses
7 depending on how they feel, and all these questions
8 are not addressed in these studies here. These
9 patients are pretty much taking it every day, and
10 we're looking at withdrawal after a certain amount
11 of time.
12 So I'm wondering, again -- and again, this
13 is not a question we can answer, but is there
14 something unusual about 208 in terms of a number of
15 patients that were retained?
16 So that really speaks to the question of how
17 safe is this drug, are we really treating
18 depression, or are we putting patients on this mild
19 dose of a mu opioid agonist, which actually I
20 believe is probably where most of the,
21 quote/unquote, "antidepressant activity is coming
22 from."
A Matter of Record (301) 890-4188 373
1 Are we really treating the depression or are
2 we just providing them a mild amount of euphoria?
3 Depending on how you define antidepressant effect,
4 is getting them better than their pre-treatment
5 baseline?
6 In terms of the REMS mitigation strategy,
7 I'm wondering if the question of safety is not
8 entirely clear for us, if there's a more
9 restrictive requirement after approval. And I'd be
10 open to ideas of course, but what comes off the top
11 of my head is kind of what's seen for clozapine at
12 this point. Patients are getting a blood draw
13 every month. They get a 30-day supply at most.
14 Is this something that the FDA is
15 considering, where perhaps patients are getting
16 U-toxes to make sure they are taking the medication
17 and that's not being diverted. And if they test
18 positive, they get another 30-day supply, but
19 patients are not given anything more than a 30-day
20 supply.
21 DR. NARENDRAN: We should let the agency
22 comment.
A Matter of Record (301) 890-4188 374
1 DR. FARCHIONE: I think that we've sort of
2 bled into a little bit voting question 2 and
3 discussion -- question number 3. But just with
4 regard to the question about is it an unusual rate
5 of retention in the studies, I don't know off the
6 top of my head what the typical rate of retention
7 is in some of the long-term studies. I know we do
8 see a little bit of attrition over time.
9 But the one thing to remember is that, by
10 the time you get to the long-term, open-label
11 study, those are usually all patients who have
12 responded to the drug and who have just continued
13 on just for the safety evaluation. So they're
14 motivated to stay in because, if they've already
15 responded, then they want to keep taking the drug.
16 DR. MATHIS: This is Mitch Mathis. That's
17 right. I think for a drug that is a controlled
18 substance with an NME that you can't get outside of
19 a clinical trial, you can retain the patients
20 easier than you can a supplement or something
21 already approved, for instance.
22 So that would explain it until proven to be
A Matter of Record (301) 890-4188 375
1 something else, to me, just the opportunity to stay
2 on a drug that has already worked for you and that
3 you've already tolerated.
4 DR. VON MOLTKE: We do actually have some
5 data compared to other studies if that would be
6 helpful.
7 So Dr. Pathak, be quick.
8 DR. PATHAK: Sanjeev Pathak, psychiatrist,
9 Alkermes. Their attention is consistent with other
10 recent adjunct programs. Slide up, please.
11 You can see it for yourself over a period of
12 a year, brexpiprazole, from the publication,
13 51 percent, and also BUP/SAM, 51 percent.
14 DR. NARENDRAN: Dr. Jain?
15 DR. JAIN: Felipe Jain, Harvard Medical
16 School. I think a number of members of the
17 committee are struggling with this question of
18 whether the drug is effective, in part, because we
19 don't have standards of what constitutes
20 effectiveness aside from the historical controls
21 that I think the company is appropriately
22 presenting.
A Matter of Record (301) 890-4188 376
1 We've heard very eloquently from several
2 patient groups, from NAMI, from DBSA, that they
3 really want new treatments that will be effective
4 for individuals. And yet, there's a disconnect
5 between that and the analysis results that the
6 company has submitted and that FDA has requested, I
7 think based on that prior literature, in which
8 we're looking at group averages in symptom
9 reduction that, on their face, are not all that
10 substantial.
11 Within trial 202, the HAM-D reduction in
12 score was just under a 3, where moving that one
13 outlier brings it down to just over a 2. Now, if
14 I'm seeing a patient in my clinic and saying I'm
15 going to give you this medication for your
16 treatment-resistant depression that is going to
17 improve your depression by a little over a 2, if
18 that's about the median if you are analyzing the
19 data in that way, that doesn't sound on its face to
20 me like an effective strategy.
21 Now, if I had the numbers on response and
22 remission across the trials and I could compare
A Matter of Record (301) 890-4188 377
1 those between placebo and BUP/SAM, following on
2 what Dr. Bodkin was saying about his results as an
3 individual provider, that would help me to
4 understand a little bit more about how to look at
5 this in terms of its effectiveness in clinical
6 practice.
7 So it seems to me that the public is asking
8 for a different set of analyses than the sponsor is
9 submitting, and the sponsor is submitting those on
10 the basis of the prior literature, looking at
11 treatment-resistant depression. And the effects of
12 it, they're just really hard to grapple with a 1 to
13 to 2-point change on the MADRS on average. If I'm
14 dealing with an individual patient, that's really
15 hard to grapple with.
16 That's one of the reasons that I'm
17 struggling so much with answering this question. I
18 know it's an active area, but it would be really
19 useful to know from your perspectives why response
20 and remission data are not required in this kind of
21 situation, and just your overall thoughts on
22 individual level of effectiveness.
A Matter of Record (301) 890-4188 378
1 DR. TEMPLE: We are very interested in the
2 distribution of effects. We're always interested.
3 In the typical depression trials, however, a
4 typical result would be, you start out at a HAM-D
5 of 24, 25. The placebo group will change by 10 or
6 12 and the drug group will change by 14. That's
7 the difference. But there are huge changes in both
8 groups.
9 Whether if we looked -- and I'm not sure we
10 have -- whether if we looked to define some kind of
11 remission rate or something like that, you'd see a
12 statistically significant difference between the
13 groups. I'm not sure. I don't think we've done
14 that, and we've relied on the mean change.
15 But there is obviously a distribution of
16 results, and we're getting increasingly interested
17 in all areas in showing what the distribution of
18 results is.
19 DR. JAIN: And thank you for that
20 clarification. Obviously, it's not a 1- to 2-point
21 reduction in absolute change from baseline. It is
22 relative to placebo.
A Matter of Record (301) 890-4188 379
1 DR. TEMPLE: It's the difference between the
2 two. Right.
3 DR. JAIN: It's the difference.
4 DR. UNGER: Yes. I'll speak on behalf of
5 the applicant here. This is Ellis Unger. If the
6 mean change is 2 points, but actually a third of
7 people respond and no one else responds, those
8 third of people are getting a 6-point difference.
9 So we've been very interested in the
10 distribution. And in the last, I don't know, 5 or
11 6 drugs we've approved for neurology indications,
12 we've shown histograms that show the distribution
13 of response, and we intend to keep doing that. I
14 think we've done that for a couple psychiatry drugs
15 also. It's very important.
16 DR. TEMPLE: We used to show cumulative
17 distributions, but it turns out nobody can read
18 those. So what we like to show is bar graphs
19 divided into change of 5, change of 6, change of 7,
20 and we are very interested in that.
21 DR. VON MOLTKE: So the sponsor does have
22 some more data with regard to comparative effects
A Matter of Record (301) 890-4188 380
1 if that would be helpful, to just quickly see the
2 slide. Dr. Pathak?
3 DR. PATHAK: Sanjeev Pathak. As regards to
4 comparative effectiveness or difference versus
5 placebo, the results are very consistent with other
6 approved adjunct therapies. And, importantly, I
7 think it's important to recognize that these are
8 patients who have gone through several failures,
9 have been suffering from symptoms for a substantial
10 period of time.
11 Slide up, please. These are deltas on MADRS
12 scales, MADRS end of treatment. At the top are the
13 three studies from BUP/SAM, 202, 205, and 207. And
14 you can see the point estimate and the confidence
15 intervals. And at the bottom are the three adjunct
16 antipsychotics and their studies. And this
17 information comes from their labels. Brex is the
18 most contemporaneous or most recently approved
19 antidepressants.
20 What you can see is that the deltas are
21 comparable and these are clinically meaningful
22 differences. And Dr. Papakostas would like to add
A Matter of Record (301) 890-4188 381
1 something.
2 DR. PAPAKOSTAS: Yes. Just briefly let me
3 add; I don't want to take too much time up, but
4 having done one of the first studies of the
5 atypicals, the most cited meta-analysis on the
6 atypical -- I just want to share, having published
7 a meta-analysis on atypicals in the American
8 Journal of Psychiatry, highly cited, this pretty
9 much maps up to the efficacy of the atypicals in
10 terms of standardized effect sizes.
11 So you see here the distribution to the
12 three studies, the two positive studies, as well as
13 the remaining studies for the other agents. That's
14 not where you see a difference. The difference in
15 my mind is in the risks that patients do not have
16 to take with this medication. Thank you.
17 DR. NARENDRAN: Agency comments? No.
18 DR. FARCHIONE: The only other thing I was
19 going to say with regard to response and remission
20 is, in fairness, that's not something we usually
21 ask for as a primary, in part, because trying to
22 settle on what's the best definition for that is a
A Matter of Record (301) 890-4188 382
1 little bit challenging in a clinical study.
2 Do you have a numerical cutoff for your
3 definition or do you define it by percent
4 improvement? If you could define it by percent
5 improvement, sometimes people who improve by
6 50 percent, if they were really sick at baseline,
7 they're still sick enough to get into another
8 trial.
9 So it's very sticky to try to figure out,
10 which is why we've just settled on the change from
11 baseline in a score. But we encourage looking at
12 that as secondary exploratory analyses, less than
13 secondary.
14 But I think part of what it feels like
15 you're asking about is, is there a way that you can
16 look at the patients who do respond and try to
17 figure out what's unique about them versus the
18 people who don't. That we would love to have for
19 any of the medications that we have approved, or
20 unapproved, or whatever. And it would be terrific
21 if there was some way that you could look at those
22 patients who did better and tell us what was
A Matter of Record (301) 890-4188 383
1 different about them that made them more likely to
2 respond, but we just don't have that.
3 If we ever get that at some point, that's
4 going to be a real revolution in psychiatry, but
5 we're just not there at the moment.
6 DR. NARENDRAN: Dr. Riley?
7 DR. RILEY: So I don't have concerns about
8 the outlier. I don't think we ought to exclude
9 participants from clinical trials purely on
10 statistical grounds alone. And I'm not really that
11 concerned about the SPCD. I know we have debates
12 that it's a relatively new approach, and some of
13 the parameters, and how they're set, and those
14 types of things. But I think, by and large,
15 there's enough acceptance, enough studies of that,
16 that I'm not too concerned about that.
17 I am still trying to wrap my head around the
18 shift to the MADRS-6 as the primary endpoint. And
19 the reason I'm particularly concerned about that is
20 this question has to do with major depressive
21 disorder, and the public comments were very clear
22 about not only mood and affect being an important
A Matter of Record (301) 890-4188 384
1 component of this, but suicidal ideation, and sleep
2 disturbance, and appetite disturbance, et cetera,
3 et cetera, none of which are measured by the items
4 that were excluded from the MADRS-10 to produce a
5 MADRS-6. So basically, you're only measuring one
6 facet.
7 I don't believe that a measure has to
8 measure every single facet of a construct to be
9 content valid or valid in general, so I think you
10 can do it with a subset of items. I think that's a
11 pretty reasonable assumption. But I think there's
12 a need here -- and I don't know if the sponsor has
13 it or if others have it, but a need to be able to
14 show that the MADRS-6 does map reasonably well onto
15 clinical indicators of major depressive disorder
16 and other measures of clinical depression. And if
17 we have that data, I would feel a little more
18 comfortable with using the MADRS-6 as a primary
19 endpoint. But I'm still trying to figure out why
20 the shift to the MADRS-6 as a primary endpoint.
21 DR. VON MOLTKE: May the sponsor comment on
22 that?
A Matter of Record (301) 890-4188 385
1 DR. NARENDRAN: Briefly.
2 DR. VON MOLTKE: We did have interest based
3 on the literature, but I will say that, once we
4 heard FDA with regard to the fact that this really
5 was not at a point yet where they were ready, we
6 backed down to the second in the hierarchy, which
7 was the MADRS-10. And it is on the MADRS-10 that
8 the 207 is being asked to be regarded as positive.
9 So we continue to be interested in MADRS-6, but for
10 this application, it's MADRS-10 for the 207 study.
11 DR. NARENDRAN: Thank you. That's clear.
12 Dr. Crawford?
13 DR. CRAWFORD: Thank you, Mr. chairman. My
14 comment was very much like Dr. Riley's, but perhaps
15 with a slightly different twist at the end. And I
16 agree with everything you've stated as well as we
17 heard the totality of everything presented from the
18 applicant, sponsor, the agency, and the powerful
19 statements from the public speakers.
20 But if we could show the question again,
21 posed to the committee, I would like clarification
22 from the agency, is this a binary yes or no
A Matter of Record (301) 890-4188 386
1 question? Because as I'm looking at the DSM IV
2 criteria, if one were to accept the design and the
3 data presented with that caveat, then the evidence
4 presented would meet at least the 5 minimum
5 criteria for the diagnosis of MDD.
6 I struggle a bit more if it is truly a
7 binary question in terms of if it were approved,
8 would the labeling be limited to mild depressive
9 episodes, or -- because the data with the MADRS-6
10 average, with all respect to the applicant, that is
11 the basis for concluding efficacy, not the MADRS-10
12 average. But we don't see what we did here about
13 the suicidal thoughts. We don't hear about a sleep
14 disturbance as a concentration and the appetite.
15 So Mr. Chair, I guess I'm asking for
16 clarification about how we should interpret this
17 question.
18 DR. MATHIS: This is Mitch Mathis. So the
19 way we usually do this is traditionally we have a
20 couple of endpoints that will measure this thing we
21 call major depressive disorder. And the Hamilton-D
22 and the MADRS both will suffice, and there are some
A Matter of Record (301) 890-4188 387
1 newer ones that we've been evaluating.
2 The sponsor in terms of meeting the
3 qualification of adequate and well-controlled for
4 substantial evidence must identify that ahead of
5 time and tell us what they expect to change before
6 they enroll in randomized patients, so that when we
7 get the results, we can compare the results with
8 some competence, that we're measuring the same
9 thing in both groups and any differences are
10 related to drug. That's how it works.
11 As they pointed out, they picked MADRS-6.
12 They did that faster than we could agree to it or
13 not agree to it, but the second was a MADRS-10
14 average. And they went to the MADRS-10 average and
15 they won. We hadn't agreed to that, but we've said
16 that we think averaging might have some utility at
17 some point. We were going to analyze that.
18 Then at MADRS end of treatment, which was
19 the third endpoint, the usual endpoint, they did
20 not make it. So for us, it is binary. You either
21 have substantial evidence or you don't, and that's
22 the decision that we're asking you for. And we've
A Matter of Record (301) 890-4188 388
1 never written a label where we said maybe the four
2 or five parts of the MADRS-10 changed, so we'll
3 label it for that and not for everything else.
4 It's major depressive disorder or not.
5 DR. NARENDRAN: I think we've had a lot of
6 discussion on question 1. I know there's a couple
7 people who have questions unless you have something
8 really burning that you wanted to make.
9 Dr. Griffin, real succinct, quick?
10 DR. GRIFFIN: Pardon?
11 DR. NARENDRAN: Succinct discussion.
12 DR. GRIFFIN: Sure. I just want to say
13 that, as a primary care physician, treating people
14 with these disorders, I understand the need for
15 more drugs or better drugs, but I also appreciate
16 FDA's charge with substantial evidence because I
17 think it's really important to get the information
18 before the drug gets out there because I think this
19 drug was tested in about a thousand people, but it
20 is going to be used in many tens or hundreds of
21 thousands of people, over many, many years.
22 So I think we want to be really, really
A Matter of Record (301) 890-4188 389
1 confident that this is an effective drug before it
2 gets licensed and used. And I think it's a very
3 interesting drug and we saw some very interesting
4 data that's intriguing, but I think the question is
5 do we have substantial evidence, and I think that's
6 really important.
7 DR. NARENDRAN: Dr. Joniak-Grant?
8 MS. JONIAK-GRANT: Elizabeth Joniak-Grant.
9 Just a quick comment, really. I'm concerned that
10 the study population was about 75 percent whites
11 and the remaining was largely blacks, leaving out
12 Asians and Hispanics, Latinx, particularly because
13 when you look at some of the results for the
14 study 202 and 207, there are differences in
15 effectiveness by race. And that just sort of
16 disappears in the discussion, and I think that's
17 something that's worth further investigation.
18 DR. NARENDRAN: Good point. Thank you.
19 I think we could vote on the question and
20 make our comments later. I think we've had a lot
21 of discussion on this particular -- Dr. Cella, go
22 ahead.
A Matter of Record (301) 890-4188 390
1 DR. CELLA: I really apologize,
2 Mr. Chairman, but this question affects my vote.
3 This is a question to probably Dr. Mathis, or
4 Unger, or Temple. The sponsor is suggesting, by
5 what they've presented, that we should consider
6 recent approvals and the effect size in those
7 recent approvals in the neighborhood of what I saw
8 as 0.2 in these typical antipsychotics, as a basis
9 for saying that that's a reasonable effect size to
10 approve a drug on.
11 Do you agree with that or should this be an
12 individual decision? There's a precedent they put
13 on a slide. I'm wanting to know if you agree with
14 that as precedent.
15 DR. UNGER: We base approvals on whether or
16 not there's substantial evidence of effectiveness.
17 We don't get into what's a clinically meaningful
18 effect size.
19 DR. TEMPLE: I don't think that's true. We
20 have to believe that whatever the effect size was
21 matters. There's a court precedent for saying
22 that, Warner-Lambert v. Heckler. So we have to
A Matter of Record (301) 890-4188 391
1 believe it's in the ballpark for that. It is in
2 the ballpark for what previous approvals have been,
3 maybe a little less, but it's not way different.
4 Then, as I always like to point out, the
5 mean isn't the whole deal. There's a distribution
6 of results, and you've got to think about that,
7 too.
8 DR. FARCHIONE: But also remember that for
9 those trials, they did have two adequate and
10 well-controlled studies with positive results in
11 each of those studies individually, and they were
12 able to demonstrate it within the trial. It
13 doesn't have anything to do with comparison across
14 trials or between different drugs. It's whether or
15 not you have the adequate and well-controlled
16 positive studies for your drug.
17 DR. TEMPLE: Where they showed an effect.
18 It's not so much whether it was 2.2 or 2.3. They
19 showed an effect it was nominally significant in
20 each of two trials.
21 DR. FARCHIONE: Right. We've got to get to
22 the p-value before we can start talking about
A Matter of Record (301) 890-4188 392
1 whether the effect is clinically meaningful or not.
2 DR. MATHIS: Does that answer your question?
3 (Laughter.)
4 DR. NARENDRAN: That was three different
5 answers, but very helpful.
6 (Laughter.)
7 DR. NARENDRAN: I assume there's no further
8 discussion on this question, and we will now begin
9 the voting process. So please press the button on
10 your microphone that corresponds to your vote. You
11 will have approximately 20 seconds to vote. Please
12 press the button firmly. After you have made your
13 selection, the light may continue to flash.
14 If you are unsure of your vote or you wish
15 to change your vote, please press the corresponding
16 button before the vote is closed, and that should
17 change it.
18 I'll read the question. Has substantial
19 evidence been presented by the applicant to support
20 the effectiveness of buprenorphine/samidorphan for
21 the adjunctive treatment of major depressive
22 disorder?
A Matter of Record (301) 890-4188 393
1 (Voting.)
2 MS. BHATT: Voting results,. 3 yes; no 20;
3 abstain zero; no voting zero.
4 DR. NARENDRAN: I think, now that the voting
5 is complete, we'll just go around the table and
6 have everyone who voted state their name, vote, and
7 if you want to, state a reason why you voted and
8 the reason you voted. We'll start from here.
9 DR. CELLA: David Cella. I voted yes, a
10 very reluctant yes based upon that interesting
11 answer to the question, but expected that will
12 travel to question 3.
13 DR. RILEY: A reluctant no from me.
14 DR. CRAWFORD: Stephanie Crawford, no for
15 the reasons stated, concerns about the MADRS-6
16 average.
17 DR. MARSHALL: Brandon Marshall. I voted no
18 because of the inconsistency of the outcomes across
19 studies 202 and 207.
20 DR. FLYNN: Kathryn Flynn. I voted no
21 because the definition of substantial evidence ,
22 meaning at least two positive studies, and I didn't
A Matter of Record (301) 890-4188 394
1 see that convincingly presented.
2 DR. JENSEN: Roxanne Jensen. I voted no for
3 some of the similar reasons mentioned, and I felt
4 that using the average endpoint for 207 versus end
5 of treatment did not meet the threshold.
6 DR. DE WIT: I voted no for the same
7 reasons, that the magnitude of the effect didn't
8 meet the standards of clinically significant to me.
9 DR. ACRI: I voted a very reluctant yes
10 because it all kind of hinged on what the
11 definition of substantial was. And if two positive
12 trials is considered substantial, then I felt like
13 they had that, but obviously the effect size was
14 fairly small and there was some controversy about
15 the methods and the analysis.
16 DR. MEISEL: Steve Meisel. I voted no, a
17 number of reasons, most of which have been
18 discussed here. We don't know the mechanism of
19 action of this drug. We have no idea, just some
20 postulates about this, and so we're just throwing
21 something we think might have some good, but we
22 don't know why even if it did.
A Matter of Record (301) 890-4188 395
1 We don't know if the impact that we are
2 seeing, which is modest, is that of an
3 antidepressant or of a narcotic euphoric effect.
4 The impact is mild. And I'm no statistician, but
5 when I took statistics back in college, I was told
6 that you can put out a graph or a statistical model
7 to prove anything you want.
8 The controversy we've heard today about the
9 stats and the fact that the sponsor changed their
10 definitions and changed their presentation all
11 along the way, despite recommendations by the FDA
12 along the way, to me is very concerning.
13 I think instead of starting off with a model
14 and let's see what this drug does, yes or no,
15 there's a likelihood here that we tried to retrofit
16 a model to prove the point. And there's no
17 evidence of that, but I think all of the
18 controversy, and the discussion, and the history of
19 the discussions with the FDA make me very
20 concerned.
21 DR. GRIFFIN: This is Marie Griffin. I
22 voted no for lack of substantial evidence. I also
A Matter of Record (301) 890-4188 396
1 didn't like including 202, which the FDA
2 persistently said was not eligible and was a pilot
3 study.
4 DR. BESCO: Kelly Besco. I also voted no
5 and just want to also commend and say that I
6 appreciate the members of the public that did share
7 their personal experiences today and just want it
8 known that I do also support the need for
9 additional new innovations around treatment for
10 resistant treatment depression.
11 But I did vote no because I think there is
12 additional need for more confirmatory evidence.
13 MS. JONIAK-GRANT: Elizabeth Joniak-Grant.
14 I voted no. I really was excited that there was
15 some attempts to try and look at depression,
16 treating it in an innovative way. It obviously is
17 a need that needs to be filled.
18 Unfortunately, though, with this one today,
19 I'm not convinced of the efficacy. I think there
20 were just too many changes sort of all along the
21 way with this study to what qualified as
22 effectiveness. I also have concerns about the
A Matter of Record (301) 890-4188 397
1 homogeneity of the population, especially when
2 whites were showing in certain things that they
3 were showing better effectiveness.
4 MS. WITCZAK: Kim Witczak, and I voted no.
5 And I think I voted no because there was so much
6 controversy and things changing. And as a
7 consumer, it's really hard because I know that
8 people are desperate for treatments. But with that
9 much controversy, and if the public was actually
10 here to hear this debate, I think it would be
11 really eye opening. And at the end of the day, we
12 pay the ultimately price because we're the guinea
13 pigs.
14 DR. FIEDOROWICZ: This is Jess Fiedorowicz.
15 I very much appreciate the need for innovative and
16 better treatments for this horrible disease. In
17 reviewing the trials, I agreed with the FDA that
18 202 should be considered an exploratory study. I
19 had concerns about risk of type 1 error in 207 with
20 the weight change in prior outcome and when it was
21 only significant with the late change and one that
22 was not approved by the FDA.
A Matter of Record (301) 890-4188 398
1 Even looking at the sponsor-selected
2 analyses, the effect size was very small, and I
3 thought could even be overestimated because there
4 is uncertainty about the integrity of the blind
5 when you have a medication that did have some
6 immediate effect that was discernible on drug
7 liking, and the inclusion of patients with pain,
8 which I didn't see measurement of.
9 So when you put that, at best, a small
10 effect against the risks and these concerns for
11 type 1 error, I voted no.
12 DR. NARENDRAN: Raj Narendran. I voted no.
13 Dr. Meisel crystallized my comments, but the
14 pharmacology to me was a question. I'm still
15 concerned about the 8-milligram didn't show effect
16 and the 2-milligram did.
17 202 in general raised questions. I was a
18 little concerned about the single subject. It at
19 least gives me pause that that subject drives the
20 results so strongly.
21 I was also concerned about the late change
22 for 207 of the outcome measure. And I just feel
A Matter of Record (301) 890-4188 399
1 like there's probably more that needs to be done.
2 A lot was learned on the fly and adjusted for 207
3 to get it to be a positive trial perhaps, but maybe
4 another trial would give me more confidence that
5 this is effective or substantial effectiveness has
6 been shown.
7 DR. JAIN: Felipe Jain. I'd like to
8 underline the reasons given by each of the members
9 of the committee who voted no, and I agree with all
10 of them, which I find myself surprised by.
11 As someone who has prescribed opiates to
12 patients with treatment-resistant depression and
13 feels that they may have some benefit within our
14 armamentarium, I still did not feel that the
15 evidence presented by the sponsor was convincing
16 of substantial effectiveness.
17 In addition to what has already been said,
18 I'd also like to underline the lack of transparency
19 regarding the individual subject-level data.
20 DR. DUNN: Walter Dunn, psychiatrist, UCLA.
21 I voted yes with a qualification. But in terms of
22 the yes vote, I was willing to accept the inclusion
A Matter of Record (301) 890-4188 400
1 of that patient from site 124, also impressed with
2 the fact that even with the removal of that
3 patient, the stage 2 data was still significant.
4 In terms of the change for 207, I was
5 convinced that the analysis using the average had
6 clinical relevance and that it remained relevant
7 even if you looked at weeks 2 through 5, 3 through
8 5, 4 through 5.
9 The qualification with the yes vote would
10 have been the recommendation that a much more
11 restrictive REMS be applied. I certainly
12 appreciate the observation that the overall
13 clinical effect size was minimal to moderate.
14 However, as we all know, there are going to be some
15 patients who are fantastic responders, and we need
16 to balance that out with potential risk for misuse
17 or abuse for this particular medication.
18 So it would have been with the
19 recommendation that, yes, we approve. However,
20 that clinical effectiveness was demonstrated, but
21 with a much more restricted REMS to what I spoke
22 before about perhaps limiting 30-day supplies,
A Matter of Record (301) 890-4188 401
1 coming in monthly for U-toxes.
2 DR. IYENGAR: Satish Iyengar. I also voted
3 no, in part, because I didn't think there was
4 substantial evidence. I also just saw too many
5 methodological issues that swayed me this way.
6 DR. HERNANDEZ-DIAZ: Sonia Hernandez-Diaz.
7 I voted no, and I didn't have a problem with type 1
8 error with exclusion or inclusion of the patient.
9 My methodological problem was more with the
10 cherry-picking of what studies to present, within
11 each study, which dose you choose to present, or
12 changes in the outcomes to present from the
13 protocols during the trials.
14 Despite this cherry-picking, if the effect
15 was huge, I would be voting yes, but I think,
16 despite that, I don't see a substantial and
17 clinically meaningful difference with placebo. We
18 could wait one more week for the placebo in some of
19 the trials, and we would be in the same room. So
20 for that reason, I voted no.
21 But having said that, I think that at the
22 same time, there is no evidence of lack of effect
A Matter of Record (301) 890-4188 402
1 of pursuing this potential effective therapy. So I
2 hope that we keep doing research and try to find an
3 effective therapy in this area of the opioids.
4 DR. HABEL: I'm Laurie Habel. I voted no
5 for many of the reasons that other members did. I
6 thought that three of the four studies did have
7 results that were suggestive of efficacy, but they
8 just weren't robust enough; 2 out of the 3 were
9 only statistically significant with certain outcome
10 measures and not with all the outcome measures.
11 And so that inconsistency bothered me.
12 DR. KULLDORFF: I'm Martin Kulldorff. I
13 voted no. I don't think there is evidence that
14 this drug works, and I think it's unethical to give
15 patients a drug if we don't know that it works. If
16 1.5 was the point estimate for the effect size
17 comparing the drug with placebo, if that's the
18 point estimate, if that's a clinically important
19 effect size, then I hope that the applicant will
20 conduct a study with appropriate sample size and
21 appropriate power.
22 If there truly is an effect size of 1.5,
A Matter of Record (301) 890-4188 403
1 then you will find out that, and then we will know
2 that it is both effective and that it is clinically
3 important if 1.5 is that. If 1.5 is not a
4 clinically important effect size, then I hope you
5 will continue your innovative thinking about these
6 drugs to find a different way to help these
7 patients.
8 In either case, I would like to thank the
9 applicant for their strong efforts of trying to
10 find something that helps patients. Thank you.
11 DR. WARHOLAK: I'm Terri Warholak, and I
12 voted no. I recognize that major depression is an
13 issue. And from the public comments, I was very
14 moved by the unmet need and that the public is
15 asking for innovative solutions. And I commend the
16 applicant for giving this a try, but I don't feel
17 that the evidence presented at us was substantial
18 in order to show efficacy. I think there were a
19 lot of methodological issues, and I think
20 Dr. Meisel's comments really captured what I wanted
21 to say.
22 I'm very concerned about the last-minute
A Matter of Record (301) 890-4188 404
1 changes, especially with the things that were
2 changed at the last minute being significantly
3 different. And I'm not convinced that there's a
4 big enough treatment effect to make a clinical
5 difference, either. But I'd like to see efforts in
6 this area to continue.
7 DR. RUHA: Michelle Ruha. I voted no. My
8 no vote really hinged on the FDA's definition of
9 substantial evidence. I think by looking at all of
10 the evidence the sponsor presented that the drug
11 probably does work, but maybe only in a specific
12 population of patients that we haven't identified
13 yet. However, that one outlier in study 202
14 actually bothered me, too, because I don't feel
15 very comfortable with any study where you eliminate
16 one subject and it changes the significance of the
17 study. So I did not feel like it met the
18 definition of substantial evidence.
19 DR. NARENDRAN: So just to summarize, it
20 sounds like people had concerns on multiple levels
21 in terms of whether a substantial effectiveness
22 standard has been met. There was a question about
A Matter of Record (301) 890-4188 405
1 the single subject, questions about the in vivo
2 pharmacology. Maybe there's opiate withdrawal
3 effects or not clearly delineated, concerned about
4 change in outcome measure, how things are done at
5 the late stage, inclusion of people in pain, racial
6 differences across outcome measures.
7 So with that being said, we'll move to the
8 next question. Question number 2 is also a voting
9 question. Has the applicant adequately
10 characterized the safety profile of buprenorphine/
11 samidorphan for adjunctive treatment of major
12 depressive disorder?
13 So this is also a voting question. Please
14 press the button on your microphone that
15 corresponds to your vote. You will have 20 seconds
16 to vote. Press the button firmly. After you have
17 made the selection, the light may continue to
18 flash. If you are unsure of your vote or want to
19 change your vote, you can change it.
20 (Pause.)
21 DR. NARENDRAN: One more person hasn't
22 voted. Everybody, press the button again. If
A Matter of Record (301) 890-4188 406
1 everybody does it again, that person's vote should
2 register.
3 (Voting.)
4 MS. BHATT: Voting results, 13 yes' 10 no;
5 zero abstain; and no voting is zero.
6 DR. NARENDRAN: We'll go around the table
7 and answer the questions. State your name, your
8 vote, and the reason. We'll start from here,
9 Dr. Ruha.
10 DR. RUHA: I voted yes. There really
11 weren't any severe safety concerns, and I thought
12 they did a good job showing even that it really
13 wasn't a huge issue with withdrawal or euphoria.
14 DR. WARHOLAK: This is Terri Warholak and I
15 voted no. I was concerned that we only had studies
16 on abuse for the single dose. I didn't see
17 anything on drug-drug interactions, and I would
18 have loved to have seen some of the studies that
19 the sponsor alluded to about what it took to
20 separate the components for abuse prevention.
21 DR. KULLDORFF: Martin Kulldorff. I voted
22 yes. One can never be completely sure about safety
A Matter of Record (301) 890-4188 407
1 before an approval of a drug, but I thought that
2 applicants did a nice, thorough job, as well as is
3 possible to do in a pre-approval process. So thank
4 you.
5 DR. HABEL: I'm Laurie Habel. I voted no
6 basically for the same reason as Terri did.
7 DR. HERNANDEZ-DIAZ: Sonia Hernandez-Diaz.
8 I voted yes because I think they did characterize
9 the safety profile. Of course, we want more
10 evidence as it accumulates if it's approved. But
11 they characterize it in the sense that they accept
12 that there is some [indiscernible] opioid effects,
13 like [indiscernible] constipation and so forth.
14 But there is some data to support that there is no
15 huge withdrawal within a large proportion of
16 patients at least.
17 DR. IYENGAR: Satish Iyengar. I voted yes,
18 in large part because I thought they did a good job
19 of addressing head on the opioid crisis and sort of
20 their approaches to that problem.
21 DR. DUNN: Walter Dunn. I voted yes based
22 on the very defined time periods that were looked
A Matter of Record (301) 890-4188 408
1 at in the studies. I voted yes with the comment
2 that I think things like long-term use in terms of
3 withdrawal or abuse should also be looked at maybe
4 perhaps with post-marketing studies and also for
5 the potential for drug-drug interactions. We know
6 that our patients aren't just taking this alone in
7 the real world.
8 DR. JAIN: Felipe Jain. I voted no, not
9 because I thought that this combination drug was
10 unsafe relative to other opiates on the
11 market -- buprenorphine's already on the
12 market -- but because I did not feel that the
13 sponsor had adequately addressed its safety in this
14 population of major depressive disorder.
15 Specifically, thinking about some of the
16 concerns raised by Dr. Sullivan in his presentation
17 and the short-term nature of the placebo-controlled
18 trials in clinical practice, treatment-resistant
19 depression patients are not going to be treated for
20 5 weeks, are not going to be treated for 6 weeks.
21 They're going to be treated continuously. And I
22 think that their safety does need to be assessed
A Matter of Record (301) 890-4188 409
1 relative to some control over a more extended
2 period of time.
3 There was a paradoxical reasoning as well
4 that I found myself trying to grapple with, which
5 was that there was a clear signal for patients to
6 immediately like the drug and for that to be
7 present on a dose-response basis, higher with 8/8
8 than with 2/2. Yet, that didn't seem to be fully
9 addressed within the analysis.
10 Some of the individual-level data regarding
11 the clinical opiate withdrawal scale, again, were
12 obscured by grouping people into categories of mild
13 withdrawal symptoms, which span a huge range of
14 symptomatology, moderate withdrawal symptoms.
15 So although I feel that there's some promise
16 for this type of approach for treatment of
17 treatment-resistant depression, I have more
18 questions regarding its safety in this population
19 than I have answers.
20 DR. NARENDRAN: I voted no because I'm still
21 concerned that it does work through the mu opioid
22 receptor. I think there is some euphoria effects.
A Matter of Record (301) 890-4188 410
1 There's probably some withdrawal associated with
2 it. I'd also like to know more about that
3 metabolite and how it accumulates in 7 days; what
4 can that mean?
5 I also think it would be very difficult to
6 screen patients for methadone or things like that.
7 Buprenorphine's prescribed by other providers, not
8 necessarily by the PCP or the psychiatrist. These
9 people are going to get put on this med, so I think
10 there's a lot of issues that could occur and create
11 problems in the real world, so I voted no.
12 DR. FIEDOROWICZ: I'm Jess Fiedorowicz. I
13 voted yes with similar caveats that were noted by
14 Dr. Dunn about a postmarketing need for a longer-
15 term controlled study. I was influenced by the
16 FDA's statement that they said they didn't believe
17 short-term trials were adequate and that the
18 tradition is to do the short-term first and require
19 other studies postmarketing. So that's why I voted
20 yes.
21 MS. WITCZAK: Kim Witczak. I voted no. I
22 voted no, I think, for a couple reasons. We don't
A Matter of Record (301) 890-4188 411
1 know the long term, so we don't know what we don't
2 know. So that's one thing, ultimately. And then I
3 heard the public speakers talking a lot about the
4 suicide, and really that being a risk, and also
5 thinking a lot about who the ideal patient is,
6 J.D., and then listening to the presentation that
7 we had by our guest outside speaker, talking about
8 that depression or with the opioid use. And I feel
9 like there isn't enough data to say that it's
10 completely safe yet.
11 MS. JONIAK-GRANT: Elizabeth Joniak-Grant.
12 I voted no for a couple of reasons. First, I
13 couldn't help but wonder sort of what happens if an
14 individual needs acute pain treatment, especially
15 if they can't communicate that they're on this
16 drug. It's like they go into an emergency room or
17 something like that.
18 I would have liked to have seen the adverse
19 events looked at by subpopulation because this is a
20 population with a number of comorbidities and also
21 looked at it in relationship to what other drugs
22 they may be taking.
A Matter of Record (301) 890-4188 412
1 I was concerned that there were few patients
2 over 65 and, in fact, in study 202, there was no
3 one included that was over the age of 65. That was
4 exclusion criteria. We see more comorbidities in
5 older populations and more drug usage. That could
6 cause problems. I was concerned that the
7 population, again, was homogeneous with mostly
8 white and mostly female participants.
9 Really, to echo what Terri said, I would
10 really like to see the data about the ability to
11 separate buprenorphine from the antagonist. I'd
12 like to see the data. I think that would be really
13 interesting because people are really good at doing
14 that. They can do it in all kinds of ways we never
15 really thought of, and I think some more
16 information about how that could operate in the
17 real world is super important.
18 DR. BESCO: Kelly Besco. I voted no as well
19 for many of the same reasons, also worried about
20 the separation aspect and not knowing what those
21 solvents are, and the fact that they could be under
22 anyone's kitchen sink, for all that we know.
A Matter of Record (301) 890-4188 413
1 Also, just another real-life practice note,
2 as an inpatient pharmacist, I see more and more
3 patients coming in on medication-assisted therapies
4 for OUD. And they come in and they need treatment
5 for acute pain. And I see physicians -- and
6 hopefully Dr. Griffin won't give me a dirty
7 look -- just throw opioids at these patients trying
8 to counteract that antagonist effect. And
9 sometimes, we get a little aggressive and we start
10 stacking the opioids and have an unintended over-
11 sedation event.
12 So I agree with Dr. Dunn, if this medication
13 or a medication like it would move forward, we
14 would need a multi-layered robust REMS program
15 around it.
16 DR. GRIFFIN: Marie Griffin. I voted no for
17 many of the same reasons. I think this is a really
18 vulnerable population with high pain, high use of
19 opioids or opioid use disorder. So I think even
20 though they're not required to have more than 5- or
21 6-weeks trials, I think really, for opioids, we
22 need to think about if they're going to be used
A Matter of Record (301) 890-4188 414
1 long term, maybe we need longer-term trials.
2 DR. MEISEL: Steve Meisel. I voted no for
3 much of the same reasons. Adequately characterized
4 safety profile, I think the characterization here
5 has been very, very narrow. First of all, the
6 issue of samidorphan, I have never heard of a
7 narcotic antagonist having an active metabolite
8 being an agonist. That seems a little weird. But
9 that has not been quantified or qualified in any
10 way, shape, or form.
11 The fact that it's got some half-lives that
12 are moderate, 4 to 5 to 7 hours; the situation that
13 Dr. Besco described, a patient coming in with acute
14 pain, they've broken their leg, and now they need
15 narcotics, if they come in on one of these drugs
16 that's used for opiate abuse syndrome, people know
17 it. If they're on naltrexone, they know it. If
18 they come in on one of these or like a drug like
19 Contrave, which is used for weight loss, people
20 don't realize there's a narcotic antagonist in
21 there. So people aren't going to realize that
22 they're trying to overcome the narcotic antagonist
A Matter of Record (301) 890-4188 415
1 within there, and that's going to increase the risk
2 of some really serious problems.
3 I think the whole issue of abuse, again,
4 it's a vulnerable population. We already heard
5 that we can get a yield of 23 percent pretty
6 simply. It won't be very long before we get a
7 yield of 60, 70, 80 percent. And I can guarantee
8 you, with the opioid crisis that we have in this
9 country, this will make headlines for people dying
10 from buprenorphine toxicity when that happens, if
11 it's a question of when.
12 DR. ACRI: I'm Jane Acri. I voted no,
13 partially because of reasons of basic pharmacology.
14 We don't know what the mechanism of action for the
15 antidepressant effect is. We don't know if it's a
16 very low level of mu agonism, if it's kappa
17 antagonism, if it's NOP agonism. We don't know,
18 and I think that really needs to be delineated.
19 I'm concerned about the active metabolite and its
20 possible accumulation and the likelihood of causing
21 physical dependence and withdrawal. I just don't
22 think that's all been adequately characterized. I
A Matter of Record (301) 890-4188 416
1 think a lot more could have been done to answer
2 those questions.
3 I also noticed -- and we never discussed
4 this -- that both compounds are metabolized by the
5 cytochrome P4503A4 and a lot of other drugs are
6 metabolized by those CYPs. And the possibility of
7 drug-drug interactions was never really discussed
8 and I didn't see any data to address it.
9 Then finally, I think it's a very vulnerable
10 population, a population that's highly at risk for
11 drug dependence in the first place, so I think it's
12 been a very risky proposition to introduce an
13 opiate in that type of population if there are
14 other alternatives.
15 DR. DE WIT: I voted yes. I thought, from
16 the point of view of abuse liability testing, the
17 applicant had done all the tests that we normally
18 do for abuse liability, both in humans and in
19 animals. And in particular, the test was done at
20 the highest possible risk population, that is,
21 opiate abusers, so they typically don't do the
22 abuse liability in the target clinical population,
A Matter of Record (301) 890-4188 417
1 but rather in a population that's known to abuse
2 that category of drugs, so they're looking at
3 diversion there. And to my mind, I saw no signal
4 for any abuse liability based on the data threshold
5 were presented.
6 On the other hand, I did wonder a little bit
7 about drug interactions. We didn't hear very much
8 about using the drug in combination with something
9 like alcohol, which would be very widely used by
10 depressed people. And I also didn't hear very much
11 about whether the formulation was tamper proof.
12 But in general, I thought they did a good job with
13 the risk assessment.
14 DR. JENSEN: Roxanne Jensen. I voted yes.
15 I think, overall, the s applicant did a good job
16 laying out the averse event, but I do think that
17 some of the comments made before we got here are
18 important, especially about what happens in
19 real-world situations, where things aren't as
20 controlled and monitored. Thank you.
21 DR. FLYNN: Kathryn Flynn. I voted yes,
22 primarily for reasons that Dr. de Wit just stated
A Matter of Record (301) 890-4188 418
1 very eloquently and I won't repeat.
2 DR. MARSHALL: Brandon Marshall. I voted
3 yes, primarily because I was convinced that
4 samidorphan is doing what it was intended to do,
5 that it reduces significantly the euphoric effects,
6 and significantly reduces the risk of abuse
7 potential, and that withdrawal was infrequent, and
8 at present, mild, and could be medically managed.
9 That said, I'd like the caveat that I would
10 like to see more data with polysubstance use,
11 particularly where we see increased risk of
12 overdose with buprenorphine, alcohol, and also
13 prescribed a non-medical use of benzodiazepines as
14 well.
15 DR. CRAWFORD: Stephanie Crawford. I voted
16 yes. In terms of the question asked, has the
17 applicant adequately characterized the safety
18 profile in treating MDD, I thought they did. That
19 stated, though, I also agree with Dr. de Wit's
20 statements. Without repeating those, if the drug
21 product were approved, however, we know that NDAs
22 do not typically have long-term data, but
A Matter of Record (301) 890-4188 419
1 postmarketing studies would be looking at certain
2 issues that have been stated by several around the
3 table.
4 I was also struck by something in
5 Dr. Sullivan's presentation, though it was only
6 based on observational studies, retrospective
7 cohort studies that longer-term greater than
8 90 days, opioid therapy at certain levels may
9 increase the risk of depression. So it's a
10 question mark, but still I thought the applicant
11 adequately characterized the safety based on the
12 indicated uses. I would have had more concerns
13 about any REMS and potential misuse in that safety
14 because the issue of the precipitated opioid
15 withdrawal scared me.
16 DR. RILEY: Bill Riley at the NIH. I have
17 to say these are tough calls, I think, on both
18 efficacy and safety. I voted yes here. I
19 understand the real-world concerns about issues
20 regarding comorbidity, and drug interactions, and
21 that sort of thing that I think need to be
22 considered.
A Matter of Record (301) 890-4188 420
1 I was also struck after listening to the
2 public comments and remembered that we already have
3 a safety issue in the real world, which is people
4 are prescribing buprenorphine for treatment-
5 resistant depression now, without this antagonist
6 for the mu opioid that I think is actually a really
7 thoughtful way to think about how to reduce the
8 safety risks of that use.
9 So I applaud the sponsors for making the
10 attempt to do that work, and I think that's the way
11 to go forward with this.
12 DR. CELLA: David Cella, Northwestern. I
13 voted yes. I don't really have anything to add to
14 the comments about the sponsor doing a presumably
15 good job of short-term due diligence, and I was
16 comforted by the obvious need and plan for a REMS
17 program if this is approved.
18 DR. NARENDRAN: Just to summarize, I heard a
19 wide range of opinions and the sponsor did a really
20 good job in characterizing the compound in, again,
21 the defined time periods, and it didn't have
22 euphoria, it didn't have abuse liability, and
A Matter of Record (301) 890-4188 421
1 didn't cause withdrawal.
2 Then I also heard that people were concerned
3 about the fact that they may have some
4 [indiscernible] and mu opioid receptors and it has
5 euphoria. It has withdrawal. I did hear
6 consistently that people were concerned about the
7 separation issue of who voted no. People were
8 concerned about the external speakers' highlighting
9 of depression, pain, being a high vulnerable
10 population for substance use disorders, which could
11 be a concern, and drug-to-drug interactions is
12 obviously a concern with other opioids, sedatives,
13 benzodiazepines. So that led people to vote no.
14 Hopefully that summarizes it, and we'll move
15 to question number 3, which is a discussion
16 question, but you've discussed everything possible
17 that you can already, but we'll move to this
18 question.
19 Related to the potential risks associated
20 with the use, misuse, and abuse of buprenorphine/
21 samidorphan in postmarketing settings, we'll start
22 with A, discuss any concerns you have about the
A Matter of Record (301) 890-4188 422
1 risks of misuse, abuse, addiction, or overdose with
2 buprenorphine/samidorphan in the intended patient
3 population or in others who may have access to the
4 drug.
5 Anybody who has comments; we'll start with
6 Dr. Besco.
7 DR. BESCO: Kelly Besco. I know we've
8 talked a lot about the patients themselves who will
9 be taking these medications, but I'd be remiss if I
10 didn't remind everyone that a large percentage of
11 our opioid addictions begin with family members
12 diverting their family members' unused medications
13 or medications perhaps they're taking from their
14 medicine cabinets.
15 I worry that if a patient would have an
16 unsuccessful trial of this medication, it could
17 remain in the home and certainly be accessed for
18 potential misuse and abuse.
19 DR. NARENDRAN: Anybody else? Comments?
20 Dr. Ruha?
21 DR. RUHA: Michelle Ruha. I just wanted to
22 comment that, clinically, I treat people who
A Matter of Record (301) 890-4188 423
1 overdose on opioids every day, and I just want to
2 remind everybody, I don't mean to completely
3 downplay it. Buprenorphine is an important opioid
4 also. But it is the safest opioid out there, and
5 this dose is extremely low.
6 So this is 2 milligrams we're talking about,
7 and I realize people could stack pills and have a
8 big cumulative dose, but it's a very small dose of
9 buprenorphine. And I admit, adults with opioid
10 overdose every day and I've never admitted a
11 buprenorphine because it does tend to have a
12 ceiling effect with respiratory depression.
13 I have admitted children who have gotten
14 into them, so I'm not saying it's risk free, but I
15 don't want us to overestimate the danger of people
16 overdosing on 2-milligram pills of buprenorphine
17 and the risk of it either.
18 DR. NARENDRAN: Dr. Besco?
19 DR. BESCO: Kelly Besco. I definitely
20 appreciate that comment and agree with you from a
21 pharmacokinetic standpoint and looking at the
22 kinetics of the medication. But I think we need to
A Matter of Record (301) 890-4188 424
1 remember, too, that access could be a gateway to
2 something down the line that would be more potent.
3 DR. NARENDRAN: If there's no other
4 questions, we'll move to -- Dr. Jain.
5 DR. JAIN: Sorry. No problem. So Felipe
6 Jain. One concern that I have pertains to some of
7 the studies that Dr. Sullivan referenced in
8 post-traumatic stress disorder, which is highly
9 comorbid with MDD, in which opiate treatment over
10 the longer term can be associated with avoidance
11 and can be seen to inhibit recovery.
12 That's partly where some of my concern about
13 doing a longer-term controlled trial in the pre-
14 marketing window stems from. I think that an
15 opiate for MDD is different from an atypical for
16 MDD. It's different than an SSRI for MDD because
17 of the known abuse potential for opiates as well as
18 this signal from PTSD.
19 DR. NARENDRAN: Any other comments?
20 (No response.)
21 DR. NARENDRAN: We'll move to the next
22 subquestion. Discuss any concerns you have more
A Matter of Record (301) 890-4188 425
1 generally about approving a product containing
2 buprenorphine and opioid for the first time for the
3 treatment of depression. Dr. Joniak-Grant?
4 MS. JONIAK-GRANT: Elizabeth Joniak-Grant.
5 Not a huge thing that I'm like, oh gosh, this is
6 the most horrible thing ever, but something that
7 keeps kind of popping into my mind is that with so
8 many of these patients also having pain, would
9 getting an opioid over time increase their pain
10 issues?
11 All the latest data suggests that taking
12 opiates long term increases pain and so how would
13 that play out with this population?
14 DR. NARENDRAN: Any other comments?
15 (No response.)
16 DR. NARENDRAN: We'll move on to the next
17 question. Discuss whether you are concerned about
18 the risk of opioid overdose if other opioids are
19 used, misused, or abused concurrently with
20 buprenorphine/samidorphan, perhaps in higher doses
21 intended to overcome buprenorphine/samidorphan's mu
22 antagonistic effects, with potential additive
A Matter of Record (301) 890-4188 426
1 opioid agonist effects that have not been fully
2 characterized. Dr. Meisel?
3 DR. MEISEL: Steve Meisel from Fairview. We
4 talked about this somewhat in one of the previous
5 questions. But there is no doubt that if you think
6 of a hospital setting, an ambulance setting or
7 something, and somebody's on this and they broke
8 their leg, or dislocated their hip, or whatever it
9 may be, and you got this narcotic antagonist on
10 board, that's going to lead to all sorts of pushing
11 of all sorts of doses, and then the narcotic
12 antagonist wears off before the narcotic does. And
13 then what?
14 Somebody gets into the operating room, and
15 they don't realize they're on whatever the brand
16 name of this thing is. They don't realize that
17 there's a narcotic antagonist that goes along with
18 that. And what's the anesthesiologist going to do
19 and how is that going to affect the care inside the
20 operating room?
21 When people are at home, I think the same
22 sorts of things happen, but in a much more subtle
A Matter of Record (301) 890-4188 427
1 way. So you go to the dentist, and now, you get
2 your prescription for Vicodin or whatever it may
3 be, but you're on this, and the dentist doesn't
4 know anything about this stuff, and they just think
5 you're on an antidepressant. Then that doesn't
6 work and then what does the person do?
7 Again, there are all sorts of unintended
8 consequences when you approve a drug like this.
9 When Contrave was approved, nobody thought about
10 this stuff, and we're now seeing all sorts of
11 problems related to that because people don't
12 realize there's naltrexone in Contrave, and nobody
13 knows how to deal with it. There are no
14 guidelines. Nobody's been able to dream up
15 guidelines, and nobody's dreamed up or even thought
16 through the guidelines with this, and it's going to
17 create all sorts of unintended consequences.
18 DR. NARENDRAN: Dr. de Wit?
19 DR. DE WIT: I'm not so worried about the
20 overdoses with combination with buprenorphine and
21 another opioid. It's a partial agonist, so it's
22 going to be its own limiting factor. If you add
A Matter of Record (301) 890-4188 428
1 another agonist, then that's going to block its
2 effects, plus the samidorphan is an antagonist, so
3 it just doesn't seem like you're going to overcome
4 the effects of this combination with another
5 opiate.
6 DR. NARENDRAN: Anybody else? Dr. Meisel
7 again?
8 DR. MEISEL: Just to respond to that, the
9 real-life situation of that is that they got the
10 narcotic antagonist in their body, and now they've
11 got acute pain. Now, you're trying to treat that
12 acute pain and you can't overcome it because of the
13 narcotic antagonist. I think, to me, that's my
14 real concern.
15 DR. NARENDRAN: Thanks for that
16 clarification. We'll move to the next
17 subcomponent. What risk reduction strategies could
18 be implemented to decrease risk associated with
19 buprenorphine/samidorphan use? Dr. Dunn?
20 DR. DUNN: I think, with every
21 pharmacological device that we use in medicine, if
22 we can treat the patient or provide benefit, we can
A Matter of Record (301) 890-4188 429
1 definitely do harm. And I think the question here
2 is, what is the harm long-term.
3 Again, as I mentioned before, I think the
4 company did a pretty good job in terms of
5 short-term safety data, but in terms of long-term
6 safety data, that's still kind of up in the air,
7 and then hence, the postmarketing recommendations
8 were for long-term follow-up.
9 I'm assuming that if this were to get
10 approved, the label would be for acute treatment of
11 a major depressive episode. But as my colleague,
12 Dr. Jain, mentioned, for these patients, they're
13 going to be on it longer for than 6 weeks. They're
14 going to be on it for longer than 12 weeks,
15 especially if it works. They're going to be on
16 this thing potentially for a lifetime, and the
17 safety aspect of that is a little bit unclear.
18 As I mentioned before, a more robust REMS
19 strategy would, I think, placate some of those
20 concerns, limiting to 30-day supplies, frequent
21 U-toxes, or perhaps even before the postmarketing
22 results came out, perhaps you could only limit it
A Matter of Record (301) 890-4188 430
1 to a 3-month or 6-month treatment, that patients
2 who were treated on this, if they don't respond in
3 that time frame, that's the one chance that you
4 got.
5 But if you responded, perhaps that's a
6 discussion that you and your clinician can have
7 about maintaining this long term. But I think that
8 clinicians need to also take responsibility in
9 terms of adequately educating the patient about the
10 risks and benefits.
11 I think the tricky part about this
12 medication is that the potential risks, if they do
13 arise, they're subtle. All right? They're
14 long-term risks. There's not something that the
15 patient is going to acutely notice. And I think
16 those types of risks are difficult to convey to the
17 patient because our brains are built to look at
18 kind of acute effects.
19 The long-term effects, long-term
20 consequences, they're harder to see. That's why we
21 have opiate epidemic. That's why we have
22 alcoholism, because those long-term effects are
A Matter of Record (301) 890-4188 431
1 harder for us to see.
2 So I think, in the beginning, a very robust
3 restrictive REMS strategy, and some of the
4 postmarketing results came out showing that the
5 risks were not as great as we feared, perhaps
6 easing those restrictions.
7 DR. NARENDRAN: Dr. Meisel?
8 DR. MEISEL: Steve Meisel. So I do want to
9 point out I've been on a number of these committees
10 relating to opioids, and I can tell you that almost
11 none of the REMS strategies are effective for
12 opioids because most of them rely upon education,
13 flyers, those kinds of things. And that's all I
14 heard today, is proposals for nice flyers, and
15 education, and websites, and that sort of stuff.
16 And we're deluding ourselves if we think that's
17 going to reduce any risk. So that's one thing we
18 should not do, replicate the failures with the
19 opioids.
20 I like your idea about your tox screens. I
21 think that might be helpful. Just like clozapine,
22 what we do with that. I think that is helpful
A Matter of Record (301) 890-4188 432
1 there. I think it would be helpful if before a
2 drug like this were to be approved, if the
3 manufacturer were forced to have a dosage form that
4 would reduce the likelihood of diversion so that
5 you couldn't separate the products from each other
6 and inject it. I know there's a lot of work going
7 on in that space in the opioid for pain world, some
8 successful partially. A lot of it is unsuccessful,
9 but that's a strategy to reduce the risk of
10 diversion.
11 In the space of what we talked about earlier
12 with patients who come in with acute pain needs and
13 nobody realizes that they're on a narcotic
14 antagonist that goes along with that, how can we
15 leverage computer systems and other forcing tactics
16 that put into people's face that this patient is on
17 a narcotic antagonist. And here is a strategy and
18 here are the tactics that you should be doing,
19 should a patient need acute pain response.
20 So to spell those out ahead of time, work
21 with our electronic health record vendors to put
22 that front and center should a patient come in on
A Matter of Record (301) 890-4188 433
1 one of these products, those things can help
2 mitigate some of the risks.
3 DR. NARENDRAN: Dr. Besco?
4 DR. BESCO: I was going to say mainly the
5 same thing, but just not as eloquent as that. But
6 no, I just will concur that I believe that the
7 concurrent proposed REMS strategies are very
8 passive and would like to see something that's more
9 active, that follows the patient throughout the
10 course of their therapy rather than just at the
11 initial, is this an appropriate use case or not.
12 DR. NARENDRAN: I'd like to reiterate that,
13 too. I think the REMS strategy is inadequate. It
14 should be at least as restrictive as what we do for
15 DATA 2000, I think. And more in line with that,
16 education, keeping track of the providers, knowing
17 how many scripts are being written, and maybe your
18 end talks could be very beneficial, at least if you
19 decide to go forward. Ms. Witczak?
20 MS. WITCZAK: Kim Witczak. Again, I know
21 this is a totally different department at the FDA,
22 but I think the communication, the outward
A Matter of Record (301) 890-4188 434
1 communication to the public is really important as
2 part of this. It might be something different
3 because I can hear the commercials right now. "Are
4 you still depressed? Your antidepressant isn't
5 working?"
6 You can see that being part of it, and I
7 don't know if that's regulatory or if it needs
8 congressional, but I think that piece of it needs
9 to be more than just a pamphlet that's given out at
10 the doctors. I think that's really something to be
11 considered.
12 DR. NARENDRAN: Thank you. We'll move to
13 question number 4, which is a voting question. Do
14 the available data support a favorable benefit-risk
15 profile of buprenorphine/samidorphan to support
16 approval?
17 Same directions; press your button on the
18 microphone that corresponds to your vote,
19 20 seconds to vote, you can always change as far as
20 it's flashing.
21 (Pause.)
22 DR. NARENDRAN: We're missing one person.
A Matter of Record (301) 890-4188 435
1 Please press the button again, everyone. Press it
2 hard so it registers.
3 (Voting.)
4 MS. BHATT: The voting results; yes 2; no
5 21; abstain zero; no voting zero.
6 DR. NARENDRAN: Do we want to go around?
7 Very quickly, state your name, your vote, and why
8 you voted. We'll start from this end of the table.
9 DR. CELLA: David Cella, Northwestern
10 University. I voted for the first question yes, as
11 I said, very reluctantly, very small effect, and
12 with the safety data, and the concerns raised
13 through the day, it just didn't put me over the top
14 in terms of risk-benefit.
15 DR. RILEY: Bill Riley, NIH. I think if the
16 safety were essentially nil, you could maybe take a
17 flyer on the minimal or the edgy aspects of the
18 benefit ratio, but they're more than nil, so no.
19 DR. CRAWFORD: I'm sorry. I was pressing
20 the yes button instead of the mic. Stephanie
21 Crawford. Having voted no in the first question
22 about substantial evidence of the efficacy, I have
A Matter of Record (301) 890-4188 436
1 to vote no on the risk-benefit.
2 DR. MARSHALL: Brandon Marshall. I voted
3 no. I just did not see evidence of benefit, maybe
4 marginal benefit, and that just did not outweigh
5 the safety concerns for me.
6 DR. FLYNN: Kathryn Flynn. I voted no. I
7 think there's clearly a lot of evidence presented
8 today that looks like, in certain situations, this
9 could work, and there is very clearly unmet need,
10 but overall, especially considering 202 as
11 exploratory or proof of concept, I should say, I
12 voted no on the ratio overall.
13 DR. JENSEN: Roxanne Jensen. I voted no for
14 efficacy, so I felt that, overall, I couldn't vote
15 yes for this.
16 DR. DE WIT: I also voted no because of the
17 efficacy issues.
18 DR. ACRI: I voted no even though I voted
19 yes for efficacy. The effect was small and,
20 obviously, there were a lot of considerations about
21 the magnitude of the benefit. And I felt that the
22 risks were equal to the benefit at least, that the
A Matter of Record (301) 890-4188 437
1 benefits did not outweigh the risks.
2 DR. MEISEL: Steve Meisel. I voted no for
3 the reasons I've already described.
4 DR. GRIFFIN: Marie Griffin. I voted no
5 because of the risk-benefit.
6 DR. BESCO: Kelly Besco. I also voted no
7 for reasons that I've already shared.
8 MS. JONIAK-GRANT: Elizabeth Joniak-Grant.
9 I voted no. I'm not convinced of efficacy and I'm
10 not wholly convinced of the safety.
11 MS. WITCZAK: Kim Witczak. I voted no. I
12 don't think we can approve something if it doesn't
13 work. And I also think, because it has the opioid,
14 we almost have to create an even higher standard
15 for approvals.
16 DR. FIEDOROWICZ: This is Jess Fiedorowicz.
17 I voted no for reasons I've already described.
18 DR. NARENDRAN: Raj Narendran. I voted no
19 for reasons described already.
20 DR. JAIN: Felipe Jain. I voted no for
21 previously described reasons.
22 DR. DUNN: Walter Dunn. I voted yes because
A Matter of Record (301) 890-4188 438
1 I voted yes on the first two questions, but also
2 the caveat that I would use this medication very
3 carefully and in probably a much more of a
4 restrictive population than what was looked at by
5 the sponsor, so perhaps for patients who have
6 failed for five different types of treatments and
7 maybe several other augmentation strategies. Even
8 then, I perhaps would only use it for a limited
9 amount of time.
10 DR. IYENGAR: Satish Iyengar. I also voted
11 no because of the efficacy issues. I had voted no
12 on the first question and also after hearing all
13 the concerns about the risks.
14 DR. HERNANDEZ-DIAZ: Sonia Hernandez-Diaz.
15 I voted no because the potential benefit doesn't
16 seem to outweigh the potential risks.
17 DR. HABEL: I'm Laurie Habel. I voted no.
18 I voted no on the first two questions, so that
19 seemed appropriate.
20 DR. KULLDORFF: I'm Martin Kulldorff, no.
21 DR. WARHOLAK: I'm Terri Warholak, and I
22 voted no for reasons already stated.
A Matter of Record (301) 890-4188 439
1 DR. RUHA: Michelle Ruha. I voted yes even
2 though I voted no on the substantial-evidence
3 question, and that was because I did think there
4 was evidence. It was more the FDA's definition of
5 substantial evidence, I felt I needed to vote no
6 the first time.
7 But I did feel like there was evidence. I
8 was actually surprised and moved a bit by the graph
9 that compared the results to the aripiprazole and
10 other atypical antipsychotics like quetiapine. It
11 looked pretty similar in efficacy there, and I was
12 moved by some of the efficacy results. And I voted
13 yes on safety.
14 I would support much stricter REMS. I agree
15 with that, possibly even the DATA ex-waiver like we
16 talked about, although I know that's for opioid
17 treatment. And I would restrict the patient
18 population to really resistant depression.
19 DR. NARENDRAN: Okay. Just to summarize, I
20 heard a lot of people voted no, mostly based on the
21 efficacy. There was a few people who also had
22 concerns about safety. The people who voted yes
A Matter of Record (301) 890-4188 440
1 also felt it was reasonable, but with a stronger
2 REMS and more restrictive prescription status.
3 Question number 5 is a discussion question.
4 What, if any, additional data are needed pre- or
5 post-approval to address outstanding issues with
6 buprenorphine samidorphan? Please be clear whether
7 you believe these data should be required prior to
8 approval. We'll start whoever's ready to go to
9 provide the discussion. Dr. Crawford?
10 DR. CRAWFORD: Thank you. I don't have any
11 additional suggestions about data, but I think I
12 would like to underscore that several of us would
13 feel better if it were a more delineated REMS
14 versus what was given within the briefing documents
15 and the applicant's description.
16 DR. NARENDRAN: Dr. Cella?
17 DR. CELLA: I would just like to suggest
18 much more clarity about responders and percent of
19 remitting patients to get to the bottom of is there
20 really this subset of patients that has a
21 tremendous benefit. I hate to see that opportunity
22 lost in those patients, but it just didn't come
A Matter of Record (301) 890-4188 441
1 through in the presentation today, and it might be
2 there based upon some of the public comment.
3 DR. NARENDRAN: Dr. Joniak-Grant?
4 MS. JONIAK-GRANT: Elizabeth Joniak-Grant.
5 In addition to some of the comments I've already
6 made about data that I'd like to see, I think it
7 would be good to see data about the super
8 responders. There's been a lot of reference to
9 them. One of the comments was about them. But
10 there's no data to say, well, are there two of
11 them? Are there 50 of them? And what do we mean
12 by a super response? So I think that data would be
13 really useful.
14 DR. NARENDRAN: Dr. Jain?
15 DR. JAIN: Yes. So I echo a call for more
16 data on response and remission. I also think that
17 it would be quite beneficial to do a trial in
18 depressed patients with comorbid pain since it's
19 the population in which prescribers are going to be
20 really quite drawn to use this. And I don't think
21 we can ignore that premarketing and not have any
22 data on that premarketing.
A Matter of Record (301) 890-4188 442
1 It would also be helpful to have greater
2 data on an individual level withdrawal symptoms and
3 greater characterization of what that is, not only
4 from a statistical perspective, but from a
5 narrative perspective in patients who are
6 experiencing symptoms such as these, particularly
7 given the FDA's analyses on the percentages that
8 showed higher side effects within the BUP/SAM group
9 than the placebo.
10 DR. NARENDRAN: Dr. Dunn?
11 DR. DUNN: Given the observation that
12 perhaps some of the mechanism is potentially purely
13 at the mu opioid receptor, I would like to see what
14 the FDA had mentioned before, long-term efficacy
15 data on patients who are on this for perhaps about
16 a year, with a concern that perhaps they're going
17 to develop tolerance to it and then lose that
18 antidepressant effect.
19 Then also along that line, if you were able
20 to show that this was truly acting at the kappa
21 opioid receptor, so perhaps a combination of
22 buprenorphine plus naltrexone still retained the
A Matter of Record (301) 890-4188 443
1 antidepressant effect, I think that would
2 contribute a lot to the safety concerns.
3 DR. NARENDRAN: Dr. Hernandez-Diaz?
4 DR. HERNANDEZ-DIAZ: Sonia Hernandez-Diaz
5 from the Harvard Chan School of Public Health. I
6 was going to say the longer-term effectiveness, I
7 think we have heard that comment many times. I was
8 going to also say that now that the outcomes have
9 been discussed and the dose is more clear, having
10 another efficacy trial focusing on more patient-
11 centered outcomes that are relevant for patients
12 and clinically meaningful. And potentially, there
13 might be studies to identify a group a priori and
14 have a study that identifies that group, and have
15 this stage 2 randomization rather than for
16 non-placebo responders, for null responders being
17 kind of dropped, and focus on a population. Like,
18 you could replicate that study in clinical
19 practice.
20 Also, in other comments, we have been trying
21 to compare informally with other adjuvant therapies
22 that in clinical practice, you will not give
A Matter of Record (301) 890-4188 444
1 placebo, or you will not stay in the same dose in
2 these patients. You will increase the dose. You
3 will add other psychotropics.
4 So I wonder if the comparison group is more
5 appropriate, that has the equipoise, not placebo,
6 but maybe considering other studies that you will
7 follow in clinical practice with these patients,
8 just a comment.
9 DR. NARENDRAN: Dr. Kulldorff?
10 DR. KULLDORFF: Martin Kulldorff, Harvard
11 Medical School. If 1.5 reduction in the scale
12 between drug and placebo is clinically important,
13 then I would like to see a randomized trial that's
14 powered to detect a difference of 1.5, and I think
15 that should be done prior to approval.
16 DR. NARENDRAN: Thank you. I would like to
17 add my comments in terms of I think it would be
18 good to see a study with in vivo pharmacologies
19 clearly defined at 2 and 2. I mean, there are
20 really good mu opioid receptor radiotracers.
21 They're really good kappa opioid receptor
22 radiotracers that could provide that kind of in
A Matter of Record (301) 890-4188 445
1 vivo pharmacology data. I think one more trial of
2 everything we have learned, perhaps SPCD, longer
3 duration, averaging their outcome measure, and to
4 prove substantial effectiveness would probably be
5 very welcome, I think. That's my last comment.
6 Anybody else have any suggestions?
7 DR. ACRI: I'm sorry. You passed me by.
8 I'm Jane Acri from NIDA. And I guess the one thing
9 that I would like to see is an actual study of
10 whether or not you produce withdrawal with
11 discontinuation because I feel like withdrawal
12 signs are one of the reasons that people keep
13 taking opioids.
14 So if I ran out of my BUP/SAM, would I just
15 take another? Would I take an oxy? Would I take
16 something to alleviate withdrawal and then get
17 hooked on another opioid?
18 So the ability to create a physical
19 withdrawal is really important in terms of
20 maintaining use and possibly transition to other
21 opioids. But I would also like to see more basic
22 pharmacology, and as you were saying, some work to
A Matter of Record (301) 890-4188 446
1 determine the mechanism of action, to look at kappa
2 opioid receptors as well as mu opioid receptors and
3 try to get to the root of what's happening in the
4 long term.
5 DR. NARENDRAN: I think, if there's no
6 further comments, I think we can close the session.
7 Thank you, everybody.
8 DR. MATHIS: Thank you for your service. We
9 appreciate it.
10 Adjournment
11 DR. NARENDRAN: Thank you. We'll meet again
12 tomorrow.
13 (Whereupon, at 5:01 p.m., the meeting was
14 adjourned.)
15
16
17
18
19
20
21
22
A Matter of Record (301) 890-4188