1

DEPARTMENT OF HEALTH AND HUMAN SERVICES

FOOD AND DRUG ADMINISTRATION

CENTER FOR DRUG EVALUATION AND RESEARCH

JOINT MEETING OF THE

CDER PSYCHOPHARMACOLOGIC DRUGS ADVISORY COMMITTEE

AND THE

FDA PEDIATRIC ADVISORY COMMITTEE

Tuesday, September 14, 2004

7:56 a.m.

Holiday Inn Bethesda

8120 Wisconsin Avenue

Bethesda, Maryland

PARTICIPANTS

Wayne Goodman, M.D., Chair

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

PSYCHOPHARMACOLOGIC DRUGS ADVISORY COMMITTEE

MEMBERS

James J. McGough, M.D.

Jean E. Bronstein, R.N., M.S. (Consumer Rep)

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

Dilip J. Mehta, M.D., Ph.D., (Industry Rep)

Lauren Marangell, M.D.

Delbert G. Robinson, M.D.

Daniel S. Pine, M.D.

Barbara G. Wells, Pharm. D.

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

PEDIATRIC ADVISORY COMMITTEE MEMBERS

P. Joan Chesney, M.D.

Deborah L. Dokken, M.P.A.

Michael E. Fant, M.D., Ph.D.

Richard L. Gorman, M.D.

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

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

Judith R. O'Fallon, Ph.D.

Victor M. Santana, M.D.

SGE CONSULTANTS (VOTING)

Norman Fost, M.D., M.P.H.

Charles E. Irwin, Jr., M.D.

Lauren K. Leslie, M.D., FAAP

Steven Ebert, Pharm. D.

James M. Perrin, M.D.

Cynthia R. Pfeffer, M.D.

Robert D. Gibbons, Ph.D.

Tana A. Grady-Weliky, M.D.

Richard P. Malone, M.D.

Irene E. Ortiz, M.D.

Matthew V. Rudorfer, M.D.

SGE PATIENT REPRESENTATIVE (VOTING)

Gail W. Griffith

GUEST SPEAKERS (NON-VOTING)

Kelly Posner, Ph.D.

PARTICIPANTS (Continued)

GUESTS (NON-VOTING)

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

FDA

Robert Temple, M.D.

Russell G. Katz, M.D.

PARTICIPANTS (Continued)

Thomas Laughren, M.D.

M. Dianne Murphy, M.D.

Anne Trontell, M.D., M.P.H.

C O N T E N T S

Call to Order and Opening Remarks:

Wayne Goodman, M.D. 5

Conflict of Interest Statement

Anuja Patel 6

Opening Comments

Thomas Laughren, M.D. 9

Committee Questions and Discussion 31

Presentation

Diane Wysowski, Ph.D. 152

Committee Discussion of Questions and Vote 163

Concluding Remarks

P. Joan Chesney, M.D. 402

Wayne Goodman, M.D. 404

P R O C E E D I N G S

Call to Order and Opening Remarks

DR. GOODMAN: Welcome to day two of this

joint two-day session of the Psychopharmacologic

Drugs Advisory Committee and the Pediatric Advisory

Committee being held on September 14, 2004, here at

the Holiday Inn in Bethesda, Maryland.

We are convened to address recent concerns

about reports of suicidal ideas and behavior

developing in some children and adolescents during

treatment of depression with selective serotonin

reuptake inhibitors and other antidepressants.

Our goal is to gather information from a

variety of sources and perspectives to help us

understand this complex situation and ultimately,

to offer the best possible recommendations to the

FDA.

Now, I would like to turn the microphone

to Anuja Patel of the FDA Center for Drug

Evaluation and Research and Executive Secretary of

this committee to read the conflict the interest

statement into the record.

Conflict of Interest Statement

MS. PATEL: Good morning. The following

announcement addresses the issue of conflict of

interest and is made a part of the record to

preclude even the appearance of such at this

meeting.

The topics to be discussed today are

issues of broad applicability. Unlike issues

before a committee in which a particular company's

product is discussed, issues of broader

applicability involve many industrial sponsors and

products.

All Special Government Employees and

invited guests have been screened for their

financial interest as they may apply to the general

topics at hand.

The Food and Drug Administration has

granted particular matter of general applicability

waivers under 18 U.S.C. 208(b)(3) to the following

Special Government Employees which permits them to

participate fully in today's discussion and vote:

Jean Bronstein, Dr. Joan Chesney, Dr. Wayne

Goodman, Dr. Lauren Marangell, Dr. James McGough,

Dr. James Perrin, Dr. Bruce Pollock. In addition,

Dr. Philip Wang has been granted a limited waiver

that permits him to participate in the committee's

discussions. He is, however, excluded from voting.

A copy of the waiver statements may be

obtained by submitting a written request to the

Agency's Freedom of Information Office, Room 12A-30

of the Parklawn Building.

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

Victor Santana have de minimis financial interests

under 5 CFR Part 2640.202 that are covered by

regulatory waiver under 18 U.S.C. 208(b)(2).

Because general topics impact so many

entities, it is not practical to recite all

potential conflicts of interest as they apply to

each member, consultant, and guest speaker.

FDA acknowledges that there may be

potential conflicts of interest, but because of the

general nature of the discussion before the

committees, these potential conflicts are

mitigated.

With respect to FDA's invited industry

representative, we would like to disclose that Dr.

Dilip Mehta and Dr. Samuel Maldonado are

participating in this meeting as industry

representatives acting on behalf of regulated

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

Maldonado is employed by Johnson & Johnson.

With respect to all other participants, we

ask in the interest of fairness that they address

any current or previous financial involvement with

any firm whose product they may wish to comment

upon.

Thank you.

DR. GOODMAN: Thank you, Anuja.

We will be starting off this morning with

a presentation from Tom Laughren who will give us

an overview and also pose the questions, the five

questions to this committee.

Following his presentation, I would invite

questions. I also think it would be a good time

before we get into the meat of our discussions to

ask representatives from the FDA questions, to

further interrogate some of the data that was

presented yesterday.

Before we get into the actual discussion

of the questions, I would like us to think of the

questions that were carried over from yesterday,

pose those, and then we will take a short break,

reconvene and start the process of discussing the

questions.

Is that clear? Okay.

Tom, are you ready?

Opening Comments

Thomas Laughren, M.D.

DR. LAUGHREN: Good morning. I would also

like to welcome everyone back to the meeting today.

I would like to do a couple of things in my few

minutes here.

First of all, what I want to do is to

briefly review what I think are some of the key

findings from Dr. Hammad's presentation yesterday,

so that you have these in mind as you are

considering the questions before you.

Then, I want to talk a little bit about

what I think the data mean and talk about what some

of the regulatory options are as you are

considering our questions, and then I want to go

over the questions and the topics again.

These are the 24 trials that we are

considering. Again, 16 of them were in major

depression, and the other 8 trials were in several

various psychiatric disorders - OCD, GAD, 1 in SAD,

and 1 in ADHD.

Again, just for summary, I think these are

the three contributions that the Division made to

this effort. Again, we went to a lot of effort to

make sure that we had complete case finding. With

the help of Columbia, we accomplished what I think

is a rational classification of these events, and

we both obtained and included patient level data in

our analysis of the suicidality data again to try

and understand some of the differences both between

trials, within programs and across programs.

These are the outcomes that we looked at

again. The focus of the analysis was on two areas,

the suicidality event data and also on the suicide

item data.

For the event data, could we have the

other slide up that we had running yesterday? Our

primary endpoint, as you recall, was the

combination of suicidal behavior and ideation,

Codes 1, 2, and 6, where 1 was suicide attempt, 2

was preparatory actions, and then 6, suicidal

ideation.

So, that was our primary endpoint, but we

also looked at secondary endpoints, at suicidal

behavior, in other words, Codes 1 and 2, and then

suicidal ideation, Code 6, and then for our

sensitivity analysis, we looked at this larger

outcome including 1, 2, and 6, but also adding in 3

and 10, where again, 3 is self-injurious behavior

where the intent is not known, and 10 is not enough

information. Again, these are the cases where

there is injury, but it is not possible to tell

whether it's self-injury or other injury.

With regard to the suicide item data, we

looked at two measures about worsening suicidality

on that item or emergence, and these again are the

cases where the patients are normal at baseline and

have some increase during the trial.

In terms of our analytical plan, the major

focus was on doing risk ratio analyses, both for

the suicidality event data and for the item data.

In both cases, we looked at individual trials, as

well as for the event data, we looked at various

pools.

We looked at both by drug, we combined all

the SSRIs, MDD trials as a group, we looked at all

of the other indications combined as a group and

also did one pooling which included all 24 trials.

For the item data, we looked again at individual

trials and then a pooled analysis over all trials.

Dr. Hammad put a lot of effort into again

trying to explain the differences that we were

seeing between trials within programs and across

programs, and I just want to spend a couple of

minutes talking about exactly what he did.

He looked for confounding within trials

using both the univariate approach and a

multivariate approach. There were a total of 17

covariates that he looked at. He was not able to

find any evidence for important confounding in that

search.

He also did stratified analysis to explore

for effect modification. The three variables that

he looked at were age, gender, and history of

suicide attempt or ideation, so basically, what he

did in each of these is to stratify on these

variables within trials to look to see if there was

basically an interaction.

Again, he did not find any evidence for

that, so basically, what that means is that on

these variables, you find the signal both in

children and adolescents, you find it both in males

and females, and you find it both in those with and

without history of suicide attempt or ideation.

Finally, he looked at 12 trial level

covariates, again, as an attempt to try and explain

the differences across trials using a

meta-regression approach. Again, that approach was

not able to explain the variability.

Now, I would say that one of the problems

in doing these kinds of explorations is that there

is very limited power, you have a very small number

of events. When you use an eyeball approach to the

data, you can't help but thinking that trial

differences might have made a difference.

I just use the TADS, the fluoxetine

situation as an example. The company had three

trials. There was no signal coming from those

three trials. If you look at the careful screening

that was done to obtain the patients for those

samples, and the exclusions of patients with prior

histories of treatment resistance, and so forth,

and then you look at the TADS sample, which is many

ways was probably more representative of the

community of patients who actually get treated,

there is quite a difference. Again, as you recall,

in the TADS trial, you see quite a striking signal

for suicidality.

So, even though quantitatively, we weren't

able to tease that out and to explain the

differences using various quantitative approaches,

it is hard to think that that may not have made a

difference.

In my next three slides, I am going to

present very briefly some of the data.

What this slide is, is presenting the risk

ratios for various poolings. So, in this column,

you have the risk ratios on our primary endpoint,

which was suicidality ideation or behavior, 1, 2,

and 6.

In the second column, you have this

expanded sensitivity analysis, 1, 2, 6, plus adding

3 and 10. The first row is all trials, so this is

a pooling across all 24 trials. In the second row,

you have the pooling of the 11 trials with SSRIs

and major depression.

Now, there are two things I want you to

notice about this slide. First of all, in every

case, the risk ratios are around 2. They range

from 1.7 to 2.2, but they are sort of in the

vicinity of 2.

Secondly, if you look at the confidence

intervals on these risk ratios, in every case, it

does not include 1, so in that sense, it is a

statistically significant finding. So, this is the

pooled data.

What I have given you in this slide are a

different set of poolings. Here, what I am doing

is pooling the individual depression trials in the

7 programs that looked at depression, and these are

the 7 programs listed here. Every row is a

separate depression program.

What I have given you here, first of all,

is the outcome on our primary endpoint a

combination of 1, 2, and 6. I have also given you,

in the second column, the outcome on suicidal

behavior, and in the third column, the outcome on

suicidal ideation.

There are a couple of things I want you to

notice about this slide. First of all, in every

instance where we have events, and we had no events

for Serzone, but in the other 6 instances where you

have events, the risk ratio is always greater than

Now, I want to turn to trying to tease

apart where that overall effect is coming from if

you break it apart by behavior and ideation. Dr.

Hammad made this point yesterday, in three cases it

appears as if the overall effect is coming from

behavior, in three cases it looks like it is coming

from ideation.

So, if you look at Celexa, here is the

risk ratio for behavior, 2.23. There is nothing

happening for ideation.

If you look at Paxil, again, it looks like

it is coming mostly from behavior.

If you look at Prozac, it looks like it is

probably coming more from behavior than from

ideation.

For Effexor, there is a signal coming from

both, but it is clearly coming more from ideation.

Here, the confidence interval is almost

significant.

For Remeron, it is all coming from

ideation, and from Zoloft, there is nothing

happening for behavior, it is all coming from

ideation.

I am not sure what this means. As Dr.

Hammad pointed out, this may simply be a small

numbers problem, but we are not seeing a consistent

finding in terms of where the overall effect is

coming from.

Finally, what I have given you in this

slide is the data from the individual other 8

trials in non-MDD indications. As you get into

these trials, the number of events you are dealing

with is very small, and just to illustrate that, I

have put the actual number of events in this slide.

So, in each of these parentheses, the

first one is the number of events for drug, and the

second one is for placebo. So, you can see the

small number of events that we are dealing with.

If you recall from the previous slide, for

Effexor, we were seeing quite a strong signal for

major depression. These are two GAD studies.

There is nothing at all happening here.

For Luvox, again, Luvox was only studied

in OCD, there was no depression trial. Just one

study in depression, only two events. They were

both happening in the drug group.

For the two non-MDD Paxil studies, one in

social anxiety, one in OCD, again, small numbers of

events, but in both cases, they were happening in

the drug group.

The same for the Prozac OCD, just one

event, but it happened in the drug group.

No events for Wellbutrin.

For Zoloft, this is the only case where

the one event is happening in placebo, and not in

drug.

It is hard to know what to make of all of

this, although the one thing that you can't help

noticing is that even though there are a small

number of events, where events occurred, they most

happen on the drug side.

Just to summarize these data, again, if

you look at various pooled analyses, the risk

ratios hover around 2. They range from 1.7 to 2.2.

In all cases for those poolings, it appears to be a

significant finding.

The signal appears to be coming mostly

from major depression, although perhaps not

exclusively. Despite those findings, there still

are these inconsistencies in this risk, both across

trials, within programs and across programs.

On the other hand, my view is--and there

isn't necessarily one consistent view coming out of

FDA on this--but my view is that this is a

reasonably consistent signal for risk. You are

seeing it in seven of nine programs. We don't see

any events in Wellbutrin. On the other hand,

Wellbutrin was only studied in ADHD, just one

trial.

There is no signal coming from Serzone,

which was studied in major depression. I am not

sure if that means that Serzone is free of risk or

it simply may mean that the events, the

ascertainment in those programs was not good enough

to pick them up. I don't know the answer.

One other point that Dr. Hammad made

yesterday, that I want to return to, is a way of

thinking about this risk is in terms of risk

difference, and if you look over all these trials

and estimate what the risk difference is, that is

the difference in the risk between drug and

placebo, so you are subtracting the placebo risk

from the drug risk, it is in the range of 2 to 3

percent.

What that means is that again, out of 100

patients treated--this is short term now,

short-term treatment--you can expect 2 or 3 out of

that 100 will have some excess of suicidality above

and beyond what would be in the background that is

due to drug.

As a clinician, what you have to do is to

balance that risk against the perceived benefit.

The problem here, of course, is that we only have,

at least from FDA's standpoint, a demonstration of

benefit for Prozac, but if you take the TADS trial

as an example of benefit, there, you can look at

the benefit difference, and the benefit difference

in the TADS trial, difference between drug and

placebo in percent of responders, using that as the

measure of benefit, it is about 25 percent.

Again, you can interpret that in the same

way, so that if you look at 100 patients who are

treated with fluoxetine, you can expect that about

25 out of 100 will have that benefit if you are

looking at response as the benefit.

So, you balance that against the risk,

which again in that trial, the risk actually was

greater than the 2 percent, it was probably more on

the order of 7 percent, but you balance that risk

against the benefit. That is the kind of calculus

that a clinician has to do.

Finally, as was pointed out, there were no

completed suicides in any of these trials.

Again, we did not see the same signal in

looking at the item data. One exploration we tried

to do to see if that could be explained by patients

dropping out, and unfortunately, that was not an

explanation. The analysis of completers did not

show really any difference from the analysis of the

patients who dropped out.

So, how should these findings be

interpreted? I think that this is an indication

that there may be some increased risk for

suicidality during short-term treatment, and I

think this is probably a class effect. Again, you

are not seeing it in every drug that we looked at,

Serzone and Wellbutrin being the two exceptions,

but I think there is enough here to suggest that

this is probably a class effect.

The signal appears to be most compelling

in major depression. It may not be limited to that

population, but again we are left with this very

unusual variation in the signal across trials,

within programs and across programs that we have

not been able really to explain.

What I want to do next is to talk about

what some of the regulatory options are, and I

first want to talk about possible labeling changes.

As you recall, we already made a fairly

major change to labeling back in March, and all of

those changes have now been implemented. There is

a fairly prominent warning statement that directs

the attention of prescribers to this possible

event.

Now, that language as it currently is

written suggests that causality has not been

established. One thing that might be done to

modify that, if there is agreement on this, we

could say that causality has now been established

for this risk in pediatric patients.

In addition to that, we could go beyond

that and provide specific suicidality findings in

the labels for different products. We could also

provide more specific information about the

efficacy findings for specific products in that

language.

There are other things to talk about in

terms of that warning statement including things

like bolding language or putting black boxes.

These are all options that are on the table.

The other option that you need to think

about, and you heard many yesterday in the open

session ask us to do this, you can think about

contraindications. The one thing I want to point

out is that in this country, for our label, a

contraindication means never. It means that that

drug will never be used in treating these patients,

it is not an option.

The other thing I want to point out is

that the term "contraindication" has different

meanings in different regulatory settings. In some

settings, it does not mean never. If you read the

fine print in the UK, for example, there is a

suggestion that specialists may still use that

drug. So, you need to keep that in mind that in

this country, a contraindication means that that

drug is never an option.

In addition to labeling changes, there are

some other obvious actions that we can and almost

certainly will take. Our plan at present is to

write a medication guide. This is basically

labeling which ideally would be attached to the

medication when it is prescribed in unit of use

packaging.

In addition to that, we will undoubtedly

have another public health advisory when we decide

on what needs to be done, and we will try and

communicate these findings to our partners.

Now, what I would like to do again is to

quickly go through the questions and the topics.

The first topic is again we would like to have your

comments on our approach to classifying these cases

and to our analysis of the data.

One of the questions for which we really

need to have you vote on is do you feel that the

suicidality data from these trials support the

conclusion that any or all of these drugs increase

the risk of suicidality in pediatric patients.

If the answer to that question is yes, to

which of these nine drugs does this increased risk

apply, in other words, is this a class effect for

all antidepressants, does it apply to certain

subclasses within this broader class, or to

specific drugs?

If this is a class risk or if it applies

to certain drugs, how should this information be

reflected in the labeling for each of these

products, and what, if any, additional regulatory

actions should the agency take?

Finally, there is this question about what

additional research is needed to further delineate

the risks and the benefits of these drugs in

pediatric patients with psychiatric illness.

At our last meeting, I suggested one type

of study that you might think about, and I am going

to make that suggestion again, because we think

that this is one study that might get at one of the

deficiencies here, and that is, not only do we not

have enough information about short-term benefit,

we also have little information about longer term

benefit or risk.

One way of getting at longer term benefit

is the randomized withdrawal study. Basically, the

way the study works is that patients who are

responders or appear to be responding to treating,

at some point in the course of treatment, are

randomized to either continue on drug or randomized

to placebo, and one looks at time to relapse as the

outcome.

Now, I know there are concerns about that

design. You know, one concern is the ethical issue

of taking patients off a medication when they

appear to be responding. I agree that is a concern,

but I think there is a way of dealing with that.

The usual randomized withdrawal trial is

done after too short a period of time on treatment.

I mean typically, they are done now after 12 weeks

or so of treatment. That is too soon. No

clinician would take a patient off of one of these

medications at that point in time.

On the other hand, at some point in the

course of treatment, whether it is six months or

nine months or a year, it seems to me that it is a

reasonable question. At some point, you reach

equipoise where the clinician has to ask the

question, well, is this long enough, you know, is

there any benefit in continuing the treatment

beyond this point in time.

Now, that is a much harder study to do, to

keep patients on treatment for nine months or a

year before you randomize them, but that would be a

way of answering that important question of whether

or not there is continuing benefit beyond that

point in time.

The other concern that has been raised

about these trials is the issue of distinguishing

between withdrawal symptoms and relapse. Again, I

agree that this is a reasonable concern, but I

think there is also a way of addressing that.

In clinical practice these days, these

drugs are tapered. One doesn't stop them cold

turkey. I think that could also be part of that

design, and that could address that issue. So,

that is one thing to think about.

Before I end, I want to leave you with two

thoughts. We clearly have an obligation at FDA to

inform clinicians and patients about the risks that

are associated with these drugs, and we take this

obligation very seriously.

Along those lines, I just want to point

out that our current regulations do not require the

same level of certainty with regard to safety in

terms of causality as is required for efficacy. In

other words, we can issue warning statements with

somewhat lesser certainty about causality than is

required to support a claim.

Secondly, as I have pointed out several

times, the lack of efficacy data in this setting

for most of these drugs needs to be part of this

discussion. On the other hand, and I am not making

your job easy, please bear in mind that depression,

whether in adults or children, is a very serious

illness that is associated with morbidity and

mortality quite apart from whatever role

antidepressants might have.

As was pointed out yesterday, this is the

major cause of death in this population, the

depression itself, so please bear that in mind.

I have very profound respect and gratitude

for the clinicians who are out there on the front

lines still willing to take care of these patients

despite what has become a very controversial and

difficult environment.

I hope that as we discuss these issues and

make a decision, that we not make it impossible for

them to practice medicine.

Thank you.

DR. GOODMAN: Thank you, Tom, for a cogent

and clear presentation.

I would like to ask committee members if

they have any questions of Tom.

Committee Questions and Discussion

DR. FOST: This is for Tom or anyone else

who has a handle on the numbers. I know there is

no precise answer to it, but it would be helpful to

me to just hear you or someone else, maybe Dr.

Shaffer, if he is still here and is allowed to

talk, this question.

Suppose there were no SSRIs, suppose they

were contraindicated, that is, prohibited,

approximately, let me just ask the question about

suicides, about completed suicides, and I

understand there is no suicides in the FDA data,

but based on everything that we know,

approximately, would there be more suicides, fewer

suicides, or the same amount if there were no SSRIs

in children?

DR. TEMPLE: There is not going to be any

way to answer that, in part because you can't do

rigorous studies of the kind that would answer

that. No one is going to let you not treat, not

institutionalize, et cetera, someone who is getting

worse and worse, and it would require long-term

studies presumably against no treatment, and it is

not easy to figure out how anybody is going to do

those.

So, you are left with the kind of data

that people have pointed out is always uncertain,

the data on suicide rates and whether they are

going up or down, so it is very hard to answer that

question.

There were no completed suicides in the

pediatric data, so that doesn't give you a clue.

You can form your own judgment about whether

increased suicidal behavior or thinking is going to

lead to suicides in a certain fraction of cases.

It is hard to imagine that it couldn't, but you

don't know what that ratio is.

The success rate of suicidal attempts is

relatively low. I gather it is higher in males

than females, but I don't think there is going to

be ways to put numbers on that.

You have to form your judgment about

whether you think the overall decline in suicides

has got something to do with therapy or has

something to do with other aspects of life in the

United States, and nobody can give you a firm

answer to that, as Dr. Wysowski said and as others

have said. So, it is very hard to answer that

question.

Certainly, some of the people who spoke

yesterday, some of the treating physicians were

quite sure that they were helping people with the

drugs, and you heard families who said that their

relatives were made much worse by the drugs.

Putting numbers on that, though, isn't feasible

based on the data we have.

DR. FOST: A related question. To those,

Dr. Shaffer and others who note a decline in

suicides in the United States, in parallel with the

increased use of SSRIs, and let's just say which

should be an increase in suicidality, suicidal

ideation due to SSRI, what is the hypothesis there,

that there is fewer suicides, but more suicidal

ideation? That is what the data seemed to suggest,

and I am confused by that.

DR. TEMPLE: Can I make another comment?

The studies you are looking at are all the

short-term studies. As Tom was pointing out, we

have none of the long term sort of relapse

prevention data. It seems entirely possible that a

drug could be causing early suicidality, but once

you are over that period, it prevents relapse,

which could have an impact.

You know, there is just literally no way

to sort that out with present data. I mean it has

never been my thought that any benefit these drugs

have consists entirely of their treatment of the

acute episode, because in adults anyway, we have

lots of data showing that the likelihood and timing

of relapse is affected by continued therapy.

As Tom said, most of those studies go

earlier than you would like to do in a pediatric

population, because they consistently show that

quite reliably. Maybe that is where their

importance is, it is very hard to know.

DR. GOODMAN: Dr. Pine is next.

DR. PINE: I have a question about some of

the regulatory options. In thinking both about a

number of the comments that were made yesterday, as

well as your comments at the end about how

difficult the decision that we will have today,

related at least in part to the dearth of data that

we really need.

Are there any options from a

pharmacovigilance standpoint as far as regulatory

actions that might increase the degree to which we

are focusing over the next time period on the

emergence of these events or bring, you know, new

data over the next months to years based on a

regulatory action?

DR. KATZ: There is the mechanism of Phase

IV requirements that say we can impose requirements

on sponsors to do various studies in Phase IV and

postmarketing environment. The question would be

what those studies would look like. I think that

is the question.

There are other obviously entities, the

NIMH and others who were set up obviously to do

large trials, and again the question is what would

those trials look like. You could do I suppose

large long-term, and again, you have heard, I

think, a lot of people say that there is a need for

long-term data.

I suppose you could do long-term

comparative trials, you can't do long-term

placebo-controlled trials, so other than the sort

of randomized withdrawal design I think that Tom

talked about.

So, there is a mechanism to require

studies.

DR. PINE: I guess I am not so much asking

about studies, and this maybe is a bit of an unfair

analogy, but in New York, for example, as well as

other states, whenever you write a prescription for

a psychostimulant, there are a whole host of

procedures that kind of go with that, that are

designed to allow monitoring of the use of

psychostimulants and the associated effects.

Is there any--again, I realize I am

thinking a little bit out of the box--is there any

form of, I don't know, computer based or monitoring

system that might give us a better handle on how

many of these events are actually happening in

regular treatment?

DR. TEMPLE: ODS should comment on that,

but it is worth just looking at, say, the study Dr.

Jick tried to do. There isn't any no-treatment

group in that. He is just comparing the risk with

one group of drugs with another, and you can

definitely do studies like that, but if you tried

to compare treated people with untreated people,

there will always be the concern of whether the

groups are fundamentally different, a very

difficult problem because people are treated.

There might be environments in which

treatment is not so common, where there is less

likelihood to treat. Maybe in those environments,

you could do something like that, but Anne wanted

to talk.

DR. TRONTELL: Just to expand briefly, you

are talking about using observational data as Dr.

Temple pointed out, where you don't have a control

group, and although you might register patients, we

have seen even in clinical trials that we have been

discussing this past day, that the issue of

ascertainment of these events is very complicated

when you actually have a clinical trial mechanism

in place to capture those events.

The other challenge that you face with

observational data, because people don't receive

the drugs randomly, there is a phenomenon called

"confounding by indication," in fact, some of your

sicker patients you might presume are the ones who

are getting the medication.

We try and control for that, but it is

very complex. I think the better option is to

think of some systematic way, and then you are in

the realm of studies, as Dr. Katz was saying.

DR. MURPHY: I just wanted to follow up on

one last thing. Because we already know that using

the system we have now for follow-up

post-exclusivity because it is already mandated

that we do one-year reporting once these products,

whether they are approved or not, so we are looking

at all-use.

We do look at that and we report that, and

we know that that is not going to inform us, you

know, to answer the questions we need to answer,

because of all the things that will impact that

reporting.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: I just wanted to mention one

related, but not quite on-point matter. We talked

yesterday about concern that the studies that had

been done to gain exclusivity might have been not

as good as we would like.

We weren't particularly talking about the

design of the studies, which we think is okay, but

let's say the approach to them. Maybe there was

too much of a rush, and so on. If we were to put

out a written request now, it would be one that

required a third arm to the study, namely, a Prozac

arm, because we know that Prozac can be shown to be

effective.

So, the study wouldn't count unless it had

been able to show that it had what we call "assay

sensitivity," the ability to tell effective drugs

from ineffective drugs. We couldn't do that before

because there wasn't anything at the time we wrote

those requests that was known to be showable in

children, but now there is. There is three studies

that all seem to show something.

So, we should have much better information

about what the pediatric population does in future

requests. That doesn't help the present

discussion.

DR. MURPHY: I wanted to address that

issue again, too, because I think I want the

committee to be very clear on the fact that the

Agency tells the company very clearly the type of

studies that need to be done.

We do give them, you know, a broader

picture of the number of patients. We tell them

what we know will be the minimum, and, in general,

I think Tom would agree that most of these studies

have come in with the numbers in each arm that we

have seen in other studies where they have shown

effectiveness.

So, the point here being that we do have

control over the types of trials that are done, the

number of patients, and the monitoring. However,

because there is a template up on your web that

basically tells you what we ask for in depression

trials.

When you look at what the safety is, as

has been pointed out many times, these trials were

not set up to answer that question. So, I think it

is those kinds of issues that we would like to hear

more about today. As Dr. Temple said, it is how

better to do these trials in the future.

Thank you.

DR. GOODMAN: Thanks for that statements,

Dianne. I just want to make sure I understand it

completely.

I think what you are saying, that if the

conditions had been different at the time, that is,

that the drug company was required to show, not

only have a study, but a study that was positive.

Then, the design would not have been any different,

the sample size would not have been any different

under those circumstances than the ones that

existed at the time.

DR. MURPHY: I think what we are saying,

that for the trials that we designed, they were the

same for the one that did show some effect, which

is Prozac, as those that did not, and that what we

don't know is if a company is putting a trial

together, and let's say we said that they had to

have 300 patients to get their exclusivity, but for

other reasons they really wanted this product

approved, and they felt the enrollment was not

going the way that they needed, would there be some

other push within that company to then go out and

get more patients, so that their enrollment would

be better versus an exclusivity where all they had

to do was meet that criteria.

I am making that number up. I think the

issues that people were trying to get at is that is

there a difference that affects behavior when you

just know you have to do certain things versus you

have another goal, which may be approval.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: The requirement for a third

arm in evidence of assay sensitivity leaves it up

to the company to decide how they are going to do a

successful study. They can look at the available

data on Prozac and say, oh, here is the number I

need, here is the kind of patients I need. That

succeeded in those three trials.

They would then know that the trial would

have to be one that can show the difference between

Prozac and placebo. That doesn't mean their drug

has to show a difference between drug and placebo.

That would be determined by the results,

and there is no obligation that the drug be

successful, but we would at least know we had a

study that was capable of detecting effective drugs

and distinguishing effective drugs from ineffective

drugs.

That would then become a requirement for

meeting the terms of the written request because

they would have to show that they had an adequate

study. Before there was an effective drug, there

was no way to do that. You couldn't tell whether

the study was a good study or not.

DR. GOODMAN: Dr. Marangell.

DR. MARANGELL: If I could go back and

address the question of what would the hypothesis

be for long term, certainly, in the absence of

data, there is some degree of speculation. I do

have a question directly to the FDA. Is it okay if

I respond?

I think the number one hypothesis would be

in the short run when you have depressed patients

who are not yet stabilized, you may see an

increased risk, and you do see certainly in this

population an increased risk of suicidality.

I imagine that what we would see with

longer term data is a substantial decrease in

suicidality over time, and that is what we are

inferring from the cohort and the epidemiologic

data. I think that clinically makes sense, as well

as mechanistically makes sense.

The question for the FDA, can you give us

a sense, I mean do we feel confident that we

actually have all the available studies now in both

children, adolescents, as well as in adults, and

what is the FDA policy on requiring review of those

studies including negative studies, when do they

come to you and when do they become publicly

available?

DR. TEMPLE: Well, let me start, others

can comment. When you submit to us an application

to change the labeling, to add a claim, say, for

pediatric use, you are clearly obliged under the

law to provide every study, successful ones,

unsuccessful ones, things that were interrupted,

and so on.

As far as we know, we are getting all

those studies. Of course, if there were something

that were done that we didn't know about, well,

then, we wouldn't know about it, but as far as we

know, we are getting them all.

So, most of the pediatric submissions to

us were associated with labeling requests or

something like that, so as far as we know, we have

all those data.

Dianne can tell you what is required under

the best BPCA, and I think there, too, they have to

provide them. We have no rule that affects whether

people have to publish results. Congress is

considering that, so are the journals and everybody

is talking about that.

Under BPCA, however, when we grant

exclusivity, we provide summarized results, and we

have done that for the drugs where the written

requests were written after the BPCA, and we have

gone back and asked the companies for permission to

summarize our analyses for all of the others where

it wasn't totally clear whether we could do it or

not.

So, the summarized result, that is not the

same as a complete study report, the summarized

results are now available publicly on all of those.

I am sure between PhRMA's commitment to provide a

registry between the journals insistence that they

will get a registry, between congressional

interest, I am quite confident that there will be a

change in the way things get published.

DR. MURPHY: The only thing that I could

add to that is that for the committee, for the

routine practice within FDA, if a company submits

an application, we review it, the studies are

negative, there is no public acknowledgment of that

unless the company for some reason wants to make

that knowledge public. We are not allowed to

comment on that.

Now, under BPCA, it said, it has a

disclosure section that says you, FDA, will

publish, as Dr. Temple is referring to, the

summaries, the medical and pharmacology summaries

up on the web--make them public, and actually, we

have chosen to do that on the web--and we have done

that.

One of the issues that has happened is

that between the enactment of the new legislation

and the old legislation, legally, things were

considered issues under the old legislation, so

even though the studies came in, we had to reissue

all those written requests to be able to say they

now were subject to this new mandate.

So, what again Dr. Temple was telling you

is that unfortunately, many of the antidepressants

came in, in that period when we had not yet issued

that letter, but despite that, we have asked the

sponsors to allow us to put those summaries up, and

they have given permission to do so.

That is why yesterday we said up on the

web now are the summaries. Again, this is not the

data. There is variations in, you know, some

medical officers will put in more information than

others in how much data is in these summaries, but

they are up now, publicly available.

DR. MARANGELL: Is that true for adults,

as well?

DR. MURPHY: No, adults are still under

the same standard. In other words, if the study is

negative, we don't talk about it.

DR. MARANGELL: So, as an example, if an

antidepressant manufacturer did a study in a new

indication for a drug that is currently available,

found increased risks of suicidality, no one would

be under any obligation to make that public?

DR. MURPHY: That is a different issue.

DR. MARANGELL: But that is the question.

DR. MURPHY: The issue is safety, and the

Agency always has the ability to make public safety

issues that arise.

Bob, do you want to say anything else

about that?

DR. TEMPLE: We consider, for example, if

someone with an antidepressant comes in for, I

don't know, obsessive compulsive disease, and we

don't buy it, we do not make those data available,

they are considered confidential commercial

information. Obviously, a lot of the people, a lot

of the public doesn't like that approach. We think

that is what we are required to do. I can't

comment on that, I am not the lawyer here.

However, companies have a separate

obligation for drugs that are marketed to report

serious and unexpected, and any serious adverse

reactions to us, and to do so promptly. A finding

of increased suicidality where that was not known,

clearly meets that test, and they would be obliged

to report it to us. If we then thought that was

true, we would add it to the label or do whatever

we are supposed to do.

So, safety data meets a different

standard. A new carcinogenicity study or

something, those do have to be reported to us.

Other studies have to be reported in the

annual report, but they are not necessarily

reported in detail, and not that much is

necessarily made of them, and they do not

necessarily become public.

DR. GOODMAN: Dr. Pollock.

DR. POLLOCK: Yes, the serious safety

issue would have to be reported while the trial is

ongoing to you, right?

DR. TEMPLE: Well, if it arises from a

trial, it has to be. Actually, the requirements

for reporting serious unexpected events in a trial

are more or less identical to the requirements

before a drug is marketed. They have to be

reported to us within 7 or 15 days.

A finding from an epidemiologic study,

there is some judgment involved in whether that

represents the kind of thing that has to be

reported promptly, but they basically do.

DR. KATZ: There is also some judgment

involved in whether or not an event is considered

to be unexpected. So, for example, suicide in a

study of patients who are at risk anyway might not

be reported to us in real-time, because it might be

considered to be expected, the blind is still

intact, you don't know if it's drug or placebo if

it is in the context of a controlled trial.

Afterwards, though, when the trial is done

and analyzed, and it turns out that there is an

increased incidence on drug compared to placebo,

that is something we would find out about.

DR. GOODMAN: Go ahead, Dr. Pollock.

DR. POLLOCK: I actually wanted to explore

your thinking a little bit about the recommendation

for a maintenance trial. I guess there are a

couple of things. One is if there is this acute

toxicity that we are concerned about, clearly, it

doesn't address that because you are dealing with

the children or the adolescents who have actually

responded, and then are withdrawn.

But I wondered if there was implicit in

your request for that, a concern that still that

the shorter half-life SSRIs seem to be, maybe not

statistically, but certainly qualitatively more at

risk in causing this phenomenon.

I was taking that as implicit perhaps in

your suggestion, maybe I am over-interpreting it,

but is there a belief that somehow--I mean it just

seems more than coincidence that signals seem a

little bit higher.

I know it has now emerged with Prozac, but

certainly, Effexor, venlafaxine stands out at one

end, then followed by paroxetine, and if there was

kind of an implicit question that you were asking,

assuming that people are still using after we are

finished, you know, those medications, that you can

require that those manufacturers actually conduct a

serious maintenance trial as part of you were

saying your Phase IV regulatory requirement.

DR. LAUGHREN: We certainly, you know,

until we saw the TADS data, were entertaining the

notion that discontinuation might be one

explanation for the bigger signal, the apparent

signal that we are seeing with Paxil and Effexor.

The TADS finding certainly challenges that

notion as a unitary explanation, since that is the

single trial among the 24 that, by itself, has a

statistically significant finding for that signal.

That doesn't mean that the other explanation isn't

possible. I mean this could be a much more complex

situation than one might seem at first glance.

But a maintenance trial is not going to

answer all those questions. I mean a maintenance

trial is only going to answer the question of

longer term benefit, but the reality is that many

clinicians, despite these concerns, are probably

going to continue to use these drugs, and we have a

dearth of information about what the longer term

benefits are. The maintenance trial is one way, I

think, of getting at that.

Now, there is this issue of how to

interpret emerging symptoms in that setting, you

know, when you take patients off the drug. Of

course, the drugs like Paxil and Effexor, that are

known to have a stronger signal for

discontinuation, obviously are a challenge in doing

that kind of trial, but I think that one could, as

one does in clinical practice, taper those patients

to try and address that, and then look for what

would be considered for relapse.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: There is another reason to do

randomized withdrawal studies. As everybody knows,

in adults, the failure rate for conventional

clinical trials of the acute episode is about 50

percent. That is, half the trials can't tell drug

from placebo, and that is true even when you

include a third arm of a drug that is known to

work. That appears to be the nature of the beast.

Nobody really has a good explanation

because if they did, they would fix it, but we at

least think it has something to do with the

environment and the discussions that go on even

informally, even if it is not planned as part of

the treatment.

In the randomized withdrawal setting, the

success rate for drugs that are known to work is

nearly 100 percent. Very few of those trials ever

fail.

There are at least two reasons. One, only

people who seem to be doing well are in the trial,

so they are enriched with a responder population.

You can make of that what you will.

The other possibility, though, is that the

environmental things that help people get better

aren't really there, they are just out living in

the community, there is nothing nurturing about it.

They are just back in their usual environment.

So, one of the attractiveness of these is

to find out whether the drugs actually provide some

benefit, even in people who seem to be doing well

on them, which seems an important question here. I

mean, as Tom has pointed out repeatedly, the

failure of most of the drugs to show effectiveness

doesn't mean they don't work. On the other hand,

we don't have evidence that they do work, and that

is not irrelevant either.

A good way to show that, if they do, is

the randomized withdrawal study. At least that has

been the history in adults, so there is a lot of

attractiveness to it.

DR. GOODMAN: Dr. Chesney.

DR. CHESNEY: Thank you. I have two

questions. The first one is for Dr.

Murphy and Dr. Temple, and the second for Dr.

Laughren. The first question addresses the

exclusivity issue. I feel like in this case, we

bypass the Phase I/Phase II stages that we would

normally go through with new drugs, so we never did

do the pharmacokinetic/pharmacodynamic dose finding

in children that we would have done had these been

new drugs.

I wondered, I probably should know this,

but could either of you explain, when we offer

exclusivity with a new drug, if it is a new drug to

children, do we require those studies, or do we

not? I am sure it is not that straightforward.

DR. MURPHY: We did required

pharmacokinetic studies. Actually, on the

template, we outline three types of studies. They

have to do two randomized, double-blind,

placebo-controlled, acute treatment trials with

recommendation at six to eight weeks for safety and

efficacy. They also are to do a pharmacokinetic

study to provide information pertinent to dosing of

study drug, and they are to do a safety study.

So, all of those were asked for. Now, if

you are asking do we go back and demand redoing

dose finding again in these, no, they were not

worded that way. It was said that the PK study

could be a traditional PK or, alternatively, a pop

PK, and actually, I don't think that the study had

any other information that would have, in essence,

told the company that they needed to redo the dose

finding, if that answers your question.

DR. CHESNEY: So, do we have dose

information on all of these drugs? Do we know what

the usual ranges are, and what excessive ranges

are, all those things?

DR. GOODMAN: Go ahead, Dr. Katz.

DR. KATZ: I think Tom mentioned this in

one of his slides yesterday. The written requests

that we issue now are very different from the

written requests we issued that probably generated

most of the trials that we are talking about here

yesterday and today.

As Dianne pointed out, for example, in

pharmacokinetics, we gave sponsors the opportunity

to generate the kinetics in kids based on so-called

population pharmacokinetic analyses, which is to

say from data generated in the controlled trials.

So, it was sort of after the fact. It was

just what is the kinetics of the doses you happen

to give in the trials.

In the earlier written requests, it was

sort of the pediatric drug development was sort of

tacked onto the adults, in other words, when the

trials were designed even, the treatment effect

size, for example, was used to calculate sample

size was taken from the treatment effect size seen

in adults. We had no information, even preliminary

information in kids.

So, we didn't have a lot of preliminary

information in those days that could inform

adequate trial design in this population, in this

setting.

Nowadays, we ask for different things. We

ask for formal PK, so we can learn before the

definitive trial design, what the kinetics are,

what doses give rise to what plasma levels. We ask

for dose finding studies, so we can determine

before we design the definitive trials what the

tolerated dose range is.

So, the written requests are much

different now than they were at the time that the

requests are generated, these data were written.

DR. MURPHY: Just to reinforce that is

that these were some of the earliest written

requests that went out, so they really, as has been

stated, and I think we tried to say this earlier

on, we are learning.

I mean because of the lack of prior

research and some fundamental scientific questions

haven't been answered, we are learning from the

trials that we have now about how to do a better

job on designing some of these trials, but these

were some of the very earliest ones that were

issued.

DR. CHESNEY: Dr. Temple, did you want to

comment on that?

DR. TEMPLE: Well, I just wanted to say

there isn't any pharmacodynamic measure to allow

you to do what is called PK/PD other than

effectiveness itself. In a lot of cardiovascular

settings, there is at least something you think

relates to the desired effects, so you can do

relatively short-term studies and get a PK/PD

relationship.

Here, your only way to do it is to insist

that every drug, every study be a dose response

study, which is of considerable difficulty. We

have trouble getting really good data even in

adults actually given the sample sizes involved,

but there isn't any measure yet. Maybe one of

these days there will be an MRI measurement or

something, but not yet.

DR. CHESNEY: Well, I don't want to

overstay my welcome, and I do have a question for

Dr. Laughren, but one does wonder about some of

these children who didn't even express ideation and

just suddenly, very early on, if they didn't have

excessive levels. I guess that is one issue I was

getting to.

Dr. Laughren, I wanted to come back to the

point Dr. Pine was making. I thought Dr.

Reisinger's point in the open session yesterday was

a very interesting one, which is that you would

have to undergo some kind of computer-based

learning program or some kind of program that

authorized you to prescribe psychoactive drugs.

Certainly, we have to do computer-based

CBLs for all kinds of things in our hospitals and

in other areas nowadays. That had a real

attraction to me, and I guess the question is what

kind of authority does the FDA have in an area like

that, can you say that anybody that prescribes

SSRIs must do a computer-based learning program on

line, or is that something that the professional

societies take on?

You offered several options, black boxes,

revised label warning, but is this a potential

option?

DR. TEMPLE: We can certainly recommend

things like that. Every labeling for a cancer drug

says that this should only be used by people who

are trained in oncology. That comes with no

enforcement on our part except that people may be

anxious about the consequences if they don't have

that training.

A labeling recommendation is certainly a

possibility. A step further to limit the drugs to

people who have been given that way, those are very

iffy questions, and it is not clear whether we can,

in fact, do that. There would have to be a debate

about it.

There are some examples of fairly

interventionist activities. As you all know, you

can't get clozapine unless you have a white blood

count, so no blood, no drug.

There are not a whole lot of other

examples like that, but there are other cases where

patients must be given a form that lists what some

of the adverse effects might be, and things like

that. You have to weigh the risk you are concerned

about with the burdensomeness to the community and

to the medical profession of those kinds of

interventions.

Putting something in labeling about what

you should know doesn't carry those kinds of

concerns, so if something sensible, suggesting that

people ought to be trained in a certain way seemed

like a reasonable thing, we could certainly

consider that.

DR. TRONTELL: I would just like to add on

to Dr. Temple's comments, because the FDA regulates

drugs, but doesn't regulate the practice of

medicine, and we walk a fine line in terms of

dealing with some drug products where we may feel,

as with clozapine, that only very tight controls on

prescribing and dispensing and use of the product

are allowed.

There are a very small handful of drugs,

they tend to be the exception rather than the rule,

where training has been required as a condition of

approval. One product in particular is the drug

product dofetilide, where, in fact, training is

required for pharmacists or clinicians. There is a

highly structured way in which that product can be

used.

Again, those have tended to be reserved

for situations where we feel the drug cannot be

safely used without that very high level of

precaution. It is extremely difficult to put those

in place for products that have already been

marketed and used by professionals.

DR. CHESNEY: The public sees your role I

think in a much broader perspective, as we heard

yesterday, and I think that is something that is

useful to clarify as to where your limits are. You

mentioned there is a fine line, and I think that is

what we are all looking for, is where does your

authority end and that of prescribing physicians

begin, I guess in a sense.

DR. TRONTELL: I don't think we yet have

an answer. I think we always have the authority of

our agency and hopefully, our ability to persuade

individuals, but I think that the actual legal

authority to do some of these is a matter of debate

within and outside of the agency.

DR. GOODMAN: Dr. Nelson.

DR. NELSON: I would like to return to the

topic of the incentives on the part of industry to

perform well-conducted trials.

There has been a lot of discussion about

the evolution of the written request and about the

improvement with three-arm studies and changes in

the ability to request that, but my understanding,

I am interested to know if this is accurate, is

that there is still two potential linkages that

don't exist that might decrease the incentive to do

a well-conducted study, and that is, absent safety

concerns, there is no tie to putting any efficacy

information in labeling, so that they receive

exclusivity if a labeling change occurs.

Second, is that there is no link of

exclusivity to a well-conducted study unless that

has changed with written request, since I read them

on the current web site, there is one asthma study

where there was members of the drug group that had

no drug level, members of the placebo group that

had measurable drug levels, and the FDA concluded

that the data was uninterpretable, but

nevertheless, exclusivity was granted.

I am wondering, is that a problem with the

written request that is now fixed, or is there

other solutions that would need to be put into

place, such as legislation, to address those two,

what I perceive as gaps.

DR. MURPHY: I think there was significant

discussion about how exclusivity should work,

should it be only if the product is approved. I

was not privy to those discussions, but I know they

occurred.

The reason for why it was put in place the

way it is, I can't give you, Dr. Nelson, but I can

tell you that one of the explanations I have heard

is that there was such little data, and FDA was

given the authority to define the trials, so again,

as you have heard, we would like to improve, and we

know we have to learn from what trials we have,

that by providing FDA the authority to define the

trials, that they hope that the trials would be,

you know, of the best that they could be, and that,

therefore, we would learn from the trials even they

were failed, because that is important information,

failing is important.

So, I guess what you would say, you are

asking if, and that is in a number of our labels,

and that is a whole other discussion, but in

situations, you know, we know that is the only

study we are going to get and this is it, failing

is put, that they failed to show effectiveness has

been put in the label in a number of situations,

and certain dosing or safety information.

As I said, about a fourth of the time, we

are describing, even irrespective of whether the

study is positive or negative, we are finding

safety signals, you know, important dosing

information, and we are able to put that

information in a label.

The intent is that the information that is

obtained, whether the product is proven to be

effective or not is important, and that safety

information, et cetera, would be obtained.

So, that is the best explanation I can

give you as to why it is set up the way it is right

now.

DR. NELSON: I understand, but let me

focus my question, I guess. Right now the efficacy

or lack of efficacy data is not in the existing

labeling that we are discussing, so, for example,

just to pick one, paroxetine, there is five

studies, and the pediatric use just says it has not

been established.

Although that is a true statement, it is a

bit misleading because many people interpret that

to mean the studies hadn't been done.

The other question is you could ask them

to do a three-arm active control study, but if they

do it badly, do they still get the money? Even if

they have done it, if they do it badly, do they

still get the money?

DR. GOODMAN: Dr. Temple wants to respond.

DR. TEMPLE: If the written request says

you need to do a three-arm study and need to show

that the trial has assay sensitivity, that is, the

ability to distinguish active drugs from inactive

drugs, and the Prozac arm doesn't beat placebo,

then, they would have failed to meet the

requirement of the written request.

We couldn't do that before, as I said,

because we didn't have a known active control, so

we wouldn't have known what to say. So, in that

case, the incentive to do a proper study becomes

quite clear. If they don't do a proper study, and

succeed in showing that, they would not get

exclusivity.

In other cases, we have insisted that the

variance be such that for, say, a blood pressure

drug, an effect size of 3 or 4 millimeters of

mercury could be detected, so if the whole thing is

done sloppily and they could not have detected such

a thing, then, they would not get exclusivity.

Some of the other things, however, that

you mentioned, don't have an obvious remedy. I

mean I guess following the example you said, we

could say, oh, by the way, people should have blood

levels showing that they took the drug. Well, we

hadn't been smart enough to think of that, and

maybe that is something we should be adding, that

is, some kind of compliance check.

That, I don't think has been part of

written requests to date. That doesn't mean it

couldn't be. The test that Congress imposed is

that if you comply with the terms of the written

request, you get exclusivity. That means if we

weren't smart enough to ask a question, that is not

considered their fault, and they are supposed to

get it.

DR. MURPHY: And we have denied

exclusivity where we thought the trials were done

sloppily, and actually, sometimes when the sponsor

said, well, we know you asked for this, but we

didn't think it was correct to keep going, so we

didn't do this for some reason, and we said, no,

you should have come in and talked to us about why

you weren't going to do it, you didn't do it, we

told you, you need to do it, sorry, you don't get

it.

So, what I guess we are trying to say is

if it's really sloppy, and they don't do what we

tell them, we deny them exclusivity. The problem I

think we are dealing with here is that we all are

learning how to better do the trials, and your

other question about whether that should go in the

label, the negative information should go in the

label, is a whole other discussion.

DR. GOODMAN: I have a list of seven other

committee members who wish to speak. After we give

them that opportunity, I am going to ask Dr.

Wysowski to come up to the podium. We had asked

her to follow up on something from yesterday. Is

there somebody else that has a burning--we have one

more and that is it--two more, that's it.

Dr. Irwin. His question has been

answered. Thank you.

Dr. Rudorfer.

DR. RUDORFER: Yes, thank you.

I would just like to revisit a couple of

issues that concern me at the front end of these

studies, and I recognize everyone from the FDA is

pointing out that this is a learning process, on

the other hand, we are faced with the dilemma of

having these particular trials to deal with.

The dosing question that was just

discussed brings to mind a concern I have related

to how the suicidal events we have been looking at

were ascertained.

As I understand it, for the most part,

these were from adverse events questionnaires and

surveys. Is that correct? I mean there was no

particular suicidal scale?

DR. LAUGHREN: Well, all of these trials

included standard depression rating instruments

like HAM-D or CDRS, and so forth, and there is a

suicide item in each of those instruments, and that

is part of what we analyzed.

But the problem is we don't really know

how those were applied. My guess is that most of

the event data we are dealing with were spontaneous

report or general questioning rather than specific

ascertainment.

That is really one of the areas that we

are trying to explore with Columbia to try and work

on a more specific instrument for improving

ascertainment for suicidality, but no, in these

trials, I don't think ascertainment was very

specific.

DR. RUDORFER: My question, as we deal

with these data, would be this. I appreciate the

very dedicated and elegant work that both the FDA

and Columbia have applied to these findings. The

question I have relates to the issue of the active

drug versus placebo groups.

Since it sounds as if much of the data

were spontaneous reports or I assume perhaps

discussion between the raters and subjects, or the

investigators and the subjects, I am wondering if

part of this is not dependent on the assumption

that the blind was kept intact throughout the

studies, and I wonder if we have any measure of

that or any sort of quality control on that issue.

DR. LAUGHREN: No, we have no idea of

that. That is typically not something that is

really ascertained. It is the assumption, but how

would you check on that?

DR. RUDORFER: In some studies, patients

and raters are asked at some point. I mean here, I

am just wondering if, in fact, if a patient

volunteered that, for instance, they were

experiencing some side effects, they come in, the

rater asks how are you doing this week, and their

first comment relates to GI distress or something

that sounds like a side effect, if they simply

don't get more attention, in other words, maybe

there is more discussion, maybe there are more

questions asked as opposed to a patient that comes

in and say, gee, I am feeling pretty good, I don't

seem to have any side effects.

Again, that would not obviate the fact

that if we find signals, then, the signals are

present. I guess I am just concerned about the

active drug versus placebo difference.

DR. GOODMAN: Let me interject. I don't

think I am as concerned about the unblinding, but

your question raises at least in my mind the

possibility that in the data, is it possible that

we would see other somatic symptoms, more side

effects reported in those patients, who also

reported suicidality than in the opposing group,

was there any attempt in the data to look at

whether there were any other--was any other

increase of adverse experience outside the target

symptoms of suicidality in those patients who

reported suicidality, the reason being that if

there was, that would suggest it was part of a

larger behavioral syndrome that was being induced

by the medication.

DR. LAUGHREN: Our analyses had to be

limited by what we had in our database, and we had

to design this database late last summer. We

didn't anticipate all of these questions. As it

was, the database we had was a very time-consuming

process to put together. It took a number of

months to get it.

They are all good questions, but we don't

have all those answers, but I agree that

ascertainment for suicidality was not optimal here.

DR. GOODMAN: But the question is at this

point, could you go back to that same data and look

to see if there is a higher rate of other adverse

experiences reported in those patients who were

also identified as experiencing or exhibiting

suicidality.

DR. LAUGHREN: Not without designing

another database and going back to the companies

and waiting for a number of months, and I am not

confident enough in the quality of the data we have

here that that would justify that additional

effort.

I mean again, these are all good

questions, but we are faced with making a decision

at this point in time with what we have, and we are

asking the committee's advice on what you think we

can do now based on what we have done.

DR. GOODMAN: No, I agree with that, I

understand that, but we were also asked what other

advice we would give in terms of future research or

studies or data that we would like to see.

DR. TEMPLE: Tom, we did look at the

association with certain kinds of things, the

activation syndrome, things like that, right?

DR. LAUGHREN: We included in our database

two other symptoms, hostility and agitation based

on the preferred terms that the companies used, and

again, we haven't looked, I suspect that there is

variability across different companies in what

actually got subsumed under those two things.

There are the only two other events, and

we don't even have the timing for that. All we

have is an indication of whether or not, at some

point during treatment, a patient experienced

agitation or hostility. We don't have all the

other somatic kinds of things that you are alluding

to. That would mean going back and trying to

create another database.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: Let me just mention one other

thing that has come up briefly and that Dianne

touched on, and that is inclusion of negative

results in labeling.

As Tom has explained at the previous

meeting and now, as a general policy, we don't

usually put in labeling the fact that a study

hasn't worked, because we don't think that proves

that it doesn't work. It just means that that

study failed.

But we are actively thinking about that

policy for the pediatric part, because the whole

point of doing the studies was the possibility that

adults and children are different, otherwise, you

wouldn't even think about doing that whole program.

All I can say is we are actively thinking about it.

It is not an easy to thing to do, however,

because what you would want to say could depend on

how good you thought the study was, and then there

is the conundrum of what do you do if there is one

study that says yes and one study that says no.

That is virtually somebody a claim, which we really

wouldn't want to do if it wasn't merited.

So, I am not going to suggest that this is

easy, but we are reconsidering this whole thing,

because the whole point of the Best Pharmaceuticals

for Children Act is to find the data and see

whether drugs work in children, and not putting

anything in seems funny, so we are reconsidering

that.

DR. GOODMAN: Dr. Perrin.

DR. PERRIN: Part of my question Dr.

Chesney eloquently asked before, but I wonder if we

can get access to the wording that you used for

cancer drugs as perhaps a guide to us for our

considerations.

My other quick question, I think back to

one of the FDA group is am I hearing you right that

if you have a drug that has been shown to be

efficacious in a particular indication, that all

trials requested in the future require an arm that

includes that drug?

DR. TEMPLE: I am not ready to say that

one would always do that, there are other ways to

try to assure quality, but in this setting, it is

reasonable to assert that we need to know whether

your trial was an adequate test of whether this

drug worked, and the only way we know to be sure

that it is an adequate test is to have an active

control, and to have that active control be

distinguishable from placebo. Then, you know this

is a trial capable of showing things.

We have determined that our future written

requests will include a requirement for a three-arm

trial, because that's the only way we know to be

sure that the trial is a trial that is capable of

showing what the answer is, and we want to be sure

we get the answer.

This comes up in written requests all the

time, how much assurance do you have to have and

how do you gain that assurance that the trial is a

useful trial, and the reason it comes up is the one

that everyone has alluded to, we don't think people

are deliberately trying to mess things up, but the

incentives to do a really good trial are greater

when you have to win.

DR. PERRIN: I am a little confused. As a

clinician, you know, typically, if I am looking at

a new medication, I want to know that it is better

than current therapy. I mean all of us are really

interested in that.

There are a number of pediatric drug

trials, not in the area of antidepressants that I

am aware of, where new drugs come on the market,

approved by the FDA, where there are only

drug/placebo trials, and not trials comparing the

new drug with currently approved FDA medications.

That is where I am confused.

DR. TEMPLE: Good question. There are two

possible uses for having an active control. One is

where you want to compare the two therapies. Now,

to do that, you would need a very, very large

study, because you would be interested in even

modest differences. That is not what we are

talking about.

We are talking about the use of the third

arm to show something about trial quality.

Actually, a third arm is extremely common in

depression trials now, because if the trial fails

to show that your drug is better than placebo,

there are two possibilities.

One is that your drug is no good, and the

other is that the study was no good, and it is very

important to somebody developing a drug to know

which of those two things it is.

If the trial shows that Prozac, say,

works, and your drug doesn't, you get rid of the

drug. If the trials shows that neither Prozac nor

your drug works, you do another study. So, it is

extremely important. But the two purposes of the

trials are quite different.

To actually do a comparison and try to

detect a small difference, you would need very,

very large groups. That is an unusual thing for

people to do, and usually, the drugs can't be

distinguished. It is very hard to do that.

DR. GOODMAN: Dr. Temple, I heard you say

before, if I heard correctly, that incentives are

different when you need to win. Were you referring

to the conditions of the six-month exclusivity

arrangement, and if you were, if the incentives

were different at that time, would you predict any

difference in the design of those trials or the

conduct of those trials?

The reason, let me say, I think that many

of us keep harping on this point, is not so much

because we think that the suicidality data would

have turned out differently. I think it is the

absence of a benefit, the absence of efficacy that

at least I am concerned about, because that is

mostly what we have in assessing benefit or those

trials, and we only have 3 out of 15 that are

positive, so if there was something about the

conditions in which those studies were designed or

conducted that might have negatively impacted the

outcome, I would like to know it.

DR. TEMPLE: Well, Russ and Tom need to

respond, but we haven't seen anything in the design

of those trials as written in protocols that makes

them look any worse than any other trials. They

seem to have reasonable size, so there is nothing

obvious.

But, you know, this is an issue that comes

up when you do so-called "non-inferiority" trials.

The incentive, you know, the point of such trials

show no difference between treatments, and as I

have written repeatedly, that is not a good

incentive to give people.

It doesn't stimulate the kind of optimal

behavior that you want, which is stimulated by the

need to try to show a difference between

treatments, and that is a problem here if you don't

need to win, to gain exclusivity, and I don't

disagree with the idea that you shouldn't need to

win, the point is to get the data. That is why the

BPCA was done that way.

On the other hand, you do want a good

trial, and one way to guarantee that the trial is a

good trial, however the drug comes out, is to be

sure that it is capable of showing something we

need to be true, namely, that Prozac seems to work.

DR. KATZ: Can I just add? One thing you

need to remember about studies done in response to

written requests is that they are very time

sensitive, or at least it's possible that they are

time sensitive.

What I mean by that is you only get

exclusivity if your study reports, your supplements

containing the data come in while you still have

some residual patent life left or exclusivity left.

So, they have to be done within a particular time

frame. In fact, the letters that we send, the

written requests include a date by which the

studies have to be submitted.

So, in some cases, there is at least

potentially motivation to do studies rapidly, so

that they are done and study reports are written,

and the supplement, which includes these data, are

submitted in time, so that they can still get their

exclusivity.

So, one at least potential question that

has been raised is enrollment so rapid or does it

need to be so rapid into these trials that maybe

not all the patients are adequately diagnosed, and

maybe they have something other than depression.

It is very, very difficult for us, if not

impossible for us, to be able to independently

corroborate diagnoses in something, in conditions

like these, so we, of course, take it on faith that

they got the right patients, but maybe, for

example, because of time constraints, they didn't

get the right patients, and that might contribute

to a negative finding even if the drugs were

effective in a true population. So, that is one

possibility.

DR. GOODMAN: I think that is a fair

answer. Anybody that wanted to comment

specifically on that? Dr. Marangell.

DR. MARANGELL: Are you aware, is there a

greater proportion of non-academic sites in these

trials?

DR. MURPHY: I don't know that we have

looked at that. I mean I know that there are

definitely, in some of these studies, very, you

know, academic sites that have been involved in

numerous or actually well-known to us

investigators.

I do want to make one thing again. One

thing that every division within FDA is told, when

writing their written requests, they are asked a

number of questions - what is the public health

benefit, what are the trials to get to that public

health benefit, and you are not to take into

consideration--and most of the time they don't even

know because you would have to go into a lot of

patent law--they don't know or are told to not pay

any attention to when the patents expire or the

exclusivity marking would go out, they must look at

it only from what are the trials that they need to

have done.

Now, what is being told to you, though, is

that--and we have written requests where the

companies come back and say, well, that is not

going to help us, because you put a date on here

that it was due by 2007, and our patent expires in

2005, and we have said, you know, we are sorry, we

need these kind of trials.

Now, would, in that situation, a company

try to compress by getting more sites or, you know,

whatever, would they try to do that trial in a

different way? Yes, possibly.

I mean that is what we are trying to

explain the balance between the way the process is

set up, it is not to be driven by the time when the

patents are expiring, the marketing exclusivity is

expiring, the divisions are to determine what the

studies are that are needed to the best of their

knowledge at that time. They are to design those

studies to answer those questions.

Do we try to be reasonable and say, gee,

we would like a 10-year follow-up study, but we

don't ask that for other--you know, we have to be

reasonable within the realm of what we would

normally ask for, for an approval product.

Again, though, maybe we can be--we say we

have to get this information because children grow

and will go through a period where that might be

effective.

DR. GOODMAN: Thank you, Dianne.

DR. MURPHY: So, you have to ask for

additional information, you might not, for adults.

I am trying to explain the process.

DR. GOODMAN: I will accept some questions

out of order if they are on this specific topic.

Dr. Rudorfer, I think you had one.

DR. RUDORFER: I just wanted to follow up

on Dr. Katz's comment about whether we are looking

at the right patients, which was an issue we

discussed some yesterday.

Just one point that I want to follow up

on. It is clear that in young people who present

with major depression, there is a disproportionate

number who go on to develop bipolar disorder, and I

think one concern that we expressed yesterday was

that the trials are very inconsistent in that

especially in terms of accounting for family

history, it sounded as if in some trials, a subject

could literally be brought to the clinic by a

parent who has bipolar disorder, and yet the child

could be included in the trial.

I realize this question might be, as we

said, a little out of the box. I would think that

if there is any way to encourage the companies to

actually try to find some of these thousands of

young people and see what has happened to them in

the 5 or 10 years since they were in the study, it

could be tremendously informative simply in seeing

whose longitudinal course has played out as what we

recognized in young adults as major depression, who

developed bipolar disorder, who developed some

other disorder, and go back and re-look at, for

instance, those who after the fact are confirmed to

have the diagnosis that we thought they did on

inclusion.

DR. GOODMAN: Dr. Pfeffer.

DR. PFEFFER: Yes, I want to I guess

continue on what Dr. Rudorfer is saying, because I

think diagnosis is critical, and I think we can

learn something about this that we have learned a

little bit about depression in other realms, too,

namely, that children are, in fact, different than

adults.

So, what appears to be adult depression

and what appears to be childhood depression may, in

fact, be quite different, so that perhaps a lack of

efficacy in most of the studies tells us something

about the nature of the developmental course, first

of all.

I agree with what you are saying about the

potential for bipolar. That is one issue that is

crucial, I think, in terms of maybe the adverse

response that children are having, but also if we

think of the number who had some suicidal thinking,

that also might be a subgroup of the children who

are in these studies also.

The other part that I want to mention is

that when I gave that talk last meeting, I talked

about the complexities about what looks like

depression in children, and not only course and

family history, but life event circumstances, and

that has not been looked at.

So, for example, children who might have

been having immediate family turmoil and looked

depressed, that is an issue that might have led to

some resistance in response, for example.

The other point I would like to make is

that we hear from some of our childhood

psychiatrist colleagues yesterday who advocate to

not ban use of these drugs, because they do see

efficacy, and it may very well be that in their

practice, with very careful assessment, careful

diagnosis, they are selecting the subgroup of

youngsters who potentially could respond, and

respond very, very well.

So, I think the question of diagnosis is

crucial, which means that in terms of the study

design, in a way, who has the most reliability to

make a diagnosis, and what kinds of questions

really are being asked and what data is being

collected that might help us even look at

predictability of response, and I don't think we

have that, such as life events, such as family

history, such as perhaps other issues that we would

need to come up with and understand.

DR. GOODMAN: Thank you.

Dr. Gorman.

DR. GORMAN: A lot of us keep saying that

children are different, and I don't think it should

come to us, then, as a surprise that children may

respond differently to medicines than adults do.

I think I would be more concerned about

the efficacy of these trials if they were all

unidirectional in the sense if they had all failed

or they had all succeeded. I have heard nothing

from the FDA to this point that says that the

playing field has been tilted in any way since one

of these drugs in this class, which may not

actually be a class, but it seems like it might be

a class, actually works for children in the bar

that the FDA sets up.

So, I am now going to address my single

question to the rushing hypothesis. After Monday

Night Football last night, I like the rushing

hypothesis. There is one small question I have to

ask.

Prozac was the first mover in this field,

therefore, I assumed it came to market first, and

probably then had the least time before its patent

extension. Is that a safe statement?

So, it came to market first. Did it have

the smallest amount of time? It was the first to

go off patent, yes or no?

DR. TEMPLE: I believe it was the first

one to go off patent by a little bit. It is off

patent now, and only, I don't know, are any of the

others off patent? So, we know it was the first

off patent, which happened sort of a year ago.

DR. GORMAN: So, that would run

counterintuitive to the rushing hypothesis, because

Prozac had to get there first, and therefore, seems

to have had the least time and would be the most

likely to be rushed to get labeling.

DR. TEMPLE: Some of the trials were done

before this program even started, I think, and they

were done a long time ago.

DR. LAUGHREN: One of the trials was done

by Emslie several years before, and the company

obtained the data and submitted those data as part

of that supplement. It was done in the early '90s,

though.

DR. KATZ: Right. The studies that we

asked for in the written requests don't necessarily

have to be done or initiated after the written

request is written.

If they have a study that is very old, but

that meets the criteria that we put into the

written request, they can use that, so they don't

have to be done specifically in response to the

written request, they have to meet the criteria

that we lay out, and it can have been submitted to

us either before the written request.

But they could have done a study many,

many years in advance before we even contemplated

written requests. If they met the criteria, they

can submit it in response.

DR. TEMPLE: But, also, remember it's a

hypothesis. We don't know why those trials fail.

It could be that children really don't respond. I

mean we don't know the actual answer.

DR. GORMAN: Well, I would love to be in

the position where I can say something nice about

the pharmaceutical industry, because it sometimes

seems to happen so rarely, but if Lilly did the

trials before there was the potential for financial

gain, because all they were doing was looking for

labeling in children without the congressionally

mandated reward for that particular behavior, and

therefore had these studies in place, maybe the

rushing hypothesis fails, but there is another

hypothesis that could be generated from that.

DR. LAUGHREN: Again, in fairness, the

Emslie trial was funded by NIMH. This was not

sponsored by Lilly. They went back and obtained

the data, and they did subsequently an independent

trial that also succeeded.

DR. GOODMAN: Dr. Newman.

DR. NEWMAN: I think Dr. Temple did a good

job of explaining why, if you have an active

treatment arm, the trial is likely to be of higher

quality when asked to demonstrate that difference.

I wonder if another approach to motivating

high quality studies would be to require that in

order to get the exclusivity, that the trial be

written up in a way that passes some sort of peer

review and be published.

That way, even published on FDA web site,

that way, if the trial is sloppy and finds the drug

doesn't work, those results would not be buried,

they would become public and that I think would

provide some motivation to do a good job.

I am a little troubled. I wrote down a

quote from Dr. Murphy. It said, "If a study is

negative, we don't talk about it." I think if

that's the case, then, there is a lot less

motivation to do a really good job on the study.

Why not require the studies be published, be

written in a way that it is of sufficient quality

that they can be interpreted, and then maybe the

quality would improve.

DR. MURPHY: But for peds now, we do.

That is the difference. That statement was for

adults. For pediatric studies, well, I should say

it is for pediatric studies that aren't done under

exclusivity, but for pediatric studies done in

response to these written requests, we now are

mandated to make them public whether they are

approved, they are not approved, or even if they

are withdrawn.

DR. TEMPLE: But you are also talking

about a level of detail in the presentation

sufficient for people to really get into was it a

high quality study, and things like that.

DR. NEWMAN: Why not?

DR. TEMPLE: it is a fairly good question.

We don't believe we have authority to insist that

things be published. We get full details, we get

all the data.

DR. NEWMAN: But you could peer review,

you could peer review them. You could send them

out and say is this something that is publishable,

and have people at FDA, who I am sure are very good

at that, say no, this gets an F, you know, this is

not good enough, send it back, or you don't get the

exclusivity.

DR. TEMPLE: Well, our reviews, when we

approve something, you get on our web site the

contents of our reviews, so you get to see what we

thought of all of the studies. If we do not

approve, however, we don't believe we have

authority to make those data public, so you don't

get to see our reviews. That is our legal

interpretation of what confidential commercial

information is, and I can't rebut it or comment on

it. It's a legal determination.

DR. GOODMAN: Dr. McGough.

DR. McGOUGH: Just on that point, does the

FDA now have the authority, if you wanted to, to

put negative studies in the pediatric label, do you

have the authority or does Congress have to do

something for you to get the authority?

DR. TEMPLE: We have authority. What I

was saying before is--and we are, as I said,

considering whether in the pediatric case, we

should do that. In other cases, we would, too, if

we thought it was important to point out the

negativity, and the negativity was a true bill.

It is just the fact that if one study

fails, doesn't necessarily say that something

doesn't work, so we have been somewhat reluctant to

just do that until it was convincing.

But as Dianne said, we are actively

thinking about amending that policy for the

pediatric setting where the whole point of getting

the studies done was to see how the drugs worked in

children, for the very same things that they have

been studied for in adults.

It is a little different from novel use or

something like that. The BCPA calls for studies of

the exact same things that have been studied in

100

adults.

DR. MURPHY: And we are putting negative

information in some of the labels already for other

types of products, but because of the complexities

that you have heard, it has been the policy for

antidepressants for children not to do that at this

point, but I think, as has been mentioned, it is

being reconsidered.

So, we have, and I have got all the labels

here that we have done, we are putting that

information in some of these labels.

DR. POLLOCK: For the new approvals.

DR. MURPHY: No.

DR. TEMPLE: For where we grant

exclusivity.

DR. MURPHY: Right. In other words, if a

product comes in and doesn't work, we have put that

information in some labels where we think it is

very clear-cut, you know, eight more studies is not

going to change this for whatever reason, and we

put that in here.

We have also put in information where it

101

hasn't worked where there are safety issues

involved, and we are not clear what those safety

issues are, but we want to tell you about them.

So, those are in the label, too, even when it

wasn't approved for that indication.

DR. GOODMAN: Dr. Santana.

DR. SANTANA: I have a comment and then a

question that really relates to a point that Dr.

Chesney made about issues regarding the boundary of

practice and FDA regulation.

My comment is that there was some comment

related to oncology and issues, how we deal with

some prescription and safety issues in oncology,

and I think we have to recognize that pediatric

oncology is unique in this country and that most of

the trials, even the exclusivity trials, of which I

have participated in some in oncology, are really

under the umbrella of research centers and academic

centers. Very little pediatric oncology is done in

the private practice.

So, by force, you are now dealing with a

group of individuals that are more specific and

102

more geared up to looking at issues that

potentially could be relevant, whereas, I think in

the other pediatric arenas that we are talking

about, that doesn't occur.

So, I think it would be a misnomer to use

pediatric oncology, maybe it should be the gold

standard, but I think we need to recognize that it

is kind of unique in the way it is practiced in

this country.

Having participated in some of the

exclusivity oncology trials, I can tell you that

they are at the same caliber and at the same

rigorous structure as any of our other oncology

trials are in the cooperative group setting, et

cetera, et cetera. So, that was just a comment to

clarify the pediatric oncology issue.

I want to get back to patients and

practicing physicians, because we have been talking

a lot about study design and how to analyze data.

I want to get back to the issue of patients,

parents, and practicing physicians, and this

boundary of what the FDA can regulate and cannot

103

regulate in terms of the practice of medicine.

I was struck yesterday by many of the

testimonies from parents and families, and

actually, there was even a gentleman who showed a

slide, in which his child was given the medication

as a free sample. I am not sure that all these

whistles and alarms that we put in labels are

really going to work unless somehow that practice

stops for certain medications that we think

potentially have a greater risk.

I wanted the FDA to address the issue

historically. Is there any ruling that you guys

can impose or potentially think about, about how

these medications are given without prescriptions,

that is, either free samples or in the marketing

world, so you could comment on that.

Secondly, does the FDA have any historical

data on successful programs? There was a mention

that maybe a med guide to parents and families

would be a way to address some of the safety issues

and bring people to a better level of

understanding.

104

Can the FDA comment on any successful

programs that they can point where this has truly

worked?

DR. TEMPLE: Just on the free sample

thing, my understanding, but I don't really know,

was that a physician did use a free sample to, you

know, like start the child out. That is not

without a prescription, it may not have been well

done and may not have had adequate follow up, but I

am not sure that it is the free sample that is

involved, it is the lack of follow up that was

described that seems like the problem.

It is not easy to know how successful our

various endeavors have been, and Anne Trontell may

want to comment on that. Some of them have effects

that are not entirely what we wanted. She

mentioned the program on dofetilide.

To start, dofetilide is a drug that is

used to prevent recurrence of atrial fibrillation,

but it is a drug that causes QT prolongation and

Torsade de Pointes, and there is no doubt about it.

The recommendation in labeling, and it is

105

enforced by the need to give out various

information requires that you come into a hospital

or outpatient facility for a couple of days to see

what your creatinine clearance is and to see

whether you are a person who has QT prolongation to

an excessive degree.

Then, after that you can go out and be

treated with it for long-term use in preventing

atrial fibrillation.

What appears to have occurred is that that

is sufficiently difficult, so that people instead

use sotalol, which doesn't have such a program, or

quinidine, neither of which are an improvement of

the situation.

So, people can avoid some of these things

if they are inconsistent across the drug classes,

so you always worry about that.

There is a very rigorous requirement for

periodic measurement of liver tests with a drug

called Bosentan, which is used for pulmonary

hypertension, and although the drug was quite toxic

when it was being developed, my most recent

106

information is that we haven't had any fatal liver

outcomes, perhaps a testimony to the fact that

people are indeed following up these patients.

Of course, this is a class of patients who

are very sick and very closely watched. You don't

know if that is typical how we are going to do.

Anne, you want to comment on some of the

other programs.

DR. TRONTELL: Sure. On the issue of

sampling, first of all, I think in some instances,

at the time of product approval, there have been

informal agreements, but no FDA authority to

restrict sampling exists to my knowledge, but there

may be agreement, you know, certainly, we don't

sample oncology drugs. There are things that just

don't happen.

On the issue of what is a successful

program, I think we struggle in the agency, because

good evaluations have not been done on a standard

basis. We have the best information for those

programs that are most restrictive, programs like

clozapine or programs like the one for thalidomide

107

to prevent pregnancy exposures.

So, the available data to us to tell us

what works tends to be only in those extreme cases

where we have put, as Dr. Temple just described,

for Bosentan, you know, very severe restrictions.

If you are asking for specific information

about medication guides, I think we have in the

general literature evidence to suggest that good

education certainly facilitates good behaviors, but

I don't think we have any evidence yet that it

guarantees that they do take place.

So, if you had questions about the

intermediate ones, I think for the most part, we

don't have information about those specific

educational programs or the reminder ones. Not

uncommonly, education is applied in the context of

these very restrictive programs that I just

described, and again, teasing out what the

education alone does is very difficult.

DR. GOODMAN: Dr. Katz.

DR. KATZ: You are hearing the

difficulties that we think we have with regard to

108

imposing various sorts of restrictions although

again, there are ways to do it although they may be

very difficult to implement.

But it occurs to me that it might be

particular difficult in this case because the use

we are contemplating in all but one case is off

label, and I don't even know what the implications

of that are. Certainly, there are legal questions

about what you can say and what you can restrict

with regard to off-label use, and I don't think

that we have thought through all the implications

of that.

DR. SANTANA: So, as a follow up to that,

since we last met in February and there was a

recommendation to do something with the label that

occurred and some warnings, has the Agency

monitored the change in practice?

I heard a number yesterday of 10 percent.

That is prescriptions, but has that been rigorously

looked at, that there was an impact of that

modification that translated to something very

tangible?

109

DR. TRONTELL: I will ask either Michael

Evans or Judy Stafford from the Office of Drug

Safety. All we have really been able to monitor

since the last advisory committee is volume of use,

but they do have some information on how that has

changed recently.

DR. GOODMAN: Ms. Griffith.

MS. GRIFFITH: I would just like to pursue

this for a moment with respect to the patient and

physician relationship. When Dr. Chesney was

proposing perhaps some sort of a computerized

programming or education, or even with respect to

these med guides, when Dr. Temple suggested that

perhaps there would be, you know, a mechanism much

like you have for other drugs, that the patient and

practitioner would be signing a consent form

outlining the risks and benefits, I want to

understand the reason that you thought that that

might be too great a deterrent to pursue, simply

because from my perspective as patient rep and

parent, it seems to me that in the course of any

treatment process for any severe illness, which as

110

we all understand depression is, you are often

asked to look at the risks and to sign some

statement to the effect that you understand what

these risks are.

You even have to do that if you get a shot

of botox, not that I know, but it just strikes me

you have put the parents now in the position of

actually doing the risk-benefit analysis. That is

where we all are.

If by providing the families with the

statement that these risks are indeed serious, I

think that what we heard yesterday was how little

awareness there was on the part of the parents that

these drugs could be lethal in certain cases.

I am arguing for more information rather

than less, not more restrictions, and I agree with

the point that Dr. Santana made, that how often

does a parent either open the box and read the

information or understand it.

So, if it is very clearly stated between

the doctor and the patient or the parent, I think

it goes a long way to satisfying the need to know

111

for parents.

DR. TEMPLE: There are gradations of

information, and we wrestle with how to do that

without being an attempt to be informative, but not

disruptive, so that, for example, a lot of drugs

have what are called med guides. These are patient

labelings that are actually, under the law,

supposed to be given out by the pharmacist.

My own view is that if you don't make it

part of the unit of use package, you might as well

not bother, but in any case, we know that there are

ways to get that information to patients either

through the proper functioning of the pharmacy or

by making it part of the package.

An enormous additional step, which has

been done in some cases, but, you know, thalidomide

is a level of risk that is sort of in its own

category, where there is a requirement that the

patient and physician discuss all these matters.

That is a very huge step, and what you might think

is reasonable for thalidomide, something that is

used infrequently, you might find more disruptive

112

than you want or more difficult than you want for

drugs that are much more widely used.

As Russ pointed out, it is particularly

tricky when the recommended use isn't even in the

label. How to write a med guide or something like

that for something you are not really recommending

and don't feel able to recommend yet, that may

sound like a bureaucratic worry, but I think it's a

serious worry.

You don't want to warn people and

simultaneously recommend a use that you don't think

is recommendable, and any discussion like that is

tantamount to recommending the use. So, we will

need to worry all of those things based on your

conversation.

MS. GRIFFITH: Could I follow up? I

understand that and I understand that there are all

sort of issues involved, liability on the part of

the physician, but I am suggesting, from the naive

perspective of the parent, I think of depression as

every bit as serious as the use of thalidomide

posing birth defect risks or, as Dr. Santana said,

113

you know, the cancer example. Parents need to be

informed about those risks.

I don't think that this is any different,

frankly, and if it is an extraordinary measure to

take, I think that it benefits both the

practitioner and the patient parent.

DR. TEMPLE: For oncologic drugs, the

label says you should be a properly trained

oncologist. There isn't anything in there that

says what you need to discuss. Patient med guides

for oncologic drugs is by far the exception, I

think because it is assumed that there always has

to be such a conversation in the course of therapy.

I guess what is being discussed is whether

there is a common practice of having that kind of

conversation in someone who is depressed, and

obviously, we all think that there should be such a

conversation. The question is now to induce it and

what to provide people to help them be sure they

ask the right questions.

DR. GOODMAN: We have a representative

from the Office of Drug Safety at the microphone.

114

Could you please state your name?

MR. EVANS: Michael Evans with the Office

of Drug Safety.

With regards to drug use, comparing the

first six months of this year to the first six

months of last year, the market rose with all ages

about 7 percent. Adolescents and children, in the

first six months of the year, still comprised

between 7 and 8 percent of that total. So, they

are still widely used in children.

DR. GOODMAN: How up to date is that data?

There must be some sort of lag time, isn't there,

between when the prescription--

MR. EVANS: It is according to IMS Health,

and this is January through June is what we looked

at. This is outpatient prescription data, which

comprises about 45 percent of all pharmacies in the

country.

DR. GOODMAN: Let me make sure I

understand. You don't see any significant drop?

MR. EVANS: No. I believe a woman

mentioned yesterday that they saw a 10 percent

115

decline. We did not see that.

DR. GOODMAN: Dr. Irwin.

DR. IRWIN: Has the rate of increase

remained the same or has it leveled off, or what is

the direction?

MR. EVANS: In February, one of our

colleagues, who gave drug use for 2002, that age

group was still 7 to 8 percent of the total. It is

still the same this year, first six months.

DR. GOODMAN: So, there has been no great

change.

I will take again other questions out of

order as long as they are directed to a

representative from ODS.

Dr. Marangell.

DR. MARANGELL: Actually, a comment

directed to this. I think it is really critical to

this discussion that we keep in mind that our goal

is to protect risk, but also that this is really a

devastating illness, and I am not sure that I

necessarily--I don't want to necessarily see

prescriptions drop. These people need to be

116

treated.

What we want to do is make sure that

people are educated of what to look for early on in

terms of risk for those people that are at risk,

children or otherwise. They are not necessarily

the same thing .

DR. GOODMAN: Other questions for our

speaker? Dr. Gorman.

DR. GORMAN: Is the data on the level of

the class or is it on the level of individual

drugs?

MR. EVANS: We looked at the class as a

whole and then we looked at each individual drug in

the class. That was according to IMS Health

National Prescription Audit, and we also looked at

the National Disease and Therapeutic Index from IMS

Health, and tried to apply those percentages to

outpatient projected prescriptions.

Only the players changed in that age group

of children 1 through 17. Paroxetine was knocked

out of the top five, but sertraline still is the

market leader, followed by fluoxetine.

117

DR. GORMAN: So, the information, there

seems to at this point only be one drug that is

efficacious in this age range, moved it up the

ladder, but didn't make it number one?

MR. EVANS: Not necessarily. This is what

we observed when we were just looking at drug use

in prescriptions outpatient, and the National

Disease and Therapeutic Index is an office-based

survey where a drug is mentioned during that survey

and linked to a diagnosis.

DR. GOODMAN: Can you differentiate

between the prescriber classes, whether it is a

primary care physician versus psychiatrist?

MR. EVANS: We did look at specialty in

MBA-Plus. Psychiatry was still about 65 percent of

the specialty, pediatrics, somewhere between 15, 20

percent still.

DR. GOODMAN: Dr. Fant.

DR. FANT: Could you comment on the

indications for the prescription, was it all

depression or other off-label--

MR. EVANS: We looked at mood disorders,

118

anxiety disorders, ADD, and other disorders. In

age group 12 to 17, it still appears there is not

really any change. It is still mood disorders,

which includes major depression, is still

two-thirds of the indications.

It looks like in the 1 to 11-year age

group, perhaps more shift to the ADD.

DR. GOODMAN: Dr. Pollock.

DR. POLLOCK: I heard rather consistently

yesterday a lot of concern about the

direct-to-consumer advertising and the role that

that has played in this, and it may not be the

purview of this committee, but I am asking if we

can address this aspect and how that plays with the

implications of labeling, that if we do put, as was

suggested, a specific negative label in terms of

the indications and certainly as a warning, let

alone a black box warning, that the amplitude of

these warnings are heard.

I mean it is almost a penalty then for the

intense direct-to-consumer advertising, which does

play, as I understand it, a huge role in driving

119

sales for some of these antidepressants.

I just wondered if you could give us some

indication, I mean is there a direct policy with

D.D. Mack how these various gradations get

translated into the few seconds that go on the tail

end of a commercial.

DR. TEMPLE: They are certainly supposed

to. The presence of a strong warning or box

warning should be reflected right in the major

statements that are made. The direct-to-consumer

comes in two flavors, written and TV.

In TV, you can't give as much information

easily, but it has to be available readily. You

can argue about whether people make use of that

availability. But the major statement would have

to reflect the balance between those two things.

You know, I am sure people have views

about whether that is done successfully or not. If

it is written, then, the written statements have to

show that balance. Any box warning has to be

reflected in it.

So, yes, it is supposed to reflect the

120

balance of information that is in the labeling, so

if the labeling changes, the direct-to-consumer

advertising should change.

DR. GOODMAN: Dr. Perrin.

DR. PERRIN: Yes. Back to the ODS person.

A number of the anecdotes yesterday suggested that

people were put on antidepressants quite off label

and probably not for major depression, but rather

for minor depression and acute depression.

You said that you have the evidence on

mood disorders that includes major depression. Can

that be disaggregated at all into other non-MDD

forms of mood disorders?

MR. EVANS: We could look at that. We

didn't, we lumped them together just for a top-line

statement at this time, because we wanted the focus

to be more on suicidality than drug use.

DR. GOODMAN: Ms. Griffith.

MS. GRIFFITH: I wanted to know, since the

data you got was January to June, and the Advisory

didn't come out until late March, is it possible to

look at the data that you got April to June to see

121

if there is a decline?

MR. EVANS: We did look at it monthly in

those months, and there was not any decline. I

mean it wasn't a change.

DR. GOODMAN: Are these new prescriptions?

MR. EVANS: These were total

prescriptions, new and refill.

DR. GOODMAN: Did you separate out by new

prescriptions in terms of the monthly rate?

MR. EVANS: We can, and we did, and we did

not see much of a decline in those, as well.

DR. GOODMAN: Dr. Chesney.

DR. CHESNEY: On the surface, this looks

not bad in the sense that 65 percent are being

written by psychiatrists, but although I am not

here as a patient representative, I do have a

daughter who has been on these medications, and I

know for a fact that most often psychiatrists do

not prescribe these medications.

My image is that at the end of the day,

they take a whole packet of prescriptions--and I

will be interested to have the psychiatrists

122

respond to this--a whole packet of prescriptions

that have been written by social workers,

pharmacists, psychologists, and sign their name.

So, I think when we are talking about

educating primary care providers, looking at this,

I am reassured, but I know that this does not

represent who is actually writing the

prescriptions.

MR. EVANS: Yes, this is a limitation of

the data, too, the data is only as good as what the

pharmacist inputs at the computer, and, you know,

if that specialty is on there, hopefully, they will

put that on there.

DR. CHESNEY: I am sure they don't, but I

think this is very important in the educational

issue, because the people who are prescribing this,

on the whole, are not child psychiatrists, and they

are family practitioners, they are ER physicians,

they are nurse practitioners, they are pharmacists.

I mean I was appalled at what happened

when we visited one of these pharmacists, but no

disrespect to pharmacists, but this is very much

123

happening out there.

DR. GOODMAN: One last question for ODS

representative from Dr. Wang.

DR. WANG: I was curious, has anyone

studied what happened after the British

contraindicated these in children, just to get a

sense of what the impact of a labeling change, such

as that, might be?

MR. EVANS: In February, they looked

through 2002, the market between 2002 and 2003

group, 15 percent with no change really in the

adolescent and children population. It was still

around 78 percent, so I don't think there was a

change much in this country.

DR. WANG: But you don't know of any data

in the British--

MR. EVANS: We didn't look at that.

DR. GOODMAN: Thank you very much. You

may step down.

We have six more presenters. I am not

taking any more, that's it.

Dr. Leslie, do you remember your question?

124

DR. LESLIE: My question goes way back and

is for the FDA. I think reading through the

materials that we received from the public, two of

the major concerns about the data that was coming

in were suicidality, et cetera, being captured

under other labels, such as emotional ability, and

then also the issue about dropouts were people

dropping out of either the placebo or the drug

groups, that were having the kinds of adverse

events that were of interest to us, and then not

being counted in the data.

So, I had two questions. One was do you

feel confident that the data you have received has

addressed those two issues for the analyses that we

looked at yesterday, and the second was what steps

could you potentially take to address those

drawbacks that were raised by the public in the

written requests that proceed from here on out.

DR. LAUGHREN: I can respond to the first

part of that. In terms of the data that we

received, if you recall, we issued letters to

companies in July of last year, which specified a

125

very clear research strategy for looking for

adverse events that might be related to

suicidality.

It included both preferred terms and

verbatim terms. All these data are electronic, so

it was a string search to look for events that

might be possibly related to suicidality. In

addition to that, we asked companies to look at all

their serious adverse event narratives, any event

that had been classified as a serious adverse

event, they would have to look at and make a

decision whether or not that might represent

suicidality.

Then, later in the year, we issued

additional requests to basically ask them to give

us all the serious adverse event narratives, so

that we could have Columbia themselves look at

those data, and also, all accidental injuries and

accidental overdoses to try and broaden the search,

to make sure that we could capture everything that

might be related.

Now, it is true, despite all of that, it

126

is possible that certain events that didn't rise to

the level of being a series adverse event might

have been captured under some other either verbatim

term or preferred term.

The other question is whether or not the

narratives that we received fully reflected the

case report forms. The narratives are created by

the companies. To try and address that, we have

sort of a second level of this contract with

Columbia that is ongoing right now.

We have done a 20 percent sampling of the

case report forms for the narratives we have, to

have them check the narratives against the case

report forms basically to see whether or not they

fully reflected, the narratives fully reflected

what was in the case report form.

In addition to that, we have asked for the

dictionaries. The dictionaries, basically,

companies, when the code data, they subsume them

under preferred terms, and once they do that, that

creates basically, a dictionary.

So, we have asked for the dictionaries

127

from all these sponsors. Columbia is currently

looking at those dictionaries to see if there are

any other additional adverse event terms that might

be of interest to look at, again to answer that

question whether or not all relevant events have

been captured.

That is a very tedious, time-consuming

process, but it is ongoing right now.

Dropouts, Dr. Hammad addressed that

yesterday. We did look at dropouts, and as he

suggested, it is true, many patients were dropping

out for these events. In a sense, it was almost a

surrogate for that endpoint.

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

comment?

DR. POSNER: I just wanted to say that, in

addition, because we asked for all of the

accidental injuries, and that would be the most

likely place, that all of these events involved

some type of injury or another, that you would find

events that were missed, so we can feel reasonably

confident that this body of data represents

128

everything we would want to look for, but we are

doing these additional steps, as well.

DR. GOODMAN: Dr. O'Fallon.

DR. O'FALLON: Yes, this sort of follows

up. We are back to what I am concerned about, the

people who came here, they are worried about the

side effects, the toxicity here. Right from the

beginning, when you told us back in February about

this study, I was worried about what wasn't

recorded in the drug companies' records, for

whatever reason.

I think that is still, no matter what we

do going forward, we have got to address the issues

there. So, what I am wondering about, I would like

to propose, and you shoot at and tell me that these

things are not feasible, but it seems to me what we

really need are somehow a standardized suicide

monitoring procedure or whatever for future studies

in mental illness, any kind of a drug that is

targeted toward the mind, we should be looking for

this type of thing, the suicidality side effect.

Then, I think we are going to have to have

129

some sort of standardized suicide coding. They

have done it in adverse events, not great, but it

could be done better for suicide coding because of

the work that has been done here.

I think this has been a wonderful outcome

in terms of the coding issues.

Now, here is one that is going to kill

everybody, but you can make suicidality a goal, a

primary goal in, say, mental illness or maybe

depression more specifically, where you really

think that this side effect or toxicity, this

adverse event is also a symptom of the disease.

In cancer, they have had to struggle with

the issue of distinguishing side effects from

symptoms of the disease for decades, some way of

going after that.

Just one more comment. This is a comment.

I believe that there was a 40 percent--in the stuff

I saw before I came out here--I think I saw 40

percent placebo effect in TADS, the TADS study.

If that is true, this is a major issue.

That is one of the reasons why we really do need to

130

have placebo arm in these different studies,

because if 40 percent of this population will have

a beneficial effect due to sugar pills, to try to

tease out true effectiveness of these medications

given their severe side effects, it is very

important that we have a placebo arm even going

forward.

I know that you are not thinking of it,

but I think some of the people in the room are

wondering why we have to go with sugar pills.

DR. LAUGHREN: Let me comment on the last

point first. We clearly agree with you about

placebo, but it is not just the act of giving a

pill. In all of these trials, there is a lot that

happens that results in improvement.

DR. O'FALLON: Yes, I know that.

DR. LAUGHREN: There is a lot of

attention, the patients have a lot of interaction

with staff. That is really the placebo effect.

But the other point you are making, I

completely agree that ascertainment is ultimately

the issue. If you don't collect the information,

131

it doesn't make any difference how carefully you

search the database, if it wasn't collected, it's

not there, so ascertainment is key.

Again, that is one of the things that we

hope to get out of this effort with Columbia is a

better guidance document for future trials to make

sure that suicidality is properly ascertained, but

it is an evolving thing in the field. I mean there

is not at this point in time an optimal way of

doing that, so we hope to get an instrument that we

can apply for future trials.

Again, I agree with you that coding of

data needs to be standardized, and again that is

one of the things that we expect to come out of

this.

DR. POSNER: Could I just add to that? We

are very committed to addressing the question that

you are referring to.

DR. GOODMAN: Would you bring the

microphone closer.

DR. POSNER: I said we are very committed

to addressing this issue in terms of suicidality

132

adverse event monitoring, and Dr. Laughren

mentioned guidelines that we are going to write and

measures that we actually have developed that we

can implement in all trials, that will help us

collect the right data and then be able to use

these consistent definitions to classify events, so

we can make sense across all of these trials.

What is important to note is that we are

working on a National Institute of Mental Health

study called TASA, Treatment of Adolescent Suicide

Attempters, which is very focus on this issue of

adverse event monitoring, and it is wonderful

because it is helping us inform the process, so we

have developed very, very rigorous standards of how

we ask these questions, what measures we use, and

how to do it consistently, and that will help

inform the guidelines and the measures that

industry and everybody else can use.

So, we have made a lot of progress in

that, I think.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: I just want to follow up on

133

the last discussion. It seems to me there are two,

somewhat separate things, and I would be interested

in people's views.

One is to make the periodic routine

question better than it is. There is a suicide

item on the score, but that didn't show anything.

Maybe that will never show anything because when it

happens, it happens abruptly and you don't happen

to pick it up at the two-week period, but it does

seem as if a better questionnaire on that question,

done routinely, might be useful.

The second part of it is how to

characterize events, what questions to ask about

those, what to write down. Is that what you are

thinking, they are two somewhat different things?

DR. LAUGHREN: I agree, they are two

separate things. There is the suicide item that is

part of every one of these instruments and sort of

standardizing how that routine information is

elicited, but then when an event occurs, you have

to ensure that the appropriate questions are asked

to flesh out that situation, so that someone down

134

the road who is looking at the data is able to make

sense of it.

DR. O'FALLON: But I would like to point

out that the monitoring procedures, especially for,

say, the first two weeks or something like this,

should include a real collaboration with the

children and their caregivers, their parents,

whoever they are living with, to be on the watch

for those and to report them immediately, and that

those things would be part of the data.

DR. LAUGHREN: Let me just respond to that

quickly. Basically, you are switching gears to a

clinical setting, I think, other than a clinical

trial.

DR. O'FALLON: You guys can write the

regulations for the clinical trial, right?

DR. LAUGHREN: Right, but obviously, the

points that you are making apply to clinical

practice, as well.

DR. O'FALLON: Yes, but they would apply I

would say in the clinical trial, because I think

that you are not possibly getting all your

135

information. If you don't come in for two more

weeks, people forget that they were scared 10 days

ago.

DR. LAUGHREN: Absolutely, I agree.

DR. POSNER: I just wanted to add that we

are also working with the CDC just on this

question, what are the right one or two questions

that need to be asked in any trial or clinical

setting to get this information to be able to

classify it appropriately, which is exactly what we

are talking about, and it is not necessarily the

best questions on the measures that were used in

this trials, but that is exactly the pertinent

point in clinical setting or in research settings

with the increased monitoring that you are

referring to.

DR. GOODMAN: I am going to need to wrap

up the remaining questions in the next five

minutes.

Dr. Irwin.

DR. IRWIN: The question I wanted to ask

specifically was related to the one that is on the

136

table right now. Yesterday, we heard from several

families and individuals about really homicidal

behavior and more violent behavior outwardly

directed, not internally directed.

Of concern to me is that the focus has

been so much on suicide, but what I wanted to know

is what kind of monitoring or what kind of tools

are in place to really measure that phenomenon in

these trials, because it seems to me that we don't

have any data that has been shown to us at least on

adverse experience or events with the clinical

trials.

DR. LAUGHREN: Our focus here clearly has

been on suicidality, and not on hostility and

violence. There was a lot less of that in these

trials than we had suicidality, and we have not

come to grips with that yet, but it is a whole

other area that needs to be fleshed out and

developed in the same way that suicidality has been

fleshed out because again we have included in our

database information on whether or not these

individual patients at some time during the course

137

of treatment were coded as having hostility or

agitation as a preferred term.

If you go back and look at what got

subsumed under that, I am sure it is going to be

quite different depending on different sponsors,

and it really requires a parallel development to

try and understand what that means.

Again, all the things we have been talking

about for suicidality apply to that domain, you

know, how do you ascertain for it, what kind of

questions do you ask to flesh it out, it is a real

problem.

DR. GOODMAN: By the way, the placebo

response rate from the TADS trial is 35 percent.

Looking at the paper again, I see 35 percent.

DR. O'FALLON: Okay.

DR. NEWMAN: Just to clarify, that is much

or very much improved in the TADS trial.

DR. GOODMAN: I don't think that one is

based on the CDRS, right, that is what you are

saying, it is based on the CGI?

DR. NEWMAN: The dichotomous outcome was

138

were they much or very much improved, so if you

said just improved, reasonably, it would have been

a lot more.

DR. GOODMAN: But that is the standard, I

mean it has to be much or very much improved to be

a responder. That is pretty much across all

clinical trials.

Dr. Pine, please.

DR. PINE: I want to return to a point

that was raised by Dr. Goodman and just call the

committee's attention to a couple things that he

said and then also raised a couple other issue

relevant to the discussion about 10 or 15 minutes

ago with FDA.

That is the issue of both how perplexing,

but also how important it is to think very

carefully about the efficacy data and the

difference between the data for fluoxetine, on the

one hand, and all the other antidepressants, on the

other hand, and how do we understand that.

Number one, just to underline that I agree

that the importance of that point cannot be

139

overstated. I guess there is a couple of issues

that were not discussed 10 or so minutes ago in

sufficient detail, and really two points to raise.

One is that I do appreciate from a

regulatory standpoint that it is very difficult to

specify exactly how one is to do an appropriate

study. We talked about a lot of the details that

we don't need to go over again except one thing was

not discussed, and that was a discussion of the

level of rigor that goes into both the training of

the investigators who are going to ascertain the

samples and document the diagnosis, on the one

hand, but then also follow the response of the

patients throughout the trial.

Then, I think the last thing to say about

that specific point is that when we look at the

data that have been published, and probably the

most extensive data are from the sertraline trial

as opposed to the TADS trial, with the sertraline

trial being a pharmaceutical-sponsored study that

appeared in JAMA, and the TADS trial being an

NIH-sponsored trial that was also published in JAMA.

140

There are some fairly clear signs that the

manner in which investigators were trained, that

the criteria for enrolling patients for the process

of evaluating the response as it was manifest

throughout the trial was quite different in those

two studies.

Again, when we come to the issue of how

important it is to compare the data for fluoxetine

and the data for the other agents, I think we need

to acknowledge that there are already signs in the

data that have been published in the reports that

have appeared in peer review, that the quality of

the studies appears to be different.

I think it is also important to note that

if we were to look at the efficacy data by industry

sponsor versus federally funded, there have been,

to the best that I can recall right off the top of

my head, two federally funded SSRI trials, both are

positive.

So, we are 2 out of 2 on that score,

whereas, if we look at all the others, we are

basically 1 out of 13 or 1 out of 14.

141

DR. GOODMAN: Those two are both in

fluoxetine, isn't that correct?

DR. PINE: That's true, so, you know, we

have a confound between federally sponsored and the

compound, but those are the data that we do have,

and I think given the issues that I just discussed,

you know, we are going to be very hard pressed to

say this is a funding or design feature issue,

which it might be, or that this is a medication

issue, which again it might be.

DR. GOODMAN: I want to make sure I am

clear on the source of the fluoxetine data for the

clinical trials. I think it was mentioned before

that some of the data that contributed, I don't

know which of the positive studies, but one of the

positive studies at least, was actually a study

that had been conducted by Dr. Emslie, and that, in

fact, was a federally funded study, is that

correct?

DR. LAUGHREN: Yes, the Emslie study was

funded by NIMH, but there was another independent

trial that was funded by Lilly, that was also

142

positive. The TADS trial was not part of our

decisionmaking, that came later. So, there was

another positive fluoxetine study that was done

entirely under Lilly's sponsorship.

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

comment?

DR. RUDORFER: Just on this last point, I

think that was back in '80s, am I correct, that the

Emslie study was first done?

DR. GOODMAN: No, '97.

DR. RUDORFER: Because we have in our

material, a fluoxetine clinical trial that goes

back to the '80s.

DR. PINE: That was a very small study, at

least as I understand it from reviewing the

material. Maybe you want to comment on the 1980

study.

DR. LAUGHREN: I am not familiar with that

one. I think the Emslie study was maybe early

'90s, but we can probably check. I am sure it is

in Dr. Dubitsky's review.

DR. PINE: Well, in the material that we

143

received, there was a study that was described,

that I recognized just from the description as a

study that was published in 1990 by Simian as the

first author.

DR. LAUGHREN: That was probably, I think

it was HCCJ. That was the study that was

terminated early. I think there were only 40

patients in that trial. That was not one of the

trials that was the basis of our approval of the

claim in fluoxetine.

DR. RUDORFER: Could I insert just another

quick safety-related question? We were talking

before about if there is a way to judge the impact

of the label change that was made in March, since

that would seem important to us in terms of whether

an additional change would be helpful.

I am wondering if the Med Watch Program

offers any clues there in terms of reports from the

clinical community of adverse events, whether there

has been a change in 2004 versus last year, for

instance.

DR. TRONTELL: We haven't looked at that.

144

The Med Watch Program, just to be very precise, has

two components. Med Watch itself involves reports

that come directly to the FDA, and don't come

through the pharmaceutical manufacturers. That is

actually a minority of the reports, 5 or 6 percent.

The Adverse Event Reporting System, which

Med Watch and the manufacturers feed into, is much

larger. The challenge was reports that came to the

agency spontaneously, that you actually can, in

fact, see a paradoxical increase in reports when

these events become known to the public, so it is

not a reliable way to tell you whether or not

things are changing, because we know not every

report comes to us and the factors that influence

reporting are changing dependent on scientific and

media attention.

DR. RUDORFER: So, within the agency, do

you see any index that you can use to judge the

impact of the label change in March?

DR. TEMPLE: The only thing that you could

measure properly is use, so if the warning sort of

made people think twice, and decided to watch and

145

wait instead of treat, you could detect that

through the data that have been described.

But the adverse events are unpredictable.

We know that not all serious events are reported,

you know, various estimates go from 1 percent to 10

percent to higher, but as Anne said, if you change

public attention to something, you can get

increased reports without having increased numbers.

It is very hard to know that, to know

about those things. What you can think about is

looking at databases that have reports of these

things, the kind of thing that Jick did and others,

but the events in question, first of all, you are

not sure how well they are described, you are not

sure whether they get into the reporting system.

It is very difficult territory. You know,

epidemiology is difficult enough, this is

particularly difficult, because you don't know how

they are classified and it is really hard.

DR. TRONTELL: Just to follow up, if you

want to see a change, you may actually have to wait

some time before you see a change in the outcome.

146

If you wanted simply to see if prescribing

practices are different, I think you might have to

go even beyond the use data we have, even beyond,

say, new prescription to see if starts on these

products are changed, to actually do some active

surveillance and survey clinicians or survey

patients to find out if, in fact, there is a

different process for introducing these products.

DR. GOODMAN: Dr. Marangell, did you have

a question? Okay.

Dr. Fant.

DR. FANT: I just want to follow up on

that and the comments and questions that were

raised by Dr. Chesney earlier.

One of the things that I was struck by

yesterday and hearing the testimony of the

families, and from my own personal experiences with

friends and family members, much of the discussion

here has been focused on the use and efficacy and

outcomes of these drugs related to major depression

in the trials that have looked at that, but the

off-label use of these medications is fairly

147

promiscuous, and the prescribers extend well beyond

those that are trained in the care of the mentally

ill.

I think that is a real problem when you

have ob-gyns prescribing it and giving it away for

ladies who just may be a little moody, you know,

when they come in, or feeling a little down, and

without having any consultation or evaluation by

someone who is specifically trained to evaluate

that, I think that is a problem and I don't think

that represents an isolated incident.

I think any labeling change considerations

really need to not necessarily be directed to

regulate how medicine is practiced, but to somehow

influence or disincentivize that kind of

unrestricted free-lance approach to how these

medicines are used.

I think that has to be kept in mind, and I

would just like to emphasize that.

DR. GOODMAN: Dr. Perrin? Dr. Irwin.

DR. IRWIN: I would just like to respond

to that. I mean I agree with you that the

148

distribution of these medications by individuals

who aren't trained is really a worrisome fact, but

I will tell you in San Francisco County, to find an

individual who is trained to see a child with a

mental health disorder is virtually impossible

unless someone walks through the door with $175 in

hand to give to a psychiatrist.

So, I think that what has happened, and

it's a fundamental problem that we are dealing

with, and it is not the purview of this committee,

that the issue of mental health problems in

children and identifying individuals to care for

them or finding individuals to care for them is

really very, very difficult, and it pushes primary

care clinicians to prescribe and make judgments,

and provide medications when they probably should

not be without appropriate consultation.

So, I think it is a real major crisis.

DR. FANT: I agree 100 percent, and I make

those comments fully cognizant of the fact that the

mental health arm of health care in this country is

probably in worse shape than health care in the

149

broader context.

Certainly, in terms of its availability,

certainly in terms of its coverage by health

insurance plans, it is not sufficient to do what it

needs to do. But I think it is important not to

make it easier for bad medicine to be practiced

under those conditions, but to somehow create

conditions that kind of force, at least under those

suboptimal conditions, some better protections and

better practices.

DR. GOODMAN: Drs. Pfeffer and Wang, and

then we are going to take a break.

DR. PFEFFER: I would like to just go to

another area of our discussion and that is to focus

on what I will call the real world issues. I think

that the speakers yesterday in their own way gave

us a representative view to some degree of that,

and I am wondering, someone mentioned, I think Dr.

Pine, about prescriptions and can they be a little

bit more either regulated or I will say focused.

I am wondering if we might consider the

option of when a physician writes a prescription

150

for a medicine, such as an SSRI, that what is also

included on the prescription is the diagnosis, so

that we would have the kind of data that we just

heard, but amplified by some knowledge at least of

what the clinician is thinking about the rationale

for the prescription.

I know we do that in New York relative to

controlled substances at times, or other, in the

clinics, for example, the prescription forms

actually have that on their forms. So, I don't

think it's any greater violation of patient's

privacy than to say a patient is already on a

medication.

It might provide us with some additional

national-like real world practice ideas.

The other comment I would like to make is

that I think it was Dr. Zito who mentioned

yesterday about a proposal of having sort of a more

widespread, sort of service oriented approach to

study the issues also.

So, I think while we are talking about

constructing drug trials that are carefully

151

controlled and carefully defined in terms of the

population, given the fact that the prescriptions

are being used much more widely, it might be

helpful for us to have a view, a focus, and how can

we study these issues also, and the studies need to

be done obviously in different ways.

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

question?

DR. WANG: No, it was covered.

DR. GOODMAN: I would like to take a

10-minute break and then we are going to return for

a presentation, and we need to at least handle one

of the questions before we break for lunch.

[Break.]

DR. GOODMAN: We are about to begin.

Please take your seats.

If you recall from yesterday, when Dr.

Wysowski was presenting, she said she had some

additional data that was provided on the Jick

study, and we asked her to defer until today to

present it. I think she should be able to do that

pretty quickly, also give you an opportunity to ask

152

her any other questions that you think are relevant

to today's discussion.

At the close of that presentation, we are

going to get down to business in addressing these

questions sequentially. In the course of doing so,

I am going to ask you to, as best as possible, to

restrict your comments and the discussion to the

question at hand. Otherwise, I may defer that to

later in the course of our discussions.

So, with that, Dr. Wysowski, could you go

ahead and present the data. Maybe you want to give

us a little bit of a sense of the context first.

Diane Wysowski, Ph.D.

DR. WYSOWSKI: Right. I don't know how

important or relevant this is at this point in

time, but I am going to ask you to switch gears

from the clinical trial data and think back to my

presentation yesterday morning, which was on

patient level controlled observational studies.

I talked about two studies, the Jick study

that was published in JAMA, and the Valuck study,

but I am going to have you focus on the Jick study

153

and recall that it was a case-controlled study. It

was done in the United Kingdom and the General

Practice Research Database, and it examined the use

of four antidepressants - amitriptyline,

fluoxetine, paroxetine, and dothiepin in suicidal

cases versus non-suicidal controls.

In their original analysis, Dr. Jick and

his colleagues used dothiepin as the reference

category.

At FDA's request, Dr. Jick and colleagues

kindly re-analyzed their multivariate data for

nonfatal suicidal behavior using amitriptyline

rather than dothiepin as a reference category.

The data are controlled for age, sex,

calendar time, and time from first antidepressant

prescription to onset of suicidal behavior. Now,

in the lefthand portion of the slide, you see their

original analysis with the risk ratios and 95

percent confidence intervals with dothiepin as a

reference category.

On the right, with amitriptyline as a

reference category, the risk ratios for both SSRIs

154

increased and became statistically significant at

the 0.05 level. The risk ratio for dothiepin was

1.21 with a 95 percent confidence interval that

included 1.

For fluoxetine, it was 1.40 with a 95

percent confidence interval of 1.03 to 1.91, and

for paroxetine, it was 1.55 with a 95 percent

confidence interval of 1.11 to 2.16.

Now, the investigators asked that their

interpretation of these results be presented

verbatim to the committee. We advised that these

post-hoc analyses be interpreted with caution.

They were not the preplanned primary analysis, and

the p-value and confidence intervals are not

adjusted for multiple hypothesis testing.

We think conservative interpretation

requires that p-values lower than 0.05 or

confidence intervals with coverage greater than 95

percent would be necessary to assert that these

results are statistically significant with overall

5 percent Type 1 error.

These results are consistent with the

155

interpretation in our report that the risk of

suicidal behavior after starting antidepressant

treatment is similar among users of amitriptyline,

fluoxetine, and paroxetine compared with the risk

among users of dothiepin, and that a possible small

increase in risk bordering statistical significance

among those starting the newest antidepressant

paroxetine is of a magnitude that could readily be

due to uncontrolled confounding by severity of

depression.

Moreover, we did not observe an increased

risk of suicide itself for the users of

amitriptyline, fluoxetine, or paroxetine compared

to users of dothiepin.

So, that is their supplementary analysis,

it was a post-hoc analysis, and that is their

interpretation of the results. It did increase the

two SSRIs to the level of statistical significance

at the 0.05 level.

DR. GOODMAN: Now, dothiepin is not a

medication available in this country. It is also a

tricyclic antidepressant, as is amitriptyline,

156

isn't that true?

DR. WYSOWSKI: That is correct, but again,

the choice of the reference category makes a

difference in the results.

DR. GOODMAN: Dr. Perrin.

DR. PERRIN: I think it is very helpful to

have these additional analyses. I really had,

though, a couple methodologic questions about this

and the Valuck study that I think are fairly quick.

One is, in the British database, how valid

or reliable are these measures of suicidal

behaviors, not achieved suicide, and similarly, how

good were the measures in the Valuck study of--I

can't remember the terminology they used off the

top of my head--but of suicidal behaviors given the

kind of database they had to work from?

DR. WYSOWSKI: Well, one of the concerns

that I had, which I expressed yesterday, was that

the possibility of missing data and incomplete

ascertainment, and misclassification, and when you

are talking about suicide ideation, which is a

softer, more subjective diagnosis, it is difficult

157

to know how many people actually come forward and

report that. I guess if it's a serious concern,

they come forward and report it.

One of the things that I think Dr. Jick

says is that there is some possibility--the general

practitioners on which the data are based, they

make these notes, and it looks like they get

entered into the computer about 90 percent of the

time is what I figure from what Dr. Jick says here.

So, if there is some misclassification in

that way for the 10 percent that don't get entered,

of the missing data, you would be concerned.

DR. PERRIN: But it requires that the

general practitioner actually puts it in his or her

notes for it to get even possibly entered. There

must be substantial variability in that phenomenon.

DR. WYSOWSKI: Well, they do have

computers, and these people were trained, and so

they achieved a level of training success to be

entered and qualified for this database, but it

sounds to me like--it says here, "Information on

patient referrals and hospitalizations available in

158

the manual medical records in the general

practitioners' offices was recorded on the computer

more than 90 percent of the time."

So, that implies that about 10 percent of

the time you are not going to find the data there.

It was in the manual record, but not on the

computer. So, there is some possibility for some

error there.

Now, how that actually works out into the

results, don't know really.

DR. PERRIN: For the Valuck study, do you

have information on their measure of suicide

attempt?

DR. WYSOWSKI: The Valuck study was based

on paid medical claims data and of 70 managed

health care organizations. I talked to Dr. Valuck

about that, and he said that he thought that the

PharMetrics integrated outcomes research database

that he used, which is the 70 managed health plans,

that the data was very good and very complete, but

one of the things that is a problem with his study

is that he cannot go back and validate through

159

medical records the information that he has on the

computer, so that is one problem, but he said that

was better than most Medicaid databases.

DR. PERRIN: Most Medicaid databases are

very poor for analyzing children's mental health

services.

DR. WYSOWSKI: Right.

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

question?

DR. ORTIZ: I just was wondering if Dr.

Wysowski could clarify the risk ratio for this. Is

it suicidal behavior, suicide attempts?

DR. WYSOWSKI: It is nonfatal suicidal

behavior, which includes ideation and attempts.

DR. GOODMAN: I believe in reading the

Jick paper, there was reference made to 15

completed suicides in the entire population.

DR. WYSOWSKI: Right.

DR. GOODMAN: And they go on to point out,

those 15 were within the 10 to 19 age group, in

fact. There were others obviously outside that

range, but that none of the 15 in that younger age

160

group were on antidepressants.

DR. WYSOWSKI: Right. Actually, they

include suicides in their study, and there were 17

cases and 157 controls, but also I think Dr.

Goodman is referring to the fact that, yeah, there

is some information on children that committed

suicide, and it was somewhere on the order of about

15.

But that makes me wonder. I wasn't able

to determine whether that 15 is a reasonable rate

or not, so, you know, whether that is

under-ascertainment or not, we don't really know.

DR. GOODMAN: That, in part, was my

question. I wasn't sure from reading the paper how

they ascertained it, but it still was striking that

none of them were on antidepressants.

DR. WYSOWSKI: However, they do say here

that causes of death in particular are routinely

recorded, so you would think that they would have

pretty good data on deaths, but you don't know, and

all the suicides, but you don't know.

DR. GOODMAN: Dr. Gibbons.

161

DR. GIBBONS: Most statisticians view

large-scale naturalistic observational studies as

statistical atrocities. I am not one of them. I

think there is a great deal to be learned from

naturalistic observational data.

Having said that, I think it is very, very

important to protect oneself from bias due to

selection, and it appeared to me here that, you

know, there is all kinds of selection bias as to

who gets on to what kind of medication.

I guess, first, there is a question, well,

first, there is a statement. These are sorts of

cases where things like propensity score matching

and other sorts of methods that have been around

for a long time for the analysis of observational

data are critically important.

Covariate adjustment typically, which I

imagine was done here, can be very misleading,

because it assumes linearity, and many of these

relations aren't linear, and, in fact, the biases

cannot be overlapping. You could have situations

where the people who got on to one drug, don't look

162

anything like the people in the other, and, in

fact, the use of covariates in a general linear

model will be more misleading than helpful.

Have they made any attempt to do this

simple analysis using some form of propensity score

matching to ensure that the likelihood of taking

the drugs is consistent between the two groups?

DR. WYSOWSKI: Yes. Dr. Jick did not, but

Dr. Valuck did, and I presented those results

yesterday. In the poster, there was no increase in

risk for any of the antidepressants, but they had

classes of antidepressants, as you recall. They

are not individual antidepressants.

For the expanded study, which was based on

24,000 patients with diagnosis of major depressive

disorder, they did find a relative risk of about

1.58 for the SSRIs, but it was not statistically

significant. They did include a propensity

matching adjustment.

DR. GIBBONS: Given that we are seeing

this consistent 1.4, 1.4 through a lot of the

analyses, it would be very interesting either to

163

have them do those analyses or perhaps the

committee would think about performing such

analyses if we could acquire the data.

DR. GOODMAN: Thank you very much.

DR. WYSOWSKI: You are welcome.

Discussion of Questions and Vote

DR. GOODMAN: It is time to roll up our

sleeves. I would like you to pull out the

questions before us. It was given out attached to

the agenda for the meeting, on one page, or you can

use it as represented in the slide handout that Dr.

Laughren gave earlier today.

Let me make a few comments first, before

we begin to address these questions. For the most

part, they are focusing on risk, specifically, risk

of suicidality, and it seems to me it is not until

Question 4 that we need to be thinking about ratio

of benefit to risk.

So, I think for the most part, the first

three questions focus entirely on risk, but in

order to come up with some recommendations in terms

of regulatory actions, we do need to consider the

164

balance between risk and benefit.

Also, in talking about suicidality, there

are some definitional questions that have come up

all along, and I think we need to be, among

ourselves, as clear as possible what we mean when

we say suicidality. Maybe there will be some

benefit from the work that the Columbia group has

done to help us make sure that we are using the

same language.

Also, I think it behooves us to try to

translate what suicidality means to the general

public. In looking at some samples of the morning

papers, front page New York Times, front page USA

Today, there are headlines about how it has been

concluded already, based on yesterday's discussion,

that the antidepressants increase suicidality in

children.

I am trying to imagine. I would be

interested in how a parent, in reading that, what

they would think, what do they mean by suicidality.

My guess is that they are going to think that it is

suicide. As we discussed yesterday, this includes

165

suicide, it includes suicide attempts, but our

definitions also includes preparatory actions and

ideation.

So, I think we need to be very clear that

we are using the same terminology, and maybe Dr.

Posner will be able to help us along the way in

that.

As I said a little bit earlier, that we

are going to try to keep very focused on the

question at hand, so I may defer some of your

questions until later in the day. I would like

very much to be able to answer at least one,

perhaps two, questions before breaking for lunch,

so that the reward will be lunch to get some of

this work done.

My sense is that Question 1 may be the

easiest question, and then they may get

increasingly more difficult, so I think we need to

pace ourselves accordingly.

Also, just as another kind of overarching

statement is that for the most part, at least in

the beginning, we are asked to focus exclusively on

166

the clinical trials, but as we begin to deliberate

on issues of risk and benefit, then, we have to

then begin to consider data from outside those

clinical trials and therefore, our task becomes

more difficult and more complex.

Any questions about process before we

begin our comments?

MS. DOKKEN: Just before we start on the

specifics of Question 1 or 2, as someone new to the

committee, who is trying to listen and learn about,

I do have a question related to guidance from both

the Chair and the staff, and I would feel more

comfortable hearing about this, I guess, before we

even get into Question 1, which is I hear still

some discomfort about the data.

On the one hand, the options of labeling

and black boxes, et cetera, but I hear another

theme, too, which was articulated in particular by

Ms. Griffith, and that is, even though we are two

advisory committees only, and even though the FDA

is regulatory, is it within our purview, and more

so, do we have a responsibility to go beyond

167

something narrow like that, like labeling, and also

talk about the issues.

As I said, Ms. Griffith referred to

education, not just of clinicians but of the

public, and how much time, you know, can we

allocate time to that, as well.

DR. GOODMAN: Dr. Katz.

DR. KATZ: One response to the last part.

I think when we get to the appropriate time in the

meeting, I think you should discuss in a very

freely ranging way what provisions you think would

be a good idea to institute in order to use these

drugs safely.

I wouldn't worry so much about the nuances

of what your responsibility is or what we can do

from a regulatory point of view. If there is

something that you recommend, and we have already

talked a little bit about this, if there are things

that are outside our purview and we are incapable

legally of instituting, we will get back to you on

that or will let you know, but I think we really

want to hear a relatively wide-ranging conversation

168

about what sorts of things you think might be

useful. If we can't do it, we will let you know,

but we would like to hear about it.

DR. GOODMAN: Any other comments about

process?

I also noticed behind me that we have

Question No. 1 projected, so you don't need your

reading glasses to address it.

Let me read it. Please comment on our

approach to classification of the possible cases of

suicidality. Here, by parentheses, that word has

been defined. Suicidal thinking and/or behaviors,

and/or analysis of the resulting data from the 23

plus 1 pediatric trials involving 9 antidepressant

drugs.

Another thing I probably should mention is

that in terms of the questions we vote on, clearly,

No. 2 is one we are going to vote on, and a yes

vote then would lead us into No. 3. I am not sure

that No. 1 requires a vote. You are asking for

comment.

Are you asking for a vote on this

169

question?

DR. KATZ: No, we are not asking for a

vote, just comments, just a general sense of the

committee.

DR. GOODMAN: I was going to actually

start and try to answer the question first, to go

out on a limb, since this is the easiest question,

I thought I would take a stab at it, in my opinion.

But if there are other comments, and I will be

going around the table, so that each of you will

have a chance to comment, so unless it's really a

process question, you will get your chance.

It is a process question.

DR. NEWMAN: This is a process question.

It seems to me it might not be a good use of time

to go around the table and have everybody comment

on it. I mean I just think they did a great job,

we should praise them, and move on to some more

substantive issues that are important.

DR. GOODMAN: You stole my words, now what

am I going to say. I do like to go around the

table even if the comment is ditto, just to give

170

everybody a chance to speak.

Let me start. I would agree that I think

that given the inherent limitations of the data, it

was a very rigorous examination, very carefully

planned, involved leading experts. There was

appropriate blinding, more than adequate training.

We saw, too, that we had further vetting by

agreement with an independent study that was

conducted by the FDA in a subsample.

I think there is always room for

criticism, but I think most of my criticisms would

be regarding the inherent limitations of the data

itself, not what was accomplished by first the

Columbia group and then the FDA's re-analysis of

the data.

So, my comments are very favorable and I

can't see that there is much room for improvement.

In fact, I am impressed with how much they

accomplished, and I was somewhat skeptical in our

last meeting, and very impressed with the outcome

and attention to detail.

So, let me now go around the room starting

171

with Dr. Fant.

DR. FANT: My only comment, I agree with

everything that you just said, my only comment is

in the same spirit of my father, when I would come

home with a 98 on my test, and he would wonder why

I didn't get 100, and I guess my question to Dr.

Posner is, is this scale applicable across various

cultures and racial groups in terms of behaviors

and actions that may be significant with respect to

the endpoint that may differentiate one group from

another, that may be useful in trying to explore

and develop those, validate this tool in different

groups?

DR. POSNER: The answer is yes, and most

of the studies that these concepts are based on

have a lot of heterogeneity in terms of race and

general populations, so absolutely.

DR. FANT: So you feel comfortable that

this tool would be just as useful in an inner city,

predominantly African-American clinic, as well as

one that serves Southwest Hispanic community versus

one that serves Native American kids on the

172

reservation?

DR. POSNER: I think that the definitions

and the classification, the underlying concepts in

the classification that were represented in what

you saw yesterday, absolutely, and again were based

in just those populations.

In the NIMH study that I referred to

earlier, for example, there is a range of all of

the populations that you just mentioned, and those

are exactly the kind of behaviors we are looking at

and the samples that we are looking at.

Again, it's overarching concepts and the

behaviors are similar across all those groups.

DR. GOODMAN: Dr. Pfeffer.

DR. PFEFFER: I think this is an excellent

scale as a classification. I think that it's a

wonderful beginning and carried out with real data

regardless of the quality of the data.

I wanted to make some points just for

clarification, and I think that is what Kelly was

just pointing out, too, in a way. This is a

classification, and I don't know that one can yet

173

extrapolate this to, let's say, use with patients.

This is a secondary means of focusing

individuals in terms of their behaviors or

thoughts, but not yet to gather the data about

them, and I think that is important to emphasize,

so that much work I think needs to be done to

create the kind of an instrument that could be used

reliably and validly to assess directly from the

patients what the nature of their thinking is and

the nature of their intent, as well as the behavior

itself.

The other point I would like to make is

that this classification does hold across all age

groups generally, but in creating a method of

interviewing and establishing the information, what

is also necessary is to consider developmental

perspective, because we do know that children's

cognitive capacities of understanding the nature of

their planned behaviors are quite different than

adults.

I will go back to the example that tends

to be thrown out, and I would agree it should be

174

thrown out, but the child who slapped herself, for

example, that was one of the illustrations.

You know, if you have a child that has

problems or immaturity and cognition, one might not

quite know what she was intending to do, and a slap

or a hit with a piece of glass or whatever, it all

can mean something quite different in the mind of a

child, so that much work needs to be done to tease

this out I think from a developmental perspective.

DR. GOODMAN: Dr. Pfeffer, I would

certainly agree with your point that the

assessment, the prospective assessment of the

patient is a different matter, but with regard to

the narrower question, as I interpret it, in terms

of the classification data, the process of

classification, would you have any additional

comments?

I think yours are more comments in terms

of what would be the next steps in implementing the

system, administering it to subjects in a

prospective study.

DR. PFEFFER: I think the classification

175

is excellent. I would also raise the question, the

children and adolescents who couldn't be evaluated,

was it due to the data itself, was it due to

questions about what category the child might fit

into, which I tend to doubt because they actually

solved those issues by discussions.

I think, generally, this is a wonderful

classification.

DR. GOODMAN: Dr. Posner, you had a

comment?

DR. POSNER: I just wanted to respond to

your first point, which is you are absolutely

right, and all of that work has been going on

simultaneously. So, we are quite well prepared and

enthusiastic actually about putting helpful

assessment tools into future studies now that ask

the right questions to be able to put events into

these kind of categories. So, there are two

separate questions and both being addressed, I

think.

I think your next question has to do with

our No. 3, the kids where we knew they hurt

176

themselves, but we couldn't say why--am I

correct--and the reason was because of the limited

data about suicidal intent.

So, these were narratives that said

superficial scratch on wrist, and that's it, and,

of course, not enough surrounding information to

infer any kind of intent, so again, that is why

that category was warranted. This is important, we

know they hurt themselves, but we just don't know

why. As the FDA told you, they put that into a

worst case scenario, sensitivity analysis, which I

guess looked very similar to the primary outcome.

Did that answer your question about those

cases?

DR. GOODMAN: Yes. Dr. Fost.

DR. FOST: No further comment, thank you.

DR. GOODMAN: Dr. Ortiz.

DR. ORTIZ: My comments I think are more

on the line of what Dr. Pfeffer had to say. I

think what Columbia has done is a wonderful start

for the pharmaceutical companies to improve their

identification of suicidal behaviors, and I think

177

the FDA has done a superb job of further analyzing

the 24 studies.

However, the 24 studies excluded children

who had suicidal ideation with the exception of 4

studies, and the vast majority excluded children

with history of family bipolar.

So, again, my concern is in regards to the

testimony yesterday that we also need to think

about families and the clinician out there

practicing, the pediatrician in Farmington, New

Mexico, who has a mother with a 12-year-old who

comes in, who says, "I want to die."

I think we need to also be thinking about

issues of side effects, of agitation, hostility,

delusions, mania, and violence, which this

particular population, I mean I think for a

classification system, it is great, but there is

other issues related to side effects and clinical

practice that I think affect suicidality

profoundly.

DR. GOODMAN: Thank you.

Dr. Malone.

178

DR. MALONE: I agree that the combined

study was done very well, and I think it is

encouraging that the original FDA study pretty much

agrees with the second study, and I think gives it

more validity that way.

DR. GOODMAN: Thank you.

Dr. Nelson.

DR. NELSON: I just have one comment, that

I think a reasonable topic for discussion going

forward by the Pediatric Advisory Committee might

be to think through what lessons have been learned

by this experience for the ability to compare

information across different drug development

programs within drug classes, because I suspect

this issue might exist in other areas.

So, I think that is worthy of focusing on

at some point in the future, just to get that on to

the docket.

DR. GOODMAN: Thank you very much.

Dr. Perrin.

DR. PERRIN: I think the classification is

great, and I would like to know a little more about

179

the discordant cases between the FDA and Columbia,

but I think that knowing more about the discordant

cases would not change the findings at all.

DR. GOODMAN: Thank you.

Dr. Grady.

DR. GRADY-WELIKY: I agree with everything

that has been said. I would just like to follow up

a bit on what Dr. Nelson said, which is that the

Columbia study actually showed us a great deal

about the role of narrative reporting, and I would

think it is very important that we look at, for

future studies, guidelines for those narratives so

we have further information.

DR. GOODMAN: Thank you very much, Tana.

Dr. Ebert.

DR. EBERT: I also agree, that the

classification I think was reasonable, and I

commend the investigators on that. As far as the

analysis, just one brief comment, and that again I

think the analysis was appropriate, but again we

have the caveat of the studies themselves being of

somewhat questionable quality, and the variability

180

of quality probably is hard to establish.

Having said that, when the studies are

analyzed, they are analyzed based on weighting

those studies on their size, which therefore gives

the greatest weight to the larger studies, not

necessarily knowing whether those are the highest

quality studies.

DR. GOODMAN: I like your last point.

Dr. Gibbons.

DR. GIBBONS: Clearly, there has been a

lot of excellent work done both in terms of the

classification and in terms of the analyses. I

think in terms of the integrity of these data, the

classification has gone about as far as you can

milk the data for, and I am not sure that we really

need to do much more in that regard.

In terms of the analysis, I think that the

analyses are very thoughtful, but I don't think

they have addressed the critical question that they

were intended to address.

I am reading from the summary minutes of

the February meeting, that "Since we are in the

181

preliminary stages of designing an appropriate

analysis of patient level data"--blah, blah, and

then it goes on, the analyses that are presented so

far are not really analyses of patient level data.

They are combinations of risk ratios in a

meta-analytic framework from study to study. They

are not patient level data.

The survival analyses that were done, to

some extent, are patient level analyses, but those

analyses are not adjusted for the effects of

covariates, and I really think there may be more to

these data than what we have seen, and would offer

that we should have a look at the data, and I would

be happy to do that.

DR. GOODMAN: So, you are making a

suggestion that there is opportunity for continued

mining of the data, and maybe we can put that in

sort of our parking lot and return to it as we get

to the recommendations.

DR. GIBBONS: Yes.

DR. GOODMAN: Any comments from the FDA

regarding the analytic questions that were raised

182

just now?

DR. LAUGHREN: We would be happy to share

the database, it is not a problem.

DR. HAMMAD: Regarding the fact that there

is no apparent patient level data, examining the

confounding on trial level was done based on the

patient level data, and also, as you said, time to

event also utilized the patient level data. Also,

examining the interaction in all trials by the

certified analysis used the patient level data, but

there might be some other things to be done putting

everything together.

DR. GIBBONS: I don't have necessarily any

expectation that a different analysis would yield a

different result, and I don't have any criticism of

the analyses that have been done. In fact, given

the time frame that were available, you have gone

way beyond any of my expectations, but I do have a

few ideas of how the data could be analyzed in a

different way, that might shed a slightly different

light.

DR. GOODMAN: Thank you.

183

Dr. Pine.

DR. PINE: Beyond the general comments

about the outstanding nature of the work, I guess I

would make only one other comment, and that relates

to some discussion between the last meeting and

this meeting, about the degree to which these

analyses were necessary or appropriate, and I guess

I would only just speak for myself to say that I

found them both helpful and in some ways necessary

to really inform on the next question that we are

going to deal with.

Again, just speaking for myself, I feel

far more comfortable being able to talk about the

second issue concerning is there or is there not a

signal, having seen the outstanding quality of the

work that was done.

DR. GOODMAN: Jean Bronstein.

MS. BRONSTEIN: I have nothing further to

comment about the study although I really thought

the analyses done with Columbia really helped me

better understand this issue than I did in

February.

184

I do at some point want to speak about

warnings and what I heard from yesterday's

testimony, and I think it really comes under the

next question rather than this one, so I will hold

that for then.

DR. GOODMAN: Thank you.

Dr. Rudorfer.

DR. RUDORFER: I would like to second the

excellent quality of the classification project. I

wonder if there is room to more formally include

informant information.

I am thinking particularly in some of the

coding, which have been described as softer, for

instance, suicidal ideation, should there be a

subcategory of suicidal ideation that is validated

by a family member as having been expressed as

opposed to just expressed by the patient.

DR. POSNER: Ideation is not typically one

of the categories that you would feel even the need

to be validated by a family member, because it is

what is going on in their head, so usually, the

most valid indicator of it is the child or

185

adolescent.

I understand what your comment is in the

narrative. Some of the narratives said, you know,

mom said that he said this, or the doctor just

said, you know, indicated what was said, but there

is no way to further break that down at this point

with the limited information that we have.

Again, I think it becomes more relevant

just from our assessment standpoint in terms of

behavior than ideation, having the supplementary

informants.

DR. RUDORFER: Thanks. The only other

thing I would add is just to re-emphasize--and

again this is not a problem with the

classification, this was a problem with the

underlying data--that your outcome was only as good

as the data that were available, and I think we are

all faced with that conundrum that those data seem

to be rather incomplete and inconsistent.

DR. POSNER: It is true, but I wanted to

highlight, in your handouts, you see the first

example I gave yesterday of the suicide attempt

186

that was clinically impressive, where the patient

took 100 pills, you have a very detailed narrative

in your handouts.

It is important to note that every one of

these narratives, many of the narratives had a lot

of supplementary information, and that was what was

so crucial about having suicidal experts, because

they can take all of that supplementary information

and say, yeah, this looks like a suicide attempt

given everything that we have.

So, I just think it is important to

highlight that there was a significant amount of

surrounding information even though stated intent

was very often not there.

DR. RUDORFER: Right. No, I appreciate

that, and I commend you for that. My concern

remains with, say, the placebo-treated subject who

walks in and verbalizes no complaints, and then the

rating process goes on and no one ever discusses

any surrounding issues because there doesn't seem

to be any cause for it.

DR. POSNER: Right, we can never make

187

sense of something that is not there.

DR. RUDORFER: Right.

DR. GOODMAN: I have already rendered my

opinion, I will just add by saying that I was not

prepared to answer Question 2 at the last meeting,

but I am now based upon the reclassification.

Dr. Chesney.

DR. CHESNEY: The good news is that I have

no further comments about the analyses other than

what the rest of the panel has said.

I was struck yesterday, as I was in

February, by the reports of the parents of a number

of children who never expressed suicidal behavior

or ideation, and yet proceeded to commit suicide,

and my second point has to do with other injurious

behavior.

We have looked at self-injurious behavior,

we have looked at obviously suicidal, but we

haven't looked at aggression, hostility, all of

those aspects of the activation syndrome that we

talked about in February.

The thing I feel relatively good about,

188

however, is that I think, had we looked at those,

it would have only strengthened the results that we

have already seen, so those are my only comments.

DR. GOODMAN: Thank you.

Dr. McGough.

DR. McGOUGH: I just agree with the

Chair's comments.

DR. GOODMAN: Ms. Griffith.

MS. GRIFFITH: I would endorse that, too,

and I agreed with Dr. Gibbons that the

classification has gone about as far as it could,

but I have a very quick question for Dr. Hammad,

because at the February meeting, you raised the

possibility that the data might not be robust

enough to render any conclusions, and your

formidable presentation yesterday, I suspect gives

you confidence, but I would just like to know if

you, indeed, feel that this is robust enough. I am

asking you a very subjective question, I am sorry

to put you on the spot.

DR. HAMMAD: Yes, you are right, it is

subjective. Of course, it depends on what do you

189

mean actually by "robust" here. I think if you

look at the individual trials, for example, you

feel that there is nothing going on, there is

nothing significant on its own, but when you see

how consistent the signal is coming from most

trials, putting this in the context of what the

rest of the process is, which is the fact that we

know now we have every event that is out there, as

well the public testified, and you still see it,

then, you can feel more comfortable about the

findings.

So, I agree with the comment that were

said before about the level of comfort that is

definitely much better than it was in February.

Also, the sort of things like the

information we have on discontinuation, for

example, also about the history of seratin [?], I

mean these two factors alone could have made the

results one way or the other, and if we did not

have information about those, and we had not tested

them, we would have spent a long time trying to

speculate how much actually impact we have.

190

So, we also got this out of the way, the

obvious, clear potential explanation for the

apparent risk. So, I think what we are saying now

is true.

MS. GRIFFITH: Thank you.

DR. GOODMAN: Dr. Leslie.

DR. LESLIE: I just want to say thank you

for the work you did.

DR. GOODMAN: Dr. Robinson.

DR. ROBINSON: In terms of classification,

I just want to second what most of us have said,

which is that I think the FDA and the Columbia

group did as good as they can with the data that

they had.

Just two comments. One is that in terms

of going forward, and Dr. Posner might obviously

have ideas about this, is that the Columbia

classification was obviously done in terms of what

you could get out of very limited data, and in the

future, sort of going forward in a prospective sort

of manner, you might have a different

classification or you might have a scale that had

191

additional items which might be very important,

which you couldn't get from retrospective data.

So, I think we still need to think about

that you can, for prospective studies, do something

maybe that is even better.

DR. POSNER: I just want to clarify one

thing, and it is a very important point. This

classification scale and scheme really is about

concepts and definitions, so we defined, suicidal

attempts were defined like this. We took the data

and put it into that category using that

definition.

Then, there are the measures that you use,

the tools, to ask the questions to ascertain that

information to be able to know whether that

definition applies. So, we do have those measures

and tools that aid in this classification that will

hopefully inform all of the studies going forward,

as you are pointing out.

So, the whole system really involves two

elements, right, the tools in which people and

clinicians and industry need to use to ask these

192

questions of these patients and families to find

out whether or not, where they go in this

classification scheme.

Does that clarify it somewhat?

DR. ROBINSON: Yes, but, for example, like

in preparatory acts, you have preparatory acts

where the person stops themselves versus somebody

else stops themselves, and often from a clinical

point of view, you know, it is like my child had

the rope up and I saw them and I stopped them, as a

clinician, that has a very different thing than

somebody saying, well, I was going to do this, and

I got the gun out, but then I told myself, no, that

is wrong, and I went to my family.

Again, for going forward, you might make

some refinements. I am just saying not necessarily

have this set in stone, because this is obviously

done for something that is sort of retrospective.

DR. POSNER: But what I am saying is we

have an assessment tool, for example, where all of

those questions, the clinician has those questions

in front of them, so have you ever done anything to

193

hurt yourself where you wanted to die, have you

ever started to do something and stopped yourself,

and then they get that information with those

definitions and the probes and the questions, and

then they can then go and decide there was a

preparatory behavior or there was a suicide

attempt.

So, all of those distinctions and

questions and helpful aids, we have certainly been

working on and intend to hopefully distribute and

even have guidelines and training days, so that

people can start to use these in their studies.

I just wanted to add to two comments I

heard about we have to look broader at more of the

other symptoms that we are talking about and

worrying about, like akathisia, agitation, and

aggression.

The study that I keep talking about, for

example, the Adolescent Suicide Attempter study,

that is NIMH-sponsored, we are working very hard on

measures and tools to look at all of these side

effects that are associated possibly with SSRIs

194

including all of those things - how to ask the

questions, how to collect it, so hopefully, we can

have tools and we can also answer some of these

related questions.

DR. GOODMAN: Dr. Marangell.

DR. MARANGELL: I would agree that the

reclassification and analysis were both clinically

and scientifically appropriate. I found them to be

rigorous. I was impressed with the blinding

procedures and would echo the thought for future

randomized, controlled trials, in conjunction with

the FDA, that there be some type of standardized

classification that is mandated across all studies.

DR. GOODMAN: Thank you.

Dr. Irwin.

DR. IRWIN: I agree and I would just like

to second what Dr. Chesney raised in terms of the

issues of aggression and violent behavior. They

don't seem to be a part of this instrument right

now. Thanks.

DR. GOODMAN: Ms. Dokken.

MS. DOKKEN: I agree with the intent of

195

the previous comments.

DR. GOODMAN: Dr. Newman.

DR. NEWMAN: I do, too.

DR. GOODMAN: Dr. Wells.

DR. WELLS: I think we have done about as

well as we can do with the reclassification and

with the re-analysis. We recognize, of course,

that the data aren't perfect, having largely to do

with how they were elicited.

I don't think that there is very much more

that we can do with the data, although there may be

a little bit more. I recall that Dr. Mosholder,

for instance, had recommended that we might want to

do an analysis using inpatient hospitalization as a

primary outcome and see what is picked up there.

We remain troubled by the inconsistencies

across the studies, even for specific drugs, we

don't understand what accounts for those

inconsistencies, but I think at this point, we need

to move forward and see if there is some decisions

that we can make with the data that we have.

DR. GOODMAN: Thank you.

196

Dr. Pollock.

DR. POLLOCK: Yes, it's a question was

there ever or is there any plan for this kind of

patient level, at least to sample, because of the

concerns about ascertainment in the adult studies,

if there was any contemplation or has there been

any probing at all of the quality of that data.

I mean we have some confidence in your

overall conclusion, but we don't have confidence,

or at least the public doesn't have confidence for

some of those adult studies, what was actually

recorded, and if it might be certainly in the

public's interest to conduct at least a quality

sampling using the same methodology in the adult

studies.

DR. LAUGHREN: As I indicated yesterday,

right now our focus is on the completed suicides

that we have in this very large database, and we

can consider this second issue, but I think right

now we are going to try and finish up with looking

at the completed suicide data.

DR. GOODMAN: Dr. O'Fallon.

197

DR. O'FALLON: I agree with what has been

said by the rest of you, I am not going to argue

with any of you, but I am still--I think there are

a couple of issues here.

One of them is that I don't think that

this re-analysis has done very much about looking

at that question about the association between

adverse events and dose changes. I think there may

have been problems. At least I didn't see very

much about that. Maybe I was missing it.

I think I am still very concerned about

the possibility that we might be underestimating

the incidence of these adverse events, the suicidal

ones. I am afraid of it because if we get the

answer wrong, we could have a very bad effect upon

medical practice. That is always the issue here

with doing research.

Obviously, there wasn't very much power to

detect, I mean it's a rare event, thank God it's a

rare event, suicidality, but there isn't a whole

lot of power to pick it up under the best of

circumstances even with the meta-analysis.

198

I think I have asked this question twice,

and I think the FDA needs to address upfront the

charges that I heard over and over again that the

FDA doesn't have all the data, that somehow or

another the companies are holding back data or

suppressing it from you.

I think that is something that has to be

made clear to the public. You have said no, if

they put in an application, we get every shred of

data they ever had even if it was 25 years old.

But there is a perception out there that I think

the FDA has to address.

DR. GOODMAN: I am not sure if those

comments are germane to the question at hand. You

may want to hold them for later.

DR. O'FALLON: Okay.

DR. GOODMAN: Dr. Santana.

DR. SANTANA: I also agree that clinically

and scientifically, you have done the best that you

can with the data that you have. I do want to move

forward, though. I mean this classification system

is an event-based, outcome system, but it really

199

doesn't get to the issue that I would have if I was

a practicing physician in this area, which is, is

this toxicity or is this lack of response, does

that lead to that common outcome.

So, as you develop your new tools, your

new questionnaires, whatever you are going to do to

validate this classification and take it forward to

new studies, I would want you to pay some attention

to try to dissect how that common outcome is

related to either toxicity or lack of response,

because I think that would be important and would

address some of the issues that the parents and

families had, you know, that they were attributing

it to toxicity, whereas, some of us may interpret

that it actually was a lack of response.

DR. GOODMAN: Thank you very much.

Dr. Wang.

DR. WANG: I agree these were very strong

process and results. I do have two small

suggestions to bound the lingering questions we

have about case ascertainment. The first is how

many cases may have been missed by the sponsor's

200

screen and never sent, and for that, you might

consider an audit of what was not sent, you know, a

sample of that.

The second is the potential ascertainment

bias due to this unblinding by side effects, and

you could check for this by seeing whether an

adverse event known to be unrelated to

antidepressants was elevated in the antidepressant

versus placebo arms. It would just at least give

us a sense of the potential magnitude of either of

these two problems.

DR. GOODMAN: Dr. Gorman.

DR. GORMAN: I would like to echo the

generally positive comments about the

reclassification as being helpful to especially

myself to understand the data, and that would then

lead to a compliment to the Office of Drug Safety

both at the global level and the individual level

for recognizing the signal through all the noise

when the data was not classified in a way to make

it as clear to them as it is now to us.

The ascertainment of the cases, I think

201

would make the signal stronger in general. I think

all the errors we are worried about, in general,

about how much information has been presented in

the narrative, might, in fact, make the signal

stronger and therefore, while I recommend to the

FDA as a comment to this that I hope this

classification system becomes generalized across

all your therapeutic areas, that the active

ascertainment for suspected or serious adverse

events in all classes become active and done in a

way that allows us to not be arguing whether we

have got as much of the signal as there is to get.

DR. GOODMAN: Thank you.

Dr. Maldonado.

DR. MALDONADO: I just have a couple of

questions, but I agree with the general consensus,

and the questions are based on these tools that

have been developed. As you know, the tools are as

good as the ones who use the tools. It is still not

very clear to me who is going to be the end user of

the tool.

I actually congratulate Dr. Iyasu for

202

doing the reproducibility and reliability within

the FDA. I am glad to see that the FDA is doing

its own studies, too. And then who are going to be

the end users, is it going to be the medical

officers on DDP who are going to be doing this

classification, or is it going to be the requesters

from the sponsors, or even primary investigators?

The more you spread that, the more

variability you have to expect, and then the study

that Dr. Iyasu did might not be relevant depending

on the user.

The other thing, maybe Dr. Posner can tell

us, where the publications for the validity of

these questionnaires and classifications are,

because again these tools are so dependent on their

validity. I am not in this field, so they may be

published and people know, but I have never seen

them, or maybe if they are not published, are they

going to be published, so people know the validity

of these two tools.

I am not just referring to a

classification, but also to the questionnaires that

203

you mentioned.

Thank you.

DR. GOODMAN: Dr. Mehta.

DR. MEHTA: I think the FDA has done a

great job of classifying data with very poor case

of confirming information. I would go one step

further, and that is, request FDA to design a case

that could confirm suicidality and also together

with a set of instructions, and give it out to

every sponsor from now on, because I suspect that

this issue we will be revisiting 10 years from now.

I think Dr. Gorman and Dr. Posner also

commented essentially the same thing.

DR. GOODMAN: Are you going to give the

references?

DR. POSNER: No, I was just going to

reiterate that we have commented many times that we

are going to write guidelines just to do that for

industry and everybody else.

DR. GOODMAN: We can't hear you.

DR. POSNER: I was just reiterating again

that we, in collaboration with the FDA, are going

204

to write guidelines for better ascertainment. I

think we should also have training meetings. We

discussed this yesterday. Whatever we can do to

make this consistent across everybody who is going

to be doing this kind of work.

DR. LAUGHREN: Right, and that applies to

both the classification and the ascertainment.

DR. POSNER: Right.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: A classification scheme like

that is presumably what a company would do,

probably with a special group set aside to do it in

a blinded way, to evaluate the data they have got.

One other observation I want to make is

that one of the ways we try to focus on things that

are important is to look closely at all the people

who leave a study prematurely in association with

an adverse reaction, and the narratives associated

with that are one of the things medical reviewers

look at most closely.

We also get a fair sample of dropouts that

were said to be for administrative reasons to see

205

if underlying those there is actually an adverse

reaction, because sometimes you want to check those

things.

So, that is one of the ways you go looking

for things you don't know enough to expect, see

what happened in those people.

DR. GOODMAN: Thank you.

I want to conclude our discussion. You

can take a seat. Thank you.

There, you have our comments. I think it

is pretty straightforward. There was a great deal

of agreement that you can't imagine a better job

being done given the starting point.

Naturally led to discussion about what to

do in the future, and I think there are some

excellent suggestions there, not only for the FDA,

but the field in general in terms of improving our

ability to detect, ascertain suicidality and

perhaps other symptoms that might be relevant and

help us sort out whether we are dealing, as was

said before, with toxicity versus an indication of

ineffectiveness.

206

So, I think that in the future, hopefully,

we will be ascertaining and classifying these data

in a prospective fashion, and obviously, a lot of

the details need to be worked out about who will be

doing what part of that job.

With that, we should head to lunch, return

at 1:00 p.m. to tackle the remaining questions. A

reminder, once again, this should be ingrained. Do

not discuss meeting questions during the lunch.

Thank you.

[Whereupon, at 12:02 p.m., the proceedings

were recessed, to be resumed at 1:00 p.m.

207

A F T E R N O O N P R O C E E D I N G S

[1:00 p.m.]

DR. GOODMAN: Would everyone take your

seats.

First, a housekeeping matter. Anuja Patel

will be passing around a sign-up sheet, so that we

can know if you need a taxi and at what time. I

assume you want that returned to you, Anuja, after

it has made its way around the table.

We are now entertaining the second

question before us. It is presented there up on

the screen. I will read it.

Do the suicidality data from these trials

support the conclusion that any or all of these

drugs increase the risk of suicidality in pediatric

patients?

Now, what I would like to do is first have

a discussion of the question, give you an

opportunity to ask any further questions of

individuals from the FDA who presented yesterday

that have bearing on this question.

Then, we will, following that discussion,

208

go around and ask for your votes. You choices are

Yes, No, or Abstain, and you are permitted to have

30 seconds, not much more, to explain the rationale

for your vote.

So, first, we are going to have a

discussion. This will be the opportunity to see if

we can extract any additional information. I also

wish to point out that there are four members of

the committee at the table who are non-voting

members. We welcome them to participate in the

discussion phase, but obviously, will not be

participating once the vote commences.

Their names are Dr. Mehta, Dr. Maldonado,

Dr. Gorman, and Dr. Wang.

In posing this question, I had a few

comments, and maybe a question to the FDA in terms

of clarification.

First, I had mentioned earlier that there

is some lack of clarity about the definition of

suicidality. In fact, as we can see on the other

screen, although there is quite clarity there, you

can set the brackets either narrowly or broadly in

209

terms of what we mean by suicidality.

For the most part, in the analysis that

was presented yesterday, the definition of

suicidality corresponded to Outcome 3, which

included evidence of suicide attempt, preparatory

actions or suicidal ideation.

So, I think before we take a vote on that

question, there should be some discussion and maybe

some guidance from the FDA, as well, is which

definition of suicidality are we adopting for the

purpose of that vote.

Second, I wanted to note that if we are

basing the information exclusively on the clinical

trials, as stated explicitly in the question, we

have no instances of suicide, so we would not be

concluding anything about suicide, only the risks

of suicidality, not completed suicide.

My feeling is--again, I pose this to the

FDA--we cannot ignore the other information we

heard from the public testimony about cases of

completed suicide, and obviously, those are not

from the trial, yet we can in some ways extrapolate

210

from the ideation and behaviors in the trials to

the risk of completed suicide that perhaps would

exist in the absence of a carefully controlled

environment, such as is the case in a clinical

trial.

So, maybe I could start by posing the two

questions to the FDA. One has to do with which

definition of suicidality should we be

entertaining, and, secondly, should we limit this

answer to what we know from the clinical trials.

DR. LAUGHREN: Our intent was that you

focus on Outcome 3. That was our primary endpoint

in the trials, so that is what we intended by

suicidality. I think for the purposes of this

question, we would like you to focus on the

clinical trials.

I mean you can subsequently address data

from other sources, but we are primarily interested

with regard to this question on the clinical

trials. I agree with you that it applies to

suicidality, not completed suicide, because

obviously, there weren't any completed suicides in

211

these trials.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: The difficulty in dealing

with the question of completed suicides is that

while, unquestionably, some of the cases reported

sound pretty interesting and persuasive on the

point, you have no idea how persuasive the decrease

in suicide that other people alleged, how large

that is.

So, how to say whether there is a net

benefit or harm on completed suicides certainly is

unclear to me. Those data are very hard to analyze

quantitatively. That is not the same as saying

that some people don't seem to get worse when they

are on these drugs, but some people seem to get

better also. So, how to put that in numbers that

addresses that question, increasing, say, the risk

of suicide, that seems very hard to do.

DR. LAUGHREN: I guess another

qualification here is that obviously, this is a

very small window in time that we are looking at.

These are short-term trials. I think it has to be

212

focused on that window in time for which we have

data. I mean in my view, that is really the

question.

DR. GOODMAN: Other comments from the

committee?

Dr. Nelson.

DR. NELSON: The question I had, let me

just be clear, because I think might have been

answered by your response, because the way the

question is framed doesn't say anything about the

timing of the suicidality and some of the

discussions of early versus late, and questions of

late decreases, et cetera.

So, by restricting the answer to this

question to the data at hand, the way I would

interpret it is it is to speak directly to the

early possible increase in signal that is seen, not

to the broader questions, which then would come in

to play perhaps in tackling Question 4, where the

risk-benefit becomes an issue.

So, I state that as a comment, I guess, in

order to make sure that I am interpreting what you

213

have said correctly, because I was going to ask

about how this would be focused on the early versus

late kind of issue that has been part of this

discussion.

DR. GOODMAN: Dr. Ebert.

DR. EBERT: Just another clarification for

my purposes, and when we talk about this, I am

assuming we are talking about an increasing risk

compared with placebo as opposed to an absolute

increase in risk, because obviously, that would

also take into account the potential efficacy of

the drug.

So, in fact, we might be, if the drug is

efficacious, seeing a net reduction in suicidality,

but we are talking here about comparing it with a

placebo.

DR. GOODMAN: I would agree with that

interpretation.

Dr. Irwin.

DR. IRWIN: Is there a word suicidality?

DR. GOODMAN: Every time I write it in

Word, it gets red underlined.

214

DR. IRWIN: It seems to me, I mean to me,

I am not certain anyone really knows what it is

that we are saying and what you are voting on, or,

to me, I would like to know what suicidality is.

DR. GOODMAN: I don't think it is in an

Oxford Dictionary either.

MS. GRIFFITH: It is not in Webster's.

DR. IRWIN: In a sense, it confounds

things by, you know, the front page of the paper

today, I think may lead to kind of a

misrepresentation.

DR. POLLOCK: Can't we just use the

explicit language?

DR. GOODMAN: That is, in part, what I

would favor, is that if we use it, I think we need

to at least parenthetically define what we mean

when we are answering the question.

Dr. Temple.

DR. TEMPLE: Yes, that is what we do. I

think that is what we actually did in labeling.

Whether we should coin a new word is debatable,

obviously, but it means suicidal behavior plus

215

suicidal ideation. That is what we use it to mean

as those items.

DR. GOODMAN: Would it be fair for us to

slightly modify the question, or do we have to take

it as it is, because what I would say, if we could

use the definition that corresponds to Outcome 3, I

would feel most comfortable, because that

corresponds to the reclassification and the way you

approach the dataset.

So, suicidality, suicide attempt,

preparatory action/or suicidal ideation.

DR. KATZ: Yes, you can certain amend the

question. We called it suicidal behavior and

ideation, but it is clearly what is embodied in

Codes 1, 2, and 6.

DR. GOODMAN: I think we have a

clarification on that and hopefully, the public

will understand what we mean, too, and that, I

think we will leave it to the press to do their job

in trying to best define what we mean and don't

mean by that term, specifically, that we are not

talking about actual completed suicide if we are

216

restricting our deliberations to the clinical

trials, because there weren't any instances.

Dr. Perrin.

DR. PERRIN: I would like to ask a related

question to help me understand how to approach this

vote, which is really not the analysis, but the

trials themselves and some I think fairly brief

questions.

My understanding from reading the reviews

in Dr. Dubitsky's presentation yesterday, and from

Dr. Hammad's review, that these are very diverse

populations in these trials, only variably well

described, so we don't really know what percentage

of these young people actually had major depressive

disorders even in the MDD trials.

We know very little about comorbid or

co-existing conditions in them. Although they

describe what the inclusion/exclusion criteria are,

we know relatively little about the concomitant

treatments for them.

Again, as I read the descriptions, they

are a lot of drugs that they might have been on

217

were excluded, on the other hand, about

three-quarters of all the samples were on some

concomitant treatment of some sort or other.

So, I take it, if I am reading this right,

a very diverse, hard to consider similar

populations across the multiple trials, which might

be the explanation for why nefazodone had

absolutely no events in 450 subjects.

I am really asking the question, am I

right in this reading, and, if so, because that

would help me understand more about the strength of

the signal given incredibly diverse samples.

DR. GOODMAN: Perhaps Dr. Dubitsky, is he

here, could answer that.

DR. DUBITSKY: I am not quite sure how to

even begin. It is a very complex question, and it

is very relevant. The diagnostic criteria did span

anywhere from DSM-III up to DSM-IV, including

DSM-IIIR, but beyond that, I think as I alluded to

yesterday, you know, some studies did use more

extensive diagnostic screening procedures, and it

is to me very unclear as to what role that may have

218

played in creating some diversity among the trials.

I think you mentioned the issue of

concomitant treatments, and it is true that most of

the patients did receive some kind of treatment, be

it something as simple as aspirin or another

antidepressant during the trial.

I think, in general, the treatment with

concomitant antidepressants other than the study

drug was fairly rare, but you can go on from there,

because you do have antipsychotics and all kinds of

other non-psychotropic medications, non-psychiatric

medications that do have psychotropic effects, and

it becomes very, very complex trying to sort that

out in terms of what the medication was, what the

actual psychotropic effect was, what the timing

was, how that may have influenced the outcome of

interest.

So, I don't have a good answer for that.

DR. GOODMAN: I think there were some

differences you pointed out and the degree of

structured interviews that were conducted, so that

may account for some heterogeneity.

219

I think even if the criteria were uniform,

the inclusion/exclusion criteria were uniform

across the studies, which they weren't, I do think

there is a great deal of heterogeneity, which has

to do with the limits of our ability to

characterize major depression in children.

DR. PERRIN: In that context, I think it

is also, we have heard the real difficulties in

distinguishing major depression and bipolar

disorder in these populations.

I guess the point that I think you are

supporting is very diverse populations,

nonetheless, a very persistent signal despite very

different populations.

DR. DUBITSKY: I think it is quite

possible that the population was very

heterogeneous. Again, to what extent that is

actually the fact, I don't really know, but there

is a distinct possibility.

DR. GOODMAN: Thank you.

I want it to be clear and make sure that

we all agree at this point and understand that as

220

we answer this question, we are restricting our

data to the clinical trials. I think that is the

intention of the FDA. I think that is reasonable

as long as we understand which question we are

asking.

We are not asking the broader question

based upon other data that has been brought to our

attention, strictly what can be gleaned from the

clinical trial data.

But I will entertain any discussion of

that point.

MS. BRONSTEIN: As I listened to the

reports on the studies and also listened to the

public testimony, I think some of the public

testimony really highlights the necessity to look

carefully at the trials.

The public is asking us very succinctly to

warn them, and I think we have done some since last

February, and I think we need to even do more in

the way of maybe even informed consent and using

family members as partners even more than we have

in the past.

221

But I want to harken on what Nami talked

about a little bit yesterday and some of the

clinicians that spoke. I am most concerned about

access for children to all of the kinds of things

that are available even with the known risks.

I guess, as somebody who is very concerned

about the consumer, I really want to focus on what

the signal is in terms of giving warning, not

necessarily restriction.

DR. GOODMAN: Dr. Marangell.

DR. MARANGELL: I certainly agree with,

for example, the next question focusing on the

clinical trials, that really is a clinical trial

question of what we know from the current data.

I think when we get to broader questions,

for example, should these antidepressants be

contraindicated in children, I think it is almost

impossible to address that question without

bringing in a broader database beyond the current

clinical trial.

DR. GOODMAN: I think that is a fair

statement.

222

I want to make sure before we try to

answer the question that we are clear about the

question. I think we are at this point.

Any further clarification needed?

[No response.]

DR. GOODMAN: Dr. Newman, I think you

wanted to add a little bit to our dataset. I will

give you an opportunity to do that.

DR. NEWMAN: I made a slide during lunch.

If I could have the pointer, too. When the FDA

staff presented the results of the pooled analysis

of the clinical trials, what they presented were

the relative risks and the 95 percent confidence

intervals, and although if you have a lot of

practice, you can look at those confidence

intervals and see what the p-value is, that does

take some mental arithmetic.

It is kind of hard to do those logs in

your head. So, these are the four risk ratios.

They are all about two, meaning that people who

were assigned to SSRI treatment in these trials had

about double the risk of these suicidality events,

223

and these are the lower and the upper 95 percent

confidence intervals, and the p-value is sort of

the measure of the strength of the signal meaning

how, if SSRIs did not cause suicidality, how often

would you see a signal this strong or stronger.

Just to show you that my little

spreadsheet way of doing it works for the people

who know some statistics, if the 95 percent

confidence interval just exactly hits 1, the lower

limit, then, that means the p-value is 0.05, which

it is here or reasonably close.

You can see that the p-value, that is the

chance of observing a signal this strong or

stronger, the highest one is about 0.04, and this

one here, which is the lowest p-value, because the

sample size is the biggest, because we are looking

at the possible events, as well as the definite

ones. This is Outcome 4, and in all the trials,

not just the MDD trials, is about 5 in 100,000, so

this is a signal strength that would occur by

chance about 1 in 20,000 times.

I think this is important because many of

224

the concerns that have been expressed by members of

the committee would be to a loss of power, they

would lead to we are not capturing all the

suicides, we are not sure that all we have is

suicidality, and heterogeneity, and maybe they

didn't even have the right disease and sloppiness,

all of those would tend to make the p-value higher.

So, I actually think the way the question

is phrased, which is does it support the conclusion

is actually a little bit weak. I think we could

phrase the question, it would be much stronger

about the data.

DR. GOODMAN: Thank you.

Dr. Nelson.

DR. NELSON: Thank you for doing that, but

I got the impression yesterday when you suggested

that, that a little statistical skirmish broke out,

so I am interested in hearing from the other

statisticians around the table just what they think

of this approach.

DR. NEWMAN: Actually, I talked to Dr.

Hammad ahead of time. Did you want to comment?

225

Okay. If the other statisticians would like to

look at this, or I could open up the spreadsheet

and show them, but I really, I don't view this as

controversial. If there are people that do, then,

I would like to hear from them.

DR. GOODMAN: Dr. Gibbons.

DR. GIBBONS: The computation is based on

the fact that these are asymptotic confidence

intervals, that is, you are assuming large sample

theory and assuming normality of the risk ratio,

and that is how Dr. Hammad did the computation, so

the probability values fall directly out of it.

Of course, it makes sense that when you

are right on the boundary of 1, the probability

should be 0.05 or close to it based on the 95

percent confidence of the asymptotic normal limit.

So, these p-values are reasonable, but be

careful about p-values. One of the reasons why

people use confidence intervals is to describe an

effect size, and a very small difference in an

effect size in a large sample will give you a

probability value that is very, very tiny.

226

So, don't interpret the difference between

0.05 and 5 times 10

-5 as being a huge difference in

effect size, but at the same time, if you are

worried about things like multiple comparisons,

like, hey, they went out and did a bunch of tests

and some of these are probably happening by chance

alone, you look at a value of 10

-5, you can do an

awful lot of comparisons.

We are all born with a fixed number of

degrees of freedom, and if you use them up too

quickly, you die a painful death, but that protects

you pretty well.

DR. GOODMAN: Thank you very much. That

is the best explanation I have heard of that yet.

Thank you.

Other committee members? Dr. Katz.

DR. KATZ: I just want to comment on the

last comment. It is true, I suppose, that 5 times

10 -4 or 10-5 protects you against a lot

of multiple

comparisons, but 0.04, which is the p-value, the

normal p-value for Outcome 3 for SSRIs and MDD, and

we haven't yet gotten to the point to which a

227

population should any conclusion apply, but the

0.04, in the face of lots of multiple comparisons

perhaps is a different kettle of fish.

DR. GOODMAN: Dr. McGough.

DR. McGOUGH: Tom, maybe a follow-up. As

far as the p-values go, if I humbly understand it,

you know, worrying about Type 1 error, we are

worrying about calling something different when it

is not, but I think the problem is the other way.

I mean I am more concerned of missing

something that is there. I am more concerned about

Type 2 error, and is there any light to be shed on

that, or are we just comfortable enough that we

have got a signal, we keep the signal, or is there

some way we can infer if we are missing a signal?

DR. NEWMAN: The Type 2 error refers to

that you have failed to find something which is

really there, and I would submit that these

p-values are very, very low, and so we have found

it, and it is there. When the p-value is 5 times

10 -5, power is not the issue at all.

You had

abundant, abundant power to find that because you

228

found it with such a very, very high level of

statistical significance.

DR. GOODMAN: Is there any more data that

people feel they need to see before answering this

question? Again, the data from the clinical

trials, not research you would like to see

conducted.

Dr. Pine.

DR. PINE: Just two brief comments, one

related to what you just asked and another related

to the other issue, and it relates to a conclusion

that Dr. Laughren gave when he was summarizing his

kind of directions to the committee, and that is

the idea that one might use a different statistical

threshold when making conclusions about safety as

opposed to when making conclusions about efficacy,

so while 0.05 has become kind of a magical number

for whatever reason, that is usually in the

discussions about efficacy, and I just wondered if

he would, you know, comment if his statements

really apply to this exact situation.

DR. LAUGHREN: My interpretation of the

229

regulations is that we don't need the same level of

certainty, so I think it applies directly here.

DR. PINE: Then, the only other comment I

would make, and this might affect the question or

it might not, and it relates to your statement

about other data from other trials.

As far as I know, there are no other

randomized, controlled trials of SSRIs in pediatric

depression. There are other trials of pediatric

anxiety disorders, and, you know, discussing them

for safety right now, I know is not really the

issue before the committee, but I think that there

has been a hint from the analyses that have been

done that perhaps the signal, so to speak, is

particularly strong in children who are suffering

from major depressive disorder.

I would just think it would be important,

if we were to go beyond major depressive disorder,

to be sure to look at trials that have not been

discussed, that are federally-funded trials in

particular as opposed to industry-sponsored trials.

DR. GOODMAN: Dr. Temple.

230

DR. TEMPLE: The analyses that Dr. Hammad

presented include all of the trials we know about

including trials that are not in major depressive

disorder, only 15 of them, or one more with TADS,

and the signal, as the previous slide showed,

actually looked slightly stronger when you add the

trials that are not in major depressive disorder.

What that means, I have no idea, but that

is how the numbers sort of come out.

DR. GOODMAN: Basically, I want to echo

that. Maybe I should clarify this point. As I

understand this question, it applies to all the

clinical trials, not just the major depression

ones.

Although the numbers were admittedly small

for some of the other anxiety disorders, like OCD,

they were very small numbers, but when you

aggregate the data, it adds to the evidence of

suicidality.

DR. TEMPLE: One of the things the

committee could think about is whether some of

those trials are more germane to this question than

231

others. The main analyses Dr. Hammad did included

them all.

For one thing, there is a lot of overlap

in these things, and maybe the people have more

than one disease, but that was the primary

analysis, and there are, of course, more data, more

numbers, more trials, so you have somewhat more

information on those than you do on the others.

Those are all good questions.

DR. GOODMAN: Dr. Pollock.

DR. POLLOCK: I just wondered if before

you bring this to a vote, if it is possible to

rewrite this, so that we are absolutely clear what

we are voting on, and it is not only--my concern is

not only that you define suicidality, but also this

may to some members appear redundant, but

pediatric, as well, that we actually talk about

clinical trials between ages of, you know, we fixed

that, because the way the warning came out, at

least into the field and into practice, in the last

few months, was all patients, that there is a risk

of increased suicidality sort of across the age

232

spectrum, and this caused a little bit of

consternation.

While it is important that the efforts

towards monitoring and that people be alert in all

ages, where we have the evidence and where we are

specifically voting, I think today is about the

evidence base that we have in those specific trials

conducted between certain ages, and that is where

the data is.

I would just again feel better if we could

just frame this question very explicitly, so it is

not subject to distortion.

DR. GOODMAN: I certainly agree that we

should further reference what is meant by "these

trials" in the question. Perhaps this list that

was supplied by Dr. Dubitsky covers it, but let me

make sure that we are voting on the right set of

trials, are they the ones that are listed here? I

am interested in a little bit of guidance about how

to properly reference the studies.

DR. TEMPLE: It is clearly the trials we

have presented to you. They are all pediatric

233

trials. If you feel that you have to say that in

there, go ahead and say it, but we will understand

it.

Those are the trials you are talking

about, and the database relates to suicidality in

pediatric patients, however, the warning you are

referring to was quite deliberately not intended to

apply only to pediatric patients, because it didn't

have anything to do with whether there is an

increased risk of suicidality. That was considered

good advice for any person getting these drugs.

You should know that sometimes people get worse and

you should monitor them closely.

So, you might want to comment on that, but

that was our intent.

DR. POLLOCK: It was conflated with this

stuff.

DR. TEMPLE: No doubt.

DR. GOODMAN: Any other data you want to

see, or discussion, before bringing it to a vote?

Dr. Laughren.

DR. LAUGHREN: Just again for

234

clarification, Question 2 is intended to follow

Question 1. Question 1 clearly states that it's

the 23 plus 1, 24 trials.

DR. GOODMAN: I apologize, I wasn't paying

attention.

DR. LAUGHREN: I am just pointing out that

Question 2 is intended to follow directly from

Discussion Point 1, which focuses on the 24 trials

for which we have presented data.

DR. GOODMAN: So, you know what we are

referring to, we now know what we are referring to,

so I think we are okay.

Ms. Griffith.

MS. GRIFFITH: But will the public know

what we are referring to, and when this is

extracted for the press, it better be as clear as

it can possibly be. Also, if I could just

reference your web site, you need to have something

very directly speaking to this and outlining it in

detail on the web site ASAP.

DR. GOODMAN: Further discussion before we

bring it to a vote?

235

Dr. Pfeffer.

DR. PFEFFER: Yes, I would agree with what

you just said in the sense that the question almost

alludes to a generalizability, and I am not yet

sure, given the discussions we had this morning,

that we are fully ready to have very generalizable

statements about the last part of the question,

suicidality in pediatric patients.

I think we have some datasets now, and we

have discussed how much they have potential

problems, but this is the existing data that we

currently have. So, as of today, this is our

knowledge base, so to speak, and we feel from what

we said this morning that we need more information

ultimately and gathered in different ways.

So, I am not sure we can say this is a

generalizable issue yet. So, that is the caveat I

would like to address.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: I am not sure what the

reference to generalizability is. This question

refers to the 23 plus 1 trials and to the evidence

236

that they do or don't provide about the risk of

suicidality, defined properly, in pediatric

patients who were the subject of those trials.

We don't think those trials have any

reference for adults. We don't know about the risk

of actual suicides, it is a fairly narrow question

because those are the new data we got. Those are

the result of the pediatric trials, and the

question is have they told us something.

DR. GOODMAN: I understand.

Dr. Pfeffer.

DR. PFEFFER: I guess what I am trying to

say is that what I mean by generalizability, I am

not talking about other ages, but specifically, I

don't yet think that it clarifies all pediatric

patients would be at risk.

DR. GOODMAN: I don't think that is the

implication of the question.

DR. PFEFFER: I am talking about the

general public.

DR. GOODMAN: It means in the trials that

were listed and presented to us.

237

DR. TEMPLE: And there could be

differences among patients in the trials. That is

true for every effectiveness trial that has ever

been done. You don't know as much as you would

like to. It's in the pediatric patients that were

in these studies, not all pediatric patients.

DR. GOODMAN: Tana.

DR. GRADY-WELIKY: I just wanted to

comment that the amount of discussion we are having

about what the question means or doesn't mean is

important to at least acknowledge and, you know, we

are a group of experts in this area, so when it

comes to the public, I think the fact that we are

struggling with it is needed to be commented on,

and I would agree with Dr. Pollock that I would

like to see the final version of the question

before voting.

DR. GOODMAN: Dr. Katz.

DR. KATZ: I just want to get back to this

question of generalizability, because we do clearly

want to be able to say something at the end of the

day about whether or not these results apply to any

238

identifiable population of pediatric patients.

If you were to conclude that, yes, the

data show that there is an increased risk of

suicidality in these specific particular pediatric

patients who enrolled, but we can say nothing

about, for example, pediatric patients with MDD, in

general, or patients with psychiatric disorders,

that would be quite problematic.

Generally speaking, we do take control

trial data and we convince ourselves that the

results apply to some relevant population that was

not studied in the trials. If all our conclusions

only applied to people in trials, we wouldn't have

very much to say about drugs.

So, there is a question we are asking.

You can take this stepwise if you like, but we will

ultimately want to know whether or not you think

that these data demonstrate that there is a risk of

suicidal behaviors or suicidality, as defined, in

some identifiable population who in the future or

who are currently being treated with the drug or

drugs.

239

DR. GOODMAN: I don't think the main text

of the question needs to be changed, but I do say

that it needs to be footnoted, so that what I would

suggest is that the statement--the question we are

answering reads as follows:

Do the suicide data from these trials--and

we should put Footnote 1--as listed in--what shall

we call this, Dr. Dubitsky? Appendix A, presented

by Dr. Dubitsky? Somebody give me another way of

describing that.

DR. TEMPLE: You could refer to it as the

23 plus 1 trials referred to in Question 1.

DR. GOODMAN: Okay, that is fine with me.

So, the 23 plus 1 trials referred to in

Question 1. Somebody is bound to ask me which 23

plus 1 trials those are, and those are in Appendix

A listed, provided by Dr. Dubitsky. We are getting

like lawyers here.

DR. MARANGELL: All available randomized

control trials in pediatrics involving

antidepressants.

DR. GOODMAN: I think we are beating a

240

dead horse, frankly. I think we all know at this

point what we are voting on, and hopefully, when it

gets translated somewhere that the press and others

will be attentive to exactly the appropriate

references.

Any other discussion? I want to get off

the question and on to information relevant to

arriving at an answer.

Any other discussion that we need to have

before you are prepared to make your vote?

If not, I am going to start, not with

myself this time, I am going to start from that end

of the room from my first voting member, Dr.

Santana, and then I am going to remind you--yes,

it's you--yes, no, you can abstain, but obviously

we would encourage you to be definitive with a yes

or a no, and up to 30 seconds in comment although

that is not necessary, a simple yes or no would be

sufficient, and we are going to be recording your

vote.

DR. SANTANA: That is why I am here. My

vote is yes, and I have no further comment.

241

DR. GOODMAN: Dr. O'Fallon.

DR. O'FALLON: I am going to abstain. I

am looking at this data, and I don't see that clear

signal that everybody sees.

DR. GOODMAN: Dr. Pollock.

DR. POLLOCK: Yes.

DR. GOODMAN: Dr. Wells.

DR. WELLS: Yes.

DR. GOODMAN: Dr. Newman.

DR. NEWMAN: I would vote yes and I would

even say that I think this particular question is

weakly phrased to say support the conclusion, and I

would also vote yes if it said do the suicidality

data from these trials prove beyond a reasonable

doubt--

DR. GOODMAN: Now, we are really becoming

lawyers.

DR. NEWMAN: -- increase the risk of

suicidality, because I really think it is

definitively shown.

DR. GOODMAN: Ms. Dokken.

MS. DOKKEN: Yes.

242

DR. GOODMAN: Dr. Irwin.

DR. IRWIN: Yes.

DR. GOODMAN: Dr. Marangell.

DR. MARANGELL: Yes.

DR. GOODMAN: Dr. Robinson.

DR. ROBINSON: Yes.

DR. LESLIE: Yes.

DR. GOODMAN: That was Dr. Leslie.

Gail Griffith.

MS. GRIFFITH: Yes, and I was convinced by

the signal exposed in the TADS data.

DR. GOODMAN: Dr. McGough.

DR. McGOUGH: Yes.

DR. GOODMAN: Dr. Chesney.

DR. CHESNEY: Yes.

DR. GOODMAN: Dr. Goodman, yes.

Dr. Rudorfer.

DR. RUDORFER: No. May I take my 30

seconds, please?

DR. GOODMAN: Yes, please.

DR. RUDORFER: In my opinion, most of the

trials we reviewed were scientifically flawed.

243

None were designed to address the question of

suicidality. What they were designed and powered

to address, namely, efficacy, most failed to do in

the major depressive studies.

I believe that we saw evidence of many

suicidal-related events, however, to assign

causality, I think that was not shown. I think

that, as we have discussed, to show a

differentiation between active drug and placebo, I

don't believe that the studies were properly

designed to do so, and we have no other

corroboration that the ascertainment of events was

equivalent, and the question of switch into mania

and akathisia, and whether those adverse events and

complications could, in fact, have resulted in or

coded as suicidal events, I think remains a real

possibility, and we simply don't have the data to

disprove that.

DR. GOODMAN: Jean Bronstein.

MS. BRONSTEIN: Yes.

DR. GOODMAN: Dr. Pine.

DR. PINE: Yes.

244

DR. GOODMAN: Dr. Gibbons.

DR. GIBBONS: Yes with a brief statement.

I think the effects are very small. I think they

are consistent across the studies, but no more so

than the actual data show in the simplest of views.

The rate of these events, Outcome No. 3 is about

double in the drug arms relative to the placebo

arm, and most of the analyses tend to corroborate

that.

Nevertheless, looking across these

studies, looking at the TADS studies, looking at

the naturalistic studies, we see a preponderance of

evidence in favor of rejecting the null hypothesis

of no difference.

I would not be totally surprised, though,

that in further analysis, we might find some

confounding factor, such as initial suicidal

ideation that might be biased across these studies.

Nevertheless, these are randomized studies, and

randomization is a very important tool, hard to

ignore.

Thank you.

245

DR. GOODMAN: Dr. Ebert.

DR. EBERT: Yes, with a footnote that we

are looking at the data collectively as a whole.

DR. GOODMAN: Dr. Grady-Weliky.

DR. GRADY-WELIKY: I also say yes with a

brief statement that yes to the question as revised

with the appropriate footnotes and definition of

suicidality as suicidal behavior and/or ideation,

and I would agree with Dr. Gibbons' eloquent

comments about the fact that it seems to be a

minimal risk, but something that we should agree

to.

DR. GOODMAN: Dr. Perrin.

DR. PERRIN: Yes, and I feel that the data

are really quite compelling given the incredibly

diverse, relatively poor studies, and that we find

a strong signal arising despite the inadequacy of

the studies is very compelling to me.

DR. GOODMAN: Thank you.

Dr. Nelson.

DR. NELSON: Yes.

DR. GOODMAN: Dr. Malone.

246

DR. MALONE: Yes, although I would like to

add the caveat that I think if you look at some of

the data we saw, that, in general, both drug and

placebo had a decrease in suicidality over the

course of the trials.

DR. GOODMAN: Dr. Ortiz.

DR. ORTIZ: Yes.

DR. GOODMAN: Dr. Fost.

DR. FOST: Yes.

DR. GOODMAN: Dr. Pfeffer.

DR. PFEFFER: Yes.

DR. GOODMAN: Dr. Fant.

DR. FANT: Yes.

DR. GOODMAN: Anuja is going to tally the

votes.

You don't get to vote, Dr. Temple.

DR. TEMPLE: I don't want to vote, but I

would like to ask a question.

DR. GOODMAN: Let me give the tally first.

A total of 27 voting. 25 Yes. 1 No. 1

Abstention.

Dr. Temple.

247

DR. TEMPLE: The committee obviously finds

these data quite convincing. I was just curious

about Dr. Rudorfer's reservation.

Do I understand that you think there may

have been an ascertainment bias, that certain clues

might make people more inclined to call this in the

treated group than the other group, is that the

nature of it?

DR. RUDORFER: That is part of it. In

terms of ascertainment, as I was mentioning at the

end of this morning, there was no systematic way of

collecting these data. We have no idea what

questions were asked in which study.

It seems to me plausible that a

placebo-treated patient, who was not volunteering,

say, somatic or other complaints, might be

subjected to less interrogation beyond the rating

scales than someone, for instance, who came in

complaining of GI side effects or other SSRI

typical side effects, and I was concerned about the

blind there.

My other larger reservation is that I

248

thought we can't have it both ways. Either we

think that these drugs are effective or they are

not, and if they are effective, then, we are

looking at a collection of studies which, for the

most part, are showing a lack of efficacy, and I

thought that is not the appropriate context in

which to evaluate the adverse effects, especially

one which we know is integral to the illness under

study.

For instance, if we were looking at, say,

a cardiovascular measure where the illness date

wouldn't necessarily be relevant, I would be less

concerned.

DR. TEMPLE: The TADS study, of course,

showed both effectiveness and an increase. That is

just one study, though.

DR. RUDORFER: I agree with you, on one

hand, yes, it is just one study. The other is that

TADS is specifically designed as an effectiveness

study, meaning very few exclusion criteria, with

the aim of following upon, but not replacing,

efficacy trials, and I am concerned that we really

249

don't have a collection of good efficacy trials to

evaluate.

DR. GOODMAN: Because of the overwhelming

affirmative vote to the last question, we don't get

to skip the next one.

Let me turn to Question No. 3 and read

that.

If the answer to the previous question is

yes, to which of these nine drugs does this

increased risk of suicidality apply? Please

discuss, for example, whether the increased risk

applies to all antidepressants, only certain

classes of antidepressants, or only certain

antidepressants.

Dr. Katz, do you want to clarify?

DR. KATZ: Yes, I do. The other grouping,

which we haven't explicitly described in this

question, would be what indications, as well.

DR. GOODMAN: I was going to add that. I

agree that I think that the other possibilities

would be the sorted or quarantine indication.

In terms of how to approach this, this is

250

a little bit more data intensive. In order to come

with an affirmative answer to the last one, you

only had to be convinced that the association was

true for one of the drugs. Now, we need to I think

have some reference and I wonder again if the

handout from Dr. Dubitsky would be appropriate to

make sure you all have handy as we take a look at

individual compounds and trials.

Again, I think this is going to be a

little bit more labor intensive.

Dr. Nelson.

DR. NELSON: It would just be helpful for

me to clarify the intent behind No. 3/4, because

one way of answering 3, when you get down to small

numbers in single trials, is to make a pragmatic

decision that you should then apply the grouped

data to individual drugs rather than just a

database decision that, in fact, you have enough

evidence, because I think we are going to take a

big group divided up into many small groups, and

the answer may be no, no, no, no for a number of

drugs where you would still decide that you would

251

want to have a class risk assessment.

So, it would be helpful. I guess I am

concerned that we don't get to the real question,

which is what to do, if we just spend time on nine

different drugs and three different indications.

DR. GOODMAN: I think that is a good

point. I also think there is a statistical

question embedded in it, in that I think it is

easier to answer in the aggregate, because that is

where we still have the stronger significance, but

if I am not mistaken, when you get down to

individual drugs or individual trials, although the

numbers may be higher, the relative risks may be

higher in the drug versus placebo group, it doesn't

reach the levels of statistical significance.

So, I would also like to have some input

from our statisticians on how we should approach

the individual trials or studies.

Dr. Katz first.

DR. KATZ: I just want to reiterate that

that is exactly what we are asking. We are asking

about the individual drugs. We want to know, the

252

numbers are small, the estimates are variable, none

of them really, for the most part, are

statistically significant on their own, but you

have already heard two sponsors with different

drugs.

One said everybody ought to get the same

label. One said the labels ought to be

drug-specific. So, we anticipated that outcome by

asking the question as we did.

DR. TEMPLE: We are asking for your best

interpretation of the data. We already know each

of the drugs is different, each of them can't be

considered statistically significant, but in the

face of that, given the whole data, what do you

think the best interpretation is.

DR. GOODMAN: I think, generally, when I

hear a class, I am thinking chemical class rather

than particularly how they are used. There is

certainly a great deal of similarity among the

SSRIs in terms of they all share high affinity for

the serotonin transporter.

When you get to the different

253

antidepressants, there is some variability. Some

have direct interactions with the serotonin

receptor, some do not, but then some experts in the

mechanism of action of antidepressants might argue

it doesn't matter what their acute receptor binding

profiles look like.

It has to do with what changes they induce

in the nervous system during chronic

administration, and some would argue that a

commonality or changes in serotonin system is

independent of the beginning.

So, I think we could get very much bogged

down on exactly what we mean by chemical class, so

perhaps, I think we are going to have to do it by

individual drug and maybe by indication, but I

would be open to other suggestions.

Dr. Pine.

DR. PINE: I guess I have a couple

thoughts, two main ones. Looking at the known

effects on brain neurochemistry of all the

medications that we have before us, one of them is

definitely a bit of an outlier in that Wellbutrin,

254

by most accounts, has clearly different chemical

effects than all the others, and I think that is

the only strong point I would make in that regard,

number one.

Number two, I think a number of people,

both yesterday and today, said the following, which

I would agree with both from the FDA and also on

the committee, that I think a lot of people had a

reasonable sense that fluoxetine was the one

medication for a lot of reasons that, you know,

might not have this effect, and yet we see the data

from the TADS trial that suggests that might be the

case.

So, to the extent that you are really

going to force us to say anything specific about

any medication, at least me personally, my feeling

was that the only feeling that one might have had

coming into the meeting was that the outlier,

besides Wellbutrin, would be fluoxetine, and I

think at least with respect to fluoxetine, the data

from the TADS trial, you know, takes that away at

least from my opinion.

255

DR. GOODMAN: Well, maybe this raises also

a question. We said that we were going to focus on

the clinical trials. Does that mean we should not

include the data from the TADS study? That's the

Plus 1. You were right, Tana, we should have been

very explicit. That's the Plus 1. I was

forgetting that that was the Plus 1.

Dr. Fant.

DR. FANT: You spoke to the question in

terms of defining what do we mean by class and how

to address the drug issues, because, you know, one

looks at Wellbutrin, but if it was an SSRI, I might

be inclined to be biased in a direction of safety

to sort of lump it in with the others, and look at

it as a class effect, but I am not sure if I am

willing to sort of roll that in without any input

from anyone else to tell me that that's off base

with the effects that we are seeing with Effexor.

DR. POLLOCK: Right, exactly, and going

the other way, Remeron is clearly not an SSRI also,

and we have data on that.

DR. GOODMAN: You have to wait until you

256

get called, because I have got other people waiting

here.

DR. POLLOCK: I am sorry.

DR. GOODMAN: Dr. Nelson.

DR. NELSON: I think we, in approaching

this question, need to be clear. Are we answering

this question as an interpretation of data issue

where you could simply take out the slides that

were provided and look at confidence intervals,

much as what we did as opposed to do we think

regardless of the data, we should apply a class

label for warning against suicidality, however that

is defined, which is really Question 4.

In answering this question, I think we

just need to be clear that it is an interpretation

of data. Even if I said that the data doesn't

support it for one drug, I may still support a

class warning. In the interests of efficiency, I

think we just need to run to get to the real

question, which is what to then do.

DR. FANT: Again, how are you defining

class, are you defining class as "antidepressant"

257

or chemical class?

DR. NELSON: All of the drugs we have been

talking about. I mean I am not a psychiatrist, so

all of the SSRIs.

DR. FANT: The reason I asked that is

because, like the Chair, I mean when I think of

class, I think of class based on mechanism of

action as opposed to therapeutic.

DR. NELSON: Correct. Simplistically, I

think of class when I go into Hippocrates or My

Palm, and it says SSRIs and has a name next to it.

That is how I think of class.

DR. GOODMAN: But there is overlap. The

point I was trying to make is that there is some

overlap.

Certainly, there are some differences and

that the SNRIs, like venlafaxine, also have potent

effects on the norepinephrine system, but they

share, they overlap, at least at some dosages, have

high affinity for the serotonin receptor, or we

don't understand exactly how bupropion works.

What I was alluding to also is their acute

258

properties may not be as relevant as what their

impact is on the adaptation of the nervous system

during chronic administration, so there may be some

independence between the initial effects and the

ultimate final pathway of the effect, because

obviously, the nervous system is functionally

coupled. These are not distinct systems for the

most part.

Dr. Marangell.

DR. MARANGELL: I think all those comments

are quite valid. I imagine that many people group

SSRIs together and will probably want a class

statement of SSRI Yes/No, antidepressants Yes/No,

and then whether or not you want to break out SNRIs

and Remeron and Wellbutrin as others.

DR. GOODMAN: I would be comfortable with

that approach.

Dr. Perrin.

DR. PERRIN: I would encourage that

approach. It seems to me, looking at the data,

that the ones that raise questions to us--to me, I

am sorry--are Wellbutrin and the nefazodone data

259

where there are basically no events.

You might argue that these are really

different drugs in that context, but my sense is

the Wellbutrin one, probably simply because this is

only kids with ADHD, and there are no kids with

depression in that population, at least to the

degree we can define it, we can't define it very

well. There certainly could have been some kids

with co-existing depression.

The nefazodone, I would like to have us

understand more about it. That is why I have asked

about it, but my guess is it is also a

population-based finding that has nothing to do

with the drug, because there were no events in the

placebo group either.

DR. GOODMAN: Dr. Gibbons.

DR. GIBBONS: I think that really this

ends up being a statistical issue. Dr. Hammad has

shown very clearly that these studies, even

combined within drug classes, are insufficient to

have reasonable power of rejecting the null

hypothesis for even a fairly major effect. You

260

know, we are out at about a risk ratio of about 4

to have reasonable power, and so I really don't

think that we have the data to be able to make

drug-specific statements, period.

Now, if the committee wants to make

statements that there is clear heterogeneity among

the effects across drugs, and even point to those

drugs that show less of a signal than others, that

seems totally reasonable to do, but to use these

limited data for a particular drug to make an

informed decision about whether or not this already

small signal has anything to do with one drug, but

not another drug, I think is reaching beyond the

available data.

DR. GOODMAN: Dr. Malone.

DR. MALONE: Yes, I agree that we have to

look at all the drugs, and I think if you look at,

say, the difference between the TADS study and the

other studies, I think the other studies were not

set up to look at suicidality very specifically,

but my impression was that the TADS study did look

at it more systematically.

261

When it was looked at more systematically,

you came up with a finding in fluoxetine that you

didn't have in the less systematic studies. So,

missing a signal or having a lower signal might

really just be ascertainment, and I think for that

reason, you have to look at it as a class.

DR. GOODMAN: Could somebody remind me, in

the analysis of SSRIs alone in major depression,

did that reach the level of statistical

significance for showing elevated risk level?

Could you please come to the microphone,

Dr. Hammad.

DR. HAMMAD: Yes, it did. I can get you

the actual number. Yes, the overall risk for

SSRI/MDD, it was 1.66, and the confidence interval

was 1.02, the lower limit, and the upper limit is

2.68.

DR. GOODMAN: So, that included

fluoxetine, paroxetine, sertraline. What am I

missing? It wouldn't be venlafaxine, it's SSRI,

right? Citalopram, I am sorry, citalopram.

Let me go back to Dr. Gibbons for a

262

moment. Given that, assuming we were just voting,

not voting, but we were just commenting on a class

of SSRIs in major depression, would you be

comfortable drawing a conclusion based upon the

data we have?

DR. GIBBONS: I think you can make the

statement within this class, you have reached

statistical significance, but I don't think you

have the data to make the statement that among the

other drugs you don't have statistical

significance. So, that's the rub.

Again, I think in all of this, you have to

explain as clearly as possible what are the

limitations of the data, so that you are not

misinterpreted as saying there isn't an effect or

there is an effect.

DR. GOODMAN: Dr. McGough.

DR. McGOUGH: I was taught 20 years ago

when we were in the tricyclic era that when you

initiate treatment of depression, there is an

increased risk of suicidality, and I think since

the SSRIs don't cause cardiac arrest when you

263

overdose, everybody forgot that need, and that is

one big problem.

As many other people have said, I think we

don't have the data based on these small, miserable

studies to say that they are safe.

The last thing that I thought of, we had a

strong argument yesterday for differential

labeling, and I think companies go to great lengths

once things are marketed to show an advantage of

their drug over their competitors, and there are

always pretty much sham studies that are set up, so

drug reps go around that can say one is better than

the other.

I don't want to let the wolf into the

henhouse by letting any company say that since my

drug hasn't been shown to cause suicidality, there

is an advantage to it compared to that other drug

over there. I think that would be a terrible,

terrible mistake.

DR. GOODMAN: Dr. Fost.

DR. FOST: Just a point of order. I am

not clear whether we are discussing Question 3 or

264

Question 4, and I want to second Skip Nelson's

suggestion that we focus on Question 4, because if

there is agreement on that, then, I don't think

there would be much value in going through it drug

by drug, condition by condition, unless somebody

wants to subtract--

DR. GOODMAN: You may be right. Let's all

take a moment to look at that.

I think the heart of the next question is

what recommendations we are making. Isn't that

really the thrust of No. 4, is the regulatory

recommendations.

DR. KATZ: Yes, I think including, more or

less, some specific recommendation. We don't need

exact language, but, in general, what concepts

ought to be conveyed in labeling, and then, of

course, any additional regulatory actions besides

just changing labeling are on the table.

DR. GOODMAN: I still think we need to

answer Question No. 3.

DR. TEMPLE: I think you are having the

right discussion about 3. Three is asking you, in

265

the face of limited data, what is the best

interpretation of these results. We already know

you don't have enough data on Effexor or any

individual drug, we knew that, and if that was the

question, we wouldn't have asked you.

The question is in the face of these

limitations, what is the best interpretation, and I

think you are having a good discussion of that

question.

DR. KATZ: But for our purposes, it is

useful if you are going to say something like,

well, we believe the findings generalize to all the

drugs, it would be useful to have some comment on,

for example, some of the drugs that have no events.

It would be useful to consider why you think those

should be included, as well.

DR. GOODMAN: Dr. Nelson.

DR. NELSON: I guess as a summary

position, I certainly don't see any reason to

question the data that has been put before us, and

I would probably just follow the confidence

intervals as a fair neophyte.

266

So, I have to defer those to my

psychiatric colleagues about the other drugs

without events and how those may or may not be

included. That is really an issue that I wouldn't

be able to address, but I think if we are going to

just follow the confidence intervals, like most of

us did last time, we should just sort of say that

and then move on.

DR. GOODMAN: Could I ask Dr. Hammad or

Dr. Laughren, which of the medications were free of

a signal, just remind us?

DR. LAUGHREN: Do you have my slides from

this morning? Slide 8 shows that there were no

events in the Serzone, the two Serzone trials.

These were trials in major depression. There were

no events in the Wellbutrin, the one Wellbutrin

trial, which was an ADHD trial.

DR. GOODMAN: Dr. Gorman.

DR. GORMAN: As one of the least

sophisticated statisticians around the table, I

harken back, as someone else did, to their early

training. I once asked a statistician what to do

267

with a zero numerator, and they said whenever you

see a zero numerator, you can always write a 3 in

there, because mathematically, it works out that

way. Don't ask me for the mathematical proof, I am

sure someone here can do it for me.

So, even in the small trials, when there

is a zero numerator, I think we can do some

interpretations. I will bring the reference for

the next committee meeting.

DR. GOODMAN: There are special

circumstances surrounding the Serzone trial that

could explain it besides the drug itself. Are

there any that could explain that outcome?

Dr. Malone.

DR. MALONE: I think we have already

talked about these studies not being designed to

look at suicide, so ascertainment could have been

different in that study than any other study, and

lack of ascertainment could be the reason they have

no events. It is really hard to know.

DR. GOODMAN: And the Wellbutrin study was

in ADD, wasn't it?

268

DR. MALONE: Yes, and I don't think in

ADHD, I am a child psychiatrist, I am not sure that

suicidality becomes a clinical focus, so in the

visits, it may not have been asked about as much,

or even paid attention to as much. I am not

surprised that in the Wellbutrin you didn't have

any events.

DR. GOODMAN: Dr. O'Fallon.

DR. O'FALLON: I looked in the back of

this book. We didn't see all those follow-up

slides, but I looked at them last night, and they

are rather useful. On page 35, the diagram, for

SSRIs, as a class in the MDD trials, and those are

the four, and you take a look in here, and it comes

up with the right confidence interval down at the

bottom.

But you can take a look and see that the

confidence interval for the whole class just barely

clears 1, so we are looking at a 0.05 level here.

Actually, the verification using the other modeling

doesn't quite even--it kind of takes away from that

a little bit.

269

DR. GIBBONS: I believe the random effect

in this case a little more than the fixed effects.

DR. O'FALLON: So, that puts the 0.05 up a

bit, like 0.052, or something like that, but at any

rate, you can take a look at that. It does show

you where your signal is in the SSRI trials in MDD.

If you look on the next page, at the top,

there is a similar diagram for I believe it's the

SSRIs in the other indications, and you can see the

signal there. That signal fails to be significant

in suicide, and that is on the key endpoint that

you were talking about, 1, 2, and 6 as the

endpoint. Does that help?

DR. GOODMAN: Dr. Marangell.

DR. MARANGELL: A couple of points in

response to recent comments. One is what is

different about Serzone. Perhaps it's not the

methodology. I mean you could hypothesize that

mechanistically, Serzone has a 5HT2 antagonist

which has been at least theoretically associated

with a decreased early agitation and early anxiety

realm of side effects, but pragmatically speaking,

270

the manufacturer has stopped producing that drug

because of other issues.

So, in terms of the use of our time, it

might be of academic interest, but I am not sure it

is going to make a clinical difference to the

people that we are trying to help.

The other point is that one of the things

I think the TADS trial very clearly indicates is if

we had the same discussion prior to that single,

relatively small, but very helpful study, we would

have said fluoxetine looks like it doesn't have a

signal.

So, the point is that one very small group

of patients can dramatically alter these numbers

that we are talking about. So, I think to finally

dissect them at the level of whether there is or is

not a statistically significant signal with these

very small numbers is likely to lead us in the

wrong direction.

The public and the FDA want something

about SSRIs versus all antidepressants, if you just

look at statistical significance of that, you could

271

probably pick any four of these drugs at random and

come up with something similar, I mean since you

have drugs that have a signal and don't within that

same class.

DR. GOODMAN: As it has been discussed, in

fact, this led to at least one of the abstentions,

and also some of the other comments, is that even

in aggregate, there are limitations in these data.

When you break it down to the individual drug, the

numbers get vanishingly small.

It would seem to me, my sense at this

point, it would be premature to identify a

particular drug that should be exempted from this

warning, the reason being, in part, that if we were

to exempt one, it would conceivably have the

unintended consequence of steering traffic in that

direction prior to us having sufficient knowledge

about the true risk, and we may inadvertently then

learn that there was a risk there at our next

meeting.

So, I think given the statistical

concerns, the small numbers, and my own clinical

272

impressions, that for the most part, when I have

seen at least--thank God I haven't seen

suicide--but I have seen suicidal behavior,

suicidal ideation, I have seen the activation

syndrome, for the most part, I have seen it with

most, I am not saying all, but with most of the

antidepressants, and I have not seen it limited to

major depression.

I have seen it in the treatment of

children with OCD, and there is, in fact, evidence

of that in these trials, that it occurred in the

OCD patients, as well, with fluvoxamine, although

the numbers were very small.

So, unless I think there is a very good

reason for us to do it, I think we are best off

talking about the class, not on a chemical basis,

but as antidepressants used in the pediatric

population.

But you can take shots at that position

now, but that is where I am leaning.

Dr. Pfeffer.

DR. PFEFFER: I would agree with that, and

273

I would just like to highlight again, on the blue

pages that we have, two issues that I would like to

highlight.

First, many of the studies did not exclude

a family history of bipolar disorder, and I think

that is an important design issue that we need to

keep in mind in even considering comparing, first

of all. In relation to Serzone, to be specific,

interestingly, they did exclude a family history,

but they also excluded history of a suicide

attempt.

So, there may be other issues besides the

chemistry, as you are saying, but could be

confounding this. When you put everything together

as a class, I think that is an interesting and

important issue, because it may say to practicing

clinicians, be wary of the SSRIs in general, but

there may be specific populations, then, that they

need to identify, for example, family history.

DR. GOODMAN: Dr. Pollock.

DR. POLLOCK: Also, even mechanistically,

mirtazapine also has a 5HT2 blocking effect, and it

274

is clearly in our group, and I am concerned about

the sloppiness of saying SSRIs, because

practitioners will say, you know, in a telegraphic

fashion, that, well, does that mean since I know

that mirtazapine and Effexor are not SSRIs, that

that might be safer.

So, I just second your idea that if we are

going to do it, then, I think on the basis of the

available evidence, I would rather not see it as,

quotes "SSRIs," unless you are entirely explicit

about the other couple of drugs, or if you just say

antidepressants, my preference is for the class in

that terms rather than trying to make it

mechanistic.

DR. GOODMAN: Dr. Ortiz.

DR. ORTIZ: My comment is that I also

agree that I think we don't have enough information

to say that particularly the bupropion is safe and

that we should stick to antidepressants. I am

wondering if we are not moving into kind of

research design recommendations, and if we

shouldn't at this point. It seems like we have

275

talked enough about 3 and 4 to vote on them.

DR. GOODMAN: Just procedurally, I am not

sure we need to take a vote on this. We haven't

been asked to take a vote on it, but I am open to

discussion.

Dr. Katz.

DR. KATZ: I think it would be useful to

have a vote. I mean the sentiment, at least the

few people who have expressed one explicitly, seems

to be that this signal should be considered to

apply to all of the drugs, for all of the

indications.

But I think it would be very useful for

us to actually get a vote on that particular

proposal.

DR. GOODMAN: Would you like to pose the

question, Dr. Katz?

DR. KATZ: Well, I think I sort of did.

DR. GOODMAN: We need to hear it again.

See if you can do it again.

DR. KATZ: I will see if I can do it under

pressure now.

276

DR. GOODMAN: We will give you a few

minutes.

DR. KATZ: Should the signal of increased

risk apply to all drugs studied or all

antidepressants including all indications studied,

words to that effect.

DR. TEMPLE: You actually proposed it. I

don't remember your exact words, but probably

someone does. I think you said something like the

best interpretation of these data is that it should

be applied to all drugs used for pediatric

depression and other conditions.

DR. GOODMAN: All antidepressants.

DR. TEMPLE: All antidepressants when used

for all of these conditions, not that we know that

to be true obviously.

DR. GOODMAN: Psychiatric conditions. We

have heard I think some examples from the open

public forum where it was used for non-psychiatric.

It may be when used in the pediatric population for

all indications, and, of course, there are very few

indications.

277

DR. TEMPLE: All the studies were for

indications pretty much that the drugs have.

DR. GOODMAN: Okay. So, we will limit it

to the indications that are under study, all the

indications, psychiatric indications.

Would somebody put that question together

for me?

In the meantime, Dr. Marangell, as we try

to draft it.

DR. MARANGELL: I was going to try and

draft it for you.

DR. GOODMAN: Please.

DR. MARANGELL: I move that we vote that

the committee's opinion is that the increased

suicidality, as previously defined, pertains to the

use of the nine antidepressants listed for all

indications studied to date.

DR. GOODMAN: I like that.

DR. McGOUGH: Can I ask a question? I

don't want to lose sight of other drugs approved

for depression. There are tricyclic

antidepressants, there are MAOIs, and I think part

278

of the decision is do we include all those, as

well, and is this a general recommendation for any

drug that is indicated for depression.

DR. PINE: Can I make a comment, too,

about both of those comments?

DR. GOODMAN: Dr. Pine, go ahead.

DR. PINE: I would feel more comfortable,

and again maybe nobody agrees with this, but I

would be interested in your thoughts, if you made

the statement more of a negative, since I think it

more accurately reflects the data.

In other words, none of the agents should

be excluded from this warning, because I feel more

comfortable and can confidently make that

statement, whereas, when you make the statement

that it applies to all, you know, particularly

agents where there is no event, I am kind of left--

DR. GOODMAN: I think that is the

corollary, but I think it would be hard to

translate the corollary into practice.

DR. PINE: The first statement was

something about a warning that we just voted on

279

already, Question No. 2, and then the second one

was no agent should be excluded from this warning--

DR. GOODMAN: Let me return to Dr.

McGough's comment for a moment in terms of whether

we should be expanding our considerations to

tricyclics and MAOIs. I don't think we can because

if we look at the history of this process today, we

are focusing on the clinical trials, although I

would agree with you that based upon clinical

experience, one would suspect similar kinds of

problems with tricyclics.

In fact, we saw that in one of the studies

based upon the British sample showed no significant

difference in relative risk between dothiepin,

which is a tricyclic.

DR. McGOUGH: I think there are actually

other reasons in addition not to use those other

classes, and I don't even think they are used very

commonly.

DR. GOODMAN: Because of cardiovascular

concerns, yes. I hear your point from a clinical

standpoint, but I think based upon the data under

280

consideration, I would agree with Dr. Marangell's

rendering of the question.

Other comments? Dr. Newman.

DR. NEWMAN: I like the phrasing except

that I wouldn't restrict it to when the drugs are

used for the indications for which they were

studied, because we heard of one girl who got one

of these drugs for migraines, and I would just say

when given to pediatric patients, and not restrict

it to the indication studies, because, you know,

also, people are getting it for sleep.

So, I would just make it very broad, when

given to children, they can increase suicidality.

DR. GOODMAN: I have to agree with that.

Dr. Gorman.

DR. GORMAN: Agree.

DR. GOODMAN: Dr. Gorman, you agree also?

DR. GORMAN: Agree.

DR. GOODMAN: Tom.

DR. LAUGHREN: Just a thought. Clearly,

when you are moving from Questions 1 through 2

through 3, going from 1 to 2, clearly the focus is

281

on the 24 trials that we looked at. I think as you

move towards 3 and 4, I personally don't see any

problem.

I mean if you have reached a conclusion

that you can expand this claim to the

antidepressant class, in other words, you are

willing to ignore findings of no events for certain

trials, I don't think it is so unreasonable to

consider expanding it to the whole class, and here

is my reasoning for that.

I think there is a great risk in steering

clinicians back to using the tricyclics as an

alternative to a safe group of drugs, which all of

us know are not, so it is just something to think

about.

DR. GOODMAN: Dr. Marangell, could you

restate that question now including all

antidepressants in the pediatric population, could

you try it?

DR. MARANGELL: I can try. I move that

the committee adopt the position that

antidepressant agents, when given to pediatric

282

patients, defined 7 to 17 years old, increases the

risk of suicidality as previously--no?

DR. GOODMAN: Let her finish and we will

get back to the age range.

DR. MARANGELL: I move that the committee

consensus is that the risk of suicidality, as

previously defined, applies to all antidepressants

when used in pediatric patients.

DR. GOODMAN: I like that. Was there a

desire to not qualify what we mean by pediatric, or

to qualify differently?

Dr. Leslie.

DR. LESLIE: I just feel that 7 to 17 is

not a good direction to go in.

DR. GOODMAN: So, it is sufficient to say

pediatric?

DR. LESLIE: And I just wanted to add the

other reason I would say that this is important to

say broadly is we have got Cimbalta and other

things coming out, and you don't want someone

saying, well, Cimbalta doesn't have this warning,

so you ought to use it instead of Luvox, et cetera.

283

DR. GOODMAN: Did anybody transcribe that?

I am comfortable with the statement. They are on

it. We are about to project it.

Dr. Katz.

DR. KATZ: I would just like to hear a

little bit more discussion about applying the

warning, whatever it is that we apply, at least

applying the result to non-psychiatric indications.

We don't have any trial data in that

population. We have reports of individual cases,

and clearly they are used for other things, but I

would just like a little more discussion or hear

what people think about extending it to all

possible indications for which these drugs might be

used.

DR. GOODMAN: Once Tom gave us some

freedom to not confine ourselves to the clinical

trials, I think led us to consideration of all

antidepressants and all possible indications, but,

sure.

Dr. Nelson.

DR. NELSON: In looking, for example, at

284

the current labeling, I think the restriction to

major depressive disorder is potentially falsely

reassuring, and it would concern me if it was

listed only for psychiatric conditions, that the

same thing would happen with the off-label use in

non-psychiatric conditions.

Even though there is no data suggesting

that, by having a warning associated with the drug,

and not with the condition, the way a clinician

could possibly read this would be here, I am giving

you this drug, but since you are not depressed, it

is not going to happen in you, so therefore, it is

safe, and we can't say that.

That is the way it would be read, so I

think by not listing it to the drug, you open up

that possible interpretation to a clinician, which,

in the absence of evidence, I think would be a

danger.

DR. GOODMAN: I agree completely, and I

think, although we haven't discussed this at great

length, and I don't know how many people would

agree with this statement, but I think at least I

285

do, and I have heard some other mention the fact

that our hypothesis of the mechanism here is some

sort of behavior toxicity that may be compounded by

an interaction with an underlying proclivity, such

as bipolar diathesis.

It may have to do something about

metabolism or drug levels. There are a number of

other factors that can contribute, but once I saw

the data in the OCD trials, although it is only a

few cases, I began to think, and also based on my

own experience, that this isn't strictly worsening

of depression or ineffectiveness in treating

depression, especially since in a number of the

cases, it seems to happen so early in the course.

So, I would agree completely that I think,

in part, our recommendations reflect a working

hypothesis that what we are seeing, although a rare

event, may represent behavioral toxicity that can

occur in individuals other than those already

diagnosed with depression.

We have the revised question up on the

screen--no, we don't.

286

DR. TEMPLE: That is my revision, it is

not exactly the same as others, and you don't have

to take it.

DR. GOODMAN: We are going to have to

change it.

Dr. Marangell, what do you think?

DR. MARANGELL: If you say the best

interpretation of the results of the 23 plus 1,

then, it would be the indication studied. If you

say what is our, you know, kind of interpretation

of where we go with those results, then, it is a

little bit broader.

DR. GOODMAN: I would like to go with

broader. I don't know if we can change it on

there.

Other comments? Dr. Chesney.

DR. CHESNEY: When you introduced applying

this principle to all antidepressants, this is a

whole new ballpark for me, because I don't know

anything about them, and we don't have any data,

and maybe I misunderstood, but if I didn't

misunderstand, could somebody explain to me why we

287

should extend whatever we decide about what we have

heard about, to things that we have heard nothing

about?

DR. GOODMAN: What is it that we haven't

heard about?

DR. CHESNEY: All the other

antidepressants. What I heard was that we extend

this, not just to SSRIs, that we know about, but to

all other antidepressants, and I don't know

anything about them.

DR. GOODMAN: We have covered most of the

field, but I would let others--

DR. MARANGELL: There are two areas of

extension in regard to the revised Question 3.

What I was actually just referring to was the

extension to pediatric use or any use in pediatric

patients as opposed to just the uses that were

studied, and the rationale for that portion is, you

know, clearly it is beyond major depression for

those of us that think that there is a consistent

signal, you see that also in some of the non-major

depression indications, such as the OCD trials.

288

So, that is the reason to extend that, and

then the feedback we got, that I heard from the

rest of the committee was that even beyond those

trials, there may be other indications, and we

don't want to try and give the signal that it is

limited to, you know, if a doc wants to use an

agent for a pain condition, that we wouldn't expect

to see it there.

In terms of other antidepressants, the

tricyclics and the MAOIs are older agents, and my

understanding is that they have never been studied

in pediatric patients, so we have no data, is that

correct?

DR. TEMPLE: It is not a bad idea to take

those questions in sequence. First, think about

the drugs that were studied and what is the best

interpretation of them, and then, you know, we

might approve a new antidepressant, and is it going

to be the only one that doesn't bear that label?

Do you like that idea? We will get to that.

DR. LAUGHREN: Actually, a clarification.

Some of the older drugs have, in fact, been

289

studied, but FDA has not seen the data, they have

never been submitted. There are at least 12 trials

in tricyclic antidepressants in pediatric

depression that I am aware of, but I don't know how

well those patients were ascertained or

suicidality, I know nothing about the safety in

those trials.

DR. GOODMAN: Dr. McGough.

DR. McGOUGH: Again, I think when you

treat depression, there is an increased risk of

suicide. Most of this is off-label, so to restrict

it to indications is kind of an illogical event,

that somebody would get Zoloft for insomnia is

beyond anything I can understand.

So, I think really the purpose of this, I

think is to really put physicians on notice that

this group of medicines can cause these problems

for whatever they think of. To be safe, I think we

are really putting the physicians, you know, the

real attempt is to inform them that they need to be

concerned, they need to monitor for whatever they

are using this for.

290

DR. GOODMAN: Dr. Robinson.

DR. ROBINSON: One of the things that I

took away from a lot of the public testimony

yesterday was how much the people said they wanted

to have known things before they got into a

treatment.

I think one of the things that we have to

deal with is that we do not know a lot about some

of the other drugs, like the tricyclics, the MAOIs,

we just don't have the information because they

were done so long ago.

Now, I personally think that it would be a

great tragedy if clinicians went from using SSRIs

to tricyclics, which have a much lower overdose.

You can kill yourself taking much, much fewer

pills.

But I think we also have to be humble

enough to say, you know, there are certain things

we don't know. We don't know about the risks for

tricyclics, we don't know about the risks of SSRIs

used for these off-label non-psychiatric

indications. We just don't know, and there is no

291

information.

In some ways, that is the most accurate

thing to tell a prospective family thinking about

these drugs, is we know for certain indications, it

has an increased risk of suicide. For this stuff,

we don't know, it might be a lot worse, it might be

better, we just don't know, but take this into

consideration.

DR. GOODMAN: Before I get to you, Dr.

Temple, I actually don't think it is such a bad

question. I think that by answering this question,

it allows us to put Question 3 behind us, and then

I think a lot of the discussion that we are having

is then what should we recommend in terms of

further regulatory action, and I think that is

really getting it to Question 4.

Would that be a reasonable way of sorting

out where we are right now, to try to maybe answer

the one that is up there, and then we can talk

about whether, in our recommendations in terms of

further warnings, that maybe that issue should be

expanded further, but this would be a good

292

transitional question, as I see it.

Dr. Temple.

DR. TEMPLE: There really are two parts.

The first part in some ways is I think what most

people have given their answer to that. The second

part I just was writing quickly referred to

tricyclics, but it is worth thinking a little bit

here about whether the best interpretation is that

it applies to some study, some new drug like we

just approved one that hasn't been tested yet.

We are going to surely have to come to

grips with that question, too, so it says

tricyclics, it could have said MAO inhibitors, it

could have said duloxetine.

That is the second question, but I think

that follows the first.

DR. GOODMAN: I would like to take off

after the second bullet for a moment. Unless there

is further discussion, I would actually like to

vote on that question that is up there.

Dr. O'Fallon.

DR. O'FALLON: Again, on page 34 of the

293

thing we got yesterday, there is a nice diagram

that shows the relative risks and their confidence

intervals for all of the nine drugs, all of the

indications, and what you will see, when you look

at it, is that every drug has at least one trial

that shows an increased drug.

You can make the argument that there is

some evidence for every single one of these.

DR. POLLOCK: If we are saying the nine

drugs, it includes Wellbutrin, which isn't one of

these.

DR. O'FALLON: Perhaps, but what I wanted

to point is that every single one of these drugs

has at least one, some more, that show an

increased--okay, eight, whatever--it says all

drugs.

DR. GOODMAN: Unless there is further

discussion, I would like to put this to a vote.

The reason, why do I want to answer this, because

it is our job to try to answer the questions that

were presented before us.

I think that if we don't answer this, it

294

is not clear from our last vote whether we were

thinking of just one drug or many drugs. So, this

is the natural sequence that we are making it clear

by this that we are talking about in aggregate when

we look at the data. We are not exempting any of

the drugs for any of these pediatric indications.

So, I think we need to answer this

question, and it should be put to a vote.

Yes.

DR. FANT: Just another way of phrasing

that which may be more palatable. If things are

just transposed a bit, to say is the best

interpretation of the results of the 23 plus 1

trials, that none of the drugs examined can be

excluded from the increased suicidality risks that

has been shown da-da-da.

DR. GOODMAN: Let me ask a statistician

and then a wordsmith. Any statistician want to

weigh in on that?

DR. GIBBONS: I don't think you want to

draw an inference beyond the data that you have.

DR. GOODMAN: So, are you supporting the

295

way it is currently phrased or suggesting it be

different?

DR. GIBBONS: I am supporting the last

statement.

DR. FANT: Basically, what it says is that

the data--it is more difficult to say the data

applies to all nine, and for me, it is easier to

say that based on the data we can't exclude any of

the drugs.

DR. GIBBONS: The non-exclusion, yes.

DR. GOODMAN: I would find that more

palatable, too, I would agree.

Does somebody want to try that?

DR. NEWMAN: I will take a stab.

DR. GOODMAN: Okay, we are working on it.

In the meantime, we will entertain other comments.

Dr. Newman.

DR. NEWMAN: I guess I will have to see

it, but I am concerned that that does sound quite a

bit weaker to say that we can't tell for sure that

this drug doesn't cause suicide or suicidality. It

is true, we need to apply this warning to the whole

296

class, but there is sort of a double negative

there, and I think it would be clearer to say the

warning applies to the whole class.

DR. GOODMAN: What I was suggesting before

is that this isn't the warning per se. I think we

can still construct the warning. I think it would

be more along the lines of what Dr. Marangell had

drafted earlier. This is just to answer the

specific question to indicate that we are not

talking about just one drug and then we can get

beyond Question 3.

Lauren, do you have something?

DR. MARANGELL: The data in aggregate

indicate an increased risk of suicidality.

Although there is variability in the results, we

are unable to conclude that any single agent is

free from risk at this time.

DR. GOODMAN: That's good. Do you want to

do that one more time? You may get the Chair of

this committee soon, you know that.

DR. MARANGELL: No, thanks.

Okay. Data in aggregate indicate an

297

increased risk of suicidality as previously

defined. Although there is variability in the

results, we are unable to conclude that any single

agent is free from risk at this time.

DR. GOODMAN: Give it a chance to be keyed

in.

DR. TEMPLE: From our point of view, any

of those are fine, because they point directly to

what the next question is going to address, so it's

okay.

DR. LAUGHREN: As long as it includes a

mention--it would be better to see it in writing,

but I didn't hear mention of pediatric patients in

all indications.

DR. GOODMAN: We are working on it over

here.

Tana.

DR. GRADY-WELIKY: Along those lines, I

wanted a point of clarification about are we

talking pediatric use indications or are we talking

pediatric use?

DR. GOODMAN: We are talking about the

298

studies. We are back to talking about the studies.

In a sense, this is the second part of the previous

vote, and what we are indicating here is that are

affirmative as already voted by the majority of the

committee, indicates concern about all the drugs in

aggregate or at least that we can't exempt any one

of them individually. Otherwise, as left, the vote

could look like we were just concerned about one of

the drugs.

DR. MURPHY: You could just say the data

in pediatric patients.

DR. GOODMAN: Dr. Irwin.

DR. IRWIN: I just wanted to be certain in

terms of the definition of pediatrics, what do we

include with that, because it can vary. Does FDA

have a definition of that?

DR. MURPHY: Yes, we have a definition

that allows you to be very flexible depending on

the state and country from which your data is

derived. Again, later, when we get to the warning,

we can be more specific, but I think right now the

phrase "pediatric data" would be sufficient.

299

DR. GOODMAN: Isn't the NIH definition up

to 21?

DR. PINE: It matters what it is being

applied for, less than 18 for some definitions, and

less than 21 for others.

DR. MURPHY: Ours is not up to 21.

DR. GOODMAN: Yours is not up to 21?

DR. MURPHY: It is up to 18 in some

guidances.

DR. GOODMAN: In these particular studies,

were they all up to, but not exceeding, age 18?

DR. MURPHY: Yes, they were. Most of

these studies included 17.

DR. GOODMAN: They included 17, but not

18?

DR. MURPHY: Is there an 18 in one of

them?

DR. GOODMAN: I am sorry, I can't hear.

DR. MURPHY: I think they say to 18. That

is where you get into argument.

DR. PERRIN: There were a couple of

studies where a couple kids got in over age 18 by

300

mistake.

DR. HAMMAD: May I say something?

DR. GOODMAN: Yes, please.

DR. HAMMAD: I think I had 85 patients

that were 18 years old out of 4,000, but I did have

it up to 18.

DR. GOODMAN: Up to and including 18?

DR. HAMMAD: Exactly, including 18.

DR. IRWIN: The reason I asked that is

because much of testimony we heard from the public,

a lot of that was in young adults that were beyond

their 18th birthday.

DR. GOODMAN: Again, that is for our next

statement.

Ms. Griffith.

MS. GRIFFITH: I agree. I think that is

problematic in that it varies from state to state.

There are legal definitions of how long a parent

can have control over a child in terms of what that

child will not have to agree to by way of

intervention.

So, I would even recommend we go up to 21

301

because, as of 18, in most of the states in the

U.S., children no longer have to comply with

pharmaceutical interventions, and I think a lot of

people would dismiss the advice once a child turns

18, and it is not warranted.

DR. GOODMAN: Any further wordsmithing of

the question? I am satisfied with it.

Dr. Newman.

DR. NEWMAN: I just think this phrasing is

quite a bit weaker than it was before because when

you say you can't conclude that an agent is free of

risk, you can never ever conclude that any agent is

free of risk, so this just doesn't say very much.

I would rather state it more

affirmatively. Although there is variability in

the results, we believe these results apply to all

antidepressants in this class.

DR. GOODMAN: Before you change it, we

need to have some other input on that.

DR. ORTIZ: This is not the warning. This

is what we are going to vote on, and it seems like

there is a lot of consensus that this is what we

302

can vote on.

DR. GOODMAN: Right.

Dr. Fant.

DR. FANT: If we put the comment in there

"within this class," I am just concerned, you know,

it is going to restrict it to say SSRIs, and the

whole point of this is to include, you know, to be

cognizant of potential risks in drugs that really,

as poor as the data is with other drugs, there is

even less data.

So, the increased risk, we can be as

strong as we want in affirming that and emphasizing

that, but I agree with the whole concept that no

company can enroll 70 kids and say we didn't see

anything, therefore, you guys ought to use us

instead of the other guy. I think we should not

facilitate that.

DR. GOODMAN: Dr. Katz.

DR. KATZ: A couple of things. First of

all, I think you have to say something about what

kind of agent we are talking about. I think if you

said any single antidepressant agent, we would know

303

what you mean, which brings me to my second point,

which is this is not language for labeling or

anything else, as someone said. This is to guide

us, to give us a sense of what you believe about

which sorts of classes and indications the risk

applies to.

I believe it is fair to say that if you

vote on this question, and you add the word

"antidepressant agent," we will know what you mean.

So, I don't know that we need much more

extensive discussion on fine-tuning the question.

We know what you are getting at, I believe, and if

you vote on this with just the addition of the word

"any single antidepressant agent," I think that

would be perfectly fine for our purposes.

DR. GOODMAN: That is the change I would

recommend, if we could just punch in that "any

single antidepressant agent."

DR. MURPHY: Dr. Goodman, when you go

around and each person votes again that 30-second

statement, they can make it clear if they have a

problem with the statement.

304

DR. GOODMAN: That is correct, and we are

talking about the antidepressants that were in the

trials.

DR. TEMPLE: That's correct. I thought

that is what you were referring to.

DR. GOODMAN: That is what we were

referring to.

DR. TEMPLE: So, we will have a second

question either now or in the next one, it doesn't

really matter about what to do with the drugs that

weren't in the studies.

DR. GOODMAN: Right.

That is going to be the last comment

before the vote.

Dr. Perrin.

DR. PERRIN: I am just wondering whether a

slight variation on that issue of risk could be,

and I can't quite have it in front of me now, that

we are unable to conclude that any single

antidepressant agent has particularly low risk of

suicidality at this time. It might be a more

accurate statement of where we are.

305

DR. GOODMAN: Is free of increased risk

maybe. It is free of increased risk.

DR. TEMPLE: Fine. That is what that

means. When say "free of risk," it means free of

increased risk, otherwise, it has no meaning.

DR. GOODMAN: That's it. We are going to

go for a vote, and if you have further comments

including about the wording of the question, you

can state them in the 30 seconds that I am going to

allow to be included.

Let's start at the opposite end of the

table this time, first, with Dr. Fant, could you

indicate Yes/No.

We did not add increase because Dr. Temple

said it was implied.

DR. FANT: Yes. No additional comments.

DR. PFEFFER: Yes. No other comments.

DR. FOST: Yes.

DR. ORTIZ: Yes.

DR. MALONE: Yes.

DR. NELSON: Yes.

DR. PERRIN: Yes.

306

DR. GRADY-WELIKY: Yes.

DR. GOODMAN: Wait, slow down. Let me do

the name first. I think we are eager to cast a

vote.

We have Fant Yes, Pfeffer Yes, Fost Yes,

Ortiz Yes, Malone Yes, Nelson Yes, Perrin Yes,

Grady-Weliky Yes.

Ebert.

DR. EBERT: Yes.

DR. GOODMAN: Gibbons.

DR. GIBBONS: Yes.

DR. GOODMAN: Pine.

DR. PINE: Yes.

DR. GOODMAN: Bronstein.

MS. BRONSTEIN: Yes.

DR. GOODMAN: Rudorfer.

DR. RUDORFER: Yes with a comment. First

of all, I wonder if--I guess it's too late to go

back, but--

DR. GOODMAN: You guessed right.

DR. RUDORFER: --if a clarification of the

phrase "increased risk" would be helpful. Again, I

307

think we have agreed that an increased risk is

likely to be small, and I wonder how that should be

conveyed, because "increased" covers a fairly wide

range.

The other thing, if I have a few seconds

left, just to reiterate my concerns. I agree

certainly with the spirit of this statement, that

we have seen that signal across all the drugs.

My concern relates, if I may use an

example, to the two citalopram studies we reviewed.

An American study found efficacy in major

depression in a pediatric sample, and found no

suicidality signal.

The European study combined data from

seven different countries, which I did not find

methodologically attractive, found no efficacy

compared to placebo, and did find a positive

suicidality signal.

S-citalopram, we have no data on the

related compound. So, again, my overarching

concern remains the fact that I think this is very

much still a work-in-progress.

308

Thank you.

DR. GOODMAN: Thank you.

Dr. Goodman, Yes.

DR. GOODMAN: Dr. Chesney.

DR. CHESNEY: Yes with the understanding

that this statement applies only to the SSRI agents

that we have been discussing. It doesn't say SSRI

anywhere.

DR. GOODMAN: Let me clarify. It applies

to all the compounds that were studied in the

trials, which includes several non-SSRIs. When you

add the non-SSRIs, the hazard ratio gets bigger,

for what that is worth.

DR. CHESNEY: In other words, we are

including imipramine and--

DR. GOODMAN: No, no, just the ones that

were involved in the clinical trials that we

reviewed, that were part of Hammad re-analysis.

DR. CHESNEY: Which were the non-SSRIs?

DR. GOODMAN: Venlafaxine.

DR. CHESNEY: All right. The answer is

Yes.

309

DR. GOODMAN: I am sorry. Let me re-ask

your vote, Dr. Chesney, based upon that

clarification.

DR. CHESNEY: Yes.

DR. GOODMAN: Dr. McGough.

DR. McGOUGH: Yes.

DR. GOODMAN: Griffith.

MS. GRIFFITH: Yes.

DR. GOODMAN: Leslie.

DR. LESLIE: Yes.

DR. GOODMAN: Robinson.

DR. ROBINSON: Yes.

DR. GOODMAN: Marangell.

DR. MARANGELL: Yes.

DR. GOODMAN: Irwin.

DR. IRWIN: Yes.

DR. GOODMAN: Dokken.

MS. DOKKEN: Yes.

DR. GOODMAN: Newman.

DR. NEWMAN: Yes.

DR. GOODMAN: Wells.

DR. WELLS: Yes.

310

DR. GOODMAN: Pollock.

DR. POLLOCK: Yes.

DR. GOODMAN: O'Fallon.

DR. O'FALLON: Yes.

DR. GOODMAN: Santana.

DR. SANTANA: Yes.

DR. GOODMAN: It was unanimous this time.

We had 27 respondents, all Yes.

Do people want a short break? Yes.

[Break.]

DR. GOODMAN: I think we have made a great

deal of progress. We have two remaining questions.

As currently constructed, neither of those

questions require a vote.

Dr. Katz.

DR. KATZ: Yes, it is very important for

us to have you vote on an extension of the question

you just voted on, which is whether or not this

should apply to all other antidepressants or

whether you simply want to limit whatever warning

or whatever conclusion we draw to just the ones

that were studied.

311

My understanding is that in this question,

the one you voted on, you limited it to

consideration of the drugs studied.

DR. GOODMAN: That is correct. Let's have

a discussion and see if others around the table

agree.

Dr. Gorman.

DR. GORMAN: For future drugs that are

coming down the pike, the Food and Drug

Administration, under the Pediatric Research Equity

Act, has the ability to demand antidepressant

studies in children prior to their release.

I don't think there can be any doubt after

the numbers we heard today that they will be used

in more than 50,000 patients and have the potential

to give a significant therapeutic advance, because

at this point, we only have one drug that is

approved.

So, I think for drugs coming down the

pike, I think there is the ability within the FDA

to ask for that information, and since I am not a

voting member of the committee, I will leave the

312

discussion about what to do with the previously

approved drugs alone.

DR. GOODMAN: Dr. Wang.

DR. WANG: Yes. It seems if you exempt

any drugs, new or otherwise, you set up this

perverse incentive, particularly for the new drugs,

to either do no studies or to do poorly conducted

studies where they don't ascertain cases or

underpower them, so I think to prevent that

perverse incentive, you have to put the onus on

them to show that they are the exception.

DR. GOODMAN: Dr. Perrin.

DR. PERRIN: While I might want to have a

similar strategy across all antidepressants, I

think it is really not appropriate for this

committee to take a stand against antidepressants

for which we have not reviewed the data.

I am very uncomfortable saying that we

know much about them when we really haven't seen

the data on them.

DR. GOODMAN: Dr. Nelson.

DR. NELSON: Although agreeing with the

313

prior point, my concern would be the message that

would be sent if you didn't apply this across all

drugs, and my own preference would be to have a

class risk warning and then any preferential

treatment ought to be on the efficacy side, so that

you then have drug-specific labeling under the

efficacy component, which would then begin to

differentiate, so that individuals can be informed

about the risk-benefit ratio by looking and

comparing those two sections, which would be one

way to direct people appropriately as opposed to

inappropriately.

DR. GOODMAN: Dr. McGough.

DR. McGOUGH: Just again, what we are

talking about here is patient safety, and I think

it is appropriate that we err on the side of being

in favor of that, and even though we haven't

reviewed data specifically, I think it is an

unanswered question, and as such, I think it is

appropriate to apply it generally to the class of

antidepressant drugs.

DR. GOODMAN: Dr. Laughren.

314

DR. LAUGHREN: Just to clarify what we are

planning to do with the general warning that we

have already implemented for 10 newer generation

antidepressants, we are planning on extending this.

In fact, it has already been extended to some of

the older drugs, some of the tricyclics.

Our plan is to extend it to all, and the

question here is whether or not we should also, if

we are thinking about adding new language to that

warning, suggesting that we have now established

causality, whether that new language should also

apply to all antidepressants.

Again, I think this concern has been

expressed by several members, that if you don't do

that, you are, in effect, directing clinicians to

use those older drugs.

DR. GOODMAN: Dr. Leslie.

DR. LESLIE: Again, my concern is not just

the older drugs, but the newer drugs, because

Cimbalta, which is coming out, is most comparable

to Effexor, which had the highest relative risk,

and I again would be very concerned with the

315

marketing directly to patients and the heavy

marketing to clinicians, that it would be marketed

as the only drug of this class without a label, and

would thus again push clinicians in that direction.

DR. GOODMAN: Dr. Marangell, did you have

a draft of the question that we were composing

before, that represented more of a general warning?

I don't think we keyed it in. I don't think it was

saved.

I think where we are in the discussion,

let me remind people, we are in Question 4, and we

are talking about recommendations regarding

additional regulatory actions.

Since the last meeting, warnings were

issued about a group of symptoms that may be part

of what some have labeled inactivation syndrome,

that were proven to be precursors of suicidality.

So, I think, in part, the question before

us is whether we want to extend that now to

conclude that there is a suicidality risk. In the

last warning that was issued, it said that there

was no established connection between the

316

medications and suicidality, and we have obviously

been deliberating and voting on that question.

We are really talking about what

additional warnings need to be posed that go beyond

the activation symptoms or syndrome that was

defined previously.

Dr. Temple.

DR. TEMPLE: Whether you do it as part of

Question 3 or part of Question 4, we unequivocally,

as you have already indicated, need to know whether

the warning, if it is modified, needs to be in the

labeling for all antidepressants or just the ones

that were in the study including tricyclics, MAO

inhibitors, duloxetine, and so on.

So, I don't know whether that is a 3

question or a 4 question, but once you answer the

question about any additional warning language, we

have to know who you think that applies to.

DR. GOODMAN: Dr. Malone.

DR. MALONE: I think it applies to all of

the drugs. Was it the Jick study that really

didn't find any difference between the risk for

317

different classes of drugs. I think that is one

reason to apply it to, say, the tricyclics which

were included in that study.

Also, some of the older drugs, like the

tricyclics and MAO inhibitors, are much more

dangerous, so if you did become suicidal, you would

actually have the drug with you, that you could use

to commit suicide easily.

I agree that new drugs, as they come on

the market, you wouldn't want to automatically give

them this undue edge that they don't have this

warning.

DR. GOODMAN: Dr. Chesney.

DR. CHESNEY: I would just like to say

that when I made my comment before the break, I

wasn't thinking of all these other factors, and on

reconsideration, I think that these are excellent

points.

I think that the wording was correct that

it is well recognized that suicide and suicidal

behavior emerges during the early stages after

treatment, and because we don't have the specific

318

studies that address this issue, it may be

reasonable to assume that this would apply to all

antidepressants at this point in time until proven

otherwise.

Somebody made that point, put the onus on

the company to show that they did not fall in that

ballpark. So, whereas before the break, I was very

alarmed that we were going to be asked to do some

things we knew nothing about, I think I now

understand the reasoning.

Thank you.

DR. GOODMAN: Thank you.

Dr. Pine.

DR. PINE: I think it is important to

recognize, speaking again as a child psychiatrist,

that there is a very legitimate concern about

discouraging people, and physicians in particular,

from moving away from the 23 plus 1, the agents

that we have been discussing, to tricyclics

antidepressants.

I think that, at least from my own

perspective, I think we want to think kind of

319

ahead, and I think we do obviously based on what

the FDA said, and based on other information, say

something about the use of those agents.

Historically, there actually is a fair

amount of data on those agents, although not nearly

the amount of data as we have reviewed for SSRIs,

and it is current standard of care in child

psychiatry not to use those agents, and that is

really based on two things.

That is based on, number one, the fact

that a number of meta-analyses have shown that the

agents are not effective over placebo, and there

has not been a single study that demonstrated

efficacy for a tricyclic antidepressant or an MAOI,

number one.

Number two, there was a lot of concern in

the 1990s about the cardiotoxicity of these agents

in children, not only the cardiotoxicity in

overdose, which I think there is little debate

about, but even questions about cardiotoxicity when

the agents were used appropriately in therapeutic

doses.

320

So, I think, on the one hand, it is very

important to say that it would not be good if

physicians were to move from the newer agents to

the older agents, on the one hand, on the other

hand, probably the strongest thing we could say

would not be to say, well, we don't know whether

these drugs cause suicidality or not, the strongest

thing we could say is that there are plenty of

reasons to discourage this.

One of the might be that the agents are

associated with suicidality like the SSRIs,

however, the cardiotoxicity and the lack of

efficacy data, I think are stronger reasons not to

move that way.

DR. GOODMAN: Dr. Gibbons, did you have

another question or comment?

DR. GIBBONS: This is a very tricky issue.

The issue is we don't want to go beyond the data

that we have, but on the other hand, with the

exception of one of these agents, we haven't shown

any adverse effects on a drug-by-drug basis.

Now, a new drug comes out on the market.

321

Given current regulatory practices, there will not

be enough data to show a risk ratio of 1.5 in that

drug, so the conclusion will be that there is no

association, and they may get an exemption.

On the other hand, you know, so what you

are doing is you are holding a much higher standard

to the drugs that were looked at in these 21

studies, for which we were able to pool over drugs

and show an effect.

I suppose you could set up a situation in

which a new drug or an old drug that was not part

of the 21 could, in fact, be removed from the list

if the study was powered, if there was enough data,

so that they could, in fact, identify a risk ratio

in the magnitude that we are looking at here, in

the 1.5 range or so, but they would have to get a

hell of a lot of patients to do it. It might not

even be practical to do that.

So, it is really a conundrum of what to

do.

DR. GOODMAN: Dr. Fant.

DR. FANT: To address the question that

322

was posed about extending to other drugs, I think

based on the data that we have looked at here, at

least from what I have seen it is impossible to

tell if the endpoints that we saw, if the signals

that we saw were due to drug-specific effects or if

they were due to factors intrinsic to the disease

process in the patient that was perturbed by

treatment by whatever mechanism.

It is kind of hard to sort out which is

the major factor in that regard. Until that can be

sorted out, until you have some information that

suggests, on a mechanistic basis, that it is

related to a drug-specific effect, I don't think

you can exclude any drug that impacts on

depression.

DR. GOODMAN: Take a moment and look at

the question or the statement, not a question, but

the statement as it is proposed.

DR. TEMPLE: Is that intended to be

labeling language?

DR. GOODMAN: My understanding is that it

is not our job to write the labeling language for

323

the FDA. That is one of the reasons I wasn't sure

that we needed to take a vote, because I don't

think that we should be writing these for you, but

I think it does help to have something in writing

here to communicate to the FDA how broad our

concerns are.

Dr. Temple.

DR. TEMPLE: So, is this your proposal?

DR. GOODMAN: Two committee members have

proposed this.

DR. TEMPLE: My initial response to that

is it doesn't really say that you should assume

this risk applies to all drugs. It says you should

use caution in pediatric patients. Well, we have

sort of said that already. It doesn't quite say

you should worry about that risk in pediatric

patients.

We can work on it, too, but we are

interested in some view of how explicit we really

should try to be even if you don't write the exact

words. That one is not so explicit perhaps.

DR. MARANGELL: I think we are just trying

324

to put forward, kind of the broadening concept of

including both prior agents and future agents that

are categorized as antidepressants and where to go

from there.

DR. GOODMAN: Dr. Katz.

DR. KATZ: Fine. I think we certainly get

the thrust of that, we appreciate it. One question

that I think we would like you to explicitly

address is whether or not you think this is the

sort of thing that belongs in a black box, which is

sort of another level of communicating risk.

Right now it is a warning, the major

language is in the warning--the language is another

section--but the major language is in the Warning

Section. Do you think this arises to the level of

a black box warning?

DR. GOODMAN: Let me ask Dr. Nelson to

comment.

DR. NELSON: Let me comment on that, and

then just make a comment on the language of the

second sentence. I think the difficulty with a

black box warning, at least my interpretation, and

325

maybe it is based on my practice, is if I see a

black box warning, I just don't do it, for example,

propofol for long-term sedation in pediatric ICUs,

no longer done because of the warning.

So, I think a black box warning may drive

people away from drugs that they might otherwise

appropriately use as opposed to a warning that is

placed upfront in the Warning Section.

My difficulty with this language, and I

know we are not going to talk language, but I would

try to be more specific. I mean one of the more

difficult things to communicate, I think even to

physicians, is risk data, and I kind of liked the

slide that we were provided, that could even in

this case, communicate data where you could have

something like out of 100 patients treated, 2 to 3

patients on average will have an increase in

suicidal behavior or ideation after initiation or

changes in treatment, period.

So, people say, well, what does this

really mean, it means that if they give 100 people

the drug, they have got to watch out 2 or 3 times.

326

That, to me, is useful information, and I would

advocate labeling that provides that kind of

information.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: I am just trying to interpret

the last statement, which is that we should give

some quantitative estimate of what the nature of

the risk is.

DR. NELSON: Well, it is more than that,

because I think it is an issue of percentage. I

mean there is a lot of literature on whether risk

is best communicated as numbers out of 100 versus

percentages versus other things, so I am explicitly

saying I find the most useful thing is real people

out of a real cohort of people helps me know what

the universe is.

DR. TEMPLE: I would add that we should

probably tell people what the baseline risk is and

what the increase compared to baseline is.

DR. NELSON: I would also link this to

similar kind of data under the efficacy, so you are

right.

327

DR. GOODMAN: Dr. Rudorfer.

DR. RUDORFER: A couple of questions.

First, about the tricyclic issue. My recollection

is there is some language already for the older

antidepressants related to risk of possible

clinical worsening early in the course of

treatment. I mean I am just wondering, if

something doesn't exist, or even if it could be

moved within the labeling to be comparable to

placement, say, of the newer--

DR. GOODMAN: Tom.

DR. LAUGHREN: Again, let me reiterate

what I suggested earlier. Our plan is to extend

this current, much broader warning statement that

is now only in these 10 current generation drugs to

all antidepressants including all the tricyclics,

all the MAOIs. Some of them actually already have

it.

So, you are right, there is that old

language in the tricyclics. That is going to be

changed to the newer language, and really the

question here is whether you are comfortable with

328

us extending this additional view, that now we have

established causality for suicidality to all

antidepressants, not just these nine drugs that

were studied in these trials, and pediatric

patients.

DR. RUDORFER: Do I understand correctly,

Tom, in the newer warning, that is where the

description of the behavioral activation is

mentioned, because again, what I am wondering is,

if that specific language really should be

applicable to the tricyclics.

I mean at our February meeting, we were

describing a syndrome that I think the sense of the

committees was that that was more specific to the

newer generation drugs.

DR. LAUGHREN: The same kinds of behaviors

are seen in SSRIs, in SNRIs. I think an argument

could be made. I mean obviously, there is a lot of

anecdotal data that the older drugs in some

patients have the same kinds of symptoms, so I mean

my inclination, in fact, some companies have

already done it on their own. They have added this

329

new language to certain older drugs. It has

already happened voluntarily by companies.

DR. RUDORFER: Just a follow-up question,

semi-rhetorical. I had asked this morning about

whether we know yet the impact of the label changes

from March, and my understanding is we don't know

yet.

Is there any rationale to giving that

additional time to see the impact of that change

before we make another change?

DR. LAUGHREN: You mean before we extend

the language to the older drugs?

DR. RUDORFER: No, I mean before we change

the warning on the newer drugs.

DR. GOODMAN: Let me try it. My opinion

on that would be that now that the committee has

decided that there is an association, at least

within the trials, that I think the language needs

to be extended to association has been established

for suicidal ideation and behavior, however we want

to say it. I think it has to be very careful how

we say it apropos of the earlier discussion. I

330

think we need to define what we mean by

suicidality.

I think our present discussion is whether

it should be extended to the other and to the

present, and I don't think our discussion is over

yet, but I think probably we are leaning in the

direction of yes, extending it and erring on the

side of safety.

DR. LAUGHREN: Just to clarify, the

current language in this warning that was

implemented as of our March advisory, states as

follows: "A causal role for antidepressants in

inducing such behaviors has not been established."

So, the question that we have been

addressing here this afternoon is whether or not we

can now say that the causality has been

established, and the further question of whether we

should extend that beyond these nine drugs to all

antidepressants.

DR. GOODMAN: Tana.

DR. GRADY-WELIKY: I just wanted to

comment on the question of the warning or black box

331

with whatever we decide on this, and I would like

to agree with Dr. Nelson that certainly having a

warning, and my preference would be for all

antidepressants, would be an important thing to do,

but I would caution against using a black box

warning because it would steer many clinicians away

from using these agents, and as we heard yesterday

from both patients and clinicians, that for many

people, antidepressants are very useful treatment.

DR. GOODMAN: Dr. Mehta.

DR. MEHTA: Just a clarification from Dr.

Laughren. If you are going to extend it to the old

drugs, I guess you will also extend the same

warning to all the new drugs, is that right?

DR. LAUGHREN: Yes.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: I just want to remind

everybody that one of the major reasons for using a

black box or a box is that you think there is

something that can be done to avoid the trouble

that you are telling people about.

Just telling people that something

332

horrible has its own value, but it seems

particularly important to get people's attention

when there is something we want you to do.

In this case, there is something that we

want you to do. We want you to pay attention, not

put the people out to pasture for three weeks and

never see them again.

So, we certainly are cognizant of the

effect this can have on prescribing, and certainly

don't want to do harm, but I just want to remind

everybody that this is potentially remediable by

seeing the patient, talking to them, being alert

for these things, one of the main reasons for

thinking that you should emphasize things by a box.

I just want to be sure that is on the table.

DR. GOODMAN: How do you deal with the

fact that with the exception of fluoxetine and some

of the other medications in some of the anxiety

disorders, these are off-label uses, how does that

gibe with the general use of a black box where, in

fact, you don't have an indication to start with?

DR. TEMPLE: It's a really good question

333

and requires great delicacy, but the box is about

the warning, and we have on occasions warned people

about things even though the use wasn't approved,

so you have to be very careful. But if it's--what

is an example--the hyperpyrexia syndrome.

DR. KATZ: One example, we have done this,

as Bob says, in cases, and usually, what we say is

here is a risk--specifically in pediatric

patients--here is a risk in pediatric patients. We

remind you that this drug is not approved for use

in pediatric patients, effectiveness has not been

demonstrated.

So, we tell people it is not approved, we

don't know if it works, but here is a particular

risk. So, there is certainly precedent for doing

that, and we have done that.

DR. GOODMAN: Ms. Bronstein.

MS. BRONSTEIN: If I heard nothing from

the public, it was they want to be warned. They

want to hear the risk. I was thinking of big bold

letters, not the black box. From what I have

understood, the black box is used for really dire

334

situations, and while suicide is a dire situation,

what we are asking is that they monitor patients

closely, that they ask their families to

participate in that monitoring process.

So, I see this more in the realm of

informed consent, and I don't know what we have

done in the past for more of an informed consent

process. Have we any drugs that we require an

informed consent?

DR. TEMPLE: We had a few where we did

that. My own personal view is that gaining

consent, after all, you have to open your mouth to

take it, is a little funny. What we have moved

more toward, but not entirely, is something where

there is a required distribution of a piece of

paper that says these things to the patient and

some acknowledgment that they have read it. It's a

little different concept for consent.

But we have done that in particular cases.

It is very burdensome. If people are worried about

discouraging the use of the drugs, they need to put

that in--

335

DR. GOODMAN: We are talking about

children. It's assent, isn't it, and it is more

complicated?

DR. TEMPLE: Well, whether it is a matter

of displaying what the risk is in a piece of paper

that someone acknowledges having received, or you

want to call it consent, it is sort of the same

thing, but it's a big step.

We have done it for thalidomide and things

like that, but it is a very big step. You have to

balance what effect you think it has on the use of

the drug.

DR. GOODMAN: Dr. Trontell.

DR. TRONTELL: Just to expand on Dr.

Temple's remarks, informed consent has been used by

the Agency for drugs other than ones with the

extensive systems like thalidomide and clozapine.

It is used for several Parkinson's disease agents,

as I recall.

It is often called, some call it a patient

agreement. It is some way of setting forward some

kind of written system recorded often in the

336

patient's chart, that basically, tries to again

assure that this conversation occurs between the

provider and the patient.

So, again, it clearly is one area where we

are again stepping into that therapeutic

relationship, where some people take exception to

the Agency doing that.

I had actually one additional remark, if I

could make, relative to boxed warnings. In

general, the Agency often uses those in context

where the adverse event is associated with

fatalities. Clearly, suicidal behavior can result

in that.

An incidental consequence of that is that

products that carry boxed warnings also have to

carry that in their advertising, and quite

frequently, that makes it quite difficult to

explicitly advertise that product in

direct-to-consumer advertising.

DR. GOODMAN: That you for that ad

information. How about in sampling, would it

affect detailing, sampling?

337

DR. TRONTELL: The requirement is if you

carry a boxed warning, that boxed warning has to go

on all the materials that are used for advertising.

I am not aware that it would impede sampling, but

effectively, the advertising that goes into popular

magazines or on the television can't easily

accommodate that, so you don't get specific

products typically advertised that have those

warnings.

DR. GOODMAN: Thank you.

Dr. Temple.

DR. TEMPLE: You can't use reminder ads

either.

DR. GOODMAN: I am sorry, I missed that.

DR. TEMPLE: You can't do something called

a reminder ad. That is where you just give the

name and don't say much about the drug. You can't

do those anymore if there is a box.

DR. GOODMAN: Dr. Newman.

DR. NEWMAN: I think those would be big

pluses. As we move into the effects of the boxes

and what effects that might have on prescribing, I

338

think we have to come back to the issue of

efficacy.

We have I think very strong evidence of

harm and really not very good evidence of efficacy,

and although I know many practitioners are

convinced that these drugs work, if you look very

closely at the TADS trial, just as an example, at

the Childhood Depression Rating Scale, the

improvement with placebo was 19 points, and the

improvement with the drug was 23.4 points.

You bring people in, you start a

medication, and you see an improvement, you are

very, very likely to believe that the drug is

effective, and the reason why we do randomized,

double-blind trials is because personal experience,

however compelling, is not a reliable way to tell

whether drugs work.

In the study where they worked, in the

TADS, the improvement over placebo was really very,

very small, and I would say not detectable by a

clinician treating individual patients.

So, it would not be that bad if use of

339

these drugs were diminished, I think, because we

don't know whether they actually help most patients

when you put together all the data, and I think it

is very important that this conversation about the

risks take place, so I would favor some sort of

informed consent process.

DR. GOODMAN: I think it was appropriate

for you to remind us of the efficacy issue as we

are starting to look about benefit, as well as

risk, as we make recommendations and regulatory

actions.

I agree and I think many of us, including

myself, have said this in the room today and

yesterday, that there is a dearth of data on

efficacy. We do have the positive trials that were

submitted, that led to the indication for

fluoxetine in major depression.

We do have the TADS data, and actually, I

may have misspoken yesterday when I said that there

was no difference between fluoxetine and placebo on

the Children's Depression Rating Scale. That is

only true by one analysis, but if you look at mean

340

comparisons, although they are small, you are quite

right, that the magnitude of the difference is

small, I think they still reach statistical

significance.

Where we lack the most data is on

long-term benefit. I think what you are hearing,

what we heard yesterday from some of the

clinicians, what you are hearing from some of the

people in this room who have treated children with

antidepressants, sure, it is contaminated by bias

and expectation. There is no question.

I have been humbled before in terms of the

limitations of my own ability to discern an effect

in the absence of a placebo-controlled study, I

grant you that, but at the same time, we are

dealing with drugs that have been out there a long

time, numerous seasoned clinicians, and there is a

very powerful impression among many, and from the

consumers themselves, that in some cases, this has

made a huge difference.

So, I worry that if we ignore that

information, even though it is not good data in the

341

way we would like to see it, that we may risk a new

increase of suicidality on the back end.

We are doing a lot here to protect it on

the front end, but we haven't done the kind of

studies that were recommended before, earlier by

Dr. Temple, and doing a discontinuation study, and

I agree we don't know the answer, but it is

conceivable that there will be some patients that

will be deprived of those medications that may be

life-saving and may even be inclined to come off

medications and have a relapse that will lead to

suicidal behavior.

I welcome other comments.

Dr. Nelson.

DR. NELSON: Although I agree with you,

Tom, I would be concerned that saying that alone

without including drug-specific efficacy data will

send the wrong message, and i think there needs to

be a balance, and I would be curious how my

psychiatric colleagues would react to the following

suggestion, that if perhaps in fluoxetine, you

included the efficacy data, which is the only drug

342

that, in fact, has passed that bar, would it be a

bad thing if, in fact, practitioners, who are not

child psychiatrists, who lack the knowledge to be

able to then, if someone fails fluoxetine, to make

an appropriate choice of any of the other agents.

If, in fact, what happened in general

practice, family practice, general pediatrics, is

fluoxetine was the first drug that would always be

chosen, hopefully, under appropriate monitoring

given the risk, but then if you failed that, you

need help, that it is not something where you will

go to multiple other drugs and start running down

the list, thinking that if you just found the right

one, you would be okay.

So, my question is can we custom craft the

labeling where you might drive that kind of medical

practice, and if we could, would that be a good

thing?

DR. GOODMAN: Dr. Malone.

DR. MALONE: I think it would be a good

idea to customize the labeling.

DR. GOODMAN: I am sorry, I missed your

343

point.

DR. MALONE: I think it would be good to

customize the labeling, but in addition, fluoxetine

is a positive study. As a clinician, when you look

at the PDR, I think it would be very helpful to

have some balanced statement regarding the number

of studies that were done, say, for depression with

a given agent, and the number of studies that were

positive and negative, because I mean I think it

has been said before, when you look at the PDR and

you see a statement that it is not indicated for

under 18, it is quite a different thing than to

know that there were five studies done and none of

them were positive.

You could add statements that negative

studies don't mean it doesn't work, but it would be

helpful for me, as a clinician, to be able to read

that information.

DR. GOODMAN: Dr. Pine.

DR. PINE: I just have a couple of

comments related to the discussions that have been

going on, the first of which is, you know, we

344

really have not talked about the efficacy data, and

I think it is very important to spend a fair amount

of time talking about that.

The second thing is, for reasons that I

will go into in a second, I would agree with some

of the sentiments related to concern about adding a

black box warning, that I would be concerned about

that, and I think probably some of the most

compelling data, at least to think about, is to

look carefully at the efficacy data for fluoxetine.

There are a couple of things to say. The

first is to remember that based on the efficacy

data alone, even before we had the TADS trial, that

those data were felt to be sufficiently compelling

to justify an indication in pediatric depression.

Now, the second thing to say about the

fluoxetine data in general is at least when I look

at the magnitude of the effect in that study,

relative not only to other effects in pediatric

depression, but relative to other effects of

psychotropic agents in many disorders, I think one

would not call that to be a small effect.

345

In fact, I think the investigators and the

journal itself called it a moderate effect, which I

think is a fair summary of that effect.

The third thing to say is, if I am

correct, although I would like to hear from the FDA

about this, thus far the only warnings that have

been given make no explicit statements about

causality, number one, and then number two, make no

explicit differentiation between the potential for

the risk being greater in children relative to in

adults.

I think just based on what I have heard,

clearly those two statements could be made

relatively strongly that there is something

different going on in kids than adults, at least

based on the level of review that we have had that

hasn't been said, and it is pointing to causality.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: Well, we still believe the

general warning about paying attention to people

still should apply to adults and children, we don't

want to change that, but as was said, it now says

346

causal relationships not established, and you don't

think and we don't think that that is a true

statement anymore about the pediatric population.

So, the thought would be to add, modify,

something to convey what the new findings are.

I want to make one other comment. There

are various ways of communicating with patients and

their families. Some kind of form to sign is one,

but there also is patient labeling that can be made

to go to every patient who gets prescribed one of

these drugs.

We have done that for a very large number

of drugs. They are called Med Guides at present,

and they can be focused on the particular concerns

that one has and the risks that there are, and we

would obviously have to dance around the fact that

the drug isn't indicated for children, but there

are still ways of doing that.

So, that is not causing a form to be

handed out in the office, but it is a way of

communicating. As I said before, the probability

that that will be distributed, we think goes up

347

when it becomes part of a package, part of a unit

of use package.

DR. MURPHY: But we need to make clear

that the Med Guide is different than the patient

package insert, that people often think about that

comes with the regular label.

What might be helpful to think about,

there is two things, what you want to warn about,

where you want to put it or how you want to put in

the label, black box, warnings, precautions, people

talk about bolding versus how do you want to get

the information out to others besides the learned

intermediary.

That is where Dr. Temple is talking about,

you know, do you want it just to be in our standard

patient information in the label, which does not

have to be given to a patient. A Med Guide would

require that it be given to the patient.

That does not address the issue that I

have heard also from you about where does the

communication between the learned intermediary and

the patient occur.

348

So, if you could think of that in three

different places, there is the learned intermediary

information, there is the information for the

caretakers between the learned intermediary, and

then there is the Med Guide where the patient gets

the information whether that person has had that

conversation with them or not.

DR. GOODMAN: One way of testing the

confidence of this group in the efficacy, albeit

unproven, of the antidepressants in the pediatric

population--here I speak specifically of

depression--is to go the next step, the step that

the British counterpart of the FDA made.

We haven't talked about that, but it

certainly has been mentioned, suggested in the

public hearings, or I would like to engage this

group in some discussion to see if there would be a

recommendation to ban the use of all the

antidepressants with the exception of fluoxetine in

the pediatric population.

I am not saying I agree with that, but I

think we should air a discussion.

349

Dr. Nelson.

DR. NELSON: A question for FDA

colleagues. One of the presentations, it might

have been Dr. Laughren might have mentioned that

contraindication means different things to

different regulatory agencies, and I would be

curious if you could just be specific and say what

does contraindication mean in Great Britain, what

does contraindication mean in the United States to

help answer that question.

DR. LAUGHREN: My understanding of what it

means in the UK is that, in general, the drug

cannot be used, but under specific circumstances,

for example, by certain specialists it may be used,

so it doesn't mean quite the same thing as it does

in this country where if we put a contraindication

in the labeling for a product, that means nobody

under any circumstances should use that. There are

no circumstances where it would be appropriate to

use that drug. That is generally the way it is

interpreted by clinicians.

DR. GOODMAN: Let's go with that for a

350

moment, a discussion of whether these drugs should

be contraindicated with the exception of fluoxetine

in pediatric depression.

Dr. Malone.

DR. MALONE: You just said it in pediatric

depression. I was going to say that two of them

are already indicated in some childhood anxiety

disorders, but apart from that, I would be against

banning the drugs. I think that would be a big

different step.

As has been said, just because studies

have not proven efficacy doesn't mean that there is

not efficacy, and if you failed on Prozac in

depression, you wouldn't have many other options.

So, I would really be against banning them.

DR. GOODMAN: Dr. Wells.

DR. WELLS: I agree with Dr. Malone, I

would not favor banning the other antidepressants

other than fluoxetine, because many of these

children will not respond to fluoxetine, and they

certainly deserve to have access to other drugs

should that occur.

351

However, I do believe that the labeling of

Paxil should provide information consistent with

the June 19th, 2003 FDA talk paper recommending

that Paxil not be used in children and adolescents

with MDD in light of the lack of proven efficacy

and the troublesome documented signal for

suicidality in that population.

I further believe that we should recommend

a similar labeling change for venlafaxine.

DR. GOODMAN: Dr. Gorman.

DR. GORMAN: I think that the black box

warnings in terms of labeling makes a lot of sense

especially since I think there is a way to also

increase transparency to the learned

intermediaries, as well as the people who use the

medicine.

I would strongly recommend that the Food

and Drug Administration reconsider an active

labeling in the pediatric section where a single

sentence where it says Pediatric Usage, "After 3

randomized, controlled clinical trials including

600 patients, this medicine was not proven to be

352

effective."

I don't think that takes a lot of space, I

don't think it's very confusing, and if you need to

then reference the summary that is available for

these pediatric studies at another place on your

web site, I think that is perfectly appropriate.

I understand that creates some other

difficulties for you, but I think that if you put

it in the Pediatric Use Section, it increases the

transparency.

So, then someone can say who wants to use

Paxil, this drug has been used in 700 children in

controlled trials, and it did not meet the bar the

FDA said, but I think it has some potential benefit

for my patient, then, they have the data to at

least start to make a decision about that.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: Do you have a view on what we

should do with a drug that had one positive and one

negative study?

DR. GORMAN: I do.

DR. TEMPLE: That's not wise guy, I mean

353

we are going to face that problem.

DR. GORMAN: I think you will face that

problem because there is one of those drugs already

there.

DR. TEMPLE: That is why I asked.

DR. GORMAN: I think you should say there

has been one positive and one negative, and that

doesn't meet the bar the FDA sets. You can then

argue. Then, people will start to argue with you

whether the bar is too high or too low, but I think

that if you make it transparent, which is my

understanding, speaking for myself and not my

academy, my understanding of the intent of the Best

Pharmaceutical for Children's Act, that this should

make--the public is paying for these drug trials

with increased consumer prices, and therefore, they

have the right to the results of that data.

I think that data should be reflected in

the label even if it's troublesome to the Agency.

DR. TEMPLE: As I said, we are moving in

that direction. We haven't quite gotten there yet,

but we have had many of the same thoughts you have

354

expressed.

DR. GOODMAN: Dr. Perrin.

DR. PERRIN: I would strongly support what

Dr. Gorman just said. It seems to me that what we

know at this point is there is a drug for which

there is moderate evidence of efficacy and a whole

bunch of drugs for which there isn't, but we don't

really think we have good enough data on those

latter drugs, both in the sense of short-term

efficacy trials, and more importantly, long-term

efficacy trials, and somehow or other we need to

get that information out in some useful way.

Second, I think we know that there are no

trials for these drugs in things like migraine,

headaches, minor acute depression, and other things

that could be very, very risky.

Third, we know something about causality,

but we know causality, unfortunately, only in the

SSRIs, but we have no evidence that other

antidepressants have any likelihood of being less

causal.

DR. GOODMAN: "Causal" with regard to

355

suicidality?

DR. PERRIN: Suicidality.

DR. GOODMAN: It isn't just the SSRIs.

DR. PERRIN: I am sorry, the nine drugs

that we are studying, my apologies. But we don't

know that it covers all antidepressants. I think

we can leave that out and say basically, we have no

evidence that any other antidepressant is better or

has less risk or whatever.

I think finally, we should say as

explicitly as we can some of the comments that Jean

has raised and others have raised about the

explicit elements where people should be particular

vigilant about risk, the timing issues, the type of

patient issues, et cetera. I think we should be

very clear about that.

DR. GOODMAN: Dr. Rudorfer.

DR. RUDORFER: I think what we are talking

about is how to titrate the warning, if you will.

One thought I had, we are clearly not going to be

able to settle the efficacy issue today because we

don't have the data before us. Again, as was just

356

mentioned, there are better efficacy data in the

anxiety disorders than we have seen in major

depression.

What I am thinking is the overriding issue

that we had coming in, which I think remains with

us, is how to discourage irresponsible use of these

antidepressants in the pediatric population

while--I don't want to speak for anybody else--but

my sense is that we don't want to discourage

appropriate responsible use of these drugs by

clinicians who are in accord with the warnings as

they now exist, that is, who are monitoring

patients responsibly.

So, one thought I had was in terms of

having a bolded section in the warning, but the

bolded part would not be the adverse effect. The

bolded part would be what we want clinicians to do,

that is, wording to the effect that close patient

monitoring is required during use of this

medication especially early in treatment and at

times of dosage changes.

That is just one thought off the top of my

357

head, again directing attention to that, and I

don't know if language such as required is

appropriate, but something along those lines to

send that message to the clinician that a higher

standard of care is required with this drug, again

certainly including other warning language as we

have been discussing, but my point being to find

the right titration where we are not scaring off

appropriate use of the drugs.

DR. GOODMAN: Ms. Griffith.

MS. GRIFFITH: We are about bouncing

around a bit when you raised contraindicating, the

possibility of contraindication and then going back

to the best methodology for informing patients and

practitioners.

I was compelled by the arguments both Dr.

Nelson and Dr. Pine made with respect to the black

box, and I am wondering if it is not now a

knee-jerk reaction when a doctor sees a black box

warning, as you suggested, Dr. Nelson, just to

absolutely refuse to use that particular

medication, and, if so, I think that that poses a

358

hazard to the ability for a doctor, a psychiatrist

hopefully, in using that drug in his or her tool

box.

I also, no disrespect intended to Drs.

Temple and Trontell, when you said that by

developing a mechanism, either informed consent or

patient information letter would be burdensome, and

I believe that Dr. Trontell said that the FDA does

not like to get in between the patient and the

provider by dictating what a provider has to do by

way of informing, I think that that is wrong. I

don't think burdensome is the right way to look at

this.

I mean it may be burdensome, but it is

absolutely necessary, and I don't think that the

FDA should feel that they can't advise doctors and

caregivers to be very forthcoming and very

interactive with the patient, so I would recommend

either a letter or--

DR. GOODMAN: Tom, you had a comment?

DR. LAUGHREN: We actually have a

precedent for an informed consent in the pediatric

359

area in child psychiatry for Cylert. There is a

consent form. The problem is that it is voluntary,

and it is not clear how much it is actually used.

Maybe some of the child psychiatrists here

who have used Cylert can comment on whether or not

they actually use that.

DR. PINE: I can tell you that the safety

issues around Cylert, when they became public, led

to a dramatic decrease in the use, and I think it

had more to do with the nature of the concern than

the process that was used to monitor. So, I don't

know that this situation is completely analogous.

DR. GOODMAN: Dr. Pollock.

DR. POLLOCK: A number of us have to leave

within the hour, and Dr. Temple, I believe has now

raised at least two specific questions that he

wants us to I think put to a vote. I am kind of

concerned, I mean I would like to at least frame

those two questions and see if we can at least

accomplish that.

As I heard them, what I would like to do

is if we go back to the text before this one, of

360

what we had for Question 3, and then I think you

were asking us does this apply to all

antidepressants, and we go around Yes/No, and then

I think the second question, which is extremely

important, is does some semblance of that

information go into a black box. I think in that

case, it might be within that black box at least

for, say, fluoxetine, within that box there is some

statement of the efficacy data.

I really am very, very concerned that the

signal-to-noise ratio, the amount that is spent in

direct-to-consumer advertising, even the patient

information sheets, none of that really gets the

attention it deserves.

We may not agree with this data, it may

not be what we wanted to find, but what we are

sitting with is no--with one exception--no evidence

of efficacy, and what we have is evidence of some

causal relationship.

As we were instructed, the evidence of a

causal relationship for a warning that we want to

really get across doesn't have to be beyond a

361

reasonable doubt. It may be beyond a reasonable

doubt for efficacy, but for warning and drawing the

appropriate attention and concern to this, I think

we really need to put that.

So, I am sort of dealing with those two

questions, if we could, Mr. Chairman, frame those

questions and vote on them before we go ahead.

DR. GOODMAN: Unless I suffered a stroke

here, I think we already voted on the first

question, Question 3, that has already been voted,

and I don't think we need to have that 3(b) in

there, because the implication of 3(a) is that we

did not exempt any of the antidepressants, at least

when we were talking in terms of the trials.

I think what you are getting at is in

whatever form the warnings take, should that be

expanded to all the antidepressants, and although

we didn't vote on that, my sense from the

discussion is the answer was Yes.

So, I think we have already answered that.

Partly because there isn't that much time left, and

I do really want to end at 5 o'clock, is unless

362

there is a compelling reason, I do not want to

subject these additional questions to a vote.

The only one that I think that perhaps

might merit a vote at this point is the black box,

but still let me remind the committee that we are

not making decisions, we are making

recommendations, and it will be up to the FDA

whether or not to implement that recommendation.

Dr. Katz.

DR. KATZ: There is one other labeling

question we either need you to take a vote on

quickly or just sort of get a general sense, but

that has to do with the contraindication question

and whether or not the sense of the room is that

they should not be or should be contraindicated.

DR. GOODMAN: I would rather start at that

end. I agree with that, and I think what I would

like to do is after just a few more moments of

discussion, equivalent to the British ban, we are

using the word contraindication, or any other

discussion before we take a vote on whether this

committee would support contraindication of all the

363

antidepressants except fluoxetine in major

depression.

Dr. Nelson.

DR. NELSON: My interpretation of the

meaning of the two words in the two different

medical systems is that contraindication in the

United Kingdom is not a ban, and is simply a way of

driving the use of these drugs into the hand of

appropriately qualified specialists.

If that interpretation is correct, then,

contraindication in our system would not be the

equivalent response, and if we think that that was

the correct thing for them to do, the question is

do we have another mechanism available to us here,

such as adding a kind of oncology type warning

about appropriately qualified specialists to

accomplish that.

In my mind, saying it should be

contraindicated here is a very different meaning.

It would be a ban in the United Kingdom perhaps,

but contraindication there was not a ban from my

interpretation of what I was told.

364

DR. GOODMAN: In the UK, they were banned

is my understanding.

Dr. Temple.

DR. TEMPLE: Well, no, there is wording

that sort of suggested if you are properly trained

and really need to, you can do it. But a

contraindication here, and you can safely vote on

that, means we think--we think you can still do it,

we don't control your pen--means we think there is

no circumstance in which anybody should use these

drugs for that purpose.

DR. GOODMAN: But at the peril of the

prescriber.

DR. TEMPLE: The prescriber doesn't have

to pay attention to our labeling, and sometimes

they don't, but it would reflect the view of us,

and presumably the manufacturer that writes it, the

rightful labeling, that you should not use these

drugs for that purpose. There is no case in which

the benefits outweigh the risks. That is what it

means.

DR. GOODMAN: More discussion on

365

contraindication?

Dr. Malone.

DR. MALONE: If you contraindicate it for

major depressive disorder, is it automatically

contraindicated for off-label, or what happens to

all these other uses?

DR. TEMPLE: Off-label use is not

discussed in labeling. That is why it is called

off-label use. We know perfectly well that people

have been using these drugs in pediatric patients

even though it is not in the label.

A contraindication actually would tend to

discourage that use, because you are then reacting

not to silence, but to a specific statement that

says you really shouldn't do this. So, it changes

the present situation, there is no doubt about it.

DR. GOODMAN: Ms. Dokken.

MS. DOKKEN: Just a quick question of

clarification. What is the impact of

contraindication on further research?

DR. TEMPLE: Well, further research is

currently very difficult because everybody has got

366

its exclusivity, that is the main impediment to me,

further research. I don't think a contraindication

necessarily means that no one is going to bother to

study it further.

I don't think we have practical experience

that gives that answer, and it might discourage it

some. I don't think we know.

DR. KATZ: But it doesn't legally preclude

it or anything. It can be done.

DR. GOODMAN: Dr. Leslie.

DR. LESLIE: Just in the interest of time,

I would like to move that we do not accept a

contraindication, but we do suggest that there be

wording to such that these medications should be

given by people trained in their appropriate use.

I would be worried about saying

subspecialist because there are a number of primary

care doctors that have no access to mental health

professionals in rural communities or urban

Medicaid areas, and you would be doing a disservice

to those populations, but if you said trained in

the appropriate use, it goes back to their

367

professional bodies to come up with appropriate

continuing medical education for the use of these

medications.

DR. GOODMAN: Now, we do not have to take

a vote on this. We can just continue to discuss

it.

DR. TEMPLE: I just want to remind

everybody that it is really hard to give

instructions for use of something that isn't

approved for use. I am not saying we can't figure

out a way out of it, but it is a very thorny

problem and we are sort of bound by our own rules,

but it kind of implies that you should use it when

you tell people how to get trained for using it.

DR. GOODMAN: Dr. Marangell.

DR. MARANGELL: I like the idea of the Med

Scrip [ph] where the patients get information and

the family gets information.

A question. Is there a way to ban direct

consumer advertising without a black box?

DR. TEMPLE: The black box doesn't ban it.

You just have to incorporate the black box or its

368

elements anyway into the label.

Our conclusion was that direct-to-consumer

advertising was legal, and that is why it became

allowed. There is nothing in the law against it.

Therefore, under at least many interpretations of

the Constitution, you are allowed to do it, and our

rules. So, I don't know if there is a way to--I

don't think there is an easy way to ban it.

We would not allow reminder ads. I don't

know whether that is important to anybody, but

those are not allowed.

DR. GOODMAN: Let me express the Chair's

view on the contraindication. I would oppose it.

If we took a vote, I would definitely oppose it. I

am repeating myself here.

I would disagree that there is no data

outside of fluoxetine. There are data supporting

efficacy. They are not great data, they are not

very good data, and they are very limited data.

There are some negative data certainly from the

clinical trials, and most of the data that we have

is anecdotal experience, but it is not from one,

369

two, or three people. It is from a multitude of

trained clinicians and patients.

Now, they could have been misled. They

could have misled themselves into their being an

effective drug when there was just a nonspecific

effect of the therapeutic encounter or time alone.

We know that, I know that, but I can't

ignore, as we can't ignore some of the public

testimony about instances that seemed to implicate

the medications in suicide.

We also can't ignore the possibility that

there is data out there that we don't have in a

form that we can analyze to our satisfaction that

points to the effectiveness and the protective

action of these drugs against suicide, particularly

in the long-term treatment of depression.

In the absence of those data, in the

absence of, say, negative studies, I would be very

reluctant to deprive patients of that opportunity.

So, I am not ready at this point, given what we

know, to ban the antidepressants.

Other comments? Dr. Fant.

370

DR. FANT: In listening to the various

comments, is it reasonable to emphasize the

concerns with a black box and include wording in

the black box that gives the serious,

knowledgeable, committed caregiver license to make

choices that someone less qualified or less

thoughtful would make under those conditions?

I mean I think there are a number of us,

when we take care of kids, institute therapies that

are more risky than other situations when we are

faced with limited options, but we do it

thoughtfully, or at least we try to do it

thoughtfully.

Is there any way to sort of strike that

balance with the black box and the wording?

DR. GOODMAN: Let's ask the FDA. What

kind of liberties do you have within the black box?

DR. TEMPLE: What you usually put in a

black box is a warning about the adverse

consequences of the use of the drug, and so you

would say we know this about use in pediatric

populations or whatever, and then therefore, it

371

says therefore, you really should monitor closely,

and I guess you could stick in there, therefore,

you should be particularly aware of the signs of

symptoms of deterioration or something like that.

We could think about that.

Those things are all possible. It is a

little tricky to sort of identify responsible

versus non-responsible physicians like, you know,

only responsible physicians, and you know who you

are, should use this drug, but you can certainly

say what they should be worried about and what kind

of thinking they should go through.

DR. GOODMAN: Dr. Maldonado.

DR. MALDONADO: Just a quick comment on a

ban or contraindication and box warnings. There

was a question about the impact in future research.

I just remembered that a ban in the UK is probably

not regulatorily different than a ban in the United

States, is that the enforcers in the United States

are lawyers, and we live in a different kind of

environment. The FDA doesn't enforce that, but

companies already know that with these restrictions

372

in the label, they are going to have tremendous

liability to do any future studies.

Again, the ban in the UK is a ban in the

UK, but physicians may take risks in the UK that

physicians in the United States may not be willing

to take.

DR. GOODMAN: Thank you.

Dr. Nelson.

DR. NELSON: One suggestion in terms of

allowing the kind of information you need would be

to have more of that information in the Pediatric

Use Section, and just refer to that section out of

the black box. You want the black box to be to the

point.

The only other question I would ask the

group is have we had enough discussion about the

Med Quick or whatever the name is for the patient

information to have formulated a view as to whether

there would be a recommendation to develop that

kind of a document, because that is one thing I

heard is a need certainly from the public

testimony.

373

The label is one thing, but having it in

language that could be understood, maybe physicians

will then read that document, would be a useful

thing.

DR. GOODMAN: Dr. Fost.

DR. FOST: I wanted to get back to that

patient information thing and follow up on Dr.

Rudorfer's comment, that both the message to the

patient and the physician has got to include this

importance of monitoring, but I haven't heard any

discussion yet of what monitoring means.

Does it mean seeing the patient once a

week, if so, by whom? Twice a week, once every two

weeks, is it phone contact monitoring? Does FDA

get into that depth of defining terms? The word

"monitoring," to me means nothing. I mean I have no

idea what it means.

So, until somebody defines it for me,

seeing it in a package insert doesn't tell me, as a

physician, what it is I am supposed to be doing,

but I have a sense that it is critical, that it is

very important that the patients be monitored

374

closely.

MS. DOKKEN: Dr. Nelson just asked about

the Med Guides. My question is they go directly to

the patient from whom, the clinician or the

pharmacist?

DR. TEMPLE: They are given out by the

pharmacist, and they are required to be given out,

but we don't think they necessarily always are, but

they can be attached to the prescribing package,

that is, to a unit of use package. In quite a

number of cases, we have converted drugs to unit of

use packaging in part so that they would carry the

Med Guide, because then they always have it.

DR. TRONTELL: I don't know if this was

your point. The patient would get the Medication

Guide presumably after they have already filled the

prescription.

MS. DOKKEN: So, it isn't necessarily an

opportunity, sort of a planned opportunity for a

conversation with the clinician.

DR. TRONTELL: If you want the patient to

be informed before they actually receive the drug,

375

you would need to be do something in addition to

the Medication Guide.

DR. GOODMAN: Dr. Chesney.

DR. CHESNEY: As we are not voting on

these issues, I just wanted to make a few brief

comments. I would strongly support the black box,

and I have to say we heard yesterday and I know we

have discussed this a number of times on the

Pediatric Advisory Committee, I think, for me,

package inserts are a legal document between the

company and the FDA, but I don't think they are

read.

We heard from a physician who presented

yesterday afternoon that he had not read the

package insert, didn't know what was in the package

insert, so I think we can put whatever we want in

there, but I think if you took a poll around this

table, most of us do not--that is not where we get

our information.

So, I think anything other than that is

important. I think the black box is important, I

think the Web Guide or Med Guide, I think attached

376

to the box is fabulous, but we were also discussing

the possibility of having a site on the FDA

internet site, which is very visible, very easily

navigated for the public to discuss adverse events.

Ms. Griffith and I, she was informing me

about the FDA web site and how difficult it is

currently to get the information on adverse events,

and I think that that is another route that if you

had advisories that were easily accessible, where

patients could go in and get that information

before or when the physician says I am going to

prescribe this, but you can go back and get

information.

So, I would just add those comments.

DR. GOODMAN: Ms. Bronstein.

MS. BRONSTEIN: I just want to reiterate

again, I think that the family and the patient must

receive something. I think having the Med Alert

come with the drug itself is a good mechanism

provided the information on there talks about

suicidality.

I think we have to be really clear about

377

what the risks are that the patient has to call the

physician about, even if the physician doesn't

invite that half of the conversation.

I guess I vote against the black box. I

want to be pretty specific about that. I think we

heard very clearly from family members that drugs

were helpful, we heard some that were very

unhelpful, but the biggest message that I heard

from the consumer is they want to be warned about

what the risk is.

DR. GOODMAN: So, did you say you would

vote against it?

MS. BRONSTEIN: I would vote against the

black box, but I would vote for the Med Alert going

directly to the patient. I think information

should go to physicians on monitoring importance

and frequency of visits.

DR. MURPHY: You can do both. I just want

to make it clear, one does not rule out the other.

MS. BRONSTEIN: I understand.

DR. MURPHY: I think you should look at

what is in the labels right now, too, because that

378

is the other thing. We already have bolded

information in the Warning Section.

MS. BRONSTEIN: I felt the bolded

information was a good beginning, but I think we

need to go further.

DR. GOODMAN: Dr. Pine. I would like us

to take a vote on the black box issue. I think we

have had sufficient discussion. It is not clear to

me what the preponderance of opinion is, and I

think we won't know until we go around the table.

First, Dr. Pine.

DR. PINE: Two brief comments, one

definitely relevant to the black box, the other

indirectly relevant.

One thing that we haven't talked about

that relates directly to the issue of restricting

access or reducing access to treatments is we

didn't talk about what are the data for other

available treatments for children who suffer from

major depression.

In some ways, some of the most disturbing

or concerning findings are the data for cognitive

379

behavioral psychotherapy from the TADS trial,

because while you can debate to some degree about

the magnitude of the effect of fluoxetine, I think

that there has been little debate about what the

study says about cognitive behavioral

psychotherapy, which currently is considered the

best documented effective psychotherapeutic

treatment for pediatric depression.

The data from the TADS trial were very

clear. That treatment was inferior to fluoxetine

alone, and it was no different from placebo. So,

that is number one.

Number two, I am sympathetic to the views

from the FDA that, you know, saying that this is an

unusual circumstance and we are not really sure

what to do, and I would, again, just speaking for

myself, say that I think it calls for some very

careful thought about how we are going to need to

think about these kinds of issues differently

relative to what the current options are, because

there are a couple of highly unusual things about

pediatric mental illness right now, one of which is

380

that the majority of treatment with psychotropic

agents, at least to the extent that we have talked

about it, is off-label use, which I think is a

problem that we all agree that we can't ignore.

The second is that the level of knowledge

in pediatric mental illness right now, in general,

is not sufficient, so that we can make very strong

statements.

So, I guess just in closing, to echo some

of the statements from Dr. Bronstein, I, too, would

not favor the black box. I am not sure what else

we need to do. I think we need to do something

clearly more than what has been done, and there

might not be a current available thing to do.

DR. GOODMAN: Let me ask a question as we

are trying to compose the question here. Is the

black box warning to contain information only about

risk, or does it also contain an evaluation of

relative benefit and risk? If the latter is the

case, then, fluoxetine would be exempt.

If what we are trying to do in the black

box is convey risk, then, it should be consistent

381

with our earlier votes and apply to all

antidepressants in the entire pediatric population.

Dr. Katz.

DR. KATZ: I think you can do both. I

think you can say in a black box--first of all,

black box is primarily for risk, but it's a

warning, right, since it clearly is intended to

convey a risk, but you can say this risk exists in

pediatric patients, and this drug has not been

shown to be effective in pediatric patients.

For a drug that has been shown to be

effective, you can just say here is the risk, and

the drug already carries the indication in the

label. So, you can tailor what is in the black box

depending upon what you know about the drug, both

for risk and effectiveness, but it is primarily for

risk, but you can handle that within the black box.

DR. GOODMAN: So, the black box for

fluoxetine, you might have a statement in terms

of--

DR. KATZ: I don't know, but you might not

say anything about effectiveness although it is

382

already in the indication. In other drugs, you

might say here is the risk and, by the way, we

remind you that it has not been shown to be

effective. We can sort of play with the language.

DR. GOODMAN: In any event, it seems like

this warning should be broader. The question

should apply to the pediatric population.

DR. KATZ: Right, it should not be limited

to depression.

DR. GOODMAN: It is not limited to

depression consistent with our earlier votes.

DR. KATZ: Right.

DR. GOODMAN: Dr. Nelson.

DR. NELSON: Each one of us may have an

opinion. Why don't we just plan then to use our 30

seconds to express it relative to what should or

shouldn't be in and how to link it rather than

trying to agree on that before we vote.

The box is fine, just in terms of what

goes in it. We may have different views, and we

could just then express that as we vote.

DR. GOODMAN: I agree with that.

383

Dr. Leslie.

DR. LESLIE: I just wondered if you could

summarize the advantages and disadvantages of the

black box, because we have kind of gone all around

it.

DR. GOODMAN: I have a volunteer. Dr.

Maldonado.

DR. MALDONADO: I think that it might

actually be very good to know if the committee

knows the criteria for precautions, one, in black

box. It seems that people are all over the place.

I think that the FDA has criteria. It may not be

very strict, but it might be good to receive that

kind of guidance before you vote.

DR. GOODMAN: Dr. Temple.

DR. TEMPLE: I wish I could remember. We

put out a draft guidance document. Basically, a

black box warning is for something you want

everybody to pay attention to, and the reason you

put it in a black box, it is more visible that way,

and people pay attention to things that stand out.

You can emphasize things by using dark

384

print, too, but the black box is prominent, it's

the first thing you see when you come to labeling.

It is always at the top. Then, in relative reform,

and it may be expanded on later, but it is supposed

to catch everybody's eye. It shows up more or less

the same way in promotion, so you see it. That is

the reason.

It is used for things that matter, for

things that can be fatal, and I would add it is

particularly attractive where there is something

you can do about it.

DR. GOODMAN: I think, in my mind, the

advantages are it attracts attention, it is an

attention getter. The disadvantages, which may

also be the advantages, it will discourage use, and

it may have some implications for the ability of

the manufacturers to promote their products, market

their products, at least in some forms, which again

might be, depending how you are looking at it, an

advantage.

We need to go to a vote. We are going to

start at this end of the table.

385

Dr. Santana--oh, Dr. Santana is gone.

Okay.

Dr. O'Fallon.

DR. O'FALLON: Because people are leaving,

could you ask them to vote on both the black box

and the Med Guide?

DR. GOODMAN: I don't think we are going

to have time to take all those votes. We haven't

finished talking about new studies.

DR. TEMPLE: We are assuming from

everything that has been said that people like the

idea of the Med Guide.

DR. GOODMAN: We don't need to vote for

that.

Dr. O'Fallon.

DR. O'FALLON: Black box, I like the idea,

yes. I vote yes. I think the idea of having the

two flavors, one for those that have been shown

efficacious and those that haven't is a good idea,

so I like the idea.

DR. GOODMAN: Dr. Pollock.

DR. POLLOCK: Yes, I also vote for the

386

black box with the same comment as Dr. O'Fallon.

DR. GOODMAN: Dr. Wells.

DR. WELLS: I would vote against the black

box because of my concerns that it would decrease

access to many patients who need to have the

medications.

DR. GOODMAN: Dr. Newman.

DR. NEWMAN: Yes, on the black box, and I

guess I would like to emphasize what Ms. Bronstein

said, that I think this is great for informing the

physician, and the Med Guide is good for informing

the patient, but it is important to have that

discussion about the risks and benefits at the time

the drug is being prescribed, and I don't think the

black box will accomplish that.

DR. GOODMAN: Thank you.

Ms. Dokken.

MS. DOKKEN: Yes, on the black box, and if

we adjourn before, I also want to say I think the

Med Guide is great, but it is too late and does not

involve that opportunity to have a discussion

between the clinician and the family and patient.

387

DR. GOODMAN: Irwin has departed.

Marangell.

DR. MARANGELL: I am actually very torn on

this issue. I think it is essential that we get

the word out. I am very concerned about a backlash

against people who really need appropriate

treatment and not getting it, particularly since

there is a dearth of specialists. I will vote

therefore no on the black box, but support the

revised bolded warning, the Med Quest, and perhaps

some additional education efforts.

DR. GOODMAN: Robinson.

DR. ROBINSON: I would vote yes in the

sense that if we are really saying that there is a

potentially fatal side effect that might occur in

2, 3 percent of children taking these drugs, I

think we have to in some way make sure that that

information gets out.

I am not really as concerned in some ways

of black box bolding. I just think that we need to

make sure that a potentially fatal side effect with

2 or 3 percent of the population needs to get out

388

there.

DR. GOODMAN: Leslie.

DR. LESLIE: Yes, on the black box, yes,

on the Med Guide, and then I also just come back to

exactly what some of the other people have said,

the importance, but I think this goes back to the

professional bodies, and I don't know what the FDA

can do to push this from bodies like the American

Academy of Pediatrics, et cetera, but guidelines

for informed consent, guidelines for follow up and

monitoring, and then it really bothers me that the

majority of education on these medications is done

in CME that is funded by pharmaceutical companies,

and I don't necessarily feel that it is always a

fair perspective.

So, I also think pushing for unbiased

reporting of results and in continuing medical

education.

DR. GOODMAN: Griffith.

MS. GRIFFITH: I have to say I, too, am

very conflicted, and I appreciate Dr. Marangell's

point of view. I have anecdotal evidence from my

389

family background that would make me very leery of

suggesting to a physician that he or she should be

over the top and overly concerned to the extent

that they precluded use of that drug.

But on the other hand, I am convinced by

the force of some of the psychiatrists who feel

that this would be beneficial, so I will vote yes.

DR. GOODMAN: Dr. Chesney.

DR. CHESNEY: Yes, this is a

life-threatening complication in a severe disease,

and I vote yes.

DR. GOODMAN: Dr. Goodman. My vote is

yes. It will make prescribing more difficult. I

anticipate there will be alarm from parents and the

child, and I think that is worth that complication,

because it will raise the threshold to prescribing

and force an engagement of a discussion, not only

about the risks, but the potential benefits and

alternatives to medication.

Dr. Rudorfer.

DR. RUDORFER: I would vote no on the

black box. I believe that while we are concerned

390

about a 2 to 3 percent increase of risk of

suicidality, I think the underlying illness carries

a 15 percent risk of suicide if left untreated, and

I fear that the black box would impede access to

treatments, and I think the appropriate warnings

could be conveyed in bolded language that would be

more likely to both be appreciated by prescribers

without scaring off patients and families and

clinicians.

DR. GOODMAN: Bronstein.

MS. BRONSTEIN: I couldn't have said it

any better, just what he said.

DR. GOODMAN: You said yes then?

MS. BRONSTEIN: No, I am saying no.

DR. GOODMAN: I am sorry, you said no.

MS. BRONSTEIN: I am saying no, and the

comments that Dr. Rudorfer has just said, said

exactly what I wanted to say.

DR. GOODMAN: Dr. Pine.

DR. PINE: I would also vote no, and I

would echo the comments of Dr. Rudorfer and then

also add that I am particularly concerned about the

391

paucity of child psychiatrists, and I am concerned

that the black box might, in particular, discourage

use by primary care physicians who might have the

necessary skills and might be the only physician

available in certain areas, but would be dissuaded

either from prescribing the agent or would force

families to travel very far to try to find a child

psychiatrist.

DR. GOODMAN: Gibbons.

DR. GIBBONS: I am going to vote no

because I am unconvinced from the data at this

point that the risk-benefit ratio is, in fact,

negative.

DR. GOODMAN: Ebert.

DR. EBERT: No, on the black box warning.

I also have some concerns about the nature of the

warning with regards to a fine line between

causality and also the cautions that would be

expected to be followed.

We have talked about generalizing this to

all antidepressants, and while I think that should

be done as far as the warnings and the cautions, I

392

am not so comfortable with doing that with regards

to causality.

DR. GOODMAN: It looks like a non-random

distribution of opinions here.

Tana Grady.

DR. GRADY-WELIKY: I am also going to vote

no on the black box and support Drs. Marangell,

Rudorfer, Pine, and Gibbons' statements.

DR. GOODMAN: Dr. Perrin.

DR. PERRIN: I am going to vote yes for

the black box, take my 30 seconds just to say that

I hope we will get a few moments to make some

recommendations to the FDA regarding special

credentialing or certification or training for

people being able to prescribe any of the

antidepressants for children.

DR. GOODMAN: Dr. Nelson.

DR. NELSON: I am going to vote yes on the

black box, but two comments. I think in this day

and age, a lot of the information we get about

drugs we pull from Palm-based databases. What

comes up first is a black box warning. If it is

393

not there, you don't find it, you don't see it, you

will go right to dose, you will miss it entirely.

So, I think that is the only way to get it out to

people.

The second is I would link that very

clearly with a discussion of the risk and benefit

under the pediatric-specific labeling data, which

would allow, then, a very, you know, the sort of

two-flavor approach and separating out fluoxetine

from other drugs, and I would hope that

practitioners who are not child psychiatrists would

start fluoxetine and then get help if they needed

it. Then, the question would be would that deal

with your 15 percent. I don't know, it's an

empirical question.

DR. GOODMAN: Dr. Malone.

DR. MALONE: I would vote yes on the black

box, and I would also encourage that it include the

efficacy data in the black box.

DR. GOODMAN: Dr. Fost.

DR. FOST: Thirty seconds, five points.

Number one, high standards--

394

DR. GOODMAN: Your vote? What was your

vote?

DR. FOST: I am coming to it. Can I start

my 30 seconds now? Number one, high standards of

informed consent however the FDA thinks they can

best be achieved. Two, Med Guide, which is not

relevant to that as we have heard. Three, yes to

the black box. Four, high standards for

monitoring, and I hope someone will explain to

someone else what that means. And, fifth, the real

black box, Dr. Temple referred to a black box that

heightens attention, but the real one, as

previously mentioned, is the one that conceals what

is going on, which is the black box of CME, and

that is at the root of the inappropriate use of

these drugs, and I realize the FDA has been quite

powerless to do anything about that.

DR. GOODMAN: Dr. Pfeffer.

DR. PFEFFER: I vote yes for the black

box, and I am also concerned that we need more

information from studies, and I am just concerned

that while I am voting yes, this may inhibit that.

395

I think that there is such variability yet in the

population studies and the methodologies that the

spirit of warning and the spirit of monitoring, I

highly agree with.

I would just hope that there would be an

effort yet to study these drugs further.

DR. GOODMAN: Dr. Fant.

DR. FANT: I vote yes on the black box,

and the comment I would like to make is that if

careful attention is paid to the wording, I don't

think the black box will have an adverse effect on

access of potentially useful medications for kids,

from knowledgeable, thoughtful providers who

thought certain drugs may be of benefit. I think

it may have a desired effect on wise cavalier use

of drugs in an unthoughtful way.

DR. GOODMAN: We have an independent

accounting firm auditing the vote at this moment.

I am prepared to read it.

This time, we have a total of 23 votes, 15

Yes, 8 No. So, more of a split decision than we

had on our previous votes.

396

We are running nearly out of time. I

think we have covered really most of what I hoped

to accomplish in No. 4. We can't subject every

recommendation to a question, I don't think we can

come up with every possible recommendation here. I

think we have made tremendous progress in giving

the FDA a sense of where we stand.

The final Question 5, maybe just take two

minutes. I think throughout, the meeting has been

punctuated by discussion about what data is

missing, what studies need to be done, and maybe

people can add to what I omit.

Certainly, we need more efficacy data. We

need safety data in which one of the intended

endpoints is assessing suicidality, so it is being

assessed appropriately and prospectively, and with

sample sizes accordingly, which, of course, could

represent some problems.

Nevertheless, it needs to be done. We

need to have long-term data including comparison

trials with fluoxetine, but also retaining a

placebo-controlled group. There are things I could

397

mention, but I think those are some of the main

clinical trials highlights.

Dr. Marangell, you want to add?

DR. MARANGELL: Any future written

requests, inclusion of suicidality assessment, and

consistent data dictionary.

DR. GOODMAN: Other comments on future

recommendations on research?

DR. O'FALLON: Maybe even having a group

that would look at suicidality events in some of

these studies.

DR. GOODMAN: Dr. Newman.

DR. NEWMAN: Everyone has called for a

long-term trial, and I just want to give one more

reason why that is important. I agree with Dr.

Nelson, who said that there should be some actual

numbers in the warnings about what the risks of the

increased suicidality are, but the 2 to 3 percent,

that is an 8 to 12 weeks, and there doesn't seem to

be evidence that it is tapering off over time.

We really need to know, you know, in a

year, is it four times that, and in two years, is

398

it eight times that, because we don't know that.

It is very important that the warning go on there,

but, you know, it is going to be hard to write that

accurately if someone is going to be on the

medicine more than three months or two months, to

know what to tell them.

DR. WELLS: I would like to recommend that

it would be very beneficial, I think, if sponsors

were encouraged to provide additional information

with regard to benefits in order to help clinicians

make the assessment of benefits versus risks.

Specifically, if they could provide, for

instance, pharmacoeconomic benefits to include cost

effectiveness information and cost minimization

data, and also humanistic benefits, such as quality

of life.

DR. GOODMAN: Other suggestions for future

research?

Dr. Perrin.

DR. PERRIN: I think there is a great deal

of value in doing a much better job with respect to

understanding the sample selection and sample

399

biases, and what we know about the samples at the

time of entry. I am not sure that I know exactly

what the sample ought to consist of, but we ought

to know something about prior history in much more

detail than we currently do.

We ought to know about the issue of the

likelihood of bipolar disease in these kids. We

need to know how accurately, reliably, and validly

to diagnose of MDD is made in these kids, and we

need to know something about the social and

environmental histories that might influence both

response to treatment and likeliness of adherence.

I will take one other moment just to plug

again, I really think that we need to find a way to

make sure the people who prescribe antidepressants

know what they are doing. I am a strong proponent

of the fact that we shouldn't allow anyone to

prescribe just by having a physician's license.

DR. GOODMAN: Dr Gorman.

DR. GORMAN: I would like to suggest that

the concern about major depressive disorder in

children is one of such importance to our country

400

that this discussion not be taking place, the study

not be taken upon by a pharmaceutical company, but

the National Institute of Health.

I think that it should be done in a way

that allows the real world application of these

medicines to be studied much closer to the TAD

trial than to the constraints of randomized,

controlled clinical trials.

DR. GOODMAN: Dr. Pfeffer.

DR. PFEFFER: I would agree with that, but

this is something I wanted to mention for a while.

It is a little bit out of the box, but given

funding concerns, and given the potential for

partnership, I wonder if there is a feasibility to

consider that drug companies do be involved in

these studies and that they be involved in a way

that they might be able to help with said costs to

the studies with the idea that it is totally

unrestricted and that, for example, NIH can be the

group leading this study, with selecting the

participants, designing the studies, et cetera.

DR. GOODMAN: Dr. Fant.

401

DR. FANT: One of the things that I have

been struck with in preparing for this meeting is

really understanding how little we know about how a

lot of these drugs work, and just the basic

pharmacology that underlines them.

We know the primary process that they seem

to perturb, you know, but reading through the

inserts, they talk about either low or no affinity

for this receptor or that receptor. Well, low

affinity can mean anything from no affinity to 25

percent occupancy.

I think more basic research on the basic

pharmacology of these drugs, the chemistry, and

having a better understanding at the

pharmacogenetic level of how different patients may

respond to a given drug.

That involves a lot of, you know, sort of

moving into the studies outside of the realm of

what we have been talking about up to this point,

but I think those will be important to better

understand what we are seeing here and perhaps be

applicable to other drugs in the future.

402

DR. GOODMAN: I am going to turn over the

meeting now to Dr. Chesney, who, as you know, is

the Chair of the Pediatric Advisory Committee. I

want to ask her help in closing this meeting.

Concluding Remarks

DR. CHESNEY: Thank you. The good news is

that I have a number of pages here of things to say

that have a great historical and philosophic

context. I am going to bypass all of that and on

behalf of the Pediatric Advisory Committee, just

make one comment which Dr. Nelson and Dr. Fost

brought to my attention.

That is the importance of having children

and families participate in research, well

constructed, well designed research, because in

most cases, the investigator has been well vetted,

if you will, is well understood his or her

credentials have to be reviewed. There is very

close monitoring that goes on in very well

constructed and high quality studies, and we get

important results, and deaths are rare in very well

constructed, high quality studies.

403

So, they wanted me to pass on that general

comment.

My only concluding remarks are in terms of

thanks. I think we would all very much like to

thank the families who took so much time to come

and gave us the emotional energy that it took for

them to relive their tragedies.

Also, to thank the psychiatrists and the

families who came and explained to us in great

detail the importance of having drugs available and

of the many good things that these drugs have done.

I particularly wanted to thank the members

of the FDA who have come under such intense

scrutiny over the last year, but have maintained

their professionalism and their integrity, and that

has meant a great deal to all of us.

I also want to thank all of the members of

the FDA who have organized and executed this

meeting over the last two days, which has taken

just an awesome effort in terms of getting the

materials out to us, getting the materials out to

everybody else, and having us here in a very calm,

404

controlled, and extremely thoughtful environment.

This has been a very intense two days, and

I wanted to thank all of them.

Finally, just to thank Dr. Rudorfer, who

chaired the February session, and Dr. Goodman, who

chaired this session, and Anuja Patel, who has

brought it all together. I think they have just

done a fabulous job of keeping us on track.

DR. GOODMAN: Hear, hear. I don't have

much to add. I echo all the comments of my

colleague. I also want to specifically thank the

members of the Psychopharmacologic Advisory

Committee and, once again, Anuja Patel. It would

be useful if you could be at my side always, make

me look as good as you did during this meeting,

keeping me organized.

It has been a very challenging two days.

I think we have made a tremendous amount of

progress. I anticipate that there could be more

meetings like this, and hopefully, the next time,

if we are asked to meet, it is after the emergence

of additional data, particularly on the side of

405

efficacy.

We didn't take a vote on the banning. I

felt, in part, that I knew how it was going to turn

out. We were not going to vote in favor of

banning. On the other hand, I think the reason

behind that was going to be based mostly on

subjective experience, not so much the data at

hand. So, I didn't subject it to a vote.

At some point, I would like to be in the

position where, like we did on some of the other

issues, is to have sufficient data before us that

we could make an informed decision and make hard

choices.

Right now, we are in a position where the

drugs are out, they are being used. There is a

widespread opinion that they not only help, but

they actually save lives, but we can't really, with

the data available to us, make the kind of informed

decision that I think would make us all feel

comfortable.

I think some of the problems that have

emerged, the suicide signal and the way it

406

appeared, are a symptom of some disparities between

our clinical practice and our clinical research

knowledge, and I hope that, over time, that gap can

be narrowed, so that our research keeps pace with

the clinical needs and the clinical practice that

is out there.

I don't know how to suggest a mechanism to

do that, and I think we have done a good deal of

damage control, and I think it is unfortunate that

we didn't have an opportunity to intercede sooner.

With that, again, I just want to thank

everybody for the participation, their attention.

This has been a really outstanding meeting from my

perspective, a terrific group of panel members who

have grappled with very difficult decisions, and

have made my job a great deal easier.

Thank you again. This meeting is

adjourned.

[Whereupon, at 5:00 p.m. the proceedings

concluded.]

- - -