1

DEPARTMENT OF HEALTH AND HUMAN SERVICES

FOOD AND DRUG ADMINISTRATION

CENTER FOR DRUG EVALUATION AND RESEARCH

PULMONARY-ALLERGY DRUGS

ADVISORY COMMITTEE

Wednesday, July 13, 2005

8:00 a.m.

Gaithersburg Hilton

The Ballrooms

620 Perry Parkway

Gaithersburg Maryland

PARTICIPANTS

Erik R. Swenson, MD, Chairman

Teresa Watkins, R.Ph., Executive Secretary

MEMBERS:

Mark L. Brantly, M.D.

Steven E. Gay, M.D., M.S.

Carolyn M. Kercsmar, M.D.

Fernando D. Martinez, M.D.

I. Marc Moss, M.D.

Lee S. Newman, M.D.

Calman P. Prussin, M.D.

Michael Schatz, M.D.

David A. Schoenfeld, Ph.D.

CONSULTANTS AND GUESTS (VOTING):

Karen Schell, RRT, Consumer Representative

Jacqueline S. Gardner

Nancy J. Sander, Patient Representative

GUEST SPEAKER (NON-VOTING):

Christine Sorkness, Pharm.D.

FDA STAFF:

Robert Meyer, M.D.

Badrul Chowdhury, M.D.

Ann Trontell, M.D., M.P.H.

Sally Seymour, M.D.

J. Harry Gunkel, M.D.

Eugene J. Sullivan, M.D., FCCP

C O N T E N T S

PAGE

Call to Order

Erik R. Swenson, M.D., Chairman 5

Introductions 6

Conflict of Interest Statement

Maryanne Killian 8

FDA Introductory Remarks

Badrul Chowdhury, M.D., Division of

Pulmonary-Allergy Drug Products 14

Guest Speaker Presentation:

An Overview of Long-Acting Beta Agonists

Christine Sorkness, Pharm.D.,

University of Wisconsin 21

Questions by the Speaker 68

GlaxoSmithKline Presentation:

Opening Remarks

C. Elaine Jones, Ph.D. 82

Salmeterol Review

Katharine Knobil, M.D. 87

Closing Remarks

C. Elaine Jones, Ph.D. 115

Questions by the Committee 116

Novartis Presentation:

Introduction

Eric A. Floyd, M.S., M.B.A. 136

Efficacy and Safety of Foradil

Gregory P. Geba, M.D. 139

Clinical Implications

James F. Donohue, M.D., Chief, Pulmonary

Division, University of North Carolina 155

C O N T E N T S (Continued)

PAGE

Questions by the Committee 168

FDA Presentation:

Salmeterol

Sally Seymour, M.D., Division of

Pulmonary-Allergy Drugs 183

Formoterol

J. Harry Gunkel, M.D., Division of

Pulmonary-Allergy Drugs 207

Questions by the Speakers 224

Opening Public Hearing:

Chris Ward, Asthma and Allergy

Foundation of America 233

Committee Discussion 238

P R O C E E D I N G S

DR. SWENSON: Good morning, everyone. I

am Dr. Erik Swenson. I am the Chairman of this

Pulmonary-Allergy Drug Advisory Committee meeting,

meeting today here to discuss the implications of

recently available information and data related to

the safety of long-acting beta agonist

bronchodilators.

Before we go around and introduce the

members of the panel, I would like to ask them to

remember that we have microphones here that have

dual functions. One is to show that you wish to

raise a question. That is the "request" option

there; then to speak is on the right-hand side.

So, in raising questions, would you please first

hit the "request" button. We will be monitoring

and call you in turn. Please do remember to use

the "speak" button when you do speak since

transcribers will need to hear you on the tapes.

With that having been said, I would like

to have members of the panel here go around and

introduce themslves. We will start with Bob Meyer

and have you introduce yourself in turn.

Introductions

DR. MEYER: I am Bob Meyer. I am the

Director of the Office of Drug Evaluation II in the

Center for Drugs.

DR. CHOWDHURY: I am Badrul Chowdhury, the

Division Director, Division of Pulmonary and

Allergy Drug Products.

DR. TRONTELL: Ann Trontell, the Deputy

Director of the Office of Drug Safety.

DR. SULLIVAN: My name is Gene Sullivan.

I am the Deputy Director of the Division of

Pulmonary and Allergy Drug Products.

DR. SEYMOUR: I am Sally Seymour, medical

officer in the Division of Pulmonary and Allergy

Drug Products.

DR. GUNKEL: Harry Gunkel, medical officer

in the Division of Pulmonary and Allergy Drug

Products.

MS. SANDER: Nancy Sander, President,

Allergy and Asthma Network, Mothers of Asthmatics.

DR. GARDNER: Jacqueline Gardner,

Professor of Pharmacy at the University of

Washington, and a member of the Drug Safety and

Risk Management Advisory Committee to FDA.

DR. SCHATZ: Michael Schatz. I am an

allergist/immunologist from Kaiser Permanente San

Diego.

MS. WATKINS: I am Teresa Watkins,

executive secretary for this committee.

DR. GAY: I am Steven Gay. I am medical

director of critical care support services,

assistant professor at the University of Michigan.

DR. MOSS: Marc Moss, associate professor

of medicine, Emory University in Atlanta.

DR. NEWMAN: Lee Newman, professor of

medicine, National Jewish Medical and Research

Center and University of Colorado Denver.

DR. BRANTLY: Mark Brantly, professor of

medicine, University of Florida.

DR. MARTINEZ: I am Fernando Martinez,

professor of pediatrics at the University of

Arizona.

DR. KERCSMAR: Carolyn Kercsmar, professor

of pediatrics, Rainbow Babies and Children's

Hospital, Cleveland, Ohio.

MS. SCHELL: I am Karen Schell. I am the

consumer representative. I am a respiratory

therapist from Emporia, Kansas.

DR. PRUSSIN: Calman Prussin. I am senior

clinical investigator in the Laboratory of Allergic

Diseases, NIAID, NIH.

DR. SCHOENFELD: David Schoenfeld,

professor of medicine at the Harvard Medical School

and professor of statistics at the Harvard School

of Public Health.

DR. SWENSON: Thank you. I would like now

to call Maryanne Killian, of the FDA. She has a

statement on conflict of interest to read.

Conflict of Interest Statement

MS. KILLIAN: Good morning, everybody.

The Food and Drug Administration is convening

today's meeting of the Pulmonary-Allergy Drugs

Advisory Committee under the authority of the

Federal Advisory Committee Act. With the exception

of the industry representative, all members of this

committee are special government employees or

regular federal employees from either agencies,

subject to the conflict of interest laws and

regulations.

FDA has determined that the members of

this advisory committee are in compliance with

federal ethics and conflict of interest laws,

including but not limited to 18 USC Section 208 and

21 USC Section 355(n)(4) which applies to FDA

people. Congress has authorized FDA to grant

waivers to special government employees who have

financial conflicts when it is determined that the

agency's need for a particular individual's

services outweighs his or her potential financial

conflict of interest.

Members who are special government

employees at today's meeting, including special

government employees appointed as temporary voting

members, have been screened for potential financial

conflicts of interest of their own, as well as

those imputed to them including those of their

employers, spouse or minor child related to the

discussions on July 13, 2005 regarding implications

of recently available data related to the safety of

long-acting beta agonist bronchodilators, and on

July 14, 2005 regarding the continued need for the

essential use designation of prescription drugs for

the treatment of asthma and chronic obstructive

pulmonary disease under 21 CFR 2.125. These

interests may include investments, consulting,

expert witness testimony, contracts, grants,

CREDAs, teaching, speaking, writing, patents and

royalties and primary employment.

In accordance with 18 USC Section

208(b)(3), four waivers have been granted to the

following participants. Please note that all

interests are in firms that could be potentially

affected by the committee's deliberations. With

regard to the July 13th meeting, Dr. Carolyn

Kercsmar for activities on a speaker's bureau. She

receives less than $10,001 per year for a grant

which is valued at less than $100,000 per year, and

for a grant for which the firm supplies products

worth approximately less than $100,000 per year;

Ms. Nancy Sander for ownership of stock currently

valued at between $25,001 and $50,000, and for

unrelated advisory board activities for which she

receives less than $10,001 per year; Dr. Steven Gay

for speaker bureau activities with four firms, from

three of which he receives less than $10,001 per

firm per year, and one for which he receives from

between $10,001 to $50,000 per firm per year. We

would also like to disclose that Dr. Erik Swenson

owns stock worth less than $5,001. A waiver under

USC 208(b)(3) is not required because the de

minimis exemption under 5 CFR 2640.202 applies.

With regard to the July 14th discussions,

Dr. Carolyn Kercsmar for activities on a speakers

bureau. She receives less than $10,001 per year

for two grants which are valued at less than

$100,000 per year, and for a grant for which the

firm supplies products worth approximately less

than $100,000 per year. She also owns stock less

than $5,001. A waiver under the USC 208(b)(3) is

not required because the de minimis exemption under

5 CFR 2640.202 applies. Dr. Fernando Martinez for

his membership on a speakers bureau. He has not

lectured or received remuneration in the past 12

months, and for membership on a related advisory

board. He has not participated or received any

remuneration to date. Dr. Michael Schatz for his

activities on a speakers bureau. He receives less

than $10,001 per year, and for a grant for which

the firm supplies product worth approximately less

than $100,000 per year. Miss Nancy Sander for

ownership of stock currently valued between $25,001

and $50,000, and for unrelated advisory board

activities for which she receives less than $10,001

per year. Miss Sander also owns stock worth less

than $5,001, again a de minimis waiver is not

required because 5 CFR 2640.202 applies. Dr.

Steven Gay for speakers bureau activities with five

firms, from three of which he receives less than

$10,001 per year, and two of which he receives from

$10,001 to $50,000 per firm per year.

We would also like to disclose that Dr.

Marc Moss' spouse owns stock less than $5,001. A

waiver under 18 USC 208(b)(3) is not required

because the de minimis exemption under 5 CFR

2640.202 applies.

A copy of the written 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, 5600 Fishers Lane,

Rockville, Maryland.

In addition, Dr. Christine Sorkness is

participating as FDA's invited guest speaker on

July 13th. She would like to disclose that she is

a researcher with regards to GlaxoSmithKline's

Advair and Novartis' formoterol. She also lectures

for GlaxoSmithKline concerning Advair and receives

less than $10,000 per year.

Lastly, Dr. Theodore Reiss is the

industry representative on the committee at the

meeting. He is acting on behalf of all related

industry. He is employed by Merck. Thank you. I

am done.

DR. SWENSON: Thank you, Miss Killian. I

would like now to turn the microphone over to Dr.

Robert Meyer of the FDA.

DR. MEYER: Thank you. Prior to more

formal introduction by Dr. Chowdhury, I wanted to,

first off, thank the advisory committee in advance

for your attendance today and for what I am sure

will be a very careful deliberation.

One of the things I wanted to mention was

that there was some speculation in the trade press

yesterday that there was a very specific purpose

and outcome hoped for by the agency in holding this

meeting today. I just wanted to be clear that the

FDA looks forward to a very open discussion of the

data available on the safety experience with the

long-acting beta agonists and any potential future

regulatory actions that might be recommended coming

out of this committee. So, thank you very much for

your attendance today.

DR. SWENSON: Thank you, Dr. Meyer. Now

Dr. Chowdhury, from the FDA, is going to give us

some introductory remarks pertinent to our

discussion today.

FDA Introductory Remarks

DR. CHOWDHURY: Good morning. Honorable

Chairperson, members of the Pulmonary-Allergy Drugs

Advisory Committee, representatives from GSK and

Novartis and others in the audience, I welcome you

to this meeting.

In this brief presentation I will

introduce you to the subject matter of this

advisory committee meeting. Members of the

committee, the objective of this meeting is to

discuss the implications of the available data

related to the safety of long-acting beta agonist

bronchodilators. There are two long-acting

bronchodilators marketed in the United States that

will be discussed in this meeting. These are

salmeterol from GSK and formoterol from Novartis.

Products containing salmeterol and formoterol are

indicated for use as bronchodilators in patients

with asthma and COPD as maintenance treatments.

These are effective drugs and form

important components of the treatment options

available for patients with asthma and COPD. But

an important array of adverse effects that has been

observed with these drugs is the occurrence of

severe asthma exacerbation. The intent of this

advisory committee meeting is to discuss this

specific finding of severe asthma exacerbation

related to these two drugs. Since the available

data pertain to asthma, the focus of this meeting

is on asthma and not COPD.

Surrogates of short-acting beta agonist

bronchodilators, such as albuterol, is not a

subject of this meeting. As you discuss and

deliberate on the safety of thee two drugs, keep in

mind the well-established efficacy of these drugs

because the use of these drugs, like any other

drug, is dependent on the risk/benefit ratio.

As you can see in the agenda, the first

presentation will be by Dr. Christine Sorkness.

Dr. Sorkness is a professor of pharmacy and

medicine in the University of Wisconsin. She will

give an overview of long-acting beta agonist

bronchodilators. We are very fortunate that Dr.

Sorkness, and expert in pharmacological drugs used

in the treatment of asthma, has agreed to speak at

this meeting. I thank her on behalf of the agency.

Following Dr. Sorkness, GSK and Novartis

will make presentations on salmeterol and

formoterol respectively, followed by FDA

presentations on these two drugs. This will be

followed by an open public hearing and committee

discussion.

As you hear these presentations you will

note that the safety signal of severe asthma

exacerbation with salmeterol was seen in

postmarketing studies, specifically the recently

halted large controlled study called the salmeterol

multicenter asthma research trial, acronym SMART,

conducted by GSK. In contrast, the safety signal

of severe asthma exacerbation with formoterol was

seen in the studies conducted by Novartis to

support registration of formoterol in the United

States. Novartis also conducted a Phase 4 study

with formoterol that did not show a clear signal of

severe asthma exacerbation, but the formoterol

Phase 4 study was much smaller compared to the

SMART study.

We are choosing to have this meeting now

because all pertinent data on salmeterol and

formoterol have only become recently available. We

also decided that it would be fruitful to discuss

these two related drugs together in one meeting.

Although salmeterol has been approved for marketing

in the United States since 1994, the study relevant

to this meeting, the SMART study, was halted by GSK

in January of 2003 and the data has been recently

fully analyzed.

Formoterol was approved for marketing in

the United States in 2001. The Phase 4 study for

formoterol was completed in March, 2004 and the

data from the study also has been recently

analyzed.

The significant regulatory actions that

the FDA has taken so far pertaining to these two

drugs, based on the available data, are in

cooperation of the results of the SMART study in

all salmeterol-containing product labels, including

the addition of a boxed warning, and not approving

formoterol 25 mcg twice daily dose for marketing in

the United States. Formoterol is currently

approved at a dose of 12 mcg twice daily. Please

note that the formoterol drug label does not

currently have warnings similar to salmeterol

because of lack of specific data related to the

marketed formoterol 12 mcg twice daily dose.

In the presentations from the industry and

the FDA you will see the data that led to the

agency regulatory actions. As you hear the

presentations, I request that you keep in mind the

questions that are in the FDA briefing book and

also attached to the agenda since you will discuss

and deliberate on these questions later in the day.

Here are the four questions that you will

be asked to discuss and deliberate later in the day

today. Question one, the product labels of

salmeterol-containing products have been modified

to include warnings related outcome the SMART

study. Based on currently available information,

what further actions, if any, do you recommend that

the agency take to communicate or otherwise manage

the risks of severe asthma exacerbations seen in

the SMART study?

Based on the currently available

information, do you agree that salmeterol should

continue to be marketed in the United States?

Question two, the label of the

formoterol-containing product does not include

warnings comparable to the warnings that are

present in the salmeterol-containing products.

Based on the currently available information,

should the label of formoterol-containing products

include warnings similar to those in the salmeterol

label?

Based on the currently available

information, do you agree that formoterol should

continue to be marketed in the United States?

Question three, what further

investigation, if any, do you recommend to be

performed by GSK that can improve the understanding

of the nature and magnitude of the risk of

salmeterol?

Question four, what further investigation,

if any, do you recommend to be performed by

Novartis that can improve the understanding of the

nature and magnitude of the risk of formoterol?

These are the four questions. We look

forward to an interesting meeting and, again, I

thank you for your time, effort and commitment to

this important public health service. Thank you.

DR. SWENSON: Thank you, Dr. Chowdhury.

At this point we would like to invite Dr. Christine

Sorkness who was just introduced. She has been

kind enough to give us a broad overview of these

drugs and I would like to turn the podium over to

her.

An Overview of Long-Acting Beta Agonists

DR. SORKNESS: Good morning. I would

first like to thank Dr. Chowdhury and Dr. Sullivan

for inviting me to speak this morning, and most of

all, for gathering this group of both clinicians,

researchers, industry colleagues and the committee

to review what I believe to be an incredibly

important topic. The risk versus benefit

considerations for the long-acting beta agonists

are the topic at hand and the committee has been

asked to discuss the implications of the available

data related to the safety of long-acting beta

agonists, as Dr. Chowdhury articulated.

It is a little bit awesome to review this

topic because of its breadth and depth, and also

because I know many of the committee members and

would acknowledge that they probably know more than

I do about this particular topic. So with that

caveat, I am going to indicate that I am just going

to review and try to set a tone for the discussions

and in particular anchor some of the available

data, at least as I see it as a researcher and a

clinician, that you might use to answer the

questions that you have been charged with.

The specific objectives that I have been

asked to address are to provide an overview of the

clinical pharmacology of the long-acting beta

agonists; to discuss the selection of therapeutic

outcomes which I believe are relevant for the

assessment of risks versus benefits of the

long-acting beta agonists; to review selected

clinical trials, selected because there are so many

which provide insight into the risk/benefit of the

long-acting beta agonists; and to outline the

controversies and the remaining questions which I

believe are related to the role.

First an overview of the clinical

pharmacology of albuterol, salmeterol and

formoterol. We have come a long way from ephedra

from China and its pharmacologic properties many,

many years ago to, certainly ephedrine and

epinephrine and isoproterenol. The three major

drugs that we use in our therapeutic armamentarium

for asthma right now are albuterol, salmeterol and

formoterol. You can see in common that they all

have a simple catecholamine ring, and there has

been great novelty from the industry of adding a

variety of different side chains to these products

to affect their oral versus inhalation efficacy

and, in particular, if you look at salmeterol and

formoterol you see that there are very large side

chains that have been attached to the basic

molecule of albuterol. This has allowed these two

products to have an extended duration of action.

Both salmeterol and formoterol are highly

lipophilic products, which may explain some of

their long duration of action, salmeterol more than

formoterol. We know that salmeterol binds within

the ligand binding cleft of the receptor which

probably allows sensitivity stimulation of the

receptor and its long duration, and there are other

speculated mechanisms of action for the long

duration of formoterol. Formoterol is a raceme and

only the RR and N tumor is active.

If you were to compare very globally the

beta adrenergic agents, this table is probably

relevant. Most of the pharmacologic studies relate

molar potency of these products to isoproterenol,

which is designated as a potency of 1. You can see

that both formoterol and salmeterol are more potent

products than isoproterenol. The pharmacologic

profile of the drugs is illustrated, with

isoproterenol an formoterol classified as full

agonists and albuterol and salmeterol as partial

agonists.

You can see that in comparison to

isoproterenol as its comparator, albuterol,

formoterol and salmeterol all have the luxury or

beta2 selectivity which is acknowledged to allow

these drugs to have primarily effects on the lung

versus the cardioselective effects that we see

primarily with activation of the beta

1 receptors.

The duration of action clearly is different in

these agents and, because of the long side chains

and mechanisms of action of formoterol and

salmeterol, we have known that these are the

longest acting inhaled bronchodilators on the

market today, with durations of action of at least

12 hours after a dose, and the bronchoprotective

effects, which specifically in this slide refer to

the prevention of bronchoconstriction induced by

exercise or non-specific bronchial challenges such

as methacholine, have, indeed, a long

bronchoprotective effect.

If you were to look at a more direct

clinical comparison of formoterol and salmeterol

based specifically on information in the package

inserts, it is believed that equipotent

bronchodilating doses of formoterol and salmeterol

are listed as above, based specifically on the

dosage form by which they are delivered. So we

believe, at least in clinical practice, that 12 mcg

of Foradil aerolizer is clinically bronchodilating

equipotent to 50 mcg delivered by Serevent Diskus.

In order to deliver these equipotent doses, the

recommended inspiratory flow rate is acknowledged

to be about 60 L/min for both products over a time

course of 2-3 seconds. As you might expect,

particularly because these drugs have been FDA

approved for individuals with much more severe

broncho-obstruction such as in COPD, probably an

inspiratory flow rate much below that can get

adequate delivery of drugs.

Both of these drugs are classified as

pregnancy category C and, indeed, enjoy the same

FDA approved indications based on the package of

information submitted to the FDA, the only

distinction being that salmeterol is approved for

the treatment of asthma and prevention of

bronchospasm for children over 4 years of age and 5

years on formoterol. Both of these agents have

been approved for EIB prevention and for

maintenance treatment of COPD, which is not part of

the agenda today.

Now, the differentiation of formoterol and

salmeterol, by and large, comes down to its

acknowledged difference in onset of action. You

can find many, many studies that would classify

different pharmacologic profiles. In summary,

formoterol probably achieves 80 percent of the

maximum bronchodilation within 5-10 minutes. It is

thought to have an onset of action quite comparable

to albuterol and acts within 3 minutes. For

salmeterol most of the data suggests that 90

percent maximum bronchodilation occurs after one

hour, with a median time to significant

bronchodilation of 30-40 minutes, and an onset

certainly at a time point of about 10 minutes.

This is a simple cartoon that segues to

the issue of the long-acting beta agonists

themselves in combination with clucocorticoids.

This is a cartoon that suggests the proposed

molecular interaction between the long-acting beta

agonists and the inhaled corticosteroids. The

long-acting beta agonist, through their activation

of the beta adrenergic receptor with adenylyl

cyclase, cyclic AMP, protein kinase A and mitogen

activated protein kinase may actually prime the

glucocorticoid receptor for greater nuclear

translocation and affinity for the binding to the

glucocorticoid regulatory element, which is

designated in this slide as GRE. Therefore, it has

been speculated by a variety of pharmacologic

models--Ikleburg[?] and others who have done very

elegant work--that actually the anti-inflammatory

effect of glucocorticoids can be enhanced with the

combination of long-acting beta agonist and,

clearly, that is certainly the rationale that

brought the combination products to the

marketplace.

Now, when we talk about risks versus

benefits of any agent, it is best to talk about the

outcomes of interest. I am going to preface my

remarks by the fact that I think the medical

community and patients have all been led to hope

for a 100 percent active and effective drug with

absolutely no side effects. I quite honestly

believe that to not be realistic. Therefore, when

we talk about risk/benefits we need to put in

perspective and weigh those issues, and I think it

is important to recognize that we may have very

safe medications that really have very poor

clinical efficacy, and I would suggest that they

have a distinct risk in their own right by their

inability to treat the disease at hand.

So, I am going to try to illustrate some

issues about what I believe to be important

outcomes and talk about some of the clinical trials

to date that teach us lessons about this as

applying this drug class to asthma.

We traditionally have used lung function

measures for management of asthma, both from the

perspective of clinical decision-making and

clinical research. There are many longitudinal

studies of lung health that have been enhanced by

measurements of lung function, particularly FEV

and FEV 1/FEC ratio. Clearly,

it has been

acknowledged that the gold standard for trial entry

for the pivotal trials reviewed by the FDA have

been traditional FEV

1's of 60-80 percent predicted

with 15 percent reversibility. Therefore, there

have been very uniform population groups that have

been studied in our clinical trials. I would

actually conjecture now, and will come back to it,

that we may need to broaden that a bit to capture a

more generalizable population.

Clearly, lung function measures have been

primary outcomes to measure efficacy because we can

standardize those procedures both on site and with

home measurements, and we have grown to believe

that we can minimize variability around the

measurements and can really get a handle and our

arms around what outcomes are important. Please

recognize as I talk about different outcomes in

asthma, I am not dispelling at all the value of

lung function measurements. I think they are still

critical but I don't believe that they are enough.

Let's start talking about what I believe

to be illustrative studies. This is a study

published by the Asthma Clinical Research Network

in which I am one of the investigators, and it was

affectionately called the SOCS trial. This is a

study that was intended to ask the question that in

a patient who was well stabilized on an inhaled

steroid and representative triamcinolone, and that

had pretty stable FEV

1's and peak flow variability,

could this patient basically be transferred to

placebo and do equally well; be converted to a

salmeterol product and do equally well; or did they

need to maintain continuance on an inhaled steroid

as represented by triamcinolone?

This is a study that enrolled individuals

whose mean FEV1 was 93 percent predicted, had very

low peak variability of about 10 percent, and

during the run-in period showed very good asthma

stability. The primary outcome of this study was

morning peak flow. That was selected because of

experience that the Asthma Clinical Research

Network had with what we believe to be an effect

size that we could power our study of a difference

of about 25 L/min, and because that effect size

correlated with other more clinically robust

endpoints in a variety of trials.

I think you can see that if you look at

the primary outcome of this trial of AM peak flow

you wee in the run-in period that all of the

patients in ultimately the three arms improved

during the run-in with triamcinolone, as you would

expect. You see the placebo group, once it was

randomized at six weeks, had deterioration in that

outcome; whereas, the triamcinolone and salmeterol

groups both had maintenance and actually

improvement in the primary outcome of peak flow,

and there was not statistically significant

difference between those two arms in this

particular outcome.

Now, there was obviously a variety of

secondary outcomes in this trial. You can see that

on the basis, in particular, of some markers of

inflammation that there was both a clinically and

statistically significant difference in favor of

the inhaled corticosteroids. Because of the

multiple comparisons used by the statistician, a p

value of 0.016 was that which was deemed to be of

statistical significance.

This is important in that it translates to

another very important secondary outcome of this

trial, that being defined as treatment failure

rates, on the left, and asthma exacerbation rates,

on the right. First, asthma exacerbation rates

were defined as increases in albuterol use,

decrease in peak flow, and the need for oral

corticosteroids. You can see with this particular

outcome that triamcinolone is the only one by the

Kaplan-Meier survival curve that, in essence, did

not have significant asthma exacerbations. Very

similarly, if you looked at treatment failure

rates, which was defined as an FEV1 less than 50

percent predicted, at least one course of

prednisone, the occurrence of emergency room or

urgent care visits or hospitalization, the same

trend could be seen. The triamcinolone was very

effective in preventing treatment failure rates but

salmeterol was quite comparable to placebo.

The summary for the ACRN investigators was

that patients with persistent asthma, well

controlled by low doses of an inhaled steroid

cannot be switched to salmeterol monotherapy

without risk of clinically significant loss of

asthma control. I think this is one of the studies

that clearly the asthma community has endorsed to

support the fact that long-acting beta agonists in

asthma should not be used as monotherapy, and I

don't believe that there is particular debate on

this issue and I think there are many studies that

illustrate similar outcomes.

This study is also important in that it

shows clear disparity between lung function

measures and other outcome measures, and leads us

to the conclusion from this study that multiple

measurements and dimensions of control are needed

to adequately assess therapies.

Therefore, I think, whether we broach

studies that are industry sponsored or NIH

sponsored, we are beginning to endorse more

composite measures of asthma control. This would

include days of asthma control; treatment failure

and asthma exacerbation criteria, as I have shown

in this study and many others. I would make as a

caveat that it becomes oftentimes very difficult to

compare trials because the specific definitions for

treatment failure versus asthma exacerbations and

mild, moderate and severe exacerbations may be a

little bit different. So, it is important for us

to anchor the definitions when we evaluate.

Other composite measurements have actually

been improvements or shifts in NAEEP defined NAEEP

defined asthma severity classification; the

achievement of total control or well controlled

status, as defined by GINA and applied to the GOAL

study; and certainly a variety of more patient

specific surveys of asthma control and quality of

life that have become important secondary outcomes

in our clinical trials.

Now, in reflecting upon the issue of more

composite clinical outcomes, the question needs to

be raised in applying an appropriate risk/benefit

relationship and assessment of how much benefit can

actually be achieved by the combination of inhaled

steroids and long-acting beta agonists. I am going

to focus my remarks on the combination because I

have told you that at least my belief is that

asthma is best treated by combination and,

therefore, the relevant studies are those that use

that.

A fairly early study that began to address

the role of inhaled steroids and long-acting beta

agonists in combination is the OPTIMA trial,

entitled, low dose inhaled budesonide as a

representative inhaled steroid an formoterol as a

representative long-acting beta agonist.

This study had both a group A and a group

B. I am going to focus on group A as a

representative trial of taking patients naive to

being on inhaled steroids and ultimately, after a

one-month run-in in which they were qualified to be

in this trial, were then continued on placebo,

Pulmicort or Oxis, as formoterol is called.

Therefore, they continued on beta agonists alone

versus being randomized to Pulmicort 100 mcg BID

and Oxis placebo or Pulmicort 100 mcg BID and

active Oxis 4.5 mcg BID.

The primary outcome of this trial was

severe exacerbation, designated by the arrow. This

was defined as the need for oral corticosteroids or

admission to a hospital or an emergency room visit

or substantial decrease in peak flow. This study

group enrolled patients who were 12 years of age

and older, not on inhaled steroids, who had to have

an FEV 1 of at least 80 percent

predicted post

bronchodilator, and actually enrolled a pre

bronchodilator mean FEV

1 group of about 90 percent.

These are the two primary outcomes of this

particular study. If you look on the left-hand

side in panel A, this is the Kaplan-Meier survival

curve and you can see that both the budesonide

alone versus the budesonide in combination win

formoterol did much better in preventing the time

to the first severe asthma exacerbation as compared

to the placebo group, which is the last curve that

you see on the slide. When you plot this, on the

right-hand side of the slide you see that actually

the two active treatments were, indeed, better than

placebo but were quite comparable to each other.

However, if you look at the other important outcome

of pulmonary function test, the morning peak

expiratory flow, you see in the top curve that the

combination product is superior to both budesonide

and placebo. So, whereas by one outcome the

exacerbation rates of the two active products were

not statistically significant, when you add in

another important secondary outcome the

combination, indeed, showed better outcomes.

Now, this same issue of looking at

prevention of asthma exacerbations has been

published by many, many authors. This is just a

representative study which looked at an analysis of

asthma exacerbations, looking at available studies

of higher dose fluticasone versus the addition of

salmeterol to low dow fluticasone.

If you look at this particular slide,

which is the probability of the time to the first

exacerbation, you see that the top curve, in green,

is salmeterol and, in red, the combination, and the

combination was clearly superior in the outcome of

time to first asthma exacerbation compared to the

long-acting beta agonist alone.

The analysis in this study group culled

out the different Ns of the spectrum of pulmonary

function impairment at baseline. As I mentioned,

typically the pivotal trials enroll patients that

have baseline FEV

1's pre bronchodilator between

40-85 percent predicted. That is what you see on

the left-hand side of all-comers that enrolled in

those pivotal trials. If you, instead, break down

patients who present with less bronchoconstriction

at baseline, for example, 60-85 percent predicted

versus 40-60 percent, you see that the trends are

not different and that either way, depending upon

the severity of obstruction in these patients, the

trend of the benefit of combination certainly could

be seen.

Now, the FACET trial also showed I think a

very important lesson about looking at the outcome

of severe exacerbations and relationships of

dose-response curves with inhaled steroids, as well

as the benefit of long-acting beta agonists. This

goes back to budesonide and formoterol as the

representative drugs in this study, and in this

study severe exacerbations were defined as a need

for oral steroids or a decrease in peak flow to

more than 30 percent baseline. So, severe

exacerbations here are predominantly due to the

need for oral beta agonists.

This is a large trial that randomized

individuals to one of four arms. If you look at

the far left, in green is the budesonide 200 mcg or

low dose inhaled steroid group; the purple bar is

budesonide 200 mcg a day plus formoterol; in

yellow, a higher dose of budesonide alone versus,

in the orange bar, the addition to formoterol. I

think what you can see is the very logical

dose-response curve that 800 mcg of budesonide

fared better than 200 mcg of budesonide but, very

importantly, you can see that the prevention of

severe exacerbations in both groups could be

enhanced by the addition of formoterol. So, again,

another study that suggests to us that combination

therapy can achieve the prevention of asthma

exacerbations.

Now, in brevity, rather than showing you

the individual studies of exacerbations to date

published, I am going to take advantage of a

meta-analysis, published by Sinn and others in

JAMA, in 2004 that looked at a systematic review

and meta-analysis of a variety of pharmacologic

therapies to reduce exacerbations.

This study clearly reviewed all of the

drugs that we know that are on the marketplace but

I am specifically going to look at two of the

analyses. This is the effect of long-acting beta

agonists alone on exacerbations and the distinct

trials that the meta-analysis chose. You can see

that the majority of these studies favored a

long-acting beta agonist over placebo, and a pooled

analysis showing a relative risk and confidence

interval that favors the long-acting beta agonists.

This is the analysis that looks at many of

what I believe to be the paradigm shifting trials

that showed the addition of long-acting beta

agonists to be better than either doubling or more

than doubling the inhaled steroids, and includes

the Matz and O'Byrne studies and Pauwels studies

that I shared with you earlier. I think you can

see that we have at least somewhat mixed results

here. Certainly the majority of trials favor the

combination of inhaled steroids and long-acting

beta agonists together versus favoring the high

doses of steroids. Some of them are right on the

line. The pooled summary obviously, here by this

graph, favors the steroids and the long-acting beta

agonists.

I would suggest that certainly some of the

differences are certainly on the basis of study

design, size of study, construct, and so forth but,

again, I think the meta-analysis supports the

individual trials as far as evidence that suggests

benefit of the combination.

Now, in switching gears, besides asthma

exacerbations, I think that the issue of the

capture of asthma control, as has been defined by

GINA and the NAEEP, is a very important outcome

that we have begun to carefully think about and to

posture in our individual trials. The GOAL trial

asked a very simple but important question, is GINA

NIH guideline based control achievable, and in what

proportion of patients with a

salmeterol-fluticasone combination compared with

fluticasone alone?

So, this is going beyond the issue of just

looking at exacerbations but overall asthma control

as defined by the guidelines. You can read this.

There is both total control and well controlled,

and it basically reflects what we, as clinicians,

hope to achieve for our asthma patients. And, the

question is can this be achieved by the therapies

that we have at hand?

The GOAL study design was very complex.

It was a year study of three strata of patients

based on whether they were either corticosteroid

naive or free for six months; whether they were on

a modest dose of a baclomethasone equivalent or

higher dose of a beclomethasone equivalent. These

were individuals that had to be at least 12, not

well controlled in the run-in period, and showed

reversibility of 15 percent. They were randomized

to either the salmeterol-fluticasone combination or

fluticasone alone via diskus, with a dose based on

the stratum.

During this complex design in phase 1, the

doses were either stepped up every 12 weeks until

total control was achieved or a maximum dose was

reached. In study phase 2 a dose of total control

or a maximum study dose was continued for 52 weeks.

It is important to recognize that all the

patients in this trial deserved to be on control

therapy. Their FEV

1's were about 75-80 percent

predicted. They had very, very obvious

bronchodilator reversibility, averaging about 20

percent, and what I would call were young adults.

So, whatever the stratum, these individuals

deserved to be stepped up with the therapies that

were used.

These are the patients who achieved well

controlled status. The triangles in dark are the

combination; the open circles are fluticasone

alone. You can see the run-in phase versus phase 1

versus phase 2 on this graph. You can see that

both study groups had a fairly brisk improvement in

achievement of well-controlled status. This

continued through the 52 weeks of the trial and was

achieved by both study arms, but was achieved to a

statistically significant greater extent with the

combination therapy.

Also importantly is exacerbation rates as

were studied in this trial as a secondary outcome.

This exacerbation was defined in this study as

either a burst of steroids or an ER or

hospitalization. You can see whether it was

steroid naive, the low dose inhaled steroid or the

moderate dose inhaled steroid stratum. Clearly,

all groups showed the trend that the combination

therapy was better at achieving prevention of

exacerbation rates as defined by the GOAL

investigators.

The results of GOAL are very important in

that significantly more patients achieved control

with combination versus fluticasone in each stratum

and in each stratum the time to achieve the first

individual week of well-controlled asthma was

significantly lower with combination than

fluticasone alone. More patients achieved control

at the same or lower dose of inhaled steroid in

each stratum for combination again verifying what

had been previously published on the inhaled

steroid-sparing effect.

I think very importantly in looking at

outcomes, we know that the majority of patients who

achieved well-controlled asthma in phase 1

maintained the status when assessed in the last 8

weeks of the study. But, also, there were some

patients that, additionally, were able to gain

control with sustained therapy. So, there may be,

very importantly, subjects who initially are able

to gain control but others that require longer

exposure to achieve this particular outcome.

Now I am going to switch gears a little

bit and talk briefly about a pediatric trial. One

of the things, at least in my mind, is that most of

the data that we have in looking at inhaled

steroids and long-acting beta agonists, whether

they be as entry therapy or as add-on therapy in

preventing the addition of inhaled steroids, has

predominantly been done in adults. Even those

studies which have enrolled individuals greater

than 12 years in age and up in general have not had

a sizeable enough cohort of the 12-18 population

that really have led to what I believe is a

substantive subanalysis. So, most of what we have

I believe is in adult studies, and I think we will

see more pediatric studies in the future.

This is a study that was recently

presented at the American Thoracic Society meeting

this summer, and was conducted by the CARE network

of the NHLBI-sponsored network. It is a one-year

prospective comparison of three control or

medications for the treatment of mild or moderate

persistent asthma in children.

In brief, the study schematic is a proof

of study concept. All children were in a one- or

two-week run-in period and then were either

randomized to an inhaled steroid alone, an inhaled

steroid at half the dose in combination with a

long-acting beta agonist in comparison to a

leukotriene receptor antagonist. In order to

achieve this particular proof of concept, the ICS

group received fluticasone by morning and evening

diskus and an evening capsule placebo. The middles

group of combination, and what I am going to call

combination in the future, received an Advair

diskus in the morning, a salmeterol diskus in the

evening and a placebo capsule, and the leukotriene

regimen active arm received montelukast at night

and two placebos.

Because this study has not been published

and there are responsibilities to editors, I am not

going to be able to share with you in slide form

all of the data, but I would like to summarize it

for you as I did at the ATS.

Inclusion criteria for this study were

children 6-14 years of age who had acknowledged

mild to moderate persistent asthma, as defined by

symptoms or beta agonist rescue use of peak flows

in the yellow zone. They needed to demonstrate

asthma by a PC20 methacholine less than 12.5 mg/ml.

Bronchodilator reversibility was collected but it

was not an entry criterion because we believed it

would bias the outcomes because one of the study

arms contained a long-acting beta agonist. These

were individuals who were naive to controller

medications. The issue was to look at whether

these three arms and how asthma control was

achieved in individuals with mild or moderate

asthma.

The percent of asthma control days during

the study period of 12 months was asthma control

days defined as a day without albuterol rescue,

without the use of non-study asthma medications, no

daytime or evening asthma symptoms, unscheduled

provider visits of school absenteeism, so a day in

which a parent and a physician both would be happy

that the asthma was well controlled and that was

the defining outcome for this trial.

In summary, I am going to focus

predominantly on the two outcomes related to the

full dose inhaled steroid arm and the combination

arm of the half dose fluticasone in combination

with salmeterol. Both of those study arms achieved

improvement in the percent of asthma control days.

At baseline this group of children had about 27

percent of the days that were asthma

controlled--so, very, very few. This actually

almost doubled or tripled during the active they

and the fluticasone group gained asthma control

days of 64 percent versus the combination of 60

percent. So, both groups adequately achieved

asthma control and these were not statistically

different.

Treatment failure was also a secondary

outcome in this trial, defined by either the third

burst of prednisone or a hospitalization or ER

visit due to asthma. There were only five

treatment failures in the fluticasone arm and eight

treatment failures in the combination arm. That

was not statistically significant. Of that, there

were no hospitalizations due to asthma in the

fluticasone group and two hospitalizations with the

combination group.

Overall, the comparison of the two groups

showed in many outcomes that the inhaled steroid

alone versus the inhaled steroid at half dose in

combination with salmeterol were comparable, as I

mentioned, in asthma control days; the time to

prednisone bursts and treatment failure status.

There were some important differences in

that if you looked at secondary outcomes such as

change in PC 20, the

improvement and ENO as a marker

of inflammation, and actually changes in maximum

bronchodilator response, the full dose of inhaled

steroid was actually statistically better.

I mention this study from the point of

view of one study looking at children that will,

hopefully, soon be published and gives us some

experience, I believe, with at least efficacy and

safety in a pediatric population.

Now, let's switch gears to potential

safety concerns that have been raised by the use of

beta agonists. That is what the committee has been

asked to really put in perspective today. It has

not been just in the last few years that safety

concerns with beta agonists have been raised.

Studies in the early '90s suggested that the

regular use of a particular beta agonist,

fenoterol, might produce adverse effects. This is

the number of subjects without exacerbation as a

Kaplan-Meier curve and you can see those

individuals treated with a regular dose of

fenoterol had more asthma exacerbations than as

needed. This study, by Taylor and others, raised

the specter of regular use of short to intermediate

beta agonists producing adverse effects.

As you well know, fenoterol never made it

to the U.S. market and albuterol has become clearly

the drug of choice as the intermediate rescue beta

agonist. Therefore, Jeff Drazen and the Asthma

Clinical Research Network felt it important as one

of its missions to try to answer the question of,

given that albuterol was the primary beta agonist

used in the marketplace, did it matter whether

patients were treated with regular beta agonists

versus as needed beta agonists. To achieve this

trial, patients either received two puffs of

albuterol four times a day plus extra as needed, or

placebo inhaler two puffs four times a day and as

needed, thus, sufficing the regularly scheduled

versus as needed paradigm. The study had a run-in,

a 16-week treatment trial and then a run-out of 4

weeks.

Now, whereas this group today is not here

to debate the issues of safety of short and

intermediate beta agonists, this trial basically

has led to many of the questions that we have asked

about long-acting beta agonists, and has led to

what I believe is a series of trials that are in

construct and will build on.

The summary from this particular study,

using again peak flow as the primary outcome and

power to find a difference of 25 L/min in the two

study arms, suggested that whether you are on as

needed albuterol or regular albuterol it really

didn't make a difference in this outcome and,

therefore, there was nothing evil about the use of

regular beta agonists. But the authors

acknowledged that clearly based on the way the

asthma community was moving, PRN beta agonists was

the more rational approach.

Whereas this was a prospective trial, at

the same time that this study was in the midst of

being carried out, Steve Liggett's group at

Cincinnati and others were working on cloning the

beta receptor. This is the beta receptor as a

G-coupled protein. As you well know there has been

a lot of interest in single nucleotide

polymorphisms at both the 27 position and the 16

position in a variety of both in vitro and in vivo

studies, looking at acute bronchodilator responses

as well as a variety of other asthma outcomes.

So, when this was cloned, the Asthma

Clinical Research investigators went back to the

BAGS trial that was still ongoing and were able to

get most of the participants to come back and be

genotyped. In that regard, the analysis showed

that there was no effect in this primary outcome at

the B27 locus. There was no effect in the B16

heterozygotes. However, there was a signal. When

the B16 Arg/Arg patients were compared to the B16

Gly/Gly patients, with a difference found in the

primary outcome variable.

So, this is a retrospective look at the

BAGS data that shows that if you were a group of

patients who received regular albuterol and you

were Arg/Arg, in yellow, your AM peak flow

deteriorated during the course of the trial, in

contrast to whether you received as needed beta

agonists and were Arg/Arg, in red, or whether you

received regular albuterol and were Gly/Gly. This

retrospective analysis was believed by the ACRN to

be hypothesis generating, not definitive and,

therefore, led to another study which I will share

with you.

At the same time, Robin Taylor reported on

the influence of beta adrenergic receptor

polymorphisms in some studies he had done looking

at, again, asthma exacerbations in this context.

If you look at the far right of all-comers in this

trial, you see that albuterol and salmeterol are

comparable and superior to placebo in preventing

exacerbations. If you look at the Gly/Gly and the

Gly/Arg groups, there were really no significant

differences. However, in those individuals that

were Arg/Arg at the B16 locus, you can see that

there were more exacerbations with those treated

with albuterol but this was not seen with the

salmeterol therapy.

So, we began to see in the asthma

community some signals, some subtle signals in

retrospective data about the issue of the potential

relevance of polymorphisms at the beta receptor.

Therefore, I told you that the Drazen trial,

retrospective, was hypothesis generating to allow

us to go forward to actually create a prospective,

randomized, placebo-controlled, double-blind trial

of regular versus minimal albuterol in each

genotype. This has affectionately been called the

BARGE trial.

In this trial, in order to minimize beta

agonist use, patients were provided with

ipratropium for rescue as a primary inhaler and

then had a backup to use albuterol of symptoms were

not relieved by ipratropium.

This is a fairly complex study design but

which we believed was important to answer the

question. First, individuals between the ages of

18 and 55 years of age who had an FEV

1 of at least

70 percent predicted, and naive to inhaled

steroids, were screened and genotyped. If they

were either found to be Arg/Arg or Gly/Gly at the

B16 they were matched on the basis of FEV

enrolled in the trial, went in a 6-week run-in

period in which individuals were all on placebo

with just rescue therapy. They were then

randomized to receive 16 weeks of active treatment

or placebo; then had an 8-week run-out; were

crossed over to the opposite trial; and then a

following run-out arm.

So, a complex study design that allowed

each patient to serve as their own control of being

on scheduled albuterol versus placebo and using the

backup rescue. These are individuals who were

about 31 years of age, had fairly normal FEV1's of

about 90 percent predicted and were matched in

pairs on the basis of the genotype of interest.

This is the data as published in Lancet.

This shows the curves of either the albuterol

modeled or raw means data versus the placebo

modeled and raw means data. In particular, if you

can look at the left-hand side of the slide, this

is the Arg/Arg group. The right-hand side is the

Gly/Gly group.

Let's look at the Gly/Gly group first. If

you look at the Gly/Gly patients in the orange line

on the top, you can see that, as you would expect,

those patients on albuterol scheduled therapy

improved by their morning peak flow during the

course of the study. In contrast, during the time

they received placebo, in green, they really showed

no improvement in their peak expiratory flow. In

contrast, the Arg/Arg patients behaved differently.

In green is the placebo and you can see the Arg/Arg

patients on placebo actually improved and those

Arg/Arg patients on albuterol, in orange, failed to

improve their peak flow during the course of the

trial.

The primary analysis with this study was

to look at the treatment differences and the mean

change in AM peak flow by genotype at week 16. You

can see that the albuterol versus placebo Arg/Arg

patients had a difference in their mean peak flow

of 10 L/min; the albuterol versus placebo Gly/Gly

comparison, a difference of about 14. Therefore,

the treatment difference of the mean Arg/Arg minus

the Gly/Gly was a difference of about 25 L/min,

which is what this study was powered to find and

what we had used in other studies to power it. So,

this was determined to be statistically

significant.

There were other outcomes that paralleled

the change in peak flow. This is looking at the

difference between regular versus placebo changes

in FEV 1 over the 16 weeks. You

can see that the

Gly/Gly subjects had an improvement in their FEV

whereas the Arg/Arg patients had a deterioration in

FEV 1. The same thing could be seen with

morning

symptoms of an increase in the Arg/Arg patients

versus a decrease in the Gly/Gly patients, and a

complementary pattern of seeing a difference in

inhaler use in the different groups, whether it be

ipratropium as first-line rescue versus albuterol.

In summary, the BARGE data concluded that

morning and evening peak flow, FEV1's, symptoms and

rescue inhaler use improved significantly in

Arg/Arg patients with asthma when beta agonists

were withdrawn, and when ipratropium was

substituted, as compared with regular albuterol

used. The pattern was reversed in the Gly/Gly

patients who actually improved with regular beta

agonist use. The authors suggested that Arg/Arg

patients, who are known to be one-sixth of

asthmatics, may actually benefit from minimizing

short-acting beta agonist use.

I included this study also because of the

important caveats from the investigators and their

conclusions. They emphasized that this study was

conducted in only individuals with mild disease,

not patients with concomitant inhaled steroid doses

and, therefore, whether this data can be

extrapolated to more severe disease or to those

patients who are on concomitant inhaled steroid

doses just could not be answered by this particular

trial, suggesting that both issues need to be

studied more in the asthma community.

Obviously, the million dollar question is,

indeed, do similar effects occur with long-acting

beta2 agonists, and what is the impact of

concurrent use of inhaled steroids? Obviously, Dr.

Chowdhury addressed the committee to really

deliberate today to answer those questions. I

don't have the answers for you and, fortunately, I

am not charged to do that. That is your tough job

today.

I would have some comments on what I

believe to be future studies that may help you to

answer those questions. Much as the BAGS trial was

hypothesis generating for BARGE, the SOCS and SLIC

trial from the ACRN did retrospectively look at

their two studies of long-acting beta agonists

alone. That was the SOCS trial that I shared with

you, and the SLIC trial which looked at combination

of inhaled steroid and long-acting beta agonists

and the tapering of such.

The data from these two retrospective

studies has been presented at meetings, suggesting

that there was a signal of a same pattern of a

difference in morning peak flow based on whether

you were Arg/Arg or Gly/Gly at the 16 locus, and

that the pattern with salmeterol, with or without

the inhaled steroid, seems to be the same.

I carefully indicated that, indeed, these

are retrospective studies, very small in design

and, clearly, will be hypothesis generating for

more robust, longer-term studies that the ACRN, and

I believe the industry, will conduct. Therefore,

the ACRN now has a study called LARGE that is in

the middle of operation that is very similar to the

BARGE study but will look at an inhaled steroid,

with or without the addition of a long-acting beta

agonist, to answer the question of whether the same

patterns in a prospective, carefully designed study

can be extrapolated.

Now, we do have some data to answer the

question on a safety issue about does regular use

of long-acting beta agonists delay awareness of

asthma progression or effect from recovery? We

have been concerned that if patients are so well

controlled with symptoms with their long-acting

beta agonists will they be aware that they are

having an asthma exacerbation, or will they fail to

recover from an exacerbation in the way that they

expected to?

This is one representative study that I

think illustrates the point. This is the Matz

article I showed you earlier of an accumulation of

data from earlier published studies. At the arrow,

the day of diagnosis is the point in time at which

the patient had an asthma exacerbation as defined

by these authors. You can see that if you look at

the change in asthma symptom score about four days

or so before the actual diagnosis of an

exacerbation these individuals began to have an

increase in symptoms, were treated in completion of

an exacerbation satisfactorily, and you see that

their symptoms decreased after the exacerbation.

In this particular trial you actually see that

there is a change in the asthma symptom score that

was different in the two different study groups.

Now, one of the things that this provides

I think is some reassuring issues that on the basis

of symptoms patients are well able to detect a

difference in their symptoms, and to know whether

they are having an asthma exacerbation, and they

recover as we expect. There seems to be no adverse

effect of the addition of salmeterol. In fact,

these patients seem, by symptoms, to recover even

quicker.

We did the same analysis with the PACT

pediatric trial that I shared with you just for

interest, to do the same pattern looking at

symptoms, the issue of albuterol use and the issue

of peak flow. We plotted the three arms of the

study to look at whether the patterns were any

different. In relevance to you today, the patients

who were on combination therapy as compared to

inhaled steroid alone had no difference in their

pattern. So, all three groups were equally able to

perceive symptoms of an exacerbation and to

adequately recover in the same kind of a pattern.

So, we are beginning to, I think, have more data

that resolves this concern that has been raised.

Now, why we are here today in particular

is to discuss the evidence for increased severe

asthma exacerbations with long-acting beta

agonists. Indeed, for these studies, as Dr.

Chowdhury outlined for you, the major issue at hand

is, indeed, severe asthma exacerbations as has been

defined by these trials. I am not going to review

them for you as you clearly have received

preliminary information and I suspect you will have

other members of the audience that will provide far

better detail of these than I can do. Suffice it

to say that these are studies that have raised

questions in the asthma community about the role of

long-acting beta agonists, and my own particular

comment on these is the fact that, whereas they are

compelling for a signal and certainly warrant a

very careful review of the trials of what they can

tell us and what they cannot tell us, it is very

difficult from these trials to discern whether

these individuals were, indeed, using concurrent

inhaled steroids during the course of the trial.

Therefore, it makes it certainly somewhat difficult

to do a full analysis and, therefore, no questions

are easily answered.

In summary, I think the committee today

has a very important job of reconciling what I

believe to be a very crucial question. How do we

all reconcile the finding of these very rare

severe, life-threatening episodes that are reported

in the SMART an formoterol trials with what I hope

to have shown you is obviously the far more global

evidence that the use of long-acting beta agonists,

particularly in combination with inhaled steroids,

results in a decrease of overall asthma

exacerbations? You all are faced with the data

that I believe show that there is very strong

evidence of the ability of inhaled steroids and

long-acting beta agonists to both achieve asthma

control and to reduce overall asthma exacerbations,

as defined by the trials that I have shown you and

others. So, that piece of data needs to be kept in

context.

I would comment that there clearly is more

evidence in adults than children so most of the

decisions made are based on adult data. I believe

that the remaining concerns about safety have to

ask the question about whether, indeed, there is an

influence of genotypic predictors, as has been

picked up as the signal with the intermediate beta

agonists. I believe that we have to look at

phenotypic predictors.

I think the era of treating all patients

equally for asthma is gone and we need to gain

insight about phenotypic predictors of responses to

all our therapy. I think this needs to include

age, severity of disease, bronchodilator

reversibility status, ethnicity and a variety of

others. Clearly, we have had some signals that

there may be ethnic differences in responses to

albuterol based on whether you happen to be Puerto

Rican or Mexican-American. So, we need to get more

information.

We need to have larger and longer trials

which incorporate multiple outcomes, including the

concurrent use of inhaled steroids, and we need to

be able to ultimately answer questions of whether

this is a class effect of a dose effect.

I don't envy the committee. I know that

you will deliberate carefully. And, I appreciate

you allowing me to provide you an overview in

anchoring your thoughts for your deliberation.

Thank you very much.

DR. SWENSON: Dr. Sorkness, I want to

thank you for a very fine talk. Since you are

going to be leaving before the day is out, I wanted

to particularly leave some time for members of the

panel to ask you questions at this moment. So, we

will take questions from the panel on the talk or

issues around it.

Questions for the Speaker

DR. MARTINEZ: Thank you so much for that

very, very nice presentation. During your

presentation you said that in the PACT trial you

were the principal investigator within the CARE

network. The decision was made, you said, to use

methacholine responsiveness as a criterion for

inclusion into the trial and not reversibility. I

am trying to quote you as best as I can, because

this could have introduced bias into the results,

unquote.

DR. SORKNESS: Yes.

DR. MARTINEZ: Are you suggesting that

some of the results of the studies that you have

shown to us, including the GOAL study in which

exacerbation was shown to be less in combination

than in use of inhaled corticosteroids alone, may

be explained by bias introduced by the fact that,

for example, in the GOAL study 15 percent

reversibility was a criterion for inclusion?

Or, a second question, has anybody tried

to separate the studies in this meta-analysis that

you presented to us between those that demanded 15

percent reversibility and those which did not?

DR. SORKNESS: It is a great question,

Fernando, and I think it allows me to clarify my

intent of saying that. Clearly, because of a

variety of reasons, whether it be historical of our

belief that bronchodilator reversibility convinces

us that this is, indeed, reversibly asthma and,

therefore, the documentation of such allows

enrollment into a clinical trial, or it convinces

us that reversibility allows other drugs to show

comparability. The majority of trials, whether

they be industry sponsored or not, clearly have

used bronchodilator reversibility as entry

criteria, and clearly most of that which I shared

with you is that. That has historically been the

context.

My point in this the fact that I believe

that there are a much broader group of asthmatics

in the world today that don't have that much

bronchodilator reversibility or may have very

little and truly have asthma. So, our assumptions

of our outcomes in the therapies are predicated on

the fact that we tend to enroll a fairly defined

population.

I think, second to that, there is

certainly some data from ACRN and other groups that

bronchodilator reversibility as a phenotype clearly

may be more predictive of response to long-acting

beta agonists or for inhaled steroids, for that

matter. So, we have isolated a particular

phenotype and enrolled them in our trials.

The ACRN, because of that and I think

because of our mission of trying to more globally

answer questions in a broader asthma population, in

general have suggested that people can be in these

studies whether you have a bronchodilator response

or PC 20 as evidence of having asthma.

Both issues

are collected by entry is not predicated on having

simply a bronchodilator effect.

In the PACT trial I wanted to emphasize

that I think, because of at least some concerns

about the generalizability of the PACT results, we

felt that PC 20

predominantly was the right entry

criteria. Bronchodilator reversibility was

collected. And, clearly, the PACT data will have

the capability of looking at both genotypic and

phenotypic predictors of responses. I can say that

about the PACT data. I haven't, Fernando, really

been privy to know whether many of these other

studies teased out bronchodilator responsiveness.

So, that is my answer to the question.

DR. SWENSON: Just for the record, that

was Dr. Martinez that posed that question. Dr.

Sorkness, to what extent are the exacerbations, as

they are detected in these multiple studies, based

on the criteria of increased use of a short-acting

beta agonist or the rescue use? Because that seems

pertinent to the question of whether long-acting

beta agonists simply just, for a while, reduce the

need for short-acting and so allow whatever

underlying process toward exacerbation to go

further without recognition.

DR. SORKNESS: As a very general comment

to this, it is a difficult question to answer

simply because whether it be asthma exacerbations

or treatment failure there is clearly, in my mind,

not a uniform definition of either in the trials

that have been described. I think, in fairness,

the vast majority of at least the mild and leading

on to the use of prednisone exacerbations in

general have been anchored by asthma action plans

that have been a combination of symptoms, albuterol

use and, on some occasions, peak flows below some

safety criteria. So, many of these studies have at

least incorporated an asthma action plan of

albuterol symptoms and peak flow leading to the use

of prednisone. So, I think it becomes kind of a

composite decision that the patient makes in

concert with the physician for those studies.

Having said that, the vast majority of the

studies, at least in my mind, that have used the

term asthma exacerbation in general have been

defined by the need for prednisone, with or without

in some cases either an ER visit or a

hospitalization, but certainly the asthma

exacerbation in many of the studies could have been

achieved simply by the issue of prednisone by that

action plan. But the definitions are very variable

and I think that does make it harder to bring all

of these together to get the best insight.

DR. SWENSON: Miss Sander?

MS. SANDER: Thank you. I need a little

bit more information on what you just said.

Whenever there is the term "rescue" medications

used, that is any and all reasons not just rescue

examples?

DR. SORKNESS: I am not sure I understand,

Miss Sander.

MS. SANDER: So, rescue would imply that

they had an emergency need for that medication.

Would it include all uses such as early

intervention, prevention of exercise?

DR. SORKNESS: In my mind, most of the

studies have in their action plans specified that

the use of albuterol to relieve symptoms and/or to

treat a peak flow at a certain safety level were

used in the definition of an action plan of going

on to treat the exacerbation. Most of the action

plans in these trials, or at least the ones

certainly from the ACRN and CARE, did not

incorporate pre-exercise intended scheduled

albuterol use in that paradigm. It was strictly

albuterol use for relief of those symptoms or

relief of a drop in a peak flow to make it return

to some baseline safety level.

MS. SANDER: Thank you. Also one other

question, were there any expectant mothers in any

of these?

DR. SORKNESS: I can't say this with

absolute confidence but I would be highly suspect

that any of the trials were conducted that did not

have a safety pregnancy test at entry and did not

have some appropriate monitoring of pregnancy

status during the trial. The vast majority of

studies that have been privy to even mandate that

if a methacholine challenge procedure is being done

at a study visit a pregnancy test be done. There

is a series of questions that coordinators and

investigators ask about the chance of a pregnancy

to make decisions as far as people continuing in

trials. So, I would be very surprised if

individuals were enrolled being pregnant.

Unfortunately, life is not perfect and I think that

there are certainly trials where a woman became

pregnant during the trial. Most of the studies I

know of, that actually required a mandated

withdrawal because of the potential influence of

pregnancy on stability of asthma. So, I don't

think there is much we can gain in insight, quite

honestly, if that is some of what you are driving

at. I just don't think it exists in these trials.

DR. SWENSON: Dr. Newman?

DR. NEWMAN: Yes, thank you for what was a

very clear presentation. I wonder if you might

comment about, from the benefit side, any

differences in these trials based on race.

DR. SORKNESS: That is a tremendous

question. I think the fairness of answering the

question is that most of the trials that I am aware

of--and I say this carefully because I don't know

the literature in its extreme--probably did not

have the ability to have a satisfactory subset of a

particular racial or ethnic group to be able to

cull out to do a reasonable racial analysis. In

the beta agonist trial by Drazen, et al., I know

for a fact that because of NIH NHLBI guidelines of

enrollment of at least a third of minority

participants, that we did do a statistical analysis

in that trial and it showed that the minority

ethnic group did not do differently on any of the

outcomes versus Caucasians, negative or benefit.

They had equal responses, as did actually a gender

analysis.

I really do not know of any other trial

that could answer your question explicitly but I

think it is very important, especially given some

of what we are learning about the potential role of

ethnicity, and that mandates that we all make a far

more serious effort for doing trials big enough

with groups to answer the question.

DR. SWENSON: Dr. Brantly?

DR. BRANTLY: Dr. Sorkness, as I recall

there were a number of bronchial biopsy studies

using ICSs. I don't recall any regarding using

either long-acting beta agonists or short-acting

beta agonists. Do they exist?

DR. SORKNESS: I am not sure I can answer

that with comfort. I actually do believe that

there are bronchial biopsy studies in individuals

on long-acting beta agonists alone and certainly on

combination. That is not clearly my area of

expertise and I really think I would be remiss in

trying to answer the question of what I know about

those studies. I am a clinical researcher.

Certainly, some of my partners do those kinds of

studies but that is clearly not an expertise that I

would feel comfortable answering. And, there may

be somebody else on the committee that clearly

knows that data far more than I.

DR. SWENSON: Dr. Prussin?

DR. PRUSSIN: Chris, on your last slide

you have a note that says, "need for larger and

longer trials which incorporate multiple outcomes."

My question is, you know, clearly long-acting beta

agonists decrease exacerbations and, yet, we have

very good data that severe pulmonary events and

death are increased. So, you can't use a trial

that is looking at exacerbations to answer the

outcome that we are interested in here. Since I

work more in a smaller frame in terms of allergic

disease, not large clinical trials, can you give me

more of an idea of what you think a large clinical

trial and multiple outcomes that we should be

looking at for these endpoints of death,

intubation, severe pulmonary outcomes?

DR. SORKNESS: Cal, I think it is

difficult and I will try to answer as best I can.

I do believe we are in an era where the most

important studies are not monotherapy in asthma

with long-acting beta agonists but with

combination. So, that is the first issue.

I believe that whereas the SMART trial and

some of the formoterol pivotal trials and others

that have raised the signal of concern are helpful

and we need to take that under consideration. I

find that the way that those studies were

constructed leave me wanting more. The methods by

which patients were accrued; the issue of whether

you really knew whether people were on inhaled

steroids concurrently and were adherent with such;

that you took into account and balanced severity of

disease at the beginning; that you truly looked at

what we believe clinically as the best that we can

ask of this array of overall asthma exacerbations

and control of disease; a year long study to deal

with seasonality, especially in kids; looking at

some markers of inflammation.

I think that we are at a stage that we

would feel better and have more confidence in the

risk/benefit relationship if we had those kinds of

trials done both in adults and children, and

particularly were able to answer in our own minds

whether the combination together--adherence, people

taking them, being on them, controlling for the

issues--that we really knew what we were doing with

those particular trials. And, I think that is the

best that we can do.

DR. PRUSSIN: Let me just follow that up.

The SMART trial was stopped because of difficulties

with accrual and slow accrual. Again, we are

talking about a huge clinical trial. In your

estimation, since this is what you do, is it

possible to do that large a trial and get the

information in a much more rigorous way, as you are

proposing? I mean in terms of accrual. Is this a

feasible endeavor to go into? Because we have been

told that SMART simply became impossible to carry

forward.

DR. SORKNESS: Yes, I think the reality is

such that it is I believe, and I am investigator so

I am asked to do these things--I think it is

impossible in this day and age to recruit large

enough subjects even in a multicenter study that

are naive to either inhaled steroids or long-acting

beta agonists at entry so that you are bringing in

this naive population to answer the question. I

don't think those patients are out there anymore

because we have done such a good job with

guidelines and because all these trials showing

that when you give people good medicines, by golly,

they get better.

So, I think that if you, indeed, enroll a

far broader population of phenotypes, of patients

that have certain entry criteria, and then you

randomized them to an inhaled steroid with and

without a long-acting beta agonist, and followed

them for long enough, I think those studies can be

constructed. And, I think that is one of the

challenges to do and I suspect that they will be

done.

DR. SWENSON: Well, thank you, Dr.

Sorkness. We appreciate very much that fine talk

and discussion. At this point we will turn the

program now over to GlaxoSmithKline and, to do so

and to introduce her colleagues, Elaine Jones will

take over.

GlaxoSmithKline Presentation

Opening Remarks

DR. JONES: Good morning. My name is

Elaine Jones and I am Vice President of Regulatory

Affairs at GlaxoSmithKline. On behalf of

GlaxoSmithKline, I would like to thank the agency

and the advisory committee for this opportunity to

review data pertinent to the discussion of the

safety of long-acting beta

2 agonists in the

treatment of asthma.

Our presentation today will focus on our

data with the inhaled long-acting beta

2 agonist

salmeterol. As we begin the presentation today, I

would like to set the stage by speaking first about

the burden of asthma. As the committee members are

well aware, asthma is a chronic disease associated

with significant morbidity and mortality. In the

United States alone asthma affects approximately 20

million patients. Asthma exerts a tremendous

societal burden as evidenced by the half million

hospitalizations and over 4,000 deaths in the U.S.

in 2002.

There are many risk factors that have been

identified that put patients at risk for an

asthma-related death. Some of these include

excessive reliance on rescue medications and use of

inhaled corticosteroids, disease severity and a

delay in seeking care. Ethnic origin is also an

important risk factor, demonstrated by the fact

that the rate of asthma deaths in African Americans

is approximately 2.5-fold higher than that of

Caucasians.

The tremendous burden of asthma has fueled

a continual development of new medications to treat

this disease and GSK has a long history in the

development of respiratory medicines. Salmeterol

was the first inhaled long-acting bronchodilator,

and its approval in the United Kingdom over a

decade and a half ago represented an important

advance in the management of asthma.

To date, regulatory authorities have

granted approval to market salmeterol in over 100

countries. In the United States there are three

salmeterol-containing products that have been

approved for marketing. These are Serevent

inhalation aerosol, Serevent diskus and Advair

diskus which contain salmeterol as one of its

components. Each one of these products has been

approved for use in patients with asthma or COPD,

and each of these approvals required a full

clinical development program.

It should be noted that the inhalation

aerosol formulation, which contained

chlorofluorocarbons, has been discontinued by GSK

as part of the phase-out of CFC-containing products

consistent with the Montreal protocol.

Worldwide approvals by regulatory

authorities have led to a great deal of clinical

experience with salmeterol. Over the last 15 years

the exposure to salmeterol is the result of the use

of salmeterol formulated as a single agent and the

use of salmeterol formulated with fluticasone

propionate in a single device. In total the

worldwide exposure is now estimated at 45.2 million

patient-years.

Based on extensive clinical experience and

a systematic review of numerous clinical trials,

evidence-based guidelines from the National Heart,

Lung and Blood Institute's expert panel report

recognize the pivotal role of long-acting beta

agonists in the treatment of asthma. While the

safety of long-acting beta

2 agonists is the topic

of today's meeting, it is important to consider the

safety of these medications in the context of their

overall benefit/risk profile. Part of the context

is provided by current asthma treatment guidelines

which position the use of inhaled long-acting beta

agonists with inhaled corticosteroids as the

preferred treatment option for patients with

moderate to severe persistent asthma.

Asthma is a serious disease with

significant morbidity and mortality and salmeterol

has become a well-established pharmacological

therapy in the management of this disease. As you

know, no medication is without risk and today's

meeting provides an important opportunity to review

safety data for inhaled long-acting beta agonists.

We look forward to discussing the safety of

salmeterol with the committee.

Salmeterol has been shown to be highly

effective in the treatment of asthma and, since its

approval 15 years ago, clinicians have accrued

considerable experience with its use. Based on the

extensive body of evidence in patients with asthma,

including 64 studies in approximately 45,000

patients in the U.S. alone, GSK believes that

salmeterol continues to exhibit a favorable benefit

to risk profile.

Dr. Kate Knobil will now provide a brief

overview of the efficacy of salmeterol, followed by

a discussion of safety data. Following Dr.

Knobil's presentation, I will return to the podium

to introduce the experts here with us today and

then we will take questions from the committee.

Salmeterol Review

DR. KNOBIL: Good morning, everyone. For

my presentation today I will first present a brief

overview of the benefits of salmeterol for the

treatment of asthma, followed by a review of the

salmeterol safety data. My review of the safety

data will focus on the postmarketing safety

surveillance studies, SNS and SMART, and the

results from epidemiology studies of salmeterol.

In addition, I will describe the ongoing studies

currently being conducted by GSK to further

evaluate the efficacy and safety of salmeterol.

Finally, I will close with an overall assessment of

salmeterol for the treatment of patients with

asthma. Given time limitations, I will not be able

to cover all of the information that is in your

briefing document. However, any questions you may

have may be addressed during the Q&A.

For several decades beta agonists have

been widely used to treat bronchoconstriction.

This slide shows the structures of albuterol and

salmeterol, and you have seen these already today,

and highlights the long lipophilic tail that helps

anchor salmeterol in the beta adrenergic receptor.

Albuterol is highly selective for beta

2 receptor,

thus having fewer cardiovascular effects than

earlier less selective beta agonists. Short-acting

beta agonists are very effective but are limited by

their relatively short duration of action of 4-6

hours.

This limitation was largely overcome by

the development of selective long-acting beta

agonists, such as salmeterol, which are effective

for at least 12 hours. In addition to having a

longer duration of action, in vitro studies have

shown salmeterol to be at least 50 times more

selective for the airway beta

2 receptor than

albuterol.

The benefit of the longer duration of

action of salmeterol can be seen in the data pooled

from the two registration studies for salmeterol

metered dose inhaler. At the time that these

studies were conducted regular albuterol use was a

common treatment for asthma and so was included as

an active comparator. For salmeterol, shown in

green, a single dose results in a clinically

significant improvement in FEV

1 within 30 minutes,

with maintenance of effect for at least 12 hours.

This is in contrast to albuterol, shown in grey,

which has a more rapid onset of effect but the

bronchodilator effect lasts only 4-6 hours.

Additionally, as shown on the right, the

bronchodilator effect of salmeterol was maintained

after 12 weeks of treatment with no diminution of

FEV 1 response over time.

Studies of up to one year in duration have

confirmed that the bronchodilator properties of

salmeterol are maintained with long-term use. In

this study, 12-hour FEV

1 area under the curve, or

AUC, was obtained after the first dose and

following 8, 20 and 48 weeks of treatment. For

salmeterol the mean FEV

1 AUC was similar at all

time points, and in all cases was significantly

greater than placebo, demonstrating maintenance of

bronchodilator effect. In addition to important

bronchodilator effects, salmeterol is very

effectiveness at reducing the symptoms of asthma.

The data shown here are from the same two studies

for salmeterol MDI that I showed previously. Over

12 weeks treatment with salmeterol resulted in a

significant reduction in asthma symptoms scores for

chest tightness, shortness of breath, wheezing and

cough compared with placebo and albuterol given 4

times daily. Although not shown here, in these and

other studies salmeterol also reduced nocturnal

symptoms associated with asthma.

Salmeterol has also been shown to be an

important treatment option for patients with asthma

who are not adequately controlled on inhaled

corticosteroids. This landmark study by Greening

and colleagues examined the effect of adding

salmeterol to inhaled corticosteroid therapy, in

this case beclomethasone, as compared to increasing

the dose of inhaled steroids. The addition of

salmeterol to a low dose of inhaled corticosteroid

was shown to result in a greater improvement in

lung function, as shown by peak expiratory flow,

then when compared to the higher dose of inhaled

steroids. In addition to the improvements in lung

function, the use of salmeterol resulted in greater

improvements in symptoms and rescue albuterol use.

The addition of salmeterol to a low to

medium dose of inhaled steroid has also been shown

to reduce the recurrence of asthma exacerbations.

Shown here again is the study by Matz and

colleagues, and was of similar design to the study

that I showed previously. When compared to

increasing the dose of fluticasone propionate, or

FP, the addition of salmeterol to the low dose of

FP significantly increased the time to the first

asthma exacerbation requiring oral corticosteroids.

Further, significantly fewer salmeterol-treated

patients experienced one or more exacerbations, 8.8

percent compared to the increased dose of FP at

13.8 percent of patients.

Another means of evaluating the patient

benefit of a medication is to assess the impact on

quality of life. In this 12-week study that was

designed to assess asthma-specific quality of life,

patients with asthma were randomized to either

salmeterol or placebo, with all patients receiving

albuterol as needed to use for symptoms.

Salmeterol MDI was shown to significantly improve

quality of life compared with placebo, and the

minimally clinically important difference of 0.5

was achieved for each domain as well as the global

score.

To summarize, the benefits of salmeterol

have been well established and salmeterol has been

accepted as having an integral role in the

treatment of asthma.

I will now move on to the safety portion

of the presentation, beginning first with the

postmarketing surveillance studies for salmeterol.

These studies are of interest because at the time

of launch of salmeterol in both the U.K. and in the

U.S. there was concern that the regular use of beta

agonists may lead to deterioration of asthma

control. This was based primarily on studies of

short-acting beta agonists, particularly fenoterol,

that suggested worsening of asthma with scheduled

use. These studies could not determine a cause and

effect relationship, however, they did bring

significant attention to the appropriate use of

this class of medications.

The first study that I will discuss is the

Serevent Nationwide Surveillance Study, or SNS,

which was performed in the U.K. between 1990 and

1992. This 16-week randomized, double-blind study

evaluated over 25,000 patients with moderate to

severe asthma. The study compared salmeterol MDI

to albuterol given 4 times daily in patients 12

years of age and older. Both treatments were added

to the patient's current asthma therapy.

At visit 1 patients were randomized in a

2:1 fashion to either salmeterol or albuterol. The

primary endpoint for SNS was combined serious

adverse events and all medical and non-medical

withdrawals. This very broad endpoint was not

restricted to respiratory events. For this

endpoint the percentage of events was similar for

the salmeterol and albuterol groups. Additional

endpoints of interest included asthma-related

deaths, hospitalizations and withdrawals. Based on

national health statistics in the U.K. and on the

2:1 randomization, 10 and 5 asthma-related deaths

were predicted in the salmeterol group and

albuterol group respectively.

In this study, 14 asthma-related deaths

occurred, with 12 in the salmeterol group and 2 in

the albuterol group, resulting in a relative risk

of 3. This difference was not statistically

significant but did raise concern. The results for

asthma-related deaths were not consistent with the

data for asthma-related hospitalizations. As you

can see here, the data for this endpoint did not

indicate an increase in risk with salmeterol. The

only statistically significant difference between

the groups was seen for the percentage of

withdrawals due to worsening asthma, with a lower

percent observed in the salmeterol group compared

with albuterol.

In light of the results of SNS, including

the asthma-related deaths and spontaneous reports,

GSK, in conjunction with the FDA, designed the

Salmeterol Multicenter Asthma Research Trial, or

SMART. The study was initiated in 1996. SMART was

a randomized, double-blind surveillance study of 28

weeks duration that was conducted at over 6,000

sites in the United States. Patients with asthma

who were 12 years of age or older, with no previous

use of inhaled long-acting beta agonists, were

included. All other asthma medications were

allowed during the study.

SMART consisted of a single clinic visit

at which patients were assessed for eligibility and

then randomized to receive either salmeterol or

placebo which was added to their usual asthma care.

Subjects were given a 28-week supply of study

medication and were not required to return for

clinic visits. Instead, patients were contacted

every 4 weeks by phone primarily to collect

information on serious adverse events, including

respiratory-related events.

The combined endpoint of

respiratory-related deaths or life-threatening

experiences was chosen as the primary endpoint.

Asthma-related death was also of interest but

because this is a rare event the sample size

required for this to be the primary endpoint was

too large to be feasible. Even with the broader

combined endpoint, it was determined that a sample

size of 30,000 patients would be required.

However, after 15,000 patients were enrolled in the

study the actual rate of primary events was found

to be approximately half of what was expected and

the target sample size was increased to 60,000

patients.

Key secondary endpoints were

respiratory-related deaths, combined asthma-related

deaths or life-threatening experiences, and

asthma-related deaths, all of which were subsets of

the primary endpoint.

Two independent review committees were

involved with SMART. They were the mortality and

morbidity review committee, or MMRC, and the data

safety monitoring board, or DSMB. Each serious

adverse event was adjudicated in a blinded fashion

by the MMRC to determine if it was respiratory

related and, if so, whether it was asthma related.

The categories for this adjudication were

unrelated, unlikely related, possibly related or

almost certainly related. Only respiratory- and

asthma-related events considered possibly related

or almost certainly related comprised the primary

and secondary endpoints. The DSMB met regularly to

evaluate blinded aggregate data which included the

cases adjudicated by the MMRC.

An interim analysis was planned when

approximately one-half of the patients had been

enrolled. At the interim analysis the study did

not reach predetermined stopping criteria, however,

there was a suggestion of worse outcomes in

salmeterol-treated patients, especially African

Americans. For this reason, the DSMB recommended

that ideally the study should be completed within 2

years or, if that was not possible, the study

should be terminated and the results disseminated.

Following discussions with the DSMB, GSK made the

decision to stop the study due to difficulties in

enrollment and the findings in African Americans.

I will now move on to the results of

SMART. Overall, the baseline characteristics of

age, sex, ethnic origin and baseline peak

expiratory flow were well matched between the

treatment groups. Approximately 70 percent of the

population was Caucasian and 18 percent was African

American. For reference, approximately 15 percent

of the patients with asthma in the United States

are African American.

Asthma medications were reported at

baseline and were similar between the treatment

groups. The most commonly reported asthma

medications were inhaled or oral beta agonists

which were reported in over 90 percent of patients.

Forty-seven of the patients reported use of inhaled

corticosteroids at baseline.

While baseline characteristics were

similar between the treatments for the total

population, this was not the case when comparing

baseline characteristics between Caucasians and

African Americans. The baseline characteristics

indicate that African Americans had more severe

asthma as measured by peak expiratory flow,

nocturnal symptoms, and history of hospitalizations

and intubations. For example, the proportion of

African Americans reporting a hospitalization for

asthma during the previous 12 months was more than

twice the percentage reported for intubation for

asthma in their lifetime. In addition to these

markers of increased severity in African Americans,

the reported use of an ICS at baseline was lower

than that in Caucasians.

The results for the primary and key

secondary endpoints will be shown on this slide.

Due to the amount of information, I will take a few

moments to summarize the data. These figures are

also available in your briefing document for

reference.

First let me orient you to the slide. The

relative risk point estimate and corresponding 95

percent confidence intervals for the primary and

secondary endpoints will be displayed graphically.

The values that correspond with these data will be

shown on the right side of the slide. The total

population will be represented in yellow, the

Caucasian subgroup in green, and the African

100

American subgroup in orange.

I will start by showing the results for

the total population as this was the primary

analysis. Then I will show the results for

Caucasians and African Americans as this post hoc

analysis was requested by the DSMB at the time of

the interim analysis.

The number of primary events, combined

respiratory-related death or life-threatening

experiences, was approximately two-thirds of what

was expected. The primary endpoint for the total

population, as shown here on the slide, was not

statistically significantly different between

treatment groups as the confidence interval

includes 1. As I review the key secondary

endpoints for the total population, which are

respiratory-related deaths, combined asthma-related

deaths or life-threatening experiences and

asthma-related deaths, it is important to remember

that each is a subset of the primary endpoint.

For the secondary endpoints, statistically

significant differences were observed between

101

treatment groups for the total population,

including asthma-related death which I will discuss

in more detail in a moment. The numbers of primary

events in the Caucasian subgroup, shown in green,

were similar between the treatment groups.

However, in the African American population, shown

here in orange, a significantly greater number of

primary events occurred in the salmeterol treatment

group.

For the key secondary endpoints the

relative risk of events was higher in African

Americans compared with Caucasians. In particular,

a significantly greater number of combined

asthma-related deaths or life-threatening

experiences occurred in the salmeterol group in the

African American population, while there was no

difference between treatment groups in the

Caucasian population.

The number of asthma deaths in SMART was

approximately half of what was expected. There was

a significantly higher number of asthma-related

deaths seen in the overall population for patients

102

receiving salmeterol compared with placebo and the

same pattern was seen in the ethnic subgroups.

While the relative risk of asthma deaths appears

similar between ethnic groups, note that there were

approximately 4 times as many Caucasians in this

study than African Americans. Therefore, the rates

for all asthma-related endpoints were much higher

in the African American population.

The effect of inhaled corticosteroids was

also of particular interest to the DSMB at the time

of the interim analysis. A post hoc analysis was

conducted to explore the association of baseline

use of ICS with the primary and key secondary

outcomes. As I mentioned previously, 47 percent of

the patients reported using inhaled steroids at

baseline. The results for the total population are

shown here, again in yellow, for reference.

Results for subjects reporting inhaled

corticosteroid use at baseline will be shown in

blue, and those not reporting ICS use at baseline

will be shown in white.

For subjects reporting ICSs at baseline

103

there were not statistically significant

differences between the treatment groups for the

primary and secondary outcomes. Patients receiving

salmeterol who did not report the use of inhaled

corticosteroids at baseline, here in white,

experienced significantly more combined

asthma-related events than those receiving placebo.

The number of deaths in those patients not

reporting inhaled corticosteroid use at baseline

was 9 in the salmeterol group versus zero in the

placebo group so direct calculation of relative

risk cannot be performed. In the patients

reporting corticosteroid use at baseline the

numbers were 4 and 3 respectively.

Although SMART was not designed to assess

the effects on inhaled corticosteroid use, these

data suggests that ICS may have had a beneficial

effect on asthma outcomes in SMART.

Finally, the data were analyzed by both

ethnicity and inhaled corticosteroid use reported

at baseline. Caucasians are shown, again, in green

and African Americans in orange. Patients

104

receiving inhaled corticosteroids are represented

by solid lines while dotted lines represent

patients not reporting inhaled corticosteroid use

at baseline. The relative risk for the primary

endpoint was higher in African Americans than

Caucasians, and those not reporting inhaled

corticosteroids at baseline had higher relative

risks within those populations.

Similar to the primary endpoint, the

relative risk for combined asthma-related events

was higher in the groups that did not report

inhaled corticosteroids at baseline independent of

ethnicity.

If we focus specifically on the number of

asthma-related deaths, shown here at the bottom of

the slide, it is evident that these events were

rare. There were more asthma-related deaths in

patients receiving salmeterol who did not report

ICS use at baseline in both Caucasians and African

Americans. Again, direct calculation of relative

risk cannot be performed for asthma-related deaths

for patients not reporting inhaled corticosteroids

105

at baseline since there were no deaths in the

placebo group for this endpoint.

SMART was not designed to determine the

effect of inhaled corticosteroids and ethnicity on

these endpoints, and the number of events in each

subgroup is quite small. Therefore, these data

should be interpreted carefully.

In summary, there were more events,

including asthma-related deaths, reported in the

patients receiving salmeterol. There was also a

suggestion that both African Americans and patients

who did not report using inhaled corticosteroids at

baseline had a higher risk of asthma-related

events. However, the number of events in SMART was

lower than expected, preventing definitive

conclusions from the data.

A careful review of the data did not

reveal any clear explanation of the results.

Possible explanations include a direct

pharmacologic effect of salmeterol; the presence of

polymorphisms in the beta receptor gene; or

patient-related factors, including a delay in

106

seeking medical care. It is well accepted that the

prevalence of patient-related risk factors is not

equally distributed across ethnic groups so the

differences in outcomes seen between the ethnic

groups in SMART may be associated with disparities

in access to medical care and asthma management and

may not reflect biological differences between the

groups. Unfortunately, none of these hypotheses

can be confirmed or refuted by the data from SMART.

While there are no clear explanations for the data,

the findings were communicated to physicians to

allow for informed treatment decisions.

In collaboration with the FDA, a number of

activities were undertaken to communicate the

results in order to inform physicians about SMART.

On the day the study was stopped a "dear healthcare

professional letter" was delivered by overnight

mail to the 229,000 healthcare professionals in the

United States who had prescribed salmeterol or

salmeterol-containing products within the previous

year. Simultaneously, notices on both the FDA and

GSK web sites were posted.

107

A second letter was sent out to health

professionals when the prescribing information for

Serevent and Advair was changed to include the

preliminary results of the interim analysis of

SMART. The information was elevated to the highest

level of prominence in the form of a boxed warning.

When the final results were obtained the labeling

for both products was updated.

This is the boxed warning that was added

to the prescribing information for Serevent and

Advair. It describes the final results of the

interim analysis of SMART. For your reference, the

full label, including the boxed warning, is

available in your briefing package.

The results of the interim analysis were

presented at the American College of Chest

Physicians meeting as a late-breaking abstract.

This was the first available national meeting with

high attendance of respiratory physicians. The

manuscript is now in press at Chest, the journal of

the American College of Chest Physicians.

Epidemiology studies offer an additional

108

method to investigate associations between drug

exposure and serious outcomes. The major advantage

of these studies is the utilization of

comprehensive medical and pharmacy databases.

These databases allow identification of a greater

number of events than can be achieved in

traditional randomized clinical trials. The

primary limitation of observational studies is the

fact that assignment of treatment is not random and

treatment effects may be confounded by differences

in baseline characteristics, including co-morbid

disease, differences in asthma severity and

selective prescribing. Since many more events can

be evaluated in an observational design, this may

be a more informative way to assess treatment

effects on the rare endpoint of asthma-related

death.

This figure displays the relative risks

from all large published cohort and case-control

studies that evaluated whether salmeterol use was

associated with the occurrence of severe

respiratory and asthma-related outcomes. The

109

dotted line represents a relative risk of 1.

The first study determine the relative

risk of respiratory-related death among patients

with asthma receiving salmeterol compared with

those receiving theophylline on the left side of

the highlighted area, or those receiving

ipratropium, which is on the right side of the

highlighted area.

The second study, which was conducted in

the United States, evaluated three endpoints,

asthma-related emergency room visits,

hospitalizations and ICU admissions, comparing

salmeterol with theophylline recipients.

The last two were separate case-control

studies that evaluated the relative risk of

asthma-related ICU admission or asthma-related

death associated with salmeterol use relative to no

use.

This last and most recent study, shown

here on the far right, included 532 pairs of asthma

deaths and matched controls and is the largest

case-controlled study evaluating asthma-related

110

death ever conducted. Notably, none of these

studies showed a significant increase in the

relative risk of these serious outcomes for

salmeterol.

GSK is committed to a comprehensive

research plan to further evaluate the safety and

efficacy of salmeterol. We believe that these

currently ongoing studies will provide valuable

information regarding the safety and efficacy of

salmeterol in patients with either asthma or COPD.

In order to address some of the issues raised by

SMART, two studies are under way.

The first is a year long clinical study

evaluating the incidence of asthma exacerbations in

460 African American subjects. Results are

expected in 2007. The second is an epidemiology

study utilizing data from 7 Medicaid plans to

examine racial variation and association of

asthma-related prescription medication use with

asthma morbidity and mortality. The results are

expected in 2006.

Studies have suggested that response to

111

short-acting beta agonists may be affected by

genetic polymorphisms in the beta

2 receptor.

However, there is one study that has suggested

there is no similar association with salmeterol and

clinical outcomes including exacerbations. This

was the Taylor study that Dr. Sorkness showed

earlier. Since there were no genetic samples

collected in SMART we are conducting two studies to

address whether clinical outcomes in patients

receiving salmeterol are affected by genotype.

The first study is a 38-week clinical

trial evaluating response by beta

2 receptor

genotype in 540 subjects with asthma. The results

are expected in 2007. The second study will

evaluate polymorphisms in beta

2 adrenergic

glucocorticoid pathways with respect to clinical

response in approximately 1,000 subjects from

completed GSK clinical trials.

Finally, while asthma has been the focus

of today's discussion, there are ongoing studies in

patients with COPD that will help address whether

the results of SMART are relevant for patients with

112

COPD. The first is a 3-year study of all-cause

mortality in approximately 6,200 subjects with

COPD. Results from this study are expected in

2006. In addition, we are conducting 2 year-long

replicate studies examining the rate of moderate to

severe COPD exacerbations, with results expected in

2007.

Asthma is a serious chronic disease with

significant morbidity and mortality. Salmeterol is

one of the most thoroughly studied medications for

asthma and has been shown to provide substantial

therapeutic benefit, including improvements in lung

function, and asthma-related quality of life, and

reduction in symptoms, rescue medication and asthma

exacerbations.

The extensive clinical trials have led

evidence-based asthma treatment guidelines to

recommend long-acting beta agonists with ICS as the

preferred option for patients with moderate,

persistent asthma. There are conflicting data for

salmeterol. SMART and SNS suggest that salmeterol

may be associated with an increased risk of rare

113

serious asthma-related events including

asthma-related death. But when large cohorts of

patients are evaluated in epidemiology studies this

association is not observed. The low number of

serious asthma events in SNS and SMART does not

allow for definitive conclusions, and the fact that

the events of concern are also those that are

experienced by patients with asthma, regardless of

treatment, makes assessment of cause and effect

relationships difficult.

Ultimately, what are the implications of

the data for patients with asthma? Well, specific

treatment decisions for an individual patient can

only be made by their physician. It is our

responsibility to provide the complete information

so that the physician can make well-informed

treatment decisions. We have done this in the

prescribing information for products containing

salmeterol.

From the time that salmeterol was

introduced in the United States in 1994 the

prescribing information has provided specific and

114

appropriate guidance on its use. This includes

that salmeterol should not be used to treat acute

symptoms. It is not a substitute for inhaled or

oral corticosteroids, and consideration should be

given to adding anti-inflammatory agents, for

example corticosteroids.

In 1995 further information was added,

including a warning that salmeterol should not be

initiated in patients with significantly worsening

or acutely deteriorating asthma. As I have already

mentioned, detailed information on the rare

asthma-related events seen in SNS and SMART had

been incorporated in the prescribing information so

that informed treatment decisions can be made.

This includes the boxed warning, as well as other

language to address the inconclusive nature of the

results and the potential for a class effect of

inhaled long-acting beta agonists.

In conclusion and based on the weight of

evidence, we firmly believe that salmeterol remains

a valuable medication that has improved the level

of care for patients with asthma and COPD.

115

Additionally, GSK is committed to further research

that will not only help better characterize the

efficacy and safety of salmeterol but will also

help better understand asthma in general. There is

a large volume of data from clinical trials of

salmeterol, as well as extensive clinical

experience with this medication. Taken together,

these continue to support a favorable benefit to

risk profile and, therefore, salmeterol should

remain available to physicians and patients.

I thank you for your time and I will now

turn the podium back over to Dr. Jones.

Closing Remarks

DR. JONES: I would like to introduce

three additional experts here with us today. Prof.

Richard Beasley is the director of the Medical

Research Institute of New Zealand. He is also an

international expert on beta agonist safety and

asthma epidemiology. Dr. Eugene Bleecker is

professor and section head, Pulmonary, Critical

Care, Allergy and Immunologic Diseases at Wake

forest University Health Sciences. Dr. Bleecker is

116

also the co-director of the Center for Human

Genomics, and was a member of the MMRC for SMART.

Finally, Dr. George O'Connor is professor of

medicine in the Division of Pulmonary and Critical

Care Medicine at Boston University School of

Medicine, and is director of the Adult Asthma

program at Boston Medical Center. In addition, he

was the chairman of the DSMB for SMART.

We would be happy to address any points of

clarification and questions. Thank you.

Questions by the Committee

DR. SWENSON: Dr. Schatz?

DR. SCHATZ: I think one possible

hypothesis based on the SNS study, where there was

increased withdrawal due to asthma in the placebo

patients, is the possibility that there ended up

being an imbalance in severity by the end of the

study due to disproportionate withdrawal of more

severe patients from the placebo group. That is

obviously not so easy to tease out but I would

question whether, in fact, one looked at baseline

severity in those who withdrew versus those who

117

didn't in both groups but particularly in the

placebo group.

DR. KNOBIL: For SNS we don't have that

cut of the data to provide for you, but it would

probably make a lot of sense that the patients who

withdrew were more severe.

DR. SCHATZ: But for SMART data--

DR. KNOBIL: Oh, for SMART we saw the same

thing in that there was greater withdrawal in the

placebo group than in the salmeterol group but,

again, we don't have that cut of the data for you

today.

DR. SWENSON: Dr. Gardner?

DR. GARDNER: I have two questions, one

related to measures of adherence within or between

the groups and whether anything has been done with

that. The second is would you give us the status

of the ongoing studies at this time? You have

listed four with expected results. Can you tell us

the enrollment at this time; how far along you are?

I understand the word "commitment" but can we see

something about progress at this time and status?

118

DR. KNOBIL: All of the studies that I

mentioned to you are right now currently enrolling.

As you might imagine, enrolling limited populations

of only African Americans takes a little bit longer

than general populations. As well, in the genetic

studies we are making sure that we have balanced

groups with the different genotypes. So, that does

take some time. So, what I can tell you is that

they are enrolling but I can't tell you when they

will be done enrolling.

DR. GARDNER: And adherence?

DR. KNOBIL: Oh, I am sorry. During the

monthly telephone contacts we did ask the

question--if you could show the slide, please? On

a scale of 0-10, with zero meaning you missed all

of your doses and 10 means you took all of your

doses, what number represents how well you followed

the study physician's instructions? In the study

the mean response, patient reported response was 8

for each group. Again, this is patient reported

and patients did not return to the clinic so we

cannot verify this. We did not ask them to bring

119

their cans in. We did not weigh canisters and we

didn't have a chronolog. But I think from studies

of chronolog, we know that patients sometimes

report taking more medication than they actually

do. So, I can't really verify the actual

compliance of the patients.

DR. GARDNER: As far as you can tell, was

there similar adherence between the African

American subjects and the Caucasian subjects?

DR. KNOBIL: Yes, as far as can tell there

was no difference between any specific group.

DR. GARDNER: Finally, on the Medicaid

study, that doesn't require enrollment. Can you

tell me where that is being done and how far along

that one is?

DR. KNOBIL: Yes, that one is being done

in seven states. What we are doing right now is

the first phase of the study, which is to see if we

can identify enough patients with high enough

proportion of African Americans to make an analysis

feasible. So, that is where that study is right

now.

120

DR. SWENSON: Dr. Moss?

DR. MOSS: I have two separate questions.

The first one has to do with the dissemination of

the information in your letters and the labeling

revisions. I think the goal of this meeting is to

disseminate information properly to the physicians

and the community. The question I have is when you

sent out those letters and put on the box labels,

do you know if that changed the prescription

practices for the physicians that received those

letters or for the general community in terms of

the prescription of your medication?

DR. KNOBIL: I don't know if that

information changed the way physicians prescribe

the medications. We don't have any specific data

to that effect.

DR. MOSS: Did overall use of your

compound decline after those letters were sent out?

DR. KNOBIL: The rate of use did not

change after those letters were sent out. But, you

know, we can't guess what would have happened; all

we saw was no change.

121

DR. MOSS: You saw no change. I think

that raises a concern about, you know, we are

trying to disseminate information and are we doing

that in the proper way?

The next question I have may be answered

by you but might also be answered by either Dr.

Bleecker or O'Connor. I think it is really hard to

figure out why someone died. Taking care of people

I see this. It is really hard to define

specifically what a cause of death is. I was just

wondering if you can give us some insight into how

you define respiratory versus asthma deaths in the

SMART study.

DR. KNOBIL: Dr. Bleecker, would you like

to address that?

DR. WEAN: While Dr. Bleecker is coming

up, I want to get back to the earlier question you

raised. I am David WEAN, Senior Vice President of

Regulatory Affairs at GlaxoSmithKline. I don't

think it appropriate to say that because we can't

posit a change in prescribing habits because of the

letters and the label that they were not effective.

122

The important thing is that the information was

disseminated in a very large way to prescribers

and, thereby, to patients. To expect that you

would see a drop-off in use of these drugs that

have therapeutic benefit because of that

communication I think would be an inappropriate

assessment about the effectiveness of that

communication.

DR. MOSS: But I think one thing we have

learned is that publishing articles and papers does

not get information out to the people that need the

information to be received. In the same way, I am

not sure that sending letters out to physicians who

get a lot of mail is an effective way of

communicating information.

DR. SWENSON: Dr. Bleecker?

DR. BLEECKER: I was asked to talk a

little bit, and George O'Connor could complement on

how the mortality review committee in SMART

adjudicated cases. I agree, it is often very

difficult, and as we did this we probably all

learned. I joined the mortality review committee

123

after about a third of the cases had been done.

The members of the committee before that had been

Roland Ingrham who was professor of medicine at

Emory. He had retired. The chairman of the

committee was Hal Nelson who is a professor of

medicine at Colorado, and the third member was

Scott Weiss who is professor at Brigham and

Williams and also head of their pulmonary and

respiratory epidemiology program.

We had data on most patients available

both from death certificates and from the medical

monitors who worked with Covance. We used that to

answer questions which were related to the cause of

death; was this respiratory related; and you heard

before the likelihood or unlikelihood of that

during Kate Knobil's presentation; and then was it

asthma related. All three of us adjudicated this

independently. If we agreed on all of these

characteristics the case was not discussed further.

All of the cases in which there was any

disagreement, ranging even between "unrelated" and

"unlikely" were discussed in detail in timely

124

conference calls. I think at times we did the best

we could on relating to that.

The least amount of information was

available on deaths toward the end of the study

that were picked up from the national death

registry. On those deaths we had to rely on death

certificates or more limited information.

DR. MOSS: Can I just follow-up on that?

Can you explain a little bit how you differentiate

asthma from respiratory deaths? A respiratory

death that is not asthma, is that pneumonia? What

were criteria to differentiate those specific

things?

DR. BLEECKER: Well, often asthma entered

into those deaths. Let me give you an example of

respiratory death. Because someone during an auto

accident had trauma to the chest--and this did

occur in some of the younger individuals--and died,

and that clearly was related to an automobile

accident and because of the nature of the event and

other things was not related to asthma. It was

more difficult if someone entered the hospital with

125

pneumonia, was intubated and in an intensive care

unit. I think under those circumstances, some of

those deaths were adjudicated depending on the

course on the ventilator or those serious events as

possibly related to asthma. So, at times it is

very hard to distinguish that from the records.

Again, I think the fact that they were performed

independently and you needed to look for

concurrence and discussion, I think a reasonable

approach was performed.

DR. SWENSON: Dr. Gay?

DR. GAY: Based on the appropriate

emphasis that you have begun to make on genetic

testing, as we have seen based on the information

that Dr. Sorkness elegantly presented before, do

you have any preliminary estimates of what you feel

would be the prevalence of Arg/Arg gene

presentation in patients with greater severity of

asthma or based on ethnic differences?

DR. KNOBIL: Yes, I would not be able to

predict the different prevalences of those genetic

subtypes and I would ask Dr. Bleecker to comment on

126

that. The one thing I would add is that in the one

genetic study we are making sure that there are

equal numbers of each genetic subtype so that we

don't have a predominance of one and very few of

another. Dr. Bleecker?

DR. BLEECKER: I would like to add a few

comments to that because I think there are some

important aspects of this. First of all, we have

centered on basically one variation within the

beta2 adrenergic gene. Looking at that gene more

carefully--and there have been published studies on

this, especially from the Liggett group as well as

some work from our laboratory which has been

presented at last year's ATS and Academy of Allergy

meetings--there is a good deal more variation in

that gene, and there are relationships between the

arginine genotype and some of that other variation.

Some of that may be very important in trying to

sort out the hypothesis of whether variation in

this gene affects therapeutic response and

potentially affects outcome.

The second important issue is there is

127

more variation, and African Americans have a higher

prevalence of the Arg/Arg genotype, about 22

percent, and that is what was seen during the

screening for the ACRN BARGE trial versus about

12-14 percent in Caucasians. The implications of

that on outcome are difficult, and I think it is

very important that the studies that were outlined

by Dr. Knobil on studying specifically an African

American cohort in which they are going to look at

outcome such as exacerbations and genotype are

critical because those kinds of studies are not

being done because of the limitations in

recruitment by the NIH NHLBI asthma clinical

network.

DR. SWENSON: Dr Prussin?

DR. PRUSSIN: I have a similar question

that I raised before with Dr. Sorkness. You have

some very nice studies that are undergoing, but do

you think they are going to address the question at

hand in terms of asthma deaths or severe pulmonary

endpoints? They are fairly small studies relative

to the SMART study and, when all is said and done

128

in three or four years, it is not clear to me at

least that you are going to have any kind of handle

on asthma death. Could you comment on that?

DR. KNOBIL: Yes, as I mentioned, it is

very difficult to prospectively study

asthma-related death because the rate is relatively

low and you need a very large N to get conclusive

results. The one study that will help us get there

though, I believe, is the Medicaid study in that we

can get information from the 7 Medicaid plans and

look at the different racial subgroups, look and

see what medications they were on and see what

contributed to those patient's deaths, whether it

is a long-acting bronchodilator, short-acting

bronchodilator or some other related medication.

So, I do believe that in that observational design

we will be able to get some more information about

asthma-related death. The other studies are mainly

going to address asthma exacerbations and other

responses such as lung function, rescue albuterol

use, and we will be able to look at those in

relationship to genetic makeup as well.

129

DR. PRUSSIN: The difficulty I have with

that though is that there is a lot of data showing

that salmeterol improved asthma exacerbations. You

have shown that. Clearly, the more severe

endpoints of death are tracking differently. So,

you are thinking of it as the tip of the iceberg,

that however exacerbations are going to track

asthma deaths are going to track and clearly they

are behaving differently. So, just because you

have certain data on asthma exacerbations in

certain subgroups, that may not be proportional or

relate to asthma. We don't have any prospective

studies ongoing or planned to really address the

endpoint that I think this committee has been asked

to address, which is asthma death. So, it could

very well be a different phenomenon than what is

causing asthma exacerbations.

DR. KNOBIL: That is true, and there are

other factors beyond asthma treatment that may be

contributing to asthma-related death as well,

including access to care or how a patient's asthma

is managed. We don't know the answers to that

130

either. So, that is why in order to actually even

look at asthma-related death an observational

design is probably going to provide more

information more quickly.

I take your point that it seems that a

prospective study would be the gold standard. The

issue there is that if the rate of asthma-related

death was similar to what we saw in SMART, in order

to see an effect you might have to have a study as

large as 800,000 patients. As you can see, that

would be very difficult to do. So, yes, it would

be nice to be able to do it prospectively but it is

going to be more difficult than doing it in an

observational design.

DR. JONES: Actually, I think Dr. Beasley

wanted to make a comment.

DR. BEASLEY: Yes, in response to that I

would like to caution in terms of considering a

prospective clinical trial as being the gold

standard when you are looking at a rare outcome

such as mortality. I think we saw that in the

SMART study, where it was required to study

131

approaching 30,000 subjects to obtain around 35

respiratory-related deaths, and there was a real

compromise in terms of design of the study to

actually achieve that and individuals were given 7

canisters at the beginning of the study without any

formal clinical follow-up, which is something which

would not be done on label in terms of salmeterol

therapy. In terms of the other issues of the label,

many of the subjects had asthma severity where

salmeterol would be inappropriate as sole therapy.

So, I think that when you try a

prospective controlled trial to look at a rare

outcome such as death you often have to incorporate

a methodology that clearly is outside the spectrum

of what is recommended clinical practice. That

clearly happened with the SMART study so that we

have to consider the SMART study results not

applicable to what would be considered good

clinical practice, and the alternative is actually

to increase the number of deaths through

case-control methodology and that was the method we

used in New Zealand to identify the risks

132

associated with fenoterol.

In that regard, I think that the Anderson

study in the BMJ is considerably more powerful in

terms of looking at the role of bronchodilator

therapy and the rare outcome of asthma mortality.

They were able to look at over 500 deaths, match

them with subjects who came from the same

proportion of the population of patients in terms

of severity, who were subjects with a previous

hospital admission, and then they were able to

stratify their analysis by looking at an even more

severe subgroup. When they did that there was

actually no increased risk associated with the use

of salmeterol therapy.

So, I think that in terms of the

epidemiological approach to a rare outcome of

asthma mortality looking at the role of medication

use, the epidemiological view is very clearly that

the case-control methodology is more powerful and

more accurate than a prospective clinical trial. I

think in some respects almost the learning point

from the experience with the SMART study was the

133

compromise that had to be obtained in terms of

clinical practice to achieve a study which, even

with 30,000 people, did not have sufficient power.

DR. SWENSON: Dr. Schoenfeld?

DR. SCHOENFELD: Do you have a tabulation

for the whole group and for each of these subgroups

and for each of the endpoints of the absolute risk

or the attributable risk? Because from a

risk/benefit point of view the relative risk isn't

very informative because, of course, you can have a

3-fold relative risk of a very, very rare event and

that may be important if you are judging something

like an environmental pollutant that you can remove

but is not really important when you are judging a

very effective drug which improves people's quality

of life. So, I wonder if you have that tabulated.

I have done some back-of-the-envelope calculations

but it would be helpful to have the actual numbers

where we looked at the percent of deaths per

patient-year or number of deaths per 1,000

patient-years, or something of that sort that would

be attributable, assuming that there is some

134

attributable risk to the use of the drug, or even

just the total risk among asthma patients per 1,000

patient-years or 10,000 patient-years, or whatever.

DR. KNOBIL: Unfortunately, we don't have

those data. We have not done those calculations.

DR. SWENSON: Dr. Martinez?

DR. MARTINEZ: Would you please clarify

what the entry criteria were for both studies, the

British one and the one done in the United States?

How was asthma defined? Were any of the usual

parameters--response to beta

2 agonists or

methacholine used to define asthma in these two

studies?

DR. KNOBIL: These studies were a little

different than normal clinical trials, say, to

opinion of the investigator if the patient have

asthma. They did not have to show reversibility.

There was no smoking criterion. In this case they

had to be 12 years of age or older. In SNS they

had to have moderate to severe asthma and in SMART

they were not allowed to have concomitant

135

administration of beta blockers. That is about

all.

DR. MARTINEZ: Was response to

bronchodilators measured?

DR. KNOBIL: It was not measured.

DR. SWENSON: Miss Sander?

MS. SANDER: Yes, when we are looking at

possible causes of death, are you able to know if

patients who died were using salmeterol as if it

was an acute bronchodilator?

DR. KNOBIL: In SMART the only question we

asked was if the patient was using the medication

as the physician told them to. That is the data

that I showed you earlier. We do not have any data

on whether they were using it as a rescue

medication.

DR. SWENSON: I realize that there are

more questions to be posed to GSK but we need to

take a break at this point. We have a general

question session in the afternoon and the rest of

these questions should be addressed at that time.

We will be back again in 15 minutes.

136

[Brief recess]

DR. SWENSON: We will resume the meeting

with Dr. Eric Floyd, from Novartis, to discuss

formoterol and its place in this controversy.

Novartis Presentation

Introduction

DR. FLOYD: Good morning. Dr. Swenson,

members of the FDA advisory committee, Dr.

Chowdhury, members of the Food and Drug

Administration and guests, my name is Eric Floyd.

I am Vice President and Global Head of Drug

Regulatory Affairs for the therapeutic areas of

dermatology, respiratory and infectious diseases.

On behalf of Novartis Pharmaceuticals Corporation,

I thank you for the opportunity to review the

current safety experience today for a long-acting

beta agonist, Foradil.

There have been numerous safety concerns

expressed publicly regarding use of long-acting

beta agonists. The purpose of our presentation

today is to discuss implications of recently

available data related to the safety profile of

137

long-acting beta agonists. This morning Novartis

would like to present to you the results and

conclusions of our review of available safety data

for Foradil. Based upon our review of clinical

trial data and postmarketing experience, we would

like to demonstrate that formoterol exhibits a

favorable risk/benefit profile.

To provide a brief regulatory overview,

Foradil was first approved for marketing in France

in 1990, and subsequently approved in other

European countries. Foradil received FDA approval

to market in the United States in 2001. Foradil is

currently approved in over 80 countries worldwide

and to date, we currently have over 13 million

person-years of exposure.

Foradil Aerolizer was marketed as a dry

powder capsule for oral inhalation and was approved

in 2001 for a dosage regime of 12 mcg twice daily

for maintenance therapy for individuals with asthma

5 years of age and above for the following

indications, acute prevention of exercise-induced

bronchospasm in individuals 5 years of age and

138

older when administered on an occasional as needed

basis, and for chronic obstructive pulmonary

disease, including chronic bronchitis and

emphysema.

Outside of the United States, Foradil is

approved at 12-24 mcg twice daily and is a highly

prescribed bronchodilator, and is also indicated

for the maintenance therapy of asthma specifically

in adults and children 5 years of age and older;

for the prophylaxis and treatment of

bronchoconstriction in patients with asthma;

prophylaxis of bronchospasm induced by inhaled

allergens, cold air or exercise; prophylaxis and

treatment of bronchoconstriction in patients with

reversible or irreversible COPD, including chronic

bronchitis and emphysema. In some countries it is

also approved as metered dose inhaler and

multi-dose dry powder inhaler, 10 mcg, Certihaler.

In order to provide a more detailed review

of our current data to date, I would like to

introduce our speakers today. Dr. Gregory Geba,

who is the Vice President and U.S. Head,

139

Respiratory, Dermatology and Infectious Diseases,

Clinical Development and Medical Affairs for

Novartis Pharmaceuticals, will present the safety

profile of Foradil.

He will be followed by Dr. James Donohue,

who is Professor of Medicine, Chief of Pulmonary

Division, University of North Carolina, Chapel

Hill, who will present the clinical implications.

In addition to the key speakers, we also

have additional advisors available to address any

specific questions you may have. Specifically from

a statistical perspective we have Dr. Gary Koch,

Professor of Biostatistics, School of Public

Health, University of North Carolina, Chapel Hill.

I would now like to turn the podium over to Dr.

Geba.

Efficacy and Safety of Foradil

DR. GEBA: Thank you, Dr. Floyd. Dr.

Swenson, committee members, members of the FDA and

interested attendees from the community, it is my

role to review with you all of the clinical data

and postmarketing data we have available for

140

Foradil, the other long-acting beta agonist being

discussed today. We firmly believe that Foradil is

a drug that provides clinical benefit with a

favorable benefit/risk profile.

As indicated in the previous presentation

by Dr. Floyd, and later on in this presentation by

Dr. Donohue, the clinical features of Foradil have

led to its inclusion in both U.S. and international

guidelines in the treatment of moderate to severe

persistent asthma.

These are the key points of the

presentation. We will describe pharmacologic

differences that exist between formoterol and

salmeterol, which may or may not have clinical

impact; a Phase 4 trial, 2307, which examined

asthma-related serious adverse events in

adolescents and adults; our integrated Foradil

clinical database; and postmarketing adverse event

data. The totality of the evidence does not

elevate concern for a safety signal for Foradil

which continues to support its favorable

benefit/risk profile.

141

These are the chemical structures of

formoterol on top and salmeterol on the bottom.

Please note that at the end of the molecule that

interacts with the beta receptor, the catechol end

of the molecule, the molecules are actually similar

in having a hydroxyl group at position 5. However,

they are different at positions 6 where you see

different side chains. Most importantly, at

position formoterol has a much longer allophanic

chain, as previously shown. Thus, although both

molecules bind to the beta

2 receptor, differences

in structure allow salmeterol to bind to an

additional site within the beta

2 receptor termed

an exosite. Prolonged salmeterol activity depends

on binding with the exosite when formoterol's

activity is independent of an exosite. In

addition, mutation of the exosite region could

affect the duration of action of salmeterol. One

of the consequences of structural differences is

that formoterol is a full agonist at the beta

adrenergic receptor, whereas salmeterol is a

partial agonist.

142

Typical experiments illustrating this

point are performed with human bronchial explants

which show that the bronchodilatory effects of

isoprenaline are decreased by prior incubation of

tissues with salmeterol but not formoterol. The

potential clinical implication of this difference

is that in the setting of beta

2 receptor

down-regulation the effect of rescue

bronchodilators may be greater for full agonists

than for partial agonists.

I would like to now move on to a

discussion of the Phase 4 trial 2307. This trial

was recently completed and is detailed in your

briefing book. Why was this study done? Protocols

040, 041 and 049 were 3 pivotal studies conducted

to support registration of Foradil in the United

Sates and 040 and 041 were 12 weeks in duration and

were conducted in adolescents and adults; 049 was

conducted for one year in children ages 5-12.

Trial participants were randomized to

receive Foradil 12 mcg BID, 24 mcg BID or placebo.

In 040 and 041 there was also a comparison to

143

regular doses of albuterol 180 mcg QID. Please

note that all groups were allowed to take

additional doses of albuterol as needed for

residual symptoms and all groups were allowed

anti-inflammatory agents.

Shown here are the proportion of patients

with asthma-related serious adverse events. These

studies showed more serious asthma-related serious

adverse events in the higher don formoterol arms

compared to the lower doses or the approved

formoterol arm as well as the placebo arm.

In light of these findings, the agency did

not approve the higher dose of Foradil. After

discussing this observation with the agency and

with their guideline, we pursued a safety study

whose primary endpoint was asthma-related SAEs.

The inclusion and exclusion criteria for protocol

2307 were identical to protocols 040 and 041 which

were our pivotal trials. Indeed, the resulting

population studied in protocol 2307 was similar to

that of the pivotal trials in adolescents and

adults, which was the population studied and

144

requested, as shown in this slide.

Apart from the trial duration which is

different, age differences were pretty similar.

FEV 1 at baseline was fairly similar.

Please note

that the proportion of black patients in this trial

mimicked the proportion of patients in the U.S.

population. There are some subtle differences in

terms of ICS use and reversibility. Actual

reversibility for 2307 was slightly less than 040

and 041 but otherwise the study populations were

very, very similar.

The design of this safety study is shown

here. Using identical entry criteria--again, these

are identical entry criteria to those employed in

our pivotal studies--after a 2-week run period,

shown here, patients were randomized to receive, in

a double-blind fashion, one of the following

treatments, either formoterol 12 mcg BID formoterol

24 mcg BID--again, 12 mcg BID was the approved

dose; 24 mcg was the higher dose. They received

either of those two doses or placebo in another

group or, in an open-label group, received

145

formoterol 12 mcg BID plus an additional up to 2

rescue doses of formoterol 12 mcg BID, which

constituted the intermediate dose arm.

Please note that to increase the rigor of

this study, after 16 weeks of treatment we planned

to contact all patients, including those who

discontinued, to record all adverse events. This

assured that all patients would be evaluated for

AEs irrespective of treatment efficacy and trial

persistence.

Results of the study are shown here.

Please note that there was a correction made to the

briefing book provided by the agency. The lower

dose arm had a somewhat higher number of patients

who reported serious asthma-related AEs. However,

after review of the specifics of these cases, the

agency excluded 2 of these events in the Foradil 12

mcg arm, reducing that number from 5 to 3. Thus,

the final event rates were 0.2 percent in the

placebo group; 0.6 percent in the low dose group;

0.2 percent in the intermediate dose group; and 0.4

percent in the high dose group. Overall, there

146

were far fewer events than expected based on the

pivotal trials.

Displayed on this slide are the point

estimates and 95 percent confidence

intervals--which I hope you can see as red on a

blue background--for the proportion of patients

experiencing asthma-related SAEs in each treatment

group. As previously mentioned, the rates are low

for all treatment groups. The 95 percent

confidence interval for all Foradil doses combined

overlaps the 95 percent confidence interval for

placebo and excludes 1 percent. These data reflect

the revised rates after an FDA adjudication.

In conclusion, the observed rate of

adverse events was far lower than we expected from

our pivotal trials despite demographics that were

similar to protocols 040 and 041. Absolute

differences between groups were very small. Higher

SAE rates in the higher dose of Foradil arm,

previously observed in adolescents and adults in

protocols 040 and 041, were not observed in this

larger, specifically designed safety study.

147

Now I would like to review with you a

comprehensive safety analysis of our clinical trial

database. We performed an extensive review of

safety based on our clinical trial database which

focused on deaths and asthma-related adverse

events. In terms of deaths, we examined all

controlled and uncontrolled trials in the Aerolizer

and Certihaler databases. All studies irrespective

of trial duration were examined to assure that the

very rare case of sudden paradoxical asthma,

culminating in the demise of a patient, was

captured.

For the analysis of asthma-related adverse

events we focused on controlled trials of greater

than or equal to 4 weeks duration so as not to

dilute the denominator for adverse events in trials

of longer duration with very short trials,

sometimes as short as 24 hours, which were

performed to assess short-term changes in lung

function.

For controlled trials the database

included nearly 6,000 patients on Foradil, shown

148

here; for uncontrolled trials over 2,700. For

trials of 4 weeks or greater in duration the number

was over 5,000 on Foradil, whereas the

placebo-controlled trial database was comprised of

over 3,700 patients who had been randomized to

Foradil.

Looking at our controlled trials, there

were 3 deaths overall, one death in the Foradil

group, representing over 1,600 patient-years of

experience; one death in the albuterol group,

representing 241 patient-years of experience; and

one death in the placebo group, representing nearly

600 patient-years of experience. The rates of

death, therefore, was 0.41, 0.17 and 0.06 in the

albuterol, placebo and Foradil arms respectively.

The Foradil death was asthma related, shown here,

representing a rate of 0.06 asthma deaths per 100

patient-years of exposure. So, here we are

expressing it as a rate per years of exposure to

the drug, which represents less than one asthma

death per 1,000 years of treatment.

In reviewing the uncontrolled clinical

149

database, it is important to note that these

studies were those that did not incorporate

comparator arms. They were all open-label and

included trials conducted as part of compassionate

use programs. In addition, patients tended to be

older, with a high proportion of elderly subjects;

had more severe asthma; used more beta agonists at

baseline; and exhibited noon-asthma-related

mortality at a higher rate, indicating a higher

degree of general medical morbidity compared to the

control database. There were 5 deaths overall, 3

of which came from one study in France which

allowed entry of severely ill patients.

Now I would like to move on to address

significant asthma exacerbations. This term

includes asthma-related adverse events which were

meaningful enough to prompt patient discontinuation

whether severe or not and, to get to Dr. Schatz'

point, included asthma-related adverse events

reported as serious whether or not they caused a

discontinuation. So, we are looking at patients

that dropped out of the trial due to an

150

asthma-related event that was meaningful enough to

stop therapy.

Displayed are the discontinuation rates

due to an asthma-related adverse event in multiple

dose, placebo-controlled trials of greater than or

equal to 4 weeks in discontinuation. These are the

discontinuation rates. Please note that there were

fewer asthma-related discontinuations in the

Foradil arms compared to the placebo arm or to

albuterol. So, the rate overall for an formoterol

doses was 7.1; for placebo it was 10.7; for

albuterol 8.1. Please note that this was

especially notable for the approved dose of

Foradil, 5.6 versus 10.7.

A reverse pattern was observed for

asthma-related serious adverse events. Foradil

patients experienced more events than placebo.

Here the imbalance was greater for the higher

Foradil dose. The numbers are shown here, 3.5 for

the approved dose versus 3.1 for albuterol, 0.9 for

placebo and a higher rate for the albuterol 48 mcg

dose. Again, this dose is the approved dose in the

151

U.S.

When both types of events are taken into

consideration, the rate of significant asthma

exacerbations, that is, asthma-related AEs

meaningful enough to cause the patient to

discontinue and asthma-related events that were

labeled as serious whether or not they caused

discontinuation, was actually lower for Foradil at

its approved dose than the rate for placebo or for

albuterol. Note that at the highest dose of

Foradil that rate was similar to the placebo rate.

But for Foradil at its approved dose the rate was

7.1 versus a placebo rate of 10.9.

In summary, based on this analysis of our

clinical trial database for asthma-related adverse

events, we observed a rate of significant asthma

exacerbations for the approved Foradil dose that

was lower than placebo rates.

I would like to now move on to a review of

postmarketing data. In order to explore the

adverse event profile of Foradil on the market and

to provide some estimates as to how it might

152

compare to other drugs in its class, we performed

an analysis of postmarketing data based on FDA AERS

database, that is the FDA's adverse events

reporting system.

We must recognize that this type of

postmarketing analysis has its limitations.

Although spontaneous reporting of ADRs remains the

most common method used for monitoring the safety

of marketed drugs and is useful for detecting

safety signals, it is limited by the fact that a

substantial percentage of ADRs are not reported.

The reporting rate also tends to be lower the

longer a drug is on the market. This is a

well-known phenomenon and is known as the Weber

effect. In addition, targeting drugs to lower or

higher risk patients may alter apparent ADR

occurrence, and notoriety associated with a drug or

class may alter reporting rates. A final concern

is that the ADR that is being reported in this

instance is the disease itself, It is a

manifestation of the disease itself.

We examined FDA adverse reports for death

153

or outcome of death and found that rates were

highest in the first years after launch and

declined each year thereafter. This analysis is

shown in your briefing book.

We will now review the reporting rates for

these and other events of interest. Please note

that reporting rates are reports with case

definition on the drug of interest divided by the

exposure worldwide since the drug was marketed in

the U.S. per 100,000 patient-years.

Relative rates of reporting can also be

assessed by simply calculating the percentage of

reports at case definition by the total number of

adverse events reported. This does not take into

consideration the exposure to the drug of interest

and may inflate the Weber effect if there is a

difference of time on the market. Shown here are

the reporting proportions, on the left, and the

reporting rates per 100,000 patient-years, on the

right, for Foradil and salmeterol.

As you can see, the reporting proportion

for formoterol, a drug that was established on the

154

U.S. market in 2001, compared to salmeterol, which

was on the market since 1994, is somewhat higher.

However, one must adjust for the exposure to the

drug which was far greater for salmeterol. When

adjusting for this higher number of exposures for

salmeterol the result ratio flips. That is, when

we adjust for actual exposure to the drug we note

that the rate for Foradil was somewhat lower than

that of salmeterol.

In conclusion, well-described

pharmacologic differences exist between formoterol

and salmeterol although the clinical relevance is

not known. In pivotal trials conducted for U.S.

registration, a potential safety signal emerged in

the Foradil high dose group, leading only to the

approval of the 12 mcg dose, that is 12 mcg BID,

and the request for postmarketing asthma safety

study 2307. Study 2307 examined asthma-related

serious adverse events in adolescents and adults

and did not provide evidence of a safety signal for

Foradil at any dose.

An analysis of the pooled Foradil clinical

155

trial database and a review of postmarketing

adverse event data, with its limitations, do not

provide evidence of a safety signal for Foradil.

The totality of the evidence, therefore, does not

elevate concern for a safety signal and continues

to support the favorable benefit/risk profile of

Foradil in the treatment of asthma.

Thank you for your attention. Dr. James

Donohue will now present the clinical implications.

Clinical Implications

DR. DONOHUE: Thank you, Dr. Geba. Dr.

Swenson, members of the advisory committee, Dr.

Chowdhury and Dr. Meyer and members of the FDA,

ladies and gentlemen, I am here today as a

clinician investigator to talk about the clinical

implications of the long-acting beta agonist class.

As an older physician, I can talk a little bit

about life before the introduction of the

long-acting beta agonist class into our clinical

practices. While the alternatives were

short-acting beta agonists, theophylline, various

epinephrine agents, oral beta agonists, each of

156

these treatments had different benefit/risk ratios

or profiles from the long-acting beta agonist

class. I would like to discuss a little bit the

implication of the roller-coaster effect on our

patients with asthma's lives, the lack of nocturnal

coverage with most of these shorter-acting agents

and issues with compliance. There have always been

issues with the short-acting beta agonists for the

need to frequently dose; special issues with our

children and whether or not they could be dosed in

the schools; the difficulty in our blue-collar

workers, of course, who need frequent dosing of

their medications.

The short-acting beta agonists have, as I

say, benefits and risks. Overdosing of the

short-acting beta agonists was associated with

tremor, particularly with the peak. There were

changes in metabolism, hypokalemia and changes in

glucose metabolism which may or may not be

clinically significant but could be under certain

circumstances. There was also tachycardia

associated with people using some higher doses or

157

people who had more co-morbidity issues. Then

also, to inform us, we had very useful data from

Saskatchewan looking at thee use of short-acting

beta agonists in Canada. These are combined data

for formoterol and albuterol.

These are the deaths per 100,000 per year

and the number of canisters. We can see that as we

start getting up in the number of canisters,

especially win formoterol, one sees an increase in

deaths due tot he short-acting beta agonists or

associated--not necessarily due to but associated

with the short-acting beta agonist class. So this

was, of course, a concern to all of us and is part

of the recommendations presently in the guidelines.

We also had different side effect profiles

of other medications available to us before the

long-acting beta agonists and drugs we would have

to consider today as substitutes. First and

foremost, theophylline, particularly the

longer-acting forms. Their safety profile is

important to look at. There was a narrow

therapeutic window, as everyone knows, with these

158

drugs. There were very, very important drug

interactions. In fact, if we look at our elderly

asthmatic population, commonly these patients would

enter the hospital with use of an antibiotic or a

medication for reflux causing a drug-drug

interaction and making the patient theophylline

toxic. Furthermore, if we look at drug

interactions in the hospital, there is a huge

safety concern about medication errors, and

what-have-you, and theophylline were always at the

top of the list.

Other medications we had were oral beta

agonists, both short-acting and long-acting and,

again, much less of an efficacy profile as compared

to the long-acting inhaled agents and a much

greater safety risk with tachycardia, tremor and

reflux. Oral corticosteroids have to be used more

and more when patients have more and more

exacerbations and I don't have to review the

laundry list of side effects that are well-known to

everyone in the room.

Now, throughout the world we have--I have

159

outlined in yellow here the G8 nations because of

last week's meeting, but we can see the variation

in prevalence of asthma symptoms as we get more

industrialized societies. We see a very large

increase in the number of patients who suffer with

airways disease.

On the other hand, there are facts that I

find very, very consoling and comforting. This is

the death rate due to asthma in the United States

going back to 1960 up to 2002. These are the

deaths per 100,000 so we can sort of get the rate

that you are asking for. Then we have the African

Americans here and the Caucasian population here.

Short-acting beta agonists were introduced

in the 1970s. We had the inhaled corticosteroids

introduced in the 1980s. There have been enormous

efforts in patient education, efforts by the expert

panels of the National Institute of Health for

guidelines and just generalized education programs,

along with the introduction of effective controller

medicines along with the long-acting beta agonists

that seems to have led to a decline, although still

160

very high, relatively higher in the African

American population, but to the general United

States population, from 5,400 to a little bit above

4,000. So, we are clearly doing something right.

What the attribution would be here to the various

things introduced is, of course, beyond my ability

to say but, clearly, all these things together have

helped us to improve the lives of our patients.

Now, what about the long-acting beta

agonists? Just briefly, you have seen an awful lot

of this data this morning and, at the risk of being

a little bit redundant, the first benefit, of

course, is in patient symptoms. These are better

bronchodilators. I think we would all agree that

the data are overwhelming on this. There is a

reduction in the roller-coaster effect, and I will

come back to that in a minute; better control

because of the longer duration of effect and

nocturnal symptoms. Because of the control, we

have less use of rescue medications. We have

improved morning lung function and also less

diurnal lung function variability. It doesn't

161

completely eradicate it but it does minimize to

some extent some of the early morning dipping that

leads to waking up or even worse outcomes. Also,

equal important perhaps, it gives us protection

against exercise-induced bronchospasm and that is

the exercise of normal daily activities in normal

life, and it is nice to have that kind of

protection on board so you don't have to

continuously dose yourself.

Looking at the formoterol data, Dr. Geba

has shown us the safety data. We saw that there

are 3 pivotal trials that you have in your briefing

documents. There is superior improvement in the

FEV 1 over placebo over the course of

12 hours.

This duration of action is sustained for 12 weeks

so there doesn't appear to be a signal that there

is any tolerance or reduced efficacy. There is a

reduced need for nighttime rescue medicine, and the

onset of action is similar to albuterol, as has

been outlined.

Just again showing the similar 12-hour

studies, this is the 040 and the 041 that you have

162

seen a moment ago. This is at week 12. Here we

see the 12-hour curve and this is the mean change

in baseline FEV 1.

Here is the short-acting

albuterol and placebo, and here is the sustained

effect of the long-acting beta agonist, in this

case formoterol. First of all, let me draw your

attention to the pre-dose or trough. Often we

power clinical trials on long-acting beta agonists

and this changes well over 10 percent here, around

12 percent, and that usually can be transferred to

as meaningful clinical improvements such as

symptoms in the morning and what-have-you. Over

the course of the day you see the roller-coaster.

That in itself means nothing but what this means

here is that as these drugs flow off here our

patients become symptomatic and have to disrupt

their daily activity to take rescue albuterol.

Also one other thing I would like to show

here is the peak effect. Patients perceive change

and when you have a nice plateau effect a lot of

side effects such as tremor are much less

perceptible to a patient.

163

Other outcomes besides bronchodilator is

rescue medication. These again are from 040 and

041, the run-in for the 3 arms, formoterol,

albuterol and placebo. We see over the course of

4, 8 and 12 weeks a nice decline in the rescue

albuterol. These parameters are less rigorously

defined but we think the minimal clinical

improvement in that parameter is 0.8 puffs per day

ranging to 1. So, it appears to be in the ballpark

of something that means something to a patient and

we could quantitate that.

Simply, formoterol reduces nocturnal

symptom scores. I am showing 040 and 041 which I

believe are the adult studies run-in and over 12

weeks and the companion study here. We see a nice

decline from about 0.5 to 0.1 in the nocturnal

asthma symptom score parameter.

To end up, one of the studies that I take

consolation from as a physician is the FACET study.

Again, points are well taken here today, we are

talking about deaths. It is very difficult in

asthma studies though to power our studies, as

164

everyone in the room has heard over and over again,

on death as the outcome. Exacerbations are

extremely important and the reason that I take a

lot of comfort from this study is that it is a

one-year study. Also, the exacerbations are

precisely defined, as Dr. Sorkness mentioned this

morning.

In this study we see--again in the

mechanical function, 835 patients, 12 months--a

marked improvement after the run-in. The patients

were in the run-in symptomatic on inhaled

corticosteroids. But this study gives us some look

at what is the added value of adding a long-acting

beta agonist--added clinical value--to inhaled

corticosteroids.

Here are the two budesonide arms. I think

there is no surprise that on the mechanical

function we do see a change of 7 or 8 percent. But

I think one takes a better message away from

looking at exacerbations. First of all, if

inflammation is ongoing and not being checked the

patient is going to know that they are going to

165

breakthrough with an exacerbation. This is our

best clinical surrogate for the so-called masking

of inflammation, that is, the breakthrough of

exacerbations. In a study of one-year duration we

have adequate duration to bring this signal out.

So, in this study, as Chris Sorkness

pointed out this morning, we have an arm with 100

BID of budesonide and with formoterol. Then, the

second arm is 400 BID with the addition of

formoterol. And, the exacerbations were described

as mild with an increase in terbutaline, the rescue

medicine, and severe, defined by a 30 percent

change in peak flow and oral prednisone. The

decline in mild exacerbations was about 29 percent

and 40 percent with the combination. But with the

higher dose, high dose budesonide reduced the

exacerbation rate by 49 percent. When one adds

formoterol to it it went to 62 percent. So, there

is a net gain there and the actual clinical

implication of that is that it appears to be

significant. We really haven't put a number on

that yet but that, in my mind, is a very reassuring

166

piece of evidence that supports the use of this

class of drug.

Using the best evidence that one has, the

expert panels--many of the members of which are in

the room here--have come together, and you have

seen this before over and over again, and have

concluded that inhaled steroids and long-acting

beta agonists have a complementary effect. One can

lower the dose of inhaled corticosteroids by using

the combination, and not everyone responds to

inhaled corticosteroids although a great majority

do.

For more severe patients we would be using

higher doses of inhaled corticosteroids and

long-acting inhaled beta agonists and, hopefully,

we will be able to avoid the use of prednisone and

its very difficult side effect profile.

So to summarize, long-acting beta agonists

have really become an established part of the

current standard of asthma treatment in the United

States and also internationally. It is an integral

part of internationally established guidelines

167

using the best evidence that we have at our

disposal at the present time. It is well

established that long-acting beta agonists have a

place in the treatment regimen for asthma but must

be conjunction with inhaled corticosteroids for

those with moderate and severe persistent asthma.

Again, I don't think at the present time

there is an alternative inhaled controller

bronchodilator or one on the immediate horizon that

is suitable for asthma. So the LABAs, the

long-acting beta agonists, have provided documented

improvement in symptoms, airway function and

quality of life and you have heard a great deal

about that today. For whatever reason, since the

introduction of long-acting beta agonists,

controllers and a national effort in education and

guidelines, asthma hospitalizations and mortality

have decreased, which is very reassuring at least

to me and I think to many others. Long-acting beta

agonists in conjunction with inhaled steroids

represent a medication category critical for

optimal care of patients with moderate to severe

168

asthma. Thank you very much.

Questions by the Committee

DR. SWENSON: The time now is open for

questions to Novartis. Dr. Schatz?

DR. SCHATZ: Some of the things we are

hearing suggest a possible disconnect between

exacerbations, as has been studied and defined

usually to include oral steroids and emergency

departments or hospitalizations, and then death or

near death. So, I guess my question has to do with

2307. How were SAEs defined? However, were

asthma-related SAEs defined?

DR. FLOYD: Dr. Geba?

DR. GEBA: To answer this question I would

like to bring up a slide that gives the definition

of asthma-related and the definition of serious.

We used predefined asthma terms, MedDRA

terms, MedDRA preferred terms. Asthma-related AEs

were defined as asthma, dyspnea, bronchospasm and

chest discomfort, cough, wheezing, etc., acute

respiratory failure and hypoxia. For the

definition of SAEs it was one of the above plus one

169

of the following, death, life-threatening

hospitalization, disability, congenital

abnormalities--this is regulatory definition, and

required intervention to prevent further

impairment--standard definition.

DR. SCHATZ: Just to follow-up,

intervention could mean oral corticosteroids?

DR. GEBA: Yes, it could be. It could be

medical intervention.

DR. SWENSON: Miss Sander?

MS. SANDER: Yes, on slide CO-10 it says

"sustained improvement in FEV

1 at 12 weeks." Was

the albuterol depicted here scheduled or was it

based on symptoms?

DR. GEBA: The albuterol dose in these

trials was scheduled.

MS. SANDER: I have another question on

the next slide. It says "Foradil reduces rescue

medication use." Is that albuterol use? And that

is for exacerbation? Is that right?

DR. GEBA: The presumption is it is

exacerbation. It was albuterol usage rescue

170

defined by the patient who was symptomatic and,

therefore, referred to rescue medication with

albuterol. Correct.

MS. SANDER: So, that phrase there,

"rescue medication" is defined as rescue needed by

the patient.

DR. GEBA: Yes.

MS. SANDER: So, not scheduled.

DR. GEBA: No.

MS. SANDER: Thank you.

DR. SWENSON: Dr. Kercsmar?

DR. KERCSMAR: I have a question about the

049 pediatric study. The slide that shows the

serious AEs looked distinctly different than the AE

profiles in the adult trials. Can you give us any

further insight into that, any other data that you

might have regarding those differences in children

in the 5-12 age range compared to the adults?

DR. GEBA: I would like to point out that

the age range of patients in our so-called adult

trials was actually 12 and on so it would include

adolescents and adults. This population of

171

patients was 5-12 so it was definitely a younger

population of patients and truly children in 049.

There was a difference in rates of exacerbations in

these trials and the point I guess that I would

like to focus on, if you could bring that slide up,

the 3 trials together, the 3 pivotal trials, slide

CS-7, please. Thank you.

What we noticed in the 3 pivotal trials

was that for the Foradil high dose arm there was a

higher event rate compared to the lower don

formoterol arm. You did notice, and we did as

well, that the rate for those events was still

higher in the lower dose of the Foradil arm for

pediatric patients.

We were most concerned, as the agency was,

and this was guided by conversations that we had

with the agency to determine whether or not there

was a dose effect going from 12 to 24 mcg. Upon

discussions with the agency, we designed a trial to

prospectively test whether or not there was a dose

effect between 12 mcg and 24 mcg, and we were

requested to study the identical patient population

172

as in 040 and 041. So, we recognize that this area

has not yet been fully analyzed and evaluated, and

we are not certain as to why the issue occurred in

this age group.

DR. SWENSON: Dr. Gay?

DR. GAY: Thank you. The data of the

SMART study suggests that a predominance of the

number of adverse outcomes began to occur after

about 90 days. That is where the split in the

Kaplan-Meier curves begins to become much more

significant. A number of the studies that you

performed are at 90 days or a little bit longer. I

wonder if you have available any post-study data of

a 3-month or 6-month follow-up that may show any

change or any increase in the number of adverse

events, such as asthma-related deaths or

significant numbers of exacerbations, that go with

your most recent study or even the earlier studies,

040 or 041?

DR. GEBA: Right, in the longer and most

recent studies there were none of those events that

would contribute to this. We only had one death in

173

all of our clinical trials of over 6,000 patients

with 16,000 patient-years of experience.

We would like to point out that there is a

continuity, one would argue. We analyzed this data

in terms of the event rates for serious adverse

events to determine whether or not they occurred

randomly during those first periods of time during

the trial allowing us to, therefore, pool and

express these rates as rates per 100 patient-years

of exposure. And, that is what we have done.

On the slide shown here is depicted the

Foradil versus placebo asthma-related serious

adverse event proportion and, as you can see, there

is a distribution in that 3-month window. Also, I

would point out that even in the briefing book if

you look at the rates of events that occurred in

SMART, you can detect already a separation by an

earlier time point than the 3-month time point.

So, we were fairly confident that we could pool the

data sets that we have, limited by the fact that we

don't have trials as long as for salmeterol, and

come up with a reasonable estimate as to the event

174

rates based on this type of analysis.

DR. GAY: Just in follow-up, I want to

make sure I understand this correctly. My question

clearly is going to lead to less rigorous data but

I am concerned about follow-up time longer than

your study duration, longer than the 12 or 16

months, not within that specific time frame.

DR. GEBA: Right.

DR. GAY: Is this data concerning that

time up to 3 months post the end of the study?

DR. GEBA: No, we did not routinely follow

patients beyond the time of their study treatment,

except for a routine visit usually performed 2

weeks post study.

DR. GAY: Thank you.

DR. GEBA: Thank you.

DR. SWENSON: Dr. Martinez?

DR. MARTINEZ: Thank you. I am going to

talk in reference to the pooled Foradil clinical

trial database in which you have shown, and correct

me if I am quoting wrongly, that for less severe

events--let's call them that way for a

175

moment--there was no difference between formoterol

and placebo or albuterol. But for more severe

events there was a difference.

So, let me propose to you, and I would

like to elicit your comments, that there could be

two different ways in which the deterioration of

asthma may occur, which may occur in different

patients with different risk factors and with

different asthma phenotypes. I am proposing this

to you as an interpretation of your data. Let's

call one of them loss of asthma control, in other

words, slow deterioration; increase in symptoms

with more wheezing, more cough. I saw that all

those elements were present in your definition.

For those, let me propose to you that there is no

difference between formoterol and placebo or

albuterol.

But it could be that there is a different

set of patients in whom these symptoms don't occur

every day and they have a more brisk, brittle form

of the disease. It is in these patients that you

see that placebo is associated with 0.3 percent.

176

These patients justify the fact that in placebo you

only see 0.3 percent and you see much more win

formoterol. This goes to issues that have been

raised by other members of the committee as to how

we can explain that in terms of the everyday

symptoms, in terms of control of asthma we see

improvement with the use of all medicines in this

class, but we are seeing in many studies a signal

that there could be more severe disease. Could it

be that we are in the face of two different forms

of expression of asthma deterioration that have

different responses to this class of medicines?

DR. GEBA: Yes, we have not done a

sufficient enough analysis of these events. I

would point to Dr. Cioppa from Novartis to respond

to that question. Thank you.

DR. DELLA-CIOPPA: Thank you. My name is

Giovanni Della-Cioppa and I am Vice President for

Clinical Research. Dr. Martinez, your explanation

is certainly one possibility. There is also

another possible explanation that we would to bring

to your attention, keeping in mind that we are

177

talking here about clinical trials, and we are

talking about clinical trials of bronchodilators.

Despite the fact that these trials are blinded,

many patients have a clear perception of an

improvement of their lung function quite rapidly.

So, one could assume that these two kinds of events

represent very similar events in terms of

magnitude, in terms of gravity, in terms of impact

on the patient's well being.

But if I am a patient on placebo, or I

think I am a patient on placebo and I start going

down the drain--as shown this morning, there are a

few days of unfolding of the event--then I will

abandon the trial and experience, therefore, a

discontinuation due to asthma. If I am a patient

on active and I know I am on active I will be more

reluctant to get out of the trial, and stay on the

trial, and the same event will be labeled by the

investigators as a serious adverse event.

It is baffling, this kind of mirroring

situation by which people on placebo get out from

the trial due to asthma and the people on active

178

who stay on the trial and get serious adverse

events. So, the two explanations are not mutually

exclusive but I just wanted to offer an alternative

explanation because we are seeing it again, and

again, and again, and we are seeing it in the

pooled database, as shown by Dr. Geba.

DR. MARTINEZ: However, as much as I can

accept your arguments, I would suggest that either

of the two explanations, or both, are potentially

reasonable for the data as it has been presented.

DR. SWENSON: Dr. Meyer?

DR. MEYER: Thank you. I just wanted to

make something explicit prior to lunch as sort of a

summary from the agency standpoint from these

morning's presentations. We have heard

presentations from two sponsors that centered

around new data from studies and, of course, we

will have our own perspective on those after lunch.

But these are very different studies. These are

almost apples and oranges. They were purposefully

so because they were meant to address very

different questions.

179

The SMART study was meant to address a

signal that was coming out from postmarketing data

where we didn't feel like that could be adequately

addressed by the postmarketing data or perhaps even

by epidemiology studies but raised even, even

pre-approval by the SNS study, a signal of very

rare, very dire events that were not well predicted

by the more common adverse events, even serious

adverse events, in shorter-term trials. Hence, a

very large, very prolonged study.

The formoterol 2307 study was not designed

to answer that kind of question. It was designed

to answer the question of events seen in

shorter-term trials, representing more sort of

common serious adverse events that were detected in

the database, and really was meant to explore a

dose effect, as was previously stated.

So, I just wanted to be very explicit

about the fact that what we are talking about here

in the end are two very interesting studies but

they are addressing, and perhaps answering to the

degree they did answer them, very different

180

questions.

DR. SWENSON: And Miss Watkins has one

announcement, after which we will adjourn for

lunch.

MS. WATKINS: I would like to remind the

committee that, in the spirit of the Federal

Advisory Committee Act and the Sunshine Amendment,

discussion about today's topic should take place in

the form of this meeting only and not occur during

lunch, breaks or in private discussions. We ask

that the press honor the obligations of the

committee members as well.

Additionally, the committee members, once

you get lunch in Salon E, there is reserved seating

for the committee members in Salon D, and we ask

that you take advantage of that.

DR. SWENSON: There is one other question

here. Dr. Newman, I apologize for leaving you out.

DR. NEWMAN: Thank you. I just want one

clarification, if I could. You emphasized the

pharmacological differences between this drug and

salmeterol. Are you suggesting that because of the

181

pharmacological differences the studies, such as

SMART, are somehow not relevant to your drug?

DR. GEBA: No, we can't go that far and I

don't want to overstate those differences. I just

wanted to point out that the molecules are

different; that the receptor binding mechanism is

somewhat different; the onset of action, those

types of things are different between the two

molecules. Whether or not that has an implication

in terms of outcomes of the sort that we have been

discussing this morning is conjecture. It cannot

be ascertained. But for completeness we included a

full discussion of the molecule and differences

from the other one that is relevant in its class.

DR. NEWMAN: Just in follow-up on that, so

in terms of the pharmacologic action and factors

such as desensitization of the beta receptor, there

still is a desensitization effect, is there not?

DR. GEBA: Well, we have some data that

distinguishes the two, and to approach that I would

ask Dr. Trifilieff to respond to that question.

Dr. Trifilieff is from basic research in Novartis.

182

DR. TRIFILIEFF: So, the question was

about agonist desensitization? In theory, full

agonists will induce much more desensitization than

partial agonists. But what you have to take into

account also is the density of the receptors

because basically a partial agonist, by definition,

would need more receptor in order to achieve the

same efficacy as a full agonist. So, in a

situation where you have low density of the

receptor a full agonist will need a 4 receptor; a

partial agonist will need 40 receptor. So, you

have a greater desensitization for partial agonists

compared with full agonists.

DR. SWENSON: I wish to thank everyone for

their participation. We will reconvene at one

o'clock rather than 12:45 as is presently on the

schedule.

[Whereupon, at 11:50 a.m., the proceedings

were recessed, to reconvene at 1:00 p.m.]

183

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

DR. SWENSON: Good afternoon, everyone.

We will resume the meeting with the FDA

presentation. To begin, Dr. Sally Seymour will

first speak, to be followed then by Dr. Harry

Gunkel, after which the panel will have the

opportunity to raise questions to the two

presenters.

FDA Presentation

Salmeterol

DR. SEYMOUR: Good afternoon. My name is

Sally Seymour and I am a medical officer in the

Division of Pulmonary and Allergy Drug Products. I

am going to be speaking to you today about the

long-acting beta agonist salmeterol. Much of what

I will present today has been presented this

morning but I am going to present you the agency's

perspective.

The objectives of my presentation today

are to discuss the regulatory history of

salmeterol, which was the first long-acting beta

agonist approved in the United States. In the

184

regulatory history I will emphasize the agency's

actions in response to the safety concerns with

salmeterol. I will then review the postmarketing

clinical studies with salmeterol which were

conducted by the sponsor, including the SNS study

and SMART. Following the discussion of the

postmarketing studies, I will briefly discuss the

postmarketing spontaneous event reports for

salmeterol. Then I will highlight the sections of

the product label which include information about

the postmarketing studies.

Let's begin with the regulatory history.

Serevent Inhalation Aerosol was approved in

February, 1994 for asthma. The indication is

long-term twice daily administration in the

maintenance treatment of asthma and the prevention

of bronchospasm in patients 12 years of age and

older with reversible obstructive airways disease.

Indications for exercise-induced bronchospasm and

for COPD were added later. However, the focus of

the discussion today, as you know, is on asthma.

As mentioned earlier, the sponsor chose to

185

discontinue Serevent Inhalation Aerosol as part of

the CFC phaseout and, thus MDI is no longer

marketed. Serevent Diskus is a dry powder

formulation of salmeterol which was approved in

February, 1997 for similar indications as the

Inhalation Aerosol. The Discus is approved in

children down to 4 years of age.

Finally, Advair Diskus, which is a

combination product of the corticosteroid

fluticasone propionate and salmeterol, was approved

in August, 2000 for asthma, and later the

indication for COPD with chronic bronchitis was

added.

Let's start at the time of the salmeterol

inhalation aerosol NDA. The NDA for salmeterol

inhalation aerosol was supported by 2 Phase 3

12-week active and placebo-controlled clinical

trials in 556 patients with mild to moderate

asthma. At 12 weeks the clinical studies

demonstrated an improvement in the salmeterol group

versus placebo in the following endpoints, FEV

1 and

peak expiratory flow rate, mean percent days and

186

mean percent nights with no asthma symptoms, and

less rescue medication use.

I would like to point out that at the time

of the NDA review the results of the SNS study were

known and considered. The SNS study, as you know,

was a postmarketing study in the United Kingdom and

I will discuss that shortly. An advisory committee

meeting was held in February, 1993 to provide

advice and make recommendations regarding the

salmeterol inhalation aerosol NDA. The advisory

committee was supportive of approval of the

salmeterol NDA in adults and subsequently

salmeterol inhalation aerosol was approved in

February of 1994 for asthma.

Shortly after approval reports of

life-threatening respiratory events and fatalities

with salmeterol use were reported. Multiple

meetings were held with the sponsor to discuss the

reports. Some of the postmarketing event reports

suggest that there was possible inappropriate use

of salmeterol. This concern led to revisions in

the label in January, 1995. The warning section

187

now included the following statements: Serious

acute respiratory events, including fatalities,

have been reported with salmeterol. Salmeterol is

not for acute symptoms. Salmeterol is not a

substitute for oral or inhaled corticosteroids.

Salmeterol should not be initiated in worsening or

acutely deteriorating asthma, and patients should

have a short-acting beta agonist for acute

symptoms.

The sponsor also conducted a physician and

patient education program which included a "dear

healthcare professional" letter. In addition, as

you know, the sponsor committed to a large safety

study, the Salmeterol Multicenter Asthma Research

Trial, or SMART. SMART was initiated in July,

1996, and I think it is important to point out in

the regulatory history what the sponsor mentioned

earlier, that the SMART study had to be amended in

June of 1999 to double the population from 30,000

to 60,000 patients because of fewer than expected

outcome events.

A planned interim analysis was performed

188

in 2002 after approximately 26,000 patients had

been enrolled. The DSMB reviewed the interim

analysis data. The data indicated that the point

estimates suggested an excess risk with salmeterol

and that African Americans may be at particular

risk. DSMB recommended to continue the study if

timely recruitment was feasible. If timely

recruitment was not feasible, the DSMB recommended

to terminate the study and disseminate the findings

to the clinical research communities within 3-6

months.

Based upon the interim analysis and

difficulty with enrollment, the sponsor terminated

SMART in January, 2003. A "dear healthcare

professional" letter was also issued in January,

2003. The sponsor submitted preliminary data from

the interim analysis of SMART to the agency and,

based upon the preliminary data the product label

was revised in August, 2003. It is somewhat

unusual for the agency to make labeling changes

based upon preliminary data, however, the agency

felt the SMART information was important to include

189

in the product label. Labeling changes included a

boxed warning and information in the clinical trial

section of the label regarding the findings of

SMART. These labeling changes were applied to all

salmeterol-containing products, including Advair.

When a safety signal is noted with a drug substance

the agency's practice is to apply labeling changes

to all products containing the drug substance

unless there are data to establish the absence of

the safety concern for a particular product.

In August, 2003 the sponsor submitted the

full SMART data set. The sponsor indicated at the

time that the National Death Index or NDI search

had been performed and noted that some of the

additional deaths were still being adjudicated. In

February, 2004 the sponsor submitted the full SMART

study report which included the adjudicated NDI

data. After review of the study report the label

was once again revised to include more details

regarding the results of SMART. That brings us up

to date on the regulatory history.

Now let's discuss the postmarketing

190

studies which I touched upon in the regulatory

history, the first of which is the SNS study. The

SNS study was a randomized, double-blind,

active-controlled, parallel group 16-week trial in

the United Kingdom. The population was 25,000

patients with asthma who were randomized in a 2:1

fashion to salmeterol 50 mcg BID or salbutamol 200

mcg QID Salbutamol is a short-acting beta agonist

known as albuterol in the United States. Note that

this was not a placebo-controlled study. It was an

active-controlled study in which both arms were

treated with regularly scheduled beta agonists.

Clinic visits were conducted at 4, 8 and 16 weeks.

Outcome measures were serious adverse events and

reasons for withdrawals.

This table displays the key findings in

the SNS study, and recall that the randomization

was 2:1. I would like you to note the following,

first, there is a numerical increase in respiratory

and asthma-related deaths in the salmeterol group

with a relative risk of 3. However, this was not

statistically significant. Second, there was no

191

difference in the respiratory and asthma-related

hospitalizations or other respiratory and

asthma-related serious events between the

salmeterol and salbutamol group. Finally, there

were significantly fewer asthma- and

respiratory-related withdrawals in the salmeterol

group and this was statistically significant.

As mentioned earlier, the results of the

SNS study, which showed a numerical increase in

respiratory- and asthma-related deaths, although

not statistically significant--these were

considered at the time of approval of salmeterol

and were discussed in the February, 1993 advisory

committee meeting. The benefit of salmeterol was

felt to outweigh the risk. Thus, salmeterol

inhalation aerosol was approved in 1994.

Following approval of salmeterol, there

were reports of serious asthma events, including

fatalities. In working with the agency, the

sponsor committed to a large safety study, the

Salmeterol Multicenter Asthma Research Trial, or

SMART, which I will discuss next.

192

SMART was a multicenter, randomized,

double-blind, placebo-controlled, parallel group

study of 28-week treatment duration. The sample

size was initially planned to be 30,000 but then

was increased to 60,000 in 1999 because of a fewer

number of events than expected. Subjects were

greater than or equal to 12 years of age with a

clinical diagnosis of asthma. They were currently

taking prescription asthma medications but no

long-acting beta agonists. Subjects were

randomized to salmeterol 50 mcg BID or placebo BID

for 28 weeks treatment in addition to usual asthma

care. Subjects underwent one clinic visit in which

they were given a 28-week supply of salmeterol or

placebo, and telephone contact was made every 4

weeks.

The primary endpoint for SMART was the

combination respiratory-related deaths and

respiratory-related life-threatening experiences.

Respiratory-related life-threatening experiences

were defined as intubation and mechanical

ventilation.

193

Ideally, the endpoints for SMART would

have been asthma-related deaths and serious asthma

exacerbations but, based upon historical data, the

number of events was expected to be low. Thus, the

primary endpoint of the study was broadened to

respiratory-related deaths and respiratory-related

life-threatening experiences. It was thought that

a respiratory-related life-threatening experience

was a marker of fatal asthma and asthma-related

deaths.

As you can see in the list of key

secondary endpoints, asthma-related deaths was one

of the important secondary endpoints, in addition

to all-cause death, asthma-related deaths and

life-threatening experiences and all-cause serious

adverse events or SAEs.

SMART was designed as a non-inferiority

trial and was designed to show that there was no

difference in the outcomes between salmeterol and

placebo. SMART was powered to rule out a 40

percent increase in the combined

respiratory-related deaths and life-threatening

194

experiences and a 3 times increase in

asthma-related deaths. These numbers may seem high

but, again, they were based upon the numbers of

expected events and the ability to power and

conduct the study. As you may recall, the SNS

study with 25,000 subjects suggested a relative

risk of 2 for respiratory- and asthma-related

deaths for salmeterol versus salbutamol.

An interim analysis was planned after

approximately half the subjects were enrolled.

Prespecified stopping criteria were the following,

a relative risk of 1.4 for the primary endpoint and

a relative risk of 3 for asthma-related deaths,

with an alpha of 0.01.

An interim analysis was performed in 2002

after 26,000 subjects were enrolled. The DSMB

reviewed the interim analysis data in a blinded

fashion. The data suggested a potential treatment

group difference, thus, the DSMB was unblinded.

After unblinding the DSMB, the data suggested an

increased risk for salmeterol use. An exploratory

subgroup analysis suggested that African Americans

195

could be at particular risk. However, the study

did not meet the prespecified stopping criteria.

The DSMB recommended to continue the study if

timely recruitment was feasible. If timely

recruitment was not feasible, the DSMB recommended

to terminate the study and disseminate the findings

within 3-6 months to the clinical and research

community. Due to difficulties with enrollment and

the interim analysis finding, the sponsor

terminated SMART in January, 2003.

Before discussing the results of SMART I

would like to note the following: The results are

from a terminated study which did not meet

prespecified stopping criteria. I think it is

important to note as I present the results that the

non-inferiority objective was not met. In fact,

the data I will show you suggested a difference in

some endpoints between salmeterol and placebo.

The results are based upon the 28-week

treatment period. The protocol specified that

investigators could report SAEs and deaths for up

to 6 months after the 28-week treatment period.

196

The initial SMART data submitted to the agency

included events collected not only from the 28-week

treatment period but also events spontaneously

reported in a 6-month post-study period. The

agency believes the data from the 28-week treatment

period is clinically the period of interest. Thus,

the results I will discuss will be based upon the

28-week treatment period.

The results also include data from the

National Death Index search. Although this was not

specified in the protocol, the agency determined

the NDI search data was acceptable to capture

outcomes during the 28-week treatment period.

Finally, the results are based upon life table

analyses to help account for censoring the subjects

during the study.

This table shows the subject disposition

for SMART. As you can see, subject disposition was

similar between treatment groups and other

categories of disposition not listed, such as loss

to follow-up, were well matched between the

treatment groups.

197

Similarly, subject demographics were

similar between the treatment groups, with the mean

age in both groups of 39 years of age; 64 percent

females and 36 percent males in both treatment

groups. Note that the ethnic origin was similar

and that the majority of the subjects were

Caucasian, with 18 percent African Americans.

The results for the primary endpoint,

which was the combined respiratory-related deaths

or respiratory-related life-threatening experiences

are shown on this table. Respiratory-related and

life-threatening experiences, again, were defined

as intubation and mechanical ventilation. The

relative risk for the total population is 1.4.

Note that the confidence interval does not exclude

1 but the lower bound approaches 1. Remember that

the study was terminated early. If the study had

continued, it is possible the confidence interval

would have tightened and excluded 1.

On post hoc subgroup analysis Caucasians

do not appear to be at increased risk, however,

African Americans had a relative risk of 4.1, with

198

confidence intervals that excluded 1. Thus,

African Americans appear to be at particular risk

for the primary endpoint.

This table shows the results for some of

the key secondary endpoints for SMART. The

following should be noted: Note that the number of

events was low. For asthma-related deaths there

were 16 deaths in 26,000 subjects. For the total

population asthma-related deaths were increased in

the salmeterol group with the relative risk of 4.37

and confidence interval that excluded 1. These

results are similar to the results of the SNS study

which showed a numerical increase in respiratory

and asthma deaths with a relative risk of 3 for

salmeterol versus salbutamol. Respiratory-related

deaths could include other causes of death, such as

pneumonia, in addition to asthma-related deaths.

For the total population respiratory-related deaths

were also increased in the salmeterol group with a

relative risk of 2.16 and confidence intervals that

excluded 1.

Subgroup analyses for asthma-related

199

deaths and respiratory-related do not suggest a

difference in risk between Caucasians and African

Americans. For combined asthma-related or

life-threatening experiences the salmeterol group

was noted to have more events, with a relative risk

of 1.7 and confidence intervals that excluded 1.

Subgroup analyses suggest that the African American

subgroup is driving the results for the total

population. In the African American subgroup the

relative risk is 4.92 with confidence intervals

that excluded 1. Although not on this slide, it is

important to note that there is no difference in

all-cause death or all-cause hospitalizations

between treatment groups.

What about the effect of inhaled

corticosteroid use on the outcomes? As you heard

earlier, SMART was not designed to look at the

effect of inhaled corticosteroid use on outcomes.

However, because of their role in the management of

asthma, information regarding inhaled

corticosteroid use is of interest. The following

should be noted though, a subgroup analysis looking

200

at the effect of inhaled corticosteroid use was not

prespecified in the protocol. All analyses looking

at inhaled corticosteroid use are post hoc

exploratory analyses. Inhaled corticosteroid use

was recorded at baseline only. Inhaled

corticosteroid use was not randomly assigned.

Baseline inhaled corticosteroid use may reflect an

imbalance in other factors that could influence

clinical outcomes. Therefore, any difference in

outcomes between groups, defined by baseline

inhaled corticosteroid use, may not be attributable

to inhaled corticosteroids. Approximately half of

the total population used inhaled corticosteroids

at baseline and 38 percent of African Americans

used inhaled corticosteroids at baseline.

This table shows the post hoc analyses for

the primary endpoint by baseline inhaled

corticosteroid use. On the left side of the table

are the results for subjects using inhaled

corticosteroids at baseline, and on the right side

of the table are the results for subjects who

didn't use inhaled corticosteroids at baseline.

201

In the total population the relative risks

are similar for subjects who used inhaled

corticosteroids at baseline and subjects who did

not use inhaled corticosteroids at baseline.

Recall that in the African American subgroup there

was a strong signal for the primary endpoint. When

analyzed by inhaled corticosteroid use there was an

increased number of events in the salmeterol group

versus placebo for the primary endpoint whether on

inhaled corticosteroids or not at baseline. Thus,

African Americans appear to be at increased risk

for the primary endpoint regardless of baseline

inhaled corticosteroid used.

This table displays the key secondary

endpoints for the post hoc exploratory inhaled

corticosteroid use analyses. Again, the numbers

are small, making it difficult to draw any

definitive conclusions. However, note again that

in the African American subpopulation there was an

increased risk for salmeterol for these secondary

endpoints regardless of baseline inhaled

corticosteroid use.

202

Although definitive conclusions regarding

inhaled corticosteroid use cannot be made from the

SMART data, the data suggests that the risk for

salmeterol exists regardless of baseline inhaled

corticosteroid use. Thus, the agency recommended a

boxed warning on Advair, the sponsor's combination

product.

To summarize SMART, SMART was a large,

simple safety study in 26,000 subjects, and was

stopped early due to interim analysis findings and

difficulty with recruitment. For the total

population the relative risk for the primary

outcome events, respiratory-related deaths or

respiratory-related life-threatening experiences,

was 1.4. The confidence intervals approached 1 but

did not exclude 1. The relative risk for

asthma-related deaths was 4.37, with confidence

intervals that excluded 1. The relative risk was

2.16 for respiratory-related deaths, with

confidence intervals that excluded 1. The data

suggests that there was a treatment group

difference favoring placebo for the endpoints shown

203

on this slide.

In a Caucasian subpopulation there were no

treatment group differences for the primary

endpoint events, but there was an increase in

asthma-related deaths and respiratory-related in

the salmeterol group. In the African American

subpopulation there was a numeric increase in the

salmeterol group for the primary endpoint events,

asthma-related deaths and respiratory-related

deaths, and also combined asthma-related or

life-threatening experiences and on that endpoint

the confidence interval actually excluded 1. As we

discussed, no definitive conclusions regarding

inhaled corticosteroid use can be made in the SMART

study.

Now that I have addressed the controlled

postmarketing studies, let's look quickly at the

postmarketing spontaneous adverse event reports for

salmeterol. The Adverse Events Reporting System,

or AERS, was reviewed for deaths reported for

salmeterol use between May, 1994 and February, 2005

and 201 deaths were reported with salmeterol use in

204

the United States. These cases were reviewed by

the Office of Drug Safety. Ninety-one of the

deaths were determined to be asthma-related and 10

were possible asthma-related deaths. After a

review of the reports the determination was that it

is difficult to draw any conclusions regarding

salmeterol use in asthma-related deaths and

postmarketing reports. In general, it is

challenging to analyze post reports for events that

are associated with the underlying disease such as

asthma-related deaths.

Next I would like to briefly mention some

of the sections of the label that have information

related to the SMART findings. The most

significant labeling change in response to the

SMART results is the boxed warning shown above, and

I think you were shown this earlier. In addition,

a summary of SMART and results of the primary and

key secondary endpoints were added to the clinical

trial section, and there are copies of the product

labels in the briefing package for details. Note

that these labeling changes were made to all

205

salmeterol-containing products.

To summarize what I have discussed, I

reviewed the regulatory history for salmeterol and

specifically focused on the agency's handling of

the safety concerns with salmeterol including

several labeling changes. I discussed the SNS

study which showed a numerical increase in

respiratory and asthma deaths in the salmeterol

group. However, these were not statistically

significant and showed fewer withdrawals due to

respiratory or asthma events, and this was

statistically significant.

I discussed SMART, which was a large,

simple safety study, stopped early due to interim

analysis findings and difficulties with enrollment.

The results of SMART suggest that there is a

difference in outcomes between salmeterol and

placebo. In the total population there was a

relative risk of 1.4 for the primary endpoint

although the confidence interval did not exclude 1.

In addition, an increase in asthma- and

respiratory-related deaths, with confidence

206

intervals that excluded 1, was also noted for the

total population.

In the African American population there

was an increase in primary events and an increase

in combined asthma-related deaths or

life-threatening experiences, and both of these

endpoints had confidence intervals that excluded 1.

SMART was not designed to assess the effects of

inhaled corticosteroid use on outcomes.

Because of the postmarketing studies, the

product labels have been updated twice, including a

boxed warning for all salmeterol-containing

products, and I have shown you the boxed warning

and the labels are in your packet.

So, based upon the safety concerns raised

in the postmarketing studies, we pose the following

questions to the committee: The product labels of

salmeterol-containing products have been modified

to include warnings related to the SMART study.

Based on currently available information, what

further actions, if any, do you recommend that the

agency take to communicate or otherwise manage the

207

risks of severe asthma exacerbations seen in the

SMART study?

Based on the currently available

information, do you agree that salmeterol should

continue to be marketed in the United States?

Note that question one is slightly

different than the question in the briefing book

and that we used the term "severe asthma

exacerbations" in this question.

Question three, also related to

salmeterol, which is what further investigation, if

any, do you recommend to be performed by GSK, the

sponsor, that can improve the understanding of the

nature and magnitude of the risk of salmeterol?

That concludes my presentation and I would

like to turn the podium over to Dr. Gunkel.

Formoterol

DR. GUNKEL: Good afternoon. My name is

Harry Gunkel. I am also a medical officer in the

Division of Pulmonary and Allergy Drug Products. I

will be reviewing with you some data pertaining to

the second of the drugs that we are considering

208

today, formoterol.

In this portion of the program, after

briefly introducing the product, I will summarize

those elements of the regulatory history that have

brought us to this point today. As Dr. Chowdhury

stated this morning, and somewhat differently from

the situation with salmeterol, the story of asthma

exacerbations win formoterol was really told in the

NDA review of the Phase 3 studies. So, we will

spend some time reviewing the results of those

studies that are relevant to today's topic.

Next, we will briefly review the Phase 4

postmarketing study with Foradil; then report the

findings from a recent review of spontaneous

postmarketing reports, and then summarize and offer

some concluding observations.

The only formoterol product approved in

the United States at this time is Foradil,

manufactured by Novartis who we heard from this

morning. Some of the data that I will show you

today will look familiar to you. Foradil is a dry

powder formulation for inhalation with the

209

aerolizer device. It is a racemate of two

enantiomers of formoterol fumarate.

So, let's briefly review some of the

relevant milestones of the regulatory history of

Foradil. In a few minutes I will review in more

detail the important Phase 3 studies and the

results. For now, I want to just point out the

findings that affected Foradil's regulatory status.

Foradil is approved for asthma, COPD and

exercise-induced bronchospasm but, as Dr. Chowdhury

noted earlier, our interest today is confined to

asthma.

The new drug application for Foradil for

use in asthma was first submitted in June of 1997.

The clinical program that comprised the NDA

investigated 2 different doses of Foradil, 12 mcg

administered twice daily and 24 mcg twice daily.

You will see as we go on that these doses, which

were used in adolescents and adults, were also used

in children.

The result from the Phase 3 studies that

concerns us today was the finding that patients who

210

had received the higher dose of formoterol

experienced more serious asthma exacerbations than

those who received the lower dose, and we will see

those specific results in just a moment. At the

same time, the reviewers found no evidence that the

higher dose consistently resulted in greater

efficacy. Therefore, only the lower dose of

Foradil, the 12 mcg BID dose was ultimately

approved in February, 2001. We saw in Dr.

Seymour's presentation that the events of concern

that occurred with salmeterol did not occur

commonly and so it was also with serious asthma

exacerbations that were observed win formoterol.

The Division was interested in further

investigation of the event but believed that

routine postmarketing surveillance would not be

eliminating when the event of concern was also the

underlying disease being treated. So, given these

circumstances, the Division asked Novartis to

commit to conduct a Phase 4 study. Novartis did so

and conducted a Phase 4 study to obtain additional

information about doses of formoterol other than

211

the one approved.

It is important to note two things.

First, the Division's judgment about the relative

safety of the higher dose of formoterol was

informed by the results of the Phase 3 studies in

the NDA and not from the results of the Phase 4

study that followed. Second, the Phase 4

formoterol study was planned and designed before

results of the SMART study were known.

With that regulatory background, let's

move on to review some of the Phase 3 data. For

all practical purposes, 3 clinical studies

comprised the pivotal evidence for the efficacy and

safety for formoterol in asthma. Dr. Geba

introduced you to studies 040 and 041 that were

performed in adolescents and adults 12 years of age

and older. The third study was 049, a study of

children 5-11 years of age. The FDA reviewer's

conclusions about the safety of the 2 doses of

formoterol were primarily formed from the results

of these 3 studies.

Studies 040 and 041 were essentially

212

identical, except that 041 included some

pharmacokinetic measurements. Both studies were

randomized, double-blind, placebo- and

active-controlled 12-week studies in asthmatics 12

years of age and older who had FEV1 40 percent or

more of predicted and 15 percent reversibility of

their bronchoconstriction.

Patients were randomized in approximately

equal proportions to 1/4 treatments, Foradil 12 mcg

BID, Foradil 24 mcg BID, albuterol 180 mcg QID or

placebo. The pediatric study, 049, enrolled

children from 5-12 years of age whose FEV

1 was

50-85 percent of predicted and who also had 15

percent reversibility.

This study was similar to the other

studies in using the same doses of Foradil, 12 mcg

and 24 mcg BID, but was different in not including

an active control group. Also note that this was a

one-year study with an objective of evaluating for

long-term safety of Foradil for children.

Before we review the adverse event results

of these studies, let's look at results of the

213

bronchodilator effects of the treatments. This

graph summarizes the results. The graph displays

FEV 1 results over the 12-hour post-dose

period at

the last study visit. Time in hours is shown

across the horizontal axis. Average FEV

1 in liters

is on the vertical axis. These results are from

study 040 and the results were essentially the same

in study 041.

Results for the placebo group are shown in

open circles and for the albuterol group in open

triangles. The other 2 curves represent the

formoterol doses. The 24 mcg dose of formoterol is

shown in closed circles and the 12 mcg dose is

shown in open squares. As you see, bon formoterol

doses were significantly better than placebo.

Although the 24 mcg dose was significantly better

than the 12 mcg dose at some individual time

points, there were no significant differences at

other time points and the results were inconsistent

between the 2 studies. There were no differences

in the areas under the curve. As mentioned

earlier, the lack of difference in efficacy between

214

then formoterol doses was weighed in evaluating the

adverse event findings.

This slide summarizes the heart of the

matter. The rates of serious asthma exacerbations

are shown in each of the 3 NDA pivotal studies. As

you see, the absolute rates of the events are low

but the differences between then formoterol doses

in each study are evident nevertheless. In each of

the 2 adult studies there were more serious asthma

exacerbations in the 24 mcg dose group than in the

12 mcg dose group, and more than in the albuterol

and placebo groups as well.

Three adult patients had events of

particular severity. One patient in study 040 who

received the 24 mcg dose, a 24 year-old male,

required intubation and mechanical ventilation for

his exacerbation. In study 041 a 66 year-old

female experienced cardiorespiratory arrest and

died, and a 49 year-old man had a respiratory

arrest but survived.

Note that the overall rates are higher in

the children in study 049, as was pointed out this

215

morning, but also that the proportionally greater

rate of events in the higher don formoterol group

is maintained in this study. Recall two factors

that might contribute to the higher incidences of

events in children. First, there was no adjustment

of doses given to children. So, these higher rates

may reflect relatively higher doses given to

children on a body weight basis. But also, the

pediatric study duration was 1 year versus 12 weeks

for the other 2 studies, allowing more time for

events to be reported.

So, to summarize the relevant issues that

arose from the Division's review of the Foradil

NDA, as a result of more serious asthma

exacerbations occurring in the higher dose

formoterol group and no efficacy advantage, that

dose was not approved. Serious asthma

exacerbations were seen consistently across the 3

pivotal studies and were more pronounced in the

pediatric study. Finally, a commitment was made to

conduct a Phase 4 study to obtain additional

information.

216

It is important to reiterate here that the

Division's concern about asthma exacerbations

associated win formoterol had been substantiated

within the NDA itself, and it was not the purpose

of the Phase 4 study to do that. That said, let's

examine the Phase 4 study and its results.

This was a randomized, parallel group,

placebo-controlled study with a 16-week treatment

period. There were 5 clinic visits during the

treatment phase of the study. For the study to

provide useful comparison to the Phase 3 results,

it was desired that the patients enrolled in this

study should be as similar as possible in

characteristics likely to affect the outcome of

primary interest, asthma exacerbations. Therefore,

patient entry criteria were in most identical to

those used in the Phase 3 studies. Note that

children were not included in this study. Indeed,

as we will see in a moment, the original study

protocol was confined to adults and it was only

upon amendment that adolescents down to 12 years

were allowed in the study. Events that might have

217

indicated recent exacerbations or altered state of

the underlying disease as indicated by changing

medications were appropriately a basis for

exclusion from the study.

Patients were randomized in approximately

equal proportions to receive 1/4 treatments during

the study, as shown. The 2 Foradil adult dosage

groups were treated in double-blind fashion, while

the fourth group, who received extra on demand

doses of Foradil, were treated in an open-label

fashion. Albuterol rescue was allowed during the

study, with more doses allowed for the 3

double-blind treatment groups.

The study was conducted over a 2-year

period between February, 2002 and March, 2004. All

told, 2,085 patients received treatment. Of those,

about 86 percent completed the study. Of the 294

patients who did not complete the study, the most

common primary reason was occurrence of an adverse

event. This was the case for 103 patients overall

or 4.9 percent of the total treated population.

Other reasons for discontinuing the study early are

218

noted on the slide.

After this study started the Global

Initiative fir Asthma promulgated guidelines for

the management of asthma. The sponsor of the study

and the investigators determined that the study

criteria were not consistent with these guidelines,

particularly in the use of inhaled corticosteroids,

and so the protocol was amended during the course

of the study to, in effect, liberalize the

concomitant use of inhaled corticosteroids.

A third amendment about half way through

the study was enacted in order to accelerate

enrollment because patient accrual was lagging

behind projections. This amendment made several

changes, including lowering the age of eligibility

to 12 years; changing the criterion for FEV

reversibility from 15 percent observed to 12

percent historical; and shortening the washout

periods required for other medications for example.

The next 2 slides display the baseline

characteristics of the 4 treatment groups. In the

far right column of these tables the same

219

characteristics from all patients in the Phase 3

studies are shown to allow us to examine whether

the populations were reasonably comparable. For

some characteristics, for example, age and gender,

the populations were similar. Note, however, that

the proportion of African American patients in the

Phase 4 study was about twice that in the Phase 3

studies. Also note the higher FEV1 reversibility

of patients in the Phase 4 study, which probably

reflects the amended entry criteria that were just

described.

For characteristics indicating acute

asthma exacerbations in recent past, the preceding

year, the study treatment groups were about the

same. These data suggest a population in

relatively good control, with fewer than 10 percent

needing an ER visit or hospitalization in the

preceding year.

On this slide is a summary of the outcomes

of primary interest from this study, adverse

events. The treatment groups and the number of

patients in each are shown across the top row. No

220

deaths at all occurred in this study. More than 50

percent of patients experienced an adverse event of

some kind. Between about 10-16 percent of patients

experienced asthma-related adverse events.

Determining whether an adverse event was

asthma-related or not was prospectively defined by

the study protocol. It was an event with one of

the following MedDRA preferred terms, cough,

wheezing, dyspnea, dyspnea exacerbated, status

asthmaticus, respiratory distress, bronchospasm,

acute respiratory failure or hypoxia. the fourth

row down shows the number of those asthma-related

AEs that met the regulatory definition of serious.

There were only 9 such events in total in this

study of more than 2,000 patients. In all 9 cases,

hospitalization was the event that categorized the

event as serious. One patient required intubation

and mechanical ventilation for his exacerbation.

He was a 51 year-old man who received the 24 mcg

dose of formoterol on the study.

Serious asthma exacerbations per se was

not a specific endpoint prospectively named or

221

defined by the sponsor in this study. I call your

attention to the next row in the table. I would

like to state that no patients with serious AEs

were excluded in this review. However, we were

interested in which patients with serious

asthma-related AEs actually had specifically

serious asthma exacerbations. There were 2 such

patients who had a serious asthma-related AE which

was not an exacerbation. The 2 that were not were

both in the low formoterol dose treatment group.

In both these patients the verbatim event that met

the asthma-related criterion was respiratory

distress. In one patient, however, the respiratory

distress was judged due to a myocardial infarction

and, in the other, due to pneumonia. Finally, the

last row in the table shows the number and

proportion of patients in each group who had an

asthma exacerbation of any kind or seriousness.

This table provides more detailed

information about the 9 patients with serious

asthma-related AEs. There were 12 events in the 9

patients, with some patients having more than 1

222

event. Of these 9 patients, 2 were African

American.

This slide more succinctly summarizes the

key results from the formoterol data we have been

considering. It shows the number and proportions

of patients by treatment who had serious asthma

exacerbations. The 4 studies of interest, the 3

pivotal Phase 3 studies and the Phase 4 study, are

shown in the 4 rows of the table.

To recapitulate, serious asthma

exacerbations occurred more frequently in the

higher don formoterol group in the Phase 3 studies

and the effect was more pronounced in the children.

No difference was seen in the Phase 4 study, and

this slide illustrates that the rate of events in

the 24 mcg dose group was, in fact, quite a bit

lower overall in the Phase 4 study than in the

Phase 3 studies.

Before concluding, let's briefly go over

the results of a recent review of postmarketing

spontaneous reports on formoterol. The Adverse

Event Reporting System database contains 180

223

domestic reports for formoterol as this past June

14th. Eleven of those reports were for

bronchospasm and obstruction events. Since the

year of marketing, 2001, there have been 4 domestic

reports of deaths in patients receiving formoterol.

Two of those deaths were caused by myocardial

infarction and the causes for the other 2 were not

reported.

The following points are offered in

conclusion: First, the observation made upon

review of the Foradil NDA that serious asthma

exacerbations occurred more frequently with 24 mcg

BID of formoterol than with 12 mcg BID was the

basis for not approving the higher dose.

Second, the serious asthma exacerbations

were more frequent overall in children who received

the same nominal doses of formoterol as adults in

children and were studied in a 1-year study.

Finally, a Phase 4 study of limited size

and restricted to adults and adolescents did not

provide any additional information about these

events.

224

With that brief review, let's restate the

questions to the committee: The label of the

formoterol-containing product does not include

warnings comparable to the warnings that are

present in the salmeterol-containing products.

Based on the currently available information,

should the label of formoterol-containing products

include warnings similar to those in the salmeterol

label?

And, based on the currently available

information, do you agree that formoterol should

continue to be marketed in the United States?

Next, what further investigation, if any,

do you recommend to be performed by Novartis that

can improve the understanding of the nature and

magnitude of the risk of formoterol? Thank you.

Questions for the Speakers

DR. SWENSON: We now have about 10 minutes

to take questions to both these speakers for the

FDA presentation. Dr. Schoenfeld?

DR. SCHOENFELD: I guess I said this

before but I will repeat it, I think that if we are

225

going to do a risk/benefit analysis at some point,

then to counterbalance the benefit the risk should

be couched in terms of the attributable risk, which

would be the difference, I guess, in the event rate

on the two treatments. I have calculated that,

roughly speaking but I just had to do it

on-the-back-of-an-envelope, that for salmeterol the

risk for asthma deaths--and it is similar for

everything else, there are about 10 events per

26,000 patient-years of follow-up. So, the risk is

1/2,600. I just wondered, you know, do you have

any tabulation in your database of all those risks

for the various subgroups, or at least could you

confirm whether I am right here?

For the other compound, the actual total

patient follow-up that is reported is something in

the order of 260 patient-years of follow-up. So,

it is so small an amount of follow-up that risk at

the rate of 1 per 2,600 would be completely

undeterminable.

But if someone in the agency is sort of

done these calculations, I would much prefer to see

226

those rather than what I do on the back of an

envelope.

DR. SEYMOUR: We have not performed those

calculations based on patient years.

DR. SWENSON: Dr. Schatz?

DR. SCHATZ: I was wondering, in the SMART

study, whether the NDI surveillance included those

who had discontinued the trial.

DR. SEYMOUR: I have to defer that to the

sponsor; I am not quite sure.

DR. KNOBIL: The NDI database included

everyone who had been enrolled in the study whether

or not they had discontinued study drug.

DR. SWENSON: Dr. Gay?

DR. GAY: I will restate a question that I

asked previously to the sponsor. We do see an

increase in adverse events in children but,

clearly, that is a much longer study. Is there any

breakdown of the timing of these events within 90

days, within 180 days, within the final 6 months of

the study?

DR. GUNKEL: I have a little bit of

227

information, not everything that you would like. I

was able to find that information for the children

who received the 24 mcg dose, the dose really of

concern. There were 11 children who had serious

asthma exacerbations in that study overall. Five

of those 11 occurred after day 84, which would be

12 weeks. So, if we try to compare the children to

the adults, for example, that were in a 12-week

study, then the rates up to that 12-week time point

are roughly comparable in children versus

adults--with all the usual caveats about the

studies weren't designed to do that, and so forth.

I don't have similar information for the 12 mcg

dose group.

DR. SWENSON: Dr. Brantly?

DR. BRANTLY: This is a question for Dr.

Seymour. Dr. Seymour, we know that socioeconomic

status plays a big role in adverse events and

asthma medications. I was wondering if there had

been any attempt to stratify the cohort for the

SMART trial as far as socioeconomic status,

particularly those individuals that were associated

228

with severe adverse events.

DR. SEYMOUR: I am not aware of any

stratification but I can ask the company if they

have done any type of socioeconomic stratification.

DR. KNOBIL: We did not collect much SES

data. What we did have, we had educational level

and we had zip code. So, we only have very crude

measures. So, we did what we could with what we

had and we didn't see any impact of income based on

zip code or educational level.

DR. SWENSON: Dr. Martinez?

DR. MARTINEZ: I have two questions for

Dr. Seymour. First, if I understand correctly from

your slide 19, discontinuations were very similar

in the salmeterol and the placebo group and,

therefore, in this case attributing the results

perhaps to different rates of discontinuation in

both groups wouldn't be fair. Am I interpreting

the data correctly?

DR. SEYMOUR: Yes, I think so. The

discontinuation rates were similar between

treatment groups.

229

DR. MARTINEZ: The second question is that

in the materials provided to the advisors there is

a stratification by the baseline percent predicted

peak flow, which is the only type of information

that I could gather with respect to severity. If

we are going to consider that 60 percent peak flow

or below is more severe than more than 60 percent

peak flow, you haven't commented at all about that.

The interpretation that at least I make is that

there appear to be different ways in which this

behaves in this post hoc type of analysis. For

combined respiratory-related death or

life-threatening experiences, it appears that it is

those subjects who have more severe disease, if

peak flow can be considered disease or even perhaps

more related to what could be considered COPD in

adults, who show more risk than those with a higher

peak flow. For asthma-related deaths, it is

difficult to say anything but there doesn't seem to

be a very clear difference between the two groups.

Has the agency interpreted that table in any way

similar or different to the one I just proposed?

230

DR. SEYMOUR: I need to know what page you

are referring to.

DR. MARTINEZ: It is page 45 of the

materials provided to us, page 45 of the section

called "salmeterol postmarketing study review,

SMART study."

DR. SEYMOUR: Just a second. I can tell

you I don't think we have formed any formal

conclusions based on this data.

DR. MARTINEZ: Just to propose an

interpretation, what I can see here is that the

rates that are observed for asthma-related deaths,

although they cannot be calculated for the less

than 60 percent, see by the absolute numbers don't

look very different. In other words, there appears

to be an increase in asthma-related deaths both for

those that have more than 60 percent and for those

that have less than 60 percent.

DR. SWENSON: Dr. Schoenfeld?

DR. SCHOENFELD: I just want to understand

whether the analysis of severe adverse events was

intent-to-treat. That is, if a patient stopped

231

medication but then still, during the 28 weeks of

follow-up in the SMART study they had an event,

they would count it, I assume? And, what about in

the other studies?

DR. SEYMOUR: My understanding of SMART is

that that is correct, it was on an ITT.

DR. SCHOENFELD: And the other studies?

DR. GUNKEL: The same.

DR. SWENSON: Dr. Moss?

DR. MOSS: I have a question that kind of

builds on what Dr. Schoenfeld talked about and a

little bit about what Dr. Meyer talked about this

morning. It seems to me that we are being asked to

compare two studies that are different in terms of

the size of the studies. So, it is very hard for

us in the Foradil study to draw conclusions be the

study is a lot smaller. I was just wondering, Dr.

Meyer, if you could talk in a little bit more depth

about why the SMART study had 26,000 people in it,

and what was the thinking of the FDA to have the

Foradil study only have 2,000 or so in it. I

realize you are looking for different outcomes but

232

what was your thought process in asking them to do

those Phase 4 studies very differently?

DR. MEYER: Sure. I think, first off, it

would be wonderful to have such a large outcome

study of formoterol but we don't, and that was not

the purpose of the Phase 4 commitment. In the

Phase 4 commitment we saw a dose response or an

apparent dose response phenomenon for the outcome

of serious adverse events in 12-week studies and we

wanted to reassure ourselves, since we were not

approving the 24 mcg dose, that in fact this was a

real finding. We also wanted further data to

relate even the 12 mcg dose to placebo in that same

kind of setting.

So, while we, again, would have liked to

have had a SMART-like study of formoterol as well,

that really wasn't the question that was being

posed in asking for the Phase 4 commitment. The

Phase 4 commitment was really to try to better

clarify what we had seen in the Phase 3 studies in

terms of an apparent dose relationship to adverse

events and whether that 24 mcg dose really would

233

prove to have a clear safety signal in relationship

to the 12 mcg dose.

Open Public Hearing

DR. SWENSON: Any further questions?

[No response]

At this point, we are moving into the open

public hearing session. To begin this I need to

read a short statement and we will have at least

one discussion or statement from a public member.

Both the Food and Drug Administration and

the public believe in a transparent process for

information gathering and decision-making. To

ensure such transparency at the open public hearing

session of the advisory committee meeting, FDA

believes it is important to understand the context

of an individual's presentation.

For this reason, the FDA encourages you,

the open public hearing speaker, at the beginning

of your written or oral statement to advise the

committee of any financial relationship that you

may have with the sponsor, its product and, if

known, its direct competitors. For example, this

234

financial information may include the sponsor's

payment of your travel, lodging or other expenses

in connection with your attendance at the meeting.

Likewise, FDA encourages you, at the beginning of

your statement, to advise the committee if you do

not have any such financial relationships. If you

choose not to address this issue of financial

relationships at the beginning of your statement,

it will not preclude you from speaking.

At this point, I would like to ask Mr.

Chris Ward to come to the podium. Mr. Ward, your

podium.

MR. WARD: Thank you. Good afternoon. My

name is Chris Ward. I am an asthma and allergy

patient and current president of the Asthma and

Allergy Foundation of America. We have not

received payment from any entity for this testimony

or for the cost of our travel and participation in

this meeting today.

On behalf of the almost 20 million

Americans with asthma, AAFA appreciates the

opportunity to testify to this advisory committee

235

concerning the safety of long-acting beta agonist

bronchodilators. Since 1953, AAFA has been

dedicated to improving the quality of life for

people with asthma and allergies. Patients, their

families and their caregivers turn to our

organization for education, research and advocacy.

AAFA appreciates the heightened vigilance

at the FDA regarding drug safety and thanks the

advisors for reviewing the available data and

meeting today to discuss potential safety concerns

with this class of drugs. Asthma, of course, is a

treatment-intense condition for many patients and

your advice to the agency today will impact

millions of individuals who depend on these

products as part of their regimen for asthma

control.

There are three key messages I would like

to convey on behalf of patients with asthma.

First, as we understand it, there are no concerns

with the efficacy of this class of drugs and their

important role in asthma control, which is

reflected in both the national and international

236

guidelines for asthma clinical care that use the

word "preferred" when recommending these products

in conjunction with inhaled corticosteroids for

moderate to severe persistent asthma. There is

strong, consistent evidence from clinical trials

that this approach leads to improvements in lung

function and symptoms, and reduces the need for

short-acting beta agonists. When we weigh this

evidence of effectiveness against the evidence of

potential risk which is, at best, still undefined,

we believe it would be difficult for asthma

patients to understand why these products would not

continue to be available to them.

Second, we believe there is an element of

scientific and clinical progress in asthma that may

be missing from this discussion. Mainly, there

seems to be progress in the pharmacogenomic

understanding of how beta agonists have different

effects in different individuals. We believe the

results of the government trial, published last

October, demonstrating different responses to the

short-acting beta agonists based on genotype is an

237

important step forward. We understand that a

similar clinical trial is just now getting underway

for the long-acting beta agonists. In other words,

from the perspective of asthma patients, there is

current and ongoing investigation into this

important clinical question. We understand very

well that a subset of patients may be at higher

risk than others. But we are only beginning to

understand why this is the case. At this point in

time then, it would seem advisable to defer any

conclusive decision about the availability of these

drugs to all patients until there are more

definitive answers.

Third, with regard to the consideration of

whether the labeling changes to salmeterol should

be approached with formoterol, it is our position

that FDA's responsibility lies in working closely

with the product sponsor to answer this question

and to determine a product label that is consistent

with the available medical evidence. In fact, we

are encouraged that both product sponsors are

working closely with the agency to further

238

understand and clarify the nature and magnitude of

potential risk of this class of drugs. We believe

the safety of asthma patients should be front and

center in this ongoing work.

In conclusion, we certainly urge all

parties to continue this important discussion.

however, AAFA believes that at this time there are

too many questions to be able to draw conclusive

decisions on medications that have been effective

for millions of patients to be able to control

their asthma.

Again, on behalf of these patients, I

thank you for the opportunity to testify on this

important issue and I am pleased to answer any

questions you might have. Thank you.

Committee Discussion

DR. SWENSON: Thank you, Mr. Ward. At

this point the meeting schedule calls for a break

but we are somewhat ahead of schedule so what I

would like to do is to move into our committee

discussion section, and we will have a break at

some mid point there. The first part then of this

239

committee discussion would be open now I think to

general questions to all parties involved and

further exposition of points that may not have been

raised earlier this morning. We will then break,

at which time we will come back to our vote on the

specific questions that are being asked of us by

the FDA. So, to begin with, we had two people

earlier this morning that I had to close out and I

will ask Dr. Kercsmar if she wishes to pose her

question from this morning.

DR. KERCSMAR: I had a question regarding

other medication use in the SMART trial that maybe

the sponsor can answer. During the 4-week

follow-up telephone calls, were there any data

obtained on other medication use, particularly the

use of short-acting beta agonists?

DR. KNOBIL: Well during each 4-week

follow-up telephone call the patients were asked if

they were continuing on the medications that they

had said that they had started at baseline or at

the previous call, and if they had started or

stopped any new medications or any of their old

240

medications. So, if anything was stopped,

obviously, they were asked which ones were stopped

and which ones were started.

During the course of the study there were

some patients who did stop and start medications

but, for the most part, throughout the study

patients remained on their beta

2 agonists. We

didn't ask them how much of their short-acting beta

agonists they were taking however so we don't have

any information on that.

DR. SWENSON: Dr. Gay, you had one

question from this morning.

DR. GAY: It was answered earlier, thank

you.

DR. SWENSON: Dr. Gardner?

DR. GARDNER: This question is for Dr.

Knobil as well. It is very impressive that you

have done such a large study and it must be as

frustrating to you as it is to us to hear people

periodically say, as they have today, well, we

really can't learn anything from this because there

were only 26,000 people there, or something like

241

that; we can't draw any conclusions. And, I know

how difficult these are to do. However, you have

given us data also from the SNS study which was

conducted in England, and published, and it is

commented that in one year of enrollment they

enrolled 25,000 people. And, in the SMART trial

over seven years you had about the same number of

people. I am wondering what changed during the

intervening approximately five to six years that

made enrollment so much more difficult here.

Because, Dr. Castle comments in her paper that

companies are often accused of delaying things so

that they can stop enrollment when they aren't

getting enough, and it is kind of an indictment by

Dr. Castle who, by the way, was a Glaxo employee.

So, can you help us understand why it seemed to

have been easier to enroll that many people in a

year in England?

DR. KNOBIL: Well, I can't really comment

on how easy it was to enroll in England since I

wasn't involved with the initial study. However, I

do know that for SMART one of the major stumbling

242

blocks was the fact that patients could not have

had any exposure to long-acting bronchodilators to

begin with. As enrollment waned over the course of

the time, we actually held focus groups with

physicians and patients to figure out why we

weren't able to get more patients in the study more

quickly. Again, what came out was that many

patients were already taking long-acting inhaled

beta agonists. They were not interested in the

placebo-controlled trial. There were a lot of new

medications coming out during the time that the

study was running. So, the interest level for the

study went way down. That is the feedback we got

from our investigators and the patients who may

have been enrolled. So, those were the major

factors.

DR. KERCSMAR: That is helpful. Thank

you.

DR. SWENSON: Dr. Schatz?

DR. SCHATZ: Well, I hate to keep going

back to the same thing but I think it is important

to understand. Now I am confused. Intent-to-treat

243

based on people who discontinue medication but are

still in the study is one thing but then there are

the people who drop out. Although the percentages

are similar statistically, numerically they are

still greater in the salmeterol group. Do you

think the surveillance for adverse events is

comparable in those who dropped out versus those

who didn't drop out?

DR. KNOBIL: There is one point of

clarification. It was statistically significantly

different in the number of patients who dropped out

of the study in the placebo group versus the

salmeterol group. More patients on placebo did

drop out.

Now, over the course of the study we did

follow up with the patients. Even if they dropped

out of the study and said they didn't want to take

study medication, we still attempted to contact

those people and get follow-up information up to

the 28 weeks and actually even beyond that as the

FDA has already mentioned. We would collect that

information up to 6 months after if we could get

244

that information. But for the deaths we did look

for all patients enrolled in the study by the NDI

database. So, at least for the deaths we were able

to follow-up on those more than, say, patients who

refused to be contacted any further.

DR. SWENSON: Dr. Knobil, I have a

question from this morning that I wasn't able to

ask and since you are standing I will go ahead and

pose it to you. That is, in the SMART study--and

you just alluded to the fact that a zip code

analysis as a proxy for socioeconomic status and

questions unrelated to the biologic activity of the

drug turned out negative, and you are going to

reiterate that that was definitely a negative

finding--

DR. KNOBIL: Yes.

DR. SWENSON: --that you found no

information by that analysis?

DR. KNOBIL: No, and to be fair, the

analysis by zip code is quite a crude way to look

at socioeconomic status and we really didn't find

any differences between the groups and nothing to

245

suggest that that was involved. However, if we had

had more detailed information that might have been

a little bit more helpful.

DR. SWENSON: Okay, and the second part of

my question then was, given that there was a

difference in the use of inhaled corticosteroids

between Caucasians and the African American

subgroup, did you make any effort to attempt any

type of matching with the two groups? For

instance, could you have pulled out a cohort of

Caucasians that had the same rate of inhaled

corticosteroid use as the African American group,

and then did that yield any further information?

You may not have done it but I am just asking.

DR. KNOBIL: Well, we discussed matching

to try to get to a better understanding of the

events, and the numbers of events are so low that

even if you found somebody to match the amount of

information that you would get out of that is not

really very helpful, and I would turn to our

statistician to see if there is anything else that

I could add to that.

246

[Not at microphone; inaudible]

DR. KNOBIL: For those of you who didn't

hear that, he said the event rate was still too low

to be able to do that.

DR. SWENSON: Thank you. Miss Sander?

MS. SANDER: Am I to take it from your

presentation that kids need less formoterol? I

think it said in one of your slides that--no?

DR. GUNKEL: That is one possible

explanation for the difference in the rates that we

saw. The other is simply the duration of the

study, but one other possibility is that they do

need, on a body weight basis, less formoterol but

we can't say that definitively and conclusively

from our data.

DR. SWENSON: Just a quick point here,

when we have this queue of questions it is

difficult for anyone else to activate their

microphone. So, if you have a question go ahead

and signify that you are ready. We will note it

and then have you in line. Thanks. Our next

question is by Dr. Newman.

247

DR. NEWMAN: Thank you. One question that

I have to the sponsors, and maybe the FDA has a

comment on this as well, that is, when I look back

at how, at the SMART study, was powered and then,

in turn, one thing we are struggling with is having

such a low frequency of serious adverse

events--first of all, I am glad there are so few

serious adverse events; fundamentally that is

good--but I wonder if I could have you help explain

to me why that frequency of serious adverse events

was so much lower than was expected. Is there

something different about the kinds of people who

were recruited into these studies that would lead

us to have this much lower rate?

DR. KNOBIL: Well, as you can imagine,

powering the study was very difficult to do because

when enrolling an asthma population there are no

statistics out there that would tell you what the

rate of events would be per asthma patient. For

example, CDC statistics give you the rate of events

for the total population, not just for the patients

who have the disease in question. So, we

248

extrapolated from that. For example, the total

number of asthma deaths that occurred--I believe it

was in 1994 at the time--divided by the estimated

number of patients in the United States, and

started from there.

Then, in order to get some information

about life-threatening events, we sent out a

questionnaire to over 100 hospitals to tell us how

many per asthma death or per respiratory death how

many intubations would you have? So, since there

was no information in the public domain for this,

this was the best way that we could get that type

of information. It turned out it was about 5

intubations per respiratory-related death. Then,

taking into account more severe asthma and the

higher rate in the African American population, we

came up with the powering that we did. If you want

anything more specific than that I will have to

turn it over to someone who was more involved with

the power calculations.

DR. SWENSON: Dr. Prussin?

DR. PRUSSIN: I also have a question for

249

the Glaxo team. You know, inhaled corticosteroid

and long-acting beta agonist therapy is really the

standard of care for all mild asthmatics, and for

the foreseeable future this is going to be the

standard of care as far as we can tell, for the

next ten years--five years, ten years. So, it is a

huge issue that involves huge numbers of patients.

You mentioned a Medicaid study. Can you elaborate

on that a little? Because I think one of the

charges we have is what types of future studies are

needed to address this and at least I am still not

clear as to the details of that study.

DR. KNOBIL: Yes, what I will do is I will

turn it over to Courtney Davis who is an

epidemiologist with us.

DR. DAVIS: Hi. I am Courtney Davis,

senior director of epidemiology for

GlaxoSmithKline. Our Medicaid study involves data

from 7 states in roughly the same time period as

the SMART trial was conducted. It is 1994 through

1999 Medicaid data. For each state the Medicaid

claims data will be matched to the death

250

certificate files for all of those 7 states. So,

that is how we will obtain asthma mortality and

respiratory-related mortality data. In addition,

of course, the claims will have the hospitalization

data equivalent to the SMART outcomes as well.

Because Medicaid is one of the only

available databases for epidemiology research that

includes race, we will be able to stratify the

analyses and look at African Americans and

Caucasians separately so we can address the

differences that were observed by race. And,

because this is longitudinal data on these

patients, including their pharmacy records, we will

be able to look at potential effect modification by

inhaled steroid use as well.

DR. PRUSSIN: Are you going to be able to

stratify asthma severity?

DR. DAVIS: The best we can will be using

proxies that will be available in the claims

database. The longitudinality of the records, of

course, vary. We have data at this point that has

come in for 5 of the 7 states. On average the

251

follow-up is about 12 months, 11.7 months of data.

So, we will be able to look back as far as we can

in terms of prior utilization that includes serious

events like hospitalizations and, of course,

prednisone use.

DR. SWENSON: Miss Schell?

MS. SCHELL: I guess I have some question

or clarification on the SMART survey itself. Since

the variability of the different types of patients

and their ability to assess themselves, and you

just did phone follow-up calls, I wonder what kind

of preparation you had for the patients prior to

enrollment on how to assess themselves, what they

considered was severe. I just don't know that they

were able to answer back consistently over their

course. Just clarification.

DR. KNOBIL: Yes, we had a whole telephone

script that was written in such a way as to be as

understandable as possible to ask patients about

whether they had been hospitalized or whether they

had been intubated. Obviously, you are not going

to use the word "intubated" but whether they have

252

had a breathing machine, something like that. I

will have to look over to someone who has been

involved in the study from the beginning but there

weren't any questions that asked them to assess

just worsening. So, there wasn't any training on,

you know, how should they assess worsening. There

was no diary card. There was no peak flow. There

wasn't anything like that. It was just mainly has

this happened to you? Have you changed your

medicines? Have you started any; have you stopped

any? Have you had any averse outcomes?

DR. SWENSON: Dr. Moss?

DR. MOSS: I want to build a little bit on

what Dr. Prussin was talking about earlier, and

this is a question for either the Glaxo

representatives or the Novartis representatives.

It seems to me that part of the results of this

SMART study and some of the Foradil studies

potentially are related to the decreased use of an

inhaled corticosteroid especially in the African

American population. Would you guys agree with

that? Do you think that some of these adverse

253

events would go away in the Serevent group if 100

percent of the patients were on inhaled

corticosteroids? Do you think that is part of the

issue?

DR. KNOBIL: Obviously it is difficult to

say based on the data from SMART since it wasn't

designed from the start to look at inhaled

steroids. However, there is a suggestion that

inhaled steroids did have a positive effect or

beneficial effect on these outcomes. So, I do

believe that if more people were treated

appropriately for their asthma in accordance with

the guidelines that we probably would have seen

fewer events. I mean, that is just speculation.

DR. MOSS: I mean, we do these studies of

26,000 people and we still are left with not a

definitive answer so at some point everyone is

going to have to base things, in their own mind, on

what they think is the right answer. I think based

on the NHLBI guidelines in terms of moderate and

severe persistent asthma, you could make a very

good case that if somebody is on a long-acting beta

254

agonist they should be on inhaled a corticosteroid.

So, if part of the goals of this meeting are to

disseminate information properly, then maybe that

is something that should be disseminated, and maybe

this would be a good network to do that in.

DR. KNOBIL: Yes, I think it is important

to note that even from the time that salmeterol has

been approved in the United States there has been

language in the label that mentions that early

consideration of anti-inflammatory therapy should

be considered. So, this is probably not a new

thing. Dr. Beasley?

DR. BEASLEY: I think we do have some

published data that can help answer your question,

and that is taken from the United Kingdom

case-control study where a proportion of asthmatics

in the control group who were taking inhaled

corticosteroids was almost the same as those taking

short-acting beta agonist drugs, suggesting that

almost all patients were taking the combination of

both an inhaled steroid and a short-acting beta

agonist drug. In that scenario, where management

255

is very close to the recommended guidelines, there

was no increased risk associated with long-acting

beta agonist therapy. So, I think that we can

deduce from a study that was of high

epidemiological quality, where management in the

control group was very close to what is recommended

in the guidelines in terms of inhaled steroid

therapy, that there is no risk observed in relation

to long-acting beta agonist therapy.

DR. MOSS: I just want to say one thing in

response to that. I just want to say that I don't

think this is the entire issue, that if everybody

was on an inhaled corticosteroid that the effect

would go away. But I think most people in this

room I think would say that that is part of the

issue, that people aren't being treated in what

would now be considered the proper manner. So, if

the goal here is to improve health of the people in

this country, and people on NIH panels and people

in this room feel that if someone is on a

long-acting beta agonist they should be on an

inhaled corticosteroid, then we should try to

256

disseminate that information.

DR. BEASLEY: There is one other bit of

data that may be relevant to the African Americans,

and that is that in New Zealand the Maori

indigenous community had a far higher rate of

mortality than the Caucasians, and it was about the

magnitude of about three-fold or more. With the

improvements in asthma management and in particular

the use of inhaled corticosteroid therapy and a

real emphasis on management within our community,

that difference has largely resolved, suggesting

that it is a reversible difference that is amenable

to improvements in management.

DR. SWENSON: Dr. Newman?

DR. NEWMAN: I would like to ask both of

the sponsors a question which I asked Dr. Sorkness

this morning but she didn't have an answer. Based

on what we know today, and I know there are other

studies going on but based on what we know today,

on the benefit side of these long-acting beta

agonists do we have reason to think that there are

any racial differences in terms of benefit? Do

257

African Americans benefit less from long-acting

beta agonists? I am trying here to weigh the issue

of benefit versus risk.

DR. GEBA: Greg Geba, Novartis. In terms

of that analysis in our own data set, we have seen

no difference in efficacy between Caucasians and

African Americans in terms of the endpoints that

were studies.

DR. NEWMAN: Were you powered to answer

the question do you think?

DR. GEBA: No. The representation of

African Americans in our trials was low. In the

most recent trial that we shared with you before,

the 2307 study, it was actually about 8 percent and

there was no difference across treatment groups in

terms of their response.

DR. NEWMAN: Just to make sure I

understand your answer, you didn't see a difference

but you didn't have adequate power to answer the

question?

DR. GEBA: No, it wasn't specifically

designed to look at that question.

258

DR. NEWMAN: So we don't know.

DR. GEBA: Right. In terms of our usage

of inhaled corticosteroids, it was about 70 percent

across all trials. There tended to be a slightly

lower incidence of any event in patients that were

taking inhaled corticosteroids, as is recommended

also in our label.

DR. KNOBIL: Similarly, we didn't have a

single trial that had enough African American

patients to reach any conclusions, but when we

pooled all of the data from African Americans there

were no differences in response between African

Americans and Caucasians.

DR. SWENSON: Dr. Kercsmar?

DR. KERCSMAR: I have another question for

GSK. If you could clarify some of your planned

prospective trials, particularly those that aim to

look at the influence of genotype or some of the

polymorphisms in the beta receptor, if you could

clarify what you hope to learn from those trials.

I believe one of the ones that you talked about is

using a combination product versus salmeterol

259

alone. The comment you made about difficulties in

enrolling patients, not wanting to be in a

placebo-controlled trial in view of the SMART data

that is out there, do you anticipate being able to

fill this trial with patients not wanting to go

into an arm that is salmeterol alone?

DR. KNOBIL: Well, I will answer the last

question first. It turns out that the genetic

trial that you have mentioned is actually a little

bit ahead of schedule so we are doing okay with

enrollment there. In this trial we are looking at

those two treatment groups, salmeterol versus the

combination, with each phenotype that has been

discussed today, the Arg/Arg, the Arg/Gly and the

Gly/Gly. I would ask Dr. Bleecker to stand up and

just tell us in a succinct way, I think better than

I could, about what we can glean from the study.

DR. BLEECKER: The trial that is ongoing

is a larger trial. There are six arms--and I could

be corrected, each with 90 individuals, and they

are not just looking at the two homozygote

genotypes at 16, the Arg/Arg, Arg/Gly and Gly/Gly,

260

but also looking at the heterozygotes. That is a

really important approach that has not been done

for the most part until now. The kind of genetic

effect, if this is the risk genotype, is that you

would expect or hypothesize some intermediate

effects from the heterozygotes.

Also, the sample size in each of the arms,

90 individuals in each of the arms, I think will

allow some better exploration of the gene and what

we would call haplotype analysis looking at a group

of snips across the gene to see if the effect is

either due to that whole haplotype or there is an

effect to variation, snips, polymorphisms and other

parts of the gene. So, I think from the point of

view of advancing the understanding of whether--and

this is still the question--there is a risk

genotype that may be associated either with varying

responses to drug or exacerbations, and being able

to make those correlations this is a very good

opportunity.

As I understand it as an outside

consultant, the African American study that is

261

looking at exacerbations over a year period will

also have the same kind of genetic analysis. So,

again, that provides an opportunity for the first

time in a large enough sample size to look at

relationships, pharmacogenetic relationships

between genotype and phenotype and response in

African Americans.

DR. SWENSON: Dr. Brantly?

DR. BRANTLY: This is a question to both

the sponsors. In thinking about mechanisms by

which long-acting beta agonists might have

increased morbidity and mortality, one sort of

thinks about possibilities like, for instance, that

these two drugs may be associated with increased

inflammation rather than decreased inflammation.

Sort of based on your experience with the

preclinical animal studies in the Phase 1 and Phase

2 studies, was there any indication that either of

these two drugs were pro-inflammatory, particularly

for instance in bronchial biopsy studies in the

Phase 1 and Phase 2 studies where there may be some

increased inflammation?

262

DR. GEBA: Greg Geba, from Novartis. I

would be happy to respond on Novartis side, if I

could ask Alex Trifilieff to come up to respond to

this question in terms of our preclinical studies.

DR. TRIFILIEFF If you look at all our

animal models that is something we look at because

beta2 agonists, especially upon acute exposure in

animal models, are anti-inflammatory. So, we

screen in different animal models and we always see

an anti-inflammatory effect. We never saw a

pro-inflammatory effect. These are the animal

model data for Phase 1 and Phase 2.

DR. DELLA-CIOPPA: We have conducted a

number of studies looking at anti-inflammatory

markers with formoterol a few years ago because

there was a rather big debate as to whether these

agents actually had anti-inflammatory activities,

and somehow they were used inappropriately based on

this assumption. Rather large studies have been

published, mainly by the group in South Hampton by

Prof. Holgate and in Sweden by Prof. Sundstrom, and

we could confirm in several severities that there

263

was no pro-inflammatory effect associated win

formoterol taken alone.

There were, indeed, some signals of a

reduction of some of the markers of inflammation

but probably not to the extent of being of clinical

relevance. But the few signals we did have went in

the opposite direction. These studies are

published.

DR. JOHNSON: I am Malcolm Johnson, from

GlaxoSmithKline. We had a similar experience in

the very early days of salmeterol development. If

anything, we saw some mild anti-inflammatory

effects in animal models. Clearly, they were not

predictive of what you might expect in man. I

think you asked the question this morning whether

there were any biopsy studies that have been

carried out to address this issue, and there are a

couple that I am aware of.

One study was conducted by Prof. Peter

Jeffrey at the National Heart Lung Institute, in

London. It was a 6-week study in mild

steroid-naive asthmatics in which patients wee

264

crossed over between either salmeterol monotherapy,

fluticasone propionate monotherapy or placebo for 6

weeks. Bronchial biopsies were taken and markers

of inflammation such as the numbers of eosinophils

and T-lymphocytes and neutrophils in the airway

tissue were assessed. I think the results were

very reassuring because in the salmeterol

monotherapy group there was no indication of a

pro-inflammatory response and, as I said, these

patients were steroid naive.

What was interesting in those studies is

that there was a signal, which is to say that there

was a reduction in neutrophils in the tissue with

salmeterol that was not seen with steroids. In the

steroid arm he saw a traditional anti-inflammatory

effect of steroids whereby eosinophils and T-cells

were reduced. So, the study was significant enough

to pick up an anti-inflammatory effect of the

steroid that we are very well accustomed to seeing.

There was no pro-inflammatory effect of salmeterol

and this effect on neutrophils in asthma, for

whatever that means.

265

DR. BRANTLY: Were any of the biopsy

studies done including blacks?

DR. JOHNSON: No, there were not. The

other thing that I think is interesting just to

quickly add is that there are some biopsy studies

that are being carried out that have looked at the

impact of adding salmeterol to inhaled steroids.

One study, again from Prof. Holgate's group, showed

that, in fact, in patients who were inadequately

controlled on low doses of steroids where ongoing

inflammation was not controlled, and that was shown

during the 3 months of the study, adding the

long-acting beta agonist salmeterol to that regimen

then controlled inflammation, and the control of

inflammation was equivalent to a much higher dose

of the steroid.

The final study, carried out in Australia,

actually looked at an index of airway remodeling.

They looked at angiogenesis in the airways and,

again, a long-acting beta agonist with a steroid

appeared to control the ongoing vascularization of

the tissue. Now, these are limited studies but, to

266

answer your question, in all of these together

there is no evidence of a pro-inflammatory effect

and there may be some emerging evidence of an

increased anti-inflammatory effect when salmeterol

is added to steroids.

DR. SWENSON: Dr. Gardner?

DR. GARDNER: I would like to go back to

the Medicaid study. It is rare that we have a

database that is able to address so many of our

questions. In the case of Medicaid, besides the

attributes you mentioned, we do have the

opportunity to look a bit at asthma management. In

fact, I have done that in some Medicaid data. So,

I would like to ask about what you have planned, in

terms of what you can see from the pharmacy data,

including adherence as measured by refill patterns.

Is that planned? If not, would you plan it?

DR. DAVIS: Yes, thanks for asking that.

Persistency of use is part of our protocol

currently. I neglected to tell you one more detail

about the study which may be of interest. Our

partner in conducting the study is Research

267

Triangle Institute so they are actually are linking

the data and doing the hands-on analysis. So, GSK

is sponsoring the study and is very actively

involved in writing the protocol as a

co-investigator but the lead investigators are

located at RTI. In addition, we have a clinical

advisory board which is also advising us on the

protocol and the clinical interpretation, and that

includes Dr. Beasley who is here today. It

includes Ann Fullbrighy[?], up at the Channing Lab

at Harvard, and it also includes Sheryl

Winwalker[?] who is in Atlanta, Georgia. So, we

have three outside clinicians who are also advising

us.

But to get back to your original question,

persistency of use is one of our variables of

interest, including the regular use of inhaled

steroids because we know from other observational

studies that intermittent use of inhaled

corticosteroids is often just a marker of severity,

not necessarily associated with improved outcomes.

DR. GARDNER: That was a good choice of

268

partner. Can you tell us again, is it 2006 when

you expect these analyses to be completed?

DR. DAVIS: That is right. The data from

5 of the 7 states have been obtained at this point

so we are waiting on 2 more states, and that

matching with the death certificates is slower in

some states than others, as you could imagine, so

once the additional data are received, and we are

hopeful that it will be in the next few months,

then the analytical portion will begin.

Also, just a final caveat, we must have

sufficient power to conduct the study or we will

not do a case-control study. The last thing we

would want to do is conduct another study which

would be underpowered and raise more questions

rather than answering them. So, we do have

criteria which must be met in order to continue the

race-specific analyses as well as the inhaled

steroid effect modification.

DR. GARDNER: One more thing, I can

certainly understand your position on that about

power, but I would encourage you, from the

269

standpoint of management, to nonetheless complete

descriptive analyses from those data because they

can be valuable in assisting many of the questions

that appear here. So, even if you don't have the

power to test a hypothesis, please do the

descriptive analyses.

DR. DAVIS: Thanks. We will.

DR. SWENSON: Dr. Prussin?

DR. PRUSSIN: I want to follow-up on Dr.

Moss' comments about salmeterol monotherapy. I

think the NIH guidelines support that. We have

heard today lots of data supporting the fact that

salmeterol monotherapy can associate with increased

exacerbation rates. Yet, when you look at the

package insert--and I know it is a little bit off

the mark and perhaps the FDA staff can tell me how

much off the mark I am--one of the issues we are

addressing is the package insert. And, if you look

at the labeling, it is "strongly advised" and I

guess I would ask the two companies here could that

labeling be made stronger, such as salmeterol

monotherapy--or whatever the long-acting beta

270

agonist--monotherapy should not be used as chronic

monotherapy in asthma, for example? Since everyone

is in agreement on this from what I have heard

today, yet the package labeling is a little softer

in terms of terminology.

DR. WEAN: I noticed Bob wasn't moving--

[Laughter]

--I think one of the issues around

labeling, and particularly around recommending

concomitant use--if I say it wrong, Bob, please

tell me--but there is a very fine line between your

label that addresses your clinical data that you

have shown in clinical studies and labeling that

takes the guise of recommending treatment, of being

treatment guidelines. I know that in discussions

we have had with the agency, while we may want to

put more information into the label, there is a

desire not to have the label be a substitute for

treatment guidelines being issued by NHLBI and

other appropriate groups of experts. So, that is

sort of the fine line that we have to walk between

strong recommendations about such concomitant use

271

but also making appropriate reflections in the

label.

DR. FLOYD: In follow-up, what we try to

do is base the label based upon the outcome of the

data that we have that is generated, but you must

maintain flexibility for physicians to be able to

use their discretion in appropriately prescribing

these drugs. So, we work with the agency to make

sure that the label is comparable to the

information that is generated based upon the data

from the Phase 3 studies.

DR. SWENSON: Miss Sander?

MS. SANDER: Yes, this is to both

sponsors. Because managing asthma or achieving

good asthma control is rarely just about taking one

drug and includes, you know, environmental control.

It includes receiving patient education that affect

your beliefs and, therefore, your behaviors and

your outcomes. I am wondering in these studies do

all patients receive the same type of education and

advice, or is it just related to the study drug?

DR. DELLA CIOPPA: You are touching upon a

272

very important point and the short answer is no.

We make a huge effort to do so, but please keep in

mind that all of these studies are in many states

in the United States but most of these studies are

in many countries, and some very diverse countries,

and the bigger the studies are and the more rare

the event we are looking for, the more countries we

go to. It is not rare that we go to 17, 20, 25

countries, not to speak of all the possible states

ion the United States. So, we do make a huge

effort to try to harmonize as much as we can the

instruction to patients, the ancillary activities

that the patients should put into place to manage

their asthma in an appropriate way. But the

cultural differences, the background differences,

the socioeconomic status differences will stay

there so that is, indeed, a considerable source of

variability in the results of the study.

On the other hand, should you go to only

one place with perfect harmonization of the data,

then another problem would occur, how could you

extrapolate? How could you extend your results to

273

other states, other countries or other cultural

situations? So, it is a balance.

MS. SANDER: So, even in one

investigator's practice would all the patients

receive the same information?

DR. DELLA-CIOPPA: Yes, they would.

MS. SANDER: They would?

DR. DELLA-CIOPPA: They receive the same

information across the study, but the way this

information is delivered may change. Within one

center we are reasonably sure that they get the

same information in the same way.

MS. SANDER: Right. Then, I guess this is

for GSK regarding the phone calls. When you did

the follow-up phone calls with the patients were

there any clues in the answers that they gave you

that some of them were struggling more than others

and were at greater risk, and was there any kind of

discussion with the patients about that? Was it

interactive or was it just a survey each time?

DR. KNOBIL: You mean, was it a real

person making the telephone calls?

274

MS. SANDER: No, no, no. The phone calls

that occurred with the patients, and you said that

you asked them, or that they were asked questions

about the use of the study drug. Am I

misunderstanding?

DR. KNOBIL: Well, what happened was a

real person did have a script as to what questions

should be asked. There were specific questions and

every person got the same questions. Now, I guess

what you are asking is if the patient said, well, I

don't know or something like that the person on the

phone would try to clarify as much as possible.

But also remember that the person on the phone was

not a physician; it is a telephone center. So,

they could only clarify what they could and they

might not pick up on every clue that someone who is

familiar with asthma might pick up on. But they

did have specific questions to ask and they got as

many answers to those questions as they possibly

could.

MS. SANDER: So, there was no trigger.

You know, if a patient had these types of answers,

275

then this patient is at greater risk and the

investigator needed to be contacted?

DR. KNOBIL: I see what you are saying,

no, the information was collected but there were no

recommendations made to the patient on what they

should do. You know, if the patient was telling

the operator that they were having problems, then

the operator would tell them to go see their

physician but there was nothing else within the

study.

DR. SWENSON: Dr. Moss?

DR. MOSS: I am going to follow-up on a

question I asked earlier to industry but I want to

ask it to the FDA. So, you can sit down and take a

little breather. Some of the questions to our

committee deal with product labels and boxed

warnings. I just wanted to make sure I understood

the role of boxed warnings and product labels. Do

you have any evidence that the product labels and

boxed warnings meet the goals that you want them

to? So, if we recommend that we are going to have

some change in a product label, as Dr. Prussin was

276

talking about, or a boxed warning, does that really

achieve what you want it to? Do you have evidence

that people change practice based on that

information? If not, then maybe that is not the

right medium to disseminate information.

DR. TRONTELL: Anne Trontell, from FDA.

There hasn't been a systematic study of the impact

of black box warnings. Some have looked at changes

in utilization measured by numbers of prescriptions

but, again, that is an imperfect surrogate for the

appropriateness of use. I believe there are some

studies forthcoming but we are not yet privy to

those results.

DR. MOSS: Getting back to what Dr.

Prussin was talking about, can you just explain a

little bit in more depth what is the goal of a

product label or a boxed warning?

DR. MEYER: That is sort of two separate

questions. The goal of the product label is to

describe the substantial evidence that led to the

FDA's decision on the approval of the drug. The

standard for approval is that a drug is safe and

277

effective for use as described in the product

labeling. So, it is to take that substantial

evidence and to put it into a format that allows

that drug to be used in a safe and effective manner

based on what we have found.

I agree with the comments made earlier

about the fact that that separates it out somewhat

from what may be very informed opinion but

opinion-based guidelines, for instance, and the

labeling is not meant to strictly restrict the

practice of medicine. The FDA is not in the

business of restricting the practice of medicine,

but it is to inform the practice of medicine.

As far as a boxed warning goes, it is

really to describe situations of serious morbidity

or mortality. The standard for putting in a boxed

warning does not require certainty about the

relationship of the drug to those outcomes. It is

really driven much more by the outcome itself. If

you are taking about death, and so on, and you have

at least a reasonable suspicion--either from animal

data or from human data be it from studies or be it

278

from postmarketing--that the drug may be

associated, even if not certainly causally with a

particularly bad outcome, that will often be enough

to place a boxed warning into the label. That is

intended to really raise to the forefront how

serious the concern is. There are some subtleties

in terms of changing how the drug can be marketed,

and so on, but the intent is really to signal right

up front to anybody using that drug that this is

something you need to consider when you use that

drug.

DR. SWENSON: Dr. Newman?

DR. NEWMAN: Just to follow on that point,

we are being asked this afternoon what we would

recommend to the agency in terms of actions for you

to take to communicate. What are the other things

in your repertoire, in addition to the things that

have been described here, when you want to

communicate?

DR. TRONTELL: Again, depending on how

broadly you look at it, there is a number of

so-called risk minimization interventions that the

279

agency has worked with sponsors to apply. In the

communication arena the agency may speak to the

public through a public health advisory, or some

other press release or talk paper. So, there are

ways we can make a more prominent announcement of

concerns.

There can be additional requirements for

patients to be specifically informed of risks,

using patient labeling that is specifically

oriented to the lay person in terms of

understanding. So, that includes such written

materials as medication guides which are required

to be handed out with prescriptions, or a patient

package insert. There is a variety of other

educational materials that could go beyond the

print media that could be considered.

Then, obviously if you wanted to get

beyond the communication realm such as we have

heard--"dear healthcare practitioner" letters and

others--you start to talk about other ways of

making the information prominent or actually trying

to direct care in a specific way.

280

DR. SWENSON: Dr. Martinez?

DR. MARTINEZ: I would like to follow on

Dr. Brantly's question before in relation to the

potential for some of the observations that have

been made to be associated with what he generically

called changes in inflammation associated with the

use of beta agonists.

Just a brief introduction, I think the

issue we are talking about here is not a global

effect, as was said before, of these medicines on

individuals because the effects that we are seeing

are seen in a very small number of individuals. I

think relevant to this issue are the data that I

would like some of the basic scientists from both

companies to comment about, presented by Steve

Liggett who has been one of the persons most

assiduously working in this area, in which

over-expression of beat adrenergic receptors in

mice, as compared to under-expression of beta

adrenergic receptors in mice, was associated with a

very paradoxical effect, contrary to what they were

really expecting. And, this is published in The

281

Journal of Clinical Investigation in August, 2003.

Contrary to their expectations, the mice

in which a beta adrenergic receptor was

over-expressed showed an increased expression of

bronchoconstrictive receptors in airway smooth

muscle, and those in which the beta adrenergic

receptor was under-expressed showed a decrease in

the expression of these bronchoconstrictive

receptors for thromboxane, for histamine, and so

forth and so on.

So, a possibility then cannot be ruled

out, and should be seriously considered as a

potential explanation for some of these effects,

that there could be individuals out there whose

genetics or phenotype of some sort that we haven't

yet figured out could be associated with them

becoming a little bit like the airways of these

mice in which these beta adrenergic receptors are

under- or over-expressed.

They didn't test in this study if

corticosteroids affected this particular under- or

over-expression so the issue of the potential

282

regulation by corticosteroids was not followed up.

So, just because this issue had been raised by Dr.

Brantly, I just wanted to know the opinion whether

the possibility could be that phenotypic and

genetic characteristics of a very small number of

subjects could make them particularly susceptible

to effects similar to those described by Dr.

Liggett with respect to beta adrenergic receptors

in the mice.

DR. BEASLEY: Yes, you raise an

interesting point. I think what Steve Liggett's

experiments were attempting to address really was

he was looking at the balance between sympathetic

and parasympathetic pathways. I think the

prediction was if you over-express beta receptors

you would see a corresponding decrease in

muscarinic receptor function. In fact, what he saw

was the opposite. In fact, there was a

recompensatory rebound in muscarinic receptor

function and, similarly, when he knocked out the

beta receptor he got a decrease in muscarinic

receptor. So, I think it underlines that, you

283

know, we know really little about how the neuronal

pathways interact. Clearly, there is more work

that needs to be done there.

I do have a slight concern in trying to

make extrapolations from animal models that require

over-expression or knockouts to what we can see in

the clinical scenario, and I think it is probably

shared by everybody. But you do raise a very

interesting question that possibly in a subtype of

patients, in a small minority, there may be an

inappropriate balance between the two pathways, and

were you to influence one pathway you might get a

reaction from the other pathway. But, as I say,

the problem is that you could model these systems

in transfected cells. You can do them in

over-expressing or knockout animal models but that

is really all they are, they are animal models and

making that extrapolation to man is obviously very

difficult. And, we clearly do not yet have those

sorts of studies, but it is a very interesting

point.

DR. MARTINEZ: I was just raising it as a

284

possible explanation but certainly I am not saying

that this is what may be going on. It is just a

potential for explaining that in a very small

minority of patients genetic and phenotypic

characteristics may make this balance, not only

with muscarinic receptors. He also studies

thromboxane receptors and histamine receptors--

DR. BEASLEY: Yes.

DR. MARTINEZ: --maybe this balance is

altered in ways that would not be expected for the

great majority of the population.

DR. BEASLEY: I agree, and I think today's

discussion about genotype implications has largely

focused on polymorphisms of the beta receptor. Of

course, we shouldn't forget there are also

polymorphisms of the muscarinic receptor, of the

glucocorticoid receptor, and when we are dealing

with the body and we are dealing with patient

response to drugs and maybe the patient's response

to disease we are dealing with an integration of

all these gentoypic polymorphisms. It is very

difficult then to predict what you would see in

285

small proportions of patients.

DR. SWENSON: I would like to ask the

Novartis personnel a question regarding the racemic

nature on formoterol. I read that I believe it is

the RR enantiomer that is the effective moiety is

not at all blocked by the SS. But that is simply

on bronchoconstrictor aspects. Have you any data

as to whether the inactive enantiomer might have

pro-inflammatory effects or some other effect that

wasn't gauged in studies that I could see? This

might bear on the problem with the danger signal

evident in the larger doses that some studies from

you have shown.

DR. TRIFILIEFF: I do not have any data to

show you today but we did look at this possible

pro-inflammatory or antagonism effect of the

inactive enantiomer for formoterol. As you said,

there was recently a clinical paper looking at the

effect of bronchorelaxation in human comparing the

different types of enantiomer for formoterol and

there was basically no antagonism effect of the

inactive enantiomer.

286

To come back to the preclinical situation,

we had basically the same situation. If we look in

our animal models or in vitro, we don't see any

antagonism activity of the inactive enantiomer.

DR. SWENSON: At this point then we will

take our break and the remainder of the meeting,

when we return in 15 minutes, will be focused on

the specific questions that the FDA wishes us to

address.

[Brief recess]

DR. SWENSON: Well, we now move to the

specific questions posed by the FDA to the panel

around warning and the potential use and studies

necessary for these two drugs that we have been

discussing today. We will go through the questions

in order that the panel already has. That is, we

will jump from one drug to the other rather than

keep all questions to one drug and then move to the

next. We will be moving back and forth with these

questions. Some of these will be, as you see,

recommendations that each of you will be able to

offer, as we go around the table, to the questions

287

that are asked. Then we will move to yes/no

questions in regards to whether we feel that these

drugs should be continued to be marketed with the

present database that we have.

I would ask the panel members to make

their yes/no votes on these questions on the basis

of the information that we have presently, that we

have heard today. Although there are significant

studies in the pipeline that may well answer these,

and both companies are making very strong efforts

in this regard I believe, but your yes/no vote

should not be based on the fact that any of those

studies will be accomplished and will provide

further answers. We need to vote yes/no on what we

have today.

So, with that, let first just see if there

are any questions or general comments by the panel.

Dr. Schoenfeld?

DR. SCHOENFELD: I have a general comment.

I made a calculation before that was wrong, which

is one of the reasons I asked for somebody to come

by and do a calculation with a computer because I

288

know how often mistakes are made, and I have made

plenty of them myself and that is why we usually

have people repeat important analyses. So, I

miscalculated the risk based on the SMART study,

and I just want to give you my current calculation,

and I think that somebody else out there ought to

check this calculation and stop me if it is wrong

because I think it is a very relevant calculation.

That is, I calculate that the attributable

risk for the treatment among the general population

is roughly 1/700, and this is in regard to asthma

deaths. That is, be an extra 1 patient in 700

would suffer an asthma death on the basis of

treatment with this drug in the general population.

And the figure among African Americans is roughly

1/200.

Now, I am enough of a Bayesian that when I

see 1/700 in the general population and 1/200 in a

subgroup that was sort of chosen out of the study

to move the 1/200 towards the 1/700. Because it

sort of caught our attention and maybe we should

focus more on the average than on what happens in

289

subgroups.

That being said, that is the kind of risk

that, if we take these data on their face, we are

facing. What I can't really judge as a

statistician is the benefit because I don't talk to

asthma patients every day; I don't know how hard it

is for them. I have no real way of knowing whether

if I was an asthma patient with a risk of 1/700 I

would take or not take. For instance, if someone

my age has a risk of 1/300 of dying from natural

causes or from all causes--something like that;

maybe 1/200. I keep track of this but I haven't

looked at it recently--

[Laughter]

--yes, I have gotten older! So, I don't

really know whether this is a risk that is too

large or not very important at all. Probably some

of the panel members who actually treat asthma

patients would know that and it may vary from

patient to patient.

DR. SWENSON: Dr. Meyer, I might ask if

you or any member of the agency want to just

290

comment, even with very little preparation, on this

question.

DR. MEYER: No, I think I would prefer not

to. I haven't seen the calculations and I think at

this point I prefer not to.

DR. SCHOENFELD: I could go over the

calculation if someone wants me to just say how I

did it. Basically, I took 10. Okay, I took

basically the data here which was 13,176 and that

is a 28-week study. That was the mistake I made

before. I multiplied by that 28 and divided by 52.

So, that is roughly half that number I guess you

would get roughly, which is roughly 7,000. Then,

basically, I divide 10 into that and I get 700.

That is how I got the 1/700. Now, how I made a

mistake doing that the first time I don't know. It

is 10 extra deaths. There were 10 extra deaths

total, 10 extra asthma deaths. I think it was 3

versus 10 I think in the data.

DR. MARTINEZ: the question I was asking

is it is 1/700 per year?

DR. SCHOENFELD: Per year of exposure. Of

291

course, every year that I am exposed I take this

extra risk of 1/700 if I want to take the SMART

study basically at face value. I guess one other

comment is that it is very hard in a large, simple

trial to know what causes a mortality difference or

a difference in any endpoint. The whole idea of a

large, simple trial is that you don't really

control what happens to the patients carefully so

you don't really know whether it is the effect of

the drug or an effect of the effect of the drug.

For instance, if a drug makes people feel better so

they go their doctor less often they may have

higher mortality. In a very carefully controlled

trial, like the Phase 3 trials that I am sure were

done you are controlling all these things. So,

there is less chance that it is some supportive

care that is being affected by the effect of the

drug that is causing the difference. But in a

large, simple trial you lose that and some people

consider that the advantages and some people

consider that the disadvantages of a large, simple

trial.

292

DR. WEAN: If I might just very

quickly--David Wean from GSK--respond to that,

because I just find I can't allow a figure such as

1/700 to be put out onto the table and potentially

be appearing in the press and public domain without

appropriate caveats. We, quite honestly, can't

fathom a cogent response to that calculation but

what we do need to say is when you look at

epidemiological data, which we presented to you, we

don't see an attributable risk of that sort. We

can also say that in trying to assess the issues

that we are asking about today in a large database

of over 25,000 patients, we had difficulty arriving

at the intended rate for asthma- and

respiratory-related death. So, I think that

calculation, quite honestly, probably does more

disservice to individuals that we are wrestling

with appropriately advising around the use of these

drugs for their asthma, and I just want to add that

balance to that back-of-the-envelope calculation.

DR. SWENSON: Dr. Prussin, did you have a

comment?

293

DR. PRUSSIN: This morning there was some

discussion about ongoing clinical trials by GSK

that we heard from their representatives, and I

just wondered if Dr. Schoenfeld might speak a bit,

since he is an epidemiologist, on what he sees as a

relevant clinical trial to address the question of

increased mortality.

DR. KAMMERMAN: Before we get to that, I

am a statistician. I am Lisa Kammerman. I am a

statistical reviewer. I think just to clarify a

little bit what Dr. Schoenfeld was doing, instead

of looking at relative risk he is a little bit more

interested in looking at the difference in the

rates, the proportions of people developing

asthma-related deaths, which is where he got the 10

from.

But in calculating any number where you

are using person years as the denominator there are

always issues involved. One of the major

assumptions is that there is a constant risk for

asthma-related death over time. So, that needs to

be cautioned also.

294

DR. SWENSON: Dr. Schoenfeld, do you want

to reply to Dr. Prussin's question?

DR. SCHOENFELD: I guess not right now

because I think that the issue of what clinical

trials should be done, or what further we should

ask these companies for is sort of a later issue.

The other thing, again, I am asking the

asthma specialists here I don't know whether this

is a large--I calculate that number. Hopefully, it

is right, but I don't have the expertise to know

whether this is a big number or a small number in

regards to the benefit of these treatments. I

mean, there are many situations in which this would

be a very small number and there are others where

it would be a large number.

DR. SWENSON: Dr. Schatz?

DR. SCHATZ: I was just going to say I

think it is also complicated because one has to ask

how representative that denominator is to the type

of patients I treat, you treat, or are out there

and I don't think we know that either. So, it is a

very complicated number.

295

DR. SWENSON: Dr. Martinez?

DR. MARTINEZ: Just a point of

clarification, what is it that exactly we are

discussing? Are we opening the discussion very

generally or are we answering any of the specific

questions? I thought that we were very generally

discussing and that is what I would like to

intervene upon.

DR. SWENSON: The plan was to move to the

specific questions but in this time period there

will be the chance for specific recommendations for

each of you, if you wish to offer those. Dr.

Meyer?

DR. MEYER: I am sorry, I just wanted to

take the opportunity to respond also to the 1/700

figure. I think our hope with the SMART study was

to try to get data that would either confirm or

refute the signal that we had coming out of the SNS

study and some of the postmarketing experience in

the early years of Serevent being marketed.

I think that the SMART study did go some

distance in terms of helping to answer that. I

296

don't think it provided nearly the kind of

precision as to what that risk might be that would

allow for us to look at a number for attributable

risk with any kind of confidence that that

represents a true number. You know, these are rare

events and this was a big study but it still

didn't, I don't think, give us a kind of precision

around what the true difference would be if we were

to take that out to the entire population. So, I

just wanted to caveat that number a little bit from

our perspective as well.

DR. LITTLE: This is Roger Little, GSK. I

am Vice President of Biostatistics for GSK. I

agree with that very much. I like the idea of

going after absolute risk. I think that is a great

goal but I think if you think about the way we

approach the study, we are trying to be

conservative. We have the car accident where that

is potentially related to asthma. We have been

trying very hard to go after these in a very

conservative way. If we want to estimate the

absolute risk I think you would look to other kinds

297

of studies. You would look to the epidemiological

study. If there was a signal nearly as strong as

the one that has been mentioned we would see that

in many other places. So, I don't think this is

the type of study to really address that. This

isn't the primary efficacy endpoint. We have

picked out the one that caused the greatest concern

perhaps, which is appropriate, in terms of looking

at this study and thinking about the risk but I

don't think this is the study to support the

absolute risk and the difference between these two

treatment groups. Thank you.

DR. SWENSON: I think we won't have any

resolution today or even in the very near future,

particularly with issues about the imprecision of

socioeconomic status and questions about the

biology of various forms of asthma. I think we

will have to leave it as still unresolved.

I would like now to move to the specific

questions. The first question is on the screen

here. We are being asked now to provide specific

recommendations to the FDA as to any further

298

actions they should take to communicate or

otherwise manage the risk of severe asthma

exacerbations seen in the SMART study. I will go

down our list in order and ask Dr. Schoenfeld to

offer any recommendations you feel strongly at this

point on this question.

DR. SCHOENFELD: I don't have any specific

recommendations there. I thought the boxed warning

seemed to--although I haven't examined it in gory

detail, it seemed like a reasonable warning to

include in the labeling. I am assuming that that

then gets communicated.

DR. SWENSON: Miss Sander?

MS. SANDER: I would have to agree.

DR. SWENSON: Dr. Gardner?

DR. GARDNER: I think this might lend

itself to a medication guide or other kind of

direct patient information that is dispensed with

the medication because it is an opportunity to

educate people about the ancillary issues, and also

to assist people in knowing what are the drivers

that may signal exacerbation or other problems that

299

would get them earlier perhaps to care. So, I

think that I would recommend some form, either in a

med. guide or patient package insert, of direct

communication to the patient about these types of

risks that includes recommendations for what might

be done to minimize them for individuals.

DR. SWENSON: Dr. Schatz?

DR. SCHATZ: I also don't at this point

that there is any specific change to recommend, but

part of that is based on not having good outcome

data as to how these communications work. So, I

would encourage answers to some of the questions

that were brought up earlier, which is to try to

understand what outcome occurs with a black box

warning. But without any of that information and

assuming that, at least on a face value, that seems

to be an appropriate way to try to educate people

who have to prescribe the medicines I wouldn't

recommend any changes right now.

DR. SWENSON: Dr. Gay?

DR. NEWMAN: Excuse me, could I just ask a

point of clarification before we go further? I am

300

sorry to interrupt.

DR. SWENSON: That is fine.

DR. NEWMAN: What warning label are we

looking at? You know, on page 34 of the materials

in that first section on the SMART study that we

got from the FDA is what looks to me like a

proposed modification of a warning label. Is that

what we are commenting on or is that what it is at

this point? Have all those deletions already been

made?

DR. SEYMOUR: There is a copy of the

product labels under separate tabs in the back that

are the current product labels. The review may

contain different language as the changes in the

label progressed throughout the years but the

current labels are included in both briefing books

at separate tabs.

DR. NEWMAN: So, I guess my question is

there is a proposed warning label change that comes

from the FDA that is in this packet. Are you

asking for comment on that?

DR. SEYMOUR: No, that has been resolved.

301

We are asking for comment on the current product

label that is in the back of the briefing books

under separate tabs.

DR. NEWMAN: Thank you.

DR. SWENSON: Dr. Gay?

DR. GAY: I would recommend a few changes.

First, I believe that we should put greater

emphasis not only on the fact that there does seem

to be a difference between ethnic populations, but

also that there does seem to be some early

difference with greater severity of disease in

asthma for those patients with peak expiratory

flows less than 60 percent predicted. So, there

should be some attempt to make a warning as well

for patients as they have greater severity of

asthma.

In addition, we have debated to some

extent the role of this package insert. If it is

to clearly attempt to make the drug as safe as

possible, the data is not quite as strong because

from the SMART data we clearly don't have enough

information about the use of inhaled

302

corticosteroids in that population. But much of

the other data presented to us for combination

therapy would seem to suggest to me that we should

in some way change the wording from "the use of an

inhaled corticosteroid should be considered" to

something more along the lines that the use of an

inhaled corticosteroid should be strongly

recommended, or the use of long-acting beta

agonists as mono or individual therapy in patients

with asthma should be discouraged and should

require the use of some type of anti-inflammatory

medication.

DR. SWENSON: Dr. Moss?

DR. MOSS: I think I am going to end up

reiterating what a few other people said, but I

have four comments about the warnings. Number one

is that I think the warning box should be left on

the salmeterol. I think it is important that it

should also be kept on the Advair since that

compound is also in Advair.

I think it is important not to stress a

race thing, a race angle, but to state that there

303

are subpopulations that may be at increased risk of

adverse events from this medication because I think

that gets back to the beta receptor issues and

probably other snips that we are just not aware of

yet.

As Dr. Gay said, I think it is important

to stress the role of inhaled corticosteroids,

reiterating what I said earlier, I don't think that

is the whole effect but, getting back to what Dr.

Schoenfeld said, if people are treated properly it

might make that attributable risk more favorable or

better by dropping the numbers in both groups down

from 13 to 3 to maybe 6 and 2 or something. So, I

think it important that people realize, as Dr. Gay

stated, that this medication should be used in

conjunction with inhaled corticosteroids.

Again, I think the FDA needs to re-thing

about how they are going to disseminate information

to physicians and the public. I am not sure

changing product labels that I don't think a lot of

people spend a lot of time reading is an effective

way of communicating information to the public, to

304

physicians and to the medical community.

DR. SWENSON: Dr. Newman?

DR. NEWMAN: I really have nothing more to

add beyond what Drs. Gay and Moss just said. I

completely agree with that. I would also agree

with the idea of considering supplemental

information for patients. This gets very

complicated very quickly. I do believe that our

patients do a good job of understanding the limits

of medical knowledge if we explain it clearly

enough and we tell them what we know and what we

don't know. I think a more direct reach in that

way to patients would be something worth

considering.

DR. SWENSON: Dr. Brantly?

DR. BRANTLY: I agree with my colleagues.

I would like to make a point that I believe is

floating around in the community--and my hope is

that it will be picked up after this meeting--the

impression that has been floating around in the

community that these drugs, and specifically

Serevent, are bad for African Americans, and I

305

think it is critical that that message not go

forward because it is likely that it will be far

more harmful to the African American community if

they avoid these types of drugs in the future. I

think that is a real message that really needs to

be considered strongly. I absolutely agree with

the agency taking that out of the black box warning

area.

DR. SWENSON: Dr. Martinez?

DR. MARTINEZ: Notwithstanding my

agreement with what my colleagues have said, I

would like to add just one more point that I

referred to before, and I think it is an issue that

is coming up I think very clearly from the data

that we have observed. I will briefly repeat what

I said before with respect to this issue.

I think there may be two dimensions to

asthma morbidity that until now were considered as

highly correlated within individuals. In other

words, asthma control, meaning the everyday

presentation of symptoms, cough at night, wheezing,

wheezing with exercise, and the likelihood of

306

having severe asthma exacerbations. In fact, I

think that it is almost implicit in the guidelines,

as they are stated now, that if you adequately

consider how much a person is controlled you in

some way, because of this implicit supposed

correlation between this and exacerbations, you are

also considering exacerbations, so much so that

exacerbations are not part of the algorithm to

determine asthma severity today, at least in the

American guidelines.

I think that what is emerging from this

data that we are observing here, and not only from

this data but from several other data that I will

not mention here, is that these two dimensions,

although correlated, are not equivalent; that there

are individuals in the community whose control with

these medications, particularly with combined

medication, is significantly improved as compared

to treatment with only inhaled corticosteroids or

with no medicine or with beta adrenergic agonists

of short duration by themselves. However, there

are individuals in whom either there is control but

307

there is still a very high risk for severe asthma

attacks, or who don't have problems of control, who

have something called brittle asthma, but are at

very high risk of developing severe asthma

exacerbations.

All that I see without yet, I agree,

definitive proof seems to indicate to me that

long-acting beta agonists as a group may have a

negative effect on the control of severe asthma

symptoms in the latter group, in what I have called

the brittle asthma group. I don't have definitive

proof, as nobody else here has, but I think the

data is clearly indicating that this is what may be

going on, and the scientific community, the FDA,

and particularly industry needs to be very worried

about this because we are going to discuss next

what is it that we need to do next with this class

of medicines. And, we would not like, I think, not

to pay attention to this possibility which could

make it difficult for these medicines to continue

to be used in the population as a whole because of

a small group that is at high risk when they

308

provide significant relief of symptoms to a very

large part of the population.

So, this idea that is relatively new

because it is not there in the guidelines I think

needs to be seriously considered in evaluating

results of any clinical trials in the future. To

summarize, there may be a population of subjects

with asthma in whom the main issue is not

day-to-day control of symptoms for which

combination therapy is the best we have today, but

in whom the main expression of the disease is

severe attacks that are not only not controlled by

these medications but may be rendered worse by

these medications.

What are the biological bases for this? I

propose one, which is the data that Steve Liggett

had proposed but there may be others. Careful

consideration to this possibility perhaps would

allow us to understand better the results we are

seeing.

DR. SWENSON: Dr. Kercsmar?

DR. KERCSMAR: I agree with most of the

309

statements that everyone has made already.

Management of asthma is an incredibly complex

problem that I think we in general tend to

underestimate. Today we have incomplete knowledge

of asthma phenotypes, as I think Fernando is

alluding to, and even less understanding of the

genetic basis of asthma, which makes the optimal

management even more difficult. Until we have

those data it makes giving a definitive answer on

what we should do with each specific class of

medication difficult, if not impossible.

So, I am not sure that I can recommend any

specific changes in the way to communicate,

although I do like the idea of perhaps reiterating

these messages to both the medical and the patient

community. The only caution I would have, which

was alluded to, is that virtually every adherence

or compliance study on asthma, particularly with

the use of inhaled corticosteroids, would indicate

that patients under-adhere to those medications and

any message that we send that might be alarmist or

hinder the use of what are appropriate medications

310

for the majority of the population would be a

disservice to that community, while we try to

understand for whom certain medications are a risk

factor.

DR. SWENSON: Miss Schell?

MS. SCHELL: I agree with everyone has

said but I would just like to add that I think it

is very important that the process include

education not only to the caregiver but to the

patient, and many of the patients that I work with

at bedside haven't an idea what the medicine is,

let alone if they read the label about the warning.

So, I think the dilemma for me is how to get this

information to them that they clearly deserve, but

to put it in a message that they can understand.

So, when you put it into fine print into the

insert, most of them can't even read it, and if you

put it on the box with the label, are they going to

be educated by the caregiver? I would like a clear

understanding of the education available to the

patient when it is dispersed to them, with the risk

involved, also what the medication is for, in a

311

language they understand. A lot of patients just

don't understand what they read when it gets into

the complications of the risk. So, that is my

dilemma when I look at it, how do we get that

information to the people that deserve it in a

clear message?

DR. SWENSON: Dr. Prussin

DR. PRUSSIN: I would agree with Dr. Gay's

comments on trying to strengthen the language on

monotherapy, discouraging its use. This drug is

already primarily being used in moderate to severe

asthmatics, and those are the groups that we talked

about being at risk. So, I am not sure if trying

to point out about subpopulations in the package

label is going to go much beyond what populations

are already being treated, in practical terms.

Lastly, I was struck primarily by the

original text, by a whole block of text. Basically

your eyes glaze over and all the numbers fade out.

You might consider either putting a table or one

single figure to try to graphically display the

relative risk data that we have been talking about.

312

I think somebody seeing a graph or a table, they

are going to gravitate to it much more than to a

text box. So, you may want to consider that. I

was originally thinking of that when I saw these

huge paragraphs of text, but maybe less so with

what you have now.

DR. SWENSON: And my comments echo much of

what has been said but I would like to state them

very briefly, and that is that the FDA really

consider a much stronger sanction of long-acting

beta agonist use in combination with inhaled

corticosteroids, and that monotherapy be highly

discouraged. I don't know whether this could be

done in some consensus with the respiratory

organizations and the agencies that have promoted

these guidelines, but to have possibly a

convergence of guidelines that are out there by

highly respected bodies match exactly what is in

the warning.

To the question of whether there is a

very, very small subset of patients that might be

adversely affected, i.e., these people with very

313

brittle asthma, consideration being given to

warnings that might suggest that patients that have

had very rapid onset of asthma leading to

respiratory arrest or need for intubation without

warning, that this class of drugs may be

potentially considered adverse. That is obviously

something that is going to have to evolve but at

least for agency consideration as maybe these

warnings change over time.

At this point we will move on to the

second part of question one, which is the vote.

Again, I will reiterate that your vote should be

based on what we presently know and not on what we

might know three or four years down the road with

many of the studies that, hopefully, will go to

completion and provide answers. I will start in

just the opposite order and ask--

MS. SANDER: Excuse me--

DR. SWENSON: Question, Miss Sander?

MS. SANDER: Excuse me. You know before

we go to the vote, if I might just add a little bit

from a patient perspective, I think some really

314

wonderful insights have been shared around the

table. If you are considering any types of changes

in any of the labeling, I would just encourage you

that when you are communicating with patients or

physicians we have to be careful what we say, and

how we say it, and when we say it because it does

have impact on what patients will wind up doing

about their disease with regard to this drug. For

example, we know from within our organization that

patients will often use Serevent to treat acute

symptoms because they don't see themselves in

rescue situations, warranting a rescue use of

albuterol. And the word "rescue" has a certain

meaning to them that is very different than what we

have talked about today around this table. So, you

know, while I think it is good for us to think

about what we are telling patients, it would be

nice to have a guide. It is also very important

for us to be careful about how we communicate. I

just wanted to echo that statement.

DR. SWENSON: Okay, if there are no

further comments, Dr. Prussin, your vote?

315

DR. PRUSSIN: Yes, should be continued.

DR. SWENSON: Miss Schell?

MS. SCHELL: Yes.

DR. SWENSON: Dr. Kercsmar?

DR. KERCSMAR: Yes.

DR. SWENSON: Dr. Martinez?

DR. MARTINEZ: Yes. May I justify my

yes--

DR. SWENSON: Certainly.

DR. MARTINEZ: --or is it only yes or no?

DR. SWENSON: No, be brief but you may.

DR. MARTINEZ: I will try to be as brief

as possible. I think the agency and industry--and

I thank them both--have presented very interesting

and important information. I would like to

summarize the way I see that information. We have

been presented with two large what I would call

surveillance studies, both coinciding with the same

type of result, which would indicate and increased

risk of asthma-related deaths in individuals who

are being treated with salmeterol without regards

to what other medicines they are receiving.

316

Dr. Beasley has several times mentioned an

epidemiologic study which I think is the strongest

study which shows that no such effect exists in a

type of different surveillance study, in which all

individuals who died with asthma are compared with

matched controls. In this case, controls in that

study were matched for the hospital in which the

subject had died or had been treated and for age.

I would like to warn, however, that that

study is completely different not only in

methodology but also in many other aspects to the

ones that were presented and are prospective

studies. Subjects in that study were much older;

42 percent of them had a specific diagnosis of

COPD. And, I would suppose that, for example, in

the SMART study any SMART doctor would not have

included subjects with COPD because what they were

asked to include were subjects with asthma. So, we

may be talking about two different things. When

you have 42 percent of subjects with COPD in one

study and perhaps not as many in the other study,

perhaps the results could be interpreted

317

differently.

I think we may have here a true signal.

As Dr. Schoenfeld has many times told us, here the

only way in which we can truly assess the signal of

risk is in terms of benefit. My evaluation today

is with respect to the patients that I see in my

clinic and the patients out there with asthma in

this country in general. It is still justified to

keep this medicine in the market. However, I say

that with a conditional, which is that there has to

be very, very accurate follow-up of the increased

risk that has been observed in these patients on

salmeterol in the future, and particularly better

understanding needs to be there if this risk is or

is not decreased by steroids. I do not think that

the data, as I see it today, justifies saying that

this risk is decreased by steroids. I am not

saying that it is not decreased by steroids. There

is no clear data to say either thing.

I am worried that the concept may get out

there that there is strong data, suggesting that if

you just give steroids this effect is not going to

318

be there when we don't have strong and definitive

data in that sense either. So, in that sense I

vote yes but with the strong conditional that a

very accurate follow-up for this issue needs to be

part of the FDA task in the future in collaboration

with industry.

DR. SWENSON: And I think most members

would agree with that. Dr. Brantly?

DR. BRANTLY: Yes.

DR. SWENSON: Dr. Newman?

DR. NEWMAN: Yes.

DR. SWENSON: Dr. Moss?

DR. MOSS: Yes.

DR. SWENSON: Dr. Gay?

DR. GAY: Yes, and I would like to state

that these drugs do seem to have a clear and

profound impact on morbidity and mortality overall

in a very positive sense. We have seen the overall

declines in some of the data presented here for the

occurrence of exacerbations and on morbidity and

mortality. There does seem to be a true signal

present but we cannot disregard the other things

319

that may contribute to this signal in these

subpopulations. This does include access to

healthcare. This does include the use of inhaled

corticosteroids in these populations. This does

include the changes in treatment and management

patterns for certain subpopulations and certain

under-represented populations. It is going to be

extremely important, and I do believe that both

companies are making good efforts to attempt to

control for those factors in the subsequent studies

that they are beginning to perform, and it is going

to be extremely important to analyze this data on

the basis of those multiple factors to see if we

can make any impact on this true signal that does

exist. DR. SWENSON: D. Schatz?

DR. SCHATZ: Yes.

DR. SWENSON: Dr. Gardner?

DR. GARDNER: Yes, with caveats.

DR. SWENSON: Miss Sander?

MS. SANDER: Yes.

DR. SWENSON: And Dr. Schoenfeld?

DR. SCHOENFELD: Yes.

320

DR. SWENSON: My vote is yes as well. I

think that we have a unanimous vote here but,

clearly, the warning is there that none of us feels

100 percent yes. I think that possibly I am

stating the obvious but wish to have it for the

record.

We will now move then to the next question

which now turns on formoterol. Here the question

is the label for formoterol an

formoterol-containing products at this point does

not include warnings comparable to the warnings

that are present in the salmeterol products and,

based on currently available information, should

the label for formoterol-containing products

include warnings similar to those to the salmeterol

label?

This will be a yes/no vote, but I think we

have enough time here if people wish to say very

briefly why they voted one way or the other.

DR. SCHOENFELD: May we comment first so

we can communicate with each other before we have

to vote?

321

DR. SWENSON: That is a reasonable request

I think if we could limit to five to ten minutes

for those that wish to make it an open discussion

here, and I suspect you wish to lead off.

[Laughter]

DR. SCHOENFELD: Well, first, again a

back-of the-envelope calculation and, again, if

anybody who has better data wants to contradict the

calculation, but as I understand it, formoterol had

the same risk as salmeterol we would expect, based

on the number of patient-years of follow-up, 0.23

events--0.23, and this is based on 527, multiplied

by 16/52 to get 162 years of follow-up, and then

multiplying that by the event rate and dividing

that by 700 we get 0.23. So, the chance of not

seeing any event is roughly 80 percent, which is

good power actually for not seeing anything. So, I

think the fact that at least in clinical trials we

didn't see anything is not surprising because if

the risk was exactly the same in the two drugs the

chances of not seeing anything would be 80 percent.

That is the point I wanted to make.

322

Now, the issue as to whether a warning

should go across the class depends upon, I guess,

how similar with think things are in the class,

which is beyond the level of my expertise. They

apparently both work similarly. One drug works

faster, which may be an advantage in terms of

preventing these problems. But if anybody has any

comment--I guess there has been a comment by the

industrial representatives but if anybody on the

panel would like to make a comment to the extent

that this is a class, I would love to hear that.

DR. SWENSON: Dr. Gardner?

DR. GARDNER: I will hold until maybe

someone can answer his question.

DR. SWENSON: Dr. Gay, do you care to

comment?

DR. GAY: Certainly. Every other drug

that works at this receptor has shown effects with

the arginine/arginine subtype of receptor of

negative outcomes. We have seen it with albuterol.

We saw it with other shorter-acting forms of beta

agonist, and we have seen it with salmeterol as

323

well. Because of this concern that with certain

genotypes or phenotypes of this receptor we have

seen this effect, I have significant concern that

it is a class effect. We have not seen a study yet

that has been powered appropriately and designed

appropriately to look at whether or not this is

potentially the case win formoterol. However, with

the fact that every other beta agonist, both

short-acting and long-acting, has similar effects

at the receptor, I have concerns that until proven

otherwise we have to make ourselves believe that

formoterol may act the same way.

DR. SWENSON: Dr. Gardner?

DR,. GARDNER: I appreciate what the

question is trying to get at but I want to raise

another one. In looking at the formoterol insert,

given the data that we saw today related to

children, I don't think that what we saw is very

well communicated in the insert as it stands, and I

would like to suggest that in addition to the

question we are addressing, which I think probably

has mostly to do with class labeling and black

324

boxes, I would like to suggest that more attention

be paid to communicating what may be a heightened

risk in children and then, of course, update it as

more data become available.

DR. SWENSON: Dr. Schatz?

DR. SCHATZ: I guess I would like to get

some clarification from FDA people regarding what

the standard is for class labeling. If, for

example, you have a signal from one drug in a class

and no adequate data in the other to say yes or no,

is that typically an indication for a class

warning? And, would you only not have a class

warning if that other drug had adequate data to

show that the signal did not exist?

DR. CHOWDHURY: I want to first clarify

that the question is actually specific to

formoterol, not necessarily all long-acting beta

agonists. So, you can't probably use the term

class labeling. The specific question is on

formoterol.

DR. SCHATZ: Yes, but I would like to

understand the meaning of the class label, my point

325

being that if one assumes one has to have a class

label until you can prove the drug doesn't do

something, then I think everyone would agree we

have different types of studies. We don't have a

study that shows that formoterol doesn't do this.

But I think this issue of class labeling--what that

means, and we are talking not clinical; we are

talking about regulatory recommendations. So, I

think I need to understand the regulatory

environment better.

DR. MEYER: Right. I think that your

recommendation has to be informed by the degree

that you do, in fact, feel, as Dr. Gay pointed out,

that this in fact does represent a class effect

because, as you say, we do not have data one way or

the other. Obviously, many drugs have

idiosyncratic effects that will not be represented

by other members of their class. On the other

hand, there are situations where we can feel fairly

confident from the pharmacology as we understand

it, or other mechanisms as we understand them, that

this would extend to other members of the class. I

326

mean, you can take certain adverse events with ACE

inhibitors for instance where you know it is a

direct result of its pharmacology.

So, I think we would defer to your

expertise in that regard but I think you would need

to feel personally convinced that this probably

does represent a class effect and, therefore, it is

only fair to put it in formoterol's labeling. On

the other hand, if you thought that the

observation, the signal that has been seen in the

SNS study and the SMART study could be due to other

considerations of the way salmeterol itself

interacts at the beta receptor that might not apply

to formoterol, then if that is a significant

unknown for you, I would think your recommendation

would be I don't think this should be extended.

But the bottom line is we do not have the data one

way or the other. If the SNS and the SMART study

did not exist for salmeterol, we would be in a

situation of having really no idea at all about

this. We have those data for salmeterol; we don't

have them for formoterol. So, you know, it is just

327

an unknown whether formoterol would have similar

findings or not. So, again, we are deferring to

your expertise, and I hope I gave you enough of an

explanation there.

DR. SWENSON: I will ask Dr. Schoenfeld

then for his vote.

DR. SCHOENFELD: Again, I am sort of

voting beyond my expertise in a way because I know

this 80 percent chance but I don't really know the

biology of these drugs. But I think the prudent

thing would be some sort of warning, boxed label,

on formoterol that says that this effect has been

seen in another member of the class and it is

unknown whether it would apply to this member, but

at least to warn people that this could be an

effect either of the class of drugs or of the way

patients act when they are, in fact, on these

drugs, which is another possibility.

DR. SWENSON: Before I ask directly for

your--

DR. SCHOENFELD: So, the answer is it

shouldn't be--

328

DR. SWENSON: Wait one minute.

DR. SCHOENFELD: Sorry.

DR. SWENSON: One can abstain if you feel

so uncertain as to whether you should vote one way

or the other. An abstention is perfectly fine.

DR. SCHOENFELD: In other words, I believe

it should be marketed but I believe it should have

some kind of warning pointing people in the

direction that this has been found with other drugs

in the class.

DR. SWENSON: Miss Sander?

MS. SANDER: When I read this question, we

are talking about a similar warning label as the

one when we are looking at salmeterol. Is that

right? I don't know all the similarities. I do

know that there is information on this in the

package insert that I do think should be brought

forward, and there may be some other information

that should be made more prominent as well that we

have learned today. I don't know if I know how to

vote on this one so I think I may have to abstain.

I think there is information that should come

329

forward; I don't know that it needs to be the same

information as with Serevent. So, if that is a yes

or a no, I am not sure.

DR. SWENSON: I think we will count it as

an abstention.

MS. SANDER: Okay, thank you.

DR. SWENSON: Dr. Gardner?

DR. GARDNER: I guess I would have to say,

based on what we saw today, my answer would be no,

although I would like to comment that I agree with

Dr. Schoenfeld that some wording pointing at what

is known about salmeterol would be useful until we

have more data.

DR. SWENSON: Dr. Schatz?

DR. SCHATZ: I really don't like the idea

of making such and important decision with such

little information, and I definitely intended to

abstain but I guess it is one of those situations

where something has to be done. I think that in

the absence of being able to say that it is not a

class effect, I am leaning, and I will therefore,

vote yes, that labeling to say that it has been

330

shown in a drug of its class; it is not known

whether it is a class effect--I am not even

convinced the effect--I see the signal. There have

been mentioned other reasons for that that may also

not even show that the drug does it. So, I am not

even convinced there, but in the sense of letting

people know what exists so they can made the best

decision, I think I am in favor of having it say

that another drug of the class has shown this

signal and it is not clear whether it extends to

the other class.

DR. SWENSON: Dr. Meyer?

DR. MEYER: I just wanted to perhaps point

out-- and maybe this will help Dr. Schatz although

I guess he eventually came down on a yes

recommendation--that one of the implications of

having one label that has these boxed warnings and

one that does not to a patient or a practitioner

may be, well, if this one is unsafe then my patient

will be better off on this other one. I guess that

is the other thing you need to consider as well.

Is that disparity something you are comfortable

331

with?

DR. SWENSON: Dr. Gay?

DR. GAY: For the reasons I have already

stated, I will vote yes.

DR. SWENSON: Dr. Moss?

DR. MOSS: I would like to reiterate what

Dr. Meyer said because I think it is important. If

we sat here today and the SMART study only had

2,000 people in it, I think people would have

totally different conclusions. We would have two

small underpowered studied that maybe didn't show

anything. I think it would be a bad message to

send to the industry that if you terminate studies

earlier that can be potentially beneficial for you.

I am not saying that Novartis did that for that

reason, but I think it is very important that if we

don't know the information and there is a

possibility that it is a class effect, I think it

is very reasonable to have a warning on formoterol

that says that a similar class of drug has shown

adverse events. It is not implicating that drug

per se but it is just, again, relaying the

332

information that there is the possibility that

there are subpopulations that may have adverse

events to this class of medication. So, I would

say yes with that reasoning.

DR. SWENSON: Dr. Newman?

DR. NEWMAN: My answer is yes. I would

just like to give you a little bit of my basis for

it. I think that Dr. Schatz said it very well and

I would underscore it with a little more emphatic

yes. We know that there are similarities in the

chemistry, the pharmacology and mechanisms of

action. We know that clinicians will use these two

drugs rather interchangeably. And, I think that

Dr. Schoenfeld used the key word, which is

"prudent." Sometimes in the absence of as complete

medical information and scientific information as

we would like, we have to recommend something that

we think is the medically prudent thing to do. So,

that is my basis, given all the caveats of the

uncertainties of the data.

DR. SWENSON: Dr. Brantly?

DR. BRANTLY: Yes, I vote for a black box

333

warning for formoterol.

DR. SWENSON: Dr. Martinez?

DR. MARTINEZ: Yes.

DR. SWENSON: Dr. Kercsmar?

DR. KERCSMAR: Yes.

DR. SWENSON: Miss Schell?

MS. SCHELL: My vote is yes. I would just

like to state that I think that it is necessary for

the patient to have that information so they can

make an informed decision.

DR. SWENSON: Dr. Prussin?

DR. PRUSSIN: Yes, with the caveat that it

has been shown in another member of the class but

hasn't been shown for this specific drug.

DR. SWENSON: And my vote is yes as well,

with that same caveat in all fairness to formoterol

that it be explicit that this has not been

established for that drug but of its class.

DR. GARDNER: Mr. Chairman, I want to

change my no vote to yes, given that my colleagues

also have expressed the caveat that caused me to

vote no. So, I agree with what you have said as

334

long as there are caveats so yes.

DR. SWENSON: Okay, fair enough. The vote

on this then was 12 yes and 1 abstention.

We move to the vote on the last question

on the slide, that is, based on currently available

information, do we agree that formoterol should

continue to be marketed in the United States. I

will ask Dr. Prussin to start the vote.

DR. PRUSSIN: Yes.

DR. SWENSON: Miss Schell?

MS. SCHELL: My vote is yes. Again, I

would like to reiterate with patient and physician

education as an important part of that.

DR. SWENSON: Dr. Kercsmar?

DR. KERCSMAR: Yes.

DR. SWENSON: Dr. Martinez?

DR. MARTINEZ: Yes, with the caveats and

conditions expressed before.

DR. SWENSON: Dr. Brantly?

DR. BRANTLY: Yes.

DR. SWENSON: Dr. Newman?

DR. NEWMAN: Yes.

335

DR. SWENSON: Dr. Moss?

DR. MOSS: Yes.

DR. SWENSON: Dr. Gay?

DR. GAY: Yes.

DR. SWENSON: Dr. Schatz?

DR. SCHATZ: Yes.

DR. SWENSON: Dr. Gardner?

DR. GARDNER: Yes.

DR. SWENSON: Miss Sander?

MS. SANDER: Yes.

DR. SWENSON: And Dr. Schoenfeld?

DR. SCHOENFELD: Yes.

DR. SWENSON: And my vote is yes as well.

So, that is a unanimous yes to that question.

We now move to recommendations to the

agency with ideas and opinions and recommendations

toward how we might further improve the

understanding of the nature and magnitude of the

risk of salmeterol first and then we will turn to

formoterol. I think that possibly much of what has

already been stated is contained in this but I

think I will go down the list to give people one

336

more opportunity to emphasize just these points

here with specific recommendations. Dr.

Schoenfeld?

DR. SCHOENFELD: I guess I really don't

know because, clearly, if you have what you

consider as a problem what you want to do is figure

out how to prevent the problem. So, I am not going

to suggest that big trials or studies would be done

to sort of determine the extent of the problem. It

seems to me that what is necessary is to try to

figure out how to prevent it and I don't think with

the study we really know what the problem is. And,

I am not really sure I know enough about this to

suggest ways that you could study the prevention of

this problem, whether this is an issue of patient

education or something else.

DR. SWENSON: Miss Sander?

MS. SANDER: My feeling is that the

overall question is what we do now and how we do

measure risks, and do we see greater risks in

certain populations of people, whether it is ethnic

or gender specific, or otherwise, or age related;

337

you know, effects on children versus on elderly

people or other groups. I think that we need to

look at things long-term. I made a few notes here

and I am trying to put them all in a capsule here,

but the challenge is that we are always going to

have questions about these medications,

particularly when I don't believe we are doing

enough to study what is happening when patients go

to the doctor in the first place to get the

medication, what kind of information they are being

given, and does that information empower them to

use the product properly at home, at school and on

the playground.

We just hear every day from families who

get their medications, go home, and have a zillion

questions that they don't seem to be able to get

answered in a manner that--well, I just think that

we could be doing a better job and I think that

whatever studies we do approach, we do need to make

certain that patients do have the information

necessary to use these drugs safely so that they do

know that Foradil and Serevent and Advair are three

338

different drugs and the instructions for using them

are going to be different. You know, I would hate

for patients to be left with the idea that they are

the same and that they are used the same.

One of my concerns is that patients are

going to use Foradil more frequently than they

should because of what they think when they hear

that it treats, you know, the breakthrough symptoms

as well as preventing symptoms. So, I think we

just need to have education included in these

studies. Anyway, I won't preach too much, except

to say that the language of asthma is very

important. What we say to patients must be said

within their hearing and their ability to follow

through whether it is a study or not.

DR. SWENSON: Well taken. Dr. Gardner?

DR. GARDNER: I think that the studies

that GSK described to us go a long way to helping

to answer some of our questions. The one thing I

have become concerned about is that they be

encouraged to stay on task and get these studies

done and reported to us in a timely fashion. We

339

are aware that these products are moving toward a

patent expiration and there may be motivation not

to finish things as quickly as we would like. If

there is a way for the agency to encourage that

they, in fact, be completed on time and reported in

a way that we can use the data, that would be my

interest.

DR. SWENSON: Dr. Schatz?

DR. SCHATZ: We are trying to consider

both science and practicality in terms of the

methodology. I think I would actually suggest

something fairly specific, which would be a

case-control surveillance design where one, in

fact, would involve hospitals with linked pharmacy

databases and actively be looking for those

patients who have these rare outcomes in a large

enough net that it doesn't take forever to have

those, to be able to match then sort of real time

with other decent controls, appropriate controls,

perhaps hospitalized and not intubated, perhaps

emergency department. And, I think in so doing you

have a chance to get both phenotypic information of

340

the type that we would like to have and even

genotypic information. It is different than a

case-control study in the way I mentioned. Without

giving this a huge amount of thought, I would put

that on the table to be considered and, of course,

the advantage of that is that that would be one

study that would be essentially done for both of

these types of medicines, as well as any other

factors that could, in fact, be associated with

these very severe and important but very uncommon

outcomes.

DR. SWENSON: I am going to step in here

simply because I want to follow on with what Dr.

Schatz had said. One recommendation I might have

is that another system with a broad database be

considered as a source for this particular

improvement information. That would be the VA, the

Veterans Administration. They are leading the way

in information management so this might be another

valuable source, akin to the state Medicare

analysis.

The other possibility would be, if we are

341

talking about a potential class effect, is to

consider that both companies merge efforts here, if

possible and practical, to look at this by

expanding numbers and resources to get at this

issue. Dr. Gay?

DR. GAY: I believe both companies at this

time, with a number of the studies that they have

commented on here today, are making good efforts

toward running the appropriate studies to help us

evaluate some of the questions we have brought up

as a committee. I would be hopeful that both

companies, with the fact that they either have

available or have in development combination

long-acting beta agonists and inhaled

corticosteroid formulations of their medications,

would look to do similar studies with those

medications to help to standardize for the lack of

inhaled corticosteroids in a number of these

studies, and to help us further evaluate where we

are in terms of some of the signals that we have

seen currently with the SMART study.

DR. SWENSON: Dr. Moss?

342

DR. MOSS: I think it is important to

remember that asthma is a heterogeneous disease,

and I think the answer is not to just to do an even

larger clinical trial. I think we would be sitting

here with probably having the same discussions.

Not to sound too NIH appropriate, I think it is

important to look at this as a translational

research project and to take the idea that there

are specific subsets of patients that may not

respond properly to this long-acting beta agonist

therapy; identifying a basic science laboratory

with a translational approach to what these

specific populations are, whether they are genetic

or acquired traits, and then study those

populations to see if there are truly harmful

outcomes in these smaller studies. I think that

would be a better way to go than just enrolling

60,000 people, or whatever, and I think the NIH

would like that too.

DR. SWENSON: Dr. Newman?

DR. NEWMAN: I wish I could roll back the

clock because I guess what I really want is for the

343

SMART study to reach its stopping point that was

set forward and really get to that point so we

would have something more complete. But I can't

have that now. That being said, I would underscore

the things that were said here. I don't have a

whole lot to add, except that as we go forward with

these additional studies and we become more

interested in both the clinical phenotypes of

asthma, as well as the genotypes of asthma, I would

encourage the sponsors not to ignore something that

we all know, which is that while the genes are

important and it is a heterogeneous disease, there

are also major environmental factors that are going

to affect an individual's risk of having severe

exacerbations and I haven't heard that really

discussed here.

People look conventionally at tobacco.

That is important. But we know that many cases of

asthma, especially in adults, have their onset in

adulthood due to occupational and environmental

factors other than tobacco smoke being involved. I

think that those ought to be weighed in when you

344

are designing studies that look at the genetic

factors. We have to understand the somewhat more

complex interactions and confounding effects, as

well as true attributable risk related to

environmental factors added to the understanding of

the genetics.

DR. SWENSON: Dr. Brantly?

DR. BRANTLY: I would echo Dr. Moss'

belief about pursuing subgroups. But I think that

one of the keys to trying to understand about this

small group of people who die is that we need to

find their phenotype. One approach to doing that

is obviously pursuing exactly what that phenotype

was at time of death, and to couple that, for

instance, with a VA study in which we actually are

able to forensically dissect what those patients

were like at that time would be a very valuable

thing. I would encourage the sponsors to consider

partnering specifically with the VA to approach

that and perhaps capture that death phenotype that

we are seeing. It may open up a wide array of

possible mechanisms which we might be able to look

345

at in the future.

DR. SWENSON: Dr. Martinez?

DR. MARTINEZ: I fully agree that at the

present time further assessment of the potential

existence of this increased risk, although

interesting, shouldn't be the center of the

attention. My general assessment of the data that

has been presented to us is that there is a signal

here, and I think the main objective will need to

be to understand what the signal is and in which

way it could be prevented.

I completely agree with Dr. Brantly that

trying to define the phenotype and perhaps the

genotype of these subjects is a great objective

that I think needs to be pursued. This will have

an additional advantage--in Spanish there is a

saying that not all bad things come to harm--which

is to better understand brittle asthma. I think

this is a good opportunity to understand who are

the subjects who have this severe form of the

disease who probably represent a significant

proportion if you think about it, of mortality for

346

this disease which is relatively low.

My recommendation also, since I have been

doing some genetic studies for the last ten years

in this disease, is that more than the more common

polymorphisms expressed in any of these potential

candidate genes, rare polymorphisms may be the

crucial factors here. Therefore, the approach of

doing this by genotyping for common polymorphisms

found in the general population may not be

successful. Perhaps a better approach could be a

more profound re-sequencing of individuals who have

this phenotype to try to determine if rare

polymorphisms are present in them that are not

present in the population as a whole. I don't see

that methodological approach mentioned and I think

it would be very important. It is more expensive

but, at the same time, it could give very

interesting results because polymorphisms that are

present, say, in 1/1,000 or 3 or 4/1,000 are not

going to be detected in linkage to equilibrium--I

am sorry to give such a complicated answer but in

linkage to equilibrium with the common ones. They

347

will have to be found specifically in each subject

who has these polymorphisms which perhaps are

increasing the risk in the way we are talking

about.

So, I strongly recommend to the industry

to search for different approaches both in genetic

and phenotypic studies to determine who the

subjects are and to ensure that in that way we can

have some sort of preventive strategy.

DR. SWENSON: Dr. Kercsmar?

DR. KERCSMAR: I agree completely with

what Dr. Moss, Dr. Brantly and what Dr. Martinez

just stated quite elegantly, that we are clearly

moving into an era of one size does not fit all for

asthma therapy, and pharmacogenetics is really

going to be what will direct us to effective and

safe therapy, and it is only through identifying

those high risk phenotypes that we will be able to

do that. So, I would agree completely with what

they said.

DR. SWENSON: Miss Schell?

MS. SCHELL: I would just like to make a

348

couple of statements. When we look at asthma, as a

practitioner at the bedside, there are many

components and asthma is very individualized on the

patient. And, when you look at the components of

asthma management, clearly, medication is one of

the components that we look at. But I also would

like to see a study that would look at factors that

could affect how much medication you are giving,

including the environmental factors, the factors of

education and compliance, all those things that

patients are not very well at doing yet and how is

that affecting how much medication you are going to

have to give them. So, basically looking at all

parts of asthma management on an individual basis I

think is important to include in a study.

DR. SWENSON: Dr. Prussin?

DR. PRUSSIN: I think it is important to

remember the public health impact of the question

we are debating. Asthma is an incredibly common

disease. Combination long-acting beta agonists and

inhaled corticosteroid therapy is the primary

therapy for moderate to severe asthma. As an

349

example of that, Advair is I believe the number one

drug for GSK. So, these are problems that deserve

the input of some resources.

I am not an epidemiologist; I am a

translational researcher. So, just the opposite of

Dr. Schoenfeld, I really can't think about study

design in a way that really is going to make sense.

But I am concerned that the small-scaled studies

that we are looking at, these translational

studies, are all very good for understanding the

biology but not for answering real-world questions.

You know, they are fine for the future and for

projecting ourselves forward and they are

important. It is the kind of work I do. But I

would like to step back and say, five years from

now or three years for now, are we going to have

answers to the questions that we are talking about

today? And, I am not clear that these are going to

do that. Again, I don't know how to; that is

beyond my expertise.

I think, clearly, one of the questions we

should be addressing is looking at combination

350

therapy versus inhaled corticosteroid alone. Is

the safety signal that we saw here with salmeterol

alone still present when you have an inhaled

steroid on board in all of the patients? I think

that is a study that could be done.

The other concern I have, just in terms of

long-term studies, that has not been addressed is

the issue of monotherapy. Again, we have talked

about it but, you know, how do we get a grasp on

it? Both companies are selling long-acting beta

agonists and I don't think we have any handle on

patients that are taking this as monotherapy and

that is a public health problem that presumably

also is a safety problem. At least we should have

a handle on the magnitude of that problem, what

percentage of patients using these drugs are using

them as monotherapy and that could be addressed as

well.

So, these are some of the issues I see as

far as what studies could be done. Again, how you

answer the specifics on those studies I don't know

but I am hesitant just to look at mechanistic

351

studies, thinking that those are going to answer

real-world questions that we need addressed that

patients and practitioners would like addressed.

DR. SWENSON: We will move to the last

question, and I think it mirrors so much the former

question that this may go quickly but we should at

least allow for discussion relevant in particular

to formoterol.

I will start just so I don't have to have

Dr. Prussin go again initially. I think the

comments that I mentioned in regards to salmeterol

hold equally for formoterol and they are in the

record. Dr. Prussin?

DR. PRUSSIN: This is in terms of question

four?

DR. SWENSON: This is in terms of

Novartis.

DR. PRUSSIN: I think really basically the

same issues are involved. I think there is a

safety signal and, obviously, the numbers need to

be greater but I think roughly the parallel issue

to what I mentioned for number three.

352

DR. SWENSON: Miss Schell?

MS. SCHELL: I will stand by what I said

earlier. Thank you.

DR. SWENSON: Dr. Kercsmar?

DR. KERCSMAR: I think the same issues

apply to this drug as well.

DR. SWENSON: Dr. Martinez?

DR. MARTINEZ: The same issues.

DR. SWENSON: Dr. Brantly?

DR. BRANTLY: The same issues.

DR. SWENSON: Dr. Newman?

DR. NEWMAN: Actually I have more a

question than even a comment. Based on what we

discussed earlier about kind of a within class

warning that we would put with this drug based on

the SMART study and leave the question unanswered

for Novartis' product, why wouldn't we be saying

here today that there should be a SMART study

equivalent, only smarter, for formoterol? Right?

I mean, why wouldn't we be proposing that?

Otherwise we are going to be proposing what is

really kind of a watered down "well, another drug

353

in the class; may have some problems and so we put

it on this label but we don't know about this

drug." Well, don't we need to know about this drug

and its safety profile based on what we have heard

today?

DR. SWENSON: Dr. Moss?

DR. MOSS: I would like to build on that a

little bit. I think if Novartis thinks that there

are differences between their drug and Serevent,

that this is not a class effect, I think it would

be prudent for you, guys, to go and figure that

out. Right now, since there is not the information

and we feel that everything should be lumped

together, it might be in your best interest to go

and figure out, either doing the large clinical

study that Dr. Newman is talking about if you feel

that it is worthwhile, or to come up with smaller

translational studies that show that there are

differences. If that was, indeed, the case then

the recommendations of a class warning would be

changed based on that. So, that is the only thing

I would add.

354

DR. SWENSON: Dr. Gay?

DR. GAY: I have to agree with Dr. Newman

and Dr. Moss. They both stated it very eloquently.

I am sure Novartis must have some concerns with the

potential addition of a black box warning and I

feel that we do need to see the data that it is

different and they will have to run some study that

shows us the differences that exist with their drug

in order to not have it present.

DR. SWENSON: Dr. Schatz?

DR. SCHATZ: Well, I may see it a little

differently than what has been mentioned. I mean,

I think what we have learned from the SMART study

is that even a study of this magnitude doesn't

answer the question, and I understand the concept

of doing it smarter but I am not totally sure it is

possible to do it smarter in the way we want and

still have it accomplished. So, I would say that

we really do need to think of another design. Of

course, the one I suggested before I still feel

would answer that question because, in addition to

other drugs, it would have other issues. So, I

355

actually don't think another SMART study is the

right answer even though I would like to have more

information.

But the only other thing I would mention

is what Cal mentioned in terms of monotherapy and

how prevalent it is. There are some data out there

that could look for that. The data like the VA and

like Kaiser where they keep track of these sorts of

things, I think we could get a handle at least in

those populations as to how common monotherapy is.

I actually have seen some of that within the Kaiser

population and it is reassuringly small in at least

that population. So, I do think the answers to

some of those questions are there if we look.

DR. SWENSON: Dr. Gardner?

DR. GARDNER: I agree about study design.

I think that Dr. Knobil's response to why they

weren't able to get more than 30,000 people would

signal to us that Novartis starting the same study

again wouldn't be able to either, and I think a

more productive way to go would be the combination

of the Medicaid analyses that are planned, which

356

will address children, and possibly the

VA--encompassing all of them, the HMO research

network or Kaiser Permanente databases carefully

analyzed by someone who understands their

potentials and limitations could help us see the

real-world issues and answer some of the questions

that we have about both of these products about

monotherapy, about combination therapy and so on.

So, I would encourage that if there are

going to be more large-scale studies done that they

be done with existing databases of real-world data

so that we can get a better handle on it.

The only other thing that occurred to me

in what we saw today was the unique signal, as I

noticed it, relative to children in the formoterol

data and I wondered whether there should be

pediatric dosing studies or some other attention

paid specifically to children since that signal

came out of there. So, I would suggest that for

this product.

DR. SWENSON: Miss Sander?

MS. SANDER: Well, I have a question first

357

and that is do black box warning make--on drugs

that have black box warnings, are they more likely

to wind up on the prior authorization list of state

health programs? Does anyone know?

DR. TRONTELL: We don't have that

information. Are you talking about state Medicaid

formularies or health plans? We don't have that

information at the agency.

DR. SWENSON: Perhaps Glaxo might know

since they have been living with a black box

warning.

DR. WEEN: Generally speaking, the black

box warning doesn't have an effect on whether or

not your drug is listed on a formulary or not. It

is just something that is pertinent to prescribing

practice and what-have-you. So, the black box

warning is not an a priori reason why you would not

be on formulary.

MS. SANDER: I just wanted to make sure

because asthma is not a prior authorization kind of

disease. So, I just wanted to make certain I

clarified that. With regard to the question at

358

hand, really I think what we are talking about here

today is trying to figure out why patients are

dying and the data doesn't tell us. But the

parents of children with asthma who have died and

loved ones, family members whose loved ones have

died of asthma do tell us that asthma is a very

deceptive disease and I think that is one of the

reasons why we are having a hard time getting our

hands around, you know, what the data means in the

real world. I think that we all need to remember

that asthma is a serious condition; that 13 people

die of asthma every single day; and how does that

fit in with this data up here? I don't really know

except that somehow in the information that we

provide patients when we are crafting these studies

we need to make certain that the terminology and

the instructions that we are giving them are

extremely clear. I would just encourage that in

your studies when you refer to albuterol you do not

refer to it as a rescue drug because it is not

reserved for what patients feel are rescue

situations only. It is to prevent exercise. It is

359

to be used at the earliest point, earliest sign of

symptoms and when we refer to it as a rescue drug

patients are waiting too long to use it. I think

that it can also cloud some of the answers that you

may get from them when you are asking them about

rescue medication use. So, that is the only advice

that I have.

DR. SWENSON: Dr. Schoenfeld?

DR. SCHOENFELD: Actually, I think that

there is place for a clinical trial. Basically, it

seems to me, there would be place for a clinical

trial comparing the two drugs without a placebo

control. The advantage of this would be that the

trial could last longer because we don't really

know the timing of these events. They may be

transitory or they may be something that sort of

happens constant over time. So, it seems to me we

would learn that as well from a trial that could

last longer than half a year and would be a

comparison of the two drugs in terms of severe

asthma or asthma-related death. I think such a

trial would be practical. I think probably that is

360

the kind of trial that NIH should probably help

with, as they have helped with other heads-up

trials of commonly used pharmaceuticals.

I think the only thing that would sort of

make this less useful would be the situation if

there is a large number of other LABAs in the

pipeline. Then this would be less useful. But if

these are the two drugs, then they should be

compared.

DR. SWENSON: Dr. Schatz?

DR. SCHATZ: I would just ask a question

though. To power it for the outcomes we are

talking about, do you have any sense for what sort

of sample sizes would be required?

DR. SCHOENFELD: You would use the

statistic of 1/700 per year--

[Laughter]

--and you could come up with a sample size

quite easily, but I don't want to do it on the back

of an envelope.

DR. SWENSON: You might need a big

envelope! With that, I believe we have come to the

361

end of the meeting. I wish to thank the personnel

from both Novartis and GlaxoSmithKline for their

excellent presentations, the panel members and the

FDA. I think if there are no further points, then

the meeting is adjourned.

[Whereupon, at 4:45 p.m., the proceedings

were adjourned.]

- - -