FOOD AND DRUG ADMINISTRATION
CENTER FOR DRUG EVALUATION AND RESEARCH
PSYCHOPHARMACOLOGIC DRUGS ADVISORY COMMITTEE
Wednesday, November 3, 1999
620 Perry Parkway
CAROLE A. TAMMINGA, M.D.
Professor, Department of Psychiatry
University of Maryland at Baltimore
Maryland Psychiatric Research Center
Spring Grove Hospital, White Building
Baltimore, Maryland 21228
SANDRA TITUS, PH.D.
Food and Drug Administration
Center for Drug Evaluation and Research
5630 Fishers Lane
Rockville, Maryland 20857
EDWIN H. COOK, JR., M.D.
Director, Laboratory of Developmental Neuroscience
Department of Psychiatry, MC3077
University of Chicago
5841 South Maryland Avenue
Chicago, Illinois 60637
ROBERTO A. DOMINGUEZ, M.D.
Professor of Psychiatry
Department of Psychiatry
University of Miami, School of Medicine
1611 N.W. 12th Avenue
Miami, Florida 33136
ABBY J. FYER, M.D.
Anxiety Disorders Clinic
New York State Psychiatric Institute
College of Physicians and Surgeons
722 W. 168th Street, Unit 82
New York, New York 10032
BARBARA GELLER, M.D.
Professor of Psychiatry
Washington University School of Medicine
4940 Children's Place
St. Louis, Missouri
COMMITTEE MEMBERS: (Continued)
ROBERT M. HAMER, PH.D.
Department of Psychiatry
R.W. Johnson Medical School, D-331 UBHC
University of Medicine & Dentistry of New Jersey
Piscataway, New Jersey 08854
ANDREW WINOKUR, M.D., PH.D.
Director of Psychopharmacology
Department of Psychiatry, MG1410
University of Connecticut Health Center
10 Talcott Notch Road
Farmington, Connecticut 06032
MARY ALTEMUS, M.D.
Weill Medical College
1300 York Avenue
New York, New York 10021
BARBARA PARRY, M.D.
Department of Psychiatry
University of California, San Diego
9500 Gilman Drive
LaJolla, California 92093-0804
SUSAN THYS-JACOBS, M.D.
St. Luke's Roosevelt Hospital Center
425 W. 59th, Suite 9C
New York, New York 10019
FOOD AND DRUG ADMINISTRATION STAFF:
RICHARD CHEN, PH.D.
RUSSELL KATZ, M.D.
THOMAS LAUGHREN, M.D.
SUSAN MOLCHAN, M.D.
ROBERT TEMPLE, M.D.
ON BEHALF OF ELI LILLY AND COMPANY:
GREGORY T. BROPHY, PH.D.
EILEEN BROWN, PH.D.
JEAN ENDICOTT, PH.D.
RAJINDER JUDGE, M.D.
CATHY SHULER, M.S.
MEIR STEINER, M.D.
GARY TOLLEFSON, M.D.
SHERRY A. MARTS, PH.D.
C O N T E N T S
NDA 18-936(S), Prozac (fluoxetine hydrochloride)
ELI LILLY AND COMPANY
Indicated for the Treatment of
Premenstrual Dysphoric Disorder
AGENDA ITEM PAGE
CONFLICT OF INTEREST STATEMENT
by Dr. Sandra Titus 7
by Dr. Russell Katz 9
FDA OVERVIEW OF ISSUES
by Dr. Thomas Laughren 10
ELI LILLY AND COMPANY PRESENTATION
Introduction - by Dr. Gregory T. Brophy 15
Overview of PMDD - by Dr. Jean Endicott 17
Efficacy and Safety of Fluoxetine in PMDD -
by Dr. Rajinder Judge 31
QUESTIONS FROM THE COMMITTEE 70
OPEN PUBLIC HEARING PRESENTATION
by Dr. Sherry A. Marts 129
COMMITTEE DISCUSSION AND VOTE 134
1 P R O C E E D I N G S
2 (8:09 a.m.)
3 DR. TAMMINGA: I'd like to call this meeting to
4 order. This is a meeting of the Psychopharmacologic Drugs
5 Advisory Committee, and we've gathered to discuss an
6 application, fluoxetine hydrochloride for the treatment of
7 premenstrual dysphoric disorder.
8 First, I'd like to have everybody at the table
9 go around and introduce themselves so that the committee can
10 refresh our memory with each other. Dr. Dominguez, do you
11 want to start? I should remind people to talk directly into
12 the microphone.
13 DR. DOMINGUEZ: My name is Roberto Dominguez from
14 the University of Miami. I'm Professor of Psychiatry.
15 DR. ALTEMUS: I'm Margaret Altemus. I'm a
16 psychiatrist at Cornell Medical College.
17 DR. HAMER: I'm Robert Hamer. I'm a statistician
18 at Robert Wood Johnson Medical School.
19 DR. GELLER: Barbara Geller. I'm a child
20 psychiatrist, Washington University in St. Louis.
21 DR. THYS-JACOBS: I'm Susan Thys-Jacobs. I'm the
22 Director of the Metabolic Bone Center, St. Luke's Roosevelt
23 Hospital and Columbia University, New York.
24 DR. COOK: Ed Cook, child psychiatrist,
25 University of Chicago.
1 DR. PARRY: Barbara Parry, Professor of
2 Psychiatry, University of California, San Diego.
3 DR. TAMMINGA: I'm Carole Tamminga. I'm a
4 professor in the Department of Psychiatry at the University
5 of Maryland.
6 DR. TITUS: I'm Sandy Titus. I'm with the FDA,
7 the Advisors and Consultants Staff.
8 DR. WINOKUR: Andy Winokur. I'm professor in the
9 Department of Psychiatry at the University of Connecticut
10 Health Center.
11 DR. FYER: Abby Fyer, psychiatrist at Columbia
12 University in New York.
13 DR. CHEN: Richard Chen, statistical reviewer,
15 DR. MOLCHAN: Susan Molchan, medical reviewer,
17 DR. LAUGHREN: Tom Laughren, team leader for
18 Psychopharm at FDA.
19 DR. KATZ: Russ Katz, Director of the Division
20 of Neuropharm, FDA.
21 DR. TAMMINGA: We're waiting for our consumer
22 representative, Gaurdia Banister.
24 DR. TITUS: I'm going to read the conflict of
25 interest statement regarding this meeting.
1 The following announcement addresses the issue
2 of conflict of interest with regard to this meeting and is
3 made a part of the record to preclude even the appearance of
4 such at this meeting.
5 Based on the submitted agenda and the information
6 provided by the participants, the agency has determined that
7 all reported interests in firms regulated by the Center for
8 Drug Evaluation and Research present no potential for a
9 conflict of interest at this meeting with the following
11 A waiver has been granted to Dr. Robert Hamer.
12 A copy of this waiver statement may be obtained by submitting
13 a written request to FDA's Freedom of Information Office
14 located in room 12-A30 of the Parklawn Building.
15 In addition, we would like to disclose for the
16 record that Drs. Andrew Winokur and Carole Tamminga have
17 unrelated interests in Eli Lilly which do not constitute
18 financial interests within the meaning of the 18 U.S.C. 208(a)
19 rule, but which could create the appearance of a conflict.
20 The agency has determined, notwithstanding these interests,
21 that the interests of the government in their participation
22 outweighs the concern that the integrity of the agency's
23 programs and operations may be questioned.
24 In the event that the discussions involve any other
25 products or firms not already on the agenda for which an FDA
1 participant has a financial interest, the participants are
2 aware of the need to exclude themselves from such involvement,
3 and their exclusion will be noted for the record.
4 With respect to all other participants, we ask
5 in the interest fairness that they address any current or
6 previous financial involvement with any firm whose products
7 they may wish to comment upon.
8 DR. TAMMINGA: Dr. Katz is the Director of the
9 Neuropharmacological Drug Products Division.
10 DR. KATZ: Thank you. I'll just be very, very
11 brief. I just really want to extend my personal welcome to
12 the committee and thanks for the work you've done prior to
13 the meeting and for the work you're going to do today.
14 I particularly want to extend a welcome and thanks
15 to our three invited consultant experts, Dr. Altemus, Dr.
16 Thys-Jacobs, and Dr. Parry, who have been gracious enough to
17 come and help us out with their expertise.
18 Once again, you know we have asked you here to
19 advise us on an application for a drug to treat an indication
20 for which there are no approved treatments. So, the
21 application presents some generic problems about how to study
22 this indication as well as, we think, interesting data-specific
23 and application-specific questions.
24 So, I really just want to say thanks for coming.
25 We look forward to an interesting discussion. I'm sure it
1 will be, and with that, I'll turn it back to Dr. Tamminga.
2 DR. TAMMINGA: Thanks, Dr. Katz.
3 Dr. Laughren will begin now with the FDA overview
4 of the issues.
5 DR. LAUGHREN: Good morning. I'd also like to
6 welcome the committee back here.
7 As Dr. Katz mentioned, we're going to be focusing
8 today on this application for fluoxetine in the disorder of
9 premenstrual dysphoric disorder, but as Dr. Katz mentioned,
10 given that there are no regulatory precedents for this
11 indication, we would like to have some general discussion about
12 this entity as an indication. Following that, we will have
13 some specific questions about this application that we'd like
14 to have discussed, and finally, as always, at the end of the
15 day, we'll want you to vote on specific questions of safety
16 and effectiveness for this application.
17 Now, whenever we consider a new indication, we
18 like that indication to have some acceptance in community.
19 We like it to be reasonably well-defined, and we like there
20 to be some reasonably well-accepted diagnostic criteria.
21 Now, PMDD of course is mentioned in DSM-IV, but
22 one potential issue for discussion is the fact that rather
23 than being in the main body of DSM-IV, it's in an appendix.
24 So, we probably ought to have some discussion of what the
25 relevance of that is.
1 Secondly, as defined, PMDD has a lot of affective
2 features, and so another question that naturally comes up is
3 whether or not this is distinct from other disorders that are
4 characterized by affective symptoms, such as, for example,
5 major depressive disorder.
6 A third question that comes up is what is the
7 relationship of PMDD to the broader category of PMS. Some
8 have suggested that this is a severe subtype of PMS, and I
9 think that merits some discussion.
10 Now, one issue which is really not the focus of
11 today's meeting, but it would be useful to have some discussion
12 on, is this question of whether or not this broader category,
13 PMS, is a candidate for a new indication. The reason I ask
14 that is that, as you are well aware, there are many companies
15 who are interested in looking at this category, and so even
16 if you were to accept PMDD as a reasonable candidate for a
17 new indication, one question is whether or not this broader
18 category of PMS would be a candidate for an indication.
19 Next I want to focus on some specific questions
20 that we'd like to have addressed with regard to this
21 application. One issue has to do with the fact that the focus
22 in these studies supporting this claim was focused primarily
23 on affective symptoms as part of this larger syndrome. For
24 example, in study 19, the visual analog scale, although it
25 included 7 items, the focus was on the 3 mood items, and
1 similarly for study 22, a 16-item visual analog scale, there
2 was a focus on the mood-4. So, one question is whether or
3 not it's appropriate to focus on that subset of a larger scale.
4 In general, the issue is whether or not one should focus on
5 a subscale when one has an instrument that's focused on a
6 broader syndrome.
7 Ordinarily in psychopharm, for example, in
8 depression or schizophrenia, in choosing a primary endpoint,
9 one would focus on the total scale, such as the HAMD or the
10 PANSS or the BPARUS. So, that's another question, whether
11 or not these should be the primary outcomes in these trials.
12 Again, if one would accept those as primary
13 endpoints in those trials, what relevance, if any, would that
14 have for the way the claim should be stated? For example,
15 should the focus be rather on the total syndrome, should it
16 be on the affective symptoms of PMDD?
17 Now, another issue that comes up is the manner
18 of dosing. In this program, of course, with fluoxetine, the
19 dosing was continuous throughout the cycle. I'm sure you're
20 aware of reports in the literature for other SSRIs where, rather
21 than continuous dosing, dosing was during the luteal phase
22 and, at least from those reports, appeared to show some benefit.
23 So, the question is, what is the relevance from a regulatory
24 standpoint of these different possible dosing strategies?
25 Would that have any relevance for us in making a regulatory
1 judgment about this application?
2 Another feature of this program was the exclusion
3 of patients who were taking oral contraceptives. One can
4 certainly understand the rationale for doing that. There is
5 some literature suggesting that oral contraceptives may in
6 themselves have some benefits in the symptoms of PMDD.
7 However, The impression one gets is that the data are not
8 entirely consistent, and it may be that there's a population
9 of patients who, even though they are taking oral
10 contraceptives, still have PMDD. So, the question then is
11 whether or not fluoxetine would have any benefits in that
12 population that was excluded from these studies.
13 Similarly, one might have a question about whether
14 or not fluoxetine has been shown to be safe in a population
15 of patients taking oral contraceptives.
16 Another issue, as was the case last month with
17 PTSD, is that this program is relatively small in terms of
18 the safety exposure. Of course, there is a substantial body
19 of systematically collected data on patients taking fluoxetine
20 for other disorders, and so again, a question as to whether
21 or not one can extrapolate from that larger database to this
22 population in terms of safety.
23 Another issue that also came up at our last meeting
24 on PTSD is the question of the appropriateness of a crossover
25 trial for a chronic psychiatric disorder. Now, we had that
1 discussion last month, and everyone I think pretty much agreed
2 that for that disorder, a crossover trial would not make much
3 sense. Now, this is also a chronic disorder, but it has some
4 unusual features that may lend itself to this design. In
5 particular, there's a very predictable cyclicity with patients
6 returning to baseline during every cycle. So, again, I'd like
7 to have some discussion of whether or not that design is
8 appropriate for this drug in particular, but also in general
9 for this condition.
10 Now, this list of questions was not intended to
11 in any way limit your discussion. Clearly, if you have any
12 other issues that you think are important to discuss, please
13 bring them up. This will be helpful to us not only in reaching
14 a judgment about this application, but again, as you know,
15 there's interest more generally in developing drugs in this
16 area. So, it would be helpful to us in advising sponsors on
17 other development programs and then ultimately in making
18 judgments about applications that we expect in the future.
19 As I said, at the end of the day, we'll want you
20 to vote on these two questions. Number one, has the sponsor
21 provided evidence from more than one adequate and
22 well-controlled investigation that supports the conclusion
23 that fluoxetine is effective for the treatment of premenstrual
24 dysphoric disorder? And has the sponsor provided evidence
25 that fluoxetine is safe when used in treatment of this disorder?
1 And I'll stop there.
2 DR. TAMMINGA: Thank you very much, Dr. Laughren.
3 You've done a good job in laying out the pivotal questions
4 for the committee to consider.
5 But the next thing that we'll do is actually hear
6 from Lilly in their presentation of the data on fluoxetine
7 in the treatment of premenstrual dysphoric disorder. I'll
8 turn this over to Dr. Gregory Brophy who will take charge of
9 the Lilly presentation. Thanks.
10 DR. BROPHY: Good morning. On behalf of Eli
11 Lilly, I'd also like to welcome you and express our appreciation
12 to the committee for their contributions today, as well as
13 for allowing us the opportunity to present data substantiating
14 the safety and efficacy of fluoxetine in the treatment of
15 premenstrual dysphoric disorder.
16 As my colleagues will elaborate, PMDD is a serious
17 disorder, one that can be clearly distinguished from other
18 depressive disorders. It's also a disorder associated with
19 significant morbidity as its symptoms characteristically can
20 adversely affect the functioning, particularly the social
21 functioning, of its sufferers.
22 I'd like to introduce our two primary speakers
23 this morning. They are Dr. Jean Endicott. Dr. Endicott is
24 the Professor of Clinical Psychology within the Psychiatry
25 Department at Columbia University. She also serves as the
1 Director of the Premenstrual Evaluation Unit at
2 Columbia-Presbyterian Hospital. Jean has a longstanding
3 clinical trial and clinical experience in this area as a PMDD
4 expert. She'll focus her discussion today primarily on a lot
5 of background information on the disease itself, in particular
6 diagnostic criteria classifying as PMDD.
7 Our second presenter will be Dr. Rajinder Judge.
8 Dr. Judge is the Medical Director within Lilly Neurosciences.
9 Dr. Judge's presentation will be primarily on the clinical
10 trials themselves, particularly focused on outcome measures,
11 as well as the results of those studies, demonstrating the
12 activity of fluoxetine in this disorder.
13 I'd like to ask, if possible, that since both of
14 these presentations build upon each other, that if we could
15 hold most of the questions until the completion of Dr. Judge's
16 presentation, other than clarifying questions, I think some
17 of the questions may well be answered in Dr. Judge's
19 In addition to these two presenters, we're also
20 honored to have another PMDD expert and one of the principal
21 investigators for the largest trial that Dr. Judge will
22 describe, Dr. Meir Steiner. Dr. Steiner is Professor of
23 Psychiatry and Behavioral Neurosciences at McMaster University
24 and will also help us address questions this morning.
25 With that, let me turn the podium over to Dr.
2 DR. ENDICOTT: Today I'm going to be focusing on
3 the menstrual cycle and a condition that is exquisitely
4 entrained with phases of the menstrual cycle, both the onset
5 and the offset of the condition. The symptomatic phase is
6 during the late luteal phase of the menstrual cycle, the period
7 after ovulation. In some women, the symptoms start earlier,
8 but the most severe symptoms are seen during this premenstrual
9 or late luteal phase of the cycle.
10 After the onset of menses, the women, within a
11 couple of days, often the first day of the onset of menses,
12 become asymptomatic. The syndrome, the disorder, goes away,
13 and during particularly the mid-follicular phase of the cycle
14 up to the time of ovulation, they're essentially symptom-free.
15 This is a unique feature of this condition among the mental
17 Now, this is not a new condition. It is not
18 something that we have discovered in the 20th century. Even
19 in ancient history, there was literature that described severe
20 changes in mood behavior that occurred just prior to the onset
21 of menses. It was mentioned in early Greek literature that
22 some women had a delay of menses and that pregnancy would be
23 a cure for it, which is rather interesting.
24 By the 1930s, the term "premenstrual syndrome"
25 was coined and was used to describe problems experienced by
1 15 women, and it was described very well. The description
2 clearly fits the current diagnostic criteria by Dr. Frank in
3 the Archives of Neurology and Psychiatry. Between ancient
4 history and the 1930s, there were other mentions of severe
5 problems with mood and behavior prior to the onset of menses
6 in the medical literature, but he coined the term "premenstrual
7 tension syndrome."
8 A great deal of work was done in the 1930s, 1940s,
9 and 1950s, and by the 1950s, the term "premenstrual syndrome"
10 came into more common usage in recognition that it was not
11 just tension, that there were other dysphoric mood states
12 associated with the syndrome.
13 In 1983, NIMH convened a workshop on premenstrual
14 syndrome, and this was in recognition that a lot of
15 investigators were beginning to study the condition and were
16 interested in coming up with some guidelines to help in the
17 study of the condition. A number of different divisions within
18 NIMH sponsored this workshop, and the workshop did yield some
19 suggestions for criteria for premenstrual changes and
20 premenstrual syndrome. The major criteria was the contrast
21 between the mid-follicular phase and the late luteal phase
22 in terms of severity and the nature of the symptoms.
23 In 1987, in response to advice of an advisory
24 group, the DSM-III-R nomenclature group included specific
25 criteria for late luteal phase dysphoric disorder in the
1 appendix of DSM-III-R as a proposed diagnostic category needing
2 further study. Of great interest and particularly relevant
3 for this group is the content was almost identical to the DSM-IV
4 criteria. The requirement that there be severe, marked
5 dysphoric mood states was included, and in fact, the DSM-IV
6 criteria adds only one symptom to the possible list, and I'll
7 go into that later.
8 In the early 1990s, as a result of this, of course,
9 there was an explosion of research in the area looking at both
10 treatment of the condition and also efforts to understand the
12 By the early 1990s, the DSM-IV nomenclature
13 committee had a work group called the Premenstrual Dysphoric
14 Disorder Work Group to review literature up to that point in
15 time, and the literature review, which is included in the DSM-IV
16 source book, included literature up to 1993. The group worked
17 together and with many advisors, and there was agreement among
18 the group in their recommendations to the nomenclature
19 committee on the suggested criteria and name of the condition.
20 There was also very good agreement on the summary of the
21 evidence and the written materials that were included in the
22 DSM-IV source book.
23 There was some lack of consensus among the members
24 of the work group regarding recommendations of the placement
25 of the condition within the nomenclature. Some recommended
1 that it be in the body of the nomenclature with the criteria.
2 Others had some reservations for various reasons, and the
3 nomenclature committee decided to put PMDD in the body of the
4 nomenclature but to include the criteria in the appendix.
5 Now, how do we conceptualize PMDD currently?
6 Currently it is thought to be in the upper range of the broader
7 category of PMS. That's partially because the work has not
8 yet been done to decide whether or not there is a discontinuity
9 between the conditions. But this is somewhat similar to the
10 concept of depression. If you think of depression and general
11 minor depression and major depression, there's no clear-cut
12 pathophysiological cutoff, but most clinicians are very
13 comfortable with thinking major depression as being different
14 from minor depression or depression in general. So, currently
15 PMDD is conceptualized as being at the upper range of severity
16 of the broader category of PMS, but there are additional
17 differences. It's not just the upper range of severity.
18 In premenstrual dysphoric disorder, the mood
19 symptoms are prominent. It's the dysphoric mood symptoms that
20 are prominent and are the primary clinical complaints of the
21 women who are seeking treatment. They're not only prominent,
22 they're severe, and they include particularly irritability,
23 low mood, and anxiety.
24 There is functional impairment associated with
25 these mood symptoms. The mood symptoms themselves are
1 associated with functional impairment particularly in
2 psychosocial relationships.
3 There are physical symptoms, just as there are
4 with the garden variety PMS. Breast tenderness and bloating
5 are there.
6 The prevalence in many studies have suggested that
7 it's around 3 to 5 percent of regularly menstruating women.
8 Some recent evidence suggests it may even be higher. It may
9 be up to 8 percent. But these are women who are having regular
10 menstrual cycles.
11 The symptoms appear regularly every cycle.
12 During the week before menses, the premenstrual period, or
13 the late luteal phase of the menstrual cycle, and they remit
14 following the onset of menses.
15 Now, in contrast with the more general
16 premenstrual syndrome, the physical symptoms tend to be most
17 prominent, particularly again the breast tenderness and the
18 bloating. Mood symptoms tend to be less severe. If they're
19 there, they're no big deal. They don't bother the women that
20 much. There's little or no functional impairment associated
21 with the syndrome, and the prevalence, of course, is much
22 broader, 20 to 80 percent.
23 Now, to go over the DSM-IV criteria, I want to
24 stress a number of features.
25 First of all, this is a chronic condition. The
1 criteria require that the symptoms occur in the late luteal
2 phase of most menstrual cycles during the past year. Most
3 women who seek treatment report that they have had it for years
4 and that it has tended to get somewhat worse over time. The
5 average in several studies has been around 8 years. Also,
6 that it remit within a few days of the onset of menses, and
7 this is an important differential diagnostic point.
8 There are 11 types of symptoms or groups of
9 symptoms in the criteria. At least 5 of the 11 symptoms must
10 have been present most of the time during each symptomatic
11 phase, but at least one of those symptoms has to be one of
12 these first 4 dysphoric moods: depression, anxiety, affective
13 lability, persistent marked anger/irritability. Now, the
14 reality again is that most women who seek treatment may have
15 one primary symptom, but they tend to have all of these, not
16 just one of them.
17 The additional symptoms are decreased interest
18 in usual activities, subjective sense of difficulty in
19 concentrating, lethargy, easy fatigability, marked change in
20 appetite. The most common is increased, but some women have
21 decreased. Hypersomnia or insomnia, and the one added
22 criteria was subjective sense of being overwhelmed and out
23 of control. That's the only criteria different between LLPDD
24 and PMDD. Therefore, any woman who meets the criteria for
25 LLPDD would have met the criteria for PMDD as well. And then
1 other physical symptoms.
2 So, you can see that the emphasis in these
3 diagnostic criteria, at least five had to be present, or on
4 the dysphoric mood changes and the associated features, the
5 physical symptoms are there, but they're not a major part of
6 the criteria.
7 The criteria continue. The syndrome must
8 markedly interfere with work, school, or usual social
9 activities and relationships. It's not sufficient just to
10 have the syndrome. There should be marked impairment and
12 It should not be merely an exacerbation of the
13 symptoms of another disorder, such as major depressive
14 disorder, generalized anxiety disorder, dysthymia. So that
15 part of the criteria is that you rule out another ongoing
16 condition that could account for the symptoms.
17 And furthermore, the criteria required that the
18 diagnosis be made provisionally until it is confirmed by
19 prospective daily ratings, and those prospective daily ratings
20 have to confirm the timing of the onset and the offset of the
21 symptoms, as well as the severity of the symptoms, and the
22 impairment during at least two consecutive symptomatic cycles.
23 Now, what about the impact on functioning? How
24 is this a clinically significant syndrome or disorder?
25 First of all, a woman who develops the disorder,
1 by age 26, may experience more than 200 symptomatic cycles
2 between then and menopause, or 1,400 to 2,800 symptomatic days,
3 depending upon the duration of her premenstrual disorder.
4 As I've mentioned, in the DSM-IV criteria the symptoms are
5 severe enough to have a significant, clinically significant,
6 impact on social, home, and occupational functioning, and I'll
7 be illustrating that with some data in the next slide.
8 The social functioning is affected more than
9 vocational functioning. Many of the women manage to push
10 themselves, spend extra time, energy and effort on their
11 vocational functioning, and it's in their social functioning,
12 particularly interpersonal relationships with mate and
13 children, in which it shows itself more.
14 Women with PMDD may report impairment of family
15 and social activities at a level similar to that of depression.
16 This is illustrated in this next slide in which women with
17 major depressive disorder are compared with women with PMDD
18 on these social adjustment scale, with the self-report scale
19 developed by Myrna Weisman, in which there are a number of
20 different dimensions measured. As you can see, the women with
21 PMDD report impairment in functioning that is nearly equivalent
22 to that, and in some cases is equivalent to that, of women
23 with major depressive disorder, particularly social
24 activities, marital activities, extended family, and
1 This is very important because some people say,
2 well, it only lasts a week to 8 or 9 days, so it must not be
3 that impairing. Frankly, it is quite impairing and it occurs
4 every cycle.
5 What do we know about the etiology of PMDD? Well,
6 just as is the case with most mental disorders, we don't fully
7 understand the etiology. However, we do know some things about
9 The most likely theories are based on observations
10 of cyclic changes in ovarian steroids do cause dramatic changes
11 in brain neurotransmitter systems, a number of them, including
12 serotonin. What has been clearly established is that in women
13 sensitive or otherwise predisposed to mood instability, the
14 normal events of the ovarian cycle -- in other words, there's
15 nothing wrong with the menstrual cycle -- the normal events
16 of the ovarian cycle may trigger severe mood changes. And
17 I'll be reporting some other information on that topic. So,
18 this is one thing that has been clearly established. The exact
19 mechanism the way the neurotransmitter systems are involved
20 is not as clearly established, but a great deal of work has
21 gone on and is currently going on in this area.
22 How is PMDD distinct from the other depressive
23 disorders, particularly major depression and dysthymia?
24 Well, first of all, the mood disturbance is
25 cyclical. It is very tightly linked to phases of the menstrual
1 cycle. It has a highly predictable onset and offset, not only
2 by phases of the cycle, but within an individual woman, you
3 will find that her onset and offset, relative to her circulating
4 gonadal hormones, is very tightly linked and very consistent
5 from cycle to cycle.
6 The most common chief complaint is irritability.
7 Although the other symptoms may be there, the women who seek
8 treatment tend to focus on irritability.
9 The cyclic occurrence of these symptoms cease
10 during pregnancy and post-menopause. This is not the case
11 with either major depression or dysthymia or the anxiety
13 Prevention or suppression of cycling gonadal
14 hormones relieves the symptoms. Again, this is not the case
15 with the other depressive and anxiety disorders.
16 Furthermore, hormone replacement therapy can
17 provoke cyclic dysphoric changes in women who have a history
18 of PMDD. This has been done in double-blind studies and is
19 clearly established. This does not happen in women who have
20 a history of major depression or dysthymia.
21 The HPA axis functions normally in PMDD. There's
22 no evidence that the HPA axis is abnormal in any way, and this
23 is unlike the documented disturbances in major depression.
24 There is great symptom stability seen across
25 cycles. Again, this is in some contrast with studies across
1 episodes of major depression in which there is somewhat less
2 symptom stability. Here the symptom stability is very stable
3 and very predictable for the individual woman.
4 And most important, recently in 1998, Ken Kendler
5 published the results of a very large study of twins, comparing
6 monozygotic and dizygotic twins, in which both premenstrual
7 related symptoms, focusing mainly on depression, and lifetime
8 major depression had been evaluated at least at two points
9 in time. What he found was that both the genetic and
10 environmental risk factors for these two conditions were not
11 closely related. They were not shared.
12 There was a large genetic contribution for
13 premenstrual mood changes but that was not accounted for by
14 major depression, lifetime major depression, and this was a
15 very important study in this area.
16 There are some other ways in which PMDD is distinct
17 from the other depressive disorders. It's most likely to
18 respond to the serotonergic antidepressants than to other
19 antidepressants. As you know, in the comparison studies
20 between the TCAs and the SSRIs, with major depression you don't
21 find that distinction. It's a clear distinction here. The
22 serotonergic antidepressants are superior.
23 Furthermore, upon treatment, the symptom
24 improvement in PMDD is very rapid, as shown within the first
25 treatment cycle, even though the women have not been on the
1 medication that long. This is in contrast with major
2 depression and dysthymia.
3 The physical symptoms shown with women with PMDD
4 are unique to that condition. Breast tenderness and bloating
5 are the most common. This is rarely seen in women with simple
6 dysthymia or major depression.
7 Upon treatment cessation, the symptoms return
8 rapidly, and the reemergence is more predictable. It's quite
9 predictable with PMDD. There have been a number of studies,
10 two of which are summarized here, about the reemergence of
11 symptoms after stopping treatment.
12 Dr. Pearlstein in 1994 published an article on
13 after 1 year of successful fluoxetine treatment, 31 women,
14 they discontinued treatment, and the PMDD symptoms, meeting
15 criteria for PMDD, returned within two cycles in 30 of the
16 31 women.
17 Kimberly Yonkers did a study published in 1997
18 in which there was double-blind randomization from sertraline
19 to placebo in women who had been on the medication 3, 6, or
20 9 cycles. So, the women did not know when placebo was going
21 to be instituted. The rates of recurrence were 66, 66, and
22 60 percent within a couple of cycles after cessation of the
23 active compound.
24 Now, in recognition of the clinical significance
25 of this condition and of the need to find effective treatments
1 for it, a very large number of compounds and interventions
2 have been studied. This is just a sampling. This is not
3 exhaustive. Other compounds have also been studied.
4 The most work has been done with the SSRIs, and
5 this is shown here in which there are 32 studies, published
6 studies, with SSRIs. The greatest number are with fluoxetine,
7 but there have been published studies -- the double-blind,
8 placebo-controlled studies are in the dark blue and the
9 open-label trials are in the light blue.
10 31 of these 32 studies were successful, were
11 effective. There was a single study with fluvoxamine in which
12 there was no difference. There are some other issues about
13 that study, but 31 out of 32 studies of SSRIs have shown the
14 SSRIs to be effective in the treatment of PMDD.
15 So, in conclusion, PMDD appears to be a distinct
16 clinical entity with exquisite onset and offset of timing and
17 clinical features and other characteristics that occurs in
18 3 to 5 percent of menstruating women and maybe even more.
19 It has clinical and biological profiles that differ from those
20 of major depression. It is a severe form, we think now, of
21 the broader category of PMS that impacts normal functioning
22 to a clinically significant degree.
23 It should be better diagnosed and treated. There
24 are plenty of women who have not had the diagnosis made. There
25 is currently no registered treatment in the U.S. for PMDD.
1 And there is an unmet clinical need for safe and effective
2 treatment for the psychological as well as the physical
3 symptoms of PMDD. There is evidence that the SSRIs meet this
4 need, and Dr. Judge will be presenting that data now.
5 Thank you.
6 DR. TAMMINGA: I'd just like to remind the
7 committee that all the slides that are shown are in the navy
8 book in front of you.
9 DR. JUDGE: Well, good morning. It's my pleasure
10 to present to the advisory committee and to the FDA this
12 As you heard from Dr. Endicott, PMDD is a disorder
13 which causes suffering to many, many American women. The data
14 I will present this morning on fluoxetine will show how highly
15 effective fluoxetine is in treating the symptoms of PMDD.
16 Firstly, I will address the efficacy with respect
17 to the PMDD studies, and I will focus on the key symptoms of
18 premenstrual dysphoric disorder, i.e., the mood symptoms, the
19 physical symptoms, and the social impairment that accompanies
20 these symptoms.
21 Secondly, as you know, fluoxetine is a drug which
22 has been marketed for over 10 years, and the safety profile
23 is very well established. I will, therefore, provide a
24 succinct summary of the safety and importantly compare that
25 to the overall fluoxetine safety database.
1 And finally, I will provide conclusions and dosing
3 These slides show the listing of the published
4 studies in PMDD for fluoxetine, firstly, the double-blind
5 studies on the left and the open-label studies on the right.
6 The first three studies here comprise the
7 application for fluoxetine in PMDD. Although these comprise
8 the application, all of the studies in the literature are
9 consistent with respect to the results for fluoxetine in
10 efficacy and safety. They have all utilized the DSM-III-R
11 criteria for LLPDD. As you heard from Dr. Endicott, as these
12 patients conform to DSM-III-R, that means that they also
13 conform to DSM-IV criteria.
14 Furthermore, all of these studies in the main
15 utilized a dose of 20 milligrams daily, and that was considered
16 an effective and safe dose for patients with PMDD.
17 Although these studies are open-label, there is
18 some nice information that can be obtained from them, namely,
19 for patients going out to longer than 6 months, as for the
20 shorter-term studies, patients going out to even up to 20 months
21 did show a maintenance of efficacy with fluoxetine.
22 Furthermore, there was also evidence from these studies to
23 suggest that when fluoxetine was stopped, even after the long
24 term, there was very quickly a reemergence of symptoms
25 following cessation of treatment.
1 Three trials, as I've indicated, comprise the
2 application for fluoxetine in PMDD, and these are listed here
3 below in more detail. These are studies C019, X022, and X037.
4 For purposes of perhaps ease of communication, I will refer
5 to these studies as studies 1, 2, and 3. All were double-blind,
6 parallel-controlled. One was a crossover trial.
7 The efficacy measures utilized in these studies
8 are listed here and spelled out in full here. For the first
9 study, number 1, the visual analog 7-item scale was utilized
10 as the primary outcome measure. For study number 2, X022,
11 a 16-item visual analog scale was utilized as the primary
12 outcome measure. For the third study, X037, an overall measure
13 of improvement, the clinical global impression, was utilized
14 as the primary outcome measure.
15 In addition to the ones I've just indicated, there
16 were a number of other scales utilized in these studies,
17 particularly in studies 1 and 2, the premenstrual tension
18 syndrome. Both patient rated and clinician rated tools were
19 also used in these studies.
20 There are a wide variety of scales utilized here.
21 There is not a gold standard of scales that is currently
22 utilized in PMDD studies, but all of the scales here are
23 appropriate and are reliable in treatment and study of PMDD.
24 I'll just go into a little bit more detail. These
25 slides list the scales that were used in these studies, the
1 main scales across the top, and across here, down here, are
2 the DSM-IV criteria for mood, for physical symptoms, and social
3 impairment. The numbers listed here list the items of these
4 scales which correspond to each of these symptoms as listed
5 by DSM-IV. This shows that all of the scales used in these
6 studies did employ items that correspond to the DSM-IV
8 So, for example, if we look at the premenstrual
9 tension syndrome scale, both the clinician rated and the
10 patient rated, listed here are items that are part of these
11 scales and that correspond to the mood symptoms of DSM-IV,
12 as listed in DSM-IV, and then over here they also contain items
13 which list physical symptoms and they also contain items which
14 list social impairment.
15 For the primary outcome variable in study 1, visual
16 analog scale-7, again the items in this scale do correspond
17 to the mood symptoms of PMDD. It also contains items
18 corresponding to the physical symptoms of PMDD. It did not
19 contain items corresponding to social impairment, but for that
20 study, the PMTS scales were utilized. So, we can glean social
21 impairment information from those scales.
22 With respect to the second study, the visual analog
23 scale 16-item was used, and this scale contained items which
24 corresponded to all of the symptoms as listed by DSM-IV, i.e.,
25 the mood symptoms, the physical symptoms, and the social
1 impairment symptoms.
2 So, all of these scales utilized are appropriate
3 and reliable to measure treatment change as listed for the
4 core symptoms for PMDD.
5 Going on to the studies for PMDD, this slide lists
6 the inclusion and the exclusion criteria for these studies.
7 First of all, the studies obviously included females 18 years
8 and over, and they had regular menstrual cycles.
9 All the patients did conform to a DSM-III-R
10 diagnosis of late luteal phase dysphoric disorder, and as you
11 heard from Dr. Endicott, as they conform to the DSM-III-R,
12 they therefore conform to DSM-IV criteria.
13 They also had to have an adequate method of birth
14 control other than hormonal. I'll make another comment on
15 that a little bit later.
16 And they also had to meet criteria for protocol
17 predefined symptom severity. For example, in study 1,
18 patients had to exhibit during the prospective cycles, during
19 which they were monitored for this baseline state, either at
20 least a 50 percent change in the core items for the mood items
21 for the visual analog scale, a 50 percent increase from
22 follicular to luteal phase, or they could exhibit, for example,
23 a 100 percent increase or more in just one of those items
24 corresponding to the mood scales.
25 The exclusion criteria. Patients were excluded
1 if they had serious health problems, and they were also excluded
2 if they were on the following medications: any psychotropic,
3 diuretic, or hormonal medication, including oral
4 contraceptives. As you've heard and just to reiterate the
5 point, it is essential to quite clearly delineate the effects
6 of fluoxetine on PMDD. As you've heard, oral contraceptives
7 can have some effect on PMDD symptoms. There's a variety of
8 literature which shows an inconsistent and variable effect
9 on PMDD symptoms, perhaps most often the physical symptoms,
10 and for that reason, rather than introduce another variable
11 into the study, it was felt prudent to exclude oral
13 Also, patients with concurrent Axis I diagnosis
14 of other disorders were excluded as appropriate.
15 Going on to these studies in more depth now, the
16 reference here on the corner of each slide indicates the study
17 to which this refers. Study 1, C019.
18 This is the first study, study 1, C019. This is
19 a double-blind, placebo-controlled, dose range-finding study.
20 After the screening period here with two cycles, patients
21 then entered a placebo single-blind period here, and this
22 provided an adequate basis for prospective monitoring for the
23 patients and adequate baseline measurements of symptoms.
24 At this point, patients who still met the DSM-III-R
25 criteria for PMDD and importantly excluded placebo responders,
1 patients were then randomized in a double-blind fashion at
2 this point to receive either fluoxetine 20 milligrams a day,
3 fluoxetine 60 milligrams a day, or placebo. For those patients
4 who received 60 milligrams a day, they were put on 60 milligrams
5 a day from day 1, straight off the bat. They did not have
6 the ability to titrate up to this dose; 60 milligrams a day
7 from day 1.
8 The study then continued for 6 treatment cycles,
9 making this a long-term study.
10 Patients were seen during each cycle twice, once
11 during the follicular phase and once during the luteal phase.
12 The primary objective of this study, as you heard
13 from the first speaker, was to assess the efficacy of fluoxetine
14 in PMDD as measured by the luteal phase Mood-3 average of the
15 visual analog scale -- and I will go into this in a little
16 more depth later for clarity -- average change from mean
17 baseline to mean treatment score.
18 Now, originally in the protocol, it did not specify
19 the VAS Mood-3 specifically. It was enlisted as just the
20 visual analog scale. As the study started, Lilly and the
21 primary investigator for this study made an agreement that
22 the most appropriate outcome measure for this protocol should
23 be the VAS Mood-3. That was decided upon and confirmed in
24 writing before the completion of the study, just after the
25 study had started in fact.
1 In addition to the primary, obviously I will show
2 you items, the second objectives of the study, further
3 measurements for the efficacy of fluoxetine in PMDD pertaining
4 to the symptom clusters for the mood items to the physical
5 items and social impairment as measured by the visual analog
6 scale and also as measured by the subtotals of the premenstrual
7 tension syndrome rating scales, patient rated and physician
8 rated, obviously, also an opportunity to assess the safety
9 and tolerability of fluoxetine in PMDD.
10 This slide here shows the visual analog scale that
11 was utilized in this study. Patients were asked to rate
12 themselves on a scale 0 to 100, ranging from no symptomatology
13 to extreme symptomatology here. The items in yellow comprise
14 the core items for mood and, therefore, the primary efficacy
15 analysis for this study. So, the primary items for mood here
16 are item 1, calm and unruffled, going to tense, uptight, uneasy;
17 number 2, happy, content, and energetic, going to extremely
18 depressed, sad, apathetic, and lethargic. Item 7 measured
19 irritability. There were three physical items score here:
20 headache, bloating and tenderness, and breast tenderness.
21 And item 4 looked at emotional lability, even-tempered to
22 extreme mood swings. So, that's the visual analog scale
24 So, the primary efficacy variable is the average
25 of the three mood symptoms here highlighted in yellow: the
1 average scores of dysphoria, irritability, and tension. And
2 secondary efficacy variables included the VAS Mood-4 average,
3 which incorporated the other emotional lability item here,
4 also the average of the physical symptoms, and then the
5 subtotals for mood, physical, and social impairment for the
6 PMTS scale.
7 This just shows in depth the PMTS scales for
8 purposes of clarity. The clinician rating scale is listed
9 on the left on both slides, and the corresponding items of
10 the patient rating scale, PMTS-P are listed on the right. So,
11 overall the range for both scales is 0 to 36. For the clinician
12 rating scale, there were 10 items which were scored from 0
13 to 4 for most of the items, apart from number 7 and 8 where
14 the items are scored from 0 to 2. The corresponding patient
15 rating scale simply asked the patient to respond a yes or no
16 to each question. Again, the items here correspond to those
17 items as per the clinician rating scale.
18 This is looking at the calculation of the efficacy
19 measures in a little bit more depth. This is a pictorial
20 representation of the follicular and luteal cycles in this
21 study. The first two cycles are the baseline placebo cycles,
22 and then the six studies are the six treatment cycles. F is
23 follicular; L is luteal. As I indicated, patients were seen
24 twice during each cycle, once in the follicular phase, once
25 in the luteal phase, and at those visits patients were assessed
1 in terms of their efficacy.
2 So, measurement of the average luteal scores here
3 for these two placebo cycles provided the mean baseline score.
4 The average of the luteal scores for these six cycles here
5 provided then the mean treatment score, and the calculation
6 of the overall efficacy measure was the mean treatment score
7 minus the mean baseline score.
8 Now, originally Lilly did plan to analyze the
9 percent change in the analysis plan. However, the percent
10 change would have assumed a normality assumption. There were
11 extreme outliers, which violated the normality assumption.
12 And therefore, it was felt appropriate to look at the mean
13 treatment change.
14 Going on to some characteristics of the patients
15 in the study, these are the baseline characteristics, and these
16 are listed in more detail in your briefing document supplied
17 to you. But essentially the age of entry for these patients
18 in these studies was mid to late 30s. Importantly for the
19 demographic variables listed here and also in your briefing
20 document, there were no differences in the groups at baseline.
21 The average VAS Mood-3 follicular and luteal
22 scores are represented here more visually. Importantly all
23 three treatment groups, with respect to their scores, are
24 similar at baseline. Moreover, as one would expect for PMDD,
25 the luteal scores are higher. This is the mean score on the
1 visual analog scale, VAS Mood-3. The luteal scores are higher
2 than the follicular scores, the follicular scores indicating
3 insignificant symptomatology, as one would expect with
4 patients with PMDD.
5 This lists the patient disposition for the study
6 with respect to the percentage of patients. Overall,
7 fluoxetine 20 milligram patients were the highest number of
8 patients who completed the study. The highest percentage of
9 patients completed the study were on fluoxetine 20 milligrams.
10 In terms of patients who dropped out for any
11 reason, these are shown here. For the patients who dropped
12 out due to an adverse event, more patients on fluoxetine 60
13 milligrams who dropped out due to an adverse events. There
14 were a low level and similar level for dropouts with respect
15 to placebo group and fluoxetine 20 milligrams. For lack of
16 efficacy, as one would expect, a higher proportion of placebo
17 patients dropped out due to lack of efficacy.
18 I'm going on to now show the efficacy measures
19 by means of a series of bar graphs. In all of these graphs,
20 fluoxetine 20 milligrams will be shown as orange, fluoxetine
21 60 milligrams will be shown as yellow, and placebo in green.
22 Moving on to the primary efficacy measures -- and,
23 again, I will concentrate on the mood symptoms, then the
24 physical symptoms, then the social impairment symptoms from
25 each study.
1 First of all, the mood symptoms in the luteal
2 phase. This looks at the mean reduction from baseline to mean
3 treatment here, so the greater the reduction, the greater
4 improvement in overall outcome.
5 This is the primary objective here, the VAS Mood-3.
6 We see here there's a greater reduction with statistical
7 significance for both fluoxetine 20 milligrams and 60
8 milligrams versus placebo here. If one looks at the individual
9 items that comprised the primary outcome, VAS Mood-3, which
10 is dysphoria, irritability, and tension, then one sees that
11 indeed in each case fluoxetine 20 and 60 milligrams are
12 statistically significantly superior in their reduction of
13 symptomatology versus placebo in each case.
14 There does appear to be some numerical superiority
15 for fluoxetine 60 milligrams versus 20, but the difference
16 between the two groups was not statistically significant with
17 respect to the two fluoxetine groups.
18 The results here are mirrored by the consideration
19 of the results seen on the PMTS scales, both the PMTS-P, the
20 patient rated scale, and the PMTS-C, the clinician rating
21 scale. Again, showing the reduction from mean baseline,
22 patients on fluoxetine on any dose, either 20 or 60, achieved
23 superior clinical improvement versus placebo, and the
24 difference between the active treatment groups and placebo
25 did attain statistical significance. Again, some numerical
1 superiority observed with fluoxetine 60 milligrams versus
2 fluoxetine 20 milligrams, but the difference between the two
3 fluoxetine arms was not statistically significant in this case.
4 Moving on to the physical symptoms, again this
5 looking at the visual analog scale, the overall physical
6 average is shown here, again mirroring the mood symptoms, a
7 statistical difference for superiority for the fluoxetine arms
8 versus placebo. Then if one looks at the individual physical
9 items which comprise the physical average, bloating, breast
10 tenderness, and headache, one sees that it is the effects of
11 breast tenderness and bloating which lead to the overall
12 significance. There does not seem to be any difference between
13 the groups with respect to headache. But as you heard earlier,
14 bloating and breast tenderness are two of the most common
15 symptoms in patients with PMDD.
16 Again, the effective results for fluoxetine in
17 the mood symptoms and the physical symptoms here are also
18 mirrored by consideration of the PMTS subtotals, for the PMTS-P
19 and the PMTS-C. Again, a significant reduction for physical
20 symptoms for both fluoxetine arms versus placebo, and again
21 some evidence of numerical superiority with fluoxetine 60
22 milligrams versus 20, but the differences were not
23 statistically significant.
24 Moving on to the social impairment. As I noted
25 earlier, the visual analog scale from this study did not measure
1 social impairment, and so we view the items from the PMTS-P
2 with respect to social impairment. Again, reduction from mean
3 baseline for the PMTS-P and the PMTS scores showing a very
4 nice improvement in social impairment for patients on
5 fluoxetine 20 and 60 milligrams versus placebo, the difference
6 between the active treatment arms versus placebo attaining
7 statistical significance.
8 So, I've shown you the subtotal scores for the
9 mood and physical symptoms and the social impairment. I just
10 want to point out now that analysis of the overall scores for
11 each of these measures, the overall visual analog scale 7-item,
12 the overall PMTS-P, the overall PMTS-C. Also I showed that
13 fluoxetine was statistically superior with respect to its
14 effects on those scores versus placebo.
15 So, efficacy was seen for both fluoxetine 20
16 milligrams and 60 milligrams for all of the symptom clusters
17 of PMDD.
18 Two pertinent questions at this point. How
19 quickly was the efficacy apparent and what was the course of
20 the treatment effect?
21 With respect to how quickly was the efficacy
22 apparent, we viewed here the efficacy seen with respect to
23 the mood symptoms and the physical symptoms at the first
24 treatment cycle. So, remember, patients were asked to take
25 medication from the first day of their menses. So, this is
1 just after a couple of weeks of treatment. We see that even
2 at the first cycle, there is superiority for fluoxetine versus
3 placebo in the mood symptoms and in the physical symptoms as
4 shown here by the primary analysis of Mood-3 average and also
5 the physical average on the visual analog scale. So, a very
6 quick response to fluoxetine was evident.
7 With respect to the course of the treatment effect,
8 this is shown here for the last observation carried forward
9 for the primary analysis, the VAS Mood-3, placebo here, this
10 line; fluoxetine 20 milligrams, the orange line here; and
11 fluoxetine 60 milligrams, the yellow line here. So, this is
12 looking at the mean reduction from baseline, and what we see
13 is that up to 6 months, at each cycle, there is a statistical
14 difference maintained between placebo and both of the
15 fluoxetine groups, both the 20 and the 60 milligram groups,
16 showing that the efficacy of fluoxetine is maintained for out
17 to 6 months.
18 So, with respect to the conclusions in the study,
19 both fluoxetine 20 and 60 milligrams a day were effective in
20 the treatment of PMDD. Statistical differences were shown
21 with respect to placebo, with respect to the primary objective,
22 the VAS Mood-3, and the secondary objectives, and I also
23 indicate there's also the consideration of the total scores
24 as well. Efficacy was seen in all of the symptom clusters
25 of PMDD. So, although mood was defined as the primary outcome
1 measure, it's also interesting to note that the mood symptoms,
2 the physical symptoms, and the social impairment associated
3 with PMDD all improved very quickly. Efficacy was
4 demonstrated in the first treatment cycle and maintained for
5 up to 6 months. There was some evidence that fluoxetine 60
6 milligrams was in general numerically greater than 20
7 milligrams, but the differences were not usually statistically
9 Moving on to the next study, this is study number
10 2, X022, and this is a double-blind crossover study. As was
11 alluded to earlier, the disorder of PMDD comprises symptoms
12 which are very closely entrained to the menstrual cycle. So,
13 the predictable nature of these symptoms emerging cycle after
14 cycle after cycle makes it a very predictable disorder with
15 discrete episodes of disorder. Furthermore, studies would
16 suggest that there is symptom stability across cycles. So,
17 symptom stability being the rule rather than exception. So,
18 these two characteristics of PMDD do make it an ideal disorder
19 to study in a crossover design. This is also evidenced by
20 the literature where a number of studies with various
21 treatments have used the crossover design in order to study
23 So, in this study after 3 cycles of screening and
24 evaluation, patients were entered into this study. Patients
25 were either randomized to the fluoxetine arm or the placebo
1 arm for 3 cycles. After this there was a 1-cycle crossover,
2 1-month crossover, and then patients were crossed over to the
3 other treatment, again for another 3 cycles here. For the
4 fluoxetine arm, patients were started on 20 milligrams, and
5 investigators, at their discretion, could titrate up in
6 increments of either 10 or 20 milligrams, according to safety
7 and efficacy, to a maximum of 60 milligrams.
8 Patients who entered here are listed here: 9 for
9 the fluoxetine group, and placebo, 10 patients. Obviously,
10 each patient acted as their own control. This enhanced
11 sensitivity allows for relatively fewer patients.
12 Originally this protocol was intended to recruit
13 30 patients, but in an earlier analysis done for purposes of
14 a scientific abstract, the investigator noted significant
15 differences between the treatment groups and elected to stop
16 the study at that point. It's important to realize that all
17 of the patients who were recruited at that time were allowed
18 to finish. That numbered a total of 19 patients. And
19 moreover, the raters who were assessing the patients and the
20 patients themselves, who obviously were assessing themselves
21 on scales remained blind to treatment assignment, first, to
22 minimize any kind of bias in this study.
23 The prime objective here was to assess the efficacy
24 in the treatment of PMDD as measured by the average within-cycle
25 change from follicular to luteal phase in the VAS Mood-4.
1 So, there are differences here with respect to the outcome
2 measures from the first study. Just to point out that patients
3 did score themselves every day in this, they did do some
4 measurements every day during this study, and they did some
5 measurements again for every visit. Again, they were seen
6 for two visits each cycle, follicular and luteal phase.
7 So, I'm going to talk about this in a little bit
8 more depth, but just to emphasize that the outcome measure
9 here was the average within-cycle change from follicular to
10 luteal phase, in the VAS Mood-4 subtotal. So, this is a 16-item
11 VAS and the VAS Mood-4 subtotal comprised the primary efficacy
12 outcome, and that comprised the mood swings, depression,
13 irritability, and anxiety.
14 Again, the secondary objectives were obviously
15 to look at the other items of the visual analog scale and the
16 PMTS scales with respect to the other subtotals, the mood
17 subtotal, the physical symptom subtotal, and the social
19 This slide shows the visual analog scale used in
20 this study. So, this is a 16-item visual analog scale.
21 Patients were asked to rate themselves from no symptoms to
22 severe symptoms, and the symptoms listed in yellow again
23 comprise the primary mood items, rapidly changing mood to mood
24 very stable; item number 8, most sad ever to most happy ever;
25 the irritability item, and the most anxious ever to most calm
2 The other items that comprise physical symptom
3 items are shown here: number 5, extreme breast pain; extreme
4 bloating; and extreme physical discomfort. You see all the
5 other items that were also in the scale.
6 This shows the daily rating form which also
7 comprised one of the secondary scales in the study. This form,
8 obviously as the name implies, was rated daily by the patient
9 and the patient rated the severity of each item on a scale
10 of 1, none, to 6, extreme, the total score ranging up to 108.
11 Listed here are those items which pertain to mood, the physical
12 symptoms, social impairment, and there are a variety of other
13 symptoms which were also scored on this daily rating form.
14 Now, just to go into the depth of how the efficacy
15 analysis was calculated, again just to reiterate the primary
16 outcome variable was the VAS Mood-4 subtotal. This is how
17 this was collected.
18 The luteal score was the average of the patient's
19 score for 7 days prior to the menses. The follicular score
20 was the average of the patient's score over the 7 days
21 post-menses. Subtracting one from the other provided the
22 within-cycle change, and this was averaged over 3 months of
23 treatment to provide the primary patient treatment outcome
25 Baseline characteristics, very similar to the
1 first study. Most of the patients were caucasian and the age
2 of these patients were similar to the first study, mid to late
3 30s. Importantly, consideration of the follicular scores here
4 for the PMTS patient total for the PMTS clinician total, and
5 for example, in the Beck's Depression Inventory, if you look
6 here, the scores are very, very low in the follicular phase,
7 indicating an absence of any significant premenstrual
8 symptomatology, as one would expect with respect to the
9 cyclicity of PMDD.
10 With respect to patient disposition, the majority
11 of patients completed this study. Very few patients dropped
12 out for any reason at all.
13 Again, I will go through the efficacy outcomes
14 with respect to the mood symptoms, the physical symptoms, and
15 then the social impairment.
16 Firstly, with respect to the mood symptoms. Now,
17 here the scale is looking at average within-cycle increase.
18 So, within-cycle increase from follicular to luteal, so
19 indicating an increase in symptomatology. So, an increase
20 in scores here would indicate an increase in symptomatology
21 and therefore deterioration in the patient's outcome.
22 For the primary outcome measure, the VAS Mood-4,
23 the 4 items on the visual analog scale, you will note there's
24 a greater increase with respect to the symptomatology seen
25 in the placebo patients shown here in green, and this increase
1 was statistically superior than the increase evident for
2 fluoxetine patients. So, the fluoxetine patients improved
3 with statistical superiority versus the placebo patients.
4 When one sees the individual items which comprise
5 the VAS Mood-4 items, again fluoxetine is superior with respect
6 to placebo in each of these items, the mood swings, depression,
7 irritability, and anxiety. And fluoxetine patients exhibited
8 far less increase in symptomatology versus the placebo
10 This was mirrored by consideration of the
11 secondary outcome variables, the daily rating form, and the
12 PMTS-P and PMTS-C. For both the DRF and PMTS-C -- that's the
13 clinician rating and patient rating -- again, evidence of
14 fluoxetine superiority with statistical significance versus
15 placebo. For the PMTS-P, quite clearly there is fluoxetine
16 superiority, but the differences did not attain statistical
18 Going on to the physical symptoms with respect
19 to this study, again shown here are the physical average for
20 the symptoms comprising from the visual analog scale. This
21 shows the individual items which made up this physical average,
22 and fluoxetine is highly effective with respect to placebo
23 for breast pain, bloating, and physical discomfort, giving
24 an overall highly statistically significant effect versus
25 placebo on the visual analog scale.
1 Again, consideration of the secondary measures
2 employed in this study further mirrored the evidence seen for
3 the primary outcome measure in that for the physical symptoms,
4 for the daily rating form completed by the patient, the PMTS-P
5 completed by the patient, the PMTS-C completed by the
6 clinician, statistical superiority for fluoxetine versus
7 placebo in each case.
8 The similar results are evidence for social
9 impairment with highly statistical significance for fluoxetine
10 versus placebo with respect to the visual analog scale here,
11 the overall social impairment. This comprised two items, work
12 efficiency and social activity. So, this is important. It
13 shows that the patients rates themselves as improving with
14 fluoxetine with respect to their efficiency at work and their
15 social activities.
16 Again, this is mirrored by the consideration of
17 the secondary outcome variable. Patients rated themselves
18 as improving with statistical significance over placebo for
19 the daily rating form, and for PMTS-P and PMTS-C, again quite
20 clearly there is numerical superiority for fluoxetine, but
21 the differences did not attain statistical significance here.
22 Going on to the same questions as we asked in the
23 first study, when was the efficacy apparent and what was the
24 course of the treatment effect, these two slides go to show
25 some evidence for those questions.
1 First of all, again, entirely consistent with the
2 first study, a consideration of the first treatment cycle,
3 within-cycle increase, showed that for the primary VAS Mood-4
4 for the physical symptoms and for the social symptoms, all
5 from the visual analog scale, at the first treatment cycle
6 was exhibited a superiority for fluoxetine versus placebo,
7 again entirely consistent with the first study.
8 The course of treatment effect is shown here just
9 for one measure, the VAS Mood-4, which is obviously the primary
10 outcome measure. For patients here for the purple, here we
11 see here for the patients who started off the treatment with
12 placebo. Now, scores higher on this graph indicate increase
13 in symptomatology. Scores in the lower half of this graph
14 indicate lower symptomatology. So, higher scores are
15 considered bad for the patient.
16 So, for the patients who started off on placebo,
17 as indicated for the first 3 cycles, their scores are in the
18 upper half of the graph, indicating significant symptomatology
19 for these patients. After the crossover -- you see then the
20 patients who then were crossed over to fluoxetine -- then their
21 scores for the next 3 cycles were in the lower half of this
22 graph, indicating improvement for the patients.
23 Exactly the opposite is evidenced for the opposing
24 group. We see here for fluoxetine, the patients who started
25 off on fluoxetine, their scores for each of these 3 cycles
1 are in the lower half of the graph, indicating very little
2 symptomatology for these patients, and after the crossover,
3 when they were switched over to placebo, we see that their
4 scores shoot up to the higher portion of the graph, indicating
5 an increase in symptomatology. So, a nice visual
6 representation of the comparative effects of fluoxetine versus
8 Now, the crossover washout phase here was a 1-cycle
9 duration, 1 month. As we appreciate, for fluoxetine, the
10 half-life is relatively long and also contains an active
11 metabolite, norfluoxetine. So, therefore, a reasonable
12 question at this point is, were there any carryover effects,
13 and if there were any carryover effects, what was the
14 implication of that carryover effect with respect to the
15 overall efficacy seen in this study? I'd like to elaborate
16 on those results.
17 First of all, with respect to the mood symptoms
18 across the treatment cycles for this study. This is just
19 looking at the VAS Mood-4 and another example of mood, the
20 DRF mood subtotal. This shows that for the overall treatment
21 effect for all cycles shown here, the overall value for
22 fluoxetine versus placebo, was highly statistically
23 significant, p, 0.002. This column here shows the possibility
24 of the carryover effect. As you see over here with the p values
25 of 0.9 and .26, there is no carryover effect evident.
1 When one then moves on to the first treatment
2 cycle, the results from the first treatment cycle only, we
3 see here again for the same items the VAS Mood-4 and the DRF
4 Mood subtotal. We see, if we look over into the carryover
5 effect column here, .09, .12, then there is a suggestion of
6 a carryover effect. But it's worth bearing in mind, again
7 just to emphasize, we're looking at the within-change from
8 follicular to luteal. So, in actual fact, a carryover effect
9 present here would actually bias against fluoxetine.
10 So, in spite of that bias, when we look at the
11 overall p value for the treatment effect at cycle 1, we see,
12 in spite of the carryover effect, which is biased negatively
13 versus fluoxetine, the differences between fluoxetine and
14 placebo still attain statistical significance.
15 Just to show the robustness of the scores in
16 another manner, I'm now going to focus on looking at the first
17 3 cycles here. I'm going to show you results from just the
18 first 3 cycles which really approximate to a parallel group
20 This is looking at the so-called first period
21 analysis. That's the analysis from the first 3 cycles of that
22 study. This is looking at a variety of measures with respect
23 to the moods on the left side and respect to the physical
24 symptoms on the right side.
25 For VAS Mood-4 subtotal, for DRF Mood subtotal,
1 for PMTS-P subtotal, for PMTS-C subtotal, overall, whichever
2 way you look at it, even in the first period analysis only,
3 statistical significance is for fluoxetine versus placebo.
4 And the same is evident for the physical symptoms. Again,
5 just looking at the first period only, statistical significance
6 is for fluoxetine versus placebo.
7 So, in conclusion for efficacy in the study, a
8 flexible dosing for fluoxetine in the range of 20 to 60
9 milligrams a day -- and the patients attained a mean dose of
10 27 milligrams in this study -- was effective in the treatment
11 of PMDD. Again, we saw statistical differences superior to
12 placebo with respect to the primary objective, the VAS Mood-4,
13 and importantly also the secondary objectives. Efficacy was
14 seen in the symptom clusters of PMDD for most variables with
15 respect to mood, physical symptoms, and social impairment.
16 Just an overall analysis of the total scores for the visual
17 analog scale 16-item, again, statistical differences for
18 fluoxetine versus placebo.
19 Improvement, as demonstrated in the first study,
20 was demonstrated in the first cycle and maintained for up to
21 3 months.
22 So, thus far, I've presented two well-designed,
23 randomized, placebo-controlled studies that have shown
24 fluoxetine is statistically significantly superior to placebo
25 in the treatment of PMDD.
1 Moving on to the third study, this is X037, study
2 number 3. This is a placebo-controlled, parallel study.
3 Initially after the screening period here, patients, first
4 of all, entered a single-blind placebo period here, after which
5 they were randomized to receive either fluoxetine, bupropion,
6 and placebo. Bupropion is a predominantly dopaminergic agent,
7 and patients were randomized to 300 milligrams a day, as 100
8 milligrams three times a day, and fluoxetine 20 milligrams
9 a day.
10 For this study, the CGI score was listed as the
11 primary outcome measure. That was specifically the CGI in
12 terms of those patients who achieved a score of 1 or 2 was
13 listed as the primary outcome measure. So, this is the
14 percentage of responders, the percentage of patients who
15 achieved a score on the CGI of 1 or 2, the primary outcome
16 measure here.
17 As we see here, for fluoxetine patients in orange,
18 there was quite clearly a trend toward significance; p, 0.07
19 for fluoxetine patients versus placebo. The differences did
20 not attain statistical differences between fluoxetine and
21 placebo, but you see that the percentage of responders between
22 the bupropion and the placebo groups is very similar, so
23 indicating perhaps, as Dr. Endicott had alluded to, some
24 evidence of the serotonergic specificity for patients with
1 When one considered any improvement on the CGI,
2 a secondary outcome measures, scores of 1, 2, or 3 -- so,
3 patients who listed any improvement when they scored 1, 2,
4 or 3 on the CGI, and then the differences between the groups
5 are statistically significant in that fluoxetine patients
6 attained the greatest number of patients who were responders,
7 with statistical superiority versus placebo. Again,
8 essentially no differences between the bupropion and the
9 placebo groups.
10 Consideration of the secondary outcome measures
11 for this study in terms of the daily assessment of functioning,
12 the GAS scores, all showed similar results to the first outcome
13 measures shown here in that the differences indicated some
14 superiority for fluoxetine but not attaining statistical
15 significance. So, I'm not going to show all of those here.
16 So, two studies have confirmed the efficacy of
17 fluoxetine in PMDD, and a third study has provided supportive
18 evidence with respect to the efficacy of fluoxetine in PMDD.
19 Importantly, the efficacy shown in these studies is entirely
20 consistent with the other double-blind studies reported in
21 the literature.
22 I'd like to move on to show you the effect size.
23 As was evidenced in these studies, there were a variety of
24 scales used because there is no one gold standard scale for
25 PMDD. But it's also interesting to note that even when one
1 makes a comparison of the effect size across the studies, you
2 see a moderate to large effect size consistently for these
3 patients. This is shown in the next slide.
4 So, effect size can be regarded as a unitless
5 measure that can compare across different studies and different
6 scales. Generally, the traditional thing is that patients
7 with an effect size of 0.5 to 0.8 have demonstrated a medium
8 to large effect of treatment.
9 Now, the circled shapes here are the primary
10 outcome measures. This lists basically the outcome effect
11 size for study 1 and study 2. Study 1, the circles are the
12 study 1 outcome measures for the mood items, the physical items
13 here, and the social impairment. The primary outcome measure
14 is this one, VAS Mood-3. For study 1, with respect to the
15 60 milligram arm, the black circle, that is the VAS Mood-3
16 here. For study 2, the primary outcome measure, the VAS
17 Mood-4, is shown here. So, for the primary outcome measures
18 for the first two studies, we see an effect size which is medium
19 to very large, consistently for these studies for the
20 fluoxetine groups. If you look at broadly the picture of
21 effect for the other effect sizes listed here with respect
22 to the other mood subtotals, the other physical subtotals,
23 and the social impairment subtotals, one sees very broadly
24 an effect size which is ranging from medium to large in the
1 So, overall in terms of efficacy, I'm going to
2 conclude on the efficacy here. The PMDD studies were
3 randomized, double-blind, and placebo-controlled. The study
4 populations were appropriate and consistent, and the outcome
5 measures were appropriate to measure changes in PMDD symptoms.
6 PMDD studies demonstrated the efficacy of
7 fluoxetine in the range of 20 to 60 milligrams a day. Again,
8 just to reiterate, 20 and 60 milligrams appeared to be similarly
9 effective, although there was some numerical superiority for
10 60 milligrams. And importantly, fluoxetine was effective in
11 treating the symptoms of PMDD for up to 6 months. The efficacy
12 of fluoxetine was also evident during the first treatment cycle
13 in all of these studies.
14 Moving on to the safety, as I have indicated, the
15 safety profile of fluoxetine is well-known. Hence, I will
16 provide a rather succinct summary.
17 First of all, with respect to the safety population
18 in the PMDD patients -- and I will compare that to the overall
19 fluoxetine safety database for the indications for which it
20 has been approved in the U.S., that is depression, OCD, and
21 bulimia, numbering almost 4,000 patients. I will also compare
22 the safety profile of the patients with PMDD to a subgroup
23 of this larger fluoxetine database, and that is the female
24 patients aged 18 to 45 years which most closely approximates
25 PMDD patients. That database is numbering almost 1,700
2 I will focus on study 1 in terms of the PMDD
3 studies. Study 1, study 2, and 3. The adverse events
4 collected here were spontaneously collected for study 1 as
5 treatment emergent adverse events. For studies 2 and 3, the
6 adverse events were collected in a different manner, and so
7 it was difficult to merge the database with respect to the
8 adverse events. So, I'm going to concentrate on study 1 when
9 I show you the adverse event profile for PMDD patients.
10 First of all, in looking at the overall study drug
11 exposure for the PMDD safety population in total, the total
12 days of exposure was over 27,000 for fluoxetine at any dose.
13 Importantly, about 50 percent of this exposure was in the
14 range of 151 to 220 days.
15 This is looking at study 1 and looking at the
16 percentage of patients who reported one or more adverse events,
17 as one would expect overall, a high level of reporting for
18 the three groups and with more patients reporting adverse
19 events in the fluoxetine arms versus the placebo arms. With
20 respect to the patients who dropped out for any adverse events
21 for this study, more patients dropped out in the fluoxetine
22 60 milligram arm as compared to fluoxetine 20 and placebo.
23 No statistical differences in the patient dropouts in the
24 fluoxetine 20 or placebo arm.
25 I just want to reiterate that the patients who
1 were on 60 milligrams in this study did start on 60 milligrams
2 at day 1. They did not have the ability to titrate up to that
3 dose. So, from what we know about fluoxetine, it may be that
4 if they had started on 20 and titrated up to 60 milligrams,
5 this higher rate of dropout would not be evident for patients
6 as shown here.
7 This is also evidenced by the fact that when the
8 dropouts did occur in the 60 milligrams, most of the dropouts
9 were, in fact, in the first portion of the study, so fairly
10 early on with fluoxetine treatment.
11 There were very few serious adverse events in this
12 study at all.
13 This lists the most common treatment emergent
14 adverse events in study 1 reported by at least 10 percent of
15 patients taking fluoxetine. This is fluoxetine 20 milligrams,
16 fluoxetine 60, on both slides. Overall, the pattern of
17 reporting of adverse events for fluoxetine are what we would
18 expect for what we know about fluoxetine. Importantly,
19 fluoxetine 20 milligrams appeared to be very well tolerated
20 with very similar differences, not statistically significant
21 to placebo, for any adverse events of any clinical
23 For patients on fluoxetine 60 milligrams, there
24 were high numbers of adverse events reported and in some cases
25 the differences between fluoxetine 60 and both placebo and
1 20 milligrams were statistically significant. As I stated
2 earlier, the fact that these patients did start out on 60
3 milligrams from day 1 may have been a factor in this.
4 Now, these slides compare the most common
5 treatment emergent adverse events in the three databases that
6 I alluded to. Firstly, for study 1, this is the incidence
7 of adverse events for the fluoxetine groups. There is a
8 combination of the 20 and 60 milligrams groups, and this is
9 compared, first of all, to the approved indications database.
10 These are patients with bulimia, depression, OCD, and then
11 this is again compared to the approved indications females
12 subgroup of that database, females aged 18 to 45. If you view
13 overall the adverse events, the pattern of adverse events is
14 as expected for fluoxetine, and importantly no unique adverse
15 events were emergent which showed any uniqueness for PMDD
17 In order to assess tolerability, it's perhaps
18 pertinent to look at patients who had dropped out due to any
19 adverse events. This shows patients who dropped out in study
20 1, study 19. We view that overall, first of all, for fluoxetine
21 20 milligrams the incidence of dropouts was low, and no one
22 particular adverse event contributed in particular to a high
23 level of reporting of dropout due to adverse event. Very few
24 patients dropped out in studies 2 or 3 due to adverse events.
25 So, again, the dropout here would be as one expected for
2 So, in conclusion for safety, fluoxetine has been
3 used in over 35 million patients worldwide and a very, very
4 large safety database does exist for fluoxetine. I've
5 provided evidence to show its safety in PMDD patients and
6 compared that with respect to the approved indications
7 database. Also extensive post-marketing surveillance for
8 this pharmacological agent has shown it to be a very safe
10 Fluoxetine patients with PMDD is therefore safe
11 and well-tolerated and, importantly, clinically comparable
12 to the known profile of fluoxetine with no unique events
13 seemingly for the PMDD patients.
14 Fluoxetine 20 milligrams appeared to be better
15 tolerated than fluoxetine 60. Overall fluoxetine 20
16 milligrams was as expected, well-tolerated and safe for PMDD
18 I'll now provide dosing recommendations based on
19 the efficacy and safety data that I have just reviewed. Again,
20 just to reiterate, fluoxetine 20 and 60 were similarly
21 effective in patients with PMDD. While fluoxetine 20 to 60
22 is safe for patients with PMDD, 20 milligrams did appear to
23 be better tolerated than 60 milligrams. So, therefore, the
24 optimal dose should be 20 milligrams for patients with PMDD,
25 and some patients may, indeed, benefit by increasing their
1 dose to 60 milligrams.
2 Before providing the concluding the comments, I'd
3 like to address one of the questions that the FDA have raised
4 with respect to the use of oral contraceptives. I've indicated
5 in these studies patients with oral contraceptives were
6 excluded for the reasons previously alluded to in that they
7 do have variable effects on premenstrual symptoms. Where they
8 do have an effect, it seems to be on the physical symptoms,
9 and that is inconsistent so. So, in order to clearly delineate
10 the effects of fluoxetine in these studies, oral contraceptives
11 were excluded. Nevertheless, for menstruating females,
12 reports anything up to 30 percent of patients may be taking
13 oral contraceptives. So, the question arises, does the
14 combination of the use of oral contraceptives in combination
15 with fluoxetine have any implications for efficacy or safety
16 for that combination?
17 First of all, what are the potential for possible
18 interactions from the pharmacokinetic point of view?
19 Fluoxetine is metabolized primarily by the P450 2D6 enzyme
20 system. The oral contraceptives are primarily metabolized
21 by the P450 3A4 enzyme system. Now, the effect of fluoxetine
22 on the 3A4 system has been investigated by virtue of in vitro
23 and in vivo studies. The in vivo studies, using the 3A4
24 substrates of midazolam and terfenadine, did not indicate any
25 clinically significant interactions between the combination.
1 So, this suggests that there is unlikely to be any potential
2 for interaction between fluoxetine and oral contraceptives.
3 I'd now like to provide evidence from clinical
4 data which further supports this. As I've indicated, there
5 is a very large efficacy and safety database for fluoxetine
6 with respect to other indications. So, although the PMDD
7 patients did not have any oral contraceptives concomitantly,
8 in the other indications, for example, depression, OCD, and
9 bulimia, there were patients who did take oral contraceptives.
10 So, by subgrouping those existing safety database with respect
11 to efficacy and safety, it was possible to try and tease out
12 any possibility of interactions between oral contraceptives
13 and fluoxetine.
14 First of all, with respect to efficacy, just to
15 reiterate, many of the women in the approved indications
16 database were taking oral contraceptives. So, in viewing the
17 efficacy of patients with depression, OCD, bulimia, and
18 comparing it for those patients who did take oral
19 contraceptives versus those that did not, there was no clinical
20 evidence that concomitant use of oral contraceptives either
21 augmented or lessened the efficacy of fluoxetine.
22 The same analysis with respect to the safety during
23 clinical trials of fluoxetine, no drug interactions were noted
24 for patients who were taking oral contraceptives.
25 Importantly, extensive post-marketing
1 surveillance has not shown any evidence for interactions
2 between fluoxetine and oral contraceptives. It's important
3 to note that fluoxetine has been on the market for over 10
4 years. Also, a search of the literature yielded no case
5 reports of such an interaction.
6 So, it would seem from the data presented, that
7 oral contraceptives and fluoxetine can be used safely with
8 respect to efficacy and no safety implications.
9 So, my concluding remarks. PMDD is a distinct
10 clinical entity which can be differentiated from depression
11 and other anxiety disorders. It can be considered a severe
12 form of premenstrual syndrome that causes impairment of
13 functioning. It's quite clearly inadequately recognized and
14 treated at the present time. For fluoxetine in PMDD, three
15 randomized, double-blind, controlled studies presented
16 support for the efficacy of fluoxetine in PMDD. The results
17 presented are entirely consistent with the numerous other
18 published studies for fluoxetine in PMDD. Safe and
19 well-tolerated at the recommended dose, and the dosing
20 recommendation is appropriately supported by the data
22 Thank you very much for your attention.
23 DR. TAMMINGA: On behalf of the committee, I'd
24 like to thank Dr. Judge, Dr. Endicott, and Dr. Brophy for their
25 very well done presentation of the data set.
1 I'd like to suggest that we take a break and
2 formulate questions, and following the break, the committee
3 will come back and address questions both to Lilly and to the
4 FDA. We'll take a break and be back by 10:15 please.
6 DR. TAMMINGA: I'd like people to take their seats
7 so we can restart the meeting please.
8 The committee has now heard a presentation from
9 Lilly about their indication. We've heard the issues laid
10 out and many probing questions laid out by Dr. Laughren. Now,
11 the committee will have a chance to ask questions to Lilly
12 about the presentation of their data.
13 I would like to encourage not only the committee
14 members, but also the advisors to satisfy every question, so
15 to speak, to Lilly because after we come back from lunch, the
16 committee will then talk about all the issues that came up
17 about the presentation.
18 One more thing I need to remind the committee of.
19 Unlike our last meeting, all the microphones are active all
20 the time.
22 DR. TAMMINGA: So, if people would just keep that
23 in mind.
24 There are a number of committee members who have
25 questions for Lilly about the presentation. I would suggest
1 that Dr. Judge actually come up to the podium, if you do not
2 mind, rather than getting up and going down all the time, and
3 you can actually address questions from the committee. I did
4 not mean to say that Dr. Judge had to answer these questions
5 all by herself.
7 DR. TAMMINGA: But I suspect all of your Lilly
8 will help out anytime. Also, we might have questions for the
9 rest of the Lilly people, but you can really moderate the
11 I might just take the chair's prerogative and ask
12 the first question. Would you remind us, Dr. Judge, what is
13 actually the half-life of fluoxetine and its major metabolite?
14 DR. JUDGE: The half-life of fluoxetine, 4 to 6
15 days. The active metabolite, norfluoxetine, up to 16 days.
16 DR. PARRY: I'd appreciate it if you could review
17 on each of the studies you presented, the authors, the site
18 of the study, and where it was published.
19 DR. TAMMINGA: If the person who's asking the
20 question could just identify themselves for a minute.
21 DR. PARRY: I'm Barbara Parry, Professor of
22 Psychiatry, University of California, San Diego.
23 DR. JUDGE: Study 1 was conducted in Canada in
24 7 centers and was published by the principal investigators,
25 Dr. Steiner, et al. in the New England Journal of Medicine.
1 Dr. Schmidt was the primary investigator for the
2 second study, X022, and that was conducted in that one center.
3 That was published in which year, my colleagues can remind
4 me. We'll get that information.
5 And the third study was conducted by -- the
6 principal investigators were Terry Pearlstein, et al., and
7 were conducted in two centers, the other investigator being
8 Dr. Stone. That was published also a number of years ago.
9 DR. THYS-JACOBS: I'm Susan Thys-Jacobs, and I
10 want to just ask a couple questions about study 19, which was
11 the multi-center trial.
12 All the studies that you had presented are
13 double-blinded and I'm assuming that the tablets looked alike.
14 But in 19 there was placebo, 20 milligram dose, and there
15 was a 60 milligram dose. When they went from single-blind
16 into double-blind phase, how did you carry that out?
17 DR. JUDGE: It was one capsule. There was always
18 one capsule for 60 milligrams, 20 milligrams, and placebo.
19 DR. THYS-JACOBS: It was one capsule for the 60.
20 DR. JUDGE: Yes.
21 DR. THYS-JACOBS: Okay.
22 Another question for 19 was that you defined
23 baseline scores as visit 3 and 5. Were the true baseline scores
24 before entering the single-blind washout different or similar?
25 Luteal mean scores.
1 DR. JUDGE: Dr. Steiner can perhaps elaborate on
2 this as well. But for the patients, as they entered the
3 screening phase, and then prior to that, the scores were
4 similar, but I don't have the scores for that on hand.
5 Dr. Steiner, can you comment?
6 DR. STEINER: There's no difference in the
7 baselines for the two or three cycles that actually screened
8 the patients before they went into the single-blind assessment
9 phase. The data that are used are for placebo nonresponders
10 that entered into the randomization.
11 DR. THYS-JACOBS: I have another question for 19.
12 You showed the data at the first treatment through treatment
13 cycle 6. At treatment cycle 6, however, at the 20 milligram
14 dose, there seemed to be a diminution effect, not major, but
15 there did, indeed, seem to be some decreased effect. Was that
16 effect significantly different from treatment cycle 1?
17 And how do you explain the fact that there was
18 a decrease of rapid decline in symptoms in treatment cycle
19 1 and then there seemed to be a gradual increase in symptoms
20 by treatment cycle 6? Do you think that women over time become
21 more tolerant to this drug?
22 DR. JUDGE: I'll make an attempt at answering
23 that, and perhaps Dr. Steiner can also comment.
24 What's interesting at cycle 1 of that study, as
25 you correctly noted, is a very, very robust and fast response,
1 and it may be evidenced by the fact for the extreme relief
2 experienced by these patients. Remember, they've had several
3 cycles at that time of prospective monitoring, and it may be
4 reflected in their extreme relief.
5 Now, that was a 6-month study and so, therefore,
6 a long-term study. Also, it may be reasonable to assume that
7 patients towards the end of the study are less able to reflect
8 or relate to their baseline levels of functioning as they would
9 earlier on in the study.
10 The important thing is that throughout the study
11 20 milligrams was statistically significant from placebo in
12 the LOCF population.
13 DR. THYS-JACOBS: Well, that was the mean and at
14 all cycles.
15 DR. JUDGE: At all cycles.
16 DR. THYS-JACOBS: No. My question was, was there
17 a difference --
18 DR. JUDGE: Was there a difference between the
19 last --
20 DR. TAMMINGA: One at a time please.
21 DR. JUDGE: My statistician, Dr. Brown, will
22 attempt to answer that.
23 DR. BROWN: This analysis looks at the comparison
24 of the treatment effect at the first treatment cycle to the
25 last treatment cycle, cycle 1 versus cycle 6. These are, of
1 course, just those patients that completed all the way through.
2 So, it's just those patients that completed that 6-treatment
3 cycle. We show the means and the standard deviations, of
4 course, for the first and the last. It's just a basic paired
5 t-test looking from the first to the last, and you can see
6 for the within-group comparisons for the placebo and the 20
7 and the 60 milligram groups, there were no statistically
8 significant differences between the first and the last. For
9 fluoxetine 60 milligrams, it's a trend for a difference.
10 Now, looking at the physical, there was a
11 difference between the first and the last for the fluoxetine
12 60 milligram group, but not the fluoxetine 20 milligram group.
13 DR. TAMMINGA: So, this is a completers only
15 DR. BROWN: Yes. This is only completers.
16 That's correct.
17 DR. TAMMINGA: So that if you look at a completers
18 only analysis, it looks considerably different than the last
19 observation carried forth, which is a slide that Dr. Judge
21 Dr. Hamer.
22 DR. HAMER: I actually have lots of questions,
23 but just one that's --
24 DR. TAMMINGA: Do you want the statistician to
25 sit down or stand up?
1 DR. HAMER: I'm not sure who's appropriate to
2 answer this one.
3 Remind me again about what direction the scoring
4 is; that is, let's look, for example, at fluoxetine 60, first
5 cycle, 23.7; last cycle, 31.6. Does that mean they got better
6 or they got worse?
7 DR. JUDGE: From baseline, the average follicular
8 scores for fluoxetine 60, as in all of them, for VAS 3, was
9 actually around 50. So, the mean here is -- this is 23, which
10 -- and so the mean here is 31, which is slightly worse than
11 23. The lower the score, the better the patient.
12 DR. HAMER: So, here for the fluoxetine 60 group,
13 they got non-significantly worse between the first treatment
14 cycle and the last treatment cycle. Right?
15 DR. JUDGE: This is a difference of 7 points on
16 this score.
17 DR. HAMER: In many of the analyses that you
18 presented earlier, you presented a lot of data in which one
19 group had a number that was bigger than another group and
20 non-significant, but nonetheless, you pointed out to us that
21 one was numerically different than another. Here you're
22 choosing not to pay attention to that numerical difference.
23 DR. JUDGE: I don't understand what you mean.
24 We are paying attention in showing that to you. There is a
25 numerical difference. Here the difference is from the
1 baseline to the mean, to the mean, there is, for example, for
2 the 20 milligrams, maybe there's a difference from here to
3 here. The last treatment cycle is about 5 points and only
4 about 7 points for fluoxetine 60 milligrams. So, it's not
5 that great.
6 DR. TAMMINGA: Dr. Dominguez.
7 DR. DOMINGUEZ: Yes. This is an unusual
8 application in that not only the total number of patients that
9 were entered into the application, which was similar to the
10 OCD application, but also from the fact that, as far as I'm
11 concerned, at least 80 percent of the strength of the treatment
12 effect is carried by one study. And perhaps you could argue
13 90 percent of it is carried by the 019 study, the other two
14 studies being relatively small.
15 Was there an extension phase to the 019 study?
16 DR. JUDGE: There was not a formal extension phase
17 to that study. Dr. Steiner can perhaps comment anecdotally
18 on what happened to patients after that study, but there were
19 no formal extensions to that study.
20 Dr. Steiner, would you like to comment?
21 DR. STEINER: In all sites, most of the subjects
22 who were completers on fluoxetine requested to stay on the
23 drug and did so. We have informal or anecdotal evidence up
24 to 1 year that women still were on the drug. I'm talking 20
1 DR. DOMINGUEZ: Was fluoxetine available in the
2 market at the time that the study was initiated in May of 1990?
3 And could this have influenced your retention rate? I am
4 not surprised with regards to the lower retention rate with
5 the placebo group and the 60 milligram group, but I am somewhat
6 surprised at the low retention rate in the 20 milligram group
7 since they seem to be doing so well. So, could the fact that
8 the medication was already available in Canada at that time
9 have influenced your retention rate? What are your thoughts?
10 DR. STEINER: I can only speculate. But the drug
11 was available.
12 DR. DOMINGUEZ: It was available.
13 DR. STEINER: Yes. At the end of the study, the
14 drug was available.
15 DR. DOMINGUEZ: At the beginning of the study,
16 it was also available. Correct?
17 DR. STEINER: Yes.
18 DR. TAMMINGA: Dr. Judge, in this study number
19 19, could you review for us again the retention rate in the
20 placebo and the 20 and the 60 milligrams?
21 DR. JUDGE: If we could go back to the primary
22 analysis that shows the patient disposition for study 19
24 Green, placebo; in orange, fluoxetine 20; and in
25 yellow, fluoxetine 60. For fluoxetine 20 milligrams, about
1 65 percent of those patients completed the study. Remember,
2 this is actually a 6-month study, which is a long-term study.
3 As you would appreciate in doing studies with
4 obsessive-compulsive disorder, depression, other studies,
5 it's actually very difficult to keep patients in a long-term
6 study. But nevertheless, 65 percent of them completed, and
7 the completer rate in terms of placebo and 60 milligram arm
8 ranged from 40 to 50 percent.
9 With respect to the dropouts due to adverse events,
10 as you pointed out, it's higher in the 60 milligram group than
11 the other groups. The dropouts due to adverse events between
12 fluoxetine 20 and placebo was not statistically significant,
13 and to be honest, with respect to other studies that we know
14 for depression, OCD, or whatever, it's not remarkably
15 different. Again, remember, this is over a 6-month study.
16 As I pointed out, for fluoxetine 60 milligrams,
17 patients were started on that drug at day 1, 60 milligrams
18 at day 1. It would be more appropriate to start them on 20
19 and titrate them up, giving them a chance to tolerate the side
20 effects. If that had taken place, the titration, one would
21 in fact expected a perhaps lower dropout rate due to adverse
23 In terms of lack of efficacy, as we predicted,
24 about 25 percent of placebo patients dropped out due to lack
25 of efficacy.
1 DR. DOMINGUEZ: Yes. I appreciate the fact that
2 this was a 24-week study and the OCD studies were 13 weeks.
3 On the other hand, one could view it as a 6-cycle study versus
4 a 13-week OCD study, and that the percentage of response in
5 these trials was very similar to the percentage of response
6 in OCD trials with regards to the active treatment and in
7 comparison also with the placebo group. Basically you had
8 no placebo effect here. You got no placebo effect in your
9 OCD application.
10 So, since 80 percent of the patients in the 20
11 milligram OCD study completed the study, I'm a little bit
12 surprised at 65 completing after 24 weeks. Just an
14 And I wanted to relate that to the availability
15 of the drug in Canada that some patients may have opted out
16 of the study, either seeing insufficient response, and also
17 based on the population that you treated, which was principally
18 a college graduate population or higher education.
19 DR. TAMMINGA: Yes, Dr. Geller.
20 DR. GELLER: I just want to comment on comparing
21 OCD to this disorder, that OCD bothers you every day of the
22 month. This bothers you just part of the month and that might
23 account some for the difference in dropouts.
24 DR. TAMMINGA: Additional questions for Lilly?
25 Dr. Fyer.
1 DR. FYER: I just want to follow up somewhat more
2 informally this observation because this is the main study
3 that the efficacy depends on. The fact is that they start
4 our randomizing 320 patients out of over 400, and then we end
5 up with 172 patients at the end. I agree that it doesn't seem
6 immediately obvious why that should be true despite the length
7 of the study. If you look at the table about attrition, it
8 doesn't look like there's such a much higher initial rate of
9 attrition in the 60 milligram group.
10 So, what I'd like to just ask is just informally
11 do the people from Lilly have any idea about why so many people
12 dropped out in that study? Because in the other studies, Dr.
13 Judge made the point that very few people did drop out. Is
14 it just because you have a more representative sample, you're
15 going to have a higher dropout rate? Or is there something
16 about that study? What were people's informal observations
17 about that situation?
18 DR. TAMMINGA: Can you comment on this, Dr. Judge?
19 DR. JUDGE: Just on that point when you said there
20 wasn't evidence of attrition early on in the study, there was
21 in fact, and I think we have a slide to show in terms of patient
22 dropouts due to adverse events for the 60 milligram arm versus
23 the 20. You'll see that most of them did, in fact, drop out
24 in the first two cycles. I'll just show you that, and the
25 slide will show you that.
1 Thereafter, the attrition rate for fluoxetine 20,
2 placebo -- placebo patients dropped out more towards the end.
3 We show this here. So, this is the patient discontinuations
4 due to adverse events by time. This is a number of patients
5 dropping out. So, for fluoxetine 60, you see that in the first
6 couple of cycles, the patients dropped out. Thereafter, it's
7 a steady dropout, not particularly different from cycle to
8 cycle, and it's fairly constant for the other two arms.
9 Do you also have the slide due to lack of efficacy,
10 the same slide? This is true for the lack of efficacy. As
11 one would expect, for placebo patients, more patients drop
12 out as the study continues over the course of time due to lack
13 of efficacy.
14 In reviewing this data, I think overall the highest
15 number of patients stayed in the study for 20 milligrams.
16 When one looks at the attrition rate for some of the long-term
17 studies for depression and OCD, there is quite a high attrition
18 rate when you refer to OCD studies. The short-term attrition
19 rates are obviously better than -- short-term attrition rates
20 would be not as high as this study. This is a long-term study,
21 and generally when one looks at long-term studies for all drugs,
22 there seems to be a high attrition rate in general. But even
23 so, 65 percent of patients on 20 milligrams were still remaining
24 at the end of the study.
25 DR. TAMMINGA: Dr. Fyer?
1 DR. FYER: Yes. I don't want to get into a picky
2 thing about 10 patients, but I think the important thing is
3 that there's a continued steady drop-off. You have a very
4 slight increase in the number of people on 60 milligrams in
5 the first week, but then at every time another 10 percent of
6 the patients are leaving the study. It's not just in the 60
7 milligram group. Again, I'd just like to ask you if you have
8 some idea as to what was going on in that study.
9 DR. JUDGE: Can I refer also Dr. Steiner here?
10 DR. STEINER: This is not unusual in PMS/PMDD that
11 you lose at the end up to a third of your population, A.
12 B, this was a very labor-intense study. These
13 women were with us for a year. They had to come twice a month,
14 and some of them just gave up after a while. And the steady
15 decline is really not unusual in these studies.
16 DR. TAMMINGA: There was some part of Dr. Fyer's
17 question that would contrast the dropout rate in this study
18 with the dropout rate in the next two. While you're up there,
19 maybe you could just more specifically --
20 DR. STEINER: This was the longest and more
21 labor-intense than the others. The requirements were
22 different. We were bringing them in. They had to be, twice
23 a month, in the clinic for up to 90 minutes, 2 hours sometimes.
24 Canada is a big country. There are all not in the big cities.
25 They're coming from far away. We had winters. All the stuff
1 that you see under "other" is transportation, distance, and
2 cold weather. And then you had to sort of drop them out because
3 if they missed two or three visits, they were out.
4 DR. TAMMINGA: Dr. Winokur.
5 DR. WINOKUR: Related to this study, you mentioned
6 early on in introducing it that a decision was made to focus
7 on the VAS Mood-3 as the primary outcome measure. I wasn't
8 clear about the rationale for that choice. I wonder if you
9 could elaborate on that a little bit.
10 DR. STEINER: If you recall -- we're talking the
11 late '80s, early '90s -- the visual analog scale was a homegrown
12 thing that we developed. At the time when we started working
13 with Lilly, we did not anticipate, nor did we know that
14 fluoxetine is going to be helpful for the physical symptoms.
15 We thought that if we lumped together the 7 symptoms and if
16 it doesn't work for the physical symptoms, we will actually
17 wash out some effect on the other 3 major components. So,
18 we picked irritability, dysphoria, and tension as the primary
19 outcome. We left the lability out because some people
20 questioned whether this is a unipolar or bipolar dimension,
21 and we then separated them out.
22 But as you have seen, we have the same statistical
23 significance for the 3 VAS. We have it for the 7 VAS, and
24 we have it for the physical separately.
25 DR. WINOKUR: My other question. This now
1 switches to safety, adverse events. Can you comment either
2 from your specific studies that you talked about or other
3 information in the literature about the occurrence of hypomanic
4 or manic symptoms in this specific population on fluoxetine?
5 DR. JUDGE: With respect to the occurrence and
6 switch into mania for fluoxetine in general, there's a very
7 low switch rate. That's evidenced by the clinical trial
8 database. In fact, one of the few double-blind studies of
9 bipolar depression with fluoxetine versus imipramine,
10 evidenced again a nice treatment effect, without any increase
11 in switch-over to mania versus placebo. So, that's for general
13 For switch-over to mania in this study, there were
14 no patients in the PMDD population that ascribed to or switched
15 to mania.
16 DR. WINOKUR: If you scale it down from full-blown
17 mania to more just --
18 DR. JUDGE: To hypomania?
19 DR. WINOKUR: -- manifestations suggestive of
20 hypomania, I'm just wondering whether there's -- I'm focusing
21 on this population because this is going to be a population
22 that's going to be extending the use of this drug.
23 DR. JUDGE: Right.
24 DR. WINOKUR: And if there's any even clue of the
25 potential for activation of --
1 DR. JUDGE: Yes. From the studies in this
2 population, there wasn't any event that we'd say, oh, my gosh,
3 this is happening.
4 Anecdotally perhaps Dr. Steiner can comment about
5 his other experience or even Dr. Endicott.
6 DR. STEINER: The exclusion criteria were that
7 they should not have an Axis I diagnosis. Therefore, we did
8 not include bipolars. But we did not have hidden bipolar II's
9 and we did not have a single switch in this study.
10 DR. WINOKUR: That's really what I'm trying to
11 focus on is excluding the known bipolars. Is there any even
12 hint? Because what we're really talking about now is extending
13 this drug to a totally separate population not known to have
14 bipolar, and that's why I think that's a crucial issue.
15 DR. STEINER: We did not witness one single case.
16 DR. JUDGE: And that was also evident for the other
17 two studies.
18 DR. THYS-JACOBS: There was no placebo effect
19 noted during the double-blinded study period 2 in this trial.
20 Was there a placebo effect going from the screening period
21 to study period 1? Was anything noted?
22 DR. JUDGE: Are you talking about study 1?
23 DR. THYS-JACOBS: 19, yes.
24 DR. JUDGE: Can you repeat your question?
25 DR. THYS-JACOBS: There was no placebo effect
1 noted during the double-blinded phase at all, and apparently
2 that was being screened for during the single-blinded study.
3 I'm asking was there a placebo effect noted during the
4 screening period into the single-blinded phase?
5 DR. JUDGE: Yes, indeed. And Dr. Steiner will
6 elaborate. But there were some patients who did drop out
7 during the screening phase because they were placebo
8 responders. So, a placebo effect was evident.
9 Dr. Steiner?
10 DR. STEINER: There were 12 placebo responders
11 during those first two cycles and they were not randomized.
12 DR. TAMMINGA: 12 out of 320 or 12 out of 450?
13 DR. STEINER: Out of 450.
14 DR. JUDGE: Remember, by this time in terms of
15 the placebo effect, patients had really undergone several
16 cycles of screening with respect to their diagnosis and
17 prospective daily ratings. So, really --
18 DR. THYS-JACOBS: There were two cycles. We're
19 talking about two cycles. There were two screening cycles.
20 Is that correct?
21 DR. JUDGE: There were screening cycles, but even
22 before then, patients in terms of their screening, that they
23 brought patients into the study -- we're talking about one
24 would expect a fairly low level of placebo responders in this
25 study because, for example, in depression or other studies,
1 we don't prospectively diagnose patients by prospective
2 measurements. This is unusual and I think serves to lower
3 the placebo response in PMDD studies in general anyway.
4 In fact, the placebo effect noted here is not
5 appreciably different from placebo effects noted in other SSRI
6 studies in the literature.
7 DR. THYS-JACOBS: Most of the studies that I know
8 of have a 25 or 20 percent to 30 percent effect. You're saying
9 that in these studies, in the PMDD trials, there is no placebo
11 DR. JUDGE: You saw that there was a placebo
12 effect, and in fact for this study, as Dr. Steiner will comment
13 as well, there is a placebo effect. The placebo effect is
14 low which can be sometimes attributed to the screening allowed
15 for these patients. But also remember, in general -- this
16 study was conducted quite a while ago, and maybe as with other
17 studies, maybe there may be a creep up of placebo effect due
18 to other phenomena, as we see with OCD, as we see with
19 depression. That may be evident, and maybe that's what you're
20 ascribing to PMDD studies.
21 DR. THYS-JACOBS: No. Most of the published
22 trials on PMDD and PMS have shown effects of anywhere from
23 20 to 70 percent. So, how are you talking about 12 out of
24 300 patients going from the screening period into the study
25 phase? That's a very small number.
1 DR. JUDGE: But there is no study that shows a
2 70 percent response rate for PMDD. When we talk about those
3 high levels of placebo responses, we're talking about studies
4 which really are not specifically PMDD but more often listed
5 as severe PMS or PMS in general.
6 Dr. Steiner, would you care to comment on that?
7 DR. STEINER: Two things. I agree with Susan that
8 -- we were very surprised. It was a low placebo response.
9 Two things to say about it.
10 The literature is really not about PMDD. It's
11 mostly about PMS, and there the placebo response was obviously
12 much higher.
13 The other thing is that between the initial
14 screening and the randomization, we have lost not only 12
15 placebo responders, we have also lost approximately 80 other
16 patients which we were not able to document whether they were
17 placebo responders and that's why they left us. They just
18 disappeared for other reasons. So, maybe it was a little bit
19 higher. But we have documented to date on 12 placebo
20 responders who stayed with us and had to be told that they
21 cannot be randomized because they're placebo responders.
22 That's what I can report on. The others are speculations.
23 I think that overall we had a very low placebo
24 response because of the extremely rigorous inclusion criteria.
25 This was the first time that that kind of rigor was actually
1 applied. So, we excluded them before they even were coming
2 into the placebo phase.
3 DR. TAMMINGA: Dr. Temple.
4 DR. TEMPLE: Just a point of terminology. The
5 placebo response is not measured in clinical trials. What
6 you measure is the response in the placebo treated group, which
7 is a mixture of true placebo response, that is, response to
8 drug taking, and the natural history of the disease. This
9 comes up a lot.
10 The idea that in depression the placebo response
11 is 60, 70 percent I think is totally unreasonable and is a
12 quirk of study design. You take people who are in the process
13 of being at the worst part of a cyclical disease and then you
14 put them in a trial, it's not surprising they regress toward
15 the mean and do other things like that. But if they got a
16 good history of regular monthly symptoms over many years, it
17 doesn't surprise me at all that when you put them in a trial,
18 nothing much happens because they're not regressing toward
19 a mean. This is something they have.
20 So, these are very study-determined. I have a
21 sort of personal quest to not call these responses in the
22 placebo group placebo response until somebody actually
23 includes a no-treatment group, that is, someone who doesn't
24 get any drug at all, and that is almost never done. So, we
25 don't really know what the placebo response is here. We just
1 know how the placebo group responds, which is not necessarily
2 the same thing at all.
3 DR. TAMMINGA: Dr. Hamer.
4 DR. HAMER: After a number of years of doing this,
5 I think I have a reasonably good idea of how to interpret these
6 sorts of studies when the studies have been designed with
7 rigorous protocols by pharmaceutical companies involving an
8 end of phase II/pre-phase III meeting in which the design,
9 the outcome measures, and the statistical analyses are planned
10 and consultation takes place. It's less clear to me how to
11 interpret studies whose purpose is registration when
12 apparently that kind of process didn't take place here.
13 So, to help me understand this, could somebody
14 from the pharmaceutical company please, in some sense, take
15 me through the history of these studies, tell me whose idea
16 they were, if indeed rigorous documentation, such as protocols
17 that specify design, outcome measures, statistical analyses,
18 in as much specificity as might take place in sort of the usual
19 situation and to what extent things like the switch apparently
20 from a measure that consisted of a total of a full-blown set
21 of visual analog scales to a measure that consisted of only
22 a subset of them, of an analysis that switched from an analysis
23 of percent change to an analysis that consisted of absolute
24 change, even though, if you analyzed the data both ways, you
25 get consistent results.
1 But there's in some sense at least the potential
2 for kind of a hidden multiple comparison issue here which,
3 thankfully for Lilly, is probably obviated by the fact that
4 there was such a strong effect that we probably don't need
5 to worry about it much.
6 But I would like to know sort of how these studies
7 got designed and what role Lilly played, if any, in the design,
8 funding, and execution of them.
9 DR. TAMMINGA: And your question pertains to all
10 three studies.
11 DR. HAMER: It pertains to all three studies.
12 DR. TAMMINGA: Dr. Judge.
13 DR. JUDGE: Well, study 1 was a protocol designed
14 collaboratively by Lilly and Dr. Steiner. The protocol was
15 put into place and agreed by Lilly and by Dr. Steiner. It
16 was Lilly monitored.
17 DR. HAMER: So, this is what we would call an
18 investigator initiated study?
19 DR. JUDGE: Yes, and Lilly funded and Lilly
21 For the other two studies, they were conducted
22 more independently. For study number 2, this was conducted
23 under an independent IND, and number 3 was exempt from IND.
24 For all three studies, protocols and analysis
25 plans were put in place by the investigators before the study
1 obviously started. Lilly, when comprising their analysis
2 plans -- it's important to note that all analysis plans were
3 put into place and very strict audit and quality controls were
4 done for each of the sites to ensure that the studies had been
5 conducted to GCP standards, had been conducted according to
6 protocol, and with the exceptions that I've stated with the
7 reasons for those exceptions. In all studies, we're confident
8 that the quality of the studies is as one would expect, good
9 quality, GCP conducted studies. Importantly, with respect
10 to the analysis plans, which were prospectively put in place,
11 before any of those Lilly personnel had information or
12 unblinded to the individual patient information.
13 DR. HAMER: Well, then to move to the blinding
14 issue, I got confused by the statement that apparently the
15 crossover study terminated early because the blind was broken
16 somewhere in the middle and then the investigators did an
17 analysis, presented an abstract, and then decided not to
18 continue the study. Is that the case that the blind was broken
19 part way through it?
20 DR. JUDGE: As you indicated, yes, but the blind
21 was not broken to individual patient assignment to the raters
22 who were rating the patients and, moreover, to the patients.
23 As you remember, the primary outcome measure was patient-rated
24 visual analog scale.
25 Now, when the analysis was done for that study,
1 it involved very few numbers of patients. In fact, the
2 investigator found a treatment effect and stopped the study.
3 In fact, there were a number of other patients enrolled in
4 the study at that time, and they were allowed to complete.
5 That numbered a total of 19, as you saw.
6 DR. HAMER: I'll save other questions until later.
7 DR. TAMMINGA: Dr. Chen.
8 DR. CHEN: Let me add some questions for this topic
9 here. Could you briefly describe the early termination for
10 the second study? How many times you have unblinded the data,
11 when you decided? Did the investigator decide, how they
12 decided to terminate the study? Do you have that knowledge
13 here for this?
14 DR. JUDGE: In addition to what I've just alluded
15 to, it was an independent decision by the investigator to
16 terminate the study at that time. But as I said, if someone
17 from my group can elaborate on how many patients exactly that
18 analysis involved. But I said there were a number of other
19 patients ongoing in the study. It wasn't that 19 had completed
20 and then the analysis involved 19 patients. Only a fewer
21 patient number had completed, and in fact when he found that
22 statistical difference, he decided to terminate the study
23 independently. Then the other patients who were already
24 enrolled in that study were allowed to complete. As I said,
25 the clinicians who rated the patients, the patients themselves
1 remained blinded in order to minimize any bias in that study.
2 DR. TAMMINGA: Dr. Fyer.
3 DR. FYER: I just want to ask a clarification.
4 So, that was an independent investigator study. Who were the
5 clinicians versus the investigator that all this was kept --
6 DR. JUDGE: Well, as with any site, there is a
7 principal investigator, and there are people who work with
8 the investigator who are the study coordinators that are more
9 involved in the actual screening of the patients, the rating
10 of the patients week by week, and assessing them per protocol.
11 DR. FYER: I'm aware of how clinical trials are
12 done generally. But what I'm interested in is, are you aware
13 on a person-by-person basis of exactly who knew and who didn't
14 know? Because very often, especially in smaller
15 organizations, there's a lot of functional overlap.
16 DR. JUDGE: Yes. In fact, for that study -- and
17 my team can correct me if I'm wrong -- it was mainly the other
18 co-investigator for that study, Dr. Su, who was actually seeing
19 more of the patients.
20 DR. FYER: And so, there was no communication
21 about the overall outcome or any issues about patterns of side
22 effects or anything of that nature.
23 DR. JUDGE: Not that I'm aware of. As I said,
24 the audit on that study has been very meticulous in terms of
25 quality assurance and quality control.
1 DR. TAMMINGA: Dr. Hamer.
2 DR. HAMER: Well, to continue Dr. Chen's question,
3 I think maybe one of the things that might be related is, in
4 the protocol, was this interim analysis that led to the early
5 termination of the study planned?
6 DR. JUDGE: This was an unplanned interim
8 DR. HAMER: Do you know if there were other
9 unplanned interim analyses?
10 DR. JUDGE: No, there were not.
11 DR. HAMER: So, if you think about spending your
12 alpha in terms of sequential analyses, there was no adjustment
13 for that here.
14 DR. JUDGE: If I can ask my statistician to comment
15 on this.
16 DR. BROWN: No, there wasn't an adjustment. But
17 like we said, the investigator initiated this unplanned interim
18 analysis at 10 patients, found a significant effect, and
19 decided to stop the study, and continued the patients that
20 were currently enrolled, so we ended up with 19 patients.
21 If we go ahead and use a penalty for an early look,
22 say, an O'Brien-Fleming type of a spending rule, and adjust
23 for those looks at the data at the 10 and the 19, we would
24 still show a significance all the way through. That would
25 be about a .01 nominal significance level we would be looking
1 at at the 19 patient level.
2 DR. HAMER: Although since this is a post hoc use
3 of an O'Brien-Fleming sequential rule, we really don't know
4 how many interim analyses we should be adjusting for since
5 they might have chosen to have done other interim analyses
6 than the ones that they did.
7 DR. BROWN: Right. They might have chosen to do
8 something else, but they did just do the one look at 10. So,
9 you're right. It is a post hoc.
10 DR. TAMMINGA: Should we understand that Lilly
11 looked into whether they did any other unplanned analyses and
12 the answer is no?
13 DR. BROWN: That's right. No, they did not.
14 DR. HAMER: Although perhaps the blind to the
15 raters was not broken, did the raters know that an interim
16 analysis had been done and that the interim analysis apparently
17 showed that fluoxetine was superior to placebo?
18 DR. JUDGE: There was, as I said, an abstract
19 generated from that interim analysis. So, anyone who viewed
20 that abstract would, in fact, know that that was the case.
21 But remember, that was on a fewer number of patients. So,
22 the abstract actually reported a fewer number of patients,
23 but the actual end of the study involved almost a double number
24 of those patients.
25 DR. HAMER: As long as I have gotten us onto the
1 crossover study --
2 DR. COOK: Can I follow up on this one?
3 DR. HAMER: Yes.
4 DR. TAMMINGA: Dr. Cook.
5 DR. COOK: I really feel the need to know very
6 specifically how this study was blinded because it just raises
7 many questions if it wasn't blind to the investigative team.
8 So, I really feel the need to have detailed knowledge of how
9 this was blinded to where it could be relatively arbitrarily
10 unblinded. Were the capsules identical? We have to get the
11 details since it's at variance with usual practice.
12 DR. JUDGE: In terms of the blinding for this study
13 -- and perhaps, Cathy, if you could comment on the actual
14 capsules. Unlike the first study where there was one capsule
15 for like 20 milligrams, 60 milligrams, and placebo, for this
16 study there were, for example, tablets corresponding to 20
17 milligrams, 30, 40, 50, 60. And in each case, if there was
18 a titration, there was a titration. So, for example, the
19 number of placebo capsules would also increase as well. So,
20 the blinding in terms of the numbers of capsules was exactly
21 identical so physicians could titrate up according to safety
22 and efficacy, and the titration would therefore involve, if
23 it was placebo, a greater number of placebo capsules; if it
24 was fluoxetine, a greater number of fluoxetine capsules.
25 Even the principal investigator was blinded to
1 individual treatment assignment.
2 And that interim analysis, the only one planned
3 for that study, was undertaken on 10 patients. There were
4 9 other patients in the study at that time, and they were
5 continued on. So, the final analysis involved 19 patients.
7 MS. SHULER: That's accurate with the exception
8 of the fact that the capsules were in 10 milligram increments.
9 DR. JUDGE: Dr. Tollefson?
10 DR. TOLLEFSON: My name is Gary Tollefson,
11 President of Neuroscience at Lilly. I want to just say a few
12 words so that we don't make this more complicated than it
13 perhaps is.
14 This was an independent investigator initiated
15 trial. However, it was done under conventional double-blind
16 methodology, as you would expect with any clinical trial.
17 Now, it happened at mid-course with the patient
18 numbers that you have seen that the primary investigator opted
19 to look at a group level for whether or not there was a treatment
20 effect. The primary investigator, as you heard, saw a very
21 robust treatment effect between the two groups, and for ethical
22 reasons elected to terminate the study at that point. So,
23 the patients that were represented as study 2 patients were
24 patients randomized prior to an unscheduled interim analysis,
25 and given the robustness of the treatment effect -- you might
1 recall the slide showing overall treatment effects -- the
2 investigator felt that it would not be prudent or ethical to
3 continue prospectively in light of the drug/placebo difference
4 that he saw.
5 DR. TAMMINGA: Dr. Hamer.
6 DR. HAMER: I'm now puzzled, as usual. You're
7 saying that they did an interim analysis after 10 patients.
8 For ethical reasons, they decided to terminate the study.
9 There were patients in that study then at that time currently
10 assigned for placebo, but ethically it didn't bother them then
11 to continue those patients on placebo double-blinded to the
12 end of the study.
13 DR. TOLLEFSON: The patients that had been
14 initially randomized to the trial were followed throughout
15 the entirety of the trial. Thus they provided the basis for
16 the interim analysis. Once that analysis was done, no
17 additional prospective patients were enrolled to go up to the
18 initially targeted patient sample of 30. So, in other words,
19 no additional randomizations occurred after the interim. The
20 interim was based on those patients previously randomized which
21 was the study cohort you saw.
22 DR. HAMER: Right, but the ethical concerns were
23 not sufficient to terminate patients on placebo in the trial
24 rather than continue to give them placebo for the rest of the
1 DR. TOLLEFSON: Yes. You remember the crossover
2 design, and it was felt that the value of the additional
3 scientific information that would be generated via the
4 crossover was justifiable in order to continue patients through
5 the rest of the trial. However, based on the phase I data,
6 separation between drug and placebo, it didn't make a lot of
7 sense to repeat that by enrolling additional patients in the
8 eyes of the investigator.
9 DR. TAMMINGA: But the decision was made on an
10 efficacy basis, not on a side effect basis.
11 DR. TOLLEFSON: That's correct by the PI.
12 DR. TAMMINGA: Dr. Temple.
13 DR. TEMPLE: No one obviously approached us to
14 see whether we thought it was ethical to continue to randomize
15 people to this non-lifesaving treatment. I'm appalled at the
16 idea that someone thought it was ethically imperative to stop
17 the study because of this. That would imply you couldn't do
18 more studies in this disease, which is really a garbled view
19 of what's ethically necessary. But nonetheless, it seems to
20 have happened.
21 DR. TAMMINGA: Dr. Katz.
22 DR. KATZ: I just had a question about the interim
23 analysis. It was done on 10 people. Are those the 10 people
24 who at the time had completed both arms of the trial, or did
25 it include data from all the 19 that were enrolled in their
1 various periods and states?
2 DR. JUDGE: It included data from 10 patients.
3 DR. KATZ: Who had completed?
4 DR. JUDGE: Yes.
5 DR. COOK: I still haven't had my question
6 answered in terms of how this blind could be broken in terms
7 of the security of the blinding codes, et cetera. This seems
8 to have been relatively arbitrary. It obviously wasn't being
9 kept separate from the investigator for the investigator to
10 be able to do this analysis. I just need assurances that we
11 know the blinding was secure.
12 DR. JUDGE: Investigators and all the raters in
13 this study were blind to the individual patient assignment
14 to drug. It is not uncommon to do interim analyses, as you
15 know, for studies to delineate a group effect.
16 DR. COOK: But if they're blind, they can't do
17 the analysis.
18 DR. TAMMINGA: Are you asking the question, Dr.
19 Cook, who was allowed to be unblinded?
20 DR. COOK: Yes.
21 DR. TAMMINGA: Who was allowed to be unblinded
22 in this study? Do we know that? Was that prospectively
24 DR. JUDGE: I don't know specifically the names
25 of those people, but the people who did the analysis were not
1 involved in the conduct of the study in terms of rating the
2 patients or ascribing treatment and seeing them from cycle
3 to cycle and providing them with treatment, monitoring the
4 adverse events, and rating their scales. So, those raters,
5 which are specifically those who saw the patients, were not
6 the ones who completed the statistical analysis. The
7 statistical analysis was done by statisticians who were not
8 part of the rating cohort of that study, and in fact patients
9 themselves who rated themselves on the primary outcome measure
10 were also kept blind to their individual patient assignment
11 as well.
12 DR. TAMMINGA: Are you worried, Dr. Cook, that
13 people might have taken other peaks at the data?
14 DR. COOK: It just seems non-standard. It is true
15 that the patients are rating themselves, but it is certainly
16 possible if anyone working with them or involved in the study
17 knows who is on which, then you violate the basic principles
18 of blinding. That's my real concern.
19 DR. JUDGE: And that was obviously a concern for
20 Lilly. It's a necessary requirement that a study is blinded
21 and kept blinded to the raters of that study, and in
22 comprehensive audit, absolutely comprehensive, a very
23 meticulous audit, assurances were made and received with
24 respect to the blinding of this study.
25 DR. TAMMINGA: Dr. Katz.
1 DR. KATZ: You probably believe that you've
2 already answered this. Maybe you have. But we've heard that
3 the PI took a look at the data. Now, did the PI have anything
4 to do with evaluating patients? Was that person involved in
5 the conduct of the trial?
6 DR. JUDGE: No. The principal rater of that study
7 was Dr. Su. In fact, the authors of that study were Su and
8 Schmidt, and Dr. Schmidt was the person whose name is on the
9 abstract. Dr. Su actually was more involved with the patients
10 from a day-to-day basis, and he was at that time not privy
11 to the individual patient assignment, as I've already alluded
13 DR. TAMMINGA: Dr. Fyer.
14 DR. FYER: I guess what makes me a little
15 uncomfortable about this is the smallness of the operation.
16 What we're talking about is a 19-patient trial with two
17 physicians involved in a relatively small sort of organization.
19 I think I would feel a lot more comfortable if
20 the sponsor had just presented the data saying, look, this
21 is what happened. We had two doctors. One was the PI. They
22 decided to take a look at the data. We don't really know to
23 what extent these individuals communicated or the ethos or
24 sort of cultural environment in a small clinic, which all of
25 us at this table have worked in this kind of environment, might
1 have contaminated this data. But the fact is that the effect
2 at 10 patients was very robust, and we looked into it. We're
3 inclined to think that this didn't occur. That kind of
4 straightforward thing I think would be a lot more reassuring
5 than what's sort of going on here.
6 The other thing that I would just raise as a point
7 is if there was a robust treatment effect at 10 patients, did
8 the sponsor consider just presenting that straightforward data
9 about which there could really be no questions at all? Even
10 though it's a smaller n, in fact we're dealing with a very
11 small n overall.
12 DR. JUDGE: Firstly, I do apologize if you felt
13 that it wasn't straightforward. I was trying to give
14 straightforward answers to these questions in that this study
15 was conducted by one center. And, yes, it's relatively small
16 compared to the other study. Patients acted as their own
17 control, so providing enhanced sensitivity. So, in crossover
18 studies in general one would expect smaller patient numbers.
19 The individuals involved in this study are very
20 well established in terms of their research field.
21 But in terms of our assurances, we also wanted
22 to assure ourselves before coming to you guys that, indeed,
23 the data was collected in a manner that is conducive to GCP
24 standards and to assure ourselves of the highest quality for
25 the data. So, regardless of anything else, we did assure
1 ourselves with very meticulous comprehensive audit that went
2 on at both sites, including the other sites in the other studies
3 as well. All three studies involved comprehensive audit from
4 Lilly personnel, which was also ascribed by also independently
5 conducted audit as well. So, regardless of that, we actually
6 did a lot to assure ourselves of the study in terms of the
7 quality and the integrity of the data.
8 DR. TAMMINGA: Dr. Hamer.
9 DR. HAMER: But good clinical practices don't
10 usually include an unplanned interim analysis, do they, by
11 the investigator?
12 DR. JUDGE: Good clinical practice assures the
13 safety and benefit for those patients done in a double-blind
14 way per protocol. And, yes, there was an unplanned analysis
15 in this study, as you heard, and that was unfortunate, but
16 we've tried to explain to you the reasons why.
17 DR. TAMMINGA: Dr. Parry.
18 DR. PARRY: First a comment. It would just be
19 helpful for purposes of references to have copies of the
20 published papers.
21 But I was interested in whether there was a
22 difference in primary outcome measure as a function of site.
23 As I recall, Meir Steiner's Canadian study, there weren't
24 differences as a function of site, but were there in the
25 Stone-Pearlstein study?
1 DR. JUDGE: In the Stone-Pearlstein study, the
2 publication did cite the CGI as an outcome measure, and it
3 involved the CGI 1 or 2 as the primary outcome measure.
4 For Dr. Su-Schmidt, for the second study, X022,
5 the publication -- it was difficult to ascertain from the
6 publication, if you just viewed the publication in the cold
7 light of day, it talks about the DRF and the VAS Mood-4 as
8 relevant outcome measures. We will have copies of that. We
9 could provide them to you perhaps during lunch, and the FDA
10 have been provided with copies of the publications.
11 DR. PARRY: But I mean, not the measures but the
12 change scores. Were there differences as a function of where
13 the study was conducted in the primary outcome measures,
14 whichever primary outcome measure was used? Whether it was
15 western Canada or eastern Canada or New Jersey or New York.
16 DR. TAMMINGA: Is your question about the number
17 1 Steiner study? Was there a site effect?
18 DR. PARRY: As I remember, there wasn't.
19 DR. JUDGE: If we're talking about the individual
20 site effect for seven centers in Canada, we've got some
21 information on that and my statistician will take you through
22 that. Dr. Brown.
23 DR. BROWN: For the primary efficacy variable,
24 the VAS Mood-3, there was, in fact, a site by treatment
25 interaction. This graph here shows the results by site for
1 the VAS Mood-3. Again, green is placebo, and orange is 20
2 milligrams. Yellow is the 60 milligrams. We can see for sites
3 3 through 7, the dose response is quite similar. Remember
4 that the placebo going above the 0 mark means that the placebo
5 group, in fact, got worse. So, the dose response is quite
7 And it appears that the interaction was being
8 driven perhaps by sites 1 and 2 where you can see in site 1
9 the 20 milligram dose group responded better, in fact, than
10 the 60, which is not consistent across the other sites. In
11 investigator site number 2, the 20 milligram group did not
12 respond as well, in fact, as the placebo group.
13 DR. HAMER: Were there particularly small numbers
14 of subjects at sites 1 and 2?
15 DR. BROWN: Here are the site-by-site patients
16 entered and randomized. So, sites 1 and 2 are a little smaller,
17 but about the same size as sites 6 and 7. Sites 3, 4, and
18 5 were the biggest sites, but the spread was pretty good among
20 DR. TAMMINGA: Dr. Dominguez.
21 DR. DOMINGUEZ: Do you have a similar breakdown
22 as far as discontinuations between sites?
23 DR. BROWN: No. I'm sorry we don't have that.
24 Well, there are completers there, but I don't have
25 discontinuations for a particular reason by site.
1 DR. DOMINGUEZ: For specific reasons per site.
2 DR. BROWN: No. Sorry.
3 DR. THYS-JACOBS: (Inaudible) or entered
4 treatment. You have 400 --
5 DR. BROWN: Pardon me. What was the question?
6 DR. THYS-JACOBS: 405 patients entered. I
7 thought the definition of entering was the treatment phase.
8 So, this is screening.
9 DR. BROWN: That's screening, and then the numbers
10 there, the 108, 104, and 108, those sum to the 320 randomized.
11 DR. THYS-JACOBS: What was the difference when
12 patients actually entered treatment phase in terms of numbers?
13 Do you have that data?
14 DR. JUDGE: Patients who entered the placebo,
15 fluoxetine at randomization numbered 108, just over 300, 320.
16 So, that's the difference between 405 as the patients went
17 to screening.
18 DR. TAMMINGA: Do you have additional questions
19 on this issue? Any additional questions for Dr. Judge? Dr.
21 DR. CHEN: I have one more question about the study
22 number 2. So, could you tell me about the resource? It seems
23 to me today you are real clear about the randomization scheme
24 and the interim analysis, but it seems to me when I reviewed
25 the study, the application, the document said it depends on
1 the investigator's decision. It was not very clear in the
3 But two weeks ago, we asked the same question about
4 the early termination of study 2. At that moment, we didn't
5 get it answered. But it seemed to me our question what's the
6 resource -- within two weeks everything becomes clear to you.
7 DR. JUDGE: No, indeed. You're referring to the
8 telephone conference that we had with the FDA, that Lilly had,
9 to understand our presentation, whether you felt that our
10 presentation was suitable. We asked you to comment on our
11 slides. You said to us that it would be important for us to
12 provide additional information on various questions, and one
13 of them was the use of oral contraceptives. One of them was
14 with respect to this question, and one of them was with respect
15 to study 19 in terms of the primary outcome measure.
16 We didn't really go through, at that time, all
17 of the answers, but we did undertake to provide to you in a
18 few days some of how we would elaborate on it today, which
19 we did just a few days after that telephone conference. So,
20 today you find we understood and then we provided all these
21 answers to you today in terms of your request to us for all
22 of those questions at that time.
23 DR. CHEN: So, are all of those, what you mentioned
24 today, documented.
25 DR. JUDGE: Yes, indeed, they are.
1 DR. TAMMINGA: Dr. Laughren.
2 DR. LAUGHREN: One of the concerns about SSRIs
3 as a group is an effect on sexual dysfunction. That wasn't
4 something that was looked at specifically in these trials,
5 but I did notice, as you were going through the various rating
6 instruments, that the PMTS did have one item that looked at
7 sexual drive and interest I believe. Do you have any data
8 on that specific item?
9 DR. JUDGE: If we could bring up those. Dr.
10 Steiner would like to comment before we bring up those slides
11 as well.
12 DR. STEINER: We actually started to look at the
13 data just a few weeks ago not because of this, because we were
14 interested in looking at it. If you look at our baseline data
15 for that particular question on the PMTS, you see that women
16 rate sexual dysfunction premenstrually as part of the symptoms
17 of PMDD. They're not interested in sex. It's not a very
18 specific question. It just asks are you interested more or
19 less or not at all. So, you see that their baseline, their
20 normal, is during the follicular phase and then they score
21 high during the late luteal phase because they're not
22 interested in sex.
23 What we have done is we have then looked across
24 the six cycles of treatment, and there was no change in this
25 course during the follicular phase. This was continuous
1 treatment with fluoxetine 20 and 60. It was no worse than
2 during the follicular phase in terms of what they score on
3 sexual functioning, and there is an improvement, an overall
4 improvement, in that question during the late luteal phase.
5 So, overall we have not seen that in this particular population
6 Prozac caused sexual dysfunction.
7 I have to say that we're now looking at how many
8 actually we lost or how many were dropped out of the study
9 because of sexual dysfunction, and I believe that there is
10 a total of 5 subjects.
11 DR. TAMMINGA: Dr. Steiner, could you be more
12 specific about when the improvement occurred in the luteal
13 phase, what percentage of that of the follicular phase -- how
14 high did it --
15 DR. STEINER: It's not back to the same normal
16 baseline, but it's halfway there.
17 DR. TAMMINGA: Dr. Parry.
18 DR. PARRY: I want to address the issue of suicide.
19 Many of these patients may actually present with suicidal
20 ideation. So, it is a potentially lethal illness. There has
21 been some, though maybe erroneously, association with
22 fluoxetine and suicide. I was wondering were patients with
23 suicidal ideation excluded from studies, and if not, was that
24 targeted? What was the effect of treatment?
25 DR. JUDGE: In general, patients with significant
1 coexisting other illnesses, generally medical, were excluded
2 and, as you heard, also patients with other Axis I diagnoses
3 were excluded.
4 Now, specifically in terms of patients presenting
5 symptoms, I'm not aware from the data that we have in these
6 studies, that anyone presented with a suicide attempt as a
7 presenting symptom. I can tell you that from these studies,
8 all of the studies, there was no one who attempted suicide,
9 and I think it's pertinent at this point to perhaps ask Dr.
10 Tollefson to comment in the overall Prozac and the suicide
11 question because, obviously, it's very important.
12 DR. PARRY: But it's also suicidal ideation --
13 DR. JUDGE: Well, remember, the scales that were
14 used -- in this study, for the big study, Dr. Steiner's study,
15 for the Su study, there was a Beck's Depression Inventory scale
16 that looked at, in the follicular phases of that study -- the
17 Beck's is the Beck's, and we saw nothing significant in the
18 follicular phase indicating that in the follicular phase the
19 patients had very, very low levels of symptomatology. You're
20 looking at a score of around 4 on the Beck's, which is really,
21 really very low.
22 In terms of Dr. Steiner's study, study number 1,
23 there was no instrument that captured the suicidal ideation
24 per se, as one would expect, for example, in the HAMD.
25 Dr. Tollefson, if you could comment on the --
1 DR. TOLLEFSON: Well, I think that when we look
2 at spontaneous events in these clinical trials, a suicide
3 attempt or a completed suicide would register as a spontaneous
4 event. None of those were observed in the clinical trial,
5 to answer your question. I think that is consistent with our
6 overall meta-analysis across not only depression but OCD and
7 bulimia where see certainly a higher emergent rate in patients
8 randomized to placebo than we do on active therapy. In fact,
9 active therapy in other indications at least is associated
10 with a reduction in suicidal ideation, as measured by HAMD
11 item 3. Those were gender indicated.
12 We did not have specific suicide indices built
13 into these prospective studies, so one would have to just really
14 rely on the adverse event data, of which there was no evidence
15 of a suicide attempt. There was one drug overdose with
16 cocaine. That is the only event that would map to "overdose"
17 in the entire cohort.
18 DR. PARRY: I think it argues for using the
19 Hamilton or some other scale. I think it argues for using
20 the Hamilton or some item that assesses that in future studies.
21 DR. TAMMINGA: Dr. Judge, would you -- Dr.
22 Steiner, do you want to say something?
24 DR. STEINER: They were screened for major
25 depression, and if they had major depression, including
1 suicidality or not, they were excluded. So, they were not
2 included. But that is not to say that we have not screened
3 them for that.
4 DR. TAMMINGA: Thank you.
5 Dr. Judge, the committee would note that the dosing
6 throughout these three studies is continuous. Certainly there
7 is some question about whether dosing needs to be continuous
8 or whether it can be intermittent. I would wonder whether
9 you would like to comment on that, and also if you could give
10 us some indication of whether Lilly is continuing to pursue,
11 in Lilly-sponsored studies, additional questions in this area,
12 if would be helpful to the committee.
13 DR. JUDGE: Thank you.
14 If you could bring up my slide with respect to
15 intermittent dosing. I've tried to attempt to note the
16 considerations with respect to intermittent dosing on one
18 Intermittent dosing. Certainly there have been
19 anecdotal reports that would indicate intermittent dosing may
20 be effective. I think in order to explore this fully, there
21 are various considerations that one has to think about.
22 First of all, in respect to safety, when we talk
23 about intermittent dosing, as I would understand you, something
24 like 10 to 14 days perhaps of drug before the onset of menses.
25 Firstly, does the intermittent administration of fluoxetine,
1 or indeed any other agent, subject that patient to repeated
2 typical adverse events month after month after month. As you
3 appreciate with fluoxetine, there is a diminution in the
4 adverse event profile with time. If you administer
5 intermittently month after month, does that subject the patient
6 to a continuous high level of adverse event reporting?
7 Furthermore, we know that fluoxetine with its
8 longer half-life -- patients don't really report
9 discontinuation symptoms, but perhaps patients on the maybe
10 more shorter half-life agents may, in fact, report
11 discontinuation symptoms. So, that's important to establish
13 With respect to efficacy, also there's a question
14 of compliance. Will patients comply to treatment that they
15 perhaps have to -- will they find it easy to take treatment
16 for a few days in the month rather than every day?
17 Importantly, this question of kindling. Is the
18 possibility of kindling phenomenon -- is that going to take
19 place in these patients? I think that's also important.
20 With respect to fluoxetine, there is one pilot
21 trial with intermittent dosing with fluoxetine. Again, Dr.
22 Steiner did this trial. This is an open label, single-blind
23 trial in which patients were administered 14 days of fluoxetine
24 treatment prior to menses, and this was compared to continuous
25 dosing 20 milligrams throughout the cycle. There was some
1 evidence of a similar response for both groups of patients.
3 However, it's worth noting that that was not
4 randomized appropriately. In terms of the patients who had
5 a past history of depression, they were given the continuous
6 trial of 20 milligrams of fluoxetine, and patients who did
7 not have a past history of affective disorder were administered
8 intermittent fluoxetine. So, that is an open trial, but
9 nevertheless to your other question, yes, Lilly is pursuing
10 other intermittent studies.
11 In the literature, all of the SSRIs, it's fair
12 to say, have the largest body of data for continuous dosing,
13 roughly at around the same dose that is also applicable to
14 the depression population.
15 DR. TAMMINGA: Could you be specific about what
16 Lilly is doing to address these considerations?
17 DR. JUDGE: There are studies underway with
18 intermittent dosing for fluoxetine.
19 DR. TAMMINGA: Lilly-sponsored studies.
20 DR. JUDGE: Indeed.
21 DR. TAMMINGA: Dr. Temple.
22 DR. TEMPLE: It's not easy to have a discontinuous
23 study of Prozac because 14 days after you stop, with a 14-day
24 half-life, you still have half the same amount of long-acting
25 metabolite on board, which raises the larger question which
1 I know will come up later, but you need to address it too.
2 Is this a sensible approach to an intermittent
3 disease? You basically are on Prozac. You didn't find any
4 abnormalities with sexual function. It's not quite clear to
5 me how hard you looked. But there are consequences to being
6 on an SSRI all the time. So, the committee is going to address
7 that later, but you may want to comment on it in advance.
8 DR. JUDGE: With respect to fluoxetine,
9 continuous dosing of fluoxetine, as one attributes to PMDD,
10 as well as other disorders, we know that fluoxetine at a dose
11 of 20 milligrams, which seems to be the optimal dose in this
12 population, is very safe and very well tolerated. There is
13 very long-term experience and a large body of evidence which
14 relates to fluoxetine with respect to that.
15 I think the question is more around intermittent
16 dosing. Yes, indeed, fluoxetine does have a longer half-life,
17 but nevertheless, if patients are given intermittent dosing
18 from day 1, do they have enough time then to reach that steady
19 state in order to have high levels when patients are off dose?
20 Furthermore, these other safety questions would
21 also be still relevant, and that is ongoing in terms of clinical
22 trials. And there are also pertinent questions to other
23 treatments that are being studied for intermittent dosing.
24 So, the questions are more around intermittent dosing I think
25 than relate to continuous dosing. I think we're fairly
1 comfortable with particularly the safety profile for
2 fluoxetine in continuous dosing. Even though the disorder
3 is just intermittent, as you say, many disorders are
4 intermittent with respect to intensity of symptoms, but
5 nevertheless, a longer-term approach prevents their
6 reemergence and relapse of those symptoms.
7 DR. TAMMINGA: Dr. Temple.
8 DR. TEMPLE: Usually you don't know when a disease
9 is going to relapse. So, you have no choice except to treat
10 continuously. This is a little different. You know exactly
11 when it's going to happen.
12 DR. TAMMINGA: That's what Dr. Endicott
13 emphasized in her presentation.
14 Dr. Fyer.
15 DR. FYER: Sort of two comments. First, on your
16 response, Dr. Judge, I think that there's a big difference
17 between a drug being overall safely tolerated the way these
18 kinds of databases are put together and the question of
19 individual people taking a drug for a very long period of time
20 and it having effects that impact their life. It may not be
21 medically dangerous. I would like to feel that the sponsor,
22 in undertaking to get this kind of indication, would really
23 take that seriously in terms of labeling and how they advertise
24 and promote the drug because we're talking about women who
25 might conceivably take this for a very long time.
1 The second comment I have is actually a question
2 to Dr. Steiner. I was glad to hear that the reason for your
3 dropouts -- it seemed to me that it was just a very hard study
4 to do and that there wasn't something odd going on with the
6 But the question I have is that in most of the
7 studies we have mean scores as opposed to responder outcome.
8 In the one study where we had the CGI used, in fact the 1's
9 and 2's -- it was a small study and there weren't significant
10 differences. I wondered if you could give us from your
11 clinical experience some idea as to whether we're dealing with
12 a situation like OCD where even the people that get better
13 don't get completely better or very few of them do, or if we're
14 dealing with a situation like major depression or panic
15 disorder where a large proportion of the responders are really
16 better, you know, just in terms of risk-benefit ratios, things
17 like that.
18 DR. STEINER: As old as I am and have been in this
19 field, this is the first time that something works. This is
20 the first time that the clinic got flowers from husbands --
22 DR. STEINER: -- because we treated something that
23 sort of restored life in some of these households. This is
24 not something to sneeze at. I don't care about statistics.
25 I'm telling you that clinically this was so impressive that
1 it was almost unbelievable.
2 You were asked who was driving this study. We
3 had 10 or 12 patients that we piloted before we went to Lilly
4 and we said, this is the first time. I have worked with a
5 lot of other compounds. They were all as good or as bad as
6 placebo, and really nothing works after works after three or
7 four cycles. This was the first time that something worked.
8 We went to Lilly and we said we must do that properly.
9 DR. TAMMINGA: Dr. Steiner, did all the patients
10 get a little bit better or did all the patients get totally
12 DR. STEINER: Most of the patients got totally
13 better. I'm talking about the completers.
14 DR. FYER: How would you comment on the third study
15 where they did do CGIs and they didn't find for the 1' and
16 2's? Remember, 1 is completely and 2 is slightly.
17 DR. STEINER: I wasn't involved in that study and
18 I don't think that I should be commenting on it.
19 DR. FYER: Somebody else?
20 DR. JUDGE: Again, for that study, there was
21 clearly a trend towards significance, .07. When one looked
22 at the overall CGI any improvement, then there was a statistical
23 separation. Again, there were small numbers of patients in
24 that study which might have accounted for some of the lack
25 of significance. But generally the results are entirely
1 consistent with the efficacy noted in the other studies.
2 DR. FYER: Respectfully, I just submit that if
3 in fact all the patients got all better, you couldn't have
4 seen that kind of CGI outcome. I don't know what happened,
5 but that's not consistent.
6 DR. TOLLEFSON: I think it might be useful if we
7 put the CGI response slide up for you to take a look at. Again,
8 remembering it's a very small sample size, which increases
9 the hurdle for showing a p value less than .05. If you look
10 just at treatment effect size, you look at the 20 to 60 milligram
11 fluoxetine arm, you're seeing a response rate on CGI 1 or 2
12 that is in excess of 50 percent. I would argue that's as good
13 as any response data, schizophrenia, depression, OCD, that
14 we see in the literature.
15 If you then expand it a little bit more to include
16 3, you see now that we're approaching 90 percent plus who had
17 CGI improvement with fluoxetine, and despite the small sample
18 size, there's a very robust statistical separation between
19 fluoxetine and placebo. You can see that there is a very strong
20 differentiation from that of bupropion.
21 So, I would argue that that's a fairly impressive
22 response rate with CGI, statistics aside for a moment, at either
23 of the score of 1 or 2, which a priori was the primary outcome,
24 or scores 1, 2, and 3 as a secondary analysis.
25 To Dr. Laughren's earlier question, which I'm not
1 sure was fully answered, we do have the sexual functioning
2 item from the PMS scale that we could up for you. I think
3 that might provide some light on the issue of either
4 improvements or, alternatively, deterioration in function
5 between drug and placebo.
6 Go ahead, Rajinder.
7 DR. JUDGE: This is looking from study 1 for those
8 patients during the study who scored on the PMTS patient rated
9 scale. So, this is looking at those patients who had sexual
10 drive increased or sexual drive decreased. For the placebo
11 group, these numbers are 58 and 51 percent, respectively.
12 For the fluoxetine group, roughly around 50 percent each group,
13 and for the fluoxetine 60 milligram arm, again roughly around
14 50 percent each group.
15 So, it seems from the data here that, overall,
16 sexual drive could increase or decrease with respect to any
17 treatment, and also that the differences between fluoxetine
18 and placebo were not statistically significant as noted here.
19 DR. TAMMINGA: Dr. Laughren.
20 DR. LAUGHREN: Is this follicular phase or luteal
21 phase data?
22 DR. JUDGE: This is luteal -- at any time, at any
23 time during any visit. And patients were scored in follicular
24 and luteal visits, so this is at any point in that study they
25 could have sexual drive increased or decreased.
1 In fact, this is consistent with other studies
2 with fluoxetine where we note that patients -- if you just
3 flick on to the second carry-on in this file, with the overall
4 depression database patients improving or not improving. It's
5 the last slide in this file.
6 So, this is looking at a meta-analysis of several
7 studies with fluoxetine. It looks at another parameter, the
8 SCL-58, and looking obviously at females only. We see that
9 for overall fluoxetine database, indeed, some patients do worse
10 with the use of fluoxetine on fluoxetine or placebo.
11 Nevertheless, some patients' sexual dysfunction remains the
12 same, but some patients actually do improve and a greater number
13 of patients do improve on fluoxetine versus placebo.
14 So, I think the sexual dysfunction question in
15 terms of what we know less about the PMDD population, we know
16 less about how these patients are at baseline, how is their
17 sexual function at baseline. That's really information that
18 is not very clear at the moment.
19 DR. TAMMINGA: Yes, Dr. Altemus.
20 DR. ALTEMUS: Do you have any data about
21 anorgasmia? Because that's really the main side effect.
22 Would that have been picked up in any of the measures?
23 DR. JUDGE: If we could go back to the first slide
24 of this file, the overall treatment adverse events for all
25 three studies. Specifically they're items relating to sexual
1 dysfunction in each of the three studies. For study number
2 1, anorgasmia was reported in 0 placebo patients, in 1
3 fluoxetine patient in 20, and again a similar number for
4 fluoxetine 60 milligrams.
5 For decreased libido, again small numbers of
6 patients reported decreased libido in this trial, as you can
7 see here. Highest numbers for fluoxetine 60, but then again
8 we're talking about 8 patients, which is 7 percent. The p
9 value was not statistically different for the placebo versus
10 fluoxetine in these groups.
11 In these studies you see perhaps a higher level
12 of reporting, but it's more important to note that the way
13 that these adverse events were not just spontaneously -- they
14 were not just treatment emergent. They were any adverse events
15 collected at any point. So, one would expect a higher level
16 of reporting.
17 Again, most importantly to note, the differences
18 between fluoxetine and placebo were not statistically
19 significant for any measure. That's the data that was evident
20 from these studies, the PMDD studies.
21 DR. TAMMINGA: Dr. Laughren.
22 DR. LAUGHREN: One comment on these data. In all
23 cases you're relying basically, I think, on spontaneous
24 reporting of those events, and you may see a different picture
25 if you have sensitive scales designed to look at sexual
2 DR. JUDGE: Yes, indeed. The interesting point
3 would be also again to exploit such scales at baseline as well
4 to see what the baseline level of functioning was.
5 DR. TAMMINGA: Any additional questions from the
6 committee for Dr. Judge or for Lilly?
7 (No response.)
8 DR. TAMMINGA: Thank you, Dr. Judge.
9 Before we break for lunch, we have one person who
10 has requested a time in open hearing to speak on this particular
11 topic. I'd like to ask Dr. Sherry Marts from the Society for
12 Women's Health Research to come forward and give her remarks.
13 Also, I would ask, if you would, in the spirit
14 of disclosure, to indicated whether or not you actually get
15 money, your organization gets money, from Lilly, and give us
16 an idea of what percent of your finance of that might be.
17 DR. MARTS: My name is Sherry Marts. I'm
18 Scientific Director for the Society for Women's Health
19 Research, which is a Washington-based advocacy group dedicated
20 to improving the health of women through research. We were
21 founded in 1990 when we brought national attention to the need
22 for research on conditions that affect women not only solely
23 because they're women but also differently from men or more
24 often than men.
25 I would like to say that we do get support from
1 Eli Lilly. They have been a longtime supporter of our efforts,
2 and they provided a grant for the consensus roundtable on PMDD,
3 which I'm going to talk about today. This was convened by
4 the society in 1998, and a peer-reviewed summary of that
5 conference was published in the Journal of Women's Health this
6 past June. I have a few copies of it. Unfortunately, I didn't
7 get the speaker's guidelines in time, and I only brought four
8 copies. But if you're willing to share, I'd be happy to hand
9 those out.
10 The group that assembled for that conference
11 included experts in psychiatry, psychology, gynecology, and
12 epidemiology. This conference report was published as a
13 peer-reviewed paper, as I mentioned. Among the conclusions
14 of that conference are that PMDD is a distinct clinical entity
15 with a clinical picture that is not a typical picture for
16 depression, mood, or anxiety disorders, and in particular,
17 internal tension, anger, and irritability are characteristics
18 of PMDD.
19 The key differences between PMDD and other
20 disorders is the clear onset and disappearance of symptoms,
21 both linked to the menstrual cycle. There's considerable
22 stability in the course of PMDD from cycle to cycle and over
23 time in the absence of treatment.
24 The conference noted that PMDD is a chronic
25 condition that in some women can worsen over time. Age of
1 onset varies and it can be any time after regular menstrual
2 cycles are established.
3 PMDD differs from other disorders in that there
4 may or may not be comorbidity with other mood disorders, and
5 unlike in depression, in PMDD the
6 hypothalamic-pituitary-adrenal axis often functions normally.
7 Blocking the menstrual cycle, as happens in pregnancy, can
8 eliminate PMDD, but has no effect on other mood disorders,
9 and after pregnancy, symptoms return once the menstrual cycle
10 is reestablished.
11 Finally, the consensus conference concluded that
12 there is biological evidence for the involvement of the
13 serotonin system in PMDD.
14 Now, I want to say that the society does not have
15 expertise in the evaluation of therapeutics. I'm not here
16 to encourage or discourage the approval of any particular
17 pharmaceutical agent for treatment. We gladly leave those
18 decisions to the experts.
19 But we want to emphasize that PMDD is a severe,
20 often debilitating disorder that is distinct from premenstrual
21 syndrome. PMDD's symptoms significantly interfere with a
22 woman's life, preventing a sufferer from functioning
23 effectively at work and at home.
24 We're concerned that this disorder may go
25 unrecognized, undiagnosed, and untreated in many women. Among
1 the barriers to diagnosis and treatment are the stigma that
2 may be attached to a condition associated with the menstrual
3 cycle. There's still a negative connotation against seeking
4 treatment for something that's just in your head or just PMS.
5 The admission that treatment is needed is seen as a weakness.
6 Because this condition is not well understood, physicians
7 may not recognize the signs and symptoms or know how best to
8 diagnose PMDD by distinguishing it from mood disorders.
9 The possibility of a medical treatment for this
10 disorder is heartening and it's evidence that women have
11 succeeded in bringing attention to this condition to their
12 health care providers.
13 We as the Society for Women's Health Research are
14 committed to reducing the stigma of diagnosis with PMDD and
15 committed to educating women that PMDD can be diagnosed and
16 treated and that symptoms are, as I said, not just part of
17 being a woman or all in your head. We encourage women to
18 consider all treatment options and to insist on treatment that
19 is effective and appropriate to the severity of their symptoms.
20 We encourage the members of the advisory panel to consider
21 the significance of this disorder for the approximately 5
22 percent of menstruating women who suffer from it.
23 Thank you.
24 DR. TAMMINGA: Thank you very much, Dr. Marts.
25 Although Dr. Marts was the only person who
1 requested a spot for remarks, this is the open public hearing
2 part of the meeting, and I'd like to ask if anybody else has
3 any statement to make.
4 (No response.)
5 DR. TAMMINGA: In that case, I think we ought to
6 break for lunch. I would like people to come back promptly
7 at 1 o'clock, and at that time the committee will have a
8 discussion both about the diagnosis and about the efficacy
9 and safety questions.
10 DR. TITUS: There is a table reserved for you in
11 the restaurant, and I would like to caution you that the
12 conversation needs to be about weather and neutral topics.
13 The topic we're discussing -- you need to come back here and
14 do it publicly.
15 (Whereupon, at 11:55 p.m., the committee was
16 recessed, to reconvene at 1:00 p.m., this same day.)
22 AFTERNOON SESSION
23 (1:03 p.m.)
24 DR. TAMMINGA: I'd like to reconvene the meeting.
25 The committee is now in the committee discussion
1 and deliberation portion of our meeting. I'd like to focus
2 our discussion on the items that Dr. Laughren raised for us
3 this morning about the questions that the FDA has of this
4 indication, both of the indication and of the drug, and remind
5 the committee that the question that Dr. Laughren would like
6 some discussion of, first of all, is the general question
7 regarding PMDD as a new indication. With some more specific
8 questions, Dr. Laughren asked about the relevance of the DSM-IV
9 appendix status for PMDD, PMDD as a distinct disorder,
10 distinguished from MDD, the relationship of PMDD to PMS, and
11 as a follow-up question, the possibility, although we're not
12 considering a drug for this today, the related question of
13 PMS as a candidate for an approved indication.
14 So, I'd like us to start this afternoon meeting's
15 deliberating about these kind of things. I would actually
16 invite our consultants, who are experts in this area, to perhaps
17 take a lead-off in the conversation. Dr. Parry.
18 DR. PARRY: Well, I think that the studies that
19 were presented this morning and most other studies that are
20 accepted for publication use the criteria for premenstrual
21 dysphoric disorder. There was a lot of deliberation going
22 into making those criteria. The term "premenstrual syndrome,
23 PMS" is amorphous and ill-defined. There's no definition of
24 it. So, to say that one is a subset of the other or how it's
25 differentiated I think is really a moot point.
1 I would strongly advise against fluoxetine being
2 considered for use in this amorphous term PMS because the
3 studies that the decision will be based on do not
4 use -- specifically and go to a lot of trouble to define criteria
5 for PMDD. So, we really don't have information on this
6 amorphous, ill-defined PMS. So, I think we should confine
7 our focus of attention on the studies and even treatments to
8 the very carefully and very rigorously defined premenstrual
9 dysphoric disorder. It's more rigorous than most other
10 psychiatric disorders. The terminology that's used in the
11 DSM-IV was based on pooling studies from across the country.
12 So, I think given that background, the focus should be on
13 premenstrual dysphoric disorder.
14 DR. TAMMINGA: Dr. Thys?
15 DR. THYS-JACOBS: I think PMS is a definitive and
16 distinct disorder. Women who have had it -- and I would say
17 across the board, there are about 70 to 80 percent of
18 premenopausal women who are suffering with premenstrual
19 syndrome. So, I think it's a very distinct disorder.
20 DR. PARRY: What's the definition?
21 DR. THYS-JACOBS: I think it's been defined as
22 an occurrence of physical and affective mood symptoms that
23 occur during the luteal phase of the cycle, as distinct from
24 the follicular phase. Now, exactly what that ratio is in terms
25 of symptomatology -- is it a 30 percent increase? Is it a
1 50 percent increase? Is it 100 percent? We all know that
2 premenstrual syndrome exists, it's a luteal phase disorder,
3 and we know that these symptoms indeed occur.
4 Is PMS distinct from PMDD? Now, there is a
5 definition in the DSM-III and the DSM-IV that cites
6 specifically that five out of a lot of these symptoms define
7 PMDD. I think it's fine. I think for a woman who has the
8 very severe end of the spectrum -- and I don't see any difference
9 between PMDD and PMS, not at all. In my practice and in my
10 research, I don't see any distinction between the two. What
11 I do see is that the women who come in who have PMDD indeed
12 have a prominence of affective symptoms. Are they more
13 symptoms than physical symptoms? I don't find that, not at
15 But I do agree that it is a distinct disorder.
16 It's distinct in terms of defining it from other depressive
17 or other mental or mood disorders.
18 DR. TAMMINGA: You're talking about PMDD.
19 DR. THYS-JACOBS: PMDD and severe PMS. I do not
20 consider them as different. Not at all.
21 DR. TAMMINGA: Dr. Altemus?
22 DR. ALTEMUS: I imagine that if this is approved,
23 in the general public and in primary care, people aren't really
24 going to know how to diagnose PMDD, and it will probably be
25 used very widely for anyone with any premenstrual symptoms
1 at all.
2 DR. THYS-JACOBS: I think the criteria are pretty
3 well spelled out in terms of the DSM-III and DSM-IV. I think
4 women know what PMS is. I don't think there's any question
5 about it. The people that I speak to who are seeing these
6 women, the gynecologists, the primary care internists, they
7 all know what PMS is. They all know that it's a group of these
8 symptoms, whichever symptoms you want to define, that occur
9 specifically during this phase of the menstrual cycle, the
10 luteal phase of the cycle, and remits with menses or the
11 follicular phase. That is PMS.
12 Is PMS distinct from PMDD? No, it is not. It
13 is not different.
14 DR. ALTEMUS: Well, I don't think we can make the
15 leap that the drug is effective for PMS. I mean, there have
16 been two studies in PMDD. Even though you intuitively think
17 that it will be effective for PMS.
18 DR. THYS-JACOBS: I'm not saying that. I'm not
19 saying that this particular drug --
20 DR. TAMMINGA: We're really just talking about
21 the diagnoses now just as a starter.
22 DR. THYS-JACOBS: -- should be indicated for
23 premenstrual syndrome. When I think of premenstrual syndrome,
24 I think of the global syndrome and I think of the physical
25 and the affective, the water retention symptoms and the pain
1 symptoms. I think of the whole global syndrome. I think what
2 we've seen is that we have defined PMDD mostly as an affective
3 syndrome with these particular group of symptoms, and this
4 drug is or is not effective for this particular group of
5 symptoms in this disorder. Is it effective across the board
6 for the entire syndrome? We haven't seen that.
7 DR. PARRY: What's your operational definition
8 in terms of severity of symptoms? They do make a distinction
9 between major depressive disorder and depressive
10 symptomatology. So, I think not only in terms of the nature
11 of the symptoms -- and you specified timing, but what about
13 DR. THYS-JACOBS: I think the timing makes the
14 syndrome. There's no question about it. It's the timing
15 during the luteal phase. That's why it was late luteal phase
16 dysphoria and premenstrual dysphoria. It's the timing of the
17 syndrome that makes this different from major depression and
18 all the other mood disorders. There's absolutely no question
19 this disorder remits, subsides, ends with the beginning of
20 menses, period. So, it's the timing. Whatever symptom you
21 want -- it could be cravings, it could be pain, it could be
22 depression -- any symptom you want that occurs during this
23 luteal phase of the cycle and recurs, you have PMS.
24 DR. PARRY: Well, but again it's severity. Would
25 you use fluoxetine to treat a little bit of breast swelling?
1 DR. THYS-JACOBS: I'm talking about PMS not versus
3 DR. PARRY: Yes, but what are your severity
4 criteria for PMS is what I'm asking.
5 DR. THYS-JACOBS: There's no severity. It's the
6 timing of those symptoms that occur during the luteal phase
7 of the cycle.
8 DR. ALTEMUS: But you're saying it's a disorder.
9 DR. THYS-JACOBS: It's a disorder. It's not a
10 disease. It's an occurrence of symptoms, whatever group of
11 symptoms the woman has for that particular phase of the cycle.
12 DR. ALTEMUS: Would you say 80 percent of people
13 have this disorder?
14 DR. THYS-JACOBS: Yes, absolutely. I think 70
15 to 80 percent of women who are walking around have suffered
16 at some point in their life with PMS. There's absolutely no
17 question. If you're out there and you're seeing women on a
18 daily basis, yes, there are women out there who are suffering
19 with this syndrome.
20 DR. TAMMINGA: Let me focus this discussion, if
21 you would, for a minute on PMDD since the question on the plate
22 -- we have sort of a minor question about PMS, but the major
23 question on our plate is the PMDD question, the suitability,
24 if you will, of the PMDD as a diagnosis and as an indication
25 for drug treatment.
1 Good, Dr. Dominguez.
2 DR. DOMINGUEZ: I would like to ask the
3 consultants, taking a step back from the discussion that just
4 happened, why was LLPDD not elevated in DSM-IV to a distinct
5 disorder? We learned from Dr. Endicott that there was a lack
6 of consensus in that area. What were the major issues that
7 drove this lack of consensus? Can the consultants or perhaps
8 even Dr. Endicott enlighten the committee as to why?
9 DR. PARRY: Well, from my view, as Dr. Endicott
10 indicated, there was consensus on the nature of the symptoms.
11 I view it that was a political decision, not a scientific
12 one. But the concern was that if it was put in the main body
13 of the text, it would implicate all women as having a disorder,
14 even though it specified that based on the estimates we had
15 from studies, only 5 percent of women met criteria, but the
16 concern was that it would stigmatize women. So, I'm happy
17 to hear other comments, but I think it was primarily a political
18 decision, not a clinical or scientific one.
19 DR. TAMMINGA: Why was the diagnosis changed from
20 LLP -- whatever that was -- to the PMDD?
21 DR. PARRY: Well, it was first listed in the
22 DSM-III-R as late luteal phase dysphoric disorder because the
23 attempt was to define it as carefully as we could into the
24 luteal phase of the menstrual cycle. But I think it was changed
25 to premenstrual dysphoric disorder because late luteal phase
1 dysphoric disorder was a bit cumbersome. It got referred to
2 as L squared PD squared, and the premenstrual terminology was
3 more user friendly.
4 DR. DOMINGUEZ: So, if the issue was political,
5 as you say, then this disorder is going to remain in the
6 appendices of future DSMs forever and ever? I think it is
7 worded as studies that need further research and further
8 verification. What else has to happen in order for this to
9 rise to a clear Axis I category?
10 DR. PARRY: I think the majority of the people
11 who were on the board wanted it to go in not under the
12 appendices, but the board, because of a few dissenting opinions
13 and because of the public controversy, overrode that. So,
14 my view would be we made a little bit of progress just getting
15 it in, and I would hope that the next step would be to put
16 it in.
17 Would you share that, Jean?
18 DR. TAMMINGA: Dr. Endicott.
19 DR. ENDICOTT: Dr. Parry and I were both on the
20 work group that were advisory to the nomenclature committee,
21 and it was fairly apparent from day 1 of that work group or
22 advisory group that there were going to be issues around this
23 political issue and that there was going to be some disagreement
24 on the ultimate recommendation. But the group was able to
25 work together very well on the subset of criteria and on the
1 evidence to support those criteria.
2 There was a desire on the part of the nomenclature
3 committee to have some kind of consensus recommendation, and
4 they finally accepted that there was not going to be a consensus
5 from our group. So, they were presented with the evidence.
6 As Barbara said, both in the advisory committee
7 and in the overall nomenclature committee, most of the people
8 were in favor of moving it up from the appendix. But there
9 were other issues going on, and this is always a committee
10 kind of decision in a general setting in which political issues
11 do get considered.
12 DR. TAMMINGA: Dr. Laughren.
13 DR. LAUGHREN: Let me try and clarify the question
14 that I'm interested in having answered. Actually there are
15 two questions.
16 One has to do with this entity PMDD and whether
17 or not that's a reasonable clinical entity to focus on as a
18 claim, an indication in labeling. In terms of this
19 application, that's the only entity that we're focusing on
20 because that's what they studied.
21 The second question really relates more to our
22 advising other companies who are interested in this area and
23 what to tell them about broader claims that they might be
24 approaching. There is some interest in looking at this broader
25 entity of PMS. My question is, separate from your answer to
1 whether or not you think PMDD is a reasonable entity, is this
2 broader entity something that, in a sense, is ready for prime
3 time in terms of an indication? Is it well enough defined?
4 Is there a consensual agreement about what the diagnostic
5 criteria are so that a company could reasonably run a
6 development program and submit an application for this more
7 diffuse claim? That's really the question.
8 DR. TAMMINGA: So, I'd like to hear from both the
9 consultants and the committee members in response to each of
10 those questions, addressing the first question about PMDD first
11 and then the broader question about PMS.
12 Abby, do you want to go first?
13 DR. FYER: Sure. Actually I guess I would take
14 it from the other point of view. What is it about this disorder
15 that makes you feel that it might not be something viable to
16 address as an indication for a drug treatment? Usually we
17 look at epidemiologic data, we look at distress and impairment,
18 et cetera, we look at distinctness from other disorders. It
19 may be slightly different in some aspects from others, but
20 I guess I'm missing the point. Maybe it's a little new in
21 the history of the explicit definition of the disorder, but
22 not really much different than panic which had a similar long
23 history unofficially and then came into --
24 DR. LAUGHREN: Actually my own bias is in favor
25 of what you said, that this entity is reasonably well defined.
1 Really more my question has to do with this broader entity
2 of PMS and whether or not that's a candidate for drug
3 development. The questions I have listed here. Whether or
4 not this entity is distinct from depressions, say, major
5 depression, I think that has largely been addressed, but other
6 committee members may want to weigh in on that. But really,
7 my main question has to do with future development programs
8 for possibly this broader entity of PMS and whether or not
9 that's a reasonable course for companies to be taking.
10 DR. FYER: You mean whether or not it's reasonable
11 to advise people to develop indications for PMS.
12 DR. LAUGHREN: Right. Is PMS at this point well
13 enough defined, well enough accepted in the community, and
14 does it have enough agreement about diagnostic criteria that
15 companies could reasonably go down that path and submit
16 applications for that entity?
17 DR. TAMMINGA: Well, we've had two answers from
18 our committee. One is yes and one is no. So, why don't the
19 two of you restate your opinions, and then the rest of the
20 committee can comment?
21 DR. THYS-JACOBS: I think PMS, premenstrual
22 syndrome, is a distinct, defined disorder. There is
23 absolutely no question that what makes this group of symptoms
24 with this disorder or this phenomenon different from any other
25 affective or any other disorder is the occurrence of these
1 particular group of symptoms during the luteal phase of the
2 cycle. I can't stress more to you that I really believe it.
3 We have all recognized this for centuries. It has been spoken
4 about, written about by historians. We've defined it.
5 I think there was, indeed, a National Institute
6 of Mental Health consensus that defined a 30 percent increase
7 from luteal to follicular phase, that you had to know that
8 there was a difference in scores, in visual analog scales,
9 that there was, indeed, this difference of symptoms during
10 the luteal phase to the follicular phase of the cycle.
11 So, we know what it is. We've defined it basically
12 as a group of symptoms that occur. It's phase-related. It's
13 temporally related to the menstrual cycle.
14 DR. LAUGHREN: Are you talking about PMS or are
15 you talking about something called severe PMS?
16 DR. THYS-JACOBS: I'm talking about premenstrual
17 syndrome, PMS.
18 DR. LAUGHREN: So, you're not making -- earlier
19 you were talking about --
20 DR. THYS-JACOBS: No. I am not making any
21 distinction in terms of severity, no, because I believe and
22 my research has shown that, that the occurrence of premenstrual
23 symptoms is abnormal. It's like being pregnant. If you're
24 pregnant, you're pregnant. If you're a little pregnant,
25 you're 1 month pregnant, versus 9 months pregnant. It's the
1 pregnancy that makes the difference in terms of the
2 reproductive cycle.
3 What I'm saying to you is that the occurrence,
4 the presence of symptomatology during the luteal phase of the
5 cycle is not normal.
6 DR. TAMMINGA: Dr. Temple.
7 DR. TEMPLE: Couldn't the relationship between
8 these two terms be described as PMS is the larger term, and
9 when the dysphoric symptoms are particularly prominent and
10 are also premenstrual and have the appropriate timing, you
11 call it premenstrual dysphoric disorder?
12 DR. THYS-JACOBS: Yes. I look at it that way.
13 DR. TEMPLE: So, one is perhaps a subset of the
15 DR. THYS-JACOBS: Yes. Yes.
16 DR. TEMPLE: And the way you get into a trial of
17 premenstrual dysphoric disorder is you meet the criteria for
19 DR. THYS-JACOBS: You focus more on the affective
21 DR. TEMPLE: If someone wanted to do PMS more
22 generally, then they would be focusing on whatever the symptoms
23 that happen to accompany this person's premenstrual syndrome,
24 but they might even not have a dysphoric disorder. They might
25 just be bloating or something. But you'd say that's still
1 PMS, but no one would say that's premenstrual dysphoric
2 disorder because they're not dysphoric.
3 DR. THYS-JACOBS: Right. I definitely agree with
5 I should actually qualify this. What I look for
6 is a difference between follicular and luteal phase. When
7 I treat a woman who has premenstrual syndrome, what I look
8 for after treatment is the equalization between luteal and
9 follicular mean scores. That's what I look for. It doesn't
10 have to be absent. I like to see absent symptoms. So, when
11 I even tell you the presence, that's not really right. It's
12 not the presence. It's the increase in symptoms during the
13 luteal phase compared to the follicular phase. That's what
14 I look for. I look for the decrease in luteal phase
15 symptomatology, those mean scores -- or you could use a visual
16 analog scale -- between one phase of the cycle and the other,
17 and I look for the equalization between these phases. Then
18 I say, yes, this woman is adequately treated.
19 DR. TAMMINGA: Dr. Hamer has a comment.
20 DR. HAMER: No. Actually, as usual, I'm just
21 confused. Did I hear you say at one time that 80 percent of
22 women had this and then at another time that it was abnormal?
23 DR. THYS-JACOBS: Yes.
24 DR. HAMER: I just wanted to make sure I heard
1 DR. THYS-JACOBS: Yes. I think 70 to 80 percent
2 of women are suffering with premenstrual symptomatology.
3 DR. HAMER: Then why isn't that normal?
4 DR. THYS-JACOBS: Well, it depends on what normal
5 is. If somebody comes in and says, I feel wonderful and their
6 vitamin D level is 0 and then you start treating them, and
7 they say, wow, I've never felt better, what's baseline, what's
8 normal? I don't have the answer to that question.
9 DR. TEMPLE: It's like asking whether presbyopia
10 is normal or not. It's a matter of definition. Everybody
11 gets it, but the lens isn't working anymore.
12 DR. TAMMINGA: The committee would like to hear
13 from both Dr. Parry and Dr. Altemus about the broader question
14 of PMS and the suitability, if you will, of PMS as a diagnosis.
15 DR. ALTEMUS: Well, I think PMDD is definitely
16 well defined, and that's definitely an indication.
17 I have a more conceptual question I guess. For
18 a drug to be approved, does there have to be a disorder with
19 diagnostic criteria? I'm just thinking of, say, pain
20 medication. Do we approve pain medications for arthritis?
21 Does there have to be a severity criteria for the drug to be
23 DR. TAMMINGA: Maybe Dr. Laughren can answer or
24 Dr. Katz can answer this question.
25 DR. ALTEMUS: I think the problem with PMS is
1 there's no severity criteria, like 80 percent of women have
2 some symptom. I'm wondering for a drug to be approved, do
3 we have to have a defined disorder with diagnostic criteria,
4 or is it possible to have, say, a pain medication approved
5 that works for all different severity of pain? Do you know
6 what I'm saying?
7 DR. KATZ: Well, you could have a pain medication
8 approved I suppose for all different severities, and they are.
9 But pain is something that everybody understands. I mean,
10 it's commonly accepted that pain is a syndrome, if you will,
11 that's commonly understood. Everybody knows what you mean
12 when you say I have pain, although there are different types
13 of pain, of course.
14 The question here that Tom is asking, that we'd
15 like to hear what everybody has to say about, is, is PMS so
16 well-defined, so clearly understood by the community and
17 accepted as a bona fide diagnosis in the community so that
18 we could reliably identify people who have PMS, know what their
19 symptoms are, and that there would be a common understanding
20 so that we could write a label that says, this drug is approved
21 for patients with PMS?
22 DR. ALTEMUS: Well, if you wanted to do that, you'd
23 have to have severity criteria, and the way you're describing
24 it right now is there are no severity criteria. It's just
1 DR. THYS-JACOBS: No. I'm not --
2 DR. KATZ: I don't know exactly what you mean by
3 severity criteria, other than everything is on a continuum.
4 First you're normal with anything and then eventually you're
5 abnormal. So, I guess in some sense any disease state is
6 defined almost by a severity criteria.
7 DR. TEMPLE: You could use presence or absence,
8 that kind of thing.
9 DR. ALTEMUS: I guess what I'm saying is I think
10 right now PMS is in the same sort of realm as pain, that it's
11 a subjective report, and there really aren't diagnostic
12 criteria for what's the severity of bloating or discomfort
13 that's defined as PMS.
14 DR. TEMPLE: You could develop one, and in fact
15 you have to develop one or you won't be able to detect
16 improvement. So, make a visual analog bloating scale and check
17 it out and see if people -- it's subjective anyway. Actually
18 nobody gave weight change measurements with this drug, which
19 I meant to ask about and forgot to. But they're subjective
20 symptoms. There are always ways to develop a scale for those
21 things. People do it when they're forced to because they want
22 to study a drug usually.
23 DR. ALTEMUS: No. I agree there are scales. But
24 right now, I'd say for PMS, as it's generally understood,
25 there's no diagnostic criteria.
1 DR. KATZ: Well, then how do you know somebody
2 has it?
3 DR. ALTEMUS: It's like pain.
4 DR. TEMPLE: But I think we heard this. They have
5 cyclical symptoms of one kind or another and they vary, that
6 come at that time --
7 DR. KATZ: Right.
8 DR. TEMPLE: -- and are not there otherwise.
9 DR. KATZ: Right, but the question is if you're
10 developing a treatment for it, do you study people who only
11 have bloating, only have breast tenderness, who are dysphoric?
12 That's what we're asking.
13 Ordinarily when we consider an application for
14 an indication, the indication is something that is generally
15 recognized as being a bona fide, reliably identifiable entity
16 by the community so that people know what Parkinson's disease
17 is, people know what major depression is. It's well understood
18 what those things consist of.
19 Here the question is, how do we define PMS? Does
20 the field in general believe that one definition as opposed
21 to another is acceptable? That sort of thing.
22 DR. TAMMINGA: Do you have another comment, Tom?
23 DR. LAUGHREN: Well, I guess another question that
24 I have, and this is really a more subjective thing. It's true
25 that we approve drugs for something like pain or for nausea
1 which are conditions which people understand. You don't have
2 to have, in a sense, diagnostic criteria.
3 I guess my question is here what we're talking
4 about is committing patients with this entity to possibly
5 decades of treatment, of continuous treatment, with a drug
6 which has some risk associated with it. The question is, if
7 this entity is so vaguely defined that it exists in 80 percent
8 of menstruating women --
9 DR. THYS-JACOBS: I can't imagine that you're
10 saying that this is vague. I think it's very clear-cut. I
11 think it's cyclical. It's recurrent. There's a group of
12 symptoms that are increased during the luteal phase of the
13 cycle. I cannot see that as vague. If this group of symptoms
14 is increased by 30 percent, 50 percent, 100 percent compared
15 to the follicular phase of the cycle, for me that is a definition
16 of what this syndrome is all about. I don't think you really
17 have to say does this person who is suffering with premenstrual
18 syndrome have this particular symptom. No, we don't look at
20 You have to go back in history and realize that
21 there have been clearly over 150 symptoms that have been
22 associated with this syndrome. We've come down to maybe we've
23 agreed to maybe 15, 17, or 20 symptoms. Do you want to look
24 at the whole syndrome? That's fine. What I look for is this
25 group of symptoms that occur and it's the change from the
1 follicular phase, and that's the definition. I think that's
2 clear. That is not vague.
3 DR. TAMMINGA: Dr. Parry, do you want to weigh
4 in on this question? This is the PMS question.
5 DR. PARRY: Well, to try to go back to the original
6 question, I do think that the disorder, premenstrual dysphoric
7 disorder, as defined in the DSM-IV, has substantial evidence
8 for clinical viability, the nature of the symptoms being
9 primarily affective in nature, the severity of the symptoms.
11 When the DSM was initially developed, the
12 definition of psychiatric disorder was that condition which
13 causes symptoms of distress as well as a certain amount of
14 impairment in social or occupational functioning. Now, for
15 research criteria, we may use a cutoff score, but as with any
16 other psychiatric disorder, it has to impair some aspect of
17 social, occupational, or even school functioning, as Dr.
18 Endicott reviewed this morning.
19 Then the other nature of the diagnostic criteria
20 is the distinct timing of the symptoms in relationship to the
21 menstrual cycle. Many women may have cyclic mood symptoms,
22 but they may bear no relationship to the timing of the menstrual
23 cycle. And retrospective reports on that are notoriously
24 unreliable. So, a prospective documentation of symptoms I
25 think is critical in making the diagnosis.
1 So, not only the nature, the severity and the
2 timing of the symptoms have been, I think, very carefully
3 described in the premenstrual dysphoric disorder, but the
4 associated features of the course of the illness, its
5 inheritance patterns have also been described, and also its
6 relationship to other psychiatric conditions and differential
7 diagnoses. So, I do think that premenstrual dysphoric
8 disorder is a well-defined, viable clinical entity.
9 I do not think that premenstrual syndrome is a
10 defined clinical entity, and I think that there would be a
11 great risk in trying to develop a drug treatment for a very,
12 what I consider, ill-defined syndrome. It would be
13 comparable, in my view, to taking someone with -- let's say,
14 developing a drug treatment -- all the work that has gone into
15 defining depression. Granted, major depression may be a
16 spectrum illness and needs to be differentiated from grief
17 or just feeling cranky or having some irritable symptoms.
18 But I don't think it's advisable to develop drug treatment
19 for something that's not a disorder that you could get recurrent
20 pain syndrome, you could get very minor symptoms, and to use
21 a drug treatment to mitigate those symptoms I have to say I
22 think would be very inadvisable.
23 DR. TAMMINGA: Dr. Geller.
24 DR. GELLER: I think it will help if we make a
25 distinction between what men accept and what women accept as
1 commonly accepted entities. I don't want to be disrespectful
2 to the meeting, but if this were a Perry Mason show, I'd ask
3 all women who think PMS doesn't exist to raise their hand.
4 I think there is such wide clinical acceptance
5 that I think that really is not so much the issue as the
6 definitional issues in terms of being scientists. We don't
7 have good measures that distinguish what the impairments are
8 with PMS from non-impairments. The challenge to the companies
9 who want to use this as an indication would be to do what Dr.
10 Temple just said, develop instruments that are validated and
11 reliable that can measure PMS symptoms, and then you can define
12 an entity. From that, you can go on to drug treatment.
13 DR. TAMMINGA: Barbara, the difference between
14 like an entity or a diagnosis, that PMS is in your point of
15 view a diagnosis, as well as commonly accepted entity.
16 DR. GELLER: I think what we generally do is we
17 make things a diagnosis if they produce impairment, and I think
18 until we have validated, reliable scales that show what the
19 impairment is, we ordinarily don't call it a diagnosis. The
20 experts are telling us that at this point in time, those scales
21 don't exist for PMS.
22 DR. THYS-JACOBS: No, that's not correct. There
23 are scales for premenstrual syndrome.
24 DR. GELLER: Well, if there are scales and you
25 can show --
1 DR. THYS-JACOBS: That's absolutely not correct.
2 DR. GELLER: -- the severity, then I'm
3 misunderstanding Dr. Parry and I need correction. I thought
4 you were saying at this point in time you didn't feel that
5 severity of PMS symptoms could be measured in a way that would
6 be appropriate for drug studies.
7 DR. PARRY: Oh, no, I did not mean to imply that.
8 I think that the scales that were reviewed this morning --
9 the visual analog scale has been found to be one of the most
10 sensitive markers, and it is a requirement to meet criteria
11 for premenstrual dysphoric disorder that ratings be done on
12 a daily basis over 2 months.
13 DR. GELLER: Right. No, for PMS, not the PMDD.
14 DR. PARRY: No. I'm talking about premenstrual
15 dysphoric disorder. The symptoms that were listed this
16 morning that are in the DSM-IV are listed by frequency of
17 occurrence based on reports made at least five different
18 centers across the U.S.
19 DR. GELLER: Dr. Parry, I think maybe we're having
20 definitional problems. I don't think anybody listening to
21 Dr. Endicott and the other presentations this morning has
22 doubts about PMDD being a distinct entity where you can show
23 severity in symptoms. My understanding is the FDA wants to
24 know about the broader question of PMS, and here it sounds
25 like there's a difference of opinion about how distinct an
1 entity that may be in terms of what we can measure with rating
3 DR. TAMMINGA: I'm hoping that Dr. Endicott might
4 be willing to weigh in on this question. Even though you're
5 not on the committee, we'd love to hear from you.
6 DR. ENDICOTT: I think I find myself falling kind
7 of in the middle in the sense that I do think that most women
8 know what PMS is and that they could give us a description.
9 We could just tap any woman in the room and most of the men
10 also could give us a clinical description of PMS that we would
11 all nods our heads and say yes. However, when it gets to the
12 issue of whether all PMS is a disorder or not, that's where
13 I do part company with Dr. Jacobs.
14 I think of it along the lines of a continuum ranging
15 from just perceptible changes usually in physical symptoms
16 that maybe get worse and worse and involve more and more
17 symptoms, and at some point along that continuum, it becomes
18 bothersome enough for the woman herself and for her mate or
19 her children or her coworkers that treatment is warranted.
20 Certainly all of us who do studies of PMDD do have women that
21 we don't put into our formal protocols because they don't meet
22 our most stringent criteria for PMDD, but they do have moderate
23 to severe PMS that is causing them problems. We tend to sedge
24 them over into another protocol or treat them openly because
25 they have sought treatment for a condition that they have
1 identified and that our prospective ratings show exist.
2 So, where is that cut-point along this continuum
3 between what is just a phenomenon that is somewhat bothersome
4 but not necessarily worthy of treatment and at what point is
5 it severe enough that a woman wants treatment, needs treatment,
6 and is willing to put up with the side effects that go with
7 most treatment?
8 I think that there's always this issue. It comes
9 up with everything else. How much pain do you have to have
10 before you take an analgesic? How bad does your headache have
11 to be? How bad does your cold, sore throat, or flu have to
12 be before you decide I got to do something about this?
13 So, I do think of PMS as ranging from something
14 that nobody would seek treatment for or think is worthy of
15 much attention, all the way up into PMDD.
16 DR. TAMMINGA: And that cut-point in your opinion
17 does not occur at the point of diagnosis of PMDD?
18 DR. ENDICOTT: PMDD is a stringent diagnosis.
19 There are certainly women who want to do something about their
20 PMS that is just below the threshold of PMDD. But the idea
21 that every woman might want to pop a pill or something -- even
22 women with PMDD, a lot of them say, well, isn't there anything
23 else I can do, lifestyle, vitamins, diet, exercise? Are there
24 other things I can do so that maybe I won't need to take
25 medication? So, for any woman there is always this issue of
1 the balance between whether or not they want to take a
2 medication or not and how severe it is, how much pain it's
3 causing in their lives, and how much impairment it's causing.
4 So, we do see women who don't quite meet our
5 criteria for the studies, but certainly are severe enough to
6 warrant treatment.
7 DR. TAMMINGA: Thanks.
8 Other comments from the committee or its advisors?
9 Dr. Fyer.
10 DR. FYER: Following up on something Dr. Endicott
11 said, years ago when Dr. Endicott started her studies, I
12 remember having a conversation with her about starting the
13 prospective follow-up. The thing that really struck me was
14 the number of people who self-reported for PMS who did not
15 turn out to have menstrual related symptoms.
16 I guess in terms of Dr. Laughren's question about
17 drug companies using PMS as an indication, I think that it
18 would be particularly important to make sure people get the
19 right kind of care, that the prospective assessment be built
20 into anything anybody does, and also the issue of, if people
21 don't have that particular menstrual related thing, to begin
22 to characterize the other things people have and what other
23 kinds of treatment may be useful for them, and sort of see
24 it as an opportunity to identify other treatment responsive
25 syndromes that are troublesome to women.
1 DR. PARRY: In that line, I think there was a study
2 done a while ago, the De Jong study at the NIH, that looked
3 at women who came in complaining of premenstrual symptoms.
4 They looked at the group in whom the diagnosis was confirmed
5 versus those in whom the diagnosis wasn't confirmed by these
6 daily prospective ratings. In the group that was not
7 confirmed, a majority of those, like 80 percent, had other
8 psychiatric diagnoses, the most frequent of which was major
9 depressive disorder. It was, I think, maybe less stigmatizing
10 to think that they had premenstrual symptoms when they really
11 had major depressive disorder.
12 DR. TAMMINGA: Additional comments, thoughts?
13 Some of the committee hasn't spoken up.
14 DR. THYS-JACOBS: I just want to mention that most
15 of the clinical trials on PMS have criteria of 2 months of
16 prospective documentation of symptoms, and their criteria
17 varies from trial to trial. Some people have used a 30 percent,
18 some people have used a 50 percent. The majority of the studies
19 have ruled out depressive disorders. So, the trials are there,
20 and I think PMS, as it stands, in research is a distinct entity.
22 There are, indeed, a number of women who come in
23 and say that they have symptoms, and when you prospectively
24 document their symptoms, it's not what it turns out to actually
25 be. But when you document for two menstrual cycles and you
1 show that there's a luteal phase increase in their symptoms,
2 that's what they have.
3 DR. TAMMINGA: Dr. Cook?
4 DR. COOK: I was just going to say I can't speak
5 because I only have one X chromosome.
7 DR. COOK: No.
8 But I did want to raise the issue, however, that
9 although people are saying women would know what PMS is, most
10 of the objections that have not been represented, other than
11 to describe them as political, particularly have to do with
12 PMS and are raised by women, not men.
13 So, I think the discussion has covered that there
14 would be some risk in developing an indication for PMS, but
15 if someone wants to go for the broader indication, I think
16 the FDA should seek some input from those who had enough impact
17 to keep what to me sounds like a very clear disorder, PMDD,
18 out of the body of DSM-IV. I think that those are people that
19 should be heard from. I don't necessarily agree with them.
20 I just would suggest that that be included.
21 DR. PARRY: One risk factor I just thought of.
22 For example, given the cyclicity of the symptoms, one very
23 important differential diagnosis is women with rapid cycling
24 mood disorders, and giving something like fluoxetine or other
25 antidepressants can potentially exacerbate the rapid cycling.
1 I think that would be one very strong indication. And 90
2 percent of patients with rapid cycling mood disorders are
3 women, and it would be one of the very adverse consequences
4 of not having a careful diagnosis.
5 DR. TAMMINGA: Dr. Hamer.
6 DR. HAMER: I'd also like then in my role as an
7 advocate to do the same thing I did last time, which is strongly
8 urge the FDA to, as it usually does, be very careful in writing
9 the labeling for exactly that reason. As this will probably
10 largely be prescribed by primary care physicians,
11 gynecologists, they really do need to be sure to eliminate
12 rapid cycling disorders so that they don't inadvertently give
13 an SSRI to them.
14 DR. TAMMINGA: Dr. Winokur.
15 DR. WINOKUR: I'll jump in and make a comment.
16 That was the direction of some of my questions earlier. I
17 think I want to also distinguish just missing women who may
18 have rapid cycling from the need to really focus down on the
19 potential effects of the SSRIs or other antidepressants for
20 the treatment of PMDD or other menstrual cycle related entities
21 because we are talking about a cyclic mood disorder. And now
22 we're talking about potentially applying these drugs that we
23 know in other populations have a potential to induce mania
24 or hypomania, as is the case for other symptoms or problems
25 that have been talked about, such as sexual dysfunction. We
1 know that sometimes symptoms have to be explicitly looked for
2 rather than just kind of observed as adverse events. So, I
3 think that future research studies might build in specifically
4 assessment of hypomanic type symptoms.
5 DR. TAMMINGA: Additional comments about PMDD as
6 a diagnosis or about PMS as a disorder?
7 (No response.)
8 DR. TAMMINGA: If not, I think we should move on
9 and talk about the appropriateness of the mood scales, of the
10 VAS Mood-3 and the VAS Mood-4, in the data that we heard
11 presented today by Lilly.
12 Dr. Parry, you already mentioned something about
13 this, that the visual analog scale was the single most
14 appropriate scale. Maybe you could launch a discussion of
16 DR. PARRY: I was just referring first, in
17 general, that the visual analog scale has demonstrated
18 reliability and validity not just in terms of premenstrual
19 dysphoric disorder but in depression and other disorders.
20 I think that the thing that you want to check is
21 that these women are turning in their scales every week so
22 that they're not retrospectively filled out.
23 I think in evaluating studies, it's important to
24 see sometimes -- some of the earlier studies done just might
25 have only a scale of 1 to 3, no symptoms, some symptoms, severe
1 symptoms. I think the advantage of at least like 100
2 millimeter line visual analog scale is it gives the subject
3 a whole range of symptoms and allows for individual variability
4 in completing these scales.
5 I think that the ones that were used in these
6 studies were very adequate. There was one case where they
7 linked I think happiness or mood and energy which may or may
8 not be associated. But for the most part, they tapped into
9 the main symptoms of depression, anxiety, rapid mood shifts,
10 and irritability. Thanks to much of the work of -- Jean
11 Endicott has developed the other scales -- it's quite adequate.
12 They have been used extensively, and I think it's important
13 to have both a clinician monitored -- as well as a subjective
14 and an objective assessment built into the scales.
15 But I'd again like to point out compared to other
16 psychiatric disorders, the daily ratings, at least for 2 months
17 to get into the study, is more rigorous than we have for most
18 other disorders. So, overall it doesn't mean there's not room
19 for improvement.
20 DR. TAMMINGA: Dr. Thys?
21 DR. THYS-JACOBS: I would definitely agree. The
22 VAS scale has been used extensively for not only PMS but for
23 PMDD, and the fact that it was Mood-3 versus Mood-4 I don't
24 think is really of any major consequence.
25 DR. TAMMINGA: What about using a subscale of the
1 VAS rather than the total VAS score? We heard from Dr. Steiner
2 this morning about the rationale for doing that.
3 DR. THYS-JACOBS: I don't think it makes that much
4 of a difference because those symptoms that were chosen really
5 are very much reflective of the total PMDD scale.
6 DR. PARRY: Yes, I do think they're much more like
7 same amoeba, different pseudopods. So, they do correlate but
8 I think it is of interest to go back, at least from a research
9 point of view, and look at the subtypes of symptoms. Now,
10 that's an area for development which specific subtypes of
11 symptoms to do an item analysis to respond to specific
12 interventions, physical versus emotional symptoms, for
14 DR. TAMMINGA: Dr. Laughren?
15 DR. LAUGHREN: Can I try and clarify again the
16 specific question that I had? I thought I heard Dr. Steiner
17 say earlier, expressed perhaps some surprise that fluoxetine
18 had as robust effects as it had on physical symptoms, as well
19 as, of course, the affective symptoms. I'm wondering if in
20 retrospect -- and again, this has to do with our advising other
21 companies in their development programs -- if you would have
22 focused on that affective subset or if you would have focused
23 on the total scale as your primary outcome in that trial.
24 Again, the question is for future companies, should we advise
25 them -- given the findings that we're seeing here, should the
1 primary outcome be the total scale or should it still be the
2 affective subscale?
3 DR. TAMMINGA: We could ask Dr. Steiner to give
4 us his opinion on that.
5 DR. STEINER: Each line on a visual analog scale
6 is its own scale. You can look at it that way. There are
7 studies were you use one line of a visual analog scale and
8 that is your rating scale. The fact that we picked seven lends
9 itself to the fact that we were able to analyze it in total,
10 which we did, which was statistically significant, lends itself
11 to take the three major emotional behavioral symptoms that
12 we wanted to analyze, and then separately the physical
14 If you look at what's available in the literature
15 for other SSRIs, it looks as if all of them work for both the
16 emotional and the physical symptoms. So, if I were to advise
17 to you guys, I would say take a combined visual analog scale.
18 DR. TAMMINGA: Dr. Katz.
19 DR. KATZ: Yes. We'd be very interested in the
20 committee's view of this whole question of the effect on the
21 physical symptoms. Certainly it wasn't expected, and I
22 suppose one possibility is that it's entirely an artifact,
23 if you will, of the primary affective effect of the drug.
24 I gather there were no objective, quote/unquote, measures of
25 bloating or weight gain or this sort of thing that were
1 examined. I gather that's true. Maybe that's not true. If
2 there are objective --
3 DR. JUDGE: We have information on weight.
4 DR. KATZ: Okay, well, that might be useful to
5 look at because this will have impact if the drug were approved
6 or might have an impact on the label and what this should be
7 indicated for, just the affective symptoms of PMDD or the entire
9 DR. TAMMINGA: Because presumably there are some
10 people who have just physical symptoms of PMDD --
11 DR. KATZ: Right.
12 DR. TAMMINGA: -- while emotional symptoms --
13 DR. KATZ: Well, not of PMDD I gather by the
14 diagnostic criteria, but of PMS.
15 In fact, it will be interesting, of course, to
16 look and see if there were any work done on the effects of
17 the drug in women who just had physical symptoms.
18 So, the whole question of the effect of the drug
19 on the physical symptoms I think is very important for us to
20 hear what the committee has to say.
21 DR. PARRY: I would just like to point out there
22 was a study. I believe it was in the archives. Now, this
23 is not PMDD but depressive symptoms in the general population.
24 When you looked at somatic symptoms of depression, a good
25 portion of those were reported by women. So, now, it may not
1 be the case in this disorder certainly, but some of the somatics
2 being a depressive equivalent I think is a distinct
4 DR. TAMMINGA: Dr. Temple.
5 DR. TEMPLE: But my understanding is -- someone
6 tell me if this is wrong -- that you can see cyclical weight
7 changes and actual edema, at least in a fraction of the
8 population. So, those presumably wouldn't be the result of
9 your improved mood if you could show a difference in those
11 DR. PARRY: I think there is documented evidence
12 of fluid and electrolyte changes with the menstrual cycle.
13 I guess I was being more broad about symptoms other than just
14 weight gain and fluid retention.
15 DR. TEMPLE: It would be extremely interesting
16 to see if a drug that's known primarily at least as an SSRI
17 had an effect on those things which certainly would have been
18 a surprise to us.
19 DR. TAMMINGA: Perhaps, Dr. Judge, you could give
20 us whatever data you have on the objective measures of physical
22 DR. JUDGE: I'll take your points about weight
23 with respect to two discrete questions. Firstly, overall
24 weight changes in the groups per se. And the information here
25 is -- this is a summary of changes in weight from two of the
1 studies. We were able to glean information on weight changes
2 from baseline to end point. What we see is for the placebo
3 group, overall there's a mean change of a gain of .2 kilograms.
4 For the fluoxetine patients, overall there was a mean loss
5 of 0.4 kilograms.
6 So, individually within the groups, if you looked
7 at the next slide please, in terms of patient numbers who had
8 potentially clinically significant changes in weight -- and
9 that is, patients who had neither a 7 percent increase or
10 decrease in weight -- we see overall for placebo very few
11 patients exhibited any such parameters, and for fluoxetine
12 a few patients again, but more than placebo, did exhibit either
13 a weight gain or a weight loss. But overall the percentages
14 were very low, 7 and 9.5 percent, respectively. Those
15 differences were statistically different from placebo. So,
16 that's the summary of changes in weight with respect to
18 Now, overall in terms of your question with --
19 DR. TEMPLE: What you're really interested in is
20 the difference between the follicular and luteal phase.
21 DR. JUDGE: Yes. That's what I was going to say
22 in terms of the discrete question with respect to the --
23 DR. TEMPLE: We already know fluoxetine can affect
24 maybe weight.
25 DR. JUDGE: Yes. I was going to say with respect
1 to your other question, especially for those patients who had
2 bloating as a reported symptom, we don't have that information.
3 DR. TAMMINGA: Dr. Altemus?
4 DR. ALTEMUS: I'd just add that at first I was
5 really surprised by that information that bloating went down,
6 that those symptoms improved. But when you think about it,
7 we don't really know what's mechanistically wrong, like what's
8 going wrong in these women, but they have normal shifts in
9 hormones but they have a very exaggerated response mentally
10 to those shifts in hormones. Certainly there are mild changes
11 in appetite and metabolism and fluid electrolytes with the
12 cycle. So, when you think about it that way, it's not that
13 surprising if they overreact to the hormone changes in terms
14 of mood, that they'd overreact in terms of appetite or
15 metabolism too.
16 DR. KATZ: The question is whether or not the drug
17 is having an effect on those independent of its effect on --
18 you may not be able to tease this out, but since they are
19 subjectively reported, if a person is feeling better from an
20 affective point of view, there might be less attention paid
21 to these other physical symptoms or they might feel as if
22 they're lessened when in fact objective measures may show that
23 they're not lessened. That's the question.
24 DR. ALTEMUS: Yes. We were talking about this
25 at lunch. I'm not even aware of a study in normal menstrual
1 cycle that shows weight changes. I'm not sure that's ever
2 been documented. It's a really common complaint, but I don't
3 think it has been shown.
4 DR. TAMMINGA: Dr. Endicott.
5 DR. ENDICOTT: There are two issues here.
6 In terms of weight gain, a couple of studies that
7 have tried to actually document the subjective feeling of
8 weight gain have generally not done so. What they have found
9 is a redistribution of water, not an actual increase, the idea
10 that the bloating is not necessarily from water retention,
11 but from a redistribution of water so that the subjective weight
12 gain may well be response to that.
13 Now, on the other issue about whether if you
14 improve the dysphoric mood changes you automatically get a
15 reflected improvement in the physical symptoms, at least in
16 one of our published studies of a drug, unnamed, we got very
17 good changes in irritability, depression, and anxiety with
18 no changes in the physical symptoms. So, it is possible to
19 demonstrate changes in the dysphoric mood symptoms without
20 changes in the physical symptoms. In fact, the women often
21 said, well, I still have my physical symptoms, but that's no
22 big deal.
23 DR. TAMMINGA: Dr. Steiner.
24 DR. STEINER: I think that bloating and the sense
25 of bloating is more subjective than the breast tenderness.
1 My surprise was that the breast tenderness -- and maybe it
2 should not be a surprise -- is a pain and you're giving an
3 SSRI. The bloatedness is also something like Jean said. When
4 we tried to measure whether there was a change in weight, there
5 isn't, although there is this complaint in the sense of
6 bloatedness. But the breast tenderness is really an
7 eye-opener. These women say I don't have that pain anymore,
8 and I think that that is a drug effect.
9 DR. PARRY: SSRIs and other antidepressants do
10 increase the pain threshold.
11 DR. TAMMINGA: We've been kind of discussing
12 around the issue of whether the indication should be for the
13 affective symptoms of PMDD or for the whole syndrome, however
14 one might define that, of PMDD.
15 Dr. Fyer.
16 DR. FYER: Well, this is not at all my area of
17 expertise, but it sounds to me, from what people have been
18 saying, is that we really don't know about the actual effect
19 on the physical symptoms, and that it would be premature to
20 put in labeling something we don't know. Rather, it seems
21 to me that it would be very nice if the sponsor would undertake
22 the responsibility of doing some well designed studies to
23 address that question. It would be, I think, a help to
25 DR. TAMMINGA: Other comments? Dr. Temple.
1 DR. TEMPLE: What should they do?
2 DR. FYER: What should --
3 DR. TEMPLE: Well, they had physical symptom
4 scores of the usual kind. Do you have a thought about how
5 they could tease that out further?
6 DR. FYER: Well, like I said, I do anxiety
7 disorders, so I'm not a measurement expert in this area.
8 But I think Dr. Katz' comment was well taken.
9 We don't know to what effect what we're seeing is the result
10 of mood changes, and it seems from what Dr. Endicott just said
11 that, in fact, there's not all that much known about what
12 actually physiologically goes on. I think it would be very
13 helpful to everybody to do studies in which they actually looked
14 at weight changes, luteal versus follicular phase of the cycle,
15 during the drug treatment, as well as before, and to try to
16 characterize that.
17 DR. TEMPLE: I suspect that's true. I guess my
18 nervousness about it is that we don't usually expect that
19 someone will solve a dilemma that's existed for 20 years before
20 they can get a practical claim that they improve it.
21 DR. FYER: Well, no. Wait a second.
22 DR. TEMPLE: So, one has to balance what's
24 DR. TAMMINGA: Dr. Cook.
25 DR. COOK: My concern is that it's new for them
1 to really raise this. They had almost wanted to take it out
2 in the design of these studies. So, even if they don't need
3 to do more, they certainly need to test the hypothesis a priori.
4 DR. TAMMINGA: Dr. Fyer.
5 DR. FYER: I'm a little surprised to hear Dr.
6 Temple take that position because I think when we talk about
7 the real world, where drugs are prescribed, I think somebody
8 raised the point before that much of the prescription for this
9 indication is going to be in non-psychiatric settings by GP's
10 where there is a predominance of managed care and limited time.
11 I really think you want to be careful about somebody saying
12 they have breast tenderness premenstrually and then being put
13 on long-term fluoxetine. Before we get into that kind of
14 situation, I think some experimental data would be useful.
15 DR. TEMPLE: That seems to me an important but
16 different question. What level of symptomatology should make
17 you want to go on an antidepressant for many years is a very
18 important question.
19 But, for example, the things that occur to me is
20 they could find a population that doesn't have any obvious
21 psychiatric component and see if they improve.
22 DR. FYER: Yes. That would be an excellent --
23 DR. TEMPLE: So, that's one thing.
24 You could also look within the trial to see whether
25 people who have better response on mood also have a better
1 response -- this is after the fact and it wouldn't be as
2 convincing, but that's another thing they might do. They could
3 look to see if there's a correlation between improvement on
4 one and improvement on the other. That wouldn't be definitive,
6 I'm just trying to think of what they could do.
7 But I would say nothing in this database would suggest a claim
8 for treating those conditions alone in the absence of the
9 dysphoric disorder. So, I don't think anybody has been
10 thinking about that.
11 DR. FYER: But I think the other issue is to what
12 extent the labeling reflects the idea that a whole syndrome
13 -- where there's the implication that people who don't have
14 predominant dysphoric symptoms will respond to this drug.
15 I think that's where the sort of gray area issue for labeling
16 comes in.
17 DR. TAMMINGA: Of course, that wasn't tested here
18 at all. The only thing that was tested here was whether --
19 it only looked at the population of people with dysphoric
20 disorder, and then we're kind of wondering about the status
21 of the physical symptoms, and that's really less clear. But
22 for sure, we're not looking at any data of people who had no
23 dysphoric disorder but PMS and had only physical symptoms.
24 Would Lilly like to suggest what studies they're
25 doing? There has been some question in the committee about
1 what studies Lilly might be currently doing that would address
2 these kinds of questions.
3 DR. JUDGE: I just want to make clear that the
4 application is for fluoxetine in PMDD, premenstrual dysphoric
5 disorder. The question was raised as to what other studies
6 Lilly are doing with respect to PMDD and specifically
7 intermittent dosing. These are currently the two
8 multi-center, multi-national studies that are ongoing, taking
9 place in Europe and in the United States.
10 Firstly, a randomized, parallel,
11 placebo-controlled study comparing two doses of fluoxetine
12 intermittently and that is defined as 14 days prior to the
13 onset of menses in approximately 250 randomized patients, and
14 this will involve many, many more patients screened, but
15 randomized will be approximately 250 patients.
16 Secondly, there is another study ongoing,
17 randomized, parallel, placebo-controlled study comparing
18 infrequent monthly dosing of fluoxetine with an alternative
19 formulation in approximately the same number of patients.
20 These are also again in PMDD patients, not
21 specifically with one or two physical symptoms or PMS,
22 specifically PMDD.
23 DR. TAMMINGA: Dr. Judge, in either of these
24 studies, do you have objective measures of physical symptoms?
25 DR. JUDGE: Again, as we considered with a group
1 of leading researchers in PMDD/PMS, it was felt that the scales
2 currently used in the trials, namely the visual analog scales,
3 namely PMTS, and perhaps one or two other scales, were indeed
4 good scales in order to measure the appropriate outcomes with
5 respect to physical symptoms and mood symptoms.
6 As you saw from the studies presented this morning,
7 the secondary objectives of measuring physical symptoms was,
8 indeed, stated as a secondary objective measure. Indeed,
9 consistently the effect was found for physical symptoms, and
10 we hope to find consistent effects in the ongoing studies as
12 DR. TAMMINGA: Dr. Fyer.
13 DR. FYER: It's really interesting that Dr.
14 Endicott just got up and said that when the objective studies
15 have been done about weight gain, what they found is
16 redistribution. Yet, you're doing these new studies, and the
17 design of the study is not really addressing what's apparently
18 known in the field. It might be nice to take these things
19 into account.
20 DR. JUDGE: The studies that I've talked about
21 are Lilly's commitment to further the work in PMDD with respect
22 to intermittent dosing. The question was raised earlier, is
23 it appropriate to pursue intermittent dosing?
24 With respect to weight change, I think that's a
25 very important question. Is weight an appropriate symptom
1 to study? We won't know that from these studies. That's a
2 different question for which we have not put a study in place.
3 Perhaps it's more appropriate for that study. I don't know
4 that someone would have a special interest in that and do that.
5 I'm not sure. But the studies ongoing are the ones that I've
6 listed here. I think that's a very important and nice
7 question, but it's not particularly going to be addressed by
8 these questions.
9 DR. FYER: Yes, that's exactly my point, that it
10 isn't going to be addressed. I think it would be nice if it
12 DR. TAMMINGA: Dr. Hamer.
13 DR. HAMER: In the existing 019 study that was
14 Dr. Steiner's study, you saw the subjects once during the luteal
15 phase and once during the follicular phase. Did they get
16 weighed each time?
17 DR. JUDGE: No, they did not get weighed each time.
18 They got weighed infrequently, and definitely at baseline
19 and endpoint, but not during every visit.
20 DR. HAMER: In the new studies, are they being
21 seen both during the follicular and luteal phases, and are
22 they being weighed?
23 DR. JUDGE: Again, there would be infrequent
24 measurements of weight.
25 DR. PARRY: The inherent problem, though, of
1 course, is that other things can affect weight, particularly
2 diet and carbohydrate craving, which you'd have to control,
3 and that's one of your outcome criteria for PMDD symptoms that
4 you're trying to monitor.
5 DR. HAMER: Sure. Just asking about weight is
6 asking about, in some sense, result not mechanism. But it
7 would be nice to design in weight measurements at those places,
8 as well as perhaps some physical measurements, various sorts
9 of circumferences and things like that, so you could look for
10 this redistribution of water.
11 DR. TAMMINGA: Dr. Altemus?
12 DR. ALTEMUS: If women on their rating forms are
13 feeling significantly less bloated with the treatment, I don't
14 think we would deny them the treatment because they don't
15 actually have a physical change in weight. It's an interesting
16 mechanistic question, but it really is not relevant to whether
17 this should be an indication for the drug.
18 DR. ENDICOTT: I'd like to comment. Weight gain
19 is not part of the criteria for PMDD for the very reasons that
20 have been mentioned, that the studies suggest that it is
21 feelings of bloatedness that are part of the criteria but not
22 weight gain, which is one of the reasons that I don't think
23 any of the studies are focusing on weight gain per se because
24 the really careful studies that have been done looking at that
25 issue have found evidence of redistribution, not of actual
2 DR. TAMMINGA: So, the physical symptoms that
3 we're talking about could more precisely be called perception
4 of physical symptoms rather than what we might really document
5 with weight gain.
6 DR. ALTEMUS: Right. Like somatic symptoms in
7 depression. People would complain of more aches and pains
8 that improve.
9 DR. PARRY: You can always look at the item
10 analysis on the Hamilton of either subjective or objective
11 measures of weight change.
12 DR. TAMMINGA: This may be a good time to
13 transition into the discussion of continuous versus
14 intermittent dosage, especially since we've seen that Lilly
15 has two current studies now, one in intermittent and the other
16 one in infrequent dosing. Any comments from the committee
17 or from its advisors on that?
18 DR. PARRY: Well, I think the jury is still out
19 with the luteal phase dosing. There are some preliminary
20 reports. I think to me probably one of the primary
21 determinants may be the half-life of the drug. I think that
22 was mentioned earlier. So, I think to put this forward, it
23 would be best -- I think to put it on luteal phase dosing at
24 this point I think would be premature.
25 DR. TAMMINGA: Dr. Thys?
1 DR. THYS-JACOBS: I agree.
2 DR. TAMMINGA: It would be --
3 DR. THYS-JACOBS: Premature.
4 DR. TAMMINGA: Premature to suggest luteal phase
6 DR. THYS-JACOBS: Right. There's not enough
7 evidence at this point in time to advise luteal phase dosing.
8 DR. ALTEMUS: I think it's also not just the
9 half-life of the drug but how long-lasting the changes are
10 in the brain, that it may take several weeks to reverse once
11 you've been on a drug for 2 weeks. That intermittent dosing
12 might work for regular depression.
13 DR. PARRY: That's why we have patients off
14 fluoxetine for at least a month before they enter studies.
15 DR. TAMMINGA: Dr. Katz?
16 DR. KATZ: I don't think we're asking whether or
17 not we should write labeling that says you should give it just
18 during the luteal phase or however often. The question is
19 given that it is an intermittent condition and that the
20 presumption that you might be able to just treat this with
21 intermittent dosing maybe even a couple of doses -- who knows
22 -- whether it's appropriate to approve it with chronic dosing,
23 in other words, whether this is something that is -- it's a
24 benefit-risk question, whether it's worth having women on this
25 drug chronically for a condition that occurs just
1 intermittently for a relatively brief period of time. That's
2 the question.
3 DR. TAMMINGA: The data that's in front of the
4 committee is only continuous data.
5 DR. KATZ: Right and the question is whether or
6 not you think that is appropriate to approve the drug on the
7 basis of that. Is it worth it? Or should they do more studies
8 that further define? Now, it is invariably true that we almost
9 never know, when we approve a drug, the perfect dosing regimen.
10 We only know what people studied. The question here is,
11 should they study more before we approve it, or can you live
12 with this?
13 DR. TAMMINGA: Dr. Altemus.
14 DR. ALTEMUS: I don't think so just because I don't
15 think we put that burden on other disorders. You could say,
16 well, can you give medication half as often for major
17 depression? I think just because it appears during 2 weeks
18 or a week of the month, I think it's pretty unlikely that --
20 DR. PARRY: Yes, I would kind of echo that, that
21 this is a recurrent, periodic illness and that the medication
22 may work as a mood stabilizer. To recommend anything less
23 than that, we just don't know if that's going to have the same
24 efficacy until that's demonstrated.
25 DR. TEMPLE: That's not the question.
1 DR. COOK: No, but I would add something, that
2 until you have evidence that the intermittent dosing is both
3 efficacious and safe, I don't think one should presume that
4 it's safer to give it in an intermittent manner. These drugs
5 have been tested for safety largely in a chronic manner, and
6 as alluded to, I wouldn't use the word "kindling," but
7 sensitization changes from this sort of dosing pattern wouldn't
8 have the safety behind them.
9 Now, with fluoxetine, of course, you have the
10 problem is anything truly intermittent, but as you're asking
11 a more general question about SSRIs, hitting it just a few
12 times a month may not be the best paradigm for long-term safety.
13 We don't know. It's intuitive that it's worth trying, but
14 the data has to be there.
15 DR. TAMMINGA: Dr. Temple.
16 DR. TEMPLE: No one thinks we could approve an
17 intermittent regimen that hadn't been studied. This goes to
18 the fundamental question of the approvability of a chronic
19 treatment for an intermittent disease where, in this unusual
20 case, you actually know when the disease is going to occur.
21 DR. THYS-JACOBS: This is not a disease. It's
22 a disorder.
23 DR. TEMPLE: Whichever. But you know when what
24 you're treating, however you will care to define it, is going
25 to occur.
1 It really goes to the fundamental question of
2 whether a drug with a half-life of 14 days if the best way
3 to treat this or an appropriate way to treat this. In asking
4 that question, I want to be very clear I'm not offering an
5 opinion. It's just something that ought to be discussed.
6 That's all. I'm not saying that we are horrified by that idea.
7 It might be the best thing in the world for everybody.
8 DR. TAMMINGA: Dr. Winokur?
9 DR. WINOKUR: Well, this is a disorder that we've
10 heard, I think, convincing evidence is associated with a good
11 deal of morbidity and distress, and I think we can evaluate
12 data from studies that have been done to form opinions about
13 efficacy and safety, tolerability, acceptability in this
14 context and then consider alternative approaches to treating
15 when such data come along. But I don't see anything that is
16 a barrier to our rendering an opinion about continuous
17 treatment for a disorder that is discontinuous, but has
18 significant implications. We've heard a lot of testaments
19 to that effect.
20 DR. TAMMINGA: Dr. Geller.
21 DR. GELLER: There was mention in the material
22 we got of a sertraline trial that was discontinuous. Is there
23 more information on that? Did people get withdrawal when it
24 was taken away and stuff?
25 DR. TEMPLE: I don't think we know yet.
1 It's also possible new information will emerge
2 later that will cause people to reconsider what the best
3 approach is.
4 DR. GELLER: I was wondering if specifically there
5 was more information about the sertraline study at this point
6 in time.
7 DR. TEMPLE: I don't think we do.
8 DR. TAMMINGA: We don't have it on our table.
9 DR. PARRY: The analogy that just comes to mind
10 is lithium, recurrent mood disorder, and I think that there's
11 substantial evidence at this point that for a recurrent mood
12 disorder that lithium works better the longer you leave it
13 on them, and there are actually risks of taking them off.
14 I might see that as the closest analogy to address that.
15 DR. TAMMINGA: If you don't mind my interrupting
16 you, I'd like to ask you the question in the case that Dr.
17 Temple proposes that in PMDD this is a predictable cyclic --
18 I mean, mania is not predictably cyclic, but this is clearly
19 predictably cyclic -- would you still hold the same opinion?
20 DR. PARRY: Yes. Well, first, mania may be a
21 rapid mood cycling. In most cases it is, but in not all cases
22 is PMDD predictable. Of course, as soon as you enter them
23 in the study, they don't get their symptoms.
24 DR. ALTEMUS: Also, I think it's premature to
25 think that because it's cyclic you should give the drug when
1 the symptoms appear. To really prove that, I bet you could
2 give the drug during the follicular phase every 2 weeks and
3 they may respond just as well. What I'm saying is it's
4 premature to think that it should be given during the luteal
5 phase just because that's when the symptoms appear.
6 DR. TAMMINGA: Additional comments? Dr. Fyer?
7 DR. FYER: I think that's an excellent point.
8 We don't know the pathophysiology. I think Dr. Parry referred
9 to the possibility that it's acting as a mood stabilizer, and
10 we don't know the important point of entry really is when people
11 are clinically ill. It might be the week before.
12 DR. TAMMINGA: The data we have on the table are
13 continuous data, and we've seen Lilly present that it's, in
14 fact, doing a couple of different intermittent designs. So,
15 we'll have some confidence that additional data will be coming.
16 I guess the question would be whether Lilly wants
17 it to be continuous dosing because it uses more drug. That's
18 not the right way to think about it.
19 DR. KATZ: No. I think Dr. Temple put it well.
20 It's simply a question of is it appropriate to approve a drug
21 for chronic use when the symptoms occur predictably
22 intermittently. That's the question. We're not bringing an
23 opinion to the table. We're simply asking the question is
24 that an appropriate type of treatment for this sort of disorder.
25 DR. PARRY: Well, yes. I think the other
1 consideration is if it's not treated in its initial stages,
2 which may be a follicular phase or early on -- and I think
3 there's data to support this, at least in PMDD -- it can get
4 worse over time, I think the point that Dr. Cook was making,
5 that it can exacerbate symptoms.
6 DR. TEMPLE: Part of the problem is it's hard to
7 know how to even talk about this. With a drug with a half-life
8 of 24 hours, you could ask the question, when do we have to
9 treat, because then you could treat for the week before, the
10 week before and during, you could modify your regimens and
11 actually ask the question. With a drug with an active
12 metabolite whose half-life is 14 days, you can't even ask the
13 question. You're treating chronically whether you like it
14 or not, and the only question is how much to give.
15 But just from my point of view, I thought what
16 Dr. Winokur said was, look, you can look at the benefit, you
17 can look at the consequences, and you can make a judgment about
18 whether the consequence of being on Prozac a lot for a long
19 time, which is obviously something that happens to a lot of
20 people in this country, is worth it in view of the benefit
21 of preventing these symptoms. I think that's what we're trying
22 to --
23 DR. PARRY: The risk of no drug treatment and drug
25 DR. TAMMINGA: Dr. Laughren.
1 DR. LAUGHREN: Let me just give an example. There
2 are drugs for chronic conditions that are given intermittently.
3 Methotrexate is given once a week for treating rheumatoid
4 arthritis. So, there are examples. That's clearly a drug
5 where you would want to limit the overall exposure. So, you
6 would want to find a regimen that involved the least exposure
7 to that drug to get a benefit. So, that's really the question.
8 Is there a more optimal dosing strategy that works that gives
9 a benefit and limits the risk?
10 DR. TAMMINGA: Dr. Fyer.
11 DR. FYER: Yes. I think this is a very
12 interesting question, but I think it comes down to sort of
13 pragmatic issues. I think what you're seeing on the committee
14 is a reflection of that. The sponsor and other people in the
15 literature have demonstrated efficacy using a certain -- or
16 made a strong case for that using a certain approach, and that
17 certainly seems better than leaving people in distress.
18 Anything else is going to be more of a question.
19 I think this almost becomes political in that would
20 I personally think that it would be best that all sponsors
21 be required to invest time and effort to find the optimum
22 approach that has the least risk, et cetera. Yes, I definitely
24 Has that and is that going to happen in the current
25 climate in this country? I doubt it.
1 So, I think if you want to know is it best to do
2 it that way, of course, it's best to do it that way, but what's
3 actually going to happen?
4 DR. LAUGHREN: We simply wanted to raise the
5 question here for two reasons really. Again, this is a
6 condition where the symptoms are cyclical. They're not
7 present continuously. And secondly, there is a suggestion,
8 I don't know how well established, that the possibility of
9 intermittent dosing may be of benefit. That's not established
10 yet. I totally agree with that. So, the question is whether
11 or not there should be encouragement to look at more optimal
12 dosing strategies for this condition.
13 DR. FYER: I personally think there should be
14 encouragement in all kinds of disorders. What you're saying,
15 Dr. Laughren, is that because of the particular pattern of
16 appearance of the symptoms, this particular question gets
17 raised. But in fact, since we don't know pathophysiology of
18 most psychiatric disorders, one could legitimately raise that
19 question about most of them.
20 DR. TAMMINGA: Dr. Laughren, is the data that the
21 sponsor presented about the studies that are ongoing
22 encouragement to the FDA or --
23 DR. LAUGHREN: We've not seen any data from any
24 other studies.
25 DR. ALTEMUS: I think just one final point about
1 that. From what we know about how antidepressants work, it's
2 not an immediate relief of symptoms when you take it. So,
3 I think it's almost counterintuitive that you would expect
4 it to work within those 2 weeks.
5 DR. LAUGHREN: Except that in this condition, it
6 appears to have a fairly rapid response compared to its rate
7 of onset in depression.
8 DR. ALTEMUS: Do we know, though? Is it within
9 2 weeks? It was a month.
10 DR. TEMPLE: It works 3 weeks later.
11 DR. ALTEMUS: But the question is does it work
12 in the first week. Is there any evidence of that?
13 DR. TAMMINGA: Your question is, does fluoxetine
14 work in the first luteal cycle?
15 DR. ALTEMUS: No. It has to start if you start
16 on day 14, would that --
17 DR. TAMMINGA: Oh, we don't have those data in
18 front of us. The only data that was presented to us this
19 morning was when the treatment was started in the follicular
20 stage, yes, there seemed to be a clear response that first
21 cycle. The major response really occurred.
22 DR. ALTEMUS: So, by then they've been on it for
23 3 weeks by the time they get to their symptomatic period.
24 DR. TEMPLE: And that's not from the
1 DR. ALTEMUS: Right.
2 DR. PARRY: On the visual analog scale data, how
3 soon do the ratings start going down?
4 DR. JUDGE: All patients starting dosing on the
5 first day of their cycle, first day of menses, and what we
6 saw for all those studies was that at the first cycle, which
7 is usually roughly around 2 weeks after beginning of dosing,
8 2 to 3 weeks, that they did have a significant effect.
9 Remember, the visits were done from cycle to cycle and the
10 patients were seen follicular, so I can't answer that question
11 whether at the first week, if we looked at the first week,
12 whether there would be significance at that point.
13 DR. PARRY: But the VASs were done daily.
14 DR. JUDGE: That was not analyzed for purposes
15 of today.
16 DR. TAMMINGA: I think we'll just have to wait
17 till your luteal dosing studies are finished and look at those
19 If there aren't any more comments on the
20 intermittency of dosing, I'd like to direct our attention to
21 the use of oral contraceptives along with fluoxetine in this
22 condition. The oral contraceptives were excluded from this
23 study. They're certainly commonly used in the age range of
24 people who would be treated for this disorder.
25 Do oral contraceptives do something promising in
1 this disorder and would we say that fluoxetine is both effective
2 and safe in PMDD people who are using oral contraceptives?
3 DR. PARRY: I think one of the reasons obviously
4 that oral contraceptives are excluded is because oral
5 contraceptives in and of themselves can induce an atypical
6 depression. They also have physical side effects. So, they
7 have to be excluded. I just don't think we have the data to
8 answer whether they can be used concomitantly.
9 If you look at what predicts onset of premenstrual
10 dysphoric disorder, if you look at clinical demographic
11 features, many women have previously been on oral
12 contraceptives and some had dysphoric mood symptoms in
13 relationship to that. Is that data different?
14 DR. TAMMINGA: So, PMDD does occur in women who
15 use oral contraceptives.
16 DR. THYS-JACOBS: Yes, it can definitely occur.
17 In fact, the evidence is either way. Some of the studies
18 show that the OCPs can actually increase affective symptoms
19 and diminish physical symptoms. What the data is on the SSRIs
20 in combination with the OCPs I'm not sure.
21 DR. TAMMINGA: We saw some data this morning, not
22 in PMDD, but in other treated populations, of oral
23 contraceptive use along with fluoxetine, and the safety data
24 looked with and without oral contraceptives looked
1 DR. ALTEMUS: And the response data too.
2 DR. TAMMINGA: Right, response data.
3 Dr. Winokur.
4 DR. WINOKUR: Yes. I would just reiterate I think
5 thought Dr. Judge's review of these issues this morning was
6 very on target. First of all, I think it would have been
7 significantly complicating and confounding to have the
8 presence of OCs in these studies, and then the second issue
9 is potential risks of combining fluoxetine with OC use. I
10 both agree that there's a pretty substantial database outside
11 of PMDD for us to extrapolate to. I can't think offhand of
12 any reason why we'd expect there would be different safety
13 risk issues. I agree that the potential drug interactions,
14 which are nontrivial with some drugs, are not, to my knowledge,
15 a significant concern. I felt that that was quite thoroughly
16 addressed in the presentation.
17 DR. TAMMINGA: Dr. Hamer.
18 DR. HAMER: It seems there are two aspects of these
19 questions. One is to help the FDA write appropriate labeling,
20 if this drug is approved, based on the studies that have been
21 done. There have been no studies done, at least among the
22 three we've been shown, in women on oral contraceptives. You
23 might consider that you should then write labeling saying that
24 this shouldn't be used in women taking oral contraceptives
25 because we have no evidence.
1 On the other hand, virtually every depression
2 trial that I've ever run, or for that matter, almost any other
3 trial, has excluded people who are suicidal, and we give
4 antidepressants to people who are suicidal all the time. Most
5 of the antipsychotic trials have excluded people abusing drugs,
6 and we give antipsychotics to drug abusing psychotics all the
8 DR. TAMMINGA: Bob, we saw data this morning of
9 fluoxetine given to women taking oral contraceptives.
10 DR. HAMER: But not for PMDD.
11 DR. TAMMINGA: Right, we didn't see PMDD plus oral
13 DR. HAMER: Right. But the point I'm trying to
14 make with the discussion of suicidality and depression and
15 so on is that that does not stop the FDA. The FDA does not
16 then write labeling saying we haven't done clinical trials
17 in suicidal people, so don't give the drug to suicidal people.
18 They should sort of use the same kinds of judgments in this
19 context too.
20 The other issue is whether to encourage or
21 discourage, in further clinical trials or in other clinical
22 trials, the use of oral contraception as an exclusion
23 criterion. And I don't have any thoughts on that.
24 DR. TAMMINGA: A more naturalistic design.
25 Do you want to say anything about that, Dr.
2 DR. LAUGHREN: Many parts of the population are
3 left out of typical development programs. Now, hopefully what
4 we see in a development program is that as you move from phase
5 II into phase III, more heterogeneous populations are enrolled
6 and you are including patients who have other disorders or
7 taking other medications and so forth.
8 Really what we want a sense from the committee
9 about on this issue is how important is this exclusion for
10 this drug for this indication for labeling. It's likely if
11 you think it's not a critical issue, that the extent of our
12 modification of labeling -- we might say, in describing the
13 clinical trials, simply that patients taking oral
14 contraceptives were excluded. There wouldn't be any
15 restriction on its use, simply a simple statement that that
16 part of the population was excluded.
17 DR. TAMMINGA: I would feel comfortable with that,
18 especially since we already have data in women with oral
19 contraceptives, even though they don't have PMDD.
20 Any other opinions on this? Dr. Temple.
21 DR. TEMPLE: Just to launch an advertisement, I
22 would say as a general matter, we agree with what was said,
23 that in phase III, as drug development progresses, exclusions
24 should drop off, if possible, to nothing, and you should look
25 at the consequence of the interaction, not avoid it. One of
1 these days I'm sure we're going to write something to that
2 effect, but I think there's a growing appreciation of that.
3 In some areas, like exclusion of the elderly and things like
4 that, there is considerable progress, but we all believe that
5 should be much more general and that, in general, trials should
6 include everybody who might get the drug.
7 DR. TAMMINGA: It's a good point to always make.
8 I think I agree with you that it is becoming more widespread.
9 What I would like us to do is to proceed into a
10 discussion of the specific studies that were presented. We
11 addressed a lot of questions this morning to Dr. Judge and
12 to the rest of the people in the sponsor's group about these
13 studies, the size of the studies, the dropout rate, the single
14 study that had a crossover design, the investigator initiated
15 nature of the studies. I'd just like to open this phase up
16 for some discussion of the committee and hear people's comments
17 on any aspect of this that they think is important.
18 I bet Dr. Hamer could launch this part of the
20 DR. HAMER: Yes, and I'm going to launch it by
21 saying that I'm really bewildered. I know how to behave when
22 I look at a set of clinical trials submitted as part of an
23 NDA whose purpose is registration, that the sponsor designed,
24 managed, supervised with prespecified endpoints negotiated
25 with the FDA. I know how to trust the results of those trials.
1 In a set of trials that was much more loosely,
2 in some sense, organized and evolved, I'm not sure I know how
3 to behave with these results. This is the intersection of
4 science and regulation, and so it's not just a scientific
6 For example, it's unclear to me how many other
7 trials there are out in the literature involving fluoxetine
8 and PMDD. I have actually been looking through the material
9 and if it's in there, which it probably is, I just can't find
10 it. But why these three trials that we're told about?
11 Again, in the usual NDA, there's half a dozen,
12 a dozen, or whatever phase II trials that I don't pay any
13 attention to because they were small and they were dose-ranging
14 and all that sort of stuff. What I pay attention to is the
15 negotiated-out phase III trials. Here I don't know what to
16 pay attention to, and I don't know if there's stuff that I
17 don't know about that I should be paying attention to. So,
18 I'm very bewildered.
19 DR. TAMMINGA: We're asked to pay attention to
20 the three studies that were presented, 1, 2, and 3.
21 DR. HAMER: What we're actually asked is, is there
22 evidence from adequate and well-controlled trials that the
23 drug is safe and effective? I don't know that we're asked
24 simply to attend to these three trials.
25 DR. TAMMINGA: Dr. Katz.
1 DR. KATZ: Well, the point about there being
2 perhaps many other trials that are negative is a good one,
3 and I'm sure the sponsor has been through the literature and
4 we can probably hear about that.
5 The question of relying on trials that were not
6 supervised or conducted by a commercial sponsor, when those
7 are submitted under NDAs, is not an infrequent one. There
8 certainly is plenty of precedent for our approving a drug on
9 the basis of a trial that wasn't conducted or supervised by
10 a commercial sponsor who submits the application. We do
11 require in those case, almost invariably, that we get the
12 complete information, the protocol, and that the study be as
13 well conducted and designed, prospective protocol, as if it
14 were run by the company. So, there's certainly precedent for
15 our relying on such trials if they're well done, if they're
16 appropriately designed, and we have the data.
17 DR. TAMMINGA: Dr. Fyer?
18 DR. FYER: Yes. Could we address the other
19 question? Because I'm just not familiar enough with the
20 literature. Are there in fact other negative trials? And
21 has the FDA conducted a literature search?
22 DR. TAMMINGA: We did actually see a slide this
23 morning from Dr. Judge. Maybe you'd like to catch us up again.
24 DR. JUDGE: These are the other double-blind
25 studies that have been conducted in PMDD for fluoxetine.
1 Besides this, there are also a number of open studies which
2 I won't show here.
3 For all of the other studies not part of the
4 submission, there is no negative study for fluoxetine in PMDD.
5 We've also attempted to try and find unpublished studies,
6 obviously because there tends to be publication bias with
7 respect to negative studies. We've also tried to look for
8 negative studies with respect to fluoxetine in PMDD, and we
9 could not find any as part of our attempts.
10 So, for example, Menkes was a study in New Zealand.
11 Ozeren's study was a study in Turkey, and Wood and Stone were
12 also studies in the United States of America. They all
13 comprised patients with a diagnostic category of DSM-III-R
14 and therefore of DSM-IV, as you heard earlier. The duration
15 of the treatment cycles ranged from 2 to 3 cycles, and all
16 utilized 20 milligrams daily. All were positive with respect
17 to the efficacy of fluoxetine in PMDD.
18 What the open studies showed was further evidence
19 for that for out longest, showing that perhaps fluoxetine is
20 effective much, much longer out, and also the fact that when
21 patients stopped treatment, even after several months, that
22 there can be very quickly a reemergence of symptoms after
23 stopping treatment. That's the summary of the open-label
25 DR. TAMMINGA: Do you want to speak to why you
1 chose the first three as part of your NDA?
2 DR. JUDGE: Yes, indeed. We attempted to find
3 the data for all of the studies, but for these studies here,
4 we were limited in terms of, first of all, access to that data,
5 sometimes lack of cooperation from the site for whatever
6 reason, and also sometimes for missing data, for example,
7 substantial missing data that was available. So, we did
8 attempt to go back to all of this data in order to provide
9 a comprehensive listing. But this is the three that we felt
10 were of adequate quality, controlled, and we had access to
11 that data. We had investigator cooperation and we could show
12 to you.
13 DR. TAMMINGA: Questions, comments for Dr. Judge?
14 Yes, Dr. Temple.
15 DR. TEMPLE: When you say they were all positive,
16 what do you mean, that they were in the right direction or
17 that they attained nominal significance or what?
18 DR. JUDGE: All of these studies, the other ones
19 listed here, for example, the two crossover and parallel
20 design, with respect to the publication, the primary objective
21 listed in that publication, there was significant evidence
22 for fluoxetine with statistical superiority versus placebo.
23 DR. TAMMINGA: One of the things that we're used
24 to seeing from data sets that don't have otherwise large safety
25 databases is large n's. In this particular study, the n's
1 are not as large as what characteristically we're used to
2 seeing. We saw this morning the presentation of effect sizes,
3 and the effect sizes were impressive. But still, the overall
4 number of patients is not great.
5 Dr. Dominguez.
6 DR. DOMINGUEZ: Yes, not only the small n sizes
7 in the other two studies, but the variability of the inclusion
8 and exclusion criteria. For example, the use of a structured
9 clinical interviews in some of the trials versus just the
10 clinical interview to exclude Axis I diagnosis.
11 The failure to obtain a urine drug screen at the
12 beginning of the trial, knowing that benzodiazepines can be
13 helpful for these individuals, and knowing that at least at
14 some time during the trial, a number of patients did report
15 using benzodiazepines at times, but that was just a listed
16 report. One cocaine overdose.
17 Again, not knowing the race and ethnic background
18 of the individuals in the larger trial. Not knowing what
19 previous treatment they had had for PMDD and what had worked
20 and what had not worked.
21 So, it's the variability in inclusion and
22 exclusion criteria which makes this set of studies much harder
23 to interpret.
24 DR. TAMMINGA: To some extent, you're really
25 agreeing with Dr. Hamer, that these are studies that are more
1 investigator initiated than drug company initiated.
2 DR. DOMINGUEZ: And they're valuable, but again
3 that variability one has to always adjust for.
4 DR. TAMMINGA: One of the things that I'm
5 wondering how to measure is the effect sizes that were reported
6 in studies. I wouldn't mind hearing from some of our
7 consultants about that. It would make me think that there
8 is some consistency to drug response despite all the
9 variability that you're bringing up, that there's a consistency
10 and a rather sizeable drug response in order to get effect
11 sizes like that.
12 DR. PARRY: Well, I know when the DSM-IV was
13 putting together the database, we looked at different
14 calculations of effect size, and no matter which way you looked
15 at the data, you pretty much got the same phenomena. So, I
16 think that it has been a pretty robust response no matter which
17 technique has been used.
18 DR. THYS-JACOBS: Effect sizes actually varied
19 from study to study. Most people use a 50 percent response
20 as a real response rate. Some studies in PMDD and PMS actually
21 looked at the visual analog scale scores and looked at the
22 difference. Not all studies actually looked at effect size.
23 DR. TAMMINGA: Any more discussion about the
24 designs of the studies that we had presented? Dr. Temple?
25 DR. TEMPLE: Well, I'm a little curious about some
1 of the conversation. Obviously, I think Dr. Hamer shows great
2 wisdom in thinking that the only really credible trials are
3 the ones we help design.
5 DR. TEMPLE: And it's hard to disagree with that.
7 DR. TEMPLE: At least sometimes we've had the view
8 that studies that use somewhat variable entry criteria and
9 yet still get the same result add to the database, and that
10 replicating the identical finding over and over again is
11 perhaps somewhat less interesting than replicating it in a
12 variety of settings. I guess I think I at least partly believe
13 that, but I'd be interested in the discussion.
14 Of course, in general, the more medicines that
15 people might be taking that you don't know about, the more
16 they interfere with showing anything. So, in a sense, even
17 that lack of knowledge is a sign of robustness, although also
18 simultaneously makes you nervous.
19 But I'd be interested in a little more of that.
20 The fact that they were different environments I wouldn't
21 say discourages me too much.
22 DR. TAMMINGA: Slightly more naturalistic one
23 might think.
24 DR. TEMPLE: Well, and there was a certain trend
25 in that direction.
1 DR. TAMMINGA: Crossover design, Dr. Hamer?
2 DR. HAMER: It's another reason to be skeptical.
3 Crossover designs really are a can of worms because they're
4 complicated by carryover effects, sequence effects, a variety
5 of things like that. In this particular design, the fact that
6 with a drug whose principal active metabolite has a half-life
7 that's probably at least 2 weeks and then to have only one
8 menstrual cycle in the middle as your recovery, when, in fact,
9 at least in depression we know that fluoxetine requires a number
10 of weeks to even start working and furthermore, a number of
11 weeks to wash out, is troublesome.
12 On the other hand, the statisticians and Dr. Judge
13 were absolutely right. In this particular trial, whatever
14 carryover effect there was would have tended to inhibit the
15 ability to show a difference between fluoxetine and placebo
16 rather than exaggerate it, and thus the fact that they did,
17 indeed, show a difference is reassuring. But I still don't
18 like them.
19 DR. TAMMINGA: I would agree really with Dr.
20 Temple about the what I would call more naturalistic design
21 of this group of trials and still seeing a robust drug response
22 is impressive from my point of view.
23 I was troubled, if you will, to use your word,
24 Bob, about the low n. But then I sort of looked back at the
25 data when I was studying it before the meeting. Although the
1 n is 19, there's really 2 or 3 cycles per person in order to
2 add up, if you will. So, if I were doing, for instance, rat
3 studies, we might count that as an n of 28 or something like
4 that. We have multiple repetitions in the same person of this
5 phenomenon we're observing.
6 DR. HAMER: They're not independent of one
8 DR. TAMMINGA: Excuse me?
9 DR. HAMER: They're not independent of one
10 another, the multiple observations from the same rat or the
11 same person. So, they're not each another degree of --
12 DR. TAMMINGA: Oh, for sure. But they're a
13 within-subject replication which is impressive.
14 Dr. Winokur.
15 DR. WINOKUR: I just wanted to ask Dr. Hamer a
16 follow-up question on his comment to help me understand better.
17 I understand the general reasons for caution about
18 interpretation of crossover studies, but it struck me in
19 initially looking at the data that in this case with the unique
20 feature of this disorder with the repetitive pattern with what
21 struck me as being a fairly solid design of starting with
22 placebo and fluoxetine and then going the other direction and
23 the results tending to support an impact of active drug
24 treatment in whichever sequence, with all of that sorting out,
25 that actually seems to be a persuasive argument. So, I'm
1 wondering in this case whether that's a particularly
2 appropriate use of this design, or am I statistically not tuning
3 into something?
4 DR. HAMER: No. In this case, in some sense, the
5 things that could have gone wrong with this crossover design
6 would have tended to obscure the drug-placebo difference.
7 So, the fact that there still was one was nice.
8 But on the other hand, there are plenty of other
9 reasons not to like crossover trials in general and including
10 prospectively before you did this, this one. One is the
11 difficulty of having a long enough washout period in the middle
12 to know that you really are washing out not just the drug but
13 the effect of the drug because, for all we know about things
14 like receptor proliferation and those sorts of things, there
15 may be a whole lot of things that need undoing over a long
16 period of time.
17 The other is that if anything goes wrong in a
18 crossover design, like you have dropouts partially through,
19 so you don't wind up with people doing the entire, full design,
20 you may wind up with only the first period analyzable, in which
21 case you now have simply a parallel group design with a tiny
23 Again, in this case, it's reassuring that they
24 analyzed just the first period data and found a significant
25 difference because, if you want to, you can sort of disregard
1 the entire crossover part of the design, and you still have
2 a supportive study.
3 DR. TAMMINGA: Any other comments or concerns or
4 extended discussions people would like to have about the design
5 of the protocol of the data sets that were presented to us?
6 (No response.)
7 DR. TAMMINGA: Any other issues that people would
8 like to bring up about the data that were presented to us that
9 speak to the question about efficacy or safety?
10 DR. COOK: I have one comment, and that is this
11 disorder would presumably start at menarche and they limited
12 the beginning to age 18. Now, it would be one thing to suggest
13 this will never be given to someone under 18, but Dr. Parry
14 has said that this is a sometimes progressive disorder in which
15 it's not clear why one would withhold treatment until age 18.
16 Obviously, these are challenging risk-benefit issues, but
17 I don't see justification for not having studied adolescents,
18 recognizing that they have reasonable expectation that there
19 will be off-label use in adolescents.
20 DR. PARRY: Well, except that generally physical
21 symptoms predominate during adolescence and you don't see the
22 mood symptoms until -- in most studies, the average age is
23 30s. Between like 30 and 38 is the mean age of symptoms, and
24 they may have been there for 5 years. But you usually don't
25 see severe mood symptoms during adolescence.
1 DR. HAMER: Speaking as the father of three people
2 who used to be adolescents, I would say that it would be pretty
3 hard to tease out mood swings due to premenstrual dysphoric
4 disorder from the normal mood swings that are part of
6 DR. PARRY: That's why you need to 2-month
7 prospective documentation.
8 DR. COOK: I have to object to that one because
9 the same thing was said about major depression in adolescence
10 and even pre-adolescence. So, particularly because you have
11 the timing with the menstrual cycle here, you could make that
13 But I'm impressed by the age of onset except
14 weren't there a lot of 18-year-olds? In other words, it seemed
15 to me there were some young adults being treated, and the
16 18-year-old cutoff was arbitrary.
17 DR. GELLER: I just wanted to say, as the other
18 child psychiatrist here, that we are beginning to tease out
19 mood disorders from normal adolescents. I think this goes
20 back to the comments that Dr. Winokur was making that the
21 labeling here has to really stress differentiating this from
22 rapid cycling mood disorders. I think without separate study
23 of the adolescent population in this regard, that's going to
24 be hard to do because a common presentation of bipolar disorder
25 in this age group is to rapidly cycle. A lot of those people
1 come in clinically and the parents tell you their child has
2 a menstrual disorder. I think this is just an age range that
3 is ripe for study and for separate study.
4 DR. TAMMINGA: So, the committee would certainly
5 like the sponsor to understand that we see studies of this
6 disorder and this drug in adolescents as important.
7 DR. GELLER: Yes.
8 DR. TAMMINGA: Do you want to say that more
9 strongly, Dr. Cook?
10 DR. COOK: I wanted to raise the issue. I hear
11 now age of onset is different. My concern is exactly what
12 Dr. Geller says. In largely primary practice, will this be
13 used off label for this. So, on the one hand, it should be
14 studied. Until it's studied, I think there should be strong
15 cautions about its use off label.
16 DR. TAMMINGA: Dr. Katz.
17 DR. KATZ: Yes. I'll just let you know. We now
18 have regulations that require sponsors to study drugs in a
19 pediatric population, and they can try and make the case that
20 for any of the particular subpopulations, the condition doesn't
21 exist. But to the extent that it does exist, they're required
22 to do it, and they have to make a commitment. If those data
23 don't come in with the specific application, they may have
24 to make a commitment that they will do it and time lines are
25 imposed and all that.
1 DR. TAMMINGA: So, we can be confident that you'll
2 receive those data.
3 DR. KATZ: At some point.
4 DR. TAMMINGA: Yes, Dr. Geller.
5 DR. GELLER: I think this has to be built in in
6 some way either to post-marketing surveillance or people
7 applying for other kinds of funding to look at the off-label
8 use in the younger population because from what we're seeing
9 as child psychiatrists, it's going to be used when other
10 diagnoses are more likely.
11 DR. TAMMINGA: Yes, Dr. Dominguez.
12 DR. DOMINGUEZ: Another area for further
13 investigation that we have heard very little from this
14 application is issues of predictors of response. For example,
15 I think the agency may want to look at certain associated
16 features of PMDD that, although may occur rarely as part of
17 the disorder, may serve as predictors of response, for example,
18 psychotic-like features during the worst time in the luteal
19 phase, suicidality, and concomitant substance abuse.
20 The issue of adding the feeling of being
21 overwhelmed and out of control to some of the rating instruments
22 I think would be important as well because it has been my
23 experience, at least clinically, that you do find a set of
24 patients with this disorder where that is the prominent
25 feature, along with the effect of instability that they present
1 with, that is the most disturbing to them. Yet, I don't think
2 that was probed at all in the data that was presented.
3 DR. TAMMINGA: Dr. Fyer.
4 DR. FYER: I just want to make two sort of comments
5 that concern me, given the small sample size. That has to
6 do with what's going to happen if and when this whole thing
7 takes place. It seems to me there are two issues.
8 One is that the sponsor has gone over this whole
9 issue about safety. I think that generalizing to the Prozac
10 or fluoxetine database, it's sort of clear that we aren't going
11 to have serious, unexpected things given the age and sex
12 distribution of many people on Prozac is the same as for this
13 requested indication.
14 On the other hand, the idea this is a chronic
15 disorder, there's a high relapse rate, people are going to
16 take this for a long period of time, I would really feel a
17 lot more comfortable if there were a serious commitment from
18 the sponsor to look carefully at quality of life issues. For
19 example, in the FDA's review, it was pointed out that a certain
20 subset of women gain weight continuously on this. That can
21 be a serious health issue as well as a quality of life issue.
22 It would be nice if systematic studies addressed exactly how
23 can people deal with this and are there alternative strategies
24 that might help that kind of thing, as well as issues of sexual
1 The other thing is that it would be in the sponsor's
2 power, if the drug were to be granted an indication and
3 marketed, to do post-marketing studies that would actually
4 look at how the drug was being used and to see if there were
5 consequences that might not be in the best interests of women
6 in this country. I would think that if an indication were
7 to be granted, that a request or a requirement from FDA that
8 such work be done, given the widespread nature of syndromes
9 related to this requested indication, would really be very
11 DR. TAMMINGA: Dr. Geller.
12 DR. GELLER: This is another child psychiatrist
13 type comment. Can the labeling include something to the effect
14 that if underlying conditions are found, they should be treated
15 before treatment is initiated just for PMDD?
16 DR. TAMMINGA: This is a question to the people
17 who design the labels.
18 DR. GELLER: Well, it's not looking for a specific
19 answer, but that kind of thinking, that at least it would
20 encourage people to think if they're going to give it off label.
21 Dr. Hamer was just asking me, who's going to give it to the
22 adolescents? Is it going to be the gynecologist or the
23 pediatrician? Pediatricians are now giving lithium. They
24 won't think anything of giving Prozac for a menstrual disorder.
25 Perhaps if there were something in the labeling pointing out
1 that underlying conditions should be treated first, we might
2 head off some of the --
3 DR. TAMMINGA: Well, certainly in the studies that
4 were presented, only people who lacked other conditions
5 actually got into the study. So, you're really suggesting
6 that something be included in labeling that's consistent with
7 the study data that were presented.
8 DR. GELLER: Right.
9 DR. KATZ: Certainly there's precedent for
10 putting statements in labeling about it's off-label use, but
11 usually when there's an affirmative finding that there is a
12 safety problem in whoever those people are or there's evidence
13 that it doesn't work in those people, those are the two
14 circumstances I can think of where we put statements in labeling
15 about off-label use.
16 DR. GELLER: I wasn't suggesting this is off-label
17 use. I just meant for all age groups that there be some
18 statement that emphasizes that we really don't have data on
19 what happens if you treat an underlying condition first. As
20 usual, the sample that was studied was relatively pristine
21 in terms of comorbid disorders that occur commonly when you
22 have PMDD. I'm going back again to Dr. Winokur's emphasis
23 on people who may be having a bipolar depression, getting the
24 drug for that and then having an exacerbation of their bipolar
1 DR. WINOKUR: Actually, since you mentioned that
2 again, let me make sure. That was only part of what I said,
3 not to have the other part left out. I'm also concerned about
4 inadvertently discovering a new phenomenon, which is people
5 who don't have formally bipolar disorder that is just not
6 detected, but have a different form of cyclic mood disorder,
7 namely PMDD, in whom a small subset but not yet detected or
8 appreciated might be stimulated in the hypomanic or manic
9 direction by a drug like fluoxetine. Again, we don't have
10 any data yet to suspect that to be the case, but I think that
11 from other clinical experiences, that's a lesson that we've
12 learned to be very cautious about. I think if it is going
13 to get out there more widely in a broader population, that's
14 one area that we don't really have adequate data to judge in
15 my opinion.
16 DR. TAMMINGA: Dr. Parry.
17 DR. PARRY: There's just one safety issue, and
18 this is not what was brought up this morning. But in reviewing
19 the materials I was sent, I just thought the description of
20 the use in pregnancy could have been more specifically
21 delineated. For example, there's the Pastuszak study and the
22 Chambers study. Though there's no major teratogenic effects,
23 the fact that mothers who are taking fluoxetine, their children
24 are in the intensive care unit and they had more respiratory
25 distress and lower birth weight and that kind of thing, I
1 thought for safety reasons should be given attention.
2 DR. TAMMINGA: Except that PMDD is not a disorder
3 that occurs in pregnancy, so there would be no reason to treat
4 it during pregnancy.
5 DR. PARRY: Yes, but you have to address that in
6 a labeling issue, and if a woman becomes pregnant and she's
7 on it, that's part of the presentation.
8 DR. TAMMINGA: Any more comments of any kind about
9 the issue in front of us? Dr. Katz?
10 DR. KATZ: Yes, I have another question before
11 you vote. It's sort of under the heading of maybe a theoretical
12 labeling question again. You know we're obviously very
13 interested in labeling, and you may just advise us to do the
14 usual good job that we do. And I appreciate that in advance.
16 DR. KATZ: But I'd just be interested in some of
17 your thoughts on the following.
18 Suppose that you have concluded or do conclude
19 that there is evidence of effectiveness. It's possible I
20 suppose to see that effectiveness as just really fluoxetine's
21 antidepressant effect, and that in some sense it could be
22 considered a global antidepressant in that it has been studied
23 up till now in major depressive disorder. Now it's being
24 studied in a considerably different disorder but that has
25 primarily an affective component. And right now the drug is
1 approved as an antidepressant and then the labeling describes
2 in whom it has been studied, major depressive.
3 One theoretical option, I suppose, for labeling
4 would be to leave the indication as an antidepressant and then
5 list after what's currently listed, which is it's been studied
6 in major depressive disorder, and now say, well, it's also
7 been studied in another sort of depression-like syndrome, as
8 opposed to giving it its own PMDD claim. I just wonder what
9 people think about that.
10 DR. PARRY: Well, I think it's important to
11 recognize that premenstrual dysphoric disorder is categorized
12 as a major depressive disorder, N.O.S.
13 DR. KATZ: Well, that's sort of what I'm asking.
14 In other words, again if you find it's safe and effective,
15 we could choose to label this as a specific treatment for PMDD
16 in addition to its current indication, or we could subsume
17 under its currently existing indication. That's the question
18 I'm asking. I'm interested in your views on that.
19 DR. TAMMINGA: Dr. Geller.
20 DR. GELLER: Just educate us. What would be the
21 down side of doing that?
22 DR. KATZ: Well, it would imply a couple of things.
23 Number one, that PMDD is a type of depression, a type of
24 depressive disorder, and it would imply that the drug is sort
25 of a global antidepressant and it works in any setting in which
1 a patient happens to be depressed, whether it's major
2 depressive, whether it's a cyclical entity, perhaps even others
3 which haven't been studied.
4 So, I don't know that there's a down side or an
5 up side. We are going to have to deal with this question,
6 and I'm just interested to know what people think.
7 DR. PARRY: I can just see, women who have
8 premenstrual dysphoric disorder often don't like to
9 acknowledge that this is a major depression. I'm just trying
10 to think ahead about the potential consequences.
11 On the other hand, if you just propose it
12 specifically for premenstrual dysphoric disorder, I guess my
13 concern would be that any woman with any kind of minor, cyclic
14 physical symptom that may or may not be related to the menstrual
15 cycle might see this as a panacea.
16 DR. TAMMINGA: Dr. Laughren.
17 DR. LAUGHREN: Except that in writing labeling,
18 we would rely very heavily on the diagnostic criteria for PMDD
19 to describe the type of patient who would be a candidate,
20 including functional impairment as part of that. So, we would
21 work very hard to avoid that possibility.
22 DR. PARRY: Yes. I'm just thinking of the
23 potential abuses that may occur irrespective of that.
24 DR. TAMMINGA: Dr. Cook?
25 DR. COOK: Yes. I would be much more comfortable
1 with a PMDD labeling. I think that's what we were here to
2 discuss, and I think that actually there's probably confusion
3 out there in terms of PMDD being just depression. I thought
4 there was data presented -- and certainly the epidemiology
5 that isn't all that was presented today -- to suggest that
6 this is a distinct disorder. This would not be like saying
7 it works for melancholic depression as well as other
8 depression. So, I think this would be reasonable with the
9 caveats that this is not PMS treatment, but PMDD, to have it
11 DR. TAMMINGA: Dr. Winokur.
12 DR. WINOKUR: I think we've heard a lot of comments
13 from the experts and from the general committee that recognize
14 PMDD to be a discrete and recognizable, diagnosable entity.
15 I think increasingly in our field and in communication to
16 more family practitioner types, we've been trying to emphasize
17 precision or care in diagnosis prior to treatment. So, I think
18 that going from data and having discrete, delineated syndromes
19 as a guide to treatment rather than encouraging the older
20 pattern of very broad spectrum triggers for treatment is really
21 something we're trying to encourage. I would much more see
22 the specific indication as being in that dimension.
23 DR. TAMMINGA: Dr. Fyer.
24 DR. FYER: I agree with what has been said about
25 the diagnosis.
1 I think that there's an additional educational
2 issue which I think is positive, but I am concerned about what
3 Dr. Parry raised. I have a question for the FDA people, and
4 that is, to what extent is it within your power, in addition
5 to just labeling about PMDD, to actually structure interactions
6 between pharmaceutical representatives, et cetera so that
7 there is real education about this sort of limitation, as
8 opposed to something that's on a package label someplace that
9 a lot of people don't read?
10 DR. LAUGHREN: The promotion has to be very
11 closely linked to what's in the label. So, to a great extent,
12 that does control the level of promotion that can go on outside
13 of labeling.
14 DR. FYER: Somehow, though, we all know that in
15 the long run there seems to be an enormous amount of what's
16 called off-label usage, and maybe again something that could
17 be considered is some post-marketing survey aspects to see
18 because I think there is a widespread potential here for
19 something that I don't think anybody is particularly interested
20 -- at least people in medicine probably not interested in having
21 unnecessary expense and side effects and maybe missing things
22 that would be treatable by other perhaps psychotherapeutic
24 DR. LAUGHREN: Again, I share that concern. As
25 I said earlier, we'll go to great lengths to try and define
1 the population that we think are candidates for this treatment.
2 Beyond that, it's hard to know what FDA can do. We don't,
3 of course, regulate the practice of medicine, so we can't
4 control off-label use in that sense. But we will try and write
5 labeling that directs clinicians to what we think is the target
7 DR. FYER: You could possibly also ask the
8 sponsors to participate in some post-marketing assessment of
9 exactly what's going on.
10 DR. KATZ: I don't know if we could, and even if
11 we could -- let's assume we learned that there was a lot of
12 off-label use going on. It's hard to know what we would be
13 able to do about it.
14 DR. FYER: Publish it so that people are aware
15 that that's going on --
16 DR. KATZ: Well, a lot of off-label use is
17 published. People think it's a good idea.
18 DR. FYER: Some of it is.
19 DR. KATZ: Well, right, right.
20 DR. FYER: Yes, and some of it isn't.
21 DR. KATZ: We could even in labeling put, just
22 as another thing that's possible to do, a patient package insert
23 in there which also tells the patient that this is for PMDD.
24 It's not for mild symptoms of PMS. Now, whether or not that's
25 actually going to affect people's behavior is another patient,
1 but you can attempt to inform the patient as well as the
3 DR. FYER: I mean, there is a trend, especially
4 with the Internet, for increasing amounts of consumer sort
5 of self-awareness and stuff. So, something like that might
6 in fact be useful.
7 DR. TAMMINGA: Dr. Temple, did you want to log
9 DR. TEMPLE: Well, do I recall correctly that this
10 will have special packaging, a different name, et cetera?
11 That really does open the possibility of a patient
12 insert that is targeted to the population. It reminds them
13 that they should think about whether they want to be on a chronic
14 drug and that their symptoms are severe enough to warrant it.
15 That very situation is one of the conditions in which we
16 believe patient labeling is important, where there's a decision
17 that the patient really needs to participate in actively.
18 In this case, you don't have to label all of the uses of Prozac.
19 You can just label this one perhaps.
20 So, we'll think about that. Special packaging
21 and labeling is a point of some controversy internally I should
22 tell you.
23 DR. TAMMINGA: Unless there's further discussion,
24 I would suggest we move ahead to a vote. Although we've
25 certainly appreciated all the comments of the consultants,
1 the consultants won't vote on the final efficacy and safety
3 So, the first question that the committee will
4 want to vote on is, has the sponsor provided evidence from
5 more than one adequate and well-controlled clinical
6 investigation that supports the conclusion that fluoxetine
7 is effective for the treatment of premenstrual dysphoric
9 I think we just ought to go around the room. Maybe
10 we'll start with you, Dr. Dominguez.
11 DR. DOMINGUEZ: I believe that there are some
12 limits to the generalizability of the data that was presented
13 with respect to race, with respect to ethnic groups.
14 Essentially the patient samples that we have been presented
15 today are in non-minority whites, and it appears from the sample
16 that clearly the drug is effective for the indication that
17 is proposed. I think that this is a clearly distinct disorder.
18 So, it should be labeled as such for this disorder.
19 DR. TAMMINGA: Dr. Hamer.
20 DR. HAMER: As uncomfortable as I am with the set
21 of studies that wasn't generated in the usual way, yes.
22 DR. TAMMINGA: Dr. Geller.
23 DR. GELLER: Yes, with the proviso that the FDA
24 do its usual outstanding job of writing the labels to take
25 into account the discussion.
1 DR. TAMMINGA: Dr. Cook.
2 DR. COOK: Yes.
3 DR. TAMMINGA: Dr. Winokur.
4 DR. WINOKUR: I vote yes.
5 If I can editorialize for a minute, I'm incredibly
6 distressed about the circumstances of the second positive study
7 that we had to consider with the investigator unilaterally
8 interrupting the study. I realize that that was not at all
9 Lilly's doing. I think the data that we are left to consider
10 are overall convincing enough. But I think it's an extremely
11 dismaying circumstance, and I think it's the kind of thing
12 that really can interrupt the progress of the kind of science
13 that we need to make these kinds of decisions.
14 DR. TAMMINGA: Dr. Fyer?
15 DR. FYER: I would vote yes with two provisos.
16 One, Dr. Geller's and that be taken seriously.
17 I'd like to say that I also found this circumstance
18 distressing and I found the sponsor's presentation of the data
19 from that trial distressing. I would hope that in the future
20 such things will be dealt with in a much more straightforward
21 fashion in sort of due respect to the members of the committee
22 and the public.
23 DR. TAMMINGA: And I vote yes as well with many
24 of the same caveats as people have talked about before, but
25 having some interest in the more naturalistic kind of data
1 that we saw today.
2 The second question. Has the sponsor provided
3 evidence that fluoxetine is safe when used in the treatment
4 of PMDD?
5 Dr. Dominguez?
6 DR. DOMINGUEZ: Yes.
7 DR. TAMMINGA: Dr. Hamer.
8 DR. HAMER: Yes, attending in particular to the
9 sorts of things Dr. Winokur has talked about.
10 DR. TAMMINGA: Dr. Geller.
11 DR. GELLER: Yes.
12 DR. TAMMINGA: Dr. Cook.
13 DR. COOK: Yes.
14 DR. TAMMINGA: Dr. Winokur.
15 DR. WINOKUR: Yes.
16 DR. TAMMINGA: Dr. Fyer.
17 DR. FYER: Yes.
18 DR. TAMMINGA: Dr. Tamminga, yes.
19 I think we've had a day and an afternoon where
20 we've discussed an important issue, both an indication and
21 a drug. We've discussed perhaps, at least for our group, a
22 new style of data, if not necessarily for groups in general.
23 And we've taken a vote and concluded the meeting.
24 Thank you all very much.
25 (Whereupon, at 3:15 p.m., the committee was