FOOD AND DRUG ADMINISTRATION

















                          JOINT MEETING OF THE




                                AND THE














                       Monday, September 13, 2004


                               8:00 a.m.




                          Holiday Inn Bethsda

                         8120 Wisconsin Avenue

                           Bethesda, Maryland








        Wayne K. Goodman, M.D., Chair

        Jean E. Bronstein, R.N., M.S.

          (Consumer Representative)

         James J. McGough, M.D.

         Philip S. Wang, M.D., M.P.H., Dr.P.H.

         Lauren Marangell, M.D.

         Dilip J. Mehta, M.D., Ph.D.

           (Industry Representative)

         Delbert G. Robinson, M.D.

         Daniel S. Pine, M.D.

         Barbara G. Wells, Pharm.D.

         Bruce G. Pollock, M.D., Ph.D.




         P. Joan Chesney, M.D., Chair

         Deborah L. Dokken, M.P.A.

         Michael E. Fant, M.D., Ph.D.

         Richard L. Gorman, M.D.

         Robert M. Nelson, M.D., Ph.D.

         Thomas B. Newman, M.D., M.P.H.

         Judith R. O'Fallon, Ph.D.

         Victor M. Santana, M.D.




         Norman Fost, M.D., M.P.H.

         Charles E. Irwin, Mr., M.D.

         Laurel K. Leslie, M.D., F.A.A.P.

         Steven Ebert, Pharm.D. (Consumer Representative)

         James M. Perrin, M.D.

         Cynthia R. Pfeffer, M.D.

         Gail W. Griffith

           (Patient Representative, Voting)

         Robert D. Gibbons, Ph.D.

         Tana A. Grady-Weliky, M.D.

         Richard P. Malone, M.D.

         Irene E. Ortiz, M.D.

         Matthew V. Rudorfer, M.D.



                        PARTICIPANTS (Continued)




         Kelly Posner, Ph.D.

         John March, M.D., M.P.H.

         Samuel Maldonado, M.D., M.P.H.

           (Industry Representative

         Barbara Stanley, Ph.D.

         Madelyn Gould, Ph.D., M.P.H.




         Robert Temple, M.D.

         Russell G. Katz, M.D.

         Thomas Laughren, M.D.

         M. Dianne Murphy, M.D.

         Anne Trontell, M.D., M.P.H

         Anuja M. Patel, M.P.H., Executive Secretary



                            C O N T E N T S


      Call to Order and Opening Remarks,

         Wayne Goodman, M.D.                                     6


      Introduction of Committee                                  9


      Conflict of Interest Statement,

         Anuja Patel, M.P.H.                                    20


      Overview of Issues:

         Dianne Murphy, M.D., Director, Office of

           Pediatric Therapeutics,

           Office of the Commissioner                           25


         Russell Katz, M.D., Director, Division of

           Neuropharmacological Drug Products,

             DNDP, CDER                                         34


      Regulatory History and Background,

         Thomas Laughren, M.D., Team Leader, DNDP, CDER         46


      Recent Observational Studies of Antidepressants

        and Suicidal Behavior,

         Diane Wysowski, Ph.D., Division of Drug Risk

           Evaluation, Office of Drug Safety, CDER              62


      Brief Report on TADS Trial,

         John March, M.D., M.P.H., Duke University              74


      Committe Discussion on TADS Trial                         93


      Characteristics of Pediatric Antidepressant Trials,

         Greg Dubitsky, M.D., Medical Officer,

           DNDP, CDER                                          107


      Classification of Suicidality Events,

         Kelly Posner, Ph.D., Columbia University              117


      OCTAP Appraisal of Columbia Classification


         Solomon Iyasu, M.D., M.P.H., Team Leader,

           Office of Counter-Terrorism and Pediatric

             Drug Development                                  140



                      C O N T E N T S (Continued)


      Results of the Analysis of Suicidality in Pediatric

        Trials of Newer Antidepressants,

         Tarek Hammad, M.D., Ph.D., M.Sc., M.S., Senior

           Medical Reviewer, DNDP, CDER                        152


      Comparison Between Original ODS and DNDP Analyses

        of Pediatric Suicidality Data Sets,

         Andrew Mosholder, M.D., M.P.H., Division of

           Drug Risk Evaluation, ODS, CDER                     200


      Citalopram and Escitalopram Product Safety Data,

         Jeffrey Jonas, M.D., Forest Laboratories, Inc.        219


      Sertraline Use in Pediatric Population: A

      Risk/Benefit Discussion,

        Steven J. Romano, M.D., Pfizer, Inc.                   232


      Wyeth Pharmaceuticals, Joseph S. Camardo, M.D.           247


      Open Public Hearing                                      255


      Summary by the Committee Chair                           444




                         P R O C E E D I N G S


                   Call to Order and Opening Remarks


                DR. GOODMAN:  I wish to welcome you to


      this two-day joint session of the


      Psychopharmacologic Drugs Advisory Committee and


      the Pediatric Advisory Committee, being held on


      September 13th and 14th here, at the Holiday Inn in


      Bethesda, Maryland.


                I am Wayne Goodman, Professor of


      Psychiatry at the University of Florida, today


      wearing my hat as chair of the advisory committee.


      As you settle in, please take this opportunity to


      put into silent mode your cell phones and any other


      devices that emit sounds in the audible range of


      human beings.


                Some of you may be surprised not to see


      Matt Rudorfer in this seat but we arm-wrestled for


      the position and he won.




                In all seriousness, his term has ended but


      we are fortunate to see him return as a voting


      consultant to the committee.




                I have some official language to read to


      you.  All committee members and consultants have


      been provided with copies of the background


      materials, from both the sponsors and the FDA, and


      with copies of letters from the public that we


      received by the August 23rd deadline.  The


      background materials have been posted on the FDA


      website.  Copies of all these materials are


      available for viewing at the FDA desk outside this




                We have a very large table, a full house


      and important topic today so I would like to start


      with a few rules of order.  Please speak directly


      into the mike when called on.  We will be keeping


      track of individuals at the table who wish to speak


      and we will call upon them in order.


                FDA relies on the advisory committee to


      provide the best possible scientific advice


      available to assist us in the discussion of complex


      topics.  We understand that issues raised during


      the meeting may well lead to conversations over


      breaks or during lunch.  However, one of the




      benefits of an advisory committee meeting is that


      the discussions take place in an open and public


      forum.  To that end, we request that members of the


      committee not engage in off-record conversations on


      today's topic during the breaks and lunch.


                Whenever there is an important topic to be


      discussed there are a variety of opinions.  One of


      our goals today and tomorrow is for the meeting to


      be conducted in a fair and open way where every


      participant is listened to carefully and treated


      with dignity, courtesy and respect.  Anyone whose


      behavior is disruptive to the meeting will be asked


      to leave.  We are confident that everyone here is


      sensitive to these issues so understand that these


      comments are as a gentle reminder.


                We look forward to a productive and


      interesting meeting.  This is an unusual meeting in


      that we have two advisory committees represented


      here, Psychopharmacological Drugs Advisory


      Committee, chaired by myself, and the Pediatric


      Advisory Committee, chaired by Joan Chesney, to my


      left.  We will now go around the table and have the




      committee introduce themselves, starting on my


      right.  Please indicate your expertise and


      affiliation.  We will start in that corner, over






                DR. TEMPLE:  Bob Temple.  I am the Office


      Director, ODE I.


                DR. KATZ:  Russ Katz, Division Director,


      Division of Neuropharmacological Drug Products,




                DR. LAUGHREN:  Tom Laughren, phychopharm.


      team leader, in the Neuropharmacological Division.


                DR. MURPHY:  Dianne Murphy, Office


      Director, Office of Pediatric Therapeutics.


                DR. TRONTELL:  Anne Trontell, Deputy


      Director, Office of Drug Safety.


                DR. FANT:  I am Michael Fant, University


      of Texas Health Science Center in Houston.  My


      expertise is neonatology and biochemistry.


                DR. PFEFFER:  Cynthia Pfeffer.  I am a


      child psychiatrist at Weill Medical College of


      Cornell University, and I have expertise in




      depression suicidal behavior in children and




                DR. FOST:  Norm Fost, University of


      Wisconsin, Professor of Pediatrics, Director of the


      Bioethics Program and Chair of the IRB.


                DR. ORTIZ:  Irene Ortiz, University of New


      Mexico, Albuquerque VA.  My expertise is in


      depression in the elderly.


                DR. MALONE:  Richard Malone, Drexel


      University College of Medicine, and my area is


      child psychiatry.


                DR. NELSON:  Robert Nelson, Children's


      Hospital of Philadelphia and the University of


      Pennsylvania.  My expertise is in pediatric


      critical care medicine and ethics.


                DR. PERRIN:  Jim Perrin, Professor of


      Pediatrics, Harvard Medical School and Head of the


      Division of General Pediatrics at the Mass. General


      Hospital.  I have shortened my expertise as being


      in general pediatrics.


                DR. GRADY-WELIKY:  Tana Grady-Weliky,


      Associate Professor of Psychiatry at the University




      of Rochester School of Medicine and Dentistry.  My


      expertise is in mood disorders and women across the


      reproductive life cycle and medical education.


                DR. EBERT:  Steven Ebert, Department of


      Pharmacy of Meriter Hospital and School of


      Pharmacy, University of Wisconsin, Madison.


                DR. GIBBONS:  Robert Gibbons, Professor of


      Statistics and Professor of Psychiatry and Director


      of the Center for Health Statistics at the


      University of Illinois, Chicago.  I only do math!


                DR. PINE:  Danny Pine, child and


      adolescent psychiatrist, National Institute of


      Mental Health intramural research program.  I am a


      clinical child psychiatrist.


                MS. BRONSTEIN:  Jean Bronstein,


      psychiatric nurse, Stanford University Hospital,


      the consumer representative.


                DR. RUDORFER:  Matthew Rudorfer, National


      Institute of Mental Health.  My areas of expertise


      are mood disorders and psychopharmacology.


                MS. PATEL:  Anuja Patel, Advisors and


      Consultants Staff, Executive Secretary for the




      Psychopharmacologic Drugs Advisory Committee.


                DR. CHESNEY:  Joan Chesney, the University


      of Tennessee, in Memphis, and Professor of


      Pediatrics, and my specialty is infectious




                DR. MCGOUGH:  Jim McGough, Professor of


      Psychiatry, UCLA.  My area is child and adolescent




                MS. GRIFFITH:  My name is Gail Griffith


      and I serve as the patient rep. on this committee,


      and I would just like to take this opportunity to


      say why I am here.  First, I am not a medical


      professional; I am a consumer.  I have suffered


      from major depression since I was a teen.  Second,


      I have a son who suffers from major depression and


      three years ago, at age 17, after he was diagnosed


      and placed on a regimen of antidepressants he


      attempted suicide by overdosing intentionally on


      all his medications.  He nearly died.  So, I know


      this illness.  I know what it does to adolescents.


                For the record, I would simply like to


      state that I have no professional ties to any




      advocacy group or any patient constituency.  I also


      wish to affirm that I have no ties to any


      pharmaceutical company, nor do I hold any


      investments in pharmaceutical manufacturers.  My


      sole responsibility is to ensure that the interests


      of concerned parents and families are represented


      at this meeting.


                DR. MARANGELL:  Lauren Marangell, Baylor


      College of Medicine.  I specialize in adult


      interventions in mood disorders, both unipolar and




                DR. ROBINSON:  I am Delbert Robinson.  I


      am from the Albert Einstein College of Medicine, in


      New York, and I specialize in psychotic disorders


      and anxiety disorders.


                DR. LESLIE:  Laurel Leslie.  I am a


      behavioral developmental pediatrician at Children's


      Hospital, San Diego and my area of expertise is in


      children's mental health services research.


                DR. IRWIN:  Charles Irwin.  I am a


      professor of pediatrics at the University of


      California, San Francisco.  I am in charge of the




      Division of Adolescent Medicine at the University,


      which is a multi-disciplinary program that cares


      for adolescents and trains large numbers of


      individuals caring for teenagers, and my research


      is in the area of risk-taking during adolescence.


                MS. DOKKEN:  I am Deborah Dokken.  I


      reside in the Washington, D.C. Metro area.  I do


      not have a specific institutional affiliation, and


      I have for several years been involved in parent


      and family advocacy and health care.


                DR. NEWMAN:  I am Thomas Newman.  I am a


      professor of epidemiology and biostatistics in


      pediatrics at the University of California, San


      Francisco, and a general pediatrician.


                DR. WELLS:  I am Barbara Wells.  I am a


      professor and Dean of the School of Pharmacy at the


      University of Mississippi.  My expertise is in


      psychiatric pharmacotherapy.


                DR. POLLOCK:  I am Bruce Pollock.  I am a


      professor of psychiatry, pharmacology and


      pharmaceutical sciences at the University of


      Pittsburgh.  I head the Division of Geriatric




      Psychiatry at the university.


                DR. O'FALLON:  Judith O'Fallon, Emeritus


      Professor of Biostatistics from the Mayo Clinic,


      with 30 years of experience particularly in cancer


      clinical trials but clinical trials methods.


                DR. SANTANA:  Good morning.  I am Victor


      Santana.  I am a pediatric hematologist/oncologist


      at St. Jude's Children's Research Hospital in


      Memphis, Tennessee.


                DR. WANG:  I am Philip Wang, Harvard


      Medical School.  I am a psychiatrist and


      epidemiologist and those are my areas of expertise.


                DR. GORMAN:  Richard Gorman, a practicing


      pediatrician for 20 years in the Baltimore suburbs,


      Chair of the American Academy's Committee on Drugs,


      and representing the American Academy of Pediatrics


      at this table.


                DR. MALDONADO:  Sam Maldonado.  I work at


      pediatric drug development at Johnson & Johnson.  I


      am one of the industry representatives to this




                DR. MEHTA:  Dilip Mehta, retired industry




      executive and industry representative on the


      Psychopharmacologic Drugs Advisory Committee.


                DR. GOODMAN:  Thank you, all, for being


      with us these two days.  Our session today is the


      second of two planned advisory committee meetings,


      convened to address recent concerns about reports


      of suicidal ideation and behavior developing in


      some children and adolescents during treatment of


      depression with a selective serotonin reuptake


      inhibitor, an SSRI, or other newer generation


      antidepressants.  Our goal is to gather information


      from a variety of sources and perspectives to help


      us understand this complex situation, and


      ultimately to offer the best possible


      recommendations to the FDA.


                I would like to thank the many groups,


      individuals and families that submitted written


      statements in advance of the meeting, many of which


      were quite informative as well as moving.  A major


      portion of today's meeting will be devoted to a


      four-hour open public hearing during which dozens


      of people from around, and even beyond, the country




      will have the opportunity to present their own


      personal or professional experiences and ideas


      about the relative risks and benefits of


      antidepressant medication in children and


      adolescents.  Although the necessary consideration


      of the clock will permit only a short time at the


      microphone for each speaker, I can assure you that


      the committee welcomes and values input from all


      viewpoints and feels it is essential to our work


      that all voices be heard.


                The committee's task is more difficult


      than usual.  Our review is not confined to whether


      one agent is safe and effective based upon the


      corresponding clinical trials submitted to the FDA.


      We are faced, instead, with assessing efficacy and


      safety for nine drugs that represent more than one


      chemical class of antidepressants, all of which are


      already available on the market.


                Although the cornerstone of the data under


      examination is derived from randomized clinical


      trials submitted to the FDA this time, following a


      reclassification of the adverse events, we find




      ourselves turning to information from a wide


      variety of sources, in particular to inform


      ourselves about the drugs' possible benefits in


      this population.  However, once we open our minds


      to consideration of data originating outside


      randomized clinical trials we rest upon a slippery


      slope in which variations in interpretation are


      introduced according to the weighting each member


      places on the merits of the source.


                For me, the difficulty in assessing the


      balance between benefit and risk is multiplied by


      the nature of the adverse events under scrutiny.


      Psychiatrists grapple, for the most part, with


      illnesses that produce significant morbidity and


      more rarely mortality except from suicide.  Nothing


      in my experience is more tragic than the loss of a


      child to suicide.  To think that I might prescribe


      an agent that contributed to that outcome is


      unbearable.  Equally unbearable is to think that I


      did not do enough to prevent it.  This is the


      essence of the dilemma before us.


                We may not have all the data we would




      like, especially to assess long-term benefit.  We


      can make recommendations about what research should


      be conducted, but we will be faced at the


      conclusion of business tomorrow to make


      recommendations based upon what we know at this


      cross-section in time.  In deliberating on the


      safety of antidepressant treatment in children, let


      us not forget the toxicity of the underlying


      disease.  Major depression remains an


      under-diagnosed, under-studied and under-treated


      serious disorder among many thousands of our


      nation's youth, leading to considerable suffering,


      disability and heartbreak in many families.


                I believe that all of us in this room


      share the desire to alleviate the suffering from


      this disorder through the successful use of


      interventions that are made available to all those


      who need them.  Despite the daunting task before


      us, I remain hopeful that with a fair and


      open-minded review of the evidence this advisory


      committee will constructively address the issues


      and ensure that interventions for this serious




      disorder meet high standards for both effectiveness


      and safety.


                Now I will ask Anuja Patel, executive


      secretary for the advisory committee, to review


      some of the ground rules for this committee and the


      public hearing.


                     Conflict of Interest Statement


                MS. PATEL:  Good morning.  Before I


      continue, I would like to notify you of a


      correction on the roster attached to the agenda.


      The following consultants, Dr. Robert Gibbons, Dr.


      Matthew Rudorfer, Dr. Richard Malone, Dr. Tana


      Grady-Weliky and Dr. Irene Ortiz will be added to


      the roster.  Amended copies of the roster will be


      available later this morning at the information


      desk outside this ballroom.


                As you know, we have a very full open


      public hearing today, and in the interest of both


      fairness and efficiency we are running it by some


      strict rules.  To make transitions between speakers


      more efficient, all speakers will be using the


      microphone and podium in front of the audience. 




      Each speaker has been given their number in the


      order of presentations and when the person ahead of


      you is speaking, we ask that you move to the nearby


      next speaker chair.  Individual presenters and


      families have been allotted three minutes for their


      presentations.  The one consolidated presentation


      has been given five minutes.  We will be using a


      timer and speakers who run over their time will


      find that the microphone is no longer working.  We


      apologize for the need for the strict rules, but we


      wanted to be fair and to give as many people as


      possible an opportunity to participate.


                The public may submit comments after this


      meeting directly to the FDA's Division of Dockets


      Management.  Instructions for submitting electronic


      and written statements are available at the


      registration desk outside this room.  The docket


      will remain open until July 29, 2005.  Thank you


      for your cooperation.


                I would like to read the meeting statement


      into the record now.  The following announcement


      addresses the issue of conflict of interest and is




      made a part of the record to preclude even the


      appearance of such at this meeting.  The topics to


      be discussed today are issues of broad


      applicability.  Unlike issues before a committee in


      which a particular company's product is discussed,


      issues of broader applicability involve many


      industrial sponsors and products.


                All special government employees and


      invited guests have been screened for their


      financial interests as they may apply to the


      general topics at hand.  The Food and Drug


      Administration has granted particular matters of


      general applicability waivers under 18 USC


      208(b)(3) to the following special government


      employees, which permits them to participate fully


      in today's discussion and vote:  Jean Bronstein,


      Dr. Joan Chesney, Dr. Wayne Goodman, Dr. Lauren


      Marangell, Dr. James McGough, Dr. James Perrin, Dr.


      Bruce Pollock.  In addition, Dr. Philip Wang has


      been granted a limited waiver that permits him to


      participate in the committee's discussions.  He is,


      however, excluded from voting.




                A copy of the waiver statements may be


      obtained by submitting a written request to the


      agency's Freedom of Infection Office, Room 12A-30


      of the Parklawn Building.


                In addition, Dr. Judith O'Fallon and Dr.


      Victor Santana have financial interest under 5 CFR,


      Part II, Sec. 40.202 that are covered by a


      regulatory waiver under 18 USC 208(b)(2).


                Because general topics impact so many


      entities, it is not practical to recite all


      potential conflicts of interest as they may apply


      to each member, consultant and guest speaker.  FDA


      acknowledges that there may be potential conflicts


      of interest but, because of the general nature of


      the discussion before the committee, these


      potential conflicts are mitigated.


                With respect to FDA's invited industry


      representatives, we would like to disclose that Dr.


      Dilip Mehta and Dr. Samuel Maldonado are


      participating in this meeting as industry


      representatives, acting on behalf of regulated


      industry.  Dr. Mehta is retired from Pfizer and Dr.




      Maldonado is employed by Johnson & Johnson.


                With respect to all other participants, we


      ask in the interest of fairness that they address


      any current or previous financial involvement with


      any firm whose product they may wish to comment


      upon.  Thank you.


                DR. GOODMAN:  We will now proceed with a


      series of formal presentations that will bring us


      to 11:45 a.m. and then a 15-minute discussion


      before lunch.  In the interest of time, I would


      like to ask my fellow committee members to restrict


      their questions after each presentation to issues


      of clarification only.  There will be time, 15


      minutes, for some discussion between 11:45 and


      12:00 and tomorrow there will be a great deal of


      time for discussion and consideration of the


      questions before us.  So, please, if you have


      questions about clarification, you can ask them


      after each presentation but restrict it to those


      kinds of issues.


                With that, I would like to introduce Dr.


      Dianne Murphy, of the FDA, who will be followed by




      Dr. Russell Katz, also of the FDA.


                           Overview of Issues


                DR. MURPHY:  Good morning and welcome to


      this very important discussion.  Before we begin


      today's important deliberations, I would ask us to


      step back and see the broader context in which this


      meeting is occurring.  I am going to spend a few


      minutes trying to describe that for you.


                There are four points I hope you take from


      this short presentation.  One is that the majority


      of medicines given to children in this country are


      prescribed off-label and have not been studied in


      all the pediatric populations in which they are




                Second, because of new legislation and


      regulations since 1998, FDA has seen an increase in


      products that are used in children being studied in




                Third, for the first 100 products,


      involving over 200 studies conducted as a result of


      the new legislation, FDA has found that


      approximately one-fourth of the time there was a




      need to change the dose, a new pediatric-specific


      adverse event was described or the product was not


      found to be efficacious despite the fact that it


      was efficacious in adults.


                Fourth, part of the reason we are here


      today is because we are finally studying the


      therapies that are being given to children.


      Children deserve the same level of evidence that is


      required for adults to determine that their use by


      them is safe, effective and properly dosed.  They


      are a heterogeneous group who undergo rapid


      metabolic, hormonal, physiologic, development and


      growth changes in comparison to us, adults, who are


      rather static and tend only to deteriorate.


                Over the last two decades FDA has actively


      supported, along with the American Academy of


      Pediatrics and many other groups, the efforts to


      encourage development of information and


      appropriate use of therapies in the pediatric




                Very quickly, and this is important to


      understand, again, the context in which some of




      this information has been brought to you, in the


      last decade we have made tremendous progress.  In


      1994 the agency published an approach that it hoped


      would help foster and encourage development of


      therapies that we be used in children.  Congress


      passed legislation in 1997 which is referred to as


      the exclusivity or the incentive to develop studies


      on products that are being used in children.


                In 1998 the FDA published the Pediatric


      Rule, which was an effort to say that if a sponsor


      is going to develop a product in adults and that


      same disease occurs in children, or condition, that


      product in most circumstances and certain


      conditions would be required to be studied.


                We are going to go more into the 2001


      adoption by FDA of Subpart D, Pediatric Ethics


      Regulations, and I wanted to bring up the Best


      Pharmaceuticals for Children Act, which you will


      hear referred to as BPCA, because it renewed the


      congressional legislation of '97 and is important


      in that, again, it is the renewal of the incentive


      to study products that are being used in children.




                Another congressional legislative


      activity, the Pediatric Research Equity Act, in


      essence confirmed FDA's authority to require


      studies in children in certain circumstances.


                In the last decade, particularly really


      since 1997, the FDA has issued over 290 written


      requests to sponsors asking them to study products


      in children because these products are being used


      in children.  We have had submitted to us over 110


      products, involving over 220 studies in children,


      and have now more than 76 new labels that have new


      pediatric information from these studies.


                The major depressive disorders were


      included in the written requests that were issued,


      and written requests were issued for the products


      you see listed here, Prozac, Zoloft, Remeron, Paxil


      Effexor, Celexa and Serzone.  Those studies were


      all conducted under this program or in response to


      this program.


                This is a list of some of the programs and


      activities that are in place at FDA to help ensure


      the quality and ethical conduct of studies and the




      approach to pediatrics.  This is really focusing


      mostly on the drugs component of this.  But there


      is, at the Commissioner's level, an Office of


      Pediatric Therapeutics.  This was enacted by the


      Best Pharmaceuticals for Children Act in 2002 and


      first staff were hired last year.  We now have in


      place an ethicist whose focus is pediatric ethics.


      You will hear a little bit more about the Pediatric


      Advisory Committee and Subpart D referrals in a


      minute.  You just heard about the exclusivity


      process which has been important in making sure


      that trials do get conducted.  I will spend a few


      moments at the end talking about disclosure


      requirements that are unique to pediatric studies


      that are conducted under exclusivity.


                This is another meeting, actually in a


      long series of meetings that have occurred to


      ensure the scientific and ethical quality of


      activities involving studies that are being


      conducted in children.  Since 1999, the Pediatric


      Advisory Subcommittee has had, including today's


      meeting, over ten meetings that have addressed over




      ten scientific issues, three ethical issues and, in


      addition, starting last year, began having specific


      safety reviews of products that have been approved,


      again under the exclusivity provisions, so that all


      adverse events that occurred in the year after


      product was granted exclusivity were reviewed.


      Again, this is just to inform you of the ongoing


      pediatric activities that are occurring at the FDA,


      some of them.


                The new advisory committee, I should say


      full Pediatric Advisory Committee is meeting for


      the first time today.  It was  chartered this year


      and is mandated to include patient and family


      organizations, and its mandated responsibilities


      include safety, labeling disputes and Subpart D


      referrals and general pediatric issues.


                The first Subpart D ethics panel met this


      past Friday and will report to this committee on


      Wednesday.  I will tell you a little bit more about




                It is important to understand that Subpart


      D, which is part of the Common Rule, was those




      extra protections for children applied to only


      federally funded activities until recently.  In


      2000, the Children's Health Act required FDA


      regulated products to be in compliance with


      additional protections for children that are


      embodied in the Subpart D of the Common Rule.


                Subpart D is fundamentally a referral


      process.  There is much more to it but it is a


      process for IRBs when they are unsure they can


      approve or under which regulation they should


      approve a study involving children.  The Pediatric


      Ethics Subcommittee reports to the Pediatric


      Advisory Committee and this is a public process.


                The disclosure of the studies that are


      conducted in children is distinct for studies that


      are conducted in children under the exclusivity


      provisions of BPCA.  I mention this because it is


      unusual in the FDA if a product is not approved


      that those studies would be disclosed.  However, we


      now have, again under BPCA, a requirement that


      within 180 days of submission of a pediatric study


      a public summary of the medical and clinical




      pharmacology reviews will be posted.  There are now


      41 pediatric summaries posted at this website.


      Basically, you can go to the FDA page and get there


      by going to the Center for Drugs or pediatric


      summary\summary review.


                The summaries of Effexor, Paxil, Serzone,


      Celexa, Zoloft and Remeron are available on the


      pediatric summary review site.  As you know, and


      will hear, Prozac is the only antidepressant that


      is approved for use in children, and it is posted


      up on FDA's site for approved applications.  That


      URL is provided for you here and in your handouts.


                The new pediatric data has taught us that


      our knowledge of pediatric therapeutics is in its


      infancy; that we must study children if we are to


      understand pediatric-specific adverse events and


      reactions or if a product is going to work in


      children.  The pharmacokinetics in children has


      proven to be more variable than anticipated.  The


      submitted studies that we are receiving are


      teaching us that we need to know more about


      pediatric endpoints, pediatric trial designs and




      how to conduct these trials, and that we will need


      to change some of our trials as we move forward.


      Ethical issues require reassessment from a


      pediatric perspective.  Therefore, at this point no


      longer shall each child be an experiment of one in


      which not much knowledge is gained.


                As we move forward, it will require our


      careful attention if we are to discover why


      children are behaving differently.  If children are


      to be appropriately treated, we will need to know


      more than how to correct those things or describe


      adverse events.  We are going to need answers to


      such fundamental questions as to why children react


      differently, what are the metabolic, physiologic


      events that are occurring that necessitate


      different dosing, or why is there a therapy that


      works in adults and does not work in children.  Our


      public policy must be more knowledge to replace our


      ignorance.  Thank you, and we look forward to your




                DR. MARANGELL:  Dr. Murphy, a quick


      question, when the FDA requests a study can you




      specify methodology and assessments that you would


      like to see included?


                DR. MURPHY:  When FDA requests a study we


      do put in that written request the trial design,


      the number of patients, the adverse events--you


      know, under exclusivity all of that does go into


      the written request.


                DR. GOODMAN:  Thank you, Dr. Murphy.  Now


      Dr. Katz?


                           Overview of Issues


                DR. KATZ:  Thank you, Dr. Goodman, and


      good morning.  I would like to welcome you to this


      joint meeting of the Psychopharmacologic Drugs


      Advisory Committee and the Pediatric Advisory




                As you know, we are here to present to you


      and to ask for your guidance in interpreting the


      results of our analyses of the relationship between


      antidepressant drug use and suicidal behavior in


      controlled trials in pediatric patients.  This


      meeting is in follow-up to the meeting of these two


      committees held in February of this year.  At that




      meeting, as you recall, we presented to you and


      obtained your endorsement of our plans to perform


      these analyses.


                At this point I would like to very briefly


      recap how we arrived to this point.  As you know,


      we first became aware of a possible relationship


      between antidepressant drug use and suicidal


      behavior in pediatric patients in May of 2003 when,


      in response to our request for further


      clarification of their data, GlaxoSmithKline, the


      manufacturer of Paxil, submitted data to us that


      suggested such a link for that drug.  As a result


      of this submission, the agency issued a public


      statement recommending that this drug not be used


      in pediatric patients with depression, and


      independently we asked all other manufacturers of


      antidepressant drugs to resubmit the relevant data


      from controlled studies with their drugs in


      pediatric patients.


                Based on our review of this data, we


      issued a statement informing prescribers that there


      was a potential relationship between all of these




      drugs and suicidal behavior, and that these drugs


      should be used with caution in these patients.


                However, at that time we also noticed that


      the data submitted to us from the various companies


      was not reported to us in a form that would permit


      definitive analyses.  Specifically, each company


      classified various behaviors as being


      suicide-related adverse events in their own


      idiosyncratic manner.  This led to questions about


      whether or not these events were, in fact,


      suicide-related and, in addition, prevented


      meaningful comparisons between drugs in this class.


                For this reason, we decided that an


      independent assessment of these possible events by


      experts in suicidology would be the most


      appropriate way to definitively answer the question


      of whether or not any, all or none of these drugs


      increased the risk of suicidal behavior in


      pediatric patients.  Let me just add that by


      definitive analyses I mean analyses that make the


      best possible use of the available data.


                It was at this time that we brought the




      issue before you.  At that meeting we presented you


      with our plans to submit blinded narrative


      descriptions of possible suicide-related events to


      a group of independent experts, to be coordinated


      by Columbia University, whose task it would be to


      classify these events as being suicide-related or


      not.  Although no formal vote was taken, this


      committee fully endorsed this effort and agreed


      that the data in hand at that time did not permit


      definitive analyses to be done.


                The committee also recommended, based in


      part on the data in hand but also, I believe


      importantly, on the basis of testimony from members


      of the public who had suffered the tragedy of loved


      ones who had committed suicide while taking these


      drugs, that the agency should ask sponsors of these


      products to warn prescribers that patients being


      treated with these drugs, especially at the


      beginning of treatment, should be closely watched


      for the emergence of signs and symptoms that might


      suggest a worsening in their clinical state.


                Since that February meeting a number of




      important things have happened.  Based on your


      advice, the agency drafted, and all of the sponsors


      of these drugs have adopted, language in product


      labeling warning about the possibility of


      significant behavioral changes at the outset of


      treatment with these drugs in both pediatric and


      adult patients, and the prescribing community and


      the public have been informed of these changes.


                Critically, this warning made clear that


      the possibility of worsening and a possible


      increased risk of suicidal behavior at the outset


      of treatment could not necessarily be attributed to


      the drugs because the data did not permit such a


      definitive conclusion.  Nonetheless, it was


      considered appropriate and prudent to inform


      prescribers and patients and their families that


      changes in behavior could occur with the onset of




                Also, the Columbia group has completed


      their task of reclassifying the potential cases of


      suicidal behavior and, importantly, we have


      completed our reanalyses of these data.  As




      promised at our February meeting, we are now ready


      to present to you the results of these analyses.


                At this point I would just like to give


      you a brief overview of the agenda for today's and


      tomorrow's session.  First Dr. Tom Laughren, of the


      Neuropharmacology Division, will provide you with a


      more detailed account of the regulatory history and


      events that have brought us here this morning.  He


      will be followed by Dr. Diane Wysowski, of the


      agency's Office of Drug Safety, who will briefly


      present the results of some recently published


      epidemiologic studies relevant to this question.


      We have provided the committees with copies of


      these published materials.  Although, of course, we


      consider our reanalyses of the controlled data to


      be the primary source of data on which your


      discussions and recommendations will be based, we


      thought it important to present at least briefly


      the available relevant epidemiologic data.


                Next, Dr. John March, of Duke University,


      will present a brief report of the Treatment for


      Adolescent Depression Study, or TADS trial, a




      recently completed trial that evaluated the effects


      of fluoxetine in adolescents with depression.  As


      you know, fluoxetine is the only drug approved in


      the United States for the treatment of depression


      in pediatric patients and, as you will see, these


      data make an important contribution to our overall


      assessment of the problem before us.


                Dr. Greg Dubitsky, again of the


      Neuropharmacology Division, will then present an


      overview of the design of the pediatric trials from


      which the data for our analyses were derived.  This


      exploration is important because similarities and


      differences in design elements among these trials


      can have important implications for whether or not


      these data can be examined as a whole, or whether


      they must be considered separately.


                Then, Dr. Kelly Posner, of Columbia


      University and the primary person responsible for


      coordinating the blinded reclassification effort,


      will present to you the methodology her group used


      to produce what we now consider to be the


      definitive data on which we have based our






                Because the reclassification of these


      clinical events was the critical activity on which


      all subsequent analyses and decisions will have


      been based, and because by its nature it involved


      subjective assessments of the primary data, we felt


      strongly that it was appropriate to ensure that the


      methodology used by the Columbia group could


      reliably and reproducibly yield similar results


      when applied by an independent group.  For this


      reason, agency scientists performed such an


      independent reclassification of a percentage of


      these cases, utilizing the Columbia classification


      schema, and Dr. Solomon Iyasu, of the agency's


      pediatric group, will present the results of this


      independent audit of the Columbia process.


                At that point, Dr. Tarek Hammad, of the


      neuropharmacology group, will then present the most


      critical results of his extensive reanalyses of the


      data as reclassified by the group of outside


      experts.  These analyses look at the data for


      individual drugs, as well as across all drugs, and




      will provide the data on which the committee


      subsequent discussions will be based.


                Finally, the last formal agency


      presentation will be given by Dr. Andrew Mosholder,


      of the Office of Drug Safety.  Dr. Mosholder's name


      is undoubtedly familiar to you.  Dr. Mosholder was


      the agency reviewer who had, prior to the February


      advisory committee meeting, performed analyses on


      the cases as submitted, that is, the


      non-reclassified cases, and had concluded that


      these drugs did, in fact, increase the risk of


      suicide-related behaviors in this population.  As


      you know, Dr. Mosholder did not present his


      conclusions at the February meeting, although the


      data on which his analyses were based were


      presented and we noted at that time that some in


      the agency had already reached a definitive


      conclusion on this question.


                There has been since that meeting


      considerable public discussion and controversy


      related to the fact that Dr. Mosholder was not


      given the opportunity to present his conclusions at




      that meeting.  The reasons for our decision at that


      time were straightforward.  As I have discussed


      today and as we have discussed publicly on numerous


      occasions, we had decided that at the time of the


      February meeting the data had not been submitted in


      a form in which we could reliably agree that the


      events described as representing suicide-related


      behavior did, in fact, represent such behavior.


      We, therefore, felt that conclusions reached on the


      basis of analyses that relied on these descriptions


      could no, in turn, be considered completely




                We felt, and still feel, that presenting


      conclusions based on potentially unreliable


      analyses could have led to errors in either


      direction, that is, resulted in a conclusion that


      the drugs were dangerous when they really were not,


      or resulted in a conclusion that the drugs were


      safe when they were not.  A mistake of either kind


      could have, in our view, disastrous consequences.


      For this reason it was, and remains, our view that


      these decisions must be based on the most reliable




      analyses possible.  Now that the definitive


      analyses have been done, however, Dr. Mosholder


      will present his own analyses and conclusions, with


      particular attention to a comparison between his


      results and Dr. Hammad's.


                Following lunch we will hear brief


      comments from several of the pharmaceutical


      companies who have antidepressant drug products on


      the market, and the day will end with the open


      public hearing.  A total of 73 members of the


      public have signed up to make statements.  As you


      have heard and as in the February meeting, we will


      again need to limit the statements from the public,


      this time to three minutes per individual.  We


      recognize that this is not much time and we


      apologize for the limit but it would be impossible


      for all those who wish to make statements to do so


      without imposing this limitation.  We appreciate


      your understanding on this point and, as you have


      heard, anyone who wishes may submit written


      testimony to the docket.


                Tomorrow the committee will discuss the




      data you will have heard.  We, of course, look


      forward to this discussion and in particular to


      your answers to the questions we have brought to


      you and which we have provided in your background


      packages.  We are, in brief, interested in your


      views on our approach to the reclassification


      effort and, critically, whether you believe that


      the analyses establish that one or more of the


      drugs studied increases the risk of suicidality in


      pediatric patients.  Importantly, if you do


      conclude that there is a signal for suicidality,


      whether for one or more of these drugs, we need to


      know what additional regulatory action, if any, you


      believe should be taken.


                The results of our reanalyses are complex


      and their interpretation is not immediately


      obvious.  They raise difficult questions, not only


      about the fundamental meaning of the results of the


      analyses for each drug, but also about the


      comparability of the various treatments and,


      therefore, whether it is appropriate to consider


      the drugs as a class for which any conclusion




      reached should globally apply or whether the drugs


      must be considered individually.  Further, the


      question of any further regulatory action is also a


      thorny one and must take into account the


      consideration of the lack of available


      effectiveness data for all of the drugs, except


      fluoxetine, although the absence of this


      effectiveness data is not easily interpreted




                Because of the complex nature of the


      evidence and because of the extraordinary


      importance for the public health of the decisions


      that we need to make, we are turning to you, the


      experts, for guidance on these matters.  We thank


      you in advance for your efforts.


                DR. GOODMAN:  Thank you, Dr. Katz.  I


      would like to invite Dr. Tom Laughren to come to


      the podium.


                   Regulatory History and Background


                DR. LAUGHREN:  Thank you, and I would also


      like to welcome everyone to the meeting today.  I


      am going to begin by briefly giving an overview of




      events leading up to today's meeting.  I am then


      going to talk about the key elements in the


      division's exploration and analysis of the


      pediatrics suicidality data.  I will then spend a


      little time talking about our March 22nd public


      health advisory and the subsequent labeling changes


      that have now been implemented.  Then I am going to


      spend a little time talking about the effectiveness


      data.  I did this at the last meeting; I will do


      this again because I think it is important to have


      these data in mind since they are an important part


      of the context of this discussion about pediatric


      suicidality.  Then I am going to quickly go over


      the questions and the issues for which we are going


      to be seeking feedback tomorrow.  I think it is


      important that you have these questions in mind as


      you hear the talks this morning.


                This slide lists a number of the people at


      FDA who have been involved in looking at these


      data.  As you can see, these people come from


      various sections of the agency.  It is a long list,


      and really the point of this slide is that we take




      this matter very seriously and we have invested a


      lot of effort into trying to understand these data.


                You heard earlier about the two laws,


      FDAMA and BPCA, that give FDA authority to grant


      additional market exclusivity for companies which


      do pediatric studies.  The point of this slide is


      that most of the data that we are dealing with this


      morning come from these types of studies, in other


      words, studies that were done to obtain additional


      marketing exclusivity.  However, we have also


      included in our analysis data from a ninth


      antidepressant drug, Wellbutrin, that was not


      studied for exclusivity.  That was one study in


      ADHD.  We are also including in our analysis data


      from the TADS trial that you will hear about in


      more detail later in the morning from John March,


      from Duke.


                As Dr. Katz pointed out, this issue first


      came to our attention based on a review of the


      Paxil supplement.  In that review, the reviewer


      noticed that events suggestive of possible


      suicidality were subsumed, along with other




      behavioral events, under the preferred term


      "emotional lability."  This led FDA to issue a


      request to the sponsor, GSK, to explain this coding


      practice.  Ultimately, that resulted in a report to


      FDA, in May of last year, on pediatric suicidality


      with Paxil.  As Dr. Katz pointed out, that report


      did suggest a signal of increased suicidality in


      association with drug use, particularly in one of


      three depression trials in that program.


                What I am going to do in this slide is


      very quickly run through subsequent events that led


      us up to the February advisory committee meeting.


      So, in June of last year we issued a public


      statement cautioning about the use of Paxil in


      pediatric patients with depression.  In July we


      issued requests to sponsors of eight other


      antidepressant products to ask them to give us the


      same kind of summary data that GSK had provided for




                In September of last year we held an


      internal regulatory briefing at FDA.  The purpose


      of this was to brief upper management about this




      signal.  The two points that I took away from that


      meeting from the standpoint of the division's work


      were, number one, there was general agreement that


      it would be important to try and classify these


      events since many of them were not clearly related


      to suicidality and we felt it would be very


      important to do a rational classification.


      Secondly, there was sentiment that we ought to try


      and obtain patient-level data information, beyond


      the summary information, in order to try and


      explain differences among trials and between




                In September and October we began to get


      responses to our July requests.  Also, in October


      we issued requests to sponsors for the


      patient-level data sets that I mentioned earlier.


      Also in October, we decided to go outside of FDA to


      get a classification of these cases accomplished.


      Then, again as Dr. Katz mentioned, in October we


      issued a second public health advisory, this time


      extending the cautionary language to all current


      generation antidepressants.  Finally, in November




      and December, having looked at the responses to the


      July request for summary data, it occurred to us


      that we may not have obtained all of the relevant


      events and so we sent additional requests to have a


      broader search for events that we would then try


      and get classified.


                That brings us up to the February advisory


      committee that we held.  At that meeting you


      advised us to basically continue with our analysis


      of the data but, in the meantime, to go ahead and


      make some labeling changes.  In March of this year


      we issued a public health advisory announcing the


      changes that we had requested.  In the meantime,


      the classification of the cases was ongoing by the


      Columbia group.  Those were completed in June of


      this year.  Then, in August of this year we


      completed our analysis of the pediatric suicidality




                In this slide what I am doing is basically


      summarizing what I think is the major contribution


      of the division to this effort.  Again, we went to


      a lot of effort to try to ensure completeness of




      case findings, that we had a complete set of events


      to have classified.  We then worked with Columbia


      University to have these events classified.


                As an aside, I would like to mention that


      this effort, conducted by Kelly Posner and her


      group at Columbia and the very exceptional group of


      outside experts that they assembled to do this,


      represents a very substantial effort that has not


      only helped us to understand these data but I think


      will have implications for the field in terms of


      developing a standard approach to classifying these


      kinds of data, and also will lead to guidance


      document that, hopefully, will improve


      ascertainment for suicidality, which was a very


      significant problem in these trials.


                Finally, the third effort that we were


      involved in was, again, in obtaining the


      patient-level data sets that allowed us to try and


      explore for confounding and effect modification, in


      essence, to try to explain some of the striking


      differences we were seeing in the signal across


      trials within programs and across programs.




                As mentioned, at the February advisory


      committee you advised us to go ahead and strengthen


      labeling, in particular for monitoring for


      suicidality, while we were completing our analysis.


      We did this and we announced the request that we


      were making in a March 22nd public health advisory.


                The changes in labeling that we requested


      have now all been implemented for the ten drugs of


      interest.  I would add here that our plan is to


      extend the standard language to all


      antidepressants, not just the current generation


      and, in fact, that has already been done for some


      of these drugs.  We are waiting to do it for the


      others until we work out the final standard


      language, which will be based on advice we get from


      you at this meeting.


                What I want to do in this slide is to very


      quickly go over the labeling changes that have been


      implemented now.  This slide focuses on the advice


      for clinicians who are using antidepressants for


      treating any condition really, whether in adult of


      pediatric patients.  So, the advice is as follows,




      first of all, we are asking clinicians to closely


      observe patients who are being treated with


      antidepressants for clinical worsening and for the


      emergence of suicidality, especially at the


      beginning of therapy but also at times of dose




                Secondly, we are asking clinicians to


      consider changing the therapeutic regimen in


      patients whose depression is either persistently


      worse or whose emergent suicidality is severe,


      abrupt in onset, or was not part of the patient's


      presenting symptoms.


                Finally, we are also asking clinicians to


      observe for the emergence of other symptoms as


      well, for example, anxiety, agitation, panic


      attacks, insomnia, irritability, hostility,


      impulsivity, and so forth.  The idea here is that


      there is a belief, not really solidly empirically


      established but a belief that many of these events


      may represent precursors to emerging suicidality.


      So, we are also asking clinicians to be alert to


      these symptoms.




                This slide focuses on advice for families


      and caregivers that also is included in labeling.


      We are asking those folks to also be alert to the


      emergence of these same symptoms and to report


      those symptoms to healthcare providers if they




                Now I want to turn to briefly describing


      the efficacy data for the 15 short-term trials that


      we looked at in our review of these pediatric


      supplements.  I am going to be focusing on primary


      outcomes in those trials.  I also want to spend a


      little time talking about the difficulty in


      interpreting negative findings in this setting and


      again I want to note that although I am not going


      to be talking about the TADS efficacy data, you


      will be hearing about the TADS efficacy data from


      John March a little bit later in the morning.


                This is kind of a busy slide but basically


      each row in this table represents a different


      trial.  Again, there was a total of 15 trials.


      This is color-coded so you can separate the


      different programs.  There were seven programs. 




      Paroxetine had three trials.  The rest all had two


      trials.  The column to look at is the far column


      where I have summarized the results on the primary




                Basically I have characterized the results


      as follows:  Where the p value on drug versus


      placebo on the primary endpoint was less than 0.05,


      I am calling it positive.  As you can see, that


      applies to the two fluoxetine trials and one of the


      citalopram trials.  If the p value fell between


      0.05 and 0.1 I am characterizing it as a trend.


      That applies to one of the sertraline trials and


      one of the nefazodone trials.  If the p value was


      greater than 0.1 on that primary endpoint I am


      characterizing it as negative.  That applies to all


      the remaining trials.  So, basically what you have


      here is three out of 15 trials meeting FDA's


      standard for being positive.


                The other point I want to make on this


      slide is that this represents FDA's view and I


      think it is a reasonable standard, however, it is


      not the only standard.  To illustrate that, I want




      to talk about two published papers, one for study


      329, the paroxetine trial, a paper that was


      published by Keller in 2001.  That paper


      characterized that trial as a positive study, the


      argument being that although it failed on the


      primary endpoint it succeeded on all the secondary


      endpoints.  So, the authors of that paper


      considered it a positive trial and many in the


      community also considered that a positive study.


                Secondly, there was a paper published on


      the two sertraline trials by Wagner et al., in


      2003, that was based on a pooling of the two


      trials.  Individually those trials did not make it


      but if you pooled them you got a significant p


      value.  Again, many in the community view that as


      evidence of effectiveness of sertraline in


      pediatric depression.  This does not meet FDA's


      standard but the point is that different folks have


      different views of the same data.


                Now I want to talk a little bit about the


      problem of interpreting negative findings in this


      setting.  First I want to turn to adult depression




      trials for drugs that we believe work.  Llooking at


      trials that on face should work, about half the


      time those trials fail. If that failure rate can be


      extrapolated to the pediatric population, the


      expectation in two study programs and most of these


      programs were two study programs--the expectation


      is that three out of four times you would fail to


      get two positive studies.  So, perhaps it shouldn't


      have come as such a surprise that many of these


      programs failed.  On the other hand, the fact that


      the overall success rate, again according to FDA's


      standard, is only 20 percent success is clearly a




                Other factors to think about in looking at


      negative trials in this setting is, first of all,


      the history of antidepressant trials in pediatric


      depression.  If you go back to the tricyclic era,


      there were 12 trials comparing tricyclics with


      placebo in this population.  All of them failed.


      There are many interpretations of that.  One, of


      course, is that these drugs simply don't work in


      that population.




                Another might be that there is even


      greater heterogeneity in this population of


      patients captured under the diagnostic criteria for


      major depression than we see in adults.  That would


      work against getting positive trials.


                Another factor to think about is the


      somewhat unusual regulatory context in which these


      studies were done.  Ordinarily, when companies do


      studies they only benefit if they get a positive


      trial.  In this setting they would win in terms of


      getting exclusivity whether the trial succeeded or


      failed.  I don't know whether or not that was a


      factor in the conduct of these trials but it is


      another thing to think about.


                Finally, at the time that we issued


      written requests for these programs we were not


      routinely asking for phase 2 dose-finding studies


      as we are now.  That, again, maybe a factor.  It is


      possible that the dose was not right in some of


      these trials.


                In any case, the bottom line in terms of


      efficacy is that I think there are plausible




      reasons for failure to find efficacy other than the


      obvious one that maybe the drugs don't work.  On


      the other hand, a very important point I believe is


      that even though most of these programs have failed


      to meet FDA's standard for approval, this is not


      the same thing as saying that we have proof that


      the drugs have no benefit.  The drugs may have


      benefit that has simply not yet been demonstrated.


      On the other hand, the failure to demonstrate


      benefit clearly is a concern, especially when we


      have a risk, as we have now seen, of emerging


      suicidality.  So, the burden is clearly on those


      who believe that these drugs do have benefit to


      show that benefit.  Tomorrow I am going to talk a


      little bit about some possible designs for looking


      at longer-term benefits with these drugs.


                Now what I would like to do is quickly


      move through the questions that we are going to be


      asking you to discuss and comment on tomorrow.


      Again, we think it is important that you have these


      in mind as you hear the presentations this morning.


                First of all, we are going to ask you to




      comment on our approach to classifying the possible


      cases of suicidality and our subsequent analyses of


      the resulting data for the now 24 trials--again,


      the additional trial is the TADS trial.


                The question then would be do the


      suicidality data from these trials support the


      conclusion that any or all of these drugs increase


      the risk of suicidality in pediatric patients?  If


      the answer to that question is yes, to which of


      these nine drugs does that risk apply?  In other


      words, is this a class effect of all


      antidepressants?  Does it apply to certain


      subclasses within this broader class or only to


      specific drugs?


                If you believe there is a class risk or a


      risk that applies only to certain drugs, how should


      this information be reflected in the labeling for


      each of these products?  What, if any, additional


      regulatory actions do you think we need to take?


                Finally, again we would like you to


      consider what additional research might be needed


      to further delineate the risks and the benefits of




      these drugs in patients with pediatric depression?


      Thank you very much.


                DR. GOODMAN:  Thank you, Tom.  I imagine


      people have questions but I want to try to catch up


      this morning to make sure that we have time for all


      the presentations.  So, I will ask you to hold your


      questions and I would like to invite the next


      speaker, Dr. Diane Wysowski, who will be looking at


      data from different sources other than clinical




            Recent Observational Studies of Antidepressants


                         and Suicidal Behavior


                DR. WYSOWSKI:  Good morning.  In this


      presentation I will be reviewing recent studies of


      antidepressants and suicidal behavior and briefly


      discuss their methods, results and limitations.


                I reviewed two types of studies,


      ecological and patient-level controlled,


      observational studies.  Ecological studies show


      increasing antidepressant use and simultaneous


      decreasing suicide rates.  However, such


      correlations do not necessarily imply causality. 




      Findings of ecological studies can be merely


      coincidental.  Numerous factors such as changes in


      risk factors, social and economic changes, more


      available counseling, changes in gun access and


      choice of a less lethal means of suicide in


      children and adolescents may coincide with


      decreases in the suicide rate in children and




                Ecological studies don't show which factor


      or factors are responsible for an observed trend.


      Furthermore, an increased relative risk of suicide


      with antidepressants in children and adolescents


      may coexist with a decreased suicide rate.  To


      better examine causality, we turned to


      patient-level controlled studies, such as


      observational studies, in clinical trials.


                For the rest of this presentation I will


      be focusing on two patient-level controlled,


      observational studies.  The first is the Jick study


      that was published this July in The Journal of the


      American Medical Association.  It is a matched case


      control design based on patient prescriptions and




      diagnoses obtained from the United Kingdom's GPRD,


      the General Practice Research Database, for the


      period 1993-1999.  The GPRD is a database of


      medical records from general practitioners of more


      than three million patients in the United Kingdom.


                For this study subjects were 10 through 69


      years of age.  Exposures studied were the most


      widely used antidepressants in the U.K.,


      amitriptyline, fluoxetine, paroxetine and


      dothiepin.  Dothiepin was chosen as the reference


      category.  From data on these antidepressants


      users, the investigators identified 555 cases of


      nonfatal suicidal behavior, defined as ideation or


      attempts.  They identified 17 cases of suicide.


      From the base group of antidepressant users, the


      investigators matched the cases with more than 2000


      controls who did not develop suicidal behavior.


      The researchers then compared the suicidal cases to


      the non-suicidal controls for initiation of each




                Controlling for age, sex, calendar time


      and time from first antidepressant prescription to




      onset of suicidal behavior, the range of relative


      risk for nonfatal suicidal behavior was 0.83 to


      1.29 for the antidepressants compared to dothiepin.


      None of these risks were statistically significant.


      Paroxetine, with a relative risk of 1.29 and a 95


      percent confidence interval of 0.97 to 1.7, had


      borderline statistical significance.


                Similar results were obtained for those


      10-19 years old.  No statistically significant


      association was found between each antidepressant


      and completed suicide.  No statistically


      significant association was found between stopping


      an antidepressant and nonfatal suicidal behavior.


                The relative risk for nonfatal suicidal


      behavior and suicide were highest for patients


      first prescribed an antidepressant within 1-9 days,


      versus 90 days or more, before the suicidal


      behavior of the case in the same time period for


      the control.  For nonfatal suicidal behavior the


      relative risk for antidepressant use within 1-9


      days was 4, with a 95 percent confidence interval


      of 2.89 to 5.74.  For suicide the relative risk for




      antidepressant use within 1-9 days was 38, with a


      wide confidence interval of 6.2 to 231.


                Reviewing the limitations of this study,


      the results are only as good as the GPRD data.


      There are concerns about possible missing data,


      possible incomplete ascertainment and


      misclassification of patients, and possible


      uncontrolled biases among the antidepressant drugs,


      such as selection by severity of depression.  There


      were no interviews of cases and controls so


      medication compliance is not systematically known.


      There was no unexposed group, and the


      antidepressant risks are only in reference to the


      dothiepin group.


                FDA asked Dr. Jick and colleagues to


      reanalyze their results with amitriptyline as the


      reference category.  They kindly responded to our


      request, and asked that their interpretation of the


      results be presented verbatim to the committee.  If


      the committee wishes to see these supplemental


      analyses I will be glad to present them in the


      question and answer period.




                Other limitations of the study include the


      fact that suicidal ideation is a more subjective,


      softer diagnosis than suicide attempts and the


      risks were not examined separately.  This was a


      study of mostly adults and there is limited


      information on children and adolescents.


                Finally, the investigators excluded


      patients with a history of 11 other


      neuropsychiatric diagnoses, calling into question


      the representativeness of the patients compared


      with those in clinical practice.


                Another patient-level controlled study


      examined the relationship between antidepressants


      and the risk of suicide attempt by adolescents with


      major depressive disorder diagnoses.  The study was


      done by investigators at the University of Colorado


      School of Pharmacy and Medicine.  Robert Valuck was


      the principal investigator.  It was presented as a


      poster at the International Society of


      Pharmacoeconomics and Outcomes Research meeting,


      this past May.


                It is a retrospective cohort study of paid




      medical claims data from the PharmMetrics


      Integrated Outcomes Database of 70 managed health


      plans for the period 1995 through March, 2003.


      Paid claims data include health care provided in


      which costs are incurred, such as for


      prescriptions, doctor visits, emergency room visits


      and hospitalizations.


                The investigators identified about 16,000


      adolescents aged 12-18 with the first major


      depressive disorder diagnosis.  They classified


      patients into cohorts by antidepressant


      prescription, those who received none over the


      entire follow-up period, which was the reference


      group, and those who received SSRIs, tricyclic


      antidepressants or other antidepressants within 30


      days of diagnosis.  They followed the cohorts for


      at least 6 months.


                The researchers used a Cox proportional


      hazards regression analysis to control for some 14


      covariates and to examine the multivariate


      relationship between antidepressant use and time to


      suicide attempt.  The majority of patients, 78




      percent, had no antidepressant filled in the 6


      months after diagnosis; 15 percent had SSRIs filled


      within 30 days of diagnosis.  And, 209, 1.3


      percent, of the 16,000 patients made at least one


      suicide attempt in the follow-up period.


                The investigators concluded that


      antidepressant treatment with any class of drugs


      did not increase the risk of suicide attempt.


      Antidepressant use for less than 6 months, compared


      with use for 6 months or more, was associated with


      a statistically significant 3-fold increased risk


      of suicide attempt.  Females, those who received


      psychotherapy within 90 days of major depressive


      diagnosis, patients with substance abuse,


      schizophrenia or another mental health disorder,


      patients with more chronic diseases and those in


      the Midwest and West were independently at greater


      risk of suicide attempt.


                Dr. Valuck and co-investigators recently


      expanded their study to include 24,000 eligible


      patients with a new diagnosed major depressive


      disorder.  This expanded study is currently being




      reviewed for publication.  The researchers added a


      propensity matching adjustment to control for


      predictors of treatment and to achieve greater


      balance among the antidepressant groups.  The


      proportion of suicide attempters, 1.4 percent, was


      about the same proportion as in their smaller




                In this expanded study the hazards ratio


      for SSRIs compared to no treatment was 1.58, not


      statistically significant.  The hazards ratio for


      tricyclics was not estimable due to small numbers


      and it was 1 for the other antidepressant category.


      The hazards ratio for multiple antidepressants was


      1.43, also not statistically significant.  The risk


      of suicide attempt declined with longer use of an


      antidepressant.  Compared with patients having less


      than 8 weeks of use, those with equal to or greater


      than 6 months of use had a statistically


      significant decline in the risk of suicide attempt.


                Concerning the limitations, the results


      are only as good as these paid claims data.  There


      are concerns about possible missing data, possible




      incomplete ascertainment and misclassification of


      patients, and possible uncontrolled selection


      biases by antidepressant group.  There were no


      interviews of patients so we don't have


      systematically collected information on medication


      compliance.  There are no data on the outcomes of


      the attempts.  We don't know how many of the


      attempters died, and there is no information on


      suicides.   Also, there is no information on


      individual antidepressants.  There were differences


      in study results between the poster and the


      expanded study, although this is probably due to


      the larger size of the expanded study.


                In conclusion, although most of the


      results for the individual antidepressants or


      classes of antidepressants were not statistically


      significantly associated with suicidal behavior, I


      do not believe that the Jick and Valuck studies


      completely rule out a possible increased risk of


      suicidal behavior with antidepressant use.  The


      studies reviewed were in agreement in showing that


      the risk of suicidal events occurred statistically




      significantly closer to diagnosis and onset of


      antidepressant treatment.  The studies did not


      provide data about the characteristics of patients


      who did not respond to antidepressants or whose


      illness worsened with them.  The studies had actual


      or potential methodological limitations.


                I conclude that more definitive studies,


      perhaps large randomized, controlled trials of


      sufficient length, are needed concerning the risk


      of suicidal behavior and suicide as related to


      antidepressant use in children and adolescents.


      With so much at stake, children and adolescents,


      their parents and physicians and society in general


      deserve to know which therapies and which


      individuals work best for treatment of depression.


      Thank you.


                DR. GOODMAN:  Thank you very much, Diane.


      My preference would be that you present those


      additional Jick data tomorrow.  I don't think we


      have time for it today.  Would that be an agreement


      by the committee as well?  So, I think we are in


      accord on that, if you could prepare to present




      that data tomorrow to us.  There is one question,


      yes, we will allow that.


                DR. PINE:  I am wondering if you could


      clarify in the Valuck 24,000 patient study, if you


      looked at the association in the less than 8-week


      treatment versus no treatment group.  Did that


      confidence interval exclude 1?  I mean, I saw that


      you gave less than 8 weeks versus prolonged


      treatment but I didn't see an odds ratio for less


      than 8 weeks versus no treatment.


                DR. WYSOWSKI:  I don't think that I have


      those data.  I don't think that they did that


      analysis.  Their results are still being considered


      for publication and we just got an abstract.  You


      saw the poster in your package.  Then, when they


      did the expanded analysis they only gave us an


      abstract of the results.  So, we don't have a lot


      of detail on the expanded study.


                DR. GOODMAN:  Dr. Fost, you had a question




                DR. FOST:  One of the theories of why


      suicide behavior might be increased shortly after




      prescribing is that the patients are at the worst


      then and that is why they are started on


      prescriptions.  If that were true, one might expect


      to see increased suicidal behavior in the week or


      two before prescribing.  Do either of these studies


      allow for that analysis?


                DR. WYSOWSKI:  I believe the Jick study


      did look at that.  Actually, no--no, I don't see


      any information on that; it is just after.


                DR. FOST:  And the data set doesn't allow


      itself for that reanalysis?


                DR. WYSOWSKI:  Well, it may.  I don't know


      whether either investigator has information on




                DR. GOODMAN:  I think that was an


      excellent question.  Now I would like to invite Dr.


      John March, from Duke University, to present


      results from the TADS trial.


                       Brief Report on TADS Trial


                DR. MARCH:  Thanks, it is a pleasure to be


      here and I would like to begin the presentation by


      thanking the committee for inviting the TADS team




      to present the TADS data, and also thank the FDA


      for the comprehensive and thoughtful way that it is


      approaching this question of enormous public health




                The TADS trial, the Treatment for


      Adolescents with Depression Study, is an


      NIMH-funded comparative treatment trial, and I am


      going to present efficacy and safety data from the


      stage-1 outcomes that were published in The Journal


      of American Medical Association several weeks ago.


      We have a detailed safety paper in preparation.  We


      have a methods paper which has been published and a


      baseline paper which looks at the sample


      composition in press, and those of you who are


      interested in the TADS trial are referred to these


      papers for further information.


                As I mentioned, this is a study funded by


      the NIMH, coordinated by the Department of


      Psychiatry at Duke University and the Duke Clinical


      Research Institute, the DCRI.  It has had the


      benefit of oversight and consultation from numerous


      consultants, a scientific advisory board.  The




      NIMH, DSMB participants included 12 sites from


      around the United States.  Lilly provided


      fluoxetine under an independent educational grant


      to Duke University, had no input into the design of


      the study, the conduct of the study, the analysis


      of the data or the preparation of the manuscript.


      My sense is that the major credit for this work, as


      for all of the research on which we base


      evidence-based practice, goes to the children and


      families who are willing to participate in


      research.  Without their participation, we would


      have no evidence at all.


                The overall objective of the TADS trial


      was to examine the effectiveness of medication and


      cognitive behavioral psychotherapy alone and in


      combination for the acute and long-term treatment


      for adolescents with DSM-IV diagnosis of major


      depression.  The design of the trial was a


      balanced, randomized, controlled study that was


      masked by use of independent evaluators; four


      groups, placebo, cognitive behavioral


      psychotherapy, fluoxetine and their combination,




      and the study involved 36 weeks of treatment, of


      which I am going to present the stage-1 data for


      the first 12 weeks of treatment.  We also have a


      year of naturalistic follow-up and we have recently


      been funded to follow these youngsters out into


      young adulthood.  That data will not be discussed




                Now, it is important, in understanding the


      generalizability of the data, to know a little bit


      about the sample composition.  The inclusion


      criteria included outpatients, both boys and girls


      age 12-17, with a DSM-IV diagnosis of major


      depressive disorder and an IQ greater than or equal


      to 80.  Youngsters with severe conduct disorder or


      substance abuse, other than nicotine, pervasive


      developmental disorders, thought disorder, bipolar


      disorder, or history of suicidality or homicidality


      were excluded from the trial.


                Because suicidality is the question at


      issue today, I thought it important to say a little


      bit more about these exclusion criteria, kids were


      excluded if they had a hospitalization within the




      previous 3 months or if they were considered to be


      high risk, which meant a suicidal event of some


      sort within the past 6 months, the presence of


      active intent or plan, or if they had suicidal


      ideation in the context of a family which was so


      disorganized that we felt that even with the


      intensive monitoring in the TADS trial framework


      that it would not be reasonable to enter them into


      a randomized, controlled research study.


                This was a moderate to moderately severely


      ill population.  We had 439 kids, as you can see,


      randomized equally into the 4 groups.  The total


      sample on the Children's Depression Rating Scale


      had a CDRS scale score of 60.  That, again, is a


      mean score of moderate to moderately severely


      depressed, with a range from mild to severe


      depression.  The T score for that mean score is 75.


      That means that these kids were more than 2


      standard deviations out from normal with respect to


      severity of depression.  The sample was multiply


      comorbid, as is characteristic of patients in the


      clinical samples.  This was the first major




      depressive episode for about 90 percent of the


      sample.  Ten percent of the sample had had more


      than one depressive episode.  The mean duration of


      major depression was 42 weeks.  Again, over 50


      percent of the sample was comorbid for another


      mental disorder, both internalizing and


      externalizing disorders; 14 percent of the sample


      had ADHD and half of those kids were on concurrent


      psychostimulant treatment.


                So, unlike the industry-funded trials, the


      TADS sample is largely representative of patients


      who are treated in clinical practice with major


      depressive disorder.  As you might expect, given


      the severity of illness and the pattern of


      comorbidity, these youngsters suffered significant


      functional impairment.  This is the children's CGAS


      rating and you can see that the mean CGAS score was


      between 40-50, so significant functional impairment


      associated with mental illness in this patient




                What did we learn in the trial?  This is a


      take-home efficacy message.  Four groups, again,




      began at a CDRS raw score of 60.  These are random


      regression analyses looking at the adjusted or


      predicted means at baseline, week 6 and week 12 of


      treatment.  Actually, all 4 treatments showed


      significant improvement, a characteristic of major


      depression.  The placebo group and the cognitive


      behavioral psychotherapy group were superimposed,


      one on top of the other.  There is no additional


      benefit from receiving cognitive behavioral


      psychotherapy, either in the slope term or at the


      end point, over receiving placebo.  There was


      significant benefit from fluoxetine alone, and the


      largest effect was associated with the combination


      condition.  The combination condition beat the CBT


      condition and the placebo condition on all 4


      efficacy measures.  Fluoxetine beat these CBT on


      all 4 measures and placebo and placebo on 3 of the


      4 measures.


                If we look at the impact of treatment


      using effect size calculations, the mean of the


      control condition minus the mean of the placebo


      condition, divided by the pooled standard




      deviation, the effect size for the combination was


      close to 1.  This is a very large effect.  For


      fluoxetine it was around 0.6, a moderate to large


      effect.  For CBT there was no difference between


      CBT and placebo, effect size calculated relative to




                If we look at response rates, defined as a


      clinical global importance measure rated by the


      independent evaluator of much improved or very much


      improved, 71 percent of the combination kids


      improved; 61 percent of the fluoxetine-treated


      patients improved; 43 percent of the cognitive


      behavioral psychotherapy treated patients and 35


      percent of the placebo-treated patients improved.


      Combination statistically was no different than


      fluoxetine.  CBT was no different than placebo.


      The two drug-containing conditions were superior to


      the two non-drug-containing conditions on responder




                If we look at the effect size calculated


      as a derivative of the odds ratio of improvement


      for the active treatments relative to placebo for




      the responder analyses, the results parallel the


      analyses on a scale or outcome variable, the


      Children's Depression Rating Scale.  The effect


      size for combination was 0.8, almost 0.6 for


      fluoxetine, and about 0.2 for cognitive behavioral


      psychotherapy.  So, again, clear superiority for


      the drug-containing conditions, with the largest


      effect reserved for the combination of medication


      management and CBT.


                Now, of interest here are the safety


      outcomes from the TADS trial.  Although we spent an


      enormous amount of time and energy on measuring


      adverse events, particularly measuring the impact


      of the treatments on the potential for harm, I


      think it is important to point out that the study


      did not have safety as a primary outcome.  With 439


      subjects randomized to 4 conditions, it is easy to


      see that for these outcomes the study is clearly




                It is I think important to separate


      ideation from behavior.  Despite the exclusion, we


      had significant suicidal ideation in the TADS




      sample.  This is looking at the Reynolds Adolescent


      Depression Scale, item 14, and 7.5 percent of the


      sample exhibited a score of 4 or above which is the


      threshold for clinical investigation on the RADS.


      CDRS item 6, serious suicidal ideation, the kind of


      suicidal ideation that leads you to consider


      hospitalization, 2 percent of the sample met CDRS


      item 13 criteria for severe suicidal ideation.  On


      the suicidal ideation questionnaire 2 measures, the


      SIQ flag which is to prompt clinical investigation,


      or elevated scores on any one of the items on the


      SIQ that would also prompt suicidal ideation, 29


      percent and 10 percent of the sample respectively


      on these measures exhibited clinically significant


      suicidality.  So, although suicidality was an


      exclusion criterion, there was plenty of suicidal


      ideation exhibited in the TADS sample at baseline.


                Now, as expected, suicidal ideation


      occurred across all 4 groups taken in the


      aggregate.  One sees here, looking at the CDRS item


      13 score, in this case greater than 1, or SIQ


      score, SIQ flag greater than 31, significant




      suicidal ideation at baseline.  It came down at


      week 6 and was significantly reduced overall at


      week 12.  This is the aggregated data across all 4


      treatment groups.


                Random regression analyses looking at


      between group differences on the SIQ, although it


      looks like these groups might be different at


      baseline, in fact there were no statistically


      significant differences on the SIQ in all 4




                One sees a different result than we found


      in the pattern for depression.  This is placebo;


      this is fluoxetine; this is CBT and this is


      combination.  The take-home messages here are


      three.  First, as we saw in the previous slide,


      suicidal ideation improves with treatment


      irrespective of which treatment one gets.  Second,


      fluoxetine and placebo are indistinguishable with


      respect to suicidal ideation, either with respect


      to the slope or at entry, indicating that


      fluoxetine, at least on average, is not provoking


      suicidal ideation.  Finally, the only treatment




      which separated from placebo with respect to


      reducing suicidal ideation was the combination of


      fluoxetine and CBT.  So, here the combination


      offers a significant advantage over medication




                Moving on to behavior, using a


      comprehensive adverse event monitoring procedure,


      we looked at the incidence of three kinds of


      harm-related adverse events.  Harm-related events,


      the broadest category, were defined as harm to


      self.  This could involve no suicidal ideation,


      ideation or attempt.  Or, harm to others which


      required aggressive ideation or actual behavior


      involving harming another person or physical


      property.  These events are subsumed one within the


      next.  So, a suicide-related event, which is a


      subset of harm-related events, involves harm to


      self, either ideation or attempt.  Then we had


      suicide attempts themselves.  What differentiates


      harm-related events from suicide-related events is


      primarily one subject with aggression and 7


      subjects, I believe, who exhibited self-injury




      without ideation, primarily cutting.


                These are the actual rates of harm-related


      and suicide-related events divided by treatment


      group.  One sees that there is a larger number of


      harm-related events and suicide-related events in


      fluoxetine-treated kids relative to placebo-treated


      kids.  The combination group is intermediate


      between harm-related and suicide-related events,


      intermediate between fluoxetine and placebo.  The


      cognitive behavioral psychotherapy group was


      roughly comparable to the placebo group.


                If you look at children who received drug,


      that is, combining the fluoxetine and the


      combination groups and comparing them to the


      placebo group, 10 percent, 22 of those


      fluoxetine-treated kids exhibited a harm-related


      event; 7 percent exhibited a suicide-related event;


      and the rates of these events overall were quite


      low, 7.5 percent of 439 kids, or 33 kids had a


      harm-related event, 24 of 439 kids, or 5.5 percent


      exhibited a suicide-related event.


                I think it is very important as we move




      through this discussion to understand that the base


      rates of these events are extremely low relative to


      the rates that we see for benefit.


                If we look at the odds ratios calculated


      from the actual rate data, the relative risk is 1.5


      for combination, 2 for fluoxetine, less than 1 for


      CBT.  Those is calculated relative to placebo.  For


      the collapsed category of fluoxetine and


      combination the relative risk is slightly greater


      than 2.  This is the only statistically significant


      relative risk in which the confidence interval


      crosses 1.  That is largely because these events


      are so rare so the power is quite low to identify


      these events.  In fact, the power for detecting a


      20 percent difference is about 10 percent.


                For suicide-related events there are no


      statistically significant differences although, as


      you can see from the graph, the odds ratios pretty


      closely track the odds ratios for harm-related




                The take-home message from this


      presentation actually is in this table--no, it is




      in the next table.  This is the table that looks at


      the suicide attempts in the trial.  We had 7 of


      them out of 439 kids, or slightly less than 2


      percent of the sample,  Two fluoxetine-treated


      kids, 4 combination-treated kids, 1 CBT and no


      placebo-treated patients made a suicide attempt.


      There probably is an imbalance in randomization


      which may in part be responsible for this.  There


      were more kids with an elevated SIQ flag randomized


      to the drug-containing than the non-drug-containing


      conditions so it is not clear what to make of this




                Here I think is where the take-home


      message lies relative to safety.  This is looking


      at the benefit/risk ratio using analyses for the


      number needed to treat and number needed to harm.


      What we see here is fluoxetine compared to placebo,


      in the first column; combination compared to


      placebo; and the collapsed category, SSRI versus no


      SSRI.  The absolute benefit increase is calculated


      as the control, the experimental event rate minus


      the control event rate.  So, it is the absolute




      benefit increase for receiving the treatment.  The


      absolute risk increase is calculated, again, as the


      experimental event rate minus the control event


      rate, that is, the risk increase attributable to


      the treatment over the placebo condition in


      absolute numbers.


                The NNT and the NNH are the reciprocal of


      the absolute benefit increase and the absolute risk


      increase respectively, 1 over the absolute benefit


      increase or 1 over the absolute risk increase.


      These are defined as the number of patients for


      benefit that would need to be treated to find one


      patient who had benefit over the benefit occurring


      from the control condition or, in the case of the


      NNH, the number of patients who would need to be


      treated to find one patient who would be harmed


      over treatment with the control condition.  So, it


      is a nice metric that combines both the absolute


      rate and the magnitude of the effect.


                One sees clearly here that 27 percent of


      patients benefit from treatment with fluoxetine,


      with an NNT of 4 which is a large effect; 27




      percent of patients benefit from treatment with


      combination, with an NNT of 3, again a very large


      effect; for SSRI versus no SSRI, combining the


      categories, 31 percent of patients benefit, again


      with an NNT of 3.


                The absolute risk increase for a suicidal


      event with respect to fluoxetine is 4.7 percent as


      compared to placebo; 2 percent for combination over


      placebo; and 3 percent for the combined category.


      NNH is number of patients that you would need to


      treat with the active treatment to find one patient


      who would be harmed over the control condition of


      21, 50 for the combination condition and 4 for the


      collapsed categories.


                From a clinical point of view, these


      patients would be easy to pick out in a crowd,


      easily identifiable who is getting better and who


      is not getting better, active treatment versus


      control.  These effects are so small and so


      uncommon that one could not possibly pick out


      patients who would be harmed by the medication


      versus patients who would commit these




      suicide-related event behaviors with placebo


      treatment.  If you calculate the NNT to NNH ratio


      looking at benefit to risk, one sees clearly here


      that the benefit tilts in favor of the treatment,


      and particularly the combination treatment.


                So, we conclude that the combination of


      fluoxetine and CBT is the most effective treatment


      for adolescents with major depression.  Fluoxetine


      alone is effective but not as effective as the


      combination of the two treatments.  CBT alone is


      less effective than fluoxetine and not


      significantly more effective than placebo.  We also


      conclude that placebo is acceptable in randomized,


      controlled trials of adolescent major depression


      and, in fact, is essential for looking at the


      adverse event outcomes, at least in this study.


      Suicidality decreases substantially with treatment.


      The improvement in suicidality is greatest with the


      combination and least for fluoxetine alone.


      Fluoxetine does not increase suicidal ideation.


      Suicide-related adverse events which are uncommon


      may occur more often in fluoxetine-treated




      patients.  CBT may protect against suicide-related


      adverse events in fluoxetine-treated patients.


      Taking both risk and benefit into account, the


      combination of fluoxetine and CBT appears superior


      as a short-term treatment for major depression in




                Now, the most practical clinical trialist,


      the kind of trial models that are used in other


      areas of medicine--cardiology, oncology, infectious


      disease for example, would much prefer a large


      simple or practical clinical trial in 2000 subjects


      to a meta-analysis of 10 under-powered subject


      trials.  So, it is our sense from looking at the


      FDA data and also the TADS data that we have a


      significant signal for drug treatment relative to


      suicidality but the evidence is not conclusive.  In


      fact, a definitive study has not been done and we


      would, as a field and as consumers of this


      information, much benefit from a


      placebo-controlled, practical clinical trial


      comparing fluoxetine to another SSRI, perhaps


      sertraline or citalopram.  This trial could be run




      easily on the child and adolescent psychiatry


      trials network, which is a clinical trials network


      that we are now putting in place to run these kinds


      of trials in the pediatric population.  Thank you.


                DR. GOODMAN:  Thank you, John.  My first


      question is will you be here tomorrow?


                DR. MARCH:  No.


                DR. GOODMAN:  Oh, you won't?  That may


      affect my subsequent questions because I am sure,


      besides myself, there will be a number of questions


      for you.  I don't know if we have time to take them


      all right now.  I wonder if there is any other


      option, Anuja.  What time do you leave today, Dr.




                DR. MARCH:  Noon.


                   Committee Discussion on TADS Trial


                DR. GOODMAN:  I am going to ask a


      question.  My understanding in looking at your


      results is that on the categorical measures of


      response fluoxetine is superior to placebo.


      However, if you look at a comparison of the mean


      scores on the CDRS fluoxetine is not superior to




      placebo.  Is that correct?


                DR. MARCH:  The slope term on the random


      regression analysis for the CDRS, the p value was


      0.08 for fluoxetine versus placebo.  Fluoxetine was


      statistically significantly different than CBT but


      not placebo.


                DR. GOODMAN:  So, from a standpoint of


      FDA, if this trial had been submitted to the FDA


      and you didn't have the CBT group, it seems to me


      that it might be classified as a negative study.


                DR. MARCH:  It would have been classified


      as a negative study using the CDRS as the primary


      endpoint, the slope term.


                DR. GOODMAN:  Do you have any comments


      about the methodology or outcome measures we are


      using and whether we are using the most appropriate


      ones in your opinion?


                DR. MARCH:  Well, I think it is actually a


      very important question.  If you look at the


      fluoxetine outcomes on the predicted endpoint, the


      week 12 endpoint on the CDRS predicted by the CDRS


      slope, on the clinical global improvement measure




      dichotomized and on the Reynolds Adolescent


      Depression Scale fluoxetine was statistically


      better than placebo.  It was simply a near miss on


      the slope term, which probably relates to the way


      the random regression analyses handle standard


      errors.  So, my sense of the story that the data is


      telling us is that fluoxetine--and also if you look


      at the effect size calculations--the story the data


      is telling is that fluoxetine is an effective


      treatment and it would be a mistake to consider


      this a negative trial.  On the other hand, the


      technical definition used by the FDA would require


      that the study be considered negative.


                DR. GOODMAN:  Dr. Perrin?


                DR. PERRIN:  On the other side of the


      equation, you made comments that the sample was


      somewhat different from some of the trials that


      have been sponsored by industry.  Could you be a


      little bit more specific about what the differences


      are, and how they might have affected the results,


      and what are the implications for meta-analyses of


      these studies?




                DR. MARCH:  I think it is a very important


      question, and we have a paper that is in press in


      The Journal of Child Medicine Psychiatry, the


      "orange journal," which describes the TADS sample


      in some detail and compares the TADS sample to


      epidemiologic samples and to treatment samples,


      both on the pharmacotherapy side, primarily the


      industry data sets, and also the cognitive


      behavioral psychotherapy trials, of which there are


      13 published at this point.  In general, our sample


      is not substantially different from either the


      epidemiologic or the treatment seeking samples,


      with the caveat that we are slightly sicker and


      slightly more comorbid, particularly relative to


      the CBT samples.


                So, given that the range of depression


      goes from mild to severe and half the sample is


      comorbid, there are plenty of patients in the data


      set who resemble the mildly ill patient all the way


      up to the severely ill, multiply comorbid patients.


      So, I think the result of the TADS trial is


      generalizable to the total sample.




                With respect to meta-analyses, it would be


      better to have a very large sample including all


      these variations, but by combining the data sets


      one gets a better picture, I believe, of the total


      variation in the patient population than simply


      using the industry data sets which tend to exclude


      the more complicated and clinically ill patients.


                DR. GOODMAN:  Dr. Newman, did you have a




                DR. NEWMAN:  Normally when you use a


      continuous outcome you have greater power than when


      you dichotomize.  I assume that is why you


      specified that as the endpoint at the beginning of


      the trial.  A reason why you might not is that the


      medication helps maybe a majority but actually


      harms a minority and then you could actually see


      that if you dichotomize the percent to improve is


      statistically significantly greater in the


      fluoxetine group but the mean may not be


      significantly greater because there are some people


      who are harmed and they drag the mean down, whereas


      they have no effect on the dichotomized variable. 




      Did you look to see whether there was evidence that


      the variation in the standard deviation in the


      effect size differed between the two groups and


      might have been greater with fluoxetine?


                DR. MARCH:  That was actually the point


      that I made, that the standard errors are larger in


      the fluoxetine-treated group than in the combo or


      the placebo--


                DR. NEWMAN:  Actually, not just the


      standard errors but the standard deviation, meaning


      that, in fact, there are some people who are harmed


      by it and that diminishes the apparent benefit when


      you average.


                DR. MARCH:  It would be impossible to look


      at the standard errors or the standard deviation


      and make a judgment about harm because there is a


      fair amount of variability data point to data point


      which is intrinsic to the disorder.  Some patients


      get better; some patients deteriorate.


                We do have in the safety paper a whole set


      of analyses looking at shifts, which I am not


      confident enough in to have wanted to present




      today.  We will try to examine what percentage of


      patients are getting worse with respect to ideation


      and behavior, and how that relates to treatment


      classification and also how it relates to other


      adverse events like mania activation, anxiety


      disinhibition and so on.  I think the secondary


      paper is going to shed a fair bit of light on these




                DR. GOODMAN:  Dr. Pfeffer?


                DR. PFEFFER:  I have two questions.  One


      is were there any differential dropout rates in the


      samples?  This also relates to the question of


      compliance in the different treatments.  My last


      question is these were intent-to-treat analyses?


                DR MARCH:  All very good questions.  The


      analyses are all intent-to-treat.  Although I


      didn't present the data, if we look at observed


      cases analyses, those who were still on their


      assigned arm at any given assessment point or


      completer analysis, the results are exactly the


      same.  About 10 percent of the kids in each


      treatment overall dropped before the week-12 data




      point.  Another 10 percent were what we call


      prematurely terminated.  That is, for ethical


      reasons.  They received an out of protocol


      treatment at some point during the first 12 weeks


      of the study.  There were no statistically


      significant differences across treatment groups in


      either the rate of dropping out or the rate of


      receiving an ancillary treatment, that is, a


      premature termination.


                You will see this afternoon when FDA


      presents its analysis of the TADS data that the


      odds ratios for being harmed by receiving


      fluoxetine are greater than we presented, or I


      presented this morning on behalf of the TADS team.


      That is because the FDA data set excluded those


      kids who were prematurely terminated and received


      another treatment.  That actually represented two


      kids in the placebo group and that, in turn,


      inflated the odds ratios in the FDA results versus


      the TADS results.  So, there is some method


      variance in there which accounts for the


      differences between the two findings.




                DR. GOODMAN:  Dr. Gorman?


                DR. GORMAN:  Does analysis of your data


      set allow interpretation of the time of onset of


      treatment to behaviors that we are studying today?


                DR. MARCH:  That is a very good question


      and, in fact, one that we will address in the


      secondary paper.  I can tell you that the majority


      of the events occurred within the first 6 weeks but


      not within the first 2 weeks.  But I don't have


      that data presented in slide form so I can show it


      to you, but it will be in the safety paper that we


      are currently preparing.


                DR. GOODMAN:  So you don't know what the


      differential rates are between the groups at this


      point, particularly between fluoxetine and placebo?


                DR. MARCH:  In terms of time?


                DR. GOODMAN:  Yes, in terms of the early




                DR. MARCH:  My general impression, looking


      at the data, is that the fluoxetine events occurred


      early and the placebo events occurred later, which


      is kind of what you would expect given what we know




      about the compounds.  But I wouldn't want to cite


      chapter and verse or have you base your decisions


      based on that because we haven't completed the


      final analyses of the data.


                DR. GOODMAN:  Dr. Rudorfer, you will be


      the penultimate questioner.


                DR. RUDORFER:  If I can go back to the


      characteristics of the patients for a moment, we


      will be looking at a number of studies that were


      submitted to the FDA by various sponsors, and it


      seems to me that the TADS inclusion criteria go


      beyond DSM-IV in terms of length of illness and


      degree of dysfunction in various spheres of life.


      Could you comment on that?


                DR. MARCH:  Sure, it is a very good


      question.  That is, the exclusion criteria included


      requirements that were designed to ensure a stable


      baseline.  So, we required at least six weeks of


      mood disorder symptoms that crossed two or three


      contexts--home, peers in school--which, of course,


      are not required in the DSM-IV criteria.  This was


      done in part to minimize the chance of a placebo




      response and to ensure that we had a sick patient


      population that would be both ethical to randomize


      and would offer some opportunity for the


      combination treatment to separate from the two


      monotherapy conditions.


                I think we actually designed a very good


      experiment, and looking at a 35 percent placebo


      response rate did exactly what we had intended to


      do.  But in that sense, this population is sicker


      perhaps than what is seen in the industry data sets


      and certainly sicker than what is seen in the CBT


      data sets on average.


                DR. GOODMAN:  I will permit two final


      questions and that is it, one from Dr. Pine and


      then Dr. Fant.


                DR. PINE:  I want to return to the first


      question from Dr. Goodman as far as how this study


      would be evaluated from an FDA perspective.  My


      sense from reading the paper is that there were two


      primary outcome measures and three analyses, and


      two of the three were positive so that the CGI


      analysis done categorically was positive, the CDRS




      analysis done categorically was positive, and it


      was only the third analysis, the CDRS continuous


      measure, that was not positive.


                So, my take on that from an FDA standpoint


      of, you know, do the primary outcome measures make


      it or not is that it would be closer to positive


      than negative.  Do I have that right?


                DR. MARCH:  You have it partially right.


      There is a CDRS slope analysis, and whether that is


      positive or negative depends on whether you treat


      the intercept term as random.  I mean, there is a


      fair amount of method variance on a subtle level


      that can tilt these things one way or the other


      when it is a near miss.  There is a CDRS endpoint


      analysis based on the predicted or marginal mean.


      There is a categorical analysis, logistic


      regression, and there is a self-report scale which


      was included, the Reynolds Adolescent Depression


      Scale, which was included because the CBT


      literature relies heavily on patient self-report.


      Three of those four measures, all but the CDRS


      slope analysis, were positive for the




      fluoxetine-placebo comparison.  The CDRS slope


      analysis, again, was a p value of 0.08, a near




                So, I think the take-home message is


      actually in the effect sizes, not in whether you


      are looking at p values or not.  Clearly,


      combination and fluoxetine have larger effect


      sizes, meaningful effect sizes relative to placebo


      as contrasted to CBT.


                DR. FANT:  For the sake of the study I


      know it was necessary to exclude certain patients


      to optimize the conditions for the study, but I


      think in real-world practice a lot of the patients


      who were excluded would be patients who would be


      prescribed medication under various conditions.  Is


      there any reason, from your standpoint or


      perspective, to think that that population of


      patients may be at a different risk for fulfilling


      suicidal attempts, ideation, and carrying it


      through to the ultimate result, or may be affected


      differently by the medication than patients that


      are not excluded from the study?




                DR. MARCH:  That is a very good question.


      That is, is there some issue to believe that there


      would be a differential treatment response in


      patients who would be excluded, particularly


      excluded for harm to self or others, as compared to


      the TADS sample of patients?  I don't know of any a


      priori reason to believe that there would be a


      differential treatment effect relative to the TADS


      sample.  I do think it is quite clear from the


      treatment and epidemiologic literature that they


      would be at higher risk for adverse harm-related


      outcomes but whether they would be more at risk


      than, say, the TADS sample patient I don't think we


      know.  My guess as a clinician is probably not.


                We are actually doing an NIMH-funded study


      called the Treatment of Adolescent Suicide


      Attempter Study, in which we are comparing a


      medication algorithm to cognitive behavioral


      psychotherapy to the combination--no untreated


      control condition obviously in this sample--to try


      to understand something about treatment for this


      particular patient population precisely because




      they have been excluded from these other trials,


      and we need additional data on their care.  So,


      that trial is now under way and should be completed


      in the next couple of years.


                DR. GOODMAN:  Thank you very much, John.


      The final question that I will take the chairman's


      privilege to ask is do you have data that you


      haven't presented yet on long-term outcome based on


      this trial?


                DR. MARCH:  The final subjects in the


      trial are out in a naturalistic follow-up window so


      that 36-week data is in the can, but that data set


      has not been cleaned and locked yet.  We expect it


      will be cleaned and locked and ready for analysis


      in the spring, and we hope to have that data in


      press by this time next year.


                DR. GOODMAN:  Thank you very much, John.


                DR. MARCH:  Thank you.


                DR. GOODMAN:  I would like to ask our next


      speaker to come forward, Dr. Greg Dubitsky.


           Characteristics of Pediatric Antidepressant Trials


                DR. DUBITSKY:  Good morning.  You just




      heard about the TADS trial from Dr. March.  I would


      like to now go on and briefly summarize the other


      studies that were included in the FDA's primary


      analysis of suicidality.


                I do want to emphasize that my review and


      this presentation are really descriptive only.  I


      am not going to touch on efficacy outcomes or


      safety outcomes.  The discussion of the risk of


      suicidality in these trials with be presented later


      this morning by Dr. Hammad.


                The study pool, again excluding the TADS


      study, consisted of 23 placebo-controlled studies


      which were conducted between 1984 and 2001.  Each


      study was done with one of nine different


      antidepressant drugs and studied patients with one


      of five different diagnostic indications; major


      depression, obsessive compulsive disorder,


      generalized anxiety disorder, social anxiety


      disorder or attention deficit disorder.


                These studies all had some features in


      common.  They were all randomized, double-blind,


      placebo-controlled.  They utilized a parallel group




      design and a flexible dosing regimen.


                I did prepare a handout to go with this


      talk which should be in your packets, at least for


      the advisory committee members.  It consists of two


      tables which summarize some of the design


      characteristics of these 23 studies.  Table 1 has


      some basic study information, to include the


      diagnostic indication, the age range that was


      studied, number of patients by treatment group, the


      duration of double-blind treatment, and the dose


      range that was used in the particular study.


                I would like to point out though that I


      don't intend for everybody to read this and


      memorize it; this is really for reference for later


      this morning when you hear about the analysis of


      these trials.


                The second table in the handout includes


      some information on screening and exclusionary


      criteria.  Some of the studies used very extensive


      diagnostic screening.  I have indicated those in


      the table.  I have also indicated information on


      whether there was a placebo lead-in, and also




      whether certain exclusionary criteria were employed


      in the various studies to include whether people


      were excluded who had a history of treatment


      resistance, current suicide risk, history of a


      suicide attempt, bipolar disorder, or family


      history of bipolar disorder.


                My review of these studies did include a


      number of other variables.  I have listed in these


      two tables the most relevant ones but in my review


      that is on the Internet I do describe some other


      characteristics which you might be interested in,


      such as the location and number of sites, whether


      stratified randomization by age group was utilized


      and other exclusionary criteria such as homicidal


      risk or the presence of psychotic symptoms.


                There were a few notable differences


      between these studies that I would like to point


      out.  One study with Prozac, HCCJ, was a very small


      study.  It was the smallest of the 23 studies, with


      only about 40 patients and it was terminated early.


                Only one of the 23 studies included an


      active control arm.  That was study 329 with Paxil




      in major depressive disorder.  That included an


      imipramine control arm.  The others only had a


      placebo control.


                Two of the studies did include inpatients


      as well as outpatients, the Celexa study, 94404,


      and Wellbutrin, 75.


                Last, I did want to point out that three


      of the studies did use a rather extensive


      diagnostic screening of the patients, much more so


      than the other studies, Prozac studies X065 and


      HCJE, and Paxil study 329.  Those three studies


      were done in major depressive disorder.


                One other difference involves the


      treatment options after patients completed the


      acute phase of double-blind treatment.  This was


      quite variable across the trials.  In eight studies


      there was a taper of acute treatment before


      discontinuation.  Seven other trials just abruptly


      discontinued treatment, and there was no provision


      for continued treatment.  Five trials did allow for


      continuation of open-label treatment, and in three


      trials patients could continue double-blind






                However, this was also very variable


      within trials.  For instance, in Paxil 329


      responders could continue double-blind treatment


      but non-responders were tapered off treatment.


      This variability in the follow-up treatment


      following the acute phase made it very difficult to


      do any analysis of suicidality-related events post


      double-blind treatment.


                I would like to point out that none of


      these studies was specifically designed to assess


      suicidality.  Suicide attempts and ideation were


      detected only through routine safety monitoring,


      that is, through treatment emergent adverse events


      and through suicide-related items on various


      depression scales, such as the HAM-D and the CDRS.


      One problem with this is that often descriptions of


      possibly suicide-related events were rather vague


      or incomplete and often made it difficult to reach


      a classification.


                I have no specific conclusions since this


      is really a descriptive review and overview of the




      studies.  I think one of the important questions


      that arises from this information though is whether


      any of these differences in design characteristics


      could contribute to any observed differences in


      suicidality risk that we observed across these


      studies.  That is a question that will be addressed


      later this morning by my colleague, Dr. Hammad.


      So, that is all I have.


                DR. GOODMAN:  Thank you for being concise


      and providing us with an outstanding handout for


      our reference.  We have one question.  Dr.




                DR. RUDORFER:  Thank you.  Could you


      clarify, of the 23 trials how many were submitted


      in response to the pediatric exclusivity rule?


                DR. DUBITSKY:  I don't have the exact


      number.  I believe most of them were but some of


      them were submitted well before pediatric


      exclusivity took place or came into effect.  I


      don't have the exact number off the top of my head.


                DR. GOODMAN:  Dr. O'Fallon?


                DR. O'FALLON:  Asking the question in a




      somewhat different way, the data that you have for


      this reanalysis, does any of that data come from


      outside, beyond the data that was submitted to the


      FDA?  That is, were you able to go in and obtain


      data from studies that were never submitted to the


      FDA for approval or whatever?


                DR. DUBITSKY:  Well, to my knowledge,


      there was one study that had not been submitted as


      part of an efficacy supplement or an approval


      package.  The other ones, I believe, were.  Dr.


      Hammad actually requested data sets for all these


      studies.  Correct me if I am wrong, but I believe


      we had relatively complete data sets to allow


      reasonable analysis for all these studies.


                DR. O'FALLON:  But I am asking whether


      there are, as some are claiming, studies that were


      done but were never submitted to the FDA.  Are


      there any of those data here, if they exist?


                DR. DUBITSKY:  There are some studies that


      are not included in this analysis, but those are


      mainly open-label continuation studies of the acute


      studies.  Also, there were a number of pediatric




      pharmacokinetic studies but I think for obvious


      reasons we didn't include those in the analysis.


      But, to my knowledge, I think we have everything.


                DR. GOODMAN:  Dr. Laughren, do you also


      want to respond to the question?


                DR. LAUGHREN:  I think I can respond to


      that.  The vast majority of these programs were


      submitted under pediatric exclusivity so the


      companies were required to submit every scrap of


      data they had as part of those supplements.  The


      only trial here that was not submitted as part of


      an application, in terms of a company trial, was


      the ADHD study for Wellbutrin.  The other study


      that we have included safety data for is the TADS


      trial and, of course, that was also independent.


      But those are the only two trials of the 24 that we


      looked at that were not submitted as part of an




                DR. GOODMAN:  Dr. Marangell?


                DR. MARANGELL:  How many of the studies


      excluded family history of bipolar disorder?


                DR. DUBITSKY:  Let's see, actually I think




      I have that in table 2.  I don't know the number


      off the top of my head.  It looks like about ten of


      the studies excluded a family history of bipolar




                DR. MARANGELL:  Thank you.


                DR. GOODMAN:  Dr. Perrin?


                DR. PERRIN:  You are saying basically that


      the extensive diagnostic screening occurred only in


      three studies, I believe.  Is that right?


                DR. DUBITSKY:  I am sorry?


                DR. PERRIN:  The extensive diagnostic


      screening occurred only I think in three


      studies--one of the points that you made.  Does


      that give us some information about the potential


      diagnostic heterogeneity and also raise questions


      about whether entrance into these studies of


      children, ages 7-17, might not have had MDD in the


      MDD studies?  My last related question is, since I


      am not a psychiatrist at all, what do we know about


      the ability to distinguish bipolar disorder from


      MDD in the 7-17 year-olds?


                DR. DUBITSKY:  Well, it is true that in




      looking across all 23 studies, those three studies


      did stand out as far as using more extensive


      diagnostic criteria.  I believe that it certainly


      is possible that we might have more confidence that


      those patients did actually have the diagnosis


      under consideration.  Whether that is actually true


      or not, I don't know and I don't know any good way


      of figuring that out.


                I am not a child psychiatrist so I can't


      answer your last question about the ability to


      diagnose.  I understand it is very tricky though.


                DR. GOODMAN:  Thank you again.  I would


      like to ask Dr. Kelly Posner to come up to the


      podium to present.  Dr. Posner is from Columbia


      University and she will be talking to us about the


      reclassification of the clinical trials data


      according to suicidality.


                  Classification of Suicidality Events


                DR. POSNER:  I would like to start by


      introducing my expert work group from Columbia that


      included myself, Dr. Maria Oquendo, Dr. Barbara


      Stanley and Dr. Madelyn Gould.  Dr. Stanley and Dr.




      Gould are here with me today.  Our statistical


      consultant was Mark Davies.


                Why was reclassification needed?  The


      problem is that the field is challenged by a lack


      of well-defined terminology and common language to


      refer to suicidal behavior, and this was reflected


      in the lack of standardized language used in the 25


      trials in question.  That is why there was


      difficulty in interpreting the meaning of all of


      these reported adverse events that occurred in


      these trials.  So, AEs that should have been called


      suicidal may have been missed and there may have


      been AEs that were inappropriately classified as




                Here are some illustrative examples of the


      difficulties in adverse event labeling in the


      field.  I want to make sure to note that these


      labels have nothing to do with the labels the


      sponsor gave these events, but just original


      investigators at the site.  Again, they are extreme


      examples just to reflect the problem.


                You see the first one, it says patient




      attempted to hang himself with a rope after a


      dispute with his father.  Investigator did not


      consider this event to be a suicide attempt but


      called it a personality disorder in this 10


      year-old patient.


                The second one is one we have all heard a


      lot about.  The patient is reported to have engaged


      in an episode of auto-mutilation where she slapped


      herself in the face, called a suicide event.  Then,


      the patient took 11 tablets impulsively then went


      to school--called a medication error.


                So, how do we address this problem?  Well,


      a common set of guidelines needed to be applied and


      we needed to look at the data consistently across


      trials using research-supported definitions and


      concepts that had reliability and validity.  We


      also needed to broaden the range of adverse events


      that we were looking at.  This was for two reasons.


      The first one is to avoid bias in readings.  We


      wouldn't have wanted the expert raters only to have


      had what the sponsors had identified as possibly


      suicidal.  Also, to identify suicidal events that




      may have been missed.


                So, what was included in this broadened


      range of events?  Of course, the events originally


      identified by the sponsors as possibly suicide


      related, all accidental injuries which included


      accidental overdoses, and serious adverse events


      which includes life-threatening events and all




                Why did we need experts in suicide?  Well,


      you all heard about the limited information


      provided in the narratives, particularly frequent


      lack of stated suicidal intent.  So, only experts


      in suicide would have allowed for inference based


      on details of behaviors and related clinical




                This is the list of our very distinguished


      international panel of experts.  I will just read


      their names very quickly, Drs. Bautrais, Brent,


      Brown, Van Herringen, King, Mazark, O'Carroll,


      Rudd, Spirido and Miller.


                So, what was the Columbia classification?


      I want to move to this slide because it goes




      through the definitions which I will just go


      through briefly.  Suicide attempt, of course, which


      is defined as a self-injurious behavior associated


      with some intent to die.  Intent can be stated or


      inferred by the rater.  It is important to know


      that no injury is needed.


                Then there was preparatory actions towards


      imminent suicidal behavior.  So, the person takes


      steps to injure himself but is stopped by self or


      other, anything beyond the threshold of a


      verbalization but not quite making it to a suicide




                Then we had self-injury behavior, intent


      unknown.  These are cases where we know there was


      some self-injury but we don't know what the intent


      was.  So, the associated intent to die is unclear


      and cannot be inferred.


                Self-injurious behavior with no suicidal


      intent is the next category.  That is where, again,


      we know there was deliberate self-harm but there


      was no intent to die so behavior is intended to


      affect other things.  This is what we think of




      self-mutilation typically.


                Suicidal ideation was the next relevant


      category, which can be passive or active thoughts,


      passive thoughts of wanting to be dead or active


      thoughts about killing oneself.


                Then we had all the other categories.


      That is essentially one rating, anything other than


      deliberate self-harm or something suicidal.  That


      could include accidents, psychiatric events or


      medical events.


                Finally, we had not enough information,


      which meant that there was insufficient information


      for a rater to be able to say whether or not there


      was some deliberate self-harm or something




                The scheme is laid out conceptually here


      for you.  I think it helps make a little more sense


      of it.  The blue boxes refer to what you will hear


      later as the FDA's primary outcome.  These are


      ratings that are considered definitively suicidal,


      suicidal behavior and suicidal ideation.  You see


      codes 1, 2 and 6.  Suicide attempt, preparatory




      actions and ideation.  The next are non-suicidal


      events, all the other events and the self-injurious


      behavior without suicidal intent, and then


      indeterminants.  The green boxes are what will be


      referred to as the sensitivity analysis, things


      that could have been suicidal but there is no way


      to know.


                So, what was done?  The classification


      methodology involved, of course, choosing the


      expert panel who had expertise in adolescent


      suicide and suicide assessment, based on reputation


      and publications.  They had no involvement in


      industry youth depression trials in question, and


      no expert rater was employed by Columbia




                We had a training teleconference to review


      classification parameters, then training


      reliability exercise to ensure appropriate


      application of classifications.  All case


      narratives were blinded to any potentially biasing


      information, and I will review that in a minute.


      There was random distribution of 427 events to 9




      expert raters.  Each case was independently rated


      by 3 raters.  Each rater received approximately 125


      events to rate, and any group of 3 raters shared


      only 5 cases.  All ratings were reviewed for


      quality assurance and identification of


      non-agreement cases.  Consensus teleconferences


      were held for any disagreement cases, and there was


      double data entry for quality assurance.


                Now, what was the consensus process I


      referred to?  If ratings did not have unanimous


      agreement, then a consensus discussion was held.


      Each case was discussed by the three raters


      involved only.  Discussion of each case was led by


      an expert other than those originally assigned the


      case.  The goal of the discussion was to reach 100


      percent agreement.  If 100 percent agreement could


      not be reached, the case then became indeterminate.


      Sometimes the original majority opinion did not


      always end up as the final consensed


      classification.  In other words, if there was a


      minority rating, sometimes that ended up being the


      final outcome.




                Now, what was rated?  Blinding of event


      narratives to avoid bias included--we received the


      narratives from the FDA blind to all potential drug


      identifying information.  This included drug name,


      company sponsor name, patient identification


      numbers, whether they were on an active or placebo


      arm, and any and all medication names and types


      because there could be associated treatments that


      might bias somebody or tip them off as to what drug


      was being talked about and, of course, primary


      diagnosis.  We also did some additional blinding of


      potentially biasing information which included the


      original label of the event given by the


      investigator or sponsor and serious or non-serious




                Rating guidelines--how was the


      classification scheme applied?  We wanted the


      experts to apply concepts using their clinical


      expertise and judgment; to use their experience to


      integrate clinical information and infer when


      appropriate.  We wanted them to have a reasonable


      certainty in order to commit to a rating, and




      rating was based on what was probable, not what was




                The guidelines for intent inference


      involved inferring if something was clinically


      impressive, and I am going to give you an example


      of that in a moment, or using two smaller pieces of


      clinical information.  The clinical information


      that could inform inference of intent included


      clinical circumstances.  That could be method used,


      number of pills; past history of suicide attempt;


      past history of self-injurious behavior or


      self-mutilation; and family history of suicide or


      suicide attempts.


                Now, here is a case example of inferred


      intent, what we call clinically impressive


      circumstances.  This is the first time you are


      actually seeing one of these real narratives.  In


      this case clinical impressiveness actually


      overruled stated intent, so you see the subject


      attempted suicide by immolation.  Her siblings


      doused the flames immediately.  She was left with


      minor burns on her abdomen and on her left




      shoulder.  The subject admitted she was angry with


      her parents for going away and leaving her alone at


      home because she was fearful.  The subject admitted


      that she had acted impulsively and had not intended


      to kill herself.


                Here are more examples.  This is another


      example actually of clinically impressive


      circumstances which was ultimately called a suicide


      attempt.  It is also important to know that we had


      no idea what the sponsor ratings were but both


      these cases were consistent with what the sponsor


      had said as well.


                This case involved a 16 year-old who


      claimed to have ingested 100 tablets of study med


      after a fight with her mother.  The patient


      informed her mother.  The mother brought her to the


      ER.  The patient reported feeling shaky.  Emergency


      room physician said she was slightly tachycardic


      with a pulse of 100.  The tox. screen was negative


      but the patient did have some illness and she


      stayed in the ER until she was asymptomatic, and


      then was later admitted to the psych. unit.




                Another example of a suicide attempt, a


      patient age 17 took an overdose of 20 tablets.  In


      the father's opinion the overdose was 5 tablets.


      The patient didn't have any symptoms of an


      overdose, not even nausea, but it was classified as


      a suicide attempt, of course.


                More overdose examples.  You see in this


      first example there were 113 tablets and it


      exemplifies how medication types were blinded so


      you see all the different numbers there.  Then, the


      next one is patient aged 15, impulsively slit her


      wrist following an altercation with her mother.


      Finally, age 17, attempted suicide by taking 8


      tablets after a fight with her father, whom she


      considered harsh and rejecting.


                Now, these are examples of self-injurious


      behavior, intent unknown.  So, this is where we did


      some harm but we just don't know why.  A patient


      aged 10 had superficial scratches, left arm,


      scratched herself with scissors.  That was all the


      information that was there essentially.


                Patient, aged 14, ingested or simulated




      ingestion of 2-3 cigarettes.  The patient was


      reported as feeling tired and playing a theatrical


      role.  Subject, aged 9, reported he had ingested 4


      of his brother's tablets on a dare.  Finally,


      patient, aged 10, swallowed a small amount of


      after-shave lotion while angry.  It is hard to know


      what to make of those without information.


                Examples of preparatory actions, age 16,


      tried to hand herself and was prevented from doing


      so by her family.  The next case, age 18, a voice


      commanded him to jump from the roof.  Although he


      went up, he did not jump.  Next one, age 10, held a


      kitchen knife to her neck while alone but did not


      cut herself.  Event was not witnessed.  Finally, a


      patient, age 18, was noted to be hostile, hopeless


      and helpless and had written suicide notes.  As I


      said, anything beyond a verbalization was


      considered a preparatory action, including writing


      a suicide note.


                These are good examples of self-injurious


      behavior, no suicidal intent.  In the first case


      the patient stated there was increased family




      tension.  She made superficial cuts on her wrist


      with an Exacto knife.  The patient and mother


      reported the cuts weren't deep and they looked like


      cat scratches.  Patient adamantly denied any


      suicidal gestures or intent.  She stated she only


      wanted a release and that cutting and hitting her


      legs offers her a release.


                The second case, denied suicidal thoughts.


      The first time she cut herself was age 16.  She


      stated she did it for attention.  Today her cutting


      was more spontaneous. She reported that cutting


      gives her a good weird feeling.


                So, what were the results?  This slide


      just refers to what we are talking to in our


      results, or referring to, and there were 427 events


      but some patients had more than one event so we


      ultimately, in the reliability data, used 378


      cases.  We employed the same severity hierarchy


      that the FDA used.  So, we just took the most


      severe event for cases that had multiple events.


                Expert rater consensus--only two of 427


      cases had no agreement among the three raters.  So,




      each rater had a different rating in only two


      cases.  Fifty-nine cases had agreement among two of


      three raters, and those had to go to


      teleconference.  There were no cases in which


      consensus was not able to be reached during the


      teleconference and they, of course, had that




                Now, discordant cases between the sponsor


      and Columbia classifications, there were 40 out of


      the 427 cases in which the sponsor and the Columbia


      classification differed.  Twenty-six new cases were


      identified that had not been identified by the


      sponsor as possibly suicide-related.  There were


      two new cases of self-injurious behavior without


      suicidal intent that had been labeled something


      other than deliberate self-harm, and 12 cases were


      originally called possibly suicidal and were


      changed to something other than possibly suicidal.


                Here it breaks it down for you further.


      Of the 26 new possibly suicide-related events, one


      was a suicide attempt; one was a preparatory act;


      13 were ideation events; four were intent unknown




      acts; and seven were not enough information to say


      whether there was deliberate self-harm.


                Here is an example of one of the newly


      identified suicidal events.  This is a preparatory


      act.  The patient, age 11, held a knife to his


      wrist and threatened to harm himself.  The patient


      was hospitalized with an acute exacerbation of


      major depressive disorder.  The reason we have this


      is because, as I mentioned before, every


      hospitalization is a serious adverse event so that


      is why this preparatory act was caught.


                The events that were changed from suicidal


      to something other included two changed to


      psychiatric; one changed to an accident; and nine


      changed to self-injurious behavior with no suicidal


      intent.  Again, our famous example, a patient


      reported to have engaged in an episode of


      auto-mutilation where she slapped herself in the


      face.  The event resolved the same day without any




                These are actually the kappa, the


      agreement between the sponsor and Columbia.  So,




      Columbia's classification of possibly


      suicide-related and the sponsor's classification of


      possibly suicide-related, the kappa was 0.77.  You


      see in the 2 X 2 table that the numbers correspond


      to the numbers that I just went through with you.


                Now, if you want to look somewhat more


      specifically or at least what we think is more


      specific, we did a comparison of what Columbia said


      was definitively suicidal and what the sponsor said


      was possibly suicidal, and the kappa was 0.69.


                Here are the reliability results of the


      ratings with the nine expert raters.  The median


      ICC was 0.86 and what the FDA will refer to as the


      primary outcome variables, you can see the numbers


      here, suicide attempts is 0.81; preparatory


      actions, 0.89; suicidal ideation, 0.97.


                Where do we go from here?  We need to


      improve our adverse event reporting for


      suicide-related events by developing consistent


      terminology; developing guidelines for


      classification of suicidality so that adequate


      information is provided by the clinician;




      utilization of research assessment tools, what


      questions to ask, how to ask, and what measures aid


      this; finally, hopefully, that will lead to


      improved, more valid identification and


      documentation of suicidality.


                DR. GOODMAN:  Thank you very much.  Dr.




                DR. CHESNEY:  Thank you.  This is a little


      bit of thinking outside the box, but we heard at


      the February meeting a number of examples of


      homicidal behavior.  I wonder if, in your


      speciality, homicidal behavior is ever identified


      as being self-injurious primarily to affect


      circumstance or to affect an internal state.


                DR. POSNER:  No, that did not represent


      any of those self-injurious, no suicidal intent


      ratings.  So, the classifications that you are


      referring to, internal state and circumstance, are


      not synonymous at all with the cases that had


      homicidal ideation or any kind of aggressive


      behavior.  It doesn't mean that it couldn't be


      looked at in another analysis but it wasn't




      represented in these cases.


                DR. CHESNEY:  I guess my more general


      question, I just wonder in the bigger question,


      homicidal behavior outcome is bound to be bad and


      self-injurious, and if it is just another factor


      that we should consider in this whole picture.


      Thank you.


                DR. GOODMAN:  Dr. Robinson?


                DR. ROBINSON:  Do you know how many of the


      events led to hospitalization and how it breaks


      down in terms of your classification?


                DR. POSNER:  Dr. Laughren, do you know?


                DR. LAUGHREN:  I don't have that figure


      off the top of my head.  There were a substantial


      number of events leading to hospitalization, I


      believe somewhere in the ballpark of maybe 40


      percent.  I don't have the exact number.  It was a


      common outcome.


                DR. POSNER:  It is important to know that


      we were very narrowly just looking at obtaining the


      most appropriate label for the particular event in


      question, and we didn't have any of the surrounding




      information or follow-up information in this


      particular piece of the project.


                DR. GOODMAN:  Dr. Wang?


                DR. WANG:  I have a question.  Do you know


      how many of the sponsors originally submitted


      reports that were categorized as serious?  The


      reason I am asking is to get a sense of how many


      cases may not have been originally reported.  I


      know you had these serious cases sent to you for


      adjudication, just to check in case there were


      cases that were being missed in what the sponsors


      were reporting, but did you look as to how many


      cases were not considered serious by the sponsor,


      jut to give us a sense of how many may be sort of


      out there in the non-serious pool?


                DR. POSNER:  Again, we were blinded to


      sponsors' classifications throughout the entire


      process.  I don't know if somebody from the FDA can


      answer that question for you.


                DR. LAUGHREN:  Again just a ballpark


      figure, I think it is probably somewhere in the


      vicinity of maybe 65-70 percent.  But you have to




      understand that a designation of serious is a


      judgment that is made by the sponsor fairly


      subjectively.  I mean, there are criteria for


      regulatory serious.  It is fatal, life-threatening,


      seriously disabling, leading to inpatient


      hospitalization.  But even though, you know, that


      on face appears to be fairly definitive, sponsors


      in many cases, in my view, made the judgment that


      if it was considered to be suicide-related it was,


      by definition, serious.


                So, if you look at many of the narratives


      that were classified as serious, I think no


      reasonable person looking at those would consider


      that, in a common sense notion, as a serious event.


      But the point is that that designation--how that


      judgment was made varied from sponsor to sponsor.


      So, you know, some of them classified many more of


      the events as serious than other sponsors.  But the


      answer to the question is that overall roughly


      two-thirds of these events that were included in


      the analysis were designated as regulatory serious.


                DR. GOODMAN:  Ms. Griffith?




                MS. GRIFFITH:  I have a question about


      cutting specifically.  It seems to me that most of


      the examples of cutting fall into self-injurious


      behavior, intent unknown or self-injurious behavior


      with no intent.  I am just curious as to are you


      confident that the reporting that you received and


      reviewed actually got to whether or not there was


      intent or no intent, and how subjective is the


      reporting likely to be?


                DR. POSNER:  Again, as you can see,


      cutting is a method that is used both in suicidal


      behavior and self-injurious behavior without


      suicidal intent.  If you remember the conceptual


      scheme, there was the category self-injurious


      behavior, intent unknown, because cutting can be


      used both ways.  I forget the exact number but


      there were 20-something cases in which they cut but


      we don't know if it was suicidal or not.  That is


      why we had to come up with a category just to


      categorize and deal with that issue.  The FDA will


      point out that the included that in the sensitivity


      analysis so just in case all of those were




      suicide-related events, they have those numbers.


                DR. LAUGHREN:  If the question you are


      asking were the narratives lacking in detail, they


      absolutely were.  These were not by any sense


      complete descriptions.  Ideally, many more


      questions would have been asked when these events


      occurred to help flesh them out.  That is why it


      was necessary to use inference as one approach to


      try and get at intent because intent was not


      included for the vast majority of these.


                DR. POSNER:  Which is why only experts in


      the field could have been able to infer from the


      surrounding information.  The narratives were


      limited with respect to suicidal intent often but